Coexisting on- and off-center Yb3+ sites in Ce Yb Fe P skutterudites 1−x x 4 12 F. A. Garcia1, D. J. Garcia2, M. A. Avila1, J. M. Vargas1, P. G. Pagliuso1, C. Rettori1, M. C. G. Passeggi3, S. B. Oseroff4, P. Schlottmann5, B. Alascio2, and Z. Fisk6 1Instituto de Física “Gleb Wataghin", UNICAMP, Campinas-SP, 13083-970, Brazil. 2Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) and Centro Atómico Bariloche, S.C. de Bariloche, Río Negro, Argentina. 3INTEC (CONICET and UNL), S3000GLN, Santa Fe, Argentina. 4San Diego State University, San Diego, California 92182, USA. 5 Department of Physics, Florida State University, Tallahassee, Florida 32306, USA 6University of California, Irvine, CA, 92697-4573, USA. 0 (Dated: January 19, 2010) 1 0 Electron Spin Resonance (ESR) measurements performed on the filled skutterudite system 2 Ce1−xYbxFe4P12 (x.0.003) unequivocally reveal the coexistence of two Yb3+ resonances, associ- ated with sites of considerably different occupations and temperature behaviors. Detailed analysis n a of the ESR data suggests a scenario where the fraction of oversized (Fe2P3)4 cages that host Yb ions are filled with a low occupation of on-center Yb3+ sites and a highly occupied T-dependent J distribution of off-center Yb3+ sites. Analysis of the 171Yb3+(I=1/2) isotope hyperfine splittings 9 reveal that these two sites are associated with a low (∼ 1 GHz) and a high (& 15 GHz) rattling 1 frequency, respectively. Our findings introduce Yb3+ in Th symmetry systems and uses the Yb3+ ESR as a sensitive microscopic probe to investigate theYb3+ ions dynamics. ] l e - r I. INTRODUCTION to probe this ion’s dynamical behavior within oversized t s cages of the Ce1−xYbxFe4P12 system (x.0.003). . at The dynamics of guest or filler ions vibrating loosely Skutterudites crystallize in the cubic LaFe4P12 struc- m insideoversizedhostcageshasbeenatopicofcurrentfo- ture with space group Im3.9 Each R ion is surrounded byeighttransitionmetalionsformingacube,andtwelve - cus in condensed matter physics. The anomalousbehav- d iorsoftheseso-calledrattler ionsraiseinterestbothfrom pnictogenionsthatformaslightlydeformedicosahedron. n Our work lies in the fact that the local point symmetry the fundamental understanding of the unusual potential o fortheRionsisT ,whichlackstwosymmetryoperations wellstheyaresubjectedto(andconsequentanharmonic- h [c ities intheir vibrationalmotions)aswellasfromthe im- (C4andC2′ rotations)10whencomparedtocommoncubic plications of such rattling on the dampening of thermal structures. Thus,the electriccrystalfield(CF) Hamilto- 1 nian(H )allowsforanadditionalsixthordertermwith transport in the material, which invites application per- CF v spectives in the field of thermoelectrics.1 Thermoelectric anextracrystalfieldparameter(CFP),B6t.11,12 Thissys- 7 tems present a complex magnetic behavior, and it is es- 8 materials, which can convert heat into electricity, are of sential to know their CF level schemes for its complete 2 great interest for energy sustainability and energy har- description.13,14 3 vesting (transformation of waste heat into useful elec- . tricity). The main obstacle is the low thermoelectric ef- ESR is a powerful microscopic tool to provide in- 1 0 ficiency of materials for heat to electricity conversion, formation about CF effects, site symmetries, valencies 0 which is quantifiedby the thermoelectric figure ofmerit, of the paramagnetic ions, g-values, fine and hyperfine 1 ZT. The highZT value is the resultofthe highSeebeck parameters.12 The ESR of excited states may be also : coefficientandthelowthermalconductivity.2 Amongthe observable, then, by measuring a R-ion ESR at differ- v i best-known cage systems displaying such characteristics ent frequencies and temperatures, one may obtain CF X are the filled skutterudite compounds RT4X12, where R ground states and, in some cases, the full set of CFP’s r is a rare earth or actinide, T is a transition metal (Fe, thatdeterminetheoverallsplittingofaR-ionJ-multiplet a Ru, Os) and X is a pnictogen (P, As, Sb). Besides ex- groundstate.15 PreviousworksonCe1−xRxFe4P12 forR hibiting a rich variety of ground states and promising = Nd, Dy, Er, Yb; (x . 0.005) succeeded in explaining thermoeletricity,3 the question of whether the R ions thelow-T ESRresultsusingsuchanexpandedH ,and CF in these compounds are sited on- and/or off-center in thefullsetofCFP’scouldbedetermined.15 However,we the oversized rigid (T2X3)4-cages is a matter of intense now found that in the case of R = Yb, as T increases, debate.4,5 There is also controversy over the extent to a second Yb3+ resonance emerges from the low-T spec- which the weakly bounded R ions can be regardedas in- tra, corresponding to a distinct site, coexisting with the dependent Einstein oscillators, and how effectively they first one. The presence of the new term in the H has CF contribute to a phonon-glasstype of heat conduction.6–8 provenessentialtoexplainthe appearanceofthis second In this work we take advantage of a uniquely favorable Yb3+ resonance. The sensitivityofYb3+ 4f-electronsto conjunctionbetweenthe chemicalandstructuralcharac- this type of CF environment make it a rare and useful teristics of skutterudites and their effect on the Electron probing ion to help the understanding of the potential SpinResonance(ESR)oftheJ =7/2multipletofYb3+, well responsible for its motion. 2 II. EXPERIMENTAL Single crystals of Ce1−xYbxFe4P12 (x . 0.003) were grown in Sn-flux as described in Ref. 16. The cubic 2.60 Ce1-xYbxFe4P12 (x = 0.0023) structure (Im3) and phase purity were checkedby x-ray powder diffraction. The Yb concentrations were deter- e2.56 Ttmiohinnee,dEMSfrR(oHme,xTtph)e,ermiHmeeaansntusdreudTse-idndeapcreSynQsdtUeanlIscDeofdocf∼-tm2hxae2gmxn2eatgmonmmete3itzeaor-f. g-valu2.52 -2 H)/H x 10 123...000 (g H= )| g=( T)H -( Tg )( 4- .2H)|(4.2) ( naturally grown crystallographic faces, as well as crys- 2.48 0.0 ( H)/H=1.3(3) g/g tals crushed into fine powder. The ESR spectra were 0.00 0.g6/0g x1 1.200-2 1. 80 a) tsapkeecntroinmeBtreurskeursiXng(9a.p4p8roGpHriza)tearnedsoQna(t3o4r.s4coGuHpzle)dbtaonda e)300 CPorywsdtaelr XX -- bbaanndd 2 mw)9 CX r-y bsatanld TK-.cTonhteroElSleRrosfpeachtrealiuomfthgeas17fl0uYxbs3y+s(tIe=m0)foirso4t.o2p.esTho.w4ed5 dth (O200 Crystal Q - band I(P)/I(0.136 4.2 K 6.4 K BNraorarodw the superposition of a narrow line and a slightly shifted wi 0.00 0.071/2 01./124 0.21 e P (W ) broad line. For single crystalsandpowderedsamples the n spectra were, respectively, fitted by the superposition of Li100 twodysonian(metalliclineshape)andtwolorentzianres- onanceswithadjustableresonancefields(H0),linewidths 0 b) (∆H), A/B ratios and amplitudes.17 0 10 20 30 40 50 T (K) III. EXPERIMENTAL RESULTS Figure 2: (color online) X and Q-bands low T-evolution of: a) g-valueand b) ∆H for the narrow and broad lines of Fig. 1andalso forapowderedcrystal. AtX-band,Inseta)shows thecorrelationbetweenδ(∆H)/H andδg/g,andInsetb)the (cid:13) microwave power dependenceof the ESR intensity. X - band (cid:13)Ce(cid:13) Yb(cid:13)Fe(cid:13)P(cid:13) (x = 0.0023)(cid:13) a(cid:13))(cid:13) 1-x(cid:13) x(cid:13) 4(cid:13) 12(cid:13) s)(cid:13) x 10(cid:13) |(cid:13) 170(cid:13)Yb(cid:13) Best fit(cid:13) unit 4.2 K(cid:13) 170Yb3+ isotope evolves into two lines, a narrow and b. 173(cid:13)Yb(cid:13)|(cid:13) |(cid:13) |(cid:13) |(cid:13) |(cid:13) |(cid:13) x 4(cid:13) (cid:13) a broad one. At low-T the measured g-values for the r 15.3 K(cid:13) (cid:13) e (a 1x7 15(cid:13)Y(cid:13)b(cid:13) |(cid:13) |(cid:13) narrow and broad lines are essentially the same, g ≃ v x 20(cid:13) 2.57, different from the g-values of 2.666 and 3.428 ex- vati 45 K(cid:13) (cid:13) pected for Γ6 and Γ7 doublets, respectively.12 At X- eri 2.2(cid:13) 2.4(cid:13) 2.6(cid:13) (cid:13)2.8(cid:13) 3.0(cid:13) 3.2(cid:13) band the narrow line displays the full hyperfine spec- n d Q - band (cid:13) (cid:13) b(cid:13))(cid:13) tra for the Yb isotopes 170Yb3+(I=0), 171Yb3+(I=1/2) orptio x 3(cid:13) 5.5 K(cid:13) atrnada1r7e3Yabs3s+oc(Iia=t5ed/2t)o,cYonbfi3r+miinongst.haFtrtohme otbhseerhvyedpesrpfience- bs 19 K(cid:13) splittings the corresponding hyperfine constants 171A = (cid:13)A (cid:13) 440(10) Oe and 173A = 120(3) Oe were obtained. These x 4(cid:13) 27 K(cid:13) values are ∼ 20% smaller than the hyperfine constants of Yb3+ in a KDGS of any system with O cubic point h symmetry.12,18 The hyperfine lines corresponding to the 9.0(cid:13) 9.2(cid:13) 9.4(cid:13) 9.6(cid:13) 9.8(cid:13) 10.0(cid:13) broad line of the 171Yb3+ isotope were not observed. H (KOe)(cid:13) Figures 2a and 2b show, respectively, the T-evolution (4.2.T .40K)oftheg-valuesandlinewidths,∆H,for the narrow and broad lines of Fig. 1 and also for a pow- Figure 1: (color online) T-evolution (4.2 . T . 40 K) of the normalized Yb3+ ESR spectra in a Ce1−xYbxFe4P12 deredsampleatX-band. Thefollowingfeaturesarenote- (x ≃ 0.0023) single crystal: a) X-band and b) Q-band. The worthy: a) forthe sitescorrespondingtothe narrow line enhancedlowfieldspectrashowstheabsenceofhyperfinelines the g-value and ∆H are frequency- and T-independent; for the broad line. b) for the broad line sites the g-value and ∆H are T- dependent, and only ∆H is frequency dependent; c) the Figures 1a and 1b show, respectively, selected X- broad line of the powdered sample is broader than that and Q-band ESR spectra for the Kramers doublet of the crystal, while the narrow line is about the same; ground state (KDGS) of the 170Yb3+(I=0) isotope in a d) angular dependent ESR experiments found these res- Ce1−xYbxFe4P12 (x ∼= 0.0023) single crystal. As T in- onances to be isotropic; e) for the various samples the creases the nearly single line observed at low-T for the relative change of the broad line linewidth, δ(∆H)/H, 3 tive intensities the broad and narrow lines correspond, (cid:13) respectively, to ∼ 95% and ∼ 5% of the Yb3+ ions fill- nits)(cid:13) 9(cid:13) Ce(cid:13)1-x(cid:13)Yb(cid:13)x(cid:13)Fe(cid:13)4(cid:13)P(cid:13)12(cid:13) (x = 0.0023)(cid:13) a(cid:13))(cid:13) ing cages. Figure 3b presents the T-dependence of the u (cid:13) relative population for the low occupied sites (narrow rb. 1000(cid:13) line). The largeobserveddropstronglysuggeststhat, as 2(cid:13) (a 6(cid:13) (cid:13) 10100(cid:13)(cid:13) (cid:13) T-increases, the low populated Yb3+ sites migrate, in a 10(cid:13) 1(cid:13) (cid:13) reversible way, to the highly populated Yb3+ ones. The y x 3(cid:13) 10(cid:13) T2 0(K(cid:13) )3(cid:13)0(cid:13) 40(cid:13) inset of Fig. 3b shows for the broad, narrow and hy- sit perfine lines the T-dependence of their intensities (×T) en 0(cid:13) (cid:13) normalized at T ≃ 4.2 K. This data reveals that the Int Broad line(cid:13) b(cid:13))(cid:13) broad line practically follows a Curie-Weiss (C-W) law, 9(cid:13) Narrow line(cid:13) T (cid:13) 6(cid:13) (cid:13) while the narrow line surprisinglydropsfaster thana C- /I(cid:13) %(cid:13)narrow(cid:13)total(cid:13) 6(cid:13) HCyuprieer-fWineei slisn(cid:13)e(cid:13) [I(T)/I(4.2)] x 0240(cid:13)(cid:13)(cid:13)(cid:13) 10(cid:13) T20 ((cid:13)K3)(cid:13)0(cid:13) 40(cid:13) (cid:13) (cid:13) Wi1tdh70eeYabr.ebeh3sTa+ovhniieaoonrnCc,se-gWsciavarerbrniyseefhulaofrrvctoaihomleirrzeaissduCapmFnpaooKgtrhntDeetrGtoiciStn.hmdeiocsmaitteeionsntmatinhgdarattthtiohanet I(cid:13) 3(cid:13) IV. ANALYSIS AND DISCUSSION 0(cid:13) 10(cid:13) 20(cid:13) 30(cid:13) 40(cid:13) T(K)(cid:13) In order to analyze our ESR data in Ref. 15 we used the expanded Hamiltonian, H : CFZ Figure 3: a) T-dependence of the ESR intensity for the two resonancesofthe170Yb3+(I=0)isotope. Theintensityofone ofthe171Yb3+(I=1/2)hyperfineline,normalized bythenat- H = W (1−|y|) xO4c +(1−|x|)O6c +yO6t ural abundance, is also shown. The inset presents the same CFZ (cid:26) (cid:20) F40 F60(cid:21) F62(cid:27) datainlogscale. b)T-dependenceoftherelativeintensityfor +g µ H·J, (1) thenarrow line. The inset displays thedata of Fig. 3a (×T) J B normalizedat4.2K.ThegreenlinesshowtheC-Wbehavior. whereamagneticmomentJwithaLandég-factorg is J considered. The CF includes the usualcubic O 21 terms h parametrizedbythexvariablethatmeasurestherelative scales at all-T with the relative change of its g-value, weightofthe4thand6thordertermsandalsotakesinto δg/g (δ(∆H)/H ≃ 1.3(3) δg/g, see inset of Fig. 2a); f) consideration the new term Ot. The relative weight y 6 saturationESR intensity measurementsshowthatat4.2 linearlyinterpolatesbetweentheO cubictermsfory=0 h K and ∼ 10 mW the broad and narrow lines present, re- and the Ot term for y= 1. This (x,y) parametrization 6 spectively, a ∼ 30% and ∼ 50% saturation (see inset of allows the entire range of the CFP’s to be accounted for Fig. 2b). within the finite intervals −1≤x≤ 1 and |y|< 1 and the All the experimental features given in Figs. 1 and 2 results do not depend on the sign of y. By diagonalizing were confirmed in crystals from different batches with H oneobtains,asafunctionofxandy,theCFwave CFZ comparable Yb concentrations. They lead us to con- functions and energies for each of the R in units of W. clude that: i) the narrow and the broad lines are, re- Fromthegroundstatewavefunctionthelowfieldg-value spectively,homogeneousandinhomogeneousresonances; can be calculated15. ii) the originof the inhomogeneity is a distribution of g- A combined analysis of the ESR data for Er3+, Dy3+ valuesoftheorderofthechangeintheg-value;iii)theT- and Yb3+ impurities in Ce1−xRxFe4P12 allowed us to independent∆H forthenarrow line indicatesthatthere pinpointtheexact(x=0.523,y=0.082)valuescorrespond- is no Yb3+ spin-lattice relaxation via exchange interac- ing to the Yb3+ narrow line observed at T ≃ 4.2 K and tionwiththeconduction-electrons;19,20iv)thesaturation g ≃ 2.575.15 However, as T increases, a second Yb3+ of the spectra at low-T suggests slow spin-lattice relax- broad line emerges from the low-T narrow line (Fig. 1) ationinvolvinglatticephononsviaspin-orbitcoupling;12 and its g-value decreases, reaching g = 2.54(1) at our and v) at low-T the Yb3+ ions behave as an adiabatic highest-T (≃ 45 K). Therefore, these two resonances spin system allowing the formation of Einstein oscilla- should be associated to two coexisting Yb3+ sites with tors inside the (Fe2P3)4-cages. different peculiarities. Figure 3a displays the T-dependence of the unsatu- In these compounds the R-ions are known to rattle rated (∼2 mW) ESR intensity for the broad and narrow at frequencies of ∼ 103 GHz6–8 which are low compared lines of the 170Yb3+(I=0) isotope of Fig. 1a. The inten- to the cage ion vibrations, but still much higher than sityofoneofthe171Yb3+(I=1/2)isotopehyperfinelines, the ESR frequencies (∼10-30 GHz). Thus, we argue normalizedbyits naturalabundance,isalsoshown. The thatthereducedhyperfineconstantforthehomogeneous inset shows the data in a log scale. From their rela- narrow line spectra results from a motional narrowing 4 mechanism22 ofon-center Yb3+ ionsrattlinginthe rigid V. CONCLUSIONS oversized(φ≃5 Å)(Fe2P3)4-cages.9 Inthe extreme mo- tional narrowing regime23 a rattling frequency of ∼ 1 In summary, our ESR results have shown that the GHz willreducein∼20%the hyperfine constant. More- over,thehyperfinestructureintheinhomogeneousbroad origin for the large g-shift of the Yb3+ KDGS, rela- tive to that in O symmetry (y = 0), may be associ- line spectrawasnotobserved,suggestingthatadistribu- h tion of Yb3+ ions are rattling at higher frequencies and ated to the B6t(O62 − O66) term in HCFZ. Coexisting narrow and broad Yb3+ resonances were observed and producing an even larger reduction of the hyperfine con- associated, respectively, to a low occupation (∼5%) of stant. Again,inthe extrememotionalnarrowingregime, a rattling frequency &15 GHz will reduce in 90 to 95% on-center Yb3+ rattling ions (∼1 GHz) and to a highly occupied (∼95%) T-dependent distribution of off-center the hyperfine splitting, and the ESR spectra would look rattling Yb3+ ions (&15 GHz). These assignments were like the observed single broad line of ∆H ≃ 30-40 Oe. basedon: i) themuchhigherexpectedYb3+ rattlingfre- Weshouldmentionthatthereportedrattlingamplitudes are . 0.1 Å.24 Hence, the broad line T-dependent shift quencies than the microwave frequency used in the ESR experiments6–8 and; ii) on the observedreduction of the andbroadeningismostlikelytheresultofaT-dependent distribution of Yb3+ ions rattling at higher frequencies hyperfine constant for the on-center Yb3+ ions and the absence of hyperfine structure in the spectra of the off- inside the (Fe2P3)4-cages. Thus, we associate the ho- center Yb3+ ionswhichwereattributedtomotionalnar- mogeneous narrow line, corresponding to the low occu- rowing effects.22,23 Although our findings relied on the pied sites, to on-center (g = 2.575) of ≃1 GHz rattling Yb3+ ionsat(x=0.523,y=0.082),whereasthe inhomoge- Yb3+ ESR results to witness the Yb3+ rattling mode, they suggest that the R ions in other skutterudites and neous broad line, corresponding to the highly occupied sites with lower g-values, to a distribution of &15 GHz clathrate compounds may be also rattling in an analo- rattling Yb3+ ions. Since this broad line is an inhomo- gous form as long as they are inside an oversized cage. However,it maynot be alwaysobservablein anESR ex- geneous resonance(distribution of g-values)andthe rat- periment. We believe that the evidence for predominant tling frequency is of the order of or higher than the ESR frequency, the rattling Yb3+ ions responsible for these off-center rattling Yb3+ ions in these skuterudites is a result that could justify the existence of Einstein oscilla- spectrashouldbe spending moretime at off-center posi- tors and help to understand the low thermal conductiv- tions in the over-size cage. ity and the strongly correlated phenomena exhibited by Sincenoemergingsecondresonancewasobservedfrom the low-T ESR spectra of Er3+ and Dy3+ ions diluted these type of materials. in Ce1−xRxFe4P12,15 it is possible that for Yb3+, with smaller ionic radius than that of Er3+ and Dy3+, larger voided excursion space may be available for the Yb3+ ions inside the (Fe2P3)4-cages which may further favor theYb3+ torattle. AT-dependentdistributionofCFPs, VI. ACKNOWLEDGMENTS that inthis T symmetry allowsfor acontinuous change h on g 25, or even a distribution of new 2nd order CFPs eff in H associated to the off-center Yb3+ sites may be We thank FAPESP-SP and CNPq for financial sup- CFZ also a plausible reasonfor the observed T-dependence of port. PS is supported by DOE grant No. DE-FG02- the inhomogeneous broad line. 98ER45707. 1 G.J.SnyderandE.S.Toberer,NatureMat.7,105(2008). 2008. 2 G. Nolas and G.Slack, American Scientist 89, 136 (2001). 9 W. Jeitschko and D. Braun, Acta Crystallogr. B33, 3401 3 See,forexample,theSkutterudite2007conferenceproceed- (1977). ings at http://jpsj.ipap.jp/journal/JPSJS-77SA.html 10 T.Inui,Y.TanabeandY.Onodera,Group Theory andits 4 T. Yanagisawa, P-C. Ho, W. M. Yuhasz,M. B. Maple, Y. Applications in Physics (Springer, Berlin, 1966). Yasumoto,H.Watanabe,Y.NemotoandT.Goto,J.Phys. 11 K. Takegahara, H. Harima and A. Yanase, J. Phys. Soc. Soc. Jpn. 77 074607 (2008).. Japan, 70, 1190 (2001). 5 T. Goto, Y.Nemoto, K.Sakai,T. Yamaguchi,M.Akatsu, 12 A. Abragam and B. Bleaney, EPR of Transition Ions T. Yanagisawa, H. Hazama, K. Onuki, H. Sugawara and (Clarendon Press, Oxford, 1970). H. Sato, Phys.Rev. B69, 180511(R) (2004). 13 Y. Nakanishi, T. Kumagai, M. Oikawa, T. Tanizawa, and 6 C. H. Lee, I. Hase, H. Sugawara, H. Yoshizawa and H. M. Yoshizawa, H. Sugawara and H. Sato, Phys. Rev. B Sato, J. Phys.Soc. Jpn.75 123602 (2006). 75, 134411 (2007); T. Yanagisawa et al., J. Magn. Magn. 7 C. B. Vining, NatureMat. 7, 765 (2008). Mater., 310, 223 (2007); W. M. Yuhasz, N. A. Frederick, 8 M. M. Koza, M. R. Johnson, R. Viennois, H Mutka, L. P.-C. Ho, N. P. Butch, B. J. Taylor, T. A. Sayles, M. B. GirardandD.Ravot,NatureMaterials, Articles,Advance Maple,J.B.Betts,A.H.Lacerda,P.Rogl,andG.Giester, online Publication, doi: 10.1038/nmat 2260, 31 August Phys. Rev. B 71, 104402, (2005); C. R. Rotundu, K. In- 5 gersent,andB.Andraka,Phys.Rev.B75,104504 (2007); Tao, Phys.Rev.B 5, 2735 (1972). W M Yuhasz, P-C Ho, T A Sayles, T Yanagisawa, N A 19 J. Korringa, Physica 16, 601 (1950); H. Hasegawa, Prog. Frederick, M B Maple, P Rogl and G Giester, J. Phys.: Theor. Phys.(Kyoto) 21, 1093 (1959). Cond.Matt.,19,076212(2007);C..PYang,H.Wangand 20 C. Rettori, D. Davidov, R. Orbach, E. P. Chock and B. K. Iwasa, Appl.Phys.Lett. 89, 082508, (2006). Ricks, Phys.Rev.B 7, 1 (1973). 14 E.A.Goremychkin,R.Osborn,E.D.Bauer,M.B.Maple, 21 K.R.Lea,M.J.M.LeaskandW.P.Wolf,J.Phys.Chem. N. A. Frederick, W. M. Yuhasz, F. M. Woodward, and J. Solids 23, 1381 (1962). W. Lynn,Phys.Rev.Lett. 93, 157003 (2004). 22 H. A. Farach, E. F. Strother and C. P. Poole, J.Phys. 15 D.J.Garcia,F.A.Garcia,J.G.S.Duque,P.G.Pagliuso, Chem. Solids 31, 1491 (1970). C. Rettori, P. Schlottmann, M. S. Torikaschvili and S. B. 23 P. W. Anderson,J. Phys. Soc. of Japan 9, 316 (1954). Oseroff, Phys.Rev.B 78, 174428 (2008). 24 D. Cao, F. Bridges, P. Chesler, S. Bushart, E. D. Bauer 16 G.P.Meisner,M.S.Torikachvili,K.N.Yang,M.B.Maple and M. B. Maple, Phys.Rev.B 70, 094109 (2004). and R.P. Guertin, J. Appl.Phys. 57, 3073 (1985). 25 F.A.Garcia,D.J.Garcia,J.G.S.Duque,P.G.Pagliuso, 17 G. Feher and A. F. Kip, Phys. Rev. 98, 337 (1955); F. J. C. Rettori, P. Schlottmann, M. S. Torikachvili and S. B. Dyson, Phys. Rev. 98, 349 (1955); G. E. Pake and E. M. Oseroff, Accepted in Phys.B 2009. Purcell, Phys.Rev. 74, 1184 (1948). 18 E. P. Chock,D. Davidov,R.Orbach, C. Rettoriand L. J.