The role of hydrophobic silane coating on Si stamps in nanoimprint lithography

Abstract

Hydrophobic silane coatings have been successfully applied to the surface of Si stamps to improve demolding in nanoimprint lithography (NIL). However, the role of the silane coating has only been studied either indirectly, by measuring adhesion or friction coefficients for Si and substrate surfaces without patterns, or collectively, by measuring the overall demolding force that does not differentiate contributions of friction dissipation, stored elastic energy, and adhesion. Here, for the first time, we present experimental evidence on the role of the silane coating in improving demolding in UV-NIL by using different silane coatings. The silane coatings were characterized by x-ray photoelectron spectroscopy, water contact angle, and friction force measurements. Then, the work of demolding was systematically measured for different silane coatings using stamps with the same micropattern but different pattern depths. Comparison of the results to the theoretical model developed for fiber-matrix debonding energy by Sutcu and Hillig [Acta Metall. Mater. 38(12), 2653–2662] indicated that with a hydrophobic silane coating, the main parameter contributing to overall demolding work shifts from adhesion to stored elastic energy and frictional dissipation as surface adhesion keeps decreasing. The results confirm that the main role of the silane coating in reducing the demolding is to reduce surface adhesion rather than friction at the stamp/substrate interface.

I. INTRODUCTION

One of the most critical process steps in nanoimprint lithography (NIL) is demolding, the process to separate the NIL stamp from the molded substrate, where most imprint defects or structural failure is generated. The main reason for the generation of imprint defects or structural failure is that most NIL resists are based on adhesive materials such as epoxy¹ and acryl.² A number of efforts have been devoted to understanding and improving the demolding process in NIL. These efforts include the development of new imprint resists mostly containing fluorinated compounds to lower the surface energy and therefore impart good anti-sticking properties,^2–6 use of low adhesion materials for stamps,^5,7 modification of geometries of stamp structures to have tapered and smooth sidewalls,⁸ and modification of the NIL process, such as partial curing of a UV resist before demolding.⁹ Among all the schemes that have been developed to enhance demolding properties, application of an anti-sticking layer on the stamp surface is the most conventional remedy.^3,10–12 As the anti-sticking agents, hydrophobic silane molecules have been the choice materials due to the small surface energy and the capability to deposit on the surface of nanoscale structures via formation of self-assembled monolayers.¹³

The figure of merit to indicate the effectiveness of silane coatings has traditionally been the surface energy measured by contact angle measurement¹⁴ and the friction coefficient measured using atomic force microscopy (AFM).¹⁵ It is generally accepted that a hydrophobic silane coating on a stamp surface lowers both the surface energy and friction coefficient, thus leading to a decrease in adhesion at the horizontal stamp/substrate interface and friction at the vertical stamp/surface interface during demolding, respectively. However, the surface energy and friction coefficient only indicate the quality of the coating, not the role of the coating in the actual demolding process. The effect of the silane coating has also been studied by measuring the demolding force, which is defined as the force required to separate the stamp from the molded substrate.¹² The measured demolding force contains collective information on the effectiveness of demolding; however, it does not provide individual contributions of friction dissipation, stored elastic energy, and adhesion to demolding.

This work presents, for the first time, direct experimental evidence on the role of the silane coating in demolding in UV-NIL. This was done by systematically measuring the work of demolding for different silane coatings and comparing the results with theoretical relationships of stored elastic energy, friction dissipation, and adhesion to the depth of stamp structures.

II. DEMOLDING THEORY

The demolding process in NIL can be understood in the analogy of the debonding process of fiber-matrix in composites. Figure 1 shows schematics of the two processes. The stress in the stamp/substrate system for NIL is mainly generated during the cooling (Thermal-NIL) or UV curing step (UV-NIL) similar to the curing step of fiber resin in the fiber-matrix composite system. In addition, adhesion between the two dissimilar materials is operative. When an external stress is applied for demolding/debonding, the elastic energy first increases and is stored in the volume near the resist/stamp (or fiber/matrix in the case of the composite) interface before demolding/debonding. In Figure S.1 of the supplementary material, the schematics for the partially debonded stage in the fiber-matrix composite system is also shown. The accumulated elastic energy is released upon separation of the interfaces. For the separation, breakage of interfacial adhesion needs to occur prior to sliding by friction at the sidewall interfaces. The only difference between demolding in NIL and debonding in fiber-matrix composite systems is that the adhesion term in NIL also includes the contribution from horizontal interfaces of the resist/stamp system, while adhesion and friction are operative at the vertical (sidewall) interfaces in both processes.

FIG. 1.

Schematics of (a) the debonding process of fiber-matrix in composites and (b) the demolding process in NIL.

According to the theoretical work on fiber-matrix debonding energy by Sutcu and Hillig,¹⁶ equilibrium demolding occurs over an incremental distance $\mathrm{d}L.$ The external work, i.e., work of demolding, is equal to the summation of the local increase in the elastic energy, the work to break the adhesion between the interface prior to sliding, and the incremental dissipative energy developing during sliding by friction. The energy balance yields

$$\frac{\mathrm{d}{W}_t}{\mathrm{d}L}=\frac{\mathrm{d}(\Delta U)}{\mathrm{d}L}+\frac{2V_r{\Gamma }_d}{a}+\frac{\mathrm{d}{Q}_f}{\mathrm{d}L},$$ (1)

where $W_t$, $\Delta U$, ${\Gamma }_d\hspace{0.17em}$, and $Q_f$ are the external work, stored elastic energy, interfacial debond energy, and frictional dissipation, all of which are presented in the units of energy per unit area. $V_r$ is the volume fraction of the resist in the patterned region and $a$ is the half of the pattern width. Integration of Equation (1) over the demolding length $L$ (equal to the stamp depth) clearly shows that the adhesion term ($\frac{2V_r{\Gamma }_d}{a}$) has a linear dependence on $L$. On the other hand, assuming that the applied external stress (σ) and the residual axial stresses in the resist and stamp in the patterned region (${\sigma }_r^r,\hspace{0.17em}{\sigma }_s^r)$ are constant during the demolding process, $\Delta U$ and $Q_f$ can be represented as third-order polynomial functions of $L$ (for details, see the supplemental information). Therefore, based on the theoretical considerations, one can hypothesize that relative contributions between the adhesion term and the other two terms to overall demolding work can be experimentally verified by examining the deviation from the linearity in the measured demolding work vs. stamp depth $L$ curves.

III. EXPERIMENTAL

A. Sample preparation

1. NIL stamp fabrication

For UV-NIL and de-bonding measurements, three 1 × 1 in.² stamps with different dimensions were fabricated by photolithography and reactive ion etching. One stamp was simply a blank silicon wafer. Two of the stamps, however, were fully covered with gratings with the same width and period (5 μm and 15 μm, respectively) but with two different depths: 570 ± 4 nm and 4.5 ± 0.2 μm. Details of the stamp fabrication are described in our previous paper.¹⁷ The stamps were first thoroughly cleaned using acetone, isopropanol, and de-ionized water and then dried using pressurized air prior to use.

2. Hydrophobic silane coating on Si stamps

Prior to the coating, the surface of Si stamps was activated by a reactive ion etching with an O₂ plasma at 150 W and an oxygen pressure of 250 mTorr for 5 min. Oxygen reactive ion etching is believed to increase the –OH bonds on the surface of silicon, which act as anchoring sites for silane molecules.

For each stamp, two different silane molecules, trifluoropropyltrichlorosilane (C₃H₄Cl₃F₃Si) labeled as F₃-TCS and tridecafluorotetrahydrooctyltrichlorosilane (C₈H₄Cl₃F₁₃Si) labeled as F₁₃-TCS, were separately applied via vapor-phase deposition as anti-adhesion coatings. The chemical structures of these two silane molecules are shown in Figure 2. Both molecules have a trichlorosilane head group that enables bonding to silanol groups on oxidized Si surfaces.¹⁸ The difference is the length of fluorinated carbon tails, which results in different surface properties for the coated surfaces. Two factors determining the quality of anti-adhesion coating include (1) the grafting density by the trichloro head group and (2) self-assembly of the long fluorinated carbon chain via steric hindrance. Trichlorosilanes (silanes with three functional chloro groups) were used due to the fact that when the silane molecules contain more than one functional group, the excessive functional groups of each molecule can also react with those of the adjacent molecules forming a 3D network of silanes. In fact, Gauthier et al. showed that trichlorosilanes most efficiently graft to the surface of silica compared with monochlorosilane, monoethoxysilane, and triethoxysilane.¹⁸ Regarding the length of the fluorinated carbon chain, a long carbon chain usually leads to a better ordered self-assembled structure.¹⁹ Since there are no side chain fluoride atoms for F₃-TCS, the resulting surface energy becomes significantly higher compared to F₁₃-TCS with side chain fluoride atoms. It is worthwhile to note that there is a silane molecule with a longer fluorinated carbon tail (Heptadecafluoro-1,1,2,2-tetrahydrodecyl)trichlorosilane, F₁₇-TCS). However, the water contact angles and the thermal degradation behavior for the F₁₇-TCS coated Si surface were very similar to those for the F₁₃-TCS coated Si surface.²⁰ Therefore, we decided to use only F₁₃-TCS in this work, which has been most widely used as anti-adhesion coating for NIL stamps.

FIG. 2.

Chemical structure of the silane molecules used.

B. Surface characterizations

1. X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) measurements were performed using a monochromatic Al Kα X-ray source at 15 kV and 10 mA on a photoelectron spectrometer (Kratos axis 165). The chemical composition of silicon coated with F₃-TCS and F₁₃-TCS is shown in Table I. As expected, the fluorine content of the F₁₃-TCS treated surface is larger than F₃-TCS. The remaining small amount of chlorine also indicates good grafting.

TABLE I.

Chemical composition of F₃-TCS and F₁₃-TCS silanes coated on silicon by XPS (at. %).

	C	O	F	Si	Cl
F3-TCS	23.33	28.23	11.51	36.48	0.44
F13-TCS	21.19	36.34	23.44	18.99	0.03

2. Contact angle measurements

DI water and hexadecane were used for static contact angle measurements. The measurements were performed on the flat (blank) silicon stamp by applying 5 μl of the liquid droplet. A minimum of 4 measurements were performed, and the results were averaged. The contact angle measurements showed a standard deviation of ±4°. The surface energy and its polar and dispersive components were calculated using the Owens-Wendt method. Sliding contact angle measurements were performed by dropping 10 μl water droplets onto a sample placed on a manual goniometer (model 07GON504 from CVI Melles Griot). The goniometer was then slowly tilted until the droplet started rolling while recording the event with a high speed camera. The images right before rolling started were used to measure receding and advancing angles, as well as their difference, which is contact angle hysteresis. A method presented by Chibowski was used to calculate the surface energy using the advancing and receding contact angles.²¹

3. Friction coefficient measurements

In order to obtain friction coefficients of silicon treated with different silanes, lateral force microscopy (LFM) was conducted using a commercial atomic force microscope (AFM, Agilent 5500). A conical silicon tip with a nominal tip radius of 10–20 nm and a typical spring constant of 0.2 N/m (Budget Sensors, model All-in-one-al) was used. Prior to the measurements, the AFM silicon tip was scanned over a silicon wafer for 15 min to make sure that no more blunting of the tip would take place during friction test experiments. The friction coefficients were determined by measuring lateral cantilever deflection versus normal load applied. The lateral deflection is measured in volts (V) with the AFM photodetector. The lateral deflection in V is converted to the unit of force by a method suggested in Ref. 22. In this method, the lateral deflection of a cantilever in V is converted to Newton (N) by determining a lateral sensitivity factor having a unit of N/V. The lateral sensitivity factor is determined by scanning the tip over a surface with a known friction coefficient. A (100) silicon wafer was used as the surface with a known friction coefficient. At room temperature, silicon (100) scanned by a silicon tip (μ_Si-Si) has a friction coefficient of 0.18 ± 0.03 as reported by Buenviaje et al.²² Using this value, the Si cantilever lateral sensitivity factor was obtained and used to convert the voltage signal to force.

C. NIL and demolding work measurements

Imprinting and demolding were carried out in different apparatuses. UV imprinting was performed using a commercial nanoimprinter (Obducat 6″) while a modified mechanical tester (Testwork 5, MTS) was deployed to measure force/displacement response during demolding. 1 × 3 in.² glass slides were used as the substrate. The glass slides were treated with the oxygen plasma for 15 min at the power and gas pressure of 150 W and 250 mTorr, respectively, to enhance its adhesion to the polymer. 20 μl of a di-acrylic UV-curable resist was manually dispensed on the surface of the stamp. This UV resist was composed of polypropyleneglycol diacrylate (70 wt. %) from Aldrich as the base, trimethylolpropane triacrylate (28 wt. %) from Aldrich as the cross-linking agent, and Irgacure 651 (2 wt. %) from Ciba as a photo-initiator. This UV resist system and some of its applications are studied elsewhere.^6,23 After dispensing the resist on the surface of the stamp, a glass slide was concentrically placed on the stamp. The stamp was previously mounted on an aluminum mold bed using a commercial epoxy glue to fit into the tensile machine fixtures. Curing was done for 10 s by using a UV flash-lamp at an intensity of 1.8 W/cm² with the wavelength range of 250–400 nm while applying a slight pressure to ensure complete filling of the stamp cavities and also similar residual layer thickness for all samples (∼1 μm as measured by AFM). After imprinting, the assembly of aluminum/stamp/resist/slide glass was transferred to a mechanical tester machine. The slide glass was clamped to the upper traverse. Then, tensile displacement at a constant rate of 0.1 mm/min was applied, while the force response was measured by a load cell. Figure 3 shows a schematic of the imprinting/ demolding tests and a typical force-displacement curve obtained during the demolding process. Demolding energy (or demolding work) was taken as the area beneath this curve. For each stamp/silane system, at least 4 successful demolding force-displacement measurements were performed, and the results were averaged. Successful demolding is defined as complete transfer of the pattern from the stamp to the substrate with no visible residues remaining on the stamp. The standard deviations for the demolding work were less than ±2 mJ.

FIG. 3.

(a) Schematic view of UV imprinting and de-bonding measurements: (1) the stamp mounted on the aluminum mold beds, (2) dispensed resist, (3) UV curing, and (4) de-bonding; (b) a typical load vs. extension graph for de-bonding. The arrow indicates the magnitude of demolding force, and the area beneath the curve yields the de-bonding energy.

IV. RESULTS AND DISCUSSION

A. Characterization of silane-coated Si surfaces

The surface compositions of the two silane coatings determined by XPS are presented in Table I. The fluorine content for the F₁₃-TCS coated surface was larger than that for the F₃-TCS coated surface. Upon networking of a trichlorosilane molecule to the substrate surface and other neighboring trichlorosilane molecules, chlorine atoms need to be detached from the molecule. Thus, the chlorine composition is indicative of the amount of non-covalently bound silane molecules or non-grafted chlorosilane groups to other silane molecules. The results show that the chlorine contents for F₃-TCS and F₁₃-TCS are 0.44 and 0.03, respectively, indicating a well-grafted coating with F₁₃-TCS compared to the coating with F₃-TCS.

Table II shows the wetting properties of Si surfaces with different silane coatings and friction coefficients between a silicon tip and the silane-coated surfaces. Figure 4 shows the measured friction force versus the normal force measured by AFM. The low surface energy for the silane-coated surfaces is imparted by the presence of –CF₂ and –CF₃ groups in the molecules.²⁴ When the silane with a longer chain length (F₁₃-TCS) was used, the reduction in surface energy was even more pronounced. However, both silanes showed almost identical contact angle hysteresis. The F₃-TCS silane reduced the friction coefficient from 0.18 for the bare Si surface to 0.16, whereas the F₁₃-TCS silane coating was more effective and reduced the friction coefficient to 0.11. It should be noted that the measured friction coefficients do not reflect sliding occurring at the vertical interface of resist/stamp systems during demolding because it represents the friction between the Si tip and silane coated Si substrate; they only indicate the quality of coatings. A smaller friction coefficient for the silane with a longer chain is explained by the increased packing density and order between the molecules, which decreases the number of energy modes such as kinks, defects, and chain distortion that can result in the dissipation of the energy required for sliding to occur.²⁵ This explanation is also corroborated by the XPS results, which showed a better-grafting for the F₁₃-TCS coated surface.

TABLE II.

Wetting and frictional properties of Si surfaces with various coatings. ${\sigma }_{\mathrm{S}},{\hspace{0.17em}\sigma }_{\mathrm{S}}^{\mathrm{P}},{\hspace{0.17em}\text{and}\hspace{0.17em}\sigma }_{\mathrm{S}}^{\mathrm{D}}$ are total, dispersed, and polar surface energies, respectively. CA and CAH represent the contact angle and contact angle hysteresis, respectively. Friction coefficients were measured between a Si atomic force microscopy tip and Si surfaces with various coatings using atomic force microscopy.

	${\sigma }_S$a (mN/m)	${\sigma }_S^P$ (mN/m)	${\sigma }_S^D$ (mN/m)	${\sigma }_S$b (mN/m)	Advancing CA	Receding CA	CAH	Friction coefficient
Bare silicon	64.15	36.76	27.39	N.A	N.A	N.A	N.A	0.18
F3-TCS	15.47	2.25	13.22	15.05	110.1	64.7	45.4	0.16
F13-TCS	11.50	1.22	10.28	9.49	122.0	77.0	45.0	0.11

^aCalculated from the Owens-Wendt method.

^bCalculated from advancing and receding contact angles.²¹

FIG. 4.

Friction vs. normal force curves measured by LFM for bare silicon and silicon treated with silanes. Friction coefficients are determined from the slope of these curves.

B. Demolding energy measurement

The measured demolding energy (or demolding work) versus stamp depth curves for various silane coatings on the stamp surfaces are shown in Figure 5. The standard deviation of each data point was less than ±2 mJ. The error bars were not included in the graph in order to more clearly show the deviation of the curves from linearity. As expected, as the stamp depth L increased, overall a larger demolding work was required to separate the stamp from the molded substrate. Comparing the curves with different silane coatings, it was found that the demolding work was the highest when a bare Si was used for the stamp and was the smallest when F₁₃-TCS was coated on the stamp surface. In addition, the curve for the stamp without any silane coating was linear, while the other two curves with silane coatings appeared to deviate from the linearity.

FIG. 5.

Demolding work of a UV resist imprinted on bare silicon and silicon treated with different silanes for the stamps with the same grating (15 μm period) structures but with different stamp depths (0 (non-structured), 570 nm, and 4.5 μm). The linear fit for the bare Si stamp and a simple cubic fit for the F₁₃-TCS coated stamp are included.

As mentioned previously, the expected hypothesis according to the theoretical work by Sutcu and Hillig¹⁶ is that adhesion contributes to the overall demolding work in a linear fashion with $L$, while the contributions from frictional dissipation and stored elastic energy follow 3rd order polynomial functions. In order to examine the hypothesis, curves were separately fitted with two different functions: one is the linear function, $W_{\mathit{d}\mathit{e}\mathit{m}\mathit{o}\mathit{l}\mathit{d}\mathit{i}\mathit{n}\mathit{g}}=W_{demolding,0}+\mathrm{b}\cdot L$, and the other is a simple cubic function of $L$, $W_{\mathit{d}\mathit{e}\mathit{m}\mathit{o}\mathit{l}\mathit{d}\mathit{i}\mathit{n}\mathit{g}}=W_{demolding,0}+\mathrm{b}\prime \cdot L^3$, where $\mathrm{b}$ is the fitting parameter. $W_{demolding,0}$ is the measured demolding work at zero stamp depth ($W_{demolding,0}$), which should be constant for a certain surface coating even with stamps of non-zero stamp depths because the area of the horizontal interfaces is identical. The fitting results for the respective fitting to both functions are shown in Figure S.2 of the supplementary material. It should also be noted that the use of the simple cubic function of $L$ for curve-fitting is just to show deviation from the linear function; it is not intended to quantitatively describe individual contributions of adhesion, frictional dissipation, and stored elastic energy. Table III shows the standard deviation for the fitting parameters ($\mathrm{b}$ and $\mathrm{b}$′) and the adjusted R² values, both of which indicate the quality of the curve-fitting and thus the deviation from the fitting functions. The adjusted R² value for the bare Si stamp was 0.99971, indicating a very good fit. The value for the F₃-TCS coated stamp was smaller, and the fit curve for the F₁₃-TCS coated stamp showed a significant deviation from linearity with the adjusted R² value of 0.98385 (See the complete fit curves provided in the supplementary material). The reversed behavior was obtained for curve-fitting with the simple cubic function of where the best fit was obtained with the F₁₃-TCS coated stamp with the adjusted R² value of 0.99931. The fit curve with a linear function for the bare Si stamp and the curve with the simple cubic function of $L$ for the F₁₃-TCS coated stamp are added in Figure 5. For the F₃-TCS coated stamp, neither a linear fit nor a simple 3rd order polynomial fit with L provided satisfactory results both visually and with adjusted R² values higher than 0.999, indicating that the contribution of adhesion to the demolding work is not negligible.

TABLE III.

Adjusted R² values obtained from fitting of demolding work versus stamp depth curves for various coatings of stamp surfaces with two fitting functions: ${\mathrm{W}}_{\text{demolding}}={\mathrm{W}}_{\text{demolding},0}+\mathrm{b}\cdot \mathrm{L}$ for linear fitting and ${\mathrm{W}}_{\text{demolding}}={\mathrm{W}}_{\text{demolding},0}+\mathrm{b}\prime \cdot {\mathrm{L}}^3$ for simple cubic fitting of $\mathrm{L}$.

Adjusted R2	Bare silicon	F3-TCS	F13-TCS
Linear fitting with $\mathrm{L}$	0.99971	0.99164	0.98385
Cubic fitting of $\mathrm{L}$	0.97930	0.99583	0.99931

According to the hypothesis set for this work, the curve-fitting results indicate that, when the bare Si stamps are used for NIL, a majority of demolding work is consumed to break adhesion at the stamp/resist interfaces, leading to a linear relationship between the demolding work and stamp depth. As the quality of the silane coating for NIL stamps improved in the order of F₃-TCS and F₁₃-TCS, the relative contribution of adhesion to the overall demolding work decreased; the demolding work versus stamp depth curves deviated more from linearity, ultimately following a cubic function of $L$. Therefore, it can be deduced that the major role of the hydrophobic silane coating for NIL stamps is to reduce adhesion at the resist/stamp interfaces and the reduction in frictional dissipation is not as significant as the reduction in adhesion. To our knowledge, it is the first experimental evidence that was obtained directly via NIL and demolding work measurements to demonstrate the role of the silane coating of stamp surfaces in NIL.

It is worth noting that a significant shrinkage of the resist upon UV curing may lead to spontaneous demolding even before any demolding force is applied. According to the simulation work by Chan-Park et al., the resist-stamp interface could be broken depending on the magnitude of the resist shrinkage upon curing.²⁶ The UV resist used in this study has a volumetric shrinkage of 6.6%.²⁷ However, there was no experimental evidence regarding whether or not any spontaneous demolding occurs. We believe that the shrinkage stress was not sufficient to break the adhesion between the UV resin and the stamp surface and that the shrinkage stress is stored in the resist and stamp as elastic energy in this study. Therefore, for simplicity, we neglected the possible effect of resist shrinkage on breaking the adhesion before the demolding force is applied in the analysis.

One other question that arose during the analysis of our results is if the demolding work with a silane coated Si stamp, which follows a cubic function of $L$, would ever exceed the demolding work with a bare Si stamp that follows a linear relationship with $L$ as $L$ increases further. The explanation is as follows: Demolding work was only investigated using stamps with shallow depths, where the contribution of the stored elastic energy and frictional dissipation terms to overall demolding work is small compared to that of the adhesion to the demolding force. Therefore, the large contribution of adhesion with the bare Si stamp showed the linear demolding work vs. stamp depth behavior. However, as the stamp depth further increases, the contribution of the store elastic energy and frictional dissipation to overall demolding work increases dramatically. It should be reminded that the delaminated portion of the resist/stamp system contributes to the frictional dissipation during demolding and the contribution of the frictional dissipation to the overall demolding work will increase with increasing aspect ratio of the structure. Therefore, with high aspect ratio structures, the demolding work vs. stamp depth curve will deviate from linearity and follow a cubic function of the stamp depth. As a result, the demolding work for the silane coated Si stamp is always smaller than that for the bare Si stamp.

This analysis of utilizing the deviation from linearity in the demolding energy versus stamp depth curve is valid for low aspect ratio structures only. However, even with high aspect ratio structures, it does not change the conclusion that the major role of the hydrophobic silane coating is to reduce adhesion at the resist/stamp interfaces. Another interesting question is at which aspect ratio of the structure, a catastrophic demolding failure would occur. In order to answer the question, it may be necessary to perform a series of experiments with stamps of different aspect ratios. Numerical simulation on the demolding process may also provide the limitation of the aspect ratio of the structures in obtaining successful demolding by comparing the resulting maximum stress or stored elastic energy in the resist with the mechanical properties of the resist.

It is also worth mentioning the durability of the silane coatings. The degradation mechanism of silane has already been vastly investigated in the literature.^3,11,12,28 It is generally believed that the free radicals generated during UV cure can cleave the fluorocarbon chains, resulting in a progressive loss of fluorine with the number of imprints. For each system measured in this study, some deviation was noticed in the demolding work. However, the deviations were erratic, and no systematic trend was observed (these deviations were included in the ±2 mJ error of the measurements). Also, only a small percentage decrease in the fluorine content of the silane layer (4–5 wt. %) was detected in the XPS measurements. This implies that the degradation of the silane layer within the number of the tests performed was negligible.

V. CONCLUSION

The role of a hydrophobic silane coating on demolding in NIL was studied by measuring demolding work with stamps having the same structures but different depths and by comparing the results with the theory developed for the fiber-matrix system. The results conclude that the major role of the hydrophobic silane coating is to reduce adhesion at the resist/stamp interfaces, while the reduction in frictional dissipation is not as significant as the reduction in adhesion. Based on our analysis, the main strategy for anti-adhesive layers for NIL stamps does not change. However, this study adds two additional values to the current knowledge on demolding and anti-adhesive layer coatings. The first one is to underline the mechanism that the silane coatings work in enhancing demolding by directly testing them in real world NIL applications. The second one is to provide an experimental methodology to study anti-adhesive coatings for NIL stamps.

VI. SUPPLEMENTARY MATERIAL

See supplementary material where the details of extending the equation derived by Sutcu and Hillig for fiber-matrix debonding energy to demolding in NIL are presented. An analysis is also provided to qualitatively demonstrate the impact of stamp depth in NIL on internal energy, adhesion, and friction dissipation parameters in that equation. Besides, line fitting results along with adjusted R² values for the demolding work vs. stamp depth curves with two different functions: (a) a linear function of $W_{\mathit{d}\mathit{e}\mathit{m}\mathit{o}\mathit{l}\mathit{d}\mathit{i}\mathit{n}\mathit{g}}=W_{demolding,0}+\mathrm{b}\cdot L$ and (b) a simple cubic function of $W_{\mathit{d}\mathit{e}\mathit{m}\mathit{o}\mathit{l}\mathit{d}\mathit{i}\mathit{n}\mathit{g}}=W_{demolding,0}+\mathrm{b}\prime \cdot L^3$ are provided for stamps treated and untreated with silane molecules.