In situ film characterization of thermally treated microstructured conducting polymer films

Abstract

Based on a low-cost fabrication routine microstructured conducting polymer films of poly (dioctylfluorene-co-benzothiadiazole) (F8BT) are prepared without any heat treatment or vacuum steps. The influence of thermal annealing at temperatures below the glass transition temperature of F8BT on such microstructured channel structures is investigated. In the applied structuring routine, a F8BT film is spin coated on a channel-type hard master structure and afterwards floated on a flat support. Thereby, the properties of the final polymeric structures, for example channel width and height, can be tuned by simply varying the polymer concentration in solution and using the same master structure. With in situ grazing incidence small angle X-ray scattering and imaging ellipsometry the installed channel structure and the influence of thermal treatment are probed. A complex interplay between a macroscopic polymer flow (reduced channel heights) and a molecular rearrangement (formation of mesoscopic crystallites) takes place during thermal annealing. © 2012 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys, 2012

INTRODUCTION

Due to their mechanical flexibility in combination with the low fabrication costs conducting polymers are very promising candidates for the future mass production of organic electronics, for example, large area organic solar cells (OSC).1 Up to now, conducting organic thin films have been used to fabricate highly efficient organic light emitting diodes (OLED), organic thin film transistors, and OSC.2–5 Conducting polymers are also interesting for application in combination with inorganic nanostructures.6, 7 As an approach to further improve the various organic devices, it has been shown that modified surface structures can lead to an enhanced performance. For example, Na et al. were able to increase the overall power conversion efficiency of polymer:fullerene solar cells by 15% due to an additional surface structuring of the photoactive polymer film.8

Therefore, different structuring routines have been developed to fabricate polymer structures on the nanoscale and microscale. On the one hand, these routines are based on the solution processability of the polymer and, however, on the softening of the polymer after reaching its glass transition temperature. Kim et al. introduced a very well-known approach called micromolding in capillaries (MIMIC).9 This method puts the master structure, which is normally based on poly(dimethylsiloxane) (PDMS), on the substrate and surrounds it by polymer solution. Due to capillary forces, the microstructures are filled constantly with solution and after complete solvent evaporation the mold can be removed leaving the final polymer structures on the substrate. This approach guarantees a perfect inverted replication of the applied master structures and has already been used to investigate organic field-effect transistors based on microrods and nanorods.10, 11 In general, MIMIC has the advantage that it does not need any high temperature or vacuum steps, which could be harmful to the polymer and would be contradictory to the needs of a cost-effective device fabrication. However, a serious drawback of the MIMIC routine is that only freestanding polymer structures can be shaped. Hence, a complete surface coverage by the polymer is not possible. In photoactive applications like OSCs or OLEDs, the polymeric material is normally sandwiched between a bottom and a top electrode. Consequently, the absence of a homogeneous polymer layer would result in short-circuit faults.

Therefore, (nano)imprint lithography (NIL) is a more appropriate approach. For several decades NIL has been widely used, allowing high aspect ratio imprints at very good accuracies.12 For this method, the thermoplastic properties of the polymer are used by pressing a mold into a polymer film, which is heated above its glass transition temperature T_g. Above T_g, the polymer behaves like a viscous liquid and can flow under pressure. Thereby, the inverse master structure is replicated in the polymer film. Using NIL based on molds made from fused silica or amorphous metals imprinted structures have been reported down to the size of 10 nm.13 Recently, the same method has also been used to mimic the phase separation in bulk heterojunction solar cells to understand the influence of the interfacial area between the two components on the charge carrier generation and on the actual device performance.14 Since a sample treatment step at an elevated temperature is mandatory during the NIL structuring routine, many conducting polymer thin films lose the potential to tune their electrical conductivity via defined post-production thermal annealing steps. As an example, the electrical characteristics of thin film field-effect transistors based on poly(3-hexylthiophene) (P3HT) depend strongly on additional thermal annealing steps, which lead to a crystallization of the P3HT polymer chains and hence an improved electrical conductivity.15, 16 McNeill et al. reported similar results for all-polymeric solar cells.17 In their case, thermal annealing of the complete device increased the overall power conversion efficiency from 0.14 to 1.20%. Annealing of the photoactive films before the deposition of the top electrode only led to smaller improvements in the efficiency. Since conducting polymer films structured via NIL already undergo thermal treatment steps, similar effects, which are individually regulated by annealing temperature and time, are impossible.

Thus, despite the existing and well-developed structuring techniques, additional simple imprinting methods, which fulfill the needs of low-cost organic electronics and do not require any heat treatment steps, are desirable. In this article, we therefore introduce a structuring approach, which does not require any high temperature steps, and apply it to conducting polymer thin films. Based on this route, we obtain structured films, which are used to investigate the influence of posterior thermal annealing on the initial polymer morphology.

This article has the following structure: In the experimental section, brief descriptions of the applied materials and of the experimental methods are given. The results and discussion section focuses on a detailed explanation of the sample preparation technique for the structured polymer films and on the influence of polymer film thickness variations on the obtained structures. The results of thermally treated structured polymer films are also presented. The article concludes with a summary of the results and a short outlook.

EXPERIMENTAL

Sample Preparation

The samples are made of the conducting polymer poly(dioctylfluorene-co-benzothiadiazole) (F8BT), which was purchased from American Dye Source (Baie D'Urfé, Canada). The molecular weight of the F8BT is 16 × 10³ g mol⁻¹ and its polydispersity is 5.6. Toluene was used as the solvent and the polymer was stirred permanently for at least 12 h to guarantee perfect solution conditions. A silicon master with channels of 2 μm width at a periodicity of 4 μm and a depth of 150 nm was applied as structured substrate. The master was fabricated by hard lithography and reactive ion etching and the total structure size was 20 × 20 mm². Standard silicon substrates were applied for the final flat supports. All substrates (master and flat support) were cleaned before coating (spin coating and floating) in an acidic hot bath at 80 °C for 15 min.18 Subsequently, all substrates were rinsed with deionized water and dried under constant nitrogen flow before coating. The structured master substrates were coated with a Süss MicroTec Delta6RC spin coater at 2000 rpm for 30 s under ambient conditions (T = 20 °C and RH = 40%). Therefore, the polymer solution was spread on the master before starting the spinning. For a substrate size of 20 × 20 mm² a solution volume of 0.25 mL was appropriate. Afterwards, the prestructured polymer films were floated off in deionized water and picked up with the flat support. Detailed information about the structuring routine can be found at the beginning of the results and discussion section. As reference samples, F8BT has also been spin cast on flat silicon supports according to the given experimental conditions. Polymer concentrations of 10, 20, and 26 g/L lead to film thicknesses of 42, 91, and 122 nm. To investigate the influence of thermal annealing on the surface properties of the structured films, the samples were placed on a hot plate in ambient atmosphere. The temperature was successively increased from 25, over 100, 110, and 120 to finally 130 °C. The temperature of the hot plate was increased with a rate of 100 K min⁻¹ for the first large temperature step and with a rate of 75 K min⁻¹ for the following smaller temperature jumps. Since the samples were kept at each temperature step for 45 min, the heating time can be neglected. The hot plate was equipped with an air pressure cooling system, which allowed the sample to be cooled from 130 to 60 °C with a rate of −50 K min⁻¹. For the further temperature decrease, the rate is reduced to −20 K min⁻¹.

Atomic Force Microscopy (AFM)

The surface topography was imaged with atomic force microscopy (AFM) in tapping-mode condition. The measurements were performed with an Autoprobe CP Research AFM instrument. All measurements were performed in ambient air and gold-coated silicon cantilevers (Ultralever cantilevers) with a resonance frequency of 75 kHz and a spring constant of 2.1 Nm⁻¹ were used. The curvature of the tip is small compared with the measured structure size. Each scanned image consists of 256 lines scanned at 1.0 Hz and has a size of 10 × 10 μm². To confirm the film homogeneity over a large area several images were taken at different sample positions. Due to a scanner-tube movement, a background subtraction had to be performed from the raw data. AFM is also used to measure the polymer film thickness of the structured films by adding an artificial scratch.

Grazing Incidence Small Angle X-ray Scattering

The in situ change of the structured polymer film surface during the heat treatment was investigated with grazing-incidence small-angle X-ray scattering (GISAXS).19 The experiments were performed at the beamline BW4 at HASYLAB, DESY (Hamburg, Germany).20 Figure 1 shows the schematic representation of the GISAXS experiment. Along the x-axis, the X-ray beam is directed on the surface of the sample under an incident angle α_i, which was chosen to be well above the critical angle of the polymer. This guarantees a full penetration of the X-rays through the polymer film. Hence, the structure of the surface and the inner film can be recorded with good statistics. The important parameters of the set-up are the sample-to-detector distance SDD, the angle of incidence α_i, the out-of-plane angle ψ, and the exit angle α_f. The wavelength of the X-rays was set to 0.138 nm and the SDD was 2.024 m. The beam had a size of 40 × 80 μm² (vert. × hor.) and the samples were placed horizontally on a goniometer, which was equipped with a hot plate. The incidence angle was set to a value of 0.405°. GISAXS is collecting structural information averaged over the whole footprint of the incoming X-ray beam, which illuminates the sample. The scattering signal was recorded by using a two-dimensional MarCCD camera with 2048 × 2048 pixels and a pixel size of 79 × 79 μm². A beamstop was placed at the position of the specular reflected beam to allow for longer measuring times and, hence, improved statistics. During thermal annealing, the samples were probed in situ with GISAXS. Directly after each temperature step short GISAXS measurements (counting time of 1 min) were performed. They have shown that an overall waiting time of 30 min at each temperature level was appropriate to reach a constant status of the sample with a constant scattering signal. After this dynamic sample evolution phase, the static GISAXS data were recorded with an accumulation time of 10 min. The IsGISAXS software by R. Lazzari was used for the analysis of the two-dimensional (2D) GISAXS data.21 All simulations are based on the distorted-wave Born approximation (DWBA). DWBA includes first order perturbations in the scattering process, which are induced by particle roughness or contrast variations.22 To mimic the channel structure of the investigated samples, a model that consists of anisotropic pyramids on top of a homogeneous polymer layer (film thickness d) is applied in the simulations. The pyramids are positioned on the lattice sites of a 2D regular lattice and their crucial dimensions are the height h, the width w, and the base angle φ of their inclined side walls. The lattice properties are constant and the length of the pyramids is equal to the lattice spacing. This guarantees a direct connection between the objects in one direction and, hence, results in a channel-like model. For the IsGISAXS simulations, following values of indexes of refraction have been used: (i) F8BT: δ = 2.787 × 10⁻⁶ and β = 8.944 × 10⁻⁹ and (ii) silicon: δ = 6.085 × 10⁻⁶ and β = 1.410 × 10⁻⁹.

Figure 1. Schematic representation of the GISAXS experiment: The incident X-ray beam hits the horizontally placed structured sample under an incident angle α_i. The reflected beam is depicted by the exit angle α_f and the out-of-plane angle ψ, which define the components of the scattering vector along the q_y- and the q_z-axis, respectively. A two-dimensional detector is used to record the scattered intensity. In addition, the coordinate system for the sample position is shown including the rotation angle ρ around the z-axis.

Optical Ellipsometry

The instrument of choice was an imaging ellipsometer (SPEM) from Nanofilm/Accurion, which was embedded in the GISAXS set-up. Further information on the experimental set-up that combines GISAXS and imaging ellipsometry can be found in literature.23 The instrument is a single wavelength ellipsometer (λ = 532 nm) with a PCSA configuration and it is suited for nulling measurements. The imaging measurements are performed at an angle of incidence of 42° using an objective with a 20× magnification. This configuration leads to a resolution of 0.27 μm per pixel.

RESULTS AND DISCUSSION

Structuring Routine for F8BT Films

Figure 2 shows the different steps of the imprinting routine. It is based on the idea of transferring a prestructured polymer film on a solid support. Therefore, a solid master structure is coated with a thin polymer film. Spin coating is very well suited because it allows a good control of the film thickness via the polymer concentration in solution or the spin parameters. Depending on the different structure properties, various materials are suitable for the master mold. One material limitation lies in its resistivity against the solvent of the polymer solution. In this study, a simple hard master made of silicon was chosen. Silicon has the advantage that it is a standard material for photolithography and reactive ion etching. Due to its chemical stability against different common solvents, it is also well suited for multiple reuses. Figure 2 already implies the wavy surface of the polymer film on top of the structured substrate. It was found that for thicker polymer films the peak-to-valley ratio of the wavy surface is reduced. However, a complete flattening was not observed for the investigated polymer films. In the following step, the floating approach is used to lift off the already prestructured polymer film from the master mold. Therefore, the coated master structure is positioned under an inclined angle in a water reservoir and the water level is slowly increased. Since the adhesion forces of the water to the substrate are stronger than the ones between the polymer and the substrate, water is forced between polymer and silicon. As the water level is increasing further, the polymer film is peeled off the substrate. After the complete polymer film has been removed from the substrate, it floats on the water surface with the channel structures facing downwards. In the next step, the water level is steadily decreased and the structured polymer film is picked up by the flat substrate. We have found that for the applied master structure dimensions similar polymer surface structures were obtained whether the floating polymer film was picked up with the structured side facing downwards or facing upwards. However, for the applied approach with the channel structures on the bottom side it is important, that the polymer film is gently dried after floating. The remaining water has to evaporate slowly, since the polymer film is still mobile on the substrate if remaining water is present. A gentle drying is also necessary, so that possible voids, which could result from evaporating water at the polymer-substrate interface, are adopted by the polymer film. GISAXS is applied to rule out the presence of such interfacial defects. Further information on this aspect is given below. Whereas the compressibility of the target polymer and its sticking coefficient to the stamp material is always a serious restriction for standard NIL, the only prerequisite in this approach is that the polymer-master-system is suited for the floating step. Given the fact that floatation is very common for the transfer of polymer films on, for example, silicon nitride membranes as used for different transmission techniques (e.g., transmission electron microscopy, TEM, or scanning transmission X-ray microscopy, STXM), a huge number of polymers have proven their abilities to fulfill this condition. Nevertheless, up to now only flat substrates were used for floating. In our investigation, we show that also structured substrates can be applied.

Figure 2. Idealized schematic overview of sample preparation: 1. A thin conducting polymer film is spin coated onto a structured silicon substrate. 2. Afterwards, the prestructured film is floated off using deionized water. 3. Then the structured substrate is replaced by a flat support. 4. The water level is decreased. 5. After drying, a structured polymer film is left on the flat support.

To demonstrate the functionality of this structuring protocol, poly(dioctylfluorene-co-benzothiadiazole) (F8BT) is chosen as an example material. This conducting polymer is already widely used for different optoelectronic devices and its properties are well-known.24–26 Since F8BT as well as its derivative F8TBT are electron conducting polymers, they play a key role for the development of all-polymeric optoelectronic devices, which require the combination of electron and hole conducting polymers.17, 27, 28 F8BT shows good solubility parameters in different solvents29, 30 and has a glass transition temperature of T_g = 130 °C.31 Also other common conducting polymers possess glass transition temperatures close to the value of F8BT (e.g., T_g (P3HT) = 110 °C).32 A similar T_g is important because it allows a good comparability of the thermal annealing experiment with other polymer types.

Surface Tuning of Structured F8BT Films

Figure 3 shows a schematic illustration of a structured F8BT film including the necessary structure dimensions. Since only elongated channel structures are investigated, the given profile view in Figure 3 depicts the relevant properties. The film thickness of the homogeneous polymer film below the surface structures is given by d, the channel width by w, the periodicity by b, the height of the channels by h, and φ describes the base angle of the inclined sidewalls of the fabricated channel structures. To investigate the surface topography of the final floated microstructured F8BT films, AFM is used (further details about the silicon master structure are given in the experimental section under sample preparation). Figure 4(a,c) show the film surface for three different structured polymer films made from polymer solutions with concentrations of (a) 10, (b) 20, and (c) 26 mg/mL. All samples are perfectly homogeneous over the total area of the master mold (20 × 20 mm²) and possess a well-defined valley-peak transition. Optical microscopy did not reveal any general cracks on a macroscopic scale either. Hence, the pre-structured films are stable enough for the floating and the water drying steps, which are the most critical ones during sample preparation. For the three different samples, the film thickness d of the homogeneous F8BT layer below the surface structure was determined with AFM. The obtained values for d are in very good agreement with the film thickness of films spin coated on bare silicon supports: (a) 42, (b) 91, and (c) 122 nm. Depending on the film thickness d, the AFM images indicate already modified channel widths w. For an enhanced illustration of the different channel heights h, the corresponding profile cuts from the AFM data are shown in Figure 4(d). Cuts perpendicular to the channel structure are selected. They show an increasing channel width w for an increasing polymer film thickness d. The corresponding values of w, which are taken at the medium channel height, are 2.1, 2.6, and 3.1 μm for the samples with increasing film thickness. In addition, the absolute channel height h, which can be calculated as the difference between medium channel and valley height, is also increasing to 18, 22, and 33 nm with increasing film thickness. However, for all investigated samples, the polymer film between the channel structures flattens out (deviations in height below 5 nm). In comparison to standard structuring methods such as NIL or MIMIC, for which the obtained polymer structure is always an inverted replica of the master structure, the changes in the characteristic topology parameters w and h make the presented structuring routine unique. On the contrary, in NIL13 and MIMIC10 different master molds are necessary for different surface topologies. Such molds are often the limiting factor for standard imprinting methods. Therefore, our observations are very promising since this imprinting method allows the fabrication of structured polymer films with different surface topographies by only tuning the polymer film thickness and using the same master mold. Moreover, the master can be reused multiple times. Although only channel-like surface imprints are shown in this article, other surface morphologies, such as, for example, pillars or cubes appear possible in principle, as well. As a drawback of this routine, one might identify that standard NIL allows direct replica of the used master molds, whereas our method always results in deviations between the exact length scales of the master and the prepared structured polymer film. For example, already for the thinnest examined polymer film, a small difference between the channel width w of the structured polymer film [w (pol) = 2.1 μm] and the corresponding channel width of the equidistant master mold [w (mas) = 2 μm] is found. However, with a proper selection of the master structure the desired polymer structure is accessible. In addition, it has to be expected that high aspect ratios—as they are observed for structured polymer films prepared with NIL or MIMIC33–35—are not easily possible with our novel method. Recently, we have shown for different spin coated conducting polymer blend films that the structure installed by phase separation strongly depends on the film thickness.36 Hence it can be expected that structured conducting polymer blend films with a laterally modified phase separation can be fabricated with this method as well.

Figure 3. Illustrated profile cut of a structured F8BT film taken perpendicular to the channels including the important structure dimensions: w corresponds to the channel width, b to the periodicity, φ to the base angle of the inclined side walls, d to the film thickness of homogeneous polymer underlayer, and h to the channel height.

Figure 4. Topographic AFM data of the structured samples with a homogeneous polymer film thickness d of (a) 42, (b) 91, and (c) 122 nm. In (d) the corresponding profile cuts along the white line in the AFM data are drawn. They indicate a decreasing valley width w and an increasing channel height h for and increasing film thickness d.

Scattering Analysis of Structured Films

An additional strong advantage of the structuring method lies in its application for large areas in combination with a high reliability. Since AFM provides information of a microscopic sample area only, experimental methods suitable for larger areas, which can still resolve nanoscopic and microscopic objects, are desired. Therefore, GISAXS is a well suited tool, which is gaining more and more attraction for the investigation of hard and soft matter nanostructures.37–42 Since a small incidence angle α_i of 0.405° is chosen for the GISAXS measurements, the footprint of the impinging X-ray beam with a height of 40 μm is extended to a distance of almost 6 mm on the sample along the beam direction (in x-direction, see also Fig. 1). Therefore, a parallel alignment of the channel structures to the incoming X-ray beam is crucial to guarantee a symmetric scattering signal on the two-dimensional detector, which is the prerequisite for a simple analysis of the scattering data. Figure 5 shows the GISAXS data of the sample with a film thickness d of 91 nm for different channel orientations with respect to the X-ray beam. The rotational angle ρ (see Fig. 2) between the channel structure and the X-ray beam is varied in five equidistant steps from −0.30° to 0.30°. For the center image, the scattering signal appears symmetric and, hence, the rotational angle is defined as ρ = 0°. Due to the strong influence of slightly disoriented channel structures on the GISAXS signal, the sample is aligned with an accuracy of 0.01° for the rotational angle ρ. In general, the GISAXS data in Figure 5 are dominated by three different features: inclined truncation rods under different angles, a circularly shaped intensity maximum around the specular peak/beamstop and prominent side maxima at the Yoneda peak.43 The Yoneda peak depicts a maximum in scattering intensity, which is defined by the critical angle α_crit of the investigated material. In case of F8BT, the Yoneda peak is located at an exit angle of α_f = α_crit = 0.135° and an out-of-plane angle ψ of 0° (see Fig. 5). The prominent inclined truncation rods result from the scattering on the inclined surface planes, which are found for our structures at the valley-channel transition (see AFM images in Fig. 4). In general, this feature does not depend on very small changes in orientation ρ of the channels and therefore the angles of the truncation rods remain constant in Figure 5. However, already for slightly higher rotational misalignments the truncation rods disappear rapidly since their reciprocal space width is small. A closer theoretical interpretation of this feature is given in the following section. The circular intensity maxima around the specular peak can be explained by the cut with the Ewald sphere.44, 45 More information can be found in the work by Yan et al., where they investigated the intersection of the grating truncation rods with the Ewald sphere by combining the corresponding mathematical expressions for the Ewald sphere with the equation for the maxima given by the specular grating truncation rods.46 In the same work a mathematical model is developed, which allows the simulation of scattering patterns of slightly misaligned channel structures. Their model also shows a strong dependence of the scattering image on a perfect sample alignment, as it was observed in our GISAXS experiments. The third prominent feature, the side maxima at the position of the Yoneda, is given by the properties of the channel structure, as well. Therefore, especially the base angle φ and the height h of the channel structure contribute most to the scattering around the Yoneda. Interestingly, Yan et al. have not observed this feature for their channel structures based on silicon.46 Nevertheless, the results from the rotation series (Fig. 5) show that experimental effort is necessary to perfectly align the channel structures parallel to the impinging X-ray beam with accuracy in rotation smaller than 0.1°. Otherwise, no symmetric scattering images are obtained, which can later be fully evaluated with theoretical simulations.

Figure 5. Series of GISAXS measurements of the structured F8BT film with a film thickness of 91 nm probed at different rotation angles ρ from −0.30° to 0.30°. For a rotational angle ρ of 0°, the scattering pattern is perfectly symmetric with respect to the q_z-axis.

Thermal Annealing of Microstructured F8BT Films

Since devices based on conducting polymer films generally show improved characteristics upon thermal annealing, its influence on the polymeric microstructures has to be investigated. In the upper row of Figure 6, the GISAXS data of the thermally treated structured F8BT film with a film thickness d of 91 nm are shown. The corresponding IsGISAXS simulations are presented in the bottom row. The sample structures are aligned parallel to the X-ray beam and the annealing temperature is successively increased from 25 to 100, 110, 120, and finally to 130 °C (Fig. 6 from left to right). More experimental details about the annealing routine are given in the experimental section. Due to the complexity of the GISAXS data a comparison of the two-dimensional scattering data appears better suited than to focus on the typical line cuts, which only contain information along one q_y- or q_z-component. In Figure 6, a clear trend from a broad scattering signal with prominent truncation rods to a more and more featureless scattering intensity image is observed. Recently, Rebollar et al. applied GISAXS to laser-induced periodic surface structures to assess morphology order over large sample areas.47 They also used IsGISAXS simulations to reconstruct the surface morphology. However, for their gratings no truncation rods have been observed. For the F8BT channels at T = 25 °C not only prominent truncation rods, a circular scattering intensity around the specular peak, and distinct side maxima at the position of the Yoneda peak are observed, but also two eye-catching intensity islands located below the critical angle of the polymer at an out-of-plane angle ψ of −0.12° and 0.12°. The corresponding IsGISAXS simulation reveals that their position is given by the dimensions of the simulated channel structure. It is important to note that for all IsGISAXS simulations the polymer structures (here anisotropic pyramids) are placed on the lattice sites of a constant two-dimensional lattice. Complementary optical microscopy images also confirm the high long range order of the fabricated channel structure. Because the IsGISAXS simulation software is based on diffuse scattering only, it does not include the additional intensity at the specular peak position. As a consequence, all simulated GISAXS images lack the circular intensity maxima resulting from the intersection of the specular peak with the Ewald sphere. Nevertheless, a good congruence between the simulated and the measured data is achieved when focusing on the truncation rods and the side maxima. This is also a good proof that no further voids (from residual air or water) are present at the polymer-substrate interface. Due to their large contrast in scattering density such interfacial structures would contribute strongly to the scattering pattern.

Figure 6. Series of GISAXS measurements (top row) of the structured F8BT film with a film thickness of 91 nm probed at different temperatures values (25, 100, 110, 120, and 130 °C). The corresponding two-dimensional simulations are illustrated in the bottom row. All simulations are based on DWBA, using models consisting of anisotropic pyramids with varying base angles and structure heights and widths positioned on regular lattice sites.

Figure 7 shows the important dimensions of the anisotropic pyramids as they are used in the IsGISXAS simulations for the different temperature steps. In Figure 7(a), the base angle φ of the channel structure is plotted against the temperature. It can be seen that the base angle is decreasing rapidly from 22 to 6.3° for the first temperature step, whereas it is decreasing slowly to 2.6, 1.8, and 1.4° for the higher temperatures of 110, 120, and 130 °C. Even on a longer time scale, only minor changes in the scattering signal have been observed for temperatures below the starting value of 100 °C. Since the base angle of the channel structure is directly linked to the angle of the prominent truncation rods, only small errors in the IsGISAXS simulations have been observed for this value. In addition Figure 7(b) presents the height h and the width w of the channel structure for the different temperatures. For increasing annealing temperatures the simulated channel heights decrease from 38.5 to 25, 15.1, 13.3, and finally to 11.2 nm. This corresponds to a dramatic relative change in height of over −70%. In comparison, the thermal influence on the width is only ∼+19% relative to the width of the as-prepared sample. For the first temperature step, it changes from 2.6 to 3.0 μm before it reaches a constant value of 3.1 μm. For the structured polymer film at 25 °C, it is found that the simulated channel width w is in good accordance with the results from the AFM data, whereas the simulated channel height h varies considerably from the topological AFM value (see Fig. 4). Due to the fact that in general the influence of the height h and the width w on the scattering signal is in comparison to the impact of the base angle φ only minor, both values from Figure 7(b) are characterized by a relatively large error given by the IsGISAXS simulations. Nevertheless, the large discrepancy in channel height for the as-prepared F8BT films [h (AFM) = 22 nm vs. h (GISAXS) = 38.5 nm] must also be a result of a mismatch between the applied perfect channel structure model for the theoretical simulations and the actual structure shape. Taking into account the profile line cuts from Figure 4(d), for example, inclined surfaces are not only present at the sidewalls of the channels but also between the channels and on top of the channel structure (formation of a dual-peak structure). Due to the softening of the F8BT channels during annealing, the model appears to fit the actual structure better and the discrepancy in channel height is reduced [after annealing: h (AFM) = 18 nm vs. h (GISAXS) = 13.3 nm; more information below]. As it was mentioned before, a constant thickness for the homogeneous F8BT layer below the polymer structures of 91 nm is assumed for all IsGISAXS simulations at the different annealing temperatures. Based on the conservation of the total amount of polymer, the extracted values for the channel width w and height h could imply a slight increase of the homogeneous layer thickness. This is however not detected within the GISAXS data due to the strong roughening of the F8BT surface upon thermal annealing and therefore not implemented in the IsGISAXS simulations (see below).

Figure 7. Resulting parameters of the IsGISAXS simulations of the sample with a film thickness d of 91 nm: (a) base angle φ of the channel structure plotted as function of the annealing temperature (dashed line shown as a guide to the eyes) and (b) height h (open symbols) and width w (filled symbols) of the anisotropic pyramids used for the simulations (dashed lines shown as guides to the eyes).

To the best of the authors' knowledge similar in situ GISAXS measurements of thermally annealed conducting polymer microstructures have not been performed yet. In contrast, Jones et al. studied the evolution of nonconducting nanoimprinted polymer patterns with critical dimension small-angle X-ray scattering (CD-SAXS).48 They investigated poly(methyl methacrylate) gratings and observed changes in the object shape only for temperatures above the bulk glass transition temperature. For the investigated F8BT structures, the evolution starts already well below the T_g of F8BT. Similar imprinted sinusoidal gratings made of PMMA and polystyrene (PS) have also been investigated by Johannsmann and coworkers using optical diffraction methods.49, 50 They have shown that the near-surface viscoelastic behavior of PMMA and PS depends on the chain length. This can result in a polymer flow of such channel structures, which already sets in at temperatures below the bulk T_g. However, structural changes due to annealing temperatures below glass transition are often reversible, whereas modifications at temperatures above T_g are irreversible.

In comparison to standard PS and PMMA structures, the surface gratings made of conducting F8BT show the tendency to form small crystallites. Therefore, the development of the F8BT crystals during thermal annealing is probed in situ with imaging ellipsometry. In Figure 8(a,e) the imaging ellipsometry images recorded at the different temperature steps are shown. Already at 110 °C the channel structure starts to lose its well-defined shape. The grainy surface morphology is a good indication for the initiation of the crystal formation. For the highest temperatures of 120 and 130 °C, a clear change to a washed out channel structure with an increased surface roughness is observed. This is an interesting effect, since not only the polymer chains crystallize and therefore pass over to an energetically more stable state, but also the superimposed artificial microscopic structures simultaneously vanish. As a consequence, a complex interplay between a macroscopic polymer flow and molecular rearrangement takes place during thermal annealing. The combination of GISAXS and imaging ellipsometry reveals that for a temperature of 100 °C mainly a polymer flow occurs whereas for higher temperatures the starting crystallization hampers this polymer flow. In addition, no further topological changes are observed during cooling of the sample to room temperature [see Fig. 8(f)]. This is in good accordance with the GISAXS measurements, which also reveal no further sample modifications due to the cooling of the sample. Nevertheless, a closer investigation of the spatial properties of the polymeric crystals is not possible within the resolution of the imaging ellipsometry set-up. Figure 9(a) therefore shows the sample surface of a thermally annealed structured sample measured with AFM. A strong formation of F8BT crystals, homogeneously distributed over the complete sample surface, is visible. No preferential crystallization between or on top of the channel structure is found. The size of the crystals is on a mesoscopic length scale, which is small in comparison to the channel size. In Figure 9(b), the corresponding line profile cut perpendicular to the channel structure is plotted as a solid line. The dashed line corresponds to a profile integrated over 3 μm vertically to the drawing direction. The line profile cut reveals surface depressions due to the crystals of up to 6 nm. Since such high surface roughness values increase the diffuse scattering in the GISAXS measurements, it explains the lack of scattering intensity at the Yoneda peak in the IsGISAXS simulations for higher temperatures (see Fig. 6).51 Unfortunately, such high roughness values could not be implemented successfully in the applied IsGISAXS models. However, a tangential fit to the integrated profile cut in Figure 9(b) reveals a gradient angle of around 1.1°, which is in quite good agreement with the equivalent base angle φ of the simulated channel structure. The formation of these mesoscopic crystallites can also be the reason that at annealing temperatures around the T_g of F8BT still slight channel superstructures are present.

Figure 8. Two dimensional imaging ellipsometry micrographs of microstructured polymer film with a homogeneous thickness d of 91 nm during heat treatment: (a) channel structure as prepared, (b–e) channel structure at the different temperature steps probed in situ with the corresponding GISAXS measurements from Figure 6 and (f) structure of the heated sample after cooling down to room temperature. The scale bar depicts a length of 20 μm.

Figure 9. (a) AFM image of the topography of the thermally annealed sample (initial film thickness d of 91 nm). The color code depicts a maximum difference in height of 25 nm. (b) The solid line illustrates the profile taken perpendicular to the channel structure along the white line from (a). The dashed line depicts the channel profile integrated over 3 μm and illustrates the averaged sinusoidal surface.

CONCLUSIONS

A structuring method based on the floating technique is presented, allowing the fabrication of structured polymer films at ambient conditions. It is especially interesting for photoactive conducting polymers since it guarantees a homogeneous underlying polymer layer as it is necessary for OSC or OLED and allows for a decoupled investigation of thermal annealing effects on conducting polymer structures. The thickness d of this underlayer can be easily controlled via the polymer concentration in solution. The properties of the surface structures can be tuned independently from specially modified master structures. By only modifying the polymer film thickness d, the width w of channels made of F8BT is varied between 2.1 and 3.1 μm and their corresponding height h is changed between 18 and 33 nm.

The influence of thermal annealing on such a structured F8BT sample is probed in situ with GISAXS and imaging ellipsometry. Therefore, a rotational series is presented, which highlights the importance of a perfect channel alignment parallel to the impinging X-ray beam. Based on IsGISAXS simulations a complete three-dimensional reconstruction of the channel properties is possible for different temperature steps of 100, 110, 120, and 130 °C. The important dimensions such as base angle φ and height h of the surface structures decrease with increasing temperature, whereas the channel width w slightly increases. By imaging ellipsometry and AFM, it was shown that for gratings made of conducting F8BT the polymer flow also sets in at temperatures well below the glass transition temperature and before the first formation of crystallites. Since similar structured polymer films have already proven their potential for actual device applications, the obtained results prove that additional focus has to be put on the post-production annealing step, even though it is performed below T_g. In the future, it is planned to adapt this structuring method to other surface structures such as cubes and cylinders.

Acknowledgements

The authors thank J. Wiedersich for his help during the BW4 experiment at HASYLAB, DESY. The financial support by BMBF project 05KS7W01 is gratefully acknowledged. M.A. Ruderer gratefully acknowledges the funding by the Bavarian State Ministry of Sciences, Research and Arts through the International Graduate School “Materials Science of Complex Interfaces” (CompInt).