Establishment of a reliable transfer process for fabricating chemical vapor deposition-grown graphene films with advanced and repeatable electrical properties

Graphene films grown by the chemical vapor deposition (CVD) method have attracted intensive attention due to their native advantages of both high quality and large quantity for commercial applications. However, previously reported graphene films have exhibited uncertain and conflicted electrical properties that greatly hinder them from being used to build reliable electrical devices because of incompatibility during the complex and multifarious transfer process. Herein, the relationship between the transfer parameters and electrical performance was systematically studied. It demonstrates that cracking during the transfer process causes significant loss of carrier mobility and hence an increase in sheet resistance. Additionally, unstable doping plays a key role in the carrier density and hence greatly influences the sheet resistance. By introducing HCl as a doping agent, graphene films with repeated sheet resistance of approximately 300 ohm sq−1 can be realized. This work establishes a facile and reliable route to fabricate graphene films with advanced and repeatable electrical properties, which is significant and essential for fair evaluation of CVD-grown graphene films and further practical applications.


Introduction
Recently, there has been great interest in graphene because of its impressive mechanical and physical properties. [1][2][3][4] Among various fabrication approaches of graphene lms, the chemical vapor deposition (CVD) method has been considered to be one of the most important approaches because graphene lms of both high quality and large quantity are obtained by its use. [5][6][7] It has been predicted that CVD-grown graphene lms will play a signicant role in the operation of the optoelectronic, electrical, and exible devices that are anticipated for the future. 8,9 It is known that before proceeding to practical devices, CVDgrown graphene lms must be transferred from catalyst substrates to desirable substrates. 10,11 As a typical example, monolayer graphene grown on a Cu substrate has been widely studied and transferred to polyethylene terephthalate (PET) substrate that can be used as exible transparent conducting lms for next-generation exible electronics. 12,13 However, this process usually leads to an unavoidable decrease in electrical performance due to uncertain cracks formation and pollution problems during the process. [14][15][16] Various transfer methods have been developed to solve those problems, including the polymer-assisted transfer method, 17,18 the thermal release tape (TRT) method, [19][20][21] and the glueassisted method. 22,23 Earlier methods involved growing graphene lm on a Ni substrate and transferring it to a Si substrate by employing polydimethylsiloxane (PDMS) as an intermediate support layer. 24 Polymethyl methacrylate (PMMA) has subsequently been widely accepted and used instead of PDMS to transfer graphene lms and other ultrathin lms, because it can be easily removed by solvents while maintaining clean and complete samples. 25 However, the size of the resultant lms in these approaches is usually limited to a few centimeters, which is not sufficiently large to be used as a next-generation transparent conducting lms for building practical devices. It was rstly reported by Samsung that large scale graphene lms with sizes up to 30 inches were successfully fabricated using thermal release tape as an intermediate transfer layer, which made it possible to build an operable and useful touch panel. 20 This impressive progress had attracted broad attention, but the difficulty of operation led to some conicting results in the reports that followed. Pressure-sensitive adhesive lms and silicone lms have been used as improved intermediate layers to produce graphene lms that exhibit more advanced completeness and electrical properties. 26 Moreover, Sony Corporation produced a 100 m-long high-quality graphene transparent conductive lms by roll-to-roll CVD and transfer with the assistance of UV-glue bonding. 23 Undoubtedly, the correct selection of the transfer method plays a key role in determining the nal device performance. 11,[27][28][29] In addition to the numerous transfer methods, another nonnegligible and irreplaceable function comes from the etchant for removing catalyst substrates, which usually contains iron chloride (FeCl 3 ), 30 iron nitrate (Fe(NO 3 ) 3 ), 31 or ammonium persulphate ((NH 4 ) 2 S 2 O 8 ). 20 For instance, Youngsoo Kim et al. reported a monolayer graphene lm with sheet resistance of 923 + 148 U sq À1 by transferring CVD-grown graphene with PMMA and etching a Cu substrate with 0.02 mol L À1 (NH 4 ) 2 S 2 O 8 solutions. 32 For the same growth and transfer process, other studies revealed diverse results with the sheet resistance of graphene lms of 1520 + 67.7 U sq À1 . 33 Similar phenomena appeared in other etching systems. Li et al. removed the Cu substrate using a 0.05 g ml À1 Fe(NO 3 ) 3 solution for 12 h and obtained graphene lms with a sheet resistance of approximately 2100 U sq À1 . 34 However, the sheet resistance of graphene lms from other reports was 425 U sq À1 . 35 Moreover, for the FeCl 3 etching system, two studies reported the sheet resistance of graphene lms as 510 U sq À1 and 725 U sq À1 . 36,37 Aer broadly reviewing current reports describing the transfer of graphene lms, it is noted that the resultant graphene lms have quite different electrical properties.
Obviously, the fabrication of graphene lms with unstable and uncertain physical properties would greatly prevent it from being used for practical applications. Thus, it is quite necessary and signicantly important to establish a clear relationship between the transfer parameters and the electrical performance. In this work, by systematic study, a standard transfer process for graphene lms was successfully established with repeated sheet resistance, which is signicantly important for evaluating the quality of CVD-grown graphene lms and also for the use of next-generation transparent conducting lms in practical applications.

Experimental
Graphene lms were synthesized through the traditional CVD method. 5 A 25 mm-thick Cu substrate was heated to 1000 C in H 2 gas with a ow rate of 100 standard cubic centimeters per minute (sccm). Aer reaching 1000 C for 10 minutes, 30 sccm CH 4 was introduced for 30 minutes to grow graphene on the Cu surface. Finally, the Cu foil with graphene on its surface was cooled to room temperature.
The CVD-grown graphene lm on the Cu was transferred to a PET substrate through different approaches, including the PDMS method, 24 PMMA method, 31 TRT method 20 and UV glue method. 23 For the typical UV glue method, a Cu foil with graphene lm on its surface was bonded onto PET lm with UV glue (HUITIAN, 3118). Subsequently, the epoxy was cured using a UV curing machine (GLINTSUN, GS-300-2PM). Then, the Cu was etched by different etchants, including ammonium persulphate (NH 4 ) 2 S 2 O 8 , iron chloride (FeCl 3 ), and iron nitrate (Fe(NO 3 ) 3 ) solutions. Finally, the graphene/PET lm was washed with deionized water several times to remove any etchant residuals and was then dried under an atmospheric environment.
The sheet resistance of graphene on a 5 cm Â 5 cm square of PET was measured with a four-point probe (Suzhou Jingge Electronic Co., Ltd., ST2258A) at room temperature. The sheet resistance mapping measurement was carried out with a sheet resistance mapping system (SURAGUS, EddyCus TF map 2525 SR). The morphology of graphene on the PET lm was characterized by eld emission-scanning electron microscopy (FE-SEM, Hitachi S-4800) and atomic force microscopy (AFM, Dimension 3100). Raman spectra were obtained on graphene lms transferred by the PMMA-assisted method using a Renishaw microspectrometer with a laser wavelength of 514 nm. X-ray photoelectron spectroscopy (XPS) measurements were performed using a Kratos Axis Ultra delay-line detector (DLD) spectrometer (Kratos Analytical). The Hall Effect Measurement System (HALL 8800) was used to measure the carrier mobility and carrier density of graphene lm with 7200 magnetic strength.  different samples reveals a similar value of approximately 8 Â 10 12 cm À2 . It different from the carrier density, the carrier mobility of different samples will exhibit a huge gap. It can be seen that the carrier mobility of graphene transferred by the UVglue method reaches the highest value of approximately 1800 cm 2 V À1 s À1 , which is six times higher than that of the PDMS method. These results clearly demonstrate that the large difference in sheet resistance originates from the change in carrier mobility while changing the transfer process.

Results and discussion
Because the graphene lms on Cu foil are fabricated through exactly the same CVD process, it is quite necessary to investigate the resultant graphene lms on PET in detail to further understand their nature. As shown in Fig. 1c-f, the corresponding surface morphologies are characterized by FE-SEM. It should be mentioned that no additional conducting layer was sputtered on the surface in order to ensure the original conditions. Greatly inhomogeneous surfaces with obvious contrast difference can occur, especially in PDMS sample. It is known that poor electrical conductivity during SEM leads to an electrical charge effect. 21,38 Here, the brighter area might suggest that a poor electrical connection exists, which is due to the cracks that occur during the transfer process. The block of electrical connection would greatly inuence the carrier transport and hence decrease the carrier mobility. For comparison, the SEM image of the UV-glue sample is quite uniform on the entire surface with slight wrinkles, which suggests improved carrier mobility. Thus, the nal sheet resistance can be significantly different for samples prepared by different transfer methods. AFM measurements were also performed to analyze the microstructures of transferred graphene lms, such as ripple or domain boundary, 39,40 as shown in the insets of Fig. 1cf. The AFM image of graphene lms transferred by the PDMS method exhibits a large oscillation with dense and obvious contrast in some areas. The contrast oscillation becomes much less pronounced in the samples transferred by the TRT and PMMA methods. Finally, it looks quite smooth on the UV-glue assisted transferred graphene lm with only slight wrinkling, which is due to the mismatch of the coefficient of thermal expansion between copper and carbon. The results are consistent with the SEM results of graphene lms using different transfer methods.
The electrical performance of graphene lms using different etchants with the same UV-glue transferring method is compared in Fig. 2. For all samples where different etchants were used, including (NH 4 ) 2 S 2 O 8 , FeCl 3 , and Fe(NO 3 ) 3 solutions, the sheet resistance of the graphene lms gradually decreases as the concentration of etching solutions was increased. Among these three etchants, the graphene lms etched in (NH 4 ) 2 S 2 O 8 exhibited the corresponding lowest sheet resistance. The carrier density and carrier mobility of these samples were investigated in detail, and as shown in Fig. 2b, the carrier mobility of graphene lms using the same etchant did not greatly change as the etchant concentration increased, and that of graphene lms etched in (NH 4 ) 2 S 2 O 8 presents the highest value among these three etchants. In contrast, the carrier density of these graphene lms increased signicantly as the corresponding etchant concentration increased, as shown in Fig. 2c. Although the carrier density of graphene lm etched in (NH 4 ) 2 S 2 O 8 is lower than that of the other samples, the lower sheet resistance is still realized in graphene lms etched in (NH 4 ) 2 S 2 O 8 , which originates from the large enhancement of carrier mobility. It can be concluded that an increase in ion concentration may not affect the carrier mobility, which remains with an unchanged electrical connection, but greatly enhances the carrier density and hence reduces the sheet resistance.
Based on the above analyses, to further improve the electrical conductivity of graphene lms, it is very important to nd a method that will increase the carrier density without the loss of carrier mobility. This requirement can be satised by introducing dopant during the transfer process, and this has been broadly studied. 20,41,42 During the above studies, it was found that the pH value decreases as the etching process proceeds. Thus, H + ions as dopant were systemically studied herein to determine if this could further improve graphene lms. Fig. 3ac presents the electrical properties of graphene lms etched in (NH 4 ) 2 S 2 O 8 solution plus different acids. The sheet resistance of graphene lms generally decreases as the concentration of additional acids increases. For HNO 3 and H 2 SO 4 as dopants, the carrier mobility of graphene lms remains smooth while the corresponding carrier density increases as the concentration of acids increases, which leads to an overall decrease in sheet resistance. The lowest sheet resistance was realized in (NH 4 ) 2 S 2 O 8 etchant with 1.2 M HCl. With increasing HCl concentration, the sheet resistance sharply increased, and with HNO 3 and H 2 SO 4 , the carrier mobility of samples in HCl linearly decreased. With careful observation as shown in Fig. 3h, the  Fig. 3i, which shows the bottom view of samples during the transfer process, abundant bubbles can be generated as the concentration of HCl increases, which suggests that the violent reaction during the etching process might cause the loss of electrical connections and hence a decrease in carrier mobility. However, the corresponding carrier density sharply increases when HCl is used at concentrations lower than 1.2 M but with a lower slope above 1.2 M. As concluded in the above paragraph, it is necessary to maintain a good electrical connection and hence the carrier mobility. Herein, a suitable HCl concentration could enable the balance between the carrier mobility and carrier density to reach the lowest sheet resistance of approximately 300 ohm sq À1 . Sheet resistance mapping measurements were also performed to reveal the exact electrical performance of 5 cm Â 5 cm graphene lms by detecting eddy current signals, as shown in Fig. 3d-g. The sheet resistance mapping of graphene lm using only (NH 4 ) 2 S 2 O 8 etchant exhibits complex color units over the entire sample, which indicates the quite disordered distribution of sheet resistance in the range of 350-700 ohm sq À1 . Aer adding different acids, the sheet resistance was signicantly reduced, and the homogeneity of the sheet resistance was greatly improved with much more uniform color distribution. Among these three types of acids, the sample with the addition of HCl exhibited the lowest sheet resistance, of approximately 300 ohm sq À1 . These results are consistent with the four-probe measurements and Hall effects measurements. The corresponding carrier density of samples in the (NH 4 ) 2 S 2 O 8 etchant plus HCl is also signicantly improved in a narrow range of 9-12 Â 10 12 cm À2 compared to 4-10 Â 10 12 cm À2 for samples in bare (NH 4 ) 2 S 2 O 8 etchant. Thus, the sheet resistance of graphene lms using HCl dopant reaches the lowest value of approximately 300 ohm sq À1 and is also in a much narrower range of 270-380 ohm sq À1 compared to the broad range of 350-700 ohm sq À1 for the graphene lms treated with bare (NH 4 ) 2 S 2 O 8 etchant. This demonstrates that the sheet resistance of graphene lms can be repeatedly realized in a quite narrow range using (NH 4 ) 2 S 2 O 8 etchant plus 1.2 M HCl dopant. For the original graphene lms, the G peak is at approximately 1582 cm À1 and the D band is at approximately 2700 cm À1 . The doped graphene will exhibit an upshi of the G band and 2D band. [43][44][45][46] Fig. 4d presents the Raman spectra of as-obtained graphene lms using different etching solutions. It clearly shows that the G band of graphene lms etched in Fe(NO 3 ) 3 and FeCl 3 is slightly upshied compared to that of the (NH 4 ) 2 S 2 O 8 sample, which is consistent with the carrier density action in  To further study the doping effect of different etchants, XPS measurements were also carried out, as shown in Fig. 4e and f. The C 1s peaks of the graphene lms etched by different etchants are divided into three symmetric Gaussian curves, which are assigned to C-C, C-O, and C]O, respectively. For the pdoping of graphene lms, the C 1s peaks corresponding to the sp 2 and sp 3 hybridized states were shied to lower energy, 20,47 which is similar to the case for p-doped carbon nanotubes. 48 The peaks at 288.372 eV, 289.049 eV, 289.084 eV, and 289.097 eV correspond to the C]O peaks for graphene lms using (NH 4 ) 2 S 2 O 8 etchants plus HCl, FeCl 3 etchants, Fe(NO 3 ) 3 etchants, and (NH 4 ) 2 S 2 O 8 etchants, respectively. The C]O peak is obviously redshied by 0.73 eV, and the C-C peak is redshied by 0.16 eV aer the addition of HCl dopant in (NH 4 ) 2 S 2 O 8 etchants, which is consistent with the Raman results.

Conclusions
In summary, the relationship between the electrical properties of graphene lms on PET substrate and the transfer parameters has been systematically investigated in detail. It was found that the guarantee of electrical connections is closely related to the carrier mobility. Additionally, the selection of etchant and dopant is very sensitive to the carrier density and hence inuences the sheet resistance. Finally, graphene lms can repeatedly reach a low sheet resistance of approximately 300 ohm sq À1 with a narrow range if a (NH 4 ) 2 S 2 O 8 etchant with 1.2 M HCl dopant is used. This work provides a facile and reliable technique to obtain graphene lms with advanced and repeatable electrical properties, which is critically important for the fair evaluation of CVD-grown graphene lms and also the practical application in the eld of electrical devices.

Conflicts of interest
There are no conicts to declare.