High Performance Polymer-Small Molecule Blend Organic Transistors

Submitted to

DOI: 10.1002/adma.((200801725))

High Performance Polymer-Small Molecule Blend Organic Transistors**

R. Hamilton1*, J. Smith2, S. Ogier3, M. Heeney4, J. E. Anthony5, I. McCulloch1, J. Veres6, D. D. C. Bradley2, T. D. Anthopoulos2

[*] Departments of Chemistry1 and Physics2, Imperial College London,

South Kensington, SW7 2AZ, (United Kingdom)

E-mail: ;

UKPETeC3, NETPark, Sedgefield,

County Durham TS21 3FD, (United Kingdom)

Department of Materials4 , Queen Mary, University of London,

Mile End Road, E1 4NS, (United Kingdom)

Department of Chemistry5, University of Kentucky,

Lexington, KY 40506-0055, (U.S.A.)

Eastman Kodak6, 1999 Lake Avenue,

Rochester, NY 146500, (U.S.A.)

We are grateful to the Engineering and Physical Sciences Research Council (EPSRC) and Research Councils UK (RCUK) for financial support. TDA is an EPSRC Advanced Fellow and an RCUK Fellow/Lecturer.

Keywords: TIPS, F-ADT, Organic Semiconductors, Organic Transistors, OFETs

Abstract.

Solution processable organic semiconductors have shown significant improvement over recent years, and are now poised for mainstream commercialisation. Although the electrical performance of the best devices are now in excess of the first generation application requirements, increasing complexity will demand improved semiconductor charge carrier mobilities. Functionalised oligoacenes have demonstrated both solution processability and high charge carrier mobility, however small molecules may demonstrate limitations in fabrication compatibility with printing techniques. Here we show that a blended formulation of semiconducting small molecule and a polymer matrix can provide high electrical performance within thin film field effect transistors (OTFTs), demonstrating charge carrier mobilities of greater than 2 cm2V-1s-1, good device to device uniformity and the potential to fabricate devices from routine printing techniques.

Main Text

Solution-deposited, robust alternatives to amorphous silicon have been pursued commercially for over a decade.[1-3] Introduction of an economically viable technology that enables large area flexible displays[4, 5] as well as ubiquitous cheap electronics such as radio frequency identification tags[6, 7] is expected to be highly disruptive to the silicon dominated market.[8, 9] Solution-deposited small molecules[10-14] and polymers[15-17] are viable approaches that promise to meet the challenge. To date, small molecules have provided the highest headline field effect mobilities,[18, 19] but device-to-device variation due to morphology issues makes large area deposition via printing difficult,[20, 21] while high solubility makes finding orthogonal solvents for further solution processing a considerable constraint. Polymers however, demonstrate excellent device uniformity[22] and solution rheology, which makes them ideal for large area printing,[5] but have not yet demonstrated the high mobility to make them truly useful in commercial devices.[23]

Combining the high mobility of crystalline small molecules with the device uniformity of polymers is very attractive and has been approached in a number of ways. These include increasing the crystallinity of a polymer such as p3HT[16] by introducing rigid units into the polymer back-bone and creating regiosymmetric monomers. Polymer-small molecule blends have previously been investigated[24, 25] in an attempt to produce ambipolar devices but rather than enhancing performance, the blending in these investigations appeared to diminish the peak electron and hole mobility of each component. Using a blend of small molecules and polymer to enhance the performance and improve deposition has been described in the patent literature.[26, 27] By blending a soluble, highly crystalline p-type small molecule organic semiconductor (OSC) with an inert or field-effect active polymer significant enhancement in performance is claimed.

Here, two acene semiconductors first described by John Anthony, TIPS-pentacene[28] and diF-TESADT[18, 29] (Figure 1), are blended with both inert and field-effect active polymers with the expectation that peak device performance will be maintained while device uniformity is improved. OTFTs fabricated from soluble blends of polymer and small molecule are cast using spin-coating, which we will show causes preferential vertical phase separation of the two components. The small molecule is forced to the exposed interface, allowing large crystals to form within the channel region of the device.

To evaluate the effect of the blend morphology devices were fabricated in a dual gate structure as well as a more conventional top gate design. The processing conditions are documented in the methods section, but Figure 1 (c) shows a schematic of the standard OTFT whilst Figure 2 (b) shows the dual gate device. Constructing transistors that have a bottom and top gate within the same device will elucidate differences in device performance between the bottom and top OSC interfaces, but will not differentiate between dielectric effects, channel injection and morphology changes. The leakage current between the two gates was always found to be less than the lowest measured off current thus allowing channels to be probed independently. Devices were operated in dual gate mode by biasing one gate to a constant voltage whilst sweeping the other to obtain transfer characteristics. This produced a shift in the threshold voltage dependent on the fixed gate voltage, which is consistent with dual gate operation.

Figure 2 (a) shows the transfer characteristics for top and bottom channels within a single device, fabricated using a 1:1 by weight blend of TIPS-pentacene and the insulating polymer poly(α-methyl styrene) spin-coated onto the bottom gate dielectric. The choice of an inert polymer matrix for the dual gate device maximised any difference in mobility between top and bottom gate which might be due to vertical phase separation. Uniform semiconductor film formation was observed, whilst the saturation mobility of holes within the bottom channel was 0.10 ± 0.05 cm2V-1s-1 increasing to 0.5 ± 0.1 cm2V-1s-1 in the top channel. Moving from bottom to top gate also showed an improvement in on/off current ratios, reduced hysteresis and threshold voltage shifts from greater than +10 V to between -5 and -10 V. Processing issues, described in the methods section, forced the use of different dielectric materials for the bottom and top channel, which could account for the difference in performance seen. However, devices were analysed by a DSIMS technique (results below) which indicate a higher concentration of TIPS-pentacene in the top-channel. The maximum mobility achieved using TIPS-pentacene : poly(α-methyl styrene) in a top-gate-only device was found to be 0.69 cm2V-1s-1, which is only slightly higher than measured in the dual gate transistors showing that the altered structure does not adversely affect the top channel operation.

In order to make an improvement in the mobility a change of polymer matrix is needed. It is suggested that phase separation (not necessarily vertical as described before) of TIPS-pentacene and poly(α-methyl styrene) causes a reduction in the effective channel width and thus a lowering of the measured mobility for devices based on insulating polymers. Replacing poly(α-methyl styrene) with the amorphous p-type polymer poly(triarlyamine) (PTAA) (FlexInk) creates conduction pathways between separate crystalline pentacene-rich regions and improves the performance of the OTFT.

Figure 3 shows typical transfer and output characteristics of top gate devices (channel length (L) of 60 μm and width (W) of 1000 μm) made from (a) TIPS-pentacene and (b) diF-TESADT, both blended with PTAA. The highest mobility devices using TIPS-pentacene had a slight deviation from ideal square law behaviour at gate voltages greater than 50 V, due to the drain being biased to -40V. Charge injection appeared to be efficient, since there is a good linear output at low drain voltages and despite the higher currents within diF-TESADT based transistors no injection problems were observed. There is also a clear improvement in hole mobility over the devices made using TIPS-pentacene : poly(α-methyl styrene) blends. The highest mobility obtained was in a diF-TESADT : PTAA device which had a saturation mobility of 2.41 ± 0.05 cm2V1s1 and a linear mobility of 1.88 ± 0.04 cm2V-1s-1.

In addition to high mobilities the important characterisation parameters for the devices remained constant and reproducible over the whole sample. A typical, non optimised sample (having lower quality evaporated source-drain electrodes but more transistors per sample) was tested and averages taken over 18 devices. The mean saturation mobility was found to be 0.66 ± 0.13 cm2V-1s-1 and the best device was 0.91 cm2V-1s-1. Similarly the mean threshold voltage was -7.2 ± 2.2 V and the on/off current ratio was 10(4.51 ± 0.32). Generally it was found that over 80% of the devices on a sample would show transistor behaviour. Table 1 clarifies the mobilities and on/off current ratios obtained for the various material and device designs employed.

Devices fabricated in air using TIPS-pentacene blends had much lower mobilities (~0.1 cm2V-1s-1) compared to those made in nitrogen. These results are consistent with the degradation of pentacene OTFT performance due to H2O and O2 acting as a charge trapping dopants introduced during annealing of the semiconductor layer[30]. However, after fabrication in nitrogen, exposure to air resulted in only a slow reduction in device performance since the permeability of CYTOP to H2O and O2 is low. Figure 4 shows how the mobility as well as on and off currents varied over a period of four weeks for a diF-TESADT based and a TIPS-pentacene based transistor. The TIPS-pentacene blend showed a slight drop in mobility after several hours while diF-TESADT blend transistors showed excellent stability and maintained a saturation mobility above 1.2 cm2V-1s-1 for the entire test. There was a gradual decrease in the on/off ratio due to increasing off current as bulk conduction in the channel became more significant.

Within the dual gate OTFTs the lower performance of the bottom gate can be attributed to two effects. Firstly, there will be a larger number of charge trapping sites on the BCB-semiconductor interface due to the oxygen plasma treatment creating polar groups on the surface. Secondly, we believe that there is vertical separation of components within the semiconductor layer. During film formation phase separation of the polymer and the small molecular material occurs, however, due to the high surface energy of the substrate, and in particular the high polar component of this energy, there is preferential crystallisation of the TIPS-pentacene (or diF-TESADT) at the semiconductor-atmosphere interface. This therefore increases the fraction of molecular solid within the conducting channel of the transistor in the top gate configuration.

Vertical profiling of the device structure by Dynamic Secondary Ion Mass Spectrometry (DSIMS) was used to confirm the distribution of the blend components. Cs ion bombardment was used to slowly sputter material from the film, and the resulting ejected ion species measured by mass spectrometer. Figure 5 shows the first signal to rise is the Si from the silyl group on the TIPS-pentacene molecule. The nitrogen signal (from the PTAA) can be seen to rise to a maximum approximately 20 - 30nm beneath the top surface. When scanned over the channel (a) the Si signal is seen to rise at a probed depth of ca. 50 nm, which could indicate an increase in concentration of TIPS at the glass substrate surface. This would explain why a field effect is still seen in bottom gate devices. When scanned over the electrodes, however, (b) the fluorine peak indicates the position of the self-assembled monolayer of pentafluorobenzene thiol (PFBT) and here TIPS seems to be excluded from the surface. The broadening in the F peak (which should in theory be a monomolecular layer <1nm thick) is due to a combination of a variable rate of etch for different materials in the layer and a slight spread in the energy density of the Cs ion beam. The gold signal rises immediately afterwards as the profiling reached the source and drain electrodes, and finally the Si signal rises once more indicating profiling into the glass substrate. A depth profile of an area between the source and drain electrodes showed the same distribution of the TIPS and PTAA, indicating that the phase separation occurs over the entire area of the spin coated sample.

The TIPS-pentacene : PTAA system was chosen to illustrate the suspected vertical phase separation, as each component contains an element that identifies it as being located at a certain depth. Using this analysis, it is apparent why a top-gate device may perform better than a bottom-gate. Although there is sufficient TIPS-pentacene in the bottom channel to allow a reasonable field-effect mobility, the higher concentration in the top channel provides an explanation of the better performance. The vertical phase separation in this system is assumed to apply to all small molecule polymer blends in this study.

The films were also studied using polarised light microscopy to observe the morphology of the crystallites and atomic force microscopy (AFM) to map the surface of the semiconductor which forms the conducting channel within the OTFT. Figure 6 shows both diF-TESADT and TIPS-pentacene based blend films. It is clear that some crystallisation of the molecular material into spherulitic-type structures has occurred. Crystallisation occurs over the whole sample when annealed for longer than 2 min at 100 °C, less than this quenches the sample leaving some amorphous regions which are not suitable for high performance devices. The AFM shows a clear difference between the amorphous part of the film, with a r.m.s. surface roughness of 6.4 nm, and the crystallites, with a surface roughness of 18 nm. This increase in surface roughness and the change in appearance of the film suggest that some of the TIPS-pentacene is forming on the top surface during crystallisation – the situation that we require for good top gate OTFT performance. The optical micrographs also show the effect of contact induced crystallisation on PFBT coated gold contacts[11]. Much larger and more spherulitic crystals are produced on the gold surfaces in the case of diF-TESADT blends where it is suggested that fluorine interactions promote the crystallisation. In the case of TIPS-pentacene blends the effect is less pronounced as the same F-F interactions do not exist, which also agrees with the better charge injection observed in the diF-TESADT devices.

Further annealing of the diF-TESADT based transistors at 100 °C after fabrication resulted in a lowering of the off current, for example after 1 hour the on/off current ratio typically increased from approximately 103.5 to 104.2. This is also consistent with an improvement in vertical phase separation on heating which would reduce bulk conduction through the semiconductor film and thus lower the current in the off state of the device.

Simple unipolar inverter circuits were constructed using TIPS-pentacene and diF-TESADT blend transistors showing their possibility for use in real devices. We have demonstrated that an inverter gain of greater than 10 and good noise margins can be achieved. Again with a view to commercial viability we have shown that devices can be fabricated on poly(ethylene terephthalate) (PET) films with very little loss of performance. Using a TIPS-pentacene : PTAA blend for the semiconductor, saturation mobilities of up to 1.13 ± 0.05 cm2V-1s-1 were measured and threshold voltages and on/off current ratios remained the same as for the devices made on glass. The key feature here was the maximisation of the substrate surface energy by oxygen plasma in order to enhance uniform film formation and vertical phase separation of the blend components.