Download as pdf or txt
Download as pdf or txt
You are on page 1of 40

Review Article

phys. stat. sol. (b) 200, 311 (1997)


Subject classification: 61.66.Fn; 71.20.Nr; 71.70.Gm; 72.20.My; 75.50.Pp; S1.62; S7

Semimagnetic Semiconductors
Based on II±V Compounds
J. Cisowski


Institute of Physics, Silesian Technical University, Katowice, ul. Krasinskiego 8,
PL-40-019 Katowice, Poland and
Department of Solid State Physics, Polish Academy of Sciences, ul. Wandy 3, PL-41-800
Zabrze, Poland

(Received July 15, 1996; in revised form December 18, 1996)

Contents
1. Introduction
2. Preparation and crystal structure
2.1 Mn-alloyed II3 ±V2 compounds
2.2 Other semimagnetic semiconductors based on II±V compounds

3. Band structure
3.1 The Bodnar model
3.2 sp±d exchange interaction
3.3 Intraionic Mn transitions

4. Transport properties
4.1 The Shubnikov-de Haas effect
4.2 Classical transport

5. Magnetic properties
5.1 Magnetization and susceptibility
5.2 d±d interaction and generalized pair approximation
5.3 Magnetic freezing

6. Conclusions
References

1. Introduction
Semimagnetic semiconductors SMSCs or, alternatively, diluted magnetic semiconductors
DMSs are semiconducting compounds in which a part of cations is replaced by transi-
tion metal or rare earth elements with partially filled d- and f-shells, respectively. The
presence of such ions randomly distributed in the host lattice gives rise to two kinds of

21*
312 J. Cisowski

interactions: (i) the sp±d (sp±f) exchange interaction between the band carriers and
localized magnetic moments of these ions and (ii) the d±d (f±f) interaction between the
ions themselves.
Among SMSCs, the most known are Mn-alloyed II±VI compounds, studied during the
last two decades [1]. In recent years, the number of SMSCs has been further augmented
by adding to II±VI compounds Fe, Co, Cr and rare earth elements as well as by using
IV±VI compounds as host semiconductors; all these materials are extensively presented
in books [2 to 5] and recent review articles [6 to 11]. Quite recently, using the molecular
beam epitaxy (MBE) technique, also Mn-alloyed III±V compounds have been success-
fully grown [12].
In view of this, SMSCs based on II±V compounds are much less known in spite of
their extremely attractive magnetic and transport properties; first of all, it should be
emphasized that these materials are characterized by the smallest cation±cation distance
among the known SMSCs (0.29 nm in (Zn1 ÿ x Mnx )3 As2 as compared with 0.38 nm ±± a
minimum for others than II±V SMSCs) which causes that both types of the exchange
interactions, mentioned above, are much stronger in II±V than in other SMSCs.
Since all the valuable results obtained for II±V SMSCs are communicated in many
original papers dispersed over various physical journals and, to our knowledge, have
been never reviewed up to now (some data on II±V SMSCs, including the values of the
sp±d and d±d exchange interaction constants which have to be revised, are presented in
the reviews devoted to other groups of SMSCs [3, 7, 8] or on the occasion of description
of ordinary II±V semiconducting compounds [13]), the aim of this paper is to summarize
the most important and reliable results obtained by various workers in this field. At the
very beginning, we briefly describe the family of ordinary II±V compounds (i.e. without
magnetic ions) which serve as host matrices for II±V SMSCs.
As is known, the elements of the II and V columns of the Periodic Table form various
chemical compounds which exhibit very interesting semiconducting properties. From the
point of view of the chemical formula, all these compounds can be divided into several
groups such as II…1† ±V…1† , II±V2 , II±V4 and II3 ±V2 which are extensively described in
the review papers [13 to 17]. It is worthy to add that the compounds of a particular
group form many ternary and quaternary alloys and their properties are, in general,
intermediate between those of the end binary compounds.
Among all the II±V compounds, the II3 ±V2 group is best known and three members
of this group, namely Cd3 As2 , Zn3 As2 and Zn3 P2 have been (in that order) successfully
alloyed with Mn to form ternary SMSCs, i.e. (Cd1 ÿ x Mnx )3 As2 (CMA) [18 to 21],
(Zn1 ÿ x Mnx )3 As2 (ZMA) [22 to 25] and (Zn1 ÿ x Mnx )3 P2 (ZMP) [26] as well as the qua-
ternary material (Cd1 ÿ x ÿ y Zny Mnx )3 As2 (CZMA) [27 to 29]; quite recently, a new II3 ±V2
SMSC but alloyed with Fe, namely (Zn1 ÿ x Fex )3 As2 (ZFA), has been also reported [30].
Apart from II3 ±V2 SMSCs, there are also other, much less known, II±V compounds
containing magnetic ions, such as CdP2 doped with Mn [31] and CdFez Sb1 ÿ z [32] in
which, contrary to other II±V SMSCs, Fe replaces Sb being the anion in CdSb.
Since the properties of the binary host materials influence the properties of the corre-
sponding SMSCs, in Table 1 we have gathered some fundamental data of those II±V
semiconducting compounds which have been so far alloyed with magnetic ions.
As can be seen from Table 1, Zn3 As2 and Zn3 P2 are ordinary wide-gap semiconduc-
tors with the direct energy gap (at the G-point of the Brillouin zone) being the difference
between the conduction and valence bands originating from a nondegenerate s-like atom-
Semimagnetic Semiconductors Based on II±V Compounds 313

Table 1
Some important physical properties of II±V compounds [13 to 17] which have been al-
loyed with magnetic elements

compound crystal lattice melting energy


structure constants point ( C) gap (eV)

Cd3 As2 body centered a ˆ 1:26461 721 ÿ0:09


tetragonal c ˆ 2:54378
Zn3 As2 body centered a ˆ 1:17786 1015 1:1
tetragonal c ˆ 2:36432
Zn3 P2 primitive a ˆ 0:80889 1160 1:6
tetragonal c ˆ 1:14069
CdP2 primitive a ˆ 0:5283 784 2:15
tetragonal c ˆ 1:9808
CdSb orthorhombic a ˆ 0:6471 456 0:53
b ˆ 0:8253
c ˆ 0:8526

ic level and a triply degenerate p-like level, respectively. In contrast to these compounds,
Cd3 As2 has an inverted band structure, since the energy difference between s-like and
p-like bands is negative as for HgTe [33, 34]. Thus, CMA with a small amount of Mn as
well as CZMA with small amounts of Mn and Zn are narrow-gap SMSCs and, being
n-type, exhibit sufficiently high electron mobilities to observe quantum oscillations of
magnetoresistance (the Shubnikov-de Haas effect) as described further on in Section 4.
The direct gap (at the G-point) of the tetragonal CdP2 is still higher than that of wide-
gap II3 ±V2 compounds, while the orthorhombic CdSb has a medium but indirect gap
with the band extrema lying along the [100] direction [17].
The organization of this paper is as follows. In Section 2, we briefly review the meth-
ods of preparation and the crystal structure of II±V SMSCs. Section 3 describes the
band structure of II3 ±V2 SMSCs taking into account their tetragonal symmetry. The
first part of Section 4 is devoted to the quantum transport in narrow-gap II3 ±V2
SMSCs, while the second part of this section deals with the classical transport in wide-
gap materials of interest. Finally, in Section 5, we discuss the magnetic properties of II±V
SMSCs studied so far.

2. Preparation and Crystal Structure


The crystal structure and composition range of known Mn- and Fe-containing II±V
alloys are presented in Table 2. All these materials correspond to their binary host semi-
conductors shown in Table 1 except for the ordered ternary compound Mg2 MnAs2 [35]
and Zn2 ÿ x Mnx Sb [36] which originates from Mn2 Sb and for x ˆ 1 is also the ordered
ternary compound ZnMnSb [37]. These materials together with CdFez Sb1ÿz are included
in Table 2 for the sake of completenes, however being typical magnetic semiconductors
and, moreover, not very well known, they will be not described in the remainder of this
article. In contrast to these materials, ZnMn2 As2 , which is also a magnetic compound
and can be treated as an ordered version of (Zn1 ÿ x Mnx )3 As2 (with x ˆ 2=3), has been
extensively studied during the last years [39 to 41] and, as characterized by an unusual
magnetic behaviour, will be presented below together with typical II±V SMSCs.
314 J. Cisowski

Table 2
Crystal structure and composition of II±V compounds containing magnetic elements

material crystal composition refs.


structure range

(Cd1 ÿ x Mnx †3 As2 tetragonal 0 < x  0:15 [18 to 21]


(Zn1 ÿ x Mnx †3 As2 tetragonal 0 < x  0:15 [22 to 25]
(Zn1 ÿ x Mnx †3 As2 hexagonal 2  x  2:5 [38 to 41]
(Cd1 ÿ x ÿ y Zny Mnx †3 As2 tetragonal 0 < x  0:06 [27 to 29]
0<y1
Mg2 MnAs2 hexagonal [35]
(Zn1 ÿ x Mnx )3 P2 tetragonal 0 < x  0:01 [26]
(Zn1 ÿ x Fex )3 As2 tetragonal 0 < x  0:015 [30]
CdP2 : Mn tetragonal x  0:005 [31]
Zn2 ÿ x Mnx Sb tetragonal 1x2 [36, 37]
CdFez Sb1 ÿz orthorhombic 0:1  z  1 [32]

2.1 Mn-alloyed II3 ±V2 compounds


To our knowledge, the first Mn-alloyed II3 ±V2 material, namely (Zn1ÿx Mnx )3 As2
(ZMA) with 2:1  x  2:5 was obtained in 1969 by Castellion et al. [38] by direct synth-
esis from the elements and was described as an ordinary wide-gap semiconductor with
the hexagonal crystal structure. The first successful attempt demonstrating that II±V
compounds can be, like II±VI compounds, also a base of a new family of SMSCs was
undertaken in 1980 by Blom et al. [18] who presented the Shubnikov-de Haas measure-
ments on samples of (Cd1 ÿ x Mnx )3 As2 (CMA) with low Mn concentration.
The first step in preparation of CMA is a melt synthesis process (performed above the
melting point of Cd3 As2 ; see Table 1) which can be made up (i) directly from the ele-
ments, (ii) using Cd3 As2 and Mn3 As2 as starting materials and (iii) mixing Cd3 As2 with
the proper amounts of Mn and As. In order to get single crystals or, at least, large
monocrystalline grains of CMA, the modified Bridgman method (slow cooling in the
presence of a temperature gradient) [18, 21] or recrystallization in the solid phase [20]
have been subsequently used.
ZMA has been grown by the modified Bridgman method exept for that the melt
synthesis process has been performed (either from the elements or by using Zn3 As2 and
Mn3 As2 as starting materials) above the melting point of Zn3 As2 ; this way, single phase
crystals of ZMA up to x ˆ 0:15 have been obtained [22 to 25].
Recently, when studying the Zn3 P2 -Mn3 As2 phase diagram, a new compound, namely
ZnMn2 As2 has been found and single crystals of this material have been prepared by the
Bridgman method [39]; this work may be treated, in a sense, as a continuation of the
above-mentioned old work [38] devoted to ZMA with high Mn concentration.
Single crystals of the quaternary compound (Cd1 ÿ x ÿ y Zny Mnx )3 As2 (CZMA) have
been grown by the Bridgman method using Cd3 As2 and Zn3 As2 with appropriate
amounts of Mn and As [27 to 29]. The crystals obtained were annealed in vapours of
Cd, As or in vacuum, in order to vary the free electron concentration.
During preparation of Mn-alloyed II3 -As2 SMSCs, which are expected to be paramag-
netic, particular attention should be paid to the formation of clusters of MnAs which is
ferromagnetic up to 317 K [42]. Such clusters have been indeed found for x  15% by
conventional X-ray studies [20, 43]; for smaller amounts of Mn, ferromagnetic clusters of
Semimagnetic Semiconductors Based on II±V Compounds 315

Fig. 1. Mn concentration as a function of the


vertical (dver ) and horizontal (dhor ) distance in
ingots of CMA [43] and CZMA [44] obtained
by the Bridgman method

MnAs can be easily detected with sus-


ceptibility measurements as discussed la-
ter on in Section 5.
Quite recently, (Zn1 ÿ x Mnx )3 P2 (ZMP)
±± a new SMSC based on Zn3 P2 has been
successfully grown for x  0:01 [26]. To
obtain ZMP, a proper small quantity of
Mn has been added to Zn3 P2 and their
mixture has been heated up to 1260  C.
After 15 min of melting, a slow cooling process has been applied, resulting in a single
phase polycrystalline material.
An important problem in the characterization of samples of a SMSC is an accurate
determination of composition and homogeneity which can be made in several ways. In
the case of Mn-alloyed II3 ±V2 compounds, the composition of the materials studied was
checked by wet chemical analysis and the homogeneity by electron microprobe measure-
ments [43, 44]; sometimes, the Mn content was verified by secondary-ion mass spectro-
scopy (SIMS) [44]. An example of the results obtained for CMA and CZMA with the
electron microprobe is shown in Fig. 1. As can be seen, short ingots prepared by the
Bridgman method are homogeneous within 5% in
both vertical and horizontal directions, while for long
crystals there is a substantial variation of the Mn con-
centration with the vertical distance when approach-
ing the top of the ingot. Therefore, samples for
further investigations should be cut perpendicularly to
the crystal growth direction and their composition
should be checked for each wafer separately.
The crystal structure of II3 ±V2 compounds which
forms a base for preparation of ternary and quatern-
ary SMSCs is quite complicated as illustrated in Fig.
2 showing the unit cell of the room temperature a-
phase of Zn3 As2 [23] as an example. The tetragonal
unit cell of a-Zn3 As2 contains 160 atoms (96 Zn and
64 As) and can be divided into 16 fluorite cubes pro-

Fig. 2. Tetragonal unit cell of a-Zn3 As2 . The cell is built up


from 16 fluorite cubes and each cube contains 4 As atoms on
a f.c.c. lattice and 6 Zn atoms on a distorted simple cubic
lattice with two vacancies (represented by open circles) con-
nected by a body diagonal [23]
316 J. Cisowski

Fig. 3. Unit cells of different phases of Cd3 As2 . When increasing temperature, the phase transitions
a ! a0 ; a0 ! a00 and a00 ! b occur at 220, 475 and 595  C, respectively [45]

vided that c  2a, where a and c are the tetragonal lattice constants (see Table 1). Each
fluorite cube contains 4 As atoms on a f.c.c. lattice (the lattice constant a0  a=2) and 6
Zn atoms inside the cube on a distorted s.c. lattice (the lattice constant  a0 =2) with 2
vacancies connected by a body diagonal.
The unit cell of Cd3 As2 is quite similar to that of Zn3 As2 , with slightly greater tetra-
gonal lattice constants (see Table 1). Above room temperature, Cd3 As2 passes through
several phase transitions [45] which are shown in Fig. 3. Apart from a-phase, also
a00 -phase is of importance as occurring in samples of CMA, CZMA and ZMP; in the
latter case, it is a consequence of the fact that Zn3 P2 itself crystallizes in a00 -phase and
its unit cell contains 40 atoms (24 Zn and 16 P) [14, 15]. However, both a and
a00 -phases are very similar and the only difference between the structures is the arrange-
ment of the above-mentioned cation vacancies which, in a00 -phase are placed on the
same face of the fluorite cube, i.e. are connected by a face diagonal. The similarity of all
phases occurring in the materials of interest allows us to refer their structure to the

Fig. 4. Compositional dependence of


the lattice constants of the fluorite-
like cell of CMA [20, 43] and ZMA
[43]. The lattice constants of the
tetragonal unit cell are given by
a ˆ 2a0 andp c ˆ 4c0 for a-phase and
by a ˆ c0 2 and c ˆ 2a0 for the a00 -
phase [45]
Semimagnetic Semiconductors Based on II±V Compounds 317

Fig. 5. Ideal hexagonal crystal structure pro-


posed for ZnMn2 As2 [39, 40]

smallest cell with the highest symmetry


which may be called aristotype or proto-
typic cell [45]. Choosing the fluorite-like
structure of b-phase (see Fig. 3) as an
aristotype, the unit cells of the other
phases can be described as composed of
deformed small aristotype cells.
Since SMSCs are ternary or quatern-
ary alloys, one expects a variation of the
lattice constants with composition. This
can be studied by X-ray or neutron dif-
fraction and the results obtained by both
methods for CMA and ZMA are gathered in Fig. 4, presenting the lattice constants of
the above-mentioned fluorite-like cell. It can be seen that the lattice constants of CMA
decrease linearly with x as expected from the Vegard law, while for ZMA these con-
stants are practically independent of x in the investigated composition range.
The crystal structure of ZnMn2 As2 (such composition corresponds to ZMA with
x ˆ 2=3) appears to be completely different from that presented above for ZMA with
x  0:2. ZnMn2 As2 is a layered magnetic semiconductor with hexagonal crystal struc-
ture which has been proposed to exhibit an ordered sequence of a doublet of Mn planes
separated by As and Zn planes as shown in Fig. 5 [39]. However, as follows from recent
neutron studies [40, 41], the magnetic order, which should follow from such an ideal
crystal structure, is far from being perfect which can be explained by the fact that the
Mn planes contain some Zn atoms and vice versa.

2.2 Other semimagnetic semiconductors based on II±V compounds

The recently obtained (Zn1ÿx Mnx )3 As2 (ZFA) has been synthetized from stoichiometric
melts of Zn3 As2 and Fe by the modified Bridgman method [30]. Based on X-ray studies,
the resulting material is claimed to be homogeneous for a small Fe content, i.e. for
x  0:015; revealing the presence of a second phase identified as Fe2 As for greater x:
The crystal structure of ZFA is very similar to that of ZMA with small x.
Apart from the above-mentioned II3 ±V2 compounds alloyed with Mn and Fe, a suc-
cessful doping with Mn of the tetragonal CdP2 , a member of II±V2 group, has been also
reported [31]. Single crystals of CdP2 :Mn have been prepared by the Bridgman method
using polycrystalline CdP2 and Mn as starting materials. It appears that homogeneous
crystals have been obtained only for a very small amount of Mn, i.e. for x < 0:5%. The
unit cell of tetragonal CdP2 (there exists also a monoclinic modification of this com-
pound [14, 15]) contains 24 atoms (8 Cd and 16 P), all being tetrahedrally coordinated.
Every Cd atom is surrounded by 4 P atoms, while every P atom by 2 Cd and 2 P atoms.
Additionally, the P atoms form spiral chains and, on the whole, the CdP2 single crystals
appear to be strongly anisotropic.
318 J. Cisowski

3. Band Structure
Qualitatively, the band structure of a SMSC resembles that of the parent semiconduct-
ing compound having the same crystal structure. As follows from theoretical pseudopo-
tential calculations [46, 47], the band structure of II3 ±V2 tetragonal semiconducting
compounds is very similar to that of cubic III±V and II±VI materials, i.e. we are dealing
with direct gap semiconductors with the band extrema occurring at the G-point. By
analogy with II±VI SMSCs [6, 8], the band structure of II3 ±V2 SMSCs at the G-point is
expected to be as that schematically depicted in Fig. 6, taking (Zn1 ÿ x Mnx †3 As2 as an
example. On the ordinary band structure of Zn3 As2 is superimposed the effect of the 3d
shell of Mn represented in Fig. 6 by the occupied (3d5 ) and unoccupied (3d6 ) levels split
by the energy Ueff  7 eV and the location of the 3d5 level is typically about Ed ˆ 3:5 eV
below the top of the valence band Ev . In reality, the 3d5 level is broadened into a
narrow band due to mixing or hybridization between Mn d-orbitals and anion p-orbitals
contained in the valence band [6, 8].
As is known, the values of the energy levels connected with the presence of magnetic
ions in SMSCs can be deduced from the photoelectron spectroscopy measurements. Pre-
liminary results of such measurements on CMA, combined with an analysis of the d±d
interaction [22], indicate that the 3d5 level should be closer to the top of the valence
band of this material than in ZMA. To our knowledge, this work, however, has not been
continued to yield values of Ed and Ueff which could be useful in determining the magni-
tude of p±d and d±d interactions in the materials of interest. On the other hand, both
Ed and Ueff in Mn-alloyed II±VI SMSCs are practically cation independent [48] and
therefore one may apply the above-cited values of Ed and Ueff also for II±V SMSCs,
reserving the most important role for the p±d hybridization potential which depends
mostly on the cation-anion distance as discussed later on in Section 5.2.
One of the attractive features of SMSCs is the tunability of band parameters by chan-
ging their composition. The most well known example for this is a linear dependence of
the energy gap on the magnetic ion content established for almost all II±VI SMSCs [6,
8]. The same linear dependence may be expected for II3 ±V2 SMSCs by analogy with
their nonmagnetic solid solutions [15, 49] as shown for (Cd1 ÿ y Zny )3 As2 and
(Cd1 ÿ x Mnx )3 A2 in Fig. 7 illustrating additionally the transition from the inverted band
structure of Cd3 As2 (x; y ˆ 0) to the simple structure which, for x; y ˆ 1, corresponds to

Fig. 6. Schematic band structure for a tetragonal SMSC with the


direct gap at the G-point, taking Zn3 As2 alloyed with Mn as an
example. 3d5 and 3d6 denote the occupied and unoccupied Mn le-
vels, respectively, separated by Ueff  7 eV
Semimagnetic Semiconductors Based on II±V Compounds 319

Fig. 7. Compositional dependence of


the energy gap in (Cd1 ÿ y Zny )3 As2
(CZA) and (Cd1 ÿ x Mnx )3 As2 (CMA)
at 4:2 K. Points deduced from experi-
ment and denoted by circles are taken
from [49] and references cited therein,
triangles down from [19, 21] and trian-
gle up from [65]. The straight lines re-
present Eqs. (1) and (2) for CMA and
CZA, respectively

the end compounds Zn3 As2 and hypothetical Mn3 As2 (real Mn3 As2 is a magnetic semi-
conductor with hexagonal crystal structure [40]). Since the hypothetical Mn3 As2 would
have E0 ˆ 1:36 eV [50], the variation of E0 (in eV) with x in CMA can be written (at
low temperatures) as
E0 ˆ ÿ0:095 ‡ 1:45x ; …1†
while for ZMA, this dependence is expected to be much weaker and given by
E0 ˆ 1:1 ‡ 0:26x : …2†
Results concerning the band edges in SMSCs as described above should be augmented
by knowledge of the dispersion relation E…k† and its behaviour in the magnetic field
which reveals the specific features of this group of materials. Therefore below, before
presenting the sp±d exchange interaction in II3 ±V2 SMSCs in greater detail, we first
briefly demonstrate the Bodnar model [33, 51] commonly accepted for the description of
the band structure of II3 ±V2 narrow-gap host semiconductors such as Cd3 As2 , followed
by the Wallace extension of this model in the presence of a quantizing magnetic field
[52].

3.1 The Bodnar model

The Bodnar model [33, 51] is a generalization of the three-level Kane model for narrow-
gap semiconductors with the tetragonal crystal structure in the frame of the k  p ap-
proach. Bodnar treated the tetragonal crystal field as an additional perturbation added
to the Hamiltonian describing a cubic semiconductor with the spin±orbit coupling in-
cluded and, moreover, he took into account the anisotropy of the interband momentum
matrix element neglecting, however, the anisotropy of the spin±orbit interaction. The
eigenvalue equation of the total k  p Hamiltonian can be expressed as
…H0 ‡ Hpert † uk ˆ E 0 uk …3†
with the unperturbed Hamiltonian
p2
H0 ˆ ‡ V0 …4†
2m0
and the perturbation part
h h

Hpert ˆ k  p‡ …5V  p†  s ‡ Vt …5†
m0 4m2o c2
320 J. Cisowski

In these expressions, uk is the amplitude of the Bloch function, the energy


E0 ˆ E ÿ  h2 k2 =2m0 (where the free electron term 
h2 k2 =2m0 with k, the wave vector and
m0 , the free electron mass, is usually neglected), p is the momentum operator, s is the
Pauli spin operator, and the total crystal potential V ˆ V0 ‡ Vt , where V0 has cubic
symmetry and Vt is caused by the tetragonal distortion.
The resulting relationship between the energy and the wave vector for a tetragonal
narrow-gap semiconductor can be then written as [51]
g…E 0 † ˆ f1 …E 0 † …k2x ‡ k2y † ‡ f2 …E 0 † k2z …6†

with
 
g…E 0 † ˆ E 0 …E 0 ÿ E0 † E 0 …E 0 ‡ D† ‡ d…E 0 ‡ 23 D† ; …7†
 
f1 …E 0 † ˆ P?2 E 0 …‡E 0 ‡ 23 D† ‡ d…E 0 ‡ 13 D† ; …8†
0
f2 …E † ˆ Pk2 E 0 …E 0 ‡ 2
3 D† ; …9†

where E0 is the energy gap between s- and p-like levels, D is the spin±orbit splitting, d
is the crystal field splitting, and P? and Pk are the interband momentum matrix ele-
ments which, as follows from experiment, are need not be distinguishable.
The level ordering at the G-point of the Brillouin zone is determined by four eigenva-
lues of the secular equation (6) for k ˆ 0, i.e: 0; E0 ; E1 and E2 with
q
E1; 2 ˆ ÿ 12 …D ‡ d†  …D ‡ d†2 ÿ 83 Dd : …10†

The level E ˆ E0 originates from a nondegenerate s-like atomic level, while the levels
E ˆ 0, E1 and E2 originate from a triply degenerated p-like atomic level split under the
simultaneous influence of the spin±orbit interaction and the tetragonal crystal field.
These splittings are relatively small as shown in Fig. 6 for Zn3 As2 , where the energy
separation between the first and second valence subbands is about 0:05 eV, and between
the second and third subbands ±± about 0:25 eV [16]. In materials with a small energy
gap, states away from k ˆ 0, have a mixed s±p type character.
Depending on the signs of E0 and d, one can describe with Eqs.(6) to (10) all possible
four-level models of a tetragonal semiconductor, i.e. the simple band structure (E0 > 0)
or inverted structure (E0 < 0) with positive (d > 0) or negative (d < 0) crystal field
splitting. In II3 ±V2 semiconducting compounds, we are dealing with the positive crystal
field splitting and the simple band structure [16], with the already mentioned exception
in the case of Cd3 As2 which has the inverted structure (see Table 1 and Fig. 7). An-
other important feature of the secular equation is that the conduction band (described
by the largest root of equation (6) is nonspherical and nonparabolic, and the Fermi sur-
face has a shape of an ellipsoid of revolution along the kz direction. Moreover, since the
functions f1 and f2 (see Eqs. (8) and (9), respectively) depend on energy in different
ways, the ellipsoid changes its shape with energy (approaching a sphere for high ener-
gies) which means that the anisotropy of physical properties depends on the carrier con-
centration n as really observed in experiment [51]. For n  4  1018 cmÿ3 , the four-level
Bodnar model itself needs an improvement by taking into account the influence of high-
er bands which results in warping of the Fermi surface [19, 51, 53].
Starting from the Bodnar model, Wallace [52] has calculated the band structure of
Cd3 As2 in the presence of a quantizing magnetic field. For an arbitrary angle q between
Semimagnetic Semiconductors Based on II±V Compounds 321

Fig. 8. Angular dependence of the cyclotron


electron effective mass for the sample of
Cd3 As2 [51]. The solid line represents
Eq. (13) with the energy gap E0 ˆ ÿ0:095
eV, spin±orbit splitting D ˆ 0:27 eV, crystal-
field splitting d ˆ 0:09 eV and matrix ele-
ment P? ˆ Pk ˆ P ˆ 7:38  10ÿ8 eV cm

the magnetic field B and the z-axis, one makes the coordinate transformation so that B
is along the new z0 -axis. The resulting secular equation, from which the Landau levels
can be calculated, is given by [52]
2N ‡ 1  2 2
1=2 f1 f2 k2z0
g…E† ˆ f1 …f1 cos q ‡ f2 sin q† ‡
l2 f1 cos2 q ‡ f2 sin2 q
P? D
 ‰…E ‡ d†2 P?2 cos2 q ‡ E 2 Pk2 sin2 qŠ1=2 ; …11†
3l2
where g…E† is given by Eq. (6), N ˆ 1; 2; 3; . . . is the Landau number and the cyclo-
p
tron radius is l ˆ  h=eB . The first term on the right-hand side of the above equation
represents the quantization in the plane perpendicular to B; the second term is unaf-
fected by the field and the third term accounts for the spin splitting of the Landau
levels. For a given Landau number N, the energy separation between the two spin-split
Landau sublevels determines the effective g-factor g* by the relation EN ‡ ÿ
ÿ EN ˆ g*mB B,
where mB is the Bohr magneton. If g B *m B  E F (with EF , the Fermi energy), an analy-
tical expression for the angular dependence of the effective g-factor can be obtained in
the form [52]
" #1=2
2m0 …E ‡ d†2 P?2 cos2 q ‡ E 2 Pk2 sin2 q
*
jg …q†j ˆ P? D ; …12†
3m*c …q† f1 …f1 cos2 q ‡ f2 sin2 q†

where m*c is the cyclotron effective mass determined, in the absence of spin splitting, by
the energy difference between the two adjacent Landau levels EN ‡ 1 ÿ EN ˆ  hwc , with
the cyclotron frequency wc ˆ eB=m*c . At low fields, the angular dependence of the cyclo-
tron effective mass is also given by an analytical formula as
( )
2
h
 d g
m*c …q† ˆ : …13†
2 dE f1 …f1 cos2 q ‡ f2 sin2 q†1=2

An example of the calculated m*c …q† dependence along with the results of careful ex-
periment for a sample of Cd3 As2 [51] is given in Fig. 8 showing a small relative anisotro-
py [m*c …90o † ÿ m*c …0o †Š=m*c …0o † ' 10%. It turns out that formula (13) may be also ap-
plied to II3 ±V2 SMSCs, since the exchange interaction practically does not affect the
values of the cyclotron effective mass [53]. On the other hand, the values of the effective
g-factor for SMSCs appear to be very much different from the band g-factor values of
322 J. Cisowski

their nonmagnetic counterparts as will be discussed below. Such behaviour of the effec-
tive mass and g-factor reflects a very characteristic feature of SMSC that the orbital
properties remain practically unchanged, while the spin properties are greatly modified
by the presence of magnetic ions.

3.2 sp±d exchange interaction

When passing from a nonmagnetic semiconductor to a SMSC, the k  p Hamiltonian (3)


has to be completed with the sp±d exchange interaction term describing the interaction
between band carriers (s or p) and localized 3d electrons of transition metal atoms. This
term can be expressed by a Heisenberg Hamiltonian as [3 to 6, 8]
P
Hexch ˆ Jsp d …r ÿ Rn † s  Sn ; …14†
n

where s is the spin operator of an electron at position r, Sn is the total spin operator of
a 3d shell of the magnetic ion at Rn , Jsp d …r ÿ Rn † is an appropriate exchange constant
and summation runs over exact positions of magnetic ions. Using the molecular field and
virtual crystal approximations and assuming that the external magnetic field B is paral-
lel to the z-axis, equation (3) can be simplified to the form
P
Hexch ˆ xs z hSz i Jsp d …r ÿ R† ; …15†
R

where s z is a component of s along the magnetic field, hSz i is the thermal average of S
along the field and summation extends now over all cation sites. If B ˆ 0, also hSz i ˆ 0
and a given material behaves like a typical nonmagnetic semiconductor.
In Mn-alloyed cubic SMSCs, the s±d and p±d exchange interactions are character-
ized by the exchange constants a and b; respectively. According to the present under-
standing of the exchange interaction between band carriers and d electrons of Mn
ions [6, 8, 48], the exchange constant a arises almost exclusively from the direct ferro-
magnetic s±d exchange and is therefore positive. On the other hand, the p±d ex-
change constant b contains both positive (bdir ) and negative (bhyb ) contributions, the
latter of the two being dominant and connected with the antiferromagnetic p±d hy-
bridization. Thus, we may write that b ˆ bdir ‡ bhyb with bdir  a and, according to
Larson et al. [48],

2 Ueff
N0 bhyb ˆ ÿ32Vpd ; …16†
…Ed ‡ Ueff ÿ Ev † …Ev ÿ Ed t†

where N0 is the number of cations per unit volume, Vpd is the hybridization potential
and the other quantities are those from Fig. 6.
In Mn-alloyed cubic II±VI SMSCs, the experimentally obtained values of N0 a and
N0 b are of the order of 0:3 and ÿ1 eV, respectively [6, 8, 48]. Similar values are reported
for II±VI SMSC0 s alloyed with Fe and Co [8], while for Cr2‡ , N0 b appears to be positive
indicating a new, ferromagnetic type of p±d interaction [55].
Calculations of the energy levels in tetragonal narrow-gap SMSCs has been done by
Neve et al. [54] by taking into account the sp±d exchange interaction described above
and treating the Wallace extension of the Bodnar model as a starting point. The basis
functions at the band edges are taken as
Semimagnetic Semiconductors Based on II±V Compounds 323

u1 ˆ iS # ; u5 ˆ ÿ…6†ÿ1=2 …X ‡ iY † # ‡…2=3†ÿ1=2 Z " ;


u2 ˆ iS " ; u6 ˆ …2†ÿ1=2 …X ‡ iY † " ;
u3 ˆ …2†ÿ1=2 …X ÿ iY † # ; u7 ˆ ÿ…3†ÿ1=2 …X ÿ iY † " ‡…3†ÿ1=2 Z # ;
u4 ˆ …6†ÿ1=2 …X ÿ iY † " ‡…2=3†1=2 Z # ; u8 ˆ …3†ÿ1=2 ……X ‡ iY † # ‡…3†ÿ1=2 Z " ;
…17†
where S and X; Y; Z are periodic Kohn-Luttinger amplitudes corresponding to s- and p-
like atomic functions, respectively. The wave function is then expressed by
P
8
wˆ u n jn ; …18†
nˆ1
where j1 ; . . . ; j8 are envelope functions. This leads to the eigenvalue equation of the
form
2 3
j1
6 j2 7
6 7
6  7
…H ÿ E  I† 6 7
6  7 ˆ 0; …19†
6 7
4  5
j8
where I is the unit matrix and the total Hamiltonian can be expressed, in general, as
the 8  8 matrix, written as [54]
 
H1 H3
Hˆ ‡ ; …20†
H3 H2
where
2 r 3
‡ ÿ 2
6 E0 ÿ c c P? k
3
Pk kz 7
6 7
6 1 7
6 7
6 cÿ E0 ‡ c 0 p P? kÿ 7
6 3 7
6
H1 ˆ 6 7; …21†
1 7
6 P? k‡ 0 ÿb p d‡ 7
6 7
6 3 7
6 r 7
4 2 1 1 1 5
Pk kz p P? k‡ p dÿ ÿ …2d ÿ b ‡ 2b0 †
3 3 3 3
2 p p 3
1 0 1 ‡ 2 ÿ 2
6 ÿ 3 …2d ‡ b ÿ 2b † ÿ p d t …b ‡ b0 ÿ d† 7
6 3 3 3 7
6 r 7
6 1 2 ÿ 7
6 ÿ p dÿ b 0 d 7
6 7
6 3 3 7
H2 ˆ 6 p 7;
6 2 ‡ 1 1 ‡ 7
6 7
6 t 0 ÿD ÿ …d ÿ 2b ‡ b0 † t 7
6 3 3 3 7
6 p r 7
4 2 2 ‡ 1 ÿ 1 5
…b ‡ b0 ÿ d† d t ÿD ÿ …d ‡ 2b ÿ b0 †
3 3 3 3
…22†
324 J. Cisowski

2 r 3
1 ‡ 1 2 ‡
6 ÿ p P? k 0 p Pk kz
3
P? k 7
6 3 3 7
6 r r 7
6 2 2 1 7
6 p Pk kz 7
6 k kz P? k‡ ÿ P? kÿ 7
6 3 3 3 7
H3 ˆ 6 r 7; …23†
6 7
6 2 ‡ 7
6 0 0 ÿ d 0 7
6 3 7
6 p p 7
4 2 ‡ 2 2 ‡ 5
t 0 ÿ …b ‡ b0 ‡ d† t
3 3 3

with c ˆ axhSz i=2, b ˆ b? xhSz i=2, b0pˆ bk xhSz i=2, c ˆ axhS i=2, d ˆ b? xhS i=2,
t ˆ bk xhS i=2 and k ˆ …kx  ky †= 2. In these expressions, hSz i and hS i ˆ


Sx  iSy are the thermal averages of the components of Mn spins, and the quantities
a, bk and b? represent exchange interaction constants for electrons with wave function
of s- (a) and p-like ( bk and b? † symmetries. Due to the lower symmetry of a tetragonal
SMSC, we have three independent constants instead of only two as in cubic II±VI
SMSCs for which bk ˆ b? ˆ b. As for hSz i and hS i, these quantities can be expressed
by hSz i ˆ hSiB cos q and hS i ˆ hSiB sin q, respectively, where hSiB is the nonvanishing
thermal average of the component of magnetic ion spin along the magnetic field B: hSiB
is proportional to the macroscopic magnetization M (the magnetization per Mn ion is
given by Mion ˆ ÿgMn mB hSiB , where gMn ˆ 2 is the Lande factor of the Mn2‡ ion and
mB is the Bohr magneton) which can be measured and/or described by phenomenologi-
cal formulae based on the Brillouin function [56, 57]. When B ˆ 0, also hSiB ˆ 0 and
then all the terms in matrices (21) to (23) containing this quantity via hSz i and hS i
vanish, leading, as expected, to the Bodnar Hamiltonian [51] which, after diagonaliza-
tion, results in the secular equation (6).
If the magnetic field is parallel to the c-axis of a tetragonal SMSC, i.e. for q ˆ 0 and,
additionally, when kz ˆ 0, the 8  8 matrix (20) decouples into two 4  4 numerical
matrices yielding, after diagonalization, two sets of Landau levels. An example of such

Fig. 9. Lowest lying conduction band Lan-


dau levels in CMA (x ˆ 0:01) at 4 K as a
function of the magnetic field [54]
Semimagnetic Semiconductors Based on II±V Compounds 325

calculation for CMA is presented in Fig. 9, where the spin split conduction band Landau
levels are plotted as a function of the magnetic field [54]. For B  4 T, one can see the
crossings of the levels - the feature which is characteristic of SMSCs as being due to the
fact that, at lower fields, the spin splitting exceeds the Landau splitting because of a
strong increase of hSiB with increasing B; for higher fields, hSiB tends to saturate (see
Section 5), the Landau splitting of a narrow-gap SMSC, as in this case, exceeds the spin
splitting leading to a sequence of levels as in the nonmagnetic counterpart.
When q 6ˆ 0, calculations of the Landau levels become much more difficult, since then
the eigenvalue problem cannot be decoupled into two 4  4 problems, but its dimension
is 8  8 even for kz ˆ 0 [54]. For an arbitrary direction of the magnetic field, one makes,
as usual, the coordinate transformation so that B is along the new z0 -axis, followed by
introducing the harmonic oscillator lowering and raising operators a‡ and aÿ defined as
[52, 54]

a ˆ l…4AB†ÿ1=4 …A1=2 kx0  iB1=2 ky0 ‡ DAÿ1=2 kz0 † …24†

with A ˆ f1 cos2 q ‡ f2 sin2 q, B ˆ f1 and D ˆ …f1 ÿ f2 † sin q cos q. This leads, with
the assumption that only the thermal component of the magnetic ion spin along the
magnetic field hSiB does not vanish and for kz0 ˆ 0, to the Hamiltonian matrix of the
form
 
Ha Hc
Hˆ ; …25†
Hc‡ Hb

where

Ha ˆ
2 r 3
1 2 ‡ ‡
6 ÿc cos q f ÿ a‡ ‡ f ‡ aÿ ÿ p …f ÿ a‡ ‡ f ‡ aÿ † …f a ‡ f ÿ aÿ † 7
6 3 3 7
6 7
6 f ‡ a‡ ‡ f ÿ aÿ ÿb cos q ÿ E0 0 0 7
6 7
6 7
6 0
p …b ‡ b0 † cos q ÿ d 7;
6 ÿ p1 …f ÿ a‡ ‡ f ‡ aÿ † 0
…b ÿ b † cos q ‡ 2d
ÿ E0 2 7
6 3 3 3 7
6 7
6 r 7
4 2 ÿ ‡ p …b ‡ b0 † cos q ÿ d 0
…2b ÿ b † cos q ‡ d 5
…f a ‡ f ‡ aÿ † 0 2 ÿD ÿ ÿ E0
3 3 3
…26†

Hb ˆ
2 r 3
1 2 ÿ ‡
6 c cos q ÿ p …f ÿ a‡ ‡ f ‡ aÿ † f ‡ a‡ ‡ f ÿ aÿ ÿ …f a ‡ f ‡ aÿ † 7
6 3 3 7
6 p 7
6 1 0
…2b ÿ b† cos q ‡ 2d 2 7
6 0 7
6 p …f ‡ a‡ ‡ f ÿ aÿ † ÿ E0 0 ÿ ‰…b ‡ b † cos q ‡ dŠ 7
6 3 3 3 7;
6 7
6 f ÿ a‡ ‡ f ‡ aÿ 0 b cos q ÿ E0 0 7
6 7
6 r p 7
4 2 ‡ ‡ 2 0
0
…b ÿ 2b† cos q 5
ÿ …f a ‡ f ÿ aÿ † ÿ ‰…b ‡ b † cos q ‡ dŠ 0 ÿD ÿ ÿ E0
3 3 3
…27†

22 physica (b) 200/2


326 J. Cisowski
2 r 3
2 1
6 c sin q F …a‡ ‡ aÿ † 0 p F …a‡ ‡ aÿ † 7
6 3 3 7
6 r 7
6 1 2 7
6 7
6 0 p b sin q 0 ÿ b sin q 7
6 3 3 7
Hc ˆ 6
6 r 
 p 7;
7 …28†
6 2 2 0 1 2 0 7
6 F …a‡ ‡ aÿ † b sin q ÿ p 
 b sin q b sin q 7
6 3 3 3 3 7
6 p r 7
6 7
4 1 2 0 2 1 0 5
p F …a‡ ‡ aÿ † b sin q b sin q b sin q
3 3 3 3

P
where now c ˆ axhS iB =2;pb ˆ b? xhS iB =2; b0 ˆ bk xhS iB =2, f  ˆ ‰…B=A†1=4 cos q
1=4 1=4 2l
…A=B† Š and F ˆ ÿP =…l 2† …B=A† sin q.
Many terms of the above Hamiltonian contain linear combinations of the harmonic
oscillator lowering and raising operators a‡ and aÿ which means that a simple vector
column of harmonic oscillator wave functions is not an eigenvector as in the case q ˆ 0o .
Neglecting various terms containing combinations of operators a‡ and aÿ and making
many other simplifications, Neve et al. [54] have finally obtained an expression for the
angular dependence of the effective g-factor for conduction electrons in tetragonal nar-
row-gap SMSCs. An example of the result of calculation for CMA is shown in Fig. 10,
where, for the purpose of comparison, the angular dependence of the band g-factor in
Cd3 As2 [57, 19] is also presented. As can be seen, the g*- factor predicted for CMA is
strongly anisotropic and, for small angles, much higher than that determined for
Cd3 As2 . Moreover, the g*-factor of CMA is strongly temperature dependent in the re-
gion of small angles and the temperature variation is far less pronounced for large an-
gles, where the value of the g*-factor approaches that expected for Cd3 As2 with the
same electron concentration (n ˆ 5  1018 cmÿ3 ).
As already stated in [54], the predicted behaviour of the g*-factor of CMA (especially
for q > 60o † may not be very reliable in view of the approximations made during the
calculations. In fact, the subsequent SdH studies of single crystals of CMA [53, 19] (part
of which will be presented later on in Section 3.3) indicated that the effective g-factor
anisotropy was not as strong as predicted by theory and therefore we have applied an-
other approach which goes beyond that taken in [54]. Bearing in mind the starting Ha-

Fig. 10. Comparison of the angular dependences


of the effective g-factor determined for Cd3 As2 .
Experimental data after [19, 51] and the solid
lines are calculated taking into account the
quantization of the free electron term [58] and
deduced for CMA (dashed lines) from approxi-
mative theoretical calculations [54]
Semimagnetic Semiconductors Based on II±V Compounds 327

miltonian matrix (20), we are looking for the eigenvector w…n† in the form of a series of
harmonic oscillator functions [28],
2 n 3 2 n ‡ 2k 3
a1 Fn a1 Fn ‡ 2k
6 a3 Fn ‡ 1 7
n 6 an ‡ 2k Fn ‡ 1 ‡ 2k 7
6 n 7 6 3n ‡ 2k 7
6 a5 Fn ÿ 1 7 6a Fn ÿ 1 ‡ 2k 7
6 n 7 6 5 7
6 a8 Fn ÿ 1 7 1 6 n ‡ 2k
P a8 Fn ÿ 1 ‡ 2k 7
6
w…n† ˆ 6 n 7 6 7
7‡ 6 an ‡ 2k Fn ‡ 2k 7 ; …29†
6 a2n Fn 7 k kˆ6ˆÿ1 6 2 7
6 a Fn ‡ 1 7 0
6 an ‡ 2k Fn ‡ 1 ‡ 2k 7
6 4n 7 6 4 7
4 a Fnÿ1 5 4 an ‡ 2k Fn ÿ 1 ‡ 2k 5
6 6
an7 Fn ‡ 1 an7 ‡ 2k Fn ‡ 1 ‡ 2k

where only for q ˆ 0o , the first column is retained. Substituting this series expansion
into the Schr odinger equation, a matrix of infinite size is obtained. Then, if a certain
cut-off is assumed, it is possible to find the eigenvalues requiring, however, lengthy com-
puter calculations. Practically, a sufficient accuracy is obtained if the series is truncated
above k  3, what reduces the eigenvalue problem to the solution of a 56  56 matrix
[28]. This way, the eigenenergies in the conduction band are calculated for given B; T , q
and Landau number N, from which the electronic g* -factor is determined for this set of
parameters. Finally, one calculates the ratio between spin and Landau splittings
n ˆ g*m*c =2m0 (with m*c given by Eq. (13)), since this quantity appears in expressions
for the SdH oscillations, allowing one to compare the theoretical model with experiment
and to estimate the sp±d exchange constants as discussed in Section 3.3.
The effect of the exchange interaction on the band structure can be much more easily
seen for wide-gap SMSCs characterized by large effective masses and small band g-fac-
tors which allow one to treat the magnetic splittings of the band states as being entirely
due to the exchange interaction. For a magnetic field directed along the nominal z-axis,
the Zeeman splittings of the conduction (DEcb ) and valence (DEvb ) bands of cubic
SMSCs can be expressed as [48, 57]
DEcb ˆ xN0 ahSz i mj ; …30†
DEvb ˆ 13 xN0 bhSz i mj ; …31†

Fig. 11. Light-induced magnetization as a


function of the photon energy in a sample of
ZMA with x ˆ 0:01 at 4:2 K at different mag-
netic fields [24]

22*
328 J. Cisowski

where mj ˆ 1=2 for the conduction band states, while, for the valence band states,
mj ˆ 3=2 and mj ˆ 1=2 for light and heavy holes, respectively. Experimentally, these
splittings can be determined from the exciton magnetoreflection structure, by assigning
an exciton to each interband transition between the Zeeman split conduction and va-
lence bands. The strongest transitions j3=2ivb ! j1=2ic , easily seen in experiment,
give directly the total conduction±valence band splitting
DE ˆ x…N0 a ÿ N0 b† hSz i ; …32†
from which the difference of the exchange constants …N0 a ÿ N0 b† can be deduced pro-
vided that hSz i is determined from magnetization. As to II±V SMSCs, this method has
been only applied to ZMP, resulting in …N0 a ÿ N0 b† ˆ 1:2 eV [26], i.e. similarly as in
II±VI SMSC0 s. Values of the sp±d exchange interaction constants for narrow-gap II±V
SMSCs, as being estimated from an analysis of the SdH effect, will be presented in Sec-
tion 4.1.

3.3 Intraionic Mn transitions


The Mn atom placed in a cation site of a SMSC lattice is subjected to the action of the
crystal field leading to the possibility of intraionic transitions between the ground and
excited states which originate from the (S ˆ 5=2; L ˆ 0) and (S ˆ 3=2; L 6ˆ 0) levels of
the free Mn2‡ ion, respectively [6]. The lowest and therefore the most important transi-
tion corresponds to about 2:2 eV and has been observed in many II±VI SMSCs. A simi-
lar transition has been also recovered in one of II±V SMSCs, namely in ZMA with 1%
Mn [24] for which light-induced magnetization has been observed, as shown in Fig. 11.
A prominent maximum seen at 2:07 eV and another feature at 2:15 eV has been inter-
preted, by analogy with II±VI SMSCs, as being connected with the intraionic transi-
tions within the antiferromagnetically coupled Mn2‡ ÿMn2‡ pairs [24].

4. Transport Properties
Among all the known II±V SMSCs, only some transport properties of CMA, CZMA and
ZMA have been investigated so far. Since CMA and CZMA with small Mn and Zn con-
tents are narrow-gap materials with high electron concentrations and mobilities, they
have been mainly studied by the Shubnikov-de Haas (SdH) effect being a quantum phe-
nomenon, in contrast to the classical transport measurements performed on ZMA which
is a typical wide-gap (E0  1:1 eV) material; this difference will be reflected below, by
presenting separately the quantum and classical transport phenomena.

4.1 The Shubnikov-de Haas effect


The Shubnikov-de Haas (SdH) effect is an oscillatory variation of magnetoresistance
which follows from the Landau quantization of the electron states in the magnetic field
as described in the previous section. The SdH study of CMA at field B  5 T [18] was
the first to show that a II±V compound such as Cd3 As2 alloyed with Mn exhibits fea-
tures characteristic of SMSCs, i.e., in contrast to ordinary semiconductors, a strongly
nonmonotonic dependence of the amplitude of SdH oscillations on the magnetic field
and temperature. The latter dependence for CMA is given in Fig. 12, showing that the
SdH amplitude at a fixed field does not decrease monotonically with temperature but
Semimagnetic Semiconductors Based on II±V Compounds 329

Fig. 12. Experimental and calculated (solid


line) temperature dependence of the SdH am-
plitude in CMA with x ˆ 0:01. Broken line re-
presents the SdH amplitude in an equivalent
nonmagnetic semiconductor [18, 19] (a.u. al-
ways means arb. units)

exhibits a much more complicated beha-


viour, even vanishing at certain tempera-
tures as explained below.
In a ternary SMSC such as CMA, a
change of Mn content influences both
the band structure and magnetic proper-
ties of the system. This inconvenience
can be, to a large extent, overcome in
quaternary alloys such as CZMA which allows one to control the band structure by the
Zn content, independently of the amount of Mn responsible for the magnetic properties.
Studies of the SdH effect in samples of CZMA with various compositions are presented
in [27 to 29] and, quite recently, in [59], under the influence of pressure in [60] and in
fields up to 25 T in [61]. An example of SdH oscillations in a sample of CZMA, mea-
sured as a function of the magnetic field at various temperatures, is shown in Fig. 13.
The general formula used to describe the SdH oscillations of magnetoresistance Dr=r0
has the form of a harmonic expansion [62],
p
Dr P1 B X r Kr
ˆC p cos …prn† cos ‰2pr…F =B ÿ g† ÿ p=4Š ; …33†
r0 rˆ1 r sinh Xr
where Kr ˆ exp …ÿa*rTD =B†; a* ˆ 2p2 m*c kB =e
h (TD is the Dingle temperature describ-
ing the collision broadening of the Landau levels), Xr ˆ a*rT =B and n ˆ g*m*=2m0 is
the already mentioned ratio between the spin- and Landau splittings, kB is the Boltz-
mann constant, g represents a phase factor and C is a constant. The oscillatory charac-

Fig. 13. SdH oscillations in a sample of CZMA


(x ˆ 0:03; y ˆ 0:1) with the electron concentration
of 3:75  1018 cmÿ3 at different temperatures. Ar-
rows indicate the positions of the first harmonic
minimum [29]
330 J. Cisowski

ter of magnetoresistance is given by the last cosine factor in Eq. (33) with the funda-
mental frequency F ˆ 1=P where the oscillation period P is related, for a spherical Fer-
mi surface, to the electron concentration n by P ˆ 2e…3p2 n†ÿ2=3 =
h:
Although Eq. (33) has been derived for an isotropic spherical Fermi surface, it is also
applicable to arbitrary shaped closed Fermi surfaces. In such a case, P and m*c are
given, in general, by the relations
 
2pe
Pˆ ; …34†
hSm E ˆ EF

 
h2 dSm

m*c ˆ ; …35†
2p dE E ˆ EF

where Sm is an extremal cross-sectional area of the Fermi surface intercepted by the


plane perpendicular to the magnetic field and EF is the Fermi energy. As already signa-
lized in the previous section, the Fermi surface of II3 ±V2 narrow-gap materials described
by the Bodnar model is a single ellipsoid of revolution which means that [51]
pg
Sm ˆ ; …36†
‰f1 …f1 cos2 q ‡ f2 sin2 q†Š1=2

leading to the angular dependence of the SdH period in the form

P …q† ˆ ‰P 2 …0 † cos2 q ‡ P 2 …90 † sin2 qŠ1=2 ; …37†

where P …0 † and P …90 † denote the SdH periods for the magnetic field parallel and per-
pendicular to the tetragonal c-axis, respectively. In high degeneracy conditions necessary
for observation of the SdH effect, the carrier concentration for the single ellipsoidal Fer-
mi surface can be expressed by both SdH periods as
 3=2 " #
1 2e 1 1 g…E†3=2
nˆ 2 p ˆ p : …38†
3p h
 P …0 † P …90 † 3p2 f1 …E† f2 …E † E ˆ EF

Due to the tetragonal symmetry of II3 ±V2 compounds including Mn-alloyed materials,
an anisotropy of the SdH period, which can be defined as KP ˆ P …0 †=P …90 †, is ex-
pected. This is illustrated in Fig. 14, where, for the purpose of comparison, some selected
experimental data for Cd3 As2 [51, 63], CMA [19, 53] and CZMA [28] are shown along
with the solid lines representing Eq. (37). It can be seen that the observed anisotropy
depends on both the electron concentration and composition. Sample CMA-2 contains
only 0:25% Mn and, in a sense, may be treated as a high electron concentration sample
of Cd3 As2 doped with Mn. Therefore when passing from sample CA-1 through CA-2 to
CMA-2, one notices a considerable decrease of the relative SdH period anisotropy (i.e.
‰P …0 † ÿ P …90 †Š=P …90 † ˆ KP ÿ 1† in Cd3 As2 with increasing n. On the other hand,
when comparing the pairs of samples with similar (CA-1 and CZMA-1) or equal n
(CA-2 and CZMA-2), one sees that substitution of Zn and Mn for Cd results also in a
strong decrease of the anisotropy. Sample CMA-2 with 1% Mn represents an intermedi-
ary behaviour between that of Cd3 As2 and CZMA.
The solid lines for CMA and CZMA in Fig. 14 have been calculated in the same way
as for Cd3 As2 , i.e. assuming that the period anisotropy follows only from the anisotropy
Semimagnetic Semiconductors Based on II±V Compounds 331

Fig. 14. Comparison of the angular depend-


ences of the SdH oscillation period for
selected samples of Cd3 As2 (CA-1 [51],
CA-2 [63]), (Cd1ÿx Mnx )3 As2 (CMA-1:
x ˆ 0:01; CMA-2: x ˆ 0:0025 [53]) and
(Cd1ÿxÿy Zny Mnx )3 As2 (CZMA-1 and CZMA-2:
x ˆ 0:005, y ˆ 0:13 [28]). The curves repre-
sent Eq. (37)

of the band structure described by the crystal field splitting d by means of the relation
 1=2  
P …0o † f1 d…3E ‡ D† 1=2
KP ˆ ˆ ˆ 1‡ : …39†
P …90o † f2 E …3E ‡ 2D† E ˆ EF

The above equation, combined with the relationship (38) between EF and n, allows one
to estimate d for each material studied. The averaged values of d found for Cd3 As2 [51,
63], CMA (0:5% Mn), CMA (1% Mn) [19, 53] and CZMA (13% Zn and 0:5% Mn) [28]
are equal to about 0.09, 0.07, 0.06 and 0.03 meV, respectively which means that the
crystal field splitting decreases quickly when alloying Cd3 As2 with increasing amounts of
Mn and Zn.
As follows from Eq. (33), the amplitudes of the consecutive harmonics of the SdH
oscillations, determined mainly by the factor Xr = sinh Xr , are modulated by the cos…prn†
term, containing the quantity n proportional to the the effective g-factor g* . As pointed
out previously, the effective g-factor g* of SMSC strongly depends on temperature and
magnetic field which means that the cos…prn† term can affect the SdH amplitude quite
significantly. In particular, when prn ˆ ip=2, where i is an odd integer, the amplitude of
r-th harmonic may vanish as demonstrated for CMA in Fig. 12 for the first harmonic,
i.e. for r ˆ 1; similar results on the SdH effect in CMA are also presented in [64, 65].
The fact that the positions of the experimentally observed zeros (or minima as dis-
cussed below) of the SdH amplitude shift with temperature T, magnetic field B and, for
anisotropic SMSCs, also with the angle q between B and the symmetry axis, allows for
a direct comparison of the theoretical calculations (for a given set of parameters) pre-
sented in Section 3.2 with experimental data. Such data, in turn, can be obtained, for
example, when keeping B constant and determining the angular dependence of the first
harmonic minimum as a function of T and vice versa, after the decomposition of the
total SdH signal into harmonics [28, 29].
Comparison between the anisotropy of the first hamonic minimum position Tmin found
in such a way for CMA (only the data for q ˆ 0 and 90 were available [19, 53]) and
that obtained from our theoretical model is presented in Fig. 15 and an analogous com-
parison for Bmin in CZMA [28, 66] ±± in Fig. 16. The dashed and solid lines in these
figures have been calculated, for a given material, with the same values of the band
332 J. Cisowski

Fig. 15. Angular dependence of the temperature


position of the first harmonic minimum at con-
stant magnetic field for three samples of CMA.
The experimental points are taken from [53].
The solid lines correspond to an anisotropic p±d
exchange interaction (j b? j<j bk j, see Table 3),
while the dashed lines represent the isotropic
case (b? ˆ bk ˆ b) [66]

structure parameters (taking, of course, into account the changes of the energy gap E0
and the crystal field splitting d with composition as discussed earlier), but for two differ-
ent cases as far as the problem of anisotropy of the exchange interaction is concerned. In
the first attempt, we have tried to describe the angular dependences of Tmin and Bmin by
an isotropic exchange interaction putting N0 b? ˆ N0 bk ˆ N0 b (dashed lines). It can be
seen that the band structure anisotropy alone (d 6ˆ 0) is unable to describe even qualita-
tively the experimental results. Therefore, we have allowed for an anisotropy between
N0 b? and N0 bk which drastically changes the character of the angular dependences of
Tmin and Bmin (solid lines) and gives a good agreement with the experimental data for
both CMA and CZMA [28, 66].
Values of the sp±d exchange interaction constants obtained for CMA and CZMA by
our fitting procedure are gathered in Table 3, where, for the sake of completeness, the
value of N0 a ÿ N0 b ˆ 1:2 eV determined for ZMP (with the nominal composition
x ˆ 0:01) from the magnetoreflectivity measurements [26] is also presented. Since the
real amount of Mn in ZMP may be lower than the nominal one (by analogy with an-
other phosphide, CdP2 : Mn [31]), the value of N0 a ÿ N0 b for ZMP may be accordingly
higher (see Eq. (31)).
In the case of CMA and CZMA, N0 a ˆ 0:4 eV, while N0 bk ˆ ÿ…2:04 to 2.28) eV and
N0 b? ˆ ÿ…1:82 to 2.08) eV, depending on composition. Nevertheless, for each composi-

Fig. 16. Angular dependence of the magnetic field


position of the first harmonic minimum at con-
stant temperature for a sample of CZMA. The so-
lid and dashed lines have the same meaning as in
Fig. 15 [28, 66]
Semimagnetic Semiconductors Based on II±V Compounds 333

Table 3
sp±d exchange interaction constants (in eV) for (Cd1 ÿ x Mnx )3 As2 , (Cd1 ÿ x ÿ y Zny Mnx )3 As2
[28, 66] and (Zn1 ÿ x Mnx )3 P2 [26]

material N0 a N0 bk N0 b? N0 a ÿ N0 b

(Cd0:995 Mn0:005 †3 As2 0:4 ÿ2:20 ÿ2:08 2:54*)


(Cd0:99 Mn0:01 †3 As2 0:4 ÿ2:28 ÿ2:00 2:54*†
(Cd0:865 Zn0:13 Mn0:005 †3 As2 0:4 ÿ2:04 ÿ1:82 2:33*)
(Zn0:99 Mn0:01 †3 P2 ÿ ÿ ÿ 1:2
*) Averaged values with N0 b ˆ …N0 bk ‡ N0 b? †=2.

tion, a constant tendency that jbk j > jb? j occurs, leading to a small average relative
anisotropy of these exchange constants …bk ÿ b? †=b?  10%. Thus, our approach based
on the band structure model of a tetragonal SMSC in a quantizing magnetic field gives
strong evidence that the p±d interaction is anisotropic in the investigated materials. As
far as the magnitude of the sp±d exchange interaction constants is concerned, it appears
that the values of N0 a and averaged j N0 b j for CMA and CZMA are higher than those
for II±VI SMCSCs, but they are not as high as presented in the review [7] (N0 a ˆ 3:8
eV and N0 b ˆ ÿ2:3 eV) and, moreover, exhibit considerably less scatter in comparison
with the values found previously for CMA with various Mn concentrations [3, 18, 19,
50].
A detailed analysis of such SdH data as those obtained for CZMA and shown in Fig.
13 indicates that, at higher fields, the amplitude of the first harmonic does not vanish
completely when n ˆ i=2; i ˆ 1; 3; . . ., as described above, but reaches a minimum;
this effect can be explained by the spin-dependent scattering, i.e. by assuming that the
scattering rates for spin-up and spin-down band carriers are not equal [67, 68]. The spin-
dependent scattering can be quantitatively described in terms of the difference Dingle
temperature dTD ˆ …TD‡ ÿ TDÿ †=2, where TD‡ and TDÿ are the Dingle temperatures for
spin-up and spin-down electrons, respectively and the results of theoretical calculations
of dTD for CZMA as a function of the square of the Mn spin hSz i (which is a good
measure of the magnetic ordering dependent on the field and temperature [29]), together

Fig. 17. Difference Dingle temperature dTD ˆ


…TD‡ ÿ TDÿ †=2 as a function of the square of the
Mn spin in three samples of CZMA with electron
concentrations of 8:98  1017 (a), 2:34  1018 (b)
and 3:75  1018 cmÿ3 (c). The solid lines a, b and
c are the theoretical curves corresponding to the
results deduced from experiment and denoted by
triangles up, triangles down and circles, respec-
tively [29]
334 J. Cisowski

Fig. 18. SdH oscillations for a sample


of CZMA (x ˆ 0:006; y ˆ 0:03) with
the electron concentration of
4:0  1018 cmÿ3 at 0:25 K [69]

with the data deduced from experiment, are shown in Fig. 17. Bearing in mind, on one
hand, many assumptions and approximations made in calculations and, on the other
hand, the lack of fitting parameters, it may be stated that the theoretical model is in
satisfactory agreement with experiment giving strong evidence that the spin-dependent
scattering is operating in CZMA.
As is known, the SdH data, and especially those of SMSCs, can be also analyzed using
the fast Fourier transform (FFT) technique which enables one to separate the contribu-
tion of each harmonic to the total SdH signal. An example of such an analysis is shown
in Figs. 18 and 19, where the SdH oscillations for a sample of CZMA recorded at 0:25 K
and their FFT spectrum, respectively, are presented. This method appears to be very
efficient allowing one to observe many SdH harmonics which can be analyzed further
yielding valuable information on the scattering mechanism and band structure of
SMSCs [69].

4.2 Classical transport


In contrast to CMA and CZMA presented above, the other member of II3 -As2 SMSCs,
namely ZMA is, similarly as its host compound Zn3 As2 , a wide-gap and low-mobility p-
type material for which mostly the classical conductivity and magnetoresistance have

Fig. 19. Fast Fourier transform (FFT) spec-


trum of the SdH oscillations for a sample of
CZMA from Fig. 18 [69]. Arrows indicate the
SdH frequencies of the consecutive harmonics
Semimagnetic Semiconductors Based on II±V Compounds 335

Fig. 20. Temperature dependence of the electri-


cal conductivity of ZMA with different composi-
tions. Full and open symbols after [25] and [71],
respectively

been studied. There exist also some Hall


data [25, 70, 71] which give the hole con-
centration and mobility at T ˆ 77 K for
0:01  x  0:13 as lying in the ranges (2 to
5) 1017 cmÿ3 and (0.1 to 2) cm2 /Vs, re-
spectively.
The temperature dependence of the elec-
trical conductivity for samples of ZMA
with Mn concentrations 0  x  0:13 [25,
71] are shown in Fig. 20. As can be ob-
served, the samples with x  0:02 have
much higher conductivity than those with
larger x suggesting that Mn in ZMA acts as a donor which compensates the p-type con-
ductivity of the Zn3 As2 matrix: On the other hand, the results in Fig. 20 indicate that
conductivity for all the samples investigated has activational character below T  100 K.
The activation energy at 4K < T < 25 K and at 25 K < T < 100 K has been found to be
equal to 0.1 to 0.9 and 0.3 to 7.5 meV, respectively, depending on composition and the
obtained data have been analyzed within a model based on the disorder of Mn atoms
over the vacancy-type sites of the cation sublattice [72, 73].
An example of the magnetoresistance (MR) data for samples of ZMA with different
compositions [74] is shown in Fig. 21. As can be seen, MR of all the samples studied
changes its sign at certain fields evidencing that ZMA has both positive and negative
contributions to MR which, however, do not depend on composition in a simple way. A
qualitative explanation of this effect is based on the phenomenon of spin-flip scattering
of holes by the magnetic moments of Mn ions which may take place during transitions
of holes between acceptor centres [74].

Fig. 21. Field dependence of the relative mag-


netoresistance of samples of ZMA with differ-
ent compositions at 4:2 K [74]
336 J. Cisowski

5. Magnetic Properties
The magnetic properties of Mn-alloyed SMSCs are, in principle, determined by the d±d
exchange interaction between the localized Mn ions forming an unordered magnetic sub-
system. Information about the exchange can be drawn from various experimental data,
including magnetization, susceptibility, specific heat and phase transitions. In recent
years, also Mn-alloyed II±V compounds have attracted considerable interest as being
characterized, on the one hand, by strong d±d interaction [21, 22, 75 to 77] and, on the
other hand, by an unusual behaviour of susceptibility interpreted as a magnetic freezing
at temperatures as high as T  200 K [24, 78 to 82].
As mentioned in Section 2, the susceptibility measurements can provide information
on the homogeneity of the materials studied and this is illustrated in Fig. 22, where
some representative results of experiments performed in similar conditions for various
Mn-alloyed II3 ±V2 arsenides [79 to 83] are shown. It can be observed that the suscept-
ibility of samples 1 and 2 exhibits an upturn in the whole investigated temperature
range, i.e. a typical paramagnetic behaviour. In contrast to these samples, the suscept-
ibility of samples 3 to 5 is much larger and, moreover, behaves, at higher temperatures,
quite different, demonstrating a downturn at T  200 K followed by an abrupt decrease
at about 317 K which correponds to the temperature of the ferromagnetic-paramagnetic
transition in MnAs [42]. Thus, one may suppose that samples 3 to 5 contain ferromag-
netic clusters of MnAs and their number depends on the crystal growth conditions. It is
worthy to note that a similar effect occurs for another Mn-alloyed arsenide, i.e. In1 ÿ x Mnx As
which can be homogeneous from the point of view of Mn distribution or can exhibit the
presence of the MnAs-like phase, depending on the growth conditions during the MBE
process [12].
As is known, some basic information on the magnetic interactions can be directly
drawn from experiment and, in particular, from magnetization and susceptibility data.
Therefore, below we present first a few examples of such data obtained for Mn-alloyed
II±V SMSCs along with a phenomenological description followed by a deeper analysis
which takes into account possible physical mechanisms of the d±d interaction and the
crystal structure of the materials of interest within our approach called the generalized
pair approximation [75].

Fig. 22. Temperature dependence of the total


mass susceptibility of Mn-alloyed II3 ±V2 ar-
senides with various nominal compositions. Data
after [79 to 83]
Semimagnetic Semiconductors Based on II±V Compounds 337

Fig. 23. High-field magnetization


of ZMA [22] and CZMA [76, 77]
with small Mn concentration
(x  0:5%†. Solid lines represent
fits to Eq. (40), containing the
modified Brillouin function, with
M0 ˆ 4:6 and 4:3 mB for ZMA and
CZMA, respectively

5.1 Magnetization and susceptibility


As mentioned in Section 3, the description of sp±d interaction in SMSCs requires a knowl-
edge of the average spin of magnetic ions hS iB and this can be achieved by measurements
of the macroscopic magnetization M given by M ˆ ÿNn g mB hS iB ; where Nn and g are the
number and the Lande factor of magnetic ions, respectively. In the case of very low concen-
trations of Mn2‡ ions, hS iB may be expressed by the standard Brillouin function
BS …g mB SB=kB T † with S ˆ 5=2 and g ˆ 2 [6]. As x increases, the ions start to interact
between each other and the standard Brillouin function is inadequate for explaining the
magnetization data. There is, however, still a qualitative similarity between the Brillouin
function and the experimentally measured magnetization which results in a very useful
phenomenological formula containing the so-called modified Brillouin function [56]
 
gmB SB
M ˆ M 0 BS …40†
kB …T ‡ T0 †
and two fitting parameters T0 and M0 , the latter being known as the apparent or techni-
cal saturation magnetization [84]. This approach has been also used for Mn-alloyed II±V
SMSCs [21, 22, 85 to 87] and the representative high-field data for ZMA [22] and CZMA
[77] with similar, very low Mn concentration (x  0:5%), are shown in Fig. 23, where
the presented results have been corrected for the diamagnetic contribution Md ˆ cd B

Fig. 24. Temperature dependence of the in-


verse paramagnetic molar susceptibility of
ZMA [22] and ZMP [26] with x ˆ 1%, cor-
rected for the diamagnetic contribution (see
Table 4). The straight lines represent fits to
the Curie-Weiss law given by Eq. (41)
338 J. Cisowski

(with cd , being the diamagnetic susceptibility), as discussed below. Fits to Eq. (40) pro-
vide the saturation magnetic moment per Mn ion M0 ˆ gmB S0 (with S0 , the saturation
value of hS iB ) as being equal to 4:1 and 4:3mB for ZMA and CZMA, respectively. These
results together with the others presented in [21, 22, 75 to 77] show that (i) even in the
limit of vanishingly small x, M0 is less than 5mB of a free Mn2‡ ion in the ground state
and (ii) M0 decreases with increasing Zn content in CZMA and both effects are con-
nected with the p±d hybridization [22, 48].
Mn-alloyed II±V compounds have been subjected to the extensive studies of the mag-
netic susceptibility c in a wide range of temperature. At sufficiently low temperatures, c
exhibits a cusp interpreted as a transition to a spin-glass state as discussed later on in
Section 5.3, while at higher temperatures, c shows a typical Curie-Weiss behaviour indi-
cating dominant antiferromagnetic Mn±Mn interactions and, similarly as in the case of
magnetization data, a reduced spin value with respect to S ˆ 5=2 of a free Mn2‡ ion, as
presented in a number of works devoted to CMA [21, 88], ZMA [22, 24, 25, 78, 80] and
CZMA [76, 77, 83]. Apart from these relatively well known SMSCs (which will be de-
scribed in Section 5.3), some susceptibility measurements have been also performed for
ZMP with x ˆ 1% [26] and CdP2 :Mn [31] as shown in Figs. 24 and 25, respectively.
Additionally, for the purpose of comparison, Fig. 24 contains the results obtained for
CMA also with x ˆ 1% [21] and all these data are corrected for the diamagnetism of the
host semiconductor.
As is known, the measured total susceptibility ctot consists of two terms, the tempera-
ture dependent contribution of magnetic ions c and the temperature independent dia-
magnetic contribution cd which should not be neglected when analyzing the magnetiza-
tion and suceptibility data for low x and high temperatures. For CdP2 ,
cd ˆ ÿ4:15  10ÿ7 emu/g is reported [89,31] and for II3 ±V2 compounds, the values of cd
are gathered in Table 4.
As follows from Eq. (40), in the low field or high-temperature limit, M is linear in B
and the net molar susceptibility of magnetic ions …c ˆ ctot ÿ cd ˆ M=B† can be ex-
pressed in the Curie-Weiss form as [90]
1 T ÿ q…x†
ˆ …41†
c C…x†

Fig. 25. Temperature dependence of the inverse


paramagnetic mass susceptibility of CdP2 doped
with various amounts of Mn [31]
Semimagnetic Semiconductors Based on II±V Compounds 339

Table 4
Diamagnetic contribution to susceptibility and sum of all interactions in Mn-alloyed
II3 ±V2 semimagnetic semiconductors
P
material cd (10ÿ7 emu/g) Jp zp (K)
p

(Cd1 ÿ x Mnx †3 As2 ÿ3 [21] ÿ164 [21, 98]


(Cd1 ÿ x Zn0:86 Mnx †3 As2 ÿ2:9* ) ÿ240 [83]
(Cd1 ÿ x Zn0:66 Mnx †3 As2 ÿ2:7*) ÿ304 [83]
(Zn1 ÿ x Mnx †3 As2 ÿ2:2 [22]; ÿ2 [24] ÿ480 [22, 98]
(Zn1 ÿ x Mnx )3 P2 ÿ1:6**) ÿ 84***
*) Interpolated between those measured for Cd3 As2 [21] and Zn3 As2 [22, 24].
**) Extrapolated assuming that cd of II3 ±V2 compounds is proportional to their molar
mass.
***) Deduced from Fig. 24.

with the Curie constant


P C…x† ˆ Nm …g mB †2 S…S ‡ 1†=3kB ; the Curie-Weiss temperature
q…x† ˆ 2xS…S ‡ 1† Jp zp =3kB and Nm ˆ 3xNA for II3 ±V2 SMSCs. In these expressions,
p
Nm is the number of magnetic ions per mol, NA is Avogadro0 s number, Jp is an ex-
change constant between the reference magnetic ion and another magnetic ion located
on the p-th coordination sphere and zp is the number of cation sites on consecutive
spheres.
As can be seen from Fig. 24, at higher temperatures, the inverse susceptibility data of
both CMA and ZMP indeed follow straight lines given by Eq. (41) allowing one to
determine the Curie-Weiss temperature and the Curie constant which depend linearly
on x. Another interesting feature which follows from the above expressions is that the
ratio
Q 2 P
ˆ 2
Jp zp …42†
C 3NA …gmB † p
contains the sum of all interactions multiplied by appropriate numbers of cation sites on
consecutive coordination spheres but does not depend on S and x, the latter being often
known only approximately. By P averaging the values of Q=C for a particular SMSC over
x, the final values of the sum Jp zp (which will be useful in the further analysis) can
p
be found and those determined for Mn-alloyed II3 ±V2 s are gathered in Table 4. It
appears that the sum of all interactions depends strongly on composition, increasing in
arsenides when passing from CMA to ZMA, i.e. with the cation substitution and being
relatively small in the phosphide ZMP which indicates the important role of the anion
forming the p-like valence band in these materials and therefore influencing the p±d
hybridization potential.
Finally, we comment on the susceptibility data of CdP2 : Mn which are shown in Fig.
25. One could expect that for such low Mn concentrations, this material should behave
like an ideal paramagnet described by the Curie law, i.e. by Eq. (41) with q ˆ 0 K:
Looking at the data in Fig. 25, one can see that only the sample with the lowest Mn
concentration (x ˆ 0:032%) follows the Curie behaviour, while for other samples the
least-squares straight lines do not pass through zero which has been interpreted as being
due to the formation of magnetic clusters [31].
340 J. Cisowski

5.2 d±d interaction and generalized pair approximation


From the point of view of the magnetic interaction, a SMSC may be treated as an
ensemble of magnetic moments (spins) arising from magnetic ions randomly distributed
over the crystal host lattice. The thermodynamic properties of such system can only be
calculated in an approximative way, one of which is the pair approximation [91, 21, 22]
applicable, in principle, to SMSCs with simple crystal structures in which each cation
site has the same arrangement of other cations which implies only one nearest-neighbour
(NN) distance with a corresponding exchange constant J1 , a unique next-nearest-neigh-
bour (NNN) distance with J2 , etc. However, as described in Section 2, the cation sublat-
tice of Mn-alloyed II3 ±V2 compounds is severely distorted, which results, in the case of
ZMA, in 12 different values for the NN distances, 28 NNN distances, etc. [23, 75] and
requires another approach, as described below.
The basic assumption of the pair approximation is that the partition function of the
system may be factorized into contributions of pairs of spins. Thus, each spin is consid-
ered to belong to one pair formed with its nearest magnetic neighbour (NMN) which
may be located on any lattice site, and the spins belonging to different pairs are treated
as noninteracting.
The Hamiltonian for a pair of magnetic ions with the orbital momentum L ˆ 0 and in
the presence of an external magnetic field B can be expressed in the Heisenberg form as
Hp ˆ ÿgmB …Si ‡ SNMNi †  B ÿ 2Ji Si  SNMNi ; …43†
where Ji ˆ J…Ri † is an exchange constant between spins Si and SNMNi , separated by a
distance Ri . Thus, the total interaction is given by H ˆfigp Hp with the summation run-
ning over all pairs fi; NMNi g.
Generalizing the pair approximation for arbitrary structures characterized by a num-
ber of inequivalent cation sites, we have calculated the total free energy F; from which
the magnetic properties of the system can be derived, as [75]
N P t P 1
Fˆ P …x† Fnj …J…Rnj †; B† ; …44†
t j ˆ 1 nj nj
where Fnj is the free energy of a pair of interacting spins separated by a distance Rnj ,
Pnj …x† is the probability of finding such a pair, N is the total number of spins and t is
the number of inequivalent cation sites.
Apart from probabilities which can be easily calculated if the crystal structure of the
system in question is known, Eq. (44) requires to specify the exchange constants, in
principle, for every kind of pair of spins. Bearing in mind a great number of various
pairs of spins in Mn-alloyed II3 ±V2 SMSCs with the corresponding number of exchange
constants, it is clear that one has to search for a distance dependence of the interaction
strength in these materials.
According to extensive theoretical studies, there are two essential exchange interac-
tion mechanisms in Mn-alloyed SMSCs [48, 92, 93], the antiferromagnetic superex-
change, i.e. an indirect Mn±Mn exchange mediated by the anion and another indirect
process ±± the Bloembergen-Rowland (BR) exchange ±± induced by the virtual interband
transitions. The simple analytical formula describing the radial dependence of superex-
change J SE …R† is given by [48]
4  
ÿ2Vpd 1 1
J SE …R† ˆ ÿ f…R=a† ; …45†
…Ev ÿ Ed ÿ Ueff †2 Ueff Ev ÿ Ed ÿ Ueff
Semimagnetic Semiconductors Based on II±V Compounds 341

where f…R=a† ( with a the lattice constant) is a material-insensitive function approxi-


mated by f…R=a† ˆ 51:2 exp …ÿ5:16R2 =a2 †. Comparison of this formula with Eq. (16)
for the p±d exchange constant gives J SE / …N0 b†2 which means that both interactions
are strongly correlated.
Rearranging Eq. (45) in order to express it as a function of r ˆ R=R1 , i.e. in units of
the first NN distance R1 and the first NN exchange constant J1SE , one gets
J SE …r† ˆ J1SE exp ‰aSE …r2 ÿ 1†Š …46†
with aSE ˆ 5:16R21 =a2 :
Similarly, the radial dependence of the BR exchange [94, 95]
J BR …r† can be written as
J BR
J BR …r† ˆ 13 exp ‰ÿaBR …r ÿ 1†Š …47†
r
ÿ 1=2
with aBR ˆ R1 2me Eg =
h, where me and Eg are the electron effective mass and the
energy gap, respectively.
Thus, the total interaction strength is given by
J…r† ˆ J SE …r† ‡ J BR …r† …48†
and contains only two unknown parameters, i.e. the first NN constants for both mechan-
isms J1SE and P
J1BR treated as adjustable parameters but with the additional condition
that the sum Jp zp should be equal to that determined experimentally from the Curie-
p
Weiss law (see the previous section).
Based on Eq. (44), one can derive the other thermodynamic quantities like magnetiza-
tion (M ˆ ÿ‰@F =@BŠT ), susceptibility (c ˆ ÿ‰@ 2 F =@B2 ŠT ) and the magnetic specific heat
(Cm ˆ ÿT ‰@ 2 F =@T 2 ŠB ) and compare them with experiment. Some representative exam-
ples of such an analysis performed for CZMA [76, 77, 96 to 98] (including CMA [21]
and ZMA [22] as the special cases of CZMA with y ˆ 0 and y ˆ 1 ÿ x, respectively),
which has allowed one to describe simultaneously all those experimental quantities, are
shown in Figs. 26, 27 and 28. An overall good agreement between our approach based
on the above equations and experiment allows us to state that superexchange is the
main exchange mechanism in Mn-alloyed II3 ±V2 arsenides at short distances and be-
comes negligible for larger distances, where the long-ranged BR mechanism plays a
major role.

Fig. 26. High-field magnetization of two


samples of CZMA with different composi-
tions. Solid lines represent the GPA calcula-
tions [96]

23 physica (b) 200/2


342 J. Cisowski

Fig. 27. Temperature dependence of the inverse


susceptibility per mol Mn of CZMA with similar
Mn content and various amount of Zn. Experimen-
tal data are taken from [21, 22, 83] and the solid
lines represent the GPA calculations [98]

The best fitted values of the exchange


constants obtained for Mn-alloyed II3 ±V2 ar-
senides are presented in Table 5. It appears
that the interaction strength of superex-
change represented by J1SE and giving the
main contribution to the total first NN con-
stant J1 , strongly increases with Zn content.
This can be partially explained by a de-
crease of the first NN (cation±cation) distance R1 which changes from 0:32 to 0:29
nm, when passing from CMA to ZMA and the remaining increase of J1SE may be
attributed to an increase of the p±d hybridization potential Vpd with increasing
amount of Zn. The substitution of Zn for Cd in CZMA, as far as Vpd is concerned,
may be treated as an indirect effect of the cation on the Mn-As bond length d, simi-
larly as in Mn-alloyed II±VI SMSCs [48]. As follows from theoretical works [48, 92],
both J1SE and J1BR are proportional to Vpd
4
(see e.g. Eq. (45)) and Vpd , in turn, is pro-
portional either to dÿ4 or to dÿ7=2 [99, 100] which finally gives a very strong depen-
dence of the exchange constants on the Mn-As bond length as being proportional
either to dÿ16 or to dÿ14 . This means that, on the whole, even a relatively small
change of all interionic distances may result in a significant change of the exchange
constants, as illustrated in Table 5 for Mn-alloyed II3 ±V2 arsenides and confirmed by
recent pressure studies of magnetization in CZMA [101]. It is also worthy to notice
that the GPA [75 to 77], which takes into account the real crystal structure of these
materials, results in considerably lower values
of the d±d exchange constant than those pre-
sented for CMA (J1 ˆ ÿ30 K) and ZMA
(J1 ˆ ÿ100 K) in the reviews [8, 9] and ob-
tained by assuming an idealized quasi-cubic
structure [21, 22].

Fig. 28. Low-temperature magnetic specific heat per


mol Mn in zero field of the same samples of CZMA
as in Fig. 27. Experimental data are taken from [21,
22, 97] and the solid lines represent the GPA calcu-
lations [98]
Semimagnetic Semiconductors Based on II±V Compounds 343

Table 5
d±d exchange interaction constants for Mn-alloyed II3 ±V2 arsenides [75 to 77, 98]

material J1SE (K) J1BR (K) J1 (K)

(Cd1 ÿ x Mnx †3 As2 ÿ12 ÿ 4 ÿ16


(Cd1 ÿ x Zn0:86 Mnx †3 As2 ÿ19 ÿ 5:5 ÿ24:5
(Cd1 ÿ x Zn0:66 Mnx †3 As2 ÿ25 ÿ 7 ÿ32
(Zn1 ÿ x Mnx †3 As2 ÿ53 ÿ11 ÿ64

5.3 Magnetic freezing


The low-field susceptibility measurements of SMSCs at sufficiently low temperatures
reveal the presence of a cusp or kink, interpreted as a transition from the paramagnetic
to the spin-glass state [6, 8] and the corresponding temperature Tf , known as the freez-
ing temperature, strongly depends on the magnetic ion concentration. In the case of Mn-
alloyed II3 ±V2 compounds, studies of the low-temperature spin freezing have been per-
formed for CMA [21, 88, 102, 103], ZMA [22, 25, 80] and CZMA [22, 88] and the results
for the first two materials are shown in Fig. 29. An interpretation of such data is usually
based on the scaling analysis which combines the average distance Rav between the mag-
netic ions with their concentration by the equation R3av x ˆ const together with the con-
jecture that Tf is proportional to the magnetic energy J of interacting Mn ions [7, 21,
22, 75, 104], leading, in fact, to an equality Tf …x†  J…x† [105].
As can be seen from Fig. 29, the spin freezing is experimentally observed even for very
small x which corresponds to a large Rav and indicates the presence of a long-range
exchange mechanism. This fact, together with the observation that the Tf …x†-depen-
dences for ZMA (with the energy gap Eg  1:1 eV) and CMA (with Eg  0) form two
separate sets of data, allow one to identify this mechanism with the BR exchange which
is long-ranged and depends on the energy gap (see Eq. (46)).
The result of our calculations based on the above-mentioned assumptions are shown
in Fig. 29 as the solid lines and appear to describe very well the available experimental
data up to about 10% Mn. A similar analysis performed up to x ˆ 0:25 in Mn-alloyed
II±VI SMSCs [106] indicates that, for x  0:125; an additional exchange mechanism
comes into play in all Mn-alloyed SMSCs which may be connected with a medium-range
superexchange [107].

Fig. 29. Spin-glass freezing temperature as a


function of the Mn concentration in CMA and
ZMA. Experimental data for CMA (triangles
up) and ZMA (triangles down) are taken from
[21, 88, 102, 103] and [22, 25, 80], respectively
and the solid lines are calculated according to
the Bloembergen-Rowland (BR) expression
(47)

23*
344 J. Cisowski

Fig. 30. Temperature dependence of


susceptibility measured on samples of
CMA [81] and ZMA [79] with x ˆ 0:1,
cooled in zero field (ZFC) and in a gi-
ven field (FC). Arrows correspond to
the temperature at which the ZFC and
FC data deviate from each other

Some very low-field susceptibil-


ity measurements performed for
ZMA [25, 78 to 80, 82] and CMA
[81] have revealed an unusual tem-
perature behaviour resembling
that of a spin-glass, as shown in
Fig. 30 for the selected samples
with a similar Mn concentration.
It can be observed that the field
cooling (FC) and zero field cooling (ZFC) data deviate from each other at a temperature
denoted by Tf , being equal to 212 and 195 K for CMA and ZMA, respectively. Tenta-
tively, such a peculiar behaviour has been attributed to the presence of various size
magnetic clusters which can persist even at elevated temperatures [80, 81]. Quite re-
cently, a deeper analysis of the anomalies in low-field susceptibility data of ZMA and
CMA has been proposed [73, 82] basing on the structural disorder in the crystal lattice
of these alloys, connected with the presence of built-in vacancies in the cation sublattice
as described in Section 2 (see Fig. 2). The existence of such vacancies enables diffusion
of a small fraction of Mn ions over the vacant sites and leads to formation of complex
centres with large magnetic moments, called clusters of polarization which, according to
a rough estimation, can remain even up to about 200 K [73].
Similar measurements of susceptibility have been performed for the recently synthe-
sized (Zn1 ÿ x Fex )3 As2 (ZFA) [30] and the results for a sample with x ˆ 0:01; cooled in

Fig. 31. Temperature dependence of susceptibil-


ity measured on a sample of ZFA with
x ˆ 0:01, cooled in zero field (ZFC) and in dif-
ferent fields (FC). Arrows correspond to tem-
peratures at which the ZFC and FC data devi-
ate from each other [30]
Semimagnetic Semiconductors Based on II±V Compounds 345

Fig. 32. Temperature dependence of the


magnetic moment in a single crystal of
ZnMn2 As2 , cooled in zero field (ZFC)
and in a given field (FC). Curves 1 and
2 correspond to the field direction paral-
lel and perpendicular to the hexagonal
c-axis, respectively. The ferrimagnetic
transition [110] is indicated by TC , the
helical ordering by Th and the spin-glass
transition by Tf [40, 41]

the field range 0 to 0.2 T, are shown in Fig. 31. These data reveal freezing of magnetic
moments in ZFA, strongly dependent on the applied magnetic field, at Tf < 50 K which
is much lower than bserved for CMA and ZMA (see Fig. 30).
It is worthy to add that a similar high-temperature spin-freezing-like behaviour of
susceptibility as in Figs. 30 and 31 has been also observed in the quaternary alloy
Cd1 ÿ x ÿ z Mnx Fez Te with x ˆ 0:37 and z ˆ 0:01 [108] and has been interpreted as being
due to formation of the bound magnetic polaron (BMP), i.e. a collective state consisting
of an impurity-bound carrier and magnetic ions polarized by the carrier in its vicinity.
In view of the works discussed above, it is very difficult at the present stage to give a
convincing explanation of susceptibility anomalies observed in some samples of II3 ±V2
SMSCs, the more so as the presence of ferromagnetic clusters in the paramagnetic ma-
trix of Mn-alloyed II±V SMSCs cannot be excluded (see Fig. 23), even for a small
amount of magnetic ions, as follows, for example, from the data obtained for another
SMSC ±± Hg1 ÿ x Cox Se ±± which indicate the presence of ferromagnetic CoSe clusters
even for x  0:87% [109].
Finally, we present some selected results of magnetic measurements performed for
ZnMn2 As2 , the ordered version of ZMA with x ˆ 2=3 [40, 41]. Fig. 32 shows a compar-
ison between the FC and ZFC data in both crystal directions, i.e. for B parallel
(curves 1) and perpendicular (curves 2) to the hexagonal c-axis, in fields 10 mT and 1 T.
Apart from an anisotropy of the magnetic behaviour, this figure shows that ZnMn2 As2
may occur in many different magnetic phases, when changing the temperature. At
TC  310 K, there is a transition from paramagnetism to ferrimagnetism [110], while at
Th  110 K, a partial antiferromagnetic ordering, i.e. a helical magnetic ordering sets in,
followed by a spin-glass transition at Tf  33 K. As pointed out in [41], the magnetic
properties of ZnMn2 As2 are reflected in its transport properties even at zero magnetic
field, since the scattering process of carriers (holes in this case) is essentially determined
by the magnetic structure of the material.

6. Conclusions
The most interesting features of semimagnetic semiconductors (SMSCs based on II±V
compounds can be summarized as follows:
346 J. Cisowski

±± the first Mn±Mn distance in these materials is the smallest one (equal to 0:29 nm
in (Zn1 ÿ x Mnx )3 As2 (ZMA) and (Zn1 ÿ x Mnx )3 P2 ) among known SMSCs and a quasi-con-
tinuous spectrum of these distances occurs, providing an unusual possibility to study the
radial dependence of the d±d interaction strength;
±± Mn-alloyed narrow-gap II3 ±V2 arsenides such as (Cd1ÿx Mnx )3 As2 (CMA) and
(Cd1 ÿ x ÿ y Zny Mnx )3 As2 (CZMA) with a small Zn content are found to be characterized
by a strong p±d interaction (with the exchange constant N0 b  ÿ2 eV, as compared
with N0 b  ÿ1 eV in II±VI SMSCs which, additionally, exhibits an anisotropy (of
about 10%) originating from their tetragonal crystal structure. On the other hand, the
s±d interaction constant N0 a  0:4 eV, as estimated for these materials, is somewhat
higher than that of II±VI SMSCs with N0 a  0:3 eV;
±± spin-dependent scattering of band electrons on the magnetic moments of Mn ions,
ordered by the field, is found to operate in CZMA;
±± Mn-alloyed II3 ±V2 arsenides exhibit a strong antiferromagnetic d±d interaction
(highly correlated with the p±d interaction) which, additionally, depends on the cation;
this is illustrated by the values of the first nearest-neighbour exchange constant which
are found to change from J1 ˆ ÿ16 K in CMA to J1 ˆ ÿ64 K in ZMA, as compared
with those of Mn-alloyed II±VI SMSCs, where J1 ˆ …ÿ5 to 15) K. The observed change
of J1 , when substituting Zn for Cd in CZMA, follows from (i) a decrease of the Mn±Mn
distance (from 0:32 nm in CMA to 0:29 nm in ZMA) and (ii) an increase of the hybridi-
zation potential Vpd , when passing from CMA to ZMA which, in turn, depends strongly
on the Mn±As bond length.
Finally, we would like to notice that the values of the sp±d and d±d exchange inter-
action constants given above have been obtained as a result of a quantitative analysis
involving many approximations and assumptions and therefore these values should be
treated as estimates, requiring further experimental and theoretical studies. We believe
that the interesting phenomena found in Mn-alloyed II±V compounds and described in
this review will stimulate a better understanding of the nature of both sp±d and d±d
interactions in these materials as well as in SMSCs in general.

Acknowledgements The author is very grateful to all of the coworkers who partici-
pated in obtaining the results presented in this article, especially H. Bednarski, W. Lub-
czy
nski, J. Kossut and A. Burian from Polish Academy of Sciences and J. C. Portal,
J. Voiron and D. Schmitt from CNRS, France.

References
[1] R. R. GaøaËzka, Proc. 14th Internat. Conf. Phys. Semicond., Edinburgh 1978, Inst. Phys.
Conf. Ser. No. 43, 133 (1979).
[2] R. R. GaøaËzka and J. Kossut, in: Landolt-Bornstein, Numerical Data and Functional Rela-
tionships in Science and Technology, Vol. 17b, Eds. O. Madelung, M. Schultz, and
H. Weiss, Springer-Verlag, Berlin 1982 (p. 302).
[3] J. K. Furdyna and J. Kossut (Ed.), Diluted Magnetic Semiconductors, in: Semicond. and
Semimet., Vol. 25, Academic Press, Boston 1988.
[4] M. Averous and M. Balkanski (Ed.), Diluted Magnetic Semiconductors, Plenum Press,
New York 1991.
[5] M. Jain (Ed.), Diluted Magnetic Semiconductors, World Scientific Publ Co., Singapore 1991.
[6] J. K. Furdyna, J. appl. Phys. 64, R29 (1988)
[7] A. Twardowski, Acta phys. Polon. A 75, 327 (1989).
Semimagnetic Semiconductors Based on II±V Compounds 347

[8] W. J. M. de Jonge, and H. J. M. Swagten, J. Magnetism magnetic Mater. 100, 322 (1991).
[9] G. Bauer, H. Pascher, and W. Zawadzki, Semicond. Sci. Technol. 7, 703 (1992).
[10] R. R. GaøaËzka, Mater. Sci. Forum 182/184, 371 (1995).
[11] T. Story, R. R. GaøaËzka, P. J. T. Eggenkamp, H. J. M. Swagten, and W. J. M. de
Jonge, Mater. Sci. Forum 182/184, 477 (1995).
[12] H. Ohno, H. Munekata, S. von Molnar, and L. L. Chang, J. appl. Phys., 69, 6103
(1991).
H. Ohno, Mater. Sci. Forum 182/184, 443 (1995).
[13] E. K. Arushanov, Progr. Crystal Growth 25, 131 (1992).
[14] D. N. Nasledov and V. Ya. Shevchenko, phys. stat. sol. (a) 15, 9 (1973).
[15] _
W. Zdanowicz, in: Landolt-Bornstein, Numerical Data and Functional Relationships in
Science and Technology, Vol. 17e, Ed. O. Madelung, Springer-Verlag, Berlin 1983 (p. 178).
[16] J. Cisowski, phys. stat. sol. (b) 111, 289 (1982).
[17] E. K. Arushanov, Progr. Crystal Growth and Charact. 13, 1 (1986).
[18] F. A. P. Blom and J. J. Neve, Proc. 1st Internat. Symp. Physics and Chemistry of II±V
Compounds, Mogilany (Poland) 1980, Eds. M. J. Gelten and L. _Zdanowicz, Eindhoven
University of Technology (The Netherlands) 1980 (p. 83).
J. J. Neve, C. J. R. Bouwens, and F. A. P. Blom, Solid State Commun. 38, 27 (1981).
J. J. Neve and F. A. P. Blom, in: Lecture Notes Phys. 152, 330 (1982).
[19] J. J. Neve, Ph. D. Thesis, Eindhoven 1984, unpublished.
[20] _
W. Zdanowicz, _
K. Kloc, A. Burian, B. Rzepa, and E. Zdanowicz, Crystal Research
and Technology 18, K25 (1983).
Z. Celin _
 ski, A. Burian, B. Rzepa, and W. Zdanowicz, Mater. Res. Bull. 22, 419 (1987).
[21] C. J. M. Denissen, H. Nishihara, J. C. van Gool, and W. J. M. de Jonge, Phys. Rev.
B 33, 7637 (1986).
[22] C. J. M. Denissen, S. Dakun, K. Kopinga, W. J. M. de Jonge, H. Nishihara,
T. Sakakibara, and T. Goto, Phys. Rev. B 36, 5316 (1987).
[23] G. C. de Vries, Crystallographic and Magnetic Properties of (C6 D11 ND3 )CuBr3 and
(Zn1ÿx Mnx )3 As2 : A Neutron Scattering Study, Netherlands Energy Research Foundation,
Petten 1989 (p. 127).
G. C. de Vries, E. Frikke, R.B. Helmholdt, K. Kopinga, and W. J. M. de Jonge,
Physica 156/157B, 321 (1989).
[24] R. Laiho, A. V. Lashkul, E. La  hderanta, L. Sa isa
 , and V. S. Zakhvalinski, Proc.
20th Internat. Conf. Phys. Semicond., Thessaloniki (Greece) 1990, Eds. E. M. Anastassaki
and J. D. Joannopoulos, World Scientific Publ. Co., Singapore 1990 (p. 759).
[25] V. A. Kulbachinski, I. V. Svistunov, S. M. Chudinov, V. D. Kuznetzov, E. K. Ar-
ushanov, V. S. Zakhalinski, and A. N. Nateprov, Fiz. Tekh. Poluprov. 25, 2201
(1991).
[26] J. Misiewicz, L. Bryja, and A. Twardowski, Japan. J. appl. Phys. 32, Suppl. 32-3, 382
(1993).
[27] W. Lubczyn _
ski, J. Cisowski, J. C. Portal, and W. Zdanowicz, Acta phys. Polon. A
73, 175 (1988).
[28] W. Lubczyn  ski, J. Cisowski, J. Kossut, and J. C. Portal, Solid State Commun. 77,
541 (1991).
[29] W. Lubczyn  ski, J. Cisowski, J. Kossut, and J. C. Portal, Semicond. Sci. Technol. 6,
619 (1991).
[30] R. Laiho, A. V. Lashkul, E. La  hderanta, V. A. Stamov, and V. S. Zakhvalinski,
J. Magnetism magnetic Mater. 140/144, 1769 (1995).
[31] G. I. Bondar, V. V. Slynko, and V. M. Frasunyak, Ukr. fiz. Zh. 35, 1700 (1990).
[32] H. Mohan and R. G. Kulkarni, Solid State Commun. 74, 1213 (1990); 75, 1001 (1990).
[33] J. Bodnar, Proc. Internat. Conf. Narrow-Gap Semiconductors, Warsaw 1977, Eds. J.
Rauøuszkiewicz, M. Go  rska, and E. Kaczmarek, Polish Scientific Publishers, Warsaw
1978 (p. 311)
[34] J. Cisowski, E. K. Arushanov, J. Bodnar, K. Kloc, and W. Zdanowicz, _ Proc. 14th
Internat. Conf. Phys. Semicond., Edinburgh 1978, Ed. B. L. H. Wilson, Inst. Phys. Conf.
Ser. No. 43, 253 (1979).
[35] R. Juza and R. Kroebel, Z. anorg. allg. Chemie 331, 187 (1964).
348 J. Cisowski

[36] G. I. Makovetski, V. M. Ryzhkovski, V. P. Dymont, and Z. L. Erofeenko, Fiz.


tverd. Tela 27, 3703 (1985).
V. P. Dymont, G. I. Makovetski, and V. M. Ryzhkovski, phys. stat. sol. (a) 107, K89 (1988).
[37] T. Kanomata, H. Endo, S. Mori, H. Okajima, T. Hihara, K. Sumiyama, T. Kaneko,
and K. Suzuki, J. Magnetism magnetic Mater. 140/144, 133 (1995).
[38] G. A. Castellion, L. A Siegel, and H. Burkhard, J. Phys. Chem. Solids 30, 585 (1969).
[39] A. A. Dvorkin, I. A. Verin, V. S. Zakhvalinski, and A. N. Nateprov, Soviet Phys. ±±
Crystal lography 36, 785 (1991).
[40] I. Laue, J. Vanacken, A. N. Nateprov, F. Herlach, M. von Ortenberg, and
E. K. Arushanov, phys. stat. sol. (b) 185, 245 (1994).
[41] I. Laue, G. Machel, O. Portugall, M. von Ortenberg, A. N. Nateprov,
E. K. Arushanov, I. Mirebeau, J. Vanacken, F. Herlach, G. Kido, and N. Miura,
Proc. 22nd Internat. Conf. Phys. Semicond., Vancouver (Canada) 1994, Ed. D. J. Lock-
wood, World Scientific Publ. Co., Singapore 1995 (p. 2581).
M. von Ortenberg, I. Laue, A. N. Nateprov, J. Vanacken, and I. Mirebeau, Physi-
ca 201B, 57 (1994).
[42] L. Pytlik and A. ZieËba, J. Magnetism magnetic Mater. 51, 199 (1985).
[43] C. J. M. Denissen, Ph. D. Thesis, Eindhoven 1986, unpublished.
[44] W. Lubczyn ski, Ph. D. Thesis, Zabrze 1990, unpublished.
[45] A. Pietraszko and K. èukaszewicz, Acta cryst. B 25, 988 (1969).
[46] P. J. Lin-Cung, Phys. Rev. 188, 1272 (1969); phys stat. sol. (b) 47, 33 (1971).
[47] K. Sieran  ski, J. Szatkowski, and J. Misiewicz, Phys. Rev. B 50, 7331 (1994).
[48] B. E. Larson, K.C. Hass, H. Ehrenreich, and A. E. Carlsson, Phys. Rev. B 37, 4137
(1988).
[49] W. Lubczyn ski, J. Cisowski, and J. C. Portal, phys. stat. sol. (a) 120, 525 (1990).
[50] H. M. A. Schleijpen, Ph. D. Thesis, Eindhoven 1987, unpublished.
[51] F. A. P. Blom, in: Lecture Notes Phys. 133, 191 (1980).
F. A. P. Blom, J. W. Cremers, J. J. Neve, and M. J. Gelten, Solid State Commun. 33,
69 (1980).
[52] P. R. Wallace, phys. stat. sol. (b) 92, 49 (1979).
[53] F. A. P. Blom, J. J. Neve, and P. A. M. Nouvens, Physica 117/118B, 470 (1983).
[54] J. J. Neve, J. Kossut, C. M. van Es, and F. A. P. Blom, J. Phys. C 15, 4795 (1982).
[55] W. Mac, Nguyen The Koi, A. Twardowski, J. A. Gaj, and M. Demianiuk, Phys. Rev.
Letters 71, 2327 (1993).
[56] J. A. Gaj, R. Planel, and G. Fishman, Solid State Commun. 29, 435 (1979).
[57] R. J. Nicholas, M. J. Lawless, H. H. Chang, D.E. Ashendorf, and B. Lunn, Semi-
cond. Sci. Technol. 10, 791 (1995).
[58] M. Singh and P. R. Wallace, Solid State Commun. 45, 9 (1983).
[59] R. Laiho, K. G. Lisunov, V. N. Stamov, and V. S. Zakhvalinski, Proc. 7th Internat.
Conf. Narrow-Gap Semicond., Santa Fe 1995, Ed. J. L. Reno, Inst. Phys. Conf. Ser. No.144,
85 (1995); J. Phys. Chem. Solids 57, 1 (1996).
[60] J. Cisowski, W. Lubczyn ski, J. C. Thuillier, and J. C. Portal, Acta phys. Polon. A
75, 301 (1989).
R. Laiho, K. G. Lisunov, M. L. Shubnikov, V. N. Stamov, and V. S. Zakhvalinski,
phys. stat. sol. (b) 198; 135 (1996).
[61] W. Lubczyn ski, J. Cisowski, and J. C. Portal, Acta phys. Polon. A 79, 377 (1991).
W. Lubczyn nski, J. Cisowski, J. Kossut and J. C. Portal, Springer Series in Solid
State Sciences, Vol. 101, Springer-Verlag, Berlin 1992 (p. 463).
[62] L. Roth and P. Argyres, in: Semicond. and Semimet.1, 159 (1966).
[63] I. Rosenmann, J. Phys. Chem. Solids 30, 1385 (1969).
_
[64] W. Zdanowicz, W. Lubczyn _
ski, J. C. Portal, and E. Zdanowicz, Acta phys. Polon. A
67, 203 (1985).
[65] W. Lubczyn ski, W. _Zdanowicz, J. Cisowski, and J. C. Portal, Acta phys. Polon. A
71, 235 (1987).
[66] W. Lubczyn ski, J. Cisowski, and J. C. Portal, Acta phys. Polon. A 80, 275 (1991).
[67] R. J. Higgins and D. H. Lowndes, in: Electrons at the Fermi surface, Ed. M. Springford,
Cambridge University Press, Cambridge 1980 (p. 393).
Semimagnetic Semiconductors Based on II±V Compounds 349

[68] M. Vaziri and R. Reifenberger, Phys. Rev. B 32, 3921 (1985).


[69] W. Lubczyn  ski, J. Cisowski, P. Gee, J. Singleton, D. K. Maude, and J. C. Portal,
Mater. Sci. Forum 182/184, 715 (1995).
[70] V. S. Zakhvalinski, Ph. D. Thesis, Kishinev 1992, unpublished.
[71] K. G. Lisunov, A. V. Lashkul, R. Laiho, V. S. Zakhvalinski, A. Ma  kinen, and
E. Lahderanta,J. Phys. C 5, 5113 (1993).
[72] R. Laiho, A. V. Lashkul, K. G. Lisunov, V. N. Stamov, E. La  hderanta, and
V. S. Zakhvalinski, J. Magnetism magnetic Mater. 140/144, 2019 (1995).
[73] R. Laiho, A. V. Lashkul, E. La  hderanta, K. G. Lisunov, V. N. Stamov, and
V. S. Zakhvalinski, J. Phys.: Condensed Matter 7, 7629 (1995).
[74] R. Laiho, K. G. Lisunov, V. N. Stamov, and V. S. Zakhvalinski, J. Magnetism mag-
netic Mater. 140/144, 2021 (1995); Solid State Commun. 93, 151 (1995).
[75] H. Bednarski and J. Cisowski, Phys. Rev. B 48, 5113 (1993).
[76] H. Bednarski, Ph. D. Thesis, Zabrze 1993, unpublished.
[77] H. Bednarski, J. Cisowski, J. Voiron, D. Schmitt, J. C. Portal, W. Lubczyn  ski,
and A. Burian, J. Magnetism magnetic Mater. 139, 95 (1995).
[78] E. La hderanta, R. Laiho, A. V. Lashkul, V. S. Zakhvalinski, S. B. Roy, and
A. D. Caplin, J. Magnetism magnetic Mater. 104/107, 1605 (1992).
[79] A. V. Lashkul, E. La  hderanta, R. Laiho, and V. S. Zakhvalinski, Phys. Rev. B 46,
6251 (1992).
[80] S. M. Chudinov, V. A. Kulbachinski, I. V. Svistunov, G. Mancini, and I. Davoli,
Solid State Commun. 84, 531 (1992).
[81] E. La hderanta, R. Laiho, A. V. Lashkul, A. Ma  kinen, and V. S. Zakhvalinski, Semi-
cond. Sci. Technol. 8, S37 (1993).
[82] R. Laiho and K. G. Lisunov, J. Magnetism magnetic Mater. 140/144, 1771 (1995).
[83] H. Bednarski, J. Cisowski, J. Voiron, D. Schmitt, J. C. Portal, and J. Heimann,
Solid State Commun. 91, 461 (1994).
[84] Y. Shapira, S. Foner, D. H. Ridgley, K. Dwight, and A. Wold, Phys. Rev. B 30,
4021 (1984).
[85] Z. Celin  ski, T Mydlarz, and W. _Zdanowicz, Acta phys. Polon. A 71, 239 (1987).
[86] W. Lubczyn ski, J. Cisowski, H. Bednarski, J. Voiron, J. C. Portal, and J.C. Pi-
coche, Acta phys. Polon. A 77, 175 (1990).
[87] W. Lubczyn  ski, J. Voiron, J. C. Picoche, J. C. Portal, J. Cisowski, J. C. Thuil-
lier, and W. _Zdanowicz, Semicond. Sci. Technol. 4, 223 (1989).
[88] Z. Celin _
 ski and W. Zdanowicz, Acta phys. Polon. A 69, 1067 (1986).
[89] I. V. Potykevich, G. I. Bondar, V. S. Koval, P. D. Ivanchuk and V. M. Frasunyak,
Ukr. fiz. Zh. 28, 1072 (1983).
[90] J. Spaøek, A. Lewicki, Z. Tarnawski, J. K. Furdyna, R. R. Gaøazka, and Z. Obusz-
ko, Phys. Rev. B 33, 3407 (1986)
[91] K. Matho, J. low-Temp. Phys. 35, 165 (1979).
[92] G. Bastard and C. Lewiner, Phys. Rev. B 20, 4256 (1979);
C. Lewiner and G. Bastard, J. Phys. C 13, 2347 (1980).
[93] S. S. Yu and V. C. Lee, J. Phys.: Condensed Matter 4, 2961 (1992).
[94] N. Bloembergen and T. J. Rowland,Phys. Rev. 97, 1679 (1955).
[95] D. J. S. Beckett, S. F. Chebab, G. Lamarche, and J. C. Woolley, J. Magnetism mag-
netic Mater. 69, 311 (1987).
[96] H. Bednarski, J. C. Portal, J. Cisowski, H. J. G. Draaisma, T. Goto, and
W. Lubczyn ski, Physica 184B, 451 (1993).
[97] H. Bednarski, J. Cisowski, D. Schmitt, J. Voiron, J. C. Portal, and W. Lubczynski,
phys. stat. sol. (a) 141, 312 (1994).
[98] H. Bednarski, J. Cisowski, J. C. Portal, J. Voiron, and D. Schmitt, Mater. Sci. For-
um 182/184, 719 (1995).
[99] W. A. Harrison, Electronic Structure and the Properties of Solids, Freemann, San Francisco
1980 (p. 253).
[100] M Go  rska and J. R. Anderson, Phys. Rev. B 38, 9120 (1988).
[101] H. Bednarski, J. Cisowski, W. Lubczyn ski, J. Voiron, and J. C. Portal, Acta phys.
Polon. 87, 205 (1995).
350 J. Cisowski: Semimagnetic Semiconductors

[102] W. J. M. de Jonge, M. Otto, C. J. M. Denissen, F. A. P. Blom, C. van der Steen,


and K. Kopinga, J. Magnetism magnetic Mater. 31/34, 1373 (1983).
[103] C. J. M. Denissen, H. Nishihara, P. A. M. Nouwens, K. Kopinga, and W. J. M. de
Jonge, J. Magnetism magnetic Mater. 54/57, 1291 (1986).
[104] H. Bednarski and J. Cisowski, Acta phys. Polon. A 82, 868 (1992).
[105] H. Bednarski and J. Cisowski, phys. stat. sol. (b) 193, K19 (1996).
[106] H. Bednarski and J. Cisowski, J. Crystal Growth 159, 1018 (1996).
[107] A. Bruno and J. P. Lascaray, Phys. Rev. B 38, 9168 (1988).
[108] Y. Irie, T. Sato, and E. Ohta, Phys. Rev. B 51, 13084 (1995).
[109] M. Averous, C. Fau, S. Charar, M. El Koldi, V. D. Ribes, J. Deportes, and Z.
Goøacki, Solid State Commun. 84, 479 (1992).
[110] A. Nateprov, I. Tomak, J. Heimann, J. Cisowski, and I. Mirebeau, J. Magnetism mag-
netic Mater., in press.

You might also like