RKKY interaction in semiconductors: Effects of magnetic field and screening
C. H. Ziener, S. Glutsch, and F. Bechstedt
Friedrich-Schiller-Universität Jena, Institut für Festkörpertheorie und Theoretische Optik, Max-Wien-Platz 1, 07743 Jena, Germany
(Received 15 April 2003; revised manuscript received 5 March 2004; published 24 August 2004)
We derive the Ruderman-Kittel-Kasuya-Yosida (RKKY) theory for semiconductors including band degen-
eracy, modulation of the band edges, and external fields. Explicit expressions are given for a three- and
two-dimensional semiconductor in a magnetic field. Screening effects are included by calculating the density
correlation function. We show that the RKKY theory with screening is equivalent to the mean-field Zener
model.
DOI: 10.1103/PhysRevB.70.075205 PACS number(s): 75.30.Hx, 75.50.Pp, 75.75.+a
I. INTRODUCTION
Diluted magnetic semiconductors (DMS) have been ex-
tensively studied for more than a decade.
1,2
The exchange
coupling between magnetic ions and electrons leads to a gi-
ant effective g factor. In recent years, ferromagnetism in Mn-
doped III-V and II-VI semiconductors has been predicted
theoretically
3–7
and measured experimentally;
8–14
for a re-
view, see Ref. 15. In low-dimensional semiconductors, fer-
romagnetism can be controlled through the modulation of the
band edges, external fields, and carrier injection by optical
pulses.
7,12,14
The goal is the development of semiconductor
devices which use the spin degree of freedom for the storage
and processing of information, known as spintronics.
16,17
The indirect exchange interaction of magnetic ions by
electrons has been explained by the Ruderman-Kittel-
Kasuya-Yosida (RKKY) theory, developed in the 1950s,
18–20
and by a mean-field Zener theory, which goes back to papers
by Stoner from the 1930s.
21
The range function of the RKKY
interaction has been calculated already in the early papers for
interaction-free electrons in three, two, and one dimension,
both in momentum
22
and in real space.
22–25
Dietl et al.,
3
as-
suming only the lowest subband to be occupied, expressed
the RKKY interaction of a low-dimensional semiconductor
in terms of the range function of the ideal three-, two-, and
one-dimensional electron gas.
However, realistic low-dimensional semiconductors are
never perfectly two or one dimensional. For example, quan-
tum wells and superlattices have an effective dimension be-
tween two and three, quantum wires between one and two.
As the translational symmetry is broken in the directions of
confinement, the range function does no longer depend only
on the difference of the coordinates. It is quite common to
treat Bloch electrons like free electrons and few papers pay
attention to the details of the band structure.
4,26–28
Moreover, the magnetic field itself leads to a quantization
of the electron and hole motion and to reduction of the di-
mensionality. For example, a bulk semiconductor in a mag-
netic field has features both of a three- and of a one-
dimensional semiconductor. To describe experiments in high
magnetic fields, it is no longer justified to assume the RKKY
interaction to be the same as in the field-free case and the
influence of the magnetic field on the range function needs to
be taken into account.
An open problem is the effect of screening. It was argued
on the basis of the diagram technique that screening plays no
role in the absence of Zeeman splitting.
29
In semiconductors,
Zeeman splitting is usually not negligible. The spin suscep-
tibility, including screening, was calculated already in the
late sixties,
30
but these results did not make their way into
the theory of ferromagnetism in semiconductors. It is also
not clear if the perturbation-theoretical character of the
RKKY theory limits its validity and in which way RKKY
theory and mean-field Zener theory are related to each other.
In this paper, we present a rigorous derivation of the
RKKY theory for semiconductors. First, in Sec. II, we derive
the RKKY interaction by means of both perturbation theory
and Kubo theory and show the equivalence. Then, in Sec. III,
we derive the effective RKKY Hamiltonian for low-
dimensional semiconductors, including the sample geometry,
band degeneracy, and external fields. To illustrate the useful-
ness of the theory, in Sec. IV we derive analytical results for
a bulk semiconductor and an ideally two-dimensional semi-
conductor in a magnetic field. The results are compared with
the range functions for the three-, two-, and one-dimensional
electron gas, and limiting cases are studied. In Sec. V, we
calculate the range function in the presence of screening and
show the equivalence of RKKY theory including screening
and mean-field Zener theory. Summary and conclusions are
given in Sec. VI.
II. THE RKKY INTERACTION
The derivation of the effective Hamiltonian by second-
order perturbation theory in the original paper by Ruderman
and Kittel is somewhat hand waving.
18
An alternative ap-
proach to the effective interaction is based upon the magnetic
susceptibility, which was pioneered by Wolff.
29,31
Here we
give a rigorous derivation of the effective Hamiltonian by
Löwdin’s perturbation theory and by linear response theory,
which gives the same result. The effective interaction is re-
lated to the density correlation function, which allows us to
systematically study the role of many-particle effects.
A. Degenerate perturbation theory
We start with Löwdin’s degenerate perturbation theory.
Suppose the Hamiltonian of a system is of the form
PHYSICAL REVIEW B 70, 075205 (2004)
1098-0121/2004/70(7)/075205(14)/$22.50 ©2004 The American Physical Society 70 075205-1