Novel Heavy Fermion Behavior in Praseodymium-Based Materials: Experimental Study of PrOs4Sb12

Material Information

Novel Heavy Fermion Behavior in Praseodymium-Based Materials: Experimental Study of PrOs4Sb12
ROTUNDU, COSTEL REMUS ( Author, Primary )
Copyright Date:


Subjects / Keywords:
Electric fields ( jstor )
Electrical resistivity ( jstor )
Fermions ( jstor )
Ground state ( jstor )
Ions ( jstor )
Magnetic fields ( jstor )
Magnetism ( jstor )
Magnets ( jstor )
Specific heat ( jstor )
Symmetry ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Costel Remus Rotundu. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
660033376 ( OCLC )


This item has the following downloads:

rotundu_c ( .pdf )



















































































































































Full Text







Copyright 2007


Costel Remus Rotundu

To my parents, Constantin and Elena Rotundu,

for their sacrifices to ensure my education.


I am dedicating this work to my dearest human beings, my parents Elena and

Constantin Rotundu. I cannot find an adequate way to express my love and gratitude to

them. I thank them for their infinite support in order to complete my education. I am

grateful to my mother and my brother Romulus Neculai Rotundu for believing in me.

I owe much to my adviser, Dr. Bohdan Andraka. He has been an incredible source

of guidance and inspiration. He was a great adviser with endless patience. My education

would not have been possible without his financial help (through DOE and NSF). I am

deeply indebted to him. Special thanks go to Prof. Yasumasa Takano for teaching me so

many experimental tricks; and for discussions, support, and great collaboration, especially

at the National High Magnetic Field Laboratory (NHMFL). He was an endless source of

energy. I would like to thank Prof. Gregory R. Stewart for letting me use his laboratory.

I thank my other supervisory committee members (Profs. Bohdan Andraka, Gregory R.

Stewart, Yasumasa Takano, Pradeep Kumar and Ion Ghiviriga) for reading this work and

for their advices. I received help with many experiments at NHMFL and our lab from Dr.

Hiroyuki Tsujii. I thank Drs. Jungsoo Kim and Daniel J. Mixson II for their technical

advice. Other people in the field I would like to acknowledge are Prof. Peter Hirschfeld,

the finest professor I ever had, who gave me insight on the theory of condensed matter

physics; and Drs. Eric C. Palm and Tim P. Murphy for their help and support over more

than 4 years at the SC\ I / NHMFL. I thank Center of Condensed Matter Sciences for the

financial support through the Senior Graduate Student Fellowship. Last and not least, I

would like to thank my high school physics teacher (Dumitru Tatarcan), who encouraged

and guided my first steps in physics.



ACKNOW LEDGMENTS .................................

LIST O F TABLES . . . . . . . . . .

LIST OF FIGURES . . . . . . . . .

A B ST R A C T . . . . . . . . . .


1 INTRODUCTION ..................................

2 THEORETICAL BACKGROUND .........................

2.1 The Crystalline Electric Field (CEF) for Cubic Group ............
2.2 Conduction Electron Mass Enhancement (m*) Mechanism in PrOs4Sb12 .
2.2.1 Quadrupolar Kondo Effect ....................... Thermodynamic Properties of the Quadrupolar Kondo


M odel . . . . . . . Relevance for the Case of Pr3+ Ion in PrOs4Sb12 .
Fulde-Jensen Model for m* Enhancement in Pr Metal .
Fluctuations of the Quadrupolar Order Parameter ......

3 PROPERTIES REVIEW OF THE PrOs4Sb12 .............

3.1 Crystalline Structure .. .. .. .. ... .. .. .. .. ... .
3.1.1 Rattling of Praseodymium Atom .. .............
3.1.2 V alence . . . . . . . .
3.1.3 Crystalline Electric Fields .. ...............
3.2 Normal-State Zero-Field Properties .. ...............
3.2.1 Specific H eat . . . . . . .
3.2.2 de Haas van Alphen Measurements .. ...........
3.2.3 R esistivity . . . . . . .
3.2.4 DC Magnetic Susceptibility .. ................
3.3 The Long-Range Order in Magnetic Fields .. ...........
3.4 Superconductivity . . . . . . .
3.4.1 Unconventional Superconductivity .. ............ The Double Transition .. ............. Temperature Dependence of Specific Heat Below Tc Nuclear Magnetic Resonance (Sb NQR) ...... Muon Spin Rotation (pSR) .. ..........
3.4.2 Conventional Superconductivity .. ............. Nuclear Magnetic Resonance (pSR) .. ....... Penetration Depth Measurements (A) by pSR . Low-Temperature Tunneling Microscopy ......


4.1 The Samples: Synthesis and ('!C i o :terization ....... ......... 49
4.1.1 Synthesis ................. . . .... 49
4.1.2 X-Rays Diffraction ('!I i o :terization ...... . ..... 51
4.2 Specific Heat Measurements .................. ..... .. 52
4.2.1 Equipment .................. ............. .. 52 Cryogenics. .................. .. .... .. .. 52 Sample Platform ................ .... .. 55
4.2.2 Thermal Relaxation Method ................. . .. 55
4.3 Magnetic Measurements .................. ......... .. 57
4.3.1 DC Susceptibility .................. ......... .. 57
4.3.2 AC Susceptibility .................. . .. 58
4.4 Resistivity .................. ................. .. 59

5 MATERIALS CHARACTERIZATION ................ .... .. 65

6 PrOs4Sb12 . . . . . . ... . . . 67

6.1 Investigation of CEF Configuration by Specific Heat in High Magnetic Fields 67
6.2 Magnetoresistance of PrOs4Sbi2 ................... ... 73

7 PrlxLaxOs4Sbl2 ............... ............ .. 93

7.1 Lattice Constant ............... ........... .. 93
7.2 DC Magnetic Susceptibility ............... .... .. 94
7.3 Zero Field Specific Heat .......... . . .... 95
7.3.1 Specific Heat of PrOs4Sb12: Sample Dependence . .... 95
7.3.2 Zero Field Specific Heat of Pr1_-LaOs4Sb12 . . 96 Evolution of T, with the La Doping . . ... 97 The Discontinuity in C/T at T . . ...... 98 The Schottky Anomaly .............. .. .. 99
7.4 Specific Heat in Large Fields .................. ..... 101
7.5 Magnetoresistance of Pr1_-La1Os4Sb12 .............. .. .. 103
7.6 Upper Critical Field H,2 .................. ....... .. 107
7.6.1 AC Susceptibility ....... . . ........ 107
7.6.2 Determination of Hc2(T) by Specific Heat Measurements in Small
M agnetic Fields .................. ........ .. 108

8 CONCLUSION .................. ................. .. 136

REFERENCES .................. ................ .. .. 139

BIOGRAPHICAL SKETCH ................... . ... 145

Table page

2-1 The relevant states for the quadrupolar Kondo effect ...... . .... 23

3-1 The a values reported by different groups, extracted from fits of specific heat
below T,. ....... .............. .............. .. .. .. 38

3-2 The a values reported by different groups from other measurements than specific
heat. . . . . . . .. . . . . 38

Figure page

2-1 Cubic point group symmetry Th. .................... ...... 27

2-2 Lea, Leask, and Wolff's representation of CEF for J 4 (Lea et al., 1962). . 28

2-3 Representation of the U4+ ions in cubic symmetry undergoing quadrupolar Kondo
effect . . . . . .. . . . 28

2-4 Mapping of the quadrupolar Kondo Hamiltonian onto the two-channel Kondo
m odel. .. ... .. .. .. ... . . .. .. .. ....... .. .. 29

2-5 S, C, C/T, and x versus T/TK of the quadrupolar Kondo model (Sacramento
and Schlotmann, 1991). .................. .. ........ 29

3-1 Crystal structure of PrOs4Sb12. ................ ....... 41

3-2 Fits of X(T) to either F3 or F1 CEF ground state, and C fitted by a two-level
Schottky anomaly (Bauer et al., 2002). .................. .... 42

3-3 Fits of X(T) to either F3 or F1 CEF ground state model, calculated S(T) in both
F3 and F1 CEF ground state models (Ti 11i i et al., 2003), and the measured
S(T) (Aoki et al., 2002). .................. .. ....... 43

3-4 p(T), x(T) and C(T) of PrOs4Sb12 (Bauer et al., 2002). .. . ..... 44

3-5 Fermi surface of PrOs4Sb12 (Sugawara et al., 2002) ............... ..45

3-6 H-T phase diagram of PrOs4Sb12 by dM(T)/dT and dM(H)/dH
measurements (T iv ii i et al., 2003) .................. .... 46

3-7 C(T) of PrOs4Sbl2 (Vollmer et al., 2003; M6asson et al., 2004) and the real part
of the ac susceptibility (\ !' i.- .. et al., 2004) presenting double SC transition.. 46

3-8 The two superconducting phases of PrOs4Sb12: phase A and phase B (Izawa et
al., 2003). The plot of the SC gap function with nodes for both phases (\! ,.1:
et al., 2003) ................... ................... .. 47

3-9 Two SC transitions in 3(T) of PrOs4Sb12 (Oeschler et al., 2003) . .... 47

3-10 T dependence of the rate 1/Ti at the 2vQ transition of 123Sb for PrOs4Sb12 and
LaOs4Sb12 (Kotegawa et al., 2003). .................. .... 48

3-11 Tunneling conductance between PrOs4Sb12 and an Au tip (Suderow et al., 2004).
The gap is well developed with no low-energy excitations, sign of no nodes in
the Fermi surface gap .............. ............ .. 48

4-1 Picture of PrOs4Sb12 large ( i-- .I (about 50 mg). ............... 60

4-2 PrOs4Sb12 samples prepared for (left panel) resistivity and (right panel) specific
heat measurements. .................. ............ ..60

4-3 Schematic view of the 3He cryostat used in the measurements performed at
University of Florida. ............... ............ .. .. 61

4-4 Schematic view of the calorimeter used in the Superconducting Magnet 1 (SC' \
1), National High Magnetic Field Laboratory. .................. 62

4-5 The sample-platform/Cu-ring assembly. .............. ... 63

4-6 Specific heat C measurement process using the relaxation time method. . 64

5-1 y(T) of PrOs4Sb12. The high temperature effective moment is 3.65PB, very close
to the one corresponding to free Pr3+, which is 3.58PB. . . ..... 66

5-2 x(T) of the non-f equivalent LaOs4Sb2. ................ ..... 66

6-1 C of PrOs4Sb12 in fields up to 8 T for H//(1 00) (upper panel), and H-T phase
diagram in fields up to 8 T for H//(1 00) (lower panel) (Aoki et al., 2002). .. 78

6-2 C of PrOs4Sbl2 in 10 and 12 T in the vicinity of FIOP transition for H//(10 0). 79

6-3 C of PrOs4Sbl2 in magnetic fields 13, 13.5, and 14 T, for //(10 0). . 80

6-4 C of PrOs4Sbl2 in 16, 20, and 32 T, for //(10 0). .. . . ....... 81

6-5 H-T phase diagram of PrOs4Sbl2 for H//(1 00) (H>8 T). . . .. 82

6-6 Zeeman effect calculations for PrOs4Sbl2 in the F1 CEF ground state scenario. .83

6-7 Zeeman effect for PrOs4Sb12 in the F3 CEF ground state scenario. . ... 84

6-8 C of PrOs4Sb12 for f//(1 10). .................. ......... .. 85

6-9 C of PrOs4Sbl2 in H=12 T, for H//(1 00) (upper panel), and H//(1 10) (lower
panel) ...................... ........ ... .... ... 86

6-10 H-T phase diagram of PrOs4Sb12 for H//(1 1 0) (H>8 T). . . ... 87

6-11 p versus T2, and p versus T for PrOs4Sbl2 .................. .. 87

6-12 p(T) of PrOs4Sbl2 in 3, 10, 15, 16, 17, and 18 T, between 20 mK and about 0.9 K. 88

6-13 p versus T2 of PrOs4Sbl2 for 3.5, 5.5, 7, 10, and 13 T . . ...... 89

6-14 a (p po+aT") versus H for PrOs4Sbl2 fields up to 18 T(upper panel). The residual
resistivity po(H) (lower panel). .................. .... 90

6-15 A (p=po+AT2) versus H. .................. .. ....... 91

6-16 The calculated p(H) of PrOs4Sb12 (Frederick and Maple, 2003), for both F3 and
FL scenarios .................. ................... .. 92

7-1 X-ray diffraction patterns of Pr1_-La0Os4Sb12 versus La content x for x=0, 0.1,
0.2, 0.4, and 1 .................. ................. .. 113

7-2 Lattice constant a of Pr1_-LaOs4Sb12 versus La content x. . .... 114

7-3 x(T) of Pr1_-LaOs4Sb12 normalized to Pr mole between 1.8 and 10 K, measured
in 0.5 T ..................... ........ ... .... 114

7-4 X(T) of Pro.33Lao.670sSb12 versus T. The Curie-Weiss fit at high temperature
(T>150 K) gives pMff 3.62/B/Pr mole. ................ ...... 115

7-5 C/T versus T for three different PrOs4Sb12 samples from different batches. 116

7-6 C/T versus T2 above T, of LaOs4Sb12 .................. .. 117

7-7 C/T versus T near T, for Pr1_-LaxOs4Sb12 for x=0, 0.05, 0.1, and 0.2. ..... ..118

7-8 C/T versus T near T, of Pr1_-La0Os4Sb12 for x>0.3 .............. .119

7-9 T, versus x of Pr1_-La0Os4Sbl2 .................. ........ .. 120

7-10 Total A(C/T) at T, and Xo versus x of Pr1_-La0Os4Sbl2 for 0>x>l. Xo is X at
1.8 K from Fig. 7-3. .................. ............. 120

7-11 C/T versus T of Pro.33Lao.670s4Sb12 fitted by F1 F5 Schottky. . ... 121

7-12 C/T versus T of Pro.33Lao.670s4Sb12 fitted by F3 F5 Schottky. . ... 121

7-13 C/T versus T of Pro.33Lao.670s4Sb12 fitted by singlet-singlet Schottky. ..... .122

7-14 C for x=0 (10 T) and 0.02 (8 and 9.5 T). .................. ... 122

7-15 f-electron specific heat of Pro.9Lao.1Os4Sb12 in magnetic fields. . ... 123

7-16 f-electron specific heat of Pro.sLao.20s4Sb12 in magnetic fields. . ... 123

7-17 f-electron specific heat of Pro.4La0.60s4Sb12 in magnetic fields. . ... 124

7-18 H-T phase diagram from C measurements for x=0, 0.02, 0.1, and 0.2. ....... .124

7-19 p(H) of Pro.95Lao.o50s4Sb12 at T 20 mK for H//I and HII (I//(00 1)). ... .125

7-20 p(H) of Pro.95Lao.osO4Sbi2 for HlI//(O0 1) at T 20 and 300 mK. ...... .125

7-21 p(H) of Pro.95Lao.0osO4Sb12 for Hil//(0 0 1) at T-20, 310, and 660 mK. . 126

7-22 p versus T2 of Pro.7La0.30s4Sb12 in 0 and 0.5 T. ................ 127

7-23 p versus T2 of Pro.7La0.30s4Sb12 in 9 and 13 T. ...... . . 128

7-24 p(H) of Pro.7Lao.30s4Sb12 for H parallel with all three ( i --1 1,graphic directions
and 1//(001) ................... .... ...... ...... 129

7-25 p(H) of Pro.7Lao.3OS4Sb12 when H//I//(O0 1) at 20, 310, 660, and 1100 mK. 129

7-26 p(H) of Pro.33Lao.670s4Sb12 at 0.35 K. ............. .... 130

7-27 AC susceptibility versus T/T, of Prl_-LaOs4Sbl2, for x=0, 0.05, 0.4, 0.8, and 1. 131

7-28 C/T versus T near T, for two PrOs4Sb12 samples from different batches in low
magnetic fields .................. ................. .. 132

7-29 C/T versus T, near Tc, of Pro.95Lao.o50s4Sb12 in low magnetic fields ....... .133

7-30 C/T versus T, near Tc, of Pro.9Lao.lOs4Sb12 in small magnetic fields. ..... ..133

7-31 C/T versus T, near Tc, of Pro.7Lao.30s4Sb12 in magnetic fields. . ... 134

7-32 -dH2V/dT versus x. .................. ............ 134

7-33 -dH21/dT/Tc versus x with the critical x,,r0.25. The inset shows A(C/T)
at T, with x,,r 0.25-0.3. .................. ........... 135

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Costel Remus Rotundu

May 2007

C'!h ,i: Bohdan Andraka
Major Department: Physics

PrOs4Sb12 is the first discovered Pr-based heavy fermion metal and superconductor.

Our high magnetic field specific heat measurements provided clear evidence for the

non-magnetic singlet < iv-i iii., electric field (CEF) ground state. This CEF ground

state precludes the conventional Kondo effect as the origin of the heavy fermion

behavior. The superconductivity in PrOs4Sb12 is unconventional, as inferred from

the double superconducting transition in the specific heat. Prl_-La1Os4Sb12 (0
crystals were synthesized and investigated in order to provide additional evidences for a

postulated CEF configuration, to discriminate between different conduction electron mass

enhancement (m*) mechanisms proposed, and to provide insight into the nature of the

superconductivity. Lanthanum doping induces anomalously small increase of the lattice

constant. The specific heat results in high magnetic fields indicated that CEF scheme

is unaltered between x=0 and at least 0.2, followed by an abrupt (but small) change

somewhere between x=0.2 and 0.4. Magnetoresistance measurements on La-doped samples

were consistent with a singlet CEF ground state of Pr. Investigation of the specific heat

discontinuity at T, and of the upper critical field slope at T, indicated that the electronic

effective mass, m*, is strongly reduced with x, between x=0 and xce0.2-0.3, followed by

a weak dependence on x for x>xr. Therefore, we have postulated that single-impurity

type models cannot account for the heavy fermion behavior of PrOs4Sb12. Investigation of

the magnetic phase diagram and magnetoresistance provided strong correlations between

a closeness to the long-range order (antiferroquadrupolar type) and m*, -ii-_-, -ri-; a

possibility of fluctuations of the antiferroquadrupolar order parameter responsible for

m* enhancement. Lanthanum has very weak effect on the superconducting transition

temperature in a stark contrast to other known heavy fermion superconductors. The study

of superconductivity provided constraints on proposed theoretical models, including the

two band model.


In the rare earth (Ce,Yb)- and actinide (U,Np)-based alloys the electronic states

have an energy orders of magnitude smaller than in ordinary metals, and since c(k)

h2k2/2m*, the effective mass m* is orders of magnitude larger than the free-electron value,

hence the term heavy fermion. There are several excellent experimental and theoretical

reviews [1-5] on heavy fermions. One hallmark of the heavy fermion character is the large

Sommerfeld coefficient 7 of the specific heat. The specific heat of metals in the normal

state at low temperature is approximated by C= T+3T3, where 7T is the electronic

specific heat and 3T3 is the lattice (Debye) contribution. For a normal metal 7 is of

order of 1 mJ/K2 mol, and for heavy fermion is from several hundred to several thousand

mJ/K2 mol. The magnetic susceptibility X at high temperatures follows the Curie-Weiss

form X=C/(T+Ocw), where C is a constant, and 6cw is the Curie-Weiss temperature.

At low temperatures X(0) ranges from ~10 to 100 memu/mol. In the 1ni iii lly of heavy

fermion metals, the electrical resistivity p at very low temperatures has a T2 dependence:

p po+AT2, where po is the residual resistivity and A is on the order of tens of /pcm/K2,

much larger than that of normal metals.

There are about 20 heavy fermion systems that are superconductors and almost all of

them are Ce- or U-based (there is one Pu-based heavy fermion superconductor: PuCoGas

The filled skutterudite PrOs4Sb12 is the first discovered Pr-based heavy fermion

compound that is a superconductor [7].

In the conventional heavy fermions, the only microscopic theories somewhat successful

in accounting for the effective mass enhancement (m*) as measured by the specific heat

are the S=1/2 and S=3/2 Kondo models. These models were initially proposed for

Ce-based systems, whose effective degeneracies of f-electrons in ( i-I ,11 iiI" electric fields

are either 2 or 4. The Kondo effect in these systems is anomalous because of strong

spin-orbit coupling. There is one f electron per Ce atom and according to Hund's rules

the total angular moment is J=5/2, which corresponds to the degenerate 6 level case.

Crystalline electric fields split this multiple either in a) 3 doublets or b) one doublet and

one quartet. Case b) can be only for cubic symmetries. Thus Ce-based heavy fermions

with a doublet CEF ground state of Ce are described by the S=1/2 Kondo model while

those with Ce in a quartet CEF ground state are described by S=3/2 Kondo model. Thus

any understanding of heavy fermion behavior needs clarification of the CEF ground state.

Unfortunately, CEF scheme is not known for U-based heavy fermions. Furthermore the

valence of U (i.e., whether the electronic configuration is f2 or f3) is not known. The high

temperature effective moments for f2 and f3 configurations that could be extracted from

the high temperature susceptibility are almost identical. Studies of CEF's done directly by

inelastic neutron scattering [8, 9] are more consistent with the f2 configuration, allowing

for a similar CEF scheme as that for Pr. Therefore, the investigation of PrOs4Sb12 with

Pr having 2 f-electrons might be relevant and help to the understanding of the large

class of U-based heavy fermions, since CEF configurations are usually known for Pr. The

non-magnetic < i lv-I !i i electric field ground state (thought as either singlet or doublet

[7]) excludes the conventional Kondo effect as the origin of the l. i, -i-rermion behavior

in PrOs4Sb12, which is considered to be the source of heavy fermion behavior in Ce- and

U-based metals. The superconductivity in PrOs4Sb12 is unconventional, but different from

that in Ce- and U-based materials. Marks of the unconventionality of superconductivity

can be inferred from the double superconducting transition and power low dependence of

the specific heat below the transition.

The main goals of this work are:

to settle the crystalline electric field ground state in PrOs4Sb12,

to bring further evidences of the heavy fermion state in PrOs4Sb12,

to differentiate between several models proposed for the conduction electron

mass enhancement (m*) and to study the relationship between the correlation and the


The outline of the dissertation is as follows: Chapter 2 presents the theoretical

framework of this thesis. This C'!i lpter begins with the theory of crystalline electric field

for the point group symmetry Th, followed by a presentation of the quadrupolar Kondo

effect. Other proposed models of the conduction electron mass enhancement are also

discussed. ('!i lpter 3 review the essential properties of PrOs4Sbl2. C'!i lter 4 gives a

brief description of the apparatus and experimental methods used. A characterization

of the materials synthesized and measured is given in Chapter 5. The experimental

data are presented and discussed in ('!i lpters 6 and 7. ('!i lpter 6 focus on PrOs4Sbl2

itself (specific heat and resistivity in high magnetic fields). ('!i lpter 7 presents the

study of Prl_-LaOs4Sbl2, 0
measurements are discussed. Finally, ('! Ilpter 8 summarizes the main findings and

contributions to the field of Pr-based L i,',-i--Fermions.


2.1 The Crystalline Electric Field (CEF) for Cubic Group

In rare earth compounds, the < ,i-- i i.: electric fields are responsible for a wide

v ii. I v of strongly correlated electron behaviors. The 4f-electrons in a rare earth ion

experience an electrostatic crystal field potential created by the surrounding electric charge

distribution (of the neighbor ions). The potential reflects the local point symmetry of the

site of the rare earth ion. In the point-charge ionic model the CEF potential at position F

due to the surrounding atoms is

VCEF (r) q (2-1)

where qj is the charge at the j"h neighboring ion, at Rj. If the magnetic ion has charge q,

at r, then the 1

HCEF q- (2-2)
i j i RjI

The sum Yi is taken over electrons in unfilled shells [10].

The CEF potential can be evaluated in terms of Cartesian coordinates or in terms

of spherical harmonics. Hutchings [10] evaluated the potential (2-1) for the simplest 3

arrangements of charges giving a cubic crystalline electric field. The three cases analyzed

were when the charges are placed at the corners of an octahedron (sixfold coordination),

at the corners of a cube (eightfold coordination), and at the corners of a tetrahedron

(fourfold coordination). In Cartesian coordinates the potential (2-1) can be written as [10]

V(x,y,z) = C4 + 4 + z4) 4] + D[(6 + + 6)

+ 5 2 4 2+ X4 2 X4 2+ 4 2+ Z24 24 6, (2-3)
+ -(xy +z +y +y +x +z )-y]), (23)
4 14

where d is the distance of the point charge q from the origin in each 3 cases. C4 and

D6 are -70q/(9d5) and -224q/(9d7) for the eightfold coordination, +35q/(4d5) and

-21q/(2d7) for the sixfold coordination, and -35q/(4d5) and -112q/(9d7) for the fourfold
coordination respectively.

In the spherical coordinates the same potential is written [10] as

V = D4{Y + [Y44( Y-4(, )]} + D'{Y6~ ( )

[Y ( ) + -4(0 )]}, (2-4)

where D' and D' are -56q v/(27d5) r4 and +32q /13/(9d7).r6 for the eightfold

coordination, +7q //(3d5).r4 and +3q V7/13/(2d7)r6 for the sixfold coordination, and

-28qv//(27d5).r4 and +16q G\ /13/(9d5).6 for the eightfold coordination respectively.
The potential contains therefore terms of order 4 and 6 in coordinates. In general, the less

symmetric is the site, the more potential terms occur in the expansion.
There are 2 general rules that can tell us the number of nonzero terms in the CF

potential. If there is a center of inversion at the ion site there will be no odd-n terms.

Secondly, if the z axis is not an m-fold axis symmetry, the potential will contain V,, [10].

However, calculating the potential terms in Cartesian coordinates and even in

spherical coordinates is tedious. A more convenient method is the so called operator

equivalent or Stevens' operator technique [11, 12]. The Hamiltonian (2-2) is of form

HcEF=- i elV(xi, Yi, Z). If f(x,y, z) is a Cartesian function, in order to find the
equivalent operator to such terms as 3 f(xj,yz,zt), the coordinates x, y, and z are

replaced by angular momentum operators J, Jy, and J respectively, taking into account
the non-commutativity of Jjs. This is done by replacing products of x, y, and z by

combinations of JJs divided by the total number of combinations. As an example we can


(x- 6xf + ) [(x ,)4 (:/ -)4 ,, )4]/2 3j(,r4 + J4]
i ii i Yi
= p(r4)O{ (2-5)

where J6=J JiJ,.

The Hamiltonian is

HCEF C4[(4 +Y+ z4) r4] + D6[(6 +6 + z6)
15(2 4 +2z4 y2 4 y2z4 z24 2 14)_ 6]. (2-6)
4 14

Using the equivalent operator representation, the Hamiltonian will be [10]

HCEF (C4/20)3j(r4) [0 + 50 (D6/224),(r6) [O 2110], (2-7)


HCEF = B[O + 50] + B [O 210o], (2-8)

where B and BO are +71elq3j(r4)/(18d5) and -lelq'y(r6)/(9d7) for eightfold coordination,
-7|1elqj(r4) /(16d5) and -31elqyj(r6)/(64d7) for sixfold coordination, and +71|elqj3(r4)/(30d5)
and -lelq'j(r6) /(d7) for fourfold coordination respectively. Also, (r4) and (r6) are the
mean fourth and sixth power of the radii of the magnetic electrons, and the multiplicity
factors aj, 3j, and 7j are for Pr3+ (2) -22 13/(32 5 112), -22/(32 5. 11 2), and
-22 17/(34 5 7.112 13) respectively [10]. Also,

o0 35J4 [30J(J + 1) 25]J 6J(J + 1) + 3J2(J + 1), (2-9)

O = 1/2(J +4), (2-10)

0 = 231J6 105[3J(J + 1)- 7]J4 + [105J2( + )2- 525J(J + 1)

+ 294]J, 5J3(J + )3 + 40J2(J + )2 60J(J + 1), (2-11)

O4 1/4[11J J(J + 1) 38](J4 4) 1/4 4)

x [11J J(J + 1)- 38]. (2-12)

In order to keep the eigenvalues in the same numerical range for all ratios of the
fourth and sixth degree terms, F(4) and F(6) are introduced [13]. The Hamiltonian is
written as
4+5044 4 21 04
HCEF = BF(4) 4 + B F(6)o6 (2-13)
F (4) F (6)
In order to cover all possible value of the ratio between the fourth and sixth degree terms
are introduced the scale factor W and the parameter x, proportional to the ratio of the
two terms

B2F(4) Wx, (2-14)

BF(6) W(1- xl), (2-15)

where -l O + o50 O 21O]0
HCEF = W x + (t |x|) (2-16)
F(4) F(6)

The term in the square bracket is a matrix whose eigenvectors and eigenvalues (crystal
field energies levels) are determined by usual diagonalization.
Praseodymium ion Pr3+ in PrOs4Sb12 has a 4f2 configuration, and then the total
angular momentum is J=4. The site symmetry of Pr ions is Th (Fig. 2-1). This is
differentiated from the cubic Oh symmetry by the fact that do not contain two types
of symmetry operations of Oh: C4 (rotations through 7/2 about the fourfold symmetry
axis) and C' (rotation through 7 perpendicular to the principle rotation axis [14]). We
recall that if the z axis is not an m-fold axis symmetry, the potential will contain VJ [10].
Therefore, the Hamiltonian will a contain new term:
HCEF W x+ (1 6- 216 0- o (21 7)
F (4) F(6) Ft+ (6)

where the coefficients F(4), F(6), and Ft(6) are 40, 1260, and 30 respectively [15]. The
new term Oj-06 has the following symmetry in Cartesian coordinates:

2Y2(x2 2_ 2) + y2 (Y2 ) + 22(z2 2). (218)

When parameter y is zero, the system reduces to Oh symmetry. The eigenvalues are
tabulated for x=0.6 and several values of y by Shiina et al. [15]

Fr: (I + 4) + 4)) + 10) (2-19)
12 6

F23 : 4) + 4)) 0)

(| + 2) +-2)) (2-20)

1F) : 1 4|)-a2 -- 2) +a2l+ )+a +4)
biT F3) + b2| T1) + b3 1) + b4| 3) (2-21)

rF ) 2 4) ali- 2) + ai +2) a2 + 4)
b2| T 3) bil T 1) b4l 1) + b3 3) (2-22)

If y 0, the eigenstates are those for Oh symmetry [13]

Fr: (I + 4) + 4)) + v 0) (2-23)
12 6

3 : ( +4)+ -4))- 0)

1(I 2)+ 2)) (2-24)

F4 TV/I T3)T /i+F)

2t( +4) -- 4)) (2-25)

F5 : + / :| 3) T Ti )

(I + 2) -I- 2)) (2-26)

Depending on the sign of the W parameter (or E), the ground state can be thought

as either F23 or F1 (Fig. 2-2).

The eigenfunctions and eigenvalues of Fi(Th) and F23(Th) are the same as those

of Fi(Oh) and F3(Oh), therefore are not affected by O from the Hamiltonian. When

y=0, FP1)(Th) has the same eigenfunctions and eigenvalues as F4(Oh), and F(2) as those

for F5(Oh). When yO/, F4 and F5 mix resulting in two F(1'2)(Th) [15]. Therefore, the

eigenfunctions and eigenvalues of CEF for Th and Oh are different. The O0 term in

Hamiltonian affect some eigenfunctions and eigenvalues, resulting in a change of the

transition probabilities of neutron scattering in Pr3+

2.2 Conduction Electron Mass Enhancement (m*) Mechanism in PrOs4Sbl2

This Section gives a review on the mechanisms believed to be responsible for the

conduction electron mass enhancement: the single-ion models, such as quadrupolar Kondo

effect, or virtual CEF excitations (Fulde-Jensen model for m* enhancement in Pr metal)

and comment on the cooperative model invoking proximity to the long-range order.

2.2.1 Quadrupolar Kondo Effect

The quadrupolar Kondo (QK) model of HF was initially proposed by Cox to explain

weak field dependence of the specific heat of UBe13 [16, 17]. Barnes [18] found that Cu2+

ions in the cuprate superconductors could lead to such a Kondo effect as well. Later, new

evidence believed to be hallmarks of a quadrupolar Kondo effect has been found in the

alloys Yl-,UPd3 [19-22] for x=0.1 and 0.2.

In UBel3, the total angular momentum of U4+ (5f2 configuration) is J=4. This

leads [13] to a F3 CEF ground state for about half the ( i -I I field parameter range (Fig.

2-2). The f2 configuration is expected also for Pr3+ in PrOs4Sb12, and, according to CEF

calculations of Lea, Leask, and Wolf [13], a F3 doublet CEF ground state is very probable

(Fig. 2-2). Therefore, the heavy fermion behavior in PrOs4Sb12 could be in principle

due to a QK effect. Also, the physical properties of UBe13 (and U-1_Th1Bel3) are highly

Table 2-1. The relevant states for the quadrupolar Kondo effect for U4+. The last two
columns are the projections of the magnetic and quadrupolar moments
respectively (Reprinted with permission from Cox and Zavadowski [23]).

Config. State J Eigenstate (J) (3J J(J + 1))
f PI3+) J 4 7[14)- 4)] |0) 0 +8
f2 I3-) J4 2)+ 2)] 0 -8

fP \- J_ | +_ 0
f7+) j 5 1) 0
l IF7_) j 5 V/-I 5\ V/I 8\ 5 0
2 6 2/2 6

CI |I8 + 2) J 5 5) I 3 ) 11 +8
S|2 2 6
C1 I|F 2) J il 5 |3l 11 +8
1 Ir + 1) J 11) + -8
c IFs 1) J5 |- 1) -18
_2 2 2

reminiscent of those of PrOs4Sb12. Thus, since the discovery of the HF state in PrOs4Sb12
its normal properties have been associated with the QK effect.
The states involved in the quadrupolar Kondo effect for U4+ are given in Table 2-1.
The doubly degenerate ground state can be treated as a two-level system (a manifold with
a pseudo-spin of '). The projected value of the electric quadrupole moment onto the F3
basis is IQzzI=| 3J J(J + 1)| = 8 and the projected value of the magnetic dipole
moment is zero, i.e. J\ = 0 (Table 2-1). Therefore, the coupling is between the electric

quadrupole moment and the conduction electrons.
The Anderson model for the relevant states of the quadrupolar Kondo effect (since
it considers only F3, F7, and Fs) is called the 3-7-8 model. Figure 2-3 shows a schematic
representation of the Anderson model relevant for U4+ ions in the cubic symmetry. The
ground state F3 (J=4, 4f2) and first excited F7 (J=5/2, 5f') mix only via the conduction
partial waves Fs (J=5/2, c'). The transition fl f2 is done by removing a conduction
electron and the transition f2 f1 is done by emitting a conduction electron. It can

be shown that only Fs conduction quartet partial waves (Table 2-1) may couple to the

impurity through hybridization [23] (or, in the group theory framework, F3 0 F7 Fs).

Applying a canonical transformation (Schrieffer and Wolff [24]) to the 3-7-8 model,

the hybridization term can be eliminated. Also, the transformation yields to an effective

exchange interaction between pseudospin- and electric quadrupole moments of the form

Hexchange = -2Jexchange3 [a8(0) + a8(0)], (2-27)

where r3 is a pseudospin- matrix for the F3 quadrupole, a(s(as) are the pseudospins

formed from the Fs + 2, Fs + 1 (Fs 2, Fs 1) partial waves (Table 2-1). The exchange

integral Jexchange is proportional to F/TrcfN(0) and is negative.

The Hamiltonian has a two-channel Kondo form; two degenerate species of conduction

electrons couple with identical exchange integrals Jexcha"ne to the local a3 = object. The

channel indices are the magnetic indices of the local conduction partial wave states. Figure

2-4 shows schematically the mapping of the quadrupolar Kondo to the two-channel Kondo

model. The two-channel quadrupolar form of the Hamiltonian tell us that the conduction

electron orbital motion can screen the U4+ quadrupole moment equally well for magnetic

spin-up and magnetic spin-down electrons. Thermodynamic Properties of the Quadrupolar Kondo Model

Figure 2-5 shows the thermodynamic properties of the quadrupolar Kondo model

[25-27]. The susceptibility (Fig. 2-5, lower panel) diverges logarithmical at T=0,

x=-(eTTH)-lln(H/TH), where TH (T/e)TK [28, 29]. Here, e is the base of In, i.e.

2.71... In the quadrupolar Kondo model this corresponds to a divergent quadrupolar

susceptibility. For T-0, the free energy in zero field is F -Tln2. Therefore, the

zero-temperature zero-field entropy is equal to ln2 [25]. The non-zero entropy at T --0

is consistent with the divergence in susceptibility and argues in favor of a non-singlet

ground state (a singlet is the ground state for the standard Kondo model). As expected,

the entropy increases monotonically with temperature and reaches .,-ii!.1 .i, ically the ln2

value (free spin) at high T. Also, the T=0 entropy increases with the field. Since the

S(T = 0) decreases with H the specific heat increases with H at intermediate T resulting

in large values of 7, common for heavy fermion systems. At high T the pseudo-spin is free,

therefore S= ln2. The entropy change AS(H)S(T = oo, H)-S(T = 0, H) increases with

H from 1 ln2 to ln2 for large H. In the C/T plots the Kondo peaks can be seen.

The initial measurements of specific heat [16] were not conclusive for a quadrupolar

Kondo effect in UBe13. Also, more recent measurements of nonlinear susceptibility [30] are

inconsistent with the quadrupolar (5f2) ground state of the uranium ion, indicating that

the low-lying magnetic excitations of UBe13 are predominantly dipolar in character. Relevance for the Case of Pr + Ion in PrOs4Sbl2

PrOs4Sbl2 has been initially reported to have a P3 doublet CEF ground state [7].

Later experiments [31 33] established the < i iiii. electric field (CEF) ground state of

the Pr3+ ion in the cubic symmetry environment of PrOs4Sbl2 (Th point group symmetry)

as the non-magnetic singlet Fi. The consequence of this is that the original formulation

of the quadrupolar Kondo effect cannot be applied to the conduction electron mass

enhancement in PrOs4Sbl2.

F1 is nearly degenerate with the F(2) triplet. Though Fi itself doesn't carry any

degrees of freedom, the pseudo-quadruplet constituted by Fi and F2) is speculated to

have magnetic and quadrupolar degrees of freedom [34], and therefore a magnetic or

quadrupolar Kondo effect is invoked to explain the enhancement of the effective mass of

the quasi-particles.

On the other hand, the model does not seem to be relevant since the predicted

properties of the quadrupolar Kondo effect are in disagreement with the measurements.

But this is a single-ion model. Possibly, intersite effects are responsible for the disagreements.

There is no lattice quadrupolar model.

2.2.2 Fulde-Jensen Model for m* Enhancement in Pr Metal

The mass of the conduction electrons can be enhanced by the interactions with

various low-lying excitations in the solid. This explains the strong dependence of the

specific heat of the Pr metal in magnetic fields found by Forgan [35].

Goremychkin et al. [32] proposed that the mass enhancement in PrOs4Sb12 can

be explained by a balance between two types of interactions, magnetic dipolar and

quadrupolar between conduction and the f electrons of Pr. The theory of Fulde and Jensen

[36] of conduction electron mass enhancement ascribes this to the inelastic scattering by

crystal field transitions in a singlet ground-state system. The mass enhancement of the

conduction electrons are due to their interaction with the magnetic excitations.

The relevant Hamiltonian describing the interaction between the conduction electrons

and the rare-earth localized moments is [36]

Hi = Isf{9L t1) J(Ri) J, (2-28)

where If is the exchange integral, gL is the Land6 factor, J is the total angular

momentum of a rare-earth ion at site R., and a are Pauli matrices.

The mass enhancement due to the inelastic transition at energy A between two levels,

|i) and Ij), is
mn* 21 (i J 2 (229) 2
S1+ ( )2 N(0) (2-29)

where gj is the Land6 factor, Isf is the exchange integral coupling the conducting electrons

to the f-electrons, N(O) is the conduction electron density of states at the Fermi level,

and (i IJ Ij) is the magnetic dipole matrix element calculated using the derived crystal

field parameters. This formula shows that for a small excitation energy A leads to a large

enhancement in m*.

2.2.3 Fluctuations of the Quadrupolar Order Parameter

In general, models involving spin fluctuations in heavy fermions around their

antiferromagnetic instability were considered by authors such as Hertz [37], Millis [38],

Moriya and Takimoto [39] (a complete review is given by Stewart [40, 41]). All these

models exhibit divergence of the low temperature specific heat.

By analogy, in PrOs4Sbl2, the quadrupole fluctuations of Pr ions are believed to

pl iv an important role in the HF-SC properties. Therefore, another model proposed (a

collective-type model) for the mass enhancement mechanism are due to the fluctuations of

the antiferroquadrupolar order parameter due to the proximity to the AFQ ordered phase.

PrOs4Sbl2 exhibits an antiferroquadrupolar ordered phase in fields between about

4.5 and 14 T. For fields 5-13 T the two lowest CEF levels are sufficiently close to form

a pseudo-doublet with quadrupolar and magnetic degrees of freedom, resulting in a long

range order.

There is no theory (to this moment) that describes the mass enhancement due

to the fluctuations of the quadrupolar order parameter. Our magnetoresistivity data

of PrOs4Sbl2 and La alloys presented in C'!i lters 6 and 7 seem to support this mass

enhancement mechanism.

21/3, I H

------ -- ------
4- I 4

-- --I--------- ----- -- -----
I- ..
a) b)

Figure 2-1. Rotational symmetry Th. In the left (a), the small bold blue segment is
assimilate with the distance between two ,i 1inl lv7 atoms belonging to the
same icosahedra. A 7 rotation with respect to (100) and a 2 rotation with
respect to (111) are allowed. The rotation with respect to (1 00) will not
turn the structure into an equivalent one. Therefore, the axes x (or (1 00))
and y (or (0 1 0)) are not equivalent (Reprinted with permission from D. Vu
Hung [42]).

Figure 2-2. Lea, Leask, and Wolff's representation of CEF for J=4 (Redrawn with
permission from Lea et al. [13]).

ff F 7


Figure 2-3.

Representation of the U4+ ions in cubic symmetry undergoing quadrupolar
Kondo effect. The model involves a doublet ground state in each of the
two electronic lowest-lying configurations: f2 having the quadrupolar or
non-Kramers F3 doublet, and fl configuration having the magnetic or Kramers
F7 doublet. The conduction electrons mix the two configurations through a
hybridization process. The Fs conduction state couples these two doublets
(Redrawn with permission from Cox and Zavadowski [231).

SO SI so-


Figure 2-4.



Mapping of the quadrupolar Kondo Hamiltonian onto the two-channel
Kondo model. a) The standard two-channel Kondo model in spin space: two
conduction electrons s,+ and s,_ couple antiparallel to the impurity spin SI.
b) In the quadrupolar Kondo case, the spin is due by the quadrupolar or
orbital deformations. The two channels come from the real magnetic spin of
the conduction electrons. The orbital motion of the electrons produces the
screening of the U4+ orbital fluctuations (Redrawn with permission from Cox
and Zavadowski [231).

nD t2 -

0 30

0, 10 0.1

0.05 7 0.3


1 -I 0 1

10 -'
10- 10- 10- 10-' 13' "C' 10 101

Figure 2-5. S, C, C/T, and X versus T/TK of the quadrupolar Kondo model
with permission from Sacramento and Schlottmann [27]).



3.1 Crystalline Structure

PrOs4Sb12 praseodymiumm osmium antimonide) is a filled skutterudite that
into a LaFe4P12-type body-centered cubic structure with the lattice parameter a 9.3068

A[43](a 9.30311 A[44] after more recent measurements), space group Im3, and Th point

group symmetry. The < i i-- 1 .)graphic arrangement of the atoms is given in Fig. 3-1. The

mass of a mole of PrOs4Sb12 is 2363.1 g, the molar volume is 242.3-10-6 m3/mol Pr and

the mass density is 9.75 g/cm3 [45]. LaOs4Sb12 is the non-fequivalent of PrOs4Sb12 with

a similar crystal structure. All the exotic phenomena of PrOs4Sb12 are thought to be

associated with its unique crystal structure. In particular, the large coordination number

of Pr ions surrounded by 12 Sb and 8 Os ions leads to strong hybridization between the

4f and conduction electrons [46]. This strong hybridization results in a rich variety of

strongly correlated electron ground states and phenomena.

3.1.1 Rattling of Praseodymium Atom

In CoAs3-type skutterudites whose name come from the cobalt arsenide mineral

that was first found in Skutterud, Norway, an alkali metal, alkaline earth, rare earth, or

actinide ion, occupies an atomic cage. The icosahedron shaped atomic cage made of Sb

atoms accommodates a rare earth ion, and the size of this cage is bigger than the radius

of the ion. Therefore, the rare earth ion will vibrate as a result of the weakly bounded

rare-earth ion in the oversized cage made of Sb ions. This anharmonic oscillation is called

rattling [47]. The consequence is a reduction of the thermal conductivity. The filled

skutterudites with the cage are favorable for a thermoelectric device possessing a high

coefficient of merit [48].

The amplitude of this vibration of the Pr ion in PrOs4Sb12 is about 8 times 'i.-.-r

than the amplitude of Os. EXAFS data [49] supports the idea of a rattling Pr filler ion

(based on the low Einstein temperature OE ~75 K) within a fairly stiff cage in this

material. Besides the dynamic movement, a static displacement was detected in which

there are two equilibrium positions for the Pr ions. The Pr ions can freeze in one of these

two equilibrium positions, and at low temperature they can pass from one position to

another through tunnel effect. It has been estimated that this displacement is about 0.07

A [49]. Goto et al. [50], based on a theory of Cox et al. [23], -,:-:-, -1. that the tunnel

effect between the two positions of the Pr ions could be linked to the appearance of the


3.1.2 Valence

At high temperature (above 150 K) the X(T) of PrOs4Sb12 can be described by a

Curie-Weiss law with an effective moment peff=2.97pB as reported by Bauer et al. [7], or

peff 3.5/B as reported by Ti- ii, et al. [51], and a Curie-Weiss temperature Ocw=-16
K [7]. The effective moment found is somewhat lower than the moment of a free ion Pr3+

which has peff=3.58/PB [52].

X-r-rv--.-..-rption fine-structure (XAFS) measurements [49] carried out at the Pr LIII

and Os LIII edges on PrOs4Sb12 sirl:., that the Pr valence is very close to 3+. Each Pr

ion has two electrons on the f shell (4f2 electronic structure).

3.1.3 Crystalline Electric Fields

In an ionic (localized) model, the cubic crystalline electric field of PrOs4Sb12

environment splits the J=4 Hund's rule multiple of non-Kramers Pr3+ into a singlet

(F1), a doublet (F3), and two triplets (F4 and s5) (in the Oh symmetry notation). The
CEF Hamiltonian in cubic symmetry was written [13] in terms of the ratio of the fourth

and sixth order terms of angular momentum operator of the CEF potential, x, and an

overall energy scale factor W. For more than two decades, the symmetry was thought as

Oh, instead of Th.
Bauer et al. [7] fitted the magnetic susceptibility data (see Fig. 3-2) by a CEF model

in which the ground state was chosen to be either the non-magnetic PF singlet (W>0)

or the non-magnetic F3 doublet (W<0). The peak present in the x(T) data was thought

to be produced when the first excited state is a triplet F5 with a energy <100 K above

the ground state and corresponds to a position x close to the crossing points on the LLW

diagram (see Fig. 2-2) where F1 or F3 are degenerate with Fs. The results are presented in

the upper panel of Fig. 3-2. The notation of the CEF energy levels corresponds with the

initially assumed Oh symmetry by Maple et al. [53] and Bauer et al. [54]

These authors use the conventional cubic < i --I I1 field model which is applicable to the

O, Td and Oh symmetries. In the Th symmetry, the non-Kramers doublet 3F corresponds

to the degenerate F2 and F3 singlet states (denoted as F23) and F4 and F5 states coincide

with F'1) and F 2) triplet states, respectively, when last term is zero in the < i -I I1 field

Hamiltonian (2-17). The singlet state F1 is the same for both cases.

The last term of equation (2-17) is unique to the Th symmetry of this material

coming from the atomic configuration of Sb ions in the crystal [15] and is absent in

the conventional cubic crystal field Hamiltonian that Maple et al. [53] and Bauer et

al. [54] used. The omitting of the last term in the Hamiltonian (see equation 2-17) has

implications in the interpretation of the inelastic neutron scattering data.

The above mentioned fit reproduces the overall shape of the low temperature peak,

and also the value of the van Vleck paramagnetic susceptibility with an effective moment

close to, but somewhat lower than that, of the free Pr3+ ion. Bauer et al. [7] fitted C

assuming a degenerate spectrum

Specific heat data was fitted [7] by a system with two levels of equal degeneracy split

by an energy 6=6.6 K (it has been assumed that the degeneracy of any level is lifted by

CEF when the local site symmetry of the Pr3+ ions is not cubic as a result of some kind of

local distortion).

The entropy in the 13-F5 case was found to be Sr3-_5=Rln2m7.6 J/(mol K) [7].

The total entropy of the broad peak just above the transition is S=f(C(T)/T)dT ,10.3

J/(mol K). The closeness in values made Bauer et al. [7] to favor the 3 ground state


Initial inelastic neutron scattering measurements [53] considering Oh symmetry

-,i--.- -1 that F3 is the CEF ground state in PrOs4Sbl2. The resistivity data measurements

were also interpreted in the framework of a F3 CEF ground state [55].

In contrast, Tayama et al. [51] obtained a somewhat better fit of the magnetic

susceptibility x(T) data by a F1 CEF ground state model (Fig. 3-3 (a)). Also, the

theoretical curves of S(T) based on F1 ground state model show increase of entropy with

fields (lower panel of Fig. 3-3 (b)). This trend is confirmed by magnetic field specific heat

measurements (Fig. 3-3 (c)) by Aoki et al. [56].

Therefore, zero or small magnetic fields data are contradictory, more experiments are

to be done in order to establish the true CEF ground state in PrOs4Sbl2.

3.2 Normal-State Zero-Field Properties

3.2.1 Specific Heat

PrOs4Sb12 was synthesized for the first time by Jeitschko et al. [43], and then by

Braun et al. [57]. It was in 2002 when Bauer et al. [7] discovered superconductivity in

PrOs4Sbl2. Since the discontinuity in specific heat is of the order of 7, this large value

discontinuity (A(C/T) IT=1.85K -500 mJ/K2 mol [7]) implies the presence of heavy

fermions both in the normal and superconducting states.

There is no consensus regarding the precise value of 7, but all the reported values

imply heavy fermion behavior. Actually, this is perhaps the strongest evidence for HF

states in PrOs4Sb12. Considering the relation A(C/7TT)=1.43 the Sommerfeld coefficient

7 is found to be -350 mJ/K2 mol. The phonon (lattice) contribution to the specific heat

C data can be described by 3T3 that is identified with specific heat of LaOs4Sb12 with

OD 304 K. 3 is related to OD by /3(1944x103)n/O(, where n is the number of atoms in
the formula unit (e.g., n=17 in LaOs4Sb12).

3.2.2 de Haas van Alphen Measurements

The Fermi surface (FS) as reconstructed by de Haas van Alphen (dHvA) measurements

[44] comparative with the bands structure (LDA+U method [58]) are presented in

Fig. 3-5. The topology of the FS of PrOs4Sb12 is very similar to that of the reference

compound LaOs4Sb12 [44] (which leaks 4felectrons). This indicates that the 4f2 electrons

in PrOs4Sb12 are well localized. The similar topology of the FS for the two compounds

is supported also by similar angular dependence of the dHvA. Three Fermi surface

sheets, including two closed (practically spherical shaped) and one multi-connected, were

identified in agreement with the calculations.

The effective masses measured by dHvA are between 2.4 and 7.6mo (mo is the

free electron mass). These values are well below the ones reported from the specific

heat measurements. These low values have been explained [59] in the framework of the

two-band superconductivity model in which band 2 corresponds to the light band detected

by dHvA measurements. Band 1 is a heavy band having most of the density of states.

The heaviest quasiparticles are seen in thermodynamic measurements (C or H,2) only.

However, the applicability of the two-band model to PrOs4Sb12 is not established. Further

more, our results presented in section 7.6 sheds some doubts in the interpretation.

3.2.3 Resistivity

Additional evidence for heavy fermion behavior in PrOs4Sb12 is provided by an

analysis of the slope of the upper critical field Hc2 near Tc. The upper panel of Fig. 3-2(a)

shows resistivity versus T. From the resistivity data in small magnetic fields (data not

shown, fields up to about 30 kOe) and from the fit of the linear part of -(dHc2/dH)T,

curve, the initial slope has been measured using the BCS relations and shown to be ~19

kOe/K [7]. This implies o ~116 A, vF=1.65x106 cm/s, and m* ~50mo. This calculation

assumes a spherical Fermi surface.

The resistivity data between 8 and 40 K revealed a T2 dependence po+AT2, with

A=0.009 pQcm/K2 [7]. The A coefficient is about two orders of magnitude smaller than

the value expected for a heavy fermion compound. Considering the Kadowaki-Woods

universal relation [60] between A and 7, A/72 = 1 x 10-5/Qcm mol2K2mJ-2. The 7 value

is only ~6.5 mJ/mol K2 [7], and this is a typical value for normal metals and is much

smaller than 7 of LaOs4Sbl2.

3.2.4 DC Magnetic Susceptibility

The X(T) data (Fig. 3-2(a), lower panel) exhibits a peak at ~3 K and saturates to

a value of about 0.1 emu/mol [7] as T-+0. This is the hallmark of a nonmagnetic ground

state. Above 150 K, X(T) of PrOs4Sbl2 can be described by a Curie-Weiss law. There is

a large discrepancy between the high temperature effective moment reported by various

research groups. The effective moment according to Bauer et al. [7] is pIff 2.97PB, and

qeff 3.5/B is the value reported by Tli-v i, et al. [51] The free ion Pr3+ has a high
temperature effective moment of 3.58/B [52]. The Curie-Weiss temperature is Ocw=-15

K [51].

From the diamagnetic onset (inset (ii), Fig. 3-4(a)) it is found that the temperature

of the superconducting transition Tc is equal to the value found from the specific heat


3.3 The Long-Range Order in Magnetic Fields

Measurements of specific heat [56] in fields up to 8 T and resistivity [55] in magnetic

fields up to about 10 T revealed the existence of a field induced ordered phase (FIOP)

above 4.5 T. In this ('! lpter a discussion of the nature of the FIOP will be presented

along with the specific heat data that completes the magnetic phase diagram. A similar

phase diagram has been obtained later by magnetization [51, 61](see Fig. 3-6) and by

thermal expansion and magnetostriction measurements [62].

3.4 Superconductivity

Experiments on PrOs4Sbl2 imply the possibility of unconventional superconductivity

(i.e., the existence of nodes in the gap of the Fermi surface). There is still some other

evidence that -,-.-.-I -I- an isotropic SC gap. We present below experimental evidence

favoring either unconventional or conventional superconductivity.

3.4.1 Unconventional Superconductivity The Double Transition

Initial specific heat measurements [7] showed a single superconducting transition at

T, of 1.85 K. Higher quality materials revealed actually two superconducting transitions

(Vollmer et al. [63], Maple et al. [53], Oeschler et al. [64]). In Figure 3-7 panels (a) and

(b) are shown specific heat of PrOs4Sbl2 presenting two superconducting transitions,

Tc2=1.75 K and Tci=1.85 K by Vollmer et al. [63], and T,2=1.716 K and Tci=1.887

K by M6asson et al. [59], respectively. Two superconducting transitions at the same

temperatures have been reported by Cichorek et al. [65] along with a speculation for

a third superconducting transition at ~0.6 K inferred from Hc measurements. It is

believed that inclusions of the free Os in the single crystal cannot be responsible for the

enhancement of HaI, though T, of pure Os is 0.66 K [66] based on sensitive X-ray and

electron microprobe studies [65].

There are two classes of explanations of the nature (intrinsic or not) of the double

transition. One argues in favor of two different parts of the sample with two different

superconducting phases, and therefore with different T,'s. Thus, the quality of the samples

is crucial. For instance it has been considered [59] that despite the sharp specific heat

transitions, the samples still present spatial inhomogeneities. One possibility would be an

inhomogeneous coexistence of two electronic configurations of Pr, 4f1 and 4f2. The high

temperature magnetic susceptibility measurements are in favor of 4f2, since they have

found [51] an effective moment Peff 3.6pB/Pr (the expected value for 4f1 is 2.54pB and

for 4f2 is 3.58/B).

Another possible scenario that is presented in this dissertation is the existence of

inhomogeneities due to the closeness of the system to a long range antiferro-quadrupolar

order: clusters with a short-range order would have different superconducting parameters

than the remaining part of the sample.

Figure 3-7 (c) shows ac susceptibility after M6asson et al. [59]. The nature of the two

transitions is not yet established. The width of the transition as measured by specific heat

and ac-susceptibility is the same, about 0.2 K.

The superconducting gap structure investigated using thermal transport measurements

in magnetic field rotated relative to the crystal axes by Izawa et al. [67] provides

another evidence for the unconventional character of superconductivity in PrOs4Sbl2.

The change in the symmetry of the superconducting gap function that occurs deep

inside the superconducting state gives a clear indication of the presence of two distinct

superconducting phases with twofold and fourfold symmetries (Fig. 3-8). The gap

functions in both phases have a point node singularity which is in contrast to the line

node singularity observed in almost all unconventional superconductors. The two-band

superconductivity (similar to that observed in MgB2) is observed in newer thermal

conductivity measurements [68].

A double transition can be seen in the thermal expansion [64] experiment (Fig. 3-9).

The two transitions are at the same temperatures at which the specific heat discontinuities

occur. Using the Ehrenfest equation OTc/9P = VTcA//AC, where Vm is the molar

volume, calculations show that the superconducting transitions T2a is decreased two

times faster under pressure than T1i. This is in favor of intrinsic nature of the two

superconducting transitions. Temperature Dependence of Specific Heat Below T,

In general, the specific heat C below T, exhibits different temperature dependence

according to the topology of the superconducting gap As(k). For an open gap the specific

heat dependence is the well known exponential e-Ao/T. Nodes in the gap, or zero points

in the gap, will be reflected in the T dependence of the specific heat as a power T3

dependence. And for zero line in the gap, the temperature dependence of C is T2

Table 3-1. The a values reported by different groups, extracted from fits of specific heat
below T, (Reprinted with permission from Grube et al. [70]).

Specific heat data a = A(0)/(kBT,)
Grube et al. [70] 3.70.2
Vollmer et al. [63] 2.60.2(F3)
Frederick et al. [71] 3.10.2(F1)
Frederick et al. [71] 3.60.2(F3)

Table 3-2. The a values reported by different groups from measurements other than
specific heat (Reprinted with permission from Grube et al. [70]).

Experiment a = A(0)/(kBTe) Gap Function
Tunneling spectroscopy [72] 1.7 Nearly isotropic
pSR [73] 2.1 Nearly isotropic
A(T) [74] 2.6 Point nodes
Sb NQR [75] 2.7 Isotropic

In all reported data the specific-heat measurements exhibit a rapid decrease of

C below the superconducting transition. This points to pronounced strong-coupling


The so called a-model [69] assumes that the superconductive properties which are

mainly influenced by the size of the gap and the quasiparticle-state occupancy could be

approximated by simply using the temperature dependence of the weak-coupling BCS gap.

The size of the gap in the Fermi surface is a freely adjustable parameter a=A(0)/kBTc,

where A(0) is Fermi-surface averaged gap at T=0. Table 3-1 presents comparative a

values obtained by different groups. An analysis using the a-model results in an extremely

large gap ratio of

a=A(0)/kBT= 3.7 and a huge specific heat jump of C/(c)>5 [70].

A summary of the published superconductive gap ratios and gap anisotropy of

PrOs4Sb12 from other measurements than specific heat are presented in Table 3-2.

Frederick et al. [71] succeeded in making a better fit for the specific heat data of

PrOs4Sb12 using a power-law function below the superconducting temperature. The fits,

using both power-law and exponential functions, cannot be considered by themselves as

proof of the superiority of one fit over the other. Nuclear Magnetic Resonance (Sb NQR)

The 121,123Sb Nuclear Quadrupole Resonance (Sb NQR) experiment [75] in zero field

shows a heavy fermion behavior and controversial conclusions regarding the nature of the

superconductivity in PrOs4Sbl2. In the SC state, 1/Ti shows neither a coherence peak just

below T, nor a T3-like power-law behavior observed for anisotropic HF superconductors

with the line-node gap. The absence of the coherence peak in 1/Ti supports the idea of

unconventional superconductivity in PrOs4Sbl2 (Fig. 3-10). The isotropic energy gap with

its size A/k=B4.8 K seems to open up across Tc below T*=2.3 K. The very large and

isotropic energy gap 2A/kBTc ~5.2 indicates a new type of unconventional strong-coupling

regime. Muon Spin Rotation (pSR)

The broken time reversal symmetry has been reported in later muon-spin relaxation

measurements. The results [76] reveal a spontaneous appearance of static internal

magnetic fields below the superconducting transition temperature, providing unambiguous

evidence for the breaking of time-reversal symmetry in the superconducting state.

This will favor the multiple superconducting phase transitions observed by specific

heat and thermal conductivity studies and support therefore the unconventionality of


Magnetic penetration depth data in single crystals of PrOs4Sb12 down to 0.1 K, with

the ac field applied along the a, b, and c directions was successfully fitted [74] by the 3He

A-phase-like gap with multidomains, each having two point nodes along a cube axis, and

parameter A(0)/kBT,= 2.6, -Ii.:.: -Iii-; that PrOs4Sb12 is a strong-coupling superconductor

with two point nodes on the Fermi surface. These measurements confirmed the two

superconducting transitions at 1.75 and 1.85 K seen in other measurements.

3.4.2 Conventional Superconductivity Nuclear Magnetic Resonance (pSR)

The temperature (T) dependence of nuclear-spin-lattice-relaxation rate, 1/TI, and

NQR frequency unravel a low-lying CEF splitting below To~10 K. In addition, the

temperature dependence of 1/Ti in PrOs4Sbl2 is an exponential one [75] (Fig. 3-10, full

symbol), which is the signature of a conventional type of superconductivity.

Figure 3-10 (open symbols) also plots the data for the conventional superconductor

LaOs4Sbl2. For an s-wave case that is actually seen in the T dependence of 1/Ti for

LaOs4Sbl2 with T,-0.75 K, in the SC state, 1/Ti shows the large coherence peak just

below Te, followed by an exponential dependence with the gap size of 2A/kBTc, 3.2 at low

T. This is a clear evidence that LaOs4Sbl2 is the conventional weak-coupling BCS s-wave

superconductor. Penetration Depth Measurements (A) by pSR

The transverse-field muon-spin rotation measurements in the vortex lattice of the

heavy fermion superconductor (HFSC) PrOs4Sb12 yields [73] an exponential temperature

dependence of the magnetic penetration depth A, indicative of an isotropic or nearly

isotropic energy gap, indicating a conventional superconductivity mechanism.

This is not seen, to date, in any other HF superconductor and is a signature of

isotropic pairing symmetry (either s- or p-wave, indistinguishable by thermodynamic

or electrodynamic measurements), possibly related to a novel nonmagnetic quadrupolar

Kondo HF mechanism in PrOs4Sb12. Also, the estimated magnetic penetration depth

A 3440(20) A [73] was considerably shorter than in other HF superconductors. Low-Temperature Tunneling Microscopy

The spectra of a direct measurement of the superconducting gap through high-resolution

local tunneling spectroscopy [72] in the heavy-fermion superconductor PrOs4Sb12

demonstrates that the superconducting gap is well developed over a large part of the

Fermi surface. The conductance has been successfully fitted by a s-wave superconductivity

model. The presence of a finite distribution of values of the superconducting gap over the

Fermi surface argue in favor of isotropic BCS s-wave behavior.

Q Sb

001 Q Os


Figure 3-1. Crystal structure of PrOs4Sb12 (Reprinted with permission from Aoki et al.


0.06 PrOs4Sb12 0.:04: I

0 10 20 30
0.04 x =0.50. W = 1.9 K X =-0.72, W =-5.4 K
S 3 (111 K) r, (313 K)
Sr,,65 K) r (130 K)
0.02 r (6 K) r. ( 1 K)
r, (O K) T,(OK)

0 50 100 150 200 250 300
(b) t 1-=g0 12
2.5 S 6.6 K

2,0 I 4

S1.5 0 5 10 15 20

S10 o A_ AC(T)= C +(T)+y'T

0.5F PrOs Sb1

0 5 10 15 20
T (K)

Figure 3-2. (a) Fits of the magnetic susceptibility x(T) of PrOs4Sb12 to CEF model with
either F3 (solid line) or F, (dashed line) ground state. The same symbols are
used in the inset, which shows x(T) bellow 30 K. In the inset, the solid line
fit saturates just above X 0.06 cm3/mol. (b) C fitted by a two-level Schottky
anomaly (Reprinted with permission from Bauer et al. [7]).






Figure 3-3.

(a) Fits of the magnetic susceptibility x(T) of PrOs4Sb12 to CEF model
with either F3 (dashed line) or Fi (doted line) ground state. The solid lane
represents the experimental data (taken from Tayama et al. [51]). (b) The
calculated entropy S(T) for H//(10 0) in both F3 and Fl CEF ground state
models (taken from T.i ,ii i et al. [51]). (c) The measured entropy S(T) for
H//(1 00) (Reprinted with permission from Aoki et al. [56]).

T (K)
T (I)

f 150 (a)

0.06 PrOs Sb,, o 5 1 15

0.04 -
H=200e (I) -
0.02- -1.0 ..
o 1,8 2.2 286 3.0

0 50 100 150 200 250 300
3.0- (b) 2.25. (iii)
1`2 2.25 1
2.5- 7,
I::;- 1,75 -
2.0 1,50 T., 1.76 K
ACITr 500 mJmol K
1.- 1.0 1.5 2.0 2,5 -



O 5 10 15 20

Figure 3-4. (a) Resistivity p(T) and susceptibility x(T) of PrOs4Sbl2 (b) Specific heat
C(T) up to 20 K [7] (Reprinted with permission from Bauer et al. [7]).

(a) band 48-hole

(b) band 49-hole


Figure 3-5. Fermi surface of PrOs4Sb12 (Reprinted with permission from Sugawara et al.

, I,

0 0.5 1

z=0 plane

P C-
1.0.5, 0.5, 0.5)


1.5 2

Figure 3-6.

H-T phase diagram of PrOs4Sb12 (Reprinted with permission from T i,. iI
et al. [51]). Open and closed symbols were determined by the dM(T)/dT
and dM(H)/dH data, respectively. Right panel, the Pr charge distributions
induced in the antiferroquadrupolar ordered phase in magnetic field (Reprinted
with permission from MWasson [45]).



Figure 3-7.

T (K)

-0.02 /
.2 -0-06
1.5 1,8 1.7 1. TK)

(a) C(T) of PrOs4Sb12 presenting double superconducting transition
(Reprinted with permission from Vollmer et al. [63]) (b) C(T) of PrOs4SbI2
presenting two superconducting transitions (Reprinted with permission from
M6asson et al. [59]) (c) The real part of the ac susceptibility of PrOs4Sb12
presenting two distinct superconducting transitions (Reprinted with permission
from MWasson et al. [59]).


I >I

1 1I




A- 1.5

^1.0 -



Figure 3-8.

The two superconducting phases for PrOs4Sb12 (Reprinted with permission
from Izawa et al. [67]). The gap function has a fourfold symmetry in A phase
and twofold symmetry in B phase. Right: The plot of the gap function with
nodes for A phase and B-phase (Reprinted with permission from Maki et al.





1.6 1.8 2.0
T (K)

Figure 3-9. Two superconducting transitions in the thermal expansion coefficient 3 of
PrOs4Sb12. The two transitions are visible for the same temperatures of the
two transitions in specific heat (Reprinted with permission from Oeschler et al.

iionnnl state

0.5 1.0

1.5 2.0




1 -;


gi LaOs4Sb12



10 I **
0.1 1 10 100
Temperature (K)

Figure 3-10. Temperature dependence of the rate 1/Ti at the 2VQ transition of 123Sb for
PrOs4Sbl2 (closed circles) and LaOs4Sbl2 (open circles) (Reprinted with
permission from Kotegawa et al. [75]).

-1000 -500 0 500
Bias voltage (IV)


Figure 3-11. Tunneling conductance between PrOs4Sb12 and an Au tip. The gap is well
developed with no low-energy excitations. The line in figure is the prediction
from conventional isotropic BCS s-wave theory using A=270 peV and
T=0.19 K (Reprinted with permission from Suderow et al. [72]).


This chapter describes the sample synthesis, characterization and the experimental

procedures used: dc and ac susceptibilities, resistivity, and specific heat measurements. A

brief description of the performed measurements is given.

4.1 The Samples: Synthesis and Characterization

4.1.1 Synthesis

The filled skutterudite antimonides studied in this dissertation are prepared using

a molten-metal-flux growth method with an excess of Sb flux [54, 57]. Since the flux is

one of the constituent elements of the compounds (i.e. Sb) the method is called self-flux

growth. High-purity starting elements (Pr and La from AMES Laboratory, 99.9', .

purity powder Os from Colonial, Inc., and 99.9'' '. purity Sb ingot from Alfa AESAR)

are used in the proportion R:Os:Sb=1:4:20, when the rare-earth element R is Pr and

La in various proportions. The R alloys used as components in the flux growth were

synthesized eventually by melting its constituent elements in an Edmund-Biihler Arc

Melter under a high purity argon atmosphere. First, small chunks of Sb were placed

inside of a quartz tube. Above that were placed the Os and the R components that were

pre-melted separately to eliminate any trace of oxide from the surface of the elements.

The Os powder was pressed in small pellets and then melted. The quartz tube was sealed

under low pressure Ar atmosphere (~20 mTorr) after the tube is pumped and flushed

3 to 5 times. The tube with the mixture was placed in a Lindberg 51333 programable

furnace (digital controlled, Tma=12000C) using the following heat treatment sequence:

temperature ramping to 9800C with a rate of 2000C/h followed by T=9800C for 24 h, then

cooling at a rate of 3C/h down to 650C. The last step was a fast cooling in the furnace

to room temperature at a ~ 2000C/h rate. The single crystals were then removed from the

.irilrilir flux excess by etching in aqua regia (HCl:HNO3 1:1). The crystals were cubic

or rectangular up to 50 mg in weight (up to ~3 mm in size) depending on the amount of

the starting elements and the cooling rate. For instance, using ~ 1 g of Os and a cooling

rate of lC/h the single-crystal mass was about 50 mg (Fig. 4-1). This large < i--I I1 later

proved to be very useful for the dc-susceptibility measurements. In the case of R being

actually an alloy, such as Prl_-La1, Pr and La are previously melted together using the

arc-melter as further described.

The poly-< i --i 11,ii: R alloys (used as one of the starting components in the

synthesis of the single crystals) were prepared by melting its constituent elements in an

Edmund-Biihler Arc Melter AM under a half atmosphere high purity Ar. The apparatus

consists of a stainless steel vacuum chamber which sits tight on a water cooled groove

crucibles in a copper base plate and with an electrode at the top. The tungsten electrode

is motor driven which can be moved freely above the crucible. The melting process can

be observed through a dark glass window. All important control functions are integrated

in the head of the electrode and ensure safe and convenient operation. When fed at

the maximum current the temperature of the electric arc in the melter can go as high

as 4000C and melts ~500 g of metals. The arc melter has a flipper, a manipulator for

turning the samples in situ. This gives the possibility to flip and again melt the sample,

ensuring its homogeneity, without opening the chamber. Before operating, the copper base

plate was thoroughly cleaned with acetone to avoid any contamination of the sample with

impurities. Right at the beginning, each of the constituent elements were well cleaned

to eliminate the oxide 1. -r on the surface. The precision in mass measurements was

0.02 mg. Starting with the radioactive or the hardest element we can adjust the relative

masses of the other components to gain the wanted stoichiometric ratio. The total mass

was from teens of milligram to ~1 g, the size of the sample bead was up to 1 cm. Right

before the elements were melted together, a zirconium button which was also used for

ignition of the arc, was melted just to ensure a even higher purity of the Ar, which was

filtered through a purifier before entering into the arc chamber. Zirconium is well known

as a oxygen absorber. The element with the highest vapor pressure was then placed on

the copper plate right below the elements with lower vapor pressures. The aim of this

was to not strike with the arc on the element with a high vapor pressure resulting in

uncontrolled vaporization of the material. The element with the lowest vapor pressure

was melted first. This reduced the mass loss, and the discrepancy between predicted and

actual stoichiometries of the synthesized alloys. To ensure an even better control of the

temperature at which the elements were melted, the copper plate in the immediate vicinity

of the place where all elements were together was first heated with a slowly increasing

current. This was done until the element with the highest vapor pressure started to

melt and to suck all the other elements. Then flipping the resulting bead and remelting

it ensured its homogeneity. This process was repeated several times. Also, for dilute

concentrations, such as Pro.98Lao.020s4Sb12, Pro.98La0.02 was needed first. Therefore we

started with master alloys (Pro.gLao.1) in order to avoid handling of very small amounts of


4.1.2 X-Rays Diffraction Characterization

X-ray diffractions of the materials verified whether the arc melting plus annealing

or the flux growth processes led to the formation of the desired <( i--I I1 structure. From

the diffraction pattern it is possible to determine the lattice constants and the presence

of the secondary phases in the material (if present in a proportion greater than 5'.).

The measurements were performed using a Phillips XRD 3720 machine at the AT ij r

Analytical Instrumentation Center ( \ AIC) at the University of Florida. Single < iv-i ,1-

and poly-< i -i~1 ,11 !i. samples were crushed and ground out into a fine powder using

a ceramic mortar. On a glass slide, about 1 cm2 of powder was glued using 7:1 .irnil

acetate-collodion mixture. The machine uses two wavelengths in the measurements: Cu

K,,11.54056 A and Cu K2 ,1.54439 A. The intensity of the ac beam is twice as great

as the a2 beam. All the measurements were taken in a 20 angle range from 200 and 1200

with a 0.020 step and a scan speed of 6/min, the machine recording 1000 counts/sec.

20 is the angle between the incident beam and the reflected one. All measurements were

performed at room temperature. The computer controlled X-ray machine records the

relative intensities of the peaks which will be plotted/di- 11 i, iI (X-ray pattern) when the

scan is completed. Also, the precise angles corresponding to specific peaks were listed. The

angles corresponding to the peaks were found from Bragg's law:

n = 2d sin 0, (4-1)

where 0 is half of the reflection angle, n is an integer (n= 1 for the first order spectrum),

d is the inter-plane distance, and A is the wavelength of the incident radiation. The lattice

constants are then calculated from d and the intersection points of the lattice planes from

the desired space group number is given in terms of the Miller indices (hkl). For a cubic

symmetry the same Bragg equation can be rewritten as

sin2 0 h2 + k2 2) (4-2)

which is derived from d(hkl) = a//h2 + k2 + 2, where a is the lattice parameter. Using

a least-squares fitting program with the wavelength, structure type, (hkl) indices and the

angles 20 of the narrowest intensity lines as input, the lattice constant can be found. All

the X-ray diffractions were taken at room temperature.

4.2 Specific Heat Measurements

4.2.1 Equipment Cryogenics

This sub-section describes the probes used in the specific heat, in house resistivity and

ac-susceptibility measurements.

In house (Stewart Lab., Physics Department at University of Florida) measurements

of specific heat were performed in the temperature range of 0.3 to 2 K usually, and in

some cases up to 10 K. A home made 3He cryostat was used. The schematic drawing is

given in Fig. 4-3. This probe was used for the measurements of specific heat in magnetic

fields as well. A specially designed dewar from Cryogenic Consultants Limited was also

used. The superconducting magnet reached 14 T at 4.2 K bath temperature. For the ac

susceptibility and resistivity measurements, another 3He cryostat was used. The resistivity

was measured between the lowest temperature of 0.3 K and the room temperature, while

the ac-susceptibility was measured between 0.3 K and about 2 K. The difference from the

one used for specific heat measurements is that this probe has no 4He pot, the pumping

being performed on the Dewar in order to reach 1.1 K for the use of the 3He cooling


Specific heat measurements at lower temperature and higher magnetic fields were

performed at the Millikelvin Facility (Superconducting Magnet 1-SC'I 1), High Magnetic

Field National Laboratory, Tallahassee, Florida using a top loading dilution refrigerator

which is permanently installed in a 18/20 T superconducting magnet. The measurement

temperature range was 20 mK to 2 K combined with a magnetic field of up to 20 T. The

small home made calorimeter (Fig. 4-5) was connected to the general purpose sample

mount provided by the facility. Resistivity measurements performed at the same facility

were done in the temperature range of 20 mK to 0.9 K. Another sample holder, a so-called

16 pin ample rotator, was used. This allows the change of orientation of the sample in

field during the experiment. This holder has 16 pins (16 connection wires to the top of

the probe) that allows up to a maximum of four different samples to be measured without

pulling out the probe from the dilution refrigerator, saving precious time. Takes up to

6 hours to insert the probe into the refrigerator and cool the sample to 20 mK. Specific

heat measurements in magnetic fields up to 32 T were performed at the 33 T, 32 mm

bore resistive magnet (Cell 9), at the same National Laboratory. Another home made 3He

probe similar to the one mentioned earlier but with slightly different dimensions in order

to fit into the magnet and also to accommodate the sample in the maximum field strength

region was used. Right before the insertion of the probe into the magnet an electrical

check was done on wire connections. The quality of vacuum and sealing was checked also

using an Alcatel ASM 10 Leak Detector. For both 33 T and 45 T measurements, a special

positioning system made it possible to center the probe inside the magnet such that it did

not touch the inner walls of the magnet. The probe was to be perfectly centered into the

maximum strength field region.

Because of the large amount of heat that had to be removed, the probe was cooled

in liquid nitrogen (LN) down to the boiling point (77.35 K). After about 2 hours, when

the probe was at thermal equilibrium with the liquid nitrogen, it was transferred quickly

into a dewar in which it fits tight. The dewar was cooled in advance in LN as well. The

dewar (with the probe inside) was filled with liquid 4He (LHe) and after several hours

(depending on the volume of the can) the temperature of the probe reached 4.2 K. After

4.2 K was attained following the procedure described above, the 4He pot was filled with

LHe from the bath by opening the needle valve, and 3He gas (a lighter isotope of He)

was transferred into the 3He pot+probe line using a home-made 3He handling system.

The handling system consists of a tank filled with 3He, a pump which helps to transfer

to and back from the 3He pot line, and pressure gauges to display the amount of 3He

left in the tank and in the transfer lines. After closing the needle valve and pumping in

the 4He line a temperature between 1 and 2 K was obtained. It was necessary to refill

the 4He pot by opening the needle valve once in several hours. In order to attain 0.3 K a

completely contained 3He cooling part using a sorption pump was required. When cooled,

gases generally adsorb to solid surfaces. The sorption pump is based on the idea that

at ~10 K almost all of the 3He gas molecules are adsorbed, whereas at ~35 K all of the

molecules desorb. The sorption pump consists of a Cu cylinder that contains activated

charcoal, which has an enormous surface area (tens of square meters per gram). The

cylinder is attached to the lower end of a metallic rod. The whole system, rod+cylinder

with charcoal, was placed inside the 3He-gas enclosure. As the charcoal was lowered

toward the 3He pot, the 3He was absorbed by the charcoal reducing the vapor pressure

and lowering the temperature of the 3He pot. After the charcoal became saturated with

3He, the charcoal was warmed up (by raising the rod with the charcoal), and the gas was

released. In about 15 minutes the gas condensed and dripped into the 'He pot again.

Then the whole process was repeated. Sample Platform

The sapphire platform is attached to the bottom of the probes to the 3He or 4He

pot, depending on the probe used (Fig. 4-5 (a)). The sample is attached to a small piece

of sapphire disc using H31LV silver epoxy cured at 150C for 1/2 h. The new assembly

of sample+sapphire sitting on the sapphire platform and attached by Wakefield grease

is shown in Fig. 4-5 (b). This ensured a good thermal contact between the sample and

the platform. The platform is thermally linked to a copper ring (silver in the case of

the platform of the calorimeter used at the SC\ I NHMFL) as schematically drawn

in Fig. 4-5. Two types of platforms were used. Each platform has four wires soldered

to silver pads attached to the ring by thermally-conductive Stycast. The two pairs of

wires are connected to the platform heater and thermometer, respectively, using EpoTek

H31LV silver epoxy. The wires ensure the mechanical support of the platform and the

thermal contact with the ring and the 3He or 4He pot by the case. They also provide the

electrical contact to a heater and a thermometer on the platform. The platform heater is

an evaporated 1li-. of 7'. Ti-Cr alloy. For measurements between 1-10 K, the platform

thermometer used was an elongated piece of doped Ge, and the platform wires were

made of a Au- 7' Cu alloy. A thin piece of Speer carbon resistor and Pt-i10'. Rh platform

wires (more mechanical resistant than the ones made from the Au-alloy) were used for

measurements between 0.4 and 2 K.

4.2.2 Thermal Relaxation Method

The specific heat was measured using the probes described earlier, employing the

thermal relaxation method [79-81]. The thermal relaxation method consists of measuring

the time constant of the temperature decay of the sample connected to the heat bath by a

small thermal link. A power P is applied (Fig. 4-6) (thermal power by a small current of

the order of pA) to the platform-sample system. The temperature of the sample, initially

at To, increases by a small amount, AT. When the current is turned off, the system

temperature T(t) decays exponentially to the base temperature To:

T(t) = To + ATexp (-t/ri). (4-3)

The time constant T1 is proportional to the Ctotai (sample+platform):

T- Ctota, (4-4)

where K is the thermal conductance of the wires linking the sample+platform at

T=To+AT and the ring at T=To. The block temperature is regulated by a block heater

(a bundle of manganin wire) and measured by a thermometer attached to the block. The

time constant is obtained by measuring the time decay of the off-balance voltage signal

from a Wheatstone bridge using a lock-in amplifier. Two arms of the Wheatstone bridge

are a variable resistance box and the platform thermometer. The bridge is balanced by

adjusting the resistance of the resistance box. This made it possible to find the resistance

of the thermometer. From an initial calibration of the thermometer R versus T:

1 (45)
S A (In R)', (4-5)

it is possible to find the temperature corresponding to the platform thermometer

resistance. The thermal conductance is given by:

AP= (4-6)

where P=IV is the power applied to the platform heater. Equation (4-3) is valid if the

thermal contact between sample and platform is ideal (i.e., sample oc). If the contact is

poor (i.e., sample ~ e), then

T(t) = To + Aexp (-t/71) + Bexp (-t/7), (4 7)

where A and B are measurement parameters and T2 is the time constant between sample

and platform temperatures. Ctotla can be calculated from TI, T2, and t. The thermal

conductivity is measured by applying a power P=IV and calculating the AT as a result

of the power applied to the heater. The specific heat of the sample can be calculated

by subtracting the addenda contribution from the Ctotl. The result is multiplied by the

molecular weight and divided by the mass of the sample.

4.3 Magnetic Measurements

Magnetic susceptibility measurements were made in order to characterize the

magnetic properties of PrOs4Sb12 and its La alloys. The direct current (dc) magnetic

susceptibility was measured using a Superconducting Quantum Interference Device

(SQUID) made by Quantum Design which can perform measurements in magnetic

fields up to 5 T, and a temperature range from 1.8 K to 300 K (350 K with special

preparations). The alternating current (ac) susceptibility was measured using a home

made apparatus. The temperature range can run from about 0.3 K to 10 K, although

normally all the measurements were done up to 2 K. Both dc- and ac-magnetic susceptibility

methods are discussed in the next subsection.

4.3.1 DC Susceptibility

All the measurements were performed in 1 or 5 kOe magnetic fields. For small

samples (mass approximately a few mg), 1 T magnetic field was used since the signal of

the sample was comparable with the signal of the plastic straw holder. In any case, in

order to avoid the straws signal subtraction, the samples were kept tightly between two

drinking straws. The principle behind the magnetic susceptibility measurements is the

Lenz's law. The magnetic moment is measured by induction: the sample moves 4 cm

through a set of superconducting pickup coils and the SQUID instrument measure the

current induced in the pickup coils. The SQUID voltage is proportional to the change

in flux detected by the pickup coils. In order to get the magnetization data curve a set

of 48 points are taken during the movement of the sample. At a given temperature this

is repeated 4 times and the signals are averaged for a better accuracy. The magnetic

susceptibility for a fixed field is X = M/H (in emu/mol), where M is the magnetic

moment, and H is the magnetic field. This is obtained from the signal measured at a

fixed field by multiplying with molecular weight of the alloy, and dividing by mass and the

applied field. Beside the magnetization at fixed field and various temperatures the SQUID

can perform measurements of magnetization in different magnetic fields at constant


4.3.2 AC Susceptibility

The apparatus consists of a primary coil of NbTi superconducting wire, 90/10 CuNi

of 0.004" with insulation, 185 turns [82] and two secondary coils made from copper

wire, wound in both sides in opposite directions of 2700 turns. The coils are attached

to the Cu block (which is in thermal contact with the 3He pot). The apparatus uses the

mutual inductance principle. The sample is subject to an alternating magnetic field of

0.1 Oe produced by the primary coil (and also the Earth's magnetic field). The resulting

electromotive force (EMF) induced in the secondary coil is detected. The background

signal is nulled by the identical secondary coil, connected in series opposition. For the

same reason the two screws are identically built. The sample is glued to one screw with

General Electric (GE) varnish 7031 which ensures a good thermal and mechanical contact

at low temperature and also can be removed easily using acetone. The ac susceptibility

measurements were performed at two different frequencies: 27 Hertz (Hz) and 273 Hz. It

was deliberately used these frequencies (not integer multiples of 60 Hz) in order to avoid

the noise coming from the common electrical outlet. In general, B=/o(H + Mv)=loH(1 +

X), with H the magnetic field, My the volume magnetization and = My/H is the
magnetic susceptibility. If the applied field H has a sinusoidal form, the time dependent

magnetization Mv(t) can be expressed as a Fourier series of the non-linear complex ac

susceptibility. Applying the inverse Fourier transform to My(t) it can be found the nh

harmonic of both real and imaginary ac susceptibility. The fundamental real component

is associated with the dispersive magnetic response which is in phase with the ac applied

magnetic field, and the fundamental imaginary component is associated with absorptive or

irreversible components which arise from energy dissipation within the sample, or in other

words the energy absorbed by the sample from the ac field. The induced EMF in coils

V(t)=-dl(t)/dt (complex, i.e. V=V'+iV") is proportional to the X = X' + iX". Therefore,

if the reference signal of the lock-in amplifier is derived from the primary driving signal,

then V' oc X", and V" oc X'. The superconducting transition temperature is determined

by a midpoint of the inductive signal deviation associated with the superconduction


4.4 Resistivity

The resistivity measurements used the same probe used for the ac susceptibility.

The sample was mounted to a sapphire disk. Four platinum wires (0.002" diameter) were

attached to the sample using silver paint (whose resistivity is much lower than that of

the sample itself) and then for a good mechanical contact with EPO-TEK H31LV silver

epoxy. The extra resistance introduced by the silver-epoxy contacts is avoided by the use

of the silver paint for electrical contacts and the epoxy just only for a good mechanical

contact between the wires and the sapphire disk. Then, the disk with the sample was

glued to the 3He block using GE varnish 7031 ensuring a good thermal contact. At each

temperature, the resistivity was obtained by averaging both absolute values for each

polarity of the current. The temperatures between 77.4 K and room temperature were

covered by measurements in liquid nitrogen (LN). The probe was immersed in LN and

the program starts collecting data, while the sample cooled towards 77.4 K. Thereafter, as

described in the previous section, the LN was removed, and liquid He was transferred. The

temperature dropped further toward 4.2 K. Further down, the resistivity was measured to

approximately 0.3 K, making use of 3He gas as described.

Figure 4-1. PrOs4Sb12 large crystal, about 50 mg (right). In the left, an Os ball with
PrOs4Sb12 single crystals attached, waiting to be etched out.

t t -I

Figure 4-2. PrOs4Sb12 samples prepared for (left panel) resistivity
specific heat measurements.

and (right panel)

Pumping line
for He3 pot-----

Pumping line
for He4 pot

Cu ring with
connection pins

He3 pot

Cu block -

Platform with

Cu cylinder with

Pumping line for vacuum can
and tube for wires

-Needle valve

SCapillary connecting
He4 pot and He4 bath

--He4 pot

Block heater

SHeat sinking and wire
connection pins

_ Vacuum can

~Block thermomet

Figure 4-3. Schematic view of the 3He cryostat used in the measurements performed at
University of Florida.

Indium ring for

2 ___ Flange to connect to the
general purpose sample
holder (SCM1)

-Block heater

Ag block with .::::::::::..::::: Block thermomet
connection pins .......
T Vacuum can
Platform with -
sample --___
----~ _&Agrng

Figure 4-4. Schematic view of the calorimeter used in the Superconducting Magnet 1
(SCi\l 1), National High Magnetic Field Laboratory.

Silver Epoxy

3/8 in.
sapphire disc


Au-7%Cu or

/ Wakefield


- Sapphire


Figure 4-5. (a) Top view of the sample-platform/Cu-ring assembly. (b) Lateral view of the
sapphire platform and sample.


T Sample
CTotal = Ge thennometer
T + AT Silver epoxy
CSa mle+ CAddenda Wakefield grease

S= The four
AT An-Cu (or Pt-Rh)

Cu (Ag) ring
T, heat reservoir

Figure 4-6. Specific heat C measurement process using the relaxation time method
(Redrawn with permission from Mixson [831).


All samples were synthesized using the self-flux growth method, described in

C! Ilpter 4. The samples are cubic shaped and of sizes ranging from 1/2 mm to 3 mm

and weighting from 1 mg to about 50 mg. X-ray diffraction was performed to verify the

desired i -I I1 structure. From the diffraction pattern it was also possible to determine

the lattice constants. In addition to this, the X-rays confirmed that the samples were

single-phase within an accuracy of 5'.

The quality of the sample is also given by the sharpness of the transition in the

specific heat. A more quantitatively measure of the quality of the sample is the residual

resistivity ratio RRR=p(300K)/p(T-+O). This ratio ranges from 50 to about 170

(PrOs4Sbi2 samples studied by M6asson et al. [59] have RRRw40.)

Due to the very small size of the samples used, the susceptibilities measured for all

concentrations and the background (susceptibility of the sample holder consisting from a

plastic drinking straw) were comparable at 10 K. At room temperature the susceptibility

was even smaller than the background, especially for dilute concentrations. In order to

avoid this background contribution, magnetic susceptibility were remeasured (for x=0,

0.05, 0.3, 0.67, 0.8 and 0.95) using bigger samples. Also, in these measurements, the

material was pressed in between two long concentric tubes such that no background

subtraction was needed.

All these additional measurements yielded to a Curie-Weiss temperature dependence

above 150 K, corresponding to an effective magnetic moment close to the one expected

for Pr3+ (Fig. 5-1), much closer to the expected value for Pr3+ than the initially reported

e/ff 2.97PB [7] for PrOs4Sbl2. The effective moment of the free Pr3+ is /eff 3.58/B [52].
New measurements by Tayama et al. [51] revealed an effective moment close to this value.

This supports the notion of an essentially trivalent state of Pr in all PrlxLaxOs4Sbl2







0.02 H=b KUe

0 50 100 150 200 250 300

Figure 5-1. X(T) of PrOs4Sb12. In the inset is the Curie-Weiss fit of high temperature
(T>150 K). The high temperature effective moment is Peff 3.65pB, very close
to the one corresponding to free Pr3+, which is 3.58pB.

0 50 100 150

200 250 300

Figure 5-2. X(T) of the non-f equivalent LaOs4Sb12.






0.0004 -

LaOs4Sb12 -





Any understanding of heavy fermion (HF) behavior requires knowledge of the

crystalline electric field (CEF) configuration. Therefore, one of the main objectives of this

thesis was to establish the CEF scheme for PrOs4Sb12 and to see how the CEF ground

state is reflected in low temperature properties of this material. The C!i lpter starts with

results on the specific heat of PrOs4Sb12 in high magnetic fields. The magnetic phase

diagram (i.e. phases that exist at a given temperature and field) will allow us to determine

the CEF scheme of Pr. High magnetic field low temperature resistivity measurements will

be used to argue for a HF state in PrOs4Sb12.

6.1 Investigation of CEF Configuration by Specific Heat in High Magnetic

The initially proposed CEF schemes [7] (either 3F or F1 CEF ground state) for

PrOs4Sb12 imply non-magnetic ground states and exclude a conventional Kondo effect,

believed to be the source of HF behavior in Ce- and some U-based metals.

The controversy between the two schemes was brought about by different experiments

that seem to favor either configuration.

As presented in C(i lpter 3, the first published results such as the zero field specific

heat, magnetic susceptibility data [7], resistivity in small magnetic fields [55], inelastic

neutron scattering data interpreted using Oh symmetry [53] favored the F3 doublet as the

CEF ground state.

On the other hand, magnetic susceptibility data of Tayama et al. [51] and entropy

changes in small magnetic fields measured by Aoki et al. [56] were better fitted by a F1

CEF ground state model.

The zero field Schottky anomaly occurring at 3.1 K can be related to the P3-P5

model, assuming these two levels are split by 6.5 K, or F1-F5 model with the splitting of

8.4 K. The difficulty in interpreting these low temperature, low field results is related to

a strong hybridization of 4f and conduction electrons, inferred from the large electronic

specific heat coefficient and the size of the discontinuity in the specific heat C at Tc.

The idea behind specific heat measurements in high magnetic fields was to suppress this

coupling between f and conduction electrons to reveal the ionic character of Pr.

In order to present our results in a proper perspective we start from recalling the

specific heat data for fields smaller than 8 T obtained by Aoki et al. [56]

Figure 6-1, upper panel, shows the low temperature specific heat to 8 T obtained

by Aoki et al. [56], the lower panel a comprehensive phase diagram known before our

measurements. 4.5 T is the lowest field at which a signature of FIOP is detectable as a

small kink (at ~0.7 K). This kink evolves into a sharp peak at 0.98 K in 6 T. The C(T)

peak grows and moves also to higher temperatures for higher fields.

The FIOP was confirmed by specific heat of Vollmer et al. [63] and magnetization

study of T li- i,, et al. [51].

A number of observations brought forward the interpretation of FIOP in terms

of antiferroquadrupolar (AFQ) order. These observations included a large anomaly in

the specific heat (corresponding to a large entropy removed by the transition) and the

very small value of the ordered (antiferromagnetic) moment (about 0.025/B at 0.25 K

in 8 T [84]) measured by neutron diffraction, and also similarities to systems displaying

quadrupolar order (e.g., PrPb3 [85]).

Figures 6-2, 6-3, and 6-4 show the specific heat in fields ranging from 10 to 32 T. The

specific heat measurements in fields up to 14 T were done using Cryogenic Consultant

Limited superconducting magnet at the University of Florida. Measurements in fields

larger than 14 T were carried out at the National High Magnetic Field Laboratory,

Tallahassee, Florida using a resistive Bitter magnet. The field was applied along the

crystallographic (10 0) direction.

The specific heat data in all three figures are after subtracting the phonon background

(fiT3 with P=(1944x 103)n/(3 [1]) corresponding to a Debye temperature (GD) of 165

K, proposed by Vollmer et al. [63] This value of HD obtained from the temperature

dependence of the specific heat of PrOs4Sb12 is somewhat controversial. Other estimates

of the Debye temperature: 304 K (Bauer et al. [7]), 320 K (Aoki et al. [56]), and 259 K

(\! i!i. et al. [53]) are based on specific heat measurements of LaOs4Sb12.

The lowest temperature of the heat capacity measurements, actual value, is chosen

relatively high in order to avoid complications associated with a nuclear contribution of

Pr. This contribution is strongly enhanced by coupling with orbital moments of f electrons

[86, 87]. It is difficult to measure specific heat by a conventional relaxation method

at temperatures where nuclear degrees of freedom dominate because of additional the

time scale entering the experiment, nuclear spin-lattice relaxation time T1 [88]. Strongly

non-exponential temperature decays at the lowest temperatures (e.g., below 0.5 K in the

field of 10 T and bellow 1.5 K in the field of 32 T) indicate the importance of nuclear

degrees of freedom and cannot be analyzed using the so-called 72 correction. Therefore,

these lowest temperature points carry large uncertainty. When the magnetic field applied

along the (1 00) crystallographic direction is 10 T, the temperature of the sharp FIOP

peak appears at 1 K (Fig. 6-2). When increasing the field from 10 T field to 12 and 13

T (Figs. 6-2 and 6-3) the ordering temperature T, decreases only slightly but C(T,) is

suppressed in a strong manner.

The results presented here [31] combined with those of Aoki et al. [56] and Vollmer et

al. [63] show that T, (peak position in C) reaches a maximum value around 9 T. Also, C

at T, is maximum somewhere between 8 and 10 T.

In 13 T a shoulder appears on the high temperature side of the FIOP anomaly. The

specific heat value at this shoulder is about 3400 mJ/K mol. This shoulder evolves into a

broad maximum for H=13.5 T. Above 13.5 T the FIOP cannot be observed anymore in

the specific heat. Thus, these results strongly imply the disappearance of FIOP before T,

reaches 0.

The broad maximum that appears in 13 T exists at all fields studied up to at least 32

T. The temperature of the maximum increases with the strength of the field (Fig. 6-4).

The magnitude of this anomaly, in fields of 13 T and larger, li,,-; between 3300 and 3500

mJ/Kmol and it is field independent. These values are within about 10''. of the maximum

value for a Schottky anomaly of a two level system with identical degeneracies [89].

The uncertainty of the specific heat measurements in these fields (and at temperatures

where nuclear contribution is small) is about 1C'. Increasing 0D from 165 K, used in

the subtraction of the phonon term, to the other extremal value proposed, 320 K would

raise the estimate of the electronic part of C by about 290 mJ/K mol at 3.5 K. Thus, the

extracted values at the maximum are well within the realistic error bar of the theoretical

3650 mJ/K mol for the two-level Schottky anomaly. The highest field used of 32 T is large

enough to split any degenerate levels, therefore the observed Schottky anomaly is due to

the excitations between two singlets. Tm is related to the energy separation of the two

levels 6 by Tm=0.4176 [89]. An extrapolation of Tm to T=0 (Fig. 6-5) determines the field

at which the two levels cross, which is somewhere between 8 and 9 T.

These result can be used to infer new information regarding the plausible (i --I I1 field

configuration of Pr. Pr can be modeled by the following single-site mean-field Hamiltonian


= -CEF gJ pBJ H J(J') J S- QL(O')O,' (6-1)

where -CEF, J and Oi represent the CEF Hamiltonian for the cubic Th symmetry,

the total angular momentum, and the i-th quadrupole moment of Pr in a sublattice,

respectively, where there are five types of quadrupolar moment operators: O0, O2, Oxy,

Oyz, and Oz j and Qi are the inter-sublattice molecular field coupling constants of

spin (exchange) and quadrupolar interactions, respectively. The thermal averages of the

angular momentum and quadrupole moment of the Pr in the counterpart sublattice are

(J') and (0O).

Using the CEF parameters proposed by Kohgi [84] for the 1F-F5 CEF configuration,

Tm (with Qi=0) and the Oy,-type quadrupolar ordering temperature T, were calculated

for (1 00) direction by Aoki et al. [90] As it is demonstrated in Fig. 6-5, the measured

phase diagram and the theoretical one (in the insets) expected for the Fj F5 model for

H//(1 0 0) are in very good agreement. In both diagrams, the crossing field is very close

to the one at which the transition temperature of the FIOP becomes maximum. Thus,

the observed correlation between the two characteristic fields constitutes a very strong

argument for the F1 singlet being the lowest CEF level.

However, the level crossing for field (1 0 0) direction is also expected for F3-F4 model,

although at somewhat different field, as demonstrated by Vollmer et al. [63].

More conclusive arguments regarding the CEF configuration can be obtained from the

study of the anisotropy of the Zeeman effect. Results of our calculations for the Zeeman

effect for 1//(100), f//(110), and Hf//(111) are shown in Fig. 6-6 for FP CEF ground

state. The plots show only the four lowest CEF levels. The higher levels are at above 100

K and 200 K from the ground state, and therefore p1 i, no role in the low temperature

properties. The calculations were done neglecting exchange and quadrupolar interactions

and considering the Th symmetry. Neglecting or retaining the last two terms in (6-1) for

the (10 0) direction lead to almost identical results for eigenvalues (Aoki et al. [90] and

our results).

There is a crossing between FP and the lowest F5 level (split by magnetic field) at

about 9 T when H//(1 00) or H//(1 1) and anti-crossing when H//(1 1 0) around

the same field. Therefore, the crossing field, extrapolated from the temperature of the

Schottky anomaly at high fields should be independent of the field direction.

Figure 6-7 shows the same calculations for the F3 CEF ground state model. For

H//(1 0 0) there is a crossing between the two lowest CEF levels, although at a field
somewhat larger than the one expected for the fl CEF ground state. However, there

is no crossing expected involving the lowest CEF levels when the field is applied along

the (1 0) or (111) direction in the 73-75 model (Fig. 6-7). Therefore, measurements

of specific heat when magnetic field is applied in any direction different than (1 00)

differentiate between the two scenarios. Measurements of the specific heat in fields to 14

T were done for H//(11 0) and are presented in Fig. 6-8. The inset to Fig. 6-8 shows the

specific heat in fields between 8 and 11 T around the AFQ transition. The specific heat

at the AFQ transition and the temperature at which AFQ occurs are maximum for 9 T.

Figure 6-10 and the inset to Fig. 6-8 -ir--.- -1 that between 9 and 12 T both the specific

heat maximum and the temperature at which this maximum occurs decrease. In H 12 T

both the AFQ transition and Schottky anomaly are visible. In fields higher than 12 T the

AFQ transition is completely suppressed. The broad anomalies from Figure 6-8 at 12, 13,

and 14 T are Schottky type.

The H-T phase diagram is presented in Figure 6-10. For H//(11 0) direction we

observe a decrease of T, values with respect to the (1 00) direction for the corresponding

fields, consistent with the previous magnetization measurements [51] (Fig. 3-6). On the

other hand, within the uncertainty of the measurement, there is no change in the position

of the Schottky anomaly at 13 and 14 T, as expected for the Fl CEF ground state and

inconsistent with the F3 scenario. Moreover, for the (1 10) orientation the Schottky

anomaly can be clearly seen already at 12 T. This lower field limit for the Schottky

maximum is probably due to competition between the two types of anomalies and lower

values of T, for the (1 0) direction (Fig. 6-9).

A straight line fit for the three Tm points results in the crossing field value of 91 T.

This value agrees, within the error bar, with the estimate for the (10 0) direction. The

existence of the crossing field for the (1 10) direction provides an unambiguous evidence

for the F1-F5 model. A small misalignment of the sample with respect to the field in either

of the measurements cannot explain essentially identical crossing fields for both directions.

In fact, the measured difference in T, values for (1 00) and (1 0) directions provides an

additional check of the alignment. Similar to the (10 0) direction, there seems to be a close

correlation between the crossing field and the field corresponding to T, maximum.

Figures 6-5 and 6-10 imply a strong competition between the field-induced order and

the Schottky peak. The FIOP transition in the specific heat abruptly disappears before

T, reaches zero. Precise magnetization measurements [44, 51], on the other hand, were

able to map T, as a function of the magnetic field all the way to Tz0. This apparent

contradiction can be explained by a very small entropy available for the FIOP transition

above 13 and 12 T for fields parallel to the (10 0) and (110) directions, respectively.

Specific heat, being a bulk measurement, can be less sensitive than magnetization

techniques in this situation. A strong competition is to be expected in the F1-F5 scenario.

The ground state pseudo-doublet formed at the level crossing carries both magnetic and

quadrupolar moments. Since a quadrupolar moment operator does not commute with a

dipolar one, the quadrupolar interactions leading to FIOP compete with the magnetic

Zeeman effect.

Therefore, the high magnetic fields measurements of specific heat [31] provided the

first unambiguous evidence for the singlet CEF ground state of Pr in PrOs4Sb12. This

result was confirmed by recent inelastic neutron scattering experiments [32] analyzed in

the Th symmetry, and our magnetoresistivity results described in Sections 6.2 and 7.2.

6.2 Magnetoresistance of PrOs4Sb12

Magnetoresistance of PrOs4Sb12 was measured to search for further experimental

evidences of the proposed CEF scheme and for possible signatures of heavy-fermion


The main indication of heavy electrons in PrOs4Sb12 is the large discontinuity in

C/T at T,. The mass enhancement inferred from specific heat measurements is of the

order of 50 [7]. This value is an estimate and there is no consensus on a precise value. An

uncertainty exists in evaluation of the effective mass directly from the low temperature

zero-field specific heat, because there is no straightforward method of accounting for the

CEF specific heat. The corresponding Schottky anomaly is strongly modified because of

the hybridization between the f and conduction electrons. The zero-field specific heat just

above T, is dominated by CEF effects.

Several other estimates of m* have been proposed. For instance, Goremychkin et al.

[32] -,r-.-. -I. m* enhancement to be about 20. However, their estimate was based on the

Fulde-Jensen model, which we do not believe is relevant to PrOs4Sb12. This enhancement

is 3-7, according to the de Haas-van Alphen measurements [44]. However, dHvA effect

was analyzed over a wide range of fields 3-17 T and did not take into account m* being

dependent of H [44].

The residual resistivity ratio RRR=p(300K)/p(T-+O) of the investigated sample was

about 150. This value is among the highest reported, implying high quality of our sample.

Both the current and the magnetic field were parallel to the (1 00) direction (longitudinal

magnetoresistance). The measurements were done using the 18 T/20 T superconducting

magnet at the Millikelvin Facility, National High Magnetic Field Laboratory, Tallahassee,

Florida. The temperature range was 20 mK to ,0.9 K, the maximum field used 20 T.

Measurements at the University of Florida were done in fields up to 14 T down to 0.35 K.

The zero-field electrical resistivity, another important characteristics of heavy

fermion metals, does not provide a straightforward support for the presence of heavy

electrons. Maple et al. [91] found that the resistivity, between 8 and 40 K, follows a

fermi-liquid temperature dependence (p=po+AT2). Our resistivity data between 8 and

16 K follows the above mentioned dependence (Fig. 6-11) with A =0.009 cm/K2 (in

agreement with A found by Maple et al. [91]). As inferred from Kadowaki-Woods (KW)

relation (A/7l y x110-5 Rcm(mol K/mJ)2) [60] this value of A implies a small electronic

specific-heat coefficient 7 ,30 mJ/K2 mol, comparable to the one measured for LaOs4Sb12.

So, evidently there is an upper temperature limit (less than 8 K) for the heavy fermion


Figure 6-12 shows the resistivity of PrOs4Sb12 in H=3, 10, 15, 16, 17 and 18 T in

a temperature range of 20 mK to 0.9 K. The resistivity below 200 mK saturates for all

fields. This temperature dependence at the lowest temperature was also observed by other

groups [91, 92]. Therefore, the resistivity for all other intermediate fields was measured to

350 inK. Figure 6-13 shows the resistivity between 350 mK and about 1.3 K for several

relevant fields (3.5, 5.5, 7, 10, and 13 T).

Maple et al. [91] proposed the following temperature dependence for fixed magnetic

field: p=po+aT", with n>2. In their study (transversal magnetoresistivity) n was ~3 for 3

T and 2.6 for 8 T. In our longitudinal case these exponents are slightly larger (e.g., 3.9 for

3 T). The exponents depend on the temperature range of the fit, i.e., n becomes smaller

when the upper temperature limit of the fit decreases. The residual resistivity po values

resulted from the fit on different temperature ranges (included in the 350 mK and 0.9 K

interval) were close to p at 20 mK. The residual resistivity po attains a maximum near

H=10 T, field corresponding to the crossing between the two lowest-energy CEF levels of

Pr (Fig. 6-14, lower panel). In this region (around 9-10 T) the lowest two singlets form a

quasidoublet possessing quadrupolar degree of freedom. These electric quadrupoles order

at sufficiently low temperatures [56] with the ordering temperature having maximum in

the crossing field [31, 56]. Resistivity is dominated therefore by the CEF effects or the

quadrupolar ordering. This ordering is completely suppressed by fields higher than 15 T.

As it can be noticed from Figs. 6-12 and 6-14 the residual resistivity po does not change

substantially in fields higher than 15 T. In fact, it can be seen than the residual resistivity

po versus H field can map the boundary of the AFQ phase, i.e. a sharp increase of po

indeed coincides with the AFQ boundary, indicated by arrows in Fig. 6-14, lower panel.

The same conclusion can be drawn from resistivity measurements for high magnetic

fields perpendicular to the current [55]. The rate of the increase of the resistivity with

temperature is still changing above 15 T (Fig. 6-12). It can be concluded that the

reduction of the temperature rate correlates with an increase of the energy between the

lowest CEF levels. A precise accounting of these changes is difficult since neither of the

functions checked out describe accurately the variation p(T) in a fixed field.

A linear dependence of p on T2 is accounted by resistivity (p po+AT2) in different

temperature ranges (above 0.4 and 0.5 K), as seen in Figs. 6-13. Using the KW ratio [60]

(however, there is no experimental or theoretical studies on A/72 for Pr-based systems)

the electronic specific heat coefficients for H=3, 10, and 18 T are about 200, 400, and

200 mJ/K2 respectively. The A coefficient deduced from the narrow range of temperature

(Fig. 6-15) increases sharply with the magnetic field and reaches a maximum near 6 T.

After a plateau between 6 and 12 T, a strong decrease is encountered. This establishes

a correlation between A and both AFQ order and CEFs with a strong increase when

approaching the AFQ boundary.

A characteristic field dependence of the residual resistivity (Fig. 6-14, low panel)

was associated both to CEF effects and long range AFQ order. The CEF effect on the

resistivity was considered by Frederick and Maple [93] using the following expression:

PCEF [Tr(PQM) + Tr(QA)]. (6-2)

The first term represents a contribution due to exchange scattering, and the second term

is the contribution due to aspherical (or quadrupolar) scattering. The aspherical Coulomb

scattering is due to the quadrupolar charge distribution of the Pr3+. Matrices Pij, QM and

Q j are defined as follows:

E p(E, Ej)
P = (63)
Z e- r 1- e-(E, -Es)' (6 3)
1 1
QJ J j2 (i j 2 J 2, (6-4)
QAi 22
In the above relations Ei are the eigenvalue of the CEF eigenstates, the Wi)'s are the

CEF eigenstates, 3=1/(kBT), and the yp's are the operator equivalents of the spherical

harmonics for L=2 (i.e., quadrupolar terms) [94]. The Qyi-matrices are normalized to each

other [95] such that

QYM = Q (2J+)J( 1 J+1)J4 180. (6-6)

The most intriguing conclusion is a strong enhancement of the A coefficient with the

magnetic field between 2 and 6 T. This could imply an enhancement of m* for fields in

this range. Low temperature resistivity calculation for F3 and Fl CEF ground state are

shown in Fig. 6-16 (upper and lower panel, respectively). Our residual resistivity (at

T=20 mK) seems to be in a better agreement with calculations for F3 than for F1 ground

state. Thus, CEF cannot account for the magnetoresistance of PrOs4Sb12. We will return

to this interesting problem while describing the magnetoresistivity of La-doped ( i--I '1-

Furthermore A seems to have a maximum value near the field separating ordered

and non-ordered phases. Note that this is not the crossing field for the lowest CEF levels.

Thus, these results -'ir-., -I a possibility of m* enhancement due to strong fluctuation of

the AFQ order parameter.

...[.. ..s.... .
4 H=
E 0".. T
"-" ..."" I" I '

02 max. in M vs T
8 4T
0 2 2 K 4 6

T i.i
6- TX

max. in M vs T

Smax. in dCfdH vs H

2 PrOsSb 1

#,oH <100>
Superconductivity H
0 1 2 3 4

Figure 6-1. Specific heat C of PrOs4Sbl2 in fields up to 8 T for f//(10 0) (upper panel).
The magnetic field phase diagram H-T of PrOs4Sb12 in fields up to 8 T for
1H//(100) (lower panel) (Reprinted with permission from Aoki et al. [56]).




E *

| 2000- *** H=10 T
"3 H= 10T
v 6000 I ,- I ,- I ,- I


S" 0* H=12T

0 1 2 3 4 5
T (K)

Figure 6-2. Specific heat C of PrOs4Sbl2 in 10 and 12 T in the vicinity of FIOP transition
for //(10 0) (Reprinted with permission from Rotundu et al. [31]).




2000 H= 13 T
8000 I I I I i

** m
2000 H= 13.5 T
8000 I I I I


4000 -

2000 H= 14 T

0 1 2 3 4 5
T (K)

Figure 6-3. Specific heat C of PrOs4Sb12 in 13, 13.5, and 14 T, for H//(100). A shoulder
appears at about 1.2-1.3 K at 13 T and the FIOP transition is suppressed at
13.5 T.


3000 -


1000 H= 16 T
4000 I I I I I

i .*** .
E 3000 .

2000 -

1000 H= 20 T
4000 I I I I I



1000 H= 32 T

0 1 2 3 4 5
T (K)

Figure 6-4. Specific heat C of PrOs4Sbl2 in magnetic fields of 16, 20, and 32 T, for
H//(1 00).


30 H/(100) i -

25 10 .

20 -


10 [

0 1 2 3 4
T (K)

Figure 6-5. Magnetic field phase diagram H-T of PrOs4Sbl2 for H//(1 00) (H>8 T).
Filled squares represent the FIOP transition. Open squares correspond to the
Schottky anomaly. The inset is the model calculation of the Schottky anomaly
assuming the singlet as the ground state[90]. The solid line represents the
FIOP boundary; the dashed line corresponds to a maximum in C (Reprinted
with permission from Rotundu et al. [31]).

H//[100] H//[ 110] H//[111]
-60 -60 -60

-70 Z -70 --70

-80 / -80 -- -80

-90 -90 --90

-100 --100 --100

-110 -110 -110

-120 -120 -120
0 10 20 0 10 20 0 10 20
H (T)

Figure 6-6. Zeeman effect calculations for PrOs4Sb12 in the F1 CEF ground state scenario.
There is crossing of the two lowest levels for H//(1 0 0) or H//( 111) at
around 9 T and anti-crossing at the same field for H//(11 0). The figure
shows only the two lowest levels, i.e. the singlet F1 and the triplet F5.

H// [100 ]








H//[ 110]

0 10 20 0 10
H (T)

H//[ 111]

-70 /





20 0 10 20

Figure 6-7.

Zeeman effect for PrOs4Sb12 in the F3 CEF ground state scenario. The effect
in strongly anisotropic. There is no crossing of the two lowest CEF levels for
H1//(11 0) or H//7(111). The figure shows only the two lowest levels, i.e. the
doublet F3 and the triplet F5.





7000 ,
H /(116) A 8T A
A H=11T 600 9T
6000 H=12T 9.5T ,
e 10T
H=13T 5000 v 10.5T '.
H=14T 11T ..1 *
5000 .

40004000 :
4000 A 1 1 -
E t 3000
S0.6 0.7 0.8 0.9 1.0 1.1

2000 w A.

0 1 2 3 4 5
T (K)

Figure 6-8. Specific heat C of PrOs4Sbl2 for H//( 1 0), H=10, 12, 13, and 14 T. The
inset shows C versus T neat T, for 8, 9, 9.5, 10, 10.5, and 11 T.

5000 H=12T
5000 -,
** H /(100)


3000 x
0 *
S 2000 I I I i | I | i |
3500 H//(110)-

0 as NO=Em

3000 ,


2500 -

0.0 0.5 1.0 1.5 2.0
T (K)

Figure 6-9. Specific heat C of PrOs4Sbl2 in H=12 T, for H//10 0) (upper panel), and
H//(1 1 0) (lower panel). The arrow indicates the AFQ transition.


T (K)

Figure 6-10. The magnetic field phase diagram H-T of PrOs4Sb12 for H//(l 1 0) (H>8 T).
The inset shows the specific heat Cma, of AFQ versus H. For a definition of
symbols see Fig. 6.5.

15 I I I I
T =1.85K
20 1O

15 0 2 4 6 8 10 12 14 16
T (K)

2- 10 P=Po+AT2
A=0.009 pncm/K2


0 50 100 150 200 250
T2 (K2)

Figure 6-11. Electrical resistivity p versus T2 for PrOs4Sb12. In the inset is p versus T
showing the superconducting transition at T-=1.85 K.

6000 *
E -E

5000 o

8 9 10 11
H (T)


4.2 I
H =10T


3.6 -H =15

o 3.4 H =16T-
3.2 H =17T
a. H =18T

3.0 H = 3T


2.6 I I
0.0 0.2 0.4 0.6 0.8 1.0

Figure 6-12. Resistivity p(T) between 20 mK and about 0.9 K of PrOs4Sb12 in 3, 10, 15,
16, 17, and 18 T (Reprinted with permission from Rotundu and Andraka


5.0 ^ 10T



3.5 3.5 T


0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
T (K2)

Figure 6-13. Resistivity p versus T2 of PrOs4Sbl2 for 3.5, 5.5, 7, 10, and 13 T.

2.5 1 1 1 5

p=Po+ aT 4

2.0 n>2 C
3 .

2 -
S1.5 0 4 8 12 16

S. H(T)
1.0 -

0.5 -

0 2 4 6 8 10 12 14 16 18

3.5 -



0 2 4 6 8 10 12 14 16 18
H (T)

Figure 6-14. Coefficient a (p=po+aT") versus H for PrOs4Sb12 fields up to 18 T is in
upper panel. The residual resistivity po(H) is shown in lower panel.



p=po +AT 2

0 2 4 6 8


10 12 14

Figure 6-15. Coefficient A (p=po+AT2) versus H.

2.5 I-

0.5 -


2 1.4 K
0.5(0.5pmaa + 0.5pACs)

S 0.35 K r. ground state
0 5 10 15 20
H (T)

41.4 K /
PrOs4 Sb12

Q 4.2 K --------

r. mag ACS
S0.35 K r ground state

0 1 I I I
0 5 10 15 20

Figure 6-16. The calculated p(H) of PrOs4Sb12, for both F3 and Fi scenarios. The vertical
line indicates the field crossing of the two lowest CEF levels. Note that
the crossing field for the F1 ground state was assumed at 3 T (lower panel)
(Reprinted with permission from Frederick and Maple [93]).


In this C'!I pter the La-alloying study of PrOs4Sb12 by dc and ac susceptibility,

specific heat and resistivity is presented. One of main objectives of this work was

to differentiate between different proposed models of the conduction electron mass

enhancement in PrOs4Sbl2. Mechanisms that have been considered range from single-ion

models, such as the quadrupolar Kondo effect [7, 16] or virtual CEF excitations [32, 36],

to cooperative models invoking proximity to a long-range order (proximity to the low

temperature state of AFQ order) [56]. While investigating the applicability of these

models, close attention was paid to whether the single-ion parameters such as the CEF

spectrum and hybridization parameters vary with the alloying.

7.1 Lattice Constant

The room temperature X-ray diffraction patterns for several La concentration are

given in Fig. 7-1. The results of the X-ray diffraction analysis were consistent with single

phase materials. A very small and monotonic increase of the lattice constant, a, with the

La content is detected (Fig. 7-2). The lattice constant is calculated from the high index

line (8 2 2) of the X-ray diffraction pattern. Calculations from smaller angle lines result

in the same dependence of a on x, but had a much larger scatter. These small changes

(about 0.0 :'.) between the end compounds is in agreement with previously reported [57]

and almost non-existent lanthanide contraction in ternary skutterudites containing Sb,

of a general form LnT4Sbl2, where T and Ln are transition element and light lanthanide,


To present this change in a proper perspective we recall that the change of the

lattice constant across Pr(Osl-xRux)4Sbl2 [97] is 10 times larger. This is despite the fact

that the atomic radii of Os and Ru are almost identical (1.35 and 1.35 A for Os and Ru

respectively), while La is much larger than Pr (1.88 versus 1.82 A). In Pr(Osl-xRu1)4Sbl2

the CEF parameters increase monotonically with x. Very small changes in lattice

constant in PrlxLa1Os4Sbl2 sil--. -i small, if any, changes in the CEF parameters

and hybridization. These parameters depend on the position of ligand atoms with respect

to Pr. These small changes of the lattice constant in Prl_-LaOs4Sb12 provide a unique

opportunity for the alloying study of superconductivity and other phenomena that are

strongly influenced by microscopic inhomogeneities associated with the lattice constants'


7.2 DC Magnetic Susceptibility

Susceptibility measurements were done on single-crystal samples of masses ranging

between a few mg to 50 mg belonging to different batches. Specific heat measurements

were performed as well on samples characterized by the susceptibility.

Figure 7-3 shows the susceptibilities only in the range 1.85 to 10 K. All data are

normalized to a Pr mole. Due to the very small size of the samples used in the initial

measurements, the measured moment of most of the samples and the background (the

magnetic moment of the sample holder consisting of a piece of a plastic drinking straw)

were comparable at 10 K. At room temperature the magnetic moment of the samples

was even smaller than the background, especially for dilute concentrations. In order to

avoid this background contribution, the magnetic susceptibilities were remeasured (for

x=0, 0.05, 0.3, 0.8 and 0.95) using several < iv--I 1- and holding them between two long

concentric straws. No background subtraction was needed this time. The Curie-Weiss

temperature was found above 150 K and the effective magnetic moment in the range

3.2-3.6PB/Pr atom. The values are in the range of moments reported for pure PrOs4Sb12.

Some discrepancy between these values and that expected for Pr3+, 3.58pB [52], can be

due to an error in mass determination. Because of the very fragile nature of these < i --i i-

some of them broke off during the measurement and small fractions moved in between

the two tubes. A further check of the magnetic moment was performed on one large

crystal for x=0.67 (~20 mg each). Figure 7-4 shows the susceptibility and the inverse of

susceptibility for x=0.67. From the Curie-Weiss fit the high temperature effective moment

is found to be 3.62MB/Pr mol, close to the value expected for Pr3.

All low temperature susceptibility data (Fig. 7-3) exhibit a broad maximum at 3-5

K due to excitations between the lowest CEF states. Our high magnetic field specific

heat study [31], and neutron [32, 84] measurements established F1 singlet as the CEF

ground state separated by about 8 K from the first excited F5 triplet. Very small changes

in the position of these maxima in the susceptibility are the first indication that CEF

are essentially unaltered by the doping as expected from the measurement of the lattice


Another interesting aspect of the susceptibility is a strong initial reduction of the

low-temperature values of X (normalized to a mole of Pr) by La. The reduction of the

maximum susceptibility from approximately 100 for x=0 to about 50 memu/Pr mol for

x=0.4 is clearly outside the error bar. The aforementioned measurements on assemblies of

crystals for x=0.8 and 0.95 also resulted in a 4 K value of about 505 memu/Pr mol for

both compositions. Some broadening and decrease in magnitude of the CEF susceptibility

are expected in mixed alloys due to increased atomic disorder. However, the very large

initial drop in the susceptibility and lack of variation above x=0.4 might indicate that

some characteristic electronic energy (analogous to a Kondo temperature) increases

sharply upon substituting La for Pr. A similar suppression of the corresponding maximum

is observed in the specific heat data discussed in the next section.

7.3 Zero Field Specific Heat

7.3.1 Specific Heat of PrOs4Sb12: Sample Dependence

The main evidence for heavy fermion behavior in PrOs4Sb12 comes from a large

discontinuity of the specific heat at T,. Specific heat provides evidence for unconventional

superconductivity. The evidence includes a power low dependence of C below Te, and

the presence of two distinct superconducting transition. PrOs4Sb12 was initially reported

to have a single superconducting transition at T-=1.85 K [7]. More recent specific heat

measurements revealed two superconducting transitions (Vollmer et al. [63], Maple et al.

[53], Oeschler et al. [64], Cichorek et al. [65]).

Before discussing zero field specific heat measurements of Prl-La0Os4Sbl2 we need

to comment on the sample dependence of the specific heat of PrOs4Sb12. This is in order

to distinguish La-induced changes from variation related to sample quality. In Fig. 7-5

we present the specific heat near T, for three representative samples from three different

batches. All three samples were obtained in an identical way. All three samples have

pronounced lower temperature transitions. The upper temperature transition is less

distinct and sample dependent. Our convention of defining Tji and Tc2 (by local maxima

or shoulders in C/T versus T) is illustrated in Fig. 7-5. The upper transition temperature

T,~ is identical for all three samples. There seems to be some sample dependence for the

lower transition temperature Tc2. However, the variation is very small considering that

each C/T at a given T in Fig. 7-5 was obtained by integrating the specific heat over

0.04T interval. The width of the transition defined, for comparison reason, by T3-Tc (Fig.

7-5) is large and approximately equal for all the samples. Finally, A(C/T) defined as the

difference between C/T at Tc2 and T3 is about 800 mJ/K2 mol.

Our observations are consistent with other, particularly more recent, reports. Almost

all recent investigations find two superconducting anomalies, more pronounced at Tc2 and

less defined at Tci. An exception to this rule are unpublished data by Aoki et al. [98]

that show a sharp peak at T1I, and only a change of slope in C/T at Tc2. The width of

the transition, ~0.2 K, defined above, is quite similar for all published data. There is

a large distribution of reported A(C/T) at Tc, from 500 to 1000 mJ/K2 mol. A usual

determination of A(C/T) by an equal area (conservation of entropy) construction cannot

be applied due to the presence of two superconducting transitions. Applying our method,

C/T(Ta)-C/T(T3), results in an average A(C/T) of 800 mJ/K2 mol for the most recent


7.3.2 Zero Field Specific Heat of Pr1_-LaOs4Sb12

In order to account for normal-electron and phonon contributions to the specific heat

of Pr1_-La1Os4Sb12 alloys, the specific heat of LaOs4Sb12 was measured. The normal state

specific heat between 1 and 10 K is shown in Fig. 7-6 in the format of C/T versus T2.

The results can be expressed by the following equation

C/T = 56 + 1.003T2 + 0.081T4 4.260 x 10-4T6, (71)

where C/T is expressed in mJ/(K2 mol) and T in K (Fig. 7-6). A significant nonlinearity

in C/T versus T2 is probably due to the rattling motion of loosely bound La atoms [99].

Values of 7 and / for LaOs4Sb12 reported by other research groups are: 7 of 36 [54], 55

mJ/K2 mol [100], 56 [44] and j=0.98 mJ/K3 mol [101].

Figures 7-7 and 7-8 present the f-electron specific heat of Pr1_-La0Os4Sb12 alloys, i.e.

the specific heat of LaOs4Sb12 and, normalizing to a mole of Pr. Note that the phonon

specific heat of pure PrOs4Sb12 in C'! Ilpter 6 was taken from Vollmer et al. [63], which was

derived by fitting the total specific heat C to a function representing phonon, conduction

electrons, and Schottky contributions. However, using the LaOs4Sb12 specific heat seems

to be more justifiable for moderately and strongly La-doped alloys and therefore this way

of accounting for phonons is consistently used in this chapter on La alloying. Evolution of T, with the La Doping

Figure 7-7 shows the specific heat for x=0, 0.05, 0.1, and 0.2 near the superconducting

transition temperature. As already discussed, the pure compound has two superconducting

anomalies in the specific heat. The specific heat data for x=0.05 exhibit a shoulder which

seems to correspond to the anomaly at T,1 for x=0. The width of the transition (T3-Tc2)

for x=0 is about 0.2 K. This width becomes slightly smaller for x=0.05. The specific heat

for x>0.3 alloys (Fig. 7-8) exhibits one superconducting transition only. The width of the

transition for this group of materials is about a half of the width of the pure material.

This reduction cannot be accounted for by the reduction of T, itself. A similar conclusion

about a drastic reduction of the width of the transition can be derived from graphs in

which T is replaced by a reduced temperature T/T,. The reduction is probably related to

a disappearance of one superconducting transition (at TI,).

Open symbols in Fig. 7-9 denote the lower temperature superconducting transition,

and the filled squares symbolize the higher temperature superconducting transition

Tc2. Lanthanum doping has a surprisingly weak impact on Tc. This weak dependence

(approximately linear) of T, on x in Pr1_-LaxOs4Sb12 is unusual for heavy fermion alloys.

For instance, heavy fermion superconductivity in UBe13 is completely suppressed by only

. La [102] substituted for U. Furthermore, since PrOs4Sb12 is clearly an unconventional

superconductor (e.g., time reversal symmetry breaking) while LaxOs4Sb12 is presumably

a conventional superconductor we would expect, while varying x, a suppression of one

type of superconductivity before the other type emerges. Figure 7-10 shows that there is

smooth evolution of T, (and superconductivity) between the end-compounds. A somewhat

stronger suppression is observed in the case of Ru replacing Os [71, 97]. But even in this

case, the T, reduction rate is small if compared with the in i ii i iy of Ce- and U-based

heavy fermions and considering the fact that Ru alloying drastically affects CEF energies

and hybridization parameters. The Discontinuity in C/T at Tc

Existence of two distinct superconducting transitions in PrOs4Sb12 makes the

determination of the discontinuity in C/T somewhat arbitrary. Furthermore, it also

complicates the interpretation of this discontinuity. Despite a substantial recent

improvement in sample quality, the question whether the two transitions correspond

to different regions of the sample becoming superconducting at different temperatures

or whether the lower transition corresponds to the change of the symmetry of the

superconducting order parameter in a homogeneous medium is not completely settled.

The comparable magnitude (equal as argued by Vollmer et al. [63]) of the anomalies at

T,~ and Th2 precludes a popular speculation that one of these transitions is associated with

surface superconductivity.

As it was already stressed, this A(C/T) is currently the main evidence for the

presence of heavy electrons. The presence of a modified Schottky anomaly near T, makes

a direct determination of the electronic specific heat coefficient unreliable. In a BCS-type

superconductor, AC/T, is related to the electronic specific heat coefficient 7. In general,

AC/T, and 7 are affected by the coupling strength of the Cooper pairs and can vary by a

factor of order of 2-3. Nevertheless, for lack of any other measure, we use this quantity as

an indication of heaviness of electrons in Pr1_,LaOs4Sbl2.

We recall that LaOs4Sb12 is also a superconductor, therefore for dilute Pr concentrations

the normalization of C/T to Pr mole (used in Fig.s 7-7 and 7-8) has no meaning.

Therefore, Fig. 7-10 shows the total A(C/T) per formula unit. There is a drastic decrease

of A(C/T) with x for 0
800 for x=0 to 280 for x=0.2 and further to about 160 mJ/(K2 Pr mol) for x=0.3 (C/T

is seven fold reduced with x, for x between 0 and 0.3). A(C/T) -1 i,-; approximately

constant with x for 0.3
LaOs4Sbl2. Therefore, these results -i,--.- -1 that the heavy fermion character disappears

above x=0.3. Therefore, there is a lack of strong correlation between the heavy-fermion

character as measured by A(C/T) at T, or AC/T, and the average T,. Thus, the results

argue also for different mechanisms responsible for the heavy fermion state and enhanced

value of T, in PrOs4Sb12. The Schottky Anomaly

The evolution of Schottky-like anomaly in pure PrOs4Sb12 near 3 K with magnetic

fields provided important information on the CEF configuration of Pr. The discrepancy

between the theoretically predicted temperature dependence and the observed one has

been explained by mixing of f- and conduction-electrons degrees of freedom, reducing the

entropy associated with the pure Schottky anomaly. The results on the discontinuity of

C/T at T,, presented in the previous section, -Ii--.- -1 that the heavy fermion character is

strongly suppressed by the La alloying. Thus, one would expect the mixing to be reduced

for strongly La-doped samples and the Schottky anomaly to approach the theoretically

predicted temperature dependence. Recall that the maximum value of the specific heat

in PrOs4Sb12 is about 7100 mJ/K mol, smaller than the theoretical value of 8500 for

singlet to triplet excitations. However, as it can be inferred from Figs. 8.7 and 8.8 the

specific heat at the Schottky-like maximum decreases noticeably with the alloying. Some

reduction and broadening of such a maximum due to CEF excitations can be explained by

disorder effects induced by the alloying. However, this effect should be small. La does not

change the local symmetry of Pr nor the distances between Pr and its nearest neighbors.

Furthermore, the main reduction takes place between x=0 and x=0.2 and, within our

error bars, there is no appreciable change beyond x=0.3. This reduction, by a factor of 2,

is about the same as the decrease of the low temperature magnetic susceptibility, discussed

in Section 7.2.

In order to study further this specific heat reduction, a large crystal of Pr0.33La0.670S4Sb12,

for which addenda heat capacity is negligible below 6 K, was investigated. First of all, the

effective magnetic moment measured at room temperature for this particular crystal

is very close to the expected value for Pr3+. Thus, the reduced specific heat cannot be

explained by incorrect Pr stoichiometry, nor by some Pr ions being in a mixed-valent

state. The results, after subtracting the specific heat of LaOs4Sb12 and dividing by 0.33,

are shown in Fig. 7-11 in the form of C/T. This graph shows also a fit to the function

describing a Schottky specific heat for a singlet-triplet excitations, scaled by a factor

a=0.44. A similar scaling was used by Frederick et al. [71] to account for the specific heat

data of Pr(Osl-xRux)4Sbl2 in terms of the singlet CEF ground state. A necessity to use

such a small scaling factor for this model of CEF (of about 0.5) was used by Frederick

et al. [71] to argue for a doublet CEF ground state. However, as can be seen from Fig.

7-12, a reasonable fit to the doublet CEF model also requires a scaling factor, although

somewhat larger (a=0.73). Finally, a fit to the singlet-to-singlet scattering requires no

scaling at all. The fit shown in Fig. 7-13 obtained with a as an adjustable parameter,

resulted in a 1.009 (within 1 .).

Full Text








Iamdedicatingthisworktomydearesthumanbeings,myparentsElenaandConstantinRotundu.Icannotndanadequatewaytoexpressmyloveandgratitudetothem.Ithankthemfortheirinnitesupportinordertocompletemyeducation.IamgratefultomymotherandmybrotherRomulusNeculaiRotunduforbelievinginme.Iowemuchtomyadviser,Dr.BohdanAndraka.Hehasbeenanincrediblesourceofguidanceandinspiration.Hewasagreatadviserwithendlesspatience.Myeducationwouldnothavebeenpossiblewithouthisnancialhelp(throughDOEandNSF).Iamdeeplyindebtedtohim.SpecialthanksgotoProf.YasumasaTakanoforteachingmesomanyexperimentaltricks;andfordiscussions,support,andgreatcollaboration,especiallyattheNationalHighMagneticFieldLaboratory(NHMFL).Hewasanendlesssourceofenergy.IwouldliketothankProf.GregoryR.Stewartforlettingmeusehislaboratory.Ithankmyothersupervisorycommitteemembers(Profs.BohdanAndraka,GregoryR.Stewart,YasumasaTakano,PradeepKumarandIonGhiviriga)forreadingthisworkandfortheiradvices.IreceivedhelpwithmanyexperimentsatNHMFLandourlabfromDr.HiroyukiTsujii.IthankDrs.JungsooKimandDanielJ.MixsonIIfortheirtechnicaladvice.OtherpeopleintheeldIwouldliketoacknowledgeareProf.PeterHirschfeld,thenestprofessorIeverhad,whogavemeinsightonthetheoryofcondensedmatterphysics;andDrs.EricC.PalmandTimP.Murphyfortheirhelpandsupportovermorethan4yearsattheSCM1/NHMFL.IthankCenterofCondensedMatterSciencesforthenancialsupportthroughtheSeniorGraduateStudentFellowship.Lastandnotleast,Iwouldliketothankmyhighschoolphysicsteacher(DumitruTatarcan),whoencouragedandguidedmyrststepsinphysics. 4


page ACKNOWLEDGMENTS ................................. 4 LISTOFTABLES ..................................... 7 LISTOFFIGURES .................................... 8 ABSTRACT ........................................ 12 CHAPTER 1INTRODUCTION .................................. 14 2THEORETICALBACKGROUND ......................... 17 2.1TheCrystallineElectricField(CEF)forCubicGroup ............ 17 2.2ConductionElectronMassEnhancement(m)MechanisminPrOs4Sb12 22 2.2.1QuadrupolarKondoEect ....................... 22 ............................. 24 25 2.2.2Fulde-JensenModelformEnhancementinPrMetal ........ 26 2.2.3FluctuationsoftheQuadrupolarOrderParameter .......... 26 3PROPERTIESREVIEWOFTHEPrOs4Sb12 30 3.1CrystallineStructure .............................. 30 3.1.1RattlingofPraseodymiumAtom .................... 30 3.1.2Valence .................................. 31 3.1.3CrystallineElectricFields ....................... 31 3.2Normal-StateZero-FieldProperties ...................... 33 3.2.1SpecicHeat ............................... 33 3.2.2deHaasvanAlphenMeasurements .................. 33 3.2.3Resistivity ................................ 34 3.2.4DCMagneticSusceptibility ....................... 35 3.3TheLong-RangeOrderinMagneticFields .................. 35 3.4Superconductivity ................................ 35 3.4.1UnconventionalSuperconductivity ................... 36 .................... 36 37 .......... 39 ................. 39 3.4.2ConventionalSuperconductivity .................... 40 ............. 40 ....... 40 .......... 40 5


........................... 49 4.1TheSamples:SynthesisandCharacterization ................ 49 4.1.1Synthesis ................................. 49 4.1.2X-RaysDiractionCharacterization .................. 51 4.2SpecicHeatMeasurements .......................... 52 4.2.1Equipment ................................ 52 ........................... 52 ....................... 55 4.2.2ThermalRelaxationMethod ...................... 55 4.3MagneticMeasurements ............................ 57 4.3.1DCSusceptibility ............................ 57 4.3.2ACSusceptibility ............................ 58 4.4Resistivity .................................... 59 5MATERIALSCHARACTERIZATION ....................... 65 6PrOs4Sb12 67 6.1InvestigationofCEFCongurationbySpecicHeatinHighMagneticFields 67 6.2MagnetoresistanceofPrOs4Sb12 73 7Pr1XLaXOs4Sb12 93 7.1LatticeConstant ................................ 93 7.2DCMagneticSusceptibility .......................... 94 7.3ZeroFieldSpecicHeat ............................ 95 7.3.1SpecicHeatofPrOs4Sb12:SampleDependence ........... 95 7.3.2ZeroFieldSpecicHeatofPr1xLaxOs4Sb12 96 ............. 97 98 .................... 99 7.4SpecicHeatinLargeFields .......................... 101 7.5MagnetoresistanceofPr1xLaxOs4Sb12 103 7.6UpperCriticalFieldHc2 107 7.6.1ACSusceptibility ............................ 107 7.6.2DeterminationofHc2(T)bySpecicHeatMeasurementsinSmallMagneticFields ............................. 108 8CONCLUSION .................................... 136 REFERENCES ....................................... 139 BIOGRAPHICALSKETCH ................................ 145 6


Table page 2-1TherelevantstatesforthequadrupolarKondoeect. ............... 23 3-1Thevaluesreportedbydierentgroups,extractedfromtsofspecicheatbelowTc. ....................................... 38 3-2Thevaluesreportedbydierentgroupsfromothermeasurementsthanspecicheat. .......................................... 38 7


Figure page 2-1CubicpointgroupsymmetryTh. .......................... 27 2-2Lea,Leask,andWol0srepresentationofCEFforJ=4(Leaetal.,1962). .... 28 2-3RepresentationoftheU4+ionsincubicsymmetryundergoingquadrupolarKondoeect. ......................................... 28 2-4MappingofthequadrupolarKondoHamiltonianontothetwo-channelKondomodel. ......................................... 29 2-5S,C,C=T,andversusT=TKofthequadrupolarKondomodel(SacramentoandSchlotmann,1991). ............................... 29 3-1CrystalstructureofPrOs4Sb12. ........................... 41 3-2Fitsof(T)toeither3or1CEFgroundstate,andCttedbyatwo-levelSchottkyanomaly(Baueretal.,2002). ....................... 42 3-3Fitsof(T)toeither3or1CEFgroundstatemodel,calculatedS(T)inboth3and1CEFgroundstatemodels(Tayamaetal.,2003),andthemeasuredS(T)(Aokietal.,2002). ............................... 43 3-4(T),(T)andC(T)ofPrOs4Sb12(Baueretal.,2002). ............. 44 3-5FermisurfaceofPrOs4Sb12(Sugawaraetal.,2002) ................ 45 3-6H-TphasediagramofPrOs4Sb12bydM(T)=dTanddM(H)=dHmeasurements(Tayamaetal.,2003) ........................ 46 3-7C(T)ofPrOs4Sb12(Vollmeretal.,2003;Meassonetal.,2004)andtherealpartoftheacsusceptibility(Meassonetal.,2004)presentingdoubleSCtransition. 46 3-8ThetwosuperconductingphasesofPrOs4Sb12:phaseAandphaseB(Izawaetal.,2003).TheplotoftheSCgapfunctionwithnodesforbothphases(Makietal.,2003). ...................................... 47 3-9TwoSCtransitionsin(T)ofPrOs4Sb12(Oeschleretal.,2003). ......... 47 3-10Tdependenceoftherate1/T1atthe2Qtransitionof123SbforPrOs4Sb12andLaOs4Sb12(Kotegawaetal.,2003). ......................... 48 3-11TunnelingconductancebetweenPrOs4Sb12andanAutip(Suderowetal.,2004).Thegapiswelldevelopedwithnolow-energyexcitations,signofnonodesintheFermisurfacegap. ................................ 48 4-1PictureofPrOs4Sb12largecrystal(about50mg). ................. 60 8


.................................. 60 4-3Schematicviewofthe3HecryostatusedinthemeasurementsperformedatUniversityofFlorida. ................................. 61 4-4SchematicviewofthecalorimeterusedintheSuperconductingMagnet1(SCM1),NationalHighMagneticFieldLaboratory. ................... 62 4-5Thesample-platform/Cu-ringassembly. ....................... 63 4-6SpecicheatCmeasurementprocessusingtherelaxationtimemethod. ..... 64 5-1(T)ofPrOs4Sb12.Thehightemperatureeectivemomentis3.65B,veryclosetotheonecorrespondingtofreePr3+,whichis3.58B. .............. 66 5-2(T)ofthenon-fequivalentLaOs4Sb12. ...................... 66 6-1CofPrOs4Sb12ineldsupto8Tfor~H//(100)(upperpanel),andH-Tphasediagramineldsupto8Tfor~H//(100)(lowerpanel)(Aokietal.,2002). ... 78 6-2CofPrOs4Sb12in10and12TinthevicinityofFIOPtransitionfor~H//(100). 79 6-3CofPrOs4Sb12inmagneticelds13,13.5,and14T,for~H//(100). ...... 80 6-4CofPrOs4Sb12in16,20,and32T,for~H//(100). ................ 81 6-5H-TphasediagramofPrOs4Sb12for~H//(100)(H>8T). ............ 82 6-6ZeemaneectcalculationsforPrOs4Sb12inthe1CEFgroundstatescenario. 83 6-7ZeemaneectforPrOs4Sb12inthe3CEFgroundstatescenario. ........ 84 6-8CofPrOs4Sb12for~H//(110). ............................ 85 6-9CofPrOs4Sb12inH=12T,for~H//(100)(upperpanel),and~H//(110)(lowerpanel). ......................................... 86 6-10H-TphasediagramofPrOs4Sb12for~H//(110)(H>8T). ............ 87 6-11versusT2,andversusTforPrOs4Sb12. ..................... 87 6-12(T)ofPrOs4Sb12in3,10,15,16,17,and18T,between20mKandabout0.9K. 88 6-13versusT2ofPrOs4Sb12for3.5,5.5,7,10,and13T. ............... 89 6-14a(=0+aTn)versusHforPrOs4Sb12eldsupto18T(upperpanel).Theresidualresistivity0(H)(lowerpanel). ........................... 90 6-15A(=0+AT2)versusH. .............................. 91 9


...................................... 92 7-1X-raydiractionpatternsofPr1xLaxOs4Sb12versusLacontentxforx=0,0.1,0.2,0.4,and1. .................................... 113 7-2LatticeconstantaofPr1xLaxOs4Sb12versusLacontentx. ........... 114 7-3(T)ofPr1xLaxOs4Sb12normalizedtoPrmolebetween1.8and10K,measuredin0.5T. ........................................ 114 7-4(T)ofPr0:33La0:67OsSb12versusT.TheCurie-Weisstathightemperature(T>150K)giveseff=3.62B/Prmole. ...................... 115 7-5C=TversusTforthreedierentPrOs4Sb12samplesfromdierentbatches. ... 116 7-6C=TversusT2aboveTcofLaOs4Sb12. ....................... 117 7-7C=TversusTnearTcforPr1xLaxOs4Sb12forx=0,0.05,0.1,and0.2. ..... 118 7-8C=TversusTnearTcofPr1xLaxOs4Sb12forx0.3 ............... 119 7-9TcversusxofPr1xLaxOs4Sb12. ........................... 120 7-10Total(C=T)atTcand0versusxofPr1xLaxOs4Sb12for0x1.0isat1.8KfromFig.7-3. ................................. 120 7-11C=TversusTofPr0:33La0:67Os4Sb12ttedby15Schottky. ......... 121 7-12C=TversusTofPr0:33La0:67Os4Sb12ttedby35Schottky. ......... 121 7-13C=TversusTofPr0:33La0:67Os4Sb12ttedbysinglet-singletSchottky. ...... 122 7-14Cforx=0(10T)and0.02(8and9.5T). ..................... 122 7-15f-electronspecicheatofPr0:9La0:1Os4Sb12inmagneticelds. .......... 123 7-16f-electronspecicheatofPr0:8La0:2Os4Sb12inmagneticelds. .......... 123 7-17f-electronspecicheatofPr0:4La0:6Os4Sb12inmagneticelds. .......... 124 7-18H-TphasediagramfromCmeasurementsforx=0,0.02,0.1,and0.2. ...... 124 7-19(H)ofPr0:95La0:05Os4Sb12atT=20mKfor~H//Iand~H?I(I//(001)). .... 125 7-20(H)ofPr0:95La0:05Os4Sb12for~H?I//(001)atT=20and300mK. ....... 125 7-21(H)ofPr0:95La0:05Os4Sb12for~H?I//(001)atT=20,310,and660mK. .... 126 7-22versusT2ofPr0:7La0:3Os4Sb12in0and0.5T. .................. 127 7-23versusT2ofPr0:7La0:3Os4Sb12in9and13T. .................. 128 10


..................................... 129 7-25(H)ofPr0:7La0:3Os4Sb12when~H//I//(001)at20,310,660,and1100mK. .. 129 7-26(H)ofPr0:33La0:67Os4Sb12at0.35K. ....................... 130 7-27ACsusceptibilityversusT=TcofPr1xLaxOs4Sb12,forx=0,0.05,0.4,0.8,and1. 131 7-28C=TversusTnearTcfortwoPrOs4Sb12samplesfromdierentbatchesinlowmagneticelds. .................................... 132 7-29C=TversusT,nearTc,ofPr0:95La0:05Os4Sb12inlowmagneticelds. ....... 133 7-30C=TversusT,nearTc,ofPr0:9La0:1Os4Sb12insmallmagneticelds. ...... 133 7-31C=TversusT,nearTc,ofPr0:7La0:3Os4Sb12inmagneticelds. .......... 134 7-32dHc2=dTversusx. ................................. 134 7-33p ............................... 135 11


PrOs4Sb12istherstdiscoveredPr-basedheavyfermionmetalandsuperconductor.Ourhighmagneticeldspecicheatmeasurementsprovidedclearevidenceforthenon-magneticsingletcrystallineelectriceld(CEF)groundstate.ThisCEFgroundstateprecludestheconventionalKondoeectastheoriginoftheheavyfermionbehavior.ThesuperconductivityinPrOs4Sb12isunconventional,asinferredfromthedoublesuperconductingtransitioninthespecicheat.Pr1xLaxOs4Sb12(0x1)crystalsweresynthesizedandinvestigatedinordertoprovideadditionalevidencesforapostulatedCEFconguration,todiscriminatebetweendierentconductionelectronmassenhancement(m)mechanismsproposed,andtoprovideinsightintothenatureofthesuperconductivity.Lanthanumdopinginducesanomalouslysmallincreaseofthelatticeconstant.ThespecicheatresultsinhighmagneticeldsindicatedthatCEFschemeisunalteredbetweenx=0andatleast0.2,followedbyanabrupt(butsmall)changesomewherebetweenx=0.2and0.4.MagnetoresistancemeasurementsonLa-dopedsampleswereconsistentwithasingletCEFgroundstateofPr.InvestigationofthespecicheatdiscontinuityatTcandoftheuppercriticaleldslopeatTcindicatedthattheelectroniceectivemass,m,isstronglyreducedwithx,betweenx=0andxcr0.2{0.3,followedbyaweakdependenceonxforx>xcr.Therefore,wehavepostulatedthatsingle-impuritytypemodelscannotaccountfortheheavyfermionbehaviorofPrOs4Sb12.Investigationofthemagneticphasediagramandmagnetoresistanceprovidedstrongcorrelationsbetween 12




Intherareearth(Ce,Yb)-andactinide(U,Np)-basedalloystheelectronicstateshaveanenergyordersofmagnitudesmallerthaninordinarymetals,andsince(k)=~2k2=2m,theeectivemassmisordersofmagnitudelargerthanthefree-electronvalue,hencethetermheavyfermion.Thereareseveralexcellentexperimentalandtheoreticalreviews[ 1 { 5 ]onheavyfermions.OnehallmarkoftheheavyfermioncharacteristhelargeSommerfeldcoecientofthespecicheat.ThespecicheatofmetalsinthenormalstateatlowtemperatureisapproximatedbyC=T+T3,whereTistheelectronicspecicheatandT3isthelattice(Debye)contribution.Foranormalmetalisoforderof1mJ/K2mol,andforheavyfermionisfromseveralhundredtoseveralthousandmJ/K2mol.ThemagneticsusceptibilityathightemperaturesfollowstheCurie-Weissform=C/(T+CW),whereCisaconstant,andCWistheCurie-Weisstemperature.Atlowtemperatures(0)rangesfrom10to100memu/mol.Inthemajorityofheavyfermionmetals,theelectricalresistivityatverylowtemperatureshasaT2dependence:=0+AT2,where0istheresidualresistivityandAisontheorderoftensofcm/K2,muchlargerthanthatofnormalmetals. Thereareabout20heavyfermionsystemsthataresuperconductorsandalmostallofthemareCe-orU-based(thereisonePu-basedheavyfermionsuperconductor:PuCoGa5[ 6 ]). ThelledskutteruditePrOs4Sb12istherstdiscoveredPr-basedheavyfermioncompoundthatisasuperconductor[ 7 ]. Intheconventionalheavyfermions,theonlymicroscopictheoriessomewhatsuccessfulinaccountingfortheeectivemassenhancement(m)asmeasuredbythespecicheataretheS=1/2andS=3/2Kondomodels.ThesemodelswereinitiallyproposedforCe-basedsystems,whoseeectivedegeneraciesoff-electronsincrystallineelectriceldsareeither2or4.TheKondoeectinthesesystemsisanomalousbecauseofstrongspin-orbitcoupling.ThereisonefelectronperCeatomandaccordingtoHund'srules 14


8 9 ]aremoreconsistentwiththef2conguration,allowingforasimilarCEFschemeasthatforPr.Therefore,theinvestigationofPrOs4Sb12withPrhaving2f-electronsmightberelevantandhelptotheunderstandingofthelargeclassofU-basedheavyfermions,sinceCEFcongurationsareusuallyknownforPr.Thenon-magneticcrystallineelectriceldgroundstate(thoughtaseithersingletordoublet[ 7 ])excludestheconventionalKondoeectastheoriginoftheheavy-fermionbehaviorinPrOs4Sb12,whichisconsideredtobethesourceofheavyfermionbehaviorinCe-andU-basedmetals.ThesuperconductivityinPrOs4Sb12isunconventional,butdierentfromthatinCe-andU-basedmaterials.Marksoftheunconventionalityofsuperconductivitycanbeinferredfromthedoublesuperconductingtransitionandpowerlowdependenceofthespecicheatbelowthetransition. Themaingoalsofthisworkare:tosettlethecrystallineelectriceldgroundstateinPrOs4Sb12, 15




whereqjisthechargeatthejthneighboringion,at~Rj.Ifthemagneticionhaschargeqiat~ri,thenthecrystallineelectriceldHamiltonianHCEFis ThesumPiistakenoverelectronsinunlledshells[ 10 ]. TheCEFpotentialcanbeevaluatedintermsofCartesiancoordinatesorintermsofsphericalharmonics.Hutchings[ 10 ]evaluatedthepotential(2{1)forthesimplest3arrangementsofchargesgivingacubiccrystallineelectriceld.Thethreecasesanalyzedwerewhenthechargesareplacedatthecornersofanoctahedron(sixfoldcoordination),atthecornersofacube(eightfoldcoordination),andatthecornersofatetrahedron(fourfoldcoordination).InCartesiancoordinatesthepotential(2{1)canbewrittenas[ 10 ] 5r4]+D6[(x6+y6+z6)+15 4(x2y4+x2z4+y2x4+y2z4+z2x4+z2y4)15 14r6]; wheredisthedistanceofthepointchargeqfromtheoriginineach3cases.C4andD6are70q/(9d5)and224q/(9d7)fortheeightfoldcoordination,+35q/(4d5)and 17


Inthesphericalcoordinatesthesamepotentialiswritten[ 10 ]as 14[Y44(;)+Y44(;)]g+D06fY06(;)r 2[Y46(;)+Y46(;)]g; whereD04andD06are56qp Thereare2generalrulesthatcantellusthenumberofnonzerotermsintheCFpotential.Ifthereisacenterofinversionattheionsitetherewillbenoodd-nterms.Secondly,ifthezaxisisnotanm-foldaxissymmetry,thepotentialwillcontainVmn[ 10 ]. However,calculatingthepotentialtermsinCartesiancoordinatesandeveninsphericalcoordinatesistedious.AmoreconvenientmethodisthesocalledoperatorequivalentorStevens0operatortechnique[ 11 12 ].TheHamiltonian(2{2)isofformHCEF=PijejV(xi;yi;zi).Iff(x;y;z)isaCartesianfunction,inordertondtheequivalentoperatortosuchtermsasPif(xi,yi,zi),thecoordinatesx,y,andzarereplacedbyangularmomentumoperatorsJx,Jy,andJzrespectively,takingintoaccountthenon-commutativityofJ0is.Thisisdonebyreplacingproductsofx,y,andzbycombinationsofJ0isdividedbythetotalnumberofcombinations.Asanexamplewecanconsider whereJ=JxiJy. 18


5r4]+D6[(x6+y6+z6)+15 4(x2y4+x2z4+y2x4+y2z4+z2x4+z2y4)15 14r6]: Usingtheequivalentoperatorrepresentation,theHamiltonianwillbe[ 10 ] or whereB04andB06are+7jejqJhr4i/(18d5)andjejqJhr6i/(9d7)foreightfoldcoordination,7jejqJhr4i/(16d5)and3jejqJhr6i/(64d7)forsixfoldcoordination,and+7jejqJhr4i/(30d5)andjejqJhr6i/(18d7)forfourfoldcoordinationrespectively.Also,hr4iandhr6iarethemeanfourthandsixthpoweroftheradiiofthemagneticelectrons,andthemultiplicityfactorsJ,J,andJareforPr3+(f2)2213/(325112),22/(325112),and2217/(345711213)respectively[ 10 ].Also, 19


13 ].TheHamiltonianiswrittenas InordertocoverallpossiblevalueoftheratiobetweenthefourthandsixthdegreetermsareintroducedthescalefactorWandtheparameterx,proportionaltotheratioofthetwoterms where1

15 ] 1:p 12(j+4i+j4i)+p 6j0i(2{19) 23:r 24(j+4i+j4i)r 24j0ir 2(j+2i+j2i) (2{20) (1)4:a1j4ia2j2i+a2j+2i+a1j+4ib1j3i+b2j1i+b3j1i+b4j3i (2)4:a2j4ia1j2i+a1j+2ia2j+4ib2j3ib1j1ib4j1i+b3j3i Ify=0,theeigenstatesarethoseforOhsymmetry[ 13 ] 1:p 12(j+4i+j4i)+p 6j0i(2{23) 3:r 24(j+4i+j4i)r 24j0ip 2(j+2i+j2i) (2{24) 4:r 8j3ir 8j1ir 2(j+4ij4i) (2{25) 5:r 8j3ir 8j1ir 2(j+2ij2i) (2{26) 21


Theeigenfunctionsandeigenvaluesof1(Th)and23(Th)arethesameasthoseof1(Oh)and3(Oh),thereforearenotaectedbyOt6fromtheHamiltonian.Wheny=0,(1)4(Th)hasthesameeigenfunctionsandeigenvaluesas4(Oh),and(2)4asthosefor5(Oh).Wheny6=0,4and5mixresultingintwo(1;2)(Th)[ 15 ].Therefore,theeigenfunctionsandeigenvaluesofCEFforThandOharedierent.TheOt6terminHamiltonianaectsomeeigenfunctionsandeigenvalues,resultinginachangeofthetransitionprobabilitiesofneutronscatteringinPr3+. 16 17 ].Barnes[ 18 ]foundthatCu2+ionsinthecupratesuperconductorscouldleadtosuchaKondoeectaswell.Later,newevidencebelievedtobehallmarksofaquadrupolarKondoeecthasbeenfoundinthealloysY1xUxPd3[ 19 { 22 ]forx=0.1and0.2. InUBe13,thetotalangularmomentumofU4+(5f2conguration)isJ=4.Thisleads[ 13 ]toa3CEFgroundstateforabouthalfthecrystaleldparameterrange(Fig.2-2).Thef2congurationisexpectedalsoforPr3+inPrOs4Sb12,and,accordingtoCEFcalculationsofLea,Leask,andWolf[ 13 ],a3doubletCEFgroundstateisveryprobable(Fig.2-2).Therefore,theheavyfermionbehaviorinPrOs4Sb12couldbeinprincipleduetoaQKeect.Also,thephysicalpropertiesofUBe13(andU1xThxBe13)arehighly 22


TherelevantstatesforthequadrupolarKondoeectforU4+.Thelasttwocolumnsaretheprojectionsofthemagneticandquadrupolarmomentsrespectively(ReprintedwithpermissionfromCoxandZavadowski[ 23 ]). Cong. StateJEigenstatehJzih3J2zJ(J+1)i 24[j4ij4i]q 12j0i0+8 2[j2i+j2i]08 2q 6j5 2iq 6j3 2i+5 60 2q 6j5 2iq 6j3 2i5 60 2q 6j5 2i+q 6j3 2i+11 6+8 2q 6j5 2i+q 6j3 2i11 6+8 2j1 2i+1 28 2j1 2i1 28 reminiscentofthoseofPrOs4Sb12.Thus,sincethediscoveryoftheHFstateinPrOs4Sb12itsnormalpropertieshavebeenassociatedwiththeQKeect. ThestatesinvolvedinthequadrupolarKondoeectforU4+aregiveninTable2-1.Thedoublydegenerategroundstatecanbetreatedasatwo-levelsystem(amanifoldwithapseudo-spinof1 2).Theprojectedvalueoftheelectricquadrupolemomentontothe3basisisjQzzj=j3J2zJ(J+1)j=8andtheprojectedvalueofthemagneticdipolemomentiszero,i.e.jJzj=0(Table2-1).Therefore,thecouplingisbetweentheelectricquadrupolemomentandtheconductionelectrons. TheAndersonmodelfortherelevantstatesofthequadrupolarKondoeect(sinceitconsidersonly3,7,and8)iscalledthe3-7-8model.Figure2-3showsaschematicrepresentationoftheAndersonmodelrelevantforU4+ionsinthecubicsymmetry.Thegroundstate3(J=4,4f2)andrstexcited7(J=5/2,5f1)mixonlyviatheconductionpartialwaves8(J=5/2,c1).Thetransitionf1!f2isdonebyremovingaconductionelectronandthetransitionf2!f1isdonebyemittingaconductionelectron.Itcan 23


23 ](or,inthegrouptheoryframework,37=8). Applyingacanonicaltransformation(SchrieerandWol[ 24 ])tothe3-7-8model,thehybridizationtermcanbeeliminated.Also,thetransformationyieldstoaneectiveexchangeinteractionbetweenpseudospin-1 2andelectricquadrupolemomentsoftheform where3isapseudospin-1 2matrixforthe3quadrupole,8(8)arethepseudospinsformedfromthe8+2,8+1(82,81)partialwaves(Table2-1).TheexchangeintegralJexchangeisproportionalto=fN(0)andisnegative. TheHamiltonianhasatwo-channelKondoform;twodegeneratespeciesofconductionelectronscouplewithidenticalexchangeintegralsJexchangetothelocal3=1 2object.Thechannelindicesarethemagneticindicesofthelocalconductionpartialwavestates.Figure2-4showsschematicallythemappingofthequadrupolarKondotothetwo-channelKondomodel.Thetwo-channelquadrupolarformoftheHamiltoniantellusthattheconductionelectronorbitalmotioncanscreentheU4+quadrupolemomentequallywellformagneticspin-upandmagneticspin-downelectrons. 25 { 27 ].Thesusceptibility(Fig.2-5,lowerpanel)divergeslogarithmicalatT=0,=(eTH)1ln(H=TH),whereTH=(=e)TK[ 28 29 ].Here,eisthebaseofln,i.e.2.71...InthequadrupolarKondomodelthiscorrespondstoadivergentquadrupolarsusceptibility.ForT!0,thefreeenergyinzeroeldisF=1 2Tln2.Therefore,thezero-temperaturezero-eldentropyisequalto1 2ln2[ 25 ].Thenon-zeroentropyatT!0isconsistentwiththedivergenceinsusceptibilityandarguesinfavorofanon-singletgroundstate(asingletisthegroundstateforthestandardKondomodel).Asexpected,theentropyincreasesmonotonicallywithtemperatureandreachesasymptoticallytheln2 24


2spin)athighT.Also,theT=0entropyincreaseswiththeeld.SincetheS(T=0)decreaseswithHthespecicheatincreaseswithHatintermediateTresultinginlargevaluesof,commonforheavyfermionsystems.AthighTthepseudo-spinisfree,thereforeS=ln2.TheentropychangeS(H)=S(T=1;H)S(T=0;H)increaseswithHfrom1 2ln2toln2forlargeH.IntheC=TplotstheKondopeakscanbeseen. Theinitialmeasurementsofspecicheat[ 16 ]werenotconclusiveforaquadrupolarKondoeectinUBe13.Also,morerecentmeasurementsofnonlinearsusceptibility[ 30 ]areinconsistentwiththequadrupolar(5f2)groundstateoftheuraniumion,indicatingthatthelow-lyingmagneticexcitationsofUBe13arepredominantlydipolarincharacter. 7 ].Laterexperiments[ 31 { 33 ]establishedthecrystallineelectriceld(CEF)groundstateofthePr3+ioninthecubicsymmetryenvironmentofPrOs4Sb12(Thpointgroupsymmetry)asthenon-magneticsinglet1.TheconsequenceofthisisthattheoriginalformulationofthequadrupolarKondoeectcannotbeappliedtotheconductionelectronmassenhancementinPrOs4Sb12. 1isnearlydegeneratewiththe(2)4triplet.Though1itselfdoesn0tcarryanydegreesoffreedom,thepseudo-quadrupletconstitutedby1and(2)4isspeculatedtohavemagneticandquadrupolardegreesoffreedom[ 34 ],andthereforeamagneticorquadrupolarKondoeectisinvokedtoexplaintheenhancementoftheeectivemassofthequasi-particles. Ontheotherhand,themodeldoesnotseemtoberelevantsincethepredictedpropertiesofthequadrupolarKondoeectareindisagreementwiththemeasurements.Butthisisasingle-ionmodel.Possibly,intersiteeectsareresponsibleforthedisagreements.Thereisnolatticequadrupolarmodel. 25


35 ]. Goremychkinetal.[ 32 ]proposedthatthemassenhancementinPrOs4Sb12canbeexplainedbyabalancebetweentwotypesofinteractions,magneticdipolarandquadrupolarbetweenconductionandthefelectronsofPr.ThetheoryofFuldeandJensen[ 36 ]ofconductionelectronmassenhancementascribesthistotheinelasticscatteringbycrystaleldtransitionsinasingletground-statesystem.Themassenhancementoftheconductionelectronsareduetotheirinteractionwiththemagneticexcitations. TherelevantHamiltoniandescribingtheinteractionbetweentheconductionelectronsandtherare-earthlocalizedmomentsis[ 36 ] whereIsfistheexchangeintegral,gListheLandefactor,~Jnisthetotalangularmomentumofarare-earthionatsite~Rn,and~arePaulimatrices. Themassenhancementduetotheinelastictransitionatenergybetweentwolevels,jiiandjji,is wheregJistheLandefactor,Isfistheexchangeintegralcouplingtheconductingelectronstothef-electrons,N(0)istheconductionelectrondensityofstatesattheFermilevel,andhijJjjiisthemagneticdipolematrixelementcalculatedusingthederivedcrystaleldparameters.Thisformulashowsthatforasmallexcitationenergyleadstoalargeenhancementinm. 37 ],Millis[ 38 ], 26


39 ](acompletereviewisgivenbyStewart[ 40 41 ]).Allthesemodelsexhibitdivergenceofthelowtemperaturespecicheat. Byanalogy,inPrOs4Sb12,thequadrupoleuctuationsofPrionsarebelievedtoplayanimportantroleintheHF-SCproperties.Therefore,anothermodelproposed(acollective-typemodel)forthemassenhancementmechanismareduetotheuctuationsoftheantiferroquadrupolarorderparameterduetotheproximitytotheAFQorderedphase. PrOs4Sb12exhibitsanantiferroquadrupolarorderedphaseineldsbetweenabout4.5and14T.Forelds5{13TthetwolowestCEFlevelsaresucientlyclosetoformapseudo-doubletwithquadrupolarandmagneticdegreesoffreedom,resultinginalongrangeorder. Thereisnotheory(tothismoment)thatdescribesthemassenhancementduetotheuctuationsofthequadrupolarorderparameter.OurmagnetoresistivitydataofPrOs4Sb12andLaalloyspresentedinChapters6and7seemtosupportthismassenhancementmechanism. Figure2-1. RotationalsymmetryTh.Intheleft(a),thesmallboldbluesegmentisassimilatewiththedistancebetweentwoantimonyatomsbelongingtothesameicosahedra.Arotationwithrespectto(100)anda2 42 ]). 27


Lea,Leask,andWol0srepresentationofCEFforJ=4(RedrawnwithpermissionfromLeaetal.[ 13 ]). Figure2-3. RepresentationoftheU4+ionsincubicsymmetryundergoingquadrupolarKondoeect.Themodelinvolvesadoubletgroundstateineachofthetwoelectroniclowest-lyingcongurations:f2havingthequadrupolarornon-Kramers3doublet,andf1congurationhavingthemagneticorKramers7doublet.Theconductionelectronsmixthetwocongurationsthroughahybridizationprocess.The8conductionstatecouplesthesetwodoublets(RedrawnwithpermissionfromCoxandZavadowski[ 23 ]). 28


MappingofthequadrupolarKondoHamiltonianontothetwo-channelKondomodel.a)Thestandardtwo-channelKondomodelinspinspace:twoconductionelectronssc+andsccoupleantiparalleltotheimpurityspinSI.b)InthequadrupolarKondocase,thespinisduebythequadrupolarororbitaldeformations.Thetwochannelscomefromtherealmagneticspinoftheconductionelectrons.TheorbitalmotionoftheelectronsproducesthescreeningoftheU4+orbitaluctuations(RedrawnwithpermissionfromCoxandZavadowski[ 23 ]). Figure2-5. 27 ]). 29


43 ](a=9.30311A[ 44 ]aftermorerecentmeasurements),spacegroupIm 45 ].LaOs4Sb12isthenon-fequivalentofPrOs4Sb12withasimilarcrystalstructure.AlltheexoticphenomenaofPrOs4Sb12arethoughttobeassociatedwithitsuniquecrystalstructure.Inparticular,thelargecoordinationnumberofPrionssurroundedby12Sband8Osionsleadstostronghybridizationbetweenthe4fandconductionelectrons[ 46 ].Thisstronghybridizationresultsinarichvarietyofstronglycorrelatedelectrongroundstatesandphenomena. 47 ].Theconsequenceisareductionofthethermalconductivity.Thelledskutteruditeswiththecagearefavorableforathermoelectricdevicepossessingahighcoecientofmerit[ 48 ]. TheamplitudeofthisvibrationofthePrioninPrOs4Sb12isabout8timesbiggerthantheamplitudeofOs.EXAFSdata[ 49 ]supportstheideaofarattlingPrllerion(basedonthelowEinsteintemperatureE75K)withinafairlysticageinthismaterial.Besidesthedynamicmovement,astaticdisplacementwasdetectedinwhich 30


49 ].Gotoetal.[ 50 ],basedonatheoryofCoxetal.[ 23 ],suggestedthatthetunneleectbetweenthetwopositionsofthePrionscouldbelinkedtotheappearanceofthesuperconductivity. 7 ],oreff=3.5BasreportedbyTayamaetal.[ 51 ],andaCurie-WeisstemperatureCW=16K[ 7 ].TheeectivemomentfoundissomewhatlowerthanthemomentofafreeionPr3+whichhaseff=3.58B[ 52 ]. X-ray-absorptionne-structure(XAFS)measurements[ 49 ]carriedoutatthePrLIIIandOsLIIIedgesonPrOs4Sb12suggestthatthePrvalenceisverycloseto3+.EachPrionhastwoelectronsonthefshell(4f2electronicstructure). 13 ]intermsoftheratioofthefourthandsixthordertermsofangularmomentumoperatoroftheCEFpotential,x,andanoverallenergyscalefactorW.Formorethantwodecades,thesymmetrywasthoughtasOh,insteadofTh. Baueretal.[ 7 ]ttedthemagneticsusceptibilitydata(seeFig.3-2)byaCEFmodelinwhichthegroundstatewaschosentobeeitherthenon-magnetic1singlet(W>0)orthenon-magnetic3doublet(W<0).Thepeakpresentinthe(T)datawasthoughttobeproducedwhentherstexcitedstateisatriplet5withaenergy<100Kabove 31


53 ]andBaueretal.[ 54 ] TheseauthorsusetheconventionalcubiccrystaleldmodelwhichisapplicabletotheO,TdandOhsymmetries.IntheThsymmetry,thenon-Kramersdoublet3correspondstothedegenerate2and3singletstates(denotedas23)and4and5statescoincidewith(1)4and(2)4tripletstates,respectively,whenlasttermiszerointhecrystaleldHamiltonian(2-17).Thesingletstate1isthesameforbothcases. Thelasttermofequation(2-17)isuniquetotheThsymmetryofthismaterialcomingfromtheatomiccongurationofSbionsinthecrystal[ 15 ]andisabsentintheconventionalcubiccrystaleldHamiltonianthatMapleetal.[ 53 ]andBaueretal.[ 54 ]used.TheomittingofthelasttermintheHamiltonian(seeequation2-17)hasimplicationsintheinterpretationoftheinelasticneutronscatteringdata. Theabovementionedtreproducestheoverallshapeofthelowtemperaturepeak,andalsothevalueofthevanVleckparamagneticsusceptibilitywithaneectivemomentcloseto,butsomewhatlowerthanthat,ofthefreePr3+ion.Baueretal.[ 7 ]ttedCassumingadegeneratespectrum Specicheatdatawastted[ 7 ]byasystemwithtwolevelsofequaldegeneracysplitbyanenergy=6.6K(ithasbeenassumedthatthedegeneracyofanylevelisliftedbyCEFwhenthelocalsitesymmetryofthePr3+ionsisnotcubicasaresultofsomekindoflocaldistortion). Theentropyinthe3-5casewasfoundtobeS35=Rln27.6J/(molK)[ 7 ].ThetotalentropyofthebroadpeakjustabovethetransitionisS=R(C(T)=T)dT10.3J/(molK).TheclosenessinvaluesmadeBaueretal.[ 7 ]tofavorthe3groundstatescenario. 32


53 ]consideringOhsymmetrysuggestthat3istheCEFgroundstateinPrOs4Sb12.Theresistivitydatameasurementswerealsointerpretedintheframeworkofa3CEFgroundstate[ 55 ]. Incontrast,Tayamaetal.[ 51 ]obtainedasomewhatbettertofthemagneticsusceptibility(T)databya1CEFgroundstatemodel(Fig.3-3(a)).Also,thetheoreticalcurvesofS(T)basedon1groundstatemodelshowincreaseofentropywithelds(lowerpanelofFig.3-3(b)).Thistrendisconrmedbymagneticeldspecicheatmeasurements(Fig.3-3(c))byAokietal.[ 56 ]. Therefore,zeroorsmallmagneticeldsdataarecontradictory,moreexperimentsaretobedoneinordertoestablishthetrueCEFgroundstateinPrOs4Sb12. 3.2.1SpecicHeat 43 ],andthenbyBraunetal.[ 57 ].Itwasin2002whenBaueretal.[ 7 ]discoveredsuperconductivityinPrOs4Sb12.Sincethediscontinuityinspecicheatisoftheorderof,thislargevaluediscontinuity((C=T)jTc=1:85K500mJ/K2mol[ 7 ])impliesthepresenceofheavyfermionsbothinthenormalandsuperconductingstates. Thereisnoconsensusregardingtheprecisevalueof,butallthereportedvaluesimplyheavyfermionbehavior.Actually,thisisperhapsthestrongestevidenceforHFstatesinPrOs4Sb12.Consideringtherelation(C=Tc)=1.43theSommerfeldcoecientisfoundtobe350mJ/K2mol.Thephonon(lattice)contributiontothespecicheatCdatacanbedescribedbyT3thatisidentiedwithspecicheatofLaOs4Sb12withD=304K.isrelatedtoDby=(1944103)n=3D,wherenisthenumberofatomsintheformulaunit(e.g.,n=17inLaOs4Sb12). 44 ]comparativewiththebandsstructure(LDA+Umethod[ 58 ])arepresentedin 33


44 ](whichleaks4felectrons).Thisindicatesthatthe4f2electronsinPrOs4Sb12arewelllocalized.ThesimilartopologyoftheFSforthetwocompoundsissupportedalsobysimilarangulardependenceofthedHvA.ThreeFermisurfacesheets,includingtwoclosed(practicallysphericalshaped)andonemulti-connected,wereidentiedinagreementwiththecalculations. TheeectivemassesmeasuredbydHvAarebetween2.4and7.6m0(m0isthefreeelectronmass).Thesevaluesarewellbelowtheonesreportedfromthespecicheatmeasurements.Theselowvalueshavebeenexplained[ 59 ]intheframeworkofthetwo-bandsuperconductivitymodelinwhichband2correspondstothelightbanddetectedbydHvAmeasurements.Band1isaheavybandhavingmostofthedensityofstates.Theheaviestquasiparticlesareseeninthermodynamicmeasurements(CorHc2)only.However,theapplicabilityofthetwo-bandmodeltoPrOs4Sb12isnotestablished.Furthermore,ourresultspresentedinsection7.6shedssomedoubtsintheinterpretation. 7 ].Thisimplies0116A,vF=1.65106cm/s,andm50m0.ThiscalculationassumesasphericalFermisurface. Theresistivitydatabetween8and40KrevealedaT2dependence0+AT2,withA=0.009cm/K2[ 7 ].TheAcoecientisabouttwoordersofmagnitudesmallerthanthevalueexpectedforaheavyfermioncompound.ConsideringtheKadowaki-Woodsuniversalrelation[ 60 ]betweenAand,A/2=1105cmmol2K2mJ2.Thevalue 34


7 ],andthisisatypicalvaluefornormalmetalsandismuchsmallerthanofLaOs4Sb12. 7 ]asT!0.Thisisthehallmarkofanonmagneticgroundstate.Above150K,(T)ofPrOs4Sb12canbedescribedbyaCurie-Weisslaw.Thereisalargediscrepancybetweenthehightemperatureeectivemomentreportedbyvariousresearchgroups.TheeectivemomentaccordingtoBaueretal.[ 7 ]iseff=2.97B,andeff=3.5BisthevaluereportedbyTayamaetal.[ 51 ]ThefreeionPr3+hasahightemperatureeectivemomentof3.58B[ 52 ].TheCurie-WeisstemperatureisCW=15K[ 51 ]. Fromthediamagneticonset(inset(ii),Fig.3-4(a))itisfoundthatthetemperatureofthesuperconductingtransitionTcisequaltothevaluefoundfromthespecicheatmeasurements. 56 ]ineldsupto8Tandresistivity[ 55 ]inmagneticeldsuptoabout10Trevealedtheexistenceofaeldinducedorderedphase(FIOP)above4.5T.InthisChapteradiscussionofthenatureoftheFIOPwillbepresentedalongwiththespecicheatdatathatcompletesthemagneticphasediagram.Asimilarphasediagramhasbeenobtainedlaterbymagnetization[ 51 61 ](seeFig.3-6)andbythermalexpansionandmagnetostrictionmeasurements[ 62 ]. 35

PAGE 36 7 ]showedasinglesuperconductingtransitionatTcof1.85K.Higherqualitymaterialsrevealedactuallytwosuperconductingtransitions(Vollmeretal.[ 63 ],Mapleetal.[ 53 ],Oeschleretal.[ 64 ]).InFigure3-7panels(a)and(b)areshownspecicheatofPrOs4Sb12presentingtwosuperconductingtransitions,Tc2=1.75KandTc1=1.85KbyVollmeretal.[ 63 ],andTc2=1.716KandTc1=1.887KbyMeassonetal.[ 59 ],respectively.TwosuperconductingtransitionsatthesametemperatureshavebeenreportedbyCichoreketal.[ 65 ]alongwithaspeculationforathirdsuperconductingtransitionat0.6KinferredfromHc1measurements.ItisbelievedthatinclusionsofthefreeOsinthesinglecrystalcannotberesponsiblefortheenhancementofHc1,thoughTcofpureOsis0.66K[ 66 ]basedonsensitiveX-rayandelectronmicroprobestudies[ 65 ]. Therearetwoclassesofexplanationsofthenature(intrinsicornot)ofthedoubletransition.Onearguesinfavoroftwodierentpartsofthesamplewithtwodierentsuperconductingphases,andthereforewithdierentTc's.Thus,thequalityofthesamplesiscrucial.Forinstanceithasbeenconsidered[ 59 ]thatdespitethesharpspecicheattransitions,thesamplesstillpresentspatialinhomogeneities.OnepossibilitywouldbeaninhomogeneouscoexistenceoftwoelectroniccongurationsofPr,4f1and4f2.Thehightemperaturemagneticsusceptibilitymeasurementsareinfavorof4f2,sincetheyhavefound[ 51 ]aneectivemomenteff=3.6B/Pr(theexpectedvaluefor4f1is2.54Bandfor4f2is3.58B). Anotherpossiblescenariothatispresentedinthisdissertationistheexistenceofinhomogeneitiesduetotheclosenessofthesystemtoalongrangeantiferro-quadrupolarorder:clusterswithashort-rangeorderwouldhavedierentsuperconductingparametersthantheremainingpartofthesample. 36


59 ].Thenatureofthetwotransitionsisnotyetestablished.Thewidthofthetransitionasmeasuredbyspecicheatandac-susceptibilityisthesame,about0.2K. ThesuperconductinggapstructureinvestigatedusingthermaltransportmeasurementsinmagneticeldrotatedrelativetothecrystalaxesbyIzawaetal.[ 67 ]providesanotherevidencefortheunconventionalcharacterofsuperconductivityinPrOs4Sb12.Thechangeinthesymmetryofthesuperconductinggapfunctionthatoccursdeepinsidethesuperconductingstategivesaclearindicationofthepresenceoftwodistinctsuperconductingphaseswithtwofoldandfourfoldsymmetries(Fig.3-8).Thegapfunctionsinbothphaseshaveapointnodesingularitywhichisincontrasttothelinenodesingularityobservedinalmostallunconventionalsuperconductors.Thetwo-bandsuperconductivity(similartothatobservedinMgB2)isobservedinnewerthermalconductivitymeasurements[ 68 ]. Adoubletransitioncanbeseeninthethermalexpansion[ 64 ]experiment(Fig.3-9).Thetwotransitionsareatthesametemperaturesatwhichthespecicheatdiscontinuitiesoccur.UsingtheEhrenfestequation@Tc=@P=VmTc=C,whereVmisthemolarvolume,calculationsshowthatthesuperconductingtransitionsTc2isdecreasedtwotimesfasterunderpressurethanTc1.Thisisinfavorofintrinsicnatureofthetwosuperconductingtransitions. 37


Thevaluesreportedbydierentgroups,extractedfromtsofspecicheatbelowTc(ReprintedwithpermissionfromGrubeetal.[ 70 ]). Specicheatdata Grubeetal.[ 70 ] 3.70.2 Vollmeretal.[ 63 ] 2.60:2(3) Fredericketal.[ 71 ] 3.10:2(1) Fredericketal.[ 71 ] 3.60:2(3) Table3-2. Thevaluesreportedbydierentgroupsfrommeasurementsotherthanspecicheat(ReprintedwithpermissionfromGrubeetal.[ 70 ]). Experiment GapFunction Tunnelingspectroscopy[ 72 ] 1.7 Nearlyisotropic 73 ] 2.1 Nearlyisotropic 74 ] 2.6 Pointnodes SbNQR[ 75 ] 2.7 Isotropic Inallreporteddatathespecic-heatmeasurementsexhibitarapiddecreaseofCbelowthesuperconductingtransition.Thispointstopronouncedstrong-couplingsuperconductivity. Thesocalled-model[ 69 ]assumesthatthesuperconductivepropertieswhicharemainlyinuencedbythesizeofthegapandthequasiparticle-stateoccupancycouldbeapproximatedbysimplyusingthetemperaturedependenceoftheweak-couplingBCSgap.ThesizeofthegapintheFermisurfaceisafreelyadjustableparameter=(0)/kBTc,where(0)isFermi-surfaceaveragedgapatT=0.Table3-1presentscomparativevaluesobtainedbydierentgroups.Ananalysisusingthe-modelresultsinanextremelylargegapratioof=(0)/kBTc=3.7andahugespecicheatjumpofC/(c)5[ 70 ]. AsummaryofthepublishedsuperconductivegapratiosandgapanisotropyofPrOs4Sb12fromothermeasurementsthanspecicheatarepresentedinTable3-2. Fredericketal.[ 71 ]succeededinmakingabettertforthespecicheatdataofPrOs4Sb12usingapower-lawfunctionbelowthesuperconductingtemperature.Thets, 38


75 ]inzeroeldshowsaheavyfermionbehaviorandcontroversialconclusionsregardingthenatureofthesuperconductivityinPrOs4Sb12.IntheSCstate,1/T1showsneitheracoherencepeakjustbelowTcnoraT3-likepower-lawbehaviorobservedforanisotropicHFsuperconductorswiththeline-nodegap.Theabsenceofthecoherencepeakin1/T1supportstheideaofunconventionalsuperconductivityinPrOs4Sb12(Fig.3-10).Theisotropicenergygapwithitssize/kB=4.8KseemstoopenupacrossTcbelowT=2.3K.Theverylargeandisotropicenergygap2/kBTc5.2indicatesanewtypeofunconventionalstrong-couplingregime. 76 ]revealaspontaneousappearanceofstaticinternalmagneticeldsbelowthesuperconductingtransitiontemperature,providingunambiguousevidenceforthebreakingoftime-reversalsymmetryinthesuperconductingstate.Thiswillfavorthemultiplesuperconductingphasetransitionsobservedbyspecicheatandthermalconductivitystudiesandsupportthereforetheunconventionalityofsuperconductivity. MagneticpenetrationdepthdatainsinglecrystalsofPrOs4Sb12downto0.1K,withtheaceldappliedalongthea,b,andcdirectionswassuccessfullytted[ 74 ]bythe3HeA-phase-likegapwithmultidomains,eachhavingtwopointnodesalongacubeaxis,andparameter(0)/kBTc=2.6,suggestingthatPrOs4Sb12isastrong-couplingsuperconductorwithtwopointnodesontheFermisurface.Thesemeasurementsconrmedthetwosuperconductingtransitionsat1.75and1.85Kseeninothermeasurements. 39

PAGE 40 75 ](Fig.3-10,fullsymbol),whichisthesignatureofaconventionaltypeofsuperconductivity. Figure3-10(opensymbols)alsoplotsthedatafortheconventionalsuperconductorLaOs4Sb12.Forans-wavecasethatisactuallyseenintheTdependenceof1/T1forLaOs4Sb12withTc0.75K,intheSCstate,1/T1showsthelargecoherencepeakjustbelowTc,followedbyanexponentialdependencewiththegapsizeof2/kBTc3:2atlowT.ThisisaclearevidencethatLaOs4Sb12istheconventionalweak-couplingBCSs-wavesuperconductor. 73 ]anexponentialtemperaturedependenceofthemagneticpenetrationdepth,indicativeofanisotropicornearlyisotropicenergygap,indicatingaconventionalsuperconductivitymechanism. Thisisnotseen,todate,inanyotherHFsuperconductorandisasignatureofisotropicpairingsymmetry(eithers-orp-wave,indistinguishablebythermodynamicorelectrodynamicmeasurements),possiblyrelatedtoanovelnonmagneticquadrupolarKondoHFmechanisminPrOs4Sb12.Also,theestimatedmagneticpenetrationdepth=3440(20)A[ 73 ]wasconsiderablyshorterthaninotherHFsuperconductors. 72 ]intheheavy-fermionsuperconductorPrOs4Sb12demonstratesthatthesuperconductinggapiswelldevelopedoveralargepartoftheFermisurface.Theconductancehasbeensuccessfullyttedbyas-wavesuperconductivity 40


Figure3-1. CrystalstructureofPrOs4Sb12(ReprintedwithpermissionfromAokietal.[ 77 ]) 41


(a)Fitsofthemagneticsusceptibility(T)ofPrOs4Sb12toCEFmodelwitheither3(solidline)or1(dashedline)groundstate.Thesamesymbolsareusedintheinset,whichshows(T)bellow30K.Intheinset,thesolidlinetsaturatesjustabove=0.06cm3/mol.(b)Cttedbyatwo-levelSchottkyanomaly(ReprintedwithpermissionfromBaueretal.[ 7 ]). 42


(a)Fitsofthemagneticsusceptibility(T)ofPrOs4Sb12toCEFmodelwitheither3(dashedline)or1(dotedline)groundstate.Thesolidlanerepresentstheexperimentaldata(takenfromTayamaetal.[ 51 ]).(b)ThecalculatedentropyS(T)for~H//(100)inboth3and1CEFgroundstatemodels(takenfromTayamaetal.[ 51 ]).(c)ThemeasuredentropyS(T)for~H//(100)(ReprintedwithpermissionfromAokietal.[ 56 ]). 43


(a)Resistivity(T)andsusceptibility(T)ofPrOs4Sb12(b)SpecicheatC(T)upto20K[ 7 ](ReprintedwithpermissionfromBaueretal.[ 7 ]). 44


FermisurfaceofPrOs4Sb12(ReprintedwithpermissionfromSugawaraetal.[ 44 ]). 45


51 ]).OpenandclosedsymbolsweredeterminedbythedM(T)=dTanddM(H)=dHdata,respectively.Rightpanel,thePrchargedistributionsinducedintheantiferroquadrupolarorderedphaseinmagneticeld(ReprintedwithpermissionfromMeasson[ 45 ]). Figure3-7. (a)C(T)ofPrOs4Sb12presentingdoublesuperconductingtransition(ReprintedwithpermissionfromVollmeretal.[ 63 ])(b)C(T)ofPrOs4Sb12presentingtwosuperconductingtransitions(ReprintedwithpermissionfromMeassonetal.[ 59 ])(c)TherealpartoftheacsusceptibilityofPrOs4Sb12presentingtwodistinctsuperconductingtransitions(ReprintedwithpermissionfromMeassonetal.[ 59 ]). 46


ThetwosuperconductingphasesforPrOs4Sb12(ReprintedwithpermissionfromIzawaetal.[ 67 ]).ThegapfunctionhasafourfoldsymmetryinAphaseandtwofoldsymmetryinBphase.Right:TheplotofthegapfunctionwithnodesforAphaseandB-phase(ReprintedwithpermissionfromMakietal.[ 78 ]). Figure3-9. TwosuperconductingtransitionsinthethermalexpansioncoecientofPrOs4Sb12.Thetwotransitionsarevisibleforthesametemperaturesofthetwotransitionsinspecicheat(ReprintedwithpermissionfromOeschleretal.[ 64 ]). 47


Temperaturedependenceoftherate1/T1atthe2Qtransitionof123SbforPrOs4Sb12(closedcircles)andLaOs4Sb12(opencircles)(ReprintedwithpermissionfromKotegawaetal.[ 75 ]). Figure3-11. TunnelingconductancebetweenPrOs4Sb12andanAutip.Thegapiswelldevelopedwithnolow-energyexcitations.ThelineingureisthepredictionfromconventionalisotropicBCSs-wavetheoryusing=270eVandT=0.19K(ReprintedwithpermissionfromSuderowetal.[ 72 ]). 48


Thischapterdescribesthesamplesynthesis,characterizationandtheexperimentalproceduresused:dcandacsusceptibilities,resistivity,andspecicheatmeasurements.Abriefdescriptionoftheperformedmeasurementsisgiven. 4.1.1Synthesis 54 57 ].Sincetheuxisoneoftheconstituentelementsofthecompounds(i.e.Sb)themethodiscalledself-uxgrowth.High-puritystartingelements(PrandLafromAMESLaboratory,99.99%puritypowderOsfromColonial,Inc.,and99.999%puritySbingotfromAlfaAESAR)areusedintheproportionR:Os:Sb=1:4:20,whentherare-earthelementRisPrandLainvariousproportions.TheRalloysusedascomponentsintheuxgrowthweresynthesizedeventuallybymeltingitsconstituentelementsinanEdmund-BuhlerArcMelterunderahighpurityargonatmosphere.First,smallchunksofSbwereplacedinsideofaquartztube.AbovethatwereplacedtheOsandtheRcomponentsthatwerepre-meltedseparatelytoeliminateanytraceofoxidefromthesurfaceoftheelements.TheOspowderwaspressedinsmallpelletsandthenmelted.ThequartztubewassealedunderlowpressureAratmosphere(20mTorr)afterthetubeispumpedandushed3to5times.ThetubewiththemixturewasplacedinaLindberg51333programablefurnace(digitalcontrolled,Tmax=1200C)usingthefollowingheattreatmentsequence:temperaturerampingto980Cwitharateof200C/hfollowedbyT=980Cfor24h,thencoolingatarateof3C/hdownto650C.Thelaststepwasafastcoolinginthefurnacetoroomtemperatureata200C/hrate.Thesinglecrystalswerethenremovedfromtheantimonyuxexcessbyetchinginaquaregia(HCl:HNO3=1:1).Thecrystalswerecubicorrectangularupto50mginweight(upto3mminsize)dependingontheamountofthestartingelementsandthecoolingrate.Forinstance,using1gofOsandacooling 49


Thepoly-crystallineRalloys(usedasoneofthestartingcomponentsinthesynthesisofthesinglecrystals)werepreparedbymeltingitsconstituentelementsinanEdmund-BuhlerArcMelterAMunderahalfatmospherehighpurityAr.Theapparatusconsistsofastainlesssteelvacuumchamberwhichsitstightonawatercooledgroovecruciblesinacopperbaseplateandwithanelectrodeatthetop.Thetungstenelectrodeismotordrivenwhichcanbemovedfreelyabovethecrucible.Themeltingprocesscanbeobservedthroughadarkglasswindow.Allimportantcontrolfunctionsareintegratedintheheadoftheelectrodeandensuresafeandconvenientoperation.Whenfedatthemaximumcurrentthetemperatureoftheelectricarcinthemeltercangoashighas4000Candmelts500gofmetals.Thearcmelterhasaipper,amanipulatorforturningthesamplesinsitu.Thisgivesthepossibilitytoipandagainmeltthesample,ensuringitshomogeneity,withoutopeningthechamber.Beforeoperating,thecopperbaseplatewasthoroughlycleanedwithacetonetoavoidanycontaminationofthesamplewithimpurities.Rightatthebeginning,eachoftheconstituentelementswerewellcleanedtoeliminatetheoxidelayeronthesurface.Theprecisioninmassmeasurementswas0.02mg.Startingwiththeradioactiveorthehardestelementwecanadjusttherelativemassesoftheothercomponentstogainthewantedstoichiometricratio.Thetotalmasswasfromteensofmilligramto1g,thesizeofthesamplebeadwasupto1cm.Rightbeforetheelementsweremeltedtogether,azirconiumbuttonwhichwasalsousedforignitionofthearc,wasmeltedjusttoensureaevenhigherpurityoftheAr,whichwaslteredthroughapurierbeforeenteringintothearcchamber.Zirconiumiswellknownasaoxygenabsorber.Theelementwiththehighestvaporpressurewasthenplacedonthecopperplaterightbelowtheelementswithlowervaporpressures.Theaimofthis 50




whereishalfofthereectionangle,nisaninteger(n=1fortherstorderspectrum),distheinter-planedistance,andisthewavelengthoftheincidentradiation.ThelatticeconstantsarethencalculatedfromdandtheintersectionpointsofthelatticeplanesfromthedesiredspacegroupnumberisgivenintermsoftheMillerindices(hkl).ForacubicsymmetrythesameBraggequationcanberewrittenas sin2=2 whichisderivedfromd(hkl)=a=p 4.2.1Equipment Inhouse(StewartLab.,PhysicsDepartmentatUniversityofFlorida)measurementsofspecicheatwereperformedinthetemperaturerangeof0.3to2Kusually,andinsomecasesupto10K.Ahomemade3Hecryostatwasused.TheschematicdrawingisgiveninFig.4-3.Thisprobewasusedforthemeasurementsofspecicheatinmagneticeldsaswell.AspeciallydesigneddewarfromCryogenicConsultantsLimitedwasalsoused.Thesuperconductingmagnetreached14Tat4.2Kbathtemperature.Fortheac 52


SpecicheatmeasurementsatlowertemperatureandhighermagneticeldswereperformedattheMillikelvinFacility(SuperconductingMagnet1{SCM1),HighMagneticFieldNationalLaboratory,Tallahassee,Floridausingatoploadingdilutionrefrigeratorwhichispermanentlyinstalledina18/20Tsuperconductingmagnet.Themeasurementtemperaturerangewas20mKto2Kcombinedwithamagneticeldofupto20T.Thesmallhomemadecalorimeter(Fig.4-5)wasconnectedtothegeneralpurposesamplemountprovidedbythefacility.Resistivitymeasurementsperformedatthesamefacilityweredoneinthetemperaturerangeof20mKto0.9K.Anothersampleholder,aso-called16pinamplerotator,wasused.Thisallowsthechangeoforientationofthesampleineldduringtheexperiment.Thisholderhas16pins(16connectionwirestothetopoftheprobe)thatallowsuptoamaximumoffourdierentsamplestobemeasuredwithoutpullingouttheprobefromthedilutionrefrigerator,savingprecioustime.Takesupto6hourstoinserttheprobeintotherefrigeratorandcoolthesampleto20mK.Specicheatmeasurementsinmagneticeldsupto32Twereperformedatthe33T,32mmboreresistivemagnet(Cell9),atthesameNationalLaboratory.Anotherhomemade3Heprobesimilartotheonementionedearlierbutwithslightlydierentdimensionsinordertotintothemagnetandalsotoaccommodatethesampleinthemaximumeldstrengthregionwasused.Rightbeforetheinsertionoftheprobeintothemagnetanelectricalcheckwasdoneonwireconnections.ThequalityofvacuumandsealingwascheckedalsousinganAlcatelASM10LeakDetector.Forboth33Tand45Tmeasurements,aspecialpositioningsystemmadeitpossibletocentertheprobeinsidethemagnetsuchthatitdid 53


Becauseofthelargeamountofheatthathadtoberemoved,theprobewascooledinliquidnitrogen(LN)downtotheboilingpoint(77.35K).Afterabout2hours,whentheprobewasatthermalequilibriumwiththeliquidnitrogen,itwastransferredquicklyintoadewarinwhichittstight.ThedewarwascooledinadvanceinLNaswell.Thedewar(withtheprobeinside)waslledwithliquid4He(LHe)andafterseveralhours(dependingonthevolumeofthecan)thetemperatureoftheprobereached4.2K.After4.2Kwasattainedfollowingtheproceduredescribedabove,the4HepotwaslledwithLHefromthebathbyopeningtheneedlevalve,and3Hegas(alighterisotopeofHe)wastransferredintothe3Hepot+probelineusingahome-made3Hehandlingsystem.Thehandlingsystemconsistsofatanklledwith3He,apumpwhichhelpstotransfertoandbackfromthe3Hepotline,andpressuregaugestodisplaytheamountof3Heleftinthetankandinthetransferlines.Afterclosingtheneedlevalveandpumpinginthe4Helineatemperaturebetween1and2Kwasobtained.Itwasnecessarytorellthe4Hepotbyopeningtheneedlevalveonceinseveralhours.Inordertoattain0.3Kacompletelycontained3Hecoolingpartusingasorptionpumpwasrequired.Whencooled,gasesgenerallyadsorbtosolidsurfaces.Thesorptionpumpisbasedontheideathatat10Kalmostallofthe3Hegasmoleculesareadsorbed,whereasat35Kallofthemoleculesdesorb.ThesorptionpumpconsistsofaCucylinderthatcontainsactivatedcharcoal,whichhasanenormoussurfacearea(tensofsquaremeterspergram).Thecylinderisattachedtothelowerendofametallicrod.Thewholesystem,rod+cylinderwithcharcoal,wasplacedinsidethe3He-gasenclosure.Asthecharcoalwasloweredtowardthe3Hepot,the3Hewasabsorbedbythecharcoalreducingthevaporpressureandloweringthetemperatureofthe3Hepot.Afterthecharcoalbecamesaturatedwith3He,thecharcoalwaswarmedup(byraisingtherodwiththecharcoal),andthegaswas 54


79 { 81 ].Thethermalrelaxationmethodconsistsofmeasuringthetimeconstantofthetemperaturedecayofthesampleconnectedtotheheatbathbyasmallthermallink.ApowerPisapplied(Fig.4-6)(thermalpowerbyasmallcurrentoftheorderofA)totheplatform-samplesystem.Thetemperatureofthesample,initially 55


Thetimeconstant1isproportionaltotheCtotal(sample+platform): whereisthethermalconductanceofthewireslinkingthesample+platformatT=T0+TandtheringatT=T0.Theblocktemperatureisregulatedbyablockheater(abundleofmanganinwire)andmeasuredbyathermometerattachedtotheblock.Thetimeconstantisobtainedbymeasuringthetimedecayoftheo-balancevoltagesignalfromaWheatstonebridgeusingalock-inamplier.TwoarmsoftheWheatstonebridgeareavariableresistanceboxandtheplatformthermometer.Thebridgeisbalancedbyadjustingtheresistanceoftheresistancebox.Thismadeitpossibletondtheresistanceofthethermometer.FromaninitialcalibrationofthethermometerRversusT: 1 itispossibletondthetemperaturecorrespondingtotheplatformthermometerresistance.Thethermalconductanceisgivenby: whereP=IVisthepowerappliedtotheplatformheater.Equation(4{3)isvalidifthethermalcontactbetweensampleandplatformisideal(i.e.,sample1).Ifthecontactispoor(i.e.,sample),then 56




82 ]andtwosecondarycoilsmadefromcopperwire,woundinbothsidesinoppositedirectionsof2700turns.ThecoilsareattachedtotheCublock(whichisinthermalcontactwiththe3Hepot).Theapparatususesthemutualinductanceprinciple.Thesampleissubjecttoanalternatingmagneticeldof0.1Oeproducedbytheprimarycoil(andalsotheEarth'smagneticeld).Theresultingelectromotiveforce(EMF)inducedinthesecondarycoilisdetected.Thebackgroundsignalisnulledbytheidenticalsecondarycoil,connectedinseriesopposition.Forthesamereasonthetwoscrewsareidenticallybuilt.ThesampleisgluedtoonescrewwithGeneralElectric(GE)varnish7031whichensuresagoodthermalandmechanicalcontactatlowtemperatureandalsocanberemovedeasilyusingacetone.Theacsusceptibilitymeasurementswereperformedattwodierentfrequencies:27Hertz(Hz)and273Hz.Itwasdeliberatelyusedthesefrequencies(notintegermultiplesof60Hz)inordertoavoidthenoisecomingfromthecommonelectricaloutlet.Ingeneral,B=0(H+MV)=0H(1+),withHthemagneticeld,MVthevolumemagnetizationand=MV/Histhemagneticsusceptibility.IftheappliedeldHhasasinusoidalform,thetimedependentmagnetizationMV(t)canbeexpressedasaFourierseriesofthenon-linearcomplexacsusceptibility.ApplyingtheinverseFouriertransformtoMV(t)itcanbefoundthenthharmonicofbothrealandimaginaryacsusceptibility.Thefundamentalrealcomponent 58




PrOs4Sb12largecrystal,about50mg(right).Intheleft,anOsballwithPrOs4Sb12singlecrystalsattached,waitingtobeetchedout. Figure4-2. PrOs4Sb12samplespreparedfor(leftpanel)resistivityand(rightpanel)specicheatmeasurements. 60


Schematicviewofthe3HecryostatusedinthemeasurementsperformedatUniversityofFlorida. 61


SchematicviewofthecalorimeterusedintheSuperconductingMagnet1(SCM1),NationalHighMagneticFieldLaboratory. 62


(a)Topviewofthesample-platform/Cu-ringassembly.(b)Lateralviewofthesapphireplatformandsample. 63


SpecicheatCmeasurementprocessusingtherelaxationtimemethod(RedrawnwithpermissionfromMixson[ 83 ]). 64


Allsamplesweresynthesizedusingtheself-uxgrowthmethod,describedinChapter4.Thesamplesarecubicshapedandofsizesrangingfrom1/2mmto3mmandweightingfrom1mgtoabout50mg.X-raydiractionwasperformedtoverifythedesiredcrystalstructure.Fromthediractionpatternitwasalsopossibletodeterminethelatticeconstants.Inadditiontothis,theX-raysconrmedthatthesamplesweresingle-phasewithinanaccuracyof5%. Thequalityofthesampleisalsogivenbythesharpnessofthetransitioninthespecicheat.AmorequantitativelymeasureofthequalityofthesampleistheresidualresistivityratioRRR=(300K)/(T!0).Thisratiorangesfrom50toabout170(PrOs4Sb12samplesstudiedbyMeassonetal.[ 59 ]haveRRR40.) Duetotheverysmallsizeofthesamplesused,thesusceptibilitiesmeasuredforallconcentrationsandthebackground(susceptibilityofthesampleholderconsistingfromaplasticdrinkingstraw)werecomparableat10K.Atroomtemperaturethesusceptibilitywasevensmallerthanthebackground,especiallyfordiluteconcentrations.Inordertoavoidthisbackgroundcontribution,magneticsusceptibilitywereremeasured(forx=0,0.05,0.3,0.67,0.8and0.95)usingbiggersamples.Also,inthesemeasurements,thematerialwaspressedinbetweentwolongconcentrictubessuchthatnobackgroundsubtractionwasneeded. AlltheseadditionalmeasurementsyieldedtoaCurie-Weisstemperaturedependenceabove150K,correspondingtoaneectivemagneticmomentclosetotheoneexpectedforPr3+(Fig.5-1),muchclosertotheexpectedvalueforPr3+thantheinitiallyreportedeff=2.97B[ 7 ]forPrOs4Sb12.TheeectivemomentofthefreePr3+iseff=3.58B[ 52 ].NewmeasurementsbyTayamaetal.[ 51 ]revealedaneectivemomentclosetothisvalue.ThissupportsthenotionofanessentiallytrivalentstateofPrinallPr1xLaxOs4Sb12alloys. 65


Figure5-2. 66


7 ](either3or1CEFgroundstate)forPrOs4Sb12implynon-magneticgroundstatesandexcludeaconventionalKondoeect,believedtobethesourceofHFbehaviorinCe-andsomeU-basedmetals. Thecontroversybetweenthetwoschemeswasbroughtaboutbydierentexperimentsthatseemtofavoreitherconguration. AspresentedinChapter3,therstpublishedresultssuchasthezeroeldspecicheat,magneticsusceptibilitydata[ 7 ],resistivityinsmallmagneticelds[ 55 ],inelasticneutronscatteringdatainterpretedusingOhsymmetry[ 53 ]favoredthe3doubletastheCEFgroundstate. Ontheotherhand,magneticsusceptibilitydataofTayamaetal.[ 51 ]andentropychangesinsmallmagneticeldsmeasuredbyAokietal.[ 56 ]werebetterttedbya1CEFgroundstatemodel. ThezeroeldSchottkyanomalyoccurringat3.1Kcanberelatedtothe3-5model,assumingthesetwolevelsaresplitby6.5K,or1-5modelwiththesplittingof8.4K.Thedicultyininterpretingtheselowtemperature,loweldresultsisrelatedtoastronghybridizationof4fandconductionelectrons,inferredfromthelargeelectronic 67


Inordertopresentourresultsinaproperperspectivewestartfromrecallingthespecicheatdataforeldssmallerthan8TobtainedbyAokietal.[ 56 ] Figure6-1,upperpanel,showsthelowtemperaturespecicheatto8TobtainedbyAokietal.[ 56 ],thelowerpanelacomprehensivephasediagramknownbeforeourmeasurements.4.5TisthelowesteldatwhichasignatureofFIOPisdetectableasasmallkink(at0.7K).Thiskinkevolvesintoasharppeakat0.98Kin6T.TheC(T)peakgrowsandmovesalsotohighertemperaturesforhigherelds. TheFIOPwasconrmedbyspecicheatofVollmeretal.[ 63 ]andmagnetizationstudyofTayamaetal.[ 51 ]. AnumberofobservationsbroughtforwardtheinterpretationofFIOPintermsofantiferroquadrupolar(AFQ)order.Theseobservationsincludedalargeanomalyinthespecicheat(correspondingtoalargeentropyremovedbythetransition)andtheverysmallvalueoftheordered(antiferromagnetic)moment(about0.025Bat0.25Kin8T[ 84 ])measuredbyneutrondiraction,andalsosimilaritiestosystemsdisplayingquadrupolarorder(e.g.,PrPb3[ 85 ]). Figures6-2,6-3,and6-4showthespecicheatineldsrangingfrom10to32T.Thespecicheatmeasurementsineldsupto14TweredoneusingCryogenicConsultantLimitedsuperconductingmagnetattheUniversityofFlorida.Measurementsineldslargerthan14TwerecarriedoutattheNationalHighMagneticFieldLaboratory,Tallahassee,FloridausingaresistiveBittermagnet.Theeldwasappliedalongthecrystallographic(100)direction. Thespecicheatdatainallthreeguresareaftersubtractingthephononbackground(T3with=(1944103)n/3D[ 1 ])correspondingtoaDebyetemperature(D)of165K,proposedbyVollmeretal.[ 63 ]ThisvalueofDobtainedfromthetemperature 68


7 ]),320K(Aokietal.[ 56 ]),and259K(Mapleetal.[ 53 ])arebasedonspecicheatmeasurementsofLaOs4Sb12. Thelowesttemperatureoftheheatcapacitymeasurements,actualvalue,ischosenrelativelyhighinordertoavoidcomplicationsassociatedwithanuclearcontributionofPr.Thiscontributionisstronglyenhancedbycouplingwithorbitalmomentsoffelectrons[ 86 87 ].Itisdiculttomeasurespecicheatbyaconventionalrelaxationmethodattemperatureswherenucleardegreesoffreedomdominatebecauseofadditionalthetimescaleenteringtheexperiment,nuclearspin-latticerelaxationtimeT1[ 88 ].Stronglynon-exponentialtemperaturedecaysatthelowesttemperatures(e.g.,below0.5Kintheeldof10Tandbellow1.5Kintheeldof32T)indicatetheimportanceofnucleardegreesoffreedomandcannotbeanalyzedusingtheso-called2correction.Therefore,theselowesttemperaturepointscarrylargeuncertainty.Whenthemagneticeldappliedalongthe(100)crystallographicdirectionis10T,thetemperatureofthesharpFIOPpeakappearsat1K(Fig.6-2).Whenincreasingtheeldfrom10Teldto12and13T(Figs.6-2and6-3)theorderingtemperatureTxdecreasesonlyslightlybutC(Tx)issuppressedinastrongmanner. Theresultspresentedhere[ 31 ]combinedwiththoseofAokietal.[ 56 ]andVollmeretal.[ 63 ]showthatTx(peakpositioninC)reachesamaximumvaluearound9T.Also,CatTxismaximumsomewherebetween8and10T. In13TashoulderappearsonthehightemperaturesideoftheFIOPanomaly.Thespecicheatvalueatthisshoulderisabout3400mJ/Kmol.ThisshoulderevolvesintoabroadmaximumforH=13.5T.Above13.5TtheFIOPcannotbeobservedanymoreinthespecicheat.Thus,theseresultsstronglyimplythedisappearanceofFIOPbeforeTxreaches0. Thebroadmaximumthatappearsin13Texistsatalleldsstudieduptoatleast32T.Thetemperatureofthemaximumincreaseswiththestrengthoftheeld(Fig.6-4). 69


89 ].Theuncertaintyofthespecicheatmeasurementsintheseelds(andattemperatureswherenuclearcontributionissmall)isabout10%.IncreasingDfrom165K,usedinthesubtractionofthephononterm,totheotherextremalvalueproposed,320KwouldraisetheestimateoftheelectronicpartofCbyabout290mJ/Kmolat3.5K.Thus,theextractedvaluesatthemaximumarewellwithintherealisticerrorbarofthetheoretical3650mJ/Kmolforthetwo-levelSchottkyanomaly.Thehighesteldusedof32Tislargeenoughtosplitanydegeneratelevels,thereforetheobservedSchottkyanomalyisduetotheexcitationsbetweentwosinglets.TmisrelatedtotheenergyseparationofthetwolevelsbyTm=0.417[ 89 ].AnextrapolationofTmtoT=0(Fig.6-5)determinestheeldatwhichthetwolevelscross,whichissomewherebetween8and9T. TheseresultcanbeusedtoinfernewinformationregardingtheplausiblecrystaleldcongurationofPr.Prcanbemodeledbythefollowingsingle-sitemean-eldHamiltonian[ 84 ]: whereHCEF,JandOirepresenttheCEFHamiltonianforthecubicThsymmetry,thetotalangularmomentum,andthei-thquadrupolemomentofPrinasublattice,respectively,wheretherearevetypesofquadrupolarmomentoperators:O02,O22,Oxy,Oyz,andOzx.JandQiaretheinter-sublatticemoleculareldcouplingconstantsofspin(exchange)andquadrupolarinteractions,respectively.ThethermalaveragesoftheangularmomentumandquadrupolemomentofthePrinthecounterpartsublatticearehJ0iandhO0ii. UsingtheCEFparametersproposedbyKohgi[ 84 ]forthe1-5CEFconguration,Tm(withQi=0)andtheOyz-typequadrupolarorderingtemperatureTxwerecalculatedfor(100)directionbyAokietal.[ 90 ]AsitisdemonstratedinFig.6-5,themeasured 70


However,thelevelcrossingforeld(100)directionisalsoexpectedfor3-4model,althoughatsomewhatdierenteld,asdemonstratedbyVollmeretal.[ 63 ]. MoreconclusiveargumentsregardingtheCEFcongurationcanbeobtainedfromthestudyoftheanisotropyoftheZeemaneect.ResultsofourcalculationsfortheZeemaneectfor~H==(100),~H==(110),and~H==(111)areshowninFig.6-6for1CEFgroundstate.TheplotsshowonlythefourlowestCEFlevels.Thehigherlevelsareatabove100Kand200Kfromthegroundstate,andthereforeplaynoroleinthelowtemperatureproperties.ThecalculationsweredoneneglectingexchangeandquadrupolarinteractionsandconsideringtheThsymmetry.Neglectingorretainingthelasttwotermsin(6{1)forthe(100)directionleadtoalmostidenticalresultsforeigenvalues(Aokietal.[ 90 ]andourresults). Thereisacrossingbetween1andthelowest5level(splitbymagneticeld)atabout9Twhen~H==(100)or~H==(111)andanti-crossingwhen~H==(110)aroundthesameeld.Therefore,thecrossingeld,extrapolatedfromthetemperatureoftheSchottkyanomalyathigheldsshouldbeindependentoftheelddirection. Figure6-7showsthesamecalculationsforthe3CEFgroundstatemodel.For~H==(100)thereisacrossingbetweenthetwolowestCEFlevels,althoughataeldsomewhatlargerthantheoneexpectedforthe1CEFgroundstate.However,thereisnocrossingexpectedinvolvingthelowestCEFlevelswhentheeldisappliedalongthe(110)or(111)directioninthe3-5model(Fig.6-7).Therefore,measurementsofspecicheatwhenmagneticeldisappliedinanydirectiondierentthan(100)dierentiatebetweenthetwoscenarios.Measurementsofthespecicheatineldsto14 71


TheH-TphasediagramispresentedinFigure6-10.For~H//(110)directionweobserveadecreaseofTxvalueswithrespecttothe(100)directionforthecorrespondingelds,consistentwiththepreviousmagnetizationmeasurements[ 51 ](Fig.3-6).Ontheotherhand,withintheuncertaintyofthemeasurement,thereisnochangeinthepositionoftheSchottkyanomalyat13and14T,asexpectedforthe1CEFgroundstateandinconsistentwiththe3scenario.Moreover,forthe(110)orientationtheSchottkyanomalycanbeclearlyseenalreadyat12T.ThislowereldlimitfortheSchottkymaximumisprobablyduetocompetitionbetweenthetwotypesofanomaliesandlowervaluesofTxforthe(110)direction(Fig.6-9). AstraightlinetforthethreeTmpointsresultsinthecrossingeldvalueof91T.Thisvalueagrees,withintheerrorbar,withtheestimateforthe(100)direction.Theexistenceofthecrossingeldforthe(110)directionprovidesanunambiguousevidenceforthe1-5model.Asmallmisalignmentofthesamplewithrespecttotheeldineitherofthemeasurementscannotexplainessentiallyidenticalcrossingeldsforbothdirections.Infact,themeasureddierenceinTxvaluesfor(100)and(110)directionsprovidesanadditionalcheckofthealignment.Similartothe(100)direction,thereseemstobeaclosecorrelationbetweenthecrossingeldandtheeldcorrespondingtoTxmaximum. Figures6-5and6-10implyastrongcompetitionbetweentheeld-inducedorderandtheSchottkypeak.TheFIOPtransitioninthespecicheatabruptlydisappearsbefore 72


44 51 ],ontheotherhand,wereabletomapTxasafunctionofthemagneticeldallthewaytoTx0.ThisapparentcontradictioncanbeexplainedbyaverysmallentropyavailablefortheFIOPtransitionabove13and12Tforeldsparalleltothe(100)and(110)directions,respectively.Specicheat,beingabulkmeasurement,canbelesssensitivethanmagnetizationtechniquesinthissituation.Astrongcompetitionistobeexpectedinthe1-5scenario.Thegroundstatepseudo-doubletformedatthelevelcrossingcarriesbothmagneticandquadrupolarmoments.Sinceaquadrupolarmomentoperatordoesnotcommutewithadipolarone,thequadrupolarinteractionsleadingtoFIOPcompetewiththemagneticZeemaneect. Therefore,thehighmagneticeldsmeasurementsofspecicheat[ 31 ]providedtherstunambiguousevidenceforthesingletCEFgroundstateofPrinPrOs4Sb12.Thisresultwasconrmedbyrecentinelasticneutronscatteringexperiments[ 32 ]analyzedintheThsymmetry,andourmagnetoresistivityresultsdescribedinSections6.2and7.2. ThemainindicationofheavyelectronsinPrOs4Sb12isthelargediscontinuityinC=TatTc.Themassenhancementinferredfromspecicheatmeasurementsisoftheorderof50[ 7 ].Thisvalueisanestimateandthereisnoconsensusonaprecisevalue.Anuncertaintyexistsinevaluationoftheeectivemassdirectlyfromthelowtemperaturezero-eldspecicheat,becausethereisnostraightforwardmethodofaccountingfortheCEFspecicheat.ThecorrespondingSchottkyanomalyisstronglymodiedbecauseofthehybridizationbetweenthefandconductionelectrons.Thezero-eldspecicheatjustaboveTcisdominatedbyCEFeects. 73


32 ]suggestedmenhancementtobeabout20.However,theirestimatewasbasedontheFulde-Jensenmodel,whichwedonotbelieveisrelevanttoPrOs4Sb12.Thisenhancementis3{7,accordingtothedeHaas-vanAlphenmeasurements[ 44 ].However,dHvAeectwasanalyzedoverawiderangeofelds3{17TanddidnottakeintoaccountmbeingdependentofH[ 44 ]. TheresidualresistivityratioRRR=(300K)/(T!0)oftheinvestigatedsamplewasabout150.Thisvalueisamongthehighestreported,implyinghighqualityofoursample.Boththecurrentandthemagneticeldwereparalleltothe(100)direction(longitudinalmagnetoresistance).Themeasurementsweredoneusingthe18T/20TsuperconductingmagnetattheMillikelvinFacility,NationalHighMagneticFieldLaboratory,Tallahassee,Florida.Thetemperaturerangewas20mKto0.9K,themaximumeldused20T.MeasurementsattheUniversityofFloridaweredoneineldsupto14Tdownto0.35K. Thezero-eldelectricalresistivity,anotherimportantcharacteristicsofheavyfermionmetals,doesnotprovideastraightforwardsupportforthepresenceofheavyelectrons.Mapleetal.[ 91 ]foundthattheresistivity,between8and40K,followsafermi-liquidtemperaturedependence(=0+AT2).Ourresistivitydatabetween8and16Kfollowstheabovementioneddependence(Fig.6-11)withA0.009cm/K2(inagreementwithAfoundbyMapleetal.[ 91 ]).AsinferredfromKadowaki-Woods(KW)relation(A/21105cm(molK/mJ)2)[ 60 ]thisvalueofAimpliesasmallelectronicspecic-heatcoecient30mJ/K2mol,comparabletotheonemeasuredforLaOs4Sb12.So,evidentlythereisanuppertemperaturelimit(lessthan8K)fortheheavyfermionbehavior. Figure6-12showstheresistivityofPrOs4Sb12inH=3,10,15,16,17and18Tinatemperaturerangeof20mKto0.9K.Theresistivitybelow200mKsaturatesforallelds.Thistemperaturedependenceatthelowesttemperaturewasalsoobservedbyothergroups[ 91 92 ].Therefore,theresistivityforallotherintermediateeldswasmeasuredto 74


Mapleetal.[ 91 ]proposedthefollowingtemperaturedependenceforxedmagneticeld:=0+aTn,withn>2.Intheirstudy(transversalmagnetoresistivity)nwas3for3Tand2.6for8T.Inourlongitudinalcasetheseexponentsareslightlylarger(e.g.,3.9for3T).Theexponentsdependonthetemperaturerangeofthet,i.e.,nbecomessmallerwhentheuppertemperaturelimitofthetdecreases.Theresidualresistivity0valuesresultedfromthetondierenttemperatureranges(includedinthe350mKand0.9Kinterval)wereclosetoat20mK.Theresidualresistivity0attainsamaximumnearH=10T,eldcorrespondingtothecrossingbetweenthetwolowest-energyCEFlevelsofPr(Fig.6-14,lowerpanel).Inthisregion(around9{10T)thelowesttwosingletsformaquasidoubletpossessingquadrupolardegreeoffreedom.Theseelectricquadrupolesorderatsucientlylowtemperatures[ 56 ]withtheorderingtemperaturehavingmaximuminthecrossingeld[ 31 56 ].ResistivityisdominatedthereforebytheCEFeectsorthequadrupolarordering.Thisorderingiscompletelysuppressedbyeldshigherthan15T.AsitcanbenoticedfromFigs.6-12and6-14theresidualresistivity0doesnotchangesubstantiallyineldshigherthan15T.Infact,itcanbeseenthantheresidualresistivity0versusHeldcanmaptheboundaryoftheAFQphase,i.e.asharpincreaseof0indeedcoincideswiththeAFQboundary,indicatedbyarrowsinFig.6-14,lowerpanel. Thesameconclusioncanbedrawnfromresistivitymeasurementsforhighmagneticeldsperpendiculartothecurrent[ 55 ].Therateoftheincreaseoftheresistivitywithtemperatureisstillchangingabove15T(Fig.6-12).ItcanbeconcludedthatthereductionofthetemperatureratecorrelateswithanincreaseoftheenergybetweenthelowestCEFlevels.Apreciseaccountingofthesechangesisdicultsinceneitherofthefunctionscheckedoutdescribeaccuratelythevariation(T)inaxedeld. AlineardependenceofonT2isaccountedbyresistivity(=0+AT2)indierenttemperatureranges(above0.4and0.5K),asseeninFigs.6-13.UsingtheKWratio[ 60 ] 75


Acharacteristicelddependenceoftheresidualresistivity(Fig.6-14,lowpanel)wasassociatedbothtoCEFeectsandlongrangeAFQorder.TheCEFeectontheresistivitywasconsideredbyFrederickandMaple[ 93 ]usingthefollowingexpression: Thersttermrepresentsacontributionduetoexchangescattering,andthesecondtermisthecontributionduetoaspherical(orquadrupolar)scattering.TheasphericalCoulombscatteringisduetothequadrupolarchargedistributionofthePr3+.MatricesPij,QMijandQAijaredenedasfollows: 1e(EiEj);(6{3) 2jhijJ+jjij2+1 2jhijJjjij2;(6{4) IntheaboverelationsEiaretheeigenvalueoftheCEFeigenstates,thejii0saretheCEFeigenstates,=1/(kBT),andtheym20saretheoperatorequivalentsofthesphericalharmonicsforL=2(i.e.,quadrupolarterms)[ 94 ].TheQij-matricesarenormalizedtoeachother[ 95 ]suchthat 76


FurthermoreAseemstohaveamaximumvalueneartheeldseparatingorderedandnon-orderedphases.NotethatthisisnotthecrossingeldforthelowestCEFlevels.Thus,theseresultssuggestapossibilityofmenhancementduetostronguctuationoftheAFQorderparameter. 77


SpecicheatCofPrOs4Sb12ineldsupto8Tfor~H//(100)(upperpanel).ThemagneticeldphasediagramH-TofPrOs4Sb12ineldsupto8Tfor~H//(100)(lowerpanel)(ReprintedwithpermissionfromAokietal.[ 56 ]). 78


SpecicheatCofPrOs4Sb12in10and12TinthevicinityofFIOPtransitionfor~H//(100)(ReprintedwithpermissionfromRotunduetal.[ 31 ]). 79


SpecicheatCofPrOs4Sb12in13,13.5,and14T,for~H//(100).Ashoulderappearsatabout1.2{1.3Kat13TandtheFIOPtransitionissuppressedat13.5T. 80


SpecicheatCofPrOs4Sb12inmagneticeldsof16,20,and32T,for~H//(100). 81


MagneticeldphasediagramH-TofPrOs4Sb12for~H//(100)(H>8T).FilledsquaresrepresenttheFIOPtransition.OpensquarescorrespondtotheSchottkyanomaly.TheinsetisthemodelcalculationoftheSchottkyanomalyassumingthesingletasthegroundstate[ 90 ].ThesolidlinerepresentstheFIOPboundary;thedashedlinecorrespondstoamaximuminC(ReprintedwithpermissionfromRotunduetal.[ 31 ]). 82


ZeemaneectcalculationsforPrOs4Sb12inthe1CEFgroundstatescenario.Thereiscrossingofthetwolowestlevelsfor~H//(100)or~H//(111)ataround9Tandanti-crossingatthesameeldfor~H//(110).Thegureshowsonlythetwolowestlevels,i.e.thesinglet1andthetriplet5. 83


ZeemaneectforPrOs4Sb12inthe3CEFgroundstatescenario.Theeectinstronglyanisotropic.ThereisnocrossingofthetwolowestCEFlevelsfor~H//(110)or~H//(111).Thegureshowsonlythetwolowestlevels,i.e.thedoublet3andthetriplet5. 84


SpecicheatCofPrOs4Sb12for~H//(110),H=10,12,13,and14T.TheinsetshowsCversusTneatTxfor8,9,9.5,10,10.5,and11T. 85


SpecicheatCofPrOs4Sb12inH=12T,for~H//(100)(upperpanel),and~H//(110)(lowerpanel).ThearrowindicatestheAFQtransition. 86


ThemagneticeldphasediagramH-TofPrOs4Sb12for~H//(110)(H>8T).TheinsetshowsthespecicheatCmaxofAFQversusH.ForadenitionofsymbolsseeFig.6.5. Figure6-11. ElectricalresistivityversusT2forPrOs4Sb12.IntheinsetisversusTshowingthesuperconductingtransitionatTc=1.85K. 87


Resistivity(T)between20mKandabout0.9KofPrOs4Sb12in3,10,15,16,17,and18T(ReprintedwithpermissionfromRotunduandAndraka[ 96 ]). 88


ResistivityversusT2ofPrOs4Sb12for3.5,5.5,7,10,and13T. 89


Coecienta(=0+aTn)versusHforPrOs4Sb12eldsupto18Tisinupperpanel.Theresidualresistivity0(H)isshowninlowerpanel. 90


CoecientA(=0+AT2)versusH. 91


Thecalculated(H)ofPrOs4Sb12,forboth3and1scenarios.TheverticallineindicatestheeldcrossingofthetwolowestCEFlevels.Notethatthecrossingeldforthe1groundstatewasassumedat3T(lowerpanel)(ReprintedwithpermissionfromFrederickandMaple[ 93 ]). 92


7 16 ]orvirtualCEFexcitations[ 32 36 ],tocooperativemodelsinvokingproximitytoalong-rangeorder(proximitytothelowtemperaturestateofAFQorder)[ 56 ].Whileinvestigatingtheapplicabilityofthesemodels,closeattentionwaspaidtowhetherthesingle-ionparameterssuchastheCEFspectrumandhybridizationparametersvarywiththealloying. 57 ]andalmostnon-existentlanthanidecontractioninternaryskutteruditescontainingSb,ofageneralformLnT4Sb12,whereTandLnaretransitionelementandlightlanthanide,respectively. TopresentthischangeinaproperperspectivewerecallthatthechangeofthelatticeconstantacrossPr(Os1xRux)4Sb12[ 97 ]is10timeslarger.ThisisdespitethefactthattheatomicradiiofOsandRuarealmostidentical(1.35and1.35AforOsandRurespectively),whileLaismuchlargerthanPr(1.88versus1.82A).InPr(Os1xRux)4Sb12theCEFparametersincreasemonotonicallywithx.VerysmallchangesinlatticeconstantinPr1xLaxOs4Sb12suggestsmall,ifany,changesintheCEFparameters 93


Figure7-3showsthesusceptibilitiesonlyintherange1.85to10K.AlldataarenormalizedtoaPrmole.Duetotheverysmallsizeofthesamplesusedintheinitialmeasurements,themeasuredmomentofmostofthesamplesandthebackground(themagneticmomentofthesampleholderconsistingofapieceofaplasticdrinkingstraw)werecomparableat10K.Atroomtemperaturethemagneticmomentofthesampleswasevensmallerthanthebackground,especiallyfordiluteconcentrations.Inordertoavoidthisbackgroundcontribution,themagneticsusceptibilitieswereremeasured(forx=0,0.05,0.3,0.8and0.95)usingseveralcrystalsandholdingthembetweentwolongconcentricstraws.Nobackgroundsubtractionwasneededthistime.TheCurie-Weisstemperaturewasfoundabove150Kandtheeectivemagneticmomentintherange3.2{3.6B/Pratom.ThevaluesareintherangeofmomentsreportedforpurePrOs4Sb12.SomediscrepancybetweenthesevaluesandthatexpectedforPr3+,3.58B[ 52 ],canbeduetoanerrorinmassdetermination.Becauseoftheveryfragilenatureofthesecrystals,someofthembrokeoduringthemeasurementandsmallfractionsmovedinbetweenthetwotubes.Afurthercheckofthemagneticmomentwasperformedononelargecrystalforx=0.67(20mgeach).Figure7-4showsthesusceptibilityandtheinverseofsusceptibilityforx=0.67.FromtheCurie-Weisstthehightemperatureeectivemomentisfoundtobe3.62B/Prmol,closetothevalueexpectedforPr3+. 94


31 ],andneutron[ 32 84 ]measurementsestablished1singletastheCEFgroundstateseparatedbyabout8Kfromtherstexcited5triplet.VerysmallchangesinthepositionofthesemaximainthesusceptibilityaretherstindicationthatCEFareessentiallyunalteredbythedopingasexpectedfromthemeasurementofthelatticeconstant. Anotherinterestingaspectofthesusceptibilityisastronginitialreductionofthelow-temperaturevaluesof(normalizedtoamoleofPr)byLa.Thereductionofthemaximumsusceptibilityfromapproximately100forx=0toabout50memu/Prmolforx=0.4isclearlyoutsidetheerrorbar.Theaforementionedmeasurementsonassembliesofcrystalsforx=0.8and0.95alsoresultedina4Kvalueofabout505memu/Prmolforbothcompositions.SomebroadeninganddecreaseinmagnitudeoftheCEFsusceptibilityareexpectedinmixedalloysduetoincreasedatomicdisorder.However,theverylargeinitialdropinthesusceptibilityandlackofvariationabovex=0.4mightindicatethatsomecharacteristicelectronicenergy(analogoustoaKondotemperature)increasessharplyuponsubstitutingLaforPr.Asimilarsuppressionofthecorrespondingmaximumisobservedinthespecicheatdatadiscussedinthenextsection. 7.3.1SpecicHeatofPrOs4Sb12:SampleDependence 7 ].Morerecentspecicheatmeasurementsrevealedtwosuperconductingtransitions(Vollmeretal.[ 63 ],Mapleetal.[ 53 ],Oeschleretal.[ 64 ],Cichoreketal.[ 65 ]). 95


Ourobservationsareconsistentwithother,particularlymorerecent,reports.Almostallrecentinvestigationsndtwosuperconductinganomalies,morepronouncedatTc2andlessdenedatTc1.AnexceptiontothisruleareunpublisheddatabyAokietal.[ 98 ]thatshowasharppeakatTc1,andonlyachangeofslopeinC=TatTc2.Thewidthofthetransition,0.2K,denedabove,isquitesimilarforallpublisheddata.Thereisalargedistributionofreported(C=T)atTc,from500to1000mJ/K2mol.Ausualdeterminationof(C=T)byanequalarea(conservationofentropy)constructioncannotbeappliedduetothepresenceoftwosuperconductingtransitions.Applyingourmethod,C=T(Tc2)C=T(T3),resultsinanaverage(C=T)of800mJ/K2molforthemostrecentresults. 96


whereC=TisexpressedinmJ/(K2mol)andTinK(Fig.7-6).AsignicantnonlinearityinC=TversusT2isprobablyduetotherattlingmotionoflooselyboundLaatoms[ 99 ].ValuesofandforLaOs4Sb12reportedbyotherresearchgroupsare:of36[ 54 ],55mJ/K2mol[ 100 ],56[ 44 ]and=0.98mJ/K3mol[ 101 ]. Figures7-7and7-8presentthef-electronspecicheatofPr1xLaxOs4Sb12alloys,i.e.thespecicheatofLaOs4Sb12and,normalizingtoamoleofPr.NotethatthephononspecicheatofpurePrOs4Sb12inChapter6wastakenfromVollmeretal.[ 63 ],whichwasderivedbyttingthetotalspecicheatCtoafunctionrepresentingphonon,conductionelectrons,andSchottkycontributions.However,usingtheLaOs4Sb12specicheatseemstobemorejustiableformoderatelyandstronglyLa-dopedalloysandthereforethiswayofaccountingforphononsisconsistentlyusedinthischapteronLaalloying. 97


102 ]substitutedforU.Furthermore,sincePrOs4Sb12isclearlyanunconventionalsuperconductor(e.g.,timereversalsymmetrybreaking)whileLaxOs4Sb12ispresumablyaconventionalsuperconductorwewouldexpect,whilevaryingx,asuppressionofonetypeofsuperconductivitybeforetheothertypeemerges.Figure7-10showsthatthereissmoothevolutionofTc(andsuperconductivity)betweentheend-compounds.AsomewhatstrongersuppressionisobservedinthecaseofRureplacingOs[ 71 97 ].Buteveninthiscase,theTcreductionrateissmallifcomparedwiththemajorityofCe-andU-basedheavyfermionsandconsideringthefactthatRualloyingdrasticallyaectsCEFenergiesandhybridizationparameters. 63 ])oftheanomaliesatTc1andTc2precludesapopularspeculationthatoneofthesetransitionsisassociatedwithsurfacesuperconductivity. Asitwasalreadystressed,this(C=T)iscurrentlythemainevidenceforthepresenceofheavyelectrons.ThepresenceofamodiedSchottkyanomalynearTcmakes 98


PAGE 100

Inordertostudyfurtherthisspecicheatreduction,alargecrystalofPr0:33La0:67Os4Sb12,forwhichaddendaheatcapacityisnegligiblebelow6K,wasinvestigated.Firstofall,theeectivemagneticmomentmeasuredatroomtemperatureforthisparticularcrystalisveryclosetotheexpectedvalueforPr3+.Thus,thereducedspecicheatcannotbeexplainedbyincorrectPrstoichiometry,norbysomePrionsbeinginamixed-valentstate.Theresults,aftersubtractingthespecicheatofLaOs4Sb12anddividingby0.33,areshowninFig.7-11intheformofC=T.ThisgraphshowsalsoattothefunctiondescribingaSchottkyspecicheatforasinglet-tripletexcitations,scaledbyafactora=0.44.AsimilarscalingwasusedbyFredericketal.[ 71 ]toaccountforthespecicheatdataofPr(Os1xRux)4Sb12intermsofthesingletCEFgroundstate.AnecessitytousesuchasmallscalingfactorforthismodelofCEF(ofabout0.5)wasusedbyFredericketal.[ 71 ]toargueforadoubletCEFgroundstate.However,ascanbeseenfromFig.7-12,areasonablettothedoubletCEFmodelalsorequiresascalingfactor,althoughsomewhatlarger(a=0.73).Finally,attothesinglet-to-singletscatteringrequiresnoscalingatall.ThetshowninFig.7-13obtainedwithaasanadjustableparameter,resultedina=1.009(within1%). 100

PAGE 101

103 ]suggestpresenceoftheeldinducedAFQorderformuchhigherLa-concentrations,ashigh0.3. 101

PAGE 102

Thus,theseresults,togetherwithpreviouslydiscussedevolutionofzeroeldproperties,particularlyweaksensitivityofthetemperatureofthemagneticsusceptibilitymaximum,providestrongargumentsforCEFenergiesandeigenstatesbeingunaectedbyLaalloying,atleasttox=0.2.Ontheotherhand,thetemperaturesoftheSchottkyanomalyin13and14Tforx=0.6areabout0.3Klowerthanthoseforx0.2,suggestingapossibilitythattheCEFenergiesandthecrossingeldforx=0.4increasebyapproximately20%.Veryrecentresultsofthehigheldspecicheatstudyofx=0.67byAndraka[ 116 ]alsoimplytheCEFenergiestobe20{25%largerthanintheundopedmaterial.Thus,thereisapossibilityofanabrupt(butsmall)changeofCEFenergiessomewherebetweenx=0.2and0.67.However,towithinourexperimentaluncertainty,weclaimCEFenergies(andeigenstates)tobeidenticalbetweenx=0and0.2,thusintheconcentrationrangewheredramaticchangesofmareanticipatedbasedonmeasurementsof(C=T). Thesehigheldresultsallowustocommentonwhethertheagreementbetweenthemagnitudeofthezeroeldanomalyinmoderatelyandstrongly(x=0.67)dilutedalloysandtheSchottkyspecicheatcorrespondingtosinglet-singletexcitationsisaccidentalormeaningful.Thespecicheatmaximumforx=0.4in14Tisapproximately1500mJ/Kmol(200mJ/Kmol).ThisvalueissignicantlysmallerthanthetheoreticalvaluefortheSchottkymaximum(3650mJ/Kmol).SimilarlythespecicheatofPr0:33La0:67Os4Sb12inmagneticeldsashighas18Tshowsamaximumwhosevalue 102

PAGE 103

103 ]. TheresistivitiesofPr1xLaxOs4Sb12forx=0.05,0.3and0.7ineldsupto18Tweremeasuredintemperaturedownto20mK.Theuncertaintyinthedeterminationoftheabsolutevalueoftheresistivitywasupto30%.Theroomtemperatureresistivitywasapproximatelyequalforallthreecrystals.Therefore,weassumedthattheresistivityatroomtemperatureis300cmforallcrystals,consistentwiththepublishedvalueforbothendcompounds,PrOs4Sb12andLaOs4Sb12[ 92 ].TheratiooftheroomtemperatureresistancetotheresistanceextrapolatedtoT=0(RRR)was50,180,and170forx=0.05,0.3,and0.7,respectively.Withtheexceptionwiththeresultforx=0.05,thesevaluesbelongtothehighesteverreportedforpureanddopedPrOs4Sb12,suggestinggoodqualitiesofoursamples.Thex=0.05crystalwasfromthesamebatchwhoseresultsofspecicheatandsusceptibilityweredescribedinpreviousSections. 103

PAGE 104

Longitudinalmagnetoresistanceforx=0.05attwotemperatures,20and300mK,isshowninFig.7-20.Thetransversemagnetoresistanceforthesamealloyatfortemperatures20,310,and660mKisshowninFig.7-21.ThetwocurvesinFig.7-20showexcellentoverlap,implyinganabsenceoftemperaturedependenceoftheresistivitybelow300mKforanyeldbetween2and18T.Thisoverlapisconsistentwithourresistivitymeasurementsofthepurecompound(Fig.6-12)forwhichtheresistivitywasatbelow300mKforalleldsbetween3and18T. Thus,the20mKresistivityabovethecriticaleldisessentiallyidenticaltotheresidualresistivity,0.0forx=0.05increasesbyafactorof2between2and10T(Fig.7-21).Thisincreaseissignicantlylargerthanthecorrespondingincreaseforx=0whichwasabout25%.Thislargerincreasein0forx=0.05coincideswithsomesuppressionoftheAFQorderwithLa,asdemonstratedbyspecicheatmeasurementinmagneticelds.Alarger0at10Tforx=0.05isduetoasmallerdegreeoftheAFQorder.Thedropin0above10T,ontheotherhand,islesspronouncedforx=0.05thanx=0.ThesetrendscontinuewithfurtherLadoping,x=0.3. Theelectricalresistivityforx=0.3inzeroeld,justaboveTc,isproportionaltothesquareoftemperature,withA=0.16cm/K2(Fig.7-22,upperpanel).ByapplyingtheKadowaki-Woodsformula,A/2=105(withAincm/K2andinmJ/K2mol),wearriveatoforder100mJ/K2mol.However,theapplicationofjust0.5T(approximatelyHc2forthisconcentration)againrevealsthesaturationofresistivityatthelowesttemperatureswhichwasseeninthepurecompound(Fig.6-12).Inhighermagneticelds,thelinearvariationinT2isgraduallyrestrainedtonarrowertemperatureintervals 104

PAGE 105

Figure7-24showstheresistivityat20mKforx=0.3upto18Tforeldsparallelto(001),(011),and(010)andI//(001).Allthreeisothermsexhibitastepcenterednear9{10Tsuperimposedonalinearbackground.Intheinvestigatedeldrangewedonotndthedomestructurecharacteristicofx=0or0.05.Notetheapproximatelyequalslopesofthecurvesbelow3Tandabove14T.Interestingly,Sugawaraetal.[ 92 ]foundthatthemagnetoresistanceofLaOs4Sb12at0.36KisapproximatelylinearinmagneticeldandhasasimilarorientationasthatshowninFig.7-24.ThelargermagnetoresistanceofLaOs4Sb12for(011)thanforthe(001)directioncorrelateswiththelargermagnetoresistanceofx=0.3for(011)thanfor(001)direction(Fig.7-24).Therefore,wecanassumethattheapproximatelylinearinHmagnetoresistanceofPr0:7La0:3Os4Sb12below3Tandabove14Tisduetonormal(non-f)electrons.Subtractingsuchlinearcontributionsresultsinidenticalcurves,almostatbelow6Tandabove13T(insettoFig.7-24).Furthermore,theresultingcurvesareidenticalforallthreedirections,arguingforveryisotropicf-electronmagnetoresistance.ThisisotropicbehaviorisconsistentwiththesingletandinconsistentwiththedoubletCEFgroundstateofPr. TheeldvariationoftheresistivityshownintheinsettoFig.7-24isconsistentwithmodelcalculationsoftheresistivityforthe1-5modelbyFrederickandMaple[ 93 ].Accordingtothesecalculationsthattakeintoaccountmagneticandquadrupolardegreesoffreedom,theresistivityshouldexhibitasharpjumpatthecrossingeld.Theindependenceofthecrossingeldonthecrystallographicdirectionisalsoconsistentwiththe1-5CEFmodel(Fig.6-6).RecallthatCEFlevelcrossinginthe3-5CEFmodeloccursforthe(100)directiononly(Fig.6-7). Figure7-25showsrawdataforthesamesamplewhen~H//I//(100)forseveraltemperatures,20,310,660,and1100mK.Nobackgroundsubtractionhasbeendone. 105

PAGE 106

93 ]. Themorediluteconcentration,x=0.67wasinvestigatedtoabout0.35Kintemperatureandineldsupto14T.Itsresistivityatthelowesttemperaturesexhibitsasimilarmagneticelddependencetothatforx=0.3(Fig.7-26). ThepresentedresistivitydatashowthatthereisnosignofFIOPorderedphaseinthex=0.3material.OurpreviousspecicheatmeasurementsindicatedthattheFIOPphasedisappearssomewherenearx=0.2.However,themaineectoftheLadopingontheAFQanomalyisthesuppressionofitssize,withasomewhatsmallereectonthetransitiontemperatureitself.Thus,itispossiblethatthiseld-inducedAFQorderpersiststoconcentrationslargerthanx=0.2,butitssignaturesinthespecicheatareundetectableduetosmallentropiesinvolved.Thecalculationsoftheresistivitypredictingthestepinthemagnetoresistancewereperformedinasingleimpuritylimit,i.e.,assumingindependentscatteringfromeachPrion.ForPrOs4Sb12thescatteringinbothsmallandlargeeldsshouldbecoherent;i.e.,onemightexpectsmallcontributionfromfionsawayfromthecrossingeldofabout9T.Atthecrossingeld,thePrlatticelosesitscoherencesincesomeoftheionswillbeintheexcitedstate;i.e.,thetranslationalperiodicityislost.Webelieve,thiscoherencemechanismisresponsibleforthedomeshapeof(H)inPrOs4Sb12andthedierencebetweenpureandLa-dopedalloys. Theunchangedeldvalueforthestepintheresistivitybetweenx=0.3and0.67suggeststhatCEFenergiesarenotsignicantlyalteredbytheLadoping.ThisisinagreementwiththealmostunchangedtemperaturepositionofthemaximuminthemagneticsusceptibilityandspecicheatbelievedtobedueexcitationsbetweenthelowestCEFlevels. 106

PAGE 107

59 65 104 ].Thetemperatureseparationbetweenthetwostepsinacsusceptibilityforx=0isabout0.14Kandisapproximatelyequaltothatbetweenthepeaksinthespecicheat.TheeldcooleddcmagnetizationofsamplesperformedbyMeassonetal.[ 59 ]showedaMeissnereectof50%,indicating(likespecicheat)bulksuperconductivity.Therefore,thereisapossibilitythatthetwotransitionsareduetoinhomogeneouscoexistenceoftwosuperconductingphasesinPrOs4Sb12.Forx=0.05bothtransitionsarevisibleandtheoverallwidthofthetransitionisabout0.13K.Theonsettemperatureisapproximatelythesameasforx=0.Fromx=0.2to0.8theacsusceptibilitydataaresimilar.Thereisanincreaseofthetransitionwidth(inthereduced 107

PAGE 108

Themostprobableoriginofthewidesuperconductingtransitionsisinhomogeneities,whoseoriginisnotclear.PrOs4Sb12andLaOs4Sb12areisostructural,butLaOs4Sb12exhibitsaverysharptransition.Therefore,theinhomogeneitiesseemtobeassociatedratherwith4felectronsofPr.OneplausiblescenarioisamixtureoftwoelectroniccongurationsofPr,4f1and4f2.However,hightemperaturemagneticsusceptibilitydatawereingeneralconsistentwith4f2congurationofPr.Also,theLIIIabsorption[ 49 ]andinelasticneutronscattering[ 32 105 ]resultsagreewithavalenceofPrcloseto+3.Anotherscenariofortheexistenceofinhomogeneitiesistheclosenessofthesystemtoalongrangeantiferroquadrupolarorder[ 56 ].Thismeansthatclusterswithashort-rangeorderwouldhavedierentsuperconductingparametersthantheremainingpartofthesample. 108

PAGE 109

71 ].IfthesetwopeaksinC=Tarerelatedtotwodierentsuperconductingphases,thencrystalNo.1hasarelativelylargefractioncorrespondingtothehigherTcphase.Theevolutionofthesespecicheatpeaksinmagneticeldsindicatesthatthereisacorrelationbetweenthesetwophases. ThetwoinsetstoFig.7-28showthecriticaleldversusTdeterminedbyspecicheatforbothsamples.ForsampleNo.1bothsuperconductingtransitionsarevisibleineldsupto0.5T.OpensymbolsmarkthehighertransitiontemperatureTc1andtheclosedcirclesmarkTc2.Thelinesrepresentingthetwotransitionsremainapproximatelyparallel,withatemperatureseparationofabout0.12K.ThemostinterestingfeatureisapositivecurvatureinHc2versusTforH<2000Oe.TheinitialslopeofHc2versusTisdHc2=dT=1T/K.However,forH>2000OetheHc2islinearinT(insettoupperpanelofFig.7-28)andtheslopeisabouttwiceaslarge,i.e.dHc2=dT=2.1T/K.ApositivecurvatureinHc2versusTnearTc,wasalsodetectedinmeasurementsofelectricalresistivity,magneticsusceptibility,andspecicheat[ 7 55 63 67 ].Thisconsistencywasusedtoargueforintrinsicproperty,andnotduetosomeartifactoftransportmeasurementsorcomingfrominhomogeneitiesinthesamples.Meassonetal.[ 59 ]consideredthispositivecurvaturetobeahallmarkofthetwo-bandsuperconductivity.Thetwo-bandsuperconductivitydescription,inwhichtwodierentbandscorrespond 109

PAGE 110

106 ]inmagneticelds. InsampleNo.2wewereunabletodeterminetheevolutionofTc1inelds,thereforeonlyonelineispresentedintheinsettolowerpanel.Withintheresolutionofourmeasurements,thereisnocurvatureinHc2versusTnearTc2,andtheslopeisfoundequaltothatofsampleNo.1foreldshigherthan2000Oe,i.e.2.1T/K.Toourknowledge,thisistheonlymeasurementofHc2(T)nearTcthatdoesnotndthispositivecurvature. Forallconcentrationsotherthanx=0onlyasinglesuperconductinganomalyinthespecicheatmeasurementscouldbeclearlydetected(mostprobablycorrespondingtoTc2inx=0).Ourx=0.02and0.05alloysexhibitasmallbutdetectablecurvature(insettoFig.7-29forx=0.05).Forx=0.05,theinitialslopedeterminedforeldssmallerthan1000Oeisabout0.9T/K.Foreldshigherthan1000Oetheslopeis1.6T/K.QuitepossiblythereisasmallcurvatureinHc(T)forx=0.1,butitcannotbeclearlyresolved(insettoFig7-30).Forx=0.3(Fig.7-31)andalloyswithx>0.3wehaveaconventionalvariationofHc2(T),i.e.,withnopositivecurvaturenearTc.Forx=0.3dHc2=dT=0.5T/K. TheevolutionofdHc2=dTisofgreatinterestsincetheslopeofHc2versusTatTc2isrelatedtotheeectivemassofCooperpairs.Figure7-32showsthedHc2=dTversusLaconcentration.Forconcentrationsx=0,0.02,and0.05twosetsofdHc2=dTvalueshavebeendetermined:theinitialslope(opencircles)andthatforsucientlylargeelds,forwhichHc2isclearlylinearinT(H>2000Oeforx=0andH>1000Oeforx=0.02and0.05)(lledcircles).ThisdHc2=dT,markedbyclosedcircles,isabout2.1T/Kforx=0anddecreasesrapidlywithx,withmostofthereductiontakingplaceforsmallvaluesofx.ThisimpliesthattheeectivemassofcarriersisrapidlyreducedbysmallamountofLa. Inthecleanlimitofsuperconductivity(l),whichisthecaseforPrOs4Sb12andLaalloys,theeectivemassdependsonp 110

PAGE 111

Bothp Usingthetwobandsuperconductivityframework,thetwoslopesinHc2(T)(Fig.7-32)forx
PAGE 112

59 108 ].However,insuchacaseonlytheuppertransitionshouldshowupinthespecicheat.Anotherpossibility,notyetconsideredinliterature,isthattheuppersuperconductingtransitionisrelatedtothephasetransitionwithanorderhigherthan2[ 109 { 111 ].Atwo-steptransitionisexpectedinthiscase.Ahigherorderphasetransition(suchasthethirdorder)wouldbeverysusceptibletoimpuritiesandimperfectionsleadingtoasecondordertransitioninsucientlyimperfectcrystals.TheexpulsionofmagneticuxwouldbeweakwhenloweringTfromTc1toTc2,followedbyamorerapidexpulsionbelowTc2.Thisscenariomyaccountforthedierentmagneticeldresponseinmagneticeldsofthetwosamples.Thepossibilityofthethirdorderphasetransitionremainsspeculativesincefewmaterialswerereportedtoexhibitphasetransitionswithanorderhigherthan2[ 109 { 111 ]. 112

PAGE 113

X-raydiractionpatternsofPr1xLaxOs4Sb12versusLacontentxforx=0,0.1,0.2,0.4,and1.Theintensitiesarenormalizedtothehighestpeak. 113

PAGE 114

LatticeconstantaofPr1xLaxOs4Sb12versusLacontentx.Thesolidlineisalinearleast-squaresttoaversusx. Figure7-3. Magneticsusceptibility(T)ofPr1xLaxOs4Sb12normalizedtoPrmolebetween1.8and10K,measuredintheeldof0.5T.Themeasurementswereperformedonindividualcrystalswithmassesrangingbetween1and5mg.Largeuncertaintyareduetolargebackground(ReprintedwithpermissionfromRotunduetal.[ 112 ]). 114

PAGE 115

Magneticsusceptibility(T)ofPr0:33La0:67OsSb12versustemperatureT.IntheinsetistheCurie-Weisstofthehightemperaturedata(T>150K)fromwhichaneectivemomentof3.62B/Prmolehasbeencalculated. 115

PAGE 116


PAGE 117


PAGE 118

112 ]). 118

PAGE 119

112 ]). 119

PAGE 120

SuperconductingtransitiontemperatureTcversusxofPr1xLaxOs4Sb12(ReprintedwithpermissionfromRotunduetal.[ 112 ]). Figure7-10. TotalC=TdiscontinuityatTcand0versusxofPr1xLaxOs4Sb12for0x1(ReprintedwithpermissionfromRotunduetal.[ 112 ]). 120

PAGE 121

Figure7-12. 121

PAGE 122

Figure7-14. 122

PAGE 123

113 ]). Figure7-16. 113 ]). 123

PAGE 124

Figure7-18. 124

PAGE 125

Figure7-20. 125

PAGE 126

114 ]). 126

PAGE 127

114 ]). 127

PAGE 128


PAGE 129

Figure7-25. 129

PAGE 130

114 ]). 130

PAGE 131

ACsusceptibilityversusT=TcofPr1xLaxOs4Sb12,forx=0,0.05,0.4,0.8,and1.SampleNo.1exhibitonesuperconductingtransitionwhilesampleNo.2havetwo(ReprintedwithpermissionfromRotunduetal.[ 115 ]). 131

PAGE 132

115 ]). 132

PAGE 133

115 ]). Figure7-30. 115 ]). 133

PAGE 134

115 ]). Figure7-32. 115 ]). 134

PAGE 135

115 ]). 135

PAGE 136

ThischaptersummarizestheresultsofthermodynamicmeasurementsonPr1xLaxOs4Sb12(0x1)asafunctionoflanthanumconcentration(x),temperature,andmagneticeld.Specicheatmeasurementsineldsbetween8and32TofPrOs4Sb12extendedthepreviouslymeasuredH-Tphasediagramupto8T[ 56 ].TheSchottkyanomaly,duetoexcitationsbetweentwolowestcrystallineelectriceld(CEF)levels,wasfoundforboth~H//(100)and~H//(110)abovetheeldwheretheeld-inducedorderedphase(FIOP)(identiedwithanantiferroquadrupolarorderedphase[ 56 ])iscompletelysuppressed.TheH-TphasediagramshowsweakmagneticanisotropyandimpliesacrossingofthetwoCEFlevelsatabout9{10Tforbothelddirections.CalculationsoftheZeemaneectinthe1CEFgroundstatescenariopredictacrossingbetween1andthelowest5energylevel,betweenbetween9and10T,whichisalmostindependentontheelddirectionSimilarcalculationsforthe3CEFgroundstatemodelpredictbothstronganisotropyofthephasediagramandnocrossingforthe(110)direction.Thus,thisworkhasestablishedthenon-magnetic1singletbeingtheCEFgroundstate.Furthermore,ourinvestigationoftheeld-inducedorderedphase(FIOP)hasprovidedevidencesforthe(near)levelcrossingasthedrivingmechanismofFIOP.Thenon-magneticsinglet1CEFgroundstatecontradictstheideaofaquadrupolarKondoeect,atleastinthepresentformulation,astheoriginoftheheavyfermionbehaviorinPrOs4Sb12. TheLa-alloystudywasperformedtoprovideinsightontheoriginoftheelectronicmassenhancement.ZeroeldspecicheatofPr1xLaxOs4Sb12showedthatthetotal(C=T)atTcisreducedmorethansevenfoldfromabout800mJ/K2molbetweenx=0and0.3andstaysapproximatelyconstantandaboutequaltothatofLaOs4Sb12,whichisaconventionalsuperconductor,forx>0.3.Similarly,measurementsoftheuppercriticaleldsuggestedtheexistenceofacrossoverconcentration,xcr0.2{0.3.TheuppercriticaleldslopenearTcdecreasesrapidlywithxforx
PAGE 137

Inordertoverifythatsingle-ionparameters,suchashybridizationandCEFspectrum(consideredbysingleimpuritymodels),arenotseverelyaectedthealloying,measurementsofthelatticeconstantandhigheldspecicheatwereperformed.X-raypowderdiractionofPr1xLaxOs4Sb12revealedananomalouslysmallincreaseoflatticeconstantwithx(0.04%betweentheendcompounds).Thelow-temperaturemagneticsusceptibilityshowedalmostnonexistentconcentrationdependenceofthelowtemperaturemaximum,believedtobeduetoexcitationsbetweenthelowestCEFlevels.Specicheatinmagneticeldsupto14Tforx=0,0.02,0.1,and0.2showedthatthetemperatureoftheSchottkyanomalyhasasimilarelddependenceforalltheseconcentrations.Therefore,CEFenergiesandeigenstatesofPrareunchangedbetweenx=0andatleast0.2,i.e.,inthealloyparameterrangewherealargechangeoftheelectroneectivemassisobserved.Theseresultsreinforceourconclusionofanon-singleimpurityoriginoftheheavyfermionbehaviorofPrOs4Sb12.Inparticular,theyareinconsistentwiththecurrentlyprevailingFulde-Jensenmodel. Ourresultsimplyastrongcorrelationbetweentheparameterscharacterizingtheeld-inducedantiferroquadrupolarorder,suchasthetransitiontemperature,thesizeofthecorrespondingspecicheatanomaly,andm,suggestingapossibilitythattheheavyfermionstateisduetouctuationsofthequadrupolarorderparameter.Thispossibilityisconsistentwithourmagnetoresistanceresults,suggestingthatmincreaseswiththemagneticeldupthetheeldvalueforwhichthelong-rangeorderisobserved. ThestudyofthemagnetoresistanceonLa-dopedsamplesconrmedthesingletCEFgroundstate.Furthermore,theyprovidedanexplanationforafew-yearoldpuzzleofanunusualmagnetoresistanceofthepurecompound,previouslyusedtoargueforadoubletCEFgroundstate. 137

PAGE 138


PAGE 139

[1] G.R.Stewart.Rev.Mod.Phys.,56:755,1984. [2] Z.Fisk,H.R.Ott,T.M.Rice,andJ.L.Smith.Nature,320:124,1986. [3] N.GreweandF.Steglich.HeavyFermions.InHandbookonthePhysicsandChemistryofRareEarths.vol.14,page343,ElsevierScience.EditedbyK.A.GschneiderJr.andL.Eyring,Amsterdam,1991. [4] H.R.Ott.ProgressinLowTemperaturePhysics.vol.11,page215,Elsevier,Amsterdam,1987. [5] P.A.Lee,T.M.Rice,J.W.Serene,L.J.Shan,andJ.W.Wilkins.CommentsCondens.MatterPhys.,12:99,1986. [6] J.L.Sarrao,L.A.Morales,J.D.Thompson,B.L.Scott,G.R.Stewart,F.Wastin,J.Rebizant,P.Boulet,E.Colineau,andG.H.Lander.Nature,420:297,2002. [7] E.D.Bauer,N.A.Frederick,P.-C.Ho,V.S.Zapf,andM.B.Maple.Phys.Rev.B.,65:100506(R),2002. [8] M.Melamud,N.Shamir,andH.Shaked.J.Mag.Magn.Mater.,7:290,1978. [9] C.Broholm,H.Lin,P.T.Matthews,T.E.Mason,W.J.L.Buyers,M.F.Collins,A.A.Menovsky,J.A.Mydosh,andJ.K.Kjems.Phys.Rev.B,43:12809,1991. [10] M.T.Hutchings.Pointchargecalculationsofenergylevelsofmagneticionsincrys-tallineelectricelds,SolidStatePhysics,AdvancesinResearchandApplications.vol.16,AcademicPress,NewYorkandLondon,1964. [11] K.W.Stevens.Proc.Phys.Soc.,A65:209,1952. [12] B.BleaneyandK.W.Stewens.Rept.Progr.Phys.,16:108,1953. [13] K.R.Lea,M.J.K.Leask,andW.P.Wolf.J.Phys.Chem.Solids,23:1381,1962. [14] T.Inui,Y.TanabeandY.Onodera.GroupTheoryandItsApplicationsinPhysics.Springer,Berlin,1996. [15] K.Takegahara,H.Harima,andA.Yanase.J.Phys.Soc.Jpn.,70:1190,2001. [16] D.L.Cox.Phys.Rev.Lett.,59:1240,1987. [17] D.L.Cox.PhysicaC,153:1642,1988. [18] S.E.Barnes.Phys.Rev.B,37:3671,1988. [19] B.AndrakaandA.M.Tsvelik.Phys.Rev.Lett.,67:2886,1991. [20] C.L.Seaman,M.B.Maple,B.W.Lee,S.Ghamaty,M.S.Torikachvili,J.S.Kang,L.Z.Liu,J.W.Allen,andD.L.Cox.Phys.Rev.Lett.,67:2882,1991. 139

PAGE 140

C.L.Seaman,M.B.Maple,B.W.Lee,S.Ghamaty,M.S.Torikachvili,J.S.Kang,L.Z.Liu,J.W.Allen,andD.L.Cox.J.AlloysCompounds,181:327,1992. [22] L.Z.Liu,J.W.Allen,C.L.Seaman,M.B.Maple,Y.Dalichaouch,J.S.Kang,M.S.Torikachvili,andM.A.LopezdelaTorre.Phys.Rev.Lett.,68:1034,1992. [23] D.L.CoxandA.Zawadowski.Adv.Phys.,47:599,1998. [24] J.R.SchrieerandP.A.Wol.Phys.Rev.,149:491,1966. [25] P.D.SacramentoandP.Schlottmann.Phys.Lett.A,142:245,1989. [26] P.D.SacramentoandP.Schlottmann.PhysicaB,163:231,1990. [27] P.D.SacramentoandP.Schlottmann.Phys.Rev.B,43:13294,1991. [28] P.B.WiegmannandA.M.Tsvelick.Pis'maZh.Eksp.Teor.Fiz.,38:489,1983. [29] A.M.TsvelickandP.B.Wiegmann.Z.Phys.,54:201,1984. [30] A.P.Ramirez,P.Chandra,P.Coleman,Z.Fisk,J.L.Smith,andH.R.Ott.Phys.Rev.Lett.,73:3018,1994. [31] C.R.Rotundu,H.Tsujii,Y.Takano,B.Andraka,H.Sugawara,Y.Aoki,andH.Sato.Phys.Rev.Lett.,92:037203,2004. [32] E.A.Goremychkin,R.Osborn,E.D.Bauer,M.B.Maple,N.A.Frederick,W.M.Yuhasz,F.M.Woodward,andJ.W.Lynn.Phys.Rev.Lett.,93:157003,2004. [33] E.D.Bauer,P.-C.Ho,M.B.Maple,T.Schauerte,D.L.Cox,andF.B.Anders.Phys.Rev.B.,73:094511,2006. [34] T.Hotta.Phys.Rev.Lett.,94:067003,2005. [35] E.M.Forgan.Physica(Utrecht)B,107:65,1981. [36] P.FuldeandJ.Jensen.Phys.Rev.B,27:4085,1983. [37] J.A.Hertz.Phys.Rev.B,14:1165,1976. [38] A.J.Millis.Phys.Rev.B,48:7183,1993. [39] T.MoriyaandT.Takimoto.J.Phys.Soc.Jpn.,64:960,1995. [40] G.R.Stewart.Rev.Mod.Phys.,73:797,2001. [41] G.R.Stewart.Rev.Mod.Phys.,78:743,2006. [42] D.VuHung.Etudetheoriquedel0etatdevortexdansdenouveauxsupraconducteurs:MgB2etPrOs4Sb12.PhDthesis,JosephFourierUniversity-GrenobleI,2006. [43] W.JeitschkoandD.J.Braun.ActaCrystallogr.B,33:3401,1977. 140

PAGE 141

H.Sugawara,S.Osaki,S.R.Saha,Y.Aoki,H.Sato,H.Inada,H.Shishido,R.Settai,Y.Onuki,H.Harima,andK.Oikawa.Phys.Rev.B.,66:220504(R),2002. [45] M.-A.Measson.LaSkutteruditePrOs4Sb12:Supraconductiviteetcorrelations.PhDthesis,CEAGrenoble,2005. [46] H.Sato,Y.Aoki,T.Namiki,T.D.Matsuda,K.Abe,S.Osaki,S.R.Saha,andH.Sugawara.PhysicaB,328:34,2003. [47] V.Keppens,D.Mandrus,B.C.Sales,B.C.Chakoumakos,P.Dai,R.Coldea,M.B.Maple,D.A.Gajewski,E.J.Freeman,andS.Bennington.Nature,395:876,1998. [48] L.Mihaly.Nature,395:839,1998. [49] D.Cao,F.Bridges,S.Bushart,E.D.Bauer,andM.B.Maple.Phys.Rev.B,67:180511,2003. [50] T.Goto,Y.Nemoto,K.Sakai,T.Yamaguchi,M.Akatsu,T.Yanagisawa,H.Hazama,K.Onuki,H.Sugawara,andH.Sato.Phys.Rev.B,69:180511,2004. [51] T.Tayama,T.Sakakibara,H.Sugawara,Y.Aoki,andH.Sato.J.Phys.Soc.Jpn.,72:1516,2003. [52] N.W.AshcroftandN.D.Mermin.SolidStatePhysics.W.B.SaundersCo.,Philadelphia,1976. [53] M.B.Maple,P.-C.Ho,E.D.Bauer,W.M.Yuhasz,F.M.WoodwardV.S.Zapf,N.A.Frederick,andJ.W.Lynn.J.Phys.Soc.Jpn.,S71:23,2002. [54] E.D.Bauer,A.Slebarski,E.J.Freeman,C.Sirvent,andM.B.Maple.J.Phys.:Condens.Matter,13:4495,2001. [55] P.-C.Ho,N.A.Frederick,V.S.Zapf,E.D.Bauer,T.D.Do,andM.B.Maple.Phys.Rev.B,67:180508(R),2003. [56] Y.Aoki,T.Namiki,S.Ohsaki,S.R.Saha,H.Sugawara,andH.Sato.J.Phys.Soc.Jpn.,71:2098,2002. [57] D.J.BraunandW.Jeitschko.J.Less-CommonMet.,72:147,1980. [58] H.HarimaandK.Takegahara.PhysicaB,359-361:920,2005. [59] M.-A.Measson,D.Braithwaite,J.Flouquet,G.Seyfarth,J.P.Brison,E.Lhotel,C.Paulsen,H.Sugawara,andH.Sato.Phys.Rev.B,70:064516,2004. [60] K.KadowakiandS.B.Woods.SolidStateCommun.,58:507,1986. [61] T.Sakakibara,T.Tayama,T.Onimaru,D.Aoki,Y.Onuki,H.Sugawara,Y.Aoki,andH.Sato.PhysicaB,163:231,1990. 141

PAGE 142

N.Oeschler,P.Gegenwart,F.Weickert,I.Zerec,P.Thalmeier,F.Steglich,E.D.Bauer,N.A.Frederick,andM.B.Maple.Phys.Rev.B,69:235108,2004. [63] R.Vollmer,A.Fait,C.Peiderer,H.v.Lohneysen,E.D.Bauer,P.-C.Ho,V.Zapf,andM.B.Maple.Phys.Rev.Lett.,90:057001,2003. [64] N.Oeschler,P.Gegenwart,F.Steglich,N.A.Frederick,E.D.Bauer,andM.B.Maple.ActaPhys.Pol.B,34:959,2003. [65] T.Cichorek,A.C.Mota,F.Steglich,N.A.Frederick,W.M.Yuhasz,andM.B.Maple.Phys.Rev.Lett.,94:107002,2005. [66] J.K.HulmandB.B.Goodman.Phys.Rev.B,106:659,1957. [67] K.Izawa,Y.Nakajima,J.Goryo,Y.Matsuda,S.Osaki,H.Sugawara,H.Sato,P.Thalmeier,andK.Maki.Phys.Rev.Lett.,90:117001,2003. [68] G.Seyfarth,J.P.Brison,M.-A.Measson,J.Flouquet,K.Izawa,Y.Matsuda,H.Sugawara,andH.Sato.J.AlloysCompounds,95:107004,2005. [69] N.Padamsee,J.E.Neighbor,andC.A.Shiman.J.LowTemp.,12:387,1973. [70] K.Grube,S.Drobnik,C.Peiderer,H.v.Lohneysen,E.D.Bauer,andM.B.Maple.Phys.Rev.B,73:104503,2006. [71] N.A.Frederick,T.A.Sayles,andM.B.Maple.Phys.Rev.B,71:064508,2005. [72] H.Suderow,S.Vieira,J.D.Strand,S.Budko,andP.C.Caneld.Phys.Rev.B,69. [73] D.E.MacLaughlin,J.E.Sonier,R.H.Hener,O.O.Bernal,B.-L.Young,M.S.Rose,G.D.Morris,E.D.Bauer,T.D.Do,andM.B.Maple.Phys.Rev.Lett.,89:157001,2002. [74] E.E.M.Chia,M.B.Salamon,H.Sugawara,andH.Sato.Phys.Rev.Lett.,91:247003,2003. [75] H.Kotegawa,M.Yogi,Y.Imamura,Y.Kawasaki,G.-q.Zheng,Y.Kitaoka,S.Ohsaki,H.Sugawara,Y.Aoki,andH.Sato.Phys.Rev.Lett.,90:027001,2003. [76] Y.Aoki,A.Tsuchiya,T.Kanayama,S.R.Saha,H.Sugawara,H.Sato,W.Higemoto,A.Koda,K.Ohishi,K.Nishiyama,andR.Kadono.Phys.Rev.Lett.,91:067003,2003. [77] Y.Aoki,H.Sugawara,H.Hisatomo,andH.Sato.J.Phys.Soc.Jpn.,74:209,2005. [78] K.Maki,S.Haas,P.Parker,H.Won,K.Izawa,andY.Matsuda.Europhys.Lett.,68(5):720,2004. 142

PAGE 143

R.Bachmann,F.J.DiSalvoJr.,T.H.Geballe,R.L.Greene,R.E.Howard,C.N.King,H.C.Kirsch,K.N.Lee,R.E.Schwall,H.U.Thomas,andR.B.Zubeck.Rev.Sci.Instrum.,43:205,1972. [80] R.E.Schwall,R.E.Howard,andG.R.Stewart.Rev.Sci.Instrum.,46:1054,1975. [81] G.R.Stewart.Rev.Sci.Instrum.,54:1,1983. [82] J.S.Kim.DopingExperimentsinHeavyFermionSuperconductors.PhDthesis,UniversityofFlorida,1992. [83] D.J.MixsonII.DieringRolesofDisorder:non-Fermi-LiquidBehaviorinUCu5xNixandCurieTemperatureEnhancementinUCu2Si2xGex.PhDthesis,UniversityofFlorida,2005. [84] M.Kohgi,K.Iwasa,M.Metoki,S.Araki,N.Bernhoeft,J.M.Mignot,A.Gukasov,H.Sato,Y.Aoki,andH.Sugawara.J.Phys.Soc.Jpn.,72:1002,2003. [85] T.Tayama,T.Sakakibara,K.Kitami,M.Yokoyama,K.Tenya,H.Amitsuka,D.Aoki,Y.Onuki,andZ.Kletowski.J.Phys.Soc.Jpn.,70:248,2001. [86] J.Kondo.J.Phys.Soc.Jpn.,16:1690,1961. [87] B.Bleaney.J.Appl.Phys.,34:1024,1963. [88] B.AndrakaandY.Takano.Rev.Sci.Instr.,67:4256,1996. [89] E.S.R.Gopal.SpecicHeatatLowTemperatures.PlenumPress,NewYork,1966. [90] Y.Aokietal:Unpublished,Privatecommunication. [91] M.B.Maple,E.D.Bauer,V.S.Zapf,E.J.Freeman,N.A.Frederick,andR.P.Dickey.ActaPhys.Pol.B,32:3291,2001. [92] H.Sugawara,M.Kobayashi,S.Osaki,S.R.Saha,T.Namiki,Y.Aoki,andH.Sato.Phys.Rev.B,72:014519,2005. [93] N.FrederickandM.B.Maple.J.Phys.Condens.Matter,15:4789,2003. [94] P.FuldeandM.Loewenhaupt.Adv.Phys.,34:589,1986. [95] Z.FiskandD.C.Johnston.SolidStateCommun.,22:359,1977. [96] C.R.RotunduandB.Andraka.J.Appl.Phys.,97:10A927,2005. [97] N.A.Frederick,T.D.Do,P.-C.Ho,N.P.Butch,V.S.Zapf,andM.B.Maple.Phys.Rev.B,69:024523,2004. [98] Y.Aokietal:Unpublished. 143

PAGE 144

W.Schnelle,A.Leithe-Jasper,M.Schmidt,H.Rosner,H.Borrmann,U.Burkhardt,J.A.Mydosh,andY.Grin.Phys.Rev.B,72:020402(R),2005. [100] S.Sanada,Y.Aoki,H.Aoki,A.Tsuchiya,D.Kikuchi,H.Sugawara,andH.Sato.J.Phys.Soc.Jpn.,74:246,2005. [101] T.Namiki,Y.Aoki,H.Sugawara,andH.Sato.ActaPhysicaPolonicaB,34:1161,2003. [102] U.Ahlheim,M.Winkelmann,P.vanAken,C.D.Bredl,F.Steglich,andG.R.Stewart.J.Magn.Magn.Mater.,76-77:520,1988. [103] Y.Yonezawa,S.Toda,Y.Aoki,K.Kuwahara,D.Kikichuchi,H.Sugawara,andH.Sato.Abstract23pPSA-1,FallMeetingoftheJapan.Phys.Soc.,2006. [104] S.Drobnik,K.Grube,C.Peiderer,H.v.Lohneysen,E.D.Bauer,andM.B.Maple.PhysicaB,359-361:901,2005. [105] K.Kuwahara,K.Iwasa,M.Kohgi,K.Kaneko,N.Metoki,S.Raymond,M.-A.Measson,J.Flouquet,H.Sugawara,Y.Aoki,andH.Sato.Phys.Rev.Lett.,95:107003,2005. [106] G.Seyfarth,J.P.Brison,M.-A.Measson,J.Flouquet,K.Izawa,Y.Matsuda,H.Sugawara,andH.Sato.Phys.Rev.Lett.,95:107004,2005. [107] G.Bruls,D.Weber,B.Wolf,P.Thalmeier,B.Luthi,A.deVisser,andA.Menovsky.Phys.Rev.Lett.,65:2294,1990. [108] D.M.Broun,P.J.Turner,G.K.Mullins,D.E.Sheehy,X.G.Zheng,S.K.Kim,N.A.Frederick,M.B.Maple,W.N.Hardy,andD.A.Bonn.cond-mat/0310613. [109] P.Kumar,D.Hall,andR.G.Goodrich.Phys.Rev.Lett.,82:4532,1999. [110] P.Kumar.Phys.Rev.B,68:064505,2003. [111] R.WernerandV.J.Emery.Phys.Rev.B,67:014504,2003. [112] C.R.Rotundu,P.Kumar,andB.Andraka.Phys.Rev.B,73:014515,2006. [113] C.R.RotunduandB.Andraka.AIPConferenceProceedings,24thInternationalConferenceonLowTemperaturePhysics,Orlando,FL,August10-17,2005(850:653,2006). [114] C.R.RotunduandB.Andraka.cond-mat/0610520. [115] C.R.Rotundu,H.Tsujii,P.Kumar,B.Andraka,H.Sugawara,Y.Aoki,andH.Sato.cond-mat/0606730. [116] B.Andraka.Privatecommunication. 144

PAGE 145

CostelRemusRotunduwasbornMarch3,1974inD^ngeni,Romania.RaisedinHulubandthenTrusesti,hewenttoBotosanitoattendtheA.T.LaurianNationalCollege,mathematics-physicssection,wherehegraduatedfromhighschoolinJune1992.HegraduatedwithaBachelorofSciencedegreeinphysics(specializingintheoreticalphysics)fromRomania'soldestuniversity,Al.I.CuzaUniversity,inJune1997.OneyearandahalflaterhegraduatedfromthesameuniversitywithMastersofScienceinphysics(specializinginnonlineartheoryphenomena).Afterworking2yearsasaresearchassistant(theorist)attheNationalInstituteofResearchandDevelopmentforTechnicalPhysicsfromthesamecity,heleftfortheU.S.tocontinuehisgraduatestudiesatUniversityofFlorida,wherehepursuedaPh.D.degreeinexperimentalcondensedmatterphysics.HeworkedunderDr.BohdanAndraka'supervisionintheheavy-fermionarea.DuringhisstayatUniversityofFloridahewasalsoappointedasPhysicsIandIIlaboratoryinstructor.AftergraduationhegotaResearchAssociatepositionwithDr.RichardGreene,attheCenterforSuperconductivityResearch(CSR)-UniversityofMarylandCollegePark. 145