Content uploaded by Eric Borguet
Author content
All content in this area was uploaded by Eric Borguet on Nov 14, 2014
Content may be subject to copyright.
Infrared second harmonic generation spectroscopy of Ge„111…interfaces
D. Bodlaki,a) E. Freysz,b) and E. Borguetc)
Department of Chemistry and Surface Science Center, University of Pittsburgh,
Pittsburgh, Pennsylvania 15260
共Received 9 December 2002; accepted 9 April 2003兲
Infrared second harmonic generation 共IR-SHG兲spectroscopy, an extension of spectroscopic SHG to
the IR, is described and applied to the investigation of germanium–dielectric interfaces in the
spectral region near the direct and indirect band gap of the bulk semiconductor. The spectrum of the
Ge(111)– GeO2interface, in the 1100–2000 nm fundamental wavelength range, is dominated by a
resonance at 590 nm. This feature is assigned to the direct ⌫25⬎⌫2transition between valence and
conduction band states. Polarization and azimuth dependent IR-SHG spectroscopy revealed that the
anisotropic contribution, containing bulk quadrupole,
, and surface,
11 , nonlinear susceptibility
terms, dominates the 590 nm resonance. S-termination of Ge共111兲significantly modifies the
interface nonlinear optical response. The IR-SHG spectrum of S–Ge共111兲presents a new, possibly
surface resonance at ⬃565 nm, in addition to the resonance inherent to the bulk Ge at 590 nm,
tentatively assigned to an interband transition of Ge atoms associated with the surface. © 2003
American Institute of Physics. 关DOI: 10.1063/1.1578619兴
INTRODUCTION
The first semiconductor transistor was based on
germanium.1The initial preference for germanium was due
to its superior carrier transport properties—the mobility of
holes and electrons in Ge are more than twice those of Si.2
Silicon, nevertheless, ultimately prevailed because the native
oxide of germanium, GeO2, does not form a stable interface
with the semiconductor and is characterized by a higher trap/
surface state density.3Interest in Ge has renewed because of
its use in high speed bipolar transistors and the ease with
which it can be integrated with Si-based devices to fabricate
emitters, modulators and receivers for optical communica-
tions, e.g., Ge-on-Si near-IR photodetectors.4,5
Of the many spectroscopic techniques used to study
chemically modified semiconductor surfaces, SHG studies
are of particular interest since they provide surface sensitiv-
ity, probe the electronic response of the interface and can be
carried out in situ.6–8 SHG can probe buried interfaces that
are difficult to probe by conventional spectroscopic
methods.9The applicability of SHG to monitor charge,
strain, microroughness as well as the progress of chemical
reactions on semiconductor interfaces has been demon-
strated.10–15 Germanium was one of the first semiconductors
to be investigated by surface SHG.16–19 While the earliest
experiments,16 did not report anisotropy in the SHG response
with respect to rotation around the azimuthal axis 共SHG-
RA兲, later work revealed the presence of strong anisotropy.17
Si and Si1⫺xGexsurfaces have been extensively studied
by SHG spectroscopy and rotational anisotropy but less
attention was paid to germanium interfaces.6,12,20–23 Re-
cently, Ohashi et al. reported the SHG spectrum of Ge–oxide
interface in the 1.1–1.6 eV region and a surface resonance at
1.16 eV.24
A detailed phenomenological theory of SHG-RAwas de-
veloped and provides a framework for our
understanding.25–28 The SHG response in the pin /pout polar-
ization combination is described by
Ipp共2
兲
共Ip共
兲兲2⬃
兩
App⫹Bpp cos共3
兲
兩
2
⫽
兩
App
兩
2⫹
兩
Bpp
兩
2cos2共3
兲
⫹2
兩
App
兩
*
兩
Bpp
兩
cos共3
兲cos共⌬AB兲,共1兲
where
is the azimuthal angle measured between the plane
of incidence and the 关21
¯
1
¯
兴direction of the single crystal.29
App , the isotropic contribution and Bpp , the anisotropic con-
tribution are specific to the pin /pout polarization combina-
tion. They depend on the nonlinear coefficients, angle of in-
cidence and the linear optical properties of the interface at
the fundamental and second harmonic wavelengths.29 App
and Bpp are complex, having a relative phase ⌬AB
App
Bpp
⫽
冏
App
Bpp
冏
exp共i⌬AB兲.共2兲
App and Bpp are related to the microscopic nonlinear suscep-
tibility elements by
App⫽Ap
冋
a1pp
⫹a2pp
冉
␥
共2
兲⫹
31
冊
⫹a3pp共
31⫺
33兲⫹a4pp
15
册
,共3兲
Bpp⫽Ap关b1pp
⫹b2pp
11兴,共4兲
a兲Present address: Department of Chemistry, University of Wisconsin, Madi-
son.
b兲Permanent address: Center de Physique Mole
´culaire Optique et Hertzi-
enne, Universite
´de Bordeaux 1, 351 Cours de la Libe
´ration, 33405 Tal-
ence, Cedex, France.
c兲Author to whom correspondence should be addressed. Electronic mail:
borguet@pitt.edu
JOURNAL OF CHEMICAL PHYSICS VOLUME 119, NUMBER 7 15 AUGUST 2003
39580021-9606/2003/119(7)/3958/5/$20.00 © 2003 American Institute of Physics
Downloaded 06 Sep 2009 to 129.32.36.173. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/jcp/copyright.jsp
The nonlinear susceptibility elements
11 ,
15 ,
31 ,
33 de-
scribe the electric-dipole response of the surface.
11 contrib-
utes to the anisotropic response (Bpp).
15 ,
31 , and
33
determine the isotropic response (App) of the SHG signal.
App and Bpp have bulk electric-quadrupole contributions, de-
scribed by the bulk nonlinear susceptibilities
␥
and
. The a
and bcoefficients as well as Apin Eqs. 共3兲and 共4兲, depend
only on the linear dielectric parameters of the interface and
the angle of incidence.29 The s-polarized SHG signal has
only anisotropic contributions and is given by the following
expression:
Ips共2
兲
共Ip共
兲兲2⬃
兩
Bps sin共3
兲
兩
2.共5兲
SHG spectroscopy has been successfully applied to
semiconductor/oxide surfaces in the visible.21–23 However,
the region near the bandgap of Si and Ge, that lies in the IR,
has not been probed. IR⫹VIS SFG is the traditional ap-
proach to probing resonances in the IR spectral region. We
propose IR-SHG as an alternative. The IR-SHG approach
has the advantage of probing resonances that exist solely at
the fundamental or second harmonic photon energies. IR-
SHG overcomes several potential limitations of IR⫹VIS
SFG, namely the much larger nonlinear optical response that
can occur at SF or VIS wavelengths, e.g., for metal and
semiconductor interfaces. The nonresonant response can ob-
scure the smaller resonant response at IR wavelengths, as is
expected to be the case for infragap states when the sum
frequency and/or visible wavelengths are resonant with su-
pragap states in semiconductors. In addition, the intense vis-
ible photon source can perturb the interface response by
electron–hole pair generation, or induce an EFISH response
by multiphoton charging of the interfacial dielectric.30 For
example, recently, doubly resonant IR⫹VIS SFG, a tech-
nique that gives access to information about the electron–
vibration coupling of surface molecules was demonstrated
experimentally.31 While these phenomena may be of interest
themselves, it is also useful to study the interface response in
the absence of these couplings. The surface SHG response
has not been probed previously beyond 1530 nm, for a num-
ber of reasons: Lack of detectors with single photon counting
capabilities at IR wavelengths, and a lack of intense IR
sources.32
We have used our short pulse 共1to2ps兲narrowband
共⬍15 cm⫺1兲, tunable optical parametric amplifier 共OPA兲,33
coupled with the near-IR single photon counting capabilities
of Si CCD devices, to extend SHG spectroscopy to the IR.
The present work focuses on the influence of the chemical
state of the surface on the nonlinear optical spectroscopy of
Ge–GeO2and Ge–S interfaces. The 1100–2000 nm funda-
mental photon wavelength range explored includes the direct
and indirect gaps of germanium.
EXPERIMENT
Sample preparation
Ge共111兲wafers 共undoped, Eagle Picher兲were degreased
by successive 10-minute sonications in trichloroethylene
共J.T. Baker reagent grade兲, acetone 共EM Science, reagent
grade兲, then methanol 共Fisher Scientific, Certified ACS
grade兲. No additional treatment was performed on oxidized
samples before experiments. All chemicals were purchased
from Aldrich Co., unless otherwise stated, and used as re-
ceived. The clean, oxidized Ge samples were hydrogen ter-
minated by dipping in 48% HF 共Mallinckrodt, reagent grade兲
five times for 10 s, each time followed by a 20 s rinse in
nanopure water. Finally, the sample was dried in N2gas.34
Sulfidation of the Ge共111兲surface was achieved by immer-
sion of hydrogen terminated Ge in (NH4)2Sat70°Cfor20
min followed by rinsing in methanol and drying by N2
flow.35,36
Infrared second harmonic generation spectroscopy
The setup for IR-SHG spectroscopic studies is depicted
in Fig. 1. To cover the 1100–2000 nm range both signal and
idler wavelengths of the OPA were used.33 The OPA was
used in a 3 KTP crystal configuration to increase the wave-
length range and overall power of the output 共⬎40
J/pulse兲.
The IR output of the OPA was spectrally filtered to remove
second harmonic photons, and divided into a reference and
sample arm. The reference arm consisted of a ZnSe crystal as
source of second harmonic generation. ZnSe is transparent
over the wavelength range scanned. In the absence of an
optical resonance, any second harmonic intensity change
should only contain information about variation in laser
source parameters. The sample arm contained the sample
with appropriate focusing and collimating lenses before and
after the sample. The OPA output was strongly polarized and
could be changed from pto sat the sample by use of a
periscope. The spot size on the sample was about 10⫺4cm2.
The beam area was corrected for the elliptical shape obtained
at non-normal incidence. The second harmonic signal, re-
flected off the sample, was collected by the collimating lens,
and then focused onto the monochromator entrance slit 共Ac-
ton Research, 300 gr/mm grating兲. An analyzing polarizer
was set to pass either s-orp-polarized SHG photons, after
short pass filtering to block IR photons. The quadratic nature
FIG. 1. IR-SHG experimental setup. The output of OPA 共40
J, 1.1–3.3
m,1to2ps,1kHz兲is filtered 共signal-idler separator and F1兲and is
focused onto the sample 共L1兲. Second harmonic generated at sample is
collimated 共L2兲, polarized 共P兲, filtered 共F2兲and focused to monochromator
共L3兲. Signal is detected by CCD camera, in single element mode. ZnSe is
used as a reference.
3959J. Chem. Phys., Vol. 119, No. 7, 15 August 2003 Infrared second harmonic generation spectroscopy
Downloaded 06 Sep 2009 to 129.32.36.173. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/jcp/copyright.jsp
of second harmonic response was verified in the power range
explored.37 The signal was detected by a 1024⫻256 pixel,
liquid nitrogen cooled CCD camera 共Princeton Instrument,
CCD30-11兲. Selected regions of the CCD chip were defined
as detection areas for SHG signals from the sample and ref-
erence, and recorded simultaneously as the OPA was
scanned, in 5 nm/step increments, over the entire wavelength
range. The normalized SHG, i.e., the ratio of the sample and
reference signals, is reported. Consecutive scans were taken
and, if necessary, several scans were averaged. Typically,
scans were reproducible to within 5% and no averaging was
required. Depending on the sample signal levels, acquisition
times were 4–10 s per step. The spectral range is limited at
shorter wavelength (2
⫽560 nm) due to low OPA power
and consequently low SHG signal levels. The long wave-
length limit,
⬃2100 nm, is set by the CCD spectral re-
sponse. Second harmonic generation rotational anisotropy
共SHG-RA兲measurements were carried out to control the ex-
actness of the crystal cut and the effects of wet chemical
treatment.29
RESULTS AND DISCUSSION
The IR-SHG spectrum of Ge–GeO2interface in the
1100–2000 nm fundamental 共NIR兲wavelength range in the
pin /pout polarization is shown in Fig. 2 共filled circles兲. The
(2) in the pin /pout polarization combination contains all the
nonzero surface nonlinear susceptibility components 共i.e.,
31 ,
33 ,
15 ,
11 surface electric dipole components and
,
␥
bulk quadrupole components兲. At the 0° azimuth both App
and Bpp are active 关Eq. 共1兲兴. The characteristic feature of the
spectrum is the peak in SHG at a fundamental wavelength of
1180 nm 共590 nm SHG兲. For comparison, the spectrum of
the reference ZnSe sample, taken by replacing Ge with a
ZnSe in the sample arm, is plotted on Fig. 2 共solid line兲. The
spectrum of ZnSe, taken as baseline for the measurement, is
flat in scanned wavelength range, showing ⬍10% standard
deviation.
The second harmonic response can be estimated from
the bulk optical properties of germanium based on
共2
兲⫽共2
兲⫺1, 共6兲
where
共2
兲is the second-order susceptibility, 共2
兲is the
dielectric constant at the SHG frequency.12 The measured
SHG response in Fig. 2 共filled circles兲follows the predicted
trend 共open squares兲qualitatively, so we attributed the en-
hancement in the second harmonic signal to the direct ⌫25
⬎⌫2transition between the valence and conduction band
states.38,39 The indirect band gap of germanium is at 1878 nm
共0.66 eV兲, the direct band gap is at 1670 nm 共0.74 eV兲.40 No
strong resonances are observed at either the indirect (2
⫽939 nm) or direct band edges (2
⫽835 nm) gaps. In-
deed, the SHG response appears more sharply peaked than
might be expected from the linear optical properties alone.
The sum frequency generation spectrum, i.e., mixing of
the 800 nm and signal 共1100–1600 nm兲photons was taken to
test whether the resonance is at the second harmonic or at the
fundamental wavelength 共Fig. 3兲. The IR-SHG spectrum
shows the resonant feature at ⬃1180 nm. In contrast, the
SFG signal increases steadily with increasing fundamental
wavelength. The absence of the resonance at 1180 nm in the
SFG spectrum suggests that there is no resonance at the fun-
damental IR wavelength. Thus, a resonance at the SHG
wavelength 共590 nm, 2.1 eV兲must be responsible for the
peak in the SHG spectrum.
With the proper choice of the polarization of the incident
fundamental and the SHG photons and the azimuthal angle
of the Ge共111兲sample, the dispersion of the isotropic and
anisotropic contribution can be probed individually. The
pin /pout , and pin /sout rotational anisotropy patterns, Fig. 4,
show threefold symmetry with three small and three large
peaks separated by 120° with no isotropic offset above the
background. Rotating the sample to the
⫽30° azimuth in
the pin /pout polarization combination turns off the aniso-
tropic contribution Bpp and only App is probed 关Fig. 4共a兲兴.
The IR-SHG spectrum of the Ge(111)– GeO2interface
recorded in the pin /pout polarization combination at
⫽0°,
Fig. 2, probes both App , and Bpp .Bpp alone cannot be
probed in the pin /pout polarization combination but can be
FIG. 2. IR-SHG spectra of Ge(111) –GeO2共䊉兲and Ge共111兲-S 共䊏兲inter-
faces. The spectra were taken in the pin /pout polarization combination at
⫽0°. The second-order nonlinear response calculated from the linear
properties of germanium is also shown 共䊐兲. The dashed vertical lines indi-
cate the resonance at 590 nm. The solid vertical line indicates the resonance
at 565 nm.
FIG. 3. IR-SHG and SFG (IR⫹800 nm) spectra of Ge(111) –GeO2. The
IR-SHG spectrum 共䊉兲and SFG spectrum 共䉱兲were taken in the pin /pout
polarization combination at
⫽0°.
3960 J. Chem. Phys., Vol. 119, No. 7, 15 August 2003 Bodlaki, Freysz, and Borguet
Downloaded 06 Sep 2009 to 129.32.36.173. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/jcp/copyright.jsp
investigated in the pin /sout polarization combination 关Fig.
4共b兲兴. The anisotropic component, Bps , of the pin /sout polar-
ization combination contains the same, bulk quadrupole
,
and surface,
11 , nonlinear susceptibility terms as Bpp .29
The pin /sout polarization combination only comprises an an-
isotropic term. Therefore, at
⫽0° no pin /sout second har-
monic signal should be generated. It is clear that by a judi-
cious choice of polarizations and sample orientation, the
individual isotropic and anisotropic components can be
probed.
The IR-SHG spectrum of the Ge(111)– GeO2interface
was taken in the 1100–1600 nm range with an appropriate
choice of polarization combination so that both isotropic,
App , and anisotropic Bpp , only the isotropic, App , and only
the anisotropic, Bps , contributions were sampled, respec-
tively 共Fig. 5兲. There is a resonance at 590 nm in the spec-
trum of the anisotropic, Bps , contribution, which coincides
with the resonance observed when both the isotropic, App ,
and anisotropic, Bpp , components were probed. The reso-
nance at 590 nm was assigned to the direct ⌫25⬎⌫2transi-
tion between valence and conduction band states.39 The as-
sociation of the resonance with Bpp but not App , indicates
that the interband transition affects mainly the
and/or
11
nonlinear susceptibility components. In contrast, App shows
no peak but increases steadily as the wavelength decreases.
The spectrum of App is suggestive of a peak at shorter wave-
length outside our experimental range. The increase of App ,
while the SHG decreases when both App and Bpp , is probed
indicates that App and Bpp are out of phase at ⬍590 nm. This
is consistent with a resonance in one component Bpp , but
not the other App , at 590 nm.
The linear optical properties of Ge can be used to predict
possible SHG resonances at shorter wavelength 关Eq. 共6兲兴.22
The bulk spectrum of Si1⫺xGexis characterized by direct
interband transitions, E0
⬘/E1and E2in the 2–5 eV range.41 In
Si they occur at 3.3 and 4.2 eV, respectively.38,39 Daum et al.
recorded the SHG spectra of the Si(100)– SiO2interface in
the 3.2–4.4 eV 共388–282 nm兲range.22 The direct bulk inter-
band transitions, E0
⬘/E1and E2were at 3.3 and 4.2 eV were
identified.22 A strong transition at 3.6 eV was also found that
had no equivalent in the bulk of the crystalline Si. The 3.6
eV feature was attributed to the unique bonding configura-
tion of Si atoms at the interface.22 Such surface resonance
can be expected at the Ge(111)– GeO2interface between 2.1
and 4.2 eV. The peak at 4.2 eV does not disperse as the Ge
content increases, whereas the peak at 3.3 eV redshifts to
lower energy as the Ge content increases.41 For pure Ge it
lies at 2.1 eV.
Chemical termination, and the changes to the SHG re-
sponse, can be used as a basis for separating the contribu-
tions of the surface and the bulk.29 Surface functionalization
should change mainly the surface properties while leaving
the bulk response intact. Recently, a number of routes to air
stable termination/passivation layers of germanium have
been reported. Notably, wet chemical preparation of H-, Cl-,
S-, and alkyl-terminations of Ge共111兲and Ge共100兲have been
described.34–36,42–45 Combined SHG and XPS experiments
have shown that H- and Cl-terminated surfaces rapidly oxi-
dize in ambient, while S- and alkyl-terminations are stable
for weeks.46 Sulfidation of the Ge共111兲surface was used here
to test the surface character of the transition observed at 590
nm 共2.1 eV兲. The IR-SHG spectra of Ge(111) –GeO2and
Ge共111兲–S in the 1100–1600 nm fundamental wavelength
range are shown in Fig. 2. The resonance at 590 nm, attrib-
uted to the direct ⌫25⬎⌫2transition between valence and
conduction band states, is present in the Ge共111兲-S spectrum
as a shoulder.40 However, the dominant feature in the
Ge共111兲–S spectra is the peak at shorter wavelength, ⬃565
nm. This peak is absent in the IR-SHG spectra of the
Ge(111)– GeO2interface. Clearly, the resonance at the
FIG. 4. SHG-RA of Ge(111)– GeO2at 1380 nm fun-
damental wavelength. 共a兲SHG-RA in the pin /pout po-
larization combination. 共b兲SHG-RA in the pin /sout po-
larization combination. The isotropic and anisotropic
components probed at the various azimuths are shown
by arrows.
FIG. 5. IR-SHG spectrum of Ge(111)– GeO2in the 1100–1600 nm range.
Both App and Bpp components are probed in pin /pout polarization at
⫽0°
共䊉兲. The isotropic, App , component is probed in the pin /pout polarization
combination at
⫽30° 共䊐兲.Bps is probed in the pin /sout polarization com-
bination at
⫽30° 共䉱兲.At
⫽0° in the pin /sout polarization combination no
SHG was generated 共⫹兲.
3961J. Chem. Phys., Vol. 119, No. 7, 15 August 2003 Infrared second harmonic generation spectroscopy
Downloaded 06 Sep 2009 to 129.32.36.173. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/jcp/copyright.jsp
⬍565 nm is the result of the surface modification, i.e., sul-
fide termination.
The apparent peak at 565 nm is most likely of surface
origin especially since it is the result of the S-termination.
Therefore, we assign the peak at ⬃565 nm to an interband
transition of Ge atoms associated with the surface. The sur-
face origin of the peak at 565 nm is further confirmed by a
recent report of a resonance at 1070 nm 共1.16 eV兲fundamen-
tal wavelength.24 The resonance was present in the SHG
spectrum of the Ge共111兲-thermal oxide interface only but not
at the Ge共111兲-native oxide interface indicating its surface
origin.24 The close proximity of the resonance reported here,
i.e., 565 nm versus 535 nm SH wavelength, suggests that we
observed the same resonance but redshifted at the Ge共111兲–S
interface. Furthermore, Lyman et al. found that passivation
of Ge共001兲in (NH4)2S共24% in H2O) results in 2.4 ML S
coverage.36 The overlayer was described as a disordered
共glasslike兲GeSxlayer residing atop a partially ordered inter-
facial layer.36 The authors pointed out that their model is
reminiscent of the often-proposed picture of the intensely
studied Si–SiO2interface; a partially ordered interfacial
SiOxwhose order breaks down for SiO2layers more than
several angstrom thick.36 It is interesting to note that the
IR-SHG spectrum of the Ge共111兲–S interface is similar to
the SHG spectrum of Si(100)– SiO2interface as the reso-
nance of surface origin lies between the bulk interband tran-
sitions of Ge, i.e., between 2.1 and 4.2 eV.22
CONCLUSIONS
We have extended surface SHG spectroscopy into the
IR; infrared second harmonic generation 共IR-SHG兲, and ap-
plied it to the investigation of germanium–dielectric inter-
faces in the spectral region near the direct and indirect band
gaps of the bulk semiconductor. The spectrum of the
Ge(111)– GeO2interface, in the 1100–2000 nm fundamen-
tal wavelength range, is dominated by a resonance at 590
nm. This feature is assigned to the direct ⌫25⬎⌫2transition
between valence and conduction band states. Polarization
and azimuth dependent IR-SHG spectroscopy revealed that
the anisotropic contribution, containing bulk quadrupole,
,
and surface,
11 , nonlinear susceptibility terms, contributes
mostly to the 590 nm resonance. S-termination of Ge共111兲
significantly modifies the interface nonlinear optical re-
sponse. The IR-SHG spectrum of S–Ge共111兲presents a new,
possibly surface resonance, at ⬃565 nm in addition to the
resonance inherent to the bulk Ge at 590 nm. This resonance
is assigned to an interband transition associated with the sur-
face Ge atoms.
ACKNOWLEDGMENTS
The authors acknowledge the generous support of the
NSF 共CHE-9734273兲. E.B. acknowledges the NSF for a
CAREER award in support of this research.
1J. Bardeen and W. H. Brattain, Phys. Rev. 74, 230 共1948兲.
2C. Kittel, Introduction to Solid State Physics, 6th ed. 共Wiley, New York,
1986兲.
3P. Balk, in The Si –SiO2System, edited by P. Balk 共Elseiver, Amsterdam,
1988兲, Vol. 32, pp. 2–20.
4J. Oh, J. C. Campbell, S. G. Thomas, S. Bharatan, R. Thoma, C. Jasper,
R. E. Jones, and T. E. Zirkle, IEEE J. Quantum Electron. 38,1238共2002兲.
5E. Maruyama, S. Okamoto, A. Terakawa, W. Shinohara, M. Tanaka, and S.
Kiyama, Sol. Energy Mater. Sol. Cells 74,339共2002兲.
6H. W. K. Tom, T. F. Heinz, and Y. R. Shen, Phys. Rev. Lett. 51, 1983
共1983兲.
7Y. R. Shen, Nature 共London兲337, 519 共1989兲.
8T. F. Heinz, ‘‘Second-order nonlinear optical effects at surfaces and inter-
faces,’’ in Nonlinear Surface Electromagnetic Phenomena, edited by H. E.
Ponath and G. I. Stegeman 共Elsevier Science, New York, 1991兲,p.353.
9G. Lu
¨pke, Surf. Sci. Rep. 35,75共1999兲.
10J. I. Dadap, P. T. Wilson, M. H. Anderson, and M. C. Downer, Opt. Lett.
22,901共1997兲.
11 J. Fang and G. P. Li, Appl. Phys. Lett. 75,3506共1999兲.
12W. Daum, H. J. Krause, U. Reichel, and H. Ibach, Phys. Rev. Lett. 71,
1234 共1993兲.
13J. I. Dadap, B. Doris, Q. Deng, M. C. Downer, J. K. Lowell, and A. C.
Diebold, Appl. Phys. Lett. 64, 2139 共1994兲.
14S. A. Mitchell, R. Boukherroub, and S. Anderson, J. Phys. Chem. B 104,
7668 共2000兲.
15S. A. Mitchell, M. Mehendale, D. M. Villeneuve, and R. Boukherroub,
Surf. Sci. 488, 367 共2001兲.
16N. Bloembergen, R. K. Chang, S. S. Jha, and C. H. Lee, Phys. Rev. 174,
813 共1968兲.
17D. Guidotti, T. A. Driscoll, and H. J. Gerritsen, Solid State Commun. 46,
46 共1983兲.
18J. A. Litwin, J. E. Sipe, and H. M. Van Driel, Phys. Rev. B 31,5543
共1985兲.
19O. A. Aktsipetrov, I. M. Baranova, and Y. A. Il’inskii, Sov. Phys. JETP64,
167 共1986兲.
20P. R. Fisher, J. L. Daschbach, and G. L. Richmond, Chem. Phys. Lett. 218,
200 共1994兲.
21G. Erley, R. Butz, and W. Daum, Phys. Rev. B 59, 2915 共1999兲.
22G. Erley and W. Daum, Phys. Rev. B 58, R1734 共1998兲.
23P. S. Parkinson, D. Lim, R. Bu
¨ngener, J. G. Ekerdt, and M. C. Downer,
Appl. Phys. Lett. 68,641共1999兲.
24H. Ohashi, H. Sano, and G. Mizutani, Jpn. J. Appl. Phys. 40, 6972 共2001兲.
25O. A. Aktsipetrov, I. M. Baranova, and Y. A. Il’inskii, Zh. Eksp. Teor. Fiz.
91,287共1986兲.
26J. E. Sipe, D. L. Moss, and H. M. van Driel, Phys. Rev. B 35, 1129 共1987兲.
27V. Mizrahi, and J. E. Sipe, J. Opt. Soc. Am. B 5, 660 共1988兲.
28G. Lu
¨pke, D. J. Bottomley, and H. M. van Driel, J. Opt. Soc.Am. B 11,33
共1994兲.
29V. Fomenko, D. Bodlaki, C. Faler, and E. Borguet, J. Chem. Phys. 116,
6745 共2002兲.
30V. Fomenko, C. Hurth, T. Ye, and E. Borguet, J. Appl. Phys. 91, 4394
共2002兲.
31M. B. Raschke, M. Hayashi, S. H. Lin, and Y. R. Shen, Chem. Phys. Lett.
359, 367 共2002兲.
32E. K. L. Wong, K. A. Friedrich, and G. L. Richmond, Chem. Phys. Lett.
195, 628 共1992兲.
33D. Bodlaki and E. Borguet, Rev. Sci. Instrum. 71, 4050 共2000兲.
34T. Deegan and G. Hughes, Appl. Surf. Sci. 123,66共1998兲.
35G. W. Anderson, M. C. Hanf, P. R. Norton, Z. H. Lu, and M. J. Graham,
Appl. Phys. Lett. 66, 1123 共1995兲.
36P. F. Lyman, O. Sakata, D. L. Marasco, T. L. Lee, K. D. Breneman, D. T.
Keane, and M. J. Bedzyk, Surf. Sci. 462, L594 共2000兲.
37V. Fomenko, J. F. Lami, and E. Borguet, Phys. Rev. B 63, 121316 共2001兲.
38Handbook of Optical Constants of Solids, edited by E. D. Palik 共Aca-
demic, San Diego, 1998兲.
39F. Wooten, Optical Properties of Solids 共Academic, New York, 1972兲.
40S. M. Sze, Physics of Semiconductor Devices, 2nd ed. 共Wiley, New York,
1981兲.
41J. Humlicek, M. Garriga, M. I. Alonso, and M. Cardona, J. Appl. Phys. 65,
2827 共1989兲.
42Z. H. Lu, Appl. Phys. Lett. 68, 520 共1996兲.
43K. Choi and J. M. Buriak, Langmuir 16, 7737 共2000兲.
44S. M. Han, W. R. Ashurt, C. Carraro, and R. Maboudian, J. Am. Chem.
Soc. 123, 2422 共2001兲.
45J. L. He, Z. H. Lu, S. A. Mitchell, and D. D. M. Wayner, J. Am. Chem.
Soc. 120, 2660 共1998兲.
46D. Bodlaki, H. Yamamoto, D. H. Waldeck, and E. Borguet, Surf. Sci. 共to
be published兲.
3962 J. Chem. Phys., Vol. 119, No. 7, 15 August 2003 Bodlaki, Freysz, and Borguet
Downloaded 06 Sep 2009 to 129.32.36.173. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/jcp/copyright.jsp