HAL Id: jpa-00218891
https://hal.archives-ouvertes.fr/jpa-00218891
Submitted on 1 Jan 1979
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Valence instabilities of Tm in its compounds and solid solutions
F. Holtzberg, T. Penney, R. Tournier
To cite this version:
F. Holtzberg, T. Penney, R. Tournier. Valence instabilities of Tm in its compounds and solid solutions.
Journal de Physique Colloques, 1979, 40 (C5), pp.C5-314-C5-320. �10.1051/jphyscol:19795108�. �jpa-
00218891�
Valence instabilities of Tm in its compounds and solid solutions
F. Holtzberg, T. Penney and R. Tournier (*)
IBM T. J. Watson Research Center, P.O. Box 218, Yorktown Heights, NY 10598, U.S.A.
(*) Centre de Recherches sur les Tres Basses Temperatures, CNRS, B.P. 166X, 38042 Grenoble Cedex, France
Résumé. — TmSe présente à la fois une valence non entière et un ordre antiferromagnétique, contrairement à la plupart des autres systèmes de valence intermédiaire qui sont non magnétiques à basse température. Nous présen- tons des mesures de constante de réseau, de susceptibilité magnétique et de résistivité effectuées sur des monocris- taux de Tm^Se dans sa région de solubilité solide 0,79 <; x <, 1. Nous trouvons des corrélations directes entre les propriétés physiques, la valence et la concentration. La fraction de caractère divalent dans Tm^Se augmente avec x, l'échantillon le plus déficitaire en thulium étant essentiellement trivalent. Des monocristaux d'alliages de TmSe avec YSe ont également été préparés. Les constantes de réseau et les susceptibilités indiquent que ces matériaux ont une valence non entière pour toutes les compositions. Lorsque la concentration de thulium augmente, on passe d'une résistivité Kondo métallique à la haute résistivité caractéristique de TmSe stœchiométrique. Bien que nous trouvions les propriétés des deux configurations Tm2 + et Tm3 + dans les propriétés magnétiques et les propriétés de transport, le fait d'obtenir ces résultats aussi bien dans la limite diluée à une impureté que dans TmSe stœchio- métrique, nous amène à conclure que tous les sites de thulium sont équivalents.
Abstract. — TmSe exhibits both non integral valence and antiferromagnetic order in contrast to most other intermediate valence systems which are non magnetic at low temperatures. We report on measurements of lattice constant, magnetic susceptibility and resistivity made on single crystals of Tm^Se over its range of solid solubility, 0.79 ^ x ;g 1. We find direct correlations among the physical properties, valence and concentration. The fraction of divalent character in Tm
xSe increases with x, the most Tm deficient sample being essentially trivalent. Alloys of TmSe with YSe have also been prepared as single crystals. Lattice constants and susceptibilities indicate that these materials have non integral valence over the entire system. With increasing Tm concentration a metallic Kondo resistivity changes smoothly towards the high resistivity characteristic of stoichiometric TmSe. Although we find the properties of Tm
z+and Tm
3+configurations in magnetic and transport measurements, since these results are observed in both stoichiometric TmSe and in the dilute single impurity limit, we conclude that all TmSe sites are equivalent.
The traditional and highly successful view of rare data [3] (SmS under pressure, S m ^ ^ Y j S , SmB
6) earths is that for each ion there is an integral number show a single line corresponding to a single non inte- of 4f electrons in a single configuration given by gral configuration. The clearest indication of the Hund's rule. The / multiplets are split, in solids, by existence of a new state is the observation of a tem- the crystalline electric field. In recent years it has perature independent susceptibility, at low tempera- become apparent that under certain conditions two tures, where a Curie divergence is normally expected.
electronic configurations may be nearly degenerate For example, although SmS under pressure is appro- and the resulting ground state has a non integral ximately 2/3 Sm
3 +and 1/3 Sm
2 +, no Curie moment number of 4f electrons. A good example of this non associated with Sm
3 +is observed [1].
integral valence state is SmS under pressure in which Unstable valence has been associated with Ce, the 4f
6-4f
55d configurations coexist [1]. Several Sm, Eu, Tm and Yb in various compounds [4]. It is of fundamental questions arise. How does one describe interest to determine the specific conditions required the coexistence of two configurations with highly for the stabilization of the non integral valence state correlated localized f electrons and delocalized d or or the coexistence of two configurations. The factors s electrons ? If one uses the Kondo approach, how involved may include crystal structure, chemical can it be applied to concentrated systems ? What are constituents, band structure, pressure, stoichiometry, the effects of other types of localization such as ran- rare earth dilution, and local environmental effects.
dom potentials or magnetic polarons ? This paper explores the instability of the Tm valence One of the most interesting experimental questions under a variety of conditions and in the context of the is whether the physical properties are those of the above considerations.
individual configurations, their average, orarecomple- The lattice constant is a good indicator of valence tely different. In Sm
1_,
cY
xS XPS [2] shows both the in rare earth compounds and alloys. In the specific f
6and f
5d configurations. However, the Mossbauer case of Tm chalcogenides, Iandelli [5] determined, on
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19795108
VALENCE INSTABILITIES OF Tm IN ITS COMPOUNDS AND SOLID SOLUTIONS C5-315
the basis of lattice constant and susceptibility measure- ments, that Tm was trivalent in TmS and TmSe and divalent in TmTe. These observations were made on powders synthesized at relatively low temperatures (1 400 OC). Subsequently, in a study on single crystals grown at high temperature, Bucher et al. [6] confirmed that the Tm ion was trivalent in TmS, divalent in TmTe but found a range of intermediate lattice constants in TmSe, 5.64 a, 5.71 A, which impli- ed a range of valence extending from Tm3+.
In general agreement with the lattice constant results, the high temperature susceptibility obtained by Bucher et al. [6] for TmS and TmTe approach the values for Tm3+ and Tm2+ respectively, while that for TmSe is intermediate. For temperatures which are high with respect to crystal field and any other interaction energies, the susceptibility may be used to determine average valence. Two configurations may coexist as a heterogeneous mixture of integral valence ions or in a homogeneous non integral valence phase.
In many systems the low temperature susceptibility clearly distinguishes these cases. For the heterogeneous case, one or the other configurations has a magnetic moment at low temperature and will exhibit a Curie type susceptibility or magnetic order. If neither a Curie susceptibility nor magnetic order is observed, the system is in the homogeneous intermediate valence state. A third possibility is a combination of integral and non integral valence ions.
TmSe is of particular importance because it orders antiferromagnetically at low temperatures but has intermediate values of lattice constant and suscepti- bility at high temperatures. The Tm2+(4f13) ion has a 'F,,, configuration which remains magnetic even in the crystal field. The Tm3+(4f12)3H, configuration is a T, singlet in the cubic field. However, it can deve- lop an induced moment when the exchange interac- tion is comparable to the crystal field splittings. Clear- ly magnetic order is consistant with a heterogeneous mixture of ~ m ' + and Tm3+. Varma [7] has suggested that in the case of Tm, where both ions may be magne- tic the homogeneous intermediate valence state can order magnetically.
The most detailed information on the magnetic structure comes from the neutron diffraction data of Shapiro et al. [S]. For a large lattice constant sample, 5.71 A, the order is type I antiferromagnetic with an ordered moment of 1.7
p,.This value is consistant with 2
,uBfrom high field measurements 191.
The small moment compared with the free ion values of 4 and 7
p,of Tm2+ and Tm3+, respectively, indi- cates that there is some interaction which splits the Hund's rule states. Shapiro et al. [8] interpret the low moment in terms of a homogeneous intermediate valence. Although this argument is perfectly plausible it does not exclude crystal field effects in addition to or instead of intermediate valence effects. They have also measured a sample with a small lattice constant (5.63 A), and find type 11 antiferromagnetic ordering
but only extending over regions of about 100 A.
The magnetic moment reported is 0.5
p,.This result is understandable, since in this sample Tm is essentially trivalent, the moment is exchange induced and the large vacancy concentration interferes with long range order.
The neutron diffraction measurements on the 5.71 A sample, as a function of temperature and magnetic field, have established that the magnetic phase diagram developed by Ott, et al. [l01 and Guertin, et al. [l l ] consists simply of an antiferroma- gnetic and a metamagnetic phase transition.
Mgller et al. [l21 have successfully modeled the magnetic system on the basis of mean field theory including cubic anisotropy and first and second neigh- bor Heisenberg exchange. The success of the mean field approach is consistant with the existence of a homogeneous intermediate valence but does not exclude inhomogeneities in the real system.
It is evident from the discussion above that the physical properties are strongly sample dependent.
As a consequence we have undertaken a study of the solid state chemistry of TmSe in relation to its phy- sical properties [13]. A similar investigation has been made by Batlogg et al. [14].
A series of single crystals of Tm,Se were grown with varying concentration,
X,including some samples prepared from large excesses of Tm metal. The result- ing crystals 'have a color range from bronze, at the nominal stoichiometric concentration, to a metallic blue for large Tm vacancy concentrations, similar to the Gd,Se system 1151. The samples were generally slowly cooled from 2 200-2 400 oC, depending on starting concentration, to 1 600-1 800 OC for anneal- ing and then rapidly cooled to room temperature.
Unlike most other crystals of the monochalcogenides, the TmSe crystals do not cleave cleanly along (100) planes and generally show some surface curvature.
It was, however, possible to find regions of an ingot with relatively parallel cleavage. The samples were analyzed wet chemically using ethylene diamine tartrate for the rare earth determination. The sele- nium was analyzed with a permanganate oxidation and back titration.
Figure 1 is a plot of lattice constant, a,, as a func- tion of concentration,
X,in Tm,Se. Near the stoichio- metric composition the lattice constant is 5.71 A which corresponds to a valence of 2.77, assuming a linear interpolation between 5.64 A for Tm3+Se and 5.95 A
for Tm2+Se. The rapid decrease of lattice constant
with decreasing Tm concentration demonstrates that
the increase in Tm vacancies reduces the amount of
Tm2 +. In order to separate the effect of vacancies and
changing valence we can compare the results with
those obtained on Gd,Se [l51 for which there is no
valence change. The overall decrease in lattice cons-
tant in this system is0.02 A for approximately the same
limiting concentration. A linear interpolation between
Er and Lu selenides fixes the lattice constant of
5.62
0.7 0.8 0.9 1 .O 1.1
COMPOSITION X
Fig. 1. - Lattice constant vs. composition for Tm,Se. The dots are from this work and the open triangles from Batlogg, et al. [14].
Tm3+ Se at 5.64 h;. We expect, therefore, that pure Tm3+Se with vacancies would have a lattice constant of 5.62 h;.
In the solid solubility range (0.79 5
X5 l ) samples of specific composition were obtained by adjusting the starting concentration to the desired value.
Adjacent regions of a single crystal were used for chemical and X-ray analysis. In an attempt to esta- blish the Tm rich phase boundary, crystals were grown from melts with up to 50 % excess Tm. These crystals had fine lines of pure Tm metal precipitate.
Crystals used for analysis were carefully examined microscopically and cleaved to exclude the regions containing Tm precipitate. Resulting analysis showed that in all cases the samples were deficient in Tm.
The data of Batlogg et al. [14], included in figure 1, also indicate a strong dependence of Tm valence on composition. The discrepancy between the two sets of data are possibly due to differences in thermal treat- ment and analytical techniques.
The peculiar growth habit of TmSe was also observed in the Sm, _,Y,S system for compositions above the critical concentration (X 2 .16), i.e., in the collapsed phase. In the SmYS system a change in average valence occurred a t elevated temperatures so that during the cooling cycle crystals become severely strained [16]. We have begun a study of the Tm-Se phase diagram using a tungsten mesh furnace and WReo,,,-WRe,,,, thermocouples [l 71. The pre- liminary results indicate that the homogeneity range extends to a eutectic at 1 350 OC on the Tm rich side and to -- 1 590 OC in the Se rich region. The extent of solid solubility has not yet been determined for all temperatures, i.e. the shape of the solid solution phase boundaries. We do observe, however, a transi- tion in the solid solution region ranging from
- 1 640 OC on the Tm rich boundary to approxima- tely 1 7300C at the Se rich phase boundary. We have not yet determined whether this transition is related to the valence instabilities in this system. In the absence of full phase equilibrium data, samples were prepared under varying conditions. These experi- ments resulted in a series of crystals which span the Tm,Se solid solution field, figure 1. The scatter indi- cates that the variation in thermal treatment can have a significant effect on crystal parameters. With the determination of the full shape and extent of the solid solution phase boundaries, we hope to be able to systematically vary the composition but also establish valid annealing temperatures in order to obtain equilibrium vacancy distributions.
The susceptibility has been measured with a Faraday balance and is shown in figure 2. To the extent that the data follow a Curie, Weiss law at high temperature, the molar Curie constant, C,, should not be significantly changed by crystal field or inter- mediate valence effects and should be a measure of the average valence. The lattice constant plotted against C, for Tm,Se, figure 3, shows a strong cor- relation between these variables, each of which should be a measure of the valence. In fact, with decreasing lattice constant, C, tends to the Tm3+
free ion value of 7.15 as a, approaches 5.62 h;, the value expected for Tm3+Se with large vacancy con- centration. Unexpectedly, a, and C, give rather
I I l l I I
0 50 1M) 150 200 250 300
TEMPERATURE ( K )
Fig. 2. - Inverse magnetic susceptibility vs. temperature for the Tm,Se system. Each curve is displaced upwards by 10 units from the one below. The lattice constant for each sample is given on the figure in the same order as the curves. The lines are the Curie-Weiss fit which yield the C, shown in figure 3.
VALENCE INSTABILITIES OF Tm IN ITS COMPOUNDS AND SOLID SOLUTIONS C5-317
C, (EMU / MOLE Tm)
Fig. 3. - Lattice constant vs. Curie constant, C,, for the Tm,Se system. The dots are derived from the slope of the susceptibility between 77 K and 293 K, while the crosses are from the data above 293 K shown in figure 6. The triangles are after Ref. [14].
different values for the valence. For example, the nearly stoichiometric samples have a valence of 2.77 as determined from a lattice constant of 5.71 A. On the other hand, the valence is 2.46 as determined from the CM of 4.70 and the free ion values. In this case the C, was determined from the data between 77 and 300 K. However, using the data from 400 to 800 K (see figure 6), CM is 5.22 giving a valence of 2.58. Part of the discrepancy between the valence determined from negative deviations of a, from Vegard's law may result from the greater compressi- bility of divalent rare earth monochalcogenides com- pared to related trivalent compounds.
In addition the determination of CM is imprecise because of the curvature in X-' vs. T, which may be due to the Kondo effect or intermediate valence.
The inverse susceptibility shows a downward curva- ture below 50 K, increasing with divalent character, figure 2. This deviation from the Curie-Weiss law together with a large negative
8implies a reduction in the moment, as seen in crystal field, and Kondo systems. It appears that
8is not determined primarily by exchange interactions.
The room temperature resistivity is approximately 200
p51.cm and not dependent on composition.
With decreasing temperature the resistivity increases gradually to the Nkel temperature, 1.8 2
T,2 3.5, and then increases more rapidly. The resistivity at 5 K is a measure of the low temperature scattering just above the ordering temperature and is plotted in figure 4 as a function of lattice constant. The resistivity tends to zero as the lattice constant decreases to 5.62 A, the lattice constant estimated for Tm3+
with vacancies. In this region the resistivity corres-
ponds to 10-20 p51 . cm per percent Tm2+ as if ~ m ~ + was in a singlet Kondo ground state. The most trivalent sample shows a normal metallic resistivity.
Fig. 4. - Resistance at 5 K vs. lattice constant for Tm,Se. Dots this work, triangles are after Ref. [14].
200
For samples having the most divalent character, the resistivity increases abruptly below T,. At very low temperatures a nearly stoichiometric sample gave a resistivity of 15 000 pQ. cm which is indicative of the localization of most of the conduction elec- trons [l 81. The divergence of the resistivity is observed only near stoichiometry where the valence is about 2.5 from CM, but somewhat larger from a,. It is of interest to establish whether this very unusual effect is due to balance between the divalent and trivalent concentrations, or the reduction in vacancies near stoichiometry or both.
The large low temperature resistivity is associated with antiferromagnetism. The transition to ferroma- gnetism under field leads to a normal metallic beha- vior [19]. These effects can be due to a new ground state below T,.
We have also made solid solutions of TmSe with non magnetic hosts to determine the effects on Tm valence instabilities of varying concentrations and environments. YSe has a lattice constant,
a, z5.71 A,
which is equal to that of stoichiometric TmSe. Lattice constants for these solid solutions, figure 5, all have this same value. At very low Tm concentrations it is difficult to estimate the valence. For larger concen- trations the valence is - 2.8 and independent of concentration. The molar Curie constant is interme- diate between the free ion values for all concentrations.
These results demonstrate that intermediate valence behavior persists even to the single impurity limit.
Further evidence for the single impurity effect can be found in the low temperature measurements of
-
*
-0 ' l I l I l I I I
5.62 5.64 5.66 5.68 5.70 5.72
LATTICE CONSTANT ( A )
TrnSe COMPOSITION X Y Se
Fig. 5. -Lattice constant vs. composition for the Tm,_,Y,Se system. The TmSe value shown corresponds to a nearly stoich~o- metric sample. The lines drawn to 5.95 8, and 5.64
A
correspond to the values expected for purely TmZ + and Tm3 + , respectively.Berger et al. on dilute Tm in YSe [20]. They find both Curie and Van Vleck behavior in the low temperature susceptibility and magnetization, proportional to Tm concentration.
The reciprocal susceptibility vs. temperature of the Tmo,,oYo~,,Se sample is shown in figure 6. The data
-
-
Tm,Y,Se
- T m 7 7 Y 2 3 S e
-
Tm, La, Se
-
p
-
TEMPERATURE ( K )
Fig. 6. - Inverse magnetic susceptibility vs. temperature for the Tm,-,Y,Se system, squares, the Tm,-,La,Se system, tr~angles ; and the Tm,Se system, dots. The labels on the figure are in the same order as the curves. Each curve is displaced upwards from the one below by 20 units. The temperature axis is the same for all curves.
The lines are a Curie-Weiss fit to the data. Their termination at zero inverse susceptibility gives the Curie-Weiss B.
above 300 K follows a Curie-Weiss law with C,
=5.69, corresponding to a valence of
W2.7 which is also the valence calculated from the lattice constant, 5.724 A. The large downward curvature below 300 K signifies that there is a reduction in moment. If the moment reduction is due to a change in valence, the lattice constant would have to increase on cooling, contrary to observation [2 l]. Alternati- vely, the low moment could results from crystal field, Kondo effect or intermediate valence inter- actions.
Also shown in figure 6 is the susceptibility of Tmo~~oLao,,oSe which gives a C,
=5.60 corres- ponding to a valence of 2.66. Pure LaSe has a lattice constant a,
=6.06 A. Comparing this value to a0=6.03 A for Tmo~,oLao~,oSe we find a valence of 2.61 which is in agreement with the valence based on C,. It should be noted that similar to TmSe but, unlike the Tm,-,Y,Se system, there is very little curvature in the temperature dependence of the inverse susceptibility (figure 6). This is a curious result, particularly recalling that the valence deduced from C, (at high T) and a, are in agreement for the alloy systems but not for TmSe.
The resistivity of several samples in the Tm, -,Y,Se system are shown in figure 7. At low Tm concentra-
TEMPERATURE ( K )
Fig. 7. - Resistance, normalized at ambient, vs. temperature for the Tm, -,Y,Se System. The sample compositions and room tem- perature resistivities in pO.cm are, in order from the top of the figure : Tm,
,,,,
Se, 180 ; Tm,,,,Y,,,,Se, 237 ; Tmo.sYo.,Se, 114 ; Tmo,4Yo.6, 120; Tm,.,Y,,,Se, 6 5 ; Tmo.oo3Yo.997~e, 42. The Tm,.,,,Se curve has been reduced by a factor of 3.VALENCE INSTABILITIES OF Tm IN ITS COMPOUNDS AND SOLID SOLUTIONS C5-319
tions a Kondo effect is observed. The resistivity mini- mum and log T behavior persist to at least 25 m01 %.
The slope of the log T region is roughly proportional to the Tm ion concentration consistent with the single impurity behavior based on magnetic properties. A detailed discussion of the Kondo effect in the yttrium system is given by Berger
et al. [20]. The Kondoeffect is also observed in the Tm,~,oLa,~,oSe sample.
The overall dependence of the resistivity with Tm concentration (figure 7) shows a gradual transition from Kondo type behavior t o the complicated beha- vior of stoichiometric TmSe discussed earlier. For Tm rich samples the low temperature resistivity increases much more rapidly than the Tm concentration, indicating a tendency toward electron localization in addition to increased scattering.
In both TmSe and its solid solution systems there is strong evidence for a reduction in moment. The origin of the reduction can be intermediate valence, Kondo, or crystal field effects. Early neutron diffrac- tion measurements failed to see crystal field levels in any of the TmX compounds [6]. Subsequently, Mook 1221 found a series of crystal field lines in the neighborhood of 1 meV in TmS. Lowenhaupt and Holland-Moritz [23] have observed an inelastic neu- tron scattering line in TmSe (ao= 5.688) which may be due to a crystal field effect [24], however, there is no indication of any crystal field levels between 10 and 300 K from the elastic constants [25]. Luthi,
et al. [26]have observed the effects of crystal field splitting on the elastic constants of both TmTe and TmSb. For Tm2+Te they find an overall splitting, T,-T,, of 16 K.
Using inelastic neutron scattering [27], all the crystal field levels have been seen in Tm3+ Sb, a model crystal-field-only system. The T, ground state lies 31 K below the first excited state and the overall splitting (T1-r4-T$2)-T2-T11)-T3) is 249 K. These results are in reasonable agreement with those obtained from elastic constants [26] which gave a T,-T4 splitting of 25 K.
The Mossbauer results of Triplett,
etal. [28] on stoichiometric (ao
=5.71 A) TmSe show a single broadened 6 line hyperfine spectrum. Triplett and Dixon [29] have observed in addition to this spec- trum a second extremely broad spectrum on a Tm,,,,Se sample. For a nearly trivalent Tm,,,Se sample, this second very broad spectrum is the domi- nant feature and must, therefore, be associated with the Tm3+ induced moment. The large distribution of hyperfine fields is the result of environmental effects due to the large number of vacancies. It follows that the second spectrum in the Tm0.,,Se sample is also due to Tm3+ near vacancies while the first spectrum is similar to that observed in the stoichio- metric sample. Triplett and Dixon suggest that since this first spectrum has only a small quadrupole contribution, TmSe must be primarily trivalent with no more than 20 % Tm2+ character. These results suggest that the spectrum arises from a homogeneous
intermediate valence, or Tm2+ and Tm3+ spectra with nearly the same hyperfine field, or from Tm3+
alone with Tm2+ unobserved.
Huang and Sugawara [30] have observed a broad temperature independent EPR line of Gd3
+in TmSe and interpreted their results in terms of interconfigu- ration fluctuations. Gambke, et al. [24] argue that this EPR observation is due to static inhomogeneities.
The compressibility of TmSe [l41 is about twice as large as that of SmS a t zero pressure [31]. This soften- ing is expected for intermediate valence systems but does not preclude a mixture of valence states.
XPS measurements [32] clearly show the presence of both configurations. Furthermore, the Tm2+
emission near the Fermi level is evidence of unstable valence. However, neither of these observations distinguish between homogeneous and heterogeneous systems.
It is clear from the results of this and other studies that two valence states coexist in Tm,Se and that the material orders magnetically at low temperature. We have shown that the valence is directly dependent on the Tm concentration. The range of solid solubility extends from
X x0.79, a composition for which Tm is essentially trivalent, to
X zz1, where the Tm valence is non integral. This paper establishes some of the conditions for the preparation and characterization of Tm,Se crystals suitable for physical measurements.
From the materials point of view two important pieces of information are still required. The first relates to the nature of the phase transition observed in thermal analysis in the 1 600-1 700 OC temperature range and its possible bearing on the formation of either a mixed or an intermediate valence state during crystal growth. The second relates to the development of information on microscopic structure and the effects of annealing on the distribution of vacancies or defects.
We have used the TmSe-YSe alloy system to study effects of Tm dilution and found that Tm has an intermediate valence, regardless of concentration.
The striking result is that single Tm ions have inter- mediate valence character. However, the magnetic and resistivity behavior are similar to that expected for a mixture of integral valence ions. A homogeneous state is formed but its properties are not qualitatively different from the magnetism and Kondo properties of the mixture.
In the case of Tm,Se there is Mossbauer and neu-
tron diffraction evidence for inhomogeneities in the
region
X< 1 where there are vacancies. For samples
approaching stoichiometry, there is no unambiguous
proof of either the homogeneous or heterogeneous
picture, although the Mossbauer data are most easily
explained with a homogeneous model. Certainly one
of the most interesting results is the extraordinarily
sharp rise in resistivity below
TN as vacancies arefilled and the Curie constant approaches the mean of
the Tm2+ and Tm3+ values. This large resistivity
may be an indication of a non integral valence ground for transport measurements, H. Lilienthal for magne- state with all Tm ions equivalent. tic data, and F. Cardone, J. Kuptsis and B. Ols(on for chemical analysis. We thank B. B. Triplett for exten-
Acknowledgments. -The authors are indebted to sive discussion and unpublished data. We thank the the following
:P. G. Lockwood and L. Nesterovskaya authors of ref. [l81 and [20] for many valuable discus- for sample preparation and X-ray data, J. Rigotty sions.
References
[l] MAPLE, M. B. and WOHLLEBEN, D., Phys. Rev. Lett. 27 (1971) 511.
[2] POLLAK, R. A., FREEOUF, J. L., HOLTZBERG, F. and EAST- MAN, D. E., P h p . Rev. Lett. 33 (1974) 820.
[3] COHEN, R. L., EIBSCHUTZ, M. and WEST, K. W., Phys. Rev.
Lett. 24 (1970) 383 ;
COEY, J. M. D., GHATAK, S. K., AVIGNON, M. and HOLTZ- BERG, F., Phys. Rev. B 14 (1976).
[4] See for example, Valence Instabilities and Related Narrow- Band Phenomena (Plenum Press, N.Y.) Ed. by R. D. Parks - (1976).
[S] IANDELLI, A., Rend. Acca. Sci. Fis. Mat. Napoli 37 (1964) 160;
IANDELLI, A. and PALENZONA, A., Colloq. Int. Cent. Natl.
Rech. Sci. 1 (1967) 397.
[6] BUCHER, E., ANDRES, K., DI SALVO, F. J., MAITA, J. P,, GOS- SARD, A. C., COOPER, A. S. and HULL, G. W. Jr, Phys.
Rev. B 11 (1975) 500.
[7] VARMA, C. M,, Rev. Mod. Physics 48 (1976) 219, ref. [4], page 20 1.
[8] SHAPIRO, S. M., M~LLER, H. B., AXE, J. D., BIRGENEAU, R. J.
and BUCHER, E., J. Appl. Phys. 49 (1978) 2101.
[9] CHOUTEAU, G., HOLTZBERG, F., MIGNOT, J. M,, PERA, 0 . and TOURNIER, R. (Ref. [4] page 483).
1101 OTT, H. R., ANDRES, K. and BUCHER, E., AZP Conf: Proc. 24 (1974) 40.
[l11 GUERTIN, R. P,, FONER, S. and MISSELL, F. P., Phys. Rev.
Lett. 37 (1976) 529.
[l21 M@LLER, H. B., SHAPIRO, S. M. and BIRGENEAU, R. J., Phys.
Rev. Lett. 69 (1977) 1021.
[l31 HOLTZBERG, F. and TOURNIER, R., U.S. Japan Seminar P. 232 (Oct. 1977, unpublished).
[l41 BATLOGG, B., KALDIS, E. and O n , H. R., Phys. Lett. 62A (1977) 270.
[l51 HOLTZBERG, F., CRONMEYER, D. C., MCGUIRE. T. R. and
VON MOLNAR, S., NBS Special Publication 364 Solid State Chem. Proceedings of 5th Mat. Res. Symp. 1972, p. 637.
[l61 TAO, L. J. and HOLTZBERG, F., Phys. RW B 11 (1975) 3842.
[l71 PLASKETT, T. and HOLTZBERG, F. (TO be published).
[l81 HAEN, P., HOLTZBERG, F., LAPIERRE, F., MIGNOT, J. M. and TOURMER, R., TO be published.
[l91 HAEN, P., HOLTZBERG, F., LAPIERRE, F., PENNEY, T. and TOURNIER, R. (Ref. [4], p. 495).
[20] BERGER, A., HAEN, P., HOLTZBERG, F., LAPIERRE, F., MIGNOT, J. M,, PENNEY, T., PEEA, O., TOURNIER. R.
(This conference) J. Physique Colloq. 40 (1979) CS.
[21] ANGELELLO, J. and LA PLACA, S., Private Communication.
[22] MOOK, H., Private communication.
[23] LOEWENHAUPT, M. and HOLLAND-MORITZ, E., J. Appl.
Phys. 49 (1978) 2124.
[24] GAMBKE, T., ELSHNER, B. and HIRST, L. L., Phys. Rev. Lett. 40 (1978) 1290.
[25] OTT, H. R., LUTHI, B. and WANG, P. S., Ref. [4], page 289.
[26] LUTHI, B., AIP Con$ Proc. 34 (1976) 7.
[27] BIRGENEAU, R. J., BUCHER, E., PASSELL, L. and TUBER- FIELD, K. C., Phys. Rev. B 7 (1971) 718.
[28] TRIPLETT, B. B., DIXON, N. S., MAHMUD, Y. and HANNA, S. S., Workshop on New Directions in Mossbauer Spectroscopy (Argonne 1977) AIP Conf. Proc. NO 38, Ed. G. T. Perlow, page 118.
[29] TRIPLETT, B. B. and DIXON, N. S., Private communication.
[30] HUANG, C. Y. and SUGAWARA, K., J. LOW Temp. Phys. 28 (1977) 229.
[31] PENNEY, T. and MELCHER, R. L., J. Physique Colloq. 37 (1976) C4-275.
[32] CAMPAGNA, M., BUCHER, E., WERTHEIM, G. K., BUCHA- NAN, D. N. E. and LONGINOTTI, L. D., Phys. Rev. Lett. 32 (1974) 885.
