Broadband photodetector based on carbon nanotube thin film/single layer graphene Schottky junction

In this study, we present a broadband nano-photodetector based on single-layer graphene (SLG)-carbon nanotube thin film (CNTF) Schottky junction. It was found that the as-fabricated device exhibited obvious sensitivity to a wide range of illumination, with peak sensitivity at 600 and 920 nm. In addition, the SLG-CNTF device had a fast response speed (τr = 68 μs, τf = 78 μs) and good reproducibility in a wide range of switching frequencies (50–5400 Hz). The on-off ratio, responsivity, and detectivity of the device were estimated to be 1 × 102, 209 mAW−1 and 4.87 × 1010 cm Hz1/2 W−1, respectively. What is more, other device parameters including linear performance θ and linear dynamic range (LDR) were calculated to be 0.99 and 58.8 dB, respectively, which were relatively better than other carbon nanotube based devices. The totality of the above study signifies that the present SLG-CNTF Schottky junction broadband nano-photodetector may have promising application in future nano-optoelectronic devices and systems.

Scientific RepoRts | 6:38569 | DOI: 10.1038/srep38569 to detect very fast optical signal with frequency as high as 5400 Hz, the rise/fall time were estimated to be 68 and 78 μ s, respectively. What is more, the on-off ratio, responsivity, and detectivity of the device were calculated to be 10 2 , 209 mAW −1 and 4.87 × 10 10 cmHz 1/2 W −1 , which are higher than other carbon nanotubes based devices. This study suggests that present SLG-CNTF Schottky junction device will have potential application in future optoelectronic devices. Figure 1a illustrates the detailed procedure for the fabrication of the SLG-CNTF Schottky junction nano-photodetector. Briefly, the fabrication starts with the self-assembly of monolayer CNTF through an Langmuir-Blodgett (LB) approach, during which uniaxial compression of subphase will lead to the self-assembly of carbon nanotube over a large area in a parallel way. The as-assembled monolayer CNTF was transferred on to a SiO 2 /Si wafer, and then partially covered with a SLG film, followed by painting of silver paste on the surface. According to previous study, only gold can forms Ohmic contact to the CNTF with negligible contact resistance 26 , therefore, in this study gold film with 50 nm was deposited on the CNTF side. Figure 1b shows a representative microscopy image of the SLG-CNTF interface. It is clear that due to their distinct contrast, both the SLG and CNTF can be easily visualized. Further scanning electron microscopy (SEM) image reveals that the as-assembled CNTF was composed of large-area flexible carbon nanotube array with only single layer [ Fig. 1c and d]. To unveil the microstructure of both SLG and CNTF, the Raman spectra of both materials were further analyzed. As shown in Fig. 1e, for the graphene, there are totally three peaks: a 2D peak (2680 cm −1 ) and a G peak (1590 cm −1 ) with an intensity ratio of I 2D :I G of 3:1, and a weak D-band scattering at ~1340 cm −1 , confirming the single-layer of the graphene 27,28 . The sheet resistance of the graphene, according to our further electrical study is around 500-800 Ω/□ . With regard to the CNTF, similar three peaks can be observed. In addition, there is an obvious signal in the range from 100-300 cm −1 which can be readily ascribed to the radial breathing mode (RBM) of single-walled carbon nanotubes 29 . Figure 2a shows a typical room-temperature I-V curve of the SLG-CNTF Schottky junction both in dark and under illumination of 980 nm, from which one can observe clearly that the SLG-CNTF interface in dark exhibits typical rectification behavior, with an on/off ratio of ~1 × 10 2 at ± 3 V. In view of the fact that there is not obvious contact resistance between the SLG and silver (Fig. 2b) 30 , and the gold can form good Ohmic contact with CNTF 26 , the above rectifying characteristic can be exclusively attributed to the SLG-CNTF interface. In order to quantitatively evaluate the SLG-CNTF Schottky junction, the barrier height was then deduced by using the thermionic-emission based diode equation 31 :

Results
where J (T, V) is the current density across the CNTF-SLG interface, V the applied voltage k B the Boltzmann's constant, T the absolute temperature, n the ideality factor = (n (q/kT)(dV/dInI)), respectively. The prefactor, Js(T) is the saturation current density and can be estimated by Js(T) = A*T 2 exp(− eφ SBH /K B T), where φ SBH is the zero bias Schottky barrier height (SBH), A* the Richardson constant, m* the effective mass of the charge carriers, respectively. For CNT, A* is theoretically calculated to be 7.12 × 10 4 Am −2 k −2 (m e * = 0.06m 0 ) 32 , using the J s value, the Schottky barrier height at the SLG-CNTF interface was calculated to be 0.78 eV. It is interesting to note that, when irradiated by IR light, such a SLG-CNTF Schottky junction exhibited obvious sensitivity at both negative and forward bias voltage [ Fig. 2a]. Figure 2b shows that the switching photoresponse of CNTF-SLG device when the IR irradiation (wavelength: 980 nm, intensity: 166 mWcm −2 ) was switched on and off repeatedly. Apparently, the device can be reproducibly switched between high and low conductance states. The on and off currents were estimated to be 5 nA to 1.2 μ A, yielding an on-off ratio as high as 240.
The above photoresponse at zero bias voltage indicated that the SLG-CNTF Schottky junction photodetector can act as a low-consumption device that is capable of detecting IR irradiation without external power supply. To further determine the response rate and explore the feasibility of the present device for practical application in optical switches, the response of the SLG-CNTF Schottky junction photodetector to high-frequency optical signal was studied. During the study, an oscilloscope was used to probe the variation of the photocurrent under pulsed light with varied frequencies. Figure 3a and b depict the photoresponse of the SLG-CNTF device to pulsed IR irradiation at frequency of 500 Hz and 5 kHz, respectively. It can be seen that the device can be alternatively switched between on-and off-state states with excellent reproducibility even under a light pulse with frequency as high as 5 kHz. Further one single cycle of photoresponse under 5 kHz light illumination was plotted in Fig. 3c. By deducing the rise and fall edges, the rise time (τ r , duration for the photocurrent to rise from 10 to 90%) and the fall time (τ f , duration for the photocurrent to decrease from 90 to 10%) were estimated to be 68 and 78 μ s, respectively, which are much faster than not only the SWCNT/PCBM heterojunction device (55/37 microseconds), but also than the graphite quantum dots/graphene heterojunction (several seconds) 33 . Understandably, such a fast response speed could be associated with the high-quality Schottky junction formed between the CNTF and SLG, which can facilitate the effective and rapid separation of the photo-generated carriers 34 . According to the normalized photocurrent at different frequencies shown in Fig. 3d, the 3 dB bandwidth is estimated to be 5400 Hz (corresponding to the frequency when the normalized photocurrent decreased from 1 to 0.707), which is much higher than the ZnO/SLG Schottky junction device (< 1500 Hz) 18 and Ge/SLG Schottky junction device (< 3000 Hz) 23 , such excellent photoresponse renders the current device a potential building block for future optoelectronic device applications.
In fact, the photosensitivity of our device is highly dependent on the bias voltage. Figure 4a plots the photoresponse under different bias voltage, from which it can be seen that both photocurrent and dark current will increase monotonously with increasing bias voltage. In order to quantitatively assess the bias voltage dependent photoresponse of the SLG-CNTF Schottky junction photodetector, two key device parameters including both responsivity (R) 35 and detectivity (D*) 36,37 , were calculated by the following formulas:   these two metrics can be further increased when the bias voltage continues to decrease. Table 1 summarizes the response speed, responsivity and detectivity of the present device and other CNTs based photodetectors. It is obvious that the SLG-CNTF Schottky junction has a high responsivity (209 mAW −1 ) and detectivity (4.87 × 10 10 cmHz 1/2 W −1 ), which are not only better than the device solely composed of pure SWCNTF 9 and MWCNTF 11 , but also than the devices assembled from SWCNT/PCBM 38 , and MWCNT/graphene nanocomposite structures 13 . Further photocurrent study at different light intensities (from 0.25 to 12 mW cm −2 ) reveals that the photocurrent (I p ) of the SLG-CNTF Schottky junction PD exhibits high dependence on the intensity of incident IR. As exhibited in Fig. 5a, the photocurrent of the SLG-CNTF Schottky junction PD increases gradually with increasing intensity. Numerical analysis of the relationship between photocurrent and light intensity reveals that the photocurrent can be described by a simple power law: I p = CP θ 39 where C is a constant for the IR light, and the exponent θ (0.5 < θ < 1) determines the sensitivity of photocurrent to the incident IR intensity. By fitting the above formula to the experimental data, θ is determined to be 0.99 [Fig. 5b]. Such a nearly integer value suggests a low density of trap states in this SLG-CNTF photodetector. One of the figure-of-merit of the device is the linear dynamic range (LDR, normally quoted in dB) which is given by:  where I photo is the photocurrent measured at a light intensity of 1 mW/cm 2 . By using the above equation, the LDR is estimated to be 58.8 dB which is higher than that of the SWCNT/PCBM heterojunction (40 dB) 38 . Such a relatively large LDR suggests the great linear response characteristic of the device. In order to further unveil the light intensity dependent photoresponse of SLG-CNTF device, the responsivity and detectivity of the nano-photodetector in the low range from 0 to 12 mW/cm 2 were studied. As plotted in Fig. 5c, the responsivity and detectivity will gradually increase with increasing intensity. On the contrary, with further increase of the intensity, the responsivity and detectivity begin to saturate, which is due to the reduced carrier recombination rate at high light intensity 40 . Figure 6 plots the responsivity and IPCE of the CNTF-SLG Schottky junction PD as a function of wavelength (To make the analysis more reliable, we kept the light power identical for all wavelengths during testing). Apparently, the SLG-CNTF Schottky junction PD shows a broadband sensitivity: namely it is nearly sensitive to all irradiation in the range from 300-1100 nm. Obviously, such a broadband spectral selectivity is primarily due to the operation mechanism. The generation of the photocurrent in the Schotty junction device is limited to by the electrical property of semiconducting carbon nanotube, whose bandgap is around 1.1 eV, corresponding to 1100 nm in wavelength. When shined by light illumination, the photons with energy larger than the bandgap of carbon nanotube (1.1 eV) can lift the electrons in the valence band to the conduction band, and therefore they can contribute to the photocurrent. In addition, there are two obvious peaks at 600 and 920 nm. Such a spectral selectivity is similar to the incident photon-to-electron conversion efficiency (IPCE), which can be calculated by the following equation where λ is the wavelength of illuminated light (320-1100 nm), I p and I d are the photocurrent and the dark current, respectively. As indicated by the red curve in Fig. 6a, the device has an IPCE exceeding 20% over a broad range (320-1050 nm), with two maximum values (43% at 600 nm, and 31% at 900 nm). The detailed origin for this phenomenon is still unknown and needs further exploration. In addition, the present PD has very good stability. It can virtually keep the same photocurrent and on/off ratio even the device was stored in ambient condition for 3 months (Fig. 6b). The operation of the SLG-CNTF Schottky junction IR photodetector can be interpreted by the energy band diagram illustrated in Fig. 7. When metallic SLG was transferred onto the CNTF, a built-in potential in CNTF near the interface (the depletion region) was formed due to the difference in work functions between the SLG and CNTF. Such a SLG-CNTF interface, like conventional metal-semiconductor junction, can allows the current to flow only in one direction, thus, leading to a rectifying behavior. When irradiated by IR light, the semiconducting CNTF will absorb photons with sufficient energy, and then form huge amount of electron-hole pairs. The photo-induced electrons-holes then drift to the SLG-CNTF interface, and are separated by the built-in electric field. Finally, the electron and holes will be collected by Ag and Au electrode, respectively, leading to the formation of photocurrent in the external circuit. Note that during this detecting process, thanks to the high quality of the SLG-CNTF Schottky junction, the photo-generated electron-holes pairs will be effectively and rapidly separated 41 . As a consequence, the present device is able to detect IR signal with a relatively rapid response.
In conclusion, we have demonstrated a highly sensitive nano-photodetector by coating a layer of SLG onto a CNTF which was obtained by an LB method. Device performance analysis reveals that the as-assembled SLG-CNTF device exhibits high sensitivity to a wide spectrum of light illumination, with two peak sensitivity at 600 and 920 nm, respectively. Specifically, the device was able to probe fast-switching optical signal with frequency as high as 5400 Hz with very good reproducibility. The rise/fall times were estimated to be 68 and 78 μ s, respectively. What is more, the on-off ratio, responsivity, and detectivity of the device were calculated to be 10 2 , 209 mAW −1 and 4.87 × 10 10 cmHz 1/2 W −1 , respectively, which are higher than that of other photodetectors. This study suggests that the current CNTF/SLG Schottky junction broadband photodetector will have potential application in future optoelectronic devices.

Synthesis and Characterization of SLG Film and CNTF. The single layer graphene (SLG) films
were fabricated via a chemical vapor deposition (CVD) method at around 1000 °C using a mixed gas of CH 4 (40 standard cubic centimeter per minute, SCCM)) and H 2 (20 SCCM) as precursors 18 . After deposition, the 25 μ m thick Cu foil covered with SLG film was spin-coated with polymethylmethacrylate (PMMA) solution (5 wt% in chlorobenzene) and the Cu substrates were etched away by the Marble's reagent solution. The mono-layer carbon nanotube array was assembled by using an Langmuir-Blodgett trough (Nima Tech., 312D). Briefly, purified nanotube soot with diameter of 1 nm was firstly dispersed in a mixed solution of trichloromethane and N,N-Dimethylformamide (DMF) with a volume ratio of 1:1. After the trough was filled with distilled water, the resultant solution (100 μ L) was then dispensed on the water sub-phase by using a 20 μ L syringe drop by drop. The carbon nanotube layer was then compressed carefully until multilayer was formed. Finally, the monolayer nanotube array was loaded onto a silicon SiO 2 (500 nm)/Si substrate by slowly lifting the substrate from the sub-phase 42 . The Raman study of both SLG and carbon nanotube was performed on a Raman spectrum (JY, LabRAM HR800).

Device Fabrication and Characterization.
To fabricate the CNTF/SLG Schottky junction photodetector, 50 nm Au electrode which served as electrical contact for graphene, was first deposited onto the one side of CNTF on SiO 2 substrate using E-beam evaporation. Then the PMMA-supported SLG were directly transferred onto the other side of CNTF on SiO 2 substrate. After drying at 100 °C for 10 min, the residual PMMA on SLG was removed by acetone. The device characteristics of CNTF/SLG were measured using an I-V characterization system (Keithley Company, SCS-4200). To determine the spectral response and time response of the Schottky junction device, a home-built system composed of a light source (Energetiq, EQ99-X), a monochromator (Princeton Instrument, SP2150) and an oscilloscope (Tektronix, DPO 5104B) was used.