Microelectronic Engineering 155 (2016) 85–91
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
Effect of nanoimprint on the elastic modulus of PMMA: Comparison between standard and ultrafast thermal NIL Michele Pianigiani a,b,c,⁎, Robert Kirchner d, Enrico Sovernigo a,b, Alessandro Pozzato a,b, Massimo Tormen a,b,⁎, Helmut Schift d,⁎ a
ThunderNIL srl, Via Foscolo 8, 35131 Padova, Italy IOM-CNR, Area Science Park, Basovizza, S.S. 14, Km. 163.5, 34149 Trieste, Italy Università degli Studi di Trieste, Piazzale Europa, 34127 Trieste, Italy d Paul Scherrer Institut, Laboratory of Micro and Nanotechnology, 5232 Villigen PSI, Switzerland b c
a r t i c l e
i n f o
Article history: Received 16 October 2015 Received in revised form 4 March 2016 Accepted 8 March 2016 Available online 10 March 2016 Keywords: Thermal NIL Pulsed-NIL PMMA PeakForce AFM Young's modulus
a b s t r a c t This paper is focused on the understanding of the effect of the nanoimprint lithography process on the elastic modulus of thin, thermoplastic films. In particular, we present the comparison between the standard and an ultrafast thermal NIL technology as well as the way both processes affect the top surface of poly(methyl methacrylate) (PMMA). The PeakForce QNM™ (Quantitative Nanomechanical Property Mapping) scanning probe technique was used to determine the Young's modulus of PMMA by comparison with a polystyrene standard. We demonstrate that imprinted PMMA, regardless of the used method, shows a 9-fold increase of Young's modulus compared to non-imprinted PMMA at least in the top 3–5 nm thick surface layer. This important finding proves that the ultrafast process with much higher temperatures, but also with much shorter process times, leads to elastic surface properties that are comparable to those of PMMA imprinted with the standard process. We have confirmed that annealing alone does not significantly influence the Young's modulus. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Thermal nanoimprint lithography (T-NIL) is a replication process able to pattern micro- and nanostructures in a range of materials. Crucial for its implementation is selecting thermoplastic resists with suitable thermo-mechanical properties, i.e., a reversible relation between temperature and viscosity. This allows performing repeatable thermocycles, by which the polymer can be molded in its viscous state and demolded in its hard (glassy) state [1]. Ideal thermoplastic molding relies on purely viscous processes and volume conservation for a given temperature, i.e., for both, the solid and the viscous state. In room temperature imprint processes it seems to be possible to compress polymers without significant flow resulting in a compaction or to have a strain hardening [2,3]. Less known is the effect which the NIL process has on the properties of the imprinted polymer, particularly, if imprint is done at very high temperatures and high shear rates, as in the case of ultrafast NIL processes [4,5]. The modification of different properties (mechanical, optical, chemical) would affect the final nanostructured polymer product, or play a role in the subsequent process steps if the material is used as a resist for pattern transfer. Examples of possible effects are local polymer
densification by reduction of free-volume [1,2], residual birefringence by orientation of polymer chains and process-induced crystallization [6,7]. Apart from changes in conformation, different kinds of degradation (chain scission, oxidation, carbonization) are possible through high temperatures and high mechanical pressure or shear [8–11]. These effects would be expected to be more severe if ultrafast thermal processes are employed which is now possible by the pulsed-NIL approach [4,5]. In this work, we examined the effect of T-NIL on the elastic modulus of the top surface of spin-coated poly(methyl methacrylate) (PMMA). By using the PeakForce QNM™ (Quantitative Nanomechanical property Mapping) mode of a Bruker atomic force microscope (AFM), the Young's modulus of as-coated resists and of imprinted and patterned resist were investigated. The PeakForce QNM™ method recently received increasing interest in polymer characterization [12–14]. For the first time, we compare the influence of an ultrafast and of a reference standard NIL process on the resists elastic modulus. The aim is to understand whether the significant differences of temperatures and shear rates between both processes are affecting the elastic properties of the resist. 2. Materials and methods
⁎ Corresponding authors. E-mail addresses:
[email protected] (M. Pianigiani), massimo.
[email protected] (M. Tormen),
[email protected] (H. Schift).
http://dx.doi.org/10.1016/j.mee.2016.03.019 0167-9317/© 2016 Elsevier B.V. All rights reserved.
In order to use nano-indentation measurements to visualize the differences in the thermoplastic film properties introduced only by the
86
M. Pianigiani et al. / Microelectronic Engineering 155 (2016) 85–91
Fig. 1. Simulation of the indentation time required to reach a certain residual layer for a 5 μm line.
nanoimprinting conditions, samples with exactly the same mold microstructure must be used. To achieve this, a single stamp suited for the pulsed-NIL was prepared and also used for the standard NIL process.
described by Stefan's squeeze flow equation [8–10]. 1 2
hðtÞ 2.1. Stamp For the ultrafast thermal NIL, the stamp requires a resistive heating layer buried below the patterned surface of the stamp. This is composed of a thin layer with a significantly higher conductivity compared to the stamp substrate. It can be obtained for instance by heavy ion implantation into the silicon, typically a few μm deep into surface. The stamp used in this study was made on a 100 mm silicon wafer being ion implanted by a commercial foundry using 31P ions and a dose of 2·1016/ cm2 at 200 keV. This leads to a sheet resistance of 4 Ω/square of the heating layer. To prevent artefacts in the AFM measurements and to obtain reliable information also at the bottom of the structures, a grating of lines of 4 μm width, 8 μm period and 525 nm height was realized by standard UV-lithography and dry etching. The stamp was functionalized with a monolayer of alkyl-trichlorosilane for easy release.
2.2. Sample fabrication by nanoimprint Previous work [5] demonstrated the imprinting of a line pattern of 5 μm width into PMMA of 100 kg/mol molecular weight Mw with an initial thickness of 200 nm with pulsed NIL. The indentation time being necessary to obtain a residual layer that is one tenth of the initial thickness is around 130 min at 160 °C for this resist while it is 3500 μs at 500 °C (Fig. 1). These times depend on the polymer viscosity η (which depends on the type of resist, temperature and shear rate), the width of the structure (w), the initial thickness (h0) and the applied pressure (p) as
¼
1 2
h0
þ
2p t ηw2
For our experiments, the samples, for both type of imprint, were prepared by spin-coating a 1050 nm thick PMMA film of about 121.2 kg/ mol molecular weight Mw on standard silicon wafer substrates. With this initial thickness, the time for a complete imprint cycle using standard T-NIL is about 15 min (including heating/cooling), however it is known that the total indentation time is in the range of 5 s [15]. This is still much larger than the roughly 100 μs required only for ultrafast T-NIL. This was confirmed by the fact that the samples were completely imprinted with comparably fast experimental conditions (see below). For the annealing experiments the PMMA thickness was about 1800 nm. Experimentally, the stamp was used first for the standard nanoimprint with a Jenopik HEX03 at a temperature of 150 °C, a pressure of 107 Pa and a holding time of 10 min. Then, the same stamp was installed in the ULISS equipment (Fig. 2) for the ultrafast imprint at a pressure of 7.5·106 Pa and a pulse duration of 100 μs, reaching temperatures estimated to be in the order of 500 °C. At this point it is possible to calculate the total energy of the two processes, which is around 104 J/cm2 for the standard T-NIL, based on the heat capacity of the hot plates and total temperature variation during the imprint. For the ultrafast process is in the order of 10 J/cm2, based on the electrical power consumed for heating. This highlights the huge saving of energy made possible by pulsed-NIL thanks to the different configuration of the press and stamps. The schematic, in Fig. 3, shows the key differences in the hot-plates/ press systems, stamps as well as operation principles and explains the extremely different consumption of energy for the two imprinting processes. The main problem of the standard T-NIL is that one has to heat
Fig. 2. a) Overview of the system “ULISS” for ultrafast NIL at 100 mm wafer scale. b) Class 10 mini environment with the pressing station within the “ULISS” system and c) Example of an ultrafast imprint on the full area of a 100 mm wafer.
M. Pianigiani et al. / Microelectronic Engineering 155 (2016) 85–91
87
Fig. 3. Heating schemes for a) standard T-NIL where the heat comes from the hotplate, and b) ultrafast T-NIL where the heating layer is in the stamp itself.
up and cool down bulky and heavy hotplates with a large mass (in the order of 1–2 kg for a plate of 150 mm diameter), while what is strictly necessary to heat is just the resist, which is in the order of a few milligrams. For this reason, heating and cooling times are long and in the range of 1 to 10 min. As earlier discussed, it is necessary to add the time of the indentation which requires minutes at 150–200 °C. Going to a higher temperature would increase further the heating and cooling time and may even result in the degradation of the polymer due to the long time at high temperature. In the approach developed by ThunderNIL, the heat comes directly from the stamp, which is thermally isolated from the bulky pressing mechanics. A short current pulse flowing below the stamp surface heats up the resist to a temperature over 500 °C to have a complete indentation within ca. 100 μs. The full imprinting, made by manual handling of the substrates requires less than 1 min. The introduction of an automatic substrate handling system is expected to reduce the imprinting cycle below 10 s.
2.3. Sample analysis by AFM After an initial selection of the correct AFM probe, AFM analysis with a Bruker Dimension Icon AFM gave a high resolution mechanical feedback at each tapping point providing spatially correlated topography and mechanical property mapping within a single scan [12–14]. While the topography is recorded in tapping mode, the AFM tip measures a force distance curve for each topography pixel (Fig. 4). The used tapping frequency of PeakForce QNM™ was 2 kHz and the line speed was 1 to 10 μm/s. The AFM tips used for this method were TAP525A with a Young's modulus range of 1–20·109 Pa and RTESPA with a range of 0.2–2 · 109 Pa [12]. The graph in Fig. 4 shows a typical force-distance curve where the modulus is fitted by the software from the dotted region using an implemented Hertzian model with adhesion (DMT: Derjaguin-Muller-Toporov). In a zone that is ideally fully elastic, this
method uses an indentation in the range of 3–5 nm to image local modulus variations. The analysis of the mechanical properties using an AFM has some problems correlated to the tip of the instrument. First is necessary to consider that every tip is different resulting in a variation on tip radius or spring constant. Thanks to the use of a hard reference substrate such as sapphire, fused silica or silicon having a Young's modulus well above the sample to be measured it is possible to set the deflection sensitivity within a few minutes. Secondly, in the remaining parts of AFM calibration one has to measure the spring constant and the tip radius. Both procedures are complicated, quite long and cannot be applied for all cantilevers. Also there is another problem: the degradation of the tip that causes a change of the radius during the work. This is important because the Young's modulus in the DMT-model is inversely proportional to the square root of the tip radius [14]. A change of the radius could happen either by wear, deformation or by collection of contaminations from the surface. For this study of the Young's modulus, a “comparative method” was used. The idea behind is to scan a reference sample (polystyrene film 12 on silicon, supplied by the Bruker), that has a known value of Young's modulus (near 2.7 GPa), and to normalize the results to this value. 3. Results and discussion The analysis of the structures is divided in two parts: general aspect, i.e., topography analysis and mechanical properties measurement. 3.1. General aspect and topography The first analysis performed with AFM was a 3D topography and 2D section analysis of the profiles of the structures imprinted with the two different methods. Both analyses did not show any differences between the two types of imprinted structures (Fig. 5).
Fig. 4. Operating principle of a Bruker Icon scan with PeakForce QNM mode scanning the surface in tapping mode with a low tapping frequency of 2 kHz while typical line speeds are in the range of 1 to 10 μm/s. The force-distance curve at each tapping pixel is defined by: surface approach (1–2), point of deepest indentation (3), force release and surface restoration (4), point of largest adhesion (5) and released cantilever (6).
88
M. Pianigiani et al. / Microelectronic Engineering 155 (2016) 85–91
Fig. 5. 3D AFM topography of line structure having 4 μm width, 8 μm period, 500 nm height and being imprinted in PMMA a) with standard T-NIL, b) with ultrafast T-NIL and c) 2D profiles of the two imprinted structures in a) and b) that show the same aspect and geometry.
To enter more in details, an analysis of roughness was performed with the use of GWYDDION 2.37 for data visualization and analysis [16]. Arithmetic average roughness, Ra, was about 1.1 nm for standard T-NIL imprinted structures and about 1.7 nm for the ultrafast imprint. Root mean squared roughness, Rq, was 1.4 nm and 2.1 nm for the standard and the ultrafast imprint, respectively (Table 1). In any case these values are very close to each other and confirm the quality of the ultrafast imprint. There is no difference between the roughness of the bottom and the top of the structures and no difference between the two types of imprint. Thus, the structures and the different parts of the structure can be well compared for their mechanical properties.
3.2. Mechanical properties After the calibration (see Section 2.3) and at the end of the scan it is possible to have a first topography image (Fig. 6a) and the correlating AFM elastic modulus image (Fig. 6b) of the polystyrene reference with the same high resolution. During the analysis of the histogram of the modulus (Fig. 6c), it is possible to change the spring constant and the radius (in the typical tolerances of the tip) to reach the correct Young's modulus value of 2.7 GPa for the polystyrene reference. This reduces the calibration time significantly. Returning on tip degradation, depending on the “landing behaviour” on the surface and the strength of the indentation of the polymer, this degradation can occur not only during the calibration step but also at the beginning and during each scan. For this reason, after the scan of the last sample it is necessary to scan another time the polystyrene. These two measurements allow for an estimation of the degradation in a linear fashion. Like it is possible to see in Fig. 7, after scanning there is a variation of the value of the measured Young's modulus for the polystyrene reference. For the presented measurements, the decrease is in the order of 1–1.5 GPa after 15 scans. The graph shows also a change of the distribution of the modulus. The increase of the radius reduced the possibility to finely recognise and resolve the modulus and for this reason the histogram of the final scan on polystyrene appears more tapered. Linear interpolating the Young's modulus between initial and final reference scan allows for the compensation of the continuous tip wear in the further data analysis.
Table 1 Roughness comparison of the imprinted structures confirming that both methods results in very comparable surface patterns and allow for a good comparison of the elastic properties between them. T-NIL
Arithmetic average Ra Root mean squared Rq
Ultrafast T-NIL
Bottom
Top
Bottom
Top
1.1 nm 1.3 nm
1.0 nm 1.4 nm
1.7 nm 2.0 nm
1.6 nm 2.1 nm
In Fig. 8, a typical example of a scan on imprinted PMMA is shown. It is possible to see the uniformity of the value of the modulus, the high number of points being analysed (512 × 512 points) and the uniformity between the top (center) and bottom (lateral) of the structure. The low modulus lines in the center are related to the edge of the imprinted lines where the tip travels across the steep gradient (525 nm high step), causing artefacts in the measurement. Finally, it is possible to determine the relative value of the Young's modulus of PMMA (Fig. 9). In that phase, due to the decision to use the relative method, all values were divided (normalized) by the value of 2.7 GPa of the reference polystyrene. This gives a relative x-axis that shows a scale being correlated to the Young's modulus of the reference material (normalized Young's modulus E*). The spun-on and not imprinted PMMA has a modulus that is E* = 0.7, the PMMA imprinted with standard T-NIL or with ultrafast T-NIL have a value around E* ~ 6.0–6.3. The imprinted PMMA, with both techniques, shows a 9-fold increase of the modulus. However, there is no significant difference in the modulus measured for the samples imprinted with the two techniques even when the temperature of ultrafast imprint is more than 400 K higher than the temperature of standard imprint (150 °C). Furthermore, there are no significant differences between the top and the bottom of the structures in both T-NIL processes. Another series of experiment shows that the increase in modulus is primarily not correlated to the imprint temperature. An annealing, with different times, at the temperature of standard T-NIL (150 °C), without the application of pressure does not change the Young's modulus of PMMA as strong as the imprint process (Fig. 10). 3.3. Discussion A 9-fold increase of the Young's modulus can only result from a significant hardening of the film. This can be due to chemical modification (e.g. crosslinking, chemical transformation) or by densification. As described in Ref. [1], densification is possible due reduction of the free volume of a polymeric film, which is associated with the unoccupied space between polymer chains. Upon compression, this free volume can be reduced, which in turn increases the density. Densification by nanoimprint has been for example observed in cases where a material is molded well below its Tg and material flow is minimized [2]. Volume reduction in other polymer molding processes, e.g., during extrusion foaming, illustrates the large variety of applications [17]. If we assume squeeze flow without any additional compaction under each protrusion, this situation would by definition be associated with volume conservation by flow of the viscous polymer. Any compression during the initial phase of molding would result in a higher degree of compaction under the protrusions than in the cavities. If densification happens after the molding, i.e., when all cavities are filled, we have to distinguish the viscous case and, during the cooling, the visco-elastic and the solidified case. In the viscous case it would mean that the polymer chains
M. Pianigiani et al. / Microelectronic Engineering 155 (2016) 85–91
89
Fig. 6. a) AFM topography, b) AFM elastic modulus map, and c) histogram of the Young's modulus of the PS reference sample surface.
reduce the space between them while they are still mobile, in the latter case that the film first solidifies and is then further compressed. In all three cases of molded structures, with homogeneous vertical compression and negligible lateral material displacement anticipated, the compaction would be again higher in the areas of protrusions than in the areas of cavities, since the vertical movement of the stamp would be the same and thus the compaction relative to the remaining thickness different. We did not observe such a difference between protrusions and cavities. This favours either a homogenous and cooperative compaction, or hardening, in all regions involving material displacement or a superficial “skin” effect in all regions. Whether such a hardening happens at elevated temperatures or at the end of the cooling, i.e., whether the modification happens in the viscous, the visco-elastic or the solid state, has to be clarified by further experiments. We demonstrated further that an annealing at the temperature of the standard TNIL, without the application of pressure, does not change the Young's modulus of PMMA, from which we conclude that the pressure is the main source of elastic modulus change and the assumed compaction. The residual layer thickness in the bottom area of the imprinted film is about 800 nm and the filled structures are about 1320 nm high (above the substrate). Considering the mold geometry, this would mean that volume conservation applies and this speaks against a homogeneous compaction. Interesting is in this regards the presence of a similar increase of the Young's modulus both in the filled structures (flow of polymer into the 525 nm deep stamp cavities) and the thinned bottom structures (squeeze flow under the stamp protrusions), which should be different if the entire imprinted film with thickness contrast is compressed by a defined indentation depth as discussed before. However, the residual layer is still large in comparison to the height of the
Fig. 7. Histograms of the elastic modulus of the polystyrene reference sample showing the data for the first scan (right peak) and after 15 scans (left peak).
stamp protrusions, therefore differences in compaction, e.g., if the entire film would be compressed, would be relatively small. In conclusion, the current experiments suggest that only a thin modified skin layer is present on top of the entire surface. A homogeneous hardening throughout the full resist thickness is only possible for a cooperative compaction of the squeezed and the filled regions and this was not observed so far. In the future, the exact depth of modification has to be analysed. Then, the 9-fold increase of the Young's modulus can be associated with a specific degree of compaction [18]. However, further experiments, e.g., by using cantilevers with specific geometry and without wear, will be needed to verify the exact size of the difference in Young's modulus, which by now can only be quantified to be significant for both types of NIL. The lack of a difference in Young's modulus between standard T-NIL and pulsed NIL with more than 400 K difference in temperature and thus orders of magnitude difference in viscosity is difficult to explain so far. Due to the missing difference between both methods, it is concluded that a degradation of PMMA due to extreme temperatures and, e.g., carbonization or crosslinking, does not apply. In addition, previous experiments indicated no chemical modification of the PMMA by the pulsed NIL process. The absence of degradation can be explained by the short time that the polymer remains at high temperature in the pulsed NIL process. In general, the decomposition rate of polymers follows an Arrhenius law, with activation energies depending on the
Fig. 8. Modulus map with 512 × 512 points of PMMA imprinted with ultrafast thermal NIL. The “roughness”, i.e., the local variation of the Young's modulus across the surface is relatively small. The lines correspond to artefacts (AFM tip crosses steep sidewall) due to the full width scan of an imprinted line in contrast to a scan of only a part of the line width as in Fig. 5a and b.
90
M. Pianigiani et al. / Microelectronic Engineering 155 (2016) 85–91
Fig. 9. Relative modulus of PMMA with/without imprint. The two broad peaks on the right side are representing the normalized Young's modulus for ultrafast NIL and for standard T-NIL for PMMA while the group of peaks on the left side represents the pristine PMMA without imprinting as well as the PS reference measurements before and at the end of the measurements.
chemical nature of the polymer. Studies of fire-safe material report degradation and ignitability of typical polymers. For instance, Ref. [11] reports the temperature of the mass-loss or decomposition peak for PMMA as a function of the heating rate using thermogravimetric analysis. Since the typical temperature reached in an ultrafast imprint process is 500–550 °C, from the mentioned study we can assume that the decomposition of PMMA sets is at these temperatures when heating rates of ~1000 K/s are applied. However, the typical heating rates of ultrafast NIL is ca. 5·106 K/s (dT = 500 K within dt = 100 μs), i.e., 5000 times faster heating compared to the required rate for a decomposition at 520 °C. This also means that there is 5000 times less activation energy available than with the rather low rate leading to decomposition [11]. This explains why the ultrafast imprinting does not result in appreciable sign of degradation. At the same time, the total energy deposited into the material during the entire pulsed NIL process is, compared to a process with 1 min heating at 170 °C, about 3000 times smaller.
quantitative values have to be verified by other methods or tips with more defined geometry. Further experiments will be needed to determine the effect for different resist thicknesses, pressures and structure sizes, as discussed in Ref. [1]. The densification of PMMA seems to be characteristic to T-NIL in general and there is no difference between standard thermal and ultrafast NIL. The huge temperature difference of the latter process does not cause any difference in the mechanical properties of the surface. From the PeakForce process we know that this modification is at least in the top 3–5 nm of the surface (indentation range) but it is not known below. In the future, we have to clarify, whether this densification concerns the entire PMMA film. This would have consequences for the pattern transfer, i.e., on etching rates, or on thermal reflow processes [19]. Although there are still questions about the origin of the Young's modulus modification in the two thermal NIL processes, the main finding of this research is that ultrafast T-NIL matches the capability of standard T-NIL but with the benefit of much shorter imprint times.
4. Conclusions We have described the principal differences on equipment and stamps between standard and ultrafast T-NIL. The PeakForce QNM method was introduced and we demonstrated its ability to measure Young's modulus differences on topography of differently imprinted polymer films. The wear of the tip could be compensated by using a relative method. There is a clear proof of a significant increase of the PMMA Young's modulus due to the imprinting process; however, absolute
Acknowledgements We acknowledge the contribution by IOM CNR that made available laboratories and microfabrication equipment and LMN at PSI for thermal NIL and AFM instruments. We acknowledge Konrad Vogelsang for performing the imprints. We are grateful to Fabio Suran for his valuable contribution in the development of the Pulsed-NIL equipment.
Fig. 10. Modulus of PMMA annealed and imprinted. The two broad peaks on the right side are representing the normalized Young's modulus for pulsed NIL and for T-NIL. The group of narrow peaks at the left side represent those for flat/structured samples annealed for different times.
M. Pianigiani et al. / Microelectronic Engineering 155 (2016) 85–91
References [1] H. Schift, Nanoimprint lithography: an old story in modern times? A review, J. Vac. Sci. Technol. B 26 (2) (2008) 458–480. [2] D.-Y. Khang, H. Yoon, H.H. Lee, Room-temperature imprint lithography, Adv. Mater. 13 (2001) 749–752. [3] P.S. Hong, H.H. Lee, Pattern uniformity control in room-temperature imprint lithography, Appl. Phys. Lett. 83 (2003) 2441–2443. [4] M. Tormen, R. Malureanu, R.H. Pedersen, L. Lorenzen, K.H. Rasmussen, C.J. Lüscher, A. Kristensen, O. Hansen, Fast thermal nanoimprint lithography by a stamp with integrated heater, Microelectron. Eng. 85 (2008) 1229–1232. [5] M. Tormen, E. Sovernigo, A. Pozzato, M. Pianigiani, M. Tormen, Sub-100 μs nanoimprint lithography at wafer scale, Microelectron. Eng. 141 (2015) 21–26. [6] H. Hlaing, X. Lu, T. Hofmann, K.G. Yager, Ch.T. Black, B.M. Ocko, Nanoimprintinduced molecular orientation in semiconducting polymer nanostructures, ACS Nano 5 (2011) 7532–7538. [7] S. Jeong, Ch. Cho, H. Kang, K.-H. Kim, Y. Yuk, J.Y. Park, B.J. Kim, J.-Y. Lee, Nanoimprinting-induced nanomorphological transition in polymer solar cells: enhanced electrical and optical performance, ACS Nano 9 (2015) 2773–2782. [8] L.J. Heyderman, H. Schift, C. David, J. Gobrecht, T. Schweizer, Flow behaviour of thin polymer films used for hot embossing lithography, Microelectron. Eng. 54 (2000) 229–245. [9] H. Schulz, M. Wissen, N. Bogdanski, H.-C. Scheer, K. Mattes, Ch. Friedrich, Impact of molecular weight of polymers and shear rate effects for nanoimprint technology, Microelectron. Eng. 83 (2006) 259–280.
91
[10] H. Rowland, A. Sun, P. Schunk, W. King, Impact of polymer film thickness and cavity size on polymer flow during embossing: toward process design rules for nanoimprint lithography, J. Micromech. Microeng. 15 (2005) 2414–2425. [11] R.E. Lyon, Advanced fire safe materials for aircraft interiors, NIST Interagency Reports 9030 (2) (1996) 248–258. [12] PeakForce QNM User Guide, (Rev. F)., https://www.bruker.com/. [13] J. Adamcik, C. Lara, I. Usov, J.S. Jeong, F.S. Ruggeri, G. Dietler, H.A. Lashuel, I.W. Hamley, R. Mezzenga, Measurement of intrinsic properties of amyloid fibrils by the peak force QNM method, Nanoscale 4 (2012) 4426 (4pp). [14] T.J. Young, M.A. Monclus, T.L. Burnett, W.R. Broughton, S.L. Ogin, P.A. Smith, The use of the PeakForce™ quantitative nanomechanical mapping AFM-based method for high-resolution Young's modulus measurement of polymers, Meas. Sci. Technol. 22 (2011) 125703 (6pp). [15] H. Schift, S. Bellini, J. Gobrecht, F. Reuther, M. Kubenz, M.B. Mikkelsen, K. Vogelsang, Fast heating and cooling in nanoimprint using a spring-loaded adapter in a preheated press, Microelectron. Eng. 84 (2007) 932–936. [16] Modular program for SPM (scanning probe microscopy) http://gwyddion.net/. [17] L.A. Utracki, R. Simha, Free volume and viscosity of polymer-compressed gas mixtures during extrusion foaming, J. Polym. Sci., Part B: Polym. Phys. 39 (2000) 342–362. [18] X. Feng, Y. Huang, A.J. Rosakis, On the Stoney formula for a thin film/substrate system with nonuniform substrate thickness, J. Appl. Mech. 74 (2007) 1276–1281. [19] A. Schleunitz, V.A. Guzenko, M. Messerschmidt, H. Atasoy, R. Kirchner, H. Schift, Novel 3D micro- and nanofabrication method using thermally activated selective topography equilibration (TASTE) of polymers, Nano Convergence 1:7 (2014) 8.
