Single-layer graphene on Al₂O₃/Si substrate: better contrast and higher performance of graphene transistors

Abstract

The fact that single-layer graphene can be visualized on 300 nm SiO₂/Si substrate using optical microscope has enabled the facile fabrication of single-layer graphene devices for fundamental studies and potential applications. Here we report an Al₂O₃/Si substrate for the fabrication of graphene devices with better contrast and higher performance. Our studies show that the contrast of single-layer graphene on 72 nm Al₂O₃/Si substrate is much better than that of single-layer graphene on 300 nm SiO₂/Si substrate. Moreover, the transconductance of single-layer graphene transistors on Al₂O₃/Si substrate shows a more than 7-fold increase, due to the smaller dielectric thickness and higher dielectric constant in 72 nm Al₂O₃ film. These studies demonstrate a new and superior substrate for the fabrication of graphene transistors, and are of significance for both fundamental studies and technological applications.

1. Introduction

Graphene is a one-atom-thick planar sheet of sp²-bonded carbon atoms that are arranged in a honeycomb crystal lattice. As a gapless semiconductor, charge carriers in graphene can be tuned continuously from electrons to holes, crossing the charge neutral Dirac point using an external electric field [1–7]. Graphene-based electronics is currently the subject of intense focus for nanoscale electronic applications due to its exceptional electronic properties including carrier mobilities exceeding 100,000 cm²/V·s and a micrometer-scale mean free path at room temperature [8–10].

Most efforts to date employ a silicon substrate as a global back gate and a 300 nm thick silicon dioxide (SiO₂) as the gate dielectric. This substrate is used primarily because the graphene can be readily visualized using optical microscope on this particular substrate due to optical interference [11–13]. The readily viewable graphene on the 300 nm SiO₂ substrate has allowed easy device fabrication and led to many interesting scientific discoveries. The requirement of 300 nm SiO₂, however, can intrinsically limit the resulted device performance in several perspectives. First, optical detection technique to visualize graphene has been demonstrated and widely used only for a SiO₂ thickness of 300 nm, but a 5% variation in thickness (e.g. to 315 nm) can significantly lower the contrast [11,14]. Second, the devices made on 300 nm SiO₂ substrate often have too small gate capacitance due to relatively large thickness and small dielectric constant, and thus require high switching voltage.

Recently, it has been predicted that the 72 nm Al₂O₃ film can be much better for visualizing graphene than SiO₂ and Si₃N₄ films by total color difference calculation [14,]. Moreover, Al₂O₃, with a dielectric constant of 9.1, is an important high-k materials with excellent dielectric properties, thermal and chemical stability [15]. Therefore, higher performance graphene transistors may be readily fabricated on 72 nm Al₂O₃/Si substrate. Here we report the integration of graphene on 72 nm Al₂O₃/Si substrate and investigation of the optical contrast and device performance. Compared to 300 nm SiO₂/Si substrate, 72 nm Al₂O₃/Si substrate could enhance the normalized optical contrast of single-layer graphene (SLG) by approximately 3 times. Furthermore, higher performance of graphene transistors can be achieved on 72 nm Al₂O₃/Si substrate with over 7-fold increase in transconductance.

2. Experimental

Al₂O₃ film was deposited on top of heavily-doped p-type silicon (100) wafers by an atomic layer deposition (ALD) technique at 250 °C using trimethylaluminum and distilled water as the source. Prior to the deposition of the Al₂O₃ layer, the native oxide layer on the Si wafer was removed with buffered oxide etching. After 72 nm Al₂O₃ film was deposited, graphene layers were mechanically peeled onto the substrate. As a control, graphene was also peeled onto standard 300 nm SiO₂/Si wafers. The substrates are then calcined at 300 °C to remove organic residue. The morphologies of the samples were characterized by atomic force microscope (AFM, Veeco Dimension 5000) and optical microscope (Olympus), respectively. The Raman spectra were recorded by a Renishaw inVia Raman spectroscopy. The thickness and refractive index of Al₂O₃ film were obtained by Sopra GES5 Ellipsometer.

Metal-insulator-semiconductor (MIS) devices were used to investigate the dielectric properties of the Al₂O₃ film. The MIS devices were made on Al₂O₃/Si substrate with circular electrodes of 100 μm in diameter, fabricated using e-beam lithography followed by e-beam metal (Ti/Au, 50/50 nm) deposition. The SLG field-effect transistors (FETs) were also fabricated by the standard e-beam lithography and e-beam evaporation (Ti/Au, 50/50 nm) process. The on-off ratio of SLG FETs is usually very low because the intrinsic SLG is a semimetal with zero band-gap. It has been demonstrated that a band gap can open up by fabricating graphene nanostructures of confined geometry such as graphene nanoribbons (GNRs)[16–21]. To this end, we have also fabricated GNR FETs on Al₂O₃/Si substrate with a recently reported procedure using chemically defined nanowires as the nanoscale etching mask [19]. The electrical transport properties were measured by a Lakeshore probe station with a home built data acquisition system. Capacitance vs voltage (C-V) measurements were performed using a Stanford Research Systems SR720 LCR Meter.

3. Results and Discussion

An AFM image of Al₂O₃ film on Si substrate indicates that the root mean square roughness for Al₂O₃ on Si substrate is less than 0.5 nm (Fig. 1a). To understand the dielectric properties of the Al₂O₃ film, metal-insulator-semiconductor (MIS) devices were used to characterize the capacitance-voltage, current tunneling and breakdown characteristics. The C-V plot for the 72 nm Al₂O₃ film (Figure 1b) shows an accumulation of charges in the silicon substrate in negative polarity and a depletion of charges in the positive polarity, consistent with p-type substrate used. The dielectric constant can then be calculated using: ε = Cd/ε₀A, where C is the insulate capacitance (highest capacitance in the C-V curve), d is thickness of the film, ε₀ is the permittivity of free space, and A is the area. The obtained dielectric constant is 7.5, consistent with previous reports on ALD Al₂O₃ films [15]. Current-voltage (I-V) measurements of the MIS device show typical Fowler–Nordheim (F-N) tunneling behaviour with a breakdown field of 6.7 MV/cm (Fig. 1c and 1d), comparable to the best quality ALD Al₂O₃ films [15]. This type of field assisted tunneling can be described by charge carriers tunneling through a triangular barrier with [15]:

$$J={AE}_{OX}^{2}exp(-B/E_{OX})$$ (1)

where

$$A=1.54\times {10}^{-6}\left\{\frac{1}{m^{\ast }{\mathrm{\Phi }}_B}\right\}$$ (2)

and

$$B=6.83\times {10}^7{(m^{\ast })}^{1/2}{({\mathrm{\Phi }}_B)}^{3/2}$$ (3)

Figure 1.

(a) AFM image of Al₂O₃ film by ALD. (b) Capacitance vs voltage (C-V) curve of an MIS device with 72 nm Al₂O₃ film on Si substrate. (c) Current-voltage (I-V) curve of the MIS device made from Al₂O₃ film, and (d) the corresponding Fowler–Nordheim (F-N) curve.

J is current density, E_ox is the oxide electric field, m^* is the effective mass of the charge carrier, which is about 0.23, and Φ_B is the barrier height. Fitting the I-V characteristics with F-N tunnelling model gives a tunnel barrier of about 2.2 eV between Al₂O₃ and Ti, comparable to previous reports of the barrier height between ALD Al₂O₃ and metals of similar work function [15, 22]. These studies clearly demonstrate that the quality of our Al₂O₃ film is good enough as high-k gate dielectrics [15].

Next we discuss the optical contrast of SLG on Al₂O₃ films. Optical contrast is the difference in visual properties that enable us to distinguish an object from other objects and the background. Remarkably, when a proper substrate is used, single layer or few layers of graphene can be observed with sufficient contrast using an normal optical microscope [23]. The origin of the contrast can be explained by Fresnel’s equations [11]. Figure 2a and 2b show the optical image of SLG samples on Al₂O₃/Si and SiO₂/Si substrate. The relative Raman spectra prove that samples are perfect SLG (Figure 2e). Importantly, the SLG on Al₂O₃/Si is much better visualized than that on SiO₂/Si substrate. To quantify the difference in the contrast of the SLG samples on Al₂O₃/Si and SiO₂/Si substrates, the color images are converted to gray-scale images with same normalized background intensity shown in Figure 2c and 2d. Indeed, the contrast intensity of SLG on Al₂O₃/Si is significantly stronger than that of SLG on SiO₂/Si. The Weber contrast C(λ) is defined as C(λ) = (R₀(λ)−R(λ))/R₀(λ) [11, 23], where R₀(λ) is the reflected light intensity of background and R(λ) is the reflected light intensity from graphene sheet. The contrast of SLG on Al₂O₃/Si is 0.070, which is 3 times larger than that of SLG on SiO₂/Si with 0.023 in our system (Figure 2c and 2d). Considering the incident light from air (n = 1) onto a graphene, SiO₂ or Al₂O₃, and Si trilayer system, the reflected light intensity from the trilayer system can be described by [11, 23]:

$$R(\lambda )=r(\lambda )r^{\ast }(\lambda )$$ (4)

$$r(\lambda )=\frac{r_a}{r_b}$$ (5)

$$r_a=(r_1{e}^{i({\beta }_1+{\beta }_2)}+r_2{e}^{-i({\beta }_1-{\beta }_2)}+r_3{e}^{-i({\beta }_1+{\beta }_2)}+r_1{r}_2{r}_3{e}^{i({\beta }_1-{\beta }_2)})$$ (6)

$$r_b=(e^{i({\beta }_1+{\beta }_2)}+r_1{r}_2{e}^{-i({\beta }_1-{\beta }_2)}+r_1{r}_3{e}^{-i({\beta }_1+{\beta }_2)}+r_1{r}_2{r}_3{e}^{i({\beta }_1-{\beta }_2)})$$ (7)

where r₁ = (n₀−n₁)/(n₀+n₁), r₂ = (n₁−n₂)/(n₁+n₂), and r₃ = (n₂−n₃)/(n₂+n₃) are the reflection coefficients for different interfaces and β₁ = 2πn₁(d₁/λ) and β₂ = 2πn₂(d₂/λ) are the phase differences when light passes through the media which is determined by the path difference of two neighboring interfering light beams. The calculated contrast spectra of SLG are shown in Figure 2f. The best contrast of graphene on Al₂O₃/Si could be obtained with 450 nm light illumination. Moreover, there is moderate contrast intensity in whole visible light wavelength, in contrast to the SiO₂/Si system where relatively strong contrast is only observed in the red region.

Figure 2.

(a) and (b) Optical image of graphene on 300 nm SiO₂/Si and 72 nm Al₂O₃/Si substrate. The dashed lines highlight the edge of the single layer graphene region. (c) and (d) the gray-scale optical image from (a) and (b). The arrows highlight the edge of the single layer graphene region. (e) Raman spectra of the single-layer graphene on 300 nm SiO₂/Si and 72 nm Al₂O₃/Si substrates. (f) The contrast spectra of single-layer graphene 300 nm SiO₂/Si and 72 nm Al₂O₃/Si substrates.

The excellent dielectric properties observed in Al₂O₃ film and the better optical contrast of SLG on Al₂O₃ film readily allows us to fabricate graphene transistors on Al₂O₃/Si substrate. To compare with previous studies, we adopted a similar back gated structure where Al₂O₃ is used the gate dielectrics and Si substrate used as the back gate. Similar devices are also fabricated on the standard SiO₂/Si substrate as a control. A schematic device diagram and an SEM image of the graphene FET are shown in Figure 3a. To evaluate and compare the device performance for devices based on SiO₂ and Al₂O₃, we have measured the transfer characteristics (Drain-source current I_ds versus gate voltage V_g). The normalized transconductance $g_m=\frac{{dI}_{ds}}{{dV}_g}·\frac{L}{W}$ can be extracted from the linear region of the I_ds−V_g curve. Significantly, the I_ds−V_g data show a significant improvement in transconductance when the Al₂O₃ is used (Fig. 3b). Figure 3c shows the measured normalized transconductance of the SLG FET as a function of gate voltage V_g. At V_ds = 0.1V, the Al₂O₃ bottom-gated device exhibits a g_m of about 22.1 μS, which is around 7 times larger than that of the SiO₂ back-gated device (g_m ~ 3.1 μS). This substantial improvement can be attributed to a much stronger gate coupling due to smaller thickness and higher dielectric constant material used in the this devices (72 nm Al₂O₃ with k = 7.5 vs. 300 nm SiO₂ with k = 3.9). In order to evaluate the mobility of the devices, a device model was used [24, 25]. The extracted carrier mobility of SLG FET on Al₂O₃ is 7400 cm²/V·s, which is similar to 8200 cm²/V·s of the FET on SiO₂.

Figure 3.

(a) SEM image of a graphene field-effect transistor on a 72-nm Al₂O₃/Si substrate, and the inset shows schematics of the device. (b) I_ds−V_g curves of graphene FET on 300 nm SiO₂/Si and 72 nm Al₂O₃/Si substrate, (c) g_m−V_g curves of graphene FET on 300 nm SiO₂/Si and 72 nm Al₂O₃/Si substrate.

To further increase the on-off ratio of the device, we have also fabricated GNR devices on Al₂O₃/Si substrate and studied the device performance. Here a SiO₂ nanowire was used as a nanoscale etch mask to define a graphene nanoribbon with a width in the 10–20 nm regime and the schematic device diagram of the GNR FET is shown in Figure 4a. The Figure 4b plots the I_ds−V_ds of the top-gated device under various gate voltages (V_g) of −8, −6, −4, −2, 0, and 2V. It shows clearly that the conductance of the nanoribbon decreases as the gate potential increases, demonstrating that the graphene nanoribbon is p-type doped, which can be attributed to edge oxidation or the physisorbed O₂ from ambient or during the device fabrication process. Transfer characteristics show the GNR transistor on Al₂O₃ dielectric can be switched on and off with approximately 6 volts of gate swing (Fig. 4c), in contrast to ~30 V required for SiO₂ back gated devices (inset, Fig. 4c) and 10–40 V for previously reported devices on SiO₂ [18–20], and the transconductance g_m of about 3.2 μS, also more than 9 time larger than that of the back-gated GNR transistor on SiO₂/Si (~0.36 μS) (Fig. 4c). The device shows a room temperature on/off ratio of ~10 at V_ds = 0.1 V, which is consistent with a graphene nanoribbon with an estimated width of 10–15 nm.

Figure 4.

(a) The schematics of the GNR FET device. (b) I_ds−V_ds curves of a GNR-FET on 72 nm Al₂O₃/Si substrate. (c) I_ds−V_g curves of the GNR-FETs on the 72nm Al₂O₃/Si substrate and the 300 nm SiO₂/Si substrate (inset).

4. Conclusions

In summary, we have demonstrated that 72 nm Al₂O₃/Si substrate can be a superior substrate to 300 nm SiO₂/Si substrate for the visualization of graphene and fabrication of graphene transistors. Compared to 300 nm SiO₂/Si substrate, 72 nm Al₂O₃/Si substrate could enhance the optical contrast of SLG by at least 3 times. Furthermore, using the Al₂O₃ film as the gate dielectrics, the back-gated graphene FETs have been fabricated to exhibit more than 7-fold increase in transconductance. This study demonstrates a new and superior substrate for the fabrication of graphene transistors, and is of significance for both fundamental studies and technological applications.