Choon-Ming Seahab,
Brigitte Vigolo*a,
Siang-Piao Chaic,
Satoshi Ichikawad,
Jérôme Gleizee,
Jaafar Ghanbajaa and
Abdul Rahman Mohamedb
aInstitut Jean Lamour, CNRS-Université de Lorraine, BP 70239, 54506 Vandœuvre-lès-Nancy, France. E-mail: brigitte.vigolo@univ-lorraine.fr
bSchool of Chemical Engineering, Engineering Campus, Universiti Sains Malaysia, 14300 Nibong Tebal, Seberang Perai Selatan, P. Pinang, Malaysia
cChemical Engineering Discipline, School of Engineering, Monash University, Jalan Lagoon Selatan, 46150 Bandar Sunway, Selangor, Malaysia
dInstitute for NanoScience Design, Osaka University, 1-3, Machikaneyama-cho, Toyonaka, Osaka 560-8531, Japan
eLaboratoire de Chimie Physique-Approche Multi-échelle de Milieux Complexes-Université de Lorraine, 1 Bd Arago, 57078 Metz, France
First published on 20th April 2016
CVD is the most efficient way to produce wafer scale monolayer graphene. Although Ni and Cu are widely reported catalysts for monolayer graphene formation, but both Ni and Cu are requiring different extreme conditions to grow graphene in single layer. Here, we show that monolayer graphene could be grown simultaneously on polycrystalline Ni and Cu under single atmospheric pressure CVD for the first time. Our catalyst system combines carbon solubility divergence between the two catalysts to limit the exposure to the carbon source. Structure and quality of the grown graphene were characterized by HRTEM and Raman spectroscopy mapping. The growth mechanism shows that the role of grain boundaries is crucial for carbon diffusion tuning. The results show that free-standing high-quality monolayer graphene can be produced in a controlled and simple way with an affordable catalyst system.
Ni and Cu are two most promising catalysts due to moderate and low carbon solubility into Ni5 and Cu6, respectively. The growth mechanism of graphene on Cu is known to be surface mediated, and Cu possesses self-limiting surface deposition which enhances the formation of predominantly monolayer graphene. However, Cu requires strict process conditions to grow monolayer graphene, including high vacuum conditions, a low carbon concentration environment and a synthesis temperature near to the melting point of Cu.7 Ni rather follows a bulk-mediated graphene formation mechanism, which unfortunately favors the formation of multilayer graphene. The formation of monolayer graphene on Ni requires high-speed cooling after CVD or the use a very thin Ni film as the catalyst.
Cu and Ni could form binary isomorphous alloys for which both metals are mutually miscible.8 Some researchers had investigated the possibility of using both Cu and Ni to grow graphene with the hope to combine the benefits of Cu and Ni. Two strategies have been reported to grow graphene from bimetallic catalysts, either by using layer-by-layer bimetallic catalyst foils9,10 or by using alloy foils.11,12 However, the synthesis process is rather complex, and it requires high vacuum condition and low carbon precursor/diluting gas to grow monolayer graphene.
In the present work, a facile technique is demonstrated to grow uniform monolayer graphenes simultaneously on both polycrystalline Ni and Cu foils by Ni–Cu bilayer catalyst. The CVD reaction was carried out under atmospheric pressure. High uniformity and quality of the crystalline structure of the grown graphene was evidenced by Raman spectroscopy mapping and High Resolution Transmission Electron Microscope (HRTEM). Our straightforward bimetallic catalyst allows the control of carbon diffusion to the area between the Ni and Cu surfaces. In particular, carbon access is reduced at the inner Ni surface, while Cu behaving as a carbon barrier. The growth mechanism of monolayer graphene facilitated by carbon diffusion through the bulk and Ni grain boundary, the driving force coming from carbon-rich surface to carbon-lacked surface. The grain boundaries are shown to play a crucial role in carbon control during the growth stage. The robustness of this bilayer catalyst allows a facile separation of the grown graphene from the catalyst and transfer to the desired support.
In general, the Ni–Cu bilayer catalyst was placed into a quartz boat with Ni surface facing upward. The quartz boat was then placed at the centre of a quartz tube (inner diameter: 30 mm). The Ni–Cu foil was then heated from room temperature in tubular furnace (Carbolite) up to the desired temperature (800–1050 °C) with a heating rate of 10 °C min−1. During the heating, high purity H2 gas (99.999%) with a flow rate of 80 sccm was flowed into the quartz tube to reduce the oxide present on the Ni surface. As the desired temperature was obtained, high purity CH4 (99.999%) was fed into the reactor at a flow rate of 20 sccm with the H2 flow rate remained unchanged. After reaction (3–7 min) the quartz boat along with the Ni foil was displaced from the heating zone to an ambient temperature zone for rapid cooling. At the same time, CH4 and H2 gases were switched off and N2 gas (99.999%) subsequently introduced into the quartz tube to quench the reaction.13
A JEOL JEM-09100IS Ion Slicer apparatus was utilized for thinning of the Ni foil in order to observe the element composition at the middle of the bulk Ni foil. First, a strip of Ni with a thickness of 250 μm was cut from the centre of the Ni foil used for CVD reaction. It was further polished with sand paper (micro-mesh 4000) to diminish the thickness down to 100 μm. The Ni strip was placed into the ion slicer apparatus and a shield belt with a thickness of 10 μm was placed onto of the strip. An Ar+ ion beam was then irradiated onto the Ni strip from the top and the Ni strip was kept oscillating by 3° at both sides until an etched hole appear. The Ni strip was transferred to a Fei Tecnai 20 transmission electron microscope for imaging and energy-dispersive X-ray spectroscopy (EDS) analysis for elementary composition between Ni grains and grain boundaries after the CVD process. A Keithley 220 programmable current source in conjunction with a Hewlett-Packard 34401A multimeter was used for the four-point probe sheet resistance measurement of graphene grown at inner surface of Ni in ambient environment and at room temperature.
The graphene samples were successfully separated from their respective metal foil and deposited onto a silicon wafer for Raman analysis. Three main bands in the Raman spectrum of graphitic materials are normally used to determine structural information: (1) D-(“disorder”) band at ∼1350 cm−1; the relative intensity of this peak is related to the degree of disorder, dangling bonds or relative sp3 carbon content, in the carbon structure; (2) G-band at ∼1580 cm−1 is the peak which indicating the sp2 graphitic bond; (3) 2D-band at ∼2680 cm−1 is the second harmonic of the D-band. The main Raman features used to evidence a graphitic material with one single sheet are: (i) a single, highly symmetric and narrow 2D-band with a single Lorentzian component; (ii) a relative intensity of the 2D- and G-bands, I2D/IG, higher than 1.4;14–16 (iii) a full width half maximum of the 2D-band (fwhm2D) below 40 cm−1.11,16,17
For conventional micro-Raman analysis, five spectra were taken randomly from each graphene sample grown at different temperatures for 5 min and at each surface of the CVD device; the corresponding I2D/IG are summarized in Fig. 1c. For Cu, 900 °C and below were too low to synthesize pristine wafer scale graphene; no signal of the 2D-band was obtained. On the other hand, for a temperature of 1050 °C, just below the melting point of Cu (1085 °C), disordered thick graphitic layers were obtained. For our CVD system, thermal decomposition of methane significantly observed at 1050 °C. The uncontrolled decomposition of methane ended up with both the catalyst and wall of the CVD reactor covered with a thick layer of disorder graphitic materials. For a temperature of 900 °C, I2D/IG are found to be above 1.4 for the graphene grown at inner surface of Ni, but the area of such samples were small and not fully covered the Ni surface. Interestingly, at a temperature of 950 °C, large area of graphene that fully covered the catalyst surface were grown on both inner surfaces of Ni and Cu, and the I2D/IG are well.
All of the regions of graphene grown from inner surfaces of Ni and Cu display a ratio I2D/IG much higher than 1.4 (Fig. 2a and c). The overall of the fwhm2D of the Raman 2D-band are below 36 cm−1 (Fig. 2b) and below 37 cm−1 (Fig. 2d) for graphene grown on inner surface of Ni and Cu, respectively, furthermore, the 2D-bands well fitted into single Lorentzian curve. The range of I2D/IG ratio and that of fwhm2D is relatively large due to the trapped moisture and air between graphene and silicon wafer causing some stress to C–C bonds of graphene. From point based Raman spectra (see ESI, Fig. S1†), the D-band is very weak compared with G-band, showing that the grown graphene contains low amount of defects. HRTEM observations were randomly carried out at the several folded edges of the graphene sheets. A clear single straight line is visibly monitored for graphene grown at both inner surface of Ni (Fig. 2e) and inner surface of Cu (Fig. 2f). Enlarged images (inserted at the bottom of Fig. 2e and f) of respective graphene in the boxed areas show how the hexagonal graphene structure is oriented. As can be seen from the Fast Fourier Transform (FFT) images the graphene film generate a single set of hexagon diffraction pattern (Fig. 2e and f (inserted at top)) indicating the present of monolayer graphene. AFM images (Fig. S2†) show that the thickness of graphene is less than 0.5 nm which is a sign of high selectivity of monolayer graphene on both samples. Since we have used polycrystalline Ni and Cu as catalyst, the roughness certainly corresponds to the grain boundary visible under AFM.
All the data obtained are pointing to the presence of monolayer graphene on both inner surfaces on Ni and Cu with CVD temperature of 950 °C and 5 min. As mentioned earlier, Ni has a relatively higher carbon solubility and higher carbon diffusion rate. Formation of monolayer graphene on Ni catalyst is difficult by using atmospheric pressure CVD conditions; and few layers of graphene are usually obtained. An alternative process involving the diffusion of carbon adatoms across the Ni film was then proposed; graphene being formed at the opposite surface.18–20 A carbon-poor environment is created to reduce the carbon-concentration and lead to successful synthesis of graphene with a minimum number of layers. Moreover, the uniformity of graphene layers was much improved under such a strategy, compared with conventional Ni catalysed CVD. However, the graphene grown with such methods is normally firmly attached onto the silicon that served as substrate, making it difficult to obtain free-standing graphene.
We propose a simple approach to facilitate the formation of monolayer graphene that enables the following separation process to obtain free-standing monolayer graphene. Cu with almost no carbon solubility is a good candidate to insulate carbon atoms.21 Our system combines the properties of Ni and Cu in order to dramatically reduce the carbon content at the inner Ni surface. Additional experiments with addition of silicon wafer to the catalyst system were carried out (Fig. S3†). Silicon wafer which has no carbon solubility and is inactive for graphene formation was used to act as a carbon shield. A silicon wafer was first placed between Cu and Ni (Fig. S3a†); no graphene formation was observed on the inner surface of Cu: i.e. no graphene could be separated. On the contrary, on inner surfaces of Ni, graphene could be observed after the Ni etched away. For the second experiment, the silicon wafer was placed on the Ni surface so that Ni was sandwiched in between Cu and silicon (Fig. S3b†). After the CVD reaction, no graphene could be found on both Ni surfaces after Ni were etched away. That means that carbon atoms involved in graphene formation mainly come through diffusion over the bulk Ni foil, neither through the leak of gases into that area nor diffusion over the Cu foil. Although, carbon diffusion was reported through the grain boundaries of 100 nm-thick Cu foil;22 our Cu foil being much thicker (25 μm) is able to efficiently inhibit carbon diffusion and the access to the inner surface of Ni is completely inaccessible through the Cu foil.
As the sequence of the two catalysts was flipped over (Ni foil wrapped over Cu foil), monolayer graphenes were as well found after the CVD reaction done at 950 °C and 5 min at inner surface of Ni and that of Cu (Fig. S4†). To further identify the path of carbon diffusion though Ni, we have investigated the characterization of the Ni foil after the synthesis process. Since the reaction was quenched by the use of high-speed cooling, this analysis provides a snapshot of the state during the reaction. Analysis was performed on a cross section of the Ni foils thinned down to around 20 nm (for the position around 90 μm from the surface). The bright field TEM image (Fig. 3a) reveals grain boundaries enriched by carbon atoms (Fig. 3b). EDS analysis confirms that carbon concentration is higher at the Ni grain boundaries; C content is higher at the grain boundary compared to that at a location within the Ni grain (Fig. 3c). A precise quantification of the C/Ni ratio is not possible because of the too high absorption difference between Ni and C element. However, C Kα line is very obvious (Fig. 3c, insert) showing the higher content of carbon near grain boundaries, especially since very low C concentration is involved in the CVD reaction.
Fig. 4 shows optical micrographs of the graphene samples transferred to silicon wafer. Their formation was obtained after 5 min (Fig. 4a and b) and 7 min (Fig. 4c and d) of reaction. Under shorter growth durations of 3 min and below (not shown here), small flakes of graphene sheet were scattered over the inner surface of Ni and cannot be transferred effectively. The optical micrographs (Fig. 4a and b) show uniform contrast across the imaged area for the graphene obtained at both inner surfaces of Ni and Cu under reaction duration of 5 min, indicative of uniform monolayer graphene coverage. The variation of colour contrast become significant as the CVD duration extended to 7 min, for graphene obtained from inner surface on Ni (Fig. 4c), indicating that the film is not uniform in thickness and consist of multi layered graphene. Comparable areas of multi layered graphene were also found on Cu surfaces (Fig. 4d).
The graphene that was obtained on the inner surface of Cu displays a similar topography as the graphene grown on the respective Ni surface. However, the same pattern of grain boundaries was not seen onto the graphene on Cu. The graphene grown on the inner surface of Cu possesses instead a weakness, where small scales of voids were appearing on the graphene film.
The Ni foil selected is a conventional and low-cost polycrystalline Ni foil. Contrary to monocrystalline Ni, it is made of grains with irregular sizes and shapes. Diffusion of carbon adatoms is consequently not uniformed towards the Ni foil. It could take place either through the bulk of the grain and also through the gap in between the grains, known as grain boundary. The solubility of carbon into bulk Ni at 950 °C is around 1.0 at%.23 Diffusion rate of carbon atoms is favoured at the grain boundaries. The carbon concentration gradient between the Ni surface facing carbon-rich CVD atmosphere (outer surface of Ni) and the carbon poor region of Ni (inner surface of Ni) would be the main force facilitating the carbon diffusion. The carbon diffusion mechanism is illustrated in Fig. 5. Carbon adatoms are diffusing faster through the grain boundary, but at the same time, some carbon adatoms are diluted into the grain due to the concentration gradient (red arrows in Fig. 5a). Our process allows the control of the carbon concentration within Ni grains leading to a uniform carbon distribution over the whole Ni surface. Combined with post CVD fast cooling, uniform graphene in term of number of layer is easily obtained. Under higher temperature (1030 °C), the diffusion of carbon through the gap between the grains being increased, much faster compared to its dilution into the Ni grain (Fig. 5b), the CVD reaction duration applied was intentionally reduced. No graphene film was obtained but instead web-liked carbon material could be observed (see ESI, Fig. S5†). The time allowed for the carbon to dilute within Ni grains was too short and carbon distribution was obviously non-uniform throughout the Ni surface. After a fast cooling, graphitic materials segregated at the locations enriched by carbon (grain boundaries) while no carbon material was formed on the surface far from grain boundary. It was here as well confirmed that the grain boundary was the major channel for the carbon supply to the interface between Ni and Cu. The sheet resistance (Rs) of monolayer graphene grown was measured to be 1900 ohm sq−1, near of the values reported in previous reports on stacked monolayer graphene.24
What is usually observed from the polycrystalline Ni in atmospheric pressure CVD is the formation of multilayer graphene.14,25 The misalignment of the distance of the lattice of Ni is reported to provide abundant nucleation sites for graphene to segregate and that leads to the formation of more layers of graphene.26 This is the main reason why Ni foil is normally forsaken to grow graphene. With our bilayer catalyst system, accumulation of the carbon at grain boundaries is reduced, inhibiting the excessive carbon segregation and thus enhancing the formation of monolayer graphene. Multilayer graphene is more likely to form on Cu due to the excessive supply of carbon. But, through our strategy, the carbon-poor atmosphere is present at the interface of Ni–Cu. The main carbon supply was from the contact with Ni.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra04197c |
This journal is © The Royal Society of Chemistry 2016 |