ELECTROLESS DEPOSITION OF COPPER
AND COPPER-MANGANESE ALLOY FOR
APPLICATION IN INTERCONNECT
METALLIZATION
by
LU YU
Submitted in partial fulfillment of the requirements
For the degree of Master of Science
Thesis Advisor: Professor Rohan Akolkar
Department of Chemical Engineering
CASE WESTERN RESERVE UNIVERSITY
May 2014
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the thesis/dissertation of
LU YU
_____________________________________________
*
Master of Science
Candidate for the _____________________________degree
.
Prof. Rohan Akolkar
(signed)____________________________________________
(Chair of the committee)
Prof. Uziel Landau
___________________________________________________
Prof. Chung Chiun Liu
___________________________________________________
___________________________________________________
___________________________________________________
4-1-2014
(date)______________________________________________
*We hereby certify that written approval has been obtained for
any proprietary material contained therein
1
Acknowledgement
To my parents, Cheng Wang and Xiaolin Yu in China for their endless love
and support. I would also like to thank Prof. Rohan Akolkar for his guidance and
mentorship throughout my master project, as well as Prof. Uziel Landau and Prof.
Chung-Chiun Liu for serving on my committee. Finally, I would like to acknowledge
funding from Atotech for the project.
2
Table of Contents:
List of Figures ................................................................................................................ 5
Abstract .......................................................................................................................... 9
Chapter 1: Introduction ................................................................................................ 11
1-1: Damascene Process for Copper Interconnect Metallization ............................ 11
1-2: Challenges of Future Interconnect Scheme ..................................................... 13
1-3: Advantages of Electroless Cu Deposition ....................................................... 14
Chapter 2: Experimental Methodology ........................................................................ 17
2-1: General Electrochemical Setup ........................................................................ 17
2-2: Electroless Deposition of Cu-Mn Alloy ........................................................... 18
2-3: Underpotential Deposition of Mn on Au and Cu ............................................. 18
2-4: Electroless Cu Nucleation on Ru ...................................................................... 19
2-5: Material Characterization Techniques .............................................................. 20
Atomic Force Microscope (AFM) ....................................................................... 20
Transmission Electron Microscope (TEM) ......................................................... 20
Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS) ....................... 21
Quartz Crystal Microbalance (QCM) .................................................................. 21
Chapter 3: Characterization of Cu-Mn Film Deposited by Electroless Deposition..... 24
Chapter 4: Underpotential Deposition of Manganese (Mn) ........................................ 30
Chapter 4-1: Underpotential Deposition of Manganese (Mn) on Gold (Au)........... 32
Chapter 4-2: Underpotential Deposition of Manganese (Mn) on Copper (Cu) ....... 40
Chapter 4-3: Conclusions......................................................................................... 43
Chapter 4-4: Future Work ........................................................................................ 44
3
Chapter 5: Electroless Cu Nucleation on Ru ............................................................... 47
Chapter 5-1: Nucleation Mechanism onto a Foreign Substrate ............................... 48
Chapter 5-2: Electroless Cu Deposition on Ru ........................................................ 51
Chapter 5-3: Conclusions......................................................................................... 59
4
List of Figures
Figure 1-1: Interconnect fabrication process (trench). (a) Deposition of SiN,
SiO2 and other low-k dielectric material. (b) Trench formation by etching.
(c) Barrier layer (TaN) and Cu seed layer deposition by PVD.
(d) Bottom-up fill by Cu electrodeposition. (Adapted from D. R. Frear1). (p.12)
Figure 1-2: SEM image of multilayer copper interconnects in an integrated circuit
chip3. Photograph courtesy of IBM Corporation, 1997. (p.12)
Figure 1-3: (a) Preferential Cu seed layer growth at feature rim by physical vapor
deposition. (b) Proposed ‘uniform’ Cu seed layer formation by electroless Cu
deposition. (p.13)
Figure 2-1: Schematic of the three-electrode electrochemical cell setup with
heating jacket for all electrochemical measurements. (p.17)
Figure 2-2: Schematic of Electrochemical Quartz Crystal Microbalance (EQCM)
setup used to investigate the underpotential deposition mechanism of Mn. (p.19)
Figure 3-1: Mixed potential transients during electroless Cu-Mn deposition on a
PVD Cu wafer. Mixed potential during 120 s deposition is always more anodic
than Mn’s standard reduction potential (-1.42 V vs. SCE). (p.26)
Figure 3-2: Negative ion TOF SIMS spectra of a Cu-Mn alloy film deposited by
electroless deposition on PVD Cu wafer for 120 s. TOF SIMS result shows there is
Mn in the electroless Cu-Mn film. (p.27)
Figure 3-3: Effect of post-annealing on the Mn content on the surface of an
electroless Cu-Mn alloy film deposited on Ru coated Si wafers. (a) as-deposited
electroless Cu-Mn film. (b) annealed electroless Cu-Mn film. TOF SIMS data
indicates that Mn is mobile and segregate at the film surface. (p.28)
5
Figure 4-1: Pourbaix diagram for Manganese (Mn) in aqueous solution1. In acid
solution, Mn electrodeposition occurs at potentials more cathodic than -1.42 V vs.
SCE (i.e., -1.18 V vs. SHE). (p.31)
Figure 4-2: Polarization and corresponding quartz crystal microbalance data
using a 5 MHz polished gold crystal in solution containing 40 mM MnSO4 at pH
of 3.8. The scan was stopped at -1.1 V vs. SCE. Scan rate: 50 mV/s. Electrode Area:
1.45 cm2. (p.33)
Figure 4-3: Polarization and corresponding quartz crystal microbalance data
using a 5 MHz polished gold crystal in solution containing 40 mM MnSO4 at pH
of 3.8. The scan was stopped at -1.34 V vs. SCE. Scan rate: 50 mV/s. The abrupt
drop of frequency after -1.3 V during the forward scan is likely due to hydroxide
formation. Electrode Area: 1.45 cm2. (p.34)
Figure 4-4: Polarization (a) and corresponding quartz crystal microbalance result
(b) using a 5 MHz polished gold crystal in solution containing 0.3 M boric acid
with (red curve) and without 40 mM MnSO4 (black curve). Scan rate: 50 mV/s. The
data shows a monolayer of Mn can be formed on Au surface through underpotential
deposition. Electrode Area: 1.45 cm2. (p.36)
Figure 4-5: Coverage of Mn UPD on Au vs applied potential. Solution contains
40 mM MnSO4 and 0.3 M boric acid. This data shows the coverage of Mn on Au is
dependent on the applied potential, reaching full coverage at -1 V. The Mn UPD
coverage was calculated by dividing the frequency at each applied potential by the
total frequency change corresponding to one monolayer Mn. Electrode Area:
1.45 cm2. (p.37)
Figure 4-6: Time-dependent frequency shift with the following potential waveform
(-1 V for 30 sec followed by holding at open circuit potential for 30 sec) for 5
6
cycles using a 5 MHz polished gold crystal in solution containing 4 mM MnSO4
and 0.3 M boric acid. The data shows Mn underpotential deposition on Au is
reversible, i.e. full Mn monolayer forms at -1 V and was stripped off at open circuit
potential. Electrode Area: 1.45 cm2. (p.38)
Figure 4-7: Pourbaix diagram for Copper (Cu) in aqueous solution12. At pH=3.8,
Cu dissolution occurs at potentials more anodic than +0.28 V vs. SHE (i.e., +0.04
V vs. SCE, marked by the red dot). To avoid Cu dissolution, Cu electrode was held
at -0.2 V vs. SCE. (p.41)
Figure 4-8: (a) Polarization and (b) corresponding quartz crystal microbalance
result using a 5 MHz polished copper crystal in solution containing 0.3 M boric
acid with (red curve) and without 40 mM MnSO4 (black curve). Scan rate: 50 mV/s.
The data shows a monolayer of Mn can be formed on Cu surface. Electrode Area:
1.45 cm2. (p.42)
Figure 4-9: Reaction scheme for the formation of Cu layer by layer atomically on
Au using Mn UPD monolayer as a sacrificial layer. (a) Mn UPD layer forms on Au.
(b) Mn monolayer is displaced by cupric ions and a monolayer of Cu forms. (c)
Mn UPD layer forms on Cu monolayer. (d) Mn monolayer is displaced by cupric
ions and two monolayers of Cu forms. (p.45)
Figure 5-1: Three different electrochemical nucleation and growth mode of metals
on foreign substrates. (a) “Volmer-Weber” 3D island growth mode. (b) “Frank van
der Merwe” layer by layer growth mode. (c) “Stranski-Krastanov” 2D and 3D
mixed growth mode. (p.50)
Figure 5-2: Schematic showing the interaction between redox reactions in
electroless Cu deposition. An autocatalytic mechanism is proposed11. (p.52)
7
Figure 5-3: Polarization curves from a ‘glyoxylic acid only’ bath using three
different substrates: pretreated Ru (black curve) and electroless Cu films plated on
Ru for 1 s (red curve) and 120 s (blue curve). The graph shows oxidation of
glyoxylic acid (GA) (reducing agent) on electroless Cu is much faster than on Ru.
(p.53)
Figure 5-4: Schematic showing the freshly formed Cu nuclei are much more
catalytic towards GA oxidation than the substrate Ru (a). By adding additives that
suppress Cu activity, nucleation is improved on Ru (b). (p.54)
Figure 5-5: Polarization curves obtained using ‘glyoxylic acid only’ bath with
and without addition of 900 ppm of ethylenediamine on Ru surface covered with
electroless Cu. Ethylenediamine significantly inhibits oxidation of glyoxylic acid
on electroless Cu. (p.55)
Figure 5-6: AFM images of Ru plated from electroless Cu bath with and without
900 ppm ethylenediamine at different plating times. Without ethylenediamine: (a)
2 sec. (b) 5 sec. (c) 30 sec. With ethylenediamine: (d) 2 sec. (e) 5 sec. (f) 30 sec.
(p.57)
Figure 5-7: Cross-section TEM of thin electroless Cu film deposited on Ru from
a bath with ethylenediamine. Bath composition is listed in Chapter 2-4 with 900
ppm of ethylenediamine. Deposition time: 30 s. (p.58)
8
ELECTROLESS DEPOSITION OF COPPER AND COPPERMANGANESE ALLOY FOR APPLICATION IN
INTERCONNECT METALLIZATION
Abstract
by
LU YU
Copper interconnects in microchips are currently formed by Cu
electrodeposition. With the trend of future device miniaturization, the copper
interconnect dimensions will shrink, requiring a drastic change in the current state-ofthe-art interconnect metallization process. The major technology hurdles include: i)
non-uniformity of Cu seed layer deposited by physical vapor deposition (PVD) on the
barrier and ii) deteriorated electromigration resistance of Cu interconnects. A
technology under development relies on vapor deposition of a uniform ruthenium (Ru)
layer on the barrier, followed by electroless deposition of a thin and continuous Cu
seed layer on Ru.
In this thesis, we report an electroless bath that enables deposition of coppermanganese (Cu-Mn) alloy film. Incorporating Mn with Cu is attractive due to ability
of Mn to improve Cu’s electromigration resistance. Results show the Mn incorporated
is mobile, a requisite for its application in interconnect metallization. Electrochemical
and Quartz Crystal Microbalance (QCM) measurements indicate that the mechanism
of Mn incorporation during electroless Cu deposition is most likely the underpotential
deposition of Mn on Cu.
9
In addition, a novel electroless Cu bath that enables high Cu nucleation
density on the Ru surface is developed. The bath enables deposition of sub-10 nm thin
Cu film on Ru. A special additive, ethylenediamine is identified that can significantly
improve electroless Cu nucleation on Ru. The mechanism by which ethylenediamine
improves nucleation is linked to the inhibition of autocatalytic plating on already
formed nuclei, thereby providing driving force for generation of new nuclei which
yield high nucleation density thin Cu films.
10
Chapter 1: Introduction
1-1: Damascene Process for Copper Interconnect Metallization
Future device miniaturization requires a drastic change of current interconnect
metallization scheme to enable metallization of sub-20 nm features. Currently, copper
interconnect metallization is accomplished by a process described in Figure 1-1. In
this process, trenches are first etched on SiO2 or other low-k dielectric substrates.
Then, a barrier layer, typically Ta or TaN, is deposited by vapor deposition to prevent
Cu diffusion into the underlying silicon dioxide. A uniform Cu seed layer is then
deposited by physical vapor deposition on the barrier layer. This Cu seed layer
provides a surface for subsequent bottom-up fill step. Finally, the features are filled
by Cu electrodeposition using an acidic Cu bath with special additives, such as
suppressors (polyethylene glycol) and accelerators (bis(3sulphopropyl) disulphide)2.
Figure 1-2 shows a multilayer copper interconnect structure in an integrated circuit
chip fabricated using the aforementioned process3.
11
Figure 1-1: Interconnect fabrication process (trench). (a) Deposition of SiN,
SiO2 and other low-k dielectric material. (b) Trench formation by etching.
(c) Barrier layer (TaN) and Cu seed layer deposition by PVD.
(d) Bottom-up fill by Cu electrodeposition. (Adapted from D. R. Frear1).
Figure 1-2: SEM image of multilayer copper interconnects in an integrated circuit
chip3. Photograph courtesy of IBM Corporation, 1997.
12
1-2: Challenges of Future Interconnect Scheme
Interconnect scaling presents numerous challenges. One important issue is the
non-uniformity of the Cu seed layer deposited by physical vapor deposition. As the
interconnect dimension shrinks, Cu preferentially deposits at the feature rims4 (see
Figure 1-3(a)), resulting in voids and defects after electrodeposition. One approach to
address this issue is to deposit a Ru layer that acts both as a diffusion barrier layer and
a seed layer5,6 (allowing direct electrodeposition4 or electroless deposition7 onto the
Ru seed layer). However, conventional acidic Cu electrolytic plating on Ru has
relatively poor nucleation density (108 to 109 nuclei/cm2) and unacceptable
uniformity6. To deposit a 10 nm thick Cu layer, nucleation density as high as 1011 to
1012 nuclei/ cm2 is needed8. Therefore, a drastic rethinking and design of process steps
are required to enable high nucleation density on Ru and thin Cu seed layer formation.
Figure 1-3: (a) Preferential Cu seed layer growth at feature rim by physical vapor
deposition. (b) Proposed ‘uniform’ Cu seed layer formation by electroless Cu
deposition.
13
1-3: Advantages of Electroless Cu Deposition
One emerging approach to deposit thin Cu film on Ru is electroless deposition.
Unlike conventional electrodeposition, electroless deposition employs a chemical
reducing agent in solution that provides electrons for reduction of metal ions. During
the deposition, an oxidation reaction of the reducing agent and a reduction reaction of
the metal ions occur simultaneously on a catalytic surface. Because no current is
passed through the seed layer, uniform Cu layer can be deposited across the entire
wafer. Also, electroless deposition is a relatively low-cost approach as compared to
physical vapor deposition. In addition, by controlling the bath chemistry and
operating conditions, additional elements can be incorporated into the electroless Cu
deposits, providing electroless Cu films improved electromigration resistance and
other superior properties. Therefore, electroless deposition is considered a promising
technique for applications in interconnect metallization.
Electroless deposition has been used for depositing various metals (Ni, Cu, Fe,
Zn, Au, Pt, Si, Co) and alloys in different applications9. For electroless Cu deposition
on Ru, literature studies are scarce7,10. Therefore, to facilitate implementation of
electroless Cu deposition in interconnect metallization, it is imperative to understand
the mechanism of electroless Cu nucleation on Ru and develop electroless Cu bath
chemistry to deposit thin Cu seed layer. Also, extending the versatility of current
electroless Cu bath to deposit alloys (i.e. Cu-Mn) would be of interest for future
interconnects metallization.
14
In summary, the objective of this work is to (i) study Cu-Mn alloy deposition
using electroless deposition, and (ii) understand the fundamentals of electroless Cu
nucleation to enable thin Cu film formation on Ru.
The structure of the thesis is as follows:

Chapter 2: experimental methodologies used in this work are described.

Chapter 3: electroless deposition of Cu-Mn alloy is described. The
content and mobility of Mn are characterized.

Chapter 4: the underpotential deposition of Mn is identified on Au and
Cu.

Chapter 5: a novel electroless Cu bath that enables thin Cu film on Ru
is reported. The mechanism for nucleation density enhancement using
this chemistry is discussed.
15
References
1. D. R. Frear, JOM, 51, 22 (1999).
2. R. Akolkar and U. Landau, Journal of The Electrochemical Society, 151, C702
(2004).
3. http://www-03.ibm.com/ibm/history/ibm100/us/en/icons/copperchip/
4. T. P. Moffat, M. Walker, P. J. Chen, J. E. Bonevich, W. F. Egelhoff, L. Richter,
C. Witt, T Aaltonen, M. Ritala, M. Leskela and D. Josell, Journal of The
Electrochemical Society, 153, C37 (2006).
5. J. Lei, S. Rudenja, N. Magtoto and J.A. Kelber, Thin Solid Films, 497, 121
(2006).
6. O. Chyan, N. A. Tiruchirapalli and T. Ponnuswamy, Journal of The
Electrochemical Society, 150, C347 (2003).
7. Q. Y. Chen, X. Lin, C. Valvede, V. Paneccasio, R. Hurtubise, P. Ye, E. Kudrak,
and J. Abys, ECS Transactions, 8, 179 (2007).
8. L. Guo, and P.C. Searson, Langmuir, 24, 10557 (2008).
9. G. O. Mallory and J. B. Hajdu, Electroless plating: fundamentals and
applications, William Andrew, 1990.
10. F. Inoue, H. Philipsen, A. Radisic, S. Armini, Y. Civale, P. Leunissen, M.
Kondo, E. Webb and S. Shingubara, Electrochimica Acta, 100, (2013) 203.
16
Chapter 2: Experimental Methodology
2-1: General Electrochemical Setup
A standard three-electrode electrochemical setup was used for all
electrochemical tests. A saturated calomel electrode (SCE) (+0.24 V vs. SHE) was
used as a reference electrode, and a platinum wire served as the counter electrode. The
150 mL electrochemical cell (Pine Research Instrumentation) had a heating jacket
connected to an external circulating bath (Anova, Inc) to control the temperature of
the electrolyte (Figure 2-1). A VersaSTAT 4 potentiostat (Princeton Applied Research)
provided power for all experiments.
Figure 2-1: Schematic of the three-electrode electrochemical cell setup with heating
jacket for all electrochemical measurements.
17
2-2: Electroless Deposition of Cu-Mn Alloy
Electroless deposition of Cu-Mn alloy was performed on either PVD Cu or Ru
coated Si wafers. The PVD Cu was immersed in 2 M H2SO4 for 10 s to remove the
surface oxides prior to electroless deposition. The electroless Cu-Mn bath consists of
36 mM copper sulfate pentahydrate (CuSO4• 5H2O, Fisher Scientific) and 40 mM
manganese sulfate (MnSO4, Acros Organics) as the source of metal ions, 0.24 M
ethylenediaminetetraacetic acid (EDTA, Fisher Scientific) as complexant, 0.19 M
glyoxylic acid monohydrate (CHOCOOH• H2O, Acros Organics) as reducing agent,
40 ppm of 2-2’ dipyridyl (Acros Organics) and 500 ppm of polyethylene glycol M.W.
4000 (Alfa Aesar). The pH was adjusted to 11.5 using 3 M NaOH and the bath was
heated up to 60 oC.
2-3: Underpotential Deposition of Mn on Au and Cu
To study the mechanism of Mn co-deposition during electroless Cu,
electrochemical tests and quartz crystal microbalance (QCM 200, Stanford Research
System) were employed (Figure 2-2). QCM reveals mass change on the crystal
surface by measuring frequency shift of a resonant quartz crystal. Frequency data was
acquired simultaneously during the electrochemical tests. Standard 1 inch 5MHz ATcut Au and Cu coated quartz crystals (Filtech) were used as the working electrode.
The geometrical area of the Au and Cu crystal surface was 1.45 cm2. The solution for
electrochemical tests consists of 4 to 40 mM MnSO4 (Acros Organics) and 0.3 M
boric acid (H3BO3, Fisher Scientific). The pH was around 3.8 with temperature
maintained at 25 oC using water circulation.
18
Figure 2-2: Schematic of Electrochemical Quartz Crystal Microbalance (EQCM)
setup used to investigate the underpotential deposition mechanism of Mn.
2-4: Electroless Cu Nucleation on Ru
PVD Ru coated Si wafers with a nominal area of 1 cm2 was used for
electroless Cu nucleation studies. The as-received wafers were pretreated in 80 g/L
borane dimethylamine (DMAB, Acros Organics) at 60 oC for 120 s, rinsed with DI
water before being transferred to an electroless Cu solution. The composition of the
electroless Cu bath is the same as shown in chapter 2-2 except without MnSO4. In
some experiments, 900 ppm of ethylenediamine (Fisher Scientific) was added to
examine its effect on electroless Cu nucleation on Ru. The final pH of the solution
was adjusted to 12.5 using 3 M NaOH and then the solution was heated up to 60 oC.
Partial bath containing all species of a complete electroless Cu solution except copper
19
sulfate was used to study the effect of additives on the glyoxylic acid oxidation
reaction.
2-5: Material Characterization Techniques
Atomic Force Microscope (AFM)
The electroless Cu plated Ru was examined using AFM (Agilent 5500)
operated at contact mode to reveal the growth mechanism and surface roughness.
AFM uses a cantilever with a sharp tip to probe the surface. The tip is typically made
of silicon or silicon nitride that has radius of curvature on the order of ~1 nm, and
when the tip is brought close to the surface, a force between tip and sample results in
deflection of the cantilever. An attractive interaction between the surface and tip will
deflect the tip towards the specimen surface, whereas a repulsive force will deflect the
tip away from the surface.
Transmission Electron Microscope (TEM)
TEM is capable of high resolution imaging because of the small de Broglie
wavelength of electrons. During TEM imaging, electron beam is passed through the
thin specimen. As the electrons are transmitted through the specimen, an image is
formed from the interaction between the electrons and specimen. The TEM sample
was prepared using focused ion beam (FIB) and analyzed by a FEI Tecnai F30 TEM
to determine the thickness of electroless Cu on Ru.
20
Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS)
TOF SIMS was used to confirm the incorporation of Mn with electroless Cu
deposits and compare qualitatively the effect of thermal annealing on the surface Mn
content. TOF SIMS employs a pulsed primary ion beam to desorb and ionize species
from the specimen surface1. The reflected secondary ion beams are then accelerated
into and analyzed by a mass spectrometer. The mass spectrometer then determines the
chemical species by analyzing their time-of-flight from the specimen surface to the
spectrometer. A PHI TRIFT V nanoTOF SIMS was used to detect the Mn content in
the electroless Cu-Mn film before and after annealing under forming gas (5% H2+ 95%
N2) environment in a rapid thermal annealer (Ulvac-Riko Mila 5000) at 300 oC for
20 min.
Quartz Crystal Microbalance (QCM)
QCM measures mass change on a quartz crystal by detecting changes of its
resonant frequency. By addition or loss of mass on the crystal surface, the resonance
is disturbed at the surface of the acoustic resonator. Since many electrochemical
experiments involve mass change, it is desirable to monitor mass change to elucidate
the mechanism of electrochemical process. In this work, a QCM 200 (Stanford
Research System) was used in combination with electrochemical measurements to
probe the underpotential deposition mechanism of Mn on Cu and Au. Since
underpotential deposition typically forms one or two monolayers of metal on the
foreign substrate, QCM with resolution down to ng/cm2 can detect such small mass
change due to monolayer formation. The relation between mass change and frequency
shift is related by Sauerbrey equation3.
∆
2
∆
21
A is the area of the working electrode, f is the resonant frequency of the quartz crystal
and µ and
are the shear modulus and density of the crystal respectively.
22
References
1. A. Benninghoven, Angewandte Chemie International Edition in English, 33, 1023
(1994).
2. A. M. Belu, D. J. Graham and D. G. Castner, Biomaterials, 24, 3635 (2003).
3. G. Z. Sauerbrey, J. Physik, 155, 206 (1959).
23
Chapter 3: Characterization of Cu-Mn Film Deposited by Electroless Deposition
Future technology nodes require shrinkage of copper interconnects to
dimensions below 20 nm. However, the narrowing of line widths raises several issues,
such as electromigration resistance. As the interconnect dimension shrinks, the
electromigration resistance of Cu interconnects is deteriorated, shortening the device
lifetime. In order to improve Cu’s electromigration resistance, one approach is to dope
the Cu interconnects with a second element. One dopant that has received great
attention is manganese (Mn), which can diffuse through the Cu interconnects and
segregate at the Cu/dielectric layer (SiO2) interface, forming a barrier layer, adhesion
promoter and oxidation retardant1. The ability of Mn to form a barrier layer eliminates
the need for sputter-deposited nitride barrier layer and simplifies the interconnect
structure. It has been shown that incorporation of about 2 at.% Mn increases the Cu
interconnect lifetime four-fold2. In addition, Mn is capable of repairing sites where the
barrier layer (Ta/TaN) is discontinuous by forming a ‘local’ manganese silicate
diffusion barrier layer.
Currently, Cu-Mn alloy is deposited via sputter deposition with Mn content
generally exceeding few atomic percent3,4. Compared to sputter deposition,
electrochemical deposition has advantages such as simpler process steps and much
lower cost. However, depositing Cu-Mn alloy electrochemically is challenging mainly
due to the large difference of standard reduction potential between Cu (+0.1 V vs.
SCE) and Mn (-1.42 V vs. SCE). Attempts to electrodeposit Cu-Mn alloy has been
reported5,6. The plated Cu-Mn alloy is usually porous, rough and spongy due to
excessive hydrogen evolution at very cathodic potentials required for Mn
24
co-deposition. These deposits are not suitable for Cu interconnect applications.
Recently, it has been shown by A. Joi, et al.7 that smooth Cu-Mn alloy with 2 at.%
Mn content can be achieved by pulse eletrodeposition. The pulsing step allows more
Mn incorporation, and at the same time, suppresses roughness evolution compared to
DC plating. However, as reported in the paper, even films deposited via pulse plating
have considerable roughness and therefore are not completely usable in interconnect
metallization.
Compared to electrodeposition, electroless deposition of Cu seed layers are
more attractive because they do not suffer from terminal effects. Additionally,
electroless deposition is simpler than electrodeposition and is of low cost.
These advantages of electroless deposition, when combined with the ability to codeposit Cu-Mn alloy, will make the process even more attractive for interconnect
applications.
To enable electroless deposition of Cu-Mn alloy, 40 mM MnSO4 was added to
the base electroless Cu solution. The pH was adjusted to 11.5. The Mn2+ ions in
electroless Cu solution are complexed by EDTA to prevent bulk precipitation.
Figure 3-1 shows mixed potential transients during 120 sec eletroless deposition of
Cu-Mn alloy on a PVD Cu substrate. After 5 sec, the mixed potential quickly reaches
-0.72 V, very close to the mixed potential observed during electroless Cu deposition
without MnSO4. Theoretically, if the mixed potential is not cathodic enough, Mn
cannot be deposited due to its fairly negative standard reduction potential (-1.42 V vs.
SCE). Based on results from Figure 3-1, one does not expect Mn to co-deposit with
Cu during electroless plating under the present experimental conditions.
25
Figure 3-1: Mixed potential transients during electroless Cu-Mn deposition on a
PVD Cu wafer. Mixed potential during 120 s deposition is always more anodic than
Mn’s standard reduction potential (-1.42 V vs. SCE).
To determine if there exists any Mn in the electroless deposited Cu film, we
employed TOF-SIMS. Figure 3-2 below shows negative ion TOF SIMS spectra of a
Cu-Mn film deposited by 120 sec electroless deposition. A Mn peak was clearly
visible at around atomic mass of 54.94, suggesting the presence of Mn in the
electroless Cu film.
26
Mn
Figure 3-2: Negative ion TOF SIMS spectra of a Cu-Mn alloy film deposited by
electroless deposition on PVD Cu wafer for 120 s. TOF SIMS result shows there is
Mn in the electroless Cu-Mn film.
Previous work by G.S. Chen, et al.1 showed that 0.4 at.% Mn can be codeposited in electroless Cu film using formaldehyde as reducing agent. The Mn
incorporated is mobile and capable of segregating at the surface after annealing as
confirmed by synchrotron radiation. Having 0.4 at.% Mn significantly strengthens the
Cu matrix and improves its electromigration resistance. To examine whether the Mn
in our electroless Cu film is mobile, TOF SIMS was employed to qualitatively
compare the Mn content on the surface before and after thermal annealing. The film
was plated on Ru coated Si wafers for the purpose of measuring resistivity change
after annealing. As shown in Figure 3-3, after annealing, the Mn content on the
surface increases almost an order of magnitude compared to as-deposited Cu-Mn film,
suggesting that Mn in our electroless Cu film is mobile and diffuses out of Cu matrix
after annealing, an important feature for potential application in interconnect
metallization.
27
(a)
(b)
Figure 3-3: Effect of post-annealing on the Mn content on the surface of an
electroless Cu-Mn alloy film deposited on Ru coated Si wafers. (a) as-deposited
electroless Cu-Mn film. (b) annealed electroless Cu-Mn film. TOF SIMS data
indicates that Mn is mobile and segregate at the film surface.
28
References
1. G. S. Chen, S. T. Chen and Y. L. Lu, Electrochemistry Communications, 12,
1483 (2010).
2. J. P.Gambino, Physical and Failure Analysis of Integrated Circuits (IPFA),
2010 17th IEEE International Symposium, IEEE, 2010.
3. J. Koike, M. Haneda, J. Lijima, Y. Otsuka, H. Sako and K. Neishi, Journal of
applied physics, 102, 043527 (2007).
4. J. Lijima, Y. Fujii, K. Neishi and J. Koike, Journal of Vacuum Science &
Technology B, 27, 1963 (2009).
5. J. Gong and G. Zangari, Journal of the Electrochemical Society, 149, C209
(2002).
6. J. Gong and G. Zangari, Journal of The Electrochemical Society, 151, C297
(2004).
7. A. Joi, R. Akolkar and U. Landau, Journal of The Electrochemical Society,
160, D3145 (2013).
29
Chapter 4: Underpotential Deposition of Manganese (Mn)
To date, most studies have focused on Mn electrodeposition (from either
chloride or sulfate-based electrolytic solutions) at very cathodic potentials (below
-1.42 V vs. SCE, see Figure 4-1) to induce Mn+2 reduction. However, the current
efficiency of such processes is typically low and the deposits obtained are spongy due
to excessive hydrogen evolution at these potentials.
As discussed in previous chapter, we can achieve electroless deposition of CuMn with small amounts of Mn in Cu at mixed potentials (-0.72 V vs. SCE) far more
positive than the Mn standard reduction potential. A possible mechanism that could
explain this discrepancy is that Mn exhibits underpotential deposition on Cu, which as
far as we know, has not been reported in literature.
In this chapter, to investigate Mn deposition at potentials anodic to the
reduction potential of Mn+2/Mn, which is applicable to electroless deposition of CuMn alloy, a combination of electrochemical tests and quartz crystal microbalance
were used to elucidate the mechanism of Mn deposition. This chapter will focus on
experimental confirmation of Mn underpotential deposition on Au and Cu substrates.
30
Figure 4-1: Pourbaix diagram for Manganese (Mn) in aqueous solution1. In acidic
solution, Mn electrodeposition occurs at potentials more cathodic than -1.18 V vs.
SHE (i.e., -1.42 V vs. SCE).
31
Chapter 4-1: Underpotential Deposition of Manganese (Mn) on Gold (Au)
Underpotential deposition (UPD) refers to the electrodeposition of metal
monolayers on a foreign substrate at potentials more positive than the metal’s
standard reduction potential2. Such phenomenon is due to the strong interaction
between the metal monolayer and the foreign substrate, and has been studied for
various metal/substrate systems in aqueous solutions, such as Cu on Au3 and Pt4, Pb
on Ag5 and Ag on Pt6, and less frequently in non-aqueous solution7. The metal
adatoms at sub-monolayer coverage are believed to exhibit significant different
electronic properties than bulk metal. In addition, UPD allows precise control of
surface coverage of metal adlayers on the substrate and therefore is a topic of
significant interest.
To determine whether Mn can be underpotentially deposited, polarization and
frequency shift data (collected using a quartz crystal microbalance) were obtained
using a gold crystal (area: 1.45 cm2) in a solution containing 40 mM MnSO4 (pH=3.8)
at room temperature (Figure 4-2). The choice of MnSO4 concentration was based on
its concentration in a typical electroless Cu solution. The pH was kept at 3.8 instead
of 11.5 as in electroless Cu solution because the addition of complexing agent to
avoid Mn2+ precipitation obscures the frequency shift due to Mn monolayer. As
shown in Figure 4-2, two characteristic peaks, A1 and A2, appeared around -0.3 V
and -0.8 V (vs. SCE). After the potential scan to -1.1 V, there was a frequency shift
of about 8 Hz. The sensitivity factor for a 5 MHz AT-cut crystal at room temperature
is 17.67 ng•Hz-1•cm-2. According to the Sauerbrey equation8, 8 Hz of frequency shift
corresponds to a mass change of 140 ng•cm-2, approximately the weight of a Mn
monolayer on Au.
32
A1
A2
Figure 4-2: Polarization and corresponding quartz crystal microbalance result using a
5MHz polished gold crystal in solution containing 40 mM MnSO4 at pH of 3.8. The
scan was stopped at -1.1 V vs. SCE. Scan rate: 50 mV/s. Electrode Area: 1.45 cm2.
In a separate experiment, we attempted to scan the potential to more negative
values to examine Mn deposition using solution containing 40 mM MnSO4 at pH of
3.8. The negative potential limit was set to be -1.34 V vs. SCE, still more positive
than Mn reduction potential (-1.42 V vs. SCE), and therefore no Mn should be
deposited. Similar to results shown in Figure 4-2, we observed a frequency shift of
about 7 Hz up to -1.1 V (Figure 4-3). However, when the potential reaches around 1.3 V, there is an abrupt drop of frequency of about 70 Hz, indicating significant
increase of mass on the electrode surface. The shift of 70 Hz frequency corresponds to
a mass increase of ~1.23 µg•cm-2, which cannot be attributed to Mn monolayer
formation.
33
One possible hypothesis for this mass increase is the formation of Mn
hydroxides at such cathodic potentials. According to the Pourbaix diagram of Mn
(Figure 4-1), Mn2+ is susceptible to forming manganese hydroxides if the pH
increases above 7. Therefore, at potentials as cathodic as -1.3 V, a local increase of
pH due to excessive hydrogen evolution could induce Mn hydroxide precipitation at
the surface. The formation of manganese hydroxides has also been found in similar
potential ranges by other investigators9.
Figure 4-3: Polarization and corresponding quartz crystal microbalance data
using a 5 MHz polished gold crystal in solution containing 40 mM MnSO4 at pH
of 3.8. The scan was stopped at -1.34 V vs. SCE. Scan rate: 50 mV/s. The abrupt
drop of frequency after -1.3 V during the forward scan is likely due to hydroxide
formation. Electrode Area: 1.45 cm2.
34
Since hydroxides formation obscures the QCM mass change due to Mn
monolayer formation, it is necessary to inhibit hydroxide formation in the potential
range of interest. Boric acid is known to be a pH buffer that can eliminate the local pH
increase at the electrode surface, thereby suppressing hydroxide formation10.
Figure 4-4 shows polarization and QCM results obtained on a gold crystal surface in a
solution containing 40 mM MnSO4 and 0.3 M boric acid at pH of 3.8. During the
forward scan, two peaks appeared, similar to what has been observed without boric
acid in solution (Figure 4-2). A frequency shift of 8 Hz during the forward scan was
observed, corresponding to a mass of 140 ng/cm2, which is approximately the weight
of Mn monolayer on Au. After -0.9 V, the frequency reached a plateau, indicating the
mass increase stopped and the surface was completely covered by a Mn monolayer.
The fact that the frequency plateaued out also indicates the hydroxides formation is
completely inhibited by the boric acid in solution.
35
(a)
(b)
Figure 4-4: Polarization (a) and corresponding quartz crystal microbalance result (b)
using a 5 MHz polished gold crystal in solution containing 0.3 M boric acid with (red
curve) and without 40 mM MnSO4 (black curve). Scan rate: 50 mV/s. The data shows
a monolayer of Mn can be formed on Au surface through underpotential
deposition. Electrode Area: 1.45 cm2.
36
Figure 4-5 shows coverage of Mn UPD layer on Au at different applied
potentials obtained using results from QCM measurements in solution containing
40 mM MnSO4 and 0.3 M boric acid. The coverage of Mn UPD on Au increased with
more negative potential applied, eventually reaching full coverage at -1 V. Correlation
between surface coverage of UPD monolayer
and UPD shift ∆
is given by the
following equation11:
∆
where A and B are parameters related to the enthalpy change of substrate-metal and
metal-metal bond formation, Gibbs free energy change involved in the substrate-water
and metal-water bond formation and lattice coordination number and work function of
metal and substrate.
Figure 4-5: Coverage of Mn UPD on Au vs applied potential. Solution composition:
40 mM MnSO4 and 0.3 M boric acid. This data shows the coverage of Mn on Au is
dependent on the applied potential, reaching full coverage at -1 V. The Mn UPD
coverage was calculated by dividing the frequency at each applied potential by the
total frequency change corresponding to one monolayer Mn. Electrode Area: 1.45 cm2.
37
According to results from Figure 4-4 and Figure 4-5, formation of a full Mn
monolayer requires potentials to be at least -1 V vs. SCE. To determine if Mn
underpotential deposition on Au is reversible, five cycles of the following potential
waveform (-1 V for 30 sec followed by open circuit potential for 30 sec) was applied
to a 5 MHz polished gold crystal in solution containing 4 mM MnSO4 and 0.3 M
boric acid. As shown in Figure 4-6, the frequency quickly (< 5 sec) dropped to and
stabilized at -7 Hz as the potential was switched to -1 V, indicating a Mn monolayer
formed and deposition stopped. At the open circuit potential, the frequency slowly
shifted back, suggesting Mn monolayer was being dissolved. After 30 sec at open
circuit potential, the next cycle yielded frequency almost exactly the same as the first
one. This shows that the underpotentially deposited Mn can be stripped from the
electrode surface and that the deposition-stripping process is very reproducible.
Figure 4-6: Time-dependent frequency shift with the following potential waveform
(-1 V for 30 sec followed by holding at open circuit potential for 30 sec) for 5 cycles
using a 5 MHz polished gold crystal in solution containing 4 mM MnSO4 and 0.3 M
38
boric acid. The data shows Mn underpotential deposition on Au is reversible, i.e. full
Mn monolayer forms at -1 V and was stripped off at open circuit potential. Electrode
Area: 1.45 cm2.
39
Chapter 4-2: Underpotential Deposition of Manganese (Mn) on Copper (Cu)
As shown in the previous chapter, a Mn monolayer can be formed on Au
surface by underpotential deposition as confirmed by voltammetry and quartz crystal
microbalance measurements. However, this does not fully account for the fact that Mn
is incorporated during electroless deposition at potentials much more positive than its
standard reduction potential. During electroless Cu-Mn deposition, Mn deposits on a
Cu-rich substrate. Therefore, this chapter will focus on studying Mn deposition
mechanism on Cu substrate using electrochemical tests and quartz crystal
microbalance.
5 MHz AT-cut Cu crystals (Filtech) were used as substrates for
electrochemical tests and QCM measurements. The Cu crystals were cleaned in dilute
sulfuric acid prior to tests. According to the Pourbaix diagram of Cu (Figure 4-7),
before introducing the Cu crystal into the electrolyte containing 40 mM MnSO4 and
0.3 M boric acid at pH 3.8, the Cu crystal were held at -0.2 V vs. SCE to avoid Cu
dissolution into the electrolyte.
40
Figure 4-7: Pourbaix diagram for Copper (Cu) in aqueous solution12. At pH=3.8, Cu
dissolution occurs at potentials more anodic than +0.28 V vs. SHE (i.e., +0.04 V vs.
SCE, marked by the red dot). To avoid Cu dissolution, Cu electrode was held at
-0.2 V vs. SCE.
Figure 4-8 shows polarization and corresponding quartz crystal microbalance
measurements conducted using a Cu crystal in solution containing 0.3 M boric acid
and 40 mM MnSO4. The potential scan was started from -0.2 V to avoid Cu
dissolution into the electrolyte. Compared to the polarization curve collected without
MnSO4 in the electrolyte, the polarization curve with MnSO4 shows a broad peak
from -0.3 V to -0.9 V. After the potential scan till -1.2 V, the electrolyte containing
only boric acid showed no frequency drop, while the electrolyte containing MnSO4
showed a frequency shift of 7 Hz, close to the weight gain due to a monolayer of Mn.
The frequency also reached a plateau at -1 V, indicating there is no mass change once
a Mn monolayer covered the Cu crystal surface. The slight frequency difference
between Mn underpotential deposition on Au and Cu might be attributed to the
41
different crystal structure, surface roughness of Cu and Au or the resolution limit of
the QCM used.
(a)
(b)
Figure 4-8: (a) Polarization and (b) corresponding quartz crystal microbalance result
using a 5 MHz polished copper crystal in solution containing 0.3 M boric acid with
(red curve) and without 40 mM MnSO4 (black curve). Scan rate: 50 mV/s. The data
shows a monolayer of Mn can be formed on Cu surface. Electrode Area: 1.45 cm2.
42
Chapter 4-3: Conclusions
Experiments described in Chapter 4-1 and 4-2 confirmed Mn underpotential
deposition on Au and Cu surfaces. This confirmation lends support to our conclusion
that Mn incorporation during electroless Cu-Mn deposition proceeds through UPD of
Mn on Cu. The fact that Mn only forms a monolayer on Cu (through UPD) explains
why the electroless Cu film has very low Mn content. Furthermore, in an alkaline
electroless Cu bath with EDTA as complexant, most of the Mn2+ is complexed by
EDTA, which may retard the kinetics of Mn UPD. Future studies are planned to gain
a more fundamental insight into the UPD kinetics, and develop strategies to control
the Mn content of electroless deposited Cu-Mn films.
43
Chapter 4-4: Future Work
Chapter 3 and Chapter 4 established the feasibility of using electroless
deposition to co-deposit Cu-Mn alloy. However, more work needs to be done in order
to increase the Mn content in the electroless Cu-Mn alloy film. We believe the low
content of Mn in the current electroless Cu-Mn film is mainly due to the strong
chelation effect between Mn2+ and EDTA in the bath that slows down the kinetics of
Mn underpotential deposition on Cu. To increase the Mn content, it is desirable to
replace EDTA with a weaker complexing agent (i.e., tartrate), as long as the Mn2+
does not precipitate in the bath. In addition, understanding how temperature and pH
affects Mn underpotential deposition kinetics would allow better design of the
electroless Cu-Mn bath to achieve higher Mn content.
Another potential application of Mn underpotential deposition is in
surface limited redox replacement (SLRR). SLRR involves first the formation of a
UPD metal monolayer (M1) on a foreign substrate, and then the monolayer is
displaced by a different metal ion in solution, which forms one monolayer of a more
noble metal (M2). The displacement reaction is self-limiting. However, by repeating
the step of forming M1 UPD monolayer on already formed M2 and followed by
displacement with M2 ions in solution, M2 can be grown one atomic layer at a time.
SLRR has been used to grow various metals taking advantage of the UPD process, i.e.
Pt growth by Pb13 and Cu UPD14, Ru growth by Pb UPD15, Cu growth by Pb UPD16.
However, most process uses Pb, a metal that is highly toxic. Mn, on the other hand, is
more environmentally benign, and can potentially replace Pb for the above process.
Figure 4-9 shows a proposed reaction scheme to use Mn monolayer as sacrificial layer
for Cu growth on Au substrate.
44
(a)
(b)
(c)
(d)
Figure 4-9: Reaction scheme for the formation of Cu layer by layer atomically on Au
using Mn UPD monolayer as a sacrificial layer. (a) Mn UPD layer forms on Au. (b)
Mn monolayer is displaced by cupric ions and a monolayer of Cu forms. (c) Mn UPD
layer forms on Cu monolayer. (d) Mn monolayer is displaced by cupric ions and two
monolayers of Cu forms.
45
References
1. http://commons.wikimedia.org/wiki/File:Pourbaix_diagram for_Manganese.svg
2. E. Herrero, L. J. Buller and H. D. Abruna, Chemical Reviews, 101, 1897 (2001).
3. F. A. Möller, O. M. Magnussen and R. J. Behm, Physical Review B, 51, 2484
(1995).
4. N. Markovic and P. N. Ross, Langmuir, 9, 580 (1993).
5. M. F. Toney, J. G. Gordon, M. G. Samant, G. L. Borges, O. R, Melroy, D. Yee
and L. B. Sorensen, The Journal of Physical Chemistry, 99, 4733 (1995).
6. F. EI Omar, R. Durand and R. Faure, Journal of electroanalytical chemistry
and interfacial electrochemistry, 160, 385 (1984).
7. L. F. Li, Y. Luo, G. G. Totir, D. A. Totir, G. S. Chottiner and D. A. Scherson,
The Journal of Physical Chemistry B, 103, 164 (1999).
8. G. Z. Sauerbrey, J. Physik, 155, 206 (1959).
9. P. Díaz-Arista, R. Antaño-López, Y. Meas, R. Ortega, E. Chaînet, P. Ozil and G.
Trejo, Electrochimica acta, 51, 4393 (2006).
10. S. L Wang, Surface and Coatings Technology, 186, 372 (2004).
11. V. Sudha and M.V.Sangaranarayanan, J. Chem. Sci., 117, 207 (2005).
12. http://commons.wikimedia.org/wiki/File:Cu-pourbaix-diagram.svg.
13. M.Fayette, Y. Liu, D. Bertrand, J. Nutariya, N. Vasiljevic, and N. Dimitrov,
Langmuir, 27, 5650 (2011)
14. S. R.Brankovic, J. X. Wang, and R. R. Adžić, Surface Science, 474, L173 (2001).
15. C. Thambidurai, Y. G. Kim and J. L. Stickney, Electrochimica Acta, 53, 6157
(2008).
16. L. T.Viyannalage, R. Vasilic and N. Dimitrov, The Journal of Physical Chemistry
C, 111, 4036 (2007).
46
Chapter 5: Electroless Cu Nucleation on Ru
Electroless Cu nucleation has been studied on Pd activated substrate1,2 and
ruthenium3,4. It was observed that Cu nucleation proceeds via Volmer-Weber island
growth mode5. For island growth, nucleation density (N) plays an important role and
the minimum thickness
at which film coalesces can be predicted by the
following equation6:
1
2√
Therefore, for a 10 nm continuous film, i.e., dcrit=10 nm, the nucleation
density should be on the order of 1012/cm2. However, none of the studies on
electroless Cu nucleation on Ru3,4 show successful formation of less than 10 nm thick
electroless Cu seed layer on Ru, a prerequisite for metallization of sub-20 nm
interconnects. In addition, there is a lack of fundamental understanding of the
electroless Cu nucleation mechanism on the Ru surface. Inoue, et al.3 studied filling
high aspect ratio TSV using an electroless Cu seed on ALD Ru. In a similar work, H.
K. Lee, et al.7 studied electroless Cu deposition on Ru for filling patterned wafers. Q.
Y. Chen, et al.4 examined electroless Cu deposition on PVD, ALD and CVD Ru, and
showed the potential of using electroless Cu as seed followed by electrodeposition to
bottom-up fill vias with critical dimension of 60 nm.
In the following section, we demonstrate enhanced electroless Cu nucleation
on Ru using additives. A mechanism is proposed that explains nucleation
enhancement.
47
Chapter 5-1: Nucleation Mechanism onto a Foreign Substrate
Electrochemical nucleation of a metal on a foreign substrate takes place at the
substrate/conducting electrolyte interface8. For 3D metal phase formation, the overall
reaction can be expressed as:
↔
(1)
The actual direction of the above reaction is dependent on the electrode potential,
which can be calculated by Nernst equation:
where
/
/
is
/
/
(2)
the reduction potential of Me2+/Me couple at electrode surface.
is the standard reduction potential of Me2+/Me couple.
of the metal ions in solution and
is the activity
is the activity of the metal, which is taken as 1.
For 3D metal phase formation, if the applied potential
>
/
,
metal
deposition proceeds due to an overpotential as driving force, and this is commonly
referred to as overpotential deposition (OPD). If the applied potential
<
/
,
dissolution of the metal phase happens. However, in some cases, metal ad-atoms can
form a 2D phase at
>
/
due
to a strong interaction between metal and
substrate, and this process is referred to as underpotential deposition (UPD). Overall,
the initial nucleation and subsequent growth of the metal are strongly dependent on
two factors, the binding energy of metal ad-atoms on foreign substrate S, ψMe-S and
the binding energy of metal ad-atoms on native substrates Me, ψMe-Me8.
48
1) If ψMe-S << ψMe-Me, metal ad-atoms will preferentially form on the native
substrates and the deposition proceeds via “Volmer-Weber” 3D island
growth mode (Figure 5-1(a)).
2) If ψMe-S >>ψMe-Me, metal ad-atoms will preferentially form on the foreign
substrates. Therefore, 2D nucleation of metal phase on the substrate will
take place in the UPD potential range, resulting in formation of a
monolayer or few monolayers of metal layer. After formation of a 2D
metal phase, if Me-S crystallographic misfit is negligible, the metal growth
will follow the “Frank van der Merwe” or layer by layer growth mode
(Figure 5-1(b)).
3) If ψMe-S >>ψMe-Me and there is significant Me-S crystallographic misfit, a
mixed 2D and 3D growth mode-“Stranski-Krastanov” (Figure 5-1(c)) will
happen. Because the 2D metal phase has a different structure compared to
3D metal phase and contains internal strain, 3D growth of metal is
energetically favored on 2D metal phase.
49
(a)
Metal
Substrate
(b)
(c)
Figure 5-1: Three different electrochemical nucleation and growth mode of metals
on foreign substrates. (a) “Volmer-Weber” 3D island growth mode. (b) “Frank van
der Merwe” layer by layer growth mode. (c) “Stranski-Krastanov” 2D and 3D
mixed growth mode.
50
Chapter 5-2 Electroless Cu Deposition on Ru
To improve electroless Cu nucleation, it is necessary to have a fundamental
understanding of its deposition mechanism. Electroless Cu deposition generally
involves a reducing agent oxidation reaction and a metal ion reduction reaction
happening simultaneously on a catalytic surface. The oxidation reactions using
formaldehyde or glyoxylic acid as reducing agent are given as:
2
2
2
2
→2
(3)
2
2
→2
(4)
While the metal ion (Cu2+) reduction reaction is given as:
2
→
(5)
Since electroless Cu deposition process is electrochemical in nature that involves
electron transfer, electroneutrality must be maintained during the deposition process.
Previous authors9 suggested that reaction (3) and (5) are independent of each other,
and therefore by obtaining the polarization curves that contain the kinetic information
of individual partial reactions, the plating rate and the surface potential in the actual
deposition process can be predicted. This theory is commonly referred to as the mixed
potential theory10.
However, our recent finding11, along with other reports12,13, indicate that
reaction (3) and (4) are not independent of reaction (5). Specifically, the freshly
formed Cu nuclei are catalytic towards the oxidation of reducing agent, and the
reducing agent produces catalytic intermediates that accelerate the reduction of cupric
ions. This mechanism is summarized in Figure 5-2. The interaction between two half
reactions leads to much faster plating rate than that predicted by the mixed potential
theory.
51
Figure 5-2: Schematic showing the interaction between redox reactions in electroless
Cu deposition. An autocatalytic mechanism is proposed11.
Similar catalytic mechanism has also been found in electroless Ni deposition
using hypophosphite as reducing agent14. Matsubara, et al. 15 studied the relation
between nucleation density (minimum transition thickness) and autocatalytic activity
of Ni-P substrates and found thinner nickel films could be fabricated at lower
substrate activity, which affects the autocatalytic mechanism. This suggests
electroless Cu nucleation on Ru could possibly be improved by suppressing the
mechanism shown in Figure 5-2.
Since substrates play an important role in the above mechanism, to determine
glyoxylic acid oxidation rates on Ru and electroless Cu surface, polarization curves
were collected from a ‘glyoxylic acid only’ bath. As shown in Figure 5-3, in the
potential range from -0.5 V to -1 V (range within which mixed potential of electroless
Cu lies), glyoxylic acid oxidation on blank Ru was much slower. The maximum
glyoxylic acid oxidation current at ~ -0.5 V is only 0.25 mA/cm2 on Ru substrate,
much slower than the oxidation current density collected on Ru surface plated with
52
electroless Cu for 1 s (0.7 mA/cm2) and 120 s (1.2 mA/cm2), respectively. Thus,
Figure 5-3 establishes the catalytic activity of Cu in enhancing glyoxylic acid
oxidation. Higher the Cu coverage on Ru, higher is the glyoxylic acid oxidation rate.
Figure 5-3: Polarization curves from a ‘glyoxylic acid only’ bath using three different
substrates: pretreated Ru (black curve) and electroless Cu films plated on Ru for 1 s
(red curve) and 120 s (blue curve). The graph shows oxidation of glyoxylic acid (GA)
(reducing agent) on electroless Cu is much faster than on Ru.
The fact that glyoxylic acid oxidizes substantially slowly on Ru explains the
poor nucleation density of electroless Cu on Ru. This is shown schematically in
Figure 5-4 (a). Suppressed redox chemistry on the Ru surface means that sustained
electroless plating occurs only on already nucleated Cu – a condition that promotes
heterogeneous nucleation and a low nucleation density number. To improve
electroless Cu nucleation on Ru, it is desirable to inhibit the Cu catalytic activity. One
53
approach is to use additives that adsorb on the Cu surface and inhibit its surface
activity. The additives’ adsorption will stifle sustained electroless plating on nucleated
Cu sites, and promote generation of fresh nucleation sites on Ru, as shown
schematically in Figure 5-4(b).
X
Figure 5-4: Schematic showing: (a) freshly formed Cu nuclei are much more catalytic
towards glyoxylic acid oxidation than the substrate Ru; (b) By adding suppressing
additives that selectively adsorb on Cu, nucleation density during electroless plating
can be improved.
It was reported that ethylenediamine acts as a strong suppressor for
formaldehyde oxidation on Cu16. Since formaldehyde and glyoxylic acid have similar
molecular structure, we choose ethylenediamine to study its suppression effect on the
glyoxylic acid oxidation on Cu. Polarization curves were collected on an electroless
Cu surface immersed in an ‘glyoxylic acid only’ solution with and without 900 ppm
of ethylenediamine. In the potential range from -1 V to -0.5 V, the oxidation current
of glyoxylic acid without ethylenediamine in solution is higher that with
54
ethylenediamine (Figure 5-5), suggesting ethylenediamine suppresses glyoxylic acid
oxidation on Cu. This effect can be explained by the adsorption of ethylenediamine
on the Cu surface16, which reduces the number of active Cu sites available for
glyoxylic acid oxidation.
Figure 5-5: Polarization curves obtained on an electrode made of electroless Cu
plated on Ru for 120 s plating time (thickness: ~80 nm). Polarization was performed
in a ‘glyoxylic acid only’ bath with and without addition of 900 ppm of
Ethylenediamine (‘En’). Ethylenediamine significantly inhibits oxidation of glyoxylic
acid on electroless Cu.
In order to understand how ethylenediamine affects electroless Cu nucleation
on Ru during the actual electroless Cu plating, we analyzed (by AFM) the samples
plated with and without ethylenediamine for different electroless Cu deposition times.
Figure 5-6 shows 2µm X 2µm AFM images for blank Ru surface and Ru deposited
55
with electroless Cu at different deposition times with and without ethylenediamine in
the bath. The blank Ru surface is smooth with average RMS roughness value of
0.15 nm. For samples plated without ethylenediamine, short time plating (2 sec)
yields Cu islands with non-uniform diameter ranging from 0.5 µm to 1 µm with
roughness increased to 1.92 nm (Figure 5-6 (a)). This suggests the Cu growth follows
3D “Volmer-Weber” island growth mode. Further nucleation and film growth (30 sec)
result in rough surface (RMS roughness 6.68 nm). However, addition of
ethylenediamine significantly changed the surface morphology and growth mode of
electroless Cu on Ru. The nuclei size is much smaller and more uniform. The RMS
roughness increases slightly to 0.47 nm after 2 sec plating, and reaches 3.5 nm after
30 sec plating, almost half of the sample plated without ethylenediamine, suggesting
the ethylenediamine improves electroless Cu nucleation on Ru.
56
No Ethylenediamine
With Ethylenediamine
(a)
(d)
(b)
(e)
(c)
(f)
Figure 5-6: AFM images of Ru plated from electroless Cu bath with and without 900
ppm ethylenediamine at different plating time. Without ethylenediamine: (a) 2 sec. (b)
5 sec. (c) 30 sec. With ethylenediamine: (d) 2 sec. (e) 5 sec. (f) 30 sec.
To test if ethylenediamine can truly promote formation of thin electroless Cu
film on Ru, cross-section TEM was taken to examine the thickness of electroless Cu
plated on Ru for 30 sec with ethylenediamine. As shown in Figure 5-7, after 30 s
57
deposition electroless Cu deposition, a uniform and thin (~10 nm) electroless Cu layer
is formed on Ru. This corresponds to a plating rate of about 20 nm/min, about half of
the plating rate without ethylenediamine in the solution (40 nm/min).
Electroless Cu
Ru
Ta
20 nm
Figure 5-7: Cross-section TEM of thin electroless Cu film deposited on Ru from a
bath with ethylenediamine. Bath composition is listed in Chapter 2-4 with 900 ppm
of ethylenediamine. Deposition time: 30 s.
The AFM and TEM images above confirm that ethylenediamine is a
promising additive for enhancing electroless Cu nucleation on Ru surface. It promotes
nucleation most likely through its adsorption on fresh electroless Cu nuclei, thereby
suppressing glyoxylic acid oxidation on Cu and promoting increased nucleation on
the substrate Ru.
58
Chapter 5-3: Conclusions
A novel electroless Cu bath that incorporates ethylenediamine as a nucleation
promoter for electroless Cu plating on Ru was developed. The enhancement in
nucleation density was explained by the effect of ethylenediamine on suppressing the
oxidation reaction of glyoxylic acid on Cu. This suppression lowered the overall
plating rate, but distributed the plating reaction uniformly over the substrate, thereby
enabling smooth and uniform electroless films of sub-10 nm thickness.
59
References
1. S. Richard, Journal of The Electrochemical Society, 117, 864 (1970).
2. J. Horkans, J. Kim, C. McGrath and L. T. Romankiw, Journal of The
Electrochemical Society, 134, 300 (1987).
3. F. Inoue, H. Philipsen, A. Radisic, S. Armini, Y. Civale, P. Leunissen, M.
Kondo, E. Webb and S. Shingubara, Electrochimica Acta, 100, 203 (2013).
4. Q. Y. Chen, X. Lin, C. Valvede, V. Paneccasio, R. Hurtubise, P. Ye, E. Kudrak
and J. Abys, ECS Transactions, 8, 179 (2007).
5. V. Sabayev, N. Croitoru, A. Inberg and Y. Shacham-Diamand, Materials
Chemistry and Physics, 127, 214 (2011).
6. L. Guo and P. C. Searson, Langmuir, 24, 10557 (2008).
7. H. K. Lee and J. Y. Hur, Metals and Materials International, 19, 821 (2013).
8. E. Budevski, G. Staikov, W. J. Lorenz and K. E. Keusler, Angewandte
Chemie-German Edition, 109, 1418 (1997).
9. M. Paunovic, Plating, 55, 1161 (1968).
10. C. Wagner and W. Traud, Corrosion, 62, 843 (2006).
11. L. Yu, L. Guo, R. Preisser and R. Akolkar, Journal of The Electrochemical
Society 160, D3004 (2013).
12. I. Ohno, Materials Science and Engineering: A, 146, 33 (1991).
13. I. Ohno and S. Haruyama, Surface Technology, 13, 1 (1981).
14. L. M. Abrantes and J. P. Correia, Journal of The Electrochemical Society, 141,
2356 (1994).
15. H. Matsubara, T. Yonekawa, Y. Ishino, N. Saito, H. Nishiyama and Y. Inoue,
Electrochimica Acta, 52, 402 (2006).
60
16. Y. M. Lin and S. C. Yen, Applied surface science, 178, 116 (2001).
61