Femtosecond and nanosecond laser fabricated substrate
for surface-enhanced Raman scattering
Adam Hamdorf,1Matthew Olson,1Cheng-Hsiang Lin,1Lan Jiang,2Jun Zhou,3Hai Xiao,4and Hai-Lung Tsai1,*
1Department of Mechanical and Aerospace Engineering, Missouri University of Science and Technology, Rolla, Missouri 65409, USA
2School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, China
3Department of Mechanical Engineering, The Pennsylvania State University Erie, Pennsylvania 16563, USA
4Department of Electrical and Computer Engineering, Missouri University of Science and Technology, Rolla, Missouri 65409, USA
*Corresponding author: firstname.lastname@example.org
Received May 31, 2011; revised July 15, 2011; accepted July 15, 2011;
posted August 3, 2011 (Doc. ID 148525); published August 22, 2011
We report a simple and repeatable method for fabricating a large-area substrate for surface-enhanced Raman
scattering. The substrate was processed by three steps: (i) femtosecond (fs) laser micromachining and roughening,
(ii) thin-film coating, and (iii) nanosecond laser heating and melting. Numerous gold nanoparticles of various
sizes were created on the surface of the silicon substrate. The 3D micro-/nanostructures generated by the fs laser
provide greater surface areas with more nanoparticles leading to 2 orders of magnitude higher of the enhancement
factor than in the case of a flat substrate. Using an He–Ne laser with a 632:8nm excitation wavelength, the surface-
enhanced Raman scattering enhancement factor for Rhodamine 6G was measured up to 2 × 107.
Society of America
OCIS codes:300.6450, 240.6695.
© 2011 Optical
The phenomenon of surface-enhanced Raman scattering
(SERS) was first observed over 35 years ago [1,2]. Since
its discovery, SERS has interested researchers because
of its ability to identify specimen information [3,4] with
enough sensitivity to be capable of single molecular de-
tection . While SERS has shown enormous potential,
one of its intrinsic drawbacks is the limited materials
capable of providing high enhancement factors (EFs).
Substrates have primarily utilized the noble metals in the
format of nanoparticles such as silver and gold and oc-
casionally copper to achieve sufficient enhancements
due to their superior optical properties . The selection
of the material depends on the application: while silver
provides the highest EFs, gold has higher chemical
stability in air.
Because the EF of the signal depends on the sub-
strate’s ability to generate surface plasmon resonance
, much research has focused on different methods
for positioning nanoparticles in ways conducive to
generating strong signals. Creating patterned surface
structures with a femtosecond (fs) laser has yielded high
EFs when used in conjunction with methods such as
lithography  or chemical plating . Laser machining
in silver nitrate solutions  or on Ag-doped materials
 has also produced exceptional EFs due to the nanos-
tructures formed by the laser processing.
In this study, fabrication of the SERS substrate was
first carried out with an fs laser to generate micro- and
nanoscale surface features. All silicon wafers were
cleaned before and after fs laser machining in an ultra-
sonic bath with a 70% ethanol solution for 10 min
and were then rinsed with distilled water. The fs laser
(Legen-F, Coherent, Inc.) had a central wavelength, max-
imum repetition rate, and laser pulse width of 800nm,
1kHz, and 120fs, respectively. The output power for the
laser was reduced through the combination of a half-
wave plate and linear polarizer. The laser beam was
focused with an objective lens (Olympus UMPLFL 10×,
NA ¼ 0:3) onto the silicon substrate mounted on a
five-axis motion stage (Aerotech) with a resolution of
about 1μm. The fs laser spot size was approximately
Low-pressure air was applied to the substrate during
fabrication to minimize the silicon spatter redepositing
on the machined area. Trenches were created on the
silicon surface by using a 72μW laser power, a 1kHz re-
petition rate, and a scan speed of 5mm=min. Horizontal
line scans were 5mm long, and each was vertically
spaced 2μm from one another so there was a 50% overlap
between the rows due to the 4μm laser spot size. A total
of 30 line scans were made on the surface, resulting in
periodic trenches over a 5000μm × 60μm area on the
silicon surface, as shown in Fig. 1, which is consistent
with previously reported results . In Fig. 1, the width
of the “stripe” is approximately 406nm, the center-to-
center distance between two stripes is 571nm, the “gap”
between two strips is 165nm, and the depth of the trench
silicon substrate after fs laser machining.
Scanning electron microscope (SEM) image of the
September 1, 2011 / Vol. 36, No. 17 / OPTICS LETTERS3353
© 2011 Optical Society of America
A 50nm thick gold film was coated on the fs laser
machined substrate using a sputter coater (Desk V,
Denton Vacuum). The vacuum of the sputter coater was
brought down to 40mTorr before starting the argon sput-
tering gas. A low target current of 8mA was used to coat
the substrates for a total of 7 min. As shown in Fig. 2, the
gold films cannot cover the trenches, and they are not
smooth and continuous. This can be attributed to the
large trenches in the middle of the silicon bridges, which
is different than sputtering on a flat surface, where the
gold will distribute evenly and will have nanoscale
breaks and islands .
The final step in the substrate processing was laser
heating and melting of the gold film using a nanosecond
(ns) laser micromachining system that consisted of a
frequency-tripled Nd:YAG laser (AVIA 355-X, Coherent)
and a computer-controlled four-axis motion stage (Aero-
tech). The central wavelength, maximum repetition rate,
and pulse width of the ns laser were 355nm, 200kHz, and
30ns, respectively. If the ns laser intensity is too high, the
gold film can be ablated and removed. Likewise, if the ns
laser intensity is too low, the gold film may not be melted
to become nanoparticles. Hence, the ns laser energy
was adjusted until nanoparticles were generated with
the following laser parameters: 2000Hz of repetition
rate, 4mm=min of scanning speed, and 120mJ=cm2of
fluence. Note that it is not the purpose of this study to
optimize the laser parameters to generate the maximum
possible nanoparticles on the substrate.
Previous research has shown that a flat gold coating
will form gold nanoparticles when treated with an ns
laser . In contrast, as shown in Fig. 3, the defocused
ns laser has generated a wider range of nanoparticle
sizes, which range from 10 to 200nm. However, ap-
proximately more than 80% of the nanoparticles are in
the range of 20 to 50nm in diameter. The 3D micro-/
nanostructures provide larger surface areas and more
nanoparticles on the surface of the substrate. In addition,
the 3D structures may help with concentrating and trap-
ping the light on the nanoparticles, leading to higher EFs.
SERS was then tested on the different areas of the sub-
strate. All substrates were incubated in Rhodamine 6G
(R6G) solutions for 1h at room temperature. This pro-
vided a consistent distribution of the R6G on the entire
substrate surface. The SERS signal was measured with a
commercial Ramanscope (Jobin Yvon) using an He–Ne
laser with a 632:8nm wavelength as the excitation
source, and the measuring laser spot size was about
10μm. The grating and objective lens used were
600line=mm and 10× (NA ¼ 0:25), respectively. A con-
centration of 10−3M was used for the normal Raman
substrate and 10−6M for all other substrates. The accu-
mulation time and the intensity of the He–Ne laser were
varied. All SERS-active substrates used a 5s acquisition
time and an intensity of 1:7mW, except the fs laser
machined, gold-coated, ns laser heated area, which used
power of 1:7μW. The uncoated substrates used acquisi-
tion times of 3 and 4s for the machined and unmachined
areas, respectively, with laser power of 17mW.
Because the fs laser affected the topography of the
substrate, it also changed the Raman spectrum. The
SERS spectrum from sputtering on a flat surface 
or by ns laser melting a flat gold coating  has already
been discussed previously. A good-quality spectrum is
also obtained by gold coating and melting over an fs laser
machined area, as shown in Fig. 4(a). Distinct Raman
peaks were produced with narrow linewidths for low
excitation powers. Figure 4(b) shows the spectrum for
the fs laser-machined, gold-coated area, which produced
a comparable signal to Fig. 4(a) but used 1000 times the
excitation power. There were not visible Raman peaks
when acquiring the signal on the machined-only area,
as shown in Fig. 4(c).
The EF was calculated with the following equation:
EF ¼ISERS× NNR
where I and N represent the intensity and number of
molecules probed, respectively, and the subscripts SERS
and NR denote the acquisition from either the SERS-
active substrate or the normal Raman control substrate.
The SERS signal EFs are shown in Table 1. The columns
in Table 1 indicate the order and type of substrate pro-
cessing, where Au, NS, and FS indicate gold coating, ns
laser melting, and fs laser machining, respectively. Note
that in the cases of Au and Au þ NS, the normal Raman
laser machining and gold coating.
SEM image of the silicon substrate topography after fs
machining, gold coating, and ns laser melting.
SEM image of the silicon substrate after fs laser
3354OPTICS LETTERS / Vol. 36, No. 17 / September 1, 2011
and SERS substrates are flat bare silicon, while in the
cases of FS þ Au and FS þ Au þ NS, the normal Raman
and SERS substrates are machined by an fs laser. There-
fore, the surface roughness of the normal Raman and
SERS substrates is close to each other. It is reasonable
to assume that the measured molecules on normal
Raman and SERS are identical, and the EF can be
simplified to the ratio of intensity of SERS and normal
Raman. The EFs given in Table 1 are the averaged values
for five substrate samples, and each was measured at 10
different locations. The variations of the data from the
averaged values are estimated to be in the range of ?30%
for all cases. It has been observed before that just sputter
coating alone could produce SERS , and the results
agree with these findings. The same was true for the ns
melting; while it was known that this method could pro-
duce SERS signals , it was necessary to find the EFs
again in this Letter for comparison purposes. Coating
gold over the fs machined area produced the lowest EFs.
However, ns annealing the fs laser grid area achievedEFs
of up to 107, much higher than all of the other SERS-
In summary, the SERS EFs up to 107were achieved by
combining fs laser machining, gold coating, and ns laser
melting. The proposed process is simple and repeatable,
which can increase the EFs 2 orders of magnitude as
compared to the previous proposed method with a flat
substrate. The 3D micro-/nanostructures can increase
surface areas and the number of gold nanoparticles with
different sizes leading to the higher EFs.
This work was supported by the U.S. Department of
Energy (DOE) under contract DE-FE0001127 and by the
National Natural Science Foundation of China (NSFC)
under grants 90923039 and 50705009.
1. M. Fleischmann, P. J. Hendra, and A. J. McQuillan, Chem.
Phys. Lett. 26, 163 (1974).
2. D. J. Jeanmaire and R. P. Van Duyne, J. Electroanal. Chem.
84, 1 (1977).
3. H. Chon, C. Lim, S. M. Ha, Y. Ahn, E. K. Lee, S. I. Chang,
G. H. Seong, and J. Choo, Anal. Chem. 82, 5290 (2010).
4. Y. S. Huh, A. J. Chung, and D. Erickson, Microfluid.
Nanofluid. 6, 285 (2009).
5. S. Nie and S. R. Emory, Science 275, 1102 (1997).
6. M. Kerker, J. Opt. Soc. Am. B 2, 1327 (1985).
7. M. Moskovits, J. Raman Spectrosc. 36, 485 (2005).
8. E. Diebold, P. Peng, and E. Mazur, J. Am. Chem. Soc. 131,
9. Y. Han, X. Lan, T. Wei, H. Tsai, and H. Xiao, Appl. Phys. A
97, 721 (2009).
10. C. H. Lin, L. Jiang, Y. H. Chai, H. Xiao, S. J. Chen, and H. L.
Tsai, Opt. Express 17, 21581 (2009).
11. Z. Zhou, J. Xu, Y. Cheng, Z. Xu, K. Sugioka, and K.
Midorikawa, Jpn. J. Appl. Phys. 47, 189 (2008).
12. J. Bonse, S. Baudach, J. Kruger, W. Kautek, and M. Lenzner,
Appl. Phys. A 74, 19 (2002).
13. C. H. Lin, L. Jian, J. Zhou, H. Xiao, S. J. Chen, and H. L. Tsai,
Opt. Lett. 35, 941 (2010).
14. L. May, C. E. Vallet, and Y. H. Lee, J. Vac. Sci. Technol. A 15,
Table 1. Enhancement Factors for
ShiftAu Au, NSFS, Au
3:6E þ 04
4:7E þ 04
6:6E þ 04
6:4E þ 04
1:8E þ 05
2:3E þ 05
2:7E þ 05
2:5E þ 05
8:1E þ 03
1:4E þ 04
1:8E þ 04
2:2E þ 04
8:7E þ 06
1:6E þ 07
2:0E þ 07
1:9E þ 07
machined, gold-coated, ns laser treated area with 10−6M R6G,
1:7μW laser power, and 5s acquisition time; (b) fs laser-
machined and gold-coated substrate with 10−6
1:7mW laser power, and 5s acquisition time; and (c) fs laser-
machined silicon with 10−3M R6G, 17mW laser power, and 3s
(Color online) SERS spectrum for (a) fs laser-
September 1, 2011 / Vol. 36, No. 17 / OPTICS LETTERS3355