Abstract
In the field of highorder harmonic generation from solids, the electron motion typically exceeds the edge of the first Brillouin zone. In conventional nonlinear optics, on the other hand, the excursion of band electrons is negligible. Here, we investigate the transition from conventional nonlinear optics to the regime where the crystal electrons begin to explore the first Brillouin zone. It is found that the nonlinear optical response changes abruptly already before intraband currents due to ionization become dominant. This is observed by an interference structure in the thirdorder harmonic generation of fewcycle pulses in a noncollinear geometry. Although approaching Keldysh parameter γ = 1, this is not a strongfield effect in the original sense, because the iterative series still converges and reproduces the interference structure. The change of the nonlinear interband response is attributed to Bloch motion of the reversible (or transient or virtual) population, similar to the Bloch motion of the irreversible (or real) population which affects the intraband currents that have been observed in highorder harmonic generation.
Introduction
A crystal electron accelerated by an electric field in an electronic band (a Bloch electron) shows a motion pattern that is distinctively different from a free electron. The motion of a Bloch electron is described in kspace by the acceleration theorem ∂_{t}k = −E^{1} (atomic units are used). Already Felix Bloch recognized that the electron motion under the influence of a constant electric field would be oscillatory instead of unidirectional because the group velocity \({v}_{n}^{k}={\partial }_{k}{\omega }_{n}^{k}\) of an electron wave packet in band n with energy ω_{n} flips the sign after crossing the Brillouin zone edges. However, these Bloch oscillations, one of the most intriguing and counterintuitive corollaries of electronic bands, are difficult to observe because electron scattering prevents extended motion in kspace for static fields below the breakdown threshold.
One option to realize Bloch oscillations is to increase the lattice constant a so that the zone edge at k = π/a is closer. This was achieved by semiconductor superlattices and allowed the first observations of Bloch oscillations^{2,3,4}. Another option is to increase the electric field sufficiently so that the zone edge is reached ultrafast before scattering destroys the wave packet. This condition can be met for intense laser pulses when the electric field is so strong that electrons cross the zone edge within one optical cycle. This was considered from the beginning as a possible mechanism of highorder harmonic generation (HHG) from transparent crystals^{5}. For most conditions, the contribution of Bloch oscillations to HHG was reported to be weaker than other mechanisms^{6}, such as recollision^{7}, multiband coupling^{8}, and motion in bands with higher spatial frequencies^{9}. For terahertz fields it was found using a numerical switchoff analysis that Bloch oscillations contribute substantially to HHG^{10}. Very recently, it has been reported that HHG produced by twocolor fields exhibit phase variations that can be associated with reaching the zone edges^{11}. On the other hand, the decoherence dynamics of strongfield processes in solids are disputed, with many recent calculations assuming the ultrafast loss of interband coherence with fewfemtosecond dephasing times^{7,12,13,14,15}. If the underlying reason for this ultrafast coherence loss is rooted in scattering, it is questionable how laserdriven Bloch oscillations can arise.
For loworder harmonics, it is commonly accepted that ionization gains importance as the damage threshold is approached^{16}. Recently, the influence of stepwise ionization on intraband currents has been discussed^{17,18,19}. When the electrons begin to cover a significant range of the first Brillouin zone, the contribution of Bloch electrons to loworder harmonics is expected to change, as visualized in Fig. 1. An electron promoted to the conduction band (n = 2) generates the current \(J(t)={v}_{2}^{k(t)}\). With a rather shortrange trajectory at low intensities, the current contains mainly fundamental frequencies. With increasing intensity, the electron motion becomes anharmonic and contains increasingly more third harmonic frequencies.
Here, this simple picture is extended to the motion of interband coherences, which give rise to interband polarization. It is found that the nonlinear optical response from interband coherences (which are responsible for the reversible or transient or virtual population) changes abruptly as the crystal electrons explore the first Brillouin zone. This is observed by an interference structure in the thirdorder harmonic generation (THG). For short laser pulses, this happens already at intensities where the contribution of intraband currents due to ionization is not yet dominant. The mechanism is different from the influence of Bloch electron motion which has been observed before in HHG^{10,11} because that is related to intraband currents, which are seeded by the irreversible (or real) population. Furthermore, the observed effect is in the realm of perturbative optics, because the iterative series^{20} converges and reproduces the interference structure. The regime of intensities might be called an almoststrongfield regime because the intensities are smaller than in strongfield laser physics, yet the nonlinear response differs substantially from conventional nonlinear optics. Conventional nonlinear optics is understood here to indicate that the excursion of band electrons is negligible. Strongfield laser physics, on the other hand, is understood here to indicate that the response cannot be treated perturbatively, which implies a Keldysh parameter γ < 1. The almoststrongfield regime, where the electron trajectories cover a significant range of the first Brillouin zone but the contribution of intraband currents is still negligible, is commonly reached in lenses and windows of highpower optical instruments, in contrast to the strongfield regime, where optical elements deteriorate quickly.
Results
Experiment
To uncover the expected interference, an intensity scan is required that extends from the regime of conventional nonlinear optics into the regime of Bloch electron motion. This introduces significant complications for the experiment, as the yield of harmonic light would cover a very wide range. To overcome this problem, noncollinear spectroscopy is used here, sketched in Fig. 2. Two visible–infrared (Vis–IR) pulses A and B are focused into 100 μmthick crystals with polarization perpendicular to the plane of incidence. A and B are overlapped spatially and temporally with a precision of ±10 μm and ±1 fs, forming a laserinduced grating^{21,22}. Deep ultraviolet (DUV) light is produced by THG in the crystals. The DUV light that is emitted collinearly to A, which is kept at a constant intensity I_{A} = 7 TW/cm^{2}, is detected with a spectrometer. The intensity of B is varied in the range I_{B} = [0, 8] TW/cm^{2}, which varies the peak intensity \({I}_{{{{{{{{\rm{peak}}}}}}}}}={\left(\sqrt{{I}_{{{{{{{{\rm{A}}}}}}}}}}+\sqrt{{I}_{{{{{{{{\rm{B}}}}}}}}}}\right)}^{2}\) in the grating. This facilitates intensity scans over a high dynamic range with little variation in spectrometer count rates.
Data evaluation
The fundamental Vis–IR pulses are strongly modified by nonlinear pulse propagation, which obscures the observation of THG mechanisms. Fortunately, THG from the beginning of the crystals strongly contributes to the fringed DUV spectrum (Fig. 2b). As previous studies^{23,24} revealed, the spectral fringes are produced by two DUV pulses that are well separated in time after the crystal. The leading pulse (labeled U) is in time with the Vis–IR pulse, whereas the trailing pulse (labeled V) propagates at the DUV group velocity^{23,24,25}. V is beneficial for the interpretation of the data because it is generated within the first few micrometers and maintains its spectrum in the subsequent linear propagation. The contribution of U complicates the interpretation of the data because it is generated after nonlinear pulse propagation modified the fundamental pulses. Due to the scattering of the Vis–IR light in the DUV spectrometer, the spectra contain also a significant background in addition to the fringed spectra \(I\left(\omega \right)={\leftU\left(\omega \right)+V\left(\omega \right)\right}^{2}\). To remove the background and to enhance the sensitivity to V, the raw spectra are inverse Fourier transformed, the side peak (alternating component) is cut out and shifted to zero, and thereafter Fourier transformed. This yields \({I}_{{{{{{{{\rm{r}}}}}}}}}\left(\omega \right)=\left{U}^{*}(\omega )V(\omega ){{{{{{{{\rm{e}}}}}}}}}^{i\omega {t}_{{{{{{{{\rm{e}}}}}}}}}}\right\), where t_{e} represents the shift to zero which corresponds to the delay between U and V after the medium. The parameters used are 87 fs for SiO_{2}, 111 fs for Al_{2}O_{3}, 190 fs for MgO.
The intensity scan reveals an interference structure in the region I_{peak} = [10, 20] TW/cm^{2} for SiO_{2} (see Fig. 2c). As this is the regime where the crystal electrons explore the first Brillouin zone (Fig. 1), this is the first indication for THG from Bloch electrons that competes with conventional THG. Corresponding experiments in Al_{2}O_{3} and MgO yield similar interferences (Fig. 3), but fine details indicate that the band structure has an influence.
Discussion
To get further insights, numerical calculations are performed using semiconductor Bloch equations (SBEs)^{1}. With restriction to the spatial dimension of the electric field vector and omitting the Coulomb interaction, the SBEs read^{20,26}
The diagonal elements \({\rho }_{nn}^{k}\) of the density matrix are the populations of the electronic bands, the offdiagonal elements \({\rho }_{nm}^{k}(n\, \ne \,m)\) are the coherences between the states with transition energies \({\omega }_{nm}^{k}={\omega }_{m}^{k}{\omega }_{n}^{k}\). The electric field E induces dynamics by coupling the electronic bands via the dipole matrix elements \({d}_{nm}^{k}\) and by moving the electrons and holes within the bands, which is realized by the coordinate transform k → k + A, where A is the vector potential defined by E = −∂_{t}A. The relaxation terms \({\left({\partial }_{t}{\rho }_{nm}^{k}\right)}_{{{{{{{{\rm{relax}}}}}}}}}\) are implemented as phenomenological damping terms (see the “Methods” section). The polarization P and the current J are calculated by
where δk is the spacing in the kgrid.
In most recent studies, the band structures are calculated by ab initio methods like density functional theory (DFT). Here, a different approach is taken. For a quantitative comparison of optical fields originating from macroscopic pulse propagation, it is essential that both the linear response (including the group velocity dispersion) and the nonlinear response (including the optical Kerr effect (OKE)) match the experiment. The linear response of DFT is known to deviate because of missing background contributions^{27}; if the OKE is correctly reproduced is typically not tested. Here, numerical refractive index data is used to incorporate linear polarization in the pulse propagation. The nonlinear polarization P^{(NL)} is used as a source term in the pulse propagation, which is calculated from Eq. (2) in Fourier space by
where χ^{(1)} is the linear susceptibility of Eq. (1) (see the “Methods” section). The dipole matrix elements of three bands are then adjusted to match the experimental data of the OKE. This supports pulse propagation using the unidirectional pulse propagation equation (UPPE) at feasible computation times, yet captures the interband and intraband dynamics consistently^{20}.
The macroscopic UPPE calculations (see the “Methods” section), using the experimentally determined pulse shapes, confirm the assumption of two separate DUV pulses (see Fig. 4). The calculated spectra are processed like the experimental spectra. The Fourierfiltered spectrum I_{r}(ω), although somewhat masked by propagation effects, still resembles the THG at the beginning of the crystal (Fig. 4d). The reason is that V originates within the first few micrometers in the crystal and maintains its spectrum in the subsequent linear propagation. The intensity scan exhibits an interference structure that is similar to the experimental data. In order to investigate the influence of the band shape, calculations are performed where the band shape contains higher frequencies (Supplementary Information Fig. 1). As expected, the band shape influences the interference structure, which might be exploited to extract information about band shapes from the data. However, the calculations limited to three bands are unlikely to reproduce the interference structure in fine detail.
The interference structure also addresses a debated inconsistency in the field of HHG from solids. Numerical calculations by several groups agree that noisy spectra of HHG are predicted, in contrast to the clean harmonics measured experimentally. The most prominent solution to this discrepancy is to assume ultrafast coherence loss realized by dephasing times below 10 fs, which helps the calculations produce clean harmonics^{7,12,13,14,15}. The interference structure vanishes for such short dephasing times (Supplementary Information Fig. 2), which does not support the assumption of dephasing times below 10 fs.
To clarify whether the interference structure is caused by a change in the mechanism of THG rather than propagation effects, the nonlinear response generated by 8fs Gaussian pulses is investigated (see Fig. 5). P^{(NL)} exhibits an interference structure in the range [8, 20] TW/cm^{2}, similar to the UPPE calculations and the experiment. The shape is influenced by the band shape (see Supplementary Information Figs. 3 and 4), but the general appearance is universal. Also, J shows an interference structure, but at lower intensities [2, 8] TW/cm^{2}. This is an indication that the interference structure observed experimentally is not due to J. This is affirmed by running the UPPE calculation with J = 0, which yields an indistinguishable result from the complete calculation displayed in Fig. 4c. This seems to contradict recent wavemixing experiments at similar intensities^{17,18}, but a crucial point may be that the SBE calculations reproduce the reversible population of the conduction band^{20,28,29} rather than assuming a stepwise ionization.
The Keldysh parameter \(\gamma=\frac{{\omega }_{0}\sqrt{{\omega }_{12}}}{\leftE\right}\), where ω_{0} is the optical frequency, is commonly used to distinguish multiphoton (γ > 1) and strongfield (γ < 1) interactions^{30}. While the former can be treated by the power series expansion of perturbative nonlinear optics, this series diverges for the latter. The transition region has attracted much attention for gases, which is sometimes referred to as the regime of nonadiabatic tunneling^{31} but has not yet received much attention for solids. The intensity range of the interference structure (γ = 1.5 at 8 TW/cm^{2} and γ = 1 at 18 TW/cm^{2}) is below the regime of strongfield laser physics in the original sense. Strongfield laser physics in the original sense is understood here to mean that the power series expansion does not converge. To test the convergence, the SBEs are solved iteratively (see the “Methods” section and ref. 20). It has been shown before that the convergence criterion of the iteration is fulfilled for γ > 1^{20}. The pseudocolor spectra produced with 100 iterations are distinguishable from those of the timedomain integration displayed in see Fig. 5a. Only the line plot of Fig. 5d reveals deviations starting at 18 TW/cm^{2} where γ = 1.
To finally reveal the mechanism that causes the modification of the nonlinear response in the almoststrongfield regime, a simplified calculation is performed neglecting the motion of the Bloch electrons. This is achieved by omitting the coordinate transform k → k + A in Eq. (1). With the Bloch electron motion turned off, the interference structure in P^{(NL)} disappears. For lower intensities, displayed in the inset of Fig. 5d, the calculations with and without Bloch electron motion agree perfectly. Thus, Bloch electron motion can be neglected at moderate intensities. Moreover, the instantaneous response P^{(NL)} = χ^{(3)}E^{3}, which is a common simplification for the OKE in transparent solids, is a very good approximation for these intensities. In the almoststrongfield regime, the THG intensity deviates from the instantaneous response model for both calculations with and without Bloch electron motion but only the full calculation generates the interference structures. At 5 TW/cm^{2}, where J exhibits interference, the electrons transverse up to 45% of the first Brillouin zone. However, this is not observed in the experiment because the influence of J is still negligible at these low intensities. At 14 TW/cm^{2}, where P^{(NL)} exhibits interference, the electrons transverse up to 75% of the first Brillouin zone. The origin of the nonlinear polarization in kspace is traced by omitting the ksummation in Eq. (2) resulting P^{(NL)}(k). In the almoststrongfield regime, the origin is shifted through the entire Brillouin zone (Fig. 5d). This underpins the interpretation of Bloch motion that affects the interband polarization. At low intensities, only the local band curvature is decisive which is highest at k = 0. When the electrons start to explore the Brillouin zone, the band curvature throughout the trajectory must be considered both for the interband and for the intraband contribution of the nonlinear response.
The spectrograms (Fig. 6 and Supplementary Information Fig. 5) show that while THG is temporally delocalized at low intensity, THG and also higher frequency generation are localized within the optical cycle at higher intensities. Some features are reminiscent of the threestep model for HHG^{32}. In particular, there are branches with positive chirp, as typically associated with short electron trajectories, followed by branches with negative chirp, as typically associated with long trajectories. However, the threestep model would predict only photons with energies of band transitions^{7,8}, which are limited to [10.3, 13.3] eV for the band structure used here. Furthermore, the highest photon energies are found at the peaks of the generating field, but the threestep model predicts them near the zero crossings. It is remarkable that the instantaneous response model fits very well for low intensity but at higher intensity both the P^{(NL)} and J exhibit dents at the field crests. These dents are clearly visible by comparing with the simplified calculation that neglects the motion of the Bloch electrons (Supplementary Information Fig. 6). These dents are reminiscent of the current generated by a single Bloch electron in Fig. 1. This strengthens the interpretation that Bloch electron motion can be regarded as a mechanism of harmonic generation not only for real electrons, which are the origin of J and which was considered from the beginning as a possible mechanism for HHG but also for virtual electrons (coherences), which are the origin of P^{(NL)}. In contrast to the original strongfield regime (γ < 1), where optical components are very easily damaged, the almoststrongfield regime is often reached in highpower lasers and other optical instruments. The results of this work, especially numerical pulse propagation at feasible computation times, will be useful for the design of such instruments.
Methods
Calculations based on SBEs
The consistent treatment of the OKE requires at least three bands^{20}. Two valence bands (bands 1 and 3) and one conduction band (band 2) are considered here. In a realistic band structure, valence bands have typically a transition energy on the order of 1 eV, but resonance effects with the Vis–IR pulse do not prevail because many valence bands exist. To avoid resonance effects for only two valence bands, \({\omega }_{3}^{k}={\omega }_{1}^{k}\) is assumed. However, only band 1 is coupled to the conduction band. Only the conduction band energy is considered to be kdependent with a tightbinding band shape:
with bandwidth b_{1} = 3 eV and lattice constant a = 0.49 nm. The bandgap is set to \({\omega }_{12}^{k=0}={\omega }_{12}=10.3\) eV,
The dipole matrix elements are matched to the OKE at low intensities as described in ref. 20. The valence band transitions are implemented with \({d}_{13}^{k}={d}_{31}^{k}={d}_{13}=16\). The valence to conduction band transitions \({d}_{12}^{k}={d}_{21}^{k}\) are implemented as^{1}
with \({d}_{12}^{k=0}=0.02\). All other dipole matrix elements are set to zero.
Relaxation is implemented as phenomenological damping terms. For the diagonal elements, the lifetime T_{1} and the collision time T_{c} are considered.
The lifetime in conduction bands of dielectrics usually exceeds 100 fs, justifying the assumption T_{1} = ∞. The terms proportional to 1/T_{c} cause a decay of the currents, while the total band population is preserved. The damping of the currents cannot be neglected, because Drude collision times are in the fewfemtosecond range. This is in accordance with the qualitative picture of excited electrons that first undergo rapid momentum relaxation and thereafter energy relaxation on a longer timescale^{33}. For the coherences,
where T_{2} is the interband dephasing time. Here it is assumed that T_{2} is identical for all coherences and independent of k. The relation between the dephasing time T_{2} and the Drude collision time T_{c} is not known. Here, T_{2} = 2T_{c} is assumed, following the phenomenological picture that if scattering occurs to an electron at position k, its interband and intraband coherences are likewise destroyed.
The numerical calculations are performed on a kgrid with 27 points. The timedomain integration is performed using the 4thorder RungeKutta (RK4) method. For the calculations with pulse propagation (Fig. 4 and Supplementary Information Figs. 1 and 2), a tgrid with 30,001 points in the interval [−250, 250] fs is used. For the calculations without pulse propagation (all other figures), a tgrid with 70,001 points in the interval [−500, 500] fs is used.
Iteration of SBEs
For dielectrics, the population transfer into the conduction band is only a small fraction of the valence band population when irreversible material changes are avoided. This justifies the assumptions \({\rho }_{11}^{k}{\rho }_{22}^{k}=1\); \({\rho }_{11}^{k}{\rho }_{33}^{k}=0\); \({\rho }_{33}^{k}{\rho }_{22}^{k}=1\), which effectively decouple the diagonal and offdiagonal elements of the SBEs. With this approximation, the nondiagonal elements of Eq. (1) can be transformed to
with
Here, the timedependent transition energy \({\omega }_{nm}^{k+A}\) is separated into a static part \({\omega }_{nm}={\omega }_{nm}^{k}\) and the dynamic part \({\widetilde{\omega }}_{nm}^{k+A}={\omega }_{nm}^{k+A}{\omega }_{nm}\). The elements D_{nm} are the corrections due to the nonlinearity. A recursive method is used for their calculation: In the nth step of iteration, the elements ρ_{nm} in Eq. (9) are calculated using the elements D_{nm} of the (n−1)th step. Each step of iteration requires (inverse) Fourier transforms and timedomain multiplications. D_{nm} = 0 is assumed in step 0.
The iteration is equivalent to a power series expansion. Accordingly, there is an upper limit for the electric field above which the iteration does not converge. In the limit of a monochromatic field with frequency ω_{0}, the convergence criterion is given by γ > 1^{30}.
The linear susceptibility follows from Eq. (9) by setting D_{12} = 0:
Pulse propagation
Macroscopic pulse propagation is calculated using the UPPE
where the hat symbol indicates the Fourier transform in the dimensions of time and transverse space. In addition to the propagation direction z, one transverse dimension (the xdimension) is included to account for the noncollinear geometry with \(K=\sqrt{{n}_{{{{{{{{\rm{R}}}}}}}}}^{2}\frac{{\omega }^{2}}{{c}^{2}}{k}_{x}^{2}}\). Numerical tables are used for the refractive index n_{R}, c is the speed of light and u is the group velocity of the Vis–IR pulse. The electric field is treated as a scalar field because all pulses are polarized perpendicular to the plane of incidence.
The UPPE is integrated numerically using the splitstep method with an xgrid with 81 points in the interval [−260, 260] μm and a zgrid with 401 points in the interval [0, 100] μm.
Subsequent to the propagation inside the crystal, the light propagating collinearly to A with an emission angle \(\frac{\alpha }{2}\) is calculated by
with \({\rm {tan}}(\frac{\alpha }{2})=\frac{{k}_{x}}{\sqrt{{(\omega /c)}^{2}+{({k}_{x})}^{2}}}\).
Data availability
The data generated in this study are provided in the Supplementary Information. Source data are provided with this paper.
References
Haug, H. & Koch, S.W. Quantum Theory of the Optical and Electronic Properties of Semiconductors, 5th edn (World Scientific, Singapore, 2009)
Feldmann, J. et al. Optical investigation of bloch oscillations in a semiconductor superlattice. Phys. Rev. B 46, 7252–7255 (1992).
Leo, K., Bolivar, P. H., Bruggemann, F., Schwedler, R. & Kohler, K. Observation of bloch oscillations in a semiconductor superlattice. Solid State Commun. 84, 943–946 (1992).
Waschke, C. et al. Coherent submillimeterwave emission from bloch oscillations in a semiconductor superlattice. Phys. Rev. Lett. 70, 3319–3322 (1993).
Faisal, F. H. M. & Kaminski, J. Z. Floquetbloch theory of highharmonic generation in periodic structures. Phys. Rev. A 56, 748–762 (1997).
Ghimire, S. et al. Observation of highorder harmonic generation in a bulk crystal. Nat. Phys. 7, 138–141 (2011).
Vampa, G. et al. Theoretical analysis of highharmonic generation in solids. Phys. Rev. Lett. 113, 073901 (2014).
Ndabashimiye, G. et al. Solidstate harmonics beyond the atomic limit. Nature 534, 520 (2016).
Luu, T. T. et al. Extreme ultraviolet highharmonic spectroscopy of solids. Nature 521, 498–502 (2015).
Schubert, O. et al. Subcycle control of terahertz highharmonic generation by dynamical bloch oscillations. Nat. Photonics 8, 119–123 (2014).
UzanNarovlansky, A. J. et al. Observation of lightdriven band structure via multiband highharmonic spectroscopy. Nat. Photonics 16, 428 (2022).
Hohenleutner, M. et al. Realtime observation of interfering crystal electrons in highharmonic generation. Nature 523, 572–575 (2015).
Kruchinin, S. Y., Krausz, F. & Yakovlev, V. S. Colloquium: Strongfield phenomena in periodic systems. Rev. Mod. Phys. 90, 021002 (2018).
Floss, I. et al. Ab initio multiscale simulation of highorder harmonic generation in solids. Phys. Rev. A 97, 011401 (2018).
McDonald, C. R., Ben Taher, A. & Brabec, T. Strong optical field ionisation of solids. J. Opt. 19, 114005 (2017).
Couairon, A., Sudrie, L., Franco, M., Prade, B. & Mysyrowicz, A. Filamentation and damage in fused silica induced by tightly focused femtosecond laser pulses. Phys. Rev. B 71, 125435 (2005).
Mitrofanov, A. V. et al. Optical detection of attosecond ionization induced by a fewcycle laser field in a transparent dielectric material. Phys. Rev. Lett. 106, 147401 (2011).
Jurgens, P. et al. Origin of strongfieldinduced loworder harmonic generation in amorphous quartz. Nat. Phys. 16, 1035 (2020).
Jurgens, P. et al. Characterization of laserinduced ionization dynamics in solid dielectrics. ACS Photonics 9, 233–240 (2022).
Pfeiffer, A. N. Iteration of semiconductor bloch equations for ultrashort laser pulse propagation. J. Phys. BAt. Mol. Opt. Phys. 53, 164002 https://iopscience.iop.org/article/10.1088/13616455/ab94cb (2020).
Pati, A. P., Wahyutama, I. S. & Pfeiffer, A. N. Subcycleresolved probe retardation in strongfield pumped dielectrics. Nat. Commun. 6, 7746 (2015).
Pati, A. P., Reislohner, J., Leithold, C. G. & Pfeiffer, A. N. Effects of the grooveenvelope phase in selfdiffraction. J. Mod. Opt. 64, 1112–1118 (2017).
Reislöhner, J., Leithold, C. & Pfeiffer, A. N. Characterization of weak deep UV pulses using crossphase modulation scans. Opt. Lett. 44, 1809–1812 (2019).
Reislöhner, J., Leithold, C. & Pfeiffer, A. N. Harmonic concatenation of 1.5 fs pulses in the deep ultraviolet. ACS Photonics 6, 1351–1355 (2019).
Babushkin, I. V., Noack, F. & Herrmann, J. Generation of sub5 fs pulses in vacuum ultraviolet using fourwave frequency mixing in hollow waveguides. Opt. Lett. 33, 938–940 (2008).
Li, J. B. et al. Phase invariance of the semiconductor bloch equations. Phys. Rev. A 100, 043404 (2019).
Kilen, I. et al. Propagation induced dephasing in semiconductor highharmonic generation. Phys. Rev. Lett. 125, 083901 (2020).
Schultze, M. et al. Controlling dielectrics with the electric field of light. Nature 493, 75–8 (2013).
Sommer, A. et al. Attosecond nonlinear polarization and lightmatter energy transfer in solids. Nature 534, 86–90 (2016).
Keldysh, L. V. Ionization in the field of a strong electromagnetic wave. JETP 20, 1307–1314 (1965).
Yudin, G. L. & Ivanov, M. Y. Nonadiabatic tunnel ionization: Looking inside a laser cycle. Phys. Rev. A 64, 013409 (2001).
Lewenstein, M., Balcou, P., Ivanov, M. Y., Lhuillier, A. & Corkum, P. B. Theory of highharmonic generation by lowfrequency laser fields. Phys. Rev. A 49, 2117–2132 (1994).
Sjakste, J., Tanimura, K., Barbarino, G., Perfetti, L. & Vast, N. Hot electron relaxation dynamics in semiconductors: assessing the strength of the electron–phonon coupling from the theoretical and experimental viewpoints. J. Phys. Condens. Matter 30, 353001 (2018).
Acknowledgements
This project was supported primarily by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) via Priority Program 1840 “Quantum Dynamics in Tailored Intense Fields (QUTIF)" (project ID 281272215) (J.R., D.K., A.N.P., I.B.) and via project B1 in the Collaborative Research Centre 1375 “Nonlinear optics down to atomic scales (NOA)" (project ID 398816777) (A.N.P.). I.B. thanks Cluster of Excellence PhoenixD (EXC 2122, project ID 390833453) for financial support.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
J.R. and A.N.P. constructed the experiment. J.R. conducted the experiment. A.N.P. developed the theory. J.R., D.K., I.B., and A.N.P. contributed to the discussion and preparation of the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Tzveta Apostolova and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Reislöhner, J., Kim, D., Babushkin, I. et al. Onset of Bloch oscillations in the almoststrongfield regime. Nat Commun 13, 7716 (2022). https://doi.org/10.1038/s41467022354013
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467022354013
This article is cited by

Threedimensional nonAbelian Bloch oscillations and higherorder topological states
Communications Physics (2023)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.