RESEARCH ARTICLE | JANUARY 01 1992

Optoelectronic measurement of semiconductor surfaces and

interfaces with femtosecond optics 
X.‐C. Zhang; D. H. Auston

J. Appl. Phys. 71, 326–338 (1992)

[Link]


View Export
Online Citation

22 June 2025 [Link]

Optoelectronic measurement of semiconductor surfaces and interfaces
with femtosecond optics
X.-C. Zhanga) and D. H. Auston
Electrical Engineering Department and Center for Telecommunications Research, Columbia University,
New York, New York 10027
(Received 13 June 1991; accepted for publication 19 September 1991)
The basic concepts and preliminary applications of optically induced electromagnetic
radiation from semiconductor surfaces and interfaces by using femtosecond optics are
discussed. This submillimeter-wave radiation provides a novel optoelectronic
technique to study semiconductor electronic surface and interface properties with a contactless
approach. The amplitude and phase of the electromagnetic radiation from the
semiconductor surfaces depend on carrier mobility, impurity doping concentration, and
strength and polarity of the static internal field. A large selection of bulk, epitaxial layer and
superlattice samples from III-V, II-VI and group-IV semiconductors has been tested.
The orientation and strength of the static built-in fields of a wide range of semiconductor
surfaces, such as surface depletion, metal/semiconductor Schottky, p-n junction and
strain-induced piezoelectric fields, can be determined and estimated.

I. INTRODUCTION strength and orientation of piezoelectric fields in strained-

layer superlattices have been probed by the optically in-
Optical and optoelectronic techniques, including Ra- duced submillimeter radiation.4
man scattering, photoluminesence, photoemission, absorp- The static built-in field associated with the semicon-
tion, and electroreflectance, have been widely used to char- ductor surfaces or interfaces provides a natural bias field
acterize semiconductor surfaces and interfaces.’ The recent for the photocurrent to producg the radiation. Unlike the
observation of the generation of a femtosecond electromag- bias field parallel to the surface, the natural bias fields, such
netic beam from a semiconductor surface provides a new as the surface depletion field, p-n junction field, or strain-

22 June 2025 [Link]

technique to characterize electronic properties of semicon- induced piezoelectric field, are normal to the surface.
ductor surfaces and interfaces.2 By analysis of the ampli- Therefore, the analysis of the radiation from the semicon-
tude and phase of the electromagnetic radiation from the ductor surface field with the field normal to the surface will
semiconductor surfaces, the transient carrier mobility, im- be different from that with the planar bias.6’7
purity doping concentration, strength, and polarity of the
static internal field (including surface depletion field, II. GENERATION AND DETECTION
Schottky field, p-n junction field, and strain-induced piezo-
electric field), and even crystal orientation can be deter- A. Radiation mechanism
mined and estimated.3 The use of the optically induced When a semiconductor surface or interface is illumi-
radiation is a promising way to measure the static field nated by an ultrafast laser pulse with the photon energy
distribution of the semiconductor. This technique provides greater than the’semiconductor band gap, photons are ab-
a noninvasive. method that can be used to measure the sorbed, creating electron-hole pairs. If there is a static elec-
semiconductor surface static field (including the piezoelec- tric field at the surface or interface, the free carriers driven
tric field and surface depletion field) and even to extend the by the static field accelerate along the field direction and
probing capability to multilayer application.4Z5 form a transient photocurrent. The rise time of the photo-
Like most optical and optoelectronic techniques, opti- current pulse is comparable to the optical pulse duration,
cally induced electromagnetic radiation uses a laser beam while the fall time is limited by the transit time of the
as a noncontact probe to characterize the semiconductor carriers across the field region (assuming the transit time is
surfaces and interfaces. These diffraction-limited electro- shorter than the carrier’s lifetime). The accelerated
magnetic beams can be optically steered by varying the charged particle or fast time-varying current radiates elec-
incident angle of the laser beam on the semiconductor, and tromagnetic waves. In the far field, the amplitude of the
they are easy to collect, collimate, focus, and detect. Co- radiation is proportional to the charge acceleration or the
herent generation and detection of the electromagnetic time derivative of the current. The radiation from the semi-
field provide both amplitude and phase information. For conductor surface has inward (transmitted direction) and
example, the radiated electromagnetic field provides infor- outward (pseudoreflected direction) components. The
mation about the impurity concentration, carrier mobility, tilted wave front of the optical beam determines the timing
surface built-in potential, and crystal symmetry. The of the photocurrent distribution. If the optically illumi-
nated area is comparable to or larger than the radiation
“Present address: Department of Physics, Rensselaer Polytechnic Insti- wavelength (approximately 0.5 mm), the electromagnetic
tute, Troy, NY 12180. pulse is directional and diffraction limited. The incident

326 J. Appl. Phys. 71 (I), 1 January 1992 @ 1992 American Institute of Physics 326
optical angle ooP, propagating direction of outward radi- where e is the electron charge, a is the absorption coeth-
ated field 8i, and inward radiated field e2 can be approxi- cient for the laser light, &X is incident photon energy in eV,
mately expressed by a generalized Fresnel law: R (e,,,) is the optical reflectivity at the surface with optical
incident angle coop,W is the depletion length, Ed(x) is the
n~(oop) sin 80P~:nl(w,l) sin &~=n~(w,l) sin e2, (1) built-in field, and a is the optical absorption length, p is the
where n2(o,[) is the index of refraction of radiated field of time-domain response function for the carrier mobility,
the sample; nl(w,P) and nl(03 are the indices of refrac- lop is the optical pulse energy per unit, and to is the carrier
tion of the optical beam and the radiated field outside of lifetime. Clearly there is no radiation along the static field
the sample, respectively. Since the difference between direction (8 = 0), and the amplitude of the radiation is
nr ( ooP) and nl (w,/) is negligible in the air, the direction of proportional to the first time derivative of the photocur-
the pseudoreflected (backward) electromagnetic radiation rent. Equation (6) does not include effects of the band
is near the specular angle of the laser beam, resulting in structure, therefore it is a good approximation under the
e,dl. moderate surface field.
The amplitude of the radiated field strongly depends
on the optical incident angle. For example, the angular-
dependent radiated field (forward) as a function of the
B. Dipole approximation
incident angle has the following form:
To study the optically induced electromagnetic radia-
E(e) &sin(e)[ 1 - R(e)]t(B). (71
tion from semiconductors, we start from the dipole approx-
imation. This is because when the carrierls velocity is much The first term comes from the radiation pattern of the
less than the velocity of light, the dipole approximation is dipole approximation, the second term from the photocar-
the simplest model with which to estimate the radiation rier density which is proportional to the optical absorption,
and gives a qualitative description for the radiation phe- and the third term from the transmission coefficient of the
nomena. The electric and magnetic components E and B electromagnetic wave. Equation (7) can be further ex-
from a charged particle in the expression of dipole radia- panded as
tion (SI units) can be written as
EA @/4%-K& 3, + (d47i-KC2f’ 3)rX (I%+‘), (2)

22 June 2025 [Link]

B=e(.irXr)/47rKc2r ‘, (3) 2 c0s(e) sin(&)
(8)
where r is the distance from the charged particle to the Xsin(B + e,) c0s(e - 02’
observation position, K is the dielectric constant, c is the where e2 = sin - ’ [sin(B)/nl] with n2 the index of refrac-
speed of the light, and + is the acceleration of the particle. tion of the radiated field. The maximum radiation direction
The first term in the E expression in Eq. (2) is a Coulomb is close to the Brewster angle, and its amplitude is propor-
field; it decreases as r z and can be ignored at far field. The tional to the optical power, the carrier mobility, and the
second term is the radiated field, and the radiation is a TM integral of the product of the built-in field and the photo-
wave. Using the equation of the current J= ev, where v is carriers.
the charge velocity, the radiation term in Eq. (2) can be
rewritten as
C. Radiation power
Eg= (sin W4mc2r).& (4)
For a moderate incident optical power density, the
where 8 is the angle between the observation direction and
screening of the surface field by the space-charge effect
particle acceleration direction. For a complete description from the photocarriers can be ignored. In this case the
of the whole radiation from a semiconductor surface, Eq. static field is independent of time and does not change even
(4) should be integrated over the optically illuminated with the optical excitation. As a result, the amplitude and
area s with an appropriate phase factor in the retarded
polarity of the radiated electromagnetic field in weak inci-
current Jr The radiation at far field (ignoring the angle 8 dent optical power is proportional to the strength and sign
variation across the optically iluminated area) can be writ- of the static built-in field.
ten as An approximate value of the electromagnetic radiation
sin 8 power from photoinduced carriers can be estimated from
(5) the following equation:
Ee=zs s 5 lr?r’j ds’y
and the angle-dependent photocurrent J, across the deple- P=Ni[ (e+)2/6mc3], (9)
tion layer can be written as where NP is the total number of photoinduced carriers and
the square of NP comes from the coherent radiation. As-
Js=g[ 1 - R(e,,) ] JeW E&)e-Wdx j-I m P(f - t’) suming 6 x 1012 carriers/pulse (2 pJ/pulse at a wave-
length of 620 nm) have been driven by the internal built-in
t’ field, and their velocity has accelerated from 0 to lo4 m/s
X e - (t’- f*)/f~~~~( t” ) &‘&” , (6)
-co within 1 ps, the electromagnetic radiation power calculated

327 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 327
 Reflected optical beam for the temporal measurement, passed through a variable
Transmitted field time-delay stage and was focused to approximately 5 pm at
the photoconductor in the dipole antenna. To achieve the
maximum amplitude of the radiation, the incident angle of
the laser beam was close to the semiconductor Brewster
angle (approximately 74” for InP).’ For the low-tempera-
ture experiment, the samples were placed in a dewar, so the
incident angle was reduced to 50” due to the diameter lim-
itation of the cryostat window. The average optical power
density was approximately 3 mW/cm” on the semiconduc-
Incident optical beam
tor surface. The outward radiated electromagnetic wave,
which was collinear with the reflected optical beam, was
FIG. 1. Electromagnetic beams generated from semiconductor surface. collimated by off-axis gold-coated parabolic mirrors and
The incident optical beam, outward radiated field, and inward radiated focused onto the photoconducting detector (dipole an-
fleld satisfy the generalized Fresnel law. tenna). The electromagnetic wave detector is a lOO+m
Hertzian dipole antenna fabricated on radiation-damaged
silicon on sapphire. Due to the slow response time (ap-
from the equation is 30 ,LLWwith K = 12.5. Clearly, Eq. (9) proximately 0.6 ps) of the lOO+m dipole antenna fabri-
does not include the carrier-carrier scattering, field screen- cated on the radiation-damaged silicon-on-sapphire photo-
ing by space-charge and other effects. conductor, the measured radiation pulses can be limited by
The radiation field is proportional to the number of the the bandwidth of the detector photoconductor, which has a
photocarriers, as shown in Eq. (9); but this is only valid peak frequency response of approximately 525 GHz. The
under the assumption of low optical excitation. When the measured signal from the detector was amplified, averaged,
incident optical power reaches a level at which the photo- and digitized by a current amplifier, a lock-in amplifier,
carrier density is comparable to the background doping and a computer, respectively. Time-dependent wave forms
density of the semiconductor, a modification of Eq. (9) is of the radiated electric field were recorded by varying the
required.8 However, the upper limit of the surface radiated time delay between the excited and gated optical pulses.
power can be estimated by calculating the surface static We can deduce the following statements: (i) The for-
energy Es which has the following form: ward and backward radiated electromagnetic waves must

22 June 2025 [Link]

E,=&E2d, (10) be TM waves; (ii) the outward radiated wave is collinear
with the reflected optical beam; (iii) the radiation fields
where d is the field length. Equation (10) can also be change sign when the laser incident angle crosses the nor-
explained as follows: A surface with a static field is a ca- mal direction; they drop to zero at normal incidence and
pacitor; it stores electromagnetic energy. With the trigger reach B maximum near the Brewster angle.
(or switch) from the laser pulses, the energy is released
and radiation is produced. If we use a GaAs sample and
assume a 10-V bias across d = 1 ,um, E, is approximately III. SURFACE STATIC FIELD
5.3 nJ/cm2. If all the energy can be released within 1 ps, A. Depletion field
one square centimeter area can generate 5.3 kW peak
power radiation. Most semiconductor surfaces have occupied surface
states. These surface states “pin” the Fermi level and bend
the energy bands near the surfaces or semiconductor/air
D. Radiation detection
interfaces, forming a charge depletion region and thus a
The optical excitation and coherence detection of elec- built-in surface electric field. The depletion field is directed
tromagnetic pulses include a balanced colliding-pulse normal to the surface, and the strength of the field is a
mode-locked (CPM) dye laser as the femtosecond optical function of the Schottky barrier potential and the doping
source to provide excitation and gating pulses, and a pho- concentration.
toconducting dipole antenna, fitted with a 3.2-mm trun- A typical expression of the surface depletion field
cated sapphire ball to enhance the collection efficiency, as Ed(x) (for an air/semiconductor interface) as a function
an electromagnetic wave receiverg-” The CPM laser pro- of the distance x vertical to the surface can be written as
duces an output pulse energy of 0.2 nJ at a repetition rate
Ed(x) = (eN/K) ( W-x), (11)
of 100 MHz and a 70-fs pulse duration at a center wave-
length of 625 nm. where N is the impurity concentration and W is the deple-
Figure 1 schematically illustrates the experimental tion width and has the form of
setup. The optical beam was split into two beams by a
W= d(2K/eN) [ V - (kT/e) I, (12)
beam splitter. In general, approximately 7 mW of average
optical power was used for the pump beam and 5 mW for where V is the potential barrier and kT/e is the thermal
the probe beam. The unfocused stronger beam, illuminat- energy. For iV = 10i8/cm3 and K = 12, the maximum
ing the semiconductor surface, was modulated by a me- value of the surface field can be as high as 5 x lo4 V/cm.
chanical chopper at a 2-kHz rate; the weaker beam, used The direction of the surface depletion field depends on the

328 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 328
 Semi-insulating InP TABLE I. The amplitude of the outward radiated field of several semi-
FIG. 2. Band diagram of conductor samples at room temperature.
Ec semi-insulating Fe:InP. Sur-
face state near the conduc- InP GaAs CdTe CdSe InSb Ge GaSb Si GaSe
----------Fermi Level tion-band edge causes Fermi
level “pinning” at the inter- Signal 100 71 33 11 8 7 2 0.5 <O.l
E” face. Photocarriers are swept
cross the depletion width Zdby
the built-in field.
GaSe at an optical wavelength of 620 nm are approxi-
mately 10 times longer than those in InP and GaAs, caus-
ing few carriers in the depletion layer to contribute to the
doping types of the impurities and the position of the sur- radiation. Among these samples, the radiation strength of
face states relative to the Fermi level. In general, the en- semi-insulating InP is 200 times stronger than that of Si.
ergy band bending near the surface is opposite in n- and The (lOO)-oriented semi-insulating iron-compensated InP,
p-type samples, resulting in opposite directions of the de- which shows the strongest optically induced electromag-
pletion fields. netic radiation, has been analyzed by Auger spectroscopy
Figure 2 schematically illustrates a band diagram of and found to have a 15-A oxide layer on the surface. Dif-
semi-insulating InP with photocarriers near the air/ ferent semiconductors show different wave forms of radi-
semiconductor interface. The surface state of a semi-insu- ated fields. With the same detection system, the amplitude
lating InP is near the conduction-band edge. As a result of of the radiation signal from semiconductors is dependent
Fermi level “pinning” at the interface, both the conduction on the magnitude and extent of the surface field, the ab-
band and valence band bend downward and form a deple- sorption coefficient for the laser light, the absorption coef-
tion layer of width ld near the interface. When an ultrafast ficient for the generated THz beam, and the dynamical
laser pulse illuminates a bare semiconductor surface with response of the photocurrent.
the photon energy greater than the band gap, photons are Figure 3 shows a typical wave form of the radiated
absorbed, creating electron-hole pairs. The built-in static field from semi-insulating InP, detected by the dipole an-
field drives the two kinds of carriers in opposite tenna in the outward direction (45” relative to the normal)
directions-the electrons to the surface and the holes to the with a 4-mm nominal diameter laser beam. The electrical
wafer. The free carriers are swept across the depletion pulses at a longer time delay are the multiple reflections

22 June 2025 [Link]

width Id, the photocurrent flows in, and a dipole layer from the semiconductor surfaces. A 16.2” diffraction angle
builds up. The rise time of the photocurrent is on the order of the radiated power has been measured at far field; this
of the laser pulse duration, and generally the decay time is value agrees well with the diffraction theory value of 15”.
the transit time of free carriers crossing the depletion layer The amplitude of the radiation field is found to be linearly
(assuming the carrier lifetime is longer than the carrier proportional to the low optical pulse energy from 2 to 100
transit time). The transient current in the depletion layer pJ. Due to the resonance behavior of the dipole detector
radiates electromagnetic waves. The measured bandwidth and slow response of the photoconductor, the bandwidth
of radiation is over 5 THz and is independent of the free- of the radiation is limited by the detection bandwidth. A
carrier lifetime of the semiconductor.‘2 The inward and recent study of the bandwidth of the radiation by using the
outward directions of the radiation beams are diffraction interferometric technique found a 160-fs electromagnetic
limited, and they satisfy generalized Fresnel laws. The out- pulse duration from the InP sample with an amplified laser
ward radiation field is collinear with the reflected optical beam.12 The frequency component ranged from 3 to 150
beam, and the inward radiation field emerges from the cm-‘.
sample after THz beam absorption in the wafer and We have measured the angular-dependent radiation by
Fresnel loss at the boundary. measuring the transmitted electromagnetic wave through a
0.4-mm-thick semi-insulating InP wafer as a function of
B. Radiation from semiconductor surfaces
A large variety of samples as radiation sources has
been selected from III-V, II-VI, and group-IV semiconduc- 6-
tors. For the semiconductors with band gaps larger than 3 I I
5 4- InP
the CPM photon energy (ti = 2 eV) , no radiation process rr
is expected, and this has been verified by ZnSe
(E, = 2.4 eV) and GaP ( Eg = 2.2 eV) .2 We have achieved
electromagnetic radiation from InP, GaAs, Gal _ Afis
(X <0.2), GaSb, InSb, CdTe, CdSe, Ge, and Si; all these
samples have band gaps lower than the incident photon
energy. The amplitudes of the radiated fields of different Time Delay (ps)
samples are listed in Table I, where the amplitudes are
normalized to that of the InP sample. Si and GaSe show FIG. 3. Wave form of the outward electromagnetic radiation from a
moderate signals, because the absorption lengths in Si and semi-insulating InP with 0,, = 45”.

329 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 329
 FIG. 4. (a) Angular-de-
optical pendent experimental
setup. D is a 100~pm di-
tb) pole antenna detector; (b) 0 2 4 6 8 IO 12
transmitted amplitude of Time Delay (ps)
radiated field El as a func-
tion of the rotated angle of
InP with measured data
FIG. 5. Radiated wave forms of both n- and p-type bulk GaSb. Opposite
(dots) and theoretical cal-
polarity of the wave form reflects the opposite direction of the depletion
culation (curve).
field of the different type doped GaSb samples.

The amplitude of the n-type sample is larger than that of

-1 .o the p-type sample. This is believed to be due to the different
-80 -40~ 0 40 80
values of the surface fields. Similar results were also
Angle (degrees)
achieved from GaAs samples (n,p z 1016/cm3). We ob-
served slight wave-form shape changes between n- and p-
type samples. The radiation pulses from thep-type samples
the incident angle. In this measurement, a 0.5 x 5-cm2 InP
have richer high-frequency components than those from
wafer was placed in the laser beam path before the dipole
the n-type samples. This difference also comes from the
detector and the wafer was rotated horizontally around its
different values of the surface built-in potential, resulting in
symmetry axis parallel to the 0.5-cm edge. The result of the
different depletion widths and different radiation wave
radiation field versus the rotation angle measurement is
forms. For p-type GaAs and GaSb samples, the surface
plotted in Fig. 4. The solid dots are experimental data and
fields are smaller and the depletion- widths are short com-
the curve is calculated from Eq. (8). The agreement be-
pared with those of n-type samples; therefore the ampli-

22 June 2025 [Link]

tween the experimental result and calculated curve con-
tude of the radiated fields from the p-type samples are
firms that the flow direction of the photocurrent is normal
weaker and the pulses are faster.
to the surface. The amplitude ratio of the TM wave (po-
We have investigated the strength and polarity of the
larization in the incident plane) to TE wave (polarization
surface fields by using several n- and p-type GaAs samples
normal to the incident plane) is over 2O:l.
with doping concentrations from 10”/cm3 to 10i9/cm3
and semi-insulating liquid-encapsulated Czochralski
C. Doped semiconductors
(LEC) GaAs samples. The results are listed-in Table II. A
The model of the dipole approximation can also be higher doping concentration reduces the radiation field due
confirmed from the measurement of the doped semicon- to the absorption of submillimeter waves from the conduct-
ductors. Since the directions of the surface depletion fields ing substrate. The absorption coefficient at submillimeter
are opposite between n- and p-type doped samples, the wavelengths is proportional to the square root of the dop-
measured radiated wave forms should have opposite polar- ing level. There are other effects of doping which have not
ities in n- and p-type samples. We have observed the op- been well characterized, such as the dependence of the
posite polarity wave forms of the radiated electromagnetic transient carrier mobility on the doping level. In Table II,
pulses from n- and p-type semiconductors, such as GaAs the n-type samples show the opposite polarity compared
and GaSb. Figure 5 shows the radiated field from n-type with the p-type samples. As we mentioned before, this dif-
(n z 10’*/cm3) and p-type (p s 10’*/cm3) GaSb sam- ference is a result of the opposite directions of their deple-
ples. The opposite polarity of the radiated fields from n- tion fields due to the different type of majority carrier. For
and p-type samples indicates opposite static surface fields. the undoped GaAs (LEC), the energy level of EL2 makes

TABLE II. The amplitude of the outward radiated field from GaAs samples with different doping concentrations. The polarities of the radiated fields
from n-type samples and p-type samples are opposite. Mobilities and resistivities are listed also.

N (cm s3)

LEC 3.5 x 10’5 6.7 x lOI 6.3 x 10” 2.0 x 10’8 1.6 x 10’9

j.6 (cm’/V s) ... 4958 263 2975 1997 47.4

p (am) ... 36.1 35.4 0.33 0.153 0.85
Doping n
Amplitude (a.u.) fn1.7 + 2.02 - :.47 + i.25 + :.1g -to*

330 J. Appt. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 330
 2
”
FIG. 6. The amplitude of the out- ?
ward radiated field E vs the dop- 3
ing concentration N from G&s
cUs1
samples plotted on a logarithmic
c9
scale. A slope of - 0.4 is mea-
sured.

0
15 16 17 18 19 20
Log(N) ki V bias (Volt)
7-V

it close to n-type; therefore, the wave form of the radiated FIG. 8. The amplitude square of the outward radiated geld vs an external
bias voltage on the Schottky barrier from n-type (positive bias voltage +
tield of undoped GaAs has the same polarity as that of
reverse bias), where V,, = kT/e - Vat the intersection point.
n-type GaAs. Figure 6 shows the amplitude of the outward
radiation from doped GaAs for a range of impurity con-
centrations N, from 1015/cm3 to 10t9/cm3 on a logarithmic fore, a bias on the Au film results in a uniform modulation
scale. A simple analysis developed in the following section of the electric field at the semiconductor surface.
predicts that the amplitude of II,,, is proportional to the For a moderate bias voltage the external potential is
inverse of the square root of the doping concentration N, mainly applied across the Schottky barrier and changes the
E cc N-O.‘. We have observed E a N- o.4in the GaAs depletion width W. For high forward bias, the high current
samples. The difference in the exponent ( - 0.5 vs - 0.4) causes the bulk resistance to reduce the potential affecting
mainly comes from the absorption and reflection of the the Schottky barrier and the built-in field cannot be com-
submillimeter wave by the substrate. pletely eliminated. For high reverse bias, the field across
the depletion region causes a breakdown of the Schottky
diode, again limiting the effect of the surface field. The
D. Metal/semiconductor Schottky barrier

22 June 2025 [Link]

externally controlled surface field can be monitored by the
Metal/semiconductor Schottky barriers have been radiated electromagnetic signal.
used to determine and estimate the sign and the strength of We can find a simple relation between the radiation
the surface field. The schematic illustration of the calibra- field and the surface depletion width W. If the depletion
tion is shown in Fig. 7. Thin gold films were evaporated on width W is much less than the optical absorption length
n- and p-type GaAs (111) wafers (n, p =: 1016/cm3). The a --I (W-g a-l ), which is generally true for carrier con-
thickness of the gold films was varied from 40 to 80 A, centrations N > 1016/cm3 the photocarrier density is
which allows greater than 60% laser beam transmission. nearly uniform within the depletion width. Furthermore
The wafers were chemically cleaned and heated in ultra- Monte Carlo calculations show that the drift velocity sat-
high vacuum to 550 “C to produce clean surfaces prior to urates with the electric field for fields greater than 5
gold deposition in the same vacuum system. An indium kV/cm.r3 The calculations are based on a colliding-pulse
contact on the back of the sample was annealed to provide mode-locked (CPM) laser pulse excitation ( 100 fs pulse
an ohmic contact. The Schottky samples show good rec- duration and 2 eV photon energy), and the temporal range
tifying behavior in current-voltage measurements. There- of the calculated velocity is over 1 ps. Therefore the drift
velocity u is saturated by the field in the depletion region
(not sensitive to the position x) . With the above approxi-
mations, the amplitude of the radiated field E, is simply
incident laser beam EM radiation
proportional to the depletion width W:
Era W, (13)
where W depends on the impurity level N and surface
potential V, as indicated in Eq. ( 12).
bsorption width
Figure 8 shows the amplitude square of the radiated
signal from n-type GaAs with a 40-A Au film versus the
applied bias voltage. The radiated field increases at a mod-
erate reverse bias on the n-type sample and saturates at
indium ohmic contact higher reverse bias. Basically, the reverse bias enhances the
internal (built-m) field and increases the depletion width,
FIG. 7. Schematic illustration of the outward radiated field from a Au/
GaAs Schottky barrier with an external bias (the inward radiation is not
while the forward bias cancels the internal field. The direc-
shown). For GaAs samples with a doping level greater than 10L6/cm3, the tion of reverse bias (as well as forward bias) of n and
depletion width is less than the photon absorption width. p-type Schottky barriers is opposite. Since I? cc W2

331 J. Appt. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 331
 a
Sb
2 6
4
% 4
s. 1 120 K v/ \
g 2
F 0
z
a -2
2 -4
-6O+-
4
Time Delay (ps) a-

FIG. 9. Wave forms of the radiated electromagnetic pulses from GaAs at I I

---v------l
1 I I I 1 I
II
three different temperatures. 0 2 4 6 8 10 12
Time Delay (ps)

a (V + hias - h-T/e) from Eq. (3) and (4), the static

built-in potential V can be estimated from the intersection FIG. 10. Wave forms of the radiated electromagnetic pulses from InSb at
POhlt where Vbias = kT/e - V in Fig. 8. We have esti- several different temperatures.
mated that the built-in field V has a value of 0.8 1 ho.05 eV
for an n-type Au/GaAs Schottky sample. In principle, the
analysis of experimental data is analogous to the electronic fields with temperature should correlate with the temper-
C-V measurement. ature dependence of the transient surface conductivity.
Similar to the analysis of the n-type Au/GaAs Since the radiated field is proportional to the first time
Schottky barrier, we have estimated V = 0.32 =f=O.1 eV for derivative of the photocurrent within the depletion width,
a p-type Au/GaAs Schottky barrier. The radiated field of the pulse shape and pulse duration of the radiated field
n-type GaAs at zero bias is about 5 times stronger than should vary with the change of the depletion width. In
that of the p type. Also the increased experimental error in general, the depletion width is increased when the temper-

22 June 2025 [Link]

the p-type sample is from the larger leakage current in our ature decreases. Due to the nature of the narrow band gap,
p-type Schottky diode. the change of the depletion width of InSb is more sensitive
to the temperature than that of the wider band gap of
E. Temperature dependence GaAs, resulting in a larger change of the radiated wave
form of InSb in Fig. IO. In addition, the temperature-de-
The amplitude of the radiation is temperature pendent complex conductivity can also affect the shape of
dependent.!4 For example, increases of the peak field value the radiated field. Both the complex conductivity and de-
of 1.7 (or 4.1) and energy value of 3.4 (or 21) of the pletion width of the undoped GaAs are less sensitive to
radiated electromagnetic wave have been found from GaAs temperature change than that of InSb.
(or InSb) when the sample temperature decreases from Figure 12 plots the radiated energy W, which is de-
280 to 80 K. These changes can be explained by the tem- fined as W(T) = JE2( T,t)dt, versus temperature. The
perature-dependent transient mobility. increase of the radiation (the field and the energy) is due
Several semiconductors have been used in the temper- to higher carrier mobility and less absorption of the radi-
ature-dependent experiment. Liquid-encapsulated Czo- ated wave by the thermal carriers at low temperature.
chralski (LEC!) undoped GaAs (n = 1016/cm3) and un- We define a and /3 as the ratios of the radiated peak
doped InSb (n = 8 X 1015/cm3 at 77 K) are the two value [y = E( T = 80 K)/E( T = 280 K)] and radiated en-
major samples used, because they are examples of rela- ergy value p = W( T = 80 K)/W( T = 280 K)], respec-
tively wide-band-gap and narrow-band-gap materials, re- tively; y and fl from several semiconductors are listed in
spectively. For example, the band gap of GaAs at room
temperature is 1.42 eV; the band gap of InSb is 0.17 eV.
The samples are ( lOO)-oriented substrate wafers. Figure 9
shows the temperature-dependent radiated field from
GaAs. The peak of the field increases monotonically with
GaAs
decreasing sample temperature. In contrast with GaAs, the
wave form from InSb varies with temperature, and the FIG. 11. Peak values of
electromagnetic pulses
peak value reaches a maximum at approximately 120 K, as from GaAs and InSb vs
l .
shown in Fig. 10. For a better comparison, the peak values temperature.
of the radiated fields versus temperature for both InSb and Mb
.
GaAs are plotted in Fig. Il. Since the radiated field is 0 % I I
proportional to the first titie derivative of the transient 100 200 300
conductivity, the change of the peak values of the radiated Temperature (K)

332 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 332
 8
0 GaAs
0 0
.
0
0 7 6
. 0 0 5%. 4
FIG. 12. Radiated electro-
InSb . magnetic energies from 3
GaAs and InSb vs temper- W2
ature.
0
lLL- 200
Temperature (K)
30
0.0 0.5
Frequency
1.0
(THz)
.

FIG. 13. Amplitude spectrum from InSb at two different temperatures.

Table III. At the temperature region greater than 180 K,

the rate of change of the radiated energy with temperature much smaller value of the transient carrier mobility ( <3
from InSb is faster than that from GaAs; when the tem- ps) than that of the steady-state value at room temperature
perature is below 180 K, InSb and GaAs have approxi- has been observed.15 The transient mobility in our mea-
mately the same rate of change. For a fixed temperature surement should reflect both the hot carrier and high-elec-
and comparable doping concentration, the radiated field or tric-field properties. We have also measured the tempera-
energy from wider-band-gap semiconductors is larger than ture-dependent radiated fields from InP, GaSb, GaAlAs,
that from the narrow-band-gap semiconductors, because and Ge. The results show similar temperature-dependent
wide-band-gap samples (such as InP and GaAs) have a behavior of the radiated field compared with that of the
larger surface depletion field than that of narrow-band-gap thermal equilibrium carrier mobility, which is consistent
samples (InSb and GaSb) . with the results from InSb and GaAs.
As we mentioned previously, the radiated field is pro- The frequency spectrum of the radiated field can be
portional to the time derivative of the surface conductivity, obtained numerically by a Fourier transform. We have
so the transient carrier mobility (-Lcan be further related to found that the bandwidth of the amplitude spectrum from
the radiation through the equation cr = enp. We can com- wider-band-gap semiconductors (such as GaAs, InP) does
pare the temperature dependence of the transient carrier not change with the temperature, while that from nar-

22 June 2025 [Link]

mobility with that of the steady-state carrier mobility. For rower-band-gap semiconductors does. Figure 13 shows the
example, for n-type InSb with a carrier concentration of amplitude spectrum of InSb at temperatures of 80 and 280
10i6/cm3, the steady-state electron mobility versus temper- K; the frequency spectrum shifts over 350 GHz. The shift
ature has a single maximum at approximately 120 K, with of the frequency spectrum (Fig. 13 ) and the shape change
a fast decrease at higher temperatures due to the phonon- of the wave forms (Fig. 10) with temperature are due to
dominated scattering and a slow decrease at lower temper- the change of the depletion width and complex conductiv-
atures due to the impurity-dominated scattering. Also for ity of the narrow-band-gap InSb.
n-type GaAs (n = 1016/cm3), a monotonic increase of its
steady-state carrier mobility versus decreasing temperature F. Anisotropic radiation
is expected within this temperature interval.
If the temperature dependence of the radiated field is One of the most interesting results is that the ampli-
dominated by the temperature-dependent transient carrier tude of the radiation varies with the crystal orientation.
mobility as E( T) a a( T) CC,LL( T), we find that the temper- This anisotropic behavior is believed to be related to the
ature dependence of the transient carrier mobility is similar anisotropic transient carrier mobility.
to that of the steady-state carrier mobility. This result is a We start from semi-insulating zinc-blende single crys-
surprise, since the transient carrier mobility is not at ther- tals. The GaAs and InP samples are ( 100) and ( 111)
mal equilibrium, and the effective temperature of the ini- oriented. The CdTe sample is ( 111) oriented. The polish-
tially photoinduced carriers can be as high as several thou- ing quality of the sample surface is not essential in this
sand degrees. The measurement of the transient carrier measurement. No substantial difference in radiation has
mobility of GaAs has been reported at low electric field, a been observed from sandblasted or mirrorlike surfaces.
Figure 14 schematically illustrates the experimental setup.
The samples were mounted on a computer-controlled ro-
TABLE III. The ratios of the radiated peak value [y = E( T = 80 K)/ tation stage with a 0.9” resolution, and it was rotated along
E( T= 280 K)] and radiated energy value [p,- W( T= 80 K)/W(T its surface normal direction. The incident angle of the laser
_-280 K)] from several semiconductors. For comparison, the relative beam was approximately 55” relative to normal. The polar-
strengths of the radiated field at 300 K are shown also.
ization of the incident la&r beam and the polarization of
GaAs InP Ge GaSb InSb the detector were set in the p plane.
A zinc-blende structure viewed from its [l 1 I] axis and
E(T= 300 K) 4.1 9.8 0.7 0.5 0.8
1.7 3.4 3.7 4.0 4.1
from its [ 1001 axis is illustrated in Figs. 15 (a) and 15 (b) ,
3.4 7.1 I1 18 21 respectively, where the solid dots and the shaded dots rep-
resent crystal cations and anions. There is a total of 24

333 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 333
 outward radiation &
reflected laser beam

I I
0 120 240 360
Rotating Angle cp (degrees)

FIG. 14. Schematic illustration of setup for crystal orientation experi-

ment. Sample was rotated around its [ll l] axis.
FIG. 16. Experimental data from ( 111 )-oriented InP. The curve is a plot
of 0.95 + 0.05 sin(3q).
symmetry operations in the zinc-blende structure. Eight of
them are 2?r/3 (120”) rotations about the body diagonals
(11 l), and three r ( 180“) rotations about the surface nor- misfit dislocation formation. For a sufficiently thin layer in
mal. the superlattice, the tetragonal distortion is the dominant
Figure 16 plots the amplitude of the radiation versus
process to accommodate the strain. The macroscopic po-
the rotating angle q from a semi-insulating ( 111 )-oriented larization Pi induced by the strain is proportional to the
InP sample; the sample clearly presents a threefold rota- off-diagonal components of the strain tensor, ej,+ The
tion symmetry at the ( 111) surface. The dots are the mea- strain-induced polarization has a form of Pi = 2ehrejk, with
sured peak value of the radiated signal and the curve is the the piezoelectric coefficient e41. Since eii = 0, the diagonal
plot of 0.95 + 0.05 sin(g,/3). The shape of the radiation strains do not induce a polarization for (lOO)-growth-axis
wave form remains unchanged during the crystal rotation, strained-layer superlattice. Any other growth direction
except for approximately a 10% amplitude variation with a strained-layer superlattices show the strain-induced
modulation period of 120”. For a semiconductor surface polarization. I6 Particularly in a ( 111 )-oriented strained-
normal with a few degrees off from the ( 111) axis toward layer superlattice, the strain-induced macroscopic polariza-
the (100) axis, the measured data can be approximately tion generates an electric field along the (111) growth
fitted by an equation of A + B sin( 3p) + C sin(4cp) where

22 June 2025 [Link]

axisI The amplitude of this piezoelectric field is given by
A, B, and C are constants, and B and C depend on the
angle of the surface normal away from the (111) axis. (14)
Currently there is no theoretical model to explain this
The sign of the off-diagonal symmetric strain tensor ejk

anisotropic behavior. One possible explanation of the an-

depends on the biaxial compression and expansion (tetrag-
isotropic behavior may be related to the anisotropic tran-
onal distortion) of the layers. The direction of the piezo-
sient carrier mobility. This model requires further study. electric field in the layer depends on the type of distortion
IV. STRAIN-INDUCED FIELD of the layer (contraction or expansion) and the grown face
of the substrate. For the layer that has a small lattice con-
A. Piezoklectric field stant and suffers expansion, the piezoelectric field from an
The misfit strain in a lattice misfit superlattice, in gen- A-face (B-face) superlattice points inward (outward).
eral, can be accommodated by the elastic deformation and The ratio of the piezoelectric fields of the well 1E, 1
and the barrier 1Eb 1 depends on the lattice constant of the
thick buffer layer grown on the substrate. The layers (bar-
(4 riers or wells) with a lattice constant close to the lattice
constant of the buffer layer have a smaller strength piezo-
electric field, since the layers have less distortion and
strain. The strength of the piezoelectric fields in some
strained-layer superlattices can be as high as lo6 V/cm.
The piezoelectric field provides a suitable bias for gen-
erating optically induced electromagnetic radiation. The
FIG. 15. A zinc-blende structure radiated fields from the strained-layer superlattices, such as
viewed (a) from its [ill] axis; GaSb/AlSb, InGaSb/InAs, and InGaAs/GaAlAs, have
(b) from ita [loo] axis.
been measured and analyzed. The structures of InGaAs/
(b) GaAlAs, GaSb/AlSb, and InGaSb/InAs samples are listed
0 (001) in Tables IV,. V, and VI, respectively. All the multiquan-
turn wells and superlattices were designed to have thick
buffer layers. For the lattice constant of the buffer layer
equal to the average lattice constant of the superlattice, the
piezoelectric field alternates in the constitute layers. For
the buffer that is lattice matched to the quantum wells (or

334 J. Appl. Phys., Vol. 71, No. I, 1 January 1992 X-C. Zhang and D. H. Auston 334
TABLE IV. The parameters of four pairs of (lOO)- and (11 l)Boriented TABLE VI. Samples no. 74Oa. 74Ob, and 740~ are the InGaSb/InAs
GaSb/AlSb superlattices. Sample no. 544a is the (lCkO)-oriented super- superlattices having (loo), (3ll)B, and (311)A orientations. They are
lattice, no. 544b is the (Ill)-oriented superlattice, and so forth in the type-II staggered superlattices. The top layers are Inc,4Gac,c,Sb.
other pairs. The top layers are GaSb.
Sample no. Lb Un0.4Gaa6Sb)(A) L, (InAs) (A) Periods (N)
Sample no. Lb 6-b) 6) L, (GaSb) 6) Periods (N)
740a,b,c 40 40 50
545a,b 300 4cHl 10
544a,b 300 300 15
595a,b 150 150 20
596a,b 150 100 20
the strain-induced electric field contributes to the electro-
magnetic radiation, the wave forms from the (lll)-ori-
ented GaSb/AlSb superlattices should be very different
barriers), the quantum wells (or barriers) have essentially from the others. For a free-standing (ill)-oriented
no distortion, leading to no strain and no piezoelectric strained-layer superlattice, piezoelectric fields exist both in
field. The orientation of the piezoelectric field in the layer the wells and the barriers, and their directions are opposite.
depends on the distortion type of the layer (contraction or With a thick ( 1.2 pm) AlSb as a buffer layer for the su-
expansion) and the growth face of the substrate. For the perlattice, the lattice distortion in the AlSb barriers is
well that has a small lattice constant and suffers expansion, much less than in the GaSb wells and the strain in the AlSb
the inward or outward direction of the field depends on barriers is negligible compared with that in the GaSb wells.
whether the superlattice is grown on the A-face or B-face Furthermore, in (11 l)-oriented GaSb/AlSb superlattices
substrate. The piezoelectric in ( 111 )-oriented strained- with the optical excitation energy of 2 eV; the radiation
layer superlattices has been confirmed by optical and op- from the quantum barriers is much weaker than that from
toelectronic techniques. ‘E-2’ the quantum wells because of the moderate,field strength
and smaller absorption coefficient (indirect band minimum
B. GaSb/AISb superlattice for AlSb, direct band minimum for GaSb) in the barriers.
Since the (ill)-oriented superlattices were grown on
Four pairs of GaSb/AlSb strained-layer superlattices (111) B-face substrates, the direction of the strain-induced
and one pair of GaSb thin-film samples grown on (100) field in the GaSb quantum wells, which have a smaller
and (11 l)B (4” off from (111) direction) -oriented GaAs lattice constant and suffer compression, points outward,

22 June 2025 [Link]

wafers by molecular-beam epitaxy (MBE) were used in opposite to the surface depletion field for p-type samples.
this experiment. The sample structures are shown in Fig. The strain-induced field overrides the surface and barrier
17. In the GaSb/AlSb quantum-well structure, GaSb fields, resulting in an outward macroscopic electric field.
(E, = 0.69 eV> forms the quantum wells while AlSb The samples were mounted in a cryostat at a temper-
(E, = 1.6 eV) forms the barriers. The width of the quan- ature of about 80 K, with their front surfaces aligned to the
tum wells varies from 100 to 400 A for different sugerlat- same plane to equalize the timing for the temporal wave-
tices. The GaSb films have a thickness of 5000 A. The form measurement. The incident angle of the laser beam
conduction-band and valence-band offsets in the superlat-
tices are approximately 0.5 and 0.4 eV, respectively, giving
strong spatial confinement for both electrons and holes.
(a) <loo> &<ill> GaSbfilms
The superlattice and film samples are not intentionally
doped; Hall measurements show that all samples arep type GoSb 5000 d
1
Fd
and have a typical impurity level of 8.2 X 10t5/cm3. AlSb 1.2 ,u m
The Fermi level in these p-type samples “pins” the
1 <100>&<111> GaAs 1
energy bands downward near the surface and forms the 0

depletion field having a direction inward to the substrates.

The samples should show the same p-type behavior in the (b) clOO> GnSblAiSb sttperlattice
electromagnetic wave forms if the only contribution to the FIG. 17. The schematic structures
GaSb 80 i 4Fd
of GaSb/AlSb superlattice and
radiation is from the surface depletion layer. A larger ra- AlSb Lb
GaSb thin-film samples. Fd and F,
N
diation amplitude from the GaSb films than that from the c---- G&b Lw are the surface depletion field and
superlattice GaSb cap layer (80 h; thick) is expected due to the strain-induced field, respec-
the greater number of photocarriers in this region. When tively. The directions of the fields
are labeled as the directions of the
arrows.

TABLE V. The parameters of three pairs of (100) and (111) B-face (c) 4 1 I > GaSbIAlSb superlattice
InGaAs/AlGaAs superlattices. The top layers are GaAs.

Sample no. Lb (InGaAs) (A) L, (AlGaAs) (A) Periods (N)

426a,b 300 400 10

427a,b 300 300 15
428a,b 150 150 20

335 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 335
 n I I fi I t I t I L I b II

0.2 0.4 0.6 0.8 1 1.2 1.4

Frequency (THz)

I t 1
0 2 4 6 8 IO 12 FIG. 19. Fourier spectra of the temporal wave forms: (a) (11 l)-oriented
Time Delay (ps) GaSb film; (b) (11 l)-oriented GaSb/AlSb superlattice sample no. 544b.

FIG. 18. Wave forms of optically induced electromagnetic radiation from

(NO)- and (11 l)-oriented GaSb films and GaSb/AlSb superlattice sam- identical to that for GaSb/AlSb. Similar to the results from
ples no. 544a and no. 544b. The wave form from the (111) superlattice is the GaSb/AlSb sample, the radiation from the (lOO)-ori-
opposite to the others due to the piezoelectric field in strained quantum ented InGaAs/GaAlAs sample is weaker than that from
wells opposite to the surface depletion fields in thep-type GaSb cap layers. the (111) sample, and opposite polarity of the wave forms
is observed. The structures of (lOO>- and (11 l)-oriented
was 55” and the optical power density on the sample was 3 [Link]/Gao.6Al~,~As samples (no. 428a and no.
mW/cm2. The radiated electromagnetic pulses at the spec- 428b) are shown in Fig. 20. The radiated signals from the
ular angle were collected by two gold-coated off-axis par- samples (no. 428a and no. 428b) are plotted in Fig. 21.
abolic mirrors and were focused on a photoconducting an-
tenna by an attached sapphire lens. Figure 18 shows the D. GalnSb/lnAs superlattice
radiation wave forms from two GaSb films and a pair of Three GaaJno.4Sb/InAs samples (with ( 100) ,
superlattices (samples no. 544a and no. 544b) at 80 K. The (31 I)& and (311)B orientations) grown by molecular-
wave forms (p-type samples) from the surface depletion beam epitaxy (MBE) were used in our experiment. The
fields of the two GaSb films and the (100) superlattice are samples grown on GaAs substrates have a l-pm AlSb

22 June 2025 [Link]

similar, while the wave form from the (111) superlattice is buffer layer followed by 50 periods Gaa&re4Sb/InAs su-
like that of an n-type sample, opposite from the others, as perlattice, as shown in Fig. 22. The thickness of each layer
we expect from the opposite direction of the strain-induced is 40 A. The cap layer is 40 A GaasIna4Sb. The
field. The radiation from the piezoelectric field in the (11 l} Gaa.,+a.$b layers with the larger lattice constant contract
superlattice clearly overrides the contribution from the sur- as a result of biaxial compressive stress, whereas the InAs
face depletion field. The amplitude (absolute value) from layers with the smaller lattice constant expand due to a
the ( 111) superlattice is 8.6 times larger than that from the biaxial tensile stress. The presence of the strain leads to the
(100) superlattice. The other three pairs of GaSb/AlSb change of the energy band structure. The valence band of
superlattice samples show similar behavior: The wave Gac61ne.4Sb is overlapping the conduction band of InAs by
forms radiated from the ( 111) superlattices are opposite to approximately 100 meV, forming a staggered structure,
those from the ( 100) superlattices. with the electron confinement in InAs and hole confine-
A pulse duration of 450 fs (full width at half-maxi- ment in Gac,,&rc4Sb at thermal equilibrium. Since the di-
mum) is measured for the ( 111) superlattice and 800 fs for rection of the strain-induced polarization macroscopically
the films (without deconvolution) . The time domain wave
forms were transformed to the frequency domain for a
better comparison. Figure 19 shows the frequency spec- (a) <ill> MQW
trum of the (111) film and the (111) superlattice. The
spectrum of the film peaks at 0.4 THz while the spectrum
of the ( 111) superlattice is completely limited by the de- 3
+ Fs
tector response, which has been calibrated by electro-optic
Cherenkov radiation from a LiTa03 crystall We also ob-
served that the spectra from the (111) superlattices with
narrower well width have richer frequency components in FIG. 20. The schematic structures of
the high-frequency region than the spectra from wider (b) <loo> MQW strained-layer InGaAVGaAlAs su-
wells, but lower amplitudes due to the poor antenna effi- petiattices.
ciency at the higher frequencies. +Fd
C. InGaAs/GaAIAs superlattice
We also measured the radiation from strained-layer
I<lOOr GaAs 1
InGaAs/GaAlAs superlattices. The experimental setup is

336 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 336
 I L I I I 1 I
a 10
24 _ <311>A
4 5 .
R 0
$j -5
3
&! -10
-4 -2 0 2 4 6 0 2 4 6 8 10 12 14
Time Delay (ps) Time Delay (ps)

a 10
FIG. 21. Wave forms of the radiated field from the (lOO)- and (1 II)- d
oriented InGaAVGaAlAs superlattices. 54 5
E 0
3 -5
depends on the orientation of the dipoles formed by the B
2 -10
displacement of the cations and anions under the strain, 0 2 4 6 8 10 12 14
the piezoelectric field which is proportional to the polar- Time Delay (ps)
ization is expected to have equal strength but opposite di-
rection in A-face (cation is on the top surface) and B-face a 10
J
(anion is on the top surface) (311) superlattices. 4 5
In contrast with the GaSb/AlSb superlattice of which %I 0
AlSb is an indirect-band-gap material, both the constituent 2 -5
layers of the GaInSb/InAs superlattice are direct-band-gap 3 -10
materials. Due to the AlSb buffer which has a lattice con- 0 2 4 6 8 10 12 14
stant between that of Ga&n&b and InAs, Ga&no,4Sb Time Delay (ps)
and InAs have opposite-oriented piezoelectric fields. The
absorption coefticient of G%.$n&b is about 3 times larger
than that of InAs, resulting in over 200 times the carriers FIG. 23. Wave forms of optically induced electromagnetic radiation from
(311).4-, (311)B- and (lOO)-oriented Ga,,Jn,,,,Sb/InAs superlattices.
absorbed in Ga&no.$b layers than in InAs layers, and the The wave form from the (31 l)d-oriented superlattice is opposite to the
G%,61no.4Sb layer suffers more strain due to more lattice

22 June 2025 [Link]

others due to the piezoelectric field in the (311)koriented superlattice
distortion than InAs on a thick AlSb buffer. If the piezo- being opposite to the (311)Boriented superlattice and the surface deple-
electric coefficients from G~.61n0,4Sb and InAs are of the tion field in the cap layer.
same order, the radiated signal from the Ga.$Q.4Sb layers
will override the signal from the InAs layers.
For a zinc-blende strained-layer superlattice, a (lOO>-
or (1 lO)-oriented sample does not have a strain-induced
piezoelectric field; the only static electric field is the surface
depletion field. Superlattices with other orientations have
the strain-induced piezoelectric field. The radiation from a
(3 1 l&oriented superlattice has contributions both from
the surface depletion field and from the piezoelectric field,
whereas the radiation from a (lOO)-oriented superlattice
has a contribution only from the surface depletion field.4
The wave forms of the optically induced radiation from
three samples are plotted in Fig. 23. The signal radiated
Cb) + ES from the (3 1 l)A sample clearly shows an opposite polarity
compared with that from the (3 1 l)B sample. The ampli-
9 FIG. 22. The schematic tude of the signal from sample (3 1 l)A is smaller than that
50 EP
c structures of Ga,,Jn,,,.,Sb/
4 InAs superlattices.
from sample (31 l)B, due to the surface depletion field
cancellation. Therefore the measured signal S’, which is
proportional to the radiated field E,, can be written as

S(m) =WG), (15)

+ ES S(~)A=WW -W$), (16)

4 S(311)~=S(Es) + W$,), (17)
EP
* where E, is the surface depletion field that does not
strongly depend on the crystal orientation, and EP is the
average macroscopic piezoelectric field in Gao.61no.4Sblay-
c3Il>B GaAs XI.
ers. The amplitudes of radiated signals from the surface

337 J. Appt. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 337
depletion field and piezoelectric field can be further ex- requiring large optically injected carrier densities. Typical
pressed by the measured signals from three samples injection levels are on the order of 1013 cm-j, which also
S(IOO), SWIM, and SWI)B as ensures that the injected carriers act only as a probe and do
not screen the field. The directional and diffraction-limited
SW =~%m~=$(~~311~~ + S(311)~)7 (18) propagation property of the radiation leads to the potential
(19)
application of probing semiconductors in the chamber en-
W$) =@311p - S(311)~).
vironment, such as in an UHV system, to monitor the
The peak amplitudes of the wave forms in Fig. 2 have the surface field with a contactless approach.
values as S(too) = 0.35; Sc3t11A = - 0.5 and S(311)B
= 1.2, separately. These measured data exactly fit with Eq. ACKNOWLEDGMENTS
( 18). The good agreement between the experimental data
and the calculated results suggests that the (3 1 l)A- and We acknowledge lab assistance from B. B. Hu, J. T.
(3 1 l)B-oriented superlattice samples are of very high qual- Darrow, and L. Xu. We are grateful to W. I. Wang, S. H.
ity. Xin, P. Mei, E. S. Yang (Columbia University), and R.
The Gae&&b/InAs superlattices used here are not Nottenburg (AT&T Bell Lab) for providing semiconduo
intentionally doped; Hall measurements show that samples tor samples and fruitful discussions. We would also like to
are weakly p type. Using the direction of the surface de- express our appreciation to P. R. Smith for assistance with
pletion field that points inward to the substrate as a refer- the fabrication of dipole detectors. This research was sup-
ence to calibrate the field orientation, we find that the pi- ported by the Office of Naval Research under Grant No.
ezoelectric field in Ga&ne,$b layers (not including cap N00014-86-K-0694.
layer) is antiparallel to the surface depletion field in the
(31 l)A sample and parallel in the (31 l)B sample. Due to ‘J. I. Pankove, OpticaI Processesin Semiconductors (Dover, New York,
the opposite direction of the piezoelectric field in the alter- 1975), Chap. 18.
‘X.-C. Zhang, B. B. Hu, J. T. Darrow, and D. H. Auston, Appl. Phys.
native layer, a vis-a-vis result can be achieved in the InAs Lett. 56, 1011 (1990).
layers of Gae61ne$b/InAs superlattices. Following a sim- 3X.-C. Zhang, J. T. Darrow, B. B. Hu, S. H. Xin, and D. H. Auston,
ilar analysis, this technique can also be extended to probe Springer Ser. Chem. Phys. 53, 198 (1990).
the field in other crystal orientations. 4X.-C. Zhang, B. B. Hu, S. H. Xin, and D. H. Auston, Appl. Phys. Lett.
57, 753 (1990).
We have also tested the radiated signals from several ‘X.-C. Zhang, J. T. Darrow, B. B. Hu, D. H. Auston, M. T. Schmidt, P.
( 1 1 1)-oriented GaSb/‘.%1Sb superlattices. The data from Tham, and E. S. Yang, Appl. Phys. Lett. 56, 2228 (1990).

22 June 2025 [Link]

the (111) GaSb/AlSb sup-lattices, similar to the (3 11) 6B. B. Hu, J. T. Darrow, X.-C. Zhang, D. H. Auston, and P. R. Smith,
Appl. Phys. Lett. 56, 886 (1990).
GaInSb/InAs superlattices, showed an opposite polarity
‘J. T. Darrow, B. B. Hu, X.-C. Zhang, and D. H. Auston, Opt. Lett. 15,
between the A-face and the B-face samples, but their am- 323 (1990).
plitudes did not fit with Eq. ( 18). This was due to the “B. I. Greene, J. F. Federici, D. R. Dykaar, A. F. Levi, and L. Pfeifler,
different sample qualities, even though they were grown Opt. Lett. 16, 48 (1991).
‘P. R. Smith, D. H. Auston, and M. C. Nuss, IEEE J. Quantum Elec-
under the same growth conditions. tron. QE-24, 255 (1988).
“Ch. Fattinger and D. Grischkowsky, Appl. Phys. Lett. 54,490 (1989).
V. SUMMARY “B. B. Hu, X.-C. Zhang, D. H. Auston, and P. R. Smith, Appl. Phys.
Lett. 56, 506 (1990).
Optically induced electromagnetic radiation from “B. I. Greene, J. F. Federici, D. R. Dykaar, R. R. Jones, and P. H.
semiconductors provides a new diagnostic tool to study the Buksbaum (unpublished).
electronic properties of the surfaces and interfaces. Since “G M. Wysin, D. L. Smith, and A. Redondo, Phys. Rev. B 38, 12 514
(i988).
the generation and detection of the radiation are coherent “B. B. Hu, X.-C. Zhang, and D. H. Auston, Appl. Phys. Lett. 57, 2629
processes, the radiation wave form provides both ampli- (1990).
tude and phase information. “M. C. Nuss, D. H. Auston, and F. Cappsso, Phys. Rev. Lett. 58, 2355
The magnitude and sign of the radiated electrical pulse (1987).
“D. L. Smith, Solid State Commun. 57, 919 ( 1986).
is determined by the strength and sign of the internal elec- “D. L. Smith and C. Mailhiot, Rev. Mod. Phys. 62, 173 (1990).
tric field. By analyzing the radiation signal, the doping type “II . K . Laurich, K. Elcess, C. G. Fonstad, J. G. Berry, C. Mailhiot, and
(n or p type) and impurity concentration, transient carrier D. L. Smith, Phys. Rev. I&t. 62, 649 (1989).
tgE A. Caridi, T. Y. Chang, K. W. Goossen, and L. F. Eastman, Appl.
mobility, and crystal orientation can be determined and Phys. Lett. 56, -659 (1990).
estimated. The detection technique is extremely sensitive ‘OK. W. Goossen, E. A. Caridi, T. Y. Chang, J. B. Stark, D. A. B. Miller,
and can detect fields of a wide range of values without and R. A. Morgan, Appl. Phys. Lett. 56, 715 (1990).

338 J. Appl. Phys., Vol. 71, No. 1, 1 January 1992 X.-C. Zhang and D. H. Auston 338