tmp2439 TMP
tmp2439 TMP
tmp2439 TMP
15 AUGUST 1999-II
I. INTRODUCTION
The drive for ever higher storage capacity has led to the
development of semiconductor lasers operating in the blue
spectral region, based on ZnSe Ref. 1 and GaN.2 Along
with the large energy gap of these materials comes a large
exciton binding energy, of the same order as k B T at room
temperature. As has been well known since the 1970s,3 excitonic gain processes are important in wide-gap semiconductors, and their importance is further enhanced in
quantum-well structures where the binding energy may be
considerably larger than k B T e.g., 35 meV in
Znx Cd1x Se/ZnSe quantum wells.
Theoretical treatments of GaAs- and InP-based lasers are
well established using a microscopic many-body approach
based on linear-response theory.4 Screening and band-gap
renormalization effects are included, assuming that the injected carriers form a completely ionized electron-hole
plasma. Such treatments have been successful in explaining
many of the observed features of mid-infrared laser diodes.
Complex valence-band effects and strain effects as well as
carrier thermalization effects have all been included at various levels of complexity. In this way, a relatively complete
understanding exists for the basic operation of these lasers.
In wide-gap semiconductors, however, the strong Coulomb interaction leads to the existence of bound-exciton
states, which persist even at elevated densities and temperatures. As such, the conventional assumption that the inversion is in the form of an electron-hole plasma with no excitons present deserves closer examination. A self-consistent
description where both bound and unbound states are treated
on an equal footing is required. Unfortunately, as far as we
are aware no comprehensive theoretical treatment of this
problem exists. Treatments based on bosonic exciton operators have been proposed,5 but this approach breaks down at
high injection when the screening of the Coulomb potential
weakens the binding and produces a population of unbound
scattering states, which clearly do not exhibit bosonic character. On the other hand, a treatment based around fermionic
electron and hole operators is complex when higher-order
excitonic correlations are important.6,7 A natural concept in
considering this issue is the degree of ionization in the inter0163-1829/99/608/557012/$15.00
PRB 60
PRB 60
system in excited semiconductors,13 we divide the total electron hole density between two terms:
.
n a n 0a n corr
a
n 0a
n 0e
ne
n 0e
n 0e n corr
e
n corr
a
n 0a n 0b
22
Z ab ,
ab
m,
exp E m,
exp
2k 2
dk,
2 eh
d m k
dk
5571
2 1 m 0
m
m k
dk
exp
2k 2
dk.
Ma
6
Here we assume that both electron and hole states in quantum wells are two-fold degenerate. The only difference between Z hh and Z ee arises from the difference between electron and hole effective masses.
Equations 26 provide a connection between the total
electron-hole density n and uncorrelated quasiparticle densities n 0e and n 0h . The quasiparticle densities in turn govern the
screening13 and therefore the strength of interaction between
particles, which uniquely defines the set of binding energies
and scattering phase shifts which enter Eqs. 5 and 6 for
the two-body partition functions. These partition functions in
via Eq. 4. Thus,
turn define the ratio between n 0a and n corr
a
to find the degree of ionization of the electron-hole plasma,
one must solve the system of equations 26 selfconsistently, together with a reasonable model of the
screened interaction.
In Sec. II we discuss the statically screened Coulomb potential which we use to model the interaction between particles in an exciton/electron-hole plasma, and present results
from the application of the variable phase method22 to scattering and bound states in this potential. In Sec. III we
present and discuss the results of calculations of partition
functions and the degree of ionization of the electron-hole
plasma. In Appendix A we derive the basic equations of the
variable phase method, which is used for calculation of scattering phase shifts and binding energies. The 2D analog of
the Beth-Uhlenbeck formula and the modified law of mass
action are derived in Appendix B.
II. STATICALLY SCREENED COULOMB POTENTIAL
There is an extensive literature dealing with different aspects of the screened Coulomb interaction in 2D systems.23
In this paper we model this interaction by the well-known
Thomas-Fermi expression for a statically screened Coulomb
potential,17
V s
e2
qJ
0 q
qq s
dq
e2 1
q H q N 0 q s ,
2 s 0 s
5572
PRB 60
potentials both signs of Eq. 7 with the screening wavenumber q s 0.2/a * , where a * 2 /( e 2 ) is the 3D exciton
Bohr radius. The scattering phase shifts are negative for the
repulsive potential and positive for the attractive potential.
For the repulsive potential all zero-energy phase shifts vanish, m (k0)0 for all angular momenta m. For the attractive potential,
lim m k m ,
k0
1
2 m 1 2 m
, 1,2, . . . . 9
q sa * c
2
m max 0, 0 2 m ,
10
where
8/ q s a * 11
2
11
N b 0 2
m1
m ,
12
8/ q s a * 1 1
0 1
.
4
2
13
PRB 60
N b 0 2 m max 0 m max1
0 2
0 1
2
0 1
0 1
2
14
For small q s a * weakly screened potential a simple approximate expression for the total number of bound states
follows from substitution of Eq. 11 into Eq. 14:
N b
20
1
2 q sa *
15
5573
2 /(2 eh a * 2 ) , and we use the same classification of energy levels as in Ref. 32, i.e., each energy level is characterized by the angular momentum quantum number m and another number which numerates different bound states for a
given m, with ( 1) being the number of nonzero nodes of
the radial wave function. For m0 (s states the calculated
energies are consistent with those obtained by J. Lee et al.29
using a variational method.
III. PARTITION FUNCTIONS AND IONIZATION DEGREE
Before we present the results of calculations of the partition functions and the ionization degree, we would like to
discuss an important consequence of Levinsons theorem for
the statistical mechanics of the 2D gas with an attractive
interaction between its particles. The bound-state sum
Z bound m, exp(Em,), entering the two-body partition
function in Eq. 5, exhibits jumps whenever bound states
disappear with increasing screening. We will now show that
these jumps do not give rise to unphysical discontinuities in
the partition function if the scattering states are properly
taken into account.
Integrating by parts the scattering term and using
Levinsons theorem in the form of Eq. 8, we can rewrite
Eq. 5 as
Z eh
exp E m, 1 2
m k
q T 0 m
m,
2
exp k 2 /q T2 kdk,
16
5574
PRB 60
Ry* 0 2
22
n 0h n 0e 4
n a * n 0h a * 2 .
eh k B T
k BT e
17
Ry*
na * 2 .
k BT
18
Equation 18 shows clearly the connection between the dimensionless screening parameter q s a * and the two main dimensionless parameters characterizing the 2D electron-hole
plasma, namely, the dimensionless density na * 2 and temperature k B T/Ry* . In addition, the role of the degree of ionization, , introduced by Eq. 3 becomes more transparent.
The parameter enters Eq. 18 explicitly, governing the
PRB 60
FIG. 6. The degree of ionization solid lines of the nondegenerate 2D electron-hole plasma as a function of the total electron
density at room temperature, calculated for the model semiconductor with the effective Bohr radius and excitonic Rydberg of a
ZnSe, and b GaAs. The arrows indicate n2/ 2M e for ZnSe and
GaAs at room temperature. Dashed lines show the degree of ionization calculated using a simple law of mass action with a single
bound state.
Ry*
Z Z ee ,
k B T eh
19
and using Eqs. 3 and 18 we obtain the following expression for the degree of ionization :
q sa *
Z eh Z ee
2
20
5575
5576
In this appendix we derive the basic equations of the variable phase approach in two dimensions from the radial
Schrodinger equation. This derivation is similar to that in
three dimensions.22
PRB 60
1. Scattering phase shifts
2 1
1 2
2
V E .
2 ab 2
A1
m , R m e im ,
m0,1,2, . . . .
A2
1
m2
R m k 2 U 2 R m 0,
Rm
A3
Am
2
k
1/2
cos k 2m1 /4 m ,
A4
A m J m k cos m N m k sin m ,
Rm
A6
where the prime indicates differentiation with respect to .
The phase function m ( ) has a natural physical interpretation as being the phase shift produced by a potential cut off
at a distance .
Differentiating Eq. A6 and substituting the resulting expression, together with Eqs. A5 and A6, into Eq. A3, we
obtain
PRB 60
[J m k cos m N m k sin m ] m A m
Am
k sin m N m k cos m
Jm
U A m J m k cos m
N m k sin m ].
A7
J m k cos m N m k sin m
Am
A m J m k sin m N m k cos m .
m
A8
d
d
N m x N m x J m x
dx
dx
2
,
x
d
U J m k cos m
d m
2
N m k sin m .
2
A10
m 0 0,
A11
m lim m .
2m1
m!
U 2m1 d ,
0.
A13
U d 0,
A14
and
W J m x ,N m x J m x
k 2m
1
m
F m k 2 2 F m 0.
Fm
5577
A12
2 U 0, 0.
A15
d
m , U I m cos m ,
d
2
2
2
K m sin m , ,
A16
m , 1/2 , 1,2, . . . .
A17
5578
N a ,N b
QN a ,N b z a a z b b ,
B1
QN a ,N b A,T Tr e H (N a ,N b ) ,
PRB 60
Q1,1 A,T g a g b
g a g b
2M M
a
b
Tr e H rel
A
2M M
a
b
d 2
n n 2 e E ,
n
B2
B12
ln z a ,z b ,A,T
ACl a ,l b z aa z bb .
l a ,l b
B3
(0)
Q 1,1
A,T g a g b
d
1
l l
ln
l a Cl a ,l b z aa z bb .
n a z a
A dz a
l a ,l b
B4
Z ab
B6
B7
and
1 2
C2,0 Q2,0 Q 1,0
2
d 2
2 E
n (0)
n e
(0)
n
B13
where the superscript 0 refers to quantities of the noninteracting system. The two-body interaction part of the partition
function is then defined by
B5
C1,0Q1,0 /A,
A
2M M
a
b
d 2
2 E
n n 2 e E (0)
n e
n
n e E e E
n
(0)
n
(0)
n
B14
Thus
(0)
Q1,1 A,T Q 1,1
A,T g a g b
A
2M M
a
b
Z ab .
B8
B15
Q1,0 A,T g a
A
2 2
d 2 k exp
A.
2k 2
A
g a 2 ,
2M a
M
a
B9
22
.
Ma
B10
E (0)
n
(0)
Q 1,1
A,T g a g b
g a g b
ga
2M
.
a
B11
B16
This yields
C1,0
2k 2
,
2 ab
A
2M
a M b
A
2 2
2M M
a
b
2
ab
Q1,0 Q0,1 .
d 2 k exp
2k 2
2 ab
ga
2M
a
gb
2M
B17
PRB 60
Z ab
B e E 0 dk g k g (0) k exp
2k 2
.
2 ab
B19
1
kR 2m1 /4 m k n
2
B21
kR 2m1 /4 n
B22
,
R d m k /dk
k (0) .
R
B23
B24
B25
(0)
gm
k k1
B26
1
d m k
R
,
dk
B28
Therefore,
1 d m k
.
dk
(0)
g m k g m
k
B27
B29
d m k
.
dk
B30
B e
EB
exp
d m k
dk
2k 2
dk,
2 ab
B31
1
2M
a M b
Z ab 2
ga
ab
gb
2M 2M
a
Z ab .
B32
or
g m k
1
R.
(0)
gm
k
B18
5579
where Z aa is defined by
A
2M
Z aa ,
a
B33
5580
Z aa
m,
g a 1 m E
m,
e
ga
g a 1 m
ga
and
m k
dk
Ia
2k 2
exp
dk,
Ma
B34
Here and in what follows the upper sign is for bosons and the
lower sign is for fermions. For the case of repulsive fermions
which do not form bound states and for g a 2, Eq. B34
reduces to Eq. 6.
Let us now calculate the two-body partition function of
(0)
, which is given by
noninteracting bosons or fermions, Q 2,0
(0)
Q 2,0
A,T
A
2
2M
a
ga
2
d 2
B35
where s(0) (k, ) and (0)
a (k, ) are symmetric and antisymmetric eigenfunctions of the noninteracting Hamiltonian of
the relative motion, i.e.,
2
cos k
A
2
sin k .
A
B37
k
(0)
Q 2,0
1 A
2 2 2
A
1
ga 2
2
M
A
2M
where
Is
1
2 2
d 2 d 2 k cos2 k exp
A
1
1
2 2 2M
4
a
2k 2
Ma
n a
B42
Z aa
1 ga
.
4 2M
B43
ga
2M
z a
a
ga
1 ga 2
z a 2
2
aa
2 M
2M
a
ab
ga
gb
2M
a
2M
Z ab z a z b .
Z aa z 2a
B44
The first two terms in Eq. B44 do not depend on interparticle interaction however, the second term depends via its
sign on the statistics of the particles. It is natural to call the
sum of two first terms in Eq. B44 an uncorrelated density;
we denote it as n 0a . Note that these terms are the first two
terms in the low-density expansion of the well-known
expression25 for the density of the noninteracting 2D Bose or
Fermi gases:
ga
2M
ln 1z a .
a
The sum of the last two terms in Eq. B44 is the interaction.
dependent correlated density n corr
a
In the low-density limit the second term in Eq. B44 is
much smaller than the first one, and within the same accuracy as Eq. B44 we can write
b n 0a n 0b 2
ab
Z ab .
B45
A
A
1
1 2 1
g a 2 Q 1,0
ga 2 .
4 M
2
4 M
n corr
a
B40
n 0a
g a 1 Is g a 1 Ia , B39
B41
ga
1
C2,0 2
2 aa 2M
can be written as
(0)
Q 2,0
A,T g a
B38
d 2 k.
2k 2
Ma
B36
(0)
a k,
d 2 d 2 k sin2 k exp
A
1
1
.
2
2 2 M
4
(0)
Q 2,0
A,T
and
Then
1
2 2
Finally,
k g a 1 s(0) k, 2
2k 2
2
g a 1 (0)
,
a k, exp
Ma
s(0) k,
PRB 60
PRB 60
H. Haug and S.W. Koch, Phys. Status Solidi B 82, 531 1977;
S.W. Koch, H. Haug, G. Schmeider, W. Bohnert, and C. Klingshirn, ibid. 89, 431 1978.
4
W.W. Chow, S.W. Koch, and M. Sargent III, SemiconductorLaser Physics Springer, Berlin, 1994.
5
E. Hanamura and H. Haug, Phys. Rep. 33C, 209 1977.
6
M. Lindberg, Y.Z. Hu, R. Binder, and S.W. Koch, Phys. Rev. B
50, 18 060 1994.
7
M.F. Pereira, Jr. and K. Henneberger, Phys. Rev. B 58, 2064
1998.
8
T.M. Rice, Solid State Phys. 32, 1 1977.
9
S. Schmitt-Rink, D.S. Chemla, and D.A.B. Miller, Adv. Phys. 38,
89 1989.
10
J. Ding, M. Hagerott, T. Ishihara, H. Jeon, and A.V. Nurmikko,
Phys. Rev. B 47, 10 528 1993.
11
M. Schmidt, G. Ropke, and H. Schultz, Ann. Phys. N.Y. 202, 57
1990; G. Ropke, A. Schnell, P. Schuck, and P. Nozie`res, Phys.
Rev. Lett. 80, 3177 1998.
12
W.-D. Kraeft, D. Kremp, W. Ebeling, and G. Ropke, Quantum
Statistics of Charged Particle Systems Akademie-Verlag, Berlin, 1986.
13
R. Zimmermann and H. Stolz, Phys. Status Solidi B 131, 151
1985; R. Zimmermann, Many-Particle Theory of Highly Excited Semiconductors Teubner, Leipzig, 1987.
14
I. Galbraith, in Microscopic Theory of Semiconductors: Quantum
Kinetics, Confinement and Lasers, edited by S. W. Koch World
Scientific, Singapore, 1996, p. 211, and references therein.
15
R. Cingolani, L. Calcagnile, G. Col, R. Rinaldi, M. Lomascolo,
M. DiDio, A. Franciosi, L. Vanzetti, G.C. LaRocca, and D.
Campi, J. Opt. Soc. Am. B 13, 1268 1996; R. Cingolani, G.
Col, L. Calcagnile, R. Rinaldi, A.L. Convertino, M. Lomascolo,
M. DiDio, I. Suemune, Phys. Rev. B 54, 17 812 1996.
16
N. Ben Brahim Aouani, L. Mandhour, R. Bennaceur, S. Jasiri, T.
Amand, and X. Marie, Solid State Commun. 108, 199 1998.
17
F. Stern and W.E. Howard, Phys. Rev. 163, 816 1967.
18
L.D. Landau and E.M. Lifshitz, Quantum Mechanics (NonRelativistic Theory) Butterworth-Heinemann, Oxford, 1997, p.
508.
19
E. Beth and G.E. Uhlenbeck, Physica Amsterdam 4, 915 1937;
see also, e.g., L.D. Landau and E.M. Lifshitz, Statistical Physics
Pergamon, New York, 1980, Pt. 1, p. 236.
20
M.E. Portnoi and I. Galbraith, Phys. Rev. B 58, 3963 1998.
21
P. Vashishta and R.K. Kalia, Phys. Rev. B 25, 6492 1982; X.
Zhu, M.S. Hybertsen, and P.B. Littlewood, ibid. 54, 13 575
1996.
22
F. Calogero, Variable Phase Approach to Potential Scattering
Academic, New York, 1967.
23
T. Ando, A.B. Fowler, and F. Stern, Rev. Mod. Phys. 54, 437
1982; for a more recent development see, e.g., C. Bulutay and
M. Tomak, Phys. Rev. B 54, 14 643 1996; K. Schrufer, C.
Metzner, M.Ch. Hofmann, and G.H. Dohler, Superlattices Mi3
5581