2.1.1.

NIL variants based on resist curing

NIL is categorized as thermal NIL and UV NIL, depending on the triggers of resist curing. UV NIL involves the use of a photopolymer resist, whose curing trigger is UV light [Fig. 2(a), i]. A typical UV NIL process is described as follows. A resist is spin-coated onto a substrate, and a stamp with an inverse shape of master mold patterns is created using PDMS. Owing to the stamp being in contact with the resist, it is filled with the resist under the desired pressure and time. Subsequently, the resist is hardened via UV exposure at room temperature. Consequently, the stamp is released, and the patterns are transferred to the substrate. UV NIL offers certain advantages over thermal NIL in terms of productivity because it can be performed at room temperature and in low-pressure conditions.

Fig. 1

Large-area, low-cost fabrication methods are introduced as follows: nanoimprint lithography, transfer printing, capillary force lithography, block copolymer self-assembly, and colloidal photonic crystal self-assembly.

Fig. 2

Types of nanoimprint lithography (NIL) depending on curing stimuli and contact types. (a) NIL variants based on resist curing. (i) NIL processes using a photopolymer resist, whose curing trigger is UV light. (ii) NIL processes using a thermoplastic polymer resist, whose curing trigger is heat. (b) NIL variants based on imprint contact types. (i) P2P NIL processes through contact between flat stamp and substrate. Reprinted with permission from Ref. [41], copyright 2021, American Chemical Society. (ii) R2R NIL processes using a pattern roller. Reprinted with permission from Ref. [45], copyright 2015, American Vacuum Society. (iii) R2P NIL processes through contact between rollers and flat substrate. Reprinted with permission from Ref. [47], copyright 2022, Multidisciplinary Digital Publishing Institute.

Thermal NIL, known as hot embossing, uses a thermoplastic polymer resist that softens when heated [Fig. 2(a), ii]^[39,40]. A thermal NIL process is conducted at a stamp temperature higher than the substrate transition temperature of the resist. The elevated temperature and sufficient pressure provide fluidity for the resist, which facilitates its entry between the patterns of the stamp. Thereafter, before detaching the stamp, the imprinting temperature is lowered below the resist temperature to allow its solidification, which leads to pattern transfer.

2.1.2.

NIL variants based on contact types

NIL can be categorized based on imprint-contact types. Three contact types for the NIL process are possible: plate-to-plate (P2P), roll-to-roll (R2R), and roll-to-plate (R2P) NIL. In this paper, we describe these methods, including their overall configurations and applications.

In P2P NIL, imprinting proceeds through contact between the flat stamp and the substrate [Fig. 2(b), i]^[41]. Recently, a study realized 30 nm features using the P2P method, thereby indicating the potential of high-resolution patterning in NIL^[42]. This method is independent of the size of the contact area because it enables the use of a stamp having the same area as the total pattern area of large-area micro/nanostructures. However, some force is required to increase the imprint pressure as the contact area expands. In addition, if the resist is deposited using the drop method (instead of spin-coating), which results in direct absorption of the resist into the stamp, air is readily trapped between the stamp and substrate. This leads to defects in the imprinted outcome. Hiroshima et al. demonstrated a solution to these bubble defects using pentafluoropropane based on gas condensation^[43,44]. They proposed two methods, namely, resin squeezing, which discharges bubbles from the mold, and gas condensation, which significantly decreases the volume of the bubbles.

R2R NIL uses a pattern roller consisting of inverse patterns to imprint patterns onto a flexible substrate on support rollers [Fig. 2(b), ii]^[45]. Considering the factors of high speed and large area, roller-based imprinting processes capable of continuous production have been proposed. R2R NIL differs from P2P NIL, which has difficulties in applying uniform pressure to the mold and ensuring uniformity in the residual layer of the resist owing to the large contact area. The contact area between the rollers and substrates on the support rollers is smaller than that of the P2P NIL. This results in a reduced imprinting force and allows to realize a continuous process, thereby facilitating large-area patterning. In addition, because the process continuously patterns a large area by spinning reusable rollers, R2R NIL is a cost-effective technology for the mass production of nanostructures^[46].

R2P NIL follows a process similar to that of R2R NIL, except that the patterning roller transfers the patterns to the substrate [Fig. 2(b), iii]^[47]. Roller-based methods have realized sub-100 nm features; however, the fabrication of sub-50 nm nanostructures is challenging because of the requirement of precise contact angle control between the resist and substrate^[48]. These methods also use the continuous resist coating mechanism rather than the spin-coating of the resist, which is mainly used in P2P NIL. However, the mechanism is more complicated and requires uniform control.

2.1.3.

Applications of NIL

NIL can be used to fabricate uniform nanopatterns on any flexible substrates through a single-step process. This technology has also been investigated for use in realizing industrial-scale applications that require mass production with low-cost materials, mask-free processes, low manufacturing cost, and high production speed^[49,50]. The polymers mainly used in NIL have wide applicability in optical/electronic devices such as 3D imaging^[51], solar cells^[52,53], light-emitting diodes (LEDs)^[54], and polarizers^[55,56] because of their ease of manufacturing, low cost, and low reactivity. In addition, NIL has been researched for fabricating metasurfaces using metals such as gold (Au)^[57]. Thus, the feasibility of various materials in NIL has been expanded to other application fields. We confirmed the characteristics of NIL in terms of high resolution, scalability, productivity, and facileness via antireflective films^[58], polarizers^[59], solar cells^[60], and metaholograms^[61].

The scalable fabrication of antireflective flexible films has been realized via thermal R2R NIL, and processing parameters (e.g., web speed and temperature) have been investigated^[62]. Jacobo-Martín et al. achieved the desired optical performance covering ranges of 400–700 nm by realizing sub-wavelength scale moth-eye structures via R2R NIL [Fig. 3(a), i and ii]^[58]. The processes are carried out at pressure set to a value of 0.2–0.3 MPa and temperatures ranging from 80°C up to 140°C. Furthermore, they defined a correlation between the process parameters with the optical and mechanical properties of the films, proving the optimization of thermal R2R NIL. Also, Ok et al. showed a dual-band infrared (IR) filter covering ranges of 6–8 and 10–12 µm by patterning a sub-wavelength scale metal disk array via UV R2R NIL. They demonstrated that R2R NIL continuously fabricates large-area micro/nanostructures on flexible substrates at high speed and low cost.

Fig. 3

Various applications using the NIL. (a) (i) Thermal R2R NIL for an antireflective film. The processes go on under a certain range of pressure (P) and temperature (T). (ii) Measured total reflection spectrum. (iii) Scanning electron microscope (SEM) image of imprinted moth-eye structures at 140°C. Inset: 3D reconstructions of atomic force microscope (AFM) images of imprinted areas. Reprinted with permission from Ref. [58], copyright 2021, Springer Nature. (b) (i) P2P NIL for an infrared nanowire grid polarizer. (ii) TM and TE transmittance spectrum. (iii) Photograph of the polarizer with 4-inch area. Reprinted with permission from Ref. [59], copyright 2018, Optica. (c) (i) Schematic of the P2P NIL for solar cells with grating structures. (ii) Power–voltage graph of solar cells with/without grating structures. (iii) Side view of the SEM image of grating structures via imprinting with vibration. Reprinted with permission from Ref. [60], copyright 2022, Elsevier. (d) (i) P2P NIL using particle-embedded resist (PER) for metasurfaces. (ii) Calculated hologram efficiency. (iii) Simulated/generated holographic images of metasurfaces. Reprinted with permission from Ref. [61], copyright 2019, American Chemical Society.

With polarization imaging and spectral analysis, IR polarizers have attracted attention in IR optics and nanofabrication. Conventional IR polarizers are mainly made by depositing nanometallic wires on IR-transparent materials such as KRS-5 and ${\mathrm{CaF}}_2$. However, these materials are brittle and hygroscopic, built on hard substrates, and are less compatible with curved compositions. Therefore, a flexible polarizer brings significant progress for applications including displays, monitors, etc. Kang et al. realized flexible 4-inch (1 inch = 2.54 cm) polarizers using thermal NIL [Fig. 3(b), i and iii]^[59]. They demonstrated the polarization characteristics, showing TM transmittance greater than 70% in wavelength ranges of 4.70–5.69 and 5.75–6.59 µm [Fig. 3(b), ii]. Also, an extinction ratio of more than 20 dB in the wavelength range of 3.6–6.7 µm is exhibited, implying great polarization characteristics. Furthermore, they emphasized functionality of the flexible IR nanowire grid polarizer due to applicability in the fields of curved surface monitoring equipment and polarized imaging equipment.

There has been growing interest in solar energy as an ideal green energy source. However, its low power generation efficiency and high cost are significant obstacles in photovoltaic technology. Gu et al. fabricated Si solar cells using vibration-assisted UV NIL [Fig. 3(c), i and iii]^[60]. They developed NIL by applying vibration to the conventional method of imprinting, which can improve the filling rate of photoresists by 25% and give higher pattern accuracy. Through this method, they realized periodic nanogratings, showing power conversion efficiency (PCE) increased by 25% when compared to bare Si solar cells [Fig. 3(c), ii]. Also, textured surfaces enhance light collection, realizing efficient light modulation including reflection control and light trapping properties^[63]. This modulation has been investigated to enhance the efficiency of a solar cell by texturing at the front of the solar cell or using pre-textured solar cell substrates^[64,65]. Amalathas et al. fabricated periodic inverted nanopyramid structures in monocrystalline Si solar cells^[66]. They realized nanostructures uniformly over a large area of $10\mathrm{mm}\times 10\mathrm{mm}$ through two UV imprint processes. This implies that NIL is a scalable and low-cost fabrication method that eliminates defect generatability due to etching. Thereafter, employing such structures, an improved PCE of 11.73% is achieved by reducing Fresnel reflection and trapping incident light into the solar cells.

Metasurfaces comprising periodic sub-wavelength nanoantennas have overcome the miniaturization limitation. Also, various materials such as ${\mathrm{TiO}}_2$^[67], perovskite^[68], and silicon^[69] have been studied for metasurfaces to realize various functions, and many manufacturing methods are being developed to reach nanoscale feature sizes and high resolution. Consequently, they have been applied to metalenses^[70], metaholograms^[71], and color printing^[72]. However, the use of EBL and holographic lithography to realize large-area metasurfaces has led to high manufacturing costs and optical diffraction limit problems such as a proximity effect^[73]. Kim et al. proposed UV NIL as a one-step printable platform without secondary operations (e.g., etching and deposition), demonstrating the feasibility of NIL for the mass production of optical metamaterials [Fig. 3(d), i]^[61]. This one-step process facilitates high productivity and low cost in the fabrication of optical devices. In addition, using a UV-curable particle-embedded resist, the refractive index of the resist itself is increased, which enables the use of the patterned resist as a metasurface. The fabricated metaholograms exhibit a conversion efficiency of 46% for $\lambda =532\mathrm{nm}$, which is comparable to that of low-loss dielectric metasurfaces [Fig. 3(d), ii and iii].

Palladium (Pd)-based ${\mathrm{H}}_2$ sensors have the potential for reversible gas-sensing capabilities and sensor mechanisms owing to their light weight and mechanical durability^[74,75]. However, as Pd structures become smaller and the surface area becomes larger, high resolution is required. Lim et al. demonstrated scalable UV NIL fabrication processes with a polycarbonate (PC) film through the fabrication of flexible and durable gas sensors^[76]. In addition, they showed that sensors with a size of $3\mathrm{cm}\times 2\mathrm{cm}$ maintain sensing performance with a response time of 57 s at 50 ppm (parts per million) ${\mathrm{H}}_2$ concentration and achieve 3500 ppm of ${\mathrm{H}}_2$ with a fast response time of 18 s.

2.2.1.

Intaglio transfer printing

Intaglio transfer printing uses an intaglio trench, which is a plate engraved in the shape of the desired patterns, to transfer the patterns onto the target substrates. Previous printing methods exhibit discrepancies between the original designs and transferred patterns owing to interfacial adhesion. Although existing transfer printing can be compatible with various materials, it fails to offer high-resolution/quality patterning. Kim et al. proposed a novel reverse-offset printing technology by patterning sub-10 µm copper nanowire (CuNW) transparent conducting electrodes^[80]. The intaglio trench allows only the desired pattern to remain on the stamp, without any morphological changes [Fig. 4(a)]. In addition, the method ensures high-resolution aligned pixels through multiple transfer printings, thus realizing that the sheet resistance of the CuNW micropatterns decreased to $31.6\mathrm{\Omega }{\mathrm{sq}}^{-1}$ as the irradiation energy increased to $3.32\mathrm{J}{\mathrm{cm}}^{-2}$. The ability to transfer patterns on various substrates regardless of the size, shape, and arrangement of pixels results in high-performance operation and miniaturization of the devices, coupled with several application possibilities^[81]. Also, Choi et al. proposed an intaglio transfer process integrated with quantum dots (QDs) to create high-resolution full-color RGB QD LED (QLED) arrays, realizing a high electroluminescence performance of 14,000 cd m⁻² at 7 V^[82].

Fig. 4

Mechanisms of different transfer printing methods. (a) Transfer printing processes using an intaglio trench, which allows only the desired pattern to remain on the stamp without any morphological changes. Reprinted with permission from Ref. [80], copyright 2022, American Chemical Society. (b) Laser-driven transfer printing processes. Selectively heating the inks via laser induces strong thrust between stamp and inks, forming a pattern on receiver. Reprinted with permission from Ref. [83], copyright 2021, John Wiley and Sons. (c) Direct roll transfer printing processes. Si nanoribbons (NRs) are released by undercut etching of the buried oxide (box) (scale bar 10 µm, red line), and then brought into direct contact with semi-cured polyimide (PI) thin film (scale bar 25 µm, black line). Reprinted with permission from Ref. [85], copyright 2021, Springer Nature.

2.2.2.

Laser-driven transfer printing

In transfer printing, the transferred patterns are significantly influenced by the interfacial adhesion strength, which is dominated by van der Waals interactions. The key role in the transfer process is to regulate interfacial adhesion states between strong for pick-up and weak for printing in a repeatable manner. Conventional pick-up and transfer processes proceed in vacuum chunks, but the size of the structure is sub-100 µm, and the limit to reduction of the vacuum system still remains a great challenge. Luo et al. presented an advanced transfer printing technique based on thermally controlled tunable adhesives^[83]. The thermally controlled tunable adhesive provides desired adhesion states in transfer processes [Fig. 4(b)]. The inks successfully attached to the adhesive of the stamp are selectively heated via a laser beam. This process causes interfacial delamination to form patterns on a target substrate owing to the strong thrust strength. By changing 100°C temperature, suction strength of 25.5 kPa and thrust strength of 32.8 kPa are offered. They demonstrated that the thermally induced air pressure change in the cavity facilitates interface bonding/debonding.

In addition to the thermal control method, a way to change the adhesion state through the unique structure of the stamp was also proposed^[84]. Kim et al. studied pressure-modulated adhesion using a stamp with four features of pyramidal relief. The structures control the adhesion strength such that the strength is reversibly switched when the stamp peels off the ink from the donor and transfers the ink to the target substrate. Here, a method using surface relief structures, developed to control the modes of adhesion and observed in geckos, is discussed. When an object is brought from the donor, the adhesion strength is maximized by increasing the contact area through the collapse of the regions between microchips. Subsequently, the elastic restoring force causes the stamp to return to its original shape, gripping the object only at the sharp points of the microtips. Consequently, the adhesion strength is minimized, and a high peeling-off velocity is promoted when transferring objects from the stamp to the target substrate. These characteristics enable rapid, reliable, and repeatable transfer printing, which indicates high productivity and potential for application.

2.2.3.

Direct roll transfer printing

Many advances in transport printing have been made for large-scale implementation of high-quality nanostructures. Zumeit et al. proposed the direct roll transfer printing as a single-step process without using any elastomer stamp [Fig. 4(c)]^[85]. Conventional transfer printing has difficulties in controlling mechanisms of viscoelastic PDMS stamps for transferring sub-100 nm thick structures. Direct roll printing avoids the use of PDMS stamps, thus reducing the complexity of processes. Si nanoribbons (NRs) are brought into direct contact with semi-cured polyimide (PI) thin film, which enhances the adhesion between NRs and receiver substrates during imprint. Also, lower process steps, shorter printing time, enhanced printing yield, and lower fabrication cost are realized compared with the conventional approach. They demonstrated the high transfer yield of $\sim 95\%$ with perfect registration, showing the possibility of printing ultrathin micro/nanostructures based on other high-mobility materials. The printed Si NRs were used as channel materials to obtain high-performance transistors, showing great mobility [>630 cm²/(V·s)], current on/off ratio (∼10⁶), and robustness after 100 cycles of bending.

2.2.4.

Water-assisted transfer printing

The transfer of nanodevices onto diverse substrates such as papers, plastics, and glass has not been easily achievable before. As properties such as flexibility, transparency, biocompatibility, and conductivity of the substrates are required for diverse applications such as flexible displays^[86] and transistors^[87], technological developments have become necessary. Here, water-assisted transfer printing methods have been proposed to overcome the difficulty in transferring multiple layers onto unconventional substrates. In addition, the method proceeds with each layer aligned at approximately 100% while maintaining the measured efficiency of the transferred devices. Peeling-off proceeds in a water bath at room temperature based on the phenomenon of water penetrating the interface between the peeling layer and donor^[88]. This process occurs in 3–4 s, following which the nanowire electronics are transferred to the target substrate. The transfer of a 4-inch wafer-scale area of microelectrodes is demonstrated, implying the scalability of transfer printing. In addition, this method offers certain advantages, such as insensitivity to mechanical and thermal stresses on the target substrate and the feasibility of diverse nonconventional substrates.

2.2.5.

Applications of transfer printing

Flexible electronic systems, including wearable devices, have been actively investigated because they can be bent, folded, and stretched without degrading their electrical performance and reliability. Here, transfer printing has been extensively studied as a technology for integrating functional devices with flexible substrates^[89,90]. Wide-ranging applications have emerged owing to the ability of the transfer printing method in enabling the assembly of functional devices onto flexible or specific-shape (e.g., curved) substrates such as polyethylene terephthalate films and PI substrates^[91,92]. The demonstrated functional components include transistors^[93,94], capacitors^[95], LEDs^[96,97], energy-harvesting devices^[98], and various sensors^[99].

Stretchable solar cells have been employed as wearable power generators because of their robust mechanical properties, which are vital for portable electronic devices^[100]. However, solar cells are unsuitable for stretchable systems because of their intrinsic rigidity and brittleness. To overcome these constraints, transfer printing is required, which involves arranging an inorganic semiconductor from a donor to a stretchable substrate. However, the transfer printing of inorganic amorphous thin-film semiconductors is challenging owing to the absence of a suitable sacrificial layer. Nam et al. developed a novel approach using a water-soluble ${\mathrm{GeO}}_2$ sacrificial layer^[100]. Stretchable inorganic amorphous thin-film solar cells were fabricated via transfer printing with a sacrificial layer, which is placed on the donor and patterned upon. When peeling off the pattern using a stamp, the sacrificial layer weakens adhesion, thereby allowing successful transfer. Owing to the layer requiring mild etching conditions, compatibility with high process temperatures, and mechanical stability, the transfer printing of the solar cells over a large area was realized. A comparison of the performances of the solar cells before and after transfer printing revealed that the etching or transfer printing process has no effect on their performance, as shown by the current density–voltage characteristics. This demonstration of stretchable solar cells with inorganic amorphous thin-film semiconductors significantly broadens the applications of solar cells as well as the capabilities of transfer printing.

Metallic nanogaps have many applications in nanoscale photonics and electronics. Nanostructures with metallic nanogaps can confine the electromagnetic field within the deep-sub-wavelength region of the gap and control/enhance the fluorescence emission. This enhancement is suited for optical systems such as surface enhanced Raman scattering (SERS) and IR sensing. Chen et al. enabled the fabrication of multiscale metallic nanogaps with reduced gap size ($\sim 5\mathrm{nm}$)^[101]. They overcome the resolution limitations of direct lithography approaches by transferring Au ink onto PDMS [Fig. 5(a), i]. After transferring the Au ink with large gap size of 120 nm to the pre-stretched PDMS, retrieving the PDMS reduces the gap size to 5 nm. They evaluated the photoluminescence (PL) enhancement by transferring again onto ${\mathrm{MoS}}_2$ flakes. Compared to the PL of ${\mathrm{MoS}}_2$ flakes decorated with a 120 nm gap (orange line), the post-assembled metallic nanogaps led to enhanced PL emission (red line) [Fig. 5(a), ii]. By realizing sub-10 nm metallic nanogaps in two transfer printings, they showed a wide variety of applicability including sensing, imaging, and energy applications.

Fig. 5

Various applications of transfer printing. (a) (i) Transfer printing integrated with post-assembly to fabricate the sub-10 nm nanogap. A gold nanodimer with 120 nm gap size is transferred onto pre-stretched polydimethylsiloxane (PDMS) substrate, which then is released. After releasing the substrate, the large gap in dimer is shrunk ($\sim 5\mathrm{nm}$). (ii) Photoluminescence (PL) spectra of MoS₂ flake with the post-assembled nanodimer (red line) and initial nanodimer (orange line). (iii) SEM images of gold nanodimer before and after releasing the stress. Reprinted with permission from Ref. [101], copyright 2020, John Wiley and Sons. (b) (i) Synthesizing and transfer printing processes of ZnO nanostructures on a polycarbonate (PC) substrate. (ii) Transmittance spectra of A series and B series. A series represents as-grown nanorod arrays, and the corresponding transferred nanorod arrays with 60 s O₂ treatment are represented as B series. Inset: SEM images of transferred nanorod arrays. Reprinted with permission from Ref. [109], copyright 2022, John Wiley and Sons.

LEDs have the potential to efficiently use energy for applications in flexible electronic devices. Recently, gallium nitride (GaN)-based LED devices, which have been used in liquid crystal displays, lighting devices, etc., have been studied for high-efficiency and high-quality light sources^[102–104]. Blue InGaN LEDs are used as alternatives for the existing infrastructure under general illumination^[105]. Kim et al. fabricated a thin, flexible, and white-light module by fabricating the phosphor in PDMS and laminating it via tape to a matching array of microscale inorganic LEDs (μ-ILEDs). Transfer printing demonstrated that the assembly of LEDs could be easily attached to plastic or glass substrates at room temperature while maintaining performance, such as emission intensity.

To effectively modulate the phase and wavefronts of light, metasurfaces comprising periodic sub-wavelength nanostructures have been exploited as holograms, lenses, color filters, reflectors, etc.^[106–108]. Specifically, in color filters, several colors are obtained by tailoring the structural parameters associated with the nanostructures. Moreover, the incoming light can be filtered into distinct colors according to the structural parameters of the nanostructures. Liu et al. demonstrated the scalability of the transfer printing process through the fabrication of $300\mathrm{\mu m}\times 500\mathrm{\mu m}$ “windmill” ZnO patterns including more than 300,000 nanorod units on PC film [Fig. 5(b), i and iii]^[109]. Further, to overcome the adhesion strength between the structures and target substrates, poly(methyl methacrylate) (PMMA) was chosen as the protection and transfer layer because of its high Young’s modulus ($\sim 5\mathrm{GPa}$) in the solidified PMMA polymer. Although there was a variation in the GaN layer interference on the substrate before and after the transfer, measured spectra showed that the major peaks of the spectrum curves are redshifted with the increase in the diameter of the nanorods [Fig. 5(b), ii]. In particular, for the transferred arrays, the lowest transmittance constantly decreases with the change in the diameter of the nanorods, implying a high color filter efficiency. In addition, they demonstrated that the presented color pallets indicate full coverage of the color gamut, indicating successful transfer printing.

2.3.1.

Light-controlled CFL

Recently, many studies have been conducted to spatially modulate the nanopixel’s height^[112,113]. To enhance functionality through this, grayscale photolithography and scanned beam lithography are used for fabrication of nanostructures. However, the methods have high production cost due to complex processes. Li et al. reported that height-controlled CFL is possible with certain photopolymers^[114]. Since it is difficult to control the polymer’s capillary rise to realize a grayscale nanopixel with conventional CFL by itself, they introduced light-controlled CFL [Fig. 6(a)]. The photopolymer spread to the substrate is irradiated with different UV doses to adjust the thickness of the polymer layer. At this step, the polymer layer’s top portion cures partially, forming an ultrathin liquid-phase polymer layer on the solidified bulk. An increase (decrease) in UV dose leads to a decrease (increase) in the liquid layer thickness. Then, the elastomeric nanocavity array and the partially cured polymer are contacted, generating a capillary force. Consequently, the capillary rise occurs until the liquid-phase layer has been solidified completely, enabling to control the height of nanoposts in the range of 5–8 nm. This mechanism shows significant potential for high-resolution, 2D patterned grayscale CFL due to the nature of light.

Fig. 6

Developments in capillary force lithography (CFL) methods. (a) Optically height-controlled CFL processes. A UV-curable polymer, NOA73, is precured by UV irradiation with different UV doses for each of the four division areas. This leads to thickness difference of the liquid-phase photopolymer layer (scale bar 1 µm). Reprinted with permission from Ref. [114], copyright 2020, American Chemical Society. (b) Continuous fabrication processes via two-step CFL. Polyurethane acrylate (PUA) resin is partially cured by UV irradiation, and top portion of the resin is completely cured with pressure.

2.3.2.

Two-step UV-assisted CFL

Micro/nanoscale hierarchical structures aid in the advancement of diverse photonic^[115,116], electronic/optoelectronic devices^[117,118] as multiscale structures improve structural functionalities compared to single-scale structures. Soft lithography approaches offer an efficient solution for the fabrication of high-quality micro/nanoscale hierarchical structures^[119,120].

Jeong et al. proposed a two-step UV-assisted CFL for fabricating monolithic micro/nanoscale structures over a large area ($\approx 5\mathrm{cm}\times 5\mathrm{cm}$)^[121]. They found that a partially cured microstructure can be further molded by sequential nanopatterning, which results in a monolithic hierarchical structure that enhances mechanical strength and chemical stability. As the fluidity of the upper part of the polymer must be adjusted well during the first molding, the appropriate curing time in the two-step UV-assisted CFL is an important factor. Later, for high fluidity, they described an improved version of CFL that involves partial curing and vacuum-assisted hydraulic filling for fabricating monolithic bridge structures^[122]. In small-scale structures, capillary forces are dominant, causing fluid flow along the wall of the stamp. However, for a partially cured polymer, capillary forces are not sufficiently high to facilitate polymer movement through the long channels connecting the structures because of the low fluidity of the polymer. They addressed this issue by moving the polymer into the channels while decreasing the hydraulic pressure through a vacuum, forming bridge structures.

Lee et al. reported continuous partial curing and tip-shaped modification approaches with microscale mushroom-shaped structures [Fig. 6(b)]^[123]. Typical methods with mushroom-shaped structures are limited owing to the failure of tips during demolding. To solve this problem, they used two-step UV-assisted CFL, which includes partial curing and tip widening. Also, by integrating with the R2R process, a continuous process is realized to increase productivity without any defects. In addition, the shape and size of the structures can be tuned by changing the geometries of the molds. This facile molding technique can serve as an alternative for fabricating dual-scale polymer micro/nanoscale structures. Consequently, the versatility of CFL, such as partial curing of two-step CFL, has broadened the range of applications to nanoscale structures of different shapes.

2.3.3.

Solvent-assisted CFL

A typical CFL process utilizes the capillary filling phenomenon of a molten polymer within the cavity of a stamp pattern at high temperatures to melt the polymer. Yu et al. proposed a solvent-assisted CFL^[124]. This method uses a solvent to soften the polymer at room temperature. The PDMS stamp functions as a solvent tank; when it comes into contact with the polymer, the stored solvent diffuses into the polymer, causing it to dissolve. Subsequently, the capillary force drives pattern formation. They tuned highly regular and reproducible micro/nanoscale polymer patterns with stamps by modulating the experimental conditions. Moreover, owing to the fast and large-area patterning ability, complex patterns also are easily fabricated using this method. In addition, this method enables the cross-patterning of polymer and organic devices without degradation.

2.3.4.

Applications of CFL

The versatility of CFL has facilitated its applications in optical research^[125,126]. In particular, remarkable features such as reusability of stamps, compatibility with various large-area substrates, and facile molding have realized complex structures. Using these structures, practical applications such as delivery systems^[127] and SERS^[128] have been demonstrated.

With the evolution of displays to ultrahigh definition (UHD), the number of pixels per inch has increased, requiring high technologies such as high density of pixels and high precision in fabrication. Conventional methods render the fabrication of μ-LEDs for compact UHD displays challenging because of their high fabrication cost/time and complicated processes. In particular, patterning processes with secondary operations, such as etching and deposition, cause system complexity. Here, we discuss CFL capable of achieving the controlled surface pattern with high precision and morphology of organic materials. Kim et al. fabricated micro-perovskite LEDs (μ-PeLEDs) for low-cost and large-area μ-LED applications, as perovskites are managed via simple processes at low temperatures [Fig. 7(a), i]^[129]. PeLEDs are suitable for display applications because they exhibit greater luminescence efficiency than conventional LEDs^[130]. However, due to the long diffusion length of excitons, PeLEDs have low external quantum efficiency (EQE). To solve this problem, the pixel size should be reduced by using a micro-patterning approach having fewer defects. They demonstrated that CFL reduces the pixel size on a 10 µm scale. They also realized μ-PeLEDs, which exhibit an EQE of 5.9% at 4 V by forming uniform pattern arrays and confirmed the performance of devices based on poly(ethylene oxide) concentrations [Fig. 7(a), ii and iii].

Fig. 7

Various applications of CFL. (a) (i) Perovskite patterning method using CFL. (ii) External quantum efficiency (EQE) spectra of micro-PeLEDs with 0.06%, 0.08%, 0.10%, and 0.12% (mass fractions) poly(ethylene oxide) (PEO). (iii) Light-emitting images of the micro-PeLEDs with 0.06% and 0.10% PEO. Reprinted with permission from Ref. [129], copyright 2022, Elsevier. (b) (i) Ag nanowire (AgNW) solution patterning using CFL. Hydrogel infiltrated at the edge of the PDMS stamp is cured on the patterned AgNWs by UV irradiation, and then washed off by ethanol. After ethanol washing, hydrogel-free AgNWs remain. (ii) Gauge factor as a function of applied tensile strain. (iii) Field emission SEM image of the patterned AgNWs on glass. Reprinted with permission from Ref. [131], copyright 2017, American Chemical Society.

As flexible strain sensors are used to monitor kinetic motions and act as on-skin electronics, high flexibility, response/recovery speed, and durability of sensors are important key factors. Also, an electrode material used on capacitive strain sensors should be chosen properly. Ag nanowire networks (AgNWs) are considered as one of the most promising materials for deformable electrodes. Lift-off lithography and laser printing are mainly used to pattern AgNWs; however, these methods have drawbacks such as limited patterning shapes and high production cost. To solve them, Kim et al. proposed patterned AgNWs using CFL as an alternative patterning technique^[131]. They progressed the CFL using the AgNW solution and embedded it in PDMS to complete the capacitive strain sensor [Fig. 7(b), i and iii]. AgNW solution is spin-coated on the glass substrate, and then annealed. After annealing and cooling, AgNWs are patterned using a PDMS stamp by CFL. The UV-curable hydrogel solution is dropped at the edge of the PDMS stamp, infiltrating space between the AgNW film and PDMS stamp. After a curing hydrogel, the cured hydrogel patterns on the AgNWs are washed off in the ethanol bath. Consequently, a pattern of hydrogel-free AgNWs remains. As a result of applying tensile force through the completed sensor, they proved that capacitance decreased as the tensile increased, showing a gauge factor of $-2.0$, which is larger than the gauge factor limit of one for the typical parallel-plate capacitive strain sensor [Fig. 7(b), ii]. The gauge factor is defined as GF = ΔC/C₀$\varepsilon$ ($C$, capacitance; $\varepsilon$, strain). They showed the development of the elastomeric substrate with elastic deformation as well as an improvement in electrical conductivity of AgNW anodes while maintaining high optical transparency.

Hydrogen gas sensors and hydrogen gas have been actively investigated as promising clean energy alternatives^[132,133]. Specifically, the absorption of hydrogen on Pd has been studied in terms of changes in mass, volume, and electrical/optical properties^[134,135]. These changes occur during the diffusion of hydrogen into Pd via the formation of a Pd hydride, which can be utilized in gas sensors to detect hydrogen. Pd-based sensors have been proven to exhibit high detection speed and high sensitivity^[136]. Zou et al. demonstrated that CFL in combination with a polymer “grafting to” approach facilitates derivation of linear assembles of Pd nanocubes for large-area sensors^[137]. In addition, the performance test of the sensor showed that the resistance increases at hydrogen gas concentrations between 0.3% and 3%, below which no response is observed, thereby demonstrating high sensitivity. This allows easy access to a self-assembly platform when fabricating large-area micropatterns. Moreover, these results indicate the possibility of fabricating nanodevices and commercializing the fabrication of devices using CFL.

3.2.1.

Fluid-assisted method

Fluid-assisted methods exploit the characteristics of fluids, such as surface tension, to induce capillary force^[175] and the Marangoni effect^[170,176]. Convective self-assembly involves a fluid-assisted process. Upon the movement of an upper blade coupled with the existence of a colloidal suspension, colloidal particles are self-assembled [Fig. 8(a)]^[177]. After a suspension is spread into the space between the blade and the substrate, the blade is withdrawn. Because the solvent evaporation induces a convective flow, the colloidal particles form arrays owing to the capillary force acting between them. For example, confined convective assembly has been demonstrated^[167]. The colloidal suspension is positioned between the substrate and the blade that are connected to a machine, which controls the lift-up rate. When the blade moves, a meniscus is formed at the edge. Hot air is blown to evaporate water. The well-arranged particles are then fixed on the glass plate. Consequently, varying the lift-up rates and colloidal suspension concentration results in the formation of 2D or 3D structures.

Fig. 8

Types of colloidal self-assembly techniques. (a) Fluid-assisted convective self-assembly. As the solvent evaporation induces a convective flow, the colloidal particles form arrays owing to the capillary force acting between them. (b) Colloidal self-assembly by using Langmuir–Blodgett method. By compressing the barrier, the monolayer is produced and then deposited on the vertically/horizontally positioned substrate. (c) Colloidal self-assembly by using Marangoni effect. Exposing a nanoparticle (NP)/ethanol (EH) suspension to a wet nitrogen (N₂) stream induces the evaporation of EH and the condensation of water (H₂O). This leads to the surface tension gradient [$\mathrm{EH}(\gamma =22\mathrm{mN}/\mathrm{m})$ and H₂O $(\gamma =72\mathrm{mN}/\mathrm{m})$], driving the NPs/H₂O toward the receding contact line. (d) Template-assisted self-assembly for plasmonic nanoarray. Photoresist is spread onto the antireflection coated (ARC) Si, and then imprinted to form a PDMS template. The template is used to align the colloidal plasmonic nanoparticles in a 2D square lattice. Right: SEM images of the fabricated master mold, PDMS nanohole arrays, and 2D plasmonic lattice. Reprinted with permission from Ref. [182], copyright 2019, American Chemistry Society.

Another fluid-assisted method is the LB method. An LB film, which is fabricated via the LB method, is an ultrathin film comprising arranged colloidal particles on a water surface [Fig. 8(b)]. Colloidal particles could form a densely packed thin film by compressing the barriers. When barrier compression occurs, a monolayer is produced and vertical/horizontal deposition occurs as the substrate is vertically/horizontally aligned. For example, well-arranged binary colloidal monolayers have been demonstrated^[168]. The first step involves placing the wafer under water at a small angle to the horizontal plane. Colloidal particles then form well-arranged arrays at the air–water interface using a diluted ethanol solution. Monolayer structures of various shapes have been fabricated depending on the diameter ratio of the colloidal particles and the potential of the hydrogen (pH) value. The pH affects the contact angle between particles of different sizes, and high crystallization is achieved. Subsequently, various self-assembled structures are transferred to the target substrate through surface lowering. In addition, the LB method can induce colloidal self-assembly on the curved substrate such as flexible foil, bulbs, and optical cables with a simple process^[169]. 600 nm silica nanoparticles are used as colloidal particles. Surface pressure depending on the area is investigated in isotherm conditions at each type of substrate. Also, the differences in the effective refractive index and transmission spectrum are observed between bare and PMMA composite films.

Although the LB method offers the advantage of manufacturing colloidal photonic crystals over a large area, it exhibits low mechanical instability. The Marangoni effect has been used to solve this problem^[178]. It uses the movement of a liquid from a place with a low surface tension to a place with a high surface tension according to changes in the surface tension (the force that causes the surface of the liquid to contract itself to occupy as small an area as possible), which induces self-assembly. Nanoparticles/ethanol suspension is spread into a thin layer on the glass substrate, and then exposed to a stream of wet ${\mathrm{N}}_2$, which induces evaporation of ethanol and condensation of water [Fig. 8(c)]^[179]. This leads to a concentration gradient and the difference between the surface tensions of ethanol and water, driving the nanoparticles/water toward the receding contact line. Also, a Marangoni flow of ethanol carrying nanoparticles is induced toward the contact line and into the water. As the condensed water dries, the ordered nanoparticle array forms on the substrate. For example, large-area self-assembled nanosphere structures have been demonstrated^[176]. The micro-propulsive injection method is used to create HCP structures. Subsequently, ethanol is added to the colloidal suspension to induce a surface tension gradient. When colloidal suspension is dropped on the air–fluid interface, surface tension induces the Marangoni force, and colloidal particles move until the entire surface area is covered by nanoparticle arrays. Consequently, high throughput is achieved, and the rapid fabrication rate and simple process facilitate its application to a large area and in various substrates such as rigid, flexible, and complex structures. In addition, a way to enhance the quality of colloidal self-assembly is demonstrated^[180]. The conventional method uses sodium dodecyl sulfate (SDS) after emulsion is introduced at the air–water interface. SDS induces a lower surface tension gradient to reduce the defects of the structure. In contrast, the proposed method uses SDS pre-saturating water before introducing colloidal solution. Through the proposed method, a fabricated structure with fewer defects is observed.

3.2.2.

Template-assisted method

Tailored substrates are fabricated via various lithography technologies. When colloidal particles slip on a tailored substrate, they are located on the patterned region by physical force. The tailored substrate used for this purpose is a template. Template-assisted methods contribute to further utilization of colloidal self-assembly by increasing design diversity. For example, template-assisted methods using physical and capillary forces have been demonstrated^[181]. First, small holes are patterned via photolithography. Subsequently, a polymer bead dispersion is added between the top and bottom of the glass substrate. When the rear end of the bead dispersion is moved, capillary force is applied to the beads, and these move continuously unless confined in the holes. Various structures, such as monomers, dimers, triangles, squares, pentagons, hexagons, and double-layered structures, are achieved by controlling the ratio of the patterned hole diameter, depth, and bead diameter.

Template-assisted methods for plasmonic nanostructures have also been demonstrated [Fig. 8(d)]^[182]. The master mold and soft mold are fabricated based on laser interference lithography and UV-curable PDMS, respectively. The soft mold is used as the template. The colloidal Au nanoparticles are confined to the template by a guiding plate. The soft mold is used to align the colloidal plasmonic particles in a 2D square lattice. Consequently, the surface lattice resonance and localized surface plasmon resonance (LSPR) wavelengths are blueshifted when the strain increases (0% to 40%).

3.2.3.

Field-assisted method

External electric or magnetic fields induce colloidal self-assembly by electromagnetic force^[183–185]. During self-assembly, a uniform driving force minimizes defects. Although this method requires additional equipment setups, well-arranged self-assembly structures are formed with a uniform driving force. In the case of the electric-field-assisted method, 2D tunable self-assembled structures have been demonstrated^[174]. Eight planar electrodes with the same height are used. For a uniform electric field, the gap between the electrodes and the substrate is controlled by varying the substrate thickness. Further, in the case of the magnetic-field-assisted method, self-assembly using the non-magnetic template method has been demonstrated^[185]. The existing magnetic field-assisted method is not able to facilitate the arrangement of particles in the desired shape owing to the magnetic properties of the target substrate. This problem has been solved via the use of ferrofluids. The target patterned substrate is immersed in ferrofluid and modulated by magnetic-field-induced electromagnetic force.

3.2.4.

3D structure method

Colloidal self-assembly also induces the formation of multiple layers or other complex structures. The layer-by-layer scooping transfer method has been demonstrated for multilayer structures^[186]. A monolayer is formed at the air–water interface by floating colloidal particles. Consequently, a multilayer is realized by repeating the monolayer transfer process onto the target substrate. Similarly, multilayer structures with different particles have also been demonstrated^[187]. The monolayer is first self-assembled on a hydrophilic petri dish. Following toluene vapor annealing, a mechanically stable monolayer is formed. If the same process is repeated for different particles, multilayer structures are formed, realizing complex shapes. For example, specific directional bonding structures have been demonstrated^[188]. The structures are fabricated with deoxyribonucleic acid (DNA)-patched particles using a previously described method^[158]. Patched DNA docks with other patched particles, and designed structures are formed. Subsequently, various bonding shapes, such as hybridized atomic orbitals, have been achieved.

4.2.1.

Pre-processing method

Pre-processing involves additional steps before BCP self-assembly. Pre-synthesized particles are used to utilize the properties of materials that are not prevalent in BCP. The pre-synthesized particles are added to the BCP solution, and the same process with pure BCP self-assembly is followed [Fig. 11(a), i]. In addition, it is possible to selectively choose a polymer domain that interacts with pre-synthesized particles. Owing to entropic interactions, which are related to block selectivity, and relative particle size, which is related to particle positioning, various types of metal and dielectric particles are combined with BCP. Particle-combined BCP self-assembly structures are achieved when the structures are thermodynamically stable^[237]. Consequently, various structures are fabricated by modulating the length of the polymer and the variety and size of inorganic materials. For example, structures with different pre-synthesized nanoparticles have been demonstrated by using BCP self-assembly^[238]. Pre-synthesized Ag particles and poly(styrene-blockvinylpyridine) (PS-b-PVP) have been used. The Ag nanoparticles selectively interact with the PS matrix. Subsequently, pre-synthesized Au, platinum, or Pd particles are added to the PVP matrix. Finally, two different nanoparticles with localized BCP self-assembled structures are formed. In addition, depending on the molecular weight of BCP, a different coated pre-synthesized nanoparticle structure is observed^[239]. This proposed method uses 7 nm PbS nanoparticles as pre-synthesized particles.

Fig. 11

Various methods for patterning substrate with BCP self-assembly. (a) BCP pre-processing method. (i) Using pre-synthesized particles to combine BCP and inorganic particles. (ii) Self-assembled BCP on pre-patterned substrate. The substrates with particular chemical patterns lead BCP to perfectly replicate the pattern. Reprinted with permission from Ref. [246], copyright 2018, John Wiley and Sons. Right-top: SEM images of angled lamellae using chemical surface patterns. Reprinted with permission from Ref. [240], copyright 2005, American Association for the Advancement of Science. The topographic substrate can induce a perfect alignment of lamellae. Reprinted with permission from Ref. [244], copyright 2008, John Wiley and Sons. Right-bottom: SEM images of (left-top) spherical BCPs aligned along the rows of spheres, which are parallel to the confining edges of the topography; reprinted with permission from Ref. [243], copyright 2002, AIP Publishing; (right-top) parallel cylinders aligned along the edges of the topography, (left-bottom) perpendicular cylinders aligned along the packed rows, which are parallel to the confining edges of the topography; reprinted with permission from Ref. [241], copyright 2004, IEEE; (right-bottom) lamellar domains oriented perpendicular to both the trench bottom and the sidewalls; reprinted with permission from Ref. [242], copyright 2007, John Wiley and Sons. (b) BCP post-processing method. (i) Deposition of gold by sputtering on the sidewall of BCP self-assembly structure. Reprinted with permission from Ref. [247], copyright 2016, John Wiley and Sons. (ii) Process of sequential infiltration synthesis (SIS) method. Certain materials selectively deposited by SIS remain after selective etching.

Another pre-processing method is using a pre-patterned substrate. Various pre-patterned substrates are prepared to enhance design variety of BCP structures^[240–244]. For example, the topologically defined silicon substrate method has been demonstrated^[245]. Substrates with various shapes, such as rectangular channels, triangular trenches, and circles, have been fabricated via EBL. BCP is then spin-coated onto the prepared substrates. When these templates are immersed in a mixed solution, various Pt nanostructures are fabricated. In addition to patterned substrates, other treated substrates can induce different self-assembled structures [Fig. 11(a), ii]^[246]. Depending on the types of pre-patterned substrates such as a chemical pattern, neutral surface, and topographic pattern, various self-assembled BCP structures are observed.

4.2.2.

Post-processing method

The post-processing method involves additional steps after BCP self-assembly. Inorganic materials have been exploited with BCP using a post-processing method. The simplest method is the deposition method. For example, secondary sputtering has been demonstrated. Au is deposited on the substrate via evaporation. PS-b-PMMA is spin-coated onto the Au-deposited substrate. After annealing, the PMMA domain is selectively etched by RIE. Thereafter, Ar-ion sputtering is used to deposit pre-deposited Au on the PS polymer sidewall. After removing the PS polymer domain, a high-aspect-ratio Au nanostructure is formed [Fig. 11(b), i]^[247].

Another post-processing method is sequential infiltration synthesis (SIS). The SIS method uses atomic layer deposition to fabricate thin films by selectively reacting with certain polymer domains [Fig. 11(b), ii]. For example, sub-10 nm patterning based on SIS has been demonstrated^[248]. A BCP pre-patterned substrate is prepared using EBL. Subsequently, another BCP layer is spin-coated on this substrate. The topcoat is then deposited on the upper BCP via initiated chemical vapor deposition (iCVD). Consequently, a quarter of the pre-patterned periodicity ${\mathrm{Al}}_2{\mathrm{O}}_3$ sub-10 nm structure is fabricated after RIE. Owing to the scalability of iCVD, sub-10 nm scale patterns can be fabricated over a large area.