betaFeedback

WorksPhysics and Astronomy

Ultrafast Successive Control and Electrical Detection of Long-Lived Valley Polarization in Room-Temperature Germanium

Abstract

Germanium (Ge) hosts four degenerate valleys in the conduction band minima, and their highly anisotropic effective masses offer a valley degree of freedom in transport phenomena via anisotropic conductivity. Using time- and polarization-resolved terahertz spectroscopy combined with deformation potential theory, we reveal a valley relaxation time reaching 10 ps near the band edge of Ge(001) at room temperature, long enough to detect the valley polarization in electrical transport measurements. We also demonstrate sub-picosecond complete reversal of valley polarization by off-resonant light pulses at extremely low energy consumption. These results pave the way for room-temperature valleytronics based on bulk indirect-gap semiconductors compatible with existing electronics.

Controlling the internal degrees of freedom of carriers is crucial for developing next-generation, low-power, high-efficiency electronic devices. While spintronics uses the spin degree of freedom, valleytronics exploits the energy band minima in momentum space, or the valley degree of freedom, which has recently attracted considerable attention. A prime example is monolayer transition metal dichalcogenides (TMDCs), represented by MoS2_2 [1, 2]. These materials exhibit degenerate bandgaps in the K and K′^\prime valleys, allowing valley-selective carrier excitation using circularly polarized light via optical selection rules for interband transitions [3, 4, 5]. Furthermore, a long valley polarization lifetime was initially expected because intervalley scattering requires a large momentum transfer with a simultaneous spin flip due to the spin-valley locking [6]. Recently, off-resonant [7, 8, 9] and coherent control [10] of valley polarization have also been reported. To realize practical electronic device applications in valleytronics, the detection of valley polarization through transport in electric circuits is indispensable, which is enabled by the valley Hall effect [11] owing to the valley-dependent Berry curvatures [6].

However, TMDC-based valleytronics faces critical challenges in valley lifetime, conversion into electric current, and compatibility with existing semiconductor electronics. At room temperature, valley polarization in MoS2_2 is lost within only 75 fs [12], owing to the strong interaction with zone-edge phonons [13], short-range interaction with defects and impurities, and exchange interaction between an electron and a hole [14, 15, 16]. Furthermore, direct observation of the valley Hall effect is notoriously difficult, so that very few studies have succeeded in directly demonstrating this effect since the initial pioneering reports [11, 17]. Moreover, obtaining high-quality samples requires encapsulation in hexagonal boron nitride (h-BN) [18], creating numerous hurdles for practical electronic device integration.

Historically, the relationship between the valley structure and transport responses has been investigated in conventional indirect-bandgap semiconductors, such as silicon (Si) and germanium (Ge) [19]. Figure 1(a) illustrates an example of a fourfold degenerate valley structure at the L point, forming conduction band (CB) minima in Ge. In these materials, because of the phonon-assisted indirect gap transition and highly anisotropic energy dispersion of each valley, the interband transitions induced by linearly polarized light are accompanied by different excitation probabilities for each valley, thereby creating valley polarization, as shown in Figure 1(b) [20, 21]. Once valley polarization is generated, the difference between the longitudinal and transverse effective masses, mℓem_{\ell}^{e} and mtem_{t}^{e}, respectively, in each valley leads to anisotropic conductivity. Recently, light-induced valley polarization dynamics in Si has been investigated using terahertz (THz) pulses [22], where anisotropic conductivity can be probed as linear birefringence and dichroism in the THz frequency range. After injecting valley-polarized carriers into the band edge using near-infrared (NIR) pulses, the relaxation time of valley polarization exceeding 1.5 ps at room temperature—20 times longer than that of MoS2_2—was revealed in the time-resolved measurements [22]. Additionally, off-resonant manipulation of valley polarization using light with a photon energy lower than the bandgap has recently been demonstrated in diamond and Si [23], highlighting its potential for applications using conventional semiconductors. Compared with the effective mass anisotropy mle/mte=4.8m_l^e/m_t^e = 4.8 in Si, Ge exhibits an even higher anisotropy (mle/mte=19.5m_l^e/m_t^e = 19.5) as well as a higher carrier mobility. Although the ultrafast Γ\Gamma–LL intervalley scattering in Ge has been intensively studied due to its strong relevance to high-performance transistors and optoelectronic devices [24, 25, 26], the valley polarization dynamics at the band bottom determined by the LL–LL intervalley scattering has evaded experimental observation at room temperature. A pioneering acoustoelectric experiment suggested a long valley relaxation time of approximately 100 ps in Ge at temperatures below 100 K [27]. Given its seamless integration with current semiconductor technologies, Ge is a promising candidate for future applications in valleytronic devices.

First Brillouin zone, light-induced valley polarization, and photocurrent detection

Figure 1. (a) First Brillouin zone of Ge and its projection corresponding to the sample orientation in our experiment. (b) Schematic of light-induced valley polarization generated by the NIR light. (c) Schematic showing the generation of a photocurrent jyj_y orthogonal to the bias field via the valley polarization and its detection as the THz linear birefringence and/or dichroism.

In this Letter, we explore possible valleytronics in bulk Ge at room temperature. We investigate the dynamics of valley polarization by exciting carriers at the indirect bandgap edge of bulk Ge(001) using wavelength-tunable NIR pulses and measuring the resulting anisotropic conductivity through the polarization rotation of the THz pulses [Fig. 1(c)]. Furthermore, by comparing our results with theoretical calculations incorporating deformation potentials, we successfully discriminate between the momentum relaxation dynamics of holes asymmetrically distributed near the Γ\Gamma point and the valley relaxation dynamics of electrons, revealing that electrons in Ge exhibit an exceptionally long valley relaxation time of nearly 10 ps. This long lifetime enables detection via time-integrated direct current (DC) transport measurements. In addition, by accelerating electrons using a linearly polarized mid-infrared light field, we demonstrate off-resonant generation and successive, sub-picosecond switching of valley polarization with ultralow energy consumption. This study paves the way for ultrafast THz information processing based on valleytronics in indirect-gap semiconductors.

The sample was a 100-μm-thick bulk Ge crystal. All experiments were performed at room temperature. First, we conducted time-resolved THz polarimetry by employing three optical pulses: NIR pump, THz probe, and NIR gate pulses, with the pulse sequence shown in Figure 2(a). NIR pump pulses were generated using an optical parametric amplifier with wavelengths tuned between 1650 and 1850 nm, corresponding to photon energies near the indirect bandgap but below the direct one [28]. The THz probe pulses were generated via optical rectification in a GaP crystal and detected by electro-optic sampling using 1030-nm gate pulses with a pulse duration shorter than 100 fs. The pump pulses were linearly polarized along the [110] or [\bar{1}10] crystal axis so that the valley polarization can be generated on the LL point [Figs. 1(a) and 1(b)]. To detect the valley polarization via THz polarization rotation, the incident THz pulse was polarized along the [100] direction, which we chose as the horizontal (xx) axis in the laboratory frame. Two wire-grid polarizers were positioned after the sample to separately detect the respective xx- and yy-polarization components of the transmitted THz pulse, ExE_x and EyE_y. From the polarization rotation Ey/ExE_y/E_x, the anisotropic conductivity σani≡σ[110]−σ[1ˉ10]\sigma_{\mathrm{ani}} \equiv\sigma_{[110]}-\sigma_{[\bar{1}10]}, which represents the conductivity difference between the [110] and [1ˉ10][\bar{1}10] crystal axes, was extracted. To eliminate the offsets independent of the pump polarization direction, the difference between the two EyE_y signals for the mutually orthogonal pump polarization directions along the [110] and [1ˉ10][\bar{1}10] crystal axes was obtained (for details, see Supplemental Material [29]). Notably, not only are electrons valley polarized, but the holes are also asymmetrically distributed around the Γ\Gamma point [35, 36] and contribute to the anisotropic conductivity.

Time sequence, THz field changes, relaxation components and times, calculated scattering times, and valley polarization ratio

Figure 2. (a) Time sequence of the NIR pump, THz probe, and gate pulses. (b) Two-dimensional plots of the THz field change parallel [ΔEx(tprobe,tpump)\Delta E_x(t_{\mathrm{probe}},t_{\mathrm{pump}})] and perpendicular [ΔEy(tprobe,tpump)\Delta E_y(t_{\mathrm{probe}},t_{\mathrm{pump}})] to the incident probe polarization. (c) ΔEy(tpump)\Delta E_y(t_{\mathrm{pump}}) at a fixed tprobet_{\mathrm{probe}} for a pump wavelength of 1700 nm (black line). Fast and slow relaxation components extracted from the double-exponential fitting (filled using light blue and dark blue colors). (d) Fast and slow relaxation times as functions of the excess photon energy ℏω−Eg\hbar\omega- E_g. (e) Solid and dashed lines represent the calculated electron intervalley scattering time τe\tau_e and hole–phonon scattering time τh\tau_h, respectively. (f) Estimated valley polarization ratio PvP_v as a function of the excess photon energy. The dashed line marks the LA phonon energy.

Figure 2(b) shows the two-dimensional (2D) plots of the change in transmitted THz field induced by the 1700 nm NIR pump as functions of the delay times of the THz probe and NIR pump pulses, tprobet_{\mathrm{probe}} and tpumpt_{\mathrm{pump}}, respectively; the left and right panels show parallel [ΔEx(tprobe,tpump)][\Delta E_x(t_{\mathrm{probe}},t_{\mathrm{pump}})] and perpendicular [ΔEy(tprobe,tpump)][\Delta E_y(t_{\mathrm{probe}},t_{\mathrm{pump}})] components to the incident field. ΔEx(tprobe,tpump)\Delta E_x(t_{\mathrm{probe}},t_{\mathrm{pump}}) represents the decrease in transmission due to photocarrier absorption, whereas ΔEy(tprobe,tpump)\Delta E_y(t_{\mathrm{probe}},t_{\mathrm{pump}}) captures the anisotropic conductivity due to the pump-induced valley polarization. In contrast to previous observations in Si [22], the diagonally oscillating signal around tpump=0t_{\mathrm{pump}}=0, indicating the field-induced photogalvanic effect [37, 38], is small in Ge; thus, we neglect it in the following discussion.

To analyze the anisotropic conductivity dynamics more quantitatively, ΔEy\Delta E_y at a fixed probe delay was measured as a function of tpumpt_{\mathrm{pump}} at various pump wavelengths. Figure 2(c) presents ΔEy(tpump)\Delta E_y(t_{\mathrm{pump}}) at a fixed tprobet_{\mathrm{probe}} for the 1700 nm pump at a fluence of 11 μJ/cm2\mu\mathrm{J}/\mathrm{cm}^2. Two distinct relaxation components, fast and slow, were isolated using a double-exponential fit convolved with a Gaussian instrument response function (see Supplemental Material [29]). The fast and slow components are plotted separately in Figure 2(c). The extracted relaxation times are plotted in Figure 2(d) against ℏω−Eg\hbar\omega-E_g, the excess energy above the bandgap (Eg=0.66E_g=0.66 eV). Both the fast and slow relaxation times increased with decreasing excess energy, which is accounted for by fewer scattering channels for small excess energies. Near the band edge, the fast relaxation time was 1.4 ps, whereas the slow relaxation time reached approximately 10 ps.

For comparison, we calculated the inverse of the L–L intervalley scattering rate of electrons τe−1(ε)\tau_e^{-1}(\varepsilon) and the hole-phonon scattering rate τh−1(ε)\tau_h^{-1}(\varepsilon) based on the deformation potential theory [34]:

τe−1(ε)=∑j=LA(LO),TA∑i=±1Z(DLL,j)2(mde)3/22πℏ3ρωjX(Nph(ωjX)+1−i2)ε+iℏωjX(1)\tau_e^{-1}(\varepsilon)=\sum_{j=\mathrm{LA(LO)},\mathrm{TA}}\sum_{i=\pm1}\frac{Z(D_{\mathrm{LL},j})^2(m_d^e)^{3/2}}{\sqrt{2}\pi\hbar^3\rho\omega_j^X}\left(N_{\mathrm{ph}}(\omega_j^X)+\frac{1-i}{2}\right)\sqrt{\varepsilon+i\hbar\omega_j^X} \tag*{(1)}
τh−1(ε)=PΓ,Oh(ε)+PΓ,Ah(ε)(2)\tau_h^{-1}(\varepsilon)=P_{\Gamma,O}^h(\varepsilon)+P_{\Gamma,A}^h(\varepsilon) \tag*{(2)}
PΓ,Oh(ε)=∑i=±13(DOh)2(mdh)3/223/2πℏ3ρωOΓ(Nph(ωOΓ)+1−i2)ε+iℏωOΓ,(3)P_{\Gamma,O}^{h}(\varepsilon)=\sum_{i=\pm1}\frac{3(D_{O}^{h})^{2}(m_{d}^{h})^{3/2}}{2^{3/2}\pi\hbar^{3}\rho\omega_{O}^{\Gamma}}\left(N_{\mathrm{ph}}(\omega_{O}^{\Gamma})+\frac{1-i}{2}\right)\sqrt{\varepsilon+i\hbar\omega_{O}^{\Gamma}}, \tag*{(3)}
PΓ,Ah(ε)=2(DAh)2kBT(mdh)3/2πℏρul2ε,(4)P_{\Gamma,A}^{h}(\varepsilon)=\frac{\sqrt{2}(D_{A}^{h})^{2}k_{B}T(m_{d}^{h})^{3/2}}{\pi\hbar\rho u_{l}^{2}}\sqrt{\varepsilon}, \tag*{(4)}

where ρ\rho is the mass density of Ge and i=±1i=\pm1 represent phonon absorption and emission, respectively. For electrons, Z=3Z=3 is the number of final valleys available for scattering; DLL,jeD_{\mathrm{LL},j}^{e} is the L–L intervalley deformation potential for a phonon mode jj, where the relevant phonons are LA, LO, and TA, with LA and LO phonons being degenerate; ωjX\omega_{j}^{X} is the phonon angular frequency at the X point; and mdem_{d}^{e} is the electron density-of-states effective mass. For holes, DOhD_{O}^{h} and DAhD_{A}^{h} denote the optical- and acoustic-phonon deformation potentials near the Γ\Gamma point, respectively. ωOΓ\omega_{O}^{\Gamma} is the optical phonon angular frequency at the Γ\Gamma point, mdhm_{d}^{h} is the hole density-of-states effective mass, ulu_{l} is the longitudinal sound velocity, and Nph(ω)N_{\mathrm{ph}}(\omega) is the phonon occupation number, 1/(eℏω/kBT−1)1/(e^{\hbar\omega/k_{B}T}-1). Because acoustic phonons near the Γ\Gamma point have sufficiently small energy, emission and absorption processes were integrated into a single expression. These formulations describe the carrier-phonon coupling strength resulting from local lattice deformations. All parameters were obtained from the literature [34] (for details, see Supplemental Material [29]). The carrier-carrier and impurity scattering mechanisms were neglected, as their rates were substantially smaller than the carrier-phonon scattering rates at photogenerated carrier densities below 1017/cm310^{17}/\mathrm{cm}^{3}. The calculated results are presented in Figure 2(e) as a function of the carrier kinetic energy ε\varepsilon for comparison with the observed relaxation times in Figure 2(d). The kinks in τh(ε)\tau_{h}(\varepsilon) near ε≈37\varepsilon\approx37 meV and in τe(ε)\tau_{e}(\varepsilon) near ε≈28\varepsilon\approx28 meV arise from the onset of emission of zone-center optical phonons and LA phonons at the X point, respectively. The fast relaxation times in Figure 2(d) align with τh(ε)\tau_{h}(\varepsilon), corresponding to the relaxation of anisotropic conductivity arising from the asymmetric hole distribution. Accordingly, the slow relaxation component is attributed to τe(ε)\tau_{e}(\varepsilon), the valley polarization lifetime of electrons. Note that the observed slow relaxation times in Figure 2(d) are systematically longer than the calculated τe(ε)\tau_{e}(\varepsilon). Especially at large excess energies, the slow relaxation times saturate whereas τe(ε)\tau_{e}(\varepsilon) continues to decrease. This is reasonable because the photoexcited hot carriers would immediately start to thermalize within each valley, so that intervalley scattering effectively takes place at kinetic energies lower than the initial excess energy. Consequently, these distinct relaxation timescales allow us to distinguish the electron and hole contributions to the anisotropic conductivity. The long relaxation time of ∼10\sim10 ps near the band edge is attributed to the valley-polarization relaxation of electrons, which is consistent with an extrapolation of previous low-temperature acoustoelectric experiments below 100 K to room temperature [27]. Remarkably, the valley polarization lifetime in Ge is an order of magnitude longer than that of Si [22], which can be attributed to its smaller density-of-states mass and higher mass density compared with Si (see Supplemental Material [29]).

Assuming a parabolic band dispersion, the relation between the anisotropic conductivity σani\sigma_{\mathrm{ani}} and valley polarization ratio PvP_v can be expressed through the Drude model as

σani=(1m∥e−1m⊥e)Ncee2τD1−iωτDPv,(5)\sigma_{\mathrm{ani}}=\left(\frac{1}{m_{\parallel}^{e}}-\frac{1}{m_{\perp}^{e}}\right)\frac{N_c^{e}e^2\tau_D}{1-i\omega\tau_D}P_v, \tag*{(5)}

where m∥em_{\parallel}^{e} and m⊥em_{\perp}^{e} are the effective masses parallel and perpendicular to the NIR polarization direction, respectively. NceN_c^{e} is the electron density, ee is the elementary charge, and τD\tau_D is the momentum relaxation time. By modeling wave propagation through an optically anisotropic medium via anisotropic conductivity, the relationship between the THz polarization rotation angle and valley polarization ratio was established for each photon energy (see Supplemental Material [29]). The estimated valley polarization ratio is presented in Figure 2(f) as a function of the excess energy ℏω−Eg\hbar\omega-E_g. Intriguingly, the valley polarization ratio increased with the excess energy and reached 4% with an abrupt transition near the LA phonon energy of 28 meV. Because LA phonons play a dominant role in indirect transitions in Ge [39], the interband transition with the emission of an LA phonon may play a key role in inducing valley polarization.

To explore the detectability of the valley polarization in DC transport, we define a conductivity anisotropy angle θani≡σyx/σxx\theta_{\mathrm{ani}}\equiv\sigma_{yx}/\sigma_{xx} (Supplemental Material [29]), which scales with PvP_v. In analogy to the spin Hall angle θH=σH/σxx\theta_H=\sigma_H/\sigma_{xx}, θani\theta_{\mathrm{ani}} characterizes the detection efficiency of valley information through transport. We evaluated θani≈0.013\theta_{\mathrm{ani}}\approx0.013 for Pv=4%P_v=4\% (Supplemental Material [29]). Compared with θH\theta_H in semiconductors used for spintronics at room temperature [40] (∼10−4\sim10^{-4} for p-Si [41] and ∼10−3\sim10^{-3} for n-Ge [42]), our results show a remarkably large conductivity anisotropy angle, suggesting that the valley polarization can be detected as a time-integrated transport response. Thus, as depicted in Figure 3(a), we replaced the THz field with a static DC bias voltage along the positive xx direction, and measured the resulting transverse photovoltage along the yy direction in a Hall bar configuration, with a lock-in amplifier referenced to the repetition rate of the pump pulses. The signal is defined as positive when the current flows along the positive yy direction. Linearly polarized NIR pump pulses at 1800 nm were focused between the electrodes without illuminating the contacts. The polarization angle of the NIR beam was rotated counterclockwise, as viewed along the incident beam, using a half-wave plate. Figure 3(b) shows the transverse voltage after subtracting the background offset, which reveals a clear dependence on the pump polarization angle and is well fitted by a 180° periodic sine curve. The oscillation phase and amplitude of the transverse voltage signal are plotted against the pump fluence in Figure 3(c) and Figure 3(d), respectively. The transverse voltage reached its maximum at a polarization angle of 45°, which was consistent with the valley-related anisotropic conductivity geometry induced by a linearly polarized NIR pump. The amplitude exhibited a nearly linear scaling at low fluences, ruling out nonlinear optical artifacts. While the transport signal started to deviate from linear scaling at higher fluences, the THz polarization rotation signal overlaid in Figure 3(d), defined as the peak value of ΔEy(tpump)\Delta E_y(t_{\mathrm{pump}}) as in Figure 2(c), remained linear up to higher fluences. This difference can be attributed to the distinct heating effects of the DC and THz pulse measurements. These results demonstrate that anisotropic conductivity is easily detectable as a DC transport signal owing to the long lifetime of the valley polarization in Ge, confirming the efficient conversion from the valley degree of freedom to charge transport that constitutes an indispensable building block for constructing valleytronic devices.

Schematic of the experimental setup and transport measurements

Figure 3. (a) Schematic of the experimental setup for transport measurements. A bias voltage is applied along the positive xx axis. The transverse voltage is positive for currents flowing along the positive yy axis. (b) Transverse voltage depending on the pump polarization direction. Dashed curves denote a sine function fitting. Red arrows show the NIR polarization directions viewed along the light propagation path (from source to sample), where 0∘0^\circ aligns with the xx axis. (c) Phase corresponding to the oscillation peak of the transverse voltage as a function of the pump fluence. (d) Pump-fluence dependence of the oscillation amplitude of the transverse voltage (dark blue, left axis) and the THz polarization rotation ΔEy\Delta E_y (orange, right axis).

Furthermore, the strong dependence of the valley polarization lifetime on the carrier kinetic energy directly observed in Figure 2(d) has another important implication for the controllability of valley polarization using an off-resonant light pulse [23]. According to the acceleration theorem, an off-resonant field oscillates the carrier momentum, imparting kinetic (ponderomotive) energy UpU_p on the Bloch electrons. Because of the effective mass anisotropy of each valley, the effective mass along the field polarization direction differs among the valleys. This implies that the imparted kinetic energy is valley-dependent; electrons in valleys with a smaller effective mass along the field polarization direction acquire a larger kinetic energy than those in valleys with a larger effective mass, as illustrated in Figure 4(a). This kinetic energy imbalance causes a difference in the intervalley scattering probabilities, as shown in Figure 2(d), thereby generating valley polarization depending on the field polarization direction.

Schematic of carrier momentum shift, pulse sequence, THz polarization rotation dynamics, and relaxation times

Figure 4. (a) Schematic of carrier momentum shift induced by the off-resonant electric field (yellow arrow) within each valley, resulting in unbalanced intervalley scattering. (b) Pulse sequence of the NIR pump, MIR drive, THz probe, and gate pulses. ΔEy\Delta E_y as a function of tdrivet_{\mathrm{drive}} for (c) single and (d) dual MIR pulses with mutually orthogonal polarizations. (e) Experimental THz polarization rotation dynamics together with the double-exponential fits (upper panel). The dashed curve indicates the slow decay component. Monte Carlo (MC) simulation of the valley polarization ratio PvP_v with its single-exponential fit (lower panel). (f) Relaxation times obtained from the experiment (orange filled squares) and the MC simulation (black open) as functions of MIR fluence.

To demonstrate this off-resonant control of valley polarization in Ge, a 4-μm mid-infrared (MIR) drive pulse was applied to the sample at a delay of approximately 100 ps after the NIR pump, as represented by the pulse sequence in Figure 4(b). The resulting valley polarization was probed using THz birefringence and dichroism. The ~100 ps delay from the NIR pump ensures that the photoexcited carriers are equally distributed among the L valleys before the arrival of the MIR pulse because it is much longer than the valley polarization lifetime (~10 ps) but much shorter than the carrier recombination time in Ge on the nanosecond scale. Figure 4(c) shows the THz polarization rotation signal ΔEy\Delta E_y at a fixed probe delay as a function of tdrivet_{\mathrm{drive}}, a fluence of 53 μJ/cm². An anisotropic conductivity signal ΔEy\Delta E_y induced by the MIR drive pulse was successfully observed, demonstrating the controllability of the valley polarization by off-resonant light without changing the carrier density (Supplemental Material [29]). The resulting PvP_v was ~0.6%, corresponding to θani∼2×10−3\theta_{\mathrm{ani}} \sim2 \times10^{-3}, which still exceeds θH\theta_H of n-Ge [42] and is thus also large enough for electrical detection. Furthermore, Figure 4(d) shows the result obtained by applying dual MIR drive pulses with mutually orthogonal polarizations and a 0.87-ps interval. After the first MIR pulse generated valley polarization, the second MIR pulse clearly switched this imbalance to the other pair of valleys. The interval between the positive and negative peaks for this reversal was approximately 0.8 ps and matched the pulse delay. This sub-picosecond complete reversal highlights the potential for ultrafast and arbitrary valley manipulation in Ge, as discussed later.

To confirm the underlying mechanism of MIR-induced valley polarization, we conducted Monte Carlo simulations of electron scattering [23]. The model incorporated two processes: the momentum shift of carriers driven by the 4-μ\mum MIR field, and the electron-phonon scattering. The latter included L–L intervalley scattering mediated by LA, LO, and TA phonons, as well as intravalley scattering mediated by optical and acoustic phonons. At each time step, we stochastically calculated the change in the electron distribution considering the inter- and intravalley scattering channels. In the intervalley process, an electron exchanged energy with a mediating optical or acoustic phonon and was randomly moved to one of the other three L valleys. In the intravalley process, an electron was scattered into a random momentum state within the same valley, and we neglected the energy of acoustic phonons. The initial carrier distribution followed the Fermi–Dirac distribution, with the chemical potential determined from the experimentally estimated carrier density. Each electron iteratively underwent the field-acceleration and scattering-determination processes with sufficiently small time steps. The MIR fluence was adjusted within the range used in the experiments. Further details are provided in the Supplemental Material [29]. Figure 4(e) shows the experimentally obtained anisotropic conductivity dynamics (upper panel) and the simulated valley polarization ratio of electrons (lower panel). As the experimental signal included both electron and hole contributions, the electron component was isolated by extracting only the slow-relaxation component using a double-exponential fit, as in Figure 2(c). The simulated valley polarization ratio was fitted by a single-exponential function to extract the relaxation time. Figure 4(f) shows the relaxation times obtained by the experiments and the Monte Carlo simulation as functions of the MIR fluence. The relaxation times are approximately 3 and 4 ps for the simulation and experiments, respectively, and are robust to the drive fluence in this range. Compared with the valley relaxation dynamics investigated in Figure 2(d), the electron system driven by the MIR pulses corresponds well to the photoexcited one with the kinetic energy of 50-100 meV. While the valley relaxation in Figure 2(d) was explained by the energy-dependent intervalley scattering rate alone, the present simulation further incorporates the field acceleration and intravalley scattering, and still reproduces the observed relaxation well.

Finally, we would like to underscore that the sub-picosecond complete valley reversal demonstrated in Figure 4(d) promises novel ultrafast information processing, which may far exceed the switching technology in spintronics. Extensive efforts have been devoted to realizing ultrafast magnetization switching in the picosecond to sub-picosecond regime using optical pulses [43, 44, 45] and in the 10-ps regime using electrical pulses [46, 47, 48, 49, 50]. However, for information processing, a key figure of merit is not the switching speed itself but the maximum repetition rate of the operation. In spin systems, the repetition rate is severely limited by heat accumulation, which causes demagnetization or, in the worst case, device failure. Therefore, to increase the repetition rate, the energy required per bit written must be suppressed to reduce total energy consumption. Very recently, ultralow power switching of an antiferromagnet driven by 40-ps current pulses was demonstrated [50]; however, the suggested energy cost of ∼1\sim1 fJ per bit relies on scaling down the device size to a 30×30×1030 \times30 \times10 nm3^3 dot. In stark contrast, the energy density imparted to the system per pulse for the valley switching in Ge presented here can be evaluated as ∫σxxEMIR2dt∼1\int\sigma_{xx} E_{\mathrm{MIR}}^2 \mathrm{d}t \sim1 aJ/μm3\mu\mathrm{m}^3 under the condition in Figure 4(d) (see Supplemental Material [29]). Note that this estimation is based on the Drude model, which is not accurate for describing the intraband electron response in the MIR range. Nonetheless, ultralow energy consumption can still be expected for the valley switching. It highlights the direct coupling between the electric field and charge carriers via intraband acceleration without additional carrier excitation, in contrast to the difficulty in electrically controlling spins. While the valley switching was demonstrated here in the nearly transparent MIR region, similar valley switching is also anticipated to occur with a much weaker field using lower-frequency, picosecond (THz) electric pulses on chip because the acquired kinetic energy UpU_p of electrons is proportional to EMIR2/ω2E_{\mathrm{MIR}}^2/\omega^2.

In conclusion, time-resolved THz polarimetry, analyzed with quantitative theoretical calculations, revealed a valley lifetime of 10 ps in Ge at room temperature. Owing to this long lifetime, the valley polarization could be read out electrically as a DC transverse voltage. We also demonstrated sub-picosecond switching of valley polarization at an energy cost of only about 1 aJ/μm3\mu\mathrm{m}^3 per pulse. These results suggest that Ge is an ideal platform for valleytronic devices for ultrafast THz information processing. Because the generation, detection, and switching of valley polarization presented here rely only on the multiple anisotropic conduction-band valleys inherent to indirect-gap semiconductors, the present scheme should also be applicable to other materials. Future studies exploring strain engineering or low-dimensional structures could further enhance the valley polarization ratio and lifetime, ultimately opening new avenues toward practical room-temperature valleytronic logic and memory devices.

Acknowledgments

A.M.S. thanks Yuto Fukushima for fruitful discussions. This work was supported by the JST FOREST (Grant Nos. JPMJFR2240 and JPMJFR213A), JSPS KAKENHI (Grant Nos. 24K00550 and 24K16988), JST PRESTO (Grant No. JPMJPR2006), and JST CREST (Grant No. JPMJCR20R4). R.M. acknowledges partial support from the MEXT Quantum Leap Flagship Program (MEXT Q-LEAP, Grant No. JPMXS0118068681). R.M. conceived the study. A.M.S. and Y.M. developed the pump–probe spectroscopy system with the help of T.F., K.A., J.Y. and

R.M. C.K. and A.M.S. prepared the sample with the support of H.A. A.M.S. performed the pump–probe experiment and analyzed the data with the help of T.F. and Y.M. and also performed the transport experiment with the help of S.A. and T.I. A.M.S. conducted the Monte Carlo simulations with the help of S.A.S. All authors discussed the results. A.M.S. and R.M. wrote the manuscript with feedback from Y.M., T.I., and all coauthors.

References

  1. [1]For a review, see e.g., J. Schaibley, H. Yu, G. Clark, P. Rivera, J. S. Ross, K. L. Seyler, W. Yao, and X. Xu, Valleytronics in 2D materials. Nat. Rev. Mater. 1, 16055 (2016). https://doi.org/10.1038/natrevmats.2016.55
  2. [2]For a review, see e.g., K. F. Mak and J. Shan, Photonics and optoelectronics of 2D semiconductor transition metal dichalcogenides. Nat. Photon. 10, 216 (2016). https://doi.org/10.1038/nphoton.2015.282
  3. [3]T. Cao, G. Wang, W. Han, H. Ye, C. Zhu, J. Shi, Q. Niu, P. Tan, E. Wang, B. Liu, and J. Feng, Valley-selective circular dichroism of monolayer molybdenum disulphide. Nat. Commun. 3, 887 (2012). https://doi.org/10.1038/ncomms1882
  4. [4]K. F. Mak, K. He, J. Shan, and T. F. Heinz, Control of valley polarization in monolayer MoS2 by optical helicity. Nat. Nanotechnol. 7, 494 (2012). https://doi.org/10.1038/nnano.2012.96
  5. [5]H. Zeng, J. Dai, W. Yao, D. Xiao, and X. Cui, Valley polarization in MoS2 monolayers by optical pumping. Nat. Nanotechnol. 7, 490 (2012). https://doi.org/10.1038/nnano.2012.95
  6. [6]D. Xiao, G.-B. Liu, W. Feng, X. Xu, and W. Yao, Coupled Spin and Valley Physics in Monolayers of MoS2 and Other Group-VI Dichalcogenides. Phys. Rev. Lett. 108, 196802 (2012). https://doi.org/10.1103/PhysRevLett.108.196802
  7. [7]F. Langer, C. P. Schmid, S. Schlauderer, M. Gmitra, J. Fabian, P. Nagler, C. Schüller, T. Korn, P. G. Hawkins, J. T. Steiner, U. Huttner, S. W. Koch, M. Kira, and R. Huber, Lightwave valleytronics in a monolayer of tungsten diselenide. Nature 557, 76 (2018). https://doi.org/10.1038/s41586-018-0013-6
  8. [8]Á. Jiménez-Galán, R. E. F. Silva, O. Smirnova, and M. Ivanov, Lightwave control of topological properties in 2D materials for sub-cycle and non-resonant valley manipulation. Nat. Photon. 14, 728 (2020). https://doi.org/10.1038/s41566-020-00717-3
  9. [9]I. Tyulnev, Á. Jiménez-Galán, J. Poborska, L. Vamos, P. St. J. Russell, F. Tani, O. Smirnova, M. Ivanov, R. E. F. Silva, and J. Biegert, Valleytronics in bulk MoS2 with a topologic optical field. Nature 628, 746 (2024). https://doi.org/10.1038/s41586-024-07156-y
  10. [10]F. Gucci, E. B. Molinero, M. Russo, P. San-Jose, F. V. A. Camargo, M. Mauiri, M. Ivanov, Á. Jiménez-Galán, R. E. F. Silva, S. D. Conte, and G. Cerullo, Encoding and manipulating ultrafast coherent valleytronic information with lightwaves. Nat. Photon. 20, 266 (2026). https://doi.org/10.1038/s41566-025-01823-w
  11. [11]K. F. Mak, K. L. McGill, J. Park, and P. L. McEuen, The valley Hall effect in MoS2 transistors. Science 344, 1489 (2014). https://doi.org/10.1126/science.1250140
  12. [12]S. D. Conte, F. Bottegoni, E. A. A. Pogna, D. D. Fazio, S. Ambrogio, I. Bargigia, C. D'Andrea, A. Lombardo, M. Bruna, F. Ciccacci, A. C. Ferrari, G. Cerullo, and M. Finazzi, Ultrafast valley relaxation dynamics in monolayer MoS₂ probed by nonequilibrium optical techniques. Phys. Rev. B 92, 235425 (2015). https://doi.org/10.1103/PhysRevB.92.235425
  13. [13]K. Kaasbjerg, K. S. Thygesen, and K. W. Jacobsen, Phonon-limited mobility in n-type single-layer MoS₂ from first principles. Phys. Rev. B 85, 115317 (2012). https://doi.org/10.1103/PhysRevB.85.115317
  14. [14]C. Mai, A. Barrette, Y. Yu, Y. G. Semenov, K. W. Kim, L. Cao, and K. Gundogdu, Many-Body Effects in Valleytronics: Direct Measurement of Valley Lifetimes in Single-Layer MoS₂. Nano Lett. 14, 202 (2014). https://doi.org/10.1021/nl403742j
  15. [15]T. Yu and M. W. Wu, Valley depolarization due to intervalley and intravalley electron-hole exchange interactions in monolayer MoS₂. Phys. Rev. B 89, 205303 (2014). https://doi.org/10.1103/PhysRevB.89.205303
  16. [16]M. M. Glazov, T. Amand, X. Marie, D. Lagarde, L. Bouet, and B. Urbaszek, Exciton fine structure and spin decoherence in monolayers of transition metal dichalcogenides. Phys. Rev. B 89, 201302(R) (2014). https://doi.org/10.1103/PhysRevB.89.201302
  17. [17]N. Ubrig, S. Jo, M. Philippi, D. Costanzo, H. Berger, A. B. Kuzmenko, and A. F. Morpurgo, Microscopic Origin of the Valley Hall Effect in Transition Metal Dichalcogenides Revealed by Wavelength-Dependent Mapping. Nano Lett. 17, 5719 (2017). https://doi.org/10.1021/acs.nanolett.7b02666
  18. [18]C. R. Dean, A. F. Young, I. Meric, C. Lee, L. Wang, S. Sorgenfrei, K. Watanabe, T. Taniguchi, P. Kim, K. L. Shepard, and J. Hone, Boron nitride substrates for high-quality graphene electronics. Nat. Nanotechnol. 5, 722 (2010). https://doi.org/10.1038/nnano.2010.172
  19. [19]For a review, see e.g., T. Ando, A. B. Fowler, and F. Stern, Electronic properties of two-dimensional systems. Rev. Mod. Phys. 54, 437 (1982). https://doi.org/10.1103/RevModPhys.54.437
  20. [20]I. Balslev, Influence of Uniaxial Stress on the Indirect Absorption Edge in Silicon and Germanium. Phys. Rev. 143, 636 (1966). https://doi.org/10.1103/PhysRev.143.636
  21. [21]A. A. Kaplyanskii, N.S. Sokolov, B.V. Novikov, and S.V. Gastev, Selective optical valley pumping in silicon and germanium. Solid State Commun., 20, 27 (1976). https://doi.org/10.1016/0038-1098(76)91691-4
  22. [22]A. M. Shirai, Y. Murotani, T. Fujimoto, N. Kanda, J. Yoshinobu, and R. Matsunaga, Valley polarization dynamics of photoinjected carriers at the band edge in room-temperature silicon studied by terahertz polarimetry. Phys. Rev. B 111, L121201 (2025). https://doi.org/10.1103/PhysRevB.111.L121201
  23. [23]A. Gindl, M. Čmel, F. Trojánek, P. Malý, and M. Kozák, Ultrafast room-temperature valley manipulation in silicon and diamond. Nat. Phys. 21, 947 (2025). https://doi.org/10.1038/s41567-025-02862-4
  24. [24]K. Tanaka, H. Ohtake, and T. Suemoto, Determination of intervalley scattering time in germanium by subpicosecond time-resolved Raman spectroscopy. Phys. Rev. Lett. 71, 1935 (1993). https://doi.org/10.1103/PhysRevLett.71.1935
  25. [25]V. G. Tyuterev and S. V. Obukhov, Ab initio calculation of electron-phonon scattering time in germanium. Phys. Rev. B 84, 035201 (2011). http://doi.org/10.1103/PhysRevB.84.035201
  26. [26]S.-H. Huang, Q.-Q. Zheng, W.-M. Xie, J.-Y. Lin, W. Huang, C. Li, and D.-F. Qi, Enhanced indirect-to-direct inter-valley scattering in germanium under tensile strain for improving the population of electrons in direct valley. J. Phys.: Condens. Matter 30, 465701 (2018). https://doi.org/10.1088/1361-648X/aae50e
  27. [27]G. Weinreich, T. M. Sanders, and H. G. White, Acoustolectric Effect in n-Type Germanium. Phys. Rev. 114, 33 (1959). https://doi.org/10.1103/PhysRev.114.33
  28. [28]W. C. Dash and R. Newman, Intrinsic Optical Absorption in Single-Crystal Germanium and Silicon at 77°K and 300°K. Phys. Rev. 99, 1151 (1955). https://doi.org/10.1103/PhysRev.99.1151
  29. [29]See Supplemental Material at [link] for our experimental setup and analysis, which includes Refs. [30–34].
  30. [34]C. Jacoboni and L. Reggiani, The Monte Carlo method for the solution of charge transport in semiconductors with applications to covalent materials. Rev. Mod. Phys. 55, 645 (1983). https://doi.org/10.1103/RevModPhys.55.645
  31. [35]T. Pfeifer, W. Kütt, H. Kurz, and R. Scholz, Generation and detection of coherent optical phonons in germanium. Phys. Rev. Lett. 69, 3248 (1992). https://doi.org/10.1103/PhysRevLett.69.3248
  32. [36]R. Scholz, T. Pfeifer, and H. Kurz, Density-matrix theory of coherent phonon oscillations in germanium. Phys. Rev. B 47, 16229 (1993). https://doi.org/10.1103/PhysRevB.47.16229
  33. [37]Y. Murotani, N. Kanda, T. Fujimoto, T. Matsuda, M. Goyal, J. Yoshinobu, Y. Kobayashi, T. Oka, S. Stemmer, and R. Matsunaga, Disentangling the Competing Mechanisms of Light-Induced Anomalous Hall Conductivity in Three-Dimensional Dirac Semimetal. Phys. Rev. Lett. 131, 096901 (2023). https://doi.org/10.1103/PhysRevLett.131.096901
  34. [38]T. Fujimoto, Y. Murotani, T. Tamaya, T. Kurihara, N. Kanda, C. Kim, J. Yoshinobu, H. Akiyama, T. Kato, and R. Matsunaga, Light-induced inverse spin Hall effect and field-induced circular photogalvanic effect in GaAs revealed by two-dimensional terahertz Fourier analysis. Phys. Rev. B 111, L201201 (2025). https://doi.org/10.1103/PhysRevB.111.L201201
  35. [39]A. Uddin, H. Nakata, and E. Otsuka, Off-equilibrium population of holes in the stress-split valence bands in photoexcited silicon and germanium. Phys. Rev. B 41, 3752 (1990). https://doi.org/10.1103/PhysRevB.41.3752
  36. [40]For a review, see e.g., J. Sinova, S. O. Valenzuela, J. Wunderlich, C. H. Back, and T. Jungwirth, Spin Hall effects. Rev. Mod. Phys. 87, 1213 (2015). https://doi.org/10.1103/RevModPhys.87.1213
  37. [41]K. Ando and E. Saitoh, Observation of the inverse spin Hall effect in silicon. Nat. Commun. 3, 629 (2012). https://doi.org/10.1038/ncomms1640
  38. [42]J.-C. Rojas-Sánchez, M. Cubukcu, A. Jain, C. Vergnaud, C. Portemont, C. Ducruet, A. Barski, A. Marty, L. Vila, J.-P. Attané, E. Augendre, G. Desfonds, S. Gambarelli, H. Jaffrès, J.-M. George, and M. Jamet, Spin pumping and inverse spin Hall effect in germanium. Phys. Rev. B 88, 064403 (2013). https://doi.org/10.1103/PhysRevB.88.064403
  39. [43]C. D. Stanciu, A. Tsukamoto, A. V. Kimel, F. Hansteen, A. Kirilyuk, A. Itoh, and Th. Rasing, Subpicosecond Magnetization Reversal across Ferrimagnetic Compensation Points. Phys. Rev. Lett. 99, 217204 (2007). https://doi.org/10.1103/PhysRevLett.99.217204
  40. [44]I. Radu, K. Vahaplar, C. Stamm, T. Kachel, N. Pontius, H. A. Dürr, T. A. Ostler, J. Barker, R. F. L. Evans, R. W. Chantrell, A. Tsukamoto, A. Itoh, A. Kirilyuk, Th. Rasing, and A. V. Kimel, Transient ferromagnetic-like state mediating ultrafast reversal of antiferromagnetically coupled spins. Nature 472, 205 (2011). https://doi.org/10.1038/nature09901
  41. [45]S. Schlauderer, C. Lange, S. Baierl, T. Ebnet, C. P. Schmid, D. C. Valovcin, A. K. Zvezdin, A. V. Kimel, R. V. Mikhaylovskiy, and R. Huber, Temporal and spectral fingerprints of ultrafast all-coherent spin switching. Nature 569, 383 (2019). https://doi.org/10.1038/s41586-019-1174-7
  42. [46]Y. Yang, R. B. Wilson, J. Gorchon, C.-H. Lambert, S. Salahuddin, and J. Bokor, Ultrafast magnetization reversal by picosecond electrical pulses. Sci. Adv. 3, e1603117 (2017). https://doi.org/10.1126/sciadv.1603117
  43. [47]K. Jhuria, J. Hohlfeld, A. Pattabi, E. Martin, A. Y. A. Córdova, X. Shi, R. L. Conte, S. Petit-Watelot, J. C. Rojas-Sanchez, G. Malinowski, S. Mangin, A. Lemaitre, M. Hehn, J. Bokor, R. B. Wilson, and J. Gorchon, Spin–orbit torque switching of a ferromagnet with picosecond electrical pulses. Nat. Electron. 3, 680 (2020). https://doi.org/10.1038/s41928-020-00488-3
  44. [48]D. Polley, A. Pattabi, A. Rastogi, K. Jhuria, E. Diaz, H. Singh, A. Lemaitre, M. Hehn, J. Gorchon, and J. Bokor, Picosecond spin-orbit torque–induced coherent magnetization switching in a ferromagnet. Sci. Adv. 9, eadh5562 (2023). https://doi.org/10.1126/sciadv.adh5562
  45. [49]K. Ogawa, H. Tsai, N. Yoshikawa, T. Matsuo, Y. Tsushima, M. Asakura, H. Peng, T. Matsuda, T. Higo, S. Nakatsuji, and R. Shimano, Ultrafast time-resolved observation of non-thermal current-induced switching in an antiferromagnetic Weyl semimetal. Nat. Mater. 25, 434 (2026). https://doi.org/10.1038/s41563-025-02402-8
  46. [50]H. Tsai, T. Matsuda, K. Kondou, K. Shimizu, T. Nomoto, T. Higo, T. Matsuo, Y. Tsushima, M. Asakura, H. Peng, D. N.-Hamane, S. Yamada, R. Tang, T. Iizuka, S. Miwa, R. Arita, M. Takenaka, and S. Nakatsuji, Picosecond ultralow-power switching device based on an antiferromagnet. Science 392, 761 (2026). https://doi.org/10.1126/science.adt3136

Paper details

vlnc:W_xgdth3ny
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