Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

Corrosion Science 49 (2007) 1777–1798

www.elsevier.com/locate/corsci

Protective coating on Mg AZ91D alloy – The


effect of electroless nickel (EN) bath stabilizers
on corrosion behaviour of Ni–P deposit
a,* a,b a
Woo-Jae Cheong , Ben L. Luan , David W. Shoesmith
a
Department of Chemistry, University of Western Ontario, Chemistry Building, 1151 Richmond St. London,
Ont., Canada N6A 5B7
b
Integrated Manufacturing Technologies Institute, the National Research Council Canada, London, Ont.,
Canada N6G 4X8

Received 1 November 2005; accepted 21 August 2006


Available online 25 January 2007

Abstract

The effects of the electroless nickel (EN) bath stabilizers; thiourea (TU) and maleic acid (MA), on
the corrosion of the EN coatings on Cu and Mg substrates were investigated in an aggressive oxy-
genated chloride environment.
Corrosion was limited to the external EN layer, and thus, the intermediate Cu layer as well as Mg
AZ91D substrate remained intact. On the EN surface, two different types of corrosion morphologies
were observed. Specimens prepared with TU showed preferential attack along particle boundaries,
while pitting corrosion was observed for specimens prepared either without a stabilizer or with
MA. The latter deposits showed higher corrosion resistances due, in part, to the formation of a
P-enriched layer on the surface. The poorer corrosion performance of EN specimens prepared with
TU was attributed to the presence of the particle boundaries and the trace impurities adsorbed on,
or included in, them. The key impurity was sulphur incorporated during formation of deposits in
TU-stabilized baths.
It was clearly shown that the corrosion behaviour of the EN specimens was predominantly influ-
enced by the morphological and surface compositional aspects of the deposits.
Crown Copyright  2006 Published by Elsevier Ltd. All rights reserved.

*
Corresponding author. Tel.: +1 519 661 2111x86248; fax: +1 519 661 3022.
E-mail addresses: wcheong@uwo.ca, cheongw@gmail.com (W.-J. Cheong).

0010-938X/$ - see front matter Crown Copyright  2006 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2006.08.025
1778 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

Keywords: A. Magnesium AZ91D alloy; Metal coating; B. EIS; SEM; C. Corrosion electrochemical behaviour

1. Introduction

Magnesium (Mg) and its alloys are plated for various reasons, but primarily to improve
their corrosion resistance. While Mg is relatively stable under alkaline conditions due to
the rapid formation of the hydroxide, Mg(OH)2, it is subject to severe attack in acidic
and chloride environments. The metal is also prone to galvanic attack when in contact
with another metal [1]. Various methods for applying protective coatings, such as electro-
chemical plating, chemical conversion coating, hydride coating, anodizing, gas phase
deposition, and organic/polymer coatings have been tried [2]. Since it requires a minimum
capital investment, chemical conversion coating has been extensively investigated. How-
ever, the use of CrO3 or chromate solutions for the conversion process is harmful to the
environment. Moreover, a conversion coating can only provide a suitable base for an
adherent organic coating [3].
Electroless deposition can also provide a protective coating for common metal sur-
faces [1,4,5]. In general, an electroless nickel (EN) deposit is less porous and more uni-
form than an equivalent electrolytically plated deposit, and thus provides better
corrosion and wear protection [6]. However, electroless deposition has the inherent prob-
lem of sudden bath decomposition, which can be aggravated in the case of reactive sub-
strate plating (i.e. Mg alloys), since the dissolved metal ions enhance decomposition. This
results in both an increase in costs and the production of environmentally hazardous pol-
lutants due to large waste generation [5]. To overcome this problem, EN baths com-
monly contain small amounts of stabilizers to prevent the homogeneous reactions that
trigger decomposition [5]. Previously, the stability of EN baths were evaluated with an
accelerated stability test in which the bath decomposition time was measured after the
addition of the optimum concentration of PdCl2 (40 mg/L) solution [8]. The EN bath
stability was increased significantly with the addition of either thiourea (TU) or maleic
acid (MA). The bath life increased from 1.3 ± 0.5 min to >264 min for an increase in
TU concentration from 0 mg/L to 2.0 mg/L and from 0.7 ± 0.1 min to 101 ± 9 min
for MA concentrations from 0 to 3000 mg/L [8]. While enhancing bath stability, stabi-
lizers may also influence deposit morphology [7,8], composition [8–11], and microstruc-
ture [8]. These changes would be expected to affect the corrosion behaviour of the EN
(Ni–P) coating.
The corrosion behavior of protective Ni–P coatings has been extensively investigated
and is considered to be dependent on many processes and features. These include the
following:

1. The process used to apply the Ni–P coating (i.e. electrolytic, electroless, and melting
and quenching processes).
2. The composition of the Ni–P coating, especially its P content [12–14].
3. The homogeneity of elemental distribution, especially P distribution in the coating sur-
face [12], and throughout the deposit [15].
4. The surface morphology of the deposit: i.e. whether it is smooth, nodular, or possesses a
cauliflower-like morphology [16].
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1779

5. The microstructure of the deposit: i.e. whether it is crystalline, microcrystalline, amor-


phous, or composed of mixed phases [17].
6. The exposure environment, especially the pH [12].
7. The presence of impurity inclusions and their distribution in the deposit; e.g., Pb or S
inclusions [12,18].
8. The nature of the substrate under the deposit [19].
9. The presence of surface defects such as microscopic pores or pits [18].

Although the corrosion behaviour of protective Ni–P deposits have been extensively
studied, only limited information is available for Ni–P deposits prepared in the presence
of the common bath stabilizers, TU and MA. In this paper, we have studied the corrosion
of such deposits over longer time periods in an aggressive chloride environment. Ni–P
coatings on both copper substrates and on magnesium AZ91D alloy substrates (with an
intermediate copper layer produced by a novel Cu immersion process) have been studied.

2. Experimental

2.1. Preparation of Ni–P deposits

Ni–P binary alloy deposits with P contents ranging from 6% to 9% (wt.%) were prepared
on a dumbbell shaped copper (99.9%) substrate in a plating bath containing 30 g/L
NiSO4 Æ 5H2O, 20 g/L NaH2PO2 Æ xH2O, 20 g/L CH3COONa, and various concentrations
of the stabilizers, thiourea (TU) or maleic acid (MA). Six different Ni–P alloy specimens
were prepared with different TU (0.5 mg/L, 0.9 mg/L, and 2.0 mg/L) and maleic acid
(180 mg/L, 1000 mg/L, and 3000 mg/L) concentrations. A reference specimen (RE) was
prepared without any stabilizer. These specimens were coded with the name of the stabilizer
followed by their concentration. For example, the EN specimen prepared in the solution
containing 0.5 mg/L of TU is coded as TU0.5. When prepared on a Mg AZ91D substrate
the EN specimen is designated as TU0.5 on Mg. During plating, the EN bath was slowly
agitated to minimize local acidity at the substrate surface. The deposition conditions
employed were a pH of 4.5, a temperature of 85(±1) C, a deposition time of 6.0 h,
and a specimen surface area to volume ratio of 7.8 cm2/L. A weight gain analysis was done
on a known surface area of specimen to determine the average thickness of the coating. To
ensure the Cu substrate or the Cu intermediate layer were not exposed during subsequent
corrosion tests, the thickness of the coating was made >50 lm, as suggested by Innes [20].
For the Mg AZ91D alloy substrate, an intermediate Cu layer was required to facilitate
deposition of the Ni–P coating. This intermediate coating was deposited using a novel Cu
immersion process in a bath containing 0.67 M CuSO45H2O and 3.3 M HF followed by a
surface sealing process.

2.2. Electrochemical measurements

The electrolyte used was a 5% NaCl solution prepared with reagent grade NaCl and
deionized water (resistivity >18 MX) from Millipore (Milli-Q academic) system. A stan-
dard three-compartment cell was used with a saturated calomel electrode (SCE) and a
2 · 3 cm2 platinum electrode as a reference and counter electrode, respectively. Before
introduction of the specimen to the corrosion cell, the solution was purged with Argon
1780 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

gas for >30 min. The specimen is only partially immersed in the electrolyte solution to
avoid crevices. Before measurement of the open circuit potential, the electrode was cathod-
ically polarized at 0.8 VSCE to remove any air-formed oxide on the surface of the Ni–P
coating. Corrosion potential (Ecorr) and intermittent electrochemical impedance spectros-
copy (EIS) measurements were then made, first in deaerated (argon) (50 ± 1 h), and then
in oxygenated (145 ± 1 h) environments.
EIS measurements were performed with a Solartron 1255 Frequency Response Ana-
lyzer coupled to a 1480 Multistat. The EIS measurements were obtained by applying a
sinusoidal perturbation of ±5 mV, and measuring the modulus of impedance and the
phase shift over the frequency range from 100 kHz to 1 mHz. The electrode potential
for each sample was held at its open circuit value during EIS measurements. Curve fitting
of the EIS data was performed using Zview software, which uses the Levenberg–Marqu-
ardt non-linear least squares method. This method is based on the complex non-linear
least squares (CNLS) program written by Macdonald [21].

2.3. Surface analysis

Surface analysis was performed before and after corrosion experiments. The surface
morphology of the Ni–P deposits was investigated using scanning electron microscopy
(Hitachi S3500N electron microscope). Energy dispersive X-ray spectrometry (IncaEnergy
400, Oxford Instrument Ltd.) was used to estimate the Ni and P contents (wt.%). The
entire 1.2 · 1.2 mm2 area was scanned to enhance the reliability of the measurements of
average P content.

3. Results and discussion

3.1. Selection of Ni–P coated specimens

The P content of the electroless deposits is shown as a function of stabilizer concentra-


tion in Fig. 1. The stabilizer concentrations used to produce specimens for corrosion test-
ing were selected so that the difference in P content was insignificant ( ±0.5 wt.%) over
the range studied. The only exception was the deposit prepared using TU2.0 that contains
1.0 wt.% less P than generally observed. Thus, with this exception, differences in the cor-
rosion behaviour are unlikely to be attributable to differences in the P content of the
deposits.

3.2. Surface layer composition (XPS)

In a previous publication [22], X-ray photoelectron spectroscopy (XPS) was used to


examine the composition of films formed on similar EN deposits after corrosion under
the same conditions as those used here. Two distinct corrosion films were observed. An
enrichment of elemental phosphorus compared to nickel was observed on the surface of
EN deposits produced with either no stabilizer or maleic acid. By contrast, extensive sur-
face oxidation was observed on EN deposits prepared with TU. Trace impurities of sulfur
were blamed for the extensive build-up of corrosion products on EN deposits formed with
TU, whereas a P enriched chemical passivation layer was formed on deposits prepared
without any stabilizer and with MA.
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1781

10

8
P Content, wt. %

Ref
4 MA
TU

0
0 1 2 500 1000 1500 2000 2500 3000
Concentration, mg/L

Fig. 1. Phosphorus content in a Ni–P binary alloy produced by EN deposition with different types and amounts
of the stabilizer.

3.3. Surface morphology and composition (SEM/EDS)

‘As deposited’ surfaces of the EN specimens are shown in Fig. 2(A1)–(G1). The surfaces
can be categorized into two types; (a) surfaces with particle boundaries and (b) smooth
surfaces. Deposits formed in an unstabilized bath or in a bath containing larger than
1 g/L MA exhibited smooth surfaces with a number of small defects (Fig. 2(A1), (F1)
and (G1)). For surfaces with particle boundaries, slightly different structures were
observed depending on the type of stabilizer and its concentration. The MA180 specimen
had regular small nodular features 100 nm in diameter (Fig. 2(E1)). The TU0.5 specimen
showed non-uniformly ordered particle boundaries (Fig. 2(B1)) while very clear uniformly
ordered boundaries were produced with TU0.9 (Fig. 2(C1)). A cauliflower like surface
with many boundaries was observed on the TU2.0 specimen (Fig. 2(D1)).
After immersion in 5% NaCl solution for 195 h (Ar-purged (50 h) and O2-purged
(145 h)), two distinctly different corrosion morphologies were observed (Fig. 2(A2)–
(G2)). For specimens with particle boundaries, preferential attack occurred along the
boundaries (Fig. 2(B2)–(E2)), and the severity of attack increased with an increase in
the concentration of TU. The blurred SEM micrograph (Fig. 2(D2)) observed for the
TU2.0 specimen may indicate the presence of a thick corrosion/oxide layer on this surface.
By contrast, evenly distributed small pits with a high number density were formed on the
surfaces of smooth deposits (Fig. 2(A2), (F2), and (G2)).
These differences in morphology of corrosion attack were correlated with the two
distinctly different surface layer compositions determined by XPS [22]. EDS analysis con-
firmed the higher O content and lower P content (compared to the bulk P content of
6 wt.%) obtained on the corroded TU2.0 specimen prepared on Mg AZ91D substrate
(Fig. 3). For the corroded RE and MA3k specimens prepared on a Mg AZ91D substrate,
the P and Ni contents were comparable to the bulk values (Fig. 3). In addition, an SEM
1782 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

Fig. 2. The SEM micrograms of Ni–P deposits surface prepared with various concentrations of TU and MA: (A)
no stabilizer; (B) 0.5 mg/L TU; (C) 0.9 mg/L TU; (D) 2.0 mg/L TU; (E) 180 mg/L MA; (F)1.0 g/L MA; (G) 3.0
g/L MA. The SEM micrograms of each specimen were obtained (1) before and (2) after corrosion experiments
in 5% NaCl purged with Ar (48 h) and O2 (145 h).
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1783

Fig. 2 (continued)

investigation of the specimens (Fig. 4(A)–(E)) after the 50 h Ar purged period indicated
that the pitting and intergranular attack initiated during the Ar-purged period.

3.4. Corrosion potential (Ecorr) measurements and electrochemical impedance spectroscopy


(EIS)

3.4.1. General corrosion behaviour and equivalent circuit models


The corrosion potentials (Ecorr) of different specimens were monitored in deaerated (Ar)
(50 ± 1 h) and then oxygenated (145 ± 1 h) 5% NaCl solution (Fig. 5). In addition, EIS
measurements were acquired periodically for every specimen (Fig. 6). For all specimens,
the absolute impedances at low frequencies (103–102 Hz) were much higher for
0 2 4 6 8 10

Ar-purged conditions than for O2-purged conditions (Fig. 6(A2)–(G2)). The increase in
the Ecorr (Fig. 5) observed on switching from Ar to O2-purged conditions, can be attrib-
uted to the increased polarization of the corrosion process by oxygen reduction leading
to an increase in corrosion rate. In addition, the introduction of oxygen leads to less noise
in the Ecorr. The noise amplitude was ±5 mV for Ar-purged conditions, but only ±0.1 mV
for the O2-purged condition. The Ecorr of the uncoated Mg AZ91D alloy substrate was
extremely negative (1.5 VSCE, Fig. 5), consistent with its instability in water. In addition,
the introduction of oxygen did not change Ecorr, indicating that the oxygen reduction reac-
tion does not influence the already rapid corrosion of the Mg AZ91D alloy (Fig. 5) in 5%
NaCl solution.
Inspection of the EIS spectra suggests three different equivalent circuit models are
required to fit the results (Fig. 7). A one time constant (s) model (Fig. 7(a)) is appropriate
for smooth deposits (RE, MA1k or MA3k) corroding under Ar-purged conditions
(Fig. 6(A), (F), and (G); plots in black). Such a circuit suggests a uniform corrosion
process on smooth deposits. The development of small pits in the latter stages of the
Ar-purged period required the fitting of the spectra with a 2 s-circuit (Fig. 7(b)), consistent
with a corrosion process occurring predominantly at the base of pits. A 2 s-circuit model
was also required to fit the spectra measured on specimens with particle boundaries when
preferential corrosion along particle boundaries occurred (Fig. 6(B)–(E)). Lastly, a 2 s cir-
cuit with an added Warburg circuit element (Fig. 7(c)) was required to fit the spectra gen-
erated from specimens prepared with >0.9 mg/L TU after prolonged exposure to
oxygenated solution (Fig. 6(C) and (D); plot in olive). Such a model is consistent with a
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1785

Fig. 4. The SEM micrograms of Ni–P deposits prepared with various concentrations of TU and A: (A) no
stabilizer; (B) 0.5 mg/L TU; (C) 2.0 mg/L TU; (D) 1.0 g/L MA; (E) 3.0 g/L MA. The SEM micrograms were
obtained after immersion in 5% NaCl purged with Ar (48 h).

corrosion process occurring preferentially at the base of deep particle boundary channels
in which O2 transport is partially rate controlling.

3.4.2. Ar-purged condition


For the deposit prepared in an unstabilized bath, Ecorr rapidly established a value of
400 mV to 425 mV followed by a slow rise (Fig. 5). A deposit grown in the presence
of a low TU concentration (TU0.5) exhibited similar behaviour, but as the TU concentra-
tion was increased, an Ecorr value in the range 450 mV to 500 mV was established more
slowly and no subsequent gradual increase was observed (Fig. 5). The MA180 specimen
behaved similarly to these last specimens. Inspection of Fig. 2 shows this latter behaviour
1786 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

Fig. 5. Corrosion potential of deposits formed in differently stabilized baths exposed to 5% NaCl solution.

is consistently observed with deposits comprising fine and nodular features (Fig. 2(C1),
(D1), and (E1)). The Ecorr behaviour in the MA1k specimen was similar to that observed
for the RE specimen. For the MA3k specimen, Ecorr was initially very low but eventually
the Ecorr rose as if undergoing a transition (Fig. 5).
A comparison of values after the Ar-purging period shows Ecorr shifted to less noble
values with an increase in TU concentration from 0 mg/L to 2.0 mg/L, Fig. 8. By contrast,
Ecorr values for the MA1k and MA3k specimens (Fig. 8) were closer to the Ecorr for the RE
specimen. The onset of pitting during the Ar-purging period forced the use of a 2 s-circuit
to fit the impedance results. Fig. 8 also shows the total surface resistances determined from
the low frequency limit of the impedance spectra. The resistances (Rt) clearly correlate
with the Ecorr values, with specimens with Ecorr < 0.45 V possessing lower resistance than
those with Ecorr P 0.40 V. This indicates that the positive shift in Ecorr under Ar-purged
conditions leads to an enhanced corrosion resistance, suggesting the formation of a corro-
sion resistant surface film/layer. The increases in Ecorr and total resistance for the RE and
the MA3k specimens can be attributed to the gradual inhibition of the anodic reaction due
 0
to the build-up of a P-enriched layer containing PO3 4 , H2 PO2 , and P , as indicated by
XPS [22]. It is likely this build-up is due to the selective dissolution of Ni at the beginning
of the immersion period [17].
The formation of a hypophosphite film has also been reported on both electroless Ni–P
(12.8% and 8.0%) alloys under natural corrosion conditions in 3.5% NaCl at pH of 10 [4]
and on electroless Ni–P (13.2%) deposits in a 0.2 N HCl solution [23]. The oxidation of
phosphorus to P1+ was considered to occur as follows [24]:
P þ 2H2 O ¼ H2 PO þ
2 þ 2H þ e

ð1Þ
The P1+ species can then be further oxidized to a P5+ species [24]
H2 PO  þ
2 þ 2H2 O ¼ H2 PO4 þ 4H þ 3e

ð2Þ
2+
and stabilized on the metal surface by an Ni cation.
By comparison, XPS analysis showed a low P/Ni ratio on the TU2.0 specimen [22], and
the low total resistance indicated that the anodic reaction (i.e. Ni ! Ni2+ + 2e) was not
inhibited in this case. The TU0.5 specimen exhibits a similar high resistance to the unsta-
bilized system, Fig. 8, but in this case XPS showed the inhibiting surface possessed a high
O/Ni ratio and only a small P/Ni ratio. The high resolution Ni 2p3/2 peak indicated most
of nickel species on the surface were bound to hydroxide [22]. The MA1k specimen exhib-
ited a particularly high Rt value suggesting the presence of an air-formed oxide, incom-
pletely removed by the cathodic cleaning step.

-80

-60
Phase Angle, °

-40

-20

0 0 0
0.01 40 0.01 50
1 80 1 100
120
100 100
Freq
uenc 160 e, h 150
y, Hz 10000 200 Tim 10000
200
-80

-60
Phase Angle, °

-40

-20
0
50
0
0.01 100

h
e,
1

m
150

Ti
100
Freq
u ency 10000
, Hz 200
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1789

Fig. 7. Equivalent circuit models for electroless Ni–P deposits exposed in 5% NaCl solution. (a) Uniform
corrosion film on the Ni–P surface, (b) corrosion film with pores of low aspect ratio and (c) corrosion film with
pores of high aspect ratio.

-0.33 Rt - TU
Rt - MA 40000
Rt - Ref
Corrosion Potential ,VSCE

-0.36 35000
Total Resistance, Ωcm2

30000
-0.39
25000

-0.42 20000

15000
-0.45
10000
Ecorr - TU
-0.48 Ecorr - MA 5000
Ecorr - Ref

0 12 1000 2000 3000


Concentration, mg/L

Fig. 8. Corrosion potential and total resistance (from EIS measurements) for deposits produced with different
amounts of stabilizer: MA – maleic acid; TU – thiourea; Ref – no stabilizer; 50 h immersion in Ar-purged 5%
NaCl.

3.4.3. O2-purged condition


The onset of O2 purging resulted in a shift of Ecorr to more positive values for all the
specimens (Fig. 5), as expected. After the initial increase, Ecorr varied for individual spec-
imens. Ecorr for the MA3.0, MA1.0 and RE specimens rapidly stabilized (Fig. 5), while the
1790 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

TU specimens relaxed to lower values (Fig. 5). This relaxation was most marked for the
TU0.9, TU2.0 and MA180 specimens, and suggests an accelerated corrosion reaction,
which is consistent with the significantly lower Rt values obtained for these specimens,
Fig. 9.
The behaviour of the three specimens prepared in TU makes an instructive series. The
relaxation in Ecorr is much less marked for TU0.5 compared to TU0.9 and TU2.0 (Fig. 5),
and so is the decrease of Rt with time (Fig. 9). Inspection of Fig. 2(B)–(D) shows that the
extent of intergranular attack is also lower for the TU0.5 specimen. However, the decrease

100000

10000
2
Rt, Ωcm

Rt - RE
Rt - TU0.5
1000 Rt - TU0.9
Rt - TU2.0
Rt - RE on Mg
Rt - TU2.0 on Mg

100
0 50 100 150 200
Immersion time, h

100000

10000
2
Rt, Ωcm

1000

Rt - RE
Rt - MA180
Rt - MA1K
Rt - MA3k
Rt - RE on Mg
Rt - 3kMA on Mg

100
0 50 100 150 200
Immersion time, h

Fig. 9. Total resistances from EIS measurements of the deposits produced from differently stabilized baths after
various periods of corrosion in 5% NaCl; 0–50 ± 1 h Ar-purged; >50 h oxygenated: (a) RE and TU specimens;
(b) RE and MA specimens.
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1791

of Rt with time was much greater for the TU0.9 specimen than that for the TU2.0 speci-
men. This may be attributed to the fact that the well defined particle boundaries present on
the TU0.9 specimen (Fig. 2(C)) are prone to the intergranular attack. Also, the decrease in
Rt for TU2.0 on Mg in the oxygenated solution was much less than that prepared on the
copper substrate (Fig. 9(a)). This difference might be attributed to the high roughness of
the Mg alloy surface and the electrochemically produced Cu intermediate layer, leading to
the production of a deposit with less well defined particle boundaries. By comparison,
noticeable differences were not observed for RE and MA specimens (Fig. 9(b)).
The EIS spectra for the TU specimens evolve considerably as Ecorr relaxes during the
O2-purging period. This is best appreciated from the phase angle plots in Fig. 6(B1)–
(D1). A second time constant evolves with time at low frequencies and for the TU0.9
and TU2.0 specimens, a Warburg impedance is required to accurately fit the spectra. This
suggests partial control of the corrosion by O2 transport within the deepening intergran-
ular sites (Fig. 6(C1) and (D1)). By contrast, the RE, MA1k, and MA3k specimens do not
obviously develop a second time constant during the O2-purging period, although MA180
shows a slight tendency to do so. This is consistent with the SEM micrograms obtained at
the end of the Ar-purged (Fig. 4) and the O2-purged (Fig. 2) periods which both show a
generally uniform distribution of numerous pits. The slight differences might be due to the
presence of the patches of surface film on the surface as indicated at the end of the Ar-
purged period.
XPS analyses showed these differences coincided with the formation of two distinctly
different corrosion films/layers after the oxygenated period [22]. Higher P/Ni ratios
(1–2) were observed on smooth specimens (RE, MA1k, and MA3k) than on the nodular
TU specimens (<0.5). The P/Ni ratios after O2-purging were much higher than those mea-
sured on sputter-cleaned surfaces for the RE, MA1k, and MA3k specimens. By contrast
the TU0.5 and TU2.0 specimens had similar P/Ni ratio as those of the sputter-cleaned sur-
face [22]. In general, corrosion of smooth deposits lead to an increase in P/Ni ratio while
those specimens exhibiting intergranular corrosion experienced little change in this ratio
[22].

The composition of corroded smooth surfaces was mainly P0 along with PO3 4 , H2 PO2
and only a minor amount of nickel oxide or hydroxide. By comparison, the intergranularly
attacked TU specimens had a high O content mostly in the form of nickel hydroxide. The
nature of these surface layers influenced the corrosion rates of the specimens. The P
enriched surface layers of the RE and MA specimens lead to the higher Rt values observed,
relative to those for the intergranularly attacked TU2.0 specimens (Fig. 9). However, a
thick generally uniform hydroxide layer also provides some corrosion resistance as in
the case of the TU0.5 specimen and the TU2.0 specimen prepared on the Mg AZ91D sub-
strate when the particle boundaries are not as well defined or subsequently corroded
(Fig. 2(B1) and Fig. 9(a)).
The use of the two time constant circuit shown in Fig. 7(b) allows the determination of
the resistance associated with pits and intergranularly corroded sites. The evenly distrib-
uted preformed defects resulted in pit formation on the surface of the RE deposit,
Fig. 4(A), and the high pore resistance (Rpo > 10,000 X cm2) (Fig. 10(a)) suggests a large
aspect ratio. The pore resistance increased slightly during the Ar-purged period as the
P-rich surface layer started to form on the surface. However, for more aggressive O2-purg-
ing conditions, Rpo initially decreased by an order of a magnitude, suggesting opening up
of the pits. As the pits propagated, the aspect ratio became larger resulting in the observed
1792 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

10000

Rpo - RE
Rpo - TU0.5
Rpo - TU0.9
Rpo - TU2.0
Rpo - RE on Mg
Rpo - TU2.0 on Mg
1000
2
Rpo, Ωcm

100

10

0 20 40 60 80 100 120 140 160 180 200


Time, h

10000
2
Rpo, Ωcm

1000

100
Rpo - RE
Rpo - MA180
Rpo - MA1k
Rpo - MA3k
Rpo - RE on Mg
Rpo - MA3k on Mg
10
0 20 40 60 80 100 120 140 160 180 200
Time, h

Fig. 10. Pore resistances of the deposits produced from differently stabilized baths after various periods of
corrosion in 5% NaCl; 0–50 ± 1 h Ar-purged; >50 h oxygenated: (a) RE and TU specimens; (b) RE and MA
specimens.

recovery in Rpo (Fig. 10(a)). The Rpo value for the RE on Mg specimen did not show the
decrease and subsequent increase, but the Rpo was similar to that of the RE specimen after
long immersion in the oxygenated solution.
For the TU specimens (including the TU2.0 on Mg specimen), a similar drop in Rpo
occurred on switching from Ar to O2, except for the TU0.5 specimen. This can be attrib-
uted to the enlarged corrosion surface area with the introduction of O2. For the TU0.5
specimen, the particle boundaries are not clearly defined (Fig. 2(B2)). Also, the high sur-
face oxide content prevented the corrosion from penetrating to the particle boundaries.
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1793

Thus, the Rpo was much smaller compared to the others (Fig. 10(a)). After the introduc-
tion of oxygen, the corrosion reaction started to occur along the particle boundaries as in
the case of the other TU specimens, and the Rpo values comparable to those of the other
TU specimens were observed through the long oxygen purged period (Fig. 10(a)).
For the MA specimens, two different behaviours were observed. A drop in Rpo was
observed for the MA180 and MA3k specimens but not for the MA1k and the MA3k
on Mg specimens on switching from Ar to O2 (Fig. 10(b)). This drop can be attributed
to the increase in the number densities and/or opening up of pits, in the more aggressive
corrosion environments, Fig. 2(E) and (G). For the MA1k specimen, the high Rpo value
(20,000 X cm2) under Ar-purging did not change on switching to O2. This is consistent
with the presence of very few pits on this specimen, Fig. 2(F2). XPS results indicated this
specimen is protected by a thick oxide layer under Ar-purged conditions and a strong sur-
face enrichment of P after O2-purging.
Steady and high film resistances (Rfilm  10,000 X cm2) were observed on all MA
specimens (Fig. 11(b)) while significantly lower film resistances were observed on most
TU specimens (Fig. 11(a)). The high values of Rfilm can be attributed to passivation by
P-enriched surface layers as indicated by XPS [22]. The film resistances for the TU samples
decreased as intergranular corrosion became severe with oxygen purging. The noticeably
lower Rfilmobserved for the TU0.9 specimen (compared to the TU2.0 specimen) can be
attributed to the much larger particle boundaries for TU0.9 compared to TU2.0. This
results in the accumulation of a thinner layer of corrosion products on the particles.

3.5. Significant factors governing the corrosion characteristics of the protective Ni–P coating

3.5.1. Microstructure vs. microscopic defects


The microstructure of the deposit as characterized previously [8] indicated that the EN
deposits are polycrystalline with mixed amorphous and nanocrystalline phases. The depos-
its with a higher degree of amorphicity or smaller grain sizes (i.e. the MA3k specimen)
exhibited slightly poorer corrosion characteristics than those with ordered nanocrystals
(i.e. the RE specimen). This result is not consistent with the literature; amorphous Ni–P
coatings passivate in acidic and neutral solution, but the presence of crystalline structures
increased corrosion rates dramatically [12,25]. This conclusion was drawn knowing that
the deposit only differs in microstructure [12] by applying heat treatment, or microstruc-
ture and P content [25] by adjusting electroplating bath composition. In the present cor-
rosion study, for a deposit with smooth morphology, the dominant factor for the
corrosion performance appeared to be the pre-formed microscopic defects/preferentially
attacked sites since the morphologies observed after etching and corrosion were similar;
i.e. both exhibited evenly distributed small pit formation (Fig. 2(A2) and Fig. 12).

3.5.2. Morphology
Specimens with a smooth morphology (i.e. the RE, MA1k and MA3k specimens) were
much more corrosion resistant than those with a nodular morphology, except for the
MA180 specimen. The variation in P content between specimens was insignificant and,
therefore, not a factor in determining the differences in corrosion behaviour. However,
the corrosion resistance decreased at the high TU concentration, consistent with published
observations [18]. Our study of deposit microstructure [8] showed TU0.9 to be more amor-
phous than the RE specimen. Clearly amorphicity is not critically important in determin-
1794 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

100000
Rfilm - RE
Rfilm - TU0.5
Rfilm - TU0.9
Rfilm - TU2.0
Rfilm - RE on Mg
Rfilm - TU2.0 on Mg

10000
2
Rfilm, Ωcm

1000

100
0 20 40 60 80 100 120 140 160 180 200

Time, h

100000

10000
2
Rfilm, Ωcm

1000

Rfilm - RE
Rfilm - MA180
Rfilm - MA1k
Rfilm - MA3k
Rfilm - RE on Mg
Rfilm - MA3k on Mg
100
0 20 40 60 80 100 120 140 160 180 200
Time, h

Fig. 11. Film resistances of the deposits produced from differently stabilized baths after various periods of
corrosion in 5% NaCl; 0–50 ± 1 h Ar-purged; >50 h oxygenated: (a) RE and TU specimens; (b) RE and MA
specimens.

ing corrosion behaviour since the corrosion rate for TU0.9 is greater than that of the RE
specimen. It appears that the presence and density of particle boundaries are more impor-
tant. As the particle boundaries become more clearly defined (Fig. 2(A2), (B2), and (C2)),
the corrosion resistance decreases, due to intergranular corrosion along these boundaries.
However, the MA180 specimen dose not fit this pattern, since despite the presence of fine
particle boundaries, the total resistance value was high and compatible with that of the
MA3k specimen.
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1795

Fig. 12. SEM microgram of a polished Ni–P deposit surface after etching with dilute HNO3 (5%) for 5 s.

3.5.3. Surface composition and impurity


The higher corrosion resistance observed for the RE specimen and the MA specimen
can be attributed to the formation of a P-enriched layer shown to be present by XPS
and mostly in the P0 valence state. Such a layer has been claimed to be a chemical passiv-
ation layer against corrosion [12,17,26], the accumulation of elemental P on the alloy sur-
face acting as a barrier against dissolution. On the contrary, the more rapidly corroding
TU specimens possessed a low P/Ni ratio but a high O/Ni ratio (>2.0).
In addition, trace amounts of the sulfur were detected by the XPS on TU specimen
before and after corrosion [9–11,22] and its presence was confirmed by high resolution
TOF-SIMS [22]. The S peaks were detected at a binding energy of 167–168 eV corre-
sponding to an oxidized form of sulfur, most likely SO2
4 [27]. This form of sulfur was pro-
posed by Eklund as the dominant dissolved sulfur species originating from the corrosion
of sulfide inclusions in stainless steel (SS) [28]. The acidity formed via this corrosion
process
þ
MnS þ 4H2 O ! Mn2þ þ SO2
4 þ 8H þ 8e

ð3Þ
was considered to activate the surface.
The stability domain of adsorbed sulfur extends beyond the usually predicted range of
stability of metal sulfides; i.e., adsorbed sulfur can exist under conditions in which no bulk
sulfide is stable [29]. The presence of the adsorbed sulfur has been shown to increase the
dissolution rate of nickel based alloys, a weakening the metal–metal bond induced by
adsorbed sulfur. This lowers the activation energy barrier for the passage of metal atoms
from the surface to the solution [29]. Furthermore, this influence can be localized if the
sulfur is adsorbed on specific sites, where it can be more tightly bonded [29]. For TU spec-
imens, sulfur adsorption may be concentrated on the particle boundaries (Fig. 2(A2)–
(D2)), where the higher adsorption density of TU during the EN deposition is expected
1796 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

due to the higher diffusive flux. Adsorbed sulfur is also thought to retard the growth of
passive films by blocking the sites utilized in oxide growth by OH adsorption. Anodic
segregation of sulfur caused by the selective dissolution of nickel could lead to large sur-
face concentrations of adsorbed sulphur on the surface [29]. These sites could be contin-
uously exposed to further corrosion. Such processes could explain the large differences
in total resistance between surfaces with particle boundaries (TU0.9, TU2.0 and MA180).
The significantly lower resistance values for TU0.9 and TU2.0 compared to MA180 could
be attributed to the presence of S on particle boundaries.
In addition to the influence of S impurities, local acidity could be enhanced at particle
boundaries by Ni2+ hydrolysis to NiOH+, Ni(OH)2, and Ni2OH3+ [30]

O2
OH-

Ni Ni2+ 2-
SO4
Sulfide oxidation
S2- + 4H2O
8H+
Sads H+
Ni2++ H2O Ni dissolution enhanced by Sads
Ni
Ni(OH)+
Ni(OH)2 Ni2+ hydrolysis

O2
Sads 8H+ OH
P + 4H2O P oxidation
PO43-
Ni Ni2+

O2

Ni-P
5 % NaCl Solution

-
O2 OH Ni dissolution is 5%
Ni2+ blocked
P-rich
P
Ni Ni-P
Ni-P
Fig. 13. Schematic models for (a) intergranular corrosion of Ni–P deposits prepared with TU and (b) chemical
passivation of Ni–P deposits prepared with no stabilizer or with MA during corrosion in oxygenated neutral 5%
NaCl.
W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798 1797

Ni2þ þ H2 O ¼ NiOHþ þ Hþ ð4Þ


2þ 3þ þ
2Ni þ H2 O ¼ Ni2 OH þH ð5Þ
The schematic in Fig. 13(a) attempts to summarize the reactions described as occurring
on an EN surface prepared with TU present. Ni2+ hydrolysis, combined with S and P oxi-
dation, lowers the pH at the particle boundaries and increases the solubility of various
Ni2+ hydrolysis products such as NiOH+ and Ni2OH3+ [30]. Thus, these Ni2+ hydrolysis
products readily diffuse out of particle boundaries and precipitate to form a thick oxide
layer on the outer surfaces as shown by XPS analyses. Due to the acidity within the cor-
roding particle boundaries, oxide/hydroxide formation is prevented and corrosion conti-
nues. On the contrary, in the absence of the particle boundaries and sulfur impurities,
the general preferential dissolution of nickel produces a P-rich chemically passive layer
which hinders further dissolution of nickel as illustrated schematically in Fig. 13(b).

4. Conclusions

The effects of the EN bath stabilizers, TU and MA, on the corrosion of the EN coatings
on copper and magnesium substrates (with an intermediate copper layer produced by a
novel Cu immersion process) were investigated in an aggressive oxygenated chloride
environment.
The dissolution or the corrosion attack of the AZ91D Mg alloy and Cu substrates were
not observed throughout the corrosion experiments, with corrosion limited to the external
EN layer.
Two different types of corrosion morphology were observed. Specimens prepared with
TU, and the MA180 specimen, showed preferential attack along particle boundaries while
those prepared with no stabilizer and MA showed a pitting-type corrosion on the smooth
surfaces. Higher corrosion resistances were observed for smooth deposits, which is attrib-
uted to the formation of a P-enriched layer on the surface. The poorer corrosion perfor-
mance of EN specimens prepared with TU is attributed to the presence of the particle
boundaries and the trace impurities adsorbed on, or included in, them.
It was clearly shown that the corrosion behaviour of the EN specimens studied was not
governed by the microstructural aspects of the EN deposit but was predominantly influ-
enced by morphological and surface compositional aspects (i.e. surface impurities).

Acknowledgements

This research is funded by the Natural Science and Engineering Research Council of
Canada (NSERC). The authors gratefully acknowledge the general technical help of John
Nagata (IMTI-NRC) and help of Mike Meinert (IMTI-NRC) and Brad Kobe (SSW) in
the acquisition of SEM micrographs. The authors would like to thank Dr. Zack Qin
(UWO) for the design of the corrosion cell and Dr. Jamie Noel (UWO) for the help in
showing problems related to electrochemical instrumentation.

References

[1] R. Ellmers, D. Maguire, Global View Magnesium: Yesterday, Today, Tomorrow, 1993, p. 28.
[2] J.E. Gray, B.L. Luan, Journal of Alloys and Compounds 336 (2002) 88.
1798 W.-J. Cheong et al. / Corrosion Science 49 (2007) 1777–1798

[3] H. Huo, Y. Li, F. Wang, Corrosion Science 46 (2004) 1467.


[4] P.-H. Lo, W.-T. Tsai, J.-T. Lee, M.-P. Hung, Surface and Coating Technology 67 (1994) 27.
[5] G.O. Mallory, J.B. Hajdu, Electroless Plating, Noyes Publications/William Andrew Publishing, LLC, New
York, 1996, pp. 35–55.
[6] R. George, S. Venkatachalam, K.N. Ninan, British Corrosion Journal 37 (2002) 37.
[7] E. Lanzoni, C. Martini, R. Ruggeri, R. Bertoncello, A. Glisenti, in: EUROCORR ’96, 1997, p. 232.
[8] W.J. Cheong, B.L. Luan, D.W. Shoesmith, Applied Surface Science 229 (2004) 282.
[9] L. Das, D.T. Chin, Plating and Surface Finishing 83 (1996) 55.
[10] I.V. Petukhov, M.G. Shcherban, Protection of Metals 35 (1999) 566.
[11] J.K. Sallo, J.S. Sallo, Journal of the Electrochemical Society 112 (1965) 1201.
[12] Z. Longfei, L. Shoufu, L. Pengxing, Surface and Coating Technology 36 (1988) 455.
[13] R.N. Duncan, T.L. Arney, Plating and Surface Finishing 76 (1989) 60.
[14] J. Flis, D.J. Duquette, Corrosion 41 (1985) 700.
[15] G. Salvago, G. Fumagalli, Metal Finishing 85 (1987) 31.
[16] G. Salvago, D. Sinigaglia, G. Fumagalli, D. Continenza, G. Taccani, in: Proceedings of INTERFINISH 80,
1980.
[17] A. Krolikowski, European Federation of Corrosion Publication 2 (1993) 119.
[18] I.V. Petukhov, M.G. Shcherban, N.E. Skryabina, L.N. Malinina, Protection of Metals 38 (2002) 370.
[19] G. Salvago, G. Fumagalli, in: 10th International Congress on Metallic Corrosion, 1987.
[20] W.P. Innes, Electroplating and Electroless Plating on Magnesium and Magnesium Alloys, Wiley-
Interscience, New York, 1974, p. 601.
[21] J.R. Macdonald, in: J.R. Macdonald (Ed.), Impedance Spectroscopy Emphasizing Solid Materials and
Systems, John Wiley & Sons, New York, 1987, pp. 179–180.
[22] W.J. Cheong, B.L. Luan, N.S. McIntyre, D.W. Shoesmith, Surface and Interface Analysis, submitted for
publication.
[23] R.B. Diegle, N.R. Sorensen, C.R. Clayton, M.A. Helfand, Y.C. Yu, Journal of Electrochemical Society 135
(1988) 1085.
[24] M. Pourbaix, in: J.V. Muylder, M. Pourvaix (Eds.), Atlas of Electrochemical Equilibria in Aqueous
Solutions, Pergamon Press Ltd., Oxford, 1966, pp. 504–515.
[25] A. Krolikowski, B. Pokrywa, Metalurgia i Odlewnictwo 16 (1990) 111.
[26] H. Habazaki, S.-Q. Ding, A. Kawashima, K. Asami, K. Hashimoto, Corrosion Science 29 (1989) 1319.
[27] K.-P. Han, J.-L. Fang, International Journal of Chemical Kinetics 28 (1996) 259.
[28] G.S. Eklund, Journal of Electrochemical Society 121 (1974) 467.
[29] P. Marcus, Sulfur-assisted corrosion mechanisms and the role of alloyed elements, in: P. Marcus, J. Oudar
(Eds.), Corrosion Mechanisms in Theory and Practice, Marcel Dekker, Inc., New York, 1995, pp. 239–263.
[30] C.F. Baes, R.E. Mesmer, The Hydrolysis of Cations, John Wiley & Son, New York, 1976, pp. 241–247.

You might also like