Introduction

There are at least four ways in which a new, exploratory technology such as OTFTs can compete with or supplement a widely used, entrenched technology such as (a-Si:H) TFTs, for which many billions of dollars have already been invested:

1. By far surpassing the performance of the entrenched technology and offering a substantial performance advantage.
2. By enabling an application that is not achievable using the entrenched technology, taking advantage of one or more unique properties or processing characteristics of OTFTs. An example for Case 2 could be a flexible AMFPD fabricated on a plastic substrate. Because of the high processing temperature used in a-Si:H deposition (approximately 360°C), which is required for the fabrication of well-performing a-Si:H TFTs, it is not possible to fabricate an AMLCD based on such TFTs on a transparent plastic substrate. OTFTs, which can be processed at or close to room temperature and thus are compatible with transparent plastics, are an enabling technology which complements the entrenched technology instead of competing with it.
3. By significantly reducing the cost of manufacturing OTFTs as compared to mainstream TFTs while delivering similar performance.
4. By leveraging a potential reduced cost advantage to create a new way of using an existing application, or to change the usage pattern or user habit for an existing application, even if performance is lower than that of the entrenched technology. An example for Case 4 could be a large-area AMFPD which uses a backplane comprising OTFTs that have been fabricated using very low-cost processes compared to a-Si:H TFTs. Because of its significantly reduced cost, such a display could have a substantially reduced lifetime compared to a conventional AMFPD, since the user would be able to replace it several times over a period equal to the lifetime of a more expensive, conventional AMFPD.

Progress in performance of OTFTs from 1986 to the present

Figure 1

Table 1   Highest field-effect mobility (µ) values measured from OTFTs as reported in the literature annually from 1986 through 2000.
Year	Mobility (cm2 V–1 s–1)	Material (deposition method) (v) = vacuum deposition (s) = from solution	Ion/Ioff*	W/L	Reference
1983	Minimal, not   reported (NR)	Polyacetylene (s)   (demonstration of field effect   in an OTFT)	NR	200	[16]
1986	10–5	Polythiophene (s)	103	NR	[17]
1988	10–4	Polyacetylene (s)	105	750	[18]
	10–3	Phthalocyanine (v)	NR	3	[19]
	10–4	Poly(3-hexylthiophene) (s)	NR	NR	[20]
1989	10–3	Poly(3-alkylthiophene) (s)	NR	NR	[21]
	10–3	–-hexathiophene (v)	NR	NR	[22]
1992	0.027	–-hexathiophene (v)	NR	100	[23]
	2 × 10–3	Pentacene (v)	NR	NR	ibid.
1993	0.05	–-di-hexyl-hexathiophene (v)	NR	100–200	[24]
	0.22†	Polythienylenevinylene (s)	NR	1000	[25]
1994	0.06	–-dihexyl-hexathiophene (v)	NR	50	[26]
1995	0.03	–-hexathiophene (v)	>106	21	[27]
	0.038	Pentacene (v)	140	1000	[28]
	0.3	C60 (v)	NR	25	[29]
1996	0.02	Phthalocyanine (v)	2 × 105	NR	[30]
	0.045	Poly(3-hexylthiophene) (s)	340	20.8	[31]
	0.13	–-dihexyl-hexathiophene (v)	>104	7.3	[15]
	0.62	Pentacene (v)	108	11	[32]
1997	1.5	Pentacene (v)	108	2.5	[33]
	0.05	Bis(dithienothiophene) (v)	108	500	[34]
1998	0.1	Poly(3-hexylthiophene) (s)	>106	20	[35]
	0.23	–-dihexyl-quaterthiophene (v)	NR	1.5	[36]
	0.15	Dihexyl-anthradithiophene	NR	1.5	[37]
2000	0.1	n-decapentafluoroheptyl-methylnaphthalene   -1,4,5,8-tetracarboxylic diimide (v)	105	1.5	[38]
	0.1	–-dihexyl-quinquethiophene (s)	NR	NR	[38]
*Values for Ion/Ioff correspond to different gate voltage ranges and thus are not readily comparable to one another. The reader is encouraged to read the details of the experiments in the cited references. †This result has not yet been reproduced.

Table 1 lists the highest field-effect mobility (µ) values measured from OTFTs as reported in the literature, annually from 1986 through 2000 and separately for each of the most promising organic semiconductors. For a specific organic semiconductor that already has an entry in Table 1 and Figure 1 for a previous year, a new mobility value is entered only if it is higher than the value of the preceding entry. We can observe an impressive increase in mobility, which was achieved either by improving the processes used for the fabrication of the transistors or by synthesizing new organic materials. A typical path to performance increase could be described as a two-stage process: 1) A new organic semiconductor is synthesized or used for the first time as the active layer in an OTFT. 2) The film deposition parameters for the semiconducting organic layer are optimized to obtain the most advantageous structural and morphological characteristics for improved performance until no more improvement seems possible. After that point, another incremental improvement in mobility usually is obtained from the synthesis and/or first OTFT application of a new organic semiconductor. This iterative procedure can be observed in Figure 1 by following the bars of the same color. Today, we have reached an important point in the plot of performance versus time. The most widely used organic semiconductors, such as pentacene, thiophene oligomers, and regioregular polythiophene, seem to have reached “maturity” as far as their performance is concerned. Their individual performance-versus-time curves have saturated (i.e., when a new, higher value is not reported in the years following a bar of a specific color, it means that there was no improvement in mobility during those years). In the past, each time such a performance saturation occurred (Figure 1), a new material was introduced whose performance broke the temporarily established upper limit in performance.

However, one can argue that today's maximum mobility at room temperature, which could be the field-effect mobility value of 2.7 cm² V^–1 s^–1 reported for holes in pentacene single crystals by Schön et al. [39], has approached a fundamental limit, at least as far as the classes of semiconducting organic materials known today are concerned. In the following section we discuss the potential existence of such a limit in conjunction with the various conduction mechanisms believed to be in operation in various organic semiconductors.

Conduction mechanisms and a potential fundamental limit to charge-carrier mobility in organic semiconductors

The boundary between band transport and hopping is defined by materials having mobilities between 0.1 and 1 cm² V^–1 s^–1 [13, 40, 41]. Highly ordered organic semiconductors, such as several members of the acene series including anthracene and pentacene, have room-temperature mobilities in this intermediate range, and in some cases temperature-independent mobility has been observed [40], even in polycrystalline thin films of pentacene [41]. That observation was used to argue that a simple temperature-activated hopping mechanism can be excluded as a transport mechanism in high-quality thin films of pentacene [41]. At low temperatures (below approximately 250 K), band transport becomes the mechanism that takes control of carrier transport in single crystals of pentacene and other acenes. Very high mobility values (from 400 cm² V^–1 s^–1 [42] to more than 1000 cm² V^–1 s^–1 [39]) have been reported. At these temperatures, the vibrational energy is much lower than the intermolecular bonding energy and phonon scattering is very low; thus, high mobility is exhibited. At or close to room temperature, phonon scattering becomes so high that the contribution of the band mechanism to transport becomes too small. At these same temperatures, hopping begins to contribute to carrier transport. Hopping of carriers from site to site becomes easier as the temperature rises. The combination of these two mechanisms explains the fact that the mobility decreases as temperature rises from a few degrees K to about 250 K, and after that the mobility begins to rise slowly [43]. This behavior is difficult or impossible to observe in polycrystalline films, in which traps attributed to structural defects dominate transport. Interestingly, a recent work [44] reports thermally activated transport for single crystals of pentacene, quaterthiophene, and hexathiophene isolated from polycrystalline films, which was attributed to Coulomb blockade-dominated transport. However, in the same work it was predicted that for materials with reduced tunnel resistance between the molecules, the Coulomb blockade model breaks down; as a result, mobilities greater than 0.5 cm² V^–1 s^–1 and exhibiting temperature-independent behavior should be attainable. This is the case for the pentacene sample in Reference [41], which exhibited a mobility of 1.5 cm² V^–1 s^–1 and had a temperature-independent mobility, in contrast to the sample having lower mobility (µ = 0.3 cm² V^–1 s^–1) and exhibiting thermally activated transport, a behavior that was attributed to thermally activated hopping conduction. There may be a reasonable explanation for the fact that the apparent mobility of the pentacene single crystal measured in Reference [44] is lower than the mobility of some polycrystalline films. It is possible that a trap-infested region is formed in the pentacene adjacent to the bottom electrodes, in the experiment of Reference [44]. We have shown that when pentacene is deposited on Au, as is the case in Reference [44], a microcrystalline, defect-infested region is formed in the channel, at and close to the edge of the gold electrode.¹ As a result of the existence of this region, there is a large concentration of traps close to the Au contact which dominates the carrier transport and drastically lowers the apparent mobility. These traps are also responsible for the gate-voltage dependence of the mobility of pentacene, as is also described below in more detail [45]. It is worth noting that the devices in Reference [41] had their source and drain electrodes deposited on top of the pentacene layer through a mask.

We can propose two possible ways of eliminating the potential fundamental limit at about 10 cm² V^–1 s^–1 for the mobility of OTFTs, imposed by the weak intermolecular forces acting among nearest-neighbor (nn) molecules. One is to strengthen such interaction. This can be done by creating a stronger bond between nn molecules. However, this must take place without breaking the conjugation of the molecules. Stronger intermolecular bonds would result in stiffer crystalline structures, and thus it would take temperatures higher than room temperature to generate substantial scattering of highly delocalized carriers by lattice vibrations. Using such a strategy, one could, in effect, produce at room temperature the high mobility that exists at very low temperatures in crystals of the acene series (e.g., pentacene). A second way would require more drastic change in the conduction path and mechanism. It would involve carrier transport along a single macromolecule that would bridge the gap between the source and drain electrodes of a TFT. Intermolecular conduction would give way to intramolecular conduction. It is well known that the mobility along the long axis of conjugated conducting and semiconducting polymers (e.g., polyacetylene) can reach values of 1000 cm² V^–1 s^–1 or more. This may require a drastic reduction in the size of the TFT channel from micron-size channels containing a large number of molecules to nano-size channels that are transcended by a single molecule. In the former case, conduction depends upon intermolecular transport, while in the latter, conduction takes place through the conjugated backbone of a single molecule and thus is intramolecular. The successful execution of at least one of the above strategies would prove that although limits can be imposed by the design and size of OTFT devices, they are not fundamental.

Modeling of the electrical characteristics of OTFTs

Figure 2

Figure 3(a) shows a typical plot of drain current I_D versus drain voltage V_D at various gate voltages V_G, which corresponds to a device using DH6T as the semiconductor, 3700 Å vapor-deposited parylene-C as the gate insulator, aluminum gate, and gold source and drain electrodes. When the gate electrode is biased negatively with respect to the grounded source electrode, DH6T insulated-gate field-effect transistors (IGFETs) operate in the accumulation mode and the accumulated charges are holes. At low V_D, I_D increases linearly with V_D (linear regime) and is approximately determined from the following equation:

ID =	WCi	µ		VG – VT –	VD		VD,	(1)
	L				2

where L is the channel length, W is the channel width, C_i is the capacitance per unit area of the insulating layer, V_T is the threshold voltage, and µ is the field-effect mobility. The latter can be calculated in the linear regime from the transconductance,

gm =		ID		VD=const	=	WCi	µVD,	(2)
		VG				L

by plotting I_D versus V_G at a constant low V_D and equating the value of the slope of this plot to g_m. Figure 3(c), which corresponds to Figure 3(a), shows such a plot, and the calculated mobility value is 0.122 cm² V^–1 s^–1 at V_D = –2 V. The value of V_D is chosen so that it lies in the linear part of the I_D versus V_D curve. For this device, L was equal to 137 µm and W was equal to 1 mm. Other devices from the same substrate produced field-effect mobilities ranging from 0.095 to 0.131 cm² V^–1 s^–1.

Figure 3

When the gate electrode is biased positively, DH6T IGFETs operate in the depletion mode, and the channel region is depleted of carriers. The current modulation (the ratio of the current in the accumulation mode over the current in the depletion mode, also referred to as I_on/I_off) for the device of Figure 3 is slightly above 10⁴ when V_G is scanned from –20 to +6 V [Figure 3(b)]. For V_D more negative than V_G, I_D tends to saturate (saturation regime) owing to the pinch-off of the accumulation layer, and is modeled by the equation dimit5.gif

ID =	WCi	µ(VG – VT)2.	(3)
	2L

In the saturation regime, µ can be calculated from the slope of the plot of |I_D|^1/2 versus V_G. For the same device as in Figure 3, the mobility calculated in the saturation regime was 0.09 cm² V^–1 s^–1.

Pentacene also exhibits typical p-type semiconductor characteristics. Figure 4 shows a graph that contains an I_D versus V_G plot and an |I_D|^1/2 versus V_G plot [45]. It corresponds to a device with channel length L = 4.4 µm and width W = 1500 µm, and utilizing pentacene as the semiconductor, 0.5-µm-thick thermally grown SiO₂ as the gate insulator, heavily doped Si (n-type) as the gate electrode, and gold source and drain electrodes. The field-effect mobility µ was 0.16 cm² V^–1 s^–1, while the threshold voltage V_T was about –30 V. The I_on/I_off ratio was above 10⁷ when V_G was scanned from –100 to +80 V. The measured mobility value is in agreement with reported hole mobilities from IGFETs based on pentacene films grown at room temperature [49, 50]. The mobility from devices with lower W/L ratios was considerably higher and reached 0.25 cm² V^–1 s^–1 for devices with W/L = 3.5 owing to fringe currents outside the channel. It is important that W/L be close to 10 or higher in order to minimize the effects of such currents; otherwise, the mobility is overestimated. An alternative way to achieve this would be to pattern the semiconductor so that its width does not exceed the width of the channel.

Figure 4

Differences can often be observed in mobility values calculated in the linear region and the saturation region. The linear region mobility can be affected by contact problems, and in such cases there are departures from the linearity of the I_D versus V_D curve which can lead to underestimation of mobility. In the saturation regime, when channel lengths are comparable to the gate insulator thickness or only a few times greater than that thickness, the I_D versus V_D curves do not saturate and exhibit an upward trend at high V_D. Calculating the mobility in the saturation region from such devices can lead to erroneously high values.

Vacuum-deposited organic semiconductor films

Figures 5(a) and 5(b) respectively show the dependence of field-effect mobility µ on the charge per unit area on the semiconductor side of the insulator Q_S, and the gate field E. The solid circles correspond to a pentacene-based device with a 0.12-µm-thick SiO₂ gate insulator thermally grown on the surface of a heavily doped n-type Si wafer that acted as the gate electrode. The open circles correspond to a similar device with a 0.5-µm-thick SiO₂ gate insulator. The mobility for the SiO₂-based devices is calculated in the saturation regime using a gate sweep, as explained in Figure 4, and is then plotted versus the maximum V_G used in each gate sweep. The maximum V_G is varied from –20 to –100 V. During all sweeps V_D is kept constant at –100 V in order to eliminate any effects that source and drain contact imperfections might have on our results. The mobility increases linearly with increasing Q_S and E and eventually saturates [Figures 5(a) and 5(b)]. Q_S is a function of the concentration of accumulated carriers in the channel region (N). Since the accumulation region has been shown in the past to be two-dimensional and confined very closely to the interface of the insulator with the organic semiconductor [52], all of this charge is expected to be localized within the first few semiconductor monolayers from this interface. An increase in V_G results in an increase in E and N. However, for the same V_G, N depends on both the dielectric constant and the thickness of the gate insulator, while E depends only on its thickness. By replacing SiO₂ with an insulator having a similar thickness but a much higher dielectric constant, we facilitated charge accumulation. An accumulated carrier concentration similar to the SiO₂ case could be attained at much lower V_G, and hence E, with all of the other parameters being similar. If mobility depends on N rather than E, a high mobility should be achieved in the devices comprising the high-dielectric-constant gate insulator at much lower V_G, and hence E, than TFTs using a comparable thickness of SiO₂. Indeed, this is what we observe in Figure 5(b). The squares in Figure 5 correspond to devices comprising room-temperature-sputtered BZT as the gate insulator [45]. The solid squares are generated by gate sweeps, while the open squares are generated by drain sweeps. From Figure 5(b) it is obvious that the applied gate field used in BZT-based devices to obtain mobility values similar to those of the SiO₂-based devices was about five times lower than the fields used in the latter devices. This clearly proves that high field is not required to obtain high mobility. Thus, the gate-voltage dependence of mobility in these devices is due to the higher concentration of holes accumulated in the channel, which was achieved with the use of insulators having a higher dielectric constant. Figure 5(a), which plots µ versus charge per unit area, Q_S, shows exactly that. The values of Q_S and N required to reach a certain mobility value are practically the same for SiO₂- and BZT-based devices, although much different gate field values were required to obtain such a mobility in each case.

Figure 5

To prove the compatibility of our device-fabrication processes with transparent plastics, we have reported the successful fabrication of pentacene-based TFTs on thin, flexible polycarbonate substrates [45]. The BZT gate insulator used was 0.128 µm thick. The performance of these TFTs was similar to that of devices fabricated on quartz or SiO₂/Si substrates. Figure 6(a) shows the characteristics of such a device (W = 1500 µm, L = 69.2 µm) [45]. Mobility was 0.2 cm² V^–1 s^–1 as calculated in the saturation region [45]. Mobility values as high as 0.38 cm² V^–1 s^–1 were measured from devices with a W/L ratio of 4. These are the highest reported mobilities from devices fabricated on transparent plastic substrates and operating at a maximum voltage of only 4 V. Figures 6(b) and 6(c) show the actual OTFT devices on the polycarbonate substrate.

Figure 6

In organic semiconductors, both the properties of the individual molecules and the structural order of the molecules in the film determine the macroscopic properties of the material. These properties can be controlled by using molecular engineering to synthesize molecules with optimal characteristics and by controlling the conditions under which these molecules assemble to form the solid. In the case of chain- or rod-like molecules, which have one molecular axis much longer than the other two, large -conjugation length along the long axis and close molecular packing of the molecules along at least one of the short molecular axes are two important conditions for high carrier mobility. Figure 7 contains proof of the above claims. By growing amorphous films of pentacene, which is achieved by keeping the substrate temperature low during deposition, we make a film that is practically insulating. When the substrate temperature is kept at room temperature during deposition, a very well-ordered film is deposited, and the mobility measured at room temperature is very high for an organic semiconductor (0.6 cm² V^–1 s^–1). Since the structure of this thin film is different from the structure of single crystals of pentacene, we must distinguish between a “thin-film phase” and a “single-crystal phase” of pentacene. When a mixture of the thin-film phase and the single-crystal phase is grown [28], the mobility is very low, possibly because of the high defect concentration resulting from the coexistence of the two phases.

Figure 7

Pentacene transistor drain–source contacts can be made in one of two configurations [Figures 2(a) and 2(b)]: top contact and bottom contact. The performance of pentacene devices with the bottom-contact configuration is inferior to that of devices with the top-contact configuration. Consequently, most high-performance pentacene TFTs reported in the literature have the top-contact configuration, and shadow masking is generally used to pattern the source and drain contacts on top of the pentacene. This is a process that cannot be used in manufacturing. A process that allows the photolithographic patterning of the source and drain electrodes on the insulator before the deposition of pentacene, according to the schematic shown in Figure 2(b), had to be developed. Furthermore, the performance of devices fabricated with such a process should be similar to or better than that of top-contact devices [Figure 2(a)]. Figure 8 shows a pentacene layer as it was grown on SiO₂ and a Au electrode. The edge between the SiO₂ and the Au in the middle photograph is marked by the end of the white (Au) area (due to variations in image contrast, the pentacene-covered Au appears different in the top two pictures). On SiO₂, far away from the Au edge, pentacene consists of fairly large grains (having sizes between 0.2 and 0.5 µm). On Au the grain size falls dramatically. Close to the Au edge but on the SiO₂ side, there is a transition region where the grain size increases with increasing distance from the edge.

Figure 8

In an organic TFT, the structure and contact behavior of the film formed on top of most of the electrode is not important for the performance of the device in the channel. It is the crystalline structure of pentacene at the electrode edge which causes the performance limitation of the bottom-contact TFT. Right at the edge of the Au electrode, there is an area with very small crystals and hence a large number of grain boundaries. Grain boundaries contain many morphological defects, which in turn are linked to the creation of charge-carrier traps with levels lying in the bandgap. These defects can be considered responsible for the reduced performance of bottom-contact pentacene TFTs. Their elimination should result in bottom-contact devices with performance similar to or better than that of top-contact devices. In a typical bottom-contact pentacene TFT, the mobility is equal to or less than 0.16 cm² V^–1 s^–1. We have used a self-assembled monolayer (SAM) of 1-hexadecane thiol to modify the surface energy of the Au electrode in an effort to improve the crystal size and ordering of the pentacene overgrowth.¹ The mobility calculated from such devices is 0.48 cm² V^–1 s^–1, which is three times larger than the mobility of the device with untreated Au electrodes. The pentacene layers for both devices were deposited in the same deposition run. Figure 9 provides an explanation for the improvement in device performance. The SAM deposited on the Au resulted in a pentacene grain size on the Au similar to the large grains grown on the SiO₂. There is no transition region at the Au edge. The trap concentration must be drastically reduced.

Figure 9

Pentacene thin films are intolerant to exposure to the various chemicals used in typical lithographic processes; therefore, they cannot be patterned using such processes. Usually shadow masks are used during deposition to pattern the material. A technique was recently proposed which uses topographic discontinuities to isolate individual devices [53]. This technique requires that the semiconducting material and the photoresist remain in inactive areas, which can be a severe limitation for many applications. We have developed a subtractive technique in which pentacene is protected from the consequent lithographic steps by means of a chemically resistant layer.² A 1-µm-thick layer of parylene-N is first used to protect the sensitive pentacene layer from the solvents and other materials used. The two-layer structure thus formed is then etched using a typical lithographic process. Photoresist is spun on top of the structure and cured. The sample is then exposed to UV radiation through a lithographic mask and subsequently developed. The structure is then etched for 30 minutes in an oxygen plasma, resulting in well-defined patterns of protected pentacene on a clean substrate. Pentacene TFTs fabricated with this method had performance similar to that of shadow-mask-patterned pentacene devices fabricated in the same deposition run.

Solution-processed organic semiconductor films

One of the first solution-processable organic semiconductors used for field-effect transistors was poly(3-hexylthiophene) [20]. Poly(3-hexylthiophene) films spun from chloroform onto interdigitated gold source and drain electrodes displayed mobilities in the range of 10^–5 to 10^–4 cm² V^–1 s^–1. These mobilities were comparable to the mobilities obtained from electrochemically prepared polythiophene field-effect transistors [54], indicating that the incorporation of insulating alkyl side chains was not detrimental to the electronic properties of polythiophene. A comparison study of poly(3-alkylthiophene)s with side chains ranging in length from butyl to decyl showed that field-effect mobility decreases with increasing chain length [55]. For films spun from chloroform, mobilities ranged from 1 to 2 × 10^–4 cm² V^–1 s^–1 for poly(3-butylthiophene) and poly(3-hexylthiophene) down to 6 × 10^–7 cm² V^–1 s^–1 for poly(3-decylthiophene).

A dramatic increase in mobility was observed when regioregular poly(3-hexylthiophene), synthesized by the Reike method [56] and having greater than 98.5% head-to-tail (HT) linkages, was used to fabricate a field-effect transistor [31]. Poly(3-hexylthiophene) can have the 3-alkyl substituents incorporated into the polymer in two types of arrangements: head to tail and head to head (HH). If poly(3-hexylthiophene) consists of both HH and HT 3-alkylthiophene moieties, it is regiorandom. Regioregular poly(3-hexylthiophene) has only one type of arrangement, either HH or HT. Mobilities as high as 0.045 cm² V^–1 s^–1 were achieved when regioregular poly(3-hexylthiophene) was solution-cast (i.e., drop-cast) from a chloroform solution. When cast from solution, highly regioregular poly(3-hexylthiophene) self-orients into a well-ordered lamellar structure with an edge-on orientation of the thiophene rings relative to the substrate. This is in contrast to solution-cast films of regiorandom poly(3-alkylthiophene)s, which are totally amorphous [56]. Spin-coated films of regioregular poly(3-hexylthiophene) are also well-ordered, but the lamellae adopt different orientations depending on the degree of regioregularity [57]. Highly regioregular poly(3-hexylthiophene) (greater than 91% head-to-tail linkages) forms lamellae with an edge-on orientation (– stacking direction in the plane of the substrate) when spun from chloroform. Mobilities of 0.05 to 0.1 cm² V^–1 s^–1 were obtained for 96% regioregular poly(3-hexylthiophene). In contrast, spun films of poly(3-hexylthiophene) with low regioregularity (81% head-to-tail linkages) consisted of lamellae having a face-on orientation (– stacking direction perpendicular to the substrate) and gave mobilities of 2 × 10^–4 cm² V^–1 s^–1. Solution-cast films of 81% regioregular poly(3-hexylthiophene) adopted an edge-on lamellar structure, resulting in an order-of-magnitude increase in mobility compared to spin-coated films. This study indicates that, in addition to the degree of order of the polymer film, the orientation of the – stacking direction relative to the substrate has a large influence on the field-effect mobility.

A variety of solvents have been investigated for the deposition of thin films of regioregular poly(3-alkylthiophene)s, including chloroform, 1,1,2,2-tetrachloroethane, tetrachloroethylene, chlorobenzene, toluene, p-xylene, and THF. The various solvents produce films with different degrees of order, uniformity, and continuity. The mobility of regioregular poly(3-hexylthiophene) was found to vary by two orders of magnitude depending on the solvent used, with chloroform giving the highest mobility [56]. Modification of the substrate surface prior to deposition of regioregular poly(3-alkylthiophene) has also been found to influence film morphology. For example, treatment of SiO₂ with hexamethyldisilazane (HMDS) or an alkyltrichlorosilane replaces the hydroxyl groups at the SiO₂ surface with methyl or alkyl groups. The apolar nature of these groups apparently attracts the hexyl side chains of poly(3-hexylthiophene), favoring lamellae with an edge-on orientation. Mobilities of 0.05 to 0.1 cm² V^–1 s^–1 from highly regioregular poly(3-hexylthiophene) have been attributed to this surface modification process [35, 57, 58]. In addition, it has been shown that deposition onto a flat substrate (top-contact device) is favorable compared to deposition onto prefabricated source and drain electrodes (bottom-contact device), yielding mobilities that are typically larger by a factor of 2 [35, 58].

Several processing techniques have been found to enhance the on–off switching characteristics of poly(3-alkylthiophene) field-effect devices without affecting their mobility. Treatment of a poly(3-hexylthiophene) solution-cast film with ammonia by bubbling N₂ through an aqueous ammonium hydroxide solution for ten hours increased the on–off ratio by more than an order of magnitude. Short thermal treatments in a N₂ atmosphere were also found to reduce the off current slightly, but higher temperatures (~150°C) caused nearly an order of magnitude decrease in mobility [31]. Evaporation of a thin layer of SiO_x onto the poly(3-alkylthiophene) film has been shown to increase the on–off ratio by as much as four orders of magnitude. Immersing the film in hydrazine has a similar effect. Both substoichiometric SiO₂ and hydrazine are reducing agents which act to dedope the poly(3-alkylthiophene) film [35, 58]. Exposure of poly(3-alkylthiophene) films to air causes an increase in conductivity and a subsequent degradation of the transistor on–off characteristics. It is possible to achieve high on–off ratios by preparing and testing devices in a dry N₂ atmosphere [35, 58].

Although most studies have focused on poly(3-hexylthiophene), other regioregular poly(3-alkylthiophene)s have also been used to fabricate field-effect transistors. Poly(3-octylthiophene) was found to have a field-effect mobility similar to that of poly(3-hexylthiophene), 0.01 to 0.03 cm² V^–1 s^–1 for devices with an indium–tin oxide (ITO) [on poly(ethylene terephthalate)] gate, polyimide insulating layer, and top-contact conductive ink source and drain electrodes. For this same device configuration, the mobility of regioregular poly(3-dodecylthiophene) was only 10^–6 cm² V^–1 s^–1 [59]. This trend of decreasing mobility with increasing side chain length is the same as was observed in a series of regiorandom poly(3-alkylthiophene)s [55].

In addition to the variation of alkyl side-chain length, modification of the chemical nature of the side chains in regioregular substituted polythiophenes has also been investigated [60]. Carboxylic acid side chains were incorporated in an attempt to use hydrogen bonding to aid self-orientation at polar substrate surfaces (e.g., SiO₂). However, solution-cast films of this polymer were nonuniform and essentially amorphous, and mobilities were in the range of 10^–5 to 10^–4 cm² V^–1 s^–1. A similar lack of crystallinity and low mobilities were observed in cast films of regioregular polythiophenes with bulky oxazoline side chains. The introduction of an optically active methyl branch off a butyl side chain produced crystalline films and field-effect mobilities of 10^–3 cm² V^–1 s^–1. The increased – stacking distance caused by the methyl branch is thought to be responsible for the reduction of the mobility by one to two orders of magnitude compared to regioregular polythiophenes with unbranched alkyl side chains.

While spin coating, solution casting, and printing are perhaps the most commercially feasible processing techniques for soluble polymeric semiconductors, the Langmuir–Blodgett (LB) technique has also been explored for the preparation of poly(3-alkylthiophene) field-effect transistors [61–63]. Initial LB studies used mixtures of regiorandom poly(3-hexylthiophene) and arachidic acid. Substantial molar percentages of the long-chain fatty acid were necessary for the formation of stable Langmuir films at the air–water interface and subsequent vertical transfer onto a transistor substrate [61, 62]. Mobilities ranged from 10^–7 to 10^–4 cm² V^–1 s^–1 for multilayer LB films, and were even lower for monolayer films. These low mobilities are not surprising considering the regiorandom nature of the polymer and the insulating nature of the fatty acid. A subsequent study showed that it is possible to form a stable Langmuir monolayer of regioregular poly(3-hexylthiophene) without the addition of a long-chain fatty acid [63]. Transistors were fabricated by horizontal (Langmuir–Schaeffer) deposition onto hydrophobic substrates. The transferred films displayed a layered structure with an edge-on orientation of the thiophene rings, and were more or less ordered depending on the spreading solvent used. Mobilities ranged from 3 × 10^–3 to 2 × 10^–2 cm² V^–1 s^–1 for multilayer films (2–5 layers) and were lower for monolayers. Although optical characterization of the films indicated a tendency of the polymer backbones to align parallel to the trough barriers, this anisotropy was not apparent in the electrical measurements. The difficulty of vertical transfer of LB monolayers of regioregular poly(3-alkylthiophene)s has been addressed by chemically modifying the side chains in order to make the polymer amphiphilic [64]. The introduction of perfectly alternating dodecyl (hydrophobic) and trioxynonyl (hydrophilic) groups allows vertical deposition of a monolayer onto a hydrophilic substrate with a transfer ratio of 1.0 ± 0.1. It is also possible to form a patterned polymer layer by deposition onto a substrate with patterned hydrophilic and hydrophobic areas. Such amphiphilic, regioregular polythiophenes are promising materials for LB field-effect transistors, especially if transfer ratios remain near unity for the deposition of a few multilayers.

Field-effect mobilities of vacuum-evaporated oligothiophenes, especially end-substituted oligothiophenes, are among the highest reported for organic semiconductors, with the exception of pentacene. Unsubstituted quinquethiophene and end-substituted quater-, quinque-, and hexathiophene display enough solubility in organic solvents to allow fabrication of field-effect devices by solution-processing techniques. Initial studies of solution-processed oligothiophene transistors gave mobilities of ~5 × 10^–5 cm² V^–1 s^–1 for quinquethiophene and ,'-diethylquaterthiophene (DE4T) [65]. Evaporated films of DE4T yielded only slightly higher mobilities, of the order of 9 × 10^–5 cm² V^–1 s^–1. According to X-ray diffraction studies, both evaporated films and films solution-cast from chloroform were highly crystalline, with the DE4T molecules arranged in a layered structure [65].

More recently, mobilities in the range of 10^–2 to 10^–1 cm² V^–1 s^–1 have been achieved using solution-processed substituted oligothiophenes [38, 66, 67]. Not surprisingly, the mobility is found to depend strongly on film morphology, which can be controlled by the processing conditions. In one study, ,'-dihexylhexathiophene (DH6T) and ,'-dihexylquaterthiophene (DH4T) were dissolved in hot chlorobenzene or 1,2,4-trichlorobenzene, then solution-cast onto bottom-contact substrates. The solvent was evaporated in a vacuum oven at temperatures between room temperature and 100°C. The resulting films were nonuniform in both thickness and morphology and yielded a range of mobilities. For DH6T films prepared in this way, the mobilities ranged from 4 × 10^–3 to 5 × 10^–2 cm² V^–1 s^–1, with the highest mobilities corresponding to trichlorobenzene and 70°C solvent evaporation. The mobilities of DH4T films were less consistent, ranging from 1 × 10^–3 to 7 × 10^–2 cm² V^–1 s^–1, but generally slightly lower than those of DH6T films prepared under the same conditions [66]. In another study, additional solvents were investigated, including 3-methylthiophene, anisole, and toluene. After DH6T was dissolved in hot solvent, the solution was cast onto a heated (50–60°C) substrate, and the solvent was evaporated under partial vacuum. Chlorobenzene, 1,2,4-trichlorobenzene, and 3-methylthiophene all yielded smooth films and mobilities occasionally as high as 0.1 cm² V^–1 s^–1 and routinely 0.03 cm² V^–1 s^–1 when the DH6T concentration was less than 0.1%. For higher concentrations, the films were thicker but rough and gave low mobilities. The less polarizable solvents anisole and toluene did not produce smooth films because the DH6T precipitated from the solution prior to any nucleation at the substrate surface. A mobility of 0.03 cm² V^–1 s^–1 was also obtained from a film of ,'-dihexylquinquethiophene (DH5T) cast from chlorobenzene onto a 50°C substrate [67]. A subsequent study reported a mobility of 0.1 cm² V^–1 s^–1 from a DH5T film solution-cast from toluene [38].

High mobilities (1.2 × 10^–2 cm² V^–1 s^–1) have also been reported for films of DH4T spun from chloroform [68]. A polycrystalline film of DH4T resulted when the SiO₂ substrate was first treated with n-octyltrichlorosilane to make the surface hydrophobic and then held at 110°C while the film was being spun. Differential scanning calorimetry (DSC) measurements indicate that DH4T is in a liquid-crystalline phase at this temperature, and this apparently facilitates long-range molecular organization. The mobilities of films prepared by spin-coating in this manner are only slightly lower than the mobilities of vacuum-evaporated films of DH4T, which are of the order of 3 × 10^–2 cm² V^–1 s^–1. Chemical modification of end-substituted thiophene oligomers to have alkoxypropyl instead of alkyl end groups has been investigated as a means of increasing oligomer solubility. In particular, field-effect transistors were fabricated from quaterthiophene and hexathiophene end-substituted with 3-butoxypropyl groups, C4OC34T and C4OC36T [38]. The solubility of these oligomers was increased by as much as a factor of 2 compared to the corresponding alkyl-substituted oligomers. C4OC34T and C4OC36T were solution-cast from chlorobenzene or 1,2,4-trichlorobenzene, and the solvent was evaporated in a vacuum oven. Although the incorporation of an oxygen atom into the end group did not seem to affect the packing and spacing of molecules in cast films, the mobilities for C4OC36T (1 × 10^–3 to 1 × 10^–2 cm² V^–1 s^–1) were slightly lower than those of DH6T, and the mobilities for C4OC34T (5 × 10^–5 to 1 × 10^–3 cm² V^–1 s^–1) were significantly lower than those of DH4T.

Other soluble organic oligomers have also been investigated as semiconducting materials for field-effect transistors. Anthradithiophene, a fused heterocycle compound similar to pentacene, is soluble in its dihexyl end-substituted form. Dihexylanthradithiophene (DHADT) transistors were fabricated by solution-casting from hot chlorobenzene, then evaporating the solvent in a vacuum oven at various temperatures. The electrical characteristics of the films were strongly dependent on the solvent evaporation temperature, with no field effect observed for room temperature, 50°C, 150°C, or 180°C. For films dried at 70°C, mobilities ranged from 6 × 10^–4 to 3 × 10^–3 cm² V^–1 s^–1. The highest mobilities, 0.01 to 0.02 cm² V^–1 s^–1, were obtained for a drying temperature of 100°C [37, 67]. In comparison, vacuum-evaporated films of DHADT gave mobilities as high as 0.15 cm² V^–1 s^–1 [37]. Transistors utilizing another thiophene-containing oligomer, trans-trans-2,5-bis-[2-{5-(2,2'-bithienyl)}ethenyl]thiophene (BTET), were fabricated by spin-coating from hot NMP. The mobility of such a device was 1.4 × 10^–3 cm² V^–1 s^–1 compared to 0.012 cm² V^–1 s^–1 for a vacuum-evaporated device [69].

Most organic semiconductors, including the thiophene oligomers discussed above, are p-type semiconductors. However, a naphthalenetetracarboxylic diimide derivative with fluorinated R groups has been reported as a solution-processable n-type organic semiconductor [38]. This compound is soluble in hot ,,-trifluorotoluene, and solution-casting results in morphologically nonuniform films, with some regions of the films giving mobilities greater than 0.01 cm² V^–1 s^–1. Vacuum-deposited films yielded higher mobilities.

Although a great deal of success has been achieved with soluble oligomers, the solubilities of these oligomers are low, requiring the solvents to be heated, and casting and spinning yield films that are nonuniform in thickness, morphology, and electrical properties. Another approach to solution-processable oligomeric materials is to begin with a precursor molecule that is soluble but not semiconducting, and then convert it to its semiconducting, insoluble form. This approach has been realized for pentacene, with initial mobilities of 0.01 to 0.03 cm² V^–1 s^–1 reported [11, 70, 71]. The pentacene precursor is soluble in dichloromethane and forms continuous, amorphous films when spun onto transistor substrates. The conversion to pentacene is accomplished by heating the films to a temperature of 140–220°C in vacuum for several minutes to two hours. Tetrachlorobenzene is eliminated in the conversion process, and electron microscopy reveals microcrystallites in the converted films. In a subsequent study, a mobility of 0.2 cm² V^–1 s^–1 was achieved by treatment of the SiO₂ substrate with HMDS prior to spin-coating the precursor, and by optimizing the conversion conditions. These optimized conditions involved spinning the precursor from a 1.5 wt% solution in dichloromethane and converting the film for five seconds at 200°C, followed by rapid quenching [72].

This precursor approach can also be applied to polymers. In fact, one of the first reported organic transistors used precursor-route polyacetylene as the semiconducting layer. The polyacetylene precursor was spun from cold 2-butanone, and conversion was achieved by heating to 80–100°C in vacuum for 12 hours. The mobilities of such transistors were low, however, in the range of 10^–4 to 10^–5 cm² V^–1 s^–1 [18, 73]. Another polymeric semiconductor that has been processed from a soluble precursor polymer is polythienylenevinylene (PTV). In an initial study, a PTV film was spun from dimethylformamide and then coated with a layer of precursor poly(p-phenylene vinylene) (PPV). Both the PTV and PPV were converted at 100–300°C in a N₂ atmosphere, with the PPV precursor supplying protons to catalyze the PTV conversion. A mobility of 6 × 10^–4 cm² V^–1 s^–1 was reported for a device prepared in this way [54]. Subsequent studies have spun the PTV precursor from chloroform and dichloromethane, and have converted the PTV at 140–160°C in a N₂ atmosphere with a flow of HCl gas to catalyze the reaction [11, 71, 74]. Although the mobilities remain fairly low, in the range of 10^–4 to 10^–3 cm² V^–1 s^–1, one advantage of this processing approach is that organic solvents can be used in subsequent device-processing steps without disturbing the converted PTV [74].

Conclusions

All in all, organic TFTs are close to passage from the research laboratory to product development and then to manufacturing of new products based on organic semiconductors.

References

Footnotes

Received August 7, 2000; accepted for publication October 17, 2000