Non-collinear high-harmonic generation in solids.
The non-collinear HHG is a well-accepted scheme that gives additional controllability on wavefront22–24, 32, polarization19,33,34, and spectrum35 especially in gaseous atomic medium. The similar approach is the HHG using an annular driver to separate driver while keeping radial symmetry of the harmonic beams36–38. Meanwhile, the non-collinear scheme has not been introduced to this recent field of solid HHG, so in-depth understanding on its key advantages has not been addressed to date. In this study, we devised a common-optics setup mediating wavefront control of the EUV high-harmonics from a solid medium. Figure 1a illustrates the experimental setup configured in this investigation to generate EUV high harmonics from a thin wafer of magnesium oxide (MgO) crystal. Our non-collinear HHG is implemented by irradiating the crystal sample with a pair of beams symmetrically split from a single driving laser. With identical vertical polarization, the two beams undergo non-collinear propagation through a series of optical components, including a Wollaston prism, a Glan polarizer, a half-wave plate, a triangular optical wedge, and a collimating lens (see Methods for details). The driving laser delivers infrared (IR) short pulses via a spatial light modulator (SLM) that is inserted to induce a sequence of linearly varying phase slopes across the laser wavefront. The driving laser is accordingly steered up and down, by the phase modulation using the SLM, vertically along the x-axis with respect to the x-y-z coordinate system embedded in the experimental setup (Fig. 1a). The oblique beams finally converging to the crystal sample, with a crossing angle of \({2\theta }_{d}\)as projected on the y-z plane, are distinguished as k1 and k2 in terms of their wavevectors.
Note that our non-collinear illumination causes generated EUV harmonics to diverge on the y-z plane as shown in our experimental data (Fig. 1c) obtained from a 150-µm thick MgO (100) crystal. In fact, the divergence angle of the q-th harmonic, measured from the surface normal of the crystal sample, varies with the harmonic order q in compliance with the conservation of two physical quantities; parity and momentum. First, the rule of parity conservation dictates the equality of q = q1 + q2 with q1 and q2 being the individual number of IR photons contributed by k1 and k2, respectively32. This shared multi-photon excitation leads to a multiplicity of possible sets of (q1, q2) for a given harmonic order q, i.e. {(0, q), (1, q-1), (2, q-2), … (q-1,1), (q,0)}, as observed in our experimental data of the 7-th (H7) and 9-th (H9) harmonics (Fig. 1d). Second, for each set of (q1, q2), the rule of momentum conservation holds as kq = q1 k1 + q2k2 with kq being the wavevector of the q-th harmonic. Third, in solid-based HHG, generated EUV harmonics are subject to strong reabsorption during the travel within the specimen, thereby leaving no concern about phase matching. More specifically, for MgO crystal, the optical penetration depth is known to be as short as about one wavelength (Fig. 1a inset) for the whole EUV spectrum ranging from 20 to 120 nm39. This implies that the main portion of EUV harmonics escaping from the crystal originates from just a few hundreds of atomic layers on the rear-side surface. Fourth, the momentum conservation indicates that the smallest value of \({\theta }_{q}\) for the q-th harmonic, propagating closest to the z-axis, satisfies the relation of \(\text{t}\text{a}\text{n}{\theta }_{q}=\text{t}\text{a}\text{n}{\theta }_{d}/q\) as verified in our experimental data (Fig. 1b). Fifth, all the non-collinear divergence of generated EUV harmonics remain uninfluenced, being not much affected by driving laser intensities up to 23 TW/cm2 (Fig. 1e). Lastly, it is worthwhile to note that the divergent character of non-collinear harmonics generation can also be scrutinized by dealing with the wave interference pattern of the two driving beams as an active grating formed on the rear surface of the specimen, in compliance with the well-established far-field wave diffraction theory 40,41.
Figure 2 presents our test results showing the dependence of measured EUV harmonic spectra upon the crystal orientation as well as the driving laser intensity under non-collinear illumination. The measured spectra of H7 – H17 all exhibit four-fold symmetry when the crystal sample is rotated with respect to the polarization direction of the driving laser (Fig. 2a), which is attributable to the face-centred cubic crystal structure of the MgO sample. Meanwhile, the yield of generated EUV harmonics becomes higher when the driving laser polarization coincides with the Γ-X direction of the crystal (Fig. 2b), which is known as the case of offering strong dipole-allowed coupling between the conduction and valence bands42. In addition, with increasing the driving laser intensity I, the yield of generated EUV harmonics builds up in proportion to \({I}^{q}\) for intensities up to 15–20 TW/cm2 (Fig. 2c). This perturbative growth of EUV harmonics however begins to weaken as the crystal sample is subject to thermal damage with increasing the driving laser intensity, and eventually ceases due to the occurrence of ablation near 30 TW/cm2. This observation indicates that EUV harmonics emitted by non-collinear irradiation exhibit similar behaviours to the normal illumination of single-beam HHG43, except generated EUV harmonics isolate from the driving laser under the conservation rule of parity and momentum.
Facile wavefront control of EUV harmonic beams
Figure 3 illustrates how our scheme of non-collinear illumination separates the EUV harmonics spatially from the driving laser in the horizonal y-z plane on the exit side of the crystal sample (Fig. 3a). Note that phase-modulation of the driving laser wavefront using the SLM permits the EUV harmonics to be steered in the vertical x-z plane (Fig. 3b), in which θdiff denotes the tilt angle of the driving laser wavefront brought on the x-z plane by the SLM. The source driving laser has a beam diameter of 2.5 mm and the MgO crystal sample is placed at a defocused position inward by an offset of 7 mm from the focal point. The focusing lens has a 50 mm focal length. The beam profile of the 7-th harmonic (H7) is measured along the Rayleigh length (1.63 mm) of the driving laser beam by adopting the knife-edge method (see Methods for details). The measured beam profile yields an elliptical cross-section with a beam diameter of 12 µm and 18 µm along the x-axis and y-axis, respectively (Fig. 3c & 3d). When θdiff varies in the range of -4 to 4 mrad by activating the SLM (Fig. 3b), the H7 harmonic makes a linear-scanning motion on the focal plane along the x-direction following the multifold phase conversion condition of \({\varPhi }_{EUV}={q\varPhi }_{IR}\), leading to a significant lateral range of -2 to 2 mm (Fig. 3e). This EUV beam scanning appears to be highly repeatable in response to the θdiff input command, with no significant variation in the spot size of the EUV beam. Accordingly, without use of extra EUV-scanning optics, elaborate dynamic control can be achieved simply by modulating the θdiff input command (Fig. 3f).
EUV Bessel vortex generation
Figure 4 shows another experimental scheme implemented in this investigation to focus EUV harmonics in the form of interferometric Bessel-like beams. The SLM screen was separated into two sections in radial direction to afford two different helical wavefronts to the driver beams. Next, the driving laser beam is spatially masked to form double-annular beams carrying a different topological charge l, the number of \(2\pi\) phase-shift around the azimuthal direction. Then, around-z-axis-symmetric non-collinear irradiation is made on the MgO crystal through an axicon lens combined with an aspheric focusing lens (Fig. 4a). The resulting EUV harmonic beam profile is monitored by moving a microchannel plate (MCP) 2-D detector along the z-axis direction (Fig. 4b). For the 7th harmonic (H7), two parity sets (q1:q2 = 3:4 and 4:3) are seen most dominant with distinct converging angles (\({\theta }_{b}\)). Therefore, the propagation of two annular EUV beams is observed along the z-axis. Note that an annular aperture is inserted before the MCP to block the IR driving laser, while letting only the H7 EUV harmonic be received. Our scheme is almost free from EUV harmonic power loss from a spectral filter (such as an aluminium membrane) owing to the non-collinear focusing configuration. With the driving laser intensity being set at 27 TW/cm2, the temporal stability of our optical configuration (Fig. 4a) was verified during a relatively long observation period of 20 min with respect to the EUV beam profile as well as the photon flux (Fig. 4c & 4d). These stable test results confirm that our Bessel-beam conversion scheme based on solid-based HHG is robust to environmental noises and less sensitive to phase matching. As a result, a stable interferometric Bessel beam of the H7 harmonic was well grown along the optical axis over a few millimetres of focal depth.
The single pixel size of our MCP detector was as large as 80 µm, being not adequate for accurate determination of the focused Bessel beam diameter. Alternatively, a ptychographic technique44 was adopted to retrieve the intensity profile of the EUV Bessel beams (Fig. 4e, see Methods for details). Figure 4g shows the retrieved beam profile with the phase map applied to the SLM. The result with an uniform phase indicates that the central lobe is well concentrated on a spot size of 390 nm in FWHM, equivalent to 3.4 times the H7 wavelength of 115 nm (Fig. 4g). Note that side lobes of lower intensities are seen not purely concentric, as they are found disturbed by the aperture stop of annular shape installed to block the IR driving laser using four suspension-legs. When the SLM applied the counter rotating topological charges of l1 = 1and l2 = -1 for each annular beam, the topological charge of the q-th harmonic Bessel beam was determined by the conservation rules of parity and orbital angular momentum: lq = q1 l1 + q2l2, thus lq = ± 1. The interference of the two driving beams with the non-zero azimuthal phase leads to formation of a spiral phase, which is transferred to the high-harmonic beams. Again, the intensity and phase distributions were disturbed by the IR beam blocks. Nonetheless, we can infer that the interference of the two driving beams with the non-zero azimuthal phase leads to formation of a spiral phase, which is transferred to the high-harmonic beams. The singularity of the spiral phase at the centre of the beam leads to the zero intensity, thereby demonstrating the EUV Bessel vortex for the first time.