Breath figure–derived porous semiconducting films for organic electronics

See allHide authors and affiliations

Science Advances  25 Mar 2020:
Vol. 6, no. 13, eaaz1042
DOI: 10.1126/sciadv.aaz1042


Porous semiconductor film morphologies facilitate fluid diffusion and mass transport into the charge-carrying layers of diverse electronic devices. Here, we report the nature-inspired fabrication of several porous organic semiconductor-insulator blend films [semiconductor: P3HT (p-type polymer), C8BTBT (p-type small-molecule), and N2200 (n-type polymer); insulator: PS] by a breath figure patterning method and their broad and general applicability in organic thin-film transistors (OTFTs), gas sensors, organic electrochemical transistors (OECTs), and chemically doped conducting films. Detailed morphological analysis of these films demonstrates formation of textured layers with uniform nanopores reaching the bottom substrate with an unchanged solid-state packing structure. Device data gathered with both porous and dense control semiconductor films demonstrate that the former films are efficient TFT semiconductors but with added advantage of enhanced sensitivity to gases (e.g., 48.2%/ppm for NO2 using P3HT/PS), faster switching speeds (4.7 s for P3HT/PS OECTs), and more efficient molecular doping (conductivity, 0.13 S/m for N2200/PS).


Organic semiconductor (OSC)–based electronic devices such as thin-film transistors (TFTs), sensors, circuits, resistors, diodes, and others have emerged as competitive to conventional technologies because of their light weight, lower potential cost, biocompatibility, and mechanical flexibility (1, 2). Particularly in the chemical sensor area, organic TFTs (OTFTs) have been used to fabricate both the circuitry driving sensor platforms (3) and direct sensing elements by exploiting TFT parameter variations such as the threshold voltage and/or the drain current (4). For the latter devices, the OSC layer chemical structure and the film morphology and microstructure are of paramount importance in achieving high sensor responsivity to analytes (5). Because charge transport in OTFTs is confined to within a few molecular layers at the dielectric/semiconductor interface, it is known that the analyte must penetrate the semiconductor layer of bottom-gate devices to effectively modulate the charge transport (6).

Thus, semiconductor films having grain boundaries/porosities nearest the channel exhibit larger sensitivities to analytes than those having smooth/amorphous film morphologies (7). Nevertheless, achieving these porous morphologies is particularly challenging when using semiconducting polymers, which typically form far more uniform channel layers than small molecules. Although sensor technologies are by far the most investigated applications for porous OSCs, other device/process technologies should benefit from more porous semiconductor layer morphologies, such as electrochemical transistors where the redox process is accompanied by ion diffusion into the bulk semiconductor (8), as well as in molecular doping of OSC films (9).

Considerable efforts to create porous OSC films and to analyze their transport/sensing properties vis-à-vis more dense films have been reported (10, 11). For example, Lu et al. (12) reported porous OTFT sensors fabricated by thermally evaporating molecular semiconductors onto polystyrene (PS) microspheres, achieving sensitivities to NH3 of >300%/ppm (parts per million), which is 100 to 1000× that of devices made from continuous films. In another approach, Wang et al. (10) prepared porous films using an OSC + insulating polymer phase separation route, where acetone extraction of the latter yields microporous semiconductor films, which are highly sensitive to NH3. In addition, Zhang et al. (13) used a cross-linked porous insulating polymer as a mask to create a porous morphology in a semiconducting polymer, demonstrating OTFTs with a ~10× greater sensitivity to NH3 than those based on analogous dense films. Although the above OTFTs all achieved excellent sensor performance, the fabrication of the porous OSC films requires tedious and expensive multistep procedures, which would be nontrivial to scale up. Furthermore, to our knowledge, there has been no demonstration of porous OSC films used for electrochemical transistors or for producing chemically doped polymer films.

Several techniques can be used to fabricate porous organic thin films such as lithography (14), colloidal templates (15), block copolymer phase separation (16), emulsions (17), and breath figures (18). Among them, the latter method has been successfully used for creating porous films of electrically insulating and amorphous polymers, such as PS and poly(methyl methacrylate), having diverse functions and applications (18). Breath figures are formed by water droplets that can be generated by directing aqueous vapors onto a cold surface. Through the evaporative cooling/water condensation on a solution deposited on a surface in a humid environment, the resulting solid films attain a porous structure (19). This process is environmentally friendly since it relies on nontoxic, easily available, and intrinsically dynamic characteristics and avoids complex fabrication processes since formation and removal of the aqueous droplets are spontaneous. The simplicity and generality of this approach for fabricating porous thin films stand in contrast to conventional subtractive approaches based on capital-intensive lithography and etching, which involve complex multistep procedures and corrosive chemicals. Amorphous polymers of sufficient molecular mass are ideal for producing breath figure films because most form robust, flexible thin films, and many polymers are water immiscible and can efficiently stabilize water droplets (20).

Nevertheless, semiconducting organic polymers and small organic molecules are reported to be unsuitable for the breath figure processing because strong aggregation/crystallization during the drying process prevents pore formation (18). There is only a single report dealing with organic semiconducting materials, and it does not report microstructural, charge transport, sensor, or ion doping data (21). However, it has been reported that PS can be blended with several semiconducting materials to enhance OTFT electrical performance and stability (22). Therefore, an intriguing question is whether PS-organic semiconducting blends are suitable for creating porous microstructures integrable with charge-carrying semiconducting films and whether these products are suitable for fabricating OTFTs, sensor platforms, and other electronic devices. Here, we demonstrate that OSC/PS blend porous films fabricated by breath figure method are suitable and universally applicable for these functions and report figures of merit significantly surpassing those based on the conventional dense films. We illustrate the generality of this approach using three types of OSCs differing in molecular structure (polymer and small molecule) and charge transport (p-type and n-type) characteristics.


Materials selection and porous film fabrication

The porous film growth process as exemplified by OTFT fabrication is shown in Fig. 1 along with the dynamics of the film formation (18). A heavily doped Si wafer with a 300-nm thermally grown SiO2 layer was used as the substrate. The substrate was cleaned ultrasonically with acetone and isopropyl alcohol and then treated with an O2 plasma for 5 min to eliminate organic contaminants and improve film adhesion. To demonstrate the breath figure fabrication method universality, three OSC materials were blended with PS (Mw = 35,000) and used to fabricate porous OTFTs: (i) p-type poly(3-hexylthiophene) (P3HT); (ii) p-type small-molecule 7-dioctyl[1]benzothieno[3,2-b]benzothiophene (C8BTBT); and (iii) n-type polymer poly[N,N′-bis(2-octyldodecyl)-naphthalene-1,4,5,8-bis (dicarboximide)-2,6-diyl]-alt-5,5′-(2,2′-bithio-phene) (Flexterra, ActivInk N2200). For solution film casting, 6 mg of PS, P3HT, C8BTBT, and N2200 were separately dissolved in 1 ml of chloroform, and the solutions were stirred at 60°C for 3 hours. Then, equal amounts of OSC/CHCl3 solution and PS/CHCl3 solution were mixed and stirred at 60°C for 1 hour. Next, the solution was spin-coated onto SiO2/Si substrates at 2000 rpm for 30 s in a humidity-controlled box maintained at an optimized relative humidity (RH) of 60% (fig. S1, A to C). Note that control films/devices based on the dense OSC/PS films were spin-coated in ambient conditions (RH, ~30%). All films were then thermally annealed at 70°C in vacuum for 3 hours and were ~30 nm thick as measured by profilometry. Last, source and drain gold electrodes were deposited by thermal evaporation through a shadow mask, affording devices with a channel length (L) of 100 μm and channel width (W) of 1000 μm.

Fig. 1 Chemical structures and fabrication process for devices with porous polymeric semiconducting films.

(A) Chemical structures of materials used in this study. (B) Schematic illustration of the porous OTFT structure and the breath figure fabrication process.

Dense versus porous P3HT/PS films and TFTs

The P3HT/PS film morphologies and the corresponding OTFTs were first characterized in detail, followed by those based on the other two PS semiconductor blends (vide infra). The surface morphologies of the P3HT/PS films were assessed by atomic force microscopy (AFM). Figure 2A shows the AFM image of the control dense P3HT/PS film clearly revealing phase separation, as indicated by the presence of characteristic spherical P3HT domains, a morphology very different from the fibrillar ones obtained with pure P3HT films (23). However, the surface of these dense P3HT/PS films is smooth with a root mean square roughness (σRMS) of only ~2 nm. Figure 2B shows the surface of a porous P3HT/PS film, which exhibits nanoscale pores ranging from 300 to 500 nm, thus considerably increasing the σRMS to 9.6 nm. The AFM cross-sectional analyses of the height images of the porous P3HT/PS films reveal that the nanopores have a depth of ~30 nm, indicating that the nanopores extend through the entire thickness of the P3HT/PS films. This morphology should thus provide expedient diffusion of analyte molecules to the field-effect transistor (FET) channel area. Furthermore, note that the porous structure extends through the entire semiconductor film, as verified by the large-scale AFM image (50 μm by 50 μm) shown in fig. S1D, corroborating the excellent uniformity of the porous films. P3HT microstructural order in the semiconductor films was next studied by two-dimensional (2D) grazing incidence wide-angle x-ray scattering (GIWAXS; fig. S1, E and F). Both the dense and porous P3HT/PS films exhibit mixed orientations of the polymer chains with predominant edge-on orientation (24). In the out-of-plane direction (see Fig. 2C), the high-order edge-on (100) reflections for the dense and porous P3HT/PS films are located at 0.380 and 0.384 Å−1, respectively. These peaks are assigned to the main chain separation between different lamella layers, and the calculated lamellar spacings of edge-on orientation are 1.65 and 1.64 nm, respectively. The peaks around 1.67 Å−1 originate from the face-on (010) lattices, corresponding to π-π stacking distances of 3.75 Å. Along the in-plane direction (Fig. 2D), the peaks at ~0.376 and 1.65 Å−1 originate from the face-on (100) and the edge-on (010) lattices, respectively. The calculated lamellar spacing (face-on) and π-π stacking distances (edge-on) are 1.67 nm and 3.81 Å, respectively (25).

Fig. 2 Morphologies, structures, and electrical performance of transistors fabricated from porous polymeric semiconducting P3HT/PS films.

AFM images and the cross-sectional analyses of (A) dense and (B) porous P3HT/PS films. (C) Out-of-plane and (D) in-plane GIWAXS line cuts of the indicated P3HT/PS thin films. Representative transfer plots of (E) dense and (F) porous P3HT/PS OTFTs.

Next, the OTFT characteristics of both dense and porous P3HT/PS devices were investigated and compared. Both devices exhibit typical p-type behavior as shown by representative transfer (Fig. 2, E and F) and output (fig. S2, A and B) plots. The electrical parameters of these devices were extracted using standard metal oxide semiconductor FET equations (26, 27), and the carrier mobility (μ) and the threshold voltage (VT) data are summarized in Table 1. Both devices exhibit a similar Ion/Ioff ratio of ~103. For the dense device, μ and VT are 1.0 × 10−3 cm2 V−1 s−1 and +4.3 V, respectively. From a transport viewpoint, the porous OTFTs should be equivalent to a dense OTFT with a reduced effective channel width, assuming that all other parameters are the same. The on current (Ion) of the porous OTFTs is slight lower (1.15 × 10−7 A) than that of the dense OTFTs (1.82 × 10−7 A) because of the reduced OSC coverage of the OTFT conduction channel. From the AFM image, the ratio of the pore area to total film area is ~45%, from which the device effective channel width Weff is estimated to be ~550 μm. Thus, the mobility of the porous OTFT is calculated to be 1.45 × 10−3 cm2 V−1 s−1 with a VT of +3.2 V.

Table 1 Electrical parameters of dense and porous P3HT/PS OTFTs exposed to the indicated NO2 concentrations.

View this table:

P3HT/PS TFT sensors

The sensor characteristics of the P3HT/PS OTFTs were evaluated using gaseous NO2 as the analyte in a concentration range of 0 to 20 ppm. In a typical experiment, dry air and 100-ppm NO2 gas in N2 (the source NO2 concentration) were mixed in the appropriate ratios, and the mixture was delivered to the test chamber by a mass flow controller. Electrical measurements were carried out after 100 s of NO2 exposure in the test chamber to ensure saturation at the desired analyte concentration. Figure 3A shows representative transfer curves of the dense and porous P3HT/PS OTFT sensors upon exposure to various NO2 concentrations, and panels B and C of Fig. 3 show the corresponding VT and μ variations, respectively. Upon NO2 exposure, the current-voltage relation (I-V) curves of both devices shift to the upper-left corner of the transfer plot along with enlarged Ioff and Ion. However, the porous OTFTs exhibit far greater response than the dense OTFTs at all NO2 concentrations. Clearly, when the NO2 vapor concentration is increased, the charge density in the conduction channel increases since VT gradually shifts in the positive direction. Simultaneously, the μ’s of both devices increase (Table 1). For example, when the NO2 concentration increases from 5 to 20 ppm, the VT and μ of the porous OTFTs increase by +15.3 and +19.3%, respectively, while those of the dense devices increase by only +6.0 and +3.3%, respectively. The increase of the carrier density in P3HT, specifically holes, upon NO2 exposure is in agreement with the oxidizing properties of this gas (28). The enhanced sensitivity of the porous structure is reasonably the result of a more accessible morphology for the NO2 molecules to approach the most responsive areas of the channel (5).

Fig. 3 Gas sensor properties of the indicated porous P3HT/PS transistors.

(A) OTFT transfer plots. Percentage change of threshold voltage VT (B) and mobility (C) as a function of the NO2 vapor concentration exposures of P3HT/PS OTFT sensors. (D) Responsivity (R)–VG plots of the indicated devices to NO2 vapor. (E) Real-time responsivity (VD = VG = −60 V) to dynamic NO2 concentrations. (F) Selectivity of the P3HT/PS OTFT sensors to H2 and CO2 at 103-ppm concentration and CO, SO2, NH3, and NO2 vapors at 20-ppm concentration.

TFT gas-sensing responsivity (R) is defined as R = [(IGasI0)/I0] × 100%, where IGas is the drain current of the transistor when exposed to the analyte and I0 is the drain current of the unexposed device (4). Figure 3D shows the responsivity of the present P3HT/PS OTFTs derived from the transfer curves. Since the transfer curve is VG dependent, the devices exhibit the maximum responsivity when VG is ~0.0 V (VD = −60 V). For [NO2] = 20 ppm and VG = 0.0 V, the responsivity of the dense P3HT/PS OTFT is ~7900%, while that of the porous device reaches ~22,000%, which is almost three times higher than for the dense OTFTs. Furthermore, this value is greater than that of recently reported NO2 OTFT sensors fabricated by directly blending a hole-transporting/electron-blocking polymer into the active layer (~20,000% for 30 ppm) (29). At the present VG = 0.0 V, even for a low [NO2] of 1 ppm, a high responsivity of ~250% is achieved for the porous device, in contrast to ~15% for the dense one. Usually, the responsivity in the OTFT saturation region is lower than that in the cutoff region. Note that this scenario is important since the saturation region has a much higher signal-to-noise ratio, and thus, the reliability of the data is greatly enhanced. When the device operates in saturation (VG = −60 V) and [NO2] = 20 ppm, the responsivity of the porous OTFT is 125%, again considerably larger (six times) than that of the dense device (~20%). Figure 3E shows the real-time response of the OTFTs functioning at VD = VG = −60 V when subjected to various absorption/desorption cycles of NO2 at different concentrations (1 to 10 ppm). The data show that the responsivity increases rapidly and markedly when the sensor is exposed to NO2 and then recovers, but incompletely, when the NO2 is replaced with dry air. For example, the responsivity of the porous OTFT is 280% after exposure to 10-ppm NO2, while the value of the dense OTFT is only 80%, indicating higher responsivity of the former OTFT structure. Figure S2C plots the responsivity of the two OTFTs as a function of NO2 concentration on a log scale, indicating excellent linear correlations, which is crucial for practical device applications. Figure S2D plots the sensitivity (S) of the two devices at different NO2 concentrations. Here, S is defined as the percentage of current change per ppm analyte concentration, thus given by the equation S = [(IGasI0)/I0]/[NO2] × 100%, where [NO2] is the NO2 concentration in ppm. The porous OTFT shows a much higher S (28.1 to 48.2%/ppm) than the dense one (6.4 to 8.3%/ppm) at all NO2 concentrations. Note also that the superiority of the porous OTFT becomes more obvious at lower NO2 concentrations and that this trend is similar to porous OTFT sensors produced using physical fabrication methods (12).

Selectivity is also a crucial parameter for a sensor and often a deciding issue for practical applications, which usually rely on specific OSC-analyte interactions. To probe the P3HT/PS OTFT selectivity to NO2, we tested other gases (20-ppm NH3, CO, and SO2 and 103-ppm CO2 and H2) with dry air as the carrier gas. The results shown in Fig. 3F reveal that the present platform is far more sensitive to NO2, demonstrating excellent selectivity. Again, the porous OTFTs achieve far greater responses than the dense ones for all of the aforementioned gases. Last, device stability is another important factor for useful sensor devices. Figure S3A shows the transfer characteristics of porous P3HT/PS OTFTs tested immediately after fabrication and after storage in air for 20 and 40 days, which demonstrate good ambient stability. This result is in agreement with the known stabilizing characteristics of PS for P3HT OTFTs (30).

OTFTs and sensors based on molecular and n-type semiconductors

To illustrate the generality of the present breath figure fabrication methodology for p-type small-molecule and n-type polymer semiconductors, we also fabricated and investigated C8BTBT/PS and N2200/PS films/devices. The surface topography of the dense and porous C8BTBT/PS and N2200/PS films examined by AFM are similar to the above P3HT/PS results (Fig. 4, A to D). The dense films are inhomogeneous with small crystallites or N2200 aggregate formation, while the porous films exhibit characteristic nanoscale pores ranging in dimensions from 200 to 500 nm. GIWAXS measurements on the dense C8BTBT/PS films (Fig. 4E and fig. S4A) reveal a ~10th-order out-of-plane (11L) Bragg peak (1.32 Å−1), conclusively indicating a highly crystalline nature. The diffraction spots of the (00L) planes along the qz direction, corresponding to a lattice spacing of 2.86 nm, indicate a highly ordered edge-on orientation of the molecular cores. GIWAXS images of the porous C8BTBT/PS films (Fig. 4F and fig. S4B) exhibit similar diffraction features, albeit less intense and generally broader, meaning a less-ordered structure (31). Both the dense and porous N2200/PS films exhibit typical (010) reflections, corresponding to π-face-on polymer orientation (fig. S4, C and D), although the latter is slightly weaker (32).

Fig. 4 Morphologies, structures, and gas sensor performance of the indicated porous transistors.

AFM images of (A) dense and (B) porous C8BTBT/PS films. The σRMS values of the dense and porous C8BTBT/PS films are 7.6 and 15.2 nm, respectively. AFM images of (C) dense and (D) porous N2200/PS thin films. The σRMS values of the dense and porous C8BTBT/PS films are 3.9 and 10.2 nm, respectively. 2D grazing incidence x-ray diffraction spectrum of (E) dense and (F) porous C8BTBT/PS films. (G) Sensitivity of the OTFTs based on the indicated films to different NH3 concentrations.

To characterize the charge transport and sensing performance of the C8BTBT/PS thin films, we fabricated and evaluated OTFTs (fig. S4, E to H); they exhibit typical p-type characteristics. As in the case of the P3HT/PS OTFTs, the output current of the porous OTFTs is somewhat lower than that of the dense OTFTs. However, the Ion reduction here is far more substantial since in the blend, the C8BTBT crystallinity is strongly suppressed. The electrical parameters of both devices are summarized in table S1. Note that the μ, VT, and Ion/Ioff metrics of the dense devices are 2.72 × 10−2 cm2 V−1 s−1, −7.5 V, and ~106, respectively, parameters close to those of spin-coated C8BTBT films (33). For the porous device, the Weff for the porous C8BTBT/PS OTFT is ~750 μm, while μ, VT, and Ion/Ioff are 4.1 × 10−3 cm2 V−1 s−1, −1.88 V, and ~105, respectively. However, despite the reduced IDS and mobility, the porous OTFTs show greatly enhanced sensing performance. The transfer curves upon NH3 exposure from 0 to 20 ppm, the real-time responses, and the gas selectivities for both dense and porous C8BTBT/PS TFTs are shown in fig. S5 (A and D), and the IDS, VT, and μ values for these devices, extracted from the characteristic transfer curves, are summarized in table S1 and plotted in fig. S5 (E and F) as a function of the NH3 concentration. For example, upon exposure to 20-ppm NH3, the dense C8BTBT/PS OTFTs exhibit a variation of 16.3% for IDS, 37.3% for VT, and 6.9% for μ, whereas for the porous devices, the variations are 30.2% for IDS, 313% for VT, and 13.8% for μ. The calculated sensitivities of the two devices shown in Fig. 4G at different NH3 concentrations also indicate that the porous C8BTBT/PS OTFT is far more sensitive, with a maximum S of 12.5%/ppm.

The sensor response of an n-type semiconducting polymer was also investigated since this class of semiconductors is not widely used in chemical sensors because of the limited air/water stability of most and the typically low carrier mobilities compared to p-type small molecules and polymers (34). From the transfer and output curves of both dense and porous OTFTs (fig. S6), the μ, VT, and Ion/Ioff parameters of the dense N2200/PS devices are 5.6 × 10−3 cm2 V−1 s−1, +21.5 V, and ~104, respectively, while those of the porous devices are 1.5 × 10−3 cm2 V−1 s−1, 15.9 V, and ~103 for a Weff of ~720 μm. The sensitivity values to NH3 of both devices are shown in Fig. 4G, and the porous ones clearly exhibit significantly greater sensitivities (about three to four times) than the dense devices when tested with NH3 over a concentration range of 5 to 20 ppm.

Electrochemical transistors and chemically doped films

We also explored the potential advantages of porous polymer film morphologies for other two organic electronics modalities: electrochemical transistors [organic electrochemical transistors (OECTs)] and chemical doping of pristine semiconducting polymers. Figure 5A shows the top-gate, bottom-contact OECT structure used in this study. Bottom Cr/Au contacts were patterned by photolithography to achieve an electrode pattern with W/L = 2660 μm/10 μm. Both porous and dense ~50-nm-thick P3HT/PS films, fabricated as discussed above, were used as the semiconductor layer. A phosphate-buffered saline (PBS; pH 7.4) solution was used as the gate electrolyte, and a tungsten probe functioned as the gate electrode. The first scan transfer curves of these OECTs (Fig. 5B) show the presence of both field-effect and electrochemical transport regions [for detailed information, see fig. S7 (A and B)]. For the dense P3HT/PS device, the transition from field-effect to electrochemical doping occurs at ~−1.68 V, a far more negative potential than that (−1.37 V) of the porous P3HT/PS OECT. Furthermore, the peak drain current of the porous OECT is ~2 mA, 30× higher than that of the dense one (0.067 mA). Note that the leakage current of all these OECTs is <10−6 A, indicating negligible contribution from electrochemical water splitting (fig. S7, A and B).

Fig. 5 Schematic of the device structures and electrical properties of porous electrochemical transistors.

(A) Schematic device structure for an OECT. (B) Transfer curves for both dense and porous P3HT/PS films. (C) Transient behavior of OECTs with dense and porous P3HT/PS films. (D) Schematic diethylenetriamine (DETA) doping of N2200/PS resistive devices. (E) Electrical conductivity of dense and porous N2200/PS films as a function of doping time. (F) Absolute I-V curves of dense and porous N2200/PS films after 20 min of doping.

For the OECT devices, a key figure-of-merit parameter is μC*, where C* is the channel capacitance per unit volume, which is affected by the materials used for device fabrication, semiconductor film morphology and thickness, and device geometry (35, 36). This parameter can be extracted from the equation gm = WdL−1μC*(VGVT), where gm is the transconductance; W and L are the channel width and length, respectively; d is the active layer thickness; and VT is the threshold voltage using devices with different W/L geometries. We estimated the μC* using this equation, where gm is peak transconductance at each testing, W/L = 2660 μm/10 μm, VG is the gate voltage corresponding to the peak transconductance, and d = 50 nm. Note that the porous morphology has less mass than the dense morphology, and thus, the μC* that we report is necessarily underestimated. For the first scan, μC* is estimated to be 2.4 and 0.3 F cm−1 V−1 S−1 for the porous and dense P3HT/PS devices, respectively. These differences demonstrate that the porous morphology enables faster doping than the dense films, and thus, the much lower transconductance and μC* for the dense P3HT/PS OECTs indicate that this device is not fully turned on during the first scan as the result of slow ion intercalation into the dense film compared to the porous one. After several scans, the stable transfer curves of the P3HT/PS OECTs are shown in fig. S7C. The transconductances gm/W, extracted from the maximum slope of the transfer curve (gm = dId/dVg), for porous and dense P3HT/PS devices are in the range of 22.5 ± 10 and 21.5 ± 8.5 mS cm−1, respectively. The corresponding μC* is estimated to be 22.5 ± 7.5 and 20.5 ± 5.5 F cm−1 V−1 s−1 for the porous and dense P3HT/PS devices, respectively. To compare our results with the literature, we have also fabricated porous/dense P3HT/PS and conventional dense P3HT OECTs using Ag/AgCl-KCl as the gate electrode/electrolyte (fig. S8) since this is a more common architecture (3538). Comparative μC* values for the tungsten/PBS and Ag/AgCl-KCl devices are shown in tables S2. The first and stabilized I-V scans of these porous/dense OECTs demonstrated similar trends as those of the tungsten-based devices but with much lower operating gate voltage, which is expected for an electroactive electrode (39, 40). Thus, for the first scan, the porous OECT turns on at ~−0.56 V and achieves a gm/W of 17.6 mS cm−1, while the dense devices turn on at ~−0.65 V and achieve a gm/W of only 4.2 mS cm−1. As expected, transfer curves of both P3HT/PS devices after several scans exhibit similar performance (fig. S8, D and E), with a μC* of 14.5 and 15.5 F cm−1 V−1 s−1 for the porous and dense P3HT/PS devices, respectively. Note that the performance for the P3HT/PS devices after stabilization are comparable to that of the control OECTs based on pristine (dense and without PS) P3HT films (table S2). Further μC* improvement can be expected by taking advantage of porous versus dense structures and using OECT semiconductors more suitable than P3HT in terms of hydrophilicity and capacity for ion uptake (35).

To quantify the advantage of structural porosity in terms of ion penetration, we investigated the temporal characteristics of the tungsten devices by applying a stepped gate voltage [0 V (off status) → −1.8 V (on state)] to determine the switching speeds. From Fig. 5C, it can be seen that the porous device can be switched on (response time, 20 s) from the first scan, whereas the dense one cannot function within the time of the experiment (60 s). From the transient behavior of the scans presented in fig. S7D, the response times of the porous devices (~4.7 s) are clearly far shorter than those of the dense ones (~88 s), reflecting far more rapid Helmholtz layer formation and the associated establishment of the porous semiconductor transistor channel (41). Note that similar conclusions can be drawn when investigating OECTs using Ag/AgCl (fig. S8F), with the porous device current saturating in ~5 s, while the dense one does not achieve saturation even after 25 s.

Chemical doping is an effective means to increase the conductivity of OSC films for fundamental studies and applications in emerging organic touch screen and thermoelectric functions (9). One variant consists of the chemical (redox) reaction between a molecule (a molecular dopant) and the semiconductor functioning as a host material, producing charged structures via electron transfer (42). Here, we studied the electron doping of the n-type polymer N2200, functioning as electron acceptor, using diethylenetriamine (DETA) vapors, functioning as the electron donor (Fig. 5D). DETA was used since it is an effective reductant for N2200 (42). Dense and porous N2200/PS films were fabricated on SiO2/Si substrates with Au as contacts. Figure 5E shows the absolute electrical conductivity of both types of N2200/PS films as a function of DETA exposure time, derived from the current measured at 10 V, and Fig. 5F shows representative absolute I-V curves collected after 20 min of doping. The conductivity of the undoped N2200/PS films is ~1.5 × 10−5 S/m and increases by >5000× to ~0.13 S/m (porous) and ~0.085 S/m (dense) after ~80 min of exposure to DETA vapors, which are values similar to those reported for doping pristine N2200 films (9, 43). The conductivity of the porous film is greater at all times, suggesting an overall greater charge density, possibly because of a larger density of dopant molecules on the pore surfaces. Furthermore, the steeper conductivity increase for the porous N2200/PS film demonstrates more efficient doping, meaning faster diffusion of the DETA molecules into the porous structure.


Here, we demonstrate a new approach to fabricate porous organic semiconducting films for diverse applications such as TFTs, sensors, electrochemical transistors, and chemically doped films. We show that the breath figure film fabrication process is very tolerant of semiconductor chemical structure and charge transport characteristics and enables the creation of films with nanoscopic circular porous features reaching down to the substrate interface. Comparison of the porous and corresponding dense film controls demonstrates the superior characteristics of the former film morphology for analyte diffusion, redox processes, and molecular inclusion. These data indicate that the porous films enhance OTFT-based chemical sensor performance, the switching speed of OECTs, and the doping process/carrier density/conductivity of OSCs. This study not only demonstrates an effective, scalable methodology for fabricating porous semiconducting films but also offers a platform for high-performance (semi)conducting devices based on them.


Device fabrication

P3HT [Mw = 43.9 kDa and polydispersity index (PDI) = 1.8] was purchased from Reike Metals Inc. and used without further purification. N2200 (Mn = 26 kDa and PDI = 1.9) was obtained from Flexterra Inc. (ActivInk N2200). The semiconductors C8BTBT, PS (Mn = 35,000), DETA, and PBS buffer solution were purchased from Sigma-Aldrich and used without further purification. For OTFTs and n-type doping experiments, 6-mg PS, P3HT, C8BTBT, and N2200 were separately dissolved in 1 ml of chloroform, and the solutions were stirred at 60°C for 3 hours. Next, equal amounts of OSC/CHCl3 and PS/CHCl3 solutions were added, and the resulting solutions were stirred at 60°C for 1 hour. Then, the solutions were spin-coated on the SiO2/Si substrates at 2000 rpm for 30 s. To obtain the porous structure, the RH must be maintained at ~60% during the film growth process. Next, the resulting films were thermally annealed at 80°C in vacuum for 3 hours. Last, gold (Au) source and drain electrodes were deposited on the films by vacuum evaporation through a shadow mask, creating a channel length of 100 μm and channel width of 1000 μm. For OECTs, bottom Cr (3 nm)/Au (25 nm) electrodes with a channel width of 2660 μm and a length of 10 μm were patterned with photolithography. The fabrication of blended semiconducting films was identical to that used for OTFT fabrication. One drop of a PBS solution (or 0.1 M KCl) was placed on the channel region, and the gate electrode tungsten or Ag/AgCl pellet was inserted directly into the solution during measurements.

Thin film characterization

Films morphologies were measured with a Veeco Dimension Icon AFM system in the tapping mode. GIWAXS measurements were performed at Beamline 8-ID-E at the Advanced Photon Source at Argonne National Laboratory. Samples were irradiated with a 10.9-keV x-ray at an incidence angle of 0.14° in air for two summed exposures of 2.5 s (totaling 5 s of exposure), and scattering x-ray was recorded by a PILATUS 1M detector located 228.16 mm from the sample. The films thickness was measured by a contact probe Dektak 150 profilometer.

Device characterization and sensor evaluation

OTFT and OECT characterizations were performed in air on a custom probe station using an Agilent B1500A semiconductor parameter analyzer. The electron mobility (μ) of OTFT was calculated in the saturation region using Eq. 1, where Ci is the capacitance per unit area of dielectric layer and W and L are channel width and length, respectively.IDS=12μFECiWL(VGSVT)2(1)For gas-sensing tests, the OTFT sensor was stored in an airtight test chamber (≈2.4 ml). A mixture of dry air and certain gas analyte (100-ppm standard NO2 gas, 100-ppm standard NH3 gas, 1000-ppm standard H2 gas, or 1000-ppm standard CO2 gas) in appropriate concentrations was introduced into the test chamber using mass flow controllers. For the n-type doping experiment conducted in ambient conditions, 4 ml of DETA were placed into a 20-ml glass vial. Next, the N2200/PS films on SiO2/Si substrates with Au as two-terminal electrodes prepared as described above were placed on top of the vial with the semiconductor film facing the vial interior. I-V measurements were conducted after different exposure times (0 to 110 min).


Supplementary material for this article is available at

Fig. S1. AFM images of porous semiconducting polymer films and 2D GIWAXS characterization.

Fig. S2. Output curves and sensor performance of P3HT/PS polymer transistors.

Fig. S3. Device stability of porous P3HT/PS polymer transistors.

Fig. S4. Structural characterization and electrical performance of indicated films and their transistors.

Fig. S5. Gas sensor performance of C8BTBT/PS organic transistors.

Fig. S6. Gas sensor and electrical performance of N2200/PS polymer transistors.

Fig. S7. Electrical performance and transient behaviors of P3HT/PS OECTs with tungsten gate electrodes.

Fig. S8. Electrical performance and transient behaviors of P3HT/PS OECTs with Ag/AgCl gate electrodes.

Table S1. Electrical parameters of dense and porous C8BTBT/PS OTFTs exposed to the indicated NH3 concentrations.

Table S2. Comparison of the μC* value of our work and the references.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: We thank AFOSR (FA9550-18-1-0320), the Northwestern University MRSEC (NSF DMR-1720139), and Flexterra Inc. for support of this research. X.Z. thanks the National Natural Science Foundation of China (grant no. U1504625) and the Youth Backbone Teacher Training Program in Henan province (grant no. 2017GGJS021). This work made use of the J. B. Cohen X-Ray Diffraction Facility, EPIC facility, Keck-II facility, and SPID facility of the NUANCE Center at Northwestern University, which received support from the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource (NSF NNCI-1542205), the MRSEC program (NSF DMR-1720139) at the Materials Research Center, the International Institute for Nanotechnology (IIN), the Keck Foundation, and the State of Illinois, through the IIN. Author contributions: B.W., A.F., X.Z., and T.J.M. conceived and designed the experiments. X.Z. and B.W. performed the experiments of transistor-based sensors and chemical doping. B.W., L.H., Z.W., and W.H. performed the experiment of electrochemical transistors. W.H., W.Z., Z.W., Y.C., and Y.M. helped analyze the data. X.Z., B.W., L.H., A.F., and T.J.M. wrote the paper. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article