Copyright 2015 American Scientific Publishers All rights reserved Printed in the United States of America Journal of Nanoelectronics and Optoelectronics Vol. 10, 1 5, 2015 Theoretical Study on Graphene Silicon Heterojunction Solar Cell Yawei Kuang, Yushen Liu, Yulong Ma, Xuekun Hong, Xifeng Yang, and Jinfu Feng School of Physics and Electronic Engineering, Changshu Institute of Technology, Changshu, 215500, China The performance of graphene based heterojunction solar cell on silicon substrate has been studied theoretically by Technology Computer Aided Design (TCAD) tools. The current voltage curves and internal quantum efficiency of this device have been calculated at different conditions using tow dimensional model. The results indicate the power conversion efficiency of graphene silicon junction cell strongly dependents on the work function of graphene and the physical properties of silicon such as thickness and doping concentration. In particular, the dark current got a sharp rise at higher concentration of 1e17 cm 3 for n-type silicon, which implies a convert of electron emission mechanism. The biggest fill factor got at higher phos doping predicts a new direction for higher performance graphene heterojunction solar cell design. Keywords: Graphene, Heterojunction Solar Cell, TCAD. 1. INTRODUCTION Graphene, a single atom layer of carbon hexagons, has attracted much attention owing to its unique structure and fascination. 1 Graphene has been produced in the form of ultrathin sheets consisting of one or a few atomic layers by chemical vapor deposition or mechanically exfoliation and can be transferred to various substrates, which open a wide field of potential applications such as high performance electronic devices, photo sensors, and smart composite. 2 In particular, the graphene film has a unique combination of high electrical conductivity and optical transparency in visible and near-infrared regions, which make it a good candidate for use in solar cells. 3 4 More recently, graphene based heterojunction solar cells have been fabricated on various substrates such as Si, 5 CdS 6 and CdSe, 7 much process has been achieved in the past several years. The graphene silicon schottky solar cells could be fabricated at room temperature which shows great potential in light harvesting and conversion application with the advantage of low cost, facile processibility and environmental amity. 8 In this structure graphene film not only serves as a transparent electrode for light transmission but also as an active layer for electron/hole separation and carrier transporting medium. 9 Although the first reported energy conversion efficiency is only 1.65%, 10 the performance of graphene silicon schottky solar cells have been improved by silicon nano-array substrate Author to whom correspondence should be addressed. adoption, 11 chemical doping 5 and Graphene/P3HT/Silicon configuration 12 ranging from 1.96% up to 10.3%. It is found that surface charge recombination as well as graphene conductivity along with the work function played important roles on determining the performance of device. However, the schottky barriers have not been studied thoroughly. The application of graphene in schottky solar cell requires much more investigation. In this paper, simulation of graphene silicon heterojunction solar cells is carried out using Technology Computer Aided Design (TCAD) tools. The simulation program solves the Poisson, the continuity and the current density equations by using a standard procedure for amorphous materials including the continuous density of state model, schottky and auger recombination mechanisms. The dependence of these parameters of optical parameters with the photon energy has been included, taking into account the doping level, graphene work function, thickness of silicon substrate and their effect on cell efficiency. 2. STRUCTURE As mentioned above, Figure 1 shows the cross section of the graphene silicon solar cell simulated using TCAD tools. The device consists three regions which are silicon substrate, SiO 2 window and graphene layer from bottom to the top area. The 10 nm graphene layer is coated onto the silicon substrate with 1 m 12 m oxide window. For this structure, most reports demonstrate that the heterojunction device fabricated from a transferred J. Nanoelectron. Optoelectron. 2015, Vol. 10, No. xx 1555-130X/2015/10/001/005 doi:10.1166/jno.2015.1815 1
Theoretical Study on Graphene Silicon Heterojunction Solar Cell Kuang et al. χ silicon E 0 ф G hv Graphene N-type silicon n ф silicon Ec Ev E f Fig. 2. Energy diagrams of graphene n-type silicon solar sell. Fig. 1. Cross section of the graphene silicon schottky solar cell. chemical vapor deposition graphene layer on silicon that could avoid high cost deposition techniques and complicated processing. 13 14 Basically, any semiconductor can form a schottky junction with a certain metal if their work function difference is big enough, and the carrier density of the semiconductor is moderate. The calculations indicate that graphene films form a schottky junction with silicon, which is favorable for producing a relatively large built-in field. 10 Under illumination, the photo excited electrons and holes are generated in the silicon substrate and then separated by the build-in electric field at the schottky junction. Electrons and holes are collected by bottom electrodes and graphene film which forms a photo voltaic action. The non-linear I V characteristic of the schottky junction can be expressed by the thermionic emission model: I s = AA T 2 e B/KT where A is the contact area, A is the effective Richardson constant, T is the absolute temperature, K is Boltzmann constant, and B is the barrier height (1) B = G (2) G is the work function of graphene, is the electron affinity for n-type semiconductor. Under illumination, the photo excited electrons and holes are generated in the silicon substrate and then separated by the build-in electric field at the schottky junction. Electrons and holes are collected by bottom electrodes and graphene layer respectively which forms a photo voltaic action for n-type silicon as shown in Figure 2. 3. PHYSICS BASED TCAD MODEL To investigate the performance of graphene based solar cell, simulation were carried out using TCAD, which divides into two steps since there is not graphene default involved in the material declination. Firstly, we deposit 10 nm aluminum films as anode electrode instead of graphene film using ATHENA tool to generate the device structure. Secondly, we redefine the anode electrode material as graphene layer using ATLAS tool. Graphene is modeled as a semi-metal, a carrier mobility of 15,000 cm 2 /Vs. 16 The carrier densities were calculated by Fermi distribution and by adjusting the values of effective masses, a thickness of 10 nm and band gap such that the cattier densities agree with experimental results. It is important to point out that for multilayer graphene film, the E K dispersion relationship is weakly parabolic rather than linear as in the case of monolayer graphene. Simulation parameters used in Atlas tool for this cell are listed in Table I. 4. RESULT 4.1. Silicon Thickness Effect We choose 100 silicon crystal as absorption area. Graphene acts as a transparent electrode that is for the intensity of photo generated carriers. A schottky junction was built in for the difference between their work functions. Figures 3(a) and (b) show the photo generation rate of schottky solar cell under AM1.5 illumination with depth for absorption layer were maintained at 2 m and 20 m Table I. Simulation parameters for graphene silicon schottky solar cell. Parameter Description Value E g Silicon band gap 1.08 ev Nc Effective density of 2.8e19 cm 3 states in CB Nv Effective density of 1.04e19 cm 3 states in VB Silicon electron 4.17 ev affinity e Silicon electron mobility, 300 K 16 No doping: 1000 cm 2 /Vs Phos doping 1e15: 1300 cm 2 /Vs Phos doping 1e16: 1076 cm 2 /Vs Phos doping 1e17: 491.1 cm 2 /Vs h Silicon hole mobility, 300 K 16 No doping: 500 cm 2 /Vs Phos doping 1e15: 675 cm 2 /Vs Phos doping 1e16: 460.9 cm 2 /Vs Phos doping 1e17: 331 cm 2 /Vs 2 J. Nanoelectron. Optoelectron. 10, 1 5, 2015
Kuang et al. Theoretical Study on Graphene Silicon Heterojunction Solar Cell (a) 2 µm (b) 20 µm Fig. 4. Current voltage curves of graphene solar cells versus different silicon thickness. is characterized from 0.021% up to 0.122% (listed in Table II) because of thicker silicon substrate. 4.2. Graphene Work Function Effect According to the theory model described in Part. 2, the battier height B is related to the difference between graphene Fig. 3. Photo generation rate of graphene silicon schottky solar cell under AM1.5 illumination versus silicon thickness: (a) 2 m (b) 20 m. respectively. The light could be absorbed in barrier layer and inside the semiconductor. Both results show the most effective absorption area located in the surface attachment about 1 m. Meanwhile, Figure 3(a) shows that 2 m thickness is not enough for full spectrum absorption compared with 20 m thickness. To investigate the effect of silicon thickness on the performance of solar cell, we calculate the current voltage curves and internal quantum efficiency (IQE) shown in Figures 4 and 5 respectively. Long wavelength has better quantum efficiency since silicon is an indirect band gap material. When the thickness of silicon crystal gets smaller, most of long wavelength light past through the device and decrease the internal quantum efficiency. On the other hand, the dark current is increased greatly due to the increase of carrier recombination rate on back electrode. It is found that the efficiency of device Fig. 5. IQE of graphene solar cells versus different silicon thickness. Table II. Efficiency versus silicon thickness, graphene work function and Phos doping concentration under AM1.5 illumination. Silicon Graphene thickness work Silicon phos J sc V oc (um) function (ev) doping (cm 3 ) (ma/cm 2 ) (V) FF (%) 2 4.6 No doping 1.431 0.035 0.26 0.021 20 4.6 No doping 3.331 0.082 0.27 0.122 20 4.7 No doping 4.667 0.118 0.31 0.284 20 4.8 No doping 5.612 0.141 0.36 0.478 20 4.8 1e15 6.055 0.187 0.42 0.793 20 4.8 1e16 6.475 0.224 0.52 1.257 20 4.8 1e17 5.722 0.158 0.58 0.874 J. Nanoelectron. Optoelectron. 10, 1 5, 2015 3
Theoretical Study on Graphene Silicon Heterojunction Solar Cell Kuang et al. Fig. 6. Current voltage curves of graphene solar cells versus graphene work function. work function G and electron affinity of silicon. Consequently, the higher work function will increase the B,and further enhance the built-in potential V bi via the equation, V bi = B V n,wherev n means the distance between Ec and Ef in silicon. Therefore, the increase of graphene work function implies the increase of V bi which corresponding to the upper limitation of V oc. As shown in Figures 6 and 7, we calculated the current voltage curves an IQE of graphene silicon schottky solar cells versus different graphene work function while the electron affinity of silicon in all devices are fixed. It is found that the power conversion efficiency and IQE are both increased monotonically according to the increase of built-in potential V bi. This result does not consistent with experiment results got by YF.Li group. 17 They increased the work function of graphene by increasing the number of graphene layers. The power conversion efficiency increases when the number of graphene layers is less than 4 and Fig. 8. Current voltage curves of graphene solar cells versus phos doping concentration. gets reduction when the number of graphene layers further increases. 4.3. Silicon Phos Doping Effect It is known that the phos doping in silicon can increase the Fermi energy level and intensity of photo generation carriers. For doping concentration below 1e17 cm 3, the shortcircuit current and open-circuit voltage are both increased due to higher barrier height compared with purity silicon substrate, which are shown in Figure 8. IQE result also indicates a higher rate for doping condition as shown in Figure 9. However, the performance of this device is completely different when the doping concentration is up to 1e17 cm 3. Firstly, the dark current got a sharp rise compared with lower doping concentration. This phenomenon may be caused by the convert of electron emission mechanism. Instead of thermionic emission mechanism, the tunneling emission mechanism dominates the carrier Fig. 7. IQE of graphene solar cells versus graphene work function. Fig. 9. IQE of graphene solar cells versus phos doping concentration. 4 J. Nanoelectron. Optoelectron. 10, 1 5, 2015
Kuang et al. emission from silicon to graphene. As a result, the opencircuit voltage gets a small reduction. Secondly, the high level doping decrease the lifetime of photo generation carriers especially for long wavelength light, these losses of carriers collection make the decrease of short-circuit current. Thirdly, we notice that the fill factor of 1e17 cm 3 doping level is the biggest though the power conversion efficiency is smaller than other two doping concentration, shownintableii. 5. CONCLUSIONS Two dimensional simulation of graphene silicon schottky solar cell is carried out by Silvaco TCAD tools. The process of structure generation and material definition is described in the front section. Furthermore, we compare the performance versus different silicon thickness, graphene work function and phos doping concentration in detail. The results show that sufficient absorption thickness, bigger work function of graphene and moderate doping in silicon are better for the improvement of power conversion efficiency. The successful implementation of high level doping condition shows a convert of electron emission mechanism from thermionic emission mechanism to tunneling emission mechanism, which has the biggest fill factor. Theoretical Study on Graphene Silicon Heterojunction Solar Cell Acknowledgments: This work was supported by National Natural Science Foundation of China (Nos. 11247028, 61404012, and 11347021). References and Notes 1. P. Avouris and M. Freitag, IEEE Journal of Selected Topics in Quantum Electronics 20, 72 2. D. Chen, H. Zhang, Y. Liu, et al., Energy and Environmental Science 6, 1362 (2013). 3. P. Gao, K. Ding, Y. Wang, et al., J. Phys. Chem. C 118, 5164 4. Z. Zhang, T. Cui, R. Lv, et al., Journal of Nanomaterials 2014 5. X. Miao, S. Tongay, M. K. Petterson, et al., Nano Lett. 12, 2745 (2012). 6. Y. Lei, R. Li, F. Chen, et al., J. Mater. Sci.: Mater. Electron. 25, 3057 7. L. Zhang, L. Fan, Z. Li, et al., Nano Research 4, 891 (2011). 8. T. Feng, D. Xie, Y. Lin, et al., Appl. Phys. Lett. 99, 233505 (2011). 9. S. K. Behura, S. Nayak, I. Mukhopadhyay, et al., Carbon 67, 766 10. X. Li, H. Zhu, K. Wang, et al., Adv. Mater. 22, 2743 (2010). 11. C. Xie, J. Jie, B. Nie, et al., Appl. Phys. Lett. 100, 193103 (2012). 12. X. Zhang, C. Xie, J. Jie, et al., J. Mater. Chem. A 1, 6593 (2013). 13. L. Lancellotti, T. Polichetti, F. Ricciardella, et al., Thin Solid Films 522, 390 (2012). 14. T. Cui, R. Lv, Z. H. Huang, et al., J. Mater. Chem. A 1, 5736 (2013). 15. Z. Arefinia and A. Asgari, Journal of Renewable and Sustainable Energy 6, 043132 16. Silvaco-Altas User s Manual (2014), 163 165. 17. Y. F. Li, W. Yang, Z. Q. Tu, etal., Appl. Phys. Lett. 104, 043903 Received: 18 April 2015. Accepted: 24 April 2015. J. Nanoelectron. Optoelectron. 10, 1 5, 2015 5