Interface studies of well-controlled polymer bilayers and field-effect transistors prepared by a mixed-solvent method

Abstract

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.