Metasurface-Enabled Holographic Lithography for Impact-Absorbing Nanoarchitected Sheets

brick-and-mortar-like patterns to result in ≈ 50–70% porosity. Nanoindentation arrays over the entire sample area reveal the out-of-plane elastic modulus to vary between 300 MPa and 4 GPa, with irrecoverable post-elastic material deformation commencing via individual nanostrut buckling, densification within layers, shearing along perturbation perimeter, and tensile cracking. Laser induced particle impact tests (LIPIT) indicate specific inelastic energy dissipation of 0.51–2.61 MJ kg − 1 , which is comparable to other high impact energy absorbing composites and nanomaterials, such as Kevlar/poly(vinyl butyral) (PVB) composite, polystyrene, and pyrolized carbon nanolattices with 23% relative density. These results demonstrate that holographic lithography offers a promising platform for scalable manufacturing of nanoarchitected materials with impact resistant capabilities.


Introduction
Hierarchical arrangements of multi scale structural constituents play a cru cial role in achieving multifunctionality and in optimizing material properties. For example, the hierarchical archi tecture of wood allows trees to com bine favorable and varied mechanical properties, such as high axial stiffness, with complex networks of xylem and phloem for nutrient transport. Inspired from nature, artificial materials such as nanoarchitected materials and nano lattices have been developed to achieve desired material properties such as enhanced damagetolerance, [1] reus able energy absorption [2] and stimulus responsivity. [3] For example, nanolat tices with ultralow relative densities of 5%, exploit nanoscale induced material size effects to derive higher strengths and stiffnesses compared with their con stituent material, others can suppress brittle failure in architected materials comprising intrinsically brittle mate rials. [4,5] Optimizing lattice architecture by, for example, beam versus platebased patterns and peri odic versus stochastic geometries, enables approaching the theoretical mechanical parameter limits in engineered mate rials. [6] Beyond these advantageous quasistatic mechanical properties, Portela et al. [7] demonstrated that nanoarchi tected pyrolytic carbon with a relative density of 23% (i.e., 77% porosity) can combine materiallevel size effects and structurelevel densification mechanisms in cellular solids to mitigate impact of 14 µmdiameter projectiles traveling from 50 to 1000 m s −1 with a 70% higher specific energy dissipa tion compared to Kevlar.
To date, the fabrication of nanolattices has been primarily accomplished via a small subset of nanofabrication techniques, the most versatile of which is twophoton lithography (TPL) that offers maskless 3D printing, complex pattern fidelity, and submicrometer resolution. [8] TPL offers feature resolution at the scale of hundreds nanometers in aswritten structures, which enables printing of arbitrarily complex 3D nanoarchi tectures in photosensitive polymer resins. One disadvantage of this approach is the long fabrication times caused by inherent limitations of serial manufacturing, with parts assembled via a layerbylayer or a voxelbyvoxel process. Alternative approaches www.advmat.de www.advancedsciencenews.com based on colloidal self assembly have also been proposed for the fabrication of nanoarchitected materials. Jiang et al., [9] intro duced nanoarchitected material synthesis that does not involve layerbylayer construction by using a selfassembled colloidal template with a conformally electroplated 100 nmthick Ni film to create 90 µmthick defectfree nanolattices. The resulting structures are restricted to inverse opal geometry and require long fabrication times on the order of multiple days.
Holographic lithography has been demonstrated as a viable platform for rapid and scalable manufacturing of nanoarchi tected materials. [10][11][12][13] The fundamental principle of holographic lithography is the utilization of light interference to generate a three dimensional pattern that is then imprinted in a photo sensitive polymer. The light interference is either achieved by multiple independent light beams or by the utilization of a dif fractive mask that splits a single incoming beam into multiple diffracted orders that interfere coherently. Approaches of holo graphic lithography for the fabrication of nanoarchitected struc tural materials include the work of Bagal et al. who utilized 3D colloidal nanolithography [14] to produce 1 µmthick nanostruc tured metal oxide films [15] with the outofplane elastic moduli of 1.19 GPa and specific energy dissipation of 0.32 MJ kg −1 , defined as energy dissipated per unit mass. This approach uti lizes a single layer of colloidal nanoparticles as a phase shifting photolithographic mask. This allows for costeffective and scal able fabrication of nanoarchitected sheets. Nonetheless, the achievable pattern fidelity is severely limited by the utilization of spherical nanoparticles. Other examples include the work of Bae et al. [16] who utilized a binary phasemodulating photo mask to produce 4 × 4 in 2 regions with 600 nm periodic pat terns in 10 µmthick negative tone photoresist by a single laser exposure. A major limitation of binary phase masks is the limited control over the generated pattern morphology caused by the their inherently low diffraction efficiencies which effec tively define the pattern's contrast and morphology. Kamali et al. demonstrated with numerical simulations that optical metasurfaces, or metamasks as we will refer to them here, can generate complex 3D nanoarchitectures [17] not achievable with conventional diffractive optics. Metasurfaces are composed of subwavelength features that locally modulate the phase of incoming light, the phase modulation can be polarization selec tive and therefore complex patterns that are not possible with diffractive masks can be generated under illumination with appropriate polarization components. For example Kamali et al., numerically demonstrated the generation of patters such as gyroid, rotated cubic and diamond which are not achievable by conventional diffractive optics. [17] Nonetheless, experimental demonstration of such patterns or the utilization of optical metasurfaces for holographic lithography is still to be desired.
In this work, we present a scalable fabrication pathway based on holographic lithography, for nanoarchitected materials that exhibit a high specific energy dissipation which is ideal for impact mitigation. In contrast to previous implementations of holographic lithography, our approach utilizes optical meta surfaces that expand the design space of possible nanaoarchi tectures. The presented fabrication methodology utilizes laser exposure of SU8 photoresist through a phase optical metasur face mask to produce 30 to 40 µmthick nanoarchitected sheets with 2.1 × 2.4 cm 2 lateral dimensions and ≈500 nmwide struts organized in layered 3D brickandmortarlike patterns. These nanoarchitected materials have relative densities ρ between 20% and 50%, elastic moduli between 300 MPa and 4 GPa, and specific energy dissipation of 0.51-2.6 MJ kg −1 when subjected to a micro projectile impact from 50 to 1000 m s −1 , a perfor mance indicator comparable to Kevlar. This performance stems from a combined effect of the constituent polymeric SU8 with high stiffness and toughness and the structural instabilities involving multiple elasticplastic deformations.

Fabrication Procedure
To fabricate nanoarchitected materials with macroscopic lat eral dimensions, we designed a laser scanning lithographic system (Figure 1a). The system is utilized to expose 1 mm thick 2 × 2 in 2 sodaline slides covered with a layer of 30-40 µm negativetone epoxybased photoresist (SU8 2050) sensitised for 532 nm light and deposited by spincoating. The system is composed of a 2D highprecision linear piezostage and a laser beam of wavelength 532 nm. The beam is directed orthogo nally to the scanning plane of the stage, which sequentially rasters the metamask placed in direct contact with the pho tosensitive material, under the fixed laser beam (Figure 1b). This scanning approach is amenable to scalable fabrication of samples with macroscale lateral dimensions because it can easily accommodate a large range of metamask dimensions. We optimized the process by controlling the width of the laser beam w, the stage scanning speed v, and the scanning step size t between sequential lines to create a uniformly accumu lated exposure dose, defined as D = P/vt, where P is the beam power, over the metamask (see Figure S1, Supporting Infor mation). For our experiments we utilized laser powers in the range of 0.23-0.25 W, the line spacing t was chosen at 0.4 mm and the scan speed u was set to 1.5 mm/s which resulted in doses between 0.38 and 0.41 J mm −2 . The metamask we utilized was composed of an array of 900 nmspaced facecentered unit cells of five 115 nmdiameter 300 nmtall crystalline Si nanopil lars spaced at s = 225 nm on top of 1 mmthick quartz substrate. Figure 1c contains scanning electron microscopy (SEM) images of a 3 × 3 array of unit cells spaced by distance p = 900 nm (top view) and a zoomedin view of a single unit cell (45° tilt view). The metamask covered a patterned area of 2.1 × 2.4 cm 2 and was fabricated by electron beam lithography and deep reac tive ion etching as descried in the Experimental Section. The utilized metamask generates a near field interference pattern that resembles struts organized in layered 3D brickandmortar like patterns as shown in Figure 1d. The light intensity pattern was calculated by finite difference time domain (FDTD) and angular spectrum propagation simulations. We exposed areas of 3 × 3 cm 2 which took a total time of 24 min. Upon laser expo sure, photoacid was generated from an oniumsaltbased visible light photoinitiator at concentrations proportional to exposure intensity. Following exposure, the sample was baked at 65° C for 270 s to promote the photoacid and polymer chain diffu sion necessary for cationic polymerization of the oxirane rings. The photoresist was developed in propylene glycol monomethyl ether acetate (PGMEA) for 2 h, after which it was washed in isopropanol to remove unreacted SU8, and then air dried. The www.advmat.de www.advancedsciencenews.com exact geometry of the resulting pattern is directly linked to the 3D intensity distribution generated by the metamask. Figure 2 shows a photograph of a typical 30 µmthick nano architected sheet produced via this process patterned over the area of 2.1 × 2.4 cm 2 on a glass substrate. The opaque region of the sheet corresponds to the nanoarchitected material, and the translucent periphery corresponds to the monolithic unpat terned SU8. The white appearance of the sheet is attributed to its strong scattering over a broad band of visiblelight wave lengths. Crosssectional SEM images demonstrate that the pat tern was preserved and successfully developed throughout the thickness of the material (Figure 2b), with the resulting archi tecture of brickandmortar arrangements of beams, with each horizontal layer shifted by half a period in both lateral direc tions respective to the underlying one (Figure 2b inset). The SEM images reveal a lateral periodicity of 900 ± 10 nm, which matches the periodicity of the metamask, and a vertical perio dicity of 1.62 ± 0.02 µm, which is 30% lower than the perio dicity of 2.39 µm predicted by finite difference time domain simulations (shown in Figure 1d). Based on the crosssectional images at different depths throughout the sample (Figure 2c), the beam diameter d increases by ≈15% from the sample surface to base. The average beam width varied between 462 ± 42 nm in the top layers to 523 ± 26 nm in the bottom layers. The unit cells in the upper 20% of the sample height have a fully opencell morphology (Figure 2c(i)), the unit cells in the center region and below (Figure 2c(ii),(iii)) preserve the laterally opencell morphology but lose the vertical interconnec tivity. Based on the measured beam diameters, we estimate that the relative density (volume fraction) ρ to vary between 0.2 and 0.35 throughout the thickness of the material.

Quasi-Static Nanoindentation and Deformation Characterization
To investigate the mechanical properties of the nanoarchitected materials, we conducted quasistatic nanoindentation experi ments at a strain rate of  ε ≈ 10 −4 s −1 with each indent compressing ≈7,850 unit cells across the 100 µm diameter flat punch indenter tip to a depth of 12 µm. A representative indentation load versus displacement curve is shown in Figure 3 and captures three dis tinct regions: 1) an initial linear elastic regime up to a strain of 0.13 followed by 2) a plastic deformation regime with a gradually  www.advmat.de www.advancedsciencenews.com increasing load plateau and discrete displacement bursts up to a strain of 0.37, and 3) partial elastic recovery of 0.027 strain during unloading. Prior to unloading, the load was held at its peak value of 115 mN for 100 s to decouple the intrinsic timedependent deformation of the viscoelastic polymer during the load removal (i.e., unloading segment). The stiffness of the material in each regime, calculated as the slope of load and displacement, is indicated by the dashed lines in Figure 3, which reveals that the postyield regime has a plateaulike response, characterized by a ≈75% lower slope (S 2 ) compared with that in the initial elastic regime (S 1 ), a pronounced steplike behavior, containing a series of discrete displacement bursts, indicated by arrows in Figure 3, and strain hardening. This response is a result of indentation induced shear stresses generated by the boundary constraints of the cylindrical punch and the densified layers beneath the indentation mark. The unloading curve indicates the significant amount of plastic deformation with the slope (S 3 ) being ≈720% stiffer than the initial elastic stiffness. The full postindentation impression mark is shown in Figure 3b and confirms the inclu sion of multiple unit cells across the flat punch diameter (inset) and the establishment of full parallel contact between the tip and the sample surface. The crosssectional SEM image of the inden tation imprint shows that 5-8 layers collapsed within the densi fied region under the indenter tip and that plastic deformation occurred mainly within the crushed area. Within the densified layers, nanostruts on the same layer were bucked to the same direction, which alternates across the depth, indicative of the strong shear stresses during the compressive loading. We also observed several tensile cracks around the contact perimeter, indicative of the development of tensile membrane stresses on sample surface. Shearoff along the imprint perimeter occurred via severe plastic deformation followed by deeper penetration of fracture from the load increase. [18,19]

Effective Modulus and Diffraction Efficiency Mapping
To probe the mechanical uniformity across the entire sample, the indents were distributed at an average spacing of 820 µm in X and 840 µm in Y on a 25 × 25 Cartesian grid across the sheet (Figure 3c). We utilized the continuous stiffness meas urement (CSM) technique by imposing a 2 nm displacement oscillation amplitude onto global displacement at a frequency of 45 Hz. Imposing this loading waveform allows for contin uous monitoring of the contact stiffness from the measure ment of the phase difference between the oscillatory force and displacement response of the system. [20] The effective inden tation modulus was calculated following the Oliver-Pharr method [20] at a depth of 10% of the sheet thickness (≈3 µm after the full contact) to avoid the substrate effect. The histo gram in Figure 3d(i) reveals that the average elastic modulus collected over the entire sample is 0.96 ± 0.52 GPa, with a range of 300 MPa-4 GPa. The distribution contains a long tail that approaches the modulus of monolithic SU8, reported to vary from 3.5 to 6 GPa. [21][22][23][24][25][26] This right skewed histogram reveals that ≈88% of the total number of indents have a modulus between 300 MPa and 1.5 GPa; within this range the distribu tion is more symmetric, with the average modulus of 0.80 GPa and a standard deviation of 0.25 GPa. According to the spatial modulus map in Figure 3d(ii), the relative uniformity of the modulus was observed in the center of the sample, whereas stiffer regions were distributed near the edges. This mechanical pattern traces the diffraction landscape of the metamask as the spatial modulus map is closely correlated to the distribution of diffraction efficiency, α of the (0,1) diffraction order of the meta surface, shown in Figure 3d  www.advmat.de www.advancedsciencenews.com of 2.07 GPa and 0.73 GPa, respectively. Further analysis of the transmitted and diffracted intensities of the metamask in the marked locations reveals that the sample location with α ≈ 0.85 received an ≈27% greater optical dose compared to the region with α ≈ 0.37. The results of these experiments reveal a strong correlation between the mechanical properties and the metamask diffraction efficiency, with those regions where the diffracted intensity is greater being stiffer, which is likely caused by the greater degree of crosslinking in the underlying polymer and microstructural variation.

Response to Impact via Laser Induced Particle Impact Testing (LIPIT)
To measure the impact absorbing capability of the fabricated material, we performed laser induced particle impact tests (LIPIT), [27] on 40 µmthick nanoarchitected sheets, during which 14 µmdiameter SiO 2 microparticles were launched with initial velocities between 50 and 10 3 m s −1 to impact the material. These experiments were performed on a sample with a relative den sity, ρ ≈ 50%, and the impacting particle size was approximately one order of magnitude greater than that of the unit cell (lateral periodicity of 900 ± 10 nm), which allows for the separation of relative scales between the impact area and the characteristic material length scale. The projectile velocities were controlled by calibrating the excitation laser pulse energy which produces a local ablation on the launching pad and propels a particle. To measure the velocities before and after impact, denoted as v 0 and v r respectively, the moving particle was tracked using a laser imaging pulse and the adjacent snapshots of its trajectory were captured by an ultrahighspeed camera (See Experimental Section for details). LIPIT experiments convey that the litho graphically patterned nanoarchitected materials exhibit near ideal energy dissipation, with a linear slope of 0.992 (R 2 = 0.99) (Figure 4a). This indicates that nearly most energy was dissi pated upon the impact to the material. Portela et al. [7] performed similar LIPIT experiments on nanoarchitected glassy carbon with similar beam dimensions in the range of 400-500 nm yet different relative density, ≈23% and reported three distinct responses as a function of impact energy W o : 1) elastic collision for impact velocities below 50 m s −1 , which corresponds to W o of <0.003 µJ, 2) particle rebound with partial cratering beyond elastic regime, and transitioning to 3) particle embedding

Fabrication and Microstructural Characterization
As shown in Figure 2a, the white macroscopic appearance of the produced samples indicates strong scattering in the visible wavelength range, an observation in contrast to the expected photonic response from highly ordered structures with perio dicity of 900 nm. This observation is consistent with the presence of morphological variations and defects within the sample. Examining the crosssectional SEM images in Figure 2 elucidated that although our fabrication method could topologi cally preserve the unit cell morphology as no collapsed or unex pected structures were observed across the sample, there were still nonnegligible morphological variations and discrepancies from the theoretical and numerically expected pattern.
The simulated light intensity (Figure 1d) predicts the beams to have a convex shape, i.e., thicker in the center cross sec tion and tapering toward both ends ; SEM images in Figure 2c demonstrate that they are slightly concave. This discrepancy can be attributed to an optical dose gradient within each beam, where its midsection experiences a higher optical dose due to the nature of the interference, which causes a greater degree of crosslinking in that region and a concomitant enhanced volumetric shrinkage of SU8. Higher crosslinking densities result in lower free volume within the polymer which in turn lead to higher volumetric shrinkage. [39] The overall dose deter mines whether the beams are interconnected and whether the resulting nanoarchitecture is a closed or an opencell cellular solid. For example, if the photoresist is underexposed, the con nections between neighboring beams do not crosslink and therefore the structure falls apart, and if it is overexposed, the pores become fully closed and eventually the structure becomes fully solid as the volume fraction approaches 100%. Our experi ments indicate that when the optical dose is sufficiently high  A linear fit of the dissipated energy is presented with a green dashed line. As a function of impact energy, four deformation regimes were exhibited: i) elastic rebound, ii) shallow, and iii) deep crater formation to iv) particle capture. b) Corresponding post-impact microscopy image of each regime (scale bar: 5 µm). c) Specific dissipation energy, W d * of nanoarchitected SU8 thin sheet compared with other bulk and nanomaterials. [7,14,[28][29][30][31][32][33][34][35][36][37][38] www.advmat.de www.advancedsciencenews.com to produce closedpore foams, critical point drying is required after the development and washing with isopropanol to retain their shape. Examples of collapsed closedcell structures are shown in Figure S4 (Supporting Information). Opencell nanoarchitectures remain robust even when dried in air, with no indication of feature collapse (Figure 2). Crosssectional images in Figure 2b,c convey the presence of slight imperfec tions in the lattice geometry such as tilted beams with respect to the vertical direction, most likely caused by the significant swelling of the crosslinked polymer network during the long development time of 120 min and the sequential nonuniform shrinkage that takes place during the washing and drying steps. The longitudinal periodicity of 1.62 µm was also ≈30% less than the theoretical value, which can be attributed to the wellknown volumetric shrinkage of SU8 induced by cross linking, reported to be as high as 40%. [39][40][41] For films attached to a substrate, this shrinkage occurs through vertical contraction because of the bi axial lateral constraint posed by the rigid substrate.
The 15% thickening of the beams toward the sample bottom ( Figure 2b) indicates a depth dependent variation of one or more of the following factors: generated light pattern, pro cessing conditions of the photoresists, or chemical gradients within the photoresist. Given the high transverse coherence of the utilized laser source and the 6 mmwide laser beam, we expect a uniform 3D pattern throughout the photoresist thick ness. During processing of the photoresist the existence of both thermal and chemical gradients cannot be excluded. Cross linking of the photoresist is achieved by a post exposure baking on a hot plate, this could result in a increasing temperature gradient toward the bottom of the sample. A higher tempera ture at the bottom would result in higher photoacid mobility and therefore to wider beams. We did not quantify the thermal gradient in this work; our analysis of the process indicates that the ambient heating conditions are crucial in defining final sample morphology. Another factor that determines the mobility of the photoacid is the remaining solvent [42,43] after the preexposure soft bake of the photoresist. Thick photore sist layers are well known to develop solvent gradients due to limited solvent diffusion at the surface of the photoresist. [44] An increase in remaining solvent towards the bottom at the photoresist leads to an increased photoacid mobility and simi larly to the temperature gradient to an increased beam width at the bottom. Quantifying this solvent gradient is outside of the scope of this work. The change in pattern morphology is a combination of effects from both thermal and solvent gradients through the thick photo resist layer during the post exposure baking step.
The nonuniform diffraction efficiency of the metamasks and the thermal and solvent gradient induced during the fab rication process results to inherent variations in pattern mor phology and crosslinking density which lead to inhomogeneity in the mechanical properties of nanoarchitected SU8. Espe cially for the lateral homogeneity, since the diffraction efficiency α of the metamask is a defining factor in the contrast of the 3D intensity modulation, beam morphology and optical dose, the inconsistencies in the metamask are directly correlated to the local variation in the material response as indicated from the comparison between two spatial maps in Figure 3d. Despite of the highly rightskewness of the histogram, the broad span of the measured quasistatic indentation modulus varying from 300 MPa to 4 GPa, indicates the significance of designing and controlling the uniformity with unique sensitivity of the meta mask. Since the production of uniform metasurfaces is not fun damentally limited, the material consistency and scalability can be achieved by the fabrication method. As the elastic modulus of a cellular material is heavily dependent on both structural dimensions and intrinsic constituent material properties, [45,46] developing precise correlation of each of these design factors with the α of the metamask and quantifying the contributions of each one to the overall material modulus are also critical for exploring and populating the mechanical property parameter space for nanoarchitected materials.

Impact Resistance and Energy Dissipation Capability
The nanoarchitected design with the solidvoid space filling topologies enables these materials to undergo local deforma tion upon global loading by axial deformation and bending of the vertical struts. Figure 3b conveys that our fabrication pathway enabled constructing nanofeatured architecture and preserving their mechanical response in the continuum level during quasistatic nanoindentation, as the scaledup nano architected sheet undergoes multiple deformation processes: multilayer collapse, full pore closure, offset shearing along the orifice of the cylindrical punch, microcracking, and localized buckling. The interplay of multiple elastic-plastic deformation modes has been observed in porous materials during quasi static indentation, [26,47] which likely corresponds to a gradually increasing stress plateau with discrete strain bursts, as exhib ited in Figure 3a. Such load-displacement response during the layerbylayer collapse and localized fracture events resemble the wellknown characteristic of a cellular solid compres sion. [45,48] The stress drops at a constant strain and the extended plateau regime leads to higher hysteresis,the enclosed area in a loading-unloading loop, which represents the energy dissi pated by the material. It clearly indicates that the enhanced local deformation and the multiple distinct plastic energy dissipa tion mechanisms available to nanoarchitected polymers enable these materials to have superior energy dissipation capabilities.
Comparing to other ballistic materials, [7,14,[28][29][30][31][32][33][34][35][36][37][38] we estimate the specific dissipation energy W d * as dissipated energy W d normalized by the crater mass measured from postmortem analysis (see Experimental Section for more details). Figure 4c conveys that the nanoarchitected SU8 has the specific energy of 0.51-2.61 MJ kg −1 across a broad range of tested velocities, a range comparable to that of other composite materials used for impact absorption, such as Kevlar/poly(vinyl butyral) (PVB) composite, [33] polystyrene, [31] and pyrolized carbon nanolat tices. [7] This enhanced energy dissipation is most likely ena bled by the synergistic interplay between the high fracture toughness of SU8 and the structural instabilities. [47,49] Benefit ting from its nanoarchitecture, the patterned SU8 sheet with ρ ≈ 50% achieves the absolute density of 600 kg m −3 , a value 2-4 times lower than that of fully dense thin polycarbonate films [30] or multilayer graphene. [35] The material and experi mental para meters of other ballistic studies are summarized in Table S1 (Supporting Information).

www.advmat.de www.advancedsciencenews.com
The variability in the specific energy dissipation of nano architected SU8 tested under the same experimental conditions can be attributed to uncertainties in the crater volume measure ments and to the local variations of mechanical properties indi cated from the indentation data, which can possibly affect the size of deformation and failure mechanisms. [30] Since the crater volume was estimated from the radial symmetric assumption of the cross sections, which were prepared by focused ion beam (FIB) milling (see Experimental Section for details), the mate rial damage or removal induced from beam local heating [50,51] and measurement uncertainty induced from the charging issue [52] was not inevitable. It has been also reported that post mortem inspection of residual impact impressions in polymers often underestimates the actual deformation volume because it does not account for the timedependent properties of the base material, for example viscoelastic recovery. [27] Figure S5 (Supporting Information) demonstrates a high specific delo calization energy observed for the nanoarchitected SU8, which accounts for additional dissipation mechanisms, such as intrinsic viscoelastic dissipation [27,35,37] and adiabatic heating of polymer, [31] in addition to energy transfer within the deformed material. Quantifying the contributions of each of these factors is critical to laying out the parameter space for the energy dis sipation capabilities in nanoarchitected materials and will be explored in future studies.

Conclusion
We demonstrate that holographic lithography enabled by optical metasurfaces and laser scanning can serve as a robust scalable fabrication platform for nanoarchitected materials. Using our developed technique to pattern SU8 negativetone photoresist, we fabricated uniform sheets with lateral dimen sions of 2.1 × 2.4 cm 2 , thicknesses of 30-40 µm, individual unit cell dimensions of 0.9 µm, and relative densities between 20% and 50%. Quasistatic and laser induced particle impact experi ments reveal the elastic modulus to be between 300 MPa and 4 GPa, as expected from the scaling of modulus with relative density, and specific energy dissipation of 0.51-2.6 MJ kg −1 , comparable to other bulk and nanomaterials such as Kevlar/ PVB composite, [33] polystyrene, [31] and pyrolised carbon nanolat tice. [7] Such comparison of specific energy density with other ballistic materials corroborates superior energy dissipation capabilities offered by nanoarchitected sheets, enabled by multiple dissipation mechanisms available to hierarchical materials. This technique is capable of continuously exposing 2.1 cm × 2.4 cm tiles at a rate of 0.45 cm 2 min −1 . This rate can be significantly increased by optimizing fabrication parameters, for example using higher laser power, parallelizing exposure with multiple beams, and employing a more efficient photoacid generator. Furthermore, it should be pointed out that although the presented investigations were limited to polymers, it is fairly straightforward to transform the polymer template to a metal or functional oxide. For examples, as demonstrated in previous works the metals or oxides can be deposited by atomic layer deposition (ALD). The polymer could be transformed to amorphous carbon by pyrolysis. More advanced options could include swelling of the polymer network with metal salts and subsequent calcination and reduction in order to generate oxides and metals respectively. [53] The developed methodology solves a critical bottleneck in the field of nanoarchitected mate rials as it offers a scalable fabrication methodology for produce sheets with macroscopic lateral dimensions in several min utes and is expected to further our understanding and advance future applications of this emerging class of materials.

Experimental Section
Sample Preparation: Soda-lime glass microscope slides (1 mm thick and 2 by 2 inch) were utilized. The first step of substrate preparation was to produce an adhesion-promoting layer on the glass. After the glass slides were cleaned in acetone and isopropanol baths, 1 mL of SU8 2000.5 was spun at 2000 rpm for 30 s, followed by a 10 min pre-exposure bake at 85 °C, 10 min UV exposure, and a 20 min post-exposure bake at 85 °C, yielding a glass substrate optimized for photoresist adhesion. Prior to application of the SU8 layer for patterning, the photoresist must be sensitized to the operating wavelength. To that end, a visible light sensitizing system, HNu 470 and HNu 254, was added such that the ratios to SU8 solids were 0.53% and 4.66%, respectively, in a 70% solids resist, all by mass. Sensitized SU8 was spun onto substrates with adhesion layers at 3000 rpm for 30 s to produce 40 µm uniform coatings. The sample was then baked at 65 °C for 10 min, then at 95 °C for 30 min.
Exposure Procedure: Glass slides prepared with solid SU8 films were placed on an acrylic block coated in Acktar black film and an index matching mineral oil to minimize internal reflection at interfaces. The acrylic block itself was affixed to a Physik Instrumente XY-translation stage. Index matching mineral oil was then deposited directly onto the SU8, and the metamask was placed, feature-side down, on the sample. Gentle pressure was applied to ensure that the sample and metamask planes were parallel. The laser was then set to exposure power (230-270 mW) and allowed to stabilise for 75 s while blocked by an external shutter. The shutter was then opened, and the translation stage raster scanned the sample at 1.5 mm s −1 relative to the stationary beam, displacing each line scanned by 0.4 mm.
Post-Exposure Sample Processing: After exposure, samples were baked at 65 °C for 3.5 min, at which point the hotplate was turned off and the sample rested as the hotplate cools for another 10 min. The sample was then placed suspended and inverted in a bath of propylene glycol monomethyl ether acetate (PGMEA) for 2 h for development. The samples were then washed in a series of baths to perform a solvent exchange from PGMEA to isopropanol (IPA). After ≈1 h in 100% IPA, samples were air dried.
Volume Fraction Estimation: Cross-sectional scanning electron images were utilized in order to estimates the volume fraction at various depths. The beam pillar widths and interlayer thicknesses where measured at the top, center and bottom of the cross-sectional image shown in Figure 2b. Under the assumption that each unit cell comprises in total four quarter beams and two halves of the interlayer plates, the volume fraction was estimated.
Quasi-Static Nanoindentation: Nanoindentation experiments were conducted with a G200 XP Nanoindenter (KLA). The nanoarchitected SU8 sheets was still attached to the glass substrate. Across the entire sheet, 25 by 25 arrays were indented with the spacing of 820 µm and 860 µm in x and y-direction, respectively. To probe the effective behavior of the architected material instead of individual unit-cell, a flat-ended cylindrical punch with 100 µm diameter was used for all the measurements. The samples were indented to displacement up to 12 µm with the displacement rate of 20 nm s −1 , which corresponds to the strain rate of  ε = 10 −4 s −1 . To dampen the time-dependent properties of the material, the indenter was held for 100 s at the peak load prior to unloading. Continuous stiffness measurement (CSM) technique, which imposed 2 nm displacement oscillation amplitude at a frequency of 45 Hz, was utilized to detect an initial contact between the indenter tip www.advmat.de www.advancedsciencenews.com Adv. Mater. 2023, 35,2209153 and the sample and quasi-continuously measure contact stiffness as a function of depth. For all experiments, the effective elastic modulus was extracted by analyzing the continuous stiffness at the depth of 10% of the sheet thickness (≈3 µm after the full contact) following the Oliver-Pharr method. [20] LIPIT: The laser induced microprojectile impact testing (LIPIT) was utilized to elucidate the high-velocity impact-energy dissipation in the nanoarchitected SU8 film. The experiment was conducted following the recent study of nanoarchitected pyrolytic carbon lattice. [7] SiO 2 microparticles with a diameter of 7.4 and 13.96 µm were selectively chosen and accelerated from a launching pad toward the sample. The launching pad consisted of a 210 µm-thick glass substrate (Corning No. 2 microscope cover slip), a 60 nm-thick sacrificial gold layer (sputter-coated), and a 30 µm-thick polyurea layer (spin-coated) where microparticles were deposited on. To propel a selective particle from the launching pad, a single pulse (532 nm wavelength, 10 ns duration) was created by an Nd-YAG ablation laser (Spectra-Physics Quanta-Ray INDI-40-10-HG) and produced a local ablation on the gold film. The polyurea layer was then rapidly expanded by the ablation, which selectively accelerated an individual projectile. The excitation laser pulse energy was calibrated to control the projectile velocity from 50 to 10 3 m s −1 . A laser imaging pulse (Cavilux, 640 nm wavelength, 30 µs duration) was used to illuminate the moving microparticle, and the adjacent snapshots of its trajectory captured by an ultrahigh-speed camera (SIMX16, Specialised Imaging) were used to measure the impact and rebound velocities.
Post-Mortem Investigation of Impact Areas: To conduct a postmortem analysis on the impacted areas, the high-resolution SEM (FEI Quanta 200) and FIB milling (FEI Versa 3D DualBeam) were taken on a plane perpendicular to the axis of projectile direction. Craters were assumed to be radially symmetric. Their volumes were measured with the geometrical relation of a spherical cap for shallow craters and a cylinder with hemispherical-end for deeper craters or regions with embedded particles.
Metasurface Fabrication: The metamask was fabricated based on 300 nm c-Si on quartz wafer. A 300 nm-thick layer of e-beam positive resist ZEP 520A (Zeon Corporation) was spun coated on the chip. A 100 keV electron beam lithography system (Raith EBPG 5200) was used to generate the metasurface mask pattern. The pattern was developed in a developer (ZED-N50, Zeon Corporation). A 50 nm layer of Al 2 O 3 was deposited using an e-beam evaporator to reverse the generated pattern by lift-off process. Then, the transferred Al 2 O 3 pattern was used as a hard mask for SF 6 /C 4 F 8 dry etching of silicon (Oxford Plasmalab System 100). Next, the Al 2 O 3 was removed by 1:1 ammonium hydroxide and hydrogen peroxide at 80 °C. Finally, a 2 µm layer of SU8 was spun coated on top of the metamask for protection.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.