A modified Damascene electrodeposition process for bottom-up filling of recessed surface features

Article abstract of DOI:10.1016/j.electacta.2010.07.045

A modification of the conventional Damascene metallization process is described whereby selective removal of the thin wetting/seed layer from the sidewalls and free surfaces enables selective nucleation and bottom-up electrodeposition of metals and alloys in recessed surface features. The process is demonstrated by filling sub-micrometer trenches with electrodeposited Ni. A conventional PVD Cu seed layer is etched to remove Cu from the sidewalls and free surface while leaving a continuous Cu wetting layer intact on the trench bottom. The underlying non-wetting barrier layer provides a conductive path for electrodeposition from contacts on the perimeter of the work piece to the trench bottom. The robustness of the bottom-up Ni electrodeposition process is greatly increased by the addition of 2-mercaptobenzimidazole (MBI) to the plating bath. The additive hinders spurious nucleation of Ni on residual Cu patches that may remain on the free surface.

Full text of DOI:10.1016/j.electacta.2010.07.045

Electrochimica Acta 55 (2010) 8527–8531
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
A modified Damascene electrodeposition process for bottom-up filling of
recessed surface features
Chang Hwa Lee, Thomas P. Moffat∗
Materials Science and Engineering Laboratory, National Institute of Standards and Technology, Gaithersburg, MD 20899, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 6 April 2010
Received in revised form 9 July 2010
Accepted 17 July 2010
Available online 23 July 2010
A modification of the conventional Damascene metallization process is described whereby selective
removal of the thin wetting/seed layer from the sidewalls and free surfaces enables selective nucle-
ation and bottom-up electrodeposition of metals and alloys in recessed surface features. The process
is demonstrated by filling sub-micrometer trenches with electrodeposited Ni. A conventional PVD Cu
seed layer is etched to remove Cu from the sidewalls and free surface while leaving a continuous Cu
wetting layer intact on the trench bottom. The underlying non-wetting barrier layer provides a conduc-
tive path for electrodeposition from contacts on the perimeter of the work piece to the trench bottom.
The robustness of the bottom-up Ni electrodeposition process is greatly increased by the addition of
2-mercaptobenzimidazole (MBI) to the plating bath. The additive hinders spurious nucleation of Ni on
residual Cu patches that may remain on the free surface.
Keywords:
Damascene process
Nickel
Superconformal deposition
Bottom-up filling
Mercaptobenzimidazole
Published by Elsevier Ltd.
1. Introduction
multilayered 3D structures without the need to maintain electrical
continuity between subsequent layers. The conventional Cu Dam-
The realization of three-dimensional (3D) metal structures has
become a central theme in the densification of microelectronics
and energy conversion devices ranging from through-silicon-vias
(TSV) and micro-electro-mechanical systems (MEMS) to bat-
tery electrodes and the metallization of photovoltaic devices
[1–8].
ascene process relies on additive-derived superconformal metal
electrodeposition to robustly fill trenches and vias on metallized
surfaces. Competitive interactions between inhibiting and accel-
erating adsorbates convoluted with area change enable void-free
superconformal filling of recessed features. The same deposition
mechanism has also been shown to be applicable to other passive
interconnect materials such as Ag and Au [16–18]. More recently,
the prospect of designing and building active, functional materi-
als in 3D metallized structures has received attention and reports
of superconformal electrodeposition of Ni and related iron-group
magnetic alloys are available [19–22]. Addition of a single addi-
tive, such as the mercaptobenzimidazole derivatives, MBIS and
MBI [20,21], or polymers such as polyethyleneimmine [19], to an
additive-free Ni plating bath has been shown to be effective in
generating void-free trench filling by a mechanism that is distinct
from that used for the Group IB metals [15–18]. The feature filling
involves an inhibition-breakdown process that initially gives rise to
a v-notch shaped growth front within the trench followed by geo-
metric leveling. Two challenges to practical implementation of the
process remain. Firstly, the plated overburden tends to vary with
feature size, which hampers subsequent CMP processing. Secondly,
in some instances the superconformal growth dynamic is unable to
compensate for non-uniformities associated with physical vapor
deposited (PVD) Cu seed layers. For example, reentrant sidewall
angles associated with Cu seed layer build up and overhang at the
trench mouth can lead to center and top voids in the final plated
structure that is especially evident in narrow features [21].
Conventional 2D (or 2.5D) processes, such as through-mask
plating, templated deposition, or LIGA (lithography, electroplat-
ing, and molding), have been utilized to manufacture patterned
microstructures made of iron group metals, Ni, Co, Fe and alloys
thereof, for use in magnetic recording heads, sensors, MEMS, etc.
[6–13]. A typical process involves metal electrodeposition on a con-
ductive substrate that is masked by a non-conducting patterned
photoresist or a porous oxide template that guides the vertical
growth of the metal deposit. Although this has proven to be a
very powerful fabrication method, there are serious constraints
to building complex multilayered 3D metal structures. For exam-
ple, electrical addressability is required for the electrodeposition
process, but some device designs might call for electrical isolation
of individual segments during subsequent processing steps. The
latter curtails the ability to build complex 3D structures. In con-
trast, the Damascene process that has been widely used in building
Cu interconnects [14,15] offers an efficient way to build complex
∗
Corresponding author. Tel.: +1 301 975 2143; fax: +1 301 926 7679.
E-mail address: thomas.moffat@nist.gov (T.P. Moffat).
0013-4686/$ – see front matter. Published by Elsevier Ltd.
doi:10.1016/j.electacta.2010.07.045

8528
C.H. Lee, T.P. Moffat / Electrochimica Acta 55 (2010) 8527–8531
More recently, attempts have been made to selectively poi-
defect tolerant directional etch systems based on monolayer resists
[26].
son or prevent electrodeposition on the conductive free surface in
order to optimize, and potentially simplify, the Damascene pro-
cess flow. Techniques ranging from PVD of a non-wetting Ta layer
[23] to derivatization of the free surface with alkane thiol films
[12,13,24] have been shown to selectively block metal deposition
reactions. Likewise, processes that couple mechanical and chem-
ical action to minimize deposition on the free surface, relative to
recessed features, have also been explored [25]. Most recently, a
mixed-mode Damascene like-variation of through-mask deposi-
tion has been explored [22]. This process involved selective etching
to remove all conductive materials from the free surface of the
trench-patterned substrate followed by electroplating where cur-
rent was directed to the trenches via a common back contact
in a manner congruent with traditional through-mask plating.
The remaining metallized sidewalls and trench bottom resulted
in superconformal feature filling. A significant limitation of this
scheme is that the entire structure remains electrically connected
via the original back plane contact. In principle, the back plane of
a single layer structure could be removed to electrically isolate
individual features. However, it is not obvious how this process
could be extended to build the intricate multilayered 3D struc-
tures.
In this paper we introduce another mixed-mode deposition
process whereby the advantages of 3D electrical addressabil-
ity associated with Damascene processing are coupled with
the bottom-up feature filling dynamic typically associated with
through-mask processing. This is accomplished by selectively
removing the wetting Cu seed layer from the free surface and trench
sidewalls while leaving the conductive barrier layer intact across
the topographically patterned work piece. The Cu seed layer that
remains on the bottom surface of the recessed features serves as
a site for electrodeposition while the non-wetting sidewalls pro-
vide an electrical conduction pathway as well as templating for
bottom-up feature filling. The bottom-up growth mode also offers
the prospect of controlling the orientation of the deposited material
by appropriate consideration of the barrier/seed layer architecture.
The properties of the interface formed between the electrodeposits
and the “non-wetting” sidewalls remain to be investigated. Partial
oxidation of the barrier layer during etching and transfer to the plat-
ing bath is likely however conventional barriers are thick enough to
ensure suitable conductivity remains. The robustness of the filling
process is dependent on nucleation being restricted to the bottom
surface of the metallized trench. This is accomplished by addition
of an inhibitor to the plating electrolyte to block sporadic nucle-
ation and growth of material on the free surface between trenches.
This strategy is analogous to that recently used in the formulation of
2. Experimental
The substrates used for this study consist of a PVD Cu seed
layer over a Ta barrier layer deposited on Si wafers that were
lithographically patterned with trench arrays with various feature
widths. Ni deposition was performed under potentiostatic con-
ditions using a three-electrode cell, with a Pt wire anode and a
saturated calomel reference electrode (SCE). A Watts-type bath
consisting of 1.0 mol L−1 NiSO4·6H2O, 0.2 mol L−1 NiCl2·6H2O, and
0.5 mol L−1 H3BO3 (pH 3.4) was used in both the absence and pres-
ence of 150 ␮mol L−1 2-mercaptobenzimidazole (MBI). Selective
etching to remove the Cu seed layer from the free-surface and
trench sidewalls was performed by immersing the substrates in
a mixture of 0.02 mol L−1 citric acid and 0.2 mol L−1 (2 wt%) H2O2
(pH 4) for 4 min prior to the metal deposition process. The citric
acid and H2O2 are well-known as a complexing agent and oxi-
dizer, respectively. The combination has been effectively used for
chemical mechanical polishing (CMP) of Cu [27,28]. Following etch-
ing, the specimens were rinsed with 18 Mꢀ cm DI water, dried
with N2 gas flow, and then immediately immersed into the plat-
ing bath with the potential applied. Cleaved cross-sections of the
Ni filled trenches were examined using a field emission scanning
electron microscopy (FE-SEM). Irregular morphologies are appar-
ent in some cross-section images. These are simply artifacts of the
specimen cleavage process that leads to ductile tearing of the elec-
trodeposited Ni and fracture of the interfaces between the Ni and
the Cu/Ta under-layers.
3. Results and discussion
FE-SEM images of cross-sectioned Cu seeded trench arrays prior
to Cu seed removal and Ni deposition are shown in Fig. 1. The trench
widths range from 150 nm to 2200 nm. Inspection reveals signifi-
cant differences in the Cu film thickness on the top, bottom and
sidewalls of the trenches. For the larger features examined, TEM
analysis (not shown) reveals 25 nm of Ta and 120 nm of Cu on the
horizontal surfaces (i.e. free surface and trench bottom) while the
sidewalls are covered with only 5 nm Ta and 10 nm Cu. The differ-
ence between the horizontal and vertical surfaces is a consequence
of the line-of-sight nature of the PVD process. This leads to signif-
icant lateral growth on concave features evident by the significant
“overhang” of the Cu seed layer at the top corners of the trenches
with the corresponding reduction of thickness at the shadowed bot-
tom corners. The reentrant nature of the upper sidewall geometry
Fig. 1. Cross-section of Cu seeded trench arrays. Note the thin sidewall coverage, build-up and overhang on the top corners of the trenches.

C.H. Lee, T.P. Moffat / Electrochimica Acta 55 (2010) 8527–8531
8529
Fig. 4. Current transients at −0.9 V SCE for Ni electrodeposition (a) on a Cu seeded
substrate in an additive-free plating electrolyte, (b) a selectively etched substrates
in an additive-free plating electrolyte, (c) a selectively etched substrates in an elec-
trolyte containing 150 ␮mol L−1 MBI.
to facilitate selective nucleation during subsequent Ni electroplat-
ing. The Cu seeded specimens were etched for 4 min in a mixture
of 0.02 mol L−1 citrate and 0.2 mol L−1 H2O2. The PVD derived
thickness variations naturally lead to rapid loss of Cu from the
sidewalls. The relative accessibility of the free surface compared to
the recessed trench bottom provides an avenue for development
of a differential, topographically dependent etch rate between the
respective surfaces. Following etching the specimens were rinsed
with DI water and transferred into an additive-free Ni plating bath
at −0.9 V SCE for 300 s for Ni plating. As shown in Fig. 3 in the
wider (e.g. 680 nm) trenches Ni is exclusively deposited on the
bottom surface. In contrast, significant Ni deposition is evident on
the free surfaces neighboring the finer (460 nm down to 150 nm)
trenches. This is attributed to some residual amount of the Cu seed
layer still present on the free surface. The remnant Cu seed on the
free surface is the result of the similar thickness of the original Cu
seed layer on the free surface and trench bottom combined with
the effect of pattern density on etchant depletion. In principle this
effect may be circumvented in a number of ways. Stronger fea-
ture height dependent etching might be induced by coupling the
existing chemistry with a hydrodynamic shear motion in a man-
ner analogous to conventional chemical mechanical planarization
process. Furthermore, modification of the etching chemistry with
Fig. 2. Schematic diagram of a modified Damascene processes that uses controlled
nucleation and substrate topography to guide a bottom-up electrodeposition pro-
cess.
represents a challenge to subsequent feature filling and effectively
narrows the processing window, thereby increasing the likelihood
of undesired void formation during subsequent feature filling by
electrodeposition. For a given seed layer thickness the negative
effect is increasingly severe for narrower features.
Additive-free Ni deposition on such seed layers results in a
continuous center-line void that forms following collision of the
opposing sidewalls that grow conformally from the top corners of
the trench. In contrast, Ni deposition in the presence of inhibitors,
such as PEI, MBIS and MBI, has been shown to result in transient
inhibition of the metal deposition on the upper corners that makes
void-free filling of sub-micrometer features possible [19–21].
As an alternative to the conventional Damascene process, the
scheme shown in Fig. 2 is explored herein. Etching time was con-
trolled to selectively remove the Cu seed layer from the sidewalls
and free surface while leaving Cu on the bottom trench surface
Fig. 3. Ni electrodeposition on Cu seeded trench patterned substrates that have been etched to remove Cu form the sidewalls and top surface. Ni deposition was performed
in an additive-free solution at a growth potential of −0.9 V SCE for 300 s.

8530
C.H. Lee, T.P. Moffat / Electrochimica Acta 55 (2010) 8527–8531
Fig. 5. Ni electrodeposition on Cu seeded trench patterned substrates that have been etched to remove Cu form the sidewalls and top surface. Ni deposition was performed
in a plating electrolyte containing 150 ␮mol L−1 MBI at a growth potential of −0.9 V SCE for (a) 300 s and (b) 600 s.
additives might be used to selectively enhance the inhibition of Cu
dissolution from the trench bottom during the etching step. Such an
approach might require an additional step to remove the inhibitor
in order to guarantee effective nucleation of Ni on Cu during the
subsequent plating operation. A simpler variant to the above is to
use an inhibitor in the Ni electrodeposition process that preferen-
tially interacts with any residual Cu on the free surface to hinder
Ni deposition thereon. This approach was explored using MBI addi-
tions to the Ni plating bath.
The influence of the etching treatment and MBI additions on
the chronoamperometric response associated with Ni deposition
on the patterned substrates is shown in Fig. 4. The transient
for additive-free Ni deposition on the etched substrate is similar
to that for an as-received Cu seeded substrate. The diminished
magnitude of the current reflects the loss of electroactive area
due to the absence of metal deposition on the etched sidewalls.
In the presence of MBI, significant inhibition is evident associ-
ated with the formation of a blocking layer that is followed by
its subsequent breakdown. The activation process corresponds to
autocatalytic breakdown of the inhibiting layer whereby the flux
of MBI required to maintain the passive state is overwhelmed by
its consumption induced by metal deposition. Similar chronoam-
perometry has been reported for Ni and Co deposition in the
presence of a sulfonated version of the molecule, known as MBIS
[20,21].
on the Cu seed layer that remains on the trench bottom is unam-
biguous. Examination of the free surface of the widest trenches
indicates that the Cu seed was effectively removed from the free
surface. For trench array with feature widths smaller than 400 nm
some remnant of the Cu seed layer is clearly evident on the free
surface segments between the trenches; However, MBI is remark-
ably effective in suppressing Ni deposition on these sites compared
to the additive-free case (Fig. 3). After 300 s, bottom-up filling has
progressed furthest in the widest features. This is in direct contrast
to the effect of MBI or MBIS on Ni deposition on continuous Cu
seed layers used in the conventional Damascene process [20,21]. At
longer deposition times, 600 s, the growth front in the finer feature
appears to catch up with bottom-up feature filling evident across
trench arrays while negligible deposition occurs on the interven-
ing free-surface areas. A few remnant Cu spots are still evident on
the free surface but little Ni deposition has occurred reflecting the
effectiveness of MBI. The near uniform growth front across the wide
feature sizes may be beneficial to any subsequent CMP process.
Looking to the future, coupling this processing scheme
with more complex additive combinations (i.e. accelerators and
inhibitors) offers the prospect of further increases in the magnitude
of the differential bottom-up growth dynamic.
4. Conclusions
The combination of selective removal of the Cu seed layer by
etching followed by Ni deposition in the presence of MBI leads to
very effective bottom-up feature filling. A series of images following
300 s and 600 s of Ni deposition are shown in Fig. 5a and b, respec-
tively. After 300 s, bottom-up Ni deposition by selective deposition
A modified or hybrid Damascene electrodeposition process
was demonstrated. Seed layer modification combined with an
inhibiting plating additive enables selective nucleation at trench
bottoms followed by bottom-up feature filling with growth con-
tours reminiscent of the through-mask plating process. Specifically,

C.H. Lee, T.P. Moffat / Electrochimica Acta 55 (2010) 8527–8531
8531
Ni electrodeposition from an MBI containing electrolyte is con-
trolled by selective nucleation on a Cu seed layer that covers
only the bottom surface of the trenches thereby enabling exclu-
sive bottom-up feature filling. The MBI plating additives serves to
preferentially inhibit Ni deposition on any remnant Cu or other
nucleation sites on the free surface, greatly enhancing the robust-
ness of the overall process. Overall, the ability to restrict nucleation
to the trench bottoms exploits the short coming of the line-of-sight
PVD Cu process. Patterning methods that enable controlled place-
ment, and perhaps texturing, of the nucleating seed layer on top of
the non-wetting conductive barrier layer promise new opportuni-
ties to design and build heretofore unrealized microstructures.
[8] J.J. Kelly, N.Y.C. Yang, Microsyst. Technol. 11 (2005) 331.
[9] P.C. Andricacos, N. Robertson, IBM J. Res. Dev. 42 (1998) 671.
[10] O. Azzaroni, P.L. Schilardi, R.C. Salvarezza, Appl. Phys. Lett. 80 (2002) 1061.
[11] T.M. Whitney, J.S. Jiang, P.C. Searson, C.L. Chien, Science 261 (1993) 1316.
[12] T.P. Moffat, H.J. Yang, J. Electrochem. Soc. 142 (1995) L220.
[13] N.S. Pesika, A. Radisic, K.J. Stebbe, P.C. Searson, Nano Lett. 6 (2006) 1023.
[14] P.C. Andricacos, C. Uzoh, J.O. Dukovic, J. Horkans, H. Deligianni, IBM J. Res. Dev.
42 (1998) 567.
[15] T.P. Moffat, D. Wheeler, M.D. Edelstein, D. Josell, IBM J. Res. Dev. 49 (2005) 19.
[16] B.C. Baker, C. Witt, D. Wheeler, D. Josell, T.P. Moffat, Electrochem. Solid State
Lett. 6 (2003) C67.
[17] E.J. Ahn, J.J. Kim, Electrochem. Solid State Lett. 7 (2004) C118.
[18] D. Josell, D. Wheeler, T.P. Moffat, J. Electrochem. Soc. 153 (2006) C11.
[19] S.-K. Kim, J.E. Bonevich, D. Josell, T.P. Moffat, J. Electrochem. Soc. 154 (2007)
D443.
[20] C.H. Lee, J.E. Bonevich, J.E. Davies, T.P. Moffat, J. Electrochem. Soc. 155 (2008)
D499.
[21] C.H. Lee, J.E. Bonevich, J.E. Davies, T.P. Moffat, J. Electrochem. Soc. 156 (2009)
D301.
References
[22] Y. Nakamaru, H. Honma, Trans. Inst. Met. Finish 87 (2009) 259.
[23] O. Luhn, A. Radisic, P.M. Vereecken, C. Van Hoof, W. Ruythooren, J.-P. Celis,
Electrochem. Solid State Lett. 12 (2009) D39;
O. Luhn, A. Radisic, C. Van Hoof, W. Ruythooren, J.-P. Celis, J. Electrochem. Soc.
157 (2010) D242.
[24] S.-K. Kim, J.J. Kim, Electrochem. Solid State Lett. 7 (2004) C101.
[25] B.M. Basol, J. Electrochem. Soc. 151 (2004) C765.
[26] M. Geissler, H. Schmid, A. Boetsch, B. Michel, E. Delamarche, Langmuir 18 (2002)
2374.
[1] Z.Y. Fan, H. Razavi, J.W. Do, A. Moriwaki, O. Ergen, Y.L. Chueh, P.W. Leu, J.C. Ho, T.
Takahashi, L.A. Reichertz, S. Neale, K. Yu, M. Wu, J.W. Ager, A. Javey, Nat. Mater.
8 (2009) 648.
[2] E. Van Kerschaver, G. Beaucarne, Prog. Photovoltaics 14 (2006) 107.
[3] J.A.D. Jensen, P. Moller, T. Bruton, N. Mason, R. Russell, J. Hadley, P. Verhoeven,
A. Matthewson, J. Electrochem. Soc. 150 (2003) G49.
[4] P.H.L. Notten, F. Roozeboom, R.A.H. Niessen, L. Baggetto, Adv. Mater. 19 (2007)
4564.
[5] A.W. Topol, D.C. La Tulipe, L. Shi, D.J. Frank, K. Bernstein, S.E. Steen, A. Kumar,
G.U. Singco, A.M. Young, K.W. Guarini, M. Ieong, IBM J. Res. Dev. 50 (2006) 491.
[6] W. Ehrfeld, Electrochim. Acta 48 (2003) 2857.
[27] M.C. Kang, H.S. Nam, H.Y. Won, S. Jeong, H. Jeong, J.J. Kim, Electrochem. Solid
State Lett. 11 (2008) H32.
[28] D.H. Eom, I.K. Kim, J.H. Han, J.G. Park, J. Electrochem. Soc. 154 (2007) D38.
[7] L.T. Romankiw, Electrochim. Acta 42 (1997) 2985.