You are on page 1of 13

Magnetic Systems: Specific Heat$

JG Sereni, Low Temperature Division, CAB-CNEA, San Carlos de Bariloche, Argentina


r 2016 Elsevier Inc. All rights reserved.

1 Introduction 1
2 Usual Contributions to Specific Heat 2
2.1 Nuclear and Phonon Contributions 2
2.2 Free Electron Mass Enhancement 2
2.3 Fermi Liquids 3
3 Phase Transitions 4
3.1 Ehrenfest Classification of Transitions 4
3.2 Effects of Magnetic Field, Pressure, and Alloying 5
3.3 The Ordered Phase Below Tord 6
3.4 Specific Heat Tails at T4Tord 7
4 Specific Heat Anomalies 7
4.1 Schottky Anomalies 7
4.2 Spin Glasses 8
4.3 Kondo Anomaly 8
4.4 Low Dimension and Anisotropic Systems 8
5 Heavy Fermions 9
5.1 Non-Fermi Liquid Behavior 10
5.2 Entropy Bottlenecks 11
6 Concluding Remarks 12
References 12
Further Reading 13

1 Introduction

Historically, Specific Heat was defined as the Heat (δQ) required to rise the temperature (T) of a unit of mass of a substance in one
degree: Cj ¼ δQ/δT, where j indicates the control parameter kept constant. This elementary definition worked hand in hand with the
development of the Thermodynamics because it is directly related to those parameters to be taken into account to describe the
state of any physical system, i.e., its energy and temperature.
From the First Law of thermodynamics one knows that the heat absorbed by a system is distributed as: δQ ¼ dU þ δW, where U
is the internal energy and W the work done by the eventual expansion (compression) or by the change of magnetization (M) in
external magnetic field (h). If only the thermal component of the system is taken into account (i.e., dV¼ dh¼ 0), the simple
expression Cv,h ¼ dU/dT can be applied, provided that the measurements are performed in adiabatic (vacuum) conditions to avoid
any external diffusion of the applied heat δQ. This constraint implies to measure at constant pressure, i.e., Cp ¼ dE/dT, where
E¼ U þ pV is the Enthalpy of the system that accounts for an eventual thermal expansion a of the system during the heating
process. The difference between Cp and Cv is given by Cp  Cv ¼ TVa2/κ, where κ the is the compressibility (see, e.g., Gopal, 1966).
Provided that a and κ are nearly temperature independent, this difference becomes negligible at low temperatures. However, this is
not strictly valid around all magnetic transitions or when the thermal energy kBT is comparable with the characteristic energy of the
system in the low temperature range. Similar precautions have to be taken into account also with the Grüneissen parameter:
GBκa/Cv which depends on the same parameters.
The Second Law refers to the progressive access of the system to the excited states by increasing temperature. The evaluation of
that process is quantized by the entropy S(T) that is related to δQ by dS¼ δQ/T. Thus, the measurement of Cp(T) provides a direct
R
knowledge of the Entropy variation of the system, S(T)¼ Cp/TdT. From this point of view, specific heat can be regarded like a
‘thermal spectrometer,’ where the excitation is given by δQ and the response by δT. The scale of energy of such excitation currently
ranges between mEv, at a few Kelvin degrees, and some nEv in the mK range of temperature. This characteristic provides an


Change History: April 2015. J.G. Sereni updated the Abstract. Extended the Section ‘Contributions to Specific Heat’ with new subsections. Extended Section
‘Specific heat at the Ordered Phase Boundary,’ now ‘Phase transitions’. Extended the Section ‘Transformation and Related Anomalies,’ now ‘Specific Heat
anomalies.’ Extended the Section ‘Phases Involving Strongly Correlated electronsv, now ‘Heavy Fermions’ with new subsections: ‘Non-Fermi Liquid behavior’
and ‘Entropy Bottlenecks.’ Updated ‘Concluding Remarks’ and references. All figures are now in ‘color’ version. Figures 1, 2, 3, and 4 were modified. Figures 5,
6, 7, and 8 were added.

Reference Module in Materials Science and Materials Engineering doi:10.1016/B978-0-12-803581-8.02787-9 1


2 Magnetic Systems: Specific Heat

extremely sensitive tool to explore low energy excitations spectra. Since S(T) counts the effective number of accessible states, its
knowledge also provides information about the occupation ratio of excited levels and about the first term of the free energy
variation dG¼  SdT þ Vdp  Mdh, which is needed to characterize new systems.
In the next Section the most usual contributions to the specific heat at low temperature are described. The characteristic
behavior of the specific heat around magnetic transitions is discussed in Section 3, including the information that can be extracted
from the Cm(T) dependence below and above the transition. Section 4 addresses to specific heat anomalies observed in magnetic
systems which are not related to long range magnetic order LRMO formation. The physics of strongly correlated electron systems
behaving as heavy Fermions (HF) is reviewed in Section 5, including some remarks on the valuable input that can be obtained by a
complementary analysis of the entropy.

2 Usual Contributions to Specific Heat

Since all allowed states of a system are equally likely, each component of a physical system contributes with its own Entropy to the
total value (i.e., S(T) ¼ ΣiSi(T)). Consequently, Cp/T¼ dS/dT contains information about the total density of excitations. To extract
the contribution of each component requires to identify the range of temperature at which their density of excitations become
dominant according to the respective thermal dependencies. Some typical contributions to the total specific heat Cp(T) are: (1) the
nuclear magnetism of some isotopes in the mK range (i.e., To0.5 K) where the tail of the nuclear contribution Cnuc is clearly
observed; (2) the conduction electrons contribution Cel is mainly detected within the 0.5oTo4 K range; (3) in presence of HF
quasiparticles Cel dominates the signal up to about 7 K; (4) phonons Cph overcome those contributions above 10 K depending on
the respective the Debye temperatures yD; (5) magnon contribution Cmag related to LRMO phases and specific heat anomalies
originated in short range correlations can be identified up to about 20 K; (6) Schottky anomalies Csch correspond to a class of
anomalies arising from the thermal population of quantum levels split by a gap of energy.
These contributions are governed by different statistics depending on the nature of the involved particles. For example,
Maxwell–Boltzman statistic applies to Schottky-type anomalies arising from nuclear or crystalline electric field (CEF) splitting.
Fermi–Dirac statistic applies to conduction electrons behaving as a Fermi Gas or to heavy quasiparticles behaving as a Fermi liquid
with enhanced effective mass. Bose–Einstein statistic applies to phonons and magnons quasiparticles. Since many of these
contributions are extensively treated in the literature (see, e.g., Gopal, 1966; Tari, 2003), we will mainly pay attention to recent
investigations on new magnetic materials at low temperature.

2.1 Nuclear and Phonon Contributions


In order to study magnetic properties originated in electronic properties, nuclear and phonon contributions have to be subtracted
by computing Cm ¼ Cp  (Cnuc þ Cph). The nuclear contribution arises from the hyperfine field hhyp produced by the electrons acting
on the nuclear magnetic moment mI ¼ gImNI. Since the nuclear magneton mNE103mB (mB is the Bohr magneton), the magnetic
moment mI is very weak. Because hhyp lifts the nuclear state degeneracy (2I þ 1), the Cnuc(T) dependence can be described as a
Schottky-like anomaly (details on this type of anomaly are described in Section 4.1). In practice, for Cm(T) studies down to the mK
range only the high temperature tail of the nuclear Schottky anomaly is taken into account, i.e., Cnucp1/T2 (see, e.g., Phillips,
1972; Sundström, 1978).
Concerning phonon contribution, its subtraction in real systems may become quite difficult because Cph(T) depends on
actually complex phonon spectra. For simple metals, Cph is extracted from the slope of a Cp/T vs. T2 representation (see Figure 1)
because Cp/T¼ Cel/T þ Cph/T¼ g þ bT2, with b ¼ 12p4 kB n=5y3D and yD the Debye temperature. The electronic contribution Cel ¼ gT is
described in the following subsection. According to the Debye function, this formula is valid up to TEyD/20 and can be extended
including a  δT5 term. In multi-atomic systems it is frequent to observe a wrong determination of yD because this formula is only
valid for pure elements or for n atoms of equivalent mass. This clearly occurs in compounds built on atoms with quite different
masses like, e.g., the high temperature superconductor YBa2Cu3O6.7. Since the atomic mass of Ba(137.3) is more than twice times
larger than Cu(63.5) the yD determination from low temperature Cp(T) measurements is dominated by the heaviest atom Ba
(yD ¼ 110 K) with a negligible contribution of Cu (yD ¼ 340 K) (Gschneidner, 1964).
The usual way to subtract Cph(T) in a magnetic system is to obtain it from a nonmagnetic isotypic compound, that for RE-based
ones is extracted from Y, La, or Lu isotypics. When those references are not accessible, a good low temperature description can be
obtained applying the heuristic function Cph ðTÞ ¼ b½1 þ b  expðν=TÞT 3  δT 5 , where ν accounts for the first correction in the
ideal Debye spectrum.

2.2 Free Electron Mass Enhancement


Although magnetism originates in electronic spin and orbital angular momenta, in the case of conduction electrons in a metal
only the former contributes to magnetic properties. The free electron contribution to specific heat Cel is described within the Fermi
Magnetic Systems: Specific Heat 3

100 150

80

C P / T [mJ / mol K 2 ]
100
60

40 β
γ 50

20

0 0
0 20 40 60 80
T 2 [K ]
Figure 1 (Color online) Examples for the Cp/T¼ g þ bT2 representation, being g and b the respective coefficients for free electron and phonon
contributions. CeSi1.85 and UAl2 are Spin Fluctuation exemplary systems. The continuous curve is the fit of CeSi1.85 using
Csf =T ¼ g þ bT 2 þ δT 2 lnðT =Tsf Þ, see the text.

theory as Cel ¼ gT, being the Sommerfeld coefficient g directly proportional to the density of states at the Fermi level, i.e., g [J/
mol K2]¼ Z(eF) [states/eV atom]/424 and consequently to the electronic mass m0 (Tari, 2003).
In the free electron model a simple parabolic density of states is used to define the m0 ¼ ℏ2 k2 =2e dispersion relation. However,
for the study of physical systems a more realistic description of the density of states is required. In a first order approximation a
simple re-normalization of the electronic mass meff allows to describe many systems, provided that they keep obeying Fermi-Dirac
statistics. Such a re-normalization corresponds to an enhancement of the measured Sommerfeld coefficient: meff/m0 ¼ g/g0, typi-
cally due to electron–phonon interaction lph, electron spin–split interaction lsf and/or local-band electron interaction lee. Since
these interactions are basically independent, an additive criterion like g¼ g0(1 þ lph þ lsf þ lee) is applied. The experimental values
of g are extracted from a Cp/T¼ g þ bT2 representation (see Figure 1 for the case of Pd). For pure metals g ranges between 0.19 and
10 mJ/mol K2 for Be and Y, respectively (Phillips, 1972).
The electron–phonon interaction is relatively weak: 0.54lph41, but it plays a relevant role in classical superconductors. In fact,
the lph term for those materials is extracted from a relationship between the superconductive temperature and yD (McMillan,
1968).
In itinerant systems, typically with 3d or 4d-Pd elements, the exchange between opposite spins of band electrons produces a
spontaneous spin–split of electrons with opposite polarization powered by the molecular field. This effect is described by the
Stoner enhancement factor S¼ 1  UZ(eF) where U accounts for the Coulomb energy (see, e.g., Blundell, 2004). As a consequence,
the Pauli susceptibility χP is enhanced as χP ¼ Sχ0 respect to the unperturbed susceptibility, with the associated increase of the bare
density of states pg0 and the re-normalization of the characteristic energy of spin fluctuation Tsf ¼ TF/S. For S{1, Tsf competes
with the thermal excitations and, if the spin fluctuation frequency osfp(kBTsf)1 is larger than thermal fluctuations, the spins
contribution looks nonmagnetic, i.e., χP like. On the contrary, for T4Tsf thermal fluctuations are faster than and the system tends
to a Curie-type p1/T paramagnetic (PM) behavior. This crossover manifests in a maximum in the susceptibility, which increases
like χ(T) ¼ χP þ bT2 on the ToTsf side. At low temperature, the associated increase of meff is accounted by the lsf term, which has the
exemplary representative in pure Pd with lsf ¼ 0.9 and g¼ 9.4 mJ K2 mol1 (Phillips, 1972), included in Figure 1. Among RE-
intermetallics an illustrative example is LaNi5 with g¼ 34 mJ K2 mol1, which is further enhanced by the presence of
instable valence Ce in CeNi5 with g ¼ 40 mJ K2 mol1 and TsfE100 K (see Sereni, 1991 and references therein).
In localized 4f and 5f system, the characteristic fluctuation energy kBTsf is strongly reduced and local details in the density of
states become relevant. For spins 1/2, the enhanced specific heat undergoes a temperature dependence at ToTsf described by
Csf =T ¼ g þ bT 2 þ δT 2 lnðT=Tsf Þ (see, e.g., De Visser, 1987). In Figure 1, two exemplary systems UAl2 and CeSi1.85 (Stewart, 1984)
are included showing the g term enhanced by more than a decade respect to pure Pd. The latter compound is fit with a CP/
T¼ 92 þ 0.84T2ln(T/130) þ 0.06T2 mJ K2 mol1 formula.

2.3 Fermi Liquids


One of the most relevant phenomena in the physics of Rare Earths (4f) and Actinides (5f) concerns the possibility of hybridization
between localized f-orbitals and band states. This mechanism may enhance meff up to three orders of magnitude. The strength of
hybridization depends on a matrix element of mixture which may reach some tenth of Ev.
Since for strong hybridization the quantum mixture involves spin and charge components of the electrons, this scenario is
known as Valence Fluctuation because the chemical valence itself is affected. In the extreme cases of hybridization strength the very
4 Magnetic Systems: Specific Heat

low values of g are observed, like in CeN (g¼ 8.3 mJ K2 mol1) and CePd7 with g¼ 9.4mJ K2 mol1, as it is extracted from the fit
in Figure 1. Notice that the latter has the same g value than pure Pd despite it contains eight atoms in the formula unit.
In a periodic disposal of those atoms, c.f., a Kondo-lattice, the Quasiparticles emerging from the quantum mixture of f–s states
form a narrow band with an enhanced electronic mass meff. That Fermi liquid FL character is recognized by the same dependence
of electronic g and Pauli-like χP parameters on the density of states Z(eF). The presence of a narrow band allow the formation of a
coherent ground state (GS), revealed by the low temperature dependence of the electrical resistivity as r ¼ AT2 well below a
characteristic r(T) maximum (which defines an energy scale kBTcoh), where the A coefficient depends on the electron–electron
interaction driving the quasiparticles formation. In that scenario A1/2pZ(eF). This means that a FL system has to fulfill the
condition that gpχPpA1/2pZ(eF). As a consequence, some relationships between those parameters can be established, like the
Wilson (p2 k2B χP =m2eff g) and Kadowaki–Woods (A/g2 ¼ 106 O cm (J/Kmol)2) ratios (Continentino, 2000).
Like for a free electron gas, where Cel/T is constant up to E10% of their characteristic Fermi energy (i.e., up to 103 K), a constant
CFL/Tpg characterizes FL systems. However, the first correction CFL/T ¼ g  b(T/TK)2 (Phillips, 1972) can be observed at much
lower temperature because of the characteristic temperature TKE100 K. Since gp1/TK, the measured values range between
20ogo200 mJ/mol K2 (Sereni, 1991). Another characteristic of these systems is the lack of CEF effects due to the delocalization of
the 4f and 5f electronic charges. As a consequence, their GS have the full Hund's N ¼ 2J þ 1 degeneracy that, in the case of Ce ions
(with J ¼ 5/2) fixes the Wilson ratio χP/g ¼ 0.036 emu K2 J1, as it is experimentally observed in at least in six Ce-IV compounds
(Sereni, 1991).
For smaller TK values (in the range of 10 K) the g term may grow up to approx 1 J K2 mol1. Within that regime, the
quasiparticles are recognized as HF (Stewart, 1984), where the moderate hybridization strength only involves the spin component.
Although the limit between valence fluctuation and HF is not quantitatively established, the deviation from FL behavior obeying
the Fermi–Dirac statistic gives the distinctive criterion. Specific heat measurements clearly detect the deviation of a HF behavior
from a CFL(T)/T dependence, thus these systems are identified as non-Fermi-liquids (NFL). The characteristics of these materials
are addressed in Section 5.

3 Phase Transitions

3.1 Ehrenfest Classification of Transitions


The specific heat is the proper parameter to define the order of a phase transition at the critical temperature T¼ Tord (Pippard,
1964). In the case of a first order transition, the first discontinuity in the free energy derivative ∂G/∂T ¼ S(T) corresponds to the
Entropy and to a divergency of Cp(Tord) ¼ T∂2G/∂T2. The energy related to the Entropy difference between both phases L¼ TDS is
R
called the Latent Heat that can be evaluated as: L¼ Cm(T)dT, after subtracting foreign contributions, like, for example, phonons.
Clausius–Clapeyron relationships: dTord/dp¼ DV/DS and/or dTord/dh¼ DM/DS provide information about the evolution of the
phase transition as a function of usual control parameters like pressure or magnetic field.
In magnetic systems, first order transitions at high temperature are frequently associated to a reduction of crystalline structure
symmetry like in the Jahn-Teller effect (see, e.g., Blundell, 2004) or magnetostrictive effects frequently observed in ferromagnetic
FM transitions driven by strong spin-phonon coupling. At low temperature, where thermal energy is not strong enough to induce
structural modifications, first order transitions can be produced by a sudden reduction of the GS degeneracy like for example in
BCC Ce-binary compounds with quartet GS (Sereni, 1991). A reduction of symmetry from cubic to tetragonal reduces the fourfold
degeneracy into two doublets with the consequent discontinuity in magnetization and entropy. Other examples of first order
transitions are those observed in the formation of magnetic dimers, spin-Peirels chains or quadrupolar transitions. The tem-
perature dependence of Cm(T) close to the transition allows to extract the critical parameters. The critical exponent describing the
temperature dependence of Cm(t) (where t¼ |T  Tord|/Tord) follows a power law: tZ, being Z is the critical exponent with a value
close to zero for specific heat (see, e.g., Stanley, 1971).
In a second order transition the ∂2G/∂T2 discontinuity is experimentally observed as a jump DCp(Tord), like those shown in
Figure 2, which is related to a change in the dS/dT slope at T ¼ Tord. According to Ginsburg–Landau theory, an order parameter c
that sets on at T¼ Tord with infinite initial slope, develops continuously till T ¼ 0 where c-c|T ¼ 0 monotonously. Taking advantage
that in FM systems c can be identified with the spontaneous magnetization Ms(T), a direct representation of c(T) can be obtained
like in the exemplary FM presented in Figure 2(a). Furthermore, because DUm pM2s one obtains Cm ðTÞ ¼ dM2s =dT. For AFM
R
systems, like the one presented in the inset of Figure 2(b), c(t) can be extracted from the internal energy Um(ToTord)¼ CmdT
evaluated as c(T)p1  Um(T)/UT ¼ 0 as shown in Figure 2(b). This procedure is in good agreement with neutron scattering
measurements. In the case of a broadened transition, as a consequence of atomic disorder for example, Tord can be determined as
the temperature of the maximum slope of Cm(T) that should coincide with the entropy compensation criterion around DCm(Tord)
(see, e.g., Stewart, 1984).
Once c(T) develops, any discontinuity of this parameter implies a jump in the internal energy of the system DUm ¼ L with the
consequent occurrence of a first order transition. Therefore, in magnetic systems any other transition below Tord is expected to be of
first order, as exemplified by CeRu2Ge2 (Bouquet et al., 2000). Many magnetic phases with incommensurate propagation vectors
Magnetic Systems: Specific Heat 5

14 1.0
0.4
12

Msat / M T=0
0.8

3/2
10 ΔM=7K

Cel / T
0.1

8 0.6

Cm [J / mol K ]
Tord /T
0.04
6 1 2 3 4 0.4

Sm / R Ln2
4
0.2
2

0 0.0
2 4 6 8 10 12
T [K]

1.0
15

1-Um(T)/ UT=0
0.8

0.6
Cm [J / mol K]

0.4
10
0.2

0.0
2 4 6 8 10 12
T [K]
5

0
2 4 6 8 10 12
T [K ]
Figure 2 (Color online) (a) Ferromagnetic example for Cm(T), DSm(T) and Mspc(T) (order parameter) evolutions. Inset: Arrhenius representation
logðCm =T 3=2 Þ ¼  DM =T for the magnon gap D evaluation, see Section 3.3. (b) Antiferromagnetic (AFM) example for Cm(T)pT3 (solid curve).
Inset: 1  Um(T)/UT ¼ 0pc(T) representation.

show a second transition upon decreasing temperature when they become commensurate with atomic periodicity because
Dc¼ DUma0. It may occur that commensuration is reached in a continuous way, in such a case no first order transition is
observed. FM transitions may produce magnetostrictive effects due to a change of volume triggered by the onset of magnetic order
(second order). In those cases a first order peak appears superimposed to DCm(Tord). Concerning the entropy, if an eventual first
order transition is due to a discontinuity in c(T) only, the total entropy will remain being DSm ¼ Rlnð2J þ 1Þ. However, if that
transition is due to a discontinuity in another physical parameter, like for example a structural modification, that foreign con-
tribution is added to the magnetic one as DStot ¼ DSm þ pDV/Tord.
The hight of DCm(Tord) provides information on the degeneracy of the ordered phase GS when it obeys the mean field MF
approximation for magnetic transitions. In that case DCm(Tord) depends on the total angular momentum J according to:
DCm(Tord) ¼ 1.5R[(2J þ 1)2  1]/[(2J þ 1)2 þ 1], with the related Entropy Sm ¼ Rlnð2J þ 1Þ, see for example Meijer et al. (1973). For
a typical doublet GS: J¼ 1/2, the value is DCm(Tord) ¼ 12.5 J/mol K and Sm ¼ 5.76 J/mol K, like in the exemplary compounds Ce
(Pd0.5Ni0.5) and CeIn3 presented in Figure 2. Thus, a first order transition can be unambiguously recognized when DCm(Tord)
exceeds the predicted MF value, like in CeRu2Ge2 (Bouquet et al., 2000). Nevertheless, the MF description fails to describe low
temperature magnetic transitions (below about 5 K) because the predicted jump does not depend on Tord, in violation of
thermodynamic principles stating that Sm(T) and Cm(Tord)-0 as T-0.

3.2 Effects of Magnetic Field, Pressure, and Alloying


The three usual control parameters: magnetic field, pressure, and alloying allows to drive magnetic systems into alternative states of
interest. As complement of χ(T) and M(T) measurements, external magnetic field h can be used for the study of the nature of the
magnetic order through Cm(T) measurements. In a FM, the applied magnetic field broadens the transition because of the
enhancement of magnetic correlations above Tord. As a consequence c(T) (or Ms(T)) sets in at higher temperature in a continuous
manner. By increasing h, the Cm(Tord) anomaly progressively tends to a Schottky-anomaly shape (see Section 4.1) that is reached
6 Magnetic Systems: Specific Heat

once h exceeds the internal molecular field hmf. Beyond that value the maximum of Cm(T) shifts in temperature proportionally to
the h increase. This transformation is due to the splitting of the doublet GS driven by h. In an AFM systems, the effect of h is quite
different because DCm(Tord) is continuously shifted to lower temperatures. Since h acts on the space-orientation of the magnetic
moments, it affects the exchange interaction Jex but not necessarily the magnetic moment intensity meff. As a consequence, when the
applied field exceeds the molecular field hmfpJex, at a critical hcr value, a meta-magnetic transition due to the flip of meff directions
occurs. Above hcr the system behaves like an induced-FM. External field effect is an important tool in anisotropic systems, because
Tord and consequently DCm(Tord) depend on the relative direction between h and the internal magnetic axes. CeCu6xAux single
crystals, for example, shows completely different behaviors of those parameters depending whether h is applied on the easy or hard
magnetic directions (see, e.g., Paschke, 1994).
Concerning calorimetric measurements, the application of external pressure p is restricted to maintain the semi-adiabatic
conditions. Two main procedures are currently used, one of them consists in the measurement of the pressure cell containing
the sample followed by the subtraction of the empty cell heat capacity (Phillips, 1968). The other, called AC-calorimetry, takes
profit of the difference between the internal heat diffusion time within the sample and its transference to the pressure cell (see, e.
g., Bouquet et al., 2000). One of the most relevant effects of p in magnetic systems concerns those with magnetic moments at the
edge of their instability. Since the hybridization of local and conduction electron states may induce the electronic charge
delocalization, the consequent collapse of the atomic volume is very sensitive to external pressure. The outstanding example of
that collapse is the g–a transition of pure Ce (Koskenmaki and Gschneidner, 1978). For moderate pressure effects a spin
screening is only induced as it is exemplified by the magnetic non-magnetic transformation in CePd2Si2 and CeRu2Ge2
(Bouquet et al., 2000). In stable moment systems one may expect a weak increase of Tord due to the reduction of the interatomic
distances with p. In this case the variation of Tord with pressure is related to DCm(Tord) through the Ehrenfest relation: dTord/
dp ¼ VTordDa/DCm. By knowing Cm(T,p) and Sm(T,P), the Maxwell relation ∂S/∂p ¼  ∂V/∂T ¼  a allows to check these results
with the thermal expansion a (Sereni, 2001a).
Two different criteria can be used in the alloying procedure, to dilute the magnetic lattice or to tune the chemical potential (c.f.,
the Fermi energy eF) by alloying nonmagnetic ligand atoms. In the former the LRMO is progressively destroyed by doping the
magnetic lattice, transforming the specific heat jump DCm(Tord) into a broad anomaly due to the atomic disorder. In the limit of
high dilution typical ‘single impurity’ behaviors, like spin glasses or diluted Kondo systems (see Section 4) can be studied. To tune
eF by alloying non-magnetic ligand atoms preserves the magnetic lattice. This is the case when the alloyed atoms have similar size,
otherwise atomic disorder may occur. The Kondo screening allows to trace magnetic phase boundaries Tord(x) down to the
Quantum Critical region. Although the DCm(Tord) jump can be mostly preserved, in Ce-based compounds it is observed that once
Tordr3 K the transition transforms into an anomaly due to quantum fluctuation effects (Sereni, 2007). Stronger alloying effects
may also produce the 4f orbital delocalization with the consequent volume collapse related to the Valence Instability. Exemplary
systems of this scenario are CePd1xNix and CePd1xRhx (Sereni et al., 1993)
Pressure and doping are currently expected to produce similar effects, but in some cases remarkable differences are observed
like in the CeCu2(Si0.9Ge0.1)2 alloy (Sereni, 2001a). While Tord and DCm(Tord) decrease reducing Ge content, applied pressure does
not affect Tord but progressively suppresses DCm/Tord (Sparn et al., 1998). This difference of the DCm/Tord on dependence on alloying
or p clearly shows that their comparison cannot be naïvely generalized.

3.3 The Ordered Phase Below Tord


A detailed study of the Cm(T) dependence at ToTord in different environments and obeying different dispersion-relations of the
spin waves is provided by Akhieza et al. (1968). According to the general Cm(ToTord)pTD/q dependence, where D is the spatial
dimensionality (typically 3D) and q the wave vector of the dispersion relation, some frequent cases can be mentioned. (1)
Isotropic FM and Ferrimagnets with q¼ 2 follow a Cm(T)pT3/2 dependence. (2) AFM with q¼ 1 show a Cm(T)pT3 dependence.
(3) Under external field or anisotropic molecular fields there is an activation energy for the spin wave energy, making Cm(T) to
become exponentially small at T-0 due to a gap DM in the magnon spectrum. If h is applied on the easy axis of anisotropy of a FM
a Cm(T)pT1/2exp(  DM/T) dependence is expected, and a T1/2exp(  DM/T) one if h is perpendicular to the easy plane. (4) In the
case of an AFM, with an easy axis of anisotropy one may find CmpT1/2exp(  DM/T), whereas in a perpendicular field it becomes
CmpT3. Experimental examples of most of these cases can be found in Sundström (1978). In order to extract DM, an Arrhenius
plot logðCm =T 3=2 Þ vs. Tord/T is used as presented in the inset of Figure 2 for the case of the anisotropic FM Ce(Pd0.5Ni0.5). The
magnon energy gap is evaluated from the negative slope of that representation.
The Cm/TD/q dependence applies to systems with doublet GS, which is the case of most RE-based compounds because CEF
effect removes the Hund's rule GS degeneracy N ¼ (2J þ 1). The exception concerns Gd (J ¼ 7/2) since its N ¼ 8 is not affected by
CEF because L ¼ 0. In that case, the GS suffers the Zeeman splitting produced by the internal molecular field acting at ToTord. The
specific heat reveals this feature showing a hump which can be described including the Brillouin function BJ(x) in the formula:
Cm ¼ Rx2BJB0 J/[BJ  xB0 J], where x¼ gJmBJHmf/kBT and Hmf is the temperature-dependent molecular field (Meijer et al., 1973). A good
example for this phenomenology is given by GdNi (Blanco et al., 1994).
Magnetic Systems: Specific Heat 7

The concept of specific heat as a thermal spectrometer is confirmed by its ability to recognize different microscopical dispersion
relations (including gaps) in the magnon spectrum. Two illustrative examples are Ce(Pd0.9Ag0.1) and CePd3Al2. The former shows
AFM behavior in its χ(T) macroscopic signal, despite of the FM CePd matrix. However, measured Cm(T)pT3/2 dependence reveals
the FM character of the magnetic interactions in coincidence with the positive Curie-Weiss temperature (yCW40). In this case, the
doping of Pd (hole-like) lattice by a Ag (electron-like) atom breaks the LR-FM coherence leading the formation of FM domains
with different phases of the propagation vector. The random distribution of domain directions compensate the local FM field and
results in a macroscopic compensated AFM-like behavior. A mirror case is given by CePd3Ga2, which orders FM at 6 K (Bauer et al.,
1993) regardless its Kondo character. Nevertheless, a Cm(T)pT3 dependence is observed for ToTord indicating the presence of
AFM interactions, in agreement with the yCWo0 value. In this case, frustration effects due to the hexagonal symmetry of the
magnetic lattice explains the difference between those microscopic interactions and the actual type of LRMO.

3.4 Specific Heat Tails at T4Tord


The boundary between PM and ordered phases is well defined by a DCm jump mostly in isotropic-3D systems. Frequently, short
range correlations set in well above Tord. The region where these precursors of the LRMO dominate the scenario can be considered
to behave as quasi-PM QPM. The limit between PM and QPM regions can be recognized in χ(T) measurements as a deviation from
the Curie law: χpT1. This criterion can be transferred to the specific heat by using the Maxwell relation: ∂S/δh¼ ∂M/∂T-∂g/
∂h¼ ∂2M/∂T2 ¼ h∂2χ/∂T2 (Stanley, 1971), from which a CmpT2 dependence is deduced (Sereni, 2001b) for the range of tem-
perature at which χpT1. Therefore the CmT2 ¼ B (const.) behavior can be used to distinguish PM and QPM regions.
In magnetic compounds characterized by strong disorder, geometrical frustration or by relevant low lying spin excitations
LRMO does not develop at finite temperature. Nevertheless, the magnetic correlations accounted by the Cm(T) tails provide
relevant information on those precursors of magnetic order regardless it occurs or not as it is discussed in Section 5.

4 Specific Heat Anomalies

Every physical phenomenon has a characteristic energy with an associated temperature T which allows to tune the experimental
control parameters for the proper study of their properties. A correct analysis of specific heat anomalies requires to distinguish
between: phase transition, thermal promotion, and physical transformation. In the former there is a critical temperature (c.f., Tord)
determined by a jump in the specific heat DCm(Tord) which defines the boundary between two phases as discussed in Section 3. The
thermal promotion produces a Cm(T) anomaly (currently known as Schottky anomaly) arising from the increase of Um(T) due to
the thermal promotion of particles into excited DCF levels by crystal electric field CEF. Concerning transformations, this concept
implies the comparison between the thermal energy kBT and a characteristic scale of energy kBT of a system. In this case, two
different regimes are clearly distinguished in two ranges of temperature, i.e., T{T and TcT. The Kondo effect with T ¼ TK and
the freezing process of magnetic moments at T ¼ Tsg in a spin glass are representative examples.
In order to distinguish between a broadened transition (by atomic disorder for example) and a transformation, some ther-
modynamic criteria are useful. In a broadened transition, Cm(T) exhibits a pronounced increase of its negative slope approaching
Tord upon cooling. On the contrary, in a transformation the Cm(T) curvature changes monotonously from positive at TcT to
negative approaching Tmax.

4.1 Schottky Anomalies


This anomaly is frequently observed in specific heat measurements related to the thermal promotion governed by Maxwell–
Boltzman statistics. These systems may arise from the splitting (D) of a degenerated nuclear or electronic (c.f., Hund's rules) GS,
where the degeneracy N is removed by hyperfine (nuclear) or CEF, respectively. The thermal dependence of the specific heat
depicted in Figure 3 is described by:

Csch ¼ RðNi =No Þ ðDCF =TÞ2 expðDCF =TÞ=½1 þ ðNi =No Þ expðDCF =TÞ2 ½1

In the limit of T{DCF one gets: CSch ¼ RðNi =No ÞðDCF =T Þ2 expðDCF =T Þ, whereas at high temperature the typical CschpT2 tail
appears (Gopal, 1966). Since the maximum value Cmax sch depends on the relative degeneracy of the ground N0 and excited Ni levels,
the gain Entropy is also fixed by those parameters DS¼ RlnðNi =No Þ. The value of Cmax sch depends directly on Ni/No and its
max
temperature scales with the splitting like Tsch E0:4DCF . In the case where Ni ¼ No, the function Csch(T) ¼ R(DCF/2T)2sech(DCF/2T)2
can be used.
8 Magnetic Systems: Specific Heat

1.0

0.8 Kondo - C K

Spin Glass

Cm / Cmax
0.6
C sg : Th 0.972 Gd 0.028

0.4
Schottky - CSch

0.2

0.0
0 1 2 3 4
T/T max

Figure 3 (Color online) Comparison between different types anomalies, normalized to their respective maxima of temperature Tmax and specific
heat Cmax.

4.2 Spin Glasses


Magnetic interactions between diluted impurities do not develop LRMO because the random orientations of the magnetic
moments frustrate the propagation of an order parameter. In that case, the molecular field produced by neighboring magnetic
moments on each magnetic site is able to freeze a magnetic configuration at a characteristic temperature: Tsg. Below that ‘glass’
temperature the system behaves in a disordered but cooperative manner. This simplified picture for spin glasses may become quite
complex in real systems when microscopic details of atomic distribution have to be accounted for (see Mydosh, 1993).
Depending on the concentration of magnetic particles, one may distinguish between three different regimes: (1) diluted, where
single spins interact magnetically according to a MF description. In this case, both Tsg and specific heat Csg contribution scale with
the magnetic impurities concentration xi, being a Csg/xi vs. T/xi representation independent of xi (Fogle et al., 1978). At low
temperature (ToTsg) Csg(T) increases nearly linear with temperature with a small T2 contribution at T-0 (i.e., Csg ¼ aT þ bT2)
(Fogle et al., 1978). A characteristic of ‘model’ spin glasses is that the maximum of Csg/T nearly coincides in temperature with the
maximum of the magnetic susceptibility, with the Csg(T) maximum placed at around 30% above (Mydosh, 1993). At high
temperature, the usual CsgpT2 tail is observed. An exemplary spin glass anomaly in specific heat is presented in Figure 3 with the
isotropic diluted alloy Th0.972Gd0.028 (Sereni et al., 1979) which also shows a small bT2 contribution to Csg(T-0). (2) Increasing
the concentration, short range correlations set on leading the formation of magnetic clusters where the scaling properties are lost.
In the trivial case of clusters with identical size and temperature formation Dcl, the associated anomaly can be described by:
Ccl ¼ RðDcl =TÞ2 eDcl =T /ðeDcl =T  1Þ2 ¼ R(D/2T)2csch(D/2T)2. (3) Further increase of magnetic impurities lead to clusters interaction
coexisting with a random (glassy) interactions.

4.3 Kondo Anomaly


In diluted magnetic alloys, local d or f states of a partially filled shell may couple AFM with conduction spins s when their energies
are close to eF. Such a f–s exchange leads to the progressive formation of a non-magnetic GS below a characteristic Kondo
temperature TK (Phillips, 1972). This local moment compensation arises continuously within a logarithmic temperature scale. A
consequence of this Kondo screening is a significant increase of the density of states Z(eF) due to the formation of a virtual-bound
state, which is observed in specific heat measurements through a significant enhancement of the g coefficient. Because of Entropy
conservation, the height of this resonant state (pg) is inversely proportional to its energy width (pTK) leading to the gp1/TK
relationship. The temperature dependence of the Kondo contribution to the specific heat (CK) is therefore linear at low tem-
1
perature (c.f., TrTK/10) and has a maximum of Cmax K ¼ 1.5 J K mol1 at T/TK ¼ 0.45 for a Jeff ¼ 1/2. In Figure 3, CK(T) is
compared with other specific heat anomalies after a CK =Cmax K vs. T/Tmax normalization. Once the singlet GS is formed, χ(T-0)
becomes Pauli-like according to χPpZ(ef)pg. The Kondo scattering of conduction electrons produces the characteristic logarithmic
increase of the resistivity at low temperature.
In dense or Kondo-lattice systems, when the conduction band is less than half filled the local magnetic moments, they may
remain under-screened. In that case LRMO may occur, but involving the fraction of degrees of freedom not yet condensed into the
Kondo state. This situation is reflected in a reduced DCm(Tord)/1.5R jump respect the MF value for Jeff ¼ 1/2.

4.4 Low Dimension and Anisotropic Systems


It is well known that one-dimensional (1D) magnetic arrangements do not develop LRMO and Cm(T) reveals that frustration
undergoing a maximum before to tend to zero at T¼ 0, see the case of 1D  CuSO4  5H2O (de Jongh et al., 1974) in Figure 4. The
Magnetic Systems: Specific Heat 9

K2
1.0 1.0

2D O 2) 6
Ba
(N
-

Ce 7
4 H O
2

Ru 3
1D -
0.8 0.8

.5

CePd0.5Ni0.5
CuSO
Cm / Cmax
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
T / T0.2

Figure 4 (Color online) Normalized Cm/Cmax versus the temperature at which Cm(T)¼0.2Cmax at a temperature defined as T/T0.2 that allows to
compare the evolution of magnetic correlations in 1D-, 2D- (s ¼1/2 Heissenberg), and 3D systems.

exact solutions for elementary cases, like chains, triangular or quadratic configurations are approached by Ising IS and Heisenberg
HS type models. Even for these cases the Cm(T) description is not trivial and have to be evaluates using numerical approximations.
Nevertheless, the morphology of those anomalies provide useful information regarding the characteristics of the ordering phase.
For example, a 2D-IS lattice is expected to exhibit a logarithmic singularity: Cm ¼ Alnt on both sides of the transition, with t¼ |
T  Tord|/Tord and the proportionality factor A depending on the type of magnetic lattice (de Jongh et al., 1974). The systems with a
2D-HS magnetic configuration can be recognized from the evolution of their DSm(T) because it is predicted that
DSR ðToTord ÞC DSm ðT4Tord Þ. A good complement is to evaluate DUm(T) and compare its distribution above and below Tord. For
example, in a 2D-triangular lattice of IS spins 1/2 the values are: (ST ¼ 0  Sord)/(Rln2) ¼ 0.48 and (S1  Sord)/(Rln2) ¼ 0.52, and
(UT ¼ 0  Uord)/kBTord ¼ 0.275 and (U1  Uord)/kBTord ¼ 0.0.55, respectively (see, e.g., Fisher, 1967).
The systems presented in Figure 4 reflect these features. In order to compare those different cases, Cm is normalized to its
maximum value Cmax. For the thermal axis, the absolute temperature is normalized to the value at which Cm(T) reaches a 20% of
Cmax, i.e., T/T0.2. As it can be seen in the figure, in 3D systems the magnetic correlations rapidly develop LRMO whereas in
anysotropic or low dimension ones the thermal energy has to be reduced significantly to reach the maximum. In spite of that, they
show a similar tendency at high temperatures towards the CmpT2 tail analyzed in the previous Subsection. In intermetallic
systems, strict low dimensionality is unlikely because the conduction electrons diffuse in all directions with their polarized spins.
Nevertheless, structural anisotropy distribute magnetic ions in preferential directions approaching in some cases a 2D behavior.
Due to its widening, to define the transition temperature in these systems requires to take into account some special criteria.
Since the thermodynamic definition of Tord corresponds to the maximum variation of ∂Um/∂T in FM systems ∂M2/∂T coincides
with the maximum of Cm(T). For AFM, DUm(T) is proportional to the magnetic correlation strength between spins: osi sj 42 pm2eff
(the effective moment). Since DUm ðTÞpm2eff pχðTÞT, one obtains: Cm(T)pd(χT)/dT (de Jongh et al., 1974). Applying the same
criterion for Tord as the temperature of the maximum variation of Um(T), the d(χT)/dT maximum corresponds to that of Cm(T).
Notice that this definition does not coincide with the maximum slope of Cm(T) 3D transitions because in that case an eventual
broadening has another origin, e.g., atomic disorder.

5 Heavy Fermions

The limit between IV and HF materials is currently established around 100rgr300 mJ K2 mol1. Such a broad empirical
estimation reveals a continuous transformation between those electronic configurations. Therefore, in order to attribute one of
those regimes to certain compound further analysis of complementary physical properties is required. As it was introduced in
Section 2.2, the formation of coherent narrow electronic bands is a characteristic of an IV periodic Kondo-lattice behaving as FL.
The crossover from IV to HF GS can be driven by alloying the RE-ligand atoms, without affecting the magnetic lattice. The
change of regime was traced by specific heat measurements on Ce-based alloys as a maximum of g(x) at a crossover concentration
xcr (Sereni, 1995). The gp(Neff)/TK relationship cannot be applied all along the concentration range because the effective
degeneracy Neff of the GS changes between IV and HF states. The former case is characterized by TK(x)4DCF and therefore the full
degeneracy of Hund's rule GS applies (Neff ¼ 6 in the case of J¼ 5/2). Around the crossover region TK(x)EDCF excited CEF levels
keep contributing (additively) to the GS properties, whereas the density of states increases driven by g(x)pTK(x). Once DCF
10 Magnetic Systems: Specific Heat

overcomes TK(x), Neff drops towards 2 (doublet GS), with the consequent decrease of g(x). Thus, the observed maximum of g(x)
can be attributed to the competition between a g(x)p1/TK(x) increase and the reduction of the GS degeneracy from 6 to 2. This
behavior is observed in Ce(Y1xZx) binary compounds with Y–Z¼ Ni–Pd, Ni–Pt, Rh–Pd, and Sn3–Tl3, respectively (Sereni, 1995).

5.1 Non-Fermi Liquid Behavior


It is currently observed that systems with gZ300 mJ K2 mol1 (or TKr30 K) deviate their Cm/T thermal dependence from the
expected for FL ones, see some exemplary systems included in Figure 5 in a double logarithmic representation. The concept of non-
Fermi-liquid NFL includes different types of behaviors with the common characteristic that they do not behave like FL. Therefore
the Sommerfeld gpZ(eF) definition does not apply anymore because Cm/T|T-0 ¼ gT becomes temperature dependent. Instead of
the fermionic-FL dependence (described in Section 2.3), for systems with 0.3rgTr2 J mol1 K2 the predicted thermal evolution
depends on different spectral distributions of the density of excitations. The most studied are: weak power laws Cm(T)/TpT1 þ l
with lE0.7  0.8, logarithmic p  lnðT=T  Þ or p¼ g|T ¼ 0  a√T (see, e.g., Stewart, 2001).
The logarithmic dependence is the most frequently observed and it is expected to diverge at T ¼ 0 in an algebraic singularity
unless the system undergoes magnetic order, which is more likely according to thermodynamic conditions. The exemplary system
for the p  lnðT=T  Þ dependence is CeCu5.9Au0.1, included in Figure 6 after (Löhneysen, 1995), that exhibits high density of low
lying spin excitations at low temperature. Most of Ce systems showing this behavior can be compared using an universal function
Cm/t ¼  Dlog(t) þ E0, with t ¼ T/T (TpTK), D ¼ 7.2 J mol1 K2 as a scaling constant and E0 accounting for high temperature
contribution like from CEF levels, as presented in Figure 6. The Cm/T|T-0 values for these systems range between 0.5 and
2 J mol1K2.

10
CP / T [J / mol K ]
2

0.1
0.1 1 10
T [K]

Figure 5 (Color online) Exemplary systems for heavy and very HF in a double logarithmic representation. The continuous curve is the fit of
CePd3B0.5 dependence using the Cm/T ¼4.2/(T2 þ 0.75) formula.

16
(Cm / T - E) T* [J / mol K]

12

0
0.01 0.1 1
t = T / T*
Figure 6 (Color online) Normalized logarithmic dependence of different Ce-NFL compounds, with TpTK and E0 accounting for high temperature
contribution like from CEF levels. Dashed line: universal Dlogðt Þ dependence representation.
Magnetic Systems: Specific Heat 11

There is a number of Uranium (5f) systems with itinerant character obeying a power law Cm(T)/TpT1 þ l with lo1 (Stewart,
2001). However, localized Ce, Pr, and Yb (4f) compounds frequently show Cm/Tp1/TQ dependence with a Q41 exponent,
reaching higher values of Cm/T|T-0. Most of these very-HF systems can be compared by using the heuristic formula Cm/T¼ G/
(TQ þ E) (Sereni, 2007) within more than a decade of temperature. Exceptionally, in the case that Q ¼ 2, like, for example, for
CePd3B0.5 with Cm/T|T-0E4.4 J mol1 K2, the lowest temperature limit recovers the FL approximation: CFL/T¼ g  b(T/TK)2, see
Figure 5. A relevant feature observed in the group of very-HF concerns the fact that three Yb systems: YbBiPt, YbCo2Zn20 and
YbCu5xAux (for 0.4rxr0.7), show values of Cm/T|T-0E7 J K2 mol1 (Sereni, 2015 and references therein). Notably, non of
them show magnetic order down to the mK range of temperature despite of their robust magnetic moments. The possibility of an
eventual upper limit for Cm/T|T-0 is discussed in the next Subsection.

5.2 Entropy Bottlenecks


In order to understand the exotic specific heat behaviors emerging in some new RE-intermetallics below a few degrees Kelvin it is
necessary to take into account the thermodynamic constraints imposed by the Third Law of Thermodynamics, which sates that
Sm(T)Z0 as T-0 (Pippard, 1964). Since zero Entropy corresponds to a singlet GS, its derivative ∂Sm/∂T also tends to zero as it
occurs in ordered systems, which is associated to a positive curvature ∂2Sm/∂T240, see in Figure 7 the case of YbCu4Au. Metallic
systems with ∂Sm/∂T¼ g (and ∂2Sm/∂T2 ¼ 0) are possible because Fermi distribution suppresses thermal excitations at T¼ 0. The
third (mathematical) possibility ∂2Sm/∂T2o0 has no physical meaning at T-0 because it would requires an increasing density of
excitations with the consequent divergence of Cm/T|T-0 and an eventual Cm|T-0a0.
These basic postulates on the entropy evolution at very low temperature allow to understand some unexpected specific heat
behaviors in that range of thermal energy. The first example addresses the upper limit of Cm/T|T-0 introduced in the previous
R
Subsection concerning very-HF. For a power law dependence with Q41 (and a Jeff ¼ 1/2), Sm ¼ Cm/TQdT will become larger than
the available degrees of freedom Rln2. This situation is exemplified by the Sm(T) evolution of YbCu4.3Au0.7 with the red curve in
Figure 7. Such a curve was evaluated from the Cm(T)/T dependence at T40.4 K showing that the Rln2 value is reached at TE0.2 K
(i.e., T40). To fulfill the thermodynamic condition that SmZ0 at T ¼ 0, the system is driven to follow a tangent Sm(T) trajectory,
indicated by the gT blue line in Figure 7, which corresponds to the maximum compatible slope (at TE0.3 K in this case) that
extrapolates to Sm ¼ 0 at T ¼ 0.
Another example where thermodynamic constraints act on the specific heat behavior at low temperature can be observed in the
way that the DCm jump decrease when Tord decreases. Driven by control parameters like RE-ligands alloying, magnetic field or
pressure, DCm decreases hand in hand with the entropy involved into the ordered phase Sord. The three possible trends are
schematically presented in Figure 8 (Sereni, 2013). Type-I includes the compounds for which Sord-0pTord, pointing to a
Quantum Critical Point QCP at T¼ 0 according to the law of corresponding states (i.e., DCm/Tord ¼ const.). Type-II shows DCm and
Sord decreasing faster than Tord, revealing that their magnetic component vanish at T40. The magnetic systems becoming super-
conductive under high pressure, like CeIn3 (Knebel, 2002), fit into this behavior. Type-III involves the few systems that show DCm
slightly decreasing with Tord with a consequent divergent DCm/Tord ratio and Cm T-0⇏0. This behavior ends at a Critical Point at
finite temperature due an entropy bottleneck at low temperature, as it occurs in Ce2(Ni1xPdx)2Sn and URu2Si2 under high magnetic
field, see (Sereni, 2013) for references.

0 2 4 6 8 10
1.0

0.8


0.6
Sm / RLn2

0.4

0.2

0.0
0.0 0.5 1.0 1.5 2.0
T [K]
Figure 7 (Color online) Entropy analysis of a very-HF system. Red curve: fit to Sm(T) from the T40.4 K range. Blue line: tangent (gT) of Sm(T)
that extrapolates to Sm ¼0 at T¼0. YbCu4Au represents Sm(T) for a magnetically ordered system. Upper T-axis: entropy evolution of the Spin-Ice
system Dy2Ti2O7.
12 Magnetic Systems: Specific Heat

Sord [degr. of freedom]


0
0
Tord [thermal energy]

Figure 8 (Color online) Different trends of the entropy in the ordered phase Sord with the ordering temperature Tord decrease. (I) Ending in a
QCP; (II) vanishing at finite temperature and (III) ending in a Critical Point at finite temperature.

Another interesting manifestation of thermodynamical constraints concerns a Cm(T)/T anomaly observed at TE1 K in a
number of Ce and Yb systems which do not order magnetically despite of their robust magnetic moments. These anomalies show a
number of common features like: (1) the coincident TmaxE1 K; (2) an important tail of Cm/T at T4Tmax described by a power law:
Cm/T ¼ G/(TQ þ A) with Q ¼ 270.2 and 0rAr2 K (Sereni, 2015); and (3) a significant amount of entropy collected within the
Cm(T)/T tail at TZTmax (c.f., SmE0.7Rln2) that coincides with the value reported for frustrated Spin-Ice systems. The origin of these
anomalies can be attributed to an interplay between the power law-divergent increase of the density of low energy excitations and
the limited amount of GS degrees of freedom. Due to this entropy bottleneck, these systems are impelled to slide into alternative
minima of free energy. This is a continuous transition because no DCm jump is observed, but its derivative D(∂Cm/∂T) shows the
discontinuity expected for a third order transition.
Deviations from the Sm ¼ 0 at T ¼ 0 condition are found in geometrically frustrated AFM systems (Ramirez et al., 1999) because
they cannot access to a singlet GS. Their remanent entropy, SRa0 at T ¼ 0, was computed by Pauling for the so-called spin-ice
systems. In this case, the full entropy Sm ¼ Rlnð2J þ 1Þ at high enough temperature (usually Rln2) has to be taken as the proper
reference value. In the exemplary spin-ice Dy2Ti2O7, a SR E0:3Rln2 value is obtained (see Figure 7) in agreement with Pauling
predictions of Sm ¼ 1=2Rlnð3=2Þ (Ramirez, 1999).

6 Concluding Remarks

The present overview on specific heat properties of magnetic systems at low temperature proves the vast applicability of this
thermal parameter to the characterization of new materials. The possibility to discriminate between different types of excitations
within a broad range of energy down to very low energy, places this experimental tool in the category of a highly sensitive thermal
spectrometer.
Hand in hand with the Entropy, specific heat is the necessary parameter to define the Free Energy that governs any physical
phenomena including itinerant and localized electrons, magnetic and nonmagnetic systems together with their associated phase
transitions or anomalies. As an ‘extensive’ parameter, specific heat provides information about the amount of degrees of freedom
involved in different processes. On the other hand, its thermal dependence reveals the intensity of the related interactions through
the ordering temperatures or the position of the respective maxima in different types of anomalies like Schottky, Spin Glasses,
Kondo, or low dimensional systems.
At the limit of very low temperature, the study of emerging phenomena, like divergent density of excitations in very HF,
Quantum Critical Points or Geometrically Frustrated configurations, are mostly based on specific heat. Since basic thermodynamic
laws dominate the physics behavior at this range of energy, also the knowledge of the thermal evolution of the entropy becomes a
relevant complement.

References

Akhieza, A.I., Bar'yakhtar, V.G., Peletminskii, S.V., 1968. Spin Waves. Doniach, S. (Ed.), Amsterdam: North Holland Pub. Co.
Bauer, E., Hauser, R., Gratz, E., et al., 1993. CePd2Ga3: A new ferromagnetic Kondo lattice. Z. Phys. B 92, 411–416.
Blanco, J.A., Gomez-Sal, J.C., Fernandez, J.R., et al., 1994. Specific Hat of GdNi1xCux compounds. Solid State Commun. 89, 389–392.
Blundell, S., 2004. Magnetism in Condensed Matter. New York: Oxford University Press.
Bouquet, F., Wang, Y., Wilhelm, H., Jaccard, D., Junod, A., 2000. Calorimetric investigation of CeRu2Ge2 up to 8 GPa. Sol. State Commun. 113, 367–371.
Continentino, M.A., 2000. Wilson and Kadowaki-Woods ratios in heavy fermions. Eur. Phys. J. B 13, 31–35.
De Visser, A., Menovsky, A., Franse, J.J.M., 1987. Physica 147B, 81–160.
Fisher, M.E., 1967. The theory of equilibrium critical phenomena. Rep. Prog. Low Temp. Phys. 30, 615–730.
Magnetic Systems: Specific Heat 13

Fogle, W.H., Ho, J.C., Phillips, N.E., 1978. Concentration and temperature dependence of the magnetic heat capacity of dilute CuMN. J. de Physique, Colloque C6 (39),
C6–901.
Gopal, E.R., 1966. Specific Heats at Low Temperatures. London: Heykood Books.
Gschneidner, K.A., 1964. Physical properties of metallic and semi-metallic elements. Solid State Phys. 16, 275.
de Jongh, J.L., Miedema, A.R., 1974. Experiments on simple Magnetic Systems. Adv. Phys 23, 1–269
Knebel, G., Braithwaite, D., Canfield, P., Laperot, G., Flouquet, J., 2002. The quantum critical point in CeIn3. High Press. Res. 22, 167.
Koskenmaki, D.C., Gschneidner Jr., K.A., 1978. Cerium. In: Gschneidner Jr., K.A., Eyring, L. (Eds.), Handbook on the Physics and Chemistry of Rare Earths, vol 1. Amsterdam:
North-Holland Publishing Company.(Chapter 4).
Löhneysen, H.V., 1995. Non-Fermi-liquid behavior in heavy-fermion systems. Physica B 206&207, 101–107.
McMillan, W.L., 1968. Transition temperature of strong-coupled superconductors. Phys. Rev. 167, 331.
Meijer, P.H., Colwell, J.H., Shah, B.P., 1973. A note on the morphology of heat capacity curves. Am. Jour. Phys. 41, 332.
Mydosh, J.A., 1993. Spin Glasses: An Experimental Introduction. London: Taylor & Francis.
Paschke, C., Speck, C., Portisch, G., Löhseysen, H.V., 1994. Magnetic ordering and Kondo compensation in the ternary heavy-Fermion compound CeCu5Au. J. Low Temp.
Phys. 97, 229.
Phillips, N.E., 1972. Low temperature heat capacity of metals. In: Schuele, D.E., Hoffman, R.W. (Eds.), CRC Critical Reviews in Solid State Sciences. Cleveland: CRC Press
Division.
Phillips, N.E., Ho, J.C., Smith, T.F., 1968. Heat capacity of a-cerium at a pressure of 11 kbar, between 0.3 and 6 K. Phys. Lett. A 27, 49.
Pippard, A.B., 1964. Elements of Classical Thermodynamics. Cambridge: Cambridge University Press.
Ramirez, A.P., Hayashi, A., Cava, R.J., Siddharthan, R., Shastry, B.S., 1999. Zero-point entropy in ‘spin ice’. Nature 399, 333–335.
Sereni, J.G., 1991. Low temperature behavior of Cerium compounds. In: Gschneidner Jr., K.A., Eyring, L. (Eds.), Handbook on the Physics and Chemistry of Rare Earths, vol
15. Amsterdam: Elsevier. (Chapter 98).
Sereni, J.G., 1995. Characteristic concentrations in Cerium ground state transformations. Phys. B 215, 273–285. and Systematics on Cerium magnetic transformations. J.
Alloys Compd 207 and 208, 229−236.
Sereni, J.G., 2001a. Thermodynamical Analysis of the Magnetic Phase Diagrams of Ce Systems. J. Phys. Soc. Japan 70, 2139–2150.
Sereni, J.G., 2001b. The specific heat of magnetic systems. In: Buschow, K.H., Gratz, E. (Eds.), Encyclopaedia of Materials: Science and Technology. Elsevier Science,
pp. 4986–4994.
Sereni, J.G., 2007. Peculiar thermal features of Ce-systems around their critical points. J. Low Temp. Phys. 147, 179.
Sereni, J.G., 2013. Thermodynamic analysis of the quantum critical behavior of Ce-lattice compounds. Philos. Mag. 93, 409–433.
Sereni, J.G., 2015. Entropy bottleneck at T-0 in Ce and related systems. J. Low Temp. Phys. 179, 126–137. and references therein.
Sereni, J.G., Beaurepaire, E., Kappler, J.P., 1993. Effects of volume and electronic concentration on the Ce(Pd1−xMx) compounds (M ¼ Ni, Rh, and Ag). Phys. Rev. B 48, 3747.
Sereni, J.G., Huber, T.E., Luengo, C.A., 1979. Low temperature specific heat of Th−Gd spinglass. Solid State Commun. 29, 671–673.
Sparn, G., Donnevert, L., Hellmann, P., et al., 1998. Pressure studies near quantum phase transitions in strongly correlated Ce systems. Rev. High Pressure Sci. Technol 7,
431–436.
Stanley, H.E., 1971. Introduction to Phase Transition and Chemical Phenomena. Oxford, Great Britain: Clarendon Press.
Stewart, G.R., 1984. Heavy Fermion systems. Rev. Mod. Phys. 56, 755–778.
Stewart, G.R., 2001. Non-Fermi-liquid behavior in d- and f-electron metals. Rev. Mod. Phys. 73, 797–855.
Sundström, L.J., 1978. In: Gschneidner Jr., K., Eyring, L. (Eds.), Handbook on the Physics and Chemistry of Rare Earths, vol 2. North Holland Pub. Co.(Chapter 5).
Tari, A., 2003. The Specific Heat of Matter at low temperatures. London, Great Britain: Imperial College Press.

Further Reading

Bredl, C.D., Steglich, F., Schotte, K.D., 1978. Specific heat of concentrated Kondo systems. Z. Phys. B 29, 327–340.
Curlik, I., Giovannini, M., Sereni, J.G., et al., 2014. Extremely high density of magnetic excitations at T-0 in YbCu5−xAux. Phys Rev. B 90, 224409.
Domb, C., Miedema, A.R., 1964. Magnetic transitions. In: Gorter, C.J. (Ed.), Progress in Low Temperature Physics, vol IV. Amsterdam: North-Holland Publishing Company.
(Chapter VI).
Löhneysen, H.v., Roch, A., Vojta, M., Wölfe, P., 2007. Fermi-liquid instabilities at quantum phase transitions. Rev. Mod. Phys. 79, 1015.
Sereni, J.G., 1998. Comparative study of magnetic phase diagrams of Ce compounds. J. Phys. Soc. Japan 67, 1767–1775.
Sereni, J.G., Gomez Berisso, N., Betancourth, D., et al., 2014. Evidence for a dimensionality crossover at the disappearance of magnetism in the Kondo lattice alloy CeCo1−
xFexSi. Phys. Rev. B 89, 035107.

You might also like