A simple method to tune graphene growth between monolayer and bilayer

,


INTRODUCTION
In addition to the fascinating monolayer graphene, bilayer graphene has attracted increasing attention. AB-stacked bilayer graphene shows a tunable band gap in a vertical electric field. 1,2 This makes it applicable in devices such as field effect transistors. On the other hand, twisted bilayer graphene also exhibits interesting physics such as the twist-dependent Van Hove singularity. 3,4 Bilayer graphene can be prepared by mechanical exfoliation of graphite. 5 However, this method is unsuitable for mass production. Chemical vapor deposition (CVD) growth on metals has been another important method to make bilayer graphene.  Graphene grown on nickel is usually multilayer because of the high solubility of carbon in Ni. In contrast, graphene grown on copper is typically monolayer due to the "self-limiting" effect 30 Cu-Ni alloy has been used to grow bilayer graphene [6][7][8] by carefully adjusting the atomic percentage. Recently, researchers have also grown bilayer graphene on copper by regulating various conditions including the carbon source, 9,10 copper purity, 11,12 pressure, 13 growth temperature 14 and gas flow rate. 15 However, these methods either require much prolonged growth processes or produce bilayer films which are not continuous because the second layer is much smaller than the first. Meanwhile, these recipes for bilayer growth are quite different from those for monolayer and usually require different setup. Therefore, high yield growth of continuous bilayer graphene films remains a topic of intensive research efforts.
Here we report an ambient pressure chemical vapor deposition (APCVD) process with the capability to tune from monolayer growth to bilayer growth. We used the sample temperature at which to start the H 2 flow as the sole control parameter. The obtained samples were characterized by optical microscopy, Raman spectroscopy, scanning tunneling microscopy (STM) and scanning electron microscopy (SEM). It was revealed that annealing the sample without the H 2 flow will increase the degree of Cu oxidation and consequently inhibit the formation of a second graphene layer.

EXPERIMENT
Graphene samples were grown on copper foils (127 µm thick, 99.9%, Alfa Aesar) in a furnace by an APCVD process as schematically shown in Figure 1(a). The furnace was closed but not air sealed or pumped. The Cu foils were heated from room temperature to 1050 • C with Ar and H 2 flow of 500 sccm and 20 sccm, respectively. The temperature was kept at 1050 o C for 40 minutes or longer. The Ar flow was on during the whole process. The starting temperature of the H 2 flow was varied as the control parameter. The other conditions were kept the same. The growth process lasted for 20 minutes with an Ar/H 2 /CH 4 flow rate of 500 /20/ 10 sccm. After growth, the samples were cooled naturally to room temperature in about 4 hours with the Ar /H 2 flow on.
Raman spectra were taken by a Renishaw inVia system with the excitation wavelength of 514 nm. Optical images of graphene on Cu were taken by an Olympus microscope (Olympus BX51). STM and STS measurements were carried out in a UNISOKU UHV STM system at 77 K with a base pressure below 3 × 10 -7 Pa. SEM measurements were carried out in a FEI Helios NanoLab 600i system.  Figure 1(b) and 1(c) show the typical optical micrographs of the samples prepared with the H 2 flow starting at 1050 • C (samples S1050) and room temperature (samples S25), respectively. As it turned out, S1050 and S25 are uniform monolayer and bilayer graphene films, respectively.

RESULTS AND DISCUSSION
To characterize the quality and layer number of the graphene samples, we have performed Raman spectroscopy measurements. Figure 1(d) shows representative Raman spectra for S1050 and S25 samples with the excitation wavelength of 514nm. The two peaks around 1590 cm −1 and 2700 cm −1 correspond to the Raman G band and 2D band of graphene, 31,32 respectively. The intensity ratio of the 2D band and G band I 2D /I G for S1050 (the dark yellow curve) is 2.2 and the full width at half-maximum (FWHM) of the 2D band is 37 cm −1 . These are characteristics of monolayer graphene. 13,27 For S25 (the orange curve), I 2D /I G is 0.83 and the FWHM of the 2D band is 60 cm −1 , characteristics of bilayer graphene. 13,27 The negligible D band around 1350 cm −1 in both curves indicates low defect density and high quality of the graphene samples.
Further analysis of the spectra reveals that the 2D band of S1050 can be well fitted by a single Lorentz curve (Figure 1(e)) while that of S25 cannot. Instead, Figure 1(f) shows that the 2D band of S25 can be excellently fitted (shown in orange curve) with four Lorentz curves, whose FWHM are all 34 cm −1 (shown in green curves). This verifies that S25 are indeed bilayer graphene samples. 14,17,26,31 Raman mapping was used to check the uniformity of the above monolayer and bilayer samples and the I 2D /I G results are shown in Figure 2(a) and 2(b). Figure 2(a) corresponds to monolayer samples S1050. The I 2D /I G value is larger than 2 all over the surface. In addition, the FWHM value of the 2D band is ∼30 cm −1 . These mapping results show that more than 98% of the surface of the S1050 samples is covered by monolayer graphene. Figure 2(b) corresponds to bilayer samples S25. The I 2D /I G value is mostly between 0.7 and 1.3. The FWHM value of the 2D band is mostly between 45 cm −1 and 65 cm −1 . These mapping results indicate that more than 95% of the surface of the S25 samples is covered by bilayer graphene. The uniformity of the samples shows that our method can be used for high yield selective growth of monolayer or bilayer graphene. Next, the stacking order of the bilayer samples was checked. Since the I 2D /I G value is ∼1, most area should be AB stacked, not twisted. 6,13,14 Results of scanning tunneling microscopy (STM) measurements are consistent with the Raman results. Most area is AB stacked as shown in the typical STM image and its Fast Fourier Transform (FFT) result in Figure 2(c). Only one sublattice of graphene shows up in the image, a consequence of Bernal stacking. 6,33,34 As comparison, both sublattices are visible on twisted bilayer graphene (Figure 2(d)). The K-point energy in the scanning tunneling spectrum (STS) results in Figure 2(e) is within 50 meV of the Fermi level. This indicates little doping of the graphene samples, consistent with the Raman results. Occasionally twisted bilayer area was also found. Figure 2(f) shows the STM image of such an area with a Moiré pattern caused by the twist between the layers. So, just by changing the temperature at which we turn on the hydrogen gas flow during the heat-up process, we can selectively grow continuous monolayer or bilayer graphene films by APCVD. If we start the H 2 flow at room temperature, bilayer graphene is obtained. If we start the H 2 flow at 1050 • C, monolayer graphene is obtained. At first thought, maybe the difference is due to the total time duration of the H 2 flow? We started the H 2 flow at 1050 • C and annealed the Cu foil for 1hr longer to compensate for the total H 2 flow time. Still monolayer graphene was obtained. So the decisive factor should be the starting temperature (not the time duration) of the H 2 flow.
Next we have done a series of growth experiments with the H 2 starting temperature in the range between 25 • C (room temperature) and 1050 o C to check at which value the transition (from monolayer to bilayer growth) occurs. Figure 3(a) shows the typical Raman spectra of these samples. When the H 2 starting temperature is above 700 • C, continuous monolayer graphene is grown FIG. 3. (a) Raman spectra of graphene samples. The H 2 flow was turned on at different temperatures. The spectra are vertically displaced for clarity. When the H 2 starting temperature was lower than 350 • C, the I 2D /I G ratio is ∼ 1.0 all over the surface, and the FWHM of the 2D band is ∼60 cm −1 , suggesting continuous bilayer graphene films. When the H 2 starting temperature was higher than 700 • C, the I 2D /I G ratio of >2. (I 2D /I G > 2.2, 2D band FWHM ∼40 cm −1 ). When the H 2 starting temperature is below 350 • C, continuous bilayer graphene is grown (I 2D /I G ∼ 1.0, 2D band FWHM ∼60 cm −1 ). When the H 2 starting temperature is between 400 • C and 600 • C, we find both monolayer and bilayer areas on the surface as shown in Figure 3(b). The darker and brighter areas are graphene bilayers and monolayers, respectively. STM measurements further verified the layer number of the graphene samples. The step height in Figure 3(c) is 0.35 nm, indicating just two layers.
What is the mechanism behind this temperature dependent behavior? It is known that hydrogen can significantly affect graphene's growth and properties. 13,15,35,36 However, since the H 2 flow conditions were kept the same after the carbon source (CH 4 ) was introduced, it should be something else that caused the difference between monolayer and bilayer growth. Due to unavoidable outgasing in the furnace at elevated temperatures and air leak during the APCVD process, oxygen exists in the furnace especially when H 2 flow is not on. The Cu foil can get oxidized and affect graphene's growth and properties. 20,37,38 Recently, people have found that oxygen in Cu can reduce the nucleation density for graphene growth. 20,37,38 To verify that oxidation is indeed the dominant difference happened to the copper foils, we heated a piece of reference copper foil to 1050 • C in the furnace with only the Ar flow (500 sccm), then turned off the power and cooled to room temperature naturally. The optical micrograph of the copper foil treated this way is shown in Figure 3(e). For comparison, the original Cu foil is shown in Figure 3(d). We can see that the copper foil changed color and was oxidized by heating without H 2 . Figure 3(f) shows the Raman spectra of the copper foils. Peaks corresponding to Cu 2 O 39 appear on the heated copper foil but not on the original one.
We believe that the transition from bilayer to monolayer growth is caused by oxygen. With the increase of the H 2 starting temperature, the oxidation of the copper foil gets more severe since the oxidation is a barrier activated process. When the temperature reaches a certain value (400 • C ∼ 600 • C), the lattice diffusion mode of the oxygen atoms in Cu 2 O is activated in addition to the the grain boundary diffusion mode, resulting in a sharp increase of the oxidation rate 40 and the oxygen solubility in Cu. 41,42 Oxygen in Cu will then hinder the carbon diffusion and nucleation, effectively inhibit the formation of a second layer. To further investigate this effect, the substrate was heated to 550 • C and maintained there for various periods of time to allow various degrees of Cu oxidation. Then the H 2 flow was turned on and the Cu foil was further heated to 1050 • C. Other conditions were kept the same as in the normal growth process (Figure 1(a)). Figure 4 shows the SEM results of the samples. The starting temperatures of the H 2 flow for the samples in Figure 4(a), 4(b), 4(c), 4(d) were 25 • C, 550 • C, 550 • C and 1050 • C, respectively. The sample in Figure 4(c) was kept at 550 o C for additional 2 hrs before starting the H 2 flow. Therefore samples shown in Figure 4(a), 4(b), 4(c), 4(d) have increasing Cu oxidation. And there is a clear trend of decreasing bilayer formation (the darker areas). In fact, additional 2 hrs of oxidation at 550 • C (annealing without the H 2 flow) can efficiently reduce the bilayer area to < 10% as shown in Figure 4(c). Because the oxidation rate at higher temperatures (>700 • C) are exponentially higher, even a few minutes without the H 2 flow will result in the absence of bilayer. This explains the transition from bilayer to monolayer growth quite well.

SUMMARY
We developed a method to readily tune large area graphene growth between monolayer and bilayer. Using the starting temperature of the H 2 flow in an ambient pressure CVD process as the single control parameter, we observed the change from monolayer to bilayer growth on Cu foil. When the H 2 starting temperature was above 700 • C, continuous monolayer graphene films were obtained. When the H 2 starting temperature was below 350 • C, continuous bilayer films were obtained. Detailed Raman and SEM study of samples treated under various conditions revealed that Cu oxidation played a crucial role in this change. This effect can be well understood since Cu oxidation will inhibit carbon diffusion thus also the formation of a second graphene layer.