Fan Zhang,
Yufeng Hu*,
Zhidong Lou*,
Xige Xin,
Meng Zhang,
Yanbing Hou and
Feng Teng
Key Laboratory of Luminescence and Optical Information, Ministry of Education, Beijing Jiaotong University, Beijing 100044, China. E-mail: yfhu@bjtu.edu.cn; zhdlou@bjtu.edu.cn; Tel: +86 10 5168 4860
First published on 21st March 2018
The properties of semiconductor/dielectric interfaces are crucial to the performance of polymer field-effect transistors. The key to fabricating high-performance polymer transistors by spin-coating is solving solvent corrosion issues, wherein the solvent of the top polymer produces a rough interface or damage on the underlying polymer layer during deposition. Herein, we propose a mixed-solvent method that employs a mixture of an orthogonal solvent of the underlying polymer and a good solvent of the top polymer as the solvent of the top polymer to prepare polymer bilayers and produce a comparative study of the trap density at the semiconductor/dielectric interface of the corresponding transistor. By changing the ratio of orthogonal solvent to good solvent, namely the degree of orthogonality of the mixed solvent with respect to the underlying polymer, the interface and film qualities of polymer bilayers can be well controlled. We applied this method to spin-coat poly(3-hexylthiophene) (P3HT) on poly(methylmethacrylate) (PMMA) with a mixture of cyclohexane (orthogonal solvent) and chloroform (good solvent). The results of morphology characterizations and electrical property studies indicate the optimal ratio of cyclohexane to chloroform for preparing high-quality P3HT/PMMA bilayers for field-effect conduction is 7:3. Transistors based on the optimal bilayers with a bottom-gate/top-contact configuration and a long channel length show good performance. The trap density at the P3HT/PMMA interface is evaluated to be 3.6 × 1012 cm−2 eV−1 from the subthreshold swing, characterizing the distribution of the interface trap levels across the bandgap in P3HT. Furthermore, based on deviations from ideality in the capacitance–voltage characteristics of the metal–insulator–semiconductor capacitor in the device, the traps at the interface are found to be acceptor-type, with the trap density determined to be 2.3 × 1011 cm−2. This value is in a good agreement with that estimated from the subthreshold swing.
However, it is usually difficult to find perfect orthogonal solvents for polymer semiconductors or dielectrics that are soluble in common organic solvents. For example, most π-conjugated polymers and solution-processable small molecules are soluble in organic solvents that can also dissolve some practically applied dielectrics, such as poly(methyl methacrylate) (PMMA) and polystyrene (PS).7 To avoid the solvent corrosion effect, a dry transfer technique11 has been utilized to construct polymer FETs based on poly(3-hexylthiophene) (P3HT) and PMMA with a bottom-gate/top-contact (BGTC) geometry. Recently, a P3HT-top and PMMA-bottom bilayer structure has been developed through vertical phase separation of a P3HT/PMMA blend during spin coating and used to fabricate P3HT/PMMA FETs.12 Herein, we propose a mixed-solvent method that employs a mixture of a perfect orthogonal solvent with respect to the underlying polymer and a good solvent for the top polymer. The degree of orthogonality of the mixed solvent with respect to the underlying polymer depends on the ratio of the orthogonal solvent to the good solvent. Accordingly, the interface morphology and roughness of the polymer bilayer can be controlled by changing the mixed solvent ratio to fabricate a high-quality bilayer. Similarly, a mixture of toluene and cyclohexane has been used to deposit poly(9,9-dioctylfluorene-alt-bithiophene) (F8T2) on PMMA to fabricate F8T2/PMMA bilayer FETs with controlled interface roughness.10 This mixed-solvent method will allow the exploration and utilization of various combinations of polymer semiconductors and dielectrics to fabricate high-performance FETs that cannot be accessed through film deposition using a solution-based process due to solvent corrosion problems.
A rough semiconductor/dielectric interface is commonly thought to introduce large amounts of defects into the semiconductor layer, which enhance the formation of charge traps at the interface and hinder charge transport.13,14 Therefore, the increasing amount of interfacial traps reduces the field-effect mobility of a semiconductor.15,16 The density of interface traps can be evaluated from the subthreshold swing of an FET device.17,18 However, whether the interface traps are donor-type or acceptor-type cannot be distinguished with this method. In this regard, it is necessary to further analyze the properties of the semiconductor/dielectric interface using a capacitance technique.19,20 A polymer FET with a dielectric layer sandwiched between a gate electrode and a semiconducting layer can be considered a metal–insulator–semiconductor (MIS) capacitor. Capacitance–voltage (C–V) measurements of an MIS capacitor provide a useful approach for studying the properties of a semiconductor/dielectric interface. This technique can rapidly evaluate the total interface-trapped charge and flat-band shift, from which the sign of the charge can be determined. To our knowledge, no comparative studies of the trap density at the semiconductor/dielectric interface in a polymer FET based on the subthreshold swing method and MIS capacitance technique have been reported.
Herein, we utilize the mixed-solvent method to produce a P3HT-top/PMMA-bottom (P3HT/PMMA) bilayer for FETs, in which both polymers have been extensively studied and are commercially available. Polymer FETs based on these two materials usually adopt a top-gate/bottom-contact (TGBC) configuration4,5,9,21–25 due to P3HT solvents being able to dissolve PMMA. A mixture of a perfect orthogonal solvent (cyclohexane) for PMMA and a good solvent (chloroform) for P3HT was used as the solvent for P3HT. The effect of solvent ratio (degree of orthogonality) on the surface morphologies of the PMMA and P3HT layers, and the electrical properties of the PMMA layers, have been investigated. We have produced P3HT/PMMA bilayers with a clear and low-roughness interface. FETs based on the optimal P3HT/PMMA bilayers with a BGTC configuration and relatively long channel length show good device performance. The trap densities at the P3HT/PMMA interfaces are evaluated from the subthreshold swings and C–V characteristics of the MIS capacitors in the devices and show good agreement.
A schematic of the spin-coating process of P3HT/PMMA bilayers on indium tin oxide (ITO)-coated glass substrates with a sheet resistivity of 15 Ω sq−1 is shown in Fig. 1(a). The substrates were first cleaned with diluted detergent, successively ultrasonicated in deionized water, acetone, and ethyl alcohol for 30 min each, and then dried with nitrogen gas. PMMA was dissolved into ethyl acetate, which is among the best solvents for PMMA, at a concentration of 50 mg mL−1, and stirred for over 12 h. The PMMA films were spin-coated on the precleaned substrates from solution at 1000 rpm for 40 s, and then heated at 150 °C for 1 h to allow the solvent to fully evaporate. P3HT was diluted in mixtures of cyclohexane and chloroform with different volume ratios at a concentration of 3 mg mL−1. P3HT layers were deposited on the annealed PMMA films by spin coating the P3HT solutions at 1000 rpm for 40 s, and then dried at 80 °C for 20 min to afford the P3HT/PMMA bilayers.
Fig. 1 (a) Schematic illustration of the spin-coating process for P3HT/PMMA bilayers; (b) structure of FETs based on P3HT/PMMA bilayers with a BGTC configuration. |
Fig. 1(b) shows the BGTC structure of the FET devices based on the P3HT/PMMA bilayers. The devices were constructed on ITO-coated glass substrates with ITO, PMMA, and P3HT acting as the gate electrode, gate dielectric layer, and active layer, respectively. Au films (thickness, 80 nm) were thermally deposited on P3HT with a shadow mask to prepare the source and drain electrodes, defining a channel with a relatively long length of 400 μm and width of 10000 μm. Furthermore, a metal–insulator–metal (MIM) structure was used to study the electrical characteristics of the PMMA films, whose surfaces were treated with mixed cyclohexane/chloroform solvents with volume ratios (Cy:Ch) of 9:1, 8:2, 7:3, and 6:4, respectively. The films were sandwiched between ITO-coated glass and thermally evaporated aluminum (Al) films, which served as negative and positive electrodes, to form an ITO/PMMA/Al structure with an area of 4.5 mm2.
The morphologies of the PMMA films and P3HT/PMMA bilayers were characterized by scanning electron microscopy (SEM), and the surface roughness values of the PMMA films were estimated using atomic force microscopy (AFM). The currents of the PMMA dielectric layers and the output and transfer characteristics of the FET devices were measured with a Keithley 4200 semiconductor analyzer. C–V measurements of the PMMA layers and P3HT/PMMA FET devices were conducted with an Agilent E4990A impedance analyzer. All measurements were performed under ambient conditions.
Cross-sectional SEM images of the bilayers showed that the thickness of the PMMA layer decreased with increasing chloroform concentration in the mixed solvent. The thicknesses of the PMMA layers prepared at Cy:Ch ratios of 9:1, 8:2, 7:3, and 6:4 were estimated to be 420, 400, 380, and 240 nm, respectively. This observation was due to more PMMA being dissolved and re-spin-coated on P3HT, leading to a reduction in PMMA thickness due to a lack of strong interactions between P3HT and PMMA. In contrast, the P3HT layer thickness increased with chloroform content. At a Cy:Ch ratio of 9:1, the P3HT layer was much thinner because it was not completely dissolved in the mixed solvent according to the solubility of P3HT in chloroform.29 As the chloroform ratio was increased to 7:3 (30%), the thickness of P3HT increased to about 100 nm as P3HT became completely soluble.
The insets of Fig. 2 show top-view SEM images of the bilayers. The solvent ratio was also observed to affect the surface morphology of the P3HT film indirectly. A smooth and uniform P3HT surface was observed when the Cy:Ch ratio was varied from 9:1 to 7:3, whereas some clusters were observed at a 6:4 ratio. This might be due to increased interface roughness caused by the dissolution of more PMMA and subsequent redeposition of the dissolved PMMA. This is expected when a thin liquid film is deposited onto a rough solid substrate10 and will be confirmed by AFM characterization.
In FETs with a BGTC configuration, a dielectric surface was extremely important because a rough surface prevents charge transport by acting as a physical trap and transport barrier, or by disturbing the morphology and the microstructures of the semiconducting layer.30,31 To understand the impact of the cyclohexane/chloroform ratio (degree of orthogonality) in the mixed solvent, we characterized the PMMA surface treated with mixed solvent containing no P3HT using AFM measurements. This strategy was feasible because the amount of P3HT in the mixed solution was low and P3HT has no chemical interactions with PMMA. A similar approach8 has been used to examine the effects of solvents for PMMA layers on the surfaces of P3HT and poly ([N,N-9-bis(2-octyldodecyl)-naphthalene-1,4,5,8-bis(dicarboximide)-2,6-diyl]-alt-5,59-(2,29-bithiophene)) (P(NDI2OD-T2)) films in devices with a top-gate architecture.
Fig. 3 shows AFM micrographs of the pristine PMMA film surface and PMMA film surfaces treated with mixed solvent with Cy:Ch ratios of 7:3 and 6:4, and cyclohexane alone. Both PMMA films spun cast from cyclohexane and the mixed solvent with 7:3 ratio exhibited similar surface morphologies to the pristine PMMA film, while the film treated with a 6:4 solvent ratio showed some clusters. The root-mean-square (RMS) roughness values were evaluated as 0.391, 0.458, 0.418, and 14.858 nm, respectively. As expected, the RMS roughness of the PMMA film treated with only cyclohexane (100% Cy) was similar to that of the pristine PMMA film, showing that cyclohexane had no impact on the roughness of PMMA owing to its complete orthogonality to PMMA. The RMS roughness of the PMMA film spun cast from mixed solvent with a 7:3 ratio was comparable with that of the pristine PMMA film, which indicated that the PMMA surface remained almost unaffected, even at chloroform concentrations as high as 30%. Nevertheless, when the chloroform ratio was further increased to 6:4, the PMMA film showed a more coarse surface with a high roughness of 14.858 nm, confirming speculation about the increased interface roughness at a solvent ratio of 6:4. Evidently, only chloroform altered the degree of orthogonality of the mixed solvent and impacted the PMMA surface quality. When the chloroform ratio reached 6:4 (40%), a rougher PMMA surface was produced by the dissolution and redeposition of more PMMA during mixed-solvent treatment. The top-view SEM images in the insets of Fig. 2(c) and (d) and the AFM images in Fig. 3(c) and (d) show a similar trend in the surface morphology of the P3HT/PMMA bilayer and the RMS roughness of the mixed-solvent-treated PMMA surface as the chloroform concentration was increased.
Therefore, by simply changing the ratio of the orthogonal solvent to the good solvent, namely the degree of orthogonality with respect to PMMA, we were able to control the interface roughness and produce well-controlled P3HT/PMMA bilayers exhibiting solvent corrosion issues. The critical chloroform content in the mixed solvent was found to be 30% (Cy:Ch = 7:3), below which the RMS roughness of the interface was relatively low, at less than 0.500 nm, and the top P3HT layer showed a smooth and uniform surface. We concluded that the optimal ratio of cyclohexane to chloroform in the mixed solvent for producing a high-quality P3HT/PMMA bilayer was 7:3, because the bilayer had a P3HT thickness desirable for FETs, in addition to a clear interface and smooth P3HT surface morphology. Therefore, this mixed-solvent method is not limited to the P3HT/PMMA system, but could also be applied to fabricating other bilayer polymer FETs to overcome solvent corrosion problems encountered in film deposition using a solution-based process.
Fig. 4(b) shows that the capacitance per unit area of all ITO/PMMA/Al devices decreased when the signal frequency was increased from 100 to 106 Hz. This capacitance reduction with increasing frequency was mainly attributed to a reduction in the dielectric constant of PMMA,20 which resulted from dipoles failing to reorient with the electric field in the high-frequency range.32 The capacitance of the PMMA layer was also observed to increase with increasing chloroform concentration. The capacitance of a dielectric film is known to be inversely proportional to its thickness. Obviously, the increase in capacitance resulted from a reduction in the PMMA thickness with increasing chloroform concentration. The capacitance of the PMMA layer treated with a 7:3 solvent ratio at 100 Hz was 3.75 nF cm−2, which was sufficient for electric charge induction. Evidently, the optimal cyclohexane/chloroform mixed solvent ratio for P3HT was 7:3, in which a high-quality P3HT/PMMA bilayer was produced and the treated PMMA layer had a low current density and sufficient capacitance for field-effect conduction.
(1) |
Fig. 5 (a) Output and (b) transfer characteristics of the P3HT/PMMA bilayer FET prepared using cyclohexane/chloroform mixed solvent at the optimal volume ratio of 7:3. |
For comparison, previously reported polymer FETs based on P3HT and PMMA5,9,11,12,21,22,24,33 and our device are summarized in Table 1, indicating structure, fabrication technique, channel length and width, and mobility value. The mobilities of the polymer FETs based on P3HT and PMMA varied over three orders of magnitude, from 10−3 to 10−1 cm2 V−1 s−1, depending on device configurations, fabrication methods, and channel lengths. The mobility has been demonstrated to increase as the channel length is reduced.34 The highest mobility value of 0.2 cm2 V−1 s−1 was achieved in the TGBC devices by spin coating with a channel length of 20 μm.24 The BGTC devices of a 120 μm-long channel constructed by a dry transfer technique11 had a mobility of 1.4 × 10−3 cm2 V−1 s−1. Mobilities ranging from 0.001–0.022 cm2 V−1 s−1 were achieved in BGTC devices fabricated by vertical phase separation.12 Therefore, despite a much longer channel length of 400 μm, our FET devices based on P3HT/PMMA bilayers prepared using a cyclohexane–chloroform mixed solvent at the optimal ratio of 7:3 exhibited good device performance.
Polymer/polymer | Structure | Method | Channel L/W (μm) | Mobility (cm2 V−1 s−1) | References |
---|---|---|---|---|---|
PMMA/P3HT | TGBC | Spin coating | 100/17200 | 9.99 × 10−3 | 5 |
PMMA/P3HT | TGBC | Spin coating | 20/1000 | 0.02–0.04 | 9 |
PMMA/P3HT | TGBC | Spin coating | 20/1000 | 0.08–0.2 | 24 |
PMMA/P3HT | TGBC | Spin coating | 50/150 | 2.5 × 10−3 | 21 |
PMMA/P3HT | TGBC | Spin coating | 40/5000 | 0.05 | 33 |
PMMA/P3HT | TGBC | Dry transfer | 30/110 | 0.013 | 22 |
P3HT/PMMA | BGTC | Dry transfer | 120/3000 | 1.4 × 10−3 | 11 |
P3HT:PMMA | BGTC | Vertical phase separation | — | 0.001–0.022 | 12 |
P3HT/PMMA | BGTC | Mixed-solvent method | 400/10000 | 4.4 × 10−3 | This work |
Subthreshold swing S is another important device parameter for quantifying how sharply a transistor can be switched from the off-state to the on-state by a gate voltage.17–19 S is defined as the change in the gate voltage needed to induce a drain-current change of one order of magnitude (eqn (2)):
(2) |
Fig. 6(a) shows the same data as Fig. 5(b), but on a logarithmic scale. The red line is a linear fit of the measured data in the voltage region from the off-state to the on-state, from which a subthreshold swing (S) of 14 V per decade (dec) was estimated for the P3HT/PMMA FET device. This value was within the range of the subthreshold swings (1–20 V dec−1) reported for organic and polymer FETs in most published studies.35
Fig. 6 (a) Logarithmic plot of the data shown for P3HT/PMMA FET in Fig. 5(b). (b) Capacitance per unit area–voltage characteristics at a frequency of 500 kHz and normalized capacitance with respect to the maximum capacitance C0 in accumulation at −40 V for the polymer FET based on the P3HT/PMMA bilayer prepared with cyclohexane–chloroform mixed solvent at the optimal volume ratio of 7:3. The voltage was only applied to the ITO gate electrode and one Au electrode. Inset: Mott–Schottky plot of capacitance vs. voltage. |
(3) |
Lastly, to further study the properties of the traps at the P3HT/PMMA interface in the FET device, we conducted high-frequency capacitance measurements19 directly on the device with a voltage only applied between the ITO gate electrode and one Au electrode, which apparently constructed an ITO/PMMA/P3HT MIS structure. The area of the MIS capacitor was 0.2 cm−2, defined by the area of the Au electrode. The capacitance of the FET device was measured by sweeping the bias voltage from −40 to +40 V with a small superimposed AC signal of 100 mV at 500 kHz on top of a preselected DC voltage. Fig. 6(b) shows typical curves of capacitance per unit area (C) vs. voltage for the MIS capacitor from its normalized value (C/C0) to the maximum value (C0) at −40 V. These C–V curves were similar to that of an ideal MIS structure.19 Accumulation, transition, and full depletion regions, but no inversion region, were clearly observed as the voltage ranged from −40 to 40 V, which was consistent with the capacitance vs. voltage behavior of polymer MIS structures.20,38 The maximum value of the capacitance per unit area at −40 V was equal to that of the PMMA insulating layer (Ci) because the P3HT semiconducting layer driven into accumulation was in a conducting state.
The difference in work function between ITO (5.1 eV) and P3HT (5.0 eV) was 0.1 eV in the P3HT/PMMA FET device and could be considered negligible.20,38 We evaluated the trap density at the P3HT/PMMA interface according to deviations in the C–V curve from that of an ideal MIS capacitor, if all charges in the dielectric were assumed to be located at the interface. This assumption was reasonable because a greater effect on the C–V behavior was expected if the charges in the dielectric were closer to the interface. In general, charges in the interface traps of polymer MIS capacitors can shift the flat band voltage, which is zero for an ideal MIS capacitor, and deform the C–V curves.20,38 The total capacitance under flat-band conditions19 for an ideal MIS structure can be expressed as follows (eqn (4)):
(4) |
Usually, the dopants in a semiconducting polymer are residual or unintentional impurities. The dopant density can be extracted based on the standard Mott–Schottky relation for an MIS capacitor in depletion (eqn (5)):39–41
(5) |
The corresponding voltage of this flat band capacitance in Fig. 6(b) was the flat band voltage, VFB, which was about 6 V for the device. The flat band voltage shift in the positive voltage direction indicated negative charges trapped at the P3HT/PMMA interface with acceptor-type interface traps.20,38 Assuming that all charges in the dielectric were located at the interfaces, the density of the interface traps, Nit, responsible for the VFB shift was calculated using eqn (6):20
(6) |
Actually, the interface trap density derived from the subthreshold swing was the same as that calculated from the capacitance measurement, but expressed in a different unit. The interface trap density in cm−2 eV−1 characterized the distribution of the interface trap levels across the energy bandgap19 between LUMO and HOMO levels in a polymer semiconductor. Therefore, it was reasonable that the interface trap density of the PMMA/P3HT FET device in cm−2 eV−1 evaluated from the subthreshold swing was larger than that in cm−2 estimated from the C–V measurements.
We investigated the performance of FET devices based on the optimal P3HT/PMMA bilayers with a BGTC configuration. The device exhibited a comparable hole mobility of 4.4 × 10−3 cm2 V−1 s−1 despite a much longer channel length of 400 μm. The subthreshold swing of the device was 14 V dec−1, from which the maximum density of the interface traps was determined to be 3.6 × 1012 cm−2 eV−1, characterizing the distribution of the interface trap levels across the bandgap between the LUMO and HOMO levels in P3HT. The results of C–V measurements on the MIS structure in the device suggested that the traps at the P3HT/PMMA interface were acceptor-type and that the density of the interface traps was 2.3 × 1011 cm−2. Notably, the interface trap density in cm−2 eV−1, evaluated from the subthreshold swing, and that in cm−2, estimated from the C–V measurements, were in good agreement. Our mixed-solvent method is not limited to the P3HT/PMMA system, but could also be applied to fabricating other polymer FETs based on various combinations of polymer semiconductors and dielectrics that show solvent corrosion problems during successive film deposition using a solution-based process.
This journal is © The Royal Society of Chemistry 2018 |