Corrosion Science 88 (2014) 372–386

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Effect of high-temperature furnace treatment on the microstructure

and corrosion behavior of NiCrBSi flame-sprayed coatings
Zoran Bergant, Uroš Trdan, Janez Grum ⇑
Faculty of Mechanical Engineering, University of Ljubljana, Askerceva 6, 1000 Ljubljana, Slovenia

a r t i c l e i n f o a b s t r a c t

Article history: The microstructure and corrosion properties of flame-sprayed and heat-treated NiCrBSi coating on
Received 27 March 2014 low-carbon steel are analyzed using an optical microscope, SEM/EDS and cyclic polarization electrochem-
Accepted 26 July 2014 ical tests. Heat treatment was performed at 930 °C, 1040 °C and 1080 °C for 10 and 20 min where
Available online 4 August 2014
densification of the coating occurred. Electrochemical corrosion tests in a 3.5 wt.% NaCl solution were
performed on a low-carbon mild steel substrate, as-sprayed coating and on furnace treated coating par-
Keywords: tially melted and fused between solidus and liquidus points at 1080 °C to evaluate the influence of heat
A. Metal coatings
treatment on corrosion resistance.
A. Nickel
B. Polarization
Ó 2014 Published by Elsevier Ltd.
B. SEM
C. Passive films
C. Pitting corrosion

1. Introduction post-processing stage. The selection of coating and substrate mate-

rials is a key factor in improving the quality of the coating micro-
Thermal-sprayed coatings offer practical and economical structure and adhesion with the substrate after the heat treatment.
solutions for the corrosion and wear protection of components. The coating must have self-fluxing properties. During the partial
Thermal-spraying processes are categorized by the different melting between the solidus and liquidus points of coating, the
heat-sources, such as combustion gas, laser, electric-arc, plasma low viscosity molten material enables fusion and porosity reduc-
source, that are used to melt and propel the feedstock material tion while the substrate material is far below the liquidus temper-
onto the surface of the substrate. The feedstock material is in the ature. The coating is very often heated using the energy of the
form of a rod, a wire or powder [1]. The combustion process with flame from a torch after spraying [5]. However, because of the dif-
oxygen-acetylene flame is also known as the low velocity oxygen ficulty to control a temperature–time cycle, the through-depth
fuel (LVOF) process, which is usually applied for the low-cost sur- microstructural properties are not reproducible in this process
face reparation of worn-out components. In the LVOF process, the [6,7]. In contrast, furnace heat treatment offers good control of
feedstock material is partially or fully melted in a stream of oxy- the process and reproducible results [8,9].
gen-acetylene flame, accelerated in a turbulent flow of gases and The corrosion-and-wear-resistant NiCrBSi coating is a promis-
deposited on the roughened substrate, thus forming a coating with ing solution to protect the low-alloyed steel substrate, which has
a lamellar, heterogeneous and porous microstructure. The solidi- a significantly lower melting temperature than the deposited alloy.
fied molten particles (or splats) are the basic structural building While heating and holding the specimen at the temperature
blocks in thermal spray coatings [1]. Splat boundaries can affect between the solidus and liquidus points of the coating alloy, the
intersplat cohesion, whereas oxides dispersed in the splat or the low viscosity eutectic phase forms and enables densification and
supersaturated oxygen can increase the hardness of the splat [2]. deoxidation, and eliminates the network of porosities [3,6,10].
The disadvantage of flame spraying is the limited cohesion/adhe- However, the low-carbon steel substrate is heated up to reach deep
sion strength of the mechanically bonded coating on the substrate into the austenite region. The period of heat treatment must be
with relatively low adhesion strength values of 11–23 MPa [3,4]. short to prevent excessive growth of austenite grains, which would
To improve the properties, heat treatment is applied as a second, lower the mechanical properties of the substrate. Heat treatment
in a furnace of NiCrBSi coatings with partial melting provides a
functionally protective and dense coating with high cohesion
⇑ Corresponding author. Tel.: +386 1 477 1203; fax: +386 1 477 1225. strength and a strong diffusion bonding with low-carbon steel
E-mail addresses: zoran.bergant@fs.uni-lj.si (Z. Bergant), uros.trdan@fs.uni-lj.si substrates.
(U. Trdan), janez.grum@fs.uni-lj.si (J. Grum).

http://dx.doi.org/10.1016/j.corsci.2014.07.057
0010-938X/Ó 2014 Published by Elsevier Ltd.
 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386 373

In the electrochemical sense, the presence of solution-path

defects, i.e. pores within coatings, enables the electrolyte to reach
the substrate surface, which accelerates the formation of a galvanic
couple between the coating and the substrate, leading to fast dis-
solution of the substrate. In addition, when the coated component
is subjected to a aggressive environment, spalling could occur due
to the extensive exfoliation effect [11]. Navas et al. [12] reported
that localized laser melting of plasma-sprayed NiCrBSi coatings
does not affect the corrosion rate, i.e. the corrosion resistance nei-
ther improves nor deteriorates. Shrestha et al. [13] analyzed the
corrosion resistance of high-velocity oxy-fuel NiCrBSi coatings,
which were subsequently vacuum fused. Lower values of current
densities during anodic polarization tests reveal that the coatings
are more corrosion resistant after vacuum fusion treatment. In this
aspect, the study of the influences of a heat treatment parameters
to obtain a denser, more stable coating has an important role in
improving the corrosion resistance. In contrast, a totally sealed
coating will reduce self-lubricating properties, leading to poorer
tribological performance in lubricated contact [14].
Recently, more expensive laser-processing methods, such as
laser surface melting [12,15,16], have attracted global interest
because of their ability to improve the characteristics of NiCrBSi
sprayed coatings to provide pore- and defect-free surfaces. How-
ever, the effect of the low-cost short-time furnace heat treatment
on the microstructure and corrosion properties of flame-sprayed
NiCrBSi coating in the presence of aggressive Cl ions lags well
behind and is, therefore, worth investigating.
Therefore, the aim of this research work was to investigate the
influence of a subsequent furnace heat treatment in a protective
argon atmosphere at different holding times on the microstructure,
the percentage of the apparent porosity area and the geometry of
the pores, as well as the influence on corrosion behavior in a
3.5% NaCl solution.

2. Experimental procedure

2.1. Processing parameters and materials Fig. 1. Substrate and powder spraying materials (a) metallographic image of low-
carbon steel substrate microstructure, (b) SEM image of NiCrBSi gas atomized
The NiCrBSi powder from Castolin Eutectic, referenced as grade particles.
Eutallow RW and produced via gas atomisation, was chosen as the
feedstock material [17]. The chemical composition of the Eutalloy
particles) and surface dimples, caused by particle–particle colli-
RW 12495 feedstock powder material and the substrate is given
sions during atomization.
in Table 1. A coating was deposited on substrate specimens made
In a cross-section, the particle microstructure contains fine
from low-carbon (0.08% C) mild steel (W. No. 1.0037, EN 10027-
cellular and dendritic microstructures (detail in Fig. 1b). Particle
2) with a predominantly ferritic microstructure with a small
size ranges between 20 and 124 lm, and they are normally distrib-
fraction of pearlite, Fig. 1a.
uted with the mean value of 72 lm (N = 98).
The nickel-based powder contains carbide-forming alloying ele-
ments Cr and B, which increase wear resistance, whereas the pres-
ence of Si favors wettability and deoxidation of the coating during 2.2. Coating deposition
the heat treatment. The partially soluble alloying elements Cr, Si
and B reduce the liquidus temperature of the coating [18]. Boron Coating was deposited onto a face surface of cylindrical speci-
is added to form a Ni–Ni3B low viscosity eutectic solution that pro- mens with a diameter of 25 mm and a length of 25 mm, using a Roto-
motes the flux in the coating. The NiCrBSi alloy has the lowest tec 80 flame-spraying torch by Castolin Eutectic. Specimens were
point of liquidus temperature at 1110 °C [17], which is far below mounted on the tube and a torch holder with a constant rotation
the liquidus temperature of a low-carbon steel substrate at approx. of 18 rev/min and feed of 2 mm/rev with stand-off distance of
1500 °C. Fig. 1b shows spherical powder particles with satellite 200 mm (Fig. 2). The powder feed was turned off after 10
particles (smaller particles that adhere on the surface of other revolutions. The spraying distance was set to 200 mm. The working

Table 1
Chemical composition (in wt.%) of NiCrBSi powder and mild steel substrate.

Chemical composition (wt.%)

Elements Ni Cr B Si Fe C Mn Mo
NiCrBSi powder Eutalloy RW 12495 78.1 13 2.08 3.22 3.09 0.51 – 0.047
Mild steel substrate (W. No.: 1.0037) 0.08 0.1 – 0.17 Bal. 0.08 0.45 –
374 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386

where y(x) is the measured profile. Rz (DIN), also known as Rtm, is

the mean value of the absolute height of five highest peaks and val-
leys in the reference length boundary.

2.5. Microstructural evaluation

For the analysis of microstructure, seven metallographic speci-

mens were prepared following the standard procedure for thermal
sprayed coatings [21]. For estimation of the porosity and coating
thickness in as-polished specimens, images were taken with a Leitz
Wetzlar optical microscope (50 magnification at aperture of 0.18)
with an Olympus Colorview digital camera. Image post-processing
with AnalysisDocu was performed for RGB separation to convert
RGB color images to grey level, for cropping, thresholding and med-
ian noise reduction, and ImageTool v3 was used for pore image
analysis. For each image, a square analysis field was selected from
the 600 dpi picture, with a width of 1000 lm. Feret diameter,
elongation and major axis values were compared between two
metallographic specimens before and after subsequent furnace
remelting. For the investigation with SEM/EDS, etching was per-
formed with a 50 mL HCl solution, 34 mL glycerol and 16 mL
HNO3. The chemical analysis was carried out with Ja EOL JXA-
8600 (SEM) electronic microscope with the addition of a microsen-
sor for energy-dispersive spectroscopy (EDS).
Fig. 2. Flame-spraying and furnace remelting set-ups and parameters.

2.6. Electrochemical measurements

pressure of the oxygen and acetylene was 4.0 and 0.7 bar,
respectively.
All electrochemical corrosion measurements were carried out
with PGZ100 Voltalab potentiostat/galvanostat controlled by
2.3. High temperature heat treatment
VoltaMaster 4 software. A classic three-electrode cell (electrolyte
volume 0.7 L) was used for the corrosion measurements. Experi-
High temperature heat treatment was performed with rela-
ments were performed in a naturally aerated, near neutral pH
tively short holding periods, i.e. 10 and 20 min. Six combinations
(7.7 ± 0.2) 3.5 wt.% NaCl (p.a. 99.8% Sigma–Aldrich), prepared with
of temperature and time values according to the general-factorial
de-ionised water. The temperature of the test medium was equal
experimental design were used to study two parameters at two
in all experiments, i.e. (22 ± 0.5) °C. Immediately prior to each
time and three temperature levels [19]. For each parameter set of
experiment, specimens were ultrasonically degreased in ethanol
the experimental run, three specimens were heat treated for fur-
and rinsed in de-ionised water for 2 min each. These steps were
ther analysis. The first temperature level was set below the solidus
necessary to achieve reproducibility and obtain reliable measure-
temperature of NiCrBSi powder alloy at 930 °C. The second temper-
ments. The specimens of 15 mm in diameter were then embedded
ature level at 1040 °C is the melting point of the eutectic stage of
in a Teflon holder, immediately transferred to the electrolytic cell
the solid austenite matrix phase. Above this temperature, the
and employed as the working electrode (WE). The area of the spec-
fusion of the coating is possible after a certain time. The highest
imens exposed to the solution was 1 cm2.
temperature level was set at 1080 °C, which is 30 °C lower than
The reference electrode (RE) was a saturated calomel electrode
the liquidus line of the NiCrBSi powder alloy.
(SCE = +244 mV vs. SHE), and the counter electrode (CE) was a plat-
Fig. 2 illustrates the experimental procedure of the coating
inum rod. All the potentials described in this work are relative to
deposition and heat treatment. Coated specimens were inserted
the SCE, unless stated otherwise.
in a hot tubular furnace chamber. After positioning three speci-
In this experiment, electrochemical techniques were performed
mens in a row, the chamber was closed tightly. Furthermore, the
in the following order:
interior was filled with pure argon gas (flow rate of 10 l/min) to
reduce the negative influence of high temperature oxidation and
(i) open circuit potential (Eocp = f[t]),
decarburization of carbon steel substrate. The temperature was
(ii) linear polarization resistance (LPR) measurements and
set below liquidus line to avoid the flow of the liquid-phase from
(iii) cyclic polarization (CP).
the coated area of the specimen. After a selected period of time,
the specimens were removed from the furnace and air-cooled.
Measurements of the open circuit potential Eocp were per-
formed for 20 min in order to stabilize the surface, whereas at
2.4. Surface roughness measurement
the end of the stabilization process the attained potential was
referred to as the corrosion potential Ecorr [22]. After the steady-
Surface roughness was measured with a Surtronic 3+ profilom-
state Ecorr, linear polarization resistance (LPR) measurements were
eter (Taylor Hobson). The selected parameters for the characteriza-
carried out, which involved disturbing the system with ±20 mV vs.
tion of the surface roughness are Ra and Rz [20]. The Ra value is the
Ecorr. According to the previous studies [23–25], a 0.2 mV s1
average arithmetic deviation from the mean line on the measuring
potential scan rate was chosen. After the measurement, Voltamas-
length L, defined by:
ter software was used to fit the data of the measured potential vs.
Z L
1 the current density (E–i curve) to a straight line. The values of the
Ra ¼ jyðxÞjdx ð1Þ
L 0 polarization resistance (Rp) were then determined as the slope of
the E–i curve (Rp = (dE/di)).
 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386 375

A cyclic polarization (CP) scan was carried out in the anodic polished state. However, for fused coatings, additional background
direction, starting at 200 mV more negative than Ecorr. Afterwards, subtraction must be performed on microscope images because of
the polarization direction was reversed at the switching potential the newly formed phases, which are visible after polishing.
Esw, which was obtained at a pre-defined threshold current density The percentage of the area of porosity was determined from
(ith = 1 mA cm2). CP tests were conducted at s scan rate of metallographic specimens in as-polished cross-sections, Fig. 4.
1 mV s1 [26–28]. From the cyclic polarization curves, essential Images were post-processed to the point at which the percentage
potential parameters were obtained. The corrosion potential (Ecorr), of the area of porosity equaled that of black pixels on the binary
corrosion current density (icorr), and cathodic as well as the anodic image. The number of the fields to be measured in order to obtain
Tafel slopes (bc and ba) were obtained by the intersection of the reliable results depends on the homogeneity, density and size of
Tafel lines extrapolations by ±100 mV around Ecorr using a com- the analyzed objects. The number of the fields to be measured in
puter least square analysis. The pitting potential (Epit) was deter- order to estimate specimen average porosity was obtained by ana-
mined with extrapolation to the potential at which the current lyzing the nine coated specimens. Five fields in each specimen
abruptly increases in the forward scan. The repassivation/protec- were determined in order to calculate the average area porosity
tion potential (Erp) was determined as the potential at which the P and standard deviation S. The relative standard error was calcu-
pffiffiffi
curve in the reverse scan intersects the forward scan. The corrosion lated as RSe ¼ S=ðP  nÞ, where n is the number of fields. The num-
tests were performed on specimens before and after coating depo- ber of fields was calculated on the specimens for which the
sition and after heat treatment at 1080 °C for 10 and 20 min. All standard error RSe of area percentage porosity is the highest (#2).
electrochemical corrosion techniques were pre-tested to ensure In Table 3, the average area percentage porosity after spraying
the repeatability of the experiments. amounts to 10.2%.
The number of the fields to be measured in order to estimate
the specimen average porosity at a 95% confidence interval, so that
3. Results and discussion
the estimation will not deviate for more than ±10% from the real
value [30] is:
3.1. Surface roughness and coating thickness
!
200 S2
Surface roughness was measured on three parallel lines in the nf ¼  ¼9 ð2Þ
y P
horizontal and vertical directions (Fig. 3) on the sample after flame
spraying and after heat treatment. The measured values of Ra and where y is deviation (y = 10%). Under 50 magnification, the num-
Rz are given in Table 2 for as-sprayed specimens and specimens ber of fields nf = 9 was chosen for further specimen analysis after
after heat treatment at 1080 °C for 10 and 20 min. Grit blasting, heat treatment.
which produces a roughened surface, is a necessary step prior to
coating deposition. Because the mechanism of coating bonding to
3.3. Porosity after high temperature heat treatment
substrate is that of mechanical interlocking to the rough asperities,
the coating adhesion is strongly influenced by Ra and Rz values.
Table 4 lists the results of area porosity percentage after furnace
After deposition, the coating’s surface consists of a large number
fusion with the average porosity P HT of the nine fields. The lowest
of small particles, remelted or/and partially remelted in the flame,
average porosity of 0.87% was measured after heat treatment at
which form a granular and relatively rough surface with an Ra
1080 °C with a holding time of 10 min.
value of 9.9 lm. After the 10- and 20-min heat treatments, a differ-
Variance analysis (ANOVA) was used to estimate the signifi-
ent reflection of light from the surface is a clear indication of a
cance of each investigated parameter, using the null-hypothesis
sealed surface. Measurements confirmed significantly lower values
test that calculates the sum of squares, interaction of parameters
of surface roughness. After 20 min of treatment at 1080 °C, the
and residuals [19]. ANOVA results for temperature and time versus
coating roughness decreases from an initial Ra of 9.9 lm to the final
area percentage porosity are given in Table 5.
Ra value of 3.3 lm, as can be seen in Fig. 3. After the heat treatment
The P-values below 0.05 indicate that both temperature and
of specimens at 930 °C and 1040 °C (under solidus temperature),
time are statistically important factors that influence porosity.
the surface roughness Ra value is on the same level as that of the
The F-value of 1.89 indicates that interactions between both
as-sprayed coating (around 9.9 lm).
parameters are weak. Fig. 5 shows the interaction chart as the
It should be noted that higher surface roughness represents a
function of temperature and time of heat treatment. As the tem-
higher sensitivity to corrosion attack [28]. This could be primarily
perature increases from 930 °C to 1040 °C, the porosity level
associated with the easier penetration of water molecules, oxygen,
decreases. The lowest area porosity was measured after 10 min
and Cl ions into isolated regions of inhomogeneities (surface cra-
at 1080 °C. The melt that forms during the fusion is a low-temper-
ters, intermetallic phases, passive film defects, pores, etc.)
ature eutectic Ni–Ni3B, and its surface tension and wetting proper-
[15,16,28].
ties can be controlled with silicon and boron, which are known as
In Fig. 3, the coating-substrate cross-sections with average layer
fluxing agents [1]. At the same time, silicon acts as deoxidation
thickness after flame spraying and after high-temperature heat
agent and boron forms hard boron carbides. In the early stage of
treatment are shown. After heat treatment at 1080 °C, the coating’s
heat treatment in the mixed zone between solidus and liquidus
thickness reduced during fusion because of the densification and
point of coating, the decrease of porosity could originate from
partial release of entrapped gases [9].
the capillary motion of low viscosity melt (eutectic), which wets
the surfaces and fills in the porosity voids and networks. The
3.2. Porosity after flame spraying results show a slight increase of porosity level when the time of
heat treatment increases to 20 min. It appears that (in addition
The microstructure after flame spraying consists of a network of to the fusion of the coatings) other phenomena occur, which lead
interconnected elongated porosities as a consequence of the to the formation of new gas pores (overheated coating). Kim
impact, rebound and the contraction of particles on the substrate. et al. [9] reported that over-fusing leads to excessive fluxing that
The latter strongly interfere with the integrity of a coated depletes the alloy of Si and B and produces large coalescent voids.
component. The porosity measurement procedure is described in Shrestha et al. [13] reported values of the porosity of fused NiCrBSi
the ASTM Standard E2109-01 [29] and is being performed in as- coatings in the range of 0.23–0.33%; in contrast, Kim et al. [9]
376 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386

Fig. 3. Surface roughness and coating thickness at different stages (a) as-sprayed, (b) 1080 °C, heat-treated 10 min and (c) 1080 °C, heat-treated 20 min.

Table 2
Surface roughness Ra and Rz values.

Cond. Grit-blasted Flame-sprayed Furnace fused at T = 1080 °C

t = 10 min t = 20 min
# No. Ra (lm) Rz (lm) Ra (lm) Rz (lm) Ra (lm) Rz (lm) Ra (lm) Rz (lm)
1 5.7 34.3 9.0 54.6 5.2 27.8 3.9 20.6
2 6.2 33.4 8.9 45.8 4.5 25.6 3.3 17.3
3 6.7 36.4 11.4 54.1 4.5 24.0 3.1 15.0
4 5.1 30.7 9.6 51.4 4.8 24.7 3.5 25.9
5 6.3 34.2 10.4 58.7 4.9 25.0 3.2 16.0
6 6.1 34.3 10.0 52.3 4.0 18.7 2.7 14.0
Avg. 6.0 33.9 9.8 52.8 4.6 24.3 3.3 18.1
St. dev. 0.54 1.85 0.94 4.32 0.4 3.03 0.40 4.44

reported values of around 2%. Previous work of other authors and 3.4. Pore shape before and after heat treatment
this research work clearly indicate that complete densification
cannot be obtained but that porosity can be greatly reduced. In this A pore is an irregular 3-dimensional object in a coating’s micro-
research work, porosity has been greatly reduced, i.e. from the structure at a random location. The area of a sectioned pore in a 2-
initial 10.2% after spraying to the final 0.87% after heat treatment D plane depends on the section’s location. To evaluate the area
at 1080 °C and 10 min of holding time. accurately, the number of test pores in a selected test field should
 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386 377

Fig. 4. Metallographic image of NiCrBSi under optical microscope (a) initial image of coating and base material in greyscale – measuring field (1000–400 lm), (b) histogram
of gray values [0–255].

Table 3
The area percentage of porosity after flame spraying applied to nine specimens.

# ni ni1 ni2 ni3 ni4 ni5 P (%) S Se RSe

1 5 9.9 10.8 10.9 10.5 10.2 10.5 0.416 0.186 0.018

2 10 6.7 10 11.6 11.2 8.5 9.6 2.021 0.904 0.094
3 15 14.5 11.2 8.9 13.2 10.5 11.7 2.214 0.990 0.085
4 20 9.2 10.6 10.5 10.2 11.5 10.4 0.828 0.370 0.036
5 25 13.3 10.4 10.5 13.1 12.5 12.0 1.410 0.631 0.053
6 30 11.3 11.7 7.8 9.8 11.5 10.4 1.645 0.736 0.071
7 35 8.9 10.3 8.7 9.5 10.6 9.6 0.837 0.374 0.039
8 40 5.9 8.4 7.8 8.9 9.5 8.1 1.380 0.617 0.076
9 45 10.8 9.4 7.7 10.1 9.8 9.6 1.159 0.518 0.054
Average 10.2 1.323

Table 4
The area percentage of porosity for nf = 9 and average PHT under different temper- semi-molten particles, the contraction and deformation of particles
ature–time conditions. leads to the formation of long, horizontally oriented pores. The
Runs Parameters Measuring fields Avg. elongation distribution of porosity is related to the coating perfor-
No # T (°C) t (min) P1 P2 P3 ... P9
mance as long horizontal pores indicate brittle lamellar micro-
P HT (%)
structures. The elongation E is the ratio between the minor and
1 930 10 2.69 2.54 1.15 ... 2 2.13
major axis lengths, E = Dmin/Dmax, where Dmin is the minor axis
2 930 20 2.79 1.81 3.53 ... 2.6 2.71
3 1040 10 0.98 1.02 0.91 ... 1 0.97 and Dmax is the major axis. If the average or median elongation dis-
4 1040 20 0.46 1.45 0.74 ... 0.95 0.90 tribution is about 1, objects are more circular; if it is about 0, the
5 1080 10 0.91 0.85 0.74 ... 0.71 0.87 pores are long and narrow.
6 1080 20 1.36 1.36 1.19 ... 1.4 1.30 Two typical binary images of coating porosity after flame spray-
ing and after heat treatment (1080 °C, 10 min) are given in Fig. 7a
and b. After remelting in a furnace, pores become smaller and close
Table 5 to spherical in shape, whereas the area percentage porosity is
Variance analysis used to estimate the influence of temperature and time on porosity reduced from 10% to values below 1%.
(nf = 9).
This analysis is focused on the comparison of the pore area, Fer-
Source of Sum of Degrees of Mean F-value P-value et diameter, major axis orientation and elongation distribution.
variation squares freedom square ImageTool V3 was used to measure and analyze the objects on
Model 14.88 5 2.98 16.97 <0.0001 the binary image, and EasyFit Professional statistics software
Temperature T 13.44 2 6.72 38.32 <0.0001 (v5.5) was employed to fit different probability density functions,
Time t 0.78 1 0.78 4.45 0.0476
which also can be used for the prediction of porosity parameters.
Interaction T  t 0.66 2 0.33 1.89 0.1771
Residuals 3.51 20 0.18 It should be noted that the number of recognized objects on a spe-
Total 18.48 29 cific image is large and diverse (Fig. 7); therefore, the statistical
evaluation represents an important and effective tool for detailed
object characterization.
be large. The Feret diameter is the diameter of a circle with the The adequacy of the fit was estimated using the Kolmogorov–
same area as the pore and is calculated as Df = (4A/p)1/2, where Smirnov test, whose statistics quantifies the distance between
Df is the Feret diameter and A is the pore area. The major axis-angle the empirical distribution function of the dataset and the cumula-
analysis reveals the presence of lamellar porosity. The major axis of tive distribution function of the reference distribution, or between
an object is the longest distance between two edge points, a line the empirical distribution functions of two datasets [31].
that runs through the centre. The minor axis also runs through Fig. 8 shows normalized probability density functions f(x) for
the centre, normal to the major axis (Fig. 6). the following parameters of the pores: the area, the Feret diameter,
Angles between substrate surface and major axis are repre- the major axis angle and the elongation. Different probability den-
sented as absolute values between 0° and 90°. The angle between sity functions were fitted to a porosity dataset, Table 6. In Fig. 8a
the surface and the major axis of a pore near 0° indicates a lamel- and b, large differences in pore area and Feret diameter distribu-
lar, horizontally oriented pore. Upon the impact of molten and tions can be observed for as-sprayed and heat-treated coatings.
378 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386

observed in the particular splats. In thermal spraying processes,

the cooling rate of particles on impact is higher than 106 K/s, which
leads to undercooled and metastable microstructures [32]. The
large differences in the microstructure morphology within individ-
ual splats indicate a wide range of dwell time (time of flight) and
particle sizes. Fig. 9c reveals that the size of cells and dendrites
in deeper-etched regions of a particular fine-grained splat is below
100 nm, while in the neighboring splat, cells and dendrite sizes are
in the range of 1–2 lm (1000–2000 nm). Gonzales et al. [6]
reported that a cNi/Ni3B eutectic phase is formed in interdendritic
regions.
After the heat treatment and air-cooling of specimens at 930 °C,
Fig. 5. Average area percentage porosity as a function of T at different holding times
no significant differences of cellular grain size were observed in
t.
individual splats, Fig. 9d–g. It appears that more stable microstruc-
tures were formed from nano-grained microstructures. At 930 °C,
no melting occurred and, consequently, splat boundaries are still
clearly visible. Furthermore, furnace treatment at this temperature
was not effective as porosity remained high. It is obvious that grain
size increased with the length of heat treatment.
After the heat treatment at 1040 °C, depending on the heat-
treatment time, the boundaries between splats gradually disap-
peared, which indicates the coatings were already slightly fused,
Fig. 9e–h. Moreover, the growth of grains is visible. In Fig. 9e and
Fig. 6. Major axis (Dmax) and minor axis (Dmin) and major axis angle definition. h, two regions can be distinguished; the darker region is more dee-
ply etched than the lighter one. This indicates that the composi-
tional difference between dendrite and interdendritic regions is a
consequence of the constitutional segregation during solidification
[33].
At1080 °C, fusion was completed as splat boundaries disap-
peared. Over time, the area of the A-region (more deeply etched
area, designated in the figure) increased and the area of B-region
decreased.
The EDS analysis was focused on four elements in the alloy, i.e.
nickel (Ni28), chromium (Cr24), silicon (Si14), and iron (Fe26), as the
weight percentages of C and B elements cannot be revealed with
this method. The average chemical composition of feedstock mate-
rial (powder), is shown in Table 1. Locations for EDS analysis on
coating after spraying are shown in Fig. 9a. EDS analysis was per-
formed on fine-grained splat (Area 1) and on dendritic splat (Area
2). From the EDS weight percentages (Table 7), there are no signif-
icant chemical composition differences within the morphologically
different splats. The microstructure is heterogeneous due to a wide
range of solidification conditions as a consequence of the chaotic
nature of the spray-deposition process. Localized elemental com-
position measurement inside the dendrite (A) and in interdendritic
(B) zones confirmed, in general, a slightly higher amount of chro-
mium inside of dendrites (A). Such results could possibly indicate
the inverse microsegregation during solidification [33].
Fig. 7. Typical binary images of porosity after flame spraying and after heat The further EDS analysis was focused on constituents of micro-
treatment; (a) a network of interconnected pores after flame spraying, (b) globular
structure after heat treatment at 1080 °C, where the developed
separated pores after heat treatment with partial remelting at 1080 °C with a
10 min time interval. phases are clearly visible on SEM micrograph, Fig. 9f. After heat
treatment at 1080 °C, the A-zone (Fig. 9f) EDS analysis revealed
The peak of the distribution is left-skewed, especially for heat- 81–82 wt.% of nickel, 9.6–10.9 wt.% of chromium and a relatively
treated specimens in which a large number of smaller pores are high content of silicon, 3.8–4.1 wt.%. However, silicon promotes
present. The left-skewed distribution in Fig. 8c also indicates that the self-fluxing properties of the coating and significantly lowers
the number of horizontally oriented pores decreased after the heat the melting point [18]. In contrast, the B-zone contains a higher
treatment. The elongation distribution in Fig. 8d indicates symmet- value of nickel with 88–90 wt.%, slightly lower chromium (7.1–
rical distribution after spraying and right-skewed distribution after 8.1 wt.%) and lower silicon (0.3–0.4 wt.%). According to the SEM/
the heat treatment, which confirms that the pores became more XRD analysis of Kim et al. [9], the matrix c-Ni phase and Ni–Ni3B
spherical (globular) during furnace treatment. eutectic phase are similar in shape and color, contributing to diffi-
cult phase characterization via the microscope.
3.5. Microstructural and chemical analyses of NiCrBSi coating The formation of new dispersed particles is visible inside the A
and B regions, Fig. 9f and i. In the A-zone, the chromium rich pre-
The SEM images of the microstructure of as-sprayed and subse- cipitates, designated as C, formed. Within the B-zone, the highest
quently heat-treated coating are shown in Fig. 9. In the microstruc- concentration of chromium was found in dark blocky precipitates,
ture in Fig. 9a–c, various sizes of dendrite and cellular grains can be designated as D.
 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386 379

Fig. 8. Fitted probability density functions, (a) area, (b) Feret, (c) major axis angle, (d) elongation; before and after heat treatment at 1080 °C for 10 min.

Table 6
Names of fitted probability density functions.

Treatment/parameter Area (lm2) Feret (lm) Major axis angle (°) Elongation Dmin/Dmax
After spraying Frechet Inv. Gaussian Johnson SB Error
HT 1080 °C–10 min LogLogistic Gumbel min.

In C-precipitates, 32 and 22 wt.% of chromium was detected in surface layer, which is subjected to severe corrosion or wear. Nev-
specimens heat-treated at 10 and 20 min, respectively. In ertheless, taking into account that the NiCrBSi coating is of a catho-
D-precipitates, 84 and 97 wt.% of chromium was detected in spec- dic nature, compared to the base-steel substrate, the porosity
imens heat-treated at 10 and 20 min, respectively. Chromium is within the coating accelerates local anodic dissolution. Thus,
easily soluble in nickel matrixes and it is likely to precipitate at porosity is considered the most important factor and should be
high temperatures. The amount of chromium is increasing with minimized in order to enhance corrosion resistance. This will be
heat treatment time in D-precipitates, Table 7. The chromium in investigated using corrosion techniques in the following sections.
D-precipitates thus reaches 97.5 wt.%. In contrast, the amount of
chromium is decreasing with heat treatment time in C-precipi- 3.6. Open-circuit potential measurements
tates. The morphology of D-precipitates is similar in appearance
to the dark blocky phases CrB and Cr7C3 identified by Kim et al. The corrosion analysis was conducted on steel-substrate speci-
[9]. The hard chromium carbide Cr7C3 dissolves when the C content mens without coating, after flame-spraying (FS) in as-sprayed con-
exceeds 0.8 wt.%. Furthermore, if the content of boron exceeds dition and after a high temperature heat treatment at 1080 °C for
2 wt.%, the chromium boride CrB phase might form. 10 and 20 min. These HT conditions were chosen because the most
Fig. 10 shows that the microstructural constituents are not effective densification occurred at 1080 °C, reflected in enhanced
evenly dispersed. The surface layer consists mainly of A-zone surface integrity with the lowest surface roughness and lowest
which (based on the literature review [6]) is most probably the porosity values.
eutectic c-Ni/Ni3B phase with a low melting point. In the process The open circuit potential curves (Eocp = f[t]) in a 3.5 wt.% NaCl
of fusion, the molten eutectic phase reached the coating as the water solution of the substrate, of NiCrBSi coating after flame
molten phase has a lower density than the solid phase. In this spraying (FS) and flame-sprayed coatings with subsequent heat
upper surface region, no B-zone can be found. Near the interface, treatment at 1080 °C (NiCrBSi + HT) at different holding times are
a higher concentration of D-precipitates is observed. This also indi- plotted in Fig. 11. From the transients obtained, it is obvious that
cates a gravity segregation during fusion heat treatment, e.g. D- the substrate exhibits the most negative values. In fact, relatively
particles being gravitationally transferred from the inside of the stable and higher anodic potential values are obtained with the
coating closer to the interface due to the density differences [33]. NiCrBSi coatings that were exposed to additional HT, especially
The distribution of phases in the coating is important in order to with the longest times of exposure, i.e. 20 min. In contrast,
correlate it with the corrosion properties and even to predict it. The specimens after flame spraying (FS) exhibited less stable Eocp tran-
most important information is the phase distribution in the thin sients over time. Furthermore, the potential of the specimen after
380 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386

Fig. 9. Microstructure of NiCrBSi coatings and chemical analysis of phases (a-b-c) after flame, (d-e-f) after 10 min of HT, (g-h-i) after 20 min of HT.

flame-spraying (FS) quickly starts to shift downward after resistance. Thus, according to the values obtained at the end of
immersion times of approx. 200 s, indicating the dissolution of OCP measurements, the corrosion resistance of the NiCrBSi coating
the specimen surface. This is logical as this specimen suffers from is expected to increase in the following order: NiCrBSi < NiC-
higher concentrations of cracks and porosity. The presence of solu- rBSi + HT_10 min < NiCrBSi + HT_20 min, where the latter achieved
tion-path defects, i.e. interconnected pores within coatings, accel- a potential ennoblement of 516 mV (Eocp = 213 mVSCE) compared
erates the formation of a galvanic couple leading to a fast to the substrate.
dissolution in anodic places. From the open circuit potential results, a good correlation with
Despite the fact that the FS specimen exhibited a descending the heat treatment conditions, i.e. time of remelting, is also
overall trend, the potential at the end of the stabilization process observed. The lower corrosion resistance of the FS specimen is
is much higher compared to the substrate (443 mVSCE vs. attributed to the higher content of pores and oxides on the very
729 mVSCE). Therefore, both specimens achieved steady-state top of the surface which has a detrimental effect leading to corro-
potential after about 20 min, indicating an equilibrium state at sion. It has also been suggested [34] that even new deposition
which the rate of oxidation (IOx) equals that of reduction (IRed). methods, such as high-velocity oxy-fuel spraying (HVOF) and
Furthermore, the potential transients of the specimen after high-velocity air fuel (HVAF) spraying, show porosity and forma-
flame spraying (FS) and FS + HT for 10 and 20 min showed more tion of oxide inter-layers of sprayed Ni-based coatings, which dras-
stable and positive potential values indicating higher corrosion tically impair the properties of the coatings. Therefore, additional
 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386 381

Table 7
Energy dispersive spectroscopy (EDS) inside the specific phase before and after heat
treatment at 1080 °C.

Zone Ni (wt.%) Cr (wt.%) Si (wt.%) Fe (wt.%)

Fig. 9a: coating after flame-spraying
A 77.46 17.04 3.50 2.01
B 78.21 15.98 2.95 2.87
Area 1 76.45 15.47 3.92 4.15
Area 2 77.87 14.98 3.68 3.50
Fig. 9f: after heat treatment at 1080 °C. 10 min
A 82.32 9.67 4.10 3.91
B 88.40 8.11 0.39 3.18
C 63.19 32.10 2.96 3.33
D 11.8 84.33 0.47 2.01
Fig. 9i: after heat treatment at 1080 °C. 20 min
A 81.15 10.86 3.89 4.09
B 90.06 7.17 0.30 2.48 Fig. 11. Effect of NiCrBSi and additional heat treatment on rest potential (Eocp = f [t])
C 70.48 21.71 6.88 4.22 of transients in a 3.5 wt.% NaCl water solution at 22 ± 0.5 °C, pH 7.7 ± 0.2.
D 2.37 97.48 0.15 0

more precisely, an increase of more than two orders of magnitude

is observed compared to the substrate specimen. The Rp values of
melting is necessary to reduce the porosity and improve their the specimens are denoted in Fig. 12 and presented in Table 8.
tribological properties in dry contact [15] and corrosion properties LPR results confirmed the lowest Rp value in a 3.5 wt.% NaCl solu-
[12]. Thus, additional HT in the furnace made it possible to obtain a tion for the substrate specimen (1.07 kX cm2), followed by the
more homogeneous, chemically stable and sealed NiCrBSi coating, flame-sprayed NiCrBSi specimen (12.11 kX cm2) and the FS speci-
which protected the material against corrosion. This will be either men with additional HT for 10 min (14.04 kX cm2). The highest
confirmed or rejected by the results of the linear polarization resis- resistance was obtained for the FS specimen after the longest expo-
tance (LPR) and cyclic polarization (CP) tests. sure to the furnace, i.e. 20 min. Here, the Rp was almost 25 times
higher than the base-steel specimen (25.96 kX cm2), showing the
3.7. Linear polarization resistance (LPR) measurements highest resistance to uniform corrosion in the tested conditions.
Furthermore, from the LPR results, a good correlation with the
The linear polarization resistance (LPR) curves of the substrate, OCP measurements was obtained, indicating the important influ-
flame-sprayed coatings and flame-sprayed NiCrBSi coatings with ence of proper heat treatment conditions on the surface oxide con-
additional heat treatment as a result of applied ±20 mV vs. Ecorr tent after NiCrBSi flame spraying. The rise in Rp indicates that the
are shown in Fig. 12. In this technique, the slope of the curve is resistance to the transfer of electrons to electro-active species in
inversely proportional to the polarization resistance (Rp) that solution increases [22]. In fact, the calculated protection efficiency
occurs at a given applied potential [24]. values in Table 8 (Eq. (3)) confirmed the extremely beneficial
From the comparison of E–i curves in Fig. 12, a strong influence effects of flame-sprayed NiCrBSi coating in the as-sprayed
of post-processing, i.e. heat treatment of FS coating, is observed; condition as well as heat-treated ones. However, the protection

Fig. 10. Nickel-matrix (c-Ni) phase and Cr-precipitate particle EDS spectrograms (a) subsurface, (b) near interface after heat treatment at 1080 °C, 20 min.
382 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386

Fig. 12. LPR curves of ±20 mV vs. Ecorr in a 3.5 wt.% NaCl solution.

efficiency among flame-sprayed specimens was the lowest for the CP curves of NiCrBSi flame-sprayed coatings in Fig. 13 show
as-sprayed condition (91.16%), indicating the highest susceptibility typical scans of passive behavior, where it is shown that in the for-
to uniform corrosion among NiCrBSi specimens due to the highest ward (anodic) scan which started at 200 mV vs. Ecorr a partial pas-
porosity. In contrast, the highest value of almost 96% after heat- sivation occurred. Afterwards, a sharp increase in anodic current
treatment for 20 min at 1080 °C indicates the high efficiency of a due to a stable pit growth or pit propagation is observed with a
smoother, more stable and homogenous coating against corrosion well-defined hysteresis in the reverse, i.e. cathodic scan.
attack. This result again confirms the important influence of subse- The results of the characteristic potentials obtained from CP
quent heat treatment of the Ni-based flame-sprayed coating, con- scans are denoted in Fig. 13b and presented in Table 8, whereas
tributing to the retardation of specimen surface dissolution. their detailed description and definition are described in Ref.
The protective efficiency (PEF [%]) of NiCrBSi coating was deter- [28]. It should be noted that nobler values of potential parameters,
mined from LPR measurements by: i.e. Ecorr, Esw and Erp, are generally obtained with NiCrBSi coatings
after being heat-treated.
Rcoating
p  Rsubstrate
p The analysis showed that among all the specimens, the flame-
PEF ¼  100 ð3Þ
Rcoating sprayed NiCrBSi specimen with an additional heat treatment of
p
20 min exhibits the noblest Ecorr potential (223 mVSCE), followed
where Rsubstrate
p represents the polarization resistance of the by the FS specimen HT for 10 min (234 mVSCE) and the specimen
substrate material, i.e. without coating, and Rcoating
p represents the after spraying (469 mVSCE), while the substrate specimen exhib-
polarization resistance of flame-sprayed and flame-sprayed and ited the most negative Ecorr value (784 mVSCE). However, it should
heat-treated NiCrBSi coating specimens. be noted that Ecorr of CP differs from OCP. This effect is probably an
outcome of the corrosion sequence itself (OCP ? LPR ? CP), a
cathodic reaction during CP or the parameters of the corrosion
3.8. Cyclic polarization (CP) scans
measurements (scan rate, duration, etc.). Additionally, differences
of Ecorr of CP and OCP among the specimens were found to be in
In this section, electrochemical characteristics analyzed from
the following order: NiCrBSi + HT_20 min (10 mV) < NiC-
cyclic polarizations (CP) curves are discussed. The CP technique
rBSi + HT_10 min (12 mV) < NiCrBSi after spraying (26 mV) < sub-
is very useful for gaining a complex picture of the active/passive
strate specimen (55 mV). These results are in good accordance
behavior of the material in order to study localized forms of
with both OCP and LPR measurements, indicating the significant
corrosion [22,28].

Table 8
Characteristic results derived from electrochemical measurements of specimens in neutrally aerated 3.5 wt.% NaCl.

Specimen Rpa (kX cm2] Ecorra (mVSCE) PEF (%) Ecorrb (mVSCE) Epitb (mVSCE) Eswb (mVSCE) Erpb (mVSCE)
Mild steel substrate 1.07 730 – 784 – 628 713
NiCrBSi-after spraying 12.11 448 91.16 469 26 48 290
NiCrBSi + HT: 10 min 14.04 223 92.38 234 32 31 175
NiCrBSi + HT:20 min 25.96 213 95.88 223 40 60 228
a
Determined from LPR measurements.
b
Determined from the CP curves.
 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386 383

of the substrate specimen shows negative hysteresis, which did

not intersect the forward anodic scan. These results are rather
logical and expected [35] since the base steel specimen (substrate)
is prone to uniform corrosion attack, having difficulty in
repassivating.
The results of the CP test of the other three (NiCrBSi) specimens
indicate an active/passive/transpassive transition in the forward
anodic scan in an aggressive chloride solution. When the scan
was reversed at a switching potential Esw (ith = 1 mA cm2), hyster-
esis was observed in the reverse scan, intersecting the forward
scan. A wider hysteresis loop and consequently a more negative
Erp are obtained for the as-is, i.e. NiCrBSi, specimen after FS,
indicating that the most severe corrosion propagation after Epit
has been exceeded. Finšgar et al. [36] proposed that pitting propa-
gation does not stop but continues at a decreasing rate until Erp is
reached, at which point pits or crevices repassivate, i.e. stop
propagating.
Although, the results from OCP and LPR tests confirmed the
most noble Ecorr and the highest Rp for the NiCrBSi specimen being
HT for 20 min, the smallest hysteresis loop was observed for the
NiCrBSi specimen that was furnace HT for 10 min. In fact, it seems
that with the heat-treated NiCrBSi specimens two aspects coexist:
Eq. (1) prior to initiation of pitting corrosion (E < Epit) the dominant
effect on the enhancement of the corrosion resistance above all
represents the surface roughness of the coating. Hence, the corro-
sion resistance was higher with the specimen being HT for 20 min
(Ra = 3.3 lm) than of 10 min HT specimen (Ra = 4.6 lm); (2) after
the initiation of pitting corrosion at the potential E > Epit, lower
porosity percentage and higher thickness takes over the dominant
effect on the repassivation ability. Thus, the fastest repassivation in
the reverse polarization scan was detected with the specimen
Fig. 13. Cyclic polarization (CP) curves measured in 3.5 wt.% NaCl water solution. being HT for 10 min, which showed higher coating thickness and
(Note: Values of marked electrochemical characteristic parameters are presented in
lower porosity values compared to the specimen being HT for
Table 8).
20 min (0.87% vs. 1.30%).
In order to evaluate the heat treatment conditioning on the
influence of high-temperature heat treatment conditions against corrosion performance more precisely, DE trends [22,28,37] as a
the advance of corrosion. Nevertheless, in the future, more system- function of specimen conditions were determined (Fig. 14). Thus,
atic investigations over a broad range of input parameters are the pitting initiation rate, i.e. region of passivity (Epit  Ecorr), ability
needed in order to more precisely characterize this phenomenon. for repassivation (Erp  Ecorr) and sensitivity to crevice corrosion
The general shift of Ecorr potential of CP in the anodic direction (Epit  Erp), were evaluated and the relative values of the measured
indicates that self-fluxing flame-sprayed NiCrBSi coatings with parameters were compared among themselves.
nickel-matrix (c-Ni phase) belong to the group of cathodic coat- According to the calculated values of DE for the heat-treated
ings, acting similarly as the anodic inhibitor. These coating catego- NiCrBsi specimens (Fig. 14), the pitting initiation rate, i.e. Epit
ries are known to provide excellent corrosion protection even in  Ecorr, is slower and the ability for repassivation, i.e. Erp  Ecorr,
extremely corrosive environments. However, the limitation of such is faster for the specimen that was furnace HT for 10 min. In addi-
coatings is that they must provide a complete barrier between the tion, a wider hysteresis and a higher potential difference of Epit -
substrate and the environment, because if the substrate (anode) is  Erp was obtained for the specimen that was heat-treated for
exposed to the aggressive environment, corrosion will be dramat- 20 min, indicating the higher sensitivity to crevice corrosion [36]
ically accelerated, resulting in the spalling of the coating. Zhao and stress corrosion cracking (SCC) [34]. Therefore, the crevice cor-
et al. [25] confirmed that even when using high-velocity oxygen rosion rate and SCC are expected to increase in the order: NiC-
fuel (HVOF) thermal spraying, corrosion develops along the paths rBSi + HT_10 min < NiCrBSi + HT_20 min. However, the general
formed by pores, inclusions, microcracks and lamella microstruc- anodic dissolution rate was found to be in the order NiC-
tures, resulting in the separation of the coating from the base rBSi + HT_20 min < NiCrBSi + HT_10 min, indicating a lower rate
material. The corrosion mode of coating was found to be exfolia- of pitting propagation beyond Epit, with a higher potential Esw for
tion and laminar peeling and the combination of two, which the NiCrBSi + HT_20 min specimen. Nonetheless, it should be noted
caused the failure of the whole coating due to extensive electrolyte that the differences are quite small between those two specimens
penetration of the coating-substrate interface. and that the NiCrBSi specimen after flame spraying exhibited the
The results of the CP test (Fig. 13) confirmed a pronounced lowest free corrosion potential Ecorr with the highest icorr, indicat-
passive plateau of the NiCrBSi specimens with evidence of a signif- ing that under steady-state conditions the corrosion attack of this
icant increase in pitting potential with heat-treated specimens. In NiCrBSi specimen would be the most severe.
contrast, the CP curve of the substrate showed a sharp current Similar corrosion behavior was reported by Navas et al. [12]
increase during anodic polarization and the highest corrosion cur- where laser surface melting led to the dissolving of plasma-
rent, indicating an active surface, i.e. non active–passive transition sprayed NiCrBSi powder particles and homogeneous microstruc-
[35]. This confirms the highest corrosion susceptibility and the tures of the whole track, with phases distributed uniformly in
highest degree of anodic dissolution of the active surface of the the Ni matrix. However, their results showed remarkably lower
substrate specimen in a test solution. Moreover, the reverse scan values of Ecorr and Erp, intersecting the forward scan in the cathodic
384 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386

determined by taking into account all of the elements, their con-

centrations and valences that are being oxidized and contributing
to corrosion:

M 1
¼ P n f ð7Þ
n i i
Mi

where ni is the valence of alloying element ‘‘i’’, fi is the mass fraction

of alloying element ‘‘i’’ (for chromium, f = 0.781, see Table 1) and Mi
is the molar mass of element ‘‘i’’ (gram/mole), whereas elements
that have a mass fraction of less than 0.01 (less than 1% by weight)
were ignored in the calculation.
Fig. 14. Comparison of DE trends as a function of heat treatment conditions
Taking into account the M/n = 21.853 g and the density of the
obtained from CP curves. P
NiCrBSi coating being 8.29 g cm3 (q = fi  qi), the corrosion rate
(in lm/year) for the NiCrBSi coating is:

region below Ecorr. The differences between the results are proba- CR ¼ 8:62  icorr ð8Þ
bly associated with different post-processing techniques, i.e. laser
Table 9 lists the results of active dissolution parameters,
and furnace melting. Therefore, it seems that low-cost furnace
including the values of corrosion potential (Ecorr), corrosion current
treatment can produce higher homogeneity of NiCrBSi coating
density (icorr) and corrosion rate (CR). The results showed signifi-
with an increased ability for repassivation and crevice corrosion
cant influence of subsequent furnace heat treatment. As discussed
resistance.
previously, the most anodic behavior of Ecorr is achieved with the
For all tested specimens, the active dissolution parameters, i.e.
specimen HT for 20 min, indicating the important role of surface
corrosion current density (icorr) and the corrosion rate (CR), were
homogeneity and enhanced roughness. The corrosion current den-
determined and listed in Table 9. The current density icorr (in
sities icorr of the substrate and NiCrBSi + HT_20 min specimens
lA cm2) is calculated according to the Stern–Geary equation
were calculated to be 15.73 lA cm2 and 1.27 lA cm2, respec-
[38], as follows:
tively. This confirms a 12-fold decrease of corrosion current after
ba  bc the NiCrBSi specimen was heat-treated at 1080 °C for 20 min. Sim-
icorr ¼ ð4Þ ilar reductions of icorr were obtained for the other two NiCrBSi
2:303  Rp  ðba þ bc Þ
specimens, which achieved corrosion currents of 2.81 and
where Rp [kX cm2] is the specific polarization resistance 2.1 lA cm2, respectively. Moreover, the calculated values of cor-
(determined from the LPR test) and ba and bc are the anodic and rosion rate (CR) were found to decrease gradually with longer
cathodic Tafel constants (in mV dec1) determined with the Tafel exposure to subsequent heat treatment. The lowest corrosion rate
extrapolation method. was obtained with the HT specimen for 20 min (10.97 lm year1)
In order to obtain quantitative information, the corrosion rate and it is almost seventeen times lower compared to the base-steel
CR in mm year1 was calculated according to Faraday’s law [39], specimen (CR = 182.76 lm year1). The specimen after FS and the
in accordance by the ASTM G-59-97 standard [40]: specimen that was HT for 10 min achieved a corrosion rate of
32.65 lm year1 and 18.12 lm year1, respectively. However, it
M 1 should be noted that corrosion rate is generally valid for uniform
CR ¼ 3:27  103    icorr ð5Þ
n q corrosion attack, which is not expected to hold completely true
at a porous cathodic layer. Nevertheless, based on the determined
where the factor 3.27  103 includes the Faraday constant and the corrosion currents, CR values can serve as quantitative comparison
metric and time conversion factors, M is the molar mass, n is the tool among the investigated specimens under steady-state
valence, q represents the density of the corroding metal and icorr conditions.
is the corrosion current density in lA cm2. For steel, The results presented in this study are consistent with other
M = 55.845 g mol1, n = 2 mol1 and q = 7.86 g cm3. Thus, Eq. (5) studies [12,14,25] in which it was found that despite the excellent
for the corrosion rate (in lm/year) for the substrate can be rewrit- corrosion resistance of the NiCrBSi coating, the reduction of defects
ten as [41]: (such as pores, inclusions and laminar microstructures) is essen-
tial. Coating pores, in particular, are superior paths for corrosion
CR ¼ 11:62  icorr ð6Þ
species to penetrate into the coating and may result in a rapid cor-
Based on the high porosity percentage and severe anodic disso- rosion attack on the substrate. All of the above makes it clear that a
lution of the substrate with the NiCrBSi specimen after spraying, it subsequent heat treatment of flame-sprayed NiCrBSi coating
was decided to also use Eq. (6). However, for the NiCrBSi coatings produces a nobler material with lower coating porosity and acts
being heat treated (avg. porosity 1%), the M/n ratio was as an effective corrosion resistant coating thus making the

Table 9
Characteristic results derived from electrochemical measurements for the calculation of corrosion rate of specimens in neutrally aerated 3.5 wt.% NaCl.

Specimen baa (mV/dec) bca (mV/dec) icorrb (lA/cm2) CRc (lm/y)

Mild steel substrate 56 126 15.73 182.76
NiCrBSi-after spraying 157 156 2.81 32.65
NiCrBSi + HT: 10 min 128 145 2.10 18.12
NiCrBSi + HT:20 min 119 211 1.27 10.97
a
Based on the Tafel extrapolation method.
b
Calculated according to the Stern–Geary equation.
c
Calculated according to the Faraday’s law.
 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386 385

penetration of water molecules and Cl ions in the near surface [6] R. Gonzales, M.A. Garcia, I. Penuelas, M. Cadenas, Ma del Rocio Fernandez, A.
Hernandez Battez, D. Felgueroso, Microstructural study of NiCrBSi coatings
more difficult.
obtained by different processes, Wear 263 (2007) 619–624.
[7] R. Gonzales, M. Cadenas, R. Fernandez, J.L. Coriyo, E. Rodriguez, Wear
behaviour of flame sprayed NiCrBSi coating remelted by flame or by laser,
4. Conclusions Wear 262 (2007) 301–307.
[8] Z. Bergant, J. Grum, Quality improvement of flame sprayed, heat treated, and
remelted NiCrBSi coatings, J. Therm. Spray Technol. 18 (2009) 380–391.
The process of high-temperature heat treatment with partial [9] H.J. Kim, S.J. Hwang, C.H. Lee, P. Juvanon, Assessment of wear performance of
remelting of NiCrBSi alloy in the region of mixed solid–liquid zone flame sprayed and fused Ni-based coatings, Surf. Coat. Technol. 172 (2003)
is an efficient method for a coating’s densification and the elimina- 262–269.
[10] N. Serres, F. Hlawka, S. Costil, C. Langlade, F. Machi, Microstructures and
tion of the porosity network, which in turn densifies the coating environmental assessment of metallic NiCrBSi coatings manufactured via
and improves the corrosion properties. Furthermore, the rough hybrid plasma spray process, Surf. Coat. Technol. 205 (2010) 1039–1046.
surface after flame spraying is greatly reduced by furnace [11] W.-M. Zhao, Y. Wang, L.-X. Dong, K.-Y. Wu, J. Xue, Corrosion mechanism of
NiCrBSi coatings deposited by HVOF, Surf. Coat. Technol. 190 (2005) 293–298.
remelting.
[12] C. Navas, R. Vijande, J.M. Cuetos, M.R. Fernández, J. de Damborenea, Corrosion
A microstructural study revealed that heat treatment below the behaviour of NiCrBSi plasma-sprayed coatings partially melted with laser,
solidus temperature was not efficient. The general morphology of Surf. Coat. Technol. 201 (2006) 776–785.
the coating is similar to as-sprayed coatings. At the solidus point [13] S. Shrestha, A. Neville, T. Hodgkiess, The effect of post-treatment of a high-
velocity oxy-fuel Ni–Cr–Mo–Si–B coating Part I: microstructure/corrosion
at 1040 °C, the porosity further decreases but the complete fusion behaviour relationships, J. Therm. Spray Technol. 10 (3) (2001) 470–479.
does not occur; moreover, the grain boundaries are still visible. At [14] S. Liu, X. Zheng, G. Geng, Dry sliding wear behavior and corrosion resistance of
1080 °C, almost complete fusion of the coating occurs, and the NiCrBSi coating deposited by activated combustion-high velocity air fuel spray
process, Mater. Des. 31 (2010) 913–917.
splat boundaries disappear. Furthermore, the network of intercon- [15] C. Navas, R. Colaço, J. de Damborenea, R. Vilar, Abrasive wear behaviour of
nected elongated porosities disappears and is replaced by a lower laser clad and flame sprayed-melted NiCrBSi coatings, Surf. Coat. Technol. 200
amount of smaller spherical pores. The through-depth distribution (2006) 6854–6862.
[16] N. Serres, F. Hlawka, S. Costil, C. Langlade, F. Machi, Corrosion properties of in
of phases after the furnace heat treatment at 1080 °C reveals the situ laser remelted NiCrBSi coatings comparison with hard chromium coatings,
differences. On the surface, the number of dark blocky phase par- J. Mater. Process. Technol. 211 (2011) 133–140.
ticles is much smaller than near the interface. [17] Produkt und Anwendungs Informationen, Castolin Eutectic, Wien Austria,
(2005), 158.
OCP transients confirmed more stable potential values over [18] ASM International, ASM Handbook, Alloy Phase Diagrams, Vol. 3, Ch.2, (2005)
longer periods after additional furnace heat treatment of NiCrBSi 83, 155, 318.
flame-sprayed specimens. It was shown that the most negative [19] D.C. Montgomery, Design and Analysis of Experiments, 5th Edition., John
Wiley & Sons, 2000.
(cathodic) value was obtained with a substrate specimen, indicat-
[20] T. Hobson, Surtronic 3+ Operator’s Handbook (1997).
ing the lowest corrosion resistance. In fact, the results have con- [21] Standard Guide for Metallographic Preparation of Thermal Sprayed Coatings,
firmed the stable potential values of heat-treated specimens, ASTM Standard E1920–03, ASM International, West Conshohocken, PA, 03.01.
whereas NiCrBSi specimen after flame spraying exhibited a sharper (2003).
[22] M. Finšgar, I. Milošev, Corrosion behaviour of stainless steels in aqueous
decrease of OCP with longer exposure times due to higher pore solutions of methanesulfonic acid, Corros. Sci. 52 (2010) 2430–2438.
content, which acts as solution-path defect, leading to faster sub- [23] M.A. Amin, M.M. Ibrahim, Corrosion and corrosion control of mild steel in
strate dissolution. The highest corrosion activity was, of course, concentrated H2SO4 solutions by a newly synthesized glycine derivative,
Corros. Sci. 53 (2011) 873–885.
detected for the base-steel, i.e. the substrate specimen. [24] M.M. Pariona, V. Teleginski, K.D. Santos, S. Machado, A.J. Zara, N.K. Zurba, R.
Linear polarization resistance results also showed the highest Riva, Yb-fiber laser beam effects on the surface modification of Al–Fe
polarization resistance (Rp) after the longest exposure to the fur- aerospace alloy obtaining weld filet structures, low fine porosity and
corrosion resistance, Surf. Coat. Technol. 206 (2012) 2293–2301.
nace, i.e. 20 min, where Rp was almost 25 times higher than the [25] W.-M. Zhao, Y. Wang, T. Han, K.-Y. Wu, J. Xue, Electrochemical evaluation of
base-steel specimen. In fact, the calculated values of protection corrosion resistance of NiCrBSi coatings deposited by HVOF, Surf. Coat.
efficiency PEF confirmed an extremely beneficial effect of a subse- Technol. 183 (2004) 118–125.
[26] M.M. Verdian, K. Raeissi, M. Salehi, Corrosion performance of HVOF and APS
quent furnace heat treatment of flame-sprayed NiCrSi coating with
thermally sprayed NiTi intermetallic coatings in 3.5% NaCl solution, Corros. Sci.
the highest value of almost 96% after being heat-treated for 20 min 52 (2010) 1052–1059.
at 1080 °C. [27] G.O. Ilevbare, G.T. Burstein, The inhibition of pitting corrosion of stainless
steels by chromate and molybdate ions, Corros. Sci. 45 (2003) 1545–1569.
Cyclic polarizations confirmed an improved, nobler material
[28] U. Trdan, J. Grum, SEM/EDS characterization of laser shock peening effect on
with lower corrosion current density of additionally heat-treated localized corrosion of Al alloy in a near natural chloride environment, Corros.
flame-sprayed NiCrBSi specimen. The highest decrease of corrosion Sci. 82 (2014) 328–338.
current density icorr was detected after heat treatment for 20 min [29] Test Methods For Determining Area Percentage Porosity in Thermal Sprayed
Coatings, ASTM Standard E2109–01, ASM International, West Conshohocken,
where the corrosion rate was reduced by as much as the factor PA, 03.01., (2003).
of 17 compared to the base-steel specimen, and by the factor of [30] M. Kališnik, Stereology fundamentals (In Slovene), Soc. Stereol. Quant. Image
3 compared to the flame-sprayed specimen. This was due to the Anal. (2002) 99–109.
[31] I. Grabec, J. Gradišek, Description of Random Phenomena (in Slovene), Faculty
more homogenous microstructure with fewer coating defects and of Mechanical Engineering, Ljubljana, 2000. 149–151.
an enhanced surface roughness. [32] J. Suffner, H. Sieger, H. Hahn, S. Dosta, I.G. Cano, J.M. Guilemany, P. Klimczyk, L.
Jaworska, Microstructure and mechanical properties of near eutectic ZrO2–
60%wt.% Al2O3 produced by quenched plasma spraying, Mater. Sci. Eng. A 506
References (2009) 180–186.
[33] F.C. Campbell (Ed.), Understanding the Basics, Fundamentals of Solidification,
Appendix B, ASM International, 2012. 429–446.
[1] J.R. Davies, Handbook of Thermal Spray Technology, ASM International,
[34] Y. Qiao, T.E. Fischer, A. Dent, The effects of fuel chemistry and feedstock
Materials Park, OH, USA, 2004.
powder structure on the mechanical and tribological properties of HVOF
[2] S. Desphande, S. Sampath, H. Zhang, Mechanisms of oxidation and its role in
thermal-sprayed WC–Co coatings with very fine structures, Surf. Coat.
microstructural evolution of metallic thermal spray coatings – case study for
Technol. 172 (2003) 24–41.
Ni–Al, Surf. Coat. Technol. 200 (2005) 5395–5406.
[35] D.C. Silverman, Tutorial on Cyclic Potentiodynamic Polarization Technique,
[3] Z. Bergant, J. Grum, Porosity evaluation of flame-sprayed and heat-treated
presented at CORROSION/98, San Diego, CA, March 23-28. (1998). (reprint of
nickel-based coatings using image analysis, Image Anal. Stereol. 30 (2011) 53–
paper available from NACE International, Houston, TX).
62.
[36] M. Finšgar, S. Fassbender, F. Nicolini, I. Milošev, Polyethyleneimine as a
[4] J. Grum, Z. Bergant, The optimization of powder flame-spraying parameters
corrosion inhibitor for ASTM 420 stainless steel in near-neutral saline media,
using a Taguchi method, Int. J. Microstruct. Mater. Prop. 3 (4/5) (2008) 682–
Corros. Sci. 51 (2009) 525–533.
700.
[37] M. Trueba, S.P. Trasatti, Study of Al alloy corrosion in neutral NaCl by the
[5] J.M. Miguel, J.M. Guilemany, S. Vizcaino, Tribological study of NiCrBSi coating
pitting scan technique, Mater. Chem. Phys. 121 (2010) 523–533.
obtained by different processes, Tribol. Int. 36 (2003) 181–187.
386 Z. Bergant et al. / Corrosion Science 88 (2014) 372–386

[38] M. Stern, A.L. Geary, Electrochemical polarization; I. Polarization curves, J. [41] F. El-Taib Heakal, N.S. Tantawy, O.S. Shehta, Influence of chloride ion
Electrochem. Soc. 104 (1957) 56–63. concentration on the corrosion behavior of Al-bearing TRIP steels, Mater.
[39] W.H. Ailor (Ed.), Handbook on Corrosion Testing and Evaluation, John Wiley Chem. Phys. 130 (2011) 743–749.
and Sons Inc., New York, 1971. 174.
[40] ASTM Standard G59–97, Standard Test Method for Conducting
Potentiodynamic Polarization Resistance Measurements, ASTM International,
West Conshohocken, PA, (2009).