f P&Y& ckm s&J Vol. 39, pp. 12734279 0 Ptrpmon Press Ltd.. 1918. Printed in Great Brim
MdSSBAUER STUDY OF ““Yb IN SOME METALS AND METAL-LIKE COMPOUNDS. ISOMER SHIET MEAS~EME~S AND ~~ED~TE VALENCE STATES OF YTIWZBIUM P. BONVILLE, P. I~~BERT, G. JEHANNO and F. GoNWEz-JIMENEEt DPh-G/PSRM, C.E.N. Saclay. Boitc Postale No. 2,91190 Gif-s/Yvette. France (Received 17 March 197g;accepted 10 May 1978) AlHract-We present here a Mssbauer effect study of the isomer shift and the magnetic behaviour of ‘%J in compounds with metallic character where the ytterbium valence state ranges from a practically divalent and diamagneticstate (Yb in Ag, Tm and Yb) to a trivalent magneticstate (Yb in Au). Intermediatevalence states, all of them probably non-magnetic,have been observed in Yb& TmA12, Yb.4 and Al as well as in the borides YbB,. Y~BIz and TmB12. In Tm metai, in addition to the quadrupoleinteraction,a weak magnetichyperfineinteractionis visible on ‘Dub
below TN.
1.lNntomoN Emphasis has been placed recently on studies of the ambivalent character of Yb in various metaliic or intermetallic matrices, and the results have been interpreted either in terms of a virtual bound state cfose to the Fermi level[ l-31, or in terms of interconfiguration fluctuations[4d]. In a review of mixed valence compounds[7], Varma introduced the concepts of pinning of the Fermi level by the f levels, of hybridization of the f and s-d states, and of the exclusion of states of more than one f efectron per atom. Miissbauer effect measurements on ‘7oyb can provide interesting information concerning both the electronic and the magnetic states of Yb. (a) Isomer shifts (IS) measurements provide information concerning the valence state of ytterbium. in spite of their relative smallness compared with the natural Miissbauer linewidth, it is now possible to measure and compare the IS for Yb with satisfactory precision if the influence on the lineshape due to dispersion[& 9) is taken into account. Also recent and precise data tables are now availabfe[9, lo]. (bf The observation of possible magnetic hyperfine structure may provide information concerning the ionic magnetic moment. Different cases have to be considered. In the case of isolated ‘7oyb impurities in a diamagnetic matrix, the existence of a focalized moment implies that the slow relaxation hypertine structure will be visible at fow enough temperatures [ 1I-131. For isolated Yb atoms in a matrix showing magnetic ordering at tow temperature, the Yb atom, if it has an intrinsic moment, will show a magnetic hypertine structure, corresponding to an important effective field, typically of the order of several megaoersteds. The situation appears more complicated in the case of a concentrated Yb compound without magnetic ordering, for the absence of a magnetic hype&e structure at low temperature may be due either to the diamagnetic tUniversidad Central de Venezuela, Caracas.
character of Yb in this compound, or to the presence of fast pararnagnetic relaxation through spin-spin or RKKY interactions. In principle, Mossbauer measurements in application of an external magnetic field, which in the latter case wouId give rise to a su~epti~ity-educed hyperfine field, could decide between the two possibilities. We present here a comparative study, on “‘Yb, of ytterbium diluted into Au, Ag, Al, Tm, TmA12and TmBtz metallic matrices (and also in insulating CaF& as welt as concentrated ytterbium in Yb met&, YbAL Yb&, YbB., YbBa and YbBr2. For Yb metal, we also give the results of a crystallographic study concerning the domains of stability of the f.c.c. and h.c.p. phases.
(a) Yb dilute samples have been studied as Miissbauer sources, with the impurities thus initially in the form of the ““Yb radioactive parent ‘Tm. Activation of thulium in Tm metal, TmA& and TmBr2 was obtained through neutron irradiation after sample preparation. The defects created during ~diation were then ei~nated by a 24 hr anneal in vacuum at 800°C. The purity of the metals was 99.9% for Tm and 99.999% for AL The boron used in the preparation of TmBr2 was 87.15% “B enriched, in order to reduce the number of defects created by the reaction ‘oB(n, a) during neutron irradiation. Two dilute alloys AJTrn were prepared by fusion in a levitation HP furnace, with diierent ~~iurn concentration kvels (c = 100 and 350 at.ppm). After irradiation for 3 days in a flux of 2.5 x 10” neutrons/cm3/s, narrow Mossbauer fines were observed so that no annealing was necessary. The dilute alloy sources AuTm (c = 170 and 500 atppm) and 4Tm (c = 500 at.Gm) were prepared in an induction furnace with thulium which had been previously neutron activated. Rsion was carried out in a Be0 crucible in a low pressure argon atmosphere. After being cold rolled, these sources were annealed at 800°C
1273
1274
P. BONVILLE etal.
in vacuum. In the case of &Tm alloys, partial oxydation of the rare earth could not be completely avoided: the presence of Tm,O, clusters was revealed through the existence of spots of high radioactivity on autoradiographic films, and through the appearance in the Mossbauer spectra of a broadened asymmetrical component, caracterizing TmzOp. This component, which has only a weak relative intensity in our best sample, was the dominant contribution in the &Tm spectra reported by Stbhr[l4]. (b) The concentrated Yb samples (Yb metal, YbAL, YbAls, YbB4, YbBs, YbBJ were studied as absorbers. All samples were checked by X-ray diffraction and a particular study presented in the next section was devoted to Yb metal which can crystallize in various forms. The YbB, and YbBrz samples were prepared by Etoumeau and Mercurio[lS]; the “m enriched YbB6 sample, used as a moving single line absorber in sources studies, contained 33 mg/cm2 of “yb and was prepared by Achard and Percheron[lS], as well as YbAlz and TmB12.
the well-known case of cobalt. In addition, DebyeScherrer X-ray spectra of well developped crystallites show that the distinct (222) line of the f.c.c. phase and (004) line of the h.c.p. phase are identically spotted; this indicates the simultaneous presence of the two stacking modes inside one crystallite and also the martensitic character of the transformation. The interpretation of X-ray spectra obtained with Cu Ka radiation leads, for the two phases, to extrapolated parameters (Nelson and Riley’s method) which are very close to those determined by Kayser, but rather different from those given by Biicher et al. for the hexagonal phase. We find: f.c.c. phase: a0 = 5.4856+ 0.0003 %r. a = 3.880~2 0.0004 A * O*kgkjA,
l
h*c.p* phase:c = 6.3%
It is seen that a0 differs only slightly from av2, showing that atomic planes perpendicular to the stacking direction are not very sensitive to the stacking mode. In addition, the value of the c/a ratio (1.6453) is significantly greater than the value corresponding to f.c.c. phase stacking (1.633). The resulting volume increment (AV/V = 0.009) is much bigger than the estimation of Bticher et ol. Owing to the fact that the Miissbauer study has to be carried out at low temperature, only the h.c.p. phase could be investigated by this technique. Our sample was prepared in this crystalline form by heating for 24 hr at 325°C and water quenching. X-Ray d&action spectra showed a low percentage of the f.c.c. phase (-5%) and gave evidence of extended coherent domains (> IOCL)for the hexagonal phase.
2.2 X-Ray study of Yb metal Existing data seemed contradictory as to the domains of stability of face-centered cubic (f.c.c.) and hexagonal closed-packed 0r.c.p.) phases of Yb metal. Stephens [ 161 found that the domain of stability of the h.c.p. phase was above 305°C where Bficher ef al. [ 171and Kayser [ 18, 191 observed a h.c.p. phase below 0°C. These apparent discrepancies, following Bficher et al., were related to the sample purity grade. Let us note however that Yb was studied only above 25°C in Ref. [16] and only below 130°C in Refs. [ 17-191. In order to clarify this point, we carried out a crystallographic study on polycrystahine needles which had 2.3 Experimental precautions in the detennination of been submitted to various thermal treatments up to isomer shifts 350°C. X-Ray measurements were performed at room ““Yb IS are generally smaller than one tenth of the experimental linewidth, and the resulting measurement temperature as we did not observe any transformation of difficulty explains the scatter of the early published data. our samples whatever their crystalline form, after leaving For precise determinations it is necessary to have single them a long time at this temperature. line spectra, or at least a spectrum with a very simple In our sample, which was 99.9% pure, we observed the hypertine structure with only one adjustable parameter, coexistence of the two phases (f.c.c. and h.c.p.) when it such as for example the energy separation of the slow had been submitted to temperatures over a wide range. relaxation doublet in cubic symmetry. In order to obtain Above 310°C the h.c.p. phase is predominant. Below under the best conditions (single line and small stat.isticaJ 300°C the f.c.c. phase is the more important and becomes scatter) a precision of about 20.01 mm.s-‘, which is to the almost exclusive phase after an llday anneal at be compared to the natural linewidth close to 2 mm.s-‘, 150°C. These observations qualitatively agree with those of Stephens but we however noticed, as did Biicher et we respected the following precautions: (a) We chose a unique reference (,TmA& source) for al., that a sample, preannealed at Iso”C and showing both the source and the absorber measurements. essentially the f.c.c. phase, rapidly transformed when (b) In the case of an absorber, the spectrum was steeped into liquid nitrogen: the X-ray pattern, obtained recorded directly with the reference source Tnu&. The at room temperature, show the domination of an h.c.p. exact position of the zero-velocity point in the spectrum phase identical to that obtained at high temperatures. was determined using a simultaneously recorded calibraThis shows that the h.c.p. phase is again the stable phase tion spectrum from a “Co in Cu source, placed at the at low temperature, so supporting recent results opposite end of the movement drive, and a single line obtained, by Vedemikov et a!.[#)], from thermopower, potassium ferrocyanide absorber whose. isomer shift is Hall effect and electrical resistivity measurements. known with precision. An a-Fe& absorber was alterIt should be noticed that in the two-phase mixtures, natively used to calibrate the velocity scale. only very well organized phases are present. We did not observe any sign of stacking faults, which contrasts with (c) In the case of a source, differential measurements
1275
Gssbauer study of “%‘b in some metals and metal-like compounds
were made by successively comparing the source under study and the reference TmAL source to the same moving single line absorber (Yb& in our experiments). The two measurements were carried out immediately one after another in strictly identical experimental conditions, p~i~ul~ly those concerning the movement, in order to eliminate possible causes of errors. (d) The theoretical lineshape fitted to the observed spectra was taken to have a lorentzian form modified by dispersive effects[8, 91, the dispersion coefficient being 25 = -0.03 for an absorber spectrum, or +0.03 for a source spectrum. (N.B. By convention, increasing positive velocities represent transitions of increasing ener8ies in the spectrum of the sample under study, for both the absorber and the source spectra.)
2.
-0.270
t
-0.263
t 0.CO.S
-0.095
* 0.020
c 0.3
0
a 0.010
0 0
-0.063
f 0.010
*0.020
t 0.010
m.031
t 0.010
0
ro.038
* 0.010
0
(
group, and Yb is in a 4g site of mm local symmetry, which could give rise to a non-axial electric field gradient (EM;). However the spectra measured at 4.2 and 0.34 K show only one sharp line, and the fit shows that the quadrupole interaction a = eV,,Q/8 is smaller than 0.3Omm.s-’ in absolute value. In addition these spectra show no magnetic hyperfine structure at these temperatures.
3.HYXMIuEPA3UMEMS fitted
3. I Cubic compounds For each of the compounds &Tm. BTm, TmAl2, TmBj2 (sources) and YbAL YbAl,, YbBs, YbB,z (absorbers), we observed at 4.2 K one sharp line ~dica~g the absence of both any quadrupole interaction (as expected for cubic local symmetry) and of any magnetic hypefine structure. The isomer shift was precisely determined in each case. In Table 2 is given a new measurement of the isomer shift of the Yb*’ absorption line in a cubic site of the CaFt insulating lattice. This value includes the dispersive correction and is more precise than our previous determination [2!]. In Table 1 we also include IS measurements from slow relaxation hypefine structures of Yb’+ ions in cubic sites, in the insulating source CaFz (“@Tm) and in the metallic source Au (““I’m). These values were obtained from our previously published spectraf21, 221, using the dispersive correction. The existence of doublets instead of single lines leads to larger error bars in the IS values.
3.3 Julie Thulium is a hexagonal metal showing antiferromagnetic order below 56K[23, 24). The Yb impurity is in a site of axial symmetry (DM) with respect to the c axis. Although the emission line of “@I’min thulium is broad and asymmetrical, it has been used as a reference source in early Mssbauer ex~~ments on “%[25]. We made a fine analysis of the lineshape and of the hyperfine interactions. The 4.2 K spectrum was fitted to an axial hyperline Hamiltonian:
where I& lies along the c axis, as do the thulium magnetic moments. The comparison of the fitted lineshape of Fig. I(b) with the lineshape obtained in the absence of I& (Fig. la) clearly shows that a magnetic interaction is present; this term however has not been taken into account in pre-
3.2 7%e YbB. compound The YbBI compound possesses the P4lmbm (&II) space .W hl.S
lh
32
-Al
Tm
CaF2
(Tm)
(Yb3*
in
-Au Tm
CaF2)
-1)
I.$. (Ref.
-G-
bn.s-‘) TmA12)
observations
:KG)
2.50
-0.116
2 0.020
2.60
-0.082
f
0.025
2.60
r0.023
r
0.008
0
0
single
tine
2.94
l0.039
? 0.015
0
0
single
line
3.14
+0.65
4.72
+0.17
+ I
0 -
0.62
0 96
single
line
unresolved
structure
;
0.03 0.04
Fitted parametersof sourcespectrarecordeda( 4.2 K with a movin8 “@Ybenriched sin@eline YbBd absorber. W. experimentalline width (natural width W, = 2.0 mm.s-‘; IS, isomer shift referred lo a TmA12source with the same conditions. A dispersion corrected line shape has been used; CY= [3 cV,Q141(21- l)] = (cV,,Q/8), quadrupole interaction; Hh,, hyperline field. PCS
Vol. 39. No. It-C
0.3
Fitted parametersof absorber spectra recordedat 4.2 K with a movingTmAll source. (Same definitionsas for Table I.)
hyperfme parameters are given for the sources in Table 1 and for the absorbers in Table 2. The
Table
1276
P.B0lWU.Eet ol. 3.4 Ihe yb meld sample The yb metal sample that we used th.c.p. phase) gave one sharp line at 4.2K. The fit shows that the axial quadrupole interaction is vanishingly small (Ial < 0.3Omm.s-‘f. This difference in quadrupole interaction between Yb metal and Tm metal probably arises as Yb is closer to the ideal hexagonal compact phase (c/a = 1.633) than Tm: c/a(Yb) = 1.645 and c/aC]rm)= 1.579.
Ftg. 1. M&sbauer spcctnun at 4.2Kof the source ‘“Fm in Tm metal. (a): &ted with an axial quadrupoieinteracti00only; fb): fitted with WIaxial quadrupo#einteractionand a hyperiioefield. vious studies. It is convenient here to stress the importance of the fitting hypothesis in the determination of IS: a single line fit would give IS = to.073 mm.s-’ with respect to TmA&, a fit using only the quad~~le interaction would give IS = -0.024 mm.s-‘, whereas a fit with both the quadrupole and the magnetic coupling gives -0.082mm.s-‘. We consider that this last value is the only significant one. The measured quadrupole interaction is almost the same at 4.2 K (T < TN) as at 60 K (T > TN). Its value is Q!= eV,,Q/8= -0.62~0.04mm.s-‘, hence V,, = (tO.tia0.04)
x 10” V.cme2.
This value does not differ much in absolute vaiue from the measurement made at 296 K using perturbed angular correlations on ?‘b in Tm metal by Rasera and LiScholz[26]: V,, = (0.453 +0.047)x lo’* V.cm-* It is in agreement, in absolute value, with a previous result using the Miissbauer Effect on ““Yb by Henning et al. [27] at 4.2 K: Vzl = (-0.63 * 0.06) x IO” V.cm-*,
but it differs from it in sign. Finally. it is of the same order of magnitude and has the same sign as the value:
giving V, = 0.52 x IO’*V.cme2, deduced by Uhrich and Barnes[M(] from Mijssbauer measurements on “?m in Tm metal. This shows that the 4f electrons do not give any contribution to the EFG, as is to be expected for an ion having essentially the 4f” configuration.
Up to the present time, almost al1 “@Yb isomer shift measurements, except those published by Russell et al.[9], have been fitted without using the dispersive correction to the lineshape. Consequently, following Russell et aI. and also our own obse~ations, a mean dispersive correction of -0.045 mm-s-’ has to be added to previously published IS measurements in absorbers, when they are given with respect to a reference source; similarly, a correction of +0.045 nuns-’ has to be added to previously published source IS given with respect to a reference absorber. No correction is however necessary when sources are compared with sources, or absorbers with absorbers. In Fig. 2 we show the IS measured in the present work, with their error bars. In the upper part of the figure we include several already published measurements[9, 10,291 with dispersive correction if necessary. Taking into account the error bars, our present results are compatible with our own previous IS evaluations modified by the dispersive correction, in the compounds YbBs, TmB,*[M, 311; Yb metal[32]; CaF2(ybz’)[211, but the precision has been greatly improved. It appears, on the other hand, that we made an error in evaluating the relative IS between Tnu% and YbAlt in Ref. 1301,and our present evaluation concerning YbA& (IS = -0.063 2 O.OlOmm.s-‘) is close to that given by Russell ef a/.[91 (-0.075*0.010 mm.s-‘). The IS values found for Yb2” and Yb” in sites of cubic symmetry in CaF2 (Fig. 2) agree with the other values for insulating lattice given by Russell et al. We support also their conclusions con~m~ng YbSO,, whose previous IS data appear to be unreliable (unresolved structure), and which is certainly not a good divalent reference absorber. Our precise measurement in a YbBs absorber (IS = - 0.263 + O.OOt? mm.s-‘) supports our previous conclusions about the divalent character of Yb in this ~om~und[~]; this value is cIose to that found by Russell et al. (- 0.266 ? 0.009 nuns-‘). The value we obtain in YbAlj (0.034+0.010 nuns-‘) is compatible with the value given by Henning1291 after the dispersive correction is added (0.015eO.030 nuns-‘). Our result in Yb h.c.p. metal (-0.095+0.02Omm.s-‘) is intermediate between the values -0.065 iO.O38[9] and - 0.105 * 0.020 [lo] with dispersive correction. However, the isomer shift in Yb metal with respect to a Tm metal source g,iven by Henning et a/.[331 (-0.145+ 0.02Omm.s-’ with dispersive correction) is not in agreement with our results, as we find comparable
Mhbauer
study of ““Yb in some metals and metal-like compounds
Evbep
VbAlp
I1
0
Ybm,&D++~Vb
Vb 6Up1 rrwtolrgl
q VbF,[‘] c... - 0.3 IxI
[XIVb, -0.1
-0.2 Ca F2 (Vb ‘+)
0
Oyb
AL2
OTm(Vb) IVb metal aVbB6
1
1277
Ag (Vb)
6&o
F,“;y”b:)
OIAL
(Vb)
On OVb
42
OTm
$1 +0.2
Al3 &(Vb)
OVbB‘
A&R)
I
1
Fig. 2. “%I isomer shift data referred to the source TmAL,using a dispersivecorrectionfor the line shape. (lie data below the velocity scale originatefrom this work. Insulating compounds are marked by crosses.)
isomer shifts in Tm and Yb. We think this discrepancy arises from the fitting hypothesis for Tm metal in relation to the non-resolved structure, as explained in the Section 3.3. 4.2 discussion of the results In insulating compounds it seems clear that there are two groups of IS values of “Vb, corresponding to the two possible valence states Yb” and Yb3+, each being characterized by a weak dispersion due to covalency effects. In metals or intermetallic compounds our results show that the situation is much more involved. Two effects can alfect the value of the electronic density 1Jl(O)l’at the nucleus, and in turn the IS value: (i) The conduction electron+ electron interation can give rise to fractional valence states of the Yb ion[ld]. A calculation performed in Eu metal(34] has shown that the occupation of the 4f shell mainly affects the screening of the 5s core electrons, and hence the value of l@5,(0))2,so giving a variable ionic contribution to the IS. We take for the value of the difference of the “oyb IS in insulators the value given by Russell et a1.[9]: AS = IS(3’) - IS(2’) = 0.362 + 0.014 mm.s-‘, corresponding to the difference J$(0)(2(4f
13) - I$(O)l’(4f
14) =
(4.0 t 1.2)x lti6 cm-‘.
(ii) The presence of conduction bands with s-character adds the quantity (&O)l’ to the electronic density at the nucleus. We may consider these two effects to be additive, the 5s shells lying too deep to participate in the conduction band. It is clear that an approximate knowledge of the second (or conduction electron) contribution to ItiO)l’ is necessary to determine the first (or ionic) contribution from the measured IS, and by comparison with insulating compounds, to determine the approximate valence of the Yb ion in the metallic matrix under consideration. Measurement of the Knight shit of the NMR absorption lines provides the quantity (&(0)(&[35] (the conduction electron charge density at the nucleus,
averaged over the Fermi surface). It can be assumed that
J&(O)& is close to (but smaller than) the conduction electron contribution :(h(O)l’ that appears in the expression of the MCssbauer transition isomer shifts. We shall now compare the results of NMR measurements on lmAu in Au metal, on ‘09Agin Ag metal [MJ and on “‘Yb in ybA2[37) with our own ‘myb IS measurements in Au, Ag and YbAl2. (a) &Yb. Ytterbium diluted in gold is trivaientI38.391 and possesses a localized moment whose relaxation has been studied by the M&sbauer effect[l2, 131. The difference in IS compard to yb&[9] taken as a trivalent insulating reference compound, corresponds to a variation of the electronic density A(#(O)(’= 1.14x lo’” crnd3. This variation represents the conduction electron contribution : Ih( at the Yb site. It may be compared with the quantity I&(0)(‘,, = 2.64 x 1026cm-’ measured by NMR on ‘“Au in gold[36]. The difference between these two values seems to be due mainly to the difference between the nuclear potentials 0% = 70 protons; Au = 79 protons), which leads to an increased electronic density at the gold nucleus. (b) &Yb. It is generally admitted that Yb is divalent in Ag[38,39]. The difference in IS compared to YbF2[9], taken as an insulating divalent reference compound, corresponds to a variation in electronic density A(#(O)l’= 1.99x 10%cmm3, which may be attributed to the contribution of conduction electrons at the Yb site in Ag. The excess in nuclear charge favours in this case the ytterbium atom (Yb = 70 protons, Ag=49 protons), which could qualitatively account for the difference between the Miissbauer result and the value obtained in an NMR measurement on ‘Q)Ag[361,l&(0)(2,, = 0.64 x lo’” cm-‘. But the hypothesis that Yb is not exactly divalent in Ag cannot be discarded[40] and in this case the MCssbauer measurement would contain, besides I (&(O)l’, a small ionic contribution due lo the fractional k valence. (c) YbA12.In the YbA12compound the difference in IS compared to YbF2 corresponds to a variation in electronic density of A(4(0)12= 2.57 x 10%cm-’ = z l&(O))2+ Al+(O)l?+ where Al#(O)l;bti accounts for
I?78
P. BoNvlLLE el al.
fractional valence. The “‘Yb NMR measurement in YbAl,[37] gives: I&(0)(& = 0.77 x lO”cm-‘; we can estimate that C I&(O)l’ is equal (or slightly superior) to this value, which leaves about 1.8~ 10zd cm-’ for Alti(O)(Li, and leads to a fractional valence f =Z2.45 for Yb in YbAl?, in agreement with the value f = 2.5 of Ref. 131. Assuming that the conduction (6s) bands are similar in YbAl,, YbAll and in TmA12, we obtain from our IS measurements the following intermediate valences: about 2.7 in YbAl, and 2.6 in TmAL. Our value in YbAl, is in better agreement with the determination using the virtual bound state model (f = 2.9[3]) than with that obtained from interconfiguration fluctuation theory cf = 2.2161). (d) &lYb. The IS we observed for ““Yb in Al metal is close to that in YbAl,. In spite of the absence of data relative to the conduction electron contribution, it may be concluded that, diluted in Al, Yb is in an intermediate valence state which is far from Yb*’ state. (e) Yb and Tm metals. Cashion et al. (411suggested the hypothesis of the existence of 0.2 holes in the 4f shell of Yb metal, that is of a fractional valence 2.2. The analogy between the “hlrb IS in Yb metal and in Tm metal suggests, as long as the conduction electron contributions are similar, the existence of a fractional valence of the same order of magnitude in the two cases. (f) YbB., YbB,*, TmB,2. Rare earth tetra- and dodecaborides are considered to have a metallic character [42-44]. Inspection of Fig. 2 shows that YbB,, YbI& and Yb in TmB,* present low temperature IS which are in the range observed for the allows such as YbAl, with intermediate valence states. For YbB4, this conclusion agrees with the magnetic measurements carried out by Berrada et al. [44]. Further quantitative evaluation of the fractional valence is however difficult in these borides due to the absence of data relative to the conduction electron contribution. 5. DCXUsSlON OF MAGNKllC PROI’ERTES Dilute ytterbium alloys (a) Ag(Yb) and AUb). We recorded the emission spectra of 4Tm and gTm sources at 0.34 K. As we did not observe either any slow relaxation magnetic structure as in the AuTm source [ 12, 131,or any relaxation line broadening, wethus conclude that, at low temperature, the ytterbium impurity possesses no magnetic moment in both silver and aluminium. (b) TmB&Jb). We made two types of Massbauer measurements on TrnI%*. First, we recorded the ‘@I’m absorption spectra of TmB,* from 20 down to 1.35 K, in order to study the magnetic behaviour of the matrix. Then we studied the ““Yb impurity from the emission spectra of ““T’m in Trn&* in the same temperature range. Single line spectra were observed in both cases. From the ‘“9Tm absorption spectrum, we conclude that no magnetic order exists in TmB12 at 1.35K. Moreover, as Tm” is a non-Kramers ions, it is very likely that the TmB,, matrix is diamagnetic at this temperature. Then, the absence of relaxation broadening of the ‘% line 5.1
implies the absence of any localized moment on the ytterbium impurity in this boride at low temperature. (c) Tm(Yb). The magnetic structure of Tm metal has been studied by neutron diffraction(23, 241. At low temperature, it presents a colinear structure along the c axis, with alternatively 4 moments along +c and 3 in the opposite direction. The effective field we found at the site of the “okra impurity following the radioactive decay of “@Tm is .H,,, = 91% 20 kOe. This result is smaller than the Yb3’ effective field in magnetic matrices by at least one order of magnitude. It clearly shows that the Yb impurity is non-magnetic in Tm metal, and that H,e comes from the s-band conduction electron polarization by the Tm3’ ionic magnetization at low temperature, through s-f exchange interaction. The small value of H.n did not enable us to separate the possible two distinct values corresponding to the two magnetically unequivalent sites, as was done for Sn impurities in Tm metal[45]; hence, the fitted value of H.n must be considered as a mean value. It is noticeable that all the hyperline parameters measured for ““Yb in Tm metal (IS, EFG, H.n) support the fact that the ytterbium configuration is not far from the diamagnetic Yb*’ ion in this metal. 5.2 Yb concentrated compounds No magnetic hypefine structure is visible in YbB4 and in YbBll down to the lowest temperature investigated (0.34 K for YbB,; 1.35K for YbB,J. This rules out the existence of any magnetic order at these temperatures. Even the existence of a Yb ionic magnetic moment seems very unlikely, as no relaxation broadening is observed. However it is not impossible that fast relaxation could be maintened by spin-spin or RKKY interactions at these low temperatures, and only the absence of an induced hyperline field in a Mssbauer experiment with an applied magnetic field (as was carried out for YbAIa[46]) would unambiguously co&m the diamagnetism of Yb in YbB. and YbB12. 6. DYNAMICPROPERTDS
We determined the apparent Debye temperature of the Yb impurity in aluminium from the variation of the Debye-Waller factor between 4.2 and 60 K. The result (0, = 225+ 10 K) is comparable with that obtained in TmA12(BD= 230~ 5 K)[M] and greatly differs from the Debye temperature of the Al lattice (0, = 428 K)[47]. This difference is probably due to the very different masses of Yb and Al. The apparent Debye temperature which we measured in the same way for the ytterbium impurity in thulium is 0, = 158K. This value is not very different from the result of specific heat measurements on the matrix (0, = 200 WWI. 7. CONCLW
REMARKS
this work, we have initially pointed out the experimental precautions necessary in order to accurately measure and compare source or absorber isomer shifts which are as small (compared to the linewidth) as for ““Yb. Our measurements accordingly contribute to the estIn
MBssbauer study of ‘“lp in some metals and metal-like compounds
ablishing of a scale of ““Yb isomer shifts in metallic matrices, where the shifts extend over a relatively large velocity range, due to the variable contribution of Yb fractional valence, and, to a lesser extent, to the variable conduction contribution
electron contribution. When the may be independently evaluated,
latter then
isomer shift measurements will provide an estimate of the fractional valence of ytterbium. Our crystallographic measurements on Yb metal confirm that the h.c.p. form, which we studied by the Massbauer effect, is the stable phase at low temperatures. A detailed analysis of hypefine interactions in a Tm metal source has been carried out and shows the presence of a small hype&e field at low temperature. Taking into account this hyperfine field enables us to correct previous isomer shift evaluations. The ‘6PTmMiissbauer study of Tn& has shown that this compound is not magnetically ordered at 1.35 K; the ““Yb spectra of YbB4 and YbBll revealed no magnetic structure respectively down to 0.34 and 1.35K. From “Vb isomer shit measurements, it appears that ytterbium presents an intermediate valence state in the three metallic borides TmB12, YbB, and YbB,*.
I. Coqblin B. and Blandin A., Adu. Phys. 17, 281 (1968). 2. C&blin B., J. Phys. (Paris) Suppl. 32. 599 (1971). 3. Havinna E. E., Buschow K. H. J. and Van Daal H. J.. Solid 4. 5. 6. 7. 8. 9. 10. II. 12. 13. 14. 15.
I279
16. Stephens D. R, 1. Phys. Chem. Solids 26,943 (1%5). 17. Bucher E., Schmidt P. H.. Jayaraman A., Andres K., Maita J. P., Nassau K. and Demier P. D., Phys. Rev. B2(10), 3911 (1970). 18. Kayser F. X., Phys. Reu. I_.& 25(10). 662 (1970). 19. Kayser F. X., Phys. Status So/i& (0) 8, 233 (1971). 20. Vedemikov M. V., Burkov A. T., DvunitkinV. G. and
MorevaN. 1.. Phys. Letts. 4EL4(4293) (1974). 21. Barely C.. Gonzalez-Jimenez F.. lmbert P. and Varret F.. 1. Phys. Chem. Solids 36.683 (1975). 22. Gonzalez-Jimenez F.. Hartmann-Boutron F. and Imbert P..
Phys. Reu. Bib, 2122(1974). 23. Koehler W. C., Cable J. W., Wollan E. 0. and Wilkinson hf. K., Phys. Rev. 126,1672(1%2). 24. Koehler W. C., 1. Applied Phys. 36, 1078(1%5). 25. Wagner F. E., Stanek F. W., Kienle P. and Either H., Zeitschrift fiir Phys. 166, 1 (1962). A., Phys. Rev. Bl, 1995 (1970). 26. Rasera R. L. and IXcbol2 27 Henning W.. Kalvius G. M. and Shenoy G. K., Phys. Rev. C2.2414 (1970). Uhrich D. L. and Barnes R. G., Phys. Rev. 164,428(1967). 2 Henning W., Zeitschrijt fiir Phys. 217,438 (1968). 30 GonzalezJimenez F., lmbert P.. Achard J. C. and Percheron A., Phys. Stotu~ Solidi (a) 19, 201 (1973). 31. Bonville P., Gonzalez-Jimenez F., Imbert P. and Varret F., 1. Phys. (Paris), Suppl. 35, CoUoque C6. 575 (1974). 32. Percheron-Guegan A., Achard J. C., Gorochov 0.. GonzalezJimenez F. and Imbert P., I. Less Corn. Metals 37, 1 (1974). 33. Henning W., Bare G. and Kienle P., Hypefie Inreractions EMed Nuclei, Pnx. Conf. Rehouot and Jerusalem, Vol. 3. p. 795. Gordon and Breach, New York (1971). 34. Kobayasi S., Fukuchi M. and Nagai S., Sol. Stare Comm. 13, 727 (1973).
35. Slichter C. P., Principles of Magnetic Resonance, Harper and Row. New York and John Weatherhill. Inc. Tokyo (1%3). Stole ?omm. 13.621 (1973). 36. Narath A., Phys. Rec. 163,232(1%7). Hirst L. L., Phys. Kondens. Marerie 11, 255(1970). 37. Gossard A. C., Jaccarino V. and Wemick J. H.. Phys. Rev. HirstL. L., AIP Conf. Proc. 24, 11 (1975). 133A.881 (1964). Sales B. C. and Wohlleben D. K., Phys. Rev. L&t. 35, 1240 38. Bijvoet J., Van Dam A. J. and Van Beek F., Sol. State (1975). Comm. 4,455 (1966). Varma C. M., Rev. Mad. Phys. 48,219 (1976). 39. Gainon D., Donze P. and Sierra J., Sol. Stale Comm. 5, 151 Wagner F. E., I)unlap B. D., Kalvius G. M., SchaUer H., (1967). Felscher R. and Spieler H., Phys. Rev. Letf. 28, 530 (1972). Russell P. B., Latshaw G. L., Hanna S. S. and Kaindl G.. 40. Murani A. P.. Sol. State Comm. 14, 199(1974). 41. Cashion J. D., Cot&hard M. A. and Prowse D. B., 1. Phys. Nucl. Phys. A210,133(1973). Chem. sol. State Phys. 8, 1267(1975). Henning W.. B&e G. and Kienle P., Zeit. jrir Phys. 241, 138 42. La Placa S., Binder 1. and Post B., 1. Inorg. Nucl. Chem. 18, (1971). 113(1961). Hirst L. L., J. Phys. Chem. Solids 31,665 (1970). Gonzalez-Jimener F. and lmbert P., Solid Srofe Corn. 11,861 43. Odintsov V. V. and Pademo Y. B., Izveslia Acad. Nauk Neog. Khim 7, 333 (1971). (1972). Gonzalez-Jimenez F.. lmbert P. and Hartmann-Boutron F., 44. Berrada A., Mercurio J. P., Chevalier B., Etoumeau J.. Haguenmuller P., Lalanne M., Gianduuo J. C. and Georges Phys. Rev. B9.95 (1974). R., Mater. Res. Bull. (USA) 11, 1519(1976). St& J., Thesis (Fakultiit fiir AUgemeine Wissenschaften der 45. Kuchma A. S. and Shpinel V. S.. Soviet Phys. JETP 35, 5% Technischen Universitit Milnchen (1975). (1972). We are indebted to Etoumeau 1. and Mercurio J. P. (Laboratoire de Chimie du Solide du CNRS. Universiti de 46. StBhr J., Phys. Reu. Bll, 3559 (1975);(see note, Ref. [45] of this paper). i)ordeaux 1, 33405Talence Cedex, France) for supplying us with YbB4 and YbB,z samples, and to Achard J. C. and 47. Kittel C.. Introduction lo Solid SIare Physics. Wiley. New York (1970). Percheron A. (Laboratoire des Terres Rares du CNRS, 92190 Meudon-BeUevue, France) for supplying us with YbA12, 48. Lounasmaa 0. V., Phys. Rev. 134, A1620(1964). YbBa and TmE112 samples.