Comparative study of organic transistors with different graphene electrodes fabricated using a simple patterning method

Two-dimensional (2D) graphene is considered to be a promising electrode material for organic field-effect transistors (OFETs)^1–10 because the use of conventional metal (Au or Pd) electrodes limits the OFET performance originated from discontinuity in the morphology, interfacial dipole barrier formation, and Schottky barrier formation at the metal/organic interface.^11–13 Studies have shown that charge injection is enhanced when graphene is used as electrodes for OFETs in comparison to conventional metal electrodes.^1–10 The enhanced charge injection has been attributed to a lower injection barrier induced from the strong π-π interaction and work function matching at the graphene/organic interface. For practical applications, scaled up fabrication processes for graphene films and their easy patterning to electrodes are needed for OFET fabrication. Both the chemical vapor deposition (CVD) grown graphene and reduced graphene oxide (RGO) films can be used for scalable OFET fabrication processes. Graphene grown through CVD is known to produce large-area films with high conductivity and in some cases close to the conductivity of pristine graphene. On the other hand, RGO is advantageous as it can be produced in large quantities at low cost through a solution-processing technique. Although both the CVD graphene and RGO can be used as electrode materials for the fabrication of OFETs, a comparative study of these two types of graphene electrodes on the OFET performance is still lacking. Such a study is of great interest for future OFET applications.

In this paper, we present a detailed comparative study of the performance of OFETs fabricated using electrodes made from either CVD or RGO large-area graphene films. For the electrode fabrication, we introduce a simple technique for the large-area patterning of graphene films using a metal etch mask along with a sacrificial polymer layer. This etch mask protects the graphene electrodes, while the unwanted film is exposed to oxygen plasma. After patterning, the etch mask is removed by a simple lift-off process in acetone (without requiring any harmful acid treatment), leaving only the graphene electrodes on the substrate. We keep the same device geometry for both CVD and RGO electrodes. For the fabrication of OFETs, we deposit pentacene thin films on patterned graphene electrodes and study the electronic transport properties. Metal electrode OFETs of the same geometry were also fabricated for comparison. The measured transfer and output characteristics of the OFETs show that although the mobility and on-current and current on-off ratio are higher for both the CVD graphene and RGO electrode compared to metal electrodes, the highest performance was achieved using the CVD graphene electrodes. Since CVD graphene has higher conductivity than RGO, our study suggests that, in addition to the strong π-π interaction at the graphene/organic interface, the conductivity of the electrodes also plays a role in the performance of OFETs.

The fabrication steps of large area graphene electrodes via a simple patterning technique are illustrated in Fig. 1. Two different types of large area graphene films were prepared: (i) CVD graphene and (ii) RGO film. The films were placed on heavily doped silicon (Si) substrates with a 250 nm thermally grown silicon dioxide (SiO₂) dielectric layer for further processing [Figs. 1(b) and 1(c)]. Single-layer CVD graphene was grown using low pressure CVD on copper.^14,15 Contemporary transfer methods were used to then transfer the graphene film to our target Si/SiO₂ substrate.^16,17 A sacrificial PMMA supported the graphene film as the copper catalyst was etched using ammonium persulfate solution, before immersion in a subsequent series of modified RCA solutions. The transferred graphene is predominantly single layer with a low defect density of ∼2 × 10¹⁰ cm⁻² and a carrier mobility on the order of 1000 cm²/V s, determined via Raman spectroscopy¹⁸ (Fig. S1 in the supplementary material) and the electrical transport measurement [Fig. 2(b)], respectively. In addition, the XPS study (Fig. S1 in the supplementary material) shows the carbon sp² fraction to be 92%.

RGO sheets were synthesized by reducing the individual graphene oxide (GO) sheets via the hydrazine hydrate method following our previously published technique.^19–21 The GO powder (Cheap Tubes Inc. 30 mg) was diluted in deionized (DI, 30 ml) water by sonication followed by a 24 h magnetic stirring. In the GO solution, 200 μL of 5% NH₃ aqueous solution was added to adjust the pH of the solution, followed by addition of 30 μL of hydrazine. Finally, the solution was heated at 90 °C for 60 min to obtain RGO (Figs. S2 and S3 in the supplementary material). RGO solution was then spin-coated on top of the Si/SiO₂ substrate at 1000 rpm for 30 s, which was repeated ∼20 times, yielding a 20–30 nm thick layer (Fig. S3 in the supplementary material). The carbon sp² fractions of the RGO are calculated to be 80% from the XPS data (Fig. S3 in the supplementary material).

After placing the graphene films (both CVD graphene and RGO) on the substrate, a polymer layer (950 PMMA C2, Micro Chem) was spin-coated at 4000 rpm for 1 min, resulting in a ∼300 nm thick layer [Fig. 1(d)]. Then, a 60 nm thick metal (Al) film was thermally evaporated through a shadow mask [Fig. 1(e)]. The sample was then exposed to the oxygen plasma etching process for 40 min, and the areas not protected by Al were etched away during the plasma exposure [Fig. 1(f)]. The metal fills the role of the “etching mask,” which protects the graphene during the plasma etching process. Finally, a simple liftoff procedure was carried out by immersing the sample in acetone at 60 °C for 1 h to obtain patterned graphene electrodes [Fig. 1(g)]. The patterned graphene films were then annealed under H₂/Ar gas at 350 °C for 2 h to remove any residual organics.²² The polymer (PMMA) used in the patterning process acts as a sacrificial layer that allows for simple acetone lift-off removal of the etch mask without the use of harmful acid treatments employed by the conventional metal etch mask processes, making the process environmentally friendly.

A pentacene film with a thickness of 30 nm was thermally deposited under vacuum at a pressure of 2 × 10⁻⁶ mbar. In order to minimize the device-to-device fluctuation from the active material morphology, all the pentacene films were deposited under identical conditions. The morphological investigation carried out using atomic force microscopy (AFM) (Fig. S4 in the supplementary material) shows that the pentacene films in the active area (between the electrodes) have a similar morphology, with an average grain size in the range of 150–200 nm. However, the morphology of pentacene is different on different electrodes. The electrical transport measurements were performed using a Hewlett-Packed (HP) 4145B semiconductor parametric analyzer connected to a probe station inside an enclosed glove box system with N₂ gas flow. A total of 30 OFET devices (10 per device configuration) were investigated for CVD graphene, RGO, and Pd electrodes.

The high quality of our two different types of graphene film on the SiO₂ substrate is shown in Fig. 2. The scanning electron microscopy (SEM) image of monolayer CVD graphene transferred on the Si/SiO₂ substrate is shown in Fig. 2(a). The transferred graphene is predominantly single layer with a low defect density ∼2 × 10¹⁰ cm⁻², determined via Raman spectroscopy¹⁸ (Fig. S1 in the supplementary material). Figure 2(b) shows a representative plot of the room temperature transfer characteristics [drain current (I_d) vs. gate voltage (V_g)] at a fixed source-drain voltage of V_d = 0.1 V. The mobility is calculated to be 971 cm²/V s from the transfer curve.

Figure 2(c) shows the SEM image of a RGO film on Si/SiO₂ substrates. Figure 2(d) shows representative transfer characteristics at a fixed V_d = 0.1 V, yielding a mobility of ∼0.1 cm²/V s. Both the films exhibit a standard ambipolar behavior, an intrinsic property of graphene. The device geometry was the same for both the CVD graphene and RGO films (channel length × width = 5 μm × 25 μm).

The CVD graphene and RGO films were then patterned using the scheme shown in Fig. 1. The optical and SEM images of the resulting patterned graphene electrodes are shown in Fig. 3. The size and shape of the pattern were determined by the shadow mask. The resulting channel length and width were 50 and 1000 μm, respectively, for all our samples. Pentacene films were deposited via thermal evaporation, and electrical characterization was then carried out.

The I_d-V_d characteristics at different V_g of a typical pentacene field-effect transistor using CVD graphene, RGO, and Pd electrodes are shown in Figs. 4(a)–4(c). All the devices show a good gate modulation with a linear behavior at low V_d (see Fig. S5 in the supplementary material for low voltage data) and a saturation behavior at higher V_d, typical of p-channel OFETs. The ohmic behavior at low bias indicates that excellent contacts are formed between the graphene and pentacene films. For comparison of device characteristics, we plotted all the curves on the same scale. The output current (at V_d = −50 V and V_g = −20 V) of the devices is 3.25, 1.73, and 0.48 μA for CVD graphene, RGO, and Pd electrodes, respectively. Compared to the control Pd electrodes, the output current is increased by approximately 4- and 10-folds for RGO and CVD graphene electrodes, respectively.

To further evaluate the performance of the transistors, we measured the corresponding transfer curve at a fixed V_d = −50 V [Figs. 4(d)–4(f)] from which the field effect mobility (μ), on-off ratio (I_on/I_off), and on-current (I_on) of the devices were obtained. The mobility is calculated using the standard formula, μ = (2LI_d,sat)/(WC_g(V_g − V_T)²), where L is the channel length, W is the channel width, I_d,sat is the saturation current, V_T is the threshold voltage, and C_g is the gate dielectric capacitance. The μ of the devices is 0.33, 0.07, and 0.005 cm²/V s for CVD graphene, RGO, and Pd electrodes, respectively. Although there is a device to device variation in V_T (Fig. S6 in the supplementary material), a negative shift in V_T was observed for CVD and RGO contacted devices compared to Pd contacted devices which could originate from the morphology of pentacene at the electrode-channel interface.²³ The I_on/I_off and I_on (I_d at V_g = −80 V) for CVD/pentacene and RGO/pentacene are 4.01 × 10⁴ and 122 μA and 1.91 × 10⁴ and 23.5 μA, respectively, whereas they are 1.83 × 10³ and 0.263 μA for Pd/pentacene. From here, we see that the I_on and I_on/I_off for CVD/pentacene and RGO/pentacene are more than an order of magnitude higher than those for control Pd electrodes.

We measured a total of 30 pentacene OFET devices, 10 each for CVD graphene, RGO, and Pd electrodes. All the devices have shown a similar saturation behavior and good gate modulation with linear behavior in the low bias regime (as in Fig. 4). Figure 5 shows a summary of the device characteristics (mobility vs current on-off ratio) for all pentacene FETs using CVD, RGO, and Pd electrodes. For the CVD/pentacene devices, the maximum (average) mobility is 0.33 cm²/V s [0.21(±0.04) cm²/V s], with the corresponding current on-off ratio varying from 3.5 × 10⁴ to 5.9 × 10⁴ with an on-current in the range of 74.1 μA to 125 μA. The maximum (average) mobility of RGO/pentacene varies from 0.07 cm²/V s [0.064(±0.005) cm²/V s], with the corresponding on-off ratio varying from 1.3 × 10³ to 2.5 × 10³ and on-current varying from 21.2 μA to 26.5 μA. The maximum (average) mobility of Pd/pentacene is 0.005 cm²/V s [0.003(±0.001) cm²/V s], with the corresponding on-off ratio varying from 2.4 × 10² to 2.6 × 10³ and on-current in the range of 1.1 μA to 2.6 μA. The maximum mobility is 5 times higher, and the on-off ratio is one order of magnitude higher for the CVD/pentacene in comparison to RGO/pentacene devices. The enhancement is much higher (mobility 60 times and current on-off ratio 2 order of magnitude) when compared with Pd/pentacene OFETs. We note that the difference in I_on/I_off is due to the differences in I_on only (Fig. S7 in the supplementary material) as there is no systematic variation of I_off for different types of electrodes. The performance enhancement depending only on the electrode/organic interface is rather impressive, considering the fact that no surface treatments were performed on our devices.

The improved performance of OFETs using graphene electrodes (both CVD graphene and RGO) over the control Pd electrode can be attributed to enhanced charge injection at the graphene/pentacene interface. Even though the work function of Pd (5.1 eV) is very close to the HOMO level of pentacene (5.0 eV), when Pd is contacted with pentacene, the effective work function of Pd is significantly lowered due to large dipole barrier formation at the Pd/pentacene interface known as the “push back effect”.²⁴ As a result, the effective work function of the Pd electrodes is significantly reduced, giving rise to a large Schottky barrier for hole injection at the electrode/organic interface. Therefore, despite this high work function, the charge injection at the Pd/pentacene interface is affected by a high injection barrier. In contrast, due to a strong π-π interaction existing at the graphene/pentacene interface (organic-organic interface), significant dipole formation does not occur, yielding a lower Schottky barrier at the graphene/pentacene interface compared to the metal-pentacene interface.²⁵ This picture is also supported by the morphological consideration of pentacene. A significant difference in the surface morphology of “pentacene on Pd” and “pentacene on graphene” is expected due to differences in molecular packing. The AFM data (Fig. S4 in the supplementary material) show that the morphology in the channel is similar for all the devices, while the morphology on the contacts is different for different electrodes. So, the difference in device characteristics is not due to the morphology in the channel; rather, it is due to the differences at the contact interface or the contact itself. Lee et al.⁴ showed that pentacene has a step-up orientation perpendicular to the graphene when it has PMMA residues, while on clean graphene, it has a lying down orientation parallel to the graphene. By comparing our AFM morphology (Fig. S4 in the supplementary material) with the literature (Ref. 4), we believe that the pentacene on graphene in our device has a step-up orientation. This is consistent with XPS data of CVD graphene which show the C-OH group possibly coming from PMMA residues, which favors the step-up orientation of pentacene. The step-up orientation is better for charge transport as pentacene on SiO₂ also has a step-up orientation, and the monolayer graphene films ensure a smooth and continuous interface between the electrode and the channel. Similarly, pentacene on RGO also have a stand-up orientation, which will result in a continuous interface.^1,3 On polycrystalline metal surfaces, however, the first few monolayers of pentacene are expected to be in parallel lying down the orientation followed by randomly oriented structures.²⁶ This gives rise to a discontinuity at the interface, giving rise to a large barrier. In addition, there are interface dipoles present in the metal pentacene junction. We note that enhanced charge injection due to the pi-pi interaction was also reported in carbon nanotube (CNT) contacted OFET devices,^27–33 with one study demonstrating the lower Schottky barrier for CNT contacted OFETs.³⁴ 

However, the performance differences between CVD graphene/pentacene and RGO/pentacene devices cannot be explained from the work function or interface consideration since the work functions are almost similar for both CVD graphene and RGO;^2,4,8,10 rather, the conductivity of the electrodes needs to be considered.⁸ RGO has many functional groups (Fig. S3 in the supplementary material), yielding a lower sp² carbon fraction and a significantly lower conductivity than CVD graphene (Fig. 2). Therefore, although RGO has a similar work function as CVD graphene, it may not have enough charge carriers for injection from the electrodes to the organic semiconducting layer in comparison to CVD graphene electrodes. Thus, the higher conductivity of CVD graphene is likely the reason why the CVD graphene/pentacene devices show much better performance than RGO/pentacene devices.

In conclusion, we presented a comparative study on the performance of pentacene OFETs using CVD graphene and RGO electrodes. The large area CVD graphene and RGO films were patterned into electrodes using a technique where sacrificial polymer layers were used for a simple, acid free patterning technique. This patterning technique can also be used for other 2D materials and carbon nanotube films. From the electrical transport measurements of the OFETs fabricated using the patterned graphene electrodes, we found that both the CVD graphene and RGO electrodes showed enhanced device performance compared to metal electrodes; however, the maximum performance enhancement was obtained from CVD graphene electrodes. Since the CVD graphene and RGO electrodes have a matching work function with pentacene, we attribute the performance enhancement of CVD graphene electrode contacted OFETs to the higher conductivity of CVD graphene. Our study suggests that, in addition to the strong π-π interaction at the graphene/organic interface, the higher conductivity of the electrodes also plays an important role in the performance of OFETs.

See supplementary material for (a) the characterization of CVD graphene, (b) the preparation of RGO, (c) the characterization of RGO films, (d) the AFM investigation of pentacene films, (e) low bias I-V curves, (f) threshold voltage analysis, and (g) on and off current analysis.

This work was supported by the U.S. National Science Foundation (NSF) under Grant No. ECCS 1102228. C.W.S. and M.I. were supported in part by the National Science Foundation under Grant No. 0955625. We thank Kirk Scammon, Jesse Thomson, and Daeha Joung for help with CVD graphene XPS data.