Incorporation of Boron Atoms on Graphene Grown by Chemical Vapor Deposition Using Triisopropyl Borate as a Single Precursor

We synthesized single-layer graphene from a liquid precursor (triisopropyl borate) using a chemical vapor deposition. Optical microscopy, scanning electron microscopy, Raman spectroscopy, and X-ray photoelectron spectroscopy measurements were used for the characterization of the samples. We investigated the effects of the processing temperature and time, as well as the vapor pressure of the precursor. The B1s core-level XPS spectra revealed the presence of boron atoms incorporated into substitutional sites. This result, corroborated by the observed upshift of both G and 2D bands in the Raman spectra, suggests the p-doping of single-layer graphene for the samples prepared at 1000C and pressures in the range of 75 to 25mTorr of the precursor vapor. Our results show that, in optimum conditions for single-layer graphene growth, that is, 1000C and 75mTorr for 5 minutes, we obtained samples presenting the coexistence of pristine graphene with regions of boron-doped graphene.


Introduction
In recent years, graphene has attracted widespread attention in scientific and technological communities due to its atomic thickness and excellent electronic, transport, and mechanical properties, which point to its many potential applications [1].With regard to applications, graphene appears to be a very promising candidate for use in electronic nanodevices.However, the fact that pristine graphene has no band gap is the most important limitation for the development of graphene-based nanoelectronics.Substitutional doping provides an effective approach to tailor the graphene electronic band structure because it opens an energy gap between the valence and conduction band.Nitrogen and boron are the most suitable candidates to be used as dopant elements for graphene [2,3] because of their size, which is similar to that of carbon, and the fact that there are five and three valence electrons available to form bonds with the carbon atoms, enabling n-type doping and p-type doping, respectively.The incorporation of these dopant elements is usually accompanied by the generation of defects [4].If, on one hand, the presence of defects can degrade transport properties in graphene, reducing the mobility of electrons or holes, on the other hand, the defects can increase the sensitivity of sensor devices based on pristine graphene [5][6][7].In fact, the sensor sensitivity may be limited by the chemical inertness of graphene.Indeed, pristine graphene has no dangling bonds on its surface, inhibiting both the adsorption of the gaseous molecules of interest and the functionalization of species.Boron-doped graphene has also been studied for applications in other technological fields [8].In fact, boron-doped graphene has been proposed as an active electrocatalyst for oxygen reduction reaction in fuel cells [9,10], as a back electrode in CdTe solar cell [11], and as support for platinum catalyst [12].For these last two applications, the borondoped graphene samples were obtained from different types of thermal processing of graphene oxide.
Several approaches have been developed for the growth of graphene and the low-pressure chemical vapor deposition (LPCVD) technique which allows the acquisition of largearea high-quality graphene film [13].The incorporation of boron atoms in graphene was obtained by different methods: exfoliation of boron-doped graphite [14], arc discharge of graphite electrodes in the presence of diborane [15], thermal decomposition of boron carbide powder [16], thermal annealing of graphene oxide in presence of boron oxide [9], hot filament CVD [17], and chemical vapor deposition (CVD) using solid or gas precursors [7,[18][19][20][21][22][23].The CVD technique could produce a single layer of boron-doped graphene with an extremely toxic and inflammable gas, that is, diborane, as the precursor [21,22].On the other hand, pristine graphene, prepared by CVD using methane, can be doped by its exposure to diborane at a high temperature in a two-stage process [20].Recently, Ruitao and collaborators proposed an alternative approach to the use of diborane [7].As precursor for the CVD growth, they used a mixture of triethylborane and hexane.The results obtained diverge considerably from other results.In fact, while the STM results obtained by Wang and collaborators [19] indicated that boron atoms are primarily incorporated in graphene lattice in the graphitic form in a random distribution, the results obtained by Ruitao and collaborators [7] also using STM indicated the formation of boron-carbon trimmers embedded within the hexagonal lattice.In both cases, a p-type conducting behavior was verified.In the case of pristine graphene samples prepared by CVD and subsequently submitted to thermal annealing in a diborane atmosphere, boron-doped single-layer graphene was obtained with the presence of boron carbide nanoparticles at the grains boundaries [17].Results for boron doping of single-layer graphene are still scarce and suggest that they are dependent on the parameters chosen for the CVD process.
In this work, boron-doped graphene was synthesized by low-pressure CVD (LPCVD) on copper foil substrates using a single liquid precursor, triisopropyl borate.We investigated the effects of the processing temperature and the vapor pressure in order to verify the homogeneity of the samples and optimize the growth parameters.The samples were characterized by Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), optical microscopy, and field emission scanning electron microscopy (FEG-SEM).Our results show that, for the optimum conditions for single-layer graphene growth, that is, 1000 ∘ C and 75 mTorr for 5 minutes, we could obtain samples presenting the coexistence of pristine graphene with regions of boron-doped graphene, which is a different figure compared to previous boron-doped graphene prepared by CVD.

Materials and Methods
Boron-doped graphene was synthesized by the LPCVD method using copper foils as the substrate (25 m thick from Alfa Aesar, 99.99% purity) and triisopropyl borate (from Sigma Aldrich, 98% purity) vapor as precursor.The processing temperature ranged from 900 ∘ C to 1000 ∘ C. The pressure was varied from 25 mTorr to 400 mTorr with a fixed exposure time of 5 minutes.The doped graphene was transferred to SiO 2 (300 nm thick)/Si and quartz substrates for subsequent characterizations using a wet transfer method with poly(methyl methacrylate) (PMMA) as the support film.Details of the sample preparation and transfer can be found elsewhere [24].The liquid precursor volatizes when exposed to low pressures.The vapor was carried throughout the furnace using 2 sccm of H 2 gas.The boron-doped graphene was then synthesized by the self-assembly of the atoms dissociated from the vapor on the catalyst surface.
The samples were characterized using a multitechnique approach including micro-Raman spectroscopy (RS), optical microscopy, scanning electron microscopy (SEM), and X-ray photoelectron spectroscopy (XPS).A micro-Raman spectrometer (NT-MDT, NTEGRA spectra) equipped with a CCD detector, cooled by a Peltier cell, and a solid-state laser, which provided an excitation wavelength of 473 nm (2.62 eV), was employed.Raman mapping was carried out by a piezoelectrically controlled XY table, which allowed for steps of 0.5 m, coupled to a single grating (600 lines/mm) Raman spectrometer (HORIBA, LabRAM HR).This setup was equipped with a He-Ne laser with an excitation wavelength of 633 nm (1.96 eV).In both systems, the laser spots on the sample surface were of the order of 0.5 m, while to minimize the heating effects due to the laser irradiation we kept the incident beam power below 1 mW.Repeated Raman measurements were performed on different regions of the sample surface and the spectra from morphologically similar regions showed overall good reproducibility.Finally, the wavenumbers of the vibrational modes of O 2 (1555 cm −1 ) and N 2 (2331 cm −1 ) molecules in the air were exploited to calibrate the band position of the graphene spectra.
A field emission scanning electron microscope (FEG-SEM, JSM-6701 F, from JEOL) operating at 1 kV was employed for morphology analysis.X-ray photoelectron spectrometry (XPS) was carried-out using Mg K  X-ray source (ℎ] = 1253.6eV) and an Alpha 110 commercial hemispherical electron energy analyzer.The measurements were performed at a grazing angle of 80 degrees.The highresolution XPS spectra were collected at a pass energy of 20 eV, and the data was processed using the CasaXPS software.It took the energy of the peak of C 1s at 284.5 eV as a reference.
A Zeiss Axio optical microscope was used for sample imaging, and the optical transmittance was measured in the range of 40 to 800 nm by using a UV-vis spectrophotometer (Lambda 950, Perkin Elmer, USA).

Results and Discussion
Typical Raman spectra recorded from samples grown using 25 mTorr of vapor pressure at three different temperatures in the 900 ∘ C-1000 ∘ C range and transferred to an oxidized silicon wafer via a PMMA route are shown in Figure 1.The spectra indicate that single-layer graphene was obtained at the highest temperature, 1000 ∘ C. The 2D band profiles of the samples deposited at the lower temperatures suggest that multilayer graphene was produced, while the presence of  defects was revealed by the occurrence of the D peak.It became progressively more important for samples prepared at lower processing temperatures.
Figure 2 shows the spectra obtained from samples transferred to SiO 2 /Si and prepared at different pressures.They were grown keeping the temperature and exposition time constant, 1000 ∘ C and 5 minutes, respectively.The Raman spectrum obtained from the samples prepared at 50 mTorr is quite similar to the spectrum of the pristine graphene layer obtained using methane as the precursor.In both spectra, the intensity ratio between the bands 2D and G ( 2D / G ) is around 4, while the bandwidth of the 2D band is 35 cm −1 .These values are typical of good quality pristine samples.A very weak D-band was observed in these spectra.It can be reasonably attributed to defects introduced during the transfer procedure.The spectrum obtained from the samples grown at 75 mTorr shows a shoulder that corresponds to the D  peak associated with the presence of defects.The other spectra of Figure 2 show an increase of the intensity of the D peak with the rise in the vapor pressure.The increase in the D-band suggests more defective samples and it is followed by a concomitant decrease in the  2D / G intensity ratio.The spectra obtained from samples prepared at pressures higher than 75 mTorr are typical of nanocrystalline graphite, as revealed by the occurrence of a broad multiphonon band centered around 3000 cm −1 [25].
In Figure 3(a), we present the optical microscopy image obtained from a sample prepared at 75 mTorr and 1000 ∘ C for 5 minutes and transferred to a SiO 2 /Si substrate, where the uniformity of the sample can be evaluated in the image.In Figure 3(b), we show the UV-vis transmittance of a sample prepared with the same parameters and transferred to a quartz substrate.Transmittance of the boron-doped sample is quite similar to that of pristine sample, also shown in the figure, and the transmitted value at 650 nm is 96.5 ± 0.2%.Based on previous reports, the experimental value for the transmittance of exfoliated graphene is 2.3 ± 0.1% [26].The difference can be attributed to defects in both pristine and doped graphene generated during sample transfers to a quartz substrate.The presence of PMMA residues and the bilayer graphene covering small areas of the sample surface can also contribute to explaining this result.
Figure 4 shows the images of boron-doped graphene on copper grown at different triisopropyl borate vapor pressure.The samples were prepared at 1000 ∘ C for 5 minutes.Figure 4(a) corresponds to pristine graphene on copper used as the reference image for comparison purpose.For growth using 200 mTorr vapor pressure (image shown in Figure 3(d)), we obtained nanocrystalline graphite, confirmed by Raman spectra as shown Figure 2.For the 50 and 75 mTorr pressures, we see the formation of dark and bright domains.It is clear that the doped samples are not homogeneous at the 3 m scale, while the pristine sample is homogenous at a much larger scale.
Raman mapping was exploited to probe the structural homogeneity of the samples.Figure 5(a) shows an optical image taken from a generic 60 m × 50 m region of the sample grown at 75 mTorr for 5 minutes at 1000 ∘ C. It shows some dark colored spots corresponding to PMMA residues, while the dashed rectangle (5 m × 20 m) is the region of the sample free of PMMA which was mapped.
The Raman map of the relative  2D / G intensity ratio is highlighted in false colors and can be seen at the left side of the Raman spectra as an inset in Figure 5(b).The Raman spectra that are plotted in Figure 5(b) are those that have the extreme values for the  2D / G (i.e., maximum) and  2D / D (i.e., minimum) intensity ratios at points 1 and 2, respectively.The intersection of the red and blue lines indicates the exact point where each spectrum was recorded.In Figure 5(c), we plot only the regions of the spectra which correspond to the G-and D  -bands.The upshift in the Gband of the spectra obtained from point 2 is clear when compared to point 1.This is evidence of the p-doping of graphene, which is in agreement with the hole doping by applying a gate voltage [27].The Raman map confirms the SEM images, since it shows that the sample consists of areas whose homogeneity is confined within a few microns square.In contrast, when one considers larger areas, the sample turns out to be not homogeneous.It is clear that the darker green zones correspond to the minimum value of  2D / G , while the brighter green zones correspond to the maximum value of this ratio.In fact, the  2D / G intensity ratio varied from 1.5 to 6.0 for the spectra obtained in the measured region, which is indicated in Figure 5(a).The blue and red spectra quoted in Figure 5(b) also correspond to the extreme values of the ratio between the D-band and G-band ( D / G ) but with the opposite behavior.In fact,  D / G is the minimum for the spectrum obtained at point 1 (blue) and the maximum for the spectrum obtained at point 2 (red).
Another important parameter is the bandwidth of the 2D band.In both spectra, the bandwidths are around 30 cm −1 which is characteristic of single-layer graphene.Besides the difference in the  2D / G ratio, other important features are the presence of the D  -band, as a peaked shoulder of the Gband, in the red spectrum plotted in Figure 5(b), as well as the ratio between the D-and G-bands. D / G is 0.6 for the blue spectrum and 1.8 for the red spectrum plotted in Figure 5(b).It was shown by theoretical calculations that both carbon and boron atoms can diffuse very rapidly on the copper surface and that boron atoms are incorporated at the growth edge forming pentagons [28].Then, the concentration of topological defects (as pentagons) in boron-doped graphene must be very high [28], explaining the higher value of the  D / G ratio and the presence of the D  -band in the Raman spectra.
The combination of these results indicates that we have a type of composite material, where the p-doped graphene layer (point 2) coexists with pure single-layer graphene regions (point 1) in Figure 5.In fact, the red spectrum presented in Figure 5(b) is similar to those previously reported in the literature for boron-doped graphene, grown from different precursors [7,17].The positions of the G-bands are slightly shifted toward a higher frequency, corroborating this picture.
The samples were also characterized by XPS.The carbon peak dominates the XPS survey spectrum obtained from graphene samples, with a small peak due to the presence of boron around 190 eV.Other peaks were also observed in the survey spectrum: the oxygen peak at 531 eV and the carbon and oxygen Auger peaks at 990 eV and 750 eV, respectively.Typical XPS B1s core-level spectra are presented in Figure 6 for samples grown at 200, 75, and 50 mTorr vapor pressure for 5 minutes at 1000 ∘ C. The spectrum was decomposed in several contributions keeping the FWHM between 1.5 and 1.6 eV and assuming a Shirley background.One difficulty that arises is the question of determining the correct value for the binding energy in the substitutional doping of graphene with boron detected via photoemission spectroscopy induced by X-rays.Few theoretical studies have been reported in the literature.Based on DFT calculations, a value between 188.5 and 190 eV for the binding energy is attributed in the XPS B 1s core-level spectra revealing the boron incorporation into substitutional sites [29].High-energy resolution XPS measurements on boron-doped graphene have associated the peak with a binding energy of 189.6 eV to substitutional doping of graphene by boron atoms [20].Wu et al. [18] deconvoluted the B 1s spectrum of boron-doped graphene obtained by CVD from a solid precursor and assigned the peak with a binding energy of 189.7 eV to BC 3 , the chemical environment of boron atoms in the case of substitutional boron doping.Thus, the peak having a binding energy of 189.4 eV can be attributed to sp 2 carbon-boron bonds in substitutional boron sites in graphene, in good agreement with previous XPS measurements [20].The peak at 187.5 eV is associated with B 4 C [30], while the two peaks corresponding to higher binding energies are assigned to BCO 2 at 191.5 eV  [30], respectively, which can be formed when boron atoms adopt a low-coordination substitutional site close to the graphene edges [19,20].The peak at 190.7 eV is assigned to BC 2 O compound in the sample grown at 200 mTorr [29].
In Table 1, the amounts of boron present in the samples grown using Cu substrate exposed at 25, 75, and 200 mTorr pressure of triisopropyl borate vapor at 1000 ∘ C for 5 minutes are reported.The total amount of boron is the range between 1 at.% and 2.5 at.% and the substitutional boron is around 0.4 at.%.In this case, we suggest that the substitutional boron is responsible for the doping.

Conclusions
In this work, single-layer graphene was synthesized using a low-pressure CVD system on copper foil substrates using triisopropyl borate vapor.Morphological results showed the formation of a kind of composite material, while Raman mapping confirms that the samples are not homogenous.The Raman results suggest that doped areas coexist with pure graphene areas.We also performed XPS measurements.The XPS spectra show the presence of not only substitutional boron but also boron bonded to oxygen and carbon (BC  O  ).The observed shift to higher frequencies of the G and 2D bands suggests the p-doping of single-layer graphene.In contrast with what was observed by Cattelan et al. [20], we attributed the composite character of the samples to a blend between doped graphene and undoped graphene, while they proposed the presence of boron carbide nanoparticles at the border grains of doped graphene.The other previous works did not analyze the homogeneity of the samples upon large areas.The lack of homogeneity appeared, from our results, to be the most important obstacle in the search for singlelayer graphene doped with boron produced by different techniques, including CVD.

Figure 1 :
Figure 1: Raman spectra obtained from samples prepared under pressure of 25 mTorr at three different temperatures and transferred to SiO 2 /Si via PMMA route.

Figure 2 :
Figure 2: Raman spectra obtained from samples prepared at different pressures for 5 minutes at 1000 ∘ C and transferred to SiO 2 /Si via PMMA route.

Figure 3 :Figure 4 :Figure 5 :
Figure 3: (a) Optical microscopy image of doped graphene prepared at 75 mTorr, 1000 ∘ C, and 5 minutes and transferred to a SiO 2 /Si substrate.(b) UV-vis transmittance of quartz substrate, pristine graphene, and doped graphene of a sample prepared with the same parameters and transferred to a quartz substrate.

Table 1 :
XPS results of the samples prepared by CVD using Cu substrate exposed during 5 minutes to different vapor pressures of triisopropyl borate at 1000 ∘ C.