Hyperfine Interactions 7 (1979) 61-69 9 North-Holland Publishing Company

NUCLEAR ORIENTATION AND NMR/ON ON IMPLANTED ll4mln IN Fe AND Co W.W. LATTIMER and N.J. STONE Mullard Cryomagnetic Laboratory, Clarendon Laboratory, Parks Road, Oxford, UK Received 8 November 1978

Nuclear orientation and NMR/ON of a 14mln implanted into Fe at an energy of 80 keV and dose of 3-5 • 1014 cm-2 is reported. The zero applied field resonance frequency v0 = 203.65(6) MHz is combined with the recently determined Bhf(InFe) of 286.8(3) kG to yield ~t(114mln) = +4.658(14) nm. The nuclear spin-lattice rel~xxation time for 114mln in iron is measured to be 88(18) s at 18 mK and the applied field dependence of the NMR/ON resonance frequency gives the Knight shift for the system as -2.4(6)%. The absence of measurable nuclear orientation in similarly prepared 114mln Co sources is discussed.

1. Introduction Indium is a chemically active element with a low melting point. Consequently difficulty is to be expected in preparing dilute alloys of In isotopes in ferromagnetic metals by thermal methods. However, an 114minF_eealloy was one of the first in which Samoilov et al. [1 ] showed the existence of large magnetic hyperfine interactions (Bhf) at nuclei of dia-magnetic impurities in ferromagnetic metals. Subsequently a nuclear orientation (NO) measurement on 114inF_ee [2] and a spin-echo NMR experiment on a 2 at.% a 13,a a SinF_~ealloy [3 ] reported results in apparently good agreement indicating Bhf = 290 -+ 5 kG. This situation was disturbed when an apparent inconsistency in analysis in the NO experiment was corrected to yield Bhf = --240 -+ 10 kG [4]. The NMR experiment has also been questioned since the reported In solubility limit in iron is 0.28 at.% [5]. In a series of detailed experiments on InFe samples prepared by a variety of thermal means, Stewart and Barclay [4] showed that the observed NO anisotropics were extremely sensitive to sample heat treatment and hypothesised that this might be associated with In clustering on grain boundaries. They obtained data corresponding to an interaction strength Bhf = --300 + 10 kG, in one extremely weak sample only, all others showing smaller anisotropies. A failure to detect NMR of the oriented nuclei (NMI~/ON), in samples which showed attenuated anisotropies, tended to confirm the idea that in alloys prepared by thermal means the In 61

62

W.W. Lattimer, N.J. Stone / NO and NMR/ON on 114mIn in Fe, Co

ions were in lattice sites subject to non-uniform hyperfine interactions. In this paper measurements on 114minFe alloys prepared by ion imt~lantation are reported. Although attenuated anisotropies are found (this being an integral over all the implanted nuclei), a strong NMR/ON resonance is observed. Implanted 114minCoo samples showed zero anisotropy and this surprising result is discussed.

2. Sample preparation The samples for implantation were polycrystalline iron and cobalt foils, electropolished to remove surface oxide. The foils were soft soldered to copper strips and secured in the target position of the electromagnetic separator using conducting silver dag. Two implantations were carried out at AERE Harwell using as source material neutron irradiated In metal. In the first implantation a contaminant beam of MbO occurred at mass 114 with concentration approximately equal to the inactive In beam and a hundred times stronger than the active 114rain beam, so that source A, of total implant dose 5 • 1014ions/cm 2, was only ~0.4/~C. For the second implantation the MbO was eliminated and source B, of dose 3 • 1014ions/cm 2, was considerably stronger, approximately 1.5/~C. In both cases implantation was done with 80 keV acceleration potential. The In range at this energy is ~150 A giving a local In concentration of ~ I atomic % in sample A and 0.5 atomic % in B. After implantation the copper strips were soldered to the cold finger of a demagnetisation system with Woods metal. The implanted foils were kept under liquid nitrogen to inhibit any thermal diffusion of the implant during this process.

3. The nuclear orientation experiments The samples were cooled to temperatures of order 16 mK by thermal contact with a pill of chrome potassium alum in a standard adiabatic demagnetisation apparatus. The decay scheme of 114rain is well established (fig. 1) so that in the expression [6] for the angular distribution of gamma radiation

w(o) =

1 + Z) A . B ~ . ( c o s 0) VeVefl

the angular momentum coupling constants,A~, are known. Two complications arise in interpreting NO measurements in this system. The first is the high radioactive heating in samples of this isotope. Because the M4 192 keV isomeric transition is highly converted there are approximately equal numbers of 724 keV, 558 keV and 192 keV gammas emitted and each is accompanied by ~6 beta particles of mean energy ~800 keV, and 5 conversion electrons, a total of ~6

W.W. Lattimer, N.J. Stone/NO and NMR/ON on 114mlnin Fe, Co

15~

keV~EC

(M

4"

1283 L.,-/~'"/o 1.

r-,

r ~,~ ~1196

63

keV ~9~ _~ . o d

% ,-,

"f v _TZs In114 \ 49

~/3-

0(3 2+

558

'~988

keY

O* 4.8Ccltt4 5n114 50 Fig. 1. Decay of 114rain"

MeV absorbed in the source. This is in contrast to the decay of 60Co involved in the 6~ NO thermometer used, in which for each gamma sequence only ~140 keV is deposited in the source. This heating will tend to set up a temperature difference between the implant foil and the thermometer which would result in an apparently attenuated anisotropy-temperature curve. Since the primary aim was to make measurements by NMR/ON no particular experimental precautions were taken to minimise this source of error. However, an estimate can be made of its likely magnitude. Woods metal and soft solder are alloys of superconductors Bi, Cd, Pb and Sn in different proportions. Steyert [7] has found the thermal conductance of similar alloys, in their superconducting state and well below 1 K, to obey the relation Q / A A T = 4.7 • 106T2"74 erg s -1 cm -2 K -1 , where A is the area of the boundary. For the radioactive heat loads and areas of our samples this expression yields for a 1/~C la4min sample AT = 4.8 mK compared with AT ~ 0.1 mK for the 6~ thermometer. This corresponds to a difference of 10 K -1 at l / T = 50. However the solder joints in this work were subject to magnetic fields of order 1 kG. Lounasmaa [8] has reported the not unexpected result that, when normal, the conductance of such solder layers is much increased. The upper limit of systematic error, due to heating, in the In alloy temperatures, reported here is therefore less than 0.5 inK, or less than +2 K -1 in lIT at the lowest temperatures reached. This is only slightly greater than the statistical error in the temperature determination.' The second possible source of attenuation results from the implantation process. A fraction of the implanted ions will find themselves in regions of severe radiation

64

W. W. Lattimer, N.J. Stone/NO and NMR/ON on 114mln in Fe, Co

damage or in the surface oxide layer which developes on the foils. The nuclei of these ions will see a different hyperfine interaction which will in general attenuate the anisotropy versus temperature curve. ll4mInF_ee. The data from source A are shown in fig. 2 and are very similar to those from source B. They were taken using a Ge(Li) detector of approximately 40 cm 3 and resolution at 1 MeV of 2.5 keV. The 724 and 558 keV 7-rays show the same anisotropy both in theory and experiment so their data have been summed. All 7-rays show anisotropy attenuated compared with the full curves. These were calculated for Bhf = --290 kG,/~N = 4.65 n.m. and corrected for detector solid angle. The empirical fits (dotted) are calculated on a naive model in which 80% of the nuclei

100o\ ~

-0.6

192 I(eV

w(o) -i

+"

/

10C

j/

-0

A' -O .1

O

10 20 210 40 50 RECIPROCAL TEMPERATURE (OK-l)

60

Fig. 2. Anisotropies o f g a m m a rays from 1 1 4 m l n as a f u n c t i o n o f lIT. The fuU curves show the anisotropies predicted from the k n o w n decay sequences with B h f = - 2 9 0 kG a n d t~N(114mIn) = 4.65 n.m. The dotted curves correspond to an attenuation o f 80%.

W.W. Lattimer, N.J. Stone /N O and NMR/ON on 114mln in Fe, Co

65

experiencing the full hyperfine field and the remainder zero field. Although this almost certainly does not represent the real situation, the data are not sensitive to differences between this and more complex site distributions. The figure of 80% is lower than that found for several other soluble implants e.g. Au, Sb, Hg, Se (all >95%) although higher than for implants of recognised low solubility such as Br and

Bi [9]. 114mlnC__oo.Nuclear orientation experiments on 114mlnCo samples, prepared at the same time as the iron samples, showed W(0) - 1 = 0.996 -+ 0.006 at 18 inK, i.e. zero anisotropy within the errors of the experiment. This is the first reported measurement on the InC_~osystem, but other implantations in Co [10] have not suggested a markedly lower proportion of nuclei in substitutional lattice sites than for Fe. On the assumption that at least 50% are in substitutional sites, this result suggests an upper limit of 40 kG for the InC___oo"hyperfine field.

4. NMR/ON experiments on 114mlnF.__ee For sample A a resonance search was made using standard techniques and a clear resonance was observed at 202.2(2) MHz in an applied field of 2.58 kOe. The resonance line shape was established by stepping through the resonant frequency using a frequency modulation amplitude of 500 kHz. Carrier wave counts gave the undisturbed anisotropy and intervals between counts allowed full nuclear relaxation. The result is shown in fig. 3 and corresponds to an integrated destruction of 55 + 5% of the total anisotropy. With the stronger source B the resonance experiment was repeated and in addition the applied magnetic field dependence of the resonant frequency was investigated. The results are given in table 1. The line measured at 4.40 kOe and the field dependence of resonant frequency are shown in fig. 4. A least squares fit, extrapolated to zero applied field (with negligible demagnetisation correction since the thin sample foil was magnetised in its plane along a long axis) yields Vo = 203.65(6) MHz for the magnetic hyperfine interaction frequency. The same fit to the variation of resonant frequency with applied field gives dv --

dBapp

-

0.693(7) MHz/kG.

Nuclear spin relaxation o f l l 4mIn in Fe. Since the ll4mIn resonance line shows large inhomogeneous broadening, an unmodulated rf signal at the resonant frequency will have negligible effect on the observed anisotropy. Thus when the frequency modulation of the rf field, set at the centre of the resonance, is reduced to zero, the nuclei relax into equilibrium with the lattice. Observation of the time dependence of the anisotropy thus allows determination of the nuclear spin-lattice relaxation time T1. With the lattice temperature at 18 mK the relaxation was approximately expo-

W.W. Lattimer, N.J. S t o n e / N O and NMR/ON on 114m/n iti Fe, Co

66

I

24 % DESTRUCTION 22

of ANISOTROPY

t"

20

"'lx

16

14 10

i6

201

202

203

I:: 206MH

Fig. 3. NMR/ON o f 114mlnFe with an applied field o f 2.58 kG (source A).

nential with the time constant r = 44 + 9 s. Because the ]14mIn nuclei are in very low concentration ('-'10 - 4 at.%) they will not establish a common nuclear spin temperature and consequently the observed relaxation is not truly exponential nor is the approximate time constant r equal to the true T1 [11 ]. However for this system, at this temperature, the relation r = 89 holds to 10%, so we have T] = 88 + 18 s.

-20

.

199

199.5

204 '~ (MHz)

1

200

200.5

2,01

29,S

203 202 201 2OO 199 I98 197 196 195 194 v (MHz) 193 2 4 202

H (kOe) 6 8 10 12

Fig. 4. NMR/ON of 114mlnFe with an applied field of 4,40 kOe (source B) and plot o f resonant frequency versus applied field.

W.W. Lattimer, N.J. Stone / NO and NMR/ON on 114mln in Fe, Co

67

Table 1 Applied field dependence of resonant frequency and percentage destruction of anisotropy (integrated over the resonance) Applied field ( k O e )

Resonanceline centre ( M H z )

Integrated destruction (%)

2.20(1) 4.40(2) 8.80(4) 15.40(3)

202.19(4) 200.52(4) 197.50(6) 192.97(4)

54(3) 41(1) 28.5(2) 26.5(1)

Fitted parameters. Slope = -0.693(7) MHz/kOe, frequency intercept = 203.65(6) MHz.

5. Analysis of the NMR/ON results During the course of this work, Campbell et al. [12] made NNR experiments on an 0.2 atomic % 113,11SlnFe alloy at 4.2 K and obtained, using the accurately known moments of these isotopes, a value 286.8(3) kG for the strength of the hyperfine field. This confirms and improves the precision of the earlier NMR" measurements and furthermore the In concentration used was below the solubility limit. A choice of assumptions exists in the interpretation of the NMR/ON results. If the Knight shift were neglected, dv/dB would yield the 114rain moment directly, then vo gives a value for ghf. However the accuracy is lower than that obtained by Campbell et al. The preferred alternative is to assume that the hyperfine anomaly between 113,115in (which have virtually identical magnetic structures dominated by the g9/2 proton) and 114rain (for which also the g9/2 proton is the dominant magnetic component) is small so that the NMR value OfBhf may be used to interpret our result. Estimates suggest any anomaly 113,SA114m will be <~0.2%. Hence, from vo we obtain g(114min ) = +4.658(5) n.m. Allowing for a possible 0.2% hyperfine anomaly gives the result /2(114mIn) = +4.658(14) n.m. which replaces the earlier value [13] of 4.7(1) n.m. This moment is then taken to determine the Knight shift using dv/dB = (1 + K) #lib, with the result K = -2.4(6)%.

6. Discussion 6.1. 114mln nuclear moment 114min is an odd-odd nucleus with proton (lg9/2) 9 and neutron (3Sl/2) 1, coupled to 5+. Simple Lande coupling gives gI = 0.9gp + 0.1g n. The Schmidt limit g factors predict a moment of +4.89 n.m. An empirical proton g-factor is readily available

68

W.W. Lattimer, N.J. Stone/NO and NMR/ON on

114m/-n

in Fe, Co

from the odd-A In isotopes in which the 9+ state shows a uniformity to 0.2% from 1~ to 11Sin with gp = +1.230(2). The 3Sl/2 neutron state is found in Sn isotopes with gn rising steadily from -1.76 for 113Sn to -2.09 for 119Sn" Taking gn = -1.84 from 11SSn leads to the value +4.62(7) for/~(114mln). This excellent agreement with experiment is expected since the high spin proton state is relatively pure and the result is insensitive to details of the neutron state. 6.2. Resonance lines and source preparation The integrated destructions and linewidths given in table 1 are well within the normal range for other samples prepared by both thermal and implantation methods. The general increase in linewidth and decrease in % destruction of anisotropy with increasing applied field Bapp, are also in agreement with other studies, the latter being a result of the reduction in the rf field enhancement factor (1 + Bhf/Bapp) and the former possibly being caused by a spread in demagnetisation effects associated with local surface irregularities in the foil sample. Thus implanted InFe samples, without annealing, show NMR/ON behaviour closely in line with other systems. This result tends to confirm the hypothesis of Stewart and Barclay [4] that their source-dependent anisotropies and failure to observe resonance were the result of the thermal sample preparation technique used, either in gettering other impurities or by agglomeration of the In. 6.3. Relaxation time and Knight shift For dilute impurities in ferromagnetic metals there exists an empirical relation T1TT~B2f = constant where, for Fe, the constant is 2 • 1018 s -1 K -1 [11]. The experimental value of T1 = 88 -+ 18 s for 114minF._~ e at 18 mK agrees well with the value of 73 s predicted by the relation. Systematic variation of the Knight shift with impurity is less well established or understood, however the result -2.4(6)% is within the normal range for dilute Fe host alloys [14] and gives no support to the much larger values of K reported on certain other systems. 6.4. InCo hyperfine interaction On the basis of simple systematics a field of order - 1 5 0 -+ 20 kG was expected for InC__o.However for both the 4s-4p series and the 5s-5p series of impurities in ferromagnets Bhf shows a change of sign in the centre of the series, this being interpreted as the result of competition between a negative conduction electron polarisation (cep) term and an increasing positive term possibly associated with overlap of the impurity with its magnetic neighbours. In the region of the sign change the resultant field is the fine balance of these contributions as is shown by the anomalous temperature dependence Of Bhf reported for GaC_o_o,GeC0_o(4p 1 and 4p 2) [15] and

W.W. La ttimer, N.J. Stone / NO and NMR/ON on 114rain in Fe, Co

69

for SnC__oo(5p2), in which Bef f reverses near 800 K [16]. With these considerations, investigation o f the InCo (5p 1) system assumes greater interest and the present surprisingly low limit o f [Bhf[ < 40 kG merits further investigation. This work was supported by a grant from the Science Research Council. We are grateful to Mr. W. Temple o f AERE Harwell for his assistance in making the ion-implanted samples.

References [i] B.N. Samoilov, V.V. Skylarevskii and E.P. Stepanov, JETP (Soy. Phys.) 11 (1960) 261. [2] R.J. Holliday, D.A. Shirley and N.J. Stone, Phys. Rev. 143 (1966) 130. [3] M. Kontani and J. Itoh, J. Phys. Soc. Jap. 22 (1967) 345. [4] G.A. Stewart and J.A. Barclay, Phys. Stat. Sol. (a) 38 (1976) 533. [5] C. Dasarathy, Z. Metall. 63 (1972) 209. [6] K.S. Krane, Nuclear Data Tables A l l (1973) 407. [7] W.A. Steyert, Rev. Sci. Inst. 38 (1967) 964. [8] O.V. Lounasmaa, Experimental principles and methods below 1 K (Acad. Press, 1974) p. 225. [9] N.J. Stone, Hyperfine Interactions 2 (1976) 45 and references therein. [10] P.T. Callaghan, N.J. Stone and B.G. Turrell, Phys. Rev. B10 (1974) 1075. [11] N.J. Stone, Hyperfine interactions in excited nuclei, vol. 1, ed. G. Goldring and R. Kalish (1971) p. 237. [12] Le Dong Khoi, P. Veillet and I.A. Campbell, J. of Phys. F5 (1975) 2184 and private communication. [13] L.S. Goodman and S. Wexler, Phys. Rev. 108 (1957) 1524. [14] M. Kopp and W.D. Brewer, Hyperfine Interactions 3 (1977) 321. [15] P. Raghavan, M. Senba and R.S. Raghavan, Hyperfine Interactions 4 (1978) 330. [16] T.E. Cranshaw, J. Appl. Phys. 40 (1969) 1481.