This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Extreme Terahertz Magnon Multiplication Induced by Resonant Magnetic Pulse Pairs

C. Huang1,2,∗, L. Luo1,∗, M. Mootz1, J. Shang3,4, P. Man3,4,
L. Su3,4, I. E. Perakis5, Y. X. Yao 1, A. Wu3,4 and J. Wang1,2,†
1Ames National Laboratory, Ames, IA 50011 USA
2Department of Physics and Astronomy, Iowa State University, Ames, IA 50011 USA
3State Key Laboratory of High Performance Ceramics and Superfine Microstructure,
Shanghai Institute of Ceramics, Chinese Academy of Sciences, Shanghai 201899, China
4Center of Materials Science and Optoelectronics Engineering,
University of Chinese Academy of Sciences, Beijing 100049, China
5Department of Physics, University of Alabama at Birmingham,
Birmingham, AL 35294-1170, USA Equal contribution. To whom correspondance shoule be addressed: [email protected], [email protected].

Abstract

Nonlinear interactions of spin-waves and their quanta, magnons, have emerged as prominent candidates for interference-based technology, ranging from quantum transduction to antiferromagnetic spintronics. Yet magnon multiplication in the terahertz (THz) spectral region represents a major challenge. Intense, resonant magnetic fields from THz pulse-pairs with controllable phases and amplitudes enable high order THz magnon multiplication, distinct from non-resonant nonlinearities such as the high harmonic generation by below-band gap electric fields. Here, we demonstrate exceptionally high-order THz nonlinear magnonics. It manifests as 7th{}^{\text{th}}-order spin-wave-mixing and 6th{}^{\text{th}} harmonic magnon generation in an antiferromagnetic orthoferrite. We use THz multi-dimensional coherent spectroscopy to achieve high-sensitivity detection of nonlinear magnon interactions up to six-magnon quanta in strongly-driven many-magnon correlated states. The high-order magnon multiplication, supported by classical and quantum spin simulations, elucidates the significance of four-fold magnetic anisotropy and Dzyaloshinskii-Moriya symmetry breaking. Moreover, our results shed light on the potential quantum fluctuation properties inherent in nonlinear magnons.

Introduction

Recently, there is growing evidence that non-thermal control of magnetism by light is feasible [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]. The THz frequency magnons [18, 19, 20, 21, 22, 23, 24] is central for emergent spin-wave interference-based devices [25] at least 1,000 times faster than the magneto-optical recording and device technologies that use thermal-magnetic excitations. Furthermore, THz control of magnetism is cross-cutting for coherent magnonics [26, 10, 11, 12, 27, 28], quantum magnetism [29, 1, 9], antiferromagnetic (AFM) spintronics [26, 2, 3], and quantum computing [30].

Microwave spin-wave excitation has been shown to induce megahertz frequency magnon multiplication [31]. However, THz frequency magnon multiplication remains an as yet unobserved nonlinear process, in which multiple THz magnons can fuse together to form a high-order magnon coherence. Recent experiments have applied THz frequency, multi-dimensional coherent nonlinear spectroscopy (THz-MDCS) to quantum systems, such as low-dimensional semiconductors [33, 32] and superconductors [35, 34, 36]. THz-MDCS study of magnetic materials have also yielded interesting results such as revealing magnon second harmonic generation [37] , upconversion [38] and mixing [39]. Thus far the results obtained are explained by calculating low-order spin susceptibilities. We place emphasis on the higher-order nonlinearity and collectivity of THz magnons, supported by classical and quantum spin simulations.

To explore magnon multiplication and control, an AFM orthoferrite is driven and probed by THz-MDCS. We use the magnetic fields 𝐁(t)\mathbf{B}(t) of intense THz pulse-pair to resonantly induce collective spin dynamics and correlation, as illustrated in Fig. 1a. The excitation by such pulse-pair enables us to measure the phase of the AFM magnonic coherent nonlinear emission, in addition to the amplitude as in regular pump-probe (PP) experiments.This approach opens the door to the possibility of driving high-order nonlinear wave-mixing, which is controlled by the interactions of multiple, long-lasting magnon excitations persisting well beyond the duration of the initial laser pulses. We emphasize three additional points here. First, by leveraging both the real-time and relative phase dependencies of coherent emission, we can effectively distinguish, in frequency space after Fourier Transform (FT), discrete spectral peaks resulting from magnon multiplications of varying orders. Second, it is important to distinguish the magnon multiplication from the conventional high harmonic generation of solids [40]. Magnon multiplication utilizes resonant excitations of collective modes leading to high harmonics that emerge after the pulses. In contrast, the conventional electric field-induced high harmonics utilize photoexcitation occurring in the transparency region which exists only during the period of laser excitation. Third, resonant THz photoexcitations have proven to be highly effective in sustaining long-lasting coherence of low-energy collective modes in quantum materials. This is evidenced by recent studies involving topological materials [41, 42], semiconductors [43] and superconductors [36].

In this Letter, we demonstrate extremely high-order nonlinear magnonics by performing THz-MDCS measurements of Sm0.4Er0.6FeO3 using an intense, resonant THz laser pulse-pair. The sensitive detection of coherent spectra enables us to observe up to sixth harmonic generation (6HG) and seventh-order spin wave-mixing (7WM) signals, which significantly exceed what has been previously reported. These distinct magnon multiplication peaks demonstrate a “super” resolution tomography of long-lived interacting magnon quantum coherences and distinguish between different many-body spin correlation functions that describe strongly-driven magnon states. By simulating the coherent dynamics generated by large pulse-pair-driven spin deviations from the equilibrium magnetization orientation, we reproduce the experimental THz-MDCS spectra and identify multi-magnon interaction originating from the dynamical interplay of four-fold magnetic anisotropy and Dzyaloshinskii-Moriya (DM) symmetry breaking.

Results

Figure 1a illustrates our THz-MDCS measurement of quasi-AFM magnons in aa-cut rare-earth orthoferrite sample Sm0.4Er0.6FeO3 (Methods) at room temperature. This sample exhibits canted AFM order in equilibrium (red arrows), with a small net magnetization pointing along the cc-axis (blue arrow). The measured linear THz response clearly shows a quasi-AFM magnon mode with energy at ωAF\omega_{\text{AF}}\sim 2.5 meV (Fig. 1b). In our THz-MDCS experiment, two collinear THz pulses E1E_{1} and E2E_{2} (red lines) with a time delay τ\tau (red arrow) are used, whose magnetic fields are polarized parallel to the sample cc-axis. The transmitted nonlinear emission excited from such intense THz pulse-pair at a fixed τ\tau is detected as the function of the gate time tt through electro-optical sampling.

Figure 1c presents representative time-domain traces of the two THz pulses, E1E_{1} and E2E_{2} (red and blue solid lines), with electric field strength up to 47.5 MV/m each. We clearly observe the THz light-driven single-order, long-lasting magnon quantum coherence. We expect that magnon-phonon interactions mainly drive the oscillation decay. Fig. 1d shows the coherent emission trace when both pulses are present, E12E_{12} (light blue line), and the extracted nonlinear signal, ENL(t,τ)=E12(t,τ)E1(t,τ)E2(t)E_{\mathrm{NL}}(t,\tau)=E_{12}(t,\tau)-E_{1}(t,\tau)-E_{2}(t) (black line), recorded as the function of gate time tt for various inter-pulse delay times τ\tau. The exemplary ENL(t,τ)E_{\mathrm{NL}}(t,\tau) at the fixed τ=4.8\tau=4.8 ps in Fig. 1d is generated by the constructive interference of THz-excited magnon with energy \sim 2.5 meV measured in the static THz spectrum (Fig. 1b). This result illustrates the simultaneous amplitude- and phase-resolved detection of the coherent nonlinear response. The subtraction of the single pulse signals E1E_{1} and E2E_{2} from the total signal E12E_{12} allows an extraction of pure coherent nonlinear emissions arising from magnonic nonlinear interactions. The measured long-time dynamics of ENL(t,τ)E_{\mathrm{NL}}(t,\tau) exhibits pronounced, anharmonic oscillations lasting for a long time, as seen within the temporal range marked by a double-arrowed solid line in Fig. 1d. This observation indicates that nonlinear effects of strongly-driven AFM magnons last well beyond the pulse duration. Fig. 1e shows a spectrum of ENL(t,τ)E_{\mathrm{NL}}(t,\tau) obtained over the long oscillation period, which clearly demonstrates the presence of high-order magnonic nonlinearities. This spectrum shows sharp magnon multiplication peaks at frequencies up to six times the magnon energy (Fig. 1b), as marked by the dashed lines. They have been absent in prior THz-MDCS studies of ErFeO3 [37, 38]. The harmonic peaks are robust from interactions among multiple long-lasting magnon quantum excitations, rather than laser fields. They appear for magnetic fields parallel to the cc-axis (black line) and are absent for perpendicular to the cc-axis (blue line).

Figures 2a and 2c show two-dimensional (2D) temporal and spectral profiles of the experimentally measured coherent nonlinear signals ENL(t,τ)E_{\mathrm{NL}}(t,\tau). For comparison, Figs. 2b and 2d present the corresponding results of our numerical simulations, to be discussed below. The oscillations along the horizontal gate time tt-axis represent the time evolution of magnon interaction for different pulse-pair time delays τ\tau. These gate time tt-oscillations are pronounced even when two incident pulses do not overlap in time, e.g., for a long inter-pulse delay τ=12\tau=12 ps. The observation of strong tt-oscillations at such large τ\tau indicates that light-induced coherence is stored in the magnon system well after the laser pulse. The oscillatory time-dependent nonlinear signal along the τ\tau-axis for a fixed tt displays coherent enhancement due to constructive interference when τ\tau corresponds to in-phase magnon excitation, while the signal diminishes due to destructive interference when τ\tau corresponds to out-of-phase magnon excitation. Fig. 2c shows the THz-MDCS spectra obtained by 2D-FT of ENL(t,τ)E_{\mathrm{NL}}(t,\tau) with respect to gate time tt (frequency ωt\omega_{t}) and inter-pulse delay τ\tau (frequency ωτ\omega_{\tau}), where a series of discrete spectral peaks can be observed.

Next, we will first discuss the peaks originating from conventional magnon rectification (MR), pump-probe (PP), and phase-reversal (R) or non-reversal (NR), followed by the newly discovered nonlinear magnonic wave-mixing and harmonic generation processes produced by high-order magnetic interactions as summarized in Table 3 (Methods). Particularly, Fig. 2c clearly shows that strong THz field driving leads to the formation of high-order nonlinear peaks never seen before, including third-harmonic (3HG), fourth-harmonic (4HG), fifth-harmonic (5HG), and sixth-harmonic generation (6HG) as well as five-wave (5WM), six-wave (6WM), and seven-wave mixing (7WM) peaks, as discussed below.

The different THz-MDCS peaks correspond to different nonlinear processes, which can be observed simultaneously and identified according to their locations in (ωt,ωτ)(\omega_{t},\omega_{\tau}) space. To analyze the measured 2D spectrum, we introduce frequency vectors for incident pulse 1 and 2 excitations, ω1=(ωAF,ωAF)\omega_{1}=(\omega_{\mathrm{AF}},\omega_{\mathrm{AF}}) and ω2=(ωAF,0)\omega_{2}=(\omega_{\mathrm{AF}},0) (red and black arrows in Fig. 2d), where ωAF\omega_{\mathrm{AF}}\sim 2.5 meV denotes to the AFM magnon mode energy. The several discrete peaks separated along the vertical axis (relative phase), as shown in Fig. 2c, demonstrate that the pulse-pair time delay τ\tau controls phase coherence between magnons excited by two incident pulses. Figs. 2e and 2f further show a zoom-in view of some high harmonic and wave mixing peaks observed in the experiment, while Figs. 2g and 2h show the corresponding peaks reproduced by the theory.

In order to compare the measured and simulated nonlinear processes, we first identify the physical origins of the series of discrete peaks seen in Fig. 2c along the vertical direction ωt=ωAF\omega_{t}=\omega_{\mathrm{AF}}, located at ωτ=ωAF\omega_{\tau}=-\omega_{\mathrm{AF}}, 0, ωAF\omega_{\mathrm{AF}}, and 2ωAF2\omega_{\mathrm{AF}}. The (ωAF,0)(\omega_{\mathrm{AF}},0) peak arises from the PP process (Table 3) that reflects the dynamics of magnon coherent populations. In particular, two field interactions during pulse 1 create a magnon population via the difference frequency process ω1ω1\omega_{1}-\omega_{1}. This population evolves during time τ\tau and then interacts with a magnon excitation during pulse 2, resulting in a single magnon 1-quantum coherence (1QC). The peak at (ωAF,2ωAF)(\omega_{\mathrm{AF}},2\omega_{\mathrm{AF}}) shows the dynamics of a magnon 2-quantum coherence (2QC), which is generated by two magnon excitations during pulse 1 via the sum–frequency process ω1+ω1\omega_{1}+\omega_{1}. This 2-magnon coherence is probed through its interaction with a magnon excitation during pulse 2.

The two peaks located at (ωAF,ωAF)(\omega_{\mathrm{AF}},\omega_{\mathrm{AF}}) and (ωAF,ωAF)(\omega_{\mathrm{AF}},-\omega_{\mathrm{AF}}) arise from the interaction of a 1QC generated by pulse 1 and a 1QC generated by pulse 2. Their interaction creates a magnon population via the difference frequency processes ±(ω1ω2)\pm(\omega_{1}-\omega_{2}), whose interaction with a magnon excitation during pulse 2 leads to a 1QC which is either phase reversed (R) or not reversed (NR) with respect to the 1QC of pulse 1. We also observe peaks along ωt=3ωAF\omega_{t}=3\omega_{\mathrm{AF}} that correspond to third harmonic generation (THG) signals. Here, the interaction of a 2QC created by pulse 1(2) with a 1QC generated by pulse 2(1) produces a THG signal at (3ωAF,2ωAF)(3\omega_{\mathrm{AF}},2\omega_{\mathrm{AF}}) and (3ωAF,ωAF)(3\omega_{\mathrm{AF}},\omega_{\mathrm{AF}}).

The measured THz-MDCS 2D spectra allow us to identify the role of the antisymmetric exchange DM interaction in the studied AFM system. This is possible through peaks generated by second-order magnon nonlinear processes facilitated by the symmetry-breaking DM interaction. The latter interaction is already known to induce a canted AFM order in the ground state. Such spin canting with respect to the AFM orientation of a spin-up and a spin-down sub-lattice results in a net magnetization along the cc-axis (blue arrow, Fig. 1a). The coherent spin dynamics along this axis then becomes anharmonic, which results in the second-order nonlinear peaks observed in the 2D spectra at ωt=2ωAF\omega_{t}=2\omega_{\mathrm{AF}} and at ωt=0\omega_{t}=0 (Fig. 2c) that involve two magnon excitations. The difference frequency processes ω1ω2\omega_{1}-\omega_{2} involving two magnon 1QC generate magnon rectification (MR) signals at (0,±ωAF)(0,\pm\omega_{\mathrm{AF}}) via the DM interaction. The latter antisymmetric exchange also generates second harmonic generation (SHG) peaks at (2ωAF,ωAF)(2\omega_{\mathrm{AF}},\omega_{\mathrm{AF}}) and (2ωAF,2ωAF)(2\omega_{\mathrm{AF}},2\omega_{\mathrm{AF}}). These SHG peaks arise from the sum-frequency generation processes ω1+ω2\omega_{1}+\omega_{2} and ω1+ω1\omega_{1}+\omega_{1}, respectively.

Finally, the THz-MDCS spectra at Figs. 2c and 2e reveal high-order nonlinear peaks involving up to six magnon quanta. In particular, we observe strong 4HG, 5HG, and 6HG peaks, and also strong 5WM, 6WM, and 7WM nonlinear signals, shown in Fig 2f. It is important to note that the 2D frequency resolution enabled by the THz-MDCS allows us to not only excite coherent magnons but also achieve a “super” resolution tomography different from the conventional high harmonic generation spectroscopies that only detect one dimensional spectra. In the 2D frequency plane, the same, high-order magnon harmonics will appear as different correlated THz peaks along multiple axes, e.g., 6HG and 5WM peaks separated in 2D space. Such strong correlation enables the deterministic assignment of higher–order nonlinear peaks, even in the presence of measurement noise or FT artifacts, which were very difficult to discern before. As discussed below, our numerical simulations shown in Figs. 2d, 2g and 2h further corroborate our discovery of the higher-order magnon multiplication peaks. We attribute these high-order nonlinear signals to the existence of both four-fold magnetic anisotropy and DM interaction in our AFM system. The origin of all observed THz-MDCS peaks is summarized in Table 3 (Methods). Note that additional experimental signals at frequencies less than ωAF\omega_{\mathrm{AF}} are not captured by our mean field model. While we do not discuss such signals here, they may arise from parametric processes involving two magnons with finite momenta 𝐪\mathbf{q} and 𝐪-\mathbf{q} such that ωAF=ω𝐪+ω𝐪\omega_{\mathrm{AF}}=\omega_{\mathbf{q}}+\omega_{-\mathbf{q}}[44].

To corroborate the above interpretation of experimental results, we use the following two sub-lattice Hamiltonian [45, 46]:

H=J𝐒1𝐒2𝐃(𝐒1×𝐒2)i=1,2(KaSi,a2+KcSi,c2)K4i=1,2j=a,b,cSi,j4γ𝐁(t)i=1,2𝐒i.\displaystyle H=J\,\mathbf{S}_{1}\cdot\mathbf{S}_{2}-\mathbf{D}\cdot\left(\mathbf{S}_{1}\times\mathbf{S}_{2}\right)-\sum_{i=1,2}\left(K_{a}\,S_{i,a}^{2}+K_{c}\,S_{i,c}^{2}\right)-K_{4}\sum_{\begin{subarray}{c}i=1,2\\ j=a,b,c\end{subarray}}S^{4}_{i,j}-\gamma\,\mathbf{B}(t)\cdot\sum_{i=1,2}\mathbf{S}_{i}\,. (1)

This classical spin model effectively describes the AFM resonance mode and THz magnon multiplication behavior in orthoferrite systems where the DM interaction is the predominant canting mechanism [47]. The first term of Eq. (1) accounts for the AFM coupling between the neighboring spins 𝐒1\mathbf{S}_{1} and 𝐒2\mathbf{S}_{2} with exchange constant J>0J>0. The second term describes the Dzyaloshinskii-Moriya (DM) interaction, with anti-symmetric exchange parameter 𝐃\mathbf{D} aligned along the bb-axis. The orthorhombic crystalline anisotropy is described by the third and fourth terms of the Hamiltonian, with anisotropy constants KaK_{a} and KcK_{c} along the aa and cc axes, respectively. To explain the observed high-order nonlinearities, we include the four-fold anisotropy K4K_{4}. Finally, the driving of the spin system by the laser magnetic field 𝐁(t)\mathbf{B}(t) is described by the Zeeman interaction (the last term of the above Hamiltonian), where γ=gμ𝐁/\gamma=g\mu_{\mathbf{B}}/\hbar is the gyromagnetic ratio, g=2g=2 is the Landé gg-factor, and μB\mu_{\mathrm{B}} denotes the Bohr magneton.

We have simulated the nonlinear spin dynamics by solving the full Landau-Lifshitz-Gilbert equations [2, 37] (LLG, Eq. (2) in Methods) derived from the Hamiltonian Eq. (1). Non-interacting magnon excitations can be described by linearizing this LLG equation in the case of small amplitude oscillations around the equilibrium magnetization. This approximation is valid for weak 𝐁(t)\mathbf{B}(t). However, the full nonlinearities described by the LLG equation, central in many spin-wave device concepts, become critical for strong fields, which drive large amplitude of magnon oscillations. Here we characterize such magnon nonlinearities by solving the full nonlinear LLG equation (Eq. (2), Methods) for two magnetic fields polarized along sample cc-axis. We obtain THz-MDCS spectra by doing 2D-FT of the nonlinear magnetization 𝐌NL\mathbf{M}_{\mathrm{NL}}, which is obtained by subtracting the magnetizations induced by two the individual pulses from the nonlinear response of the net magnetization |𝐌||\mathbf{M}| under intense pulse-pair excitation (Methods). Figs. 2b and 2d present the 2D temporal profile 𝐌NL(t,τ)\mathbf{M}_{\mathrm{NL}}(t,\tau) and spectrum 𝐌NL(ωt,ωτ)\mathbf{M}_{\mathrm{NL}}(\omega_{t},\omega_{\tau}), respectively. The simulated 𝐌NL(ωt,ωτ)\mathbf{M}_{\mathrm{NL}}(\omega_{t},\omega_{\tau}) is fully consistent with the measured spectrum in Fig. 2c. This overall agreement of the two-time/two-frequency dependencies between theory and experiment allows us to assign the different discrete peaks observed in the experiment to specific nonlinear processes and magnetic interactions. The agreement also corroborates the magnonic origin of the observed high-order nonlinearities and symmetry-breaking peaks. Note that the remaining discrepancy with experimental results underscores the significance of considering the quantum properties of the spin in describing THz-MDCS experiments discussed in more detail below.

Figures 3a - 3f show the field dependence of the nonlinear spectral signals ENL(ωt,τ)E_{\mathrm{NL}}(\omega_{t},\tau) at magnonic multiplication frequencies ωt\omega_{t}= 0, 1, 2, 3, 4, 5ωAF\omega_{\mathrm{AF}}. The ENL(ωt,τ)E_{\mathrm{NL}}(\omega_{t},\tau) traces are taken at τ\tau = 4.8 ps to maximize the sensitivity similar to Fig. 1e. These high-order magnon multiplication peaks clearly grow nonlinearly with increasing field strength and follow power-law dependencies. By fitting (solid lines) the 5ωAF\omega_{\mathrm{AF}} peak amplitude in Fig. 3a, we see that is clearly scales as E5E^{5}. Note that 0ωAF\omega_{\mathrm{AF}} and 1ωAF\omega_{\mathrm{AF}} peaks scale as E2E^{2} and E3E^{3}, respectively, confirming that they originate from second-order and third-order nonlinear processes as discussed above.

Figures 3g and 3h compare a pulse-pair and single pulse pumping for generating high-order magnon multiplication peaks at 2, 3, 4ωAF\omega_{\mathrm{AF}} (dashed vertical lines). The extracted ENL(ωt,τ)E_{\mathrm{NL}}(\omega_{t},\tau) spectra at fixed τ=\tau= 4.8 ps in Fig. 3g is presented for 100% (purple line), 70% (green line), and 50% (orange line) of the maximum THz field strength and grow nonlinearly with increasing field. Compared to the ENL(ωt,τ)E_{\mathrm{NL}}(\omega_{t},\tau) spectra, the single–pulse signal, E2(ωt)E_{2}(\omega_{t}) (Fig. 3h), does not show any significant high-order magnon multiplication peaks. Magnonic nonlinearity and interaction effects are hidden in the response to a single strong pulse, but become prominent in the THz-MDCS. In the latter case, the single-pulse responses are subtracted out, so the pure nonlinear signals are determined by correlations between two pulse excitations, rather than by single pulse nonlinearities. As a result, THz-MDCS allows us to resolve the nonlinear interactions between coherent magnons excited by different pulses and associate them with different peaks in 2D frequency space.

Discussion

Figure 4 presents our numerical calculations in more detail. Particularly, it provides direct insight into the physical origin of the different THz-MDCS spectral peaks by comparing the contributions from different magnetic interactions and anisotropies. We compare the spectra 𝐌NL(ωt,ωτ)\mathbf{M}_{\mathrm{NL}}(\omega_{t},\omega_{\tau}) calculated in an electric field strength of 45\sim 45 MV/m between (i) the full-term calculation (Fig. 4a), (ii) a calculation without DM interaction, D=0D=0 , where the ground state has no spin canting (Fig. 4b), and (iii) a calculation without four-fold magnetic anisotropy, i. e., by setting K4=0K_{4}=0 in the Hamiltonian (1) (Fig. 4c). All even-order wave-mixing signals vanish when the DM interaction is switched off, consistent with earlier works [37]. In addition, the third and higher odd-order nonlinear magnon signals are strongly suppressed (Fig. 4b). The third- and higher-order nonlinear magnon signals are also strongly suppressed if K4=0K_{4}=0 (Fig. 4c). The suppression of high order signals in the simulations (Figs. 4b and 4c) indicates that the high-order nonlinear peaks observed in THz-MDCS experiment are generated by the four-fold magnetic anisotropy as well as the DM interaction. This interpretation can be seen more clearly by plotting a spectral slice along ωτ=ωAF\omega_{\tau}=\omega_{\mathrm{AF}} in Fig. 4d, where the full-term calculation (black line) is compared with simulations where (i) the DM interaction is switched off (blue line), or (ii) the four-fold magnetic anisotropy is switched off (red line). This observation reveals the existence of a four-fold magnetic anisotropy and also demonstrates that the high order magnonic nonlinearities arise from the dynamical interplay between DM interaction and four-fold magnetic anisotropy.

Figures 4e and 4f underscore the significance of considering the quantum fluctuation properties of spins in describing the THz-MDSC spectra. While the simulations based on the classical spin LLG model in Fig. 2d can replicate the positions of the peaks in the THz-MDCS spectra, the strengths of the high-order nonlinear signals using classical spins are much weaker compared to the experimental data in Fig. 1e and Fig. 2c. For instance, the experimental data (gray squares) at ωτ=2ωAF\omega_{\tau}=2\omega_{\mathrm{AF}} and ωτ=3ωAF\omega_{\tau}=3\omega_{\mathrm{AF}} exhibits substantially larger high-harmonic generation signals in Fig. 4f compared to the result of the classical simulations (blue diamonds). To understand this we have also calculated 𝐌NL(ωt,ωτ)\mathbf{M}_{\mathrm{NL}}(\omega_{t},\omega_{\tau}) for the quantum spin version of Hamiltonian (1). This is done by replacing the classical spins 𝐒i\mathbf{S}_{i} in Eq. (1) by quantum spin operators 𝐒^i\hat{\mathbf{S}}_{i} at site ii of spin magnitude S=5/2S=5/2. The resulting two-site quantum spin Hamiltonian has (2S+1)2=36(2S+1)^{2}=36 eigenstates |Ψn|\Psi_{n}\rangle and eigenenergies EnE_{n}. The inset of Fig. 4e shows the eigenenergy spectrum of the quantum spin model. We have used the same parameters for KaK_{a}, KcK_{c}, and K4K_{4} as in the classical simulations and adjusted JJ and DD such that the magnon frequency, defined by ωAF=E1E0\omega_{\mathrm{AF}}=E_{1}-E_{0} in the quantum spin model, matches the one observed in the experiment (Table 2, Methods). The energy spectrum is anharmonic, however, the difference between the eigenenergies is given by multiples of the magnon energys such that one expects a harmonic energy spectrum.

To model the transmitted magnetic field, we calculate the dynamics of the magnetization along cc-direction, Mc(t)Ψ(t)|(S^1,c+S^2,c)|Ψ(t)M^{c}(t)\equiv\langle\Psi(t)|(\hat{S}_{1,c}+\hat{S}_{2,c})|\Psi(t)\rangle. Here |Ψ(t)|\Psi(t)\rangle is the quantum state at time tt which is obtained via exact diagonalization. The calculated nonlinear magnetization MNLc(t,τ)=M12c(t,τ)M1c(t,τ)M2c(t)M_{\mathrm{NL}}^{c}(t,\tau)=M_{12}^{c}(t,\tau)-M_{1}^{c}(t,\tau)-M_{2}^{c}(t) is presented in Fig. 4e (red line) at a fixed inter-pulse delay of τ=\tau= 4.8 ps as in Fig. 1e. We observe high-harmonic generation peaks (vertical dashed lines) up to seventh-order. Compared to that, the result of the corresponding classical simulation in Fig. 4e (blue line) exhibits about one-order of magnitude smaller second to fifth-order harmonic generation signals. To identify the origin of the strong magnonic high-harmonic generation signals of the quantum spin simulation, we express |Ψ(t)|\Psi(t)\rangle after the magnetic field pulse-pair excitation in terms of the eigenstates |Ψn|\Psi_{n}\rangle, yielding |Ψ(t)=naneiEnt|Ψn|\Psi(t)\rangle=\sum_{n}a_{n}\,e^{-iE_{n}t}|\Psi_{n}\rangle with anΨn|Ψ(t=0)a_{n}\equiv\langle\Psi_{n}|\Psi(t=0)\rangle. The dynamics of magnetization can then be written as Mc(t)=j,kMj,kcajakei(EkEj)tM^{c}(t)=\sum_{j,k}M^{c}_{j,k}\,a_{j}^{*}a_{k}e^{-i(E_{k}-E_{j})t} with magnetic dipole matrix elements Mj,kcΨj|(S^1,c+S^2,c)|ΨkM^{c}_{j,k}\equiv\langle\Psi_{j}|(\hat{S}_{1,c}+\hat{S}_{2,c})|\Psi_{k}\rangle. As a result, the peaks in the spectrum arise from transitions between different eigenstates. The contribution of the transitions between the different eigentstates to Mc(t)M_{c}(t) is determined by strength of the magnetic dipole matrix elements Mj,kcM^{c}_{j,k}. Since the difference between eigenenergies (inset of Fig. 4e) corresponds to multiples of the magnon frequency, we obtain a harmonic McM^{c} spectrum as observed in Fig. 4e.

Figure 4f directly compares the magnonic high harmonic generation observed in the simulations and experiment by plotting the ratio between the peak strengths of the nnth harmonic and the fundamental harmonic. The result of the experiment (gray squares) is compared with the ratios deduced from the quantum-spin simulation (red circles) and classical simulation (blue diamonds). The quantum spin simulation agrees well with the ratios from experiment while the classical simulation produces substantially smaller high-harmonic generation peaks. Our results from the quantum spin calculation depicted in Fig. 4f emphasize the importance of further investigation into the pronounced higher-order magnon nonlinearities observed in our THz-MDCS experiment, which notably exceed the predictions of the classical spin model. Additional factors also need to be considered in the future quantitative analysis of high-order magnon nonlinear peaks, including the THz electro-optic sampling response functions [51] and THz propagation effects [52, 53]. But we expect neither of these effects to substantially modify the peak strengths of the 2ωAF\omega_{\mathrm{AF}} and 3ωAF\omega_{\mathrm{AF}} nonlinear magnon peaks, which constitute the focus of the quantum spin calculations.

At last, we elaborate three key distinctions of the THz magnon multiplication observed in this work. First, magnon multiplication are generated by utilizing resonant excitation of collective spin wave modes with long-lasting quantum coherence that persists well after the laser pulses. High harmonic generation (HHG) in solids often occurs during the period of laser excitation. Second, THz-MDCS achieves super-resolution tomography of magnon interaction and nonlinearity. Our results reveal four-fold magnetic anisotropy and the Dzyaloshinskii-Moriya interaction in creating interactions among multiple long-lasting magnon quantum excitations. Such interaction leads to the high-order magnon multiplication. Third, the magnon multiplication peaks in the 2D spectrum can arise from resonant transitions between different quantum spin eigenstates [54]. Consequently, magnon multiplication nonlinearity can potentially provide direct information about quantum spin systems. Our results provide compelling implications of considering the quantum fluctuation properties of spins.

In summary, we demonstrate the extremely high-order magnonic multiplication and wave-mixing at THz frequency by driving a canted antiferromagnetic state. Our experimental results are reproduced by both classical and quantum spin simulations, signifying the importance of four-fold magnetic anisotropy and DM symmetry breaking, also the fluctuation properties of quantum spins in the higher order magnon multiplication. Our demonstration of magnonic nonlinearity and collectivity through THz coherent spectroscopy can be further extended to pursue nonlinear quantum magnonics in parallel with nonlinear quantum optics.

Methods

Sample

Polycrystalline sample Sm0.4Er0.6FeO3\text{Sm}_{0.4}\text{Er}_{0.6}\text{FeO}_{3} was prepared by the conventional solid state reaction method, using the high purity oxide powder Sm2O3\text{Sm}_{2}\text{O}_{3} (99.99%), Er2O3\text{Er}_{2}\text{O}_{3} (99.99%) and Fe2O3\text{Fe}_{2}\text{O}_{3} (99.99%) as the starting materials. The mixture was pressed into pellets and sintered in air at 1400 K for 20 h. After confirmation by X-ray diffraction that the products had converted to the orthoferrite structure, the powder was then compacted into the rods. The rods were sintered in a vertical tube furnace at 1500 K for 15 h in air. Single crystal growth was performed in the optical floating-zone furnace (FZ-T-10000-H-VI-P-SH, Crystal System Corp).

The sample measured in the experiment is aa-cut Sm0.4Er0.6FeO3\text{Sm}_{0.4}\text{Er}_{0.6}\text{FeO}_{3} single crystal with size around 10 mm×\times5 mm×\times1 mm. At room temperature, the canted spin antiferromagnetic order leads to a small net magnetization along c{c}-axis (Fig. 1a). The magnetic resonance mode (quasi-AFM mode) can be excited if 𝐁THz\mathbf{B}_{\text{THz}} is parallel to this net magnetization. The in-plane axis is determined by rotating the sample to achieve the strongest magnon oscillation signal. In the experiment, we fixed sample c{c}-axis perpendicular to THz electric field.

THz multi-dimensional coherent nonlinear spectroscopy

To generate and characterize THz-MDCS spectra, the Sm0.4Er0.6FeO3\text{Sm}_{0.4}\text{Er}_{0.6}\text{FeO}_{3} single crystal sample was excited by two broadband THz laser pulses. The measured coherent differential transmission ENL(t,τ)=E12(t,τ)E1(t,τ)E2(t)E_{\mathrm{NL}}(t,\tau)=E_{\mathrm{12}}(t,\tau)-E_{\mathrm{1}}(t,\tau)-E_{\mathrm{2}}(t) is plotted as a function of the gate (real) time tt and the delay time between pulses 1 and 2, τ\tau, which controls the relative phase accumulation between the two phase-locked THz fields. The time-resolved coherent nonlinear dynamics is then explored by varying the inter-pulse delay τ\tau . By measuring the electric fields in the time-domain through electro-optic sampling (EOS) by a third pulse, we achieve phase-resolved detection of the sample response as a function of gate time tt for each time delay τ\tau. The nonlinear response signals, separated from the linear response, leads to enhanced resolution that enables us to observe high order magnonic nonlinearity.

Experimental details

Our THz-MDCS setup is driven by a 800 nm Ti: Sapphire amplifier with 4 mJ pulse energy, 40 fs pulse duration, and 1 kHz repetition rate. The majority part of laser output (2.7 W) is split into two beams with similar power (1.35 W each), and the inter-pulse delay between them is controlled by a mechanical delay stage. Two beams are recombined and focused into a MgO doped LiNbO3 crystal to generate intense THz pulses through the tilted pulse front scheme. The intense THz pulse has a peak field strength \sim 47.5 MV/m, central frequency \sim 1 THz (4.1 meV), and bandwidth \sim 1.5 THz [48, 49, 50]. The focused THz spot size at the sample position is \sim 1.3 mm. The rest part of the laser output (\sim 1 mW) is used to detect THz fields in time-domain by EOS using a 2 mm thick (110) oriented ZnTe crystal.

To acquire the THz nonlinear signal, a double modulation scheme has been used. Two optical choppers modulate two intense THz beams at a frequency of 500 Hz (pulse 1) and 250 Hz (pulse 2), respectively. Such a scheme allows to isolate nonlinear THz signals. The FT is performed at 7 – 28 ps range of the long lasting time-domain oscillatory traces as shown in Figs. 1c and 1d. The range is chosen to ensure avoidance of the stimulating THz pulses and echo pulses. The high order THz magnon multiplication obtained were possible by (1) choosing the Sm-doped magnetic material with the bigger four-fold magnetic anisotropy compared to un-doped ones, (2) applying an experimental scheme of 2 pulses vs 1 pulse driving and (3) optimizing the detection sensitivity of 2D coherent nonlinear signals.

THz–MDCS signal simulations

To simulate the THz–MDCS signals, we solve Landau–Lifshitz–Gilbert (LLG) equations derived from the Hamiltonian (1[37]:

d𝐒idt=γ1+α2[𝐒i×𝐁ieff(t)+α|𝐒i|𝐒i×(𝐒i×𝐁ieff(t))],\displaystyle\frac{\mathrm{d}\mathbf{S}_{i}}{\mathrm{d}t}=\frac{\gamma}{1+\alpha^{2}}\left[\mathbf{S}_{i}\times\mathbf{B}_{i}^{\mathrm{eff}}(t)+\frac{\alpha}{|\mathbf{S}_{i}|}\mathbf{S}_{i}\times\left(\mathbf{S}_{i}\times\mathbf{B}_{i}^{\mathrm{eff}}(t)\right)\right]\,, (2)

which describe the nonlinear coherent spin dynamics induced by the magnetic field 𝐁(t)\mathbf{B}(t). Here, 𝐁ieff(t)=1γH𝐒i\mathbf{B}_{i}^{\mathrm{eff}}(t)=-\frac{1}{\gamma}\frac{\partial H}{\partial\mathbf{S}_{i}} corresponds to the time-dependent effective local magnetic field at site ii and α\alpha is the Gilbert damping coefficient. We numerically solve the full nonlinear LLG equations (2) with input THz magnetic field pulses 1 and 2 polarized along cc-axis for realistic orthoferrite material parameters summarized in Table 1. The parameters used match the magnon mode frequency, free-induction decay, and nonlinear magnon signals observed in the experiment. We then calculate the net magnetization 𝐌𝐒1+𝐒2\mathbf{M}\propto\mathbf{S}_{1}+\mathbf{S}_{2} of the two AFM sub–lattices. To directly simulate our THz-MDCS experiment, we calculate the nonlinear differential magnetization 𝐌NL\mathbf{M}_{\mathrm{NL}}, which corresponds to the measured 𝐄NL\mathbf{E}_{\mathrm{NL}}. 𝐌NL\mathbf{M}_{\mathrm{NL}} is obtained by computing the net magnetization induced by both pulses, 𝐌12(t,τ)\mathbf{M}_{\mathrm{12}}(t,\tau), as a function of the gate time tt and the inter-pulse time τ\tau, as well as the net magnetizations resulting from pulse 1, 𝐌1(t,τ)\mathbf{M}_{\mathrm{1}}(t,\tau), and pulse 2, 𝐌2(t)\mathbf{M}_{\mathrm{2}}(t). The nonlinear differential magnetization is then given by 𝐌NL(t,τ)=𝐌12(t,τ)𝐌1(t,τ)𝐌2(t)\mathbf{M}_{\mathrm{NL}}(t,\tau)=\mathbf{M}_{\mathrm{12}}(t,\tau)-\mathbf{M}_{\mathrm{1}}(t,\tau)-\mathbf{M}_{\mathrm{2}}(t) for the collinear geometry used in the experiment. The THz-MDCS spectra are obtained by Fourier transform of 𝐌NL(t,τ)\mathbf{M}_{\mathrm{NL}}(t,\tau) with respect to both tt (frequency ωt\omega_{t}) and τ\tau (frequency ωτ\omega_{\tau}). To analyze these 2D spectra, we introduce the frequency vector (ωt,ωτ)(\omega_{t},\omega_{\tau}). Since both pulses resonantly drive magnon 1-quantum coherences (1QC) characterized by the magnon frequency ωAF\omega_{\mathrm{AF}}\sim 2.5 meV (Fig. 1b), the frequency vectors of the two pulses 1 and 2 can be written as ω1=(ωAF,ωAF)\omega_{\mathrm{1}}=(\omega_{\mathrm{AF}},\omega_{\mathrm{AF}}) and ω2=(ωAF,0)\omega_{\mathrm{2}}=(\omega_{\mathrm{AF}},0). The observed spectral peaks and corresponding nonlinear processes are summarized in Table 3.

Parameter value used
Exchange constant JJ 5.0 meV
Anisotropy constant KaK_{a} 1.2×1021.2\times 10^{-2} meV
Anisotropy constant KcK_{c} 0.46×1020.46\times 10^{-2} meV
Anisotropy constant K4K_{4} 1.8×1041.8\times 10^{-4} meV
Antisymmetric exchange constant DD 7.6×1027.6\times 10^{-2} meV
Total spin number SS 5/2
α\alpha 1.3×1031.3\times 10^{-3}

Table 1. Parameters used in the classical simulations

Parameter value used
Exchange constant JJ 2.5 meV
Anisotropy constant KaK_{a} 1.2×1021.2\times 10^{-2} meV
Anisotropy constant KcK_{c} 0.46×1020.46\times 10^{-2} meV
Anisotropy constant K4K_{4} 1.8×1041.8\times 10^{-4} meV
Antisymmetric exchange constant DD 0.50.5 meV
Total spin number SS 5/2

Table 2. Parameters used in the quantum spin simulations

signal nonlinear process frequency space
MR ω1ω2\omega_{1}-\omega_{2} (0,ωAF)(0,\omega_{\mathrm{AF}})
MR ω2ω1\omega_{2}-\omega_{1} (0,ωAF)(0,-\omega_{\mathrm{AF}})
PP (ω1ω1)+ω2(\omega_{1}-\omega_{1})+\omega_{2} (ωAF,0)(\omega_{\mathrm{AF}},0)
NR (ω1ω2)+ω2(\omega_{1}-\omega_{2})+\omega_{2} (ωAF,ωAF)(\omega_{\mathrm{AF}},\omega_{\mathrm{AF}})
2QC 2ω1ω22\omega_{1}-\omega_{2} (ωAF,2ωAF)(\omega_{\mathrm{AF}},2\omega_{\mathrm{AF}})
R (ω2ω1)+ω2(\omega_{2}-\omega_{1})+\omega_{2} (ωAF,ωAF)(\omega_{\mathrm{AF}},-\omega_{\mathrm{AF}})
2HG 2ω12\omega_{1} (2ωAF,2ωAF)(2\omega_{\mathrm{AF}},2\omega_{\mathrm{AF}})
2HG ω1+ω2\omega_{1}+\omega_{2} (2ωAF,ωAF)(2\omega_{\mathrm{AF}},\omega_{\mathrm{AF}})
5WM 3ω2ω13\omega_{2}-\omega_{1} (2ωAF,ωAF)(2\omega_{\mathrm{AF}},-\omega_{\mathrm{AF}})
3HG ω1+2ω2\omega_{1}+2\omega_{2} (3ωAF,ωAF)(3\omega_{\mathrm{AF}},\omega_{\mathrm{AF}})
3HG 2ω1+ω22\omega_{1}+\omega_{2} (3ωAF,2ωAF)(3\omega_{\mathrm{AF}},2\omega_{\mathrm{AF}})
6WM 4ω2ω14\omega_{2}-\omega_{1} (3ωAF,ωAF)(3\omega_{\mathrm{AF}},-\omega_{\mathrm{AF}})
4HG 3ω2+ω13\omega_{2}+\omega_{1} (4ωAF,ωAF)(4\omega_{\mathrm{AF}},\omega_{\mathrm{AF}})
4HG 2ω2+2ω12\omega_{2}+2\omega_{1} (4ωAF,2ωAF)(4\omega_{\mathrm{AF}},2\omega_{\mathrm{AF}})
7WM 5ω2ω15\omega_{2}-\omega_{1} (4ωAF,ωAF)(4\omega_{\mathrm{AF}},-\omega_{\mathrm{AF}})
5HG 4ω2+ω14\omega_{2}+\omega_{1} (5ωAF,ωAF)(5\omega_{\mathrm{AF}},\omega_{\mathrm{AF}})
6HG 5ω2+ω15\omega_{2}+\omega_{1} (6ωAF,ωAF)(6\omega_{\mathrm{AF}},\omega_{\mathrm{AF}})

Table 3. Nonlinear magnon processes contributing to the THz-MDCS spectra

Data Availability

All the data generated in this study are provided in the Source Data file.

References

  • [1] Li, T. et al. Femtosecond switching of magnetism via strongly correlated spin–charge quantum excitations. Nature 496, 69–73 (2013).
  • [2] Kampfrath, T. et al. Coherent terahertz control of antiferromagnetic spin waves. Nat. Photon. 5, 31–34 (2011).
  • [3] Jungwirth, T., Marti, X., Wadley, P. & Wunderlich, J. Antiferromagnetic spintronics. Nat. Nanotechnol. 11, 231–241 (2016).
  • [4] Chovan, J. & Perakis, I. E. Femtosecond control of the magnetization in ferromagnetic semiconductors. Phys. Rev. B 77, 085321 (2008).
  • [5] Kapetanakis, M. D., Perakis, I. E., Wickey, K. J., Piermarocchi, C. & Wang, J. Femtosecond coherent control of spins in (Ga,Mn)As ferromagnetic semiconductors using light. Phys. Rev. Lett. 103, 047404 (2009).
  • [6] Wang, J. et al. Memory effects in photoinduced femtosecond magnetization rotation in ferromagnetic GaMnAs. Appl. Phys. Lett. 94, 021101 (2009).
  • [7] Lingos, P. C., Wang, J. & Perakis, I. E. Manipulating femtosecond spin-orbit torques with laser pulse sequences to control magnetic memory states and ringing. Phys. Rev. B 91, 195203 (2015).
  • [8] Patz, A. et al. Ultrafast probes of nonequilibrium hole spin relaxation in the ferromagnetic semiconductor GaMnAs. Phys. Rev. B 91, 155108 (2015).
  • [9] Lingos, P. C., Kapetanakis, M. D., Wang, J. & Perakis, I. E. Light-wave control of correlated materials using quantum magnetism during time-periodic modulation of coherent transport. Commun. Phys. 4, 60 (2021).
  • [10] Nova, T. F. et al. An effective magnetic field from optically driven phonons. Nat. Phys. 13, 132–136 (2017).
  • [11] Olejník, K. et al. Terahertz electrical writing speed in an antiferromagnetic memory. Sci. Adv. 4, eaar3566 (2018).
  • [12] Li, J. et al. Spin current from sub-terahertz-generated antiferromagnetic magnons. Nature 578, 70–74 (2020).
  • [13] Kubacka, T. et al. Large-amplitude spin dynamics driven by a THz pulse in resonance with an electromagnon. Science 343, 1333 (2014).
  • [14] Mashkovich, E. A. et al. Terahertz light-driven coupling of antiferromagnetic spins to lattice. Science 374, 1608 (2021).
  • [15] Fitzky, G. et al. Ultrafast control of magnetic anisotropy by resonant excitation of 4f electrons and phonons in Sm0.4Er0.6FeO3\text{Sm}_{0.4}\text{Er}_{0.6}\text{FeO}_{3}. Phys. Rev. Lett. 127, 107401 (2021)
  • [16] Afanasiev, D. et al. Ultrafast control of magnetic interactions via light-driven phonons. Nat. Mater. 20, 607 (2021).
  • [17] Hortensius, J. R. et al. Coherent spin-wave transport in an antiferromagnet. Nat. Phys. 354, 17 (2021).
  • [18] Zhang, Z. et al. Generation of third-harmonic spin oscillation from strong spin precession induced by terahertz magnetic near fields. Nat. Commun. 14, 1795 (2023).
  • [19] Baierl, S. et al. Nonlinear spin control by terahertz-driven anisotropy fields. Nat. Photon. 10, 715–718 (2016).
  • [20] Behovits, Y. et al. Terahertz Neel spin-orbit torques drive nonlinear magnon dynamics in antiferromagnetic Mn2Au. Nat. Commun. 14, 6038 (2023).
  • [21] Kurihara, T., Bamba, M., Watanabe, H., Nakajima, M. & Suemoto, T. Observation of terahertz-induced dynamical spin canting in orthoferrite magnon by magnetorefractive probing. Commun. Phys. 6, 51 (2023).
  • [22] Mashkovich, E. A. et al. Terahertz Optomagnetism: Nonlinear THz Excitation of GHz Spin Waves in Antiferromagnetic FeBO3. Phys. Rev. Lett. 123, 157202 (2019).
  • [23] Baierl, S. et al. Terahertz-Driven Nonlinear Spin Response of Antiferromagnetic Nickel Oxide. Phys. Rev. Lett. 117, 197201 (2016).
  • [24] Khan, P., Kanamaru, M., Matsumoto, K., Ito, T. & Satoh, T. Ultrafast light-driven simultaneous excitation of coherent terahertz magnons and phonons in multiferroic BiFeO3. Phys. Rev. B 101, 134413 (2020).
  • [25] Yuan, H., Cao, Y., Kamra, A., Duine, R. A. & Yan, P. Quantum magnonics: When magnon spintronics meets quantum information science. Phys. Rep. 965, 1–74 (2022).
  • [26] Schlauderer, S. et al. Temporal and spectral fingerprints of ultrafast all-coherent spin switching. Nature 569, 383–387 (2019).
  • [27] Chumak, A. V., Vasyuchka, V. I., Serga, A. A. & Hillebrands, B. Magnon spintronics. Nat. Phys. 11, 453–461 (2015).
  • [28] Pirro, P., Vasyuchka, V. I., Serga, A. A. & Hillebrands, B. Advances in coherent magnonics. Nat. Rev. Mater. 6, 1114–1135 (2021).
  • [29] Takayoshi, S., Murakami, Y. & Werner, P. High-harmonic generation in quantum spin systems. Phys. Rev. B 99, 184303 (2019).
  • [30] Lachance-Quirion, D., Tabuchi, Y., Gloppe, A., Usami, K. & Nakamura, Y. Hybrid quantum systems based on magnonics. Appl. Phys. Express 12, 070101 (2019).
  • [31] Koerner, C. et al. Frequency multiplication by collective nanoscale spin-wave dynamics. Science 375, 1165–1169 (2022).
  • [32] Maag, T. et al. Coherent cyclotron motion beyond Kohn’s theorem. Nat. Phys. 12, 119–123 (2016).
  • [33] Kuehn, W., Reimann, K., Woerner, M. & Elsaesser, T. Phase-resolved two-dimensional spectroscopy based on collinear n-wave mixing in the ultrafast time domain. J. Phys. Chem. 130, 164503 (2009).
  • [34] Mootz, M., Luo, L., Wang, J. & Perakis, l. E. Visualization and quantum control of light-accelerated condensates by terahertz multi-dimensional coherent spectroscopy. Commun. Phys. 5, 47 (2022).
  • [35] Luo, L. et al. Quantum coherence tomography of light-controlled superconductivity. Nat. Phys. 19, 201 (2023).
  • [36] Huang, C. K., et al. Discovery of an Unconventional Quantum Echo by Interference of Higgs Coherence. arXiv:2312.10912. (2024).
  • [37] Lu, J. et al. Coherent two-dimensional terahertz magnetic resonance spectroscopy of collective spin waves. Phys. Rev. Lett. 118, 207204 (2017).
  • [38] Zhang, Z. et al. Terahertz field-driven magnon upconversion in an antiferromagnet. Nat. Phys. (2024).
  • [39] Zhang, Z. et al. Terahertz field-induced nonlinear coupling of two magnon modes in an antiferromagnet. Nat. Phys. (2024).
  • [40] Borsch, M. et al. Super-resolution lightwave tomography of electronic bands in quantum materials. Science 370, 1204 (2020).
  • [41] X. Yang, et al., Light Control of Surface–Bulk Coupling by Terahertz Vibrational Coherence in a Topological Insulator, npj Quantum Materials 5, 13 (2020).
  • [42] C. Vaswani, L. L. Wang, D. H. Mudiyanselage, Q. Li, P. M. Lozano, G. D. Gu, D. Cheng, B. Song, L. Luo, R. H. J. Kim, C. Huang, Z. Liu, M. Mootz, I. E. Perakis, Y. Yao, K. M. Ho, and J. Wang, Light-Driven Raman Coherence as a Nonthermal Route to Ultrafast Topology Switching in a Dirac Semimetal, Phys. Rev. X 10, 021013 (2020).
  • [43] Z. Liu, C. Vaswani, L. Luo, D. Cheng, X. Yang, X. Zhao, Y. Yao, Z. Song, R. Brenes, R. J. H. Kim, J. Jean, V. Bulović, Y. Yan, K.-M. Ho and J. Wang, Coherent band-edge oscillations and dynamic longitudinal-optical phonon mode splitting as evidence for polarons in perovskites. Phys. Rev. B 101,115125 (2020).
  • [44] Zakharov, V. E., L'vov, V. S. & Starobinets, S. S. Spin-wave turbulence beyond the parametric excitation threshold. Sov. Phys. Usp. 17, 896–919 (1975).
  • [45] Shapiro, S. M., Axe, J. D. & Remeika, J. P. Neutron-scattering studies of spin waves in rare-earth orthoferrites. Phys. Rev. B 10, 2014–2021 (1974).
  • [46] Shane, J. R. Resonance frequencies of the orthoferrites in the spin reorientation region. Phys. Rev. Lett. 20, 728–732 (1968).
  • [47] Herrmann, G. F. Magnetic resonances and susceptibility in orthoferrites. Phys. Rev. 133, A1334 (1964).
  • [48] Vaswani, C. et al. Terahertz second-harmonic generation from lightwave acceleration of symmetry-breaking nonlinear supercurrents. Phys. Rev. Lett. 124, 207003 (2020).
  • [49] Yang, X. et al. Terahertz-light quantum tuning of a metastable emergent phase hidden by superconductivity. Nat. Mater. 17, 586 (2018).
  • [50] Yang, X. et al. Light control of surface-bulk coupling by terahertz vibrational coherence in a topological insulator. npj Quantum Mater. 5, 13 (2020).
  • [51] Gallot, G. et al. Electro-optic detection of terahertz radiation. J. Opt. Soc. Am. B 16, 1204 (1999).
  • [52] Grishunin, K. et al. Terahertz Magnon-Polaritons in TmFeO3. ACS Photonics 5, 1375 (2018).
  • [53] Metzger, T. W. J. et al. Effect of antiferromagnetic order on a propagating single-cycle THz pulse. Appl. Phys. Lett. 121, 252403 (2022).
  • [54] Zhang, J. & Tanimura, Y. Coherent two-dimensional THz magnetic resonance spectroscopies for molecular magnets: Analysis of Dzyaloshinskii-Moriya interaction. J. Chem. Phys. 159, 014102 (2023).

Acknowledgments

This work was supported by the Ames National Laboratory, the US Department of Energy, Office of Science, Basic Energy Sciences, Materials Science and Engineering Division under contract No. DEAC0207CH11358 (project design and THz-MDCS spectroscopy). Synthesis and basic characterizations of crystals was supported by National Natural Science Foundation of China (NSFC, No.52272014), Science and Technology Commission of Shanghai Municipality (No. 21520711300) and CAS Project for Young Scientists in Basic Research (YSBR-024). The quantum spin model simulations were supported by the U.S. Department of Energy, Office of Science, National Quantum Information Science Research Centers, Co-design Center for Quantum Advantage (C2QA) under contract number DE-SC0012704.

Author Contributions

C.H. and L.L. performed the THz spectroscopy measurements with J.W.’s supervision. M.M., and Y.Y simulated the THz magnon nonlinearity using the quantum spin model. M.M., and I.E.P analyzed the THz spectra using the classical spin model. J.S., P.M., L.S. and A.W. grew the samples and performed crystalline quality and other basic characterizations. The paper is written by J.W., C.H., and M.M. with discussions from all authors. J.W. conceived and coordinated the project.

Competing interests

The authors declare no competing interests.

[Uncaptioned image]

Fig. 1. THz-MDCS of magnonic multiplication in Sm0.4Er0.6FeO3. a, Experimental schematics. Two phase-locked THz pulses with inter-pulse delay τ\tau are focused onto the sample. The magnetic fields of this THz pulse-pair are polarized parallel to the sample cc-axis to excite the quasi-AFM mode. The coherent nonlinear emission signals are detected through electro-optic sampling. b, The static THz transmission measurement of the sample, which describes the linear magnonic response, shows a peak at the magnon mode energy \sim 2.5 meV (0.6\sim 0.6 THz). c and d, Typical four channel THz signals measured by THz-MDCS at a fixed delay time τ=4.8\tau=4.8 ps. Two individual THz pulses used here, E1E_{1} (red line) and E2E_{2} (blue line), are shown in c. The coherent emission signal when both pulses are present, E12E_{12} (light blue line), as well as the extracted nonlinear signal ENLE_{\text{NL}} (black line), obtained by subtracting the two individual pulse contributions from E12E_{12} (zoom in 100×100\times), are plotted in 𝐝\mathbf{d}. The Fourier transform of ENLE_{\text{NL}} is performed for the time window after the pulse indicated by the two dotted red lines. e, The ENLE_{\text{NL}} spectrum shows up to six–magnon multiplication (dashed lines) for THz magnetic fields parallel to the cc-axis (black solid line). The magnon nonlinear peaks are absent for THz magnetic fields perpendicular to the cc-axis (blue solid line).

[Uncaptioned image]

Fig. 2. Super-resolution tomography of magnonic multiplication and correlation. a, Two-dimensional false-color plot of the measured coherent nonlinear transmission signals ENL(t,τ)E_{\mathrm{NL}}(t,\tau) as a function of the gate time tt and the delay τ\tau induced with the electric field strength 47.5 MV/m. b, The simulated ENL(t,τ)E_{\mathrm{NL}}(t,\tau). c and d, The ENL(ωt,ωτ)E_{\mathrm{NL}}(\omega_{t},\omega_{\tau}) spectra from 2D-FT of time domain images a and b. The THz-MDCS spectra show PP, R, NR, and 2QC peaks, which are generated by third-order nonlinear processes. Second-order nonlinear processes lead to MR and SHG peaks. Importantly, strong THz driving also leads here to formation of high-order nonlinear signals including THG, 4HG, 5HG, 6HG as well as 5WM, 6WM and 7WM peaks. The frequency vectors for two incident pulses 1 and 2, ω1=(ωAF,ωAF)\omega_{1}=(\omega_{\mathrm{AF}},\omega_{\mathrm{AF}}) and ω2=(ωAF,0)\omega_{2}=(\omega_{\mathrm{AF}},0), are marked as red and black arrows. e and f, Zoom-in view of new high harmonic and wave mixing peaks observed in our experiments which involve up to six magnon quanta (main text). g and h, Corresponding zoom-in peaks reproduced by theory.

[Uncaptioned image]

Fig. 3. THz field dependence of magnonic multiplication. af, Magnonic multiplication peak amplitudes as a function of THz fields by recording spectral signals ENL(ωt,τ)E_{\mathrm{NL}}(\omega_{t},\tau) (Fig. 1e) at frequencies of 0, 1, 2, 3, 4, 5ωAF\omega_{\mathrm{AF}} and inter-pulse delay τ=4.8\tau=4.8 ps. Solid lines are power-law fitting of the experimental data (markers). The error bars marked in (a)-(f) are obtained by averaging background noise following the same data taking protocol and fall within the marker sizes for the lower orders. g, ENL(ωt,τ)E_{\mathrm{NL}}(\omega_{t},\tau) spectra at fixed inter-pulse delay τ=4.8\tau=4.8 ps for 100%  (purple line), 70% (green line), and 50% (orange line) of THz field excitations. h, Corresponding single pulse spectra E2(ωt)E_{2}(\omega_{t}) at the same field strengths.

[Uncaptioned image]

Fig. 4. Simulations of high-order nonlinear magnonic correlations. ac, MNL(ωt,ωτ)M_{\mathrm{NL}}(\omega_{t},\omega_{\tau}) for (a) the full-term calculation, (b) a calculation without DM interaction, D=0D=0, where the ground state has no spin canting, i. e., 𝐌=0\mathbf{M}=0 in the ground state, and (c) a calculation without fourfold anisotropy in the Hamiltonian, K4=0K_{4}=0. d, MNL(ωt,ωτ)M_{\mathrm{NL}}(\omega_{t},\omega_{\tau}) at fixed ωτ=ωAF\omega_{\tau}=\omega_{\mathrm{AF}}. The full-term calculation (black line) is compared with simulations where the DM interaction is switched off (blue line) and the four–fold magnetic anisotropy is switched off (red line). (e) MNLc(ωt,τ)M^{c}_{\mathrm{NL}}(\omega_{t},\tau) at a fixed inter-pulse delay of τ=4.8\tau=4.8 ps. The result of the full quantum spin simulation (red line) is shown together with the result of the classical simulation (blue line). The quantum spin simulation yields stronger high-harmonic generation peaks (vertical dashed lines) compared to the classical simulation. Inset: Eigenenergies EnE_{n} of the unperturbed quantum spin Hamiltonian which has (2s+1)2=36(2s+1)^{2}=36 eigenstates for the studied two-site spin-s=5/2s=5/2 system. (f) Ratio between the strength of nnth harmonic and fundamental harmonic generation peaks. The ratios extracted from the experimental mHHG spectrum in Fig. 1e (squares) are compared with the results of the quantum spin simulation (circles) and the classical simulation (rectangles).