X-ray based tools for the investigation of buried interfaces in organic electronic devices

Graphical abstract

Highlights

► We fabricated orthogonal soluble polymer stacks and probed the buried interface by X-ray reflectivity. ► Depending on the used solvent of the organic semiconducting material the interface morphology changed significantly. ► Grazing incidence X-ray diffraction exhibits the molecule alignment in the investigated polymer stack. ► The buried interface roughness within the polymer stack was correlated to the OTFT performance containing the stack.

Abstract

X-ray reflectivity combined with grazing incidence diffraction is a valuable tool for investigating organic multilayer structures that can be used in devices. We focus on a bilayer stack consisting of two materials (poly-(3-hexylthiophene)) (P3HT) and poly-(4-styrenesulfonic acid) (PSSA) spin cast from orthogonal solvents (water in the case of PSSA and chloroform or toluene for P3HT). X-ray reflectivity is used to determine the thickness of all layers as well as the roughness of the organic–organic hetero-interface and the P3HT surface. The surface roughness is found to be consistent with the results of atomic force microscopy measurements. For the roughness of P3HT/PSSA interface, we observe a strong dependence on the solvent used for P3HT deposition. The solvent also strongly impacts the texturing of the P3HT crystallites as revealed by grazing incidence diffraction. When applying the various PSSA/P3HT multilayers in organic thin-film transistors, we find an excellent correlation between the determined interface morphology, structure and the device performance.

1. Introduction

Knowledge about the interface morphology of an organic multilayer arrangement is crucial for electronic devices. This is because structural as well as morphological properties of interfaces can significantly impact the device performance. For instance, in organic solar cells, a rough interface between the semiconducting materials is preferable due to the improved probability of electron–hole separation at the donor/acceptor interface. In this context, Yan et al. used the resonant soft X-ray reflectivity technique to probe a polymer/polymer interface and subsequently correlated the morphology to the device performance of their solar cell [1]. Also in organic light emitting diodes, controlling the interface morphology is of relevance for improving the outcoupling efficiency and the internal quantum efficiency [2].

In organic thin film transistors (OTFTs), a smooth dielectric/semiconductor interface is beneficial, as there the charge transport occurs mostly in the first few monolayers of the semiconducting material closest to the dielectric [3,4]. Previous studies demonstrated a device performance enhancement by the insertion of an additional modification layer or passivation layer into the device architecture [5–7]. This modification layer creates a threshold voltage shift or enhances the charge carrier mobility, which was attributed to modifications of the interface morphology and the crystallographic order within the semiconducting materials [8–11].

One possible way to realize organic multilayer structures is the deposition of polymer layers on top of each other from orthogonal solvents. This approach has been used in the literature to improve the recombination efficiency in organic light-emitting devices [12–14]. Here a large variety of structures can be envisioned, e.g., by ink-jet printing of orthogonally soluble polymers on top of each other [15–17].

Such multilayer stacks naturally contain buried interfaces, whose non-destructive characterization is difficult. Here X-ray reflectivity (XRR) [18] can be a highly useful tool, as besides providing information on the average thickness of the various layers, it also allows a characterization of the surface and, most importantly, the buried organic/organic and organic/inorganic hetero-interfaces [19–24]. Beyond that, combining XRR with grazing incidence X-ray diffraction (GIXD) [25] allows the simultaneous study of interface-induced thin film structures and textures [26–28].

To demonstrate the potential of the combination of those techniques, we have investigated double-layer structures consisting of the water-soluble poly (4-styrenesulfonic acid) (PSSA) onto which poly-(3-hexylthiophene) (P3HT) is spin-cast either from chloroform (CHCl₃) or from toluene (C₇H₈). The choice of the latter is motivated by being a classical organic semiconductor material used in solar cells and OTFTs. Its measured charge-carrier mobility is reasonably large (typically between 0.01 and 0.1 cm²/V s), and very dependent on the morphology and the crystallographic properties of the P3HT film [29–31]. The latter is influenced by molecular weight of the polymer chains [7,32,33], the deposition technique of the film [5,33], the solvent used [15,16,33,34] and the annealing temperature of P3HT [35]. PSSA was primarily employed because it can be spin-cast from water.

To benchmark the XRR results, the surface morphology of the top P3HT layer was additionally investigated by atomic force microscopy (AFM) and to correlate interface morphology and P3HT texturing with charge transport properties, PSSA/P3HT bilayers were also included into bottom-gate top contact OTFT structures.

2. Materials and methods

2.1. Materials and thin film preparation

Poly(4-styrenesulfonic acid) (PSSA) was purchased from Sigma Aldrich (CAS: 28210-41-5) and dissolved in deionized water with a concentration of 3.6 g/l. Regioregular-poly-(3-hexylthiophene) (P3HT) was purchased from Rieke Metal (Sepiolid P200, CAS: 156074-98-5). According to the manufacturer it shows a head-to-tail regioregularity higher than 98% and a molecular weight of ∼30.000 g/mol. Both materials were used without further purification. For thin film preparation, the P3HT was dissolved either in the low boiling point solvent chloroform (short: P3HT(chloroform), T_b = 61 °C) with a concentration of 6 g/l or in the high boiling point solvent toluene (short P3HT(toluene), T_b = 111 °C) with a concentration of 10 g/l. The different concentrations were chosen to realize a P3HT layer thickness of ∼40 nm on top of the 8 nm thick PSSA layer. As substrates, doped Si-wafers (size 20 mm × 20 mm), with a 150 nm thermally oxidized SiO₂ layer were obtained from Siegert Consulting e.K. (Aachen, Germany). The substrate had a surface roughness of ∼0.5 nm (as measured by XRR). This smooth and flat surface made them ideal substrates for the investigations. The substrate was chemically cleaned by RF O₂-plasma etching for 30 s immediately before deposition of the first layer. PSSA was then spin cast at 2000 rpm for 15 s followed by 3500 rpm for 40 s onto the substrate under ambient conditions. The PSSA layer was annealed at 80 °C in high-vacuum for 2 h to reduce the residual water. The subsequent spin casting of P3HT was done in an Ar atmosphere using a home-built spin-coater at ∼1500 rpm for 40 s. Then the sample was annealed in Ar for 5 min at 80 °C (i.e., above the glass transition temperature of P3HT of T_g = 12.1 °C [36]) to remove solvent residuals from the thin film [37]. Besides the multilayer stack also single polymer thin films were characterized in terms of their crystallographic properties and layer morphology (layer thickness, layer roughness and electron densities). Fig. 1 shows the investigated polymer arrangement on the substrate with the chemical structure of the two polymer materials.

Fig. 1.

Multilayer structure of the investigated polymer stack on the Si/SiO₂ substrate together with the chemical structure of poly-(3-hexylthiophene) (P3HT) and poly (4-styrenesulfonic acid) (PSSA).

2.2. Structural investigations

Specular X-ray reflectivity measurements were performed on a Panalytical Empyrean Reflectometer equipped with a 1/32° slit, a 20 mm beam mask (axial width) and a multilayer mirror (equatorial divergence less than 0.055°) on the primary side, using Cu Kα radiation (λ = 0.154 nm). A small receiving slit of 0.1 mm, a 0.02 rad Soller slit and a PANalytical PIXCEL^3D detector (used as a point detector) were used on the secondary side. The goniometer radius of the Reflectometer was 240 mm. The 2-Theta step size was set to 0.004° to get sufficient resolution during the experiment. The experimental data were fit with the X´Pert Reflectivity 1.3 software (PANalytical) [38], which uses the Parratt formalism to simulate the data [39]. The reported errors are of statistical origin and are related to the numerical error of the fitting parameters [40]. The surface roughness and the interface roughness of the specimen were determined using the Croce and Névot approach [41].

Grazing incidence X-ray diffraction (GIXD) measurements were performed with a commercial four-circle Bruker D8 Discover diffractometer upgraded with the Bruker Ultra GID add-on and a sealed copper tube (λ = 0.154 nm). The incidence angle (α_i = 0.17) of the primary beam was optimized to maximize the scattering intensity from the sample and the beam height finally was set by a 0.6 mm slit [42]. The results of in-plane GIXD measurements are presented in the form of integrated intensities along the q_z direction (q_z = 0.2 − 3 nm⁻¹) with respect to the in-plane component q_p of the scattering vector q extracted from reciprocal space maps [43].

The atomic force microscopy (AFM) measurements were performed with a MFP 3D system of Asylum Research in tapping mode under ambient conditions. Cantilevers NSG30 from NT-MDT with a force constant of about 40 N/m, a resonance frequency of about 300 kHz, a tip radius of around 10 nm and an opening angle at the apex of about 10° were used. Five independent images (10 μm × 10 μm) have been measured for each sample to allow sufficient statistics. The images were processed afterwards with the data analysis software Gwyddion [44]. For the characterization of the surface roughness, the one dimensional height–height correlation function (HHCF) was calculated along the fast scan axis, x, of the images and then averaged over all scan lines. Because of the self-affinity of the surfaces, the HHCF was fit to [45],

$$C(x)={\sigma }^2{e}^{-{(\frac{x}{\xi })}^{2\alpha }}$$ (1)

to obtain the three main roughness parameters, that are (i) the root mean square (r.m.s.) roughness σ, (ii) the lateral correlation length ξ, and (iii) the Hurst parameter α. The latter parameter describes how jagged the layer surface is [45]. This analysis has been shown to yield surface roughness parameters which are in good agreement with the data obtained by the integral XRR technique [46].

2.3. Device fabrication

For device characterization, ∼50-nm-thick gold source and drain electrodes were deposited on top of the polymer stack by a shadow mask in a high-vacuum set-up operated inside an Ar glove-box. The resulting channel length and width were 25 μm and 7 mm, respectively. The devices were fabricated and characterized in the glove box, without ever exposing them to ambient air. Initial measurements one week and control measurements one month after device fabrication resulted in the same trends regarding the device performance. A Keithley KE2623A dual source-meter was used to measure the data, which were evaluated with a home-made software package. The mobility of the investigated polymer transistors were extracted in the saturation regime neglecting the impact of the contacts and the dependence of the mobility on the gate-voltage. Due to that and the often considerable hysteresis (see below), we refer to effective mobilities. The device characteristics were measured for two simultaneously prepared sets of samples, each of which contained 4 bottom gate-top contact transistors, yielding consistent results.

3. Results and discussion

3.1. XRR and AFM results

3.1.1. Individual organic layers on Si/SiO₂–substrate

In order to investigate a complex multilayer system by XRR analysis, it is beneficial to first characterize the individual layers on a substrate. Hence, each organic material was dissolved in the desired solvent and spin cast onto the Si/SiO₂–substrate. The preparation condition was kept the same for all following test samples as well as for the devices. Optical inspection suggested homogenous film coverage of all samples. Fig. 2a shows the reflectivity data of the PSSA(water) layer on the substrate and the corresponding AFM image (Fig. 2b) of the same sample. The variations of the reflectivity with the large period originate from the PSSA layer. The thickness of that layer is extracted from the XRR fit (red¹ line) to be 8.9 ± 0.1 nm (Table 1). The inset in Fig. 2a reveals rapid oscillations (Kiessig fringes [47]) coming from the SiO₂ layer, whose thickness is determined to be 147.2 ± 3.5 nm. The fit surface roughness of the PSSA layer was 0.3 ± 0.1 nm and the PSSA layer had an electron density of 414 ± 31 nm⁻³. From the fit, the roughness of the SiO₂/PSSA interface was determined to be 0.4 ± 0.1 nm. The SiO₂ layer was considered in all following XRR fits. The parameters obtained for the SiO₂ layers in all of those fits were within the error bars of the example in Table 1.

Fig. 2.

(a) X-ray reflectivity (filled squares) of the Si/SiO₂/PSSA-stacks as a function of the scattering vector q_z. The red line is the corresponding XRR fit; only every tenth data point is designated by a square. The insert in the graph illustrates the thickness oscillations of the SiO₂ substrate at higher q_z values. (b) 10 μm × 10 μm AFM image showing the morphology of the PSSA layer on top of the silicon oxide.

Table 1.

Layer thickness, d, r.m.s. roughness, σ, and total electron density, ρ, of the investigated layers extracted from the XRR data shown in Fig. 2 (top). Also the surface roughness, σ, lateral correlation length, ξ, and Hurst parameter, α, obtained from AFM are included. Note that the reported errors reflect only statistical effects and numerical deviations between data and fits. They do not account for systematic errors and differences between the used techniques (beyond the different underlying physical processes the fact that XRR is an integral method, while with AFM one samples only a small section of the surface).

Sample	XRR			AFM
	d (nm)	σ (nm]	ρ (nm−3)	σ (nm)	ξ (nm)	α
SiO2–substrate	147 ± 3.5	0.4 ± 0.1	670 ± 11
PSSA(water)	8.9 ± 0.1	0.3 ± 0.1	414 ± 31	0.12 ± 0.02	100 ± 35	0.6 ± 0.1

The AFM surface morphology investigations on the same sample confirm the assessment from the XRR analysis that the PSSA surface is very smooth. The extracted r.m.s. surface roughness of 0.12 ± 0.02 nm is in reasonable agreement with the XRR data. The lateral correlation length is 100 ± 35 nm and the Hurst parameter 0.6 ± 0.1.

Furthermore, investigations on P3HT spin-cast from two different solvents onto the Si/SiO₂–substrate, were performed. These data are useful for the later comparison of devices, where P3HT is directly spin-cast onto the SiO₂ dielectric with those containing an additional PSSA layer. The top XRR graph in Fig. 3a shows the result for the P3HT layer prepared from toluene and the bottom XRR graph for that spin-cast form chloroform. Here, the rapid oscillations again correspond to the silicon oxide layer of the substrate. Interestingly, the graphs exhibit completely different behavior concerning the oscillations originating from the P3HT layers. The top curve for P3HT(toluene) comprises only few oscillations (fringes) descending, while the bottom curve for the P3HT(chloroform) sample shows many well pronounced interference fringes. This is not a consequence of different layer thicknesses, as the XRR fits (red lines) reveal essentially the same layer thicknesses for both samples (cf., Table 2). Instead the extracted surface roughnesses are significantly different. The three times larger surface roughness for the film cast from toluene (cf., Table 2) causes the slow reflectivity variations to vanish at large q. Both XRR simulations yield an interface roughness between SiO₂ and P3HT with a roughness value of 0.4 ± 0.1 nm. AFM morphology investigations on the same samples show a similar relative increase of the surface roughness compared to the PSSA layer and also confirm the significantly larger surface roughness for the P3HT film cast from the high boiling-point solvent toluene (Table 2 and AFM image of Fig. 3b). The AFM image of the P3HT film cast from chloroform shows wide protrusions more than 10 nm high and a few 100 nm resulting in an overall r.m.s. roughness of 6.4 ± 1.6 nm (ξ = 160 ± 25 nm, α = 0.8 ± 0.1) (Fig. 3c). Neglecting these isolated spikes, the r.m.s. roughness becomes significantly smaller and amounts to only 1.3 nm on a laterally shorter correlation length of 70 ± 5 nm. The result is then well comparable to the data from the XRR investigations (Table 2). The spikes are not reflected in the XRR data, where it has to be stressed that XRR is an integral method. They, however, also could be a consequence of degradation effects of the P3HT layer during the XRR measurements [48].

Fig. 3.

(a) X-ray reflectivity for the Si/SiO₂/P3HT-stacks as a function of the scattering vector q_z. The upper line illustrates the stack with the P3HT layer prepared from toluene (open circles) and the lower curve is for the P3HT layer prepared from chloroform (filled squares). The red lines are the corresponding fits. The data for P3HT(toluene) are shifted by 10³ nm⁻⁴ for the sake of clarity; only every tenth data point is designated by a circle/square. (b) AFM image of the P3HT(toluene) morphology and (c) AFM image for P3HT(chloroform), both 10 μm × 10 μm.

Table 2.

Layer thickness, d, r.m.s. roughness, σ, and total electron density, ρ, of the investigated layers extracted from the XRR data shown in Fig. 3a. Also the surface roughness, σ, lateral correlation length, ξ, and Hurst parameter, α, obtained from AFM are included (Fig. 3b and c). Note that the reported errors reflect only statistical effects and numerical deviations between data and fits. They do not account for systematic errors and differences between the used techniques (beyond the different underlying physical processes the fact that XRR is an integral method, while with AFM one samples only a small section of the surface).

Sample	XRR			AFM
	d (nm)	σsurface (nm)	ρ (nm−3)	σ (nm)	ξ (nm)	α
P3HT(toluene)	38.5 ± 0.5	5.3 ± 0.5	376 ± 21	10.4 ± 2.0	225 ± 40	0.8 ± 0.1
P3HT(chloroform)	39.0 ± 0.1	1.6 ± 0.4	357 ± 21	1.3 ± 0.1	70 ± 5	0.9 ± 0.1

3.1.2. Multilayer stacks

With the properties of the individual layers known, next the morphological characteristics of the multilayer stacks will be discussed. These consist of the water soluble PSSA layer on top of the Si/SiO₂ substrate onto which P3HT is spin-cast either from toluene or from chloroform. The corresponding XRR data together with the fits are shown in Fig. 4a and the parameters extracted from the fits are summarized in Table 3. The fast oscillations in both characteristics again originate from the SiO₂. The other interference fringes in the XRR fits arise from the 39.9 ± 0.5 nm thick P3HT(toluene) layer on top of the 8.1 ± 0.5 nm thick PSSA layer for the first specimen and a 39.1 ± 0.9 nm thick P3HT(chloroform) layer on top of an 8.5 ± 0.5 nm thick PSSA layer for the second specimen (see Fig. 4a). The individual layer thicknesses agree well with the thicknesses of the single layer investigations.

Fig. 4.

(a) X-ray reflectivity of the Si/SiO₂/PSSA/P3HT-stacks as a function of the scattering vector q_z. The upper line illustrates the stack with P3HT prepared from toluene (open circles) and the lower curve is for P3HT dissolved in chloroform (filled squares). The red lines are the corresponding fits. The data for P3HT(toluene) were shifted by 10³ nm⁻⁴ for the sake of clarity; only every tenth data point is designated by a circle/square. (b) Electron density profile of the Si/SiO₂/PSSA/P3HT multilayer stack perpendicular to the substrate surface. The open circles illustrate the stack with the P3HT layer prepared from toluene and the filled squares shows the P3HT layer prepared from chloroform.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 3.

Layer thickness, d, r.m.s. roughness, σ, and total electron density, ρ, of the investigated multilayer stacks extracted from the XRR data shown in Fig. 4a.

Sample	dPSSA (nm)	σInterface (nm)	ρPSSA (nm−3)	dP3HT (nm)	σsurface (nm)	ρP3HT (nm−3)
PSSA(water)/P3HT(toluene)	8.1 ± 0.5	0.2 ± 0.1	410 ± 10	39.9 ± 0.5	4.7 ± 0.5	375 ± 17
PSSA(water)/P3HT(chloroform)	8.5 ± 0.5	1.2 ± 0.3	414 ± 12	39.1 ± 0.9	0.9 ± 0.2	367 ± 19

Again, the fringes due to the P3HT layer decay rapidly with q for the samples cast from toluene. This is fully consistent with the larger surface roughness of the P3HT(toluene) film extracted from the fits (σ_rms = 4.7 ± 0.5 nm for P3HT(toluene) vs. σ_rms = 0.9 ± 0.2 nm for P3HT(chloroform)). In that sample, however, the fringes due to interference at the PSSA layer are much better preserved at large q than for the film cast from chloroform. This observation has to be related to an interface roughness of the interface between the PSSA and the P3HT layer. Indeed, the fits reveals a significantly larger roughness of the buried PSSA/P3HT interface for the P3HT(chloroform) sample (σ_rms = 1.2 ± 0.3 nm) than for the stack containing the P3HT(toluene) layer (σ_rms = 0.2 ± 0.1 nm). I.e., when using the high boiling point solvent toluene, a larger surface but smaller interface roughness is obtained. This can be explained by the significantly longer time it takes for toluene layer to dry allowing for significant rearrangements of the P3HT chains at the surface. The more polar chloroform, on the other hand, can be expected to more strongly modify the PSSA layer when spin-casting P3HT. Possible processes could be that chloroform swells or partially re-dissolves the relatively polar PSSA surface. A clear identification of these processes that occur during the spin-casting is, however, beyond the possibilities of the XRR experiments, which we need to perform ex-situ, i.e., on dried films. The details of the process notwithstanding, the consequence of the stronger interaction between chloroform and the PSSA layer is a larger roughness of the buried PSSA/P3HT interface.

Fig. 4b shows the variation of the total electron density perpendicular to the sample surface of the two multilayer arrangements. The electron density profile was calculated with the parameters from the XRR results by using the effective density model [49]. The interface roughness of each individual layer results in smearing along the electron density profile. One can clearly observe the difference between the two samples. The higher roughness at the interface between PSSA and P3HT(chloroform) leads to a gradual change in the electron density, which is clearly smeared out. The smaller surface roughness of the P3HT(chloroform) layer leads to a step-like change of the electron density at the surface. In contrast, the smooth interface between PSSA and P3HT(toluene) results in a rapid change in the electron density profile at that interface and the rough surface of the P3HT(toluene) layer results in a corresponding electron density profile, which is somewhat smeared out.

To obtain an independent second set of results for the surface roughness, in Fig. 5, AFM images of the samples are compiled with their HHCFs. The HHCF presented in Fig. 5c and d have been fit to Eq. (1) to obtain the surface roughness σ, the lateral correlation length ξ and the Hurst parameter α. The slight deviation for x-values larger than several ξ can be attributed to the lack of statistics for large distances. For better statistics, five independent 10 × 10 μm² AFM images of each sample were analyzed. Table 4 contains the resulting average values and the standard deviations of the extracted roughness parameters. The obtained r.m.s. roughness, the lateral correlation length ξ, and the Hurst parameter of the P3HT films in the multilayer stack (Table 4) agree well to the results for the P3HT layers directly grown on SiO₂ (Table 2).

Fig. 5.

(a and b) 10 × 10 μm² AFM micrographs of Si/SiO2/PSSA(water)/P3HT(toluene) and Si/SiO2/PSSA(water)/P3HT(chloroform) samples. (c)The height–height correlation function, C(x), corresponding to the data from (a), and (d) that for the data from (b). The black dots show the experimental data and the full line is the fit using Eq. (1).

Table 4.

Mean values of the surface parameters (rms roughness, σ, lateral correlation length, ξ, Hurst parameter, α, statistically calculated from a certain number of AFM images for each sample. (Fig. 5c and d).

Sample	σsurface (nm)	ξ (nm)	α
PSSA(water)/P3HT(toluene)	8.4 ± 0.4	200 ± 5	0.8 ± 0.1
PSSA(water)/P3HT(chloroform)	0.6 ± 0.1	60 ± 15	0.7 ± 0.1

3.2. Grazing incidence X-ray diffraction results

Grazing incidence X-ray diffraction experiments were performed to get insight into the preferred orientation of the molecules within the SiO₂/PSSA/P3HT multilayer stacks. Fig. 6a shows the intensities as a function of the in-plane component of the scattering vector, q_p, extracted from the measured reciprocal space maps integrated over the out of plane component, q_z. Only diffraction features from P3HT were observed. Also single layer investigations reveal no crystallographic order of PSSA. For P3HT spin-cast on SiO₂/PSSA from chloroform, a diffraction feature at q_p = 3.7 nm⁻¹ was observed (full circles in Fig. 6a). This feature corresponds to the d₁₀₀ spacing of P3HT crystallites [31]. Since the second frequently observed diffraction feature at q_p = 16.8 nm⁻¹ is missing (d₀₂₀ spacing of P3HT crystallites), we conclude that the crystalline parts of the P3HT film consist mainly of crystallites with [0 1 0] orientation parallel to the sample surface. Fig. 6b illustrates this alignment of the P3HT molecules on the sample, which is frequently denoted as face-on alignment. Interestingly, the arrangement of the P3HT crystallites in the thin film does not change for the P3HT layer deposited from chloroform onto the PSSA film (full rectangles in Fig. 6a).

Fig. 6.

(a) Scattered intensities as a function of the in-plane direction q_p of the scattering vector q of the investigated single layers of Si/SiO₂/P3HT and of the Si/SiO₂/PSSA/P3HT multilayer samples. Integration of the 2D reciprocal space maps has been performed over the q_z-direction (q_z = 0.2–3 nm⁻¹); (b) sketch of the face-on alignment of the P3HT molecules prepared from chloroform solution and (c) sketch of the edge-on alignment of the P3HT molecules prepared from toluene solution.

In the GIXD measurements of the P3HT(toluene) layer deposited either onto the SiO₂ surface (green triangle) or onto the PSSA layer (black diamond) no diffraction feature from the (1 0 0) planes of the P3HT crystallites was observed. Instead, the second expected diffraction feature of P3HT at q_p = 16.8 nm⁻¹ appears within the integrated intensities (Fig. 6a). This suggests a dominating [1 0 0] orientation of the P3HT crystallites with respect to the sample surface, also known as edge-on alignment (Fig. 6c). This is consistent with the results of Chang et al., who demonstrated a preferred edge-on alignment of P3HT molecules dissolved in high boiling point solvents [50]. Please note, that no conclusion about the orientation of the molecules in the amorphous state can be given.

3.3. Device performance

To obtain a first impression, how the above-described structural and morphological parameters correlate with device performance, we fabricated a series of OTFTs containing the differently spin-cast P3HT layers. Representative transfer characteristics for these devices are shown in Fig. 7, where the top panel refers to devices in which the P3HT layer was deposited from chloroform (Fig. 7a) and the bottom panel to devices, where toluene was used as a solvent (Fig. 7b). The open, black squares refer to single layer devices and the filled, red squares to OTFTs containing PSSA/P3HT double layers. The main (effective) device parameters extracted from the transfer characteristics are summarized in Table 5.

Fig. 7.

Representative transfer characteristics of the thin film transistor devices containing the Si/SiO₂/P3HT films and Si/SiO₂/P3HT/PSSA films spin-cast either from chloroform (a) or from toluene (b). The drain voltage was set to V_D = −40 V.

Table 5.

Main device parameters extracted from the saturation region of transfer characteristics of P3HT and PSSA/P3HT OTFTs spin-cast either from chloroform or toluene solution and measured at V_D = −40 V. The values are averaged over all investigated devices. All reported values have been obtained for the sweep from positive to negative voltages, which is the first sweep in the measurement procedure. The larger relative errors (standard deviations) for devices cast from chloroform are a manifestation of the larger scattering in the obtained data.

Device	μsat/10−3 cm2/(V s)	VT,sat (V)
Si/SiO2/P3HT(toluene)	5.9 ± 0.9	15 ± 2
Si/SiO2/PSSA(water)/P3HT(toluene)	24.0 ± 2.0	1 ± 2
Si/SiO2/P3HT(chloroform)	3.0 ± 2.0	2 ± 4
Si/SiO2/PSSA(water)/P3HT(chloroform)	0.6 ± 0.3	5 ± 5

The investigated devices are all characterized by a non-negligible hysteresis. For the devices not containing a PSSA layer, this is attributed to using the P3HT as received (i.e., without further purification) and also to trapping at the semiconductor/SiO₂ interface. The further increase of the hysteresis in devices containing a PSSA layer is most likely a consequence of residual water molecules present in the PSSA layer also after the annealing process and the high acidity of the layer, which creates a situation to some extent reminiscent of that encountered in certain transistors with polymer electrolyte gate dielectrics [51,52]. Independent of the large hysteresis, the general mobility trends reported in Table 5 are also clearly reflected in the actual evolution of I_DS as a function of V_G − V_T, especially for small gate voltages [53].

For the devices with P3HT directly grown on SiO₂ (i.e., devices not containing a PSSA layer) we find somewhat higher effective mobilities for P3HT films grown from toluene. This is consistent with the edge-on alignment of the polymer chains in the crystalline parts of the film discussed in the previous section and the resulting π–π stacking in the direction of charge transport [26,30,54].

The differences in device performance increase dramatically when including a PSSA layer on top of the SiO₂ (PSSA/P3HT devices): While the effective mobility decreases for P3HT spin-cast from chloroform, it significantly increases (by ∼ a factor of 4) when depositing P3HT from toluene. As the GIXD measurements demonstrated that the alignment of P3HT is independent of the subjacent layer, we attribute this effect to the differences in the roughness of the PSSA/P3HT interface (1.2 nm when casting P3HT from chloroform and 0.2 nm when casting it from toluene). The structure of that interface is of particular importance, as it is located exactly where the conducting channel is formed [55]. The smoothening of the semiconductor/dielectric interface by the PSSA layer also explains the increase of the effective mobility in the PSSA/P3HT devices with P3HT cast from toluene compared to those not containing a PSSA layer. In this case one, however, cannot exclude that beyond the smoothening effect also the modified chemical structure (and thus trap distribution) of the PSSA/P3HT interface as well as the different dielectric constant of PSSA compared to SiO₂ does impact the mobility [56]. Our observation that the dielectric/semiconductor interface strongly influences the charge-carrier mobility is, in fact, consistent with several reports in the literature, where the interface roughness has been shown to play an important role [55,57–59]. In contrast to the present paper, where we directly characterize the buried interface, in those studies the interface roughness has, however, only been inferred from the surface roughness determined before the deposition of the active layer: Steudel et al. showed that for pentacene transistors with SiO₂ dielectrics, the decrease of the dielectric/semiconducting interface roughness resulted in a mobility improvement in the devices [57]. A detailed interface roughness vs. mobility investigation was also performed by Jo et al. with PS-b-PMMA block copolymer as interface layer on SiO₂ substrates [58]. It clearly showed a dependence of the charge carrier mobility on the interface roughness between pentacene and PS-b-PMMA layer. In that study only small changes of the interface roughness improved the charge carrier mobility significantly. In addition, Chua et al. reported a critical interface roughness of 0.7 nm between bilayers of orthogonally dissolved polymers (determined from AFM power spectra) where the charge carrier mobility drops significantly at the polymer/polymer interface [59].

Another aspect that can be relevant for the improvement in the device cast from toluene is the higher P3HT surface roughness observed there. This results in a rougher P3HT(toluene)/Au interface at the source and drain contacts, which can be expected to facilitate carrier injection due to local field enhancements and a larger injecting area. The reduced contact resistance in the PSSA/P3HT(toluene) devices then results in a larger source–drain current yielding a larger extracted effective mobility [60,61].

4. Conclusion

We show that a combination of X-ray reflectivity investigations with grazing incidence X-ray diffraction provides a valuable and non-invasive tool for simultaneously determining the surface and the interfacial morphology as well as the preferred texturing in polymeric thin-film stacks spin-cast from orthogonal solvents. A particular strength of the approach is that XRR is capable of analyzing buried interfaces. This is shown here explicitly for a stack consisting of Si/SiO₂ as a substrate, a water soluble PSSA layer as a modification layer, and the semiconducting P3HT deposited from different solvents. The surface of the P3HT layer was studied also by AFM, where the obtained trends compare well with those obtained from XRR.

The choice of the solvent for spin-casting P3HT determines not only the alignment of the P3HT crystallites (edge-on when using toluene vs. face-on for chloroform, as determined from the GIXD measurements), but also the roughness of the P3HT surface and the P3HT/PSSA interface. Interestingly, the observed trends for surface and interface roughness are opposite: Dissolving P3HT in chloroform decreases the P3HT surface roughness and increases the PSSA/P3HT interface roughness compared to using toluene as a solvent. This can be explained, on the one hand, by the different boiling points of the solvents and, on the other hand, by their different polarities. The different roughness of the various interfaces also has a distinct impact on the hole mobilities observed in OTFTs containing the above described multilayer stacks.