Nanoimprinting of pressure-intolerant tilted nanostructures assisted by an electric field for high performance optical coupler

Abstract

Tilted nanostructures, with their unique light-modulating capabilities, find broad applications in optical fields such as optical couplers, and escalating performance demands of devices impose formidable challenges regarding structural fidelity and fabrication cost-effectiveness, owing to their inherent pressure-intolerant properties. Here, a novel electric-field-assisted nanoimprinting technique is proposed. The electric-induced “interface-tension-driven” active resist filling of template microcavities is fundamentally different from the traditional nanoimprinting technique. This paradigm shift enables simultaneous mitigation of structural deformation induced by pressure and preclusion of residual layers, thereby resolving the intrinsic conflict between low-stress deformation control in tilted nanostructures and high-pressure residual layer elimination. Furthermore, a demolding path optimization strategy is proposed to effectively minimize stress-induced damage to tilted nanostructures during the demolding process. Based on the optimized filling and demolding processes, the developed nanoimprinting prototype enables real-time process parameter adjustment according to target geometries, resulting in high-fidelity, residual-layer-free fabrication of various tilted nanostructures. The integration of custom-fabricated tilted nanostructures into augmented reality (AR) systems results in remarkable performance enhancement, highlighting the substantial application potential of this technique in high-performance optical couplers and advanced nanomanufacturing.

Introduction

Tilted nanostructures, featuring unique asymmetric geometries, exhibit exceptional light-field modulation capabilities, enabling precise manipulation of the phase, amplitude, and polarization state of incident light^1–3. This modulation mechanism transcends the inherent physical limitations of conventional optical elements, demonstrating broad application potential in chiral optics⁴, optical sensing⁵, optical communications⁶, and optoelectronic chips^7,8. Notably, tilted nanogratings—leveraging their superior diffraction efficiency and wavefront manipulation properties—serve as the preferred technical pathway for diffractive optical couplers, significantly enhancing the performance of augmented reality (AR) display systems and high-resolution imaging systems^9–11. The fabrication of tilted nanostructures, which serves as a critical bridge linking structural design and functional realization, has garnered extensive attention. Current fabrication methods for tilted nanostructures, such as oblique etching¹², laser processing¹³, nanoinscribing¹⁴, and generative nanoimprinting¹⁵, are constrained by high costs or complex process control, making large-scale mass production challenging.

Undoubtedly, nanoimprinting technique enables the direct patterning of functional materials through template replication, providing a low-cost, high-throughput technical approach for the large-scale and mass production of tilted structures^16–18 (Fig. 1c). However, the inherent requirement for external pressure in conventional UV-nanoimprinting technique (c-N) of tilted nanostructures presents a fundamental constraint, forcing a compromise between achieving high structural fidelity and eliminating the residual layer, which consequently limits improvements in optical performance. Primarily, the stress-sensitive characteristics of tilted nanostructures render them prone to deformation under external nanoimprinting pressure, resulting in reduced structure fidelity^19,20. To mitigate deformation, current approaches typically rely on thicker residual layers for stress buffering^21,22. Nevertheless, residual layers significantly degrade device performance^23,24. The residual layers introduce unwanted substrate layer thickness fluctuations, which weaken optical field localization effects and cause a surge in transmission loss²⁵. Consequently, extensive research has been conducted on residual-layer-free nanoimprinting technologies. Current research focuses primarily on template structure design²⁶, material optimization²⁷, and process improvement²⁸ to achieve residual-layer-free nanoimprinting, as well as novel techniques such as reversal nanoimprinting²⁹, contrast exposure³⁰, and selective template surface treatment³¹. However, all of these methods have inherent limitations, and they tend to increase process complexity and cost while reducing processing efficiency. Moreover, these methods are primarily designed for vertical nanostructures and rely on mechanical force application, which is inappropriate for tilted structures. Nanoimprinting high-fidelity residual-layer-free tilted nanostructures confronts a paradox: the inherent stress sensitivity of these structures demands minimal stress in the absence of a residual buffer layer, while interfacial residual layer elimination necessitates substantial pressure to drive resist flow, which is an impossible task for traditional nanoimprinting techniques. Thus, developing a novel nanoimprinting technique that simultaneously ensures structure fidelity and residual-layer-free characteristics has become pivotal for overcoming the optical coupling performance bottleneck in nanomanufacturing.

Fig. 1. Schematics of tilted-nanograting optical coupler and e-N method.

a Schematic of a tilted-nanograting optical coupler, b influence of structural fidelity and residual-layer thickness on the diffraction efficiency, c limitations of c-N method, and d schematic of the core components and flow diagram of the e-N system

In this paper, an electric-field-assisted nanoimprinting (e-N) technique is proposed for fabricating high-fidelity residual-layer-free tilted nanostructures for an efficient optical coupler. An electric field applied between the template and substrate provides the driving force for both structure filling and template contacting the substrate. This approach allows self-organized resist filling of microcavities using electric-induced interfacial forces, ensuring complete resist filling through precise volume control. Most importantly, the interfacial-force-dominated filling mechanism eliminates volume-force-induced structural deformation, guaranteeing structure fidelity. Additionally, the electric-interfacial force acting directly on the resist enhances the effective disjoining pressure of the liquid film, making it more susceptible to instability. This leads to spinodal dewetting, thereby achieving residual-layer-free patterning. To counter the increased demolding stress caused by geometric interlocking between the template and the formed structures in high-angle-tilted residual-layer-free structures, a demolding optimization strategy is proposed to minimize stress concentration during demolding. Based on these processes, an e-N prototype was developed that allows for real-time adjustment of process parameters according to target structure geometries, resulting in the successful fabrication of a variety of high-fidelity residual-layer-free tilted nanostructures. The customized tilted optical coupling structures have been successfully integrated into AR glasses, demonstrating superior display performance. This e-N technique also demonstrates significant application potential in the fabrication of structures for optical metasurfaces and chiral optics.

Results

Tilted nanostructures have significant application potential in optical couplers due to their exceptional capability in modulating diffraction efficiency. As illustrated in Fig. 1a, when a vertically incident light beam interacts with these tilted structures, it excites multiple diffraction orders. The diffraction efficiency can be maximized in either -1st (or +1^st) order by precisely optimizing structural parameters (tilt angle, linewidth, etc.).^10,32. Following diffraction-based modulation by the tilted nanostructures, the light beam propagates efficiently via total internal reflection at predetermined angles within the waveguide substrate, thereby achieving highly efficient optical coupling.

The structural parameters play a critical role in determining the performance of the optical coupler; therefore, a simulation model was established to investigate the influence of tilted angle ($\theta$), line width ($w$), height of the nanostructures ($H$), and residual layer thickness ($h$) on diffraction efficiency (Fig. 1b and Figs. S1–2). The optimal parameters for a diffraction grating structure designed for 532 nm light are: a tilt angle of 40°, a linewidth of 200 nm, a structure height of 600 nm, and a period of 400 nm. Firstly, vertical pressure during nanoimprinting causes structural deformation, which increases the tilt angle and gradually reduces diffraction efficiency. Secondly, the residual layer can affect optical coupling performance. On the one hand, the non-uniformity of residual layers may induce unwanted diffraction orders, resulting in energy dispersion and stray light generation. On the other hand, the residual layer thickness has a direct impact on the efficiency of target diffraction orders. The results indicate that diffraction efficiency decreases periodically with increasing residual layer thickness, peaking when there is no residual layer. As shown in the Fig. 1b, the optimal diffraction efficiency can be achieved when maintaining the optimal tilt angle and realizing a residual layer-free structure.

Manufacturing technology serves as the critical bridge connecting the design of tilted grating coupling and its functional realization^33–35. There are two major challenges in fabricating high-fidelity residual-layer-free tilted nanostructures. First, achieving high-fidelity nanostructures necessitates overcoming the intrinsic stress sensitivity of tilted structures while ensuring sufficient resist filling into template cavities. Furthermore, to achieve residual-layer-free structures, dewetting must occur by overcoming the surface tension of the liquid film under precise control of resist volume, thereby enabling direct template-substrate contact. Second, the demolding process necessitates the use of optimized demolding pathways to mitigate elevated interfacial stresses that can lead to structural deformation, which is particularly evident in high-angle-tilted nanostructures³⁶.

For c-N technology, the external volume-pressure is inherently unavoidable. The stress sensitivity of tilted nanostructures makes it impossible to ensure structure fidelity during residual-layer-free nanoimprinting. Therefore, maintaining such fidelity requires the buffering effect of a thick residual layer, thus making it impossible to achieve both high-fidelity and residual-layer-free fabrication of tilted nanostructures (Fig. 1c). Additionally, c-N technologies employ fixed implementation modes (roll-to-roll, roll-to-plate, and plate-to-plate), making it impossible to dynamically adjust process parameters according to the nanoimprinted structures. As a result, residual-layer-free high-angle-tilted nanostructures are susceptible to deformation or even cracking due to stress during the demolding process. Consequently, constrained by its inherent reliance on externally applied pressure, c-N struggles to achieve both high-fidelity residual-layer-free tilted nanostructures, thus constraining the performance enhancement of optical couplers based on such structures.

As depicted in Fig. 1d, the electric-field-assisted nanoimprinting system addresses these challenges through electric field-driven mechanics. The electric field applied between the template and substrate generates the driving forces, while the integrated flexible template-roller-substrate assembly enables continuous nanoimprint processing. The figure depicts the nanoimprinting process, with defined orientations of the structures’ tilt direction, the nanoimprinting direction, and the demolding direction. Initially, the flexible template contacts the substrate at one edge. As the roller moves, the template progressively advances via line-contact propagation until full substrate coverage is achieved. To ensure high-fidelity structure formation during the nanoimprinting process, the resist fills the microcavities of the template driven by interfacial tensile forces from the electric field. Meanwhile, the electric-induced interfacial forces act directly on the liquid film by increasing its effective disjoining pressure, thereby inducing film instability and rupture, which triggers spinodal dewetting and ensures the formation of a residual-layer-free film during the nanoimprinting. After the template makes full contact with the substrate, UV curing is performed. Subsequently, the roller moves in the opposite direction for peel-off demolding. The demolding direction is aligned with the tilt direction of the formed nanostructures, minimizing stress concentration and preventing defects. This process ultimately yields high-fidelity, residual-layer-free tilted nanostructures.

During the e-N process, the electric field induces significant surface/interface effects in the resist. Firstly, this is manifested as a notable change in the contact angle³⁷. As shown in Fig. 2a, the contact angle of the resist with the template material surface decreases from 79° to 30° when an electric field is applied, increasing capillary filling within the template microcavities. The resist is a high-refractive-index composite system doped with TiO₂ nanoparticles. Additionally, due to the different dielectric constants of the resist and the template structure layer, dielectrophoretic forces act on the resist.

Fig. 2. Principles of electric-field-assisted high-fidelity formation for residual-layer-free tilted nanostructures.

a Wetting behavior of resist on template surface under the electric field, b resist filling driving force under an electric field, c composition of the filling driving force under different electric field intensities, d simulation of filling behavior under different driving forces, e structure fidelity of the residual-layer-free tilted structures fabrication via e-N and c-N methods, f electric-field-induced interfacial dewetting for residual-layer-free formation, and g effective disjoining pressure under varying electric field intensities

As shown in Fig. 2b, the driving force $F_d$ for resist filling in the template cavities comprises electrocapillary force $F_{ecf}$, dielectrophoretic force $F_{dep}$, and electrostatic attraction force $F_{ea}$. The electrowetting effect enhances capillary force, resulting in $F_{ecf}$, which is essentially a Laplace pressure:

$$F_{ecf}=\frac{2\gamma \cos {\theta }_E}{d}$$ 1

Where $\gamma$ is the surface tension coefficient of the resist, ${\theta }_E$ is the contact angle of the resist under an electric field³⁸, and $d$ is the linewidth of the structure.

$F_{dep}$ induced by dielectric constant differences is essentially the Maxwell stress:

$$F_{dep}=\frac{1}{2}{\epsilon }_0{\epsilon }_{eff}\left({\epsilon }_{eff}-1\right)E^2$$ 2

Where ${\epsilon }_0$ is the dielectric constant of vacuum, ${\epsilon }_0=8.85\times {10}^{-12}F/m$, ${\epsilon }_{eff}$ represents the effective dielectric constant of the resist, which is determined by the dielectric properties of both the polymer matrix and the TiO₂ nanoparticles, and is taken as 41. The detailed analysis is provided in Note S1.$E$ is the electric field intensity.

$F_{ea}$ is an interaction between electrodes:³⁹

$$F_{ea}=\frac{1}{2}{\epsilon }_0{\epsilon }_r^2{E}^2$$ 3

Where ${\epsilon }_r$ denotes the average dielectric constant between the two electrodes. The template structure layer thickness is 5 μm ($h_1$), with its relative dielectric constant ${\epsilon }_1=4$; the resist thickness is 230 nm ($h_2$), with its effective dielectric constant ${\epsilon }_{eff}=41$. The average dielectric constant of the two-layered material is thus given as ${\epsilon }_r=\frac{\left(h_1+h_2\right)\cdot {\epsilon }_1\cdot {\epsilon }_{eff}}{h_1\cdot {\epsilon }_{eff}+h_2\cdot {\epsilon }_1}$.

$F_{ecf}$ and $F_{dep}$ act as interface forces on the resist surface, whereas $F_{ea}$ acts as a volume force across the entire inter-electrode region. For nanostructures with a 40° tilt angle, 200 nm linewidth, and $l:d=3$, theoretical analysis based on Henry’s law of gas dissolution predicts a critical driving force threshold of 0.749 MPa for residual-layer-free filling (Note. S2)⁴⁰. Figure 2c illustrates the total filling driving force during the fabrication of this structure using the e-N method, under varying electric field intensities, along with the corresponding distributions of interface driving force and volume driving force. Among these, the interface driving force is dominant, with the volume driving force making only a minor contribution. Under the condition $E=8\times {10}^{6}V/m$, the total driving force meets the filling requirement with interface forces ($F_{ecf}+F_{dep}$) contributing 98.35% and volume forces ($F_{ea}$) contributing only 1.65%.

A phase field simulation model, as shown in Fig. 2d-i, was developed to quantify the effect of each driving force during resist filling. Figure 2d-ii to 2d-iv present the filling states under different driving forces at a time variable of 1: the resist surface profile under total driving force $F_d$ (Fig. 2d-ii), filling behavior driven by interface forces (Fig. 2d-iii), and liquid morphology under electrostatic attraction force (Fig. 2d-iv). The results show that the liquid rise height induced by interface forces is significantly greater than that induced by volume forces. The fluid completely fills the template microcavities under the total driving force. These simulation results confirm that the electric-field-induced interfacial force-dominated mechanism ensures effective filling while also allowing the complete formation of tilted structures.

Tilted nanostructures exhibit remarkable external stress sensitivity due to their unique geometry, which is primarily associated with volume forces. Therefore, during the nanoimprint process, precise control of the applied pressure is required to ensure complete structure formation. Mechanical analysis reveals that bending moments acting on the cantilevered end of these structures cause lateral deflections (Fig. S3). According to $P-\Delta$ effects, the axial force further amplifies the bending deformation, leading to a total displacement that rapidly increases with rising axial pressure (Note S3)⁴¹. This coupled compression-bending behavior increases the titled structure’s susceptibility to deformation and lateral instability, significantly degrading the nanostructure’s morphological fidelity.

As the volume driving force constitutes only a very small percentage of the total driving force during the e-N process, its impact on the structural morphology is minimal. Figure 2e illustrates the change rate of the structure’s vertical height caused by the volume force introduced at various electric field intensities in the e-N process. The results show that in the e-N process, the deformation rate is extremely low. For instance, at $E=8\times {10}^{6}V/m$, the change rate is only 0.12%, which is negligible; such an ultra-low deformation effectively maintains the structure fidelity. By comparison, c-N generally employs an external volume-force of 0.3–0.5 MPa. The absence of a thick residual layer for stress buffering makes template structures particularly susceptible to deformation under such pressures. Therefore, a finite element simulation is performed to study the deformation of the tilted structure under different pressures. The figure shows the deformation of a tilted structure with a 40° tilt angle, $l:d=3$, and a duty cycle of 0.5 under different pressures. When the pressure is 0.3 MPa, the deformation reaches 12.4%, which is two orders of magnitude greater than that in the e-N process. Such deformation significantly reduces the fidelity of structure replication. Experimental validation (Fig. S4) using SEM imaging confirms the pronounced structural compressive deformation predicted by the simulation. Furthermore, parametric analysis demonstrates that the stress-sensitive characteristics of the tilted structure are related to its tilted angle and aspect ratio. These findings provide guidelines for optimizing pressure in tilted nanostructure fabrication; precise pressure control must be tailored to specific geometric parameters (tilt angle, aspect ratio, etc.).

For optimizing optical coupling performance, controlling the residual layer thickness of nanoimprinted tilted structures is a critical process challenge. The key to eliminating the residual layer is to precisely control the resist volume so that it never exceeds the template cavity’s capacity. According to the traditional nanoimprint extrusion model, even under this condition, sufficient pressure and processing time are still required to compress the residual layer to an ultra-thin state for achieving residual-layer-free formation⁴². However, in this proposed approach, electric-field-induced interfacial forces directly acting on the resist alter the stability of the liquid film, making it more unstable and triggering spinodal dewetting. This mechanism enables the fabrication of tilted nanostructures with no residual layer. We model the evolution of the resist film on the substrate during nanoimprinting using lubrication approximation theory, incorporating the dielectrophoretic and electrocapillary forces acting on the liquid to construct modified Navier-Stokes equations (Note S4)⁴³. Then the evolution of the liquid film thickness can be described as:

$$\frac{\partial h}{\partial t}=\nabla \left[\frac{h^3}{3\mu }\nabla (\gamma {\nabla }^{2}h-{\prod }_{eff})\right]$$ 4

Where $h$ is the liquid film thickness, and $\mu$ is the liquid viscosity. The stability of the liquid film is governed by the effective disjoining pressure, which is defined as:

$${\prod }_{eff}={\prod }_{vdw}+{\prod }_{edl}+{\prod }_{dep}+{\prod }_{ecf}$$ 5

Here, ${\prod }_{vdw}$ denotes the Van der Waals force, ${\prod }_{edl}$ the electrostatic double-layer force, ${\prod }_{edl}$ the dielectrophoretic force and ${\prod }_{ecf}$ the electrocapillary force.

When ${\prod }_{eff}>\gamma {\nabla }^{2}h$, which means that the effective disjoining pressure exceeds the liquid-gas surface tension, the liquid film becomes dynamically unstable⁴⁴. During the nanoimprinting, as the template-substrate gap gradually decreases and the liquid fills the template cavities, the liquid film thins progressively. Particularly in the contact region between the protruding structures of the template and the substrate, the liquid film is the thinnest. When its thickness further decreases, dewetting occurs under the action of the effective disjoining pressure—i.e., the liquid film ruptures and forms dry spots. As the process continues, the dry spots gradually expand until the template is in direct contact with the substrate, ultimately achieving no liquid film residue between the contact regions. After curing, enabling the formation of residual-layer-free tilted nanostructures upon curing (Fig. 2f). Figure 2g illustrates the relationship between the effective disjoining pressure and the surface tension of the liquid film under different electric field intensities. When $E>5.3\times {10}^{6}V/m$, the system satisfies the aforementioned instability criterion but triggers interfacial destabilization.

Tilted nanostructures exhibit significantly higher stress during the demolding process than the vertical structures due to the geometric interlocking between the template and replicated structures, with stress further escalating as the tilt angle increases. Additionally, demolding stress exhibits an inverse relationship with residual layer thickness—it increases drastically in residual-layer-free scenarios⁴⁵. Consequently, for demolding residual-layer-free tilted nanostructures, a customized demolding path strategy is required to effectively reduce stress and mitigate the risks of structural deformation or interfacial delamination.

This work employs a peel-off demolding method, which is illustrated in Fig. 3a. After the nanoimprint is completed, the flexible template is directionally peeled off from the substrate edge until it is completely separated from the substrate. The demolding-induced stress primarily originates from three interfacial interactions: adhesion between the template and fabricated structures $F_a$, frictional forces during template disjoining $F_f$, and impact stresses from mechanical engagement. While adhesion and frictional forces, which depend on the surface properties, can be mitigated by hydrophobic treatment (which is beyond the scope of this study), this work focuses on optimizing the demolding path to reduce impact stresses induced by the template during disjoining.

Fig. 3. Peeling-off demolding path optimization strategy for tilted nanostructures.

a Schematic of peeling-off demolding, b simulation model of stress distribution in fabricated tilted nanostructures during demolding, and effect of demolding position, c demolding angle and demolding speed on demolding stress

This study explores tilted nanostructures with a 40° tilt angle and $l:d=3$, where the geometric relationship between the peel-off direction and the grating direction is the primary factor influencing demolding stress. Figure 3a depicts typical scenarios in which the relative angle between the peel-off and grating directions is 0°, 45°, 90°, 135°, and 180°. For a single-period structure, the peel-off demolding process can be regarded as the template rotating along a circular arc path with radius $R$ and demolding angle $\alpha \approx \arctan \frac{h}{R}$.

The demolding stress analysis is performed using the finite-element model depicted in Fig. 3b-i. The edges and corners of the structure are chamfered to reduce the mesh distortion and convergence. Figure 3b-ii presents the maximal stress distribution nephograms of fabricated structures under different demolding locations during the demolding process for the demolding speed of 1 mm/s and demolding angle of 25°. The quantitative data shown in the figure indicate that the peak demolding stress at the 0° position is an order of magnitude greater than at other orientations, while the 180° position yields the lowest stress. Figure 3c reveals that demolding stress decreases with increasing demolding angle when the demolding speed is fixed at 1 mm/s and the demolding location at 180°, with minimal stress achieved when the demolding angle approaches the structure’s tilt angle. Additionally, Fig. 3c shows that the demolding stress increases with speed, which can be attributed to impulse effects from impact loading during high-speed disjoining. In summary, for high-angle-tilted nanostructures, using a 180° demolding location, larger demolding angles, and slower demolding speeds effectively reduces demolding stress in residual-layer-free tilted nanostructures.

An e-N prototype was developed based on the optimization of electric-driven interfacial force-filling processes and demolding trajectories. The exterior design of the device is shown in Fig. 4a, and its core modules, comprising an optical system, precision motion system, programmable power supply, vacuum system, and control system, are shown in Fig. 4b. The core nanoimprinting component is a flexible roller, which consists of a driven roller, a flexible conductive template, and a substrate chuck platform. The flexible conductive template is a multilayer structure made up of a backing layer, conductive layer, and structural layer that is fabricated in situ on the prototype (Fig. 4c). The template assembly is mounted on a specialized fixture, with one end secured to the roller shaft and the other end aligned with the substrate plane on the template holder (Fig. 4d), forming a continuous wrap around the roller to constitute the flexible roller system (Fig. 4e). The electric field applied between the template and substrate acts as an interfacial driving force for resist filling and facilitates uniform template-substrate conformal contact. In contrast to traditional nanoimprinting technologies (roll-to-roll, roll-to-plate, plate-to-plate) with fixed suboptimal (non-adjustable) demolding angles, the roller enables precise vertical and horizontal movement, allowing real-time adjustment of the peeling-off angle during the demolding process to achieve large demolding angles that effectively mitigate demolding stresses. This technological advancement provides critical process support for residual-layer-free nanoimprinting of high-angle-tilted nanostructures, allowing for the replication of high-fidelity complex nanostructures.

Fig. 4. Electric-field-assisted nanoimprinting prototype system.

a External view of the equipment, b internal structural composition of the equipment, and the core nanoimprinting component—the flexible-roller mechanism (e), composed of the flexible conductive template (c), drive roller shaft, and substrate support platform (d)

High-fidelity residual-layer-free fabrication of high-angle-tilted nanostructures has been successfully achieved using the self-developed e-N equipment. Figure 5a displays the mold with an array of tilted nanostructures. A working template was replicated from this mold and subsequently used for nanoimprinting to obtain the final nanostructure. Figure 5b shows the successfully nanoimprinted structures (40° tilt angle, 200 nm linewidth, and 600 nm length). The nanoimprinting process is thoroughly described in the Experimental Section. In comparison to the template structures, the fabricated structures have excellent fidelity and are free of residual layers.

Fig. 5. Fabrication and characterization of high-fidelity, residual-layer-free tilted nanostructures.

a Tilted structures on the mold, b nanoimprinted tilted nanostructures, c uniformity analysis of fabricated tilted nanostructures, d, e Tilted nanostructures with different morphologies (width, tilt angle, and height), and f SEM/EDS characterization of residual-layer-free tilted nanostructures

The structural uniformity of nanoimprint samples is critical for device performance. Figure 5c depicts a tilted nanostructure sample, with a structural region diameter of 8 mm. Five measurement locations were selected on the sample, and three structural periods were randomly chosen at each location to measure the tilt angle and residual layer thickness of the tilted structure. The measurement results presented in Fig. 5c show that the tilt angles and residual layer thicknesses of the fabricated tilted nanostructures are highly consistent and uniform across the entire substrate. This uniformity is attributed to the precisely controlled uniformity of both the template thickness and initial resist thickness, which ensures a homogeneous distribution of electric field forces over the substrate surface, which is essential for maintaining structural uniformity and fidelity in the fabricated tilted nanostructures.

Implementing the aforementioned approach, the residual-layer-free various tilted nanostructures were fabricated. Figure 5d presents successfully replicated nanostructures with 250 nm linewidth, 30° tilt angle, and 300 nm length, while Fig. 5e shows structures measuring 200 nm linewidth, 20° tilt angle, and 500 nm length. The fabricated structures have a high pattern fidelity with the template structures, with no visible residual layers on the surface and sharply defined structural edges. Furthermore, this method can be extended to the fabrication of 2D tilted nanostructures (nanohole and nanorod arrays), as demonstrated in Figure S5. These results systematically validate the technique’s capability to fabricate complex high-fidelity residual-layer-free tilted nanostructures, paving the way for the scalable manufacturing of complex micro-nano structures. The fabrication capability of this method in terms of structural resolution, aspect ratio, and tilt angle adequately meets the current requirements for optical coupling devices. For more extreme processing limits, follow‑up studies will systematically investigate factors such as master fabrication capability and material properties.

Elemental analysis was conducted on the fabricated nanostructures with small tilt angles to further evaluate residual layer thickness, as their structural geometry allows access to the bottom of the trench during the scanning. Figure 5f demonstrates the residual-layer-free tilted nanostructures with a line width of 500 nm and a tilt angle of 10°, fabricated using a high-viscosity nanoparticle-doped resist. The resist is doped with high-refractive-index TiO2 particles—a standard approach in optoelectronics for enhancing polymer refractive index. Cross-sectional and top-view SEM images (Fig. 5f-i & ii) reveal the tilted nanostructures with superior structural collimation and edge sharpness, as well as a near-zero residual layer. The Ti elemental line scan (Fig. 5f-ii) and corresponding area mapping (Fig. 5f-iii) exhibit minimal signal intensity within inter-structure grooves, contrasting sharply with control samples containing residual layers (Fig. S6). This differential Ti distribution provides conclusive evidence that our method can achieve residual-layer-free nanoimprinting even with high-viscosity functional resists.

Tilted-nanograting optical coupling demonstrates high diffraction efficiency and superior display performance when integrated into AR displays. Leveraging these developments, the high-angle-tilted residual-layer-free structure, depicted in Fig. 4b, has been successfully integrated into AR displays. As shown in Fig. 6a, the AR displays have in-coupling and out-coupling regions, with tilted nanogratings serving as in-coupling structures.

Fig. 6. Applications of high-fidelity, residual-layer-free tilted optical coupler in AR displays.

a Nanoimprint-fabricated AR displays, b tilted nano-gratings in the in-coupling region fabricated by e-N, c efficiency analysis of AR Displays fabricated by e-N and c-N, d process and performance analyses of AR displays fabricated by the two nanoimprinting methods, and e, f display performance of the AR displays fabricated by e-N

Figure 6b presents residual-layer-free tilted nanostructures fabricated via the e-N method. While the template configuration and film thickness conditions remain consistent with those shown in Fig. S4, our method successfully maintains structural geometric integrity under residual-layer-free conditions. This stands in sharp contrast to the deformation observed in Fig. S4, clearly demonstrating its superior structural fidelity. Currently, commercially available AR displays predominantly utilize c-N processes for fabrication. Figure S7 presents a cross-sectional SEM image of a mature commercial product (supplied by Mojie Technology) fabricated using an identical template. In such conventional processes, a relatively thick residual layer is typically retained to preserve structural fidelity; however, noticeable deformation of the structure is still observable. To systematically evaluate the performance advantages of our proposed method, we conducted a comprehensive performance comparison between 20 sets of AR displays fabricated using our approach and 20 sets of commercially available AR displays produced via c-N. The corresponding comparison results are displayed in Fig. 6c. The average efficiency of AR displays fabricated by e-N is 594.634 with a standard deviation of 24.654, while that of those prepared by c-N is 484.733 with a standard deviation of 51.327. The efficiency of e-N samples is improved by 22.67%, and the lower standard deviation indicates better sample uniformity, which is crucial for consistency in the case of mass production.

In addition to increased efficiency, comparative analyses of process and other key performance improvements were conducted (Fig. 6d). In the evaluation of efficiency uniformity across different positions in the out-coupling region of AR displays, the standard deviation of e-N samples was 21.88% lower than that of c-N samples, indicating a significant improvement in outcoupling uniformity and thereby enabling superior display performance. Modulation transfer function (MTF) measurements showed that the e-N process outperforms the c-N process by 39.68%, indicating a significant increase in optical signal transmission efficiency. In terms of process improvement, the template lifespan was doubled due to the absence of fatigue damage caused by sustained mechanical pressure—an issue inherent in c-N. Additionally, precise control over resist distribution reduced material usage by 60%, directly contributing to cost reduction.

The fabricated AR displays demonstrate exceptional performance (Fig. 6e, f), exhibiting sharp image delineation with high luminance, indicating a promising application potential in navigation. These achievements not only validate the practical viability of this technique for fabricating high-tilted-angle, residual-layer-free, high-fidelity nanostructures in AR applications but also provide a universal fabrication approach for high-precision manufacturing of photonic devices and photonics-integrated chips.

Conclusions

In this paper, an electric-field-assisted nanoimprinting technique is proposed and demonstrated for fabricating high-fidelity, residual-layer-free, high-angle-tilted nanostructures for a high-performance optical coupler. The proposed method utilizes an applied electric field to generate electro-interfacial forces that actively drive the resist filling process, which effectively avoids structural damage caused by volumetric forces, thereby ensuring high-fidelity pattern replication; furthermore, these interfacial forces induce dewetting in the liquid film, establishing the essential condition for fabricating residual-layer-free nanostructures. This approach resolves the fundamental conflict between the low-pressure requirements for pattern fidelity and the high-pressure demand for residual layer-free structure. A demolding pathway optimization strategy is developed to effectively mitigate stress-induced structure damage during demolding processes. A self-developed nanoimprint prototype was used to demonstrate the fabrication of various tilted nanostructures using dynamic parameter modulation based on template geometries. This work not only provides novel solutions for enhancing the performance of tilted nanograting optical couplers but also establishes a pivotal technical pathway for high-precision manufacturing in domains such as optical metasurfaces, chiral photonics, and photonics chips.

Materials and methods

The nanoimprinting process

Silicon master molds were fabricated via electron-beam lithography, etching, and subsequent surface hydrophobic treatment. Flexible working templates were obtained by replicating the silicon master molds. Silicon wafers were selected as nanoimprinting substrates to facilitate dicing and structure characterization, while glass substrates were used for fabricating AR displays. The resist had an acrylate-based formulation that was diluted before use to achieve the desired thickness. In this study, the resist was deposited via a spin-coating process, with its thickness precisely controlled by regulating the spin speed and duration. This step was carried out using an EVG101 coating system. A high-refractive-index resist was prepared using a TiO₂-particle-doped system. The process parameters for fabricating 40° tilt-angle grating waveguides are as follows: initial resist thickness of 230 nm (the specific parameters were as follows: 1.5 ml of resist was dispensed onto the wafer substrate, followed by a two-stage spin sequence—first at 500 rpm for 10 s, then at 4500 rpm for 60 s), applied electric field intensity of 8 V/μm, demolding angle of 25°, and demolding speed of 1 mm/s.

Simulation

In investigating the effects of structural fidelity and residual layer thickness on diffraction efficiency, the grating structure was segmented into multilayers to enhance simulation accuracy, with the refractive indexes of the grating and substrate set to 1.9 and 1.5, respectively. Finite-element stress analysis of tilt-structural sensitivity and demolding stress was performed using ABAQUS, with Young’s modulus of the structural layer material set to 50 MPa. The filling-driving force contributions and dynamic filling processes for tilted structures during nanoimprinting were investigated using the COMSOL Multiphysics software.

Characterization

The wetting angle was measured by a contact angle measurement device, OCA15EC, provided by Data Physics Instruments, Germany. The SEM images were obtained using a Hitachi S-3000N scanning electron microscope. The performance of the AR displays was characterized using a comprehensive performance testing platform.

Author contributions

Conceptualization: Y.F. and C.W. Methodology: Y.F., C.W., and J.S. Investigation: Y.F., C.W., and X.Z. Visualization: Y.F., H.S., and X.C. Simulation: Y.F. and S.L. Funding acquisition: C.W. and J.S. Project administration: H.T. and X.C. Supervision: X.L. and J.S. Writing—original draft: Y.F. Writing—review and editing: C.W. and J.S.

Conflict of interest

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41378-026-01231-2.