Superconformal electrochemical deposition of gold for metallization in microelectronic devices

Article abstract of DOI:10.1149/1.2196670

A novel process for superconformal electrochemical deposition of gold has been proposed and demonstrated using 3-mercapto-1-propanesulfonic acid sodium salt (MPS) as additive in a sulfite-based gold electrolyte. At the optimized process conditions true bottom-up gold deposition was achieved, with which aggressive features such as those that are straight-walled and even re-entrant were completely filled without voids. In the process, the additive MPS functions as a suppressor that hinders electrochemical deposition of gold, as opposed to its function as an accelerator in electrochemical deposition of copper. The results obtained appear to suggest that the driving force leading to superconformal deposition of gold with the sulfite-based gold electrolyte in the presence of the grain refiner thallium and the additive MPS may primarily stem from the depletion of MPS inside features of interest as a result of its consumption and limited supply.

Full text of DOI:10.1149/1.2196670

Journal of The Electrochemical Society, 153 ͑7͒ C467-C471 ͑2006͒
C467
0
013-4651/2006/153͑7͒/C467/5/$20.00 © The Electrochemical Society
Superconformal Electrochemical Deposition of Gold
for Metallization in Microelectronic Devices
,
z
Zhongmin Hu* and Tom Ritzdorf*
Semitool, Incorporated, Kalispell, Montana 59901, USA
A novel process for superconformal electrochemical deposition of gold has been proposed and demonstrated using 3-mercapto-
1
-propanesulfonic acid sodium salt ͑MPS͒ as additive in a sulfite-based gold electrolyte. At the optimized process conditions true
bottom-up gold deposition was achieved, with which aggressive features such as those that are straight-walled and even re-entrant
were completely filled without voids. In the process, the additive MPS functions as a suppressor that hinders electrochemical
deposition of gold, as opposed to its function as an accelerator in electrochemical deposition of copper. The results obtained appear
to suggest that the driving force leading to superconformal deposition of gold with the sulfite-based gold electrolyte in the
presence of the grain refiner thallium and the additive MPS may primarily stem from the depletion of MPS inside features of
interest as a result of its consumption and limited supply.
©
2006 The Electrochemical Society. ͓DOI: 10.1149/1.2196670͔ All rights reserved.
Manuscript submitted November 23, 2005; revised manuscript received March 7, 2006. Available electronically May 4, 2006.
While copper is widely used for metallization in silicon-based
additive on gold electrochemical deposition behavior, based on the
superconformal deposition results and electrochemical behavior of
gold deposition in the presence of the additive.
1-5
microelectronic devices, gold is the metal of choice for metalliza-
tion in compound semiconductor devices because of its unique
6
chemical, physical, and electrical properties. Traditionally, gold
Experimental
metallization ͑e.g., interconnection͒ is performed using through-
mask electrochemical deposition ͑electroplating͒. However, gold
metallization using damascene electrochemical deposition is needed
in many circumstances, especially when fine ͑submicrometer scale͒
structures are to be formed.
The additives MPS ͑90%͒ and 2-mercaptoethylamine ͑MEA,
95%͒ were purchased from Aldrich and used as received. The
sulfite-based gold electrolyte ͑Sel-Rex Neutronex 309i͒ was pur-
chased from Enthone-OMI. The electrolyte is referred to as base
gold electrolyte in the present paper. The electrolyte contains gold
sulfite, sodium sulfite, a pH buffer, and thallium as grain refiner.
Gold electrochemical deposition was performed at 50°C with the
sulfite-based gold electrolyte at pH 9.4 in the absence and presence
of additives under investigation. Deionized water with a specific
resistance of ϳ18 M⍀ cm was used whenever needed.
One of the challenges associated with damascene electrochemi-
cal deposition is how to achieve void-free filling of features ͑i.e.,
cavities͒ that include trenches and vias. Since the introduction of
1
,2
copper for interconnect metallization,
chemical deposition ͑or superfilling͒ of copper has been intensively
studied.
superconformal electro-
3
,7-19
As a result of the efforts of both industrial and aca-
demic communities, several proprietary copper electrolytes, where
organic additives are used to achieve superconformal filling, have
been made commercially available. Several models describing su-
Voltammetric, chronoamperometric, and chronopotentiometric
investigations were carried out using an EG&G 263A potentiostat/
galvanostat, coupled with an EG&G model 616 rotating electrode
assembly and a conventional three-electrode electrochemical cell
which consists of a gold-deposited platinum rotating disk electrode,
a saturated calomel electrode ͑SCE͒, and a platinized titanium piece
as the counter electrode. The geometric surface area of the platinum
3
,10-16
perconformal deposition of copper have also been developed.
In addition, superconformal filling of copper using chemical vapor
2
0,21
deposition in the presence of iodine
chemical deposition of silver
and superconformal electro-
have been demonstrated.
2
2-24
2
Prompted by the industry’s interest in damascene electrochemical
deposition of gold and by the observation that features of interest
could not be satisfactorily filled using commercially available gold
electrolytes, we investigated superconformal gold deposition using
disk is 0.126 cm . Prior to each distinct use, the platinum disk elec-
trode was electrochemically cleaned and then deposited with ca.
0.5 ␮m gold in the base gold electrolyte. Visually, the deposited
gold was essentially mirror-smooth. All current densities reported in
the present paper were calculated using geometric surface area, and
potential values are presented with respect to the SCE.
25
sulfite-based gold electrolytes. Superfilling of gold in submicrome-
ter trenches using a cyanide-based gold electrolyte has also been
2
6
reported.
Electrochemical deposition of gold on wafer segments was
performed with the rotating electrode assembly, coupled with a
homemade adaptor that affixes a wafer segment onto the assembly
In the present paper, we report the development of superconfor-
mal electrochemical deposition of gold for gold metallization in
microelectronic devices through damascene processing. We first
describe the issues we encountered when using a commercially
available sulfite-based gold electrolyte for gap filling and discuss
the possibilities of using additives such as thiol and disulfide
compounds that are capable of interacting with gold and altering
interfacial properties to achieve superconformal deposition. We
then discuss the results obtained with the use of representative thiols
that were selected on the basis of their potential interaction with
gold and the consequent influence on heterogeneous electron trans-
fer. Specifically, we found that with the use of 3-mercapto-1-
propanesulfonic acid sodium salt ͑MPS͒, gap-filling capability of the
gold electrolyte could be dramatically improved. At the optimized
process conditions true bottom-up gold deposition was achieved,
with which complete filling of aggressive features was accom-
plished. Finally, we made an attempt to rationalize the effect of the
2
7
and defines exposed surface area using a sealed ring contact. On
full wafers, deposition of gold was performed using Semitool’s
electrochemical deposition ͑ECD͒ tool, which allows the operator
to process wafers at predetermined process conditions such as tem-
perature, electrolyte flow rate, wafer rotation, and applied current.
Prior to electrochemical deposition, all wafers ͑either 100 or 150
mm diam͒ were deposited with a gold seed layer using physical
vapor deposition. The effect of process conditions, including the
chemical nature and concentration of the additives, on gap filling
was evaluated based on cross-sectional scanning electron micro-
graphs ͑SEM͒ taken with a focused ion beam/scanning electron mi-
croscope ͑FEI Dual Beam 820͒.
Results and Discussion
Unlike a through-mask electrochemical deposition process where
the electrically conductive seed layer lies only at the bottom of the
features into which a metal is deposited and thus the deposition is a
naturally bottom-up process, a damascene electrochemical deposi-
tion process usually requires superconformal deposition capability
*
Electrochemical Society Active Member.
E-mail: zhu@semitool.com
z
Downloaded on 2015-02-24 to IP 35.8.11.2 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C468
Journal of The Electrochemical Society, 153 ͑7͒ C467-C471 ͑2006͒
Figure 1. Cross-sectional SEMs of trenches deposited with gold using the
2
base gold electrolyte at current densities of ͑A͒ 4 mA/cm and ͑B͒
2
0
.08 mA/cm that correspond to nominal deposition rates of 0.25 and
Figure 2. Effect of additive MPS on cathodic polarization at a gold-
deposited platinum electrode. ͑A͒ Linear-scan voltammograms obtained in
the absence and presence of MPS: ͑a͒ no MPS, and ͑b͒ 0.3, ͑c͒ 1.2, ͑d͒ 2.4,
0
.05 ␮m/min, respectively.
2
and ͑e͒ 4.8 mL/L MPS. ͑B͒ Plot of electrode potential at 4 mA/cm against
MPS concentration. Initial scan potential 0 mV. Scan rate 5 mV/s. Electrode
rotating speed 100 rpm.
for achieving complete fill of features, especially those that are ag-
gressive in shape ͑e.g., straight-walled or re-entrant͒ and in aspect
ratio. Shown in Fig. 1 are representative cross-sectional SEMs ob-
tained with the base gold electrolyte. The SEM in Fig. 1A is the
result of gold deposition at the process condition optimized for nor-
mal through-mask deposition. Obviously, because of the premature
pinch-off at the entrance, the features were left with large voids.
After the process parameters such as current density, agitation, and
the duty cycle of pulsed power supply were optimized for the
present damascene electrochemical deposition, marked improve-
ment in gap filling was observed ͑Fig. 1B͒. While the optimization
of process conditions resulted in a significant improvement in gap
filling, from subconformal to near-conformal, at the expense of
deposition rate, the features of interest were still left with seam
voids. The nominal deposition rate used for achieving this near-
conformal deposition is 0.05 ␮m/min, which is only one-fifth of the
deposition rate typical of the through-mask gold deposition process.
At the optimized process conditions, the base gold electrolyte
may be used to fill nonaggressive features such as V-shaped ones.
For aggressive features such as those that are straight-walled or
re-entrant, seam voids would be unavoidable with the base gold
electrolyte that deposits conformally at best. As a result of the above
observation we explored superconformal deposition of gold with
candidate additives.
Among the thiols and disulfides available, the amine-terminated,
carboxylic acid-terminated, and sulfonic acid-terminated species are
especially intriguing. In addition to their reasonably high solubility,
these thiol and disulfide species bear terminal groups that may in-
teract electrostatically and/or chemically with gold species in solu-
tion. Sulfonic acids exist fully deprotonated in most circumstances,
even in strongly acidic conditions; surface-tethered sulfonate species
are capable of interacting electrostatically with the positively
charged free ͑uncomplexed͒ gold ions in solution. At the operational
pH of ca. 9.4, amines are partially protonated. Surface-tethered
amine species are expected to be able to interact electrostatically
with negatively charged gold–sulfite complex ions and perhaps
chemically with free gold ions. In the sulfite-based gold electrolyte
that is slightly alkaline, carboxylic acids are essentially completely
deprotonated. Surface-tethered carboxylate species would be ca-
pable of interacting electrostatically with free gold ions as well. In
this regard, carboxylic acid-terminated thiols and disulfides are ex-
pected to be similar to sulfonate-terminated thiol and disulfide spe-
cies.
In the present work that is limited in scope, we have focused our
investigation on sulfonic acid- and amine-terminated thiol additives,
or MPS and MEA. Our experimental results obtained with these two
additives indicate that the introduction of MEA into the base gold
electrolyte resulted in no appreciable improvement in gap filling
Additive selection.— It is generally believed that superfilling
with a metal ͑e.g., copper͒ in electrochemical deposition is deter-
mined by the interaction of additives with the surface of the metal
substrate and, in some cases, with the to-be-deposited metal ions in
the electrolyte. In other words, superconformal deposition is prima-
rily governed by the chemical nature of the additives used. Thiols
and disulfides are known to adsorb strongly on gold as the corre-
͑
SEM images not shown͒. The use of MPS, on the other hand,
brought about marked improvement in gap filling ͑vide infra͒. As a
consequence, we further investigated the effect of MPS on gold
electrochemical deposition behavior and optimized gap-filling pro-
cess conditions.
2
8
sponding thiolate through the cleaving of S–H and S–S bonds,
spontaneously forming monolayers on gold surfaces.
2
8-33
The
Process condition optimization.— Superconformal
electro-
chemisorption of unfunctionalized or functional group-terminated
thiols and disulfides on gold yields surfaces that are vastly different
in water wettability, electrostatic interaction ability, and complex-
ation capability and may accordingly influence the heterogeneous
electron-transfer reactions involved. It is therefore rational to ex-
plore the use of these types of compounds as additives for the
present application.
chemical deposition of metal is not only governed by the chemical
nature of the additives used but is also affected by the concentration
of the additives in the electrolyte and by other process conditions.
Presented in Fig. 2A are cathodic polarization curves obtained ͑a͒ in
the base gold electrolyte and ͑b–e͒ in the presence of the additive
MPS at various concentrations as indicated in the caption. A plot of
2
the electrode potential at a current density of 4 mA/cm vs the con-
Downloaded on 2015-02-24 to IP 35.8.11.2 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 ͑7͒ C467-C471 ͑2006͒
centration of the additive MPS is shown in Fig. 2B. Obviously, the
C469
concentration of the additive MPS has a significant effect on depo-
sition potential; the deposition potential becomes more cathodic
with increasing MPS concentration and then gradually levels off
with the further increase in MPS concentration.
From the viewpoint of process control, an additive concentration
in the leveled-off region is preferred over the preceding region
where a slight change in additive concentration would cause a
marked change in deposition potential ͑Fig. 2͒ and most likely a
significant change in gap-filling capability. Accordingly, our inves-
tigation of the effect of additive concentration on gap filling has
focused in this region. The results obtained at the concentration of
3
mL/L additive are presented in Fig. 3, which shows the realization
of superconformal deposition of gold by detailing the profile evolu-
tion of gold deposit inside the features of interest. The thickness of
gold deposit indicated in the caption is nominal, calculated from
surface area and coulombs passed. The results were observed on full
wafers. In addition, as can be clearly seen, the features shown in
Fig. 3 are significantly more aggressive than those shown in Fig. 1.
Besides the chemical nature and concentration of the additive
under investigation, other process conditions are also important in
affecting gap filling. In general, process conditions that favor mass
transport of metal species of interest would benefit gap filling. For
instance, raising gold concentration and/or temperature benefits
mass transport of gold and thus promotes gap filling at a given
deposition rate. For electrochemical deposition of gold, however, the
degree of freedom in changing these process conditions may be
limited to a large extent, especially when a sulfite-based gold elec-
trolyte is used. For instance, raising gold concentration and/or tem-
perature is limited by bath stability. As compared with nonpulse
plating, pulse plating at a lower duty cycle generally produces better
fill results, which may be explained by the fact that pulse plating at
a lower duty cycle and a moderate pulse frequency allows the tran-
sient concentration of gold inside the features of interest to be maxi-
mized. Additionally, current density, which determines deposition
rate, has a pronounced effect on gap filling. In general, deposition at
a lower current density benefits gap filling. Increasing current den-
sity may make a superconformal deposition process become a con-
formal or even subconformal deposition process. Presented in Fig. 4
are SEMs obtained at two different current densities, with other
process conditions being the same. As is evident, at the lower cur-
rent density the features under investigation were completely filled
͑
Fig. 4A͒, whereas at the higher current density the features were
left with voids ͑Fig. 4B͒. In essence, the effect of both duty cycle
and current density on gap filling is the reflection of the impact of
mass transport of gold species on gap-filling capability.
Additive role.— In superconformal deposition of copper, thiol
and disulfide species such as MPS and bis͑3-sulfopropyl͒disulfide
͑
SPS͒ function as accelerators that make copper deposition faster at
a given deposition potential. This behavior of thiol and disulfide
species is particularly conspicuous in the presence of suppressors
such as polyalkylene glycols and/or levelers such as polyamines that
inhibit the deposition of copper. The role of thiol and disulfide spe-
cies in the deposition of gold might be different from that of the
species in the deposition of copper. The difference might be due to
the differentiation of their interaction with gold and copper. It has
been suggested that compared with a S–Cu bond that is quite ionic,
Figure 3. Cross-sectional SEMs showing the profile evolution of gold de-
posit inside trench features: ͑A͒ before electrochemical deposition of gold,
and ͑B͒ 0.23, ͑C͒ 0.47, ͑D͒ 0.55, and ͑E͒ 2.0 ␮m gold deposited in the
presence of 3 mL/L MPS.
34
a S–Au bond is strongly covalent in character.
It is well known that the presence of thiolate monolayers on gold
29
strongly hinders heterogeneous electron transfer. In the process of
electrochemical deposition where the substrate surface is continu-
ously refreshed, thiols and disulfides may only sparsely adsorb on
the substrate surface instead of forming a fully covered monolayer.
However, even the sparsely adsorbed organic species on the sub-
strate surface would exert an appreciable effect on heterogeneous
electron transfer. As shown in Fig. 2, in the deposition of gold the
presence of MPS actually causes an increase in cathodic polariza-
tion, characteristic of a suppressor that makes the deposition slower
at a given deposition potential.
One of the prevailing models proposed to describe superconfor-
mal deposition of copper is based on the assumption that thiol and
disulfide species such as MPS and SPS accumulate inside features of
interest as deposition progresses and the surface area inside the fea-
Downloaded on 2015-02-24 to IP 35.8.11.2 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C470
Journal of The Electrochemical Society, 153 ͑7͒ C467-C471 ͑2006͒
Figure 5. Chronoamperometric transients of gold deposition in the base
gold electrolyte at gold substrates: ͑a͒ underivatized; ͑b͒ derivatized in the
gold electrolyte containing 3 mL/L MPS for 2 min; and ͑c͒ derivatized in
the gold electrolyte containing 20 mL/L MPS for 10 min. Deposition
potential −350 mV. Electrode rotating speed 100 rpm.
was derivatized in the gold electrolyte containing 20 mL/L MPS for
1
0 min, gold deposition was hindered to the extent that the onset of
Figure 4. Cross-sectional SEMs of wafers deposited with gold at current
2
the deposition did not occur until about 20 min after the application
of the deposition potential. It also took much longer for the deposi-
tion, after its onset, to reach the characteristic steady-state current
density observed on the underivatized substrate.
densities of ͑A͒ 1.6 mA/cm that corresponds to a nominal deposition rate of
2
0
.10 ␮m/min, and ͑B͒ 2.4 mA/cm , corresponding to a nominal deposition
rate of 0.15 ␮m/min.
Based on the findings reported in the literature that thiols and
disulfides spontaneously adsorb on gold and form monolayers that
28,29,32,33
tures decreases. This model has been successfully used to explain
the observed superconformal deposition of copper in features and
the following momentum deposition that leads to the formation of
bumps on top of the filled features in the presence of both accelera-
tor ͑thiol or disulfide species͒ and suppressor ͑polyalkylene glycols
are resistive to heterogeneous electron transfer,
it is rational
to attribute the hindrance of the derivatized substrates to gold depo-
sition to the presence of MPS on the substrates and ascribe the
difference in hindrance degree to MPS coverage on the substrates.
The coverage of MPS on substrate c is expected to be significantly
higher than that on substrate b, as the former was derivatized longer
in a higher concentration of MPS solution. Consequently, gold depo-
sition on substrate c was more strongly hindered than on substrate b.
It is also reasonable to attribute the gradual increase in current den-
sity on the derivatized gold substrates after the onset of gold depo-
sition to the reduction in MPS coverage on the substrates. The re-
duction in the surface coverage of MPS might be due to the
incorporation of MPS into the gold deposit and/or its dissolution
into the gold electrolyte. However, considering the strong interaction
of MPS with gold, it might be more reasonable to suspect that the
reduction in MPS surface coverage was primarily due to its incor-
poration into the gold deposit as deposition progressed.
After the elimination of the adsorbed MPS, gold deposition on a
derivatized gold substrate would eventually come close to that on an
underivatized one. The trend can be seen from Fig. 5. Figure 5 also
shows that the current density on the derivatized substrates actually
surpassed the characteristic current density on the underivatized sub-
strate. The higher current density on the derivatized substrates was
most likely due to the larger electrochemical surface area as a result
of the increase in deposit surface roughness. On the derivatized gold
substrates, gold deposition might have first occurred at localized
small areas such as the pinholes of the MPS adlayers, resulting in a
rougher surface and a correspondingly larger electrochemical sur-
face area. The effect appeared to be particularly noticeable with
substrate c, where the highest current density reached about 1.1
times the characteristic current density of the underivatized sub-
strate. After the elimination of MPS from the gold substrate, the
surface roughness on the derivatized substrate might improve as the
deposition further progressed, accounting for the observation that
the current density on substrate c afterward decreased and leveled
off ͑Fig. 5͒.
11-15,35
such as polyethylene glycols͒.
In the case of gold deposition,
because of the strong interaction of thiol and disulfide species with
gold, the surface-tethered thiol and disulfide species ͑as thiolate͒
might be readily buried in the deposit during deposition. As a con-
sequence of the incorporation, coupled with the limited supply of
the species into features, the local concentration of the species inside
features would be lower than that on the surface of the substrate
onto which gold is deposited. The reduction in the concentration of
the inhibiting species inside features would allow faster deposition
of gold in the features, particularly at the bottom of the features
where the concentration of the inhibiting species is the lowest.
In an attempt to understand the role of thiol and disulfide species
such as MPS in the deposition of gold, we performed gold deposi-
tion under potentiostatic and galvanostatic conditions on gold sub-
strates underivatized and derivatized with MPS. The deposition was
carried out in the base gold electrolyte without the addition of MPS
on a gold-deposited platinum rotating-disk electrode. For deposition
on an underivatized gold substrate, the gold-deposited rotating-disk
electrode was used without further modification. For deposition on a
derivatized gold substrate, the electrode was modified with MPS at
room temperature in the gold electrolyte containing MPS. The modi-
fied electrode was rinsed with deionized water before being trans-
ferred to the base gold electrolyte for gold deposition.
Figure 5 shows the chronoamperometric transients observed on
the underivatized and derivatized gold substrates. At the experimen-
tal condition indicated in the caption, gold deposition on the un-
derivatized substrate a occurred upon the application of the deposi-
tion potential and reached a steady-state deposition rate ͑or current
density͒, which is representative of the current density used to
achieve superconformal deposition of gold, in about 1 min. On the
substrate b that was derivatized in 3 mL/L MPS gold electrolyte for
2
1
min, gold deposition was noticeably hindered; it took nearly
0 min for the deposition to reach the steady-state current density
Consistent with the chronoamperometric results, the chronopo-
tentiometric results also suggest that the presence of MPS on a gold
surface suppresses the deposition of gold, and the degree of suppres-
observed with the underivatized substrate. On the substrate c that
Downloaded on 2015-02-24 to IP 35.8.11.2 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 ͑7͒ C467-C471 ͑2006͒
C471
superconformal deposition of gold in the presence of thallium,
which functions as a depolarizer, and MPS was most likely driven
by the depletion of MPS inside the features of interest as a result of
its consumption through incorporation into the deposit in conjunc-
tion with its diffusion-limited supply. On this basis, it can be ex-
pected that other water-soluble thiol and disulfide species such as
carboxylic acid-terminated thiols and disulfides as well as sulfonic
acid-terminated disulfides that strongly adsorb on gold and hinder
heterogeneous electron transfer would have similar effects on the
deposition of gold and promote superconformal deposition.
Semitool, Inc., assisted in meeting the publication costs of this article.
References
1
. D. Edelstein, J. Heidenreich, R. Goldblatt, W. Cote, C. Uzoh, N. Lustig, P. Roper,
T. McDevitt, W. Motsiff, A. Simon, J. Dukovic, R. Wachnik, H. Rathore, R.
Schulz, L. Su, S. Luce, and J. Slattery, Tech. Dig. - Int. Electron Devices Meet.,
Figure 6. Chronopotentiometric transients of gold deposition in the base
gold electrolyte at gold substrates: ͑1͒ underivatized; ͑2͒ derivatized in the
gold electrolyte containing 3 mL/L MPS for 2 min; ͑3͒ derivatized in the
gold electrolyte containing 20 mL/L MPS for 10 min; and ͑4͒ derivatized in
the gold electrolyte containing 20 mL/L MPS overnight. Deposition current
1
997, 773.
2. S. Venkatesan, A. V. Gelatos, V. Misra, R. Islam, B. Smith, J. Cope, B. Wilson, D.
Tuttle, R. Cardwell, I. Yang, P. V. Gilbert, R. Woodruff, R. Bajaj, S. Das, J. Farkas,
D. Watts, C. King, P. Crabtree, T. Sparks, T. Lii, C. Simpson, A. Jain, M. Herrick,
C. Capasso, S. Anderson, R. Venkatraman, S. Filipiak, B. Flordalice, K. Reid, J.
Klein, E. J. Weitzman, and H. Kawasaki, Tech. Dig. - Int. Electron Devices Meet.,
2
density 2 mA/cm . Electrode rotating speed 100 rpm.
1
997, 769.
sion depends on the coverage of MPS on the surface ͑Fig. 6͒. At the
galvanostatic condition the elimination of the surface-adsorbed MPS
appeared to be significantly faster than at the previous potentiostatic
condition, although both of them are representative of the deposition
current and the corresponding deposition potential under which su-
perconformal deposition of gold was achieved. As shown in Fig. 6,
steady-state deposition on the derivatized substrates was reached in
less than 2 min on all the derivatized substrates, including the one
derivatized in 20 mL/L MPS overnight. In the galvanostatic depo-
sition, which more closely resembles the actual gold deposition on a
wafer than the potentiostatic deposition, the surface-adsorbed MPS
might be more readily buried in the deposit because of the faster
initial deposition rate resulting from the higher deposition potential
3. P. C. Andricacos, C. Uzoh, J. O. Dukovic, J. Horkans, and H. Deligianni, IBM J.
Res. Dev., 42, 567 ͑1998͒.
. P. Singer, Semicond. Int., 21, 90 ͑1998͒.
. M. E. Gross, C. Lingk, W. L. Brown, and R. Drese, Solid State Technol., 42, 47
4
5
͑1999͒.
6. R. Williams, Modern GaAs Processing Methods, Artech House, Boston, MA
͑1990͒.
7
. K. Kondo, T. Matsumoto, and K. Watanabe, J. Electrochem. Soc., 151, C250
2004͒.
. T. P. Moffat, D. Wheeler, and D. Josell, J. Electrochem. Soc., 151, C262 ͑2004͒.
͑
8
9. M. Hayase, M. Taketani, K. Aizawa, T. Hatsuzawa, and K. Hayabusa, Electrochem.
Solid-State Lett., 5, C98 ͑2002͒.
0. Y. Cao, P. Taephaisitphongse, R. Chalupa, and A. C. West, J. Electrochem. Soc.,
1
1
48, C466 ͑2001͒.
1
1. D. Josell, D. Wheeler, W. H. Huber, J. E. Bonevich, and T. P. Moffat, J. Electro-
chem. Soc., 148, C767 ͑2001͒.
͑
Fig. 6͒.
The mechanism that is responsible for superconformal deposition
12. A. C. West, S. Mayer, and J. Reid, Electrochem. Solid-State Lett., 4, C50 ͑2001͒.
1
3. T. P. Moffat, D. Wheeler, W. H. Hubber, and D. Josell, Electrochem. Solid-State
Lett., 4, C26 ͑2001͒.
of gold in the presence of such species as MPS is certainly a subject
that needs to be further explored. For instance, the grain-refiner
thallium that functions as a depolarizer in the electrolyte may play
an important role in affecting the gap-filling capability of the elec-
trolyte. It was observed that the introduction of thallium into the
gold electrolyte that contains MPS but no thallium significantly im-
proves deposit surface roughness. However, the results obtained so
far appear to suggest that the driving force leading to superconfor-
mal deposition of gold was primarily due to the depletion of thiol or
disulfide species inside the features of interest. This is analogous to
the copper superfilling model proposed based on the assumption of
1
4. Y. H. Im, M. O. Bloomfield, S. Sen, and T. S. Cale, Electrochem. Solid-State Lett.,
6
, C42 ͑2003͒.
15. A. C. West, J. Electrochem. Soc., 147, 227 ͑2000͒.
1
1
6. M. Georgiadou, D. Veyret, R. L. Sani, and R. C. Alkire, J. Electrochem. Soc., 148,
C54 ͑2001͒.
7. T. P. Moffat, J. E. Bonevich, W. H. Huber, A. Stanishevsky, D. R. Kelly, G. R.
Stafford, and D. Josell, J. Electrochem. Soc., 147, 4524 ͑2000͒.
18. Y. Jin, K. Kondo, Y. Suzuki, T. Matsumoto, and D. P. Barley, Electrochem. Solid-
State Lett., 8, C6 ͑2005͒.
9. M. Hasegawa, Y. Negishi, T. Nakanishi, and T. Osaka, J. Electrochem. Soc., 152,
C221 ͑2005͒.
0. K.-C. Shim, H.-B. Lee, O.-K. Kwon, H.-S. Park, W. Kob, and S.-W. Kang,
J. Electrochem. Soc., 149, G109 ͑2002͒.
1
2
3
differential inhibition by diffusion-controlled additives. The mo-
21. D. Josell, S. Kim, D. Wheeler, T. P. Moffat, and S. G. Pyo, J. Electrochem. Soc.,
50, C368 ͑2003͒.
2
1
mentum deposition, which is common in superconformal deposition
of copper, was not observed in the present superconformal deposi-
tion of gold.
2. T. P. Moffat, B. Baker, D. Wheeler, J. E. Bonevich, M. Edelstein, D. R. Kelly,
L. Gan, G. R. Stafford, P. J. Chen, W. F. Egelhoff, and D. Josell, J. Electrochem.
Soc., 149, C423 ͑2002͒.
2
3. B. C. Baker, M. Freeman, B. Melnick, D. Wheeler, D. Josell, and T. P. Moffat,
J. Electrochem. Soc., 150, C61 ͑2003͒.
Conclusions
2
2
4. E. J. Ahn and J. J. Kim, Electrochem. Solid-State Lett., 7, C118 ͑2004͒.
5. Z. Hu and T. Ritzdorf, in Electrochemical Processing in ULSI and MEMS, H.
Deligianni, S. T. Mayer, T. P. Moffat, and G. R. Stafford, Editors, PV 2004-17,
p.13, The Electrochemical Society Proceedings Series, Pennington, NJ ͑2005͒.
6. D. Josell, C. R. Beauchamp, D. R. Kelly, C. A. Witt, and T. P. Moffat, Electrochem.
Solid-State Lett., 8, C54 ͑2005͒.
Superconformal electrochemical deposition of gold has been
demonstrated by utilizing a thiol species, MPS, as an additive in a
sulfite-based gold electrolyte. The developed process was proven to
be a true bottom-up process with which aggressive features such as
straight-walled and re-entrant ones can be successfully filled without
seam voids. The achievement of superconformal deposition of gold
allows us to completely fill fine features of interest and thus makes
the implementation of gold metallization through damascene pro-
cessing feasible.
2
2
2
7. L. W. Graham, Ph.D. Thesis, Rensselaer Polytechnic Institute, New York ͑1999͒.
8. C.-J. Zhong and M. D. Porter, J. Am. Chem. Soc., 116, 11616 ͑1994͒.
29. M. D. Porter, T. B. Bright, D. L. Allara, and C. E. D. Chidsey, J. Am. Chem. Soc.,
109, 3559 ͑1987͒.
3
0. C. D. Bain, E. B. Troughton, Y.-T. Tao, J. Evall, G. M. Whitesides, and R. G.
Nuzzo, J. Am. Chem. Soc., 111, 321 ͑1989͒.
Voltammetric, chronoamperometric, and chronopotentiometric
studies reveal that the thiol species MPS in the present superconfor-
mal deposition system functions as a suppressor that hinders elec-
trochemical deposition of gold, in contrast to its function as an ac-
celerator in a superconformal copper deposition system. The
experimental results, coupled with the expected strong interaction of
the thiol species MPS with gold substrates, lead us to believe that
3
1. C. D. Bain and G. M. Whitesides, J. Am. Chem. Soc., 111, 7164 ͑1989͒.
32. J. J. Hickman, P. E. Laibinis, D. I. Auerbach, C. Zou, T. J. Gardner, G. M.
Whitesides, and M. S. Wrighton, Langmuir, 8, 357 ͑1992͒.
3. C.-J. Zhong and M. D. Porter, Anal. Chem., 67, 709A ͑1995͒.
3
3
3
4. H. Rieley and G. K. Kendall, Langmuir, 15, 8867 ͑1999͒.
5. T. Ritzdorf, D. Fulton, and L. Chen, in Advanced Metallization Conference (AMC)/
1
999, M. E. Gross, T. Gessner, N. Kobayashi, and Y. Yasuda, Editors, p. 101,
Materials Research Society Conference Proceedings, Warrendale, PA ͑2000͒.
Downloaded on 2015-02-24 to IP 35.8.11.2 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).