High Quality Monolayer Graphene Synthesized by Resistive Heating Cold Wall Chemical Vapor Deposition

Abstract

The growth of graphene using resistive-heating cold-wall chemical vapor deposition (CVD) is demonstrated. This technique is 100 times faster and 99% lower cost than standard CVD. A study of Raman spectroscopy, atomic force microscopy, scanning electron micro­scopy, and electrical magneto-transport measurements shows that cold-wall CVD graphene is of comparable quality to natural graphene. Finally, the first transparent flexible graphene capacitive touch-sensor is demonstrated. Chemical vapor deposition (CVD) of monolayer graphene on copper1, 2 has emerged as one of the most competitive growth methods for securing the industrial exploitation of graphene, due to its compatibility with Si and roll-to-roll technologies.3 Recently, there has been tremendous progress in controlling the morphology,4-6 functionalization,7-10 and growth of heterostructures of intrinsic and doped graphene.11 However, the low-throughput and the very high production cost for high-quality CVD graphene are central challenges for the industrial exploitation of this material.12, 13 The most common CVD approach is to use a hot-wall system where Cu foils are heated at temperatures ≈1000 °C in a quartz tube furnace through which the precursor hydrocarbon gas flows. The long processing time, that can take a few hours, limits the throughput of graphene by this method. At the same time the typical cost of graphene produced in this way is in excess of 1 cm−2, whereas its retail price ranges from 4.57 cm−2 to 21 cm−2 (see the Supporting Information). Therefore, a way forward to increase the throughput and reduce the production cost is to grow graphene in a cold wall CVD system which heats selectively only the Cu foils. Few types of cold wall CVD have been investigated so far for the growth of graphene3, 14-19 such as magnetic induction heating CVD,14 rapid thermal annealing CVD using halogen lamp heating,15, 16 Joule heating CVD,17, 18 and resistively heated stage CVD.19 Of all these methods, the resistively heated stage CVD approach allows for faster, more efficient heating and cooling, shorter growth time, and less gas consumption. This method provides a more uniform substrate heating, it reduces the chemical reactions which can take place in the gas phase at high temperature known to contaminate graphene and it allows for very fast cooling rates, which have been shown to enhance the quality of graphene grown by CVD on copper foil.20 Furthermore, this type of cold-wall CVD system is found in manufacturing plants of the semiconductor industries. Most importantly we show that with this method truly high quality monolayer graphene can be reproducibly grown. To date, virtually nothing is known on the growth mechanism of monolayer graphene by cold-wall CVD, as well as on its quality and suitability for flexible electronic applications. Therefore, understanding the growth and properties of graphene obtained with cold-wall CVD is imperative to enable the exploitation of this material and facilitate the birth of novel graphene-based applications. Here we report a completely new mechanism for the growth of graphene by resistively heated stage cold-wall CVD which is markedly different from the growth mechanism of graphene in a hot-wall CVD. Through a combined study of Raman spectroscopy, atomic force microscopy (AFM), and scanning electron microscopy (SEM) we elucidate the early stage formation of graphene by monitoring the transition from disordered carbon adsorbed on Cu to graphene. We also demonstrate for the first time (1) high-throughput production, (2) ultralow cost, and (3) high quality monolayer graphene grown on Cu foils by resistively heated stage cold-wall CVD. Our technique merges short deposition time (approximately few minutes) with high-efficiency heating of a cold-wall CVD system, resulting in ≈99% reduction in graphene production cost. The Raman spectra of our graphene films shows a low defect related peak and in devices with an area of 5600 μm2 fabricated on standard SiO2 substrates we measure a charge carrier mobility of 3300 cm2 V−1 s−1 and the quantum Hall Effect typical of single layer graphene. In contrast, the quality of graphene grown by hot-wall CVD is often gauged only by carrier mobility,1, 2, 4, 5, 21-23 giving little information regarding the large area properties of the film. Therefore, to better quantify the quality of graphene films for electronic applications, we introduce an electronic quality factor (Q) accounting for the area across which the carrier mobility is measured. Using Q as a gauge we show that graphene grown by cold-wall CVD has enhanced quality compared to the material grown by hot-wall CVD. Finally, we demonstrate that graphene grown by cold-wall CVD is suitable for the next generation electronics by embedding it into the first transparent and flexible graphene capacitive touch-sensor that could enable the development of artificial skin for robots. Studies of the growth mechanism of graphene on copper (using methane) in a hot-wall CVD have thus far suggested the direct growth of 2D films involving several steps. The first step is the direct formation of 2D nuclei of graphene24 from the adsorbed carbon species resulted from the catalytic decompo­sition of methane on the copper surface. These graphene nucleation sites subsequently grow with the addition of carbon to their edges to form islands and large domains.25 The growth parameters such as the temperature, pressure, growth time, and gas flow are tuned to let graphene domains grow until they coalesce and a continuous graphene film is attained.26 Though it has been suggested that after the growth of the first layer the catalytic copper surface becomes passivated and limits the growth of other layers, several studies of low-pressure CVD have reported the growth of bilayer27 and trilayer,28 as well as multilayers for atmospheric pressure CVD.29 Nevertheless, the thickness of the grown layers in a hot-wall CVD is always limited to few nanometers or less. Our experiments show that the growth mechanism of graphene in cold-wall CVD is markedly different from that of the hot-wall CVD described above. Specifically, over a range of growth temperatures that we have investigated, we always observed a thick carbon film (100 nm), which forms in the early stages of the growth (see Figure 1a, top left), that becomes progressively thinner with increasing the growth time (see top inset in Figure 1b) and finally evolves into graphene islands (see Figure 1a, top right). The time required to form graphene decreases from 6 min at 950 °C to 20 s at 1035 °C (see the Supporting Information). To elucidate the initial stage of graphene growth, that is the adsorption of carbon on the Cu substrate, we focus on the slow graphene formation at 950 °C. Graphene films were obtained using a commercial cold-wall CVD system (see the Supporting Information for details on the design and stability of critical parameters needed for the growth of high quality graphene with this process). The films were transferred from the Cu foils to SiO2/Si substrates using a wet transfer method.1, 30 Full details of the growth and transfer procedures are provided in the Supporting Information. Similar studies for films grown at higher temperatures are presented in the Supporting Information. Figure 1b shows the Raman spectra of films grown at 950 °C for growth time (tG) ranging from 1 to 6 min. For all the samples we observe the characteristic peaks of sp2 bonded carbon atoms: the D-peak at ≈1340 cm−1, the G-peak around 1600 cm−1, the D′-peak around 1620 cm−1, and the 2D-peak at ≈2700 cm−1. For short tG (i.e., 1–4 min) the D- and G-peaks have considerable higher intensities than the 2D-peak, which is typical of disordered carbon films.31, 32 As tG increases we observe changes in intensities, sharpness, and positions of the D and G peaks, and for tG > 4 min a well-defined 2D-band emerges. At the same time, AFM measurements show a reduction in the film thickness from 116 to 2.7 nm with increasing tG from 1 to 6 min (see top inset of Figure 1b), which suggests the desorption of carbon from the film. Lorentzian fitting of the D-, G-, and 2D-peaks allows us to ascertain the structural ordering within the films by analyzing the band intensities (ID,G,2D), the full width at half maximum (FWHM(D,G,2D)) and the peak positions (Pos(G,2D)). According to the three stage model for classification of disorder,33-37 the evolution of ID/IG, FWHM(D,G) and Pos(G) allow us to assess the ordering/amorphization in carbon materials ranging from graphite and amorphous carbon33, 34 to few-layer and monolayer graphene.35-37 For tG = 1 min, the presence of a 2D peak with Pos(2D) = 2683 cm−1 and FWHM(2D) = 88 cm−1, the absence of a doublet in the D and 2D peaks, together with the overlap of G and D′ peaks indicate the formation of nanocrystalline graphite with no 3D ordering. Figure 1c shows that I2D and IG increase with increasing tG, whereas the ratio ID/IG decreases from ≈3.9 to 0.2 (see Figure 1d). At the same time Pos(G) down-shifts from 1601 to 1590 cm−1 (see Figure 1e) and a significant reduction of FWHM(D,G) occurs (see Figure 1f). The evolution of IG,2D, ID/IG, Pos(G), and FWHM(D,G) with increasing tG is consistent with the stage 1 ordering trajectory leading from nanocrystalline graphite to graphite. In this regime the size of sp2 clusters (La) increases with increasing ordering and can be estimated using the Tuinstra–Koenig relation ID/IG = C(λ)/La where C(532 nm) ≈ 4.96 nm.38, 39 Using this relation we estimate La ≈ 2 nm for tG = 1 min, which increases to La ≈ 25 nm for tG = 6 min as shown in Figure 1d. For tG > 6 min the 2D-peak intensity is larger than two times the intensity of the G-peak and it can be fitted with a single Lorentzian, with Pos(2D) = 2678 cm−1 and FWHM(2D) = 30 cm−1 indicating the formation of monolayer graphene.35, 40, 41 This conclusion is supported by AFM measurements showing the formation of islands with a thickness of 2.7 nm, which corresponds to monolayer graphene and accounts for fabrication residues and substrate effects.42 Furthermore, electrical transport measurements performed on continuous films with a similar Raman spectra and AFM thickness show the quantum Hall effect typical of monolayer graphene as discussed later. To provide further insights into the transition from nanocrystalline graphite film to graphene islands we monitor the evolution of the density, size, and separation of the islands using SEM observations combined with a simple counting algorithm described in the Supporting Information. Figure 2a shows the evolution from a continuous film to discrete islands with increasing growth time for 950 °C. These images have been performed on the same samples used for the Raman measurements in Figure 1. The average island area within the same range of growth times is shown in Figure 2b, whereas the average separation between islands at initial fragmentation then from 4 to 10 min is shown in Figure 2c. An initial reduction in island size suggests desorption of material from the surface. The observed saturation in the island separation of 7.23 μm indicates that there is no further nucleation of islands after the initial fragmentation. After 7 min we see a maximum in island size of 19.7 μm2. Raman measurements confirm that these islands are composed of graphene. SEM analysis of films grown for 1000 and 1035 °C reveals a similar behavior of the saturation in island separation and a maxima in island size (see the Supporting Information). We observe that an increase in growth temperature leads to a reduction in the time required to achieve the maximum island size and to form a monolayer graphene as shown in the inset of Figure 2c. A similar behavior has been also observed in other CVD graphene growth studies,24, 26 which showed that the growth rates of graphene islands are determined by competing atomic phenomena such as adatom mobility and attachment to the islands edges versus desorption, as well as being affected by the microscopic substrate roughness.26 The counterintuitive decrease in island area with time can be understood within the desorption controlled regime26 where the growth is a thermally activated process with a barrier energy Ea = (Edes + Eatt – Ed – Ead)/2 and with the density of graphene islands Ni ≈ PCH4· exp(2Ea/KT), with Edes the desorption energy of a carbon monomer on the Cu surface, Eatt the barrier of attachment for the capture of a monomer by supercritical nucleus, Ed the activation energy of surface diffusion of a monomer, Ead the activation energy for dissociative adsorption of CH4 on Cu, PCH4 the methane partial pressure, K the Boltzmann constant, and T the growth temperature. Figure 2d shows that when the island area decreases with time, Ni has a dependence on growth temperature which is typical of the desorption controlled regime with an activation energy of 1.66 eV. The desorption model is also consistent with the formation of holes inside the islands at 8 min of growth (see Figure 2a). The observed transition from a disordered carbon film adsorbed on Cu to graphene is very likely due to the combination of high temperature, low pressure, and the presence of the catalytically active surface of Cu, which induces the conversion to graphene as well as the thinning process of the carbon film. Previous studies43-45 have also investigated the high temperature conversion of amorphous carbon (a-C) films into graphene. In situ transmission electron microscopy (TEM) and molecular dynamics (MD) studies43 have reported the high temperature conversion of amorphous carbon (a-C) into graphene patches of 100 × 300 nm2. It was shown that a-C can rearrange into graphene through a phase of glasslike carbon which takes place within a time frame from 1 to 15 min, in the temperature range of 326–926 °C. Another study44 showed that graphene can be grown in a solid-state transformation of a-C in the presence of a catalytically active metal at temperature up to 720 °C. In this case rearrangement processes take place in two or 3D unordered network structures in which a huge number of bonds are broken and newly formed. Finally, a third study showed the metal-catalyzed crystallization of a-C to graphene by thermal annealing at 650–950 °C.45 It was shown that part of the carbon source is crystallized into graphene with the rest outgassing from the system. Furthermore, this study also reports that for long annealing times no carbon or graphene remains on the surface due to significant desorption of C atoms under the low pressure and high temperature ambient. Similarly to these studies we have a film of nanocrystalline graphite on top of a catalytically active metal in low pressure and high temperature conditions, as well as comparable time frames for the conversion to graphene. Having established the initial stages of graphene formation, we investigate the transition from graphene islands to a continuous film. Figure 2c shows that the island size reaches a maximum with the growth time and a further increase in the growth time leads to a decrease in the island size. To grow continuous graphene monolayer films we adopted the two stage growth described by Li et al.,26 where increasing methane flow rate after the formation of the islands is shown to fill the regions between islands while suppressing further nucleation sites. As our objective is to minimize growth time, we selected the growth temperature of 1000 °C where maximum island size and island separation are reached in the shortest time (40 s). Using the grown graphene islands as nucleation sites, we find that increasing the methane flow rate and growth time to 5 min allows the islands to merge into a continuous graphene monolayer film of up to 8 cm2 in area. SEM, AFM, and Raman measurements confirm that the continuous films are monolayer graphene. Figure 2e shows the morphology of the graphene monolayer after the complete coalescence of the islands studied by SEM and AFM. The analysis of the Raman measurements performed on the continuous films is presented in Figure 1, where the green highlighted regions in panels (c) to (f) indicate the values of IG,2D, ID/IG, La, Pos(G), and FWHM(D,G,2D) for a 1 × 1 cm graphene film. Raman mapping measurements shown in the Supporting Information demonstrate the uniformity and high-quality of the continuous films. The total processing time of this procedure is about 20 min (see the Supporting Information); this includes (1) heating up time for the CVD system from room temperature to the growth temperature, (2) Cu foil annealing time, (3) graphene nucleation and growth time, (4) cooling down time for the system to room temperature. The demonstrated processing time is significantly shorter than the processing time needed by hot-wall CVD (typically >70 min).1, 2, 4, 5, 21, 22 We estimate the total cost of graphene production by cold-wall CVD to be <0.37 cm−2 (see the Supporting Information). Compared with other CVD studies and neglecting the base cost of copper we see a reduction in the production costs of 98.83%–99.89%. This extraordinary reduction in the production cost, together with the possibility of reconstitution of high purity copper from etchant solutions by electrolysis that can yield up to 99% of the original foil,46 open a new way forward to accelerate the commercialization of graphene. To ascertain the quality of the electronic properties of graphene produced by cold-wall CVD we characterized the charge carrier mobility in transistor devices fabricated on SiO2/Si substrates. Using the model we estimate the effect mobility to be 3300 cm2 V−1 s−1 at K and cm2 V−1 s−1 at room temperature. This across a large area is comparable to the mobility in area devices of graphene grown by CVD to or (typically few and on 2, 4, 5, 21, The quality of cold-wall CVD graphene as compared to that grown with other methods is using the electronic quality factor (Q) for the area across which the carrier mobility is is as the effect mobility V−1 by the area of the As shown in the Supporting graphene grown by heating cold-wall CVD has Q ranging from 4 × to × whereas most reports of monolayer graphene grown by hot-wall CVD have Q ranging from to 7 × cold-wall CVD grown graphene has a of Q from a high quality growth This is in to the of Q over three of reported for hot-wall CVD grown graphene. Figure shows the in a large Hall × 25 see fabricated on standard substrates. The Si substrates are doped and as the a to the we the carrier from × to 6 × The charge is at indicating low in our Figure shows the and the Hall the at 13 T and at a temperature of are when the energy is within a that to the = + typical of the quantum Hall effect of monolayer graphene with and At the same time we observe = where the energy is within the = for and indicating that the graphene quality is high to observe the in the At the same time, shows well-defined which are with the A of the Hall and charge carrier shows that up to = 6 are at high (see Figure The = 1 is down to as low as 5 The presence of these at low is an of low in the graphene which further the high electronic quality of cold-wall CVD graphene. In the of this we demonstrate that graphene produced by this novel method is suitable for the next generation flexible and transparent In such is the method for an the the capacitive have the time and the to However, graphene-based flexible capacitive have been demonstrated so far due to the from of graphene and layers on flexible substrates. We a novel fabrication procedure that the high quality of graphene (see the Supporting us to demonstrate for the first time a flexible and transparent capacitive using graphene for the top and Figure shows a of a capacitive fabricated on a flexible and transparent The of two of graphene by as in Figure The graphene were fabricated on the Cu foil and to nm thick by their transfer to the transport measurements show that the typical across is ≈ and the A of the fabrication procedure and electrical is provided in the Supporting Information. The of the are at the between the graphene and a As pressure is to an of the the the between the graphene resulting in an increase in Figure shows an of in for when one is with a The maximum in occurs on the with changes to the To the of the we and an with a and the in shown in Figure A in of = 6 was observed with a to the original after The in demonstrated indicates a fast to and of the we the and of the devices by the substrate and the of the graphene across the Figure shows the in the two of the graphene as the is a cm for This was performed for graphene and to the of After only changes of less than in the are which show no significant of the of the flexible These measurements demonstrate the and of our graphene and its suitability for use in flexible In we have shown a new growth mechanism of graphene by cold-wall CVD, which with the formation of a thick carbon film in the early stages of the growth, that becomes progressively thinner with increasing the growth time and finally evolves into graphene At the same time we demonstrate an high-throughput and cost efficient growth procedure for high quality monolayer graphene using cold-wall CVD. Finally, we use graphene as material and demonstrate the first flexible and transparent graphene capacitive using processing that are with transparent and flexible electronic its for the industrial exploitation of graphene cold-wall CVD are found in semiconductor manufacturing our could to new of flexible electronics and new for the of graphene-based The from and from the and As a to our and this provides information by the materials are and be for are or from information than be to the The is for the or of information by the than be to the for the