Schottky barrier diodes from semiconducting polymers

zyxwv zyxwv MOLECULAR ELECTRONIC DEVICES Schottky barrier diodes from semiconducting polymers zyxwvutsr zyxwvutsrqpo zyxwvutsrqpo zy H.L. Gomes D.M.TayIo r Indc,xitzg- lernzs: Schoitlcy-burrier diodes, I- V cliui.ucter'i.ctics,Senziconducting polymer,v. Polymer ,films - Abstract: Schottky barrier diodes based on All poly(3-methylthiophene)/Au have been fabricated and their electrical behaviour investigated. ILV characteristics revealed a dependence on the fabrication conditions, speciGcally on the time under vacuum prior to evaporation of the rectifying contact and post-metal annealing at elevated temperature. The available evidence is consistent with the formation of a thin insulating layer between the metal and the polymer following these procedures. Long periods under vacuum prior to deposition of the aluminium eleclrode reduced the likelihood of such a layer forming. Capacitance-voltage plots of the devices were stable to voltage cycling, so long as the forward voltage did not exceed -1 V. Above this a srnall degree of hysteresis was observed, which is attributed to the fillingiemptying of interface states or traps in the polymer. 1 Introduction Although the first semiconducting polymer was prepared some 20 years ago [I], it was the discovery three years later [2] that the films could be doped which led to intense research activity in this field. This early research focussed mainly on the mechanism of charge transport in these materials and led to an understanding of the main features required to obtain highly conducting polymers. Already the outcome of this work is seen in commercial applications such as antistatic coatings, capacitors and batteries. More recently, under the umbrella of molecular electronics, much of the research has been directed towards the development of stable, processable polymers that exhibit optoelectronic and semiconducting properties with the ultimate goal of developing electronic devices and circuits entirely from polymers, e.g. the electroluminescent devices reported by Friend and coworkers 131. Devices such as MISFETs [4-61 and Schottky barrier diodes [7, 81 have been fabricated from almost all semiconducting polymers. MISFETS generally operate in the accumulation mode [4] with only polyacetylene MISFETs apparently showing any evidence for inversion [SI. Schottky barriers formed from semiconducting polymers have been shown to have good rectification ratios and solar cells made from thiophene oligomers were reported [9] to have a quantum yield close to unity under low illumination (pW/cm'> although this fell to less than 0.15'%1at higher levels of illumination. Clearly, these devices are still far from being competitive with those based on inorganic materials but the results were sufficiently encouraging to promote considerable research effort in an attempt to understand barrier properties. In a recent paper [lo], we reported the results of a systematic investigation into both the DC and AC behaviour of the Schottky barrier formed at the interface between aluminium and electropolymerised poly(3-methylthiophene). Here we summarise the main findings of that work and present new results showing the effect of fabrication conditions on the barrier properties. f-- zyxwvutsrq 0IEE, 1907 IEE Proceclding.s online no. 19971003 Paper received 21st Dccember 1995 H.L. Gomes is with Uniddde de Ciencias Exactas e Humanas, Universidade do Algarve, Campus de Gambelas, 8000 Faro, Portugal D.M. Taylor is with the School of Electronic Enginecring and Computer Systems, University of Wales, Dean Street, B;ingor, Gwynedd LL57 IUT, UK IEE Eroc.-C'ircuit.s Devices Sysl., Vol 144, No. 2, April IY97 metal Fig. 1 2 p-type semiconductor Clus,r.icSchottky harrier to U p-type smziconductor Theoretical The energy band diagram for a classic Schottky diode formed on a p-type semiconductor is shown in Fig. 1. A metal of work function I $ ~contacting the polymer has created a depletion layer at the semiconductor surface [l I]. Hole transport across the interface is controlled by the associated potential barriers, qvy0in the semiconductor and q5/, froin the metal. Based on this 117 zyxwvutsrqpo zyx zyxwvut zyxwvutsrqp zyxwvu zyxwvut classical model, the current-voltage characteristic is described by where J , is the reverse saturation current density, q the electronic charge, k Boltzmann's constant, T the absolute temperature and V the applied voltage. When transport is limited by thermionic emission across the interfacial potential barriers, Jo is given by are themselves independent of frequency. The situation may be further complicated by the presence of traps and interface states [12]. Furthermore, we have shown [ 131 that Mott-Schottky plots for electropolymerised 3methylthiophene display a significant change in slope which was interpreted in terms of a second, much deeper acceptor state which only became active when the band bending was sufficiently great. I where A* is the modified Richardson constant. When transport is limited by diffusion, then Jo is given by (3) where NV is the effective density of states in the valence band, ,U the hole mobility and E,,, the maximum electric field in the depletion region, i.e. Fig.2 Simple equivalent c i m d used here to model the Schottky diode R , and C, represent the depletion region while R, and C, represent the bulk polymer E,,, aaL (Vs0 - V )3 (4) In practice, the forward characteristic of a polymer diode follows the more general behaviour (5) where the ideality factor n is in the range 1.2 to 8. A number of mechanisms have been proposed to explain the departure of n from unity, one of which is the presence of an interfacial layer between the metal and the semiconductor [12]. Other factors include the high series resistance of the bulk polymer between the Schottky barrier and the ohmic gold contact [lo, 121. It is also the case that reverse currents increase more rapidly with applied voltage than suggested by eqns. 2-4. We have suggested that, following Rhoderick [ 121, this may be due to image-force lowering of the interfacial barrier, whence, for negative values of applied voltage V, the reverse current density J R has the form 3 Experimental The polymer films for this study were prepared in a class two semiconductor clean room by electropolymerisation onto a working electrode using a classic threeelectrode cell. The working electrode in this instance was formed by depositing 30nm of gold onto a clean microscope slide previously coated with 3 nm of chrome to improve adhesion. The metals were deposited by evaporation in a turbomolecular pump system. The reference and counter electrodes were made from silver wire and platinum foil, respectively. A 0.5M solution of 3-methylthiophene dissolved in propylene carbonate was prepared, to which was added 0.1 M tetrabutylammonium fluoride as the supporting salt. Electropolymerisation was conducted in two steps: 2.0V was applied across the cell for the first 5s and then 1.6V for 38min depending on the film thickness required. The working electrode was then transferred to a second monomer-free cell and undoped with a voltage in the range -0.1 to +0.2V until the current was less than 0.1pA. The films were then washed with acetone and within 5min placed in the turbosystem for deposition of the aluminium counter electrode. Generally, films were left under vacuum for several hours prior to metal evaporation, although in some cases this time was reduced to as low as 8min to determine what effect, if any, this process step had on diode behaviour. Some devices were also subjected to a post-metal anneal at 90°C under vacuum. After preparation, the devices were mounted in a temperature-controlled sample holder located inside an earthed, steel chamber evacuated to less that torr for 24h prior to taking measurements. The DC characteristics of devices were generally obtained using a PFG model 605 Function Generator and a Keithley model 6 15 electrometer, manually incrementing the voltage firstly in the forward direction (positive voltage to the gold electrode) and then in the reverse direction. Readings were taken after allowing the currents to stabilise for 2min to enable any slow states to equilibrate. For some devices, the characteristics were obtained automatically using a Keithley model 487 picoammeteri voltage source in which the voltage was incremented at zyxwvu zyxwv where N A is the acceptor density, e, and E~ the highfrequency and static relative permittivities, respectively, E, the permittivity of free space and VC = { Vso (kTi 4)).Now In J R is expected to vary as ( V , - Vs0)1'4, behaviour which has been observed experimentally [lo, 131. The capacitance-voltage relation of a Schottky barrier diode is often used to determine the doping density N A in the semiconductor since the capacitance per unit area, C, of the depletion region is given by the MottSchottky relation - (7) where E E ~is the absolute permittivity of the semiconductor. Unfortunately, the relatively low bulk conductivity of semiconducting polymers poses a problem in the accurate measurement of the depletion capacitance C,. As can be seen from the simple equivalent circuit in Fig. 2, the measured admittance Cp in parallel with Rp is strongly frequency-dependent even when the capacitances and resistances of the bulk and depletion regions 118 IEE Proc -Circuits Devices Syst , Vol 144, No 2 April 1997 the faster rate of 10mV/s. These latter measurements were carried out under dry argon. Small-signal admittance measurements over the range lOHz to 3MHz as well as capacitance-voltage (C-V) measurements were carried out with a Hewlett-Packard model 4192A impedance analyser using a test signal of 50mV. When obtaining C-V data, the applied voltage was incremented steadily from the highest reverse bias (usually -2V) to the highest forward bias (+2V) and then the sweep was reversed. 4 negative-going sweeps a negative contribution is made. Furthermore, in forward bias, the depletion region capacitance is expected to increase rapidly with increasing bias leading to increasing values of In and large hysteresis. At high reverse bias, when the depletion capacitance is small, 1, is also small and makes little contribution to the total current, hence little hysteresis occurs. Normally, the two zero-crossing points would be expected to straddle zero applied voltage. As indicated above, that this does not occur is due to the presence of a small offset voltage in the device. Interestingly, the magnitude of this offset voltage was dependent on preparation conditions as can be seen from Fig. 4. The two characteristics presented here were obtained from two devices formed in the same polymer film. Prior to metal evaporation, this film was left under vacuum for 3h by which time the pressure in the chamber had decreased to 6 x 10 torr. With half the film protected by a metal mask, electrodes were then evaporated onto the remainder. Curve ( U ) in Fig. 4 is typical for devices from this group. Subsequently, the vacuum was broken and the mask moved across so that electrodes could be deposited on the remainder of the film. This time, however, evaporation was carried out after 8min with the pressure at 1 x torr. Curve (b) in Fig. 4 is typical of the plots obtained from this second group of devices. Clearly, the time under vacuum prior to aluminium evaporation has had a major influence on device behaviour. zyxwvutsr Results and discussion The electropolymerisation technique developed in this programme resulted in dense, relatively smooth films although scanning electron microscopy did reveal the presence of rounded humps of the order of lOOnm in size projecting slightly out of the film surface [lo]. zyxwvutsrqpon zyxwvutsrqp zyxwvutsrqpon zyxwvutsrqp zyxwvutsrqpon -1oL -2.0 -1.5 I -1 0 -0.5 0 I 0.5 1.0 1.5 20 applied voltage V Fig.3 Dynamic J-V plots obtained for CL diode ,formed by evtrporuting the aluminium electrode soon ujter ph&g in vacuum The measurements were made under dry argon by applying a triangular waveform to the device commencing from OV. The sharp minima on this semilog plot signify a change in the polarity o f t h c current A typical J-V characteristic collected automatically using thie Keithley model 487 system is shown in Fig. 3. The voltage sweep commenced at zero bias and was incremented initially to +2V in the forward direction. For the first three points the current flowed in the opposite sense to the applied bias owing to a small offset voltage associated with the device. Between 0.02 and 0.03V the current reversed and then increased in the direction consistent with a positive applied bias. Upon decrementing the voltage from +2V, the currents for particular values of forward voltage were generally smaller and current reversal was observed between 0.18 and 0.17V. A similar pattern was seen for reverse bias, i.e. currents were higher when the voltage was increased. Upon reducing the voltage from -2V to zero, the current regained its intitial value. Identical curves were obtained if the voltage cycle was repeated. Two interesting features of these curves are a the shift in the voltage at which the current reverses (these zero-crossing points are clearly identified as the sharp minima in the plots) h the general offset in the curves towards positive bias. The shifts in the zero-crossing points are readily explained by considering the contribution of the displacement current ID (= C U'Vldt) to the total measured current. For positive-going voltage sweeps Z , makes a positive contribution to the measured current, while for -101 I -12 -2.0 -1.5 -1.0 -0.5 0 0.5 1.0 1.5 20 applied voltage, V Fig.4 Dynamic J-V plots obtuined as in Fif 3 .:bowing the eflect o j time under vacuum prior to evaporation o j the a uminium electrode Thc devices were formed after (a) 3 hours and (b) X mins under vacuum For clarity curve (h) has been shifted vertically upwards by multiplying the currents by a factor of 250 First, the zero-crossing points are well-separated in curve (b) indicative of a much larger depletion capacitance. This is consistent with our previous finding that atmospheric oxygen is a reversible dopant [lo] in poly(3-methylthiophene). After 3 h under vacuum, residual oxygen would be removed from the film thus reducing the doping density and hence the depletion capacitance. This is confirmed by the much smaller separation of the zero-crossing points displayed by curve (a). Secondly, it is seen that the shorter the time under vacuum prior to metallisation the larger the initial offset voltage. Under these conditions it is likely that, when the chamber is insufficiently evacuated prior to metallisation, an interfacial layer forms between the zyxwvutsr zyxwvutsr zyxwvutsrqp zyxwvutsrqp IEE Proc.-Circuits Devices Syst., Vol. 144, No. 2,April 1997 119 deposited metal and the polymer. Polarisation within such a layer would then be a possible cause of the offset voltage. Fig. 5 shows the J-V characteristics of three diodes obtained under vacuum by iiicrementiiig the voltage manually at the much slower rate of 50mV every 2min. Now, the effect of the displaccmcnt current has been eliminated, as expected. Interestingly, the slow scan coupled with placing devices under vacuum appears to have removed the offset voltage. Good rectification ratios were observed in all devices left under vacuum for long periods prior to deposition of the aluminium electrode. The highest ratio (-50 000 at 2V) was shown by the annealed device DLl.As discussed in our previous publication [lo], this is because of an order of magnitude reduction in the reverse currents, probably arising from the formation of a thin insulating layer bctwcen the metal and polymer during the annealing process. case (device F) a reversal of the rectification occurred. This behaviour is reminiscent of that observed in a sibcon MIS device with a thin insulating layer between the metal and semiconductor. According to Green and Schewchun [14], the plateau signifies a transition from inversion to depletion to accumulation at the silicon surface upon increasing the forward voltage, the plateau arising from the absorption of the applied voltage by the depletion region It is generally assumed that at high forward bias device CLirrents are limited by the bulk coiiductivity of the polymer. By fitting numerically generated capacitance and loss curves based on the equivalent circuit in Fig. 2 to experimental data such as that in Fig. 7 the bulk resistance per unit area of the device R , (expressed in Qcni’), may be estitnated. The values so obtained for four of the devices tested here are given in Table 1 together with estimates of bulk resistance based on the current flowing at an applied forward bias of 2V. It is clear that the latter values are several orders of magnitude greater than the former zyxwvutsrq zyxwvu zyxwvutsrqponm zyxwvutsr zyxwvutsrqponm zyxwvutsr zyxwvutsrqponml zyxwvutsr -12 -3 , -2 -1 0 applied voltage, Fig. 5 under - I - 1 - 2 1 I 3 V Steadystate J- V chamcteristics of deu’ce.c il, L, ciiid D,,ohiniiirti wuiurii. In all cases, the pol)iiici- was placed under v x u u i i i for sexera1 hours prior to deposition of thc aluminiitin electrode. Onc of the del ices. D(,. ~ a subjected $ to a post-metal anncal a t 90°C % device A + devicc L, 0 device D, 100 101 102 io3 frequency, 104 105 io6 Hz zyxwv zyxwv Fig.7 Fiequeticy depenLLnce.c or cuppncilance and loss jbr device C The continuous curvcs arc the best tit based on the equivalent circuit in Fig. 2 capacivance 5 loss Table 1: Comparison of bulk resistance values obtained from DC and AC measurements -1 2 -2.0 -1.5 Qcm2 RbAc,Qcm2 Thickness, p m D 0.13 190 10 P 0.18 200 5 E 0.20 530 5 C 0.80 27000 36 In addition t o being of different thicknesses, the samples w e r e u n d o p e d t o different extents -1.0 -0.5 0 0 5 1.0 1.0 20 applied voltage, V Fig. 6 Stcady-stutc J- V clzuractcvistics obtuiiwd under vaciunz f b v devices F, L , and N that M J ~ E undeer vacuum for o n l j a short time prior to the deposition of the aluminium electrode x devicc I; 0 device L , + dcvice N When Schottky barriers are formed after about 10 min under vacuum, the resulting characteristics obtained manually (see Fig. 6) are different from those in Fig. 5. Although no offset voltage can be seen, a pronounced plateau appears in forward bias and in one 120 R,,,, Device Possible reasons for this discrepancy include current limitation at the gold electrode or the presence of an interfacial layer between the aluminium and the polymer which limits direct current flow but not an alternating current. It has already been suggested that an interfacial layer may be responsible for the high ideality factors of polymer Schottky diodes. Indeed, when such a layer is present Rhoderick [12] shows that eqn. 5 applies but with y1 given by IEE Proc.-Circuits Devices Syst,, Vol. 144, No. 2,April 1997 zyxwvutsrqpo zyxwvutsrqp zyxwvutsrq zyxwvutsrqpo zyxwvutsrq zyxwvutsrqponmlkj where D , is the density of surface stakes, Dss the density of' occupicd surface states and (9) where b and E, are, respectively, the thickness and absolute permittivity of the Interfacial layer. Thus we sec that the ideality factor would now depend strongly on the presence of interface states and would be expected to vary with both bias and temperature, as indeed is observed experimentally [lo]. Furthermore, the presence of the layer can lead to inversion at the semiconductor surface as discussed above. If 6 is large enough even a reversal of the rectification can occur. It is interesting in this contcxt to note that Garnier and coworkers [ 1 S] reported a reversal in the rectification of Ag/athienyl oligothiophene diodes after thermal annealing in air at 1SO"C. They suggested that this was caused by the transformation of the polymer from p-type to ntype. The results are more readily explained, howcver, by the formation of an insulating layer between polymer and electrode during the anneal process. The interfacial layer will also control the reverse current density JI, in the device so that [12] Thus, for reverse voltages the device current should increase exponentially with bias so long as Dsn is a slowly varying function of bias. This is indeed seen to be the case over most of the reverse bias range in Figs. 5 and 6. It should be noted though, that for an interfacial layer less than I n m thick to have any observable effect on the characteristic, D, would have to be of the order of l o t 7m 2V I, devices. They attributed such shifts to the field-induced drift of dopant ions in the device. The consequence of such an effect within the polymer, however, would be to 'flatten' the C-V curve not to shift it. This is so because in reverse bias, negatively charged acceptors would drift into the depletion region thereby maintaining a higher capacitance than would otherwisc be ineasured. In forward bias, the dopant ions would drift away from the depletion region, thus reducing its capacitance below that which would be expected. The translation observed by these workers was probably related to the drift of impurity ions within the insulating oxide layer. The small degree of hysteresis observed in Fig. 8 is likely to arise from the charging and discharging of interface states or traps in the depletion region. x1013 zyxwvuts -1.5 -1.0 -0.5 0 0.5 10 applied voltage, V Fig.9 Mott-Schottky plots ohtuitzed ut d@wnl fiequencic.c Tlic systematic variation is caused by the lowfrequency dispersioii evidciit in Fig. 7. Thc general shapc of thc plot is retained and shows two distinct regimes, t h c transition from one to the othcr occurring ncar zero bias I inonHz x SOOHz zyxwvutsrqpon zyxwvutsrqp zyxwvutsrq + 200147 3 2013L E W 83 0 " ____ 0 05 10 applied voltage, V (~'u/~"cilioizc~,-.-lIi~~~(i~e plot,r ijht(~ii~ed under vm1cuum -20 Fig.8 -1.5 -1.0 -0 5 1.5 20 The voltagc was sciinncd From -2V Lo -1 2 V at IOmVis and 5OinV/s The mcasuremcnt frcqueiicy was IOOHz. The sinall degree of hysterehis Teen in forward biar is only prcsent when the voll~igeis ramped a b m e about I V 0 SO mV/s I'oI1lv:s Fig. 8 shows C ~ - Vcurves obtained at lOOHz for two different voltage sweep ratcs. The capacitance increases from a low value in reverse bias to a high value in forward bias, consistent with the presence of a depletion layer in the device. A small degree of hysteresis was observed when the forward voltage exceeded --I V but this decreased with dccreasing swecp rate. We did not observe the large reversible translation of the curves along the voltage axis that was reported by Ziemelis et al. [ 161 for Si/Si02/poly(3-hexylthienylenej!Au MIS IE,5 Proc.-C'irmir.s Dcvir.c.s S'),,\r., Cid 144. IVO. 2, A p d 1997 The effect of the dispersion in Fig. 7 on the C-V plots is to cause a general increase in capacitance with decreasing frequency. According to the equivalent circuit in Fig. 2, so long as RI, << RI, the low-frequency capacitaiice should asymptote to the depletion capacitance. The lack of agreement between theoretical curves and experimental data obtained at low frequencies, suggests that the results may be influenced by interface states and/or trapping effects. The effect is to introduce a frequency-dependence in the MottSchottky plots (Fig. 9) but without changing significantly the nature of the C-V dependence. For all devices, the Mott--Schottky plots decreased in two stages as the bias was swept from reverse to forward with a change of slope occurring near zero bias. In an effort to minimise the effects of both the main dispersion and the low frequency dispersion only the data obtained at a frequency in the range IO-IOOHz, i.e. some two orders of magnitude below the main relaxation frequency, were used to estimate doping densities in the polymer [lo]. According to eqn. 7 a semiconductor uniformly doped with a single acceptor species should give a linear Molt-Schottky plot. Interestingly, for our devices such plots yield two almost linear regions. Whilst such behaviour may be indicative of a lower doping density near the interface, we have dismissed this as the likely 121 cause because similar results were obtained for a wide range of diodes varying in both thickness and doping density. We have argued previously [lo] that the observed dependence was caused by the presence of a second dopant species in the polymer which provided a set of acceptor states deep in the forbidden gap. Such states only become active when band-bending near the polymer surface is sufficient for their energies to fall below the Fermi level. Based on this analysis, it was shown that the density of the shallow acceptor ranged from (1 to 20) x 1023m-3while that for the the deeper state was about an order of magnitude higher ranging from (9 to 200) x 1023m-3.Both were sensitive to the degree of undoping. A third possible explanation is that the presence of an interfacial layer is affecting the capacitance of the Schottky barrier: not directly (it is a series capacitance of high value) but because it results in only a fraction of the applied voltage appearing across the depletion region. In this case, eqn. 7 must be replaced by For diodes prepared after several hours under vacuum, no such plateau was seen. However, when one of these devices, D, in Fig. 5 , was annealed at high temperature, the improved rectification ratio arose from a decrease in the reverse current, again suggesting that an interfacial layer had formed during the anneal. Indeed, close inspection of the forward current did reveal the presence of a small plateau [lo] which is consistent with this view. None of our diodes yielded linear Mott-Schottky plots. Rather, two regimes were identified which we have attributed previously [IO] to the presence of two sets of acceptor states, one shallow and one deep. The present work shows that for some devices, at least, the existence of a thin interfacial layer between the metal and the polymer may control device behaviour. Further detailed work is now necessary to distinguish the various contributions. zyxwvutsrq zyx zyxwvutsrqpo zyxwvut zyxwvutsr zyxwvutsrqp zy zyx zyxwvutsr zyxwvutsrqp 6 Acknowledgments The authors are grateful to JNICT (Portugal) for their continued support of this work. where C is the effective capacitance of the barrier after correcting for the main frequency dispersion, C, is the capacitance of the interfacial layer and a = qsD,,l~, where D,, is the concentration of interface states in equilibrium with the semiconductor [12]. Thus, if C, >> C, a linear plot is again expected but with a modified slope. Once again though, an interfacial layer l n m in thickness will only affect the estimate of doping density when D,,is of the order of 10’7m-2V-1.The model does suggest though that for a device in which a >> 1 an increased slope can be expected in forward bias when C, >> C,. Since C, will generally exceed C,, a change of slope similar to that observed in Fig. 9 could also occur if the Fermi level at the semiconductor surface passed through and activated a second set of interface states. A final possibility is that the presence of the interfacial layer allowed the polymer surface to go into weak inversion. This would certainly be consistent with the anomalous I-V plots in Fig. 6 but not with the more normal plots such as those in Fig. 5. 5 Conclusions This work shows, not surprisingly, that fabrication conditions play a major role in determining the electrical behaviour of Schottky barrier diodes formed from semiconducting polymers. When polymers are left for only short times under vacuum prior to evaporating the rectifying contact then the dynamic I-V characteristics show clear evidence for a higher doping density in the device, probably owing to the presence of absorbed atmospheric oxygen. We have also argued that the larger offset voltage seen in such measurements arises from polarisation in an interfacial layer created during electrode deposition. Steady state measurements on these devices yield anomalous I-V characteristics, the forward currents displaying a plateau similar to that observed in MIS diodes, thus providing further evidence for the presence of an insulating interfacial layer in these diodes. 122 7 References 1 ITO, T., SHIRIKAWA, H., and IKEDA, S.: ‘Simultaneous polymerization and formation of polyacetylene film on the surface of concentrated soluble Ziegler-type catalyst solution’, J. Po/ynz. Sei. Polym. Chem. Ed., 1974, 12, pp. 11-12 2 CHANG, C.K., FINCHER, C.R., PARK, Y.W., HEEGER, A.J.; SHIRIKAWA, H., LOUIS, E.J., GAU, S.C., and MACDIARMID, A.G.: ‘Electrical conductivity in doped polyacetylene’, Plzys. Rev. Lett., 1977, 39, pp. 1098-1101 3 BURROUGHES, J.H., BRADLEY, D.D.C., BROWN, A.R., MARKS, R.N., MACKAY, K., FRIEND, R.H., BURNS, P.L., and HOLMES, A.B.: ‘Light emitting diodes based on conjugated polymers’, Nature, 1990, 347, pp. 539-541 4 TAYLOR, D.M., GOMES, H.L., UNDERHILL, A.E., EDGE, S., and CLEMENSON, P.I.: ‘The effect of oxygen on the electrical characteristics of field effect transistors formed from electrochemically deposited films of poly(3-methylthiophene)’, J. Phys. D, Appl. Phys., 1991, 24, pp. 2032-2038 5 BURROUGHES, J.H., JONES, C A . , and FRIEND, R.H.: ‘New semiconductor device physics in polymer diodes and transistors‘. Nature, 1988, 335, pp. 137-141 6 ASSADI,, A., SVENSSON, C., WILLANDER, M., and INGANAS, 0.: ‘Field-effect mobility of poly(3-hexylthiophene)’, Appl. Plzys. Lett., 1988, 53, pp. 195-197 7 GOMES. H.L., and TAYLOR, D.M.: ‘Frequency-dependent admittances of Schottky barriers at the aluminium-poly(3-methylthiophene) interface’, Mol. Cvyst. Liq. Cryst., 1993, 236, pp. 151156 8 KANICKI, J.: ” in SKOTHEIM, T.A. (Ed.): ‘Handbook of conducting polymers’ (Marcel Dekker, New York, 1986) 9 GLENIS, S., TOURILLON, G., and GARNIER, F.: ‘Influence of the doping on the photovoltaic properties of thin films of poly3-methvlthioohene’. Thin Solid Films. 1986. 139. DD. 221-231 10 TAYLOR, D.M., and GOMES, H.L.: ‘Electricaf characterisation of the rectifying contact between aluminium and electrodeposited poly(?-methylthiophene)’, J. Phys. D,Appl. Phys., 1995, 28, pp. 25562568 11 SZE, S.M.: ‘Physics of semiconductor devices’ (Wiley, New York, 1981) 12 RHODERICK, E.H.: ‘Metal semiconductor devices’ (Clarendon Press, Oxford, 1978) 13 GOMES, H.L., TAYLOR, D.M., and UNDERHILL, A.E.: ‘Charge transport in poly(3-methylthiophene) Schottky barrier diodes’, Synfh. Met., 1993, 55-57, pp. 4076-4081 14 GREEN. M.A.. and SHEWCHUN. J . : ‘Caoacitance oronzrries of MIS tunnel diodes’, J Appl Phys, 1975, 46, pp 5185-5i90 15 GARNIER, F , HOROWITZ, G I and FICHOV, D ‘Conlugated polymers and oligomers as active material for electronic devices’, Synth. Met., 1989, 28, pp. C705SC714 16 ZIEMILIS, K.E., HUSSAIN, A.T., BRADLEY, D.D.C., and FRIEND, R.H.: ‘Electro-opticalproperties of polymeric semiconductor-devices constructed from poly(3-hexylthienylene)’, Synth. Met., 1991, 41, pp. 1045-1050 zyx I E E Proc.-Circuits Devices Syst., Vol. 144, No 2,April 1997