Ultrafast 3D nanofabrication via digital holography

The study addresses the need for high-resolution, complex 3D nanostructures and the limitations of conventional two-photon lithography (TPL), which relies on serial point scanning and is too slow and costly for large-scale applications. Prior attempts to increase throughput using multiple beams (interference, microlens arrays, diffractive optical elements) lack individual focus control and are largely limited to periodic structures. Programmable beam shapers have been used, but throughput is constrained by limited foci (<10) due to insufficient peak power from common femtosecond oscillators. Projection-based TPL can improve throughput but struggles with overhangs and complex 3D structures due to its layer-by-layer nature. Regenerative femtosecond laser amplifiers offer much higher peak power but have low repetition rates (1–10 kHz), which alter polymerization thresholds and kinetics, raising concerns about resolution and structural integrity. The paper proposes a multi-focus, digital holography-based TPL platform using a 1 kHz fs regenerative amplifier and a custom photoresist to overcome these challenges and achieve ultrafast, scalable 3D nanofabrication.

- Multi-beam interference, microlens arrays, and DOE approaches enable parallelization but lack individual control and are suited mainly for periodic structures (refs 18–20). - Programmable beam shaping (SLM/DMD) allows dynamic control of individual foci but has been limited by device speed and insufficient peak power from oscillators, typically supporting <10 foci (refs 21,22). - Projection-based TPL using depth-resolved fs light sheets improves throughput but requires layer-by-layer fabrication and struggles with overhangs without supports (ref 25). - Low-repetition-rate regenerative amplifiers (1–10 kHz) provide high peak power (~10 GW) but reportedly increase polymerization threshold and reduce dynamic range, altering kinetics (refs 24,26–28). - Proximity effects in multi-spot DLW and strategies to mitigate them have been reported (ref 35). These works motivate a system that combines high peak power with individually programmable multi-focus control and kinetics-aware processing to preserve resolution and enable complex 3D builds.

System architecture: A Ti:sapphire fs regenerative amplifier (800 nm, 1 kHz, 100 fs, 4 W) feeds a DMD-based holographic scanner. Binary holograms synthesized by a weighted Gerchberg–Saxton algorithm generate up to 2000 individually controlled foci at the DMD Fourier plane; a spatial filter removes unwanted orders. A reflective blazed grating (600 lines/mm) and 4-f pre-compensation (L1=225 mm, L2=250 mm) counteract DMD-induced angular dispersion. A second 4-f relay (L4=200 mm) projects foci to the objective; high-NA oil immersion objective performs dip-in fabrication in index-matched photoresist. Substrates (FTO glass) are mounted on a 6-DOF hexapod for stitching beyond the DMD work volume (~299×554×760 μm^3). Hologram control and scanning: Holograms are synchronized at 1 kHz (per-pulse updating) enabling random-access 3D scanning; each focus has independent amplitude, phase, and position control. The theoretical voxel rate is number of foci × 1 kHz. Photoresist design: Monomer base: BPADA (68 wt%) + PETA (32 wt%). Photoinitiator: a bis-donor, symmetrically substituted conjugated chromophore (two-photon cross-section ~800 GM at 800 nm), chosen to favor 2PA and suppress multiphoton ionization at peak intensities of 3.3–22.7 TW/cm^2. Inhibitor: 4-hydroxyanisole (MEHQ) 70 ppm. Initiator concentration: 0.4 wt%. UV-Vis spectra measured; Raman used to assess degree of crosslinking (DC). Single-pulse exposure model: With 1 kHz repetition, exposure spacing relative to diffusion time thresholds (~20 ms for PETA to diffuse ~150 nm) determines voxel growth. Multi-pulse exposures (>20 pulses per voxel) lead to diffusion-driven tapering and linewidth expansion until reaching a steady-state diffusion limit (~700 nm lateral). Single-pulse exposure per voxel minimizes diffusion-induced broadening and roughness, achieving higher resolution while maximizing rate. Experimental validation printed nanowires with 1, 20, 100, 200 pulses per voxel at equal total dose, showing increased tapering/roughness with more pulses. Avalanche ionization onset was inferred at >~10 nJ per pulse (>25.9 TW/cm^2); printing was conducted at 3–7 nJ to ensure 2PA-dominant polymerization. Multi-focus optimization: Proximity-induced overpolymerization observed when foci are closer than ~3 μm; mitigated by enforcing minimum inter-focus spacing ≥3 μm, single-pulse exposure, and WGS-based intensity uniformity (up to 99%). System can support >2000 foci (potentially >4000 with trade-offs in focus quality and intensity uniformity). Lower magnification objectives can expand work volume at slight cost to minimum step size. Fabrication protocol: Substrate cleaning (sonication in glass cleaner, DI water, IPA; oven dry at 80°C). Dip-in printing in custom resist; post-development in PGMEA 15 min and IPA 10 min; air dry. Fragile structures optionally freeze-dried (−40°C freeze then −84°C lyophilization). SEM (Pt sputter coat, 5–10 kV), optical microscopy, and confocal Raman used for characterization. Functional materials and devices: Magnetic resist prepared by dispersing 3.0 wt% PAA-functionalized Fe3O4 (8 nm) into the custom resist via planetary micro-milling; printed micro-gears actuated by 10 mT Helmholtz coils in PVP aqueous medium. Mechanical testing of metastructures via rheometry: time-sweep oscillatory tests (1% strain, 1 Hz), static compression (0.1 s^-1), cyclic loading (20 cycles). Carbonization in N2 at 900°C (5°C/min ramp, 1 h hold) yielding ~73% isotropic linear shrinkage.

- Parallel TPL via digital holography with up to 2000 individually programmable foci at 1 kHz achieves 2,000,000 voxels/s. - Achieved sub-200 nm resolution: 90 nm lateral and 141 nm axial, among the best reported for parallel/projection TPL. - Printing speed for low filling-ratio (1–12%) structures: 4.5–54.0 mm^3/h with 2000 foci, surpassing most parallel TPL methods, while maintaining grayscale accuracy >99%. - Custom photoresist exhibits single-pulse polymerization threshold of 1.27 nJ and a wide dynamic range of 12.46; DC increased from 16.2% to 48.5% as single-pulse energy increased 1.37→6.85 nJ. - Single-pulse exposure eliminates diffusion-induced tapering/roughness seen with multi-pulse per voxel at 1 kHz, preserving feature fidelity and breaking the traditional resolution–throughput trade-off. - Energy efficiency: average power per focus array 20–400 mW (for 100–2000 foci), markedly lower than DOE (~4 W) and projection-based (~1.5 W) high-throughput systems. - Large-scale metastructures: 1.08×1.08×1 mm^3 octahedral truss lattice (6.82×10^5 unit cells, 7.9 μm unit size, ~700 nm linewidth) fabricated with 100 foci; demonstrated ~30% compressibility and recovery. - Mechanical properties: Dynamic moduli measured; Young’s modulus of two truss designs measured as 35 MPa and 80 MPa; 20–30% mechanical resilience over cycles; carbonized lattice exhibits G' = 22.7 kPa and Young’s modulus = 497.3 MPa with ~73% isotropic shrinkage. - Complex devices: 36-focus grayscale writing of alphanumeric features in 150 ms; centimeter-scale diffractive surfaces (1×4 cm^2) producing mono/rainbow color; arrays of woodpile annuli, egg-beater structures, microlenses. - Magnetic micromachines: Arrays of micro-gears (voxel pitch 500 nm) printed in 49 s via 32 foci; individual gear fabricated in 290 ms; actuated at 10 mT to realize translation, 5 Hz rotation, and 1 Hz flipping.

The work demonstrates that combining a 1 kHz high-peak-power regenerative amplifier with DMD-based digital holography and a high-2PA, wide-dynamic-range photoresist enables true multi-focus, random-access TPL without sacrificing resolution. The single-pulse per voxel strategy addresses diffusion-driven broadening inherent to low-repetition-rate sources, thereby maintaining nanoscale feature fidelity while maximizing voxel rate. Mitigation of proximity effects via ≥3 μm inter-focus spacing and holographic intensity uniformity allows scaling to thousands of simultaneously active foci. These advances directly address the central challenge of scaling TPL from lab-scale prototyping to practical, large-area, complex 3D structures with low material fill. The resulting platform delivers state-of-the-art resolution, high throughput, precise grayscale control, and improved energy efficiency, expanding the applicability of TPL to photonics, metastructures, micro-robotics, and other fields.

This study introduces an ultrafast multi-focus TPL platform based on digital holography and a 1 kHz regenerative amplifier, achieving up to 2000 independently controlled foci, 2×10^6 voxels/s, and 90/141 nm lateral/axial resolution. A custom photoresist with low single-pulse threshold and wide dynamic range, combined with a single-pulse exposure strategy, overcomes diffusion and proximity limitations at low repetition rates. The system fabricates centimeter-scale diffractive surfaces and millimeter-scale metastructures with reproducible mechanical performance, and rapidly produces functional micromachines. Future improvements suggested by the results include expanding the number of foci (e.g., via lower magnification objectives to increase work volume) while managing focus quality, and leveraging programmable holography for advanced beam shaping and error compensation to further enhance throughput and functionality across application domains.

- Proximity effects: Over-polymerization and even bulk solidification can occur when many foci are closely spaced; mitigated by enforcing minimum inter-focus spacing ≥3 μm and using single-pulse exposure. - Focus quality and uniformity trade-offs at very high parallelization: While >2000 (potentially >4000) foci are possible, higher counts can slightly degrade focus quality and intensity uniformity, increasing risk of overpolymerization. - Work volume constraints: The DMD-scanner work volume (~299×554×760 μm^3) necessitates stitching for larger parts, adding system complexity. - Avalanche ionization risk at high pulse energies: Evidence of avalanche-induced polymerization above ~10 nJ per pulse (>25.9 TW/cm^2) imposes an upper bound on per-pulse energy for high-fidelity features. - Composite (magnetic) photoresist increases writing threshold (~8.6×), which can reduce process window in systems with lower peak power (though handled here by the amplifier).