Demonstration of near-field scanning photoreflectance spectroscopy
Charles Paulson and A. B. Ellis
Department of Chemistry, University of Wisconsin, Madison, Wisconsin 53706
Leon McCaughan
Department of Electrical and Computer Engineering, University of Wisconsin, Madison, Wisconsin 53706
Brian Hawkins, Jingxi Sun, and T. F. Kuech
a)
Department of Chemical Engineering, University of Wisconsin, Madison, Wisconsin 53706
͑Received 10 April 2000; accepted for publication 26 July 2000͒
A near-field scanning optical microscope ͑NSOM͒ was developed to perform photoreflectance ͑PR͒
spectroscopy experiments at high spatial resolution ͑ϳ1
m͒. Representative PR spectra are shown,
along with an image illustrating the capability of observing contrast in images due to the strength of
a PR feature. It was found that sufficiently high intensity light from the NSOM tip can produce
photovoltages large enough to limit the spatial resolution of the electric field determination by PR.
The photovoltage effect is measured as a function of light intensity, and the results are discussed in
terms of a simple photovoltage expression. © 2000 American Institute of Physics.
͓S0003-6951͑00͒01839-8͔
The photoreflectance ͑PR͒ spectroscopy technique is a
form of electromodulation spectroscopy used to determine
the electronic properties at the surface of direct band gap
semiconductors, e.g., surface electric field in the depletion
region, the surface Fermi level E
F
, doping density, the band
gap energy E
g
, and other critical point energies.
1
The semi-
conductor is modulated at a frequency
, with an above-band
gap light source, the pump. A tunable light source serves as
the probe and is reflected from the modulated area of the
sample’s surface. The reflected probe light contains both a dc
and an ac component, with the ac component having the
frequency of the pump modulation. The modulated reflectiv-
ity, normalized by the dc reflectivity, is the PR signal. The
magnitude of ⌬R/R is typically 10
Ϫ2
–10
Ϫ4
and depends on
the wavelength of the probe light. As the probe wavelength
is scanned, some samples will exhibit Franz–Keldysh oscil-
lations ͑FKOs͒ in the ⌬R/R spectra due to modulation of the
internal electric field.
2
The extrema in the FKOs are given by
n
ϭ
4
3
ͫ
E
n
ϪE
g
ប
ͬ
3/2
ϩ
͑1͒
and are related to the magnitude of the electric field F by
ប
ϭ
ͩ
e
2
ប
2
F
2
2
ͪ
1/3
. ͑2͒
Here, (ប
)
3/2
is the electrooptic energy, n is the index of the
nth extreme of the oscillations, E
n
is the energy of the probe
photon, E
g
is the gap energy,
is an arbitrary phase factor,
and
is the reduced mass of an electron in the sample. F can
be determined from a plot of (4/3
)(E
n
ϪE
g
)
3/2
versus in-
dex number n, and should be a straight line with slope equal
to (ប
)
3/2
, which is proportional to F.
We have developed and demonstrated a near-field scan-
ning PR measurement technique based on near-field scan-
ning optical microscopy ͑NSOM͒.
3
The tapered, aluminum-
coated tip used here had an aperture diameter of ϳ100 nm,
and was held at a constant distance of ϳ10–40 nm from the
surface, permitting the simultaneous measurement of surface
electronic properties and surface topography when the tip is
scanned. The optical signals can be gathered with a spatial
resolution that is nominally limited by the aperture size.
Typically, only 1–10 mW of light is launched down the fi-
ber, since higher power levels are apt to damage the aperture
by heating and thus degrade the resolution. A drawback to
NSOM-based measurements is that only 10
Ϫ6
–10
Ϫ4
of the
incident light passes through the small aperture, requiring
that the optical detection system be sensitive to signal levels
in the 1–1000 nW ͑and smaller͒ range for reflection mea-
surements made with unity collection efficiency.
The photoreflectance NSOM ͑PRNSOM͒ system uses a
chopped HeNe laser ͑632.8 nm͒ for the pump and a cw tita-
nium sapphire ͑TiS͒ laser ͑that is tuned from 770 to 905 nm͒
for the probe in this experiment. The two laser beams are
combined using a beam splitter and coupled into a single-
mode fiber optic, ensuring that the pump and probe beams
have strong spatial overlap at the NSOM tip.
The light was reflected at near-normal angles to the
sample surface, and a high collection efficiency was realized
using a large ͑7.5 cm diam͒, fast (f/No.ϭ1) lens placed on
axis with the fiber, one focal length away from the tip–
sample interface. A long-pass filter was placed in front of the
detector to remove the HeNe light from the reflected signal.
Detection was accomplished with a large ͑1cm
2
͒ unbiased
silicon photodiode and a precision current amplifier followed
by a lock-in amplifier. The probe laser intensity was con-
trolled to produce a constant power output which aids in the
normalization of the PR signal.
4
Both the photoluminescence
͑PL͒ and the PR signals are passed by the filter and detected.
The PL signal generated by the pump laser and detected by
the lock-in amplifier was independently measured by block-
ing the probe beam. The magnitude of the PR signal was
then corrected by subtracting the pump-induced PL back-
a͒
Author to whom correspondence should be addressed; electronic mail:
kuech@engr.wisc.edu
APPLIED PHYSICS LETTERS VOLUME 77, NUMBER 13 25 SEPTEMBER 2000
19430003-6951/2000/77(13)/1943/3/$17.00 © 2000 American Institute of Physics