Swansea University E-Theses

_________________________________________________________________________

Effect of plate characteristics on ink transfer in flexographic

printing.

Hamblyn, Anja

How to cite:
_________________________________________________________________________
Hamblyn, Anja (2015) Effect of plate characteristics on ink transfer in flexographic printing.. thesis, Swansea
University.
http://cronfa.swan.ac.uk/Record/cronfa42827

Use policy:
_________________________________________________________________________
This item is brought to you by Swansea University. Any person downloading material is agreeing to abide by the terms
of the repository licence: copies of full text items may be used or reproduced in any format or medium, without prior
permission for personal research or study, educational or non-commercial purposes only. The copyright for any work
remains with the original author unless otherwise specified. The full-text must not be sold in any format or medium
without the formal permission of the copyright holder. Permission for multiple reproductions should be obtained from
the original author.

Authors are personally responsible for adhering to copyright and publisher restrictions when uploading content to the
repository.

Please link to the metadata record in the Swansea University repository, Cronfa (link given in the citation reference
above.)

http://www.swansea.ac.uk/library/researchsupport/ris-support/
 Swansea University
Prifysgol Abertawe

Effect of Plate Characteristics on

Ink Transfer in Flexographic Printing

Anja Hamblyn

Dipl.-Ing. (FH), MSc

Subm itted to Swansea University in fu lfilm e n t o f the

requirem ents fo r the Degree o f Doctor of Philosophy

Swansea University

June 2015
 ProQuest Number: 10821217

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is d e p e n d e n t upon the quality of the copy subm itted.

In the unlikely e v e n t that the a u thor did not send a c o m p le te m anuscript

and there are missing pages, these will be noted. Also, if m aterial had to be rem oved,
a n o te will ind ica te the deletion.

uest
ProQuest 10821217

Published by ProQuest LLC(2018). C opyright of the Dissertation is held by the Author.

All rights reserved.

This work is protected against unauthorized copying under Title 17, United States C o d e
M icroform Edition © ProQuest LLC.

ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 4 8 1 0 6 - 1346
s$S frN Sx?
m i
I LIBRARY
 Sum m ary
Flexography is gaining market share from other printing processes by continuously improving its
performance through innovations of materials, technologies and processes. However, the
exploitation of the new developments is restricted by a lack of knowledge of their underlying
science. This research investigated the role of printing plate properties, namely feature geometries
resulting from imaging technologies, surface texturing of printing areas and their combined effect,
on ink transfer.

White light interferometry was used for the plate characterisation. New methods for the
determination of top geometry and surface area of halftone dots were developed, the latter using
post-processing with image analysis. Laboratory- and industrial-scale print trials investigated the
effect of plate parameters on print quality, notably optical density, print uniformity and defects,
under different process conditions.

The top geometry of halftone dots on the printing plate was found to be a material parameter and
independent from imaging technology. This was the first time that the predominantly concave
geometry has been quantified. Concavity together with a new dot deformation mechanisms were
suggested as a cause for the printing defects of halos and partially unprinted areas in the halftone
dot itself.

The ink transfer mechanisms for patterned printing areas were dependent on the ink type used. In
conjunction with lower-viscosity water-based ink many surface patterns performed in an analogous
manner to anilox cells and increased ink transfer to the prints. Higher-viscosity UV-curing ink
remained atop pattern features and caused unprinted areas on the substrate. Micro-patterns
imaged with "flat-top" imaging technology improved the ink laydown of halftones, but deteriorated
solid prints compared to pattern patches imaged by standard digital technology.

The new understanding of ink transfer from different feature geometries and surface texturing
gained through this research can be used as basis for simulations and optimise flexography as a
mass-production process for high-quality applications.

I2
D eclaration and Statements

Declaration

This work has not previously been accepted in substance for any degree and is not being

concurrently submitted in candidature for any degree.

Signed (Anja Hamblyn)

Date 12/12/2015 ^

Statement 1

This thesis is the result of my own investigations, except where otherwise stated. Other sources are

acknowledged by explicit references. A bibliography is appended.

Signed (Anja Hamblyn)

Date 12/12/2015

Statement 2

I give consent for the thesis, if accepted, to be available for photocopying, inter-library loan,

electronic storage and distribution, and for the title and summary to be made available to outside

organisations.

Signed ^ (Anja Hamblyn)

Date 12/12/2015

I3
Acknowledgem ents
This research has been enabled by the Knowledge Economy Skills Scholarship (KESS), a pan-Wales

higher level skills initiative led by Bangor University on behalf of the higher education sector in

Wales. It was funded in parts by the Welsh Assembly Government's European Social Fund

convergence programme for West Wales and the Valleys, and by Swansea Printing Technology as

industrial partner. Thanks to the KESS teams in Swansea and Bangor, in particular Cassy Froment,

Jane Kelly and Penny Dowdney who kept their cool despite all administrative resistance.

Generous and much appreciated support in form of materials was provided by Asahi Photoproducts,

Innovia Films, SGS International and SunChemical. Thanks also for professional and light-hearted

conversions to David Galton (Asahi), Wayne Peachey (SGS), Brian Crombie and Michael Simoni

(SunChemical).

I would like to offer my special thanks to David Gethin for his ever calm, patient and positive

influence, especially during the writing and correcting of this thesis. The reader is indebted to him

for making this volume more palatable, and I myself for much more. Thanks to Tim Claypole for the

opportunity to pursue a doctorate at the Welsh Centre for Printing and Coating (WCPC).

The help of Lorraine Leung (Swansea University), Justine Rexer and Sullivan Delanoe (Grenoble INP-

PAGORA, Grenoble Institute of Technology, Grenoble, France) with the small-scale print trials,

digitising and measuring of samples is very much appreciated. Thanks to Nick Croft, Dan Curtis, Peter

Davies, Davide Deganello, Simon Hamblyn, Chris Phillips and Sam Rolland for the support with

instruments and software.

A huge thank you to WCPC staff, home and visiting students - past and present - who sweetened

the occasionally bitter pill of research: Bahar Aslan, Andreas von Berchem, David Beynon, Ben

Clifford, Glyn Davies, M att Everett, Sakulrat Foulston, Neil Graddage, Christine Hammett, Charles

Jones, Tatyana Korochkina, Tim Mortensen, Ben Mogg, Adam Rees, and especially my fellow

doctorate sufferers Alexandra Lyashenko, Fotios Pelesis and Ingmar Petersen.

I would like to thank the following people for their assistance in obtaining and translating relevant

literature: J. Aspler (Canada), N. Bornemann and the staff at Universitatsbibliothek Potsdam

(Germany), T. Gotsick and J. Seymour (USA), J. T. Guthrie (UK), J. Johnson (Sweden), M. Kaplanova

(Czech Republic), S. Shilko (Belarus) and the staff at Converter E Cartotecnica (Italy).

Simon and Ava. Never ever again and at least three times more! I would not and could not have

done it without you.

14
Contents
Summary................................................................................................................................................................ 2
Declaration and Statem ents.............................................................................................................................. 3
Acknowledgements.............................................................................................................................................. 4
Contents................................................................................................................................................................. 5
List of Tables......................................................................................................................................................... 8
List of Figures........................................................................................................................................................ 9
Abbreviations and Symbols...............................................................................................................................15
Chapter 1 Introduction and Background.................................................................................................... 18
1.1 Aim and M otivation.......................................................................................................................... 18
1.2 Flexographic Printing Process........................................................................................................ 19
1.2.1 Printing Plates.......................................................................................................................... 19
1.2.1.1 Digital Flexo Plate Structure and Imaging Processes......................................................19
1.2.1.2 Oxygen Inhibition of Imaging Process...............................................................................23
1.2.2 Plate Image and Optical Density........................................................................................... 26
1.2.3 Printing U n it..............................................................................................................................28
1.2.4 Ink Delivery System and Anilox Roll...................................................................................... 29
1.2.5 Ink Composition and Drying Mechanisms...........................................................................31
1.3 Objectives and Thesis Layout...........................................................................................................32
Chapter 2 Literature Review......................................................................................................................... 34
2.1 Ink Transfer......................................................................................................................................... 34
2.1.1 Wettability and Ink Transfer Mechanisms...........................................................................34
2.1.2 Effect of Material Parameters on Ink Transfer................................................................... 36
2.1.2.1 Printing Plate.........................................................................................................................36
2.1.2.2 Ink and Substrate.................................................................................................................41
2.1.3 Effect of Process Parameters on Ink Transfer..................................................................... 42
2.2 Flexographic Printing Plates............................................................................................................. 44
2.2.1 Geometry of Dot Tops.............................................................................................................44
2.2.1.1 Effect of Top Geometry on Ink Transfer.......................................................................... 45
2.2.1.2 Related Publications............................................................................................................ 47
2.2.2 Texturing of Printing Surfaces............................................................................................... 48
2.3 Print Characterisation....................................................................................................................... 50
2.3.1 Introduction to Halftone M odels.......................................................................................... 50
2.3.2 Printing Defects........................................................................................................................ 52
2.3.2.1 Doughnuts and Halos.......................................................................................................... 52
2.3.2.2 Uncovered Areas.................................................................................................................. 54
2.3.2.3 Fingering Instabilities.......................................................................................................... 55
2.4 Conclusions......................................................................................................................................... 57
Chapter 3 Methodology................................................................................................................................. 60
3.1 Surface Profilometry......................................................................................................................... 60
3.1.1 Determination of Feature Dimensions and SurfaceRoughness..........................................61
3.1.2 Determination of Planar Surface Area of PrintingFeatures...............................................62
3.1.2.1 Planar Surface Area in Wyko Vision32..............................................................................63
3.1.2.2 Planar Surface Area in ImageJ............................................................................................ 66
3.2 Surface Energy................................................................................................................................... 68
3.2.1 Steady-state Surface Tension..................................................................................................68
3.2.2 Steady-state Surface Energy...................................................................................................68
3.3 Rheology............................................................................................................................................. 70
3.4 Printing.................................................................................................................................................71
3.4.1 Printing on Laboratory Printability T ester...........................................................................71
3.4.2 Industrial-scale Printing..........................................................................................................72

I5
 3.4.2.1 Setup of the T-Flex 5 0 8 ....................................................................................................... 73
3.4.2.2 Thin Film Pressure Sensors.................................................................................................74
3.4.2.3 Test for Cyclic Variations..................................................................................................... 75
3.5 Print Characterisation....................................................................................................................... 75
3.5.1 Visual Inspection of Samples by Microscopy.......................................................................76
3.5.2 Determination of Planar Surface Area and Volume of Printed Dots............................... 76
3.5.3 Determination of Optical Density by Spectrophotometry.................................................78
3.5.4 Digitisation of Prints by Scanning...........................................................................................79
3.5.5 Digital Image Analysis in ImageJ............................................................................................80
3.5.5.1 Determination of Print Density.......................................................................................... 80
3.5.5.2 Determination of Print Uniform ity....................................................................................82
3.6 Design of Experiments...................................................................................................................... 85
3.7 Closure................................................................................................................................................ 85
Chapter 4 Dot Top Geometry of Halftones.................................................................................................86
4.1 Materials - Plates, Substrate and Inks.......................................................................................... 86
4.2 Printing.................................................................................................................................................92
4.2.1 Printing Conditions................................................................................................................... 92
4.2.2 Printing Force by Thin Film Sensor.........................................................................................93
4.3 Print Characterisation....................................................................................................................... 95
4.3.1 Planar Surface Area and Volume of PrintedDots.................................................................95
4.3.2 Optical Density..........................................................................................................................97
4.4 Results and Discussions....................................................................................................................98
4.4.1 Characteristics of Plate Features...........................................................................................98
4.4.1.1 Planar Surface Area of Plate Features.............................................................................. 98
4.4.1.2 Geometry of Plate Features..............................................................................................102
4.4.2 Characteristics of Printed Features..................................................................................... 107
4.4.2.1 Ink Distribution across Printed Features.........................................................................107
4.4.2.2 Planar Surface Area of Printed Features.........................................................................115
4.4.2.3 Volume of Printed Features..............................................................................................120
4.4.2.4 Optical Density of Printed Features................................................................................ 125
4.4.2.5 Applicability and Comparison of Halftone M odels.......................................................127
4.4.3 Conclusions............................................................................................................................. 137
4.5 Closure.............................................................................................................................................. 138
Chapter 5 Meso-Patterns on Printing Plates........................................................................................... 139
5.1 Meso-Pattern Effect using aPrintability Tester.......................................................................... 140
5.1.1 Materials - Plates, Substrates and Inks..............................................................................140
5.1.2 Printing..................................................................................................................................... 144
5.1.3 Print Characterisation............................................................................................................145
5.1.3.1 Definition of Fingering Instabilities..................................................................................145
5.1.3.2 Introduction to DoE Analysis Approach..........................................................................147
5.1.4 Results and Discussions......................................................................................................... 149
5.1.4.1 Plate Characterisation........................................................................................................149
5.1.4.2 Effect of Ink T yp e................................................................................................................152
5.1.4.3 Effect of Substrate Type.....................................................................................................152
5.1.4.4 Effect of Plate M aterial......................................................................................................153
5.1.4.5 Effect of Anilox Volume......................................................................................................157
5.1.4.6 Effect of Printing Force......................................................................................................157
5.1.4.7 Effect of Printing Speed.....................................................................................................159
5.1.4.8 Effect of Surface Patterning..............................................................................................160
5.1.5 Conclusions............................................................................................................................. 166
5.2 Meso-Pattern Effect using anIndustrial Printing Press.............................................................. 167

I6
 5.2.1 Materials - Plate, Substrates and In k ................................................................................. 167
5.2.2 Printing and Print Characterisation......................................................................................167
5.2.3 Results and Discussions..........................................................................................................170
5.2.4 Conclusions............................................................................................................................. 174
5.3 Closure............................................................................................................................................... 175
Chapter 6 Microcell Patterns on Printing Plates..................................................................................... 176
6.1 Microcell Pattern Effect using a PrintabilityTester..................................................................... 177
6.1.1 Materials, Printing and Print Characterisation..................................................................177
6.1.2 Printing and Print Characterisation......................................................................................179
6.1.3 Results and Discussions..........................................................................................................180
6.1.3.1 Plate Characterisation........................................................................................................ 180
6.1.3.2 Effect of Plate Material andImaging Technology........................................................... 181
6.1.3.3 Effect of Microcell versusMeso-patterns........................................................................ 184
6.1.3.4 Effect of Pattern Scaling.....................................................................................................191
6.1.4 Conclusions............................................................................................................................. 194
6.2 Microcell Pattern Effect using an IndustrialPrinting Press......................................................... 196
6.2.1 Materials, Printing and Print Characterisation.................................................................. 196
6.2.2 Results and Discussions..........................................................................................................198
6.2.2.1 Plate Characterisation........................................................................................................ 198
6.2.2.2 Microcells on Solids............................................................................................................ 202
6.2.2.3 Microcells on Halftones.....................................................................................................208
6.2.3 Conclusions............................................................................................................................. 213
6.3 Closure...............................................................................................................................................214
Chapter 7 Conclusions and Recommendations....................................................................................... 215
7.1 Conclusions....................................................................................................................................... 215
7.2 Further W ork..................................................................................................................................... 219
7.3 Industrial Recommendations......................................................................................................... 220
A Appendix...................................................................................................................................................222
A .l Macros for ImageJ............................................................................................................................ 222
A.2 Screening of Scanning Parameters............................................................................................... 224
A.2.1 Selection of Scan Colour....................................................................................................... 224
A.2.2 Selection of Scan Resolution................................................................................................ 225
A.2.3 Determination of Temporal Stability and Consistency.................................................... 225
A.2.4 Determination of Linearity and Gamma Correction.........................................................229
A.2.5 Selection of File Form at........................................................................................................ 231
A.2.6 Selection of ROI Size...............................................................................................................232
A.3 Macro - Print Uniformity and Optical Density inImageJ........................................................... 233
A.4 Striation of Printing Plate Surface................................................................................................. 233
A.5 Area Transfer Ratios........................................................................................................................ 235
A.6 Volume Transfer Ratios...................................................................................................................237
A.7 Plate Characterisation..................................................................................................................... 239
A.8 Example of Full Data Set for Main Factor andInteraction Effects.............................................242
A.9 Printing Defects of AFP-DSH Plate................................................................................................ 242
A.10 Print Characterisation..................................................................................................................... 245
Glossary.............................................................................................................................................................. 246
Bibliography...................................................................................................................................................... 248

17
List of Tables
Table 3.1: Measurement parameters for Spectrolino with ColorScout A+ measurement table 79
Table 3.2: Scan parameters for digitisation of prints....................................................................................80
Table 4.1: Parameter combinations for the generation of different feature geometries on printing
plate...................................................................................................................................................................... 87
Table 4.2: Ink surface tension of SunChemical Solarflex Nova SL Pro DK03 series process colours ....91
Table 4.3: Rheometer settings for ink pre-conditioning and measurement geometry (inks of
industrial printing press studies).....................................................................................................................91
Table 4.4: Rheometer settings for ink viscosity determination (inks of industrial printing press
studies)..................................................................................................................................................................91
Table 4.5: Combination of print parameters for dot geometry tria l......................................................... 92
Table 4.6: Anilox specifications for dot geometry trial................................................................................ 93
Table 4.7: Combinations of parameters for which planar surface area and volume of printed dots was
determ ined..........................................................................................................................................................95
Table 4.8: Comparison of model errors for different halftone models at 100 Ipi line ruling 129
Table 4.9: Comparison of model errors for Yule-Nielsen and Expanded Murray-Davies halftone
models.................................................................................................................................................................133
Table 4.10: Comparison of model errors for different halftone models at 150 Ipi line ruling............ 135
Table 5.1: Selected properties of plate materials used for meso-pattern study on the IGT-F1.......... 140
Table 5.2: Specifications of meso-patterns.................................................................................................. 141
Table 5.3: Material properties of substrates used for surface patterning studies on the IGT-F1.......142
Table 5.4: Rheometer settings for ink pre-conditioning and measurement geometry (inks for IGT-F1
studies)................................................................................................................................................................143
Table 5.5: Rheometer settings for ink viscosity determination (inks for IGT-F1 studies)..................... 143
Table 5.6: Parameter levels for experimental plan of meso-pattern study on the IGT-F1...................144
Table 5.7: Parameter levels for experimental plan of meso-pattern trial on the T-Flex 50 8 .............. 169
Table 5.8: Volume specification of anilox bands........................................................................................ 169
Table 5.9: Combination of print parameters for meso-pattern trial on the T-Flex 5 0 8 ....................... 170
Table 6.1: Selected properties of plate material used for micro-pattern study on the IGT-F1........... 177
Table 6.2: Surface patterns of the three printing plates used for micro-pattern study on the IGT-F1
............................................................................................................................................................................. 178
Table 6.3: Specifications of microcell patterns............................................................................................179
Table A .l: Scan conditions for determination of scanner's temporal stability and consistency.........227
Table A.2: Distance and depth of striae........................................................................................................235
Table A.3: Exemplary DoE data set of plain solid reference (meso-pattern trial on the IGT-F1)........242
List of Figures
Figure 1.1:
Market shares won by flexography from other printing processes.......................................18
Figure 1.2:
Schematic of terminology referring to the structure and geometry of printing plates 20
Figure 1.3:
Schematic of digital plate-making process..................................................................................20
Figure 1.4:
Comparison of conventional Gaussian and SQUAREspot laser imaging technology...........21
Figure 1.5:
Effects of oxygen inhibition on feature geometry of curing halftone do ts...........................23
Figure 1.6:
Schematic of industrial solutions against oxygen inhibition.................................................... 25
Figure 1.7:
Typical image elements in functional printing and graphical printing.................................. 26
Figure 1.8: Representation of different tonal values by dots in the unit square.....................................27
Figure 1.9: Schematic of physical and optical dot gain.................................................................................28
Figure 1.10: Schematic of flexographic printing unit.................................................................................... 28
Figure 1.11: Enclosed ink delivery system ......................................................................................................29
Figure 1.12: Selected anilox engraving geometries...................................................................................... 30
Figure 1.13: Schematic of interaction between plate feature and cell opening size.............................. 30
Figure 2.1: Contact angle 0 at triple contact line of liquid, solid and vapour phase............................... 35
Figure 2.2: Schematic of the pressure distribution in the printing nip and film splitting by the
mechanisms of cavitation and filam entation................................................................................................ 36
Figure 2.3: Mechanisms of plate deformation - top expansion and shoulder barrelling...................... 37
Figure 2.4: Folding-over of model halftone dot under axial load............................................................... 37
Figure 2.5: Flexographic process variables.....................................................................................................42
Figure 2.6: Comparison of microphotographs of standard digital to "flat-topped" plate features and
their resultant print quality................................................................................................................................45
Figure 2.7: Microphotographs of halftone dots suggesting concave dot geometry on conventional
and standard digital plate..................................................................................................................................45
Figure 2.8: Mechanisms of liquid pinning to sharp edge of conical frustum............................................ 47
Figure 2.9: Apparent liquid pinning to imperfect sharp e d g e .....................................................................47
Figure 2.10: Secondary asperities on the sides of a conical frustum........................................................ 48
Figure 2.11: Schematic of underlying assumptions for different halftone models................................ 51
Figure 2.12: Printing defect of doughnut and halo....................................................................................... 53
Figure 2.13: Schematic of potential cause for h a lo...................................................................................... 54
Figure 2.14: Different manifestations of fingering instabilities in prints................................................... 56
Figure 2.15: Schematic of potential ink transfer mechanisms depending on dot geom etry............... 58
Figure 2.16: Schematic of potential ink transfer mechanisms depending on surfacepatterning 59
Figure 3.1: False colour image of a chequer pattern on a printing p la te ..................................................61
Figure 3.2: Display option "2D analysis" in WYKO Vision32 software........................................................62
Figure 3.3: Challenge of determining the surface area for different feature geometries...................... 63
Figure 3.4: Customised display option in WYKO Vision32 software for determination of planar
surface area and volume.................................................................................................................................... 64
Figure 3.5: Comparison of 3-D appearance of halftone dot for different restoration and threshold
sequences in Vision32.........................................................................................................................................65
Figure 3.6: Image processing steps in ImageJ for determination of planar surface area of an isolated
printing featu re.................................................................................................................................................... 67
Figure 3.7: Schematic viscosity-shear rate diagram illustrating Newtonian, shear-thinning and shear-
thickening properties of fluids.......................................................................................................................... 70
Figure 3.8: Schematic of IGT-F1 printability tes te r........................................................................................72
Figure 3.9: Schematic of T-Flex 508 printing unit with UV lam p ................................................................ 73
Figure 3.10: Temporal variation in solid optical density measured on consecutive sheets to test for
cyclic variations....................................................................................................................................................75
Figure 3.11: Microscopic image of printed dot showing irregular contour of the main ink volume and
UCAs....................................................................................................................................................................... 76
Figure 3.12: Manual selection of ROI for determination of planar surface area and ink volume 77

19
Figure 3.13: Schematic of (natural) volume algorithm in Vision32 software and matching up of actual
ink volume with algorithm by data inversion............................................................................................... 78
Figure 3.14: Original image of prints versus positioning template for ColorScout A + ........................... 78
Figure 3.15: Example of GSL distribution in histogram for a printed and subsequently digitised
sample..................................................................................................................................................................81
Figure 3.16: Correlation of optical density (spectrophotometry) and MGSL (ImageJ) for experimental
and computed data pairs..................................................................................................................................82
Figure 3.17: Illustration of deficiency of Coefficient of Variation approach for print uniform ity 83
Figure 3.18: Correlation of uniformity ranking obtained by visual judgement and uniformity
parameter StDev obtained by image analysis software ImageJ.................................................................84
Figure 4.1: Image for print trial on dot geometry.........................................................................................88
Figure 4.2: Methodology for quantification of dot geom etry....................................................................89
Figure 4.3: Location of WU profiles along which the cup depth of the dot top was measured...........89
Figure 4.4: Comparison of surface striation in halftone dots and solid for Asahi plate......................... 89
Figure 4.5: Ink viscosity of SunChemical Solarflex Nova SL Pro DK03 series process colours............... 91
Figure 4.6: Possible locations along plate's bearer bars for force measurement using thin film sensor
.............................................................................................................................................................................. 93
Figure 4.7: Comparison of consistent setup of anilox-plate cylinder nip using anilox force................. 94
Figure 4.8: Comparison of consistent setup of printing nip using mean printing force......................... 94
Figure 4.9: Comparison of ink transfer (planar surface area and volume of printed halftones) for
different inks.......................................................................................................................................................96
Figure 4.10: WLI images showing printed dot shapes dependent on dot size and printing conditions
.............................................................................................................................................................................. 97
Figure 4.11: Plate gain for Asahi plate (standard digital imaging technology).........................................99
Figure 4.12: Plate gain for MacDermid plate (standard digital imaging technology)............................. 99
Figure 4.13: Plate for MacDermid Lux plate (LUX "flat-top" imaging technology)................................ 100
Figure 4.14: Plate gain for Kodak plate (Flexcel NX "flat-top" imaging technology).............................100
Figure 4.15: Comparison of dot shape for different plate materials and imaging technologies.........101
Figure 4.16: Profiles of 10% and 50% nominal area coverage dots at 100 Ipi line ruling..................... 102
Figure 4.17: Cup depth for Asahi plate (standard digital imaging technology)..................................... 103
Figure 4.18: Cup depth for MacDermid plate (standard digital imaging technology).......................... 103
Figure 4.19: Cup depth for MacDermid Lux plate (LUX "flat-top" imaging technology)...................... 104
Figure 4.20: Cup depth for Kodak plate (Flexcel NX "flat-top" imaging technology)............................104
Figure 4.21: Schematic of potential cause for cupping............................................................................. 105
Figure 4.22: Images of printed dots at 10% nominal area coverage and 100 Ipi line ruling for different
printing conditions showing the evolution of doughnut and halo defects............................................ 108
Figure 4.23: Comparison of printed halo with central diameter of dot cupping on printing plate.... 109
Figure 4.24: Schematic of potential interaction of cupping and h alo .....................................................109
Figure 4.25: Images of printed dots compared for different dot top geometries at 50% nominal area
coverage and 100 Ipi line ruling showing UCAs and ink residue around the printed dot edges 110
Figure 4.26: Schematic of the potential interaction of cupping and air entrapment............................I l l
Figure 4.27: Potential modelling sequence for the future study of printing defects such as halos and
UCAs.................................................................................................................................................................... 113
Figure 4.28: Comparison of ink residue around yellow and black dots showing varying amounts of
pigments............................................................................................................................................................114
Figure 4.29: ATRs for dots of 100 Ipi line ruling (1 thou engagement)....................................................116
Figure 4.30: ATRs for dots of 100 Ipi line ruling (4 thou engagement)....................................................116
Figure 4.31: ATRs under different printing conditions for 10% dots of 100 Ipi line ruling...................117
Figure 4.32: Comparison of magenta ink buildup on dots of 20% nominal area coverage at 100 Ipi line
ruling for two different plate materials........................................................................................................117
Figure 4.33: Comparison of ATRs for line rulings of 150 Ipi and 100 Ipi for Asahi p la te ...................... 119

I 10
Figure 4.34: VTRs for dots of 100 Ipi line ruling (150 ft/m in; 4 thou engagement)...............................121
Figure 4.35: VTRs for dots of 100 Ipi line ruling (300 ft/m in, 4 thou engagement)...............................121
Figure 4.36: Relationship of ink volume transferred and planar surface area of printed dot at 100 Ipi
line ruling...........................................................................................................................................................122
Figure 4.37: Mean film thickness for dots of 100 Ipi line ruling...............................................................122
Figure 4.38: Comparison of VTRs for line rulings of 150 Ipi and 100 Ipi for Asahi p la te ..................... 123
Figure 4.39: Relationship of ink volume transferred and planar surface area of printed dots
compared for 150 Ipi and 100 Ipi line ruling................................................................................................ 124
Figure 4.40: Mean film thickness compared for dots of 150 Ipi and 100 Ipi line ruling......................124
Figure 4.41: Comparison of cup geometry for similar dot sizes at different line rulings....................125
Figure 4.42: Optical density under different printing conditions for 70% dots of 100 Ipi lineruling. 126
Figure 4.43: Relationship of optical density and area coverage on print for dots of100 Ipi line ruling
sorted by imaging technology........................................................................................................................ 126
Figure 4.44: Relationship of optical density and area coverage on print compared for 150 Ipi and
100 Ipi line ruling.............................................................................................................................................. 127
Figure 4.45: Comparison of actual and calculated reflectance at 100 Ipi line ruling obtained from
different halftone models...............................................................................................................................128
Figure 4.46: Comparison of print defects included in solids produced by different plate materials. 129
Figure 4.47: Comparison of actual and calculated reflectance obtained from Murray-Davies model
for 100 Ipi line ruling sorted by imaging technology..................................................................................130
Figure 4.48: Comparison of actual and calculated reflectance obtained from Beer's Law for 100 Ipi
line ruling...........................................................................................................................................................131
Figure 4.49: Comparison of actual and calculated reflectance at 150 Ipi line ruling obtained from
different halftone models............................................................................................................................... 134
Figure 4.50: Comparison of actual and calculated reflectance obtained from Beer's Law for 150 Ipi
line ruling...........................................................................................................................................................135
Figure 4.51: Printed dot structure of highlights, midtones and shadows in photo-engraving 136
Figure 5.1: Schematic comparison of scale between meso- and micro-patterns................................. 139
Figure 5.2: Illustrations of the 12 meso-patterns and reference solid....................................................141
Figure 5.3: Viscosity of the UV-curing and water-based inks used for surface patterning studies on
the IGT-F1...........................................................................................................................................................143
Figure 5.4: Parameters and their potential interactions in meso-pattern study on the IGT-F1..........144
Figure 5.5: Classes of fingering instabilities as observed in the meso-pattern trials............................145
Figure 5.6: Revelation of fingering instabilities underneath bead p a tte rn ............................................ 147
Figure 5.7: Comparison of optical and histogram data for two print samples produced by checker
pattern at 45° t i l t ............................................................................................................................................. 148
Figure 5.8: Actual area coverage of meso-patterns on printing plate and coverage loss compared to
the artwork for the AFP-DSH p late................................................................................................................149
Figure 5.9: Actual area coverage of meso-patterns on printing plate and coverage loss compared to
the artwork for the AWP-DEF plate...............................................................................................................150
Figure 5.10: Comparison of nominal polka dot design and plate geometry captured by W LI............ 150
Figure 5.11: Main effect of ink type on MGSL and StDev compared for plain solid reference and all
surface patterns................................................................................................................................................150
Figure 5.12: Nonuniformity of ink films printed with polka dot pattern using different inks and
printing conditions (AFP-DSH plate, APCO substrate)................................................................................151
Figure 5.13: Main effect of substrate type on MGSL and StDev compared for plain solid reference
and all surface patterns...................................................................................................................................152
Figure 5.14: Main effect of plate type on MGSL and StDev compared for plain solid reference and all
surface patterns................................................................................................................................................ 153
Figure 5.15: Chequer pattern at 45° tilt on AWP-DEF plate material. Little change between fingering
defect regimes with printing conditions observed for water-based ink on APCO substrate.............. 155

I 11
Figure 5.16: Chequer pattern at 45° tilt on AWP-DEF plate material. Strong change between fingering
defect regimes with printing conditions observed for water-based ink on APCO substrate 156
Figure 5.17: Main effect of anilox volume on MGSL and StDev compared for plain solid reference and
all surface patterns.......................................................................................................................................... 157
Figure 5.18: Main effect of printing force on MGSL and StDev compared for plain solid reference and
all surface patterns.......................................................................................................................................... 158
Figure 5.19: Main effect of printing speed on MGSL and StDev compared for plain solid reference
and all surface patterns...................................................................................................................................159
Figure 5.20: Main effect of surface patterning on MGSL and StDev for all surface patterns compared
to the plain solid reference............................................................................................................................. 161
Figure 5.21: Effect of the meso-patterns on fingering demonstrated on the example of a single set of
printing conditions (water-based ink on APCO substrate, AWP-DEF plate m aterial)...........................164
Figure 5.22: Image for meso-pattern trial on the T-Flex 5 0 8 ................................................................... 168
Figure 5.23: Parameters and their interactions of meso-pattern trial on the T-Flex 5 0 8 ....................169
Figure 5.24: Mean optical densities for plain solid reference and all surface patterns in meso-pattern
trial on the T-Flex 5 0 8 ...................................................................................................................................... 171
Figure 5.25: Comparison of printing defects (areas of missing ink) in meso-pattern trial on the T-
Flex 5 0 8 ...............................................................................................................................................................171
Figure 5.26: The 15 largest main factor and interaction effects on optical density for the example of
chequer pattern at 45° tilt on the T-Flex 5 0 8 .............................................................................................. 172
Figure 6.1: Illustrations of microcell patterns.............................................................................................. 179
Figure 6.2: Parameters and their potential interactions in micro-pattern study on the IGT-F1 179
Figure 6.3: Actual area coverage on printing plate and coverage loss compared to the artwork for
MacDermid Lux Plate 1 .................................................................................................................................... 180
Figure 6.4: Actual area coverage on printing plate and coverage loss compared to the artwork for
MacDermid Lux Plate 2 .................................................................................................................................... 180
Figure 6.5: Actual area coverage on printing plate and coverage loss compared to the artwork for
MacDermid Lux Plate 3 .................................................................................................................................... 181
Figure 6.6: Comparison of mean MGSL for MacDermid Lux (Plate 1) and Asahi plates....................... 181
Figure 6.7: Surface profile of negative hexagon pattern on different plate types................................ 182
Figure 6.8: Comparison of mean StDev for MacDermid Lux (Plate 1) and Asahi plates....................... 183
Figure 6.9: Effect of printing force on MGSL for MacDermid Lux (Plate 1) and Asahi plates.............. 184
Figure 6.10: Effect of printing force on StDev for MacDermid Lux (Plate 1) andAsahi plates.............. 184
Figure 6.11: Mean MGSL and mean StDev for plain solid reference andall surfacepatterns on
MacDermid Lux (Plate 2 ).................................................................................................................................. 185
Figure 6.12: Main effect of surface patterning on MGSL and StDev for all surface patterns on
MacDermid Lux (Plate 2).................................................................................................................................. 185
Figure 6.13: Correlation of area coverage on MacDermid Lux (Plate 2) with mean MGSL..................186
Figure 6.14: Correlation of recess volume on MacDermid Lux (Plate 2) with mean MGSL for all
surface patterns except V50/50, tiles and positive hexagons.................................................................. 187
Figure 6.15: Example profiles of selected microcell and meso-patterns featuring recesses 187
Figure 6.16: Lack of defects in prints made with UV-curing ink (APCO substrate, MacDermid Lux
(Plate 2)).............................................................................................................................................................. 188
Figure 6.17: Correlation of area coverage on MacDermid Lux (Plate 2) with mean StDev..................188
Figure 6.18: Correlation of recess volume on MacDermid Lux (Plate 2) with mean StDev for all
surface patterns except V50/50, tiles and positive hexagons...................................................................189
Figure 6.19: Transition of fingering defect regimes with area coverage in descending order (water-
based ink on APCO substrate, MacDermid Lux (Plate 2 )).......................................................................... 190
Figure 6.20: Scaling example of chequer meso-pattern designs.............................................................. 191
Figure 6.21: MGSL and StDev of entire experimental plan for plain solid reference and all surface
patterns on MacDermid Lux (Plate 3 )............................................................................................................192

I 12
Figure 6.22: Main factor effects of patterning on MGSL and StDev for different pattern scales on
MacDermid Lux (Plate 3)................................................................................................................................. 192
Figure 6.23: Transition of printing defects with pattern scaling for chequer pattern at 45° tilt (APCO
substrate)...........................................................................................................................................................193
Figure 6.24: Samples indicating ink squeeze at high printing force (APCO substrate)..........................193
Figure 6.25: Image for microcell trial on the T-Flex 5 0 8 ............................................................................ 197
Figure 6.26: Actual area coverage of micro- and meso-patterns on printing plates investigated on the
T-Flex 5 0 8 ...........................................................................................................................................................198
Figure 6.27: Surface profile of solid superimposed with MC09P_L microcell pattern ......................... 199
Figure 6.28: Example of microcells superimposed on halftone dots of all three plate types 200
Figure 6.29: Effect of microcells on cup depth of halftone dots on p late.............................................. 200
Figure 6.30: Area coverage of halftone dots with and without microcells based on dot edge 201
Figure 6.31: Area coverage of halftone dots with and without microcells based on actual data points
on MacDermid Lux plates...............................................................................................................................201
Figure 6.32: Comparison of optical densities for plain solid reference and microcells patches for all
plate materials..................................................................................................................................................202
Figure 6.33: Difference in optical density between solids without and with microcells on Asahi plate
for different printing conditions....................................................................................................................203
Figure 6.34: Difference in optical density between solids without and with microcells on MacDermid
plate for different printing conditions.......................................................................................................... 203
Figure 6.35: Difference in optical density between solids without and with microcells on MacDermid
Lux plate for different printing conditions................................................................................................... 204
Figure 6.36: Correlation of actual area coverage and optical density for the microcell patterns for all
plate materials..................................................................................................................................................204
Figure 6.37: Printing defects (UCAs) in printed reference solids caused by air entrapm ent.............. 205
Figure 6.38: Orientation of UCAs in prints along lines similar in frequency to striae on Asahi printing
p late....................................................................................................................................................................206
Figure 6.39: Change of size and distribution of UCAs with increasing engagement (Asahi plate) 207
Figure 6.40: Schematic of redistribution of entrapped air bubbles by microcells................................. 207
Figure 6.41: Distribution of UCAs in relationship to microcell location on different printing plate
materials (MC16P pattern ).............................................................................................................................208
Figure 6.42: Difference in optical density between halftones without and with microcells (Asahi
plate)...................................................................................................................................................................209
Figure 6.43: Difference in optical density between halftones without and with microcells
(MacDermid plate)........................................................................................................................................... 209
Figure 6.44: Comparison of print uniformity produced by halftone dots imaged without and with
microcells on Asahi and MacDermid plate m aterials.................................................................................210
Figure 6.45: Difference in optical density between halftones without and with microcells
(MacDermid Lux p la te )....................................................................................................................................211
Figure 6.46: Comparison of print uniformity produced by halftone dots without and with microcells
on MacDermid Lux plate m aterial.................................................................................................................211
Figure 6.47: Optical densities of halftones without and with microcells for all plate materials 212
Figure A .l: Normalised histogram of GSL distribution for patch "reference solid" at different scan
resolutions.........................................................................................................................................................226
Figure A.2: Water-based ink prints - MGSL and StDev of test scans in consecutive order..................228
Figure A.3: UV-curing ink prints - MGSL and StDev of test scans in consecutive order..................... 229
Figure A.4: Reference greyscale as part of the Color Checker Classic..................................................... 230
Figure A.5: Comparison of normalised MGSL (determined by scanning) and normalised luminance
(obtained by spectrophotometry) showing non-linear manipulation of scanner data using "no colour
correction" option............................................................................................................................................ 230

I 13
Figure A.6: Normalised histogram of GSL distribution for patch "reference solid" at different ROI
sizes.....................................................................................................................................................................232
Figure A.7: Comparison of surface striation in halftone dots and solids on printing plates............... 233
Figure A.8: Comparison of ATRs for line rulings of 150 Ipi and 100 Ipi for Kodak plate........ 235
Figure A.9: Comparison of ATRs for line rulings of 150 Ipi and 100 Ipi for MacDermid plate.236
Figure A.10: Comparison of ATRs for line rulings of 150Ipi and 100 Ipi for MacDermid Lux plate.... 236
Figure A .ll: Comparison of VTRs for line rulings of 150Ipi and 100 Ipi for Kodak plate..................... 237
Figure A.12: Comparison of VTRs for line rulings of 150Ipi and 100 Ipi for MacDermid plate 237
Figure A.13: Comparison of VTRs for line rulings of 150Ipiand 100 Ipi for MacDermid Lux plate.... 238
Figure A.14: Nominal versus actual feature geometries of original image on printing plates for the
meso-pattern trials on the IGT-F1.................................................................................................................239
Figure A.15: Chequer pattern without tilt on AFP-DSH plate material. Strong change between
fingering defect regimes with printing conditions observed for water-based ink on APCO substrate
.............................................................................................................................................................................243
Figure A.16: Polka dot pattern on AFP-DSH plate material. Strong change between fingering defect
regimes with printing conditions observed for water-based ink on APCO substrate.......................... 244
Figure A.17: Failure of meso-patterns to produce closed ink films on the T-Flex 5 0 8 ......................... 245

I 14
Abbreviations and Symbols

A b b r e v ia t io n s

2-D two-dimensional ISO International Organization for

Standardization
3-D three-dimensional
IR infrared
ANSI American National Standards Institute
MGSL mean greyscale level (histogram)
ATR Area Transfer Ratio
Ipi lines per inch
BS British Standard
LAMS laser ablation mask system
CIE Commission International de
I'Eclairage (International Commission OWRK Owens, Wendt, Rabel and Kaelble
on Illumination) (approach)

CMYK cyan, magenta, yellow and black ROI region of interest

("key") - the process colours
STC Surface Tension Component
CV Coefficient of Variation (approach)

D50 standard illuminant (daylight) StDev standard deviation of mean greyscale

level (histogram)
D65 standard illuminant (daylight - noon)
TIFF Tagged Image File Format
DoE Design of Experiments
thou one thousands of an inch (25.4 pm)
dpi dots per inch
UCA uncovered area (printing defect)
DTP dot top peak
UV ultraviolet
ft foot (0.3048 m)
VTR Volume Transfer Ratio
FT mean film thickness
WLI white light interferometry
GSL greyscale level (histogram)

I 15
L a t i n Sy m b o l s

Symbol Meaning Unit

A% area coverage % (percent)

a relative area (Noffke-Seymour equation) -

ATR Area Transfer Ratio -

C* chroma (CIE L*C*h* colourspace) -

Ca capillary number -

CV Coefficient of Variation -

D optical density -

D mean optical density -

E effect (DoE) [case dependent]

AE uncertainty of effect (DoE) [case dependent]

FT mean film thickness m

GSL greyscale level -

h nip height m

L* lightness (CIE L*C *h* colour space) -

N number of data points -

n n-value (Yule-Nielsen equation) -

R roughness m

Ra average roughness (entire surface area)

Rz average roughness

(10 greatest separations of surface points)

R2 correlation coefficient -

r radius m

t film thickness m

u velocity m- s 1

V v power factor (Expanded Murray-Davies equation) -

w w power factor (Expanded Murray-Davies equation) -

Y relative linear luminance (CIE XYZ colour space) -

z adjustment parameter (viscous fingering equation) -

I 16
G reek Sy m b o l s

Symbol Meaning Unit

/? (beta) reflectance factor -

y (gamma) linearity correction factor -

V (eta) viscosity Pa-s

6 (capital theta) contact angle ° (degree)

Oa advancing contact angle ° (degree)

0a,o advancing contact angle on a flat, horizontal surface ° (degree)

o (sigma) standard deviation [case dependent]

os (sigma) surface free energy of a solid mN/m

0| surface tension of a liquid

Osl interfacial tension between a solid and a liquid

cd dispersive component of surface free energy

op polar component of surface free energy

r (tau) transmission factor -

(p (phi) corner sharpness angle ° (degree)

I 17
Chapter 1 Introd uctio n and Background
Traditionally printing processes are used fo r the production of graphical items such as books,

magazines, calendars, wallpapers, laminates and packaging. Since printing is a fast, consistent and

relatively cheap means o f mass production (Hubler et al., 2002; Kempa et al., 2008), it also has

potential to expand into the m anufacture o f functional devices, fo r example photovoltaics, light

em itters, transistors, batteries and antibody sensors (Sauer, Bornemann and Dorsam, 2011; Hecker,

2013). For both, graphical and functional device printing, it is of the utm ost im portance to be able to

predict the outcom e o f the printing process in order to achieve the target quality and avoid wasting

resources during the printing. However, printers are continuously challenged by developments in

the printing m arket such as new materials, technologies and processes. Their introduction by

industry is often not accompanied by a good theoretical understanding o f the impact on the

underlying process mechanisms. Thus, technologies are not fully exploited due to a lack of

knowledge about the process conditions th a t could lead to optim isation.

1.1 Aim and Motivation

Over the last tw o decades the flexographic printing process has gained im portance w ithin the

printing m arket. It replaces o th e r printing technologies, because it can achieve large volume

production runs in short periods o f tim e w ith ever improving quality and low er costs (Figure 1.1).

The printing plate as the carrier o f the actual image inform ation stands at the heart of the printing

process.

Print Buyer Projections:

Flexo Replacing Other Print Processes

Other

Silkscreen replaced by flexo

Letterpress replaced by flexo

Cravure replaced by flexo

Offset/litho replaced by flexo

0% 10% 20% 30% 40% S0%

Figure 1.1: M a rk e t shares w on by flexography from other printing processes (LPC, 2010)
Recent years have seen numerous commercial developments in flexographic printing plate

technology including material properties, surface characteristics and imaging processes all of which

claim to improve print quality and increase efficiency. Ink transfer mechanisms are proposed based

on the outcomes of selected commercial print runs, which only serves to highlight inadequate

insight into the underlying science. The research reported in this thesis focusses on developing the

understanding of how the ink transfer from the printing plate is affected by plate properties, namely

surface geometry and patterning. The next section will give an overview of the flexographic printing

process, before discussing the research objectives and thesis structure in more detail.

1.2 Flexographic Printing Process

A summary of the current state of the flexographic printing process as relevant to this work is

provided in the following section. Particular emphasis is placed on the printing plate, plate-making

and -imaging. The second half of the section is concerned with the components of a flexographic

printing press and the printing ink as significant material parameter.

1.2.1 Printing Plates

For this work only the digital plate-making process is of importance as it currently prevails on the

market and has been employed to produce the plates used in this research project.

1.2.1.1 Digital Flexo Plate Structure and Imaging Processes

The general terminology used to describe the structure of the printing plate is provided in Figure 1.2.

The image information is contained in the form of raised features, and the ink transferred from the

image area at the top of these features. The creation of a digital printing plate for flexography

involves six main steps: (Dykes, 1999; Liu and Guthrie, 2003; Birkenshaw, 2004)

• Imaging

• Back exposure

• Front exposure

• Washout

• Drying and stabilisation

• Finishing and post-exposure

| 19
 Image area Line

Rate floor

Figure 1.2: Schematic of term inology referring to the structure and geom etry of printing plates
(adapted from S. Hamblyn, 2004)

- «
(a) (b)

(c) (d)

<e) (f)

Figure 1.3: Schematic of digital plate-m aking process: During laser imaging the carbon mask layer is partially removed
(a). Back exposure w ith UV-light cures the plate floor (b) to which the halftone dots, created by UV-light transmission
through the mask layer, are anchored in the fro nt exposure step (c). Uncured plate m aterial is washed out using a
suitable solvent and brushing (d). The solvent is then evaporated (e) and the plate surface finished under UV-light (f).

| 20
In its most simple form the digital printing plate consists o f a photopolym er layer sandwiched

between a form -stable polyester backing sheet and carbon mask layer. During the first step o f the

digital imaging process the image inform ation is transferred directly from the artw ork in the

com puter to the mask layer using laser ablation or sublim ation (Figure 1.3a). In the case o f laser

ablation the mask layer is sensitive to infrared (IR) radiation and an IR-laser physically destroys the

mask layer to reveal th e photop olym er underneath. This m ethod is referred to as laser ablation

mask system (LAMS). Mask layers fo r laser sublim ation are designed to change th e ir chemical

com position under laser exposure, w hile remaining intact. Imaging can take place using an IR- or UV-

laser. IR-sensitive layers contain sublim ation dyes which tu rn increasingly transparent proportional

to the IR-radiation received. UV-sensitive layers change th e ir density under UV-radiation and

therefore allow d iffe re n t am ounts of UV-light to advance through the mask layer to the

photopolym er. There is a second mask technology which is non-integrated and involves ink-jet

printing o f the image negative onto the photopolym er sheet (Dykes, 1999). No fu rth e r processing of

the imaged mask layer is required beyond the step of ink-jet printing.

Several types o f laser imaging technology are available on the market. The most common one is the

conventional Gaussian laser characterised by the high intensity at the centre o f the beam which falls

o ff tow ards the edges in a bell-shape pow er profile creating soft dots (Figure 1.4). The proprietary

SQUAREspot technology by Eastman Kodak (Rochester, New York, USA) utilises a beam o f uniform

lower pow er w hich is th o u g h t to result in hard dots w ith sharp edges. (Anderson, 2009)

Conventional Gaussian Imaging sQUAAEspot Imaging Technology

-15mlaunfc
10#mtcroni
spot swath

2400 dpi dot 2400 dpi dot -

Figure 1.4: Comparison of conventional Gaussian and SQUAREspot laser imaging technology
by the laser beam 's pow er profile and resultant pixel shape (Kodak, 2012)

Before the printing features can be imaged on the plate, a base fo r them has to be created by

exposing the photop olym er sheet to UV-A radiation through the backing sheet (Figure 1.3b). Once

I 21
the floor of the printing plate is established, the front exposure of the polymer takes place through

the mask layer (Figure 1.3c). Where the UV-A light penetrates through the mask, the curing of the

polymer takes place. Areas which are covered by the mask, and therefore not exposed, remain

uncured and soluble. The incident angle of the light and its diffraction within the material determine

the shape of the halftone dot on the printing plate. Light entering other than perpendicular results in

asymmetrical dots leaning towards the light source. Extended diffraction creates a wider base of the

dot and shallower shoulder angles. (Fioravanti, 2003) The duration of the exposure has to be

matched to the requirements of plate material and image. If the exposure takes place for too long,

small printing features in particular might get enlarged and the plate surface starts to crack (Liu,

Guthrie and Bryant, 2002). If the exposure is not long enough, the printing features do not connect

sufficiently to the plate floor and are removed during washout along with the mask layer and non­

cured polymer material (Figure 1.3d). Depending on the plate material a water- or solvent-based

solution is used for the washout combined with brushing.

Once the washout is completed, the remaining solution is evaporated from the plate surface and the

bulk of the material through drying in an oven (Figure 1.3e) which improves the feature quality. This

step is of particular importance for solvent-washable plates, as they swell and deform as a result of

solvent-polymer interaction. During a subsequent stabilisation phase the form stability of the plate

material is re-established. This drying and stabilisation step takes about 24 hours for solvent-

washable printing plates. Mechanical and thermal polymer removal methods aim to eliminate

washout and drying. In mechanical removal, instead of washing out the non-cured photopolymer

after the imaging step, uncured photopolymer is melted and absorbed with a cloth revealing the

hardened elements on the plate. The thermal removal evaporates the uncured material. (Novakovic,

Dedijer and Mahovic Poljacek, 2010)

After the drying of the plate, a residual tackiness of under-cured polymer might remain on the plate

surface. This can be removed by immersing the plate into a chlorine bath or exposing it to UV-C light.

This is called the chemical or light-finishing step. Finally, the plate is exposed to UV-A light again in

order to cure any remaining parts of the plate that might have been left un- or undercured during

the previous exposure steps (Figure 1.3f). After this the plate is ready to be cut to size and mounted

onto the plate cylinder.

Although the majority of digital printing plates follow this imaging process, a small proportion is

manufactured by other laser technologies. In direct laser imaging the energy of the laser beam is

sufficiently powerful to initiate the polymerisation reactions without requiring the exposure to UV-

light (Kipphan, 2001). For direct laser engraving, the entire polymer sheet, disregarding any printing

I 22
features, is cured during a fro n t exposure step. In the next step the polymerised material is partially

rem oved by a high-energy laser beam to expose printing features. This process is comparable to the

mechanical engraving of rubber plates. (Dykes, 1999; Birkenshaw, 2004)

1.2.1.2 Oxygen Inhibition o f Im aging Process

The key constituents of the digital photopolym er plate usually are

• a functionalised oligom er as the main structural com ponent of the polym er network,

• a m ono- or m ultifunctional m onom er as reactive diluent, and

• polym erisation initiators. (Decker, 1993; Liu and Guthrie, 2003)

UV-light initiates the free radical photopolym erisation in the material regions exposed. The process

takes place in fo u r steps: radical generation, initiation, propagation and term ination of the chain

reaction. Details o f the process can be found w ith , fo r example, Hageman (1989), Selli and Bellobono

(1993), Zum brum , Wilkes and Ward (1993) and Thompson (1998).

yV
< (U V )>
02 \ 02

Mask layer
Uncured photopolymer

Cured photopolymer

Polyester backing sheet

Figure 1.5: Effects of oxygen inhibition on feature geom etry of curing halftone dots
(dashed lines indicate nominal geom etry)

Oxygen inhibition is the prem ature term ina tion o f the polym erisation reaction by the interference of

m olecular oxygen from the am bient air. The oxygen inactivates the in itia to r radicals as well as the

interm ediate products of the initiation and propagation reactions, before these have the

o p p o rtu n ity to contribu te to the grow th o f the polym er chain. The inhibition is characterised by a

longer induction period before the start o f the polym erisation, reduction o f the rate of

polym erisation, thereby necessitating longer exposure times, and incom plete polym erisation of the

unsaturated educts. (Hageman, 1989; Selli and Bellobono, 1993) The significance o f oxygen

inhibition fo r the plate-making process is th a t the resultant geom etry of the printing features is
misshapen and smaller in size compared to the nominal geometry (Figure 1.5). Due to the rounding

o f the feature top, standard digital imaging methods resulting in this kind of appearance are usually

referred to as "round-top" technologies.

In the printing industry the problem of oxygen inhibition has been approached from two very

different angles. Firstly, rather than resolving the underlying issue during the plate-making process,

compensation is applied to the artwork by which the dots are imaged larger in the mask layer than

nominally required. If feature loss occurs during polymerisation, the level would only fall back to

about the nominally intended value instead of below (Mahovic Poljacek et al., 2013; Tomasegovic,

Mahovic Poljacek and Cigula, 2013b). However, this only addresses the symptoms of the problem

and requires knowledge of the magnitude of feature losses during the inhibited polymerisation.

(Kipphan, 2001; Charlesworth, 2008)

Solutions for the prevention of oxygen inhibition which are already in place for processes similar to

plate-making include:

• high-intensity light sources to decrease the necessary exposure time for the polymer and

providing a smaller timeframe for the oxygen to diffuse into the material;

• usage of oxygen barriers to prevent the diffusion of molecular oxygen into the material

(barriers can be polymer films directly applied to the polymer layer, sheets of glass or

coverage with liquids);

• removal of air from the exposure chamber through evacuation;

• displacement of air by inert gas or carbon dioxide in the exposure chamber;

• usage of oxygen scavengers;

• usage of surface-active initiators on the polymer;

• conversion of molecular oxygen into a state in which it readily forms complexes with

acceptors and can support the chain reaction of the photoinitiated polymerisation. (Broer,

1993; Selli and Bellobono, 1993; Takimoto, 1993; Shibanov, 2012)

I 24
 Figure 1.6: Schematic of industrial solutions against oxygen inhibition.
(a) Plate-im aging of standard digital plates affected by oxygen inhibition. Prevention of oxygen inhibition by
(b) high-intensity light exposure, (c) air replacem ent through inert gas, carbon dioxide or air evacuation,
(d) deaerated liquid as oxygen barrier, (e and f) different im plem entations of polym er film as barrier layer.

Several of these possible solutions have been adopted by d ifferent m anufacturers o f digital plate

materials. The m ethod o f using high-intensity UV-light is employed by DuPont and the Flint Group

(Figure 1.6b). PRPflexo replaces the m olecular oxygen in the imaging cham ber w ith carbon dioxide

gas (Figure 1.6c), although gas conditioning in general is considered to be not efficient enough. Most

oth e r companies use some means of oxygen barrier. Exposing the plate under deaerated w a te r is

less com m on during the main exposure steps, but is currently used by some companies fo r the post­

exposure of the printing plate (Figure 1.6d). The more w idely used barriers are polym er films. W hile

MacDermid laminates the barrier film on to p o f the ablated mask layer (MacDermid LUX technology;

Figure 1.6e), Kodak provides an integrated solution in which the mask layer and barrier layer are
integrated into a sheet separate from the main plate m aterial. The Kodak mask layer is imaged on

th e barrier film , before the film is laminated onto the polym er layer (Kodak Flexcel NX technology;

Figure 1.6f). Toyobo also employs the m ethod o f having the mask above the barrier layer. This

m ethod m ight result in a slight enlargem ent o f the printing feature, since the exposure light can

diffuse w ith in the barrier layer, thereby being projected into regions o f the photopolym er it was not

supposed to penetrate. Figure 1.6f shows the nom inally polymerised dot w ith an indication o f the

possible enlargem ent by the dashed lines. (Shibanov, 2012) These approaches to prevent oxygen

in hibition allegedly create plate features w ith a fla t top and sharp corners, thus being described by

the suppliers as "fla t-to p " technologies.

1.2.2 Plate Image and Optical Density

Functional device printing aims to transfer the same ink film thickness w ith all image elements

independent of th e ir size and geom etry (Figure 1.7a). Graphical printing on the other hand has to be

able to produce d iffe re n t gradients (Figure 1.7b) o f each colour component, fo r example in a

photographic image. To reproduce a range o f colours from a standard process colour ink set o f cyan,

magenta, yellow and black (CMYK), the image is separated into its CMYK components and each

separation divided into very small printing and non-printing areas. The human eye is unable to

resolve the tiny elements and therefore averages the colour impression of the visible area covered

w ith printed elements and the uncovered substrate in between (Figure 1.7c). An image region of

50% area coverage w ould be perceived as a 50% colour gradient (tonal value).

Figure 1.7: Typical image elem ents in functional printing (a) and graphical printing w here a gradient (b) has to be
replaced by halftone dots of different size (c) (Beynon, 2007)

The printing elements can be of d ifferen t shape such as circular, elliptical, diamond or square, but

are generally referred to as halftone dots. The desired area coverage can be achieved by several

different com binations of halftone dot size, num ber and spacing, so-called screening m ethods. In

I 26
this study only the conventional screening em ploying am plitude m odulation is used. By screening

conventionally, the image area is divided into unit squares whose size is determ ined by the spatial

w avelength, i.e. the line ruling. The am plitude m odulation refers to the varying area coverage of the

d o t w ith in the unit square dependent on the tonal value represented (Figure 1.8). In the case of

round dots, the dot diam eter reaches the edge o f the unit square for theoretical halftone values of

78.5%. At higher values neighbouring dots join together creating a closed printing area th a t is

characterised by holes. These halftones are generally referred to as shadows due to th e ir dark

resulting printed colour. The smallest, independent halftones are known as highlights which includes

area coverages up to 20% fo r the purpose o f this work. The halftones in the middle o f the range are

suitably called m idtones and only join each oth er at th e ir bases on the printing plate. "Solid" refers

to an area coverage o f 100%. Highlights, m idtones and shadows exhibit differences in polym erisation

during the plate-m aking process w hich w ill lead to variations in th e ir physical behaviour during the

printing process.

A
unit square

i
• • • %
10% 20% 30% 80% 90%

Figure 1.8: Representation of d ifferent tonal values by dots in the unit square

The designed nom inal area coverage deviates from the "optically effective area coverage" o f the

prints fo r tw o reasons. Firstly, plate deform ation and ink spreading cause actual dot enlargem ent

which is known as physical dot gain (Figure 1.9). Secondly, light gathering underneath ink films

creates the impression o f dot enlargem ent which is referred to as optical dot gain. Together physical

and optical dot gain account fo r the tonal value increase of printed halftones. (Kang, 1999; Kipphan,

2001) The to tal am ount o f light absorbed by the p rint is measured as optical density, D\

D = - log/? Equation 1.1

where (3 is the reflectance fa cto r o f the sample. The optical density is dependent on ink film

thickness and area coverage.

I 27
 Ink
A Substrate

Nominal
area coverage

Physica dot gain Optical dot gain

Figure 1.9: Schematic of physical and optical dot gain

1.2.1 P rinting Unit

The »rinting ink is fed into the anilox chamber which is in intim ate contact w ith the anilox roll (Figure

1.10 The ink floods the finely engraved cells on the anilox surface. The metering blade (also called

revese angle doctor blade) removes excess ink from the surface o f the anilox roll ensuring th a t ink

only emains in the engraved anilox cells.

Printing Plate & Impression

Plate Cylinder Cylinder

Enclosed
Anilox Chamber

Substrate

Anilox Roll

Figun 1.10: Schematic of flexographic printing unit containing ink chamber, anilox roll, plate cylinder w ith printing plate
and impression cylinder
The printing plate is fixed onto the plate cylinder using double-sided m ounting tape (also referred to

as cushion or backing tape). The raised plate features make contact w ith the anilox surface and pick

up ink from the anilox cells in the nip between plate and anilox cylinder.

In the printing nip (betw een plate and impression cylinder) the ink is transferred onto the substrate

under light pressure. The three cylinders (anilox, plate and impression) can all be moved relative to

each other to create d iffe re n t nip engagements. If plate and impression cylinder just about touch

and the mean contact force between them is near enough zero, the position is referred to as "kiss

engagem ent" o r "kiss contact". During printing this nip is form ed by cylinder engagement, and a

printing pressure is applied.

Several of these printing units can be installed in sequence on a printing press which allows several

types or colours o f ink to be transferred in d ifferent patterns next or on top of each other.

(Birkenshaw, 2004; W yatt, 2004)

1.2.4 In k D elivery System and Anilox Roll

There are tw o main types o f ink delivery systems - open ones w ith separate doctor blade and

enclosed ones w ith integrated doctor blade. The closed ink delivery system can be found in most

applications, because o f its health, economic and quality advantages (Figure 1.11). (Birkenshaw,

2004)

Metering Blade
Inb Supply

Anilox Roll

Containment Blade
Inb Return
Endosed
Anilox Chamber

Figure 1.11: Enclosed ink delivery system w ith anilox cham ber, doctor blades and anilox roll

The main com ponent o f the ink delivery system is the anilox roll. The anilox is a metal cylinder which

is engraved using laser technology or mechanical means. Traditionally, hexagonal cells shaped like

honeycombs orientated at 60° were used. However, there are now many different proprietary

| 29
engraving geom etries (Figure 1.12). A fter engraving the anilox roll is coated w ith chrome or ceramic

to make its surface more durable. (Birkenshaw, 2004)

Hexagonal Pyramid Quadrangular Tri-helical Z-flow

Figure 1.12: Selected anilox engraving geom etries (Flexcor, 2012)

The p rinting fe a tu re is well supported by several

cell walls.
VAAAAAAAS VAAAAAAAy VAAAAAAAS

The printing fe a tu re is insufficiently supported by

the cell walls. Increased fe a tu re deform ation
causes plate wear. Early stages o f dot dipping
VA7V_/VAA7VA7W m ig h t occur.

Dot dipping occurs and too much ink is

transferred to the p rinting fe a tu re and its sides.
This results in deviations in the printed image.
W \ J

k The circled features are n o t supported and w ill

experience dot dipping.

Figure 1.13: Schematic of interaction b etw een plate feature and cell opening size (images by Niggem eier, 2002)

The anilox roll is characterised by the cell geom etry, the num ber o f cells and volum e per unit area

carried. For stable and uniform ink transfer, each feature on the printing plate has to be sufficiently

supported by several anilox cell walls during contact (Figure 1.13). If the printing feature is

insufficiently supported by the cell walls, it is deform ed against the cylinder which increases the

stress on the plate m aterial and may lead to plate failure. If the size o f the printing feature becomes

sm aller than the anilox cell opening, the entire feature can be immersed into the cell and the sides

o f the feature w ill be inked. This is known as "d o t dipping". The dipping increases the am ount o f ink

| 30
transferred from this feature and results in a larger printed area. This highlights the importance of

matching the anilox screen ruling to the halftone line ruling on the plate typically at a minimum ratio

off three-to-one (Cherry, 2007). It also indicates how different feature geometries and surface

texturing might lead to altered ink transfer depending on parameter combinations.

1.2.5 Ink Composition and Drying Mechanisms

The term "ink" traditionally refers to a liquid that includes dispersed pigments and is used to make

graphical information visible on a substrate. Although this terminology is still appropriate for

graphical printing, its original definition does not apply anymore to the liquids used in the creation of

functional applications. However, since the fundamentals of the flexographic printing process

remain the same in both cases, no differentiation between the liquids is made, and for the purpose

o f consistency the term "ink" shall be used for any liquid transferred throughout this work.

There are three main groups of printing inks in flexography: solvent-based, water-based and

radiation-curing inks. The radiation-curing inks can be divided into those curing under ultra-violet

radiation (UV-curing) and by electron beam exposure (EB-curing). (Leach and Pierce, 2004)

Solvent- and water-based inks contain resin, colorant (dye or pigment) and additives dispersed in

their respective ink vehicle (solvent or water) (Wyatt, 2004). Such inks are dried by force-

evaporating the vehicle from the resin-pigment mix. This can be achieved by using hot-air or infrared

dryers.

UV-based inks cure through photopolymerisation of the ink composites. The ink contains a mix of

unsaturated oligomers and monomers to be cross-linked as well as a photoinitiator to start the

polymerisation process. The cross-linking leads to a quick hardening of the ink film, thereby allowing

high production speeds on a wide variety of substrates. The composition of the UV-curing ink is very

similar to that of photopolymer printing plates and follows the same polymerisation principles.

(Hargreaves, 2004)

EB-curing inks also harden through photopolymerisation, but they do not require photoinitiators,

since the free radicals necessary to start the polymerisation process can be created directly by the

high-energy photon impact. Although EB-curing inks do not contain any volatile organic compounds,

they are still considered to be hazardous in the curing process as the electron gun does not only emit

high-energy short-wavelength radiation, but also X-rays. The inks are very expensive, but even thick

ink films can be cured in a very short time.

I 31
1.3 Objectives and Thesis Layout
Various new types of halftone geometry resulting from different imaging technologies and the

surface texturing on printing areas allegedly improve the volume and consistency of ink released

from a printing plate. While several such proprietary systems are experiencing increasing popularity

in the market, the underlying mechanisms to validate the claims of improved performance have yet

to be established. Since earlier studies of the flexographic printing plate have barely touched on

these areas of interest, this research into new plate characteristics was instigated. It builds on

previous works involving conventional and standard digital plate technologies (Bould, 2001;

S. Hamblyn, 2004; Beynon, 2007). The research questions addressed are outlined below and

compiled in more detail in section 2.4.

• Which dot top geometries exist? Is the applicability of halftone models independent from

top geometry?

• Which effect do different dot top geometries and surface patterns have on ink transfer and

laydown - in particular on optical density, print uniformity and printing defects?

• Which interactions with other material and process parameters can be observed?

An experimental investigation was undertaken using a printability tester as well as an industrial

printing press to establish whether the industrial success claims of their recent panaceas, i.e. "flat-

top" dots and microcell patterning, are holding true and what the ink transfer mechanisms behind

the solutions are. This will place the technology on a firmer scientific footing which in return will

underpin the implementation of improved parameters for the manufacture and use of printing

plates. The ultimate goal is the contribution to a competitive advantage of flexography over other

printing processes and the possibility to advance beyond limitations in graphic and functional device

printing.

T h e s is La y o u t

Following this introductory background, Chapter 2 provides a review of the literature relevant to this

work. The main part of the review is the printing plate as parameter for ink transfer in the

flexographic printing process with special emphasis on different geometries of halftone dots and

surface texturing. From this the guiding hypotheses for the research presented in this volume were

derived. The experimental and analytical procedures employed to achieve the research goals are

described in Chapter 3. This includes several newly developed methods for the characterisation of

printing plates and ink depositions on prints.

I 32
The main chapters are thematically divided into the investigation of dot geometry and surface

patterning. Chapter 4 is dedicated to the results of a print trial studying how dot geometries

produced by different imaging technologies affect ink transfer on an industrial printing press.

Chapter 5 details the findings of print trials on a printability tester and industrial press researching

in-house surface patterning of solid printing areas and whether they have any effect on print quality.

The two topics are synthesised and explored in more detail in Chapter 6 where a particular type of

commercial microcell patterning is scrutinised in conjunction with standard digital and "flat-top"

imaging technology. For this purpose halftone and solid prints were produced in print trials on a

printability tester and industrial printing press. The work is concluded in Chapter 7 with an appraisal

of the research findings with regard to the hypotheses laid out earlier. The chapter contains

recommendations for better industrial practice in the application of the various imaging

technologies and surface patterning. Lastly, suggestions are made for further research which

emerged from this work.

| 33
Chapter 2 L ite ra tu re Review
The purpose of this work was to study the effect of plate characteristics on ink transfer in

flexographic printing. The emphasis was placed on novel plate-making technologies which alter the

topography of the plate surface. As these cannot be scrutinised in isolation, other material and

process parameters were also taken into account in the experimental investigations. The impact of

these factors was determined by means of print quality measures.

The aim of this chapter is to review previous work that has been carried out on ink transfer in

flexographic printing. For this purpose the chapter is divided into three parts. The first section

examines wettability and ink transfer mechanisms in general, as well as the effect of material and

process parameters relevant to this research. The second section focusses on dot top geometry and

surface texturing of the printing plate, as these constitute the main focus of this work. The final

section is dedicated to print characterisation, notably halftone models and typical flexographic

printing defects. Historically, there are extensive literature reviews on several of these topics.

However, only a reference to the most relevant reviews and articles together with a critical summary

thereof will be provided in the respective sections.

2.1 Ink Transfer

Literature reviews on ink transfer mechanisms and process parameters affecting it have been

conducted by S. Hamblyn (2004) and Beynon (2007). The studies featured findings obtained

principally by experiments employing model setups or (for sources published until approximately

1992) the letterpress printing process, the predecessor of flexography that uses rigid printing plates.

The validity of these review findings for the flexographic printing process with photopolymer

printing plates was later confirmed in trials under laboratory as well as industrial conditions for

many different combinations of printing inks and substrates.

2.1.1 Wettability and Ink Transfer Mechanisms

The fundamental principle of most printing processes is the transport of an ink through the roller

train of the printing press onto the substrate. A successful process outcome requires that the ink

wets the respective surfaces, while allowing the transfer of ink through the roller train. Traditionally,

the wettability is quantified through the equilibrium contact angle, 0, at the triple contact line

formed by a liquid, a solid and the surrounding equilibrium vapour phase (Figure 2.1) (de Gennes,

1985). The shape of the liquid drop, the triple contact line and thereby the contact angle are

dependent on several factors. In the ideal case of a "small, axisymmetric sessile drop on a flat,

|3 4
horizontal, smooth, homogeneous, isotropic, and rigid" (Shanahan, 1995) solid, the contact angle

can be directly related to the surface energy o f the solid, as, the surface tension of the liquid, ah and

th e interfacial tension between solid and liquid phase, ash through Young's equation (Equation 2.1).

W e ttin g takes place if the surface energy o f the solid is larger than or equal to the surface tension of

th e liquid. (Young, 1805; de Gennes, 1985)

Gsl = Gs — Oi • COSO Equation 2.1

<71

Figure 2.1: Contact angle 0 at triple contact line of liquid, solid and vapour phase

Incom plete ink transfer between w etted surfaces relies on a film splitting process, when the physical

bounding surfaces are drawn apart. H iibner (1991) defined tw o different classes of ink splitting

based on the fluid volum e involved. The 1st order process is a lamella separation and refers to the

splitting in a divergent film section, as can be found in the exit region o f a roller pair, and this is

appropriate fo r solid printing patches. The 2nd order process is a point separation during which a

localised ink volum e is split, as is the case fo r halftone dots.

The fundam ental ink splitting mechanisms are identical fo r both classes. At high printing speeds the

ink splitting is preceded by the form ation of cavitation bubbles in the liquid film (Figure 2.2),

facilitated by the occurrence o f negative pressures tow ards the exit o f the roller nip which exert

tension on the liquid. The critical speed fo r the onset o f cavitation is dependent on the geom etry of

the roller system, the respective roller velocities and the ink viscosity. Examples fo r cavitation nuclei

are contam inations, volatile compounds in the ink, air held in surface voids and entrained at the nip

entrance. The cavitation bubbles grow and create liquid filam ents when coalescing. The filam ents

are elongated by the diverging rollers and eventually rupture. (Banks and M ill, 1954; M ille r and

Myers, 1958; Taylor and Zettlem oyer, 1958; Myers, M iller and Zettlem oyer, 1959; De Grace and

Mangin, 1987; VoG, 2002; Barrow e ta l., 2003; Naito e ta l., 2004; Vlachopoulos, 2009; Deganello

e ta l., 2011).

I 35
 M A X IM U M
PRESSURE

•o

Figure 2.2: Schematic of the pressure distribution in the printing nip and film splitting
by the mechanisms of cavitation and filam entation (Zettlem oyer and M yers, i9 6 0 )

Numerous attem pts at m odelling the location o f the ink split relative to the tw o encompassing

surfaces have been made. None o f the models developed was universally valid fo r all printing

processes, substrate and ink com binations (W alker and Fetsko, 1955; Zaleski, Schaeffer and

Zettlem oyer, 1971; Zang, 1992), possibly because each model was only based on data from selected

ink and substrate types investigated on test printing presses. However, good agreement was

achieved on the effects o f differen t process and material parameters on ink transfer, thus

introducing high level process guidelines based on a scientific background. Following this early work,

a num ber o f studies have been undertaken to explore particular parameters in more detail.

2.1.2 Effect of M aterial Param eters on In k Transfer

The three materials participating in the final ink transfer step on the printing press are the printing

plate, ink and substrate. Key properties o f these materials, notably plate geom etry, elasticity, surface

energy and roughness, ink viscosity, substrate roughness and porosity, w ill be discussed in this

section.

2.1.2.1 P rinting Plate

The printing plate is a core process com ponent and, beside the image carried, defined by chemical,

m orphological, mechanical and topographical properties. Furtherm ore, beyond the conditions

encountered during plate-m aking th a t have a large impact on the m anifestation o f these properties

(Liu, G uthrie and Bryant, 2002; Galton, 2003; S. Hamblyn, 2004; Harri and Czichon, 2006; Harri, 2009;

I 36
Andersson, Johnson and Jarnstrom, 2009; Johnson e ta l., 2009; Tomasegovic, Mahovic Poljacek and

Cigula, 2013a), the dynamics in the printing nip can also alter the plate characteristics, tem porarily

or perm anently (M irle, 1989; S. Hamblyn, 2004; Olsson et al., 2006; Beynon, 2007). W hile the effects

o f some plate characteristics are well understood, others remain ambiguous or have not been

investigated sufficiently as the follow ing overview w ill show. A comprehensive literature review on

the influence of plate-making parameters on plate properties has been conducted by S. Hamblyn

(2004) and is om itted here.

Dot surface Dot shoulder

Impression
cylinder

Shoulder barrels to
Expansion of
form part of dot
dot surface after
surface after
compression
compression

Deformation by Deformation by
expansion barrelling

(a) (b)

Figure 2.3: Mechanisms of plate deform ation - (a) top expansion and (b) shoulder barrelling
(Bould, Claypole and Bohan, 2004a)

Figure 2.4: Folding-over of m odel halftone dot under axial load (Gotsick, 2014)

I 37
In tonal coverage, typical measurands of the dot geometry are height, shoulder angle and the size of

the printing area. Bould (2001) used numerical modelling to identify two mechanisms by which the

printing area and overall dot shape are altered in the printing nip: expansion of the dot top and

barrelling of the dot shoulder (Figure 2.3). The increase of contact area between plate and substrate

relates directly to the degree of dot deformation. The relative top expansion is almost constant for

all dot sizes, but decreases once dot surfaces start to join together at higher area coverage. The

smaller the dot size, the larger is the relative contribution of shoulder barrelling to the deformation.

Deeper dots with steeper shoulders experience less barrelling. (Bould, Claypole and Bohan, 2004a)

This leads to less increase in the contact area of the dot and thereby the dot gain. On the other hand

these deeper, steeper dots tend to have a narrower base which renders them more susceptible to

deformation potentially resulting in more dot gain (Liu, Guthrie and Bryant, 2002). Gotsick (2014)

confirmed the dual mechanism of the shoulder angle on macro-models of dots and identified the

critical angle for which the dot folds over under axial load (Figure 2.4). This dot failure changes the

shape of printed dots from round to oblong and causes significant dot gain. It is therefore important

to control the shoulder angle in the polymerisation process by adjusting the exposure parameters

(Liu, Guthrie and Bryant, 2002). Conversely, Johnson et al. (2009) observed no influence of shoulder

angle on ink transfer in their experiments, but suggested that the dual effects cancelled each other

out. Gaining further understanding of the mechanisms causing dot gain is a major driver of the

research presented in this work.

The structural response of the plate material to mechanical compression is often generically

attributed to the property of plate "hardness" which on closer inspection comprises several

individual material properties, notably indentation hardness (Shore A), Young's modulus and

Poisson's ratio. The polymers used in flexographic plates are nearly incompressible, denoted by their

Poisson's ratio being close to 0.5, and have a low Young's modulus. Bould, Claypole and Bohan

(2004a) determined that the expansion of the dot top is governed by Poisson's ratio and the

shoulder barrelling by Young's modulus. Because the plate is a polymer, Young's modulus is a

dynamic property which was found to increase under high nip pressure or through multiple

compressions of the printing plate, thus the material exhibits "stress-hardening" (Mirle, 1989; Bould,

Claypole and Bohan, 2004a; S. Hamblyn, 2004). This mechanical property impedes deformation and

thus reduces tone gain (Meyer, Durholz und Butterich, 1996; Liu, Guthrie and Bryant, 2004;

S. Hamblyn, 2004; Johnson et al., 2009). At the same time, stiffer solid printing patches conform less

well against substrate unevenness, thereby reducing ink transfer and consequently optical density

(Johnson et al., 2003).

| 38
Some studies concluded that the deformation of more elastic printing plates, i.e. those having a

lower Young's modulus, takes place primarily within the bulk of the material rather than in the

individual printing feature, thereby lowering the contact force on ink and substrate.The thicker the

printing plate, the more compression it can tolerate within the bulk. (Mirle 1989; Bould et al., 2004b;

Holmvall and Uesaka, 2007) However, other studies saw no effect of plate hardness on ink transfer

parameters investigated (Quinn et al., 1997; Johnson et al., 2009). Quinn and colleagues do not

specify their hardness parameter, whereas Johnson and colleagues used the indentation hardness.

Although the indentation hardness is related to Young's modulus, it might be insufficient to describe

the dynamic structural response of printing plates. Furthermore, the plate hardness has to be

viewed in conjunction with the mounting tape, as this alters the elasticity of the whole system and

thus the apparent hardness of the printing plate (Meyer, Durholz und Butterich, 1996; Holmvall and

Uesaka, 2007).

Overall, the plate deformation accounts for only a fraction of the print dot gain, whereas ink

spreading constitutes the major cause (Bould, 2001). S. Hamblyn (2004) determined that shallower

shoulder angles increase the ink-carrying capacity of the dot, and the larger ink volume held leads to

increased tone gain by ink spreading. He also emphasised that if the dot top is smaller than the

opening of the anilox cell, dot dipping occurs and increases dot gain by additional ink transfer from

the dot shoulders.

The effect of the plate surface energy on ink transfer is still not fully established. This decreases the

accuracy of models describing the ink distribution on and ink transfer from the plates. It also results

in the pairing of non-compatible plate materials and inks leading to wettability problems, lack of ink

transfer and print quality; a particular issue for uncommon non-standard inks such as used for

functional printing or special graphical applications. Mirle (1989) and Hejduk (2010) transferred

larger ink volumes by respectively increasing the hydrophilic component and overall surface energy

of the plate. Pluhar (2004) found that the best ink transfer occurs from plate materials with a lower

surface energy, although he did not specify which quality criteria of ink transfer this referred to.

Other studies saw no effect of the printing plate surface energy on ink transfer (Quinn et al., 1997;

Liu, Guthrie and Bryant, 2004; Johnson e ta l., 2009). On the one hand, the lack of any effects is

attributed to incomplete ink transfer and ink remaining on the printing plate. Thus, during

consecutive plate inking steps an ink-ink rather than a plate-ink contact takes place and the plate

surface energy is rendered inconsequential (Liu and Guthrie, 2003). On the other hand the contact

angle as a measure of the surface energy is a dynamic property and influenced by the process

parameters, notably printing speed and pressure (Chadov and Yakhnin, 1988; Liu and Shen, 2008),

| 39
and other material properties, notably molecular orientation, chemical heterogeneity and surface

roughness (Shanahan, 1995 and 1996). The surface roughness in particular is often not sufficiently

taken into account.

Commonly, plates are imaged by digital means (refer to section 1.2.1.1) and the macro-roughness in

the form of regular surface striae (as created by the tracing of the laser during plate-imaging)

decreases surface energy and ink transfer (Johnson et al., 2009; Heijduk, 2010). Micro-roughness

(anisotropic or isotropic) of the printing plate increases the contact area with the ink, thereby raising

the adhesion force at their interface and the apparent surface energy (Mirle, 1989; Bassemir and

Krishnan, 1990; Pluhar, 2004; Johnson etal., 2009; Hejduk, 2010). Although increased micro­

roughness enhances the wettability of hydrophilic surfaces, it hampers the wettability of

hydrophobic ones. Surfaces with nano-roughness are more likely to suspend liquids over air pockets

trapped in surface cavities which increases the contact angle. (Jung and Bushan, 2006)

Jung and Bushan's findings are not unexpected, because it is known from research outside the field

of printing that contact angles on rough surfaces tend to be larger than the Young angle (Kwok, Ng

and Neumann, 2000). Wenzel (1936) described the wettability case of a liquid which completely

displaces the air in the cavities of the rough surface on which it is placed. Cassie and Baxter (1944)

expressed the case of the liquid not displacing the air, but sitting on top of the air pockets. Each of

them developed an extension to Young's equation (Equation 2.1) to accommodate the respective

contact angles measured (Whyman, Bormashenko and Stein, 2008). However, the Wenzel and

Cassie-Baxter equations are rarely applied to printing plates. A few studies investigate the principles

for offset printing plates (Rousset e ta l., 2001; Tian, Song and Jiang, 2013), but only a single study

exists for flexographic printing plates (Dedijer et al., 2012).

Dot geometry, plate thickness, hardness and surface energy can be altered significantly by

dissolution and swelling if the plate material is not chemically resistant to the solvents and ink

components it is in contact with (Shanahan, 1996; Galton, 2003; Liu and Guthrie, 2003; Pluhar,

2004). Features altered in this way are temporally less stable under compression and can be

permanently damaged. Such an incompatibility of materials becomes more likely with the

introduction of new functional inks based on different chemistries in comparison to graphic inks.

(Theopold et al., 2012)

Studies on the effect of printing plate properties on ink transfer remain inconclusive. Although good

progress has been made in the numerical modelling of plate deformation, the models usually do not

take into account interactions with the ink. The generic use of the term "hardness" for the plate

| 40
properties of indentation hardness, Young's modulus and Poisson's ratio, as well as the disregard of

th>e mounting tape as part of the entire plate system, might have led to the vastly different findings

re:garding the effect of plate hardness on print quality. The determination of surface energy remains

problematic, because of a lack of fundamental understanding in general. In the field of printing,

interaction effects of surface energy and roughness are not sufficiently taken into account despite

th<e availability of advanced approaches such as the Wenzel and Cassie-Baxter equations. All of these

areas provide scope for more research.

2.1.2.2 Ink and Substrate

The ink and substrate are the two other key process components besides the printing plate. The

lower the ink viscosity, the more likely the ink is to spread on non-porous substrates and penetrate

into substrate pores (Fetsko and Walker, 1955; Walker and Fetsko, 1955; Damroth etal., 1996).

Generally a viscosity decrease in printing inks results from shear-thinning as a consequence of high

printing speeds (Zang, 1992), but can also be the result of a rise in temperature. Olsson et al. (2007)

observed that lower viscosity ink penetrates better into porous substrate where larger ink volumes

are then immobilised. Since the ink split occurs in the mobile phase of the ink, the relative ink

transfer is increased. The low-viscosity ink also spreads and levels better, thereby decreasing print

unevenness on the smaller scale.

Different from most other authors, Beynon (2007) concluded from his experiments that the

decrease in optical density with lower viscosity inks was due to a reduction in ink pigment

concentration and ink transfer. However, the data he presents suggests an increase in ink transfer

with decreasing viscosity. He also observed a strong interaction between ink viscosity and area

coverage. Lower viscosity led to reduced ink spread on porous substrate for highlights and midtones

which was attributed to better hydraulic impression into the bulk of the substrate instead of lateral

flow. Shadow halftones saw increased ink spread which was thought to be caused by a rolling nip

contact and ink buildup in the non-printing areas of the plate. However, the supporting graphs were

presented in such a way that it was difficult for the reader to draw the same (or any) conclusion

regarding the relationship between ink viscosity and area coverage.

The contact area between ink and substrate, and thereby ink transfer, is influenced by substrate

roughness. Smooth materials facilitate an increased contact area and enhance ink transfer. (Fetsko

and Walker, 1955; Walker and Fetsko, 1955; De Grace and Mangin, 1983; Laksin and Parris, 1997;

Holmvall eta l., 2011) Surface voids created by the roughness of non-porous substrates can

potentially hold more ink, but the ink transfer to these substrates is hampered by air held in the

|4 1
voids which is d iffic u lt to displace in the printing nip. Paper porosity and the pressure gradient

between ink and the air in the capillaries govern the ink absorption into porous substrates. Pores of

larger volum e can take up more ink, but are prone to release it again once tension is exerted on it.

(De Grace and Mangin, 1983; Zang, 1992)

2.1.3 Effect of Process Param eters on In k Transfer

There are many parameters th a t have the potential to affect print quality delivered by flexography

(Figure 2.5). However w ith in the scope o f this research, the process parameters of printing pressure

and speed, anilox volum e and geom etry as well as am bient tem perature are discussed in this

section.

Pre-press & Mounting Press Anilox

P re-prt‘« M oun ting Configuration >Type Pressure (<iap) Pressure ((rap)
Halftoning Vertical ity Machine mechanical precision Run length Ruling
Screen ruling Hori/ontality Concentricity Web tension Volume
Layout T wist Diameter uniformity Register control Depth-to-open ratio
Bouncing I i \ lit ion type Parallelism Ambient conditions Cells shape I (ieometry
Deflection ( iroup* alignment • Register Temperature Humidity Cells orientation
Colour separation Speed Vibrations Mechanical wear
Cleaning
Substrate (paper) Plates Mounting Tape Doctor Blade Material
Roughness ■Topography ( (imposition ( '(imposition Configuration Porosity
Absorbance Structure Thickness Angle! s) R i Highness
Cupillanf) Thickness Thickness variation Material Surface tension
Porosity Hardness Resilience Ink release
Compressibility Resilience Temperature stability
Surface energy (dif components! Roughness Adhesion Ink
Chemical uniformity Hacking “Baking Type / Drying process
pll Viscosity
T cmperaturc Pigment charge
Moisture content pH
Optical Properties Drying rate
Whiteness Tack build-up
Colour Rhctdogy
Fluorescence Lightfastness
Set-ofT Rub resistance
Age

Figure 2.5: Flexographic process variables according to Barros (2004)

The printing pressure (between printing plate and substrate) affects the roughness o f porous

substrates, as larger forces are able to compress the fibre netw ork and even fibre alignm ent which

enlarges the contact area between the substrate and ink. Furtherm ore, increased pressure leads to

lateral ink spreading on the surface and, in the case o f porous substrates, into the bulk by capillary

sorption. (Fetsko and W alker, 1955; W alker and Fetsko, 1955; De Grace and Mangin, 1983; Laksin

and Parris, 1997; Johnson et al., 2003; S. Hamblyn, 2004; Beynon, 2007; Bould et al., 2011)

I 42
Many studies observed an increase in print optical density with printing pressure in the lower

pressure range. In the higher pressure range the effect is less pronounced and eventually levels off

(Mirle, 1989; Johnson e tal., 2003; S. Hamblyn, 2004; Holmvall and Uesaka, 2008a). There are a few

exceptions to these findings. Bohan et al. (2003) deemed the effect of printing pressure insignificant

in comparison to pressure effects at other locations in the printing press. Increased pressure

between the ink chamber and anilox roll was found to improve ink transfer. This was attributed to

the load on the doctor blade changing the doctoring angle and the deflection thereby reducing the

wiping action of the doctor blade against the anilox cylinder. No explanation was provided for the

reduction of ink transfer observed with increased pressure between anilox roll and printing plate.

Olsson et al. (2006) report the highest optical densities at intermediate printing pressure. However,

they state that the optical density measured does not correspond to the amount of ink transferred

as determined by atomic absorption spectrophotometry. They provide no further clarification.

An increase in anilox volume leads to a rise in optical density attributed to an increased total ink

supply for transfer (Damroth et al., 1996; Lindholm et al., 1996; Fouche and Blayo, 2001; S. Hamblyn,

2004, Beynon, 2007; Bould et al., 2011). The ink transfer between the anilox roll and printing plate is

dependent on the geometry of the anilox cells (S. Hamblyn, 2004; Cherry, 2007; Bould etal., 2011).

Smaller cells provide more resistance to the extraction of ink. In general, higher printing speeds

reduce the filling of the anilox cells from the reservoir and their emptying to the plate. It was

thought that speed also affects the amount of ink removed from the anilox cell by doctor blade

action, the so-called "Kunz effect" (Kunz, 1975). Analogous observations were made by Davies and

Claypole (2004) for the rotogravure process.

Furthermore, higher printing speeds decrease the nip dwelling time (typically milliseconds). As this is

associated with the impression of ink into porous substrates, the ink transfer is also reduced. (Fetsko

and Walker, 1955; Walker and Fetsko, 1955; De Grace and Mangin, 1983; Zang, 1992; Damroth

et al., 1996; Fouche and Blayo, 2001; Johnson et al., 2003; S. Hamblyn, 2004). In contrast Quinn et al.

(1997) and Olsson et al. (2007) observed an increase in ink transfer with printing speed. The latter

postulated that a shorter nip dwelling time caused less substrate compression which allows pores to

remain open and take up more ink. However, the author of this work suggests instead an interaction

with the construction of the laboratory testers used in these two studies.

The effect of printing pressure, speed and anilox volume on ink transfer from halftones on

conventional and standard digital printing plates under industrial conditions followed the same

trends as for solids, but showed interactions with area coverage of the halftones (Johnson etal.,

I 43
2Q03; Hamblyn, 2004; Beynon, 2007; Holmvall and Uesaka, 2008b; Borbely and Szentgyorgyvolgyi,

2Q11).

Temperature affects ink transfer in two ways which compete with each other. On the one hand, it

lovwers ink viscosity and thereby facilitates ink spread and absorption. On the other hand, higher

temperatures promote faster ink-setting, so that the immobilised ink is anchored in place. This may

impact print quality, e.g. by reducing print gloss through faster and less favourable setting of ink

components. (Olsson et al., 2007)

Despite extensive research into flexographic process parameters, a number of factors remain

unclear regarding their effects on ink transfer. The varying effects of, for example, printing pressure

and speed suggest that there might be interactions with other process and material parameters

which have not yet been considered or sufficiently explored. The constant introduction of new

materials to the printing process does not simplify the matter. This provides numerous starting

po ints for future research.

2.2 Flexographic Printing Plates

This section is dedicated to the plate parameters which constitute the main focus of this work,

namely the geometry of the top of halftone dots and texturing of printing surfaces.

2.2.1 Geometry of Dot Tops

As overviewed in section 1.2.1.2, plate-making technology now has the capability to facilitate the

production of different dot top shapes, notably round or flat. Conventional analogue plates have

printing features with mostly level tops and a sharp edge to the dot shoulder. The introduction of

standard digital printing plates brought dome-shaped dots without obvious distinction between dot

top and shoulder (Figure 2.6). These are a product of oxygen inhibition during the polymerisation

process. New digital imaging technology replicates conventional plate-making in that it prevents

oxygen inhibition and therefore leads to the creation of level printing features. (Anderson and

Schlotthauer, 2010) The method is referred to as "flat-top imaging technology" (Shibanov, 2012).

The differentiation of "flat" and "round" dot geometries is generally accepted, as this is perceived

from microphotographs such as in Figure 2.6. However, microphotographs taken perpendicular to

the dot top indicate the existence of concave geometries (Figure 2.7). Only three publications could

be identified which refer to concave dot tops. Hornschuh (2005) attributes the "doughnut" printing

defect (refer to 2.3.2.1) to erroneous plate-making which creates printing features with top edges

higher than the centre. Claypole et al. (2008) observed ridges atop the edges of printing fields on

| 44
conventional and standard digital plates. Sievers (2011) described how all three types o f top

geom etry (convex, fla t and concave) can be achieved on the same standard digital plate material by

varying the intensity o f light exposure.

No publications were found which quantify the actual top geom etry and investigate the resultant

p rint quality. In general very little research into the printing behaviour o f features w ith d iffe re n t dot

tops is available. All scientific publications identified which directly compare the effects o f top

geom etry are addressed below.

1 Conventional digital plate showing a 7% dot at toalpl a Ftat-topped dot plate showing a a% dot at tojlpi

Figure 2.6: Comparison of microphotographs of standard digital (left) to "flat-topped" (right) plate features
and th e ir resultant print quality (inset pictures) (Charlesworth, 2008)

Figure 2.7: M icrophotographs of halftone dots suggesting concave dot geom etry on
(a) conventional (Harri and Czichon, 2006) and (b) standard digital plate (Johnson et al., 2009)

2.2.1.1 Effect o f Top Geometry on In k Transfer

During the imaging of the design onto the printing plate, conventional plates experience an increase

in feature size, i.e. dot diam eter or line w idth, compared to the nominal values, whereas features on

standard digital plates decrease. On "fla t-to p p e d " plates almost com plete fid e lity is achieved (DFTA,

2002; Beynon, 2007; Yusof et al., 2007; Claypole et al., 2008; Valdec, Zjakic and M ilkovic, 2013). The

negative plate gain on standard digital plates is usually addressed by increasing the feature size using

correction curves (M ahovic Poljacek et al., 2013; Tomasegovic, M ahovic Poljacek and Cigula, 2013b).

I 45
For the ink transfer from the plate to the substrate, little difference in relative dot gain was seen

between conventional and standard digital plates (Claypole etal., 2008; Gilbert and Lee, 2008). In

absolute terms conventional plates performed better at low printing engagement and line ruling

where feature sizes were closer to target. Standard digital plates without correction excelled at high

engagement and line ruling, because they compensated the effects of increased deformation and ink

spread with negative plate gain. (Yusof et al., 2007) Corrected digital plates caused significant dot

gain which was thought to result from identical printing engagement leading to more relative dot

deformation and the lack of shoulder sharpness facilitating more ink spread (DFTA, 2002).

Studies attributing standard digital plates with higher dot circularity on the plate, improved print

contrast and reproducible tonal range, but less print uniformity have to be treated with caution.

Beynon (2007) concluded that halftone dots on conventional plates have a less circular dot shape

due to a lack of intimate contact between the imaging film and plate allowing light leakage. This is

disputable, because light leakage and scattering have the potential to improve non-circularity caused

by irregularities in the film mask. Furthermore, the effect of oxygen inhibition on the circumference

of standard digital halftones was completely ignored. Galton (2005a) investigated the effect of

imaging technology on the uniformity of halftone prints using the approach of Design of

Experiments. He attempted to calculate the effects of seven factors using only eight experiments

which resulted in a position whereby the individual effects could not be resolved from their

interactions. Therefore, the finding that conventional plates result in better halftone print uniformity

remains unclear. Gilbert and Lee (2008) compared prints from conventional and standard digital

plates. However, their prints were produced under different printing conditions and by several press

operators casting doubt about the consistency of their experimental approach. Furthermore, the

authors' results lack any critical explanation accounting for cause and effect.

Only one article compared the effect of standard versus "flat-topped" digital plates in a rigorous way

(Valdec, Zjakic and Milkovic, 2013). The dots on the standard digital plate were very sensitive to

changes in printing pressure and produced higher dot gain. Numerical modelling of the dots without

ink showed that the flatter the dot, the more uniform was the pressure distribution across the

printing surface and within the bulk of the material, and the less shoulder barrelling occurred.

However, a significant stress concentration took place at the sharp transition of top to shoulder

which indicated that "flat-topped" dots were unable to deform as continuously as round ones.

(Holmvall and Uesaka, 2008b; Anderson and Schlotthauer, 2010) The exploration of ink transfer in

flexography by numerical modelling remains incomplete to date and the approach to doing this can

be better informed by experimental evidence typical of that which will be gathered in this work.

| 46
2.2.1.2 Related Publications

In the absence o f research on the ink behaviour on varying dot top geometries, inform ation can be

obtained from related fields of investigation. Analogous to the behaviour of an ink drop on a

halftone dot under compression, a simple model fo r the advancement o f a liquid across a sharp edge

o f a conical frustum was developed by Gibbs (1873). The contact line of a growing liquid drop moves

across a flat surface at an advancing contact angle, 0ao (Figure 2.8a). W ith fu rth e r enlargem ent the

contact line w ill be pinned to the sharp edge (of angle (p), while the contact angle relative to the

frustum increases (Figure 2.8b). If the contact angle reaches a critical value 6a (Figure 2.8c), the drop

collapses down the side o f the cone (Figure 2.8d). The critical contact angle was calculated by Gibbs

as a purely geom etrical extension to the contact angle in Young's equation (Equation 2.1):

6a = 9a o + (1 8 0 ° - (p) Equation 2.2

a.o

A
Figure 2.8: Mechanisms of liquid pinning to sharp edge of Figure 2.9: Apparent liquid pinning to im perfect sharp
conical frustum - (a) drop inflation, (b) contact line edge (Oliver, Huh and M ason, 1977)
advancem ent atop frustum , (c) liquid pinning to frustum
edge and (d) drop collapse (Extrand and M oon, 2008)

The "Gibbs inequality" has been confirm ed experim entally (Oliver, Fluh and Mason, 1977; Extrand

and M oon, 2008; Chang et al., 2010; Toth et al., 2011). Oliver et al. specified th a t no edge could ever

be perfectly sharp (Figure 2.9). Therefore, the contact line advances along the radius o f the edge

whenever the condition o f the advancing contact angle 6ao is fulfilled. For any angle smaller than

6ao the contact line remains pinned in position. If the drop can m aintain equilibrium , its

advancement is continued on the new surface under the same conditions. In the dynamic case the

critical angle 6a w ill be exceeded and the liquid collapses down the sides. The contact line pinning

I 47
takes place in the case o f invading as w ell as retreating liquids. The form er is governed by the

advancing, the la tte r by the receding contact angle. (Extrand and M oon, 2008; Chang e ta l., 2010)

This is significant fo r the ink transfer process to and from halftones of any geom etry as well as for

the continuous buildup o f ink on the d ot shoulders. For example, more ink buildup w ould be

expected fo r halftones w ith round to p geom etry and shallower shoulders, because the critical

contact angle to overcome pinning can be achieved m ore easily under compression.

Gibbs' model was also found to be valid fo r piilars w ith triangular or square base, although the

critical contact angle was form ed at d iffe re n t tim es around the circum ference o f the non-circular

edge. It was firs t achieved along the sides o f the edge and last at its corners. The liquid could spread

down the sides o f the pillar at angles higher than the critical contact angle w ith o u t collapsing the

drop, because the entire volum e was still suspended from the corners of the edge by pinning.

However, circular pillars held more liquid (normalised to the surface area) than square and triangular

ones at the instant o f collapse. (Toth et al., 2011) Furtherm ore, the suspended volum e was not only

dependent on the prim ary shape of the feature (side angles and edge geom etry), but also on any

secondary asperities present (Figure 2.10). Secondary texturing was found to be hierarchically

dom inant in the determ ination o f w e ttin g properties and im proved the feature's ability to suspend a

liquid drop. (Extrand, 2005) These tw o observations are of relevance for halftones dots w ith irregular

circum ference, such as those imaged w ith SQUAREspot laser technology which suffer from jagged

edges, and dots w ith high surface roughness. It implies th a t the form er have low er and the latter

higher ink-carrying capacities.

a u

Figure 2.10: Secondary asperities on the sides of a conical frustum (Extrand, 2005)

2.2.2 Textu rin g of Printing Surfaces

Surface texturing on the printing plate falls into one o f tw o categories: increased m icro- or nano­

roughness, and patterning. Roughness is a plate p roperty and its effect on ink transfer has already

been discussed above. Popular commercial im plem entations of patterning are MicroCells

(EskoArtworks, Ghent, Belgium) and DigiCap (Eastman Kodak). They have in comm on th a t they are

| 48
 made up of small recesses varying in size and pitch in the surface of the printing area. From here on

they will be referred to as "microcell patterning". The similarity in appearance to the engraving of

anilox cylinders led to the assumption that the microcell functionality is analogous to the land areas

and cells on the anilox. It is claimed that microcells carry increased amounts of ink which raise ink

transfer and anchor the ink film more uniformly to the substrate (Samworth, 2001 and 2009; Kodak,

2010), but there is little scientific evidence to support this model. Patterning similar to microcells

(Weichmann, 2002; Samworth and Cadogan, 2007; Stolt, Zwadlo and Rozzi, 2010) is used to control

the ink film thickness by only transferring ink from the land areas of patterned surfaces. The pattern

recesses are not considered to be ink-carrying, but are claimed not to be visible in the final print due

to ink coalescence.

Only one publication on the investigation of microcells in an academic context was found

(MacDermid Printing Solutions, 2011). In this work a print trial was conducted into the effects of

microcells on print quality. Of 32 microcell patterns 15 achieved significantly higher optical densities

than a plain reference solid when imaged on a standard digital plate. In combination with "flat-top"

imaging technology, 22 microcell patterns outperformed the plain solid. (Cook, Recchia and Gotsick,

2014) No explanation was provided why some patterns performed better than others.

Very few other relevant publications on liquid transfer from patterned surfaces are available.

Griesheimer (2013) investigated the ink transfer mechanism (filamentation in particular) from

< patterned surfaces using a model setup. His "patterning" consisted of halftone dots at a screen

ruling of 48 and 60 lines/cm and varying area coverage. This choice of pattern is not representative

of commercial surface patterning and holds more relevance for ink transfer from halftone patches.

Other studies are limited to quasi-static liquid transfer between individual features rather than

entire pattern assemblies (Gupta et al., 2007; Huang et al., 2008). Nevertheless, they emphasise the

significance of different feature geometries and dimensions on liquid transfer, for instance pillar

structures releasing liquid more easily than cavities, and shallower cavities improving relative liquid

transfer.

Stolt and colleagues (2010) plainly admitted that knowledge about the underlying science of surface

patterning is lacking, which has not changed to date. The only academic investigation of microcells

| might be biased, since the motivation for the investigation originated from the company which

commercially distributes microcell patterning. In the light of increasing popularity of surface

II.
; patterning in the printing industry, specific studies of the ink transfer from patterned surfaces are
i
| required to gain an understanding of any underlying mechanisms that are present and hence to

! optimise the printing process.

| 49
2. 3 Print Characterisation
Thie first part of this section introduces halftone models which serve the purpose of relating print

optical density to the area coverage of the printed ink film. This approach is imperative in industry

w here the determination of area coverage by other techniques is not viable. The second part of the

section is concerned with typical flexographic printing defects, namely doughnuts and halos,

uncovered areas and fingering instabilities.

2.3.1 Introduction to Halftone Models

The tool most widely used to monitor the consistency of industrial printing processes is the

comparison of the area coverage of halftone control patches against target values. The area

coverage is calculated from the reflectance factor measured for the respective patch using halftone

models (mathematical equations expressing the relationship between parameters). This has the

advantage of speed and simplicity over optical methods. A comprehensive review of halftone

models was provided by Wyble and Berns (2000). Only mono-chrome approaches which are used as

part of this research are introduced below.

The Murray-Davies model established itself as the standard method due to its straightforwardness

and is implemented in the majority of spectrophotometers:

P H a lf tone, M urray-Davies ~ (1 — ^ % ) ‘ Psubstrate “h ' Psoiid Equation 2.3

where A % is the area coverage, Psubstrate, Psoiid and P Haiftone are the reflectance factors of the

substrate, reference solid and halftone area to be determined respectively. The disadvantage of the

Murray-Davies equation are the model assumptions that all printed areas are characterised by sharp

edges (i.e. hard dots) and uniform ink film thickness (Figure 2.11) as well as that no light scattering is

taking place in the substrate (Murray, 1936). Changes in reflectance are purely a result of variation in

area coverage. In practice neither assumption holds true and leads to the calculation of an area

coverage different from the optically effective one.

The Yule-Neilson equation (Equation 2.4) accounts for light scattering and dot softness by

introducing an additional parameter, n, which is dependent on substrate type (Yule, 1943; Yule and

Neilsen, 1951). This additional parameter can only be established empirically for each individual

substrate. However, Pearson (1980) determined that the n-value for most paper substrates lies

between 1.4 and 1.8 (with a majority at 1.7).

P H a lf tone,Yule-Nielsen = [ ( 1 “ ^ % ) ’ Psubstrate + ’ Psoiid] **

I 50
 Murray-Davies Beer- Lambert- Bouguer Noffbe- Seymour
100«o

so%

10*0

Figure 2.11: Schematic of underlying assumptions (dot sharpness and ink distribution) for d ifferent halftone models.
M urray-D avies disregards ink spread and is based on uniform ink films of constant thickness and w ith sharp edges. Beer-
Lam bert-Bouguer assumes com plete ink spread to 100% area coverage and accordingly changing ink film thickness.
Noffke-Seym our considers partial ink spread to dots w ith varying ink film thickness and soft edges.

In the Expanded Murray-Davies model the effects of light scattering and dot softness are included by

the separate parameters w and v (Arney, Engeldrum and Zeng, 1995), but the equations are

arranged in a way th a t the parameters are interchangeable and no inform ation on the underlying

physics o f the process can be deduced. The Expanded Murray-Davies equation (Equation 2.5) has the

same form as the Murray-Davies equation (Equation 2.3) w ith the exception th a t the substrate

reflectance,/?s, and the reflectance of the solid reference,/?;, become functions of area coverage

and have to be determ ined by Equation 2.6 to Equation 2.8.

(1 — A 0/o) • p s { A % ) + A 0/o ■P i{A o /o} Equation 2.5

P H a l f tone, E x p a n d e d M u r r a y - D a v i e s

P i { A % ] — P s u b s tr a te ’ U - ( 1 “ Tl n k ) ' ^ % ] [ 1 (1 — T/nfc) • Ao/0] Equation 2.6

Equation 2.7
P s{A % ) - P s u b s tra te * [1 T/n f c ) ( l “ (1 ~ ^ % ) W)]

• [ i - ( i - r /nfc) d - ( i - z i o /0r ) ]

Equation 2.8
_ / Psoiid \ 1/2
'P s u b s tra te '

where zlnk is the ink transm ittance o f the reference solid (Equation 2.8).

A continuous tone model based on the Beer-Lambert-Bouguer law (Equation 2.10) was suggested by

Seymour and Noffke (2012; Seymour, 2013a). It is based on the supposition th a t the printed ink

volum e is conserved, but spread out over the substrate until it form s a uniform solid ink film (Figure

I 51
2.11). The optical density then becomes solely dependent on ink film thickness. In its assumption of

perfect ink spread the Beer model represents the other extreme to the Murray-Davies model which

assumes no ink spread whatsoever.

Psoiid \ A% n Equation 2.9

P H a lf tone, Beer ~ ( n 1 ’ Psubstrate
hcfr/i /
^PsubstrateJ

For a more accurate representation of the actual limited ink spread in a soft dot (Figure 2.11);

Seymour and Noffke (2012) combined the two models to form the Dot Spread equation (Equation

2.10), now known as the Noffke-Seymour equation (Seymour, 2013b).

Phlalf tone,Noffke-Seym our = (1 — a 2^ ’ Psubstrate Equation 2.10

( Psoiid \ a^ a2 o
' a2 ’ I « J ’ Psubstrate
' Psubstrate'

where c^and a 2 are the relative area of the halftone dot on the printing form and spread out in the

print respectively. The equation cannot currently be solved in closed form. Neither the Beer nor the

Noffke-Seymour equations take into account the optical dot gain.

There is currently no universal halftone model which predicts accurately the printed area coverage

based on optical density. Only empirical studies reveal which model provides the best fit to data

from different printing processes, substrates, halftone dot shapes, line rulings and so on. It is not

known how innovations such as "flat-top" imaging technology and surface texturing of the printing

plate in flexography change the suitability of the traditional halftone models. This is a topic of

academic interest as well as industrial significance which merits exploration.

2.3.2 Printing Defects

Printers differentiate between many defects impacting print quality (Mathes, 2011 and 2012), but

only a few of them are directly relevant to this research and addressed below.

2.3.2.1 Doughnuts and Halos

"Doughnuts" are printed halftone dots which assume a ring-shape due to lack of ink in their centre

(Figure 2.12 left). It is generally assumed to be caused by the ink being displaced outward and away

from the top of the printing feature during substrate contact, a hypothesis supported by the fact

that doughnuts are more prominent at high impression pressure (Mathes, 2012). Hornschuh (2005)

I 52
suggested th a t the "D onald-Effekt" (sic! phonetic error) at the start of a print run is the result of only

th e edges o f concave dot structures printing. W ith increasing run length the centre of the dot also

transfers ink and the doughnut effect decreases. Industry does not necessarily differentiate

doughnuts from halos (Mathes, 2012).

S. Ham blyn (2004) used concave dot tops to explain a printing defect o f alternative form . "Halos1"

are ink-free annuli found in printed dots (Figure 2.12 right). Although he observed only convex dot

to p s on his printing plates, S. Hamblyn hypothesised th a t during contact between the printing plate

and substrate a concave dot shape is created. The hydrodynam ic squeeze film effect in the ink

betw een printing plate and substrate raises the pressure at the dot centre above the am bient

pressure at the dot edge and causes the dot centre to cave in on itself (Figure 2.13). The concave dot

centre seals o ff ink w ith in the cup, and the ink underneath the cup brim is squeezed out leaving an

ink-free area under the edge (halo). The effect was thought to be supported by the polym er in the

centre o f the dot being softer than the surrounding m aterial. This could either be an inherent

p roperty based on the imaging process, or be caused by stress-hardening o f the outer polym er layer

under loading. An indication o f the latte r was provided by halos only occurring at the higher printing

pressures investigated, but not the low er ones.

The literature research has revealed th a t no w ork has been published on the origin of halos or

potentia l interaction effects w ith dot geom etry.

w l | f
Figure 2.12: Printing defect of doughnut w ith approxim ately circular lighter area in the centre (left)
and halo w ith crescent-shaped lighter area in the dot centre (right) (Hornschuh, 2005)

1 The d e fin itio n o f "h a lo " provided is adopted fro m S. Hamblyn (2004) and em ployed in th a t sense th ro u g h o u t
this research. The industrial de fin ition o f "h a lo " as an o u tline of ink buildup along the edges o f p rinted areas
(M athes, 2012) is n ot the in te rp re ta tio n used here.

I 53
 Reduced ink volume
in contact zone
Substrate
t
Ink squeeze

(b )

Deformation of
dot centre dot shoulder

Figure 2.13: Schematic of potential cause for halo; (a) inked dot before impact, and
(b) dot deform ation during impact w ith substrate (adapted from S. Hamblyn (2004)

2 3 .2 .2 Uncovered Areas

Randomly occurring small holes in printed ink film s are referred to as "pinholes" (Mathes, 2011) or

more recently in academic circles as "U ncovered Areas" (UCAs). UCAs can be immensely im portant

in printed functional devices w here th e ir presence can lead to com plete device failure.

Barros, Fahlcrantz and Johansson (2006) saw very good correlation between flexo-printed UCAs and

the location of topographical depressions in paper and paperboard surfaces. On printing, it was

found th a t small substrate cavities could be bridged by ink, but above a critical lateral dimension or

fo r low ink volum e supplied, ink transfer failed (Naito et al., 2004; Barros and Johansson, 2006;

Holmvall e ta l., 2011). M ettanen (2010) a ttrib ute d only 20% o f UCAs to the topography of super

calendered and new sprint paper in offset and gravure printing. The remaining 80% were thought to

be caused by othe r substrate properties and process parameters. Mesic and colleagues (2006a,b)

compared the num ber o f UCAs occurring in flexo-printed ink films on plastic-coated paperboard of

diffe re n t age and degradation after corona tre a tm e n t. Since more UCAs were found on older

degraded substrates, an uneven change of substrate surface chem istry and energy was assumed to

affect UCAs, but conclusive pro o f could not be provided. Overall UCAs were a ttrib u te d to w e tta b ility

issues and surface topography in equal parts. Naito et al. (2006) observed the ink laydown process

through the reverse of a clear plastic substrate. They reported the creation of UCAs in those
locations where air bubbles entrapped between substrate and ink film on the printing plate

prevented ink transfer.

No publications were identified which investigate the role of the printing plate as a cause for UCAs,

and whether UCAs are limited to solid printing patches or could also occur in printed halftones.

2.3.2.3 Fingering Instabilities

At very low printing speeds, ink splitting occurs as a stable smooth layer separation governed by

hydrodynamic principles. Above a critical velocity, the ink-air interface at the exit of the printing nip

destabilises, and printing defects occur. (MacPhee, 1997) These fingering instabilities demonstrate

either transient or steady-state behaviour leading to different classes of printed patterns (McCloud

and Maher, 1995). One such steady-state pattern is "ribbing" which has long been studied for

different coating processes (Pearson, 1960). Ribbing is characterised by a regularly undulating ink

film thickness transverse to the print direction (Figure 2.14a). A literature review can be found in

Vlachopoulos (2009). It took until the late 1980s for Rabaud and colleagues (1991) to relate this so-

called "printer's instability" to a larger group of hydrodynamic defects first described by and named

after P. G. Saffman and G. I. Taylor (1958). In general Saffman-Taylor instabilities are created when a

less-viscous fluid (air) penetrates the interface to a more-viscous fluid (ink) forming channels in the

ink known as "fingers". The onset of fingering instabilities is governed by a critical capillary number,

Cacrit, which is dependent on the fluid properties and speed, or the system geometry:

Equation 2.11

where r] and a are the dynamic viscosity and surface tension of the ink, u is the tangential velocity of

the rollers, h0 is the nip height, r is the radius of the interface curvature at the nip, and zx and z2 are

adjustment parameters for the respective parameter conditions (Pearson, 1960; Savage, 1984;

Rabaud et al., 1991; Sauer, Bornemann and Dorsam, 2011). If the printing conditions divert from the

steady state at which ribbing occurs, the fingers become transient and branched (Figure 2.14 b,c).

Comprehensive literature reviews on Saffman-Taylor instabilities have been conducted by Saffman

(1986), McCloud and Maher (1995). The term "viscous fingering" is currently establishing itself in the

printing sector to exclusively describe branched printed patterns. "Ribbing" continues to be used for

patterns of discrete, straight fingering. This convention will be followed in this work.

| 55
 ; • •;> • • ♦ <,
•••-«•*+ **, * *•*
«-v-«
• * . V .* <

(a) (b) (c)

Figure 2.14: D ifferent m anifestations of fingering instabilities in p rin ts -

(a) ribbing, viscous fingering w ith (b) small and (c) large beads

Saffm an-Taylor instabilities have been known to occur in other printing processes, although possibly

categorised as one m anifestation o f the print defect "m o ttlin g " (Mathes, 2011). Behler (1993)

a tte m p te d to dem onstrate on the model of a lifting Hele-Shaw cell2 how viscous fingering affected

th e edge structure o f printed halftone dots. VoR (2002) used analytical and experim ental methods to

investigate Saffman-Taylor instabilities in halftone and solid flexographic prints. The experim ental

emphasis lay on the correlation o f anilox topography w ith the frequency of ribbing observed in the

solid prints. No correlation was found, because Voft failed to recognise that the final ink transfer step

betw een printing plate and substrate will determ ine the appearance o f the instabilities.

M ore recently the necessity to understand the mechanics behind Saffman-Taylor instabilities has

gained more significance in flexographic, gravure and lithographic printing of electronic devices

(Bornem ann, Sauer and Dorsam, 2010). Since the instabilities cannot be avoided under sensible

production conditions, the parameters affecting fingering have to be optim ised to achieve the best

production outcom e possible (Reuter et al., 2007; Bornemann, Sauer and Dorsam, 2011; Sauer,

Bornemann and Dorsam, 2011; Hernandez-Sosa et al., 2013). The graphics printing industry

addressed the problem by introducing surface texturing on printing plates. That anisotropic,

patterned surfaces are able to change the appearance o f fingering has been dem onstrated by

several authors using Hele-Shaw cells3 w ith textured plates (literature review provided by McCloud

and Maher, 1995). However, fundam ental studies o f the interactions between textured surfaces and

hydrodynam ic instabilities in a system setup th a t is appropriate fo r printing processes are missing to

date.

2 In general the Hele-Shaw cell consists o f tw o parallel plates containing tw o im m iscible fluids o f d iffe re n t
viscosity (Saffman, 1986). Behler (1991) used a liftin g Hele-Shaw cell w ith parallel plate separation w hich
allow ed th e air to penetrate the fo rm in g ink fila m e n t fro m th e sides.
3 The Hele-Shaw cells in question had stationary plates and th e less-viscous fluid was injected in to th e cell w ith
m ore-viscous fluid through a small hole in o f th e plates (McCloud and M aher, 1995).

I 56
2.4 Conclusions
This literature review has shown that extensive research into flexographic printing has been

conducted previously. However, due to the complexity of the process, each study can only

investigate a small piece out of the big jigsaw puzzle that is flexography with findings having limited

validity within the boundary conditions set. From the literature the following challenges have been

identified:

• More detailed understanding of the plate deformation mechanisms causing dot gain is

required as a precursor for accurate numerical modelling.

• Past numerical modelling of plate deformation did not take into account interaction effects

with the ink and, in some models, disregarded the mounting tape as part of the entire plate

system.

• The generic use of the term "hardness" for the plate properties of indentation hardness,

Young's modulus and Poisson's ratio might have led to the vastly different findings regarding

the effect of plate hardness on print quality.

• The determination of surface properties, including surface energy and roughness, remains

problematic, because of a lack of fundamental understanding in general. In the field of

printing, the interaction effects of surface energy and roughness are not sufficiently taken

into account despite the availability of advanced approaches, such as the Wenzel and Cassie-

Baxter equations.

• Despite extensive research into flexographic process parameters, no ultimate consensus has

been reached regarding the effects on ink transfer, which suggests currently unknown

interactions with other process and material parameters.

• There is no method for the quantification of halftone top geometries. Furthermore, very

little research into the printing behaviour of features with different dot tops and the

resultant print quality is available.

• There is currently no universal halftone model which predicts accurately the printed area

coverage based on optical density. It is not known how different halftone dot geometries

affect the applicability of the traditional halftone models.

• No research into the underlying science of the effect of surface patterning on print quality

has been conducted to date.

• The origin of the halo printing defect is still uncertain. The influence of the dot top geometry

on the defect is unknown.

• The role of the printing plate in the creating of the UCAs in prints is unknown.

I 57
 • The effect o f textured surfaces on the onset of hydrodynamic destabilisation o f a liquid

meniscus between rollers has not yet been explored in a printing context.

This review has identified a num ber o f challenges. However, the w ork in this thesis w ill be confined

to a subset o f these and include a detailed exploration o f tw o factors, namely dot geom etry and

surface patterning:

D o t G e o m etr y

• Are dot tops produced by standard digital technology convex; those produced by "fla t-to p "

imaging technology level? Do concave dot tops exist? (Figure 2.15)

• Do "fla t-to p p e d " halftones achieve higher print quality by transferring larger ink volumes,

causing less physical dot gain and m ore uniform ink lavdown?

• Do dots imaged w ith SQUAREspot laser technology have a less circular dot top? If so, how

does this affect the ink transfer from these dots?

• Do concave dot tops cause the printing defect o f halos?

• Is the effect of process parameters, such as printing pressure and speed, on ink transfer

independent from the geom etry o f the dot top?

• Is the applicability of halftone models independent from the geom etry o f the dot top?

convex flat concave

(a)

Halftone
dot

(b)
Substrate
Ink

co

Figure 2.15: Schematic of potential ink transfer mechanisms depending on dot geom etry -
(a) inked halftone dots w ith convex, flat or concave dot top, (b) deform ation of halftones
against substrate, and (c) resultant ink distribution on substrate

I 58
Surface Pa t t e r n in g

Does surface patterning o f the printing plate increase the ink transfer to the substrate as

quantified by the optical density of the prints? Does it reduce printing defects, notably UCAs

and fingering instabilities? (Figure 2.16)

Do microcell patterns act analogous to anilox cells and increase the ink-carrying capacity of

the printing plate?

Are the p rint results dependent on pattern design?

Is the effect of process parameters, such as printing pressure and speed, on ink transfer

independent from the geom etry o f the dot top?

Is the efficiency o f surface patterning increased when combined w ith fla t-to p imaging

technology?

Patterned plate Plain plate

Substrate

Inb

Figure 2.16: Schematic of potential ink transfer mechanisms depending on surface patterning -
(a) inked plate patches w ith and w ith o u t surface patterning, and (b) resultant ink distribution on substrate

I 59
C hapter 3 Methodology
Thiis chapter is divided into sections reflecting the process steps employed in the investigations

performed for this work: material characterisation, printing and print analysis. It was necessary to

develop innovative applications of measurement techniques, in order to gain an in-depth

understanding of the image transfer process.

M a t e r ia l Ch ar a c t e r is a t io n

The most important materials used in the print trials, namely printing plate, substrate and ink, were

characterised with regard to their respective properties of interest for the print trials and analysis. As

the main focus of this work, the printing plates received the most extensive characterisation using

whiite light interferometry (WLI). Furthermore, it is common practice within the field of printing

research to establish basic material properties such as surface roughness (plate, substrate), rheology

(ink) and surface energy/tension (for all materials), because they can have significant influence on

printability and print quality.

Pr in t in g

The print trials were performed on an IGT-F1 printability tester and under near-industrial conditions

on a Timsons T-Flex508 printing press. In order to improve the efficiency of data capturing and

analysis, the technique of Design of Experiments (DoE) was employed. This method allowed the

identification of the most important printing parameters (including materials) with regard to the

target parameters of print quality.

P r i n t A n al ysi s

The print analysis followed two different lines of inquiry. The first one saw the digitisation of the

prints by scanning. Through the use of image analysis software the acquired digital images were

investigated for print uniformity and density. The scanning methodology was refined in an extensive

test series. The second line of inquiry involved digital microscopy, spectrophotometry and surface

profilometry. Data on optical density, amount and structure of ink transferred was obtained.

3.1 Surface Profilometry

The surface profile is the key parameter explored in this work using WLI which performs contactless

optical surface profilometry. The interferometer employed was a Wyko NT2000 (Veeco Instruments,

Tuscon, USA) with the supporting software WYKO Vision32 (Version 2.303), and the working

| 60
principles may be found in Olszak, Schmit and Heaton (2001). It was used to determ ine the surface

roughness o f printing plates and substrates, the dimensions and geom etries o f printing elem ents on

the plates, as well as the planar surface area and volum e o f printed features. The WLI data sets

shown in this w ork are all false colour images obtained in the Vision 32 software (Figure 3.1). The

colour coding is such th a t red shades in the image depict raised areas, whereas recesses are shaded

blue. Surface points fo r which no data could be captured remained w hite. No data m ight result from

surface points which are too steep or very rough, and thus reflect the incident light beam in such a

manner th a t it cannot be detected by the microscope objective.

The descriptions of the measurem ent techniques are provided in the follow ing sections. The

determ inatio n o f feature geom etries and planar surface area were the most significant fo r this work.

Figure 3.1: False colour image of a chequer pattern on a printing plate

(blue pixels represent troughs, red pixels represent ridges)

3.1.1 D eterm ination of Feature Dimensions and Surface Roughness

The Vision32 softw are has integrated display and analysis functions w ith which feature dimensions

and geom etry can be determ ined. For this w ork the "2D analysis" option was used to select a cross

section through the 3-D data set and to display it as a height-w idth-diagram (Figure 3.2). W ithin the

diagram, cursors may be positioned on opposite sides of the feature under investigation and the

distance between the tw o cursors noted as feature w idth. Furtherm ore, the cursors were placed at

the bottom and top of the feature to establish its height. Either measurement was dependent on the

available content of the data set, e.g. the height o f a printing feature could only be assessed if the

data set included the relative flo or height o f the printing plate.
The surface roughness of a m aterial is o f interest, because it provides inform ation on the potential

contact area between tw o surfaces during the printing process, the surface volum e in which ink can

be carried, and tolerances fo r the determ ination of feature dimensions. The WYKO Vision32

softw are is able to calculate d ifferen t predefined roughness parameters from the surface data of

w hich tw o have been used in this work, notably Ra and Rz (Lippold and Podlesny, 1998).

Veeco X Profile
X 192 2 tm

i 2 *9 o.y m ...
CJ
F :: 7 §
u ’M .m
Rp -mo B9
R* *17.71 urn

Ct*v* -15 J mai

To th, S W

A vm H i « tO .il urn
' ’ i, • Are* 0075.53 un*3

Y Profile
X 1D61um
; y _____ ___________
r ■ t t T- 1 I 1"1 --- ------- ------- ---- I -
w <M IH m n* »?
055 tun
Rj 0.13 um
Rl 1.60 um
ftp >10 03 tun
R\ -tl M um
X 15915 - lira
V 155.39 . urn Angle -5.«M tursii
Ht - m C w vr -13.70 mai
Uf*t - wn tr im , \OOf
Anifc Av»Ht -10.76 um
T ilk Subregion M IB 1* IB ,%
> BO Aw* W unvi

Note: X offset 62 Y offset 45

Figure 3.2: Display option "2D analysis" in WYKO Vision32 softw are showing 3-D representation of a halftone dot,
2-D cross sections along tw o different axes, surface roughness and various other param eters

3.1.2 D eterm in ation of Planar Surface Area of P rinting Features

For numerous printing features it was not sufficient to determ ine th e ir dimensions in order to gain

inform atio n on the surface area involved in the printing process. This held particularly true for

features w ith rounded edges. For example, the surface area o f a flat top (Figure 3.3a) could be

assumed to be the prim ary area from which ink was transferred. On the other hand, a m ultitude of

d iffe re n t surface areas could be determ ined fo r convex dots depending on the extent o f top

curvature and sharpness o f shoulder edges (Figure 3.3b). There is currently no standard in the

printing sector which provides a single or separate methods covering both cases. The challenge lay in

developing a robust approach using the Vision32 software which was applicable to any type of

geom etry.

I 62
 Figure 3.3: Challenge of determ ining the surface area for d ifferent featu re geom etries. The flat top of
halftone dot (a) can be assumed to be the prim ary printing area, whereas the roundness of the halftone
top (b) allows m ultiple possibilities for the definition of the printing area.

The next tw o sections contain an outline o f the procedures fo r determ ination of planar surface area

of halftone dots on the printing plate. Interconnected shadow halftones receive a very similar

tre a tm e n t to isolated dots, thus only a short explanation o f necessary adjustm ents to the m ethod is

provided at the end o f the second section.

3.1.2.1 P lan ar Surface Area in Wyko Vision32

Previously the planar surface area was determ ined based on the assumption o f standard geom etric

shapes (Bould, 2001). For this a 3-D profile o f the convex halftone dot on the plate was obtained by

WLI and a 2-D cross section of this data set exported into a M icrosoft Office Excel spreadsheet.

There all data points th at fell below a certain surface height (i.e. the dot shoulders) were removed in

order to define the dot top. The expanse o f the remaining data points was defined as the diam eter

of the dot top and the planar surface area calculated as a circle w ith such a diam eter.

However, S. Hamblyn (2004) acknowledged th a t printing features rarely correspond to standard

geom etric shapes, and developed a m ethod em ploying built-in functions o f the WYKO Vision32

software. The principles o f his technique were adopted fo r this work, and a new customised display

and analysis function designed fo r the Vision32 software. The display comprised a 3-D view o f the

region of interest (ROI), the height histogram o f the data points w ith in the ROI and an analysis box

providing inform ation on planar surface area4 and volum e5 (Figure 3.4).

4 The planar surface area fu n ctio n is called "la te ra l surface area" in Vision32.

I 63
 as -F1-
Histogram

»0
MOO

■w -

1000-

too 300
*> 30 -10
Height

My Group
Lateral Surf Area 0 028678 mm 2
Volum e 130359.2969 um3

Figure 3.4: Customised display option in WYKO Vision32 softw are for determ ination of planar surface area and volume
showing a 3-D representation of a halftone dot, a height histogram of the surface pixels and the measurand parameters

Firstly, if several printing features were contained w ithin the original data set, a single dot was

isolated using the subregion option in the 3-D viewing box. The utm ost care was taken to avoid

tiltin g the sample during the m easurem ent stage. The option of tilt removal in the Vision32 software

was not used, thus preventing height skew in the data set. The heights were then displayed in a

histogram which characteristically shows one to three peaks in the height distribution. The peak at

the largest height corresponded to the dot top. The peak at medium height denotes the valley

between tw o adjacent dots on the plate. The th ird and lowest peak represents the plate floor. The

la tte r was not usually captured, as measurements were norm ally perform ed w ithin large fields of

adjacent dots.

To isolate the planar surface area from the rest o f the dot, the histogram peak fo r the dot to p and

the surface roughness Rz were used. This is based on the assumption th a t the planar surface

consists o f data points w ith heights centred around the dot top peak (DTP) and ranging from

(D T P ~ ~ R z ) 1:0 + AH histogram data below (^DTP — ^ R ^ was masked (i.e.

tem porarily removed from the ROI shown), leaving only the data points constituting the dot to p fo r

analysis. The quality o f the data displayed could be improved fu rth e r by manually masking stray data

5 The volum e fu nction referred to is th e "n a tu ra l vo lu m e " and explained in section "3.5.2".

| 64
points outside the main surface area, e.g. specks o f dust or other contam inations, however manual

intervention should be avoided to ensure a consistent approach and this was adopted in this work.

The planar surface area was then determ ined by the software based on the number o f pixels

constituting the remaining image.

(a) O riginal data set w ith missing data points

(b) Data set a fte r restoring and (d) Data set a fte r thresholding

before thresholding and before restoring

(c) Data set a fte r restoring and thresholding (e) Data set a fte r thresholding and restoring

Figire 3.5: Comparison of 3-D appearance of halftone dot for d ifferent restoration and threshold sequences in Vision32

I 65
Since this method is dependent on the pixel count, it is important that minimum data points are

missing from the original data set. This was the case for materials having low surface roughness, and

the method worked well for these. The rougher the material under investigation, the higher was the

likelihood that the data set contained gaps within the dot top where light scattering prevented data

capture. The lack of data points inevitably resulted in an underestimation of the planar surface area.

It was therefore necessary to introduce additional processing steps, before the planar surface area

could be determined.

The Vision32 software contains an algorithm to infill missing data points by interpolation of

surrounding data. The quality of the restored results is dependent on the amount of data missing.

Figure 3.5 illustrates the effect that the order of restoring and thresholding had on the original data

set (Figure 3.5 a). Only interior points were restored across a field of no more than five missing data

pixels. The limitation of pixel number improved the data quality, but failed to infill all missing points

within the dot area. Restoring the data first, had the disadvantage that the application of the height

threshold created new missing data points within the area of interest, thereby leading to the original

problem (Figure 3.5 b,c). A subsequent second data restoration was not possible. Applying the

height threshold before restoration, in many cases resulted in fields of missing data larger than the

restore option was able to fill, thereby yet again failing to resolve the original problem (Figure

3.5 d,e). This exhausted the capabilities of Vision32 without leading to the desired result of creating

a data set of the dot top without missing pixels. Therefore, the final approach was to export the data

set and to develop a new technique to infill the missing data.

3.1.2.2 Planar Surface Area in Image]

Strategically, image analysis may lend itself to improving the data set for the determination of the

planar surface area, and ImageJ (version 1.44o) (Abramoff, Magalhaes and Ram, 2004) was adopted

to explore this as well as being used for other parts of this work. Within the calculation sequence,

once the data for the dot top was isolated in the Vision32 software through height thresholding (no

data restore function applied), the data set was exported as a false colour image in TIFF. The created

file was then processed in ImageJ. For ease and speed of analysis, an ImageJ macro containing the

necessary processing steps was written (Appendix A .l). User interaction with the macro was possible

only to permit the selection of the correct ROIs within the images.

The main function of the macro was to separate pixels belonging to the image background from

those of the same colour representing the missing data within the planar surface area (both were

white on export from Vision32) (Figure 3.6a). Pixels within the background area were flood-filled in

| 66
the colour black, thereby distinguishing them from points representing the dot top and the missing

data points contained w ithin the top area (Figure 3.6b). The colour-separated pixels could then be

counted and stored to form a histogram. Other processing steps in the macro solely served the

purpose o f presenting the image and pixel count in such a way th a t the user could access the ROI

and count efficiently. For this step the image was converted to binary black-white (originally black

pixels rem aining black and any othe r colour pixels rendered w hite) (Figure 3.6c). The user was then

prom pted to select the ROI, and only the num ber o f w hite pixels was displayed in the histogram

(corresponding to the planar surface area in “ pixels" unit). Finally, the pixel count was converted into

area (expressed in “ pm 2") using calibration inform ation recorded on image generation by the

Vision32 softw are.

Figure 3.6: Image processing steps in ImageJ for determ ination of planar surface area of an isolated printing feature:
(a) original image w ith data points of the dot top (blue) and missing data points (w hite), (b) colour separation of
background pixels (black) and missing data points w ith in dot top (w hite) by flood-filling the background in colour black,
and (c) rendering all surface pixels (previously blue and w h ite) in colour w hite thereby including the missing data points
of the dot top in the surface area

The macro was tested on data sets obtained from a very smooth plate material which did not result

in missing data points. The relative difference in planar surface area calculated by the Vision32

softw are and the ImageJ macro was less than 0.1% for all samples and is attributed to the rounding

I 67
of ithe conversion factor between units. For consistency purposes this method was adopted for all

plate materials regardless of their surface roughness. The number of individual halftone dots

investigated for each line ruling and area coverage was dependent on the relative standard error of

the planar surface area calculated. The number was steadily increased (up to a maximum of 30),

until the standard error fell below 5.0%.

The planar surface area of shadow halftones, which were joined together, could not be determined

by this version of the macro, since it was not possible to isolate the individual features from each

other. Instead the planar surface area of the recess between halftones was computed with a second

ImageJ macro built on the principles set out above (Appendix A .l) and then the halftone area

calculated by subtracting from the square based on line ruling (refer to section 1.2.2). The two

macros are distinct in that one counts the white foreground pixels for halftones and the other the

black background pixels for the recesses. Both account for missing data points in the printing areas.

3.2 Surface Energy

Numerous methods exist for the determination of a surface energy (Adamson and Gast, 1967). For

this work a DAT 1100 Dynamic Absorption and Contact Angle Tester (FIBRO system ab, Stockholm,

Sweden) was available. This limited the options for the investigation of steady-state surface tension

of liquids to the pendant drop method and of steady-state surface energy of solids to the sessile

drop method.

3.2.1 Steady-state Surface Tension

For a detailed description of the pendant drop method and calculations involved therein refer to

Andreas, Hauser and Tucker (1938). In this work the method was employed to determine the steady-

state surface tension of the inks for the print trials, as well as to characterise the test liquids for the

sessile drop method (see below). The surface tension was calculated automatically by the algorithms

implemented in the DAT 1000. The method was found to be very consistent. The relative standard

error of the surface tension obtained by this method on 16 drops of each ink sample was 0.1%

maximum.

3.2.2 Steady-state Surface Energy

While the surface tension of a liquid can be found relatively easily using the pendant drop method,

problems arise for the establishment of a solid's steady-state surface energy by the sessile drop

method. The latter relies on a conversion of the Young contact angle which in itself is reliant on the

I 68
employment of the correct interfacial tension between solid and liquid. However, the nature of the

interfacial interactions and thereby the mathematical expressions describing these relationships are

still not fully known (Della Volpe etal., 2004; Zenkiewicz, 2007b). Two different schools of thought

on the treatm ent of the interfacial tension dominate the field of surface science. The Equation of

State approach takes the view that the interfacial tension and thereby the contact angle are only

dependent on the total liquid surface tension and the solid surface energy (Spelt and Neumann,

1987; Spelt et al., 1996; Kwok and Neumann, 1999). The Surface Tension Component (STC) approach

tries to reflect the origin of the total interfacial tension by dividing it into its contributing forces

(Fowkes, 1962; Owens and Wendt, 1969; Kaelble, 1970; Rabel, 1971; van Oss, Chaudhury and Good,

1988; Good, 1992; Della Volpe et al., 2004). Both methods are limited by the assumptions they are

based on, and practical results obtained are dependent on the test materials used (Kwok, Li and

Neumann, 1994; Zenkiewicz, 2007a; Hejda, Solar and Kousal, 2010). Nevertheless, their results are

considered to be very similar if the uncertainty of the methods is taken into account (Della Volpe

et al., 2004).

In the field of printing research, the STC method named after van Oss, Chaudhury and Good is

gaining importance (Mirle, 1989; Jarnstrom et al., 2007: Debeljak et al., 2013), but the slightly less

complex approach of Owens, Wendt, Rabel and Kaelble (OWRK) remains more popular (Quinn et al.,

1997; Pluhar, 2004; Kaplanova and Hejduk, 2006; Griesheimer, 2013; Stahl, 2013). Since the surface

energy components were not the focus of this work, where required, it was decided to use OWRK

which limits the surface energy components to a dispersive part, a d, (based on molecular van der

Waals interactions) and polar part, crp, (based on dipole-dipole interactions and hydrogen bonds):

° total = <Jd + ( T p Equation 3.1

The solid surface energy is determined graphically employing Equation 3.2 which in essence is a

linear equation. The dispersive part of the surface energy is deduced from the graph intersection

with the ordinate, and the polar part from the graph slope. For a good explanation of the derivation,

refer to Griesheimer (2013).

Equation 3.2

I 69
The contact angle determ ined by contour analysis o f the sessile drops as well as the known

dispersive and polar properties of the test liquids are inputs to the w ork model. To im prove the data

quality, it is necessary to choose a well-balanced set o f at least tw o test liquids, one o f which should

be predom inantly polar and the othe r one dispersive (Della Volpe et al., 2004; Hejda, Solar and

Kousal, 2010). In this w ork w ater, ethylene glycol and diiodom ethane were used to characterise the

surface energy of printing plates and substrates. The consistency of the sessile drop m ethod was

verified on three d iffe re n t substrates (tw o polypropylene films, one standardised coated paper)

using w ater as the test liquid. The relative standard error o f the contact angle measured did not

exceed 2.8% fo r any sample (based on 22 or 24 drops deposited). The approach was used to

determ ine the surface energy of printing plates and substrates used in this work.

3.3 Rheology
A comprehensive introduction to the differen t rheological properties, equations and rheom eters as

m easurem ent apparatus is provided by Barnes (2000). Of relevance to this w ork was forem ost the

property o f shear viscosity which describes the ink's resistance to flow under a prescribed strain

rate. In flexography, shear viscosities typically range between 0.05 and 0.5 Pa-s (Hubler et al., 2002)

which renders it a low-viscosity process. In Newtonian liquids the shear viscosity is independent

from the strain rate (Figure 3.7). Many flexographic printing inks display non-Newtonian, shear-

thinning behaviour characterised by a decrease in shear viscosity w ith increasing strain rate, i.e.

higher processing speed causes the ink to flo w more easily. Furthermore, viscosity is tem perature-

dependent. A rise in tem perature due to frictio n effects in the running printing press decreases the

viscosity (Olsson et al., 2007).

i i viscosity

shear-thickening

Newtonian

shear-thinning

shear strain rate

Figure 3.7: Schematic viscosity-shear rate diagram illustrating New tonian,

shear-thinning and shear-thickening properties of fluids

I 70
For this research the viscosity was determined using the shear rheometer Bohlin Gemini HR Nano

(Malvern Instruments, Malvern, UK). The lower part of the test system is formed by a stationary

Peltier plate, a thermal element which maintains the test liquid at a constant temperature. The

upper part of the system comprised a metal cone, the size and angle of which are selected in

accordance with the test fluid. The measurement space was enclosed by a solvent trap cover to

prevent the evaporation of ink components and any interaction with the environment.

Before testing, the inks were homogenised in their containers by stirring. As this might affect the

rheological properties, it was of essence to pre-condition the inks at very low strain rate on the

rheometer prior to the actual viscosity measurement. This was followed by a time at rest to regain

equilibrium. In order to capture any non-Newtonian behaviour during the test, the shear rate was

step by step increased from a minimum to a maximum value, and the corresponding viscosity

recorded. Tests were performed on three samples of each ink type and the relative standard error

(averaged for all strain rates) did not exceed 2.8% for any of them. The actual instrument settings for

the tests are provided in the materials part of each main chapter of this work.

3.4 Printing
The printing experiments for this work were conducted on a narrow-web industrial-scale printing

press under conditions as close to industrial ones as possible in order to ensure industrial relevance

of the research findings. Where applicable, the studies on the industrial-scale printing press were

preceded by experiments on a laboratory printability tester for the screening of material and process

parameters. The two different size machines employed and their related particulars are explained in

the following sections.

3.4.1 Printing on Laboratory Printability Tester

The IGT-F1 printability tester (IGT Testing Systems, Amsterdam, Netherlands) is a laboratory-scale

printer that simulates flexographic printing. The tester features the same main components (Figure

3.8) as an industrial printing press. However, its ink delivery system is reduced to a doctored anilox

roll, and the ink is applied to the anilox using a pipette. The printability tester is pressure-driven,

meaning the substrate carrier was advanced solely by the pressure exerted on it, when the anilox

and plate cylinders were engaged and rotating against each other. The ink was pressed onto (and in

the case of porous materials into) the substrate through the pressure in the printing nip. Adjustable

settngs on the IGT-F1 were: number of anilox roll revolutions before engagement with plate

cylinder, printing speed, force between anilox roll and plate cylinder as well as between plate and

impression cylinder. Samples produced with UV-ink were cured in a Jenton JA150SF UV-curing

I 71
conveyor unit (Jenton International, W hitchurch, UK). W here other ink types were used, the samples

were air-dried.

Doctor Blade Printing Plate &

Plate Cylinder

Anilox Roll
Substrate

Impression
Cylinder

Figure 3.8: Schematic of IGT-F1 printability tester containing anilox roll,

doctor blade, plate cylinder w ith printing plate and impression cylinder

3.4.2 Industrial-scale Printing

For the industrial-scale printing experim ents the T-Flex 508 (Timsons, Kettering, UK) flexographic

printing press was used. The T-Flex 508 was designed to run substrates from reel to reel. The

substrate firstly travels through an alignm ent unit to ensure lateral registration. Consistent web

positioning was essential fo r the fine alignm ent o f the image produced in consecutive printing units.

A corona trea tm en t unit follow s the web alignm ent unit, but its use was optional. The substrate then

proceeded through all fo ur printing units (Figure 3.9), before being rewound.

The w eb path, curing/drying station and air extraction were set up fo r the printing w ith UV-curing

inks. W ithin each printing unit in operation, ink was circulated continuously from an ink bucket

underneath the unit into the enclosed anilox chamber. The ink was transferred from the cham ber to

the anilox roll, printing plate and finally to the substrate. On exit from the printing unit, the ink was

cured im m ediately under UV radiation. This avoids the potential occurrence o f ink levelling and

thereby loss o f detail in ink laydown which is the essence of this work.
 Enclosed Printing P late &
Anilox Cham ber P late Cylinder

Impression Guide
Cylinder Cylinder

Anilox Roll

Substrate

Figure 3.9: Schematic of T-Flex 508 printing unit w ith UV lamp containing anilox chamber,
anilox roll, plate cylinder w ith printing plate, impression and guide cylinder

The'e are three main differences between the laboratory p rin te r and industrial-scale printing press.

Firsly, the ink supply by the anilox on an industrial-scale press is very consistent and is

chaacteristically part of a closed cham ber ink supply system. On the printability tester ink is

sup)lied manually using a pipette and this leads to a source of variability, although the utm ost care

was taken to supply ink in a consistent manner. For the print trials on either machine, it is assumed

thai the entire land area o f the anilox roll is wiped clean and th a t no ink is scooped out of the anilox

cell: under the doctor blade. Furtherm ore, the industrial press roller contact is set by engagement

andnot load, thus making the engagement independent of the image.

Thefollow ing three sections contain details on the general setup o f the T-Flex, the use of pressure

seniors to establish the cylinder setup and the m ethod to check fo r any cyclic variation in the print.

3.42.1 Setup o f the T-Flex 508

The accurate fixing of the printing plate to the plate cylinder was achieved using a double-sided

adhesive tape and a plate m ounting station (Heaford, Altrincham , UK) equipped w ith high-

m agiification cameras to achieve precise alignm ent.

On he printing press itself, the most im p orta n t parameters th a t had to be established during setup

w en the engagement between the anilox roll and plate cylinder as well as the engagement between

plat? and impression cylinder. Firstly, the gap between plate and impression cylinder was set using a

feehr gauge w hile bringing them together. The gauge was only slightly thicker than the com bination

I 73
of adhesive tape and printing plate, thereby bringing the mounted plate within micrometer range

from the substrate supported by the impression cylinder. Then the distance between anilox roll and

plate cylinder was set in the same fashion.

In the next step the rotating anilox roll and ink chamber were engaged for filling, and the anilox was

brought into contact with the printing plate. Any skew in the anilox cylinder was adjusted until the

full length of the revolving printing plate received ink evenly and sufficiently. Then the ink chamber,

anilox roll and plate cylinder system was moved towards the rotating impression cylinder. After the

first contact any skew was removed and the initial distance between printing plate and impression

cylinder was adjusted so that they made "kiss contact", and minimal ink transfer took place in this

zero reference position. The final engagement was then set by moving the two cylinders together

beyond the kiss contact. The engagement could only be adjusted manually on the T-Flex 508 in

increments of one thousands of an inch (abbreviated as "1 thou") which corresponds to 25.4 pm.

3A.2.2 Thin Film Pressure Sensors

The kiss engagement at the printing nip is dependent on the printer's perception of minimal, even

ink transfer across the cylinder. Although the operator subjectivity cannot be removed from the

process, crosschecks of engagement with other methods allow verification of good press setup, e.g.

with regard to skew of cylinders. Since it is difficult to measure the engagement in the form of

absolute distance between cylinders, the pressure in the cylinder nip served as indicator for

engagement and uniformity across the cylinder width.

Engagement pressure was measured by an ELF force and load measurement system (Tekscan,

Boston, MA, USA) using a single element, thin-film FlexiForce sensor. The sensor was placed on a

continuous raised part of the image on the printing plate. The press was then inched forward and

the sensor carried into the nip between the cylinders. The pressure was recorded the entire time,

thus providing data for the nip inlet, nip itself and nip outlet. The maximum pressure recorded was

used as an engagement indicator. If too large a pressure difference was detected between the two

ends of a cylinder, the skew could still be adjusted before the print run. The pressure data obtained

was also important for the comparability of printing results obtained from different printing plates.

The latter will be elaborated on in Chapter 4.

I 74
3.4.2.3 Test fo r Cyclic Variations

S. Hamblyn (2004) revealed cyclic variations in the print quality produced on the T-Flex 508. The

optical density o f the prints varied at a frequency o f four copies and a range of 0.15 between

maxim um and m inim um density values. As this occurrence m ight be setup-dependent, prints from

the main trials fo r this w ork were scrutinised fo r any periodic deviations in optical density obtained

by spectrophotom etry.

The range o f solid densities measured on eight consecutive sheets of the dot geom etry study (refer

to Chapter 4) was below 0.03 (4.0%) fo r all plate materials investigated. This was not significant

enough to require fu rth e r analysis o f cyclic variation, since this was w ith in the noise range of the

printing press. The range o f solid optical densities observed fo r 16 consecutive prints o f the meso-

pattern study (refer to Chapter 5) was larger w ith 0.10 (11.0%). However, the extrem e values o f the

range occurred on d ifferen t sheets fo r neighbouring printing patches which rules out cyclic variation

o f the printing press (Figure 3.10). The cause o f the tem poral deviations observed could not be

identified. To account for any variation, spectrophotom etric data was obtained on eight consecutive

sheets and averaged.

♦ High anilox volum e B Low anilox volum e

0.9

> 0.7

£ 0.4
O 0.3
0.2
0.1

1 6 11 16
Sheet

Figure 3.10: Tem poral variation in solid optical density measured on consecutive sheets to test for cyclic variations

3.5 Print Characterisation

In order to evaluate the effect of plate characteristics on ink transfer, both qualitative and

quantitative measures need to be considered. Print characterisation starts w ith the visual inspection

of samples by microscopy which yields qualitative inform ation. This can help to in te rp re t the data

I 75
obtained by the follow ing quantitative methods. The quantitative data comprises printed planar

surface area, ink volum e transferred, optical density and p rint uniform ity. The respective approaches

and apparatus are explained.

3.5.1 Visual Inspection of Samples by Microscopy

The VHX-1000 digital microscope (Keyence, Osaka, Japan) was used fo r the inspection o f the prints.

The high-resolution Keyence possesses a coaxial light source, i.e. the light path was in line w ith the

optical path, which im proved the contrast o f the printed ink layer and substrate in the images

captured. Furtherm ore, the Keyence was designed for capturing colour inform ation at up to

1,000 tim es m agnification which rendered it ideal fo r small printed features.

3.5.2 D eterm in ation of Planar Surface Area and Volume of Printed Dots
One way to determ ine the planar area o f a printed dot w ould be to measure its diam eter under a

microscope and to calculate the surface area by assuming the dot to be a perfect circle. As can be

seen from Figure 3.11, real dots are irregular around th e ir circum ference and m ight contain UCAs

w ithin the dot. Accounting fo r these voids by geom etric techniques w ill give only a crude

approxim ation of the surface area o f such a dot. Another possibility would be the processing o f the

microscopic image in ImageJ where the exact num ber o f pixels making up the printed d o t could be

established and then converted to planar area. However, this option is only viable fo r images w ith

sufficient contrast between dot and substrate. In the case o f the dot in Figure 3.11 any isolation

a tte m p t failed due to the varied coloration and brightness o f the dot pixels. A required additional

manual selection of pixels was not available in ImageJ.

----------------Contour

Printed dot

UCA

Lent zlooxiooo

Figure 3.11: Microscopic image of printed dot showing irregular contour of the main ink volum e and UCAs
It was there fo re decided to adapt S. Hamblyn's (2004) m ethodology of using WLI and the built-in

functions o f the Vision32 softw are to determ ine the planar surface area and ink volume of a printed

dot. The process steps were analogous to the m easurem ent of the planar surface area of a halftone

dot on the plate (above), but - instead o f isolating the dot top from the shoulders of the plate

feature - the ink volum e transferred was isolated from the substrate by height thresholding.

Although the m ethod was suited for the thicker, defect-free, uniform ink film s investigated by

S. Hamblyn, it had to be refined fo r the dot structures deposited in this work.

The step o f isolating data points representing the ink from those representing the substrate by

height thresholding was replaced by the manual exclusion o f substrate pixels from the data set. For

this the outline of the printed dot (Figure 3.12a) was traced w ith a polygon tool to form a closed

contour. The substrate pixels outside the contour were masked to not appear in the image analysis.

Any unprinted areas w ith in the dot were separately traced and excluded. All remaining data points

denoted ink deposition and were used fo r the calculation of planar area and ink volum e by Vision32

functions (Figure 3.12b). Before the volum e function could be applied, an inversion of the height

inform ation in the data set had to be perform ed. This additional step was necessary, because by

definition the (natural) volum e calculated by the software is the "volum e th a t the surface would

hold if it were covered just to the surface of the highest peak" (Lippold and Podlesny, 1998), i.e., the

volume above the printed dot instead o f the ink volum e itself (Figure 3.13a). The data inversion

rectified this problem by matching both volumes (Figure 3.13b). The built-in function o f planar

surface area remained unaffected by the data inversion.

The m ethod o f manual ROI selection achieved very consistent results. Ten repetitions of masking the

same data set o f a printed dot resulted in a relative standard error o f 0.1% fo r the planar surface

area and o f 0.6% fo r the ink volume.

Figure 3.12: M anual selection of ROI for d eterm ination of planar surface area and ink volume
(a) printed dot before and (b) a fte r the manual masking selection of data points

I 77
 Figure 3.13: Schematic of (natural) volume algorithm in Vision32 softw are (a) showing difference betw een
actual and calculated ink volum e. Matching up of actual ink volum e w ith algorithm by data inversion (b).

3.5.3 D eterm in ation of Optical Density by Spectrophotom etry

For the colour measurements a Spectrolino m easurement head by Gretag M acbeth (Regensdorf,

Switzerland) was used. For measurements of sm aller samples printed on the IGT-F1 the Spectrolino

was utilised in hand-held mode, w hile for repetitive measurements o f large T-Flex prints the

instrum ent was attached to a ColorScout A+ m easurem ent table (ColorPartner, Kiel, Germany). The

ColorScout A+ softw are contained a tem plate option which allowed fo r exact positioning of the

instrum ent head in the XY-plane fo r every sample. This was achieved by m atching three registration

marks on the printed sheet w ith the alignm ent marks in the tem plate (Figure 3.14). This enabled the

repeatability o f measurements at the same point o f an image, as well as the direct comparison of

colour inform ation on several consecutive sheets or those produced under d iffe re n t printing

conditions.

^ J ^ W C P C

ura

[fin

2 =3*
si

(a) (b)
Figure 3.14: Original image of prints (a) versus positioning tem p la te for ColorScout A+ (b)
including the three cross-hair marks for alignm ent and positioning

All measurements were taken in accordance w ith BS ISO 12647-1:2004 (Table 3.1). The Spectrolino

measured the spectral response of the sample in the wavelength range o f 360 to 730 nm at intervals

o f 10 nm (36 values in total). From the spectral response the software calculated the optical density

separated by the channels CMYK. The selection o f sample backing was dependent on the
experiment. At the time that the studies on the IGT-F1 were conducted, no standardised backing

was available. Instead a sheet of non-standard, white card was used. This did not pose a problem, as

all sample substrates were opaque and the colour influence of the background was therefore

negligible. The T-Flex 508 print samples measured on the ColorScout A+ all profited from the matte,

opaque white backing (L*<92, C*<3 according to BS ISO 12647-1:2004) which was part of the

measurement table. The dimensions of the printed patches were optimised to the smallest size

feasible (6 mm) in order to fit more elements onto the same printing plate, while at the same time

not to compromise the necessary sampling aperture of the Spectrolino (4 mm in reflection mode).

Table 3.1: Measurement parameters for Spectrolino with ColorScout A+ measurement table

Measurement parameter Value

Mode Reflection

White base Absolute

Backing [dependent on experiment]

Geometry 45°/0°

Observer angle 2°

Illumination (calculated) D50

Filter D65

Polarisation No

Instrument spectral response ANSI Status T

3.5.4 Digitisation of Prints by Scanning

The prints had to be digitised in order to be able to perform the print uniformity analysis on the

digital images (outlined in the next section). For the digitisation the Epson Perfection V700 Photo

(Epson, Suwo, Japan), a flatbed scanner with dual lens system for colour images, was used.

According to the manufacturer the maximum optical resolution achievable with the scanner is 4,800

by 9,400 dpi for reflective and contact sheet scanning at a maximum colour depth of 48 bit (just

under 281.5 trillion) colours. With regard to archiving of the digitised data and potential future

analysis, it might appear desirable to retain as much printed image information as possible, i.e. to

scan as large an area in original colours at as high a resolution as possible. However, this approach

will result in particularly long scanning times and large digital files which could not be handled easily

by different types of software. A compromise between scan colour, resolution, sample and analysis

area became inevitable, while at the same time a significant loss of image information expressed in

deviant measurands had to be avoided. Besides the optimum scan parameters it had to be

I 79
established how stable and consistent the scanning was performed over time. Therefore, the print

digitisation was preceded by an extensive screening of scanning specifications as outlined in

Appendix A.2. The final parameters employed for this work are given in Table 3.2.

Onliy after completion of the scanning task had it been discovered that - unbeknown to the user -

the scanner software applies a non-linear algorithm for image manipulation which has a significant

effect on the scan results and image analysis measurands. Details on the linearity problem and its

resolution are provided in Appendix A.2.4. A non-linearity inversion was applied manually to all

relevant scan data series and new measurands calculated. All the data presented in this work is

based on the corrected linearised scans.

In performing the scans it should be noted that great care was taken to remove any contaminations

fro m the prints and scanner glass before scanning. At the beginning of each scan session the glass

was cleaned with lint-free wipes and isopropyl alcohol. Before each scan the glass and print were

duslted with a soft cloth.

Table 3.2: Scan parameters for digitisation of prints

Scan parameter Value

Scan mode reflective on document table

Scan type photo

Scan quality "best"

Scan colour 8-bit greyscale

Scan resolution 2,000 dpi

Sample area scanned 20 by 15 mm

Sample area analysed 10 by 10 mm

Compression none

File format saved multi-page TIFF

Gamma correction 1.66 (applied post-scanning)

3.5.5 Digital Image Analysis in Image)

The density of prints can also be estimated using ImageJ. This approach has the added advantage

that the print uniformity of the image can be quantified at the same time.

3.5.5.1 Determination of Print Density

The prints were digitised according to the parameters outlined in the previous section. A macro

(Appendix A.3) was employed to generate a histogram of greyscale levels (GSL) for the ROI selected

| 80
in each image level o f the multi-TIFF file. Based on the 8-bit greyscale image, 256 d iffe re n t bins were

included in the histogram comprising the range o f colours from black (0) across the shades o f grey to

w h ite (255) (Figure 3.15).

25000

20000

= 15000

•S
Q. 10000

5000

100 150 200 250

G re y s c a le le v el

Figure 3.15: Example of GSL distribution in histogram for a printed and subsequently digitised sample
(D = 0 .4 6 ,MGSL = 154.04)

The spectrophotom eter as well as the scanner capture reflectance values ((3=0...1) o f the printed

samples which correspond linearly to luminance values (Y = 0...100). The reflectance coefficient

obtained by the spectrophotom eter is an average value of the w hole sample area covered by the

measurement aperture and converted into optical density, D, using Equation 1.1. The scanner's

reflectance coefficients are transform ed into GSLs (GSL = 0...255) on the basis o f a linear relationship

and captured in the histogram. The mean greyscale level (MGSL) is the average value of the

histogram and thereby the average reflectance coefficient obtained by the scanner. The MGSL

corresponds to the optical density which is the average reflectance coefficient o f the sampled area

obtained by the spectrophotom eter. Therefore, the MGSL is related to optical density in an

exponential manner. (Hall, 1979) This was confirm ed experim entally using data from the first meso-

pattern trial on the p rintab ility tester (refer to chapter 5.1). Twelve patterned patches were printed

using the IGT-F1 p rinta bility tester in 64 experim ents featuring d ifferent com binations o f six process

parameters. Figure 3.16 compares the 768 experim ental data pairs w ith the theoretical values

com puted using Equation 3.3 below. Clearly the correlation is very good. Optical densities obtained

through spectropho tom etry w ill be denoted as such, whereas optical densities calculated from GSLs

w ill be denotes as "p rin t density".

 MGSL = 255 • 1 ( T d E qu atio n 3.3

♦ Experim ental ■ Computed

300

0 0.5 1 1.5 2 2.5 3

O p tic a l d e n s ity (-)

Figure 3.16: Correlation of optical density (spectrophotom etry) and MGSL (ImageJ)
for experim ental and com puted data pairs (com putation using Equation 3.3)

3.5.5.2 Determ ination o f P rin t Uniformity

Dependent on the type o f print nonun iform ity, many diffe re n t approaches fo r its quantification can

be found in the literature (Vo(3, 2002; Galton, 2004b; Dube et al., 2005; Hallberg, Odeberg Glasenapp

and Lestelius, 2005; Sadovnikov et al., 2005 a,b; Sadovnikov, Lensu and Kalviainen, 2007; Barros and

Johansson, 2007; Galton and Rosenberger, 2007; Rosenberger, 2010). An essential part o f the

m ajority of models is the mean optical density, D, and its standard deviation, oD, across a sample

area. Alternatively, the MGSL and its standard deviation, StDev (Equation 3.4), can be employed.

Often they appear as the Coefficient o f Variation, CV (Equation 3.5).

Equation 3.4
n
S tD e v ^ ( G S L t - MGSL)

where n is the num ber of pixels in the image.

on
On S tD ev
OLUVV Equation 3.5
CV = -=- = ----------
D MGSL

| 82
The CV has previously been used to characterise the uniformity of layers transferred in functional

device printing (Stahl, Sauer and Dorsam, 2012). However, uniformity indicators obtained with this

equation need caution in their interpretation. The mean optical density acts as a scaling factor and

distorts the true uniformity value. Assuming two samples containing identical printing defects

(i.e., = StDev), but different thicknesses of the underlying ink film (i.e., * MGSL), the sample having

the higher MGSL will result in a lower CV and therefore could be interpreted as more uniform. Also,

two apparently identical CV values might belong to samples of rather different print uniformity

whose results were biased because of the MGSL magnitude (Figure 3.17).

Condition 1 Condition 2

jar ■ ' v , • ^ " -

l : •?.,?-,%
% *rvV * *!

;'t’ ^ '■ '• •“'s ' ’ V t' \

* *

k* jf f 3 * *
%') ' «*''« i •’*1
. 4f-;v ; , .<V,.* , * a

MGSL = 141.43, StDev = 21.125, CV = 0.149 MGSL = 110.94, StDev = 16.68, CV = 0.150

•Condition 1 •Condition 2

25000

20000

15000

* 10000
Q.

5000

0
0 50 100 150 200 250
Greyscale level

Figure 3.17: Illustration of deficiency of Coefficient of Variation approach for print uniformity
on tw o examples produced under different printing conditions. Identical CV values represent
strongly differing print uniformity better captured by the StDev parameter.
Hence it was decided to adapt the ISO M ottle approach according to ISO 13660:2001 which uses the

standard deviation on its own. However, instead of dividing the sample image into smaller tiles,

calculating the standard deviation for each tile, and then averaging them over the whole image as

specified in the ISO standard, for this work only a single StDev was calculated for an image. The

smaller the StDev, the more uniform was the print. Graphically high homogeneity of GSLs was

represented by a narrow, unimodal distribution with very short tails in the histogram. Skewed, bi- or

multimodal distributions with longer tails were the result of irregular GSLs in the ROI and implied

nonuniformity of the print.

A small investigation was conducted to check the correlation between print uniformity and StDev.

Ten people were asked to rank prints of nine different patterned patches produced on the IGT-F1

printability tester from "most uniform (1)" to "least uniform (9)". All participants had normal or

corrected-to-normal vision. The viewing conditions were the same for everyone (ambient lighting -

indoors, daylight, overcast). The level of experience with this kind of task ranged from "expert" to

"novice". The decision on the worst and two best prints was unanimous. The ranking of the six

samples in between was more varied and depended on the individual's perception of random and

periodic defects in the prints. The ranking results were averaged for all observers. The correlation of

visual ranking and StDev for the samples produced a correlation of R2 = 0.79 (Figure 3.18). Taking

into account that the scale for visual ranking was a linear one and unable to capture the severe

difference between some of the print defects as well as the participants' variable sensitivity, then

the correlation could be regarded as sufficiently good to allow the use of StDev as a measurand for

print uniformity.

30

bO

</)

Visual Ranking

Figure 3.18: Correlation of uniformity ranking obtained by visual judgement ("most uniform (1)" to "least uniform (9)")
and uniformity parameter StDev obtained by image analysis software ImageJ ("most uniform (0)", scale open-ended)

| 84
3.6 Design of Experiments
The majority of print trials conducted for this work were laid out as experimental plans according to

DoE. This allowed quantifying the effects and interactions of printing parameters, such as plate

material, printing speed and pressure, on quality criteria of prints, such as optical density and

uniformity, in a simpler and more comprehensive manner. The first order, full-factorial plans were

analysed according to Roch (2007) and Montgomery (2012). The latter provides an introduction to

DoE and contains all related equations used in this work (calculation of main and interaction effects,

variance and confidence intervals forjudging the significance of effects).

Throughout this work all confidence intervals were determined by a two-sided significance test

based on Student's t-distribution and a 5% error of probability. For sample sizes smaller than 13, the

standard deviation was determined as the "alternative measure of spread" (Oakland, 1987).

3.7 Closure
This chapter has detailed the experimental methodologies employed in this research. WLI made it

possible to characterise the printing plates non-destructively prior to printing. The plate surface

roughness and selected feature dimensions were determined by standard procedures. A new

method has been developed for the calculation of the planar surface area of rough materials in the

image analysis software ImageJ after data preparation in the WLI software Vision32. Descriptions of

the printing on the laboratory printability tester IGT-F1 and the industrial-scale printing press T-

Flex 508 have been presented. A refined method for the determination of planar surface area and

volume of printed dots using WLI was introduced. Furthermore, the standard method of measuring

optical density of prints by spectrophotometry was supplemented by a newly developed method for

digitised prints in ImageJ which allowed analysing print uniformity at the same time. The robust

design of the developed methods has been established.

The procedures for plate analysis and printing formed the basis of the research conducted on dot

geometry and surface patterning of printing plates presented in the next three chapters. The

analysis of the planar surface area and volume of printed dots features in the investigation of dot

geometry. Optical density and print uniformity are the key aspects to the investigation of surface

texturing on printing plates.

| 85
Chapter 4 D ot Top G eom etry o f Halftones
One of the recent key areas of improvement in flexography has been the imaging technology as a

means to control the dot geometry. "Flat-topped" halftones are claimed to be capable of superior

print image quality because of higher ink-carrying capabilities and lower susceptibility to the plate

deformation mechanisms causing physical dot gain in the prints. However, there is no satisfactory

scientific explanation or justification to support these claims, and research on the detailed effect of

dot geometry on the ink transfer mechanisms is lacking.

The work reported in this chapter seeks to fill in part of the research gap by comparing the influence

of dot geometries resulting from different imaging technologies on halftone ink transfer. Four sets of

printing plates (two imaged by standard digital technologies, two by "flat-top" technologies) were

characterised through WLI and then used in a print trial on an industrial printing press. The ink

transfer was evaluated using dot gain and film thickness parameters (calculated from the area and

volume of ink transferred, determined by WLI on the prints), as well as the resultant optical density

of the prints. Special attention was directed to the types of printing defects created and how they

affect the applicability of selected halftone models.

4.1 Materials - Plates, Substrate and Inks

P late Pr o p e r t ie s

In order to generate different feature geometries on the printing plate, four separate combinations

of plate material, imaging technology and resolution were employed (Table 4.1). The plate materials

were ty Asahi Photoproducts Europe (Shenfield, UK), Eastman Kodak and MacDermid (Waterbury,

Connecticut, USA). An overview of the plate-making technologies was given in section 1.2.1.1. The

standa'd digital, water-washable Asahi AWP-DEF and the solvent-washable MacDermid Digital Rave

were ciosen for the comparison of the ink transfer from standard digital printing features on plates

of different chemistry (as indicated by their washability). The same MacDermid material was also

imagec under identical conditions with LUX technology (proprietary to MacDermid), i.e. the plate

was co/ered with an oxygen barrier layer during imaging, thereby preventing oxygen inhibition and

suppossdly resulting in "flat-topped" printing features. Using the same MacDermid plates imaged

under different conditions served to directly compare the effect of imaging technology on dot

geome:ry and ink transfer. Finally, a Kodak Flexcel NX material was imaged with Flexcel NX

technoogy (oxygen barrier layer located underneath imaging layer) to facilitate a second group of

"flat-tcpped" printing features. It was not possible to achieve the same imaging resolution for the

| 86
Kodak plates, because Flexcel NX is a proprietary technology which is supposed to be processed by

Eastman Kodak's own platesetters exclusively. These only operate a SQUAREspot laser (Kodak, 2012)

at the lower resolution of 2,400 dpi. The Asahi plates were imaged by Creation Reprographics

(Daventry, UK) and all the other plates by SGS Packaging Europe (Hull, UK). For each parameter

combination a set of four printing plates was produced (one for each process colour CMYK) and the

sets were referred to as Asahi, Kodak, MacDermid and MacDermid Lux.

Table 4.1: Parameter combinations for the generation of different feature geometries on printing plate

Property Asahi Kodak MacDermid MacDermid Lux

Material name AWP-DEF Flexcel NX Digital Rave

Washability water solvent

Surface roughness, Ra 171 nm 83 nm 89 nm 53 nm

Surface roughness, Rz 6.2 pm 1.0 pm 1.8 pm 1.7 pm

Steady-state
38.1 mN/m 36.2 mN/m 44.6 mN/m
surface energy

Thickness 1.7 mm

Imaging technology Standard digital Kodak Flexcel NX Standard digital MacDermid LUX

Imaging resolution 4,000 dpi 2,400 dpi 4,000 dpi

Laser geometry Gaussian SQUAREspot Gaussian

All printing plates were based on the same digital image (Figure 4.1). The following elements were

relevant for the investigation of dot geometry:

• Halftone scales in the colours CMYK

o 16 area coverages (1, 2, 3, 4, 5 ,1 0 , 20, 30, 40, 50, 60, 70, 80, 90, 95,100% ) each at

o 6 line rulings (85, 100, 120, 150, 175, 200 Ip i; 33, 39, 47, 59, 69, 79 lines/cm

respectively).

The range of line rulings chosen is representative of that found in industrial flexography. 85 Ipi is

used for coarser packaging printing; the intermediate line rulings of 100 to 150 Ipi are standard

practice in graphic printing; the higher resolutions of 175 and 200 Ipi are the industrial goal, but

cannot currently be achieved consistently by all plate-making processes. The choice of area

coverages comprises highlights, midtones and shadows to allow observation of any influence of

halftone structure on dot geometry and print quality. The lowest coverages of 1 to 5% were used to

determine the smallest dot size that can be imaged on the plate.

|8 7
 WCPC
I \ ji

tit

Figure 4.1: Image for print trial on dot geom etry showing halftone scales in CMYK, full tone patches, photographs
and various other test elem ents (image parts contained in red box taken into account for image analysis; red box not
included in actual image) (0.3x magnification)

For the plate characterisation, WLI measurements were perform ed on the follow ing plate patches;

• colour black - all halftone patches at all line rulings, and

• colours cyan, magenta and yellow - halftones o f 50% area coverage fo r all line rulings.

For each patch, up to 30 printing features were investigated depending on the standard error of

results. The planar surface area o f each feature was determ ined according to the m ethod stated in

section 3.1.2.2, and the isolated dot tops were analysed for th e ir geometry. The investigation found

th a t the top geom etries on the fo u r plate types were not lim ited to simple round or fla t shapes. The

full range from convex to concave dot tops was encountered. A new term , "cupping" was introduced

fo r non-planar dot geometries, which has not previously been described in literature. Cupping holds

great industrial significance as it potentially increases the ink-carrying capabilities o f the printing

plate (explained in section 4.4.1.2) and affects the print uniform ity o f halftone dots (sections

4.4.2.1.1 f.). The measurand of top geom etry was height (here referred to as "cup depth"). The

| 88
height o f concave dot tops was measured between the apex o f the cup brim and the bottom o f the

cup. For the purpose o f comparison, the cup depth of convex dot tops was calculated as the height

difference o f the top edge to the apex (Figure 4.2). For every dot four values were obtained, tw o

along each o f the profiles A and B (Figure 4.3).

reference heights

convex dot top concave dot top - "cupping"

Figure 4.2: M ethodology for quantification of dot geom etry using dot apex and threshold height for convex dot top
(positive sign) as w ell as cupping brim and bottom for concave dot top (negative sign)

Profile B

Profile A

Figure 4.3: Location of WLI profiles along which the

cup depth of the dot top was measured

Halftone

Figure 4.4: Comparison of surface striation in halftone dots (50% dot at 100 Ipi) and solid for Asahi plate
( llO x m agnification)

| 89
The determination of top geometry was impeded by surface striation that was encountered on the

Asaihi and both MacDermid plates. Figure 4.4 shows an example of striae in halftone and solid areas

on the Asahi plates (complete tables of images and striation parameters for all plate types can be

fouind in Appendix A.4). The striae are created by the laser beam during the ablation of the plate

mask layer (Johnson et al., 2009; Hejduk, 2010). The Kodak plate was free of striae, because its mask

is alblated separately on the barrier layer and only then laminated onto the printing plate. It was not

possible to consistently select dot profiles that did not contain striae. Therefore, the striation was

treated like any other surface point and included in the determination of top geometry. If the

mimimal surface point was required for the measurement, it was taken in the striation valley. The

maximal point was found on the striation apex.

S u b s t r a t e P r o p e r t ie s

The film substrate Rayoface WPA59 by Innovia Films (Wigton, UK) was used for the print trial,

because ink cannot penetrate into this non-porous substrate. This permitted easier quantification of

the ink volume transferred and ink laydown by optical methods. The Rayoface substrate was a high

gloss, white, biaxially oriented polypropylene film with proprietary top coating The roughness of the

print side was Ra = 149 nm and Rz = 1.46 pm, and its surface energy was 43.2 mN/m.

I n k P r o p e r t ie s

For the investigation of the interaction between inks and plate features, it would have been

desirable to employ several ink systems based on fundamentally different chemistries, such as UV-

curing, solvent- and water-based ink. However, since the printing press was primarily designed for

UV-curing inks, each of the four printing plates within a set was printed with one of the UV-curing

process colours of the series Solarflex Nova SL Pro DK03 by SunChemical (Slough, UK). UV-inks have

the advantage that they only experience minimal volume shrinkage during curing, because no

solvent loss is taking place. This provides clearer information on the amount of ink transferred

during the printing process. The inks were characterised in terms of their surface tensions (Table 4.2)

and viscosity (instrument settings in Table 4.3 and Table 4.4). All four inks displayed a similar

magnitude of shear viscosity between 0.7 and 1.7 Pa-s (Figure 4.5). The black and cyan inks were

approximately Newtonian. The yellow and magenta inks had slight shear-thinning properties which

distinguished them from the black and cyan inks.

| 90
 Table 4.2: Ink surface tension of SunChemical Solarflex Nova SL Pro DK03 series process cdours

Ink Cyan Magenta Yellow Black

Steady-state surface tension (m N /m ) 32.5 35.3 31.9 32.4

Table 4.3: Rheom eter settings for ink pre-conditioning and m easurem ent geometry
(inks of industrial printing press studies)

Pre-conditioning Measurement geometry

Type controlled rate Geometry cone/plate
Shear rate 0.01 s 1 Cone 4° / 40 mm
Time 60s Gap size 150 pm
Equilibrium time 10
Temperature 25°C (isotherm al)

Table 4.4: R heom eter settings for ink viscosity determ ination (inks of industrial printing press studies)

Type shear-controlled rate

Minimum shear 0.1 s'1
Maximum shear 100 s 1
Delay time 10 s (decreasing tim e)
Integration time 10 s
Mode up
Samples 21
Repeats 2
Temperature 25°C (isothermal)

■ Sun Black ■ Sun Cyan ■ Sun Magenta D S un Yellow

1.6
1.4
1.2 ^ y q—□—
o
_ _ ■ ■
0.8
0.6
0.4
0.2

0.1 1 10 100
S h e a r s tra in r a te ( 1 /s )

Figure 4.5: Ink viscosity of SunChemical Solarflex Nova SL Pro DK03 series process colours

| 91
4.2 Printing

4.2.1 P rinting Conditions

The print trial was perform ed on the T-Flex 508 industriai-scale printing press. The trial was designed

to test each set of plates under a broad range o f printing conditions, notably printing speed and

plate-to-substrate engagement (Table 4.5). The engagement between anilox roll and plate was not

included as a prin ting param eter, because its setup could not easily be recreated fo r different plates.

For each plate set, all param eter com binations of speed and engagement were executed in the order

o f increasing engagement, firstly at low er and then at higher speed. A fter printing at the highest

engagement o f 4 thou, the engagement was reduced and an additional experim ent run at 2 thou.

This served to check w hethe r the system recovered fully once the load was removed from the

printing plate.

Table 4.5: Combination of print param eters for dot g eom etry trial

Experiment number Printin g speed Engagement plate- substrate

(ft/m in ) (m /m in) (thou) (pm)

1 1 25.4

2 2 50.8

3 150 45.7 3 76.2

4 4 101.6

5 2 (Hysteresis) 50.8

6 1 25.4

7 2 50.8

8 300 91.4 3 76.2

9 4 101.6

10 2 (Hysteresis) 50.8

The fo u r process colours were all applied in the same print run using laser-engraved anilox rolls

(Table 4.6) and cured at 70% pow er (standard settings) o f the UV-lamps. This arrangem ent led to ten

experiments per plate set and a to ta l o f 40 individual experim ents fo r the investigation of dot

geometry. The printing order o f plate sets was Kodak, MacDermid, Asahi and MacDermid Lux. All

plates were m ounted onto the printing cylinders w ith the com bination tape 1015 (3M, Bracknell, UK)

which was suitable fo r the halftones as well as fo r the solid printing elem ents on the image.

| 92
 Table 4.6: Anilox specifications for dot geom etry trial

Colour Anilox volume (cm3/m 2) Screen ruling (Ipi) Print order

Yellow 3.10 Unit 1

Cyan 2.39 Unit 2

1,200
Magenta 3.70 Unit 3

Black 3.32 Unit 4

4.2.2 P rin ting Force by Thin Film Sensor

To facilitate a direct comparison o f the printing features on different plates, all were printed at

identical speeds and engagements. To ensure th a t the engagement setting resulted in identical

printing forces fo r all plates, these were checked using thin film sensors to measure the "anilox

fo rce " in the anilox-plate nip and the "p rintin g force" in the plate-substrate nip (determ ined at

1 thou engagement). The form er force prom otes the inking o f the printing plate, whereas the la tte r

initiates the ink transfer to the substrate. The thin film sensor was placed on the plates' bearer bars

(Figure 4.6), as this provided a sufficiently large continuous area fo r testing.

bearer bars

possible locations
for thin film sensors

Figure 4.6: Possible locations along plate's bearer

bars fo r force m easurem ent using thin film sensor

When the press was set up, the anilox pressures fell w ithin a similar range fo r all fo u r plate sets

which im plied a consistent setup o f the engagement between anilox and plate (Figure 4.7). The

mean printing force between the plate and substrate is shown in Figure 4.8 where the mean force is

about 30 N and very consistent fo r the Asahi and Kodak plates. The MacDermid plates showed some

inconsistency. This was due to low spots on the printing plates w hich did not make sufficient contact

w ith the anilox or substrate at low er engagement. In order fo r the low spot to print, it was necessary

| 93
to increase the printing pressure to achieve kiss contact across the entire image, thereby potentially

over-impressing othe r parts of the printing plate such as the bearer bars. This is a com m on issue in

flexography and results in increased mean printing force for plates containing low spots.

■ Cyan ■ Magenta □ Y e llo w ■ Black

Asahi Kodak M acDermid M acDermid Lux

Figure 4.7: Comparison of consistent setup of anilox-plate cylinder nip using anilox force for all plate and colour types

■ Cyan ■ Magenta □ Yellow ■ Black

70

_ 60

<u 50
u

c
1 30
c

10

0
Asahi Kodak M acDerm id M acDerm id Lux

Figure 4.8: Comparison of consistent setup of printing nip using mean printing force for all plate and colour types
4.3 Print Characterisation

4.3.1 Planar Surface Area and Volume of Printed Dots

An inspection of the prints with regards to the quality of ink laydown was first performed with the

Keyence digital microscope (l,000x magnification). The use of optical microscopy allowed the

comparison of 2-D real colour images with 3-D false colour, topographic images obtained by WLI.

The planar surface area and volume of printed dots were determined from WLI data for the different

parameter combinations of printing speed and engagement for a representative selection of

halftone line rulings and area coverages (Table 4.7). The 100 Ipi and 150 Ipi were chosen as these are

two of the most frequently used line rulings in industry. The 10, 30, 50 and 70% halftone patches are

characteristic of highlight, midtone and shadow regions on the print. The 90% patch was eliminated

from the analysis, because the printed ink film could not be separated into individual features of

which to calculate the ink volume transferred.

Table 4.7: Combinations of parameters for which planar surface area and volume of printed dots was determined

Speed

150 ft/m in 300 ft/m in

Engagement Engagement

Colour Line ruling 1 thou 4 thou 1 thou 4 thou

Black 100 Ipi, 150 Ipi V V V V

Cyan, magenta, yellow 100 Ipi V V

In order to determine whether significant ink transfer deviations exist for the inks of different colour,

CMYK prints were compared for selected parameter combinations (Table 4.7). Figure 4.9 illustrates

the results for the Asahi prints representative for all plate materials (the results for the other plate

materials were as or even more consistent than the Asahi ones). The inks followed the same trends

in printed planar surface area and volume for the two combinations of printing parameters. The

slopes of the graphs are representative of the printed ink film thickness and were very consistent for

the respective parameter combination. Any deviations between the inks were within the range

expected under the given measurement uncertainty (see below) and through error propagation

resulting from the setup of the inks' separate printing units on the printing press. Therefore, it was

decided to concentrate the analysis of printed features on those produced with black ink.

The method of determining planar surface area and volume of printed dots based on WLI data was

in itself very consistent (standard errors below 1%). However, the measurement uncertainty

I 95
increased w ith do t size, printing speed and engagement. The small dots printed at low speed and

engagement had an approxim ate dome-shape which could easily be isolated from the substrate

background (Figure 4.10). W ith increasing dot size, printing speed and engagement, the dot shapes

became more irregular resulting in isolated islands of ink w ith in the dot area surrounded by very thin

residual ink film s. Here, it was extrem ely difficu lt to separate inked from unprinted areas. This

discrim ination had a large effect on surface area and volum e measured, but was m inimised by a

consistent approach to the manual separation of these features (refer to section 3.5.2).

♦ Black B Cyan A M agenta X Y e llo w

50000
45000
^ 40000
3 35000
| 30000
•% *
£ 25000
2 20000
W ♦
I 15000
> 10000
5000 ■ 4 * -------------------------------------------------------------------
0 1------------------ 1------------------ 1------------------ 1------------------ 1------------------ 1------------------ r

0 5000 10000 15000 20000 25000 30000 35000 40000

Planar surface area on print (p m 2)

(a)

♦ Black HCyan A M agenta X Y e llo w

50000
45000
zr 40000
3 35000
| 30000
£ 25000
° 20000
I 15000
> 10000 &
5000
0
0 5000 10000 15000 20000 25000 30000 35000 40000
Planar surface area on print (pm 2)
(b)

Figure 4.9: Comparison of ink transfer (planar surface area and volume of printed halftones) for different inks; printing
conditions.-(a) 150 ft/m in speed at 1 thou engagem ent and (b) 300 ft/m in speed at 4 thou engagem ent

I 96
 10% nominal area coverage 70% nominal area coverage

Speed: 150 ft/m in , engagement: 1 thou

inb film
0

s u b itr a te /^
discontinuities \ 1 /

Speed: 300 ft/m in , engagement: 4 thou

Figure 4.10: WLI images showing printed dot shapes dependent on dot size and printing conditions
(Asahi m aterial, 100 Ipi line ruling; 210x m agnification)

4.3.2 Optical Density

The optical densities (separated into th e ir CMYK components) were obtained fo r the halftone

patches fo r all the line rulings and printing param eter com binations. Eight consecutive sheets of

each experim ental number were measured and the mean optical density value for each halftone

patch calculated in order to minim ise effects of copy-to-copy variation. For this very large num ber of

measurements the Gretag M acbeth Spectrolino spectrophotom eter m ounted onto the

I 97
ColorScout XY measurement table was used, which allowed for consistent positioning of the

measurement head on each halftone patch. The analysis was not performed on the entire data set

acquired, but matched to the selection of halftones taken into account for the investigation of planar

surface area and volume of printed dots (Table 4.7).

4.4 Results and Discussions

4.4 .1 Characteristics of Plate Features

The plate features were characterised for planar surface area, to provide information on the transfer

of tihe design to the plate during the imaging process and the geometry of the feature top.

4.4.1.1 Planar Surface Area of Plate Features

The plate gain for all plate types is presented in Figure 4.11 to Figure 4.14. The two "round-topped"

plates, Asahi (Figure 4.11) and MacDermid (Figure 4.12), experienced similarly large coverage losses,

i.e. the dot sizes on the plates were smaller than specified in the digital artwork. The magnitude of

the loss was dependent on the dot size. Smaller dots, i.e. highlights and dots at higher line ruling,

had the most extensive losses relative to the nominal area coverage. This was due to the large ratio

of dot circumference to dot area, at which oxygen inhibition of the polymerisation process occurred.

At 200 Ipi the Asahi plate could establish dots down to 5% coverage. With decreasing line ruling the

lowest area coverage that could be held on the plate improved and reached 2% at 85 Ipi. In absolute

terms the smallest dot size corresponded to approximately 25 pm2 on the Asahi plate. The

MacDermid plate fared slightly better with all area coverages imaged at 85 and 100 Ipi. The lowest

area coverage at 200 Ipi was 3%, and the smallest absolute dot size was approximately 20 pm2. The

shape of the dot circumference was nearly circular for these two materials (Figure 4.15).

In stark contrast, the MacDermid Lux plate (Figure 4.13) achieved almost complete design fidelity.

The area coverages of the lower line rulings were within 1% of the nominal value. At higher line

rulings, plate gain of up to 4% occurred due to the addressability of the laser. As the dots reduce in

size, they approach the dimensions of the imaging laser beam, and it becomes increasingly difficult

to address the laser in such a precise fashion, that it only ablates the mask area required. Thus, a

slightly larger area than desired is ablated and imaged. The laser traces can be seen as steps around

the dot circumference which is roughly circular (Figure 4.15).

| 98
 20 40 60 80 100

-5 \ A
t *
I

X
A ♦ 85 Ipi
s
OP
N
I/I -10 s X
.Q A ■ lO O Ipi
c X
'ro
X A 120 Ipi
M
(U X 150 Ipi
m-15
CL
X 175 Ipi

• 200 Ipi

-20

-25
Nominal area coverage (%)

Figure 4.11: Plate gain for Asahi plate (standard digital imaging technology)

20 40 60 80 100

-5

♦
♦ 85 Ipi
VP
o'-
-10
■
S3 A ■ 100 Ipi
ro
A 120 Ipi
ro *
W J X
01
+-> X X 150 Ipi
-15 X A
o. X 175 Ipi
X
X
A • 200 Ipi
X A
-20

-25
Nominal area coverage (%)

Figure 4.12: Plate gain for M acDerm id plate (standard digital imaging technology)

| 99
 U i!! *
i 1
i

SP
O'" _ ♦ 85 Ipi
J3 -5
ro
■ 100 Ipi
c
'ro
bO A 120 Ipi
<U
S -10 X 150 Ipi
Q.
X 175 Ipi

• 200 Ipi
-15

-20
20 40 60 80 100
Nominal area coverage (%)

Figure 4.13: Plate for M acDerm id Lux plate (LUX "flat-top " imaging technology)

9
X
m < A
I J ♦ ♦ ♦

SP
¥ X
o'
U1 ♦ ♦ 85 Ipi
-Q
ro
X ■ 100 Ipi
c
’ro
euo
A I A 120 Ipi
♦
tu
*->
_ro -10 X 150 Ipi
Q.
X 175 Ipi

• 200 Ipi
-15

-20
20 40 60 80 100
Nominal area coverage (%)

Figure 4.14: Plate gain for Kodak plate (Flexcel NX "flat-top " imaging technology)

| 100
The trend in plate gain fo r the MacDermid Lux plate was not observed fo r the Kodak plate (Figure

4.14). The highlights fo r some line rulings exhibited coverage loss, w hile they exhibited gain for

others. In the m idtones and shadows almost all features suffered losses w ith a sharp reversal in

trend occurring between 70% to 80% nominal area coverage. This was attributed to the SQUAREspot

technology which ablates distinctive square pixel areas o f significant size (due to lower laser

resolution) from the mask layer. Once the halftone dots start joining on the printing plate, the

discrete laser pixels cannot resolve the contour of the holes between the dots anymore. Thus,

smaller holes than expected are produced and the coverage losses in the shadow halftones improve.

The SQUAREspot technology creates very d iffe re n t dot shapes. Smaller dots and non-printing areas

in the shadows approxim ate rectangular contours. The basic shape of the midtones is round, but the

square laser pixel combined w ith the low er resolution create a stepped pattern around the

circum ference o f the dot (Figure 4.15). All area coverages were established fo r the line rulings on the

Kodak and MacDermid Lux plates. The smallest absolute dot size was 195 pm 2 and 155 pm 2

respectively.

Asahi MacDermid

1 50

1 00

0 50

0 00

- -0 50

-1 00

Kodak MacDermid Lux

Figure 4.15: Comparison of dot shape for d ifferent plate materials and imaging technologies
(50% dot at 150 Ipi; 350x magnification)

| 101
4.4.1.2 Geometry o f Plate Features

The cupping depths fo r all plate types are presented in Figure 4.17 to Figure 4.20. The w ater-

washable Asahi plate, which theoretically falls into the category o f "round-topp ed" plates, was the

only material on which the full range o f dot geom etries could be observed (Figure 4.16). Dots below

10% nominal area coverage could not be investigated, because they were either not established on

the printing plate or too small to distinguish the top geom etry against the surface roughness.

Halftone dots smaller than approxim ately 8,000 pm 2 had pronounced convex top shapes (Figure

4.17), and the large positive values resulted from the increased surface roughness o f this plate

m aterial. The roughness necessitated the isolation o f the dot top data at a low er height in the

histogram and this revealed m ore points on the dot shoulders. Consequently, the height difference

between the dot shoulder and its apex increased. W ith increasing surface area, the convex shape

transitioned to concave.

—— Asahi Kodak M acDerm id M acDermid Lux

5
0 50 100 150 200
W id th (p m )

(a)

•Asahi ■Kodak M acDerm id ■MacDermid Lux

0.5

-0.5

i -1.5

-3.5

-4.5
0 50 100 150 200
W id th (p m )

(b)

Figure 4.16: Profiles of (a) 10% and (b) 50% nom inal area coverage dots at 100 Ipi line ruling

| 102
 * ♦ 85 Ipi
A
■ 100 Ipi

A 120 Ipi

X * x X 150 Ipi

X 175 Ipi

• 200 Ipi
♦ ♦ ♦

20000 40000 60000 80000

Planar surface area on plate (pm 2)

Figure 4.17: Cup depth for Asahi plate (standard digital imaging technology)

0.0

-0.5

- 1.0
x xxr
X X X
Aa
x
-1.5 * ~ A A A ----- ♦ 85 Ipi
E
- 2.0 ■ 100 Ipi
CL
<u A 120 Ipi
TJ -2.5
Q. ♦ ♦ ♦
X 150 Ipi
-3.0
X 175 Ipi
-3.5
• 200 Ipi

-4.0

-4.5
20000 40000 60000 80000
Planar surface area on plate(pm 2)

Figure 4.18: Cup depth for M acDerm id plate (standard digital imaging technology)

| 103
 0 .0

-0.5

- 1.0

-1.5
► rrrr ♦ 85 Ipi
- 2.0 ■ 100 Ipi
♦ ■
-o -2.5 A 120 Ipi
Q.
X 150 Ipi
5 -3.0
X 175 Ipi
-3.5
• 200 Ipi

-4.0

-4.5
20000 40000 60000 80000
Planar surface area on plate (pm 2)

Figure 4.19: Cup depth for M acDerm id Lux plate (LUX "flat-top " imaging technology)

- 1.0

-1.5
♦ 85 Ipi
♦
■ 100 Ipi

T3 -2.5 A 120 Ipi

Q.
3 X 150 Ipi
u
3.0
X 175 Ipi
-3.5
• 200 Ipi

-4.0

-4.5
20000 40000 60000 80000
Planar surface area on plate (pm 2)

Figure 4.20: Cup depth for Kodak plate (Flexcel NX "flat-top " imaging technology)

The MacDermid (Figure 4.18), MacDermid Lux (Figure 4.19) and Kodak plates (Figure 4.20) resulted

exclusively in concave dot tops w ith significant cupping. The dot shape and cup depth were

dependent on dot size.

| 104
The change in dot geom etry fo r the plates o f d iffe re nt washability and the fact th a t the extent of the

cupping was sim ilar fo r the MacDermid plates indicates th a t the top geom etry was independent of

the type of imaging process used, but subject to the chemical com position of the plate material.

Polymers in general are subject to shrinkage during curing. In an uncured polym er sheet only weak

interm olecular attraction forces exist which pull the molecules together up to a distance of about

10'1A. During polym erisation the carbon-carbon covalent double bonds at the end of molecules are

opened and tw o molecules joined to a chain through a new single covalent bond. This bond is much

stronger than the attraction forces and closes the gap between molecules to a distance o f about 1 A.

(Thompson, 1998) The effect is an increase of m aterial density and volum e shrinkage. Additional

shrinkage results from b etter packing o f the molecules during the polym erisation process which is

dependent on tem perature and steric interactions between molecules. (Broer, 1993; Hikmet,

Zwerver and Broer, 1992) Any variance in curing degree and resultant polym er shrinkage creates

material regions under tension or compression (Fuh, 1997).

The cupping o f halftone dots is the result o f this shrinkage during the plate-m aking process. The

follow ing shrinkage mechanism is suggested: A fter the main exposure of the plate m aterial, a single

printing dot consists mostly o f cured polym er (Figure 4.21a). It is enclosed by uncured polym er w ith

a buffer zone o f partially cured m aterial separating the tw o. The cured m aterial tries to shrink in

volume, but it is anchored in place by the uncured and partially cured polymer. This creates tension

w ithin the halftone dot. Once the uncured m aterial is washed out, the anchor is released and the

tension w ith in the dot is reduced by a contraction o f the cured polym er (Figure 4.21b). The

shrinkage is most notable in the centre of the dot, because its sides are restricted in th e ir m obility by

partially cured molecules. The effect m ight not have been observable for the highlights on the Asahi

plate, because the polym er at the dot to p was removed during washing out. The m orphology of

printing dots could not be explored fu rth e r w ith in the scope o f this work, thus a separate study is

recommended.

uncured

cured

Figure 4.21: Schematic of potential cause for cupping, (a) cured polym er under tension a fte r main exposure;
(b) release of tension through contraction a fte r washing out creates cupping

| 105
For all plate types, there is an almost linear relationship between cup depth and area coverage for

highlights and midtones up to the point at which the maximum negative cup depth occurs (Asahi -

Figure 4.17, MacDermid - Figure 4.18, MacDermid Lux - Figure 4.19, Kodak - Figure 4.20). The

cupping effect became increasingly visible with area coverage, potentially because attraction forces

in the dot sides had less effect at the dot centre. Beyond this point, i.e. for larger dots in the

midtones and shadows, the cupping depth decreased. This was attributed to the reduction in

intermediate dot height (refer to Figure 1.2 for definition). With increasing area coverage, these

larger dots progressively join at their base, thereby reducing the intermediate relief depth (Beynon,

2007) and with it the volume of the dot column for which a density differential in the material could

be created. The reduction in differential might result in less contraction of the dot.

Overall, the flattest dot tops were observed on the Asahi plate which is nominally a "round-topped"

plate, whereas the other plates had concave top geometries. Therefore, the differentiation of "flat-"

and "round-topped" plates by imaging technology and the terms themselves should be abandoned.

Although cupping depths of up to 3.5 pm may appear insignificant compared to 500 pm nominal

relief depth on the printing plate, the depression for the 50% Kodak dot in Figure 4.16 represents a

volume of about 31,400 pm3 if this cup could be filled with ink. If one square metre was imaged with

this type of dot, their combined cup volume could take up 15% of the nominal ink volume supplied

by the anilox.

Besides the very common measurands of dot height, shoulder angle and planar surface area, future

investigations of halftone dots should also always take into account the cupping. The method

introduced in this study has been aimed at the quantification of the cup depth. Cup width and

volume as well as the radius of the cup rim could be included as additional measurands.

Furthermore, the geometry of the cup shoulder (transitional region between dot top and shoulder)

is expected to have a significant effect on ink transfer, but WLI could not capture data points on the

shoulders of all plate types, as they were too steep (Figure 4.16).

| 106
4.4.2 Characteristics of Printed Features

4.4.2.1 Ink Distribution across Printed Features

The WLI and Keyence microscopic images were used to obtain information on the ink distribution

across printed halftones and solids. The emphasis was placed on nonuniformities and printing

defects, notably halos, uncovered areas and ink residue around the dot edges, and the following

sections set out potential mechanisms to explain their occurrence. The effects of engagement and

printing speed on the ink laydown were taken into account.

4.4.2.1.1 Halos
Halos, or the onset thereof, (Figure 4.22) were observed for all four plate sets, but the shape of the

halo was strongly dependent on plate type and printing conditions. It became more pronounced

with increasing speed and engagement. The defect created by the Asahi plate was approximately

circular and enclosed very little ink, thus it is better classed as a doughnut defect which is

characterised by a ring-shaped ink deposition. Since the Asahi plate was the only one of the four

investigated which featured convex dot tops for 10% halftones, it may be assumed that the

doughnut was created by complete displacement of ink underneath the top apex.

The Kodak, MacDermid and MacDermid Lux plates resulted in ring-shaped halos. Overlaying images

of the halos with an outline of the concave dot top on the printing plate showed that their diameters

were very similar (Figure 4.23). This implies an interaction between the halo defect and cupping by

which the cup brim of the printing dots seals off some ink within the cup. The ink on top of the cup

brim is displaced outward with the ink on the dot shoulders, thereby leaving a crescent-shaped ink-

free impression on the substrate. That the halo diameter was slightly smaller than the cupped zone

on the plate indicates that some additional deformation of the dot top has taken place. It is possible

that, as described by S. Hamblyn (2004), a depression in the dot top could have occurred at its

centre under hydrodynamic pressure causing the cup edges to fold in (Figure 4.24). In the actual

printing process a rolling contact between the plate and substrate takes place which further

complicates the deformation and deflection of the plate features, especially at higher engagements.

Confirmation of this mechanism requires further investigation through the application of numerical

simulation, and this needs to include the distribution of ink around the dot as substantial gain may

be attributed to this mechanism. A suggested modelling sequence will be outlined for an analogous

problem at the end of the next section.

| 107
 Experiment 1 Experiment 4 Experiment 9

150 ft/m in , 1 thou 150 ft/m in , 4 thou 300 ft/m in, 4 thou
Asahi
Kodak
MacDermid
MacDermid Lux

Figure 4.22: Images of printed dots at 10% nominal area coverage and 100 Ipi line ruling for different
printing conditions showing the evolution of doughnut and halo defects (250x m agnification)

| 108
 MacDermid MacDermid Lux Kodak

Figure 4.23: Comparison of printed halo w ith central diam eter (red circle) of dot cupping on printing plate
(10% nominal area coverage at 100 Ipi line ruling; printed at 300 ft/m in and 4 thou; 250x magnification)

Cupping edge

Substrate

Folding-in of
cupping edge

Figure 4.24: Schematic of potential interaction of cupping and halo, (a) concave dot top before impact,
and (b) folding-in of cupping edge during impact w ith substrate

4.4.2.1.2 Uncovered Areas

A second defect, which predom inantly occurred fo r larger features, was regions o f missing ink w ith in

the dot. Examples o f UCAs (at the centre o f the dot and protruding from the halo) can be seen fo r

50% dots of all plate types in Figure 4.25. The m inim um feature size at which the defect became

apparent differed between line rulings. It ranged from 30% nominal area coverage (23,500 pm 2

printed dot area) at 85 Ipi line ruling to 70% nominal area coverage (10,500 p m 2) at 200 Ipi. The

| 109
UCAs became more pronounced w ith area coverage. Some of the UCAs appeared to contain ink

residue, whereas others solely fram ed substrate which never seemed to have been in touch w ith ink.

This prom pted several potential causes fo r these UCAs. Since they manifested themselves only for

dots having an area in excess o f a certain size, surface depressions and nonuniform surface

chem istry o f the substrate (as suggested in the literature) were ruled to be unlikely reasons, as these

would affect the appearance of th e printed dot independent from its size.

Asahi MacDermid

Kodak MacDermid Lux

Figure 4.25: Images of printed dots compared for d ifferent dot top geom etries at 50% nominal area coverage and
100 Ipi line ruling showing UCAs and ink residue around the printed dot edges (printed at 300 ft/m in and 4 thou)

| 110
A nother possibility w ould be the form a tion of cavitation bubbles during the separation process of

plate and substrate. At very high separation speeds as experienced during the printing process the

stresses in the ink filam ent become very high. One mechanism to relieve the stresses is the

appearance and grow th of bubbles which provide extra volume in the bulk o f the liquid (Gay, 2002).

Nuclei o f bubbles are already present in form o f entrapped air, foam ing of ink during transport to

and w ith in the anilox chamber, or small contam inants on printing plate and substrate (Poivet et al.,

2003 and 2004). However, the size decrease of UCAs w ith increasing printing pressure, prom pts a

diffe re n t mechanism relating to the cupping o f the dot top.

Cupping edge

Substrate

Air
entrapm ent

Figure 4.26: Schematic of the potential interaction of cupping and air entrapm ent. Before the impact w ith the
substrate the ink does not fill the concave dot top entirely, but eith er (a) coats the contour of the dot top or
(b) sits atop the cupping edges. During impact (c) air is entrapped w ithin the dot cup by the cupping edge forming
a seal w ith the substrate, thus preventing ink transfer.

The concept o f the origin o f halos in the printed dot shown in Figure 4.24 can be expanded by

assuming th a t the concave dot top was not filled level w ith ink, but either coated w ith an ink layer

follow ing the contour of the cup (Figure 4.26a) or only inked atop the cupping edges (Figure 4.26b).

| 111
Due to the concave shape, on contact with the substrate a small volume of air is entrapped with the

ink in the cup (Figure 4.26c). The air bubble prevents the ink transfer between plate and substrate,

and an ink-free area remains in the print. This is supported by the observation that the size of the

UCAs decreased with increasing printing pressure which would cause a compression of the

entrapped air or its partial displacement outside the cup. Direct observations of air bubbles

entrapped between the ink film on the printing plate and the substrate were made by Naito et al.

(2006).

All plate materials exhibited deep cupping for nominal area coverages between 20% and 50%

(except for the Kodak plate were it occurred at about 70%), whereas the UCAs were more

pronounced for 50% dots and upwards. However, as the cup depth decreased for the midtones, the

dot width increased more rapidly, thereby providing a rising cup volume and potentially more space

for air to be entrapped in. This led to an increase in the size of UCAs in the higher midtones. For the

shadows, the appearance of UCAs did not decline much further which suggests that the shallow cup

depths became insignificant, and the creation of UCAs in the shadows was governed by the same

mechanisms as for UCAs in solids. Since shadow halftones feature joined surfaces, the cupping might

not have provided a seal for entrapping the air bubbles in the cup, and the bubbles could move

between printing features or be released completely.

The detailed mechanisms remain a research challenge for the future and their study might employ

numerical modelling. This will be a very complex task due to the multiple intricate steps involved in

the printing process each of which might have to be covered by one or several individual simulations

using varied numerical approaches, e.g. the finite element method for the deformation of the plate

feature against the substrate in the printing nip and the Lattice-Boltzmann method for the ink

behaviour during this deformation. A potential modelling sequence is depicted in Figure 4.27 and

outlined below.

The first models of the sequence should be concerned with the filling of the anilox cells with ink,

wiping action of the doctor blade and levelling of the ink in the cells thereafter, but these steps are

omitted in Figure 4.27. There the starting point for the first simulation is the printing dot and inked

anilox cells (a). The dot top is then brought into contact with the anilox which causes the polymer to

deform against the cells as well as ink to squeeze out of the cells and across the deformed dot top

(b). The output of this first model is the input to the second model in which the dot is disengaged

from the anilox and the recovery of the dot shape studied (c). Additionally, the ink film splitting has

to be taken into account. Ink is drawn out of the anilox cells forming filaments between the two

surfaces. If the ink withdrawal from the anilox cells is incomplete, the filaments will elongate and

| 112
eventually rupture leaving depositions of ink in the cells and on the dot top (d). The potential

levelling o f these ink depositions on the dot have to be simulated before the inked dot can be input

into the next series o f models describing the ink transfer process to the substrate (e). Analogous to

the inking process, the dot to p engages w ith the substrate and is thereby deform ed (f). The ink is

squeezed across the substrate and dot top. When in the next model the dot is removed from the

substrate (g), the deform ation of the dot top is released, the ink extends into filam ents between dot

and substrate and either full or partial ink transfer takes place (h). The final model step is the ink

levelling on the substrate (i). The size and geom etry o f the final printed dot in the model can be

compared directly to the actual d ot produced in the printing process in order to validate the

modelling sequence.

1 Anilox
(a ) I ! | C0 1

(d) (e) 1 (0 I I Substrate

Figure 4.27: Potential modelling sequence for the future study of printing defects such as halos and UCAs. The starting
point is the dot top and inked anilox cells (a). In the first m odel the dot is brought into contact w ith the anilox cells and
deform ed (b). Upon disengagement from the anilox, the dot recovers its shape and th e ink is drawn into filam ents (c).
A fter the filam ent rupture (d), ink film levelling has to be taken into account in a separate model, before the inked dot
can be input into the next simulation (e). The modelling sequence for the dot impacting w ith the substrate (f), dot
recovery and ink filam entation upon w ith draw al (g) as well as ink film splitting are analogous to the anilox sequence. Ink
film levelling on the substrate is the final m odel step (i).

| 113
 4.4.2.1.3 Ink Residue around Dot Edges

Ink residue (Figure 4.25) around the main body o f the printed dot can be observed fo r all halftones

independent o f size and plate type fo r which there are a num ber of potential causes. Firstly, the

I substrate m ight have been dew etted when the ink was retracted during ink film splitting. Ink

components, which adhered to the substrate surface or were trapped by it, resisted retraction and

formed residual films. Secondly, m onom ers and oligomers which constitute the ink's vehicle could

leach from the ink bulk and form a thin film around the main body o f ink after curing. The film

appears transparent if no pigments were removed from the ink bulk.

pigments

Ini? residue

Figure 4.28: Comparison of ink residue around yellow and black dots showing varying amounts of pigments
(M acDerm id, 50% nom inal area coverage at 100 Ipi line ruling; printed at 150 ft/m in and 1 thou; 450x m agnification)

The tw o mechanisms suggested are not m utually exclusive, but the extent to which each occurs

depends on the ink used (Figure 4.28). Yellow ink contains large pigments (visible in the microscopic

image around the edge o f the dot) which m ight o ffe r m ore resistance to dew etting and would also

hinder the ink vehicle in leaching. The result is a narrow er region o f ink residue. The nano-particles in

black ink potentially lead to the opposite behaviour, in th a t they w ithdraw freely during dew etting as

| 114
well as slightly leach with the ink vehicle. Thus wider regions of ink residue are created. For each

particular ink type, the width of the residue regions remains almost constant. Only a slight increase

in width occurs with area coverage (compare Figure 4.22 with Figure 4.25). That it is not just an

optical effect which makes the dots appear larger, can be confirmed with the WLI data. WLI is able

to capture the ink residue around the dot (refer to Figure 4.10), but it is difficult to quantify its film

thickness against the surface roughness, since it is in the region of 0.1 pm.

4A.2.2 Planar Surface Area of Printed Features

The planar surface area of the printed features was determined from WLI data and compared to that

on the plate. The effect of engagement and printing speed on the printed surface area were

investigated for line rulings of 100 Ipi and 150 Ipi, chosen because these are two of the most

frequently used line rulings in industry.

4.4.2.2.1 Line Ruling of 100 Ipi

The planar surface area of printed dots tends to be larger than the surface area of the features on

the plate. Bould (2001) and S. Hamblyn (2004) identified three mechanisms for this physical dot gain:

• ink spreading through squeezing action in the printing nip,

• top expansion and

• shoulder barrelling of the plate feature through deformation under pressure.

According to S. Hamblyn increasing the engagement enhances the effect of these mechanisms.

Smaller plate features are subject to the largest increases relative to their original size, because their

material volume is insufficient to stabilise them against deformation. They also have the larger

relative ink-carrying capacity which adds to the ink squeezing effects.

This gain process has been identified for conventional flexographic plates and was found to be

applicable to the digital plates used in this work. However, the magnitude of the effects differed

significantly for the four plate sets investigated. Following S. Hamblyn (2004), the physicaldot gain

was captured in the Area Transfer Ratio (ATR) as expressed in Equation 4.1.

Planar surface area on prin t (um2) Equation 4.1

Area T ra n s fe r Ratio (ATR) = —-------------- —
Planar surface area on plate (pm1)

At low printing speed and engagement the ATR was typically 1.15 for all plate materials (Figure

4.29). A clear difference occurs for the smallest dots (nominal area coverage of 10%, planar surface

| 115
areas below 10,000 (am2). Also there is a clear distinction between the "round-topp ed" Asahi and

MacDermid materials w here the ATR = 1.7 and the "fla t-to p p e d " Kodak and MacDermid Lux

materials fo r w hich the AIR = 1.2. This was attribu ted to the significant divergence in profile for the

10% dots (Figure 4.16a) compared to more sim ilar larger dots (example of 50% dot in Figure 4.16b).

The difference between the plates became even more apparent at the higher engagement o f 4 thou

(Figure 4.30). The larger dots approached ATRs o f 1.45, except fo r the MacDermid Lux plate which

remained low er at ATR = 1.2. The nominal 10% dots ranked Asahi (ATR = 2.5), MacDermid (2.1),

Kodak (1.6) and MacDermid Lux (1.3). The MacDermid Lux plate was the most stable fo r all dot sizes

under the d iffe re n t printing conditions, resulting in consistently low ATRs.

3.0

2.5

•£ 2.0
(0
DC
k_ ♦ Asahi
V
<4- 1 c
l/> 1 S
C
ro A ■ Kodak
♦
1.0 A M acDermid

X Lux
0.5

0.0
10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure 4.29: ATRs fo r dots of 100 Ipi line ruling (printing conditions: 150 ft/m in , 1 thou engagem ent)

3.0

2.5

■S 2.0
ro
DC
♦ Asahi
01
1.5
■ ■ Kodak

1.0 A M acDermid

X Lux
0.5

0.0
10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure 4.30: ATRs for dots of 100 Ipi line ruling (printing conditions: 150 ft/m in , 4 thou engagem ent)

| 116
S. Hamblyn (2004) showed th a t raising the printing speed decreases the ATR through the reduction

o f ink available fo r squeezing. However, in this study the higher speeds produced sim ilar or larger

surface areas indicating increased ink transfer (Figure 4.31). Rather than being a true effect o f the

change in printing speed, this has been attribu ted to the result o f ink buildup along the edges o f the

printing features over the course o f the printing trials. Image analysis of dots printed at 2 thou and

2 thou "Hysteresis" (experim ent numbers 2 and 5 in Table 4.5) confirm ed th a t the diam eter of the

printed dots had increased between the tw o nom inally identical experim ents, thereby indicating ink

buildup.

3.0
2.8

.2 2.4

3 1.6
< 1.4

■
rH rH rH rH rH
*1
o o O o o o o o O O O o O O o
LO LO O o LO m o o LO O O LO LO O O
rH ro no rH no no rH ro ro rH rH ro ro

Asahi Kodak M acDermid Lux

Figure 4.31: ATRs under different printing conditions for 10% dots of 100 Ipi line ruling
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

AWP-DEF AFP-DSH

Figure 4.32: Comparison o f magenta ink buildup on dots of 20% nom inal area coverage at 100 Ipi line ruling for tw o
d ifferent plate materials (red outline of dot top in the centre and dot shoulders surrounding it)

| 117
Particularly when using high printing forces, the ink is not only spread across the substrate during

impact, but also pushed from the dot top onto the dot shoulder, where it can build up with every

consecutive impression. Figure 4.32 shows microscopic images of the ink buildup on two different

printing plates (refer to section 5.1.1 for material specifications), both having completed print runs

of a similar number of copies in excess of 12,000 impressions (the scoping print trial was specifically

designed to investigate ink buildup, but is not included in this thesis). On the AWP-DEF material the

magenta ink buildup is located on the dot shoulder near the top, whereas the ink spilled all the way

down the dot shoulders on the AFP-DSH material and into the valleys between the dots. Ink

remaining around the circumference of the dot potentially acts like an increase of plate surface area,

thereby enlarging the printed dot.

The extent to which the ATRs increased with printing speed differed between the "round-" and "flat-

topped" plate materials. This indicates deviations in the progression of ink buildup and led to two

hypotheses to describe the mechanism. Firstly, the cupping of the "flat-topped" dots served as

restriction to the advancement of ink onto the dot shoulder, as the ink would have had to present

higher advancing contact angles first, before it would have been possible for it to overcome the

topographic obstacle (Oliver, Huh and Mason, 1977; Extrand and Moon, 2008). Secondly, the

cupping edges in contact with the substrate sealed off the ink contained within the cup. Thus less ink

was available to be squeezed out of the cup thereby reducing ink buildup and ATR. This second

model is also consistent with the mechanism leading to the creation of the halo defect (refer to

section 4.4.2.1.1).

The difference between the ATRs at different speeds, and thereby the ink buildup, was largest for

the Asahi plate. This was possibly due to the convex shape of the dot top (Figure 4.16). As discussed

in section 2.2.1.2, it is known that sharp surface features, e.g. as part of the feature geometry or

surface roughness, are able to pin advancing liquid fronts in place. The Asahi plate was characterised

by higher surface roughness than any of the other plates, with the potential associated impact on

the contact angle formed with the ink. However, this might not have been sufficient to pin the

contact line of the ink to the dot top. The accumulated effect was visible as increased ATR difference

for the two speeds. In the image sequence of printed Asahi dots (Figure 4.22) the enlargement in dot

diameter attributed to ink buildup is clearly visible.

The second largest difference in ATRs was observed for the MacDermid plate followed by the Kodak

and MacDermid Lux plate, which meant the latter experienced the least ink buildup. This is in

agreement with the manufacturers' claims that these plate materials require less frequent cleaning

during print runs (MacDermid Printing Solutions, 2012; Kodak, 2013) All three plates exhibited

| 118
p ro m in e n t concave cup shapes on the dot top (Figure 4.16) potentially prom oting the sealing-off and

pinning o f the ink. The slightly higher ATR values fo r Kodak over MacDermid Lux suggested th a t the

ragged edges o f the Kodak dot form ed a less tig h t seal w ith the substrate, where the ink was able to

squeeze out and onto the d o t shoulder. The in te rm itte n t and undulating appearance of the Kodak

d o t's halo (Figure 4.22) supports this.

Despite its sim ilarity to the Kodak and MacDermid Lux cup geometries, the MacDermid dot achieved

results closer to the convex Asahi dot. There are a num ber o f mechanisms th a t could lead to this,

and they require fu rth e r investigation using numerical modelling, geometrical and morphological

analysis th a t w ill allow the m aterial parameters to be explored systematically.

4.4.2.2.2 Line Ruling of 150 Ipi

The ATR results fo r the line ruling o f 150 Ipi closely follow ed the findings of 100 Ipi, but increased

m ore significantly w ith engagement and speed. Figure 4.33 provides an ATR comparison fo r the tw o

line rulings on the Asahi plate under d iffe re n t printing conditions which reflects the results fo r the

o th e r plate materials (refer to Appendix A.5).

u .u

5.0
♦ 1501 pi 150_1

£ 4.0 ■ 1501 pi 150_4

ro
DC
▲ 150lpi 300_1
a>
3.0
X 1501 pi 300_4

2.0 X lOOIpi 150_1

i f T .
* 2 1a • lOOIpi 150_4
1.0
+ lOOIpi 300_1

0.0 -lO O Ipi 300_4

0 10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure 4.33: Comparison of ATRs for line rulings of 150 Ipi and 100 Ipi for Asahi plate
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

The ATRs increased more significantly w ith engagement than speed. They also rose sharply fo r the

smallest dots, observable in particular fo r dots o f 10% nominal area coverage (approxim ately

880 pm 2 planar surface area on the plate). Dot dipping and picking up of additional ink volum e on

| 119
the dot shoulder was an unlikely cause, since all dots investigated had a larger diameter than the

anilox cell opening. However, different to its counterpart at lOOIpi, a 10% dot at 150 Ipi will only

make contact with three or four anilox cells at a time and could deform critically against the cell

walls, thereby releasing ink from the anilox cells with its shoulders. This effect became negligible for

dot sizes of 10,000 pm2 upwards, and the ATR curves levelled off.

4A.2.3 Volume of Printed Features

The volume of the printed features was determined from WLI data and compared to the planar

surface area of the plate and printed features. The ink volume transferred from the printing plate is

an indicator for the ink-carrying and ink-releasing capabilities of the plate material. The ink volume

deposited for a given printed area represents the average ink film thickness. The effects of

engagement and printing speed on volume were taken into account.

4.4.2.3.1 Line Ruling of 100 Ipi

The trends observed in the ATRs were approximately reproduced by the Volume Transfer Ratios

(VTR) from the printing plate (equation by S. Hamblyn, 2004):

Volume T ra n s fe r Ratio (VTR){pm) Equation 4.2

Volume on p rin t (pm 3)

Planar surface area on plate {pm2)

At lower engagement the VTR for all plate materials lies between 0.8 and 1.1, and there was no

particular order by plate material to the results. Only for the smallest dots of 10% nominal area

coverage did the standard digital plates from Asahi and MacDermid produce larger VTRs of 1.3 to

1.5. At the higher engagement of 4 thou and 150 ft/m in speed, all VTRs increased, especially for the

10% area coverage on the Asahi plate (Figure 4.34). The VTRs of the two "flat-topped" materials

remained similarly lower. The most extreme printing conditions (300 ft/m in and 4 thou) saw the

largest differences between the plate materials (Figure 4.35). The VTRs ranked from Asahi over

MacDermid and Kodak to MacDermid Lux. The latter was the only material which experienced a

decrease in VTR with changing printing conditions.

| 120
 2 .5

| 2.0

ro
* 1.5
♦ Asahi
♦
A ■ Kodak
1.0
OJ ▲ MacDermid
E
"5 0.5 X Lux
>

0.0
10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure 4.34: VTRs for dots of 100 Ipi line ruling (printing conditions: 150 ft/m in , 4 thou engagem ent)

2.5

2.0

1.5
♦ Asahi

■ Kodak
1.0
▲ M acDermid

o 0.5 X Lux

0.0
0 10000 20000 30000 40000 50000
Planar surface area on plate (p m 2)

Figure 4.35: VTRs for dots of 100 Ipi line ruling (printing conditions: 300 ft/m in , 4 thou engagem ent)

The sim ilarity o f the VTR trends to the ATR ones suggested th a t the ink volum e transferred was a

function of the planar surface area printed. This was confirm ed through a very high correlation

coefficient (R2= 0.97) for the ink voiume-area graph (Figure 4.36). The slope of the graph represents

the mean film thickness (FT) (Equation 4.3) and was approxim ately constant. M ost FT values were

between 0.7 pm and 0.9 pm w ith a mean FT=0.79 pm (Figure 4.37).

Volume on p r i n t ( p m 3) Equation 4.3

Mean f i l m thickness ( FT) (pm) =
P l a n a r s u r f a c e area on p r i n t ( p m 2)

| 121
This implies that, although the printed surface area itself was dependent on plate m aterial and

printing conditions, the film thickness was not. Therefore, the ink transfer would prim arily be

governed by the ink and substrate properties. This is in agreement w ith previous assumptions that

the surface characteristics o f the printing plate play a m inor role once the plate is inked, because

some ink would always remain on the printing plate after the impression, and the plate-ink contact

would turn into an ink-ink contact (Liu and Guthrie, 2003).

♦ Round B F Ia t

50000
E
a.
40000

Q.
C 30000
o
T3
O)
20000
c
kro- 10000
4-*
<D
E « * ■
_3
O 0
>
10000 20000 30000 40000 50000 60000
c
Planar surface area on print (pm 2)

Figure 4.36: Relationship of ink volum e transferred and planar surface area of printed dot at 100 Ipi line ruling (all plate
materials under all printing conditions included)

♦ Round B F Ia t

_ 0.9
I 0.8
X 0.7
o>
J 0.6
2 0.5
*->
E 0.4
0.3
0.2
0.1
0
10000 20000 30000 40000 50000 60000
Planar surface area on print (p m 2)

Figure 4.37: M ean film thickness for dots of 100 Ipi line ruling (all plate materials under all printing conditions included)

| 122
4.4.2.3.2 Line Ruling of 150 Ipi

Just like the ATR results, the VTR trends at 150 Ipi follow ed th e ir 100 Ipi counterpart closely. Figure

4.38 provides a VTR comparison fo r the Asahi plate which was representative o f all plate materials

(refer to Appendix A.6). The smallest dots produced by the Asahi and MacDermid plates (10%

nom inal area coverage, approxim ately 880 pm 2 planar surface area on the plate) resulted in

disproportio nally large VTRs. Analogous to the ATRs, this was attributed to the increased

susceptibility o f smaller dots to deform against the anilox and pick up additional ink w ith the dot

shoulders. The VTRs decreased significantly between 10% and 30% nominal area coverage due to

dot stabilisation against deform ation through th e ir own material size and volume as well as

increasing support by neighbouring dots. The larger dots produced sim ilar VTRs to dots o f the same

size at 100 Ipi.

4.5

_ 4.0

I 3.5 " ♦ 150lpi 150_1

■ 1501 pi 150_4
nj 3.0
oc
(U 2.5 A 1501 pi 300_1
>4-

g 2-° X 1501 pi 300_4

I—
1.5 X lOOIpi 150_1
E
2. 1.0
o
f *► • lOOIpi 150_4
>
0.5 + lOOIpi 300_1

0.0 -lO O Ipi 300_4

10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure 4.38: Comparison of VTRs for line rulings of 150 Ipi and 100 Ipi for Asahi plate
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

The data comparison o f the relationship o f ink volum e transferred to planar surface area printed

showed the sim ilarities between 150 Ipi and 100 Ipi line ruling (Figure 4.39). At 150 Ipi, the

correlation coefficient o f the tw o measurands was a very high at R2 = 0.98. The mean film thickness

(FT=0.76 |im ) was slightly low er than at 100 Ipi (0.79 (im) (Figure 4.40) which m ight be due to a

difference in the dot geometries. For dots o f sim ilar planar surface area, the shape o f the cupping

differed depending on the ir associated line ruling (Figure 4.41). At 100 Ipi the 10% dots had a

comparable planar surface area to the 30% dots at 150 Ipi, but th e ir cups were slightly deeper and

could therefo re potentially hold more ink.

| 123
Having explored the geom etry o f the printed dots, the next stage is to consider the impact on print

density, and this w ill be dealt w ith in the next section o f this chapter.

♦ 150 Ipi ■ 100 Ipi

_ 50000
E
r 40000

Q.
C 30000
o
-o
QJ
£ 20000
**-
i/t
c
£ ioooo
<u
E
i o
>
10000 20000 30000 40000 50000 60000
Planar surface area on print (pm 2)

Figure 4.39: Relationship of ink volume transferred and planar surface area of printed dots compared for 150 Ipi and
100 Ipi line ruling (all plate m aterials under all printing conditions included)

♦ 150 Ipi ■ 100 Ipi

i 0.8

10000 20000 30000 40000 50000 60000

Planar surface area on print (pm 2)

Figure 4.40: M ean film thickness compared for dots of 150 Ipi and 100 Ipi line ruling
(all plate materials under all printing conditions included)

| 124
 30% at 150 Ipi 10% at 100 Ipi

U)

-150 -100 -50 100 150

W idth (urn)

150 Asahi 150 Kodak 150 M acDermid 150 Lux

100 Asahi 100 Kodak 100 M acDerm id 100 Lux

Figure 4.41: Comparison of cup geom etry for similar dot sizes at d ifferent line rulings
(left: 30% nominal area coverage at 150 Ipi; right: 10% at 100 Ipi)

4.4.2.4 Optical Density o f Printed Features

The optical density of the printed features was determ ined by spectrophotom etry. This was

measured fo r the conditions th a t correspond to the physical measurements o f dot features. Thus it

was possible to explore the relationship between physical measurands and print density.

4,4.2.4.1 Line Ruling of 100 Ipi

For all p ate materials an increase in optical density w ith engagement and printing speed was

observed (Figure 4.42). The trends were sim ilar to those found fo r the printed area coverage

(compare section 4.4.2.2.1). Figure 4.43 shows the relationship of optical density to area coverage of

the p rinted features sorted by the diffe re n t imaging technologies and taking into account all printing

conditions. The very strong correlation indicates th a t optical density was directly dependent on

printed area coverage, which thereby explains the analogous effects o f engagement and speed on

optical density. It also supports the Murray-Davies halftone model which states that diffe re n t optical

densities are generated by changing area coverage. The largest deviation between plate types

occurred between 60% and 80% printed area coverage fo r all plate materials, and was attrib u te d to

the increasingly nonuniform ink distribution w ith dot size. The printing defects were slightly more

pronounced fo r "fla t-to p p e d " Kodak and MacDermid Lux plates which expressed itself in slightly

lower opt ical density.

| 125
 Kodak MacDermid

Figure 4.42: Optical density under d ifferent printing conditions for 70% dots of 100 Ipi line ruling
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

♦ Round B F Ia t

1.0
0.9
_ 0.8

i oj
ui 0 .6
♦
I 0.5
5 0.4
o. 0.3

0.1
0.0 -i-------------------------- 1-------------------------- 1-------------------------- r
0% 20% 40% 60% 80% 100%
Area coverage on print (%)

Figure 4.43: Relationship of optical density and area coverage on print for dots of 100 Ipi line ruling
sorted by imaging technology (all plate materials under all printing conditions included)

4.4.2.4.2 Line Ruling of 150 Ipi

The findings regarding optical density o f features printed at 150 Ipi line ruling confirm ed the results

discussed fo r 100 Ipi (Figure 4.44). However, no density difference could be distinguished between

"fla t-" and "rou nd -to pp ed" plate materials at 150 Ipi. This was attrib u te d to the more coherent ink

distribution o f these dots fo r all plate types as a result o f th e ir reduced size.

| 126
 ■ 1 0 0 Ip i ♦ 1 5 0 Ipi

1.0
0.9
_ 0.8

i oj
'wi 0.6
| 0.5
ro
o 0.4
1. 0.3
O
0.2
0.1
0.0
0% 20% 40% 60% 80% 100%
Area coverage on print (%)

Figure 4.44: Relationship of optical density and area coverage on print compared for 150 Ipi and 100 Ipi line ruling (all
plate materials under all printing conditions included)

4.4.2.5 Applicability and Comparison o f Halftone Models

Halftone models are used to calculate the area coverage o f a halftone patch from the optical density

(i.e. reflectance) measured on it. They are the most widely used tool to determ ine area coverage as

part of process control in industry. The previous section established th a t a very good correlation of

optical density and area coverage (prerequisite fo r the Murray-Davies halftone model) was observed

fo r the fo u r plate materials. This section compares the prediction o f this particular model w ith the

experim ental data. O ther models were also evaluated to determ ine w hether they are better able to

cope w ith the particular challenges posed by the printed features o f the different plate types. The

models are Yule-Nielsen, Expanded Murray-Davies, Noffke-Seymour and Beer's Law, and th e ir

descriptions are provided in section 2.3.1.

The equations o f the Expanded Murray-Davies and Noffke-Seymour models cannot be solved fo r the

area coverage in closed form , as is possible fo r the Murray-Davies model. Thus, all model equations

were rearranged to calculate the expected halftone reflectance produced by a given area coverage,

which was then compared to the actual reflectance. The model error was determ ined using Equation

4.4.

Equation 4.4
X i ( P a c t u a l,i ~ P m o d e l.i)
E rro r =
N
N
where factual is the actual reflectance measured by spectrophotom etry, /3modei is the reflectance

calculated from the respective model, and N is the num ber of points in the data set. The Yule-

| 127
Nielsen n-value as well as the parameters v and w in the Expanded Murray-Davies model were

optim ised fo r the best fit of model to actual data, i.e. the m inim um model error. The respective

values employed are provided w ith the error results below.

4.4.2.5.1 Line Ruling of 100 Ipi

Figure 4.45 illustrates the wide range o f modelled reflectance values at 100 Ipi line ruling obtained

from the diffe re n t halftone equations. The corresponding model errors are listed in Table 4.8 and

each model is discussed separately below.

♦ Actual ■ Murray-Davies A Expanded Murray-Davies

X Yule-Nielsen XBeer • N offke-Seym our

100 % ■ ----------------------------------------------------------------------------------------------------------------------------

90% -

80%

70% -

60% -
<U
u

S 50% -
v
H-
<u
DC

40% -

30% -

20% -

10% -

0% -
0% 20% 40% 60% 80% 100%
Area coverage on print (%)

Figure 4.45: Comparison of actual and calculated reflectance at 100 Ipi line ruling obtained from d ifferent halftone
models (all plate materials under all printing conditions included)

| 128
 Table 4.8: Comparison of model errors for different halftone models at 100 Ipi line ruling

Expanded Noffke-
Model Murray-Davies Yule-Nielsen Beer
Murray-Davies Seymour

Parameters - i;=0.5705; w=0 n=1.4877 - -

Error 0.0847 0.0462 0.0432 0.1077 0.1028

Asahi MacDermid

5 . *
0 te #

Kodak MacDermid Lux

Figure 4.46: Comparison of print defects included in solids produced by d ifferent plate materials
(100% nominal area coverage at 100 Ipi line ruling; printed at 150 ft/m in and 1 thou)

M u r r a y -D a v ie s M o del

The standard Murray-Davies model overestim ates the reflectance value fo r all halftone dots at

100 Ipi line ruling. The prim ary cause is the result o f the assumption th a t the solid reflectance, Psoiid>

which is at the heart o f the Murray-Davies equation, is obtained from a printed patch which features

100% nominal area coverage. However, independent of plate type all solid patches suffered from

| 129
considerable p rint defects in the form o f UCAs (Figure 4.46). The defects reduced area coverage and

increased reflectance of the reference solid which consequently increased all modelled halftone

reflectance values. Basing the Murray-Davies equation on such a flawed measurand would inevitably

lead to deviating results in halftone reflectance, unless all the corresponding area coverages in the

calculation dem onstrated sim ilar flaws. However, this was not the case.

The defects in the 50% and 70% halftones o f all plate materials, notably halos, residual ink film s and

UCAs, were com parable in appearance to the UCAs at 100% area coverage, and th e ir respective

reflectances increased in line w ith each other. Thus, actual and modelled reflectance show a low er

deviation. The highlights were characterised by very uniform ink film s w ith little or no defects which

resulted in reduced actual reflectance out of line w ith the reference reflectance provided by the

flaw ed 100% area coverage. Consequently, the actual and modelled reflectances deviate.

A dditionally, the midtones and shadows printed w ith "fla t-to p p e d " Kodak and MacDermid Lux plates

suffered from more irregular ink film s than those printed w ith standard digital plates. This rendered

them m ore sim ilar in appearance to the reference solid, th e ir reflectances aligned b etter and

reduced the model erro r (Figure 4.47).

♦ Actual round ■ Murray-Davies round A Actual fla t X Murray-Davies flat

100% H

90%

80%

70%

«
U 60%

S 50%
_a»
V04)-
“ 40%

30%

20%

10%

0%
0% 20% 40% 60% 80%
i
100%
Actual area coverage

Figure 4.47: Comparison of actual and calculated reflectance obtained from M urray-Davies model
for 100 Ipi line ruling sorted by imaging technology

| 130
B e e r ' s La w

Beer's Law represents the oth e r extrem e to the Murray-Davies Model by assuming th a t the printed

ink film spreads out com pletely and covers the entire substrate. The m ajority o f modelled

reflectance values significantly underestim ated the true values. The only exception was the values

calculated fo r highlights on standard digital plates which were overestim ated (Figure 4.48). The

overall model e rro r was the largest out o f all models investigated (Table 4.8). The underestim ate was

a direct result o f ignoring the substrate's reflectance in the equation, thereby producing overall

reflectance values which were too low.

♦ Actual round ■ Beer round A Actual fla t X Beer fla t

1 0 0 % | 1

q n°>£

a n 0/ .

7 n o / %

c n o t

cnoz
S T *
A C \o /
&
O fW ’* * % . * 1
j U /O

o n oz

*
1 n%

0 % - . ~r i i
1
0% 20% 40% 60% 80% 100%
A re a c o v e ra g e on p rin t (% )

Figure 4.48: Comparison of actual and calculated reflectance obtained from Beer's Law for 100 Ipi line ruling

The highlights achieved the best data fit, because th e ir printed structure resembled m ost closely the

theoretical dot on which Beer's Law was based. Unlike the oth e r models, this theoretical dot is not

characterised by the actual printed area coverage but the dot size on the printing plate. The model

presumes th a t all plate dots (independent o f size) carry an ink film of identical thickness, have very

sharp edges, and there is no ink spread or printing defects. The actual printed dots th a t got closest

| 131
to this description were the highlights. The printed structure of the larger dots increasingly deviated

from this ideal dot and the model error increased.

The modelled reflectance of the highlights on the standard digital plates (Asahi and MacDermid) was

higher than for the "flat-topped" plates (Kodak and MacDermid Lux), because their size on the

printing plate was significantly smaller. The real reflectance of these dots was reduced by additional

ink transfer from the ink buildup around the dot tops which created a proportionally larger printed

dot.

N o f f k e -S e y m o u r M odel

The Noffke-Seymour equation was designed to estimate the reflectance value of the full range of

halftone structures - from hard printed dots which did not spread out over the substrate to soft dots

which completely spread out. The true spectral reflectance is expected to lie somewhere between

these extremes, and the model has previously been used successfully to estimate it (Seymour and

Noffke, 2012). However, this success could not be repeated in this work, and the modelled

reflectances were an even larger overestimate (model error = 0.1028) than the Murray-Davies

results (model error = 0.0848).

The Noffke-Seymour model is subject to the same problem discussed for the Murray-Davies

equation above, that the reflectance of the reference solid is increased because of the UCAs it

contains. Since the halftones do not exhibit any defects to the same extend as the solid, i.e. their

reflectances have not increased in line with the solid reflectance, their modelled values are

overestimated. The problem aggravates in the Noffke-Seymour equation, because the solid

reflectance is increased further by the adjustment for ink spread. The model assumes that dot gain

in the prints is the result of ink spread across the substrate which is accompanied by a thinning of

the ink film. This in return increases the reflectance. However, it has been shown in section 4.4.2.3

that the mean ink film thickness is approximately constant for all halftones. Thus the real halftone

reflectance is solely governed by the area coverage, and the ink spread adjustment based on ink film

thickness in the Noffke-Seymour equation overestimates the modelled reflectances. The basic

concept behind the Noffke-Seymour model has potential, but the equation itself requires revision, as

in its current form it does not represent an improved alternative to the Murray-Davies equation.

| 132
Y u l e - N ie l s e n M odel a n d Ex p a n d e d M u r r a y - D a v ie s M odel

The parameters of the Yule-Nielsen and Expanded Murray-Davies models were optimised to achieve

the best fit of modelled to actual data. The two model errors were of similar magnitude (model error

Yule-Nielsen = 0.0462; Expanded Murray-Davies = 0.0432) and only half as large as the Murray-

Davies model error.

The improved fit was only achievable because of the retrospective parameter optimisation process

step which at the same time constitutes one of the disadvantages of these two models. However,

following Pearson's (1980) recommendation for the optimal Yule-Nielsen factor n = l.7, the resultant

model error of 0.0478 would only be slightly larger than the error of the optimal fit and still

represent an improvement over the Murray-Davies model.

The largest deviations in the fit occurred for the highlights (about 15% actual area coverage) where

both models overestimated the reflectance. If these dots were excluded from the data set, the

model errors could be further improved (Table 4.9). This was yet another indicator of the

dissimilarity of highlights compared to midtones and shadows. As has already been explained for the

Murray-Davies model, the solid reflectance entered into the models was increased due to the

printing patch containing defects. Modelled and actual reflectance values of printing dots with

similarly flawed structure were more likely to correspond to each other. This could be seen for

midtones and shadows in Figure 4.45. The more coherent highlights had lower actual reflectance

values which were not anticipated by the models.

Table 4.9: Comparison of model errors for Yule-Nielsen and Expanded Murray-Davies halftone models

Model Expanded Murray-Davies Yule-Nielsen

Data range incl. 10% dots excl. 10% dots incl. 10% dots excl. 10% dots

Parameters l?=0.5705; w=0 17=0.4711; w=0 n=1.4877 71=1.4101

Error 0.0462 0.0405 0.0432 0.0381

4.4.2.5.2 Line Ruling of 150 Ipi

At 150 Ipi line ruling the modelled reflectance values showed similar trends to the ones at 100 Ipi

line ruling (Figure 4.49). The Murray-Davies and Noffke-Seymour models produced the largest error

by significantly overestimating reflectance (Table 4.10). Beer's Law, the Expanded Murray-Davies

and Yule-Nielsen models resulted in good fits of modelled and actual reflectance. Compared to the

100 Ipi line ruling, Beer's Law was able to perform better with the 150 Ipi data set, because all

halftone dots retained a more consistent, less flawed structure approximating the sharp dot which

| 133
stands at the base of this model. Further, Beer's Law was the only model which resulted in

significant variations o f calculated reflectance values fo r "fla t-" and "round-topped" plate materials

(Figure 4.50). This was owed to the large deviation in the plate area coverage fo r d iffe re n t plate

types which was entered into the equation, whereas the printed dots were more similar in area

coverage and structure. The o th e r fou r models em ploying actual printed area coverage showed little

variance between plate types. It was particularly notable that the best fit of Beer's Law, the

Expanded Murray-Davies and Yule-Nielsen models occurred fo r the midtones and shadows, while

the reflectance o f the highlights (about 15% actual area coverage) was almost always overestim ated.

An exclusion of these dots led to an im provem ent in model error (Table 4.10), as previously

observed at 100 Ipi line ruling.

♦ Actual ■ Murray-Davies ▲ Expanded Murray-Davies

X Yule-N ielsen XBeer • Noffke-Seym our

100% m

90%

80%

70%

60%

50%
(u
DC

40%

30%

20%
%
i

10%

0%
0% 20% 40% 60% 80%
!
100 %
Area coverage on print (%)

Figure 4.49: Comparison of actual and calculated reflectance at 150 Ipi line ruling obtained from d ifferent halftone
models (all plate materials under all printing conditions included)
 Table 4.10: Comparison of m odel errors for d ifferent halftone models at 150 Ipi line ruling

Expanded Noffke-
Model Murray-Davies Yule-Nielsen Beer
Murray-Davies Seymour

incl. 10% dots

Parameters - 17=0.8223; w=0 n=1.7314 - -

Error 0.1039 0.0411 0.0393 0.0746 0.1396

excl. 10% dots

Parameters - 17=0.6761; w=0 71=1.6078 - -

Error 0.1034 0.0318 0.0296 0.0689 0.1412

♦ Actual round ■ Beer round A Actual fla t X Beer fla t

100% n

90%

80%

70%

60%
i
50%
<u X
DC
40% X
X
X f t
30%

20% *i* A

10%

0%
0% 20% 40% 60% 80%
»
100%
Area coverage on print (%)

Figure 4.50: Comparison of actual and calculated reflectance obtained from Beer's Law for 150 Ipi line ruling

The deviation in model fit fo r d iffe re nt halftone area coverages m ight have been based in the origin

o f the Murray-Davies model. The equation was developed to fit data obtained from the p h o to ­

engraving process which features particular dot shapes across the halftone scale (M urray, 1936).

There the highlight dots are m ostly round, m idtones are curvilinear quadrangles and shadows

| 135
framed round unprinted areas (Figure 4.51). Due to ink squeeze, the centre o f the dots printed

lighter than th e ir perim eter, but fo r the calculations an average density was assumed fo r the entire

dot. The do t structures observed in this trial were predom inantly round in the highlights and

midtones, and square in the shadows. Thus deviations caused by dot structure w ould have been

expected to occur in the midtones and shadows, whereas they are most severe in the highlights.

Figure 4.51: Printed dot structure of (a) highlights, (b) midtones and (c) shadows in photo-engraving w here darker areas
represent higher ink film thickness (M u rray, 1936)

The halftone scale is divided into tw o or more d iffe rent classes o f dot structures - sharp highlights

versus softer m idtones and shadows (specifically: m idtones w ith UCAs; midtones and shadows w ith

UCAs and residual ink films; solids w ith UCAs). No differentiatio n of this or a similar kind is currently

addressed in any halftone model. So far, alternative halftone models introduce expansions and

additional parameters to the basic Murray-Davies equation which apply equally to the entire

halftone scale. N either had it been investigated w hether it would be more appropriate to apply

diffe re n t halftones models or at least correction factors to selected ranges o f the halftone scale.

The issue is hereby recommended fo r future research, as it holds particular industrial relevance in

the form o f im proved models fo r more precise process control in graphic and functional device

printing. The starting point o f such research should be the validation of existing halftone models

(selective popular models are reviewed fo r example by Kang (1999), Wyble and Berns (2000)) w ith

em pirical data obtained from samples printed under different conditions in the m ajor printing

processes. This w ill allow the identification o f the halftone models which in general lack accuracy

and those which perform well under particular conditions or fo r certain parts o f the halftone scale.

The reasons fo r the perform ance then have to be identified, e.g. w hether the models sim ulate the

underlying physical principles inappropriately or a difference in halftone dot structure exists across

the halftone scale. In the form er case the model equations would have to be adjusted or new

| 136
equations developed, whereas in the latter case different halftone models might have to be applied

across the halftone scale. The new model solutions can then be validated against the empirical data

and further changes made where necessary.

4.4.3 Conclusions
The plates produced by standard digital imaging technologies suffer severe coverage losses. The

plates imaged with "flat-top" technologies almost achieved complete design fidelity, although the

lower imaging resolution and SQUAREspot laser technology led to deviations in Kodak's higher

midtones and shadows. Furthermore, the ragged edges of the Kodak dots, which translated into

severely destabilised edges of the printed dots, would leaves this plate less suitable for the

manufacturing of printed electronics (especially those orientated diagonally to the print direction)

which requires stable, uniform ink films.

The "flat-top" imaging technologies resulted in concave dot tops. This was the predominant shape

on all four plate types which renders top geometry independent of imaging technology. A new

method for the quantification of the top geometry by cup depth was introduced and should be

considered to be adopted within a standard for halftone characterisation. The Asahi and MacDermid

plates led to higher dot gain attributed to increased ink buildup on the plate and ink squeeze on the

prints, but achieved a more uniform ink distribution. The Kodak and MacDermid Lux plates resulted

in less dot gain, yet suffered from more pronounced UCAs and unstable edges in the printed dots.

This reduced the amount of ink transferred per unit area. The severity of printing defects increased

with area coverage, printing speed and engagement for all plate types. The effect was reduced

through the usage of finer line rulings on the printing plate.

For most of the criteria investigated a differentiation in results was possible between the two groups

of printing plates, standard digital Asahi and MacDermid versus "flat-topped" MacDermid Lux and

Kodak. This stood in contrast to the very similar cupping geometries of the latter three materials, a

distinction which might have arisen from differences in the geometry of the cupping edge and dot

shoulder. Due to the very high industrial relevance, this research question merits a further, more

focussed investigation.

The ink distribution of the printed dots influenced the optical density of the prints. More stable and

uniform printed dots increased the optical density. Therefore, the Asahi and MacDermid materials

had a slight advantage over Kodak and MacDermid Lux. However, the latter two exhibited slightly

improved agreement with the halftones models. The Murray-Davies model fitted the Kodak and

| 137
MacDermid Lux data better, because the defects found in their printed dots resembled the faults in

the reference solids more closely. All five models investigated were found lacking in accuracy. The

optimisation of equation parameters only achieved improved fit for subsets of data. One of the

reasons identified is the existence of different classes of dots within the halftone scale - a problem

which has not yet been addressed by any halftone model.

4.5 Closure
A study was conducted on the industrial-scale printing press T-Flex 508 to investigate the effect of

different imaging technologies on the output quality of graphic halftone printing. Two plate sets

imaged by standard digital technologies and two by "flat-top" technologies were included in the

trial. Since the predominant top geometry identified on all plates was concave, the terms "flat-

topped" and "round-topped" do not truly reflect the dot shape and are misleading. No imaging

technology or plate material investigated exhibited ultimate superiority with regards to print quality.

Each sample bore its particular advantages and disadvantages. The "flat-top" technologies produced

dots closer to target size on plate and print. Yet, it was the standard digital plates which resulted in

more stable, more uniform printed dots and higher optical density. Therefore, careful considerations

are required to match a plate type to the desired quality output of each printing application.

| 138
Chapter 5 Meso-Patterns on P rinting Plates
The previous chapter highlighted the recurring problem of defective UCAs in printed patches and its

im pact on process control through halftone models. Surface patterning o f printing plates is claimed

to prevent this type of defect (M iller and Zmetana, 2005) by pattern recesses functioning analogous

to anilox cells (Samworth, 2001 and 2009; Kodak, 2010). Allegedly, the volum e o f ink transferred is

raised and the ink film anchored m ore uniform ly to the substrate. This w ork sought to gain a deeper

insight into w hethe r surface patterning affects ink transfer and, if so, w hich mechanism governs it,

as w ell as how it depends on diffe re n t m aterial, printing and pattern parameters.

The investigation was divided into tw o parts: an initial study on a p rin ta b ility tester follow ed by a

confirm ation tria l on an industrial printing press. Eleven different surface pattern designs, influenced

by Stolt, Zwadlo and Rozzi's (2010) recom m endation o f regular geometries, were created in-house

and explored in this chapter. The nominal w idth of the pattern features was 50 pm and thereby

slightly larger than the commercial microcell patterns (< 25 pm) studied in the next chapter. For

easier diffe re n tia tio n o f the tw o d iffe re n t pattern scales, the form er w ill be referred to as meso- and

the latter as m icro-patterns (Figure 5.1).

M e s o -p a tte r n (5 0 p m ) M ic r o -p a tte r n (2 5 p m )

Figure 5.1: Schematic comparison of scale betw een meso- and m icro-patterns (200x magnification)

M eso -P a t ter n Effect u s in g a P r in t a b il it y T ester

The aim o f this study was to establish w heth er surface patterning applied to solid printing areas of a

flexographic printing plate has any effect on print quality. The print trial was carried out on the IGT-

F1 p rin ta b ility tester using DoE and six d iffe re n t input variables, one o f them being the pattern

design.

M eso -P at te r n Effect u s in g a n I n d u s t r ia l P r in t in g P ress

The study was repeated on the T-Flex 508 printing press. This served to confirm w h e th e r the findings

obtained from prints made w ith the IGT-F1 were transferable to industrial-size setups.

| 139
5.1 Meso-Pattern Effect using a Printability Tester
This study was designed for the systematic investigation of surface patterning in conjunction with

different printing parameters, substrate and plate materials. The main focus of this trial was to

explore whether patterning had an effect on mean print density and uniformity. Closer inspection of

the various types of printing defects produced provided the first information cn the mechanisms of

surface patterning in ink transfer.

5.1.1 Materials - Plates, Substrates and Inks

P la tes

The two plate materials employed were AFP-DSH and AWP-DEF (both by Asahi Photoproducts).

These were chosen because the former has a solvent-washable composition, whereas the latter is

water-washable. The two underlying polymer chemistries also had a direct effect on the plate

surface energy and roughness (Table 5.1).

Table 5.1: Selected properties of plate materials used for meso-pattern study on the IGT-F1

Property AFP-DSH AWP-DEF

Washability solvent water

Surface roughness, Ra 154 nm 257 nm

Surface roughness, Rz 1.2 pm 3.0 pm

Steady-state surface energy 45.0 mN/m 38.1 mN/m

Hardness* 69 Shore A 70 Shore A

Thickness 1.7 mm

* hardness according to publications by Asahi Photoproducts (2011 a n d 2 0 1 1 a )

The plates were imaged with 12 printing patches in a standard digital process by V&W Graphics

(Alford, UK). Every patch contained a different surface pattern each of which was designed to force

changes in the ink splitting process and hence the ink transfer to the substrate. One of these patches

remained without surface patterning and served as a solid area reference patch. The other eleven

patches contained meso-patterns based on a nominal feature size of 50 pm (Figure 5.2 and Table

5.2). The hole and polka dot patterns were based on midtone and shadow halftone screens

commonly found in flexography. The grid, hexagon and chequer patterns are stylised geometries

based on these halftone screens. They served to observe the effect of edge geometry on ink transfer

compared to the standard flexographic halftones. The intermittent tile and continuous line patterns

at different orientations have been further stylised and simplified. Together with the two chequer

| 140
patterns they were introduced to investigate w hether the directionality of patterns affects the ink

transfer. The actual area of pattern features on the printing plates was captured w ith WLI to confirm

com pliance w ith the nominal design and w ill be discussed in section 5.1.4.1.

Grid 50/50 Lines SO'150 lines 50/50 Lines 50/50 Holes Reference
diagonal horizontal vertical 95% 133lpi 100%

Hexagon 50/25 Hexagon 50/25 Polkadots 50/25 Tiles 100/50/25 Chequer 50/50 Chequer 50/50
neqative positive 45* tilt no hit

Figure 5.2: Illustrations of the 12 m eso-patterns and reference solid (76x m agnification)

Table 5.2: Specifications of m eso-patterns

Surface pattern Nominal pattern dimensions Nominal area coverage (%)

Reference solid - 100.0

95% 48 pm diam eter/191 pm pitch 95.0

Vertical lines

Horizontal lines 50 pm track/50 pm gap 50.0

Diagonal lines

Grid 50 pm by 50 pm w id th /5 0 pm track 75.0

Chequer (no tilt)

50 pm by 50 pm w idth 50.0
Chequer (45° tilt)

Tiles 100 pm by 50 pm w id th /2 5 pm gap 53.3

Polka dots 50 pm diam eter/100 pm pitch 40.3

Positive hexagons 50 pm w id th /2 5 pm gap 44.5

Negative hexagons 50 pm w id th /2 5 pm track 55.5

Substrates

The substrates chosen were a coated paper APCO (Papierfabrik Scheufelen, Lenningen, Germany),

denoted as APCO, and a highly absorbent paper Ford's Gold Medal (Arjo Wiggins Fine Papers,

Manchester, UK), denoted as Ford. The two substrates were chosen, because they have significant

differences in their material properties (Table 5.3).

Table 5.3: Material properties of substrates used for surface patterning studies on the IGT-F1

Property APCO Ford

Class coated uncoated, absorbent

Surface roughness, Ra 0.40 pm 4.16 pm

Surface roughness, Rz 4.89 pm 40.21 pm

Steady-state surface energy 35.6 mN/m _ *

* n o t possible to determ ine surface energy o f Ford due to very high absorbability

I nk

Two fundamentally different ink systems with a distinct contrast in their rheological properties were

chosen for the investigation of interaction effects with the surface patterns. The first ink was an in-

house formulation on a water base. It contained 6% low-molecular polyvinyl alcohol,0.5% surfactant

"Tween 20" and 5 mg/l dye "Crystal violet" dissolved in phosphate-buffered saline solution. Its

surface tension was 39.9 m N/m. Secondly, a commercial UV-curing ink FlexoCure Gemini (Flint

Group UK, Wrexham, UK), colour cyan, with a surface tension of 37.0 mN/m was employed. The UV-

curing ink is a typical flexographic ink, whereas the water-based formulation is representative of inks

used for biomedical functional device printing. Since the print analysis is based on the optical

properties of the inks, it would have been preferential to use two ink systems which achieve equal

colour strength in print. However, in practice this is very difficult to match, and the investigative

focus was placed on the rheological properties of the inks rather than their colour in order to cover

the wide range of flexographic applications.

The ink viscosity characterisation was performed on the Bohlin rheometer (instrument settings

according to Table 5.4 and Table 5.5). The water-based ink exhibited a nearly constant viscosity of

about 6-10'3 Pa-s at all strain rates and thereby can be considered to be Newtonian (Figure 5.3). The

viscosity of the UV-curing ink was on average 120 times higher than for the water-based ink, but at

around 0.8 Pa-s the ink could still be classed as a low-viscosity liquid. At higher shear rates it

demonstrated slight shear-thinning properties, but was mostly Newtonian over the rates explored.

| 142
 Table 5.4: Rheometer settings for ink pre-conditioning and measurement geometry (inks for IGT-F1 studies)

Pre-conditioning Measurement geometry

Type controlled rate Geometry cone/plate
Shear strain rate 0.01 s 1 Cone
Time 60s (water-based ink) 2 ° / 55 mm
Equilibrium time 10 (UV-curing ink) 4° / 4 0 mm

Temperature 25°C (isothermal) Gap size 150 pm

Table 5.5: Rheometer settings for ink viscosity determination (inks for IGT-F1 studies)

Viscosity measurement
Temperature 25°C (isothermal)

Type shear-controlled rate

Minimum shear strain rate 0.1 s'1
Maximum shear strain rate 100 s 1
Delay / integration time
(water-based ink) 5 s (decreasing time) / 5 s
(UV-curing ink) 10 s (decreasing time) / 1 0 s
Mode up and down
Samples 21
Repeats 2

UV-curing ink B Water-based ink

0.1

0.01

0.001
0.1 1 10 100
Shear strain rate (1/s)

Figure 5.3: Viscosity of the UV-curing and water-based inks used for surface patterning studies on the IGT-F1

| 143
5.1.2 Printing
The param etric study was based on an exploration o f six factors which were all adjustable at tw o

levels (Figure 5.4). The resulting full-factorial plan consisted of 64 individual experim ental points

which took into account all possible param eter com binations. This allowed studying the effects o f

the main factors and th e ir interactions on print density and uniform ity obtained through image

analysis. The three material parameters, namely printing plate, substrate and ink, have been

described above. Printing was executed under varying conditions on the IGT-F1. The process

parameters were anilox volum e, printing force and speed. The tw o levels fo r all parameters are

specified in Table 5.6. Based on a scoping study, the anilox force remained constant at 125 N, and

one inking revolution o f the anilox was used throughout. The water-based ink was air-dried after

printing, whereas the UV-ink was cured on the Jenton conveyor unit.

Material Process
parameters parameters

Printing plate

Substrate Anilox volume

Printing force

Printing speed

Figure 5.4: Parameters and th e ir potential interactions in m eso-pattern study on the IGT-F1

Table 5.6: Param eter levels for experim ental plan of m eso-pattern study on the IGT-F1

Factor Lower level Upper level

Plate material AWP-DEF AFP-DSH

Substrate APCO - coated paper Ford - uncoated, absorbent paper

Ink UV-curing water-based

Anilox volum e 8 cm 3/m 2 (350 Ipi) 12 cm3/m 2 (300 Ipi)

Printing force 50 N 150 N

Printing speed 0.2 m/s 0.8 m/s

Note: the analysis was conducted separately fo r each pattern geom etry

| 144
5.1.3 P rin t Characterisation

5.1.3.1 Definition o f Fingering Instabilities

This research has revealed fingering defects resulting from interfacial instabilities at the printing nip

exit in all prints from the m eso-pattern trials. Dependent on the scale o f interest, fingering can lead

to improved u nifo rm ity on a large scale w hile presenting transfer variation at the scale of the plate

patterning th a t may be classed as a defect. In this section the term s fo r the d ifferent fingering

instabilities are defined as they w ill be used in the fu rth e r print analysis.

viscous fingers dendrites beads

Figure 5.5: Classes of fingering instabilities as observed in the m eso-pattern trials. Ribs separate steady-state discrete
fingers. Finger tip bifurcation driven by external disturbance leads to viscous fingering. Stabilised bifurcation creates
m ore ordered dendritic defects. Beads are potentially superimposed on fingering defects as result of dew etting.
(7x m agnification)

The fingering defects can be divided into fou r classes: beads, dendrites, viscous fingers and ribs

(Figure 5.5). The latter three are classic examples o f hydrodynam ic instabilities of the Saffman-Taylor

type (Ben-Jacob et al., 1985; McCloud and Maher, 1995). As has been explained in section 2.3.2.3,

these instabilities can be created when a less viscous fluid invades a more viscous fluid along the

paths of least resistance. A more viscous fluid would be able to shift the less viscous fluid while

m aintaining a uniform interface. (Fernando, 2012) In this w ork the printing ink was displaced by the

| 145
surrounding air at the nip exit between the printing plate and substrate. Depending on the

conditons acting on the system, the ink meniscus in the nip might separate in a continuous stable

fashion which does not lead to any defects. However, it is more likely that the meniscus is

destabilised and air breaks through the interface nonuniformly.

The simplest pattern is the rib structure which can be interpreted as steady-state discrete fingers.

The ribs are distributed evenly retaining a definite shape. However, if the initial finger tip is

disturbed by external factors, it splits to form two fingers which propagate separately and in

different directions. Over time the finger tips continue bifurcating. Some fingers will eventually die

off, while others progress unhindered, creating the branched structure known as "viscous fingering".

Ben-Jacob and colleagues (1985) suggest that viscous fingers transition to dendrites if the bifurcation

stabilises under very large driving forces or in the presence of an anisotropic enclosing surface.

Dendrites are characterised by a linear main stem from which smaller side branches protrude at a

roughly normal angle.

The bead structure might not appear to fall into the category of Saffman-Taylor instabilities. But if a

digital image of it is transformed from greyscale to black-and-white by thresholding, the underlying

fingering instabilities (a mix of dendrites and viscous fingers) are revealed (Figure 5.6). The nature of

the beads themselves is not entirely clear, but the following mechanism is proposed. It might be

possible that the advancing air concentrates the ink at certain locations between adjacent fingers.

The additional amounts of ink form liquid bridges between the printing plate and substrate at the

nip exit where the liquid bridges are drawn out into filaments by the diverging cylinder surfaces.

While the filaments are being stretched, further ink is pulled into the filament by dewetting the

areas surrounding its contact zones at both cylinders. In this case, the dewetting might be aided by

the difference between the ink surface tension (39.9 mN/m) and the substrate surface energy

(35.6 mN/m) which indicates that the substrate is not easily wetted by the ink. Finally, the filament

breaks, and the impression of a bead remains in its place. The other defects might not exhibit this

feature, because the ink concentrations between fingers to not break up into filaments at the nip

exit, but the ink film splits continuously with the advancing meniscus. Potential future studies of the

issue could explore the effect of surface tension-surface energy combinations by using different test

liquids, substrates and plate materials.

| 146
 Figure 5.6: Revelation of fingering instabilities underneath bead pattern -
(a) original image, (b) thresholded black-and-w hite image (7x magnification)

5.1.3.2 Introduction to DoE Analysis Approach

Quantitatively the fingering instabilities are defined through the uniform ity param eter StDev

(explained in section 3.5.5.2). The effects o f m aterial and process parameters on StDev are analysed

using DoE. To gain a clear understanding o f the challenges in the analysis approach employed in this

and the follow ing studies, the effect of a single process param eter change on print un ifo rm ity and

density is dem onstrated on tw o samples produced by the chequer pattern at 45° tilt using the w ater-

based ink on the APCO substrate (Figure 5.7). Although the only process param eter altered was the

printing force, significant difference in print density, u n ifo rm ity and class o f fingering instability can

be observed in the image. This is reflected in the histogram o f GSLs, resultant MGSL and StDev.

Discounting any o th e r factors fo r this example, the effect (E) o f printing force on MGSL is calculated

as the difference in MGSLs observed at the tw o levels investigated. By increasing the force from 50N

to 150N, the MGSL decreases by 14.3 GSLs (EMGSL = -14.3) which is equivalent to a rise in p rint density

and can be observed as a darker image. The effect of printing force on StDev is also a negative one

(EstDev = -6.5) describing th a t the StDev decreases due to improved print u niform ity (also evident in

the images).

| 147
 Condition 1 Condition 2

Anilox volume 8 cm 3/m 2

Printing speed 0.8 m/s

Printing force 50N 150N

Image

(lOx magnified)

MGSL 134.9
m m 120.6

StDev 24.1 17.6

Histogram
-C ondition 1 •Condition 2

20000

15000

u 10000

50 100 150 200 250

Greyscale level

Figure 5.7: Comparison of optical and histogram data for tw o print samples produced by checker pattern at 45° tilt

The full-factorial experim ental plan this study was based on facilitated the exploration of 63 main

factor and interaction effects fo r each o f the surface patterns and the plain reference solid, revealing

detailed inform ation about the underlying processes of ink transfer. An exemplary data set o f effects

referring to the MGSL of the plain solid reference is provided in Appendix A.8.6 It illustrates th a t the

significance of effects decreases w ith increasing num ber of factors participating in an interaction.

Four-factor and higher interaction effects are usually negligible. Therefore, the discussion w ill be

lim ited to significant main and low er order interaction effects which are the most relevant to this

study.

‘ The com plete effect data fo r MGSL and StDev o f all surface patterns, as w ell as the u ncertainty o f th e effects
stating th e ir significance, is provided on th e CD-ROM attached to this thesis.

| 148
5.1.4 Results and Discussions

5.1.4.1 Plate Characterisation

The image analysis o f the plates' WLI data revealed significant losses in area coverage due to oxygen

inhibition of the polym erisation process. The AFP-DSH material (Figure 5.8) was seemingly more

affected w ith area coverage losses between 12% and 47% than the AWP-DEF plate (Figure 5.9) w ith

coverage losses betw een 4% and 27%. The considerable difference between the plates is attributed

to th e ir surface shape and roughness (Figure 5.10). The rougher AWP-DEF material scattered the

incident beam of light which allowed data capturing on the shoulders of the printing features that

w ould otherw ise be too steep to be measured. Furtherm ore, the feature tops were quite large and

dome-shaped w ith gently declining slopes which provided a larger area suitable fo r data capture. On

the sm oother AFP-DSH m aterial m inim al light scattering took place and only areas perpendicular to

and on the centreline o f the incident beam were recognised. The sharply truncated feature tops

were significantly sm aller than on the AFP-DSH plate and provided less area fo r data capture. The

steep feature shoulders became invisible to the microscope, thus data points missing from the

shoulder region impede the quantification of the precise shape of the dot and how it may contribute

to the surface from w hich ink is transferred.

Overall, all plate images differed strongly from the nominal artw ork fo r both plate materials. For

example, the corners o f the shapes fo r the grid, tile and hexagon patterns were missing. The

features became m ore rounded, and the gaps between them widened. A com plete list of images

(WLI data prepared fo r image analysis) can be found in Appendix A.7. These shapes, as measured,

were used in the p rin t analysis.

■ Actual area coverage ■ Coverage loss

100

Figure 5.8: Actual area coverage of meso-patterns on printing plate and coverage loss
compared to the artw ork for the AFP-DSH plate

| 149
 ■ A c tu a l a re a c o v e ra g e ■ C o v e ra g e loss

100
90
^ 80
'Z 70
jo 60
§j 50
8 40
S 30
< 20 46
10
0

Figure 5.9: Actual area coverage of meso-patterns on printing plate and coverage loss
compared to the artw ork for the AWP-DEF plate

Nominal AFP-DSH AWP-DEF

Figure 5.10: Comparison of nominal polka dot design and plate geom etry captured by WLI (red data points correspond
to raised areas from which w ill be printed; 76x pattern magnification; 180x sample magnification)

■ MGSL ■ StDev

Figure 5.11: M ain effect of ink type on MGSL and StDev compared for plain solid reference and all surface patterns

| 150
 UV-curing Anilox volume

8 cm /m 12 cm /m

Water-based Anilox volume

ink 8 cm /m 12 cm /m

Figure 5.12: Nonuniform ity of ink films printed w ith polka dot pattern using different inks and printing conditions
(AFP-DSH plate, APCO substrate, low printing speed). At low anilox volum e and printing force, areas of missing ink can
be observed for both ink types. For all the o th er printing conditions, the UV-curing ink achieves good area coverage
which is reflected in improved optical density and print uniform ity. On the o th er hand the uniform ity of w ater-based
prints decreases significantly due to the occurrence of fingering instabilities under high printing force and anilox volume.
(6x m agnification)

| 151
5.1.4.2 Effect o f In k Type

Figure 5.11 shows the main effect o f the ink type on MGSL and StDev. When compared w ith all other

effects th a t w ill be presented below, this showed the most significant influence on these o utput

param eters fo r the plain solid reference and all surface patterns. As expected by th e ir different

colour strengths, changing from the UV-curing to the water-based ink increased the MGSL by over

70 levels fo r most printing patches, because the rich blue pigm entation o f the UV-ink created a

much higher opacity than the transparent dye in the water-based ink. At the same tim e, the higher

StDev fo r the water-based ink indicated a drastic decrease in print u n iform ity resulting from printed

patterns w ith in the ink film s on all substrates which are associated w ith fingering instabilities (Figure

5.12).

5.1.4.3 Effect o f Substrate Type

Figure 5.13 shows the effect on MGSL and StDev when the substrate is changed from APCO to Ford.

Out of all the effects presented in this subchapter, the substrate type had the smallest influence on

MGSL, but was im p ortan t fo r StDev. The very high surface roughness of the Ford paper provided less

contact area fo r the ink and allowed only partial ink transfer to the raised parts o f the substrate

(Holmvall et al., 2011). The substrate recesses remained ink-free, thus increasing MGSL and StDev.

■ MGSL ■ StDev

40 14

-10
-20

-30
-40 -14

A° \<A \<p \<£ A A kcP

Figure 5.13: M ain effect of substrate type on MGSL and StDev compared for plain solid reference and all surface patterns
(In conjunction w ith the ink type, a strong negative interaction effect was observed (EMGSL = -9.7 and

EstDev = -5.1 on average). This implies that even though individually Ford paper and water-based ink

hampered print quality, togethe r these trends were m oderated. The reason fo r this was that the

llow-viscosity water-based ink was readily absorbed by the Ford paper, thereby spreading from the

peaks of the surface into the valleys and covering more o f the substrate. The more viscous UV-ink

failed to cover all o f substrate, as it remained on the top of the paper fibres.

The contribu tio n to the calculation o f the main substrate effect by those individual experim ents

which benefitted from the interaction between water-based ink and Ford substrate was very large

fo r certain surface patterns. Thus, an overall negative MGSL main effect was achieved fo r the plain

solid reference, 95% and horizontal line patterns. The u n iform ity of prints produced by the

horizontal line pattern im proved in particular w ith this ink-substrate com bination, because the

pronounced printing defects form ed by the water-based ink on APCO could not evolve on Ford. This

was a ttrib u te d to the fast ink absorption into the substrate and its unavailability for the form ation of

fingering instabilities at the printing nip exit.

5.1.4.4 Effect o f Plate M a te rial

The change in plate material from AWP-DEF to AFP-DSH had a small positive effect on MGSL, but

resulted in very large StDev effects fo r some surface patterns (Figure 5.14). As both plates had the

same thickness and hardness, these material properties were ruled out as causes fo r the difference.

■ MGSL ■ StDev

14

>
01
o
*->
CO
c
o
-10
it
-20 - -7

-30
-40 -14

Figure 5.14: M ain effect of plate type on MGSL and StDev com pared for plain solid reference and all surface patterns

| 153
jAs identified in Table 5.1, Figure 5.8 and Figure 5.9, the other properties to consider were surface

energy and roughness as well as area coverage. For successful ink transfer to take place in the

printing press, the printing plate should have a higher surface energy than the ink, but a lower

surface energy than the substrate (Thompson, 1998). Noting surface energy, the AFP-DSH was more

likely to be wetted with ink from the anilox roll. However, the surface energy was determined using

the static contact angle which does not reflect the highly dynamic conditions in the printing press.

Studies suggests that printing speed and pressure affect the influence of surface energy on ink

transfer (Chadov and Yakhnin, 1988; Liu and Shen, 2008), but this could not be explored further

within the scope of this work.

With regard to image transfer, the lower surface energy of AWP-DEF might have been compensated

for by its larger planar surface area and roughness. The plate characterisation (Figure 5.8 and Figure

5.9) showed higher area coverage for the AWP-DEF which would translate directly into a larger

contact area between the ink and substrate. More extensive ink film transfer in general results in

lower MGSL. The influence of the lack of area coverage on MGSL was observable for the line, tile,

polka dot and positive hexagon patterns on AFP-DSH in particular (Figure 5.14). This was

accompanied by a rise in StDev, because the smaller pattern features did not carry sufficient

amounts of ink and were spaced too far apart in order to create closed ink films by coalescence

resulting in print nonuniformity.

The increased surface roughness on AWP-DEF (Table 5.1) potentially served a dual purpose. Firstly, it

increased the contact area on which ink could be received from the anilox, thereby raising the

plate's ink-carrying capability and the amount of ink transferred in the printing process. Secondly, it

might have acted as a surface patterning on the nano-scale improving print uniformity by raising the

anisotropy of the printing surface. The anisotropy affected the transition between the regimes of

different defect dynamics thereby producing fingering with the AWP-DEF plate which remained

similar in scale and visually unobtrusive for the majority of printing conditions and surface patterns

(Figure 5.15). (Ben-Jacob et al., 1985; McCloud and Maher, 1995)

The AFP-DSH plate on the other hand produced a wide range of striking printing defects which had a

detrimental influence on StDev (refer to Figure 5.16 for an example of the chequer pattern at 45° tilt

and to Appendix A.9 for the chequer pattern without tilt together with the polka dot pattern which

are representative of the range of fingering patterns produced by all meso-patterns on the AFP-DSH

plate).

| 154
 Speed = 0.2 m/s Speed = 0.8 m/s

Force = 50 N Force = 150 N Force = 50 N Force = 150 N

— ---- -r—

ftnvtebui

fflR n i
Anilox volume: 8 cm3/ m

/ . t e n ''[ H i;
Anilox volume: 12 cm3/m 2

Figure 5.15: Chequer pattern at 45° tilt on AWP-DEF plate m aterial. Little change betw een fingering defect regimes w ith
printing conditions observed for w ater-based ink on APCO substrate (7x magnification)

| 155
 Speed = 0.2 m/s Speed = 0.8 m/s

Force = 50 N Force = 150 N Force = 50 N Force = 150 N

1
Anilox volume: 8 cm3/m

M il I
Anilox volume: 12 cm3/m 2

Figure 5.16: Chequer pattern at 45° tilt on AWP-DEF plate m aterial. Strong change betw een fingering defect regimes
w ith printing conditions observed for w ater-based ink on APCO substrate (7x magnification)

| 156
5.1.4.5 Effect o f Anilox Volume

The second largest influence on MGSL was the anilox volum e (Figure 5.17). More ink supplied by the

anilox led consistently to an increase in the am ount of ink transferred, and higher ink film

thicknesses appeared darker (Dam roth et al., 1996; Lindholm e ta l., 1996; Fouche and Blayo, 2001;

S. Hamblyn, 2004; Beynon, 2007; Cherry, 2007). Furtherm ore, w ith rising ink film thickness the

opacity o f the ink layer increased and the contrast between unevenly printed areas diminished.

The perform ance of the AFP-DSH plate was im proved by interaction w ith the higher anilox volum e

w hich was a ttribu ted to a p roportion ally larger increase in ink transfer compared w ith the AWP-DEF

plate. This m ight have been caused by ink m igration onto the feature shoulders which then acted

sim ilar to an increase in area coverage o f the plate. This could be investigated in future trials by

capturing images of the plate before and a fter printing, sim ilar to the work done during earlier

scoping trials (refer to Figure 4.32).

■ MGSL ■ StDev

40 -T r 14

Figure 5.17: M ain effect of anilox volum e on MGSL and StDev compared for plain solid reference and all surface patterns

5.1.4.6 Effect o f Printing Force

Increasing the print force im proved the ink transfer as indicated by decreasing MGSL (Figure 5.18)

(Bohan e ta l., 2003; Johnson e ta l., 2003; S. Hamblyn, 2004; Beynon, 2007; Cherry, 2007; Holmvall

and Uesaka, 2008a) and a num ber o f mechanisms can contribute to this. At higher force the printing

plate may mould more closely to the substrate and increased the contact area. At the same tim e fo r

a porous but incompressible substrate, potentially m ore ink is impressed into the substrate and

immobilised, becoming independent from the ink split ratio between the substrate and printing

| 157
plate. Furtherm ore, on Ford paper a negative interaction effect may take place, attributed to the

compression and sm oothing o f the substrate under higher force. This w ill not only increase the

contact area, but also prom ote the capillary action to take up more ink into the bulk of the

substrate.

MGSL ■ StDev

40 14
30
20 >
<D
10 o

0
-10 HJ <u
!fc
-20
-30
-40 -14

Figure 5.18: M ain effect of printing force on MGSL and StDev compared for plain solid reference and all surface patterns

W hile noting the com m ents in the paragraph above, the force increase aided p rint uniform ity in

several ways. Firstly, the rise in contact area decreased the portion of non-printed substrate.

Secondly, lateral ink flo w and coalescence to closed ink film s was encouraged. However, the printing

force affects the final ink splitting process and manifests itself in various types o f printed patterns, in

particular fo r prints w ith water-based ink (Figure 5.12).

It is possible th a t the reduction in nip height under higher printing force affects the hydrodynamic

stability o f the ink meniscus at the nip exit. A thinner gap is thought to increase the frequency at

which fingering patterns occur (Vo(3, 2002; Bornemann; 2013) which w ill result in a perceivable

im provem ent in p rin t uniform ity. Although a reduction in StDev was observed fo r higher printing

forces in this trial, a frequency analysis o f the prints to confirm the correlation o f printing force,

finger frequency and p rin t uniform ity was not part of this work. However, the prints indicate that

the uniform ity im provem ent results from a change in printed pattern type w ith printing force rather

than the increase in finger frequency (refer to Figure 5.16).

5.1.4.7 Effect o f P rin ting Speed

Previous investigations found th a t the ink transfer decreased at higher speeds (Damroth et al., 1996;

Fouche and Blayo, 2001; Johnson e ta l., 2003; S. Hamblyn, 2004; Cherry, 2007). This was attrib u te d

to a hampered filling o f the anilox cells w ith ink and a less favourable ink splitting ratio between the

plate and substrate. In this investigation the increase in printing speed had a negative effect in all

but tw o cases (Figure 5.19), im plying th a t m ore ink was transferred.

■ MGSL ■ StDev

Figure 5.19: M ain effect of printing speed on MGSL and StDev com pared for plain solid reference and all surface patterns

The same effect was observed in the trial on the IGT-F1 discussed in section 6.1, as well as by Quinn

(1997) and Olsson (2007). On all three occasions p rin ta b ility testers were used. This suggests th a t

the particular construction o f the testers (open doctoring systems) causes a noise factor which

influences the calculated speed effect on MGSL and StDev. Through hydrodynam ic action, it is

possible th a t the do ctor blade was deflected during rotation o f the anilox cylinder, where increasing

printing speed leads to higher force levels between the doctor blade and anilox. This w ill in return

result in a gap change through blade deflection, thereby allowing more ink to pass under the doctor

blade. The availability o f diffe re n t levels o f ink at the printing nip exit w ill im pact on the critical

capillary num ber (Equation 2.11) and the ink splitting process which may create d iffe re n t fingering

patterns as shown in Figure 5.16 and Appendix A.8. The change in printed pattern characteristics

was in agreement w ith the literatu re th a t suggests an increasing dom inant frequency o f fingering

instabilities at higher speed (Vo(3, 2002; Bornemann, 2013). In the prints fro m this trial, smaller

pattern structures were also observed w ith increasing speed which led to a more homogeneous

appearance recorded as a reduction in StDev.

| 159
An improvement in print density and uniformity with speed was more notable for the water-based

ink. This was attributed to its low viscosity which may in turn allow better filling of the anilox cells

even at higher speed. The more viscous UV-ink requires more time to invade and evacuate the anilox

cells, thereby performing less well at increased speed. However, further work is required to gain a

clear insight into the mechanisms that are present.

The combinations of printing conditions (anilox volume, printing force and speed) lead to distinctive

regimes for the formation of different types of fingering instabilities. The sensitivity of fingering

defects to printing conditions is discussed here using the example of the chequer pattern at 45° tilt

printed with water-based ink on the APCO substrate using the AFP-DSH plate material (Figure 5.16).

At low printing force and anilox volume, the printing speed had little effect on the type of fingering

defect (beads in both cases), but increased the finger frequency. At higher printing force, two

completely different classes of fingering instabilities evolved dependent on printing speed. Low

speed led to dendrites, whereas high speed resulted in ribs (the same type of pattern was observed

for low and high anilox volume). Again the finger frequency increased with speed. The anilox volume

had no effect at higher printing force, but led to a gradual transition between defect regimes (Sinha

and Tarafdar, 2009) for the prints produced at low printing force and high anilox volume. The tops

and the bottoms of the printed patches still exhibit beads (or at least traces thereof), while the

centre already shows the new patterns of dendrites or ribs which are characteristic of the regimes

otherwise observed at higher printing force. This suggests that the regime transition is force-driven

and the anilox volume affects the local pressure in the printing nip. Small variations in the printing

conditions have a large impact on the fingering instability created.

5.1.4.8 Effect o f Surface Patterning

Compared to the material and process parameters, the surface patterns had a more subtle effect on

the values of MGSL and StDev (Figure 5.20). With the exception of the line patterns which hampered

MGSL and StDev severely, all surface patterns increased print density, but aggravated uniformity. It

has been suggested that surface patterns act similar to anilox rolls in carrying additional ink in the

pattern recesses (Samworth, 2001). This would increase the ink volume supplied compared to the

plain solid reference. Differences were observed for UV-curing and water-based ink.

| 160
 ■ M G SL ■ S tD e v

40

30
20
10
to
0
-10
-20

-30
-40 -14

Figure 5.20: M ain effect of surface patterning on MGSL and StDev

for all surface patterns compared to the plain solid reference

Images of the prints produced w ith UV-curing ink at low printing pressure and low anilox volume

show th a t partial ink films w ith individual pattern features have been reproduced (Figure 5.12).

Increasing the anilox volum e and printing pressure respectively leads to an increase in closed film

area which suggests tw o d iffe re nt mechanisms. The first proposed mechanisms sees the larger

anilox volum e supply more ink causing flooding o f the pattern recesses w ith ink and consequently

additional ink transfer from non-printing areas on the plate. That would mean pattern recesses act

like anilox cells under certain process conditions.

The second proposed mechanism is instigated by the printing force. If the UV-ink is only deposited

on the printing features, but not in the recesses, it is necessary fo r the form ation of closed,

homogenous printed areas to squeeze the ink out from underneath the features and into the gap

between where it coalesces w ith the adjacent ink films. If the gap between features is too wide or

the ink volum e available not large enough, the ink squeeze is insufficient. Consequently, the ink films

do not connect and the substrate remains visible, disturbing print uniform ity. If the printing force is

too high, all the ink m ight be squeezed out from under the pattern feature, thereby creating a lighter

impression in its place on the print. The hypothesis o f the second mechanism is supported by the

MGSL and StDev data from the individual printing experiments. It shows th a t the surface patterns

produce lighter and less uniform prints compared to the solid reference under all printing

conditions; at best the prints are similar.

| 161
The print uniformity of samples produced using water-based ink show distinct fingering patterns for

all conditions. At low anilox volume and printing force individual features can be discerned in the

prints (Figure 5.12). The finger pitch is approximately four times as large as the feature pitch on the

printing plate, and a potential mechanism is ink film dewetting followed by coalescence rather than

the actual ink transfer from the plate features. The ink surface tension and substrate surface energy

also point to this. With increasing printing force and anilox volume, the ink film coalesces. Although

this reduced the nonuniformity arising from visible substrate, it also facilitated the evolution of

striking fingering instabilities.

Under all printing conditions, higher MGSLs were achieved using surface patterning compared to the

plain reference solid which implies improved ink transfer by pattern recesses acting analogous to

anilox cells. This could prospectively be corroborated by capturing WLI data of the printing plate

between inking and printing, as well as after printing, since this would provide information on

whether the pattern recesses are filled completely or only partially, and how much of the ink is

released from the recesses during printing. Furthermore, this raises a number of questions

concerning the detailed mechanisms of ink transfer to and from the pattern patches themselves

which should be explored in the future in order to gain further detailed insight.

The line patterns achieved the worst effects on MGSL and StDev, because they failed to create

closed ink films in the prints under most parameter combinations. This suggests that neither the UV-

curing nor the water-based ink fully occupied the recesses in these patterns; even the highest

volume of ink supplied by the anilox being insufficient to flood the gap between the elongated cell

pattern entirely (Samworth, 2001).

In summary, patterning has a small positive effect on MGSL for prints produced with UV-ink,

whereas the effect is much larger and negative for prints produced with water-based ink, because

the surface patterns act analogous to anilox cells for this ink type. Thus, the net effect of the entire

experimental plan is negative (Figure 5.20). The print nonuniformity arising from the partial ink

laydown is improved in conjunction with higher anilox volume and printing force (both are negative

2-factor interaction effects), but worsened through the emergence of distinctive printing defects,

when all three factors interact (positive 3-factor interaction effect). Imaging the patterns on the

AWP-DEF plate material had a negative effect on MGSL and StDev which was attributed to an

increase in ink supply by the material leading to better coalescence of individually printed areas.

There was a positive interaction effect between the surface pattern and the substrate. As already

explained for the main effect of the substrate, the rougher surface of the Ford paper leads to a

| 162
 smaller contact area for ink transfer, increasing MGSL. This influence was amplified with decreasing

area coverage by the surface patterns.

Although the main factor effect demonstrates that surface patterning has an adverse influence on

print uniformity, by taking into account all interaction effects, homogeneity improvements can be

achieved for almost any combination of material and process parameters through the use of surface

patterning. However, without prior tests it is currently not possible to predict which surface pattern

has the most beneficial impact at a given combination of parameters. For example, if APCO substrate

is printed with an AWP-DEP plate and water-based ink at low anilox volume and printing pressure,

then the grid pattern results in the best print uniformity at lower speed (StDev = 10.69 compared to

: reference StDev = 11.32), but the tile pattern is better at higher speed (StDev = 9.38 compared to

] reference StDev = 17.57). The latter is a significant improvement, while the former is a marginal

change which might be accompanied by a disadvantageous shift in the type of fingering pattern

produced. Furthermore, the surface pattern which improves print uniformity might not be the one

to result in best optical density at the same time. Using the exemplary process parameters, the

highest density would be produced by the polka dot pattern at the lower speed (MGSL= 144.78

compared to reference MGSL = 182.07) and at higher speed (MGSL = 100.69 compared to reference

MGSL = 136.23).

' The effect of surface patterning on printed patterns is illustrated on a single print of the water-based

ink on APCO substrate in Figure 5.21. Excepting the vertical and diagonal line patterns which reflect

the image of the printing plate, the reference solid and all the other surface patterning resulted in

different classes of fingering patterns. The reference solid suffered from viscous fingering which was

also the dominant defect class for the chequer pattern at 90°, patterns made up of holes, tiles,

positive hexagons, negative hexagons and horizontal lines; the latter two adorned with beads. The

images strongly suggest that the orientation of the surface patterning has a direct influence on the

propagation of the instabilities. Although the viscous fingers roughly followed the direction of the

recesses on the printing plate when bifurcating, the actual angle of propagation was usually smaller

than the one on the printing plate (see the tile and chequer pattern without tilt in Figure 5.21).

| 163
 Reference solid Holes

I<1 til )
If!

Vertical lines Diagonal lines

s
^ v i ' ♦ * v l
%%A \

§
Horizontal lines Grid

n ■m rm %* ;

W - n'
|hu\|hy
Chequer (no tilt) Chequer (45° tilt)
* ff
1 ii f 1141* *i
m m m m I ififfB l# t
M lk*J
I If -Nt i . f i I *1+5’-»*: *■" (§9
IfW
f W I J
tW
lT I Vi f ; *
lp
| \ ^
I H
IS llM IH M 1M
I l'tiul m¥..- I;m V jf& f I ni
f
f *1 % lr^Hf fs* I f l
Figure 5.21: Effect of the m eso-patterns on fingering dem onstrated on the example of a single set of printing conditions
(w ater-based ink on APCO substrate, AWP-DEF plate m aterial, speed = 0.2 m /s, force = 50 N, anilox volum e = 8 cm3/ m 2)
(76 .4x pattern magnification; 6.5x sample m agnification) (continued on next page)

| 164
 Polka dots

n >v > d Y :iW .!/**■i f f

Positive hexagons Negative hexagons

i |
w S f i v i w iV

Figure 5.21: (continued from previous page)

W hen the pattern orientation was changed to align the non-image areas o f the printing plate w ith

the printing direction, the finger propagation seemed more contained and linear. The rib instabilities

fo r the grid and chequer pattern at 45° tilt are an example fo r this. The orientation of the polka dot

pattern also created less obstructed channels in the p rint direction which resulted in mostly linear

main fingers. However, the gaps to the left and right o f these channels seemed influential enough to

slightly destabilise the main finger and give rise to short side branches and occasional bifurcation.

That these systems are very sensitive to small changes in conditions can also be seen by comparing

the polka dot and positive hexagon patterns. They are nom inally very sim ilar and strongly resemble

each oth e r on the printing plate (the sharp corners of the hexagons were not reproduced on the

plate). Nevertheless, the positive hexagons were the slightly larger printing features and protruded

m inim ally more into the plate channels orientated in the print direction. This small difference was

sufficient to cause the transition from dendritic ribs to branched viscous fingers.

Overall the patterns seem to be able to affect the frequency of the fingering instabilities in the

prints, although more detailed evaluation is required, fo r example using one- and two-dim ensional

frequency analysis. Sometimes the shift in frequency was accompanied by stabilisation towards ribs,

som etim es by changes towards appearances tha t are even more branched. However, the latter was

not necessarily a disadvantage and on occasion resulted in at least as uniform an impression as the

reference solid (see the tile pattern). That this is also beneficial fo r functional device printing was

| 165
concluded by Reuter et al. (2007): if the instabilities cannot be avoided, then the parameters should

be chosen towards increasing frequency and branching of viscous fingering, as this will improve the

overall uniformity of the ink film transferred.

5.1.5 Conclusions
Eleven meso-patterns were investigated under a variety of printing conditions on the IGT-F1 to

determine whether they are able to alter print density and uniformity compared to a plain reference

solid. This was found to be the case and governed by two different ink transfer mechanisms

dependent on the ink type used. The dual pattern functionality in flexographic printing has not been

documented in literature previously:-

• Lower-viscosity water-based ink floods the recesses of the surface patterns analogous to

anilox cells, thereby increasing the ink-carrying capacity of the printing plate and the amount

of ink transferred to the substrate. Fingering instabilities in the ink transfer process create

striking printed patterns. The types of fingering patterns produced and their frequency could

be controlled by pattern geometry and orientation, as well as the process parameters.

Improvements in print uniformity resulted from the increase in defect frequency.

• Higher-viscosity UV-curing ink is more likely to be held on the pattern features. The creation

of closed ink films has to be achieved through the mechanism of ink squeeze under pressure

in the printing nip causing the coalescence of neighbouring printed features.

The best pattern for print uniformity did not necessarily correlate with the one for highest print

density, and the print quality was highly dependent on the combination of surface patterning,

material parameters and printing conditions. Material and process parameters had a larger influence

on the print quality than the surface patterns. All the parameters have to be well-matched,

otherwise the process output deteriorates.

| 166
 5.2 Meso-Pattern Effect using an Industrial Printing Press
The study using the IGT-F1 demonstrated the potential of surface patterning to improve print

.quality. Although the literature suggests that trials performed on certain printability testers are

representative of industrial-scale printing (Aspler, 2004), there is also substantial experiential

evidence that printing on full-scale equipment will lead to improved results. Thus the study in this

subchapter investigated whether the same effects of surface patterning on ink transfer and

distribution could be observed if the experiment was scaled up to near industrial conditions.

5.2.1 Materials - Plate, Substrates and Ink

The same plate materials (AFP-DSH and AWP-DEF; Table 5.1) and surface patterns (Figure 5.2) were

1 used. The 12 patterns were organised in a band and seven such bands were repeated next to each

other to make up the final design of the printing plate (Figure 5.22). Each band was later printed

with a different anilox volume. The actual feature sizes were determined from WLI data and

presented in section 5.1.4.1.

Details of the materials (substrate Rayoface WPA59 by Innovia Films and the magenta UV-curing ink

of the series Solarflex Nova SL Pro DK03 by SunChemical) are provided in section 4.1.

5.2.2 Printing and Print Characterisation

The parametric study was based on six factors, namely surface pattern, plate material, mounting

tape, anilox volume, plate-substrate engagement and printing speed (Figure 5.23), which were all

investigated at two levels (Table 5.7). The first mounting tape, 1015 by 3M , is suitable for printing

solids and halftones on the same printing plate. The second tape was the softer, more compressible

1115 by 3M which is recommended for halftone printing. The printing und curing took place on the

T-Flex 508 industrial printing press by Timsons. The anilox employed contained seven bands of

different cell volumes (Table 5.8). The other anilox specifications were identical for all bands: laser

engraving, 60° screen angle and 500 Ipi screen ruling. All bands were printed at the same time, but

i only the prints using anilox volumes of 3.04 and 8.31 cm3/m 2 were analysed. Although only the

\ higher volume is comparable with the specifications used on the IGT-F1, the lower volume was
j
‘ representative of conditions generally found in industrial colour printing. The experimental order is

outlined in Table 5.9.

| 167
 Figure 5.22: Image for m eso-pattern trial on the T-Flex 508:
(a) seven identical bands containing the printing patches w ith 11 surface patterns and plain reference solid are arranged
in columns (0.2x magnification); (b) image detail corresponding to the red box in (a) showing the positive and negative
hexagon pattern patches on neighbouring bands (to scale)

Visual inspection o f th e patches directly after printing revealed th a t most surface patterns had failed

to create a closed ink film on the substrate. Since the optical density o f the prints was governed by

area coverage rather than ink film thickness, it was deemed unnecessary to digitise the prints fo r

analysis of print u n ifo rm ity w ith image analysis software. Inform ation on optical density and the

effects o f printing param eters on it could be obtained more efficiently by spectrophotom etry.

Therefore the optical density o f printed patches was measured on eight consecutive sheets fo r each

possible com bination o f param eters listed in Table 5.7.

| 168
 Material Process
parameters parameters

Surface pattern

Printing plate Anilox volume

Plate-substrate
Mounting tape
engagement

Printing speed

Figure 5.23: Parameters and th e ir interactions of m eso-pattern trial on the T-Flex 508

Table 5.7: Param eter levels for experim ental plan of m eso-pattern trial on the T-Flex 508

Factor Lower level Upper level

Surface patterning Reference solid (no pattern) Pattern

Plate material AWP-DEF AFP-DSH

M ounting tape Soft (1115) Combination (1015)

Anilox volume 3.04 cm 3/ ir i2 8.31 cm3/m 2

Printing engagement 3 thou 76.2 pm 5 thou 127 pm

Printing speed 100 ft/m in 30.5 m /m in 300 ft/m in 91.4 m /m in

Table 5.8: Volum e specification of anilox bands

Band 1 2 3 4 5 6 7

Anilox volume (cm3/m 2) 1.63 3.04 4.20 5.49 6.59 8.31 1.63
 Table 5.9: Combination of print param eters for m eso-pattern trial on the T-Flex 508

Sp<2ed Engagejment
Experiment number Tape Plate
(ft/m in) (m /m in) (thou) (pm)

1 3 76.2
100 30.5
2 5 127.0
AWP-DEF
3 3 76.2
300 91.4
4 5 127.0
M edium
5 3 76.2
100 30.5
6 5 127.0
AFP-DSH
7 3 76.2
300 91.4
8 5 127.0

9 3 76.2
100 30.5
10 5 127.0
AWP-DEF
11 3 76.2
300 91.4
12 5 127.0
Soft
13 3 76.2
100 30.5
14 5 127.0
AFP-DSH
15 3 76.2
300 91.4
16 5 127.0

5.2.3 Results and Discussions

The spectrophotom etric analysis confirm ed the visual observation th a t fo r no com bination of

printing parameters did any of the patterns achieve as high an optical density as the reference solid

(Figure 5.24). Although the reference prints were not perfectly homogenous and included numerous

defects, the UCAs in the printed pattern patches were more extensive and significantly decreased

the optical density o f th e ir prints (Figure 5.25 shows the exem plary comparison of the reference

solid and chequer pattern at 45° tilt). The UCAs are likely to be the combined result o f entrapped air

bubbles which prevented com plete ink transfer and non-printed areas in the patterns which remain

if not enough ink volum e or pressure is applied to fo rm closed ink films through the mechanism of

ink squeeze (as discussed fo r the IGT-F1 in section 5.1.4.8).

| 170
Figure 5.24: M ean optical densities (averaged over all printing conditions o f the entire experim ental plan) fo r plain solid
reference and all surface patterns in m eso-pattern trial on the T-Flex 508

Reference solid Chequer pattern (45° tilt)

s u b s tra te 4 *
d is c o n tin u ity *

• SB • O - w

. *0 1 1
( ^ d e fe c t

I Oi
r
* ? r - 'in
X , * • % •^ ^

fllm
»

Figure 5.25: Comparison of printing defects (areas of missing ink) in m eso-pattern trial on the T-Flex 508
(AWP-DEF plate m aterial, soft tape, high anilox volum e, 100 ft/m in printing speed a t 8 thou engagem ent)

To facilitate a comparison between the trials on the IGT-F1 and T-Flex 508, the DoE analysis o f the

data has been undertaken. The com plete data set on optical density fo r all surface patterns, as well

as the uncertainty o f the effects stating th e ir significance, is provided on the CD-ROM attached to

this thesis. All surface patterns resulted in main factor effects having an identical im pact and very

sim ilar magnitude. As an example, the 15 largest effects fo r the chequer pattern at 45° tilt are

provided in Figure 5.26. They are all significant based on the effect uncertainty o f AE=±0.003. For

this pattern the mean optical density fo r the entire experim ental plan was D=1.04, and effects of

ED1=0.10 and ED2=0.90 correspond to changes in MGSL by approxim ately EMGSL1=10 and

E m g s li= 6 5 levels respectively. Only the most relevant main and interaction effects fo r all surface

patterns are discussed below.

 0 .6

0 .4

I 0.2
c
01
^ 0.0
re
u
‘■S. - 0.2
o
o -0.4
(1)
56 -0.6

- 0.8

- 1.0

s>0+ *
<#
.<f~ & r f ^ A
&
<^° < /" <^°+ <P <r « /
X ^ q*
^ J? ..
<*> o \*

Figure 5.26: The 15 largest main factor and interaction effects on optical density for the exam ple of
chequer pattern at 45° tilt on the T-Flex 508 (main effects in blue, interaction effects in orange)

In general, all m aterial and process parameters th a t increased ink transfer from the plate to the

substrate or ink squeeze on the substrate increased optical density. M ore ink provided by the anilox

led to higher ink transfer to the plate and then to the substrate where better area coverage was

achieved through ink squeeze (ED=+0.40). The slightly harder com bination tape compressed less in

its core during impact. Thus the printing plate itself had to deform more against the substrate. The

contact area o f plate and substrate increased and w ith it the ink squeeze in the printing nip

(Ed=+0.19). The soft m ounting tape on the other hand relieved the printing plate through core

compression. This is more im p orta n t in halftone printing to avoid dot distortion and create sharper

printed dots. The interaction of high anilox volum e and com bination tape increased the effect on

optical density through ink squeeze even fu rth e r (ED=+0.07). The last main factor w ith positive effect

on optical density was the engagement which increased the contact area between the plate and

substrate (ED=+0.16), thereby creating higher area coverage and supporting ink squeeze towards

more homogenous ink distribution.

Surface patterning had the largest negative influence on optical density (ED=-0.89). On the printing

plate, patterned patches had low er area coverage than the reference solid which translated into

reduced ink coverage and optical density in the prints. Patterns featuring holes rather than

freestanding features, such as the 95%, grid and negative hexagon patterns, fared slightly better

| 172
than the others. Due to the interconnectivity of the features, these patches were closer to forming

continuous ink films in the print than isolated features which had to connect the separate ink

depositions on the substrate through ink squeeze. Since the ink squeeze action is the dominant

mechanism for surface patterns in conjunction with UV-curing ink, it can be concluded that the

design characteristics of the meso-patterns, namely geometry, area coverage and feature pitch,

were not suited to this particular ink type under the printing conditions encountered on the

T-Flex 508, which contributed to the pattern failure. To confirm this, a further print trial was

conducted where the printing plates were imaged with micro-patterns, and this will be dealt with in

Chapter 6.

The negative patterning effect was amplified when including the interaction with the plate material

(Ed=-0.15), because the pattern area coverage was significantly lower on the solvent-washable AFP-

DSH (Figure 5.8) than on the water-washable AWP-DEF (Figure 5.9). Furthermore, the increased

surface roughness of AWP-DEF potentially improved its ink-carrying capabilities. Consequently,

larger amounts of ink transferred to the plate, hence to the substrate, improving the ink squeeze

action and optical density (reduction of optical density with usage of AFP-DSH expressed by

Ed=-0.19).

The rise in printing speed decreased the optical density (ED=-0.23). One potential cause identified for

this was the ink film splitting closer to the substrate at higher speed, thereby reducing the ink film

thickness printed (De Grace and Mangin, 1983). It has also been observed previously that a

reduction of ink volume in the anilox cells occurs due to their filling being hampered by

hydrodynamics at higher speed, and ink being scooped out of the cell through the wiping action of

the doctor blade (Kunz, 1975). Less ink in the cells leads to less ink transferred to the substrate. The

negative interaction effect of printing speed and anilox volume (ED=-0.06) implies that the larger

anilox volumes were impacted further by the change in speed due to more ink being doctored out of

these cells with larger opening. However, there was a positive interaction of speed and surface

patterning (ED=+0.09) which meant that the individual reduction in optical density by usage of

surface patterning and higher speed was cumulatively less severe. Potentially, while the reference

solid was unable to deform sufficiently into the anilox cells containing reduced fluid levels to ink up

fully, the pattern features deformed better or dipped into the cells, picking up proportionally more

ink. That this beneficial effect had its limit is suggested in the negative interaction of patterning and

anilox volume (ED=-0.12). Assuming that the anilox cells with larger opening contain lower ink levels

(as discussed above), the pattern features also were increasingly unable to retrieve ink from the

depths of the cells and the ink transfer was reduced. However, no trend in the correlation of the

| 173
latter tw o interaction effects and the type of surface pattern could be observed. Furthermore, the

magnitude of most interaction effects fell within the noise range identified for the T-Flex508

(AD=0.1) and have to be treated with caution.

Overall there was a good agreement of parameter effects on optical density between the studies on

the IGT-F1 and T-Flex 508. The deviant influence of printing speed was attributed to particular

differences in the construction of IGT-F1 and T-Flex 508. On the IGT-F1, the ink supply is dependent

on the operator and can easily be overdosed. The deflection of the doctor blade due to non-rigid

mounting was aggravated by overdosage, thereby allowing the flooding of the anilox roll (cells and

land area) with ink. Consequently, this ink swamped the patterned patches on the printing plate,

creating higher ink coverage and making meso-patterns a viable option for the improvement of print

density. The effect was particularly severe for low-viscosity water-based ink. On the other hand, the

ink supply on the T-Flex 508 is consistent through design which renders surface patterning less

successful.

A second difference between the printing methods was the anilox specifications. The anilox screen

ruling used on the IGT-F1 was 300 and 350 Ipi (84.7 and 72.6 pm respectively) which was

significantly larger than the nominal pattern size of 50 pm. Pattern features were therefore likely to

dip into the anilox cells and pick up more ink on the feature shoulders. On the T-Flex 508 the

nominal pattern width on the plate and anilox cell pitch (50.8 pm) were similar. Thus, it is possibly

that dot dipping was a less frequent occurrence, reducing ink coverage on the plate and then the

print. Furthermore, the anilox volumes investigated on the IGT-F1 (8 and 12 cm3/m 2) were larger

than those on the T-Flex 508 (3.04 and 8.31 cm3/m 2), thereby facilitating increased ink transfer to

the printing plates.

5.2.4 Conclusions
The investigation of eleven meso-patterns was scaled up to near-industrial conditions. The effects of

material and process parameters on print quality observed on the T-Flex 508 confirmed the findings

of the IGT-F1 study. Using UV-curing ink, the surface patterns failed to create closed ink films on the

substrate, because the ink squeeze mechanisms was potentially insufficient due to a lack of ink

volume or inappropriate pattern design, notably too large a distance between pattern features or

freestanding pattern geometry.

| 174
5.3 Closure
A comprehensive investigation of the influence of plate surface patterning on the ink transfer from

solid printing patches has been undertaken under laboratory and near-industrial conditions. Meso-

patterns with nominal feature sizes of 50 pm were examined for their effect on print quality, namely

optical density and print uniformity, in conjunction with different material and process parameters.

For the first time, two functionalities of surface patterns were identified in conjunction with the ink

type used. Lower-viscosity water-based ink is carried on the plate surface as well as in the pattern

recesses analogous to anilox cells, and the pattern geometry can be used to shift the frequency of

printing defects towards improved print uniformity. Higher-viscosity UV-curing ink remains on the

pattern features and depends on the ink squeeze across the substrate to create closed ink films. The

successful application of this mechanism requires an optimisation of the pattern design to facilitate

ink coalescence.

Under the conditions investigated on the T-Flex 508, the meso-patterns failed to create closed ink

films. On the IGT-F1, meso-patterns were found to be capable of improving optical density and print

uniformity, although both quality criteria were not necessarily obtainable at the same time. The

effectiveness of the patterns was highly dependent on material and process parameters, and some

parameter combinations had particularly adverse influence on the amount and consistency of the

ink transferred. In general, the process parameters were able to affect optical density and print

uniformity more than the application of surface patterning, and very specific parameter

combinations were required to gain a benefit from patterning. The directions of the parameter

effects determined for the IGT-F1 and T-Flex 508 were in agreement with previous knowledge of

printing solids and halftones in flexography, and can be employed in the optimisation of printing

processes.

| 175
Chapter 6 M icrocell Patterns on P rinting Plates
In the previous chapter two different functionalities of meso-patterns were identified dependent on

the ink type used. In conjunction with UV-curing ink, the surface patterns created only partial ink

films on the substrate due to a lack of ink coalescence. This was attributed to inadequate pattern

design, notably geometry, insufficient area coverage and feature pitch, in order to allow ink

coalescence by a squeezing action.

In this chapter the design issues identified for the meso-patterns were addressed in the form of

commercial microcell patterns. The microcells have in common that they are made up of small

recesses in the surface of the flexographic printing plate analogous to the land areas and cells on the

anilox cylinder. Their area coverage and feature frequency is higher than for the meso-patterns, and

the recess diameter smaller than the nominal width of the meso-features. This chapter aims to

explore the ink transfer functionality of these micro-patterns superimposed on solid printing patches

and halftone dots. For this purpose, again, investigations were conducted on a printability tester and

industrial-scale printing press. An overview of the two trials is provided below.

M ic r o c e l l Pa t t e r n E ffe c t u s in g a P r in t a b il it y T ester

The IGT-F1 printability tester was used to explore a total of three pattern sets. The plate-imaging

was performed using "flat-top" imaging technology on appropriate plate material supplied by

MacDermid. In order to establish whether this material behaved fundamentally differently in

conjunction with surface patterns, the print trial for meso-patterns was repeated and formed the

first pattern set. The second printing plate employed a selection of meso- and four microcell

patterns. The particular microcell specifications were chosen by the reproduction house for their

claimed successful application in previous industrial print jobs. The last printing plate consisted of

four of the most efficient meso-patterns at three different size ratios (50%, 100% and 200%). The

shrinkage of meso-patterns by 50% made their dimensions comparable to the ones of microcells and

bridged the gap between meso- and micro-patterns in the investigation. Furthermore, this served to

gain additional information on the effect of pattern-scaling on print quality.

M ic r o c e l l Pa t t e r n Ef f e c t u s in g a n I n d u s t r ia l P r in t in g P ress

The study conducted on the T-Flex 508 printing press examined the effects of four microcell patterns

under varying printing conditions at a practical commercial scale. "Flat-topped" and "round-topped"

solid printing patches and halftones superimposed with microcells were investigated on three

different plate types. The quality characteristics that were applied comprised optical density and

print uniformity.

| 176
 6.1 Microcell Pattern Effect using a Printability Tester
The main aim o f this study was to establish w hether the functiona lity o f microcell patterns is

fundam entally d iffe re nt to th e ir meso-scale counterparts. D ifferent meso- and m icro- pattern

designs were investigated.

6.1.1 Materials, P rinting and P rin t Characterisation

The plate m aterial "Digital Rave" by MacDermid was provided and imaged by SGS Packaging Europe.

The key plate properties are listed in Table 6.1. The imaging was perform ed using MacDermid LUX

technology and an imaging resolution of 4,000 dpi. Details o f the pattern sets fo r all three plates are

!listed in Table 6.2. The first pattern set (Plate 1) is identical to the set o f the previous chapter and

' served to establish w hether the d iffe re n t plate materials and imaging technologies in conjunction

w ith surface patterning had an effect on print quality. The second pattern set (Plate 2) compares

meso-patterns and microcells. The third pattern set (Plate 3) contains m eso-patterns at three

d iffe re n t scales, the smallest o f which is sim ilar to the m icro-patterning and links the tw o pattern

groups investigated. For the illustrations o f the m eso-patterns refer to Figure 5.2. Specifications of

the microcell patterns are provided in Figure 6.1 and Table 6.3. The actual pattern geometries on the

printing plates were characterised using WLI. This is illustrated in Appendix A.7 and w ill be discussed

in section 6.1.3.1.

The substrates APCO and Ford as well as the UV-curing ink FlexoCure Gemini and the in-house

water-based Newtonian ink were selected again. The m aterial properties are detailed in section

5.1.1.

Table 6.1: Selected properties of plate m aterial used for m icro-pattern study on the IGT-F1

Property MacDermid Digital Rave

W ashability solvent

Surface roughness, Ra 54 nm

Surface roughness, Rz 1.1 pm

Steady-state surface energy 44.6 m N /m

Hardness* 60 Shore A

Thickness 1.7 mm

* hardness according to publication by MacDermid Printing Solutions (2012)

I 177
 Table 6.2: Surface patterns of the three printing plates used for m icro-pattern study on the IGT-F1

Surface pattern Nominal pattern dimensions P latel Plate 2 Plate 3

Reference solid - V V V

Vertical lines 50 pm track/50 (im gap V V

Positive hexagons 50 pm w id th /2 5 pm gap V V

Negative hexagons 50 pm w id th /2 5 pm track V V

Horizontal lines 50 pm track/50 pm gap V

Diagonal lines 50 pm track/50 pm gap V

Polka dots 50 pm diam eter/100 pm pitch V

Chequer (no tilt) 50 pm by 50 pm w idth V

25 pm by 25 pm w idth V

Chequer (45° tilt) 50 pm by 50 pm w idth V V V

100 pm by 100 pm w idth V

24 pm diam eter/95.5 pm pitch V

95% 48 pm diam eter/191 pm pitch V V V

96 pm diam eter/382 pm pitch V

25 pm by 25 pm w id th /2 5 pm track V

Grid 50 pm by 50 pm w id th /5 0 pm track V V V

100 pm by 100 pm w id th /1 0 0 pm track V

50 pm by 25 pm w id th /12 .5 pm gap V

Tiles 100 pm by 50 pm w id th /2 5 pm gap V V V

200 pm by 100 pm w id th /5 0 pm gap V

MC 09P_H V

MC 09P_L V
Microcells
MC 12P V

MC 16P V

| 178
 Figure 6.1: Illustrations of microcell patterns (200x magnification)

Table 6.3: Specifications of microcell patterns

Microcell pattern Nominal size (approx.) Nominal area coverage (%)

MC 09P_H 19 (im diam eter recesses 64.0

MC 09P_L 19 pm diam eter recesses 89.9

MC 12P 22 pm diam eter recesses 76.0

MC 16P 25 pm diam eter recesses 78.4

6.1.2 Printing and P rin t Characterisation

All three printing plates were subjected to the same full-factorial experim ental plan involving five

factors (each adjustable on tw o levels) (Figure 6.2). The same param eter levels (except fo r the

change in plate m aterial) selected fo r the m eso-pattern studies on the printability tester were

adopted (refer to Table 5.6). The prints from the 32 experim ental points were digitised, and the

measurands MGSL (density) and StDev (print uniform ity) obtained through the image analysis

software. The com plete effect data fo r MGSL and StDev o f all surface patterns, as well as the

uncertainty o f the effects stating th e ir significance, is provided on the CD-ROM attached to this

thesis.

Material Process
parameters parameters

Substrate Anilox volume

Printing force
Figure 6.2: Parameters and
their potential interactions
in m icro-pattern study on
Printing speed the IGT-F1

| 179
6.1.3 Results and Discussions

6.1.3.1 Plate Characterisation

Using LUX technology, oxygen inhibition o f the plate polym erisation was prevented during imaging

and the digital designs were transferred to the printing plate at alm ost perfect fid e lity (compare

images in Appendix A.7). A slight gain in area coverage o f up to 3% was noted for all pattern patches

(Plate 1 - Figure 6.3, Plate 2 - Figure 6.4 and Plate 3 - Figure 6.5). Only the microcell patterns

MC09P_H and MC12P had slightly higher increases of 6% and 4% respectively. The actual

geom etrical shapes o f the features were m aintained on the printing plate w ith in practicability o f the

imaging process defined by the resolution o f the laser creating the imaging mask on the plate.

■ N om inal area coverage ■ Coverage gain

90

2^ 70
<D
CuO
£ 50
75
30
50
10

-10
.o$
# 4 ? </ ^ 4 ? *
* ^ <5 jy

Figure 6.3: Actual area coverage on printing plate and coverage loss compared to the artw ork for M acDerm id Lux Plate 1

■ Nom inal area coverage ■ Coverage gain

90

95

■
45

-10

Figure 6.4: Actual area coverage on printing plate and coverage loss compared to the a rtw ork fo r M acDerm id Lux Plate 2
 N o m in a l a re a c o v e ra g e ■ C o ve ra g e g a in

90

£ 70
<u
M
ro 50
95 95
75 75 75
30
50 50 50 53 53 53
10

-10

<v &C?\°" <z>

C°\°" ( j? -& w ^
<r
&' ^

* S' V

Figure 6.5: Actual area coverage on printing plate and coverage loss compared to the artw ork for M acDerm id Lux Plate 3

6.1.3.2 Effect o f Plate M a te ria l and Im aging Technology

For the investigation o f the effect o f plate material and imaging technology, m eso-pattern designs

identical to the ones in the previous chapter have been used. Despite the significant differences in

plate geometries, the mean MGSL (Figure 6.6) and mean StDev (Figure 6.8) from the experim ent

were very sim ilar or better fo r the MacDermid Lux (Plate 1) compared to the Asahi plates. A

difference o f 10 GSLs w ithin the range observed correlates approxim ately to a difference of 0.1 in

optical density which corresponds to the lim it o f discernibility fo r the untrained eye.

MacDermid Lux ■ AWP-DEF ■ AFP-DSH

120
115
110
-00i 105
g 100
! 95
S 90
s 85
80
75
70
vy \<o jy $0

* * / / J J < < /> > *

Figure 6.6: Comparison of mean MGSL (averaged over all printing conditions of the entire experim ental plan) for
M acDerm id Lux (Plate 1) and Asahi plates

| 181
The pattern fid e lity on the MacDermid Lux plate had tw o advantages w ith regards to the dual

fu n ctio n a lity of surface patterns identified in the previous chapter. On the one hand, it led to an

increased area coverage which governed the transfer o f UV-curing ink. This translated directly to

increased amounts o f ink supplied to the substrate and im proved mean MGSL o f the prints fo r the

m a jority o f pattern patches (Figure 6.6). On the o th e r hand, deeper pattern recesses were

established (Figure 6.7) which, dependent on cell evacuation, potentially allowed larger volum es of

water-based ink to be transferred, thereby creating darker prints. The lack o f area coverage and

recess volum e on the AWP-DEF plate seemed to have been compensated by its higher ink-carrying

capabilities, and the mean MGSL fo r this Asahi material and the MacDermid Lux plate were very

similar.

M acDermid Lux AWP-DEF AFP-DSH

0
5

E -10

-15

" -20

-25

-30

-35
0 50 100 150 200 250
W idth (pm)

Figure 6.7: Surface profile of negative hexagon pattern on different plate types (missing WLI data on fea tu re shoulders
of M acDerm id Lux and AFP-DSH plates was interpolated from captured data at peaks and troughs)

The higher mean StDev from the experim ent reveals th a t the MacDermid Lux plate produced less

uniform prints than the AWP-DEF material and most patches on the AFP-DSH plate (Figure 6.8). One

reason fo r this m ight be the bigger ink transfer from the MacDermid Lux plate w hich aided the

evolution o f fingering instabilities (refer to Figure 6.19 fo r examples). Furtherm ore, larger volum es of

ink am plified the contrast o f the printed patterns, thereby enhancing the inhom ogeneities. This can

be quantified in the difference between maximum and m inim um GSLs observable fo r the print

samples. Secondly, the printed patterns created w ith the MacDermid Lux plate were by th e ir nature

less uniform . This m ight be due to the very low surface roughness o f the plate which provided few

| 182
points to which the receding ink could be pinned in order to create finer filam ents during ink film

splitting. Thus, the form a tio n o f broad, low -frequency fingering patterns was facilitated.

■ MacDerm id Lux ■ AWP-DEF AFP-DSH

35

UI
30

> 25
<U
9 20
S 15 i L i y I I i i , ■■ ■■ i i ■■ i
01
S 10

0
5 il|y u
Figure 6.8: Comparison of mean StDev (averaged over ail printing conditions of the entire experim ental plan) for
M acDerm id Lux (Plate 1) and Asahi plates

In general the direction and magnitude o f the main factor effects were sim ilar on all three plates.

Therefore, the param eter effects determ ined fo r the Asahi plates on the IGT-F1 in the previous

chapter remain applicable fo r the MacDermid Lux plates. However, the printing force was an

exception. On the MacDermid Lux plate, the force had no significant or only a smaller effect on

MGSL (Figure 6.9) and StDev (Figure 6.10), when the printing force was changed from 50 to 150N.

The effect was positive fo r some and negative fo r other printing patches. On the Asahi plates,

printing force had a significant negative effect throughout. The Asahi plates suffered from a loss in

area coverage and exhibited reduced edge sharpness. This made them more sensitive to

deform ation under pressure, whereas the MacDermid Lux features retained th e ir geom etry and

were able to distribute the pressure load b etter (Anderson and Schlotthauer, 2010). Furtherm ore,

the deform ation o f the softer MacDermid Lux plate took place w ith in the plate layers rather than at

the top of the printing features (M irle, 1989).

| 183
 M a c D e rm id Lux I A W P -D E F ■ AFP-DSH

10

-25
A° K<*> K
<*> <*> <*> <*> <*> jc? o’ e)

4^ jC T 9

Figure 6.9: Effect of printing force on MGSL for M acDerm id Lux (Plate 1) and Asahi plates

■ M acDerm id Lux ■ AWP-DEF ■ AFP-DSH

Figure 6.10: Effect of printing force on StDev for M acDerm id Lux (Plate 1) and Asahi plates

6 .1 .3 .3 E ffect o f M ic ro c e ll versus M e s o -p a tte rn s

To investigate microcell patterns in comparison w ith m eso-patterns, as well as the plain reference

solid, Plate 2 contained seven o f the original m eso-patterns and four solid printing patches

superimposed w ith microcells (patterns as outlined in Table 6.3). The patches including microcells,

which are in essence patterns o f recesses analogous to the 95% and negative hexagon meso-

patterns, produced lighter prints (Figure 6.11) and had less im pact on MGSL (Figure 6.12) than most

o f the other pattern geometries. There was only a very small num ber o f process param eter

com binations fo r which the microcells achieved higher density.

| 184
 ■ MGSL ■ StDev

Figure 6.11: M ean MGSL and mean StDev (averaged over all printing conditions of the entire
experim ental plan) for plain solid reference and all surface patterns on M acDerm id Lux (Plate 2)
(microcell patterns - orange; m eso-patterns - blue)

■ MGSL ■ StDev

Figure 6.12: M ain effect of surface patterning on MGSL and StDev for all surface patterns on M acDerm id Lux (Plate 2)
(microcell patterns - orange; m eso-patterns - blue)

| 185
In the previous chapter it could be concluded th a t m eso-patterns act analogous to anilox cells in

conjunction w ith water-based but not UV-curing ink, and the study of m icro-patterns corroborates

this. Using UV-curing ink which remains on the printing features, the mean MGSL improves w ith the

area coverage of the pattern on the printing plate, as this results in higher area coverage on the

prints (Figure 6.13). The ink volum e th a t the pattern recesses can potentially hold is of little

consequence w ith UV-curing ink, because the recesses are not being flooded (Figure 6.14). The

opposite can be observed fo r prints w ith water-based ink. The low er the pattern area coverage and

the larger the recess volum e7, the darker are the prints. The microcells had significantly higher area

coverage (Figure 6.4) than m ost o f the m eso-patterns which provided a small advantage fo r prints

w ith UV-ink. However, they also had a low er recess volume, as illustrated by Figure 6.15, which

strongly hampered water-based prints and led to the overall reduced patterning effect on MGSL

(Figure 6.12).

♦ UV-curing ink ■ W ater-based ink

160

140

120
E 100
80

60

40

20

20 40 60 80 100
A c tu a l a re a c o v e ra g e (% )

Figure 6.13: Correlation of area coverage on M acDerm id Lux (Plate 2) w ith mean MGSL
(averaged over all printing conditions of the entire experim ental plan involving the respective ink type)

7 The recess volum e was estim ated assuming conical recesses fo r th e microcell, 95% and negative hexagon
patteirns, and pyram idal recesses fo r th e grid and chequer patterns (choice inform ed by WLI data) on a square
area <of ten m illim e tres side. The recess d e p th and w id th was obtained fro m the WLI data and examples o f
recess profiles are shown in Figure 6.15. No volum e data was calculated fo r th e tile and positive hexagon
patteirns w hich contained freestanding features.
 ♦ UV-curing ink ■ W a te r-b a se d ink

160

140

120
ft 100
80

40

20

0.E+00 l.E+08 2.E+08 3.E+08 4.E+08 5.E+08

Recess volume (urn3)

Figure 6.14: Correlation of recess volume on M acDerm id Lux (Plate 2) w ith mean MGSL (averaged over all printing
conditions of the entire experim ental plan involving the respective ink type) for all surface patterns except V 5 0 /5 0 ,
tiles and positive hexagons

MC09P_L --------95% Hexaneg

? -l°
3 -15
H -20
01
o .2 5

-30
-35
-40
0 50 100 150 200 250 300
Width (urn)

Figure 6.15: Example profiles of selected microcell and m eso-patterns featuring recesses

The effect o f surface patterning on p rint u n ifo rm ity was also highly dependent on the ink type used

(refe r to section 5.1.4.8). The high opacity of the UV-curing ink did not allow the observation of any

fingering instability in the prints featuring closed ink film s (Figure 6.16). The only visible printing

defects fo r UV-curing ink were UCAs which remained where the patterns failed to create closed ink

film s on the substrate. The extent of UCAs was directly dependent on the pattern area coverage, as

high area coverage aids the ink film coalescence and thereby print u n iform ity (Figure 6.17). The

observable trend in print unifo rm ity im provem ent w ith decreasing recess volum e relates directly to

| 187
area coverage (Figure 6.18), because the recesses volum e is inversely proportional to the area

coverage. The microcell patterns led to few er UCAs in the UV-prints due to th e ir high area coverage,

small recess volum e and feature pitch. Thus the calculated effect o f microcell patterning on print

un ifo rm ity (Figure 6.12) was governed by the data obtained from prints made w ith water-based ink

featuring fingering instabilities which reduced p rin t uniform ity, representing a w orst case result.

Name Reference solid MC09P L

Figure 6.16: Lack of defects in prints made w ith UV-curing ink (APCO substrate, M acD erm id Lux (Plate 2),
speed = 0.2 m /s, force = 50 N, anilox volum e = 8 cm 3/ m 2; 7x m agnification)

♦ UV-curing ink ■ Water-based ink

40 60 100
Actual area coverage (%)

Figure 6.17: Correlation of area coverage on M acDerm id Lux (Plate 2) w ith mean StDev
(averaged over all printing conditions of the entire experim ental plan involving the respective ink type)

| 188
 ♦ UV-curing ink ■ W a te r-b a se d ink

30

0.E+00 l.E+08 2.E+08 3.E+08 4.E+08 5.E+08

Recess volume (pm 3)

Figure 6.18: Correlation of recess volum e on M acDerm id Lux (Plate 2) w ith mean StDev (averaged over all printing
conditions of the entire experim ental plan involving the respective ink type) for all surface patterns except V 5 0 /5 0 , tiles
and positive hexagons

Although the microcell patterns did not perform exceptionally by im proving the print density of

water-based prints, they led to im provem ents in print uniform ity. The patterns MC09P_H and

MC09P_L achieved a small negative main factor effect implying more homogeneous prints than the

; reference solids (Figure 6.12). However, the effects were too small to be recognised as statistically

; significant. Also not significant was the positive effect o f the 95% and grid patterns which, together

w ith the tw o larger microcell patterns (MC12P and MC16P), had the least detrim ental effect on print

uniform ity. These six patterns have in com m on that they are made up o f holes entirely surrounded

by a land area. Thus a trend may be deduced from this factor, although the negative hexagon

pattern, which by defin ition is also a hole pattern, did not fit w ith this observation. This may be

a ttrib u te d to the larger size of the land area: the negative hexagons have a land area coverage of

55.7%, whereas none o f the others fell below 75% (Figure 6.4). This is confirm ed by the correlation

o f area coverage w ith print u n ifo rm ity (Figure 6.17). The recess volum e had little impact on print

u n ifo rm ity (Figure 6.18).

The dem arcation in area coverage was accompanied by a transition in printing defects. The solid

reference (100% area coverage), 95% (95.4%) and microcell patters (90.5% to 76.0%) exhibit defects

o f viscous fingering (Figure 6.19). The grid pattern (75.5%) showed the transitional form ation of

small beads on top o f the viscous fingers. The negative hexagons and open patterns w ith

freestanding features (55.7% to 47.1%) exhibited extensive beading. It appears th a t hole patterns

are able to counteract the evolution o f filam ents during ink film splitting which leads to beading. The

| 189
transition in fingering patterns m ight not only be due to effects of area coverage, but also to the

difference in scale between microcell and m eso-patterns as discussed in the next section.

Name Reference solid MC09P L

Area coverage 100 % 95.4% 90.5%

Name MC16P MC12P MC09P H

Area coverage 81.5% 75.9%

Name Negative hexagons 45 Chequer

Area coverage 75.5% 55.7% 53.0%

(
Figure 6.19: Transition of fingering defect regimes w ith area coverage in descending order (w ater-based ink on APCO
substrate, M acDerm id Lux (Plate 2), speed = 0.2 m /s, force = 50 N, anilox volume = 8 cm3/ m 2; 7x m agnification)

| 190
6.1.3.4 Effect o f Pattern Scaling

In crder to bridge the scaling gap between m icrocell and m eso-patterns, this th ird part o f the study,

using Plate 3 (patterns as outlined in Table 6.3), investigated fo u r m eso-pattern designs at three

different scales, the smallest of which was comparable to the size o f the microcell patterns. The

scaling o f the fou r patterns was adjusted from 100% (original feature size o f about 50 pm) to 50%

(25 pm) and 200% (100 pm). The proportions o f land and recess areas as well as the nom inal area

coverage were m aintained at the three scales (Figure 6.20). The actual area coverage was shown in

Figure 6.5.

♦♦♦♦♦♦♦< ♦ ♦ ♦ ♦ ♦
♦♦♦♦♦♦♦<
♦ ♦ ♦ ♦ ♦ ♦ ♦ <
♦ ♦ ♦ ♦ ♦
♦ ♦ ♦ ♦ ♦ ♦ ♦ <
♦♦♦♦♦
► ♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦

♦♦♦♦♦
► ♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ <
::::::::::::::::::: ♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
<
<
::::::::::::::::::: ♦ ♦ ♦ ♦ ♦ ♦ ♦ <
♦♦♦♦♦♦♦< ♦♦♦♦♦
♦ ♦ ♦ ♦ ♦ ♦ ♦ <
♦ ♦ ♦ ♦ ♦ ♦ ♦ <
♦♦♦♦♦
Figure 6.20: Scaling example of chequer m eso-pattern designs -
(from left to right) 50%, 100% and 200% scaling (50x m agnification)

Pattern scaling had an im pact on MGSL as well as on StDev (Figure 6.21). The main factor effect of

patterning suggests th a t scaling had no consistent influence on MGSL fo r three o f the fo u r patterns

investigated (95%, grid and chequer at 45° tilt) (Figure 6.22). Only fo r the tile pattern did the

m agnitude of the patterning effect increase steadily w ith decreasing feature size. This was directly

related to the p rint uniform ity, because smaller patterns resulted in low er StDev and more

homogeneous prints. The larger patterns (200% scaling, about 100 pm feature size), as well as the

original tile pattern (100% scaling, 50 pm), showed increased value, meaning a deterioratio n o f print

uniform ity. This observation is very much in agreement w ith the claims o f Sam worth (2001), Stolt,

Zwadlo and Rozzi (2010) th a t m icro-patterning is required to im prove uniform ity. The origin o f the

im provem ent in print quality can be attrib u te d directly to the tw o functionalities o f surface patterns

identified in the previous chapter (refer to section 5.1.4.8)

| 191
 ■ MGSL ■ StDev

Figure 6.21: MGSL and StDev of entire experim ental plan for plain solid reference and all surface patterns
on M acDerm id Lux (Plate 3) (5 0 /1 0 0 /2 0 0 denoting pattern scale in percent)

MGSL ■ StDev

20 7.0
15
10 3.5 >
IS) QJ

5 Q
*->
IS)

0 0.0 c
o
<v -5 (V
5t
-10 -3.5
-15
-20 -7.0

Figure 6.22: M ain factor effects of patterning on MGSL and StDev for different pattern scales on M acD erm id Lux (Plate 3)
(5 0 /1 0 0 /2 0 0 denoting pattern scale in percent)

The printing defects observed fo r the UV-curing ink are related to the reproduction o f the surface

patterning from the plate in the prints (Figure 6.23). At the largest scale individual printed elements

are discernible fo r the chequer and tile patterns. This printing defect is due to a lack of ink

coalescence between neighbouring features through squeeze action, confirm ed by the fact that

higher anilox volum e and printing pressure facilitated better coalescence. At m edium scale an ink

film disrupted by small UCAs is created which transform s to a very uniform , closed ink film at the

smallest scale caused by the ink squeeze. Figure 6.24 shows the 95% and grid patterns at the largest

| 192
scale which exhibit printed patterns of areas w ith increased ink film thickness. The geom etry of

these areas corresponds to the non-printing regions o f the pattern patches on the printing plate.

This suggests th a t the printing force was o f sufficiently high magnitude to squeeze out most of the

ink from under the printing features and into the pattern recesses to coalesce.

45Chequer 50 45Cnequer100 45Chequer 200

*■•„V ♦* *V «f v
* * V.
I' |V“ *i ~7 I 5 * t4 . . • ■■ v

c . « y ♦"J * J , 0 Jit
TJ
c/>
u I V * ! I f t f if i l l
<
_Q

(TJ m J i® |[ ■l * t ' j
.* % , T u i A
Hi*
* .$: f Hi ;•
J l i l i k t | t i l 4*1 V i
« V it * ♦ * V % * **•> « J '

Figure 6.23: Transition of printing defects w ith pattern scaling for chequer pattern at 4 5 s tilt (5 0 /1 0 0 /2 0 0 denoting
p att ern scale in percent) (APCO substrate, speed = 0.2 m /s, force = 50 N, anilox volum e = 8 cm3/ m 2; 7x magnification)

95% 200 Grid 200

Fiigure 6.24: Samples indicating ink squeeze at high printing force (200 denoting pattern scale in percent) (APCO
substrate, speed = 0.2 m /s, force = 50 N, anilox volum e = 8 cm3/ m 2; 7x magnification)

| 193
The print uniformity of water-based ink samples was governed by fingering instabilities which

changed appearance with pattern scaling. At large scale the printing defects observed for all

patterns were beads. Reducing the pattern scale also changed the bead pattern (size of the actual

beads as well as the intervals at which they occurred). Further reduction resolved the beads and the

underlying viscous fingering pattern remained. Depending on pattern and printing conditions, the

viscous fingers transformed to ribs at the smallest pattern scale. The example provided in Figure

6.23 demonstrates the transition from beads to ribs for the chequer pattern. This suggests that plate

features act as anchor points during ink film splitting. Ink filaments pin to the features and cause a

local increase in the ink volume deposited after the filament breakup. The smaller the pattern

features and the higher the frequency of their occurrence, the more ink filaments are created during

film splitting. More and thereby finer filaments result in less obtrusive printing defects. This is in

agreement with Griesheimer (2014) who performed initial investigations into the effect of pattern

scale on filament formation using halftone dots, and observed that more secondary filaments are

created at smaller scale. Furthermore, this mechanism corroborates the findings of the previous

section that the microcell patterns of smaller recess diameter and feature pitch have a particularly

beneficial effect on print quality.

6.1.4 Conclusions
This investigation formed an extension to the meso-pattern trial conducted on the IGT-F1 and

discussed in the previous chapter which sought to determine the effect of different meso-patterns

on print density and uniformity. The three more detailed studies in this chapter served to

• verify the parameter effects for meso-patterns created by a "flat-top" imaging technology

on a different plate material,

• compare the effectiveness of microcell versus meso-patterns, and

• identify the effect of pattern scaling.

In general, as for the findings on the Asahi plates, all results achieved by the patterns on the

MacDermid Lux plates were highly dependent on the combination of experimental parameters used.

The MacDermid plates imaged with LUX technology allowed almost perfect fidelity in the transfer of

the image information from the artwork to the plate. MacDermid Lux improved the impact of

surface patterning on print density because of the increase in the pattern area coverage and recess

volume which corroborates the dual functionality of surface patterns depending on ink type used.

| 194
A good correlation was found between area coverage and print density for UV-prints as well as

recess volume and print density for water-based prints. This applies equally to microcells, micro- and

meso-patterns. It is possible that transitions between pattern functionalities take place dependent

on pattern design, material and process parameters. The smaller the pattern scale and the shallower

the pattern recesses, the more likely the recesses are to be filled with ink independent of ink type.

The print uniformity was determined by pattern geometry, area coverage, feature size and pitch.

The creation of closed ink films with UV-curing ink relies on the ink squeeze in the printing nip. Hole

patterns, such as microcells, and small feature pitch, as achieved by very small pattern scaling,

improve ink coalescence of neighbouring depositions, because the ink has to be displaced over a

shorter distance. The appearance of fingering instabilities (ribs, viscous fingers, beads) in water-

based prints also profits from lower feature pitch. The pattern features act as anchors for ink

filaments during film splitting. The smaller the feature size and pitch, the finer is the ink

filamentation and the more uniform the resultant printed ink film. This was confirmed by comparing

the prints of different scale meso-patterns. Patterns with lower area coverage, isolated features or

large-diameter recesses led to a transition in printing defects to very distinct beads which was

attributed to the formation of large filaments pinned to the pattern features during the film splitting

process.

| 195
6.2 Microcell Pattern Effect using an Industrial Printing Press
The knowledge gained in the study of microcell patterns on the IGT-F1 was explored further on the

T-Flex 508. Since the print trial of meso-patterns on the T-Flex 508 failed to produce solid prints

(refer to section 5.2.3), it was decided not to repeat an entire stand-alone study for microcells on the

press. Instead printing patches containing microcells were included in the test image investigating

dot geometry on the T-Flex 508 (refer to Chapter 4).

This setup on the one hand allowed studying the influence of microcell patters on the optical density

and print uniformity produced by halftones and solids under different printing conditions. On the

other hand, the influence of imaging technology in conjunction with microcell patterning could be

investigated. A previous trial had indicated that the use of MacDermid LUX "flat-top" imaging

technology with microcells improved print optical density compared to standard digital technologies

(MacDermid Printing Solutions, 2011; Cook, Recchia and Gotsick, 2014). The work in this chapter will

be aimed at exploring this independently.

6.2.1 Materials, Printing and Print Characterisation

Detailed information on the experimental order, material and process parameters was provided in

Chapter 4. Due to the small number of variable parameters, no experimental plan was required for

the execution and analysis of this study. The MacDermid Lux ("flat-topped")8, MacDermid and Asahi

plates (both "round-topped") carried identical test images (Figure 6.25) of which the following

elements were relevant for the investigation of microcell patterning:

• Solid patches with

o microcell patterns MC09P_L, MC09P_H, MC12P and MC16P and

o in-house patterns Grid (50 pm track and gap) and Chequer (50 pm at 45° tilt);

• Halftone scales in CMYK consisting of

o 10 area coverages (20, 3 0 ,40 , 50, 60, 70, 80, 90, 95,100% ) each at

o 100 Ipi line rulings without microcells and 100 Ipi with microcells MC09PJH.

The two meso-patterns were included in the investigation for the purpose of reference and

comparison with the previous trial. By default microcells cannot be superimposed on halftone dots

below 20% nominal area coverage, so that only halftones of 20% upwards were taken into account

for the analysis. The plate features were characterised using WLI and this will be explained in section

6 .2 .2 .1 .

8 The Kodak plate was excluded from the study, since the proprietary Kodak Flexcel NX imaging technology did
not allow the superimposition of microcells in the image.

| 196
 U"rMf tHfrrforPrintingand(oattnn

A /
WCPC( imntfanArpaffnaChamnCymru

Figure 6.25: Image for microcell trial on the T-Flex 508 (image parts contained in red boxes taken into account for image
analysis: red boxes not included in actual image) (0.35x magnification)

The visual inspection o f the prints was perform ed on the Keyence digital microscope. Based on the

experiences w ith m eso-patterns reproduced on the T-Flex 508, it was deemed sufficient to

determ ine the optical density of the relevant image elements under investigation by

spectrophotom etry (eight consecutively printed sheets fo r each experim ental num ber were

measured). Conclusions on print u n ifo rm ity were drawn from visual and optical density data. Data

was obtained fo r the halftone scales fo r all fo u r process colours CMYK, but only the analysis o f the

black prints is presented here due to sim ilarity in the results.

6.2.2 Results and Discussions
The results are presented separately fo r microcells on solids and halftones. The inform ation

obtained from the m easurem ent o f optical density is explained in conjunction w ith observations

from p rint uniform ity.

6.2.2.1 Plate Characterisation

The area coverage o f patterned solids fo r the Asahi and MacDermid plate materials was expected to

be smaller than the nominal value due to oxygen in hibition during the polym erisation process. On

the Asahi plate the three smaller microcells patterns exhibited significant increases in area coverage

(Figure 6.26) which was the result of insufficiently cross-linked polym er molecules on the plate

surface being eroded during washout. Thus the depth o f the m icrocells was decreased (Figure 6.27)

and, taking into account the higher surface roughness o f the AWP-DEF m aterial, the area coverage

increased. The MC09P_H and MC09P_L microcells w ere shallow such th a t peaks and troughs

became com parable w ith surface roughness, thus approxim ating 100% area coverage. The

smoothness o f the MacDermid m aterial did not allow the capture o f WLI data from the cell

shoulders, therefore the area coverage determ ined was significantly smaller than fo r the other tw o

plate materials. On the MacDermid Lux plate the surface remained intact and the microcells were

distinctive (Figure 6.27). Actual and nominal area coverage alm ost matched.

■ Nominal ■ Asahi ■ M acD erm id ■ M acD erm id Lux

Chequer45 Grid MC09P_H MC9P_L MC12P MC16P

Figure 6.26: Actual area coverage of micro- and m eso-patterns on printing plates investigated on th e T-Flex 508

| 198
 MacDermid M acDerm id Lux

W id th (p m )

Figure 6.27: Surface profile of solid superimposed w ith MC09P_L microcell pattern (missing WLI data on microcell
shoulders of M acDerm id and M acDerm id Lux plates was interpolated from captured data at peaks and troughs)

Examples of microcells superimposed on a halftone dot are shown in Figure 6.28. The

superim position o f microcells on to halftones had varying effects on dot geom etry fo r the different

plate m aterials (Figure 6.29). On the Asahi plate the masked areas used to create microcells during

plate-im aging reduced the oxygen inhibition. Improved polym erisation o f the plate surface took

place which decreased the removal of plate m aterial during washout. The resultant dot geom etry

shifted from convex and relatively fla t dots to more concave w hich exhibited the effects of polym er

shrinkage9. M idtones and shadows exhibited erratic changes in cupping depth due to the num ber

and placem ent of microcells on the dot. The algorithm s in the imaging software align the centre of

the microcell array w ith the centre o f the halftone dot. Then the dot surface is progressively filled

w ith cells from the centre outwards to the dot edges. The algorithm s prohibit the coincidence of

m icrocell and dot edge in which case no fu rth e r cells are placed. In this way the distance between

the microcell and dot edge varies (Figure 6.28). M icrocells placed very close to the dot edge hamper

polym erisation, decrease the edge height and thereby reduce the height difference of edge to

centre (i.e. cupping depth). The fu rth e r away from the edge the microcell is located, the better the

edge form atio n and deeper the cupping.

On the tw o MacDermid plates the microcells had less effect on the cupping (Figure 6.29), because

the height difference o f cup edge to centre o f the dot was created by a narrow er rim around the dot

circum ference which was not wide enough to experience significant changes through microcells

9 D etailed explanations o f d o t to p geom etries, polym er shrinkage and cupping are provided in section 1.2.1.

| 199
placed near the edge. In general, where the interference o f microcells w ith the dot edge could be

ruled out, there was a tendency o f slightly decreased cupping depths fo r dots w ith microcells

(independent of imaging technology). The internal tensions, created by polym er shrinkage th a t led

to cupping, were partly released by the breakup o f the dot surface w ith microcells. Less tension was

exerted on the dot surface and the cupping reduced.

Asahi MacDermid MacDermid Lux

Figure 6.28: Example of microcells superimposed on halftone dots of all three plate types
(WLI data - red colour represents raised areas; dot of 70% nominal area coverage, 100 Ipi; 150x magnification)

♦ Asahi ■ Asahi MC A M acDermid

X M a cD e rm id MC X MacDerm id Lux • M acDerm id Lux MC

2.5

2.0

1.5

1.0

? 0.5
3 .
.c 0.0
Q.
10000 20000 300 R) 40000 000 ^ 60000 70000
-S -0.5
Q.
<3 -i-o i* k
-1.5
X A
^ X*
- 2.0

-2.5

-3.0
Planar surface area (pm 2)

Figure 6.29: Effect of microcells on cup depth of halftone dots on plate ("M C" denotes microcells)

The planar surface area o f dots w ith microcells could not be determ ined consistently fo r all plate

types. The erosion of the dot surface on the Asahi and MacDermid plates due to oxygen inhibition

rendered the WLI data points of microcells, cupping bottom and rim too similar in height, and no

clear peak was visible in the height histogram which to use as reference fo r the separation of the dot
to p and shoulder. If the microcells were included in the planar surface area, the area coverage

obtained showed no difference between halftones w ith and w ith o u t microcells (Figure 6.30) despite

th e ir influence on top geom etry (Figure 6.29). An exclusion o f the microcell area was only possible

fo r the M acDerm id Lux plate and showed increasing coverage losses w ith halftone coverage (Figure

6.31).

♦ Asahi ■ Asahi MC A M acDermid

X M acDermid MC X M acDerm id Lux • M acDermid Lux MC

100

90

<u
00
ro
§| 50
o
ro 40
<u
< 30
20

0 20 40 60 80 100
Nominal area coverage (%)

Figure 6.30: Area coverage of halftone dots w ith and w ith o u t microcells based on dot edge (MC denoting microcells)

♦ W ith o u t microcells I W ith microcells

100
90
80
sP
70
0)
QO 60
ro
Ol 50
>
o
u 40
ro
30
<r
20
10
0
20 40 60 80 100
Nominal area coverage (%)

Figure 6.31: Area coverage of halftone dots w ith and w ith o u t microcells based on actual data points
on M acDerm id Lux plates

| 201
6.2.2.2 Micro cells on Solids

This section details the findings on the effect and functiona lity of microcell patterns superimposed

on solid printing patches. The printing experim ent was analysed for optical density (shown in Figure

6.32 to Figure 6.36) and print unifo rm ity (shown in Figure 6.37 to Figure 6.40). The tw o meso-

patterns (grid and chequer at 45° tilt) perform ed worse than the microcells patterns and reference

solid under all printing conditions (refer to Appendix A .10). The reason was the same as fo r the

pattern failure in the m eso-pattern trial conducted on the T-Flex 508 and studied in the previous

chapter: the area coverage o f the patterns and the printing conditions were insufficient to form

closed ink film s on the substrate. These tw o patterns were therefore elim inated from the analysis.

Figure 6.32 shows th a t the optical densities o f the plain solids were very similar fo r all plate types

under the given printing conditions providing an equal point o f reference. The follow ing graphs

(Asahi - Figure 6.33, MacDermid - Figure 6.34, MacDermid Lux - Figure 6.35) display the difference

in optical density between the plain solid reference and the microcell patterns (negative values

correspond to higher optical density achieved through microcells). A difference of AD =0.1

approxim ates the lim it o f discernibility fo r the untrained eye. W ithin the exploration of process

param eters, it was found th a t fo r all microcell patterns and plate materials either no significant

density difference could be observed between the plain reference solid and microcell patterns, or

the solid achieved higher optical densities, because the microcell patterns led to increased UCAs in

the prints (refer to Appendix A.10).

l Asahi ■ MacDerm id ■ M acDermid Lux

1.6
1.4

X 12
l
c
-g 0.8
2 0.6
4-»
Q.
O 0.4

0.2
0
Solid MC09P H MC09P_L MC12P MC16P
Patterning

Figure 6.32: Comparison of optical densities for plain solid reference and microcells patches for all plate materials
(printing speed of 150 ft/m in at 4 thou engagem ent)

| 202
 0 .3 5

0.3

X 0-25

c 0.2
M C 09P _H
It 0.15
MC09P_L
£ 0.1
MC12P
o 0.05
MC16P

-0.05
150_2 150_3 150_4 300_2 300_3 300_4
Printing conditions

Figure 6.33: Difference in optical density betw een solids w ith o u t and w ith microcells on Asahi plate fo r different printing
conditions (printing conditions: "speed (ft/m in )_en g ag em en t (thou)")

0.35

0.3

c
k<u
-
0.25

0.2
I t
rrn i
MC09P_H
Q)
It 0.15 I T I F T I H
TJ MC09P_L
0.1
MC12P
q 0.05
MC16P
0 i t
-0.05
150_2 150_3 150_4 300_2 300_3 300 4
Printing conditions

Figure 6.34: Difference in optical density betw een solids w ith o u t and w ith microcells on M acDerm id plate for different
printing conditions (printing conditions: "speed (ft/m in )_ e n g a g e m e n t (thou)")

| 203
 M C 09P _H

MC09P_L

MC12P
m i i 1 1 i n i i MC16P
0

-0.05
150_2 150_3 150_4 300_2 300_3 300_4
Printing conditions

Fijure 6.35: Difference in optical density betw een solids w ith o u t and w ith microcells on M acDerm id Lux plate for
different printing conditions (printing conditions: "speed (ft/m in )_en g ag em en t (thou)")

1.35

1.3

t 1-25
♦ Asahi

1.2 ■ M acDermid

▲ M acDerm id Lux
1.15

1
0 20 40 60 80 100
Area coverage (%)

Figure 6.36: Correlation of actual area coverage and optical density for the microcell patterns for all plate m aterials
(printing speed of 150 ft/m in at 4 thou engagem ent)

The correlation o f the microcells area coverage on the printing plate and the print optical density

confirm s the findings from the IGT-F1 study. Since the UV-curing ink is only held on the pattern

features, the optical density increases w ith area coverage (Figure 6.36). The advantage o f the Asahi

plate was the reproduction of shallow microcells (see Figure 6.27) which made them more likely to

be inked up fully by the anilox, thereby achieving higher area coverage in the prints. The slight trend

discrepancy in the data o f the MacDermid plate is due to the challenges encountered during the

d eterm ination o f the area coverage from WLI data fo r this material (refer to 6.2.2.1). The true area

| 204
coverage m ight be larger than the determ ined one, thereby placing the outlying MacDermid data

points closer to the range described by the Asahi and MacDermid Lux plates (Figure 6.36).

Figure 6.37 shows prints of the reference solids w ithin which UCAs are present. The discontinuities

in the ink film m ight have been caused by dew etting o f the substrate, but the surface energy o f the

substrate (43.2 m N /m ) was sufficiently larger than the ink surface tension (32.4 m N /m ) to prom ote

good w etting. It was therefore most likely th a t the UCAs were caused by air bubbles entrained into

the printing nip. The bubbles were trapped during the contact between the plate and substrate,

because the non-porous film substrate provided no means by w hich the air could escape, thereby

preventing ink transfer.

Legend Asahi

defect
;b
• • *

/ * • -IT *

inb film
P IK*
1 %.
Us* t l *f
mm * e-

*
•♦
* %
*

•4
iiLjijS •*
* S5 ■ '• -
• . • * *
I « &4
P i
\
m
;
l-
*♦
r f .f, V f » * -* „
* e
a-’-- ■
*

^K2S0.00um (Pt
*1 *

MacDermid MacDermid Lux

Figure 6.37: Printing defects (UCAs) in printed reference solids caused by air en tra p m e n t
(printing speed of 150 ft/m in , 2 thou engagem ent)

Although fairly random in size and distribution, the UCAs were distinctive fo r the diffe re n t plate

materials. The Asahi print stood out from the others, because a more regular pattern was

recognisable in the defect distribution. The air bubbles were orientated along lines parallel to the

| 205
print direction (Figure 6.38). The average line pitch was 69 pm. This correlates very well w ith the

striae found on the printing plate10 which were also aligned in print direction at a pitch o f about

65 pm (this distance was measured w ith the plate lying flat and increased slightly through surface

expansion during m ounting on the round printing cylinder). Both MacDermid plates also exhibited

striae, but since these were orientated perpendicular to the print direction, they might not have

exerted the same influence on the trapped air bubbles as was observed fo r the Asahi plate. This

raises the question w hether the striation patterns can be regarded as surface texturing in th e ir own

right, an aspect which was not part o f the current research, but m erits fu rth e r investigation.

Print (microscopic image)

1 V -.T * ,> A ,

g S *X v ® ^ * l 1 Vi
(100x m agnification)

L*ot 2 rOOxXiO ____________

Figure 6.38: O rientation of UCAs in prints along lines similar in frequency to striae on Asahi printing plate
(red lines superimposed on microscopic image to highlight alignm ent of UCAs)

Expanding on the air entrapm ent model (explained in section 4.4.2.1.2), at low engagement the

trapped air bubbles nestle into any free space created by unevenness o f the printing plate against

the substrate and very irregular UCAs are created in the prints (Figure 6.39). W ith increasing

engagement the UCAs decrease and become more regular in shape and distribution, attributed to

the air bubbles being pressed into the remaining caverns between printing plate and substrate

which are not closed up fully under pressure. M icrocells are thought to constitute such caverns and

cause a defined redistribution of the trapped air bubbles in the printing nip (Figure 6.40). The regular

frequency o f the locations to which the air bubbles are consigned corresponds to the m icrocell pitch

on the plate and this now improves p rint uniform ity. Indications o f this additional mechanism were

found on all plate materials (Figure 6.41), although it could not be observed in the samples o f the

trial conducted on the IGT-F1. The effect m ight have been masked or even com pletely erased by the

10 The explanation o f the origin o f striae is provided in section 4.1. The com plete tables o f s tria tio n images and
param eters fo r all plate types can be found in Appendix A.4.

| 206
use o f higher-volum e anilox rolls on the IGT-F1 (8 and 12 cm3/m 2) compared to the T-Flex 508

(3.32 cm 3/m 2). The low er ink volum e supplied w ould affect the ink distribution and squeeze across

printing plate and substrate.

2 thou engagement 3 thou engagement 4 thou engagement

L «ns 2 1 0 0 X I 0 0 0

Figure 6.39: Change of size and distribution of UCAs w ith increasing engagem ent. At 4 thou engagem ent the spacing of
UCAs in the prints corresponds to intervals of microcells on printing plate. (Asahi plate, MC16P, 150 ft/m in )

(a )

Air b ub ble
\ * A Microcell

Ink

Substrate

(b )

Figure 6.40: Schematic of redistribution of entrapped air bubbles by microcells; (a) air bubbles entrained into printing
nip by ink film , and (b) redistribution of entrapped air bubbles into microcells.

| 207
 Asahi MacDermid MacDermid Lux

l «nt ZIOOXIOOO

Figure 6.41: Distribution of UCAs in relationship to microcell location on d ifferent printing plate materials
(red highlight = microcell location) (MC16P pattern, 150 ft/m in , 4 thou)

6.2.2.3 Microcells on Halftones

This section details the findings on the effect o f microcells superimposed on halftone dots. The

printing experim ent was analysed fo r optical density (shown in Figure 6.42 f. and Figure 6.45) and

print uniform ity (shown in Figure 6.39, Figure 6.44 and Figure 6.46). The ink transfer from halftone

dots with microcells follow ed the same trends w ith regards to changes in printing speed and

engagement as was observed for plain halftones. Figure 6.42 and Figure 6.43 show the difference in

optical density between halftones w ith o u t and w ith microcells on the Asahi and MacDermid plates

(positive values correspond to higher optical density achieved w ith o u t microcells). The microcells

failed to im prove ink transfer under any printing condition. A consistent density difference existed

fo r haltones o f 20% to 60% nominal area coverage w ith sim ilar values fo r all printing conditions. On

both pates a maximum difference was achieved fo r dots o f 70% nominal area coverage. The

shadovs produced a large variety o f results depending on plate material and printing conditions.

| 208
 0.20

♦ 150_2
c
<u
k.
0.15
CD
■ 150_3

-a A 150_4
>
> 0.10
X 300_2

0.05 X 300_3

• 300 4
n nn
0 20 40 60 80 100
Nominal area coverage (%)

Figure 6.42: Difference in optical density betw een halftones w ith o u t and w ith microcells (Asahi plate)
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

0.25

0.20

♦ 150_2
5 0.15
QJ ■ 150_3
v fc
-a A150_4
> 0.10
c
cu I X 300_2
Q
0.05 I X 300_3

• 300 4

0.00
20 40 60 80 100
Nominal area coverage (%)

Figure 6.43: Difference in optical density betw een halftones w ith o u t and w ith microcells (M acD erm id plate)
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

The failure of the microcells on these materials was attributed to tw o mechanisms. Firstly, the actual

cell valleys did not ink up fully when in contact w ith the anilox. The decrease in ink volum e supplied

reduced the overall ink spread on the substrate (Figure 6.44). UCAs created by microcells could only

be partially closed by ink squeeze. This accounted fo r the maximum difference in optical density fo r

dots of 70% nominal area coverage which marked the transition between m idtones and shadows.

W ith o u t microcells the ink volum e carried by the 70% dots was large so th a t by ink squeeze the dots

joined together on the substrate and a higher optical density was created. The reduction in ink

| 209
transfer w ith microcells allowed no or little joining of printed dots, thereby lowering optical density.

Secondly, microcells constituted additional sites at which air could be entrained into the printing nip.

The air was not sufficiently redistributed in the shallow microcells o f the Asahi and MacDermid plate

(see also Figure 6.44), thereby hampering ink transfer, creating UCAs and decreasing optical density.

W ithout microcells W ith microcells

Asahi

MacDermid

me,:;;::.!
Figure 6.44: Comparison of print uniform ity produced by halftone dots imaged w ith o u t and w ith microcells on Asahi and
M acD erm id plate materials (70% nominal area coverage; printing speed of 300 ft/m in at 4 thou engagem ent)

The MacDermid Lux plate was the only one able to im prove optical density w ith the use of microcells

(Figure 6.45). For the halftones from 20% to 60% nominal area coverage, the density difference

changed continuously exhibiting the benefit o f microcells. Between 70% and 80% nominal area

coverage the microcells became less effective, before regaining th e ir efficiency at 90% coverage. At

95% the perform ance becomes reversed and the plain halftone w ith o u t microcells outperform s its

patterned counterpart. This results pattern was repeatable fo r all printing conditions and may be

a ttrib u te d to several mechanisms.

| 210
 0 .2 0

0.15

♦ 150_2
S 0.10
■ 150_3

T3 A150_4
> 0.05
c X 300_2
<u
Q
0.00 t X 300_3
y y i 1 • 300 4

-0.05 jl J
20 40 60 80 100
N o m in a l a re a c o v e ra g e (% )

Figure 6.45: Difference in optical density betw een halftones w ith o u t and w ith microcells (M acD erm id Lux plate)
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

It was shown in section 4.4.2.1.3 th a t the plain halftones printed by the MacDermid Lux plate

suffered from th in residual film s around the printed dot perim eter w ith increasing size (Figure 6.46).

Dewetting was identified as one potential cause for these films which did not contribute to the

optical density. Halftones w ith microcells also resulted in residual films, but they were in te rm itte n t

and less extended (Figure 6.46). The microcells possibly interrupted the dew etting mechanism acting

tow ards the centre o f the dot by causing the break-up o f the ink film into m ultiple filam ents which

locally fixed m ore ink around the dot perim eter. The effect was an increase in printed area which

actually contributed to optical density. This was also observed at 70% and 80% nominal area

coverage. However, the net effect was reduced by the overall decrease in ink volum e transferred

(see above fo r explanation on 70% dots on Asahi and MacDermid plate).

W ithout microcells W ith microcells

(ESULHO taaa

Figure 6.46: Comparison of print uniform ity produced by halftone dots w ith o u t and w ith microcells on M acDerm id Lux
plate m aterial (50% nominal area coverage; printing speed of 300 ft/m in at 4 thou engagem ent)

I 211
At 90% nominal area coverage the circum ference cf the non-printing areas and thereby any

dew etting edge effects become less significant in comparison w ith the size of the printed area. UCAs

hampering optical density and p rint unifo rm ity gain importance. Therefore, it is thought that the

governing mechanism fo r print quality im provem ent changes to the redistribution of entrained air

by the microcells. The microcells took up the trapped air in th e ir recesses (Figure 6.40) which led to

very small, less visible UCAs and im proved optical density. For nominal area coverages higher than

90% this turned into a disadvantage, because the cumulative area o f the UCAs caused by entrained

air and microcells on patterned surfaces exceeded the area of UCAs caused by entrapped air on plain

surfaces.

Overall these findings could be interpreted as supportive o f the industrial claim th a t microcells have

to be imaged by "fla t-to p " imaging technology, because only the "fla t-to p p e d " MacDermid Lux plate

achieved im proved optical densities w ith microcells. Yet, a direct comparison o f the optical densities

achieved by the three plate materials w ith and w ith o u t microcells (Figure 6.47) shows th a t there was

very little difference between the plate types. It was the plain Asahi halftones which achieved

slightly be tte r optical densities than the oth er plate types. This emphasised again the significance of

plate com position and topography over imaging technology and surface patterning.

♦ Asahi ■ Asahi MC A M acDermid

X M acDerm id MC X M acDerm id Lux • M acDermid Lux MC

1.4

1.2

1.0
♦
|<u 0.8
TJ
s 06 hr
♦-»
Q.
° 0.4

0.2
k +i
, 1
0.0
20 40 60 80 100
Area coverage (%)

Figure 6.47: Optical densities of halftones w ith o u t and w ith microcells for all plate materials (printing speed of
300 ft/m in at 4 thou engagem ent; conditions chosen for highest optical density achieved in print trial)
6.2.3 Conclusions
As for the meso-patterns studied on the T-Flex 508, microcells failed to achieve improved print

quality under the given process conditions. It was confirmed that the optical density of solids and

halftones was mainly dependent on the area coverage on the printing plate. All microcell patterns

decreased area coverage and with it optical density. Of the microcells, the pattern of smallest and

least frequent recesses achieved the darker prints. Only for selected midtones on the MacDermid

Lux plate were microcells able to achieve higher optical densities than their plain counterparts. This

was thought to be caused by microcells preventing the dewetting of the printed dot perimeter.

In the previous chapter it had been identified that UV-curing ink remains on the top of plate

features. Ink squeeze across the substrate is required to coalesce ink depositions and create closed

ink films. Insufficient ink squeeze results in UCAs, additional to those created by air entrapped in the

printing nip. In general, the print uniformity of halftones with microcells was not only hampered by

the pattern recesses acting as non-printing area, but also by them constituting additional cavities

which allowed air to be entrained into the contact zone. If sealed off by the cupping of the dot top,

this air caused further defects of the printed dot.

For the solid printing patches superimposed with microcells an additional mechanism to the already

established pattern functionality could be observed. That this had not been apparent on the IGT-F1

might be due to the reduction in available ink volume on the T-Flex 508 affecting the ink distribution

on the printing plate and ink transfer. On the T-Flex 508 the microcells improved print uniformity by

acting as redistribution sites for the entrapped air. Air bubbles, which would otherwise have caused

UCAs in the prints, were shifted into the pattern recesses where they had less adverse effects on

optical density and print uniformity.

The industrial call for imaging of microcells with "flat-top" technologies for improved optical density

was shown to be unfounded. In the case of solid printing patches the optical density was primarily

subject to highest area coverage which was achieved by "flat-" or "round-top" imaging technology

depending on pattern. Very little advantage of "flat-top" technology was observed for halftones

superimposed with microcells. Under the given combination of material and process parameters

employed in this study, the overall best optical densities were achieved by plain printing features as

these provided the highest area coverage.

| 213
6.3 Closure
A comprehensive study of the influence of m'crocell patterning on ink transfer of solids and

halftones has been undertaken. It showed that m'crocells, just like micro- and meso-patterns, hold a

dual functionality dependent on ink type. An additional mechanism was identified by which

microcells act as redistribution sites for air entrained into the printing nip. The relocation of

entrapped air into the pattern recesses by a squeeze action decreases the number of UCAs caused

by air bubbles in the prints, thereby improving print uniformity. This has not been reported

previously.

With regards to the aims of achieving higher optical density and improved print uniformity, surface

pattern recommendations can be expressed based on the dual pattern functionalities:-

• Ultimately, plain printing patches achieve the highest print quality with UV-curing ink. The

decreased area coverage of surface patterns reduces the ink volume transferred, and

pattern recesses create UCAs in the prints. The least detrimental effect on print quality had

microcell and other micro-patterns comprising high area coverage, small feature size and

pitch, because these provide the best conditions for ink film coalescence under squeeze

action.

• Using water-based ink, it is possible to increase ink transfer without significantly

deteriorating print uniformity by employing microcells or other hole patterns (independent

of recess geometry), as long as the area coverage remains higher than 75%. However,

printed patterns in the form of fingering instabilities cannot be avoided entirely.

Several questions concerning the detailed mechanisms of ink transfer to and from the pattern

patches remain and merit future exploration. Until then it continues to be advisable for printers to

perform a trial run with a selection of surface patterns under production conditions to identify the

optimum surface patterning for the respective print job. The guidelines provided above will help to

make an informed decision whether any profit might be gained from surface patterning and narrow

down the selection of patterns.

Above all, the potential of plate chemistry and topography, two factors which are already an

inherent part of the printing process, should not be neglected as their optimisation might hold the

key to improvements in print quality beyond the possibilities of surface patterning.

| 214
Chapter 7 Conclusions and Recommendations
The market share of flexography in graphic and functional device printing is increasing, but several

inherent problems to the printing process are currently setting limits to its expansion, one of which

are printing defects caused by interaction with the printing plate. The printing industry addressed

the issue by introducing solutions such as "flat-top" imaging technology and surface texturing to

printing plates. Selected industrial print trials demonstrated that these innovations are able to

resolve some of the issues under particular process conditions. However, the underlying

mechanisms were not understood. A review of the literature available revealed a big gap in this

subject area and insufficient information transferrable from related research fields. This work has

contributed to the body of knowledge by systematically investigating the role of dot geometry and

surface patterning on printing plates in ink transfer. The quantification of the dot top geometry

presented is the first of its kind.

To achieve this, print trials were conducted under laboratory and near-industrial conditions.

Different types of printing plates imaged by standard digital and new "flat-top" imaging technology

were employed to investigate ink transfer from varying dot geometries and surface patterns (meso-

and micro-scale designs). New procedures to characterise features on the printing plate and in the

prints by WLI and image analysis were developed to meet the special requirements of the materials

used. During the investigation of dot geometry, the surface area and geometry of the halftone dot

was related to the size, volume and uniformity of the printed dots produced under varying printing

conditions. The effect of dot uniformity on the validity of selected halftone models was

demonstrated. Surface patterning of the printing plate was studied under the aspect of pattern and

process parameters affecting optical density and print uniformity.

The following sections contain the conclusions obtained from this research in answer to the

knowledge gaps identified in the literature review, as well as recommendations for future work in

this field and better industrial practice.

7.1 Conclusions
D o t G eometry

• Previous observations that oxygen inhibition of the standard digital imaging process creates

area coverages smaller than nominal were confirmed using Asahi and MacDermid plate

materials. The largest losses occurred in the midtones and increased with line ruling.

Halftones imaged with "flat-top" imaging technology using a Gaussian laser beam

| 215
 (MacDermid LUX) had area coverages closest to nominal with only small plate gain in the

midtones. SQUAREspot laser beams (Kodak) resulted in intermediate area sizes and a saw

tooth-like pattern around the circumference of the halftones, whereas the aforementioned

imaging technologies produced circular dots. The geometry of the halftone circumference

directly influenced the edge stability of the printed dot. The tooth-like edges of the Kodak

halftones produced the least stable dot edges, characterised by intermittent thin residual ink

films, which are of particular significance to functional device printing. No indications were

found that the ragged edges render the dot more susceptible to fingering instabilities during

printing.

• The predominant dot geometry identified on printing plates imaged with standard digital as

well as "flat-top" imaging technology was concave, most probably due to a polymer

shrinkage mechanism. The new term of cupping was suggested to describe this geometry

and a method based on cup depth introduced for its quantification. The cupping depth was

found to be more dependent on the composition of the plate material than the imaging

technology. The cupping effect is most pronounced in the midtones, which have the largest

free bulk volume deformable under tension. Only on the water-washable Asahi plate a

transition from convex to concave dot tops took place in the midtones, because the

insufficient polymerisation of the dot edges under oxygen inhibition led to their subsequent

washout.

• In conjunction with the cupping, a third mechanisms of dot deformation on the printing

plate, besides shoulder barrelling and top expansion, was identified: sealing off the ink

within the cup at the dot top creates hydraulic pressure which depresses the centre of the

dot and causes the cupping brim to fold in on itself. Ink atop the cup brim is displaced

outward and into the cup leaving an ink-free annulus on the substrate which is the origin of

the halo defect. If air bubbles as well as ink are sealed off inside the cup, the bubbles

prevent ink transfer creating UCAs in the prints.

• A rise in printing pressure led to increased dot gain from plate to print due to increased

deformation of the halftones on the plate and ink squeeze. The dot gain was smaller for the

plates produced by "flat-top" imaging technology, because they experienced less dot

shoulder barrelling and thereby increase in printing area under pressure. The standard

digital plates resulted in higher dot gain due to enhanced ink-buildup around the dot top

facilitated by ink migration down the dot shoulders. For the first time it had been reported

that the ink volume transferred is solely dependent on the size of the printed area and not

the printing area on the plate. The mean ink film thickness across the printed dots lies

| 216
 between 0.7 and 0.8 urn for all plate materials. Ink volumes containing printing defects,

notably halos, UCAs and unstable geometry of the dot edges, resulted in lower optical

densities for halftones. The effect of printing speed could not be observed in isolation, as it

was masked by the ink buildup around the halftones.

• Five halftone models were investigated for their precision in predicting halftone reflectance

from the actual halftone coverage and solid reflectance on the prints. Retrospective

optimisation rendered the Expanded Murray-Davies and Yule-Nielsen models the overall

most precise. Beer's Law proved a good fit for small, defect-free halftones as their printed

area coverage was the most similar to the one on the printing plate taken into account for

the model. The Murray-Davies model, which is most popular in industry, led to increased

errors, because it could not account for the deviation in reflectance caused by the printing

defects encountered. The Noffke-Seymour model was subject to the same problem, but

resulted in further inaccuracies due to the adjustment of the reflectance in the equation for

ink film thinning by ink spread which could not be observed in the actual prints. In general,

the validity of the halftone models was less dependent on the geometry of the dot top on

the plate than the printing defects encountered. The scale of printed halftones was divided

into sharp, uniform highlights, and flawed midtones and shadows containing UCAs and thin

residual films. Since the halftone models investigated were better in predicting either one or

the other, a new approach in the development of halftone models taking into account these

different classes of halftones is called for.

S u r f a c e P a t t e r n in g

• For the first time different functionalities of surface patterns were reported in conjunction

with the ink type used. The effect of surface patterning on print quality, notably optical

density and print uniformity, is dependent on pattern geometry, area coverage, feature size

and pitch, as well as other material and process parameters.

o Lower-viscosity water-based ink tends to flood pattern recesses which thereby act

analogous to anilox cells and increase the ink volume transferred. The higher the

recess volume, the more ink can be transferred and the higher the optical density of

the prints. However, the recess volume shall not exceed the volume of ink supplied

by the anilox, as this would prevent the pattern recesses to be filled completely and

cause UCAs in the prints. The appearance of fingering instabilities in the prints can

be controlled by the surface pattern parameters. Microcell and other hole patterns

| 217
 with area coverages above 75% achieved print uniformity comparable to the plain

solid, but with improved optical density,

o Higher-viscosity UV-curing ink remains on the pattern features. For the creation of

closed ink films in the prints, the most beneficial surface patterns are microcells and

other micro-patterns with high area coverage, small feature size and pitch as they

aid ink coalescence. Failure to coalesce leads to UCAs in the prints,

o An additional mechanism by which microcell patterns improve optical density and

print uniformity on the T-Flex 508 was by acting as redistribution sites for entrapped

air bubbles. Air that would otherwise have formed UCAs in the prints and decreased

print quality is compressed into the microcells where it has less effect on ink

transfer. The smaller the opening of the microcells and the larger the overall area

coverage, the better the print quality achieved. Large microcell openings prevented

the formation of closed ink films.

• The optical density and print uniformity achieved by surface patterning on the IGT-F1

printability tester was highly dependent on printing conditions, material and pattern

parameters. The surface patterning had to be well-matched to the other parameters in

order to achieve an improvement in print quality. The optimisation of optical density

together with print uniformity was not always possible. In some cases the sole adjustment of

IGT-F1 printing conditions achieved better results than the use of surface patterning. In

general, an increase in anilox volume, printing force and speed improved optical density and

print uniformity.

• Different classes of hydrodynamic instabilities of the Saffman-Taylor type were identified in

the printing defects, namely beads, dendrites, viscous fingers and ribs. The transitions

between fingering patterns were dependent on printing conditions, material and pattern

properties. Surface patterns with extended grooves parallel to the print direction (e.g.

chequer tilted at 45°, polka dots and positive hexagons) resulted in more linear dendrites or

ribbing along this orientation, whereas surface patterns with grooves of changing direction

or orientation other than parallel to the print direction caused highly-branched viscous

fingering. In the case of the latter, the pattern geometry controlled emerging finger growth,

bifurcation and dominant frequency. The bead pattern atop viscous fingering defects

resulted from local dewetting of the ink film and formed the base of filaments during ink film

splitting. High-frequency surface patterns, which offered more sites for filament pinning,

created finer secondary filaments and more uniform prints.

| 218
• The efficiency of surface patterning in combination with flat-top imaging technology differed

for solid and halftone prints:

o For three of four microcell patterns on solid printing patches, "flat-top" imaging

technology did not improve the print quality, because it creates deeper, more

defined microcells which do not fill up with ink and cause UCAs in the prints.

Standard digital technology on the other hand results in rudimentary microcells of

only a few micrometres. These are more likely to make complete contact with the

anilox cylinder, therefore facilitating increased ink transfer with the feature tops and

recesses, while at the same time retaining the capability to act as redistribution sites

for entrapped air.

o In halftone printing, microcells reduced the area coverage on the printing plate and

thereby the volume of ink transferred. Furthermore, they acted as additional

sources for entrapped air which caused more UCAs. However, microcells imaged by

"flat-top" technology achieved an improvement in print quality for highlights and

midtones by reducing the residual ink film around printed dots. This was attributed

to microcells interrupting the dewetting of the substrate by causing ink to break up

into multiple filaments and remain pinned at the dot edge.

Further W ork
G eo m etry

• WLI was found to be insufficient to capture the full geometry of dot tops and shoulders on

the printing plate. An advanced method has to be developed in order to capture the cupping

geometry, cup and dot shoulders sufficiently, (section 4.4.1.2)

• Detailed knowledge of the deformation of concave dot tops in the inking and printing nips,

as well as the ink transfer to and from different top geometries is currently not available.

Since direct observation of the mechanisms is not possible at this size scale, numerical

modelling is suggested for the future exploration of these issues. This might also allow the

systematic investigation of the origins of printing defects, namely halos and UCAs within

dots, (sections 4.4.2.1.1 f.)

• The research question of why halftones on MacDermid, MacDermid Lux and Kodak plates

lead to widely different area transfer ratios despite very similar surface geometries remains

open. This could be the missing link between the behaviour of "flat-" and "round-top"

plates. It might require the implementation of the two processes suggested above as well as
 morphological analysis of the respective plate materials to answer the question, (section

4.4.2.2.1)

• A new halftone model which takes into account different classes of printed dots has to be

developed, (section 4.4.2.5.2)

S u r f a c e P a t t e r n in g

• Confirmation of the microcell functionality to redistribute air bubbles entrapped in the

printing nip could be obtained by printing against glass and observing the ink film formation

from the reverse of the glass with high-speed camera. This might also allow the observation

of differences in microcell functionality between patterns imaged by standard digital and

"flat-top" imaging technology, (section 6.2.2.2)

• Further exploration of surface pattern functionalities could employ WLI to determine the

amount of ink contained within pattern recesses between inking and printing, as well as

after printing of the plate. This would provide information on whether the pattern recesses

are filled completely or only partially, and how much of the ink is released from the recesses

during printing, (section 5.1.4.8)

• Certain striation patterns on the printing surface caused by the laser imaging of the mask

layer also exhibited an effect on the redistribution of UCAs. Closer investigation of the striae

as potential integrated surface texturing to the imaging process is required, (section 6.2.2.2)

• Halftones with microcell patterning showed less residual films around printed dots which

was attributed to microcells preventing dewetting. Confirmation of this mechanism might be

obtained by observing the transfer of ink from a sufficiently large model dot with patterned

surface in an extensional rheometer or similar apparatus, (section 6.2.2.3)

• The general effect of surface patterning on ink transfer mechanisms might be observable in

a radially lifting Hele-Shaw cell of which one plate is replaced by a patterned printing plate.

During the opening of the cell, ink filamentation and meniscus destabilisation can be

captured by high-speed camera. High-speed recording is also suitable for the direct

observation of the printing nip in a printing press (access permitting).

7.3 Industrial Recommendations

• Neither "flat-top" nor standard digital imaging technology can be endorsed fully. Standard

digital halftones suffer from tonal value decrease on the plate due to oxygen inhibition of

| 220
 the imaging process, but the loss can be compensated for during the preparation of the

digital data for plate-making. The dot gain from plate to print is large. "Flat-top"

technologies produce features closer to nominal size and the dot gain from plate to print is

small, varying less under different printing conditions. This stable operation latitude requires

a rethink by the printer who tends to regulate ink transfer on the press with the pressure

setting. However, "flat-topped" plates result in more printing defects and thereby cannot be

synonymous of print quality.

• The terminology of "flat-top" and "round-top" imaging technology is misleading, because

dot geometry is independent from imaging technology. The quantification of the dot

geometry, e.g. by cup depth, width, volume and rim radius, has to be added to the common

measurands of halftone dots, notably dot height, shoulder angle and planar surface area.

The method suggested in this work (based on cup depth) could serve as a starting point for

the development of an advanced method. An international standard for the characterisation

of flexographic printing plates is still missing. Its significance has to be emphasised in the

light of these findings.

• Surface patterning is not a panacea for print quality. The effect of the plate's surface

patterning on ink transfer is strongly dependent on material and process parameters. In the

closure of Chapter 6, guidelines on the use of surface pattern designs for UV-curing and

water-based ink were provided which will help to make an informed choice. In order to

avoid waste during the print run caused by prints not meeting the quality target, a

fingerprint trial should be performed in advance to identify which pattern (if any at all)

produces the print quality desired in combination with the printer's most commonly used

inks, substrates and press settings. The results can then be referred to for future print runs

to select the optimum parameter combinations.

• Fingerprints like this can also serve as reference for the calculation of parameters for

halftones models, notably the Expanded Murray-Davies and Yule-Nielsen equation. These

models are more accurate than the more widely used Murray-Davies equation.

| 221
A Appendix

A .l Macros for Image)

M acro - Planar Su rface A rea o f Is o l a t e d P r i n t i n g F e a t u r e s in Im a g e J

runf'Options...", "iterations=l count=l black edm=Overwrite");

setBackgroundColor(255, 255, 255);

setForegroundColor(0, 0,0 );

while (nlmages>0)

setTool ("point");

waitForUser("Fill area"/'Select the NON-ROI and click 'OK'");

getSelectionCoordinates(xPoints,yPoints);

x = xPoints[0];

y = yPoints[0];

floodFill(x, y);

run("8-bit");

setThreshold(0,199);

run("Convert to Mask");

setTool("polygon");

waitForUser("Select areas","Select the area to analyse and click 'OK'");

getSelectionBounds(x,y,width,height);

setKeyDown("alt");

run("Histogram", "bins=256 x_min=255 x_max=255 y_max=Auto");

waitForUser("Copy pixel count"," Has the pixel count been copied?");

runf'Close");

run("Open Next");

| 222
M acro - P l a n a r S u r f a c e A r e a o f N o n - P r i n t i n g P l a te F e a t u r e s in Im a g e J

run("Options...", "iterations=l count=l black edm=Overwrite");

setBackgroundColor(255, 255, 255);

setForegroundColor(0, 0,0 );

while (nlmages>0)

setTool ("point");

waitForllser("Fill area","Select the ROI and click 'OK'");

getSelectionCoordinates(xPoints,yPoints);

x = xPoints[0];

y = yPoints[0];

floodFill(x, y);

run("8-bit");

setThreshold(200, 255);

run("Convertto Mask");

setTool("polygon");

waitForUser("Select areas","Select the area to analyse and click 'OK'");

getSelectionBounds(x,y,width,height);

setKeyDownfalt");

run("Histogram", "bins=256 x_min=255 x_max=255 y_max=Auto");

waitForUser("Copy pixel count"," Has the pixel count been copied?");

run("Close");

run("Open Next");

| 223
A.2 Screening of Scanning Parameters
This section contains the considerations and tests performed for the screening of scanning

parameters. The specifications investigated were scan colour, resolution, temporal stability and

consistency, linearity and gamma correction, file format for data storage, and ROI size for image

analysis.

A.2.1 Selection of Scan Colour

The main purpose for the digitisation of the prints for this investigation was to use ImageJ as a fast

and simple means to gain comparable information on print uniformity and optical density from the

image histogram. The scan option of 48-bit colour was immediately discarded as it was questionable

whether it is indeed possible for the scanner to discern 281.3 trillion different colours. Furthermore,

the large amount of data generated would be very disadvantageous from the practical viewpoint of

digital data handling in ImageJ as well as in other computer programmes. This left the scan options

of 24-bit colour, 16- and 8-bit greyscale. All three options were investigated in ImageJ, and it became

apparent that the software employed certain standard procedures for the generation of the

histograms, which was not consistent for all colour options.

ImageJ converted the 24-bit colour image to an 8-bit greyscale image, before the histogram was

calculated. Therefore, the resultant histogram had only 256 levels and the advantage of the

16.8 million colour digitisation was lost. Furthermore, the colour channels scanned were RGB. Any

image manipulation would take place only within the realm of these channels. This is of less practical

significance for the field of printing, since the process colours are CMYK, and any colour analysis,

such as spectrophotometry, is more commonly performed for the CMYK than the RGB channels.

Thus the long scanning time and large disc space for the digitisation to 24-bit colour images became

rarely justifiable.

The histogram of the 16-bit greyscale image was calculated correctly for 65,536 levels, but then was

truncated for the graphical display. All levels equal to zero at either end of the histogram were

discarded. All the remaining levels and their values were proportionally distributed over 256 bins

and displayed as a histogram. By gathering several greyscale levels into one bin, the possibility for

later calculations in other software based on the histogram data was forfeited. Although, it was

possible to generate the full histogram of 65,536 levels with the help of a macro in ImageJ, this large

number of data points could not be handled by every computer programme. For example, Microsoft

Office Excel limits the number of data points displayable in a chart to 32,000 within a series.

Finally, the 8-bit greyscale image was resolved in an 8-bit greyscale histogram of 256 levels. This

number of data points was small enough to be handled by most computer programmes.

| 224
Furthermore, the colour difference between two levels is smaller than what the human eye can

easily discern, therefore rendering this approach sufficient for the field of printing.

A.2.2 Selection of Scan Resolution

Digitisation was mainly required for prints investigating surface patterning on printing plates. For the

selection of the scan resolution, the smallest feature size expected in the prints had to be taken into

account. Assuming a one-to-one transfer of the image from digital file to printing plate to substrate,

the smallest printed structures would have had a width of 25 pm (based on the original in-house

artwork for plate-imaging). According to the Nyquist-Shannon sampling theorem, the scanning

frequency had to be at least twice as high as the largest frequency encountered in the image, in

order to reconstruct the original information sufficiently from the sampled data (Unser, 2000). Using

this as a reference, the maximum scan width had to be 12.5 pm and the minimum scan frequency

800 lines/cm (approx. 2,000 Ipi).

To compare the effect of scan resolution, the same print of six surface pattern patches was scanned

at eight different resolutions (500; 1,000; 1,500; 2000; 2500; 3000; 3500; 4000 dpi) and 8-bit

greyscale. The GSL histograms were plotted normalised to pixel number, and were almost identical

for all scan resolutions. Figure A .l shows the histograms of the patch "reference solid" which were

the most divergent of all patterns. Only the resolution of 500 dpi produced slightly deviating results,

thereby prompting the use of at least 1,000 dpi. The relative standard deviation of the MGSL and

StDev did not exceed 0.2% and 0.9% respectively for any pattern. A similar observation was made by

Stahl (2013) for the Epson Perfection V750 Photo scanner. Above 2,000 dpi the scan time and file

size increased considerably. Due to memory restrictions in ImageJ, very large file sizes had to be

avoided, and the setting of 2,000 dpi was selected as the scan resolution.

A.2.3 Determination of Temporal Stability and Consistency

A series of scans was performed with the aim of analysing the scanner temporal stability and

consistency. The same print containing three pattern patches was scanned at 8-bit greyscale and

2,000 dpi resolution in several test runs, called "a" to "h" (Table A .l). The runs differed in the

number of patches captured in each scan, the inclusion of a standard Lanetta paper (a high-gloss

coated paper with a black and white half serving as a black-white reference), and the number of

consecutive scans. One pattern patch was contained in every scan and was analysed using the MGSL

and StDev of the histogram distribution.

| 225
 4000 dpi

3500 dpi

3000 dpi

2500 dpi

2000 dpi

1500 dpi

1000 dpi

500 dpi

0 50 100 150 200 250

G re y s c a le le v el

Figure A .l: Normalised histogram of GSL distribution for patch "reference solid" at different scan resolutions

Figure A.2a shows th a t the MGSL o f the scanned image decreased fo r the first six prints of run "a".

This was attrib u te d to a prolonged w arm -up period of the scanner lamp beyond the tim e which was

autom atically allocated by the scanner software. A fter th a t the MGSL increased continuously in an

almost linear fashion fo r all scan runs. This was not an indicator of scanner instability, but was most

likely caused by changes in the printed ink itself. The water-based ink used was an in-house model

liquid which contained dye. The pigments in the dye were not lig h tp ro o f and faded under the

continual exposure of the strong scanner lamp, thereby increasing the MGSL. This conclusion was

supported by the increasing StDev o f the histogram distribution w ith scan num ber (Figure A.2b). The

only exception to the linear trend was the first scan of each run. It exhibited a larger MGSL than the

rest o f the run, most likely caused by the lamp cooling down slightly between scan runs. Breaks of

about one m inute increased the MGSL by half a level and breaks o f tw o m inutes by one level (run

"h").

In order to confirm the findings o f the water-based ink and exclude fading issues, scan runs were

analysed fo r a UV-curing ink (scan runs "A " to "E" in Table A .l). Figure A.3 shows clearly th a t the

scanner stabilised afte r the initial w arm -up period, and then scanned very consistently, as long as its

lamp did not cool down between scans. The cool-down effect became obvious again in the first scan

o f each p rint run, as well as in run "E". During the last run most scans were taken at intervals of 30

| 226
seconds. The two scans which were performed about two minutes after the previous scan stood out

with slightly higher MGSL However, the standard deviation within that run was well below 0.2%.

It can be concluded that the scanner was capable of performing very stably and consistently

depending on a constant lamp temperature. Practically that required the running of the scanner

several times before the actual scanning, and a steady exchanging of samples between scans to

avoid too large an influence of the lamp cooling down.

Table A.1: Scan conditions for determination of scanner's temporal stability and consistency

Ink Run No. of patches Reference No. of scans Duration

a 1 - 12 9 min

b 3 - 16 32 min

c 1 Lanetta 16 33 min
TJ
Q
i/)J
CO d 3 Lanetta 16 53 min
_o
L.
40)
-» e 1 - 16 13 min
CO
$ f 1 Lanetta 16 32 min

g omitted * 5 min

h 1 - 12 10 min

A 3 Lanetta 15 54 min

CuO
B omitted * 50 min
c
k_
13 C 3 - 15 30 min
u1
>
Z> D 1 Lanetta 15 31 min

E 1 - 15 12 min

| 227
 161

160

159

158

155

154

153

152
0 20 40 60 80 100 120
No. of scan
(a)

30.5

♦ a

■ b

q 29.5 Ac
4->
(/>
Xd

Xe

28.5 • f

+ h

40 80 100 120
No. of scan
(b)

Figure A.2: W ater-based ink prints - MGSL (a) and StDev (b) of test scans in consecutive order

| 228
 121

120
119

118

H7 ♦ A

2 116 ■C
115 ▲D
114 XE

113

112
0 20 40 60 80
No. of scan
(a)

No. of scan

Figure A.3: UV-curing ink prints - MGSL (a) and StDev (b) of test scans in consecutive order

A.2.4 D eterm ination of Linearity and Gamma Correction

All scans were perform ed in good faith th a t if the option "no colour correction" was selected in the

scanner software, the scanner w ould indeed not apply any corrections to the image. However after

com pleting the analysis o f the scanned images, the author received a copy o f a dissertation (Stahl,

2013) in which sim ilar w ork had been done, revealing th a t the scan software applies an algorithm

which affects the linearity o f the scan system. These autom atic non-linearity algorithm s (called

gamma corrections) serve to adjust the tonal values o f an input image to the o u tp u t m edium by

taking into account the human eye's perception (Busch, 2003; Stone, 2003; Urban, 2005).

229
One possible and simple test fo r gamma correction and a subsequent restoration o f linearity was

outlined by Stahl (based on Urban, 2005). A reference greyscale is characterised w ith a calibrated

spectropho tom eter (in this w ork a Gretag Macbeth Spectrolino) and digitised by the scanner under

investigation. The grey reference used fo r this w ork was a Color Checker Classic (X-Rite, Grand

Rapids, M l, USA) w ith six fields of d ifferen t GSLs (Figure A.4). The mean greyscale level o f each patch

determ ined by scanning and image analysis was then compared to the corresponding relative

luminance value, Y (CIE XYZ colour space), obtained through spectrophotom etry. The scanner

settings tested were "no colour correction" and colour correction w ith gammas y=1.0 (linear); 1.8

and 2.2.

Figure A.4: Reference greyscale (bottom row) as part of the Color Checker Classic

0.9

_i 0 .6
in No co rre c tio n
| 0.5
Gamma=1.0
■g 0.4
Gamma=1.8

Gamma=2.2

0.1

0 0.2 0.4 0.6 0.8 1

Normalised luminance (spectrophotometer)

Figure A .5: Comparison o f normalised MGSL (determ ined by scanning) and normalised luminance (obtained by
spectrophotom etry) showing non-linear m anipulation of scanner data using "no colour correction" option

| 230
From Figure A.5 it can be seen that the "no colour correction" option actually produces a non-linear

graph similar to the one of correction y=1.8. This would correspond to the adjustment performed on

output to a Macintosh computer screen. This finding had a large impact on this work as all scanned

data was only available in non-linear form which rendered all consequent measurands calculated

from it unreliable. Since it was not feasible to re-scan the entire data with linear gamma correction,

the correction inversion had to be performed on the existing encoded data using the inversed

gamma function (based on 8-bit greyscale):

_ ( GSLi,encoded^ o r r Equation A .l
u ^ L i,linear ~ ^ 5 5 --- ) '

A more detailed analysis of the graphs in Figure A.5 showed that a gamma of y=1.8 did not fully

correct the non-linearity of the data. This might be attributed to other non-linear influences which

affect the encoded data, such as an offset in the analogue-to-digital converter or reduced input

signal range of the scanner sensors (Urban, 2005). Gammas within the range of y={1.66,..,1.71)

resulted in R2= l. This corresponded to the range in gamma determined by Stahl (2013) for the Epson

Perfection V750 Photo (y=1.68). For this work a gamma of y=1.66 was chosen, as this resulted in a

slightly better fit for lighter greyscale values. Once the linear GSLs were restored, they had to be

applied to every single histogram that had been recorded.

A.2.5 Selection of File Format

The largest prints to be scanned contained seven bands of 12 pattern patches each. The size of the

scanner bed was designed for a maximum document size of 216 by 297 mm. A single sample band

made up of 12 pattern patches was about 313 mm long and exceeded these dimensions. Cutting the

bands apart and placing them next to each other was not a viable option. Therefore, each scan run

included a group of six pattern patches only. The scanned data of the two groupings was then saved

within one multi-page compression-free TIFF-file for easier data management. Every page of the file

corresponded to a single image level. The different image levels could be viewed separately in

Microsoft Windows Photo Viewer, ImageJ and several other computer programmes. A slight

disadvantage of multi-page TIFF-files was the increased file size which made it desirable to scan

smaller ROIs. However, printed patches occasionally contained contaminations which rendered part

of the image unsuitable for image analysis. Therefore tests on scanned sample area were conducted

to determine the ROI size.

| 231
A.2.6 Selection of ROI Size
The same six pattern patches as fo r the test o f scan resolution were scanned in th e ir entirety in the

chosen colour o f 8-bit greyscale and 2,000 dpi resolution. Seven d iffe re n t ROI sizes, ranging from

100 m m 2 to 900 m m 2, were selected and applied to the scanned images (alignm ent of ROI centre

w ith centre o f scanned patch). The largest ROI size o f 25 by 25 mm was selected to be smaller than

the patch size (30 by 30 mm) in order to avoid any edge effects (uneven or faulty ink application

along the edges o f printed features) which could ham per the data collection (Bornemann, 2013;

Stahl, 2013). Then the GSL histograms (normalised to pixel number), MGSL and StDev o f the

histogram d istrib u tio n were obtained fo r all ROI sizes and compared w ith each other.

It was found th a t the ROI size had only a small impact on the histogram and resulting measurands.

The histogram distributions were very sim ilar (Figure A.6). The relative standard deviation o f the

MGSL was less than 0.9% fo r all patterns. The relative standard deviation o f the StDev was higher

w ith a maximum o f 4.0% fo r one particular pattern. This was due to several contam inations

contained w ith in the scanned area which stood out in a darker colour against the rest o f the prints

and caused slight shifts in the histograms. Depending on the ROI size used, all, part or none o f the

contam inations w ere included in the ROI, thereby having varying effects on the results. But it was

exactly for this reason th a t the ROI had to be kept small; in order to be able to avoid measurements

on contam inated areas o f the prints. Therefore a ROI o f 10 by 10 mm (787 by 787 pixels) w ith in a

scanned image o f 20 by 15mm was chosen fo r the image analysis in this work.

90000

80000

70000

— 10x10
60000
15x10
50000
15x15
40000
20x15
30000 20x20

20000 25x20

10000 25x25

100 150 200 250 300

Greyscale level

Figure A.6: Norm alised histogram of GSL distribution for patch "reference solid" at d ifferent ROI sizes
(sizes stated in millim etres for x- and y- direction of rectangular ROI)

| 232
A.3 Macro - Print Uniformity and Optical Density in ImageJ

run("Set M easu re m e nts..."," mean standard min skewness kurtosis redirect=None decimal=3");

run("Specify...", "w idth=787 height=787 x=800 y=600 slice=l centered");

while (nSlices>0)

w a itF orU se rf'M o ve ROI","Move ROI and click 'OK'");

run("M easure");

run("H istogram ", "slice");

waitForUser("Copy data","Has the data been copied?");

runf'C lose");

run("N ext Slice [>]");

A.4 Striation of Printing Plate Surface

Halftone Solid

s s
9
i

«/)
<

Figure A .7: Comparison of surface striation in halftone dots (50% dot at 100 Ipi) and solids on printing plates
( llO x m agnification) (continued on next page)

| 233
 Halftone Solid
Kodak
MacDermid
MacDermid Lux

Figure A.7: (continued fro m previous page)

| 234
 Table A.2: Distance and depth of striae

Plate type Distance (pm) Depth (pm)

Asahi 64.8 0.3

Kodak - -

MacDermid 90.0 0.3

MacDermid Lux

• m ajor striation 90.1 0.1

• m inor striation 9.8 0.04

Area Transfer Ratios

♦ 150lpi 150_1

£ 4 ■ 150lpi 150_4
<o
DC
k. A 1501 pi 300_1
0)
CO J3
C
ro X 150lpi 300_4
XIO O Ipi 150_1

• lOOIpi 15 0 4

+ lOOIpi 300_1

-lO O Ipi 300_4

0 10000 20000 30000 40000 50000
Planar surface area on plate (p m 2)

Figure A.8: Comparison of ATRs for line rulings of 150 Ipi and 100 Ipi fo r Kodak plate
(printing conditions: "speed (ft/m in )_en g ag em en t (thou)")
 ♦ 150lp 150_1

■ 150lp 150_4

A 1501 p 300_1

X 1 5 0 lp 300_4

XIO OIp 150_1

* 1 • lOOIp 150_4

+ 100lp 300_1

-lOO Ip 300 4
10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure A .9: Comparison of ATRs for line rulings of 150 Ipi and 100 Ipi for M acDerm id plate
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

♦ 150lp 150_1

■ 1501 p 150_4

A 150ip 300_1

X 1 5 0 lp 300_4

XIO OIp 150_1

• lOOIp 150_4
*11 1 •
+ 100lp 300_1

«100lp 300 4
10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure A.10: Comparison of ATRs for line rulings of 150 Ipi and 100 Ipi for M acDerm id Lux plate
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")
A.6 Volume Transfer Ratios

4.5

_ 4.0

3 3.5 ♦ 150lp 150_1

o
vp 3.0 ■ 1501 p 150_4
re
oc
53 2.5 ▲ 150lp 300_1
H-
</>
X 1 5 0 lp 300_4
kg- 20
« 1.5 XIO OIp 150_1
£
= 1.0 • lOOIp 150_4
o
>
0.5 + 100lp 300_1

0.0 -lO O Ip 300 4

0 10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure A .11: Comparison of VTRs for line rulings of 150 Ipi and 100 Ipi for Kodak plate
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

4.5

_ 4.0

I 3.5 ♦ 1501 p 150_1

o
V<P ■ 150lp 150_4
D 3.0
Q£
53 2.5 ▲ 150lp 300_1
»4-
\A
g 20 X 1 5 0 lp 300_4
k.
« 1.5 XIO OIp 150_1
E
ir • lOOIp 150_4
-o 10
> 0.5 + 100lp 300_1

0.0 -lO O Ip 300 4

0 10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure A .12: Comparison of VTRs for line rulings of 150 Ipi and 100 Ipi for M acDerm id plate
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

| 237
 4 .5

4.0

Volume Transfer Ratio (|im )

3.5 ♦ 150lpi 150_1

3.0 ■ 1501 pi 150_4

2.5 A 1501 pi 300_1

2.0 X 1 5 0 lp i 300_4

1.5 XIO OIpi 150_1

1.0 • lOOIpi 150_4

0.5 + lOOIpi 300_1

0.0 -lO O Ipi 300_4

0 10000 20000 30000 40000 50000
Planar surface area on plate (pm 2)

Figure A .13: Comparison of VTRs for line rulings of 150 Ipi and 100 Ipi for M acDerm id Lux plate
(printing conditions: "speed (ft/m in )_en gagem ent (thou)")

238
<
u
p fl
u
u

QJ
QJ

CU
+->
4-»
CS
Figure A.14: Nominal versus actual feature geometries of original image on printing plates for the meso-pattern trials on the IGT-F1
(some images not presented at real angle; 76x nominal pattern magnification; 180x actual sample magnification)

s a u n |b ; u o z u o h
s a u n |e u o 3 e ;G
s a u n le 3 ! l j a A
| 239
 | 240
* *
! * ■

iV. 4

K * mm «
i£M»

S3|OH UN* .06) J a n b a ip UN* oSt0 JanbaqD

 | 241
L * # w i
• • •
# # #
% •
# # # §

• J ?
8
W 'W 1 I # #
# # # #
f * 4 # I • •
# •
\ \ # t A * t '
# * * * i
# . ♦ ♦ A *
t * 4 # * . # _# f
1

• • • •
• • • «
I _
• « • •
■
« • • *

« • • •
~ 1 • • % m
■

r • • »
" I _ • • • • i
*

- X X X X X XI
•••
•••
•••
•••

T ttttt
bA A A A A Jrrrrrn

sal!l Slop e>||Od suoSexaq aAjijsod suoSexaq OAjieSdN

A.8 Example of Full Data Set for Main Factor and Interaction Effects
This is the exemplary data set fo r main factor and interaction effects in the experim ental plan

investigating the MGSL o f the plain solid reference in the m eso-pattern study conducted on the

IGT-F1 (Table A.3). Positive effects are set in green, negative effects in red; effects th a t could not be

recognised as significant (uncertainty o f effect AE = 0.489) are marked grey. The nom enclature of

factors involved in main fa cto r/in te ra ctio n effects is: 1 - ink, 2 - substrate, 3 - printing plate, 4 -

anilox volum e, 5 - printing force, 6 - printing speed.

Table A.3: Exemplary DoE data set of plain solid reference (m eso-pattern trial on the IGT-F1)

Mean Main effects 2-factor interaction effects

0 1 2 3 4 6 12 1 13 14
ft 5
90.375 98.319 -2.540 2.971 -21.943 -5.028 -13.031 -9.237 1.479 -4.665
2-factor interaction effects
16 23 24 25 26 34 35 36 45
15
-1.987 -11.868 -3.769 -0.199 -2.145 -2.074 -5.570 1.504 -1.192 2.800
2-factor 3-factor interaction effects
46 56 123 124 125 126 134 135 136 145
4.206 1.354 -5.632 1.552 1.216 -3.560 -4.267 2.586 0.225 0.180
3-factor interaction effects
146 156 234 235 236 245 246 256 345 346
3.613 2.379 -1.591 -1.494 1.590 -1.242 -1.381 -0.260 -0.183 1.598
3-factor 4-factor interaction effects
356 456 1234 1235 1236 1245 1246 1256 1345 1346
-1.765 -0.690 -0.165 -2.012 1.721 -0.755 -0.328 -0.656 -0.195 1.684
4-factor interaction effects 5-factor interaction effects
1356 1456 2345 2346 2356 2456 3456 12345 12346 12356
-1.459 -0.824 1.316 2.593 1.867 0.704 1.885 1.075 2.343 2.187
5-factor interaction effects 6-factor interaction effect Uncertainty of effects
12456 13456 23456 123456 AE
0.726 0.939 -0.846 -1.128 ±0.489

A.9 Printing Defects of AFP-DSFI Plate

Refer to next page.

| 242
 Speed = 0.2 m/s Speed = 0.8 m/s

Force = 50 N Force = 150 N Force = 50 N Force = 150 N

Figure A.15: Chequer pattern w ith o u t tilt on AFP-DSH plate m aterial. Strong change betw een fingering defect regimes
w ith printing conditions observed for w ater-based ink on APCO substrate (7x m agnification)

| 243
 Speed = 0.2 m /s Speed = 0.8 m/s

Force = 50 N Force = 150 N Force = 50 N Force = 150 N

Anilox volume: 8 cm3/m
Anilox volume: 12 c m 3/ m 2

Figure A .16: Polka dot pattern on AFP-DSH plate m aterial. Strong change betw een fingering defect regimes w ith printing
conditions observed for w ater-based ink on APCO substrate (7x m agnification)

| 244
 Figure A.17: Failure of meso-patterns to produce closed ink films on the T-Flex 508 (printing speed of 150 ft/m in, 4 thou engagem ent)

jije s v
p ju jjaQ D eiA i
xrr| pjiujacpeiAi
| 245
Glossary
Cupping: Top geometry of concave halftone dots which features the centre of the dot top below the
top edge.

Dot dipping: Immersion of a dot on the printing plate into the cell of an anilox roll if the cell opening
is larger than the dot.

Dot gain, optical: Apparent increase in printed area coverage of halftones due to light gathering.

Dot gain, physical: Increase in printed area coverage of halftones compared to the corresponding
area coverage on the printing plate due to plate deformation and ink spreading.

Dot hardness: -> Sharpness

Dot softness: -> Sharpness

Doughnut: (Printing defect) printed halftone dots which assume a ring-shape due to lack of ink in
their centre.

Halftone: Image element of varying size, shape and pitch into which a printing area is separated
(screened) in order to create tonal gradients in the printed image. The smallest, isolated halftones
are known as highlights, because they print faint areas. In this work this includes area coverages up
to 20%. The halftones in the middle of the range are suitably called midtones. Halftones consisting of
conjoint elements and resulting in dark printed areas are referred to as shadows (approximately 75%
area coverage and above). Unscreened areas creating 100% area coverage are called solids.

Halo: (Printing defect) ink-free annulus found in printed halftone dots.

Highlight: -> Halftone

Kiss contact or kiss engagement: Printing press setting at which printing plate and substrate on the
impression cylinder just about touch and the mean contact force between them is near enough zero.

Light gathering: Absorption of scattered photons by ink films upon exiting the substrate.

Line ruling: Pitch of halftones.

Meso-pattern: Surface patterning on the printing plate which is made up of recesses or raised
features with approximate nominal width of 50 pm; the feature geometry and pitch may vary.

Microcell: Commercial surface patterning of printing plates (EskoArtworks) which is made up of

small recesses varying in size and pitch.

Micro-pattern: Surface patterning of printing plate which is made up of recesses or raised features
with approximate nominal width of 25 pm or smaller, including microcells; the feature geometry and
pitch may vary.

Midtone: -> Halftone

| 246
Oxygen inhibition: Premature termination of a polymerisation reaction by the interference of
molecular oxygen from the ambient air with the reagents.

Plate gain: Increase in area coverage of halftones on the printing plate compared to the
corresponding area coverage in the digital image design.

Ribbing: (Printing defect) undulating ink film thickness transverse to the print direction, also referred
to as "printer's instability".

Screen ruling: Pitch of anilox cells.

Shadow: -> Halftone

Sharpness: Property of halftone dots and ink films describing the angle formed between the top and
sides of the feature surface. Perfectly sharp features have right angles. Sharpness is also referred to
as hardness in case of ink films. Hard printed dots are characterized by right angles, whereas soft
dots feature obtuse angles.

Solid: -> Halftone

SQUAREspot: Proprietary laser technology by Eastman Kodak which creates a laser beam of uniform
lower power with rectangular cross section.

Surface patterning: Secondary design superimposed on ink-transferring areas of a printing plate to

create additional surface texturing which is claimed to improve print quality.

Tonal value increase: Difference between the observed area coverage of halftones on the print and
the corresponding area coverage on the printing plate. Sum of optical and physical dot gain.

UCA. uncovered area: (Printing defect) randomly occurring small holes in printed ink films.

Viscous fingering: (Printing defect) ramified channels of lower ink film thickness transverse to the
print direction, also referred to as "Saffman-Taylor(-like) instability".

| 247
Bibliography
Abramoff, M. D., Magalhaes, P. J. & Ram, S. J., 2004. Image Processing with ImageJ. Biophotonics
International, July, 11(7), pp. 36-42.

Adamson, A. W. & Gast, A. P., 1967. Physical Chemistry o f Surfaces. New York: Wiley-lnterscience.

Anderson, J., 2009. Don't Overthink It. Flexo platemaking is complex only if you let it become so.
package Printing, March Volume.

Anderson, J. & Schlotthauer, E., 2010. Kodak's Flat Top Dots. The effect o f flexo dot shape on p late
stress and wear. s.I.:Kodak.

Andersson, C., Johnson, J. & Jarnstrom, L., 2009. Ultraviolet-Induced Aging of Flexographic Printing
Plates Studied by Thermal and Structural Analysis Methods. Journal o f Applied Polymer Science,
112(3), pp. 1636-1646.

Andreas, J. M., Hauser, E. A. & Tucker, W. B., 1938. Boundary tension by pendant drops. Journal o f
Physical Chemistry, 42(8), pp. 1001-1019.

Arney, J. S., Engeldrum, P. G. & Zeng, H., 1995. A Modified Murray-Davies Model of Halftone Gray
Scales. 1995 TAGA Proceedings. TAGA, pp. 353-363.

Aspler, J. S., 2004. Print Quality Test Method for Linerboard. In: 2004 TAGA Proceedings. TAGA,
pp. 427-452.

Banks, W. H. & Mill, C. C., 1954. Some observations on the behaviour of liquids between rotating
rollers. Proceedings o f the Royal Society A, 233(1154), pp. 414-419.

Barnes, H. A., 2000. A Handbook o f Elementary Rheology. Aberystwyth: University of Wales,

Institute of Non-Newtonian Fluid Mechanics.

Barros, G. G., 2004. Optical Topographical Characterisation fo r Flexographic Printability Assessment.

Karlstad: Karlstad University.

Barros, G. G., 2006. Influence o f Substrate Topography on Ink Distribution in Flexography. Karlstad:
Karlstad University.

Barros, G. G., Fahlcrantz, C.-M. & Johansson, P .-A ., 2006. Topographic Distribution of Uncovered
Areas (UCA) in Full Tone Flexographic Prints. TAGA Journal, 2(1), pp. 43-57.

Barros, G. G. & Johansson, P .-A ., 2006. Prediction of Uncovered Area occurrence in flexography
based on topography - A feasibility study. Nordic Pulp and Paper Research Journal, 21(2), pp. 172-
179.

Barros, G. G. & Johansson, P.-A., 2007. Mottle and ink distribution in full-tone flexography prints on
coated linerboard. Journal o f Pulp and Paper Science, 33(4), pp. 217-226.

Barrow, M. S. et al., 2003. A study of the tensile properties of liquids in confined spaces using an
atomic force microscope. Proceedings o f the Royal Society o f London. Series A: M ath e m a tic a l and
Physical Science, 459(2039), pp. 2885-2908.

Bassemir, R. W. & Krishnan, R., 1990. Practical Applications of Surface Energy Measurements in
Flexography. FLEXO, July Issue, n.p.

| 248
Behler, H., 1993. Die Randstruckturen von Druckpunkten - eine experimentelle Untersuchung der
Farbspaltungsstrdmung. Darmstadt: Technische Hochschule Darmstadt.

Ben-Jacob, E. e ta l., 1985. Experimental Demonstration of the Role of Anisotropy in Interfacial

Pattern Formation. Physical Review Letters, 55(12), pp. 1315-1318.

Beynon, D. G., 2007. Plate to Substrate Ink Transfer in the Flexographic Printing Process. Swansea:
Swansea University.

Birkenshaw, J. W., 2004. Printing processes. In: R. H. Leach & R. J. Pierce, eds. The Printing Ink
M anual. 5th edition. Dordrecht: Kluwer Academic Publishers, pp. 14-85.

Bohan, M. F. J. et al., 2003. Evaluation of Pressures in Flexographic Printing. In: 2003 TAGA
Proceedings. TAGA, pp. 311-320.

Borbely, A. & Szentgyorgyvolgyi, R., 2011. The influence of printing pressure on flexographic print
quality. In: N. Enlund & M. Lovrecek, eds. Advances in Printing and M edia Technology. Proceedings
o f the 3 8 th International Research Conference. Darmstadt: iarigai, pp. 87-92.

Bornemann, N., 2013. Characterization and Investigation o f Large-Area, Ultra-Thin Gravure Printed
Layers. Darmstadt: Technische Hochschule Darmstadt.

Bornemann, N., Sauer, H. M. & Dorsam, E., 2010. Thin Film Behavior after Ink Transfer in Printing
Processes. In: Anonymous. LOPE-C (Large-area, Organic and Printed Electronics Convention)
Proceedings. Frankfurt a. M., Germany, 31 May-2 June 2010. Frankfurt a. M.: LOPE-C.

Bornemann, N., Sauer, H. M. & Dorsam, E., 2011. Gravure Printed Ultrathin Layers of Small-Molecule
Semiconductors on Glass. Journal o f Im age Science and Technology, 55(4), n.p.

Bould, D. C., 2001. An Investigation into Quality Improvements in Flexographic Printing. Swansea,
University of Wales Swansea.

Bould, D. C., Claypole, T. C. & Bohan, M. F. J., 2004a. An investigation into plate deformation in
flexographic printing. Proceedings o f the Institute o f M echanical Engineers Part B: Journal o f
Engineering M anufacture, Volume 218, pp. 1499-1511.

Bould, D. C., Claypole, T. C., Bohan, M. F. J. & Gethin, D. T., 2004b. Tone Gain due to Deformation of
the Flexographic Plate. In: Anonymous. Advances in Printing and M ed ia Technology. 31 st
International Research Conference. Copenhagen, Denmark, 5-8 September 2004. Copenhagen:
Danish School of Media and Journalism.

Bould, D. C., Hamblyn, S. M., Gethin, D. T. & Claypole, T. C., 2011. Effect of impression pressure and
anilox specification on solid and halftone density. Proceedings o f the Institute o f M echanical
Engineers Part B: Journal o f Engineering M anufacture, Volume 225, pp. 699-709.

Broer, D. J., 1993. Photoinitiated Polymerization and Cross-Linking of Liquid-Crystalline Systems. In:
J. P. Fouassier & J. F. Rabek, eds. Polymerisation Mechanisms. Barking: Elsevier Science Publishers,
pp. 383-443.

Busch, D. D., 2003. M astering Digital Scanning with Slides, Film, and Transparencies. Boston: Course
Technology/Cengage Learning.

Cassie, A. B. D. & Baxter, S., 1944. Wettability of porous surfaces. Transactions o f the Faraday
Society, Volume 40, pp. 546-551.

| 249
Chadov, A. V. & Yakhnin, E. D., 1988. Influence of roughness on wetting and the transfer of liquid
from one solid surface onto the other. Colloid Journal o f the USSR, 50(4), pp. 771-775.

Chang, F.-M., Hong, S.-J., Sheng, Y.-J. & Tsao, H.-K., 2010. Wetting Invasion and Retreat Across a
Corner Boundary. Journal o f Physical Chemistry C, January, 114(3), pp. 1615-1621.

Charlesworth, K., 2008. Flexo turns to flat-top dots. PrintWeek, February, Issue 21 February, pp. 29-
BO.

Cherry, J. A., 2007. Ink Release Characteristics ofA nilox Rolls. Swansea: University of Wales Swansea.

Claypole, T. C. et al., 2008. Flexo printing of fine lines. In: 2 00 8 TAGA Proceedings. TAGA, pp. 252-
266.

Cook, B., Recchia, D. A. & Gotsick, T., 2014. M eth o d fo r improving print performance o f flexographic
printing elements. US Patent 8669041.

Cusdin, G., 1999. Introduction. In: Anonymous. Flexography: Principles and Practices. 5th edition.
Ronkonkoma: Foundation of Flexographic Technical Association, pp. 1-36.

Damroth, G. et al., 1996. The effect of UV flexo ink viscosity, anilox cell volume, and press speed on
printed density and dot gain. In: 1996 TAGA Proceedings. TAGA, pp. 86-101.

Davies, G. R. & Claypole, T. C., 2004. The Effect of Speed on Image Transfer in Rotogravure Printing.
In: Anonymous. Advances in Printing and M ed ia Technology. 3 1 st International Research Conference.
Copenhagen, Denmark, 5-8 September 2004. Copenhagen: Danish School of Media and Journalism.

de Gennes, P. G., 1985. Wetting: statics and dynamics. Reviews o f M odern Physics, 57(3, Part 1), pp.
827-863.

De Grace, J. H. & Mangin, P. J., 1983. A Mechanistic Approach to Ink Transfer: Part I: Effect o f
Substrate Properties and Press Conditions. In: W. H. Banks Jr, ed. Advances in Printing Science and
Technology. Proceedings o f the 1 7 th International Conference o f Printing Research Institutes. London,
Plymouth: Pentech Press, pp. 312-332.

De Grace, J. H. & Mangin, P. J., 1987. A Mechanistic Approach to Ink Transfer: Part II - The Splitting
Behaviour of Inks in Printing Nips. In: W. H. Banks Jr, ed. Advances in Printing Science and
Technology. Proceedings o f the 1 9 th International Conference o f Printing Research Institutes. London,
Plymouth: Pentech Press, pp. 146-161.

Debeljak, M., Hladnik, A., Cerne, L. & Gregor-Svetec, D., 2013. Use of Effect Pigments for Quality
Enhancement of Offset Printed Speciality Papers. Color Research and Application, June, 38(3), pp.
168-176.

Decker, C., 1993. New Developments in UV-Curable Acrylic Monomers. In: J. P. Fouassier & J. F.
Rabek, eds. Polymerisation Mechanisms. Barking: Elsevier Science Publishers, pp. 33-64.

Dedijer, S. et al., 2012. Flexographic ink composition and its wetting influence on flexo printing plate
and printed substrate (PE FOIL). Papnpar, 56(4), pp. 24-30.

Deganello, D. et al., 2011. Level-set method for the modelling of liquid bridge formation and break­
up. Computers and Fluids, Volume 40, pp. 52-51.

Della Volpe, C. et al., 2004. The solid surface free energy calculation. I. In defense of the
multicomponent approach. Journal o f Colloid and Interface Science, Volume 271, pp. 434-453.

| 250
DFTA, 2002. Technical comparison of different flexo plate-making processes. Flexo and Gravure
International, Volume 3, pp. 14-19.

Dube, M., Mairesse, F., Boisvert, J.-P. & Voisin, Y., 2005. Wavelet Analysis of Print Mottle, submitted
to IEEE Transactions on Im age Processing.

Dykes, Y., 1999. Flexographic Printing Plates. In: Anonymous. Flexography: Principles and Practices.
5th edition. Ronkonkoma: Foundation of Flexographic Technical Association, pp. 1-60.

Extrand, C. W., 2005. Modeling of Ultralyophobicity: Suspension of Liquid Drops by a Single Asperity.
Langmuir, 21(23), pp. 10370-10374.

Extrand, C. W. & Moon, S. I., 2008. Contact Angles on Spherical Surfaces. Langmuir, September,
24(17), pp. 9470-9473.

Fernando, H. J. S. (Ed.), 2012. Handbook o f Environmental Fluid Dynamics. Volume One: Overview
and Fundamentals. Boca-Raton: CRC Press.

Fetsko, J. M. & Walker, W. C., 1955. Measurement of Ink Transfer in Printing Coated Paper. American
Ink M aker, November, 33(11), pp. 38-42/67-69.

Fioravanti, M., 2003. The future of exposure technology. Flexo and Gravure International, Volume 1,
pp. 72-74.

Flexcor, 2012. Anilox screens. Available from: < http://www.flexcor.com.au/html/rollers/rollers.htm>

[11 February 2015]

Fouche, L. & Blayo, A., 2001. Transfer Characterization of UV Flexo Inks. In: 2001 TAGA Proceedings.
TAGA, pp. 426-443.

Fowkes, F. M., 1962. Determination of interfacial tensions, contact angles, and dispersion forces in
surfaces by assuming additivity of intermoleclar interactions in surfaces. Journal o f Physical
Chemistry, 66(2), p. 382.

Fuh, J. Y. H. et al., 1997. Post-Cure Shrinkage of Photo-Sensitive Material Used in Laser Lithography
Process. Journal o f M aterials Processing Technology, Volume 63, pp. 887-891.

Galton, D. J., 2003. A study of the effects of the process parameters on the characteristics of
photochemical flexographic printing plates. Pigment and Resin Technology, 32(4), pp. 235-247.

Galton, D. J., 2004a. Investigation into the Causes for "Mottle". In: 2004 TAGA Proceedings. TAGA,
pp. 409-427.

Galton, D. J., 2004b. A study of the effects of the process parameters on the flexographic printing
problem "dot bridging" - Part I: experimental design and linear run length. Pigment and Resin
Technology, 33(5), pp. 293-301.

Galton, D. J., 2005a. Investigation into the Causes for "Mottle". In: 2005 TAGA Proceedings. TAGA,
pp. 202-221.

Galton, D. J., 2005b. A study of the effects of the process parameters on the flexographic printing
problem "dot bridging" - Part II: experimental design and print contrast/density. Pigm ent and Resin
Technology, 34(2), pp. 94-100.

Galton, D. J. & Rosenberger, R. R., 2007. A novel solution for the characterisation of corrugated
printing quality. In: 2 0 0 7 TAGA Proceedings. TAGA, pp. 403-414.

| 251
Gay, C., 2002. Stickiness - Some Fundamentals of Adhesion. Integrative and Comparative Biology,
Volume 42, pp. 1123-1126.

Gibbs, J. W., 1874-1878. On the Equilibrium of Heterogeneous Substances. Transactions o f the

Connecticut Academy o f Arts and Sciences, Volume 3, pp. 108-248, 343-520.

Gilbert, E. D. & Lee, F., 2008. Flexographic Plate Technology: Conventional Solvent Plates versus
Digital Solvent Plates. Journal o f Industrial Technology, 24(3), n.p.

Girard Leloup, L., 2000. Statistical Analysis of the Possibilities and Limits of Flexographic Process
Modelling. In: 2000 TAGA Proceedings. TAGA, 469-486.

Good, R. J., 1992. Contact angle, wetting, and adhesion: a critical review. Journal o f Adhesion Science
and Technology, 6(12), pp. 1269-1302.

Gotsick, T., 2014. Theoretical and Empirical Models o f Print Gain. Analyzing Shoulder Angle and
Compression. Available from < http://www.flexoglobal.com/blog/2014/01/20/theoretical-empirical-
models-of-print-gain/> [20 November 2014]

Griesheimer, S., 2013. Farbspaltungsphanomene von Druckfarben an strukturierten Oberflachen am

Beispiel des rlexodrucks. Darmstadt: Technische Hochschule Darmstadt.

Gupta, C., Mensing, G. A., Shannon, M. A. & Kenis, P. J. A., 2007. Double Transfer Printing of Small
Volumes of Liquids. Langmuir, 23(5), pp. 2906-2914.

Hageman, H. J., 1989. Photoinitiators and Photoinitiation Mechanisms of Free-Radical

Polymerisation Processes. In: N. S. Allen, ed. Photopolymerisation and photoim aging science and
technology. Barking: Elsevier Science Publishers, pp. 1-54.

Hall, E. L., 1979. Computer Im age Processing and Recognition. New York: Academic Press.

Hallberg, E., Odeberg Glasenapp, A. & Lestelius, M., 2005. Quantification of Banding on Printed
Currogated Board Using Spatial Frequency Analysis. Packaging Technology and Science, Volume 18,
pp. 89-95.

Hamblyn, S. M., 2004. The Role o f the Plate in the Ink Transfer Process in Flexographic Printing.
Swansea: University of Wales Swansea.

Hargreaves, I., 2004. Ultra-violet and electron-beam curing systems. In: R. H. Leach & R. J. Pierce,
eds. The Printing Ink M anual. 5th edition. Dordrecht: Kluwer Academic Publishers, pp. 636-677.

Ha rri, L., 2009. Microscopic Studies of the Influence of Main Exposure Time on Parameters of
Flexographic Printing Plate Produced by Digital Thermal Method. Microscopy Research and
Technique, Volume 72, pp. 707-716.

Ha rri, L. & Czichon, H., 2006. Microscopic Studies of the Influence of Technological Conditions on
Technical Parameters of Photopolymer Flexographic Plates. Microscopy Research and Technique,
Volume 69, pp. 675-683.

Hecker, K. & Breitung, S., 2013. Organic and Printed Electronics. Applications, Technologies and
Suppliers. 5th edition. Frankfurt a. M.: VDMA Verlag.

Hejda, F., Solar, P. & Kousal, J., 2010. Surface Free Energy Determination by Contact Angle
Measurements - A Comparison of Various Approaches. In: J. Safrankova & J. Pavlu, eds., WDS'10
Proceedings o f Contributed Papers: Part III - Physics. Prague: Matfyzpress, pp. 25-30.

| 252
Hejduk, J., 2010. The Im pact o f Physical Properties o f Rubber Roller Surfaces on the Effectivity o f Ink
Transfer (Annotation). Pardubice: University of Pardubice.

Hernandez-Sosa, G. et al., 2013. Rheological and Drying Considerations for Uniformly Gravure-
Printed Layers: Towards Large-Area Flexible Organic Light-Emitting Diodes. Advanced Functional
M aterials, Volume 23, pp. 3164-3171.

Hikmet, R. A. M., Zwerver, B. H. & Broer, D. J., 1992. Anisotropic polymerization shrinkage behaviour
of liquid-crystalline diacrylates. Polymer, 33(1), pp. 89-95.

Holmvall, M., 2010. Nip Mechanics, Hydrodynamics and Print Quality in Flexo Post-Printing.
Sundsvall: Mittuniversitetet.

Holmvall, M. & Uesaka, T., 2007. Nip Mechanics of Flexo Post-Printing on Corrugated Board. Journal
o f Composite M aterials, 41(17), pp. 2129-2145.

Holmvall, M. & Uesaka, T., 2008a. Striping of Corrugated Board in Full-Tone Flexo Post-Printing.
A ppita Journal, 61(1), pp. 35-40.

Holmvall, M. & Uesaka, T., 2008b. Print Uniformity of Corrugated Board in Flexo Printing: Effects of
Corrugated Board and Halftone Dot Deformation. Packaging Technology and Science, Volume 21, pp.
385-394.

Holmvall, M., Uesaka, T., Drolet, F. & Lindstrom, S. B., 2011. Transfer of a microfluid to a stochastic
fibre network .Journal o f Fluids and Structures, October, 27(7), pp. 937-946.

Hornschuh, M., 2005. Diskussion uber Probleme im drucktechnischen Alltag. Auszuge aus dem DFTA-
Fruhjahrs-Symposium. Available from < http://www.manfred-hornschuh.de/downloads/
gesdok05.pdf/> [29 January 2015]

Huang, W.-X.et al., 2008. Simulation of Liquid Transfer Between Separating Walls for Modeling
Micro-Gravu re-Offset Printing. International Journal o f H eat and Fluid Flow, Volume 29, pp. 1436-
1446.

Hubler, A. C. et al., 2002. High Volume Printing Technologies for the Production of Polymer
Electronic Structures. In: Anonymous. POLYTRONIC 2002. Proceedings o f the 2 nd International IEEE
Conference on Polymers and Adhesives in Microelectronics and Photonics. Piscataway: IEEE, pp. 172-
176.

Hiibner, G., 1991. Ein Beitrag zum Problem der Flussigkeitsspaltung in der Drucktechnik. Darmstadt:
Technische Hochschule Darmstadt.

Jarnstrom, J. et al., 2007. Alternative methods to evaluate the surface energy components of ink-jet
paper. Colloids and, Volume 294, pp. 46-55.

Johnson, J., Ratto, P., Lestelius, M. & Jarnstrom, L., 2003. Dynamic Nip Pressure in a Flexographic Cl-
Printing Press. In: 2003 TAGA Proceedings. TAGA, pp. 357-374.

Johnson, J. et al., 2009. Some Properties of Flexographic Printing Plates and Aspects of Print Quality.
Appita Journal, October-December, 62(5), pp. 371-378.

Jung, Y. C. & Bhushan, B., 2006. Contact angle, adhesion and friction properties of micro- and
nanopatterned polymers for superhydrophobicity. Nanotechnology, Volume 17, pp. 4970-4980.

Kaelble, D. H., 1970. Dispersion-Polar Surface Tension Properties of Organic Solids. Journal o f
Adhesion, Volume 2, pp. 66-81.

| 253
Kang, H. R., 1999. D igital Color Halftoning. Bellingham, WA: SPIE - The International Society for
Optical Engineering.

Kaplanova, M. & Hejduk, J., 2006. Influence of Rubber Surface Tension on Ink Transfer in
Flexography. In: Anonymous. Conference proceedings o f Printing Technology SPb '06. St. Petersburg:
St. Petersburg State University of Technology and Design, pp. 59-61.

Kempa, H. et al., 2006. On the applicability of different mass printing methods for the deposition of
organic functional materials. ln:Anonymous. OEC-06. Proceedings o f the Organic Electronics
Conference. Cambridge: cintelliq.

Kempa, H. et al., 2008. Progress in Mass-Printed Electronics. In: Anonymous. TPE 08. 3 rd International
Symposium "Technologies fo r Polymer Electronics". Rudolstadt: TITK (Thuringisches Institut fur Textil-
und Kunststoff-Forschung e.V.), p. 146 ff.

Kipphan, H. (Ed.), 2001. Handbook o f Print M edia. Berlin, Heidelberg: Springer.

Kodak, 2010. Kodak DigiCap NX Screening. Achieve higher densities and greater efficiency in
flexographic printing. Rochester: Eastman Kodak Company.

Kodak, 2012. Accuracy, stability, and repeatability. Kodak SQUAREspot Imaging Technology.
Rochester: Eastman Kodak Company.

Kodak, 2013. Kodak Flexcel NX System. Anything is possible. Rochester: Eastman Kodak Company.

Kunz, W., 1975. Ink Transfer in Gravure Process. In: 1975 TAGA Proceedings. TAGA, pp. 151-176.

Kwok, D. Y., Li, D. & Neumann, A. W., 1994. Evaluation of the Lifshitz-van der Waals/Acid-Base
Approach To Determine Interfacial Tensions. Langmuir, Volume 10, pp. 1323-1328.

Kwok, D. Y. & Neumann, A. W., 1999. Contact angle measurement and contact angle determination.
Advances in Colloid and Interface Science, Volume 81, pp. 167-249.

Kwok, D. Y., Ng, H. & Neumann, A. W., 2000. Experimental Study on Contact Angle Patterns: Liquid
Surface Tensions Less Than Solid Surface Tensions. Journal o f Colloid and Interface Science, Volume
225, pp. 323-328.

Laksin, M. & Parris, J., 1997. Optimization of Ink Lay in UV Flexographic Printing. In: 1997 TAGA
Proceedings. TAGA, pp. 861-871.

Leach, R. H. & Pierce, R. J., 2004. The nature of printing ink. In: R. H. Leach & R. J. Pierce, eds. The
Printing Ink M anual. 5th edition. Dordrecht: Kluwer Academic Publishers, pp. 1-13.

Lindholm, G., Gustafsson, M., Girard Leloup, L. & Vuillermoz, S., 1996. Ink transfer in flexo evaluated
by X-ray fluorescence. In: 1996 TAGA Proceedings. TAGA, pp. 102-117.

Lindner, A., Derks, D. & Shelley, M. J., 2005. Stretch flow of thin layers of Newtonian liquids:
Fingering patterns and lifting forces. Physics o f Fluids, Volume 17, pp. 072107-1 - 072107-13.

Lippold, S. & Podlesny, J., 1998. WYKO Surface Profilers. Technical Reference M anual. Tuscon: Veeco
Metrology Group.

Liu, F. & Shen, W., 2008. Forced Wetting and Dewetting of Liquids on Solid Surfaces and their Roles
in Offset Printing. Colloids and Surfaces A: Physicochemical Engineering Aspects, March, 316(1-3), pp.
62-69.

| 254
Liu, X. & Guthrie, J. T., 2003. A review of flexographic printing plate development. Surface Coatings
International PartB: Coatings Transactions, June, 86(2), pp. 91-99.

Liu, X., Guthrie, J. T. & Bryant, C., 2002. A study of the processing of flexographic solid-sheet
photopolymer printing plates. Surface Coatings International Part B: Coatings Transactions,
November, 85(4), pp. 313-319.

Liu, X., Guthrie, J. & Bryant, C., 2004. A Study of the Relevance of Plate Quality and Print Quality to
UV Flexographic Folding Carton Printing. Surface Coatings International Part B: Coatings
Transactions, 87(3), pp. 167-174.

LPC, 2010. PRIMIR "Benchmarking and W orldwide M a rk e t Trends fo r Flexographic Printing". Austin:
LPC.

MacDermid Printing Solutions, 2011. The Effect o f M acDerm id LUX on the Solid Ink Density.
Enhancement o f EskoArtworks MicroCell Surface Patterning Technology, s.I.: MacDermid Printing
Solutions.

MacDermid Printing Solutions, 2012. M acDerm id Digital Rave. s.I.: MacDermid Printing Solutions.

MacPhee, J., 1997. A Unified View of the Film Splitting Process, Part 1. American Ink M aker, 75(1),
pp. 42-49.

MacPhee, J. & Lind, J. T., 2002. Insight Into the Relationship Between Print Density and Ink Film
Thickness. In: 2002 TAGA Proceedings. TAGA, pp. 479-496.

Mahovic Poljacek, S., Cigula, T., Tomasegovic, T. & Brajnovic, O., 2013. Meeting the Quality
Requirements in the Flexogrpahic Plate Making Process. International Circular o f Graphic Education
and Research, Volume 6, pp. 62-69.

Mathes, H., 2011. Flexo troubleshooting. Most common flexo printing issues, Part 1 /2/3. Flexo and
Gravure International, Volume 2 /3 /4 , pp. 12-13/20-22/22-23.

Mathes, H., 2012. Flexo troubleshooting. Most common flexo printing issues, Part 4. Flexo and
Gravure Global, Volume 1, pp. 10-11.

M cCloud, K. V. & Maher, J. V., 1995. Experimental perturbations to Saffman-Taylor flow. Physics
Reports, Volume 260, pp. 139-185.

M esic, B., Lestelius, M. & Engstrom, G., 2006a. Influence of Corona Treatment Decay on Print Quality
in Water-borne Flexographic Printing of Low-density Polyethylene-coated Paperboard. Packaging
Technology and Science, Volume 19, pp. 61-70.

M esic, B., Lestelius, M. & Engstrom, G., 2006b. Occurrence and Causes of Uncovered Areas in W ater-
Borne Flexographic Print on PE-extrusion-coated Packaging Papers. In: Anonymous, Proceedings o f
the International Printing and Graphic Arts Conference. Peachtree Corners: TAPPI.

Mettanen, M., 2010. M easurem ent o f Print Quality: Joint Statistical Analysis o f Paper Topography
a n d Print Defects. Tampere: Tampere University of Technology.

M<eyer, K.-H., Durholz, R. & Butterich, K., 1996. The printing plate as a quality characteristic in the
reproduction o f the printed im age in flexographic printing. Stuttgart: DFTA Technology Centre at the
T&chnical College of Printing.

Miiller, J. C. & Myers, R. R., 1958. A Photographic Study of Liquid Flow in a Roll Nip. Transactions o f
th e Society o f Rheology, Volume 2, pp. 77-93.

| 255
Miller, S. & Zmetana, S., 2005. Advanced Prepress Technologies fo r Flexographic Printing. Burnaby:
Creo Inc.

Mirle, S. K., 1989. Studies on Flexography: Characterization o f Surface, Compositional and

M echanical Properties o f Photopolymer Plates; Printability Experiments and M ath em atical Modelling
o f Flexographic Printing. Bethlehem: Lehigh University.

Montgomery, D. C., 2012. Design and Analysis o f Experiments. 8th edition (International Student
Version).Singapore: John Wiley and Sons Singapore.

Muller, P. H., Neumann, P. & Storm, R., 1979. Tafeln der m athematischen Statistik. 3rd, improved
edition. Leipzig: VEB Fachbuchverlag Leipzig.

Murray, A., 1936. Monochrome Reproduction in Photoengraving. Journal o f the Franklin Institute,
June, 221(6), pp. 721-744.

Myers, R. R., Miller, J. C. & Zettlemoyer, A. C., 1959. The Splitting of Thin Liquid Films. Kinematics.
Journal o f Colloid Science, Volume 14, pp. 287-299.

Naito, I. et al., 2004. Examining the Mechanism of Ink Transfer. In: Anonymous. Advances in Printing
and M ed ia Technology. 31 st International Research Conference. Copenhagen, Denmark, 5-8
September 2004. Copenhagen: Danish School of Media and Journalism.

Naito, I., Tsuzaki, Y., Ohto, A. & Nitta, Y., 2006. Studies of Ink Trapping III. Direct detection of small
air bubbles in ink layer. Acta Graphica, 18(1-4), pp. 19-24.

Nase, J., Derks, D. & Lindner, A., 2011. Dynamic evolution of fingering patterns in a lifted Hele-Shaw
cell. Physics o f Fluids, 23(12), n.p.

Niggemeier, G., 2002. Around the Anilox Roller. Available from

<http://www.zecher.com/fileadmin/user_upload/Knowledge/Basics/AniloxBasics/AniloxBasics_ENG.
pdf> [12 February 2015]

Novakovic, D., Dedijer, S. & Mahovic Poljacek, S., 2010. A Model for Improving the Flexogrpahic
Printing Plate Making Process. Teknicki vjesnik, 17(4), pp. 403-410.

Oakland, J. S., 1987. Statistical Process Control. A Practical Guide. London, Melbourne,
Johannesburg, Auckland: Heinemann Professional Publishing.

Oliver, J. F., Huh, C. & Mason, S. G., 1977. Resistance to Spreading of Liquids by Sharp Edges. Journal
o f Colloid and Interface Science, May, 59(3), pp. 568-581.

Olsson, R., Yang, L., van Stam, J. & Lestelius, M., 2006. Effects on ink setting in flexographic printing:
coating polarity and dot gain. Nordic Pulp and Paper Research Journal, 21(5), pp. 569-574.

Olsson, R., Yang, L., van Stam, J. & Lestelius, M., 2007. Effects of elevated temperature on
flexographic printing. In: N. Enlund & M. Lovrecek, eds. Advances in Printing and M ed ia Technology.
Proceedings o f the 3 4 th International Research Conference. Darmstadt: iarigai.

Olszak, A. G., Schmit, J. & Heaton, M. G., 2001. Interferom etry: Technology and Applications.
Available from < http://cmi.epfl.ch/metrology/files/Wyko/DOC3_AN47_lnterferometry.pdf>. [27
January 2015]

Owens, D. K. & Wendt, R., 1969. Estimation of surface free energy of polymers. Journal o f Applied
Polym er Science, August, 13(8), pp. 1741-1747.

| 256
Pearson, J. R. A., 1960. The instability of uniform viscous flow under rollers and spreaders. Journal o f
Fluid Mechanics, 7(4), pp. 481-500.

Pearson, M., 1980. n Value for General Conditions. In: 1 980 TAGA Proceedings. TAGA, pp. 415-425.

Pluhar, J., 2004. Die Suche nach dem universellen Material. Flexo + Tief-Druck, Volume 5,
pp. 103-105.

Poivet, S., Nallet, F., Gay, C. & Fabre, P., 2003. Cavitation-induced force transition in confined viscous
liquids under traction. Europhysics Letters, April, 62(2), pp. 244-250.

Poivet, S. et al., 2004. Force response of a viscous liquid in a probe-tack geometry: Fingering versus
cavitation. The European Physical Journal E, Volume 15, pp. 97-116.

Quinn, J. A., Valenzuela, D. P., Micale, F. J. & Lavelle, J. S., 1997. Does the Surface Energy of the Plate
Affect Ink Transfer? CONVERTER e Cartotecnica, 10(unk.), pp. 135-143.

Rabaud, M., Couder, Y. & Michalland, S., 1991. Wavelength selection and transients in the one­
dimensional array of cells of the printer's instability. European Journal o f Mechanics - B/Fluids, 10
(2 (Suppl.)), pp. 253-260.

Rabel, W., 1971. Einige Aspekte der Benetzungstheorie und ihre Anwendung auf die Untersuchung
und Veranderung der Oberflacheneigenschaften von Polymeren. Farbe und Lack, 77(10), pp. 997-
1006.

Reuter, K. et al., 2007. Influence of process parameters on the electrical properties of offset printed
conductive polymer layers. Progress in Organic Coatings, 58t4), pp. 312-315.

Roch, F., 2007. Lecture notes fo r "Statistische Versuchsplanung", Sommersemester 2 0 0 7 . Leipzig:

Hochschule fur Technik, Wirtschaft und Kultur Leipzig, unpublished.

Rosenberger, R. R., 2010. M eth o d of, and apparatus fo r, measuring the quality o f a surface o f a
substrate. US Patent 2010/0231708.

Rousset, E., Baudin, G., Cugnet, P. & Viallet, A., 2001. Screened Offset Plates: A Contact Angle Study.
Journal o f Imaging Science and Technology, 45(6), pp. 517-522.

Sadovnikov, A., Lensu, L., Kamarainen, J.-K. & Kalviainen, H., 2005a. Quantified and Perceived
Unevenness of Solid Printed Areas. In: A. Sanfeliu & M. Lazo Cortes, eds. Progress in Pattern
Recognition, Im age Analysis and Applications. Proceedings o f the 1 0 th Iberoamerican Congress on
Pattern Recognition, CIARP2005. Berlin, Heidelberg: Springer-Verlag, pp. 710-719.

Sadovnikov, A. et al., 2005b. Mottling Assessment of Solid Printed Areas and Its Correlation to
Perceived Uniformity. In: H. Kalviainen, J. Parkkinen & A. Kaarna, eds. Im age Analysis. Proceedings o f
the 15th Scandinavian Conference, SCIA 2007. Berlin, Heidelberg: Springer-Verlag, pp. 409-418.

Sadovnikov, A., Lensu, L. & Kalviainen, H., 2007. Automated Mottling Assessment of Colored Printed
Areas. In: H. Kalviainen, J. Parkkinen & A. Kaarna, eds. Image Analysis. Proceedings o f the 15th
Scandinavian Conference, SCIA 2007. Berlin, Heidelberg: Springer-Verlag, pp. 621-630.

Saffman, P. G., 1986. Viscous Fingering in Hele-Shaw cells. Journal o f Fluid Mechanics, Volume 173,
pp. 73-94.

Saffman, P. G. & Taylor, G. I., 1958. The penetration of a fluid into a porous medium or Hele-Shaw
cell containing a more viscous liquid. Proceedings o f the Royal Society o f London. Series A:
M ath em atical and Physical Science, 245(1242), pp. 312-329.

| 257
Sam worth, M., 2001. Flexographic Printing Plate Having Im proved Solids Rendition. US Patent
6,213,048.

Samworth, M., 2009. Printing Plates Containing Ink Cells in Both Solid and H alftone Areas. US Patent
7,580,154.

Samworth, M. & Cadogan, P., 2007. M eth o d o f Controlling Ink Film Thickness on a Printing Plate. US
Patent 2007/0002384.

Sauer, H. M., Bornemann, N. & Dorsam, E., 2011. Viscous fingering in functional flexo printing: an
inevitable bug?. In: Anonymous. LOPE-C (Large-area, Organic and Printed Electronics Convention)
Proceedings. Munchen, Germany, 28-30 June 2011. Munchen: LOPE-C.

Savage, M. D., 1984. Mathematical Model for the Onset of Ribbing. AlChE Journal, November, 30(6),
pp. 999-1002.

Selli, E. & Bellobono, I. R., 1993. Photopolymerization of Multifuntional Monomers: Kinetic Aspects.
In: J. P. Fouassier & J. F. Rabek, eds. Polymerisation Mechanisms. Barking: Elsevier Science
Publishers, pp. 1-32.

Seymour, J., 2013a. Noffke-Seymour Halftone M odel. Personal communication, 11 January 2013.

Seymour, J., 2013b. The color o f a bunch o f dots, Part 4. Available from
<http://johnthemathguy.blogspot.co.uk/2013/07/the-color-of-bunch-of-dots-part-4.html>
[5 January 2015]

Seymour, J. & Noffke, P., 2012. A Universal Model for Halftone Reflectance. In: 201 2 TAGA
Proceedings. TAGA, pp. 1-40.

Shibanov, V., 2012. Flat top dot technology spreads! Flexo and Gravure Global, 1(1), pp. 6-7.

Sievers, W., 2011. Curing o f photo-curable printing plates with f la t tops or round tops by variable
speed exposure. US Patent 2011/0104615.

Sinha, S. & Tarafdar, S., 2009. Viscous Fingering Patterns and Evolution of Their Fractal Dimension.
Industrial and Engineering Chemistry Research, Volume 48, pp. 8837-8841.

Spelt, J. K., Moy, E., Kwok, D. Y. & Neumann, A. W., 1996. The Theory of Surface Tension
Components and the Equation of State Approach. In: A. W. Neumann & J. K. Spelt, eds. Applied
Surface Thermodynamics. New York: Marcel Dekker, pp. 293-332.

Spelt, J. K. & Neumann, A. W., 1987. Solid Surface Tension: The Equation of State Approach and the
Theory of Surface Tension Components. Theoretical and Conceptual Considerations. Langmuir, 3(4),
pp. 588-591.

Stahl, S., 2013. Einfluss der Druckform und der Fluidviskositat a u fd ie Schichtdicke von im Tiefdruck
hergestellten Funktionsschichten und deren Charakterisierung. Darmstadt: Technische Hochschule
Darmstadt.

Stahl, S., Sauer, H. M. & Dorsam, E., 2012. Investigation in the homogeneity of gravure printed
polymer films for printed electronics. In: N. Enlund & M. Lovrecek, eds. Advances in Printing and
M ed ia Technology. Proceedings o f the 3 9 th International Research Conference. Darmstadt: iarigai, pp.
47-51.

Stolt, P., Zwadlo, G. L. & Rozzi, W. A., 2010. Enhanced Relief Printing Plate. US Patent 2010/0143841.

| 258
Stone, M. C., 2003. Field Guide to D igital Color. Natick: A K Peters/CRC Press.

Takimoto, Y., 1993. Comparative Advantages and Drawbacks of Radical and Cationic Polymerization
in UV Curing. In: J. P. Fouassier & J. F. Rabek, eds. Polymerisation Mechanisms. Barking: Elsevier
Science Publishers, pp. 269-299.

Taylor, J. H. & Zettlemoyer, A. C., 1958. Hypothesis on the Mechanism of Ink Splitting During
Printing. TAPPI, December, 41(12), pp. 749-757.

Theopold, A., Neumann, J., Massfelder, D. & Dorsam, E., 2012. Effects of solvents on flexographic
printing plates. In: N. Enlund & M. Lovrecek, eds. Advances in Printing and M edia Technology.
Proceedings o f the 3 9 th International Research Conference. Darmstadt: iarigai,pp. 159-167.

Thompson, B., 1998. Printing M aterials: Science and Technology. Leatherhead: Pira International.

Tian, D., Song, Y. & Jiang, L., 2013. Patterning of controllable surface wettability for printing
techniques. Chemical Society Reviews, 42(12), pp. 5184-5209.

Tomasegovic, T., Mahovic Poljacek, S. & Cigula, T., 2013a. Surface properties of flexographic printing
plates related to UVC post-treatment. Journal o f Print and M edia Technology Research, 2(4), pp. 227-
233.

Tomasegovic, T., Mahovic Poljacek, S. & Cigula, T., 2013b. Impact of Screen Ruling on the Formation
of the Printing Elements on the Flexpgraphic Printing Plate. Acta Graphica. Journal fo r Printing
Science and Graphic Communication, 24(1-2), pp. 1-12.

Toth, T. et al., 2011. Suspension of W ater Droplets on Individual Pillars. Langmuir, Volume 27, pp.
4742-4748.

Unser, M., 2000. Sampling - 50 Years After Shannon. Proceedings o f the IEEE, April, 88(4), pp. 569-
587.

Urban, P. M., 2005. M etam e re und multispektrale M ethoden zur Reproduktion farb ig er Vorlagen.
Hamburg: Technische Universitat Hamburg-Harburg.

Valdec, D., Zjakic, I. & Milkovic, M., 2013. The influence of variable parameters of flexographic
printing on dot geometry of pre-printed printing substrate [sic]. Teknicki vjesnik, 20(4), pp. 659-667.

van Oss, C. J., Chaudhury, M. K. & Good, R. J., 1988. Interfacial Lifshitz-van der Waals and Polar
Interactions in Macroscopic Systems. Chemical Reviews, 33(6), pp. 927-941.

Vlachopoulos, G., 2009. Phenomena Affecting Ink Transfer in Offset Printing. Swansea: Swansea
University.

VoB, C., 2002. Analytische Modellierung, experimentelle Untersuchungen und dreidimensionale

Gitter-Boltzm ann Simulation der quasistatischen und instabilen Farbspaltung. Wuppertal: Bergische
Universitat Gesamthochschule Wuppertal.

Walker, W. C. & Fetsko, J. M., 1955. A Concept of Ink Transfer in Printing. American Ink M aker,
December, 33(12), pp. 38-44, 69-71.

Weichmann, A., 2002. M eth od o f Varying the Ink Density o f the Full Tone in Offset Printing Within a
Rotoary [sic] Printing Machine. US Patent 2002/0152911.

Wenzel, R. N., 1936. Resistance of solid surfaces to wetting by water. Industrial and Engineering
Chemistry, 28(8), pp. 988-994.

| 259
Whyman, G., Bormashenko, E. & Stein, T., 2008. The rigorous derivation of Young, Cassie-Baxter and
Wenzel equations and the analysis of the contact angle hysteresis phenomenon. Chemical Physics
Letters, 450(4-6), pp. 355-359.

Wyatt, F. C., 2004. Flexographic inks. In: R. H. Leach & R. J. Pierce, eds. The Printing Ink M anual. 5th
edition. Dordrecht: Kluwer Academic Publishers, pp. 547-598.

Wyble, D. R. & Berns, R. S., 2000. A Critical Review of Spectral Models Apllied to Binary Color
Printing. Color Research and Application, 25(1), pp. 4-19.

Yule, J. A. C., 1943. Theory of the Halftone Process. III. Dot Size and Halftone Density. Journal o f the
Franklin Institute, November, 236(5), pp. 473-487.

Yule, J. A. C. & Neilsen, W. J., 1951. The Penetration of Light into Paper and its Effect on Halftone
Reproduction. In: 1951 TAGA Proceedings. TAGA, pp. 65-76.

Yusof, M. S., Ahmad Zaidi, A. M., Claypole, T. C. & Gethin, D. T., 2007. The Effects of Printing Plate on
the Reproduction of Fine Solid Line Printing in Flexography. In: Anonymous. Printing Future Days
2007. Proceedings o f the 2 nd International Student Conference. Chemnitz: Institute for Print and
Media Technology at Technische Universitat Chemnitz, pp. 214-218.

Zaleski, M., Schaeffer, W. D. & Zettlemoyer, A. C., 1971. An Extended Application of the Walker-
Fetsko Ink Transfer Equation. In: W. H. Banks Jr., ed. Advances in Printing Science and Technology.
Proceedings o f the 1 1 th International Conference o f Printing Research Institutes. London, Plymouth:
Pentech Press, pp. 175-184.

Zang, Y. H., 1992. Asymmetric splitting and ink transfer: a new ink transfer model. In: Anonymous.
Proceedings o f the 6 th International Printing and Graphic Arts conference. Peachtree Corners: TAPPI,
pp. 103-112.

Zenkiewicz, M., 2007a. Comparative study on the surface free energy of a solid calculated by
different methods. Polymer Testing, Volume 26, pp. 14-19.

Zenkiewicz, M., 2007b. Methods for the calculation of surface free energy of solids. Journal o f
Achievements in M aterials and M anufactoring Engineering, September, 25(1), pp. 137-145.

Zettlemoyer, A.C. & Myers, R. R., 1960. The Rheology of Printing Inks. In: F. R. Eirich, ed. Rheology.
Theory and Application. Volume 3. New York, London: Academic Press, pp. 145-188.

Zhang, L., yun Zhang, M. & ping Yang, J., 2011. Development and Comparison of Models for
Predicting Printing Quality in Flexographic Printing. Advanced M aterials Research, Volume 174, pp.
231-234.

Zumbrum, M. A., Wilkes, G. L. & Ward, T. C., 1993. Characterization of the Dynamic Mechanical and
Dielectric Properties of UV- and EB-Cured Coatings. In: J. P. Fouassier & J. F. Rabek, eds.
Polymerisation Mechanisms. Barking: Elsevier Science Publishers, pp. 101-151.

| 260