ON THE DOUBLE-EXCHANGE INTERACTIONS IN La 1 _xba x Mn03

Transcription

1 R 712 Philips Res. Repts 25,8-16,1970 ON THE DOUBLE-EXCHANGE INTERACTIONS IN La 1 _xba x Mn03 Abstract 1. Introduetion by F. K. LOTGERING Magnetic and electrical measurements on Lal_",Ba",Mn03 and Lal_",Ba",Mnl_",Ti",03 with 0,:;;;x':;;; 0 25 are reported. From the difference between the paramagnetic Curie temperatures of the Ti-free and Ti-containing materials with x = 0'25, the double-exchange interaction is estimated to be about 5 times as strong as the superexchange interaction. The magnetic transition temperatures for x = 0 05 are essentially not in agreement with De Gennes' magnetic phase diagram. As found by Jonker and Van Santen 1.2), the substitution of Mn 4 + in antiferromagnetic 3), semiconducting 1) LaMn03 by the replacement of trivalent La by divalent Me = Ca, Sr or Ba according to La 3 + 1_XMe2+ x I Mn H l_xmn 4 + x I leads to ferromagnetism and metallic conduction. The saturation moment at T = 0 is near to the value corresponding to à parallel alignment of the spins of the Mn H and Mn 4 + ions for compositions in a region around x = 0 3. Such a combination of ferromagnetism and metallic conduction has been attributed by Zen er 4) to "double-exchange" interactions between the MnH and Mn 4 + ions occupying octahedral sites. The theory has been extended by Anderson and Hasegawa 5.6), who pointed out that double-exchange interaction is not of the Heisenberg type but depends on cos (cpj2) where cp is the angle between the spin directions. A detailed treatment of the influence of double-exchange interactions on the magnetic and electrical properties of La1_xMexMn03 has been given by De Gennes 6,7). As shown by Jonker 2), double exchange is not essential to ferromagnetism in manganites with perovskite structure, The Lal_xBaxMnl_xTix03 materials, in which double exchange cannot occur because the manganese is exclusively present as Mn H, also show ferromagnetism. The mixed-crystal series La1_xBaxMn03 and La1_~BaxMnl_xTix03 have pseudo-tetragonally distorted perovskite structures for small x and both become cubic at x Rj The ferromagnetic-saturationmoments increase with x and approach theideal values as soon as the non-distorted, cubic phase appears 2). The aims of the present work were (1) to estimate the double-exchange contribution to the interactions in Lal_xBaxMn03 by a comparison with Lal-xBaxMnl_xTix03;

2 DOUBLE-EXCHANGE INTERACTIONS IN La 1 _ x Ba x Mn0 3 9 (2) to test experimentally De Gennes' theoretical treatment 6.7) of the transition from a semiconducting antiferromagnet to a metallic ferromagnet for Lal_xBaxMn03 with increasing x. 2. Preparation, chemical analysis and X-ray diagrams of the samples Samples of Lal_xBaxMn03 and Lal_xBaxMnl_xTix03 with 0 ~ x ~ 0 25 were prepared by usual ceramic techniques. The materials were prefired at 1000 C in air and the final firing of pellets pressed hydrostatically under a pressure of 14 kbar/cm" as specified in table 1. All samples were rapidly cooled and the surface layer, which may be reoxidized, was ground off. The preparations were well sintered and showed densities of 90 to 96 % of the X-ray densities. The Mn 4 + content of the samples was determined in the usual way by titrating the oxidizing power. The error in the values given in table I is estimated to be less than gramion Mn 4 + per mole. Chemical analysis of the first sample of LaMn03 given in table I gave an Mn content of ± gramatom/mole and an oxidizing power of ± gramatom oxygen/mole. The fact that the latter figure is smaller than the oxidizing power of (1' )/2 = gramatom oxygen/mole, corresponding to the lower limit of the Mn content, shows that the sample did not contain tetravalent manganese. The X-ray patterns for 0 10 ~ x ~ 0 25 showed a cubic perovskite phase (cell edge a') and very weak superstructure reflections that could be indexed using a tetragonal cell 8) with edges a = b = a' V2 and c = 2a'. The deviation from an ideal perovskite is very small and will be neglected. We found a' = 3 90 A and 3 95 A for the Ti-free and Ti-containing series, respectively, and a' proved to be constant within 0 02 A for 0 10 ~ x ~ The X-ray patterns for x = 0 and 0 05 could be indexed using an orthorhombic cell related 8) to a pseudo-tetragonal perovskite cell (a' = c' =1= b' ; P' near to 90 ). The lattice constants were in agreement with literature data 2.9). 3. Experimental results The Xm -l_t curves measured on the samples x = 0 and 0 05 of the Ti-free (fig. 1) and Ti-containing series are of the type reported by Jonker 2). The two straight parts correspond to the cubic perovskite phase at high temperatures and the pseudo-tetragonally distorted modification at lower temperatures.. The data given in fig. 1 for x = 0 were measured on an LaMn03 sample free from Mn 4 + (sec. 2). For temperatures below 150 "K we measured magnetization versus field curves obeying relations a = (fo + Xg H, and the values of Xg were used to draw the Xm -l_t curve in fig. 1. The latter curve is typical of an antiferromagnet with T N = 150 "K and the (fo-t curve (fig. 1, inset (a))

3 10 F. K. LOTGERING TABLE 1. Preparation and Mn 4 + content analytically determined in the samples Mn x final firing 4 + per molecule h 1350 oe in CO 2 + A + H 2 (95 : 4,5: 0'5) h 1350 oe in CO h 1350 oe in CO 2 + A + H 2 (95 : 4 5 : 0 5) h 1200 oe in CO h 1200 oe in air idem idem idem I idem x final firing per molecule h 1350 oe in CO 2 + A + H 2 (90 : 9 : 1) Mn2+ 6h 1350 oe in CO Mn h 1350 oe in CO 2 + A + H 2 (90: 9: 1) Mn h 1350 oe in CO 2 + A + H 2 (95 : 4,5: 0 5) Mn h 1350 oe in CO 2 + A + H 2 (90 : 9 : 1) Mn idem Mn idem Mn idem 0 000

4 DOUBLE-EXCHANGE INTERACTIONS IN La 1 _ x Ba x Mn X;;;I a) LaMn03 t 71~~", o er 2 er --=---- T(OK) LIO 20 5 t20~ 0 6PB/melee. b) La()ogj3a(}orfn 03 -H(kOe) LIOO T (OK) Fig. 1. LaMn03 (sample free from Mn 4 +, see sec. 2) and LaO'9SBao.osMn03' Reciprocal molar susceptibility Xm -1 and magnetization (J (gauss cm 3 g-l) measured at fields indicated as a function of temperature. Inset: (a) LaMn03: parasitic ferromagnetic saturation (Jo as a function of temperature; (b) LaO'9SBao.osMn03: magnetization (J as a function of the applied field at 4 2 "K. shows that the spontaneous magnetization (jo disappears at T N.("parasitic ferromagnetism "). All samples with x =1= 0 investigated were ferromagnetic at 4 "K. The materials Lal_xBaxMn03 with 0 10 :::;;; x:::;;; 0 25 were saturated at about 15 koe. For the other samples, however, no saturation was obtained at the highest fields applied (30 koe), and the spontaneous magnetization was determined by extrapolation to H -+ 0 (examples are given in fig. 1, inset (b), and in fig. 2). The moment determined for LaO'9SBao.osMno'9STio.os03 in this way is very small (~ 0 25,uB/molecule).The occurrence of ferromagnetism was confirmed by the observation of a slightly asymmetrie hysteresis loop at 4 5 "K on a field-cooled sample. The (j-t curves measured at 15 koe show a tail that makes a determination of Tc difficult (see e.g. fig. 2). The a 2 -T curves measured at 1 to 3 koe show a linear part that was used for the determination of Tc by extrapolation to (j (examples are given in figs 1 and 2). The paramagnetic measurements on all samples at temperatures up to 1100 "C showed that a Curie-Wei ss law holds at temperatures above ~ 500 C. The paramagnetic Curie temperature (J is found to be considerably higher than Tc. Forexample, we found Z, =230 Kand (J = 355 OKforLao.ssBao'lSMn03. This material can be strongly magnetized in an external field at temperatures far above Tc. For instance, we measuredc = 16 gauss cm" g-l (being 22 5% of the saturation at T = 0) at 20 koe and 260 OK(being 30 OKabove Tc). The a-h curve is slightly curved, but no remanence was observed, indicating that the substance was indeed paramagnetic.

5 12 F. K. LOTGERING 24 cr 20 t ID H(kOe) cr t 100 lso T(OK) Fig. 2. LaO.9Bao.lMno.9Tio.lÛ3' Magnetization a (gauss cmê g-l) as a function of the applied field at 4'3 "K and as a function of temperature. Figure 3 gives a number of magnetic and electrical properties as a function of the composition. From the log e-x curves for La1_xBaxMn03 at 120 and 290 "K it can be-seen that the temperature coefficient of (! decreases with x,and changes sign for x = The Seebeck coefficient a decreases also with x and becomes small «20!LVjdeg) for x > Such conducting behaviour is essentially similar to that reported by Van Santen and Jonker 1) for La1_xSrxMn03 and may be described as a change from semiconduction to metallic conduction at x = 0 20 to The metallic character of the conduction in the sample with x = 0 25 was confirmed by resistivity measurements down to 4 2 "K, which showed that log (! decreased from -0,92 at 293 "K to -1,25 at 78 ok and at 4 2 ok. All La1_xBaxMn1_xTix03 compositions investigated are semiconductors (fig. 3). 4. Discussion We start with an estimation of the superexchange and double-exchange contributions to the interaction in the practically cubic perovskites by comparing the paramagnetic Curie temperatures ofthe Ti-containing and Ti-free materials. Jonker 2) has remarked that the superexchangeinteractions between the Mn3+ ions is positive in cubic perovskites, and this has been discussed by Goodenough et al. 10), Havinga 11) and Bokov et al. 12). Havinga 13) has given an empirical relation between the magnitude of the Mn3+0Mn3+ interaction and the cell edge from which it can be seen that the superexchange inter-

6 DOUBLE-EXCHANGEINTERACTIONSIN Lal_xBa~xM_nO-,3:!..._ I_3 Fig. 3. Lal_xBaxMn03 and Lal_xBaxMnl_x03. Molar Curie constant Cm and ferromagnetic moment Jtf at 4 5 "K (in JtB/molecule), paramagnetic and ferromagnetic Curie temperatures o and Tc, resistivity 12(Q cm) at temperatures indicated and Seebeck coefficient IX (in!lv/deg) at room temperature as functions of Mn 4 + or Ti 4 + content. The Mn 4 + concentration is the value determined chemically. The paramagnetic quantities Cm and 0 refer to the cubic (hightemperature) phase, 0dlst to the distorted (Iow-temperature) modification. action in Lao.7sBao.2sMn03 (a A) and Lao.7sBao.2sMno.7sTio.2s03 Ca = 3 95 A) is about equally strong. The difference of 120 "K between ()= 390 "K for Lao.7sBao.2sMn03 and ()/(l - x) = 200/0 75 = 270 "K for Lao 7sBao 2sMno.7sTio.2s03 (the factor (1 - X)-l being a correction for the dilution of Mn by Ti) can then be attributed to double exchange. The ratio of the double-exchange and superexchange contributions to () is then 120/270 = 0 45 for x = Since there are x N = 0 25 N double-exchange bonds and 3 N superexchange bonds per N manganese ions, this ratio of the two contributions is 0 45 (3/0 25) F:::i 5; in other words, calculated per bond, the double exchange dominates. In the case of a cubic lattice, De Gennes' equation for the paramagnetic Curie temperature reads 7): ()= ()super + ()double = ()super + 1 6x bik, where b is the transfer integral. The equation was derived for free carriers and for x«1. These two conditions cannot be realized experimentally be-

7 14 F. K. LOTGERING cause the material is semiconducting for x ~ Applying De Gennes' equation at x = 0 25 (free carriers but x not «1) and x = 0 05 (x«1 but no free carriers), we find the effective values b = 0 03 and 0 06 ev, respectively, using the I-valuesgiven in fig. 3 (referring to the cubic phase at x. 0'05). The values found are' a factor ~ 10 smaller than estimated by Anderson 6). We will now discuss the magnetic behaviour of the pseudo-tetragonal materials with small x having a ferromagnetic moment fl-f at 4 5 "K much smaller than the value fl-s for parallel Mn spins (fig. 3). The influence of double-exchange interactions on the spin ordering for low charge-carrier concentration has been treated in detail by De Gennes 7), who distinguishes the two possibilities of free carriers (metallic conduction) or bound states of the carriers (semiconduction). In the first case the configuration at T = 0 consists of a regular spin configuration which can be derived from the antiferromagnetic layer structure in LaMn03 by canting the two sublattice magnetizations so that a spontaneous moment results. According to De Gennes the spin canting disappears at a temperature Tc below the Néel temperature T N so that the substance is antiferromagnetic for Tc < T < T N The magnetic transition temperatures are given schematically as a function of x in De Gennes' magnetic phase diagram (fig. 4). In the case of bound states, De Gennes finds at T = 0 local spin distortions in the antiferromagnetic layer structure. An irregular spin configuration results and a net moment arises from the parallel alignment of localized moments. According to De Gennes, the magnetic phase diagram can be expected to be very similar to that given in fig. 4. T t Paromagn. Fig.4. Magnetic phase diagram according to De Gennes 6, ') giving the Néel temperature TN. the ferromagnetic and paramagnetic Curie temperatures Tc and (J, and the spontaneous magnetization Jl-f at T = 0 as a function of the Mn 4 + content x (schematically). _X

8 DOUBLE-EXCHANGE INTERACTIONS IN La 1 _ x Ba x Mn We should like to remark that, whereas the occurrence of double exchange is essential in the first model, this is not the case in the second. It can easily be shown that local ferromagnetic Heisenberg coupling in the antiferromagnetic layer structure can also give local spin distortions. The ferromagnetism of LaO 9SBao osmno 9STio.os03 must be attributed to an ordering of this type because double exchange cannot be present in this material. De Gennes 7) has given an elegant explanation of certain neutron-diffraction patterns reported by Wollan and Koehler 3). These authors observed a mixture of ferromagnetic and antiferromagnetic phases on non-stoichiometric samples of LaMn03H or (La,Ca)Mn03H with 0 09 to 0 20 gramion Mn 4 + fmole, and this is precisely the pattern expected for the spin-canted model in the case of free carriers. However, it seems to us unlikely that the charge carriers in these samples can be considered free because it will be seen from both ref.land the present paper that the materials do not conduct metallically unless the Mn 4 + concentration exceeds a content of 0 20 to 0 25 gramionfmole *). In the case of bound carriers another neutron-diffraction pattern is expected 7) so that we doubt whether De Gennes' interpretation is correct. From fig. 3 it can be seen that #f«#s for LaO.9SBao.osMn03 and Lal_xBaxMnl_xTix03 with x = 0 05 and From De Gennes' diagram (fig. 4) one would then expect Tc < T N and Tc«T N (x = 0) = 150 "K. However, the X-1-T curves of these substances do not show any indication of a magnetic transition at temperatures above Tc (see, for example, fig. I), and Tc is about equal to the Néel temperature of LaMn03 (150 OK). This means that the observed behaviour is essentially not in agreement with De Gennes' model. The fact that the Tc-x curve joins smoothly with the Néel temperature of LaMn03 for both systems seems to point to a continuous transition from parasitic ferromagnetism at x = 0 (sec. 1) to normal ferromagnetism for increasing x. Since the origin of parasitic ferromagnetism in perovskites is not yet established 9), it is difficult to give an explanation of such a transition at the moment. Acknowledgement The author thanks Mr G. H. A. M. van der Steen for preparing the samples and performing the measurements. Eindhoven, May 1969 *) We also prepared LaMn03+J. containing gramion Mn 4 + fmole. Although the Mn 4 + concentration is larger than in Wollan and Koehler's samples, and the substance was found to be ferromagnetic with a saturation moment of 97 % of the theoretical value, the substance was semiconducting and showed a high resistivity at Iow temperatures (log e = 3 5 at 100 OK).

9 16 F. K. LOTGERING REFERENCES 1) G. H. Jonker and J. H. van Santen, Physica 16,337, 599, ) G. H. Jonker, Physica 22,707, ) E. O. Wollan and W. C. Koehier, Phys. Rev. 100, 545, ) C. Zener, Phys, Rev. 82, 403, ) P. W. Anderson and H. Hasegawa, Phys. Rev. 100, 675, ) P. W. Anderson, in G. T. Rado and H. Suhl (eds), Magnetism, Academic Press, New York, 1963, Vol. 1, p ) P. G. de Gennes, Phys, Rev. 118, 141, ) M. A. Gilleo, Acta cryst, 10, 161, ) J. B. Goodenough, Landolt-Börnstein, 6 Aufi., Bd. 11/9, p. 2, ) J. B. Goodenough, A. Wold, R. J. Arnott and N. Menyuk, Phys. Rev. 124,373, ) E. E. Havinga, Philips Res. Repts 21, 432, ) v. A. Bokov, N. A. Grigoryan, M. F. Bryzhina and V. V. Tikhonov, Phys, Stat. sol. 28, 835, 1968.