1. Introduction

Graphene, a two-dimensional and sp²-bonded carbon material, has attracted considerable attention because of its remarkable properties, notably superior electrical and thermal conductivity, excellent mechanical stability, and high optical transmittance [1–4]. Defects in graphene, such as sp³-bonding, vacancies, grain boundaries, wrinkles, and cracks, can be generated during growth, transfer, and device fabrication processes [5, 6]. Since defects in graphene can modify its fundamental characteristics, such as its chemical, mechanical, magnetic, and electrical properties, defect engineering can be used to fine tune its intrinsic properties or open the local bandgap [7–10]. However, these defects degrade the elastic modulus and mechanical strength of the material [7]. In addition, defective regions in graphene are more chemically reactive than pristine graphene owing to their stronger binding energy [8, 9]. In this respect, the electrical characteristics and chemical reactivity of graphene can be enhanced by combining a controlled introduction of defects with other processes, such as chemical doping and film deposition [11–16]. Oh et al. reported on the effects of AuCl₃ doping on defective graphene, which reduced the sheet resistance and improved the long-term stability of chemical doping [11]. Seo et al. demonstrated the growth of high quality cadmium sulfide on ultraviolet (UV)/ozone-treated graphene, employed as a seed layer [12].

Cadmium telluride (CdTe), a II-VI compound semiconductor material, has a direct bandgap of ~1.5 eV, and it has been considered for numerous applications in optoelectronic devices such as solar cells, photodetectors, and X-ray and gamma-ray detectors for medical imaging because of its excellent optical properties, which include a high light absorption coefficient and a high specific power [17–19]. Adding a small fraction of Zn in CdTe produces alloys (Cd_(1-x)Zn_xTe) well-suited to radiation detection because of their high resistivity, high mean atomic number, and high density [19]. Furthermore, flexible CdTe thin-film-based solar cells fabricated on metal foil, polymer, and thin glass substrates have been demonstrated recently [20–22]. Mahabaduge et al. prepared superstrate CdTe devices on 100-μm-thick thin glass, yielding a cell efficiency of 16.4% [22]. In spite of advantageous optical properties however, CdTe has seldom been considered for photodetecting applications. Recently though, the excellent crystallinity, high light harvesting efficiency, and charge collection efficiency of CdTe have been exploited to fabricate nano- and microstructured photodetectors. However, the fabrication of nano- and microwire-based devices typically involves time-consuming and low-yield processes, such as contact printing, dielectrophoresis, and electron-beam lithography [23–26]. Using a contact printing method, Xie et al. fabricated heterojunction p-CdTe nanoribbon/n-silicon nanowire-based photodetectors sensitive to visible to near-infrared illumination [23]. The direct growth of self-aligned nano-/microstructure on substrates is essential to simplify and increase the yield of nano-/microelectronics fabrication processes.

In the work reported here, graphene was employed as a seed layer to induce the self-aligned growth of microstructured CdTe thin films. Graphene has considerable advantages over other seed layers in this context; in particular, the resulting structure grows in very small thickness increments because of the ultra-thin graphene layer, which is an excellent barrier material. In addition, the structure can be used as a back contact electrode for CdTe-based solar cells, while restricting growth to specific patterned graphene areas allows the selective fabrication of micro-structured CdTe devices. In this paper, we demonstrate the growth of self-aligned CdTe thin films on a pre-defined graphene layer containing defects generated by UV/ozone treatment. Patterned CdTe photodetectors were fabricated from the grown structures, and their electrical properties and photoresponses were investigated.

2. Experimental details

The fabrication process for the patterned CdTe photodetectors is shown in Fig. 1. Commercial monolayer graphene, grown on Cu foil by chemical vapor deposition (CVD) method, was transferred to the SiO₂/Si substrate using a poly(methyl methacrylate)-based wet transfer method [Fig. 1(a)]. The wet-transfer process was repeated to obtain bilayer graphene. Details of this typical wet-transfer method can be found elsewhere [11, 27]. Conventional photolithography processes, namely the spin-coating of photoresist (PR), UV exposure, and development, were performed to pattern the graphene. The exposed graphene was removed by reactive ion etching (RIE5000, SNTEK) in an oxygen plasma for 5 s at a power of 100 W under 20 sccm oxygen flow, as shown in Fig. 1(b). Defects in the bilayer graphene were generated by UV/ozone treatment for 30 min in the chamber of an UV cleaner (PSDP-UV4T, NOVASCAN), where wavelengths of 185 nm and 254 nm were emitted to generate ozone and promote chemical reactions. Dry air, containing less than 3 ppm H₂O, was injected into the chamber to control the humidity. The patterned CdTe structure was grown on the defective graphene by close-spaced sublimation (CSS) method under Ar ambient [Fig. 1(c)]. The temperatures of the CdTe source (99.999%, Alfa Aesar) and the substrate were maintained at 600 and 540 °C, respectively, where the temperature difference between the substrate and CdTe powder source drives CdTe growth. The pressure of the CSS chamber and the CSS growth time were 0.5 Torr and 5 min, respectively. The details of the CdTe growth procedure under CSS can be found elsewhere [28]. The metal contacts were also defined by a standard photolithography and lift-off process. Cu/Au (20/80 nm thick) electrodes were deposited using an electron-beam evaporator and annealed at 400 °C for 10 min under Ar ambient to form Ohmic contacts.

 

Fig. 1 Schematic of the fabrication process for patterned CdTe devices: (a) transfer of graphene to a SiO₂/Si substrate; (b) formation of graphene patterns by photolithography and oxygen plasma treatment; (c) selective growth of CdTe by CSS method; (d) deposition of Cu/Au electrodes by photolithography.

Download Full Size | PDF

The thickness and quality of the bilayer graphene were characterized using micro-Raman spectroscopy in backscattering geometry. The laser source was a 532-nm diode-pumped solid-state (DPSS) laser (Omicron-Laserage). Scanning electron microscopy (FE-SEM, S-4400, Hitachi) was utilized to investigate the surface morphology of the deposited CdTe thin films. Micro-photoluminescence (PL) spectroscopy was performed with a 532-nm DPSS laser to analyze the crystal quality of the CdTe thin films. Electrical measurements were conducted with a semiconductor parameter analyzer (4155C, Agilent) connected to a probe station. The photoresponse was characterized using a 365-nm UV lamp (15 W, UVItec LTD.). Neutral density filters (10, 30, 50, and 70%, Optosigma) were used to control the light intensity, which was measured with a laser power meter (FieldMax II-TO, Coherent).

3. Results and discussion

Figure 2(a) presents an optical microscope image of the patterned graphene on the SiO₂/Si substrate after oxygen plasma etching. The graphene regions that appear turquoise are well defined because the PR layer protected them from the reactive oxygen plasma. The role of the UV/ozone treatment was then to introduce defects in a controlled manner, because populated defects can promote crystal growth on the surface of the graphene. Figure 2(b) shows the Raman spectrum of the patterned graphene before and after UV/ozone treatment. Distinct G, 2D, and D bands are observed. The intensity ratios of the 2D to G bands and of the D to G bands are denoted I_2D/I_G and I_D/I_G, respectively. The latter is approximately 0.18 before UV/ozone treatment, indicating that defects may be introduced during the wet-transfer and photolithography processes. A small I_D/I_G and an I_2D/I_G of approximately 1.7 are typical characteristics of high-quality bilayer graphene. I_D/I_G increases sharply to approximately 1.7 after 30 min of UV/ozone treatment, indicating that the defects in the graphene become populated, while I_2D/I_G decreases from 1.7 to 0.76, as has been reported in other studies in which defects were introduced into graphene [7, 11]. Figure 2(c) is an optical micrograph of the self-aligned CdTe thin-film patterns that were grown by CSS for 5 min. There is a possible incorporation of Te into the graphene during CSS growth. The size and shape of the patterned CdTe structures correspond to those of the graphene in Fig. 2(a), demonstrating that the underlying defective graphene acts as a seed layer. Since CVD-grown graphene can be transferred onto arbitrary substrates, our growth method for patterned CdTe structures can be used to prepare substrate-configured devices on any material. An SEM image of the CdTe surface morphology is shown in Fig. 2(d), in which large grains (~5 μm) are observed. The thickness and growth rate of the CdTe layer are approximately 4.5 μm and 0.9 μm/min, respectively. The final structure of the patterned CdTe devices with Cu/Au Ohmic metallizations is shown in Fig. 2(e). The room-temperature micro-PL spectrum in Fig. 2(f) contains a single peak at 1.5 eV—in good agreement with the bandgap of CdTe [20]—whose full width at half maximum is 61 meV (34 nm), comparable to values previously reported for CdTe thin films on graphene and even for CdTe nanostructures [25–27, 29].

 

Fig. 2 (a) Optical micrograph of the patterned graphene. (b) Raman spectra of graphene before and after UV/ozone treatment. (c) Optical micrograph of the patterned CdTe grown by CSS method. (d) SEM image of the CdTe surface. (e) Optical micrograph of TLM-patterned CdTe after the deposition of Cu/Au electrodes. (f) Micro-PL spectrum of the CdTe grown on defective graphene.

Download Full Size | PDF

The optical and scanning electron micrographs of our CdTe devices obtained after the deposition of the Cu/Au electrodes [Figs. 3(a) and 3(b)] show that these are well defined. Since a high Schottky barrier originates from the high electron affinity of CdTe (4.5 eV), Cu was used to form an Ohmic contact because of its acceptor behavior in CdTe layers and high work-function [17]. Devices were prepared with distances between the neighboring electrodes of 10, 20. 40, and 80 μm. A cross-sectional view of the final structure of our devices is shown in the inset of Fig. 3(b). Current-voltage (I-V) characteristics were recorded to calculate the sheet resistance of the CdTe layer and the contact resistance of the interface between the CdTe and the annealed Cu/Au electrode. Figure 3(c) shows that the I-V behaviors are quasi-linear for all four electrode separations, the resistance increasing with increasing the distance between the transmission line measurement (TLM) patterns. This relationship was fitted using a linear function, as shown in Fig. 3(d), yielding the sheet and contact resistance via R_T = (R_Sh·L)/W + 2R_C, where R_T is the total resistance of the device, R_Sh is the sheet resistance, W is the width of the TLM pattern, L is the distance between the TLM patterns, and R_C is the contact resistance at the interface between the CdTe and the Cu/Au electrode. The values obtained for R_Sh and the specific contact resistance are 1.24 × 10⁸ Ω/sq. and 1.85 × 10² Ω·cm², respectively. The R_Sh obtained in this work is comparable with previously reported values [30, 31]. Since Cu acts as a p-type dopant in CdTe, the doping effect caused by the diffusion of Cu in the gap between TLM patterns might enhance the electrical properties of CdTe, and thus reduce its R_Sh. Also, the underlying graphene may contribute to the electrical conductance.

 

Fig. 3 (a) Optical and (b) scanning electron micrograph of the fabricated CdTe device. Inset: cross-sectional structure of the final CdTe-based photodetectors. (c) I-V characteristics of CdTe devices with different distances between the Cu/Au electrodes. (d) Resistance at + 1 V of the CdTe devices as a function of the distance between the neighboring electrodes.

Download Full Size | PDF

Figure 4(a) depicts the experimental setup used to measure the photoresponse of our patterned CdTe photodetectors. These were exposed to UV light (365 nm), passed through neutral density filters to tune its intensity to 0 (dark ambient), 1.6, 7.0, 23, 69, and 269 μW/cm². Figure 4(b) shows the I-V characteristics at each light intensity of the patterned CdTe devices with 10 μm gaps. Since the number of photo-induced free carriers increases with the light intensity, so do the resulting photocurrents. This relationship can be described by a power law that characterizes the photodetector. Figure 4(c) shows that the photocurrents at + 1 V closely follow the power law I = AP^B, where A is determined by the wavelength, P is the light intensity, and B represents the device’s photoresponse to UV illumination. Non-unity exponent values (B) have been reported for CdTe nanoribbons (0.2) and microwires (0.40) [25, 29]. Here, we report the highest B (0.47) among the reported values, which indicate that our CdTe thin films grown on graphene is of high quality. Figure 4(d) shows the dynamic photoresponse to UV light of our patterned CdTe photodetectors at + 1 V under various light intensities. On-off ratio of our CdTe photodetectors was as high as ~2.8 at a intensity of 269 μW/cm². Although the photocurrents are proportional to that of the light intensities, the photoresponse is highly reproducible for all intensities. The average response time, defined as the time necessary for the photocurrent to reach 90% of its saturated value from the baseline, is found to be 10.4 s, slower than has previously been reported for other CdTe nanostructure-based photodetectors [24, 25, 29]. Post-growth processes, such as CdCl₂ activation or nitric-phosphic acid etching, will therefore be investigated as a means to improve the optoelectronic characteristics of these patterned CdTe devices.

 

Fig. 4 (a) Schematic of the experimental setup used to measure photoresponses. (b) I-V characteristics under different UV light intensities of a CdTe photodetector with a 10 μm gap. (c) Photocurrent intensity at + 1 V under different UV light intensities. (d) Time-dependent photocurrents at various UV illumination intensities.

Download Full Size | PDF

4. Conclusion

Self-aligned CdTe thin-film structures were successfully grown by CSS method, using patterned graphene as a seed layer. Graphene defects were intentionally introduced through UV/ozone treatment. The higher chemical reactivity of defects in graphene enables the selective growth of high quality CdTe thin films, as confirmed here by micro-PL spectroscopy and SEM. The excellent electrical properties and photodetecting performance of the resulting devices—delivering intense, reproducible photocurrents with fast response/decay—suggest that our graphene-integrated patterning technique is a promising advance in view of fabricating graphene-semiconductor thin film devices on a wide range of substrate materials.