FERROMAGNETISM IN THE RHGD AND PTGD SYSTEMS

Rhodium and platinum have bccn studied with up to 33% additions of gadolinium. The magnetic propenies of the constituents in tbe layers of a eutectic structure are different from the bulk properties of tbc constituents. The compound GdPts ordcrs ferromagnetically at 13.9 K.


INTRODUCTION
Recently Matthias et al. (1) demonstrated bulk superconductivity in the eutectic forrned between pure Ir and Ylr2. The eutectic has a superconducting transition temperature that is much higher than that of either constituent. This enhancement is correlated with a dramatic drop in the Debye temperature at the eutectic composition which is thought to be caused by the longrange strain occurring from the lattice mismatch at the interfaces between the layers of the constituents (2). This led to a study of the magnetic properties of the corresponding system with the yttrium replaced by gado-1 inium, to a survey of dilute additions of most rare earth elements to iridium, and a survey of dilute additions of gadolinium to ruthenium, osmium, rhodium, and platinum (3). W e report here on further work on the RhGd and PtGd systems. The IrGd is somewhat more compl icatedthan r e ported in reT.
(3) and. quite naturally, resemble s the RhGd which remains puzzling. nie PtGd system i s now fully described.
This work was unde rtaken to elucidate the magnetic behavior of chese systems from their dilute limic, through the eutec t ic composicion, to the first compound . That has turned out to be difficult, although in a qualitative sense the results are clear.
What is equally important is that once again the need for metallugical examination of samples is clear, and specifically the difficulties encountered with preparing dilute magnets is apparent.

EXPERIMENTAL
The samplea were prepared in an argon-atmosphere arc furnace. Samples with less than 2000 ppm Gd (all % and ppm are atomic) were prepar ed in more than one dilution step. Weight losses upon mel ting were negligible. Hence, all compositions reported are calculated and respresent an upper limit to the gadolinium concentration. Annealed samples were wrapped in tantalum foil and held at iooooc for 8-14 days. Most of ehe samples were checked by Debye-Scherrer x-ray analysis and several were sub jec ted to me tallographic examination. Magnetic measurements down to 1.4 K were made in a vibrating sample magnetometer with the f ield produced by a superconducting solenoid having a maximum field of 54 kOe. In this work on dilute magnets most of them were run in the nominal zero field of the magnet becauee of the esse of separating the many magnetic transitions that occur (see ref. 3).
This field is moderately reproducible and is usually 50-100 Oe.

I . !!Gd
The 111<>st dilute compound that we found in PtGd alloys was GdPt5 with the CaCu5 structure. Althoug~we found no report of it in the literature, its existence is not surprising. A magnetization cooling curve of GdPt5 is shown in Fig. l, and the sample can be seen to order ferromagnetically at Tc m 13.9 K. This curve is not affected by annealing the sample. The saturation moment is 7. l+O. l 1.18 for a Gd atom which is in reasonable agreement with the theoretical value of gJ ~ 7.0. The approach to saturation is dependent on the alignment of the field direction with the cooling direction of the sample in the arc furnace. The m(H) curve rises initially about three times faster when the field is perpendicular to the solidification direction tban when parallel. The aamples ss.turate at -20 kOe, and the anisotropy is gone at that field. The x-ray examination showed that the sample was pure GdPt5 before annealing, while after, some pure Pt was present. Metallographie examination showed a few percent of a second phase that was unaf fected by the anneal. lt could be seen in the metallography that a marteneitic transformation was induced near the edge where the samples were cut mechanically.
The preparation of powder for the x-rays can be expected to induce this transfot"lllation.
Hence the powders were annealed at 600°c for one hour.
At the composition Gd.09Pt.91 a sample is formed that is completely a lamellar eutectic structure of GdPt5 and Pt layers with a layer thickness of <:'.5000 ft.
The Pt layers will of course contain dissolved Gd up to its solubility limit. A cooling magnetization curve of an unannealed sample is shown in Fig. 1. Two transitions are observed at Tc= 12.8 K and 7.5 K. The higher Tc must be associated with the GdPt5 lamellae and the lower with the very dilute PtGd lamellae. The reduced Tc in GdPt5 could be a strain-;ffect of the nature of the coherent strains discussed in ref. 2 or could be simply an inhomogeneously broadened transition because the sample can be viewed as Pt wich a Gd concentration modulation. The ordering in the dilute phase could occur from some conduction electron polarization that arises in the neighboring layers or may be true dilute magnetic impurity ordering where ehe non-equilibrium, high concentration of Gd atoms is present. lt seems unlikely that Gd clusters are fonning in ehe Pt layers because the existence of the layered str.ucture is, in fac t, the thermodynamic response to the solubility problem. After annealing, the GdPt5 Tc moves up to the value for the pure compound and the dilute Tc is severely broadened. This suggests simply that the Gd atoms are moving out of the dilute layers into better positions in the GdPt5 structure or equivalently that the lattice strain is being relieved.
As the Gd concentration is reduced, this same eutectic structure with its two magnetic transitions persists down to at least 1000 ppm Gd. Metallographie examination of a 2000 ppm sample shows that the Gd is in fact not all dissolved. This is in disagreement with the recent work of Hardiman et al. (4) who report no magnetic ordering of the Gd in Pt over concentration levels that overlap our levels. On the other hand they do have to deal with a troublesome "cluster" line in their electron paramagnetic resonance studies.
At concentration levels below -1% Gd, the dilute transition is usually not seen in cooling curves but shows up in the warming curves of field cooled samples. lt is then quite broad, resembling the transition in the annealed sample curve in Fig. 1. This dilute transition is obviously weakened as the amount of eutectic present decreases to what is likely the limit of one layer of GdPt5 between grains of Pt. However, even at 1000 ppm Gd the 12.8 K transition is clear. We conclude that the true solubility of Gd is below 1000 ppm and that the Gd that does dissolve does not order magnetically.
We report no magnetic parameters on the eutectic magnetic phases because it is difficult to do this on an inherently two-phase system.
The m(H) curves at Gd levels below -3000 ppm show no sign of saturation at 54 kOe, but even at that field they suggest a moment greater than 7 µ5. This could be caused by problems of eddy-current signals that appear in our magnetometer as the samples become very dilute.

II. RhGd
The RhGd is more complicated than the PtGd system. The Gd-Rh-Phase diagram is known (5). Rowe;er, we were not able to find any trace of the reported GdRh3 compound. We were able to retain some GdRh5 in the CaCu5 structure by quenching the samples. However, there was a second unidentified phase. After annealing, the GdRh5 disappeared and Rh and an unidentified phase appeared which was consistent with the metallographic examination.
Because of the difficulty of correlating various phases with up to three magnetic transitions in some samples, we simply present Tc' s versus nominal composition in Table I. We have made a tentative identification of the Tc ' s in Table I with various phases which is 76,56 K -GdRh2, 49 K -GdRh2 or unidentified phase present in eutectic layers, 33 K -dilute Gd in Rh in eutectic layers, 20 K -GdRh5, -10 K -ordering of dilute Gd in Rh grains. These identifications are only suggestions.
lt is clear that this is a more complicated system than PtGd and it seems likely that no more insight can be gained by the difficult task of more sorting out of phases.
lt is however obvioue that eutectics are magnetically different from their constituent phases. Also, there may be ordering at-10 K of  Interestingly, the properties of the dilute samples are independent of field direction. Some extra solubility is obtained at even 1800 ppm by quenching, since annealing brings up another magnetic transition. Further studies are underway on the dilute limit.

CONCLUSIONS
The layered and strained structures present in samples at eutectic compositions modifies the magnetic properties of the constituent phases and may represent a new type of magnetic material. As always, the characterization of the samples is crucial.

ACKNOWLEDGMENTS
We are grateful to Prof. B. T. Matthias for many provocative discussions. We thank E. G. Zukas and R. A. Pereyra for metallurgical examinations of samples.