Scaling up graphene fabrication is a critical step for realizing industrial applications of chemical vapor deposition (CVD) graphene, such as large-area flexible displays and solar cells. In this study, a roll-to-roll (R2R) graphene transfer system using mechanical peeling is proposed. No etching of graphene growth substrate is involved; thus, the process is economical and environmentally benign. A prototype R2R graphene transfer machine was developed. Experiments were conducted to test the effects of relevant process parameters, including linear film speed, separation angle, and the guiding roller diameter. The linear film speed was found to have the highest impact on the transferred graphene coverage, followed by the roller diameter, while the effect of separation angle was statistically insignificant. Furthermore, there was an interaction effect between the film speed and roller diameter, which can be attributed to the competing effects of tensile strain and strain rate. Overall, the experimental results showed that larger than 98% graphene coverage could be achieved with high linear film speed and large guiding roller diameter, demonstrating that a large-scale dry graphene transfer process is possible with R2R mechanical peeling.
Since the discovery of graphene in 2004, numerous studies have been reported to investigate its extraordinary properties, as well as the synthesis and transfer methods of this carbon-based two-dimensional (2D) material [1–8]. One of the goals of graphene research is to utilize this material in the electronics industry for a wide range of applications, e.g., flexible transparent electrodes as a replacement of indium tin oxide in large-area display panels. Such applications require a reliable method for fabricating large-scale graphene of high quality. Chemical vapor deposition (CVD) has the proven capability for growing large-area monolayer graphene sheet using metal foil as the catalytic growth substrate [4,9]. After growth, the critical step is to transfer the as-grown CVD graphene from the metal surface to the target substrate for device fabrication.
The roll-to-roll (R2R) technique traditionally used in printing has recently gained popularity as it is finding advantages in manufacturing thin film electronics [10–13]. Similarly, R2R designs have been reported for high-throughput graphene fabrication, including both large-scale CVD graphene growth and polymer-assisted graphene transfer . Hesjedal  is among the first to use a R2R system for CVD graphene growth. Continuous graphene was grown on a polycrystalline copper foil under the atmospheric pressure. Many other reports on R2R graphene growth followed shortly after using atmospheric pressure CVD [15,16], conventional low-pressure CVD [17,18], and plasma-enhanced CVD [19,20].
As the next essential step, graphene transfer needs to be implemented using a R2R system; however, most of the current studies for graphene transfer involves wet chemical etching, where the as-grown graphene is submerged in a chemical bath to etch away the metal growth substrate [17,21–25]. An early R2R graphene transfer study was conducted by Juang et al.  with CVD graphene grown on nickel. Bae et al.  reported an R2R system using wet chemical etching with thermal release tape as the polymer backing layer to transfer graphene from copper foil to a polyethylene terephthalate (PET) substrate. Wet chemical etching was also used by Kobayashi et al.  to transfer a 100 m long CVD graphene from copper foil with photo-curable epoxy and PET. Besides wet chemical etching, electrochemical delamination, also known as the hydrogen bubbling method, and surface-energy-assisted hot water delamination have both been used for R2R graphene transfer [23,24].
Wet chemical etching requires dissolution of the metal growth substrate. The process is neither economical nor environmentally benign. Hot water delamination and hydrogen bubbling both involve extra equipment for liquid handling and drying, introducing additional cost and time to the process. In addition, hot water delamination relies on graphene defects to initiate the formation of copper oxide. The method is less compatible with high quality graphene and requires long time to operate. For high-speed graphene production, a mechanical transfer method is much more beneficial. Mechanical exfoliation is a common method for separating graphene patches from bulk graphite [26–30]. Small-scale dry peeling of double-cantilever graphene samples has been used to measure the interfacial facture toughness of as-grown graphene on copper [31–33]. Juang et al. used a cold rolling process for R2R transfer of few-layer graphene samples .
In this study, an R2R dry transfer process using mechanical peeling is proposed for single-layer graphene fabrication. An R2R machine is designed and experiments are conducted to examine the effect of process variables including the linear film speed, separation angle between the copper foil and polymer film, and the diameter of a guiding roller. The effect of each process parameter is investigated along with the interactions among them in the R2R graphene dry peeling process.
The Dry-Peeling Machine.
Figure 1 shows the prototype dry peeling machine that was developed in this study. Figure 1(a) shows a schematic of the machine. There are four rollers in the system, an unwinding roller carrying the prepared sample (R1), an idler roller guiding the film (R2), and two rewinding rollers (R3 and R4) collecting the separated copper foil and graphene/polymer film. Each rewinding roller is actuated by a NEMA 34 stepper motor (STP-MTRH-34066, Cumming, GA) controlled by a micro-controller (Arduino Uno based on ATmega328P) and a dual H-bridge micro-stepping drive (STP-DRV-6575, Cumming, GA). The power was provided by a 48 V linear power supply (STP-PWR-4810, Cumming, GA). Rotary incremental quadrature encoders (TRDA-2E500VD, Cumming, GA) were individually attached to the idler rollers via timing belts to monitoring the speeds of the copper and polymer films, respectively. A customized tension roller by Dover Flexo Electronics Inc. (TR0-2.25-25-6:00-RD3, Rochester, NH) with both a 5 V analog output signal and a physical tension indicator (TI23E-25, Rochester, NH) was installed in the R2R system to monitor the in situ tension developed in the PET carrier film after graphene delamination. Figure 1(b) shows a pictorial view of the prototype machine.
Dry-Peeling Graphene Sample Preparation.
The graphene samples used in this study were grown following a low-pressure chemical vapor deposition procedure . Oxygen-free copper foil of 127 μm thick (Alfa Aesar 13380, Tewksbury, MA) was used as the copper growth substrate. Glacial acetic acid (Sigma-Aldrich 537020, St. Louis, MO) was used to remove the oxidation layer on copper foil before the CVD graphene growth. A 1:1 volume ratio of hydrogen and methane at 5 sccm were used with a peak growth temperature of 1030 °C.
Instead of using rolls of graphene samples that would be cost prohibitive, CVD graphene samples of 12 mm × 25 mm in size were used in this study. Figure 2 illustrates a three-step sample preparation procedure for R2R graphene dry peeling experiments. As-grown CVD graphene specimens were first hot laminated on commercial PET/ethyl vinyl acetate (EVA) film (Scotch Thermal Laminating Pouches TP3854-100, 3M, Maplewood, MN) using a hot laminator (GBC HeatSeal H425, Aylesbury, UK), as shown in Fig. 2(a). Next, the PET/EVA layer on one side of the sample was manually removed, as shown in Fig. 2(b), to examine the graphene growth quality. PET/EVA film was chosen as the polymer support for its strong bonding strength to graphene to achieve complete graphene delamination by mechanical peeling. It was also used as the polymer backing layer for R2R graphene transfer and for adhesion energy measurements in previous studies [23,24,34]. The last step of the sample preparation process involved sandwiching the prepared copper/graphene/EVA/PET samples between a top PET carrier film (MYLAR® A (TEKRA, New Berlin, WI), 254 μm thick) and a bottom copper carrier foil (Grainger 4 UGU3, 76.2 μm thick) with the help of a permanent double-sided pressure-sensitive adhesive (PSA, Scotch Tape 6137H), as shown in Fig. 2(c). The prepared Cu/graphene/EVA/PET roll was then loaded onto the R2R machine for experiment.
An added benefit of using discrete sample patches for testing is that they can be placed to leave adequate room for making process adjustments during the R2R peeling process such that the effect of these adjustments can be easily identified. Moreover, CVD graphene samples of smaller sizes could be grown with more consistent quality due to the fact that the entire samples could fit in the evenly distributed temperature zone in a high-temperature tube furnace that is used for graphene growth. A more uniform contact between the graphene growth substrate and carbon feedstock could be achieved with small sample patches.
A decentralized automatic control scheme was implemented for the R2R system to control the linear film speeds of both the copper foil and the polymer backing layer, as well as the tension in the polymer carrier film after the graphene delamination. Figure 3 shows a diagram of the control scheme. It was essential to keep the copper foil and polymer film travel at the same linear speed so that there is no sagging or bending of the two separated films. Maintaining a constant rpm for the two motors would lead to gradual increase in the linear film speed due to the growing diameter of the rewinding rolls. Different thicknesses of the copper foil and graphene/EVA/PET films could also lead to different linear film speeds. Therefore, instantaneous measurements from both encoders were fed into the microcontroller every 0.2 s and compared with the target value specified for each experiment. Corrections on the microstepping drive output frequency were made if there was any detected deviation from the target speed. The tension transducer was used to monitor the in situ tension developed in the PET film that carried the delaminated graphene. Based on the geometry of the R2R system, the compression force measured by the tension roller was converted to the film tension. To protect the transferred graphene on the polymer backing layer, a 0.2% elongation was used as the maximum allowable value. If the measured tension exceeded this threshold, the speed of the stepper motor driving the polymer film would be reduced to release the tension build-up.
Parametric studies were conducted on the R2R dry peeling machine to understand the effects of process parameters including linear film speed, separation angle, and the diameter of the guiding roller (R2 in Fig. 1(a)), as well as the interaction among them. The effect of film speed was first studied by fixing the separation angle and roller diameter and varying the film speed from 0.42 to 1.72 m/min. Similarly, the effect of separation angle was studied by fixing the film speed and roller diameter and varying the separation angle from 30 deg to 45 deg. In addition, a two-level full-factorial experiment was conducted to determine the average effect of each parameter and the interaction effects among them. Table 1 summarizes the experimental conditions of the tests.
The samples transferred under different conditions were characterized with scanning electron microscope (SEM) (Quanta FEG 600) using water under the low vacuum mode for surface charge dissipation. The image contrast was significantly improved compared with the characterization under the high vacuum mode. Park Scientific atomic-force microscopy (AFM) was used to characterize sub-50 nm features on as-grown CVD samples. Raman spectroscopy (Alpha 600) was also used to confirm the presence of graphene on PET/EVA film with an argon-ion laser of 488 nm wavelength for excitation. Graphene coverage on PET/EVA after the transfer was used as a response variable of the experiment. The image processing software Image J was used to convert SEM images of transferred graphene to 16-bit grayscale images and a rectangular area of interest was selected for characterization, following a procedure developed in Ref. .
Results and Discussion
Characteristics of As-Grown Graphene.
Figure 4 shows typical surface characteristics of the graphene grown in this study. The low-magnification SEM images shows mosaic polycrystalline copper grains along with parallel tooling marks generated during the copper rolling process. As the magnification increases, graphene ad-layers become visible, which have a darker tone and quasi-hexagonal shape. Furthermore, graphene wrinkles formed as a result of thermal mismatch are observed in the high-mag SEM image. The AFM image shows intrinsic copper steps after the CVD graphene growth. Copper surface restructured as graphene islands expanded and coalesced at 1030 °C. These characteristics confirmed that high quality graphene specimens were obtained in the CVD growth process used in this study.
Characteristics of Transferred Graphene.
Figure 5 shows an SEM image of dry peeled graphene on PET/EVA and its corresponding Raman spectrum. Darker regions seen in Fig. 5(a) were transferred graphene and bright regions were vacancies where electric charges accumulated on the exposed insulating EVA layer. Graphene ad-layers and wrinkles were also clearly visible in the SEM image. The delaminated ad-layers were likely composed of more than two layers of graphene, but only the top few layers were peeled off copper. Figure 5(b) shows a single-scan Raman spectrum of transferred graphene on PET/EVA, where pronounced G (1580 cm−1) and 2D (2690 cm−1) peaks were clearly observed . When the focus of the laser beam was lowered from the top graphene layer into the EVA layer, the intensity of G and 2D peaks was drastically reduced while other peaks remained relatively constant. This suggested that these other peaks in the Raman spectrum were due to the EVA backing layer and confirmed the presence of transferred graphene on top of the PET/EVA backing layer.
The Effect of Film Speed.
Figure 6 shows the effect of film speed while the separation angle and roller diameter were fixed at 45 deg and 23 mm, respectively. As can be seen from the SEM images, large patches of graphene were transferred to the PET/EVA film as the linear film speed increased from 0.43 m/min to 1.72 m/min. In addition, parallel cracking marks were observed in the graphene transferred at lower speeds. At the highest speed of 1.72 m/min, the parallel cracking marks disappeared and a continuous graphene sheet was transferred with microcracks that could only be seen at the higher magnification.
Using Image J, the transferred graphene coverage rate was calculated under each film speed on a surface area of ∼100 × 100 μm, and the results are shown in Fig. 7. The relationship between the graphene coverage rate on PET/EVA and the linear film speed is almost linear. From 0.43 m/min to 1.72 m/min film speed, the graphene coverage rate increased from less than 50% to approximately 90%.
Effect of Separation Angle.
With the film speed and guiding roller diameter kept constant at 2 m/min and 23 mm, respectively, the angle of separation between the copper foil and polymer support film was varied from 30 deg to 45 deg. Since a higher film speed was shown to yield a higher graphene coverage rate, a linear speed of 2 m/min was used for all the experiments with different separation angles. Figure 8 shows the SEM images of transferred graphene under these conditions. As a result, higher coverage rates were observed in all the SEM images than the previous cases. Figure 9 shows the measured graphene coverage rate at different separation angles. The coverage rate increased from the 30 deg to 35 deg separation angle setting. Beyond the 35 deg angle setting, the graphene coverage rate appeared to have a decreasing trend with the increasing angle. In other words, 35 deg separation angle produced marginally higher graphene coverage based on current R2R setup. However, this trend was not statistically significant.
Average Main and Interaction Effects.
To explore the average main and interaction effects, a two-level full-factorial design of experiment was conducted with film speed, separate angle, and guiding roller diameter as the parameters. Graphene coverage was again used as the response variable. Table 2 summarizes the experimental results. Each condition was repeated three times. The average and standard deviation were obtained from three samples under the same condition.
Figure 10(a) shows the average main effects of the process parameters. The most obvious trend in the data is that a higher film speed resulted in a higher graphene coverage rate, as has been observed previously. A smaller guiding roller diameter corresponds to a slightly decreased graphene coverage rate. The effect of the separation angle between copper foil and polymer film does not show an obvious trend. Figure 10(b) shows the interaction effects among the process parameters. It is seen that film speed and roller diameter appears to have the strongest interaction effect, followed by speed and angle and angle and diameter. The interaction effect of film speed and roller diameter suggests that the speed effect is much stronger when a smaller roller diameter is used. A statistical analysis was conducted using the experimental data. The result confirmed the above observations. The main effects of Speed and Diameter are statistically significant at the 5% significance level. The interaction effect of speed and diameter (Speed*Diameter) is significant with a statistical significance of 6.56%.
where y0 is the distance from the graphene layer to the neutral plane and R is the bending radius. In our study, y0 is the same for both the small and large rollers. Therefore, a smaller roller will result in a larger tensile strain. With the parameters used in this study, the strain experienced by the graphene layer is determined to be εsmall = 0.58% for the smaller roller and εlarge = 0.26% for the larger roller. In addition to the tensile strain generated by the roller, the roll-to-roll transfer process also introduces tension due to the stretching of the film. As strain increases, more cracks will be formed . Thus, a smaller guiding roller will introduce more cracks and leave more cracked graphene patches on the copper surface.
The interaction effect between the film speed and roller diameter reveals the competing significance of the two parameters. At a low film speed, roller diameter plays a more significant role affecting the graphene cracks and thus the coverage rate. At high speeds, the adhesion energy between the PET/EVA film and graphene becomes so high that even cracked graphene is able to adhere to the PET/EVA film. Therefore, the linear film speed is the dominant factor on the graphene coverage at higher film speeds, whereas avoiding cracks by choosing a large guiding roller diameter is critical to successful graphene transfer at low film speeds.
Mass production of high-quality graphene has always been a challenge for the industry to realize graphene-based electronics. R2R fabrication has been suggested as an efficient and scalable manufacturing technique for CVD graphene growth and transfer. In this study, R2R transfer of CVD graphene by mechanical dry peeling was proposed. A prototype R2R dry transfer machine was developed and experiments were conducted to determine the effect of linear film speed, separation angle, and guiding roller diameter on transferred graphene coverage rate. It was found that the linear film speed had the highest impact on the graphene coverage rate due to the strain rate effect of the polymer carrier layer. A larger guiding roller is generally preferred for the R2R graphene transfer process because it introduces lower tension strain on the graphene layer. The effect of separation angle is statistically insignificant over the range of values tested in this study. Overall, a larger than 98% graphene coverage rate was achieved with the R2R dry peeling machine under 3 m/min film speed and 51 mm guiding roller diameter, demonstrating that R2R dry peeling is a promising technique for future high-throughput graphene fabrication.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. The authors would also like to acknowledge Dr. Richard Piner for assistance with SEM and Raman characterization of the graphene samples.
National Science Foundation (EEC-1160494).