Slow light nanocoatings for ultrashort pulse compression

The study addresses the challenge of dispersion control for femtosecond laser pulses in the ultraviolet, visible, and near-infrared (<1.3 µm), where transparent materials are normally dispersive and elongate pulses. Existing solutions—angular-dispersive prism/grating pairs, reflective chirped mirrors, and photonic-crystal-fiber-based compressors—add complexity, path length, and beam deviations. The research question is whether an ultrathin transmissive nanocoating can provide anomalous group delay dispersion (GDD) upon transmission to directly compensate normal material dispersion and compress ultrashort pulses. The purpose is to develop a compact, broadband, high-transmission coating that can be applied to conventional optics to render them ultrashort-pulse compatible, thereby simplifying femtosecond systems. The importance lies in enabling anti-/non-dispersive optics and broadening applications of femtosecond pulses without complex optical arrangements.

Prior dispersion compensation relies on prism pairs and gratings (angular-dispersive systems), reflective dispersive mirrors (chirped mirrors), and photonic crystal fibers for anomalous dispersion, each with setup complexity and alignment demands. Recent dielectric metasurfaces have shown utility for spatial wavefront control and, in Fourier-transform setups, for temporal pulse control. Theoretically proposed ultrathin dispersive approaches include plasmonic metasurfaces and flat nanodisk Huygens metasurfaces, but they face trade-offs in transmission, dispersion sign/magnitude, higher-order dispersion, or bandwidth. Slow-light concepts in photonic crystals and Mie-resonant scatterers indicate potential for strong dispersion. The present work leverages high-aspect-ratio silicon nanopillar arrays to achieve high transmission, anomalous dispersion with low higher-order dispersion, and broadband operation around 800 nm.

Design principle: Uniform circular amorphous silicon nanopillars in a periodic square array on fused silica form a transmissive compressor. The working mechanism is described via a 2D photonic crystal cross-section and waveguide mode picture: incident light couples primarily to HE11-like and HE12-like modes. Near the HE12-like mode cutoff at kz = 0, reciprocity enforces vanishing group velocity (slow light), and the bending of its dispersion from the vacuum light line into this slow-light region induces anomalous GVD. Coupling transitions from HE12-like to HE11-like near cutoff, producing a broadband region with approximately constant anomalous GDD. Modeling and optimization: A large parameter space (nanopillar diameter, height, periodicity) was surveyed using rigorous coupled-wave analysis (S4). Finite-difference time-domain (FDTD, Lumerical) simulations then fine-tuned nanopillar radius, height, and unit cell to maximize negative GDD, minimize higher-order dispersion, and maintain high transmission. Modal dispersion, coupling fractions, and field profiles were analyzed; far-field response and angle sensitivity up to 5° for s-polarization were evaluated. Theoretical limit: A lower bound for constant anomalous GVD/GDD was derived by parameterizing the dispersion within the working range as parabolic and constraining maximum group velocity to c at the high-frequency edge. This yields GVDmin proportional to (Δk)^2 and connects GDDmin to coating thickness L and the effective refractive index n+ at the high-frequency edge. For an 80 nm bandwidth centered at 800 nm and n+ ≈ 1, a theoretical limit of GVDmin ≈ −264 fs^2 µm^-1 is predicted, with practical realizations achieving about half due to coupling and non-ideal parabolic dispersion. Fabrication: A 610 nm-thick amorphous silicon layer was deposited on a 500 µm fused silica substrate via PECVD. Negative e-beam resist (ma-N 2403) and conductive polymer (ESPACER 300) were applied for EBL patterning (Elionix ELS-F125), followed by development (MIF 726). Nanopillars were etched using ICP-RIE (SF6/C4F8 chemistry). The resist mask was removed in piranha solution. Devices covered ~3.1 mm^2 and are compatible with DUV lithography for scalable manufacturing. Parameter tuning and resonances: The nanopillar diameter tunes the broadband TM dipole (Mie-type) resonance to set the central wavelength. Array periodicity (475 nm) was chosen subwavelength in air and substrate to suppress higher diffraction orders and preserve beam profile. Nanopillar height controls dispersion magnitude but certain heights can induce reflectivity; a narrowband Fabry–Perot resonance present in the working range can be suppressed by tuning periodicity. The interplay between Mie and Fabry–Perot resonances was analyzed; residual narrow signatures are negligible for ultrashort pulse envelopes due to their narrow spectral width. Characterization: Spectral-domain group delay was measured by white-light spectral interferometry using a tungsten lamp source in a Michelson interferometer; spectral interference between sample and reference arms yields the sample’s group delay profile. Time-domain pulse measurements employed noncollinear SH-FROG with a 100 µm BBO crystal; iterative ptychographic retrieval with measured fundamental spectra as constraints was used to reconstruct intensity and phase. Care was taken to avoid SH-FROG GDD sign ambiguity by ensuring the incoming pulse GDD magnitude exceeded that of the compressor. Two-photon absorption in the silicon coating was evaluated and remains <1% for incident intensities beyond 50 GW/cm^2.

• Simulations (FDTD): From 760–840 nm, group delay follows a near-linear profile with constant anomalous GDD ≈ −71 fs^2 and <2 fs deviation; a more linear sub-band (780–825 nm) yields anomalous GDD ≈ −82 fs^2 with <0.5 fs deviation. Predicted transmission is near unity across the band.
• Experimental spectral characterization (white-light interferometry): Three devices with nanopillar diameters 147±7 nm, 158±7 nm, and 170±6 nm showed full working ranges of 705–775 nm, 740–810 nm, and 770–850 nm, respectively. Full-range average GDDs: −71±1 fs^2, −61±2 fs^2, and −64±2 fs^2. Linear working ranges: 735–770 nm, 775–805 nm, 800–840 nm with average GDDs of −120±2 fs^2, −128±6 fs^2, and −127±4 fs^2. Average transmission in the working ranges ≈ 79–81%.
• The coatings compensate the GDD of up to ~2 mm of fused silica around 800 nm (SiO2 GDD ≈ +36.2 fs^2 mm^-1 at 800 nm). The compressor’s GDD magnitude per unit thickness is ~5800× that of glass.
• Residual narrow Fabry–Perot signatures limit the linear working range high-frequency side but enhance anomalous GDD in the linear band. A broadband transmission dip on the low-frequency side reduces average transmission to ~80%; simulations including a 4 nm diameter tolerance reproduce this, suggesting near-unity transmission with improved fabrication.
• Above the working range, the realized coatings can induce positive GDD up to +83 fs^2, enabling compact pulse stretchers when spectrally shifted.
• Time-domain pulse compression (SH-FROG): Using a Ti:sapphire oscillator, the compressor (162±6 nm diameter device) reduced pulse FWHM from (48.3±1.2) fs to (37.5±1.3) fs (≈11 fs shortening). SH-FROG-retrieved group delay profiles agree with white-light interferometry within uncertainty.
• Measured GDDs over 760–840 nm: Incoming pulse +191±5 fs^2; substrate alone +210±4 fs^2; compressor+substrate +154±4 fs^2. Extracted optics GDD: substrate +17±6 fs^2; compressor −58±6 fs^2.

The work demonstrates that nanostructured transmissive coatings can provide strong, broadband anomalous GDD to counteract normal material dispersion in the visible–NIR, directly addressing the need for compact dispersion compensation without angular-dispersive optics or reflective compressor components. By exploiting slow-light behavior near the HE12-like mode cutoff and mode mixing with the HE11-like mode, the coatings achieve nearly constant negative GDD across bandwidths up to 80 nm around 800 nm while maintaining high transmission. Experimental spectral and temporal measurements confirm both the magnitude and functional impact: direct pulse shortening and measured negative GDD of the coating. Compared to chirped mirrors, the approach offers competitive dispersion per thickness and easy integration on transmissive optics (e.g., substrates, beamsplitters), simplifying beam paths and alignment. The ability to tune the central wavelength by adjusting nanopillar diameter and to control dispersion magnitude via height, along with potential for inducing positive GDD outside the main band, underscores versatility for compressors or stretchers. Agreement between SH-FROG and white-light interferometry validates the characterization approach. These results suggest practical routes to anti-/non-dispersive optics for femtosecond systems and open pathways to more complex temporal shaping via advanced nanophotonic designs.

The study introduces and validates ultrathin slow-light nanocoatings based on amorphous silicon nanopillar arrays that impart broadband anomalous GDD upon transmission in the 700–850 nm region, enabling compensation equivalent to up to ~2 mm of fused silica and direct compression of femtosecond pulses. Simulations and experiments show near-constant negative GDD (≈−61 to −71 fs^2 over up to 80 nm) with linear working sub-bands achieving ≈−120 to −128 fs^2, and average transmissions around 80% with a clear path to improvement. Time-domain SH-FROG confirms pulse shortening from ~48 fs to ~38 fs through the coated substrate. The approach is compact, compatible with conventional optics, polarization- and angle-tolerant within a few degrees, and highly adaptable via geometric tuning. Future research could pursue stacked or more intricate metasurface designs, inverse-design and machine-learning optimization to broaden bandwidths, tailor arbitrary temporal phase profiles, and control complex pulse shapes for applications such as coherent control and nonlinear optics (e.g., high-harmonic generation).

• Residual narrow Fabry–Perot resonance within the working range limits the linearity on the high-frequency side, though it enhances anomalous GDD in that sub-band; periodicity tuning can suppress it further.
• Average transmission in the working band is ~80%, attributed to fabrication tolerances (e.g., ±4 nm diameter variations); improved fabrication should approach unity transmission.
• Practical devices achieve about half the theoretical slow-light dispersion limit due to non-ideal coupling and non-parabolic dispersion near cutoff.
• Certain nanopillar height ranges can induce unwanted reflectivity, constraining realizable thicknesses.
• Angle and polarization dependence were only characterized up to 5° for s-polarized light; broader angular/polarization performance is not fully detailed.
• Device area demonstrated is 3.1 mm^2; while scalable with industrial lithography, large-area uniformity is not experimentally shown here.