Corrosion Science xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

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

The oxidation performance for Zr-doped nickel aluminide coating by

composite electrodepositing and pack cementation
⁎
Yuhua Zhoua, Xiaofeng Zhaoa, , Chunshan Zhaoa, Wei Haoa, Xin Wanga,c, Ping Xiaob
a
Shanghai Key Laboratory of High Temperature Materials and Precision Forming, Shanghai Jiaotong University, Shanghai 200240, China
b
Material Science Centre, School of Materials, University of Manchester, Grosvenor Street, Manchester M1 7HS, UK
c
Konca Solar Cell Co., Ltd., 168 Yanfeng Road, Yanqiao Town, Huishan District, Wuxi 214174, China

A R T I C L E I N F O A B S T R A C T

Keywords: Ni-Zr composite coatings were prepared by composite electrodepositing with the Zr particle loadings from 0.1 g/
A. Metal coatings L Zr to 0.7 g/L Zr. Then the Ni-Zr coatings were aluminized by the pack cementation and heat-treatment to
B. Thermal cycling obtain Zr-doped β-NiAl coatings. The effects of Zr on the coating microstructure and oxidation performance were
C. Electrodeposited films investigated. The spallation resistance for Zr-doped NiAl coatings was better than the coating without Zr. The
C. Kinetic parameters
coating of 0.5 g/L Zr loading exhibited the best spallation resistance duo to a lower growth rate of the oxide scale
C. Interfaces
and the formation of a mechanically interlocking oxide pegs structure.
C. Oxidation

1. Introduction RE-doped aluminide coating on any type of substrate without depend-

Although NiAl-based coatings especially Pt modified β-NiAl coat-
ings have been widely used as an oxidation resistant coating or a bond
coat in thermal barrier coatings due to the excellent oxidation
resistance [1,2], the cost of the Pt is too high for industrial applications.
Researchers found that doping small amounts of reactive elements (RE),
e.g. zirconium (Zr), to β-NiAl bulks and coatings not only has beneficial
effects on the oxidation behavior, but also drastically saves the cost
compared with that of Pt modified β-NiAl coatings [3,4].
To apply RE-doped β-NiAl coating on the nickel-base superalloy, the
chemical vapor deposition, the physical vapor deposition and pack
cementation techniques were developed [5–8]. Among those methods,
pack cementation method has been widely employed because of its low
cost and easy production. However, pack cementation is a diffusion
process and the alloy elements from the substrate would migrate to the
coating, which can affect the oxidation performance [9,10], and make
the discernment of the roles of RE from the effects of the other alloy
elements difficult. In addition, the content of RE has a significant
influence on the oxidation performance, but the controlling of the
content and distribution of Zr in the coating is difficult. Composite
electroplating, due to the low-cost of instrumentations as well as the
possibility to control the content of the doping particles, has attracted
much attention in recent years [11–14]. When a certain RE particles-
doped Ni coating is electroplated, subsequent pack cementation
aluminizing is applied. This combined process can control both of the
content and distribution of the RE dopants. Moreover, it can obtain the
ing on the superalloy composition.
In the published literature, a lot of efforts have been made during
the last decades to study the impact of RE, such as Zr, Hf, Y, et al. (or
their oxides) which present in the alloys or coatings on the oxidation
resistance [3,4,6–8,15,16]. A number of theories and mechanisms have
been proposed to explain the effect of RE on the oxidation resistance,
such as the oxide scale spallation, the growth rate of the oxide scale, the
transformation of the θ-Al2O3 to α-Al2O3 in the transient oxidation
stage, and the oxide scale stress [17–20]. However, the exact mechan-
isms behind these observations are still in debate, which depend either
on differences in the composition, the microstructure of alloys and the
coatings investigated, or on the differences in the concentration and
techniques of RE doping. For the electrodeposited composite coatings,
the small RE (or RE oxide) particles dispersed exhibited better high
temperature oxidation resistance compared to RE-free coatings, which
have been reported recently. For example, Zhou et al. [21] showed that
the Y2O3-dispersed chromizing coating by electrodepositing and pack
cementation had a superior oxidation resistance. Xu et al. [11] found
that CeO2-dispersed δ-Ni2Al3 coating significantly increased the scale
spalling resistance during cyclic oxidation. However, to the author’s
knowledge, there is no report about the oxidation behavior of the Zr
particles dispersed β-NiAl coatings prepared by electrodepositing and
pack cementation.
Therefore, the Zr-doped β-NiAl coatings were prepared by the
composite electrodepositing and pack cementation in this paper. The
effects of Zr on the aluminide coating microstructure, the oxidation

⁎
Corresponding author.
E-mail address: xiaofengzhao@sjtu.edu.cn (X. Zhao).

http://dx.doi.org/10.1016/j.corsci.2017.04.008
Received 30 September 2016; Received in revised form 13 April 2017; Accepted 17 April 2017
0010-938X/ © 2017 Elsevier Ltd. All rights reserved.

Please cite this article as: Zhou, Y., Corrosion Science (2017), http://dx.doi.org/10.1016/j.corsci.2017.04.008
Y. Zhou et al. Corrosion Science xxx (xxxx) xxx–xxx

Table 1 same, 0.5 μm colloidal alumina was used to polish the surface of the
The compositions and parameters of the nickel-plating. annealed sample. The polished samples (after cleaned in acetone) were
placed in alumina crucibles and cyclically oxidized at 1150 °C using a
Composition (g/L) Parameters
chamber furnace up to 60 h in a laboratory air atmosphere. Every 12 h,
NiSO4·6H2O: 150–300 pH: 4–4.5 the crucibles were removed from the furnace, air-cooled outside the
NiCl2·6H2O: 10–30 Temperature: 55 °C furnace at room temperature for approximately 15 min. At least five
H3BO3: 20–40 Current density: 50–100 mA/cm2
samples for each kind of coating were prepared for examining the
CH3(CH2)10CH2OSO3Na: 0.5 Time: 100 min
Zr: 0/0.1/0.3/0.5/0.7 Agitation: 300 rpm oxidation performances. Samples were removed from the furnace at the
specified intervals for spallation and stress measurements, and then

Fig. 1. The cross-section images of electrodeposited composite coatings with various Zr particle loadings in nickel-plating bath: (a) without Zr, (b) 0.1 g/L, (c) 0.3 g/L, (d) 0.5 g/L and (e)
0.7 g/L. (f) The Zr volume fraction in the composite coatings as a function of Zr particle loadings in nickel-plating bath.

spallation resistance, the growth rate of the oxide scale, the transforma-
tion of the θ-Al2O3 to α-Al2O3 in transient oxidation stage and the oxide
scale stress were investigated. Moreover, the influence mechanisms of
Zr on the growth rate of the oxide scale and the oxidation spallation
resistance were discussed.
returned for further cyclic oxidation. Some samples were withdrawn
from cycling, mounted in epoxy and polished to a mirror finish for
observing the thickness of the oxide scale.
2.2. Characterization

2. Experiment The scanning electron microscopy (SEM, Inspect F50, FEI) in

2.1. Materials
Hastelloy X alloy was used as the substrate and its nominal
composition was 46.8Ni-22Cr-18Fe-9Mo-1.5Co-0.6W-1Mn-1Si-0.1C
(in wt.%). First, the substrates were electrodeposited by nickel-plating
bath with Zr loadings of different concentration. The average particle
size of Zr powder was ∼2 μm. The detailed electrodepositing para-
meters were shown in Table 1. The substrate of
20 mm × 25 mm × 4 mm was used as cathode and a nickel sheet of
99.99% purity with a dimension of 20 mm × 30 mm × 2 mm as the
anode. The substrate and the nickel sheet were ground up to 600#
sandpaper, and the Hastelloy X substrate surface was pretreated by
etching in the HCl solution prior to the electrodepositing. Second, the
electrodeposited samples were pack cementation by the powder of
97 wt.% Al (75–150 μm) and 3 wt.% NH4Cl at 650 °C for 10 h under an
argon atmosphere. According to the literature [11,22], the aluminide
coatings developed on the pure Ni or Ni-base alloys at temperatures
below 700 °C were mainly composed of δ-Ni2Al3. To obtain β-NiAl
phase, the aluminized samples were annealed at 1050 °C for 2 h in a
flowing argon atmosphere. Oxides formed on the surface of the
annealed coatings although annealing was performed in argon atmo-
sphere. In order to ensure the surface condition of the coating was the
conjunction with energy dispersive spectroscope (EDS, Oxford
Instruments) was used to determine the microstructures and chemical
compositions of the samples. X-ray diffraction (XRD, Ultima IV, Rigaku)
with CuKα radiation was used for the phase identification. The volume
fraction of Zr particles in the as-electrodeposited Ni coating, defined as
the ratio between the Zr particles area and the total area of the coating,
was examined by SEM. At least 10 cross-section images at 1000 times
magnification were taken from each sample, which were then analyzed
using an image processing software (ImageJ). The spallation of the
oxide scale was examined by using an optical microscopy (BX51M,
Olympus) and SEM. More than five images by SEM were taken from
each sample to estimate the spallation degree of the oxide scale, which
is defined as the ratio between the spalled area and the total area of the
oxide scale in the image. The analysis was performed using the image
processing software, Image J. The stress and θ-Al2O3 phase in the oxide
scale were measured by photo-stimulated luminescence spectroscopy
(PSLS). The PSLS spectra were collected using a LabRAM HR Evolution
Raman microprobe system (Horiba, France) with a 50× objective lens
and a 532 nm laser source at room temperature. For each sample, at
least 25 positions were collected on the centre of the oxide scale surface
and the average values were taken. All the spectra were fitted by the
Lorentzian-Gaussian function using the Labspec 5 software to obtain the
peak position and peak area.

2
Y. Zhou et al.
Fig. 2. The cross-section micrographs of aluminized coatings with various Zr loadings in nickel-plating bath: (a) without Zr, (b) 0.1 g/L, (c) 0.3 g/L, (d) 0.5 g/L and (e) 0.7 g/L. The (f) is
chemical compositions of point A determined by the EDS in aluminized coating for the 0.1 g/L Zr loadings. The (g)–(i) are the elemental mappings of the cross-section for the aluminized
coating from 0.5 g/L Zr loadings.
3. Results
3.1. The as-electrodeposited coatings
The cross-sections of electrodeposited Ni coatings with different
proportions of Zr metal particles are presented in Fig. 1a–e. The average
thickness of the as-electrodeposited coatings was 65.98 ± 1.22 μm.
When the Zr particle loading in nickel-plating bath was 0.1 g/L, the
distribution of bright Zr particles in the coating could be observed in
sparse distribution, and the particles were not visible in most areas.
When the Zr loadings increased from 0.3 g/L to 0.7 g/L, the bright Zr
particles were homogeneously dispersed in the coatings. Moreover, the
volume fraction of Zr particles increased with the particle loadings in
nickel-plating bath from 0.1 g/L to 0.7 g/L, as plotted in Fig. 1f.
3.2. The aluminized coatings
Fig. 2a–e compares the cross-section micrographs of aluminized
samples for both pure nickel and Zr-doped composite coatings. All the
aluminized coatings consisted of two layers, i.e., the outer Ni2Al3 layer
(the XRD patterns in Fig. 3a) and the inner pure nickel or Zr-doped Ni
composite layer that was not aluminized (for example, the aluminized
coating of 0.5 g/L Zr loading, there is no Al in the inner layer, see the
elemental mappings in Fig. 2g–i). The average thickness of the outer
layer was 73.04 ± 0.86 μm and the inner layer was 30.11 ± 2.05 μm.
In addition, voids around the Zr particles were observed on the Zr-
3
Corrosion Science xxx (xxxx) xxx–xxx
doped aluminized coatings. But no voids were seen in the aluminized
coating without Zr. In fact, the Zr particles have transformed to the
compound containing the Zr, Al and Ni (Fig. 2f), which would result in
the volume change. This volume change might not match the volume
change owing to the conversion of the Ni layer into the δ-Ni2Al3 [14],
which would contribute to the formation of voids around the Zr
compound.
3.3. The post-heat treatment coatings
Fig. 4a–e shows the micrographs of post-heat treatment coatings. All
the coatings consisted of three layers, i.e., Al rich β-NiAl (Al > 50 at.
%) outer layer, Ni rich β-NiAl (Ni > 50 at.%) interlayer and γ’-Ni3Al
layer. The average thickness of the outer layer was 58.83 ± 2.97 μm,
21.00 ± 1.34 μm for the inner layer, and 11.67 ± 0.78 μm for the γ’-
Ni3Al layer. Based on the elemental mappings by EDS (Fig. 4f–l), the
coating was without any alloy elements from the substrate, such as Fe,
Cr, Co and Mo. Particularly, in addition to the existence in the coating,
part of element Zr diffused to the coating/substrate interface
(Fig. 4b–e). The XRD analysis in Fig. 3b combining the chemical
compositions in Table 2 revealed that the bright zone in the outer
layer of the coatings (Fig. 4b–e) was Al5Ni2Zr phase. Moreover, the
voids were found in the outer layer in all the coatings, and large
number of voids also existed in the coating/substrate interface of the
0.7 g/L Zr loading (Fig. 4e). Supposedly the δ-β phase transformation,
as well as the unequal inter-diffusion fluxes of the Zr, Al and Ni duo to
Y. Zhou et al.
Fig. 3. The XRD patterns of the aluminized coatings (a) and the post-heat treatment
coatings (b).
the different diffusion rates would lead to the void formation and
development. The Zr content in the post-heat treatment coating, which
was analyzed by the EDS (using area-scanning, the schematic diagram
can be seen in Fig. 5a) and resulted from at least five images for each
sample, increased with an increasing of the Zr particles in nickel-plating
bath from 0.1 g/L to 0.7 g/L (Fig. 5b).
3.4. The performance of oxidation spallation resistance
Figs. 6 and 7 a–m illustrate the optical and SEM images of the oxide
scale surface after oxidation at 1150 °C. The spallation degree, defined
as the ratio of the spalled area over the total area of the oxide scale, is
plotted as a function of the oxidation time in Fig. 7n. After 12 h, the
coating without Zr suffered the worst spallation (Figs. 6 a, 7 a and n).
The coatings of 0.1 g/L Zr and 0.3 g/L Zr loadings exhibited large
amount of oxide spallation after 24 h (Figs. 6 a and b, 7 f and g and n).
The spallation degree for the coatings of 0.5 g/L Zr and 0.7 g/L Zr was
similar and smaller than other coatings until 48 h (Fig. 7n). After 60 h,
the coating of 0.7 g/L Zr delaminated from the substrate. However, the
oxide scale was remained intact for the coating of 0.5 g/L Zr loading
4
Corrosion Science xxx (xxxx) xxx–xxx
(Fig. 6d), which indicated that the coating prepared from 0.5 g/L Zr
loading exhibited the best spallation resistance.
3.5. The transient alumina and the growth kinetics of the oxide scale
3.5.1. The transient alumina
It is well established that the oxidation resistance of the nickel-
aluminide coating depends on the formation of an adherent, continuous
and dense oxide scale, predominantly α-Al2O3. However, numerous
studies have clearly identified that there existed a transient stage during
oxidation [3,17,18,23]. In the early stage of oxidation (transient stage),
transient metastable phases such as γ, δ or θ-Al2O3 phases will form.
When the oxidation time is extended, transient alumina will transform
to α-Al2O3 until the coating surface is covered with a continuous and
protective α-Al2O3 layer. The growth rate of transient alumina, such as
θ-Al2O3, is sensibly higher than that of α-Al2O3 [3,24]. From this point
of view, the growth of the oxide scale is also controlled by the
formation of transient alumina. In order to evaluate the transient
alumina, the PSLS spectra of each oxide scale after oxidation for
10 min, 30 min, 60 min and 90 min were acquired. As shown in
Fig. 8a, after oxidation for 10 min, apart from the α-Al2O3, the
characteristic peaks of θ-Al2O3 (∼14,540 and ∼14,603 cm−1) were
found. The θ-Al2O3 peaks were different to those found in Ref. [25]
(∼14,575 and ∼14,650 cm−1) due to different residual stresses in the
oxide scale. After oxidation for 90 min (Fig. 8b), for the 0.3 g/L Zr,
0.5 g/L Zr and 0.7 g/L Zr samples, θ-Al2O3 peaks were almost dis-
appeared, which indicated that the phase transformation of θ to α-
Al2O3 was basically accomplished. Labspec 5 software was used to fit
the spectrum curve (Fig. 8c) and obtained the peak areas of the α-Al2O3
and the θ-Al2O3. Then the θ-Al2O3 content in the oxide scale was
calculated using the following equation:
A14540 + A14603
Cθ − Al2 O3 =
A14540 + A14603 + AR1 + AR2 (1)
where A14540 and A14603 are the peak areas at 14,540 and 14,603 cm−1
for θ-Al2O3, respectively, and AR1 and AR2 are the peak areas of the
characteristic R1 and R2 peaks for α-Al2O3. The evolution of the θ-
Al2O3 content with oxidation time for different samples is presented in
Fig. 8d. The θ-Al2O3 content decreased with an increasing of the Zr
particles in the nickel-plating bath from 0.1 g/L to 0.7 g/L, which
means Zr accelerates the phase transformation of θ to α-Al2O3.
3.5.2. The growth kinetics of the oxide scale
The oxide scale growth follows the relationship [26]:
h2 = 2kp t (2)
where h is the average thickness of the oxide scale after oxidation for
time t and kp is the parabolic growth rate constant having units of
cm2 s−1. The average oxide scale thickness was calculated through the
cross-sectional area of the continuous oxide scale divided by the length
of interface (the internal oxidation and oxide pegs in the coatings also
were considered) in the SEM image. At least 5 SEM images were used to
obtain average thickness datum for each sample. The average thickness
of the oxide scale versus square root of oxidation time can be seen in
Fig. 9 and the values of the growth rate constant can be given by the
linear fitting method.
Generally, the parabolic rate law cannot be employed when the
spallation of the oxide scale is not negligible. According to the
spallation condition in Figs. 6 and 7, the coating without Zr, 0.1 g/L
Zr and 0.3 g/L Zr doped coatings exhibited large amount of oxide
spallation after 12 h and 24 h, respectively. In addition, the coating of
0.7 g/L Zr loading delaminated from the substrate after 60 h (Fig. 6d).
When the cross section of the 0.7 g/L Zr sample after oxidation for 12 h
Y. Zhou et al. Corrosion Science xxx (xxxx) xxx–xxx

Fig. 4. The cross-section images of post-heat treatment coatings with various Zr particle loadings in nickel-plating bath: (a) without Zr, (b) 0.1 g/L, (c) 0.3 g/L, (d) 0.5 g/L and (e) 0.7 g/L.
The (f)–(l) are the elemental mappings of the cross-section for the coating from 0.5 g/L Zr loadings.

Table 2 delaminated from the substrate) presented in Table 3 indicated that Zr

Chemical compositions of points 1–4 in Fig. 4b–d. doped coatings showed a lower oxidation growth rates, as compared to
the un-doped NiAl coating. Also, the 0.3 g/L and 0.5 g/L Zr doped
Point Elements content (in at.%)
coatings exhibited a lower growth rate than other Zr doped coatings.
Al Ni Zr

1 56.03 29.04 14.93 3.6. The oxide scale stress

2 61.19 24.48 14.33
3 57.18 28.89 13.92
4 – 81.76 18.24 It is widely accepted that the oxide scale stress consists of the
growth stress (induced by the oxide scale growth) and the thermal
expansion mismatch stress (a consequence of differences in the thermal
was prepared in an epoxy mount and polished for SEM analysis, the expansion mismatch between the oxide scale and substrate during
coating had delaminated from the substrate (as shown in the inset in cooling). The oxide scale stress is a dynamic competition between the
Fig. 9). Therefore, the data of the oxide scale thickness for the 0.7 g/L stress generation and concurrent relaxation through creep, crack or
Zr sample after oxidation for 12 h were not used. The kp values (for a spallation. Measurements of the stress in the oxide scale were made by
given time period before high spallation happened and the coating PSLS technique at room temperature. The stress was obtained from the

5
Y. Zhou et al.
Fig. 5. (a) The schematic diagram of area-scanning by the EDS and (b) the Zr content (at.%) in the post-heat treatment coatings for samples with 0.1 g/L–0.7 g/L Zr loadings.
frequency shift [27]:
σ = Δυ 5.07
where σ is the average stress of the oxide scale, and Δν is the frequency
shift of the characteristic R-lines. Since the R2 luminescence line of α-
Al2O3 had a nearly linear dependence on stress [28,29], the stress in
oxide scale was estimated from the R2 frequency shift in this study (the
stress-free frequency of the R2 came from the single crystal sapphire
standard).
The average compressive stresses in the oxide scale of these coatings
after oxidation between 12 h and 48 h are shown in Fig. 10a (the
negative sign represents the compressive stress). The oxide scale stress
of the coating without Zr at 12 h, the oxide scale stresses of the 0.1 g/L
Zr and 0.3 g/L Zr doped coatings at 24 h were markedly decreased,
which resulted from the stress relaxation caused by oxide spallation
(Figs. 6 and 7). Before spallation (e.g. 12 h), the oxide scale stresses of
all Zr doped coatings were similar.
In order to better observe the oxide scale stress of the coating
without Zr doping, the stress of the initial oxide period before 12 h in
which the oxide scale has not spalled was provided, as shown in
Fig. 10b. The spallation of the oxide scale on the coating without Zr
occurred after oxidation for 4 h so that at that moment the measured
stress was lower than that of the other coatings. However, before
oxidation for 4 h, especially oxidation for 1 h, the oxide scale stresses of
the Zr-doped and the un-doped NiAl coatings were basically in the same
Fig. 6. Representative optical images of the spallation conditions for the coatings without Zr, 0.1 g/L Zr, 0.3 g/L Zr, 0.5 g/L Zr and 0.7 g/L Zr loadings after oxidation at 1150 °C for (a)
12 h, (b) 24 h, (c) 36 h and (d) 60 h.
Corrosion Science xxx (xxxx) xxx–xxx
level.
(3)
4. Discussion
4.1. Factors influencing the oxide scale growth kinetics
The data of early stage oxidation in Fig. 8 showed that the θ-Al2O3
content decreased with an increase of the Zr particles in the nickel-
plating bath from 0.1 g/L to 0.7 g/L, which suggests that Zr accelerates
the phase transformation of θ to α-Al2O3. The θ to α-Al2O3 transforma-
tion can be influenced by a number of factors and overall the
transformation mechanisms are far from being fully understood
[18,24,30–33]. Brumm and Grabke [24] found that the content of
dopant (Cr) can affect the θ to α-Al2O3 phase transformation. The
higher the dopant (Cr) content (that means more formation of Cr oxide
at the initial oxidation) and the more α-Al2O3 nuclei were present, the
shorter was also the transition time of the θ to α-Al2O3 transformation.
Zhao et al. [33] found that at the early stage of high-temperature
oxidation, the RE (Dy) in the coating was preferentially oxidized to
form RE oxides because of its higher oxygen affinity than that of Al. The
formation of RE oxides provided nucleation sites for the growth of α-
Al2O3, which accelerated the θ to α-Al2O3 phase transformation. In this
paper, after oxidation for 10 min, when combining the results of PSLS
spectra in Fig. 8a and SEM characterization in Fig. 11a and c, it can be
concluded that these platelets or needles were θ-Al2O3 (the morphology
6
Y. Zhou et al.
Fig. 7. SEM images (a–m) of the spalled oxide surface for the coatings of without Zr, 0.1 g/L Zr, 0.3 g/L Zr, 0.5 g/L Zr and 0.7 g/L Zr loadings after oxidation at 1150 °C for 12 h, 24 h,
36 h and 48 h. n is the spallation degree of coatings as a function of oxidation time at 1150 °C.
of θ-Al2O3 can reference [30,32]) for the un-doped NiAl coating. For the
coating of 0.7 g/L Zr loading, oxides containing Zr (Table 4) formed at
the coating surface (Fig. 11b and d) and θ-Al2O3 distributed between
these oxides, as shown in Fig. 11d (this phenomenon also can be seen in
the samples of 0.1 g/L Zr, 0.3 g/L Zr and 0.5 g/L Zr loadings). The
higher the Zr content, the more oxides containing Zr for α-Al2O3 nuclei,
and more benefits were for the θ to α-Al2O3 transformation. Because θ-
Al2O3 has a higher growth rate than α-Al2O3 [3,24], then the change of
the θ-Al2O3 content in the oxide scale would lead to a significant
difference in the oxide scale thickness and the growth rate. Fig. 12
presents the parabolic rate constant for different coatings. During the
transient regime (such as 0–1.5 h), kp values for the coating without Zr
was much higher than other coatings. One reason is the relatively high
θ-Al2O3 content (Fig. 8) and another reason is the formation of a large
internal oxidation (the inset of Fig. 12a). After the transient regime
(when the θ-Al2O3 basically transformed into α-Al2O3), kp values
reduced except the 0.1 g/L Zr doped coating. It is because of the
formation of a large internal oxidation in the 0.1 g/L Zr doped coating,
seen the inset in Fig. 12b. So the high θ-Al2O3 content, as well as the
internal oxidation, lead to a higher oxide growth rate for the coating
without Zr and 0.1 g/L Zr doped coating. For the 0.3 g/L Zr, 0.5 g/L Zr
and 0.7 g/L Zr doped coatings, although θ-Al2O3 content decreased
with an increase of the Zr content in the baths and coatings, kp value for
7
Corrosion Science xxx (xxxx) xxx–xxx
the 0.7 g/L Zr doped coating was higher than 0.3 and 0.5 g/L Zr doped
coatings. It is owing to a large number of oxide pegs formed at the
oxide/coating interface for the 0.7 g/L Zr doped coating (Fig. 13).
These internal oxide pegs contribute to a higher growth rate constant
(Figs. 9 and 12 b) for the 0.7 g/L Zr doped coating compared with the
0.3 and 0.5 g/L Zr doped coatings.
4.2. Factors influencing the spallation of the oxide scale
In general, the spallation is driven by the residual stress in the oxide
scale and countered by the interfacial adherence between the oxide
scale and the coating [26,34]. In this paper, before the oxide scale
spallation, the residual stress in the oxide scale (Fig. 10) was almost the
same. This signified that the residual stress may be not the decisive
factor to induce the spallation of the oxide scale. Therefore, there must
be other reasons for the difference in the spallation behavior of the
oxide scale.
The benefits of RE (such as Zr, Hf and Y) additions were well
documented in previous studies [3,6,15,35–37]. RE additions, in
particular, markedly reduce the scale growth rate. Furthermore, RE
additions may improve the oxide scale adhesion duo to the mechanical
interlocking of the oxide pegs formed at the oxide/metal interface.
From Fig. 9, it can be seen that Zr-doped coatings showed a lower
Y. Zhou et al. Corrosion Science xxx (xxxx) xxx–xxx

Fig. 8. (a) and (b) PSLS spectra of each oxide scale after 10 min and 90 min transient
oxidation at 1150 °C. (c) A typical spectrum from oxide scale containing both θ-Al2O3 and α-
Table 4
Chemical compositions of points 5–7 in Fig. 11d.

Point Elements content (in at.%)

O Al Zr Ni

5 42.02 57.98 – –
6 34.26 49.89 8.15 7.70
7 35.99 52.39 5.20 6.42

Table 5
Chemical compositions of points 8–10 in Fig. 14d.

Point Elements content (in at.%)

O Al Zr
Fig. 9. Plots of oxide scale thickness versus oxidation time t0.5 and the inset is the 0.7 g/L
Zr doped coating after oxidation at 1150 °C for 12 h (the coating had delaminated from
8 42.05 47.94 10.01
the substrate).
9 43.83 43.28 12.89
10 42.74 49.37 7.89
Table 3
The values of the oxidation parabolic rate constant kp for the samples after oxidation at
1150 °C during a given time period.
oxidation growth rates when compared to the un-doped NiAl coating.
Samples Given time period (h) kp (cm2 s−1)
Besides, many pegs were found in the Zr-doped coatings after oxidation
Without Zr 0–12 (4.18 ± 1.25) × 10−10 (Figs. 13 and 14).
0.1 g/L Zr 0–12 (1.29 ± 0.47) × 10−10 It was important to note that although the 0.3 g/L and 0.5 g/L Zr
0.3 g/L Zr 0–24 (1.56 ± 0.16) × 10−11
0.5 g/L Zr 0–24 (1.22 ± 0.13) × 10−11
0.7 g/L Zr 0–8 (3.44 ± 0.22) × 10−11

8
Y. Zhou et al.
Fig. 10. Average stress in the oxide scale formed on the coatings of the different Zr particle loadings for oxidation at 1150 °C with the oxidation time: (a) between 12 h and 48 h, (b)
before 12 h.
doped coatings exhibited a similar growth rate (Fig. 9) and oxide scale
stress (Fig. 10), the 0.5 g/L Zr doped coating presented a relatively
better spallation resistance. It is because of more pegs (Fig. 14) forming
in the 0.5 g/L Zr doped coating than that in the 0.3 g/L Zr doped
coating (As the studies by Hong et al. [36] suggested, a higher Zr
content in the aluminide coating had more pegs). The high-magnified
image of the oxide peg in the 0.5 g/L Zr doped coating and the
corresponding chemical compositions of the oxide peg are presented
in Fig. 14d and Table 5. It can be inferred that the oxides pegs mostly
were composed of Zr-rich oxide core and an outer alumina sheath, and
this result was similar with the published reports [15,36]. Oxide pegs
play a role of ‘anchoring’ at the interface between the oxide scale and
coating, which makes the oxide scale adhesion of 0.5 g/L Zr doped
coating stronger. In other words, for determining coating lifetime, the
contribution from the oxide scale adhesion appears to be much more
important than the effect on the oxide scale growth.
However, too many pegs are not beneficial. Such as the pegs formed
in 0.7 g/L Zr doped coating, it intruded deeply into the coating and
induced further internal oxidation (Fig. 14c). This kind of pegs would
accelerate the coating oxidation and finally result in internal failure of
the coating [15]. At the same time, for the 0.7 g/L Zr doped coating, the
coating delaminated from the substrate (Fig. 6). It is due to the voids at
9
Corrosion Science xxx (xxxx) xxx–xxx
the interface of the coating/substrate in the post-heat treatment coating
(see Fig. 4e). During oxidation, oxygen permeates the coating/substrate
interface along the voids and creates oxides at the interface, which
destroys the interface continuity and reduces the bonding strength of
the interface, finally results in the delamination of coating from the
substrate (the coating/substrate interface condition after oxidation can
be seen in Figs. 13 c and 14 c).
For the un-doped NiAl coating and 0.1 g/L Zr doped coating, the
heavy internal oxidation also was found (the insets in Fig. 12). But it
was different from the internal oxidation of 0.7 g/L Zr doped coating
which arose from the deep oxide pegs by the over-doping of RE Zr.
Kaplin and Brochu proposed that an extension of the transient oxidation
stage can lead to internal oxidation [38]. When the oxide scale on the
coating is incomplete or loose, oxygen can pass through the scale and
diffuse into the coating along grain boundaries, which could develop
the internal oxidation. During the transient oxidation stage, an incom-
plete Al2O3 layer (θ-Al2O3 mixed α-Al2O3) gradually changed into a
continuous and dense layer of α-Al2O3 with the extension of the
oxidation time (Fig. 8a and b). However, the θ-Al2O3 content decreased
from 0.1 g/L Zr to 0.7 g/L Zr doped coatings (Fig. 8d). That means more
θ-Al2O3 formed for the un-doped NiAl coating and 0.1 g/L Zr doped
coating at the same transient oxidation time, which allows for oxygen
Y. Zhou et al.
Fig. 11. Surface morphology (secondary electron) of oxide scale after oxidation at 1150 °C for 10 min: (a) low-magnified, without Zr, (b) low-magnified, 0.7 g/L Zr, (c) high-magnified,
without Zr and (d) high-magnified, 0.7 g/L Zr.
diffusion passing the oxide scale easier into the coating to form the
internal oxidation. Due to the large volume expansion caused by serious
internal oxidation, the locale high stress generates at the oxide scale/
coating interface in or near the internal oxidation zone. The locale high
stress is susceptible to the crack nucleation and propagation, and then
causes the oxide scale to spall from the coating for the un-doped NiAl
and 0.1 g/L Zr samples.
All in all, 0.5 g/L Zr doped coating exhibited a lower growth rate
(Fig. 9), and appropriate pegs (Fig. 14) which markedly improved the
oxide scale adhesion. So the 0.5 g/L Zr doped coating presented a better
spallation resistance (Figs. 6 and 7).
5. Conclusions
The Zr-doped β-NiAl coatings were prepared by the composite
electrodepositing and pack cementation. The effects of Zr on the
aluminide coating microstructure and oxidation performance were
investigated. The following conclusions can be drawn:
10
Corrosion Science xxx (xxxx) xxx–xxx
(1) Incorporation of Zr in the β-NiAl coating improved the spallation
resistance of the oxide scale and reduced the oxidation rate. The
best spallation resistance and the lowest oxidation growth rate were
exhibited by the coating generated from the 0.5 g/L Zr solution.
(2) The Zr content of the β-NiAl coating had no influence on the stress
in oxide scale. The stresses for the Zr-doped and the un-doped NiAl
coatings were all approximately the same at the stage before the
oxide scale has spalled.
(3) At the transient oxidation stage, the Zr in the coating was
preferentially oxidized to form oxides containing Zr. The formation
of oxides containing Zr provided nucleation sites for the growth of
α-Al2O3, which accelerated the θ to α-Al2O3 phase transformation.
This led to a significant difference in the θ-Al2O3 content and the
growth rate of the oxide scale.
(4) The formation of oxide pegs at the interface of the oxide scale/
coating can improve the oxide scale adhesion. However, too many
pegs can induce internal oxidation, which were not beneficial for
the scale adhesion.
Y. Zhou et al. Corrosion Science xxx (xxxx) xxx–xxx

Fig. 12. (a) Parabolic rate constants (kp/cm2 s−1) evolution during oxidation, and the inset is the internal oxidation for the coating without Zr after oxidation for 90 min. (b) The
enlargement of the parabolic rate constants in (a) for 0.1 g/L Zr, 0.3 g/L Zr, 0.5 g/L Zr and 0.7 g/L Zr doped coatings, and the inset is the internal oxidation for the 0.1 g/L Zr doped
coating after oxidation for 12 h.

Fig. 13. The oxide pegs at the oxide scale/coating interface after oxidation for 90 min: (a) 0.3 g/L Zr, (b) 0.5 g/L Zr and (c) 0.7 g/L Zr doped coatings.

11
Y. Zhou et al. Corrosion Science xxx (xxxx) xxx–xxx

Fig. 14. The oxide pegs at the oxide scale/coating interface after oxidation for 24 h: (a) 0.3 g/L Zr, (b) 0.5 g/L Zr, (c) 0.7 g/L Zr doped coatings and (d) the high-magnified image of the
oxide pegs in the 0.5 g/L Zr doped coating.

Acknowledgements aluminide coating prepared by pack cementation, Corros. Sci. 110 (2016) 284–295.
[11] C. Xu, X. Peng, F. Wang, Cyclic oxidation of an ultrafine-grained and CeO2-
dispersed δ-Ni2Al3 coating, Corros. Sci. 52 (2010) 740–747.
The authors would thank for the financial support from the “Qianren [12] Y.B. Zhou, J.F. Sun, S.C. Wang, H.J. Zhang, Oxidation of an electrodeposited Ni-
Plan” Platform Funding, the Program for Professor of Special Appointment Y2O3 composite film, Corros. Sci. 63 (2012) 351–357.
[13] Z. Dong, X. Peng, Y. Guan, L. Li, F. Wang, Optimization of composition and
(Eastern Scholar) at Shanghai Institutions of Higher Learning (No. structure of electrodeposited Ni-Cr composites for increasing the oxidation resis-
SHDP201303), the National Natural Science Foundation of China (No. tance, Corros. Sci. 62 (2012) 147–152.
51271120). [14] X. Tan, X. Peng, F. Wan, The mechanism for self-formation of a CeO2 diffusion
barrier layer in an aluminide coating at high temperature, Surf. Coat. Technol. 224
(2013) 62–70.
References [15] D.Q. Li, H.B. Guo, D. Wang, T. Zhang, S.K. Gong, H.B. Xu, Cyclic oxidation of β-NiAl
with various reactive element dopantsat 1200 °C, Corros. Sci. 66 (2013) 125–135.
[16] Y.Q. Wang, J.L. Smialek, Marc Suneson, Oxidation behavior of Hf-modified
[1] J.R. Nicholls, Advances in coating design for high-performance gas turbines, MRS
aluminide coatings on Inconel-718 at 1050 °C, J. Coat. Sci. Technol. 1 (2014)
Bull. 28 (2003) 659–670.
25–45.
[2] K. Shirvani, S. Firouzi, A. Rashidghamat, Microstructures and cyclic oxidation
[17] B.A. Pint, Progress in Understanding the Reactive Element Effect Since the Whittle
behaviour of Pt-free and low-Pt NiAl coatings on the Ni-base superalloy Rene-80,
and Stringer Literature Review, Metals and Ceramics Division, Oak Ridge National
Corros. Sci. 55 (2012) 378–384.
Laboratory, Oak Ridge, 2017.
[3] S. Hamadi, M.P. Bacos, M. Poulain, A. Seyeux, V. Maurice, P. Marcus, Oxidation
[18] B.A. Pint, M. Treska, L.W. Hobbst, The effect of various oxide dispersions on the
resistance of a Zr-doped NiAl coating thermochemically deposited on a nickel-based
phase composition and morphology of A12O3 scales grown on β-NiA1, Oxid. Met.
superalloy, Surf. Coat. Technol. 204 (2009) 756–760.
47 (1997) 1–20.
[4] M. Zagula-Yavorska, J. Sieniawski, J. Romanowska, Oxidation behaviour of
[19] R.J. Christensen, V.K. Tolpygo, D.R. Clarke, The influence of the reactive element
zirconium-doped NiAl coatings deposited on pure nickel, Arch. Mater. Sci. Eng. 58
yttrium on the stress in alumina scales formed by oxidation, Acta Mater. 45 (1997)
(2012) 250–254.
1761–1766.
[5] J. Romanowska, M. Zagula-Yavorska, J. Sieniawski, J. Markowski, Zirconium
[20] I.J. Bennett, W.G. Sloof, Modelling the influence of reactive elements on the work of
modified aluminide coatings obtained by the CVD and PVD methods, Open J. Met. 3
adhesion between a thermally grown oxide and a bond coat alloy, Mater. Corros. 57
(2013) 92–99.
(2006) 223–229.
[6] H.B. Guo, L.D. Sun, H.F. Li, S.K. Gong, High temperature oxidation behavior of
[21] Y.B. Zhou, H. Chen, H. Zhang, Y. Wang, Preparation and oxidation of an Y2O3-
hafnium modified NiAl bond coat in EB-PVD thermal barrier coating system, Thin
dispersed chromizing coating by pack cementation at 800 °C, Vacuum 82 (2008)
Solid Films 516 (2008) 5732–5735.
748–753.
[7] X.S. Zhao, C.G. Zhou, Effect of Y2O3 content in the pack on microstructure and hot
[22] X. Tan, X. Peng, F. Wang, The effect of grain refinement on the adhesion of an
corrosion resistance of Y-Co-modified aluminide coating, Corros. Sci. 86 (2014)
alumina scale on an aluminide coating, Corros. Sci. 85 (2014) 280–286.
223–230.
[23] G.C. Rybicki, J.L. Smialek, Effect of the θ-α-A12O3 transformation on the oxidation
[8] Y.F. Yang, C.Y. Jiang, H.R. Yao, Z.B. Bao, S.L. Zhu, F.H. Wang, Preparation and
behavior of β-NiAl + Zr, Oxid. Met. 31 (1989) 275–304.
enhanced oxidation performance of a Hf-dopedsingle-phase Pt-modified aluminide
[24] M.W. Brumm, H.J. Grabke, The oxidation behavior of NiAl-I. Phase transformations
coating, Corros. Sci. 113 (2016) 17–25.
in the alumina scale during oxidation of NiAl and NiAl-Cr alloys, Corros. Sci. 33
[9] H.M. Tawancy, A.I. Mohamed, N.M. Abbas, Effect of superalloy substrate compo-
(1992) 1677–1690.
sition on the performance of a thermal barrier coating system, J. Mater. Sci. 38
[25] Q.Z. Wen, D.M. Lipkin, D.R. Clarke, Luminescence characterization of chromium-
(2003) 3797–3807.
containing θ-alumina, J. Am. Ceram. Soc. 81 (1998) 3345–3348.
[10] Y.H. Zhou, L. Wang, G. Wang, D.L. Jin, W. Hao, X.F. Zhao, J. Zhang, P. Xiao,
[26] A.G. Evans, D.R. Mum, J.W. Hutchinson, G.H. Meier, Mechanisms controlling
Influence of substrate composition on the oxidation performance of nickel

12
Y. Zhou et al.
durability of thermal barrier coatings, Prog. Mater. Sci. 46 (2001) 505–553.
[27] R.J. Christensen, D.M. Lipkin, D.R. Clarke, Nondestructive evaluation of the
oxidation stresses through thermal barrier coatings using Cr3+ piezospectroscopy,
Appl. Phys. Lett. 69 (1996) 3754–3756.
[28] J. He, D.R. Clarke, Determination of the piezospectroscopic coefficients for
chromium-doped sapphire, J. Am. Ceram. Soc. 78 (1995) 1347–1353.
[29] A. Selcuk, A. Atkinson, The evolution of residual stress in the thermally grown oxide
on Pt diffusion bond coats in TBCs, Acta Mater. 51 (2003) 535–549.
[30] K.M.N. Prasanna, A.S. Khanna, R. Chandra, W.J. Quadakkers, Effect of θ-alumina
formation on the growth kinetics of alumina-forming superalloys, Oxid. Met. 46
(1996) 465–479.
[31] H.J. Grabke, Oxidation of NiAl and FeAl, Intermetallics 7 (1999) 1153–1158.
[32] H.B. Guo, D. Wang, H. Peng, S.K. Gong, H.B. Xu, Effect of Sm, Gd, Yb, Sc and Nd as
reactive elements on oxidation behaviour of β-NiAl at 1200 °C, Corros. Sci. 78
(2014) 369–377.
13
Corrosion Science xxx (xxxx) xxx–xxx
[33] X.Y. Zhao, H.B. Guo, Y.Z. Gao, S.X. Wang, S.K. Gong, Effects of Dy on transient
oxidation behavior of EB-PVD β-NiAl coatings at elevated temperatures, Chin. J.
Aeronaut. 24 (2011) 363–368.
[34] S. Bose, High temperature coatings, Oxidation, Elsevier Science & Technology
Books, 2007, pp. 29–52.
[35] B.A. Pint, The role of chemical composition on the oxidation performance of
aluminide coatings, Surf. Coat. Technol. 188–189 (2004) 71–78.
[36] S.J. Hong, G.H. Hwang, W.K. Han, K.S. Lee, S.G. Kang, Effect of zirconium addition
on cyclic oxidation behavior of platinum-modified aluminide coating on nickel-
based superalloy, Intermetallics 18 (2010) 864–870.
[37] B.A. Pint, Experimental observations in support of the dynamic-segregation theory
to explain the reactive-element effect, Oxid. Met. 45 (1996) 1–37.
[38] C. Kaplin, M. Brochu, Effects of water vapor on high temperature oxidation of
cryomilled NiCoCrAlY coatings in air and low-SO2 environments, Surf. Coat.
Technol. 205 (2011) 4221–4227.