Sputtered gold films for surface-enhanced Raman scattering

L. Maya, C. E. Vallet, and Y. H. Lee

Citation: Journal of Vacuum Science & Technology A 15, 238 (1997); doi: 10.1116/1.580518
View online: http://dx.doi.org/10.1116/1.580518
View Table of Contents: http://scitation.aip.org/content/avs/journal/jvsta/15/2?ver=pdfcov
Published by the AVS: Science & Technology of Materials, Interfaces, and Processing

Articles you may be interested in

Recyclable surface-enhanced Raman scattering template based on nanoporous gold film/Si nanowire arrays
Appl. Phys. Lett. 105, 011905 (2014); 10.1063/1.4889850

Dynamic plasmonic tweezers enabled single-particle-film-system gap-mode Surface-enhanced Raman scattering

Appl. Phys. Lett. 103, 191119 (2013); 10.1063/1.4829617

Surface acoustic wave velocity of gold films deposited on silicon substrates at different temperatures
J. Appl. Phys. 110, 023503 (2011); 10.1063/1.3606412

Gold surface with gold nitride–a surface enhanced Raman scattering active substrate
J. Appl. Phys. 105, 054302 (2009); 10.1063/1.3082871

Synthesis of highly oriented piezoelectric AlN films by reactive sputter deposition

J. Vac. Sci. Technol. A 18, 1609 (2000); 10.1116/1.582394

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 131.193.242.67 On: Sun, 30 Nov 2014 08:35:00
 Sputtered gold films for surface-enhanced Raman scattering
L. Maya,a) C. E. Vallet, and Y. H. Lee
Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, Oak Ridge,
Tennessee 37831-6119
~Received 12 September 1996; accepted 20 December 1996!
Sputtered gold films in a pure form or as nanocomposites in silica or silicon nitride were screened
for surface-enhanced Raman scattering ~SERS! activity using Rhodamine 6G as a probe. The films
were prepared by sputtering pure gold or solidified Au–Si alloys in plasmas generated in a dc glow
discharge apparatus. The plasmas were produced with argon, nitrogen, or argon–oxygen as the
sputtering gas to directly deposit gold films or in the latter case a gold oxide intermediate. The alloys
produce nanocomposite films in a silicon nitride or silica matrix depending on the plasma gas. SERS
activity was detected in some of the films thus leading to a search for the critical parameters that
controlled this phenomenon. The films were characterized by profilometry, x-ray diffraction, and
atomic force microscopy. SERS activity was found to be correlated to crystallite size in the 10–25
nm range and to roughness larger than 15 nm, and it was independent of film thickness. Sputtered
gold films, particularly those containing the gold as a nanocomposite in silica are attractive media
for SERS because of excellent adherence, ruggedness, and simplicity in preparation. © 1997
American Vacuum Society. @S0734-2101~97!03402-7#

I. INTRODUCTION perimental protocols to generate appropriate degrees of ag-

The discovery1 and initial development2,3 of surface-
enhanced Raman scattering ~SERS! about 20 years ago
marked the beginning of considerable scientific interest that
is still quite active as reflected in recent reviews.4,5 SERS is
a phenomenon observed in the course of conducting Raman
spectroscopy of molecular species adsorbed on suitable sur-
faces, most commonly, specially treated noble metals and it
involves as the name implies a manyfold enhancement of the
Raman-active vibrational modes. The accepted understand-
ing of the enhancement mechanism identifies two distinct
components; an electromagnetic one, stemming from the ex-
citation of surface plasmon polaritons and a chemical effect
arising from charge transfer between the surface and the ad-
sorbed species. SERS is of importance because it enables the
study of molecular interactions at the liquid–solid interface
to provide a deeper understanding of phenomena associated
with catalysis, separations, transport, diffusion, etc. and can
be applied as a sensitive tool to detect a wide variety of
analytes.6 One of the critical issues in applying SERS as a
tool is the development of surfaces with the right character-
istics, such as roughness in the range of 3–100 nm7,8 which
may be reliably produced. Initially, surfaces used for SERS
were produced by electrochemical means, later by metal
evaporated unto rough surfaces and most recently the ap-
proach being explored involves the use of noble metal col-
loids in suspension4 or dispersed on chemically modified
surfaces.8,9 The importance of microstructure was recognized
early on but there was no simple way to generate or measure
the roughness of the surfaces. The initial electrochemical
work followed empirical observations to generate an ad-
equate degree of roughness; similarly, use of noble metal
colloids10 in a flow injection analysis apparatus required ex-
a!
Electronic mail: mayal@ornl.gov
gregation and therefore roughness to obtain adequate SERS
signals. A more elaborate procedure to generate roughness in
a well defined pattern involved11 the creation of microscopic
silica posts that were then coated on the tips with evaporated
silver deposits. Simpler approaches evolved through the use
of microparticles such as silica or alumina12 which were used
as supports to evaporated noble metal films or inmobilized
colloid on derivatized surfaces.8
The advent of scanning probe microscopies following the
discovery and development work of Binning et al.13 has pro-
vided an accessible tool to image and measure surfaces down
to atomic resolution. We have employed this technique in the
present study to explore the suitability of sputtered gold films
for SERS which were generated taking advantage of our ex-
perience with nanocomposite gold films produced directly or
through an intermediate gold oxide precursor.14,15
II. EXPERIMENT
Films of gold or gold oxide in a pure form or as nano-
composites in silica or silicon nitride matrices were gener-
ated by sputtering in a plasma produced in a simple parallel
plate dc glow discharge apparatus described previously.16 In
a typical run, sputtering was conducted for a few hours while
passing 10 ml/min of a purified gas ~nitrogen, argon, or ar-
gon:oxygen 60:40! and maintaining a pressure of 0.26 kPa.
Typical plate voltage and current densities were 400 V and
1.8 mA/cm2, respectively. The sputtering targets were, pure
gold foil, or solidified Au/Si alloys containing 31 at % Si or
95 at % Si. Films between 0.05 and 4.0 mm in thickness were
generated by sputtering for 1–18 h. Substrates consisted of
mica, quartz, and acid-rinsed ~5 vol % HF! single crystal
silicon having a @111# orientation. The coating on silicon was
done on the mirror-polish side with the exception of a few
specimens coated on the unpolished side. Coating took place

238 J. Vac. Sci. Technol. A 15(2), Mar/Apr 1997 0734-2101/97/15(2)/238/5/$10.00 ©1997 American Vacuum Society 238

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 131.193.242.67 On: Sun, 30 Nov 2014 08:35:00
 239 Maya, Vallet, and Lee: Sputtered gold films for SERS 239

FIG. 2. Rhodamine 6G spectra on ~A! a SERS active film and ~B! an inactive
FIG. 1. Experimental setup to study SERS. film. Test solution concentrations were 1025 and 1023 M for A and B,
respectively.

without any external heating. The backside temperature of

was indeed the case in the course of the present study involv-
the cathode immediately after extinguishing the plasma did
ing the examination of over 50 samples and SERS activity
not exceed 100 °C in any run. Some films were subjected to
was expressed as a measurable signal ~which within an order
rapid thermal annealing in order to induce crystallite growth
of magnitude! was about 104 of the response of Rhodamine
by heating a few seconds at 400 °C. The films were screened
6G in the absence of an active surface. This is illustrated by
for SERS activity using Rhodamine 6G as a probe. The test
the spectra given in Fig. 2 showing the response of an active
consists of depositing a 10 ml aliquot of a dilute solution,
film and that of an inactive surface; note that the concentra-
either 1023 or 1025 molar solution in methanol on the sur-
tion used in the latter is 100-fold higher than that of the first
face to be tested. The drop typically spread into a 1 cm
and produces a signal just barely above the noise level. The
diameter spot which was then allowed to evaporate in air. A
aim of the present study was then the identification of the
schematic representation for SERS measurement used in the
significant parameters leading to the preparation of not only
present study is shown in Fig. 1. Further details are given in
an active surface, but one which is easily prepared, mechani-
Ref. 17. Additional characterization included x-ray diffrac-
cally rugged, and reusable. The influence of different param-
tion ~XRD! to establish average crystallite size from line
eters examined will be discussed in turn.
broadening, profilometry for film thickness and atomic force
microscopy ~AFM! to establish film microstructure including
a measure of film roughness. Microscopy was conducted us- A. Film thickness
ing a Digital Interments Nanoscope III operated in air in the SERS response was found in films ranging in thickness
contact mode with a Si3N4 tip. Further details will be given covering the whole range studied here; namely, 0.05–4.0
in a separate report. The images were acquired on 1 mm31 mm. This confirms the notion that SERS is truly a surface
mm surfaces with a scan rate of 7–10 Hz. The mean rough- phenomenon which can be elicited even in films a few na-
ness R a of a 1 mm2 area was calculated from images ob- nometers in thickness7–9 or the top layer of bulk metal.5
tained using the following equation:

R a5 S DE E
1
L xL y
Ly

1
Lx

1
u f ~ x,y ! u dx dy,
where f (x,y) is the surface relative to the center plane and
L x and L y are the dimensions of the surface.
III. RESULTS AND DISCUSSION
Consistent with the current understanding of the SERS
phenomena, a surface would be active if it fulfills the re-
quirements of surface morphology that are appropriate for
the excitation of the plasmons and/or energy transfer to ad-
sorbed species. A gradual response is not expected but a
situation whereby a signal is either present or absent. This
B. Substrate
All films showed random crystallographic orientation in-
dependent of the substrate; similarly, SERS activity was also
found to be independent of the substrate, even for films de-
posited on unpolished single crystal silicon which had an
average roughness of 120 nm provided that the film thick-
ness completely covered that roughness. This is in contrast to
the fact that the crystallographic texture of gold films evapo-
rated on silicon substrates having different crystallographic
orientations are affected by the substrate.18,19 This difference
is apparently due to the fact that sputtering involves a more
energetic environment that creates multiple nucleation sites
and leads to a morphology with poorly oriented texture.20

JVST A - Vacuum, Surfaces, and Films

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 131.193.242.67 On: Sun, 30 Nov 2014 08:35:00
 240 Maya, Vallet, and Lee: Sputtered gold films for SERS
FIG. 3. Micrographs of SERS active films with corresponding section profiles.
C. Plasma source
As mentioned in the experimental section all the deposi-
tions were conducted in a dc glow discharge apparatus with
the exception of two specimens, examined for comparison
purposes, which were generated in a magnetron sputtering
system using argon as the sputtering gas and a heated sub-
strate ~200 °C! of glass precoated with a thin chromium film
to improve adherence. These two specimens had a pro-
nounced @111# texture and were SERS inactive. These speci-
mens have the characteristics of evaporated gold films20 and
were quite smooth.
D. Sputtering targets
Sputtering of gold in an argon or nitrogen plasma trans-
ports the pure metal to the substrate; on the other hand,
argon–oxygen ~argon in the mixture prevents excessive
buildup of an insulating oxide layer on the target! produces
stoichiometric films of gold oxide, Au2O3 , which can be con-
verted into elemental gold by heating >350 °C or by chemi-
cal means through exposure to methanol or moist air.15 Sput-
tering of the Au/Si alloys in nitrogen generates a gold
dispersion in silicon nitride while use of argon–oxygen
plasma produces a dispersion of gold oxide in silica. The
silica or nitride matrix are amorphous and readily detected
by their infrared spectra. The combination of targets and
plasma compositions: Au–N2 , Au–Ar, Au–Ar O2 , and Au/
Si–Ar O2 produced some SERS active films, while
Au/Si–N2 did not. Those that were not active provided some
clues as to the significant parameters affecting SERS activ-
ity. Considered in this context were crystallite size as derived
J. Vac. Sci. Technol. A, Vol. 15, No. 2, Mar/Apr 1997
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 131.193.242.67 On: Sun, 30 Nov 2014 08:35:00
240
by XRD and microstructure as established by AFM, to be
discussed in the following sections. Argon sputtering was
only performed with the pure gold target.
E. Crystallite size
A preliminary correlation emerged which indicated that
all SERS active films, irrespective of origin, had average
crystallite size between 10–25 nm. On the other hand, films
with smaller crystallite size, typically gold nanocomposites
in silicon nitride, were inactive. These films did not show
any crystallite growth even after annealing, evidently be-
cause the matrix surrounding the crystallites prevents crystal
growth; on the other hand, annealing treatments of pure gold
films even for brief periods, about 30 s at 400 °C were suf-
ficient to cause growth to an average crystallite size larger
than 30 nm.
F. Microstructure AFM
A selected number of specimens were subjected to micro-
scopic analysis. Micrographs of surfaces representative of
SERS active and inactive response are given in Figs. 3 and 4.
The results of the microscopy work are collected in Table I
in which a measure of roughness is given.
The surfaces are typical of sputtered films, showing bot-
ryoidal morphology, the difference between the films in Figs.
3 and 4 is the roughness of the surface which evidently is the
determining factor controlling the presence of SERS activity.
It appears that the critical limit for the appearance of SERS
takes place when the film roughness reaches a value of about
16 nm. Figure 3 shows 3D images of two SERS active sur-
 241 Maya, Vallet, and Lee: Sputtered gold films for SERS 241

FIG. 4. Micrograph of SERS inactive films with corresponding section profiles. Note the smaller height scale on lower images.

faces and the corresponding surface profiles. The images are
somewhat different but the relevant characteristics are illus-
trated by the profiles that shows protruding areas with
heights of about 30–40 nm above the zero plane. The sepa-
ration between protruding aggregates is similar to their width
and relatively small compared to the wavelength of the inci-
dent light, a factor significant to SERS activity.7 The films
shown in Fig. 4 are smooth over large areas with variation in
heights within 10 nm except deep cracks.
The correlation between roughness parameters and SERS
activity observed here apply to these sputtered gold films and
the experimental conditions used such as exciting light fre-
quency. Most likely a different set of parameters would ap-
ply to silver films but the significant fact is the detection of a
critical roughness for the appearance of SERS.
Additional proof that the roughness of the film is the criti-
cal parameter determining whether SERS activity occurs was
derived from a simple experiment, whereby a SERS active
film ~that in Fig. 3! was removed cleanly from the single
crystal silicon surface by means of adhesive cellulose tape.
The back side of the film, initially in contact with the silicon
~identified in the table as film H! was exposed and was found
to be very smooth ~roughness of 0.8 nm!. This surface
proved to be SERS inactive. Additional checks following the
same procedure on different films confirmed these observa-
tions. Given the fact that XRD is not surface sensitive and
the x rays penetrate to depths comparable to the thickness of
the films, then the crystallite size derived from the XRD is
just an indication of SERS activity since very smooth films
may have appropriate crystallite sizes but be inactive.
G. General considerations
The fact that sputtered films are rougher than evaporated
films was established in a study of vacuum deposited gold
films.20 In that case the roughness was established by elec-
trochemical means by comparison of the current required to
reduce the gold oxide generated in a given macroscopic geo-

TABLE I. Microstructural analysis of selected samples.

Sample A B C D E F G H
a b a a
Target Au Au/Si Au/Si Au Au/Si Au Au/Si Au
Plasma N2 Ar/O2 Ar/O2 Ar/O2 N2 Ar Ar/O2 N2
Film Au Au/SiO2c Au/SiO2c Au Au/Si3N4 Au Au/SiO2c Au
SERS Yes Yes Yes No No No No No
Roughness 23.0 16.6 16.3 13.1 11.0 10.0d 6.5 0.8
R a ~nm!
a
Target containing 69 at % Au.
b
Target containing 5 at % Au.
c
From intermediate containing Au2O3 .
d
Data obtained with different microscope tip.

JVST A - Vacuum, Surfaces, and Films

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 131.193.242.67 On: Sun, 30 Nov 2014 08:35:00
 242 Maya, Vallet, and Lee: Sputtered gold films for SERS 242

metric area; thus, a smoother surface such as an evaporated roughness parameters necessary to produce a SERS active
or annealed film would require a smaller integrated current. surface.
That observation correlates well with our findings as derived
by AFM. The fact that sputtered films are rougher can be
turned into an advantage as for the generation of SERS ac- ACKNOWLEDGMENTS
tive surfaces. The emphasis of previous studies19,20 was ac- This research was sponsored by the Division of Materials
tually the opposite whereby smooth surfaces were desired. Sciences, Office of Basic Research, U. S. Department of En-
Many specimens were prepared in the course of the ergy, under Contract No. DE-AC05-96OR22464 with
present study, choosing a set of conditions, i.e., plasma and Lockheed–Martin Energy Research Inc. The authors are
sputtering target would be dictated by the end use, whether grateful to J. C. Fister for providing the gold films on chro-
ruggedness or simplicity in preparation are required. All sur- mium precoated glass and to J. P. Young for helpful discus-
faces could be rinsed and used repeatedly but those consist- sions.
ing of a gold nanocomposite in silica were superior in being
1
quite adherent to the different substrates, as established with M. Fleischmann, P. J. Hendra, and A. J. McQuillan, Chem. Phys. Lett. 26,
the simple test of placing on the film and pulling a strip of 163 ~1974!.
2
D. L. Jeanmaire and R. P. Van Duyne, J. Electroanal. Chem. 84, 1 ~1977!.
cellulose adhesive tape; thus use of a gold oxide intermediate 3
M. G. Albrecht and J. A. Creighton, J. Am. Chem. Soc. 99, 5215 ~1977!.
step seems justified in gaining a rugged surface and the flex- 4
J. J. Laserma, Anal. Chim. Acta 283, 607 ~1993!.
5
ibility of conversion to gold either by thermal or chemical A. Otto, I. Mrozek, H. Grabhorn, and W. Akemann, J. Phys. Condensed
means; if the application is not demanding, simple sputtering Matter 4, 1143 ~1992!.
6
T. Vo-Dinh, Sensors Actuators B 29, 183 ~1995!.
of pure gold in either nitrogen or argon will produce a useful 7
R. G. Freeman, K. C. Grabar, K. J. Allison, R. M. Bright, J. A. Davis, A.
film. A rugged surface would be particularly useful for mul- P. Guthrie, M. B. Hommer, M. A. Jackson, P. C. Smith, D. G. Walter, and
tiple uses in harsh environments as would be required for M. J. Natan, Science 267, 1629 ~1995!.
8
sensors installed to monitor chemical processes or in radio- K. C. Grabar, R. G. Freeman, M. B. Hommer, and M. J. Natan, Anal.
Chem. 67, 735 ~1995!.
logical areas. 9
G. Bar, S. Rubin, R. W. Cutts, T. N. Taylor, and T. A. Zawodzinski, Jr.,
Langmuir 12, 1172 ~1996!.
10
A. Berthod, J. J. Laserna, and J. D. Winefordner, Appl. Spectrosc. 41,
IV. CONCLUSIONS 1137 ~1987!.
11
T. Vo-Dinh, M. Meier, and A. Wokaun, Anal. Chim. Acta 181, 139
Sputtered gold films can be readily prepared and used as ~1986!.
12
SERS active surfaces. Sputtering conditions should be such J. M. Bello, D. L. Stokes, and T. Vo-Dinh, Appl. Spectrosc. 43, 1325
~1989!.
that deposition takes place close to room temperature to pre- 13
G. Binning, H. Rohrer, C. Gerber, and E. Weibel, Phys Rev. Lett. 49, 57
serve the microstructure. X-ray diffraction provides a pre- ~1982!.
14
liminary indication of whether a film might be suitable for L. Maya, W. R. Allen, A. L. Glover, and J. C. Mabon, J. Vac. Sci.
SERS. Different combinations of plasma gas and substrate Technol. B 13, 361 ~1995!.
15
L. Maya, M. Paranthaman, T. Thundat, and M. L. Bauer, J. Vac. Sci.
composition might be used, but it is advantageous to use an Technol. B 14, 15 ~1996!.
argon–oxygen plasma and a target having the Au–Si eutectic 16
L. Maya, J. Vac. Sci. Technol. A 11, 604 ~1993!.
composition, 31 at % Si, to generate robust films showing
17
S. Dai, Y. H. Lee, and J. P. Young, Appl. Spectrosc. 50, 536 ~1996!.
18
A. K. Green and E. Bauer, J. Appl. Phys. 52, 5098 ~1981!.
good adherence to a variety of substrates and without the 19
Y. Golan, L. Margulis, S. Matlis, and I. Rubinstein, J. Electrochem. Soc.
requirement of an intermediate chromium film. Atomic force 142, 1629 ~1995!.
microscopy is a powerful tool to identify the critical set of 20
Y. Golan, L. Margulis, and I. Rubinstein, Surf. Sci. 264, 312 ~1992!.

J. Vac. Sci. Technol. A, Vol. 15, No. 2, Mar/Apr 1997

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 131.193.242.67 On: Sun, 30 Nov 2014 08:35:00