Letter pubs.acs.org/NanoLett
Colloidal Antireflection Coating Improves Graphene−Silicon Solar Cells Enzheng Shi,† Hongbian Li,*,‡ Long Yang,‡ Luhui Zhang,† Zhen Li,§ Peixu Li,§ Yuanyuan Shang,† Shiting Wu,† Xinming Li,§ Jinquan Wei,§ Kunlin Wang,§ Hongwei Zhu,*,§ Dehai Wu,§ Ying Fang,‡ and Anyuan Cao*,† †
Department of Materials Science and Engineering, College of Engineering, Peking University, Beijing 100871, P. R. China National Center for Nanoscience and Technology, 11 Beiyitiao Street, Zhongguancun, Beijing, 100190, P. R. China § School of Materials Science and Engineering, Tsinghua University, Beijing 100084, P. R. China ‡
S Supporting Information *
ABSTRACT: Carbon nanotube-Si and graphene-Si solar cells have attracted much interest recently owing to their potential in simplifying manufacturing process and lowering cost compared to Si cells. Until now, the power conversion efficiency of graphene-Si cells remains under 10% and well below that of the nanotube-Si counterpart. Here, we involved a colloidal antireflection coating onto a monolayer graphene-Si solar cell and enhanced the cell efficiency to 14.5% under standard illumination (air mass 1.5, 100 mW/cm2) with a stable antireflection effect over long time. The antireflection treatment was realized by a simple spin-coating process, which significantly increased the short-circuit current density and the incident photon-to-electron conversion efficiency to about 90% across the visible range. Our results demonstrate a great promise in developing high-efficiency graphene-Si solar cells in parallel to the more extensively studied carbon nanotube-Si structures. KEYWORDS: Graphene, silicon, antireflection layer, TiO2 colloid, solar cell
R
oxide sheets have been used as transparent conductive electrodes for polymer-based and other types of solar cells.11−15 From the success of CNT-Si solar cells, one would expect that a graphene-Si model also works well. In particular, compared to cylindrical CNTs, the flat graphene sheet seems to be more compatible to be integrated with planar Si. In fact, solar cells made by transferring single to few-layer graphene to Si substrate have been reported with improved cell efficiencies by methods such as chemical doping on graphene.16−20 Although the up-to-date efficiencies of graphene-Si cells have reached 7.7−8.6%,17,20 this efficiency is still under 10% (a criterion for commercialization possibility) and also much lower than the CNT-Si cells (15%).8 In addition, the degradation (or stability) of graphene-Si cells has not been investigated previously. Whether the performance of grapheneSi cells could catch up and compete with the CNT-Si cells remains a question now. Here, we show that, by involving an antireflection TiO2 layer through a simple solution spin-coating process, the efficiency of graphene-Si (G-Si) solar cells can be enhanced to 14.5%, comparable to the best CNT-Si cells. The resulting TiO2-G-Si cell shows excellent device parameters including an open-circuit voltage (VOC) 0.612 V, a short-circuit current density (JSC) of
ecently, there is a growing interest in developing carbon nanotube (CNT)−silicon heterojunction solar cells, and the power conversion efficiency has been continuously improved to the range of 10−15% during the past several years.1−10 The devices are typically made by depositing a transparent single-walled CNT film on the surface of a n-type single-crystalline Si wafer to form CNT-Si junction and subsequent chemical doping on the CNT film and junction to optimize electronic property. Compared to traditional Si solar cells involving high-temperature dopant diffusion and additional metal grids as top contact, the fabrication of CNT-Si heterojunctions is a low-temperature process based on commercial wafers while still leading to high efficiency. More studies have revealed diffusion-dominated p−n junction transport mechanism.4,9 Chemical doping by various agents including oxidative acids, SOCl2 and H2O2, has been explored and proven to be effective to improve the solar cell efficiency.2,5,6,8−10 Very recently, our group showed that spincoating an antireflection colloidal layer on top of the CNT film could significantly boost the short-circuit current and result in enhanced cell efficiency of more than 15%.8 Graphene can be considered as a structure obtained by unrolling a CNT into a flat sheet. Its two-dimensional structure, atomic thickness, and high carrier mobility make graphene an ideal electrode material to be applied in a variety of thin film devices. Similar to CNTs, graphene layers synthesized by chemical vapor deposition (CVD) or reduced from graphene © 2013 American Chemical Society
Received: January 28, 2013 Revised: March 15, 2013 Published: March 21, 2013 1776
dx.doi.org/10.1021/nl400353f | Nano Lett. 2013, 13, 1776−1781
Nano Letters
Letter
Figure 1. Fabrication process of TiO2-G-Si solar cells. (a) Illustration of the spin-coating process in which a colloidal TiO2 was applied to a G-Si cell as antireflection coating. (b) Optical images of a TiO2-G-Si solar cell. The region enclosed by Ag paste (red dashed line) is defined as an active area. The enlarged optical microscope image shows the surface morphology of the TiO2 coating, with many cracks formed after colloidal drying.
32.7 mA/cm2, a fill factor (FF) of 72%, and an incident photon to electron conversion efficiency (IPCE) up to 90% across the visible region. We also studied key factors that influence the cell stability including the diode ideality factor, series resistance, and Schottky barrier height. We conclude that simple colloidal antireflection technique not only applies to porous CNT spiderweb-films, but also works well on two-dimensional smooth graphene. Compared to our previous work on CNTSi,8 our high efficiency graphene-Si solar cells represents another promising model for future development of low-cost, high performance nanomaterial-Si photovoltaics. Monolayer graphene was synthesized on copper foil substrates by the CVD method described in literature21,22 and then transferred to a n-type single-crystalline Si wafer covered with 400-nm-thick oxide and with a gallium−indium eutectic (E-GaIn) back contact (see Experimental Section in the Supporting Information for details). Typically the transferred graphene size was 1−2 cm, and rectangular to squareshaped regions with areas of several to 15 mm2 were defined by pasting mm-thick Ag paste on the SiO2, as illustrated in Figure 1a. The enclosed region was defined as the device area receiving incident photons (for calculating the current density later), while the Ag paste around could block light completely. To form the G-Si junction, HF vapor was evaporated to the enclosed area to etch the SiO2 in between, rendering a direct GSi contact. A colloidal solution containing small size TiO2 nanoparticles (3−5 nm, which make uniform solution and film) was spin-coated into the G-Si window to form a very thin coating on the entire graphene surface. The optical image showed that, after the coating of TiO2, the color of that area changed from light-gray to dark-blue, indicating strong inhibition of light reflection from the Si surface (Figure 1b). The TiO2 layer was uniform across the window and produced many microcracks after drying. The resulting solar cell can be illustrated as a sandwich structure consisting of a TiO2 layer on the top, a Si wafer at the bottom, and a monolayer graphene sheet in the middle (Figure 2a). Cross-sectional scanning electron microscopy (SEM) image reveals that the TiO2 coating has an average thickness of about 65 nm and good adhesion to the Si substrate (Figure
2a). From the top view, there are many narrow curled cracks with widths of
