Growth and Characterisation of Antiferromagnetic Ni₂MnAl Heusler Alloy Films

Abstract

Recent rapid advancement in antiferromagnetic spintronics paves a new path for efficient computing with THz operation. To date, major studies have been performed with conventional metallic, e.g., Ir-Mn and Pt-Mn, and semiconducting, e.g., CuMnAs, antiferromagnets, which may suffer from their elemental criticality and high resistivity. In order to resolve these obstacles, new antiferromagnetic films are under intense development for device operation above room temperature. Here, we report the structural and magnetic properties of an antiferromagnetic Ni₂MnAl Heusler alloy with and without Fe and Co doping in thin film form, which has significant potential for device applications.

1. Introduction

Antiferromagnetic spintronics has been attracting a lot of attention due to its potential for THz operation and low power consumption [1]. Currently, antiferromagnetic metals, e.g., IrMn₃ and PtMn₃, and semiconductors, e.g., CuMnAs, have been commonly used [2]. However, these materials are fragile against local compositional changes and may not be suitable for miniaturisation. For further investigation towards device applications, it is critical to develop new antiferromagnetic materials, which are robust against nanofabrication, such as oxidation, atomic mixing, edge roughness and post-annealing.

Antiferromagnetic Heusler alloys are good candidates for applications with precise controllability by potential atomic substitutions. In particular, ternary Heusler alloys, e.g., Pt₂MnGa [3], Ni₂MnAl [4,5] and Mn₂VSi [6], have been reported to exhibit antiferromagnetism as schematically shown in Figure 1. Among them, Ni₂MnAl has been studied and has a lattice constant of 0.5812 nm [5], similar to commonly used seed layers and substrates. However, the Néel temperature has been reported to be between 40 °C [7] and 80 °C [4], which will need to be increased for device applications via atomic substitution as previously reported, e.g., for compensated Mn_2.4Pt_0.6Ga ferrimagnet [8]. These studies have been based on epitaxial Ni₂MnAl films [9,10] and bulk samples, but not on polycrystalline films with their suitability for devices, e.g., magnetic sensors and recording. Recent reduction in a device size can utilise a single grain in a polycrystalline film to avoid any electrical and/or magnetic scattering at a grain boundary and/or magnetic domain wall.

In this paper, we report on the optimised growth and annealing conditions for polycrystalline Ni₂MnAl films sputtered at room temperature (RT). The optimisation involves (i) post-annealing between ~400 and 500 °C for up to 2 h and (ii) atomic substitution for crystallisation. Point (ii) is highly significant for the tuning of structural and magnetic properties, such as the Néel temperature, blocking temperature and the lattice constant.

2. Experimental Procedures

100 nm thick Heusler alloy films capped with 2~5 nm of Ru or Ta were deposited onto Si substrates using RF generated remote plasma sputtering. The layers were deposited using a PlasmaQuest high target utilisation sputtering system (HiTUS) with a base pressure of 3 × 10⁻⁵ Pa. The plasma was generated by an RF field of 13.56 MHz in an Ar atmosphere of 3 × 10⁻¹ Pa and steered onto the target with a DC bias (V_T) ranging from −250 to −990 V. V_T controlled the deposition rate and the resulting atomic mixing. The typical deposition rate from the stoichiometric Ni₂MnAl target using a bias voltage of 990 V was 0.06 nm/s. The growth rate also controls the grain size. Ni₂MnAl (25)/CoFe (10) and Co₂FeSi (10) (thickness in nm) bilayers were also grown using HiTUS in a similar manner.

The composition of the films was analysed using inductively coupled plasma optical emission spectroscopy (ICP-OES) by InterTek Ltd., confirming near stoichiometry (with some Al deficiency), Ni:Mn:Al = 54.1:28.9:17.0 and 56.1:29.6:14.3 for V_T = −900 and −300 V, respectively. The crystalline structures were characterised using out-of-plane (OP) and grazing incidence in-plane (IP) X-ray diffraction (XRD, Rigaku SmartLab) with a Cu Kα source and a Ge(220) 2-bounce monochromator. The Heusler-alloy films were annealed in an Ar gas flow (2 L/min.) between 235 and 700 °C for up to 9 h. After each annealing step magnetisation curves were measured using an alternating gradient force magnetometer (AGFM, Princeton Measurements Model 2900) at RT.

3. Results and Discussion

3.1. Ni₂MnAl Films

From XRD measurements, a Ni₂MnAl(220) principal peak was found once samples were annealed at 700 °C as shown in Figure 2. A well-defined (220) peak was developed OP within the first 90 min of annealing. However, after a further one hour of annealing, no further OP crystalline growth was observed as shown in Figure 2b. As similar to Fe_2+xV_yAl [11], longer annealing induces phase segregation in the Ni₂MnAl films as confirmed by the appearance of a ferromagnetic phase. The OP lattice constant of Ni₂MnAl was measured to be (0.575 ± 0.001) nm. This is about 1% lower than the theoretically predicted value of 0.5812 nm [5]. There was a transition in the IP (220) peak when the sample was annealed for more than 1.5 h, corresponding to a 1% change in the IP lattice constant from (0.577 ± 0.001) nm to (0.571 ± 0.001) nm, which almost agree with the OP results. The lattice constants of the sputtered polycrystalline Ni₂MnAl films are almost 2% lower than the theoretical value, which may be due to the lower crystallinity and the formation of the segregated phases in the films. This is supported by the absence of additional superlattice (200) or (111) peaks in the XRD patterns. Magnetic measurements confirmed that these films show no magnetic response, indicating that they are in an either antiferromagnetic, paramagnetic or compensated ferrimagnetic state.

In order to quantify the crystallinity of the Ni₂MnAl films measured by XRD, we used a Q factor defined as

$$\mathrm{Q}\mathrm{factor}=\frac{\mathrm{Intensity}\left(\mathrm{cps}/°\right)}{\mathrm{Full}\mathrm{width}\mathrm{half}\mathrm{maximum}\left(\mathrm{FWHM}\right)}$$

(1)

The (220) peaks were used to quantify the A2 crystalline ordering (atomic mixing between Ni, Mn and Al), i.e., the increase in Q factor represents the increase in A2 ordering, which typically leads to B2 ordering (atomic mixing between Ni and Mn) with a coexisting (200) peak and L2₁ perfect ordering with coexisting (200) and (111) peaks [12].

Given the requirements for device fabrication, annealing temperatures above 500 °C are impractical. Therefore, our optimisation criteria were as follows: (i) we identified the post-annealing condition below 500 °C to maximise the Q factor. The increase in the A2 ordering leads to the formation of the pseudo-B2 phase, which is theoretically [5] and experimentally [10] predicted to exhibit antiferromagnetism; (ii) we have substituted some constituent elements to investigate the effect on the crystallisation temperature.

The Ni₂MnAl samples were post-annealed for up to 6 h at temperatures in the range 250 to 700 °C. There was no L2₁ crystallisation observed in any of the samples post-annealed at a temperature below 700 °C (see Figure 3). The presence of a small number of A2 ordered nanocrystals was indicated in all the as-deposited samples by a weak (220) Ni₂MnAl alloy peak at 44.0° with a FWHM of 0.7°. Rocking curves were used to measure the texture of the (220) peak, indicating that the as-deposited grains were aligned within ±15° from the sample surface only for the epitaxial films as reported previously [10]. In a similar Heusler alloy of Fe_2+xV_yAl films [11], a B2 phase has been formed when the Q-factor increases over one order of magnitude as compared with the initial A2 crystallisation occurred. In this study we applied this criterion to identify the phase transformation in Figure 3.

The corresponding exchange bias are shown in Figure 4 for the Ni₂MnAl (25)/CoFe (10) and Co₂FeSi (10) (thickness in nm) bilayers post-annealed at 250 and 400 °C for 2 h, which were in the A2 and B32a phase, respectively, as shown in Figure 3. These magnetisation curves were measured after field cooling to 100 K under an in-plane magnetic field of 20 kOe. No clear exchange bias effect can be seen within the measured temperature range in Figure 4. However, an increase in the coercivity was observed only for the sample annealed at 400 °C forming the B32a phase. This may be indicative of coupling between layers, suggesting the Ni₂MnAl layer is antiferromagnetic but the corresponding Néel (and blocking) temperature is below 100 K possibly due to the coexisting disordered phases. Further increase in the annealing temperature leads to the higher crystallinity of Ni₂MnAl as shown in Figure 3 but it induces interfacial mixing in the bilayers causing reduction in the saturation magnetisation. By maintaining the sharp interface between Ni₂MnAl and the neighbouring ferromagnetic layer, an exchange bias can be measured at 100 K or above [10].

3.2. Ni₂MnAl Films with Atomic Substitutions by Al

As the sputtered Ni₂MnAl films are ~10% Al deficient, one 6 mm diameter Al peg was added to a pre-drilled hole in the sputtering target, achieving Ni:Mn:Al = 49:27:24. With a target utilisation of over 90%, this technique allows for the variation of the film composition in the HiTUS. Figure 5 shows the Q factors for the Al-doped Ni₂MnAl post-annealed at 500 °C, which was found to be the maximum annealing temperature to avoid Al segregation. Similar to the films without Al doping, Ni₂MnAl crystallises at a relatively high temperature. The Al-doped Ni₂MnAl films have therefore the optimum A2 crystal structure after 2~4 h of annealing at 500 °C. The corresponding lattice constants were found to decrease from 0.585 to 0.575 nm after 1 and 4 h of annealing, respectively. This confirms that the lattice constants are similar to those without Al doping except for the sample annealed for 1 h. This agrees with the increase in the Q factors shown in Figure 5, indicating the effect of the Al doping on the optimised crystallisation temperature and crystallinity of Ni₂MnAl is negligible.

3.3. Ni₂MnAl Films with Atomic Substitutions by Fe

In order to decrease the Ni₂MnAl crystallisation temperature, an attempt to substitute Ni in the lattice by Fe was made because its covalent radius is greater than that of Ni. This results in the lattice being strained and this may decrease the manganese–manganese nearest neighbour distance to induce antiferromagnetic coupling. However, no crystallisation was found. Fe_2+xV_yAl has been studied previously [11], showing that the ordering temperature is lower than that of the Ni₂MnAl thin films. Therefore, it is predicted that doping with Fe can lower the ordering temperature and induce antiferromagnetic ordering in the Ni₂MnAl thin films.

The Ni₂MnAl sputtering targets were doped with two or four Fe pegs with a diameter of 2 mm, effectively making a Ni_2−xFe_xMnAl target. The sputtered Ni₂MnAl films doped with Fe were found to exhibit ferromagnetic behaviour even in the as-deposited state. Almost negligible magnetic moments of (1.5 ± 0.1) µemu, approximately 3 × 10⁻² emu/cm³, was measured for the samples grown from the Ni₂MnAl sputtering target doped with two Fe pegs and (2.7 ± 0.1) µemu with four Fe pegs. This small moment was probably induced by Fe segregation. The films were then post-annealed for up to three hours at temperatures between 400 and 500 °C. However, no crystallisation of the Ni₂MnAl Heusler alloy was detected but the average saturation magnetisation increased to (2.0 ± 0.1) µemu and (3.4 ± 0.1) µemu after annealing at 500 °C for 3 h for the films doped with two and four Fe pegs, respectively. This suggests that the Fe segregation is promoted by post-annealing.

3.4. Ni₂MnAl Films with Atomic Substitutions by Co

Co-doping was also tested because its covalent radius is greater than that of Ni but is smaller than that of Fe. This may also decrease the manganese–manganese nearest neighbour distance to induce antiferromagnetic coupling as predicted for the case of the Fe doping. The corresponding lattice constants for one Co peg were found to decrease from 0.584 to 0.580 nm after 1 and 5 h of annealing, respectively. Co₂MnAl has hence been reported to have an ordering temperature lower than Ni₂MnAl [13,14], suggesting that the Co doping may also lower the ordering temperature and promote antiferromagnetic ordering.

The Ni_2−xCo_xMnAl sputtering targets were used adding one or two Co pegs with a diameter of 6 mm onto the Ni₂MnAl target, achieving Ni:Co:Mn:Al = 56:9:20:15 and 51:15:20:14 for one and two Co pegs, respectively. The sputtered films exhibited paramagnetic behaviour in the as-deposited state. The films successfully crystallised after post-annealing at temperatures between 400 and 500 °C for two hours as shown in Figure 6. Both samples exhibited the A2 ordering as identified by the (220) peak at 43°. No clear (200) peak was visible at 37.5° which again suggests the absence of B2 ordering.

As shown in Figure 7, the crystallisation also occurs at 300 °C, which is 150 °C lower than that for films without Co doping, which is favourable for device applications. The corresponding lattice constants were measured to be 0.8~2.9% larger than those for films without Co doping.

The crystallinity was also found to improve with increasing Co doping as shown in Figure 7. The magnetic response of the samples changed to ferromagnetic after post-annealing at 500 °C as shown in the inset of Figure 6, showing a magnetic moment of (76.2 ± 0.1) µemu, approximately 1.5 emu/cm³, which is larger than that of the Fe substituted films. A possible cause for the ferromagnetism is the formation of Co₂MnAl grains because the films were annealed at the ordering temperature [15]. No further crystalline ordering was observed with increasing annealing and the corresponding magnetic properties were unchanged. This observation supports the hypothesis of the formation of Co₂MnAl as the cobalt content can be segregated from the Ni₂MnAl matrix, thus preventing further significant formation of grains. To eliminate such segregated grains, uniform compositional distributions within a film may be required, which is difficult to achieve using the doping method of the sputtering target used in this study. Even so, our study suggests a significant potential of Ni_2−xCo_xMnAl for robust antiferromagnetism at room temperature.

4. Summary

We have grown and characterised a series of Ni₂MnAl Heusler alloy films. The films with one Co peg post-annealed at 500 °C for 2 h were found to show the highest crystalline ordering within the A2 phase but exhibited weak magnetic moments. For the demonstration of room-temperature antiferromagnetism, Fe and Co have been used to partially substitute for Ni. The Fe substitution showed an increase in the magnetic moment with increasing Fe content, which may be due to Fe segregation. On the other hand, Co substitution can effectively reduce the crystallisation temperature down to 300 °C but the corresponding magnetisation measurements proved that the B2-ordered (Ni,Co)₂MnAl films showed ferromagnetic Co₂MnAl segregation as well. The segregated phases need to be eliminated by further compositional optimisation to achieve the stoichiometry while maintaining the reduced crystallisation in the pseudo-B2 phase temperature for antiferromagnetic spintronics.