betaFeedback

WorksPhysics and Astronomy

Atomistic spin model of single pulse toggle switching in Mn2_2Rux_xGa Heusler alloys

Abstract

Single femtosecond pulse toggle switching of ferrimagnetic alloys is an essential building block for ultrafast spintronics. Very different element-specific demagnetization dynamics is believed to be a hard limit for switching in ferrimagnets. This suggests that ferrimagnets composed of two ions of different nature, such as rare earth transition metal alloys, are necessary for switching. However, experimental observation of toggle switching in Mn2_2Rux_xGa Heusler alloys, has contested this limit since Mn ions are of the same nature. To shed some light into this question, we present an atomistic spin model for the simulation of single pulse toggle switching of Mn2_2Rux_xGa. The magnetic parameters entering in our model are extracted from previous experimental observations. We show that our model is able to quantitatively reproduce measured magnetization dynamics of single pulse toggle switching. We demonstrate that differently to previous understanding toggle switching in Mn2_2Rux_xGa is possible even when both Mn sublattices demagnetization at very similar rate.

Single pulse femtosecond toggle switching in ferrimagnets has attracted a lot of attention as a promising solution for low energy, faster memory applications. [1, 2, 3, 4] It has already been demonstrated in micro and nanostructures [5, 6], to switch magnetic tunnel junctions [7], as passive component to induce switching in ferromagnets [8] and even using picosecond electric pulses [9]. Toggle switching has been mostly found in one class of material, systems composed of transition metals rare-earth, e.g. GdFeCo [1] and TbFeCo [10] alloys, and Gd/Co[11] and Tb/Co stacks [12]. Integrating all optical switching with spintronics using this class of material has been proposed[13], however, Mn2_2Rux_xGa alloys, the second class of material showing toggle switching, is better suited [14]. While for GdFeCo intense research has provided a large amount of experimental data permitting the validation of a number of theoretical models, for Mn2_2Rux_xGa, there only exist a few experimental works showing ultrafast magnetization dynamics and switching [14, 15, 16, 17].

The Mn2_2Rux_xGa Heusler alloy crystallises in the cubic space group F4ˉ\bar{4}3m with the magnetic Mn atoms on the 4a and 4c sites (Figure 1a). Spins at 4a and 4c sites are coupled antiferromagnetically, whereas the respective Mn 4a- and 4c sublattices are coupled ferromagnetically [18]. The Ga atoms appear at the 4b sites and Ru atoms at the 4d sites of the lattice, and their magnetic contribution can be neglected. Ultrafast magnetization dynamics of the magneto-optically active Mn 4c sublattice was first measured by Bonfiglio et al. [15]. In a subsequent work, they concluded that Mn 4c sublattice shows an ultrafast demagnetization (∼100\sim100 fs) followed by either a secular equilibrium or by a fast remagnetization (∼1\sim1 ps). By using a phenomenological model based on the so-called four temperature – electron, phonon, and spin temperatures of Mn 4a and 4c – these dynamics were interpreted as a signature of strong exchange-driven relaxation [16]. Spin temperature models however are unable to describe angular momentum transfer and

Crystal structure, exchange-constant schematic, and magnetic moments of Mn2RuxGa

Figure 1. a) Crystal structure of Mn2_2Rux_xGa with the two magnetic Mn-sublattices represented as blue and red arrows. b) Schematic of the exchange constants J4a−4aJ_{4a-4a}, J4a−4cJ_{4a-4c} and J4a−4cJ_{4a-4c}. c) Mn-specific atomic magnetic moments, symbols correspond to experimental data [19] and lines to the values used in the atomistic spin model.

switching.

Experiments carried out by Davies et al. demonstrated that switching is possible in Mn2_2Rux_xGa, but only when the initial temperature (T0T_0) lies below the magnetization compensation temperature (TMT_\mathrm{M}) [20]. Based upon a phenomenological model for the magnetization dynamics of two-sublattice magnets [21], the authors argued that in order to link static thermodynamic properties (equilibrium TMT_\mathrm{M}) to highly non-equilibrium dynamics (switching), both sublattices should demagnetize at the same rate, conserving the total angular momentum during the whole process. This picture explains their observation of the switching onset for a wide range of system parameters, including switching with picosecond-long pulses. This interpretation collides with the phenomenology proposed by Bonfiglio et al. [16], which states that both sublattices only demagnetize at the same rate after ∼1.5\sim1.5 ps. Still, since data on the magnetization dynamics of the switching process was missing, the question remained open. An apparent step forward along this direction was made by Banerjee et al. [14] by measuring the dynamics of the Mn 4c sublattice when switching occurs. They were unable to observe an explicit switching of the sign of the magneto-optical signal, however the dynamics of the Mn 4c sublattice behaved like those observed by Bonfiglio [16]. These observations differ strongly with the well-known element specific signal switching measured in GdFeCo [22]. Recently, however, Banerjee and co-workers have clearly observed the dynamics of magnetization switching of the Mn 4c sublattice [17], in clear disagreement to their own previous observations [14]. This disagreement could be related to the different strength of the resetting magnetic field, stronger in the latter. Similarly to the previous works [14, 16, 20], the understanding of the physics behind the switching process rested on phenomenological arguments. An attempt to describe switching in Mn2_2Rux_xGa using a first-principles model exists [23]. Although this model provides useful insights on the potential origin of switching, due to its simplicity it is unable to describe the temperature dependence of the switching condition (T0<TMT_0 < T_M) observed by Davies et al. [20] and Banerjee et al. [14]. A quantitative model able to account for thermodynamic aspects of the magnetization switching dynamics in Mn2_2Rux_xGa is necessary but so far missing.

In this work we address this issue by presenting an atomistic spin model to describe ultrafast magnetization and switching in Mn2_2Rux_xGa alloys. Atomistic spin models have demonstrated themselves to be able to describe the dynamics of the magnetization switching in transition metal rare-earth alloys [1, 24] and multilayers [25, 26]. Atomistic spin models are based on a semiclassical spin Hamiltonian, which is defined by the exchange and anisotropy constants as well as the atomic magnetic moments (see Supplemental Material for details). In a minimal model for Mn2_2Rux_xGa, one needs to determine at least three exchange constants, J4a−4aJ_{4a-4a}, J4c−4cJ_{4c-4c} and J4a−4cJ_{4a-4c} (see Figure 1 b)). Within a Heisenberg spin model, the values of the exchange constants determine the critical temperature TcT_c, and in ferrimagnets, besides TcT_c, the magnetization compensation temperature, TMT_M, if it exists. The reported TcT_c range from Tc=625T_c = 625 K [18] (x=0.68x = 0.68), Tc=450T_c = 450 K (x>0.5x > 0.5) [27], Tc=550T_c = 550 K (x=0.7x = 0.7) [16]. We fix TcT_c to about 630 K which is at the upper end of reported critical

temperatures. Here, we use values of the exchange constants already determined experimentally [18], but slightly re-scaled such that our model reproduces the experimental values of both TcT_c and TMT_M (Figure 2). The anisotropy constant in Mn2_2Rux_xGa has been reported to be site-specific, with dz,4c=1.1664⋅10−23d_{z,4c} = 1.1664 \cdot10^{-23} J and dz,4a=0d_{z,4a} = 0 [18], which we use in our model. We note that the role of the anisotropy in the ultrafast magnetization dynamics is minimal since dz/J≪1d_z/J \ll1 and it is included to fix the average magnetization towards the zz-axis. Any choice of relatively small values of the anisotropy constant would yield the same simulation results.

Figure 2. Magnetic properties of Mn2_2Rux_xGa alloys gained from atomistic spin model simulations. a) The equilibrium magnetization for a Mn2_2Ru0.68_{0.68}Ga alloy as function of temperature of the Mn4a_{4a} sublattice in red, the Mn4c_{4c} sublattice and the total magnetization Mn4a+4c=∣Mn4a∣−∣Mn4c∣_{4a+4c} = |\mathrm{Mn}_{4a}| - |\mathrm{Mn}_{4c}|. b) displays the magnetization compensation temperature TMT_M as a function of Ru-concentration, xx. Black dots correspond to experimental measurements [14] and red crosses from our model simulations. The red line is a linear fit of the simulated results and is guidance to the eye.

Lastly values for the site-specific Mn atomic magnetic moments are needed. Based upon experimental data, we assume that the atomic moment of the Mn 4a sublattice stays constant at μs,4a=2.88μB\mu_{s,4a} = 2.88\mu_B for the range of Ru-concentration studied here [19] (Figure 1c). Differently, the Mn 4c sites have been shown to have a stronger dependence on the Ru-concentration [19] (Figure 1c). Here we use values of μs,4c=4.71μB⋅x\mu_{s,4c} = 4.71\mu_B \cdot x (Figure 1c) that are both close to the experimentally measured values and provide good results for TMT_M (see Figure 2). These values not only compare well to those found in experiments [19, 28] but also agree with the general trend found in first-principles calculations [23, 29]. Simulations of our model are able to reproduce well the observation of an increasing TMT_M for an increasing Ru-concentration. For x<0.66x < 0.66, our model starts to deviate from the experiments [14] (Figure 2 b)).

The shared wisdom of single femtosecond pulse toggle switching in ferrimagnets is based on experimental observations in only one class of material, transition metal rare-earth compounds. Despite their structural or morphological differences, in those materials the conditions for switching are based in the same working principles, the rare-earth spin sublattice response to the laser pulse is slower than that of the transition metal. The physical reason behind this difference relies in the core character of the highly localized 4f4f electron spins in Gd, plus the absence of orbital magnetic moment (L=0L = 0) [30]. The laser only excites them indirectly, by their coupling to the 5d6s5d6s itinerant electrons. The transition metal spins respond quickly to changes in the electronic structure band due to the quick temperature rise.

Based on this it has been argued for long that toggle switching is only possible for two sublattice compounds with components that demagnetize at different enough rates. Since data on site-specific dynamics is unavailable, it is unclear whether or not this criterion holds in Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga}. It is tempting to draw similarities to GdFeCo to explain switching in Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga}. For example, since Mn 4a4a spins are localized, they could play the role of the slow rare earth, while Mn 4c4c have a more delocalized character, and thus play the role of the transition metal. In this work we show differently, that Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga} switches even though its two sublattices show similar demagnetization times, unlocking different demagnetization times as a necessary condition for switching. Another hard constrain for switching Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga} is that it is only possible when the initial temperature lies below TMT_M [14, 20]. Davies et al. [20] found that this condition is robust and that can be explained assuming that the dynamics is exchange-driven, which has as a consequence that dM4a/dt=−dM4c/dtdM_{4a}/dt = -dM_{4c}/dt. This condition also holds, when the coupling to the heat-bath is similar for both sublattices and they demagnetize at similar rates. Here, we demonstrate that switching is possible when the initial demagnetization dynamics is dominated by the coupling to the heat-bath instead the exchange relaxation.

We use the atomistic stochastic-Landau-Lifshitz-Gilbert equation (sLLG) [31] to describe the magnetization dynamics of the Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga} alloys and the two-temperature model (TTM) to describe the electron temperature TelT_{\mathrm{el}} and the phonon temperature TphT_{\mathrm{ph}} (see Supplemental Material (SM) for details) [32, 33]. The parameters defining the TTM for Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga} are not established yet. We use parameters similar to GdFeCo alloys, which are close to those used by Bonfiglio et al. within the 4TM [16]. The complete set of system parameter used in our model are summarized in table II of the SM. Within atomistic spin dynamics models, the demagnetization time scales with τ∼μat/(αTc)\tau\sim\mu_{\mathrm{at}}/(\alpha T_c) [21, 34]. In two-sublattice magnets, the parameters, α\alpha, TcT_c and μat\mu_{\mathrm{at}} are sublattice-specific. While TcT_c and μat\mu_{\mathrm{at}} are determined by equilibrium properties as discussed before, the value of the damping parameter is related to the coupling to the heat-bath. One method to estimate this value from experiments is to photo-excite magnetization precession. These experiments have been conducted in Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga} and measured by time resolved Faraday effect as a function of the applied field and temperature. From the decay of the precession the intrinsic damping parameter has been determined to have values smaller than 0.02 far from TMT_M [15]. We chose a damping value (α4c=0.01\alpha_{4c} = 0.01 and α4c=0.013\alpha_{4c} = 0.013) that reproduces the ultrafast magnetization dynamics of the Mn 4c4c sublattice during a switching event as measured by Banerjee et al. [17] and allows switching for x>0.85x > 0.85 similar to the experimental findings of Banerjee et al. [14]. We note, that Davies et al. [20] find switching already for x=0.75x = 0.75 which could be reproduced for a different choice of damping values (discussion in the SM). With this our model is able to quantitatively reproduce the recently measured demagnetization dynamics of the 4c4c lattice by Banerjee et al. (Figure 3).

Comparison between experimentally measured Mn4c dynamics and atomistic spin model simulations

Figure 3. Comparison between the experimentally measured Mn4c_{4c} dynamics for a switching event (points) [17] and our atomistic spin model simulations (lines).

How does our model compare to the current understanding of switching in Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga}? Previous works [16, 17, 20] have suggested, that exchange-driven dynamics dominate the first step of the demagnetization and switching process. This is assumed since the inter-lattice antiferromagnetic exchange is stronger than in (for example) transition metal rare earth alloys. Exchange-relaxation stems from processes driving both sublattices to a mutual equilibrium. Thus, it is unlikely that exchange processes play a role on the first steps of the dynamics. Moreover, since exchange processes describe transfer of angular momentum between sublattices, it is more likely that those processes dominate when the sublattice magnetic order is small rather than when its saturated. Near equilibrium, relaxation by coupling to the heat-bath dominates, while for situations of non-equilibrium and reduced magnetic order, the exchange-relaxation dominates.

Figure 4 shows the site-specific dynamics for three characteristic Ru-concentrations of Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga} for x=0.76x = 0.76 (a), x=0.86x = 0.86 (b) and x=0.96x = 0.96 (c). These three cases represent alloys below/at/above the threshold of x=0.9x = 0.9, which defines the experimentally found switching condition [14].

A closer look to mz,totalm_{z,\mathrm{total}} in Figure 4 indicates that up to the first picosecond mz,totalm_{z,\mathrm{total}} is not conserved, although it only changes slightly due to the similar demagnetization dynamics of the Mn 4a and 4c sublattices (note, that the scale of the right y-axis is much smaller than the scale of the left axis). This means that relaxation by coupling to the heat-bath dominates. This relaxation can in turn be interpreted as excitation of ferromagnetic (optical) magnons (following Banerjee and co-workers[17]). However, as the sublattice magnetization reduces to small values (∼1\sim1 ps), the value of mz,totalm_{z,\mathrm{total}} stays constant in time. This means that exchange-relaxation dominates the dynamics (conservation of total angular momentum). This can be interpreted as excitation of antiferromagnetic (acoustic) magnons. Total angular momentum is conserved for a relatively long period of time (few picoseconds). Once the system has reached the exchange relaxation dominated regime (≈1\approx1 ps) the interpretation of switching of Davies and co-workers remains valid. The exchange relaxation results in ∣Δm4a∣=∣Δm4c∣|\Delta m_{4a}| = |\Delta m_{4c}|, but since ∣m4a,0∣<∣m4c,0∣|m_{4a,0}| < |m_{4c,0}|

Figure 4. Site-specific magnetization dynamics of the 4a and 4c Mn sites (red, blue; left axis) after a 100 fs laser pulse excitation at t=0t = 0 (Gaussian pulse peak) for an increasing Ruthenium concentration (α4c=0.01\alpha_{4c} = 0.01, α4a=0.013\alpha_{4a} = 0.013). The total magnetization (mz,total=mz,4a+mz,4cm_{z,\mathrm{total}} = m_{z,4a} + m_{z,4c}) is shown as a black line (right axis). All laser parameters were the same for all three simulations. a) shows the no-switching scenario for a Mn2_2Ru0.76_{0.76}Ga alloy. b) shows the non-deterministic scenario for a Mn2_2Ru0.86_{0.86}Ga alloy, with a prolonged demagnetization state. c) shows the switching scenario for a Mn2_2Ru0.96_{0.96}Ga alloy.

means that m4am_{4a} crosses zero before m4cm_{4c} does and therefore giving rise to successful magnetic switching. Differently to the interpretation of Davies et al. [20] this condition, ∣m4a,0∣<∣m4c,0∣|m_{4a,0}| < |m_{4c,0}|, holds because the first steps of the sublattice demagnetization are driven by direct coupling to the heat-bath with relaxation at similar speeds.

Our picture can be easily checked. We argue that switching happens if ∣m4a∣<∣m4c∣|m_{4a}| < |m_{4c}| when the system enters the exchange relaxation regime, which dominates the dynamics for small values of the sublattice magnetic order. The sign of the total magnetization at that moment can be controlled by the coupling to the heat-bath of each sublattice, namely the value of α4a(4c)\alpha_{4a(4c)}. By increasing α4a\alpha_{4a} the demagnetization due to the coupling to the heat-bath increases so that ∣m4a∣<∣m4c∣|m_{4a}| < |m_{4c}| can be fulfilled even for smaller values of xx. The effects of this change of α4a\alpha_{4a} on mtotalm_{\mathrm{total}} would change it from a behaviour as shown in Figure 4 a) towards the one in Figure 4 c). The full result of this analysis is shown in Figure 6 in the Supplemental Material. Where we observe that by increasing the value of α4a\alpha_{4a}, a wider range of Ru-concentrations switch and thus lowering the switching threshold set by the compensation temperature, (x=0.90x = 0.90 in our model).

Switching behavior as a function of ruthenium concentration and absorbed laser energy for three values of the 4a-sublattice damping

Figure 6. Switching behavior as color of Mn2RuxGa\mathrm{Mn}_2\mathrm{Ru}_x\mathrm{Ga} alloys as function of the Ruthenium concentration xx and the absorbed laser energy for different dampings α4a\alpha_{4a} of the 4a4a-sublattice. Red areas indicate switching behavior, blue marks areas without switching and grey indicates a prolonged transient ferromagnetic state or a demagnetized state. The damping α4c=0.01\alpha_{4c}=0.01 of the 4c4c-sublattice was kept constant while α4a\alpha_{4a} was varied from α4a=0.011\alpha_{4a}=0.011 (top) to α4a=0.013\alpha_{4a}=0.013 (middle) to α4a=0.015\alpha_{4a}=0.015 (bottom).

Similar to the experimental work of Davies et al. [20] we investigate switching as function of the starting temperature T0T_0 and the pulse duration (Figure 5). Our results compare qualitatively well to the experimental observations of Ref. [20]. Quantitative comparison between our model, based on the experimental measurements of Banerjee et al. [14], and Davies et al. are difficult without modifying some of the material parameters used in the model. The main difference between the two experiments is the value of the compensation temperature for similar Ru-concentrations ($T_{M,x=0.75} = 370 K [20] and TM,x=0.7=245T_{M,x=0.7} = 245 K [14]). In order to mimic the conditions of Davies et al., we use Mn2_2Ru0.98_{0.98}Ga, which features a compensation temperature of TM=375T_M = 375 K (experimentally TM,x=0.75=370T_{M,x=0.75} = 370 K). The pulse-duration switching threshold decreases as the system base temperature approaches a threshold temperature with a similar slope as experiments. Experimentally this threshold temperature is close to the compensation temperature, while in our model is a bit further. However our findings suggest that the condition T0<TMT_0 < T_M is not a limiting, sufficient condition for switching in Mn2_2Rux_xGa.

Figure 5. Switching behavior as color of Mn2_2Ru0.98_{0.98}Ga as function of the starting temperature T0T_0 and the pulse duration. Red areas indicate switching behavior, blue marks areas without switching and grey indicates a prolonged transient ferromagnetic state or a demagnetized state.

To summarize, we have presented an atomistic spin model able to describe single pulse toggle switching in Mn2_2Rux_xGa Heusler alloys. The parameters of the spin model are based on experimental measurements and can reproduce key material parameters such as TcT_c, M(T)M(T) or the Ru-concentration dependence of TMT_M. We show that our model is able to quantitatively reproduce measured magnetization dynamics of single pulse toggle switching and we demonstrate that differently to previous understanding toggle switching in Mn2_2Rux_xGa is possible even when both Mn sublattices demagnetization at very similar rate.

Acknowledgments

We gratefully acknowledge support by the Deutsche Forschungsgemeinschaft through SFB/TRR 227 "Ultrafast Spin Dynamics", Project A08.

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Supplementary Material

Details of the estimation of the system parameters

We model Mn2_2Rux_xGa Heusler alloys based on the Heisenberg spin Hamiltonian:

H=−∑i≠jJijSi⋅Sj−∑idzSz,i2(1)\mathcal{H}=-\sum_{i\ne j}J_{ij}\mathbf{S}_i\cdot\mathbf{S}_j-\sum_i d_z S_{z,i}^2 \tag*{(1)}

Here Si=μi/μs,i\mathbf{S}_i=\boldsymbol{\mu}_i/\mu_{s,i} represents a classical, normalized spin vector at site ii. Here, μs,i\mu_{s,i} is the atomic magnetic moment of the site ii. Si\mathbf{S}_i couples to its nearest neighboring spin Sj\mathbf{S}_j via the coupling constant JijJ_{ij}, and dzd_z represents the on-site anisotropy with easy-axis along the zz-axis.

The considered Heisenberg exchange parameters are the ferromagnetic exchange couplings J4a−4aJ_{4a-4a} and J4c−4cJ_{4c-4c} as well as the antiferromagnetic J4a−4cJ_{4a-4c}. The values of the exchange and the uniaxial anisotropy for Mn2_2Ru0.68_{0.68}Ga are based on the values of a molecular field model [35] derived in Ref. 18. The exchange parameters are adjusted to yield a Curie temperature of about 630 K which is at the upper end of reported Curie temperatures, for instance, it is close to the value reported in Ref. 18 of Tc=625T_{\mathrm{c}}=625 K for x=0.68x=0.68, where xx stands for Ru-concentration. Additional experiments find an approximately constant Curie temperature of 450 K for x>0.5x>0.5 [27], whereas Ref. 16 reports Tc=550T_{\mathrm{c}}=550 K for x=0.7x=0.7. Furthermore the TcT_{\mathrm{c}} measured in Ref. 27 features a peak around x=0.4x=0.4 which could explain why our parameters only reproduce TcompT_{\mathrm{comp}} for x>0.68x>0.68. Thus, a dependence of the exchange parameters on Ru-concentration may therefore be possible for lower values of xx. Since we are mostly interested in Ru-concentration for which switching has been experimentally demonstrated (x>0.7x>0.7), we restrict our study to Ru concentrations of x=0.6−1.0x=0.6-1.0. We note that we have increased J4a−4aJ_{4a-4a} by 10% to reproduce the magnetization compensation temperature as reported in Ref. [14] while keeping Tc=630T_{\mathrm{c}}=630 K. Using exchange constants with the ratios as stated in Ref. 18 couldn’t reproduce the experimentally measured Mn4c_{4c} dynamics of C. Banerjee et al. [17] that are shown in Figure 3 (main text). However, the fact that there are reports of largely varying different Curie temperatures [16, 18, 27] and compensation temperature [14, 20] indicate that there seems to be a wide spectrum of valid exchange constants between different experiments, so that an adjustment of 10% of one of the parameters seems reasonable.

The uniaxial anisotropy is also considered to be site-specific, in particular, the anisotropy energy density Kz,4c=216K_{z,4c}=216 kJm−3^{-3} and Kz,4a=0K_{z,4a}=0 kJm−3^{-3} taken from Ref. [18]. Assuming a unit cell size of (≈0.6\approx0.6 nm)3^3 [36] we obtain dz,4c=1.1664⋅10−23d_{z,4c}=1.1664\cdot10^{-23} J as on-site anisotropy and dz,4a=0d_{z,4a}=0. The anisotropy is included to yield an alignment along the zz-axis after demagnetization or switching. Since it is much smaller the Heisenberg exchange it does not have meaningful impact on the switching itself. Furthermore ref. 15 finds, based on XMCD experiments, different gg-factors, g4a=2.05g_{4a}=2.05 and g4c=2.00g_{4c}=2.00 for the 4a4a- and 4c4c-sublattice. However since both γ\gamma and μs\mu_s are proportional to gig_i this does not enter the LLG.

The spin dynamics of this system are described by the atomistic stochastic-Landau-Lifshitz-Gilbert equation (sLLG) [31]

(1+αi2)μs,iγ∂Si∂t=−(Si×Hi)−αi(Si×(Si×Hi)).(2)\frac{(1+\alpha_i^2)\mu_{s,i}}{\gamma}\frac{\partial\mathbf{S}_i}{\partial t}=-(\mathbf{S}_i\times\mathbf{H}_i)-\alpha_i\left(\mathbf{S}_i\times(\mathbf{S}_i\times\mathbf{H}_i)\right). \tag*{(2)}

Where γ\gamma represents the gyromagnetic ratio and αi\alpha_i is a site-specific atomic damping parameter. Here, we also draw on experimental observations to estimate the values of the damping parameters. Photo-excited spin precession was observed by time resolved Faraday effect as a function of the applied field and temperature. From the decay of the precession the intrinsic damping parameter was also determined to have values smaller than α=0.02\alpha=0.02, far from compensation [15]. Here, we decided to use α4c=0.01\alpha_{4c}=0.01 and α4c=0.013\alpha_{4c}=0.013 since it reproduces experimental observations as discussed in the main text. The temperature dynamics are described using the two-temperature model (TTM) that describes the electron temperature TelT_{\mathrm{el}} and the phonon temperature TphT_{\mathrm{ph}} via a pair of two coupled differential equations [32, 33]:

Cel∂Tel∂t=−gep(Tel−Tph)+Pl(t)(3)C_{\mathrm{el}}\frac{\partial T_{\mathrm{el}}}{\partial t}=-g_{\mathrm{ep}}\left(T_{\mathrm{el}}-T_{\mathrm{ph}}\right)+P_l(t) \tag*{(3)}
Cph∂Tph∂t=+gep(Tel−Tph).(4)C_{\mathrm{ph}}\frac{\partial T_{\mathrm{ph}}}{\partial t}=+g_{\mathrm{ep}}\left(T_{\mathrm{el}}-T_{\mathrm{ph}}\right). \tag*{(4)}

CelC_{\mathrm{el}} and CphC_{\mathrm{ph}} represent the specific heat of the electron- and phonon system and Pl(t)P_l(t) describes the absorbed energy of the electron system, coming from the laser. Since experimental data on the electron and phonon temperature dynamics is missing, we use similar parameters to typical GdFeCo values for our TTM and similar ones to the experimental work by Bonfiglio et al. (Ref. [16]). Table I provides an overview of the used TTM-parameters in comparison to Ref. [16] and to parameters used in GdFeCo from Ref. 37 and Ref. 38.

TTMUnitMn2_2Rux_xGaMn2_2Rux_xGa[16]GdFeCo[38]GdFeCo[37]
CphC_{\mathrm{ph}}J/m3^3K3×1063 \times10^62.27×1062.27 \times10^63×1063 \times10^63×1063 \times10^6
gphg_{\mathrm{ph}}J/m3^3Ks6×10176 \times10^{17}8×10178 \times10^{17}2×10172 \times10^{17}17×101717 \times10^{17}
γ\gammaJ/m3^3K2^2350484714700

Table I. Two temperature model parameters comparison between GdFeCo and Mn2_2Rux_xGa between different sources.

The laser pulse is assumed to be Gaussian shaped with a FWHM of 100 fs. The electron temperature TelT_{\mathrm{el}} yielding from the TTM is used to scale the temperature effects in the spin system. This is done by including a Langevin thermostat, which adds an effective field-like stochastic term ζi\zeta_i to the effective field Hi=ζi(t)−∂H∂SiH_i = \zeta_i(t) - \frac{\partial H}{\partial S_i} with white noise properties [39]:

⟨ζi(t)⟩=0and⟨ζi(0)ζj(t)⟩=2αikBTelμs,iδijδ(t)γ.(5)\langle\zeta_i(t) \rangle= 0 \quad\text{and} \quad\langle\zeta_i(0)\zeta_j(t) \rangle= \frac{2\alpha_i k_{\mathrm{B}}T_{\mathrm{el}}\mu_{\mathrm{s},i}\delta_{ij}\delta(t)}{\gamma}. \tag*{(5)}

The complete set of system parameter used in our model are summarized in Table II.

H\mathcal{H}ValueUnitTTMUnit
J4a−4aJ_{4a-4a}1.28×10−211.28 \times10^{-21}[J]CphC_{\mathrm{ph}}3×1063 \times10^6 [J/Km3^3]
J4c−4cJ_{4c-4c}4.0×10−224.0 \times10^{-22}[J]CelC_{\mathrm{el}}γel⋅Te\gamma_{\mathrm{el}} \cdot T_{\mathrm{e}} [J/Km3^3]
J4a−4cJ_{4a-4c}−4.85×10−22-4.85 \times10^{-22}[J]γel\gamma_{\mathrm{el}}350 [J/K2^2m3^3]
γ\gamma1.76×10−211.76 \times10^{-21}[JTs\frac{\mathrm{J}}{\mathrm{T}s}]gepg_{\mathrm{ep}}6×10176 \times10^{17} [J/sKm3^3]
dzd_z1.17×10−231.17 \times10^{-23}[J]
μs,4a\mu_{\mathrm{s},4a}2.88[μB\mu_{\mathrm{B}}]
μs,4c\mu_{\mathrm{s},4c}4.71⋅x4.71 \cdot x[μB\mu_{\mathrm{B}}]
α4a\alpha_{4a}0.013
α4c\alpha_{4c}0.01

Table II. Table of the Heisenberg spin Hamiltonian parameters (left) and the two temperature model (TTM) (right).

Switching behaviour in dependence of Gilbert damping parameters

Figure 6 shows the switching behavior of Mn2_2Rux_xGa alloys as function of the Ru-concentration xx and the absorbed laser energy for different dampings α4a\alpha_{4a}. Red areas indicate switching behavior, blue marks non-switched simulations and grey areas indicate simulations with a prolonged transient ferromagnetic state or a demagnetized state.

The simulation was counted as switched, when the 4a-sublattice crossed mz=0m_z = 0 and reached a threshold of mz,4a<−0.12m_{z,4a} < -0.12 after 15 ps (starting at positive mzm_z values). Otherwise it was counted as demagnetized if ∣mz,4a∣<0.12\lvert m_{z,4a}\rvert< 0.12, or remagnetized if mz,4a≥0.12m_{z,4a} \mathrel{\ge} 0.12 (see Figure 4 for examples).

The damping α4c=0.01\alpha_{4c} = 0.01 was kept constant while α4a\alpha_{4a} was varied from α4a=0.011\alpha_{4a} = 0.011 (top) to α4a=0.013\alpha_{4a} = 0.013 (middle) to α4a=0.015\alpha_{4a} = 0.015 (bottom). Figure 6, shows clearly distinguish behaviors for the simulated alloys depending on the Ru-concentration xx. For low absorbed laser energies below $5.5 \cdot 10^8$ J/m3^3 no switching occurs for all Ru concentrations considered here. This is due to the insufficient energy to temporarily demagnetize both sublattices. For all three cases, we find that for low Ruthenium concentrations below a damping-dependent threshold value the alloy does not switch, independent of the laser energy. Only above that threshold we find deterministic switching. In the top panel with α4c=0.01\alpha_{4c} = 0.01 and α4a=0.011\alpha_{4a} = 0.011, we find the threshold Ru-concentration for switching to be around x≈0.9−0.95x \approx0.9 - 0.95. When the damping of the 4a sublattice is increased to α4a=0.013\alpha_{4a} = 0.013 the threshold moves to x≈0.85−0.9x \approx0.85 - 0.9, which approximately corresponds to the switching threshold found in Ref. 14. Finally, for α4a=0.015\alpha_{4a} = 0.015 (bottom) the switching threshold decreases to x≈0.8x \approx0.8. Therefore we find, that by increasing the element specific damping discrepancy the switching threshold moves towards lower Ruthenium concentrations. Our results compare best to the experiments by Banerjee and co-workers (threshold around x=0.8−0.9x = 0.8 - 0.9) [14] when choosing α4c=0.01\alpha_{4c} = 0.01 and $\alpha_{4a} = 0.013` (Fig. 6 middle). We note that in our model the only parameter that directly depends on xx is μs,4c\mu_{s,4c}, which impacts the

speed of the 4c-sublattice. Our model shows that by increasing xx, μs,4c\mu_{s,4c} also increases and in turn the sublattice demagnetization speed continuously slows down, up to the point where the demagnetization speed difference in the 4a- and 4c-sublattice is large enough to enable switching behavior. The model also shows that this relatively different demagnetization speed can also be controlled by the intrinsic site-dependent damping parameters, which influences the xx-dependent threshold between switching and non switching behavior.

References

  1. [1]T. A. Ostler, J. Barker, R. F. L. Evans, R. W. Chantrell, U. Atxitia, O. Chubykalo-Fesenko, S. El Moussaoui, L. Le Guyader, E. Mengotti, L. J. Heyderman, F. Nolting, a. Tsukamoto, A. Itoh, D. Afanasiev, B. A. Ivanov, a. M. Kalashnikova, K. Vahaplar, J. Mentink, A. Kirilyuk, T. Rasing, and V. V. Kimel, Ultrafast heating as a sufficient stimulus for magnetization reversal in a ferrimagnet, Nature Communications 3, 666 (2012).DOI
  2. [2]A. V. Kimel and M. L. Writting magnetic memory with ultrashort light pulses, Nature Reviews Materials 4, 189 (2019).DOI
  3. [3]A. El-Ghazaly, J. Gorchon, R. B. Wilson, A. Pattabi, and J. Bokor, Progress towards ultrafast spintronics applications, Journal of Magnetism and Magnetic Materials 502, 166478 (2020).
  4. [4]J. Barker and U. Atxitia, A Review of Modelling in Ferrimagnetic Spintronics, Journal of the Physical Society of Japan 90, 81001 (2021).DOI
  5. [5]L. Le Guyader, M. Savoini, S. El Moussaoui, M. Buzzi, A. Tsukamoto, A. Itoh, A. Kirilyuk, T. Rasing, A. V. Kimel, and F. Nolting, Nanosecond sub-100 picosecond all-optical magnetization switching in gd feco microstructures., Nat Commun 6, 5839 (2015).DOI
  6. [6]A. El-Ghazaly, B. Tran, A. Ceballos, C.-H. Lambert, A. Pattabi, S. Salahuddin, F. Hellman, and J. Bokor, Ultrafast magnetization switching in nanoscale magnetic dots, Applied Physics Letters 114, 232407 (2019).
  7. [7]J.-Y. Chen, L. He, J.-P. Wang, and M. Li, All-Optical Switching of Magnetic Tunnel Junctions with Single Subpicosecond Laser Pulses, Physical Review Applied 7, 21001 (2017).DOI
  8. [8]J. Gorchon, C.-H. Lambert, Y. Yang, A. Pattabi, R. B. Wilson, S. Salahuddin, and J. Bokor, Single shot ultrafast all optical magnetization switching of ferromagnetic Co/Pt multilayers, Applied Physics Letters 111, 42401 (2017).
  9. [9]Y. Yang, R. B. Wilson, J. Gorchon, C.-H. Lambert, S. Salahuddin, and J. Bokor, Ultrafast magnetization reversal by picosecond electrical pulses, Science Advances 3, 10.1126/sciadv.1603117 (2017).DOI
  10. [10]T.-m. . M. Liu, T. Wang, A. H. Reid, M. Savoini, X. Wu, B. Koene, P. Granitzka, C. E. Graves, D. J. Higley, Z. Chen, G. Razinskas, M. Hantschmann, A. Scherz, J. Sto, A. Tsukamoto, B. Hecht, A. V. Kimel, A. Kirilyuk, T. Rasing, and H. A. Du, Nanoscale con fi nement of all-optical magnetic switching in tbfe co - competition with nanoscale heterogeneity, Nano Letters 15, 6862 (2015).DOI
  11. [11]M. L. M. Lalieu, M. J. G. Peeters, S. R. R. Haenen, R. Lavrijsen, and B. Koopmans, Deterministic all-optical switching of synthetic ferrimagnets using single femtosecond laser pulses, Physical Review B 96, 220411 (2017).DOI
  12. [12]L. Avilés-Félix, A. Olivier, G. Li, C. S. Davies, L. Álvaro-Gómez, M. Rubio-Roy, S. Auffret, A. Kirilyuk, A. V. Kimel, T. Rasing, L. D. Buda-Prejbeanu, R. C. Sousa, B. Dieny, and I. L. Prejbeanu, Single-shot all-optical switching of magnetization in Tb/Co multilayer-based electrodes, Scientific Reports 10, 5211 (2020).
  13. [13]M. L. M. Lalieu, R. Lavrijsen, and B. Koopmans, Integrating all-optical switching with spintronics, Nat Commun 10, https://doi.org/10.1038/s41467-018-08062-4 (2019).
  14. [14]C. Banerjee, N. Teichert, K. E. Siewierska, Z. Gercsi, G. Y. P. Atcheson, P. Stamenov, K. Rode, J. M. D. Coey, and J. Besbas, Single pulse all-optical toggle switching of magnetization without gadolinium in the ferrimagnet mn2ruxga, Nature Communications 11, 4444 (2020).DOI
  15. [15]G. Bonfiglio, K. Rode, K. Siewerska, J. Besbas, G. Y. P. Atcheson, P. Stamenov, J. M. D. Coey, A. V. Kimel, T. Rasing, and A. Kirilyuk, Magnetization dynamics of the compensated ferrimagnet Mn2RuxGa, Physical Review B 100, 104438 (2019).DOI
  16. [16]G. Bonfiglio, K. Rode, G. Y. P. Atcheson, P. Stamenov, J. M. D. Coey, A. V. Kimel, T. Rasing, and A. Kirilyuk, Sub-picosecond exchange-relaxation in the compensated ferrimagnet Mn2RuxGa, Journal of Physics: Condensed Matter 33, 135804 (2021).DOI
  17. [17]C. Banerjee, K. Rode, G. Atcheson, S. Lenne, P. Stamenov, J. M. D. Coey, and J. Besbas, Ultrafast double pulse all-optical reswitching of a ferrimagnet, Phys. Rev. Lett. 126, 177202 (2021).DOI
  18. [18]C. Fowley, K. Rode, Y.-C. Lau, N. Thyagarajah, D. Betto, K. Borisov, G. Atcheson, E. Kampert, Z. Wang, Y. Yuan, S. Zhou, J. Lindner, P. Stamenov, J. M. D. Coey, and A. M. Deac, Magnetocrystalline anisotropy and exchange probed by high-field anomalous hall effect in fully compensated half-metallic mn2ruxGa thin films, Phys. Rev. B 98, 220406 (2018).
  19. [19]D. Betto, N. Thiyagarajah, Y.-C. Lau, C. Piamonteze, M.-A. Arrio, P. Stamenov, J. M. D. Coey, and K. Rode, Site-specific magnetism of half-metallic mn2ruxga thin films determined by x-ray absorption spectroscopy, Phys. Rev. B 91, 094410 (2015).
  20. [20]C. S. Davies, G. Bonfiglio, K. Rode, J. Besbas, C. Banerjee, P. Stamenov, J. M. D. Coey, A. V. Kimel, and A. Kirilyuk, Exchange-driven all-optical magnetic switching in compensated 3d ferrimagnets, Physical Review Research 2, 32044 (2020).DOI
  21. [21]J. H. Mentink, J. Hellsvik, D. V. Afanasiev, B. A. Ivanov, A. Kirilyuk, A. V. Kimel, O. Eriksson, M. I. Katsnelson, and T. Rasing, Ultrafast spin dynamics in multisublattice magnets, Physical Review Letters 108, 057202 (2012).pubmed.ncbi.nlm.nih.gov/22400955
  22. [22]I. Radu, K. Vahaplar, C. Stamm, T. Kachel, N. Pontius, F. Radu, R. Abrudan, H. Dürr, T. Ostler, J. Barker, R. Evans, R. Chantrell, A. Tsukamoto, A. Itoh, A. Kirilyuk, T. Rasing, and A. Kimel, Ultrafast magnetism as seen by x-rays (2012) p. 82601M.DOI
  23. [23]G. P. Zhang, R. Meadows, A. Tamayo, Y. H. Bai, and T. F. George, An attempt to simulate laser-induced all-optical spin switching in a crystalline ferrimagnet, AIP Advances 10, 125323 (2020).arxiv.org/abs/2011.02001
  24. [24]F. Jakobs, T. A. Ostler, C.-H. Lambert, Y. Yang, S. Salahuddin, R. B. Wilson, J. Gorchon, J. Bokor, and U. Atxitia, Unifying femtosecond and picosecond single-pulse magnetic switching in Gd-Fe-Co, Physical Review B 103, 104422 (2021).DOI
  25. [25]C. Xu, T. A. Ostler, and R. W. Chantrell, Thermally induced magnetization switching in Gd/Fe multilayers, Physical Review B 93, 54302 (2016).
  26. [26]S. Gerlach, L. Oroszlany, D. Hinzke, S. Sievering, S. Wienholdt, L. Szunyogh, and U. Nowak, Modeling ultrafast all-optical switching in synthetic ferrimagnets, Phys. Rev. B 95, 224435 (2017).DOI
  27. [27]H. Kurt, K. Rode, P. Stamenov, M. Venkatesan, Y.-C. Lau, E. Fonda, and J. M. D. Coey, Cubic mn2Ga thin films: Crossing the spin gap with ruthenium, Phys. Rev. Lett. 112, 027201 (2014).pubmed.ncbi.nlm.nih.gov/24484042
  28. [28]L. Yang, B. Liu, F. Meng, H. Liu, H. Luo, E. Liu, W. Wang, and G. Wu, Magnetic properties of Heusler alloy Mn2RuGe and Mn2RuGa ribbons, Journal of Magnetism and Magnetic Materials 379, 1 (2015).DOI
  29. [29]I. Galanakis, K. Özdoğan, E. Şaşıoğlu, and S. Blügel, Effect of disorder on the magnetic properties of cubic Mn2RuXGa compounds: A first-principles study, Journal of Applied Physics 116, 33903 (2014).DOI
  30. [30]F. Frietsch, J. Bowlan, R. Carley, M. Teichmann, S. Wienholdt, D. Hinzke, U. Nowak, K. Carva, P. M. Oppeneer, and M. Weinelt, Disparate ultrafast dynamics of itinerant and localized magnetic moments in gadolinium metal, Nature Communications 6, 8262 (2015).
  31. [31]U. Nowak, Classical Spin Models in Handbook of Magnetism and Advanced Magnetic Materials (John Wiley and Sons, Ltd 2007) (2007).
  32. [32]M. I. Kaganov, I. M. Lifshitz, and L. V. Tanatarov, Relaxation between electrons and crystalline lattices, JETP 173 (1957).
  33. [33]J. K. Chen, D. Y. Tzou, and J. E. Beraun, A semiclassical two-temperature model for ultrafast laser heating, International Journal of Heat and Mass Transfer 49, 307 (2006).DOI
  34. [34]I. Radu, C. Stamm, A. Eschenlohr, F. Radu, R. Abrudan, K. Vahaplar, T. Kachel, N. Pontius, R. Mitzner, K. Holldack, A. Föhlisch, T. a. Ostler, J. H. Mentink, R. F. L. Evans, R. W. Chantrell, A. Tsukamoto, A. Itoh, A. Kirilyuk, a. V. Kimel, and T. Rasing, Ultrafast and distinct spin dynamics in magnetic alloys, SPIN 5, 1550004 (2015).DOI
  35. [35]J. S. Smart and J. H. Van Vleck, Effective field theories of magnetism, Physics Today 19, 77 (1966), https://doi.org/10.1063/1.3048415.
  36. [36]N. Thiyagarajah, Y.-C. Lau, D. Betto, K. Borisov, J. M. D. Coey, P. Stamenov, and K. Rode, Giant spontaneous hall effect in zero-moment mn2ruxga, Applied Physics Letters 106, 122402 (2015).
  37. [37]J. Barker, U. Atxitia, T. A. Ostler, O. Hovorka, R. W. Chantrell, O. Chubykalo-Fesenko, and R. W. Chantrell, Two-magnon bound state causes ultrafast thermally induced magnetisation switching, Scientific reports 3, 3262 (2013).
  38. [38]A. Mekonnen, A. R. Khorsand, M. Cormier, A. V. Kimel, A. Kirilyuk, A. Hrabec, L. Ranno, A. Tsukamoto, A. Itoh, and T. Rasing, Role of the inter-sublattice exchange coupling in short-laser-pulse-induced demagnetization dynamics of gdco and gdfeo alloys, Phys. Rev. B 87, 180406 (2013).DOI
  39. [39]U. Atxitia, O. Chubykalo-Fesenko, R. Chantrell, U. Nowak, and A. Rebei, Ultrafast spin dynamics: The effect of colored noise, Physical Review Letters 102, 057203 (2009).pubmed.ncbi.nlm.nih.gov/19257543

Paper details

vlnc:W_sd4qayxc
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