The effect of copper pre-cleaning on graphene synthesis

Graphene was synthesized by low pressure chemical vapor deposition using a copper foil (25 μm, 99.8%, Alfa Aesar) as a catalytic metal substrate. After the pre-cleaning process, the copper foil was annealed at 1000 ° C for 30 min under a 10 sccm hydrogen atmosphere (∼330 mTorr) to increase the grain size and to obtain a smooth surface, followed by graphene synthesis under 5 and 10 sccm of methane and hydrogen atmospheres for 40 min (∼450 mTorr) while maintaining the same temperature. During the cooling of the chamber, 10 sccm of hydrogen was flowed until the furnace was cooled to room temperature.

After the growth, the graphene was transferred from the copper to a SiO₂/Si substrate. Poly(methyl methacrylate) (PMMA) (4.5% in anisole) was spun onto the sample at 2500 rpm for 1 min, followed by etching the copper with copper etchant (CE-100, Transene). The film of PMMA/graphene was then washed with DI water. To further remove the residual iron particles which come from the copper etchant (FeCl₃), the film was floated on 10% HCl for 20 min, followed by neutralization with deionized water. After the PMMA/graphene was transferred onto the target substrate, the PMMA was removed by acetone vapor.

The sheet resistance values of the graphene films (1 × 1 cm²) were obtained using the van der Pauw method. The Hall voltage to extract the carrier concentration was measured with a magnet of 2000 G. To observe the surface morphology, high resolution field emission scanning electron microscopy (HRSEM, SUPRA™ 40, Zeiss) was used. The surface morphology of the copper foil was examined by atomic force microscopy (Dimension 3100, Veeco).

To estimate the continuity of the graphene films, copper etching tests were carried out. A drop of copper etchant (APS, Transene) was put on a graphene/copper foil for 5 s, followed by washing with DI water and blowing with N₂ gas to dry off the water. If the graphene layer does not completely cover the copper surface, the copper etchant will leak through, which results in etching pits on the surface.

Figure 1 compares the results of graphene growth with and without the Cu pre-cleaning step. In figure 1(a), SEM images of graphene grown on pristine copper foils (referred to as G-pristine Cu) of the same batch purchased from Alfa Aesar, but from four different runs under the same growth conditions, are shown. As mentioned previously, many white impurity particles with a large variation of size (10 nm–2 μm) can be seen in figure 1(a) (I shows the larger ones, in II–IV these particles can also be seen). Using SEM with energy-dispersive x-ray spectroscopy (EDX) (figure S1(a) available at stacks.iop.org/Nano/24/365602/mmedia) and Auger electron spectroscopy (AES) (figures S1(b) and (c) available at stacks.iop.org/Nano/24/365602/mmedia), it was found that these white particles on the copper surface are mostly metal particles, such as Si, Ca, Pt, Ru, and Ce. These impurities appear to play two roles during graphene growth: (1) as possible nucleation sites for mono- and bilayer graphene growth (for example, in figure 1(a), II and III, it can be seen there are white particles in the center of the bilayer graphene flakes) and (2) as possible obstacles to prevent continuous graphene growth (for example, in figure 1(a), III, the white thin lines are the incomplete graphene regions; some white particles can also be found in these regions; and in figure 1(b), I, the incomplete graphene region (when transferred onto SiO₂/Si substrate) is very likely due to the presence of the impurity particles which prevent graphene from growing underneath). Figure 1(b) shows the optical images of graphene grown on these pristine copper substrates transferred to 300 nm SiO₂/Si substrates (see SI method, available at stacks.iop.org/Nano/24/365602/mmedia). Figure 1(b) I corresponds to an incomplete region where graphene growth was hampered by the presence of impurity particles. In this region the graphene was not complete due to the presence of impurity particles, whereas in figure 1(b) II there are many broken regions in the graphene film. In our experiments, it appears that the graphene grown on the pristine copper foil tears more easily during transfer.

Zoom In Zoom Out Reset image size

There are two possibilities for the origin of these impurity particles. First, they could come from impurities distributed inside the copper foil and be driven out to the surface during the high temperature growth. Another possibility is that these impurity particles are on the Cu surface even before the CVD synthesis. To test which possibility is dominant, we used various etchant solutions to etch the Cu surface, and found that the impurity particles were significantly removed when Ni etchant (or nitric acid) was used. This suggests that the impurity particles have been on the Cu surface from the beginning (further discussion is presented later in this work). Figure 1(c) shows the SEM image of graphene synthesized on a pre-cleaned copper foil (referred to as G-cleaned Cu). A very clean and highly continuous film can be seen. In addition, the occurrence of bilayer graphene is much reduced. The optical image of this graphene transferred to 300 nm SiO₂/Si also confirms highly continuous graphene film (figure 1(d)). From these images it is very clear that the Ni etchant pre-cleaning is very effective in removing the impurity particles on the surface. Furthermore, the synthesis results with different batches of Cu became reliable when the pre-cleaning step was used.

In figure 2, a schematic diagram of the copper pre-cleaning process is shown (figure 2(a)) together with the SEM characterization of the corresponding Cu foil surface at each stage (figure 2(b)). As mentioned previously, our hypothesis is that most of the impurity particles were on the Cu surface. To remove these impurities, the foil was briefly dipped in Ni etchant solution for 30–90 s depending on the amount of impurities observed (see figure S2 available at stacks.iop.org/Nano/24/365602/mmedia). (Note: we have observed that Cu foil having more impurities requires a longer surface cleaning time. However, if the cleaning time is more than 2 min, the copper foil becomes severely damaged and is not useful for growth of graphene.) The main component of the Ni etchant is nitric acid, which reacts with Cu according to

$$\begin{eqnarray} &&\hspace{-11.5pc}{\mathrm{Cu(s)+4HNO}}_{3}(\mathrm{aq})\nonumber\\ &&\hspace{-1.5pc}\rightarrow \mathrm{Cu}({\mathrm{NO}}_{3})_{2}(\mathrm{aq})+2{\mathrm{NO}}_{2}(\mathrm{g})+2{\mathrm{H}}_{2}\mathrm{O}. \end{eqnarray} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

During this reaction, the Cu on the surface is being oxidized and dissolved in the solution; at the same time NO₂ gas is violently generated, helping to push the impurity particles away. In this way, the original surface of the Cu foil, together with the impurities, is scraped off and a new, clean but very rough surface is obtained. After cleaning, the copper foil is dipped in DI water for further washing. This washing process is carried out three times, each time with fresh DI water. It should be noted that this step is very important because if any residual material from the solution is left on the copper surface, these impurities affect the graphene growth as further nucleation sites. To remove the DI water from the copper completely, the copper foil is briefly washed with acetone and isopropanol and dried by blowing with N₂. The rough copper surface turns into a clean and smooth surface after postannealing under a hydrogen atmosphere, and is ready for growth of graphene.

In our experiments, various acidic and basic etchants were used to clean the copper foil, such as acetic acid, hydrochloric acid, chromium etchant (Chromium Etchants 1020, Transense), nitric acid, and sodium hydroxide. The SEM images in figure 3 show the copper surface after treatment with several etchants followed by the graphene growth. We found that nitric acid (5.4% (w/w)) and Ni etchant are most successful in removing the impurity particles, whereas the other etchants are not as effective. With acetic acid treatment, there is also a dark region along the copper rolling line (figure 3(a)). We confirmed that the dark regions are not bilayer graphene after transfer to 300 nm SiO₂/Si. These regions might be related to the residuals from acetic acid or some impurities. In table S1 (available at stacks.iop.org/Nano/24/365602/mmedia) the etching ability of each etchant is tabulated. Since it was found that the impurity particles contain Si, we used NaOH to clean the copper substrate to remove the Si. However, after NaOH treatment and graphene growth we still found Si impurities on the graphene/copper surface. There could be two reasons why nitric acid (or Ni etchant) is more effective than other etchants. First, it is a strong reactant and oxidizing agent, and can remove many different types of impurities. Second, during the etching, NO₂ bubbles are vigorously generated, thereby effectively preventing the impurities from depositing back on the Cu surface. The latter reason could be more crucial to result in effective cleaning.

Zoom In Zoom Out Reset image size

Different cleaning times were also investigated to obtain optimized results. In general, we found that 30 s was long enough to obtain a clean graphene surface, while with shorter times typically the white impurity particles were not completely removed (figure S2 available at stacks.iop.org/Nano/24/365602/mmedia). Nevertheless, it was also found that for certain batches of copper foil, the amount of impurity particles is much larger and requires a longer cleaning time. This is illustrated in figure S3 (available at stacks.iop.org/Nano/24/365602/mmedia), where Cu foils from two different batches (but with the same catalog number, Alfa Aesar, 25 μm, 99.8%) are shown for comparison. Even though the foils have the same purity (99.8%), after graphene growth it can be seen that batch I has a lot more impurity particles compared to batch II, both in SEM (figures S3(a) and (c) available at stacks.iop.org/Nano/24/365602/mmedia) and in AFM (figures S3(b) and (d) available at stacks.iop.org/Nano/24/365602/mmedia). It was found that a longer cleaning time (90 s) with simultaneous ultrasonication was necessary to obtain a reproducible clean surface for batch I while 30 s was long enough for batch II.

To evaluate the continuity of the as-grown graphene, we carried out a simple etch test using copper etchant (APS-100, Transense) on the graphene/copper surface [15]. After the CVD growth, if the copper surface is not completely covered by graphene, the copper etchant will leak through the openings and etch the copper underneath (as illustrated by the schematic diagram in the lower insets of figures 4(a) and (b)). For comparison, graphene samples grown from a pre-cleaned copper substrate (figure 4(a), G-cleaned Cu) and from a pristine copper substrate (figure 4(b), G-pristine Cu) were characterized by this method side by side. Figure 4(a) shows a typical AFM height image of G-cleaned Cu after the etching test; no etch pits could be found on the surface, indicating that graphene is highly continuous when grown using these pre-cleaned copper substrates. On the other hand, it can be seen in figure 4(b) that there are many etch pits for the graphene sample grown using a pristine copper substrate. This observation is consistent with the understanding that the impurity particles on the copper surface prevent the graphene film from closing up to form a seamless layer [15].

Zoom In Zoom Out Reset image size

The quality and continuity of the graphene were further examined by Raman characterization. Figures 5(a) and (b) display the Raman map of G-band intensity (1590 cm⁻¹) divided by Si peak intensity (520 cm⁻¹) for a 50 μm × 50 μm area for G-cleaned Cu and G-pristine Cu transferred to 300 nm SiO₂/Si, respectively. The color of the Raman map for the G-cleaned Cu film shows good uniformity over the whole area (figure 5(a)), and this is typical for all G-cleaned Cu samples. On the other hand, it is more common to observe broken graphene regions in the G-pristine Cu samples, and one of these regions is characterized in figure 5(b). As can be seen, the broken regions in the graphene film also result in folded or rolled-up regions. Figures 5(c) and (d) present the corresponding Raman spectra of specific sample positions in figures 5(a) and (b), respectively. It is noted that the D-band in the Raman spectra of both samples is quite low, indicating the good crystalline structure of the graphene samples.

Zoom In Zoom Out Reset image size

Electrical measurements were also carried out to characterize the effect of copper cleaning. Figure 6 shows the mobility versus the carrier concentration for the G-cleaned Cu samples (red circles) and G-pristine Cu samples (blue squares). The mobility carrier concentration values were obtained by Hall measurements using large area samples (1 × 1 cm²). Most G-cleaned Cu samples have high carrier mobilities and low carrier concentrations, whereas G-pristine Cu samples show lower mobilities and high carrier concentrations. The difference in the carrier concentrations for these two types of samples could come from the fact that there are more broken regions in the G-pristine Cu samples with a lot of active edge sites to be doped during the graphene transfer, whereas the basal plane of graphene is not easily functionalized or doped compared to the edges of graphene [16]. In addition, the sheet resistance values of the G-cleaned Cu are between 241 and 295 Ω/sq, whereas the G-pristine Cu samples have a much wider sheet resistance variation, between 368 and 650 Ω/sq. The higher sheet resistance values and their larger variation in magnitude could be due to the fact that G-pristine Cu has more incomplete/broken regions in the film.

Zoom In Zoom Out Reset image size

Finally, it is worth pointing out that this pre-cleaning method is very effective in removing the impurity particles on the Cu surface allowing the growth of continuous graphene with higher quality and reproducibility. As some of these impurity particles also serve as nucleation sites for the growth of bilayer or multilayer graphene, this method can also help to reduce the amount of bi- or multilayer graphene. However, the bilayer or multilayer growth cannot be completely eliminated, since this growth is also determined by the growth conditions. Further effort is still needed to gain better control over the growth of monolayer versus bilayer or multilayer graphene.