Organic semiconductors, comprising small molecules and conjugated polymers, are a promising materials family for next generation electronic devices such as transistors, light-emitting diodes and solar cells. Processes at interfaces play a very important role in the operation of these devices. Because in organic semiconductors the microstructure is intimately tied to their electronic properties, understanding the effect of surfaces and interfaces on the morphology of these materials is an important component of optimizing their performance. In thin film transistors (TFTs) for instance, mobile charge is transported within a few nanometers of the semiconductor/dielectric interface, which in the case of bottom-gated devices is the substrate surface. There is significant experimental evidence demonstrating that controlling the molecular packing and orientation at the dielectric interface can improve the charge carrier mobility in thin films of poly(3-hexylthiophene) (P3HT), a model semicrystalline, semiconducting polymer, by several orders of magnitude.1, 2 Previous studies have demonstrated that high mobilities can be realized in many semiconducting polymers through modification of the dielectric interface with a self-assembled monolayer (SAM), deposition from high boiling point solvents, and thermal annealing.3–7 The effects of these processing parameters on the microstructure of the polymer however is not well understood at a quantitative level, especially at the interfaces. In fact, while significant improvements in device performance have been achieved, a primarily qualitative picture has been painted of the changes occurring at the dielectric interface as a result of these different processing techniques.
It has been suggested that overall crystalline order plays a significant role in charge transport. Although many studies have attempted to quantify the total degree of crystallinity (DoC) in polymer thin films using a variety of techniques, there is no consensus as to what is exactly measured.8–11 For instance, a thorough analysis of the UV–vis absorption spectrum of P3HT allows us to infer the fraction of aggregated polymer.12 Crystallites and aggregates are however different entities. Although aggregates comprise a few π-stacked conjugated segments that modify photophysical properties, crystallites are composed of enough π-conjugated segments to result in diffraction peaks. As such, aggregates can in principle be found in the non-crystalline regions of the film and a measure of film aggregates is not equivalent to a measurement of film crystallinity.13 Furthermore, most studies measure bulk crystallinity, which may not reflect the microstructure at interfaces.
To draw comparisons between the microstructures resulting from various processing conditions, a quantitative method is needed to characterize the changes that are occurring in molecular packing, DoC and orientation. In this manuscript, we present a quantitative X-ray diffraction (XRD)-based method that allows a thorough characterization of polymer thin film microstructures. XRD has been used extensively to analyze polymer microstructures.14 Although there exist some important examples of using quantitative XRD to study semicrystalline, semiconducting polymer microstructures,11, 15–19 X-ray analysis in this field is often limited to qualitative observations. Moreover, only a few studies are able to make direct observations about interface structure using XRD.17, 18
The quantitative XRD technique presented here allows one to discern the microstructure at the interfaces and compare it to the microstructure found in the bulk of polymer thin films. The technique involves the measurement and analysis of XRD-based pole figures for the quantification of overall film texture and DoC as a function of film thickness. A vertically stratified, three-layer model can accurately describe the crystallinity data, and is used to explain the crystallization behavior in the P3HT films. The model suggests that while there are highly textured crystallites that nucleate directly from the dielectric interface, the film crystallinity is significantly depressed in the immediate vicinity of this interface. We also relate the microstructural characterization to charge transport measurements obtained through the use of field effect transistors (FETs). The transistor measurements highlight the importance of the formation of transport paths through crystalline regions at the interface.
All polymers and solvents were received and used without further purifications. Thin films for field-effect devices were grown on highly n-doped Si wafers with 200 nm of thermally grown oxide and a SAM of octadecyltrichlorosilane (OTS). Substrates were sonicated for 10 min each in acetone, methanol and isopropanol, followed by a DI water rinse, a drying step and a 20 min. UV-ozone treatment. OTS treatment was performed by submerging cleaned substrates into a 5 mM solution of OTS in hexadecane for 8–10 min. Followed by subsequent rinsing in heptane, acetone and isopropanol. Films of regioregular P3HT (Mw = 108 kDa; PDI = 1.61) with different thicknesses were produced by spin casting at 800 rpm for 2 min. from polymer solutions in 1,2-dichlorobenzene with concentrations ranging from 0.06 to 1.0 wt % that had been filtered through an Acrodisc 13-mm syringe filter with a 0.45 μm PTFE membrane. Polymer solutions were allowed to sit on the OTS treated substrates for 1 min. prior to spin casting. Finally, 80 nm Au contacts were thermally deposited onto the polymer films at a rate of 1 Å s−1 to form devices with channel lengths of 50–200 μm.
Device characterization was carried out in a vacuum probe station evacuated to 10−4 mbar at temperatures ranging from 80 to 300 K. Mobilities were extracted from transfer curves in the linear regime at Vd = −10 V for room temperature measurements and Vd = −2 V for temperature dependent measurements. Film thicknesses were determined using Atomic Force Microscopy in non-contact mode with a XE-70 Park system and silicon tips with spring constants of 45 N m−1 and resonant frequencies of 335 kHz. Thickness measurements obtained by AFM were confirmed by X-ray reflectivity.
X-ray scattering was performed at the Stanford Synchrotron Radiation Lightsource on beam line 7–2 (high resolution grazing incidence), 2–1 (high resolution specular scan), and 11–3 [two-dimensional (2D) scattering with an area detector, MAR345 image plate, at grazing incidence]. The incident energy was 8 keV for beam lines 7–2 and 2–1, and 12.7 keV for beam line 11–3. The diffracted beam was collimated with 1 milliradian Soller slits for high resolution in-plane scattering and with two 1 mm slits for specular diffraction. For both grazing incidence experiments, the incidence angle was slightly larger than the critical angle, ensuring that we sampled the full film depth. Scattering data are expressed as a function of the scattering vector q = 4π × sin(θ)/λ, where θ is half the scattering angle and λ is the wavelength of the incident radiation. Here, qxy (qz) is the component of the scattering vector parallel (perpendicular) to the substrate.
RESULTS AND DISCUSSION
P3HT films ranging from 5 to 100 nm thick spun on OTS-treated SiO2 and untreated SiO2 were characterized using various XRD geometries. Specular diffraction was used to characterize out-of-plane lamellar spacing and calculate the crystalline coherence length in the direction normal to the substrate surface [Fig. 1(A,B)]. Three orders of the (h00) peaks, attributed to the alkyl-stacking repeat, are visible for films on OTS treated SiO2. When films are spun on untreated SiO2, the third order is observed only in the thicker films. The crystalline coherence length can be estimated from FWHM values of these peaks using an approach based on the method of integral breadths [inset of Fig. 1(A,B)].20 Using this technique, for films thicker than approximately 10 nm, the (100) crystalline domain size in the direction normal to the substrate can be estimated as approximately 10 nm for films on OTS-treated SiO2, and slightly larger (12 nm) for films on bare SiO2. A coherence length of 8–10 nm for film thicknesses greater than 20 nm on OTS-treated SiO2 was confirmed using a more rigorous approach based on the Warren–Averbach Fourier transform peak shape analysis technique.21, 22 These values of domain size are in agreement with values found in literature.18
Specular geometry only collects diffraction from a small slice of reciprocal space and will miss details of film texture, namely diffraction from misoriented crystalline domains. Representative 2D-GIXD images of the thinnest film (ca. 6 nm) and the thickest film (ca. 100 nm) are shown in Figure 1(C) (OTS-SiO2) and (D) (SiO2) (see Supporting Information Fig. S1 for all of the 2D-GIXD images). The diffraction patterns show three orders of (h00) peaks lying along the vertical qz axis. Also, visible is the (0k0) pi-stacking peak along qxy axis (ca. 1.7 A−1). This fiber texture is typical for thin spin-cast films of semicrystalline P3HT and other polythiophenes.1, 23, 24 In the 2D-GIXD images, off-axis diffraction is manifest as arcing associated with the family of (h00) Bragg reflections. For films on both OTS-treated and untreated SiO2, arcing of the (h00) peaks increases with increasing film thickness, giving qualitative evidence of the degradation of the crystalline fiber texture as the films become thicker. Indeed, the decrease of P3HT texture on increasing film thickness has been noted previously.11, 17–19
Although 2D-images are important for qualitative texture characterization, raw intensity collected with a flat detector in grazing incidence geometry is distorted,25, 26 and as a result the detector image is not a direct map of reciprocal space. The distortion makes it difficult to use the intensity on the detector for quantitative analysis. We have developed a method to measure XRD pole figures for fiber textured thin films using three separate measurements, taking advantage of the benefits of an area detector.15, 16, 27 By appropriately combining data from grazing incidence diffraction patterns with data from local-specular diffraction patterns (where the proper choice of incidence angle prevents distortion near the Bragg reflection),27 we obtain diffracted intensity that represents the true intensity of a Bragg reflection across the entire span of polar angles (−90° to 90°). The polar angle χ is defined such that 0° corresponds to the substrate normal and 90° corresponds to the substrate plane. In the present case, the polar angle is the only relevant angle, as our films are isotropic in-plane and the diffracted intensity is in turn independent of the azimuthal angle. Because of the resolution limits of a 2D area detector, a high-resolution measurement in rocking geometry allows us to capture a “resolution limited peak” (vide infra) which provides information about the fine structure near χ = 0°, and allows us to be quantitative when comparing intensities between different samples of the same material. Additional details of the measurement method are discussed in Supporting Information.
Pole figures of P3HT films of varying thicknesses are shown in Figure 1(E,F) for films on OTS-treated SiO2 and untreated SiO2, respectively. Data are normalized for thickness in order to display the average DoC per nanometer of the film. For all samples, the intensity is highest at χ = 0° and quickly decreases as the value of χ increases, confirming the texture typical of spun cast P3HT films and other semicrystalline polythiophenes: the majority of crystallites are oriented with the (h00) repeat direction nearly perfectly normal to the plane of the substrate. As thickness is decreased, the width of the pole figures also decreases. This observation confirms in a more quantitative fashion what we observed with the 2D GIXS images: the fiber texture of the P3HT films improves as films are made thinner, for both the OTS-treated and the untreated SiO2 surface.
We can use pole figures to compare the relative DoC between films using eq 1 below.
Here, DoC refers to the DoC, t is the total film thickness, and Δβ and Δθ refer to the horizontal and vertical angular acceptance of the diffracted beam, respectively. The diffracted intensity is separated into two parts. The first term of the right hand side of eq 1 refers to the intensity collected at χ = 0° [which appears as a spike in Fig. 1(E,F)], within the angular resolution of our detector. We refer to this peak as the “resolution-limited peak.” Ipeak and Ibase correspond to the maximum and minimum intensity of the resolution-limited peak, which is modified by the angular acceptance of the diffracted beam (controlled by beamline geometry). The second term of the sum in the right-hand side of eq 1 refers to the diffracted intensity away from the resolution-limited peak, which varies more slowly with χ. This diffracted intensity is collected from the 2D image plate and is integrated over the acceptance angle. The sum of the thickness-normalized integrated intensity of the resolution-limited peak and the slowly varying peak is directly proportional to the DoC of the film. The relative DoC calculated from the complete pole figures for P3HT films on OTS-treated SiO2 and untreated SiO2 is shown in Figure 2(A,B). It should be emphasized that the values in Figure 2(A,B) are not absolute values of crystallinity because the unit cell of P3HT (and hence the structure factor) is not accurately known. However, we verify that P3HT crystallizes in the same polymorph in all films and a relative DoC can be measured. The most crystalline film in each dataset is given a value of unity, against which other films are compared. In this case, films spun on different surface treatments were treated independently, but the datasets can be directly compared: the thickest film on SiO2 is on average 84% as crystalline as the corresponding film on OTS treated SIO2, and the thinnest film on SIO2 is on average 63% as crystalline as the corresponding film on OTS treated SiO2. The average crystallinity increases with increasing thickness before beginning to level off in thicker films, with the thinnest P3HT film approximately 20% as crystalline as the thickest film, regardless of surface treatment.
In addition to providing a means to measure the DoC, pole figures simultaneously capture details of the crystalline texture. It is instructive to divide the pole figures into components based on different aspects of the crystalline texture and monitor separately the contributions of these components to film crystallinity. We attribute the resolution-limited intensity at χ = 0° to perfectly oriented crystallites nucleated at the very flat dielectric interface.1 This intensity is proportional to the “interface nucleated” crystallinity and is shown plotted against film thickness in Figure 2(C,D) for films spun on OTS and SiO2, respectively. Here, interface nucleated intensities are normalized by the average thickness of the crystallites (∼9 nm), or by film thickness for films thinner than 9 nm. The remainder of the pole figure intensity is attributed to crystallites nucleated elsewhere (such as off existing crystallites), which we call “bulk” crystallinity. Although not perfectly oriented, crystallites contributing to the bulk crystallinity are still textured, as illustrated by the majority of the intensity existing within a few degrees of χ = 0° [note the log scale of y-axis in Fig. 1(E,F)]. The contribution of this intensity is shown in Figure 2(E,F). This bulk intensity is normalized by film thickness. The normalized intensities in Figure 2(C–F) can be compared. Hence, we can discuss relative contributions of crystalline regions in the films to the overall crystallinity, according to film texture.
The relative contribution of bulk crystallinity to the total crystallinity increases with increasing film thickness, and in the case of thicker films (>15 nm), crystallinity from the bulk constitutes the largest contribution to overall DoC. In other materials systems, this behavior is often explained by the presence of disordered interface regions, which account for an increasing portion of the film as the overall thickness decreases.28–30 In our case, crystallization is more complicated: the semiconductor/dielectric interface is host not only to disordered material but also to interface-nucleated crystalline domains.
For films with a thickness greater than 15 nm the interface nucleated crystallinity remains nearly constant as a function of thickness. When film thickness is on the order of crystallite size (ca. 8–10 nm), there is a drastic increase in the population of perfectly oriented crystallites. The reason for this increase is not well understood, but has been observed previously.19 Below 10 nm, overall film crystallinity is strongly suppressed due to interfacial effects propagating throughout the full film thickness.
We can use a modified, vertically stratified, three-layer model that incorporates independent interface and bulk layers to calculate the relative DoC in our thin films. The model contains three fitting parameters: interfacial thickness (the sum of the polymer–substrate interface thickness and the polymer–air interface thickness), bulk misoriented crystallinity and interface misoriented crystallinity (which is located in the interfacial layer whose thickness is the fitting parameter described above). Additionally, there is a component to the crystallinity arising from perfectly oriented crystallites, which is extracted directly from the pole figures and is therefore not a fitting parameter. For our fitting procedures we ignore the peak in interface-nucleated crystallinity that occurs at film thickness near 10 nm, which was discussed separately above. As stated previously, the thickness of these crystallites is assumed to be equal to the crystallite size calculated using our XRD data. We assume that the amount of perfectly oriented crystallinity in the bulk of the film is negligible compared to that present at the interface, as surmised from the large ratio Ipeak/Ibase.
The relative DoC data for films spun on both OTS treated SiO2 and untreated SiO2 were fit using the model described by eq 2:
where T is the total average DoC per unit thickness, Pi is the contribution to DoC due to perfectly oriented crystallites nucleated at the interface, Mi is the contribution to DoC due to crystallites existing with the interface layer but not perfectly oriented, Mb is the contribution to DoC in the bulk, tc is the thickness of the crystallites in the alkyl stacking direction, ti is the combined thickness of the polymer–substrate and polymer–air near-interface layers (where crystallinity is suppressed), tb is the thickness of the bulk region of the film, and t is the total film thickness. All values of DoC contribution are in units that can be compared and are calculated per unit thickness. It should be noted that the substrate surface has two distinct effects: on one hand it causes the nucleation of a population of crystallites directly off the substrate interface, and on the other hand it inhibits the crystallization in a layer of thickness ti. Figure 3 illustrates the model and fitting parameters are given in Table 1. The fits to the crystallinity data for films spun on OTS-treated SiO2 and untreated SiO2 are superimposed onto data shown in Figure 2. The model is able to capture the major features of the overall crystallinity plots, as well as the individual contributions from highly oriented and bulk crystallinity, when treated as separate components.
Table 1. Parameters Used in Fits of Three-Layer Crystallinity Model to Experimental Data
The use of a three-layer model for describing the semicrystalline microstructure of the films is further supported by previous findings on the kinetics of crystallization in the presence of interfaces. A bilayer model has been previously suggested and has been shown to accurately describe the crystallization behavior in thin polymer films.31–33 The model consists of a bulk-like layer and an interfacial layer, referred to as a reduced mobility layer (RML) which exists in close proximity to the polymer/substrate interface. The RML is characterized by a reduction in molecular mobility and an increase in the local glass transition temperature due to the interactions of the polymer with the substrate. Together these factors act to inhibit polymer mobility during crystallization, significantly reducing the crystallization rate at the polymer/substrate interface. We suspect that this reduction in crystallization rate lowers the DoC near the interface relative to that of the bulk. Computer simulations of polymer melts also suggest disorder at the polymer/air interface, leading us to adopt a three-layer model. Indeed, a decrease in density and an alignment of chains parallel to the interface are found using a lattice Monte Carlo model within 1–2 nm of the polymer/air interface in a variety of semi-crystalline and amorphous polymers.34–36 It is likely that this perturbation remains upon crystallization and manifests as a less crystalline layer near the top interface. Together these experimental and theoretical results are consistent with the findings of our three-layer model. As we do not know the exact length scale over which these interfacial effects operate nor the extent of crystallinity suppression at the polymer–air interface relative to the polymer–substrate interface, our three-layer model assumes comparable misoriented crystallinity (Mi) near the polymer–air interface (ti,a) and near the polymer–substrate interface (ti,b), and estimates the combined thickness of the two (ti = ti,a + ti,b). These near-interface layers constitute the “interface misoriented crystallinity”: crystallites which exist in close proximity to the interfaces and may impinge on them, but are not nucleated on the substrate, else they would be perfectly oriented. The total combined thickness of the interfacial layers is found to be ∼6–8 nm, likely split with a few nanometers of inhibited crystallinity at both the top and bottom interfaces.
Interestingly, in our system, the perfectly oriented, interface-nucleated crystallinity appears to be decoupled from the bulk crystallinity. This observation implies that the surface chemistry of the buried interface has a local effect, which does not propagate more than one crystalline coherence length, or approximately 10 nm, into the film. As an illustration of this effect, we compare the pole figures of P3HT on OTS-treated silicon against those of P3HT on untreated silicon (Supporting Information Fig. S2). The pole figures are nearly overlapping for larger angles, but differ significantly near χ = 0°. Accordingly, the values for the bulk crystallinity are nearly identical, while the values for interface nucleated crystallinity are higher for films on OTS-treated substrates, shown in Figure 2(C–F). These data show that OTS acts as a nucleation layer and promotes the formation of strongly textured crystallites at the dielectric interface.1, 37 Moreover, the fact that the bulk crystallinity is identical for P3HT on OTS-treated silicon and P3HT on untreated silicon confirms that crystalline domains in the bulk do not extend to the buried interface, else the population would be affected by the interface surface chemistry. Our data is in agreement with previous observations by Porzio et al.,17 who found OTS to have an enhancing effect on P3HT chain alignment and crystallization, with the effect limited to within approximately 10 nm of the silanized interface. In contrast to OTS-treated surfaces, untreated SiO2 strongly inhibits the interface-induced crystallization of P3HT, as demonstrated by the very low density of interface-nucleated perfectly oriented crystals, approximately two orders of magnitude less than the crystallite density found in the bulk. Disorder in the presence of the SiO2 interface has been observed previously in P3HT films17 and in other non-conjugated polymers.38 The benefits of the silane treatment on P3HT microstructure likely arise due to the interactions of the alkyl chains in the polymer with those of OTS.37 As a result, the density of interface nucleated crystallites on an OTS-treated surface is approximately twenty times higher than on bare SiO2.
It has been suggested that structural order is inhibited at the polymer/semiconductor interface in P3HT due to quick aggregation of polymer chains during the spin-coating process.39 Our model suggests that interfaces inhibit the formation of crystallites and that the polymer/dielectric interface is approximately 30% as crystalline as the bulk in films spun on OTS and 20% as crystalline in films on bare silicon oxide. As discussed above, the bulk crystallinity is found to be almost equal for the two surface treatments, in agreement with our hypothesis of a three-layer model in which the interfacial chemistry only affects crystallization in a thin layer in close proximity to that interface. Although contradicting results have been reported as to the degree of structural order at the polymer/air interface relative to the bulk in thin films of P3HT,39–43 it has been suggested that a thin disordered region exists at the polymer/air interface as well.39 Ho et al.39 used near-edge X-ray absorption fine-structure spectroscopy (NEXAFS) to demonstrate disorder in both the thiophene ring orientations and hexyl side chains in thin films of regioregular P3HT within the technique's ∼2.5 nm probing depth of the polymer/air interface. Additionally, Porzio et al. used depth dependent XRD to show that P3HT is less ordered near the air/polymer interface than in buried regions of the film.18
The described three layer model has been developed using P3HT, but should be applicable to other flexible chain, semicrystalline polymers. However, differences in polymer chain architecture, including chain stiffness and sidechain density, may result in different crystallization habits. We have previously performed similar measurements in PBTTT, and found that crystallites exhibit comparable trends in texture as a function of thickness.16 Similar to P3HT, mobility also drops sharply in extremely thin films due a suppression of overall film crystallinity, however PBTTT mobility also falls in thicker films. We have attributed this to the higher inherent crystallinity of PBTTT which causes the connections between crystalline grains, and hence film texture, to have a greater impact on charge carrier mobility. Finally, PBTTT may also be thermally annealed to a liquid crystalline phase, after which crystallites extend through the film thickness, a situation without an analogous P3HT morphology.
In an attempt to understand the implications of semicrystalline film microstructure on charge transport, the electrical properties of P3HT thin films of varying thicknesses were characterized by fabricating FETs on OTS-treated, thermally grown SiO2. Figure 4(A) summarizes the field-effect mobility of P3HT films with different thicknesses, as extracted from transfer curves in the linear regime. We observe that mobility is correlated with interface nucleated crystallinity. Hole mobility does not change dramatically as a function of thickness for films much thicker than the characteristic crystallite size in the lamellar stacking direction (∼9 nm), despite the observation of an increasing overall DoC. Near 9 nm, a modest but reproducible increase in mobility is observed, which can be attributed to the increase in highly oriented crystallites observed in diffraction. Finally, the mobility drops quickly for films below 9 nm as the misoriented crystallinity decreases, as previously observed by Joshi et al.19 Although the interface nucleated crystallinity remains similar to that measured in thicker films, we expect that the near complete suppression of bulk crystallinity for very thin films, which would otherwise impinge at the interface and participate in transport, impedes overall film transport. In fact, very thin films are essentially entirely made of the more disordered interfacial layer and have no bulk layer. Finally, it should be added that we characterize the relative DoC by analyzing the XRD peaks corresponding to lamellar stacking. In the lamellar stacking direction, charge transport however is essentially suppressed. Nevertheless, lamellar-stacked crystallites also exhibit π-stacking, therefore the higher fraction of interface nucleated crystallites after treating the substrate surface with OTS implies a larger fraction of π-stacked molecules, which exhibit faster charge transport compared to amorphous polymer regions. Duong et al.13 also showed that the crystallites at the polymer/dielectric interface are more edge-on in character and more ordered along the polymer chain backbone, thereby helping with intergrain transport. When taken together with the results presented here, we conclude that the quality of the crystallites at the interface is better, but that the quantity is smaller.
It is well documented that TFT mobility of semicrystalline polythiophenes is higher when the surface is treated with an OTS monolayer.3, 23, 44 It has been shown that the origin of the effect is at least partially structural,45 which is supported here by the substrate surface dependence of interface nucleated crystallinity. We find that an OTS monolayer encourages the nucleation of highly oriented crystallites at the substrate interface, which is the important interface for charge transport in a bottom-gate TFT. In fact, while the total interface crystallinity is only 30% higher in P3HT films on OTS-treated SiO2 compared to bare SiO2, the interface-nucleated crystallinity is approximately 20 times higher on OTS-treated SiO2 compared to bare SiO2. A number of studies have documented the trend of better crystalline texture resulting in improved electronic performance.16, 23, 46–48 Our data confirms and expands that structural details of semicrystalline polymers have an important role in controlling charge transport.
To further elucidate the relationship between charge transport mechanisms and microstructure, we measure the temperature dependence of mobility for temperatures ranging between 80 and 300 K [Fig. 4(B)] in P3HT films of varying thicknesses deposited on OTS-treated SiO2. Complete modeling of the transport properties of the semiconductor requires the measurement of the mobility as a function of charge density as well as temperature. Such a procedure is beyond the scope of this manuscript however general information about the processes limiting charge transport can be inferred from Figure 4(B).
As commonly observed, mobility is thermally activated and can be accurately described by a simple Arrhenius-type equation, μ = μo × exp(−Ea/kT) (Supporting Information Table S1). In semicrystalline polymers, where transport occurs via a mobility edge mechanism,6, 49–51 the activation energy is related to the energy distribution and concentration of electronic traps. In all of the films studied here, the mobility versus 1/T curves are nearly parallel suggesting that the differences in mobility between the thickest and the thinnest films are not likely to be attributed to charge trapping and hence they must be due to the prefactor. In the mobility edge model, the prefactor is affected by the mobility of charges above the mobility edge. In semicrystalline materials however there are preferential transport paths, such as crystallites connected by low-angle grain-boundaries. As a result, the prefactor increases as the volume density of such preferential paths increases. The transport data suggest that a reduction in the density of preferential transport paths through ordered regions of the semiconductors is mainly responsible for the mobility decrease in the thinnest P3HT films, which is in agreement with our structural data. Indeed, in such very thin films, a bulk region ceases to exist and all the film can be considered as “interfacial”, where crystallinity is strongly suppressed [Fig. 2(A)].
In conclusion, we have investigated the effects of vertical confinement and interfaces on the microstructure and charge transport properties in thin films of P3HT, paying particular attention to dielectric surface treatment. XRD-based pole figures were constructed and used to quantify crystalline texture as well as the DoC of P3HT thin films as a function of thickness. The overall DoC was found to decrease with decreasing film thickness for films spun onto both OTS-treated SiO2 and untreated SiO2. Additionally, the crystalline material was found to show increasingly edge-on character as films became thinner. Pole figures also allowed for the separation of diffracted intensity according to crystalline texture. While the amount of bulk (misoriented) crystals increased with increasing film thickness, the density of crystallites nucleated at the substrate surface remained constant. Modeling the thickness dependence of the crystallinity allows us to determine that the interface is approximately 20–30% as crystalline as the bulk and also allows us to quantify the effect of substrate chemistry on microstructure. In the case presented here, the substrate/semiconductor interface hosts a population of perfectly oriented crystallites, whose formation is strongly enhanced (20×) by treating the SiO2 substrate with OTS. Transport measurements suggest that this population of interface-nucleated crystallites plays an important role in charge transport in field-effect transistors. Hence, these observations offer a possible explanation for the higher mobility measured in P3HT deposited on OTS-treated surfaces compared to bare SiO2. Although SAMs are shown to improve order and result in higher charge carrier mobilities, the interface is still significantly more disordered than the bulk of the film. In fact, the increased amount of interfacial disorder observed in very thin (∼4 nm) P3HT films causes a sharp decrease in field-effect mobility. Future efforts to improve performance in organic electronic devices should therefore focus not only on developing new materials with better intrinsic properties but also investigating ways to improve order in the areas most important for charge transport: interfaces.
Portions of this research were carried out at the Stanford Synchrotron Radiation Lightsource, a national user facility operated by Stanford University on behalf of the US Department of Energy. S. Himmelberger would like to thank the NSF for its support in the form of a Graduate Research Fellowship. A. Salleo gratefully acknowledges funding from NSF grant DMR-1205752. The authors thank Martin Heeney at Imperial College for supplying the P3HT used in this study.