Fabrication and physical properties of [Fe/Fe₄N]_N multilayers with high saturation magnetization

Magnetic materials with giant saturation magnetization have attracted many of researchers and condensed matter physicists for decades. As described by the famous Slater-Pauling curve, the material with the highest saturation magnetization Ms is Fe₆₅Co₃₅ alloy (Ms ∼1950 emu/cc). However, this was challenged in 1972 by an announcement of the giant magnetism of α”-Fe₁₆N₂ with its Ms about 18% higher than that of Fe₆₅Co₃₅, in the compound of polycrystalline films.¹ α”-Fe₁₆N₂ which was first reported by Jack in 1951, is an ordered nitrogen martensite obtained by quenching the austenite.² In 1989, epitaxially grown Fe-N films demonstrated that the saturation magnetization of Fe₁₆N₂ is in the range of 2230-2290 emu/cc.³ After that, three kinds of Fe-N materials were investigated: foil, powder and films. However, many controversial results on the saturation magnetization for the α”-Fe₁₆N₂ phase have been reported, which range from 1750 emu/cc to 2290 emu/cc.^4–9 The theoretical study does not support the giant magnetic moment of iron atoms.¹⁰ Therefore, this material has been considered as one of the most mysterious ones in the magnetic community.

In 2010, a new model of partial localized electron states was proposed to explain the giant saturation magnetization of α”-Fe₁₆N₂, whose Ms is 2130 emu/cc.^11,12 The group of Japanese also synthesized the single-phaseα”-Fe₁₆N₂ nanoparticles, which exhibit a saturation magnetization (Ms) of 234 emu/g at 5K and a magnetocrystalline anisotropy constant (Ku) of 9.6*10⁻⁶ erg/cc.¹³ The results again attracted the researcher’s attention and showed a new path for a possible candidate of the rare-earth-free permanent magnet material with a high Ms, which is very important for a green energy society. Besides, α-Fe(N) with interstitial N atoms in the Fe lattice, has high saturation magnetization similar to that of FeCo.¹⁴ Moreover, Fe-N is a potential material in spintronics.¹⁵ 

The lattice of α”-Fe₁₆N₂ is a body centered tetragonal structure where N atoms alternatively occupy the octahedral sites, and its lattice can be thought of as consisting of the alternatively arranged γ-Fe and γ’-Fe₄N cells.² Though γ-Fe is not stable at room temperature, α-Fe, the stable body-centered-cubic, can be regarded as a face-centered-cubic with the compressed c axis. In this paper, [Fe/Fe₄N]_N multilayers were designed to approach the α”-Fe₁₆N₂ structure, and their magnetic properties were studied. After annealing treatments, high saturation magnetic induction of 1850 emu/cc was obtained in the Fe-N multilayers.

The Fe-N thin films were grown on MgO (200) single-crystal substrates. Besides Ar, N₂ with different flow ratios to Ar was introduced in the chamber and then Fe-N films were obtained by the DC reactive magnetron sputtering. Pure iron (99.99%) was used as the sputtering target. The base pressure in the deposition chamber was lower than 4*10⁻⁸Torr, and the substrate temperature was 350 ^oC. The recipe of [Fe/Fe₄N]_N multilayers is described as follows. An α-Fe layer was deposited first on the MgO substrate. After the shutter closed, N₂ was introduced in the chamber for 30 seconds and then γ’-Fe₄N was prepared by the reactive sputtering with the N₂ flow ratio of 20%, which was identified as the ratio of N₂ flow to Ar flow. After the deposition of Fe₄N, the shutter closed again and the N₂ inlet was turned off for 70 seconds before Fe deposition. Since the lattice constant of Fe₄N is 0.385 nm, the thickness of the individual Fe₄N layers were set at 0.76 nm, 1.52 nm, 2.28 nm and 3.04 nm, respectively. The Fe layers were of the same thicknesses as the corresponding Fe₄N layers. The recipe was repeated different times for different Fe and Fe₄N thickness, and then the thickness of the multilayers was kept at 30 nm. Finally, a 3-nm-thick Pt layer was capped on the multilayers. For directly comparing the saturation magnetization, all the samples were prepared using the same recipe. That is, the shutter was opened and closed even in the preparation of Fe and Fe₄N films. Finally, samples were annealed at different temperatures, from 350 ^oC to 500 ^oC in a 4*10⁻⁸Torr vacuum for 60 min, in order to adjust the position of the nitrogen atoms.

The structure of the samples was examined by x-ray diffraction (XRD) with Cu Kα radiation. The film thickness was determined by the sputtering rate, which is calibrated by using a surface profiler. The magnetic properties were measured by a vibrating sample magnetometer (VSM) with a resolution of 5*10⁻⁶ emu at room temperature. The M–H loops were measured with the applied field parallel to the film plane.

Figure 1 shows X-ray θ-2θ scans for 30 nm Fe-N films at different N₂ flow ratios of 5 %, 10 %, 15 % and 20 % at 350 ^oC. Phase transformation from α-Fe to γ’-Fe₄N is clearly seen. For the Fe-N film prepared at the N₂ flow ratio of 5 %, there is only a strong Fe (200) peak at 65^o, but no Fe (110) peak at 44.7^o which is its preferred orientation. This indicates the epitaxial growth of the Fe film on MgO, since the diagonal of the Fe lattice is 0.405 nm which is similar to the MgO lattice of 0.42 nm. With the increase of the N₂ flow ratio, the Fe peak becomes weak, but the γ’-Fe₄N (200) peak appears and turns strong. When the N₂ flow ratio reaches 20 %, only Fe₄N peak is observed. The absence of (111) peak implies the epitaxial growth of γ’-Fe₄N.

Figure 2 shows the angular dependent magnetic properties of the γ’-Fe₄N film which was prepared at the N₂ flow ratio of 20 %. The switching field increases with the angle between the applied field and the in-plane direction, following 1/cosθ law of the magnetization reversal of the domain wall motion. Single crystal films were obtained. The saturation magnetization of Fe₄N is 890 emu/cc, and that of the Fe film is 1400 emu/cc. The Ms of iron film was slightly smaller than other groups’ results.^16,17 This might be due to the inaccurate control of film thickness by the shutter opening and closing. The dependence of saturation magnetization on the N₂ flow ratio is consistent with the transformation of the Fe phase to the Fe₄N phase as shown in the Figure 1.¹⁸ 

By using α-Fe and γ’-Fe₄N layers, [Fe/Fe₄N]_N multilayers were fabricated. The total thickness of the [Fe/Fe₄N]_N multilayers was 30 nm, and the thickness of the Fe and Fe₄N layers were the same. Figure 3 shows X-ray θ-2θ scans for 30 nm Fe-N multilayers with the different periods N. Fe (200) peak appears in all the samples and its intensity becomes stronger with the increase of the period thickness. As shown in Fig. 3(d), there is no Fe₄N peak in [Fe/Fe₄N]₂₀, where the Fe₄N layer (0.76 nm) is very thin. Considering that the lattice constant of Fe₄N is 0.385 nm, the layer is only two-lattice high. Hence the lattice structure is hard to form and/or to be detected. However, the Fe₄N (200) peak appears when the thickness increases to 1.52 nm (Fig. 3(c)), about four-lattice height. With the further increase of the thickness, the Fe₄N (200) peak becomes stronger and stronger (Fig. 3(b) and 3(a)). Epitaxial growth between Fe₄N and Fe was demonstrated.

Fig. 4 shows the influence of the periodic number N of the multilayers on the saturation magnetization. Ms of all the samples was around 1100 emu/cc. Except that of the [Fe/Fe₄N]₁₀, however, the saturation magnetization of the other three samples were below the ideal value of 1150 emu/cc, the average one of Fe and Fe₄N. The results imply the reactive sputtering process was not well controlled and then the partial Fe layer was nitridized. Moreover, interface roughness might be another reason, leading to N diffusion into the Fe layer.

In order to adjust the position of the nitrogen atoms, the annealing treatments of different temperatures, from 350 ^oC to 500 ^oC, were carried on to the [Fe/Fe₄N]_N multilayers for 60 min. Figure 5 shows the XRD θ-2θ scans for the [Fe/Fe₄N]₅ after post-annealing at different temperatures. There is no change from the as-deposited multilayers in the scan after 400 ^oC annealing. Tiny migration of Fe (200) can be observed above the annealing temperature of 450 ^oC. Moreover, a slight doming is seen. The doming developed to a peak after 500 ^oC annealing. The peak, which is located at 63.3^o, should not be iron oxide, since there is a 3-nm-thick Pt cover layer. It is reported that Fe₄N cannot form at the substrate temperature of 400 ^oC though excellent epitaxial growth of Fe₄N can be realized at the substrate temperature of 350 ^oC.¹⁹ That is, it is easy for the N atom to escape from the Fe₄N lattice at a temperature higher than 400 ^oC.²⁰ Therefore, the appearance of doming at 450 ^oC and the Fe₄N peak at 500 ^oC should be due to the N atom diffusion from the Fe₄N to the Fe layer. The diffusion is enhanced with the temperature increase. Moreover, N atoms tend to locate close to Fe₄N side. This leads to an obvious deviation from the Fe (200) peak. The interstitial N atom leads to the expansion of the Fe lattice. According to the lattice expansion, the N content in the Fe lattice is about 2.4 at%.² However, no peak of α”-Fe₁₆N₂ is found. This might be due to the small content of the phase. Since there is a transition layer between the Fe and the Fe₄N layers, a gradual change of the lattice is expected. Fe₁₆N₂ could form in the transition region.

Fig. 6 shows the magnetic hysteresis loops of the Fe film and the post-annealed [Fe/Fe₄N]₅ multilayer. The saturation magnetization of annealed [Fe/Fe₄N]₅ is 1850 emu/cc. As a comparison, the saturation magnetization of pure Fe is 1400 emu/cc (that of Fe₄N is 890 emu/cc as showed in Fig. 3). Thus, the saturation magnetization of [Fe/Fe₄N]₅ above 500 ^oC annealing, increases by 32 % on the base of the saturation magnetization of iron. The inset shows the hysteresis loops of the as-deposited and post-annealed samples of [Fe/Fe₄N]₅. Above 500 ^oC annealing, saturation magnetization increases. It should be mentioned that 500 Oe is the maximal external magnetic field shown here, but the multilayers required the magnetic field of 15 kOe to be saturated. Therefore, the samples have not been fully magnetized at the applied field of 500 Oe. The Fe(N), when the N atoms are in the Fe lattice as the interstitial atom, has larger saturation magnetization than that of pure Fe.^7,14 In the current experiment, the diffusion of the N atom into the Fe lattice from the Fe₄N lattice could generate a similar effect, increasing the magnetization. Therefore, the result reveals a new approach to preparing high saturation magnetization Fe-N materials, by driving the nitrogen atom from the Fe₄N lattice to the Fe lattice during high temperature annealing.

[Fe/Fe₄N]_N multilayers with high magnetic moment have been fabricated by DC reactive magnetron sputtering. By introducing N₂ flow into the sputtering chamber, high quality γ’-Fe₄N film can be prepared. Then, [Fe/Fe₄N]_N multilayers are fabricated by alternative epitaxial growth of the Fe and the Fe₄N layers. The saturation magnetization of the as-deposited [Fe/Fe₄N]_N is slightly below the average value of Fe and Fe₄N. However, above 500 ^oC annealing saturation magnetization of [Fe/Fe₄N]_N increases to 1850 emu/cc, about 32 % higher than that of iron. This is due to the N atom which was diffused from the Fe₄N to the Fe layer at 500 ^oC annealing. This is a new method to prepare high saturation magnetization Fe-N materials.

This work is supported by the NSF of China (Grant No.11174056).