PhysicsLettersA172 (1993) 291—295 PHYSICSLETTERS ANorth-Holland

Time differentialperturbedangularcorrelationstudyof impurity distribution in dilute NiFe andNiPd alloys

P. DecosterandM. RotsInstituutvoorKern- enStralingsfysika,KatholiekeUniversiteitLeuven.B-3001Leuven,Belgium

Received21 September1992; revisedmanuscriptreceived3 November1992; acceptedforpublication4 November1992Communicatedby J. Flouquet

Impurity interactionsin diluteternaryalloysN Fe(In)andNiPd(In) werestudiedby thetimedifferential perturbedangularcorrelationmethod.A strongInPd attractionandaweakInFerepulsionweredeterminedin anickelhost.TheIn/Cdprobesensedmagnetichyperfinefield shiftscausedby Fe impuritiesin threenearestneighbourshellsaroundtheprobe,while no magnetic-momentperturbationwasfounddueto Pdimpurity atomsin a nickel host.

1. Introduction We appliedTDPAC to measureboth the efg andthe magnetichyperfine field distribution in dilute

The perturbedangularcorrelation (PAC) tech- NiFe andNiPd alloys,as function of impurity con-niquedetectsimpuritiesneighbouringa radioactive centration,usingthe diamagneticprobe“In. De-(here“In) probeatom in a dilute cubic ferromag- tails onthemethodanddataanalysisareto befoundneticnickel alloy via the inducedelectric field gra- in ref. [1]. The alloys werepreparedby rf-meltingdient (efg) andmagnetichyperfinefield (hif) shifts. highpurityconstituentsinanargonatmosphere,afterOur recent interestwas focusedon Ni alloys with evacuationto better than 1 0~Ton. After cold-early3d-transitionelementssuchasCr, V, etc. [1], rolling the sample,the “In activity was deposited,theso-called“weak” ferromagnets[2—4].Thoseearly the foil remelted,followedby annealingunderhy-3d-impuritiesinducelong rangemagnetizationper- drogenat 900°Cfor at leasttwo hours.Thesamplesturbation, although the impurity momentsthem- were cooleddownto room temperatureslowly. Forselvesare small. The largestcontribution to the this experimentnickelalloyscontaining1.0, 1.5, 2.5changein totalmagneticmoment,measuredthrough and5.0at%Feand0.2, 0.3, 0.7, 0.8, 1.3, 1.5 and2.5the hyperfinefield at the “Cd probe, comesfrom at% Pd were studied.a strongreductionof hostmomentsin a sphereoffour to five shellsaroundthe impurity. Thistype ofalloy wasfurthercharacterizedby astrongrepulsive 2. ResultsIn—impurity interaction.

On the otherhand,NiFe alloys fall on the main Two contrastingtypes of hyperfine field distri-branchof theSlater—Paulingcurve.Theyarethe so- butionshavebeenobserved,both in the paramag-called“strong” ferromagnets.For Fe impurities the netic and ferromagneticphase,dependingon thelocal magneticmomentexceedsthe hostmoments solute.andthey inducenearlyno changeor magnetization In the paramagneticphase,abovethe Curie tem-disturbanceon neighbouringshells. The changeof peratureT~1= 360°C and up to an impurity con-totalmomentin thealloy issolelydeterminedby the centrationof 5.0at%, the 3d-impurityFe causesaimpurity defect moment ~u= /2Fe— /1Ni• Finally, weak electric field gradientat the “Cd probe.TheNiPdalloysareuniquesincethesealloysare neither datawerefitted assumingthat a fraction of probesstrongnor weak ferromagnets[5]. j experiencea Gaussiandistributionö aroundzero

0375-9601 1931$ 06.00© 1993ElsevierSciencePublishersB.V. All rights reserved. 291

Volume 172,number4 PHYSICSLETTERSA 4 January1993

quadrupoleinteractionstrength,while the remain- Table 1

ingfractionfocontributea constanttotheR ( I) func- Parametersextractedfrom fitting of R( t) spectrafor differentconcentrationsof Fe impurity, c, in a Ni matrix (ô(eQV~/h):

tion. This two-fractionmodelsuggests:while the In absolutedistribution on quadrupole interaction strength;J~:probeatomsarelocatedat substitutionallattice sites, probability of jth site) measuredat “Cd at temperature

a fractionf of them senseFe impurities in various T=400(l0)°C.

distantneighbouringshellsanda fractionf0 of theprobesaresituatedin a cubic undisturbedenviron- c .f 5(eQV~/h)

(at%) (%) (%) (MHz)ment.A typical result for the experimentalanisot -_____________________________________________ropy factorR(t) is shownin fig. la. Thefitting re- 1.0 60(3) 40(3) 3.8(6)

sults are displayedin table 1, as a function of Fe 1.5 59(7) 41(7) 3.1(5)5.0° — 100(4) 8.0(6)concentration. _______________________________________________

From the NiFe phasediagram [6] a strong en- ° T=430°C.hancementof Curietemperaturewith increasingsol-uteconcentrationcanbenoted,giving at CFe= 5.0at%a T~of 430°C,asconfirmed in our measurements.Remarkablyalso for this highestconcentrationes- ForNiPdalloys,onthecontrary,fig. lb showsvery

sentiallythe samespectrumwasobservedasappli- clearly that the “Cd nuclei are surroundedby Pdcablefor low concentrations.This observationand impuritiesandexperienceseveralsharpquadrupolethe Gaussiandistribution aroundzero interaction interactionfrequencies.In additionalsoherepartofstrengthclearly reflectsthefactthat Fe is nota charge the probes(J~)feel no electric field gradient.Tabledisturbancein the nickel lattice. 2 displaysthe parameters,as a function of Pd im-

purity concentration,obtainedfor the characteriza-tion of the different atomic configurations:quad-rupoleinteractionstrengthseQV~)/h,corresponding

~ fractionsJ andasymmetryparameters~,. The dis-020 tributionsô, in interactionstrengthat eachsite i are

foundtobe nearlyzero.Basedon argumentsrelated-0.25 with the probabilityandsymmetryof the particular

atomic configurations,the quadrupoleinteraction

-0.30 strength eQV~/h 14(1) MHz first appearing

could be attributedto first nearestneighbourInPdpairs,while eQV~/h~36(1)MHz and~2=0.6 is

-4--. ___________________________________ associatedwith two Pd atomsat first nearestneigh-

turecanbeextractedfrom inspectionof table2:thereexistsa largeenhancementin thepopulationsofboth

—0.0configurationsascomparedto theoneexpectedfroma randomimpurity distribution.Thisfact evidences

01 bour distanceof the probe [7]. An interestingfea-—0.1

a strong attraction of Pd atoms towardsthe firstnearestneighbourshellaroundtheindiumprobenu-

—0.2cleus.In ref. [7] we estimatedfrom the absenceofany temperaturedependencein the fractionsJ an

—0.3 _______________________________________________I I I I energyof the InPdpair attractionof at least — 100

0 100 200 300 1.00 meV.Time(ns) Thedifferentbehaviourof thePdandFe impurity

Fig. I. TDPAC quadrupolespectrafor “Cd in (a) Ni98 5Fe,5 aboveT~is alsoobservedin theferromagneticphase.and (b) Ni987Pd,3at400°C(site 1: eQV~/h=l4(1) MHz; LocalisedFemomentsin the nearvicinity of thesite2: eQV~~/h=36(1)MHz, ~= 0.6). diamagneticIn probewill resultin well-resolvedsat-

292

Volume 172,number4 PHYSICSLETTERSA 4 January1993

Table2Parametersextractedfrom fitting ofR(t) spectrafor different concentrationsof Pdimpurity, c, in a Ni matrix (eQV~/h:quadrupoleinteractionstrength;J

1: absolutedistributionequalszero;~, asymmetryparameter,f:probabilityof ith site) measuredat “Cd attem-peratureT=400(l0)°C.

c f, eQV~,’~/h f2 eQV~/h C2 f3 eQV~/h(at%) (%) (%) (MHz) (%) (MHz) (%) (MHz)

0.2 88.3(8) 11.7(8) 14(1) — — — — —

0.3 83.2(4) 16.7(3) 15.7(3) — — — — —

0.7 73(2) 20(2) 14(1) 7(2) 36.3(6) 0.59 — —

0.8 58(1) 36(1) 13.5(4) 7(2) 36.3(6) 0.6 — —

1.3 31.9(6) 50(1) 14.1(3) 19(1) 36.4(2) 0.59 — —

1.5 36.8(9) 52(2) 14.1(2) 12(2) 34.0(4) 0.6 — —

2.5 15(1) 53(4) 16.3(4) 20(3) 35.8(8) 0.6 12(4) 29.8(1)

~i~i~ ~ .: ~.0 100 200 300 400 500 ‘0 20 30 40 50Time(ns) Frequency (MHz)Fig. 2. Magnetichyperfinefield distributionofanannealed“InNi985Fe,~5alloy in timeandfrequencydomain.

ellite hyperfinefields, ascanbe noticedfrom fig. 2 suggestedalso from the Miedema—Królasmodelboth in time andfrequencydomain.Themain fre- [8,9]. The measuredhyperfinefield shift 4,’ due toquenciesandtheir populations,resultingfrom least oneFe atom in the ith shell aroundthe probe,rel-squaresfits of the time spectra,are summarizedin ativeto thenegativevaluein purenickel, equalsre-table3. Identificationof the different sites,charac- spectively+4.3(9), — 1.59(9) and +0.4(2) T forterizedby v, canbeobtainedby applyinga binomial i= 1, 2 and3. Thetransferredhff at Cd (whichcar-distribution pershell (fcc coordinationnumber12, ries no local moment)in a nickel matrix canbe ex-6, 24, 12, 24, respectively).Thefrequencyv0 cor- pressedbyrespondsto the purenickel hyperfinefield. The fre- 11T — AKquencyv, representsoneFeatomin thesecondnear- hf — 4 ‘~j aPi,est neighbour shell around In, while v2 and v3correspondto one, respectivelytwo, impuritiesat with Ag1 the conductionelectron polarisationcon-third nearestneighbourdistanceandv4 to one Fe tributionpermagneticmomentdueto a nickel atomatombeingfirst nearestneighbourto theprobe.The in the ith neighbourshell of the probewith coordi-agreementbetweenthe experimentalpopulationfor nationnumbern. TheAll1 canbe derivedfrom theeachsatellite frequency,A~, andthe corresponding hy~ethnefield satellitepositions,the hif shift A,’,binomial distributionper shell,A ~, indicatesa ran- knowingthecorrespondingimpurity configurations.dom Fe distribution in nickel, except for the first Therelationshell. The emptyingof the first nearestneighbour 41 = Ag. ( fL . —

shellcanbeexplainedby a weakInFerepulsion,as Ni Fe

293

Volume172, number4 PHYSICSLETTERSA 4 January1993

Table 3 tively two Pd atomsat first nearestneighbourdis-Derivedparametersfor annealedNiFe alloys asa function ofconcentration(E experimental;B binomial per coordination tancefrom theprobeare,however,notwell resolvedshell), from the main purenickel frequency[7]. Thislack

ofclearsatellitestructureis unexpectedsincea smallc (at.%) positive moment for Pd in nickel was found by

Crangleand Parson [11], Sadron [12] and Ste-1.0 1.0 1.5

_____________________________________________ fanou and coworkers [5]. Since local repulsionbe-<8j’> (MHz) 0.06(1) 0,13(1) 0.10(3) tween In and Pd is excludedby our quadrupolein-v0 (MHz) 15.59(6) 15.68(6) 15.44(2) teraction results, other explanations have to be found.A~(%) 69(2) 72(2) 58(4)A~(%) - - - Pdmayhavea localisedmomentequalto thenickelv, 11.9(2) 11.9(3) 11.7(7) moment [13] or a negativedefectmoment~Pd—UNi

A~ 4.9(4) 7.0(3) 9(1) on Pd existsbut is compensatedby a positivemo-A~ 6 6 8 ment disturbance at neighbouringNi sites, resulting

16.4(4) 16.5(4) 16.5(5) in no hyperfine field shifts [14].18(2) 10(1) 21(3)19 19 25

V3 17.6(2) 17.5(3) 18.0(9)8.4(7) 8.2(6) 7(2) 3. Conclusion2 2 5— 26.7(8) 25(2)

A~ - 2.2(3) 5(1) For both impurities, Fe and Pd, the indium—A~ - 11 15 impurity bindingenergycouldnotbe deducedby ap-

plying the Arrheniuslaw [15]. From concentrationdependent quadrupole interaction measurements a

resultsin conductionelectronpolarisationhif shifts strong InPd attraction is reflected in a strongof All, = —2.1(4) T/~B-atom,AH2= 0.73(9)TI~~- suppressionofthedefectfreefractionfo,while a weakatom andAH3= —0.22(9) T/u3-atom.Thosecon- repulsionfor indium—iron pairs is suggestedby theduction electronpolarisationhif shifts, positive as deviation,observedin the magneticspectra,of thewell asnegativerelativetothepurematrix value,are correspondingpopulationfrom a randomimpurityof the RKKY type [10]: oscillatory anddecreasing distribution.The impurity effectson the hfi param-with increasingprobe—impuritydistance. eterforbothtypesof solutesPdandFeis completely

Sincefirst nearestneighbourInFe pairswerede- different.Fe createsa negligibleefgat the “Cd probetected(max 5% for CFe= 1.5 at%) in the magnetic but induceswell-resolvedmagnetichyperfinefieldspectra,the absenceof a well-definedquadrupolein- shifts up to the third nearneighbourshell. Thisob-teractionfrequencyin the paramagneticphaseis jl servationis in agreementwith the fact that Fedoeslustratedoncemore. The conclusionis that Fe im- not inducechargeor magnetizationdisturbancesonpurities do not representa chargedefectand thus neighbouringatoms.On the contrary,we identifiedinduceno substantialelectricfield gradientin nickel. different first nearestneighbour Pd—In configura-On the contrary,wehavetoconsiderin themagnetic tionsvia quadrupoleinteractionmeasurements,butphase of the NiPd dilute alloys Ni99 8Pd02 and those configurationsinduceno well-resolvedmag-Ni98 7Pd,3, a combined magneticand electric hy- netic satellitehyperfinefields.Togetherwith thede-perfineinteraction,sincethe quadrupoleinteraction termination of the near neighbourimpurity inter-cannotbeneglected.Indeedthedatain fig. 3 cannot action type and the identification of their atomicbe fitted with a single frequencywith finite distri- configurations,the presentpaperreportsthe exper-bution. Only a fit including several“combined in- imental verification of a RKKY-like oscillation interaction”sites,different in the anglesfi betweenthe thecontributionsto the transferredmagnetichyper-symmetryaxesof the magnetichyperfinefield and fine field in nickel.theelectricfield gradient,givessatisfyingresults.Themagneticfrequenciescorrespondingto one,respec-

294

Volume172,number4 PHYSICSLETTERSA 4January1993

(a)0.1

0.0

I’

a:—0.1

2.43-02 a

Ii 2~3 3c~0 4c~ SdO

Time (ns)02. __________________________ ___________________________ —

.200.1.U)

I~

L~o.o.

IIIII~ .10 a-0.1. 387 ______________________________ ______________ .00 200 400 0 20 40

Time (ns) Frequency (MHz)

Fig. 3. Magnetichyperfinefield distributionfor “In in Ni98,7Pd,,3measuredatroomtemperaturein time ((a), (b)) andfrequency

domain(c). Thetimefit improvesusingacombinedhyperfineinteraction(a) insteadofthefit with asingledampedmagneticfrequency(b) asindicatedbyx

2.

References [7] P.Decosteretal.,tobepublished.[8]K.KrOlas,Phys.Lett.A85 (1981) 107.

[91A. MiedemaandAK. Niessen,CALPHAD 7 (1983)27.[1] P.Decoster,G.DeDoncker,M. RotsandA.Z. Hrynkiewicz, [l0]M.B. Stearns,Phys.Rev.B 9 (1974)2311.J. Phys.Condens.Matter 3 (1991)7575.

[11] J.CrangleandD. Parson,Proc.R. Soc.A 255 (1960)509.[2] G.G.Low andM.F. Collins, J.AppI.Phys.34 (1963)1195. [12] C.Sadron,Ann.Phys. (Paris) 17(1932)371.[3] M.F. CollinsandG.G.Low,Proc.Phys.Soc.86 (1965) 535.

[131 D.Hunter,A.S. Arrott, R.I. Grynszpan,P. Dassonwalleand[4] J.B. Comly,T.M. HoldenandG.G.Low,J.Phys.C2 (1968) P. Langlois,J.Phys. (Paris) Colloq.C 8 (1988)131.

458. [14]J.W. Cable andH.R. Child,Phys.Rev.B 1 (1970)3809.[5] D.N. Stefanou, A. Oswald,R. Zeller andP.H. Dederichs, [15]A.Z. HrynkiewiczandK. Krdlas, Phys.Rev. B 28 (1983)

Phys. Rev. B35 (1987) 6911. 1864.[6]M. Hansen, Constitution of binary alloys, 2nd Ed.

(McGraw-Hill, NewYork, 1958).

295