A review of roll-to-roll nanoimprint lithography
© Kooy et al.; licensee Springer. 2014
Received: 7 April 2014
Accepted: 14 June 2014
Published: 25 June 2014
Skip to main content
© Kooy et al.; licensee Springer. 2014
Received: 7 April 2014
Accepted: 14 June 2014
Published: 25 June 2014
Since its introduction in 1995, nanoimprint lithography has been demonstrated in many researches as a simple, low-cost, and high-throughput process for replicating micro- and nanoscale patterns. Due to its advantages, the nanoimprint lithography method has been rapidly developed over the years as a promising alternative to conventional nanolithography processes to fulfill the demands generated from the recent developments in the semiconductor and flexible electronics industries, which results in variations of the process. Roll-to-roll (R2R) nanoimprint lithography (NIL) is the most demanded technique due to its high-throughput fulfilling industrial-scale application. In the present work, a general literature review on the various types of nanoimprint lithography processes especially R2R NIL and the methods commonly adapted to fabricate imprint molds are presented to provide a clear view and understanding on the nanoimprint lithography technique as well as its recent developments.
Recent developments in semiconductor and flexible electronics applications have observed a rapid increase in demands for lower cost, higher throughput, and higher resolution micro/nanofabrication techniques. This is due to the fact that conventional techniques such as electron beam lithography (EBL) have a low throughput  for mass production and other alternatives such as extreme ultraviolet lithography and focused ion beam lithography are very costly, limiting the technology only to large organizations .
Nanoimprint lithography (NIL) was introduced by Prof. S.Y. Chou and the team in 1995  as a simpler, low-cost, and high-throughput alternative to micro- and nanofabrication. In the NIL process, a prefabricated mold containing an inverse of the desired patterns is pressed onto a resist-coated substrate to replicate the patterns via mechanical deformation. Hence, many replications may be produced from a single prefabricated mold using this method. As the NIL process is based on direct mechanical deformation, its resolution is not constrained to the limitations of light diffraction or beam scattering factors as observed in conventional nanolithography methods . In terms of patterning capability, various 2D and 3D structures  with feature sizes ranging from several micrometers [6, 7] down to sub-50-nm scale [8–10] have been demonstrated. Due to its promising potential, the NIL process has been added into the International Technology Roadmap for Semiconductors (ITRS) for 32- and 22-nm nodes  and has been widely researched and improvised by many researchers ever since, resulting in several variations of the process.
In contrary to the thermal NIL process, the UV NIL process involves imprinting onto a layer of liquid photopolymer resist and curing using UV exposure, which causes resist hardening due to cross-linking in the polymer instead of manipulating the phase change via resist temperature . The remaining imprint mechanism, however, is similar to the thermal NIL process. A typical UV NIL process is also illustrated in Figure 1 for comparison purposes. The UV NIL process has several prominent advantages over the thermal NIL process, which include the capability of UV NIL to be conducted at room temperature without the need of elevated temperature imprinting [5, 14], which helps eliminate the issues resulting from thermal expansion variations between the mold, substrate, and resist. In addition, the imprinting process involves a less viscous liquid photoresist, which allows the process to be conducted at a lower imprint pressure compared to thermal NIL processes [11, 14–16]. The lower viscosity of the resist also allows the resist to fill in the mold cavity in a shorter time, and the elimination of the temperature cycle also improves the process throughput .
In addition, Chou and the team  also introduced the usage of a single XeCl excimer laser pulse to melt a thin layer up to 300 nm of the silicon substrate surface, where the molten silicon layer will then be imprinted using the mold. This NIL process is named laser-assisted direct imprint (LADI). Similar to thermal NIL in concept, the molten silicon layer will fill in the mold cavity under suitable imprinting pressure, transferring the patterns to the silicon substrate. The embossing time is reported to be less than 250 ns. A similar concept is also observed in , where a CO2 infrared laser is used to soften a thermoplastic resist for the NIL process.
Imprint forces used in P2P NIL processes from research publications for several different imprint areas
Additionally, air bubble entrapment issues are also commonly observed in P2P NIL, particularly in large-area, single-step processes [21, 26] as air is easily trapped in the gaps between resist and mold cavities, resulting in defects on the imprinted structures. The risk of defects is increased when the mold contains depressions or when the resist is deposited as droplets rather than spin-coated, which allows air to be trapped easily , which results in the need to conduct the imprinting process under vacuum to prevent trapping of air bubbles as observed in [5, 8, 21]. However, vacuum or reduced atmosphere chambers are difficult to be implemented in a system with a continuous web feed. Hiroshima and the team had been working on this matter and introduced the usage of pentafluoropropane as ambient to solve the bubble defect problem [27–29].
Alternatively, in multiple-step imprinting, smaller wafer sizes are used to pattern over a larger area in the form of a matrix (also known as SSIL) as observed in the work of Haatainen and the team [30, 31], which reduces both the required force and air bubble issue observed in a single-step imprinting. However, such system is typically more complicated as it requires highly accurate mold alignment during imprinting.
On the contrary, in R2P NIL, a roller press mechanism is used to provide the imprinting force onto a rigid surface as shown previously in Figure 3. Since a roller press mechanism is utilized in roller-based NIL, the actual contact area during imprinting is only a line along the roller in contact with the substrate rather than the entire stamp area in P2P NIL. This very much reduces the required imprinting force in the NIL process [32, 33], which may go as low as 200 N to achieve an imprinting pressure of approximately 1 bar for an imprinting width of 300 mm . Additionally, due to the line contact, the roller-based NIL process has the advantage of reduced issues regarding trapped air bubbles, thickness variation, and dust pollutants, which also greatly improve its replication uniformity [34, 35].
As for the R2R NIL process, an imprint roller with a patterned surface (or wrapped with a flexible mold) is used to imprint onto a flexible substrate on a supporting roller instead of a flat plate in R2P NIL processes. The entire process is based on the roll-to-roll manufacturing concept, which has the advantages of continuous process and high throughput [39, 40] and, hence, provides a highly promising solution for industrial-scale applications. While R2P methods have great advantages over conventional P2P NIL in terms of imprint force, throughput, and size of equipment, it still has several limitations in realizing a continuous imprinting process . Even though studies have been conducted to allow continuous imprinting in R2P systems as observed in [36, 37], the throughput of the process remains lower in R2P NIL since time is needed to lift and return the imprint roller in position. This also requires an additional high-precision linear drive system for positioning and alignment, which makes it less favorable compared to R2R NIL.
Despite the advantages, it is noted that there are several challenges in realizing the continuous R2R NIL process. One of the main challenges is the fabrication of the special flexible mold, which will be discussed in further sections. In addition, an integrated continuous resist coating mechanism is also required in a continuous R2R NIL process as the substrate is continuously being fed for imprinting. This poses a challenge as it would require a more complicated mechanism and uniformity control  as compared to spin coating, which is much simpler and has been used in almost all studies on P2P and some non-continuous R2P systems [14, 18, 21–25, 35, 48–50]. Selection of resist material is also important as it needs to have good coating properties and low viscosity [4, 40]. The issue, however, is not observed in studies involving direct imprinting onto a polymer substrate , although such method tends to require higher imprinting force and elevated temperature as compared to their UV-based counterparts.
Compared to P2P NIL, the mold separation at the end of the imprinting process requires less force. However, in the study of Dumond and the team , R2R NIL demolds with the parts and imprint mold moving in circular motion. This relative movement can cause a collision and damage the parts in the process. More attention should be paid when designing the microstructure for the R2R NIL process. In recent development of the R2R nanoimprint lithography device, the separation of the cured resin from the mold is generally assisted by a deflection roller and a certain amount of web tension.
One of the most important key items in the nanoimprint lithography process is the imprint mold or stamp, which contains the inverse of the desired patterns on the imprinted output. Ever since NIL's introduction in 1995, the performance of the NIL process in terms of resolution and feature size is determined primarily by the mold as the resist is shaped according to the mold cavity via direct mechanical contact [3, 11]. As the patterns are transferred from the mold to imprint at 1× scale (feature sizes of imprint and mold are the same) in the NIL process, the fabrication of the mold tends to be difficult as the feature sizes go down to lower ranges of nanometer scale [11, 26]. As a result, the fabrication of the NIL molds remains as one of the critical bottleneck factors in further development of the NIL process, particularly the roller-based variants [37, 40, 53]. Additionally, the material selection for the NIL molds is also crucial in overcoming critical issues such as the well-known mold sticking issue and thermal expansion mismatch issue (for thermal NIL processes) as well as to prolong its lifespan [4, 9, 40].
For P2P and R2P (using a flat mold) NIL processes, the micro/nanostructures are normally patterned onto rigid substrates such as silicon or quartz using conventional techniques (i.e., EBL) [3, 21, 22, 48] or even nanoimprint lithography , where the patterns are then etched into the substrate using reactive ion etching (RIE) to be used as a flat mold in the NIL process. Other techniques such as focused ion beam (FIB) was also explored by Taniguchi and the team  to fabricate molds for the NIL process, which was reported to be suitable for speedy fabrication of 3D molds with a depth resolution down to 10 nm. To prevent the sticking issues from occurring during imprinting, the surface of the mold is usually coated with a thin layer of anti-stick coating such as fluorinated silanes [21, 55] or polybenzoxazine . In some studies, the patterned resist layer is used directly as the mold surface (with or without anti-stick coating) without etching process as observed in the works of Mohamed  and Ishii and Taniguchi .
Alternatively, a flat mold may also be conducted using a soft mold, where a polymer imprint replica of the master mold is used as the mold for the imprinting process as observed in the work of Plachetka et al.  and Ye et al. . The imprint replica is usually made using a polymer cast molding technique, where the process is as follows: First, the solution of a polymer with low surface energy such as PDMS is poured onto the patterned master and then spin-coated to achieve a uniform and the desired thickness. The PDMS-coated master is then put in the vacuum for several hours to release the trapped air bubbles to allow complete filling of cavities, before being cured at an elevated temperature (120°C for 15 min for Sylgard® 184 PDMS ) and peeled off to be used as the soft mold. Soft mold imprinting provides a simple and good alternative to the conventional wafer imprinting as multiple copies of the soft mold are easily produced using a simple and low-cost method , besides the fact that the low surface energy of PDMS allowed it to be used directly for imprinting without the need for anti-stick layers [16, 58].
However, unlike P2P and R2P NIL processes which utilize a flat mold, continuous R2R and R2P (using a roller mold) NIL processes require a roller mold for imprinting. Out of all the available fabrication techniques, a flexible mold is generally used in the application of a roller mold. It is a challenge [35–37] as the mold has to be flexible enough to be bent and wrapped around the imprint roller yet has sufficient strength and modulus to imprint onto the polymer resists [40, 60]. The mold should also have good wear resistance properties as it will be used to imprint polymer resists over a large number of cycles in repetition. Hence, material selection is important as its properties determine the above requirements as well as several issues commonly observed in NIL processes.
Metallic layers (i.e., nickel) and silicone-based polymer castings (i.e., PDMS) are commonly used due to their flexibility. Silicone-based molds usually have sufficient modulus to imprint onto liquid resists in UV NIL processes [15, 16, 61], whereas thermal NIL imprinting, which requires higher mold modulus, usually utilizes metal-based molds such as nickel [32, 42, 45]. In addition, the mold material should also have low surface energy to ensure that the resist does not adhere to the mold surface during the separation process which will result in defects and mold damage. Low surface energy also reduces friction and ensures a clean de-molding process, which also helps improve its life cycle . Nevertheless, polymers such as ethylene tetrafluoroethylene (ETFE)  and PDMS [15, 26, 35] are commonly used as a flexible mold as an alternative to nickel due to their low surface energy (15.6 and 19.6 dyn/cm, respectively ) and ease of fabrication as compared to metal molds . However, according to Odom and the team from Harvard University, the low elastic modulus of PDMS mold will lead to feature deformation of the transferred patterns due to high loading imprint force .
Most of the other studies, however, use a simpler approach for fabrication of flexible molds for the R2R and R2P NIL processes, where a replica of a master mold is used as the flexible mold for the roller imprint process. In general, the desired structures are first patterned onto a silicon or quartz substrate using conventional nanolithography techniques such as EBL and followed by the RIE process, similar to its P2P variant. The replication of the master mold can then be conducted using several methods. One of the common techniques involves deposition of an anti-stick layer onto the master mold, followed by a layer of metal such as nickel directly onto the master mold, where it will then be peeled off to be used as a flexible mold in the roller nanoimprint process as observed in [32, 43, 46]. In some cases such as in , an imprint replica of the master mold is first obtained using nanoimprint lithography (step-and-repeat technique) onto a resist-coated wafer, where a nickel layer is then deposited onto the imprint and peeled off to be used as the flexible mold in the imprint process published in .
Alternatively, the imprint replica of the master mold may also be produced via the polymer cast molding technique using non-sticking polymers such as PDMS or ETFE to be used as the flexible soft mold for the imprint process as observed in the work of a few research groups [7, 15, 35]. It is highlighted in the work of Ye et al.  that polymer cast molds (typically made of PDMS) are usually more preferable in the UV-based roller imprinting process due to their advantages of being low cost, low surface energy (fewer sticking issues), chemically inert, elastic, and simpler to produce as compared to metal molds.
One of the important challenges of producing roller molds is the surface planarity of the attached flexible mold . A similar uniformity is needed to achieve imprint rollers in order to prevent transmission of low-frequency and long-range surface waviness onto the replicated pattern.
Since its introduction back in 1995, the rapid development of the nanoimprint lithography process has resulted in a number of variants in the process, which can be categorized based on its two main operation features: resist curing and type of imprint contact. To date, in terms of resist curing, there are two fundamental types of processes: thermal NIL and ultraviolet (UV) NIL. As for the types of imprint contact, the process can be categorized into three common types: plate-to-plate (P2P) NIL, roll-to-plate (R2P) NIL, and roll-to-roll (R2R) NIL.
From literature, roller-based NIL processes, particularly the R2R NIL process, show a highly promising future to be implemented as a full-scale production process due to their high throughput and wide-area patterning capability as observed in several research works. However, detailed information on R2R NIL, particularly regarding process and stability control, is still limited as there are still many challenges and issues to be solved in the R2R NIL process. Nevertheless, further extensive and thorough studies on the process are crucial to solve these challenges to realize the implementation of R2R NIL for commercial applications in the near future.
The authors would like to thank Universiti Sains Malaysia for funding this research work through the USM Delivering Excellence (DE2012) Grant.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.