Smirnov 1993

Handbook on the Physics and Chemistry of Rare Earths, Vol.
16
edited by K.A. Gschneidner, Jr. and L. Eyring
© 1993, Elsevier Science Publishers B. E
Chapter 106
THERMAL CONDUCTIVITY OF RARE EARTH
COMPOUNDS
I.A. SMIRNOV and V.S. OSKOTSKI

Russian Academy of Sciences, A.F. Ioffe Physico-Technical Institute,
St. Petersburg, Russia
Contents
1. Introduction 110
2. Brief summary of thermal conductivity of solids 111
2.1. Thermal conductivity of a crystal lattice 112
2.2. Electron contribution to the thermal conductivity 114
3. Influence of the magnetic structure on the thermal conductivity of lanthanide compounds 120
3.1. Electron scattering by spin disorder in the paramagnetic region ( T > Try, Tc, T~) and its
influence on the thermal conductivity of lanthanide materials 120
3.2. Electron scattering by spin disorder in the paramagnetic region in Kondo systems
(electrical conductivity, thermal conductivity) 124
3.3. Magnetic phase transitions (temperature region 2: T,,~ TN, Tc, T~) 125
3.3. I. Thermal conductivity of magnetic lanthanide materials in the region T ~ TN, Tc 126
3.3.2. Behaviour of the thermal conductivity of lanthanide materials at a SOPT 127
3.4. Phonon-magnon scattering, heat transfer by magnons in lanthanide compounds
( T < TN, Tc) 129
4. Thermal conductivity of spin glasses 133
5. Thermal conductivity of systems with heavy fermions, intermediate valency of lanthanide ions,
dilute and concentrated Kondo lattices t40
5.1. Main principles, theoretical premises 140
5.2. Experimental results 143
5.2.1. Nonmagnetic heavy-fermion systems 144
5.2.2. Magnetic heavy-fermion systems 148
5.2.3. Lorentz-number anisotropy in heavy-fermion systems 150
5.2.4. Lorentz number in dilute Kondo systems 151
5.2.5. Thermal conductivity of systems with an intermediate valence 152
6. Influence of paramagnetic lanthanide ions on the thermal conductivity of ordered and
disordered systems 152
6.1. Phonon scattering by paramagnetic levels split by a lattice crystal field 153
6.1.1. Temperature dependence of -Atcre s 154
6.1.2. Methods for separating --Alcr~s 155
6.1.3. Arrangement of PLnls in a crystal lattice 156
6.1.4. Choice of materials for investigation 156
107
108 I.A. SMIRNOV and V.S. OSKOTSKI
6.1.5. Schemes of level splitting in a lattice crystal field for selected materials 157
6.1.6. Analysis of experimental thermal conductivity data 158
6.2. Influence of the crystal field effect on the electron part of the thermal conductivity 177
7. Thermal conductivity of lanthanide compounds with structural phase transitions initiated by
the cooperative Jahn-Teller effect (CJTE) 179
8. Thermal conductivity of rare earth superconductors 182
8.1. Main principles 182
8.2. Thermat conductivity ofheavy-fermion systems 184
8.3. Thermal conductivity of rare earth compounds in which superconductivity and
magnetism coexist 188
8.4. Some words on the thermal conductivity of high-temperature superconducting rare
earth compounds 193
9. Thermal conductivity of rare earth glasses and crystalline materials with glass-like behaviour
of ~cL(T) 195
9.1. Thermal conductivity of rare earth glasses 195
9.1.1. Thermal conductivity of insulating rare earth glasses 196
9.1.2. Thermal conductivity of metallic rare earth glasses 200
9.2. Thermal conductivity of crystalline materials with "glass-like" behaviour of xL(T) 201
9.2.1. "Amorphisation" of ~cL arising due to a complex crystallographic structure 203
9.2.2. "Amorphisation" of ~cL due to the strong phonon scattering by lattice defects 204
10. Thermal conductivity of rare earth compounds at high temperatures 206
10.1. Phonon spectrum peculiarities and ~L of rare earth compounds 206
10.2. Phonon electron scattering at T~> O 209
10.3. •o of rare earth compounds at high temperatures 211
10.4. Other effects 215
1I. Conclusion 216
References 216
Symbols and abbreviations

a lattice constant Her critical magnetic field for
(ac) crystal plane destroying superconductivity
a, c crystal axes H~I , Ho2 low and high critical magnetic field
Ad de Gennes factor H~ magnetic field of tc saturation
C heat capacity Hs0 high magnetic field for suppression
CJTE cooperative Jahn--Teller effect of carriers
Ce electron heat capacity ! exchange integral
d density J total angular moment
D thermal diffusivity k electron wave vector
DKL dilute Kondo lattice k0 Boltzmann constant
e electron charge l free path length of phonons
f Fermi function L Lorentz number
F Fermi integral L orbital moment
g Land6 factor LA longitudinal acoustic
g(ev) electron density of states at Fermi LO longitudinal optic
level L~f Lorentz number of the crystal field
h h/2~, h is the Planck constant effect
H magnetic held Leff effective Lorentz number
HFS heavy-fermion system LN nonparabolic Lorentz number
HTSC high-temperature Lo Sommerfeld value of the Lorentz
superconductivity number
THERMAL CONDUCTIVITY OF RECs 109
nL m generalised Fermi integral W0 electron thermal resistivity due to

k
scattering by impurities and defects
m electron mass
effective electron mass x concentration
m*
effective mass of heavy electrons XMeL
magnetic percolation limit of
m,~
concentration
m? effective mass of light electrons
M atomic mass
MPL magnetic percolation limit c~ thermal electromotive force
MPT magnetic phase transition ~N nonparabolic thermal
MTLS magnetic two-level system electromotive force
normal state of a metal fl = ko T~
He electron number from a level in a electronic specific heat coefficient
conduction band 6 linear expansion coefficient
N normal process of phonon A resonance energy of phonons
scattering AGes change of thermal conductivity due
N degeneracy of an f-level to resonance scattering of phonons
NM nonmagnetic eF Fermi energy
N. number of free electrons per atom % width of a forbidden band
q, q l , q2, q3 phonon wave vectors Sg effective width of a forbidden band
Q electron number for an f-level ,90 angle of magnetic moment rotation
r atomic radius O Debye temperature
R rare earth element Oj low Debye temperature
R electrical resistivity 02 high Debye temperature
REC rare earth compound K thermal conductivity
S spin moment ~% bipolar thermal conductivity
SOPT spin-orientation phase transition K~ electron thermal conductivity
T temperature G~c exciton thermal conductivity
TA transverse acoustic lattice thermal conductivity
/~L
TO transverse optic tc,~ magnon thermal conductivity
To superconduction critical /('MPL magnetic percolation limit of
temperature thermal conductivity
Tc Curie temperature Kphot photon thermal conductivity
T~ temperature of "freezing" /£tm total thermal conductivity
Tfcr ferroelectric phase transition 2 phonon wave length
temperature k~ = e v / k o T, residual Fermi level
TK Kondo temperature #B Bohr magneton
T. Jahn-Teller phase transition v electron band density of states
temperature p electrical resistivity
T~ N6el temperature Pl ideal electrical resistivity
Z spin-reorientation temperature Pmagn magnon electrical resistivity
T~, T:~ temperatures of the beginning and Pph electrical resistivity due to phonon
the end of the reorientation process scattering
T* high-temperature limit of coherent Dspd electrical resistivity due to spin-
interaction between heavy disorder scattering
fermions /)tot total electrical resistivity
U Umklapp-process of phonon P0 residual electrical resistivity
scattering a electrical conductivity
V mean phonon velocity relaxation time of phonons
W thermal resistivity Zb boundary relaxation time of
we electron thermal resistivity phonons
w~ ideal electron thermal resistivity % colloid relaxation time of phonons
due to scattering by phonons "gd point defect relaxation time of
WL lattice thermal resistivity phonons
rD dislocation relaxation time of % paramagnetic-ion relaxation time

phonons of phonons
q~e conduction-electron relaxation r~ molecular impurity relaxation time
time of phonons of phonons
• t, four-phonon relaxation time of ~ quasilocal vibrational relaxation
phonons time of phonons
~N normal-process relaxation time of % Umklapp-process relaxation time
phonons of phonons
co phonon frequency
1. Introduction
Rare earth compounds (RECs) include a wide set of new classes of materials with
unusual and sometimes unique properties. By now several thousand of RECs have
been synthesized: oxides (Arsen'ev et al. 1983), chalcogenides, hydrides, borides,
carbides, phosphides, pnictides, halogenides (Kost et al. 1983), zirconates, hafnates,
niobates, tantalates, antimonates (Arsen'ev et al. 1985), carbonates, oxalates, nitrates,
titanates (Komissarova et al. 1984), silicates, germanates, phosphates, arsenates,
vanadates (Bondar' et al. 1983), sulphates, selenates, tellurates, chromates and others.
Luckily for the authors, thermal conductivity has been measured by now in no more
than hundred RECs. The main feature of RECs is that they include the lanthanide
ions with the unfilled inner shells containing from zero (La-4f °) to 14 (Lu-4f 14)
electrons. The ions with the unfilled inner shells have noncompensated momenta,
therefore one expects magnetic ordering in compounds with such ions. However,
because of the specificity of filling and behaviour of the 4f shells (see table 4) not all
RECs are magnetics. So, one can separate the RECs in two groups: magnetics and
nonmagnetics. The 4f-shells do not overlap in RECs (the radius of the 4f shell is
0.3 ,~, which is ~ 0.1 of the interatomic distance) and they form localized atomic-
like states with concentration ~ 1022 cm 3 (Smirnov 1981). The 4f levels are involved
in some RECs:
(a) in a forbidden band (lanthanide semiconductors) in which they are present as
"impurity levels" (Smirnov 1981, Smirnov and Oskotski 1978). [These "impurity
levels" are unique, because in the usual semiconductors (Ge, Si, A3B5, A2B6) it is
impossible to create such a large concentration of impurity levels. In lanthanide
semiconductors the 4f levels determine the main kinetic effects, optics and unusual
phase transitions] or,
(b) in a conduction band, forming systems with Kondo impurities, concentrated
Kondo lattices, heavy fermions and intermediate valencies of lanthanide ions depend-
ing on the position of the 4f levels relative to the Fermi level (and on the lanthanide
ion concentration) (see section 5).
All of the above listed nonstandard classes of lanthanide materials show their
specific character in nonstandard behaviour of physical parameters. For example,
(1) coexistence is possible of magnetism and intermediate valency (Matlak and
Nolting 1983), magnetism and superconductivity (see section 8.3), magnetism and
ferroelectricity (¥enevtsev and Lubimov 1990):
THERMAL CONDUCTIVITY OF RECs i !1
(2) it is possible to induce by a magnetic field the K o n d o effect (Sandalov and

Podmarkov 1985) and superconductivity (Buzdin and Moschalkov 1986), and, by
powerful optical pumping, magnetic phase transitions in magnetically ordered lan-
thanide compounds (Golovenchits et al. 1981, Golovenchits and Sanina 1982);
(3) induced ferromagnetism (arising as a result of competition between the
exchange interaction effect and the crystal field effect) is observed (Guertin and Foner
1978, Bucher et al. 1972);
(4) in systems with heavy fermions destruction of the superconductivity by intro-
ducing nonmagnetic impurities occurs and, on the other hand, the superconduction
transition temperature (T~) increases with increasing magnetic ion concentration
(Buzdin and Moshchalkov 1986);
(5) in a number of systems there is a "re-entrant" superconductivity (with two and
even three superconduction transition temperatures);
(6) all systems have unusual dependences (on temperature, magnetic and electric
fields, and pressure) of the various parameters of transport, optical, magnetic and
other effects.
From the history of thermal conductivity 0c) investigations in solids one can see
that tc has always "responded" on the appearance of new effects. We will try to review
the unusual character of t¢ in RECs. Unfortunately, tc of many interesting RECs has
not yet been measured and many mysteries have no answer at this moment. This
review is the first major review devoted to ~c of RECs. There is only one prior review
devoted to ~c and other thermal properties of rare earth chalcogenides (Smirnov
1972), in which data are collected up to 1972.
This review is not a reference book on ~c of RECs. Here we are trying to show on
typical examples how the unusual properties of RECs reveal themselves in the
behaviour of ~c. Naturally, because of the limited size of this review, we could not
analyse all literature data on tc of RECs. Before we analyse the specific behaviour of
tc in RECs, we will briefly discuss in the next section the usual (classical) regularities
in the behaviour of lc of standard solids. This should help readers to understand
what new features are added to the whole picture of the behaviour of the thermal
conductivity by the results on ~c of RECs. This review is mainly devoted to discussing
the experimental work.
2. Brief summary of thermal conductivity of solids
The total thermal conductivity lCtot of insulators, metals and semiconductors

consists of the following contributions:
insulators: /('tot = lgL~

metals: ~Ctot = ~cL + x e ,
semiconductors: /('tot = KL -[- ICe -~- /£b -~- K'ph -~- /£exc,
where bcL is the lattice, G is the electron, ~cu the bipolar, bCphthe photon and Gxc the
exciton contribution to the thermal conductivity. A contribution ~cm, the magnon
thermal conductivity, is possible in magnetic insulators, metals and semiconductors

at low temperatures due to heat transport by spin waves. At present the main
theoretical propositions of the thermal conductivity K of solids are obtained, compre-
hended and described in Oskotski and Smirnov (1972), Berman (1976), Smirnov and
Tamarchenko (1977) and other monographs, and many reviews.
2.1. Thermal conductivity o f a crystal lattice
A thermal flow is carried by phonons and its value and temperature behaviour
depend on the contribution of different scattering mechanisms. IcL(T) of semiconduc-
tors and insulators is similar, while ~cL of metals is significantly smaller in the low-
temperature region because of strong phonon-electron scattering (fig. 1). At low
temperatures ( T ~ O, O is the Debye temperature) phonons are scattered, mainly,
by crystal boundaries and ~L oCT 3. At T > O p h o n o n - p h o n o n scattering results in
a temperature dependence tCLOCT 1. At T < O exists a temperature region with
~cL oc exp(O/aT). In metals at T ~ O due to phonon-electron scattering t% oc T 2. As
it has been noted above, different temperature dependences could occur in
other temperature regions of ~cL(T) due to contributions from different scattering
mechanisms.
For three-phonon interactions one distinguishes two types of collisions: normal
processes (N processes), in which the total momentum is conserved and the direction
of flow does not change (these processes lead to infinite thermal conductivity); and
Umklapp processes (U processes), in which the sum of the wave vectors is not
conserved and changes sharply, leading to a finite thermal resistivity of a crystal. In
U processes the following conditions are fulfilled:
col + co2 = c03, (1)
ql + q2 = q3 = 2~zb, (2)
where co is an angular frequency, ql, q2 and q3 are phonon wave vectors, and b is a
reciprocal lattice vector. The theory of lattice thermal conductivity meets with
difficulties with N processes, especially at low temperatures, when long-wavelength
phonons prevail and N processes are the main type of p h o n o n - p h o n o n scattering.
Although N processes themselves do not produce a thermal resistance, they influence
~L indirectly by redistributing phonons of different modes. Three-phonon N and U
processes are mainly anharmonic processes. However, processes of higher orders are
XL
2 .),,I..X.j- exP O/aT

I Z,
1,3//-._~ Fig. 1. Schematicshape of the tempcraturedependenceo f l ¢ L for metals

T (1) and insulators (semiconductors)(2).
T H E R M A L CONDUCTIVITY OF RECs 113
possible when simultaneously four of more phonons interact (Oskotski and Smirnov
1972, Berman 1976, Smirnov and Tamarchenko 1977). To satisfy condition (2),
interacting phonons are to have wave vectors near rib. At T > O there are many
such phonons, and U processes occur often, their number increases proportional
to T, and lCLOCT *. At T ~ O the number of phonons with wave vectors near
nb decreases exponentially giving rise to a fast increase of ~CL by the law tcLoc
e x p ( O / a T ) (a is a coefficient of order two). At low and medium temperatures the
acoustic phonons of the spectrum participate in the heat transport and scattering
processes. At high temperature optical phonons are excited. Then t% can increase
due to additional heat transport by optical phonons, or decrease due to additional
scattering of acoustic phonons by optical ones (Blackman 1935, Gurevich 1959,
Leroux-Huqou and Yeyssie 1965, Devyatkova and Smirnov 1962, Slack 1979).
Usually theoretical analysis of tel data is carried out in the Debye model in the
relaxation-time approximation, where 1eL has the form
tel = GT3 [o/r 1 siXnh2(½x)dx. (3)

20 ~-
Here x = h~O/ko T, G is a constant, T- ~ is the sum of the reciprocal relaxation times

of phonons in different scattering processes: r - 1 = E~i- 1, where ~i = %, *a, re, VD, %,
Z~., Z[, rp, %, ZN, ZF, the relaxation times of phonon scattering, correspondingly, on
crystal boundaries, point defects and impurities, conduction electrons, dislocations,
colloids, molecular impurities, quasi-local vibrations, paramagnetic ions, U, N and
four-phonon processes. Each z- ~ has its own frequency and temperature dependence
(Oskotski and Smirnov 1972, Berman 1976, Smirnov and Tamarchenko 1977). In
fig. 2 the temperature regions are marked in which the various scattering mechanisms
are significant. Callaway (1959) has shown how it is possible to account in eq. (3) for
the relaxation time of N processes. Holland (1963) has proposed a mechanism which,
for ~ 1 and ~R 1, takes into account some details of phonon spectra and the separated
heat flows transferred by transverse and longitudinal acoustic phonons. To conform
XL
X rnQx
T d , T u , T N ,7"R', ' g e , T p
/ \
Tb ~ , ~ u ,'t'N,
TF
F'ig. 2. Schematic shape of the temperature dependence
of ~cL for solids. The temperature regions are marked,
I ~_ where various phonon scattering processes play an
o T appreciable role. O is the Debye temperature.
114 I.A. SMIRNOV and V.S. OSKOTSK1
experimental and theoretical values of ~CL,according to eq. (3) one has to introduce
a great number of fitting constants (for example, A1-Edani and Dubey (1986) have
introduced twenty constants for ~cL of GdS). The more mechanisms of phonon
scattering are introduced, the greater is the number of fitting constants. Sometimes
some of the constants can be evaluated by independent theoretical calculations or
measured experimentally. In this review we will try to use this method of analysis as
seldom as possible and limit ourselves only to the physical nature of the behaviour
of ICL(T). Thus, analysing experimental data on teL(T) of RECs we consider the results
"standard" if ~CL(T) has a value and a temperature dependence predicted by the
theory for ordinary solids (Oskotski and Smirnov 1972, Berman 1976, Smirnov and
Tamarchenko 1977), and "nonstandard" if there are deviations from the theory.
2.2. Electron contribution to the thermal conductivity
The electron thermal conductivity of metals and semiconductors is determined by

the Wiedemann-Franz law
Ke = LT/p, (4)
where L is the Lorentz number and p is the electrical resistivity. In the low-
temperature region ( T < O/10) the electron thermal resistivity, We= 1/G , can be
presented in the form (Smirnov and Tamarchenko 1977)
G = Wo + (5)
where Wo is the thermal resistivity caused by electron scattering by impurities and

defects (Wo = po/Lo Z with Po the residual electrical resistivity, L o the Sommerfeld
value of the Lorentz number and Lo = ½(kc/e)Z~r 2 = 2.45 x 10 -a W •/K 2) and Wi is
the ideal thermal resistivity caused by electron scattering of phonons (Oskotski and
Smirnov 1972, Berman 1976, Smirnov and Tamarchenko 1977)
(T~[ 0 D 0 [2~2 01 0)1

w,=B -- T - 5 17 . (6)
Here B and D are constants, eV is the Fermi energy, and 15 and I7 are integrals of
the type
fo E z" dz
I,(x) = (ez - 1)(1 - e z)" (7)
These integrals are tabulated in Smirnov and Tamarchenko (1977). Equation (5) is
obtained by analogy with the case of the electrical resistivity Ptot = Po + Pi (where Pi
is the ideal electrical resistivity caused by electron scattering by thermal vibrations).
At low temperatures WE oCT 2, so eq. (5) can be presented in the form
W~ = po/Lo T + ocT2 = fi/T + c~T2, (8)

We T = ~T 3 + ft. (9)
The temperature dependence, We = f ( T 3 ) , is a straight line with a slope of c< The

intercept on the We-axis is/L The parameter c~ must be a constant for a certain class
of materials and independent of sample purity (Smirnov and Tamarchenko 1977).
The constant /~, on the other hand, changes from sample to sample and depends
indirectly (through Pc) on its purity. Figure 3 shows schematically the temperature
dependence of the different contributions to eq. (8). Since W~ increases and Wo
decreases with T, We has a minimum at a certain temperature. The Lorentz number
of metals of any purity for T >> O, and in samples with impurities and defects for
T ~ O, is a constant and equals Lo. At T = 0 for pure metals L = 0 (see fig. 4 and
eq. (10)). In the intermediate temperature range, L is determined by the temperature
and the purity of the sample. At temperatures both lower as well as higher than O
the ratio L/Lo in metals could be presented in the form (Oskotski and Smirnov 1972)
L[po (T) 5 (O)l/[po (T, D~], (10)

Lo = 7~0 + I5 ~ 4AEo+X\o ~FJ]
where the function
= 1+ 15 -
with 15 and 17 integrals of the type (7), A o is a constant which depends on an

interaction constant between electrons and the lattice, the lattice constant, @, atomic
mass and O. D is also a constant. According to Olsen and Rosenberg (1953)
D/@ = 2-1/3N. 2/3 (11)
We
kk...d'//
y....jWo
r Fig. 3. The temperature dependenceof different contributions to eq. (8).
1.0 ~l I I I I l~----I'~'--'~
"--
o 0.6
-4
.II?'-'-,d" I
d // I
0.2
_b,? ',
., "~1 I Fig. 4. The dependenceof L/Loon T/O(Oskotski and
z ~ I I I I I I
0.4 0.8, 1.2 1.6
Smirnov 1972) calculated using eq.(10) for N, = 1.
Curve 1: po/4A=O,curve 2: 0.0038, curve 3: 0.0191,
T/0 curve 4: 0.0957.
116 1.A. S M I R N O V and V.S. O S K O TS K I
where N a is the number of free electrons on an atom. For T/O > 0.6 (Olsen and
Rosenberg 1953)
Pi = Ptot PO -- Ao r i o . (12)
The parameter po/4Ao in eq. (10) is a measure of the purity of a material. For very
pure metals po/4Ao = 0. Figure 4 shows, e.g., dependences L/L o = f ( T / O ) calculated
using eq. (10) for a number of values of po/4Ao. A decrease of L compared to L o is
connected with the inelastic character of electron scattering by phonons in this
temperature region. At very low temperatures in metals with impurities and for
T > O electrons are scattered elastically (Smirnov and Tamarchenko 1977). In some
metals (e.g., monovalent and noble ones) for T > 0 one finds that L < Lo. One
connects this effect with an additional inelastic scattering mechanism electron
electron scattering (Smirnov and Tamarchenko 1977). In the thermal conductivity
theory any deviation of L from Lo indicates the presence of some new elastic or
inelastic scattering mechanism for the current carriers.
A more complicated behaviour of the Lorentz number is observed in semiconduc-
tors. L/L o behaves differently for elastic and inelastic scattering of electrons and in
the presence of complicated energy band and interband scattering.
The Lorentz number of a semiconductor with a parabolic band, due to elastic
scattering of electrons can be written as (Oskotski and Smirnov 1972, Smirnov and
Tamarchenko 1977)
= IFr+ - (r + )2FL
where F represents the Fermi integral Fro(#* ) = ~x'(eU*+ 1) 1 dx (the integrals are
tabulated in Smirnov and Tamarchenko (1977)). /x* is a reduced Fermi level, # * =
ev/ko T [#* is determined from experimental data on the thermal electromotive force,
c~, by the equation
c~= \(r + ~-)F~+ 1/2(#*) - #* (13a)
(Oskotski and Smirnov 1972)], r is a scattering parameter (r is an exponent in the

dependence of electron relaxation time on energy, e.g., r = 0.5 and - 0 . 5 , respectively,
for scattering by optical and acoustic phonons). In nondegenerated samples (#* < 0,
1#1 ~> 1) the Lorentz number is equal to
L = (r + 2 . 5 ) ( k o / e ) 2 (14)
and depends only on the scattering parameter. In the case of strong degeneration
(/~*>> 1) L = L o . Figure 5 (curve 1) shows the dependence L/(ko/e) 2 = f ( # * ) for
r = -0.5.
In the case of a nonparabolic band described by the Kane model (Kane 1957) for
the elastic scattering mechanism of electrons by acoustic vibrations, the Lorentz
f i [ I"
z
3.0
2o b.
Fig. 5. The dependences L/(ko/e)2 - f(ll*)
calculated by eq.(13) (curve 1) and eq.(15)
(curves 2 and 3) for r= -0.5. Curve 1: /3=0,
1.0 _ . L . _ _ _ _ L I A__ I ---L parabolic band (a); curve 2:/3 = 0.05, nonpara-
-4 0 4 8 12 16 bolic band (b); curve 3:/3 = 0.2, nonparabolic
/z*
band (b).
number has a more complicated form (Oskotski and Smirnov 1972)
2LI °LI-- (1L2)2 (15)

LN = OL1 ,
-2
where "Lm (z, #*) are generalised Fermi integrals:

k
These integrals are tabulated in Smirnov and Tamarchenko (1977). Here

= ko Te*, z = e/ko T, f is the Fermi function and e* is the effective width of the
forbidden band of interaction, which is near to the real forbidden band width, eg, for
a number of narrow-band semiconductors. #* is also determined from c~by equations
taking account of the band nonparabolicity. For r = - 0 . 5
1L1 _ # * OL1
2 --2
c~N= (ko/e) 0L1 (15a)
--2
Figure 5 (curves 2 and 3) shows the dependences L/(k/e) 2 =f(ff*) calculated using
eq. (15) for a number of parameters/3. A nonparabolic band appreciably changes the
L value except for the case of strong degeneration.
All inelastic scattering mechanisms of electrons (electron electron scattering, scat-
tering by optical lattice vibrations, intervalley transitions and so on) lead to a
decreasing Lorentz number compared with the case of elastic scattering. For example,
in degenerated semiconductors (and metals) we always have L < L 0 (Oskotski and
Smirnov 1972, Smirnov and Tamarchenko 1977).
In materials with a complex electron band structure (subbands of "heavy" and
"light" carriers) strong interband scattering (e.g., s d ) is possible at a certain position
of the chemical potential relative to the bottom of the "heavy" subband. This
scattering strongly influences the value of L (fig. 6) (Kolomoets 1966). The value of
118 I.A. S M I R N O V a n d V.S. O S K O T S K I
40
3.5
30
2.5
-a -4 0 4 B 12
1 + I +
o I b I
Fig. 6. The dependence of the Lorentz number on 7
for the case of high degeneration of current carriers at
S = 32 (1), 8 (2). (a), (b), and (c) depict different positions
of the Fermi level, eF, relative to the bottom of the
EF band of heavy carriers (eo), 7 = (eF -- eo)/koT.
L depends on the parameter S:

IMlhl2m~ 3/2
S = iMl12m,3/2, (17)
where Mlh and M1 are matrix elements of transitions between states of corresponding
subbands and m* and m* are the effective masses of heavy and light carriers.
Let us consider briefly other contributions to the thermal conductivity. Bipolar
thermal conductivity, tCblp, appears in semiconductors in the region of the intrinsic
conductivity due to diffusion of electron hole pairs fi'om the hot to the cold end of
a sample.
The p h o t o n contribution, Kphot, to the thermal conductivity, Ktot, can be significant
at m e d i u m and high temperatures in materials which are semi-transparent and
transparent in the infrared region. In insulators 1%hotoc T 3, and in semiconductors
tCphot(T ) shows a curve with a maximum. At low temperatures K'phot = 0. At high
temperatures, due to the intrinsic conductivity, the concentration of carriers increases
exponentially, and the absorption coefficient increases in semiconductors so that
1Cphot decreases to zero. The m a g n o n component, ~m, of the thermal conductivity will
be considered in detail in section 3.4. Exciton thermal conductivity (~cexo)- diffusion
of excitons from the hot to the cold end with a consecutive recombination - could
be expected at high temperatures. This has been predicted by Pikus (1956).
The following information a b o u t defects in solids can be obtained from ~:(T)
measurements:
(1) Type of defects: vacancies, bivacancies, complexes, clusters, impurity atoms,
dislocations, isotopes and so on.
(2) A value for the impurity and defect concentrations down to small values
(~1012_1013 c m - 3).
(3) Kinetics of ordering processes of defects in solids.
(4) Kinetics of the formation and decay of bivacancies, complexes and clusters.
F r o m analysis of KL(T) one can obtain information about the contributions of
different phonon groups (optical or acoustic) to the heat transport and about the
nature of phonon interactions with free carriers. The electron component of the
thermal conductivity contains information on the scattering mechanism and its
character (elastic or inelastic), and allows one to determine the kind of energy band
structure (parabolic, nonparabolic), the presence of additional subbands with light
and heavy carriers, and the nature of the electron interaction with phonons and
other electrons. F r o m data on the bipolar thermal conductivity one can determine
the forbidden-band width at medium and high temperatures. Useful information on
material properties can be obtained from tgphot , /£exc and ~cm.
In recent years a new class of materials rare earth compounds (RECs) has
been investigated intensively. These materials possess nonstandard physical proper-
ties, including the thermal conductivity. New features were observed in ~cL, ~co and
lCm. Analysing data on the thermal conductivity of RECs, in this review we will try
to attract more attention to nonstandard effects characteristic of the solids containing
ions with f-electrons.
We will consider in more detail results for ~CL, Kc, Km. Table 1 shows materials,
effects and parameters, which can be analysed using experimental data on ~cL,~ce and
~m' This review is devoted, mainly, to the discussion of situations enumerated in
table 1.
TABLE 1
Effects influencing the thermal conductivity of rare earth compounds.
Ktot = ~L + Ke + Km
Superconductingsystemsd Influence of magnetic ~__ Systemswith Kondo -- I Magnonheat
with heavyfermions [I phasetransitions impurities transport
Phononscatteringon ~_ I ConcentratedKondo I Phononscattering

superconduetin9systems[ I spin disorder lattices;systemswith by magnons
heavy fermions
Systemswith coexistencel I Phonon scattering }_ I Scatteringof current
of magnetism and on paramagnetic I Systemswith __ carriersby magnons
superconductivity I rare earth ons intermediate valency
of rare earth ions
Influence of
Jahn-Te er effect
L
| I Electron scattering
on spin disorder
I Spin g'asses
Scatteringof current
carriers on ground
I R. . . . . rth glasses }'-- state levelssplit
by crystalfie d
Systems with homo-
geneousand nonhomo-
geneous intermediate
va ency of rare earth ions
120 1.A. SMIRNOV and V.S. OSKOTSKI
3. Influence of the magnetic structure on the thermal conductivity of lanthanide

compounds
One can divide the influence of magnetic properties on thermal conductivity into
three temperature regions (fig. 7):
Region 1 (T > TN, To T~). Phonons and electrons are scattered by disordered spins
(spin-disorder scattering).
Region 2 (T_~ TN, To T~). Thermal conductivity can change sharply.
- Region 3 (T < TN, To ~). There is electron and phonon scattering by magnons
and heat transfer by magnons.
Let us consider the behaviour of ~c of lanthanide compounds in these regions.
3.1. Electron scattering by spin disorder in the paramagnetic region (T > TN, To Ts)
and its influence on the thermal conductivity of lanthanide materials
In great detail the influence of electron scattering by spin disorder has been
investigated by Gratz and Zuckerman (1982a, b), Gratz and Novotny (1983), Gratz
(1982), Bauer et al. (1985a, b, 1986, 1987) and Gratz et al. (1985). Experimental data
on the thermal conductivity of RA12 (R = Y, La, Ce, Pr, Nd, Sm, Gd, Tb, Yb, Dy,
Ho, Er, Tm, Lu), LaCu6, CexLal xCu6 have been analysed. Let us consider data
for RA12. The compounds RA12 crystallise in a cubic lattice of MgCu2-type. RA12
with R = Y, La, Yb, Lu are nonmagnetics. CeA1 a is antiferromagnetic (TN = 3.8 K)
I I I I
I< - = - >1
]
i k TN ,Y C Ts
' 1>
.• Electran
scattering
on magnons
x~ Phonon
scattering
on spin disorder
Etectron
t scattering
Phonon r~ scattering
on m a g n o n s onspin disorder
_•aMagnon heat
nsport [ ~
Fig. 7. The influence of certain scattering or transport mechanisms on the contributions to thermal
conductivity(the symbolis givenin the lower right corner of each rectangle)in three temperature regions.
7~, 7c and T~are the N6el, Curie and spin reorientation temperatures, respectively.
(Barbara et al. 1977). Other compounds are ferromagnetics with Tc in the range 6
168 K. The R atoms (excluding Ce, Eu and Yb) have valency + 3. RA12 are metals.
The measured tc is equal to
~tot = ~cL + 1%. (18)
At temperatures T ~ O thermal phonons in metals are mainly scattered by conduc-

tion electrons. According to a theory of Klemens (1969), the predominant phonon
electron scattering is given by
1 % = b T 2. (19)
At these temperatures ~c¢ is determined by conduction electron scattering on static
defects (Klemens 1969) and
~Ce= aT. (20)
So, at T ~ O
~Ctot = a T + bT 2 or Iqot/r= a + bT. (21)
The coefficients a and b can be evaluated from the dependence/qot/T = f(T). Figure 8
shows such a dependence for YA12 (Bauer et al. 1986). The resulting b shows that
the contribution of ~cL to fqot is about 5% at 20 K. For higher temperatures a
numerical calculation has been done using approximate theoretical equations
(Klemens 1969). It turns out that, at 300 K, ~cL contributes only about 2% to ~qot.
So, practically the experimental value of/~tot for RA12 is equal to ~ce.
The influence of the electron-spin disorder scattering on i% in the paramagnetic
region (i.e. the magnetic contribution to the thermal conductivity) is estimated by
comparing ~qot-~ tre of the magnetic and nonmagnetic compounds of the RA12 family.
YA12 is chosen as a nonmagnetic reference material. In the paramagnetic region
(Dekker 1965) p(T) can be represented as
, o ( g ) =- jo o -}- P p h Jr- Dmagn, (22)
where the terms are the residual, the lattice and the magnon electrical resistivity,
respectively.
Pmag,1= Pspd OCconst.(g -- 1 ) 2 j ( j + 1) = const. Aa, (23)
where Pspa is the electrical resistivity due to spin-disorder scattering (Dekker 1965),
g the Land~ factor, J the total angular momentum, and A a the de Gennes factor.
a~
o
O 5 10 15 20 25 Fig. 8 ~:,oUTasa function of Tin YAl~(Baueret al.
~£ T,K D86).
The electronic thermal resistivity We = 1/tee in the paramagnetic region for RAI2
can be written as:
(a) for nonmagnetic compounds (NM) (YAI2, LuAI2)
We = W~,d
NM+ W¢,ph,
NM (24)
(b) for magnetic compounds (M) (GdA12, etc.)

M
= + wo,ph + Wo,m.+n. (25)
The indices (e,d), (e,ph), (e,magn) indicate free-carrier scattering by defects, phonons
and magnons, respectively.
The difference (W M - W~ M) corresponds to the contribution to the electronic
thermal resistivity from electron scattering by the disordered spins
A W ~-- M NM
We---We = ( W ~U, d - We,d
NM) + (We,ph
M -- We,ph
NM ) -I- W~m . g n . (26)
Let us suppose that We,ph
M and Wc,ph
NM are equal in all related RA12 compounds, then
AW = (w+Ma-- We,d
NM) + We,magn. (27)
We,a in magnetic and nonmagnetic compounds is proportional to T-a [see eq. (20)]:
W~Md= BM/T, WeN,M = B•M/T, where B = 1/a [see eq. (20)]. According to Bauer et al.
(1986)
We,mag n = Wspd ~ const.(g -- l)2J(J + 1)/T= const. Aa/T. (28)
So, finally we have
A W = (B M -- BNM)/T+c o n s t . Aa/T. (29)
Experimentally AW is determined as the difference 1/toM - 1/~cTM, where tcNM is the

thermal conductivity of YA12 (see, e.g., fig. 9). Figures 10 and 11 show the dependence
of P+pd, eq. (23), and AW'T, eq. (30), on the de Gennes factor
A W ' T = A W T - (BM- B NM)= Wsp d r. (30)
As one can see in these figures, the experimental dependences Pspa(Aa) and W+pa(Aa)
W 103, cm K/rnW
I l ~ ~ t i I I -I I I
40
3O
~ GdAI2 t
YAI 2 Fig. 9. VV~= l#c mMand We = 1/KM as functions of

0 I I l z I I ~ I t I I I T, respectively, for YA12 and GdA12 (Bauer et al.
0 1O0 200 T, K 1986).
(g-1)2 .] ( 3 + 1 )
0 5 10 15
80
E 60
u
&
& 40
20
Fig. 10. The dependence of P~p,1 on tile de Gennes l~.ctor for

Er HO Dy Tb Gd compounds RA12 (Bauer et al. 1986).
(9-'i)2"J(J +1)
0 5 10 15
0 I.,'~" I I I
'~ I I I I [ I Fig. 11. The dependence of AW'eT on the de Gennes
Tm Er Ho Dy Tb Gd factor for compounds RAI2 (Bauer et al. 1986).
14 I I I I I I 1 I I I I I I
10
u
&
2 I I I t I I I I I I I I I Fig. 12. The dependence of W of compounds RA12 on

LQ Pr Pm Eu Tb HO Tm Lu the atomic number of the lanthanides at T = 270 K
Ce Nd Sm Gd Dy Er Yb (Bauer et al. 1985a, b, 1986).
agree well with the theory for RA12 for the heavy lanthanide elements [cf. eq. (23)
(Dekker 1965) and eq. (28) (Bauer et al. 1986)].
Experimental data for W, as shown in fig. 12, show a large influence of electron
spin disorder scattering on t~, of RA12. The dotted line is considered as a reference,
corresponding to W of a hypothetical nonmagnetic material. Additional scattering
of electrons by disordered spins causes an increase in W compared to the reference
value. GaAI2 has the largest W. W decreases from GdA12 in both directions to
nonmagnetic LaA12 and LuA12. "A jump" of W is observed for CeA12. CeA12, unlike
other RAlz, is a Kondo material with peculiar physical properties. The experimental
data shown in fig. 13 confirm the theoretical dependence, eq. (29).
Table 2 shows data for ,Ospd (obtained from electrical conductivity data) (Gratz and
T,K
250 150 100
I .I I _
10
GdAI/ / SmAI2
E
I// 7
u
eo
t I i I I i I I I I I I I d_.
5 10
103/1, K -1
Fig. 13. The dependence AW(I/T) for compounds RA12
(Bauer et al. 1985a,b, 1986).
TABLE2
Electrical resistivity due to spin-disorder scattering of electrons.
RA1z GdAl/ TbA12 DyA12 HoA12 ErA12 TmA12
P*pd(gf~cm) 43.7 32.8 16.6 13.2 5.5 1.7

Pspa (I"t~ cm) 59.0 35.0 23.5 13.5 6.5 1.5
p*pa:calculated from the Wiedemann Franz law.
Pspd: experimental results.
Zuckerman 1982a,b, Gratz 1982, Gratz and Novotny 1983, Gratz et al. 1985, Bauer
et al. 1985a,b, 1986) and P*pa (calculated using the W i e d e m a n n - F r a n z law from
experimental AW')./)spa -~ Lo A W ' T (Bauer et al. 1986). As one can see in table 2, the
agreement between Pspa and P*pais good.
3.2. Electron scattering by spin disorder in the pararnagnetic region in Kondo systems
(electrical conductivity, thermal conductivity)
The theoretical dependence p(T) has been obtained by Coornut and Coqblin (1972)
for compounds with concentrated K o n d o lattices and carrier scattering by disordered
spins (Pspd), taking into account the crystal field effect. According to this calculation
the temperature dependence pspa(ln T) could be divided into two regions separated
by a m a x i m u m at a temperature approximately equal to the value of the splitting
by the crystal field.
CeAI2 is a prototype of materials with concentrated K o n d o lattices. For TN > 3.8 K
(the paramagnetic region) strong scattering of carriers by disordered spins is observed
(Bauer et al. 1986). Figure 14 shows the dependence p~pd=f(ln T) [p~pd =

p(CeA12) - p(YA12)], which agrees well with the conclusions of Coornut and Coqblin
(1972). It turns out that the dependence AW T = f ( l n T) lAW =
W(CeA12)-W(YA12)] for CeA12 has the same shape as Psp,~ (Bauer et al. 1986)
(fig. 14). Similar dependences AW(T) have also been obtained in other systems with
concentrated K o n d o lattices: CeCu2 (Gratz et al. 1985) and CeCu 6 (Bauer et al.
1987). All this allows us to conclude that the observed behaviour AW T = f ( l n T) is
typical for compounds with Kondo lattices and electron scattering by disordered
spins,
So, the influence of electron scattering by spin disorder on tee of RECs has been
considered in sections 3.1 and 3.2. We have failed to find data on the influence in
the temperature region T > TN, Tc of phonon scattering by spin disorder on teL.
3.3. Magnetic phase transitions (temperature region 2." T ~ TN, Tc, T~)
Magnetic phase transitions (MPTs) can be divided in two groups (Belov et al.
1976, 1979, Belov 1972):
group 1: MPTs of the order disorder type. These transitions are connected with
the destruction of ferro- and antiferromagnetic ordering (observed transitions:
ferromagnetic-paramagnetic and antiferromagnetic-paramagnetic at temper=
atures Tc and TN, respectively).
group 2: MPTs of the order order type. These transitions are connected with a
magnetic structure type change (e.g., ferromagnetic-antiferromagnetic, collinear
antiferromagnetic-noncollinear antiferromagnetic and so on). Such transitions
resemble structural transitions in crystals and sometimes they are called magneto-
structural phase transitions.
The spin-orientation transitions can be distinguished by their unique features from
the large family of magneto-structural phase transitions. Unfortunately, the thermal
conductivity of RECs has not been investigated for all groups of magnetic phase
transitions. Reliable data are available for the M P T of the order disorder type and
for the spin-orientation phase transition (SOPT).
E
u
::L
7i
66
58
| ~ 2 i i -l--xJ i i --
-,
\\--
12 "7
8 %
%
oE
®
Fig. 14. Dependence of P~pa (1) and AW~T (2)

50~
-- I 1 I l" ~ I I 1 I
xa
on In T for CeAIz (Bauer et al. 1986). Pspa-
5 10 20 50 100 200300 p(Ce AI2)- p(YAla). AWe= We(CeAI2)
T, K - W2(YAI2).
I26 I.A. S M I R N O V a n d V.S. O S K O T S K I
3.3.1. Thermal conductivity of magnetic lanthanide materials in the region T,,~ TN, Tc
in this region there are no specific effects in the thermal conductivity of lanthanide
magnetic materials. Their ~c behaviours are analogous to the K of other magnetic
materials. Very often the phase transition from the paramagnetic to the ferromagnetic
or antiferromagnetic state does not manifest itself in the temperature dependence of
tcL and Ke. However, there are cases where at TN and Tc appreciable variations are
observed in the temperature dependence and the value of KL and Kc. There are
qualitative considerations (which, however, are not always confirmed experimentally),
why a variation of IcE and ~ce is observed at TN and Tc. It is important to know which
scattering and transfer mechanisms prevail near TN and Tc and which variations of
these mechanisms are connected with the reconstruction of a magnetic system. For
example, if for T > TN (Tc) there is a very strong electron scattering by disordered
spins in a material with/£tot ' ~ /£e, then an increase of the value/(tot is possible at the
transition temperature TN (Tc).
Let us consider some examples of 1% and ~cc behaviour of RECs at Tc and TN:
(1) EuO (insulator, 1(.tot = I(.L)" No variation of ~cL is observed at Tc (fig. 15) (Salamon
et al. 1974, Martin and Dixon 1972). It is interesting to note that in EuO a strong
decrease of the thermal diffusivity, D, at Tc is observed (Salamon et al. 1974). This
decrease is fully compensated for in 1(.cby an increase of the heat capacity, C, in this
temperature range 0eL = DCd, d is the density of the material). For T ~< Tc in EuO
there is no additional heat transfer by magnons (because of strong m a g n o n - m a g n o n
interaction), p h o n o n - m a g n o n scattering is low and KL is determined mainly by
p h o n o n - p h o n o n scattering.
(2) PrA12, SmA12 [metals, ~Ctot = ~c~ (Bauer et al. 1987, (Mfiller et al. 1983)] (see
fig. 16). ~Ctot changes at Tc in PrA12 and does not change in SmA12. Possibly, this is
connected with a different influence of electron scattering by the spin system on the
thermal conductivity (for T > Tc electrons are scattered by disordered spins, for
T < Tc by the disordered spin system). This situation has been discussed above.
(3) DyS [ p o o r metal, /(:tot = /£L Jr- lee (Novikov et al. 1975)] (see fig. 17). Anomalies
in KL(T) and Kc(T) are observed at T = TN. The behaviour of ~co(T) of DyS near TN
is analogous to that in PrA12 (cf. figs. 16 and 17) and could be explained by electron
scattering by the spin subsystem. It is more complicated, however, to explain the
behaviour of tCL(T) of DyS near TN.
0.50
0.3
0.45 •
%
u
0.40
6
0.2
0.35
Fig. 15. D e p e n d e n c e Ktot(T)=J%(T ) and D(T) for a
- I E ~ I "{ I I I I
60 70 80
m o n o c r y s t a l of E u O . Tc = 69.33 K ( S a l a m o n et al.
T,K 1974).
I I I I I ----
150
E
u
~" 100
E
2.
o 50
0 " I --L I I I
0 50 100 t50 200 250 300 Fig. 16. Dependence K,ot(T)~-I%(T)for PrAI2 (I) and
T,K SmA12 (2). Tc is the Curie temperature (Bauer et al. 1986).
0 gO 100 T, K
I I lllll 1 I I r tll~
O.1 -- "~
0.05 ,/
v / ,,,/./ ~-
/ // TN
/ Fig. 17. Temperature dependence of the thermal
o.ol ~1 ~ , , , 1 I ~ i I,,~,1, conductivity of DyS. Curve 1: ~ct,,t, 2: Kc, 3: iq.
5 10 20 50 t00 T,K (Novikov et al. 1975).
3.3.2. Behaviour of the thermal conductivity of lanthanide materials at a SOPT

At a S O P T the orientation of the magnetic moments relative to the crystallographic
axes changes under influence of external parameters (temperature, magnetic field,
pressure) (Belov et al. 1976, 1979, Belov 1972). If the S O P T occurs at a variation of
temperature or magnetic field, it is called, correspondingly, spontaneous or induced
transition. Transitions with spin reorientations are characterised by the ordering
parameter 00, the angle of magnetic moment rotation relative to the crystal axes. A
S O P T could be of the first (0 changes abruptly) or of the second kind (0 changes
gradually). A great number of SOPTs are observed in RECs (orthoferrites, ferrites
garnets, intermetallics) (Belov et al. 1976, 1979). The nature and characteristic features
of the SOPTs have been investigated most completely in the orthoferrites RFeO3.
The crystal symmetry of orthoferrites is described by the orthorhombic space group
17
- Pbnm. The orthoferrites have a distorted perovskite structure.
The most frequent reorientational transitions in the lanthanide orthoferrites are
transitions with spin and magnetic m o m e n t u m reorientations in the (ac) crystal plane.
At the temperatures Tls and T2~, which correspond to the beginning and to the end
of the reorientation process, one observes phase transitions of second kind. The
values of TI~ and Tzs in R F e O 3 vary over a wide range depending on the atomic
number of the lanthanide (Belov et al. 1976). At TI~ and T2~ there are anomalies
128 I,A. SMIRNOV and V.S. OSKOTSKI
of the elastic modulus, the heat capacity, the magnetic susceptibility, the frequencies
of soft spin modes become zero, and so on (Belov et al. 1976, 1979). The character
of the transitions changes significantly in (ac)-reorientation in an external magnetic
field H. When applying H along the a- and c-axes of a crystal the temperature of
the first phase transition is shifted and the other disappears (Belov et al. 1976). To
our regret, we could not find data on measurements of lc in a magnetic field for this
phase transition type.
The influence of a SOPT on ~c has been investigated in single crystals of SmFcO3,
Smo.6Gdo.4FeO 3 and Smo.6sEro.15Tbo.2FeO 3 (Barilo et al. 1984). Figure 18 shows
data on ~c(r), Oo(T) and Av/vt = f ( T ) (the relative variation of the sound velocity) in
Smo.6Gdo.~FeO3. Analogous results are obtained for two above-mentioned composi-
tions as well. For SmFeO3 and Smo.65Ero.l~Tbo.2FeO3 T~s = 454 and 303 K and
T2s = 487 and 358 K, respectively (Barilo et al. 1984).
RFeO3 are insulators and the experimentally measured ~ctot = ~CL.As one can see
in fig. 18a, at the temperatures T~s and T2~ ~c has minima. At Tt~ spins are oriented
along the a-axis, at T2s along the c-axis (fig. 18b). Anomalous behaviour of to(T) of
Smo.6Gdo.4FeO 3 in the SOPT region is connected with unusual behaviour of the
sound velocity (fig. 18c). Both anomalies are due to a strong spin phonon interaction.
As an example of the SOPT influence on ~cL we have listed the most impressive
experimental data. However, for other types of SOPTs t¢e could behave differently
at T = ~: from sharp changes of the ~cL value to the complete absence of an effect.
Besides data on the influence of the S O P T on tcL of RFeO3 there is information
I~F-- 'I I I--q

I 'a.
.fi 10
J
~' 6 I i I [ li
I
90 . . . . [-- --
o~ 60
30
0 I I ~ I
L
I I Q- Oxis
Fig. 18. Properties of Smo.6Gdo.4FeO 3 in the

SOPT region as functions of T (Barilo et al.
-2 xis 1984). (a) ~q. (b) 0o, the angle between the c-
axis and F, the light magnetisation vector in
-41- II ~ I I
200 300 400
the spin-reorientation regions. (c) A v / v t. The
T,K dashed lines indicate TI~ and T2~.
in the literature on the behaviour in the S O P T region in RCu2 (R = Sm, Ho, Dy,
Er, Tm) (Gratz et al. 1990, Gratz and N o v o t n y 1985). Theoretical calculations of KL
near the S O P T have been carried out by Buchelnikov et al. (1987).
3.4. Phonon magnon scattering, heat transfer by magnons in lanthanide compounds

( T < TN, Tc)
Magnetic materials insulators and semiconductors - at 7 ' < Tc (TN) can display
two effects:
(1) A decrease of 1eL as a result of phonon scattering by magnons.
(2) An additional thermal conductivity due to magnon heat transfer (tOm).
Since in metals bcL ~ ~cc, it is difficult to analyse the influence of phonon magnon
scattering on ice. In practice, one can consider only the contribution tq,. Electron
magnon scattering has to decrease the value tce in these metals. In magnetic lanthanide
insulators a p h o n o n - m a g n o n scattering effect or an additional heat transfer by
magnons could prevail. It is possible that both effects are comparable. As a rule, Km
contributes at lower temperatures while phonon m a g n o n scattering contributes at
higher temperatures (near Tc (TN)).
Let us first consider data on the thermal conductivity of insulating magnetic
lanthanide materials. Strong phonon m a g n o n scattering occurs when magnon and
phonon dispersion branches intersect. They perturb one another, branches split, and
an effective gap A appears between them (fig. 19) (Rives et al. 1969, Sheard 1976,
Kittel 1958). As a result, the contribution to the heat transfer of phonons with the
energy of the intersection decreases. In an external magnetic field the magnon
dispersion branch shifts to higher frequencies. The distribution of phonons carrying
the heat has a m a x i m u m near energy 4k0 T. If the intersection of the phonon and
magnon branches at H = 0 occurs at an energy less than 4ko T, then the thermal
conductivity first decreases and then increases with increasing H (curve l of fig. 20).
At different mutual dispositions of the dispersion curves variant 2 of fig. 20 is possible.
In the works of Rives et al. (1969), Sheard (1976), Dixon (1976, 1981), Dixon et al.
(1974) and Dixon and Landau (1976) some versions of p h o n o n - m a g n o n scattering
are considered: one p h o n o n - o n e magnon, two m a g n o n s - o n e phonon.
z3t
(00
,/< Fig. 19. Schematic shape of magnon (1) and phonon (2) dispersion
curves for a hypothetic fcrromagnet (Rives et al. 1969, Sheard 1976).
q(K) (3) The region of strong phonon-magnon interaction.
x (H)/x (o)
Hs
l
t - - 2 ( ( H s ) = %L
/ (magnon-- phonon
....... inter'action = O )
Fig. 20. Scheme of the dependence of ~c(H)/~:(O)on

magnetic field. H~ is the field in which ~c(H)/K(O)
saturates (gflHs>>koT). Curves 1 and 2: strong
X (bts)=X L
p h o n o n - m a g n o n interaction, 3: magnon heat
I (3~m=O)
transport.
In a high magnetic field (H > H~) the ratio ~c(H)/~c(O) saturates (fig. 20), and the
influence of p h o n o n - m a g n o n scattering on ~ci~ completely disappears, and ~c(Hs)
becomes equal to ~cL. As a rule, ~cL(Hs) is larger than ~cL(H= 0), since at H = 0
p h o n o n - m a g n o n scattering decreases ~i,. A high magnetic field (H > H~) could
suppress lcm as well (fig. 20, curve 3) (Rives et al. 1969, Martin and Dixon 1972,
McCollum et al. 1964). Then lc(H~)= ~cL. When analysing experimental data on the
thermal conductivity of lanthanide magnetic materials in a magnetic field one has
to keep in mind that:
(l) If heat transfer is to occur by magnons or p h o n o n - m a g n o n scattering it is
necessary to reach magnetic fields H >~ H~. At weak magnetic fields one could
erroneously take phonon magnon scattering for ~Cn~(fig. 20, curves 1, 2 at H < H~).
(2) Zeeman splitting of paramagnetic levels occurs in magnetic fields. At low
temperatures a decrease of ~cL due to resonance phonon scattering by these levels is
possible.
The contribution of /£m to /£ and the influence of p h o n o n - m a g n o n scattering on
~:L have been studied for a great number of lanthanide magnetics. References on
pioneer studies are found in Dixon and Landau (1976), Charap (1964), Slack and
Oliver (1971), Martin and Dixon (1972). In this review we will consider data on ~c of
the lanthanide insulating magnetic materials GdC13 (Tc = 2.2 K), EuO (Tc ~ 69 K),
HoPO~ (TN ~ 1.39 K) and the garnets DyA1G (TN = 2.5 K) and YIG ( 7 c ~ 545-
560 K), and for these materials we will analyse the influence of ~m and p h o n o n -
magnon scattering on their thermal conductivity.
In the ferromagnet GdCI 3 and the antiferromagnet H o P O 4 heat transfer by
magnons is insignificant. The dependence ~c(H)/~c(O) (figs. 21a, b) is analogous to
curves 1 and 2 in fig. 20.
By contrast, in EuO the effect of phonon magnon scattering is small and the heat
at low temperatures is transferred mainly be magnons. At T = 0.93 K the contribution
of icm to ~c in EuO is about 75%, fig. 22. This value is a record for the investigated
magnetics at the present time. However, simple dependences like those shown in figs.
21 and 22 (Dixon and Landau 1976, Walton et al. 1973, Metcalfe and Rosenberg
THERMAL CONDUCTIVITY OF' RECs 131
F--T--T--T ~I~T-'T'~
0.5
1.0 . . . . . . . . . .
L}'8 U I 1 I 1- I 1
1.5
0.4 . . . . . .
.~ 0 a.
2.0
0.4
0 I0 20 30 40 0 5 10 15
H, kOe H, kOe
Fig. 21. (a) Dependence A~c/~(H) for GdC13 (Rives et al. 1969). Curve 1:1.3 K, 2:0.46 K, Tc= 2.2 K,
AIc= I~(H)- t~(0). (b) Dependence tc(H)/t¢(0)for HoPO~ (Parsons 1976). T = 0.54 K.
1.0
0.8
0.6
3::
0.4
4
0.2
0
0 20 40 60
Fig. 22. Dependence K(H)/K(O)for EuO (Martin and Dixon
H , kOe 1974). T = 0.93 K.
1.0-- i I I i i i i
(3. b.
0.8
-~° 0.6
\\
.* \ \ B
'),
• k B
~ 0.4 \ ~] " "-Z"
0,2 2
0 I I I I I I I I
0 2 4 6 8 10 2 4 6 8 10
AH AH
Fig. 23. Dependence K(H)/K(O) for YIG (Walton et al. 1973) at temperatures 0.458 K (a) and 0.273 K (b).
A = #B/l% T. Curve 1: according to a theory without taking into account phonon-magnon scattering and
magnon heat transport. Curve 2: according to a theory which takes into account phonon magnon
scattering and magnon heat transport. B is the contribution from phonon-magnon scattering.
1979) are n o t always observed in experiments. For example, in garnets, parallel with
the p r e v a i l i n g heat transfer by m a g n o n s , the effect o f m a g n o n - p h o n o n scattering
p l a y s a significant role a n d its c o n t r i b u t i o n increases with temperature (figs. 23a, b)
(Dixon a n d L a n d a u 1976, Walton et al. 1973).
Peculiarities in the p h o n o n and m a g n o n dispersion curves can have influence on

the considered effects (Slack and Oliver 1971). Figures 24a, b shows dependences
~c(H, T) for GdCI3 (Dixon 1976, 1981) and D y A I G (Dixon and L a n d a u 1976), respec-
tively, and helps to illustrate more clearly our considerations on the contribution of
1~m and the role of p h o n o n m a g n o n scattering in t¢ of lanthanide magnetic materials.
At low temperatures in a magnetic field H = H~ one observes in GdC13 an increase
(+AK) and in DyA1G a decrease (-A~c) of the thermal conductivity. In GdC13
p h o n o n m a g n o n scattering is excluded as scattering mechanism and in DyA1G Gn
is suppressed.
Theoretical calculations have been carried out by Callaway (1959) of the two parts
of ~L(T) with T < To TN and T > To TN in a high magnetic field. It turns out that
b o t h parts can be described by using c o m m o n constants in the theoretical formulae
for p h o n o n scattering by crystal boundaries, point defects and p h o n o n s (with taking
into account the U and N processes). This points out that tc(H~, T) obtained at low
temperatures in both materials is equal to the real ~cL.
Let us consider experimental data on the thermal conductivity of the magnetic
lanthanide metals. As it has been noted at the beginning of this section, it is rather
complicated to separate Gn from the experimentally measured/£tot. In these metals
/('tot ~- K:L -[- Ke -t- 1Cm . (31)

The main difficulties arise in the separation of teL. At very low temperatures (where
/cm is most manifested) p h o n o n s are scattered, mainly, by crystal boundaries and
i% decreases rapidly with decreasing temperature (KLoc T3). Therefore, at these
I I I 1 I
I I I I I I I I I- 500 2.5K
200
1 H=O z ~
;o
E 50
u
--'-,'/I
2O
"~ 0.1 E
10 --2
5
/6~i ,~x • 3
O.011- 13. 2 b,
°1 I 1 I I I 1 1 I 1 I I I I i
0.2 1 10 100 200 O.1 0.5 1.0 5.0
T,K T,K
Fig. 24. (a) The temperature dependence of the thermal conductivity for GdC13 (Dixon 1976, 1981) at
H = 0 (1) and H = 3.5 kOe (3). The solid line (2) represents a calculation according to Callaway (1959)
taking into account phonon scattering by crystal boundaries, point defects and phonons (U and N
processes). The heat flow and magnetic field are directed ahmg the c-axis of the crystal. (b) The temperature
dependence of the thermal conductivity for DyA1G (Dixon and Landau 1976) at H - 0 (dotted line) and
H = 35 kOe (points 1). H II[111], heat flow II[1001. The solid line (2) represents a calculation taking into
account phonon scattering by crystal boundaries, point defects and phonons (U and N processes). The
curve for H = 0 is given for averaged experimental values.
temperatures m o r e often 1co ~ 1% (see, e.g., section 3.1). 1% is calculated in the first
a p p r o x i m a t i o n by the W i e d e m a n n - F r a n z law
1% = Lo T / p , (32)
where L 0 is the Sommerfeld value of Lorentz number. So, Km in magnetic metals is
calculated using the scheme:
(1) ~cc is supposed to be negligible in c o m p a r i s o n with t%.
(2) L etf is calculated from the dependence
Left :/¢tot T i p . (33)
If Left exceeds Lo, it is considered that there is a contribution of ~cm to/Cto~.
(3) ~:eo is calculated using eq. (32).
(4) ~cm is determined by
~cm = t g o t - x°. (34)
Let us consider, for example, the separation of Km in the alloy ErNi (Tc ~ 105 K)
(Mori et al. 1984). Figures 25a, b, c illustrate the procedure of the determination of
~¢m' One could expect a decreased ~ce in the magnetic lanthanide metals due to
electron m a g n o n scattering. However, we have not succeeded in finding reliable
literature data on this question.
Another way for a direct separation of 1¢m in magnetic metals exists, but for some
reason n o b o d y has used it. It is possible to suppress ~Cm in a strong magnetic field
(H > H~) and determine ~Ce, and then to calculate bcm. In principle, the magnetic field
could influence t% as well, but because of the low mobility of current carriers in the
lanthanide metals it is practically impossible to reach a sufficiently high magnetic
field (H~0) for the suppression of carriers (H~o > H~).
4. Thermal conductivity of spin glasses
A spin glass is a new magnetic state differing from the ferro- and antiferromagnetic
state. A spin glass is formed by introducing magnetic impurities into a nonmagnetic
x 10-2
a. I r----S_
:x: 3
E 0.02
E x10-8 t 1 --.2.
~'2 [b Le, / ~, 00"
\ 6t- ;..-'~-. 1 ¢ 000~.
1
•
i I } o2 / ~ "~ -I ~ o.oo, I I
5 10 15 0 5 10 15 5 10 20
T,K T,K T,K
Fig. 25. The temperature dependence for ErNi: (a) of the total thermal conductivity (Mori et al. 1984),
(b) of the effective Lorentz number calculated using eq. (33) and (c) of/¢m calculated using eq. (34).
134 I.A. SMIRNOV and V.S. OSKOTSK!
(or weakly magnetic) matrix. In spin glasses the orientation of impurity spins varies
randomly in space, just as atoms are randomly arranged in glasses (Maletta 1982,
Fisher 1983, 1985, Mydosh 1978, Korenblit and Shender 1984, Maletta and Zinn
1989). The resulting magnetic structure is called "spin glass" (fig. 26). The spin-glass
state is the result of competition between ferro- and antiferromagnetic interaction in
a disordered system. The macroscopic moment is zero. Spin waves in spin glasses
are not observed. Therefore ~Cmequals zero. Spin glasses can be metals or insulators.
By structure they can be crystalline (metals and insulators) or amorphous materials
(metallic glasses). Classical spin-glass systems are alloys: CuMn, AuFe, ZnMn, AuCr,
AuMn, AgMn, MoFe, RhMn, pdFe, PtCo and others (Fisher 1983). There are also
a great number of spin glasses based on rare earth elements: rare earth metal alloys,
metal and insulator crystals, and metallic amorphous systems (Fisher 1983, 1985,
Mydosh 1978, Durand and Poon 1979).
Figure 27 shows a scheme of the formation of metallic spin glasses (Mydosh 1978).
A small number of magnetic impurity ions in a nonmagnetic matrix forms an impurity
Kondo system. At a critical concentration xcr (in metal systems on the average equal
to 50 ppm) the system changes to a spin-glass state. At higher concentrations
I+,,,
ltt,+
-rl
i x I Fig. 26. Schematicpicture of the magnetic momcnts in ferro-
magnetic (FM), antiferromagnetic(AM) and spin glass (SG)
FM AFM SG systems.
Kondo regime Spin glass Long range order

_ _ A _ _ A /
/ /
Interaction Cluster formation Clusters in the Ferromagnetic,

Single single spins (pairs, triplets, I spin glass anti ferromagnetic
impurities etc, ) matr x
Magnetic
N .50 ppm percolation
limit
TK=To (C)
~._ Concentration of magnetic impurity (C) >

Fig. 27. Scheme of formation of metallic spin glasses (Mydosh 1978).
(x > XMpL,magnetic percolation limit) the system becomes magnetically ordered. The
scheme of fig. 27 shows that interpretation of experimental data on thermal conduc-
tivity (specially on ~ce)is rather complicated, because two influences on the Lorentz
number are mixed: an effect which is specific to the spin-glass character, and the
Kondo effect (see section 5).
p in metal spin glasses varies with temperature (at low temperatures) as T 3/2 (more
often) and as T 2 (more seldom). Patterson (1978) and Sharma and Tripathi (1980)
have calculated the Lorentz number theoretically for metallic spin glasses with taking
the temperature dependencies of p into account. It turns out that at tow temperatures
L / L o ~- 0.98 (at p ~ T 3/2) (Patterson 1978), (35)
L / L o ~- 0.55-0.44 (at p ~ T 2) (Sharma and Tripathi t980). (36)
Unfortunately, we were unable to find experimental studies on the thermal conductiv-
ity of lanthanide metallic spin glasses which attempted to verify eqs. (35) and (36).
Figure 28 shows experimental data for the metallic spin-glass alloys AgCr. As one
can see, the experimental and theoretical results agree rather well (Ducastel and Pitsi
1986).
Let us consider data on the thermal conductivity in spin-glass insulators. In them
~Ctot = ~cL. A typical representative of spin-glass insulators is EuxSrl xS. When doping
the Heisenberg ferromagnet EuS (Curie temperature Tc = 16.6 K, lattice of NaC1-
type) with nonmagnetic Sr (SrS is diamagnetic with NaCl-type lattice) the ferromagnet
order is destroyed and a spin glass is formed. According to the magnetic phase
diagram of EuxSrl_~S (Maletta 1982) (fig. 29a):
(a) for x > 0.7 the system behaves like a simple dilute Heisenberg ferromagnet
(region I);
(b) for x~ < x < 0.7 the system is in a reentrant or "frustrated" ferromagnetic state
(region II);
(c) for xp < x < x~ ~- 0.51 spin-glass behaviour is found (region III);
(d) for x < xv superparamagnetism of isolated Eu clusters is observed (Xp = 0.136)
(region IV).
The thermal conductivity of the EuxSrl_~S system in the temperature region
0.06-30 K and in magnetic fields up to 7 T has been measured by Arzoumanian
et al. (1983, 1984), L6hneysen et al. (1986) and Lecomte et al. (1983, 1984, 1986). The
compositions in the different regions of the magnetic phase diagrams have been
investigated: I (x = 1), II (x = 0.54), III (x = 0.25, 0.40 and 0.44), IV (x = 0.1, 0.017,
0.005) (fig. 29a). Figures 30 and 31 show data on lCtot(Z ) for Eu~Srx ~S single crystals
from all phase-diagram regions.
-~ 0.9 _ _ ~- 7 --v~-
0.8 I I I I I I I I Fig, 28. The temperaturedependenceof L/Lo for the met-

O .4 8 12 allic spin glass AgCr (1) (Ducastel and Pitsi 1986) and
T,K (2) calculationusing eq. (35).
136 I.A. SM1RNOV and V.S. OSKOTSKI
o.l I I
L I I ~T'--T-~T I I
I I
_,
I pN I ZM
x:
l
I
,
I ~
/I
... I SG
5 _ b.~ 1 I I 7- I I --T----
4
3
2
rE 1
-fi
0
-1
Sr5 I[ I I I I I I 1 Eus
I o,2
Xp ~ X
0.4
x¢
0.6 O.8
t
1,0
I
Fig. 29. (a) Magnetic phase dia-
gram for Eu=Srl_xS (Maletta
1982). (b) Dependence of 1¢(H)/1~(0)
Dilute Spin glass Ferromagnet on x = l . 6 K and H = 6 T (based
superparamagnet J I I on the experimental data of Belay
I L Reentrant ferromagnet (1972) and Belov et al. (I979).
4=16.9T'~
J / z " X=(Z25
I 0-1 / / / H=O
// Tc //~'/
uE / / / X=O54
H=6T//~,'
2
T// H=O
10.2
I 1 J I I ]11J L
2 5 10 20 Fig. 30. The temperature dependence of Jq,,t for EuxSrl-xS single
T,K crystals (x = 0.25 and 0.54) (Lecomte et al. 1983).
Let us c o n s i d e r d a t a for ~ m e a s u r e d in zero m a g n e t i c field (H = 0). S o m e r e g u l a r i t i e s

c a n be n o t e d :
(1) N o a n o m a l i e s of lCL n e a r Tf (the "freezing" t e m p e r a t u r e for the s p i n - g l a s s
system) h a v e b e e n o b s e r v e d (for c o m p o s i t i o n s w i t h x = 0.44 a n d 0.54 the v a l u e s Tf
THERMAL CONDUCTIVITY Of" RECs 137
:0017 /// ~'NN

I0 -I
lo -2
J,,
2.10-3 i i x alibi I I ~
10
Fig. 31. The temperature dependence of Got for EuxSr I ~S single
T,K crystals (x = 0.44 and 0.017) (Arzoumanian et al. 1983, 1984).
are essentially equal, respectively 1.8 and 2 K). The same effect has been observed
also in the thermal conductivity of metallic a m o r p h o u s spin glasses (see, e.g. Herlach
et al. 1981).
(2) N o anomalies of ~CL have been observed at Tc---4.5 K for samples with
x = 0.54.
(3) At low temperatures (for the samples with x = 0.25, 0.44 and 0.54) ~cc obeys a
dependence T 18 2 and not T 3, as would be expected for p h o n o n scattering by
crystal boundaries.
(4) D a t a on the t e m p e r a t u r e dependence of ~CL can be described satisfactorily in
the model of a magnetic two-level system (MTLS) ( A r z o u m a n i a n 1983, 1984). This
model explains well the behaviour of teL(T) in metal spin glasses (see, e.g., Herlach
et al. 1983). The m o s t interesting data on 1%(T) have been obtained in magnetic
fields (along the heat flux) (figs. 29b, 30, 32, 33). In the spin-glass region III (x = 0.25,
0.4) at low temperatures an a n o m a l o u s l y large increase of tc in a magnetic field is
observed (fig. 29b). F o r compositions in region IV (x = 0.005 and 0.017) a magnetic
field does not have an influence on the ~c value (figs. 29b, 32). In region II (x = 0.54)
the ratio t¢(I-I)/~c(O) increases but not so m u c h as in the spin-glass region (cf. data for
x = 0.44 in fig. 29b). In EuS (region I) the value of ~c(T) in a magnetic field does not
increase, but, on the contrary, decreases (figs. 29b, 32, 34, 35) (Lecomte et al. 1984,
I I I
0 --
5
I I I
Fig. 32. Magnetic field dependence of the thermal conductivity of
2 4 6 Eu:,SQ ~S (1-4) (L6hneysen et al. 1986) and EuS (5) (Lecomte et al.
H(T) 1984) at 1.6 K plotted as lc(H)/~(0);x = 0.4 (1), 0.25 (2), 0.54 (3), 0.005 (4).
138 I.A. S M I R N O V a n d V.S. O S K O T S K I
1.5 7 1 I ~ I
1.4
s
~: 1.3
I' 1.2
11
1.0 i i i 1 Fig. 33. Magnetic field dependence of the thermal conductivity for
i i
0 2 4 6 Euo.s4Sro.46S at three fixed temperatures T= 1.65 K (1), 3.85 K (2),
H(T) 13.5 K (3) (Lecomte et aL 1983).
O I O ~ F t
0.5
n-
O Fig. 34. The magnetic field dependence ~c(H)/~(O) for EuS at

0 2 4 6
H(F) T= 1.65 K (Lecomte et al. 1984).
~: 10-2
E
tA
2xlQ-3 1 I i I Fig. 35. The temperature dependence of the thermal conductivity of EuS
2 5 in various magnetic fields: H = 0 T (1), 0.77 T (2), 3.22 T (3), 6.45 T (4)
T,K (Lecomte et al. 1984).
1986, M c C o l l u m et al. 1964). A thermal conductivity which decreases in magnetic

fields at low temperature, is not unique. This effect has been discussed in section 3.4.
The effect of a giant increase of the thermal conductivity in a magnetic field in the
spin-glass region needs special explanation. It cannot be explained, for example, by
an increase in the heat capacity C, because, on the contrary, C in magnetic field
decreases (fig. 36) (L6hneysen et al. 1986). M o r e probably, the strong increase of ~Ctot
in a magnetic field is connected with field-induced m a g n o n heat transport.
As it has been noted above, one does not observe spin waves in a spin glass, and
hence ~cm is absent. The magnetic field, however, orders the spin a r r a n g e m e n t (fig. 37)
and K;m arises. A sufficiently high magnetic field must lead to "freezing" ~cm (like in
the case of EuS) and the ratio ~c(H)ffc(O) must begin to decrease at H > Her (fig. 38).
However, experiments in such high field have not yet been carried out. A r z o u m a n i a n
et al. (1983) and Lecomte et al. (1983) have given a n o t h e r explanation of the observed
THERMAL CONDUCTIVITY OF RECs I39
10-2 m
#
l
l
& m-3
I
u
II
10-4 II
II
211
10-5 I
t
I I I Fig. 36. The heat capacity C of Euo.s4Sro.,6S as a function
0.1 1.0 10 of temperature, T, in two magnetic fields: H = 0 T (1), 6.6 T (2)
T,K (L6hneysen et al. 1986).
I÷\
I , , #/#
SG SG (H I SG (H2)
Fig. 37. Magnetic moments in spin glasses in various
H2 > H1 magnetic fields: H = 0 (1), H 1 (2), Hz (3); H2 > HI.
~T= const i
8 I
A / / ~ l \\\
I \\
ii
1/ L-, i \
I
Hcr H Fig. 38. Schematic shape of the dependence of ic(H)ffc(O).
effect. T h e increase of the r a t i o ~c(H)/t¢(O) in the spin-glass region is connected with

decreasing p h o n o n - m a g n o n scattering, since spins o r d e r in the m a g n e t i c field. H o w -
ever, this contradicts, a c c o r d i n g to L e c o m t e et al. (1986), the e x p e r i m e n t a l d a t a on the
absence of the influence of a m a g n e t i c field on samples with x = 0.1 a n d 0.017 outside
the spin-glass region. B o t h h y p o t h e s e s need further e x p e r i m e n t a l a n d theoretical
study.
5. Thermal conductivity of systems with heavy fermions, intermediate valency of

lanthanide ions, dilute and concentrated Kondo lattices
5.1. Main principles, theoretical premises
As it has been noted already in the Introduction the systems noted in the section
title are a m o n g the most interesting and unusual materials known. They are, as a
rule, metals. Figures 39a, b, c show schemes of their electron structures and figs.
39a', b', c' show, as an example, schematically the unusual behaviour of p(T) of these
classes of materials (almost all physical parameters behave anomalously). One can
conclude from the band pictures that peculiarities of the behaviour of ~c in these
materials must reveal themselves mainly in G (especially in the temperature depen-
dence of the Lorentz number). Considerable difficulties arise, however, in the separa-
tion of t¢e and /£e from ~tot. In m a n y cases the contribution ~cL to /Ctot is rather
significant over a wide temperature interval, due to low mobilities (and consequently
not a small p). An incomplete exclusion of teL could give rise to an overstated value
L if calculated according to the W i e d e m a n n - F r a n z law [L = (tqot - ~cL)T/p]. Unfortu-
nately, the separation of ~ce and ~c~ with a high enough precision is a rare case, and
experimentalists are satisfied with approximate methods not always yielding a true
temperature dependence L(T). The more reliable data on L(T) can be obtained in
the low- and superlow-temperature regions ( T < 4 K and T < 1 K, respectively),
where one could consider ~cL ~ tcc with a high precision.
In a number of theoretical works the behaviour of ~ce(T) and L(T) of systems with
concentrated (CKL) and dilute (DKL) K o n d o lattices have been investigated (Fisher
1971, N a k a m u r a et al. 1987, Bhattacharjee and Coqblin 1988, Cox and Grewe 1988,
c~ b c
Fig. 39. Schematic shape of the electron band

structure (a, b, c) and p(T) (a', b', c') for metals
with Kondo impurities (a,a'), concentrated
Kondo systems (including systems with heavy
i,
fermions) at T < TK (b, b') and for compounds
T T T
with homogeneous intermediate valency of
(3' c' rare earth ions (c, c').
Belitsky and Goltscv 1991, Bhattacharjee et al. 1989, Raki et al. 1990), and the
specific behaviour of L(T) intrinisic only in these systems has been predicted (or
explained, if experimental data had been obtained previously). Analysing experimen-
tal data on L(T) of C K L and D K L systems one must remember that in normal
metals L = Lo due to the scattering of electrons by static defects, and L < L 0 in thc
low-temperature region due to
(a) inelastic scattering by long-wavelength acoustic phonons (see section 2, fig. 4)
and
(b) scattering of electrons by split crystal field ground state levels of paramagnetic
rare earth ions (see section 6.2). L can be larger or smaller than Lo at interband
scattering of the current carriers in the presence of a complex band structure with
heavy and light bands (see section 2, fig. 6). The last case is close to the situation in
the class of materials considered in this section and reminds one of the situation
shown in figs. 39b, c.
Discussing results on ~ of heavy-fermion systems one must distinguish two temper-
ature regions: a low-temperature one, T < TK, and a high-temperature one, T > T~,.
The K o n d o temperature, TK, corresponds to the delocalization of f-electrons due to
Kondo interaction with free electrons:
TK = (eF/ko ) exp [-1/2lg(e F)]. (37)
g(f;F) is the state density at the Fermi level, ev, I is the exchange integral of the s f
interaction. TK is usually of the order of only a few or dozens K. The region T < TK,
in its turn, is divided in two intervals: T < T* and T*~< T~< TK. Here T* is the
temperature below which one must consider the interaction between heavy fermions
as being coherent. In this temperature region the law p ocDT 2 (p = Po + DT2) is
fulfilled. (In systems with heavy fermions D is about two orders larger than in normal
metals.) T* is smaller than TK. For example, in CeAI3 TK --~ 5 K and T* ~- 350 m K
(Flouquet et al. 1985). At T < TK a peak in the state density arises due to the
Abrikosov-Suhl resonances (fig. 39b). At T-~ TK the peak spreads out, and its ampli-
tude sharply decreases. At T >> TK the singularity in the state density disappears and
the system turns into a K o n d o impurity state. The theoretical formulae describing
the variation of L/Lo with temperature and magnetic field in the region T < T* are
taken from Belitsky and Goltsev (1991). These formulae take into account scattering
of current carriers by the usual impurities (defects, impurity atoms) [eq. (38)] and by
so called "exchange energy defects" [eq. (39)]. The first defects do not give rise to,
and the second defects give rise to, a change of the local exchange integral between
localized f-electrons and conduction electrons. An example of a defect of the second
type are substitutional impurities - nonmagnetic rare earth ions in a Kondo lattice.
16 q0 7c2T2
L(T, H) = 1 + [1 + ~ J ( J + 1)(g#,H/ro)2], (38)
15no rg
16 7rz T 2
L ( T , H ) = 1 + 15 T~- [1 - ½ J ( J + I)(g#RH/To)2], (39)
where qo = Q/N (N is the degeneracy of the f-level, Q the number of electrons in the
f-level), Nn c is the number of electrons in one level in the conduction band, To is like
by nature to T*,
To = (no/vo) exp(1/VoI ), (40)
where vo is the band-electron state density.
It follows from eqs. (38) and (39) that in the coherent region at low temperatures
L > Lo, which is in contrast to normal metals with impurities, where L = Lo. Such
an effect is only possible in a Kondo lattice. For the standard metals another equation
for L/L o is used, the correction to L o is proportional to T/eF and is small. Unfortu-
nately, at present there is no clear physical explanation for the fact that L exceeds
Lo in K o n d o systems at low temperatures. In the extreme case of strong inelastic
fermion-fermion scattering eq. (39) turns into
L(O, O)/Lo = 36/7r2 -- 3 _~ 0.648. (41)
This was shown in some other works [see, e.g., Cox and Grewe (1988)]. With
decreasing temperature L/Lo increases, reaches L = Lo, and then behaves according
to eqs. (38) or (39). Such a behaviour is the result of competition of different scattering
mechanisms for current carriers (inelastic fermion-fermion, elastic scattering by static
defects and the peculiar scattering characteristic for a K o n d o lattice). For T > TK
one expects that in the Kondo-impurity region L < L o (fig. 40), see Nakamura et al.
(1987), Bhattacharjee and Coqblin (1988), Cox and Orewe (1988). Thus, we have
schematically presented the dependence L/Lo = f ( T ) for the heavy-fermion system in
fig. 41.
-- I I I I !
1.5
o 10
0.5
0 l ~__ 1 1
Fig. 40. Calculated temperature dependence of L / L o for
0 0.2 0.4 0.6 heavy-fermion systems based on Ce. x is the concentration
of Ce. Curve 1: concentrated Kondo system ( x - 0 . 9 8 ) ,
T/T K 2: dilute Kondo system (x = 0.2) (Nakamura et al. 1987).
TK
1.0
L
Coherent I Incoherent
scattering I scattering
t
Fig. 41. Schematic shape of the temperature
dependence of L / L o for a heavy-fermion system.
5.2. Experimental results
To date ~c has been investigated in a large n u m b e r of C K L and D K L systems.

Table 3 gives a list of the most i m p o r t a n t c o m p o u n d s investigated. It includes also
data on the classical heavy-fermion systems UBe13 and UPt3 to make the picture
complete. We will discuss ~c of these c o m p o u n d s in section 8.2, analysing ~ of the
superconducting heavy-fermion systems. Also presented in table 3 are the crystal
structures, the values of 7, the electron specific-heat coefficient (C~ = 7T), which gives
information on the electron state density (and hence on the value of the effective
mass) at the Fermi level. (For nonmagnetic materials 7 is defined as 7 = lim(C/T),
while for magnetic materials 7 is defined by extrapolation to zero temperature from
above the ordering temperature.) According to a c o n t e m p o r a r y classification (De
Visser et al. 1987) the c o m p o u n d s can be separated into two groups: the true heavy-
fermion systems (7/> 400 m J / m o l K 2, m*/m ~ 102-103) and C K L s with "half-heavy"
fermions (7 -G<400-200 m J / m o l K 2) (in standard metals ~ ~ 1-10 m J / m o t K2). The i~
of a n u m b e r of D K L systems has been measured: in C e x L a l _ x C u 6 (Bauer et al.
1987), C%Lal_xA12 (Steglich 1976, Moeser and Steglich 1975) and CePd3, a com-
p o u n d with an intermediate valence (Schneider and Wohlleben 1981). All systems
with heavy and "half-heavy" fermions can be divided into three groups: nonmagnetics,
magnetics and superconductors (see table 3). The most typical representative of these
TABLE 3
Cerium and uranium materials related to concentrated Kondo lattice systems for which the thermal
conductivity has been measured.
Compound 7, Crystal State Refs.*

mJ/mol K 2 structure
CeA13 1620 hexagonal nonmagnetic [1 6]

CeCu 6 1600 orthorhombic nonmagnetic [7-9]
CeCuz Siz 1100 tetragonal superconducting [3, 10-13]
UBe13 1100 cubic superconducting [9, 4, 5, 14, 15]
UPt a 422 hexagonal superconducting [5, 10, 16-20]
CeRu2Si2 350 tetragonal nonmagnetic [5, 2 l ]
CeB6 ~ 255 orthorhombic antiferromagnet [5, 23, 24]
CeA12 135 cubic antiferromagnet [3, 24]
CeCu2 82 orthorhombic antiferromagnet [25]
CePt z Si2 80-86 tetragonal nonmagnetic [26-28j
*References: [1] Flouquet et al. (1985), [2] Ott et al. (1984a), [3] Sparn et al. (1985), [4] Jaccard and
Flouquet (1987), [5] Flouquet et al. (1986), [6] Andres et al. (1975), [7] Brfick et al. (1986), [8] Peysson
et al. (1986b), [9] Jaccard and Flouquet (1985), [10] Steglich et al. (1985a), [11] Franz et al. (1979),
[12] Franz et al. (1978), [13] Schneider et al. (1983), [14] Jaccard et al. (1985a), [15] Ravex et al. (1987),
[16] De Visser et al. (1987), [17] Jaccard et al. (1985b), [18] Sulpice et al. (1986), [19] Franse et al. (1985),
[20] Steglich et al. (1985b), [21] Amato et al. (1989), [22] Peysson et al. (1985), [23] Marcenat et al.
(1990), [-24] Bauer et al. (1986), [25] Gratz et al. (1985), [26] Bhattachrjee et al. (1989), [27] Raki et al.
(1990); [-28] Ayache et al. (1989).
groups are: CeA13 [nonmagnetic; although Berth et al. (1987) have observed in CeAl3
weak magnetic ordering], CeAI=, CeB6 (magnetics) and CeCu2 Si2 (superconductor).
Data on ic of the first two groups are considered in this section, data on superconduc-
tors will be discussed in section 8. Let us note some general features of these systems.
(1) Ce has an integer valence in all compounds listed in table 3.
(2) At low ( T < 4 K ) and superlow ( T < 1 K) temperatures ~L ~ ~Ce and thus the
experimental tCto t is equal to G.
(3) At elevated temperatures one cannot neglect the contribution of tcL in metals
with a small free path length for electrons.
(4) 7£tot of heavy-fermion systems, as a rule, is much smaller than l£tot of related
rare earth compounds which are not heavy fermions (see, e.g., figs. 42 and 12).
5.2.1. Nonmagnetic heavy-fermion systems

In the classical heavy-fermion compound CeA13 (TK ~-- 5 K, T* ~ 0.35 K) (Jaccard
and Flouquet 1985, 1987, Sparn et al. 1985) at T < 1 K (the region T<~ T*) L/Lo
behaves according to the theoretical dependence (38) (fig. 43). At ~- 50 mK there is a
maximum (L/L o ~ 1.1) and near 350 mK a minimum (connected with inelastic fer-
mion-fermion scattering). At T > 1 K the ratio L/Lo tends to one and then, in the
I I l - I I
20
E
u
%
:~ 10
0 I I I I I
0 50 100 150 200 250 300 Fig. 42. The temperature dependence of ~ctot in CeCu z
T,K and YCu2 (Gratz et al. 1985).
-~'-'-[-- I I i
1.1
o 1.0
O.9
0.8
i ! I I I
0 0.5 1,0 Fig. 43. The temperature dependence of L/L o in CeA13 (Sparn
T,K et al. 1985, Jaccard and Ftouquet 1987).
Kondo-impurity region, L/Lo exceeds one. At T < 1 K it is assumed that lee ~ %, as

has been noted above.
The increase of L/Lo at T < 500 mK occurs due to intensifying (with decrease of
temperature) current carrier scattering by static defects (since Po becomes more
important than D/T 2) and the specific effect of L/Lo increases (over one), which is
characteristic for the heavy-fermion systems (Belitsky and Goltsev 1991). In this
temperature interval tCeOCT -1 (fig. 44) (Flouquet et al. 1985). In CeA13 leE and K~
have been separated between 0.1 and 50 K (fig. 45) (Ott et al. 1984a). % is calculated
using the Wiedemann-Franz law under the supposition that L = Lo. It turns out~
that the contribution iCE to ~Ctot at T > 1 K is rather large, but it becomes negligible
at T < 1K. If at T > 5 K ( T > TK) the ratio L/Lo begins to exceed one, as is expected
according to the theory for the Kondo-impurity region (Nakamura et al. 1987,
Bhattacharjee and Coqblin 1988, Cox and Grewe 1988), then the ~cL contribution to
/£tot of CeA13 is small.
An interesting result should be noted. The separation of the lattice thermal conduc-
tivity carried out above gives for CeA1 a ~cLoc T, and not T 2 (as for phonon scattering
by electrons) as well as not T 3 (as for boundary scattering).
CeCu2Si2 is another classical nonmagnetic heavy-fermion superconductor
(To ---0.6 K, TK --~ 10 K, T* ~- 1 K) (Sparn et al. 1985, Steglich et al. 1985a, Franz et al.
I I I I I I
1.5
I
E
u 1.0
E
2.
2
N
0.5
0 I I I I Fig. 44. The temperature dependence of %o, in CeAI~

O 1.0 2.0 3.0 (Sparn et al. 1985, Jaccard and Flouquet 1987). t% ~ ~:~
TtK
at T < I K .
10 ' I ' I ' I i
.,
1.0 H
xtot ,if,"
E
-~. 10-'1
//
/
10-2 / xk_
//
/
i0-3 i I ~ i , i i
10-2 10-I 1.0 10 100 Fig. 45. ]-'he temperature dependence of the various contributions
T,K to ~c in CeA13 (Ott et al. 1984a).
1978, 1979, Schneider et al. 1983). The low-temperature data on ~c used for the
analysis of L(T)/L o are obtained at a magnetic field (H = 2.5 T) in which the sample
is in the n o r m a l state (fig. 46) (Sparn et al. 1985). As one can see, up to T_~0.7 K
the ratio L/Lo = 1 and then L/L o becomes > 1. The "excess" A~c of Ke in c o m p a r i s o n
with the calculated curve (curve 1) is connected by Sparn et al. (1985) with the neglect
of the ~cL contribution to h2tot, At( is p r o p o r t i o n a l to T 2 and at 1 K A~c _ 0.2 m W /
K cm, which coincides with •L extrapolated from the high-temperature region for
CeCu2Si2 obtained by F r a n z et al. (1978). (Similar behaviour of L(T)/Lo is observed
in the heavy-fermion superconductor U P t 3 (De Visser et al. 1987).) F r a n z et al. (1978)
have attempted to separate ~cL and i% of CeCu2Si 2 in the interval 1.5-300 K (fig. 47).
The contribution ~cL to Ktot turns out to be rather large over the whole temperature
interval. L(T)/Lo calculated for CeCu2 Si2 from 1%(T) (fig. 47) and p(T) (Franz et al.
1978) is shown in fig. 48. In region I, where CeCu2Si 2 can be considered as C K L
system, the ratio L/Lo < 1. In region III, corresponding to a K o n d o - i m p u r i t y system,
L/Lo > 1, which agrees well with theoretical predictions ( N a k a m u r a et al. 1987,
Bhattacharjee and Coqbtin 1988, Cox and Grewe 1988, Belitsky and Goltsev 1991,
Bhattacharjee et al. 1989) and with the scheme of the behaviour of L(T)/Lo (fig. 41).
CeRu2 Si2 (TK ~-- 24 K, T* -~ 700 m K ) is a "half-heavy" fermion system. It is
I 1
/
/
/
0.8
2
le/A X~ T
/
O/
" ......
uE
0.4
E
Fig. 46. The temperature dependence of 1~in polycrystal-

¢'// line CeCu2.02Sil.98 (Sparn et al. 1985). The solid line
I 1
gives ~ce calculated using the Wiedcmann-Franz law with
O
O 0.5 1.o L = Lo, The dotted line gives the experimental result at
T~K H = 0, points the experiment result at H = 2.5 T.
1 7"--T I'--
100 x to t ~ ~ "
Y 7-,
!
3O
-.V.. ~o
/
///
•~ 3
/11 Xe
/
/ Fig. 47. Experimental (~Ctot)and calculated (KL and ~ce)thermal
1 I I I conductivities as functions of temperature in the polycrystalline
10 30 100 3 0 0 1000 sample CeCu2Si2 (Franz et at. 1978). Curve 1: lq, calculated
T,K according to Leibfried and Schl6mann (1954).
_o - I
Fig. 48. The temperature dependence of L/L o in

0 , I I A I I l CeCu2Si z (Franz et at. 1978). I is the coherent state
0 100 200 300 region, II the intermediate region, and III the Kondo-
T,K impurity region.
~9, #.Q • c m
4.0 - B.5
t L-- I l--t I ( t I i
x:
3.5
£
u 5.5
E
3.0
4.5
2.5 I
2.0 1 I I I I L I I I 1 3.5 Fig. 49. The magnetic field dependence of ~%and p in a

0 2 4 6 8 10 single crystal of CeRuzSi / (Amato et al. 1989). ~c, pLc,
H,T HHe, T= 640 inK.
1.2 i i 1 l i l I I
o
+co
._i
0.8
I
I I I I I
I
i Ii t
Fig. 50. The magnetic field dependence of L/L o in a single
2 4 6 8 10 crystal of CeRu2Si 2 (Amato et al. 1989). ~c, pLc, HIlt,
H,T T = 640 mK.
i s o s t r u c t u r a l with CeCu2 Si2, b u t it is n o t a s u p e r c o n d u c t o r . C e R u 2 Si 2 u n d e r g o e s a

m e t a m a g n e t i c - l i k e t r a n s i t i o n at T < 10 K in a n a p p l i e d m a g n e t i c field H* _~ 8 T a l o n g
the c-axis. It t u r n s out t h a t the m a x i m u m of p(H*) coincides with the m i n i m u m of
~:e(H*) (fig. 49) ( A m a t o et al. 1989). At low t e m p e r a t u r e s ( T < 1 K) ~cL ~ ~e. F i g u r e 50
shows the d e p e n d e n c e L(H)/Lo for T < T* (T* = 640 m K ) for C e R u z S i 2. L/Lo < l
up to H - ~ 7 T, which suggests the presence of s t r o n g inelastic f e r m i o n - f e r m i o n
scattering characteristic for a h e a v y - f e r m i o n system. T h e increase of L/Lo at H > 7 T
can be explained, possibly, in the f r a m e w o r k of the theories d e v e l o p e d by N a k a m u r a
et al. (1987), C o x a n d G r e w e (1988) a n d Belitski a n d G o l t s e v (1991).
148 I.A. SMIRNOV and V.S. O S K O T S K |
5.2.2. Magnetic heavy-fermion systems

An evaluation of the 1% contribution to xtot has been carried out for CeCu2
(TK ~ 17 K, TN = 3.5 K) by the method considered in section 3.2 (Gratz et al. 1985).
It turns out that in a rather wide temperature interval 1% ~ ~cc. Figure 51 shows the
dependence L(T)/L o of CeCu2 in the temperature interval 3.5 300 K (Gratz et al.
1985). At low temperatures (but for T > TN) there is a m a x i m u m L/L o ~_ 1.11 at about
20 K. A minimal value of L/Lo = 0.8 is reached at ~- 80 K. Thus, CeCu2 has a L(T)/
Lo dependence which is typical for a heavy-fermion system (cf. with L(T)/L o for
CeA13, fig. 43).
In CeA12 (TK --~ 5 K, TN --~ 3.8 K) /% ~ 1<¢ only at T < 4 K (i.e. for T ~< TN) (Sparn
et al. 1985). Figure 52 shows L(T)/Lo of CeA12 at T > 5 K. L/Lo = 1 at T = 0.8 and
4 K. Between 0.8 and 4 K the ratio L/Lo < 1, which is probably caused by inelastic
scattering of the current carriers by magnons. This conclusion is confirmed by the
results of Marcenat et al. (1990) on ~c of CeB 6. CeB 6 is a typical representative of
C K L systems. In contrast to classical nonmagnetic K o n d o systems, it undergoes
two magnetic phase transitions, at T1 = 3.3 K and T2 = 2.4 K. Three phases are
distinguished (Effantin et al. 1985):
- p h a s e I (T > T1, paramagnetic),
- phase II (T2 < T < TI, antiferro-quadropolar ordered phase) [a partial magnetic
ordering is observed when applying a magnetic field (Effantin et al. 1985)],
I l I t I
1.A
o 1.0
_J
u
0.6
0.2
I I I I f
50 100 150 200 250 3oo Fig. 51. The temperature dependence of L / L o in
T,K CeCu 2 (Gratz et al. 1985).
I J! o , i
E 2 TI - i
?
1 FCL=Lo L < Lo L >/ Lo Fig. 52. The temperature dependence of tc in
I/ a CeA12 single crystal (Marcenat et al. t990).
oV
0
' ,
2
,
3
',
4 5
The solid line gives the calculation with
L = Lo, points the experimental result; the heat
T,K flow is along [100].
T H E R M A L C O N D U C T I V I T Y OF RECs 149
- phase III ( T < T2, antiferromagnetic, T2 = 7;~).

Figure 53 shows experimental data o n K t o t ( r ~ H) of CeB 6 (Marcenat et al. 1990).
An evaluation of L(T, H)/Lo for CeB6 was carried out for the temperature region
T < TN, where ~ci,,~ t¢c. It turns out that experimental data on ~ce(T,H) can be well
described (see fig. 54) by the following equation
1% 1 = po(H)/Lo T + [p(H) - po(H)]/O.45Lo T, (42)
where the first term arises due to electron scattering by static defects and the second
term is connected with inelastic electron scattering by magnons (i.e. L/L o = 0.45).
L/Lo does not depend on temperature (between 0.3 and 1.4 K) and on magnetic
field (between 0 and 7 T).
Let us note an interesting feature in the behaviour of Igtot(H ) of CeB 6 not connected
with L(T, H)/L o but rather pertaining to the subject of section 3. In zero magnetic
field tqot does not change at T = T1, but has a minimum at T = TN (figs. 53 and 55).
Two peaks are clearly seen in the temperature dependence of the heat capacity of
CeB6 a large peak at TN and a small one at T1 (fig. 56) (Peysson et al. 1985). A
"weighted" redistribution of these peaks occurs in a magnetic field. At a field H = 8 T
the peak at TN disappears and only one large peak at T~ remains (fig. 57).
The minimum of/qot(T, H) at H = 7 T shifts to T1 (fig. 55).
l I I I I "-I l
H=TT
/~----~f/~ "H=3T x t ° t = oT I- bT 2
1.0
E
rX\,.o,
XX TN \ T~
H=11"2
T~ Fig. 53. The temperature dependence of ~qot for CeB 6
10 -1 single crystals in various magnetic fields H (Marcenat
0 0.5 1,0 1.5 2.0 2.5 3.0 3.5 4.0 et al. 1990). H and VTI[[ll0]. For T > T N lqo t -
T,K aT+ b T 2.
1 T / ~ t ~ ' ~ =7T
~d
1.0
Fig. 54. The temperature dependence of Ko in CeB6 single

crystals (Marcenat et al. 1990) at H - 0 and 7T. H and
VTH [110]. Points give the experimental results, the solid line
o~
gives the calculation using eq. (42) (the same good agreement
10-1 J I I I of experiment and theory has been obtained by Marcenat
0.5 1.0 15 2.0 et al. (1990) for the curves in fig. 53 at H = 1.2 and 3 T). Not
T,K all experimental points of Marcenat et al. (1990) are shown.
I 0°
10-I H=7T /
.
uE 10-2
_~H=OT
10-3
- [ T1
TN
10-4 , ~ I , , I 1"
10° 101 102 Fig. 55. The temperature dependence of/Ctot in CeB 6 (Peysson et al.
T,K 1985) at H = 0 and 7 T.
2
10
101
-6
E
10 °
'.eB6
t
I
I
I
/
I0 -I --
LAB6
/
10-2 , I , i { h i I Fig. 56. The temperature dependence of the heat capacity (C) of
10 q 10 ° 101 10 2 CeB6 (Peysson et al. 1985). Data on C(T) for LaB6 are shown for
T,K comparison•
H:OT H=8T
3.0
"1
2.0
Q.
£9
1.0
0
L I
5
1
10 0 5
I
10 Fig. 57. The temperature dependence of the heat capacity (C) of CeB 6 at
T,K two magnetic fields (Peysson et al. 1985).
5.2.3. Lorentz-number anisotropy in heavy-Jermion systems

O v e r the last few years m e a s u r e m e n t s of the t h e r m a l c o n d u c t i v i t y of the h e a v y -
f e r m i o n s y s t e m s C e C u z Si 2 ( S c h n e i d e r et al. 1983) a n d C e P t 2 Si 2 ( B h a t t a c h a r j e e et al.
1989, R a k i et al. 1990) h a v e b e e n c a r r i e d o u t o n single crystals a n d the a n i s o t r o p y
of L/L o has b e e n a n a l y s e d . U n f o r t u n a t e l y , in these w o r k s ~;L a n d ~;e h a v e n o t b e e n
separated. Therefore, it is only possible to draw general conclusions about the

L(T)/Lo anisotropy:
(1) Anisotropy of the Lorentz number is strongly revealed at low temperatures.
(2) Anisotropy of L(T)/Lo is observed experimentally in C e C u 2Si 2 (L/Lo along
[100] is greater than along [001]) and in CePtzSi.~ (L/L o along [00l] is greater than
along [110]).
(3) Theory agrees with experiment only qualitatively (Bhattacharjee et al. 1989,
Raki et al. 1990).
5.2.4. Lorentz number in dilute Kondo systems

Let us consider Lal_xCe~A12 (Steglich 1976, Moeser and Steglich 1975) and
Lal_ x C e x C u 6 (Bauer et al. 1987) as examples of dilute Kondo systems. The terminal
compounds of the La~ ~Ce~AI2 system are LaAI2, a superconductor (To = 3.3 K),
and CeAI2, a C K L system. Ce at concentrations of x = 0.0099 and 0.0150 is used as
Kondo impurity in LaA12 (Moeser and Steglich 1975). At such concentrations of Ce
impurities superconductivity is not observed. Figure 58 shows the dependence L(T)/
L 0 for Lal _~Ce~AI2 obtained by Moeser and Steglich (1975) taking into account the
contribution of ~ci. to ~tot. As in the case of CKL, the maximum of L(T)/Lo (with
L > Lo) is observed in the low-temperature region. A similar dependence is observed
in Lal x C e x C u 6 , where Ce ( x = 0.03 and 0.1) also is a Kondo impurity (fig. 59)
(Bauer et al. 1987, Onuki et al. 1985a, b). The contribution of ~L to Ktot is taken into
account in the calculation of L(T)/Lo as in the previous case (Bauer et al. 1987). So,
one can draw the general conclusion that the presence of a maximum in L(T)/Lo
(with L > L0) is the same necessary indication of a dilute Kondo system as is the
presence of a minimum in p(T). However, in the case of L(T)/L o one must keep in
mind that the presence of a large amount of non-Kondo impurities in the sample,
along with the Kondo impurities, could give rise to the disappearance of the maxi-
mum in L(T)/Lo (with L > Lo) due to the strong influence of current carrier scattering
by static defects.
I i i
13
1.2 . ~
o Z
1.1 /
..J
1.0 ----
0.9 Fig. 58. The temperature dependence of L/Lo in La 1 ~C%A12 (Moeser
, I I I and Steglich 1975) for various values of the concentration, x: ( r - 52) (1),
0.3 1.0 T, K 10 0.0099 (r = 79) (2), 0.0150 (r = 47) (3), where r = p(300 K)/p(12 K).
21,,,3,, I
O/: 1 I I I /
5 10 100 Fig. 59. The temperature dependence of L/L o of Ce, La 1 xCu 6 (Bauer et al.
T,K 1987) for various x values: 0 (1), 0.03 (2), 0.I (3).
152 I.A. S M I R N O V a n d V.S, O S K O T S K I
5.2.5. Thermal conductivity of systems with an intermediate valence

One could expect nonstandard behaviour of L(T)/Lo in these systems too. Unfortu-
nately, this case has not been discussed theoretically and it is not possible to draw
any conclusions from the analysis of experimental data on thermal conductivity. In
the classical compounds with an intermediate valence of the tanthanide atoms (SmB6,
TmSe) ~cL >> ~co (especially in the low-temperature region). The contribution of ~¢L to
~qot has not been taken into account during investigations of metals from this class
of materials [e.g., CePd3 (Schneider and Wohlleben 1981)].
6. Influence of paramagnetic lanthanide ions on the thermal conductivity of ordered

and disordered systems
As it has been noted in the Introduction, a principal special feature of ions with
partly filled inner shells (d- and f-elements are related here) is the existence of nonzero
spin (S), orbital (L) and total (J) momenta (see table 4). A crystal electric field
removes the degeneration of the orbital momentum orientations and instead of one
energy level in a crystal there is a system of levels. The magnitude of the splitting is
determined by the value of the electric field gradient, the electron shell configuration
and its location inside the ion. In a magnetic field, H, additional splitting of the levels
(the so-called Zeeman splitting) is possible. The energy difference between the split
levels equals AE = g#uH, where g is the Land6 factor and #3 the Bohr magneton.
Lattice vibrations can transfer an ion from one level to another with the absorption
of a phonon. Thus paramagnetic rare earth ions with unfilled inner shells are defects
which reduce the thermal conductivity of a crystal lattice. There are two approaches
for an evaluation of the variation of the thermal conductivity of compounds with
TABLE 4
E l e c t r o n c o n f i g u r a t i o n s , spin (S), o r b i t a l (L) a n d t o t a l ( J ) m o m e n t a of l a n t h a n i d e i o n s R 3 ~, R 4 + a n d R 2~ .
Ion 4f n S L J Ground term
L a 3 . , Ce 4+ 4f ° 0 0 0 a So
Ce 3+ 4f 1 ½ 3 ~ 2F5/2
Pr 3+ 4f 2 i 5 4 3H4
N d 3+ 4f3 3 6 9 4j9/2
pm3 + 4f * 2 6 4 5 J4
Sm3+ 4f5 5 5 5 6H5/2
E u 3+, S m z+ 4f 6 3 3 0 7F o
G d 3+, E u 2+ 4f 7 -~ 0 7 8Sv/2
Tb 3+ 4f ~ 3 3 6 v F6
D y 3+ 4f 9 25- 5 1@ 6H15/2
H o 3+ 4f t° 2 6 8 5J 8
Er3+ 4fll 3 6 wls ~J15/2
T m 3+ 4f 12 1 5 6 3H 6
Y b 3+ 4f a3 ½ 3 ~ ~F7/2
Lu 3 ~, Yb 2 + 4f 1~ 0 0 0 1 So
paramagnetic ions: a resonance approach (Orbach 1960, 1962, McClintok ct al. 1967,
McClintok and Rosenberg 1968, Oskotski and Smirnov 1971, 1972, Oskotski et al.
1972, 1982, Luguev et al. 1975b, Vasil'ev et al. 1984b, Smirnov et al. 1985, 1989) and
a coherent one taking into account the formation of spin phonon excitations in the
crystal spectrum (Elliot and Parkinson 1967, Iolin 1970, Kokshenev 1985). Both
theoretical approaches give roughly the same result - essentially a decrease of the
thermal conductivity of crystals containing paramagnetic ions. The resonance is more
obvious and we will use it in the following.
We will also consider separately the influence of internal crystal field and external
magnetic field on tc of compounds with lanthanide ions.
6.1. Phonon scattering by paramagnetic levels split by a lattice crystal field
Let us consider the two-level scheme (fig. 60) (Smirnov et al. 1989) and a phonon
energy distribution function which is a product of the Planck function and the
phonon density function (phonon spectrum) (fig. 61). A phonon with a resonance
energy he) = A can be absorbed in an intermediate process (fig. 60). Because of that
a narrow band of phonons (shaded in fig. 61) is in practice no longer involved in the
heat transport process, decreasing ~c by (-AGe~) (fig. 62). To absorb a phonon it is
necessary to have A less than OD (i.e. the energy A lies inside the phonon spectrum).
The f-shells are found deeply in atoms, screened by outer shells and their splitting
by a crystal field is small (~100 K), which is just inside the phonon spectra.
Phonons can be scattered as well by the levels of the d-shell split by the crystal
field. But interactions of the crystal field with the d- and f-shells are significantly
different. The d-shells are external and the crystal field acts on them stronger than
the spin orbit interaction. The d-levels in crystals are split according to the following
scheme. The orbital eigenstates formed by a crystal field from states with different
- -Ii _ Fig. 60. The two-level system. A is the splitting by the lattice crystal field, ~ is the
spread of the levels with temperature, ~ oc T - L/2 (Oskotski et al. 1972, Luguev
et al. 1975b).
(~)
/
+--~
L
T
~ ~
koT Fig. 61. The phonon energy distribution function.

154 I.A. S M I R N O V and V.S. O S K O T S K I
(-~ xres)
TmQx
/ i
'7'/ !%',""
i \
/ i \" T Fig. 62. Theoretical temperature

-A~cre~ for the two-level model.
dependence of
projections of the orbital momentum are distant from one another by ~1000 K.
These levels are degenerate according to the orientations of the total spin of the
d-shell. The spin-orbit interaction removes this degeneration and splits the levels by
~100 K (this is inside the phonon spectrum). For this to occur an eigenstate has to
have a nonzero average orbital momentum. So, the paramagnetic levels influence tcL
only in special cases. It must as well be noted that crystals with d-elements are often
ferromagnetics with high Tc and the crystal field does not split the levels.
In f-elements the spin-orbit interaction is essentially stronger than the l a t t i c e
orbit one. This leads to the splitting of states with a total momentum Y = L + S and
to a more simple level structure than in the case of the d-elements. Only ions with
L # 0 and Y # 0 can interact with phonons. This condition is not satisfied for the ions
La 3+, Ce 4+, Lu 3+, Yb 2+ (L = J = 0), EU 3+, Sm 2+ ( J = 0), Gd 3+ and Eu 2+ ( L = 0 )
(see table 4). First studies of the influence of paramagnetic lanthanide ions (PLnIs)
on ~cL were performed at low temperatures on holmium and cerium ethylsulphates
with hexagonal symmetry (McClintock et al. 1967, McClintock and Rosenberg 1968,
M o r t o n and Rosenberg 1962). In this case the paramagnetic levels are split in the
lattice crystal field by some degrees and could be split further by an external magnetic
field (for details see section 6.1.6.6). The typical splitting in cubic crystals is about
100 K, so the effect on ~ci~is essential at medium and high temperatures, in this
section we are mainly considering the experimental data concerning the influence of
PLnIs on ~L in crystals of cubic symmetry.
6.1.1. Temperature dependence o f - A~cr~~

The temperature dependence of - AGes - the reduction of the thermal conductivity
due to resonance phonon scattering due to PLnIs has a resonance form (fig. 62) in
the two-level model (fig. 60) (Oskotski and Smirnov 1971, Oskotski et al. 1972,
Luguev et al. 1975b, Smirnov et al. 1989, Smirnov 1972, Golubkov et al. 1973,
Vasil'ev et al. 1978). The temperature region T > Tin,x, T > O is the most informative
(Luguev et al. 1975b, Smirnov et al. 1989). The temperature dependence and absolute
value of -A~cr~ ~ depends on:
(1) The lanthanide ion concentration.
TABLE 5
The temperature dependence of - AKro~in the high-temperature region on the concentrations and arrange-
ments of the PLnIs in the lattice.
Small concentration High concentration of PLnIs

of PLnIs
Ordered Disordered arrangement of PLnIs in the lattice
arrangement of
PLnIs in the lattice Phonon-phonon Phonon-phonon
scattering is greater than scattering is less than
phonon-impurity one phonon-impurity one
I II III IV
T-2 T-O.5 T-1 T-O.5
(2) The arrangement of lanthanide ions in the crystal lattice.

(3) The relative contribution of the different mechanisms of phonon scattering
(table 5).
6.1.2. M e t h o d s f o r separating - A G e ~
There are two methods for separating - A G e ~ : a theoretical one and an experimen-
tal one. Theoretically - A G , ~ is determined in the Callaway model as the difference
of ~CL(1) and ~Ce(2), where XL(1) is calculated taking into account phonon scattering
by sample boundaries, defects and phonons (N and U processes), and tce(2) is
calculated taking into account the same processes plus resonance scattering of
phonons by the split paramagnetic levels of the lanthanide ions (Oskotski et al. 1982,
Vasil'ev et al. 1978, 1984b, Neelmani and Verma 1972, Arutyunyan et al. 1986, 1987):
- AG~s = teL(2) -- 1eL(1). (43)

Experimentally -- AGes is determined as differences between ~cL of crystals containing
PLnIs and that of crystals with the same concentration of lanthanide ions with zero
L or J. For example:
--AKr~ ~ = K L ( P r S ) - - x L ( L a S ) ,
- A G ~ = ~cL(TbS) - tcL(GdS),
- Axre~ = xe(ErS) - XL(LuS).
All lanthanide elements can be grouped around three reference elements: around
La - - Ce, Pr, Nd (group I); around Gd - - Sm, Eu (group II) and Tb, Dy, Ho
(group III); and around Lu - - Er, Tm, Yb (group IV). The errors in the evaluation
of --AlCre s are small, because the properties of the reference elements and the corre-
sponding compounds do not differ too much, and the differences in their masses and
ionic radii are small. The maximum differences in the atomic mass and ionic radii
are:
Group l AM = M(Nd) - M(La) = 5.34 g,

Ar = r(Nd) - r(La) = -- 0.066 ,~.
G r o u p II A M = M(Sm) - M(Gd) = - 6 . 8 5 g,
Ar = r(Sm) - r(Gd) = 0.026 A.
G r o u p I I I A M = M(Ho) -- M(Gd) = 7.68 g,
Ar = r(Ho) - r(Gd) = - 0.044 A.
G r o u p IV A M = M(Er) -- M(Lu) = - 7.74 g,
Ar = r(Er) - r(Lu) = 0.033 A.
F o r example, for the substitution of Pr by Sc in PrS the values of A M and Ar are

m u c h higher: A M = 95.951 g, Ar = 0.183 ~.
6.1.3. Arrangement of PLnls in a crystal lattice

P L n l s can be arranged in a crystal lattice in different ways (fig. 63). in all cases
the appearance of -Alcros can be expected. Crystals where P L n l s are the m a i n ions,
and not defects, are of special interest. Here, in an ideal lattice without impurities
an additional thermal resistance W2 appears due to p h o n o n scattering by the para-
magnetic levels of the PLnIs.
6.1.4. Choice of materials for investigation

Let us analyse all cases of table 5. F o r this purpose experimental data on ~(T) are
presented for the different groups of materials:
- Small concentrations of P L n I s (these ions are impurities): Y~ ~LnxA15012
(Ln = Gd, Tb, Dy, Er, Tm, Lu) (Oskotski et al. 1972, Vasil'ev et al. 1984a, b,
Smirnov et al. 1985, A r u t y u n y a n et al. 1987, D z h a b b a r o v et al. 1978, Parfen'eva
et al. 1979, Smirinov et al. 1988). These c o m p o u n d s are insulators 0qot = 1eL),
and they have a cubic lattice with the space g r o u p O~°-Ia3d.
Large concentration of P L n I s - they form the matrix of a c o m p o u n d with
ordered arrangement of P L n I s in a lattice:
(a) PrS (LaS is the reference material) (Vasil'ev et al. 1978, Oskotski et al. 1982,
Smirnov et al. 1988, 1989). This is a metal ('~tot = 1£L+ 1£e) with a cubic lattice of
NaCl-type.
(b) PrTe~.33 (LaTe1.33 is the reference material) (Luguev et al. 1975b, Smirnov
et al. 1985, 1988, 1989, Oskotski et al. 1982, Vasil'ev et al. 1976). This is a
degenerated semiconductor (tqot = tcL + tcc) with a cubic lattice of Th 3 P~-type.
Large concentration of P L n I s disordered arrangement of P L n I s in a lattice:
(a) PrTe~.5 (LaTel.s is the reference material) a defect lattice, where the
p h o n o n p h o n o n scattering is stronger than the p h o n o n - d e f e c t one (Luguev
et al. 1975b, 1978, Oskotski et al. 1982, Smirnov et al. 1985, 1988, 1989, Vasil'ev
et al. 1976). These are insulators (/¢tot ~-KL) with cubic structure of Th3P4-type.
(b) Glasses a solid where there is complete disorder. The p h o n o n p h o n o n
scattering is less than the p h o n o n - i m p u r i t y one. Pr2S3(Ga203)2
[ L a a S 3 ( G a 2 0 3 ) 2 as the reference material] (Smirnov et al. 1985, 1988, 1989,
1990, Parfen'eva et al. 1990a); P r P s O 1 4 (LaPsO14 as the reference material)
(Parfen'eva et al. 1990b).
PLn ][ I
l Impurities I Regutar tot tice
A+B+R
--F- ode'ec' I 0efec, 1 Goss,,o.y
IL tattice I - - ~ disordered
t A+ R AfR+ vacancy
l NO regutor
Pr2 [Gttice
R2 R2
WL;1 +W2 -'vacGncy

0
W :I/xL=WI+W 2
Fig. 63. Schematic arrangement of PLnIs in a crystal lattice. A and B are the lattice ions, R represents
lanthanide ions. W1 is the usual defect thermal resistance, and Wz is the thermal resistance due to
resonance phonon scattering on PLnls.
In all these materials the lanthanide ions have a valence of + 3. All the c o m p o u n d s
have a cubic structure. In the investigated temperature range there are no magnetic
transitions.
6.1.5. Schemes of level splitting in a lattice crystal fleld for selected materials
D a t a on level splitting of the g r o u n d states of Pr, Er, Ho, D y ions in a lattice field
of crystals PrS, PrzTe3, P r g T % and garnets have been obtained from n e u t r o n
scattering (Turberfield et al. 1971, Parfen'eva et al. 1990b), optical measurements
(Slack and Oliver 1971), and from an analysis of the Schottky heat capacity (Csoh)
(fig. 64) (Vasil'ev et al. 1983, Mitarov et al. 1975). Figure 64 shows that only a few
z/,K
0~26OK
40E
- - - - - - Cd)
_ _ (d)
. - - (t)
300
(d)
0 "a 200K 0eff ~ 177 K
200
- - ( t ) _ _ (d)
100 _ _ C d ) ..............
0
PrS Pr2Te 3 Pr3Te 4 Y2.8 DY0.2 AI5 O12 Pr2S3 " ( G a 2 0 3 ) 2
Fig. 64. Splitting of the levels in a crystal field for PrS (Vasil'ev et al. 1983, Turberfield et al. 1971);
P r z T e 3 and P%T% (Mitarov et al. 1975);Y2.sDy0.2A15012(Slack and Oliver 1971);and Pr2S3(GazO3)2
(Parfen'eva et al. 1990b). d stands for doublet, t for triplet. Only levels with an energy less than 0 give
contributions to phonon scattering.
low-energy levels in the garnets are capable of scattering phonons. The ions Dy 3+
in Y3A15012, Er 3+ in Er3A15OI2 and H o 3+ in Ho3A15012 have eight levels (Slack
and Oliver 1971).
Difficulties have arisen with regard to the level scheme of the glass Pr2 Sa (Ga2 O3)2.
The average energy position of the broadly spread first excited level has been
determined from an analysis of the Csch data using the model of Bhattacharjee and
Coqblin (1982) for the paramagnetic ions in the amorphous material. This quantity
is approximately equal to 30 K (Smirnov et al. 1990). The energy positions of the
next excited levels in the PrzS3(Ga203) z glass have been successfully determined
from experimental inelastic neutron scattering data (Smirnov et al. 1990). The temper-
ature dependence of the neutron spectrum indicates the existence in the disordered
glass structure of a certain principal type of a nearest-neighbour environment of ions
Pr 3 + (with a statistical distribution) producing a crystal field potential. So on the
basis of Csch and neutron scattering data it has been possible to suggest an approxi-
mate energy level scheme for PrzS3(Ga203)2 glass of 0, 30, 60-72, 96 108 K, which
is shown in fig. 64.
6.1.6. Analysis of experimental thermal conductivity data

It has been necessary to establish from an analysis of experimental data on the
thermal conductivity of many materials with PLnIs the following:
(1) Do the theoretical temperature dependences of -Atq.es(T) for low and high
temperatures ( T > , < O) given in fig. 62 and table 5 agree with experiment?
(2) Is the observed effect of •L decreasing analogous to a Schottky effect in the
heat capacity?
(3) How is Tma x [-in the dependence - A t C r e s = f ( T ) ] connected with A1, the distance
to the first excited level?
(4) How much influence does a defect of the lattice have on the value of -A~c~e~?
Below we consider separately for high and low temperatures, the experimental
data on the thermal conductivity of compounds with PLnIs.
6.1.6.1.Regions of high temperatures. T > Tin.x, T > O.

(a) Lanthanide ions are impurities (Oskotski and Smirnov 1971, Oskotski et al.
1972, 1982, Vasil'ev et al. 1984a, b, Smirnov et al. 1988, 1989, Parfen'eva et al. 1979).
Figure 65 shows the dependence of ~cL on the atomic number of Ln 3÷ in
Y2.8Lno.2A15Ot2 (Ln= Gd, Tb, Dy, Er, Tm, Lu) at 300 K (Smirnov et al. 1988~
1989). The results can be approximated by two straight lines: line A for paramagnetic
ions and line B for nonparamagnetic ions. Similar dependences KL= f ( L n 3+) were
obtained in the temperature range 100-300 K for all investigated PLnI impurities.
Figure 66 shows data on ~L(T) for Y2.sDyo.2A150~2.
Figure 67 shows a comparison of ~cL of magnetic (ErA1G), nonmagnetic (YA1G,
GdA1G, LuA1G) garnets and YAIG with various concentrations of Er impurity (Slack
and Oliver 1971). Data on ~L of YA1G with Er impurity have also appeared in
Arutyunyan et al. (1987, 1984), Petrunin et al. (1989). One can see that ~cL of the
-Axre s
/
"7
"7
E
Fig. 65. Dependence of KL on the atomic number of

Gd Tb Dy Ho Er Tm Yb Lu the impurity ion R for Yz.sRo.zA15Olz at T = 300K
3+
Ln (Smirnov et al. 1988, 1989).
11
B
10
~o
IE 9
Fig. 66. The temperature dependence of KL for

I I I_ I,_ Y2.sDyo.zAIsOlz. Curve A gives the experiment and B
200 250 300 T,K data obtained with the procedure of fig. 65.
I I I
1 ~ 3 2
9.54
% &77
~ --6
0
0
~~ 100
. . .
I
.
200
.
I °-7
300
T,K
Fig. 67. The temperature dependence of the thermal resistance of the crystal lattice, WL= 1/1%, for
R3AtsO12 (R=Er (1), La (2), Gd (3), Y (4)) (Slack and Oliver 1971), and Y3_xErxAlsO12 (x = 0.32 (5),
0.148 (6), 0 (7)) (Oskotski et al. 1972).
crystals with Er impurities b e c o m e s smaller than the tcL of the n o n m a g n e t i c garnets.

From these figures one can see that in garnets with PLnls the value of --A/ere s is
sufficiently large at all investigated temperatures. Figure 68 shows the dependence
--Afire s• = f ( T ) in Y2.s Ro.2AlsO123+ (R 3+ = D y or Tb). It turns out, that (--AK) oc T 2,
as predicted by the theory (see table 5, column I).
(b) Large concentrations of PLnls (ordered arrangement of PLnls in a lattice)
( L u g u e v et al. 1975b, O s k o t s k i et al. 1982, S m i r n o v et al. 1985, 1988, 1989, Vasil'ev
et al. 1976, 1978, Neelmani and Verma 1972, Arutyunyan et al. 1986, 1987, Slack
and Oliver 1971). (In this section experimental data on ~cL in Pr3Te 4 are discussed.
Data for PrS will be described in section 6.1.6.2.)
Figure 69 s h o w s e x p e r i m e n t a l data o n xL(T ) in P r 3 T % , La3Te4 and -A/~re s as
the difference ~cL(Pr3Te4)--KL(La3Te4). It turns out that (-AGes)oCT -°'5 (see
fig. 70), as it is predicted by theory (table 5, column II). Arutyunyan et al. (1987) have
O1
2.0 --
02
5,K T 2
E 1.5 - -
N
2
1.0 --
O
200 300 400 Fig. 68. The temperature dependence of --Alq.es for
T,K Y2.sRo3.~A150,2 . R = D y (1) and Tb (2).
T H E R M A L C O N D U C T I V I T Y OF RECs 16l
...................r ................ 1- ........ -q-1
~ Q L~3re4
1.2
7
.--7 0.8
N
.,q
I
v \ 0
.5
0.4 ¸
Fig. 69. The temperature dependence of

I ) 1 1% for La3Te4, Pr3Te 4 and At<,~
100 200 300 400
(Luguev et al. 1975b, Oskotski et al,
T,K 1982, Smirnov et al. 1988, 1989).
--y-
0,5
~E 0.4
.,~ 0.3
I
0.2 I I _1 L_
100 200 300 400 Fig. 70. The temperature dependence of -AI,-re , for
T,K Pr3Te 4 (Smirnov et al. 1988, 1989).
observed a similar temperature dependence, ( - A ~ e s ) < T -°6, in the temperature

range 200--1000 K in Y3 xErxAlsO12 for x = 0.9 and 1.5, which are sufficiently large
that these samples cannot be considered as small impurity concentrations (Luguev
et al. 1975b).
Unusual temperature dependences (-AtCres)oc T 2 must be noted for EuA1G and
DyA1G in the temperature range 20 300 K (Slack and Oliver 1971) and T 1.4 in
ErA1G (Arutyunyan et al. 1987) in the range 200--1000 K, which do not agree with
the theory of Luguev et al. (1975b). These observations need a special explanation.
(c) High PLnI concentration. Disordered arrangement of PLnIs in a lattice
(phonon -phonon scattering is stronger than the phonon impurity one) (Luguev et al.
162 I.A. S M I R N O V and V.S. OSKOTSKI
1975b, 1978, Oskotski et al. 1982, Smirnov et al. 1985, 1988, 1989, Vasil'ev et al.
1976).
The R/Te 3 phase has a defect lattice with ~ 1021 cm -3 vacant sites (every ninth
R-site is empty). The vacancies are statistically distributed (Golubkov et al. 1973).
Figure 71 shows data on ~cL(T) for Pr2Te3, La2Te3 and -A~r~=tc(Pr2Te3)
-~c(La2Te3). (-A~c~jocT -a (fig. 72), as is predicted by the theory (table 5,
column III).
(d) High PLnI concentration. Disordered arrangement of PLnls in a lattice (pho-
non-phonon scattering is less than the phonon-impurity one) (Smirnov et al. 1985,
1988, 1989, 1990, Parfen'eva et al. 1990a, b). (Data in this section concern both high-
and low-temperature ranges.) These scattering conditions are satisfied in glasses.
Figure 73 gives ~L in the glasses La/S3(GazO3)2, LazSg(Ga203) 3 (group 1) and
Pr2S3(Ga203)2, Pr2S3(Ga203)3 (group 2) and - A t ~ = ~cL (glass with Pr)-tcL
(glass with La) (the difference between the averaged values of the curves in fig. 73 of
the second and first group). Figure 74 shows the temperature dependence of the heat
capacity of the glasses La2S3(Ga203)2 and PrzS3(GazO3) 2 in a plot of C / T 3 versus
In T. The presence of resonant scattering of phonons by the PLnI (Pr 3+) in these
glasses is indicated by the following experimental facts:
I I I I
3
• 2 ~
\ ~ • Pr2Te3
E
- ~ xres "'l .....

i q- - I
100 200 300 400 Fig. 71. The temperature dependence of ~L for La2Te3,
T,K Pr2Te 3 and --A~resres (Smirnov et al. 1988, 1989).
] I
1.2
"7
E 0.8
N
i
\
\
\
0.4 I I _
I00 20O 300 Fig. 72. The temperature dependence of --Axre s for PrzTe 3
T,K (Smirnov et al. 1988, 1989).
r I i ~T F -[ ---[-j
0.8 ---- ,,o ~ o / - -
•~ ! ¥ ~ ~
o.oF /
~- F T.~" ,1 -
7 F]~6o"~ ,~4
2 o.21__o - ' - . Fig. 73. The temperature
s-" I ~..,,... (-~7~res) dependence of KL and -A~c,o,
of the glasses R2S3(Ga203)2
F- (R=La (1), Pr (3)) and
o|1 3 ~
5 10
I I "~'--I
30 50 100
I.
300
RzS3(GazO3) 3 (R=La (2),
Pr (4)). (Smirnov et al. 1988,
T,K 1989),
10-2
5
E
10-3
£9
10 -4 I I t_ I I I __ Fig. 74. The temperature dependence of C / T 3 for the

3 5 10 20 30 50 lO0 glasses PrzS3(Ga203) z (1) and La2S3(GazO3) 2 (2)
T,K (Smirnov et al. 1989).
(1) In P r 2 S 3 ( G a 2 0 3 ) 2 a Schottky contribution to the heat capacity is observed

(fig. 74).
(2) A considerable variation of glass composition (from R z S 3 ( G a z O 3 ) / to
R2S3(Ga203)3 does not have m u c h influence on the ~cL value. But tel of glasses with
Pr and La differ greatly.
(3) The dependence (-A~cre~(T))oc T -°'s2 (fig. 75) agrees well with the prediction
of the theory (see fig. 62 and table 5, column IV).
The 1eL(T) of glasses of the intermediate compositions (PrzS3)y(LazS3)l y
( G a 2 0 3 ) / ( y = 0.1, 0.3 and 0.5) has been measured by Smirnov et al. (1992a) (fig. 76).
---Q~x----r-- ~ i 1
0.1
"T
"7"
E
-q
0.0.5
1 L I L I
20 30 40 50 100 Fig. 75. The temperature dependence of -AK,.~ for
T,K Pr2S3(GazO3) 2 (Smirnov et al. 1988, 1989).
-'--1---'I I I I I
0.8 --
I / 1/ o 6~ o
o¢#.~
0.6
i1 ~%~///t"
E I 2~ / 1I
_1
0.';
o,t.i + 1
o"," l • 2
0.2 o 3
Fig. 76. The temperature dependence of t<L for
(PrzS3)y(La2Sa)l_y(Ga203) 2 (Smirnov et al. 1992a) for the
1 I I I I I I
3 4 5 ~0 152{ values y = 0.1 (1), 0.3 (2), 0.5 (3). Curve I is for Laz $3(Ga203)2,
T,K curve II for P r z S 3 ( G a / O 3 ) a.
The glass (Gd2S3)o.3(La2S3)o.7(Ga2S3) 2 has been used as a reference material,

because in this glass there is no resonance phonon scattering from the paramagnetic
levels of Gd 3 + ions with L = 0.
Figure 77 shows data on the heat capacity of (Pr 283)o.3(La2S3)o.7(Ga203)2. An
additional contribution to the heat capacity is observed in the Pr glass in comparison
with the La glass. There is good agreement of experimental data (points 1 in fig. 77)
with the sum of the partial contributions of the experimental values of C of the
glasses Pr2S3(Ga203) 2 and La2S3(Ga203)2, i.e. 0.3C[Pr2S3(Ga203)2] 4-
0.7C[LazS3(GdzO3)2] (the solid curve 2 in fig. 77). This proves that the observed
additional contribution to the heat capacity, AC, is due to the Schottky-effect, i.e.
AC = Csch.
(-AKres) has been determined for this glass with x = 0.3 using the relation
(--A/Ores) = t C L l - ( P r 2 8 3 ) o . 3 ( L a 2 8 3 ) o . 7 ( G a 2 0 3 ) 2 ] - 1¢o . (44)
The value of ~:o (fig. 78) is obtained by correcting the xL reference curve of
(Gd2 $3)o.3 (La2 $3)o.7 (Ga2 O3)2 by taking into account the dependence of the thermal
lO
5
-~1.o
d
0.5
Fig. 77. The temperature dependence of the heat capac-

0.1 ity for (Pr/S3)0,3(La2Sa)o.7(Ga203) 2 (Smirnov et al.
1992a). Points (l) give experimental results, the
£ solid line (2) is calculated from the formula
0.3C[Pr2S3(Ga203)2] + 0.7C[La2 $3(Ga203)2].
I I I IIIII IIIII ~
3 5 10 20 Curve I represents heat capacity of La2S3(Ga203)2,
T,K II the heat capacity of Pr2Sa(Ga203) 2.
I I I I I I I l l II
d
0.7
F 0.5
# Fig. 78. The temperature dependence of t% tor

0.3 (Pr2S3)0.3(LazS3)o.7(Ga203) 2 (1) and (Gd2S3)o. 3
(La2S3)o.v(Ga203) 2 (2) (Smiruov et al. 1992a).
The solid line (I) represents values of ~co (the correct-
ion of the reference curve for (Gd2S3)o.3(La2S3)o. 7-
0.1 (Ga203) 2 taking into account the dependence of the
II I I I I I I i-
3 45 m ~520 304o5o lOO lattice thermal conductivity on the mass and the ionic
T,K radius of the lanthanide ion).
conductivity of the mass and the ionic radius of the lanthanide ion (Smirnov et al.
1991a). The resulting (--A~Cr~s)ocT -°'5 (for T > Tm~x in the range 5-50 K) agrees well
with theory (table 5, column IV).
Resonance scattering of phonons by Pr ions has also been observed in the phos-
phate-glass PrP 5 O14, during an investigation of its thermal conductivity (LAP5 O14
has been used as the reference material) (fig. 79) (Parfen'eva et al. 1990b). The thermal
conductivity of phosphate glasses is much smaller than that of R2 S3(Ga2 03)2 and
has nearly the same value and temperature dependence as the thermal conductivity
of quartz glass (Zeller and Pohl 1971).
"--'T--FT-- ~ - T q - - ' - F - T q - - - - - - - F
J
0.3
0.2
/
0.1
I I I I I I I I I I Fig. 79. The temperature dependence of ~q, for the
4 5 10 15 20 30 4050 100 glasses LaPsO1, and PrPsO14 (Parfen'eva et al.
T,K 1990b).
A Schottky contribution to the heat capacity of PrPsO14 has also been observed
(fig. 80).
The temperature dependence ~L(T) and C(T) for LaPsO14 show that phonon
scattering by PLnIs is observed in the temperature range where their thermal proper-
ties could be explained by using a two-level model (Karpov and Parshin 1985).
6.1.6.2. Influence of PLnls on Ice. Low temperatures (T <<,Tmax, T < O).

(a) PLnIs are impurities (Vasil'ev et al. 1984a, b, Arutyunyan et al. 1987).
Analysis of the experimental data on ~ce of solid solutions of garnets
Y2.sR0.2A15 Or2 (R = Gd, Dy, Er, Tin, Lu) has been carried out using the Callaway
5,10 2
10 PrPs014~ "2
5~ 5.0 ~ _2
05
-
Sc Z
/~T
16
"
///
Fig. 80. The temperature dependence of the heat capacity

2 3 5 10 50 100 300 of the glasses LaPsOI¢ and PrPsO14 (Parfen'eva et al.
T,K 1990b).
model (Callaway 1959) in the temperature range 2-100 K (Vasil'ev et al. 1984a, b).
For nonmagnetic solid solutions with R = Gd, Lu scattering of phonons by bound-
aries, point defects, phonons (N and U processes) are taken into account. The results
of the calculations are shown in fig. 81 (solid lines come from theory, points show
experimental data). It is noted that all three curves for garnets with Gd and Lu
impurities are calculated by the variation of only one parameter of the phonon
defect scattering, taking into account the concentration and the mass of the sub-
stituted atoms. The other parameters for the curves 1-3 in fig. 81 are identical.
Similar calculations have been carried out for Y1 xRxA15012 (x = 0.2) with impuri-
ties PLnIs R = Er, Dy, Tm (Vasil'ev et al. 1984a, b). The scattering of phonons by
paramagnetic ions has been accounted for in these cases together with a consideration
of the above scattering mechanisms. The results for Yz.sDyo.2A15 O12 are shown in
fig. 82. An analogous analysis of the experimental data for magnetic garnets and
garnets with paramagnetic ions has been carried out by Neelmani and Verma (1972)
and Arutyunyan et al. (1986, 1987).
Figure 83 shows the results of a comparison of the values of -AGes for
Y2.sDy0./A15012 obtained by calculations (for T < 100 K) and from experimental
data (for T > 100 K). Curve 1 in fig. 83 presents the differences between the experimen-
tal values ~cL (points in fig. 82) and values calculated as described above without
taking into account the phonon scattering by paramagnetic ions (curve l in fig. 82).
Curve 2 in fig. 83 presents the experimental data shown in fig. 68 (points 1). As one
can see, the coincidence of the two (-AGes) curves is good in actual magnitude and
also for the temperature dependences in the low- and high-temperature ranges.
(b) A high concentration of PLnIs (ordered arrangement of PLnls in a lattice)
(Smirnov et al. 1988, 1989, Vasil'ev et al. 1978).
Figure 84 shows data on tcc of LaS and PrS and --A/ere s = ~cr(PrS)-tce(LaS).
LaS and PrS are metals. For them /(tot K'L-]-Ke, where ~cL is calculated from the
:
Wiedemann Franz law, and the Lorentz number is determined by eq. (4) (Oskotski
and Smirnov 1972, Smirnov and Tamarchenko 1977, Wilson 1965). The influence of
300 i i I I
200
tO0
Fig. 81. The temperature dependence of KL for the solid alloys

I I I I I Y3_xRxA15012. R = G d , x = 0 . 2 (1), R = L u , x = 0 . 1 (2), x = 0 . 2 (3).
5 10 3 0 50 100 Solid lines are calculated using the Callaway model, the points are
T,K experimental data (Vasil'ev et al. 1984a, b).
250 1 7---F'--T--I--~
200
I 2 !
°
150
E
g
100
\
I / ,,',
50
3 ~/ '~"k
' Fig. 82. The temperature dependence of ~h. for
Yz.sDyo.2AlsO12 (Vasil'ev et al. 1984a, b). Curve l: a
I "-% calculation without resonance, 2: a calculation with
I I 1"_ I I I
3 5 10 30 50 100 resonance, 3: --A~Cres as difference between 1 and 2,
T,K 4: experimental points.
I O0 1- 1
30
E
)--lo
.q
i
Fig. 83. The temperature dependence of --A,'Cres

for Y2.sDy0.2A15012. Curve 1 is the low-temper-
ature part, -AG~., = ~L (the experimental curve 4
........ I I from fig. 82) - ice (the calculated curve 1 of fig. 82),
10 30 100 300 curve 2 is the high-temperature part (T> 100 K),
T,K --AKre s - - t h e experimental data from fig. 68.
the crystal field effect on tee has not been taken into account in the separation of the
lattice thermal conductivity from tetot of PrS and Pr3Te4 (see section 6.2). Evaluations
have shown that taking this effect into a c c o u n t hardly changes the value of -eXte,.es
and does not affect the main conclusions. A theoretical analysis of teL of PrS in the
Callaway model, with resonant p h o n o n scattering by paramagnetic Pr 3 ÷ taken into
account, was carried out by Vasil'ev et al. (1978) (fig. 85). The solid line in fig. 85 is
from experiment, and the dotted line a calculation using the equations of Oskotski
et al. (1972), Luguev et al. (1975b) and Vasit'ev et al. (1978). The theoretical and
experimental curves are matched at the point Tmax. One can see that the experimental
r I V-
30
LaS
'7
"7 PrS
E 20 Trnax
.-g
,..h. ,
N
J
10 f ~" --....(~{~4X res )
Fig. 84. The temperature

OI I I I dependence of Kn for LaS, PrS
0 100 200 300 and AG~~ (Vasil'ev et al.
T, K 1978).
1 I
20 -- t
I
10
Fig. 85. The temperature dependence

I I of --A/ere s for PrS (Vasil'ev et al.
0 100 200 300
1978). Curve 1 experiment, 2 theor-
T,K etical calculation.
a n d theoretical t e m p e r a t u r e d e p e n d e n c e s of - A G e ~ agree well at T < Tmax a n d

T> T,nax.
O n the basis of the analysis of the e x p e r i m e n t a l d a t a on ~CL(T) and -AKre~(T ) of
a large g r o u p of l a n t h a n i d e m a t e r i a l s one m a y c o n c l u d e that t h e o r y a n d e x p e r i m e n t
agree sufficiently well at b o t h low ( T < Tmax, T<<.O) a n d high ( T > Tm,~, T > O)
temperatures.
6.1.6.3. Is the effect of resonant phonon scattering by paramagnetic impurities on ~CL

an analogue of the Schottky effect in the heat capacity? Let us compare the temper-
ature dependences of Csch and -AGo,, in crystals (PrS) and glasses (Pr 2 S 3(Ga2 O3)2).
(a) Crystals. Figure 86 shows values of C of PrS and LaS and Csch =
C(PrS)- C(LaS) (Vasil'ev et al. 1983). The experimental data Cs~h(T) agree with the
theoretical results using the level splitting of the ground state of Pr 3 + in PrS obtained
in neutron studies (fig. 87) (Turberfield et al. 1971). It follows from fig. 87 that:
(1) The shapes of temperature dependences of Csch and - A G e s for high and low
temperatures are similar.
(2) The temperatures of the maxima of Cs~h and - AGes coincide.
(3) The effect of -AKro~ and the Schottky-effect differ in sign: t¢L decreases, whereas
C increases (fig. 88).
So, one can draw the conclusion that the effect of resonant phonon scattering by
50 I I
PrS
40
30
"7
E
d 20
L~
10 \
lI \
x \ ~/C Sch
!
I
Fig. 86. The temperature depen-

dence of the heat capacity of PrS
and LaS, and the Schottky contri-
100 200 bution Csch (Vasil'ev et al. 1983).
T,K
i I I
10 [-5 -
5
E
--3 1
! "%...
J
U
[ I- Theory
~ - Exper.
I I
20 _
b I
I
I 1
!
10
Fig. 87. The temperature depen-

I
dence (a) of Csch (Vasil'ev et al.
1983) and (b) of -Atc,e~ (Smirnov
et al. 1988, 1989) for PrS.
o I I (c) Crystal field levels of Pr ~+ in
0 100 200 300 PrS obtained from neutron data
T,K (Turberfield et al. 1971).
CSch
Tmax
y<8 # ~ . T>O
//'L ~' I
I Fig. 88. A schematic representation of the temperature

0
Xres dependences of Csch and --AKre~ for a two-level system.
172 I.A. S M I R N O V and V.S. OSKOTSKI
levels of the paramagnetic ions split in a crystal field is an analogue of the Schottky
effect in the heat capacity.
(b) Glasses. The above-stated conclusion is also confirmed by measurements of
Cs~h and --AKre s in the glass P r 2 S a ( G a 2 0 3 ) 2. However, in glasses there are some
peculiarities (Smirnov et al. 1992a). The maximum temperatures of Csch and -A~cre ~
do not coincide (fig. 89) and the value A T = Tm,~(CsCh)- Tin,x(-- A~c~e~)increases with
increasing broadening of the split levels, Yo (fig. 90). When 7o = 0 (no broadening,
crystal case) A T--+ 0. In the Pr2 S3(Ga2 $3)2 glass where A T_~ 10 K, this corresponds
to Yo -~ 0.5.
--~F-F-I I I I I I ~-
I
¢1.
0.4
x::
~Z I
u
0.2- I
I
I
0.2 -- ~ ~
x:
E
0.1
i b.
Fig. 89. The temperature dependence (a) of Csch and (b) of
I II __1 I I I II
3 5 10 20 50 --A~Jres for the glass PrzS3(Ga203) 2 (Smirnov et al.
T,K 1992a).
I t I I I I I-I~F I n I-r-"
\ C. -
20
E
15
I0
5/
\o I I ....... I-
2° F
5.0
x k 0 0. o.8 .2
Fig. 90. (a) The theoretical dependence of the position of

the maximum of Cs~h and -A~crc ~ in the glass
I I I I I I I I I I I I I I l
0.5 1.0 1.5 Pr2 Sa(Ga2 03)2 on the broadening of the lowest split levels,
7o Y (Smirnov et al. 1992a). (b) The dependence of AT on y.
6.1.6.4. Connection of Tm~, in -A~c~os ( T ) with A 1, the separation of the .first excited
level from the ground state. Figure 91 shows data on Tm,x= f ( T ) for a number of
materials with PLnls. It turns out that T ~ ~-~A~/k. So, it is possible to exclude
from the heat transfer (thermal conductivity)of a solid that or another phonon group
(cf. fig. 61), and to construct a material with certain thermal properties by selection
of PLnIs.
6.1.6.5. Influence of lattice dejects on the -At¢res value.

(a) Defects in solids. Let us consider as a model defect system Pr2 Te3-Pr3 Te4 (and
as a reference system LazTe3-La3Te4). In these systems:
(1) A continuous transition takes place from a defect free lattice Ln3Te~ to a defect
one [the defects (vacancies) are in the cation subsystem]. In LnzTe 3 the vacancy
concentration reaches ~ 1 0 2 1 c m - 3 (Golubkov et al. 1973).
(2) The same type of crystal structure (Th3P4) is kept through the entire system:
the lattice constant changes insignificantly (from 9.485 A in PraTe4 to 9.482 • in
PrzTe3) (Golubkov et al. 1973).
(3) The energy of the crystal field levels of the ground state of Pr 3 + ions varies
insignificantly through the system (cf. fig. 64) (Mitarov et al. 1975).
Let us consider in more detail the experimental data on -Alcre.~ of the system
PraTe3 Pr3Te4 obtained for T > Tmax, T > O (these have been partly discussed
earlier). Let us analyse data on tcL of this system at 100 K (Luguev et al. 1975b, 1978,
Oskotski et al. 1982, Smirnov et al. 1988, 1989, Vasil'ev et al. 1976). A summary of
the data is shown schematically in fig. 92 (Smirnov et al. 1988, 1989). It turns out
that - m / g r c s of the heavily defect lattice Pr2Te3 significantly exceeds -A~c~ of the
defect-free lattice Pr3Te 4 (fig. 92). The -A~C~e~ value increases continuously with an
increase of the number of defects in the lattice. This effect can be explained with the
help of the diagrams in figs. 92b, c, which are analogous to fig. 61.
In Pr3Te 4 (figs. 92b, b') all Pr ions (let us designate them Pr2) have the same
environment. Phonons with an energy At do not take part in the heat transport
(fig. 92b'). The ions Prl and Pr2 are nonequivalent in Pr2Te3 (fig. 92c, c'), since Prl
in contrast to Pr2 has a vacancy as a neighbour (cf. scheme of the crystal lattice in
I I I I [ I I
80
6O
E 40
FI Y2.8Ero.2 AI5012 J
20
Fig. 91. Dependence of T,I,~ on A 1
(energy of the first crystal field level)
0 40 80 120 for a number of investigated mater-
z11,K ials (Smirnov et al. 1988, 1989).
el.
S~E. 2 ~ (-A Xres)
2 PrTey
I, , ?°°Kl
1.33 y-~ 1.5
0 vacancy-lP" i021cm
-3
¢1.' OJ
N(~)
b.
Pr 2 I
N(~) I
Fig. 92. A schematic presentation of the

influence of crystal lattice defects on
--Al¢rc~ in the system PrzTe3 Pr3T%.
Pr2 Prl AI AI' Details are explained in the text.
fig. 92a'). They are in different crystal fields and as a result there are two types of
levels, Aa and A'~ (fig. 92c). This leads to "spreading" of the resonance band and to
more phonons not taking part in heat transport (fig. 92c').
(b) Glasses. As has been noted above, in glasses, where there is short-range order,
but no long-range order, a more significant decrease of ~cL due to resonance phonon
scattering by PLnIs than in Pr2 Te3-Pr3 Te4 is observed. Short-range order in glasses
will split the levels of Pr (Smirnov et al. 1990). Due to the absence of long-range
order in glasses the first excited level, in contrast to crystals, is separated from the
ground state by a value which varies from ion to ion according to the Gauss law
(Bhattacharjee and Coqblin 1982)
p(A) = ~
1
exp
[ (A-Aoa)2~
~- j, (45)
where Aot is the average position of the excited level and A the halfwidth of the
T H E R M A L C O N D U C T I V I T Y O F RECs 175
distribution function (fig. 93). The large effect of resonant phonon scattering by PLnIs
in glasses can be explained qualitatively using the scheme of levels shown in figs.
94a, b (as has been done for the system PrzTe3-Pr3Te~). A strong spreading of the
first excited level is a consequence of a system of close lying levels (A1, A'~, Ai' and
so on) of different Pr ions with different environments (fig. 94a). This can lead to a
broader resonance band than in the Pr2Te3-Pr3Te4 system, and as a result more
phonons not taking part in heat transport (see fig. 94b).
6.1.6.6. Phonon scattering by paramagnetic levels split by a magnetic field. As noted

at the beginning of this section, an additional splitting of levels can occur in a
magnetic field (the so-called Zeeman splitting). This allows one to vary the resonant
phonon energy and this leads to a contribution of certain phonon groups to the
thermal conductivity with the help of a magnetic field. The results on the influence
of a magnetic field at low temperatures on ~g of materials with PLnIs are now
classical and have been described in reviews and books (Oskotski and Smirnov 1972,
Berman 1976). This effect was observed for the first time in holmium ethylsulphate,
I\
A ~
#/ /101
o ~,
Fig. 93. Positions of the first excited level in a glass (dashed line).
-_--2S
N (el
If'' w-s
/1\\
A1 AI' AI" '41'"
Fig. 94. Scheme illustrating the influence of
phonon scattering by PLnIs in a glass.
H o ( C 2 H 5 S O 4 ) . 9 H 2 0 (Morton and Rosenberg 1962, McClintok et al. 1967), and

then in cerium ethylsulphate (McClintok and Rosenberg 1968). To date ~L(H) has
been measured in a large number of materials with paramagnetic ions.
Let us consider briefly the results on ~ce(H) in hohnium ethylsulphate (Morton
and Rosenberg 1962, McClintok et al. 1967). Holmium ethylsulphate is an insulator
with hexagonal symmetry; according to spectroscopic data the lowest energy level is
a doublet• A singlet lies higher by 8.6 K. The g-factor is large (g = 15.4 for ground
doublet splitting by a magnetic field along the hexagonal axis). This allows a large
splitting in moderate magnetic fields• The magnetic field splits the ground doublet
as shown in fig. 95. Phonons can cause transitions, A, between the split doublet
levels, and, B1 and B2, between the doublet and singlet levels. At H = 0 a transition
Do is possible. The energy of resonant phonons for transition A is he) = g#BH, for
transitions B1 and B2 this energy is he)= D O+_½g#nH. Bands with finite widths
appear in the phonon spectrum at frequencies corresponding to the transitions A,
B1 and B2, giving decreased contributions to the thermal conductivity due to resonant
scattering (figs. 95c, d, e), i.e. the situation is analogous to that discussed above.
Figure 95a shows the dependence of the thermal conductivity of holmium ethylsul-
I I I
2 a.
15 ~~ 3.02K
E
10
u
I I I I
B1 20
H,kOe
40
N(E)
A B
A,B
Fig. 95. Properties of holmium ethylsulphate.(a) The magnetic

field dependence of the thermal resistivity. (b) Dependence of
the energylevelson magnetic field.(c, d, e) Schemesof contribu-
B A tions of different parts of the phonon spectrum to 1%(maxima
of curves are near hco~ k0T).
phate on a magnetic field H. At low temperatures ( T < 8.6 K, when k0 T-~ ½D0) in
weak H there are two clear peaks in the dependence WL(H). Let us try to explain
this effect qualitatively using the scheme in figs. 95c, d, e. First we divide the curve
WI.(H) (fig. 95a) into four parts: 1 2, 2 3, 3 4, and 4-5. For weak magnetic fields
the band A lies in the range of small frequencies, the bands B 1 and B2 lie on the
right side of the phonon distribution function, N(e) (fig. 95c). This corresponds to a
position of WL(H) near point 1 in fig. 95a. With increasing H the band A shifts to
the right and the band B1 shifts to the left (fig. 95c). This leads to an increase in the
number of phonons with a small contribution to ~cL. At the same time the band B2
shifts to the right and its contribution to 1eL is insignificant. So, ~CL decreases (WL
increases). This decrease continues until the moment when bands A and B1 coincide
(fig. 95d). This process corresponds to the part 1-2 of the WL(H) dependence (fig. 95a).
If the intensities of resonant scattering of phonons in the bands A and B1 are high,
their superposition (fig. 95d) leads to a decrease in the sum of the intensities of
scattering and WL(H) decreases (part 2-3 in fig. 95a). With a further increase in H,
the bands A and B1 are separating from each other (fig. 95e). This leads at first to a
sharp increase of WL(H ) (part 3-4 in fig. 95a) and then to a smooth decrease of
WL(H) (part 4 5 in fig. 95a). At greater values of/-/ the resonant bands shift to a
frequency range where there are small number of phonons, so WL(H) ceases to
depend on H. At the same time the transitions between levels which have occurred
without a magnetic field, disappear and WL(H) at high H is less than a t / 4 = 0.
6.2. Influence of the crystalJield effect on the electron part of the thermal conductivity
Up to now we have discussed the influence of the crystal field effect on ~cL (due to
phonon resonance scattering by the rare earth ions) and the heat capacity (Schottky
effect). It is known that the crystal field can significantly influence the behaviour of
the elastic constants (Lfithi et al. 1973), the thermal expansion coefficient (Ott and
Lfithi 1976), the magnetostriction and thermoemf (Sierro et al. 1975), the electrical
conductivity (Lfithi et al. 1973, Friederich and Fert 1974, Andersen et al. 1974) and
the electron part of the thermal conductivity ~ce (Smirnov and Tamarchenko 1977,
Wong 1978, Matz et al. 1982, Mfiller et al. 1982).
Let us consider very briefly the influence of a crystal field on G. (In a somewhat
different way this phenomenon has been considered in section 3.1 when discussing
spin phonon scattering and its influence on G in the paramagnetic region.) A theory
of this effect has been elaborated on by Wong (1978), Fulde and Peschel (1972),
Bhattacharjee and Coqblin (1988) and was experimentally verified with RSb (R =
La, Er, Tm) by Wong (1978), with PrCus by Wong (1978) and Matz et al. (1982),
with PrPt5 by Wong (1978), with LaA13 and PrA13 by Mfiller et al. (1982) and with
(La, Pr)Sn by Sfitsch and Umlauf (1981). It has been supposed by Wong (1978), Matz
et al. (1982) and M/iller et al. (1982) that at low temperatures in the investigated
alloys ~cL ~ ~ce and therefore ~qot= •e. (The question whether such a supposition and
method of separation of tcL and ~ce is right, has been discussed in section 3.1.) For
these temperatures
I/V~= I/rG= Wo + W~= A / T + BT 2 or ~G/T=(A + BT3) ~ (46)
(Smirnov and Tamarchenko 1977). Here A = po/L, Po is the residual electrical resis-
tance and W~ the ideal thermoresistance due to electron scattering by lattice
vibrations.
Figure 96 shows, as an example, the experimental data for PrPt5 with the effect
of the crystal lattice field, and as a reference material LaPt5 without such effect (it
turns out that for LaPt 5 the second term in eq. (46) is insignificant). Values ~ce for
PrPts, calculated with the theoretical formulae of Wong (1978) taking into account
the crystal lattice field effect by using the experimental data of the energy level
splittings shown in fig. 96b, are presented as well. tc~ decreases appreciably due to
the crystal field effect. A good agreement of experiment and theory is noted.
Figure 97 shows data on Lcf/Lo = f ( T ) in PrAt3. Lcf is the so-called Lorentz
number of the crystal field effect. It is determined by the equation
LCf =/gcf/Wcf T~ (47)
where p~f and l / V c f a r e the "magnetic" parts of the electrical and thermal electronic
resistance which depend upon the crystal field effect, p~f and l/V~fare determined as
differences of p and We in PrA13 and LaA13:
per = p(PrA13) - p(LaA13), (48)
W~f = We(PrA13)- We(LaA13). (49)
At low temperatures Lcf < Lo, which points out the inelastic character of the current
carrier scattering at F1-F6 transfer. With increasing temperature the doublet F6
becomes occupied (fig. 97b) and this, probably, results in the elastic character of the
scattering at higher temperatures. The fact that L~f exceeds L o for T > 20 K could
be due to the neglect of the ~cL contribution to the thermal conductivity.
I I
a. ' --v-r~
9 oO o
I
L
I
E I
u
E
~z
5 C4
Fig. 96. (a) The temperature dependence of KL/T for
Y b.
3 I I I
LaPt s and PrPt s . Solid line is a theoretical calculation of
5 10 15 /re for PrPt 5 taking into account the crystal lattice field.
T,K (b) The energy level splitting in PrPts (Wong 1978).
1.4 1 | 1 I i I I a.
2
1.0 .........
~-L;---"
u_ 0.6
o
_1
0.2
/
I I 1 I I I
Fig. 97. (a) The temperature dependence of the ratio Lcf/L o for
20 40 60 80 PrA13. (b) The energy level splitting in PrA13 (Mfiller et al.
T,K 1982).
7. Thermal conductivity of lanthanide compounds with structural phase transitions

initiated by the cooperative Jahn-Teller effect (CJTE)
According to the Jahn Teller theorem (Jahn and Teller 1937) any structure (besides
linear) with an electron ground state degenerate in the orbital quantum number is
unstable and tends to reduce the symmetry, to split the ground state and to remove
the degeneracy. The Jahn-Teller effect manifests itself as a phase transition of first
or second order at a transition temperature TjT (we are using ~T instead of the usual
T~, since we are using T~ for the superconduction transition temperature). The J a h n -
Teller effect has been observed experimentally in solids on defects and as cooperative
effect on basic atoms. The C J T E is investigated using different methods: resonance,
electronic, optical, acoustic, magnetic, thermal (Sturge 1967, Gehring and Gehring
1975). It manifests itself distinctly in the temperature dependence of the thermal
conductivity. The most convenient materials for investigation of the CJTE are
compounds with lanthanide elements. In lanthanide paramagnetic ions, because of
the strong localization of f-shell inside the ion and its weak interaction with the
crystal lattice (see section 6), the Jahn-Teller splitting of levels (2W in fig. 98a) turns
out to be small and TjT equals a few degrees [-in the case of d-elements this interaction
is much stronger, and Tn- can be of the order of 1000 K; in M g M n 2 O a , e.g.,
TjT --~ t000 K (Sturge 1967, Gehring and Gehring 1975)]. At these temperatures m a n y
scattering mechanisms are insignificant, which considerably simplifies the investiga-
tion of the influence of a CJTE.
II ~I 11
1.O
10 -2
7 I ca.
Fig. 98. (a) Energy levels in the tetragonat and orthorhombic
phase of TmVO4. (b) The temperature dependence of 1eLfor
single crystals of TmVO4 (Daudin and Salce, 1982). Curve 1
gives the calculation according to Callaway (1959) taking into
account the various scatterings: (I) phonons by phonons (U
and N processes), crystal boundaries, point defects, disloca-
DFt horhombic Tetrogonal tions and (II) resonance scattering by split levels of the ground
state and first excited level. Curve 2 gives the calculation
,I ll! ii il according to Callaway (1959) taking into account scattering I
10-2 10 -1 1.O 10 10 2
but not scattering II. Curve 3 gives the experimental values
T,K of KLfrom Daudin and Salce (1982).
The CJTE is observed in a great number of lanthanide compounds: zircons RXO4

(R = Tm, Tb, Dy; X = V, P, As), intermetallic compounds (TmGd, TmZn, CeAg and
others), compounds with lanthanide impurities (LaSb : Dy) DySb, PrA10 3 and others
(Sturge 1967, Gehring and Gehring 1975, Ray et al. 1981, Bloch and Davidov 1982,
Harley et al. 1973). Experimentally the influence of the CJTE on KL has been
investigated, mainly, in the insulators RXO4. The structural transition from tetrago-
nal to orthorhombic symmetry occurs in these crystals at decreasing temperature to
TjT (to a few Kelvins) and the two-fold degenerate ground state splits in two (see
fig. 98a). The first measurements of ~c(T) in TmVO4 and DyVO4 have been carried
out by Parsons and Rosenberg (1972). Then K(T) of these crystals and also of TbVO4,
HoPO4 and TmAsO4 has been measured by Parsons (1976), Daudin et al. (1980,
1981a, b, 1984) and Daudin and Salce (1982). These authors and Parsons and Rosen-
berg (1972) have discussed the influence of a magnetic field on the thermal conductiv-
ity of these crystals. Published data on K(T, H) have been analysed theoretically by
Daudin et al. (1984) and Mutscheller and Wagner (1984a, b, 1986, 1987).
The classical subject of CJTE studies is TmVO4. It has no additional phase
transitions besides the CJTE (TjT = 2.15 K). Figure 98b shows experimental data on
ic (T) in TmVO4 (Daudin and Salce 1982). Since TmVO4 is an insulator, the
experimentally measured l£totequals 1eL.Figure 98a shows the energy levels of TmVO4
in the high-temperature (tetragonal symmetry) and in the low-temperature (ortho-
rhombic symmetry) phase (Cooke et al. 1972). At T < TjT the doubly degenerate
ground state splits into two levels with 2W-~ 3 K (Cooke et al. 1972). The distance
to the next excited level is practically the same in both phases and equals ~ 50 K
(Knoll 1971). At T = TjT a sharp change in the value and of the temperature depen-
dence of tel of TmVO4 is observed (fig. 98b). Near TjT (for T < TjT) tCLoCT 5"5-+°5
(fig. 99). Figure 98a (Daudin and Salce 1982) shows the results of a theoretical analysis
of KL(T) for TmVO4 using the Callaway model (Callaway 1959) taking two groups
of phonon scatterers into account: the first one is the scattering of phonons by
phonons (U and N processes), crystal boundaries, point defects and dislocations; and
the second one the resonance phonon scattering by the split levels of the ground
state and by the first excited level. There is a direct analogy with the case of phonon
scattering by the paramagnetic levels of lanthanide ions split by the lattice crystal
field (see section 6).
Comparing experimental data and theoretical curves of ~cL one can draw the
following conclusions:
5,_
° o
1.0
E • TjT
;%", I/~/
1 0 -1 ",.f'"'-I [
Fig. 99. The temperaturedependenceof KL/T 3 for a single crystal of
1 10 TmVO4 (Daudin and Salce 1982). Dashed lines show calculated
T,K values of 1%for boundary scattering (the so-called Casimir limit).
(1) At T ~ TjT strong scattering of phonons by the split ground state occurs. With
decreasing temperature this scattering at first increases in the narrow temperature
region corresponding in fig. 98a to the region of the transition from the unsplit
doublet to the one split by 2W,, and then vanishes due to "freezing" of the resonance
phonons (see fig. 99, T < TjT). Consequently, the main scatterer of phonons is not
the sample grain boundaries, but the boundaries of the Jahn-Teller domains. The
Jahn-Teller domains are the regions in which the equivalent orthorhombic axes are
directed along different second-order axes in the basal plane of the tetragonal crystal
cell. In T m V O 4 the orthorhombic axes a' and b' are rotated by 45 ° relative to the
axes a and b of the nondistorted lattice (Kazey and Sokolov 1986). The sample grain
boundaries begin to scatter phonons only at T ~ 100 m K (see figs. 99 and 100).
(2) At T > TjT (approximately to 6 K, i.e. ~3TjT ) 1¢L continues to experience the
influence of phonon resonance scattering on the above mentioned level system (see
fig. 98b).
A similar effect has been observed in the measurement of the linear expansion
coefficient 5 in TmVO4 (Kazey and Sokolov 1986). The contribution to the value of
5 from the C J T E does not vanish at T_~ ~T, but continues up to T ~ 3Tj T. Kazey
and Sokolov (1986) have connected such a behaviour of 5(T) with the presence of
significant distortions in the tetragonal crystal lattice of T m V O 4 at T > TjT.
Other explanations of the observed decrease of ice in TmVO4 in the region T > T~T
have been proposed - the influence of a soft mode, pseudospin thermal conductivity
and so on (Daudin and Salce 1982).
A strong increase of ~cL has been observed on applying a magnetic field H along
the c-axis of a crystal of TmVO4 (H II c) over a wide temperature region T < TjT to
T > Tjv (figs. 100, 101) (Daudin and Salce 1982, Daudin et al. 1984). tcL increases
i I i I i I I
10 2 __
101 H=70k~iO0~e~f'~
E
/o o
~ I0 °
N
F i
p J
10 -I j m
d
d
I0-2
0::7
,/ Fig. 100. The temperature dependence of lcL for
a single crystal of TmVO4 at H = 0 and 70 kOe,
10-3 I HEIc (Daudin et al. 1984). The dashed line
I I , I i I
i0 -2 i0-I 100 101 102 givesthe calculated 1% due to grain boundary
T,K scattering.
,o I-f", , ' - - -
E..E. 10 -1
\
10-2
Fig. 101. The magnetic-field dependence of KL for a single crystal of

0 20 40 60 80 TmVO4 (Daudin et al. t984). Data are given for the temperatures 150
H, kOe and 700 mK.
with H up to 20-30 kOe, and then becomes saturated (fig. 101). The magnetic field
strongly influences the structural phase transitions induced by the CJTE (Gehring
and Gehring 1975, Cooke et al. 1972, Kazey and Sokolov 1986, Vekhter et al. 1987,
Melcher et al. 1973). Three cases can be realized (Vekhter et al. 1987):
(1) Full suppression of the transition. The external field splits or mixes the electron
states in such a way that the local Jahn-Teller distortions are reduced, TjT decreases
(sometimes it reaches zero).
(2) The external field increases local distortions, and causes a homogeneous
distortion and makes the phase transition indistinct.
(3) The phase transition in the external field remains, but the temperature interval
of the ordered phase increases.
The effects observed in TmVO4 can be explained, more probably, on the basis of
the first supposition.
In addition, a reorientation of the Jahn-Teller domains occurs in the magnetic
field, and at some critical mean field He, the sample reaches a single domain state
(Cooke et al. 1972, Kazey and Sokolov 1986, Metcher et al. 1973). Ho~ in TmVO4 is,
apparently, equal to ~ 70 kOe, because at this mean field the following condition is
fulfilled: phonons are scattered on the sample grain boundaries but not on the
boundaries which separate the domains (fig. 100, dashed line).
It is necessary to note that, as the conclusion of this section, all the above-
mentioned peculiarities of ~cL(T) behaviour found in TmVO4 are observed as well in
other investigated compounds of the family of rare earth zircons.
8. Thermal conductivity of rare earth superconductors
8.1. Main principles
Let us consider qualitatively some conclusions of the theory of thermal conductivity

of standard superconductors which may be useful in analysing tc data of rare earth
T H E R M A L C O N D U C T I V I T Y O F RECs t83
superconductors. This theory has been developed by Geilikman (1958), Geilikman

and Kresin (1958, 1959, 1972) and Bardeen et al. (1959).
The most important properties of superconducting electrons are that they do not
transport energy and do not interact with phonons. At T < Tc the number of heat
transporting electrons decreases by an exponent determined by the energy gap A(T)
at the Fermi surface and the electron thermal conductivity
~c~~ exp [ - A(T)/T]. (50)
The relative contributions to the total thermal conductivity, ~c~ot,from ~c~ and ~c[ in
the superconductivity region and their temperature dependences are determined,
mainly, by A(T) and Tc (superconducting and normal states are marked, respectively,
by the indices s and n).
Let us consider the case of a metal with t<~ ~ ~c~.
(a) The value Tc is small and electrons are mainly scattered by defects. The electron
thermal conductivity in a normal state ~cn oc T. In the superconducting state at T < T~
the value ~cs decreases rapidly due to the decreasing number of normal electrons. At
very low temperatures, when ~:~is practically negligible, ~c[ shows up. ~[ is somewhat
larger than ~c[ at the same temperature due to absence of phonon-electron scattering.
It is possible at very low temperature to reach the region ~c[ oc T 3, where boundary
scattering dominates (see fig. 102).
(b) The value To is not so small and electrons are scattered by defects and lattice
vibrations. At T < T~, ~c~ decreases at first (as in the previous case) and then can
increase slightly due to a weakening of the electron phonon scattering. At very low
temperatures ~c[ can show up on the background of the small ~ (fig. 103).
In the case ~c~>>~c~ at T < T~ the ~c[ becomes greater than the ~c~ due to a decrease
of the phonon-electron scattering (fig. 104). A combination of variants of
figs. 102 104 is possible.
"[¢
J/
T
~T3/ I Fig. 102. Schematic shape of the temperature dependence of l<[,
~c~ and K~ for a metal when electron-impurity interaction is the
P, T main electron scattering mechanism. KL ~ ~co.
f \
/ \
/
\\
Fig. 103. Schematic shape of the temperature dependences of 1c[,

~c2and tc~ for a metal when electron impurity and electron-phonon
~-T interactions are the main electron scattering mechanisms. ~L ~ Kc.
]'c
/
/
~ //
/
Fig. 104. Schematic shape of the temperature dependences of ~ t

I, T and t¢~ot for the case KL ~>1%.
An external magnetic field H > Her (critical field) destroys superconductivity and
transfers the s u p e r c o n d u c t o r to the n o r m a l state. F o r pure metals Her can be small.
In superconductors of the second kind at T < To in a mixed state (in a magnetic
field between He1 and Ho2, where Hol is the lower and H~2 the upper critical field)
a new effect in the thermal conductivity is possible: p h o n o n and electron scattering
by Abrikosov vortex lines penetrating the s u p e r c o n d u c t o r (Red'ko and Chakalski
1987).
8.2. Thermal conductivity of heavy-fermion systems
Steglich et al. (1979) have observed in CeCu2Si 2 a new effect in solid state physics:
heavy-fermion superconductivity (HFS) with Tc = 0.56 K. In the following years the
HFS has also been observed in UBe~3 (To = 0.854 K) (Ott et al. 1983) and UPt3 (To =
0.520 K) (Stewart et al. 1984). For the sake of completeness we will consider data on
the thermal conductivity of all three heavy-fermion superconductors.
HFS has a number of distinguishing features. Let us note some of them:
(1) In these systems triplet superconductivity is possible with a parallel orientation
of the electron spins in a pair with total spin 1 (superconductivity in HFS-systems
is connected with Cooper pairs, formed by fermions with a very large effective mass
(m*/mo ~ 100-1000, m0 is the free-electron mass)).
(2) A non-phonon mechanism of superconductivity is possible (Gudak 1985).
(3) The heavy-fermion superconductors of the second kind (such as CeCu2Si2,
UBe~3 and UPt3) have a ratio Ho(O)/Tc(O) significantly higher than the classical
superconductors. According to the theory H¢(O)/Tc(O) does not depend on the type
of superconductor. So, at Tc = 1 K the magnetic field Ho(0) which destroys supercon-
ductivity (paramagnetic limit) is 16 kOe. H~(0) values in excess of the paramagnetic
limit is a characteristic of HFS systems. For example, in CeCu2Si 2 it is two times
and in UBe13 it is more than six times as large.
Magnetic impurities suppress superconductivity and nonmagnetic impurities
weakly influence T~ in classical superconductors. In HFS systems even nonmagnetic
impurities suppress superconductivity. For example, the higher the concentration of
Ce 3 + ions in the alloy CexLa~ _xCu2Si2, the higher is T~ (Buzdin and Moshchalkov
1986). So, a highly unusual situation arises: a decrease of the number of magnetic
centres leads to a sharp decrease of To.
Can one obtain from the thermal conductivity some independent information
about the nature of superconductivity in HFS systems? It turns out that from the
temperature dependence of ~c~ at T ~ To one can decide if there is triplet or singlet
coupling in the HFS system. For triplet coupling a vanishing energy gap occurs only
at points on the Fermi surface, whereas for singlet coupling vanishing energy gaps
can form entire lines on the Fermi surface (Volovik and Gor'kov 1984, 1985, Varma
1985, Schmitt-Rink et al. 1986). This is the only way to determine the coupling type
at the present time (Volovik and Gor'kov 1984, 1985). One can obtain indirectly
information about the behaviour of the gap, A, from the temperature dependences
of the heat capacity and the thermal conductivity of HFS systems (see table 6 and
fig. 105) (Varma 1985).
Let us consider the experimental data on the thermal conductivity of CeCu2Si 2
(Sparn et al. 1985, Steglich et al. 1985a, b, Franz et al. 1978, 1979, Schneidner et al.
TABLE 6
Influence of the superconducting-gap shape on the thermal properties of semiconductors.
Properties at T ~< T~ Quasi-particle spectrum Gap vanishes at Gap vanishes at lines

has a gap over the points on the Fermi on the Fermi surface
entire Fermi surface surface (fig. 105b) (fig. 105c)
(fig. 105a)
Heat capacity o~ e x p ( - A / T ) o~(T/A) 3 oc(T/A) 2
Thermal conductivity 1¢~ oc e x p ( A / T ) oc(T/A) 3 oc(T/A) 2
Kz Kz Kz
Fig. 105. Schematic shape of the supercon-

ducting gap A(k) in momentum space (Buzdin
Kx Ky Kx and Moshchalkov 1986) (a) in a standard
superconductor; and in a heavy-fermion super-
A
conductor with (b) vanishing gaps at some
a b c points and (c) on lines on the Fermi surface.
1983), UBe13 (Sparn et al. 1985, Jaccard and Flouquet 1987, Flouquet et al. 1986,
Jaccard et al. 1985a, b, Ravex et al. 1987, Alekseevski et al. t986) and UPt3 (De
Visser et al. 1987, Floquet et al. 1986, Steglich et al. t985a, Jaccard et al. 1985b,
Sulpice et al. 1986, Franse et al. 1985). At T ~ T~ all three compounds have ~[ ~ ~c~
and the ~c~ dependence on T has the form
tc~ = ~ T 2 + f i T (51)
The values of ~ and fl are given in table 7. The temperature dependence tc~oc T 2
which is observed in all three compounds, corroborates the presence of singlet
coupling (see tables 8 and 6), however, the temperature dependence of the heat
capacity C oc T 3 (see table 8) in CeCu2 Si2 and UBe13 contradicts this conclusion (see
table 6). Experimental data on the absorption coefficient of ultrasound in UPt3
(Bishop et al. 1984) and on the reciprocal time of the spin relaxation in CeCu2Si2
(MacLaughlin et al. 1984) and UBe13 (Clark et al. 1984) confirm the conclusion
about singlet coupling.
At present there is no clear understanding of the nature of the linear term in ~c~.
In a number of works the appearance of this term is connected with the presence of
impurities in the samples (or the nonsuperconducting phase), because the contribution
TABLE 7
Experimental values of the coefficients c~ and fl in the dependence tc~ = fiT+ c~T2 for T < T~.
Compound c~ Refs.* /? Refs.*

(mW/K 3 cm) (mW/K 2 cm)
CeCu 2 Si 2 2.8 [1] 0.7 [1] a

1.8 [2]
UBe13 0.38 [1] 0.03 [4]"
0.53 [3]
UPt3 19.5 [1, 5, 6] 0.55 [1, 6-8]
20 [7, 8]
URu 2 Si2 0.96 [9] 0.2 [9]
aThe contribution fiT for CeCu2Si 2 and UBe13 has also been observed by Franse et al. (1984) and Jaccard
and Flouquet (1987).
*References: [1] Jaccard et al. (1985a), [2] Steglich et al. (1985b), [3] Alekseevski et al. (1986), [4] Ravex
et al. (1987), [5] Steglich et at. (1984), [6] Jaccard et al. (1985b), [73 Sulpice et al. (1986), [8] Schmitt-
Remi et al. (1986), [9] Lopez de la Torre et al. (1988).
TABLE 8
Experimental temperature dependences of C and xso for T < T~.
Property CeCu 2 Si 2 Refs.* UPt 3 Refs.* UBe13 Refs.*
C T3 [1] T2 [1, 2] T3 [3]

Ke
T2 [4-6] T2 [1, 4, 7] Tz [4, 8-10]
*References: [1] Steglich et al. (1985a), [2] Sulpice et al. (1986), [3] Ott et al. (1984b), [4] Jaccard et al.
(1985a), [5] Steglich et al. (1985b), [6] Franse et al. (1984), [7] Jaccard et al. (1985b), [8] Sparn et al.
(1985), [9] Varma (1985), [10] Alekseevski et al. (1986).
of the linear term to the thermal conductivity decreases with increasing purity of the
material. A linear term is also observed in the heat capacity at T ~ T~ in a number
of works [e.g., in UBe13 (Ravex et al. 1987) and UPt3 (Sulpice et al. 1986)]. Figures 106
and 107 show, e.g., data on x[ of UPt3 (Schmitt-Rink et al. 1986, Clark et al.
[ I i
6
f,./ //
T¢
me /,J
N
0 I I 1 I t Fig. 106. The temperature dependence of
0 100 200 300 400 500 x~/T for UPt3 (Schmitt-Rink et al. 1986,
T,mK Clark et al. 1984).
- - T ] ]-- I
25-
20-
15-
=L
N 10-
/ /
/
5--
I I I I Fig. 107. The temperature dependence of

0 0.05 0.10 0.15 0.20 x~/T for UBe13 (Volovik and Gor'kov 1984,
T, K 1985).
188 l.A. SMIRNOV and V.S. OSKOTSKI
1984) and UBe13 (Volovik and Gor'kov 1984), which illustrate the T 2 and 7' depen~-
dences in the low-temperature region.
8.3. Thermal conductivity of rare earth compounds in which superconductivity and

magnetism coexist
For a long time it was thought that superconductivity and magnetism could not
coexist (Ginzburg 1957). A small amount of magnetic impurities (up to ~ 1%) destroys
superconductivity (Matthias et al. 1958, 1959, Suhl et al. 1959). Superconductivity is
destroyed also during a phase transition from the superconducting to the magnetic
state. However, at the beginning of the 1970s two classes of ternary compounds,
RxMo6X8 (x = 1 or 1.2, X = S, Se) and RRh4B4, in which magnetism and supercon-
ductivity coexist, were discovered by Hamaker et al. (1979) and Fischer et al. (1979).
Several dozens of magnetic superconductors are known at the present time. In the
system RxMo6Xs (rhombohedral-hexagonal crystal structure with rhombohedral
angle near 80 °) all compounds with R from La to Lu (excluding Ce and Eu) are
superconductors (Fischer et al. 1975, Shelton et al. 1976). Coexistence of magnetism
and superconductivity is observed in R1.2Mo6S8 (R = Gd, Tb, Dy, Er) (Ishikawa
and Fischer 1977, Moncton et al. t978) and RxMo6Se 8 ( x = 1, 1.2; R = Gd, Er)
(McCallum et al. 1977a, b). In the system RRh4B4 (tetragonal crystal structure)
superconductivity is observed in compounds with R = Y, Nd, Sm, Er, Tm, Lu and
superconductivity coexists with magnetism in compounds with R = Nd, Sm, Er, Tm
(Matthias et al. 1977, Vandenberg and Matthias 1977).
Different electrons are responsible for magnetism and superconductivity in the
case of their coexistence. Superconductivity arises because of the formation of Cooper
pairs of outer shell electrons, while magnetism is connected with the magnetic
moments of the inner electron shells of the lanthanide ions. As has been shown by
Buzdin et al. (1984), competition between superconductivity and ferromagnetism
leads to the appearance of a new type of magnetic order in the superconducting
phase a magnetic domain structure, which is an average between ferromagnetism
and antiferromagnetism (Buzdin and Moshchalkov 1986).
Since then extensive investigations of the physical properties of systems with
coexistence of superconductivity and magnetism have been carried out. The thermal
conductivity of the RRh4B 4 system (R = Sm, Er, Tm, Lu) (Ott et al. 1980, Hamaker
et al. 1981a, b, Odoni and Ott 1979, Odoni et al. 1981) has been investigated in
detail. Table 9 gives some parameters of these materials which are necessary for a
discussion of the results on to. T~I and T~2 are, respectively, the high and the second
critical superconducting temperatures. As one can see from table 9, the width of the
temperature interval of coexistence of superconductivity and magnetism is small:
~0.5 1 K for TmRh4B4 and SmRh4B4 and ~0.05 K for ErRh4B4. The last interval
has been fixed only by means of precise measurements of the magnetic susceptibility
and the linear expansion coefficient ( O t t e t al. 1978, Woolf et al. 1979). Therefore
one can try to search for an influence of this effect on ~c only in TmRh4B4 and
SmRh4B4. Nevertheless, we will consider in this section, to make the picture com-
plete, all four compounds from table 9. LuRh4B~ can be considered a reference
TABLE 9
Data on T~a, T~z, Tc and TN for some compounds RRe,~B,~.
Compound To1 Refs.* T~2 Refs.* Tc Refs.* TN Refs.*

(K) (K) (K) (K)
SmRh4 B4 2.72 Ill 0.87 [1, 2]

2.68 [2]
ErRh4B 4 8.7 [3-5] 0.98 [~3,4] 0.93 [3, 4, 6]
TmRh~B~ 9.8 [7, 8] 0.4 [7, 8]
LuRh 4 B4 1 t.6 [4]
*References: [1] Hamaker et al. (1979), [2] Ott et al. (1980), [3] O t t e t aI. (1978), [4] Woolf et al. (1979),
[5J Odoni and Ott (1979), [6] Fertig et al. (1977), [7] Hamaker et al. (1981a), [8] Hamaker et al. (1981b).
material, because it undergoes only the superconducting transition. Figure 108 shows
the temperature dependence of ~Ctot of LuRh4B~ (Odoni et al. 1981). tel and 1£e are
separated into the normal and superconducting state contributions by calculation.
At T ~ To a sharp increase of tc~ (tc[) occurs due to decreasing phonon scattering by
the electrons, and a decrease of tc2 (tos) occurs. At the lowest temperatures the thermal
conductivity in the superconducting state 0d) is only due to ~c[. At T < 2 . 5 K
td = tc[ and is proportional to T a, which is typical for boundary phonon scattering.
ErRh4B ~ is a prototype of compounds with two superconducting transition temper-
atures (T~ and To2) - "reentrant" superconductors (see table 9 and fig. 109). Figures
110 and 111 show the K(T) dependences. Unlike LuRh4B4, the contribution of ~cL
t o /('tot is small near T~. However, at low temperatures/£tot becomes equal to tc[ and
proportional to T 2"s (as in the case of LuRh4B4 (see figs. 110 and 11 la)). At T < 1.2
1.4 K lqot increases with decreasing temperature and at T ~ T~(T¢2) it has a maximum.
At T < To (To2) the behaviour of the thermal conductivity changes radically. Now
the main heat carriers are not phonons (~cL decreases sharply again due to strong
phonon-electron scattering), but electrons, ~cL ~ ~ce and/£tot = Ne' This conclusion is
proved by the temperature dependence of tgtot = /£e OC r~ which is typical for electron
scattering by static defects and impurities. On extrapolation of T to zero, tro(T) tends
5
Tc
xn
4
E 3
ff~ e
d 2
Fig. 108. The temperature dependence of 1¢

1 ! .... for LuRh4B,~ (Odoni et al. 1981). Solid line
is the experimental iCtot at H = 0, points give
0 the experimental lgtot at H = 5 kOe, dashed
0 5 10 15 lines show calculated values x2, 1~~, Jc~ and
T,K 1~,
180 I" Tc2 T C 12kOe Tel
c~:a_120600~~
H=0 ROe
Fig. 109. The temperature dependence of the electrical
o I 1 i I I I i i resistivity, R, for ErRh4B~ without a magnetic field and
0 1 2 3 4 5 6 7 8 lo in a longitudinal magnetic field H = 12 kOe (Ott et al.
T,K 1978).
10
Tq Xt°t // -
Tc~ I ~/~I//- xe
~c
fi t
t /
I ,~ 4 I
I Fig. 110. The temperature dependence
of ~ for ErRh4B 4 (Odoni et al. 1981).
ilJ....'(, , a / i "i'-'?- ~.__l.~__e_l__gr- 7- q-- The solid line is the experimental value.
0 2 4 6 8 10 12 14 The dashed line gives the calculated
T.K ~", 1<~,t¢[ and t~.
f m I '
1.0
E
Xn // ~,S
O.1 T2.B2,XS
,(, Lt, ,o.
0.02
0.03 0.1 1.0 10
T,K
f
Fig. 111. (a) The temperature dependence of 1¢ for ErRh4B 4 in
0 0.2 0,4 0.6 0.8 1.0 1.2 region I of fig. 110 (Odoni and Ott 1979). (b) A more detailed
T,K picture near the phase transition.
to zero (fig. 11 lb). Besides, ~qot(T) can be described well by the Wiedemann-Franz
law with L = Lo. Near the phase transition to(T) of ErRh4B4 shows hysteresis
(fig. 11 lb), similar to the case of R(T) (fig. 109).
Figures 112-115 show ~c(T) and R(T) dependences for SmRh4B4 and TmRh4B 4
with the coexistence of superconductivity with antiferro- and ferromagnetic ordering,
respectively (Hamaker et al. 1979, 1981a, b, Ott et al. 1980). At H = 0 in the regions
of superconductivity (II), of superconductivity and magnetism coexistence (I), and
also at the points Tc (0.4 K for TmRh4B4) and TN (0.87 K for SmRh4B4) no anomalies
in ~cs behaviour are observed except for the decrease of the value n in the dependence
~cs oc T" in the region of superconductivity and magnetism coexistence from _~2.8
(typical for tc[) to ~-1 0doc T) with decreasing temperature. Ott et al. (1980) have
connected this effect with the appearance of a magnon contribution against a back-
ground of small tcL.
In the normal state [which requires a field ~ 2 kOe for SmRh4B 4 (fig. 112) and
~ 3 kOe for TmRh4B 4 to quench superconductivity (fig. 114)] ~cnoc T, as in the case
10 t I I
10 °
T I
10-1
//~s I
10-2
r/N T:I
i0-3 / I tl / Fig. 112. The temperature dependence of 1~for SmRh4B 4 without
0.01 0.1 1.0 10 a magnetic field and in a longitudinal magnetic field. The numbers
T,K by the curves are the values of H in kOe (Hamaker et al. 1981b).
N C1
100 - ~
l ----T--" I I
80
~ fF /r
I
f
I
I
J / / // I
/ / f
60 L / J
i/1.7~1.5 11.0 ~ 0.4
/ I J
40
I I f, I H=O kOe
/ i I
I I -4 Fig. 113. The temperature dependence of
20 I I
I I the electrical resistivity, R, for SmRh4B 4
!
I
I without a magnetic field and in a longitu-
0 / I I I / /J I dinal magnetic field. The numbers by the
0 0.5 1.0 1.5 2.0 2.5 3.0 curves are the values of H in kOe
T,K (Hamaker et al. 1979, O t t e t al. 1980).
I I --
10 °
31
10-1
E
"~. 10~2
7 T
10.3
,/,% Fig. ll4. The temperature dependence of ~: for

10 . 4
TmRh 4 B4 without a magnetic field and in a longitudinal
0.01 0.1 1.0 10 magnetic field. The numbers by the curves are the values
T,K of H in kOe (Hamaker et al. 1981a, b).
TC Tq
400
I ~ 1 I
300
i', /" / /
\\ x\ // I I 2~ "0
c~ 200
::L
E \: x\ ,~ f
100 . ,. / 9 /7 3 Fig. 115. The temperature dependence of the electri-
3
7~\ "~" I cal resistivity, R, for TmRh4B4 without a magnetic
_0 I I"'-. I I I .'tl I I I field and in a longitudinal magnetic field. The num-
0 2 4 6 8 10 bers by the curves are the values of H in kOe
T,K (Hamaker et al. 1981a, b).
of ErRh4 B 4, and this can also be explained by the prevailing contribution of % at

these temperatures.
It is interesting to observe the influence of a magnetic field on ~c in the presence
of the specific magnetic d o m a i n structure typical for the m a g n e t i s m / s u p e r c o n d u c t i o n
coexistence region. A n o m a l o u s behaviour of tc in a magnetic field has been observed
in T m R h ¢ B 4 (fig. 114):
(1) A w e a k magnetic field strongly influences both the value and the temperature
dependence of the thermal conductivity.
(2) Large hysteresis is observed in the dependence lc(T) for T > Tc, which
disappears at T-~ T~, and for T < To tcs is proportional to T.
(3) The phase transition temperature shifts with a magnetic field to higher temper-
ature. At H = 3 kOe, ~c(T) of T m R h 4 B 4 behaves like ~c(T) of ErRh4B 4 in the interval
Tc <. T<~ To2 in zero magnetic field (see fig. 111).
So, the unusual behaviour of ~c(T) of TmRh4B4 in a magnetic field can be explained
by a series of phase transitions induced by the magnetic field. This field leads to
magnetic ordering, which starts the destruction of superconductivity.
8.4. Some words on the thermal conductivity of high-temperature superconducting rare

earth compounds
In the past few years investigation of high-temperature superconductors (HTSC)

has become one of the fashionable branches of solid state physics. More than a
hundred papers have been published on the thermal conductivity of HTSCs (this is
only an insignificant part of the several thousand papers on other physical properties
of HTSCs). Nevertheless, because of the limited size of this review we cannot discuss
in detail even the principal articles on ~c for HTSCs. Recently several reviews and a
number of general articles have been devoted to this aspect of HTSCs (Fischer et al.
1988, Kirichenko et al. 1990, Jezowski and Klamut 1990, Uher 1989, 1990, Regueiro
and Castle 1990, Florent'ev et al. 1990). One can find references to the majority of
published works on ~c of HTSCs in these review articles. The theory of ~c in HTSCs
is considered by Tewordt and W61khausen (1989). Investigations of ~c in HTSCs
containing rare earth elements (system 1-2-3:RBa2Cu3OT_6 (R = Y, Eu, Gd, Ho,
Sm, Er, Dy); system 2 1 4:R2 xSrxCuO4) can be separated into four stages:
(1) Imperfect, often multiphase, ceramic samples have been measured. Everyone
associated with these papers has strived to measure something and to publish it as
soon as possible. These data are not trustworthy at present.
(2) Measurements have been carried out on well prepared and certified ceramic
samples of high density. These results are trustworthy.
(3) Measurements have been carried out on small single crystals (plates with sizes
~0.5 x 0.5 mm). Some difficulties have arisen. First, many of the single crystals that
have been prepared are imperfect. Second, the classical, precise methods for ~c
measurements were developed for large samples. Thus, it has been necessary to
develop a new technique, but significant errors are associated with it.
(4) The quality of single crystals has been improved, and their dimensions have
grown (but not enough to use the classical methods of ic measurement). The precision
of the new methods has increased due to technological progress. The anisotropy of
the thermal conductivity has been investigated. The ~c of single crystals is 8 l0 times
higher than the tc of ceramic samples.
A number of peculiarities of ~c(T) has been observed in HTSCs. Let us consider
some of them.
(1) In contrast to classical superconductors lCL~> ~Cein HTSCs (in polycrystalline
HTSCs it is 10 25%, in single crystals 30-50%).
(2) A nonstandard temperature dependence of ~c(T) is observed for T > T~ and
T < T~ (fig. 116). For T > Tc ~c_~ const. For T < T~, the value of ~c initially increases
with decreasing T, passes through a maximum and then decreases. Between 10 and
1 K , 1¢ oc T2; at T < 1 K a linear temperature dependence (~coc T) is observed in some
samples. For the greatest part the peculiarities of tqot in HTSCs are connected with
tcL, because, as has been noted above, t¢e > tee. Independence of ~cc of T for T > T~
194 I.A. SM1RNOV and V.S. O S K O T S K I
Io x
ATA,~/ T~
T3
f//.,("I// B Fig. 116. Schematic shape of ~c for a HTSC (on a

logarithmic scale). Curves A and B are possible tem-
perature dependences of lc for different groups of
log T materials and samples.
can be explained either by strong phonon-electron scattering or phonon-defect

scattering. Oxygen vacancies, twins and structural inhomogeneities serve as defects.
The increase of ~cL for T < T~ can be explained by a decreasing phonon-electron
scattering in the superconductivity region (cf, figs. 116 and 104). There is another
explanation of this effect. Ginzburg (1989a, b) hypothesised that the convective heat
transport by normal current carriers, which is negligible in classical superconductors,
becomes significant in HTSCs. For T < To, ~cincreases at first and then, as the normal
carriers "condense", decreases. The higher the contribution of ~ce to /{tot, the higher
the maximum of ~c. The ~cL oc T 2 dependence at low temperatures in crystals can be
explained by phonon scattering by conduction electrons (in superconductors by
normal carriers), dislocations and in glasses or heavily disordered systems by reso-
nance phonon scattering of two-level systems, see section 9.l. An estimation of
Regueiro and Castello (1990) shows that the second version is more likely in the case
of HTSCs, because at these temperatures in the investigated HTSC samples there
are no longer any normal carriers, and a model taking phonon scattering by disloca-
tions into account leads to an unrealistically high dislocation density. By the way,
the same temperature dependence ~coc T 2 has been observed in superconducting
heavy-fermion systems (section 8.2). However, for these materials it is ~ce(T), and for
HTSCs it is ~L(T). Very likely, the oxygen vacancies are responsible for the structure
"disorder" in HTSCs. The dependence ~ oc T has been observed in a number of
HTSC samples at superlow temperatures. Mostly this is connected with scattering
of normal electrons on static defects. In this temperature region, tcL becomes insignifi-
cant and ~ce is appreciable (Ke oc T). It is not clear where the normal electrons come
from [according to the estimate of Kirichenko et al. (1990) and Uher (1990) they
make up ~15% in the 1-2-4 HTSCs]. Either they are inherent to HTSCs due to
the high anisotropy of the gap in the electron exciting spectrum, or they are present
because of impurities. As noted earlier, the linear term of ~c(T) has been observed in
the superconducting heavy-fermion systems and in the systems in which superconduc-
tivity and magnetism coexist. In the first case, as in the case of HTSCs, this term is
connected with scattering of normal current carriers by defects, and in the second
case with the contribution lcm to ~qot. Probably, structural defects are responsible for
the linear term in ~cL(T) of HTSCs. It should be remembered that in 1-2-3 systems
the linear term is not observed up to T-~ 50 m K (Regueiro and Castello 1990). A
different interpretation of the low-temperature data on ~ of a number of HTSCs is
given in some works (Uher 1990, Florent'ev et al. 1990, Sparn et al. 1989). The ~c(T)
dependence is approximated by the equation ~c(T)= aT+ bT 3. The second term is
connected with boundary scattering of phonons.
Probably, additional independent experiments are necessary to explain the origins
of the different K behaviour of HTSCs at low temperatures.
Concluding this section, we present in fig. 117 the experimental 1trot(T) of the 1-2-3
system for a polycrystalline sample (the average/£tot value) and for a single crystal
(the maximum Ktotvalue known at the present time) (Florent'ev et al. 1990).
9. Thermal conductivity of rare earth glasses and crystalline materials with glass-like
behaviour of 1%(T)
9.1. Thermal conductivity of rare earth glasses
Glasses based on the rare earth elements have been investigated intensively in the
last fifteen years because of their interesting and prospective practical spectral-
luminescent properties [see, e.g., Reisfeld and Bornstein (1977), Reisfeld et al. (1977,
1979), Kamarzin et al. (1983a, b), Sokolov et al. (1988)]. However, data on the thermal
conductivity in the literature are scarce: Smirnov et al. (1985, 1988, 1989, 1991a),
Parfen'eva et al. (1990a, b, c), Guessons and Mazuer (1982). The unusual thermal
conductivity and heat capacity of glasses and amorphous systems at superlow, low
and high temperatures has attracted the attention of theoreticians and experimental-
ists [see, e.g., Anderson et al. (1972), Phillips (1972, 1987), Karpov and Parshin (1983,
1985), Pohl and De Yoreo (1985), Jagannathan et al. (1989), Galperin et al. (1988),
Krasny et al. (1990), Freeman and Anderson (1986), Love and Anderson (1990)].
Rare earth glasses possess all peculiarities of ~c(T) and C(T) inherent in usual glasses
but they show also a new effect - resonance scattering of phonons by the paramag-
netic lanthanide ions [see Smirnov et al. (1985, 1988, 1989, 1992a), Parfen'eva et al.
(1990a, b)], which has been discussed in detail in section 6.
Below we will consider briefly the main results on the thermal conductivity of rare
earth glasses which in principle can be explained in the framework of contemporary
theory (Anderson et al. 1972, Karpov and Parshin 1983, 1985). We will discuss
separately the data for insulating and metallic rare earth glasses.
196 I.A. S M I R N O V and V.S. O S K O T S K ]
m:-F-VI-I-r----F~I t ~ =~qqq----- I -
- ~f ~. -
._ ..i \ _
- 3 I "/ I X
/ I \\
- / ~ I \ "
/ "/ \ I \
O.1
E
~J I
t
- I
Fig. 117. The temperature d e p e n d e n c e of

lctot for a single crystal s a m p l e of
O.O1 - Y o . 9 s Y b o . 0 2 B a a C u 3 0 ? - x (l) and for a poly-
crystalline Y B a 2 C u 3 0 7 x (2). D a t a for C u O
--I llil I I I I Ill ]
are s h o w n for comparision (3) (Florent'ev
lO lOO
T,K et al. 1990).
9.1.1. Thermal conductivity of insulating rare earth glasses

According to the existing theory the temperature dependence of Ice in insulating
glasses can be represented as a kind of staircase (fig. 118) (Karpov and Parshin 1983,
1985, Galperin et al. 1988). At low temperatures (T < 0.1-5 K) the heat is transported,
mainly, by the thermal phonons scattered by two-level systems. Here Kc is propor-
tional to T 2 (region I in fig. 118). The first "step" (region II, T ~ 5 10 K), where teL =
const., is connected with resonance scattering of thermal phonons by anharmonic
oscillators, which show Van Hove singularities (Galperin et al. 1988) in their densities
of states. In region III the increase of KL is connected with "subthermal" phonons
with energies much less than the energy of the phonons scattered resonantly by the
I i i
i I I
I j I
rrr I I:2 I ~ I
I i I
i I
T1 X ~T3
>...--; ,,
I- I t I
I I I
_ _ l I t ..J___~.. Fig. 118, Schematic shape of the theoretical

T t e m p e r a t u r e dependence of ~cc for glasses.
two-level systems. Here ~L is proportional to 7:. The second "step" (region IV) is
connected with phonon scattering by boundaries of clusters and structural units of
the glass. In this region the conditions that the free path length 1 is constant and
that l does not depend on temperature must be fulfilled. The increase in thermal
conductivity in region V (fig. 118, A~{) is connected with the appearance of photon
thermal conductivity. (The heat capacity of glasses in region I has a temperature
dependence C oc 7; which with increasing temperature turns into the standard depen-
dence C oc T 3 and then at high temperatures into C = const.)
The experimental data on ~<i~ of rare earth glasses are based, mainly, on
RzS3(GazS3) x (R = La, Ce, Pr, Nd; x = 2 or 3) glasses (Smirnov et al. 1985, 1988,
1989, 1991a, Parfen'eva et al. 1990a c). In the system R2S3 R 2 0 3 - G a 2 O a GazS3
(R = lanthanide) glass formation is observed over a wide region (Sokolov et al. 1988,
Guittard et al. 1986). Figure 119 shows as example the glass formation region for
R= La.
Glasses based on GazS 3 are formed with lanthanides from La to Er (Loireau-
Lorach et al. 1976), and glasses based on G a 2 0 3 are formed with lanthanides from
La to Nd (Sokolov et al. 1988). Lanthanides following Nd can enter into a glass
composition together with some light lanthanides. The width of the glass formation
region of the system (R2 O3)(Ga2 X3)x (X = S, O) depends on R and decreases to the
side of the heavy lanthanidcs (Loireau-Lorach et al. 1976).
Data on t<L of the above-mentioned rare earth glasses are related to the regions
III V distinguished in fig. 118. Let us consider t<i. of La2S3(Ga203)2 (fig. 120)
LO202S
Lo2S3 ~ ~ ../Lo203
/- /.A
Fig. 119. The glass region (I) in the system

Z La2Sa-La203 G a 2 0 3 - G a 2 S 3 (Sokolov et al.
0.2 0.5 0.8 1988, Guittard et al. 1986).
I I i U I I il I ~ ]
J
I / I
0.8 _ ~,,/ I /-
~L 0.6
0.4 / _
/I I I I I I I I Fig. 120. The temperature dependence of
3 5 10 20 50 lO0 200 300 xL for La2S3(Ga203) 2 (Parfen'eva et al.
T,K 1990c).
(Smirnov et al. 1992a, Parfen'eva et al. 1990c). The temperature dependence of ~cL in
the interval 3-5 K 0%oc T) is typical for region III (fig. 118) [In this temperature
interval the heat capacity C oc T 3 (Smirnov et al. 1992a)]. The free path length of
phonons l (l = 3tcL/Cv ) (Smirnov et al. 1991a) calculated from experimental data on
~L, C and the sound velocity v at T>~ 30-40 K begins to saturate (fig. 121), which is
typical for region IV. And finally in the interval 100-400 K the law Ax oc T 3 is
fulfilled, which is typical for region V. The low-temperature plateau in the dependence
lcL(T) typical for insulating glasses is realized for R2S3(Ga203) x at lower temper-
atures ( T < 4 K).
As shown above (see fig. 73) a variation of the concentration x from 2 to 3 in
glasses R2 $3 (Ga2 O3)x (with R 3 + = La 3 +, Pr 3 ÷) practically does not influence the
value of ~&. On the other hand, substitution of La 3 ÷ by Ce 3 + or Nd 3 ÷ (at constant
x) changes t% significantly in region IV (but has little effect on the value of ~L in
region III) (fig. 122). (The large influence of substitution of La 3÷ by Pr 3÷ on ~cL
caused by strong resonance scattering of phonons by the paramagnetic levels of Pr
is discussed in section 6.1.).
In region IV ~L decreases linearly with the lanthanide atomic number (fig. 122b),
which shows that there is some decrease of the structural unit sizes with decreasing
ionic radius of the lanthanide ion. This effect is purely geometrical and is not
connected with the decrease of xL due to resonance phonon scattering on the
paramagnetic levels of Ce 3 ÷ and N d 3 +. This is confirmed by the absence of the
Schottky effect in the heat capacity of these glasses in the discussed temperature
region (Smirnov et al. 1992a). The split excited levels of the ground states of the ions
Ce and N d lie high and do not scatter the phonons. An analogous influence on ~cL
is observed for the partial substitution of La 3+ by Gd 3÷ in the glass (Gd2S3)0.3
(La283)0.7(Ga2O3)2 (fig. 123) (the Gd 3 + ion has L = 0 and, consequently, resonance
phonon scattering is absent). From figs. 120 and 122 one can see that ~CL(T) of
the considered glasses is somewhat different from the classical temperature depend-
~1 .... t '' '1 i I
10-3!
10-4
E
10- 5
10-6
. _
10-7 I
,I .... I , , ,I I I I Fig. 121. The temperature dependenceof the phonon free path
5 10 50 100 300 length in the glass La2S3(Ga2Oa) (1) (Smirnov et al. 1992b)
T,K and in fused quartz (2) (Zellerand Pohl 1971).
- - - ~ - - T I I I I i I
0.7
0.5
0.3
0.1
Fig. 122. (a) The temperature dependence of ~L for
L a C e Pr Nd the glasses R2S3(Ga203) 2 (R = La (l), Ce (2), Nd (3))
Q. (Parfen'eva et al. 1990c, Smirnov et al. 1992b).
I I I I I I I I I (b) Dependence of t%(T) on the atomic number of R
3 5 10 20 50 1OO in the glasses R2Sa(Ga203)2. T= 20 K (1), 30 K (2),
T,K 40 K (3), 100 K (4).
I I I I I I I I I
0,7
_S O.5
Fig. 123. The temperature dependence of •L for the

o.3 I [I I I I I I I I glass (GdzS3)o.3(La2Sa)o.7(Ga203) 2 (1) (Smirnov et al.
3 5 10 20 50 100 1992b) and for the glass (La2Sa)(Ga2Oa) / (dashed
T,K line 2). For T < 10 K, ~L of both glasses coincide.
ence of the insulating glasses. In the rare earth glasses a small peak in ~cL (character-
istic for crystalline solids) is observed near 10 K. This peak becomes m o r e p r o n o u n -
ced for substitution of La by a heavy lanthanide.
In all a b o v e - m e n t i o n e d glasses the La ions are substituted isovalently. F o r the
nonisovalent substitution of La ions ~:L of the glasses decreases m u c h stronger. F o r
example, when one considers ~:L(T) of the glass (2EuS)o.5(LazS3)o.5(Ga203)2
(fig. 124) (Smirnov et al. 1992a), with the La 3 + ion substituted by E u 2 + ions. In this
glass, apparently, not only variation of the short-range order p a r a m e t e r s occurs, but
also f o r m a t i o n of ionic vacancies as additional p h o n o n scattering centres (Smirnov
et al. 1992a) (the resonance p h o n o n scattering effect o n E u 2 + levels is absent, because
E u 2 + has L = 0). All these glasses belong to the glasses with a "large" ~CLvalue. The
classical insulating glasses (of quartz-type) have a ~:L which is 2 - 3 times smaller.
Besides, these glasses have a noncharacteristic (for the classical insulating a m o r p h o u s
solids) small m a x i m u m of ~:L in the low-temperature region (see figs. 120, 122 and
123). This could point to the presence of small crystalline regions in the glasses of
R z S 3 ( G a 2 0 3 ) x - t y p e (although according to X-ray data these materials are ideal
glasses).
200 LA. S M I R N O V and V.S. OSKOTSKI
- - - - F T - I - - T - - T ~
0.7
I /
/
E
~- 0.5
t! 1
t
0.3
Fig. 124. The temperature dependence of KL for the
I I I _I I_£__J._I_L__
5 10 50 100
glass (2EuS)o.5(LazS3)o.5(Ga/O3)z (1) (Smirnov et al.
T,K 1992b), for the glass La2S3(Ga203) 2 (I).
0.7 m
~--~'-0.5
/
3 /
0.3
Fig. 125. The temperature

dependence of t¢L for the glasses
0.1
f La2S3(Ga203) 2 (1), LaPsO~4
f
(2) (Parfen'eva et al. 1990c,
1 ~ I I it r t I I Smirnov et al. 1992b) and fused
3 5 10 50 1O 0
quartz (3) (Zeller and Pohl
T,K 1971).
However, in the family of rare earth glasses representatives may be found which
have similar properties (both in magnitude and temperature dependence of •L) as
the classical insulating glasses. The glasses LaPsO14 and PrPsO14, which were
discussed in section 6.1 (see also fig. 125) are an example.
9.1.2. Thermal conductivity of metallic rare earth glasses

There are only a few papers devoted to investigation of the thermal conductivity
of metallic rare earth glasses (Guessons and Mazuer 1982). No peculiarities (compared
with standard metallic glasses) have been connected with the presence of the rare
earth elements. The "glass" behaviour of metallic glasses must display itself in both
the value and the temperature dependence of tcL (as it is for insulating glasses). The
metallic glasses differ from the insulating ones by:
(a) the presence of electron thermal conductivity (1trot= ~L + ~C~);and
(b) the presence of phonon electron scattering decreases the value of tq. and
influences its temperature dependence (L6hneysen et al. 1981).
This causes difficulties in the separation of ~L from ~qot. It is known that it is not
easy to separate ~cL from 7Ctotwith a great deal of precision in normal (crystalline)
metals, because ~co>>ICe. For metal glasses this task is even more difficult because ~CL
is smaller than in crystalline metals. According to the experimental temperature
dependence of ~ce all metallic glasses can be divided in two groups (Matey and
Anderson 1977a, b):
(a) a group with a plateau (~cL = const.) in the region 1 10 K; and
(b) a group without such a plateau.
It should be kept in mind that the plateau of ~Ce in the region 1-10 K is typical
for insulating glasses. In a number of works (Love and Anderson 1990, Matey and
Anderson 1977a, b, Zaitlin and Anderson 1975, Dreyfus et al. 1968) attempts have
been made to explain qualitatively the absence of the plateau in metallic glasses (in
this model the plateau can be absent under certain conditions in the insulating
glasses). According to the model the low-frequency phonons (L, with frequency less
than 10m Hz) contribute to ~cL only at low temperatures, and high-temperature
phonons (H) contribute only at high temperatures. The presence or absence of the
plateau in teL(T) depends on the relative contributions of the L and H phonons (see
fig. 126b). The plateau is observed for a large ratio of L and H contributions (Matey
and Anderson 1977a, b). Unfortunately, there are very few experimental data on tel
of the metallic rare earth glasses. Love and Anderson (1990) give data on ~c(T) of the
glasses Ce~oo_~Al~ (x = 12, 43, 70, 77 and 86). These glasses belong to the second
group, they have no plateau in the region 1 10 K. Figure 126 shows, as example,
~c(T) of two glasses of this system (Love and Anderson 1990).
9.2. Thermal conducdvity of crystalline materials with "glass-like" behaviour of ~cL( T )

During the last few years many cases were observed where ~cL of crystals behaves
as ~cL of glasses in some narrow, or over a wide, temperature range. As is known,
100 I l I )--
_ e"~,.°~ ~'
1o
E
u
• ~
E
1.0
Ld_ Fig. 126. (a) The temperature dependence of ~ce for the
metallic glasses Ce23 A177 (1) and Ces7 A143 (2) (Guessons
and Mazuer 1982). (b) Schematic shape of the contribu-
0,1 / ) I I I
1 10 100
tions to ~L of different phonon groups (L and H) (Matey
T~ K and Anderson 1977a, b).
the distinctive features of glasses are a small t¢L and a characteristic temperature
dependence of ~cL. The small ~L value in amorphous (and in some crystalline)
materials is caused by strong phonon scattering. Different types of scatterers can be
responsible for one or another temperature dependences of KL. Let us consider what
factors in crystals could result in decreasing ~cL and a change in its temperature
dependence. Let us divide them in three groups:
(a) first group: factors leading to "amorphous" behaviour of ~L over the whole
temperature region;
(b) second group: factors leading to "amorphous" behaviour of ~cL in separate
temperature regions;
(c) third group: factors leading to the "amorphous" behaviour of ~cL only at high
temperatures.
Let us consider some of these factors. Factors in the first group are: a high
concentration of point defects (e.g., vacancies) strongly distorting the crystal lattice
(Smirnov et al. 1973, 1992b), a decrease of crystallite sizes to some critical value
(Spear et al. 1981) (this effect is essential, as a rule, in thin films), large fluctuation of
composition, a crystal lattice with a great number of atoms in the elementary cell
(Slack et al. 1971, Cahill et al. 1989, Golikova and Tadzhiev 1986, Berezin et al. 1974,
Petrov et al. 1969, 1975, Golikova et al. 1974), the presence in ferroelectric crystals
of sporadic regions with a local electric polarization (De Yoreo et al. 1985) and so
on. A rule was established empirically for ferroelectrics (De Yoreo et al. 1985): if the
dielectric constant and refraction coefficient sharply change at the point of the
ferroelectric phase transition (Tfer), then ~cL(T) behaves as in a normal crystal. If at
the point Tfer these parameters do not change sharply, then ~:L(T) behaves as in
amorphous materials. Effects characteristic for the second group [-where "amorphisa-
tion" is revealed as regions with ~cL = const, on the curves ~cL(T)] can be initiated by
doping with impurities that have a much larger mass than the matrix mass [this
leads to local modes in the phonon spectrum and resonance phonon scattering on
them (Cahill et al. 1989)], by phonon scattering on colloids and large clusters of
impurity atoms (De Gofir 1986), by phonon scattering by inhomogeneities of the
crystal lattice which are caused by irradiation with small doses of neutrons, electrons,
ions, 7-rays (at large doses full amorphisation can be reached) (Oskotski and Smirnov
1972, De Gofir 1986, De Goar et al. 1986), by phonon scattering on disordered crystal
regions, crystallographic domains, soft modes arising at various phase transitions
[see, e.g., Andres et al. (1984, 1985), Salce and Boatner (1986)], by strong acoustic
phonon scattering by optical phonons (Oskotski and Smirnov 1972) and so on. The
third group is related to the case (possible only at high temperatures) when the
phonon free path length reaches interatomic distances, i.e. becomes less than a
phonon wavelength (the minimal mean free path of phonons in a crystal must be of
the order of the phonon wavelength) - the so-called case of minimal thermal conduc-
tivity (Slack 1979). In this case the ~cL of crystal and amorphous materials are about
the same in magnitude and ~L ceases to depend on temperature (Slack 1979). Let us
consider a few of the most interesting examples of "glass" behaviour of 1eL(T) in
crystalline rare earth compounds.
9.2.1. "Amorphisation" o f 1¢L arising due to a complex crystallographic structure

Let us consider data on ~&(T) of RB66 (R = Y, Gd) (Slack et al. 1971, Cahill et al.
1989, Golikova and Tadzhiev 1986, Tfirkes et al. 1986, Raychaudhuri et al. 1980).
The crystal lattice of RB66 is cubic and is rather complicated (Richards and Kasper
1969). The elementary cell of YB66 (lattice constant a = 23.440 A) consists of 1584
boron atoms and 24 yttrium atoms. Inside the cell the boron and yttrium atoms are
disorderly arranged (i.e. the structure is nearly amorphous). This complicated charac-
ter of the lattice causes the "glassy" behaviour of t%(T) in these compounds (Slack
et al. 1971, Cahill et al. 1989, Petrov et al. 1969, 1975) over the entire investigated
temperature interval (1-1000 K). Figures 127 and 128 show ~L(T) of YB66, YB61.7
and GdB66 (Cahill et al. 1989, Golikova and Tadzhiev 1986) [YB66 and GdB66 are
I I I
c.
10
1.0
10 -1
E
u 1
10 - 2
T3J/r',,ge
1.91 /
/~
10-3 /I /~z,
10-1
a.
I
2
I
[
10-4
,//
10-5 ,,/¢-
10-6
'/J
10-7 I I I I
10 100 1000 300 500 700 900 10-1 1 10 10 2 103
T,K T,K T,K
Fig. 127. (a) The temperature dependence of t% for YB66 (1) and GdB66 (2) (Golikova and Tadzhiev 1986,
Richards and Kasper 1969). Data on lcc of fused quartz are shown for comparison (3). The dashed lines
give the minimal thermal conductivity (at l = 2) for GdB66 (2') and fused quartz (3') (Petrov et al. 1969,
Smirnov et al. 1972). (b) EL for YB66 (1) and for amorphous boron (2) (Richards and Kasper 1969). (c) EL
for YB66 (1, 2), /?-boron (3) and fused quartz (4) (the low-temperature region) (Golikova and Tadzhiev
1986, Raychaudhuri et al. 1980).
I I I I I I
1.0
10-2 p-B
E
u
_7 10-.4
YB66
10 - 6 Fig. 128. The temperature dependence of the flee path length, l,
I
I I I I I I
in YB66 (Cahill et al. 1989). Data for the crystal /?-boron are
3 10 30 100 300 1000 shown for comparison (at 1 K l is approximately equal to the
T,K sample size).
semiconductors with P3oo ~ 10 4 ~ cm (Golikova and Tadzhiev 1986), therefore/(;tot :':
~:L]. At all temperatures ~cL(T) behaves as in classical insulating glasses (fused quartz,
amorphous boron).
The plateau in the dependence 1eL(T) of YB66 occurs at a higher temperature than
in the case of a classical insulating glass. This is connected with the large value of
O in YB66 (in YB66 O _~ 1340 K, in fused quartz O -~ 500 K). The higher the value
of O, the higher the temperature region of the plateau (Cahill et al. 1989).
The free path length of acoustical phonons in YB66 tends to a constant value
already at T>~ 80 K (fig. 128) (Slack et al. 1971). In addition to tcL(T), the "amor-
phous" character of YB66 has been observed in the temperature dependences of the
electrical resistance, the thermoemf and the heat capacity (Cahill et al. 1989, Golikova
and Tadzhiev 1986). Thus, in these compounds not only the phonon spectrum is
"amorphisised", but the electron spectrum as well. As has been noted above, the
"amorphous" behaviour of ~cL(T) in YB66 is observed over the wide temperature
interval 1-1000 K. The "amorphisation" is also observed in the behaviour of ~cL(T)
of a similar material, the tetrahedral crystal /~-B with 105 atoms in the elementary
cell (Petrov et al. 1969). However, in this case the amorphisation reveals itself only
at high enough temperatures. At low temperatures ~L(T) behaves just as in normal
crystalline solids (it has maximum at low temperatures). It is very likely that the
large number of atoms in the elementary cell of YB66 is the cause of the more
complete "amorphisation" of YB66 in comparison with/~-B. At high temperatures in
YB66 (as in classical insulating glasses) the condition l = 2 begins to be fulfilled [the
so-called case of minimal thermal conductivity (Slack (1979)] (see fig. t27a, dashed
lines).
9.2.2. "Amorphisation" of 1¢L due to the strong phonon scattering by lattice defects
An unusual situation arises in tanthanide compounds with an inhomogeneous
variable valency of the lanthanide ions with phonon scattering by crystal lattice
defects. We will consider as an example Sm3S 4. [In Sm3S4, P3o0 ~- 10~ cm. The
electronic contribution to tqot is insignificant over the whole temperature interval,
hence one can consider ~Ctot = xL (Smirnov et al. 1972)]. This compound contains the
ions Sm z+ and Sm 3+ in the ratio 1"2 ( S m Z + S m ~ + S 4 ) , statistically distributed
through the cation sublattice. Because of the appreciable difference in the ionic radii
of Sm z+ and Sm 3+ one of them can be considered as an "impurity" (let us call it
an A-type defect), scattering phonons and decreasing ~cL.The ions Sm 2 + and Sm 3 +
exchange electrons by hops, which change the valencies (hopping conductivity). With
increasing temperature the frequency of hops increases (but the ratio of Sm z + and
Sm 3 + concentration remains constant). At the valency variation the radii of the ions
change, deforming the lattice around them. The deformation cloud fluctuating in the
lattice from one ion to another ion (let us call it a type-B defect) is the cause of
additional scattering of phonons, decreasing the thermal conductivity. Both defects
of A- and B-type decrease the ~cL to values approaching the thermal conductivity of
a glass. At high temperatures, when the "concentration" of B-type defects increases
appreciably, strong phonon-defect scattering can lead to an "amorphous" temper-
ature dependence of tcL(T) (~cL = const). This is observed experimentally (fig. 129). [At
F - T - - r - r - r ~ - - I ~'--Y-
5
E 1,6B
~: 0,84
#
o 1 a I [ l i i I I I I l I_I__L__}~_L. Fig. 129. The temperature dependence of ~cc
200 400 600 800 for Sm3S 4 (T/irkes et al. 1986). Points show
T,K measurements for different experimental sets.
the highest temperatures, when l-~ 2, ~L(T) is probably influenced by the effect of
"minimal thermal conductivity" (Slack 1979).]
Ytterbium hydride YbHx (2 < x < 3) also belongs to the compounds with inhomo-
geneous variable valency of the lanthanide ions. In YbHx there are different ions,
Yb 2+ and Yb 3+ in the ratio 1:2 (Smirnov et al. 1992b). YbHx has a face centred
cubic lattice. % is small over the entire temperature range investigated, so/£tot = K:L"
but YbHx differs from Sm 3 S4 having besides the A- and B-type defects a rather large
concentration of vacancies in the hydrogen sublattice (let us call them C-type defects).
The concentration of hydrogen vacancies decreases as x varies from 2 to 3 (at x = 2
it is ~-33%, at x = 3 it is 0). Figure 130 shows the tCL(T) dependence for the two
compounds from the system YbH~ with x = 2.41 and 2.57. The vacancy (C-type
defect) concentrations are, respectively, ~ 2 0 and --~14% . One can see that the value
and the temperature dependence of XL(T) (at low temperatures ~:LOCT 1'4 1.6) is
similar to KL(T) of classical insulating glasses (see fig. 129). Thus, only the combined
influence of the three types of defects (A, B and C) results in the full "amorphisation"
of KL(T). When the concentration of C-type defects decreases, xL of YbHx increases
and tends toward ~:L of Sm3S4 (curve 4 in fig. 130).
Data on p(T) and cffT) of YbH~ compositions support the above conclusion of
"amorphisation" as well. There are many experiments showing the "amorphous"
character in the thermal conductivity behaviour in some temperature regions, but
we cannot discuss them because of space limitations on this review.
This section was devoted to the thermal conductivity of rare earth glasses with
5,0 l " -
IT - T - - - I
T19 //
,
3
I I I II
"-/ o,,~,
/ /o,, r. ,
20 I - .," 4' -... _
Fig. 130. The temperature dependence tcL for

0.3 YbHx with x = 2.41 (1), 2.57 (2) and 2.63 (3)
(Andres et al. 1984) (phonons are scattered
by defects of A, B and C type), for Sm3S 4 (4)
(T/irkes et al. 1986) (phonons are scattered
ol p I I'6 I I I I I I by defects of A and B type), for the glasses
3 5 lo 50 loo 2o0 300 La2S3(Ga203) 2 (5) and LaP5OI4 (6) (see
T,K fig. 125).
206 I.A. SMIRNOVand V.S. OSKOTSKI
different distortions of the crystal structure (so-called crystal inhomogeneities). In

section 4 the thermal conductivity of spin glasses with magnetic inhomogeneities has
been discussed.
10. Thermal conductivity of rare earth compounds at high temperatures
Up to now the discussion of the thermal conductivity of rare earth compounds

has been primarily devoted to low and superlow temperatures. The thermal conduc-
tivity of rare earth compounds at medium and high temperatures has been discussed
partly in the sections devoted to glasses (section 9) and to phonon scattering by
tanthanide ions (section 6). In this section we consider the behaviour of tc of these
materials at high temperatures, i.e. at T>~ O. Unfortunately, for many of these
experiments the measurements have not been exact enough, and the samples have
been of poor quality, because the main purpose of the measurements has been for
technical applications. For this reason it is necessary to carry out complex purposeful
investigations of the thermal conductivity on samples of high quality at high temper-
atures. Optical phonons can contribute to ~x. (or decrease ~cL due to additional
phonon scattering) in this temperature region. ~cL can vary due to a change in the
sample volume with temperature. "Amorphisation" of ~:c is also observed when one
reaches the conditions for "minimal thermal conductivity" (see section 9).
In the high-temperature region one can expect peculiarities of t% due to variation
of the Lorentz number near an electron zone boundary of a material (see section 2)
and to other scattering mechanisms of electrons. Peculiarities can arise in ~Ctotdue to
bipolar, photon and exciton contributions and possible heat transport by induced
vacancy diffusion. Known experimental data on K of rare earth compounds confirm
the regularities observed in other types of materials. However, many effects are
original, have their own distinctive features. We will separate these results into two
groups:
(1) experimental data, and
(2) interesting possible effects which have not yet been observed experimentally.
10.1. Phonon spectrum peculiarities and ~cL of rare earth compounds
Phonon spectra of compounds with rare earth ions with intermediate valency
exhibit anomalies (Mook and Holtzberg 1981, Mook and Nicklow 1978, Alekseev
et al. 1989, Kikoin and Mishchenko 1988) which are capable of affecting the behaviour
of ~cL(T) at medium and high temperatures, but, unfortunately, such experiments
have not yet been carried out. Let us consider, for example, features of the phonon
spectra of Smo.TsYo.25S (Mook and Holtzberg t981, Mook and Nicklow 1978) and
SmB6 (Alekseev et al. 1989), which are typical representatives of compounds with
intermediate valency (figs. 13 lb, 132b). In these compounds:
(1) the longitudinal acoustic (along [111] in both compounds) and optical (in
Smo.75Yo.25S) modes are strongly softened (compare spectra a and b in figs. 131 and
132) (in all known crystals LA-phonons have higher energy than TA-phonons near
LO
6
N
I
a~
LA
0
S Fig. 131. Phonon spectra (a) of semiconducting SmS and
7 L [- L (b) of Smo.75Yo.25 (Mook and Holtzberg 1981, Mook and
q Nicklow 1978, Kikoin and Mishchenko 1988).
~-" 6
' li . . . . .
I I I I lit I I I I
0 0.2 0.4 0 0.2 0.4 Fig. 132. P h o n o n spectra (a) of LaSb and (b) of StuB 6 (Alekseev
Phonon wave vector et al. 1989).
the zone centre. The LO mode of Smo.vsYo.25S has a lower energy than the TO
mode over the entire Brillouin zone (Mook and Nicklow 1978, Mook and Holtzberg
1981). This is rather unusual - such an effect has been observed only in Smo.75 Yo.25S
(Mook and Holtzberg 1981, Mook and Nicklow 1978) and YbS (Roedhammer et al.
1978).
(2) In both compounds a nondispersive mode arises between LA and LO modes
(figs. 131b and 132b). Alekseev et al. (1989) and Kikoin and Mishchenko (1988) have
suggested that this mode is typical for compounds with intermediate valency of the
lanthanide ions. Mook and Nicklow (1978), on the other hand, have contended that
this mode is connected with local vibrations of the Y ions.
The above noted features of the phonon spectra must manifest themselves in both
the magnitude and the temperature dependence of IcL. Let us hope that future
experiments will prove this supposition.
The hexaborides of the rare earth elements show interesting properties. RB 6
crystallizes in the CaB6 structure, where the cations and the B6-octahedrons form a
lattice of CsCl-type (fig. 133). In such a lattice two Debye temperatures are observed:
a low-temperature one (O1 -~ 150 K) connected with the "soft" metal sublattice and
a high-temperature one (O -~ 600 K) related to the "hard" boron sublattice (Paderno
et al. 1991, Schell et al. 1982, Smith et al. 1985, Takegahara and Kasuya 1985~
Peysson et al. 1986a, Klochkov and Shvarzman 1982, Korsukova et al. 1987). The
vibrations of the lattices depend upon each other. The phonon spectra of RB 6 have
some peculiarities (see, e.g., fig. 132) (Alekseev et al. 1989, Paderno et al. 1991, Schell
et al. 1982, Smith et al. 1985, Takegahara and Kasuya 1985, Peysson et al. 1986a).
The principal one is the nondispersive behaviour of the LA and TA modes at energies
higher than ~- 150 K. All this can be reflected in the ~cL(T) behaviour. Impurities and
defects can decrease ~cL significantly at low and medium temperatures, because they
are scattered, mainly, by long-wavelength phonons; and the short-wavelength acous-
tic phonons do not transport heat, since their group velocity ~o~/~q ~-0 (see figs. 131
and 132). This has been proven experimentally for StuB6 (Konovalova et al. 1984)
and for LaB6 (Neshpor et al. 1976). Figure 134 shows, as an example, data on t¢L of
single crystal and polycrystalline samples of SmB6. It is difficult to say now which
phonons are responsible for ~cL in RB6 at high temperatures (Konovalova et al. 1984,
Neshpor et al. 1976, Tanaka 1974, Markov et al. 1978, Aivazov et al. 1979, 1980,
Bryushkova et al. 1985). ~cL is rather high in RB 6 at high temperatures. For example,
~cL -~ 15 W/m K in LaB 6 at 1300 K (Tanaka 1974).
_ _ R
Q -- B Fig. 133. Crystal structure of R B 6.

150
. 100
E
50
Fig. 134. The temperature dependence of KL for single

5 l0 15 20 25 crystal (1) and polycrystalline (2) samples of SmB6
(Konovalova et al. 1984).
One could suppose a contribution of the optical phonons to KL at high

temperatures (Slack 1979, Devyatkova and Smirnov 1962), but this is not too likely.
It is possible that in RB 6 there are two types of p h o n o n spectra due to the presence
of two sublattices: the "low-temperature" spectrum (connected with metal atoms)
and the "high-temperature" spectrum (connected with boron atoms). The boron
sublattice vibrations give an insignificant contribution to KL at low temperatures.
However, this contribution becomes the main one at high temperatures. Unfortu-
nately, it is difficult to draw from the literature data a simple conclusion about the
influence of the different sublattices in RB 6 on KL, because measurements of the
thermal conductivity have been carried out often on poorly characterised samples
over narrow temperature intervals. Difficulties also arise in the separation of ~cL from
the experimental/£tot.
10.2. Phonon electron scattering at T>~ 0
Only phonons with q less than 2k (k is the absolute value of the electron wave
vector) can interact with electrons. In nondegenerate semiconductors the wave vector
of most electrons is much smaller than the wave vector of the thermal phonons at
not very low temperatures. Therefore, only the longest-wavelength phonons can
interact with electrons giving rise to a small contribution to the thermal conductivity.
The electron wave vectors at the Fermi level in metals are more likely, rather than
the wave vectors of phonons, to make a significant contribution to the heat transport.
These phonons can interact with electrons, and thus, an effect of phonon-electron
scattering is to be expected in metals at all temperatures. One fails, however, to
observe phonon-electron scattering in pure metals at high temperatures because of
the complexity of the ~ce separation. One can separate tG and ~cL and observe p h o n o n -
electron scattering in rare earth metals and metaMike compounds which contain
rare earth elements, due to the low mobility of electrons which increases p consider-
ably and hence decreases G ) (Oskotski and Smirnov 1971, Khusnutdinova et al.
1971, Luguev et al. 1975a).
Let us consider, as an example, KL(T) of the model system LaBTe4-La2Te3 (or
otherwise LaTex with 1.33 ~< x ~< 1.5) (Luguev et al. 1975) (this system has been
considered already to some extent in section 6). ~cL is separated from ~Ctot by the
standard method (Oskotski and Smirnov 1972, Luguev et al. 1975a). ICe of this system
presents a surprise again (as noted in section 6). One might expect a decrease of ~c
with variation of the composition from LaTe1.33 to LaTe1.5, due to an increase in

the number of vacancies in the La lattice. However, the opposite is observed (fig. 135).
One can explain the unusual behaviour of ~¢L(x) in this system only by considering
additional phonon-electron scattering. Figure 136 shows the dependence of the
thermal resistance of the crystal lattice, WL= 1/1¢L, on temperature for three composi-
tions of the LaTe x system. A calculation has been carried out on the basis of the
Callaway (1959) model. One can see that due to phonon-electron scattering ~cL of
LaTe1.33 (and nearby compositions) is smaller than tcL of LaTe1.5 (fig. 135).
We note that the minimum of 1eL (in isotherms of ~cL(x)) in the LaTe x system is
smoothed out as temperature increases and the phonon-electron scattering decreases
compared to other scattering mechanisms (Oskotski and Smirnov 1972, Luguev et al.
1975a).
A large contribution to WL from phonon-electron scattering is observed in LaS
and GdS up to ~ 7 0 0 K (Khusnutdinova et al. 1971).
2.52
E
1.68
0.84
I I I ~ I
1.33 1.41 1.50
1.37 1.45 Fig. 135. The temperature dependence of ~cL for the system LaTe x
X ,,~ (1.33 ~ x ~< 1.5) (Luguev et al. 1975a). T = 100 K.
I I I I
LaTe1.33
1.2
0,72
0.24 ~ /
"''-i I [ l
1.20 -LaTer47
0.72 3 ~ ' / ~ ~. "~" "~ ~
0.24 ~ ~ "'"
I I I I
1.68 LaTe1.38
7 V //~ Fig. 136. The temperature dependence of the lattice thermal

resistance, WL = 1/~L, in LaTea.33 , LaTe1.47 and LaTe1.38.
Solid lines are the result of calculation, points of experiment.
072
The phonon scattering processes considered in the calcula-
0.24 ~ / 1 tions are: (1) phonons by phonons, (2) phonons by phonons
I I I I plus phonons by electrons, (3) phonons by phonons plus
0 100 200 300 400 phonons by defects, (4) phonons by phonons plus phonons
T,K by electrons plus phonons by defects.
THERMAL CONDUCTIVITY OF RECs 21 !
10.3. tc~ of rare earth compounds at high temperatures
As has been noted in section 2, one can obtain from an analysis of data on ~cc
information about scattering mechanisms of electrons in a material, the electron
spectrum, and so on. Let us consider some examples of such an analysis in the high-
temperature region. Difficulties arise in the separation of ~cL and tee from ~Ctot for
metals and metal-like rare earth materials. At T > O one usually supposes L = L0
and calculates lee by the W i e d e m a n n - F r a n z law. However, because of a large ~c
value and a small t~L value both estimated contributions have large errors associated
with them. Over the last thirty years a great number of measurements at high
temperatures have been carried out on the hexaborides of rare earth elements and
their solid solutions (Neshpor et al. 1976, Tanaka 1974, M a r k o v et al. 1978, Aivazov
et al. 1979, 1980, L'vov et al. 1963, 1965, Samsonov et al. 1975, Samsonov 1964).
[The thermal conductivity of hexaborides has been measured at low temperatures
by K o n o v a l o v a et al. (1984), Flachbart et al. (1982a, b) and Peysson et al. (1985,
1986c).] The problem of separation of ~co and ~L arises in this case as well.
Bryushkova et al. (1985) have carried out an evaluation of the L/Lo ratio for the
hexaborides and their solid solutions for temperatures from 300 to 2000 K. The
classical method of separation of ic~ and ~cL by analysis of the dependence tqot = f ( a )
(or is the electrical conductivity, ~ = l/p) at constant temperature has been used
(Smirnov and Tamarchenko 1977). Such an evaluation has been completed for RB 6
(fig. 137) and the solid solutions LaxEut xB6 and L a x G d l _ x B 6 (fig. 138). One can
determine L/Lo of L a x G d l _xB6 in the temperature region 1000-2000 K using data
on t%t(T) (fig. 139) and p (fig. 140) (see also fig. 141). The ratio L/Lo is smaller than
one between 300 and 1500 K. This is probably connected with inelastic scattering
of carriers by optical vibrations of the crystal lattice. The resultant dependence
L(T)/Lo is typical for metals with a large O value (see section 2). Data of Neshpor
et al. (1976) seem to contradict the above ideas on L(T)/Lo.
More interesting results are obtained from an analysis of ~ce of rare earth com-
pounds carried out for the sake of determining the features of the electron zone
I I [
50
TbB6 { / ~ "
40
E
EuB6 //~CeB 6 B ~ ~ ~ 6
5 3o
R
IA//I VbBA \ L : _Xt_ot
2O ~ aT
~ GdB6
J" StuB6 I I I Fig. 137. ~;tot for R B 6 as a f u n c t i o n of G at
10
20 40 60 3 0 0 K ( B r y u s h k o v a et al. 1985). ~Ctot=
(7, 105 .C2-1 • rn-1 lcL + L~T.
212 I.A. S M I R N O V and V,S. O S K O T S K I
T l V-
4O
x:
E 30
y O 2 - -
10 I I I Fig. 138. ~:to~ as a function of o tbr LaxEu I ~B 6 (1)

0 20 z,O 6O
and La.,Gdl _~B~ (2) at T = 300 K (Bryushkova et al.
o , 1 0 5 f 2 - 1 . m -1 1985).
J I I I i t [ I I
50
4O
() 7
E
o 6
•- - - . a _ . . _ . . _ ~ _ _ o o • s
S
£' SO
x 4 ~ ¢ ~' x 3
x x x × x "~'~-- I/ 2
20
:~, I ~ j ~ ~ I i i i i I
300 ~ooo ~soo 2000
T,K
Fig. I39. The temperature dependenceof Kt,,t for LaxGdl xB6 at x = 0 (1), 0.1 (2), 0.3 (3), 0.5 (4), 0.7 (5),
0.9 (6), 1 (7) (Bryushkovaet al. 1985).
boundary in SmS (Shadrichev et al. 1976, Oskotski and Smirnov 1977, Sergeeva et al.
1972), in CesSe~, NdsS4 (Smirnov et al. 1972), in GdS, ErS (Vasil'ev et al. 1975) and
in LaSb (Goncharova et al. 1968). Let us consider data on the thermal conductivity
of semiconducting n-type SmS (Shadrichev et al. 1976, Oskotski and Smirnov 1977,
Zhuze et al. 1973, Smirnov 1972). Unfortunately, the data on ic of SmS and a number
of its solid solution alloys obtained at low temperatures (T < 100 K) (Benbachir et al.
1985), do not give any information on the material parameters.
SmS (crystal structure of NaCl-type) has a homogeneity region, approximately,
from 50 to 54 at% Sm, which does not extend to the S-side (Oskotski and Smirnov
1977, Zhuze et al. 1973, Smirnov 1972) where it changes from a semiconductor to a
semimetal (figs. 142a, b). With increasing temperature the Fermi level in SmS can
change its position and "scan" the conduction band. If the conduction band in SmS
is complex [which consists of s- and d-subbands (see Golubkov et al. 1973), and the
s-subband is lower than the d-subband], then the Lorentz number undergoes sharp
changes as the Fermi level approaches the bottom of the d-band. This can lead to
increasing as well as decreasing L / L o values (see section 2). Oskotski and Smirnov
l ] T 1 [ [
E 3 _ 7 -
?o
c2
I I I I
300 600 900 1200 1500 1800
T,K
Fig. 140. The temperature dependence of p for La~GdI _xB6 at various x. Designations (1) (7) see the
figure caption of fig. 139.
I I I I I I I I
o 1.0
-J
ol
0,6 81 82 Fig. 141. The temperature dependence of the ratio L/Lo
I.--I I.--..I O2 for the hexaborides of lanthanides (1) and the solid solu-
I I I I I I I I
500 1000 1500 2000 tion alloy La~Gdl -xB6 (2). 01 and 02 are the two Debye
T,K temperatures.
(1977) have carried out an analysis of ~ce(T) for two samples of SmS from the
homogeneity region (figs.142b, c). Experimental values of L (see fig. 143a) have been
calculated by using the equation
Loxp = (Ktot - I/W°)/~T~ (52)
where KTtot and a are measured experimentally, WL° is determined by extrapolation of

WL from the low-temperature region T < 300 K (fig. 142c) to the high-temperature
region. Curve 1 of fig. 142c is calculated for all temperatures by
wL = l / ( ~ t o t - L~T), (53)
where L is determined by standard formulae for the single-band model (taking into
account the electron gas degeneration) for the scattering of carriers by acoustic
vibrations of the crystal lattice (Oskotski and Smirnov 1972, 1977, Smirnov and
Tamarchenko 1977). According to Oskotski and Smirnov (1977) a deviation of the
WL values from W ° at T > 300 K can be attributed to neglecting the interband
scattering. The Lexp value (fig. 143a) can only be explained by interband scattering
0.72 . 1I ~ - ~ ~ 3 3 6.3
' 4JJ~ I ~ , . . . . . .,
I 2 _J ,o3 a.j °4
o i , 1,%1 o . f'jJ 'a ,o t 4

& 5~52so4a ~t y" ./" ~ ] I--7,,",/
o,o,osm ~ .~ ~0~o~-1-- /
/ ~ ' * J " z rq-r-; ,-r'~-r/

/l"J ~ ~ 3 54 50 46 42
0['~'~''Y ' ' ' ] ~" I ~ I i

0 200 400 600 800 1000
T,K
Fig. 142. The dependence on the Sm concentration of the electrical conductivity a and the current carrier
concentration n (a), ~% (b) and WL (c) in the homogeneity region of SmS, where 1 through 6 are the sample
numbers. (a) and (b) are the result of measurements at 300 K (Shadrichev et al. t976, Oskotski and
Smirnov 1977, Zhuze et al. 1973, Smirnov 1972). The interval G gives the homogeneity limits.
a. I s - band\ f ~ d-band
4.0
Jj \~|, / O.leV
3.0
2
2.0 1 'i~ 4f 6 . . . . . . . ~..k O.Ogev
0.23 eV b
1.0
~ I , , , 5m10'5 S
0
4 202 4 68
;ra
Fig. 143. (a) Dependence of L/L o on Y for the sample Sml.015S (Shadrichev et al. 1976, Oskotski and
Smirnov 1977, Zhuze et al. 1973). Curve 1: calculation for a model with the heavy-electron band lower
than the light-electron band (here the interband interaction does not influence L/Lo). Curves 2 and 3:
calculation for a model with the light-electron band lower than the heavy-electron band (2 without and
3 with the interband interaction taken into account). Curve 4: experimental data. (b) Schematic shape of
the conduction band of Sml.015S. Solid circles give impurity levels. Values of energy distances have been
obtained from analysis of data on ~ct (Shadrichev et al. 1976).
of electrons from the "light" lower s-band into the "heavy" upper d-band. The Fermi
level lies i n t h e d - b a n d . T h e t h e o r e t i c a l L/Lo r a t i o (fig. 143a) d e p e n d s u p o n t h e
p a r a m e t e r S (see s e c t i o n 2, eq. (17))• C u r v e 3 i n fig. 143a is c a l c u l a t e d w i t h S = 8. F o r
the experimental L/Lo v a l u e t o a g r e e w i t h t h e t h e o r e t i c a l o n e it is n e c e s s a r y t o t a k e
a l a r g e r S value.
10.4. Other effects
We will now consider some effects in the lc of rare earth compounds, some of
which have already been observed, others will probably be seen in later experiments
in the high-temperature region. A theoretical analysis of experimental data has
been carried out in the Callaway model for 1eL and GdS and LaS (crystal lattice of
the NaCl-type) in the temperature region 300 800 K taking into account the
contribution to the heat transport from longitudinal and transverse acoustic modes
(Dubey 1981, A1-Edani and Dubey 1986). In GdS and LaS, as in other compounds
with NaCl-structure (Singh and Verma 1972), heat is transported, mainly, by the
transverse acoustic phonons. Slack (1979) has shown that lee of solids at high
temperatures can significantly depend on a volume variation. This conclusion has
been proven experimentally many times in solid Ar, Ne, Kr and Xe [comparative
measurements have been carried out at constant pressure and constant volume
(Clayton and Batchelder 1973, Weston and Daniels 1984)]. Anomalous volume
variations have been observed in a number of rare earth compounds either at
phase transitions [-e.g., in SmS at a hydrostatic pressure of -~6.5 kbar at 300 K
(Smirnov and Oskotski 1978)] or over a wide enough temperature interval [e.g., in
Smo.83Gdo.lvS , where the lattice constant changes from 5.7 to 5.83~ in the
temperature interval 300-900 K (Jayaraman et al. 1975)]. Unfortunately, experi-
ments on the influence of a volume variation on ~ce in rare earth compounds have
not been carried out yet.
The photon contribution Kphot can give a significant contribution to ~C~o t in
semitransparent (or transparent) IR-radiation materials. Hence, it is necessary to
be convinced that Xphot is absent, before an analysis of the possible heat transport
contributions by vacancies, excitons and so on is made. This makes the conclusions
of Pillai and George (1988, 1991) connecting an additional thermal conductivity of
RCoO3 (R = La, Nd, Gd) in the temperature range 300-1100 K with the exciton
contribution, Gxc, somewhat tentative, because the tCphot contribution has not been
considered. The thermal conductivity of PrF3, LaF3, LaF 3 + Nd 3 ÷ single crystals
have been measured between 300 and 1400 K by Remizova (1975, 1978). These
materials are transparent in the IR part of the spectrum [e.g., LaF3 is fully
transparent between 0.13 and 13gm (Remizova 1978)]. An experimental and
theoretical evaluation of /£phot is given, t~phot gives the major contribution to h2tot
(e.g., at 1400 K /£phot is ---85% of ~Ctot in LaF3). However, an additional thermal
conductivity remains after subtraction of •phot and ~cL from tqot. Remizova (1975,
1978) explains its origin by the ionic conductivity of F - ions, which have a high
mobility due to a great number of defects (Sher et al. 1966). It seems more realistic
to explain the additional thermal conductivity in LaF 3 and P r F 3 by vacancy heat
transport (Zinov'ev and Masharov 1973, Strzhemechny and Kalnoy 1983), because
(as has been noted above) there are a great number of Schottky defects in these
compounds with the concentration increasing with temperature (Sher et al. 1966).
Heat transport by vacancies has been observed in the premelting region in solid
Kr and Ar by Bondarenko et al. (1982) and in Pt by Zinov'ev et al. (1968) and
Slack (1964).
l l . Conclusion
Investigation of the thermal conductivity of RECs has contributed much new

information, and thus extended our knowledge on the thermal conductivity of solids
as a separate scientific branch of solid state physics. This new information is obtained
because of the specific influence of the 4f-levels of the lanthanide ions on ~. The
limited size of this review does not allow us to discuss the K of a large group of
RECs introduced into technological applications over the last few years. These are:
medium and high-temperature thermoelectric materials, hydrogen absorbers, laser
materials, new materials for powerful tiny magnets, new luminescent materials, oxides,
coating materials, superlow-temperature materials and so on. The ~c of some of the
materials listed above has been considered in this review when discussing certain
thermal conductivity effects. Other topics not fully reviewed include numerous works
on the thermal conductivity of the high-temperature superconductors, and papers
devoted to the thermal conductivity of elementary rare earth metals.
In the process of writing this review it has become clear that the many-sided
theoretical interpretations and critical analyses of the published data demand a new
extensive review.
The investigation of the thermal conductivity of the rare earth compounds is still
far from completion. There are many interesting problems and mysteries awaiting
solutions.
References
Aivazov, M.I., T.I. Bryushkova, V.S. Mkrtchyan Andres, E.E., I.V. Volchok, A.I. Zvyagin and
and V.A. Rubanov, 1979, Tekh. Vys. Temp. 17, S.V. Startsev, 1984, Fiz. Nizk. Temp. 10, 398.
330. Andres, E.E., I.V. Volchok, A.I. Zvyagin,
Aivazov, M.I., S.B. Aleksandrovich, T.I. Bryush- V.B. Kokshenev and S.V. Startsev, 1985, Fiz.
kova, V.S. Mkrtchyan and V.A. Rubanov, 1980, Nizk. Temp. 11,769.
Tekh. Vys. Temp. 18, 429. Andres, K., J.E. Graebner and H.R Ott, 1975,
A1-Edani, M.C., and K.S. Dubey, 1986, Phys. Phys. Rev. Lett. 35, 1779.
Status Solidi b 137, 449. Arsen'ev, P.A., L.M. Kovba, Kh.S. Bagdasarov,
Alekseev, P.A., A.S. Ivanov, B. Dorner, H. Shober, B.F. Dzhurinski, A.V. Potemkin, B.I. Pokrov-
K.A. Kikoin, E.S. Konovalova, A.S. Mish- ski, F.M. Spiridonov, V.A. Antonov and
chenko, V.N. Lazukov, Y . B . Paderno, V.V. Ilyukhin, 1983, Rare Earth Compounds
A.Y. Rumyantsev and I.P. Sadikov, 1989, Euro- (Systems with Oxides of I I I ! Elements)
phys. Lett. 10, 457. (Nauka, Moscow) p. 278. In Russian.
Alekseevski, N.E., A.V. Mishin, A.S. Rudenko and Arsen'ev, P.A., V.B. Glushkova, A.A. Evdokimov,
A.A. Sorokin, 1986, Pis'ma Zh. Eksp. & Teor. E.K. Keller, V.B. Kravchenko, M.V. Krav-
Fiz. 43, 533. chinskaya, V.A. Krzhizhanovskaya, A.K. Kuz-
Amato, A., D. Jaccard, J. Sierro, P. Haen, P. Lejay netsov, Kh.M. Kurbanov, A.V. Potemkin,
and J. Flouquet, 1989, J. Low Temp. Phys. 77, P.A. Tikhonov and M.N. Tsnitlin, 1985, Rare
195. Earth Compounds (Zirconates, Hafnates, Nio-
Andersen, H.N., P.E. Gregers Hansen, E. Holm, bates, Tantalates, Antimonates) (Nauka, Mos-
H. Smith and O. Vogt, 1974, Phys. Rev. Lett. cow) p. 270. In Russian.
32, 1321. Arutyunyan, S.R., Kh.S. Bagdasarov,
Anderson, P.W., B.I. Halperin and C.M. Varma, A.P. Dodokin and A.M. Kevorkov, 1984,
1972, Philos. Mag. 25, 1. Kvant. Elektron. 11, 1284.
Arutyunyan, S.R., Kh.S. Bagdasarov, Berman, R., 1976, Thermal Conductivity in Solids
A.P. Dodokin and A.M. Kevorkov, 1986, Fiz. (Clarendon Press, Oxford).
Tverd. Tela 28, 957. Berth, S., H.R. Ott, F.N. Gydax, B. Hitti,
Arutyunyan, S.R., Kh.S. Bagdasarov, E. Lippelt, A. Schenck, C. Baines, B. van den
A.P. Dodokin and A.M. Kevorkov, 1987, Fiz. Brandt, T. Konter and S. Mango, 1987, Phys.
Tverd. Tela 29, 1874. Rev. Lett. 59, 2991.
Arzoumanian, C., A.M. de GoEr, B. Salce and Bhattacharjee, A.K., and B. Coqblin, 1982, in:
F. Holtzberg, 1983, J. Phys. (Paris) Lett. 44, Crystalline Electric Field Effect in f-Electron
L39. Magnetism (Plenum, New York) p. 1.
Arzoumanian, C., B. Salce, A.M. de GoEr and Bhattachatjee, A.K., and B. Coqblin, 1988, Phys.
F. Holtzberg, 1984, in: Phonon Scattering in Rev. B 38, 338.
Condensed Matter, eds W. Eisenwenger, K. Bhattacharjee, A.K., B. Coqblin, M. Raki,
Lal3man and S. D6ttinger (Springer, Berlin) L. Forro, C. Ayache and D. Schmitt, 1989,
p. 460. J. Phys. (Paris) 50, 2781.
Ayache, C., M. Raki, B. Salce, D. Schmitt, Bishop, D.J., C.M. Varma, B. Batlogg, E. Bucher,
A.K. Bhattacharjee and B. Coqblin, 1988, Z. Fisk and J.L. Smith, 1984, Phys. Rev. Lett.
J. Phys. Colloq. (Paris) 49, C8-791. 53, 1009.
Barbara, B., J.X. Boucherle, J.L. Buevoz, Blackman, M., 1935, Philos. Mag. 19, 989.
M.F. Rossignol and J. Schweizer, 1977, Solid Bloch, J.M., and D. Davidov, 1982, Phys. Rev. B
State Commun. 24, 48 i. 26, 3631.
Bardeen, J., G. Rickayzen and L. Tewordt, 1959, Bondar', I.A., N.V. Vinogradova and
Phys. Rev. 113, 982. L.N. Dem'yanets, Zh.A. Ezhova, V.V. Ilyukhin,
Barilo, S.N., S.P. Ges, L.E. Soshnikov and V.Yu. Kara Ushanov, L.N. Komissarova,
T.V. Smirnova, 1984, Phys. Status Solidi b 122, E.V. Lazarevski, B.N. Litvin, P.P. Mel'nikov,
47. D.A. Murashov, V.M. Orlovski, K.K. Palkina,
Bauer, E., E. Gratz and G. Adam, 1985a, in: Proc. M.A. Petrova, I.A. Rozanov, N.N. Chudinova
Conf. on Electronic Structure and Properties and A.A. Fotier, 1983, Rare Earth Compounds
of Rare Earths and Actinide Intermetallics, eds (Silicates, Germanates, Phosphates, Arsenates,
G. Helscher, G. Wiesinger, E. Gratz, C. Vanadates) (Nauka, Moscow) p. 284. In
Schmitzer, P. Weinberger and R. Gr6sssinger Russian.
(North-Holland, Amsterdam) pp. 81-83. Bondarenko, A.I., V.G. Manzhely, V.A. Popov,
Bauer, E., E. Gratz, H. Kirchmayr, N. Pillmayr
M.A. Strzhemechny and V.G. Gavrilko, 1982,
and H. Novotny, 1985b, J. Less-Common Met.
Fiz. Nizk. Temp. 8, 1215.
111,369.
Brfick, E., A. Nowack, N. Hohn, H. Paulus and
Bauer, E., E. Gratz and G. Adam, 1986, J. Phys.
A. Freimuth, 1986, Z. Phys. B 63, 1.
F 16, 493.
Bryushkova, T.I., B.A. Merisov and
Bauer, E., E. Gratz and Y. Peysson, 1987,
V.S. Mkrtchyan, 1985, Tekh. Vys. Temp 23,
J. Magn. & Magn. Mater. 63-64, 303.
Belitsky, V.I., and A.V. Goltsev, 1991, Physica B 1016.
172, 459. Buchelnikov, V.D., Yu.A. Kuzavko and
Belov, K.P., 1972, Ferrits in High Magnetic Fields V.G. Shavrov, 1987, Fiz. Nizk. Temp. 13, 1074.
(Nauka, Moscow). In Russian. Bucher, E., J.P. Maita and A.S. Cooper, 1972,
Belov, K.P., A.K. Zvezdin, A.M. Kadomtseva and Phys. Rev. B 6, 2709.
R.Z. Levitin, 1976, Usp. Fiz. Nauk 119, 447. Buzdin, A.I., and V.V. Moshchalkov, 1986, Exotic
Belov, K.P., A.K. Zvezdin, A.M. Kadomtseva and Superconductors (Znanie, Moscow) p. 64. In
R.Z. Levitin, 1979, Orientational Transitions in Russian.
Rare Earth Magnetics (Nauka, Moscow). In Buzdin, A.I., L.N. Bulaevski, M.L. Kulich and
Russian. S.V. Panyukov, 1984, Usp. Fiz. Nauk 144,
Benbachir, K., J. Mazucr and J.P. Senateur, 1985, 597.
Solid State Commun. 54, 965. Cahill, D.G., H.E. Fischer, S.K. Watson,
Berezin, A.A., O.A. Golikova, M.M. Kazanin, R.O. Poht and G.A. Slack, 1989, Phys. Rev. B
T. Khomidov, D.N. Mirlin, A.V. Petrov, 40, 3254.
U.S. Umarov and V.K. Zaitsev, 1974, J. Non- Callaway, J., 1959, Phys. Rev. 113, 1046.
Cryst. Solids 16, 237. Charap, S.N., 1964, Phys. Rev. Lett. 13, 237.
Clark, W.G., Z. Fisk, K. Glover, M.D. Lan, L.S. Parfen'eva and I.A. Smirnov, 1978, Fiz.
D.E. Mac Laughlin, J.L. Smith and C. Tien, Tverd. Tela 20, 297.
1984, in: Proc. LT-17 Conference (North- Effantin, J.M., J. Rossat-Mignod, P. Burlet,
Holland, Amsterdam) p. 227. H. Bartholin, S. Kunii and T. Kasuya, 1985,
Clayton, F., and D.N. Batchelder, 1973, J. Phys. J. Magn. & Magn. Mater. 47-48, 145.
C 6, 1213. Elliot, R.J., and J.B. Parkinson, 1967, Proc. Phys.
Cooke, A.J., S.J. Swithenby and M.R. Wells, 1972, Soc. 92, 1024.
Solid State Commun. 10, 265. Fertig, W.A., D.C. Johnston, L.E. de Long,
Coornut, B., and B. Coqblin, 1972, Phys. Rev. B R.W. McCallum, M.P. Maple and
5, 4541. B.T. Matthias, 1977, Phys. Rev. Lett. 38, 987.
Cox, D.L., and N. Grewe, 1988, Z. Phys. B 71, Fischer, H.E., S.K. Watson and D.G. Cahill, 1988,
321. Comments Condens. Matter. Phys. 14, 65.
Daudin, B., and B. Salce, 1982, J. Phys. C 15, 463. Fischer, K.H., 1971, Phys. Status Solidi b 46, 11.
Daudin, B., A.M. de G6er and S.H. Smith, 1980, Fischer, K.H., 1983, Phys. Status Solidi b 116,
in: Phonon Scattering in Condensed Matter, 357.
ed. H.J. Maris (Plenum, New York) p. 93. Fischer, K.H., 1985, Phys. Status Solidi h 130, 13.
Daudin, B., R. Lagnier and B. Salce, 1981a, Fischer, O., A. Treyvaud, R. Chevrel and
J. Magn. & Magn. Mater. 25, 197. W. Sergent, 1975, Solid State Commun. 17, 721.
Daudin, B., B. Salce and S.H. Smith, 1981b, Fischer, O., M. Ishikawa, M. Pelizzone and
J. Phys. Colloq. (France) 42, Suppl. 12, C6-277. A. Treyvaud, 1979, J. Phys. Colloq. (Paris) 40,
Daudin, B., W. Mutscheller and M. Wagner, 1984, C5-89.
Phys. Status Solidi b 122, 73. Flachbart, K., M. Reiffers, S. Janos, Y.B. Paderno,
David, G., D.G. Cahill and R.O. Pohl, 1987, Phys. V.I. Azorenko and E.S. Konovalova, 1982a,
Rev. B 35, 4067. J. Less-Common Met. 88, L3.
De GoSr, A.M., 1986, in: Phonon Scattering in Flachbart, K., M. Reiffers, S. Janos, Y.B. Paderno,
Condensed Matter, Vol. V, eds E.A. Anderson V.I. Azorenko and E.S. Konovalova, 1982b,
and J.P. Wolfe (Springer, Berlin) pp. 6 14. J. Less-Common Met. 88, Lll.
De Go,r, A.M., N. Devismes and B. Salce, 1986, Florent'ev, V.V., A.V. Inyushkin, A.N. Taldenkov,
in: Phonon Scattering in Condensed Matter, O.K. Mel'nikov and A.B. Bykov, 1990, Sverkh-
Vol. V, eds A.C. Anderson and J.P. Wolfe provodirnost Fiz. Chim. Tekh. 3, 2302.
(Springer, Berlin) pp. 70-72. Flouquet, J., P. Haen, C. Marcenat, P. Lejay,
De Visser, A., A. Menovsky and J.J.M. Franse, A. Amato, D. Jaccard and E. Walker, 1985,
1987, Physica B 147, 80. J. Magn. & Magn. Mater. 52, 85.
De Yoreo, J.J., R.O. Pohl and G. Burns, 1985, Flouquet, J., P. Haen, F. Lapierre, D. Jaccard and
Phys. Rev. B 32, 5780. G. Remenyl, 1986, J. Magn. & Magn. Mater.
Dekker, AJ., 1965, J. Appl. Phys. 36, 906. 54-57, 322.
Devyatkova, E.D., and 1.A. Smirnov, 1962, Fiz. Franse, JJ.M., P.M. Frings, A. de Visser,
Tverd. Tela 4, 2507. A. Menovski, T.T.M. Palstra, P.M. Kes and
Dixon, G.S., 1976, in: Phonon Scattering in Sol- J.A. Mydosh, 1984, Physica B 126, 116.
ids, eds L.J. Challis, V.W. Rampton and A.F.G. Franse, J.J.M., A. Menovsky, A. de Visser,
Wyatt (Plenum, New York) pp. 246-248. C.D. Bredl, U. Gottwick, W. Lieke,
Dixon, G.S., 1981, J. Phys. C 14, 2147. H.M. Mayer, U. Rauchschwalbe, G. Sparn and
Dixon, G.S., and D.P. Landau, 1976, Phys. Rev. F. Steglich, 1985, Z. Phys. B 59, 15.
B 13, 3121. Franz, W., A. Griegel, F. Steglich and
Dixon, G.S., J.J. Martin and N.D. Love, 1974, D. Wohlleben, 1978, Z. Phys. B 31, 7.
AIP Conf. Proc. 18, 1073. Franz, W., F. Steglich and D. Wohlleben, 1979,
Dreyfus, B., N.C. Fernandes and R. Maynard, J. Phys. Colloq. (Paris) C5-40, C5-342.
1968, Phys. Rev. Lett. A 26, 647. Freeman, JJ., and A.C. Anderson, 1986, in: Pho-
Dubey, K.S., 1981, J. Thermal Anal. 20, 447. non Scattering in Condensed Matter, Vol. V,
Ducastel, R., and G. Pitsi, 1986, Phys. Status eds A.C. Anderson and J.P. Wolfe (Springer,
Solidi b 134, 799. Berlin) pp. 32-34.
Durand, J., and SJ. Poon, 1979, J. Phys. Colloq. Friederich, A., and A. Fert, 1974, Phys. Rev. Lett.
(Paris) C5-231. 33, 1214.
Dzhabbarov, I., S.I. Ivanov, V.V. Medved, Fulde, P., and I. Peschel, 1972, Adv. Phys. 21, 1.
Galperin, Y.M., V.G. Karpov and V.N. Solov'ev, Gratz, E., E. Bauer, H. Novotny, H. Mfiller,
1988, Fiz. Tverd. Tela 30, 3636. S. Zemirli and B. Barbara, 1987, J. Magn. &
Gehring, G.A., and K.A. Gehring, 1975, Rep. Magn. Mater. 63-64, 200.
Prog. Phys. 38, 1. Gratz, E., N. Pitlmayr, E. Bauer, H. Mfiller,
Geilikman, B.T., 1958, Zh. Eksp. Teor. Fiz. 34, B. Barbara and M. Loewenhaupt, 1990, J. Phys.
1042. Condens. Matter 2, 1485.
Geilikman, B.T., and B.Z. Kresin, 1958, Dokl. Gudak, O., 1985, Pis'ma Zh. Eksp. & Teor. Fiz.
Akad. Nauk SSSR 123, 259. 42, 244.
Geilikman, B.T., and B.Z. Kresin, 1959, Zh. Eksp. Guertin, R.P., and S. Foner, 1978, Phys. Lett. A
& Teor. Fiz. 36, 959. 68, 116.
Geilikman, B.T., and B.Z. Kresin, 1972, Kinetic Guessons, A., and J. Mazuer, 1982, J. Phys. (Paris)
and Nonstationar Effects in Superconductors Lett. 43, L801.
(Nauka, Moscow). In Russian. Guittard, M., A.M. Loireau-Lorach, F. Berguer,
Ginzburg, V.L., 1957, Sov. Phys.-JETP 4, 153. S. Barnier and J. Flahaut, 1986, Mater. Res.
Ginzburg, V.L., 1989a, Pis'ma Zh. Eksp. & Teor. Bull. 62, 191.
Fiz. 49, 50. Gurevich, V.L., 1959, Fiz. Tverd. Tela 1, 147.
Ginzburg, V.L., 1989b, J. Supercond. 2, 323. Hamaker, H.C., L.D. Woolf, H.B. Mac Kay,
Golikova, O.A., and A. Tadzhiev, 1986, J. Non- Z. Fisk and M.B. Maple, 1979, Solid State
Cryst. Solids 87, 64. Commun. 32, 289.
Golikova, O.A,, V.K. Zaitsev, N.M. Orlov, Hamaker, H.C., H.B. Mac Kay, M.S. Torikachvili,
A.V. Petrov, L.S. Stil'bans and E.N. Tkalenko, L.D. Woolf and M.B. Maple, 1981a, J. Low
1974, Phys. Status Solidi a 21, 405. Temp. Phys. 44, 553.
Golovenchits, E.I., and V.A. Sanina, 1982, Fiz. Hamaker, H.C., H.B. Mac Kay, L.D. Woolf,
Tverd. Tela 24, 375. M.B. Maple, W. Odoni and H.R. Ott, 1981b,
Golovenchits, E.I., V.A. Sanina and Phys. Lett. A 81, 91.
T.A. Shaplygina, 1981, Zh. Eksp. & Teor. Fiz. Harley, R.T., W. Hayes, A.M. Perry and
80, 1911. S.P.R. Smith, 1973, J. Phys. C 6, 2382.
Golubkov, A.V., E.V. Goncharova, V.P. Zhuze, Herlach, D.M., E.F. Wassermann and J. K~istner,
G.M. Loginov, V.M. Sergeeva and 1981, Phys. Rev. B 24, 204.
I.A. Smirnov, 1973, Physical Properties of Rare Herlach, D.M., E.F. Wassermann and
Earth Chalcogenides (Nauka, Leningrad). In R. Willnecker, 1983, Phys. Rev. Lett. g0, 529.
Russian. Holland, M.G., 1963, Phys. Rev. 132, 2461.
Goncharova, E.V., V.P. Zhuze, V.V. Zhdanova, Iolin, E.M., 1970, Fiz. Tverd. Tela 12, 1159.
T.B. Zhukova, I.A. Smirnov and Ishikawa, A., and O. Fischer, 1977, Solid State
E.V. Shadrichev, 1968, Fiz. Tverd. Tela 10, Commun. 24, 747.
1322. Jaccard, D., and J. Flouquet, 1985, J. Magn. &
Gratz, E., 1982, Phys. Scr. T 1, 97. Magn. Mater. 47-48, 45.
Gratz, E., and H. Novotny, 1983, in: Proc. Int. Jaccard, D., and J. Flouquet, 1987, Helv. Phys.
Conf. on Magnetism of Rare Earths and Actin- Acta 60, 108.
ides, Bucharest, 1983, eds. E. Burzo and H. Jaccard, D., J. Flouquet, Z. Fisk, J i . Smith and
Rogalski, p. 17. H.R. Ott, 1985a, J. Phys. (Paris) Lett. 46, LSll.
Gratz, E., and H. Novotny, 1985, Physica B 130, Jaccard, D., J. Flouquet, P. Lejay and
75. J.L. Tholence, 1985b, J. Appl. Phys. 57, 3082.
Gratz, E., and M.J. Zuckermann, 1982a, J. Magn. Jagannathan, A., R. Orbach and O. Entin-Wohl-
& Magn. Mater. 29, 181. man, 1989, Phys. Rev. B 39, 13465.
Gratz, E., and M.J. Zuckermann, 1982b, Trans- Jahn, H.A., and E. Teller, 1937, Proc. Roy. Soc.
port Properties of Rare Earth Intermetallic London A 161, 220.
Compounds, in: Handbook on the Physics and Jayaraman, A., P.D. Dernier and L.D. Longinotti,
Chemistry of Rare Earths, Vol. 5, eds K.A. 1975, High Temp.-High Pressures 7, 1.
Gschneidner Jr and L. Eyring (North-Holland, Jezowski, A., and J. Klamut, 1990, in: Studies of
Amsterdam) pp. 117-216. High Temperature Superconductors, ed. A.
Gratz, E., E. Bauer, B. Barbara, S. Zemirli, Norwkar (Nove Science PUN., New York)
F. Steglich, C.D. Bredl and W. Lieke, 1985, pp. 263-309.
J. Phys. F 15, 1975. Kamarzin, A.A., A.A. Mamedov, V.A. Smirnov,
A.A. Sobol, V.V. Sokolov and Krasny, V.P., N.P. Kovalenko and V.V. Mikho,
I.A. Shcherbakov, 1983a, Kvant. Electron. 10, 1990, Physica B 162, 122.
1560. Lecomte, G.V., H. v. L6hneysen and W. Zinn,
Kamarzin, A.A., A.A. Mamedov, V.A. Smirnov 1983, J. Magn. & Magn. Mater. 38, 235.
and V.V. Sokolov, 1983b, Fiz. Tverd. Tela 25, Lecomte, G.V., H. v. L6hneysen and W. Zinn,
1664. 1984, in: Phonon Scattering in Condensed
Kane, E.O., 1957, J. Phys. & Chem. Solids 1, 249. Matter, eds W. Eisewenger, K. Lagman and S.
Karpov, V.G., and D.A. Parshin, 1983, Pis'ma Zh. D6ttinger (Springer, Berlin) p. 466.
Eksp. & Teor. Fiz. 38, 536 [JETP Lett. 38, 648~]. Lecomte, G.V., H. v L6hneysen, J. Wosnitza and
Karpov, V.G., and D.A. Parshin, 1985, Zh. Eksp. W. Zinn, 1986, in: Phonon Scattering in Con-
& Teor. Fiz. 88, 2212 [Soy. Phys.-JETP, 61, densed Matter, eds A.C. Anderson and J.P.
13081. Wolfe (Springer, Berlin) p. 251.
Kazey, Z.A., and V.I. Sokolov, 1986, Zh. Eksp. & Leibfried, G., and E. Schl6mann, 1954, Nachr.
Teor. Fiz. 91, 1799. Akad. Wiss. G6ttingen A 2, 71.
Khusnutdinova, V.Y., V.S. Oskotski, Leroux-Huqou, P., and J.J. Veyssie, 1965, Phys.
I.A. Smirnov and V.M. Sergeeva, 1971, Phys. Status Solidi 8, 561.
Status Solidi b 48, 353. Lockwood, A., C. Wood, J. Vandersande,
Kikoin, K.A., and A.S. Mishchenko, 1988, Zh. A. Zoltan, L. Danielson, V. Rawag and
Eksp. & Teor. Fiz. 94, 237. D. Wittenberg, 1986, J. Less-Common Met.
Kirichenko, Y . A . , K.V. Rnsanov and 126, 113.
E.G. Tyurina, 1990, Sverkhprovodimost Fiz. L6hneysen, H.v., D.M. Herlach, E.F. Wasser-
Khim. Tekh. 3, 1385. mann and K. Samwer, 1981, Solid State Com-
Kittel, C., 1958, Phys. Rev. 110, 836. mun. 39, 591.
Klcmens, P.G., 1969, Thermal Conductivity (Aca- L6hneysen, H.v., R. van den Berg, J. Wosnitza,
demic Press, London). G.V. Lecomte and W. Zinn, 1986, J. Magn. &
Klochkov, L.A., and E.I. Shvarzman, 1982, in: Magn. Mater. 54-57, 189.
High Temperature Borides and Silicides (IPM Loireau-Lorach, A.M., M. Guittard and
AN USSR, Kiev) pp. 82 87. In Russian. J. Flahaut, 1976, Mater. Res. Bull. 11, 1489.
Knoll, K.D., 1971, Phys. Status Solidi b 45, 553. Lopez de la Torre, M.A., S. Vieira, R. Villar,
M.B. Maple and M.S. Torikachviti, 1988, Phys-
Kokshencv, V.B., 1985, Phys. Status Solidi b 132,
ica C 153-155, 457.
375.
Love, M.S., and A.C. Anderson, 1990, Phys. Rev.
Kolomoets, N.V., 1966, Fiz. Tverd. Tela 8, 999.
42, 1845.
Komissarova, L.N., V.M. Shatski and G.Ya.
Luguev, S.M., V.S. Oskotski, L.N. Vasil'ev,
Pushikina, L.G. Shchorbakova, L.G. Mam-
V.N. Bystrova, T.I. Komarova and
surova and G.E. Sukhanova, 1984, Rare Earth
I.A. Smirnov, 1975a, Fiz. Tverd. Tela 17, 2941.
Compounds (Carbonates, Oxalates, Nitrates, Luguev, S.M., V.S. Oskotski, V.M. Sergeeva and
Titanates) (Nauka, Moscow) p. 234. In Russian. I.A. Smirnov, 1975b, Fiz. Tverd. Tela 17, 2697.
Konovalova, E.S., Y.B. Paderno, N.I. Perepelitsa, Luguev, S.M., T.I. Komarova, V.N. Bystrova and
E.M. Dudnik, S. Janos, K. Flachbart and I.A. Smirnov, I978, Izv. Akad. Nauk SSSR
M. Reiffers, 1984, Fiz. Tverd. Tela 26, 2138. Neorg. Mater. 14, 46.
Korenblit, I.Y., and E.F. Shender, 1984, lzv. Lfithi, B., M.E. Mullen, K. Andres, E. Bucher and
Vyssh. Uchebn. Zaved. Fiz. 27(10), 23. J.P. Maita, 1973, Phys. Rev. B 8, 2639.
Korsukova, M.M., L. Lundstrom and L'vov, S.N., V.F. Nemchenko and Y.B. Paderno,
L.E. Tergenius, 1987, Solid State Commun. 63, 1963, Dokl. Akad. Nauk SSSR 149, 1371.
187. L'vov, S.N., V.F. Nemchenko and Y.B. Paderno,
Kost, M.E., A.L. Shilov and V.N. Mikheeva, 1965, in: High Temperature Inorganic Com-
S.I. Uspenskaya, V.I. Novokshonov, pounds (in Russian) (Naukova Dumka, Kiev)
K.E. Mironov, M.N. Abdusalyamoba, pp. 445-450.
A.A. Eliseev, G.M. Kus'micheva and MacLaughlin, D.E., C. Tien, L.C. Gupta,
G.B. Seifer, 1983, Rare Earth Compounds A. Aarts, F.R. de Boer and Z. Fisk, 1984, Phys.
(Hydrides, Borides, Carbides, Phosphides, Rev. B 30, 1577.
Pnictides, Chalcogenides, Pseudochalcogen- Maletta, H., 1982, J. Appl. Phys. 53, 2185.
ides) (Nauka, Moscow) p. 268. in Russian. Maletta, H., and W. Zinn, 1989, Spin glasses, in:
Handbook on the Physics and Chemistry of Mook, H.A., and F. Holtzberg, 1981, in: Valence
Rare Earths, Vol. 12, eds K.A. Gschneidner Jr Fluctuations in Solids, eds L.M. Ealikov, W.
and L. Eyring (North-Holland, Amsterdam) Hanke and M.B. Maple (North-Holland,
pp. 213-356. Amsterdam) p. 113.
Marcenat, C., D. Jaccard, J. Sierro, J. Flouquet, Mook, H.A., and R.M. Nicklow, 1978, Phys. Rev.
J. Onuki and T. Komatsubara, 1990, Physica B 18, 2925.
B 163, 147. Mori, K., Y. Ishikava and K. Sato, 1984, J. Phys.
Markov, Yu.M., G.N. Trokhina, E.E. Zernova Soc. Jpn. 53, 664.
and V.V. Maksimovski, 1978, Izv. Akad. Nauk Morton, I.P., and H.M. Rosenberg, 1962, Phys.
SSSR Neorg. Mater. 14, 79. Rev. Lett. 8, 200.
Martin, J.J, and G.S. Dixon, t972, Phys. Status Mfiller, H., E. Hegenbarth, W. Matz, E. Mrosau
Solidi b 54, 707.
and A. Schmeltzer, 1982, in: Crystalline Electric
Matey, J.R., and A.C. Anderson, 1977a, Phys.
Field Effects in f-Electron Magnetism, eds R.P.
Rev. B 16, 3406.
Guertin, W. Suski and Z. Zotnierek (Plenum,
Matey, J.R., and A.C. Anderson, 1977b, J. Non-
New York) pp. 98-93.
Cryst. Solids 23, 129.
M/iller, H., E. Hegenbarth and S. Mahmoud,
Matlak, S., and B. Nolting, 1983, Solid State
Commun. 47, 11. 1983, Phys. Status Solidi b 115, K69.
Matthias, B.T., H. Suhl and E. Gorenzwit, 1958, Mutscheller, W., and M. Wagner, 1984a, in: Pho-
Phys. Rev. Lett. 1,449. non Scattering in Condensed Matter, eds W.
Matthias, B.T., H. Suhl and E. Gorenzwit, 1959, Eisenmenger, K. Lassman and S. D6ttinger
Phys. Rev. Lett. 2, 92. (Springer, Berlin) p. 298.
Matthias, B.T., E. Gorenzwit, J.M. Vandenberg Mutscheller, W., and M. Wagner, 1984b, Solid
and H. Barz, 1977, Proc. Nat. Acad. Sci. US State Commun. 50, 577.
74, 1334. Mutscheller, W., and M. Wagner, 1986, Phys.
Matz, W., B. Lippold, E.A. Goremychkin, Status Solidi b 134, 39.
A. Andreeff, H. Greissman and T. Frauenheim, Mutscheller, W., and M. Wagner, 1987, Phys.
1982, in: Crystalline Electric Field Effects in Status Solidi b 144, 507.
f-Electron Magnetism, eds R.P. Guertin, W. Mydosh, J.A., 1978, J. Magn. & Magn. Mater. 7,
Suski and Z. Zotnierek (Plenum, New York) 237.
pp. 69-82. Nakamura, A., N. Kawakami and A. Okiji, 1987,
McCallum, R.W., D.C. Johnston, R.N. Shelton, Phys. Lett. A 120, 241.
W.A. Fertig and M.B. Maple, 1977a, Solid State Neelmani, and G.S. Verma, 1972, Phys. Rev. B 6,
Commun. 24, 501. 3509.
McCallum, R.W., D.C. Johnston, R.N. Shelton Neshpor, V.S., B.A. Fridlender, V.N. Paderno,
and M.B. Maple, 1977b, Solid State Commun. Y.B. Paderno and M.A. Nerus, 1976, Techn.
24, 391. Vys. Temp. 14, 903.
McClintok, P.V.E., and H.M. Rosenberg, 1968, Novikov, V.I., V.P. Zhuze, V.M. Sergeeva and
Proc. Roy. Soc. London A 302, 419. S.S. Shalyt, 1975, Zh. Eksp. & Teor. Fiz. 68,
McClintok, P.V.E., I.P. Morton, R. Orbach and
2159.
H.M. Rosenberg, 1967, Proc. Roy. Soc. London
Odoni, W., and H.R. Ott, 1979, Phys. Rev. Lett.
A 298, 359.
A 70, 480.
McCollum, D.C., R.L. Wild and J. Callaway,
1964, Phys. Rev. A 136, 426. Odoni, W., G. Keller and H.R. Ott, 1981, Physica
Melcher, R.L., E. Putte and B.A. Scott, 1973, B 108, 1227.
Phys. Rev. Lett. 31, 307. Olsen, J.L., and H.M. Rosenberg, 1953, Adv.
Metcalfe, M.J., and H.M. Rosenberg, 1979, Phys. Phys. 2, 28.
Lett. A 33, 211. Onuki, Y., Y. Shimizu, T. Komatsubara,
Mitarov, R.G., V.V. Tikhonov, L.N. Vasil'ev, A. Sumiyama, Y. Oda, H. Hagano, T. Fujita,
V.S. Oskotski, A.V. Golubkov and Y. Maeno, K. Satoh and T. Ohtsuka, 1985a,
I.A. Smirnov, 1975, Fiz. Tverd. Tela 17, 496. J. Magn. & Magn. Mater. 52, 344.
Moeser, J.N., and F. Steglich, 1975, Z. Phys. B Onuki, Y., Y. Shimizu, M. Nishihara, Y. Mashii
21, 165. and T. Korrlatsubara, 1985b, J. Phys. Soe. Jpn.
Moncton, D.E., G. Shirane, W. Thomlinson, 54, 1964.
M. Ishikawa and O. Fischer, 1978, Phys. Rev. Orbach, R., 1960, Philos. Mag. 5, 1303.
Lctt. 41, 1133. Orbach, R., 1962, Phys. Rev. Lett. 8, 393.
Oskotski, V.S., and I.A. Smirnov, 1971, Fiz. Solids, eds L.J. Challis, V.W. Rampton and
Tverd. Tela 13, 1635. A.F.G. Wyatt (Plenum, New York)
Oskotski, V.S., and I.A. Smirnov, 1972, Crystal pp. 240-242.
Defects and Thermal Conductivity (Nauka, Parsons, M.W.S., and H.M. Rosenberg, 1972, in:
Leningrad). In Russian. Proc. Int. Conf. on Phonon Scattering in Solids,
Oskotski, V.S., and I.A. Smirnov, 1977, Semicon- ed. H.F. Albany (CERN Saclay, Paris)
ductor-metal phase transition in samarium pp. 326-329.
monosulphate, in: Rare Earth Semiconductors, Patterson, J.D., 1978, Phys. Status Solidi b 90,
eds V.P. Zhuze and I.A. Smirnov (Nauka, K85.
Leningrad) pp. 105 145. In Russian. Petrov, A.V., M.S. Germaidze, O.A. Golikova,
Oskotski, V.S., L.S. Parfen'eva and I.A. Smirnov, A.Y. Kiskachi and V.N. Matveev, 1969, Fiz.
1972, in: Proc. Int. Conf. on Phonon Scattering Tverd. Tela 11, 907.
in Solids, ed. H.J. Albany, Paris, pp. 254-259. Petrov, A.V., V.M. Orlov, V.K. Zaitsev and
Oskotski, V.S., L.N. Vasil'ev and I.A. Smirnov, V.A. Feigelman, 1975, Fiz. Tverd. Tela 17, 3703.
1982, in: Synthesis and Properties of Rare Petrunin, G.I., V.G. Popov and M.I. Timo-
Earth Compounds (AN USSR, Sverdlovsk) shechkin, 1989, Fiz. Tverd. Tela 31(7), 139.
pp. 86-94. In Russian. Peysson, Y., C. Ayache, B. Salce, J. Rossat-Mig-
Ott, H.R., and B. Lfithi, 1976, Phys. Rev. Lett. 36, nod, S. Kunii and T. Kasuya, 1985, J. Magn.
600. & Magn. Mater. 47-48, 63.
Ott, H.R., W.A. Fertig, D.C. Johnston, Peysson, Y., B. Daudin, M. Dubus and
M.B. Maple and B.T. Matthias, 1978, J. Low R.E. Benenson, 1986a, Phys. Rev. B 34, 8367.
Temp. Phys. 33, 159. Peysson, Y., B. Salce, C. Ayache and E. Bauer,
Ott, H.R., W. Odoni, H.C. Hamaker and 1986b, J. Magn. & Magn. Mater. 54-57, 423.
M.B. Maple, 1980, Phys. Lett. A 75, 243. Peysson, Y., B. Salce and A.M. de Go6r, 1986c,
Ott, H.R., H. Rudiger, Z. Fisk and J.L. Smith, in: Phonon Scattering in Condensed Matter,
1983, Phys. Rev. Lett. 50, 1595. Vol. V, eds A.C. Anderson and J.P. Wolfe
Ott, H.R., O. Marti and E. Hulliger, 1984a, (Springer, Berlin) pp. 251-253.
607. Phillips, W.A., 1972, J. Low Temp. Phys. 7, 351.
Ott, H.R., H. Rudiger, T.M. Rice, K. Ueda, Z. Fisk Phillips, W.A., 1987, Rep. Progr. Phys. 50, 1657.
and J.L. Smith, 1984b, Phys. Rev. Lett. 52, 1915. Pikus, G.E., 1956, Zh. Tekh. Fiz. 26, 36.
Paderno, Y . B . , N.Y. Shitsevalova and Pitlai, C.G.S., and A.M. George, 1988, Proc. Nat.
V.B. Muratov, et al., 1991, Borides, preprint
Acad. Sci. India, Sect. A 54, 824.
91 - Nt (IPM AN USSR, Kiev). In Russian.
Pillai, C.G.S., and A.M. George, 1991, Int.
Parfen'eva, L.S., I. Dzhabbarov, S.I. lvanov,
J. Thermophys. 12, 207.
I.A. Smirnov and V.V. Medved, 1979, in:
Pohl, R.O., and J.J. de Yoreo, 1985, in: Physics of
Refractory Rare Earth Compounds, ed. K.E.
Disordered Materials, eds D. Adler, H.
Mironov (Nauka, Novosibirsk) pp. 139 141. In
Fritzsche and S.R. Ovshinsky (Plenum, New
Russian.
York) pp. 529-540.
Parfen'eva, L.S., I.A. Smirnov, V.V. Tikhonov,
V.S. Iskotski and V.V. Sokolov, 1990a, in: Raki, M., L. Forro, C. Ayache, D. Schmitt,
Physics and Chemistry of Rare Earth Semicon- A.K. Bhattacharjee and B. Coqblin, 1990, Phys-
ductors (Nauka, Novosibirsk) pp. 106-108. In ica B 163, 93.
Russian. Ravex, A., J. Flouquet, J.L. Tholence, D. Jaccard
Parfen'eva, L.S., I.A. Smirnov, V.V. Tikhonov and and A. Meyer, 1987, J. Magn. & Magn. Mater.
K.M. Khalmedov, 1990b, in: Abstracts 5th All- 63-64, 400.
Union Conf. on Physics and Chemistry of Rare Ray, D.K., K. Westerholt, S. Methfessel and
Earth Semiconductors (Universitet, Saratov) S.K. Ghatak, 1981, Solid State Commun. 38,
p. 31. 783.
Parfen'eva, L.S., V.V. Sokotov, I.A. Smirnov, Raychaudhuri, A.K., J.M. Peech and R.O. Pohl,
V.V. Popov, N.N. Efremova and L.D. Finkel- 1980, in: Phonon Scattering in Condensed
shtein, 1990c, in: Physics and Chemistry of Rare Matter, ed. H.J. Maris (Plenum, New York)
Earth Semiconductors, ed. K.E. Mironov pp. 45-48.
(Nauka, Novosibirsk) pp. 109-111. In Russian. Red'ko, N.A., and B.K. Chakalski, 1987, Zh.
Parsons, M.W.S., 1976, in: Phonon Scattering in Eksp. Teor. Fiz 92, 1090.
Regueiro, M.D.N., and D. Castello, 1990, invited ids, eds L.J. Challis, V.W. Rampton and A.F.G.
talk in: Int. Conf. LT-19, Brighton, England, Wyatt (Plenum, New York) pp. 154 162.
p. 631. Shelton, R.N., R.W. McCallum and H. Adrian,
Reisfeld, R., and A. Bornstein, 1977, Chem. Phys. 1976, Phys. Lett. A 56, 213.
Lett. 47, 194. Sher, A., R. Solomon, K. Lee and M.W. Muller,
Reisfeld, R., A. Bornstein, J. Flahaut, M. Guittard 1966, Phys. Rev. 144, 593.
and A.M. Loireau-Lozach, 1977, Chem. Phys. Sierro, J., E. Bucher, L.D. Longinotti, H. Taka-
Lett. 47, 408. yama and P. Fulde, 1975, Solid State Commun.
Reisfeld, R., A. Bornstein, J. Bodenheimer and 17, 79.
J. Flahaut, 1979, J. Lumin. 18-19, 253. Singh, D.P., and G.S. Verma, 1972, Phys. Rev. B
Remizova, L.A., 1975, Tekh. Vys. Temp. 13, 207. 6, 4013.
Remizova, L.A., 1978, Tekh. Vys. Temp. 16, 773. Slack, G.A., 1964, J. Appl. Phys. 35, 339.
Richards, S.M., and J.S. Kasper, 1969, Acta Crys- Slack, G.A., 1979, Solid State Phys. 34, 1.
tallogr. B 25, 237. Slack, G.A., and D.M. Oliver, 1971, Phys. Rev. B
Rives, J.E., G.S. Dixon and D. Walton, 1969, 4, 592.
J. Appl. Phys. 40, 1555. Slack, G.A., D.M. Oliver and F.H. Horn, 1971,
Roedhammer, P., W. Reichardt and F. Holtzberg, Phys. Rev. B 4, 1741.
Smirnov, I.A., 1972, Phys. Status Solidi a 14, 363.
1978, Phys. Rev. Lett. 40, 465.
Smirnov, I.A., 1981, Zh. Vses. Khim. Ova. 26, 602.
Salamon, M.B., P.R. Garnier, B. Golding and
Smirnov, I.A., and V.S. Oskotski, 1978, Usp. Fiz.
E. Buechler, 1974, J. Phys. & Chem. Solids 35,
Nauk 124, 241.
851.
Smirnov, I.A., and V.I. Tamarchenko, 1977,
Salce, B., and L.A. Boatner, 1986, in: Phonon
Electronic Thermal Conductivity in Metals and
Scattering in Condensed Matter, eds A.C. Semiconductors (Nauka, Leningrad). In
Anderson and J.P. Wolfe (Springer, Berlin) Russian.
pp. 272 274. Smirnov, I.A., L.S. Parfen'eva, V.Y. Khusnut-
Samsonov, G.V., 1964, Refractory Compound of dinova and V.M. Sergeeva, 1972, Fiz. Tverd.
Less-Common Metals with Nonmetals (Metal- Tela 14, 844.
lurgiya, Moscow). In Russian. Smirnov, I.A., L.S. Parfen'eva and V.M. Sergeeva,
Samsonov, G.V., T.I. Serebryakova and 1973, Neorg. Mater. 9, 485.
V.A. Neronov, 1975, Borides (Atomizdat, Mos- Smirnov, I.A., V.S. Oskotski and L.S. Parfen'eva,
cow). In Russian. 1985, J. Less-Common Met. 111, 353.
Sandalov, I.S., and A.N. Podmarkov, 1985, Zh. Smirnov, I.A., V.S. Oskotski and L.S. Parfen'eva,
Eksp. & Teor. Fiz. 88, 1321. 1988, in: Actual Problems of Physics and
Schell, G., H. Winter, H. Rietschel and F. Gompf, Chemistry of Rare Earth Semiconductors (AN
1982, Phys. Rev. B 25, 1589. USSR, Makhachkala) pp. 4-23. In Russian.
Schmitt-Rink, S., K. Miyake and C.M. Varma, Smirnov, I.A., V.S. Oskotski and L.S. Parfen'eva,
1986, Phys. Rev. Lett. 57, 2575. 1989, High Temp. High Pressures 21, 237.
Schneider, H., and D. Wohlleben, 1981, Z. Phys. Smirnov, I.A., V.V. Tikhonov, P.A. Alekseev,
B 44, 193. V.N. Lazukov, L.N. Vasil'ev, V.V. Sokolov and
Schneider, H., Z. Kletowski, F. Oster and L.S. Parfen'eva, 1990, in: Abstracts 5th All-
D. Wohlleben, 1983, Solid State Commun. 48, Union Conf. on Physics and Chemistry of Rare
Earth Semiconductors (Universitet, Saratov)
1093.
p. 33.
Sergeeva, V.M., E.V. Goncharova, N.F.
Smirnov, I.A., L.S. Parfen'eva, V.S. Oskotski, L.N.
Kartenko, M.A. Demina, I.A. Smirnov, R.N. Vasil'ev, V . V . Tikhonov, K.M. Khal-
Andryushin and Y.K. Misyurev, 1972, Izv. medov, V.V. Sokolov, P.A. Alekseev and V.N.
Akad. SSSR Neorg. Mater. 8, 2114. Lazukov, 1992a, to be published.
Shadrichev, E.V., L.S. Parfen'eva, V.I. Smirnov, I.A., L.S. Parfen'eva, T.B. Zhukova,
Tamarchenko, O.S. Gryaznov, V.M. Sergeeva K.M. Khalmedov, V.S.Oskotski, I.N. Kulikova,
and I.A. Smirnov, 1976, Fiz. Tverd. Tela 18, E.A. Sovestnov and V.A. Shaburov, 1992b, to
2380. be published.
Sharma, K.C., and R.S. Tripathi, 1980, Phys. Smith, H.G., G. Dolling and T. Goto, 1985, Solid
Status Solidi b 102, K161. State Commun. 53, 15.
Sheard, F.W., 1976, in: Phonon Scattering in Sol- Sokolov, V.V., A.A. Kamarzin, X. Bach and
S. Methfessel, 1988, in: IVth int. Soviet-West Superconductivity and Applications, Buffalo,
Germany Symp. on Studies of Rare Earth NY.
Compound (Tbilisi) pp. 192 196. Uher, C., 1990, in: Int. Conf. on Transport Prop-
Sparn, G., W. Lieke, U. Gottwick, F. Steglich and erties of Superconductors, Rio de Janeiro, April
N. Grewe, 1985, J. Magn. & Magn. Mater. 29-May 4.
47-48, 521. Vandenberg, J,M., and B.T. Matthias, 1977, Proc.
Sparn, G., M. Baenitz, S. Horn, F. Steglich, Nat. Acad. Sci. USA 74, 1336.
W. Assmus, T. Wolf, A. Kapitulnic and Varma, C.M., 1985, J. Appl. Phys. 57, 3064.
Z.X. Zhao, 1989, Physica C 162-164, part 1, Vasil'ev, L.N., I.N. Kulikova, L.S. Parfen'eva,
508. N.N. Stepanov and E.V. Shadrichev, 1975, Fiz.
Spear, W.E., G. Willeke, P.G. le Comber and Tverd. Tela 17, 2928.
Vasil'ev, L.N., S.M. Luguev and V.S. Oskotski,
A.E. Fitzgerald, 1981, J. Phys. Colloq. (Paris)
1976, Fiz. Tverd. Tela 18, 906.
42, C4-257.
Vasil'ev, L.N., T.I. Komarova, L.S. Parfen'eva and
Steglich, F., 1976, Z. Phys. B 23, 331.
I.A. Smirnov, 1978, Fiz. Tverd. Tela 20, 1077.
Steglich, F., J. Aarts, C.D. Bredl, W. Lieke, Vasil'ev, L.N., A.V. Gorobets, A.V. Golubkov,
D. Meschede, W. Franz and H. Sch/ifer, 1979, V.S. Oskotski, I.A. Smirnov and V.V. Tikh-
Phys. Rev. Lett. 43, t982. onov, 1983, Phys. Status Solidi a 80, 237.
Steglich, F., C.D. Bredl, W. Lieke, U. Rauch- Vasil'ev, L.N., I. Dzhabbarov, V.S. Oskotski,
schwalbe and G. Sparn, 1984, Physica B 126, L.S. Parfen'eva, V.V. Popov and I.A. Smirnov,
82. 1984a, Fiz. Tverd. Tela 26, 2710.
Steglich, F., U. Ahlheim, J.J.M. Franse, N. Grewe, Vasil'ev, L.N., I. Dzhabbarov, L.S. Parfen'eva,
D. Rainer and U. Rauchschwalbe, 1985a, V.S. Oskotski and I.A. Smirnov, 1984b, High
J. Magn. & Magn. Mater. 52, 54. Temp. High Pressures 16, 45.
Steglich, F., U. Rauchschwalbe, U. Gottwick, Vekhter, B.G., V.N. Golubev and M.D. Kaplan,
H.M. Mayer, G. Sparn, N. Grewe, U. Poppe 1987, Pis'ma Zh. Eksp. & Teor. Fiz. 45, 136.
and J.J.M. Franse, 1985b, J. Appl. Phys. 57, Venevtsev, Yu.N., and V.N. Lubimov, eds, 1990,
3054. Magnetoferroelectric Materials (Nauka, Mos-
Stewart, G.R., Z. Fisk, J.O. Willis and J.L. Smith, cow) p. 181. In Russian.
I984, Phys. Rev. Lett. 52, 679. Volovik, G.E., and L.P. Gor'kov, 1984, Pis'ma
Zh. Eksp. & Teor. Fiz. 39, 550.
Strzhemechny, M.A., and S.E. Kalnoy, 1983, Fiz.
Volovik, G.E., and L.P. Gor'kov, 1985, Zh. Eksp.
Nizk. Temp. 9, 318.
& Teor. Fiz. 88, 1412.
Sturge, M.D., 1967, Solid State Phys. 20, 92 21 l. Walton, D., J.E. Rives and Q. Khalid, 1973, Phys.
Suhl, H., B.T. Matthias and E. Gorenzwit, 1959, Rev. B 8, 1210.
J. Phys. & Chem. Solids 19, 346. Weston, H.T., and W.B. Daniels, 1984, Phys. Rev.
Sulpice, A., P. Candit, J. Chaussy, J. Flouquet, B 29, 2709.
D. Jaccard, P. Lejay and J i . Tholence, 1986, Wilson, A.H., 1965, The Theory of Metals, 2nd
J. Low Temp. Phys. 62, 39. Ed. (Cambridge University Press, Cambridge).
Sfitsch, P., and E. Umlauf, 1981, J. Low Temp. Wong, Y.H., 1978, Phys. Rev. B 17, 3899.
Phys. 42, 11. Woolf, L.D., D.C. Johnston, H.B. Mac Kay,
Takegahara, K., and T. Kasuya, 1985, Solid State R.W. McCallum and M.B. Maple, 1979, J. Low
Commun. 53, 21. Temp. Phys. 35, 651.
Tanaka, T., 1974, J. Phys. C 7, L177. Zaitlin, M.P., and A.C. Anderson, 1975, Phys.
Tewordt, L., and T. W61khausen, 1989, Solid State Rev. B 12, 4475.
Commun. 70, 839. Zeller, R.C., and R.O. Pohl, 1971, Phys. Rev. B
4, 2029.
Turberfield, K.C., L. Passell, R.J. Birgeneau and
Zhuze, V.P., E.V. Goncharova, N.E. Kartenko,
E. Bucher, 1971, J. Appl. Phys. 42, 1746.
T.I. Komarova, L.S. Parfen'eva, V.M. Sergeeva
Tfirkes, P.R.H., E.T. Swartz and R.O. Pohl, 1986,
and I.A. Smirnov, 1973, Phys. Status Solidi a
in: Proc. Int. Conf. on the Physics and Chemis- 18, 63.
try of Boron and Boron-Rich Solids, Albuquer- Zinov'ev, V.A., and S.L Masharov, 1973, in:
que, 1985, eds D. Emin, T. Aselage, C.L. Beckel, Thermophysical Properties of Materials
I.A. Howard and C. Wood, AlP Conf. Proc. (Nauka, Moscow) pp. 49 51. In Russian.
140, 346. Zinov'ev, V.A., R.P. Krentsis and P.V. Geld, 1968,
Uher, C., 1989, Invited Talk, 3rd Ann. Conf. on Fiz. Tverd. Tela 10, 2820.

Smirnov 1993

Uploaded by

Copyright:

Available Formats

Smirnov 1993

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Smirnov 1993

Uploaded by

Copyright:

Available Formats

Handbook on the Physics and Chemistry of Rare Earths, Vol.

I.A. SMIRNOV and V.S. OSKOTSKI

Symbols and abbreviations

nL m generalised Fermi integral W0 electron thermal resistivity due to

rD dislocation relaxation time of % paramagnetic-ion relaxation time

(2) it is possible to induce by a magnetic field the K o n d o effect (Sandalov and

2. Brief summary of thermal conductivity of solids

The total thermal conductivity lCtot of insulators, metals and semiconductors

insulators: /('tot = lgL~

thermal conductivity, is possible in magnetic insulators, metals and semiconductors

2.1. Thermal conductivity o f a crystal lattice

2 .),,I..X.j- exP O/aT

1,3//-._~ Fig. 1. Schematicshape of the tempcraturedependenceo f l ¢ L for metals

tel = GT3 [o/r 1 siXnh2(½x)dx. (3)

Here x = h~O/ko T, G is a constant, T- ~ is the sum of the reciprocal relaxation times

2.2. Electron contribution to the thermal conductivity

The electron thermal conductivity of metals and semiconductors is determined by

where Wo is the thermal resistivity caused by electron scattering by impurities and

(T~[ 0 D 0 [2~2 01 0)1

W~ = po/Lo T + ocT2 = fi/T + c~T2, (8)

The temperature dependence, We = f ( T 3 ) , is a straight line with a slope of c< The

L[po (T) 5 (O)l/[po (T, D~], (10)

with 15 and 17 integrals of the type (7), A o is a constant which depends on an

c~= \(r + ~-)F~+ 1/2(#*) - #* (13a)

(Oskotski and Smirnov 1972)], r is a scattering parameter (r is an exponent in the

number has a more complicated form (Oskotski and Smirnov 1972)

2LI °LI-- (1L2)2 (15)

where "Lm (z, #*) are generalised Fermi integrals:

These integrals are tabulated in Smirnov and Tamarchenko (1977). Here

L depends on the parameter S:

Phononscatteringon ~_ I ConcentratedKondo I Phononscattering

3. Influence of the magnetic structure on the thermal conductivity of lanthanide

At temperatures T ~ O thermal phonons in metals are mainly scattered by conduc-

(b) for magnetic compounds (M) (GdA12, etc.)

Experimentally AW is determined as the difference 1/toM - 1/~cTM, where tcNM is the

YAI 2 Fig. 9. VV~= l#c mMand We = 1/KM as functions of

Fig. 10. The dependence of P~p,1 on tile de Gennes l~.ctor for

2 I I I t I I I I I I I I I Fig. 12. The dependence of W of compounds RA12 on

RA1z GdAl/ TbA12 DyA12 HoA12 ErA12 TmA12

P*pd(gf~cm) 43.7 32.8 16.6 13.2 5.5 1.7

(Bauer et al. 1986). Figure 14 shows the dependence p~pd=f(ln T) [p~pd =

Fig. 14. Dependence of P~pa (1) and AW~T (2)

3.3.2. Behaviour of the thermal conductivity of lanthanide materials at a SOPT

I~F-- 'I I I--q

Fig. 18. Properties of Smo.6Gdo.4FeO 3 in the

3.4. Phonon magnon scattering, heat transfer by magnons in lanthanide compounds

Fig. 20. Scheme of the dependence of ~c(H)/~:(O)on

~ 0.4 \ ~] " "-Z"

Peculiarities in the p h o n o n and m a g n o n dispersion curves can have influence on

/('tot ~- K:L -[- Ke -t- 1Cm . (31)

4. Thermal conductivity of spin glasses

Kondo regime Spin glass Long range order

Interaction Cluster formation Clusters in the Ferromagnetic,

~._ Concentration of magnetic impurity (C) >

0.8 I I I I I I I I Fig, 28. The temperaturedependenceof L/Lo for the met-

Let us c o n s i d e r d a t a for ~ m e a s u r e d in zero m a g n e t i c field (H = 0). S o m e r e g u l a r i t i e s

:0017 /// ~'NN

O Fig. 34. The magnetic field dependence ~c(H)/~(O) for EuS at

1986, M c C o l l u m et al. 1964). A thermal conductivity which decreases in magnetic

Hcr H Fig. 38. Schematic shape of the dependence of ic(H)ffc(O).

c~= \(r + ~-)F~+ 1/2(#) - # (13a)