Microelectronic Engineering 86 (2009) 12821285

Contents lists available at ScienceDirect

Microelectronic Engineering
journal homepage: www.elsevier.com/locate/mee

Gold cleaning methods for electrochemical detection applications

Lee M. Fischer a,*, Maria Tenje a, Arto R. Heiskanen a, Noriyuki Masuda a, Jaime Castillo a,
Anders Bentien b, Jenny mneus a, Mogens H. Jakobsen a, Anja Boisen a
a
b

DTU Nanotech Department of Micro- and Nanotechnology, Technical University of Denmark, Building 345 East, DK-2800 Kongens Lyngby, Denmark
Grundfos A/S, Poul Due Jensens Vej 7, DK-8850 Bjerringbro, Denmark

a r t i c l e

i n f o

Article history:
Received 29 September 2008
Received in revised form 26 November 2008
Accepted 26 November 2008
Available online 3 December 2008
Keywords:
Gold cleaning
Electrochemistry
Cyclic voltammetry
Impedance spectroscopy

a b s t r a c t
This work investigates methods for obtaining reliably clean gold lm surfaces. Nine gold cleaning methods are investigated here: UV ozone photoreactor; potassium hydroxidehydrogen peroxide; potassium
hydroxide potential sweep; sulfuric acid hydrogen peroxide; sulfuric acid potential cycling; hydrochloric
acid potential cycling; dimethylamine borane reducing agent solutions at 25 and 65 C; and a dilute form
of Aqua Regia. Peak-current potential-differences obtained from cyclic voltammetry and charge transfer
resistance obtained from electrochemical impedance spectroscopy, as well as X-ray photo-electron spectroscopy are used to characterize surface cleanliness. A low peak-current potential-difference and charge
transfer resistance indicates a cleaner surface, as does a higher percentage of elemental gold on the electrode surface. The potassium hydroxide potential sweep method is found to leave the gold surface the
cleanest overall.
2008 Elsevier B.V. All rights reserved.

1. Introduction
Gold is widely used in microfabrication for a variety of applications [1]. The interaction between gold surfaces and thiol-terminated molecules has been widely studied and leveraged for
biological purposes [2,3]. Gold is also used as an electrochemically
active surface on which the behaviors of chemical or biological
samples are studied [4,5]. As such, in electrochemical experiments,
the quality of the active surface will affect the measurements. The
peak currents in cyclic voltammetry and the frequency response
during electrochemical impedance spectroscopy will be dependent
on the surface composition of the gold. Upon exposure to a noncleanroom laboratory environment the gold surface is subject to
numerous ambient contaminants, affecting binding kinetics of thiols in addition to electrochemical effects [6]. It is for this reason
that gold, especially evaporated gold, must be cleaned immediately
before it is chemically modied for these uses.
This work describes the application of nine treatments meant to
clean the surface of the gold. Following cleaning, the gold samples
were characterized using cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). Surface composition was
investigated using X-ray photo-electron spectroscopy (XPS). The
characterization methods used here investigate the electrochemical activity of the gold surface and analyze its elemental composition. The results obtained here are to be applied to future
* Corresponding author. Tel.: +45 4525 5759; fax: +45 4525 7772.
E-mail address: Lee.Fischer@nanotech.dtu.dk (L.M. Fischer).
0167-9317/$ - see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.mee.2008.11.045

electrochemical detection studies to ensure reliably clean gold surfaces and maximize stable coverage of thiol monolayers.

2. Electrode fabrication
Gold electrode chips were fabricated on 100 mm (1 0 0) silicon
wafers wet oxidized at 1050 C for 1.5 h to form 500 nm of SiO2
as an electrical insulating layer. Standard UV lithography and liftoff processes were used to pattern the 300 nm Au on 10 nm Ti electrode layer deposited by electron-beam evaporation. A 500 nm
PECVD silicon nitride layer was deposited using a STS PECVD reactor. A second UV lithography step followed by a SF6 and O2 STS RIE
plasma etch dened the 1 mm2 electrode and 3 mm by 2 mm clip
contact pad openings in the nitride layer. Acetone removed the
bulk of the resist and a 10 min plasma ashing step removed residual resist. Finally, the wafers were removed from the cleanroom for
dicing.

3. Cleaning methods
Two gold etches and a variety of common gold cleaning methods were investigated. All compounds mentioned are diluted with
Milli-Q water. Each cleaning method was applied to four gold samples for statistical relevance. The applied treatments are described
below, including solution concentrations and voltammetry parameters. Cleaning times are explicitly stated for non-voltammetric

L.M. Fischer et al. / Microelectronic Engineering 86 (2009) 12821285

1283

techniques, while all voltammetric techniques were accomplished

in under 2 min.

only spent 2 min in this solution as the gold was visibly etched
after 8 min. Abbreviated Aqua Regia in the results.

3.1. Ultraviolet ozone cleaning

4. Characterization methods

Gold samples spent 40 min in a Merck Eurolab PR-100 UV ozone

Photoreactor. This treatment is commonly used when working
with biological samples as a way to restore or regenerate the gold
surface after experiments [7,8]. Abbreviated UV in the results.

4.1. Electrochemical characterization

3.2. Potassium hydroxide and hydrogen peroxide

This method has been used by Heiskanen et al. as an intermediate step in a gold cleaning protocol [4]. Samples spent 10 min in a
solution of 50 mM KOH and 25% H2O2 before rinsing with Milli-Q
water. Abbreviated KOH + H2O2 in the results.
3.3. Potassium hydroxide potential sweep
This is the second part of the cleaning protocol used by Heiskanen et al. [4]. After the treatment described above samples were
placed in 50 mM KOH and connected to a potentiostat. The electrode potential was swept from 200 to 1200 mV (vs. Ag/AgCl)
once, at 50 mV/s scan rate, and then rinsed in Milli-Q water. Abbreviated KOH sweep in the results.
3.4. Sulfuric acid and hydrogen peroxide

A CH Instruments 660C Electrochemical Analyzer was used to

perform all CV and EIS. The probe solution consisted of 10 mM ferri/ferro-cyanide [Fe(CN)6]3 /4 redox couple and 200 mM KCl (reference/counter-electrolyte) in Milli-Q water. A chlorinated silver
wire formed the Ag/AgCl pseudo-reference electrode. All electrochemical measurements were conducted in a Dr. Bobs glass cell
(Gamry Instruments) with 6 mL of probe solution without stirring.
Fresh probe solution was used for every different cleaning method
sample set.
Before and after cleaning each electrode was subjected to CV of
ve cycles sweeping the potential from 300 to 700 mV. Additionally, EIS was performed by applying a 10 mV AC voltage on a DC
bias of 280 mV at frequencies from 0.1 Hz to 100 kHz and measuring the frequency response. Each cleaning technique and accompanying measurements were performed on four separate samples to
generate statistically signicant results.
4.2. X-ray photo-electron spectroscopy

A very mild version of a piranha clean is formed from 50 mM

H2SO4 and 25% hydrogen peroxide [9]. Gold samples spent
10 min in this treatment before being rinsed with Milli-Q. Abbreviated H2SO4 + H2O2 in the results.

A Ka X-ray photo-electron spectrometer from Thermo Scientic

was used to conduct XPS to obtain an atomic percent (at.%) composition readout of the electrode surface following each of the cleaning methods, as well as the uncleaned electrode. A nal sample
was argon-sputtered and scanned until a stable composition reading was obtained, indicating the bulk gold had been reached.

3.5. Sulfuric acid potential cycling

5. Results

Cycling the electrode potential in a weak sulfuric acid solution

until a stable CV scan is achieved is a very common electrochemical cleaning technique [5,10]. Sample potential was cycled from
400 to 1400 mV (vs. Ag/AgCl) at a rate of 100 mV/s in 50 mM sulfuric acid until the CV becomes stable (approximately 12 cycles).
Abbreviated H2SO4 CV in the results.

5.1. Electrochemical results

Following fabrication and dicing, all samples spent 2 weeks on a
shelf in the lab. It is known that, in ambient lab storage conditions,

3.6. Hydrochloric acid potential cycling

Gold is known to form a stable compound with chlorine [1,11],
thus this cleaning method is actually more of an electrochemical
etch. Samples are subjected to only three potential cycles from
500 to 1500 mV (vs. Ag/AgCl) at 100 mV/s scan rate in 50 mM
HCl since gold is visibly removed from the surface after 10 cycles.
Abbreviated HCl CV in the results.
3.7. Reducing agent solutions
This technique is based on the principle of trying to reduce the
gold oxide on the electrode surface to form metallic gold. This solution contains the reducing and electron-supplying agents from an
electroless deposition recipe for gold [12]: 800 mM KOH,
750 mM Na2CO3, and 50 mM dimethylamine borane (DMAB).
One set of samples spent 10 min in this solution at room temperature, abbreviated DMAB@25C, and another set at 65 C, abbreviated DMAB@65C.
3.8. Aqua Regia
A weak version of the extremely powerful Aqua Regia metal
etchant was formed from 1 M HNO3 and 3 M HCl [11]. Samples

Fig. 1. (a) A sample cyclic voltammagram. The potential-difference in the positions

of the peak anodic (Epa) and cathodic (Epc) currents, labeled DEP, is used as a
measure of cleanliness. (b) The Randles equivalent circuit model of the electrochemical system. The charge transfer resistance, Rct, provides a measure of surface
cleanliness.

1284

L.M. Fischer et al. / Microelectronic Engineering 86 (2009) 12821285

Table 1
Potential difference and charge transfer resistance for each gold cleaning method. A
negative sign indicates a percentage decrease from uncleaned values.
Cleaning methods

DE p
%D

UV
KOH + H2O2
KOH sweep
H2SO4 + HO2
H2SO4 CV
HCI CV
DMAB@25C
DMAB@65C
Aqua Regia

7.1
20.3
28.4
9.3
9.8
17.2
8.1
18.7
8.8

Table 2
Description of preliminary AFM results.
Sample description

RMS roughness
(nm)

Island width
(nm)

Island pitch
(nm)

2 weeks old uncleaned

2 years old uncleaned
2 weeks old
KOH + H2O2
2 years old KOH
sweep

1.10
0.30
0.70

40
100
40

100
200
80

0.65

40

80

Rct

rE
12.0
3.1
3.3
3.5
3.1
5.8
1.0
2.2
6.2

%D
40
63
71
38
6
52
27
69
58

rR
31
9
6
9
17
10
4
6
18

the gold surface is quickly covered in particles [13], thus we used

this as standard contamination for our tests.
The potential-difference between the peak cathodic and anodic
currents, DEP, is used as a measure of electrochemical cleanliness
of the electrode surface, seen in Fig. 1a, with a smaller DEP indicating a cleaner surface. Theoretically, for single-electron transfer
reactions such as in the [Fe(CN)6]3 /4 couple on a perfect gold surface, the potential-difference should be DEP = 58 mV [14], and we
interpret any increase in this value to be caused by surface imperfections or contaminations.
The frequency response from the EIS measurements were t to
the Randles equivalent circuit seen in Fig. 1b [14], from which the
charge transfer resistance, Rct, was extracted. Since Rct is a measure
of the resistance encountered by the electron as it travels between
the redox species and electrode surface, a lower Rct indicates a cleaner and more electrochemically active surface.
Table 1 shows the change in DEP and Rct as a percent difference
(%D) from their original, uncleaned sample values, while the standard deviation (r) is given as a percentage of the original DEP or Rct
values. All samples exhibited values of DEP = 99.2 3.6 mV and
Rct = 259 41 X before cleaning. Considering this, and the theoretical minimum value DEP = 58 mV, a perfectly clean gold surface
would have a percentage decrease in DEP of %D = 41.5%.
The large standard deviation in the UV cleaned samples indicating possible contamination of the UV photoreactor due to widespread use. The KOH + H2O2 and KOH sweep methods appear to
deliver the cleanest gold, while the very common H2SO4 CV method delivers a surface less clean than expected. The HCl CV, Aqua

Regia, DMAB@25C and DMAB@65C methods appear to have similar

effects on the electrode surface.
5.2. XPS results
Automated peak-tting analysis in the Ka spectrometer provided elemental composition information. The at.% composition
is plotted in Fig. 2, ignoring trace signals. We expect a larger percentage of elemental gold on the surface to indicate a cleaner surface, while carbon indicates contamination. The shift in the oxygen
peak indicates the majority of oxygen found on the surface is in the
form of metal oxide. Chlorine signals are only observed in the samples cleaned in acidic chlorine environments, and in the UV
cleaned sample, possibly due to contamination of the photoreactor.
Again, the KOH sweep sample is believed to be the cleanest as it
has the highest amount of elemental gold on the surface, but the
H2SO4 CV sample has the second highest gold percentage, which
is unexpected after the electrochemical results. The HCl CV etch
and Aqua Regia etch contain a lower than expected amount of elemental gold, while the DMAB@25C and DMAB@65C methods appear to behave as etches and do not reduce the surface gold oxide.
The H2SO4 CV and HCl CV results appear contradictory between
the electrochemical and XPS measurements. One possible explanation of this is that potential cycling in acidic solution affects the
surface charge of the gold. Since the redox couple being used is
negatively charged it may affect the electrochemical measurements. Additional electrochemical characterization using a positively charged redox couple probe solution will be employed in
future studies to address this issue.
5.3. AFM results
Preliminary contact-mode AFM was performed on two samples.
One sample was from the batch of microfabricated electrodes used
in this paper, stored for 2 weeks in the lab, then scanned before and
after the KOH + H2O2 cleaning method. The second was a 2-yearold gold sample from the same deposition machine, scanned before
and after treatment in the combined KOH + H2O2 and KOH sweep
method. Table 2 describes the results where samples are characterized by RMS roughness, width of the gold islands, and pitch of the
gold islands, giving an indication of the surface structure. Uncleaned sample scans, especially the 2-year-old sample, appear
noisier with less visible islands, while after cleaning the islands appear clear and the surface smooth. Also note that both samples exhibit nearly identical roughness, width, and pitch after cleaning,
despite their prior differences. Further AFM work is to be done,
but this is promising initial evidence that the surface structure is
improved after cleaning, in addition to the electrochemical properties and chemical composition.
6. Conclusions

Fig. 2. Elemental surface composition of gold samples following each cleaning

technique.

After investigating these cleaning methods we can conclude the

KOH + H2O2 and KOH sweep combined cleaning method is the best

L.M. Fischer et al. / Microelectronic Engineering 86 (2009) 12821285

technique, while the H2SO4 CV method appears to be acceptable

when investigated using XPS. The H2SO4 + H2O2 method severely
oxidizes the gold; while the HCl CV and Aqua Regia etches do
not appear to produce a signicantly cleaner surface. Finally, the
attempt to reduce the surface gold oxide using the electroless-style
solution in DMAB@25C and DMAB@65C is ineffective and merely
behaves as an etch. Preliminary AFM supports the effectiveness
of these cleaning techniques and will be investigated further.
Acknowledgments
Fabrication of the gold electrodes was accomplished at DTU
Danchip, while XPS analysis was performed at Ris DTU. The
authors would like to acknowledge Lene Hubert for her assistance
in XPS analysis, and the entire staff at DTU Danchip for their maintenance efforts and microfabrication expertise.
References
[1] T.A. Green, Gold Bull. 40 (2007) 105.

1285

[2] M. Mrksich, Cell. Mol. Life Sci. 54 (1998) 653.

[3] B. Leca-Bouvier, L.J. Blum, Anal. Lett. 38 (2005) 1491.
[4] A.R. Heiskanen, C.F. Spgel, N. Kostesha, T. Ruzgas, J. mneus, Langmuir 24 (16)
(2008) 9066.
[5] C. Spgel, A. Heiskanen, J. Acklid, A. Wolff, R. Taboryski, J. mneus, T. Ruzgas,
Electroanalysis 19 (23) (2007) 263.
[6] J. Kang, P.A. Rowntree, Langmuir 23 (2007) 509.
[7] N.S. McIntyre, R.D. Davidson, T.L. Walzak, M. Westcott, A. Pekarsky, J. Vac. Sci.
Technol. A 9 (3) (1991) 1355.
[8] C.G. Worley, R.W. Linton, J. Vac. Sci. Technol. A 13 (4) (1995) 2281.
[9] T.M.H. Lee, I.-M. Hsing, Anal. Chem. 74 (2002) 5057.
[10] M.F.M. Noh, I.E. Tothill, Anal. Bioanal. Chem. 386 (2007) 2095.
[11] K.R. Williams, K. Gupta, M. Wasilik, J. Microelectromech. Syst. 12 (6) (2003)
761.
[12] P.T. Tang, Fabrication of Micro Components by Electrochemical Deposition,
Doctoral Dissertation, Technical University of Denmark, 1 March 1998.
[13] T. Ishida, S. Tsuneda, N. Nishida, M. Hara, H. Sasabe, W. Knoll, Langmuir 13
(1997) 4638.
[14] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and
Applications, second ed., John Wiley and Sons, Hoboken, NJ, 2001.