Content uploaded by Charles Bahr
Author content
All content in this area was uploaded by Charles Bahr on Mar 24, 2014
Content may be subject to copyright.
Near-field photoconductivity: Application to carrier transport in InGaAsP
quantum well lasers
S. K. Buratto, J. W. P. Hsu,a) E. Betzig, J. K. Trautman, R. B. Bylsma, C. C. Bahr,
and M. J. Cardillo
(AT&T Bell Laboratories, Murray Hill, New Jersey 07974
~Received 1 April 1994; accepted for publication 23 August 1994!
A new contrast method in near-field scanning optical microscopy in which the near-field probe is
used to excite photocurrent in a semiconductor sample is described and demonstrated. The use of
near-field optics results in an order-of-magnitude improvement in spot size and a fivefold
improvement in resolution over previous methods of photocurrent imaging. The application of this
near-field photoconductivity technique to a multiquantum well laser provides direct visualization of
carrier transport throughout the structure, yielding information on growth inhomogeneities, carrier
leakage and isolation, and the overall quality of p-njunctions. © 1994 American Institute of
Physics.
In this letter we describe and demonstrate a new method
for generating an image in near-field scanning optical mi-
croscopy ~NSOM!in which the near-field radiation is used to
excite photoconduction. The resolution of the near-field pho-
toconductivity ~NPC!technique, for this particular sample ~a
multiquantum well laser!, is at least 53better than in previ-
ous photoconductivity imaging as well as in electron-beam
induced current ~EBIC!imaging, which are both .1
m
m.1
Additional advantages of NPC over EBIC reside in the in-
herent flexibility of NPC. NPC affords the ability to tune the
energy of the exciting light around the absorption edge of the
sample and does not require a vacuum environment. The
NPC image yields a map of minority carrier transport with a
resolution of approximately 250 nm. The technique is appli-
cable to a wide variety of electronic and optoelectronic ma-
terials such as transistors, light emitting diodes, photodetec-
tors, and lasers, where understanding transport properties on
a nanometer scale is important.
Such applications have been made possible by the devel-
opment of a near-field optical fiber probe2which provides a
factor-of-104enhancement in throughput over earlier de-
signs. Since then several variations of the basic NSOM ex-
periment have been reported for super-resolution ~subdiffrac-
tion limited!microscopy in transmission or reflection
including: Faraday effect imaging,3localized spectroscopy,4,5
collection mode microscopy,6–8 and photolithography.4The
NPC technique represents a new imaging method using the
super-resolution of NSOM.
The combination of near-field excitation with photocon-
ductivity results in a tenfold decrease in excitation spot size
~l/8, l5632 nm!over previous photoconductivity
experiments,9for which the spot size was of the order of 1
m
m in diameter. When the subwavelength aperture is placed
within approximately one radius away from the surface,10 the
spot size will be determined by the aperture diameter and not
the wavelength of the exciting light. The resolution of the
NPC experiment is then determined by a convolution of this
excitation spot size with the absorption length of the near-
field radiation, and the subsequent diffusion of the excited
carriers from this excitation volume.
The NSOM apparatus used for NPC has been described
in detail elsewhere8and is only briefly described here. The
subwavelength aperture exists at the end of an aluminum
coated optical fiber tapered to a small point.2The probe tips
are fabricated in a commercial pipette puller using a CO2
laser as the heat source. The aluminum coating is applied via
thermal evaporation. The tip size can be varied over a wide
range ~,20 to .200 nm!but was kept at approximately 80
nm for our experiments. The tip was maintained in the near
field of the sample surface using shear force feedback.11 This
was accomplished by mounting the tip in a small piezotube
and oscillating the tube/tip combination at its resonance fre-
quency. A laser beam scattered by the dithering tip was then
synchronougly detected using a lock-in amplifier. The ampli-
tude and phase of the applied dither ~;2 nm peak to peak!
changed as the tip approached within 50 nm of the surface,
and the demodulated output of the lock-in served as the input
to the feedback circuit. For the experiments described here
the tip–surface gap was maintained at approximately 15 nm.
The shear force feedback signal yielded a topographic image
of the surface with approximately 20 nm lateral resolution
and was collected simultaneously with the NPC image.
The InGaAsP MQW laser is shown schematically in Fig.
1 and has also been described in detail elsewhere.12 The
MQW structure includes six 4 nm InGaAsP ~Eg50.84 eV!
wells and five 9 nm InGaAsP ~Eg51.04 eV!barriers. The
laser active region ~200 nm31
m
m!consists of this MQW
stack sandwiched between two 70 nm InGaAsP ~Eg51.04
eV!layers for further confinement of the laser mode. The
curved mesa which houses the active region is formed by
defining, etching, and regrowing p-InP and n-InP current
blocking layers. The effect of these current blocking layers is
to channel the injected carriers into the MQW for conversion
into laser light. NPC images of the active region and sur-
rounding mesa provide a map of the minority carrier trans-
port and are sensitive to growth uniformity, nonradiative de-
fects, and the position of the laser’s p-i-njunction relative to
the MQW, all of which are important to the device physics.
a!Present address: Department of Physics, University of Virginia, Charlottes-
ville, VA 22901.
2654 Appl. Phys. Lett. 65 (21), 21 November 1994 0003-6951/94/65(21)/2654/3/$6.00 © 1994 American Institute of Physics
Downloaded¬20¬Mar¬2007¬to¬128.111.125.82.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://apl.aip.org/apl/copyright.jsp
NPC experiments were performed with the excitation
perpendicular to the laser p-i-njunction with a HeNe laser
~l5632 nm!as the excitation source and a maximum power
of 2 mW coupled into the optical fiber probe. The probe
attenuation was measured to be approximately 106~2nW!by
collecting the probe throughput by far-field optics ~0.4 NA
microscope objective!. The photocurrent was collected at the
n-InP and p-InP contacts with the laser unbiased. The excit-
ing light was amplitude modulated at a frequency of 500 Hz
and the photocurrent was synchronously detected with a sec-
ond lock-in amplifier. Modulation was necessary to detect a
signal above the large background signal induced by scat-
tered light from the feedback laser ~also a HeNe!.
An NPC image of an InGaAsP laser is shown in Fig.
2~a!. The 632 nm ~1.95 eV!source is sufficient to excite
carriers in all InP alloys present in the device. Figure 2~b!
shows two one-dimensional traces through the image of Fig.
2~a!along the lines denoted by the arrows A and B in Fig.
2~a!. The two traces of Fig. 2~b!illustrate the of our NPC
experiment. This can be seen by the separation between the
two signals in trace B and by the rising edge ~on the n-InP
side!largest signal, both of which are approximately 250 nm.
The asymmetry of the largest peaks of the two traces in Fig.
2~b!and the difference in signal strengths will be explained
later in the text.
A variety of features are observed around the laser’s
mesa structure as seen in Fig. 2~a!. The p-njunctions formed
by the p-InP contact layer and the top n-InP blocking layers
~see Fig. 1!are clearly visible with a strong signal and are
separated by the 1
m
m mesa width. The signal here is close
to uniform on each side of the mesa, indicating essentially
uniform film growth as expected by the planar interface be-
tween the p-InP contact and the top n-InP blocking layer. The
p-njunctions formed by the bottom p-InP blocking layers
and the n-InP contact layer ~see Fig. 1!, however, are not as
well defined. The signal along this junction is nonuniform
and has a maximum current that is a factor of 5 smaller than
the top p-njunction. The nonuniformity of the photocurrent
along the junction can be explained by considering the shape
of the etched mesa. The curvature of the mesa results in a
surface with large steps and a high step density. Growth on
such a surface leads to both a large density of defects and a
FIG. 1. InGaAsP MQW laser. The active region is depicted in the expanded
region enclosed in the dashed line. The 4 nm InGaAsP ~Eg50.84 eV!quan-
tum wells are the solid regions and the 9 nm InGaAsP ~Eg51.04 eV!bar-
riers are unshaded. The active region is located at the top of the mesa
structure in the depletion region of a p-i-njunction and the mesa is sur-
rounded by alternating n-InP and p-InP blocking layers. The laser is
mounted with the p-InP region closest to the ceramic heat sink.
FIG. 2. NPC of InGaAsP MQW laser. ~a!shows the NPC image ~632 nm
excitation!on the left and the shear force image on the right. The grey scale
is linear and the laser active region is shown in each image. The large signal
on the upper right of the shear force image is the In solder. ~b!shows line
cuts through the NPC image in two places, across the top and bottom p-n
junctions of the blocking layers ~trace A!and across the laser p-i-njunction
~trace B!. The rising edge on the n-InP side of the large peak in trace A
illustrates the 250 nm resolution. The asymmetry of each peak is explained
in the text.
2655Appl. Phys. Lett., Vol. 65, No. 21, 21 November 1994 Buratto
et al.
Downloaded¬20¬Mar¬2007¬to¬128.111.125.82.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://apl.aip.org/apl/copyright.jsp
nonuniform distribution of dopants and these effects lead to a
poor quality p-njunction as seen in the image. The overall
decrease in signal can be explained in terms of the distance
between the junction and the contact ~the n-InP contact is
503farther from the active region than the p-InP contact!
despite the higher electron mobility ~
m
e54600 cm2/V s
whereas
m
h5150 cm2/V s at 300 K13!. The combined result
of these two points is an expected decrease of a factor of 2 in
signal from the top junction which is half of the observed
result. The remaining loss in signal is most likely due to the
poor quality of the bottom p-njunction which is evident in
the image. Note also the absence of signal in the center of the
blocking regions. This indicates that the blocking layers are
providing isolation of the carriers in this region from the
p-InP and n-InP contacts as designed. Electron-hole pairs
created within these layers are electrostatically trapped and
are unable to induce a current at the laser contacts.
The image in Fig. 2~a!also shows contrast in the vicinity
of the active region where the blocking layers converge as
well as within the active region itself @see trace B of Fig.
2~b!#. The photocurrent observed within the active region is
small relative to the other p-njunctions because a large frac-
tion of the excited carriers lose their excess momentum, be-
come trapped in the MQWs, and do not result in a current.
The signal in the active region is also asymmetric with the
right side of the active region exhibiting a steep gradient of
NPC signal. This is indicative of dopant diffusion into the
active region which provides a path for carrier leakage past
the MQW. Note also that there is a higher NPC signal on the
p-InP side of the active region than on the n-InP side @see the
plateau in trace B of Fig. 2~b!#. Minority carriers excited on
the p-InP side leak past the MQW active region and are
detected as a current. This result is supported by collection
mode NSOM8which showed strong electroluminescence in
the n-InP region of this same laser, a further indication of
carrier leakage.
The signal observed in NPC depends on several param-
eters, the excitation spot size, the absorption length, the car-
rier diffusion constant Dand the carrier lifetime
t
. A signal is
observed when the excited electron and hole are separated
such that they induce an internal electric field in the sample
which is sensed as a current. The source volume for the NPC
signal which results from the near-field excitation is deter-
mined by the aperture size, the penetration of the near-field
radiation into the crystal, and the probability that a photon
generates an electron-hole pair. For a probe with diameter a,
the illuminated area is of the order of a2. In near-field optics
it is the evanescent modes, which decay exponentially with
distance from the probe, that result in super-resolution.10
Thus the excitation volume which results in
super-resolution14 is of the order of a3. The total source vol-
ume for photoconductivity is then a combination of this ex-
citation volume and carrier diffusion. The spatial dependence
of the photocurrent Icc is given by1
Icc5I0exp~2x/L!,~1!
where L5(D
t
)1/2 is the carrier diffusion length, D5kT
m
/q
the carrier diffusion constant,
t
the carrier lifetime, and I0
the current observed from the center of the depletion region.
Assuming a quantum efficiency of 8% for the evanescent
modes15 and an intensity of the order of 10 nW yields a
carrier generation rate of the order of 1025/cm3s. Using this
rate it is possible to estimate the carrier lifetime from previ-
ous photoconductivity ~far-field excitation!experiments on
InP16 to be of the order of 10 ps. This estimate of the carrier
lifetime implies a diffusion length of 350 nm for electrons
~De'120 cm2/s!and 70 nm for holes ~Dh'4cm
2
/s!. Thus
the expected resolution in p-InP is 430 nm (Le1a) and the
resolution in n-InP is 150 nm (Lh1a). This difference in
carrier diffusion for n-type and p-type material can be readily
observed in Fig. 2~b!by the asymmetry in the peaks. On the
nside of a p-njunction the NPC signal rises faster than on
the pside. The expected resolution of 150 nm on the n-InP
side is 30% higher than the observed resolution of 250 nm.
This discrepancy could be due to a larger penetration depth
of the evanescent modes or slower carrier relaxation than
postulated, both of which lead to a higher Lh. In any case the
resolution is limited by carrier diffusion.
In summary, NPC applied to InGaAsP MQW lasers re-
sults in important information concerning p-njunctions, de-
fects, and growth inhomogeneities. The skeleton of the mesa
structure is clearly visible by this technique, not only near
the active region but also around the blocking layers. Con-
trast along the bottom p-njunction of the blocking layers
indicates a preponderance of defects, especially along the
portion of highest curvature. The degree of carrier isolation
provided by these blocking layers can also be mapped by
NPC. The NPC results provide valuable insight into the de-
vice physics including carrier leakage paths and nonradiative
defects.
1T. Fuyuki, J. Phys. D 13, 1503 ~1980!; H. J. Leamy, J. Appl. Phys. 53, R51
~1982!.
2E. Betzig, J. K. Trautman, T. D. Harris,, J. S. Weiner, and R. L. Kostelak,
Science 251, 1468 ~1991!.
3E. Betzig, J. K. Trautman, R. Wolfe, E. M. Gyorgy, P. L. Finn, M. H.
Kryder, and C.-H. Chang, Appl. Phys. Lett. 61, 142 ~1992!.
4E. Betzig and J. K. Trautman, Science 257, 189 ~1992!.
5R. D. Grober, T. D. Harris, J. K. Trautman, E. Betzig, W. Wegscheider, L.
Pfeiffer, and K. West, Appl. Phys. Lett. 64, 1421 ~1994!.
6E. Betzig, M. Isaacson, and A. Lewis, Appl. Phys. Lett. 51, 2088 ~1987!.
7M. Isaacson, J. A. Cline, and H. Barshatzky, J. Vac. Sci. Technol. B 9,
3103 ~1991!.
8S. K. Buratto, J. W. P. Hsu, J. K. Trautman, E. Betzig, R. B. Bylsma, C. C.
Bahr, and M. J. Cardillo, J. Appl. Phys. ~in press!.
9For a review of spatially resolved photoconductivity, see C. J. R. Shep-
pard, Scanning Microsc. 3,15~1989!.
10 U. Durig, D. W. Pohl, and F. Rohner, J. Appl. Phys. 59, 3318 ~1986!;E.
Betzig, A. Harootunian, A. Lewis, and M. Isaacson, Appl. Opt. 25, 1890
~1986!.
11 E. Betzig, P. L. Finn, and J. S. Weiner, Appl. Phys. Lett. 60, 2484 ~1992!.
12 J. E. Geusic, R. L. Hartman, U. Koren, W.-T. Tsang, and D. P. Wilt, AT&T
Tech. J. 71,75~1992!.
13 S. M. Sze, Physics of Semiconductor Devices, 2nd ed. ~Wiley, New York,
1981!.
14 It can be shown that although there is a contribution to the photoconduc-
tivity from propagating modes, excitation by these modes produces a con-
stant background that does not vary with the position of the tip. A more
detailed analysis of the NPC signal and resolution will be published in a
longer paper.
15 Using P5P0exp~2
a
z!where P/P0is the fraction of the initial light
power absorbed,
a
is the absorption coefficient of InP ~
a
51
m
m!, and zis
the propagation into the crystal, which yields P/P0'0.08 for z580 nm.
16 P. S. Weiss, P. L. Trevor, and M. J. Cardillo, J. Chem. Phys. 90, 5146
~1989!.
2656 Appl. Phys. Lett., Vol. 65, No. 21, 21 November 1994 Buratto
et al.
Downloaded¬20¬Mar¬2007¬to¬128.111.125.82.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://apl.aip.org/apl/copyright.jsp