Fermi-Level Modulation of Chemical Vapor Deposition-Grown Monolayer Graphene via Nanoparticles to Macromolecular Dopants.

It is critical to modulate the Fermi level of graphene for the development of high-performance electronic and optoelectronic devices. Here, we have demonstrated the modulation of the Fermi level of chemical vapor deposition (CVD)-grown monolayer graphene (MLG) via doping with nanoparticles to macromolecules such as titanium dioxide nanoparticles (TiO2 NPs), nitric acid (HNO3), octadecyltrimethoxysilane (OTS) self-assembled monolayer (SAM), and poly(3,4-ethylene-dioxythiophene):polystyrene sulfonate (PEDOT:PSS). The electronic properties of pristine and doped graphene samples were investigated by Raman spectroscopy and electrical transport measurements. The right shifting of G and 2D peaks and reduction in 2D to G peak intensity ratio (I 2D/I G) assured that the dopants induced a p-type doping effect. Upon doping, the shifting of the Dirac point towards positive voltage validates the increment of the hole concentration in graphene and thus downward shift of the Fermi level. More importantly, the combination of HNO3/TiO2 NP doping on graphene yields a substantially larger change in the Fermi level of MLG. Our study may be useful for the development of graphene-based high-performance flexible electronic devices.

Introduction

Graphene, a two-dimensional (2D) material, has received massive attention in the nanoscience field due to its extraordinary physical, electronic, and photonic properties.[1] It is a fundamental structural component of carbon allotropes such as fullerene, carbon nanotubes, and graphite.[2,3] In the honeycomb lattice structure, each carbon atom in graphene is sp2 hybridized and three carbon atoms are covalently bonded with three distinct neighboring carbon atoms. Therefore, one free electron lies in pz-orbital and perpendicular to plane plays an important role in the high electrical conductivity of graphene.[4,5] The excellent transparency and flexibility of graphene material is highly suitable for transparent conducting electrodes (TCEs) used in various electronics and optoelectronics devices such as display systems, sensors, spin devices, photovoltaic cells, and storage systems.[6−10] Recently, various synthesis approaches have been investigated for the preparation of monolayer graphene (MLG).[11] Among all, growth of high-quality and large-area graphene films on metal substrate by chemical vapor deposition (CVD) is a cost-effective and efficient method.[12,13]

However, the linear dispersed band structure and low carrier density of graphene limit its potential applications in future electronic devices. Therefore, identifying a suitable method to control the charge carrier concentration and thus Fermi level of graphene is an important step. In this context, researchers have adopted various doping strategies, including adsorption of gas molecules, ultraviolet irradiation, chemical doping, and electrostatic-field doping, for the tuning of charge carrier density in graphene.[14−18] However, thermal and ultraviolet irradiation doping techniques induced defects in the lattice of graphene and thus deformed its electronic structure, resulting in significant reduction in graphene conductivity. On the other hand, electric-field doping requires a high voltage for a long time to bring about a substantial change in the electronic properties of graphene. Recently, a novel plasma doping technique has been studied for the improvement of chemical species reactivity in a graphene device.[19,20] Graphene is highly sensitive to surface dopants, as its sp2 hybridized carbon atoms can easily react with the surrounding atmosphere.[21] Therefore, modifying the surface of graphene via doping with donor or acceptor chemical species is not only a simple and effective technique but also facilitates a significant change in the carrier concentration of graphene.[22,23] Graphene can be chemically doped by either substitution of carbon atoms with heteroatoms or sharing of charge between graphene and dopants. However, the former method is a destructive and stable doping approach, while charge transfer doping is a nondestructive method and preserves the intrinsic properties of graphene. Therefore, charge transfer doping, which has a great ability to modulate the carrier concentration, became the primary approach for doping of graphene.[24,25] The controlling of the carrier concentration and Fermi level of graphene depends on the type (n- or p-type) and concentration of the chemical dopants. In this context, Lu et al. theoretically investigated the tuning of the Fermi level of graphene via charge transfer doping with a range of dopants (e.g., AuCl3, FeCl3, SbF5, HNO3, MoO3, Cs2O, O2, and OH).[24] It was reported that the shifting of graphene Fermi level depends on the electron affinity and ionization potential of the chemical dopants. Except for the −OH dopant, charge transfer doping of graphene with other dopants did not change the sp2 hybridization and preserved the intrinsic properties of graphene. Another experimental study demonstrated the effect of copper chloride (CuCl) molecule doping on the electronic properties of CVD-grown graphene.[26] It was optically observed that the improvement in electrical conductivity and shift of Fermi level are directly correlated with the doping level. The Fermi level of graphene downshifted up to ∼0.64 eV, and simultaneously, the electrical conductivity is improved by more than two times at the highest concentration of CuCl. Seo et al. found that surface modification of graphene with heptadecafluoro-1,1,2,2-tetrahydrodecyl-trichlorosilane (HDF-S) self-assembled monolayers (SAMs) substantially increased the work function of graphene from 4.56 to 5.50 eV.[27] Meanwhile, the change in graphene work function depends on the amount of HDF-S molecules adsorbed on it. Raman spectroscopy results imply that the structural properties of graphene were preserved after HDF-S SAM treatment. Very recently, Yu et al. found that the Fermi level of graphene can be tuned by the adsorption of azobenzene molecules.[28] Herein, chemical modification of the graphene surface by azobenzene molecules with different dipole moments and dipole orientations induced both n- and p-type doping; at the same time, the lattice structure of graphene was maintained.

Here, we chemically modified the surface of CVD-grown MLG with various types of dopants, including nanoparticles (TiO2 NPs), small molecules (HNO3 and OTS SAMs), and macromolecules (PEDOT:PSS) to realize the tuning of its Fermi level. Raman spectroscopy, electrical transport measurement, and atomic force microscopy (AFM) were employed to characterize the change in electronic properties and surface morphology of MLG. The results demonstrate that all of the dopants induced p-type doping, and thus, the carrier concentration and Fermi level of graphene are substantially modulated. Our study presents a better understanding of the tuning of graphene Fermi level via doping with a range of dopants.

Results and Discussion

The schematic illustration of the doping of the MLG device with different dopants is shown in Figure .In order to study the effect of doping on the electronic structure of graphene, Raman spectroscopy measurement was carried out at different positions of each sample before and after doping. Figure a shows the Raman spectra of pristine and doped MLG devices, in which the standard D, G, and 2D peaks of the graphene samples are normalized and fitted with the Lorentzian function. Introduction of a small D peak in the Raman spectra of pristine graphene suggests the presence of defects or disorder, whose intensity is slightly changed after the doping treatments.[30,31] The G peak of pristine is positioned at ∼1579 cm–1, and involves the E2g phonon mode at the Γ point of the Brillouin zone; the 2D peak is located at 2671 cm–1, and involves transverse phonon emission near the K-point of the Brillouin zone.[31] In order to understand the charge transport mechanism between graphene and the doping sources, the G and 2D peak positions of the pristine sample compared with doped graphene samples are shown in Figure b. The G and 2D peak positions, intensities, and line widths are sensitive to the number of graphene layers, doping, and laser energy. The shifting of G and 2D peaks towards higher frequencies or lower frequencies is attributed to the p- or n-type doping of graphene.[30] In our case, the G peak position of MLG shifted from ∼1579 to ∼1583 cm–1 (HNO3), ∼1582 cm–1 (TiO2 NPs), ∼1585 cm–1 (HNO3/TiO2), ∼1582 cm–1 (PEDOT:PSS), and ∼1583 cm–1 (OTS), validating the change in doping level of graphene.[32,33] Since the carrier doping modifies the 2D phonon frequencies, the 2D peak of MLG shifted from 2671 to 2674 cm–1 (HNO3), 2673 cm–1 (TiO2 NPs), 2677 cm–1 (HNO3/TiO2), 2672 cm–1 (PEDOT:PSS), and 2673 cm–1 (OTS), confirming hole doping. The shifting of the 2D peak towards higher frequencies reflected that the Fermi level of graphene shifts downward from the Dirac point (EF = 0).[34] After doping, a small reduction in the full width at half-maximum (FWHM) of the G peak is observed, which may be due to the removal of Kohn anomaly at the Γ point, while the FWHM of the 2D peak increased due to the phonon confinement effect (Figure b). Similar to the Raman peak shifts, the intensity ratio of 2D and G peaks (I2D/IG) is an important parameter to estimate the change in doping level of graphene. The I2D/IG value of the pristine sample is more than two times, and the FWHM of the 2D peak is close to 33 cm–1 (Figure c), confirming that our pristine graphene is monolayer.[31] Due to p-type doping, the reduction of the 2D peak intensity and I2D/IG value is attributed to the increase of charge carrier scattering in graphene.[32] On the other hand, the ID/IG ratio defines the degree of disorder and is inversely proportional to the crystalline size of graphene lattice.[35] As shown in Figure c, the ID/IG value of MLG increased after doping treatment, which confirms that defects would generated at the interface between the dopants and graphene. After Raman spectra analysis, it is our primary speculation that each dopant induced p-type doping effects on CVD-grown MLG devices and the Fermi level downshifted from the Dirac point. However, it is quite difficult to distinguish the n- or p-type doping of graphene purely on the basis of Raman spectroscopy results, because in both doping cases the G peak position shifted in the same direction.[36] Therefore, for a better understanding of the doping type and change in the Fermi-level position of graphene, we discuss the electrical transport measurements in the following paragraphs.

Figure 1

Schematic illustration of the doping of an MLG device with different dopants.

Figure 2

(a) Normalized D, G, and 2D peaks of pristine and HNO3-, TiO2 NP-, HNO3/TiO2-, PEDOT:PSS, and OTS SAM-doped MLG. (b) Normalized G and 2D peaks of pristine and HNO3-, TiO2 NP-, HNO3/TiO2-, PEDOT:PSS-, and OTS SAM-doped MLG (c) I2D/IG and ID/IG ratios of MLG as a function of dopants.

The change in surface morphology of MLG after doping with HNO3, TiO2 NPs, PEDOT:PSS film, and OTS SAM molecules was investigated by AFM. Figure a–e displays the AFM images of pristine and doped graphene samples. Figure a shows the AFM image of pristine MLG on SiO2/Si substrate; the surface of graphene contains few defects and wrinkles, which may have arisen during the transfer process. It is clearly evident from Figure b–e that the surface morphology of doped graphene samples is slightly changed as compared to pristine graphene. The variation in surface morphology of doped graphene samples may be due to the different doping sources having their own specific nature and structure.

Figure 3

AFM micrograph of (a) pristine-, (b) HNO3-, (c) 1.0 mg TiO2 NP-, (d) 1.04 w/v PEDOT:PSS-, and (e) OTS-doped MLG (inset shows the height of the scale bar).

The effect of nanoparticle, small molecule, and macromolecule doping on the electronic properties of the MLG device was investigated through the electrical transport measurement. The schematic diagram of the electrical measurement setup for graphene devices is illustrated in Figure a. A fixed drain-source voltage VDS ∼ 1 V was applied between the source and drain electrodes, while a variable gate voltage was applied for the electrical characterization of the devices. Figure b shows the resistivity vs gate-source voltage (VG) curve of the pristine and doped MLG devices. The Dirac point of the device corresponds to the voltage at which the resistance is maximum. The pristine graphene Dirac point was positioned at ∼4 V, signifying unwanted p-doping due to the adsorption of unwanted water or oxygen molecules on the MLG surface.[37] Upon doping with HNO3, TiO2, HNO3/TiO2, PEDOT:PSS, and OTS SAMs on the MLG surface, the Dirac point shifted to the right, indicating that the electrons are transferred from graphene to the dopants, resulting in hole doping.[24] The Dirac point of the MLG device as a function of dopants is shown in Figure c. In order to understand the charge trapping and interface properties of the device, the hysteresis of the graphene devices is monitored using the resistivity vs gate voltage curve, as illustrated in Figure S1 of Supporting Information. It is observed that the Dirac point of the graphene devices did not shift upon sweeping of the voltage, which likely validates that a nominal amount of water molecules is trapped at the interface of graphene devices.

Figure 4

(a) Schematic diagram of the electrical measurement setup of the graphene devices. (b) Resistivity vs back-gate voltage (VG) characteristics curve of the MLG before and after doping. (c) Dirac point of the MLG device as a function of dopants.

Recently, a DFT study demonstrated that the adsorption of TiO2 on the graphene surface imposes either n- or p-type doping depending on whether the Ti or O atoms of TiO2 are close to the carbon atoms of graphene.[38] The ground-state structure of the TiO2 monolayer displays that the Ti layer is sandwiched between two oxygen layers, and thus, oxygen atoms exposed to the graphene surface cause the Fermi level to be downshifted relative to the Dirac point. Upon HNO3 doping over graphene, HNO3 molecules are divided into nitrogen dioxide (NO2), nitrate (NO3) radicals, and water (H2O) molecules, and these radicals are adsorbed on the graphene surface.[39] The NO2 and NO3 radicals are computed to have a single occupied DOS below the Fermi level of graphene, which allows the transfer of electrons from graphene to this state, thereby inducing p-type doping. The deposition of the PEDOT:PSS film significantly modulates the electronic properties of graphene, which can be explained by two different mechanisms.:[40,41] (a) In acidic doping, the sulfonic acid (S(=O)2–OH) group of the PSS chain attracts the electrons from graphene, (b) while in surface charge doping, the electrons are transferred from graphene to the PEDOT:PSS film due to the higher work function (more than 5.0 eV) of PEDOT:PSS than the graphene (∼4.6 eV), which leads to the hole doping of graphene. As reported, the functionalization of the graphene surface with oxygenated species (defects) offers unwanted doping, and these defects act as nucleation sites for the SAM molecules.[42] After OTS doping, the electron-withdrawing Si(OCH3)3 group of the OTS molecules reacts with the oxygen species in atmospheric condition; as a result, the poly-siloxane (Si(OH)3) group is formed, and the density of the hole carriers increases in graphene. One −OH group chemically reacts with the other −OH group and leads to the formation of an Si–O–Si bond. This self-polarization process changes the magnitude of the dipole at the graphene/OTS interface and tunes the graphene Fermi level.

Considering the Dirac point, the carrier concentration of the MLG before and after chemical treatment can be calculated by the expression n = Cg (VCNP)/e,[1] where Cg is the gate capacitance per unit area with an estimated value of ∼115 aF cm–2, e is the electronic charge, and VCNP is the charge neutrality point (i.e., Dirac point) of the sample. Figure a demonstrates the carrier concentration of MLG as a function of dopants. It is noted that the hole concentration of MLG is significantly increased after p-doping and is tunable up to 1.01 × 1013 cm–2 by combined doping with HNO3 and TiO2 NPs, which is consistent with other p-type-doped graphene systems.[24,43]

Figure 5

Variation in (a) carrier density, (b) Fermi level, and (c) electron and hole mobility of MLG as a function of dopants.

After doping, improvement in the carrier concentration of graphene is directly related to a shift of Fermi level, which is evaluated by the expression ,[5] where the Planck’s constant (ℏ) is 6.62 × 10–34 m2 kgs–1, n is the charge carrier density, and the Fermi velocity |vF| is 1.1 × 106 m/s and is assumed to be constant. The Fermi level of graphene can move upwards or downwards from the relative Dirac point (EF = 0) depending upon the n- or p-type doping of graphene.[44]Figure b highlights the Fermi level of CVD-grown MLG as a function of the doping sources. We have calculated the Fermi level of pristine graphene to be ∼27 meV, which is slightly deviated from the Fermi level of intrinsic graphene and may be due to the adsorption of dipolar water molecules or unwanted doping during the transfer process of the graphene device.[45−47] After doping, the Fermi level of graphene shifted more downward, and the maximum variation of ∼405 meV is observed for the HNO3/TiO2 NP-doped MLG device; a similar shift in Fermi level was reported for other hole-doped graphene systems.[48] We have also compared our results with the others, as given in Table S1 of the Supporting Information.

The carrier mobility of CVD-grown MLG before and after doping was calculated by the equation ,[1] where σ (1/resistivity) is the conductivity of the graphene device and VG is the applied back-gate voltage. The hole and electron mobilities of pristine and doped graphene samples were measured by the linear fitted slope of their respective conductivity–voltage curve. As illustrated in Figure c, the hole and electron mobilities of CVD-grown MLG are reduced after doping with HNO3, TiO2 NPs, HNO3/TiO2, PEDOT:PSS, and OTS SAM molecules.[15,24] Upon doping, reduction in the carrier mobility of MLG could be attributed to the creation of charge impurity scattering centers, as well the short-range disorder in the graphene lattice. It is clearly observed that the HNO3/TiO2-doped MLG device shows a minimum hole mobility, likely due to the high scattering effect. On the other hand, due to strong p-type doping, the electron conductivity of graphene largely disappears, and therefore measurement of electron mobility is not feasible.[49] In particular, the carrier mobility of the PEDOT:PSS-doped MLG device is almost similar to that of pristine graphene, because it may be possible that the dipole field created at the interface of graphene and the PEDOT:PSS film screened the charge impurity scattering effects.

Conclusions

We have demonstrated the chemical doping of CVD-grown MLG with various dopants (nanoparticles to macromolecules), including HNO3, TiO2 NPs, OTS SAMs, and PEDOT:PSS. These dopants significantly modulate the Fermi level of CVD-grown MLG. Raman spectroscopy and electrical transport measurements were used to characterize the doped graphene. Raman spectra show the up-shifting of G and 2D peak positions as well as reduction of I2D/IG value after doping with various dopants, revealing the p-type doping of MLG. The electrical transport measurements demonstrate the shifting of Dirac points towards positive voltage after doping, which validated the change in Fermi level of graphene. The shift of Fermi level is analyzed as a function of the dopant. Among the dopants, the largest variation in Fermi level is observed for HNO3/TiO2 NP-doped MLG, which suggests that the combined doping approach is highly effective for modulation of the graphene Fermi level. Our work shows that the doping of graphene with nanoparticles to macromolecules is a promising way to tune the Fermi level of MLG for many applications, including conductive, transparent, and flexible electronic devices.

Experimental Section

Fabrication of MLG Devices

In our study, the CVD method was used for the synthesis of MLG on a 25 μm thick polycrystalline copper (Cu) substrate. Further, the grown MLG film was transferred onto the SiO2 layer (300 nm thick) capped over the highly p-doped Si substrate (p++-Si). Details about the growth of the MLG film and its transfer process have been discussed in our previous paper.[29] In order to fabricate the graphene field effect transistors, the source and drain electrodes of the Au film (30 nm thick) were thermally evaporated onto the graphene using the shadow mask method.

MLG Doping and Characterization

In order to investigate the influence of different types of dopants on the Fermi level of CVD-grown MLG, the fabricated MLG device was cut into five samples, in which one was used as pristine and remaining four samples were used for chemical treatment. The chemical dopants such as HNO3 (Molychem, Product code-16560), TiO2 NPs (Alfa Aesar, Product code-39953), OTS SAM (Sigma-Aldrich, Product code-376213), and PEDOT:PSS (Sigma-Aldrich, Product code-483095) were used as received without further purification in our experimental study. For the molecular doping approach, an 8M solution of HNO3 and a 20 μL solution of OTS SAM were spin-coated on the channel regions of two different MLG devices at 3000 and 4000 rpm for 60 s, respectively. In the metal oxide NP doping process, 10 mg/mL TiO2 was suspended in distilled water, and then, a suspension of 100 μL of TiO2 NPs was dropped over the channel region in order to dope the MLG device with 1.0 mg TiO2 NP concentration. The combination of the HNO3 and TiO2 NP doping approach was also used for the improvement of the electrical transport properties of MLG. The macromolecule doping of MLG was achieved via coating a thin, transparent, and conducting film of 1.04 w/v PEDOT:PSS polymer. Finally, all of the doped MLG devices were annealed at 60 °C for 1 h to complete the removal of moisture from their surfaces. Raman spectroscopy measurement was performed on undoped and doped MLG devices by micro-Raman spectroscopy (WiTec α 300R, Germany) with laser excitation wavelength 532 nm. The laser excitation power was kept at below 1 mW to avoid sample heating and introduction of any defects in the graphene lattice. The surface morphology of the MLG devices before and after doping was recorded by AFM (Asylum MFD-3D, U.K.) in noncontact mode. The electrical characterization of pristine and doped graphene samples was performed using a Keithley source measure unit model (2612A, USA) under ambient conditions.