In situ stress measurements during aluminum deposition from Al Cl3 -EtMeImCl ionic liquid

Article abstract of DOI:10.1149/1.2168048

In situ stress measurements were made during aluminum electrodeposition from Lewis acidic aluminum chloride-1-ethyl-3-methylimidazolium chloride (Al Cl3 -EtMeImCl) using the wafer curvature method. The underpotential deposition of Al on (111)-textured Cu shows a stress response that is consistent with a two-step process where the first step involves the desorption of Al Cl4- from the Cu surface. This results in a tensile surface stress which is consistent with adsorbate-induced stress models that appear in the literature. The tensile stress is eliminated when the full incommensurate monolayer of aluminum is formed. On (111)-textured Au, the full Al monolayer results in a tensile stress that can be attributed to both lattice misfit and Al-Au alloying. The bulk deposition of Al on (111)-textured Au appears to be governed by Stranski-Krastanov three-dimensional growth. Initially a compressive stress is observed that we attribute to capillarity effects. This is followed by a rapid increase in tensile stress that is consistent with nuclei coalescence and grain boundary formation. In this region of nucleation and growth, the total stress is the superposition of both stress mechanisms and each is dependent on the deposition potential.

Full text of DOI:10.1149/1.2168048

Journal of The Electrochemical Society, 153 ͑4͒ C207-C212 ͑2006͒
C207
0013-4651/2006/153͑4͒/C207/6/$20.00 © The Electrochemical Society
In Situ Stress Measurements during Aluminum Deposition
from AlCl3-EtMeImCl Ionic Liquid
,z
b,
G. R. Stafford,a, O. E. Kongstein,
*
a,**
*
and G. M. Haarberg
aNational Institute of Standards and Technology, Materials Science and Engineering Laboratory,
Gaithersburg, Maryland 20879, USA
bDepartment of Materials Technology, Norwegian University of Science and Technology, NO-7491
Trondheim, Norway
In situ stress measurements were made during aluminum electrodeposition from Lewis acidic aluminum chloride-1-ethyl-3-
methylimidazolium chloride ͑AlCl3-EtMeImCl͒ using the wafer curvature method. The underpotential deposition of Al on ͑111͒-
textured Cu shows a stress response that is consistent with a two-step process where the first step involves the desorption of AlCl4−
from the Cu surface. This results in a tensile surface stress which is consistent with adsorbate-induced stress models that appear
in the literature. The tensile stress is eliminated when the full incommensurate monolayer of aluminum is formed. On ͑111͒-
textured Au, the full Al monolayer results in a tensile stress that can be attributed to both lattice misfit and Al–Au alloying. The
bulk deposition of Al on ͑111͒-textured Au appears to be governed by Stranski–Krastanov three-dimensional growth. Initially a
compressive stress is observed that we attribute to capillarity effects. This is followed by a rapid increase in tensile stress that is
consistent with nuclei coalescence and grain boundary formation. In this region of nucleation and growth, the total stress is the
superposition of both stress mechanisms and each is dependent on the deposition potential.
© 2006 The Electrochemical Society. ͓DOI: 10.1149/1.2168048͔ All rights reserved.
Manuscript submitted June 28, 2005; revised manuscript received November 14, 2005. Available electronically February 21, 2006.
Electrodeposited films tend to develop sizable mechanical
stresses as a result of the nucleation and growth process or from the
use of solution additives and alloying elements needed to achieve
desired deposition rates and mechanical properties. Often these
stresses can approach or exceed the yield stress of the bulk material
and can lead to loss of adhesion and the generation of bulk and
surface defects. The stress observed at room temperature on a coated
surface is typically the result of two different phenomena. The first,
thermal stress, is present when the film is deposited at a temperature
different from the service temperature and the film and substrate
have different thermal expansion coefficients. The second, intrinsic
stress, is caused by the manner in which the film grows and can arise
from lattice-mismatched epitaxial growth, nuclei coalescence and
grain growth during deposition, incorporation of impurities or side-
reaction products, and phase transformations accompanied by vol-
ume changes. Although similar growth morphologies and stress de-
velopment are inherent to both electrodeposited films1 and those
grown from the gas phase,2-6 the electrodeposition community has
only recently made efforts to better understand the microstructural
contributions to film stress.
The direct electrodeposition of aluminum–transition metal alloys
from ambient-temperature ionic liquids has recently received con-
siderable attention. These electrolytes are prepared by combining
AlCl3 with certain unsymmetrical quaternary ammonium chloride
salts such as 1-ethyl-3-methylimidazolium chloride ͑EtMeImCl͒.
Progress in this area has recently been reviewed.7 Among the alloys
that have been successfully electrodeposited are Al–Cr,8,9
Al–Mn,10,11 Al–Ti,12-15 and Al–V.16 Many of these alloys may be of
technological importance, particularly in thin-film form, because
they show improved resistance to chloride-induced pitting corrosion
relative to Al. Although these alloy films may exhibit several of the
electrochemical properties required of functional coatings, the
stresses associated with their formation are virtually unknown. The
purpose of this paper is to examine the stresses associated with the
earliest stages of aluminum deposition from the ambient temperature
aluminum chloride-1-ethyl-3-methylimidazolium chloride ͑AlCl3-
EtMeImCl͒ ionic liquid. The in-plane stress generated in the alumi-
num deposit during growth was calculated from the perpendicular
deflection of a flexible cathode. The deflection was measured by a
laser beam reflected off of the back surface.17-21 This and similar in
situ techniques have been widely used to examine adsorbate-induced
surface stress,21-28 underpotential deposition,29-33 electrochemical
insertion and intercalation reactions,17,18,20,34 and bulk film
deposition.35-38
Experimental
A schematic of the components of the in situ stress measurement
apparatus is shown in Fig. 1. The light source was a 1-mW helium–
neon laser ͑JDSc Uniphase, model 1108P͒. A beam splitter ͑50%
reflected, 50% transmitted͒ was placed in the path of the beam to
direct the laser to the back of the cantilever working electrode, yet
allow the beam reflected from the cantilever to reach the position
sensing detector ͑PSD͒. The incident and reflected beams were ini-
tially coincident. Two mirrors were placed in the path of the re-
flected laser beam in order to increase the optical lever. A duo-lateral
PSD with dimensions 20 ϫ 20 mm ͑DLS-20 from UTD Sensors,
Inc.͒ was used to measure the position of the reflected beam. The
four photocurrents from the PSD were amplified, measured by a
National Instruments A/D card, and transferred to a Macintosh
Power PC computer. The signals were converted into vertical and
horizontal positions on the PSD. The stress calculation utilized only
the vertical position of the laser.
The cantilever was a borosilicate glass slide ͑D 263, Schott͒
measuring 60 ϫ 3 ϫ 0.108 mm. The glass had a Young’s modulus
of 72.9 ϫ 109 N m−2 and a Poisson ratio of 0.208, as specified by
the vendor. To one side of this substrate a 4-nm-thick adhesion layer
of titanium and a 250-nm film of gold were vapor-deposited by
electron-beam evaporation. The glass–metal interface provided the
reflective surface for the laser beam. Prior to use, the electrodes
were boiled in distilled water and then held in a hydrogen flame for
approximately 1 s. The Au electrodes had a strong ͑111͒ crystallo-
graphic orientation. The 200 reflection was not apparent in ␪-2␪
X-ray scans, and rocking curves of the 111 reflection generally
yielded a full width at half-maximum on the order of 2°. The depo-
sition of Al was examined on both ͑111͒-textured Au and Cu elec-
trodes. The Cu electrodes were formed by electrodepositing
0.25 ␮m of Cu from an additive-free copper sulfate electrolyte
͑0.01 mol L−1 CuSO4 in 0.1 mol L−1 H2SO4͒ onto the evaporated
cCertain trade names are mentioned for experimental information only; in no case does
it imply a recommendation or endorsement by the National Institute of Standards and
Technology.
*
Electrochemical Society Active Member.
Electrochemical Society Student Member.
**
z E-mail: gery.stafford@nist.gov
Downloaded on 2015-01-28 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C208
Journal of The Electrochemical Society, 153 ͑4͒ C207-C212 ͑2006͒
Applied Research Corp. ͑PARC͒ model 273 potentiostat/galvanostat
that was controlled by a Macintosh Power PC computer and Lab-
View software.
We examined both the changes in surface stress of the Au canti-
lever in response to changes in surface chemistry in the double layer
and aluminum underpotential deposition ͑UPD͒ region as well as the
in-plane stress developed in electrodeposited aluminum films. Be-
cause the Au electrode is on the side away from the laser, compres-
sive stress displaces the cantilever toward the laser while tensile
stress displaces the cantilever away from the laser. The relationship
between the force per cantilever beam width, Fw, exerted by pro-
cesses occurring on the electrode surface and the radius of curvature
of the cantilever, is given by Stoney’s equation40
Est2s
Fw
=
= ␴avgtf
͓1͔
f
6͑1 − ␯ ͒R
s
Figure 1. The laser path: ͑1͒ He–Ne laser, ͑2͒ beam splitter, ͑3͒ electro-
chemical cell, ͑4͒ mirrors, and ͑5͒ PSD.
where Es, , and t are the Young’s modulus, Poisson ratio, and
thickness of the glass substrate, respectively, and R is the radius of
vs
s
curvature of the cantilever. In the case where the force on the can-
tilever is due to surface processes, Fw is the surface stress, i.e., the
reversible work per unit area needed to elastically stretch a pre-
existing surface, and has units of J/m2 or N/m. For solid electrodes
in solution, the surface stress is determined by the charge density at
the surface, which in turn is strongly influenced by the electrode
potential and adsorbates on the electrode surface. In the case where
Au electrodes. Examination by X-ray diffraction indicated that the
Cu deposit had retained the ͑111͒ crystallographic orientation of the
Au substrate.
The electrochemical cell was a single-compartment Pyrex cell. A
schematic is shown in Fig. 2. A glass disk was joined to the back of
the cell to allow the cell to be held and positioned by a standard
mirror mount on the optical bench. The three electrodes were held in
place by a segmented aluminum disk. Each of the three segments of
the disk acted as a current collector for one of the electrodes; con-
sequently, they were electrically isolated from each other. After po-
sitioning the electrodes, the disk segments were held together by
plastic screws. When the disk was placed into the collar of the
electrochemical cell, each segment of the disk rested on a Au-plated
tungsten wire, which passed through the wall of the glass cell. The
entire electrochemical cell was then sealed by a Teflon screw cap
onto a compression O-ring. The cap was long enough to contact the
segmented aluminum disk so that slight pressure was maintained
between the disk segments and the tungsten wire feed-throughs in
order to maintain electrical continuity.
The electrolyte was a 55–45 mole ratio of AlCl3-EtMeImCl. It
was prepared and purified using the procedures outlined in Ref. 39.
The counter electrode was an aluminum wire placed parallel to and
in the same solution as the working electrode. The reference elec-
trode was also an aluminum wire placed in the same solution as the
working electrode and positioned as shown in Fig. 2. All potentials
reported here are with respect to the Al reference. The electrolyte
was added to the cell while in a dry box containing less than 2 ppm
oxygen and was then sealed. The cell was then removed from the
dry box, placed in the optical mount, and positioned on the optical
bench. Potential control was maintained using an EG&G Princeton
the force on the cantilever is the result of bulk metal deposition, then
avg
f
Fw is equal to the average biaxial stress in the electrodeposit ͑␴
͒
multiplied by the average deposit thickness ͑tf͒. The film thickness
was calculated from the charge, assuming 100% current efficiency
and uniform current distribution. This simplified version of Stoney’s
equation requires only the elastic properties of the substrate. The
properties of the Au film and the electrodeposit can be ignored if the
combined thickness is less than 5% of the substrate thickness.41 We
have thus limited our experiments to deposit thicknesses less than
1 ␮m. Equation 2 is Stoney’s equation in terms of the PSD output
Es · t2snair dPSD
Fw
=
͓2͔
6͑1 − ␯ ͒2LnelDPSD
s
where dPSD is the vertical coordinate of the reflected laser beam onto
the PSD, DPSD is the distance of the PSD from the electrode, and L
is the length of electrode submerged into the electrolyte down to
where the laser strikes the electrode. Because the electrochemical
cell is filled with electrolyte, a correction must be made to account
for the difference in the refractive index between the electrolyte
inside the cell ͑nel = 1.5͒ and the air outside the cell ͑nair = 1.0͒.
During a measurement Fw is determined from dPSD using Eq. 2. In
the configuration used for these experiments, this apparatus can re-
solve surface stresses on the order of 0.025 N/m. A more detailed
description of the optical bench and stress measurement can be
found in Ref. 42.
Results and Discussion
It has previously been reported that the UPD of Al onto Cu͑111͒
is characterized by two pairs of well-resolved peaks, indicating the
formation and dissolution of two different adsorbed structures.43
High-resolution scanning tunneling microscopy imaging at 0.455 V
revealed an ordered adlayer which was interpreted to be a layer of
AlCl−4 that had been oxidatively adsorbed on the Cu͑111͒ surface
with the face of the tetrahedra adjacent to the copper surface. It was
concluded that the first UPD wave involved the desorption of AlCl4−
from the Cu͑111͒ surface. The second cathodic peak was associated
with a fully discharged monolayer of aluminum on the Cu surface.
Figure 3 shows the linear sweep voltammetry and the surface
stress ͑Fw͒ for the UPD of Al onto ͑111͒-textured Cu. Although the
quality of the voltammetry with respect to peak resolution is inferior
to that reported on single-crystal Cu͑111͒,43 it is clear that the UPD
process involves two distinct electrochemical steps. The surface
stress curve in Fig. 3 is similar to that reported for Cu UPD on
Figure 2. Schematic of electrochemical cell and aluminum disk electrode
holder.
Downloaded on 2015-01-28 to IP 128.6.218.72 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, 153 ͑4͒ C207-C212 ͑2006͒
C209
Figure 3. Linear sweep voltammetry and surface stress associated with the
UPD of Al onto ͑111͒-textured Cu in 55–45 mole ratio of AlCl3-EtMeImCl.
Sweep rate was 20 mV/s.
Au͑111͒ in aqueous sulfate electrolyte,29,31 where the change in sur-
face stress in the tensile ͑positive͒ direction as the potential was
swept cathodically was attributed to the desorption of sulfate from
the electrode surface. A simple surface-induced charge redistribution
model has been used to explain the stress behavior in both the sul-
fate adsorption and the Cu UPD regions. Ibach23,26 has suggested
that the loss of bonds at a clean metal surface causes an increased
charge density between the remaining surface atoms, thereby in-
creasing their attractive interaction and resulting in a tensile stress at
the surface. This tensile stress, which has been estimated to be
2.7 N m−1 for Au͑111͒,44 has been observed in most free-electron
metals and is a likely driving force for surface reconstruction.28 As
negative charge is removed from the electrode with an increase in
potential the surface stress moves in the negative ͑compressive͒ di-
rection. Likewise, when negative charge is added to the electrode
with a decrease in potential the surface stress moves in the positive
͑tensile͒ direction.
Figure 4. ͑a͒ Linear sweep voltammetry and surface stress associated with
the UPD of Al onto ͑111͒-textured Au in 55–45 mole ratio of AlCl3
-EtMeImCl, ͑b͒ extended down to potential for bulk Al deposition. Sweep
rate was 20 mV/s.
The adsorption of species on the surface can also be expected to
alter the surface stress because the local interaction of each adsor-
bate will alter the bond strength between neighboring atoms on the
surface. Electron donors would be expected to cause tensile stress
because they increase the bond charge density between the underly-
ing metal atoms. Electron acceptors such as adsorbed anions cause
compressive stress because they reduce the electron density in the
metal surface. For example, the adsorption of a sulfate ion increases
the compressive trend in the surface stress while its desorption
moves the surface stress in the tensile direction. In the present case,
anion adsorption ͑presumably AlCl−4͒ at more positive potentials
causes a compressive shift in the surface stress of the Cu electrode.
As the potential is swept cathodically and AlCl−4 is desorbed from
the Cu surface, the stress moves in the tensile direction, which is
consistent with the charge-induced surface stress model described
above.
The formation of the complete Al monolayer results in a com-
pressive change in the surface stress. Because the aluminum atom is
12% larger than copper, the Al monolayer is expected to be incom-
mensurate with respect to the copper surface. As a consequence,
misfit stress is not expected to play a role in the stress development.
Al is likely to have a more negative potential of zero charge than
Cu; thus, AlCl−4 may readsorb onto the electrode surface once the
aluminum monolayer has formed. This would move the surface
stress in a negative ͑compressive͒ direction. In summary, we con-
clude that the stress response at positive potentials reflects the
adsorption/desorption of AlCl−4 from the Cu surface. The compres-
sive stress associated with Al monolayer formation is likely due to
the combination of monolayer formation and readsorption of AlCl4−
onto the newly formed Al surface.
Figure 4a shows the linear sweep voltammetry as well as the
surface stress for the UPD of Al onto ͑111͒-textured Au. The volta-
mmetry shows a single cathodic peak at 0.165 V and two anodic
peaks at 0.24 and 0.45 V. Reports in the literature indicate that
Al–Au alloys can be formed in the UPD region.45-47 The general
consensus in the literature is that the first anodic peak is associated
with dealloying rather than partial removal of the UPD layer. This
conclusion is based on the fact that the height of the first oxidation
wave increases with the time at which the electrode is held at
+0.10 V. Integration of the single cathodic peak in Fig. 4a results in
a charge of about 0.72 mC/cm2, which is consistent with the theo-
retical value of 0.66 mC/cm2 for a complete monolayer of Al on Au.
This suggests that alloying can be minimized if the potential is sim-
ply swept through the UPD region and not held for a time to allow
alloying to occur.
When the potential of the Au electrode is decreased from
0.8 to 0.3 V, the surface stress ͑Fw͒ decreases slightly in the com-
pressive direction. This is somewhat surprising because the charge
distribution model for surface stress predicts a tensile stress as elec-
trons are added to the electrode. Further, any anion desorption pro-
cess that might occur is also expected to produce a tensile stress. A
possible explanation for the decrease in surface stress in this poten-
tial region is that the Au͑111͒ surface is undergoing surface recon-
struction. As the potential is reduced further, a tensile stress of about
1.2 N/m is associated with the cathodic UPD process at 0.165 V.
The stress continues to increase until the first anodic wave occurs,
after which the stress returns to a value somewhat higher than its
original value.
Downloaded on 2015-01-28 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C210
Journal of The Electrochemical Society, 153 ͑4͒ C207-C212 ͑2006͒
One possible source of this tensile stress in the Al UPD region is
the small misfit between the Al and Au lattice; the Al lattice is only
0.74% smaller than that of Au. If the Al atoms occupy the three-fold
sites of the Au surface, then the Al monolayer will be under a slight
tensile stress. One can estimate this stress, assuming that continuum
elasticity properly describes the Al monolayer. The surface stress
caused by the misfit is described in Ref. 26 by
Y
͑111͒
␶ =
␧
mf d
␪
͓3͔
͑111͒
1 − ␷
͑111͒
where Y
is Young’s modulus for single-crystal Al, ␷
is the
͑111͒
͑111͒
Poisson ratio, ␧ is the misfit strain, d
is the height of the
͑111͒
mf
monolayer, and ␪ is the number of monolayers on the surface. The
elastic constants, calculated from the elastic compliances for Al,48
are Y
= 72 GPa and ␷
= 0.365. Inserting these values into
͑111͒
͑111͒
Eq. 3 for a monolayer of aluminum results in a misfit stress of
+0.24 N/m, which is about 20% of that measured experimentally.
The fact that the cantilever position does not immediately return
to its original value suggests that some alloying occurs between Al
and Au. However, the cantilever will return to its original position if
left for several minutes at a potential of 0.8 V. The formation of
Al–Au alloys has been examined in some detail at higher tempera-
tures in AlCl3–NaCl electrolyte.45 The presence of both AlAu2 and
Al2Au5 has been confirmed after 2 h at 250°C. An Arrhenius plot of
the high-temperature data suggests that at 25°C an Al–Au diffusion
zone of 0.27 nm will form after 60 s. The stress associated with
alloy formation can be estimated from the molar volumes of the
constituents and reaction product
Figure 5. Chronoamperograms for Al deposition onto ͑111͒-textured Au in
55–45 mole ratio of AlCl3-EtMeImCl. The potential was stepped from
+0.3 V to
a deposition potential ranging from −0.05 to −0.25 V
in 0.025-V increments.
the traditional way to shed light on the nucleation kinetics.49 The
data showed a reasonable fit to the theoretical curve for instanta-
neous nucleation.
Figure 6a shows the Fw associated with the bulk deposition of
aluminum for a film thickness up to 260 nm. The negative Fw indi-
cates that the Al films develop a compressive stress in the earliest
stages of growth. The incremental film stress, given by the slope of
the Fw—thickness curve, is also seen to increase in the negative
direction with increasing deposition overpotential. At a given de-
posit thickness that is dependent on the electrode potential, the can-
tilever Fw begins to move in the tensile direction until a net tensile
stress is established. Figure 6b shows the average film stress ͑Fw
divided by the film thickness͒ obtained from the Fw—thickness
curves in Fig. 6a. As the overpotential is increased, the compressive
stresses in the early stages of deposition increase from −50 to
− 120 MPa. The stress then becomes tensile and both the thickness
at which the transition occurs and the magnitude of the tensile ex-
cursion are a function of deposition potential. A maximum tensile
stress of 70 MPa was observed in the film deposited at −300 mV.
The growth stresses associated with Volmer–Weber or island
growth from the gas phase have been divided into three distinct
modes:2-6 low-mobility columnar growth, epitaxial growth, and
high-mobility island growth. The low-mobility columnar films de-
velop tensile stresses, epitaxial Volmer–Weber films are completely
compressive, while high-mobility films show a tensile to compres-
sive transition as the film thickens. The general observation in the
latter case is that the stress progresses from compressive to tensile
and then back to compressive. This has been referred to as CTC
behavior. The initial compressive stress occurs in the discrete-nuclei
stage of growth and is due to the capillarity stress of these small
particles. The rapid development of tensile stress is associated with
nuclei coalescence and grain boundary formation while the final
compressive stage occurs during thickening of the continuous film.
The nucleation transients in Fig. 5 indicate that the aluminum
films grow by three-dimensional nucleation and diffusion-controlled
growth. As a consequence, one would expect to observe a compres-
sive to tensile transition in the stress transient. As noted in the lit-
erature, the compressive stress in the earliest stages of deposition is
due to the capillarity stress associated with small particles. This is
similar to the pressure inside a small drop of liquid which is related
to the drop radius by the Laplace equation. A similar treatment can
be used to characterize the strain in a solid particle. For a spherical
solid particle
⌬V
␴
= −EЈ
͓4͔
imc
imc 3V
where EiЈmc is the biaxial modulus of the intermetallic and ⌬V is the
change in volume when the intermetallic forms. The molar volumes
for Al and Au are 9.99 and 10.2 cm3/mol, respectively, whereas the
molar volume for orthorhombic AlAu2 is 9.553 cm3/mol. If one
assumes an EЈimc of 200 GPa and includes the ⌬V/V of −0.056, then
alloying will produce a force on the cantilever of +3.7 N/m for each
nanometer of intermetallic formed. The 0.27 nm of AlAu2 formed in
60 s, estimated from the high-temperature kinetic data, would result
in a surface stress of +1.0 N/m, which is close to that observed
experimentally. It is clear that both surface alloying and lattice misfit
can account for the tensile stress observed during Al UPD on Au.
Figure 4b shows that when the potential is decreased to the point
of bulk Al deposition, the stress moves in the compressive direction.
When the bulk Al is electrochemically stripped, the stress moves in
the tensile direction, reflecting the condition established by the first
monolayer. The changes in stress during dissolution are far more
sluggish than those during deposition. This may be another indica-
tion that alloying is taking place. When the potential reaches
800 mV, there is a net tensile stress of about 0.5 N/m, again reflect-
ing the alloying between Al and Au. With time at elevated potential,
the stress returns to its original starting value of zero, suggesting that
the surface alloy does not permanently alter the Au cantilever elec-
trode.
The intrinsic stress associated with the bulk deposition of alumi-
num was examined by pulsing the potential of the cantilever elec-
trode to a series of cathodic values. Figure 5 shows the chrono-
amperograms associated with those potential pulses. The transients
show the current increase and Cottrell relaxation that is typical of
three-dimensional nucleation as the spherical diffusion fields overlap
into a single linear diffusion front. One can conclude that aluminum
deposition occurs by three-dimensional nucleation followed by
diffusion-controlled growth. This suggests that the growth follows
Stranski–Krastanov growth where the initial wetting layer trans-
forms to island growth as the deposit thickens. The nucleation chro-
noamperograms were also normalized to the current/time maxima in
Downloaded on 2015-01-28 to IP 128.6.218.72 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, 153 ͑4͒ C207-C212 ͑2006͒
C211
Figure 7. Nuclei strain calculated from Eq. 7 using the maximum compres-
sive stresses taken from the stress-thickness transients ͑as in Fig. 6b͒. The
data points are an average of three independent measurements. The error bars
represent the standard deviation of those measurements.
value and the nuclei develop a positive stress-free strain. An in-
plane, biaxial compressive stress ͑␴͒ is generated that is related to
the strain in the particle ͑␧͒ by
␴ = −E ␧
͓7͔
Ј
where E is the biaxial modulus. Figure 7 is a plot of the strain that
Ј
was calculated from the maximum compressive stresses taken from
stress-thickness transients similar to those shown in Fig. 6b. A value
of 113 GPa was used for the biaxial modulus ͑E ͒ of aluminum.
Ј
This should be viewed as an estimate of the strain since a uniform
deposit thickness was assumed in calculating the average film stress.
The strain is seen to increase with increasing deposition overpoten-
tial. The apparent decrease in strain at the highest overpotentials is
most likely the result of nuclei coalescence that occurs before the
full strain is established in the nuclei ͑see Fig. 6b͒. The 0.05–0.1%
strains that were calculated from the compressive stresses are con-
sistent with values that might be expected due to capillarity. A more
detailed description of the film stress induced by surface stresses for
the growth of cylindrical islands has been derived in the
literature.3,53
The stress curves can also be used to estimate the nucleation
density at the point of coalescence if one assumes that the minimum
in the stress-thickness curve occurs when the nuclei first touch and
begin to form grain boundaries.53 In other words, the capillarity
stress remains constant while tensile stress due to coalescence be-
gins to move the average film stress in the positive direction. If it is
assumed that the nuclei are hemispherical in shape, of equal size,
and are distributed to form a square grid, then the nucleation density
and nuclei radius can be calculated from Eq. 8 where N is the nucle-
ation density and t is the average deposit thickness
Figure 6. ͑a͒ Force per cantilever beam width, Fw, and ͑b͒ the average film
stress, ␴avg, plotted as a function of deposit thickness ͑up to 260 nm͒ for Al
f
deposited onto ͑111͒-textured Au in 55–45 mole ratio of AlCl3-EtMeImCl,
as a function of deposition potential.
2f
⌬P = −
͓5͔
r
where f is the surface stress, r is the particle radius, and ⌬P is the
pressure difference. Using the definition of the bulk modulus of a
solid, K = −⌬PV/⌬V where V is the volume, and assuming that the
nuclei are crystalline and have a cubic structure with lattice param-
eter a, the equilibrium lattice parameter as a function of the distance
r from the particle center is2
2f
a͑r͒ = ao 1 −
͓6͔
ꢀ
ꢁ
3Kr
0.068
t2
0.5
N1/2
N =
,
r =
͓8͔
where K is the bulk modulus of the particle and ao is the equilibrium
lattice parameter of the bulk material. The atomic distances associ-
ated with small particles are compressed when compared to the
equilibrium parameter of the bulk material. In fact, the earliest sur-
face stress determinations were made from lattice parameter mea-
surements of small metallic particles, typically less than 15 nm in
diameter.50-52 It is important to note that these small particles are in
equilibrium and do not exert a stress on the cantilever substrate
͑other than a possible misfit stress͒. The basic mechanism for stress
generation is associated with particle growth. When the particles
nucleate on the surface, they have a lattice parameter that is dictated
by the particle radius. As the particle grows, the equilibrium lattice
parameter increases. Once the nuclei become rigidly attached to the
substrate, the lattice parameter cannot adjust to the equilibrium
Figure 8 shows a plot of both the nucleation density and nuclei
radius at coalescence as a function of the deposition overpotential
from the nominal deposit thickness taken from the stress minima in
Fig. 6b. The nucleation density varies from 108 to 1010 cm−2, which
is consistent with values that are typically measured microscopically
on electrodeposited films.49,54 The nucleation density is also seen to
increase with deposition overpotential, which is consistent with
electrocrystallization theory and experimental observation reported
in the literature. Similarly, the radius of the nuclei, calculated from
the nucleation densities, range in size from 60 to 500 nm. At coa-
lescence, this would result in aluminum grains on the order of
0.1–1 ␮m diam, which are very realistic values.
Downloaded on 2015-01-28 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C212
Journal of The Electrochemical Society, 153 ͑4͒ C207-C212 ͑2006͒
5. W. D. Nix and B. M. Clemens, J. Mater. Res., 14, 3467 ͑1999͒.
6. R. Abermann, Vacuum, 41, 1279 ͑1990͒.
7. G. R. Stafford and C. L. Hussey, in Advances in Electrochemical Science and
Engineering, Vol., 7, R. C. Alkire and D. M. Kolb, Editors, p. 275, Wiley-VCH
Verlag GmbH, Weinheim, Germany ͑2002͒.
8. T. P. Moffat, J. Electrochem. Soc., 141, L115 ͑1994͒.
9. M. R. Ali, A. Nishikata, and T. Tsuru, Electrochim. Acta, 42, 2347 ͑1997͒.
10. G. R. Stafford, J. Electrochem. Soc., 136, 635 ͑1989͒.
11. T. P. Moffat, G. R. Stafford, and D. E. Hall, J. Electrochem. Soc., 140, 2779
͑1993͒.
12. G. M. Janowski and G. R. Stafford, Metall. Trans. A, 23A, 2715 ͑1992͒.
13. G. R. Stafford, J. Electrochem. Soc., 141, 245 ͑1994͒.
14. T. Takenaka, A. Hoshikawa, and M. Kawakami, in International Symposium on
Molten Salt Chemistry and Technology-1993, PV 93-9, p. 184, M.-L. Saboungi and
H. Kojima, Editors, The Electrochemical Society Proceedings Series, Pennington,
New Jersey ͑1993͒.
15. T. Tsuda, C. L. Hussey, G. R. Stafford, and J. E. Bonevich, J. Electrochem. Soc.,
150, C234 ͑2003͒.
16. G. R. Stafford, T. Tsuda, and C. L. Hussey, J. Min. Metall., B, 39, 23 ͑2003͒.
17. J. Scarminio, S. N. Sahu, and F. Decker, J. Phys. E, 22, 755 ͑1989͒.
18. J. M. Rosolen and F. Decker, J. Electrochem. Soc., 143, 2417 ͑1996͒.
19. G. G. Lang and M. Seo, J. Electroanal. Chem., 490, 98 ͑2000͒.
20. K. Y. Chung and K.-B. Kim, J. Electrochem. Soc., 149, A79 ͑2002͒.
21. M. Seo and Y. Serizawa, J. Electrochem. Soc., 150, E472 ͑2003͒.
22. L. Jaeckel, G. Lang, and K. E. Heusler, Electrochim. Acta, 39, 1031 ͑1994͒.
23. H. Ibach, J. Vac. Sci. Technol. A, A12, 2240 ͑1994͒.
Figure 8. Plots of nucleation density and particle radius at coalescence,
calculated from the thickness taken at maximum compressive stress in Fig.
6b, as a function of deposition overpotential. The data points are an average
of four independent measurements. The error bars represent the standard
deviation of those measurements.
24. R. Raiteri and H.-J. Butt, J. Phys. Chem., 99, 15728 ͑1995͒.
25. T. A. Brunt, E. D. Chabala, T. Rayment, S. J. O’Shea, and M. E. Welland, J. Chem.
Soc., Faraday Trans., 92, 3807 ͑1996͒.
Conclusion
26. H. Ibach, Surf. Sci. Rep., 29, 193 ͑1997͒.
27. W. Haiss, R. J. Nichols, J. K. Sass, and K. P. Charle, J. Electroanal. Chem., 452,
199 ͑1998͒.
28. W. Haiss, Rep. Prog. Phys., 64, 591 ͑2001͒.
29. W. Haiss and J. K. Sass, J. Electroanal. Chem., 386, 267 ͑1995͒.
30. T. A. Brunt, T. Rayment, S. J. O’Shea, and M. E. Welland, Langmuir, 12, 5942
͑1996͒.
31. W. Haiss and J. K. Sass, Langmuir, 12, 4311 ͑1996͒.
32. W. Haiss and J. K. Sass, J. Electroanal. Chem., 410, 119 ͑1996͒.
33. C. Friesen, N. Dimitrov, R. C. Cammarata, and K. Sieradzki, Langmuir, 17, 807
͑2001͒.
34. S. N. Sahu, J. Scarminio, and F. Decker, J. Electrochem. Soc., 137, 1150 ͑1990͒.
35. R. Weil, Plating, 58, 137 ͑1971͒.
36. M. A. Butler and D. S. Ginley, J. Electrochem. Soc., 134, 510 ͑1987͒.
37. H. Feigenbaum and R. Weil, J. Electrochem. Soc., 126, 2085 ͑1979͒.
38. W. Haiss, R. J. Nichols, and J. K. Sass, Surf. Sci., 388, 141 ͑1997͒.
39. J. S. Wilkes, J. A. Levisky, R. A. Wilson, and C. L. Hussey, Inorg. Chem., 21,
1263 ͑1982͒.
40. G. G. Stoney, Proc. R. Soc. London, Ser. A, 82, 172 ͑1909͒.
41. G. Moulard, G. Contoux, G. Motyl, G. Gardet, and M. Courbon, J. Vac. Sci.
Technol. A, 16, 736 ͑1998͒.
42. O. E. Kongstein, U. Bertocci, and G. R. Stafford, J. Electrochem. Soc., 152, C116
͑2005͒.
43. G. R. Stafford, V. D. Jovic, T. P. Moffat, Q. Zhu, S. Jones, and C. L. Hussey, in
Twelfth International Symposium on Molten Salts, PV 99-41, p. 535, P. C. Trulove,
H. C. De Long, G. R. Stafford, and S. Deki, Editors, The Electrochemical Society
Proceedings Series, Pennington, New Jersey ͑1999͒.
44. V. Fiorentini, M. Methfessel, and M. Scheffler, Phys. Rev. Lett., 71, 1051 ͑1993͒.
45. B. S. Radovic, R. A. H. Edwards, and J. N. Jovicevic, J. Electroanal. Chem., 428,
113 ͑1997͒.
46. C. A. Zell, F. Endres, and W. Freyland, Phys. Chem. Chem. Phys., 1, 697 ͑1999͒.
47. J.-J. Lee, I. T. Bae, D. A. Scherson, B. Miller, and K. A. Wheeler, J. Electrochem.
Soc., 147, 562 ͑2000͒.
48. G. Simmons and H. Wang, Single Crystal Elastic Constants and Calculated Ag-
gregate Properties: A Handbook, The M.I.T. Press, Cambridge, MA, ͑1971͒.
49. B. Scharifker and G. Hills, Electrochim. Acta, 28, 879 ͑1983͒.
50. C. W. Mays, J. S. Vermaak, and D. Kuhlmann-Wilsdorf, Surf. Sci., 12, 134 ͑1968͒.
51. H. J. Wasserman and J. S. Vermaak, Surf. Sci., 22, 164 ͑1970͒.
52. H. J. Wasserman and J. S. Vermaak, Surf. Sci., 32, 168 ͑1972͒.
53. R. C. Cammarata, T. M. Trimble, and D. J. Srolovitz, J. Mater. Res., 15, 2468
͑2000͒.
In situ stress measurements were made during aluminum elec-
trodeposition from AlCl3-EtMeImCl using the wafer curvature
method. The UPD of Al on ͑111͒-textured Cu shows a stress re-
sponse that is consistent with a two-step process where the first step
involves the desorption of AlCl4−. This gives rise to a tensile shift in
the surface stress. The tensile stress is eliminated when the full
incommensurate monolayer of aluminum is formed. The situation is
very different on ͑111͒-textured Au. In this case a compressive shift
is observed as the potential is swept cathodically in the potential
region positive of aluminum UPD. This is inconsistent with charge
distribution models that appear in the literature. The full Al mono-
layer results in a tensile stress that can be attributed to both lattice
misfit and Al–Au alloying. The bulk deposition of Al on ͑111͒-
textured Au appears to be governed by Stranski–Krastanov three-
dimensional growth. Initially a compressive stress is observed that
we attribute to capillarity. This is followed by a rapid increase in
tensile stress that we attribute to nuclei coalescence and grain
boundary formation. In this region of film formation, the total stress
is the superposition of both stress mechanisms and each shows a
dependence on the deposition potential.
Acknowledgments
The authors gratefully acknowledge the technical contributions
of Thomas Moffat, Jonathon Guyer, Charles Hussey, and Vladimir
Jovic.
National Institute of Standards and Technology assisted in meeting the
publication costs of this article.
References
1. R. Weil, Plating, 58, 50 ͑1971͒.
2. R. Koch, J. Phys.: Condens. Matter, 6, 9519 ͑1994͒.
3. J. A. Floro, E. Chason, R. C. Cammarata, and D. J. Srolovitz, MRS Bull., 27, 19
͑2002͒.
4. A. L. Shull and F. Spaepen, J. Appl. Phys., 80, 6243 ͑1996͒.
54. A. Radisic, A. C. West, and P. C. Searson, J. Electrochem. Soc., 149, C94 ͑2002͒.
Downloaded on 2015-01-28 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).