Modeling, optimization, and comparative analysis of trivalent chromium electrodeposition from aqueous glycine and formic acid baths

Article abstract of DOI:10.1149/1.1933688

Trivalent chromium electrodeposition has been studied as an alternative to hazardous hexavalent chromium electroplating. Although limited by inability to deposit thick coatings with acceptable quality, it is useful for decorative chromium electroplating. In the present work, we illustrate the application of statistical design of experiments (DOE) in empirical modeling and optimization of electrodeposition of trivalent chromium from baths of chromium complexed with glycine. DOE shows that variations in current efficiency with glycine and chromium chloride concentrations can be explained by a quadratic model. The composition of the bath was optimized with respect to current efficiency to obtain faster deposition. We compare the effects of process parameters (pH, temperature, current density, and pulsed current) on current efficiency and deposit characteristics for both glycine and formic acid-containing baths. Experiments reveal that whether pulsed current increases or decreases current efficiency is determined by how current density influences current efficiency, and the range of lower and upper current density levels.

Full text of DOI:10.1149/1.1933688

C504
Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
0
013-4651/2005/152͑7͒/C504/9/$7.00 © The Electrochemical Society, Inc.
Modeling, Optimization, and Comparative Analysis of Trivalent
Chromium Electrodeposition from Aqueous Glycine and
Formic Acid Baths
a
b, ,z
Anil Baral and Robert Engelken *
a
Environmental Science Program and College of Engineering, Arkansas State University (Jonesboro), State
University, Arkansas 72467, USA
b
College of Engineering, Arkansas State University (Jonesboro), State University, Arkansas 72467, USA
Trivalent chromium electrodeposition has been studied as an alternative to hazardous hexavalent chromium electroplating. Al-
though limited by inability to deposit thick coatings with acceptable quality, it is useful for decorative chromium electroplating. In
the present work, we illustrate the application of statistical design of experiments ͑DOE͒ in empirical modeling and optimization
of electrodeposition of trivalent chromium from baths of chromium complexed with glycine. DOE shows that variations in current
efficiency with glycine and chromium chloride concentrations can be explained by a quadratic model. The composition of the bath
was optimized with respect to current efficiency to obtain faster deposition. We compare the effects of process parameters ͑pH,
temperature, current density, and pulsed current͒ on current efficiency and deposit characteristics for both glycine and formic
acid-containing baths. Experiments reveal that whether pulsed current increases or decreases current efficiency is determined by
how current density influences current efficiency, and the range of lower and upper current density levels.
©
2005 The Electrochemical Society. ͓DOI: 10.1149/1.1933688͔ All rights reserved.
Manuscript submitted August 11, 2004; revised manuscript received February 15, 2005. Available electronically June 17, 2005.
Attempts to deposit chromium from aqueous trivalent chromium
tials of the individual stages. Not all Cr͑II͒ ions produced are dis-
charged immediately because the total limiting current for chro-
mium deposition ͓Cr͑III͒ → Cr͑0͔͒ is about one third of the total
solutions can be traced to the early 1900s. Since 1933, many patents
and papers have been issued for deposition of chromium alone and
as an alloy with elements like nickel, manganese, iron, cobalt, and
11
limiting current of Cr͑II͒ formation. This suggests that there are
unreacted Cr͑II͒ ions adsorbed on the cathode or in the double layer.
In other words, electroreduction of Cr͑II͒ is a rate-limiting stage in
chromium electrodeposition.
1
-7
molybdenum. Trivalent chromium plating was commercially in-
troduced in the U.S. in 1976 and, by 1981, there were more than 70
8
users of the process. Trivalent chromium electroplating was first
used only for decorative purposes due to thickness limitations. Dur-
ing the 1980s and 1990s, studies were performed on electrodeposi-
tion of chromium from trivalent chromium baths, most of which
contained carboxylic acids. Besides aliphatic carboxylic acids,
It is difficult to electroplate trivalent chromium from aqueous
solutions because aqua-chromium complexes have a low thermody-
namic stability and a high kinetic inertness. This is why trivalent
chromium is usually complexed with amino acids, amides, formates,
or oxalates. Hydrolysis and olation occur close to the cathode and
9
10
thiocyanate and hypophosphite were also studied as complexants.
10
14
Hwang obtained deposits that were bright, attractive, and white,
although not exactly the same as the blue-white color of conven-
tional hexavalent chromium deposits, using baths containing hypo-
phosphite. Baths containing an aluminum salt as a buffer have
inhibit chromium deposition. In the presence of complexants,
complexation reactions compete with olation reactions and mini-
mize formation of olation products. Glycine is a strong complexant
that can compete with olation reactions more effectively than most
11
12
gained attention. Edigaryan and Polukarov found that aluminum
sulfate was a better buffer than boric acid and yielded hard but
ductile layers with a thickness up to 50 ␮m. Amino acids such as
other ligands.
Olation reaction
1
2
glycine as complexing agents yielded good quality deposits.
In acidic aqueous baths, electrodeposition of trivalent chromium
3
+
2+
͓
Cr L͑H O͒ ͔ → ͓Cr L͑H O͒ OH͔ + H+
͓4͔
͓5͔
2
4
2
3
11
occurs in two steps
0
E =−0.407 V
Cr͑III͒ + e − ——→ Cr͑II͒
͓1͔
͓2͔
0
E =−0.913 V
Cr͑II͒ + 2e − ——→ Cr͑0͒
͓
6͔
The overall reaction can be shown as
0
where L = ligand ͑e.g., H2O, glycine͒.
E =−0.744 V
Traditionally, optimization of a property ͑for example thickness,
current efficiency, or appearance͒ in electrochemistry has been car-
ried out by varying one variable at a time while keeping the rest of
the variables constant. The problem with this approach is that it is
difficult to arrive at optimal solutions of multivariable problems by a
bird’s-eye view of two-dimensional graphs alone. Statistical design
of experiments ͑DOE͒ is one valuable approach to empirical mod-
eling and optimization of the parameter of interest in a multivariable
Cr͑III͒ + 3e − ——→ Cr͑0͒
͓3͔
For potentials less negative than −0.92 V, only hydrogen evolu-
11
tion and reduction of Cr͑III͒ to Cr͑II͒ occurs on the cathode. In the
range of −0.92 to −0.950 V, reduction of Cr͑II͒ to Cr͑0͒ occurs,
depending upon concentration of Cr͑II͒. At cathodic potentials more
negative than −1.0 V, the rates of the reactions of Cr͑III͒ to Cr͑II͒,
and Cr͑II͒ to Cr͑0͒ become equal and the concentration of Cr͑II͒
remains constant.
system. DOE has been used widely in scientific research and indus-
13
15,16
Danilov and Protsenko demonstrated that deposition begins
only when the limiting current for reduction of Cr͑III͒ to Cr͑II͒ is
attained because of the large difference in standard reduction poten-
trial applications.
However, application of DOE in electrochem-
istry has been almost nonexistent. The purpose of this work is to
illustrate the use of DOE in modeling and optimization of elec-
trodeposition of trivalent chromium from glycine-containing baths
with respect to current efficiency. We compare and contrast both the
effects of parameters on current efficiency and the characteristics of
deposits between glycine and formic acid-containing baths.
*
Electrochemical Society Active Member.
E-mail: bdengens@astate.edu
z
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
C505
Table I. Composition of glycine bath.
Chemicals
Weight
CrCl ·6H O
13.87 g
2.63 g
1.22 g
5.35 g
2.91 g
1.19 g
4.83 g
100 mL
3
2
NH CH COOH
2
2
H BO
3
3
NH Cl
4
NaCl
KBr
AlCl ·H O
3
2
Total solution volume
Experimental
Operating conditions and parameters.—Electrodeposition of
trivalent chromium was carried out separately with glycine and
formic acid baths using galvanostatic and pulsed currents. The
composition of the glycine bath was optimized with respect to
current efficiency using statistical software called DESIGN EXPERT
to improve the rate of chromium deposition. The nominal com-
positions of the glycine and formic acid baths are given in Table I
and Table II. Glycine was chosen from a host of amino acids
including ␣-alanine and ␤-alanine. Although ␣-alanine- and ␤-
alanine-containing baths yielded metallic deposits, their physical
appearance was not as good as those with glycine due to black
streaks. ␣-alanine and ␤-alanine are structurally similar to glycine,
Figure 1. UV-visible absorbance spectra of the chromium-glycine mixture
and chromium chloride solution before and after boiling.
by rinsing with distilled water. At the end of electrodeposition, sub-
strates were rinsed with distilled water. Galvanostatic current was
generated using an EG&G/PAR model 362 scanning potentiostat/
galvanostat. Pulsed current electroplating was performed using a
function generator ͑BK Precision͒ in tandem with an oscilloscope
with an additional -CH2
difference.
−
in the molecular structure being the only
͑ELENCO-1252͒ and the potentiostat. To investigate the effects of
pulsing the current, electrodeposition was carried out at different
on-off ratios and frequencies. The effects of temperature, pH, cur-
rent density, and duration of plating on the current efficiency and
quality of the films were compared between glycine and formic acid
baths. Current efficiency was determined based on the weight dif-
ference before and after deposition, as determined by microbalance
weighing ͑CAHN C-32 microbalance͒. Since X-ray photoelectron
spectroscopy ͑XPS͒ analysis showed that the chromium mass con-
centrations in the deposited film were about 80%, chromium current
efficiency was adjusted accordingly. The deposited films were char-
acterized by visual observations, X-ray diffraction ͑XRD͒ ͑Cu-K␣͒,
scanning electron microscopy ͑SEM͒, energy dispersive X-ray spec-
troscopy ͑EDS͒, and XPS. XPS data were collected using a Mg
anode ͑Mg-K␣͒.
Glycine H N-CH -COOH
2
2
␤
-Alanine H N-CH -CH -COOH
2 2 2
␣
-Alanine CH -CH͑NH ͒-COOH
3
2
Ammonium chloride, sodium sulfate, and sodium chloride were
used as conductivity salts. Glycine, formic acid, and sodium oxalate
were complexing agents. Boric acid and aluminum chloride were
1
7
used as buffers. Boric acid also acts as a brightening agent. Bro-
mide salts or triethylene glycol were used to prevent the oxidation of
17,18
Cr͑III͒ into Cr͑VI͒ at the anode.
The role of aluminum chloride
12
was to improve bath stability and quality of chromium deposits.
El-Sharif et al. proposed that aluminum chloride prevents olation
1
2
Preparation of glycine baths.—First, chromium chloride, gly-
cine, ammonium chloride, sodium chloride, and boric acid were
dissolved in ultrapure water and boiled for 30 min. The solution
was cooled to room temperature. Aluminum chloride and potassium
bromide were then added. No pH adjustment was required ͑pH 0.5
to 0.6͒. Chromium forms complexes slowly when Cr͑III͒ salts
are mixed with potential bonding ligands due to the kinetic stability
of aqua-chromium complexes. This is why it is necessary to boil
solutions to speed up the complexation reactions. For example,
when chromium chloride was mixed with glycine, the visible spectra
showed two peaks at ␭ = 589 nm and ␭ = 422 nm after 30 min
reaction during electroplating. As pH in the vicinity of the cathode
3
+
rises due to reduction of hydrogen ions, ͓Al͑H O͒ ͔ will hydro-
2
6
lyze and exert its buffering effect to prevent formation of ␮-hydroxy
14
bridged chromium ͑III͒ species that lead to electrochemical deac-
tivation of chromium complexes.
Graphite was used as an anode, and copper or brass as a cathode.
2
The working area was 4 cm in the case of the optimization of the
2
glycine bath and 2 cm for the rest of the experiments. The non-
working area was insulated by plastic tape. The solution was agi-
tated by a magnetic Teflon-coated stir bar and stirrer. Before elec-
troplating, the substrates were etched in 10% sulfuric acid followed
1
2
of mixing. At the same time, the visible spectra of solutions contain-
ing only chromium chloride showed two maxima corresponding
to ␭ = 596 nm and ␭ = 426 nm. The fact that the peaks of the
1
2
Table II. Composition of formic acid bath.
chromium-glycine mixture were slightly shifted to shorter wave-
lengths than those of the chromium chloride solution suggests the
formation of a chromium-glycine complex in the former. Upon
standing for 5 days, a gradual shift towards shorter wavelengths was
Chemicals
Weight
CrCl ·6H O
10.65 g
3.0 mL
3
2
observed both in chromium-glycine mixtures ͑␭ = 571 nm and
1
HCOOH
␭2
= 412 nm͒ and chromium chloride solutions ͑␭ = 577 nm and
Na ͑COO͒
2.0 g
1
2
2
␭2
= 414 nm͒. When the chromium-glycine mixture was boiled for
H BO
3.71 g
9.68 g
7.1 g
3
3
30 min, an almost identical shift was observed on the spectra as that
NH Cl
4
obtained from the 5-day-old ͑␭ = 572 nm and ␭ = 414 nm͒ ͑Fig.
Na SO
1
2
2
4
KBr
1.55 g
0.5 mL
100 mL
1͒. This suggests that boiling accelerates ligand exchange and
achieves the equilibrated state quickly. Similar results were also ob-
tained from boiling the chromium solution alone. The degree of shift
Triethylene glycol
Total solution volume
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C506
Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
Figure 2. A typical central composite design for three variables.
Figure 3. Interaction graph obtained from screening experiments.
obtained for the chromium chloride solution in 5 days was similar to
that obtained by boiling for 30 min. When the glycine bath was
boiled, glycine reacted with chromium chloride and chromium com-
plexes, and hydrochloric acid was formed, decreasing pH
of sodium chloride and boric acid were set at 14.5 and 12.2 g/l,
respectively.
3
+
−
2+
−
͓
Cr͑H O͒ ͔ 3Cl + GlyH → ͓Cr͑H O͒ Gly͔ 2Cl + 2H O + HCl
Optimization.—The main purpose of optimization is to determine
optimum factor settings to elicit the maximum response, in our case,
current efficiency. It is possible to build approximate models repre-
sentative of the true process and identify near-optimal combinations
of factors. For a single response process, path of steepest ascent
͑PSA͒, simplex, and response surface methods ͑RSM͒ are the most
widely used methods for optimization. These are basically statistical
curve-fitting methods based on experimental data. In this study, a
central composite design, which is one of the response surface meth-
ods, was employed for optimization. A central composite design
consists of three groups of design points: ͑i͒ two-level factorial or
fractional factorial deigns points; ͑ii͒ axial points; and ͑iii͒ center
points ͑Fig. 2͒. Each variable is varied over five levels ͑i.e., con-
centrations͒. A matrix of 20 combinations of concentrations was
generated using the central composite design by incorporating six
center points. The alpha ͑␣͒ value which was the measure of the
rotatability of the design was set at 1.68179. In a central composite
design, the alpha ͑␣͒ value represents the distance from the center of
the design space to an axial point. Rotatability means that the varia-
tion in the response predicted by a model is constant at a given
distance for the center of the design.
2
6
2
4
2
͓
7͔
Preparation of formic acid baths.—First, ammonium chloride,
boric acid, potassium bromide, and sodium sulfate were dissolved in
ultrapure water by heating the solution slightly. Chromium chloride
was added, followed by the addition of formic acid and sodium
oxalate. The solution was boiled for 5 min, and the temperature was
lowered to 50°C. The solution was stirred at 50°C for 1 h and cooled
to room temperature, followed by the addition of triethylene glycol.
The solution was stirred overnight at room temperature. The next
day, the pH was adjusted to 1.5 using sodium or potassium hydrox-
ide, and the solution was stirred for 1 h at 50°C. Finally, the pH was
adjusted to 2.5 or above. It was required that these steps be followed
closely to obtain reproducible results. Electrodeposition of chro-
mium was carried out at 25°C.
2
0
Optimization of glycine baths.—Statistical experimental design
and optimization of bath constituents with respect to current effi-
ciency were carried out using statistical software called the DESIGN
EXPERT ͑version 6.06, Stat-Easy Inc., Minneapolis, MN͒ in two
steps: ͑1͒ screening and ͑2͒ optimization. The purpose of screening
was to identify bath constituents that significantly affect current ef-
ficiency. Once identified, these were subjected to optimization.
Screening experiments were conducted using a two-level factorial
Results and Discussion
Screening.—The results of residual analyses showed no viola-
tions of model assumptions ͑normal plot of residuals, outlier-T plot,
residual vs predicted plot͒. The results of analysis of variance
1
9
design, whereas optimization experiments were carried out using a
central composite design.
20
͑
ANOVA͒ indicated that glycine and chromium chloride are the
Screening.—There were seven constituents in a glycine bath ͑see
Table I͒. Since boric acid was used as a buffer and sodium chloride
as a conductivity salt, their influence on current efficiency would be
expected to be minimal and, hence, they were excluded from
the two-level factorial design. The remaining five bath constituents
most influential bath constituents. The test for curvature showed the
presence of significant curvature in the response. There is significant
interaction between glycine and chromium ͑Fig. 3͒. This suggests
that the effect of glycine on current efficiency depends on the con-
centration of chromium chloride. The effect is more pronounced at
higher concentrations of glycine. It was found that current efficiency
depends on the glycine-chromium chloride molar ratio. As the ratio
increased above 1, current efficiency decreased gradually. At a molar
ratio of 1, most complexes have one glycine ligand attached to one
chromium ion. As the molar ratio increases, the concentrations of
complexes with two or three glycine ligands also increase, while
complexes with one glycine ligand decrease. Complexes with mul-
tiple glycine ligands are less electrochemically facile for chromium
͑
and interferes with Cr͑III͒ deposition. Glycine forms a chelate ring
with chromium through its nitrogen and oxygen atoms.
͑
or factors͒ were included. Two-level factorial designs are useful
19
for estimating the main effects and interactions. Here, each factor
variable͒ is set at two levels ͑that is, concentrations͒. Resolution V
͑
was considered in designing a fractional factorial design. Resolution
is the measure of degree of confounding for the main effects and
interactions. The higher the resolution the better is the design to
estimate higher order interactions. Four center points were also
included in the design to check for curvature. A design matrix of
21
5
−1
2
fractional factorial design was generated by DESIGN EXPERT.
III͒ electrodeposition than the complex that has one glycine ligand,
Altogether there were 20 runs ͑baths͒. Each run represented a
combination of concentrations of bath constituents. Trivalent chro-
mium from each bath was electroplated for 5 min using a graphite
2
+2
anode and copper cathode at 0.175 A/cm and 30°C. Concentrations
Cr͓͑H N-CH -COO͒͑H O͒ ͔ Monoglycine ligand complex
2
2
2
4
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
C507
Figure 6. Effect of temperature on current efficiency at current density
2
0
.15 A/cm .
Figure 4. 3D graph of reduced quadratic model.
Optimization.—Based on the results of statistical analysis, a
quadratic model with “probability Ͼ F ” value of less than 0.0001
was chosen from a group comprised of linear, two-factor interaction,
quadratic, and cubic models. A linear model also had a low
probability Ͼ F value ͑0.0066͒, but it showed a significant lack of
fit ͑probability Ͼ F = 0.0036͒. Bath constitutents found to be im-
portant were chromium chloride and glycine. In agreement with
screening experiments, aluminum chloride was not an influential
bath constituent. The lack-of-fit test showed that the data fit the
+
Cr͓͑H N-CH -COO͒ ͑H O͒ ͔ Diglycine ligand complex
2
2
2
2
2
Cr͓͑H N-CH -COO͒ ͔ Triglycine ligand complex
2
2
3
In our study, it was observed that there was no decline in current
efficiency when the molar ratio decreased to below 1. However,
2
1
2
quadratic model well. This is further supported by higher R and
El-Sharif et al. noted that the total amount of chromium that can
be electroplated was highest at a molar ratio of 1:1. At other ratios,
the amount deposited decreased gradually. The fact that there was no
decline in current efficiency when lowering the ratio below 1, but a
decrease in the total amount of chromium that could be electro-
plated, implies that there should be uncomplexed chromium in the
glycine bath which cannot be electroplated. Strong interaction be-
tween glycine and chromium occurs in the model because chromium
is electroplated from the chromium-glycine complex and the amount
of suitable complex varies depending on the glycine-chromium mo-
lar ratio.
2
adjusted R values which indicate a good explanation of the vari-
2
ability by the selected model. The prediction R of 0.7066 was not
2
as close to the adjusted R of 0.9135 as one would like. This sug-
gests that model reduction might have been necessary. To improve
the model, backward elimination regression was employed with the
␣
-value ͑significance level͒ set at 0.05.
From the backward elimination regression, a reduced quadratic
2
model was obtained. The reduced model had a prediction R of 0.81
2
which was close to the adjusted R of 0.88. Adequacy of the model
Figure 7. Effect of duration of chromium electroplating on current efficiency
at current density 0.15 A/cm2 and temperature 30°C for glycine baths and
25°C for formic acid baths.
Figure 5. Effect of pH on current efficiency at current density 0.15 A/cm2
and temperature 30°C for glycine baths and 25°C for formic acid baths.
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C508
Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
Figure 10. Comparison of amount of chromium deposited by pulsed current
and constant current plating from a glycine bath ͑A = pulsed current for 10
2
2
min at average current density = 0.15 A/cm ͑minimum = 0 A/cm and
2
maximum = 0.30 A/cm ͒, B = constant current for 10 min at current
density = 0.15 A/cm , C = constant current at 0.30 A/cm for 5 min͒.
2
2
Current efficiency ͑%͒
=
6.84 + 0.10 ϫ chromium chloride ͑g/L͒
−
−
+
0.28 ϫ glycine ͑g/L͒
5.24 ϫ 10−4 ϫ chromium chloride ͑g/L͒2
1.62 ϫ 10−3 ϫ chromium chloride ͑g/L͒ ϫ glycine ͑g/L͒
Figure 8. Effect of current density on current efficiency at temperature 30°C
for glycine baths and 25°C for formic acid baths.
͓
8͔
After selecting the appropriate model and testing model diagnos-
tics, optimization was performed by a numerical method available
with DESIGN EXPERT software. Numerical optimization can be repre-
sented by a general nonlinear algorithm with constraints applied to
the main objective function. In this case, the main objective was to
maximize current efficiency under given constraints.
Solutions obtained from numerical optimization for both unre-
duced and reduced quadratic models showed different optimal con-
centrations of bath constituents. Since the reduced model is the most
appropriate model, the optimal set of factors obtained from the re-
duced model was selected for further study. Thus, 138.78 g/L chro-
mium chloride, 26.30 g/L glycine, and 48.37 g/L aluminum chloride
were selected as the optimal combination. The final composition of
the glycine bath obtained after optimization is given in Table I. The
current efficiency obtained from experimental electrodeposition of
chromium with the selected bath composition was in close agree-
ment ͑within 5%͒ with the current efficiency predicted from the
reduced quadratic model.
was checked by residual analysis ͑normal plot of residuals, residuals
vs predicted plot, outlier-T plot, leverage, and Cook’s distance͒,
which showed no violations of model assumptions for both the un-
reduced and reduced model. A three-dimensional graph of the re-
duced quadratic model is shown in Fig. 4. The reduced model con-
tained fewer but important terms. The dependence of current
efficiency on chromium chloride and glycine concentrations can be
described by the following empirical equation in the range of con-
centrations studied:
Effects of process parameters.—The effects of pH, temperature,
current density, duration of plating, and pulse plating on current
efficiency were studied for formic acid and glycine baths. The
current efficiency remained more or less constant in the pH range
0
5
.3-0.7 with a slight decrease above pH 1.0 for glycine baths ͑Fig.
͒. The physical appearance of the chromium films deteriorated at
pH above 1.5 with the appearance of black patches. No metallic
deposits were obtained at pH’s below 0 and above 2. In contrast,
increase in pH resulted in a gradual decline in current efficiency
with an initial increase up to pH 2.5 in a formic acid bath ͑Fig. 5͒.
Although high pH levels ͑3- 3.8͒ reduced current efficiency and,
hence, the thickness of deposits, deposit quality was superior ͑bright
Table III. EDS analysis of typical deposits obtained from glycine
and formic acid baths.
Figure 9. Comparison of amount of chromium deposited by pulsed current
and constant current plating from a formic acid bath ͑A = pulsed current for
2
2
1
5 min at average current density = 0.15 A/cm ͒ ͑minimum = 0 A/cm and
2
Glycine bath
Formic acid bath
Mass conc. Atomic conc.
maximum = 0.40 A/cm ͒, B = constant current at current density of
0
2
2
.15 A/cm for 15 min, C = constant current at 0.40 A/cm for 5 min, 38 s͒.
Mass conc.
Atomic conc.
Note: The current efficiencies for constant current plating ͑B͒ in Fig. 8 are
not all exactly equal because they were obtained from baths of different ages.
Freshly prepared baths yielded higher current efficiency than the used baths.
However, the age of electrolytic baths would not alter the nature ͑trend͒ of
relationships among A, B, and C as observed in Fig. 9.
Element
͑%͒
͑%͒
͑%͒
͑%͒
Cr
O
Ca
88.6
11.0
0.4
70.9
28.6
0.4
85.7
14.1
0.25
65.1
34.7
0.25
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
C509
Table IV. Elemental composition of deposits obtained from glycine baths as measured by XPS after sputtering away 100 Å of chromium film.
Pulsed current
Constant current
5 min͒
Constant current
Constant current 2
͑10 min, 1000 Hz, 2
͑
͑40 min͒
͑10 min, 0.15 A/cm ͒
Avg c.d. = 0.15 A/cm ͒
White patch
Atomic
Atomic
conc.
͑%͒
Atomic
Atomic
Atomic
Mass conc.
Mass conc.
conc.
Mass conc.
conc.
Mass conc.
conc.
Mass conc.
conc.
Elements
͑%͒
͑%͒
͑%͒
͑%͒
͑%͒
͑%͒
͑%͒
͑%͒
͑%͒
Cr 2p
O 1s
C 1s
N 1s
Ca 2p
80.0
4.9
14.1
0.1
50.4
10.1
38.6
0.3
77.0
7.2
15.9
45.6
13.8
40.6
85.5
3.6
9.4
0.5
1.03
60.7
8.2
28.9
1.3
84.0
3.3
11.5
0.3
57.2
7.3
34.0
0.8
75.24
8.58
15.75
0.06
43.74
16.21
39.64
0.13
0.81
0.65
0.95
0.92
0.81
0.38
0.29
metallic luster͒. The decline in the deposition rate with increase in
pH was due to olation reactions that formed ␮-hydroxy bridged
chromium ͑III͒ species, deactivating chromium complexes from
formic acid baths. The decline of current efficiency with increase in
current density is an advantage because it causes uniform metal
distribution. In contrast, metal distribution would not be uniform
for glycine baths because more metal was deposited in high current
density areas such as edges and corners than in low density areas
such as the center of the substrate. El-Sharif et al. noted that
the thickness of the deposit was 40 ␮m at the edge compared to
12
which chromium can be deposited. The effect of temperature on
current efficiency was identical at temperatures above 25°C in for-
mic acid and glycine baths ͑Fig. 6͒. In both cases, deposition rate or
current efficiency declined with increase in temperature. An increase
in temperature causes the increased reduction of hydrogen ions at
12
30 ␮m at the center. This behavior is similar to that of hexavalent
1
2
−
the cathode, leading to increase of double-layer OH concentra-
tion. This can form inert oxy-and hydroxy complexes of the
chromium-glycine complex. However, the initial increase in current
efficiency in formic acid baths with the rise of temperature from 15
to 25°C was probably due to a decline in pH from 2.6 to 2.5, which
increased current efficiency as evidenced by Fig. 5. In other words,
the effect of pH is dominant over the effect of temperature at low
temperatures.
chromium electroplating, where an increase in current density
causes an increase in current efficiency, which results in a poor
metal distribution.
The current efficiency of approximately 9.5% was obtained for
2
formic acid baths at pH 2.5, 25°C, and 0.15 A/cm after 10 min of
plating. For glycine baths, the current efficiency was approximately
2
8
.3% at 30°C, pH 0.4-0.6, and 0.15 A/cm , and increased to 14.4%
2
12
at 0.4 A/cm . The current efficiency obtained by El-Sharif et al. at
.4 A/cm was about 10% for glycine baths, lower than obtained in
The effect of the duration of deposition on current efficiency is
illustrated in Fig. 7. In both glycine and formic acid baths, a decline
in current efficiency was observed as the duration increased. How-
ever, the rate of decline was more rapid in formic acid baths ͑pH
2
0
this study. Based upon the trend ͑Fig. 8͒, a further increase in current
density may increase current efficiency. However, because higher
current density caused the deposit to crack and flake, it was difficult
to determine current efficiency based on the amount deposited.
These current efficiencies are comparable to those of hexavalent
2.5͒ than in glycine baths. The rate of decline was relatively slower
at higher pH levels ͑3-3.8͒ in formic acid baths. Results obtained
with glycine baths did not agree with those obtained by El-Sharif et
12
chromium electroplating. However, current efficiencies greater than
al., which showed no such decline even with a deposition time of
0 min. Cathode surface pH increases with increase in duration of
11
40% have been reported for trivalent chromium-based baths.
6
22
Aqueous trivalent chromium-based baths have thickness limitations
with extended electroplating due to low current efficiencies as a
result of competition from hydrogen ions and gradual decline in
current efficiency due to olation reactions.
deposition, as was established by Tu et al. They found that the pH
quickly rose to about 8.4 from an initial pH of 3.5 in 45 min in
carboxylic acid-based baths. This rise in pH causes the formation of
the oxy-and hydroxy chromium ͑III͒ species, consistent with the
reduction in glossiness of the deposit with increase in deposition
time because of accumulation of chromium oxide/hydroxide on the
deposit in the case of formic acid baths. This is why it is difficult to
obtain thick films even with extended plating in formic acid baths.
However, there was no decrease in glossiness of deposits obtained
from glycine baths with extended plating, even though the rate of
deposition declined gradually.
To compare the effects of constant current vs pulsed current plat-
ing on the rate of deposition, three sets of experiments were con-
ducted for formic acid and glycine baths. They were ͑A͒ pulsed
current plating at different frequencies with an average current den-
2
sity of 0.15 A/cm and on-off ratio of 50:50 for glycine baths and
3
8:62 for formic acid baths; ͑B͒ constant current plating at
2
2
0.15 A/cm ; and ͑C͒ constant current plating at 0.3 A/cm with half
of the total plating time as with pulsed or constant current plating.
Experiments were designed in such a way that the total charge
passed through the system was the same in all three sets. In the
Current efficiency varied with current density as indicated in
Fig. 8. Whereas an increase in current density led to an increase in
current efficiency in glycine baths, the opposite was true for
Table V. Elemental composition of deposits obtained from formic acid baths as measured by XPS after sputtering away 100 Å of chromium film.
Constant current
5 min͒
Constant current
Constant current 2
Pulsed current ͑10 min, 1000 Hz,
2
͑
͑40 min͒
͑10 min, 0.15 A/cm ͒
Avg c.d. = 0.15 A/cm ͒
Mass conc.
Atomic conc.
Mass conc.
Atomic conc.
Mass conc.
͑%͒
Atomic conc.
͑%͒
Mass conc.
Atomic conc.
Elements
͑%͒
͑%͒
͑%͒
͑%͒
͑%͒
͑%͒
Cr 2p
O 1s
C 1s
N 1s
Ca 2p
76.0
5.0
18.8
0.3
43.5
9.2
46.7
0.6
73.4
10.9
15.4
0.3
41.6
20.0
37.8
0.6
82.3
5.2
11.4
0.3
54.7
11.3
32.7
0.7
78.5
7.0
13.9
0.6
48.0
13.9
36.8
1.3
0.8
0.7
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C510
Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
nificantly more chromium at both 50 Hz and 1000 Hz than did
constant current plating ͑B͒ ͑Fig. 10͒. This contrasting behavior can
be explained by analyzing the effect of current density on current
efficiency.
For the formic acid bath, a current density vs efficiency graph
͑Fig. 8͒ revealed that current efficiency decreased with an in-
crease in current density. The minimum and maximum current
2
densities were set at 0 and 0.4 A/cm , respectively, for pulsed
2
current plating. It follows from Fig. 8 that with 0.4 A/cm , the cur-
2
rent efficiency was considerably lower than that at 0.15 A/cm in
constant current plating. It follows that pulsed current plating should
result in less deposition than constant current plating, even though
the average current density and duration of plating are the same.
Since increases in current density increased the current efficiency
for the glycine bath, pulsed current plating resulted in more chro-
mium being deposited because the current density fluctuated be-
2
Figure 11. Intensity of ␣-Cr peaks of deposit ͑formic acid bath͒ annealed in
air at different temperatures ͑annealing time = 30 min for 300-400°C, for
other temperatures 15 min͒.
tween 0 and 0.3 A/cm and the current efficiency was higher at
2
2
0.3 A/cm than at 0.15 A/cm used in constant current plating. The
amount deposited using pulsed current with an average current den-
sity of 0.15 A/cm was also less than the amount deposited using a
2
2
23
constant current at 0.30 A/cm for half of the time. Tu et al. found,
by using a carboxylic acid-based bath, that pulse plating improved
the current efficiency but offered no explanation. These results show
that whether pulsed current plating increases or decreases the
amount deposited relative to constant current plating, with the same
amount of charge being passed, depends on how current density
influences current efficiency and the minimum and maximum cur-
rent density values chosen. If the maximum current density lies on
the ascending part of the current density-current efficiency curve,
the pulsed current increases the amount deposited relative to con-
formic acid bath, the amount of chromium deposited decreased in
the order B Ͼ A Ͼ C ͑Fig. 9͒. This was particularly evident at 50
Hz. As the frequency increased, the difference in amount deposited
by pulse current plating ͑A͒ and constant current plating ͑B͒ became
less significant, as evidenced by the overlapping of error bars
͑
± standard deviation͒ ͑Fig. 9b and c͒, although average values still
followed the order B Ͼ A Ͼ C. In the glycine bath, the opposite
phenomenon was observed, with the amount deposited decreasing
in the order C Ͼ A Ͼ B. Pulsed current plating ͑A͒ deposited sig-
Figure 12. SEM micrographs show-
ing ͑a͒ microcracking and ͑b͒ macroc-
racking on the deposit obtained from
2
constant current plating at 0.15 A/cm
for 5 and 40 min, respectively, from a
glycine bath and ͑c͒ the deposit ob-
tained by pulsed current plating ͑aver-
2
age current density 0.15 A/cm , 1000
Hz͒ for 10 min from glycine bath.
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
C511
Figure 13. SEM micrographs of de-
posits obtained by constant current
plating for ͑a͒ 5 min; ͑b͒ 40 min; and
͑
1
c͒ 10 min and by pulse plating for ͑d͒
0 min at 0.15 A/cm from a formic
2
acid bath.
stant current. However, if the maximum current density lies on the
descending part of the current density-current efficiency curve,
pulsed current lowers the amount deposited relative to constant cur-
rent. In other words, the efficiency obtained from pulse plating is
approximately the average of current efficiencies obtained in cases B
and C.
few angstroms of the surface/enviroment interface. White patches,
which often occurred in the deposits obtained from glycine baths
and were present at high current density areas such as tips and edges
of the substrates, contained slightly less chromium. Although the
elemental composition of the white patch ͑Table IV͒ was similar to
that of the shiny metallic deposit, the appearance was drastically
different. An SEM micrograph of the white patch revealed poorly
coalesced chromium nodules, and the color was similar to that of
burning as obtained with hexavalent chromium processes at high
current densities.
Deposit characterization and analysis.—EDS analyses of typi-
cal chromium deposits obtained by galvanostatic plating for 10 min
2
at 0.15 A/cm are shown in Table III. The mass and atomic concen-
trations of chromium were slightly higher in the deposits obtained
from glycine baths.
X-ray diffraction ͑XRD͒ of as-deposited chromium films from
glycine and formic acid baths revealed no chromium peaks. Exami-
nation of XRD spectra of the deposits annealed at 350°C in air for
Results of XPS analysis ͑after sputtering away 100 Å of the
deposits͒ are presented in Tables IV and V. The chromium mass
concentration was approximately 80%. In agreement with EDS, the
chromium content in the deposits was higher for glycine baths than
formic acid baths. Carbon and oxygen were found to occur in sig-
nificant quantities ͑Tables IV and V͒. Calcium and nitrogen were
present as contaminants. The chromium mass concentration in de-
posits obtained from glycine baths declined slightly from 80 to 77%
as the duration of plating increased from 5 to 40 min ͑Table IV͒. A
similar trend was also obtained with formic acid baths ͑Table V͒.
This is consistent with the appearance of the deposits changing from
shiny metallic to dull and ash-colored, indicating more oxygen and
carbon after 40 min of plating at pH 2.5 in formic acid baths.
However, a surface scan by XPS revealed a much smaller mass
concentration for chromium ͑approximately 20%͒ and significant
mass concentrations for carbon ͑32-51%͒ and oxygen ͑22-36%͒.
These high surface concentrations for carbon and oxygen are not
surprising because these elements are easily adsorbed on/in the first
3
4
0 min indicated the presence of a ͑100͒ ␣-Cr peak at a 2␪ value of
4.46° ͑Cu-K␣͒. This suggests that as-deposited chromium is amor-
phous. Chromium can exist in several phases ͑␣, ␤, ␦, and ␴͒. Of
17
these, ␣-Cr is the most stable and predominant phase. Other
phases are unstable, occur under specific conditions, and eventually
1
7,24,25
get converted to ␣-Cr.
Increasing the annealing temperature
from 300 to 450°C increased the intensity of the ␣-Cr chromium
peak in deposits obtained from both the formic acid and glycine
baths ͑Fig. 11͒. A slight shift in the ␣-Cr peak was observed with
increases in annealing temperature. Annealing in air at 500°C de-
stroyed the films.
The deposits obtained from glycine baths were shiny and
metallic in appearance. Good quality deposits were obtained when
plated for a short duration ͑less than 5 min͒ and at low current
2
densities ͑Ͻ0.15 A/cm ͒. At higher current densities, deposits be-
came brittle and often flaked, especially around the edges. The
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C512
Journal of The Electrochemical Society, 152 ͑7͒ C504-C512 ͑2005͒
thickness increased with extended electroplating, but the deposits
became brittle around edges and corners even though current densi-
ties were low ͑Ͻ0.15 A/cm ͒. This was because, as thickness
started to increase, microcracks were converted into macrocracks
due to stresses. An SEM micrograph ͑Fig. 12a͒ indicated microc-
racks at the deposit which later turned into macrocracks ͑Fig. 12b͒
with an increase in thickness. The deposits obtained by constant
current plating were smooth with no nodular growth ͑Fig. 12a͒.
However, deposits obtained by pulsed current revealed small nodu-
lar growths ͑Fig. 12c͒.
edges of the substrate. Chromium mass percentages were slightly
higher in deposits obtained from glycine baths than in those from
formic acid baths. The electrodeposition of trivalent chromium from
formic acid and glycine baths exhibited contrasting results with re-
spect to effects of various parameters such as pH, temperature, cur-
rent density, duration of plating, and pulsed current on current effi-
ciency. Pulsed current plating may increase or decrease current
efficiency relative to constant current plating, depending on how
current density affects current efficiency. The results provided here
are reproducible as long as freshly prepared electrolytic baths are
used.
2
Thin, shiny, metallic deposits were obtained with formic acid
baths when electroplated for a shorter duration ͑less than 5 min͒.
The thin chromium deposits were microporous as evidenced by the
presence of small pores ͑Fig. 13a͒. Thin deposits were generally
smooth with few nodular growths. However, as thickness increased
with plating time, granular deposits were obtained. At higher mag-
nifications ͑5000͒, an SEM micrograph ͑Fig. 13b͒ revealed nodular
growths in parallel rows. Unlike glycine baths, no brittle deposits
were obtained at higher current densities. As the duration of electro-
plating increased, deposits gradually turned from shiny metallic to
dull metallic to dark gray. Continuous microcracking was observed
on the deposits obtained from both pulsed current ͑1000 Hz, 50:50
Acknowledgments
We gratefully acknowledge Dr. John Frick and the Institute of
Hazardous Materials Management ͑Rockville, MD͒ for providing a
Graduate Student Research Grant to support this work. We also
appreciate the financial support provided by the EPA/AR EPSCoR
Program, and both the Environmental Science Program ͑Dr. Jerry
Farris-Director͒ and College of Engineering ͑Dr. Rick Clifft-Interim
Dean͒ at Arkansas State University.
Arkansas State University assisted in meeting the publication costs of this
article.
2
on-off ratio, average current density = 0.15 A/cm ͒ and constant
2
References
current plating ͑0.15 A/cm ͒. Transverse microcracks were more
1
. R. R. Lloyd, Trans. Electrochem. Soc., 89, 443 ͑1946͒, as mentioned in J. C.
Crowther, and S. Renton, Electroplat. Met. Finish., 28, 6 ͑1975͒.
. M. R. Zell, U.S. Pat. 2, 801, 214 ͑1957͒.
visible on a thick deposit ͑Fig. 13b͒ than on a thin deposit ͑Fig. 13a͒.
Similar to the deposits obtained from glycine baths, the deposits
obtained from formic acid baths by constant current plating were
smooth with less nodular growth ͑Fig. 13c͒ than with the deposits
obtained with pulsed current for the same plating duration of 10 min
2
3. T. Kudryastev and I. A. Bodrov Trudy, Inst Electrochem., 26, 105 ͑1959͒.
4. T. Ishaguro, J. Electrochem. Soc. Jpn., 23, 215 ͑1955͒.
5
6
7
8
. A. J. Deyrup, U.K. Pat., 981, 481 ͑1965͒
. A. J. Deyrup, U.S. Pat., 2, 801, 823 ͑1961͒.
. W. H. Safranek, U.S. Pat., 3, 111, 464 ͑Patent applied for 29/9/61͒.
. L. Gianelos, Plat. Surf. Finish., 69, 30 ͑1982͒.
͑
Fig. 13d͒.
Conclusions
Based on the statistical design of experiments ͑DOE͒, we carried
9. A. K. Hsieh, K. M. Chen, and M. F. Chung, Met. Finish., 91, 45 ͑1993͒.
0. J.-Y. Hwang, Plat. Surf. Finish., 78, 118 ͑1991͒.
1
1
1. A. A. Edigaryan and Y. M. Polukarov, Prot. Met., 34, 95 ͑1998͒.
12. M. El-Sharif, J. McDougall, and C. U. Chisholm, Trans. Inst. Met. Finish., 77, 139
1999͒.
out empirical modeling and optimization of the current efficiency of
trivalent chromium electrodeposition from glycine baths. The results
indicated that a quadratic curve-fitting model could be appropriate to
model the variation of current efficiency with concentration of chro-
mium chloride and glycine, the most influential bath constituents
with respect to current efficiency. The strong statistical interaction
between chromium chloride and glycine, as indicated by DOE, is
consistent with chromium being electroplated from chromium-
glycine complexes and the amount of suitable complex ͑monogly-
cine ligand complex͒ depending upon the molar ratio of glycine and
chromium. As-deposited chromium was found to be amorphous.
Due to a decline in the rate of deposition with time, it was difficult
to obtain thick deposits of good quality from formic acid baths.
Thin, shiny, and metallic deposits were obtained when electroplated
for a short duration. Extended electroplating resulted in dull deposits
in the case of formic acid baths. In the case of glycine baths, ex-
tended plating resulted in brittle deposits, especially around the
͑
1
1
3. F. I. Danilov and V. S. Protsenko, Prot. Met., 37, 224 ͑2001͒.
4. A. M. Smith, A. Watson, and D. H. Vaughan, Trans. Inst. Met. Finish., 71, 106
͑
1993͒.
15. O. Sreekumar, N. Chand, and S. C. Basappa, J. Biosci. Bioeng., 88, 334 ͑1999͒.
1
1
1
1
6. C.-J. Shieh, P. Keohler, and C. Akoh, J. Food. Sci., 61, 97 ͑1996͒.
7. H. Dasarathy, C. Riley, and D. Coble, J. Electrochem. Soc., 141, 1773 ͑1994͒.
8. T. Hayashi, and A. Ishihama, Plat. Surf. Finish., 66, 36 ͑1979͒.
9. D. Montgomery, Design and Analysis of Experiments, 5th ed., John Wiley & Sons,
New York ͑2001͒.
0. R. H. Myers and D. Montgomery, Response Surface Methodology: Process and
Product Optimization Using Designed Experiments, 2nd ed., John Wiley & Sons,
New York ͑2002͒.
2
2
1. J. McDougall, M. El-Sharif, and S. Ma, J. Appl. Electrochem., 28, 929 ͑1998͒.
22. Z. Tu, Z. Yang, J. Zhang, M.-Z. An, and W.-L. Li, Plat. Surf. Finish., 80, 79
1993͒.
͑
2
2
2
3. Z. Tu, Z. Yang, and J. Zhang, Plat. Surf. Finish., 77, 55 ͑1990͒.
4. R.-Y. Tsai and S.-T. Wu, J. Electrochem. Soc., 137, 638 ͑1990͒.
5. M. H. Gelchinski, L. Gal-Or, and J. Yahalom, J. Electrochem. Soc., 129, 2433
͑1982͒.
Downloaded on 2015-01-08 to IP 131.111.185.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).