betaFeedback

WorksPhysics and Astronomy

All-optical control of second-harmonic generation in ββ-BaB2_2O4_4 via coherent, terahertz-driven acentric lattice displacement

Abstract

Dynamical control of the nonlinear optical properties of solids – with light itself – will be essential for future ultrafast photonic technologies. Previously, methods to modulate nonlinear processes including second-harmonic generation (SHG) have relied primarily on non-resonant light-matter interaction or photo-generation of hot electrons in nanoscale materials. However, these approaches are typically constrained by limited interaction lengths and the initial frequency conversion is relatively weak under equilibrium conditions. Here, an approximately 30% modulation of efficient phase-matched SHG in bulk beta-barium borate (beta-BaB2O4) is achieved through transient lattice deformation by intense terahertz (THz) pulses that are tuned to resonance with an infrared-active phonon mode. The effect originates from modification of the index of refraction ellipsiod and the corresponding nonlinear phase-matching conditions, rather than from direct modulation of the nonlinear susceptibility through THz-mediated chi^(3) processes. This mechanism, of resonant selective lattice excitation, points toward novel THz-control schemes to tune the nonlinear optical response in materials.

Introduction

Nonlinear optical frequency conversion – including second harmonic generation (SHG) – greatly extends the spectral range of conventional lasers, underpinning key scientific applications across a wide range of fields from high-resolution microscopy to precision optical metrology to quantum information science[1, 2, 3, 4, 5, 6]. For communications and technological applications, integrated photonic circuits have emerged[7, 8, 9, 10], in which the nonlinear optical response can be precisely modulated. Control over optical nonlinearities, including SHG, using applied electric fields has now been demonstrated in optical crystals[11], polymers[12], semiconductors[13, 14, 15], atomic-layer systems and heterostructures[16, 17], and artificial photonic structures[18, 19]. However, modulation of the nonlinear optical response with this approach is fundamentally limited by the speed of conventional electronics to GHz modulation rates, creating a functional barrier to the next generation of optoelectronic devices. In principle, orders-of-magnitude higher speeds can be achieved through all-optical approaches, which would enable ultrafast data transfer and real-time optical signal processing.

Indeed, ultrafast, all-optical SHG modulation has been demonstrated in nanostructures and two-dimensional materials[20, 21, 22, 23]. However, these achievements appear impractical for applications in solid-state devices, as the modulation generally relies on either photoexcitation of hot electrons[22, 24] with significant collateral heating, or non-resonant light-matter interaction[25, 26, 27, 28, 29], which is limited by material damage thresholds. Further, while high contrast has been achieved, the absolute SHG signals themselves remain relatively weak due to limited interaction lengths. On the other hand, while bulk three-dimensional materials can deliver strong SHG signals suitable for applications, high-contrast, ultrafast, all-optical modulation has not yet been reported.

In this work, high-contrast modulation of strong SHG is achieved by resonant excitation of the β\beta-barium borate (β\beta-BaB2O4\mathrm{BaB}_{2}\mathrm{O}_{4}, BBO) crystal lattice using intense terahertz (THz) pulses tuned to an infrared (IR)-active phonon mode. Theoretical considerations — based on group theory and ab-initio calculations[30, 31, 32] — indicate that both the linear properties and the second-order optical nonlinearity in BBO originates from localized electronic density within the B3O6\mathrm{B}_{3}\mathrm{O}_{6} anionic rings in the crystal unit cell. Therefore, displacement of the anionic ring can be predicted to significantly influence SHG and targeted displacements might lead to strong, controlled modulation of the SHG signal. While this parameter cannot be selectively accessed with conventional stimuli (e.g., strain, static electric fields), selective displacement can be achieved with intense THz pulses, resonantly coupled to IR-active phonons, as illustrated in Figure 1a. With this approach, THz pulses can be expected to enable ultrafast control over the linear and nonlinear optical properties of BBO, enabling efficient modulation of the SHG process.

THz-driven optical SHG modulation dynamics

Figure 1. THz-driven optical SHG modulation dynamics of β\beta-BaB2_2O4_4: a, Illustration of the planar B3_3O6_6 anionic unit surrounded by Ba atoms within the unit cell of β\beta-BaB2_2O4_4 in the crystallographic coordinates x′x' and y′y', along with the

The potential for control by THz-driven resonant excitation is investigated here through the SHG of a femtosecond NIR probe laser. By measuring the time-resolved SHG efficiency as a function of the fundamental input wave polarization and delay with respect to the THz excitation, the modulated SHG signal yields the overall out-of-equilibrium nonlinear optical response. And importantly, under appropriate input conditions, high-contrast SHG modulation exceeding 30% can be reached.

Macroscopically, this effect can be explained by an electro-optic, THz-induced rotation of the principal optical axes (perpendicular to the direction of propagation), which alters the SHG phase matching conditions. Density functional perturbation theory (DFPT) calculations validated by the experiment can then be used to obtain an accurate estimate of the frequency dispersion of the electro-optic coefficient in the THz range, as well as the phonon normal-mode amplitude of 0.67 A˚ (amu)1/20.67\,\text{\AA}\,(\text{amu})^{1/2}, which displaces the B3O6\mathrm{B}_{3}\mathrm{O}_{6} anionic unit from its equilibrium position. This quantitative link between the atomic-scale structural dynamics and macroscopic nonlinear optical response may facilitate engineering of materials with specific transient nonlinear optical responses in the future.

Results

Experimental design and THz excitation geometry

To investigate the SHG modulation by selective lattice excitation on ultrafast timescales, intense THz pulses are used to drive the material while its linear and nonlinear response are probed with an overlapping femtosecond near-infrared (NIR) pulse (Figure 1a-Figure 1b and § 11 and § 12 of the Supplementary Information for details). Prior to probing the dynamics, the BBO crystal is characterized in the chosen experimental geometry in which the THz pulse is polarized along the ordinary axis (o-axis) of the BBO crystal. Steady-state optical spectroscopy (see Supplementary Figure 3 and § 13 of the Supplementary Information for details) yields the imaginary part of the dielectric function ε2(ω)\varepsilon_{2}(\omega) and the corresponding IR-active phonon modes (Figure 1d). The dominant IR-active EE-symmetry phonon mode is found at a frequency of ωQ=4.32\omega_{Q} = 4.32 THz, consistent with other recent independent characterizations by THz time-domain spectroscopy[33]. This mode is of particular interest, as density functional perturbation theory (DFPT) calculations indicate that it leads to significant coplanar motion of anionic [B3O6]3−[\mathrm{B}_{3}\mathrm{O}_{6}]^{3-} rings while the Ba2+\mathrm{Ba}^{2+} cations will exhibit in nearly motion in the opposite direction with comparatively smaller displacements, due to their large atomic mass[31] (Figure 1a).

Transmission spectrum in the THz regime of a 30 μm-thick BBO crystal measured by FTIR spectroscopy

Supplementary Figure 3. Transmission spectrum in the THz regime of a 30 μ\mum-thick BBO crystal measured by FTIR spectroscopy. Black dotted line denotes the fit curve.

phonon displacement vectors projected in the crystallographic x′x^{\prime}-y′y^{\prime} plane associated with the IR-active mode at ωQ=4.3\omega_{Q} = 4.3 THz. Phonon displacement vectors have been calculated using density functional theory. b, Schematic of the experimental setup. The THz pulse interacts with the nonlinear crystal while the SHG response is probed with a temporally synchronized ∼\sim50 femtosecond NIR pulse (FF beam) converted in a SH pulse. Arrows define the orientation of the ordinary and extraordinary axes with respect to the crystallographic axes x′x^{\prime}, y′y^{\prime}, and cc (θ=29.3∘\theta= 29.3^{\circ}). c, Temporal waveform of the THz electric field and d, corresponding normalized Fourier spectral amplitude (Sp. Ampl.). e, Real and imaginary components (ε1\varepsilon_{1}, ε2\varepsilon_{2}) of the dielectric function of BBO (β\beta-BaB2_{2}O4_{4}) measure using FTIR spectroscopy. f, THz-driven SH intensity modulation dynamics (blue curve). g, Normalized Fourier spectral amplitude (Sp. Ampl.) of the SH harmonic modulation dynamics is shown in panel e.

Electro-optic sampling (EOS) of the driving THz pulse is used to verify spectral overlap with the EE phonon mode required for targeted excitation. In-situ EOS measurements of the THz temporal waveform ETHz(t)E_{\mathrm{THz}}(t) yield a peak electric field strength of ETHzpeak=8.5E_{\mathrm{THz}}^{\mathrm{peak}} = 8.5 MV/cm, and Fourier analysis reveals a spectrum heavily weighted around 4.3 THz, with additional weaker frequency components ranging from 0 – 10 THz (Figure 1c-Figure 1d). The spectral content ensures efficient coupling of the THz pulse to the desired EE phonon mode. While there are additional IR-active phonon modes spanned by the broadband THz pulse, they have lower absorption cross-sections[33] and are not expected to contribute significantly to the driven lattice dynamics (Figure 1e).

Following initial characterization, time-resolved experiments are performed by focusing the THz pulse into a 300 μ\mum thick BBO crystal cut for type-I phase-matched SHG of a ∼\sim50 femtosecond NIR laser pulse at the fundamental wavelength of 800 nm. In the experiment, 20 nJ NIR pulses are focused to a spot size of 35 μ\mum at the BBO crystal with corresponding intensity of ∼8×1010\sim8 \times10^{10} W/cm2^{2}, and at equilibrium a maximum SHG energy conversion efficiency of ∼4\sim4 % is observed.

By varying the delay between the THz pump pulse and polarization of the femtosecond NIR probe pulse, the influence of the phonon dynamics on the SHG response is investigated in the time domain as function of polarization. A key observable is the time-dependent, normalized relative strength of the SHG intensity, ΔIrel=ΔISH/ISH,0\Delta I_{\mathrm{rel}} = \Delta I_{\mathrm{SH}}/I_{\mathrm{SH},0}. Here, ΔISH=ISH,THz−ISH,0\Delta I_{\mathrm{SH}} = I_{\mathrm{SH,THz}} - I_{\mathrm{SH},0}, ISH,THzI_{\mathrm{SH,THz}}, and the equilibrium SHG intensity, ISH,0I_{\mathrm{SH},0}. By convention, time-zero occurs when the peak of the femtosecond NIR pulse coincides with the maximum of the THz electric field. Under optimized input polarization (detuned from peak conversion efficiency) and delay the THz pulse induces a modulation in the relative SHG strength exceeding 30% (Figure 1f), whereas no significant THz-induced modulation is observed in the NIR probe intensity (Supplementary Fig. 2).

The time-dependent SHG exhibits a damped oscillatory behavior persisting for over 4 ps, with Fourier analysis revealing a distinct peak slightly below the phonon frequency, ωQ\omega_{Q}, coincident with the positive maximum of real part of the dielectric function, ε1\varepsilon_{1}, see Figure 1g. Tuning the spectrum of the THz driving field with spectral filters to above and below the phonon frequency ωQ\omega_{Q}. When the spectral components of the driving THz field are above ωQ\omega_{Q}, the SHG modulation is suppressed. Similarly, when the THz cut-off frequency is below ωQ\omega_Q, modulation of the SHG is also strongly suppressed and can be attributed to weak excitation of lower frequency phonon modes (Figure 2a).

Phonon-induced SH intensity modulation dynamics of β-BaB2O4

Figure 2. Phonon-induced SH intensity modulation dynamics of β-BaB2O4. a, THz excitation of the BBO IR-active phonon modes. On resonance, a pronounced spectral amplitude (amp.) response is observed (grey curves). For comparison, low and high frequency off-resonant THz excitation (yellow and green curves) leads to a significantly reduced modulation of the second harmonic intensity. b-c, The amplitude of the SH intensity modulation is directly proportional to the resonant peak field ETHzpeakE_{\mathrm{THz}}^{\mathrm{peak}} of the THz pump and is inverted upon reversing the polarity of the applied THz field.

The phase dependence of the SHG modulation is further investigated by returning to the nominal THz pulse (tuned to resonance) and inverting its polarity. A corresponding inversion in the relative SHG intensity is observed, indicating that the dynamics are resonantly driven and phase-locked to the THz field (Figure 2b). Additionally, scaling the THz field strength results in a linearly dependent effect, suggesting that the SHG modulation is directly proportional to the phonon amplitude, which is itself linearly dependent on the THz driving field strength[34, 35, 36] (Figure 2c).

Polarisation dependence

At equilibrium, the expression for the type-I SHG in the undepleted pump approximation in BBO of length LL by an electromagnetic wave of intensity II, frequency ω\omega, and polarisation angle α\alpha is given by (see Supplementary Information Sec. 4):

ISH,0(α)=2ω2dNL2L2neno2c3ε0(sin⁡(12ΔkL)12ΔkL)2cos⁡4(α)I2(ω)(1)I_{\mathrm{SH},0}(\alpha)=\frac{2\omega^{2}d_{\mathrm{NL}}^{2}L^{2}}{n_{e}n_{o}^{2}c^{3}\varepsilon_{0}}\left(\frac{\sin\left(\frac{1}{2}\Delta kL\right)}{\frac{1}{2}\Delta kL}\right)^{2}\cos^{4}(\alpha)I^{2}(\omega) \tag*{(1)}

where no(ω)n_o(\omega) is the ordinary refractive index at the fundamental (ω\omega) and ne(2ω)n_e(2\omega) is the extraordinary refractive index at second harmonic (2ω2\omega), dNLd_{\mathrm{NL}} is the nonlinear optical coefficient, and Δk\Delta k represents the momentum mismatch between the fundamental and second harmonic waves (Δk=0\Delta k = 0 in perfect phase matching conditions). The polarisation angle α\alpha defines the orientation of the fundamental wave with respect to the ordinary axis, as illustrated in Figure 3a. According to Eq. (1), and consistent with equilibrium experimental observations, the SHG intensity ISH,0(α)I_{\mathrm{SH},0}(\alpha) exhibits a cosine-to-the-fourth power dependence on input polarisation, resulting in a two-fold symmetric angular distribution with maximum intensity at α=0\alpha= 0 and α=180∘\alpha= 180^\circ (Fig. 3b).

Polarization resolved THz-induced SHG modulation

Figure 3. Polarization resolved THz-induced SHG modulation: a. The angle α\alpha between the NIR pulse polarization and the BBO ordinary principal axis is varied while the THz field is kept parallel to the ordinary axis. b. The SHG intensity as a function of α\alpha is fit by cos⁡4α\cos^4 \alpha curve (black dashed line), which is the expected angular dependence in type-I SHG. c. THz-induced SHG modulation ΔISH(α)\Delta I_{\mathrm{SH}}(\alpha) versus α\alpha at fixed THz-NIR delay time of 2.4 ps. The orange solid curve denotes an increase of the SH intensity (ΔISH(α)>0\Delta I_{\mathrm{SH}}(\alpha) > 0) while the blue line a reduction (ΔISH(α)<0\Delta I_{\mathrm{SH}}(\alpha) < 0). d. The positive and negative modulation quadrants are interchanged by delaying the THz-NIR by one phonon half period.

Out-of-equilibrium, the SHG can in principal be modulated by two distinct effects. The dynamics can result from a time-dependent variation in the second-order nonlinear coefficient, which is proportional to the driving THz field and can be considered effectively as a third-order χ(3)\chi^{(3)} effect. Notably, under certain circumstances, this χ(3)\chi^{(3)}-driven effect has been used to describe THz-Field Induced Second Harmonic Generation (TFISH) [37, 38, 39, 40]. Alternatively, the dynamics can result from phonon-mediated electro-optic effects in the linear index of refraction (a χ(2)\chi^{(2)} effect involving THz and phonon frequencies) – and subsequent modification of the phase-matching conditions (Δk\Delta k), affecting the SHG efficiency, without inducing a time-dependent χ(2)\chi^{(2)} or χ(3)\chi^{(3)} effect, through the second-order nonlinear coefficient involving the fundamental field frequency (Secs. 5 and 6 of the Supplementary Information). This mechanism has been described as a cascade of χ(2)\chi^{(2)} processes [26, 41]. Under strong-field THz excitation, both χ(2)\chi^{(2)} and χ(3)\chi^{(3)} effects are always present, however, it can be expected that only one of the effects dominates the nonlinear response depending on the specific equilibrium phase-matching conditions.

To identify the origin of the observed dynamics, the NIR probe pulse polarization is scanned at a time delay of t=2.4t = 2.4 ps - corresponding to a local maximum of the SHG dynamics. Notability, the angular dependence of the measured SHG modulation, ΔISH(α)=ISH,THz(α)−ISH,0(α)\Delta I_{SH}(\alpha) = I_{SH,THz}(\alpha) - I_{SH,0}(\alpha), now exhibits a four-fold symmetry, with the two lobes peaking at α≈30∘\alpha\approx30^\circ and α≈330∘\alpha\approx330^\circ having a positive modulation (ΔISH(α)>0\Delta I_{SH}(\alpha) > 0), corresponding to an increase in SHG intensity (Figure 3) and vice-versa for the lobes observed at α≈150∘\alpha\approx150^\circ and α≈210∘\alpha\approx210^\circ. Further, as the temporal delay between the THz and NIR pulses is scanned over one oscillation cycle, the sign of ΔISH(α)\Delta I_{SH}(\alpha) is inverted in all lobes, passing through a common zero-crossing where no modulation is observed (ΔISH(α)=0\Delta I_{SH}(\alpha) = 0), as shown in Figure 3d.

Importantly, this angular and temporal dependence of the SHG signal in the type-I phase matched geometry is incompatible with a TFISH interpretation involving an effective χ(3)\chi^{(3)}-effect resulting from a THz-induced modulation of the second-order χ(2)\chi^{(2)} susceptibility. Evidence of this inconsistency is obtained from comparative measurements of SHG modulation in BBO where the phase-matching conditions are not met (see Supplementary Figure S5). Here, due to the limited coherence length, the the χ(3)\chi^{(3)} effects to be isolated, as electro-optically induced changes in the index of refraction ellipsoid are negligible in the overall SHG response. In the phase-mismatched conditions, the χ(3)\chi^{(3)}-effects dominate, resulting in a polarization dependence in the SHG modulation with features fundamentally different from those observed in the phase matched case. Additionally, the equilibrium signal strength and magnitude of the modulated intensity is orders of magnitude weaker (see Secs. 7 and 8 of the Supplementary Information for full explanation). Therefore, in the current experiment, the strength of the effect and the 4-fold symmetry leads to the conclusion that the dynamic effect is dominated by the phonon-mediated alteration of the type I phase-matching conditions – a cascaded 2nd-order effect.

Supplementary Fig. 5, polar plots and experimental geometries

Figure S5. Supplementary Fig. 5. Isolating the TFISH contribution to THz-driven SHG modulation. a, Off–phase-matched geometry: the NIR and THz pulses propagate along the crystal cc axis; SHG is collected along yy. b, Phase-matched SHG geometry. c, Static “off-phase-matched” SHG: experimental data and fit curve with the function: ISHG∝∣d22Ex2−d22Ey2∣2I_{\mathrm{SHG}} \propto\lvert d_{22}E_x^2-d_{22}E_y^2\rvert^2. d, Static phase-matched SHG: experimental data with fit: ISHG∝∣d22Ey2∣2I_{\mathrm{SHG}} \propto\lvert d_{22}E_y^2\rvert^2, where the Ex2E_x^2 is not present because not phase-matched. e, THz-induced SHG modulation in the off–phase-matched case: data and fit: ΔISHG∝a1Ex4+a2Ey4+a3Ex2Ey2+a4Ex3Ey+a5ExEy3\Delta I_{\mathrm{SHG}} \propto a_1E_x^4+a_2E_y^4+a_3E_x^2E_y^2+a_4E_x^3E_y+a_5E_xE_y^3 modelling TFISH-driven modulation of the χ(2)\chi^{(2)} tensor components, see Ref.[14]. The a3a_3 is dominant and produces lobes oriented at 45∘45^\circ; a4a_4 and a5a_5 are over one order of magnitude smaller. The maximum SHG intensity modulation is below 1% at the maximum THz field. f, THz-induced SHG modulation under phase-matched conditions: the angular symmetry differs qualitatively from panel e and cannot be described by TFISH alone.

Analytical model of the cascaded χ(2)\chi^{(2)} dynamics

Resonant excitation of the EE phonon mode by the THz pulse displaces the (B3O6)3−(\mathrm{B}_3\mathrm{O}_6)^{3-} anionic rings far from equilibrium positions (Figure 1a), thereby inducing an electro-optic modulation of the refractive index along the crystal’s principal axes (x′x^\prime, y′y^\prime, z′z^\prime; see Figure 1a). Under equilibrium conditions, BBO is a negative uniaxial crystal (nx′=ny′≠nz′n_{x^\prime}=n_{y^\prime}\ne n_{z^\prime}), but the phonon excitation breaks its axial symmetry, leading to biaxial behaviour where all three refractive indices become distinct (nx≠ny≠nzn_x\ne n_y\ne n_z) – see § 15 of the Supplementary Information for details. The dynamical refractive indices along the xx and yy axes in the crystal-frame become: nx′THz=nx′+δnn_{x^\prime}^{\mathrm{THz}}=n_{x^\prime}+\delta n and ny′THz=ny′−δnn_{y^\prime}^{\mathrm{THz}}=n_{y^\prime}-\delta n, where the magnitude of the phonon-induced refractive index modulation is:

δn=nx′,y′3r22Q/2(2)\delta n = n_{x^\prime,y^\prime}^{3}r_{22}Q/2 \tag*{(2)}

Here, QQ is the phonon amplitude and r22r_{22} is the phonon-optic coupling coefficient, relating the modulation of the refractive index to the phonon amplitude. To interpret the index of refraction ellipsoid in the context of phase matching, which governs the efficiency of the SHG χ(2)(2ω,ω,ω)\chi^{(2)}(2\omega,\omega,\omega)-process, a projection normal to the direction of propagation in the laboratory frame is taken from which the indices along the ordinary and extraordinary axes, known as the principal axes, are found.

In the experiment, the BBO crystal was cut for optimal type-I phase matching at equilibrium (SHG at 800 nm), with its crystal axes oriented at θ=29.3∘\theta=29.3^\circ and φ=90∘\varphi=90^\circ (following the conventions in Ref. 42). The cut angle defines the direction of propagation in the crystal system and ensures that the refractive index of the ordinary fundamental wave (non_o) matches that of the extraordinary second harmonic wave (nen_e). In the dynamical case, the phonon-induced modulation of the index of refraction ellipsoid (δn\delta n in the crystal coordinate system) leads to a rotation of the ordinary and extraordinary axes in the laboratory frame by an angle γ=κ(θ)r22Q\gamma=\kappa(\theta)r_{22}Q, as illustrated in Figure 4a. Here, κ(θ)\kappa(\theta) is a geometrical parameter that depends on the crystal orientation θ\theta.

Figure 4. Theoretical modeling of SHG modulation and FDTD simulation for the phonon propagation. a. The BBO principal axes with corresponding refractive indices non_o and nen_e undergo a rotation in the laboratory frame by an angle γ0\gamma_0 due to the lattice displacement QQ. b. Spatio-temporal evolution of the normalized (norm.) THz-driven phonon amplitude Q(z,t)Q(z,t) inside the bulk BBO crystal calculated using FDTD method. Top panel: FDTD results for the time-integrated phonon amplitude QQ as a function of depth in the crystal (red curve) and the exponential fit of the decay, Q(z)=Q0e−z/δphQ(z)=Q_0e^{-z/\delta_{ph}} (black dotted curve), to estimate the phonon penetration depth δph\delta_{ph}. c. The phonon-driven rotation of the principal axes, γ0\gamma_0, results in a polarization-dependent SHG modulation with cos⁡3(α)sin⁡(α)\cos^3(\alpha)\sin(\alpha) angular dependence (Eq. (2)). Dashed black lines are fits (based on Eq. (3)) of the measurements as a function of the normalized THz field

While the resonant phonon excitation rotates the ordinary and extraordinary axes, impacting the phase-matching conditions in the SHG process, the magnitudes of the ordinary and extraordinary refractive indices – apart from a third-order correction – remain nearly unchanged. (Derivations of the modulation in the refractive index δn\delta n, the effective rotation angle γ\gamma of the principal axes, and all relevant details are given in § 15 and § 16 of Supplementary Information).

To quantitatively model the dynamics, it is also essential to consider the spatio-temporal propagation of the phonon inside the BBO crystal. The phonon amplitude decays exponentially along the direction of propagation. This spatial dependence of the phonon amplitude can be described by Q(z)=Q0e−z/δphQ(z)=Q_0e^{-z/\delta_{ph}}, where Q0Q_0 is the initial phonon amplitude at the crystal surface, and δph\delta_{ph} is its penetration depth. By accounting for the phonon-induced rotation of the ordinary and extraordinary optical axes, as well as the penetration depth of the phonon, the variation in SHG intensity as a function of the input polarization angle α\alpha (at fixed THz-NIR delay) is the given by:

ΔISH(α)=4ISH,0(0)γ0Λ(δph,L)Lcos⁡3αsin⁡α,(3)\Delta I_{SH}(\alpha)=4I_{SH,0}(0)\gamma_0\frac{\Lambda(\delta_{ph},L)}{L}\cos^3\alpha\sin\alpha, \tag*{(3)}

where ISH,0(0)I_{SH,0}(0) is the equilibrium SH intensity at α=0\alpha=0, and γ0=κ(θ)r22Q0\gamma_0=\kappa(\theta)r_{22}Q_0 is the maximum rotation angle of the optical axes induced by the phonon amplitude Q0Q_0 at the fixed delay. The 4-fold dependence on the input polarization, ∼cos⁡3αsin⁡α\sim\cos^3\alpha\sin\alpha, is consistent with the alternating sign and angular symmetry observed experimentally, and the strength of the effect depends on the BBO thickness, LL and Λ(δph,L)\Lambda(\delta_{ph},L), which includes the phonon penetration depth.

strength at a fixed delay of 2.4 ps between NIR probe and THz pump-pulse. d. Estimate of γ0\gamma_0 as a function of peak THz pulse field strength using the fits to polarization resolved SHG modulation measurements.

A key parameter is the phonon penetration depth δph\delta_{ph}, which can be estimated independently using finite-difference time-domain (FDTD) methods to calculate the spatio-temporal dynamics of the phonon amplitude. With FTIR measurements of the linear optical properties of the BBO crystal, EOS measurements of the THz driving electric field, and the phonon equation of motion, the normalized amplitude Q(z,t)Q(z,t) of the phonon at ωQ\omega_Q is simulated using FDTD and shown as a function of the crystal depth and time in Figure 4b. Integrating the phonon amplitude in time and fitting the depth-dependent profile with an exponential decay (Figure 4c) yields a phonon penetration depth of δph=30.2\delta_{ph} = 30.2 μm.

Using δph\delta_{ph}, angularly dependent SHG measurements (at fixed time delay) for THz pulses of varying intensity are then fit with Eq. (3), keeping ISH,0(0)I_{\mathrm{SH},0}(0) as a constant parameter. These fits yield γ0\gamma_0 as a function of the THz field amplitude (Figure 4d-Figure 4e). The rotation of the ordinary and extraordinary axes, γ0\gamma_0, is found to vary linearly with applied THz field amplitude, consistent with the linear dependence of the IR-active phonon amplitude Q0Q_0 on the THz field strength. The value of γ0\gamma_0 achieved for the maximum THz field strength in the experiment is γ0,max=0.216\gamma_{0,\mathrm{max}} = 0.216 rad.

Microscopic Mechanisms of Phonon-Driven Nonlinear Optical Modulation

The polarization-resolved measurements in conjunction with the analytical analysis yield the rotation of the optical axes due to the resonant THz-phonon excitation. This, in turn, permits quantification of the associated modulation of the refractive indices δn\delta n in the crystal coordinate system, but does not immediately allow for independent retrieval of the electro-optic coefficient r22Er^{E}_{22} or a determination of the phonon displacement QQ. At the microscopic level, the effect of the THz-induced coherent lattice vibrations on the local electronic environment within the anionic units remains hidden, as does the electro-optic coefficient r22Er^{E}_{22}.

To access this information density functional theory (DFPT) can be used in the frozen-phonon approximation, to calculate the electronic band structure [32] and, by extension, the dielectric function corresponding to a lattice displacement Q0Q_0 along the phonon coordinates. The resulting dielectric function can then be used to extract absolute values for the refractive indices nx′THzn^{\mathrm{THz}}_{x'} and ny′THzn^{\mathrm{THz}}_{y'} in the (x′,y′,z′)(x',y',z') coordinate system of the crystal as a function of the phonon displacement Q0Q_0 (Figure 5a). Starting from equilibrium (Q0=0Q_0 = 0), both indices are found to be nx′=ny′=1.705n_{x'} = n_{y'} = 1.705, in good agreement with experimental measurements and previous ab-initio calculations[32]. With a finite phonon amplitude Q0Q_0, the refractive indices respond linearly as expected: nx′THzn^{\mathrm{THz}}_{x'} increases, while ny′THzn^{\mathrm{THz}}_{y'} decreases with increasing Q0Q_0 (see Figure 5a). Using the DFPT-calculated dependence of δn\delta n on Q0Q_0, the phonon-optic coupling coefficient r22r_{22} can be estimated from the relation, r22=2δn(Q0)/nx′3Q0r_{22}=2\delta n(Q_0)/n_{x'}^3Q_0, yielding a value of r22=0.0049r_{22}=0.0049 Å−1^{-1}(amu)−1/2^{-1/2}. This coefficient is especially relevant, as it can be used to calculate the maximum phonon amplitude at maximum THz field strength using the experimentally determined rotation of the optical axes γ0,max\gamma_{0,max} at the crystal surface, Q0,max=γ0,max/(κr22)=0.67Q_{0,max}=\gamma_{0,max}/(\kappa r_{22})=0.67 Å(amu)1/2^{1/2} (Figure 5a). This value is consistent with expectations based on experiments targeting different effects in various materials that are nevertheless similar in that they also require THz excitation of large amplitude phonons. In these independent experiments, performed at X-ray free-electron lasers, it was possible to make a direct structural determination of the phonon amplitude[35, 43].

Frozen-phonon DFPT calculation results and phononic-modulation spatial charge-density modulation

Figure 5. Frozen-phonon DFPT Calculation Results and Phononic-modulation Spatial Charge-density modulation. a. Computed refractive indexes nx′THzn_{x'}^{\mathrm{THz}} and ny′THzn_{y'}^{\mathrm{THz}} along the x′x' and y′y' axes of the unit cell as a function of Q0Q_0. Vertical dashed

Finally, the electro-optic coefficient r22E(ω)r_{22}^{E}(\omega) can be deduced from phonon-optic coupling coefficient r22r_{22} using the conventional linear relation between the THz field and QQ in the frequency domain, which leads to the following expression:

r22E(ω)=r22E(0)(ωQ2−ω2)(ωQ2−ω2)2−(Γω)2,(4)r_{22}^{E}(\omega)=r_{22}^{E}(0)\frac{\left(\omega_{Q}^{2}-\omega^{2}\right)}{\left(\omega_{Q}^{2}-\omega^{2}\right)^{2}-(\Gamma\omega)^{2}}, \tag*{(4)}

where ωQ\omega_Q and Γ\Gamma are the phonon linewidth and the r22E(0)r_{22}^{E}(0) is the DC electro-optic coefficient (see § 19 of the Supplementary Information for details). Approaching the phonon resonance, the r22E(ω)r_{22}^{E}(\omega) peak value is 55 pm/V at ω∗=ωQ2−ΓωQ=4.22\omega_*=\sqrt{\omega_Q^2-\Gamma\omega_Q}=4.22 THz. Notably, the spectrum of the time-dependent SHG modulation dynamics closely matches the EO coefficient, with shared maxima just below ωQ\omega_Q (dashed line Figure 5b). This frequency dependence provides further evidence that the SHG modulation arises from the electro-optic effect. Indeed, in phonon-mediated EO processes, the response is governed not by the imaginary part of the dielectric function, ε2(ω)\varepsilon_2(\omega), which peaks at the phonon frequency ωQ\omega_Q, but by the real part, ε1(ω)\varepsilon_1(\omega), whose maximum occurs at ω∗\omega_*[44].

lines, which delimit the shaded light-blue region, indicate the maximum amplitude estimated in the experiment, Q0,max⁡∼0.67 A˚(amu)1/2Q_{0,\max}\sim0.67\,\text{\AA}(\mathrm{amu})^{1/2} at 2.4 ps. b. Frequency dependence of the electro optic coefficient r22E(ω)r^{E}_{22}(\omega), shown together with the spectrum of the SHG-modulation dynamics for comparison. c. Planar modulation of the normalized (norm.) charge density Δρ(x′,y′)\Delta\rho(x',y') in the unit cell of β\beta-BaB2_2O4_4 induced by phonon displacement for Q0,max⁡Q_{0,\max} and −Q0,max⁡-Q_{0,\max}. The electronic density is significantly affected by the O atomic displacement of the B3_3O6_6 unit.

The peak value of the vibrational electro-optic coefficient can also be estimated experimentally by combining the analytical expression for the SHG modulation with the phonon equation of motion (see § 19 of the Supplementary Information). Using this approach and the broadband THz waveform applied in the experiment, a peak value of r22Er^{E}_{22} of ∼140\sim140 pm/V is obtained in relatively good agreement with that obtained by DFPT (see Supplementary Figure S7). To exclude the involvement of other phonon modes, additional experiments were performed using a narrowband THz pulse, which yields a similar value of ∼170\sim170 pm/V (see Supplementary Figure S8). The discrepancy of less than a factor of 3 between theory and experiment, which exists in both the broadband and narrowband case could therefore stem from the calibration of the absolute THz field amplitude, phonon parameters, and on the optical constants involved in the analysis—including their large frequency dispersion near the phonon resonance and dependence on the geometrical parameters in the experiment.

Experimental data and calculated relative SH modulation

Figure S7. Experimentally measured ΔIrel\Delta I_{\mathrm{rel}} and calculated one using Eq. (58).

Four-panel plots showing THz pump waveform, amplitude spectra, THz-driven modulation, and corresponding amplitude spectra

Figure S8. SHG modulation with a narrowband THz pulse. a, Time-domain THz pump waveform measured by EOS. b, Amplitude spectra corresponding to a. c, THz-driven ΔIrel\Delta I_{\mathrm{rel}} and calculated one using Eq. (58). d, Amplitude spectra corresponding to curves in c.

On the other hand, the theoretical treatment could also be slightly flawed, as extrapolating the theoretical electro-optic coefficient to the DC limit gives r22E(0)=6.6r^{E}_{22}(0)=6.6 pm/V, which is larger than the commonly reported experimental DC value of approximately 2.5 pm/V for BBO [45].

Regardless of the small discrepancy that remains, relative agreement between the experimental and theoretically determined electro-optic coefficient lends credence to the DFPT calculations, by which additional parameters can be obtained in the vicinity of the phonon resonance. In particular, the interaction between the phonon and the local electronic environment can be explored computationally. Here, DFPT-calculated differential charge density maps Δρ(x′,y′)\Delta\rho(x',y') confirm that the phonon mode predominantly drives cooperative displacements of the anionic [B3_3O6_6]3−^{3-} rings within the lattice plane, while the Ba2+^{2+} cations remain largely stationary (Figure 5c). This selective atomic motion produces a marked redistribution of charge density around the planar [B3_3O6_6]3−^{3-} units. The electronic density around the oxygen atoms is largely influenced by the phonon dynamics, suggesting that the local electronic environment of these atoms plays a central role in modulating the refractive indices nx′THzn^{\mathrm{THz}}_{x'} and ny′THzn^{\mathrm{THz}}_{y'}.

Discussion

Coherent lattice dynamics driven by intense THz pulses tuned to resonance with an infrared-active phonon mode have been shown to strongly modulate SHG in BBO on ultrafast timescales with the phase of the driving optical field. Analytical modeling, and first-principles DFPT calculations,, provide both a macroscopic overview and a comprehensive description of the physics at the microscopic level. At the atomic level, DFPT calculations indicate that the THz driven phonons induce acentric displacements of the B3O6\mathrm{B}_3\mathrm{O}_6 anionic rings in the BBO unit cell. These displacements are phase-locked to the phonon dynamics (and driving field) and estimated to be as large as ∼0.5 A˚−1 (amu)−1/2\sim0.5\ \mathrm{\AA}^{-1}\,(\mathrm{amu})^{-1/2}, leading to a redistribution of local charge around the oxygen atoms. The impact of the transient charge redistribution on the refractive index ellipsoid of BBO is quantified through DFPT calculations. Additionally, the phonon-mediated electro-optic rotation of the ordinary and extraordinary optical axes that governs SHG is quantified using an analytical model. The observed SHG-modulation spectrum and angular dependence are consistent with the predicted EO response, exhibiting a resonance immediately below the IR-active phonon frequency. From a macroscopic perspective, it is the link between the rotation of the optical axes – which are defined at equilibrium by the experiment geometry – and the optical phase-matching conditions that permit strong modulation of the SHG signal in the bulk material that exceeds 30%.

Excellent agreement between theoretical predictions and experimental observations show that an approach that combines microscopic computational and macroscopic analytic treatment can be leveraged to capture the interplay between phonons and electronic states underlying nonlinear optical phenomena in solids[30]. These findings also establish DFPT as a powerful framework in which the optical properties of out-of-equilibrium crystal structures can be accurately evaluated.

Beyond fundamental insight, these findings may impact the design of new materials with tailored transient nonlinear responses, advancing ultrafast photonics and THz technologies. The ability to dynamically modulate SHG with high efficiency and precision through lattice-driven mechanisms opens new opportunities for ultrafast optical signal processing, high-speed data communication, and nonlinear optical switching. By leveraging lattice degrees of freedom[46, 47], this work introduces a resonant pathway for ultrafast, high-efficiency control of nonlinear optical responses in three-dimensional bulk materials. Further, by general application of DFPT as a predictive tool for the optical-electronic-lattice interaction, combined with analytical treatment and experiments in the future, it can be expected that a broad range of nonlinear optical responses in other materials may be optimised and tailored for specific functionalities.

Methods

Experimental details

Intense THz pulses were generated through optical rectification in a DSTMS organic crystal using the signal beam of an optical parametric amplifier, driven by 50 fs, 800 nm pulses at a 100 Hz repetition rate. The SHG dynamics in β\beta-BaB2_2O4_4 were investigated using a time-delayed replica of the 800 nm pulses. The experimental setup is shown in Supplementary Fig. 1; further details are provided in § 11 of the Supplementary Information.

All experiments were performed at room temperature. The sample used was a commercially available β-BaB2_2O4_4 single crystal (5 mm × 5 mm × 0.3 mm), oriented at θ=29.3∘\theta= 29.3^\circ and ϕ=90∘\phi= 90^\circ. Crystal orientation and crystallographic quality were verified by Laue X-ray diffraction (Supplementary Figure S4).

Laue diffraction and sample orientation

Figure S4. Laue diffraction and sample orientation: a, Laue diffraction pattern of the single crystal 300 μ\mum thick BBO sample (θ=29.3∘\theta= 29.3^\circ, ϕ=90∘\phi= 90^\circ) used in the experiment to verify the experimental geometry. The THz electric field is aligned along the aa-axis. In the Laue measurement, the crystal was rotated by ∼−θ∘\sim-\theta^\circ around the aa-axis to confirm the R3cR3c symmetry. Red dots indicate simulated Laue diffraction peaks generated using Laue diffraction software. b, Crystal unit cell from DFPT calculations, shown in crystallographic coordinate system x′x', y′y' (in-plane) and z′z' (out-of-plane). The x′x'-axis corresponds to the crystallographic aa-axis. The EE-mode at 144 cm−1^{-1} is doubly degenerate, with branches E1E_1 and E2E_2 having dipole moment directions indicated by the orange arrows.

FTDT Spatio-Temporal Simulations of Phonon Propagation

The THz-driven phonon propagation along the crystal depth has been simulated in time domain by computing Maxwell’s equations based on FTDT method[48]. The material optical response along the o-axis at THz frequencies was modeled including the dielectric function of IR-active phonons, whose parameters have been experimentally determined using FTIR spectroscopy (see § 13 of the Supplementary Information). The dominant feature is the phonon at ωQ\omega_Q, but we include additional 13 phonons for a faithful description of the optical response (phonon parameters are listed in Supplementary Table 1). For each ii-th phonon mode, the equation of motion in the time domain is given by[35]

Q¨i+2ΓiQ˙i+ωQ,i2Qi=ZETHz(t)(5)\ddot{Q}_i + 2\Gamma_i\dot{Q}_i + \omega_{Q,i}^{2}Q_i = Z E_{THz}(t) \tag*{(5)}

Where Γi\Gamma_i is the damping rate, ωQ,i\omega_{Q,i} is the phonon frequency and Z=ωQ,iϵ0(ε0−ε∞)Z = \omega_{Q,i}\sqrt{\epsilon_0(\varepsilon_0 - \varepsilon_\infty)} is the phonon effective charge, being ε0\varepsilon_0 and ε\varepsilon the dielectric constant at the low and high-frequency limit respectively and ϵ0\epsilon_0 the vacuum permittivity.

Lattice equations and the electric field component from Maxwell’s equations in media are related by displacement field as:

D=ϵ0ε∞ETHz+∑iωQ,iϵ0(ε0−ε∞)Qi(6)D = \epsilon_0\varepsilon_\infty E_{THz} + \sum_i \omega_{Q,i}\sqrt{\epsilon_0(\varepsilon_0 - \varepsilon_\infty)} Q_i \tag*{(6)}

By FTDT calculation we numerically solve the above equation at points of the grid through leapfrog time-step scheme. In the simulation, the THz electric field ETHz(t)=E0cos⁡(ω0t)e−t22σ2E_{THz}(t) = E_0 \cos(\omega_0 t)e^{-\frac{t^2}{2\sigma^2}} is a Gaussian field having a width σ=100\sigma= 100 fs and a center frequency ω0=5\omega_0 = 5 THz. The plot of the spatiotemporal propagation of the phonon at ωQ\omega_Q is show in Figure 4b.

DFPT Calculations of Linear and Nonlinear Optical Properties Under Phonon Displacement

To quantify the linear and nonlinear optical responses due to phonon displacement, first-principles calculations were perfomed in the framework of density-functional perturbation theory (DFPT) implemented using PHonon package[49] in Quantum ESPRESSO[50, 51], with PBE SG15 Optimized Norm-Conserving Vanderbilt (ONCV) pseudopotentials[52, 53].

DFPT calculations were conducted within the structural unit cell [42 atoms, R3c (No. 161) space group] of the BBO crystal at room temperature has been taken from Ref. [54]. For the Brillouin zone sampling, 4x4x4 k-point grid and a plane-wave energy cutoff of 80 Ry have been used and the structural parameters have been relaxed until the force acting on each atom was less than 0.0025 eV/Å and the pressure was below 0.5 kbar. Though phonon energies and corresponding displacement vectors at the Brillouin zone center, the macroscopic dielectric tensor, and Born effective charges were computed.

The refractive index ellipsoid were computed as the square-root of the macroscopic dielectric tensor, calculated with DFPT in the long-wavelength limit (q=0q = 0). This is valid in non-absorbing materials, in which the refractive index dispersion is small, as in the case of BBO in the optical optical region.

At zero displacement along the phonon coordinates (Q0=0Q_0 = 0), the DFPT-calculated dielectric tensor in the crystal frame is:

ξ0=(ξx′x′0000ξx′x′0000ξz′z′0).(7)\xi_0 = \begin{pmatrix} \xi_{x'x'}^0 & 0 & 0 \\ 0 & \xi_{x'x'}^0 & 0 \\ 0 & 0 & \xi_{z'z'}^0 \end{pmatrix}. \tag*{(7)}

which corresponds to a refractive nx′=ξx′x′=1.705n_{x'} = \sqrt{\xi_{x'x'}} = 1.705 and nz′=ξz′z′=1.705n_{z'} = \sqrt{\xi_{z'z'}} = 1.705. Previous DFPT studies have reported a refractive index of nx′=ny′=1.692n_{x'} = n_{y'} = 1.692 at 800 nm, see Ref.[32], which is within 0.5% of our DFPT-calculated value. This close agreement validates the use of the long-wavelength approximation in our approach. Additionally, the small discrepancy might by due to the fact our DFPT method, Ref.[49], includes local-field effects and goes beyond the independent-particle approximation employed in Ref.[32]. Experimentally, the refractive index has been measured to be nx′=1.66n_{x'} = 1.66, which is within 3% of our calculated results.

The dielectric tensor as a function of the phonon amplitude, was computed with the same approach but assuming structures displaced along the relevant IR-active phonon mode displacement vectors.

The dielectric tensor as a function of the phonon amplitude has the following form:

ξQ=(ξx′x′ξx′y′0ξx′y′ξx′x′000ξzz)→R(π2)(ξx′x′000ξx′x′000ξz′z′)(8)\xi_Q = \begin{pmatrix} \xi_{x'x'} & \xi_{x'y'} & 0 \\ \xi_{x'y'} & \xi_{x'x'} & 0 \\ 0 & 0 & \xi_{zz} \end{pmatrix} \xrightarrow{R\left(\frac{\pi}{2}\right)} \begin{pmatrix} \xi_{x'x'} & 0 & 0 \\ 0 & \xi_{x'x'} & 0 \\ 0 & 0 & \xi_{z'z'} \end{pmatrix} \tag*{(8)}

which diagonalized with a x′−y′x' - y' plane π2\frac{\pi}{2} rotation around the z′z' axis, as in the analytical model (see § 16 of the Supplementary Information). For the dielectric tensor in Eq. (7), the refractive indeces as a fuction of phonon amplitude were calculated.

For the electronic density calculations in Figure 4b, a plane-wave kinetic energy cutoff of 320 Ry has been used. To evaluate the nonlinear optical coefficients of BBO (see Supplementary Figure S6) finite-field approach was used[55]. In this method, the electric field is explicitly included in the DFPT calculations, and the nonlinear coefficients are obtained numerically by computing the second derivative of the polarization with respect to the applied electric field.

DFT calculated modulation of nonlinear coefficients

Figure S6. a, DFT calculated modulation of the nonlinear coefficients d22d_{22} as a function of lattice displacement Q0Q_0. b, Modulation of the nonlinear coefficients Δd26\Delta d_{26}, Δd35\Delta d_{35}, Δd25\Delta d_{25}, and Δd36\Delta d_{36} as a function of Q0Q_0. At the equilibrium structure, these coefficients are approximately zero.

Acknowledgments. The authors thank the PSI-LNO GL group for support during the experiments and R. Li for helpful discussions. This work was supported by the Swiss National Science Foundation (SNSF) under grant number 10001644 and SNSF Spark grant number 221173.

Author Contributions Statement. F.G. and C.V. conceived the project and designed the experiment. F.G., C.V. and G.M. performed the THz spectroscopy measurements. F.G., C.V., A.T., G.N. and L.S. carried out the THz experiments. F.G. performed the FDTD simulations. N.C. carried out the DFPT calculations. N.F., N.C. and F.G. developed the analytical model and interpreted the data. F.G. prepared the figures. F.G., A.L.C. and C.V. wrote the manuscript with input from all authors. All authors discussed the results.

Competing Interests Statement. The authors declare no competing interests.

Supplementary Information: All-optical control of second-harmonic generation in β\beta-BaB2O4\mathrm{BaB}_2\mathrm{O}_4 via coherent, terahertz-driven acentric lattice displacement

Flavio Giorgianni; Nicola Colonna; Gabriel Nagamine; Leonie Spitz; Guy Matmon; Alexandre Trisorio; Nicolas Forget; Carlo Vicario; Adrian L. Cavalieri

Experimental Setup and THz Field Characterization

A schematic of the experimental geometry is shown in Supplementary Figure 1a. The measurements are made using a 100 Hz, 20 mJ, 50 fs Ti:Sa laser system coupled to an optical parametric amplifier (OPA). The output energy of the OPA signal beam was approx. 3.9 mJ per pulse. The OPA signal, at a wavelength of 1.5 μ\mum, was employed to generate intense single-cycle THz pulses through optical rectification in a large-aperture DAST crystal, with clear aperture exceeding 10 mm in diameter. The OPA signal beam diameter was approximately 9 mm at the crystal surface. Metallic multi-mesh filters were inserted into the THz optical path to isolate the THz pulses from the residual OPA beam and to shape the THz pulse spectrum.

Experimental setup and THz field characterization

Supplementary Figure 2. Experimental Setup and THz field characterization. a, Setup for THz generation and modulation of second harmonic in BBO. LS: linear stage; WP: half-wave plate, DM, dichroic mirror, SH-BPF: band-pass filter for SH signal, T-Filter: THz spectral filters, TPs: THz polarizers, BBO: BBO crystal sample. b, THz waveforms for 10 THz LPF, 3 THz LPF, and 4.2 THz HPF+10 THz LPF. c, Intensity profile ∣E(t)∣2|E(t)|^2 of the THz pulse with the 10 THz LPF. d, Horizontal and e, vertical THz beam transverse profile of THz pulse at the sample position with the 10 THz LPF. Orange and light blue solid curves are the experimental data measured with a THz camera while black dotted lines represent a Gaussian fit. The estimated FWHM are 90 μ\mum and 85 μ\mum, respectively.

The THz beam was expanded and then focused onto the β\beta-BaB2O4\mathrm{BaB}_2\mathrm{O}_4 (BBO) crystal sample using three off-axis parabolic mirrors with respective focal lengths of 1", 6", and 2" (see Supplementary Figure 1a), which permits tight focusing of the THz beam onto the sample. A pair of wire-grid THz polarizers are used to vary the THz electric field strength at the sample (TPs in Supplementary Figure 1a). The rotation angle of first polarizer is adjustable, while the second polarizer was fixed to transmit only the vertically polarized component.

For the broadband THz-pump experiments (see Figure 1c–Figure 1f in the main text), isolation of the OPA signal beam was achieved using a filter set consisting of two 20 THz low-pass filters (LPFs) and one 10 THz LPF (all from QMC Ltd). This configuration provided a measured extinction ratio of ∼107\sim10^7 for the residual 1.5 μ\mum OPA pulse. To selectively tailor the THz pump spectrum relative to the phonon resonance (see Figure 2a of the main manuscript), the 10 THz LPF in the filter set was replaced as follows: (i) for isolating frequency components below the phonon resonance, a 3 THz LPF was used; (ii) for isolating components above the resonance, a combination of a 4.2 THz high-pass filter (HPF) followed by a 10 THz LPF was used. The temporal profile of the THz pulse at the sample position was measured using electro-optic sampling (EOS) with a (110) GaP crystal of 200 μ\mum thickness. The THz waveforms measured by EOS for these three filter configurations are shown in Supplementary Fig. 1b.

The broadband THz electric field strength was estimated from measurement of the temporal profile of the THz pump, the THz pulse energy, and the focused THz spot size. The intensity temporal profile of the broadband pump pulse computed as the square of the THz electric field waveform is shown in Supplementary fig:1c. The transverse spatial profile of the THz beam was measured using a THz micro-bolometric camera (NEC IRVT0830). The THz energy per pulse was measured with a Gentec THZ12D-3S-VP-D0 energy meter.

The THz pulse in Figure 1 of the main manuscript has the following parameters: THz energy/pulse: 3.5 μ\muJ, focused spotsize (FWHM): 90 μ\mum from Gaussian fit of the beam profile (see Supplementary Fig. 1d-e).

A THz field strength, ETHz≈8.5E_{THz} \approx8.5 MV/cm, was estimated using the following expression1:

ETHz=z0Wπw2∫−∞∞g2(t) dt(9)E_{THz} = \sqrt{\frac{z_0 W}{\pi w^2 \int_{-\infty}^{\infty} g^2(t)\,dt}} \tag*{(9)}

where z0=377z_0 = 377 Ω\Omega is the free space impedance, WW is the THz energy/pulse, ww is the beam waist and g2(t)g^2(t) is the temporal intensity profile of the THz pulse with peak value normalized to 1, (see Supplementary Fig. 1b). The temporal profile g(t)g(t) was measured by EOS and the integral, ∫−∞∞g2(t)dt=0.086\int_{-\infty}^{\infty} g^2(t)dt = 0.086 ps. The THz field strengths for pump spectra above and below the phonon resonance were determined from EOS measurements. The THz pulse below the phonon resonance (obtained using the 3 THz LPF), has a peak electric field that is 4 times smaller than the peak electric field of the broadband THz pulse obtained with the 10 THz LPF alone (see Supplementary Fig. 1b). Above resonance (implemented using the 4.2 THz HPF in combination with the 10 THz LPF), the peak THz field amplitude was 0.16 times the peak field of the broadband THz pump pulse.

Evidence of THz-field-induced second harmonic intensity modulation in BBO

To verify that the observed SH intensity modulation does not result from variations in the intensity of the transmitted fundamental frequency (FF) beam potentially due to THz-induced changes in the Fresnel coefficients, the transmitted NIR pulse intensity was measured as a function of the time delay relative to the THz pump pulse (see Supplementary Fig. 2). No significant modulation is observed in the NIR pulse intensity, confirming that the SHG modulation originates from the resonant THz-driven phonon excitation.

Experimental measurement of THz-field-induced modulation of SH and FF intensities

Supplementary Figure 2. Experimental measurement of THz-field-induced modulation of SH and FF intensities. The absence of significant modulation in the FF beam intensity confirms that the observed modulation effect originates from the SH process.

THz Spectroscopy of BBO and Laue X-ray Diffraction

To characterize the BBO infrared-active phonons, optical transmission measurements in the THz regime were performed using Fourier transform infrared (FTIR) spectroscopy at the IR beam line of the Swiss Light Source at PSI. For these measurements, a 30-μ\mum-thick reference BBO crystal was used with the same orientation (θ=29.3∘\theta= 29.3^\circ and ϕ=90∘\phi= 90^\circ) as the crystal in the main THz experiment, but with reduced thickness to maximize the THz transmission. Supplementary Figure 3 shows the transmission spectra in the THz range for light polarised along the ordinary-axis, which is consistent with previously reported work[2, 3, 4].

The real ε1(ω)\varepsilon_{1}(\omega) and imaginary part ε2(ω)\varepsilon_{2}(\omega) of complex dielectric function, depicted in Figure 1d of the main manuscript, were found by fitting the optical spectra using the Kramers-Kronig (KK) consistent dielectric function model with Reffit software[5]. The fit procedure for the transmission spectrum was based on the conventional dielectric function:

ε(ω)=ε∞+∑iωp,i2ωo,i2−ω2−iΓiω(10)\varepsilon(\omega) = \varepsilon_{\infty} + \sum_{i} \frac{\omega_{p,i}^{2}}{\omega_{o,i}^{2} - \omega^{2} - \mathrm{i}\Gamma_{i}\omega} \tag*{(10)}

Where ω0,i\omega_{0,i} (cm−1^{-1}), ωp,i\omega_{p,i} (cm−1^{-1}) and Γi\Gamma_i (cm−1^{-1}) are the phonon strength (or plasma frequency), the eigenfrequency and the linewidth of the ii-th Lorentz oscillator, respectively. ε∞\varepsilon_{\infty} is the high frequency dielectric constant. The black dashed line in Supplementary Figure 3 is the fit and the corresponding parameters are listed in Supplementary Table 1. These parameters were used to model the BBO optical response in the finite-difference time-domain (FDTD) simulations (see Eq. (5) in Methods).

iiω0,i\omega_{0,i} (cm−1^{-1})ωp,i\omega_{p,i} (cm−1^{-1})Γi\Gamma_i (cm−1^{-1})iiω0,i\omega_{0,i} (cm−1^{-1})ωp,i\omega_{p,i} (cm−1^{-1})Γi\Gamma_i (cm−1^{-1})
147.1444.338.638119.6241.7214.68
255.2933.985.829144.85217.026.49
361.9133.595.1710188.78102.1034.17
471.8041.774.3411245.15176.4660.90
584.07736.306.1412340.2621.0319.23
692.2660.615.8213383.20185.155.33
7103.2967.297.2514469.3045.6547.70
ε∞=2.89\varepsilon_{\infty} = 2.89

Table 1. Supplementary Table 1. Fit parameters of ε(ω)\varepsilon(\omega) (Eq. (10)). The parameters of the phonon at ωQ1=4.32\omega_{Q1} = 4.32 THz are highlighted in blue.

Phonon resonances are found at ωQ1=4.32\omega_{Q1} = 4.32 THz (144 cm−1^{-1}); ωQ2=5.65\omega_{Q2} = 5.65 THz (189 cm−1^{-1}), ωQ3=7.34\omega_{Q3} = 7.34 THz (245 cm−1^{-1}) consistent with those in Ref.4, which show peaks at ωQ1=4.65\omega_{Q1} = 4.65 THz, ωQ2=5.65\omega_{Q2} = 5.65 THz, ωQ3=7.34\omega_{Q3} = 7.34 THz. For ωQ1\omega_{Q1}, the discrepancy of 0.3 THz between the measurements here and those reported in Ref. 4 could originate from sample variation and/or instrument resolution. The current measurements use a 30 μ\mum-thick uncoated BBO and the FTIR was performed with a resolution of 0.014 THz. While the measurements in Ref.4 were performed using a BBO crystal with a 400 nm antireflection coating (which could also affect the THz transmission) and transmission spectra were collected with a lower frequency resolution (0.05 THz). The 30μ\mum reference BBO in the current measurements was also evaluated using Raman spectroscopy, which yields EE-symmetry modes at 143 cm−1^{-1}, 187 cm−1^{-1} and 245 cm−1^{-1}, which closely matches (to within 2 cm−1^{-1}) the values reported in Ref.6. Sample quality and orientation of the 300μ\mum thick BBO used in the experiment were verified by Laue X-ray diffraction. Establishing the crystal orientation is essential for defining the crystallographic coordinate system (x′,y′,z′)(x',y',z') and its relation to the laboratory frame (x,y,z)(x,y,z), which is required for both the DFPT calculations based on BBO crystal structure and the analytical model. The Laue pattern shows that the crystallographic aa-axis corresponds to the yy of the laboratory frame (ordinary axis), which coincides with the x′x'-axis in the crystal frame since ϕ=90∘\phi= 90^\circ (Supplementary Figure S4). The crystal is R3cR3c symmetric, so the EE phonons are doubly degenerate. DFT calculations indicate that the THz-driven EE mode at 144 cm−1^{-1} has two orthogonal in-plane dipole moments: E1E_1 along x′x' and E2E_2 along y′y' (Supplementary Figure S4). When the THz pump is polarised along the a-axis, the field couples only to the E1E_1 branch and the orthogonal E2E_2 branch is not excited.

Polarization angular dependence of static second harmonic generation

For BBO, the second-order nonlinear dNLd_{NL} tensor has 3 independent coefficients[7] that are defined in the crystal coordinate system (x′x', y′y', z′z'):

dNL=(0000d15−d22−d22d220d1500d15d15d33000)(11)d_{NL} = \begin{pmatrix} 0 & 0 & 0 & 0 & d_{15} & -d_{22} \\ -d_{22} & d_{22} & 0 & d_{15} & 0 & 0 \\ d_{15} & d_{15} & d_{33} & 0 & 0 & 0 \end{pmatrix} \tag*{(11)}

where d15=0.16d_{15} = 0.16 pm/V, d22=−2.2d_{22} = -2.2 pm/V, d33∼0d_{33} \sim0.

In the laboratory coordinate system, the FF is linearly polarized and propagates in the zz-direction. The polarization is defined by the angle α\alpha of the polarization vector with respect to the vertical, yy-axis, which corresponds to the ordinary axis of the static BBO crystal in the laboratory frame. In this case:

ElabFF(α)=E0 u,(12)E^{FF}_{lab}(\alpha) = E_0\,u, \tag*{(12)}

where u=(sin⁡α,cos⁡α,0)u = (\sin\alpha,\cos\alpha,0) is the polarization vector and E0E_0 is the amplitude of the FF beam. Experimentally, the polarisation is varied using a half-wave plate.

The polarization vector in the crystal frame is found using the transformation uxtal=Ruu_{xtal} = Ru where θ\theta is the polar angle - which is the angle of the lab frame zz-axis with respect to the crystal cc-axis - and ϕ\phi is the azimuthal angle -which is the angle of the lab frame xx- and yy-axes with respect to the crystal ordinary axis (for the BBO crystal used in the experiments, θ=29.3∘\theta= 29.3^\circ and ϕ=90∘\phi= 90^\circ).

R=(−cos⁡θcos⁡ϕsin⁡ϕcos⁡ϕsin⁡θ−cos⁡θsin⁡ϕ−cos⁡ϕsin⁡θsin⁡ϕsin⁡θ0cos⁡θ)(13)R = \begin{pmatrix} -\cos\theta\cos\phi& \sin\phi& \cos\phi\sin\theta\\ -\cos\theta\sin\phi& -\cos\phi& \sin\theta\sin\phi\\ \sin\theta& 0 & \cos\theta \end{pmatrix} \tag*{(13)}

With this transformation, it follows that the effective nonlinear tensor deffd_{eff} in the laboratory frame, is:

deff=RTdNLf3→6(Ru),(14)d_{eff} = R^T d_{NL} f_{3\to6}(Ru), \tag*{(14)}

Where f3→6f_{3\to6} is the operator[8] that maps the square of the electric field components (x′,y′,z′)(x',y',z') onto (x′2,y′2,z′2,2y′z′,2x′z′,2x′y′)(x'^{2},y'^{2},z'^{2},2y'z',2x'z',2x'y') in the crystal frame.

For Type-I phase matching (o+o→eo+o\to e), the SH harmonic is polarized along the extraordinary axis (xx-axis in the lab frame). Therefore, for this component one finds:

ESHx(α)∝(sin⁡θ d15−cos⁡θsin⁡3ϕ d22)cos⁡2α E02(15)E_{\mathrm{SH}}^{x}(\alpha) \propto\left(\sin\theta\,d_{15}-\cos\theta\sin3\phi\,d_{22}\right)\cos^{2}\alpha\,E_{0}^{2} \tag*{(15)}

For this particular SH component, the nonlinear tensor elements in dd can be written as a scalar coefficient deffd_{eff}.

ESHx(α)∝deffcos⁡2α E02(16)E_{\mathrm{SH}}^{x}(\alpha) \propto d_{eff}\cos^{2}\alpha\,E_{0}^{2} \tag*{(16)}

The polarisation-dependent Type-I generated SH intensity ISH,0I_{\mathrm{SH},0} is then proportional to deff2d_{eff}^{2} and cos⁡4α\cos^{4}\alpha as experimentally observed (Figure 3b in the main manuscript).

Analytical model: THz Vibrational Modulation of Optical Properties

The THz-driven phonon modulates the refractive index ellipsoid ηi\eta_i through the electro-optic tensor rijr_{ij}. The BBO crystal structure is 3m3m, therefore the electro-optic tensor in the Voigt notation in the crystal coordinate system (x′,y′,z′)(x',y',z') is written as[8]:

rijEO=(0−r22r130r22r1300r330r510r5100−r2200)(17)r_{ij}^{EO}= \begin{pmatrix} 0 & -r_{22} & r_{13}\\ 0 & r_{22} & r_{13}\\ 0 & 0 & r_{33}\\ 0 & r_{51} & 0\\ r_{51} & 0 & 0\\ -r_{22} & 0 & 0 \end{pmatrix} \tag*{(17)}

With an applied field, the components of the index of refraction ellipsoid undergo a variation Δηi=rijEO⋅Ej\Delta\eta_i=r_{ij}^{EO}\cdot E_j, where EjE_j are the electric field components in the crystal frame driving the electro-optic effect. The electro-optic effect generated by a temporally modulated electric field can occur through two distinct mechanisms: electronic and vibrational that are described by “electronic” electro-optic coefficient rijelr_{ij}^{el} and the “vibrational” electro-optic coefficient rijvibr_{ij}^{vib}, Ref. [9]. When a THz electric field ETHzE_{\mathrm{THz}} is applied, the change in the index of refraction ellipsoid tensor elements can then be rewritten as: Δηi=(rijel+rijvib)⋅ETHz,j\Delta\eta_i=(r_{ij}^{el}+r_{ij}^{vib})\cdot E_{\mathrm{THz},j}. In BBO, the vibrational contribution is known to be dominant – even for off-resonant electric fields at low frequencies and it is strongly enhanced when approaching the phonon frequency at ∼4.3\sim4.3 THz.[9] Evidence of this effect is found in the experiment through the observation of a significant increase in SH modulation on resonance, when the THz pump spectral components overlap with the phonon frequency (2a-b of the main manuscript). Therefore, near the phonon resonance, the vibrational contribution can be considered to be dominant and only rijEO=rijvibr_{ij}^{EO}=r_{ij}^{vib} is considered in the electro-optic tensor[9]. Since the phonon amplitude scales as Qj∝ZETHz,jQ_j\propto ZE_{\mathrm{THz},j}, where ZZ is the phonon effective charge, the vibrational term rijvibETHz,jr_{ij}^{vib}E_{\mathrm{THz},j} can be rewritten as rijQjr_{ij}Q_j. Here, rijr_{ij} is the phonon–optic coupling coefficient, which describes the modulation of the refractive-index ellipsoid element as a function of the phonon amplitude.

In the laboratory frame (x,y,z)(x,y,z), the extraordinary axis is along xx and the ordinary axis is along yy, while the light propagates along the zz direction. In the experiment, the THz electric field polarisation is fixed and aligned along the ordinary, yy-axis. The THz polarization in the Jones formalism is ETHz=(0,ETHz,y,0)E_{THz} = (0, E_{THz,y}, 0). The transformation matrix RR is used to convert the THz components from the laboratory frame (x,y,z)(x,y,z) to the crystallographic frame (x′,y′,z′)(x',y',z'):

R=(−cos⁡θcos⁡ϕsin⁡ϕcos⁡ϕsin⁡θ−cos⁡θsin⁡ϕ−cos⁡ϕsin⁡θsin⁡ϕsin⁡θ0cos⁡θ)(18)R = \begin{pmatrix} -\cos\theta\cos\phi& \sin\phi& \cos\phi\sin\theta\\ -\cos\theta\sin\phi& -\cos\phi& \sin\theta\sin\phi\\ \sin\theta& 0 & \cos\theta \end{pmatrix} \tag*{(18)}

The THz electric field in the crystal frame is then ETHzxtal=RETHzE_{THz}^{xtal} = R E_{THz}. Under coherent phonon excitation, the index ellipsoid becomes ηQ=η0+ΔηQ\eta_Q = \eta_0 + \Delta\eta_Q, where η0\eta_0 is the index ellipsoid at the equilibrium [7]. In crystal frame (x′,y′,z′)(x',y',z'), η0\eta_0 can then be written in matrix form as:

η0=(1nx′20001nx′20001nz′2),(nx′=ny′).(19)\eta_0 = \begin{pmatrix} \frac{1}{n_{x'}^2} & 0 & 0 \\ 0 & \frac{1}{n_{x'}^2} & 0 \\ 0 & 0 & \frac{1}{n_{z'}^2} \end{pmatrix}, \quad(n_{x'} = n_{y'}). \tag*{(19)}

Given that BBO is uniaxial, for the modulation of the index ellipsoid ΔηQ\Delta\eta_Q, only the phonon-optic coupling coefficient r22r_{22}, which is related to rijEOr_{ij}^{EO}, is significant and the others can be neglected[8, 10].

As the phonon dipole moment is along the THz electric field direction[3], the phonon amplitude is Q0∝ZETHzxtal≡ZETHz,y′Q_0 \propto Z E_{THz}^{xtal} \equiv Z E_{THz,y'}. Thus, the modulated refractive index ellipsoid, calculated from the electro-optic tensor in Eq. (17), can be written in the matrix form as:

ηQ=(1nx′2+r22Q0cos⁡ϕ−r22Q0sin⁡ϕ0−r22Q0sin⁡ϕ1nx′2−r22Q0cos⁡ϕ0001nz′2).(20)\eta_Q = \begin{pmatrix} \frac{1}{n_{x'}^2} + r_{22}Q_0\cos\phi& -r_{22}Q_0\sin\phi& 0 \\ -r_{22}Q_0\sin\phi& \frac{1}{n_{x'}^2} - r_{22}Q_0\cos\phi& 0 \\ 0 & 0 & \frac{1}{n_{z'}^2} \end{pmatrix}. \tag*{(20)}

Since ϕ=90∘\phi= 90^\circ, a finite value of phonon amplitude Q0Q_0 leads to the emergence of off-diagonal terms, and the crystal becomes biaxial associated with an optical anisotropy in the x′x'-y′y' plane, as the degeneracy between x′x' and y′y' directions is lifted.

The index ellipsoid in the crystal frame can be written as ηQ=η0−r22Q0⋅A\eta_Q = \eta_0 - r_{22}Q_0 \cdot A, where:

A=(010100000).(21)A = \begin{pmatrix} 0 & 1 & 0 \\ 1 & 0 & 0 \\ 0 & 0 & 0 \end{pmatrix}. \tag*{(21)}

The dynamical refractive index ellipsoid in the laboratory frame ηTHzLab\eta_{THz}^{Lab} can then be determined upon cartesian axis transformation: ηTHzLab=RTη0R−r22Q0RTAR=RTη0R−ϵRTAR\eta_{THz}^{Lab} = R^T\eta_0R - r_{22}Q_0 R^TAR = R^T\eta_0R - \epsilon R^TAR, with ϵ=r22Q0\epsilon= r_{22}Q_0 and RTR^T the transpose of RR. The two terms of ηTHzLab\eta_{THz}^{Lab} are analysed separately:

RTη0R=(cos⁡2θnx′2+sin⁡2θnz′20cos⁡θsin⁡θ(1nz′2−1nx′2)01nx′20cos⁡θsin⁡θ(1nz′2−1nx′2)0sin⁡2θnx′2+cos⁡2θnz′2),(22)R^T\eta_0R = \begin{pmatrix} \frac{\cos^2\theta}{n_{x'}^2} + \frac{\sin^2\theta}{n_{z'}^2} & 0 & \cos\theta\sin\theta\left(\frac{1}{n_{z'}^2} - \frac{1}{n_{x'}^2}\right) \\ 0 & \frac{1}{n_{x'}^2} & 0 \\ \cos\theta\sin\theta\left(\frac{1}{n_{z'}^2} - \frac{1}{n_{x'}^2}\right) & 0 & \frac{\sin^2\theta}{n_{x'}^2} + \frac{\cos^2\theta}{n_{z'}^2} \end{pmatrix}, \tag*{(22)}

and

ϵRTAR=ϵ(0−cos⁡θ0−cos⁡θ0sin⁡θ0sin⁡θ0)(23)\epsilon R^{T} A R = \epsilon\begin{pmatrix} 0 & -\cos\theta& 0 \\ -\cos\theta& 0 & \sin\theta\\ 0 & \sin\theta& 0 \end{pmatrix} \tag*{(23)}

With the definition 1ne(θ)2=cos⁡2θnx′2+sin⁡2θnz′2\frac{1}{n_e(\theta)^2} = \frac{\cos^2\theta}{n_{x'}^2} + \frac{\sin^2\theta}{n_{z'}^2} and 1no2=1nx′2\frac{1}{n_o^2} = \frac{1}{n_{x'}^2}, and reducing the analysis to two dimensions, considering that the light polarization lies in the x-y plane of the laboratory frame, it follows that:

ηTHzLab=(1ne(θ)2ϵcos⁡θϵcos⁡θ1no2)(24)\eta_{\mathrm{THz}}^{Lab} = \begin{pmatrix} \frac{1}{n_e(\theta)^2} & \epsilon\cos\theta\\ \epsilon\cos\theta& \frac{1}{n_o^2} \end{pmatrix} \tag*{(24)}

This dynamical index ellipsoid in the lab frame is diagonalized by a rotation of γ0\gamma_0, and the new refractive indexes are:

noTHz=no+(r22Q0)2no3γ0/2(25)n_o^{\mathrm{THz}} = n_o + (r_{22}Q_0)^2 n_o^3\gamma_0/2 \tag*{(25)}

and

neTHz=ne(θ)−(r22Q0)2ne3(θ)γ0/2,(26)n_e^{\mathrm{THz}} = n_e(\theta) - (r_{22}Q_0)^2 n_e^3(\theta)\gamma_0/2, \tag*{(26)}

with γ0=κϵ+O[ϵ]2=κr22Q0\gamma_0 = \kappa\epsilon+ O[\epsilon]^2 = \kappa r_{22}Q_0 with κ=cos⁡θ1ne(θ)2−1no2\kappa= \frac{\cos\theta}{\frac{1}{n_e(\theta)^2}-\frac{1}{n_o^2}}.

The primary result is that the dynamical principal axes of the crystal in the laboratory frame are no longer aligned with the xx and yy crystal axes upon phonon excitation with amplitude Q0Q_0. Instead, they are rotated by an angle γ0\gamma_0, which is proportional to Q0Q_0. Importantly, the magnitude of the refractive indices are not significantly altered by Q0Q_0, since the correction is to 3rd-order (r22Q0)2γ0∝(r22Q0)3(r_{22}Q_0)^2\gamma_0 \propto(r_{22}Q_0)^3.

Analytical model: Vibrational Modulation of Second Harmonic Intensity

For the Type-I phase-matched BBO crystal (ϕ=90∘\phi= 90^\circ, θ=29.3∘\theta= 29.3^\circ), the SHG intensity is proportional to ∣EyFF∣4\lvert E_y^{FF}\rvert^4, where EyFFE_y^{FF} is the projection of the amplitude of the 800 nm FF on the lab-frame y-axis, which is parallel to the unperturbed crystal-frame ordinary axis. The SH will be polarised along the crystal extraordinary axis, which is parallel to the lab-frame x-axis.

A linearly polarised FF propagating along the z-direction in the laboratory frame has field components (x,y)(x,y) given by the following expression:

EFF(α)=E0FF(sin⁡α,cos⁡α)(27)E^{FF}(\alpha) = E_0^{FF}(\sin\alpha,\cos\alpha) \tag*{(27)}

where α\alpha is the angle between that y-axis and the input polarisation.

Due to the birefringence of the BBO crystal, for input fundamental fields that are not aligned with the principal axes, the polarisation will evolve with propagation distance ζ\zeta in the crystal. In the static case, at equilibrium, after travelling a distance ζ\zeta, the evolution of the field components is determined using the linear phase retardation matrix Γ(ζ)\Gamma(\zeta) such that:

EFF(ζ,α)=Γ(ζ)EFF(α)whereΓ(ζ)=(ei2πneζ/λ00ei2πnoζ/λ)(28)E^{FF}(\zeta,\alpha) = \Gamma(\zeta)E^{FF}(\alpha) \quad\text{where} \quad\Gamma(\zeta) = \begin{pmatrix} e^{i2\pi n_e\zeta/\lambda} & 0 \\ 0 & e^{i2\pi n_o\zeta/\lambda} \end{pmatrix} \tag*{(28)}

In this case, and in the undepleted pump approximation, the Type-I (o+o→eo + o \to e) SH intensity from a crystal length LL as a function of input polarisation α\alpha can be written as:

ISH,0(L,α)∝∣∫0L(EyFF(ζ,α))2e−ikne,2ωζ dζ∣2(29)I_{\mathrm{SH},0}(L,\alpha) \propto\left\lvert\int_0^L \left(E_y^{FF}(\zeta,\alpha)\right)^2 e^{-ikn_{e,2\omega}\zeta}\,d\zeta\right\rvert^2 \tag*{(29)}

And upon substitution:

ISH,0(L,α)∝∣∫0L(E0FFcos⁡α)2ei4πno,ωζ/λe−ikne,2ωζ dζ∣2(30)I_{SH,0}(L,\alpha) \propto\left|\int_{0}^{L} \left(E_{0}^{FF}\cos\alpha\right)^{2}e^{i4\pi n_{o,\omega}\zeta/\lambda}e^{-ikn_{e,2\omega}\zeta}\,d\zeta\right|^{2} \tag*{(30)}

For perfect Type-I phase matching, Δk=0\Delta k = 0, no,ω=ne,2ωn_{o,\omega} = n_{e,2\omega} and the exponential term reduces to 1, and the SH intensity is:

ISH,0(L,α)∝L2(E0FF)4cos⁡4α(31)I_{SH,0}(L,\alpha) \propto L^{2}\left(E_{0}^{FF}\right)^{4}\cos^{4}\alpha \tag*{(31)}

In the dynamic case, the principal axes of the BBO crystal are rotated, however, the χ(2)\chi^{(2)} tensor remains unchanged. This is confirmed by DFPT calculations, which show that even for relatively large phonon amplitudes the variation of the χ(2)\chi^{(2)} tensor elements remains negligible with respect to the static d22d_{22} value (see Supplementary Figure S6). Therefore, the relevant components of the input field remain the x and y-components in the laboratory frame. Again, the linearly polarized input field can be expressed as:

EFF=E0FF(sin⁡α,cos⁡α)(32)E^{FF}=E_{0}^{FF}(\sin\alpha,\cos\alpha) \tag*{(32)}

However, now, upon THz excitation the principal axes are rotated by an amount γ\gamma. To calculate the propagation phase-advance, the components are rotated onto the new principal axes coordinate system by RγR_{\gamma}, before applying the linear phase retardation and then rotated back to the original lab-frame coordinate system to find the FF components as a function of the propagation distance ζ\zeta. In this case the fundamental field components are given by the expression:

EFF(ζ,γ)=RγTΓ(ζ)RγEFF(33)E^{FF}(\zeta,\gamma)=R_{\gamma}^{T}\Gamma(\zeta)R_{\gamma}E^{FF} \tag*{(33)}

Here, RγR_{\gamma} is the rotation matrix associated with the rotation of the ordinary and extraordinary axes (see Figure 4a of the main text), and is written as:

Rγ=(cos⁡γ−sin⁡γsin⁡γcos⁡γ)(34)R_{\gamma}=\begin{pmatrix}\cos\gamma& -\sin\gamma\\ \sin\gamma& \cos\gamma\end{pmatrix} \tag*{(34)}

and RγTR_{\gamma}^{T} is its transpose.

If the rotation of the principal axes is considered to be constant (γ0\gamma_{0}) and again in the undepleted pump approximation, the intensity of the SH polarized along lab-frame x-axis after propagating to a depth zz inside the crystal is:

ISH,THz(z,α,γ0)∝∣∫0z(EyFF(ζ,γ0))2e−ikne,2ωζ dζ∣2(35)I_{SH,THz}(z,\alpha,\gamma_{0}) \propto\left|\int_{0}^{z}\left(E_{y}^{FF}(\zeta,\gamma_{0})\right)^{2}e^{-ikn_{e,2\omega}\zeta}\,d\zeta\right|^{2} \tag*{(35)}

Upon substitution, three terms appear in the integral with distinct arguments in the exponential corresponding to Type-0, Type-I and Type-II wave mixing processes. If the Type-0 and Type-II processes are neglected due to a lack of phase matching, i.e. ne,ω≠ne,2ωn_{e,\omega} \ne n_{e,2\omega} and ne,ω+no,ω≠2ne,2ωn_{e,\omega}+n_{o,\omega} \ne2n_{e,2\omega}, and only the Type-I process is considered, the integral in the intensity term becomes:

ESH,THz(z,α,γ0)∝E02cos⁡2(α−γ0)2sin⁡(Δkz/2)ΔkeiΔkz(36)E_{SH,THz}(z,\alpha,\gamma_{0}) \propto E_{0}^{2}\cos^{2}(\alpha-\gamma_{0})\frac{2\sin(\Delta kz/2)}{\Delta k}e^{i\Delta kz} \tag*{(36)}

Where Δk=(ω/c)(no,ω−ne,2ω)=(ω/c)Δn\Delta k=(\omega/c)(n_{o,\omega}-n_{e,2\omega})=(\omega/c)\Delta n.

For small rotations of the principal axes (γ0≪π/2\gamma_{0}\ll\pi/2), the expression for ESH,THzE_{SH,THz} can be expanded.

ESH,THz(z,α,γ0)∝E02z sinc⁡(Δkz/2)eiΔkzcos⁡α[cos⁡α+2γ0sin⁡α+O[γ02]](37)E_{SH,THz}(z,\alpha,\gamma_{0}) \propto E_{0}^{2}z\,\operatorname{sinc}(\Delta kz/2)e^{i\Delta kz}\cos\alpha\left[\cos\alpha+2\gamma_{0}\sin\alpha+O[\gamma_{0}^{2}]\right] \tag*{(37)}

And, finally, the SH intensity to 1st^{\mathrm{st}}-order in γ0\gamma_{0} becomes:

ISH,THz(z,α,γ0)=∣ESH,THz(z,α,γ0)∣2∝E04z2sinc⁡2(Δkz/2)[cos⁡4α+4γ0cos⁡3αsin⁡α](38)\begin{aligned} I_{SH,THz}(z,\alpha,\gamma_{0}) &= \left|E_{SH,THz}(z,\alpha,\gamma_{0})\right|^{2}\\ &\propto E_{0}^{4}z^{2}\operatorname{sinc}^{2}(\Delta kz/2)\left[\cos^{4}\alpha+4\gamma_{0}\cos^{3}\alpha\sin\alpha\right] \tag*{(38)} \end{aligned}

Assuming perfect Type-I phase matching, Δk=0\Delta k = 0, the intensity of the SH is given by the expression:

ISH,THz(z,α,γ0)∝E04z2cos⁡3α(cos⁡α+4γ0sin⁡α)(39)I_{SH,THz}(z,\alpha,\gamma_0) \propto E_0^4 z^2 \cos^3\alpha\left(\cos\alpha+4\gamma_0\sin\alpha\right) \tag*{(39)}

At equilibrium, γ0=0\gamma_0=0, and the conventional expression for SH intensity is recovered, in agreement with the angular dependence observed in the experiment (Figure 3b of the main manuscript). Based on equation S31, the SH intensity modulation can be approximated as:

ΔISH,THz=ISH,THz−ISH,0≈4ρE04z2γ0cos⁡3αsin⁡α(40)\Delta I_{SH,THz}=I_{SH,THz}-I_{SH,0}\approx4\rho E_0^4z^2\gamma_0\cos^3\alpha\sin\alpha \tag*{(40)}

where ρ=ω2dNL2nω2c2\rho=\frac{\omega^2d_{NL}^2}{n_\omega^2c^2} is the proportionality constant and, d22d_{22} the nonlinear coefficient of Type-I SHG{sup}11`.

In the dynamical case, for a finite phonon amplitude Q0Q_0 and rotation of the principal axes γ0\gamma_0, the analytical expression for the SH intensity modulation in Eq. (40) matches the angular dependence observed in the experiment.

In the presence of optical absorption due to IR-active vibrations, the THz electric field undergoes exponential attenuation within the crystal. As a results, the phonon amplitude decreases as a function of the propagation depth ζ\zeta and can be approximated with an exponential decay, Q(ζ)=Q0e−ζ/δphQ(\zeta)=Q_0e^{-\zeta/\delta_{ph}}, where δph\delta_{ph} is the phonon propagation depth. Consequently, along the propagation, the parameter γ\gamma, being proportional to QQ, also undergoes exponential depletion according to γ(ζ)=γ0e−ζ/δph\gamma(\zeta)=\gamma_0e^{-\zeta/\delta_{ph}}. Then, including the propagation dependence of γ(ζ)\gamma(\zeta), the SH field ESH,THzE_{SH,THz} can be written as:

ESH,THz(z,α,γ0,δph)=∫0z(Ey(ζ,γ0,δph))2e−i2noπζλ/2 dζ(41)E_{SH,THz}(z,\alpha,\gamma_0,\delta_{ph})=\int_0^z\left(E_y(\zeta,\gamma_0,\delta_{ph})\right)^2e^{-\frac{i2n_o\pi\zeta}{\lambda/2}}\,d\zeta \tag*{(41)}

In this case, the expression for the SH intensity modulation for z=Lz=L is found to be:

ΔISH,THz=4ρE04Lγ0lscos⁡3αsin⁡αΛ(δph,L).(42)\Delta I_{SH,THz}=4\rho E_0^4L\gamma_0l_s\cos^3\alpha\sin\alpha\Lambda(\delta_{ph},L). \tag*{(42)}

After substituting the static SH intensity at α=0\alpha=0, ISH,0(0)=ρE04zL2I_{SH,0}(0)=\rho E_0^4zL^2, the modulation becomes:

ΔISH,THz(α)=4ISH,0(0)Lγ0cos⁡3αsin⁡αΛ(δph,L),(43)\Delta I_{SH,THz}(\alpha)=4\frac{I_{SH,0}(0)}{L}\gamma_0\cos^3\alpha\sin\alpha\Lambda(\delta_{ph},L), \tag*{(43)}

which corresponds to Eq. (3) of the main manuscript. Equation S35 is similar to Eq. (40), but it includes the depletion factor:

Λ(δph,z=L)=(4e−z/δph)⋅(−4π2Δn2+4ez/δphπ2Δn2−λ2δph+λ2δph2cos⁡(2πzΔnλ)−2πΔnλδphsin⁡(2πzΔnλ)4π2Δn2δph+λ2δph3)(44)\begin{aligned} \Lambda(\delta_{ph},z=L) &=\left(4e^{-z/\delta_{ph}}\right)\\ &\quad\cdot\left(\frac{-4\pi^2\Delta n^2+4e^{z/\delta_{ph}}\pi^2\Delta n^2-\frac{\lambda^2}{\delta_{ph}}+\frac{\lambda^2}{\delta_{ph}^2\cos\left(\frac{2\pi z\Delta n}{\lambda}\right)}-\frac{2\pi\Delta n\lambda}{\delta_{ph}\sin\left(\frac{2\pi z\Delta n}{\lambda}\right)}}{\frac{4\pi^2\Delta n^2}{\delta_{ph}}+\frac{\lambda^2}{\delta_{ph}^3}}\right) \tag*{(44)} \end{aligned}

To evaluate the depletion factor, the wavelength-dependent refractive indices for BBO at room temperature were taken from Ref. 12.

(43) shows that the SHG modulation intensity ΔISH,THz(α)\Delta I_{SH,THz}(\alpha) scales with the parameter γ0\gamma_0, and therefore with the phonon amplitude Q0Q_0, as expected and reproduces the same angular dependence observed experimentally. This equation has been used to fit the polarizationresolved differential SHG modulation (see Figure 4b in the main text) and to extract the value of γ0\gamma_0. A key quantity in this analysis is the phonon penetration depth, δph\delta_{\mathrm{ph}}, which is determined to be 30.2 μm30.2\ \mu\mathrm{m} by finite-difference time-domain (FDTD) simulations (Figure 4c in the main text).

Higher-order Nonlinear Processes

Alongside the phonon-mediated electro-optic modulation of the optical axis - a χ(2)\chi^{(2)} process that modulates the SHG through phase matching – higher-order THz-mediated χ(3)\chi^{(3)} effects. These processes have been described within the framework of Terahertz-Field–Induced Second Harmonic (TFISH). It should be noted, however, that in previous TFISH works[13, 14, 15, 16], either the phonons were not resonantly excited, or critically, the SHG process was phase-mismatched. In this case, the frequency conversion is confined to a coherence length that is typically on the order of microns – commensurate with a few wavelengths of the fundamental beam. Within this short-interaction regime the THz-driven phonons directly modulate the second-order nonlinearity resulting in a χ(3)\chi^{(3)} effect, while second-order electro-optic variation of the refractive index ellipsoid and its impact on non-existent phase matching conditions are negligible.

In the current experiment, χ(3)\chi^{(3)} effects are found to have negligible impact on SHG modulation under optimal phase matching conditions, as they cannot reproduce the angular dependence observed in the experiment, the χ(3)\chi^{(3)} SHG modulation increases and decreases uniformly as a function of time delay for all input polarisations (rather than exhibiting a vice-versa behaviour between neighbouring lobes as observed in the experiment), and finally, the magnitude of the χ(3)\chi^{(3)}-effects on the SHG intensity modulation are much smaller to those observed in the experiment.

To isolate and study the χ(3)\chi^{(3)}-effects, additional experiments were performed in a c-cut (θ=0∘\theta= 0^\circ) BBO crystal (MolTech GmbH, 500 μm500\ \mu\mathrm{m} thick), which is far from Type-I phase-matching conditions (Supplementary Figure 5a-Figure 5b). The crystal quality and orientation were verified by Laue x-ray diffraction. In this configuration, the THz and the NIR pulses propagate along the optic axis (c-axis) and the refractive index is non_o along xx- and yy-axes in the laboratory frame. Therefore, the crystal is isotropic and the coherence length is identical for all input polarization states.

In the phase-mismatched case, the equilibrium SHG conversion is two orders of magnitude lower than in the phase-matched case and the nonlinear coefficient d22d_{22} dominates the SHG conversion. Analytically, the SH intensity along the lab-frame xx-axis is, ISHG,x∝∣d22Ex2−d22Ey2∣2I_{\mathrm{SHG},x} \propto\lvert d_{22}E_x^2-d_{22}E_y^2\rvert^2, where Ex=E0FFsin⁡αE_x = E_0^{FF}\sin\alpha and Ey=E0FFcos⁡αE_y = E_0^{FF}\cos\alpha are the projections of fundamental field EFFE^{FF} along the lab-frame xx- and yy-axes. From the expression, a 4-fold symmetry in input polarisation angle α\alpha is expected, with local maxima on the x- and y-axes. Indeed, at equilibrium in the phase-mismatched case, the observed SHG has a polarization dependence in agreement with the analytical expression (Fig S5c).

In contrast, when equilibrium SHG measurements are performed under optimal Type-I phase matching conditions, only the SH due to the Type-I process (SH along the extraordinary in the lab frame, xx-axis) is generated by the component of the FF field projected along the ordinary, yy-axis (o+o→eo+o\to e). Contributions from the other input fundamental field components are negligible due to a lack of phase matching. Analytically, this results in a static SHG signal ISHG∝∣d22Ex2∣2∝cos⁡4αI_{\mathrm{SHG}} \propto\lvert d_{22}E_x^2\rvert^2 \propto\cos^4\alpha with 2-fold angular symmetry (see Supplementary Figure S5).

In the dynamical case, the THz excitation modulates the intensity of the SH. The effect can be considered from two perspectives, either the THz field can be viewed to modulate the 2nd-order nonlinear susceptibility such that χ(2)→χ(2)(t)\chi^{(2)} \to\chi^{(2)}(t), or the effect can be seen to occur through a 4-wave mixing process involving the 3rd-order nonlinear susceptibility χ0(3)\chi^{(3)}_0 and the THz and fundamental fields. The second scenario is commonly used in the description of TFISH experiments and is why THz-Field Induced Second Harmonic is often known as a 3rd-order χ(3)\chi^{(3)} effect. Within the TFISH framework, the intensity of the SHG signal has three terms, a time-dependent heterodyne term which depends linearly on the THz electric field, a homodyne term and a DC offset term[17]:

ISHG(t)∝∣(χ(2)+χ(3)ETHz(t))EFF2∣2∝∣χ(2)EFF2∣2⏟DC+2Re⁡(χ(2)χ(3)ETHz(t)EFF4)+∣χ(3)ETHz(t)EFF2∣2⏟Homodyne∝∣ETHz(t)∣2(45)\begin{aligned} I_{\mathrm{SHG}}(t) &\propto\left\lvert\left(\chi^{(2)}+\chi^{(3)}E_{\mathrm{THz}}(t)\right)E_{\mathrm{FF}}^2\right\rvert^2 \\ &\propto\underbrace{\left\lvert\chi^{(2)}E_{\mathrm{FF}}^2\right\rvert^2}_{\mathrm{DC}}+2\operatorname{Re}\left(\chi^{(2)}\chi^{(3)}E_{\mathrm{THz}}(t)E_{\mathrm{FF}}^4\right)+\underbrace{\left\lvert\chi^{(3)}E_{\mathrm{THz}}(t)E_{\mathrm{FF}}^2\right\rvert^2}_{\mathrm{Homodyne}\propto\lvert E_{\mathrm{THz}}(t)\rvert^2} \tag*{(45)} \end{aligned}

In the alternative perspective, where the THz electric field is assumed to modulate the 2nd2^{\mathrm{nd}}-order susceptibility tensor, χ(2)(t)\chi^{(2)}(t) can be written as:

χ(2)(t)=χ(2)+Δχ(2)(t) and ISHG(t)∝∣χ(2)(t)EFF2∣2,(46)\chi^{(2)}(t)=\chi^{(2)}+\Delta\chi^{(2)}(t)\ \text{and}\ I_{\mathrm{SHG}}(t)\propto\left|\chi^{(2)}(t)E_{\mathrm{FF}}^{2}\right|^{2}, \tag*{(46)}

where χ(2)\chi^{(2)} is the static nonlinear susceptibility tensor and Δχ\Delta\chi is the THz induced, time-dependent modulation.

The pictures are equivalent, as the time-dependent change in the nonlinear susceptibility is assumed proportional to the THz field, Δχ(2)(t)∼χ(3)ETHz(t)∼χ(3)Q(t)\Delta\chi^{(2)}(t)\sim\chi^{(3)}E_{\mathrm{THz}}(t)\sim\chi^{(3)}Q(t).

If only a single input FF polarization is used, the heterodyne term in the time-dependent intensity might erroneously be considered as the source of the SH modulation in the phase-matched experiment. The effect is linearly dependent in strength on the THz electric field, it is phase-locked to the THz electric field, and it has a frequency dependence that matches the THz-driven phonon. However, the strength of the 3rd3^{\mathrm{rd}}-order TFISH effect is expected to be much weaker than the 2nd2^{\mathrm{nd}}-order effect, and perhaps even more conclusively, when the angular symmetry of the intensity modulation is included in the analysis, it can be shown that 3rd3^{\mathrm{rd}}-order processes cannot be the underlying source of the observed dynamics in the phase-matched case.

To investigate the polarization dependence, the perspective employing a time-dependent χ(2)(t)\chi^{(2)}(t) will be used for analysis and interpretation of the dynamical SH signal in both the phase-matched and phase-mismatched geometries. By convention, the nonlinear susceptibility is expressed in terms of the nonlinear coefficient d=12χ(2)d=\frac{1}{2}\chi^{(2)}, so that the dynamical SH field amplitude is:

ESH,THzxtal∝(d0+Δd)ExtalFFExtalFF,(47)E_{\mathrm{SH,THz}}^{\mathrm{xtal}}\propto(d_{0}+\Delta d)E_{\mathrm{xtal}}^{\mathrm{FF}}E_{\mathrm{xtal}}^{\mathrm{FF}}, \tag*{(47)}

where ExtalFFE_{\mathrm{xtal}}^{\mathrm{FF}} is the fundamental field in the crystal-frame, d0d_{0} is the nonlinear coefficient at the equilibrium and Δd\Delta d is the modulation induced by the THz field.

By applying a coordinate transformation from the lab-frame to crystal-frame, evaluating the SH fields that are generated in the BBO and then transforming back to the lab-frame, the resultant SH field components can be related to the input polarisation of the fundamental field in the lab frame, which is defined by α\alpha,

ESH,THzlab(α)=RT(d0+Δd)f3→6(R(EFFlab(α))).(48)E_{\mathrm{SH,THz}}^{\mathrm{lab}}(\alpha)=R^{T}(d_{0}+\Delta d)f_{3\to6}\left(R\left(E_{\mathrm{FF}}^{\mathrm{lab}}(\alpha)\right)\right). \tag*{(48)}

Here, f3→6f_{3\to6} is the operator that maps vector components (x′,y′,z′)(x',y',z') of the input field onto the (x′2,y′2,z′2,2y′z′,2x′z′,2x′y′)(x'^{2},y'^{2},z'^{2},2y'z',2x'z',2x'y') terms corresponding to the square of the input field. The components of the fundamental field in the lab frame are, E0FF(sin⁡α,cos⁡α,0)E_{0}^{\mathrm{FF}}(\sin\alpha,\cos\alpha,0), with α\alpha the polarization angle with respect to the yy-axis. RR is the rotation matrix that maps the field components from the lab frame to the crystal frame and is a function of the crystal geometry (θ=29.3,ϕ=90\theta=29.3,\phi=90).

Finally, the SH intensity as a function of input polarisation can be written as:

ISH,THz(α)∝∣ESH,THzlab(α)∣2∝∣RT(d0+Δd)f3→6(R(EFFlab(α)))∣2(49)I_{\mathrm{SH,THz}}(\alpha)\propto\left|E_{\mathrm{SH,THz}}^{\mathrm{lab}}(\alpha)\right|^{2}\propto\left|R^{T}(d_{0}+\Delta d)f_{3\to6}\left(R\left(E_{\mathrm{FF}}^{\mathrm{lab}}(\alpha)\right)\right)\right|^{2} \tag*{(49)}

The output intensity in the dynamical case contains interference terms involving the static 2nd2^{\mathrm{nd}}-order nonlinearity and time-dependent 2nd2^{\mathrm{nd}}-order nonlinearity — or equivalently the χ(3)\chi^{(3)} nonlinear susceptibility.

When the xx-component in the lab frame is isolated with a polariser (corresponding to the extraordinary axis in the phase matched geometry), and for generality if all elements in the Δd\Delta d tensor are allowed to be non-zero, the polarisation dependent intensity difference includes the five following terms (Ez=0E_{z}=0):

ΔISH,THzx(α)=ISH,THzx−ISH,0x\Delta I_{\mathrm{SH,THz}}^{x}(\alpha)=I_{\mathrm{SH,THz}}^{x}-I_{\mathrm{SH,0}}^{x}
∝(a1)Ex4+(a2)Ey4+(a3)Ex2Ey2+(a4)Ex3Ey+(a5)ExEy3,(50)\propto(a_{1})E_{x}^{4}+(a_{2})E_{y}^{4}+(a_{3})E_{x}^{2}E_{y}^{2}+(a_{4})E_{x}^{3}E_{y}+(a_{5})E_{x}E_{y}^{3}, \tag*{(50)}

where:

Ex=E0FFsin⁡α ;Ey=E0FFcos⁡α.(51)E_x = E_0^{FF}\sin\alpha\ ;\quad E_y = E_0^{FF}\cos\alpha. \tag*{(51)}

In Eq. (50), the a1,…,a5a_1,\ldots,a_5 coefficient depends on the static and modulated nonlinear tensor elements. Given the initial symmetry of the BBO crystal (d15,d22,d33≠0d_{15},d_{22},d_{33}\ne0), the THz induced variation in the intensity of the second harmonic can have an angular dependence with terms proportional to, cos⁡4α\cos^4\alpha, sin⁡4α\sin^4\alpha, cos⁡2αsin⁡2α\cos^2\alpha\sin^2\alpha, cos⁡αsin⁡3α\cos\alpha\sin^3\alpha, and cos⁡3αsin⁡α\cos^3\alpha\sin\alpha, and the proportionality will depend to 1st or 2nd order on the time-dependent second order nonlinearity, or the 3rd-order susceptibility χ(3)\chi^{(3)}.

The angular dependence of the SH intensity modulation in the phase-mismatched case is shown in Supplementary Figure S5. Here, as previously stated, the TFISH, χ(3)\chi^{(3)}-effects are isolated and the angular dependence can be fit with an equation of the form in S42. The dominant fit parameters are found to be associated with the cos⁡4α\cos^4\alpha, sin⁡4α\sin^4\alpha terms, which scale the lobes on the xx- and yy-axes, and the cos⁡2αsin⁡2α\cos^2\alpha\sin^2\alpha term, which scales the lobes along the diagonals. Small adjustments to the rotation of the polar fit can be made using the remaining terms. In the time-domain, the onset of THz excitation results in a uniform decrease of the SHG signal on the axes, ΔISHG<0\Delta I_{\mathrm{SHG}} < 0, which is consistent with a THz-induced decrease of the d22d_{22} coefficient. At the same time, the onset of THz excitation results in a uniform increase of the SHG signal on the diagonals, ΔISHG>0\Delta I_{\mathrm{SHG}} > 0.

The angular dependence in the Type-I phase-matched case is shown in Supplementary Figure S5. Here, in contrast to the phase-mismatched case, the strongest variation in the SH intensity is found at ±30∘\pm30^\circ and the signal on the axes is minimised. Moreover, the sign of the effect alternates between lobes such that if one lobe experiences an increase in SH intensity, the neighbouring lobe exhibits a decrease. If the dynamics were explained by TFISH or, equivalently, χ(3)\chi^{(3)} effects, one would expect that the cos⁡4α\cos^4\alpha, sin⁡4α\sin^4\alpha and cos⁡2αsin⁡2α\cos^2\alpha\sin^2\alpha terms would all play a role in the SH modulation, yet there is no clear signature of any of these terms in the angular dependence of the phase-matched measurements. This is especially notable for the sin⁡4α\sin^4\alpha term, which involves a heterodyne term that mixes two Type-I phase matched processes, and would be expected to have the strongest response.

Instead, only the cos⁡3αsin⁡α\cos^3\alpha\sin\alpha term from the TFISH scenario is present in the phase-matched angular dependence. A closer look at the elements in the proportionality coefficient of the cos⁡3αsin⁡α\cos^3\alpha\sin\alpha term, reveals component that is linear in THz field strength and scales with O(Δd)O(\Delta d) or O(χ0(3))O\left(\chi_0^{(3)}\right), and a component that is quadratic in THz field strength and scales with O2(Δd)O^2(\Delta d) or O2(χ0(3))O^2\left(\chi_0^{(3)}\right).

The homodyne contribution can be excluded because the SH intensity modulation varies linearly with the THz field strength (Figure 2c of the main manuscript). Furthermore, as the homodyne contribution is proportional to ∼Q2\sim Q^2, it would need to be driven by a phonon at half the observed modulation frequency. However, in the experiment, when the driving THz was filtered to excite only a phonon at 2.15 THz, the intensity modulation is nearly extinguished (Figure 2, main manuscript).

There are also reasons to suggest that the heterodyne term does not contribute significantly to the phase-matched intensity modulation either. Importantly, the heterodyne interference term involves an SH component proportional to dd that results from the Type-I phase-matched process, but the other component, proportional to the Δd\Delta d time-dependent nonlinearity, results from a Type-II process, which is NOT phase matched.

In summary, the angular dependence in the phase-matched case does not show the features one would expect from the phase-mismatched reference measurements, and the features that are expressed do not appear to have originated from a χ(3)\chi^{(3)} process.

The features in the phase-matched case are however fit perfectly by the cascaded 2nd2^{\mathrm{nd}}-order model involving electro-optic modification of the index of refraction ellipsoid and its effect on phase-matching in the usual SHG process with static 2nd2^{\mathrm{nd}}-order nonlinearity.

Numerical estimate of the χ(3)\chi^{(3)} contribution in the phase-matched case

Qualitative analysis of the angular symmetry in the phase-matched experiment and comparison with the symmetry observed in the phase-mismatched experiments fails to provide evidence of χ(3)\chi^{(3)} or TFISH effects as the definitive source of the THz-driven SH modulation in BBO. However, higher order effects are always present, and additional insight can be gained by a more quantitative evaluation of the strength of these effects.

In the TFISH framework, the intensity of the SHG signal is proportional to three terms, a time-dependent heterodyne term, a time dependent homodyne term and a DC offset term:

ISHG(t)∝∣(χ(2)+χ(3)ETHz(t))EFF2∣2∝∣χ(2)EFF2∣2⏟DC+2Re⁡(χ(2)χ(3)ETHz(t)EFF4)Heterodyne ∝ETHz(t)+∣χ(3)ETHz(t)EFF2∣2⏟Homodyne ∝∣ETHz(t)∣2(52)\begin{aligned} I_{SHG}(t) &\propto\left|\left(\chi^{(2)}+\chi^{(3)}E_{THz}(t)\right)E_{FF}^{2}\right|^{2}\\ &\propto\underbrace{\left|\chi^{(2)}E_{FF}^{2}\right|^{2}}_{\mathrm{DC}}+2\operatorname{Re}\left(\chi^{(2)}\chi^{(3)}E_{THz}(t)E_{FF}^{4}\right)_{\mathrm{Heterodyne}\ \propto E_{THz}(t)}+\underbrace{\left|\chi^{(3)}E_{THz}(t)E_{FF}^{2}\right|^{2}}_{\mathrm{Homodyne}\ \propto|E_{THz}(t)|^{2}} \tag*{(52)} \end{aligned}

To link the intensity modulation to the optically driven phonons in the BBO, according to Ref.[18], χ(3)ETHz(t)\chi^{(3)}E_{THz}(t) is identical to a time-dependent perturbation of the second-order non-linearity and can be rewritten in terms of the effective nonlinear coefficient deffd_{eff} and the phonon displacement QQ:

χ(3)ETHz(t)=2∂deff∂QQ(t)→Δd=∂deff∂QQ(53)\chi^{(3)}E_{THz}(t)=2\frac{\partial d_{eff}}{\partial Q}Q(t)\to\Delta d=\frac{\partial d_{eff}}{\partial Q}Q \tag*{(53)}

Therefore, the heterodyne contribution is proportional to the amplitude of the phonon, which is assumed proportional to the THz field.

DFT calculations of the phonon displacements can then be used to estimate the time-dependent 2nd2^{\mathrm{nd}}-order nonlinear coefficients, Δd\Delta d. Calculations of the phonon-amplitude dependent nonlinear coefficients are shown in Figure S6. By combining these coefficients with the phonon penetration depth δph\delta_{ph} (calculated independently by DFPT), a simple two-layer model of the BBO crystal can be used to estimate the strength of the 3rd3^{\mathrm{rd}}-order effects. This model consists of a uniformly pumped layer with a constant phonon amplitude Q0Q_0 that is δph\delta_{ph} thick with the additional time-dependent nonlinearities, while the remainder of the crystal is considered to be at equilibrium.

The Type-I phase-matched component, which is proportional to d222∣EYF∣4d_{22}^{2}|E_{Y}^{F}|^{4} will be affected by the time-dependent term Δd22\Delta d_{22}, which is expected to have a maximum value of Δd22=0.1 pm/V\Delta d_{22}=0.1\ \mathit{pm}/V with d22=2.9 pm/Vd_{22}=2.9\ \mathit{pm}/V. In the 2-layer model which is expected to overestimate the strength of the effect, the relative 3rd-order intensity modulation of this component is calculated to be 0.68%, far below the modulation observed in the experiment. Furthermore, in terms of the angular dependence, this modulation will result in a uniform scaling of the original equilibrium 2-fold symmetric pattern, which is not observed in the experiment.

The strength of the Type-I/Type-II heterodyne component can also be calculated using the other time-dependent nonlinear components, Δd25\Delta d_{25}, Δd26\Delta d_{26}, and Δd35\Delta d_{35}. With the same 2-layer model, the relative intensity modulation that can be expected from this term is approximately 0.3%. While this value is again expected to be an overestimate, it is still far below the modulation observed in the experiment.

Again, the estimates of the strength of the χ(3)\chi^{(3)}-effect fail to support a TFISH explanation of the intensity modulation in the phase-matched experiment.

Estimation of the vibrational electro-optic coefficient and its frequency dependence

This section describes in detail the calculation of the vibrational electro-optic coefficient magnitude near the phonon resonance, as obtained from DFPT and from experimentally estimated values.

Calculation of the EO coefficient from DFPT: The modulation of the refractive index Δn\Delta n is related to the phonon coordinate QQ by:

Δn=nx3r22Q2(54)\Delta n = \frac{n_x^3 r_{22} Q}{2} \tag*{(54)}

see Eq. (2) in the main manuscript. Here, r22r_{22} denotes the phonon–optic coupling coefficient that quantifies the coupling between the phonon amplitude and the resulting refractive-index modulation.

From the DFPT results, Δn\Delta n varies linearly with QQ by which a value of r22=0.0049 A˚−1 (amu)1/2r_{22} = 0.0049\ \mathring{\mathrm{A}}^{-1}\,(\mathrm{amu})^{1/2} can be extrapolated from a linear fit, as discussed in the main text.

The standard vibrational EO coefficient, which is denoted here as r22Er_{22}^{E} describes the variation of the index of refraction with respect to the applied electric field strength. This coefficient can be given in SI units (pm/Vpm/V).

The conventional linear relationship between the phonon coordinate QQ and the electric field ETHzE_{\mathrm{THz}}: Q(ω)=ϵ0VωpωQ2−ω2−iΓωETHz(ω)Q(\omega) = \frac{\sqrt{\epsilon_0 V \omega_p}}{\omega_Q^2-\omega^2-i\Gamma\omega} E_{\mathrm{THz}}(\omega) is required to convert between the two coefficients.

Substituting this relationship into Eq. (54) yields

Δn=nx32r22ϵ0VωpωQ2−ω2−iΓωETHz=nx32r~22EETHz(55)\Delta n = \frac{n_x^3}{2}r_{22}\frac{\sqrt{\epsilon_0 V}\omega_p}{\omega_Q^2-\omega^2-i\Gamma\omega}E_{\mathrm{THz}} = \frac{n_x^3}{2}\widetilde{r}_{22}^{E}E_{\mathrm{THz}} \tag*{(55)}

Here, VV is the volume of the unit cell, ε0\varepsilon_0 is the vacuum permittivity, ωQ\omega_Q, ωp\omega_p and Γ\Gamma are the phonon frequency, phonon plasma frequency and the linewidth, respectively. Here, r~22\tilde{r}_{22} is the complex EO coefficient in SI units, and taking its real part:

r22E=Re⁡{r~22EO}=r22ε0Vωp(ωQ2−ω2)(ωQ2−ω2)2−(Γω)2=r22E(0)(ωQ2−ω2)(ωQ2−ω2)2−(Γω)2.(56)r^{E}_{22} = \operatorname{Re}\{\tilde{r}^{\mathrm{EO}}_{22}\} = r_{22}\frac{\sqrt{\varepsilon_0 V}\omega_p(\omega_Q^2-\omega^2)}{(\omega_Q^2-\omega^2)^2-(\Gamma\omega)^2}=r^{E}_{22}(0)\frac{(\omega_Q^2-\omega^2)}{(\omega_Q^2-\omega^2)^2-(\Gamma\omega)^2}. \tag*{(56)}

Note that Eq. (56) corresponds to the usual expression of the vibrational EO coefficient (see e.g. Ref. 19), also shown in in Eq. (55).

Using r22=0.0049 A˚−1(amu)12r_{22}=0.0049\,\text{\AA}^{-1}(\mathrm{amu})^{\frac{1}{2}} from the DFPT-derived refractive-index modulation, and ωP=6.57\omega_P=6.57 THz, ωQ=4.32\omega_Q=4.32 THz and Γ=0.19\Gamma=0.19 THz determined by infrared spectroscopy, a maximum positive value of r22−max⁡E=55 pmVr^{E}_{22-\max}=55\,\frac{\mathrm{pm}}{V} at ω∗=ωQ2−ΓωQ=4.25\omega_*=\sqrt{\omega_Q^2-\Gamma\omega_Q}=4.25 THz is obtained. As discussed in the main text, the EO coefficient does not peak exactly at the phonon frequency ωQ\omega_Q, the real part attains its maximum just below resonance at ω∗\omega_* which is also consistent with the frequency peak of the SHG modulation response (see Figure 5b of the main text).

A comparison of the calculated EO coefficient with experimental values can be performed in the DC limit. Extrapolating Eq. (56) to ω→0\omega\to0 yields r22E(0)=6.6 pm/Vr^{E}_{22}(0)=6.6\,\mathrm{pm/V}, which is larger than the commonly reported experimental DC value of approximately 2.52.5 pm/V for BBO (Ref. 20). This discrepancy likely reflects uncertainties in the experimental determination of the phonon plasma frequency, as well as the limitations of the single-phonon DFPT approach in accurately describing the low-frequency electro-optic response, which includes both electronic contributions and the cumulative effect of multiple phonon modes.

Experimental estimation of the EO coefficient from broadband THz pumping: Experimentally, the value of r22Er^{E}_{22} can be estimated from the measured SHG modulation dynamics. As discussed in the main text, the relative SHG modulation, at fixed polarisation angle α\alpha is: ΔIrel(t,α)=ΔISH(t,α)/ISH,0(α)\Delta I_{\mathrm{rel}}(t,\alpha)=\Delta I_{\mathrm{SH}}(t,\alpha)/I_{\mathrm{SH},0}(\alpha), where ISH,0(α)=ISH,0(0)cos⁡4αI_{\mathrm{SH},0}(\alpha)=I_{\mathrm{SH},0}(0)\cos^4\alpha. Measurements made at α=30∘\alpha=30^\circ were used to evaluate the EO coefficient.

Analytically, the modulation of the SH dynamics ΔISH(α,t)\Delta I_{\mathrm{SH}}(\alpha,t) in the time domain as a function of α\alpha (Eq. (2) in the main text) is:

ΔISH(α,t)=4ISH,0(0)γ(t)Λ(δph,L)Lcos⁡3αsin⁡α,(57)\Delta I_{\mathrm{SH}}(\alpha,t)=4I_{\mathrm{SH},0}(0)\gamma(t)\frac{\Lambda(\delta_{\mathrm{ph}},L)}{L}\cos^3\alpha\sin\alpha, \tag*{(57)}

therefore,

ΔIrel(α,t)=4γ(t)Λ(δph,L)Lsin⁡αcos⁡α=4κ(θ)r22Q(t)Λ(δph,L)Ltan⁡α.(58)\Delta I_{\mathrm{rel}}(\alpha,t)=4\gamma(t)\frac{\Lambda(\delta_{\mathrm{ph}},L)}{L}\frac{\sin\alpha}{\cos\alpha}=4\kappa(\theta)r_{22}Q(t)\frac{\Lambda(\delta_{\mathrm{ph}},L)}{L}\tan\alpha. \tag*{(58)}

Here, κ(θ=29.3∘)=65\kappa(\theta=29.3^\circ)=65 is the crystal geometrical factor, Λ\Lambda is the parameter that accounts for the finite phonon penetration depth δph=30.2 μm\delta_{\mathrm{ph}}=30.2\,\mu\mathrm{m}, within a crystal of thickness L=300 μmL=300\,\mu\mathrm{m} and Q(t)Q(t) is the time dependent phonon coordinate. Q(t)Q(t) can be calculated using the phonon equation of motion leaving r22r_{22} as the only free parameter to find agreement with the experimentally measured relative modulation.

The phonon dynamics are described by:

Q¨+ΓQ˙+ωQ2Q=ε0VωpE~THz(t),(59)\ddot{Q}+\Gamma\dot{Q}+\omega_Q^2Q=\sqrt{\varepsilon_0V\omega_p}\widetilde{E}_{\mathrm{THz}}(t), \tag*{(59)}

where ε0\varepsilon_0 is the vacuum permittivity, V=596 A˚3V=596\,\text{\AA}^3 is the unit cell volume, and ωQ=4.32\omega_Q=4.32 THz, ωp=6.57\omega_p=6.57 THz and Γ=0.19\Gamma=0.19 THz are the phonon frequency, phonon plasma frequency and the linewidth measured by THz-FTIR spectroscopy. E~THz(t)\widetilde{E}_{THz}(t) is the THz field transmitted through the air-sample interface: E~THz(t)=12π∫t~(ω)ETHz(ω)e−iωtdω\widetilde{E}_{THz}(t)=\frac{1}{2\pi}\int\widetilde{t}(\omega)E_{THz}(\omega)e^{-i\omega t}d\omega, where t~(ω)\widetilde{t}(\omega) is the complex Fresnel transmission calculated from the BBO refractive index (determined by THz FTIR), ETHz(ω)E_{THz}(\omega) is the spectrum of the THz pump field measured by electro-optic sampling (EOS).

Solving Eq. (59) for Q(t)Q(t) and substituting into Eq. (58), agreement between the calculated relative modulation and measurement is obtained for a phonon-optic coupling coefficient, r22=0.0130r_{22}=0.0130 Å−1^{-1} (amu)1/2^{1/2} (Figure S7) with corresponding EO coefficient, r22−maxE=145r^{E}_{22-\mathrm{max}}=145 pmV\frac{\mathrm{pm}}{\mathrm{V}}.

Estimation of EO coefficient from narrowband THz pumping: Broadband THz pumping contains spectral components far from the phonon resonance which may interfere with the determination of r22Er^{E}_{22}. Therefore, an important cross-check is to evaluate r22Er^{E}_{22} with a narrowband THz excitation using band-pass THz filters to isolate spectral components centered around the phonon frequency.

The optical setup was based on a 1-kHz repetition-rate Ti:sapphire laser. THz radiation was generated in a DSTM crystal and driven by OPA signal pulses with an energy of 0.44 mJ per pulse. The resulting THz waveform and the corresponding field-strength estimation were obtained via EOS using a 200-μm-thick GaP crystal. The THz electric field was determined using the conventional expression (see, e.g., [21]), including the Fresnel transmission coefficient. The measured temporal waveform and its Fourier amplitude spectrum are shown in Supplementary Fig. 8a and Supplementary Fig. 8b, respectively.

The relative SH modulation dynamics ΔIrel(α,t)\Delta I_{\mathrm{rel}}(\alpha,t), were recorded at the polarization angle (α=30∘\alpha=30^\circ), and the corresponding Fourier amplitude spectrum is shown in Supplementary Fig. 8c and Supplementary Fig. 8d, respectively.

Following the same procedure as in the broadband THz case, the phonon optic coupling coefficient was found to be r22=0.0160r_{22}=0.0160 Å−1^{-1} (amu)1/2^{1/2}, which corresponds to r22−maxE=178r^{E}_{22-\mathrm{max}}=178 pmV\frac{\mathrm{pm}}{\mathrm{V}}, comparable to the value obtained using the broadband THz excitation.

Supplementary References

  1. Blanchard, F. et al. Improved terahertz two-color plasma sources pumped by high intensity laser beam. Opt Express 17, 6044–6052 (2009).

  1. Valverde-Chávez, D. A. and Cooke, D. G. Multi-Cycle Terahertz Emission from β-Barium Borate. J Infrared Millim Terahertz Waves 38, 96–103 (2017).

  1. Li, R. et al. Optical properties of barium borate crystal in the THz range revisited. Opt Lett 50, 686–689 (2025).

  1. Liu, J., Guo, X., Dai, J. and Zhang, X.-C. Optical property of beta barium borate in terahertz region. Appl Phys Lett 93, 171102 (2008).

  1. Kuzmenko, A. B. Kramers–Kronig constrained variational analysis of optical spectra. Rev Sci Instrum 76, 083108 (2005).

  1. Ney, P., Fontana, M. D., Maillard, A. and Polgár, K. Assignment of the Raman lines in single crystal barium metaborate. J Phys: Condens Matter 10, 673–681 (1998).

  1. Sutherland, R. L. Handbook of Nonlinear Optics (CRC Press, Boca Raton, 2003).

  1. Yariv, A. Quantum Electronics (John Wiley & Sons, New York, 1988).

  1. Li, R. Vibrational contributions to the electro-optic effect of BBO, KTP and RTP crystals. Comput Mater Sci 230, 112529 (2023).

  1. Abarkan, M., Salvestrini, J. P., Fontana, M. D. and Aillerie, M. Frequency and wavelength dependences of electro-optic coefficients in inorganic crystals. Appl Phys B 76, 765–769 (2003).

  1. Boyd, R. W. Nonlinear Optics (Third Edition) (Academic Press, Burlington, 2008).

  1. Zhang, D., Kong, Y. and Zhang, J. Optical parametric properties of 532-nm-pumped beta-barium-borate near the infrared absorption edge. Opt Commun 184, 485–491 (2000).

  1. Mankowsky, R., von Hoegen, A., Först, M. and Cavalleri, A. Ultrafast Reversal of the Ferroelectric Polarization. Phys Rev Lett 118, 197601 (2017).

  1. Grishunin, K. A. et al. THz Electric Field-Induced Second Harmonic Generation in Inorganic Ferroelectric. Sci Rep 7, 687 (2017).

  1. Bodrov, S. B., Sergeev, Yu. A., Korytin, A. I. and Stepanov, A. N. Terahertz-field-induced second optical harmonic generation from Si (111) surface. Phys Rev B 105, 035306 (2022).

References

  1. [1]Boyd, R. W. Nonlinear Optics (Third Edition) (Academic Press, Burlington, 2008).
  2. [2]Dutt, A., Mohanty, A., Gaeta, A. L. and Lipson, M. Nonlinear and quantum photonics using integrated optical materials. Nature Reviews Materials 9, 321–346 (2024).DOI
  3. [3]Chang, D. E., Vuletić, V. and Lukin, M. D. Quantum nonlinear optics - Photon by photon. Nature Photonics 8, 685–694 (2014).
  4. [4]Hickstein, D. D. et al. Self-organized nonlinear gratings for ultrafast nanophotonics. Nature Photonics 13, 494–499 (2019).DOI
  5. [5]Casacio, C. A. et al. Quantum-enhanced nonlinear microscopy. Nature 594, 201–206 (2021).
  6. [6]Parodi, V. et al. Nonlinear Optical Microscopy: From Fundamentals to Applications in Live Bioimaging. Frontiers in Bioengineering and Biotechnology 8, (2020).DOI
  7. [7]Bogaerts, W. et al. Programmable photonic circuits. Nature 586, 207–216 (2020).DOI
  8. [8]Lukin, D. M. et al. 4H-silicon-carbide-on-insulator for integrated quantum and nonlinear photonics. Nature Photonics 14, 330–334 (2020).DOI
  9. [9]Moody, G., Chang, L., Steiner, T. J. and Bowers, J. E. Chip-scale nonlinear photonics for quantum light generation. AVS Quantum Science 2, 014702 (2020).DOI
  10. [10]Liu, X. and Fu, H. Highly-coherent second-harmonic generation in a chip-scale source. Light Science & Applications 13, 20 (2024).DOI
  11. [11]Widhalm, A. et al. Electric-field-induced second harmonic generation in silicon dioxide. Optics Express 30, 4867–4874 (2022).
  12. [12]Chen, S., Li, K. F., Li, G., Cheah, K. W. and Zhang, S. Gigantic electric-field-induced second harmonic generation from an organic conjugated polymer enhanced by a band-edge effect. Light Science & Applications 8, 17 (2019).DOI
  13. [13]Puckett, M. W. et al. Observation of second-harmonic generation in silicon nitride waveguides through bulk nonlinearities. Optics Express 24, 16923–16933 (2016).DOI
  14. [14]Timurdogan, E., Poulton, C. V., Byrd, M. J. and Watts, M. R. Electric field-induced second-order nonlinear optical effects in silicon waveguides. Nature Photonics 11, 200–206 (2017).DOI
  15. [15]Franchi, R., Castellan, C., Ghuliñyan, M. and Pavesi, L. Second-harmonic generation in periodically poled silicon waveguides with lateral p-i-n junctions. Optics Letters 45, 3188-3191 (2020).
  16. [16]Huang, B. et al. Electrical control of 2D magnetism in bilayer CrI₃. Nature Nanotechnology 13, 544–548 (2018).DOI
  17. [17]Seyler, K. L. et al. Electrical control of second-harmonic generation in a WSe₂ monolayer transistor. Nature Nanotechnology 10, 407–411 (2015).DOI
  18. [18]Lee, K. T. et al. Electrically Biased Silicon Metasurfaces with Magnetic Mie Resonance for Tunable Harmonic Generation of Light. ACS Photonics 6, 2663–2670 (2019).DOI
  19. [19]Cai, W., Vasudev, A. P. and Brongersma, M. L. Electrically Controlled Nonlinear Generation of Light with Plasmonics. Science 333, 1720–1723 (2011).DOI
  20. [20]Klimmer, S. et al. All-optical polarization and amplitude modulation of second-harmonic generation in atomically thin semiconductors. Nature Photonics 15, 837–842 (2021).DOI
  21. [21]Liu, D. et al. Light-Triggered Reversible Tuning of Second-Harmonic Generation in a Photoactive Plasmonic Molecular Nanocavity. Nano Letters 23, 5851–5858 (2023).DOI
  22. [22]Taghinejad, M., Xu, Z., Lee, K. T., Lian, T. and Cai, W. Transient Second-Order Nonlinear Media: Breaking the Spatial Symmetry in the Time Domain via Hot-Electron Transfer. Physical Review Letters 124, 013901 (2020).DOI
  23. [23]Li, G. C. et al. Light-induced symmetry breaking for enhancing second-harmonic generation from an ultrathin plasmonic nanocavity. Nature Communications 12, 4326 (2021).DOI
  24. [24]Belvin, C. A. et al. Exciton-driven antiferromagnetic metal in a correlated van der Waals insulator. Nature Communications 12, 4837 (2021).DOI
  25. [25]Bodrov, S. B., Sergeev, Yu. A., Korytin, A. I. and Stepanov, A. N. Terahertz-field-induced second optical harmonic generation from Si(111) surface. Physical Review B 105, 035306 (2022).DOI
  26. [26]Cornet, M., Degert, J., Abraham, E. and Freysz, E. Terahertz-field-induced second harmonic generation through Pockels effect in zinc telluride crystal. Optics Letters 39, 5921–5924 (2014).DOI
  27. [27]Chen, J., Han, P. and Zhang, X.-C. Terahertz-field-induced second-harmonic generation in a beta barium borate crystal and its application in terahertz detection. Applied Physics Letters 95, 011118 (2009).DOI
  28. [28]Bodrov, S. B., Sergeev, Yu. A., Korytin, A. I., Burova, E. A. and Stepanov, A. N. Terahertz pulse induced femtosecond optical second harmonic generation in transparent media with cubic nonlinearity. Journal of the Optical Society of America B 37, 789–796 (2020).DOI
  29. [29]Ovchinnikov, A. V, Chefоnov, O. V, Mishina, E. D. and Agranat, M. B. Second harmonic generation in the bulk of silicon induced by an electric field of a high power terahertz pulse. Scientific Reports 9, 9753 (2019).DOI
  30. [30]Cammarata, A. and Rondinelli, J. M. Contributions of Correlated Acentric Atomic Displacements to the Nonlinear Second Harmonic Generation and Response. ACS Photonics 1, 96–100 (2014).DOI
  31. [31]Li, R. Vibrational contributions to the electro-optic effect of BBO, KTP and RTP crystals. Computational Materials Science 230, 112529 (2023).DOI
  32. [32]Lin, J., Lee, M.-H., Liu, Z.-P., Chen, C. and Pickard, C. J. Mechanism for linear and nonlinear optical effects in β-BaB₂O₄ crystals. Physical Review B 60, 13380–13389 (1999).
  33. [33]Li, R. et al. Optical properties of barium borate crystal in the THz range revisited. Optics Letters 50, 686–689 (2025).
  34. [34]Vicario, C., Trisorio, A., Allenspach, S., Rüegg, C. and Giorgianni, F. Narrow-band and tunable intense terahertz pulses for mode-selective coherent phonon excitation. Applied Physics Letters 117, 101101 (2020).DOI
  35. [35]Kozina, M. et al. Terahertz-driven phonon upconversion in SrTiO₃. Nature Physics 15, 387–392 (2019).DOI
  36. [36]Melnikov, A. A. et al. Coherent phonons in a Bi₂Se₃ film generated by an intense single-cycle THz pulse. Physical Review B 97, 214304 (2018).DOI
  37. [37]Mankowsky, R., von Hoegen, A., Först, M. and Cavalleri, A. Ultrafast Reversal of the Ferroelectric Polarization. Physical Review Letters 118, 197601 (2017).
  38. [38]Grishunin, K. A. et al. THz Electric Field-Induced Second Harmonic Generation in Inorganic Ferroelectric. Scientific Reports 7, 687 (2017).DOI
  39. [39]McDonnell, C., Carletti, L., Vincenti, M. A., Della Valle, G., De Angelis, C., Celebrano, M., and Ellenbogen, T. THz Field Induced Second Harmonic Generation in Epsilon Near Zero Indium Tin Oxide Thin Films. Nano Letters 25, 12201-12206 (2025).DOI
  40. [40]Itoh, H. et al. Terahertz Field Control of Electronic-Ferroelectric Anisotropy at Room Temperature in LuFe₂O₄, Physical Review Letters 135, 106504 (2025).
  41. [41]Stegeman, G. I., Hagan, D. J. and Torner, L. χ⁽²⁾ cascading phenomena and their applications to all-optical signal processing, mode-locking, pulse compression and solitons. Optical and Quantum Electronics 28, 1691–1740 (1996).
  42. [43]Gerber, S. et al. Femtosecond electron-phonon lock-in by photoemission and x-ray free-electron laser. Science 357, 71–75 (2017).DOI
  43. [44]Chelladurai, D. et al. Barium titanate and lithium niobate permittivity and Pockels coefficients from megahertz to sub-terahertz frequencies. Nature Materials 24, 868–875 (2025).DOI
  44. [45]Goodno, G. D. et al. Investigation of β-BaB2O4 as a Q switch for high power applications. Applied Physics Letters 66, 1575–1577 (1995).DOI
  45. [46]Disa, A. S., Nova, T. F. and Cavalleri, A. Engineering crystal structures with light. Nature Physics 17, 1087–1092 (2021).
  46. [47]Fu, W. et al. Phononic integrated circuitry and spin–orbit interaction of phonons. Nature Communications 10, 2743 (2019).DOI
  47. [48]von Hoegen, A., Mankowsky, R., Fechner, M., Först, M. and Cavalleri, A. Probing the interatomic potential of solids with strong-field nonlinear phononics. Nature 555, 79–82 (2018).DOI
  48. [49]Baroni, S., de Gironcoli, S., Dal Corso, A. and Gianozzi, P. Phonons and related crystal properties from density-functional perturbation theory. Reviews of Modern Physics 73, 515–562 (2001).DOI
  49. [50]Giannozzi, P. et al. QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials. Journal of Physics: Condensed Matter 21, 395502 (2009).DOI
  50. [51]Giannozzi, P. et al. Advanced capabilities for materials modelling with Quantum ESPRESSO. Journal of Physics: Condensed Matter 29, 465901 (2017).DOI
  51. [52]Hamann, D. R. Optimized norm-conserving Vanderbilt pseudopotentials. Physical Review B 88, 085117 (2013).
  52. [53]Schlipf, M. and Gygi, F. Optimization algorithm for the generation of ONCV pseudopotentials. Computer Physics Communications 196, 36–44 (2015).DOI
  53. [54]Materials Project. Materials Data on Ba(BO₂)₂, doi:https://doi.org/10.17188/1276454.
  54. [55]Umari, P. and Pasquarello, A. Ab initio Molecular Dynamics in a Finite Homogeneous Electric Field. Physical Review Letters 89, 157602 (2002).

Paper details

vlnc:W_xmghhdh6
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