Abstract

Microelectronic fabrication of Si typically involves high-temperature or high-energy processes. For instance, wafer fabrication, transistor fabrication, and silicidation are all above 500°C. Contrary to that tradition, we believe low-energy processes constitute a better alternative to enable the industrial application of single-molecule devices based on 2D materials. The present work addresses the postsynthesis processing of graphene at unconventional low temperature, low energy, and low pressure in the poly methyl-methacrylate- (PMMA-) assisted transfer of graphene to oxide wafer, in the electron-beam lithography with PMMA, and in the plasma patterning of graphene with a PMMA ribbon mask. During the exposure to the oxygen plasma, unprotected areas of graphene are converted to graphene oxide. The exposure time required to produce the ribbon patterns on graphene is 2 minutes. We produce graphene ribbon patterns with ∼50 nm width and integrate them into solid state and liquid gated transistor devices.

1. Introduction

Working with 2D materials such as graphene requires novel methods to fabricate ribbon patterns. Among the traditional methods are a metallic or resist mask to selectively protect graphene in plasma etch exposure [15] and focused ion-beam (FIB) etching [6, 7]. In an oxygen plasma at 200 mTorr and 50 Watts, the etch rate of graphene is about 1 layer per second [8], and a 5- to 10-second plasma etch exposure is typically employed to selectively etch graphene with a hydrogen silsesquioxane (HSQ) resist [8] or metallic mask [9]. At a shorter time (<4 seconds) of plasma etch exposure, graphene oxide can be generated [10]. Major drawbacks of the traditional methods are the lack of adaptability of FIB for mass production of devices, the usage of harsh acid treatment to remove the HSQ resist [5] or metal mask [9, 11], and overetching of graphene from the edges underneath the metallic ribbon mask [9]. With a HSQ ribbon mask, the resultant width of graphene ribbon pattern is ∼10 nm smaller than the resist mask [8]. A polymethyl-methacrylate (PMMA) resist mask can also be employed to pattern graphene, and the edge roughness of the resultant pattern is ∼5 nm [3]. The PMMA mask has been adopted in the fabrication of graphene constrictions or quantum dots [3,1214] where they exploit the formation of “S” shaped edges, which is possibly due to the strong and normally undesired effect of etching from the edges. Despite its adoption to fabricate quantum dots, the PMMA resist mask has not been widely adopted to make patterns on graphene. A metallic mask is preferred instead of PMMA to make nanoscale patterns with widths or diameters smaller than 50 nm [9, 11, 15].

Novel methods to fabricate ribbon patterns include metallized DNA origami [16, 17], inorganic nanowire mask [18, 19], PMMA as sacrificial layer for a metallic mask [20, 21], and block-copolymer lithography [22, 23]. Major drawbacks of the novel methods are the lack of control on the creation of arbitrary patterns with block copolymer, obtaining sub-100 nm resolution with a PMMA sacrificial layer, and the lack of control of placement of a DNA nanostructure or nanowire on the graphene substrate.

On the other side, plasma etch exposure is also useful to reduce the number of layers in multilayer graphene down to single layer [2427], to control the electronic bandgap of graphene by chemical functionalization [28, 29], and for controllably engineering atomically thin material systems with monolayer precision [3032].

In the present work, novel parameters for plasma patterning of graphene/graphene oxide are identified to enable the usage of a PMMA ribbon mask cured at low temperature (<115°C). Low-temperature curing is unconventional in electron-beam lithography and PMMA typically shows low resistance to oxygen plasma etching [33]. PMMA is therefore normally considered inadequate as a mask for selective plasma etching at sub-50 nm resolution.

2. Materials and Methods

Graphene (Figure 1) was synthesized by chemical vapor deposition (CVD) on copper foil [3436]. Impurities in the CVD chamber can be translated to the graphene sample, and thus, an oxygen plasma treatment of the chamber was completed before introducing the copper foil sample for CVD synthesis of graphene. Briefly, the CVD process began by annealing the copper foil at 800°C under a flow of hydrogen gas of 1000 sccm at 300 Torr for 30 minutes. Then, the temperature was set to 1000°C. When the temperature stabilized at 1000°C in all three monitored zones of the CVD furnace, hydrogen gas was purged at 0.1 Torr before exposing the copper foil to a flow of methane of 50 sccm at 1 Torr. In this process, graphene was deposited by the reduction of methane. After 3 minutes of methane flow, the gas was purged at 10 mTorr and the furnace heater was shut off to begin the cooling process. The cooling process was done initially under flows of hydrogen and argon of 1000 sccm each at 300 Torr for 5 minutes. For the rest of the cooling process (∼75 minutes), only the hydrogen gas was kept active. The introduction of argon during the initial stages (5 minutes at 1000°C) of cooling yields a D peak (∼1300 cm−1) of increased intensity and a 2D peak (∼2650 cm−1) of reduced intensity in the Raman characteristics of graphene (Figure 1), which is consistent with reports of graphene with a high degree of disorder [3739].

A PMMA (MicroChem) thin film, produced by spin coating, protected the front side of the graphene (a) during the etching of graphene on the backside of the copper foil in a dilute nitric acid solution (Table 1); (b) during the transfer of graphene to an oxide wafer (Table 2); and (c) during the oxygen plasma patterning of graphene ribbons on silicon dioxide (Figure 2).

The curing temperature of PMMA on graphene/copper foil and the temperature for drying of PMMA/graphene on the oxide wafer were either room temperature (21°C) in a vented hood (∼12 hours) or 37°C on a hot plate (∼3 hours).

After the transfer of graphene from the copper foil to an oxide wafer, we need to characterize the graphene through electrical measurements; this requires the fabrication of a graphene ribbon with electrodes, for which four lithography steps were implemented:(a)Patterning on top of the oxide wafer (SiO2/Si++) and before graphene deposition by direct-write optical lithography (Microtech LW405) of back gate electrodes, as interface to the highly doped silicon (Si++).(b)Patterning of drain-source (D-S) electrodes and two auxiliary (Aux) electrodes after graphene deposition by optical lithography. Graphene covers most of the surface of the oxide wafer (∼1 cm2), and (D-S-Aux) electrodes were deposited on top of graphene. Back gate electrodes are employed for electrical measurements in air conditions. Auxiliary electrodes are employed as gate electrodes for electrical measurements in aqueous conditions.(c)Etching away by a selective exposure to oxygen plasma with a photoresist mask of the areas surrounding the electrodes (except the zone between drain and source) to avoid electrical short circuit. Graphene has a good electrical conductivity and may cause short circuits between electrodes.(d)Defining the ribbon pattern between drain and source electrodes by a selective exposure to oxygen plasma with a PMMA (e-beam resist) ribbon mask.

Electrodes were based on palladium, ∼30 nm thick, and deposited by sputtering (AJA sputtering system). S1805 (MICROPOSIT S1800, film thickness ∼0.5 µm) was chosen as photoresist; thicker resists left more residue impurities on graphene. The resist was spin coated at 3000 rpm for 30 seconds and cured at 90°C or 115°C for 1 minute. Acetone was employed as a resist remover in the liftoff process. 1165 solvent is the conventional remover of Shipley resists. However, we found that it causes detachment of graphene from the oxide substrate.

In order to perform the electron-beam (e-beam) lithography (Raith 150 e-beam system), the PMMA resist was cured on a hot plate and we tested three temperatures: room temperature (21°C for 12 hours), 37°C (3 hours), and 115°C (2 minutes). We did not observe significant differences in the outcome of the plasma etch due to the curing temperature of PMMA resist. IPA/water (7 : 3) was used as the developer of PMMA [4143] ribbon patterns at room temperature (21°C). The optimal development time of PMMA in IPA/water was different for different curing temperatures: ∼10 s at room temperature, ∼15 s at 37°C, and ∼20 s at 115°C. We chose PMMA A2 950K, which originates films with ∼60 nm thickness and therefore produces PMMA nanoribbons with good aspect ratio and stability since we need to fabricate nanoribbon patterns of width <100 nm. The optimal development time depends on the curing temperature, the chosen optimized e-beam dose parameters, and the resultant e-beam current during patterning.

Optimal parameters for the e-beam processing of PMMA nanoribbons on graphene included a high-voltage source of 30 kV, aperture of 10 µm, magnification of 2000, working distance of 10 mm, step size of 4 nm, area dwell time of 0.372 μs, and beam current of 34.38 pA. We test and suggest the following doses: 80, 100, and 120 µC/cm2.

The graphene/PMMA ribbon mask was exposed to an oxygen plasma at low pressure (<40 mTorr) in a reactive ion etching (RIE) chamber (MARCH CS-1701) (Figure 2). Sometimes, there were delays of ∼5–25 seconds in the activation of the plasma at low pressure. If the delay was longer, we turned off the plasma. Before reactivating the plasma, (a) we activated the vacuum pump to evacuate the gas; (b) set the oxygen gas pressure to 160 mTorr at 20 sccm; (c) activated again the vacuum pump to evacuate the gas; and (d) set the oxygen gas pressure to 30 mTorr at 1 sccm.

3. Results and Discussion

Polymer or metal masks are typically used to etch graphene. Physical vapor deposition of the metal mask inherently involves high-temperature molecular events and strong binding of the metal to graphene. Therefore, we opted for a PMMA-based mask, a polymeric material employed in e-beam lithography as a resist. However, PMMA is well known to have a very low resistance to plasma oxygen [33]. For example, a double layer of PMMA (PMMA 495K A2/PMMA 950K A4) can be employed to pattern a 65 nm width graphene ribbon [15], but a metallic mask is preferred to obtain smaller ribbons with widths smaller than 50 nm [9, 11, 15]. Therefore, different etch processing parameters are needed to enable PMMA as a nanoribbon mask for 50 nm or smaller widths. To find optimal parameters for the effective plasma patterning of a graphene nanoribbon, we analyze the events involved in the selective etching process of a 2D material. During the plasma etching process, at certain conditions of RF power and gas pressure, an atom of PMMA ejected at high kinetic energy can remove other PMMA atoms, this process constituting a chain reaction. The effect of this chain reaction on the material depends on the number of atoms of the material and, therefore, will have a stronger effect in the etching resistance of a 3D material (PMMA thin film) than in a 2D material (graphene). We suggest this is an important factor for which PMMA can show an etching resistance to oxygen plasma as poor as that of graphene. Therefore, to find conditions of improved etching resistance for PMMA and for selective patterning of graphene, we should decrease the energy and the number of oxygen ions that initiate the chain reaction. Our aim was to find the lowest power, the lowest oxygen pressure, and the shortest time needed to etch graphene.

Firstly, we explored much lower levels of power to test the plasma etch resistance of graphene. We hypothesized that the strength of adhesion of graphene to the substrate can influence the etch resistance of graphene. We do not report the influence of different values of temperature (during graphene transfer or resist curing) on the resistance of graphene to plasma etching. However, we found that baking dry graphene/SiO2 at 150°C or 180°C makes it significantly more difficult to remove graphene by exposure to oxygen plasma, so that higher power (>50 Watts) is needed to etch graphene. Working with temperature levels lower than 115°C at all stages of the fabrication process, graphene has a lower strength of adhesion to the oxide wafer and is easier to etch.

We found that 8 Watts is the minimum level of power needed to effectively pattern graphene in a reproducible way; at this level of power, our graphene sample on 300 nm thick SiO2 becomes optically transparent after ∼2 minutes of plasma exposure (oxygen, 20 sccm, 160 mTorr).

The next step is finding an adequate level of plasma pressure to optimize the etch resistance of PMMA. The lowest pressure at which the RIE instrument works is typically 40 mTorr, which is the level of pressure used to evacuate the gas present in the chamber before beginning a plasma etch process. However, by setting a lower pressure (30 mTorr) and a flow of gas oxygen of 1 sccm, we were still able to produce plasma. We found that the oxygen plasma produced at a minimum power of 8 Watts, 30 mTorr, and 1 sccm converts graphene to graphene oxide in two minutes. The structural change was noticeable by light microscopy and verified by Raman spectroscopy (Figure 3). An increase in the magnitude of the D peak (∼1300 cm−1), becoming larger than the G peak (1530 cm−1), in the Raman characteristic of graphene ribbons (Figure 3) with respect to that of pristine graphene (Figure 1), agrees with previous reports on the patterning of graphene ribbons; that is, the increase in the D peak is obtained during the process of patterning [15, 44, 45].

After we found low power (8 Watts) and low pressure (∼30 mTorr) parameters of oxygen plasma to effectively etch graphene, we needed testing the etching resistance of the PMMA ribbon masks on an oxide substrate (SiO2) as well as on graphene/SiO2. Thus, we fabricated PMMA ribbon masks of three different widths: 20, 50, and 100 nm. We were able to produce 20 nm width PMMA ribbons on an oxide substrate (Figure 4), but we were not able to produce them on graphene. We suggest, at very small width (20 nm), the PMMA ribbon mask detach from graphene during the development process. We were able to produce 50 nm (Figure 5) and 100 nm (not shown) width PMMA on an oxide substrate (SiO2) as well as on graphene. Notice that the development process of PMMA nanoscale patterns is normally done at low temperature (–4°C) [46, 47]. However, in the present work, the development process was done at room temperature (21°C).

When PMMA is supported on SiO2, all PMMA ribbon widths were able to withstand an exposure to the oxygen plasma (30 mTorr, 1 sccm, 8 Watts) for 2 minutes. By SEM (scanning electron microscopy), we did not observe a significant change in the width of PMMA ribbons after exposure to oxygen plasma at those conditions (Figure 4). Moreover, we found that 16 Watts is the minimum power needed to remove PMMA at 30 mTorr, 1 sccm, and 2 minutes.

Finally, when PMMA is supported on graphene/SiO2, we tested the plasma etch (oxygen, 8 Watts, 30 mTorr, 2 min) with ribbon masks of 50 nm (Figures 5(a) and 5(b)) and 100 nm width (not shown) and characterized the resultant graphene ribbon patterns using SEM imaging (Raith 150 e-beam system) and electrical measurements (HP 4145A) (Figures 5(c)5(f)).

The current-voltage (IdsVg) characteristic at air conditions (Figure 5(e)) is similar to other reports of graphene ribbon electron devices. Han [8] reported a Imax/Imin value around ∼1.6 (Vg = 0–20 Vdc) for graphene ribbon with widths of 49 and 71 nm at 200 K. While Jeong et al. [45] reported a Imax/Imin value of ∼3.1 (Vg = 0–20 Vdc) for an array of graphene ribbons with sub-10 nm width at room temperature. In the present report, the Imax/Imin is around ∼1.4 (Vg = 0–20 Vdc) for graphene ribbons with width of 50 or 100 nm at room temperature. At lower temperature and smaller ribbon width, the Imax/Imin is expected to increase. For example, for a very small ribbon width of only 9 to 13 atoms width, the Imax/Imin is around ∼1000 (Vg = 0–20 Vdc) [48].

In the IdsVg characteristic in water, the gate modulation is more effective (Figure 5(f)). An electric double layer formed at the interface of an aqueous solvent and graphene has been suggested as a dielectric that allows a more effective gate modulation of graphene [4951]. The gate modulation in water solvent allows a Imax/Imin of ∼2.2, similar to the one obtained in air (Vg = −20 to +20 Vdc) but with a smaller gate voltage (Vg = 0 to 5 Vdc) applied.

4. Summary and Conclusions

A high-temperature treatment of the PMMA/graphene membrane can cause a poor or null yield of graphene transfer from the copper foil to the oxide wafer. We attribute this to a strong adhesion of PMMA to graphene, after curing on copper foil, and a shape memory effect of the polymer, which causes poor adhesion of the PMMA/graphene membrane to the oxide substrate during the drying process. We were able to overcome that obstacle with low-temperature curing of PMMA and low-temperature drying at 21 or 37°C. A low level of temperature for resist curing (<115°C) was also important to enable the usage of PMMA nanoribbons as a mask during the oxygen plasma patterning of graphene. Normally, under exposure to oxygen plasma, a PMMA nanoribbon mask degrades faster than graphene. We suggest the plasma etching process in a PMMA nanoribbon mask, a 3D material, is critically dependent on the effect of a chain reaction, and this effect is weaker when the plasma is set to low levels of pressure and energy. We showed that a low temperature of PMMA curing (<115°C) combined with low oxygen gas pressure (∼30 mTorr) and low power (∼8–12 Watts) produced ribbon patterns on graphene using a PMMA mask (∼50 nm width, ∼60 nm thick) in a reactive ion etching chamber (MARCH CS-1701). We did not analyze the role of intrinsic defects in the CVD graphene, which can play an important role in the plasma etch process. The method could be adapted to other types of polymer or UV resists. Other types of gases or chemical functionalization could also be explored to produce nanoribbon patterns on 2D materials.

Data Availability

The data used to support the findings of this study are included within the article.

Disclosure

The submitted manuscript has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory (“Argonne”). Argonne, a U.S. Department of Energy Office of Science laboratory, is operated under Contract DE-AC02-06CH11357. The U.S. Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the government.

Conflicts of Interest

The authors declare no conflicts of interest exist.

Acknowledgments

The Center for Nanoscale Materials was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract DE-AC02-06CH11357. The authors also acknowledge financial support from Argonne National Laboratory’s Laboratory-Directed Research and Development Strategic Initiative.