E-ALD of Cu nanofilms on Ru/Ta wafers using surface limited redox replacement

Article abstract of DOI:10.1149/1.3454213

This paper describes the formation of Cu nanofilms on a Ru/Ta-coated wafer using electrochemical atomic layer deposition (E-ALD). The initial steps involved cleaning and oxide removal from the Ru/Ta substrate using ultrahigh vacuum electrochemical system. Auger spectroscopy was used to follow the relative amounts of oxygen, Ru, and Cu on the wafer: as-received, after electrochemical treatment, after ion bombardment, and after Cu deposition. An automated flow cell electrodeposition system was employed to grow Cu nanofilms with up to 200 cycles using surface limited redox replacement (SLRR) reaction. The open-circuit potential was used to follow the exchange of Pb for Cu in the SLRR reaction. Electron probe microanalysis was used to determine the homogeneity of the Cu films. The use of a complexing agent (citrate) greatly improved the homogeneity. Different concentrations of citrate were investigated, containing 2 or 4 mM citrate. Atomic force microscopy images of the Cu films showed the ingress morphology to be the same as the egress when 4 mM citrate was used. A prominent Cu(111) peak was displayed in the X-ray diffraction pattern for 200 cycles of Cu grown on the Ru/Ta-coated wafer.

Full text of DOI:10.1149/1.3454213

D466
Journal of The Electrochemical Society, 157 ͑8͒ D466-D471 ͑2010͒
0013-4651/2010/157͑8͒/D466/6/$28.00 © The Electrochemical Society
E-ALD of Cu Nanofilms on Ru/Ta Wafers Using Surface
Limited Redox Replacement
a
Chandru Thambidurai,a Daniel K. Gebregziabiher,a, Xuehai Liang,
*
Qinghui Zhang,a Valentina Ivanova,b Paul-Henri Haumesser,b and
a,**,z
John L. Stickney
aChemistry Department, University of Georgia, Athens, Georgia 30602, USA
bCEA, LETI, MINATEC, F38054 Grenoble, France
This paper describes the formation of Cu nanofilms on a Ru/Ta-coated wafer using electrochemical atomic layer deposition
͑E-ALD͒. The initial steps involved cleaning and oxide removal from the Ru/Ta substrate using ultrahigh vacuum electrochemical
system. Auger spectroscopy was used to follow the relative amounts of oxygen, Ru, and Cu on the wafer: as-received, after
electrochemical treatment, after ion bombardment, and after Cu deposition. An automated flow cell electrodeposition system was
employed to grow Cu nanofilms with up to 200 cycles using surface limited redox replacement ͑SLRR͒ reaction. The open-circuit
potential was used to follow the exchange of Pb for Cu in the SLRR reaction. Electron probe microanalysis was used to determine
the homogeneity of the Cu films. The use of a complexing agent ͑citrate͒ greatly improved the homogeneity. Different concen-
trations of citrate were investigated, containing 2 or 4 mM citrate. Atomic force microscopy images of the Cu films showed the
ingress morphology to be the same as the egress when 4 mM citrate was used. A prominent Cu͑111͒ peak was displayed in the
X-ray diffraction pattern for 200 cycles of Cu grown on the Ru/Ta-coated wafer.
© 2010 The Electrochemical Society. ͓DOI: 10.1149/1.3454213͔ All rights reserved.
Manuscript submitted February 15, 2010; revised manuscript received May 21, 2010. Published June 25, 2010.
Metallic copper is widely used for interconnects in ultralarge-
has only recently been possible with the development of SLRR, an
outgrowth of work by Adzic,19 Mrozek et al.,20 Vasilic et al.,21,22
Viyannalage et al.,23 and Kim et al.,24-26 which has also been re-
ferred to as monolayer ͑ML͒ restricted galvanic displacement. In an
SLRR, ions of the metal of interest are introduced over the substrate
coated with an atomic layer of a less noble metal ͑a sacrificial layer͒
at open-circuit potential ͑OCP͒. The more noble element then takes
electrons from the sacrificial layer, dissolving it and plating an
atomic layer in its place. E-ALD cycle chemistries, SLRR, have
been developed for many metals, including Cu,23,25,27,28 Pt,24,29-31
Ag,32-35 and Ru.36
This paper describes the use of an ultrahigh vacuum ͑UHV͒ sur-
face analysis instrument to study the electrodeposits. The use of
UHV surface analysis techniques to study electrodes has been re-
ferred to as UHV-electrochemical ͑EC͒, with the electrochemical
experiments generally taking place in an attached stainless steel
antechamber.37-41 UHV-EC was used here to study the cleaning pro-
cedures for a Ru/Ta-coated wafer using cyclic voltammetry ͑CV͒
and ion bombardment. The UHV-EC configuration allowed the
transfer of the substrates from the electrochemical cell directly into
the UHV analysis chamber without exposure to ambient. In addition,
an E-ALD cycle was developed for the deposition of a Cu nanofilm
on Ru/Ta. While developing the procedure for Cu deposition, the Cu
coverage decreased from ingress to egress. The coverage gradient
evidenced that the mechanism for the SLRR may not always be
electron transfer from atom to ion ͑the direct exchange pathway͒ but
rather from deposit to ion or electron transfer through the substrate.
That is, the electrons may come from any point on the surface, travel
through the deposit, and transfer where the activity for the deposit-
ing ion is greatest, usually at the ingress initially, for the flow cell
used in the present study. Deposit morphology was studied using
atomic force microscopy ͑AFM͒, elemental composition using elec-
tron probe microanalysis ͑EPMA͒, and deposit structure using
glancing incidence X-ray diffraction ͑XRD͒.
scale integration due to its low electrical resistivity ͑1.67 ␮⍀ cm͒
and electromigration.1 However, to avoid Cu diffusion through di-
electric interlayers, or into silicon, a barrier layer is used. The im-
portant issues in the formation of barrier layers are good adhesion to
both Cu and the interlayer dielectric, and low electrical resistivity.2
In addition to this, the International Technology Road Map for
Semiconductors demands a thinner barrier layer ͑Ͻ5 nm͒ for it and
integrated circuit devices, in general, to continue scaling.3,4 Cur-
rently, Ta/TaN films are used as barrier layers, although they tend to
be more resistive than desired for direct Cu electroplating, especially
in high aspect features. In addition, a Cu seed layer must be depos-
ited on the Ta/TaN layer to ensure good Cu electrofilling.3
Ru has received significant attention as a possible barrier layer
material, for an advanced technology node, as it might obviate the
need for a Cu seed layer before the dual damascene process.5-7 Ru
has an order of magnitude lower resistivity ͑7.6 ␮⍀ cm͒ compared
with Ta ͑13.5 ␮⍀ cm͒ and TaN ͑Ͼ200 ␮⍀ cm͒. More impor-
tantly, Ru, like Ta, shows negligible solid solubility with Cu even at
high temperatures, based on the binary phase diagram.8
Copper deposition on Ru using techniques such as sputtering,9
electrodeposition,2,8 molecular beam epitaxy,10 physical vapor
deposition,3 and metallorganic chemical vapor deposition11,12 has
been studied over the past 10 years. In this paper, the electrodepo-
sition of Cu on a Ru/Ta-coated wafer has been investigated using the
electrochemical form of atomic layer deposition ͑ALD͒, where the
cycles were based on surface limited redox replacement ͑SLRR͒.
E-ALD is the electrochemical analog of ALD, a nanofilm deposition
methodology based on the formation of materials one atomic layer
at a time, using surface limited reactions.13 E-ALD is based on the
use of electrochemical surface limited reactions called underpoten-
tial deposition ͑UPD͒. By definition, UPD is the deposition of one
element on another at a potential before ͑under͒ that needed to de-
posit the element on itself: bulk deposition.14-18
Experimental
Advantages with the use of E-ALD should include low cost,
conformal growth, and atomic layer control over deposit thickness.
In addition, most deposits are formed under near equilibrium condi-
tions, which suggest that the deposits should not require annealing.
The layer-by-layer electrodeposition of pure metals using E-ALD
Substrates were 7 nm thick films, 90% Ru and 10% Ta, on 50 nm
of SiO2 on a Si͑100͒ wafer. The films were formed by physical
vapor deposition. The substrates were obtained from LETI
͑Grenoble, France͒. The as-received substrates were cleaned using
CV in degassed 10 mM HClO4 ͑Aldrich, doubly distilled͒ to remove
any oxide layer formed in transit. The UHV-EC system26,42,43 con-
sisted of a surface analysis chamber attached to a stainless steel
electrochemical antechamber via a 4.5 in. o.d. UHV gate valve. The
UHV-EC instrument allowed experiments to be performed without
*
Electrochemical Society Student Member.
Electrochemical Society Fellow.
**
z E-mail: stickney@chem.uga.edu
Downloaded on 2015-03-27 to IP 138.251.14.35 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, 157 ͑8͒ D466-D471 ͑2010͒
removing the sample from the instrument. The antechamber was
D467
backfilled with high purity Ar gas ͑99.999%͒ to atmospheric pres-
sure while electrochemical experiments were performed. Subse-
quently, the electrochemical cell was removed from the antecham-
ber, through a second 4.5 in. o.d. UHV gate valve, into a secondary
antechamber. That process allowed the pump down of the sample in
the first antechamber and its transfer to the analysis chamber for
characterization without exposure to air. After characterization, the
substrate was further cleaned by Ar+ ion bombardment to remove
any remaining contaminants. The analysis chamber was equipped
with optics for low energy electron diffraction ͑Princeton Research
Instruments, Inc.͒ and a cylindrical mirror analyzer for Auger elec-
tron spectroscopy ͑AES, Perkin-Elmer͒.
The Cu nanofilm was deposited using an electrochemical flow
electrodeposition system consisting of pumps, valves, an electro-
chemical flow cell, a potentiostat, and specialized software ͑Electro-
chemical ALD L.C.͒.44,45 The auxiliary electrode was a gold wire
embedded in the cell wall, facing the deposit, while the reference
electrode was Ag/AgCl ͑3 M NaCl, Bioanalytical Systems, Inc.͒
inserted in the egress.
Solutions used in the UHV-EC system experiments were 0.1 mM
Cu͑ClO4͒2 ͑Aldrich, 98%͒ in 10 mM HCl ͑Sigma Aldrich͒ and 0.1
mM Pb͑ClO4͒2 ͑Aldrich, 99.995%͒ in 10 mM NaClO4. The same
solutions were used in the flow cell system but with a higher con-
centration of 1 mM Pb and Cu solutions. The blank solution used in
the flow cell system was 50 mM NaClO4. Sodium citrate
͑Na3C6H5O7͒ was used as a complexing agent. Electrochemical re-
actions in the UHV-EC system were performed in a Pyrex glass cell.
A Au wire counter electrode and a Ag/AgCl ͑3 M NaCl͒ reference
electrode ͑Bioanalytical Systems͒ were used. A ␮Autolab Type III
potentiostat/galvanostat was used for potential control. All potentials
were reported vs Ag/AgCl ͑3 M NaCl͒.
Ru
Cl
O
Cu
d
c
b
a
50
250
450
650
850
eV
Figure 1. AES spectra of ͑a͒ as-received Ru/Ta substrate, ͑b͒ after CV in 10
mM HClO4 at 5 mV/s, ͑c͒ after cold IBB, and ͑d͒ after Cu UPD.
there was always oxygen on the surface ͑from Auger͒ even when
extreme electrochemical reductions were performed. This raised the
question, was the oxide reduced from TaN in solution? Was the
surface reoxidized upon sample transfer? Or, was the oxide never
reduced under the electrochemical conditions used? The previous
TaN work suggests that, either way, the presences of Ta on the
surface should always be accompanied by oxygen ͑Fig. 1b͒, while it
is known that the Ru surface oxide can be electrochemically
reduced.36,46-48
In an attempt to remove more of the oxide, and any carbon that
might be present, the surface underwent ion bombardment ͑IBB͒
with 200 eV Ar+ ions. It can be seen from Fig. 1c that the oxygen
signal did go down significantly, to a ratio of 0.08, relative to Ru.
The peak for Ru has essentially the same energy as that for C, so the
only way to judge the amount of C on the surface is to look at the
ratio of the Ru peaks at 231 eV to that at 273 eV, as the higher the
ratio, the less C present.48 After IBB, the ratio suggested that all C
had been removed.
Figure 1d displays an AES spectrum for the Ru/Ta surface after
Cu UPD at Ϫ30 mV, in 0.1 mM Cu͑ClO4͒2. The charge suggested a
Cu coverage of 0.90 ML. One ML is defined here as one deposited
atom for each substrate surface atom. In this paper, the substrate has
been approximated as a Au͑111͒ surface with an atomic density of
1.17 ϫ 1015 atoms/cm2. Auger indicated a Cu to Ru peak ratio of
0.09 and a small peak for chloride at 178 eV. The chloride signal
resulted from the 10 mM HCl electrolyte,49,50 rather than the per-
The solutions were degassed with high purity Ar or N2 gases in
the UHV-EC and flow cell deposition systems, respectively, for 1 h
before electrochemical studies. In the UHV-EC system, Cu UPD
was performed at Ϫ30 mV for 1 min to form the first Cu atomic
layer. It was followed by the deposition of Pb UPD at Ϫ475 mV for
1 min to form the first sacrificial atomic layer. Pb UPD on Cu UPD
was then exposed to the Cu2+ solution at OCP for redox replace-
ment, completing one ALD cycle. This basic cycle, derived from
previously published results,27 was used to form the Cu atomic lay-
ers. After confirmation of a working Cu cycle, using the UHV-EC
instrument, that chemistry was transferred to the flow cell deposition
system to grow thicker nanofilms.
A Scintag PAD-V diffractometer, with Cu K␣ radiation ͑␭
= 1.5418 Å͒, was used to obtain glancing angle XRD patterns.
EPMA was run on a JEOL 8600 wavelength dispersive scanning
electron microprobe. The morphology of the surface was character-
ized by intermittent contact mode AFM ͑Molecular Imaging, Pico-
Plus͒. AFM image resolution was 512 ϫ 512, and the image was
processed using WSxM software.
2.0E-05
Cycle 1
Cycle 2
Results and Discussion
Figure 1 shows the Auger spectra for a Ru/Ta-coated wafer ͑a͒
as-received, ͑b͒ after cycling in blank solution, ͑c͒ after cold ion
bombardment, and ͑d͒ after Cu UPD at Ϫ30 mV. The ratio of oxy-
gen to Ru Auger peak heights was calculated to be 0.41 for the
as-received wafer ͑Fig. 1a͒. The high ratio was the result of surface
oxide formation on both Ru and Ta of the Ru/Ta film, after exposure
to air. In an attempt to remove the oxide layer, the electrode was
immersed in 10 mM HClO4 and scanned negative from the OCP, 55
mV ͑Fig. 2͒. During the first reductive scan of the experiment, a
significant reduction current was evident between Ϫ100 and Ϫ350
mV, which was then absent in the subsequent scan. After those two
cycles in the blank solution, the O to Ru ratio decreased to 0.27 ͑Fig.
1b͒, still a significant ratio. The currents on the left side of Fig. 2
were primarily due to hydrogen evolution and oxidation.
0.0E+00
-2.0E-05
-4.0E-05
-6.0E-05
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
OCP = 0.055V
E (V)
Figure 2. CV of Ru/Ta wafer in 10 mM HClO4 at a scan rate of 5 mV/s. The
Similar studies of TaN surfaces ͑not yet published͒ indicated that
OCP for Ru/Ta wafer in blank solution was 55 mV.
Downloaded on 2015-03-27 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D468
Journal of The Electrochemical Society, 157 ͑8͒ D466-D471 ͑2010͒
8.E-05
6.E-05
4.E-05
2.E-05
0.E+00
0.89ML
0
5
10
15
20
25
30
t (s)
Figure 3. I-t trace of Cu UPD stripping at 0.15 V in 10 mM HClO4. The Cu
UPD was deposited using Ϫ30 mV and UPD charge was calculated as 0.90
ML.
Figure 5. ͑Color online͒ Three ALD cycles for Cu deposition on Ru/Ta
wafer. Pb was deposited at Ϫ450 mV and the average deposition charge was
calculated as 0.80 ML. The Cu replacement reaction was done in OCP for
10 s.
chlorate ions ͓Cu͑ClO4͒2͔, as perchlorate ions adsorb only weakly.
Chloride, however, strongly adsorbs on Cu and can limit Cu
oxidation.50,51 The oxygen evident in Fig. 1d would not be expected
on a Cu surface, which had been in contact with chloride, suggesting
that the oxide is associated with Ta at the surface and that the Cu did
not cover the Ta oxide. After the characterization of the Cu UPD
layer, the sample was transferred back to the electrochemistry ante-
chamber, where the Cu UPD was stripped in the blank solution ͑Fig.
3͒ at 150 mV, resulting in a charge equivalent to 0.89 ML, which is
essentially the same as observed for Cu UPD. This stripping charge
was consistent with the Cu not being oxidized and the presence of
Ta oxide on the surface.
The initial ALD cycle chemistry for depositing Cu atomic layers
on the Ru/Ta surface, using SLRR, was developed for deposition of
Cu on Au,27,36 and started with Cu UPD at Ϫ30 mV for 1 min, this
time on the Ru/Ta substrate. The cycle consisted of Pb UPD at Ϫ475
mV for 1 min, on the Cu coated surface, from the Pb2+ ion solution.
The resulting Pb sacrificial atomic layer was then replaced with a Cu
atomic layer by immersion in the Cu2+ ion solution at OCP for redox
replacement. The replacement was followed by monitoring the OCP
and was assumed complete when it reached the formal potential for
Cu2+/Cu of about 0.0 V. This cycle was repeated to grow Cu nano-
films.
Cycles performed with the UHV-EC instrument used a simple Pyrex
cell, where the solution exchange was performed by draining and
then refilling the cell. The result was that some of the sacrificial Pb
atomic layer was spontaneously oxidized when the solution was
drained, and the deposit was allowed to go open circuit in the back-
fill gas. The result was a greatly reduced amount of Pb UPD to
exchange for Cu2+ ions, and accounting for the minimal amounts of
Cu deposited in the two SLRR cycles.
To avoid problems with loss of potential control and to grow
thicker Cu films, the automated flow cell deposition system was
used to grow the Cu films with up to 200 cycles. In the flow depo-
sition system, potential control was not lost between solutions, so no
Pb was oxidized before the introduction of the Cu2+ solution. Figure
5 displays the current–time ͑I-t͒ curves for three ALD cycles of Cu
deposition on Ru/Ta wafer, using the flow cell deposition system,
where essentially the same Pb coverages were clearly obtained in
each cycle. A flow rate of 10 mL/min was used. Larger flow rates are
frequently used by this group, 40 mL/min, for example; however,
the 7 nm thick Ru/Ta films proved too fragile for faster pumps and
peeled off.
The basic E-ALD program was as follows: Pb solution was
pumped into the cell for 8 s at Ϫ450 mV and held for 8 s to produce
0.8 ML of Pb UPD. A blank solution was then pumped through the
cell, at the same controlled potential, to remove excess Pb2+ ions.
The potential was then allowed to go open circuit, and the Cu2+
solution was flushed through for 10 s, at OCP, allowing Cu2+ ions to
oxidize the Pb and form a Cu atomic layer. The cycle was completed
by pumping a blank solution through the cell to remove the resulting
Pb2+ ions as well as any excess Cu2+ ions.
Figure 4 displays the Auger ratio for Cu/Ru and the Cu charge in
ML, for Cu UPD, one SLRR cycle, and two SLRR cycles, formed
using the UHV-EC instrument. UPD results in just under 0.9 ML of
Cu, while one and two cycles suggested little additional growth.
The above cycle was repeated two, four, and eight times. After-
wards, the resulting Cu films were stripped in a blank solution, at
400 mV, to determine the amount of Cu deposited during the SLRR.
To determine the background charge from Ru/Ta wafer during the
stripping steps, the same process was performed in a blank solution,
using the same potential ͑400 mV͒. The background charge was
only 0.04 ML. Figure 6 shows the Cu coverage as a function of two,
four, and eight cycles: 0.83, 1.95, and 3.89 ML, respectively. The
linear growth vs the cycle number evident in Fig. 6 is a clear sign of
an ALD process. The average Cu coverage calculated from this
graph was 0.49 ML per ALD cycle. Based on the average Pb depo-
sition charge ͑0.80 ML͒, the replacement efficiency was calculated
to be 61%. The reasons for less than 100% efficiency are associated
with side reactions, both during Pb deposition and during the ex-
change. It is assumed that the losses are associated with the reduc-
tion of small amounts of oxygen or protons. Extra charge was likely
Figure 4. AES ratios of Cu to Ru and Cu coverage charges vs the number of
cycles.
Downloaded on 2015-03-27 to IP 138.251.14.35 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, 157 ͑8͒ D466-D471 ͑2010͒
D469
200
4.5
4
0mM sod.citrate, 10s replacement time
2mM sod.citrate, 20s replacement time
4mM sod.citrate, 40s replacement time
0
-200
-400
-600
3.5
3
2.5
2
1.5
1
0.5
0
blank strip
0
20
40
60
80
time,s
0
2
4
6
8
10
Figure 8. OCP measurement as a function of ͓Cit3−͔ added to the Cu plating
bath.
number of ALD cycles
Figure 6. Cu stripping charge as a function of ALD cycle.
change pathway͒. However, the fact that the deposit thickness is
greater near the inlet suggests, at least with the hardware and con-
ditions used by this group, that a different mechanism or pathway
may be at work. Instead of the direct exchange pathway, electrons
may transfer through the substrate. The substrates used to make the
deposits are conductive, so there is no necessity for the electrons of
an atom to transfer or tunnel directly to an ion in solution for a direct
place exchange. Instead, deposition may be a function of the Cu2+
ion activity gradient in solution. That is, reduction of ions occurs
where the Cu2+ activity at the surface is greatest. In a flow cell, that
would initially be at the ingress, where convection is present.
The gradient in ion activity at the ingress is transient, a function
of the time it takes to fill the cell and for the ions to diffuse to the
surface. If the cell could be filled homogeneously on a time scale
which is short compared with the exchange reaction, the deposit
should be homogeneous. Possible solutions include speeding up the
solution introduction so the composition over the surface is more
homogeneous. Alternatively, the exchange reaction could be slowed
down on the time scale of solution introduction. As noted, the solu-
tion was pumped at only 10 mL/min, instead of the 40 mL/min used
for previous deposits on Au, because of the instability of the Ru/Ta
film. With the present hardware and substrate, the solution could not
be delivered faster and still obtain a deposit.
Figure 8 shows the OCP as a function of time for exchange of a
Pb atomic layer with a 1 mM Cu solution. The leftmost curve ͑solid
line͒ shows a reaction time of 3–4 s, that is, it takes 3–4 s for the
potential to shift from that used to deposit the Pb to that character-
istic of a Cu surface, about as much time as it takes to pump solution
into the cell. That is, the reaction is over before the Cu2+ ion solu-
tion has even filled the cell. The rate of the exchange process varies
from element to element, and with solutions and hardware. Some
elements take considerably longer to exchange than others. Cu is
notably faster than Ru or Pt.27
As the solution could not be introduced faster, slowing down the
exchange rate of Pb for Cu was considered. The most direct method
would be to drop the concentration of Cu2+ ions. The Cu2+ ion
concentration was 1.0 mM and could easily have been dropped to
0.1 mM; however, if it was dropped much lower, there might not
have been enough Cu2+ ions in the cell to form a Cu ML. To drop
the Cu2+ ion concentration by a couple of orders of magnitude, a
complexing agent was introduced. In that way, the total amount of
Cu ions in the cell could remain at the 1 mM level but be present as
a complex, with only a small amount present as Cu2+ ions. In the
present studies, citrate was chosen to complex the Cu2+ ions. Two
solutions were prepared, both 1 mM Cu2+ ion solutions, but one
with 2 mM citrate and one with 4 mM citrate. Two hundred E-ALD
cycle films were grown with both solutions for exchange with the Pb
sacrificial atomic layers. Figure 8 displays the shifts in OCP during
measured during Pb deposition, as Pb is known as a catalyst for
oxygen reduction, and it is very hard to remove all the O2 from
solution by purging with N2, such as used in these studies.
The same ALD program for Cu deposition was used to grow
thicker films, including 25, 100, and 200 cycle deposits, to better
characterize the morphology, crystal structure, and deposit distribu-
tion. The EPMA signals were reported in terms of atomic percent-
age, with the assumption that the sample was homogeneous. As the
samples were not homogeneous but consisted of a film on a sub-
strate, it is best to consider the atomic percentage as a relative indi-
cation of coverage of an arbitrary unit. The remaining percentage
signal came from the substrate. The average Cu coverage for a 200
cycle Cu film was 33%. Figure 7a displays the Cu coverage across
the deposit, from ingress to egress, which is strikingly skewed to-
ward the ingress. This turns out to be an extreme case, but a similar
pattern has been observed for many of the E-ALD deposits formed
using SLRR with the flow cell hardware.27 Excess deposition at the
inlet has been observed with EPMA and optical microscopy, to some
degree, with E-ALD metal nanofilms of Cu, Ru, and Pt, by this
group.
The model of an SLRR is that a depositing ion takes the elec-
trons from a sacrificial atom͑s͒ on the surface, dissolving the sacri-
ficial atom and reducing the depositing ion onto the surface to take
its place. That is, one Pb atom is oxidized by one Cu2+ ion, and the
depositing Cu atom takes the place of the Pb atom ͑the direct ex-
100
0mM
2mM
4mM
90
80
70
60
50
40
30
20
10
0
a
b
b
0
0.5
1
1.5
2
2.5
3
Deposit position, cm
Figure 7. Cu atom % as a function of deposit position. The EPMA was done
on 200 cycle Cu film deposited using 0, 2, and 4 mM ͓Cit3−͔ baths.
Downloaded on 2015-03-27 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D470
Journal of The Electrochemical Society, 157 ͑8͒ D466-D471 ͑2010͒
100.86 nm
359.35 nm
870.14 nm
1.0µm
1.0µm
1.0µm
Figure 9. ͑Color online͒ AFM images of
200 cycle Cu film deposited using differ-
ent ͓Cit3−͔ baths. Image resolution was
512 ϫ 512 and the image was processed
and analyzed with WSxM software.
0.00 nm
0.00 nm
0.00 nm
b. Inlet, 2 mM[Cit3-]
a. Inlet, 0 mM[Cit3-]
c. Inlet, 4 mM[Cit3-]
66.47 nm
120.28 nm
160.32 nm
1.0µm
1.0µm
1.0µm
0.00 nm
0.00 nm
a’. Outlet, 0 mM[Cit3-]0.00 nm
b’. Outlet, 2 mM[Cit3-]
c’. Outlet, 4 mM[Cit3-]
the exchange of Pb for Cu. With no citrate present, as noted above,
the majority of the exchange took 3–4 s to reach the Cu potential
plateau. However, the exchange times were closer to 15 and 30 s for
the solutions containing 2 and 4 mM citrate, respectively. The pla-
teau potentials shifted negatively with increasing citrate, as ex-
pected, given the corresponding decrease in Cu2+ activity.
though the Ru film was so thin that most of the peak can be ascribed
to Cu, suggesting a primary ͑111͒ growth habit. A peak correspond-
ing to Cu͑200͒ planes was also observed without interference. There
was no evidence of Pb in the XRD pattern or from EPMA.
Conclusion
Figure 7b shows EPMA data from across two 200 cycle Cu films
formed using 1 mM Cu2+ and 2 and 4 mM citrate. In addition, Fig.
7a was measured from a 200 cycle deposit where no citrate was
present. The relative atomic Cu % was displayed from ingress to
egress. With no citrate present, the deposit was an order of magni-
tude thicker at the ingress, relative to the rest of the surface ͑Fig.
7a͒. This was also evident from optical microscopy, as light scatter-
ing indicated significant roughening near the ingress. Layer-by-layer
growth is not expected to produce such an inhomogeneous deposit
and rough morphology. By adding 2 and 4 mM citrate, deposit mor-
phology became specularly reflective and the thickness homoge-
neous ͑Fig. 7b͒, with the atomic percentage essentially the same
from ingress to egress. The use of citrate in the Cu bath appears to
have slowed down the exchange process, allowing the Cu2+ ion
gradient to be removed before a significant fraction of the Pb had
reacted.
Cu nanofilms were formed on a prospective barrier layer mate-
rial, Ru/Ta, using E-ALD with SLRR cycles. UHV-EC methodolo-
gies were used to examine the surfaces resulting from initial surface
treatments considered as possible substrate cleaning procedures. An
oxide layer present on the as-received Ru/Ta wafers was partially
removed by electrochemical reduction. Other Ta work suggested
that the oxide remaining after electrochemical reduction was Ta ox-
ide, not Ru oxide. The Ta oxide is difficult to reduce in an aqueous
solution, while Ru can be readily reduced electrochemically.
Using an automated electrochemical flow cell deposition system,
Cu deposits were formed on these Ru/Ta substrates using E-ALD.
Initial deposits showed significantly greater thickness near the in-
gress than on the rest of the substrate. It is proposed here that the
SLRR mechanism manifested here may involve the transfer of elec-
trons through the substrate, rather than direct exchange. This con-
The homogeneity and morphology of the Cu citrate films were
also studied using AFM. Figure 9 displays AFM images of 200 cycle
Cu films, deposited using 1 mM Cu solutions with 0, 2, and 4 mM
citrate. As expected, the Cu film deposited without citrate was very
rough at the ingress, indicating that significantly more than an ML
was formed there each cycle. The deposit formed using 4 mM citrate
showed the least roughing at the ingress. At the egress, none of the
baths displayed excessive roughening, although the crystallite size
appears larger for the solution without citrate. Evidently, the use of
the complexing agent worked well for the production of a more
homogeneous and smooth deposit. However, in previous studies of
the E-ALD formation of Cu nanofilms on Au substrates, 200 nm
terraces were observed.27 These surfaces showed a surface consist-
ing of 40 nm grains under these imaging conditions. The difference
may have to do with the nature of the Ru/Ta substrate and the
presence of the Ta oxide, although the citrate may have acted as a
grain refiner, decreasing the crystallite size, as well.
1000
CU(111)/RU(101)
800
600
400
RU(002)
CU(200)
200
0
35
40
45
50
55
60
65
Grazing angle XRD was also used to characterize the Cu film
structure. Figure 10 displays the XRD pattern for a 200 cycle Cu
film deposited using a Cu solution with 4 mM citrate. The peak at
43.35° corresponded to both the Cu͑111͒ and Ru͑101͒ planes, al-
2-theta
Figure 10. XRD pattern for the Cu film deposited using 4 mM ͓Cit3−͔ bath.
Downloaded on 2015-03-27 to IP 138.251.14.35 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, 157 ͑8͒ D466-D471 ͑2010͒
D471
17. A. A. Gewirth and B. K. Niece, Chem. Rev. (Washington, D.C.), 97, 1129 ͑1997͒.
18. E. Herrero, L. J. Buller, and H. D. Abruña, Chem. Rev. (Washington, D.C.), 101,
1897 ͑2001͒.
19. R. Adzic, Encyclopedia of Electrochemistry, Vol. 1, p. 561, Wiley-VCH, Weinheim
͑2003͒.
20. M. F. Mrozek, Y. Xie, and M. J. Weaver, Anal. Chem., 73, 5953 ͑2001͒.
21. R. Vasilic and N. Dimitrov, Electrochem. Solid-State Lett., 8, C173 ͑2005͒.
22. R. Vasilic, L. T. Viyannalage, and N. Dimitrov, J. Electrochem. Soc., 153, C648
͑2006͒.
23. L. T. Viyannalage, R. Vasilic, and N. Dimitrov, J. Phys. Chem. C, 111, 4036
͑2007͒.
24. Y.-G. Kim, J. Y. Kim, D. Vairavapandian, and J. L. Stickney, J. Phys. Chem. B,
110, 17998 ͑2006͒.
25. J. Y. Kim, Y.-G. Kim, and J. L. Stickney, J. Electrochem. Soc., 154, D260 ͑2007͒.
26. Y.-G. Kim, J. Y. Kim, C. Thambidurai, and J. L. Stickney, Langmuir, 23, 2539
͑2007͒.
clusion was based on the increased deposition at the ingress, relative
to the rest of the deposit. That is, if the electrons travel through the
substrate, they should result in the reduction of Cu ions where ever
the activity is highest, with the flow cell that would be near the
ingress. These results suggest the need for more detailed studies of
the kinetics for the SLRR. By using citrate as a complexing agent,
the exchange of Pb for Cu was slowed, allowing the solution to fill
the cell homogeneously before the majority of the sacrificial atomic
layer had been oxidized. This resulted in the formation of homoge-
neous 200 cycle deposits, as evidenced using EPMA, optical mi-
croscopy, and AFM.
Acknowledgments
27. C. Thambidurai, Y.-G. Kim, N. Jayaraju, V. Venkatasamy, and J. L. Stickney, J.
Electrochem. Soc., 156, D261 ͑2009͒.
28. C. P. daRosa, E. Iglesia, and R. Maboudian, J. Electrochem. Soc., 155, D244
͑2008͒.
29. F. H. B. Lima, J. Zhang, M. H. Shao, K. Sasaki, M. B. Vukmirovic, E. A. Ti-
cianelli, and R. R. Adzic, J. Solid State Electrochem., 12, 399 ͑2008͒.
30. R. E. Rettew, J. W. Guthrie, and F. M. Alamgir, J. Electrochem. Soc., 156, D513
͑2009͒.
Support from the National Science Foundation, Division of Ma-
terials Research is gratefully acknowledged.
University of Georgia assisted in meeting the publication costs of this
article.
References
1. P. C. Andricacos, C. Uzoh, J. O. Dukovic, J. Horkans, and H. Deligianni, IBM J.
Res. Dev., 42, 567 ͑1998͒.
2. T. P. Moffat, M. Walker, P. J. Chen, J. E. Bonevich, W. F. Egelhoff, L. Richter, C.
Witt, T. Aaltonen, M. Ritala, M. Leskelä, et al., J. Electrochem. Soc., 153, C37
͑2006͒.
3. M. Damayanti, T. Sritharan, Z. H. Gan, S. G. Mhaisalkar, N. Jiang, and L. Chan, J.
Electrochem. Soc., 153, J41 ͑2006͒.
4. R. Chan, T. N. Arunagiri, Y. Zhang, O. Chyan, R. M. Wallace, M. J. Kim, and T. Q.
31. J.-F. Huang, Electroanalysis, 20, 2229 ͑2008͒.
32. S. Y. Sayed, B. Daly, and J. M. Buriak, J. Phys. Chem. C, 112, 12291 ͑2008͒.
33. L. A. Porter, Jr., H. C. Choi, A. E. Ribbe, and J. M. Buriak, Nano Lett., 2, 1067
͑2002͒.
34. A. Gutes, I. Laboriante, C. Carraro, and R. Maboudian, J. Phys. Chem. C, 113,
16939 ͑2009͒.
35. N. Dimitrov, and R. Vasilic ECS Trans., 1͑12͒, 33 ͑2006͒.
36. C. Thambidurai, Y.-G. Kim, and J. L. Stickney, Electrochim. Acta, 53, 6157
͑2008͒.
Hurd, Electrochem. Solid-State Lett., 7, G154 ͑2004͒.
5. T. N. Arunagiri, Y. Zhang, O. Chyan, M. J. Kim, and T. Q. Hurd, J. Electrochem.
Soc., 152, G808 ͑2005͒.
6. T. N. Arunagiri, Y. Zhang, O. Chyan, M. El-Bouanani, M. J. Kim, K. H. Chen, C.
T. Wu, and L. C. Chen, Appl. Phys. Lett., 86, 083104 ͑2005͒.
7. D. Josell, D. Wheeler, C. Witt, and T. P. Moffat, Electrochem. Solid-State Lett., 6,
C143 ͑2003͒.
8. O. Chyan, T. N. Arunagiri, and T. Ponnuswamy, J. Electrochem. Soc., 150, C347
͑2003͒.
9. J. Oh, J. Lee, and C. Lee, Mater. Chem. Phys., 73, 227 ͑2002͒.
10. R. Naik, M. Ahmad, G. L. Dunifer, C. Kota, A. Poli, K. Fang, U. Rao, and J. S.
Payson, J. Magn. Magn. Mater., 121, 60 ͑1993͒.
37. A. T. Hubbard, J. L. Stickney, S. D. Rosasco, M. P. Soriaga, and D. Song, J.
Electroanal. Chem. Interfacial Electrochem., 150, 165 ͑1983͒.
38. M. D. Lay, K. Varazo, and J. L. Stickney, Langmuir, 19, 8416 ͑2003͒.
39. T. E. Lister, L. P. Colletti, and J. L. Stickney, Isr. J. Chem., 37, 287 ͑1997͒.
40. M. Muthuvel and J. L. Stickney, Langmuir, 22, 5504 ͑2006͒.
41. M. Muthuvel and J. L. Stickney, J. Electrochem. Soc., 153, C67 ͑2006͒.
42. D. W. Suggs and J. L. Stickney, Surf. Sci., 290, 362 ͑1993͒.
43. J. L. Stickney, J. Y. Kim, and Y.-G. Kim, ECS Trans., 1͑30͒, 41 ͑2006͒.
44. J. L. Stickney, in Advances in Electrochemical Science and Engineering, R. C.
Alkire and D. M. Kolb, Editors, p. 1, Wiley-VCH, Weinheim ͑2002͒.
45. T. L. Wade and T. Sorenson, in Interfacial Electrochemistry, A. Wieckowski, Edi-
tor, p. 757, Marcel Dekker, New York ͑1999͒.
11. K. Weiss, S. Riedel, S. E. Schulz, M. Schwerd, H. Helneder, H. Wendt, and T.
Gessner, Microelectron. Eng., 50, 433 ͑2000͒.
46. M. B. Vukmirovic, R. L. Sabatini, and R. R. Adzic, Surf. Sci., 572, 269 ͑2004͒.
47. J. Lee, I. Oh, S. Hwang, and J. Kwak, Langmuir, 18, 8025 ͑2002͒.
48. E. M. Stuve, J. W. Rogers, Jr., D. Ingersoll, D. W. Goodman, M. L. Thomas, and
M. T. Paffett, Chem. Phys. Lett., 149, 557 ͑1988͒.
49. J. L. Stickney and C. B. Ehlers, J. Vac. Sci. Technol. A, 7, 1801 ͑1989͒.
50. C. B. Ehlers, I. Villegas, and J. L. Stickney, J. Electroanal. Chem. Interfacial
Electrochem., 284, 403 ͑1990͒.
12. K.-K. Choi and S.-W. Rhee, Thin Solid Films, 397, 70 ͑2001͒.
13. S. M. George, Chem. Rev. (Washington, D.C.), 110, 111 ͑2010͒.
14. D. M. Kolb, M. Przasnyski, and H. Gerischer, J. Electroanal. Chem. Interfacial
Electrochem., 54, 25 ͑1974͒.
15. D. M. Kolb, in Advances in Electrochemistry and Electrochemical Engineering, H.
Gerischer and C. W. Tobias, Editors, p. 125, John Wiley & Sons, New York ͑1978͒.
16. K. Juttner and W. J. Lorenz, Z. Phys. Chem., Neue Folge, 122, 163 ͑1980͒.
51. J. Y. Kim, Y.-G. Kim, and J. L. Stickney, J. Electroanal. Chem., 621, 205 ͑2008͒.
Downloaded on 2015-03-27 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).