Spatially Controlled Nucleation of SingleCrystal Graphene on Cu Assisted by Stacked Ni Dong Ding,† Pablo Solís-Fernández,‡ Hiroki Hibino,§ and Hiroki Ago*,†,‡ †
Interdisciplinary Graduate School of Engineering Sciences and ‡Global Innovation Center (GIC), Kyushu University, Fukuoka 816-8580, Japan § School of Science and Technology, Kwansei Gakuin University, Hyogo 669-1337, Japan S Supporting Information *
ABSTRACT: In spite of recent progress of graphene growth using chemical vapor deposition, it is still a challenge to precisely control the nucleation site of graphene for the development of wafer-scale single-crystal graphene. In addition, the postgrowth patterning used for device fabrication deteriorates the quality of graphene. Herein we demonstrate the site-selective nucleation of single-crystal graphene on Cu foil based on spatial control of the local CH4 concentration by a perforated Ni foil. The catalytically active Ni foil acts as a CH4 modulator, resulting in millimeter-scale single-crystal grains at desired positions. The perforated Ni foil also allows to synthesize patterned graphene without any postgrowth processing. Furthermore, the uniformity of monolayer graphene is significantly improved when a plain Ni foil is placed below the Cu. Our findings offer a facile and effective way to control the nucleation of highquality graphene, meeting the requirements of industrial processing. KEYWORDS: chemical vapor deposition, single-crystal graphene, crystal growth, field-effect transistors such as the CH4 concentration,14 the presence of impurities on the Cu,11,12,18−21 surface irregularities and defects,22−25 and the crystal structure of the Cu.26 Patterns of poly(methyl methacrylate) (PMMA) were used as controlled nucleation sites, but the graphene quality was degraded due to contamination from the PMMA.27 Recently, controlled growth of large graphene grains has been attained by locally feeding the CH4 gas using a gas nozzle and Cu−Ni alloy to speed up the growth rate.28 However, this method requires a complicated setup and the use of an alloy as catalyst, which hinder its upscaling. Here, we report that strategically located Ni foils enable one to produce spatially defined nucleation of millimeter-scale, pure monolayer, single-grain graphene at a high growth rate. The local tuning of the CH4 feedstock is realized by perforated Ni foil masks, allowing for the localized growth of large graphene grains as well as for the growth of patterned graphene without the need of lithography processes. Moreover, when a plain Ni foil is placed below the Cu, bi/few-layer graphene areas, which are usually observed in the graphene grown on Cu foils,1,3,13,15−17,29−31 are completely suppressed. Our method offers a low-cost, simple, and powerful way to control the
A
mong various methods to synthesize monolayer graphene, chemical vapor deposition (CVD) on Cu foils is the most promising to produce large-area graphene at relatively low cost.1−3 However, CVD-grown graphene usually consists of a number of grains separated by grain boundaries (GBs), which negatively impact the physical and electronic properties of graphene.1,4,5 Several studies have been devoted to decrease the nucleation density for obtaining large and isolated single-crystal graphene grains.6−13 This can be partly achieved by lowering the CH4 feedstock or pretreating the Cu surface.11,14 More recently, it was shown that the presence of oxygen in the Cu decreases the graphene nucleation density, allowing to synthesize millimeter and even centimeter-sized single-grains.15−18 However, decreasing the nucleation density generally results in unpractically long CVD times to obtain a complete graphene coverage.16,17 Compared with the attempt to increase the grain size, little attention has been paid on controlling the position of the graphene nucleation sites on the Cu foil. This lack of control induces the presence of untraceable GBs when the randomly scattered large-grains are merged to form a continuous graphene layer. Controlling the position of the nucleation allows to obtain single grains either at a predetermined point for a complete coverage without GBs or at predefined positions for the mass fabrication of device arrays and circuits at increased processing speeds. However, this can be a complex task, as graphene nucleation depends on a variety of factors, © 2016 American Chemical Society
Received: September 16, 2016 Accepted: November 23, 2016 Published: November 23, 2016 11196
DOI: 10.1021/acsnano.6b06265 ACS Nano 2016, 10, 11196−11204
Article
www.acsnano.org
Article
ACS Nano
Figure 1. (a) Schematic of the production of a spatially controlled, single-crystal graphene grain on a Cu foil. The Cu foil is sandwiched between two Ni foils during the CVD, which control the local CH4 concentration. (b) As-grown, millimeter-sized single graphene grain obtained by using the method shown in (a). The position of the hole (2 mm diameter) in the upper Ni foil is indicated by the white circle. Graphene was visualized by slightly oxidizing the Cu foil after the CVD growth. (c) A graphene grain obtained by using a Ni mask with a 6 mm hole. Magnified images before (d) and after the transfer (e). (f−j) Simultaneous site-selective nucleation of three graphene grains on a 15 × 10 mm2 Cu foil by a Ni mask with three holes. Images show the as-grown grains (f) and those after the transfer (g). (h−j) Magnified images of the individual grains from (g). (k) Schematic of a conventional CVD growth of graphene on Cu foil without the assistance of Ni foils. (l) As-grown graphene using the same CH4 concentration as in (c−e) but without placing Ni foils on either side of the Cu. (m) Optical micrograph of the sample shown in (l) after the transfer. The graphene is monolayer but contains many bi/few-layer graphene areas.
nucleation of large single-crystals of monolayer graphene and to directly grow patterned graphene, which are expected to stimulate the development of practical applications of CVD graphene.
single-crystal graphene, obtaining a large graphene grain at a specific area of the Cu foil. As shown in Figure 1b, a single graphene grain was obtained in the center of a ∼17 × 17 mm2 Cu piece by using a Ni foil with a 2 mm-diameter hole. This corresponds to a nucleation density as low as 0.34 nuclei/cm2, which is the lowest value reported for Cu foils as far as we know (see Supporting Information, Table S1).10−18,29,31,32 It is worth mentioning that this density can be further reduced by using larger Cu and Ni foils. By increasing the size of the hole, the growth rate of the graphene can be increased, although the accuracy on the position slightly decreases. However, even when a Ni foil with a relatively large hole (6 mm diameter) was used, a single graphene grain was still obtained, located only ∼1 mm off the center of the hole (Figure 1c, 60 ppm of CH4, 90 min CVD reaction). We checked the reproducibility of this site-selective nucleation. Figure S2 and Table S2 of Supporting Information summarize the result of 12 CVD runs with different CH4 concentrations and growth times using a Ni foil with a 6 mm circular hole. We found that a reasonably wide range of the growth conditions give a single graphene grain (around 67% of the performed CVD runs with a 6 mm hole, see Figure S2, Supporting Information). There is a tendency that the number of grains increases with CVD time or CH4 concentration (see Table S2, Supporting Information). However, densities attained
RESULTS AND DISCUSSION Site-Selective Synthesis of Single-Crystal, Bi/FewLayer-Free Monolayer Graphene. Figure 1a shows a schematic of the process employed for the CVD growth of single-crystalline graphene at predefined positions of the Cu foil. The Cu foil was oxidized just before the CVD to decrease the graphene nucleation.15 This preoxidized Cu foil was then sandwiched between two Ni foils during the CVD, in contrast to conventional CVD growth that uses only a preoxidized (or pristine) Cu foil. The upper Ni foil has a circular hole at the center (diameter varies between 2 and 6 mm, see Supporting Information, Figure S1a,b) and is employed to modify the spatial concentration of CH4 in the surroundings of the Cu. Ni has a higher catalytic activity toward CH4 decomposition and a much higher C solubility than Cu.30 As we shall discuss later, this induces a decrease of the CH4 concentration in the vicinity of the Ni. Thus, the local CH4 concentration becomes higher at the center of the Ni hole, triggering the graphene nucleation on the Cu just below the hole, while suppressing it on the rest of the Cu surface. This allows us to control the nucleation site of 11197
DOI: 10.1021/acsnano.6b06265 ACS Nano 2016, 10, 11196−11204
Article
ACS Nano were always much lower than those when the top Ni foil was absent (see Supporting Information, Figure S3). When using a smaller hole with 2 mm diameter, a single grain was always obtained under the tested conditions (Supporting Information, Table S3). The magnified image of the grain (Figure 1d) clearly shows that the shape is hexagonal, suggesting the formation of singlecrystalline graphene. The grain also shows a uniform optical contrast after being transferred onto a SiO2 substrate (Figure 1e), except for the presence of some polymer residues due to the transfer process. This implies that the grain is a uniform monolayer of graphene. As we shall see later, the absence of bi/ few-layer areas is a direct consequence of the presence of the bottom Ni foil. The result of the CVD dramatically changed when no Ni foils were used (Figure 1k), with a large number of graphene grains nucleated at random positions across the Cu surface (Figure 1l; note that for this figure, the CVD time was shortened to avoid a complete merging of the grains). Supporting Information Figure S3 shows that an increase of CH4 concentration increases the total area of graphene. Thus, the almost full coverage of the Cu surface is an indication of a higher local CH4 concentration when no Ni is used (Figure 1l). Another difference is the presence of many bi- to few-layer graphene areas when no Ni foils are employed, as indicated by the optical contrast seen after transfer onto SiO2 (Figure 1m). Thus, our method can offer a facile way to control the nucleation of graphene without the formation of multilayer grains, which are widely seen in the literature for the graphene grown of on Cu foils.1,11,16,32−34 It should be noted that our approach here is completely different from those reported for graphene growth using Ni− Cu alloys instead of pure Cu foils.28,35−37 In our case, there is only a slight physical contact between the Cu and Ni foils without any external pressure, and thus no substantial alloying occurs between the Ni and Cu foils. Accordingly, they are easily detached from each other after the CVD. This was confirmed by X-ray diffraction (XRD) measurements of the Ni foils after the CVD, as will be discussed later. We also confirmed that the Ni foils can be repeatedly used without any decrease on their efficiency nor requiring any further treatment between the successive CVD runs (see Figure S4, Supporting Information). By simply arranging multiple holes in the Ni mask at selected positions, it is also possible to obtain several graphene grains on the Cu surface at preselected positions. Figure 1f−j shows one example of three isolated graphene grains grown using a Ni mask having three circular holes. Figures 1f and 1g show asgrown graphene grains and those after the transfer, respectively. In spite of the relatively large size of the employed Cu foil (∼150 mm2), graphene nucleation occurred only at the uncovered areas. This result is essential for speeding up the growth process, allowing to grow single-grain graphene arrays even in a wafer scale, which is promising for automating the processing of graphene for practical applications at industrial scales. Especially interesting is the combination of the multiholed Ni masks with graphene growth techniques on epitaxial Cu(111) films that allow to control the orientation of the graphene grains, potentially leading to a seamless stitching of the different grains.38 The quality of the graphene grown on the Cu foil sandwiched between two Ni foils was investigated by Raman spectroscopy after transferring to a SiO2 substrate (Figure 2a). The spectra were collected at different positions of the sample surface, indicated by the colored circles in the optical image
Figure 2. (a) Raman spectra of a single graphene grain grown on Cu sandwiched by Ni foils, after transfer on SiO2. Each spectrum was collected from the spots marked in the optical image with the same color. Upper-right side of the image (dark) is the graphene, while the lower-left side (light) is the bare SiO2. (b,c) Raman mapping images of the same area shown in (a). Intensity ratios of ID/IG (b) and I2D/IG (c). (d) Optical image of an as-grown graphene grain (center) and LEED patterns collected at the marked positions.
(Figure 2 inset). No significant D band (∼1350 cm−1) was observed in the spectra, except for some specific positions (green spectrum) that correspond to wrinkles formed during the CVD growth and/or transfer to the SiO2 substrate (Figure 2b). This D-band is usually related to the presence of defects, GBs, and/or wrinkles, and thus its absence clearly indicates the high quality of the graphene.39 The relative intensity of the 2D band (∼2680 cm−1) with respect to the G band (∼1585 cm−1) was ∼2, supporting that the graphene is monolayer (see Figure 2c).30 To confirm that the isolated graphene grain obtained below the Ni hole is single crystal, low-energy electron microscopy (LEEM) was performed for an as-grown grain on the Cu foil. An optical image of the grain is shown in the center of Figure 2d. Electron diffraction patterns were collected across the grain surface, at the areas marked by the blue circles. All the diffraction patterns related to the graphene (six diffraction spots marked by dotted yellow circles) are oriented along the same 11198
DOI: 10.1021/acsnano.6b06265 ACS Nano 2016, 10, 11196−11204
Article
ACS Nano direction, implying the growth of a single-crystal graphene grain. Together with the hexagonal diffraction pattern originated in the graphene, some other additional diffraction spots were also observed (Figure 2d). These additional spots change depending on the measured position and are probably originated from the faceting of the underlying polycrystalline Cu foil surface. Interestingly, the CVD graphene extended continuously over the GBs of the Cu (see e.g., the dark horizontal line intersecting the graphene hexagonal domain), as reported elsewhere.15 Role of the Upper Ni Foil. In the previous section, we have shown that the graphene growth on a Cu foil can be tuned by the presence of nearby Ni foils. In this section, we will discuss the role of the Ni foil placed above the Cu, and how it enables the site-selective nucleation of graphene. It should be noted here that there is only a slight physical contact between the upper Ni foil and the Cu foil, and due to the roughness of both foils, there exists a small gap (30 times). Some examples are shown in Figure 5a−e; the patterns of monolayer graphene, such as arrays of circles and squares as well as the word “Graphene”, were easily obtained (Figure 5a−e). Although the graphene is not necessary singlecrystalline in this case, it is still important to show the possibility of the direct synthesis of patterned graphene without the aid of any lithography processes, avoiding the postgrowth etching of graphene, that increase the complexity of the processing and can deteriorate the quality of the graphene. To demonstrate the potential of this method, an array of graphene field-effect transistors (FETs) was produced without using any lithography processes, including the patterning and deposition of electrodes. The graphene channels for the FET were directly grown on a 4 in. sapphire wafer with a thin Cu film on top, and the metal electrodes were deposited using a shadow mask after transferring the graphene to a SiO2/Si substrate. These devices show the ambipolar transfer curve expected for graphene (Figure 5f).46 The carrier mobility was 2000−3000 cm2/(V s) at 300 K probably due to the difficulty of transferring such large graphene (the FET channel length and width are millimeter scale) without wrinkles and impurities. Mechanism of Site-Selective Nucleation of Monolayer Graphene. From the experimental observations, the mechanism of site-selective nucleation of monolayer graphene on the stacked Cu catalyst is presented in Figure S15, Supporting Information. The key of our strategy is to utilize a perforated Ni foil mask on top of the Cu to locally control the carbon source supply. The high catalytic activity of Ni metal toward CH4
cm2 can be readily obtained (Figure 4f inset and Figure S13, Supporting Information). Finally, it is worthwhile to note that the bottom Ni foil also prevents the growth of graphene on the backside of the Cu foil (see Figure S14, Supporting Information). In conventional CVD using Cu foils, graphene is formed on both sides of the Cu, requiring to remove the graphene from one of the sides by plasma etching or any other method before the transfer to another substrate.45 This additional processing is inconvenient and can potentially damage/contaminate the graphene on the other side. By simply placing a Ni foil below the Cu foil, the graphene formation can be completely suppressed at the backside, which makes the graphene transfer process easier, faster, and more efficient. Patterned Growth of Monolayer Graphene. Our Ni mask method is not limited to the site-selective nucleation of single-crystal graphene. The method can be further applied for the direct synthesis of patterned graphene, as shown in Figure 5. This is done by either prolonging the reaction time or increasing the CH4 concentration so that the graphene extends its growth front until the edges of the Ni mask. Thus, we can synthesize graphene patterns with arbitrary shapes by simply placing the appropriate Ni mask on the Cu foil. In addition, as we discussed before, the Ni masks can be used in many cycles 11201
DOI: 10.1021/acsnano.6b06265 ACS Nano 2016, 10, 11196−11204
Article
ACS Nano
For the experiments shown in Figures 3d,e and S9, we used the following conditions: Low CH4 dose: CH4 23 ppm, reaction time 120 min. Middle dose: 95 ppm, 120 min. High dose: 95 ppm, 240 min. Graphene Transfer. For the transfer from the Cu foil, the graphene surface was covered with a PMMA film by spin coating. After removing the Cu film by immersing in an aqueous solution of APS, the PMMA/graphene stack was transferred onto a SiO2/Si wafer. Finally, the PMMA film was removed by dipping the substrate into acetone. Characterizations. Raman spectra were measured with a Nanofinder30 (Tokyo Instruments) using a 532 nm excitation laser. The crystallinity and crystallographic orientation of Ni foils were measured by an X-ray diffractometer (XRD) (RIGAKU RINT-III) using Cu−Kα radiation (1.5418 Å). Crystal orientations of the as-grown graphene films were characterized by LEED in an ultrahigh vacuum chamber (8 × 10−9 Pa) using a BDL600IR (OCI, Canada). The chemical composition of the Ni and Cu foils was measured by XPS (Kratos AXIS-165). The depth profile was measured by etching the foils with Ar ions. Back-gated graphene FETs were made with a transferred graphene pattern using Au metal electrodes deposited through a shadow mask. Transport properties were measured using a B1500A semiconductor analyzer (Keysight Technologies) at room temperature and under vacuum conditions (∼1.5 × 10−4 Pa).
decomposition results in a selective nucleation of graphene at center of the Ni hole (arrows 1 and 2 in Figure S15, Supporting Information). On the other hand, the underlying Ni foil can trap the excess carbon species diffused from the top surface of Cu, possibly through GBs of the Cu foil. In addition, the presence of the bottom Ni foil prevents the bottom surface of the Cu to act as carbon supply to the top surface, as evidenced by the absence of graphene on the backside. These effects are expected to suppress the formation of bi/few-layer graphene and reduce the nucleation density of monolayer graphene (arrows 3 and 4).
CONCLUSIONS We demonstrate the possibility to control the location of singlecrystal graphene grains. This was achieved by stacking a perforated Ni foil mask on top of the Cu foil during the CVD. The presence of the Ni foil tunes the local concentration of CH4, decreasing the nucleation density as well as controlling the nucleation sites on the Cu foil. This allowed us to increase the CH4 concentration, speeding up the growth rate up to values comparable to the best ones obtained so far. The presence of a second Ni foil below the Cu completely suppressed the formation of bi/few-layer graphene in monolayer graphene. This also inhibited the growth of graphene on the back side of the Cu foil. Together, the use of the Ni foils realized one of the lowest reported nucleation densities for graphene on Cu foils (
