Академический Документы
Профессиональный Документы
Культура Документы
semiconductors
This content has been downloaded from IOPscience. Please scroll down to see the full text.
(http://iopscience.iop.org/0268-1242/5/3S/045)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 170.140.26.180
This content was downloaded on 24/08/2015 at 01:02
INVITED PAPER
0 A Pankratov
P N Lebedev Physical Institute, USSR Academy of Sciences, Leninsky Prospect 53,
117924 Moscow, USSR
1 L+
where for the band-inverted heterojunctionf(z) changes
its sign. Assumingf( - CO)< 0 andf( + CO)> 0, we put
the coordinate origin in such a way thatf(0) = 0 (figure
L- i C c
1).
Under the constraint (3), the transformation
2 1 t-
L
I converts (2) into a pair of decoupled equations:
Figure 1. Energy diagram of a band-inverted
heterojunction made of Pb, - .Sn,Te(Se). Labels L F mark
odd and even energy bands.
(5)
by Pb, -,Ge,Te, which reveals a ferroelectric phasetran- where
sition. The antiferromagnetic system can be realised by
doping with magnetic impurities, e.g. Mn. Such a system
is of principal importance because it allows one to model
the effect of a parity violation in quantum electrodynam-
Equation (5) is thecommonSchrodingerequation
ics (QED) in (2 + 1) dimensions (two space plus onetime)
with a specific potential, which is a linear combinationof
[12]. The field-theoretical viewpoint on the band-
the square of the derivative of the same function W,(z). It
inverted system was used by Fradkin et a1 [13], although
possesses a special symmetry and represents the known
theprinciple role of themagneticimpurities was not
Wittensupersymmetric quantum mechanics [14]. The
mentioned.
supersymmetric potential permits the factorisationof ( 5 ) :
S205
0 A Pankratov
theboundstatesdisappearandthesupersymmetry is
said to be broken. It is easy to verify that atthese energies
the conic interface spectrum (9) touches the L- bands where L, = (ch/lel H)112is the magnetic length. Formula
(taken at zero momentum p,) of bulk materials on both (14) follows from (9) after a substitutionpI + (huL/LH) x
sides of the junction (figure 2). (271). To estimate the splitting it is instructive to note
From (9) and (1 1) we find that the number of bound that ~ u J L H= ( ~ ~ h 0 , / 2 ) ~where
/ ~ , cg and W, arethe
states per unit area is given by fundamental gap and cyclotron frequency of the bulk
material either to the left or to the right. Since usually
N = (A;/pi - l)(Az( + CO)+ A2( - ~0))/4nh~uf.(12) h o c , the splitting (14) is much larger than that of the
Assuming the factor Ai/& - 1 is of the order of unity, we band states. In a field H = 1 T the bulk fundamental gap
get an estimate of N Y 10l2 cm-2 for typical values E* ?: 0.1 eV is overlapped by only five Landau levels. As a
2: 0.1 eV and uI 2: 2.2 x lo7 cm S - . result of thegiantsplitting,onecan realise at finite
From (9) and (12) it follows that bound states exist temperature the quantum regime for interface states and
p
,
only when I I < I A, 1. Otherwise, the superpotential ( 6 ) the classical region for band states. Then one can expect
does not change sign and the supersymmetry is broken at that quantum oscillations of the longitudinal magnetor-
any energy. This restriction implies that in Pb,-,Sn,Te esistance due to resonant scattering by interface Landau
alloy the L- and L bands move towardseachother levels will not be masked by band effects.
with increasing x. In a weak field (classical limit T 9 hv,/L,), using the
Experimentally, the value of p
, for Pb, -,Sn,Te has spectrum (14) we get thediamagnetic susceptibility of
still not been uniquely determined. According to Pichler interface electrons [4] :
et al [16], I p o l < IAoi (type I band diagram), although
Murase et al [17], supported by a recent paper [18],
insist on type 11. However, Pb, -,Sn,Se experiments [l91
favour the type I diagram. where the chemical potential p is counted from the conic
For Hg,-,Cd,Te thetypeIbanddiagram iswell point. The temperature dependence x ( T ) is non-mono-
established, although there is some controversy regard- tonic and shows a maximum at T,,, ?: 0.641p i .
ing the value of the valence band offset [20, 211. When Circular polarised light propagating parallel to the
pL = 0 thecorresponding EFA equation for the Kane magnetic field along the z axis (Faraday configuration)
S206
Supersymmetry in narrow-gap semiconductors
) - PsgnH
~ ( 0= (19)
where E? is the EFA Hamiltonian, the cps determine the ~ ( n )= f [ P 2 + 2 4 1 - ~ ~ / d ~ ) h ~ : / L ~ ] ~ .
z-dependent part of the bulk and interface wavefunctions,
and ~ ~ , , ( nis) the energy of the nth Landau sub-band. An important feature of (19) is that the zero level ~ ( 0 )
crosses the gap 2P when H is switched from left to right.
If this levelis populated by electrons, this switching
4. Magnetic impurities in a band-inverted heterojunc- requires some work. This means that the interface elec-
tion: antiferromagnetic domain walls trons possess a magnetic moment. At finite temperature
this moment per unit area is given by [l21
Does a possibility exist to create a gap in an interface
electron spectrum (9)? The degeneracy in a conic pointis
ensured by the time-reversal symmetry and can be re-
moved only by magnetic impurities. To provide a gap we
where 4,, = hc/l e I is a fluxoid.
need the off-diagonal interband matrix element in the
EFA Hamiltonian (1). Symmetry analysis [123 shows that
M is directed along z and is produced by interface
it can be produced by an exchange interaction with an electrons not by magnetic impurities. It creates an inter-
odd-spin density (because of the opposite parity of L* nal magnetic field which can evoke a Hall current in the
bands).Homogeneouslydistributedantiferromagneti- absence of an external field H . This current is identical to
cally ordered impurities will provide the imaginary term the so-called abnormal vacuumcurrent discovered in
iP = constant in theHamiltonian of aband-inverted QED C221 in (2 + 1) dimensions. Instability of the (2 1) +
QED vacuumstate with respect to thegeneration of a
junction
magnetic field was also found [23]. However, in (2 1) +
H=(
A(z) + cp(z) vu-p + iP QED the magnetic moment is not an observable (it is a
HC -A@> + vacuum property). In our case we have a third dimen-
sion, which allows us to measure its deflection from the z
and instead of a linear interface spectrum (9) we get a
axis, and M becomes an observable.
massive spectrum
+
~ ~ ~= ( p [P~ ) u:pl(l - cp&Ai)]2. (18)
Another way to obtain an analogous interface spec- 5. Ferroelectric domain walls
trum in the case A(z) = constant is to consider an inho-
mogeneous configuration of an exchange field P(z) like The thirdexample of a supersymmetric heterostructure is
that of figure 3. It is important that, similar to We(z), the a domain wall in semiconductor ferroelectrics such as
field P(z)shouldchangeits sign. Suchasoliton-like Pb, -.Ge,Te. Below the Curie point the polarisationU is
configuration is indeed an antiferromagnetic domain directed along one of the [11l] axes. Symmetry analysis
wall. In this case thenon-degenerateinterfacestates [ 12,241 shows that via the optical deformation potential
the polarisation U produces an off-diagonal perturbation
in the k - p Hamiltonian:
H=(A
A vu*jj- i u - u
HC -A
(an isotropic form is given for simplicity). We will discuss
the simplest case of oppositely directed domains, when
the domain wall is perpendicular to U. In this case the
term iu U = ia,u(z) changes its sign as shown in figure 3.
The loss of an inversion centre lifts the spin degenera-
cy in a homogeneous ferroelectric, leading to the spin-
split bulk bands (figure 4)
Figure 3. Soliton configuration of an exchange field-the
antiferromagnetic domain wall. The same configuration of
an electric polarisation corresponds to oppositely directed
k &*(pZ, pI) = f [A2 + ufipf + (U f uIpJ 23 112 .
domains in a semiconductor ferroelectric. (22)
S207
0 A Pankratov
1 2 3 4 5
H, / H
S208
Supersymmetry in narrow-gap semiconductors
[S] Pakratov 0 A, Pakhomov S V and Volkov B A 1987 H, von Ortenberg M and Kriechbaum M 1985
Solid State Commun. 61 93-6 Superlatt. Microstruct. 1 1-9
[S] Chang Yia-Chung, Schulman S W, Bastard G, Guldner [l71 Murase K, Shimomura S and Takaoka 1985 Superlatt.
Y and Voos M 1985 Phys. Rev. B 31 2257-560 Microstruct. 1 177-82
[7] Lin-Liu Y R and Sham L J 1985 Phys. Rev. B 32 [l81 Nishijima Y 1989 J. Appl. Phys. 65 935-40
5561-3 [l91 Shotov A and Sjelivanov Y 1990 Semicond. Sci. Technol.
[8] Cade N A 1985 J . Phys. C; Solid State Phys. 18 5 S27
5135-41 [20] Tran Minh Duc, Hsu C and Faurie J P 1987 Phys. Rez;.
[S] Korenman V and Drew H D 1987 Phys. Rev. B 35 Lett. 58 1127-30
6446-9 C213 Shin C K and Spicer W E 1987 Phys. Rev. Lett. 58
[lo] Agassi D and Korenman V 1988 Phys. Rev. B 37 2594-7
10095-106 C221 Jackiw R 1984 Phys. Rev. D 29 2375-7
[11] Volkov B A and Pankratov 0 A 1986 Pisma Zh. Eksp. C231 Cea P 1985 Phys. Rev. D 32 2785-93
Teor. Fiz. 43 99-101 (Engl. transl. 1986 JETP Lett. C241 Bangert E, Bauer G, Fantner E J and Pascher H 1985
43 130-3) Phys. Rev. B 31 7958-78
[l21 Pankratov 0 A 1987 Phys. Lett. A 121 360-6 C251 Faurie J P, Woo K C and Rafol S 1986 Proc. Znt.
[l31 Fradkin E, Dagotto E and Boyanovsky D 1986 Phys. Winter School (Mauntendorf, Austria), Springer
Rev. Lett. 57 2967-70 Series in Solid State Sciences vol. 67, ed. G Bauer, F
[l41 Witten E 1981 Nucl. Phys. B 188 513-54 Kuchar and H Heinrich (Berlin: Springer) pp 24-32
[l51 Gendenshtein L E and Krive I V 1985 Usp. Fiz. Nauk [26] Clemens H, Ofner P, Krenn H and Bauer G 1987 J .
146 553-90 Cryst. Growth 84 571-6
[l61 Pichler P, Fantner E J, Bauer G, Clemens H, Pascher
S209