Magnetic Intermetallic Nitrides

Isn’t it puzzling that most of our (digital) memories depend on the existence of the right materials, in particular magnetic compounds? Remember that a ferromagnetic iron oxide, magnetite, coated on a polymer resulted in the world’s first magnetic tape and was used to eternally capture Mozart’s Symphony no. 39 in E flat major played by the London Philharmonic Orchestra and conducted by Sir Thomas Beecham. The event took place at the concert hall of the chemical giant BASF in Ludwigshafen (Germany) on November 19, 1936, and BASF provided both the organic polymer and the magnetic oxide [1]. Since then, magnetic data storage has become increasingly important. Not only is maghemite the most crucial magnetic pigment [2], our “digital life” would simply be non-existent without the ability to store and retrieve essential data. We may consider ourselves lucky that the capacity of hard discs still doubles from year to year [3], the performance increase going back to superior physical technologies such as giant magnetoresistance [4]. Materials chemistry, however, is at the forefront when it comes to providing the world with even better magnetic compounds, for example magnetic nitrides.

What do chemists mean with the word nitride, and what is so fascinating about them? Nitrides are not just “compounds with nitrogen”, nitrides contain the anion N^3– with an oxidation state of –3. Thus, the latter ion adopts a noble-gas configuration and is isoelectronic with neon, just like the oxide anion, O^2–, with a –2 oxidation state. Note that the number of well-characterized nitrides is orders of magnitude smaller than those of the oxides, despite nitrogen being almost as electronegative (Pauling scale: 3.0) as oxygen (3.5), so we would expect a plethora of nitrides, just like there is a multitude of oxides. The reason for the paucity of the nitrides, however, is given by the enormously strong triple bond of the N₂ molecule: almost 1000 kJ/mol are needed for disruption [5]. Thus, there is no nitrogen corrosion although the atmosphere contains almost 80% nitrogen. Whenever N₂ gets “activated” (that is, split into nitrogen atoms), however, it gets almost as reactive as O. High temperatures are typically needed to result in new nitrides, or one better starts with nitrogen-containing, not-too-stable molecules in the first place, for example the ammonia molecule, NH₃. Making nitrides is both tricky and truly rewarding [6].

For a long time, chemists have classified the nitrides into three groups. There are, first, ionic nitrides such as Li₃N = (Li⁺)₃N^3–, held together by Coulomb forces between cations of the simple (base) metals and the nitride anion. Second, we find the covalent nitrides such as the explosive S₄N₄ [7] with covalent bonds between the two non-metals. Binary compounds such as GaN [8] are often put into this category, too, because the covalent part of the chemical bonding in GaN (the most important optoelectronic material, by the way) is much stronger than the ionic part, despite the oversimplifying notion Ga³⁺N^3–. Finally, there is the fascinating group of metallic nitrides in which nitride ions occupy empty crystallographic sites of the metal lattices. Because there are many more metals than non-metals and because the compositions of the metallic nitrides may adopt various stoichiometric ratios, the metallic nitrides are numerous. For the magnetically archetypal element iron, for example, the socalled iron-rich nitrides alone [9] are Fe₂N, Fe₃N, Fe₄N and Fe₁₆N₂ (strangely not called Fe₈N, for historical reasons).

Among these, the phase correctly designated as γ-Fe₄N is of paramount importance and has been studied experimentally [10] and theoretically [11]. Not only is γ-Fe₄N highly inert in terms of chemical reactivity, it is also mechanically hard. In addition, the unusually large ratio of bulk and shear modulus indicates that it is a ductile material [12], so its properties would make it an almost ideal candidate for high-performance magnetic recording heads. Its cubic unit cell is depicted in Figure 1, with a central nitride ion on Wyckoff position 1b. This N^3– is octahedrally coordinated by six nearest Fe atoms, sitting on the cube’s face centers on position 3c. In addition, eight more Fe atoms (which we designate as Fe’) are found on the corners, position 1a. Any crystallographer immediately recognizes the similarity with the CaTiO₃ perovskite structure. Clearly, N plays the role of the Ti atom, Fe corresponds to O, and Fe’ has replaced the Ca atom. In order to indicate the similarity with the perovskite type, one might write Fe₄N as Fe’NFe₃ which also clarifies that it is an anti-perovskite because the roles of cation and anion are exchanged.

Figure 1. Unit cell of the cubic crystal structure of γ-Fe₄N with the center N atom (position 1b) in green and the Fe atoms in red (face centers, 3c) and in orange (corners, 1a). The lattice parameter is a = 3.79 Å.

Coming back to the magnetic properties of Fe₄N, the very large saturation magnetization of 208 emu/g (almost as large as for bcc-Fe) and also the low coercive field of 5.8 Oe have attracted the materials scientists [13]. The crystal structure itself suggests that the physical properties of Fe’NFe₃ may be fine-tuned by substituting the Fe’ corner atom by another M atom to come up with yet another magnetic phase. Whether a metal atom M may substitute the Fe’ atom depends on the atomic size because the corner position is significantly larger than the face-center position. In addition, the nitrogen affinities of both metals need to be taken into account, nowadays by applying state-of-the-art electronic-structure theory. Thus, instead of running hundreds of time-consuming and expensive chemical reactions, density-functional GGA total-energy calculations using pseudopotentials allow us to conveniently scan the chemical phase space. For example, if we were to replace the Fe’ atom with a precious Rh (rhodium) atom, theory predicts that RhFe₃N should be enthalpically stable and also ferromagnetic, with a slightly exothermic formation enthalpy [14], if one starts with the lowest-energy educts FeRh, Fe and FeN. This is shown in Figure 2.

Figure 2. DFT enthalpy-pressure diagram for the formation of ferromagnetic RhFe₃N (red curve). The enthalpy of the starting materials FeRh/Fe/FeN defines the (black) zero line while the enthalpy of Rh/2 Fe/FeN is given as the green curve. Ferromagnetic RhFe₃N is the most stable phase over the entire pressure range.

In addition, theory clearly favors Rh to go on the corner, not the face-center position. Nonetheless, synthesizing RhFe₃N is not that easy – which doesn’t surprise the nitride chemist. The most straightforward way is a “coupled reduction” route [15] which requires the reduction of a metal oxide and the simultaneous formation of an intermetallic alloy. Then, RhFe₃N is always accompanied by a few percent FeRh alloy but its presence may be minimized by an improved two-step ammonolysis combining a high-temperature sintering step and a low-temperature nitriding reaction [16]. The best RhFe₃N samples have been characterized using X-ray Rietveld refinements and result in a larger a = 3.83 Å due to the large Rh atom. SEM/EDX techniques confirm the absence of other impurities. Not only is the Rh atom found exclusively on the corner position, a few Fe atoms also go on this lattice site, surprisingly enough. Thus, both X-ray diffraction as well as ⁵⁷Fe Mößbauer experiments indicate that the correct composition must be written as Rh_0.8Fe_3.2N [16]. Magnetic susceptibility measurement evidence a typical ferromagnetic behavior with a Curie temperature of about 505 K. The hysteretic loop close to zero Kelvin suggests a coercive field of one fifth of that of γ’-Fe₄N. Together with a remanence of about 0.52 T, Rh_0.8Fe_3.2N is correctly classified as a semi-hard ferromagnetic material, with a low-temperature saturation magnetization of 8.3 μ_B [15].

At this point, the opportunities for new materials chemistry, both experimentally and theoretically, seem to be endless. There is a plethora of theoretical synthetic targets, all differing in the M atom used in MFe₃N formulas, and they all differ in their availability, individual stability, magnetic properties, price and so forth. One might, for example, take the relatively large and cheap main-group element gallium (Ga), and this results in the antiferromagnetic phase Ga_0.9Fe_3.1N [17] with a = 3.80 Å. Its SEM picture is given in Figure 3. Ga_0.9Fe_3.1N, however, Ga is just one compound out of an entire series dubbed Ga_xFe_4–xN with 0 < x < 0.9 [17], and chemists would call these compounds berthollide (that is, stoichiometrically variable) but not daltonide. Upon decreasing the Ga amount, a continuous change from antiferromagnetic to ferromagnetic behavior results, as depicted in the hysteretic loops in Figure 4, such that the magnetic properties may be synthetically adjusted. In the entire series the course of the lattice parameters does not follow Vegard’s rule but shows a minimum for Ga_0.5Fe_3.5N. Once again, quantum mechanics helps to clarify this observation: different magnetic orderings are responsible for the difference in molar volumes. Also, finite-temperature calculations including the phonon contributions make us understand why high temperatures are needed to suppress the formation of nonmagnetic GaN: only at high temperatures the Gibbs formation energy of the ferromagnetic compound turns negative [18].

Figure 3. Scanning electron microscopy photograph of a freshly prepared Ga_0.9Fe_3.1N sample.

Figure 4. Hysteretic loops of various compounds of the series Ga_xFe_4–xN showing that the magnetic saturation moments increase with a lowering gallium concentration.

Upon moving down main-group III, the In_xFe_4–xN series can also be synthesized but its lattice parameters do follow Vegard’s rule. The higher homologue indium does not inhibit the formation of a ferromagnetic state because the indium-richest phase In_0.8Fe_3.2N is also a ferromagnet, with a Curie temperature of about 662 K [19], possibly due to the widened lattice: as a rule of thumb, spin polarization is often enhanced by “stretching” magnetic atoms. In fact, there is an intimate connection between the structural properties and the electronic structure, as exemplified for PdFe₃N. When its thermal expansion (a very wide a = 3.86 Å) is theoretically modeled by first-principles phonon dynamics, it becomes clear that the partially covalent Fe–N interaction suppresses the lattice vibrations of the PdFe₃ matrix, a most astonishing finding [20]. Moreover, the experimentally determined lattice expansion shows a kink close to PdFe₃N’s ordering temperature. Such phenomena define tomorrow’s research: not only do some nitrides possess exciting magnetic and mechanical (hardness) properties, there are clear indications that magnetism influences structure, and structure influences magnetism. It would be highly rewarding to synthesize a metallic nitride whose magnetic and mechanical properties may be switched between different states by either subjecting the material to a magnetic field or, alternatively, to a mechanical stress.

References

1. http://www.magnetbandmuseum.info/
2. R. Dronskowski, Adv. Funct. Mater. 2001, 11, 27.
3. M. Albrecht, J.-U. Thiele, A. Moser, Physik Journal 2003, 10, 25.
4. G. Binasch, P. Grünberg, F. Saurenbach, W. Zinn, Phys. Rev. B 1989, 39, 4828.
5. A. F. Holleman, N. Wiberg, Inorganic Chemistry, Academic Press, San Diego 2001.
6. R. Niewa, F. J. DiSalvo, Chem. Mater. 1998, 10, 2733.
7. S. H. Irsen, P. Jacobs, R. Dronskowski, Z. Anorg. Allg. Chem. 2001, 627, 321.
8. W. Klemm, W. Tilk, H. Jacobi, Z. Anorg. Allg. Chem. 1932, 207, 187.
9. B. Eck, R. Drons­kows­ki, M. Takahashi, S. Kikkawa, J. Mater. Chem. 1999, 9, 1527.
10. H. Jacobs, D. Rechenbach, U. Zachwieja, J. Alloys Compd. 1995, 227, 10.
11. C. A. Kuhnen, R. S. De Figueiredo, V. Drago, E. Z. da Silva, J. Magn. Magn. Mater. 1992, 111, 95.
12. Z. Wu, J. Meng, Appl. Phys. Lett. 2007, 90, 241901.
13. S. K. Chen, S. Jin, T. H. Tiefel, Y. F. Hsieh, E. M. Gyorgy, D. W. Johnson, Jr., J. Appl. Phys. 1991, 70, 6247.
14. J. von Appen, R. Dronskowski, Angew. Chem. Int. Ed. 2005, 44, 1205.
15. A. Houben, P. Müller, J. von Appen, H. Lueken, R. Niewa, R. Dronskowski, Angew. Chem. Int. Ed. 2005, 44, 7212.
16. A. Houben, V. Šepelák, K.-D. Becker, R. Dronskowski, Chem. Mater. 2009, 21, 784.
17. A. Houben, J. Burghaus, R. Dronskowski, Chem. Mater. 2009, 21, 4332.
18. J. Burghaus, M. Wessel, A. Houben, R. Dronskowski, Inorg. Chem. 2010, 49, 10148.
19. J. Burghaus, J. Kleemann, R. Dronskowski, Z. Anorg. Allg. Chem. 2011, 637, 935.
20. D. Music, J. Burghaus, T. Takahashi, R. Dronskowski, J. M. Schneider, Eur. Phys. J. B 2010, 77, 401.