PHYSICAL REVIEW B VOLUME 42, NUMBER 11 15 OCTOBER 1990-I

Generalized Franz-Keldysh theory of electromodnlation

H. Shen
GEO-Centers, Inc. , U. S. Army Electronics Technology and Device Laboratory, Fort Monmouth, New Jersey 07703

Fred H. Pollak
Physics Department, Brooklyn College of the City University of New York, Brooklyn, ¹w York 11210
(Received 10 August 1989; revised manuscript received 18 June 1990)
We present a detailed generalization of the Franz-Keldysh theory of electromodulation which
takes into account the presence of a dc surface electric field, 8&,. %'e demonstrate rigorously that
when the ac modulating electric field is small compared to Cd, , the period of the Franz-Keldysh os-
cillations yields 8d, . This result is independent of the relation between the width of the space-
charge region and the penetration depth of light. The effects of broadening also are considered.

I. INTRODUCTION tween the width of the space-charge region (SCR) and the
penetration depth of the light. The theory yields the rela-
There has recently been considerable interest in the use tion between the magnitude of the electromodulation sig-
of electromodulation' (EM) to study semiconduc- nal and the fields C„and Cd, . The effects of finite modu-
tors' ' and semiconductor microstructures. ' Elec- lation as well as the influence of broadening also are con-
tromodulation is the most useful of various optical modu- sidered. Calculations for the case of GaAs with various
lation methods since it yields, in general, the sharpest doping levels will be presented.
structure'
fields. ' and is sensitive to surface (interface) electric
' While most of the work in EM has made II. THEORY
use of the sharp, derivativelike features there is growing

or interface electric fields. ''

activity in the latter area, i.e. , EM as a probe of surface
A very useful aspect of
A. No broadening

Keldysh oscillations '''

EM to study these effects is the observation of Franz-
(FKO). ' ' Early
1. Electroreflectance
In electroreflectance methods' such as Schottky bar-
work on FKO was reported using electroabsorption in
rier, metal-insulator-semiconductor, or semiconductor-
junctions. ' '
the nonuniform field of the space-charge region of p-n
These authors pointed out that the period electrolyte configurations, the electric field is modulated
by +8„about bd, . In this case the variation in the
of the FKO was a direct measure of the maximum dc
field in the p-n junction for the case of small ac modula- dielectric function 5e (E, Cd„b„}can be expressed as
tion. However, no detailed theory in terms of Airy func- 5e (E, @d„b'„)=e(E,@d, +6"„) e(E, @d, —8„—
) (la)
tions was presented. Bhattacharya et al. ' observed that
in the reflection mode (electroreflectance and = b e(E, @d, + 8„) be(E, 6d, —6„),
photoreflectance) with small ac modulation the FKO
were directly related to the maximum field in the surface (lb)
space-charge region, i.e. , the dc surface field 6d, . These
authors presented a detailed theory which contained an
where E is the photon energy and
error although the final result was correct. Bottka and
' also have investigated the FKO related to
be(E, 6)=e(E, C) —e(E, O) . (lc)
co-workers
the surface space-charge layer. However, they have er- It has been shown that for an Mo (three-dimensional)
roneously concluded that the period of the FKO is relat- critical point the quantity b, e(E, 6) of Eq. (lc) can be
ed to the average dc field rather @d,. The use of FKO is written as'
becoming significant in evaluating surface electric fields
and related quantities such as Fermi-level pinning
be(E, 6 ) =(B/E )(A'0)' [G (71)+iF(g) ], (2a)

and carrier concentrations. ' Thus, it is important to where the quantity B (related to matrix element effects) is
establish the correct relationship between the period of defined on p. 172 of Ref. 1. The parameters AO and g are
the FKO and the field in the space-charge region.
(A'8)'=e A 6 /2p (2b)
In this paper we demonstrate rigorously that for low-
field ac modulation (6'„) in the presence of a large dc g=(E0 —E)/AB, (2c)
electric field the period of the FKO is a direct optical
measure of the dc surface electric field, Bd, . The surface where p~. is the reduced effective mass in the direction of
,

field can be due to Fermi-level pinning or can be external- 8 and Eo is the energy gaP. The Parameters G(rl) and
ly applied. This result is independent of the relation be- F(g } are electro-optic functions given by'
42 7097 1990 The American Physical Society
7098 H. SHEN AND FRED H. POLLAK 42

G(il) =m[Ai'(il)Bi'(rl) —gAi( 71)Bi(g)]+ ri'~ H( —il), 5e (E, 6'd„A'„)

(2d)
— hw
6(z))e' "dz
—AW e(E,
2itt b,
F(g) =ir[Ai' (il) —ilAi'(il)] —( ri)'—H( —il) . (2e)

In Eqs. (2d) and (2e) Ai(ii), Bi(rl), Ai'(ii), and Bi'(il) are
Airy functions and their derivatives and H( — rl) is the
+4ittbW f 0
be(E, 6(z))e' "'dz (6)

unit step function. '

By neglecting broadening effects we In Refs. 6 and 7 Bhattacharya et al. assumed that
are in the regime where I «AO, where I is a phenome- KW «1.
nological broadening parameter. From the linear relationship between 6 and z [Eq. (3)]
We now consider the case in which C„«
6d, . We re- we can write
ferred to this situation as the low-field Franz-Keldysh = W(1 —y),
z (7a)
(LFFK) criterion. Several approximations will be made.
We denote as z the distance into the SCR as measured y= 8/6"d, , (7b)
from the surface and p ( =e~ND N„~ )— as the net charge
density, where ND and N~ are the donor and acceptor where W is the width of the SCR for the case of the dc
concentrations, respectively. We assume that no surface electric field 6d, . Also we can write from Eq. (2c)
significant free carriers are created in an electromodula- s y~2/3 (7c)
tion experiment, i.e., p remains a constant. For the SCR
the abrupt junction approximation is made so that 8(z) is where
a linear function of z given by '
(7d)
6 (z) = Cd, [( W —z) / W] (3a)
(7e)
= (pleo)( W—z), 3b
In addition we define
where 6d, is the surface dc electric field, W is the width g=h'„/6' , =db W/W .
of the SCR and e0 is the static dielectric constant.
Using Eqs. (2a) and (7a)—(7f), the expression of Eq. (6b)
In the nonuniform field regime the quantity
can be rewritten as
5e(E, 6d„6„)can be expressed as
5eE'(E, e„,C„)
5e (E, bd„@„)=[ 4itt(eo/p— )(@d,)' (8;,)](J, +J,„)
(8a)
be(E, 6(z —b, W))e'
W+b, W
2lK "'dz where
0
J, (old„(, 0) =G, (gd„(, 0)+iF, (71'd„$, 0), (8b)
—
f0
b, e(E, C(z +b, W))e' "'dz, (4) J„(rid„(,0) = G.„,(old„(, 0)+ iF,„(rid„(,0) . (8c)
The terms J, (rid„$, 0) and J,„(old„(,0) represent
the
where tt(E) is the complex propagation vector of the light
contribution of the "surface" (J, ) and "average" (J,„)
in the solid. The quantities W+b, W and W — b W are
electric fields to the FKO for the unbroadened case
the widths of the SCR for the case of Cd, +8„and
— respectively. In Eq. (4) we have made use of (I =0). However, as we will show in the section below
bd, 8„, the period of the FKO from both these terms yields 8d,
the linear relation between 8 and z. In Ref. 8 the authors
as long as the modulation is in the LFFK regime. Com-
erroneously assigned the limits of this integral to be be-
tween 0 and W. Thus, they were in the regime of modu-
plete expressions for G„F„G,
„, and F,„are given in the
Appendix.
lation from flat band which is not appropriate for LFFK.
By making an appropriate change of variables in Eq. 2. Photorejlectance
(4) we can write
In photoreflectance the electromodulation is from W to
5e (E, bd„@„) W— 5 W since the photoinjected electron-hole pairs
lower the dc electric field. Thus, Eq. (1b) becomes
2ilt e' — b, 6'(z))e' "'dz
w
—AW e(E, 5e (E, Ad„@„)= b e(E, @d, ) be(E, Ad, —6— „), (9a)
5e (E, hd„@„)=( —,' )5e (E, 8d„@„). (9b)
—e ' ' w
be(E, @(z))e' 'dz
b, W
3. Results for GaAs
We now make the low-modulation field approximation
(LFFK), i.e. , vb W «
1. Neglecting terms of order In order to evaluate G„G,„, F„and F,„ for GaAs
= 0. 7
we
(b, W) or higher, Eq. (5) can now be written as use the following parameters: Vb; V,
42 GENERALIZED FRANZ-KELDYSH THEORY OF ELECTROMODULATION 7099

TABLE I. Values of W, 6&„AOd„aW, and (ao/2)W for various net carrier concentrations
IND
—N„ for GaAs.
I

W &dc A'Od,
(cm ') {A) {10 V/cm) (meV) aW (~, /2) W

1x10" 10 200 1.3 5. 1 27.5 0. 15

1x10" 3 220 4.3 11 8.7 0.071
1x10" 1 020 13 23 2.8 0.033
1 X 10' 322 43 51 0.87 0.015

6p = 1 3.4 ~ 8. 85 &( 10 @II

=0.055m p, where Vb; is
' F/cm, The extrema in the FKO are given by
the built-in surface potential due to Fermi-level pinning = P+ (4/3) [(E —Eo ) /A'8'
m rr d]
and mp is the free-electron mass. The above value of p~I is
related to transitions from the heavy-hole valence where m is the index of the mth extrema, P is an arbi-
band. ' The depletion width W and dc surface electric
field ( 6'd, ) can now be expressed as ' trary phase factor, and E is the photon energy of the
mth oscillation.
IV = [(2eo/p)( —V,„, kT/e)— ]' (loa) Plotted in Fig. 2 is the quantity
Eo)/fi0d, —
Vb;
(4/3~)[(E ] as a function of m for G,
Cd, =[(2p/eo)( Vb,
—V, „, kT/e))—'/ (lob) (open circles) and F, (closed circles) as obtained from
Figs. 1(a) and 1(b). The solid lines are least-squares fits to
p=e ND N„ I, — (10c) a linear function. All four curves yield a slope of 1.0.
%'e take This demonstrates that over a wide range of depletion
where V, „, is an externally applied voltage.
widths (see Table I) the period of the FKO of J, for g 1 «
In Table I we have listed IV, 8, ', R8, ', aIV, and (LFFK regime) yields the dc surface electric field C~, .
(ao/2)IV as a function of net-impurity concentration Plotted in Figs. 3(a) and 3(b) are G, „(solid line) and
—N„ for GaAs. The parameters a and ao are F,„(dashed line) as a function of —rid, for
IND
IND
—N„I =1X10"cm and 1X10' cm,
respective-
defined in Eqs. (A3) and (A5) in the Appendix, respective-
ly. For the latter curve the FKO are rapidly damped out.
ly.
Plotted in Fig. 1 are G, (solid line) and F, (dotted line) In Figs. 4(a) and 4(b) we display (4/3m)[(E
/ as a function of m as
Eo)/—
as a function of —rid, for )=0.01 for the cases of fiOd, ] deduced from G, „(open
IND
—N, l=lX10' cm [Fig. 1(a)] and 1X10' cm circles) and F,„(closed circles) for the two different car-
[Fig. 1(b)]. The two curves are almost identical in both rier concentrations. The solid lines are least-squares fits
period and amplitude. to a linear function. As in the case of Fig. 2 the slope is
1.0. Thus, in contrast to Refs. 8 and 10, we have demon-
strated that the "average field" contribution to the FKO
also yields the dc surface electric field and not the average
0.5
( =0.01 — Gg
--- FS

CV

Vl D
CD

o 4—
UJ
I

0— pJ 2
Q—
P sP

2
-0.5
-2 0 2 6 8
S
fl. dc Index m
FIG. 1. Theoretical
values of the unbroadened electro-optic FIG. 2. The quantity (4/3m)[(E —Eo)/A'Hd, ]
~
as a func-
functions G, and F, with (=0.
01 for the case of tion of FKO index m for G, and F, with /=0. 01 for
IND N~ = 1 X 10" cm
I
d INn —
' an— N„ = I X 10" cm I IND N„I =1X10"cm an—
d IND N„I=1X10' cm—
 H. SHEN AND FRED H . POLLAK 42

0.5 0.0i —Gs

E=O. Iq
--- FS

INp-Na I
= I x IOisc„s

'I

I
'I I
INP NA I
= I x IOI8
II
II
II
II

-O. g -0. 5 I

-2 8 -2 0 8
R.dc qS
FIG. 3. Theor
eoretical values of the unbroadened el t FIG. 5. Theor
eoretical of thee unbroadened
values o u electro-o
n or the caseof X—
d N, —N„I=Ix10" an D Nq I
= 1 &—10" cm &

dc electric field. dependence of thee envelope function

11
%e
havee also explored the si
d
Fitgures 5(a) and 5(b) d' p 1 ~ G
N
5( ) o
lhf- r —
e field, as first
gd, & 4 the erio
11a d'ifferent in re
n ig.
0 ig.
' '
e.
~ ~

r e ect is due to the 'n omo-

D A
cm, e y. Thee amplitude
respectively. of F G, for 'nn re ation to thee penetration
e depth f h
case is consider abl sm t e period of e e same
~

Fi . 5( ) 5(b) , 1(a), and 1(b). Th

~

a or
the other F G. Th e situation of Fi .
' '
1. t fi tfFKO i e1d a correct value of the dc surface
s arting to approach m d m at
re important in relation to F G, .

amped in r 1 tion t o F'ig. 1(b). T e reason for this

ore
( =O.OI —
---
Gs
Np-NaI=IxIO' cm s FS

rrrv

III ~ l
= l. 5
0.0-------
CD

x I. 5)
O
UJ V
I

E
UJ l" = 2.0
0.0-
(x5)

0 -0. 5 -8 I

-4
I I I

0 4
I I

8
0 6 IO I2
Index rn C

FIG.. 4. The quantity (4/3~

tton of FKO
d IN
'

N —1X10" I IN
un,
FIG. 6. Theo
eoretical
an,
—N„ =1x1o"
I
values of the
with
ec ro-optic
42 GENERALIZED FRANZ-KELDYSH THEORY OF ELECTROMODULATION 7101

B. Eft'ects of Lorentzian broadening EM signal to Cd, and 6„. Broadening effects also have
been considered.
The electric-field-induced change in the dielectric func-
tion e in the presence of Lorentzian broadening may be APPENDIX
obtained from the unbroadened change, 5e(E, bd„b'„)
We take into account the relation between the penetra-
[=5e(E, D)], using' tion depth of the light and the width of the space charge
region W. The complex propagation vector of the light ~
5e(E, A', r)=(1/ rr)
f '
(E E'— ) +I
dE', (12a) can be written as

where I represents the energy broadening or collision a +ik),

=(2n. /s(. )(n (Al)
frequency. A contour integral of Eq. (12a) yields 1~=a +i (a/2), (A2)
5e(E, 6, I ) = 5 (eE +i r, 6 ) . (12b) where

The broadened quantities G„F„G,„, and F,„can thus a =(2mn/A, ) (A3)

be written as and A, is the wavelength of the light, n and k are the real

G, (q'„, g, r)=(i/~) f"--„G,(q'„—

[(~'„)', o]r
)2+r' q
g,
(13a)
and imaginary parts of the complex index of refraction,
and a is the absorption coefficient.
&e now consider the specific example of the funda-
mental (direct) band gap of GaAs(EO). In the region of
" ' $, d7)', )', 0]I
F;(ri'„g, I') =(1/n. ) f-- F;(q„—
[(qd,'
q )'+r'
(13b) Eo the quantity n does not vary very much with photon
energy but there is a considerable change in a. From
Ref. 29 and taking a value of ED=1.42 eV (S731 A) we
where i =s or av and I is in units of fi9d, . find that
In order to illustrate the effects of broadening we have
plotted in Fig. 6 the values of G, (rid„g, I ) and a =3X10 cm (A4)
F, (ri'd„), I ) for ~ND N„~=1X— 10' cm for /=0. 01
For e as a function of photon energy E we use the expres-
and I =1.0, 1.5, and 2.0, in units of fi8d, =5. 1 meV (see
sion
Table I). The former quantity is represented by the solid
1/2
line while the latter is the dashed line. For I =1.0 it is E
a(E) =ao H(E EQ), — (A5)
still possible to observe about eight or nine FKO. The
effects of I are quite evident above — gz, = 8 as compared
0

to Fig. 1. Only two or three FKO can still be seen for where H(E Eo) is the u— nit step function. The quantity
I =1.5 while for I =2. 0 the FKO are completely ao can be evaluated from a(E) listed in Ref. 29. We find
damped out. In this case the electromodulation signal is
in the third-derivative functional form regime. Similar 0'0=5 X 10 cm (A6)
results have been obtained for ND N„~ = I X 10' cm- The quantity ~ W can be expressed as
for which AO&, =51 mev. The infiuence of the damping
on the envelope function of the FKO could be used to vW=aW+bW, (A7)
evaluate the broadening parameter I .
where
In conclusion we have developed a detailed generalized
theory of FKO in electromodulation which correctly b =(a /2)(%Os /E )1/2( ris )1/2H( res (AS)
takes into account of the presence of a dc surface electric
field (Ad, ) when the ac electric-field modulation is small The expressions for G„G„G„,and E,„can be written
compared to 6d, . The modulation of the dielectric func- as
tion consists of two terms which represent the contribu- ''
tions of the "surface" (J, ) and "average" ( electric J„) G, =(1/2() f — y' Me 1 c,
'dy, (A9)
fields to the FKO. Using a model numerical calculation F, =(1/2g) f — y' 'Ne ' r'dy, (A10)
in the vicinity of the direct band gap of GaAs we have 1 g

demonstrated that both terms yield FKO related to the M = G cos[2a W(1 —y)] Fsin[2a W(1 —y—
) ], (Al 1)
dc surface electric field independent of the penetration of — —
the light in relation to the width of the SCR. This result
=
N G sin[2a W(1 y)]+F cos[2a W(1 g)],
follows from the fact that Airy functions can be ex-
G, „=—2a WI4 —2b WI3, (A13)
pressed in terms of periodic sine and cosine functions.
The period of such functions is not affected by the ex- F,„=2aWI3 —2b WI4, (A14)
ponential envelope function related to the penetration 1/3~ —2b W( 1 —y)d
depth of the light. Even for relatively large modulation OX
(A15)
(/=0. 15) the first few FKO can be used to evaluate 6d, . +1/3~ —2b W(1 —y)d+ (A16)
4
Our theory also relates the magnitude of the observed
7102 H. SHEN AND FRED H. POLLAK
See, for example, M. Cardona, Modulation Spectroscopy
(Academic, New York, 1969), and references therein.
2See, for example, B. O. Seraphin, Semiconductors and Semimet-
als, edited by R. K. Willardson and A. C. Beer (Academic,
New York, 1972), Vol. 9, p. 1, and references therein.
See, for example, D. E. Aspnes, in Handbook on Serniconduc-
tors, edited by T. S. Moss (North-Holland, Amsterdam, 1980),
Vol 2, p. 109, and references therein.
~ R. A. Batchelor, A. C. Brown, and A. Hammett, Phys. Rev. B
4See, for example, F. H. Pollak, in Proceedings of the Society of
Photo-Optical Instrumentation Engineers, San Jose, 1981
(SPIE, Bellingham, 1981), Vol. 276, p. 142, and references
therein.
5P. M. Raccah, J. W. Garland, Z. Zhang, V. Lee, D. Z. Xue, L.
L. Ables, S. Ugur, and W. Wilensky, Phys. Rev. Lett. 53, 1958
(1984).
R. N. Bhattacharya, H. Shen, P. Parayanthal, F. H. Pollak, T.
Coutts, and H. Aharoni, Solar Cells 21, 371 (1987).
7R. N. Bhattacharya, H. Shen, P. Parayanthal, F. H. Pollak, T.
Coutts, and H. Aharoni, Phys. Rev. B 37, 4044 (1988); also,
Proceedings of the Society of Photo Optic-al Instrumentation
Engineers (SPIE, Bellingham, 1987), Vol. 794, p. 81.
8R. Glosser and N. Bottka, in Proceedings of the Society of
Photo-Optical Instrumentation Engineers (SPIE, Bellingham,
1987), Vol. 794, p. 88.
N. Bottka, D. K. Gaskill, R. S. Sillmon, R. Henry, and R.
Glosser, J. Electron. Mater. 17, 161 (1988).
' D. K. Gaskill, N. Bottka, and R. S. Sillmon, J Vac. Sci. Tech-
~
nol. B 6, 1497 (1988).
P. M. Raccah, J. W. Garland, S. E. Buttrill, Jr. , L. Francke,
and J. Jackson, Appl. Phys. Lett. 52, 1584 (1988).
H. Shen, S. H. Pan, Z. Hang, J. Leng, F. H. Pollak, J. M.
Woodall, and R. N. Sacks, Appl. Phys. Lett. 53, 1080 (1988).
H. Shen, F. H. Pollak, J. M. Woodall, and R. N. Sacks, J. Vac.
Sci. Technol. B 7, 804 (1989).
' C. Van Hoof, K. Deneffe, J. DeBoeck, O. J. Arent, and G.
Borghs, Appl. Phys. Lett. 54, 608 (1989).
H. Shen, F. H. Pollak, and J. M. Woodall, J. Vac, Sci. Tech-
nol. (to be published).
~eF. H. Pollak and O. J. Glembocki, in Proceedings of the So
ciety of Photo Optical Instrum-entation Engineers (SPIE, Bel-
42
lingham, 1988), Vol. 946, p. 2.
In bulk or thin-film materials low-field EM is related to the
third derivative of the optical constants. However, as dis-
cussed in Ref. 16 for isolated quantum wells EM yields a first
derivative.
' A. Frova and P. Handler, Phys. Rev. 137, A1857 (1965).
' P. Handler, Phys. Rev. 137, A1862 (1965).
41, 1401 (1990).
2'See, for example, S. M. Sze, in Physics of Semiconductor De
Uices, 2nd ed. (Wiley, New York, 1981), p. 248.
J. L. Shay, Phys. Rev. B 2, 803 (1970).
H. Hasegawa, H. Ishii, T. Sawada, T. Saitoh, S. Konishi, Y.
Liu, and H. Ohno, J. Vac. Sci. Technol. B 6, 1184 (1988).
The dominance of the heavy-hole valence transitions in FKO
was demonstrated by M. Chandreskhar and F. H. Pollak,
Phys. Rev. B 15, 2127 (1977).
Values of the conduction and heavy-hole effective masses were
taken from Numerica/ Data and Functional Relationships in
Science and Technology, Vol. 17a of Landolt-Bornstein, edited
by O. Madelung, M. Schulz, and H. Weiss (Springer, New
York, 1982).
In Refs. 8 and 10 Bottka and co-workers have erroneously
given the net carrier concentration p = e No+ i)I„~.
If there is a photoinduced voltage the expression for @~, is
given by
6~, =[(2p/e0)( V~,
—V,„, —V kT/e)]' ~'—
,
where V~ is the photoinduced voltage.
Equation (9) of Ref. 7 incorrectly stated the expression for the
extrema in the FKO as
m n =P+ (16/3)[(E E)/0%8~, ]'~', —
although the correct equation was used to interpret the exper-
imental results.
B. O. Seraphin and H. E. Bennett, in Semiconductors and
Semimetals, edited by R. K. Willardson and A. C. Beer
(Academic, New York, 1967), Vol. 3, p. 499.