Electrodeposited gallium alloy thin films synthesized by solid state reactions for CIGS solar cell

Article abstract of DOI:10.1149/2.096202jes

The solid-state reactions of electroplated gallium thin films with copper, gold and molybdenum were investigated using XRD, SIMS, TEMEDS and linear strip voltammetry. Due to the low melting temperature of gallium metal, the activation energy for gallium diffusion is lower than the other metals used. Hence Ga diffuses faster and reacts to form alloy phases with the other metals investigated. In a thin film stack of Ga with another metals (Me), (Me = Cu, Au and Mo), equilibrium compounds are formed, which compounds form depends on the relative composition and diffusion time. If there is excess Ga, or excess Me, it can further react with the metastable compounds to form an equilibrium compound. In addition, TaN was found to act as an effective barrier layer to gallium diffusion.

Full text of DOI:10.1149/2.096202jes

Journal of The Electrochemical Society, 159 (2) D129-D134 (2012)
D129
0013-4651/2012/159(2)/D129/6/$28.00 © The Electrochemical Society
Electrodeposited Gallium Alloy Thin Films Synthesized by Solid
State Reactions for CIGS Solar Cell
S. Ahmed,a,∗,z K. B. Reuter,a Q. Huang,a,∗ H. Deligianni,a,∗ L. T. Romankiw,a,∗∗ S. Jaime,b
and P. P. Grandb
aIBM T. J. Watson Research Center, Yorktown Heights, New York 10598, USA
bNexcis Photovoltaic Technology, Rousset, France
The solid-state reactions of electroplated gallium thin films with copper, gold and molybdenum were investigated using XRD,
SIMS, TEM/EDS and linear strip voltammetry. Due to the low melting temperature of gallium metal, the activation energy for
gallium diffusion is lower than the other metals used. Hence Ga diffuses faster and reacts to form alloy phases with the other metals
investigated. In a thin film stack of Ga with another metals (Me), (Me = Cu, Au and Mo), equilibrium compounds are formed, which
compounds form depends on the relative composition and diffusion time. If there is excess Ga, or excess Me, it can further react
with the metastable compounds to form an equilibrium compound. In addition, TaN was found to act as an effective barrier layer to
gallium diffusion.
© 2011 The Electrochemical Society. [DOI: 10.1149/2.096202jes] All rights reserved.
Manuscript submitted February 23, 2011; revised manuscript received November 14, 2011. Published December 22, 2011. This was
Paper 1642 presented at the Las Vegas, Nevada, Meeting of the Society, October 10–15, 2011.
Gallium is an element from the group IIIA. It has been used in
the semiconductor and the electronics industries due to its unique ex-
ploitable physical properties e.g. electrical and thermal properties.1–3
There are many applications of gallium in optoelectronics (LED’s),
telecommunications, aerospace and many other technology products.
There has also been a high demand for gallium as a thermal inter-
face material. It is used as highly thermally conducting contact layer
between silicon chips and metallic cooling plates.4–6 Gallium forms
compounds in various combinations with aluminum, indium, phos-
phorus, arsenic, and antimony, (e.g., gallium arsenide and indium
gallium arsenide phosphide) with valuable semiconductor and opto-
electronic properties.7,8 Some of these compounds form the basis for
electronic devices, for example as light-emitting diodes and semicon-
ductor lasers.
It has recently been reported that gallium plays an important role
in achieving high efficiency solar cells if it is incorporated in CuInSe2
as a partial replacement of indium to form CuInGaSe2 (CIGS).9–14
CIGS based solar cells have been considered one of the most promis-
ing thin film solar cell technologies. The absorber layer band gap
increases from 1.02 eV for CuInSe2 to 1.1–1.2 eV by a partial sub-
stitution of In with Ga, leading to substantial increase in efficiency.9
However, gallium has a low melting point of 29.78◦C and thus readily
diffuses to form an alloy with other metals e.g. Cu, In, Au, Sn, Zn etc.
already at room temperature.15–19 In photovoltaic applications, e.g.
in CIGS fabrication, Ga can be incorporated in the film using both
vaccum (PVD)9–11 and electrochemical methods.20–23 Gallium is gen-
erally considered a difficult metal to electrodeposit without excessive
hydrogen generation at the cathode. The electrode potential required
for gallium electrodeposition overlaps with the hydrogen evolution
overpotential. As a result the hydrogen evolution causes low plating
efficiency and porosity in the film. Surface roughness and microde-
fects in the film can also become very high due to the hydrogen
evolution.
Experimental
A conventional three electrode configuration with a platinum mesh
counter electrode and a Hg|Hg2Cl2 (SCE) reference electrode, to
which all potentials are referenced, was used. Different substrates
(working electrodes) were used in a rotating disk electrode (RDE)
system (Pine Research Instrumentation, USA). In all the electroplat-
ing and voltammetry experiments, the working electrode was rotated
at 540 rpm. The working electrodes consist of (i) a 6 mm diameter gold
disk electrode (Pine Research Instrumentation, USA), (ii) PVD copper
and (iii) gold seed layers on a glass/Mo and Si(100)/TaN respectively.
Except for the standard gold disk electrode, all other substrates were
electrically connected through a contact on to the front surface using
spring loaded pins.
The electrolyte for gallium plating consisted of 0.25 mol dm−3
Ga3+ and 0.5 M sodium citrate of pH ≥ 10. The pH was adjusted
using a NaOH solution. Deionized water (≥18 Mꢀ-cm) was used for
all solution preparation and rinsing. For both cyclic and linear sweep
voltammetry, the potential of the working electrode was controlled
using a CHI 660D potentiostate (CH Instruments, Austin, TX, USA).
Gallium thin films were electroplated galvanostatically using an AU-
TOLAB Galvanostat/Potentiostat with 10 mAcm−2 current density
at room temperature which varied between 19–23◦C. These electro-
plated films were self-annealed at room temperature for a different
length of time and then characterized using linear sweep voltammetry
(LSV), XRD and microscopy. For the electrochemical identification
of the Ga-Me intermetallics LSV was carried out over a potential
range of +2V to −2V (SCE) at 0.05Vs−1 in 0.5M NaOH at different
intervals.
The morphology and microstructure of the deposited film were
characterized using a Zeiss Leo Scanning Electron Microscope
(SEM). A Philips X-ray powder diffractometer (XRD) with a Cu
x-ray source (Kα = 1.54 Å) was used for phase identification of the
Ga-alloy with Cu, Au and Mo. X-ray fluorescence (XRF) was used
for thickness measurement of the film. A standard sample for XRF
was made and the composition was measured by RBS. A FEI SB
200 Focus Ion Beam (FIB) station was used to prepare samples for
cross-sectional SEM and TEM. A JEOL 3000F Transmission Elec-
tron Microscope (TEM) equipped with an Energy Dispersive X-ray
Spectrometry (EDS) was used for microstructure and chemical com-
position analysis of the deposit.
The efficiency of CIGS solar cell, in addition to the other factors,
is also dependent on the solid state reaction and stability of Ga in
the CIGS layer.24 In this study, we have investigated electroplated
Ga thin films and the alloying of gallium with other elements at room
temperature during self-annealing. Stable compounds of gallium form
with copper and gold.
Results and Discussion
∗
Electrochemical Society Active Member.
Electrochemical Society Fellow.
∗∗
Fig. 1 shows a typical cyclic volammogram for the electrodepo-
sition of gallium on gold in 0.25 mol dm−3 Ga3+ and 0.5 mol dm−3
z E-mail: shafaat.ibm@gmail.com
Downloaded on 2014-06-19 to IP 155.198.30.43 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D130
Journal of The Electrochemical Society, 159 (2) D129-D134 (2012)
A thin film, especially an electroplated film, has large number of
defects, e.g. dislocations, vacancies and grain boundaries. These de-
fects can act as a rapid diffusion path. Inter-diffusion at the interface
of the two films is possible even at low temperatures due to the pres-
ence of these defects. The formation of compounds is presumably the
result of both rapid diffusion and lattice diffusion. It has been shown
that the inter-diffusion coefficient is highest for the lowest melting
Figure 1. Cyclic voltmmogram of a polished gold substrate in a Ga bath
consisting of 0.25 mol dm−3 Ga3+ and 0.5 mol dm−3 sodium citrate with pH
≥ 10 at a scan rate 0.05 Vs−1 from +0.5 V to −2.0 V (SCE). The scaled up I-V
in the inset clearly show the onset of cathodic reduction at −0.8 V vs. SCE.
sodium citrate at pH = 13. The potential was scanned from 0.5V to
−2.0 V (SCE) at 0.05 Vs−1. As the potential is scanned in the negative
direction a cathodic current corresponding to the electrodeposition of
gallium was observed at −0.8V (SCE). Initially the current is con-
trolled by reaction kinetics but as the potential is increased further a
local maximum in current (∼11 mA cm−2) is reached at a −1.6 V.
This is due to mass transport control, i.e. diffusion of the Ga-ions to
the electrode surface becomes rate limiting. As the scan continues to
a potential more negative than − 1.65 V, the magnitude of the current
decreases. This is due to the depletion of the Ga-ions in the region
adjacent to the electrode. The current eventually reaches a plateau at
10.5 mA cm-2 when the thickness of the depletion layer (i.e. the re-
gion adjacent to the electrode which is depleted of Ga ions) becomes
approximately constant. As the potential is made more negative, an
increase in current is again observed due to the hydrogen gas evolution
2H + + 2e− → H2.
The potential scan is reversed at −2.0V. At −1.5 V the rest potential
is reached and no further electrodeposition of gallium occurs. As the
potential is scanned to values more positive than the rest potential,
the current becomes anodic and the gallium begins to be oxidized
according to the reaction Ga →Ga3+ + 3e −. The current increases
to a peak value and eventually decreases because of depletion of the
gallium on the gold surface.
To study the solid state reaction of Ga with copper and gold, a 100–
150 nm Ga film were galvanostatically deposited. The substrate was
silicon with Mo/TaN/Ta/Cu and TaN/Au deposited by physical vapor
deposition. The Ga-deposition was carried out using a bath containing
0.25 M Ga3+ and 0.5 M sodium citrate with pH ≥ 10 using 10 mAcm−2
current density. Fig. 2a and 2b shows the topography and cross-section
images of a typical ∼130 nm Gallium on Si/Mo/TaN/Ta/Cu substrate
respectively. The surface topography of electroplated gallium on cop-
per shows particle type features (Fig. 2a). The dimensions of these
features are in the range of 50–200 nm which is further magnified
and shown in the inset of Fig. 2a. These particle type features on the
surface indeed caused the film to be rougher which can be seen in the
cross-section micrograph (Fig 2b). The surface undulation and these
particles, marked by an arrow, can also be seen in the cross-section
(Fig. 2b). Fig. 2c shows an x-ray fluorescent profile (energy, eV vs.
count) of the sample where 8.04 and 9.26 eV corresponds to Cu-Ka
and Ga-Ka respectively. These films were aged at room temperature
for different lengths of time and then characterized by XRD and LSV.
The XRD data was then correlated with the characteristics anodic
peaks obtained by LSV. The LSV was used as an electrochemical
spectroscopy method for detecting gallium alloy phases.
Figure 2. (a) Topdown SEM micrographs of ∼130 nm gallium thin films
electroplated on Si/Mo/TaN/Ta/Cu substrate from 0.25 mol dm−3 Ga3+ and
0.5 mol dm−3 sodim citrate with pH ≥ 10 using 10 mAcm−2 current density
at room temperature. Higher magnification image is shown in the inset. (b)
Cross-sectional SEM of the film. (c) A typical XRF profile of the film.
Downloaded on 2014-06-19 to IP 155.198.30.43 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, 159 (2) D129-D134 (2012)
D131
Figure 3. A typical XRD spectrum of Si/Mo/TaN/Ta/Cu substrate.
Figures 3b–3d show XRD spectra of ∼130 nm gallium films stored at room
temperature for 6.84×104 s, 4.34×105 s and 2.59×106 s respectively.
Figure 5. X-ray diffraction spectra of (a) Si/TaN/100 nm gold substrate, (b)
and (c) ∼150 nm electroplated gallium on this substrate after aging at room
temperature between 23–25◦C for 1200 s and 7200 s respectively.
metal. This is in agreement with the finding25 that suggests that the
activation energy of self-diffusion of metals is proportional to their
melting point (Tm): Q = 34 Tm. According to this relationship, the
self diffusion at room temperature requires lower activation energy
than for the higher melting point metals.
Figure 3a shows a typical XRD spectrum of Si/Mo/TaN/Ta/Cu
substrate. Figures 3b–3d shows XRD spectra of ∼130 nm gallium
films stored at room temperature for 6.84×104 s, 4.34×105 s and
2.59×106 s respectively. During this period the films were held and
stored at room temperature (18–20oC). The XRD data (Fig. 3) show
that with aging, the intensity of the Cu-Ga alloys phases, e.g. CuGa
(101) and CuGa (102) at 2θ equals to 35.23 and 44.67 degrees, re-
spectively, increases. However the intensity of the pure copper phases,
e.g. Cu (111), Cu (200) and the gallium intermetallics of CuGa2 (102)
at 2θ equals to 43.47, 50.67 [JCPDS#03-1018], and 46.55 degrees,
[JCPDS # 65–1636] respectively, decreases or disappears (Fig. 3c).
This decrease in the intensity of the Cu and Cu2Ga phases indicates
that the inter-diffusion is still in progress and with further aging time,
the remaining copper will react with the remaining CuGa2 and will
transform to CuGa intermetallic. Indeed, the XRD spectrum obtained
after 2.59×106 s only shows the intensity for the CuGa phase and
there is no intensity for Cu and CuGa2. However, the XRD after af-
ter 5.18×106 s shows no difference compared with the XRD after
2.59×106 s.
As the gallium is a low melting point metal (29.7◦C), a few degrees
variation in room temperature plays a key role in its alloying with
copper. The gallium films stored at room temperature 23–25◦C (∼5◦C
higher than before) show accelerated diffusion and alloy formation
with copper in contrast with the film stored at room temperature of 18–
20oC. Fig. 4 shows x-ray diffraction spectra of ∼130 nm electroplated
gallium on Si/Mo/TaN/Ta/Cu substrate obtained after 1.08×104 s aged
at 23–25◦C. Due to the aging of the gallium film at ∼5◦C higher than
before, the intensity of CuGa phases appeared higher than the CuGa2
phases (Fig. 4) at the end of 1.08×104 s. Whereas the sample aged
Figure 4. X-ray diffraction spectra of (a) Si/Mo/TaN/Ta/Cu substrate and (b)
∼150 nm electroplated gallium on that substrate obtained after 3-hours at
23–25◦C.
Downloaded on 2014-06-19 to IP 155.198.30.43 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D132
Journal of The Electrochemical Society, 159 (2) D129-D134 (2012)
Figure 6. I-V curves obtained by LSV over a potential ranging from −2V
to +2V (SCE) at 0.05Vs−1 in 0.5M NaOH. (a) shows I-V curves obtained
by LSV of Si/Mo/TaN/Ta/100 nm-Au substrate (γ), Si/Mo/TaN/Ta/100 nm-
Au/150 nm-Ga after aging at RT = 23–26◦C for 30 minutes (ψ) and after 120
minutes (λ). (b) shows I-V curves obtained by LSV of Si/Mo/TaN/Ta/100 nm-
Cu substrate (χ), Si/Mo/TaN/Ta/100 nm-Cu/Ga after aging at RT = 23–26◦C
for 3-hours (σ).
∼5◦C lower showed lower intensity of CuGa as compared to the
CuGa2 (Fig. 3) even after 6.84×104 s of aging at the end of plating.
This data suggests that the increase of room temperature closer to
the melting temperature of Ga accelerated its diffusion and alloy
formation with copper at a faster rate (Fig. 4) than the sample that was
stored at lower temperature (Fig. 3).
Figure 7. A cross-sectional SEM image (a) and x-ray spectrum (b) of elec-
troplated ∼ 100 nm gallium on glass/Mo/Cu substrate after 7 days.
For a given composition of copper and gallium in the layers, the
following solid state reaction can progressively form.
is formed by the reaction of the second compound with the excess of
the higher melting metal and so on. However the Au-Al and Au-Sn
systems are an exception and do not follow the same rule.26
The study of the Au-Ga interdiffusion and alloying during room
temperature aging was also carried out to understand the activity
of gallium with the different metal system. Fig. 5 shows an x-ray
diffraction spectrum of ∼150 nm electroplated gallium on Si/TaN/100
nm-Au substrate after aging at room temperature of 23–25◦C for (a)
1200 s and (b) 7200 s. After 1200 s both the AuGa2 and GaAu phases
have formed and the amount of gold phases has been decreasing.
However, at the end of 7200 s of aging at that room temperature, the
intensity of AuGa2 started to decrease and the GaAu phase plus a new
gallium-gold phase GaAu2 dominate in the spectrum. In addition, over
these 7200 s periods the intensity of the Au (111), Au (311) and Au
(222) becomes zero.
Cu + 2Ga → CuGa2
CuGa2 + Cu → 2CuGa
2CuGa2 + 6Cu → 4Cu2Ga
The stochiometry of the resulting compound depends on the ratio
of the concentration of both copper and gallium and the time of diffu-
sion. Marinkovic et al.15 showed that when two or three compounds are
formed, the inter-diffusion co-efficient of the first compound, which
is formed directly from the starting metals, is about two orders of
magnitude higher than the second compound, which is formed by the
reaction of the first compound with the excess of the high melting
metals. Similarly, the interdiffusion coefficient of the second com-
pound is two order magnitudes larger than the third compound, which
Downloaded on 2014-06-19 to IP 155.198.30.43 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, 159 (2) D129-D134 (2012)
D133
Figure 8. (a) SIMS profiles from the sample of glass/Mo/40 nm-Cu/100 nm-Ga. (b) Annular dark field image, arrow indicating the surface, corresponds to arrow
in (c). (c) EDS profile taken along the line shown in (b); (d) Bright field image labeled with the compositions measured by EDS.
This data suggests that there is no un-reacted gold in the film
and all the gold in the films must have been alloyed with gallium at
the end of a 2-hour of self-annealing. It can be seen that the inter-
diffusion and phase formation of gold-gallium system was faster than
the copper-gallium system. These data well agreed with previous
findings in the literature15 as the diffusion coefficient of Ga in Au
(1.6 × 10−12 cm2 s−1) in Au-Ga thin films is larger than the diffusion
coefficient of the Ga in the Cu (3.1×10–15 cm2s−1) in Cu-Ga system
at room temperature. This means that the reaction rate of gold with
gallium is higher than with the copper. V. Simicand et al.16 reported
that the interdiffusion at room temperature leads to the formation
of intermetallic binary compounds of gallium and gold. Depending
on the at. wt.% of Ga and Au, composition, and the diffusion time,
the different Au-Ga intermetallic compounds are formed. For a given
composition, the same compounds are formed irrespective of the dif-
fusion medium. They have shown that if Au is coupled with more than
33 wt% of Ga, then the compounds formed after 60–120 s remain un-
changed for 360 days but if the wt% of Ga ≤ 33% then the compounds
formed between 4 days to 3-months have transformed to equilibrium
compounds over the period of time. Our data, however, does not quite
agree with Simicand’s observation. In our case, we used 30–60 wt.%
Ga and observed that at first an intermediate metastable compound
(CuGa2 or AuGa2) is formed by the reaction of Ga with Cu and Au.
These metastable compounds further react with Cu or Au and formed
an equilibrium intermetallic compound (e.g. CuGa or AuGa).
anodic peaks obtained by LSV. These anodic peaks would be the
characteristic peaks for the particular phases or compounds which
are formed during self-annealing. The anodic peaks towards the more
positive potential correspond to more noble alloy compounds.
Fig. 6 show the I-V curves obtained by LSV over a potential
ranging from −2V to +2V (SCE) at 0.05Vs−1 in 0.5M NaOH.
Fig 6a shows I-V curves obtained by LSV of Si/Mo/TaN/Ta/100 nm-
Au substrate (γ), Si/Mo/TaN/Ta/100 nm-Au/150 nm-Ga after aging
at RT = 23–26◦C for 1800 S ψ) and after 7200 minutes (λ). It can
be seen from the I-V curve γ that at the potential more negative than
−1.648 V and more positive than 1.182 V (vs. SCE) the current in-
creases exponentially due to the hydrogen and oxygen evolution re-
spectively and the peak at 1.61 V (SCE) presumably represents the
oxidation of gold. In addition to these characteristics responses of I-V
curve γ , the I-V curve ψ and λ shows 4 and 1 anodic peaks respec-
tively. The anodic peaks in curve ψ at −1.09 V, −0.789 V, 0.609
V and 0.927 V (SCE) are thought to be the oxidation of absorbed
hydrogen on the surface, the stripping of a thin gallium films from
the surface, AuGa2 and AuGa intermetallic compound respectively.
On the other hand, the anodic peaks in curve λ at 0.88 V (SCE) most
likely represents the stripping of the GaAu compound. Gold is a noble
metal and its oxidation can be expected at a more positive potential.
However, the Me intermetallics with gallium are expected to be less
noble than the gold. Depending on the composition of the stripping
potential, Au2Ga is expected to be more noble than AuGa or AuGa2.
Fig 6b shows I-V curves obtained by LSV on Si/Mo/TaN/Ta/100
nm-Cu substrate (χ), Si/Mo/TaN/Ta/100 nm-Cu/Ga after aging at RT
= 23–26◦C for 1.08×104 s (σ). It can be seen from the I-V curve χ
that at the potential more negative than the −1.56 V (SCE) the current
increases exponentially due to the hydrogen evolution. However, the
peak at 0.0 and 0.78 V (SCE) is probably due to the oxidation of
LSV can be used to identify these Cu-Ga and Au-Ga intermetallic
compounds. The time dependent XRD study during self-annealing of
Ga-Me (Me = Cu, Au) showed that depending on the interiffusion
coefficient of the Ga-Me system, the composition and diffusion time,
different intermetallic compounds form. The compounds, which have
been identified by XRD during aging, can be correlated with the
Downloaded on 2014-06-19 to IP 155.198.30.43 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D134
Journal of The Electrochemical Society, 159 (2) D129-D134 (2012)
Cu to Cu2O and CuO respectively. In addition to these characteristics
responses of I-V curve χ , the I-V curve σ shows 5 anodic peaks at
−1.07 V, −0.796 V, 0.03V, 0.78V, 1.21 V (SCE) respectively assumed
to be corresponding to the oxidation of hydrogen, Ga or Ga2O3 strip-
ping, oxidation of remaining Cu to Cu2O and CuO and oxidation or
striping of the CuGa intermetallics.
solid state reaction with Me at a faster rate than the sample stored at
lower temperature.
Our study shows that if at. wt% of Me ꢀ Ga, then an equilibrium
compound of MexGa is formed where x>1. However, if at. wt%
Ga>> Me then the equilibrium compound of GaMe (1:1) is formed
and remaining Ga diffused into the next layer (in this case Mo). We
found that TaN acts as an effective diffusion barrier layer to the gallium
diffusion.
The Ga-intermetallics formed after alloying could be identified
using Linear Sweep Voltammetric Stripping. The anodic stripping
peak of CuGa shift towards the more positive potential (more noble)
but in AuGa system the anodic peaks are shift to a negative (less
noble) potential.
A cross-section SEM image and x-ray spectrum of electroplated ∼
100 nm gallium on glass/Mo/Cu substrate captured after 10.08×104
s at the end of electroplating are shown in Fig 7. The samples were
held at room temperature (17–20◦C) during the aging period. It can
be seen from the XRD data that Ga readily diffuses to the Cu and Mo
layers and forms CuGa, CuGa2 and GaMo3 intermetallics during self-
annealing. The SEM image (Fig. 7a) shows an irregular penetration of
gallium or its alloys with copper into the Mo layers. Sputtered Mo has
a columnar structure and its grain boundaries normal to the substrate
are considered to be a potential path for the diffusion of Ga or its
alloys. As gallium and copper are lower melting point metals than
Mo, their diffusion coefficients are expected to be higher and thus
diffuse to the Mo very readily. However, when Ga was plated on a
substrate containing a TaN barrier layer (Fig. 3) there was no Ga-Mo
alloy detected even after 5.184×106 s aging. Therefore, TaN acts as
a diffusion barrier for gallium. Since major peaks of Mo-Ga phases
overlap with the peaks of Mo and GaCu, it is hard to distinguish
Mo-Ga phases in the XRD spectrum. However SIMS and chemical
analysis at the nanoscale using a TEM/EDS, confirmed that the Ga
diffused into the Mo.
Acknowledgment
The authors gratefully thank Andrew Kellock at the IBM Almaden
Research Center and the Materials Research Laboratory (MRL) at the
IBM Thomas J. Watson Research Center.
References
1. Pierre de la Breteque, Industrial and engineering chemistry, Vol. 56(6), 54 (1964).
2. K. Wade and A. J. Banister, The Chemistry of aluminum, gallium, indium, and
thallium, Pergamon Texts in Inorganic Chemistry Volume 12, Pergamon Press, 1975.
3. Teng Li, Yong-Gang Lv, Jing Lui, and Yi-Xin Zhou, Forschung im Ingenieurwesen,
70(4), 243–251 (2005).
4. S. V. Garimella, Y. K. Joshi, A. Bar-Cohen, R. Mahajan, K. C. Toh, V. P. Baelmans,
J. Lohan, B. Sammakia, and F. Andros, IEEE Trans.On Components and Packaging
Technologies, 25(4) 569–575, (2002).
5. K. Wade and A. J. Banister, Pergamon Texts in Inorganic Chemistry Volume 12,
University of Durham, Pergamon Press, 1975.
6. S. Moorthy Babu, L. Durai, R Dhanasekaran, and P. Ramasamy, Bull. Mater. Sci.,
13(1&2), 43–49 (1990).
A SIMS profile plotted with thickness (nm) vs. secondary ion
intensity (c/s) of the sample (Fig. 7) shows the depth profiles of Ga,
Mo and Cu (Fig. 8a). The first ∼100 nm of the top surface is gallium
rich. However, as the depth penetration progresses the concentration
of Ga is rich up to ∼50 nm inside the Mo layer and then falls off.
The SIMS profile also shows that both Ga and Cu diffuse into the Mo
through out entire thickness of the 600 nm Mo.
The sample was further analyzed by TEM and EDS as shown in
Fig. 8b–8d. The composition, measured by EDS along the line shown
by an arrow in Fig. 8b, is plotted as a function of the thickness vs. the
number of counts (Fig. 8c). The first 15 nm of the profile were on the
organic coating layer and thus shows no counts for Ga, Cu and Mo
(Fig. 8c). Further along the profile, 25–35 nm from the surface into the
Mo layers shows a high concentration of Ga. These data agree with
the data obtained by SIMS. However, further along the profile, within
the Mo shows only the presence of Mo which does not quite agree
with the SIMS data. The discrepancy between the EDS and SIMS data
could be explained by the limited sampling size of EDS compared to
the SIMS. In addition, SIMS has lower detection limits. The Ga and
Cu observed in the Mo layer by SIMS may be in the intergranular
region and the EDS may have been done on individual Mo column.
One final explanation may be found in the image in Fig 8d. We
observed islands forming on the Mo layer and these islands contained
Cu while next to the island no Cu is detected, only Mo and Ga.
However, why the copper accumulated in a localized bump yet has to
be resolved by a further study.
7. P. K. Pendey, S. N. Sahu, and S. Chandra, Editors, Hands book of Semiconductor
Electrodeposition, pp. 206, Marcel Dekker, New York 1996.
8. S Chandra and Neeraj Khare, Semicond. Sci. Technol., 2, 214–219 (1987).
9. M Contreras et al., Prog. Photovolt., 7, 311–316 (1999).
10. R Wieting, AIP Conf. Proc., 462, 3–8 (1999).
11. R. M. Burgess, W. S. Chen, W. E. Devaney, D. H. Doyle, N. P. Kim, and B. J.
Stanbery, “Electron and proton radiation effects on GaAs and CUInSe2 thin film
solar cells,” Photovoltaic Specialists Conference, 1988, Conference Record of the
Twentieth IEEE, 2, 909 (1988).
12. A Jasenek et al., Thin Solid Films, 387, 228–230 (2001).
13. W.
Chen
et al.,
Photovoltaic
Specialists
Conference,
377,
1987,
1445
Conference
Record
of
the Nineteenth
IEEE,
(1987).
14. T. Takahashi et al., J. Mater. Sci. Lett., 8, 882 (1989).
15. Z. Marinkovic and V. Simic, Thin Solid Films, 217, 26–30 (1992).
16. V. Simic and Z. Marinkovic, Thin Solid Films, 34, 179–183 (1976).
17. S. Nakahara and E. Kinsbron, Thin Solid Films, 113, 15–26 (1984).
18. M. Purwins, R. Enderle, M. Schmid, P. Berwian, G. Muller, F. Hergert, S. Jost, and
R. Hock, Thin Solid Films, 515, 5895–5898 (1976).
19. R. S. Crisp, W. G. Henry, P. A. Schroeder, and R. W. Wilson, Journal of chemical
and Engineering data, 11(4), 556 (1966).
20. A. Kampmann, J. Rechid, A. Raitzig, S. Wulff, M. Mihhailova, R. Thyen, and K.
Kalberlah, Mat. Res. Soc. Symp. Proc., Vol. 763, B8.5.1 (2003).
21. Yusuke Oda, Takashi Minemoto, Hideyuki Takakura, and Yoshihiro Hamakawa,
Mater. Res. Soc. Symp. Proc., Vol. 1012, Y12–39 (2007).
22. M. Ganchev, J. Kois, M. Kaelin, S. Bereznev, E. Tzvetkova, O. Volobujeva, N.
Stratieva, A. Tiwari, Thin Solid Films, 511 – 512, 325 – 327 (2006).
23. I. M. Dharmadasa, N. B. Chaure, G. J. Tolan, and A. P. Samantilleke, J. Electrochem.
Soc, 154(6), H466–H471 (2007).
Conclusions
The solid state reaction of electroplated Ga thin films with Me (Me
= Cu, Au) have been observed during aging at room temperature.
The stoichiometry of the resulting equilibrium compounds depends
on the ratio of at. wt% of both Ga and Me as well as the time of
diffusion. We observed that the increase of room temperature closer
to the melting point of gallium (29.7◦C) accelerated its diffusion and
24. G. Hanket, W. Shafarman, and R. Birkmire, Proc. WCPEC-4, 560 (2006).
25. S. Z. Bokshtein, Structure and Properties of Metal Alloys [in Russian], Metallurgiya,
Moscow 53 (1971).
26. Z. Marinkovic and V. Simic, J. Less-Common Met., 115, 225 (1986).
Downloaded on 2014-06-19 to IP 155.198.30.43 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).