Thermal nanoimprint lithography (NIL) [1
] is a high-throughput and low-cost soft lithography technique by which a surface pattern on a typically hard mold is physically imprinted into a thermoplastic material, which is often a polymer. The polymer can be deposited as thin film on a substrate or used as a free-standing thicker foil (in this last case, the process was originally introduced as “hot-embossing”. However, very often NIL and hot-embossing are used in the literature with the same meaning when the replicated features are of sub-micrometric dimension. In this paper, we use the term NIL to refer to both cases). The process is based on heating the material above its glass transition temperature (Tg
) and making it flow into the mold cavities by applying an adequate pressure. After cooling down below Tg
, the pressure is released to complete the replica process.
Usually, high-resolution and large-area NIL molds are expensive and difficult to fabricate. Even though they are typically made of silicon or other hard materials like nickel or quartz, after a number of imprinting cycles, they start cracking and, therefore, become unusable. In order to preserve the mold without affecting the throughput, intermediate molds were introduced. Intermediate molds are replicas of the original mold that are themselves used as molds to transfer the topographies to the final material. This interest is also present at an industrial level. For example, the company Obducat AB has recently patented [3
] an imprinting apparatus to perform a two-step process involving intermediates for typical topographies ranging from gratings of 80 nm linewidth up to micrometric pillars.
Intermediate molds are generally produced in plastic or soft materials by soft-lithography techniques and must guarantee high-fidelity copies and a sufficiently high number of processes before undergoing degradation [4
]. While polymers such as poly-(methyl-methacrylate) (PMMA) or polystyrene (PS) are frequently used in thermal NIL, they are not recommended as material for intermediate molds because they are generally sticky and tend to crack during the release step [5
]. Differently, beyond the ease of fabrication, elastomeric materials can ensure good elastic adaptation and conformal contact with the substrate, which leads to intimate contact without voids [6
Polydimethylsiloxane (PDMS) is widely used as soft-mold material because of its low surface energy and mechanical properties, which allow conformal contact and easy release from the initial mold and the final imprinted film. However, the low Young’s modulus of PDMS often limits the replica process if the topographies are very small (i.e., hundreds of nm or less) and with high spatial density. Furthermore, PDMS often degrades after a few cycles of patterning [5
]. Since their high stiffness, high temperature resistance, and unique anti-adhesive properties, fluorinated polymers have optimal properties for soft plastic molds fabrication. Barbero and co-workers [8
] demonstrated that ethylene(tetrafluoroethylene) (ETFE) can sustain multiple embossing (at T = 160 °C and P ≅ 1 bar) to transfer gratings (period = 2 µm, and ridge width = 700 nm) to PMMA films. However, owing to the typical dimension of its crystallization domains (100 nm) [9
], ETFE allowed a maximal resolution of ≈150 nm. In the paper of Greer coworkers [10
fluorinated ethylene–propylene (FEP) was exploited as material for soft-nanoimprint lithography. They could transfer 45 nm-diameter pits to the nano-imprint specific resist mr-NIL210 XP and to the photoresist Nano SU-8 3005, but the low FEP glass transition temperature (80 °C) greatly limited the number of compatible thermoplastic materials.
Perfluoropolyether (PFPE)-based elastomers are a unique class of fluorinated polymers whose structure is formed by linear chains based on multiple strong carbon fluorine bonds, which entail high stiffness and temperature resistance. Prior curing fluoropolymers are viscous liquids at room temperature and are characterized by a very low surface energy. This facilitates the filling of nanoscale cavities and guarantees an anti-adhesive behavior [11
]. PFPEs are inert and exhibit high durability and toughness, high gas permeability, and low toxicity [12
], with additional features of chemical and thermal stability. These characteristics minimize degradation under use and provide good lubricity, which reduces the contact surface wear. PFPE-based elastomers are, therefore, promising for NIL but only very rarely used to this end. In Reference [11
], PFPE was successfully tested against silicon and PDMS as mold for thermal nanoimprinting of polycarbonate (PC) sheets. More specifically, the process was performed at T = 170 °C and P = 5 bar for transferring gratings with period of 1 μm (250 nm depth), 400 nm (330 nm depth), and 300 nm (330 nm depth). They also stated that, owing to its gas permeability, PFPE molds prevent air-trapping issues, which allowed for the use of lower imprint pressures [14
]. However, a long-time process (≅30 min) at a rather high temperature (20° more than PC Tg
= 150 °C) were required to fabricate a high-fidelity replica. To the best of our knowledge, only in a short conference proceeding PFPE was exploited to fabricate an intermediate mold, but as a film on a silicon substrate. The patterning was performed by casting and thermal-curing [15
]. In this paper, the initial mold was a silicon substrate with a 200-nm-period grating (with 120-nm-depth) on its surface and the final replicas were obtained in poly-(vinyl phenyl ketone) (PVPK) by thermal NIL (T = 90 °C and P = 70 bar). The authors found that PFPE molds could successfully transfer the grating pattern and also replicate the roughness present along the ridge edges. For this reason, they speculated that PFPE could in principle be suitable for features with dimension far beyond 100 nm. However, the intermediate mold fabrication process was rather long (1–3 h of annealing) and led to very poor adhesion of the fluoropolymer to the silicon support [15
]. This last issue might compromise the final NIL step.
In this paper, we propose and characterize by using a scanning electron microscopy and an atomic force microscopy through a two-step process based on the use of a free-standing patterned film of perfluoropolyether (PFPE) as an intermediate mold to replicate sub-micrometer features from a silicon mold to the final thermoplastic polymer. We compare PFPE elastomeric molds with more standard molds made of PDMS, which demonstrates better resolution and fidelity of the replica process. Lastly, we test PFPE molds for transferring isolated grooves and ridges with sub-100-nm lateral dimension.
Molds with sub-100-nm resolution were obtained in PMMA by electron-beam lithography (EBL) starting from commercial p-doped silicon wafers (SYLTRONIX, Archamps, France). Each mold was initially processed by EBL to generate arrays of ridges and grooves of 100 mm2
area into a 50-nm-thick PMMA film. PMMA was spun over a 5 nm-thick titanium layer and was previously deposited on the silicon wafer by thermal evaporation. After cleaning with nitrogen flow, the molds were systematically characterized by optical microscopy (Carl Zeiss Microscopy, Jena, Germany) and atomic-force microscopy (Veeco Instruments Inc., Santa Barbara, CA, USA). The mold with 600-nm-period grating was fabricated by using laser interference lithography (LIL). SPR220 (Microposit, Shipley European Limited, Shipley, UK) was spun onto a silicon wafer with a spin speed of 4000 rpm for 30 s. The sample was exposed to a 50 mW helium cadmium (HeCd) laser, which emitted a TEM00
single mode at a 325 nm light source with a beam incidence angle of 165.7° and an exposure dose of 77 mJ/cm. Resist developing was performed by immersing the sample in an MF319/Milli-Q water (10:1) solution for 15 s [16
Polydimethylsiloxane (PDMS) intermediate mold fabrication
. The precursor PDMS polymer (SYLGARD 184, The Dow Chemical Company, Pittsburg, CA, USA) was mixed with its curing agent at a ratio of 10:1 and poured onto the nanostructured mold. The uncured replica was then left resting for 10 min to reduce surface inhomogeneities. Then it was baked in an oven for 2 h at 80 °C. After thermal curing, the replica was gently removed from the master using scalpel and tweezers [17
PerFluoroPolyEther (PFPE) intermediate mold fabrication
. PFPE resin (FLUOROLINK®
MD 700, Solvay Speciality Polymers, Bollate, Italy) was mixed with 3% w
photoinitiator Darocure 1173 (C10
, 405655 Sigma Aldrich, Milano, Italy), poured on top of the PMMA surfaces, and crosslinked by UV-light (365 nm, 25 mW·cm−2
). The exposure was performed in two steps, which was reported in References [18
]. Briefly, the samples were kept for 180 s in nitrogen atmosphere and then were kept for 60 s in air. After curing, the PFPE films were easily peeled off and cleaned with nitrogen flow.
Cyclic Olefin Copolymer (COC) nanoimprinting
. COC foils (thickness 140 µm, Microfluidic ChipShop GmbH, Jena, Germany) were imprinted using an Obducat Nanoimprint 24 system (Obducat, Lund, Sweden) with the PFPE intermediate molds. After cleaning with 2-propanol, the COC substrates were placed on top of the molds and softened by raising the temperature up to 150 °C. A pressure of 50 bar was then applied for 300 s before cooling down to 70 °C, i.e., below the glass transition temperature of the copolymer (Tg
= 134 °C). Lastly, the pressure was released and the mold was detached from the imprinted COC with a scalpel [20
Scanning Electron Microscopy. Molds, intermediate molds, and final replicas were analyzed with a LEO 1525 field emission scanning electron microscope. In order to enhance the topography of substrate surfaces, image acquisition was carried out by using an Everheart-Thornley detector.
Contact Angle Measurements. Substrate wettability was evaluated by contact angle measurements acquired with a CAM 200 instrument (KSV Instruments, Helsinki, Finland). A deionized water drop was deposited on top of each substrate through a micro-syringe. All these measurements were performed in air at room temperature. Data are reported as mean ± SD.
Atomic Force Microscopy
. Sample topographies were characterized by an atomic force microscope (Veeco Innova Scanning Probe Microscope, Veeco Instruments Inc., Santa Barbara, CA, USA), operating in tapping mode. The scan frequency was set at 0.977 Hz and the scanning areas were 5 × 5 μm2.
At least three areas were analyzed per sample (512 point/line each). At least three PFPE intermediate mold and COC replicas were imaged for each topography type. A silicon nitride tip with a nominal spring constant in the range of 0.2–0.8 N/m and a resonant frequency of 45–95 kHz was used. All the measurements were performed in air at room temperature and raw scan data were leveled by surface subtraction to remove possible sample tilts. Data were analyzed by the Gwiddion software (Gwiddion 2.47 version, Brno, Czech Republic, “Profile” tool) and reported as mean ± SD. Full Width Half Maximum values were measured from the AFM profiles by the “Analysis: Peaks and Baseline: Multiple Peak Fit- Gaussian fit” tool of the software Origin (OriginLab Corporation, Northampton, MA, USA https://www.originlab.com
, version 9.0).
Statistical analysis. Data are reported as average values ± the standard deviation (mean ± SD). Data were statistically analyzed by the GraphPad PRISM 6.1 program (GraphPad Software, San Diego, CA, USA). Student’ t-test (unpaired) analysis was used to compare distributions. The mean values obtained in each repeated experiment were assumed to be normally distributed regarding the true mean. Statistical significance refers to results where p < 0.05.
In conclusion, we have introduced and characterized an innovative two-step thermal NIL process based on the use of intermediate molds made of PFPE to replicate sub-100 nm features from a silicon mold to a final thermoplastic material (COC).
PFPE elastomeric molds were compared with molds made of the more standard PDMS, which demonstrated better resolution and fidelity of the replica process. More specifically, we showed that, in case of 600-nm-period nanogratings, PDMS could not successfully reproduce the topography likely because of its rather low Young’s modulus. On the contrary, the more rigid PFPE allowed the nanograting to be successfully copied.
Lastly, in order to characterize the fidelity of the complete two-step transfer protocol, we used sub-100-nm patterns specifically designed to test the process with extreme geometries. We found that isolated lines of lateral dimension of ≈80 nm (in case of ridges) and ≈60 nm (in case of grooves) and with aspect ratio = 1 can be considered at the moment as the minimum feature size that can be transferred to the thermoplastics.
Given our results and the possibility to increase the rigidity of PFPEs by adding chemical groups or tailoring the PFPE/crosslinker mix, we believe that sub-50 nm resolution and more would be possible. However, more investigations with different topographies such as nanoposts, nanoholes, aspect ratios, and spatial densities are required to fully unravel the real potential of PFPEs as mold or intermediate mold material for thermal NIL.