Technical Comments

Ferromagnetic Superlattices

Science  11 Sep 1998:
Vol. 281, Issue 5383, pp. 1571
DOI: 10.1126/science.281.5383.1571a

Kenji Ueda et al. (1) fabricated a ferromagnetic double perovskite compound La2FeCrO6, which they built as a single atomic layer superlattice along the [111] crystallographic direction of the known antiferromagnetic perovskites LaFeO3 and LaCrO3. The unusual behavior of the material bears a strong resemblance to my theoretical predictions (2) for precisely that artificial compound, so it is instructive to follow the comparison as far as the current data allow.

My objective was specifically to predict, entirely from first principles computational theory, new compounds that would have the property of being half-metallic. This designation (3) means that, in a magnetically ordered material where electrons with different directions of their spin (up and down) are not equivalent, materials with one spin direction would be metallic, while those of the other spin direction would be insulating. Although discussed for over a decade, such half-metallic behavior only now seems to have been clearly observed (4). I chose to search in the class of double perovskites, one member of which Ueda et al. has succeeded in constructing.

For La2FeCrO6, I obtained three magnetically ordered states that were electronically metastable (at least) and differed rather little in energy. The ferromagnetic one (where the Fe and Cr spins are parallel) involved high-spin Fe and Cr ions, had a total magnetization corresponding to 7.15 Bohr magnetons (μB) per Fe-Cr pair, and was not half-metallic. Two separate ferrimagnetic phases (antiparallel Fe and Cr spins) were obtained, both of which are half-metallic. The most stable one again involved high-spin ions, but because they were antiparallel, the net moment was 2 μB per Fe-Cr pair. Ueda et al. report a saturation magnetization around 6 μBper pair, suggesting that it is the ferromagnetic state that is most stable. In my calculation, one of the ferrimagnetic phases was lower in energy than the ferromagnetic phase. This extra energy gain that stabilizes the ferromagnetic phase could come from a structural distortion, which was not included in my calculations and has not yet been measured. The difference between 6 and 7.15 μB may be a result of the imperfect crystallinity of the films.

Progress on understanding these double perovskites will occur quickly if experiment and theory interact closely. I predict, for example, that this material will be conducting; transport properties have not yet been reported. Fabrication of La2FeCrO6, however, demonstrates that this class of novel materials can be stabilized in an ordered crystal structure, even if the ordered structure is thermodynamically unstable (as this one is), and even if the two cations are similar in size and have equal charges (unequal sizes and charges are known to enhance ordering tendencies). Further pursuit of new members in this class of double perovskites may turn up realizations of stoichiometric half-metallic ferromagnets.


In their report (1), Ueda et al. seem to use the magnetic units incorrectly in interpreting their magnetization data.

They state that the saturation magnetization of the LaFeO3-LaCrO3 superlattice is 2 emu/g [figure 3 in (1)]. This does not, however, correspond to a macroscopic moment size of 3 μB per site, but only to 0.09 μB per site. In fact, 1 emu/g = 1 emu/g · 10−3 Am2/emu · 481.6 g/mol · 1/2N A · 1 μB/9.27 × 10−24 Am2 = 0.043 μB per site. Moreover, one could argue that the value of the saturation magnetization is not 2 emu/g, as stated, but rather 0.2 emu/g [magnetization M at T = 100 K minusM(T > T c)]. Thus, the macroscopic moment size of the superlattice is as small as 0.009 μB per site.

This correction makes it questionable that ferromagnetic order occurs in the LaFeO3-LaCrO3 superlattice. The small macroscopic moment might well be explained with canted antiferromagnetic or ferrimagnetic order, but not with ferromagnetic order resulting from a 180° Fe3+-O-Cr3+superexchange interaction, as proposed by Ueda et al.


Response: With regard to the comment by Pickett, we have measured the conductance of our samples [LaCrO3-LaFeO3 (1/1) superlattice]. The material is an insulator, and sufficiently large saturation magnetizations (>2 μB per pair site) have been observed in our sample. We agree with the possibility that the structural distortion stabilizes the ferromagnetic state, as Pickett points out. We would like to exploit new members in the double-perovskite materials. Pickett's prediction and suggestion is encouraging for the design of half-metallic ferro(ferri)magnets.

With regard to the comment by Meijer, the saturation magnetization value of 3 μB per site was determined from the result of a saturation magnetization-hysteresis (M-H) measurement at 6 K, shown in figure 4B of our report (1). The saturation magnetization of ∼3 μB per site was estimated by using 2 · 10−5 emu and the size and thickness of our sample (1 mm · 1 mm · 550 angstrom as measured by a superconducting quantum interference device). Other samples with different sizes and thickness (2 mm · 3 mm · 700 angstrom) also showed saturation magnetization between ∼2 to 3 μB per site at 6 K, 1 T. The value of ∼3 μB per site is large enough for us to determine that the interaction between Fe3+ and Cr3+ is ferromagnetic. The ferrimagnetic order does not explain the observed macroscopic moment. In figure 3 in (1), the magnetization is not saturated because of a low magnetic field of 0.1 T and high temperature of 100 K. This figure should not be used for calculating saturation magnetization. As Meijer points out, the ordinate (emu/g) of that figure was miscalculated; however, this error does not affect the conclusion of our report.


Related Content

Navigate This Article