betaFeedback

WorksPhysics and Astronomy

THz-induced phonomagnetism in diamagnetic quantum paraelectric KTaO3_3

Abstract

The current efforts striving to develop new ways of data manipulation are aimed at ultrafast control of magnetization in magnetic materials, as well as at inducing magnetic moments in diamagnetics. We demonstrate that in the diamagnetic quantum paraelectric KTaO3_3, the electric field of circularly polarized THz pulses with an amplitude of ∼300 \sim 300\,kV/cm induces a transient magnetic-like response by resonantly exciting its degenerate soft polar phonon. This phonon-mediated response was measured using the THz pump—optical probe technique via the time-resolved magneto-optic Faraday effect. Our detection scheme was set up to cancel out the major part of the electro-optic Kerr effect which also usually significantly contributes to the transient response. The Kerr-effect-related signal was further suppressed by subtracting the experimental data related to oppositely circularly polarized THz radiation. Thus, we were able to unambiguously identify a temperature-dependent magnetic-like behavior of the KTaO3_3 crystal manifested by the extracted Faraday rotation. We developed a theoretical model describing well quantitatively the measured curves of the transient Faraday effect signal. However, their amplitudes exhibit an unexpected temperature dependence, which might be a key to a deeper understanding of the observed phonomagnetic effect.

In the context of ongoing efforts to bring up novel efficient mechanisms of data manipulation and operation for future uses in data storage and computational applications, much attention has been paid to interactions of light with magnetic order, especially on an ultrafast scale. Indeed, it is well known that ultrashort laser pulses allow for a very fast control of magnetic states in solids. Previously, magnetic phase transitions induced by ultrafast laser heating were studied [1]. Later, non-thermal control of magnetization by instantaneous photomagnetic pulses was used [2]. In this case, circularly polarized fs laser pulses excited and coherently controlled the spin dynamics in magnets via the inverse Faraday effect, generating an effective magnetic field whose direction depended on the handedness of the circularly polarized laser beam.

More recently, transient changes in crystal structure or in magnetic order were achieved by non-linear phonon coupling and/or by phonomagnetic effects [3]. It was demonstrated that resonant pumping of polar phonons by intense THz (or infrared) radiation may trigger a non-linear coupling with another phonon of different symmetry (e.g. a Raman-active phonon), which in turn can cause a transient change in the crystal structure or in the magnetic order [4, 5, 6, 7, 8, 9, 10]. Metal-insulator phase transitions were induced in this way [11], and even a non-equilibrium superconducting phase near room temperature was observed [12, 13, 14, 15].

In 2017, Nova et al. [16] excited two orthogonally polarized phonons near 7 THz in an antiferromagnetic ErFeO3_3 using intense infrared radiation polarized along the [110] crystallographic direction. They have shown that non-linear mixing of two orthogonally polarized phonons enhances magnetization and excites the magnon at 0.75 THz in this material [16]. Juraschek et al. [17, 18] developed their reasoning and proposed to pump degenerate phonons in non-magnetic substances using circularly polarized THz radiation in order to induce a quasi-static magnetic moment as a consequence of a circular motion of atoms. The authors calculated the orbital magnetic moments of axial phonons in 33 different compounds and predicted the largest magnetic moments in CsH, SrTiO3_3 and KTaO3_3 [18]. Based on the assumption that circularly polarized phonons with time-dependent polarization P(t)\mathbf{P}(t) lead to a magnetization M∝P×∂tP\mathbf{M} \propto\mathbf{P} \times\partial_t\mathbf{P}, the concept was named dynamical multiferroicity [17, 19].

The possibility to generate magnetization via orbital magnetic moment μph\mu_{\mathrm{ph}} of coherently excited circularly polarized phonons [17, 18] was recently confirmed by Basini et al. [20] who drove the degenerate mode in SrTiO3_3 near room temperature. The peak of the pumping spectrum was set at 3 THz, coinciding with the phonon frequency. Their measurements yielded two remarkable results:

  1. The measured response was four orders of magnitude higher than the theoretically expected value. This also suggests the phonon-induced magnetic moment μph\mu_{\mathrm{ph}} to be four orders of magnitude higher than predicted in Ref. [18]. To explain the discrepancy, various mechanisms of transfer of angular momentum from the phononic to the electronic degree of freedom were proposed [21, 22, 23]. Merlin even proposed that the large magnetic-like signal is generated by a non-Maxwellian field that breaks the time-reversal symmetry [24, 25].

  1. Their detected signal featured two maxima, at 0.6 and 6 THz. The maximum at 6 THz (i.e., twice the pumping THz frequency) was mostly caused by the THz-field-induced electro-optic Kerr effect (TKE) associated with the non-linear polarization Pi,NL=ε0χijkl(3)EjTkTlP_{i,\mathrm{NL}} = \varepsilon_0\chi_{ijkl}^{(3)}E_jT_kT_l (here χijkl(3)\chi_{ijkl}^{(3)} is the third-order non-linear electric susceptibility and EjE_j, Tk,lT_{k,l} are electric field components of the probing optical and pumping THz electromagnetic waves, respectively) [20, 26]. The peak at 0.6 THz was attributed to a combination of the TKE and the THz-field-induced magneto-optic Faraday effect (TFE).

In this article, we present results of measurements of the THz-field induced magneto-optical TFE and electro-optical TKE on a single crystal of KTaO3_3. KTaO3_3 has large Born effective charges and it exhibits a temperature dependence of the soft mode (SM) phonon frequency very similar to that of SrTiO3_3 [27, 28]. The advantage of KTaO3_3 over SrTiO3_3 is that its SM damping is very low (≤60\le60 GHz below 50 K) and its crystal structure remains cubic down to absolute zero. This enables measurements at low temperatures where the SM potential is shallow so an efficient coupling with the THz field is expected, leading to higher phonon amplitudes and higher values of μph\mu_{\mathrm{ph}}. Also, according to theoretical analysis [18], the magnetic moments owing to the SMs in KTaO3_3 and SrTiO3_3 should be the highest among all perovskites. Being aware of the difficulties caused by the superposition of the TFE with the concomitant TKE, we optimized the experimental setup to reject this unwanted contribution without any need of material’s response modelling. Thus we were able to identify unambiguously the magnetic-like TFE response directly from the acquired data.

A (001)-cut KTaO3_3 single crystal in the form of an optically polished parallelepiped with dimensions approximately 7×8×0.637 \times8 \times0.63 mm3^3 and edges oriented along [100] and [10] axes was used (these directions also coincided with xx and yy axes of our laboratory frame). Its quantum paraelectric behavior was verified through the temperature dependence of the dielectric permittivity, see Fig. S2 in Supplementary information (SI) [27].

The pump-probe experiments were performed on the TELBE high-field THz facility at the Helmholtz-Zentrum Dresden-Rossendorf [29, 30] where a linear electron accelerator generates tunable intense multi-cycle THz pulses at a repetition rate of 50 kHz. We used pulses with an energy of ∼10 μ\sim10\,\muJ, an electric field amplitude of ∼300\sim300 kV/cm in the focus, a central frequency of Ω0=0.7\Omega_0 = 0.7 THz and a full-width-at-half-maximum (FWHM) of ∼65\sim65 GHz pertaining to the power spectrum (see Fig. S5 in [27]). Experiments were performed with an either linearly or circularly polarized THz pump beam. In the former case, the pump polarization was oriented along the [100] axis of KTaO3_3; only TKE can be induced. In the latter case, a quartz quarter-wave plate was used to transform the linearly polarized pumping pulse into a circular one; in fact, it involved an elliptical imperfection, namely a difference of about 20% in perpendicular electric-field amplitudes along xx and yy axes. We carried out experiments with both right-hand-polarized (RHP) and left-hand-polarized (LHP) pump. For probing we used time-synchronized femtosecond Ti:sapphire laser pulses centered at the wavelength of ωp=800\omega_{\mathrm{p}} = 800 nm and linearly polarized along the [110] crystallographic direction. Polarization-sensitive balanced photodetection was employed to detect the THz-pump-induced optical component FF polarized along the [11ˉ0][1\bar{1}0] direction, perpendicularly to the incident probe field EE, as described in End Matter (Figure 4). The data traces F(t)F(t) were recorded as a function of the pump-probe delay tt; the photodiodes had a broadband sensitivity, so they measured both the transient signal induced at the incident probe beam frequency ωp\omega_{\mathrm{p}} (main band) centered in the THz spectrum at Ω≈0\Omega\approx0 and the one induced at the sidebands, ωp±2Ω0\omega_{\mathrm{p}} \pm2\Omega_0, which appeared in the traces as an oscillation at Ω≈2Ω0\Omega\approx2\Omega_0. The experiments were performed in transmission geometry at temperatures down to 11 K. A complete experimental schema with more technical details is provided in SI (Fig. S4 in [27]).

Figure 4. Detection scheme. Incident probe beam: E=(E,E,0)/2\mathbf{E}=(E,E,0)/\sqrt{2}; output probe beam: Eprobe=(E−G+F,E−G−F,0)/2\mathbf{E}_{\mathrm{probe}}=(E-G+F,E-G-F,0)/\sqrt{2}. Vectors E\mathbf{E}, F\mathbf{F} and G\mathbf{G} indicate the polarization state of the probe beam after the interaction with the sample. In the case of the Kerr effect the FF and GG are phase shifted by π/2\pi/2, the output polarization is elliptical and the measured difference signal is proportional to ℜ(G∗F)\Re(G^{*}F), vanishing in the first order. In the case of the Faraday effect, G=0G=0 and FF is in phase with EE, thus the difference signal is ℜ(E∗F)\Re(E^{*}F).

It was shown experimentally and theoretically [20, 26] that in THz pump—optical probe experiments, the TKE always generates an electromagnetic contribution, for both linearly and circularly polarized THz pump. Therefore, for studying efficiently the TFE, it is desirable to conceive an experimental scheme minimizing the TKE contribution. Compared with the experimental setup of Ref. [20], in our setup, we turned the incident polarization of the probe beam by 45∘45^\circ to the [110] direction which allowed us to avoid a half-wave plate between the sample and the photodiodes. Indeed, this scheme allowed us to efficiently cancel out the TKE contribution for the circularly polarized pump, as briefly described in End Matter and in detail in SI [27] by solving the nonlinear wave equation. In fact, the phase of the TKE-induced contribution is essentially π/2\pi/2 shifted compared to the phase of the unperturbed probe wave, making the detection insensitive to the first-order TKE contribution. Even more importantly, for this particular experimental arrangement, the TKE contribution to the main band vanishes. Finally, the differential signal from balanced photodiodes can be written as

ΔΓRHP/LHP=±ΔΓTFE+δΓTKE,(1)\Delta\Gamma_{\mathrm{RHP/LHP}} = \pm\Delta\Gamma_{\mathrm{TFE}} + \delta\Gamma_{\mathrm{TKE}}, \tag*{(1)}

where the right- and left-hand-polarized pump is denoted by RHP and LHP, respectively, and they induce a positive and negative TFE signal ΔΓTFE\Delta\Gamma_{\mathrm{TFE}}. In contrast, the sign of the second-order residual TKE contribution δΓTKE\delta\Gamma_{\mathrm{TKE}} is independent of the pump handedness (see End Matter). After the subtraction ΔΓdif=ΔΓRHP−ΔΓLHP=2ΔΓTFE\Delta\Gamma_{\mathrm{dif}} = \Delta\Gamma_{\mathrm{RHP}} - \Delta\Gamma_{\mathrm{LHP}} = 2\Delta\Gamma_{\mathrm{TFE}}, the signal is free of the TKE, provided that the RHP and LHP pumps are indeed circularly polarized or slightly elliptically polarized with the principal axes along [100] and [10] crystallographic directions of KTaO3_3 (which is true in our case, as shown in Sec. IV of SI).

Figure 1 displays experimental results obtained for nominally circularly polarized pump pulses as a function of temperature. In the plots all curves clearly contain a quasi-static component with Ω≈0\Omega\approx0 and an oscillating one at Ω≈2Ω0\Omega\approx2\Omega_0. Since the 2Ω02\Omega_0 component is forbidden for TFE, it must be due to the residual TKE, and this part of the signal provides us with an order-of-magnitude estimate of the amplitude of the detected TKE. The quasi-static part of the signal has a fast Gaussian-like shape between roughly 0 and 10–15 ps and a long-lived contribution, extending up to 30 ps for higher temperatures (81 K) and exceeding 60 ps for the lowest temperatures. The long-lived component clearly exhibits the same sign for RHP and LHP pump at all temperatures and thus it is supposedly the TKE residual which should vanish by subtracting ΔΓRHP−ΔΓLHP\Delta\Gamma_{\mathrm{RHP}} - \Delta\Gamma_{\mathrm{LHP}}. The fast Gaussian-like component has again the same sign for the two pump polarizations at high temperatures (61–81 K) but, upon further cooling, the sign of LHP signal changes. The sign reversal upon the LHP to RHP switching corresponds to the time-reversal induced by the THz pump to the crystal and thus we observe here an onset of the TFE. Indeed, between 10 and 50 K the SM resonance frequency increases from 0.6 to 1 THz (see Fig. S1 in [27]), so its component with a non-zero orbital moment can be resonantly driven by the circularly polarized pump pulses inducing a net magnetization, as predicted in Ref. [31] and as expressed by Eqs. (S23) and (S24) in [27].

Figure 1. Time dependences of the differential signal ΔΓ\Delta\Gamma at various temperatures in KTaO3_3 upon pumping with (a) LHP and (b) RHP THz pulse. The scale in arbitrary units is the same for the two plots.

As discussed above and in End Matter, calculating the difference between the LHP and RHP data, the TKE contribution is subtracted and a pure TFE response is obtained. However, the oscillations at 2Ω02\Omega_0 have a period of ∼0.7\sim0.7 ps and in the TELBE setup a relative time delay (of the order of a few hundreds of fs) between the arrivals of LHP and RHP pump pulses can arise leading to a slightly different time origin of the parts (a) and (b) in Figure 1. This shift of the time origin may be due to several factors, namely the rotation of the quarter-wave plate and imperfections in parallelism of its surfaces, as well as software corrections of the pulse time jitter in the TELBE beamline. In this context we introduced an additional parameter, representing the time delay between the LHP and RHP data at a given temperature. Its resulting values, not exceeding 300 fs, were chosen so as to minimize the 2Ω02\Omega_0 component of the ΔΓdif\Delta\Gamma_{\mathrm{dif}} curves. This leads to time traces which can be interpreted as due to the sole TFE effect; these are shown in Figure 2(a) as a function of temperature. (For illustration, the inset shows results of subtracting the raw data without any time shift.) We also stress that the time shift values we used do not influence the quasi-static peak originating in the TFE (see Fig. S7 in SI). The corresponding TFE spectra calculated by Fourier transformation are shown in Figure 2(b). Its inset then displays the temperature dependence of the amplitude of the quasi-static Faraday signal. This quantity was calculated as an average of the two lowest-frequency values of the main panel at each temperature. We note that the error bar, defined as the difference of these values, is similar to (or smaller than) the size of the data symbols.

Figure 2. (a) Lines: THz-induced Faraday rotation calculated as a difference of the signals ΔΓRHP−ΔΓLHP\Delta\Gamma_{\mathrm{RHP}}-\Delta\Gamma_{\mathrm{LHP}} as shown Figure 1 at various temperatures. A time-shift correction was applied to the raw time-domain data (as described in the text). Symbols: fits of the data by the theoretical model given by Eq. (S47) of SI (see text for details). Inset: difference traces of the raw time-domain data (i.e., without the time-shift correction). (b) Fourier transform of the time-domain traces in the main panel of (a). Inset: temperature dependence of the integral Faraday signal.

For a more complete analysis, we also measured the TKE ΔΓLP\Delta\Gamma_{\mathrm{LP}} using a linearly polarized THz pump beam; in this case, no magnetic contribution can be induced. At the same time, the magnetic contribution in the TFE measurements with a circularly polarized pump should be canceled upon summing the time traces measured with the two circular polarizations of the pump: ΔΓRHP+ΔΓLHP\Delta\Gamma_{\mathrm{RHP}}+\Delta\Gamma_{\mathrm{LHP}}. A comparison of these two types of curves for selected temperatures is shown in Figure 3. The long-lived component (previously discussed in the literature in terms of THz-induced electric polarization and correlations between polar nanoregions which develop especially at low temperatures [32, 33]) exhibits the same decay for both data sets. The component calculated from experiments with circularly polarized pump shows a sig-

Figure 3. Comparison of the sums of traces ΔΓRHP+ΔΓLHP\Delta\Gamma_{\mathrm{RHP}}+\Delta\Gamma_{\mathrm{LHP}} obtained for RHP and LHP THz pumps (solid line) with independently measured signals ΔΓLP\Delta\Gamma_{\mathrm{LP}} for a linearly polarized THz pump (dashed line). The plots are vertically shifted for clarity (thin dash-dotted lines indicate zero signal level for each part).

nificantly stronger sideband (2Ω02\Omega_{0}) component since, in principle, different terms contribute to the signal in this case.

Note that the Faraday rotation ΔΓTFE\Delta\Gamma_{\mathrm{TFE}} shows a non-monotonic temperature dependence, as inferred from Figure 2(a,Figure 2b). Its values are relatively low at 81 K and they increase with cooling down to 31 K, where the TFE intensity exhibits a maximum. Further cooling below 31 K is accompanied by a decrease in the TFE signal. This is in qualitative agreement with the theory of dynamical multiferroicity with a temperature-dependent soft optical phonon [17]. We performed a fit of the time-domain Faraday traces shown in Figure 2(a) using the formula (S47) derived in SI [27]. For the calculations, we used a circularly polarized time domain waveform with a Gaussian envelope, which matches very well the experimentally measured pump waveforms (see Fig. S5). We also used the experimentally determined temperature- and frequency-dependent linear dielectric susceptibility χ(Ω;T)\chi(\Omega;T) and the related THz refractive index N(Ω;T)N(\Omega;T) of KTaO3_3 (see Fig. S3 in [27]); these quantities enter Eq. (S47) through the functions VV and WW defined by Eqs. (S32) and (S33). The only fitting parameters were the experimental time origin and the amplitude of the trace for each temperature, D(T)D(T), introduced in Eq. (S47). The fit yielded an excellent agreement between the experimental time traces and the theoretical model, except for the measurement at 22 K [see Figure 2(a)]. We attribute this small discrepancy to the fact that at this temperature the SM frequency matches the bandwidth of the pump pulse and, due to these resonance conditions, some additional higher-order nonlinear effects can take place. The temperature dependence of the amplitude of the traces D(T)D(T) follows an unexpected trend which is discussed in detail in End Matter.

To conclude, we clearly demonstrated a phonomagnetic response of KTaO3_3 by means of experiments where the soft phonon was coherently driven by strong-field circularly polarized THz pulses and a magnetic-like response was detected via the time-resolved magneto-optical Faraday effect. A proper experimental setup and an in-depth theoretical analysis of nonlinear interactions enabled us to eliminate the influence of the electro-optic Kerr effect, which was predominant in a similar earlier experiment on SrTiO3_3 [20]. Our approach reveals a straightforward method of identifying the Faraday phonomagnetic effect, which will be very useful for further research in chiral phonons and phonomagnetism. Although our model is very appropriate for quantitatively describing the mechanism of inducing the magnetic-like response of KTaO3_3 and the shapes of the concomitant Faraday effect transiently induced by strong THz pulses, it does not explain the observed temperature dependence of the amplitude of the Faraday effect. This provides an opportunity for follow-up studies which should bring a deeper understanding of the clearly demonstrated phonomagnetic effect in diamagnetic quantum paraelectric KTaO3_3.

Acknowledgments

Acknowledgments—This work was supported by the Czech Science Foundation (Project No. 24-10791S) and by project TERAFIT - CZ.02.01.01/00/22_008/0004594 co-financed by European Union and the Czech Ministry of Education, Youth and Sports. Parts of this research were carried out at ELBE at the Helmholtz-Zentrum Dresden - Rossendorf e. V., a member of the Helmholtz Association. MB and SB acknowledge support from the Knut and Alice Wallenberg Foundation (Grant No. 2019.0068) and from the SNSF Ambizione project PZ00P2_216089.

Data availability—Data supporting the findings of this article are publicly available [34].

End Matter

Principle of detection

In Figure 4 we show the principle of detection. In the laboratory frame xyzxyz (which coincides here with the crystallographic axis frame of KTaO3_3), the incident optical probing pulse E\mathbf{E} is linearly polarized along the [110] direction, i.e., E=(E,E,0)/2\mathbf{E}=(E,E,0)/\sqrt{2}. The THz-pump-induced optical field may have components polarized both along [110] (denoted as GG in Figure 4) and along [11ˉ\bar{1}0] (denoted as FF), i.e., the total field transmitted through the sample reads Eprobe=(E−G+F,E−G−F,0)/2\mathbf{E}_{\mathrm{probe}}=(E-G+F,E-G-F,0)/\sqrt{2}, F,G≪EF,G\ll E.

Moreover, owing to the frequency mixing in the nonlinear interaction, each THz-pump induced polarization component may consist of a wave oscillating at the original probe frequency ωp\omega_p (main band) and of a wave oscillating at shifted frequencies ωp±2Ω0\omega_p\pm2\Omega_0 (sidebands), where Ω0≈0.7\Omega_0\approx0.7 THz is the central frequency of the THz pump pulse. These waves will be denoted as F0F_0, G0G_0 (main band) and F2F_2, G2G_2 (sidebands). The detection part consists of a Wollaston polarizer, which separates the probe beam into two parts, one with horizontal (Eprobe,xE_{\mathrm{probe},x}) and the other with vertical (Eprobe,yE_{\mathrm{probe},y}) polarization, and of a pair of balanced photodiodes, which deliver the difference signal ΔΓ\Delta\Gamma between the two beams:

ΔΓ=Γx−Γy∝∣Eprobe,x∣2−∣Eprobe,y∣2=2ℜ(E∗F)−2ℜ(G∗F).(2)\Delta\Gamma=\Gamma_x-\Gamma_y\propto\left|E_{\mathrm{probe},x}\right|^2-\left|E_{\mathrm{probe},y}\right|^2=2\Re(E^{*}F)-2\Re(G^{*}F). \tag*{(2)}

Note that ℜ(A∗B)=0\Re(A^{*}B)=0 if AA and BB are mutually π/2\pi/2 shifted. TKE and TFE have different origins and therefore the polarization state of the output probe beam is different in the two cases. In SI [27] we solve the appropriate wave equation for both linearly and circularly polarized THz pump pulses. In principle, TKE leads to an elliptical polarization of the output probe beam, which means that the component FF is phase shifted with respect to EE. We further show that in the case of a circularly polarized pump (a) F0=0F_0=0 [see Eq. (S14)], (b) the phase shift of F2F_2 and G0G_0 is approximately equal to π/2\pi/2, and (c) the phase shift is zero for G2G_2 (see also Table I for a summary of these properties). In this case, the first order signal ℜ(E∗F)\Re(E^{*}F) vanishes and only the second order contribution, ℜ(G0∗F2)\Re(G_0^{*}F_2), can be detected (this signal is caused by a very small tilt of the principal axes of the polarization ellipse with respect to [110] direction owing to the π/2\pi/2-shifted GG-component). Note that this term contributes merely to the sidebands. The main band component induced by the Kerr effect might appear in the measured signal only due to imperfect polarization states of the pump and probe beams, due to the finite bandwidth of the probe pulse (i.e., finite temporal pulse length), or due to a very weak non-phase-matched term (see SI for further details). In the case of TFE, F0F_0 oscillates in phase with EE and F2=0F_2=0, G=0G=0 (see SI for the derivation); the measured signal consists of the main band only, and it is proportional to ℜ(E∗F0)\Re(E^{*}F_0), see also Table I. In this way our experimental scheme leads to a linear Faraday signal and it rejects the Kerr component as much as possible.

Next we calculate the difference between the signals for RHP and LHP pump. The rejection of the Kerr contributions is then inherent due to the symmetry of the RHP and LHP induced signals as shown in the last column of Table 1. It follows that for TKE ΔΓRHP−ΔΓLHP=0\Delta\Gamma_{\mathrm{RHP}}-\Delta\Gamma_{\mathrm{LHP}}=0, while for TFE ΔΓRHP−ΔΓLHP=2ΔΓTFE\Delta\Gamma_{\mathrm{RHP}}-\Delta\Gamma_{\mathrm{LHP}}=2\Delta\Gamma_{\mathrm{TFE}}.

ComponentMain bandSidebandsRHP vs. LHP
TKE – linear pump
FFiF0≠0iF_0\ne0iF2≠0iF_2\ne0NA
GGiG0≠0iG_0\ne0iG2≠0iG_2\ne0NA
TKE – circular pump
FFF0=0F_0=0iF2≠0iF_2\ne0F2,RHP=F2,LHPF_{2,\mathrm{RHP}}=F_{2,\mathrm{LHP}}
GGiG0≠0iG_0\ne0G2≠0G_2\ne0G0,RHP=G0,LHPG_{0,\mathrm{RHP}}=G_{0,\mathrm{LHP}}
TFE – circular pump
FFF0≠0F_0\ne0F2=0F_2=0F0,RHP=−F0,LHPF_{0,\mathrm{RHP}}=-F_{0,\mathrm{LHP}}
GGG0=0G_0=0G2=0G_2=0NA

Table 1. Properties of THz-pump-induced components (see Figure 4) for various interactions. The imaginary unit denotes an approximate π/2\pi/2 phase shift with respect to the incident probe field; its absence represents zero phase shift. All the theoretical details are explained in SI [27].

Note that if the probe beam impinges on the sample linearly polarized along a crystallographic axis of KTaO3_3 e.g. along [10], and the half-wave plate is positioned only after the sample, which was the case in [20], the main difference compared to our current case shown in Figure 4 lies in the properties of TKE with a circular pump. Indeed, as shown in SI, in this case, F0≠0F_0 \ne0, see Eq. (S12) in [27], which leads to a more intense residual pollution of the Faraday signal by the main band of TKE. Moreover, another consequence of this experimental setup is that for the circular pump, F2F_2 and G2G_2 will feature a general phase shift with respect to the unperturbed probe wave; this phase shift will depend on the magnitudes of various nonlinear susceptibility coefficients.

Origin of the D(T)D(T) dependence

The temperature dependence of the amplitude of the traces D(T)D(T) is shown in Figure 5. We emphasize that the temperature dependence of the Faraday effect due to the temperature evolution of the KTaO3_3 susceptibility is included in our model [see Eq. (S47) in Ref. 27], and if the susceptibility were the only source of the temperature variation of the Faraday signal, D(T)D(T) would be constant. Clearly, the observed shape of D(T)D(T) exceeds the scope of our model. Roughly speaking, as the temperature drops from 80 to 11 K, the amplitude of the observed signal becomes eight times stronger than predicted, e.g., by the simplified expression in Eq. (S48). Interestingly, this dependence is similar to the temperature variation of the static linear electric susceptibility in KTaO3_3 which also increases monotonically towards 0 K (see Figure 5). Below we discuss the possible origins of the observed shape of the D(T)D(T) dependence (see Figure 5).

Figure 5. Comparison of temperature dependencies of the linear dielectric susceptibility value of our KTaO3_3 crystal measured at 1 MHz (blue line, left scale) and of the amplitude DD of the fit of the Faraday rotation.

(i) It is unlikely that it comes from the temperature dependence of the parameter gg in (S47) which is proportional to the Verdet constant. As a rule, in diamagnets, the temperature dependence of the Verdet constant is very weak, as it is mostly due to the temperature dependence of the optical refractive index [35].

(ii) We theoretically estimated the temperature dependence of the gyromagnetic ratio (parameter γ\gamma), see the description in SI [27] and Fig. S10 therein. We found that γ\gamma can vary at most by 5% in the temperature range between 0 and 100 K, far less than in the observed temperature dependence of the amplitude D(T)D(T).

(iii) It has been suggested in recent papers [21, 31] that an orders-of-magnitude enhancement of the phonomagnetic effect could be expected if an additional related contribution by free electrons is taken into account. However, it appears that, in order to induce an appreciable effect, the concentration of free electrons should be sufficiently high to shift the Fermi level very close to (or even above) the conduction band edge (see Fig. 4 in [21]). In principle, conduction band electrons may be present in the system due to ionized oxygen vacancies or due to multiphoton ionization by strong-field THz pulses. In our case, the steady-state conductivity of our insulating KTaO3_3 crystal is negligible and so must be the concentration of electrons. Dynamical generation of conduction band electrons by strong pulses at 0.7 THz would require at least 30-photon absorption to overcome the energy difference of 0.1–0.2 eV for oxygen vacancies levels. So this mechanism seems also very unlikely.

(iv) One can consider a contribution of the nonlinear polarization component in the formula (S21). In the case we take into account the third-order (Kerr-like) polarization PNL(3)\mathbf{P}_{\mathrm{NL}}^{(3)}, leading to an additional phonomagnetic contribution, such as B(3)=γP˙×PNL(3)\mathbf{B}^{(3)}=\gamma\dot{\mathbf{P}}\times\mathbf{P}_{\mathrm{NL}}^{(3)}, and we carry out the calculation of the Faraday signal, we find again a vanishing 2Ω02\Omega_0 component and a non-zero Ω≈0\Omega\approx0 component featuring an additional susceptibility term, see Eq. (S50), as required by the temperature dependence shown in Figure 5. However, since this nonlinear contribution to the phonomagnetic field is additive with the linear one, it would have to dominate in order to imprint the required temperature dependence to the measured signal (cf. Fig. S6). At the same time, the third order polarization would also lead to a 4Ω04\Omega_0 component, which was not observed.

(v) Recent papers by Merlin [24, 25] suggest that, by symmetry, phenomena analogous to the phonomagnetic effect discussed here may arise from a term T×T∗T\times T^*, which has apparently the same symmetry as the polarization-induced magnetic field given by (S21), and which is non-zero for the circular THz polarization. However, the magnitude and temperature behavior of the related coupling constant is completely unknown.

References

  1. [1]E. Beaurepaire, J.-C. Merle, A. Daunois, and J.-Y. Bigot, Phys. Rev. Lett. 76, 4250 (1996).
  2. [2]A. Kimel, A. Kirilyuk, P. Usachev, R. Pisarev, A. Balbashov, and T. Rasing, Nature 435, 655 (2005).
  3. [3]M. Först, C. Manzoni, S. Kaiser, Y. Tomioka, Y. Tokura, R. Merlin, and A. Cavalleri, Nature Physics 7, 854 (2011).
  4. [4]M. Först, R. Tobey, S. Wall, H. Bromberger, V. Khanna, A. L. Cavalleri, Y.-D. Chuang, W. Lee, R. Moore, W. Schlotter, et al., Phys. Rev. B 84, 241104 (2011).
  5. [5]M. Fechner, A. Sukhov, L. Chotorlishvili, C. Kenel, J. Berakdar, and N. Spaldin, Phys. Rev. Mater. 2, 064401 (2018).
  6. [6]M. Gu and J. Rondinelli, Phys. Rev. B 98, 024102 (2018).
  7. [7]A. Disa, M. Fechner, T. Nova, B. Liu, M. Först, D. Prabhakaran, P. Radaelli, and A. Cavalleri, Nature Physics 16, 937 (2020).
  8. [8]D. Afanasiev, J. Hortensius, B. Ivanov, A. Sasani, E. Bousquet, Y. Blanter, R. Mikhaylovskiy, A. Kimel, and A. Cavignia, Nature Materials 20, 607 (2021).
  9. [9]A. Stupakiewicz, C. Davies, K. Szerenos, D. Afanasiev, K. Rabinovich, A. Boris, A. Caviglia, A. Kimel, and A. Kirilyuk, Nature Physics 17, 489 (2021).
  10. [10]A. Disa, J. Curtis, M. Fechner, A. Liu, A. Von Hoegen, M. Först, T. Nova, P. Narang, A. Maljuk, A. Boris, et al., Nature 617, 73 (2023).
  11. [11]M. Rini, R. Tobey, N. Dean, J. Itatani, Y. Tomioka, Y. Tokura, R. Schoenlein, and A. Cavalleri, Nature 449, 72 (2007).
  12. [12]D. Fausti, R. Tobey, N. Dean, S. Kaiser, A. Dienst, M. C. Hoffmann, S. Pyon, T. Takayama, H. Takagi, and A. Cavalleri, Science 331, 189 (2011).
  13. [13]D. Nicoletti and A. Cavalleri, Adv. Optics and Photonics 8, 401 (2016).
  14. [14]M. Mitrano, A. Cantaluppi, D. Nicoletti, S. Kaiser, A. Perucchi, S. Lupi, P. Di Pietro, D. Pontiroli, M. Riccò, S. Clark, et al., Nature 530, 461 (2016).
  15. [15]E. Rowe, B. Yuan, M. Buzzi, G. Jotzu, Y. Zhu, M. Fechner, M. Först, B. Liu, D. Pontiroli, M. Riccò, et al., Nature Physics 19, 1821 (2023).
  16. [16]T. Nova, A. Cartella, A. Cantaluppi, M. Först, D. Bossini, R. V. Mikhaylovskiy, A. Kimel, R. Merlin, and A. Cavalleri, Nature Physics 13, 132 (2017).
  17. [17]D. Juraschek, M. Fechner, A. Balatsky, and N. Spaldin, Phys. Rev. Mater. 1, 014401 (2017).
  18. [18]D. Juraschek and N. Spaldin, Phys. Rev. Mater. 3, 064405 (2019).
  19. [19]K. Dunnett, J.-X. Zhu, N. Spaldin, V. Juričić, and A. Balatsky, Phys. Rev. Lett. 122, 057208 (2019).
  20. [20]M. Basini, M. Pancaldi, B. Wehinger, M. Udina, V. Unikandunnanni, T. Tadano, M. Hoffmann, A. Balatsky, and S. Bonetti, Nature 628, 534 (2024).
  21. [21]R. Geilhufe and W. Hergert, Phys. Rev. B 107, L020406 (2023).
  22. [22]N. Shabala and R. M. Geilhufe, Phys. Rev. Lett. 133, 266702 (2024).
  23. [23]N. Sellati, J. Fiore, and L. Benfatto, https://arxiv.org/abs/2605.27127.
  24. [24]R. Merlin, Phys. Rev. B 110, 094312 (2024).
  25. [25]R. Merlin, PNAS nexus 4, pga f002 (2025).
  26. [26]M. Basini, M. Udina, M. Pancaldi, V. Unikandunnanni, S. Bonetti, and L. Benfatto, Phys. Rev. B 109, 024309 (2024).
  27. [27]See https://link.aps.org/supplemental/... for supplemental material.
  28. [28]H. Vogt, Phys. Rev. B 51, 8046 (1995).
  29. [29]B. Green, S. Kovalev, V. Asgekar, G. Geloni, U. Lehnert, T. Golz, M. Kuntszch, C. Bauer, J. Hauser, et al., Scientific Reports 6, 22256 (2016).
  30. [30]A. Pashkin, J.-C. Deinert, et al., The European Physical Journal Plus 138, 158 (2023).
  31. [31]R. M. Geilhufe, V. Juričić, S. Bonetti, J.-X. Zhu, and A. Balatsky, Phys. Rev. Research 3, L022011 (2021).
  32. [32]X. Li, P. Peng, H. Dammak, G. Geneste, A. Akbarzadeh, S. Prosandeev, L. Bellaiche, and D. Talbayev, Phys. Rev. B 107, 064306 (2023).
  33. [33]B. Cheng, P. Kramer, Z.-X. Shen, and M. Hoffmann, Phys. Rev. Lett. 130, 126902 (2023).
  34. [34]Dataset on Zenodo, link to be provided.
  35. [35]P. A. Williams, A. H. Rose, G. W. Day, T. E. Milner, and M. N. Deeter, Appl. Opt. 30, 1176 (1991).

Paper details

vlnc:W_taw6kggh
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

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