A comparative study of air cushion and roll-to-plate pressing in wafer-level nanoimprint lithography

Abstract

Nanoimprint lithography (NIL) has recently regained significant industrial momentum, driven by the burgeoning demand for high-resolution optical components, AR/VR devices, and advanced semiconductor fabrication. Among current wafer-level NIL strategies, Air Cushion Press (ACP) and Roll-to-Plate Press (R2P) represent two predominant mechanical approaches. This study provides a comprehensive comparative analysis of ACP and R2P to evaluate their efficacy in high-volume wafer manufacturing. Our findings demonstrate that the vacuum-assisted environment in ACP effectively mitigates air entrapment, enabling defect-free imprinting with a rapid cycle time of approximately one minute. Conversely, the R2P process, typically operated under atmospheric conditions, necessitates a protracted bonding phase—often lasting tens of minutes—to evacuate trapped air between the mold and the substrate. Furthermore, ACP achieves superior uniformity in both imprinting pressure and residual layer thickness (RLT), which is critical for maintaining pattern fidelity during subsequent etching processes. While the flexible molds used in both methods are prone to gradual deformation and feature distortion over repeated cycles, the extent of such degradation is significantly lower in ACP than in R2P. While R2P offers a cost-effective solution for large-area patterning on flexible substrates where post-etching is not required, ACP demonstrates clear technical superiority in throughput, uniformity, and precision for high-end wafer-level NIL applications.

Introduction

Different from nano-embossing, nanoimprint lithography (NIL) integrates the principles of imprinting and lithography, enabling the formation of nanopatterns by mechanically pressing a mold into a resist layer. These imprinted patterns can subsequently serve as etching masks, allowing pattern transfer to the underlying substrate through dry etching. Since its invention in 1995 by Chou et al. [1–3], NIL has been recognized as a revolutionary paradigm and a promising alternative to conventional lithographic techniques [4, 5]. Nevertheless, its contact-based nature has long been viewed as a limitation for semiconductor foundries, which traditionally favor the non-contact exposure methods of photolithography. Compared with photolithography, the molding process of NIL offers distinct advantages, including ultrahigh resolution, excellent pattern fidelity and repeatability, as well as lower energy consumption and reduced manufacturing cost. However, it also poses inherent challenges in defect control and high-precision alignment during large-scale manufacturing. Initially developed for semiconductor applications, NIL has lagged behind the rapid advancements of photolithography over the past decades. A major milestone occurred in 2023 when Canon announced its plan to begin mass production of 3D NAND chips using NIL by 2025 [6–8], marking the first formal adoption of nanoimprint technology in high-volume semiconductor manufacturing. Beyond semiconductors, NIL has seen rapidly expanding applications in diverse fields such as optics [9, 10], photonics [11], and biotechnology [12, 13], owing to its unique ability to directly fabricate complex three-dimensional (3D) micro- and nanostructures in functional materials—including optical and biological materials—through a single imprinting step [14, 15]. These imprinted structures can function directly as optical or bio-integrated components without additional etching steps. Furthermore, when the residual layer is removed by dry etching, the resulting nanostructures are fully compatible with standard semiconductor fabrication processes, comparable to those produced by photolithography. Importantly, NIL eliminates the wet development step required in photolithography, thereby simplifying the overall process and improving environmental sustainability [16, 17].

As its name suggests, NIL involves applying external pressure to transfer nanoscale features from a mold into an imprint resist coated on a substrate. In the commonly used UV-curable NIL, capillary forces alone are sufficient to fill the nanoscale relief structures of the mold once it comes into contact with the liquid resist, while external pressure is primarily applied to ensure conformal contact between the mold and the substrate. Early implementations of NIL employed the mechanical plate-to-plate (P2P) pressing method developed by Chou et al. [18, 19], who later introduced the roll-to-plate (R2P) and air cushion press (ACP) techniques. In the P2P configuration, nanostructures are replicated by pressing the mold and substrate together between two rigid plates. Although this method enables high imprinting pressure and is relatively easy to integrate with precise alignment modules, it inherently carries a high risk of damaging rigid molds and substrates due to direct mechanical contact. Electrostatic-assisted imprinting has also been explored as an alternative to mechanical force for mold pressing [20]. Canon’s nanoimprint lithography system can be regarded as a P2P-based approach derived from the step-and-flash imprint lithography (SFIL) developed by Wilson et al. [21]. This technology integrates the advantages of nanoimprint lithography with the step-and-repeat exposure concept of photolithography, making it particularly suitable for semiconductor manufacturing that requires die-by-die processing. Using this method, a single-die master fabricated by electron beam lithography can be repeatedly replicated to produce wafer-scale nanostructures, which can then serve as reusable wafer-level molds.

Beyond semiconductor device fabrication, wafer-level processes are crucial for achieving low-cost, high-throughput manufacturing. Among various approaches, the air cushion press (ACP) and roll-to-plate press (R2P) methods are two representative techniques widely employed for wafer-scale nanoimprinting. The R2P process can be implemented using either a roller mold [22] or a plate mold [23] configuration. In the roller mold approach, a flexible mold is mounted on a rotating cylindrical roller that sequentially contacts the substrate during motion, thereby transferring the imprint patterns. However, the line contact between the curved mold and the flat substrate is inherently non-conformal, often leading to non-uniform residual layer thickness (RLT) and limiting its applicability to etching-based pattern transfer [24]. In contrast, the plate mold configuration employs a roller to apply pressure onto a flat mold and substrate, enabling improved contact uniformity across the imprint area. The ACP method, which evolved from the plate-to-plate NIL, replaces rigid pressing plates with compressed gas to achieve uniform contact. The use of pressurized gas ensures an even pressure distribution across the entire wafer surface, which is essential for maintaining uniform RLT and high pattern fidelity in wafer-level nanoimprinting.

Although both the air cushion press (ACP) and roll-to-plate press (R2P) techniques have been successfully applied in wafer-level nanoimprinting, their underlying mechanisms, operating conditions, and performance characteristics differ substantially. The choice between these two methods critically affects imprinting uniformity, pattern fidelity, and process throughput—key factors determining their suitability for industrial-scale manufacturing. To elucidate these differences, this study systematically compares ACP and R2P under controlled wafer-level imprinting conditions through both simulation and experimentation. The performance of the two methods is evaluated in terms of pressure uniformity, residual layer thickness (RLT) uniformity, pattern fidelity, throughput, and air-trapped defect formation. The results provide fundamental insights into how their distinct pressure transmission modes and contact mechanisms influence imprint quality and long-term pattern stability, offering practical guidance for selecting appropriate wafer-level NIL methods and advancing the development of next-generation nanoimprint equipment.

Results and discussion

Imprinting process

In this paper, imprint experiments are conducted using R2P and ACP equipment from Prinano Co., Ltd. (See the Methods section). Figure 1(a) and (b) schematically illustrate the imprinting processes of the R2P and ACP methods, respectively. For comparison, both methods utilize disposable polymer molds as working molds. They are replicated from the same master silicon mold via reverse imprinting of a low-surface-energy, perfluorinated UV-curable resin onto PET sheets, using the R2P and ACP processes, respectively. Working molds are mounted onto custom-designed fixtures (as Fig. 1(c) and (d) shown), which are subsequently installed in the R2P and ACP nanoimprint machines (as Fig. 1(e) and (f) shown) as required. The wafer, spin-coated with the imprinted resist, is mounted on the wafer stage of the imprint machine. In R2P, the PET mold is secured within the frame by clamping assemblies integrated with a parallel spring array (Fig. 1(c)), which dynamically balances tensile forces during imprinting to prevent localized overstretching and maintain uniform stress distribution and then the frame is installed in the imprint machine (Fig. 1(e)). The rigid roller is typically wrapped with a flexible material, such as silicone or rubber, allowing it to accommodate slight variations in the substrate surface and maintain uniform line-contact pressure. When the roller descends and applies controlled force onto the PET mold, it induces conformal contact between the wafer and the PET mold. During this operation, the tension-balancing springs elongate to absorb mechanical stress, thereby preventing plastic deformation or overstretching of the PET material. The roller maintains linear motion across until the PET mold achieves full conformal contact with the wafer surface. UV exposure is used to solidify the imprinted resist sandwiched between the PET mold and the wafer. For demolding, the roller executes a reversed trajectory while the trailing edge of the frame is progressively lifted, causing the PET mold peeling from the wafer, starting until complete separation occurs. The system then resets with the roller returning to its initial position. The cured resist with the imprinted patterns is transferred onto the wafer.

Fig. 1.

a and b schematics of R2P and ACP process; c and d working mold frames of R2P and ACP; e R2P and f ACP Install the frames in the nanoimprint lithography machine

In the ACP process, the PET working mold, affixed to the frame in a relaxed state (Fig. 1(d)), is mounted onto the imprint equipment (Fig. 1(f)), serving as a flexible separation membrane that divides the imprint chamber into upper and lower compartments. Due to gravity, it sags slightly, forming a profile where the center is lower than the edges. The wafer is secured onto the wafer stage in the lower chamber. After both chambers are evacuated to achieve vacuum conditions, regulated compressed air is systematically introduced into the upper chamber. This pressure differential drives the PET working mold into progressive conformal contact with the mold surface, initiating at the geometric center and propagating radially toward the edges. During this continuous conforming process, the sample stage ascends dynamically based on real-time feedback, compensating for potential deformation of the PET mold caused by unsupported pressure differentials. Upon completion of the conforming process and UV-exposure, the upper and lower chambers are vented to atmospheric pressure, and the sample stage executes a programmed descent, enabling automatic separation of the substrate and working mold from the edge toward the center.

Air-trapped defects and throughput

During the imprinting process, air can become trapped between the mold and substrate, leading to the formation of bubbles[25], or voids[26], in the resist layer. The selection of the imprinting method inherently dictates the contact approach, which directly affects the efficiency of air bubble removal.

A 6-inch silicon wafer featuring a hexagonal microcolumn array (period: 2 μm; height: 2.5 μm; top/bottom diameter: 1.5/1.8 μm), specifically designed for patterned sapphire substrate (PSS) fabrication, is employed as the master mold. The R2P imprinting process is carried out under atmospheric conditions. During imprinting, the mold and the wafer gradually come into contact from one side to the other, with the imprinting roller applying force along the movement direction without lingering on the mold surface after contact is established. As shown in Fig. 2 (a), this progressive contact method effectively expels most of the air between the mold and the wafer. A small amount of air will inevitably remain trapped between the PET sheet and the mold. If the trapped air volume is small enough, it can completely dissolve into the imprint resist [25]. However, if the volume is larger, the air cannot fully dissolve, leading to the formation of bubbles or voids. Figure 2(b) shows a top-view microscope image of the prepared PET mold. The light regions indicate air bubbles trapped between the PET sheet and the perfluorinated resin film. Furthermore, during the subsequent imprint process of the PET mold, the trapped bubbles tend to break, as illustrated in Fig. 2(c). Upon demolding, the presence of trapped air bubbles leads to the formation of voids and significant defects on the imprinted surface (Fig. 2(d)).

Fig. 2.

a Schematic illustration of air expulsion and entrapment during R2P imprinting; b Microscope image of air bubbles trapped between the PET sheet and the perfluorinated resin layer of the working mold fabricated via R2P imprinting; c Microscope image of a ruptured air bubble on the working mold following subsequent R2P imprinting; d SEM image of an air bubble defect on the imprinted PSS microstructures; (e-i)-(e-iv) Correlation between imprinting speed and the size of trapped air bubbles; f Correlation between imprinting speed and the total area of trapped air bubbles on a 6-inch wafer

The imprinting speed is a key factor affecting the formation of bubble defects. As observed in Fig. 2(e) and 2(f), both the size of the trapped bubbles and the total area of air bubbles decrease with the reduction in the roller’s translational speed. The faster the roller moves, the larger the size and total area of the bubbles. At a roller speed of 20 mm/s, individual bubble defects can reach sizes up to 19.3 × 10^− 3 mm², with the total bubble coverage on a 6-inch silicon wafer approaching 38 mm².This occurs because the faster movement of the roller reduces the time available for air to escape. Additionally, the increased speed may induce turbulence or pressure variations, further aggravating air entrapment. As the roller speed is reduced to 10 mm/s, large bubbles begin to break up into multiple smaller bubbles, which tend to elongate in the direction of roller movement. For an 6-inch wafer, the optimal imprint speed for PSS structures is found to be below 0.2 mm/s. When the speed is below this threshold, trapped air bubbles are hardly found (Fig. 2(e-iv)). However, this slow speed results in a time-consuming process, taking approximately 750 seconds for a single 6” wafer, making it impractical for large-scale production.

A vacuum imprinting environment is an effective way to eliminate trapped air bubbles. However, as shown in Fig. 1(e), the bulky mechanical structure of the R2P imprinting system demands a large-capacity vacuum chamber, significantly increasing both the equipment cost and the vacuum evacuation time. In contrast, the entire ACP imprinting process is conducted within a small sealed chamber (Fig. 1 (f)), simplifying the establishment of vacuum conditions. To improve efficiency, minimizing the chamber volume allows for the rapid achievement of high vacuum conditions within seconds. Both chambers are simultaneously vacuum‑pumped, while the absolute pressure of the lower chamber is kept marginally higher (slightly poorer vacuum) than the upper chamber. This pressure setting avoids premature downward sagging and pre-contact of the flexible PET membrane toward the wafer stage before formal pressurization. When compressed air is subsequently injected into the upper chamber, the enlarged net pressure difference reliably drives the PET mold to make conformal contact starting from the center and propagating radially outward across the wafer surface. This setup achieves a high‑vacuum imprinting environment.For a 6-inch wafer imprint, full contact between the mold and substrate is established approximately 10 s after applying gas pressure, with demolding completed in just 10 s. Given that the exposure time is also on the order of several seconds, the entire imprint cycle takes less than one minute. No trapped air is observed, both macroscopically and microscopically, during imprinting at a vacuum of -98 kPa.

Uniformity of imprint pressure and residual layer thickness

To investigate the pressure uniformity of ACP and R2P, a pressure-sensitive papers (FUJI Film, 3LW) with a pressure measurement range from 0.2 to 0.6 MPa and a total paper thickness of 200 μm is placed between working mold and a silicon wafer. The pressure-sensitive paper consists of a layer of micro-encapsulated color-forming material. When pressure is applied to the paper, it ruptures the microcapsules. Consequently, the density of ruptured microcapsules is dictated by the applied pressure, leading to difference in color intensity across the paper. Increased pressure results in a higher density of ruptured microcapsules, intensifying the color appearance.

Figure 3(a) presents the imprint pressure distribution on pressure-sensitive paper in the R2P method under an applied load of 250 N. Significant non-uniformity is observed, with local pressures ranging from 0.3 MPa in the lightest regions to 1.2 MPa in the darkest regions, corresponding to a fourfold variation across the nanoimprint field. Uneven pressure points (different shades of color) can also be observed at the same lateral region. The pressure non-uniformity arises from surface roughness of the wafer, dust particles, or imperfections on the substrate, working mold, and roll surfaces, leading to localized high-pressure zones. Additionally, the pressure sharply increases at the start and end positions of imprinting (the upper area and lower area of the wafer surface). The primary reason is that the imprinting force originates from the constant weight of the roller throughout the entire imprinting process. At the beginning and end of the process, the roller contacts the circular wafer over a smaller area, leading to a significantly higher pressure per unit area compared to the central region. This results in a non-uniform pressure distribution across the wafer. To ensure uniform pressure during the R2P process, real-time adjustment of the imprinting force is required. It is observed that when the UV-curable imprint resist is thinned to several hundred nanometers, premature detachment between the mold and substrate can occasionally occur after the roller passes but before UV curing is complete. This behavior is attributed to the reduced adhesive force of the thin resist layer, which becomes insufficient to sustain conformal contact once the roller has moved on. As the resist thickness decreases, both capillary and van der Waals forces are significantly weakened, thereby lowering the interfacial bonding strength. Without adequate adhesion, mechanical disturbances—such as stress relaxation induced by roller movement—can readily trigger separation.

Fig. 3.

Photographs of pressure-sensitive paper (covering a 6-inch wafer area) imprinted by a R2P and b ACP.

In the ACP process, both the upper and lower chambers are first evacuated, after which compressed air is introduced into the upper imprint chamber until the preset pressure is reached. The resulting pressure drives the PET mold into conformal contact with the substrate. Owing to the isotropic nature of the gas pressure, uniform force is distributed across the entire surface of the PET mold. At a nominal pressure of 0.3 MPa, the pressure-sensitive paper exhibits a uniform color across the wafer, confirming consistent pressure distribution over the entire imprinting area (Fig. 3b). Additionally, the ACP imprinting chamber is smaller, allowing for faster pressure application. Unlike the R2P process, the ACP process maintains uniform pressure throughout the imprinting cycle until demolding, thereby preventing the premature separation between the mold and substrate observed in R2P.

Compared to photolithography, a distinctive characteristic of NIL is the presence of a resist layer between the imprinted features and the substrate, known as the residual layer [27]. This residual layer is crucial in NIL to ensure the complete filling of all cavities on the mold and to protect the mold from direct contact with the substrate. A thin and uniform residual layer is critical for subsequent pattern transfer during the dry etching process. The two key factors influencing the residual layer are the spin-coated resist thickness and the uniformity of the imprinting pressure. Under the same resist thickness and with working molds of identical parameters, we compared the influences of ACP and R2P on the uniformity of the residual layer. Based on the mold parameters of PSS, UV-curable resist with a uniform thickness of 1.5 μm is spin-coated onto the silicon wafers, not only to ensure complete filling of the mold cavities but also to facilitate observation of variations in the residual layer thickness. as shown in Figs. 4(a), Residual layer thickness variations on 6-inch wafers imprinted by R2P and ACP are examined by acquiring cross-sectional SEM images at five locations: one at the wafer center and four at the wafer edge, as shown in Figs. 4(b) and 4(c), respectively. The residual layer is artificially colored to make it easier to distinguish in the SEM images. As illustrated in Fig. 4c, for R2P, the residual layer thickness increases significantly from 114 nm at the beginning of the imprint to 492 nm at the end, representing a variation of over 300%. This is because, at the contact line between the mold and the substrate, a portion of the liquid resist is driven by the lateral movement of the roller, resulting in an increase in both the resist film and residual layer thickness. Substantial variations in residual layer thickness across the wafer surface present challenges for subsequent etching processes. As for ACP, the residual layer thickness at the five locations is consistently close to 210 nm, with a variation of less than 5% (Fig. 4b). There are two reasons for the uniform residual layer thickness of ACP. Firstly, the imprint resist primarily experiences downward air pressure transferred from the mold, without any lateral forces involved. Secondly, due to the uniform distribution of air pressure, the resist is subjected to consistent pressure across all areas. This consistency prevents the resist from moving from higher-pressure regions to lower-pressure regions, thereby avoiding fluctuations in both the resist and residual layer thickness. This facilitates the subsequent processes following imprinting nanostructures.

Fig. 4.

a Residual layer thickness distribution at five measurement points on 6-inch wafers imprinted by ACP and R2P; b Five SEM observation positions for ACP imprinting; c Five SEM observation positions for R2P imprinting

Pattern fidelity

The fidelity of the imprinted patterns and their repeatability over multiple imprinting cycles are key factors in evaluating the reliability of the nanoimprint process. Currently, PET sheets are widely used as substrates for imprint working molds in industrial applications due to their flexibility, low cost, and ease of processing [28]. Working molds are most commonly fabricated by imprinting a perfluorinated UV-curable resin onto PET sheets. While PET has a higher elastic modulus than PDMS [29], it is still much lower than that of rigid substrates such as quartz. This allows PET working molds to achieve good conformity with the substrate during imprinting, but also introduces a potential risk of deformation.

As shown in Fig. 5a, reference marks with a spacing of 40 mm were replicated onto both the PET working mold and the silicon substrate. In the R2P process, alignment of the marks at one end (Fig. 5b) revealed a one-dimensional misalignment of approximately 5 μm at the opposite end (Fig. 5c). This offset corresponds to a dimensional deviation of only about 125 ppm and is attributed to slight deformation of the PET substrate during mold replication and imprinting. Although no measurable change in the nanostructure periodicity was detected (Fig. 6), the cumulative dimensional variation of the PET mold over a relatively large distance resulted in a micrometer-scale positional error between the corresponding alignment marks. These results suggest that PET working molds are well suited for single-layer nanoimprint patterning, but their dimensional stability may be insufficient for applications requiring sub-micrometer alignment accuracy or multilayer overlay registration over large areas.

Fig. 5.

Evaluation of the dimensional fidelity of the PET working mold in the R2P process. a Schematic illustration of the reference marks replicated onto the PET working mold and the silicon substrate with a spacing of 40 mm. b Optical microscope image showing alignment of the corresponding reference marks at one end. c Optical microscope image of the corresponding reference marks at the opposite end. d Magnified view of the aligned reference marks at c, showing a one-dimensional misalignment of approximately 5 μm

Fig. 6.

a, b SEM images of the imprinted microcolumn and grating structures after the 1st, 10th, and 20th imprint cycles using the R2P and ACP processes, respectively. Schematic illustrations of the corresponding demolding mechanisms are also included

Consistent with the alignment-mark measurements, SEM characterization (Fig. 6) revealed that the pattern periodicity remained essentially unchanged throughout repeated imprinting cycles for both the R2P and ACP processes, indicating negligible dimensional change of the replicated mold at the local scale. In contrast, progressive changes in feature geometry were observed with increasing imprint cycles. Specifically, the replicated structures exhibited a gradual increase in aspect ratio, manifested by a reduction in linewidth accompanied by an increase in feature height.

For the R2P process, the grating linewidth decreased from 217 nm to 188 nm (− 13.4%), while the feature height increased from 454 nm to 475 nm (+ 4.6%) after 20 imprint cycles. A more pronounced deformation was observed for the microcolumn structures, where the column diameter decreased from 1.75 μm to 1.33 μm (− 24.0%) and the height increased from 2.42 μm to 2.88 μm (+ 19.0%). These results indicate substantial elongation and thinning of the mold features during repeated use. In contrast, ACP exhibited significantly improved structural stability. For the grating structures, the linewidth decreased only slightly from 218 nm to 213 nm (− 2.3%), while the height increased from 455 nm to 461 nm (+ 1.3%) after 20 imprint cycles. Similarly, for the microcolumn structures, the diameter decreased from 1.75 μm to 1.67 μm (− 4.6%), whereas the height increased from 2.39 μm to 2.62 μm (+ 9.6%). Compared with R2P, ACP reduced the dimensional evolution of both nano- and microscale features, demonstrating substantially improved mold durability.

The observed structural evolution is attributed primarily to mechanical deformation of the mold features during repeated demolding. Because the cured mold material possesses a relatively low Young’s modulus (~ 60 MPa), localized tensile and shear stresses generated at the mold–resist interface can gradually stretch the features, leading to feature narrowing and elongation over successive imprint cycles. Notably, the magnitude of deformation strongly depends on the demolding mechanism. In the R2P process, demolding occurs progressively as the roller retracts while simultaneously lifting one side of the frame supporting the PET mold. For an 8-inch substrate, the lifted side is elevated by approximately 50 mm relative to the opposite side at the end of demolding, resulting in a large separation angle between the mold and substrate. This geometry enhances mechanical interlocking between the mold features and the replicated structures and increases the tensile stress imposed on the mold features during release. By contrast, ACP employs a fundamentally different demolding mechanism. Separation is initiated at the outer perimeter of the mold–substrate interface, after which compressed gas gradually penetrates between the mold and substrate, supporting the mold and promoting uniform release across the entire area. Throughout the demolding process, the mold and substrate remain nearly parallel, with only a few millimeters of separation at complete release. This near-parallel demolding configuration minimizes tensile and shear stresses at the mold–resist interface, thereby significantly suppressing feature stretching and deformation. As a result, ACP provides markedly improved mold lifetime and pattern fidelity compared with the conventional R2P process.

The key performance metrics of ACP and R2P wafer-level nanoimprint lithography are summarized in Table 1 for straightforward comparison.

Table 1.

Comparison of ACP and R2P wafer-level nanoimprint lithography

Performance metric	Air Cushion Press (ACP)	Roll-to-Plate Press (R2P)
Operating environment	Small sealed chamber with vacuum-assisted imprinting	Imprinting under atmospheric ambient conditions
Air-entrapment defects	Effectively suppressed under − 98 kPa vacuum; no observable bubble or void defects	Inherent air trapping; bubble defect size and total area increase with roller speed; bubble-induced void defects form on imprinted patterns
Cycle time (6-inch wafer)	Whole imprint cycle < 1 min, with ~ 10 s for mold-substrate contact and ~ 10 s for demolding, in addition to UV exposure.	~ 750 s for bubble‑free imprint at roller speed below 0.2 mm/s; much longer cycle under bubble‑suppression conditions.
Residual layer thickness (RLT)	RLT ~ 210 nm at five measured positions; variation < 5%, no lateral resist flow driven	RLT varies from 114 nm to 492 nm across wafer; variation > 300%, resist accumulates by roller lateral movement
Demolding mechanism	Mold-substrate separate nearly in parallel, separation initiates from edge toward center; low tensile/shear stress on mold features	Large-angle progressive peeling by roller retraction and frame lifting; high local tensile and shear stress applied to mold structures
Microcolumn mold degradation after 20 cycles	Diameter − 4.6%, height + 9.6%	Diameter − 24.0%, height + 19.0%
Grating mold degradation after 20 cycles	Linewidth − 2.3%,height + 1.3%	Linewidth − 13.4%, height + 4.6%
Representative application scenarios	High-end wafer-level NIL requiring high throughput, uniform residual layer and high pattern fidelity; semiconductor, AR/VR optics, photonic devices	Cost-effective large-area patterning on flexible substrates; large RLT variation may reduce post-etch uniformity

To verify the scalability of the ACP process, a PET working mold replicated from a 12-inch silicon master was used imprint a matching 12-inch glass substrate. As illustrated in Fig. 7, a highly uniform pattern transfer was achieved across the entire substrate without perceptible defects, underscoring the potential of ACP for large-area, high-throughput nanoimprint manufacturing.

Fig. 7.

Large-area nanoimprint results using the ACP process. Photographs of the a 12-inch silicon master mold, b replicated PET working mold, and c 12-inch imprinted glass substrate

Conclusion

In summary, this work provides a systematic comparison of Air Cushion Press (ACP) and Roll-to-Plate Press (R2P) technologies for wafer-level nanoimprint lithography. The results demonstrate that ACP offers significant advantages in imprinting efficiency, pressure uniformity, residual layer thickness (RLT) control, and pattern fidelity. The vacuum-assisted imprinting mechanism effectively eliminates air entrapment, enabling defect-free pattern transfer with a cycle time of approximately one minute, whereas the atmospheric R2P process requires substantially longer bonding times to remove trapped air. ACP also exhibits superior imprint uniformity and reduced mold deformation during repeated imprint cycles, leading to improved pattern durability and process stability. These findings provide practical guidance for selecting appropriate NIL strategies in emerging applications such as advanced semiconductor manufacturing, AR/VR optics, and high-performance photonic devices.

Methods

Acrylate-based UV-curable nanoimprint resist (PL-R-PC1000), perfluorinated UV-curable working mold resin (PL-R-UPM), Air Cushion Press (ACP, PL-A) and Roll-to-Plate Press (R2P, PL-R) nanoimprint lithography systems were all obtained from Prinano Technology Co., Ltd., Hangzhou, China (https://www.prinano.com/). Identical resist preparation and UV curing parameters were applied for both ACP and R2P throughout all comparative experiments to guarantee fair comparison between the two imprint approaches; LLLW thin films provided by FUJIFILM Corporation (Japan) were utilized as auxiliary imprinting materials, and the imprinting pressure was set within the range of 0.2 MPa to 0.6 MPa during all experimental tests. All other reagents were purchased from commercial suppliers and used as received without further purification unless otherwise specified. All of the SEM images were obtained using a field-emission scanning electron microscope (ZEISS, ULTRA-55). A KEYENCE VHX-7000 digital microscope was used to measure the alignment mark error of patterned samples.

Biographies

Guiyan Yang

Department of Materials Science and Engineering, College of Engineering and Applied Sciences, Nanjing University, Nanjing 210046, P. R. China; National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093, P. R. China; Engineering Research Center of Photoresist Materials, Ministry of Education, School of Chemistry&Chemical Engineering, Nanjing University, 210046 Nanjing, P. R. China.

Si Chen

Department of Materials Science and Engineering, College of Engineering and Applied Sciences, Nanjing University, Nanjing 210046, P. R. China; National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093, P. R. China; Engineering Research Center of Photoresist Materials, Ministry of Education, School of Chemistry&Chemical Engineering, Nanjing University, 210046 Nanjing, P. R. China.

Bo Xu

Department of Materials Science and Engineering, College of Engineering and Applied Sciences, Nanjing University, Nanjing 210046, P. R. China; National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093, P. R. China; Engineering Research Center of Photoresist Materials, Ministry of Education, School of Chemistry&Chemical Engineering, Nanjing University, 210046 Nanjing, P. R. China.

Mengmeng Deng

The Prinano Technology Co., Ltd. (Hangzhou, China)

Xinyu Wang

Department of Materials Science and Engineering, College of Engineering and Applied Sciences, Nanjing University, Nanjing 210046, P. R. China; National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093, P. R. China; Engineering Research Center of Photoresist Materials, Ministry of Education, School of Chemistry&Chemical Engineering, Nanjing University, 210046 Nanjing, P. R. China.

Xiaofeng Chen

The Prinano Technology Co., Ltd. (Hangzhou, China)

Zongbin Hao

Department of Materials Science and Engineering, College of Engineering and Applied Sciences, Nanjing University, Nanjing 210046, P. R. China; National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093, P. R. China; Engineering Research Center of Photoresist Materials, Ministry of Education, School of Chemistry&Chemical Engineering, Nanjing University, 210046 Nanjing, P. R. China.

Haixiong Ge

Department of Materials Science and Engineering, College of Engineering and Applied Sciences, Nanjing University, Nanjing 210046, P. R. China; National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093, P. R. China; Engineering Research Center of Photoresist Materials, Ministry of Education, School of Chemistry&Chemical Engineering, Nanjing University, 210046 Nanjing, P. R.

Author contributions

G.Y. Y and H.X.G. conceived and designed the research framework; S.C., B.X. and M.M.D. conducted all ACP and R2P nanoimprint lithography experiments and collected raw experimental data. X.Y.W., X.F.C. and Z.B.H. performed data analysis, mechanism discussion and prepared all figures. H.X.G. wrote the main manuscript text. All authors reviewed and approved the final version of the manuscript.

Funding

This research did not receive funding.

Data availability

The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval and accordance

Not applicable.

Consent to participate

Not applicable.

Consent to publish

Not applicable.

Competing interests

The authors declare no competing interests.