Laser-assisted direct roller imprinting of large-area microstructured optical surfaces

The study addresses the need for low-cost, high-throughput fabrication of large-area microstructured polymer films used in advanced optical functions (antireflection, polarization, photonic crystals, light extraction, diffusion) and emerging flexible optoelectronics. Conventional nanoimprint approaches (UV-NIL, soft lithography, thermal nanoimprinting, TNI) face limitations for large-area continuous production: high equipment complexity, need for coatings or UV exposure, and especially, in TNI, long cycle times and energy usage due to bulk heating/cooling and risks of microstructure damage by polymer reflow. Roller imprinting is promising for continuity but is constrained by the need to heat and cool during contact, leading to defects or the need for costly belt molds. The research proposes and evaluates laser-assisted direct roller imprinting (LADRI), which locally heats only the mold–film interface with a scanned laser during rolling, hypothesizing that it enables rapid, large-area replication with minimal film distortion. The study investigates how laser power density, scan speed, feature size, and contact pressure affect replication speed and fidelity, supported by transient heat-conduction and polymer-flow analyses.

Prior work established NIL as a low-cost alternative to photolithography, with UV-NIL enabling high-resolution patterning on coated resists but requiring optics and curing systems. Soft lithography suits thermocurable polymers but is laboratory-oriented. TNI allows direct replication on bulk thermoplastics by heating above Tg and cooling, yet suffers from cycle-time limits and reflow-related damage due to bulk thermal inertia. Direct surface-heating strategies sought to overcome these issues: LADI used laser heating through quartz molds to replicate nanostructures; subsequent methods included thin-film current heating and laser-assisted mold-surface heating for rapid replication on bulk polymers or glass. Roller imprinting offers continuous large-area processing, but conventional TNI-based roller nanoimprinting struggles because the polymer must be cooled below Tg before demolding; belt-mold systems separate heating and cooling but are expensive and complicate flatness control. This context motivates LADRI, combining laser scanning with roller imprinting to achieve localized, transient heating and rapid cooling via the mold body.

LADRI system: A roller mold with a microstructured Ni stamper surface is pressed against a PMMA film by a quartz backup roller while a 1070 nm single-mode fiber laser (up to 100 W) is scanned via a galvanometer and focused (f-theta lens) through the quartz roller and film onto the mold surface. Key components and parameters: backup glass (quartz) roller diameter 50 mm; PMMA film thickness 75 µm (Tg ~100 °C); roller mold: 200 µm-thick Ni stamper wrapped on a 70 mm-diameter Al-alloy cylinder; laser spot diameter ~340 µm at the mold surface; measured mold absorptivity ~29% (from reflectance). The laser heats the mold surface locally; after the beam moves, rapid cooling occurs by conduction into the Ni stamper and roller body while the film base remains near room temperature. Experimental replication: Diffraction grating patterns (EB-lithography master → Ni electroformed stamper) with pitches 250–500 nm and 230 nm square holes were replicated across 100 mm × 100 mm PMMA areas. Typical operating conditions included 100 W laser power and scan speeds around 6.1 m/s; additional measurements covered up to 8 m/s to calibrate simulations. Pattern fidelity was verified by SEM and AFM; patterned images (e.g., "UTokyo") were produced via laser shuttering. Additional target structures included: (i) subwavelength conical antireflection (AR) structures (several hundred nm) and (ii) microlens arrays (MLA) with ~10 µm lens diameter for light extraction; for MLAs, contact pressure was varied to study filling behavior. Heat-conduction simulation: A 3D FEM (ANSYS) modeled flat layers of Ni, PMMA, and SiO2 with transient heat conduction: ρ c ∂T/∂t = λ(∂²T/∂x²+∂²T/∂y²+∂²T/∂z²)+Q, with Gaussian volumetric heat input localized at the Ni–PMMA boundary. Material properties (examples): Ni (ρ=8.9×10³ kg/m³, c=440 J/kg·K, λ=91 W/m·K), SiO2 (2.2×10³, 710, 1.4), PMMA (1.2×10³, 1500, 0.2). Thermal contact resistances were set to zero. Initial temperature 25 °C. The heat transfer coefficient to ambient was tuned to 65,000 W/m²·K to match experiments at 100 W, 8 m/s; with 29% absorption, internal heating was set to 29 W. Irradiation time per point estimated as t_spot = d_spot/V_scan (e.g., 0.34 mm / 3.2 m/s ≈ 0.106 ms). Replication speed U_rep = w_rep·V_scan when local temperature exceeded 100 °C. Polymer-flow consideration: Simulations and experiments indicated that within several tens of micrometers, the heat conduction timescale and the viscous flow timescale of the molten PMMA surface are comparable, influencing filling in features like MLAs. Contact pressure effects on filling depth for MLAs were quantified experimentally.

• LADRI enables rapid, large-area replication by locally heating only the mold–film interface; replicated diffraction patterns covered areas of 100 mm × 100 mm with high fidelity and minimal film crinkling at irradiated/non-irradiated boundaries.
• Replication speed increased with higher laser power density and scan speed; however, sufficient irradiation time is required to allow polymer to fill the full depth of microstructures, indicating an optimal operating window.
• Transient heat-conduction FEM showed that shorter irradiation times more efficiently raise PMMA surface above Tg due to Ni’s higher thermal conductivity drawing heat inward; calibrated with HTC ≈ 65,000 W/m²·K and measured absorptivity of 29% at 1070 nm (internal heating ≈ 29 W for 100 W incident).
• Analytical relations: t_spot = d_spot / V_scan; replication width and replication speed U_rep = w_rep·V_scan correlate with scan speed, matching experiments under calibrated conditions.
• Maintaining the film base near room temperature (~25 °C) prevents global thermal distortion, thereby avoiding crinkles and preserving film flatness and thickness, unlike conventional TNI.
• Application demonstrations: • Antireflection structure: reflectivity reduced from ~4% to ~0.5% after replication. • Light-extraction MLA (~10 µm lenses): output light intensity increased by a factor of ~1.47; replication degree was strongly governed by contact pressure.
• Polymer-flow analysis indicated that heat conduction and molten PMMA flow speeds are comparable over tens of micrometers, explaining pressure dependence in MLA filling.

The results support the hypothesis that localized laser heating during roller imprinting can overcome TNI’s thermal inertia and reflow challenges while enabling continuous, large-area patterning. By confining heating to the mold–film interface and allowing rapid post-irradiation cooling via the mold body, LADRI achieves higher replication speeds without inducing film-wide thermal stresses, eliminating crinkling and preserving flatness. The combined experimental–simulation approach clarifies the process window: increased power and scan speed enhance throughput, but adequate dwell (t_spot) is needed to ensure full-depth filling, especially for deeper or larger features. The calibrated FEM accurately predicts replication width versus scan speed, providing a tool for process design. Application tests validate functional benefits: AR nanocones yield near an order-of-magnitude reflectivity reduction, and MLAs improve light extraction, with filling governed by contact pressure—highlighting the need to co-optimize pressure with thermal input. Overall, LADRI addresses key scalability and quality issues in microstructured film manufacturing for optical devices.

This work introduces laser-assisted direct roller imprinting (LADRI) as a high-throughput, low-distortion method for fabricating large-area microstructured optical films. The study experimentally demonstrates uniform replication over 100×100 mm areas and validates process scaling with laser power and scan speed, guided by a transient heat-conduction FEM. Optical applications show significant performance gains: AR reflectivity reduced from ~4% to ~0.5% and MLA light intensity increased by ~1.47×. The process inherently minimizes film crinkling by keeping the film’s bulk near room temperature. Future work should address continuous-operation thermal management (e.g., active roller cooling to control mold temperature rise), refine dimensional accuracy considering Ni–PMMA CTE mismatch, and further couple polymer-flow modeling with process parameters (pressure, dwell) to optimize filling for diverse feature sizes and aspect ratios.

• The publication focuses on demonstration; continuous, long-duration runs likely lead to mold temperature rise and thermal equilibrium effects not fully characterized here.
• Thermal contact resistances were neglected in simulation; real interfaces may introduce additional resistances affecting temperature profiles.
• The calibrated heat transfer coefficient is empirically fit to specific conditions (100 W, 8 m/s); generality across tools and environments needs verification.
• Dimensional fidelity may be affected by thermal expansion mismatch between Ni and PMMA; this and absolute dimensional accuracy were not fully quantified.
• Optimal pressure–temperature–time windows for deep or high-aspect-ratio features require further systematic study, especially under varying contact pressures and scan strategies.