See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318294922

Tailoring Network Properties of Castor Oil-Based Binder through Factorial

Design

Conference Paper · July 2017

DOI: 10.2514/6.2017-4694

CITATIONS READS
0 84

4 authors, including:

Luciene Villar
Instituto de Aeronáutica e Espaço
15 PUBLICATIONS   88 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Study of metal bioleaching from domestic sewage sludge View project

Development of alternative polyurethanes for solid rocket propellant View project

All content following this page was uploaded by Luciene Villar on 26 December 2018.

The user has requested enhancement of the downloaded file.

 AIAA 2017-4694
AIAA Propulsion and Energy Forum
10-12 July 2017, Atlanta, GA
53rd AIAA/SAE/ASEE Joint Propulsion Conference

Tailoring Network Properties of Castor Oil-Based Binder

through Factorial Design

Lourenço H. B. Vidotto1
Instituto Tecnológico de Aeronáutica, São José dos Campos, SP, 12.228-900, Brazil

Aron E. Silvestrini2
Universidade de São Paulo, Escola de Engenharia de Lorena, Lorena, SP, 12.602-810, Brazil

Elizabete Y. Kawachi3
Instituto Tecnológico de Aeronáutica, São José dos Campos, SP, 12.228-900, Brazil
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

and

Luciene D. Villar4
Instituto de Aeronáutica e Espaço, São José dos Campos, SP, 12.228-904, Brazil
Instituto Tecnológico de Aeronáutica, São José dos Campos, SP, 12.228-900, Brazil

Properties of castor oil-based binder were evaluated by using a factorial experimental

design for assessment of two factors: molar ratio between isocyanate and hydroxyl groups
(NCO/OH), and relative mass concentration of 1,4-butanediol (BDO) in castor oil. The aim
was to obtain empirical models based on these factors that could be used for prediction and
tailoring of network properties, namely crosslinking density. To optimize modeling,
responses on mechanical and thermal properties were included. By overlaying the contour
plots of secondary-order models, two possible regions were identified inside which higher
degree of crosslinking density compared to HTPB-based binder may be obtained, with the
comparable values of tensile strength and modulus. In spite of that, some unfavorable
aspects related to the use of castor oil on binder formulation are lower elongation and higher
glass transition temperature (Tg).

Nomenclature
Ci = mass concentration of the ith component
E = Young modulus
E” = loss modulus
IOH = hydroxyl number
Mc = molar mass between crosslinks
m1,2 = mass of swollen disk before drying
m2 = mass of swollen disk after drying
Q = swelling degree
Tg = glass transition temperature
Tonset = onset temperature
tan  = damping
Tonset = onset temperature
V1 = molar volume of the solvent

1
Doctoral Student, Departamento de Ciências Fundamentais, Praça Mal Eduardo Gomes 50, Vila das Acácias.
2
Undergraduate Student, Estrada Municipal do Campinho, s/n, Ponte Nova.
3
Full Professor, Departamento de Ciências Fundamentais, Praça Mal Eduardo Gomes 50, Vila das Acácias.
4
Senior Research Engineer, Divisão de Química, Praça Mal Eduardo Gomes 50, Vila das Acácias, and AIAA
Associate Member.
1
American Institute of Aeronautics and Astronautics

Copyright © 2017 by Lourenço Henrique Bittar Vidotto, Aron Elizeu Silvestrini, Elizabete Yoshie Kawachi, Luciene Dias Villar. Published by the American Institute of Aeronautics and Astronautics, Inc., with permi
 m = mass loss on TGA
 = axial point at factorial design
 = solvent-polymer interaction parameter
rup = elongation at rupture
i = absolute viscosities of the ith pure component
m = absolute viscosities of the mixture
ν2 = volume fraction of the gel
1 = density of the solvent
2 = density of the dried swollen polyurethane
rup = tensile strength at rupture
e = crosslinking density

I. Introduction

C ASTOR oil has been used in development of solid rockets since the mid-sixties. Being a naturally hydroxylated
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

compound, this plant oil can be employed as a polyol without any further chemical modification. One of the
first reported applications of castor oil in the field of solid rocket was in replacement of polyether diol for propellant
binder and liner composition of cast-in-case motors.1
Although castor oil can be used directly as a polyol, Indian Space Research Organization (ISRO) has registered a
process designated to provide a polyurethane prepolymer with tailored molecular mass by conducting the
polymerization of 12-hydroxyestearic acid,2 a castor oil fatty acid ester. Even though, this process was aimed at
replacing hydroxyl-terminated polybutadiene (HTPB), propellants prepared with those prepolymers have not
reached the desired performance.3 Despite of that, this versatile vegetable oil continues to be used as an additive in
solid propellant formulations, either as a triol crosslinking agent3 or as a bonding agent associated with polyamine
compounds.4,5 Some studies have also indicated the use of castor oil in the binder formulation of hybrid
propellants.6-8
Castor oil contains approximately 90% of a triol
fatty acid known as ricinoleic acid (Fig. 1). It also
has some diols, which provide a molar ratio of
0.3:0.7 in diol to triol functions, respectively.3 The
presence of secondary hydroxyl groups produces a
crosslinked network when reacting with
diisocyanates. This is an interesting feature of castor
oil-based binder, since crosslinking can act as a
barrier against migration, known to be a critical issue
for propellant-to-liner adhesive bondline. In addition,
a crosslinked binder can be less prone to moisture
sorption, which may improve shelf life of solid Figure 1. Ricinoleic acid chemical structure.
rocket motors. On the other hand, high levels of
crosslinking can compromise mechanical and thermal properties, especially ultimate elongation and glass transition
temperature (Tg). To overcome this drawback, chain extenders may be incorporated into binder, in order to provide
some linear regions in the chemical structure of the polyurethane binder, resulting in more suitable final
characteristics. The knowledge of the castor oil's contribution to binder network properties can be useful for the
employment of this compound in propellant matrixes. Thus, this study was aimed to evaluate the network behavior
of castor oil-based polyurethane binder by tailoring its formulation.
With this in mind, this study has assayed the effect of two independent variables, called as factors, namely the
NCO/OH (isocyanate/hydroxyl) molar ratio, and the relative mass concentration between a diol, 1,4-butanediol
(BDO), and castor oil as a triol. A factorial design at two levels, with central and axial points was planned to
examine the network characteristics by means of mechanical, physicochemical, and thermal properties. A
compatibility study between BDO and castor oil by viscosity measurement was also carried out. HTPB-based
binders were prepared as reference materials.

2
American Institute of Aeronautics and Astronautics
 II. Experimental
Binder formulations for the factorial desing were synthesized in an one-step process, by mixture of castor oil
(Resinol EXP PU) (A. Azevedo, Brazil) with 1,4-butanediol (BDO) (Aldrich), in the presence of dibutyltin dilaurate
(DBTDL) (CESBRA, Brazil). The curing agent was isophorone diisocyanate (IPDI) (SNPE). Properties of the
polyols used are summarized in Table 1. Polyols were previously dried in a vacuum oven for at least 72 hours.
Compatibility of castor oil with BDO was verified before polyurethane synthesis. A series of mixtures of both
polyols were prepared varying from 0/100 to 100/0 parts of BDO to parts of castor oil, respectively. Viscosity
measurement was performed in these solutions (Brookfield DV-II+ Pro, spindle SC4-34) at 25oC to verify if the
additive rule, shown in Eq. (1), was obeyed.9

ln m  Ci ln i (1)

Where i and m are absolute viscosities of the ith pure component and the mixture, respectively; Ci is the mass
concentration of the ith component.
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

Table 1. Properties of the polyols.

Triol/Diol Hydroxyl Number, IOH
Polyol Designation Viscosity (cP)
(hydroxyl groups) (mmol.g-1)
Castor Oil Resinol EXP PU Triol (secondary) 669  7 2.75  0.11
1,4-Butanediol BDO Diol (primary) 68  1 21.7  0.1

The influence of the factors NCO/OH molar ratio (factor 1, designated as x1) and BDO:castor oil relative mass
concentration (factor 2, designated as x2) in tailoring of polyurethane network was evaluated through a Central
Composite Rotatable Design (CCRD) of experiments at two levels (1 and +1) with central (0) and axial points (
and +). The CCRD design is shown in Table 2. Central points were made in triplicate.

Table 2. Central Composite Rotatable Design applied to castor oil-based binder.

Level
Factor
 1 0 +1 +
NCO/OH molar ratio, x1 0.85 0.87 0.925 0.98 1.0
BDO:castor oil, x2 0 0.02 0.06 0.10 0.12

Each polyurethane formulation was submitted to mechanical and thermal characterization. Uniaxial tension test
(Zwick 1474, STANAG 4506-00) was performed in six replicate dogbone specimens (ASTM D412-15 model C) at
a velocity of 500 mm.min-1. Simultaneous TGA/DTG-DSC (TA Instruments SDT-Q600) analyses were performed
in samples of approximately 10 mg at a heating rate of 10oC.min-1 from room temperature to 800oC under nitrogen
atmosphere (100 mL.min-1). DMA measurements were carried out on TA Instruments DMA Q-800 analyzer.
Sample dimensions were approximately 35 mm x 12 mm x 2 mm. Analyses were conducted at flexure mode by
using single cantilever clamp with amplitude of 30 m and frequency of 1 Hz. Temperature range varied from 120
to +80oC at rate of 3oC.min-1. Samples for TGA/DTG-DSC and DMA were analyzed without replicates.
Crosslinking density was measured through solvent swelling of polyurethane small disks (0.25  0.05 g) in
chloroform, following an earlier-described method.10 The swelling degree (Q), molar mass between crosslinks (Mc),
and crosslinking density (e) were calculated after measuring the swelling ratio, using Eqs. (2) to (4), respectively.
The solvent-polymer interaction parameter () was assumed to be 0.228 for the chloroform-polyurethane pair.11
Swelling measurements were conducted by using five specimens for each binder formulation.

ρ2  m1,2  m2  1
Q = 1+   = (2)
ρ1  m2   2

3
American Institute of Aeronautics and Astronautics
   2V1 ( 2 3   2 )
1

Mc = 2 (3)
ln( 1   2 )   2   22

2
e = (4)
Mc

Where ρ1 and ρ2 are the densities of chloroform (1.47 g.mL-1) and dried swollen polyurethane at 25oC,
respectively; m1,2 and m2 are the mass of swollen disk before and after drying, respectively; ν2 is the volume fraction
of the gel; V1 is the molar volume of the solvent (81.21 mL.mol-1 for chloroform). Measurement of ρ2 was carried out
by application of Archimedes' principle.

III. Results and Discussion

Prior to the preparation of castor oil-based polyurethanes, the compatibility between both polyols (BDO and
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

castor oil) was verified. Figure 1 shows the experimental and estimated viscosities of BDO and castor oil mixtures
for concentrations varying from 0 to 100 % (w/w). It can be observed that additive rule (Eq. 1) is obeyed for most of
the concentration range. The experimental
linear fit has effectively the same angular 7
coefficient as the theoretical fit. However, there Experimental
ln (Absolute viscosity), cP
is a tendency of detachment from estimated Estimated
values for BDO concentrations between 25%
6
and 80% (w/w), which might indicate a lower
y = (6.64  0.08) + (0.022  0.002) x
compatibility in this region. This possible lack y = 6 .5 0  0 .0 2 3 x 2
R = 0.9502

of compatibility is, however, not very

significant, since diol/triol relative mass 5
concentration used for the purpose of chain
extender is usually lower than 1:4.
Once the compatibility between BDO and
castor oil has been verified, thus ensuring 4
homogeneous binder to be obtained, synthesis 0 20 40 60 80 100

of castor oil-based binders following factorial BDO concentration, % w/w

design (Table 2) was carried out. Table 3 Figure 2. Experimental and estimated absolute viscosities
presents the results obtained for mechanical and as a function of BDO concentration in castor oil.
physicochemical properties, the latter ones
being related to swelling tests.
Thermal properties obtained by TGA/DTG-DSC and DMA analyses are presented in Table 4. TGA data
comprises initial temperature of thermal degradation, defined as T onset; mass loss for the thermal events identified,
which were in number of three (m1, m2, m3); in addition to the residue percentage. Data on Tg are identified
according to thermal analysis used for their calculation, i.e. from DSC or from DMA, in this case from loss modulus
(E") or tan  peaks. In both tables, formulations from 1 to 4 represent the orthogonal design, 5 to 8 correspond to
axial points, and 9 to 11 correspond to central points, done in replicates in order to estimate the experimental error.
To avoid any systematic bias, all the formulations were prepared in a randomized order, which is shown alongside
formulation number in parenthesis (Table 3 and 4).
Results from Tables 3 and 4 showed significant changes in the properties following changes in x1 (NCO/OH
molar ratio) and x2 (BDO:castor oil relative mass concentration). This observation indicates that an empirical model
for prediction of response-variables can be obtained from the selected factors. In addition, a good reproducibility of
binder formulation was observed for central point replicates, confirming that the preparation process of the
polyurethane binder was well controlled.
In order to evaluate the goodness-of-fit of second-order empirical models, F-test was applied on measured
properties (Table 5). Due to the similar information obtained from analysis of Q, Mc, and e, only the last one was
selected for this statistical evaluation, and any further analyses. Concerning mass loss from TGA, only the first event
(m1), which is directly correlated to thermal stability, was considered.

4
American Institute of Aeronautics and Astronautics
 Table 3. Mechanical and physicochemical properties for factorial design of castor oil-based binder.
Factor Mechanical Properties Physicochemical Properties
Formulation rup rup  Q Mc e
x1 x2
(MPa) (%) (MPa) (adim.) (g.mol-1) (10-4 mol.mL-1)
1 (2nd) 1 1 1.04 ± 0.11 245 ± 17 0.67 ± 0.02 7.17 ± 0.07 4995 ± 96 1.73 ± 0.03
2 (8th) +1 1 3.22 ± 0.45 219 ± 11 1.85 ± 0.03 4.74 ± 0.05 2167 ± 50 3.99 ± 0.09
3 (11th) 1 +1 1.54 ± 0.44 519 ± 27 0.77 ± 0.08 15.59 ± 1.6 21434 ± 3986 0.41 ± 0.08
4 (6th) +1 +1 15.15 ± 2.32 303 ± 5 3.92 ± 0.23 5.47 ± 0.09 2915 ± 99 2.96 ± 0.10
5 (5th)  0 0.93 ± 0.14 536 ± 17 0.45 ± 0.07 15.45 ± 1.05 21000 ± 2531 0.42 ± 0.06
6 (7th) + 0 9.13 ± 2.02 273 ± 12 2.58 ± 0.06 4.49 ± 0.03 1938 ± 31 4.45 ± 0.07
7 (3rd) 0  1.34 ± 0.16 158 ± 14 1.29 ± 0.05 5.09 ± 0.04 2511 ± 44 3.44 ± 0.06
8 (10th) 0 + 8.29 ± 1.07 387 ± 6 2.02 ± 0.19 8.00 ± 0.29 6182 ± 436 1.40 ± 0.09
9 (1st) 0 0 2.26 ± 0.61 300 ± 28 1.10 ± 0.06 5.96 ± 0.06 3464 ± 74 2.49 ± 0.05
10 (4th) 0 0 2.33 ± 0.55 313 ± 29 1.05 ± 0.01 6.98 ± 0.22 4737 ± 295 1.83 ± 0.11
11 (9th) 0 0 3.49 ± 0.33 311 ± 12 1.52 ± 0.01 5.88 ± 0.08 3375 ± 88 2.56 ± 0.07
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

Table 4. Thermal properties for factorial design of castor oil-based binder.

Factor TGA Tg (oC)
Formulation Tonset m1 m2 m3 Residue
x1 x2 DSC E” tan 
(oC) (%) (%) (%) (%)
1 (2nd) 1 1 307.2 27.5 56.1 16.1 0.0 −21.1 −16.3 −2.8
2 (8th) +1 1 306.7 30.3 54.1 12.3 3.0 −12.7 7.4 4.7
3 (11th) 1 +1 307.9 43.8 46.8 8.3 0.6 −11.4 4.7 13.1
4 (6th) +1 +1 307.3 45.6 44.2 9.1 0.7 −0.7 8.0 19.3
5 (5th)  0 307.8 36.2 52.1 11.1 0.4 −18.7 11.9 5.9
6 (7th) + 0 306.5 40.4 50.6 9.1 0.0 −3.2 4.2 16.1
7 (3rd) 0  306.9 24.2 57.5 16.5 2.5 −8.9 15.4 3.0
8 (10th) 0 + 307.5 47.5 40.7 9.2 2.0 −2.5 4.2 19.6
9 (1st) 0 0 306.7 37.7 46.7 14.2 0.9 −9.8 4.0 8.7
10 (4th) 0 0 306.8 38.2 49.0 12.0 0.6 −11.9 6.7 7.9
11 (9th) 0 0 307.5 38.7 50.5 10.8 0.0 −10.7 3.6 9.4

The ratio between mean square residual and mean square error (Fcalc) was compared to Ftab for degrees of
freedom 1 = 3 and 2 = 2 at a 0.05 significance
level (F3; 2; 0.05 = 19.16). Values of coefficient of Table 5. Statistical evaluation of goodness-of-fit of
2 2
determination (R ) and maximum R , taking into second-order models.
account the discount of pure error, were also R2 max. R-sq
Response-variable Fcalc
calculated. Residual plots (data not shown) have (%) (%)
not indicated any deviation from normal rup 2.26 97.95 99.53
distribution.
rup* 19.28 97.84 99.93
Calculated F ratios (Fcalc) were lower than
tabulated values (Fcalc < F3; 2; 0.05), except for the  2.48 94.05 98.74
elongation at rupture (rup), where Fcalc > F3; 2; 0.05 e 0.38 97.11 98.16
(19.28 > 19.16). Most of the coefficients of
2 Tonset (TGA) 0.17 76.73 81.50
determination were close to maximum R (Table
5) with two exceptions: Tonset, obtained from m1 0.92 99.79 99.91
TGA curves, and Tg, obtained from DSC curves.
In spite of these statistical discrepancies, second- Tg (DSC)† 12.11 89.37 99.44
order model was considered suitable to be Tg (E")‡ 0.06 99.05 99.13
applied for prediction of properties of castor oil-
based binder. As a consequence, the effects Tg (tan ) 1.59 99.35 99.81
considered for second-order model were the * Large residue for formulation 5 (predicted value 504; residue 32).
mean of orthogonal results, the linear (L) and † Large residue for formulation 7 (predicted 12.7; residue 3.8).
quadratic (Q) effects of factors 1 and 2, and the ‡ Large residue for formulation 10 (predicted 4.8; residue 1.9).

5
American Institute of Aeronautics and Astronautics
 interaction between those factors, represented as 1 . 2 (Table 6). A significance level of 0.05 was applied. For
reasons of simplification, only the empirical model for T g obtained from E" curves is presented. Effects with
positive values indicate that change of the considered factor from low level to high level, for instance, from 1 to
+1, increases the response value. The opposite occurs with negative effects. However, it is important to consider that
final predicted value for a given property is the sum of all significant effects and their interactions.
Analysis of significant effects revealed that most of the response-variables are not affected by quadratic terms of
selected factors, as showed by p-value < 0.05 (Table 6). This is the case for thermal properties (Tonset and Tg), and
for modulus and crosslinking density. On the other hand, prediction of tensile strength and elongation is well
represented by second-order empirical models. In addition, interaction effect was significant for all mechanical
properties, being not significant for crosslinking density and thermal properties. Effects' significance on response-
variables can also be analyzed by observation of contour plots, as presented in Fig. 3 for mechanical properties and
crosslinking density, and in Fig. 4 for thermal properties.

Table 6. Coefficients of second-order empirical model for response-variables. Values in parenthesis are p-
value at 0.05 significance level.
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

rup  rup E e Tonset m1 Tg

Effect
(MPa)  (MPa) (10-4 mol.mL-1) (oC) (%) (oC)
391 13696 58.8 -7.9 341.3 -19.8 23.0
Mean
(0.004) (0.000) (0.002) (0.000) (0.000) (0.000) (0.001)
-832 -29123 -128.6 1.4 -67.7 64.0 -162.0
1 (L)
(0.000) (0.000) (0.001) (0.000) (0.022) (0.001) (0.000)
-1185 23003 -211.2 -47.2 4.0 375.0 -203.0
2 (L)
(0.000) (0.000) (0.022) (0.002) (0.081) (0.000) (0.000)
443 15644 71.8 11.6 32.9 -17.8 130.0
1 (Q)
(0.023) (0.008) (0.225) (0.818) (0.535) (0.816) (0.464)
633 -12222 151.0 13.8 68.1 -697.0 -290.0
2 (Q)
(0.032) (0.083) (0.121) (0.861) (0.418) (0.002) (0.309)
1270 -21111 218.9 32.6 -6.7 -109.0 431.0
12
(0.002) (0.011) (0.036) (0.666) (0.931) (0.357) (0.137)

Analysis of contour plots presented in Fig. 3 and Fig. 4 showed that level curves were the result of a two-
dimension plot of a surface with curvature in the case of mechanical properties, contrasting with two-dimension plot
of a flat plane in the case of crosslinking density and thermal properties (Tonset and Tg). A slight deviation from the
thermal properties behavior described above was observed for m1, since level curves showed the tendency to
converge to a minimum value close to the region (+1, 1). This observation is in agreement with coefficients
presented in Table 6, where quadratic term for factor x2 is significant, thus resulting in a surface with a pronounced
curvature around this factor.
Increase on tensile strength at rupture was observed when factors x1 and x2 were raised from low (1) to high
levels (+1) (Fig. 3a). However, the increase in this property was more expressive when x1 was fixed at high levels.
Same behavior was observed for Young modulus (Fig. 3c). In the case of elongation at rupture (Fig. 3b), increase
was observed when x1 was decreased and x2 was increased, with the elongation increase being more expressive
when x1 was fixed at low levels (1). In the opposite direction, crosslinking density was increased when x1 was
increased and x2 was decreased (Fig. 3d). Unlikely mechanical properties, increase in crosslinking density was
independent of the level where x1 or x2 was fixed. This is explained by the fact that interaction between factors x 1
and x2 has no effect on this response-variable (Table 6).
Thermal stability of binder can be established based on higher initial degradation temperature, and lower mass
loss at the beginning of degradation. These characteristics are represented here as Tonset and m1, respectively.
Another important feature of binder is to render solid propellant low values of Tg. Changes on x1 and x2 factors
from high (+1) to low levels (1) resulted in a more thermally stable binder in respect to mass loss and Tg (Fig. 4b
and c, respectively). As for Tonset changes required for increasing thermal stability were from high (+1) to low levels
(1) for x1, and from low levels (1) to high levels (+1) for x2 (Fig. 4a). Compared to its HTPB-based counterpart, at
NCO/OH ratios of 0.85, 0.925, and 1.0 (Table 7), castor oil-based binder showed to be less stable and to have a
hardly lower Tg. For this reason, the use of castor oil in binder formulations need to be carefully planned in order to
balance the improvement of network properties obtained from crosslinks with the drawback on thermal properties.

6
American Institute of Aeronautics and Astronautics
 0.12 0.12
2 8 16 600.0 400.0

BDO : castor oil

BDO : castor oil

0.08 12 0.08
500.0

0.04 0.04 250.0

2 4 300.0

1 1
0.00 200.0
0.00
0.85 0.90 0.95 1.00 0.85 0.90 0.95 1.00
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

Molar ratio NCO/OH Molar ratio NCO/OH

a) b)
0.12 0.12
4.0
1.0 2.0 1 3
0

BDO : castor oil

BDO : castor oil

0.08 0.08
3.0

0.5
0.04 0.04

1.5 4
2
0.00 0.00
0.85 0.90 0.95 1.00 0.85 0.90 0.95 1.00

Molar ratio NCO/OH Molar ratio NCO/OH

c) d)

Figure 3. Contour plot for (a) tensile strength at rupture; (b) elongation at rupture; (c) Young modulus; (d)
crosslinking density.

0.12 0.12 0.12

308.0 307.2 -5 5
10
45
BDO : castor oil

BDO : castor oil

BDO : castor oil

0.08 306.8 0.08 40 0.08

307.6

0
0.04 0.04 0.04
35
30 -15

307.0 306.6
25 -10
0.00 0.00 0.00
0.85 0.90 0.95 1.00 0.85 0.90 0.95 1.00 0.85 0.90 0.95 1.00
Molar ratio NCO/OH Molar ratio NCO/OH Molar ratio NCO/OH
a) b) c)

Figure 4. Contour plots of (a) Tonset (TGA); (b) m1, and (c) Tg (E”).

7
American Institute of Aeronautics and Astronautics
 Table 7. Mechanical, physicochemical and termal properties of the reference HTPB-based binders.

σrup εrup E Q υe Tonset Δm1 Tg

R
(MPa) (%) (MPa) (adim.) (10-4 mol.mL-1) (oC) (%) (oC)
1.000 1.12 ± 0.15 283 ± 62 1.42 ± 0.02 8.35 ± 0.45 1.30 ± 0.14 309.4 16.0 -68.7
0.925 1.01 ± 0.08 578 ± 43 0.84 ± 0.01 12.30 ± 0.19 0.62 ± 0.02 311.5 16.0 -68.9
0.850 1.08 ± 0.06 1189 ± 9 0.35 ± 0.02 27.14 ± 1.16 0.15 ± 0.01 313.4 17.4 -69.9

Although analysis of each response-variable

is possible to be accomplished on a single
basis, it is a very difficult task to carry on, due
to the interactions between the investigated
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

factors. For this reason, a more valuable

analysis can be obtained by evaluating several
responses all at once. Since no unique
combination of factors results in a desired
response for all variables, overlaying of
empirical models for mechanical responses and
crosslinking density was carried out (Fig. 6).
Limits were set based on the reference HTPB-
based binders (Table 7). Limits for crosslinking
density were arbitrarily chosen to be higher
than those presented for HTPB-binder. Since Figure 6. Overlaid contour plots for mechanical properties and
thermal properties related to thermal stability crosslinking density.
showed not to be critical for castor oil-based
binder tailoring, Tonset and m1 were not included on overlaying. In the case of Tg, response on this variable was not
included in this analysis because reference values were out of range obtained for castor oil-based binder.
White areas in Fig. 6 correspond to properties within the established limits. Thus, inside these areas, high levels
of crosslinking density, from 1.5 to 4.0 x10-4 mol.mL-1, could be obtained in conjunction with the same range of
mechanical resistance (tensile strength and modulus) as found for HTPB-binders. Limits for elongation had to be
lowered, up to 600%, in order to be included in the overlaying analysis, since increase in crosslink density resulted
in decrease of elongation.

IV. Conclusion
Castor oil can be employed as a binder component to act as crosslinking agent, in order to avoid undesired
diffusion of compounds and to increase tensile resistance. However, due to low molecular mass of this oil,
elongation is expected to decrease. To overcome this drawback, chain extenders can also be incorporated into binder
formulation. Due to the fact that several properties had to be examined all at once, a factorial experimental design
was employed in this study, allowing not only to overlaying the response-variables, but also to find out which are
the significant factors for prediction of the responses. Mixtures of BDO and castor oil were compatible up to relative
mass concentration of 1:4, which comprises a large range of mass proportion of diol:triol to be employed in binder
formulation. Tensile strength and elongation at rupture were explained by second-order empirical models, whereas
all other responses were predicted by first-order models. Interaction between the factors NCO/OH molar ratio and
BDO:castor oil relative mass concentration was significant at 0.95 confidence level for the mechanical properties.
Crosslinking density and thermal properties were not affected by interaction of the investigated factors. As for the
results of the tailoring experiments conducted, it was possible to establish two possible regions inside which factors
can vary and satisfy the condition for obtaining a castor oil-based binder with higher degree of crosslinking density
than HTPB-based binder. Inside these regions tensile strength and modulus were the same as for the reference
binder, but elongation was lower, despite the presence of BDO. Regarding thermal degradation, a slight decrease in
thermal stability was observed, but the higher values of Tg were the main concern about using castor oil in binder
formulations. However, the presented results rely solely on diol/triol properties. The use of castor oil as crosslinking

8
American Institute of Aeronautics and Astronautics
 agent in HTPB-based binders may easily overcome the unfavorable aspects pointed out here and be benefited from
the better network properties tailored by this study.

Acknowledgments
Financial support from Agência Espacial Brasileira (Brazilian Space Agency) is acknowledged. L. D. Villar
thanks PROAP/CAPES. A. E. Silvestrini thanks CNPq/PIBIC for his scholarship (Grants 800412/2014-1 and
800631/2016-1).

References
1
Gustavson, C., Greenlee, T. W., and Ackley, A. W., “Bonding of composite propellant in cast-in-case rocket motors,” Journal of
Spacecraft, Vol. 3, n. 3, 1966, pp. 413-418.
2
India Space Research Organisation, ISRO, Bangalore, India, G.B. Patent Application for a “Production of Esthertype Polyols,”
Patent No. 1,524,782, filed 24 Jan. 1976.
3
Sekkar, V., and Raunija, T. S. K. “Hydroxyl-terminated polybutadiene-based polyurethane networks as solid propellant binder-
state of the art,” Journal of Propulsion and Power, Vol. 31, n. 1, 2015, pp. 16-35.
Downloaded by UNIVERSITY OF ILLINOIS on July 18, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-4694

4
Dennis, J. D., Shark, S. C., and Villarreal, J. K., “Experimental investigation and analysis of the sol-brid tactical rocket motor
concept,” AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, AIAA, Reston, 2010, pp. 1-28.
5
Dennis, J. D., and Villarreal, J. K., “Design and validation testing of the solbrid rocket motor,” AIAA Aerospace Sciences
Meeting including the New Horizons Forum and Aerospace Exposition, AIAA, Reston, 2011, pp. 1-10.
6
Gomes, S. R., Rocco Jr., L., and Rocco, J. A. F. F., “Combustion evaluation of PU solid fuel based in prepolymer technology
additivated with paraffin and aluminum for hybrid rocket engines,” AIAA/ASME/SAE/ASEE Joint Propulsion Conference &
Exhibit, AIAA, Reston, 2012, pp. 1-7.
7
Gomes, S. R., Rocco Jr., L., Rocco, J. A. F. F., and Iha, K., “Evaluation of polyurethane binder additivated with paraffin and
tested with a swirl injector,” AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, AIAA, Reston, 2012, pp. 1-8.
8
Rocco Jr., L., Rocco, J. A. F. F., Gomes, S. R., and Iha, K., “Experimental study of vegetal based polyurethene fuel filled with
paraffin particles for hybrid rocket motors,” AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, AIAA, Reston,
2013, pp. 1-9.
9
Bagchi, S., Nema, S.K., and Singh, R.P., “Ultrasonic and viscometric investigation of ISRO polyol in various solvents and its
compatibility with polypropylene glycol,” European Polymer Journal, Vol. 22, n. 10, 1986, pp. 851-857.
10
Villar, L.D.; Cicaglioni, T., Diniz, M. F., Takahashi, M. F. K., and Rezende, L. C., “Thermal aging of HTPB/IPDI-based
polyurethane as a function of NCO/OH ratio,” Materials Research, Vol. 14, n. 3, 2011, pp. 372-375.
11
Oberth, A. E., “Equilibrium swelling of polyurethanes elastomers,” Rubber Chemistry and Technology, Vol. 63, 1990, pp. 56-
65.

9
American Institute of Aeronautics and Astronautics

View publication stats