Hyperfine lnieraclions 60 (1990) 991-994 991

MAGNETIC HYPERFINE FIELD AND ELECTRIC FIELD GRADIENT FOR '"Cd AT THE Kr/Fe INTERFACE

P. DECOSTER, G. DE DONCKER, H. PATTYN and M. ROTS Instituut voor Kern- en Stralingsfysica, K.U. Leaven, Celeslijnenlaan 200 D, B-3030 Leaven, Belgium

Polycrystalline iron foils implanted to high l<r (loses we,e doped with "qn and the magnetic hyperfine field ,as well as the electric field gradient measured. From the present TDPAC experiments we observed a substantial fraction of probes in a "defect" site, with its hyperfine parameters AIBhr I = 6.9% and

Vzz = 1.12 10'TV/cm 2, in close similarity with those expected for the l(r/Fe interface.

In recent years different investigations dealt with the precipitation or bul)ble formation of rare gases in metals produced either by nuclear reaction or ion implantation. Nowadays it is rather well established that the bubbles are overpressurized and in solid phase, epitaxially aligned with the metal matrix/1,2/ . In the present study we report on preliminary results on a search for hyperfi,m inl.eractioll prol)es h)cate(I at the interface of a ,'are gas inclusion in iron. In view of the interest in hyl)erfine interaction i)arameters at. surfaces, this approacll may be most valuable.

A polycrystalline Ire-foil l)re-implanted with krypton to a dose of 2 1016 ions/cm~ was subsequently doped with ~Uln (dose 4 10 ~a ions/era '~) and post-implanted with the same initial amount of kryl)ton. The iml)lantation energies were matched for an ol)timal depth profile overlap and the implantation was performed at room temperature. TDPAC exl)eriments, in a conventional four detector set-up, were done on the sampl~ as-implanted ms well ,as annealed for 30 rain. at temperatures Ul) to 500(_]. Along with zero magnetic field 0!

-01

O2

C~ NO

.01

O2

oS

O0

�9 as , ~ p l o n l e d

5O IOO 150

Time [n~ , )

Fig. 1. Time spectra ['or Kr-- implanted Fe-foils as a function of annealing temperature TA, observer,1

with the mln-probe.

mcasurcmcnts, we pcrformcd cxpcrimcnts in an external field, simultaneously ill tile transversal and longitudinal detector configuration. The latter configuration was inten- ded to decouple the electric quadrupole interaction from the magnetic interaction. Earlier data on rain implanta- tion in pure iron, already indicated that as-implanted the substitutional fraction is almost negligible. As a first result we mention the observation of a well-defined spin precession pattern attributed to ahnost 50% of the in([ium probes, in an ~- implanted sample containing the high Kr--dose. Moreover annealing does not change this fraction dr,'~tically, in contrast with the behaviot, r for In implan- ted pure iron saml)[es. Some typical PAC spectra obtained after annealing, are shown in fig. 1. A re~onable fit to the data results in a magnetic interaction frequency w B=555.8(1)Me of 6=1 .23(3)%. The observed frac-

t ion of probes in this hyperfine interaction environment is

57% as-implanted and increases towards 70% after annea- ling of 30 minutes at 100C. 'l'his fraction remains ahnost unchanged during annealing below 350C. At the annealing temperature T A = 400C a completely different PAC

pattern w~ observed, reflecting a broadly distributed hyperfine interaction with the parameters: w B = 175(2)Me

and 6w B = 21(2)Mc. The spin precession pattern dis-

appears at T A = 500C and the indium activity escapes

fi'om the foil at T A = 600C.

�9 J.C. Baltzer A.G., Scientific Publishing Company

992 P. Decoster et al., MHF field and EFG for mCd at the Kr/Fe interface

A n i l n p r o w u l fit. I.l~ <l;d+a set. c tmh l ho ~d~l.ained hy a~'r ` I.Wt~ prohe sil.cs fin. samplc~ annealed below T A = 300C. The first fraction of probe atoms experiences a

well--defined interaction frequency of WB= 555.3(2)Mrad/s and width 6= 2.8(1.1)Mrad/s,

while the other is somewhat ill-defined with a broadened width. Obviously a combined magnetic and electric interaction should be considered. Therefore, we me~ured simultaneously the spin precession under an external magnetic field of Bext = 0.1 T in both the transversal and longitudinal detector configuration. In the latter case the external field is applied along one of the detectors and theremre the Larmor precession of the spins around the magnetic h f i -~ is results in an unperturbed TDPAC spectrum. In a polycrystalline sample, however, a possible electric field gradient is randomly oriented relative to the external field axis and remains therefore reflected in the precession pattern. In fig.2 we show the result for the transversal and longitudinal geometry.

Notice that the latter has been 02-

Of-

r f 00-

-Of-

o ~ o TRANSVERSAL

2; s ob s'o ,;o ~io ,to Time (ns l

LONGITUDINA

. 0 5 -

- . 0 5 -

r ime(ns}

Fig. 2. Spin precessmn pattern for Kr~loped ge in an external magnetic field of 0.IT in the transversal and longitudinal detector geometry.

000- IX:

measured double-sided and the corresponding perturbation pattern, shown unh~lded, corresponds to tile interdetecLor angle of 900. This spectrum can be nicely fitted with two quadrupole sites: ft= 0.64 with Wo= 3.53(5) Mrad/s and f2= 0.36 with

WQ= 1.1(1) Mrad/s, meaning however

that the magnetic interaction is negligible. In view of the spectum obtained in the transversal geometry this is an unexpected result, because a substantial fiaction of probes experience a large magnetic hyperfine field. These magnetic sites contribute to the Iongitndinal spectrum ,as a constant anisotropy, whenever the ratio y=wB/W Q is large. Therefore the h fraction may correspond to probes on such "quasi-substitutional" lattice positions, where they contribute to the transversal spect, rum by a well-defined precession pattern.The time dependent anisotropy flmction for a combined interaction, with the electric field gradient, randomly oriented in respect to the magnetic interaction axis, can not be written in closed form/3/ . llowever, simulations with the parameters y and WQ illustrate the

spectrum shape to be expected in the longitudinal geometry

llere we did not perform data analysis along those lines, but derive fi'or, comparison with those s imulat ions/3/an apl)roximate value for tile y ratio. Then the ft fraction should be associated to probe sites experiencing a combined interaction with y_< 1, in order to produce the observed spectrum shape of the longitudinal measurement.As a consequence those probes shoukl then experience a reduced magnetic hyperfine field, but remain almost unobservable in the transversal spectrum, due to incoherent superposition of the different frequency components.

As mentioned above for the anneal sequence, the transversal spectrum was analyzed assuming two probe sites, both experiencing a combined hyperfine interaction, but with a

P. Decoster et al., MHF field and EFG fat" mCd at the Kr/Fe interface 993

.'r y 0"al.io, .y~' I() ":~ ~L~ i ,d ic; t l . ( .d I~y I.he f.s fracl.ion~ connt.;tined iu I.h(~ hmgiI.LJdiu-'tl ..q)ec:l.uurJL Because of tile transversal geometry, only even fi'equency harmonics of tile magnetic interaction are observed, which are shifted due to the presence of tile quadrupole interaction. The latter being small relative to the former, in first order approximation we expect frequency components of the type and close to:

2w B:~6wQ and 2w B~18wQ with negligible contributions centered around w B. Indeed, when a(hnitting more than one

probe site, the data could be fitted nicely in this model, with the result:

- a "quasi-substitutional" site (fraction 65%) with wB= 554.7(1)Mrad/s,

WQ= 0.42(14)Mrad/s and 5-- 9.2(1.1)mrad/s

- a "defect site" (fraction 20%) with WB=519.9(2.2)Mrad/s, WQ= 2.7(2)Mrad/s and

6= 10(5)Mrad/s - a n "undefined site" (fraction 15%) with wB= 130(2)Mrad/s alld 6= 41(7)Mrad/s.

In addition almost 40% of tile probes remain undetected, most probably due to sites where the interaction ratio y is anou,d one. Then the correlation auisotropy (lies out quickly due to destructive interference among the many frequency counl)oneuts.

After correction for the external field we obtain for tile qua.si-substitutional site (s) a

hf-field shift A[I ]hr[=-6 .8 10 -'~ anJd <V~,z>= 0.13('l)10~ '2. I,'or the defect, site (d) however the hf-field shift equals A I I]hf[ =-6.89(4)% or -2 .6T while

<Vzz>= 0.86(6)I0'TV/cm ~. The measured electric field gradient vaJues are mean values because of their random orientation relative to Bext. We note that eft values for tile lie-decorated vacancy clusters in fee metals are an order of magnitude larger /4/. Such decorated configurations may eventually be responsible for tile unobserved fraction mentioned above, which then should correspond to the fi fi'action seen in the longitudinal spectrum. After annealing at 320C tile defect site becomes almost undetectable, while tile other sites have the characteristics : WB= 553.4(8)Mrad/s, WQ= 0 and ~= 10(5)Mrad/s

together with wB= 111 (4)Mrad]s and ~= 70(12)Mrad/s.

Tile present results can be compared with CEIVlS data /5 / at the Fe (110) surface, where for a clean surface A Bhf = - 2 . 0 3 T (first layer) o r - 0 . 1 1 T (second layer) has beet found along with* Vzz= 8.10JTV/cm ~. Although this comparison s rather good, whether or not the hf-field at Fe surfaces has been measured here remains to be verified.

We like to argue that the "defect site" carl be identified ,as probes located at tile interface of the krypton inclusion. Indeed, a remarkable high fraction of probes are implanted at well-defined sites, where they experience a hyllerfine field comparable with the one at substitutional sites in pure iron. This ollservatio, suggt~ql,s that indeed Kr-bubl)le formation liappens, thereby consuming the vacancies produced in the indium implantation. As a conseqnence only a minor fraction of probes, eventually those responsible for the undefined site, will be associated with vacancy complexes due to radiation damage. Kr-decoratcd vacancy clusters around the In-probe should have an efg v a l u e / 4 / a l l ordcJ" of magnitude larger than we me~ured hcrc for the "defect site". Furthermore, the quadrupolc interaction strength as well as tile hyl)erfine field shift, correspond quite well witll tile vahle observed at the Fc (110) surface in CEMS-cxperimcnts/5/ .

As a final remark we suggest that both fractions showing a well-defined precession in the TDPAC pattern (quasi-substitutional and defect site) eventually may be interpreted originating both from tile Fe/Kr-intcrface, the latter at tile sur[acc the [ormer in the outermost Fe layers. Indeed it is known from experiments/6/ as well as recent cl'g calculations/7/, that the cubic charge symmetry is restored some Lwo layers bencaLh the metal surface. Calculations o[ the magnetic hypcrfinc field/8/, on the other baud, seem to predict substantia[ dependence on the symmetry of the surface. No characterization of tile

*) the value quoted ia / ,5/ has bee,, corrected for the quadrul ,ole mome.t.

994 P. Decoster el aL, MHFJIeld and EFG for mCd at the KtVFe interface

surface type , probed here at the Kr-inclusion, was available ,as yet and therefore quantitative comparison with theoretical estimates may be somewhat premature. Anyhow the present results do not suffer from uncertainties due to residual gas adsorption, the main limitation in earlier work on "qu~i--clean" surfaces.

The present experiments are the first results on impurity hyperfine field parameters at the surface of a magnetic metal matrix. Our approach seems promising to the study of surface phenomena, without using expensive apparatus to produce clean surfaces. Indeed one only needs an ion implanter to create inert gas bubbles into the metal matrix and a probe with an affinity to occupy sites at the interface of those inclusions. The insolubility of indium in iron may well be responsible for the relatively large fraction of interface positions observed in the present experiment.

References.

I. A. yore Velde, J. Fink, Th. Miiller-II(~inze.rling, J. I'lfiger, B. Scheeren mId G. l,iuker, Phys.lLev.Lett. 83 (1984) 922

2. R.. C. Birtcher and W. J~'~ger, Nucl.lnstr.Meth. B15 (1886) 3. F. Pleiter, A. R. Are,~ds and II. G. Devare, llyp.lnt.3 (19;/7) 87 4. M. Deicher, G. Grfibel, W. Reiner and Th. Wiche,'t, Ilyp.hlt.15/16(1983)467 5. M. Przybylski and U. Gradmmm, I'hys.Rev Lett. 59 (1987) ! 182 6. J. Kor~chi, Ilyp.lnt. 40 (1988) 89, and ,'eferences therein. 7. B. Lindgren, Hyp.Int. 34 (1987) 217 8. S. Ohnishi, M. Weinert and A. J. Freeman , Phys.Rev.B30, (1984) 36