Integrated Optoelectronic Devices Based on Conjugated Polymers

See allHide authors and affiliations

Science  12 Jun 1998:
Vol. 280, Issue 5370, pp. 1741-1744
DOI: 10.1126/science.280.5370.1741


An all-polymer semiconductor integrated device is demonstrated with a high-mobility conjugated polymer field-effect transistor (FET) driving a polymer light-emitting diode (LED) of similar size. The FET uses regioregular poly(hexylthiophene). Its performance approaches that of inorganic amorphous silicon FETs, with field-effect mobilities of 0.05 to 0.1 square centimeters per volt second and ON-OFF current ratios of >106. The high mobility is attributed to the formation of extended polaron states as a result of local self-organization, in contrast to the variable-range hopping of self-localized polarons found in more disordered polymers. The FET-LED device represents a step toward all-polymer optoelectronic integrated circuits such as active-matrix polymer LED displays.

Solution-processible conjugated polymers are among the most promising candidates for a cheap electronic and optoelectronic technology on plastic substrates. Polymer LEDs exceeding peak brightnesses of 106 cd m–2(1) and high-resolution video polymer LED displays (2) have been demonstrated. One of the main obstacles to all-polymer optoelectronic circuits is the lack of a polymer FET with sufficiently high mobility and ON-OFF ratio to achieve reasonable switching speeds in logic circuits (3) and to drive polymer LEDs.

Conjugated polymer FETs (4) typically show field-effect mobilities of μFET = 10–6 to 10–4 cm2 V–1 s–1, limited by variable-range hopping between disordered polymer chains and ON-OFF current ratios of <104 (5). This is much too low for logic and display applications, and therefore all previous approaches to drive polymer LEDs have used polycrystalline (2) or amorphous silicon (a-Si) (6) technology. Recently, a polymer FET with a mobility of 0.01 to 0.04 cm2V–1 s–1 and an ON-OFF ratio of 102 to 104 using regioregular poly(hexylthiophene) (P3HT) was described (7). The high mobility is related to structural order in the polymer film induced by the regioregular head-to-tail (HT) coupling of the hexyl side chains. However, a clear understanding of the transport mechanism giving rise to the relatively high mobilities is still lacking.

Here, we report a considerably improved P3HT FET reaching mobilities of 0.05 to 0.1 cm2 V–1 s–1 and ON-OFF ratios of >106, the performance of which starts to rival that of inorganic a-Si FETs and enables us to demonstrate integrated optoelectronic polymer devices. As an example, we have chosen a simple pixel-like configuration in which the FET supplies the current to a polymer LED. This allows us to assess the prospects of active-matrix addressing in all-polymer LED displays.

To construct the multilayer device (Fig.1A), we first fabricated the FET by spin-coating a film of P3HT (500 to 700 Å) (8) onto a highly doped n+-Si wafer with a 2300 Å SiO2gate oxide (capacitance C i = 15 nF cm–2). Au source-drain contacts were deposited onto the P3HT through a shadow mask. Then, a layer of SiOx was thermally evaporated through another, mechanically aligned, shadow mask to define the active LED area on the finger-shaped Au FET drain electrode acting as the hole-injecting anode of the LED. A single layer of poly[2-methoxy-5-(2′-ethyl-hexyloxy)-p-phenylenevinylene] (MEH-PPV) was spin-coated on top. Evaporation of a semitransparent Ca-Ag cathode completed the device. No photolithographic steps were involved. The device is a part of an active-matrix LED display pixel (9), which would require an additional transistor T2 and a storage capacitor C s controlling the voltage on the gate of T1 (Fig. 1A).

Figure 1

(A) Cross section of the integrated P3HT FET and MEH-PPV LED. The device is a part (shown inside the dashed area in the top left corner) of a full active-matrix polymer LED pixel. The lamellar structure of the regioregular P3HT and its orientation relative to the SiO2 substrate and the direction of the in-plane FET current I d are shown schematically. (B) Photograph of a FET-LED with one of the four “pixels” switched on. The MEH-PPV layer (orange) was made to cover the substrate only partially in order to make the underlying (blueish) P3HT layer visible.

When the FET was switched on (Fig. 1B), the current flowed from the FET source electrode (right contact probe) to the LED cathode (left contact probe). Evidently, the LED could be switched on and off by the FET gate voltage V g (Fig.2A). At V g = –50 V, the FET supplied a current density of ∼10 mA cm–2 to the LED, resulting in a brightness on the order of 1 cd m–2. From the linear relation between the FET currentI d and the detected photocurrentI p (Fig. 2B), we estimate the external LED quantum efficiency to be ηext ≈ 0.01%. This value is much smaller than in an optimized MEH-PPV LED with ηext = 1% (10). This is attributed to nonoptimized carrier injection at the Au and Ca-Ag electrodes, the use of a single-layer LED, and absorption losses in the 20-nm Ca-Ag cathode. With an optimized LED (ηext > 1%) (10), the current density of 10 mA cm–2 supplied by the FET, which is regarded as a design rule for polymer LED displays, should be sufficient to achieve a practical video brightness of 100 cd m–2. If the channel length L were reduced from 75 μm to ∼20 μm, the width W of the FET could be scaled down to integrate the device fully along one edge (with lengthd) of the LED anode—that is, Wd—such that it occupies only a small area.

Figure 2

(A) Brightness of the LED (circles) and drain current I d supplied by the FET to the LED (triangles) as a function of the FET gate voltage V g. I d is normalized by the area A of the LED. (B) Linear relationship between I d and the photocurrentI p detected by a Si photodiode above the LED (A = 300 μm by 430 μm, channel length L = 75 μm, channel width W = 1500 μm, source-drain voltageV sd = −70 V).

Our simple device does not address the complex issues of integrating a full pixel structure (9), which would also require the use of polymeric insulators (11). Moreover, to lower the high FET operating voltage, it is necessary to reduce the gate insulator thickness and to improve the subthreshold characteristics (see below). However, the device clearly demonstrates that high-mobility P3HT FETs have sufficient driving current to switch polymer LEDs of similar size (Wd). The relative crudeness of our integration scheme illustrates the ease of fabrication and the robustness of solution-processed conjugated polymers for multilayer integrated devices. (Our first attempts to realize the FET-LED with an oligomer FET deposited by vacuum sublimation failed, as the buried FETs degraded during the deposition of subsequent layers. No degradation was observed for P3HT devices.)

In the following, we focus on the improvement and transport mechanism of the P3HT transistors as the key element of the integrated device. A typical improved P3HT FET switches on sharply aroundV 0 = 0 to 4 V, with good subthreshold slopes of 1 to 1.5 V decade–1 (Fig.3A). The ON-OFF ratio betweenV g ≈ 0 V andV g = –60 V exceeds 106, the OFF current being limited by gate leakage. The film conductivity σ is <10–8 S cm–1. From the transfer characteristics in the saturation regime, we extract mobilities μFET = 0.05 to 0.1 cm2 V–1s–1, depending on the details of device preparation. This improvement of the ON-OFF ratio by two orders of magnitude and of the mobility by a factor of 2 (7) is attributed to the optimized device fabrication reducing unintentional doping and promoting self-organization of the polymer. All solution preparation and device processing steps were performed in a dry N2 atmosphere, because the conductivity of P3HT films was found to increase upon exposure to air for a few minutes. Deposition of a layer of SiOx (Fig. 1A) onto the surface of such air-exposed films restored the low conductivity, which suggests that the substoichiometric SiOx attracts dopants from the P3HT surface layer. From capacitance-voltage measurements on n+-Si–SiO2–P3HT–Au diodes, we estimate the residual bulk doping level of carefully processed P3HT films to be ∼5 × 1015 cm–3. The mobility increase is partly attributed to predeposition treatment of the Si-SiO2 substrate with the silylating agent hexamethyldisilazane replacing the natural hydroxyl groups on the SiO2 surface with apolar methyl groups. Although the atomic interface structure is not known, this is believed to promote phase segregation of the polymer at the interface (Fig. 1A). The deposition onto a flat surface was also found to be advantageous relative to coating the polymer onto a substrate with prefabricated source-drain contacts (7). Films were spin-coated at 2000 rpm from a solution of P3HT (0.7 to 0.8 weight %) in CHCl3.

Figure 3

(A) Transfer characteristics in the saturation regime (v sd = −80 V) and (B) output characteristics of a typical P3HT FET (L = 75 μm, W = 1500 μm).

Regioregular P3HT with μFET ≈ 0.1 cm2V–1 s–1 and σ < 10–8 S cm–1 clearly does not follow the “universal relationship” between conductivity and mobility (μFET∝ σδ, δ ≈ 0.7) that is characteristic of variable-range hopping in more disordered conjugated polymers (5). The mobility is of the same order of magnitude as in polycrystalline oligomer FETs with related structures such as α-sexithiophene (11) and approaches the mobility of molecular single-crystal FETs (∼1 cm2 V–1s–1) (12). This strongly suggests that the origin of the exceptionally high polymer mobilities reported here is the formation of extended current-transporting states similar to those in structurally related oligomers. Extended states cannot be formed unless there is at least short-range, if not microcrystalline, order. X-ray diffraction shows that on a local scale, regioregular P3HT has a lamellar, phase-segregated structure with alternating layers of conjugated backbones and interdigitated side chains parallel to the substrate (7, 13), favorable for in-plane FET transport (Fig. 1A) (14). Extended-state formation is believed to be the result of relatively long conjugation lengths along polymer chains (8) and π-π stacking of adjacent chains. Grain boundaries between the small crystalline domains with more disordered chain conformation, residual doping, and other structural or chemical defects give rise to localized trap states.

This electronic structure justifies the analysis of the FET characteristics in terms of a model that has proven very successful for a-Si FETs (15) and has already been applied to oligomer FETs (12, 16). The model assumes that the current is transported by extended (that is, high-mobility) states above a mobility edge E c, below which there is a tail of localized states with much lower mobility, but it makes no further assumptions about the nature of the localized states, nor about the transport mechanism above E c. Hence, the model may also be applicable to polarons in a conjugated polymer. The transconductance dI d/dV gin the saturation regime is given byEmbedded Image(1)(15), where the field-effect mobility μFET is defined byEmbedded Image(2)and where n ind(n ind S) is the total density of induced interface carriers with elementary charge e [at the source (S) electrode], and n f is the fraction ofn ind in current-transporting states aboveE c with mobility μ0. Neglecting the small variation of the Fermi level E F withV g,n ind S increases linearly withV g, that is,n ind SC i/e(V gV FB) above the flat-band voltageV FB.

If the tail of localized trap states is steep,E F is below the typical tail state energyE t and varies little withV g. Then, μFET ∝ μ0 exp[–(E cE t)/kT] is constant, resulting in a linear dependence ofdI d/dV g on (V gV T) above the threshold voltage V T. This is observed in high-quality a-Si (17) and in most oligomer FETs. However, if E F enters the distribution of localized states, for example, for a relatively broad distribution, μFET exhibits a strong dependence onn ind. This seems to be the case in P3HT FETs.

Around room temperature, the transconductance of a typical P3HT FET is linear above ∣V g∣ > 30 V (∣V T∣ = 10 to 20 V), corresponding to a constant field-effect mobility of μFET = 0.05 to 0.1 cm2 V–1 s–1 (Fig.4A). However, there is an extended nonlinear region between the transistor switch-on atV 0 and V T in which μFET depends on V g. At low temperatures, μFET depends on V gin the whole observable voltage range (18). At fixed values of n ind S, the mobility calculated from Eq. 1(19) shows a roughly exponential temperature dependence proportional to exp(–E a/kT) with positive deviations below ∼150 K (Fig. 4B). Below ∼80 K, measurements were not possible because of increased turn-on voltages indicating disorder-induced carrier localization. The activation energies E a = 60 to 100 meV are similar to those of oligomer and a-Si FETs (17) but are small relative to hopping activation energies in low-mobility polymers (E a > 0.2 eV) (5). The decrease ofE a with increasingn ind S (Fig. 4C) suggests that the dominant exponential temperature variation of μFET reflects the distribution of localized states—that is, the ration f/n ind(T) (see Eq. 2)—and not a possible temperature dependence of μ0(T), which is not expected to depend onn ind.

Figure 4

(A) TransconductancedI d/dV g in the saturation regime (L = 75 μm, W = 1500 μm,V sd = –70 V) at 320 K (crosses) and 144 K (squares). (B) Field-effect mobility atn ind S = 6 × 1012cm–2 as a function of temperature between 330 and 84 K. (C) Activation energy E a of the mobility extracted between 300 and 160 K as a function ofn ind SC i/e(V gV 0).

We do not attempt here to deduce information about the distribution of localized states, which would require careful device modeling and independent spectroscopic measurements. Therefore, an estimate of the magnitude of the trap-free mobility μ0and the corresponding mean free path length cannot be given at present. [For a-Si with μFET = 0.1 to 1 cm2V–1 s–1, μ0 ≈ 10 cm2 V–1 s–1 has been estimated (17).] We can conclude, however, that our model involving extended polaron states induced by local self-organization and disorder-induced localized states allows a consistent understanding of the exceptionally high room-temperature mobilities (close to those of oligomers with much higher polycrystalline order) and of the thermally activated temperature dependence. The latter mainly reflects the distribution of localized states rather than any intrinsic properties of polaron hopping (20).

Our model provides a consistent description of transport in high-mobility polymer FETs as well as polycrystalline oligomer FETs. For the latter, recent results, which contradict earlier experiments (20), have shown that as the crystalline order of the organic layers improves, the mobility becomes independent of temperature (12, 21). For the much less perfectly ordered P3HT, an increase in the crystallite grain size and a reduction of the density of localized states will result in lower threshold voltages, reduced activation energies, and possibly even higher mobilities.

The high-mobility polymer FET demonstrated here with a high ON-OFF ratio and good postprocessing robustness approaches the requirements for all-polymer, multilayer electronic and optoelectronic integrated circuits. With some further improvement of FET performance, polymer logic circuits and active matrix addressing in all-polymer LED displays might be feasible.

Note added in proof: Since submission of the final version of the manuscript, we have become aware that an organic LED driven by a polymer FET has independently been demonstrated (22).


View Abstract

Navigate This Article