Analysis And Control Of Copper Plating Bath Additives
And By-Products
Beverly Newton, Edward Kaiser
Dionex Corporation
1228 Titan Way
Sunnyvale, CA 94086
Abstract. New copper plating bath chemisties are being developed to meet the emerging need of plating copper into submicron features on semiconductor wafers. These chemistries are designed to provide a fast, efficient, fill for even the most
challenging wafer terrain. It has been found that maintaining the concentration of the additives in these plating baths at
certain levels is critical to the performance of the bath. Plating technology for semiconductor applications requires rigid bath
control and disciplined methodology.
Establishing correlations between what is found in the plated film and bath chemistry control parameters is fundamental in
producing interconnects that are consistent and reliable. To establish these correlations, it is important to have a clear
understanding of the chemical composition of the bath. It is theorized that the "suppressor" bath components help moderate
the deposition rate of the copper fill and the "leveler" additives improve the topology of the copper overfill. Too much or
too little of these components in the bath can be detrimental to the quality of the copper deposition and may result in "fill
failure" leading to a higher than necessary scrap rate for the wafers. Indirect bath measurements, such as Cyclic
Voltammetric Stripping (CVS), tell an incomplete story as these techniques only measures the combined effect of the
additives and by-products on the plating quality. High Performance Liquid (HPLC) and Ion Chromatography are analytical
techniques which provide important information on the concentration, chemical balance and trend measurement of major
constituents such as additives, brighteners, boosters, stabilizers, carriers, levelers, inhibitors, accelerators, transition metals,
metal complexes and contaminants in the plating bath. This information provides for improved device quality, reduced
scrap rate and reduced costs of bath maintenance.
This, however, is not the end of the story. In addition to additives, copper plating baths also contain process byproducts.
This paper will cover the development of analytical methods using metal free liquid chromatography to quantify the
components and any related by-products found in copper plating baths used for small-featured semiconductors..
electrodeposition by-products to help optimize the
process and improve yields [2].
New copper plating bath chemistries have been
optimized to meet the emerging needs of this
application. The new products are designed to
provide a fast, efficient fill for even the most
challenging wafer terrain. An example of these new
products is the Enthone-OMI Cubath SC plating
chemistry. New analytical methodologies have also
been developed using HPLC and 1C for evaluating
these new electroplating products. Some typical
applications for the analysis of the copper
metallization plating bath are the determination of
SPS additive concentrations and the analysis of SPS
additive electrolysis by-products.
INTRODUCTION
The use of copper metallization for small-featured
semiconductor devices has set the stage for increased
requirements for the purity, plating effectiveness and
plating speed of electroplating bath chemicals [1].
Increased requirements have also been set for
monitoring and optimizing copper electroplating
baths used in the manufacture of these devices.
Although electrochemical deposition (BCD) is not a
new technology, the requirements for precision and
quality of the deposition in the semiconductor
industry are, by far, more critical than those currently
practiced in other industries. HPLC and Ion
Chromatography (1C) provide the analytical
capabilities to monitor plating bath additives and
CP683, Characterization and Metrology for ULSI Technology: 2003 International Conference,
edited by D. G. Seiler, A. C. Diebold, T. J. Shaffner, R. McDonald, S. Zollner, R. P. Khosla, and E. M. Secula
© 2003 American Institute of Physics 0-7354-0152-7/03/$20.00
514
macroporous copolymer with a very high
hydrophobic surface area. A mobile phase composed
of acetonitrile and deionized water was used to elute
the analytes of interest.
For the determination of chloride, lonPac AS159\im analytical column (250 x 4 mm) was used. The
packing material of the lonPac AS 15 is composed of
a highly crosslinked macroporous core and an anionexchange layer grafted to the surface. An anion selfregenerating ultra suppressor (ASRS®) from Dionex
Corporation operated in the recycle mode was used
to reduce the conductivity of the eluent. An EG40
eluent generator equipped with an EluGen® EGCKOH cartridge was used to produce potassium
hydroxide eluents. The EG40 eluent generator
system electrolytically produces high-purity
potassium hydroxide (KOH) eluents using deionized
water as the carrier stream.
For the determination of copper, an lonPac CS12A
analytical column (250 x 2 mm) was used. The
packing material of the lonPac CS12A is composed
of a highly crosslinked macroporous core and a
cation-exchange layer grafted to the surface.
PLATING BATH CHEMISTRY
Copper plating on semiconductor wafers requires
filling of sub-micron features. This is achieved by
optimizing the choice and concentration of a variety
of additives called accelerators, levelers and
suppressors [3]. See Table 1. The "accelerator"
additives are aptly named due to their ability to
influence the rate of deposition of copper into the
small features. A controlled increase in accelerator
concentration in a plating solution causes an increase
in the rate of copper deposition at constant potential.
SPS, bis-(sodium sulfopropyl)-disulfide, is a typical
component of accelerators used in many commercial
and proprietary plating formulations.
As the bath becomes aged, the bath additives are
consumed over time by the process and also broken
down into electrolysis by-products. This requires
that the additives be replenished at frequent intervals
to ensure that the bath is operating in the optimal
range of concentration of these additives. In order to
control the concentration of the additives in the
plating bath, some plating tools use a "bleed and
feed" format where a certain amount of bath is
removed or "bled off' and then fresh bath and
additives are added to replenish the active chemicals.
Other tools simply require that the entire bath be
replaced at a certain time interval. In either case, it is
critical to track the concentration of these additives
and their by-products in the bath. If by-products are
allowed to build up uncontrolled, their concentration
in the bath will eventually reach a level where
plating defects termed "fill failures" will occur [4].
Chemicals, Solutions and Samples
Samples of the acid copper plating bath and
additives were from Enthone-OMI, (New Haven,
CT, USA). The acid copper-plating bath consisted of
copper sulfate, sulfuric acid, and hydrochloric acid.
The proprietary additives were used at Enthone's
recommended operating concentrations.
EXPERIMENTAL
RESULTS AND DISCUSSION
Chromatographic System
Ion chromatography was applied to the
determination of chloride in the acid copper-plating
bath. The excess of sulfate from the copper sulfate
hampers determination of chloride. The lonPac AS 15
column anion exchange column was selected for this
separation because it had sufficient capacity for the
high-sulfate matrix and strong retention for sulfate.
Figure 1 shows a separation of chloride for a 25-jiiL
sample of the bath diluted 1:100. Good separation is
achieved for trace chloride in the presence of an
excess of sulfate, Figure 1. This method has been
demonstrated to give reliable quantification of
chloride for over 500 injections, and was applied to
the analysis of chloride from a plating mini
marathon.
It was of interest to develop a reliable means for
quantifying the amount of copper in the plating bath.
The determination of this analyte is simplified
All chromatography was performed on a Dionex
(Sunnyvale, CA, USA) DX-500 ion chromatograph.
The system consists of a gradient pump (GP50), and
an autosampler (AS50) with chromatography
compartment. Detection was by means of a fixed
wavelength
absorbance
detector
(AD25),
conductivity detector (CD20) or evaporative light
scattering detector (PL-ELS 1000) Polymer Labs Inc.
(Amherst, MA, USA). The instrumentation was
controlled and the data were acquired with the
Dionex PeakNet® chromatography software using a
personal computer.
All columns used in this study were manufactured
by Dionex Corporation. The separation of the
accelerator and suppressor additives was performed
on an lonPac NSl-lO^im analytical column (250 x 4
mm). The packing material is a highly crosslinked
515
because it is present at high concentration and lacks
any interfering analytes. A 50-jiiL bath sample
(diluted 1:100) was separated with an lonPac CS12A
cation exchange column and detected with
absorbance detection at 750 nm, Figure 2.
The suppressor is one of three proprietary
additives added to improve the quality of the copper
deposition. The determination of this additive is
challenging because it does not have a strong
chromophore, and therefore cannot easily be
determined by UV photometric detection.
Evaporative light scattering detection (ELSD) was
explored as an alternative detector [5-7]. With
ELSD, the analytes of interest are separated on a
column and the effluent is passed through a
nebulizer. The mobile phase is evaporated and a
plume of non-volatile solute particles is formed. A
light beam is passed through the analyte particles and
the scattered radiation is detected. The response is
proportional to concentration.
Prior to detection with the ELDS, the suppressor
additive was separated from the other bath
components with the lonPac NSl-lOjiim polymeric
reversed phase column. Figure 3 shows the analysis
for 100 \iL of a fresh plating bath sample containing
5 ml/L suppressor. The suppressor peak at 6 minutes
is well resolved from the large matrix peak at 2
minutes. A coefficient of determination (r2) value of
0.9993 was calculated for a three level calibration
curve from 4 to 12 mL/L suppressor in the plating
bath. As the bath ages a second compound was
detected as shown in Figure 4.
Chromatographic conditions were optimized for
the determination of the Accelerator additive. Two
peaks at 5.8 and 6.5 min were attributed to the
Accelerator. The Suppressor was not detected under
these conditions.
To verify proper quantification of the additives,
increasing concentrations of the Accelerator were
spiked into the acid copper Makeup solution. Three
replicate injections were made at each calibration
level. Spikes of 1.67, 4.2 and 11.1 ml/L for the
Accelerator yielded coefficients of determination (r2)
values of greater than 0.99. The method was
validated by analyzing the same sample on two
separate days. Precision for n=18 yielded relative
standard deviation (RSD) values less than 10%.
A method using a shallow gradient from 0.25% to
9% Acetonitrile at 2 ml/min provided better
resolution of the two Accelerator peaks. A diode
array detector was utilized for this analysis because
of its ability to simultaneously gather spectral data
from a range of 190 nm to 800 nm. This allowed
precise determination of absorption maxima for the
analytes of interest.
Figure 5 shows the PDA output for a 500-^L
injection of a fresh bath spiked with 1 ml/L of
Accelerator. This is a display of the absorbance
detector output from 200 to 325 nm. The time axis
begins at 2.5 min to avoid obscuring the display with
the large matrix peak. Under these separation
conditions, the two Accelerator peaks are better
resolved from the matrix peak. No byproducts were
detected because this sample was taken prior to
electrolysis. A wavelength of 246 nm was selected to
give the best signal-to-noise ratio for the two
Accelerator peaks.
An aged bath sample was analyzed using the
shallow gradient method and a 10-^L injection size.
The Accelerator peak #1 elutes at -llmin. as with
the 500-^L injection. Two additional species, that
were not present from the analysis of a fresh bath
sample, were detected at 3.4 and 6.2 min. This
suggests that these species are byproducts that were
formed from electrolysis. These peaks were labeled
Byproduct A and Byproduct B, Figure 6. Closer
examination of the PDA spectra reveals a stronger
signal in the lower wavelengths, especially for
Byproduct A.
To identify the byproducts, possible candidates
were generated by synthesis and by electrolyzing the
additives. These compounds were spiked into the
spent bath as a means of identifying the byproduct
peaks. There was a good match in retention time and
spectral characteristics when one of these
synthesized compounds was spiked in for Byproduct
A. In the same way, Byproduct peak B was also
identified with the spike. The area of Accelerator
peak 1 also increased in this case because it was
needed for the synthesis of Byproduct peak B.
CONCLUSION
High Performance Liquid Chromatography and Ion
Chromatography provide unique capabilities for the
analysis of additives and detection of by-products in
acid copper plating baths. A non-metallic (i.e. noncorrosive) HPLC system is required for long-term
reliable analysis. Recent articles have shown that
HPLC can be used for rapid control of the
concentrations of multiple additives in copper baths.
[8] The concentration of the bath additives and the
active by-products can be confirmed and used to
control the plating quality.
516
ACKNOWLEDGMENTS
Determination of Chloride in Acid Copper
Plating Bath
The authors gratefully acknowledge the helpful
technical discussions with Richard Carey and Deyan
Wang of Shipley, Richard Hurtubise and Paul Figura
of Enthone-OMI, and John McConville of Polymer
Labs.
2-
uS
i I.
3.
4.
5.
6.
7.
8.
V
lonPa*AG15,AS15,4
30 mM Potassium
EG4
1.2
^copperplate
Detection:
Suppressed
recycle
1. Unidentified
2. Chloride 0.47
v^_
Peaks
—-———*
3. Carbonate 0
2.
L4
0— V*,_ l\^J
REFERENCES
1.
1
Column
Eluent
Eluent
Flow
Sample
R. Hurtubise, E. Too, and C. C. Cheng, Future Fab
Intl., 243-245 (1998).
S. Heberling, D. Campbell, and S. Carson, PC Fab.,
12(8), 72 (1989).
Z. Sun, and G. Dixit, Solid State Technology, 97-102
(November, 2001).
J. Reid, et. al, Solid State Technology, 86-103 (July,
2000).
G. Guiochon, A. Moysan, and C. Hoi ley J. Liq.
Chromatogr., 11,2547(1988).
J. M. Charlesworth, Anal Chem., 50 1414 (1978).
R. MacRae and J. Dick, J. Chromatogr., 210, 138
(1981).
K. Hong and H. Choi, Solid-State Technology, 57-59
(October, 2002).
5
10
15
Minute
20
25
30
4. Sulfate
17020
FIGURE 1. Determination of the chloride additive in an
acid copper plating bath.
Determination of Copper in Acid Copper
Plating Bath
Column
2.0-1
Eluent
Flow
Inj.
Sample
Detection:
s
1.5mA 1.00.5-
Key words: acid copper, chloride, suppressor,
interconnects, semiconductor, liquid chromatography
lonPac CS12A, 2mm
100 mM Methanesulfonic
0.38
50 uL
67 g/L copper sulfate diluted
Absorbance, 750
1. Copper
ISO calibration data
unavailable
6
8
Minute
10 12 14
FIGURE 2. Determination of the copper component in an
acid copper plating bath.
lonPac, PeakNet, EluGen, and ASRS are registered
trademarks of Dionex Corporation. CUBATH is a
registered trademark of Enthone OMI Inc.
Determination of Suppressor in Acid Copper
Plating Bath Before Electrolysis
Column
Eluent
lonPac NS1.10 ^m
40% Acetonitrile for 2 min
to90%Acetonitrilefrom2to7m
Flow
Inj.
Detection:
Peaks
500
LOmL/min
100 L
Evaporative Light Scattering
Copper sulfate/sulftiric acid
2. Suppressor
Mnute
FIGURE 3. Analysis of the suppressor component in a
newly made acid copper plating bath.
517
Determination of Suppressor in Acid Copper
Plating Bath After Electrolysis
Column
100'
L
Eluent
Flow
Inj.
Detection:
mV500
Peaks
3
o3
5
Minute
lonPac
NS1,
^m
40% Acetonitrile for 2 min
to 90% Acetonitrile from 2 to 7
1.0 mL/min
10(y.
Evaporative Light Scattering
1. Copper sulfate/sulfuric acid
2. Suppressor
15
10
17031
FIGURE 4. Analysis of the suppressor component in a
used acid copper plating bath . Reprinted with permission
from the Electrochemical Society.
Analysis of Acid Copper Plating Bath After
Electrolysis (Shallow Gradient)
Analysis of Acid Copper Plating Bath Before
Electrolysis (Shallow Gradient)
Column:
lonPac®NS1andNG1,10nm
2.0 mL/min
Sample Volume: 500 jiL
Detection:
Absorbance, 200-325 nr
Peaks:
1. Copper Matrix
2. Accelerator-Peak 1
3. Accelerator-Peak 2
FIGURE 5. Analysis of the accelerator additive in a
newly made acid copper plating bath. Reprinted with
permission from the Electrochemical Society and European
Semiconductor magazine.
Additive
Chloride
Accelerator
Suppressor
Leveler
lonPac®NS1andNG1,10|im
Eluent:
0.25% to 9% Acetonitrile
in150mNsulfuricacidin18m
Flow Rate:
2.0 mL/min
Sample Volume: 10 ^L
Eluent:
Flow Rate:
Column:
Detection:
UV PDA 200-325 n
Peaks:
1. By-product A
2. By-product B
3. Accelerator-Peak 1
FIGURE 6. Analysis of a used acid copper plating bath
showing the by-products of the electrolysis of the
accelerator additive. Reprinted with permission from the
Electrochemical Society and European Semiconductor
magazine.
TABLE 1. Typical Copper Plating Bath Additives[4]
Function
Other Names
Helps polymers to absorb
None
on the substrate surface
Promotes superfilling of
Brightener, Anticopper in features
suppressor, MD
Moderates activity of the
accelerator
Levels overgrowth of
copper due to superfilling
518
Ductilizer, MLo
Grain refiner, over- plate
inhibitor
Characteristics
Ionic
Sulfur containing molecules
such as sulfonic acids or
disul fides
Poly ethyl enegly col polymer
mw>1000
High MW polymers with
amine or amide functional
groups