betaFeedback

WorksPhysics and Astronomy

Giant tunability of magnetoelasticity in Fe4_4N system: Platform for unveiling correlation between magnetostriction and magnetic damping

Abstract

Flexible spintronics has opened new avenue to promising devices and applications in the field of wearable electronics. Particularly, miniaturized strain sensors exploiting the spintronic function have attracted considerable attention, in which the magnetoelasticity linking magnetism and lattice distortion is a vital property for high-sensitive detection of strain. This paper reports the demonstration that the magnetoelastic properties of Fe4_4N can be significantly varied by partially replacing Fe with Co or Mn. The high quality Fe4_4N film exhibits large negative magnetostriction along the [100] direction (λ100λ_{100}) of -121 ppm while Fe3.2_{3.2}Co0.8_{0.8}N shows λ100λ_{100} of +46 ppm. This wide-range tunability of λ100λ_{100} from -121 to +46 across 0 allows us to thoroughly examine the correlation between the magnetoelasticity and other magnetic properties. The strong correlation between λ100λ_{100} and magnetic damping (αα) is found. The enhanced extrinsic term of αα is attributable to the large two magnon scattering coming from the large magnetostriction. In addition, the density of states at the Fermi level plays a primal role to determine both λ100λ_{100} and the intrinsic term of αα. Thanks to the giant tunability and the bipolarity of magnetoelasticity, magnetic nitrides are candidate materials for high-sensitive spintronic strain sensors.

† Present address: Department of Advanced Materials Science and Engineering, Faculty of Engineering Sciences, Kyushu University, Fukuoka 816-8580, Japan

Flexible electronic devices have rapidly become important in the coming trillion-sensor era, in which wearable devices will spill over. Flexibility is required not only for semiconductor-based electronic devices, but also for spintronic devices, and the vast research field of flexible electronics involves a variety of research topics coming from technological demands and scientific interests, e.g. device fabrication and operation on flexible substrates, and improved performance and new functionalities under the application of mechanical strain1-7. As we look at the current flexible spintronics, miniaturized strain sensors exploiting the spintronic function6 have attracted much attention, where the magnetoelasticity linking magnetism and lattice distortion is a vital property for high-sensitive and multi-directional strain detection. Through the inverse magnetostriction effect, the magnetization vector of a magnetic material changes depending on the magnitude and the direction of mechanical stress applied to the magnet. Thus, large magnetoelastic materials and the elucidation of underlying physical mechanism are keys for the development of high-sensitive devices.

Although there are several magnetoelastic materials, the spintronic applications need to show large magnetoresistance (MR) effect as well as large magnetostriction. Tb-Dy-Fe alloys8 and Fe-Ga alloys9 are well known ferromagnets exhibiting large magnetostriction. However, since no large MR effect has been reported in any giant MR (GMR) or tunneling MR (TMR) devices with Tb-Dy-Fe or Fe-Ga, these materials are excluded from consideration. The recent first-principles calculation10 has predicted that Fe4N\mathrm{Fe}_4\mathrm{N}, a ferromagnetic iron nitride exhibits the magnetostriction constant several times larger than that of typical ferromagnetic metals such as Fe, Co, and Ni, and there is a report on experimental investigation for magnetostriction of Fe4_4N although the reported value was smaller than the theoretical prediction11. In addition, a large TMR effect in a magnetic tunnel junction12, anomalous Hall effect13, and anomalous Nernst effect13,14 have been reported for systems with Fe4_4N. These features make ferromagnetic nitrides a promising material group not only for spintronic devices15,16, but also for flexible spintronic devices. Moreover, epitaxial growth of ferromagnetic nitrides on flexible mica substrates and modulation of magnetic properties by stress application have been reported17, and the Fe4_4N does not contain minor metal elements and rare earth elements, which is another advantage from the viewpoint of large-scale fabrication of wearable devices. Thus, it should be clarified to what extend the magnetoelasticity can be enhanced and controlled for ferromagnetic nitrides in order to achieve best performance of flexible spintronic devices.

Here we report that the magnetoelastic properties of Fe4_4N can be significantly varied by partially replacing Fe with Co or Mn. As shown in Figure 1a, the Co substitution in Fe4_4N corresponds to the electron doping whereas the Mn substitution leads to the electron depletion. This variation in the number of valence electrons gives rise to the shift of the Fermi level (EFE_{\mathrm{F}}) position, and also makes the Fe4_4N a suitable platform to examine the correlation between the magnetoelasticity and other magnetic properties experimentally and theoretically. By measuring various magnetic properties such as magnetostriction, saturation magnetization (MsM_{\mathrm{s}}), magnetic anisotropy constant (K1K_1), and magnetic damping constant (α\alpha), we experimentally and theoretically reveal the strong correlation between magnetostriction and magnetic damping.

Concept of element substitutions and structural analysis for magnetic nitrides

Figure 1. Concept of element substitutions and structural analysis for magnetic nitrides. (a) Crystal structures of magnetic nitrides; Co substitution into Fe4N\mathrm{Fe}_{4}\mathrm{N} corresponds to electron doping and Mn substitution to hole doping. Out-of-plane and in-plane XRD patterns of the (b) Fe3.2Co0.8N\mathrm{Fe}_{3.2}\mathrm{Co}_{0.8}\mathrm{N}, (c) Fe4N\mathrm{Fe}_{4}\mathrm{N}, and (d) Fe3MnN\mathrm{Fe}_{3}\mathrm{Mn}\mathrm{N} films. Red and blue lines correspond to the results with the out-of-plane and in-plane configurations, respectively.

Co or Mn substitution in Fe4N\mathrm{Fe}_4\mathrm{N} films

Fe4N\mathrm{Fe}_4\mathrm{N}, Fe4−xMnxN\mathrm{Fe}_{4-x}\mathrm{Mn}_x\mathrm{N} (x=0.1x = 0.1, 0.40.4, and 1.01.0), and Fe4−yCoyN\mathrm{Fe}_{4-y}\mathrm{Co}_y\mathrm{N} (y=0.2y = 0.2, 0.50.5, 0.80.8, 1.71.7, and 2.22.2) films with a thickness of approximately 23 nm were grown on SrTiO3(STO)(001)\mathrm{SrTiO}_3(\mathrm{STO})(001) substrates. The exact values of thicknesses, which were estimated from the x-ray reflectivity, were given in the Supplementary Information. Figure 1b, Figure 1c, and Figure 1d display the x-ray diffraction (XRD) patterns for the representative samples of the Fe3.2Co0.8N\mathrm{Fe}_{3.2}\mathrm{Co}_{0.8}\mathrm{N}, Fe4N\mathrm{Fe}_4\mathrm{N}, and Fe3MnN\mathrm{Fe}_3\mathrm{MnN} films, respectively. The red (blue) lines denote the out-of-plane (in-plane) XRD patterns. The 001 and 002 peaks of nitrides appear in the out-of-plane XRD patterns while the 100 and 200 peaks are observed in the in-plane XRD patterns. These facts indicate the epitaxial growth of the nitride films on the STO(001) substrates. From the structural characterization using XRD as well as reflection high-energy electron diffraction patterns (not shown here), it was confirmed that all the samples are epitaxially grown and no secondary phase exists.

Figure 2a shows the magnetization curves of the Fe4N\mathrm{Fe}_4\mathrm{N} film measured at room temperature. When an external magnetic field (HH) is applied in the in-plane [100] direction (solid curve), the remanent magnetization (MrM_\mathrm{r}) is equal to MsM_\mathrm{s}. On the other hand, MrM_\mathrm{r} is decreased when HH is applied in the in-plane [110] direction (dashed curve). This difference originates from the cubic symmetry of magnetocrystalline anisotropy in Fe4_4N with the [100] easy magnetization axis18,19. From the magnetization curves, MsM_s and K1K_1 were evaluated for all the samples, where K1K_1 was estimated from the area enclosed between the [100] and [110] magnetization curves.

Magnetization, magnetostriction and magnetization dynamics of magnetic nitrides

Figure 2. Magnetization, magnetostriction and magnetization dynamics of magnetic nitrides. (a) In-plane magnetization curves of the Fe4N\mathrm{Fe}_{4}\mathrm{N} film measured at room temperature, where MM is normalized by MsM_s. The solid (dashed) curve denotes the result with HH applied in the in-plane [100][100] direction (in-plane [110][110] direction). (b) Illustration of experimental setup of an optical cantilever method. (c) ϕ\phi-dependence of ss at a rotational HH of 175 Oe applied to the Fe4N\mathrm{Fe}_{4}\mathrm{N} film. (d) HH dependence of λ100\lambda_{100} for the Fe4N\mathrm{Fe}_{4}\mathrm{N} (red line), Fe3.2Co0.8N\mathrm{Fe}_{3.2}\mathrm{Co}_{0.8}\mathrm{N} (blue line), and Fe3MnN\mathrm{Fe}_{3}\mathrm{MnN} (green line) films. (e) HH dependence of f0f_0 and (f) Δf0\Delta f_0 for the Fe4N\mathrm{Fe}_{4}\mathrm{N} (red line), Fe3.2Co0.8N\mathrm{Fe}_{3.2}\mathrm{Co}_{0.8}\mathrm{N} (blue line), and Fe3MnN\mathrm{Fe}_{3}\mathrm{MnN} (green line) films with the IP configuration. The inset of (e) shows the resonance spectrum together with the fitting result for the Fe4N\mathrm{Fe}_{4}\mathrm{N} film measured at H=1400H = 1400 Oe. (g) HH dependence of f0f_0 and (h) Δf0\Delta f_0 for the Fe4N\mathrm{Fe}_{4}\mathrm{N} (red line), Fe3.2Co0.8N\mathrm{Fe}_{3.2}\mathrm{Co}_{0.8}\mathrm{N} (blue line), and Fe3MnN\mathrm{Fe}_{3}\mathrm{MnN} (green line) films with the OOP configuration. The inset of (g) shows the resonance spectrum together with the fitting result for the Fe4N\mathrm{Fe}_{4}\mathrm{N} film measured at H=17H = 17 kOe.

Magnetoelasticity and magnetization dynamics for nitride films

Figure 2b illustrates the experimental setup for measuring the magnetostriction: optical cantilever method20. A rectangular-shaped sample with the size of 10 mm ×\times 20 mm was fixed on one side and illuminated with a laser beam from the vertical direction of the film surface. A rotational HH of up to 175 Oe was applied in the in-plane direction to induce the magnetostrictive effect, and the degree of deflection of the sample was measured by the position of the reflected laser light (ss). Since ϕ\phi is the angle between the [100] direction of the nitrides and HH, i.e. ϕ=0∘\phi= 0^\circ is the [100] of nitrides, this measurement setup corresponds to the evaluation of magnetostriction along [100] (λ100\lambda_{100}). The ϕ\phi dependence of ss at H=175H = 175 Oe is plotted in Figure 2c.

The obtained sinusoidal curve was numerically fitted and the displacement of ss (Δs=s(ϕ=0∘)−s(ϕ=90∘)\Delta s = s(\phi= 0^\circ) - s(\phi= 90^\circ)) was obtained, allowing us to evaluate λ100\lambda_{100} with high accuracy (see Supplementary equation (1)). Figure 2d shows the HH dependence of λ100\lambda_{100} for the Fe3.2_{3.2}Co0.8_{0.8}N, Fe4_4N, and Fe3_3MnN films. At the saturated state of magnetization by applying H=175H = 175 Oe, a large magnetostriction, λ100=−121\lambda_{100} = -121 ppm, is obtained for the Fe4_4N film, which is three times larger than the previous report (−40-40 ppm11) in absolute value and is close to the theoretical value of −130-130 ppm by the first-principles calculation10. Also, it is noted that λ100=−121\lambda_{100}=-121 ppm is much larger than the values for Fe-Co alloys without high temperature annealing21, which are representative ferromagnets for the conventional spintronics, and is even in the same order of magnitude as that for the representative magnetostrictive material Fe-Ga. In contrast to Fe4_4N, Fe3.2_{3.2}Co0.8_{0.8}N and Fe3_3MnN exhibit small positive and negative λ100\lambda_{100}, respectively.

The magnetization dynamics of the Fe4_4N, Fe3.2_{3.2}Co0.8_{0.8}N, and Fe3_3MnN films is given in Figure 2e-Figure 2h. Figure 2e and Figure 2f plot the HH dependence of the ferromagnetic resonance (FMR) frequency (f0f_0) and resonance linewidth (Δf0\Delta f_0) at the in-plane HH applied along the [100][100] direction, which is called the IP configuration. The inset of Figure 2e displays the FMR spectrum of the Fe4_4N at H=1400H=1400 Oe. The HH dependences of f0f_0 and Δf0\Delta f_0 are fitted by the following equations:

f0(H)=(γ/2π)(H+Ha,ip)(H+4πMeff)andΔf0(H)=αip(γ/2π)(2H+Ha,ip+4πMeff)+(γ/2π)Hinh\begin{aligned} f_0(H) &= \left(\gamma/2\pi\right)\sqrt{(H+H_{\mathrm{a,ip}})(H+4\pi M_{\mathrm{eff}})} \quad\text{and}\\ \Delta f_0(H) &= \alpha_{\mathrm{ip}}\left(\gamma/2\pi\right)(2H+H_{\mathrm{a,ip}}+4\pi M_{\mathrm{eff}})+\left(\gamma/2\pi\right)H_{\mathrm{inh}} \end{aligned}

where γ\gamma is the gyromagnetic ratio, Ha,ipH_{\mathrm{a,ip}} is the in-plane magnetic anisotropy field, MeffM_{\mathrm{eff}} is the effective magnetization, HinhH_{\mathrm{inh}} is the inhomogeneous field, and αip\alpha_{\mathrm{ip}} is the damping parameter estimated in the IP configuration. The fitting of f0f_0 as a function of HH was done with the fixed γ\gamma value of 1.76×1071.76\times10^{7} rad−1^{-1} Oe−1^{-1} with the assumption of g-factor of 2, which enabled to evaluate Ha,ipH_{\mathrm{a,ip}} and MeffM_{\mathrm{eff}} as fitting parameters. Those values were used for the fitting of Δf0\Delta f_0 as a function of HH. Finally, αip\alpha_{\mathrm{ip}} was evaluated from these IP configuration measurements. Similar analyses were carried out for the results at HH applied in the out-of-plane direction (OOP configuration) as shown in Figure 2g and Figure 2h, where the following equations were used: f0(H)=(γ/2π)(H−Hs)f_0(H) = (\gamma/2\pi)(H - H_s) and Δf0(H)=αoop(γ/2π)(2H−2Hs)+(γ/2π)Hinh\Delta f_0(H) = \alpha_{\mathrm{oop}}(\gamma/2\pi)(2H - 2H_s) + (\gamma/2\pi)H_{\mathrm{inh}}, where HsH_s is the saturation field of magnetization by the out-of-plane HH. αoop\alpha_{\mathrm{oop}} corresponds to the intrinsic term of magnetic damping, which is defined as intrinsic α\alpha while αip\alpha_{\mathrm{ip}} involves both intrinsic and extrinsic terms22,23. This paper defines Δα=αip−αoop\Delta\alpha= \alpha_{\mathrm{ip}} - \alpha_{\mathrm{oop}}, and Δα\Delta\alpha expresses the extrinsic contribution of magnetic damping constant. Although one may think the nonlocal spin relaxation process24 due to the effect of spin pumping to adjacent layers as a component of Δα\Delta\alpha, that contribution is negligibly small because of small spin-orbit interaction of Ti capping layer in the present layer stacking.

Composition dependences of magnetic properties

Figures 3a, 3b, and 3c plot the xx and yy dependences of MsM_s, K1K_1, and λ100\lambda_{100}, respectively. MsM_s decreases by the Mn substitution into Fe4N\mathrm{Fe}_4\mathrm{N} because Fe4−xMnxN\mathrm{Fe}_{4-x}\mathrm{Mn}_x\mathrm{N} exhibits ferrimagnetic spin structures25. Although Fe4−yCoyN\mathrm{Fe}_{4-y}\mathrm{Co}_y\mathrm{N} is a ferromagnet26,27, the Co substitution also reduces MsM_s, which results from the Co atom having a smaller spin magnetic moment than Fe. K1K_1 decreases by the Mn substitution for x>0.4x > 0.4. In the case of Co substitution, K1K_1 decreases with increasing yy and its sign changes to negative at y>0.75y > 0.75. The most striking features are found in the composition dependence of λ100\lambda_{100}. First, as mentioned above, the large λ100\lambda_{100} of −121 ppm is achieved for the present high-quality Fe4N\mathrm{Fe}_4\mathrm{N} epitaxial film. Next, the value of λ100\lambda_{100} takes the negative sign for 0≤x≤1.00 \le x \le1.0 and 0≤y≤0.50 \le y \le0.5, and becomes positive at y≥0.8y \ge0.8. More specifically, λ100\lambda_{100} values can be modulated over a wide range from −121 ppm for Fe4N\mathrm{Fe}_{4}\mathrm{N} to +46 ppm for Fe2.3Co1.7N\mathrm{Fe}_{2.3}\mathrm{Co}_{1.7}\mathrm{N} with λ100\lambda_{100} close to 0 near y=0.75y = 0.75. This is clear demonstration that the Co substitution in Fe4N\mathrm{Fe}_{4}\mathrm{N} realizes the large tunability and bipolarity of magnetoelastic property. Another finding is that there is no strong correlation between K1K_{1} and λ100\lambda_{100} (see the Co-rich composition) although both are parameters related to the spin-orbit interaction.

Figures 3d and 3e plot the xx and yy dependences of αip\alpha_{\mathrm{ip}} and αoop\alpha_{\mathrm{oop}}, respectively. The value of αip\alpha_{\mathrm{ip}}, the sum of intrinsic and extrinsic α\alpha, shows the maximum value of 0.04 for the Fe4N\mathrm{Fe}_{4}\mathrm{N} film and rapidly decreases with the Mn or Co substitution. The value of αoop\alpha_{\mathrm{oop}}, corresponding to intrinsic term of α\alpha, exhibits the composition dependence similar to that of αip\alpha_{\mathrm{ip}} except for x=1.0x = 1.0 and y=2.2y = 2.2. From the values of αip\alpha_{\mathrm{ip}} and αoop\alpha_{\mathrm{oop}}, Δα\Delta\alpha is obtained as a function of composition as shown in Fig. 3f, in which the composition dependence of the ∣λ100∣\lvert\lambda_{100}\rvert value is also plotted. One can see that the composition dependences of Δα\Delta\alpha and ∣λ100∣\lvert\lambda_{100}\rvert are fairly in agreement with each other. Moreover, a correlation is also suggested between αoop\alpha_{\mathrm{oop}} and ∣λ100∣\lvert\lambda_{100}\rvert except for the Mn or Co rich region. The values of HinhH_{\mathrm{inh}} are shown in the Supplementary Information.

Origins of correlation between magnetostriction and magnetic damping

The finding of correlation between ∣λ100∣\lvert\lambda_{100}\rvert and α\alpha in the magnetic nitrides, especially the strong correlation with extrinsic α\alpha allows us to deeply understand the underlying mechanisms of magnetostriction and magnetic damping. The origin of α\alpha is a long-standing issue28 and the correlation between them, which was reported also for the other systems29,30,31,32, has not been understood yet. In this study, we performed the first-principles calculations for ∣λ100∣\lvert\lambda_{100}\rvert and intrinsic α\alpha. Because the precise ferrimagnetic spin structure has not been determined for Fe4−xMnxN\mathrm{Fe}_{4-x}\mathrm{Mn}_{x}\mathrm{N}, calculations were performed only for ferromagnetic Fe4−yCoyN\mathrm{Fe}_{4-y}\mathrm{Co}_{y}\mathrm{N}. Fe and Co can exist at both of the two occupied sites33,34 as shown in Figure 1a.

Figure 4a shows the representative results for the density of states (DOS) for the majority and minority spins of Fe4N\mathrm{Fe}_{4}\mathrm{N}, Fe3.2Co0.8N\mathrm{Fe}_{3.2}\mathrm{Co}_{0.8}\mathrm{N}, and Fe2Co2N\mathrm{Fe}_{2}\mathrm{Co}_{2}\mathrm{N} around EFE_{\mathrm{F}}. Note that the contribution from minority spin states much larger than majority spin especially around EFE_{\mathrm{F}}. In addition, the peak shift toward lower energy side observed for minority spin states by replacing Fe with Co. The yy dependence of theoretical MsM_{\mathrm{s}} is given by the black solid circles in Figure 4b and increasing Co content will lead to the reduction of total magnetic moment. However, the calculated MsM_{\mathrm{s}} is larger than the experimental values measured at 4 K (open star). Here, we notice that if the position of EFE_{\mathrm{F}} is shifted to the higher energy side, the calculated results are closer to the experimental results. This is simply explained due to the minority spin dominates the conduction band, hence shift toward higher energy side lead to decrease of the total magnetic moment. The red triangles and blue squares represent the results with EFE_{\mathrm{F}} shifted by 0.2 and 0.3 eV, respectively. The EFE_{\mathrm{F}} shift by 0.2 eV is appropriate because the EFE_{\mathrm{F}} shift by 0.3 eV gives rise to the calculated value lower than the experiment at the Co-rich composition. The EFE_{\mathrm{F}} position with the 0.2 eV shift is close to the peak top of the minority spin DOS of Fe4_4N as shown in Figure 4a.

Six-panel first-principles calculations on electronic structures and magnetic properties of Fe4-yCoyN

Figure 4. First-principles calculations on electronic structures and magnetic properties of Fe4−y_{4-y}Coy_yN. (a) Density of states (DOS) for the majority and minority spins of Fe4_4N (red line), Fe3.2_{3.2}Co0.8_{0.8}N (blue line), and Fe2_2Co2_2N (green line) around EFE_\mathrm{F}, predicted by first-principles calculations. (b) Composition dependence of theoretical and experimental MsM_\mathrm{s} values in Fe4−y_{4-y}Coy_yN. The black circles, red triangles, and blue squares represent the calculation results of shifting EFE_\mathrm{F} by 0, 0.2, and 0.3 eV, respectively. The open stars indicate the experimental MsM_\mathrm{s} values measured at 4 K. (c) Relationship between αcal\alpha_\mathrm{cal} and δ\delta for Fe4_4N, Fe3.2_{3.2}Co0.8_{0.8}N, and Fe2_2Co2_2N. The position of δ=0.01\delta= 0.01 eV is denoted by the dotted line. (d) Composition dependence of αcal\alpha_\mathrm{cal} at δ=0.01\delta= 0.01 eV (black squares) and αoop\alpha_\mathrm{oop} (purple inverted triangles). (e) Dependence of EtotE_\mathrm{tot} (black squares) and MAE (red circles) of Fe4_4N on cc without modulation of NvN_\mathrm{v}. (f) Calculated yy dependence of λ100\lambda_{100} for Fe4−y_{4-y}Coy_yN. The black circles and blue triangles show the results when Nv=0N_\mathrm{v} = 0 (no EFE_\mathrm{F} shift) and 0.4 / f.u., respectively. The green squares represent the

Using the DOS with the 0.2 eV EFE_\mathrm{F} shift and the Kamberský torque correlation model35, intrinsic α\alpha values were calculated for Fe4−y_{4-y}Coy_yN. Figure 4c shows the calculated values of α\alpha (αcal\alpha_\mathrm{cal}) on the scattering rate (δ\delta), in which the intraband component is dominant in the small δ\delta region whereas the interband component is dominant in the large δ\delta region. From the resistivity of ∼\sim110 μΩ\mu\Omega cm for Fe4−y_{4-y}Coy_yN36, the magnitude of δ\delta is estimated to be ∼\sim0.01 eV (denoted by the dotted line in Figure 4c), suggesting that the intraband component is dominant for Fe4_4N. Figure 4d compares αcal\alpha_\mathrm{cal} at δ=0.01\delta= 0.01 eV with the experimental values of αoop\alpha_\mathrm{oop} (intrinsic term of α\alpha).

The calculation qualitatively explains the trend of decreasing αoop\alpha_\mathrm{oop} with increasing Co amount.

The magnitude of αcal\alpha_\mathrm{cal} is proportional to D(EF)D(E_\mathrm{F})37. Thus, the EFE_\mathrm{F} shift away from the peak position of the minority spin DOS of Fe4_4N leads to the decrease in magnetic damping.

Figure 4e shows the total energy (EtotE_\mathrm{tot}) and magnetic anisotropy energy (MAE) of Fe4_4N as a function of the lattice constant perpendicular to the film surface (cc) without modulation of the valence electron number (NVN_\mathrm{V}). Assuming Etot=gc2+hc+iE_\mathrm{tot} = gc^2 + hc + i, where gg, hh, and ii are constants, λ100\lambda_{100} can be calculated from the following equation38:

λ100=−23h⋅d(MAE)dc(1)\lambda_{100}=-\frac{2}{3h}\cdot\frac{\mathrm{d}(\mathrm{MAE})}{\mathrm{d}c} \tag*{(1)}

Figure 4f shows the yy dependence of calculated λ100\lambda_{100} for Fe4−y_{4-y}Coy_yN using equation (1), where the black circles represent the results without modulating NVN_\mathrm{V} and the blue triangles represent the results in case that NvN_{\mathrm{v}} is increased by 0.4 / f.u. The experimental values are also given by the green squares for comparison. By increasing NvN_{\mathrm{v}} as in the case of calculation for α\alpha, the calculated results are fairly in agreement with the experimental results. One sees that the experimentally obtained composition dependences of α\alpha and λ100\lambda_{100} for Fe4−y_{4-y}Coy_yN (Figure 4d and Figure 4f) are well explained by the first-principles calculations. It is important to note that the sign and magnitude of λ100\lambda_{100} strongly depend on the derivative of MAE on the lattice parameter cc (d(MAE)/dc). Using second order perturbation analysis (see Supplementary Fig. 3), we clarified that the dominant contribution to the d(MAE)/dc is given by MAE from the spin conserving term in the minority-spin state, implying the crucial role of minority D(EF)D(E_{\mathrm{F}}). Therefore, D(EF)D(E_{\mathrm{F}}) plays a pivotal role to correlate between intrinsic α\alpha and ∣λ100∣|\lambda_{100}|.

In addition to the discussion about the correlation between “intrinsic” α\alpha and ∣λ100∣|\lambda_{100}|, we can discuss the correlation between “extrinsic” α\alpha and ∣λ100∣|\lambda_{100}| observed experimentally. The extrinsic α\alpha is partly attributable to two-magnon scattering. A previous paper22 mentioned that magnetostriction coupled with nonuniform stresses is a source of effective field for additional spin wave dynamics, which enhances two-magnon scattering and increases the extrinsic α\alpha.

Figure 5a displays the cross-sectional high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) image of the STO(001)/Fe4_4N. The inset is a fast Fourier transform (FFT) image. The image shown in Figure 5b is obtained by selecting appropriate diffraction spots from the FFT image and applying inverse FFT. The yellow marks denote the misfit dislocations, and the inhomogeneous strain is introduced around them. Since the lattice mismatches between STO\mathrm{STO} and Fe4−yCoyN\mathrm{Fe}_{4-y}\mathrm{Co}_y\mathrm{N} do not change significantly (−2.8-2.8 and −4.0%-4.0\% for y=0y = 0 and 44, respectively), we consider the misfit dislocation density hardly depends on yy. Therefore, the extrinsic α\alpha is enhanced in a material with large ∣λ100∣\lvert\lambda_{100}\rvert, and is also a useful indicator to investigate the magnitude of magnetostriction constant.

Microstructure of the Fe$_4$N film on the STO substrate

Figure 5. Microstructure of the Fe4_4N film on the STO substrate. (a) Cross-sectional HAADF-STEM image of the STO(001)/Fe4_4N. The inset shows a FFT image. (b) Inverse FFT image obtained from the inset of (a). The yellow marks show the misfit dislocations.

Giant tunability and bipolarity of magnetoelasticity

We demonstrated that the large negative λ100\lambda_{100} can be achieved for the epitaxially grown Fe4N\mathrm{Fe}_4\mathrm{N} film and the magnetostriction properties can be significantly varied by partially replacing Fe with Co or Mn. The magnitude of λ100\lambda_{100} was modulated in a wide range from −121-121 to +46+46 ppm. This fact means the giant tunability and bipolarity of magnetostriction in the ferromagnetic nitride. We found that the correlation between α\alpha and ∣λ100∣\lvert\lambda_{100}\rvert experimentally and theoretically. The first-principles calculations indicated that D(EF)D(E_{\mathrm{F}}) was the leading term to determine the magnitudes for both intrinsic α\alpha and λ100\lambda_{100}, suggesting that appropriate control of D(EF)D(E_{\mathrm{F}}) is indispensable for developing giant magnetoelastic materials. As for the correlation between extrinsic α\alpha and ∣λ100∣\lvert\lambda_{100}\rvert, the magnetostriction is a source of the two-magnon scattering, resulting in the large extrinsic α\alpha with increasing λ100\lambda_{100}. We believe that the findings obtained on the Fe4N\mathrm{Fe}_4\mathrm{N} research platform can give a guiding principle for exploring new giant magnetoelastic materials.

We address several advantages of ferromagnetic nitrides as a material for spintronic strain sensors. The giant tunability and bipolarity of magnetostriction are useful to improve the sensitivity of strain detection. The Fe4N\mathrm{Fe}_4\mathrm{N} with large negative λ100\lambda_{100} exhibits the alignment of magnetic moments along the orthogonal direction to the strain direction whereas the Fe2.3Co1.7N\mathrm{Fe}_{2.3}\mathrm{Co}_{1.7}\mathrm{N} with large positive λ100\lambda_{100} exhibits the moment alignment along the strain direction. For GMR/TMR-based strain sensors with two or more ferromagnetic layers, these characteristics are essential to obtain the large variation in device resistance determined by the relative angles between the adjacent ferromagnetic layers under the strain application. We expect that the large difference in magnetostriction between the Fe4N\mathrm{Fe}_4\mathrm{N} and the Fe2.3Co1.7N\mathrm{Fe}_{2.3}\mathrm{Co}_{1.7}\mathrm{N} significantly improves the device performance. As mentioned in the introductory part, the Fe4N\mathrm{Fe}_4\mathrm{N} shows the TMR by combining with a standard tunnel barrier material MgO, and the epitaxial Fe4N\mathrm{Fe}_4\mathrm{N} layer can be grown on the flexible substrate. Also, we may realize the largescaled device integration thanks to the fact that Fe4N\mathrm{Fe}_4\mathrm{N} does not contain minor metal elements and rare earth elements. Therefore, we conclude that magnetic nitrides are leading candidate materials for high-sensitive spintronic strain sensors.

Methods

Sample fabrication

The nitride films with a thickness of approximately 23 nm were grown on STO(001) substrates using molecular beam epitaxy (MBE). STO substrates were ultrasonically cleaned in the order of acetone, ethanol, and deionized water, and then dipped in buffered hydrofluoric acid. Fe and Co were supplied by using electron beam guns, Mn by a Knudsen Cell, and N by a radio-frequency (rf) radical source (Eiko engineering: ER-1000). The deposition rate was fixed at 0.2 Å/s for the total of Fe+Co and Fe+Mn, and the composition ratio was modulated by changing the supply rate ratio. High purity (6N) N₂ gas was supplied at 1.0 sccm and the rf input power was fixed at 300 W. The deposition temperature was set at 450 °C. For the Fe₄₋ᵧCoᵧN (ᵧ = 1.7 and 2.2) films, the temperature was lowered to 400 °C to prevent nitrogen desorption. Subsequently, a 2-nm-thick Ti capping layer was formed on the nitride layers in an MBE chamber to prevent surface oxidation.

Structural characterization

The structures of the samples were characterized by XRD with a Cu-Kα radiation source (SmartLab: Rigaku Corp.). The film thicknesses were evaluated by an x-ray reflectivity method. The surface morphologies and crystal growth orientations were also monitored by reflection high-energy electron diffraction during the film deposition. The microstructure of the STO/Fe4_4N was observed by HAADF-STEM (JEM-ARM200F: JEOL Ltd.). The composition ratios of the nitride films were evaluated using a field emission electron probe microanalyzer (JXA-8530F: JEOL Ltd.) with the standard samples for calibration.

Magnetic property measurement

The magnetization curves at room temperature were measured using a vibrating sample magnetometer (TM-VSM2614HGC-KIT, Tamakawa Co., Ltd.). The values of MsM_{\mathrm{s}} at 4 K were measured employing a superconducting quantum interference device magnetometer (MPMS3: Quantum Design Inc.). The magnetostriction was evaluated using an optical cantilever system30 (Toei Scientific Industrial Co., Ltd.). FMR spectra were measured at room temperature using a vector network analyzer (N5222A: Agilent Technology) and coplanar wave guides. Frequency-domain FMR measurements were carried out, sweeping rf current in the range of 1 ∼\sim 20 GHz and varying HH. For the IP configuration measurements, an rf-compatible prober station (TKSPHE-3V10: Toei Scientific Industrial Co., Ltd.) was used with 200 ≤H≤\le H \le 2400 Oe applied in the in-plane [100] direction of the films. For the OOP configuration measurements, a physical property measurement system (PPMS: Quantum Design Inc.) was used and 13.0 ≤H≤\le H \le 22.5 kOe was applied in the perpendicular direction to the film surface.

First-principles calculation

We performed first-principles density functional calculation using VASP[39] to obtain electronic structures and MAE of Fe4−yCoyN\mathrm{Fe}_{4-y}\mathrm{Co}_{y}\mathrm{N} series. The projector augmented-wave (PAW) potential was used to describe the behavior of core electrons[40]. The generalized gradient approximation proposed by Perdew, Burke, and Ernzerhof was adopted for the exchange and correlation energies[41]. A unit cell containing 4 Fe (Co) and 1 N atoms was constructed. The virtual crystal approximation (VCA) was used to describe the mixing between Fe and Co atoms[42]. We used the plane-wave cutoff energy of 350 eV for the wave function expansion and 30×30×3030 \times30 \times30 k-point mesh for wave vector integration in the first Brillouin zone. The damping was calculated using Torque Correlation Model by Kamberský35. The MAE was calculated by including spin-orbit interaction self-consistently. The magnetostriction constant was estimated by calculating MAE for a particular range of the lattice constant cc with fixing the volume of unit cell.

Data availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request. Source data are provided with this paper.

Code availability

Computer codes used in this study are available from the corresponding author upon reasonable request.

Acknowledgements

The authors thank S. Kanai, S. Fukami, R. Y. Umetsu, and Y. Sakuraba for support to measure FMR, S. Mizukami and H. Kurebayashi for their valuable comments, I. Narita for technical support to do the composition analysis, and K. Omura and Y. Murakami for help to perform the XRD measurements. This work was partly supported by the Collaborative Research Center on Energy Materials, Institute for Materials Research (IMR), Tohoku University, and the Cooperative Research Project of the Research Institute of Electric Communication, Tohoku University. The magnetization curve and XRD measurements were carried out at the Cooperative Research and Development Center for Advanced Materials, IMR, Tohoku University. The first-principles calculations of the magnetic damping and the magnetostriction were performed on a Numerical Materials Simulator at NIMS. This work was supported by JSPS KAKENHI Grant Numbers JP22K18894 and JP22H04966, and Initiative to Establish Next-generation Novel Integrated Circuits Centers (X-nics) under Grant JPJ011438, the Ministry of Education, Culture, Sports, Science and Technology (MEXT).

Author contributions

Film growth, XRD measurements, and magnetization curve measurements were carried out by K.I. K.I. and T.S. performed magnetostriction measurements with support from Y.E. K.I. and T.S. performed FMR measurements. Y.S. carried out TEM observations. First-principles calculations were carried out by I.K. with support from Y.M. K.I. and T.S. wrote the draft. All the authors discussed the results and commented on the manuscript. T.S. conceived the project.

Competing interests

The authors declare no competing interests.

Correspondence and requests for materials should be addressed to Keita Ito.

Composition dependences of magnetic properties

Figure 3. Composition dependences of magnetic properties. Composition dependences of (a) MsM_s, (b) K1K_1, (c) λ100\lambda_{100}, (d) αip\alpha_{\mathrm{ip}}, (e) αoop\alpha_{\mathrm{oop}}, and (f) Δα\Delta\alpha together with ∣λ100∣\lvert\lambda_{100}\rvert for the Fe4−xMnxN\mathrm{Fe}_{4-x}\mathrm{Mn}_x\mathrm{N} and Fe4−yCoyN\mathrm{Fe}_{4-y}\mathrm{Co}_y\mathrm{N} films.

experimental results.

Supplementary Information

Supplementary Figure S1 shows the x-ray reflectivity (XRR) profile for the Fe4N\mathrm{Fe}_{4}\mathrm{N} film. The black and red lines correspond to the obtained data and the fitting line, respectively.

Figure S1. XRR profile of the Fe4N\mathrm{Fe}_{4}\mathrm{N} film.

Supplementary Table S1 summarizes the film thicknesses of the nitride films estimated from the XRR. The accuracy of the estimated thicknesses is high because the fitting was done with high precision.

SamplesThickness (nm)
Fe4_4N27
Fe3.9_{3.9}Mn0.1_{0.1}N21
Fe3.6_{3.6}Mn0.4_{0.4}N25
Fe3_3MnN25
Fe3.8_{3.8}Co0.2_{0.2}N22
Fe3.5_{3.5}Co0.5_{0.5}N21
Fe3.2_{3.2}Co0.8_{0.8}N25
Fe2.3_{2.3}Co1.7_{1.7}N21
Fe1.8_{1.8}Co2.2_{2.2}N22

Table S1. The thicknesses of the nitride films estimated from the XRR.

According to the model for the optical cantilever method, the displacement of reflected position of the laser light (Δs\Delta s) is expressed by the following equation20:

Δs=Cλ100⋅hfilmhsub⋅EfilmEsub⋅1+νsub1+νfilm,\Delta s = C\lambda_{100} \cdot\frac{h_{\mathrm{film}}}{h_{\mathrm{sub}}} \cdot\frac{E_{\mathrm{film}}}{E_{\mathrm{sub}}} \cdot\frac{1+\nu_{\mathrm{sub}}}{1+\nu_{\mathrm{film}}},

where CC is the coefficient determined by the measurement system geometry, λ100\lambda_{100} is the magnetostriction constant along [100] direction, hsubh_{\mathrm{sub}} (hfilmh_{\mathrm{film}}) is the substrate (film) thickness, EsubE_{\mathrm{sub}} (EfilmE_{\mathrm{film}}) is the Young’s modulus of the substrate (film), and νsub\nu_{\mathrm{sub}} (νfilm\nu_{\mathrm{film}}) is the Poisson’s ratio of the substrate (film). Supplementary Table 2 summarizes EE and ν\nu values of SrTiO3\mathrm{SrTiO_3}S1, Fe4N\mathrm{Fe_4N}S2, Fe3MnN\mathrm{Fe_3MnN}S3, Fe3CoN\mathrm{Fe_3CoN}S4, and Co4N\mathrm{Co_4N}S5. For the evaluation of λ100\lambda_{100} for all the samples, the EE and ν\nu values of Fe4N\mathrm{Fe_4N} were used because accurate EE and ν\nu values for each composition are unknown.

The ν\nu values of Fe4N\mathrm{Fe_4N}, Fe3MnN\mathrm{Fe_3MnN}, Fe3CoN\mathrm{Fe_3CoN}, and Co4N\mathrm{Co_4N} are close, while the EE values of Fe3MnN\mathrm{Fe_3MnN}, Fe3CoN\mathrm{Fe_3CoN}, and Co4N\mathrm{Co_4N} are larger than that of Fe4N\mathrm{Fe_4N}. This means that obtained λ100\lambda_{100} values of the Fe4−xMnxN\mathrm{Fe_{4-x}Mn_xN} and Fe4−yCoyN\mathrm{Fe_{4-y}Co_yN} films are slightly overestimated. However, the overestimation of λ100\lambda_{100} does not affect the main conclusion of this study.

MaterialsEE (GPa)ν\nu
SrTiO3_32780.24
Fe4_4N1620.36
Fe3_3MnN (Ferromagnetic)2020.31
Fe3_3CoN2170.32
Co4_4N2390.31

Table 2. EE and ν\nu values of SrTiO3_3, Fe4_4N, Fe3_3MnN, Fe3_3CoN, and Co4_4N.

As mentioned in the main text, the fitting functions for the resonance linewidth (Δf0\Delta f_0) involve the inhomogeneous magnetic field (HinhH_{\mathrm{inh}}): Δf0(H)=αip(γ/2π)(2H+Ha,ip+4πMeff)+(γ/2π)Hinh\Delta f_0(H) = \alpha_{\mathrm{ip}}(\gamma/2\pi)(2H + H_{\mathrm{a,ip}} + 4\pi M_{\mathrm{eff}}) + (\gamma/2\pi)H_{\mathrm{inh}} for the IP configuration, and Δf0(H)=αoop(γ/2π)(2H−2Hs)+(γ/2π)Hinh\Delta f_0(H) = \alpha_{\mathrm{oop}}(\gamma/2\pi)(2H - 2H_{\mathrm{s}}) + (\gamma/2\pi)H_{\mathrm{inh}} for the OOP configuration. Thus, we evaluated the values of HinhH_{\mathrm{inh}} together with αip\alpha_{\mathrm{ip}} and αoop\alpha_{\mathrm{oop}}. In the case of IP configuration, the measured values of Δf0\Delta f_0 are remarkably scattered as seen in Figure 2f, which gives rise to the estimated values of HinhH_{\mathrm{inh}} with significant errors, e.g. Hinh=113±506H_{\mathrm{inh}} = 113 \pm506 Oe for x=0.5x = 0.5. Thus, we cannot discuss the values of HinhH_{\mathrm{inh}} for the IP configuration because of low accuracy of estimation. In contrast to the IP configuration, the OOP configuration leads to the values of HinhH_{\mathrm{inh}} with relatively small errors. Supplementary Fig. 2 plots the composition dependence of HinhH_{\mathrm{inh}} estimated in the OOP configuration. One sees that the values of HinhH_{\mathrm{inh}} are negligibly small and do not show a clear composition dependence.

Composition dependence of inhomogeneous magnetic field

Figure Supplementary 2. Composition dependence of inhomogeneous magnetic field (HinhH_{\mathrm{inh}}) estimated in the out-of-plane configuration.

Supplementary Fig. 3 shows the spin-resolved magnetocrystalline anisotropy energies (MAE) for Fe4_4N and Co4_4N in the second-order perturbation of spin-orbit interaction as a function of cc lattice parameter. We clarified that the dominant contribution to the d(MAE)/dc\mathrm{d}(\mathrm{MAE})/\mathrm{d}c is given by competing term of negative trend of dMAE(↓⇒↓)/dc\mathrm{dMAE}(\downarrow\Rightarrow\downarrow)/\mathrm{d}c and positive trend of d(MAE(↑⇒↓)/dc\mathrm{d}(\mathrm{MAE}(\uparrow\Rightarrow\downarrow)/\mathrm{d}c, where σ⇒σ′\sigma\Rightarrow\sigma^{\prime} indicates the spin-transition process in the second order perturbation from the initial spin σ\sigma state to the intermediate spin σ′\sigma^{\prime} state. For Fe4_4N, the former is much more dominant than the latter, hence give the large negative magnetostriction. This means that the anisotropy of orbital moment of Fe related to the MAE(↓⇒↓)\mathrm{MAE}(\downarrow\Rightarrow\downarrow) significantly contributes to the magnetostriction of Fe4_4N. However, the latter is slightly more dominant for Co4_4N. Nevertheless, both cases implying the crucial role of minority-spin D(EF)D(E_{\mathrm{F}}). Therefore, D(EF)D(E_{\mathrm{F}}) plays a major role to correlate between intrinsic α\alpha and ∣λ100∣\lvert\lambda_{100}\rvert.

Spin-resolved MAE for Fe$_4$N and Co$_4$N

Figure 3 (PDF p. 37). Spin-resolved MAE for Fe4_4N and Co4_4N in the second-order perturbation of spin-orbit interaction as a function of cc lattice parameter.

Supplementary references

S1. Pandech, N., Sarasamak, K., & Limpijumnong, S. Elastic properties of perovskite ATiO3\mathrm{ATiO}_{3} (A = Be, Mg, Ca, Sr, and Ba) and PbBO3\mathrm{PbBO}_{3} (B = Ti, Zr, and Hf): First principles calculations. J. Appl. Phys. 117, 174108 (2015).

S2. Takahashi, T., Burghaus, J., Music, D., Dronskowski, R., & Schneider, J. M. Elastic properties of γ′\gamma^{\prime}-Fe4N\mathrm{Fe}_{4}\mathrm{N} probed by nanoindentation and ab initio calculation. Acta Mater. 60, 2054-2060 (2012).

S3. Wu, H., Sun, H., & Chen, C. Superior magnetic and mechanical property of MnFe3N\mathrm{MnFe}_{3}\mathrm{N} driven by electron correlation and lattice anharmonicity. Phys. Rev. B 91, 064102 (2015).

S4. Wu, Z & Meng, J. Elastic and electronic properties of CoFe3N\mathrm{CoFe}_{3}\mathrm{N}, RhFe3N\mathrm{RhFe}_{3}\mathrm{N}, and IrFe3N\mathrm{IrFe}_{3}\mathrm{N} from first principles. Appl. Phys. Lett. 90, 241901 (2007).

S5. Azouaoui, A., Benzakour, N., Hourmatallah, A., & Bouslykhane, K. Study of structural, electronic, magnetic, elastic, and thermodynamic properties of Co4N\mathrm{Co}_{4}\mathrm{N} antiperovskite by first principles and Monte Carlo simulation. Phys. Status Solidi B 258, 2000339 (2021).

References

  1. [39]Kresse G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169-11186 (1996).pubmed.ncbi.nlm.nih.gov/9984901
  2. [40]Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953-17979 (1994).pubmed.ncbi.nlm.nih.gov/9976227
  3. [41]Perdew, J. P., Burke, K., & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865-3868 (1996).pubmed.ncbi.nlm.nih.gov/10062328
  4. [42]Eckhardt, C., Hummer, K., & Kresse, G. Indirect-to-direct gap transition in strained and unstrained SnₓGe₁₋ₓ alloys. Phys. Rev. B 89, 165201 (2014).

Paper details

vlnc:W_n2dvwew4
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents