Effect of lanthanide substitution on superconductivity of La1.85−xRxCa1.15Cu2O6−δ (R  Pr, Nd and Y)

PHYSICA ELSEVIER

Physica C 268 (1996) 279-286

Effect of lanthanide substitution on superconductivity of Lal.85_xRxCal.15Cu206_ (R = Pr, Nd and Y) Minoru Takemoto a,19 Naoki Ohashi a,* 9 Takaaki Tsurumi Junzo Tanaka b

a9

OSalTIUFukunaga

a

a Department of Inorganic Materials, Faculty of Engineering, Tokyo Institute of Technology, 2-12-10-okayama, Meguro, Tokyo 152, Japan b National Institute for Research in Inorganic Materials, 1-1 Namiki, Tsukuba, lbaraki 305, Japan Received 12 March 1996; accepted 23 May 1996

Abstract We have prepared solid solutions LaLss_xRxCas.15Cu206_~(R = Pr, Nd, Y) by annealing at l l00°C in 30 atm of oxygen atmosphere and measured their superconducting transition temperature T~. Single phase solid solutions were obtained in the compositional region of x < 1.2 for R = Pr, Nd and x < 0.2 for R -- Y. The highest T~ was observed at 25 K for the sample of no R substituted ( x = 0.0), and Tc decreased with increase in x for all samples. No superconducting transition was observed in the compositional range, x > 0.6 for R = Pr, x > 0.2 for R = Nd and x > 0.08 for R = Y. Ionic valence of Cu ions (2 + p) detected by iodometric titration method were almost constant ((2 + p) ~ 2.06) in all samples. Structural refinement revealed that La, R and Ca ions were redistributed on eight- fold oxygen coordinated M(I) site with change in x. It was found that the decrease in Tc was correlated with an average ionic valence v(1) in M(1) site, and the suppression of T~ was observed at v(l) ~ 2.1 independently with R ions. We concluded that the origin of the T~ decrease in these system is mainly due to change in local disordering of ionic valence at M(1) site. Keywords: La2SrCu206-type; Superconductivity; Rietveld analysis; Occupation factor

1. Introduction In the La2SrCu206-tYpe structure (Fig. 1), lanthanum and strontium ions can be substituted by other lanthanide ions and alkaline earth ions [1-7]. These cations occupy both eight fold oxygen coordinated M(1) site and nine-fold oxygen coordinated

* Corresponding author. Fax: + 81 3 5734 2514; e- mail:[email protected]. 'Present address: Department of Applied Chemistry, Kanagawa Institute of Technology, 1030 Shimo-ogino, Atsugi, Kanagawa 243-02, Japan.

M(2) site. Smaller cations preferably occupy the M(1) site and larger cations preferably occupy the M(2) sites, so that the cationic substitution induces cation redistribution and change in occupation factors on both sites. Superconducting La2SrCu206-type cuprate can be obtained on the following two conditions: (1) annealing in high pressure oxygen and (2) containing Ca ion instead o f Sr ion. In the previous paper [8], we have reported that superconducting transition temperatures T~ were decreased with increasing x in Lalss_xR~Caj.15 Cu206_ 8(R = Pr, Nd). W e have observed no change of the oxygen content(6 - 8 ) with increasing x. The

0921-4534/96/$15.00 Copyright © 1996 Elsevier Science B.V. All fights reserved. PH S 0 9 2 1 - 4 5 3 4 ( 9 6 ) 0 0 3 8 8 - 7

280

M. Takemoto et aL / Physica C 268 (1996) 279-286

origin of decrease in Tc cannot be simply explained by the decrease in O(2p) hole concentration because oxygen content is generally considered to be an indicator of the hole concentration. Kinoshita et al. [9] have investigated annealing effect of LaL85Ca115CU2Oy and showed that the sample annealed under 400 atm O 2 was a superconductor(Tc = 40 K) but the sample annealed under 1 atm O 2 was a nonsuperconductor. They have showed that the superconducting sample has y = 6.04, while the nonsuperconducting sample has y = 5.99. They have also examined the sample of Lal.85Srl.15CU2Oy annealed under 400 atm O 2 and obtained values of y = 6.38. It should be noted that the Sr containing sample showed nonsuperconducting but metallic character of electrical conduction. Combining the previous results by the present authors [8] with those by Kinoshita et al. [9], it is generally expected that no direct relationship exists between oxygen content y (or 6 - ~) and the superconducting Tc for the La2SrCu206 type compounds. In the present study we will discuss on the origin of decrease in Tc in relation to local structure in La2SrCu206-type superconductor.

2. Experimental The starting powder materials, La203 (99.99%), Pr6Oll (99.9%), Nd203 (99.9%), Y203 (99.9%), CaCO 3 (99.9%), and CuO (99.9%) were mixed with ethanol in an agate mortar. The mixed powder was calcined at 1050°C in oxygen gas flow for 24 h with intermediate grinding. The calcined powder was pressed into pellets of 10 mm diameter and 2 mm thick in size under pressure of 200 MPa by cold isostatic pressing. The pellets were sintered at 1080°C in oxygen flow (1 atm) for 12 h. The samples were annealed in oxygen gas under pressure of 30 atm at 1100°C for 50 h, at 700°C for 50 h and cooled down to 300°C for 50 h. Crystal phases and lattice constants of the samples were determined using an X-ray diffractometer of RAD-IIa (Rigaku Co., Japan). The lattice constants were calculated by a least square method using Si as an internal standard. Oxygen deficiency was determined by iodometric titration using KI solution and Na2S203 solution.

Electric resistivity was measured by dc four-probe method in the temperature range of 5 to 273 K. Magnetic susceptibility was measured using a SQUID magnetometer (Quantum Design, USA). Temperature dependence of magnetic susceptibility was measured under an applied field of 100 Oe for R = Pr and Nd and 20 Oe for R = Y in the temperature range between 5-35 K by zero field cooling and field cooling modes. Tc'S described in this paper were all determined by the on-set temperature. Crystal structure was refined by the Rietveld method using the program RIETAN [10]. Diffraction intensity data were collected in the region of 20 = 20-120 ° (Cu K et) at 0.02 ° interval. Two cation sites M(1) and M(2) were assumed to be occupied by Ca 2÷ and a virtual ion " L n " which consists of La 3÷ and R 3+ ions in the nominal composition, because it was impossible to refine the occupation factors of two sites simultaneously occupied by three kind of ions (La 3+, R 3+ and Ca 2+).

3. Results and discussion

The lattice constants are shown in Fig. 2. The a-axis was almost constant while the c-axis decreased with increasing x for all series. The decrease in the c-axis corresponds to decrease in average ionic radius of La 3+ and R 3+ with increasing x [11]. The c-axis decreased rapidly with increasing x for the samples substituted by R 3÷ having smaller ionic radii. The slopes of c-axis versus x, designated as dc[R]/dx, were determined to be - 1 . 7 3 x 10 -2 nm for R = Nd and - 2 . 8 0 X 10 - 2 n m for R = Y. The difference between d c [ N d ] / d x and d c [ Y ] / d x is due to the difference in ionic radii among La 3+, Nd 3+ and y3+ ion, since the ionic radius of R ion, r[R], has a relation that r[La 3+ ] > r[Nd 3+ ] > r[Y 3+ ] and, further, the oxygen content was independent of R ion as described below. As for the samples of R = Pr, there is a possibility that Pr ion takes the mixed valence state between Pr 3+ and Pr 4÷. The slope d c [ R ] / d x was determined to be - 1.34 × 10 -2 nm, which was larger than dc[Nd]/dx. Supposing dc[R]/dx depends on r[R], this result indicates that the ionic radius of Pr is intermediate between r[La 3÷ ] and riNd 3÷ ]. It is consequently considered that the Pr ion is almost trivalent; this result is consistent to

M. Takemoto et a l . / Physica C 268 (1996) 279-286

• Cu • M(1) [] M(2) C) o(1)

•

0(2)

A 0(3)

Fig. 1. Crystal structure of LaLss_xRxCal.15Cu206_ ~ (R = Pr, Nd and Y).

the magnetic susceptibility measurement for Prl.9Srl.jCu206+~ reported by Ohyama et al. [12]. Solid solution with La2SrCu206-type structure was obtained in the composition region of 0.0 < x < 1.2 for R - - P r , Nd and 0 . 0 < x < 0 . 2 for R = Y in La1.ss_xRxCaL15Cu206_8. The R content x at the solid solution limit decreased when La ion was substituted by smaller R ion. As shown in Fig. 1, the La2SrCu206-type structure is constructed by stacking of fluorite layer (containing M(1) site), CuO 2 layer and rock-salt layer (containing M(2) site). Structural stability of La2SrCu206-type structure depends on matching of lattice constant of each layer (perpendicular to c-axis). Lattice parameter for rock salt and fluorite layer decreased with the increase in x, since averaged ionic radius for both M(1) and M(2) decreased with increasing x: that is consistent to the systematic decrease in c axis, as discussed above. However, length of a-axis is almost constant in all samples, i.e., d a[R]/d x ~ 0, as shown in Fig. 2. This result is inconsistent to the decrease in averaged ionic radius of cations. On the other hand, it is expected that lattice parameter of CuO 2 layer is not expanded nor compressed by the substitution of

281

La 3÷ by R 3÷, since no hole nor electron is introduced in antibonding Cudx2_y2-O(2p) orbitals in the CuO 2 layer; that is consistent to the result as d a [ R ] / d x - 0 . Thus, it is suggested that rock salt and fluorite layer substituted by smaller R ion are expanded along a-b plane because of incompressibility of the CuO 2 layer along a-b plane. As a result, the structure become unstable and solid solution limit decreased with decreasing ionic radius of R 3+ ion. The oxygen content ( 6 - 6) was constant and independent of x: ( 6 - 6 ) was 5.984+0.005 for R = Pr, 5.986 + 0.005 for R = Nd and 5.979 + 0.009 for R = Y, respectively. Thus, ionic valence of Cuions are calculated to be 2.059 ___0.005 for R --- Pr, 2.061 + 0.005 for R = Nd and 2.054 _ 0.009 for R = Y, which are constant and have almost no relation to kind of R-ions. In La2SrCu206-type cuprates, the oxygen content changed with cation substitution in solid solution systems. Adachi et al. [13] have reported that the oxygen content (6 - 6) was 6.01 for LaLsSr0.2CaCu206_~ and 5.95 for LaL3Sr0.7 Ca0.sY0.sCu206_ 8, respectively. Annealing conditions such as temperature, oxygen pressure and annealing time can affect the value of oxygen content. The constant oxygen content in the present samples probably resulted from long duration of annealing under 30 atm oxygen. Temperature dependence of electric resistivity is shown in Fig. 3 for superconducting samples. The

1,94~" "

~

.

•

.

.

.

.

.

.

i

.

.

.

.

.

1.93

1.9 H?YI, . . . .

0

0.5

1.0

1.5

x

Fig. 2. Lattice parameters a and c of Lal.ss_~R~CaHsCu206_~ (R = Pr, Nd and Y). The symbols r,, [] and C, represent samples for R = Pr, Nd and Y, respectively.

282

M. Takemoto et a l . / Physica C 268 (1996) 279-286

-: otJ 0

E 4

,?, o

,-

~-1

0

(a) R=Pr x = 0.00 O 0.10@ 0.20 • 0.60 n

0

3

~. 2

0

'

i~O'l

'(b).

_

I

I

I

I

I

0

i

o,,

E 6 0 m, o

I

-3

0

. o 'o

4

.Z-2

0.0

-3

earl m

'~

4

[.

,

.!14,ToO?

l0 20 30 40 50

,I I"

-30

10

T/K Fig. 3. Temperature dependence of electric resistivity for LaL85_xR~Cat 15Cu206_ ~ with (a) R = Pr, (b) Nd and (c) Y.

sample with x = 0.0 showed the highest T~ (25 K). Tc decreased with increasing x and superconductivity disappeared at higher doping level for all series. Semiconductive behavior at lower temperature became to be remarkable with increase in R content x above T~. Fig. 4 indicates the temperature dependence of magnetic susceptibility. T~ was reduced with increasing x in all series. Fig. 5 shows the relation between Tc and x in L a l . 8 5 _ x R x C a l . l s C U 2 0 6 _ s. Tc decreased approximately linearly with increasing x in all series. The critical content x~ at which superconductivity disappeared was estimated to be 0.65, 0.3 and 0.1 for R = Pr, Nd and Y, respectively. T~ decreased more

0

o.o2 •

.

0

0

x =0.00

if'

r

--

0.040

~.u

~,o -21

20 T/K

30

Fig. 4. Temperature dependence of magnetic susceptibility of Lalss_xRxCaHsCu206_,~ with (a) R = P r , (b) Nd and (c) Y. Open and closed circles represent the data collected by zero-field cooling mode and field cooling mode, respectively.

3 0

-

•

-

•

-

•

-

•

-

254 2O 15 10 5 0 0.2 0.4 0.6 0.8 1.0 X

Fig. 5. Tc vs. R ion content x for Lal.85_xRxCaHsCu206_8 (R = Pr, Nd and Y). Open dots correspond to the Tc determined by electric measurement (Fig. 3) and closed dots correspond to those determined by magnetic measurement (Fig. 4). Solid lines are obtained by least-square fits.

M. Takemoto et al. / Physica C 268 (I 996) 279-286

steeply, that is, the slope I d T J d x l r increased when R-ion has smaller ionic radius; IdTc/dxlp~ = 37 K, IdTc/dXlNd ----81 K and IdTc/dxl v = 248 K. Suppression of Tc by the substitution of lanthanide ions is also seen in (Y,Pr) Ba2Cu307_ ~ [14-16]. Takenaka et al. [17] have explained that the suppression of Tc in this compound is due to hybridization of Pr(4f) orbital with O(2p) orbital in CuO 2 plane and this hybridized orbital consumes holes belonging to the CuO 2 plane by forming tetravalent Pr 4÷ ion. However, the Pr ions in the present system are estimated to be Pr 3÷ state from the [dc/dXlR tendency. Besides, trivalent Nd 3+ and y3+ are very stable. In the LaL85_rRxCaLlsCU 2 06_ 8(R = Pr, Nd and Y) samples, thus, the change in T~ cannot be explained by decreasing hole concentration due to the R(4f)-O(2p) hybridization. The Y124-type superconductors RBa2CuaO 8 (R = lanthanide ions and Y ion) show decrease in Tc with increasing ionic radius of R ion [18,19], which is opposite tendency to the present samples. From structural analysis, it was reported that anisotropic shrinkage of C u - O bond lengths in CuO2-1ayer with changing kind of R ion plays most responsible roles for increasing Tc in Y 124-type superconductors [20]. Large positive pressure dependence of Tc in YBa2Cu408 (R = Y) has been reported [21-25]. In this case, the increase in Tc is due to anisotropic shrinkage of the bond lengths between plane Cu ion and apical O ion in CuO 5 pyramid. T~ of cuprate superconductor can be affected by local structural change. Therefore, we analyzed local structure in LaL85_xRxCaHsCu206_ ~ (R = Pr, Nd and Y) samples by the Rietveld refinement method. Refined crystallographic parameters and bond lengths are listed in Table 1 for selected samples of LaL85_xRxCaLtsCu206_8 (R = Pr, Nd and Y). Refinement of precise distribution for La 3+, R 3+ and Ca 2÷ in these sites has not been achieved in this work. However, the occupation factors, listed in Table 1, are considered to be in agreement with the actual distribution of the trivalent ion (La 3÷ and R 3+) and divalent ion (Ca2+), since atomic scattering factors of " L n " ions are much larger than that of Ca ion. Ca-ion preferably occupied M(1) site in LaL85Cal.15Cu206_ ~ ( x = 0.0). The occupation factor of Ca in M(1) site decreased and that in M(2) site increased with increasing x, and the occupation

283

factor of " L n " in M(1) site correspondingly increased. Accordingly, we can estimate the average valence of ions in M(1) site, v(1), as a measure of the mixed state of divalent ion (Ca 2÷ ) and trivalent ions (La 3÷, R 3+) at M(1) site. These results correspond to the preferential occupation of smaller ions in M(1) sites. Distance between M(2) and Cu parallel to c-axis was remarkably decreased with the increase in x and decrease of ionic radius of R ion. This result means the decrease in c-axis in due to the decrease in M(2)-O distance a n d / o r Cu-O(2) distance. The Cu-O(2) distance showed a tendency to decrease with increasing x, even if the large error was taken into account. On the other hand, M(2)-O(2) distance parallel to c-axis showed no systematic relation with x. The M(2)-O(2) distance was considered to be not compressible because of structural stability of rock salt layer, since the distance (0.230-0.244 nm) was extremely shorter than both M(2)-O(2) distance not parallel to c-axis (0.274-0.276 nm) and M(2)-O(1) distance (0.262-0.266 nm). Thus, it is indicated that the decrease in c-axis by the substitution was mainly due to the decrease in Cu-O(2) distance: this means that CuO 5 pyramid is anisotropically compressed. Izumi [26] has reported the anisotropic compression of CuO 5 pyramid in LaL89CaLl~Cu206-type compound under high pressure. Pressure effect on Tc has been measured in La2_xCal+xCu206 and their T~'s were not so dependent on pressure [27]. We could not make a simple comparison between La2SrCu206-tYpe and Y124-type superconductor, because the latter has CuO-chain as charge reservoir which supplies holes to CuO 2 plane. However, the length between plane Cu ion and apical oxygen ion is considered to be one of the most striking parameters for Tc. However, no apparent change in length of Cu-O(2) was found because this estimation includes large error. More precise structural analysis is necessary to discuss the effect of bond length on Tc in detail. Among crystallographic parameters, the occupation factors at M(1) site changed most remarkably. We adopted average valence of cations, v(1), at the M(1) site as a parameter controlling superconducting T~. As shown in Fig. 6, the Tc versus v(1) dependence can be plotted by a common curve for all lanthanide substitutions examined in the present

M. Takemoto et a l . / Physica C 268 (1996) 2 7 9 - 2 8 6

284

~

N

o

~

d

d

N

d

o

d

dSd

d

~ 1

= ~ ~

. ;

~

d 2 ~ d g

~~

~

~

d~

6

~

o

o

--

r.,--

~

~ d

~ ~

d

d

~ e

~

d

m d

o

o

~

~

o

= d

~ d

~

~

~

d

Z 11

d

o d

o •

E

~

o

~

~

~

~.,

~

~

"

~

"

<=;

0

0

M. Takemoto et al./ Physica C 268 (1996) 279-286 3

0

,

-

-

-

,

l--~.

/

")L

20~

-

,

• Nd

'X

i

-

APr * Y

,=

2.05

2.10

2.15

v(1) Fig. 6. The relation between Tc and v(1), which is average valence of ions in the M(1) site.

study. The relation in Fig. 6 suggested the following aspects: (1) The suppression of Tc with v(1) is independent of the R ions. (2) There is a critical value of v ( 1 ) ~ 2.1 where Tc goes to zero. The relation will be changed by the chemical composition of C a / S r and oxygen stoichiometry 8 in La2SrCu206 type superconductor. For example, Kinoshita et al. [9] have shown that g[La(1)]= 0.09(v(1)=2.09) in superconducting Lal.85Cal.15 Cu20 y (Tc = 50 K), while g[La(1)]=0.56(v(1)= 2.56) in nonsuperconducting Lal.ssSrl.15CU2Oy. Although the general trend suggested that smaller v(1) value gives higher T~, a critical v(1) value to suppress superconductivity will be changed by the character of the samples, i.e. chemical composition and oxygen content. The increase in v(1) generated the disorder in CuO 2 layer which suppressed Tc; this phenomenon has been also observed in R = Nd sampies, LaLss_xNdxCaLlsCu206_ 8, from a viewpoint of temperature dependence of Hall effect [28]. Therefore, the occupation factors are effective parameters to be modified in order to obtain higher Tc in La2SrCu206-type superconductor, and local structural analysis and superconducting Tc in this superconductor present new aspect to understand superconducting transition in cuprates.

4. Conclusion We have prepared superconducting solid solution, Lal.85_xRxCal.15Cu206_8 ( R - - P r , Nd and Y) and

285

studied effect of R substitution on superconductivity. Smaller R content at solid solution limit was resulted in smaller ionic radius of R ion. The a-axis was constant and c-axis decreased with increasing x for all R substituted system examined in the present study. Superconducting Tc decreased in all samples with increasing R content x. The suppression of T~ in this system cannot be explained by simple scheme of hole concentration originated from oxygen content, because oxygen contents were 5.98 ~ 5.99 for all solid solution samples including Pr, Nd and Y substituted series. The Rietveld refinement indicated that redistribution of La, R and Ca ions occurred and occupation factor of La and R ions in M(1) increased, while Ca ion in M(1) decreased with increasing x. We conclude that the origin of the suppression of superconductivity causes from the redistribution of ions in M(1) site. We defined v(1) as an average ionic valence of ions in M(1) site and found that Tc became to zero at v(1) ~ 2.1 independently of the substituted R ion.

Acknowledgements The authors wish to thank Professor S. Okuma (Research Center for Very Low Temperature System, Tokyo Institute of Technology) for useful discussion. They also thank Professor T. Atake and Dr. T. Shirakami for measurement of Meissner effect using SQUID magnetometer in The Research Laboratory of Engineering Materials, Tokyo Institute of Technology. This study was partly supported by Special Coordination Funds for Promoting Science and Technology Agency of Japan (the Frontier Ceramic Project) and Grant in Aid for Scientific Research from Ministry of Education Science and Culture of Japan.

References [1] N.Nguyen, L.Er-Rakho, C.Michel, J.Choisnet and B.Raveau, Mat. Res. Bull. 15 (1980) 891. [2] M.Hiratani, T.Sowa, Y.Takeda and K.Miyauchi, Solid State Commun. 72 (1989) 541. [3] E.A.Hayri and J.Z.Larese, Phys. C 170 (1990) 239.

286

M. Takemoto et al. / Physica C 268 (1996) 279-286

[4] D.B.Currie, M.T.Weller, S.Rowles and D.H.Gregory, Mat. Res. Bull. 25 (1990) 1279. [5] PH.Labbe, M.Ledestert, V.Caignaert and B. Raveau, J. Solid State Chem. 91 (1991) 362. [6] J.M.Navarro, A.Fuertes and P.Gomez-Romero and J.Rodriguez-Carvajal, Solid Sta~ Commun. 81 (1992) 677. [7] D.H.Gregory, M.T.Weller, C.V.N.Rao and D.B.Currie, Mat. Res. Bull. 27 (1992) 1309. [8] M.Takemoto, N.Ohashi, O.Fukunaga, H.Ikawa and J.Tanaka, Advanced Materials '93 VI/A: Superconductors, Surfaces and Superlattices 297. [9] K.Kinoshita, H.Shibata and T.Yamada, Phys. C 176 (1991) 433. [10] F.Izumi, J. Crystallogr. Soc. Japan. 27 (1985) 23 [in Japanese]. [l l] R.D.Shanon, Acta. Crystallogr. A 32 (1976) 751. [12] T.Ohyama, N.Ohashi, O.Fukunaga, H.lkawa, F.Izumi and J.Tanaka, Phys. C 249 (1995) 293. [13] S.Adachi, T.Sakurai, Y.Yaegashi, H.Yamanchi and S.Tanaka, Phys. C 192(1992) 351. [14] L.Soderholm, K.Zhang, D.G.Hinks, M.A.Beno, J.D.Jorgensen, C.U.Segre and I.K.Schuller, Nature 328 (1987) 604. [15] J.K.Liang, X.T.Xu, S.S.Xie, G.H.Rao, X.Y.Shao and Z.G.Dian, J. Phys. B 69 (1987) 137. [16] Y.Dalichaouch, M.S.Torikachvili, E.A.Early, B.W.Lee, C.L.Seaman, K.N.Yang, H.Zhou and M.B.Maple, Solid State Commun. 65 (1988) 1001.

[17] K.Takenaka, Y.Imanaka, K.Tamasaku, T.Ito and S.Uchida, Phys. Rev. B 46 (1992) 5833. [18] D.E.Morris, J.H.Nickel, J.Y.T.Wei, N.G.Asmar, J.S.Scottk, U.M.Scheven, C.T.Hultgren, A.G.Markelz, J.E.Post, P.J.Heaney, D.R.Veblen and R.M.Hazen, Phys. Rev. B 39 (1989) 7347. [19] S.Adachi, H.Adachi, K.Setsune and K.Wasa, Phys. C 175 (1991) 523. [20] K.Mori, Y.Kawaguchi, T.lshigaki, S.Katano, S.Funahashi and Y.Hamaguchi, Phys. C 219 (1994) 176. [21] E.Kaldis, P.Fischer, A.W.Hewat, E.A.Hewat, J.Karpinski and S.Rusiecki, Phys. C 159 (1989) 668. [22] B.Bucher, J.Karpinski, E.Kaldis and P.Wacher, J.LessComm. Met. 164/165 (1990) 20. [23] A.W.Hewat, P.Fischer, E.Kaldis, E.A.Hewat, E.Jilek, J.Karpinski and S.Rusieski, J.Less-Comm. Met. 164/165 (1990) 39. [24] Y.Yamada, T.Matsumoto, Y.Kaieda and N.M6ri, Japan. J. Appl. Phys. 29 (1990) L250. [25] Y.Yamada, J.D.Jorgensen, S.Pei, P.Lightfoot, Y.Komada, T.Matsumoto and F.Izumi, Phys. C 173 (1991) 185. [26] F.Izumi, Trans. Am. Crystallogr. Assoc. 29 (1993) 11. [27] Y.Yamada, K.Kinoshita, T.Matsumoto, F.Izumi and T.Yamada, Phys. C 185-189 (1991) 1299. [28] M.Takemoto, N.Ohashi, T.Tsurumi, O.Fukunaga and J.Tanaka, Phys. C 265 (1996) 171.