Effects of trifluoromethyl substituents on interfacial and bulk polarization of polystyrene gate dielectrics

Polymers capable of static electronic polarization, known as electrets,¹ are of interest in applications such as electromechanical energy harvesting^2–4 and as a means of pre-setting parameters^5–7 and storing information^8–10 in organo-electronic devices. By using these materials as gate dielectrics in organic field-effect transistors (OFETs),^11–15 one produces “nonvolatile” transistors, in which the parameter being adjusted is the threshold voltage V_th, the voltage above which the transistor is considered to be “on” and displays charge carrier mobility that is relatively independent of gate voltage. Changing V_th can alter the power requirements of a device and is therefore useful in many applications, such as changing the sensitivity to sensor inputs^16–18 or alternating between different logic states in a circuit.^10,19,20 With the growing interest in using organic and polymer semiconductors for sensing and logic, polymer electrets offer a compelling alternative to inorganic materials for nonvolatile OFET gate dielectrics.

Many leading electrets such as poly(tetrafluoroethylene) and polyvinylidene fluoride are fluorinated.^5–7,21,22 The presence of fluorine increases the hydrophobicity of the polymers and stabilizes stored charges against dissipation by hydrolysis, but also presents fabrication challenges because of the relative insolubility of fluorinated polymers in nonfluorinated solvents. On the other hand, polystyrene (PS) is also a known and easily processed electret^23,24 but has limited charge storage stability. This work considers the contributions of a fluorinated substituent, trifluoromethyl (TFM), on the performance of styrene polymers in pentacene OFET gate dielectrics. The OFETs serve as a convenient probe of both interfacial and bulk polarization of the electret, providing information that is relevant to non-transistor applications of electrets as well. In previous work, we established a system for creating multilayered dielectric stacks of PS-based polymers with modified styrene monomers to enhance morphological and electronic properties.²⁵ In this work, we have applied this approach to poly(3-trifluoromethyl styrene) (P3TFMS) to prepare dielectrics to compare directly the electronic effects of the added TFM unit on interfacial and bulk electronic effects. We find that the TFM substituent markedly and independently affects both the initial V_th and the nonvolatile, shifted V_th observed after the application of static gate voltage, depending on its position adjacent or apart from the pentacene. This work provides evaluation of this dielectric polymer for its separate effects on the initial interfacial polarization of the OFET and subsequent bulk, static polarization.

The homopolymer P3TFMS and a crosslinkable copolymer, poly(3-trifluoromethyl styrene)-co-4-vinylbenzocyclobutene, were radical-polymerized to provide polymers appropriate for use in single-layer and multilayer dielectrics. The corresponding crosslinkable polystyrene-co-4-vinylbenzocyclobutene was also synthesized and commercially available atactic polystyrene (m.w. 50 000, Polysciences) was used. The synthesis of the prepared polymers was consistent with previous work²⁵ and is detailed in the supplementary material. The crosslinkable copolymers had 90% by number of the trifluoromethyl or styrene monomer and 10% of the 4-vinylbenzocyclobutene crosslinker.

OFETs were fabricated on silicon wafers in a vertically stacked geometry [Figs. 1(a) and 1(b)], consistent with previous work.²⁵ Single polymer layer dielectrics of PS or P3TFMS were deposited on gold gate electrodes by spin-coating from 80 mg/ml CHCl₃ solutions under dry glovebox conditions. This yielded layer thicknesses of ∼1 μm based on previous work²⁵ and as confirmed via a Filmetrics F20-NIR thin film analyzer. Semiconducting channels of pentacene (thickness = 40 nm) followed by gold source and drain electrodes defining channel dimensions of 6 mm width and 0.25 mm length were deposited by thermal evaporation through shadow masks on top of the dielectric layers. AFM measurements of the pentacene top surfaces showed comparable morphologies and surface roughnesses for the PS and P3TFMS-based devices (Fig. S1). Additional information on the device fabrication is included in the supplementary material.

The effects of charge injection and trapping in OFETs containing single layer dielectrics of both PS and P3TFMS are shown in Fig. 1. Output curves of the as-produced devices were measured for fixed gate voltages V_G [Figs. 1(c) and 1(d)], followed by a transfer curve measurement with fixed drain-to-source voltage V_DS = −90 V and sweeping the gate-to-source voltage V_G from 0 to −70 V [Figs. 1(e) and 1(f)]. Charge trapping in the dielectric layers was then driven by a static electric potential difference applied across the dielectric layer by grounding the gate electrode and setting the source and drain electrodes to −70 V for 5 min. This voltage was found to reliably affect the properties of the OFETs and 5 min was sufficient to allow full charging to occur (see supplementary material and Fig. S2). After this process, output and transfer curve measurements were repeated to determine the difference in electronic properties [Figs. 1(c)–1(f)]. The output curves all showed larger currents for given voltages during characterization following charging. Threshold voltages and mobilities were extracted from the transfer curves via a linearized regression of the saturation regime (⁠$VG<Vth$⁠), according to the approximate relationship²⁶ $ID∝VG−Vth2$⁠. Examples of these fits are shown in Figs. 1(e) and 1(f), and the results of these measurements on ensembles of both PS and P3TFMS devices are shown in Fig. 2. The key results of these measurements and analysis are that the P3TFMS devices show much smaller (negative) initial threshold voltages than the PS-based devices and also show much larger (positive) threshold voltage shifts $ΔVth$ in response to charging.

To distinguish interfacial and bulk contributions to the above effects, we investigated bilayer dielectrics using analogous techniques to compare the effects of the styrene and 3-trifluoromethylstyrene monomers incorporated into the polymers. These included PS/P3TFMS bilayers with the 3-trifluoromethylstyrene adjacent to the semiconducting layer (TFM adjacent) and apart from the semiconductor (TFM apart), as shown in Figs. 3(a) and 3(b), and control bilayers formed from a crosslinkable PS layer and a pure PS layer. A bottom crosslinkable layer of thickness ∼500 nm was spin-coated from a 40 mg/ml CHCl₃ solution, followed by thermally activated crosslinking and subsequent spin-coating of the second layer, again from a 40 mg/ml spinning solution, to yield total bilayer dielectric thicknesses comparable to the single layer structures.

Examples of the output and transfer curves for the two PS/P3TFMS bilayer geometries before and after charging at −70 V for 5 min are shown in Fig. 3. Data for a representative PS/PS control bilayer are shown in Fig. S3 of the supplementary material. Once again, the output curves all showed larger currents for given voltages during characterization following charging. The results for $Vth,i$ and $ΔVth$ for the three bilayer geometries are shown in Fig. 4. Notably, while the bilayers with P3TFMS adjacent to the semiconductor showed $Vth,i$ values comparable to the single layer P3TFMS devices shown in Fig. 2, those with TFM apart from the semiconductor had $Vth,i$ values similar to both the single-layer PS and bilayer PS samples. In contrast, both P3TFMS-containing bilayer geometries showed similarly larger $ΔVth$ shifts than PS-only devices. These results suggest that $Vth,i$ is governed by interfacial effects, and $ΔVth$ is an effect arising from the interior of the film.

To assess the stability of the electronic effects of static charge trapping, we subjected OFETs to multiple cycles of characterization. The devices were characterized by output curves, charged via the process described above, recharacterized by output curves, and then had a discharging voltage applied (“discharged”) by grounding the gate and setting the potential at the source and drain to 70 V and holding for 5 min. After each of these procedures, the state of the device was assessed via a transfer curve measurement. At the end of this series of measurements, another set of output curves was taken to verify that the device was still operating normally. This multistep process was repeated several times on each device.

The evolution of V_th of single-layer devices during this process is shown in Fig. 5 for both PS and P3TFMS dielectrics. Of specific interest is the degree to which a device retained its ΔV_th after a discharge process. Starting from the ΔV_th after the first charging process, pure PS devices on average retained 72% of their initial ΔV_th relative to the initial ΔV_th after post-charging output curve characterization but only 40% after the first discharge process. For the same processes, P3TFMS devices showed greater post-charging stability, retaining 96% of their initial ΔV_th after output curve characterization and 64% after discharging. Figure S4 shows the data from these measurements for bilayer samples. Samples containing a TFM polymer retained on average 71% of their initial shift after output curves were taken and 40% after discharge, compared with PS control bilayers which retained 75% and 30%, respectively.

In these polystyrene-based dielectric systems, by replacing the styrene unit with 3-trifluoromethyl styrene as the most prevalent monomer, we added properties associated with the trifluoromethyl group which altered the electronic and charge storage capabilities of the polymer. We found significant differences between the polymers in the electronic properties of single-layered systems both before and after charging (Fig. 2). We observed a difference in the initial or as-prepared V_th of OFETs of nearly 19 V (18.4 ± 6.4 V), and after charging, P3TFMS devices had 15 V (14.7 ± 6.2 V) larger ΔV_th shifts than PS devices. The data for the bilayer dielectrics showed similar statistically significant differences in both V_th,i and ΔV_th. Mobilities of devices were extracted from the transfer curve data and were initially 0.025 ± 0.003 cm²/V s and 0.022 ± 0.004 cm²/V s for PS and P3TFMS, respectively. After charging, mobilities increased to 0.045 ± 0.004 cm²/V s and 0.025 ± 0.003 cm²/V s for PS and P3TFMS, respectively. Average surface trap density was calculated from the subthreshold swing^28,29 in the region V_g = ( V_th + 10 V) to (V_th +17 V). We found that PS- and P3TFMS-semiconductor interfaces had trap densities of 59 ± 3 × 10¹³/m² V and 37 ± 2× 10¹³/m² V, respectively.

These data allow us to separate these phenomena into a surface effect on the initial performance of the device and a bulk effect on the magnitude of ΔV_th. The observed change in initial properties only occurred in devices for which the TFM-bearing polymer was adjacent to the semiconducting layer, making this a surface effect. Possible surface contributions to V_th in OFETs include trap states and dipoles at the dielectric-semiconductor interface.³⁰ To estimate the maximum possible effect of these dipoles, we note that the trifluoromethyl group introduces an electric dipole moment of ∼2.5 D to the monomer.³¹ Thus, a perfectly aligned layer of such dipoles at the number density of P3TFMS would produce an electrostatic potential difference of order 4 V. This is insufficient to produce the observed difference in V_th,i and suggests that this difference comes from a reduction in the surface density of trap states in P3TFMS.²⁶ This is verified by the difference in number density of traps in the two materials as extracted from the subthreshold swing.

We observed small but statistically significant differences in the mobilities of single layered devices after charging. The pentacene on both PS and P3TFMS layers showed an increase in mobility, consistent with the charging process filling trap states that had been inhibiting the mobility. The PS-based devices showed a greater mobility increase than P3TFMS-based structures, consistent with molecular dipoles on the TFM groups near the semiconductor interface inhibiting the mobility in P3TFMS devices, as compared with PS devices. In all devices, the overall magnitude of the mobility (<0.05 cm²/V s) was consistent with grain-to-grain transport being the dominant effect, with grain sizes of less than 0.5 μm (see supplementary material).^32–35 

The introduction of static electric charge by the application of a charging voltage results in a built-in electric potential difference across the dielectric layer which adds to the applied gate voltage of the OFET when it is operated and is observed as a change in the threshold voltage.^6,36 The larger ΔV_th that we observe after the application of a static charging field, arising in any dielectric stack that contains TFM-bearing polymers, is consistent with this being a bulk effect. There is a statistically significant difference in ΔV_th in both single and double layers containing a TFM-bearing polymer as compared with non-TFM bearing controls with identical layer geometries (Figs. 2 and 4). Further, we find that there is no measurable difference between the threshold voltage shifts of bilayers with TFM-bearing polymers in different locations, showing that this effect is independent of position. As we are well below the glass transition temperature for P3TFMS of ∼65 °C,^27,37 any major reorientation of the polymer will occur on timescales significantly longer than those probed here.^38,39 For these charging fields and dipole strengths, while we would expect small transient alignment of the molecular dipoles, primarily from reorientations of the phenyl rings, these are only expected to be stable on the timescale of several nanoseconds.^40–42 Therefore, bulk polarization should be negligible in this system and the ΔV_th values we observe come from the presence of injected static charge. The charge densities necessary to produce the observed potential differences are quite dilute, corresponding to 1 bound electronic charge per 2.6 ± 0.3 × 10⁵ monomers in PS and 1 charge per 6.9 ± 0.7 × 10⁴ monomers in P3TFMS.

During the repeated charging and discharge cycles, P3TFMS single layers showed greater capacity to accept and retain charge as compared with PS. Bilayers containing TFM polymers showed threshold voltage shift magnitudes, our measure of accepting charge, comparable to P3TFMS single layers, despite having only half as much TFM. This suggests that there may be extra charge storage capacity coming from the internal interface, a conclusion supported by several studies reporting observation of significantly increased dielectric properties in the presence of interfaces in polymer blends.^43,44 In addition, the bilayers showed threshold voltage shift stability consistent with all-PS devices, suggesting that the presence of PS in the bilayers provides mechanisms of charge dissipation which are not present in the P3TFMS.

In summary, replacement of the styrene monomer with the 3-trifluoromethylstyrene monomer results in a polymer that leads to more stable V_th shifts via beneficial interface and bulk charge trapping properties. This makes P3TFMS potentially attractive for electret applications. The reduction in bulk charge trap states and the increase in stability of charged devices make this a promising matrix for other functionalized comonomers bearing chargeable functional groups.²⁵ Finally, voltage-tunable organoelectronic systems such as thermoelectrics placed adjacent to TFM polymers would benefit from the added stability of the TFM subunit after charging as induced static electric potentials could be more robust.

See supplementary material for extended methods describing polymer synthesis (supplementary material 1), OFET fabrication and characterization (supplementary material 2), analysis of the impact of surface molecular dipoles on mobility (supplementary material 3), AFM images of pentacene layers on top of PS and P3TFMS (supplementary material 4), additional OFET transport data (supplementary material 5), and nuclear magnetic resonance (NMR) spectra of fluorinated polymers (supplementary material 6).

We thank James West and Ugur Erturun for assistance with Kelvin probe measurements and Patricia McGuiggan for assistance with AFM measurements. This work was supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under Award No. DE-FG02-07ER46465.