Functionalised hexagonal-domain graphene for position-sensitive photodetectors

Graphene’s unique photoresponse has been largely used in a multitude of optoelectronics applications ranging from broadband photodetectors to wave-guide modulators. In this work we extend the range of applications to position-sensitive photodetectors (PSDs) using FeCl3-intercalated hexagonal domains of graphene grown by atmospheric pressure chemical vapour deposition (APCVD). The FeCl3-based chemical functionalisation of APCVD graphene crystals is affected by the presence of wrinkles and results in a non-uniform doping of the graphene layers. This doping profile creates multiple p–p+ photoactive junctions which show a linear and bipolar photoresponse with respect to the position of a focused light spot, which is ideal for the realization of a PSD. Our study paves the way towards the fabrication of flexible and transparent PSDs that could be embedded in smart textile and wearable electronics.


Introduction
The isolation of a single layer of graphene [1] has triggered an immense response in the scientific community in the past decade [2,3]. Both the exceptional electrical and optical properties of this material have been studied in detail and exploited in many areas of pure and applied research. In particular, its broadband absorption and field effect tunability make graphene an excellent platform for optoelectronic devices [4]. The breadth of such devices includes highresponsivity [5] and high-speed [6] broadband photodetectors (PDs), wave-guide coupling [7] and transparent and flexible electrodes [8]. The photoresponse of graphene PDs has been investigated with many techniques, though so far no reports have been made of its application in position sensitive detectors (PSDs). Such devices traditionally consist of a semiconductor (Si or Ge) junction with four contacts, which exploits the lateral photoeffect [9], that is the photovoltage generated at the junction plane in the presence of localized illumination, in addition to the conventional photovoltage effect that is formed across the junction. This effect has been extensively studied in the past [10,11] and it is at the base of PSDs used currently in many applications, such as: laser alignment, motion control, automation and scanning probe microscopy. The detection of focused x-rays via field-effect in a graphene transistor has been shown [12] though no other uses of graphene for PSDs in the UV-visible-NIR spectral range have been reported.
In this work we present the first chemical functionalisation of multilayer hexagonal domains of graphene grown by atmospheric pressure chemical vapour deposition (APCVD) [13] and demonstrate its use as an all-graphene PSD. Pristine graphene presents fundamental limitations in optoelectronic applications, such as an low intrinsic conductivity and the lack of bandgap. Chemical functionalisation [14] has been employed in the past years to overcome such limitations providing researchers with a stable and robust platform, thanks to the opening of an energy bandgap [15] or an increase in the conductivity of graphene. Amongst the different forms of functionalisation, intercalation with FeCl 3 [16], is used to induce strong p-type doping in the graphene layers, making it particularly suitable for use as transparent electrodes [17,18]. It has been recently reported that FeCl 3intercalated graphene, employed in a light emitting device, gives an enhanced light emission of 60% compared to standard graphene electrodes and up to 40% enhancement compared to commercial conductive polymers [8] while having an unprecedented stability in ambient conditions [19]. Here we intercalate multilayer (two to three layers) hexagonal crystals of graphene with FeCl 3 and characterize them using Raman spectroscopy and atomic force microscopy (AFM). After intercalation, we fabricate a multi-terminal device and characterize it using scanning photocurrent mapping (SPCM) [20]. The formation of wrinkles in the as-grown graphene is associated with the observed photoresponse of the device, in agreement with the charge density distribution measured across the device, using Raman spectroscopy mapping. Several junctions between regions of high (p + ) and low (p) doping are observed and the position of such p-p + junctions correlates well with the observed photocurrent (PC). Furthermore, the photoresponse scales linearly with the position of the excitation source and changes sign at the centre of the p-p + junctions. The observed linearity and bipolarity make our device suitable for position-sensitive detection of focused light. Our findings, combined with the high stability of FeCl 3intercalated graphene [19] and the ability to grow millimetresized single crystals [13], fill a technological gap in the inventory of all-graphene devices and open a route towards the fabrication of flexible PSDs for a multitude of novel applications.

Graphene growth and intercalation
Multilayer hexagonal domains of graphene were grown on copper by APCVD [21] using a melting/re-solidification pretreatment step to reduce nucleation density and increase domain size. Growth was carried out at~ 1075 C using diluted CH 4 (1000 ppm) as the carbon precursor and a high H 2 /CH 4 (50/25 sccm) to aid multilayer formation. The crystals where then transferred to highly doped Si substrate capped with 285 nm thermal SiO 2 using PMMA-supported electrochemical delamination [22,23]. The delamination step was performed using a 0.5 M NaCl solution and glassy carbon anode at 0.5 A, followed by rinsing in de-ionized water.
The intercalation with FeCl 3 was performed using a previously reported vapour phase method in vacuum [16]. More specifically, graphene on SiO 2 /Si and anhydrous FeCl 3 (Sigma-Aldrich, powder, 99.99% trace metals basis) are placed in a glass tube of a multi-zone resistive heated furnace. The glass tube is sealed after evacuating it to a pressure < -10 5 . Hence the sample is heated at a temperature of  360 C and the FeCl 3 at the sublimation temperature of  315 C for 12 h. The furnace is then left to cool and the sample removed.
After the intercalation, samples were washed in acetone and isopropyl alcohol in order to remove FeCl 3 residues from the substrate. The quality of the growth, the transfer and the degree of intercalation with FeCl 3 was assessed via optical inspection, AFM and micro-Raman spectroscopy. All measurements were performed in atmosphere and at room temperature. Raman spectra were acquired using a Renishaw spectrometer, with 532 nm excitation laser with incident power density of -0.3 1 MW cm −2 through a ×50 objective lens. The scattered light was dispersed by a 2400 g mm −1 grating and recorded by a CCD with 5 s integration time. Raman maps were acquired with the same setup performing a raster scan of the area in m 0.5 m steps. AFM topography and phase image were acquired with a Bruker Innova AFM system, operating in the 'tapping' mode using a sharp (radius of curvature <10 nm) highly doped silicon tip from Nanosensors with a nominal resonance frequency of 330 kHz.

Devices fabrication and characterization
Multi-terminal PDs were fabricated on the same substrates. Metal contacts were defined via electron-beam lithography using 300 nm thick PMMA as resist, followed by electronbeam evaporation of Ti/Au (5/50 nm) and lift-off in acetone. The sample was then contacted to the chip carrier by mean of wedge bonding using m 20 m thick gold wire. SPCM was performed using a custom-built setup [20]: a laser beam (l = 375 nm in ) was focused on the sample by a ×50 objective lens to a ∼260 nm spot diameter. The sample was mounted on a motorized microscope stage and raster-scanned under the laser beam. The photoresponse of the device was measured in short-circuit configuration for each point, producing a two-dimensional map of the photogenerated current (see also figure 3(a)). The light of the laser was modulated at a frequency of 33.25 Hz and the current measured via an Ametek 7270 DSP Lock-in amplifier locked at the same frequency.

Results and discussion
3.1. Functionalization of graphene single crystals Figure 1(a) shows the optical micrographs and Raman spectra of a pristine, as-transferred, multi-layer hexagonal domain of graphene grown by APCVD. Each layer grows in a stacked sequence, where the multi-layers appear at the centre of the first single-layer indicating that they share the same nucleation site [24]. The Raman spectrum of the first layer (1) shows the two main features of graphene: the G peak at~-1585 cm 1 , originating from the resonant E 2g mode and the 2D peak at -2700 cm 1 , originating from the double-resonant A 1g mode [25]. The shape of the 2D band, a single Lorentzian peak, and intensity ratioĨ I 1.9 2D G confirms the single-layer nature of the graphene. The spectrum of the second layer (2) shows the same features but with a ratioĨ I 3.36 2D G , which deviates from the expected ratio for two AB-stacked graphene layers [26]. This is due to the rotation of the crystallographic axes with respect to the underlying layer, producing an effective decoupling of the two stacked layers [21,27]. The third layer (3) is slightly twisted with respect to the second, as shown by optical inspection, and the spectrum shows a 2D band that can be fitted with the convolution of four Lorentzian peaks, which is characteristic of twisted bilayer graphene with a twist angle <  3 [21,28]. These observations are common to all the crystals we examined.
Functionalization with FeCl 3 strongly affects the electronic characteristics of graphene [14,17]. FeCl 3 intercalates between the layers of graphene forming a stacked sequence of graphene/intercalant/graphene [29]. Charge transfer to the FeCl 3 layer causes p-doping of the graphene [16], which results in the non-adiabatic removal of the Kohn anomaly at the Γ point and the consequent stiffening of the E 2g mode of graphene [30]. This stiffening is observed in the upshift of the G peak [31][32][33] in the Raman spectrum of graphene. Figure 1(b) presents the optical micrograph and the Raman spectra of the intercalated flake. Optically we can see each hexagonal crystal, with a change in contrast due to the presence of FeCl 3 . The Raman spectra are acquired in the same locations as in figure 1(a). The first layer (1) shows an upshift ), both indicating small doping of the graphene [33]. The second (2) and third (3) layers show the same characteristic spectra. We observe a split of the G band into three peaks, each corresponding to different doping levels: the G 0 -peak is the signature of pristine graphene; the G 1 -peak is given by a graphene layer in contact with one FeCl 3 layer (stage-2) and the G 2 -peak is given by one layer of graphene sandwiched between two FeCl 3 layers (stage-1). The 2D-band is also indicative of the effective decoupling of the graphene layers by the intercalant, as can be seen in the third spectrum, the multi-peak structure is reduced to a single Lorentzian peak, signature of the loss of stacking order within the graphene layers. The peaks related to the FeCl 3 molecules lie at significant lower energies (100-400 cm −1 ) [34], therefore they do not interfere with the modes of graphene. These results confirm the successful intercalation of single crystal APCVD graphene with FeCl 3 .
To evaluate the quality of our intercalated APCVD graphene we used AFM. Figures 2(a) and (b) show the topography image of a pristine single crystal, with a second layer grown at the centre, as seen optically (inset). Here we observe a series of parallel wrinkles (green arrows), which match the topography of the re-solidified copper used as metal catalyst during the growth. More interestingly, we observe larger wrinkles running in the perpendicular direction with respect to the first ones, originating from the centre of the crystal (blue arrows) and ending close to the middle of the hexagon's side. Wrinkles in APCVD graphene have been previously reported [23,35] to form during the cooling stage due to the different thermal expansion coefficients of graphene and copper. In the same images, residues of PMMA from the transfer process appear as white dots, as previously reported [36].
Intercalation with FeCl 3 changes the topography of the single crystals. Figures 2(c) and (d) show the AFM topography and phase images of an intercalated single crystal. In this case the image was acquired after fabrication of metal contacts. Focusing on the topographic features we immediately notice the absence of the substrate-related wrinkles, while a number of bubble-like structures are still present and the cross-directional wrinkles can still be seen (blue arrows). The disappearance of small wrinkles is attributed to the intercalation process: by separating the graphene layers, the intercalant allows them to relax. The AFM phase image can readily distinguish between different materials, as it represents the phase lag between the tip excitation signal and its motion that is due to the (viscoelastic) damping properties of the sample [37]. In figure 2(d) it can be seen that the bubble-like structures observed in topography show a clear phase contrast. Since no annealing was performed after the transfer of graphene form the copper substrate, the PMMA residues observed before intercalation are still present during the process. The distribution and density is, indeed, comparable with what is observed in the pristine crystal, suggesting that those structures are caused by the contamination of graphene, arising from the transfer process. Clustering of FeCl 3 around these areas can contribute to the observed contrast. The presence of wrinkles, in particular along the crossed lines which extends from the centre of the crystal to the sides of the hexagon, is still clearly visible.

APCVD graphene PDs
Having successfully intercalated single crystal multi-layer graphene, we proceed to study its optoelectronic properties. Figure 3(a) shows a schematic representation of a multiterminal device. Here the contacts have been positioned in parallel to the sides of the top-layer hexagon, as shown in the optical micrograph ( figure 3(b)). The resistance across two opposing pair of contacts, namely 1-4 and 3-6 in figures 3(a) and (b), was =  W -R 660 2 1 4 and -=  W R 670 2 3 6 , in agreement with previous measurements [16]. The photoresponse of the device was characterized using SPCM, employing different pairs of contacts. Figures 3(c) and (d) show the SPCM maps acquired with two different pairs of contacts (1-4 and 3-6, respectively). Both SPCM maps show PC I ph being generated across the whole device, with a net change in sign appearing at the centre of it, in the direction orthogonal to the contacts. The dashed black lines in panels (c) and (d) mark the position of the maxima of the PC. Measuring across both pair of contacts and summing the two signals we obtain a SPCM map which displays a clear fourfold symmetry of the PC, as shown in figure 3(e) where the black dashed lines are the same as in panels c and d. Superimposing these lines, extrapolated from the SPCM map, onto the AFM maps of the same device, shown in figures 2(c) and (d), we see that they match with the observed crossed wrinkles in the graphene crystal.
The role of grain boundaries and wrinkles in graphenebased PDs has been studied via near-field PC nanoscopy, where the presence of grain boundaries was associated with a reversal in the sign of the PC while enhanced PC was observed in the presence of wrinkles [38]. The growth of hexagonal domains by APCVD is known to give high-quality, defect-free, graphene. Therefore, we do not expect any grain boundaries to be present across the device. All our observations point towards the fact that the observed PC is related to the intercalation spatial inhomogeneity, and thus to the doping inhomogeneity, of the graphene crystals. The presence of wrinkles is likely to create clusters of FeCl 3 , increasing the level of doping in those regions, while sign reversal is related to a sharp change in doping, forming a pp + junction in the highly doped graphene. These junctions can therefore act as photoactive centres giving the observed photoresponse for the overall device.
To confirm these hypotheses we acquired a Raman map of the device (see methods). In figures 4(a)-(d) we report the fit of Lorentzian peaks to the G-band region, as previously discussed (see figure 1(b)). Figures 4(a) and (b) show the height and position of the G 0 peak, respectively, while the G 1 band is shown in figures 4(c) and (d). The presence of a blueshifted G 0 peak agrees very well with the boundaries of the crystal and indicates areas with partial intercalation of FeCl 3 within the graphene layers. We observe that the G 1 peak height decreases significantly across a sign reversal in PC (the cross-shaped structure). This decrease in height is also accompanied by a redshift of the G 1 peak, from 1612 cm −1 to~-1605 cm 1 , in the same region, as shown in figure 4(d). We used the model developed by Lazzeri et al [30] to extrapolate the total density of holes in doped graphene from the position of the G-band peaks. It has been shown that this model gives an estimation of the charge density within 10% of the values extrapolated with transport techniques [17]. The results are shown in figure 4(e): here we see that a gradient of charge density is present along the two lines determined by the PC maps, confirming our hypotheses.
The PMMA residues observed in figures 2(a)-(d) do not contribute to the observed doping, and therefore to the photoresponse, since the charge transfer due to residual PMMA (∼10 11 cm −2 , see [36]) is negligible compared to the high level of doping induced by the intercalation with FeCl 3 (>4 10 13 cm −2 ).

Towards graphene-based PSDs
The unique PC distribution, with a four-fold symmetry, observed in our device (figure 3(e)), can be used for positionsensitive applications. Commercial PSDs exploit the photovoltage generated parallel to a semiconductor junction in the presence of local illumination [9]. This effect has been extensively studied in the past [10,11] and is characterized by a bipolar linear photoresponse as function of illumination position. Figures 5(a) . We can clearly see that the extremes of I ph are located where n h crosses n h , i.e. in the presence of a p-p + junction. At the same time, bipolar linear regions are present (green lines), where the PC changes sign at the centre of the photoactive junction. This behaviour is very similar to what is observed in a lateral photoactive junction [9]. In this case the junction is formed between areas of different doping in the functionalised graphene, induced by the inhomogeneity of the FeCl 3 intercalation driven by the wrinkles in the pristine APCVD single crystals. Therefore, exploiting a multi-terminal geometry as the one adopted here, it is possible to use these linear regions to determine the position of a focused light spot on the device.

Conclusion
In conclusion, we have shown that multilayer APCVD-grown hexagonal crystals of graphene can be intercalated with FeCl 3 and used as PSDs. We characterized the degree of intercalation and the resulting doping of graphene, showing the formation of multiple p-p + junctions associated with the presence of wrinkles in the pristine APCVD crystals. We then characterized the photoresponse of a multi-terminal device showing a strong spatial correlation between the observed PC and the p-p + junctions. The photoresponse is found to be a linear function of the laser spot position and changes sign at the centre of each junction. This behaviour, combined with a four-fold pattern in the spatially resolved PC, makes these devices ideal candidates for position sensitive detection of focused light. Furthermore, all measurements have been repeated after ∼11 months exposure of the device to environmental conditions, showing no change in the photoresponse, in agreement with the previously reported stability of this material [19]. These findings pave the way to additional functionality in graphene-based optoelectronic devices and open a new route towards flexible, lightweight, transparent and highly stable PSDs, with possible employment in smart textiles and wearable electronics [39].