baru per 72
TRANSCRIPT
-
7/29/2019 baru per 72
1/5
Electrochimica Acta 51 (2005) 877881
Characterisation of a thiosulphatesulphite goldelectrodeposition process
M. J-Liew, S. Sobri, S. Roy
School of Chemical Engineering and Advanced Materials, Institute of Nanoscale Science and Engineering,
University of Newcastle, Merz Court, Newcastle upon Tyne NE1 7RU, UK
Received 19 October 2004; received in revised form 20 April 2005; accepted 8 May 2005
Available online 2 August 2005
Abstract
Electrodeposition of soft gold is an important process in the fabrication of micro devices for electronics, optics etc. Traditional gold
electroplating is based on a gold cyanide process which is not applicable for the stringent requirements in state of the art micro device
manufacture. Newcastle University has been involved in the development of an industrial process based on a mixed ligand electrolytethe
gold thiosulphatesulphite system. Here we present methods forthe formulation of this electrolyte in the laboratory which ensurebath stability
and process compatibility. In addition, we have carried out spectrophotometry to elucidatethe possible reasons of its chemicalstability. Standard
rotating disk and cyclic voltammetry has been carried out to determine the electrochemical behaviour of the gold thiosulphatesulphitesystem.
The changes in electrochemical behaviour as the bath ages are also discussed.
2005 Elsevier Ltd. All rights reserved.
Keywords: Gold electrodeposition; Thiosulphatesulphite electrolyte; UVvis spectroscopy; Cyclic voltammetry; Rotating disc electrode
1. Introduction
Soft gold electrodeposition has traditionally been per-
formed usinggold cyanide or gold sulphite electrolytes. How-
ever,both these electrolytes have inherent problems regarding
long term sustainability as well as process incompatibility
[14]. This has led to search for alternative electrolytes [5,6].
However, the main problem encountered in replacing gold
cyanide baths has been that the stability of the goldcyanide
complex is very high (stability constant of 1038) and all other
ligands exhibit lower stability constants with gold in solution
[7].Over the past few years, there has been some interest
in gold deposition from an electrolyte containing two dif-
ferent complexants, sulphite as well as thiosulphate. This
electrolyte wasoriginallyproposedby Osaka and co-workers,
who used it to deposit soft gold, which was comparable
to gold deposited from a sulphite electrolyte [8,9]. They
Corresponding author. Tel.: +44 191 222 7274; fax: +44 191 222 5292.
E-mail address: [email protected] (S. Roy).
also added phosphoric acid as a buffering agent and thal-
lium ions to soften deposits. The desirable attributes of this
thiosulphatesulphite electrolyte was that it was stable near-
neutral pH which provided compatibility with photoresists
and that it was more stable than either gold sulphite or gold
thiosulphate electrolytes.
This was a breakthrough, mainly because thestability con-
stant for a goldsulphite complex is of the order 1010 [7], and
for thiosulphate it is of the order 1028 [7], both of which are
significantly smaller than that for cyanide complexes. Both
electrolytes are unstable at neutral or slightly acidic pHsin
sulphite, disproportionation of gold occurs and in thiosul-phate sulphur precipitation occurs. Some researchers have
attributed the stability of the thiosulphatesulphite electrolyte
to the formation of a new bi-ligate compound, i.e. gold is
complexed with both thiosulphate and sulphite as opposed to
either one of them [10].
In a subsequent study, our group developed a similar elec-
trolyte containing gold thiosulphate and sulphite; however,
we did not use phosphoric acid as buffer or thallium, as used
by the previous researchers. This electrolyte was shown to be
0013-4686/$ see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2005.05.060
-
7/29/2019 baru per 72
2/5
878 M. J-Liew et al. / Electrochimica Acta 51 (2005) 877881
stable, operated at near-neutral pH and showed good resist
compatibility [11,12]. The feasibility of large scale produc-
tion using this electrolyte hasbeen reported in an earlier paper
[11]. The deposit properties were found to be comparable to
those obtained using commercially available sulphite elec-
trolytes [11].
In this paper we report on (1) the method of formulationof the electrolyte, and (2) the chemical and electrochemi-
cal behaviour of the fresh and aged thiosulphatesulphite
electrolyte. The gold complexes existing in solution have
been examined by ultravioletvisible (UVvis) spectroscopy
and compared to those existing in gold thiosulphate and
gold sulphite solutions. Cyclic voltammetry was carried
out to examine the electrochemical behaviour and mass
transfer characteristics were studied using a rotating disc
electrode.
2. Electrolyte formulation
As gold thiosulphatesulphite, i.e. the combined lig-
and complex, is not commercially available, the electrolyte
was prepared in our laboratory using ACS reagent grade
quality chemicals. The gold thiosulphatesulphite elec-
trolyte was prepared by first dissolving both the complexing
agents together (0.42 M Na2SO3 and 0.42 M Na2S2O3) with
deionised water. Once dissolved, 0.05 M of gold chloride
trihydrate (or chloroauric acid), HAu(III)Cl4 (Sigma Chem-
icals) was slowly added to the solution containing the two
sulphur ligands. The (Au(III)Cl4) subsequently underwent
a homogeneous reduction reaction to form a ligand with the
thio complexes in solution, which was identified by spec-troscopy.
The most important aspect during electrolyte formula-
tion was controlling the pH. This is because the thiosul-
phate ion can disproportionate near pH 6.0, by the following
reaction:
S2O32 S0+SO3
2 (1)
At still lower pHs, the sulphite ions can form SO2. It is
therefore, imperative to maintain a pH close or above 7.0
during solution preparation. Titration reactions of solutions
containing 0.84 M Na2SO3,0.84MNa2S2O3, or both chemi-
cals, i.e.0.42M Na2SO3 and 0.42M Na2S2O3 against H2SO4showed that sodium sulphite acts as a buffering agent. The
order for mixing reactants with water, therefore, was first
to add sodium sulphite, then sodium thiosulphate and there-
after chloroauric acid. During the dissolution of these salts
in high purity water (18M cm), the pH of the electrolyte
was monitored using a Model 8521 pH meter (HANNA
Instruments) with PHM-090-010G Quick Flush glass bod-
ied pH electrode probe (Russell) to ensure that the pH was
maintained at 7.5. The pH was adjusted to 7.5 with the addi-
tion of either 0.1M NaOH or 0.1 M H2SO4, as and when
necessary.
3. Experimental
3.1. Chemical characterisation of complexes
The gold complexes in solution were identified using
ultravioletvisible spectroscopy using a Shimadzu DV-160A
spectrophotometer. In these experiments, the spectra wererecorded after filling a standard quartz cuvette (cell path
length of 1.0 cm) with a particular electrolyte, and perform-
ing a fast scan over the wavelength range of 200500 nm. The
quartz cuvette wasrinsed and dried after each scan before fill-
ing it with solution.
In these experiments, the spectra for 0.42 M Na2SO3 and
0.42M Na2S2O3 were gathered first and used as baseline
corrections for subsequent traces. UV spectroscopy analysis
was repeated for all the other constituents of the combined
electrolyte, i.e. diluted HAuCl4, gold sulphite prepared in the
laboratory (0.05 M HAuCl4, 0.42 M Na2SO3), and the same
solution as sourced from a supplier (Engelhard-CLAL) and
gold thiosulphate (0.05 M Na3Au(S2O3)). Finally, the UVspectrum of the gold thiosulphatesulphite was measured. A
comparison of all the spectra was used to identify the species
in solution. Except for pure HAuCl4 (pH 1.3) and ECF60 (pH
9.5) all other solutions were standardised to a pH of 7.0 0.5
so that differences in pH would not distort the positions of
the peak in the spectra.
3.2. Electrochemical experiments
Electrochemical studies were carried out using cyclic
voltammetry and polarisation measurements in an H-cell
and a rotating disk electrode (RDE) using a conventionalthree-electrode system. The disc electrode was used for char-
acterising the fresh electrolyte and the H-cell was used to
characterise the aged one. The anode and cathode compart-
ments of the H-cell were separated by a glass frit. The cath-
ode, i.e. working electrode, was a 1 cm 1 cm gold foil and
the anode was a 2.5 cm 2.5 cm platinised titanium sheet.
The electrolyte volume in the two compartments was approx-
imately 25.0 ml.
The rotating disk electrode was a 0.2 cm diameter plat-
inum rotating disk electrode (EDI101) embedded in Teflon in
conjunction with a CTV101 speed control unit (Radiometer
Copenhagen). The counter electrode was a 2.0 cm 1.5cm
platinum sheet. A saturated mercury sulphate electrode
(SMSE), Hg/Hg2SO4 in saturated K2SO4 reference was con-
nectedto a LugginHaber tipfilled with theelectrolyteplaced
0.2 cm from the surface of the cathode. All potentials are
reported with respect to this reference electrode. The disk
was polished using a Struers Dap-7 (Struers Ltd.) polishing
machine with a 2400 grit silicon carbide (SiC) paper, rinsed
with distilled water and thoroughly dried before each exper-
iment.
Cyclic and linear sweep voltammetry in the H-cell
and RDE were carried out using a potentiostat (Ministatt,
Sycopel) and PCI-100 data acquisition system controlled by
-
7/29/2019 baru per 72
3/5
M. J-Liew et al. / Electrochimica Acta 51 (2005 ) 877881 879
a PC using ECPROG software (Sycopel). The cathode was
polarised between 0 and2000 mV, and rotation speed of the
RDE ranged from 400 to 2000 rpm. It was assumed that the
ohmic drop in the polarisation experiments was negligible
since the distance between the working and reference elec-
trode was small. All experiments were carried out at room
temperature.
4. Results and discussion
4.1. Chemical characterisation
When the electrolyte was formulated, it was observed that
the Au(III) chloride salt, when dissolved in high purity water,
was bright yellow-orange in colour. As this coloured solution
was slowly added to the reaction mixture containing sodium
thiosulphate and sulphite complexes, upon mixing, the colour
disappeared and the final solution was clear. The other two
solutions, i.e. gold sulphite and gold thiosulphate, were both
colourless.
Fig. 1 shows the UVvis absorption spectra of all the gold
electrolytes obtained from the spectrophotometer in plots of
absorbance, A, against wavelength, , in nm. An absorption
peak at 210 nm was detected for the HAuCl4 solution con-
taining 0.6 mM Au(III) ions, followed by shoulder at 280 nm.
The shape of the spectrum is similar to that found in the
literature by previous authors [6,13] but at different wave-
lengths, i.e. although they reported the peak and shoulder
at wavelengths of 240 and 313 nm, respectively. This may
be because their source of Au(III) ions were derived from
sodium tetrachloroaurate,which would have a higher pH thanour solution.
The spectra for gold sulphite solution prepared from lab-
oratory chemicals as well as the supplier (marked as ECF60,
Englehard) are observed to exhibit an absorbance peak at
Fig. 1. UVvis absorption spectra of various gold electrolytes. The spec-
tral lines, as denoted in the figure represent the following electrolytes: (1)
aurochloric acid; (2) gold sulphite electrolyte from a vendor (ECF60); (3)
gold sulphite electrolyte made in the laboratory; (4) gold thiosulphate made
with laboratory reagents and (5) gold thiosulphatesulphite.
270 nm. The absorbance peak for the gold thiosulphate solu-
tion, on the other hand, lies at 285 nm. The peak for the gold
thiosulphatesulphite solution is observed at 283 nm, similar
to that of gold thiosulphate. UVvis spectra were also col-
lected from solutions while carrying out electrodeposition in
an H-cell for over 20 min. It was found that the absorbance
peak remained between 285 and 287 nm throughout the pro-cess, which showed that the gold remained ligated to the
thiosulphate in solution during the electrodeposition process.
This suggests that the gold in the combined solution exists as
a goldthiosulphate complex. This is consistent with the fact
that gold forms a far more stable complex with thiosulphate
than with sulphite [7].
4.2. Electrochemical characterisation
Cyclic voltammetry data for a gold thiosulphatesulphite
electrolyte at a rotating disk electrode at a rotation speed of
400 rpm and a scan rate of 10 mV/s is presented in Fig. 2a.
Although a cathodic current is observed at electrode poten-tials below 500 mV, with a shoulder around 700 mV, no
gold was deposited at these potentials. Gold reduction was
observed near the peak at 1200 mV. Anodic currents are
Fig. 2. (a) Cyclic voltammogram at a RDE for gold thiosulphatesulphite
electrolyte, at 400 rpm. Scan rate 10 mV/s and (b) cyclic voltammogram at
a RDE for gold sulphite electrolyte at the same rotation speed and scan rate.
-
7/29/2019 baru per 72
4/5
880 M. J-Liew et al. / Electrochimica Acta 51 (2005) 877881
Fig.3. Cyclic voltammogramobtainedat RDEfor goldthiosulphatesolution
(Na3Au(S2O3)2) at 400rpm, 10 mV/s.
observed when the electrode potential exceeds100 mV, and
two clear anodic peaks are observed; one at about 50mVandanother at 400 mV. Thecyclic voltammetry for a gold sul-
phiteelectrolyte is shown in Fig.2b, whichis clearly different,
(1) no current is observed until a high cathodic overpotential
is attained (i.e. 1000 mV) and (2) no current was observed
at all during the anodic half of the cycle. These data are in
direct contrast with that observed in Fig. 2a and suggest that
it is unlikely that gold is complexed with the sulphite ligand.
In order to identify some of the peaks observed in
the experiment described in the previous paragraph, cyclic
voltammetry was carried out with a solution containing 0.5 M
Na3Au(S2O3) (sodium gold thiosulphate) only. The cyclic
voltammogram for this solution is exhibited in Fig. 3. Thedata show similar characteristics to the thiosulphatesulphite
solution, except that the cathodic shoulder at 700 mV and
the anodic peak at 50 mV are absent. This finding also sup-
ports the conjecture that gold is reduced from a gold thiosul-
phate species.
The polarisation data for gold deposition from the
thiosulphatesulphite electrolyte at a rotating disk electrode
at different rotation speeds is illustrated in Fig. 4. These data
were collected to determine the limiting current and the scan
rate was set at 10 mV/s. As is often observed for reduction of
metal from complexes, there is no clear mass transfer limit-
ing current plateau [14]; the arrow marks the potential which
was identified to correspond to where the limiting current
occurred. The inset in Fig. 4 shows the Levich plot derived
from the limiting current experiments. However, since the
data do not pass through the origin for rotation speed equal
to zero, a LevichKoutecky plot was used to determine the
diffusion coefficient, which is shown in Fig. 4b. The diffu-
sion coefficient, as calculated from the data was found to
be 1.2 106 cm2/s. In addition, polarisation data for gold
thiosulphate was also collected and analysed [15] (which are
not included here for brevity). The diffusion coefficients for
gold thiosulphate from those experiments were found to be
6.3 107 cm2/s [15]. The diffusion coefficient of the mixed
Fig. 4. (a) Limiting current data of gold deposition from a thiosulphate
sulphiteelectrolyte at different rotationspeeds.Scanning rate10 mV/s. Inset:
Levich plot ofIL vs.1/2
and (b) a LevichKoutecky analysis of the limitingcurrent data.
ligand electrolyte lies somewhere between the diffusivities of
gold thiosulphate and gold sulphite electrolytes (the diffusiv-
ity of gold sulphite was found to be 6 106 cm2/s [15,16]).
Our chemical and electrochemical experiments with the
thiosulphatesulphite electrolyte indicate that gold is com-
plexed with the thiosulphate ligand, as has been proposed
by other researchers previously [17,18]. The UVvis spec-
tra show that the spectrum of the combined gold electrolyte
and gold thiosulphate are very similar. In addition, the elec-
trochemical behaviour (cyclic voltammetry as well as lim-
iting current polarisation data) for gold deposition from thethiosulphatesulphite electrolyte is similar to those observed
for gold thiosulphate. Therefore, we believe, gold is dis-
charged from a goldthiosulphate complex.
4.3. Studies on aged electrolyte
In order to induce accelerated aging, the thiosulphate
sulphite electrolyte was used extensively in an industrial pro-
cess (for over 23 weeks) during which period the equivalent
of 2 months of total production was simulated. At the end
of this period, there was no discernible change in the pH,
-
7/29/2019 baru per 72
5/5
M. J-Liew et al. / Electrochimica Acta 51 (2005) 877881 881
although a significant change was observed in the electrode-
posits. The deposits were rough, dark and were beginning to
become powdery. The electrolyte was then collected and left
to ageover 6 months. This aged electrolyte wasthen collected
and their pH was measured; it was found to be 5.2.
In order to determine if there was any significant change
in the electrochemical properties of the electrolyte, cyclicvoltammetry experiments were carried out in an H-cell.
Although the anode and cathode compartments were mon-
itored to see if any changes occurred during electrolysis, no
significant visible effects were found. The polarization data,
however, showed two separate peaks; one at0.25 V and the
second one at 0.65 V, which are much more anodic com-
pared to the fresh electrolyte.
It was noted that gold deposits were of acceptable quality
only in the fresh electrolyte, where a significant quantity of
sulphite is present in the electrolyte. The overpotential for
the commencement of gold deposition was at lower cathodic
potentials in the aged electrolyte. The difference in poten-
tial for gold electrodeposition could be due to an adsorbedspecies, which yields a smooth and dense deposit. We believe
this species is sulphite, since sulphite electrolytes are char-
acterised by a high cathodic overpotential for gold reduction
and anodic currents are entirely absent.
A second aspect of bath aging was the decrease of solution
pH with time (even in the absence of electrolysis). Although
the pH decreased very slowly (over months), this rendered
the electrolyte less stable and prone to precipitation. How-
ever, precipitates were not visible to the naked eye and no
turbidity was detected (as is usually observed when gold
disproportionates). This indicates that a slow homogeneous
reaction occurs within the solution which leads to a decreasein pH. It is important to understand the role of homogeneous
reactions, since they determine long term bath stability.
5. Conclusions
This study shows the chemical and electrochemical prop-
erties of a gold thiosulphatesulphite solution which can
be used for soft gold deposition. UVvis spectroscopy
showed that the spectrum for gold thiosulphate and gold
thiosulphatesulphite are very similar. Cyclic voltammetry
at a rotating disc electrode and limiting current data showed
that electrochemical data for the thiosulphatesulphite elec-
trolyte is similar to that obtained for gold thiosulphate solu-
tion. These results strongly support that gold is deposited
from a goldthiosulphate complex. It was found that an aged
electrolyte showed changes in chemical and electrochemical
behaviour.
Acknowledgements
The authors would like to thank Dr. T.A. Green for his
advice on solution chemistry and P.A. Christensens group
for help of UVvisspectroscopy. This work wassupported by
EPSRC GR/M64314 and an HEFCE/EPSRC JIF-4NESCEQ
grant.
References
[1] J. Traut, J. Wright, J. Williams, Plat Surf. Finish. 81 (1994) 52.
[2] K. Kosaki, M. Matsuoka, Y. Seiwa, S. Orisaka, K. Nishitani, M.
Otsubo, in: M. Datta, K. Sheppard, D. Snyder (Eds.), The Elec-
trochemical Society Proceedings Series, PV 92-3, Pennington, NJ,1992, p. 317.
[3] H. Watanabe, S. Hayashi, H. Honma, J. Electrochem. Soc. 146
(1999) 574.
[4] A. Gemmler, W. Keller, H. Richter, K. Ruess, Plat. Surf. Finish. 81
(1994) 52.
[5] H. Honma, Y. Kagaya, J. Electrochem. Soc. 140 (1993) L135.
[6] H. Honma, K. Hagiwara, J. Electrochem. Soc. 142 (1995) 81.
[7] L.G. Sillen, A.E. Martell, Stability Constants of MetalIon Com-
plexes. Part II. (Inorganic Ligands), The Chemical Society, London,
1971.
[8] T. Osaka, A. Kodera, T. Misato, T. Homma, Y. Okinaka, J. Elec-
trochem. Soc. 144 (10) (1997) 3462.
[9] T. Osaka, M. Kato, J. Sato, K. Yoshizawa, T. Honma, Y. Okinaka,
O. Yoshioka, J. Electrochem. Soc. 148 (2001) C659.
[10] T. Inoue, S. Ando, H. Okudaira, J. Ushio, A. Tomizawa, H. Take-hara, T. Shimazaki, H. Yamamoto, H. Yokono, Proceedings of the
45th IEEE Electronic Components Technology Conference, 1995,
p. 1059.
[11] T.A. Green, M.-J. Liew, S. Roy, J. Electrochem. Soc. 150 (3) (2003)
C104.
[12] M.J. Liew, S. Roy, K. Scott, Green Chem. 5 (2003) 376.
[13] X. Xu, C.L. Hussey, J. Electrochem. Soc. 139 (1992) 3103.
[14] P. Bradley, S. Roy, D. Landolt, J. Chem. Soc., Faraday Trans. 92
(20) (1996) 4015.
[15] M.-J. Liew, Thesis: Novel Gold Electrodeposition Process for Micro
and Opto Electronics, University of Newcastle, 2002.
[16] S. Caprodossi, S. Roy, in: T. Osaka (Ed.), Proceedings of the Sympo-
sium on Electrodeposition, PV 99-34, The Electrochemical Society,
Pennington, NJ, USA, 2000, p. 145.
[17] M. Kato, K. Niikura, S. Hoshino, I. Ohno, J. Surf. Finish. Soc. Jpn.42 (1991) 729.
[18] M. Kato, Y. Yazawa, Y. Okinaka, Proceedings of the SUR/FIN95
(1995) 805.