Research Article www.acsami.org
Evolution of a Native Oxide Layer at the a‑Si:H/c-Si Interface and Its Influence on a Silicon Heterojunction Solar Cell Wenzhu Liu,†,‡ Fanying Meng,† Xiaoyu Zhang,† and Zhengxin Liu*,† †
Research Center for New Energy Technology, Shanghai Institute of Microsystem and Information Technology (SIMIT), Chinese Academy of Sciences (CAS), Jiading, 201800 Shanghai, People’s Republic of China ‡ University of Chinese Academy of Sciences (UCAS), Shijingshan, 100049 Beijing, People’s Republic of China S Supporting Information *
ABSTRACT: The interface microstructure of a silicon heterojunction (SHJ) solar cell was investigated. We found an ultrathin native oxide layer (NOL) with a thickness of several angstroms was formed on the crystalline silicon (c-Si) surface in a very short time (∼30 s) after being etched by HF solution. Although the NOL had a loose structure with defects that are detrimental for surface passivation, it acted as a barrier to restrain the epitaxial growth of hydrogenated amorphous silicon (a-Si:H) during the plasma-enhanced chemical vapor deposition (PECVD). The microstructure change of the NOL during the PECVD deposition of a-Si:H layers with different conditions and under different H2 plasma treatments were systemically investigated in detail. When a brief H2 plasma was applied to treat the a-Si:H layer after the PECVD deposition, interstitial oxygen and small-size SiO2 precipitates were transformed to hydrogenated amorphous silicon suboxide alloy (a-SiOx:H, x ∼ 1.5). In the meantime, the interface defect density was reduced by about 50%, and the parameters of the SHJ solar cell were improved due to the post H2 plasma treatment. KEYWORDS: native oxide layer, epitaxial growth barrier, H2 plasma, microstructural evolution, surface passivation
■
INTRODUCTION A silicon heterojunction (SHJ) solar cell with intrinsic hydrogenated amorphous silicon (a-Si:H) passivation layers is intensively investigated because of the high conversion efficiency (Eff) benefiting from its combined structure with monocrystalline silicon (c-Si) and ultrathin a-Si:H layers.1 For such solar cells, a low defect density at the a-Si:H/c-Si interface (Dit) is particularly crucial to achieve a high open-circuit voltage (Voc).2,3 Besides, according to calculations with Automat for the simulation of heterostructures software (AFORS-HET), the fill factor (FF) of an SHJ solar cell is also sensitive to the variation of Dit, expecially on the condition that Dit is higher than 1.0 × 1012 cm−2·eV−1.4−7 At present, in order to passivate the c-Si surface with a high density of dangling bonds, an intrinsic aSi:H layer with a typical thickness of 3−5 nm is the most common material deposited onto both its surfaces using plasma-enhanced chemical vapor deposition (PECVD) or hotwire chemical vapor deposition (HWCVD). For the a-Si:H/c-Si heterojunction design, performances of the SHJ solar cell are extremely sensitive to some recombination mechanisms; among them, the bulk recombination within the epitaxial emitter and the interface recombination at the a-Si:H/c-Si interface are the most significant mechanisms responsible for the Eff loss.8 Miniscule amounts of bonded hydrogen within the intrinsic aSi:H layers due to the epitaxial growth account for the high bulk defect density.9 However, the defects within the a-Si:H/cSi interface are much more difficult to investigate and far away © XXXX American Chemical Society
from clarifying the microstructural origins. Wolf and co-workers claimed there are likely nanoscale hydrogen platelets at the aSi:H/c-Si interface which affect the light induced degradation,10 and they also demonstrated the dominant defects at the a-Si:H/ c-Si interface are silicon dangling bonds.11 On the other hand, H2 plasma treatments have recently been applied to SHJ solar cell fabrication12,13 and achieved higher Eff. However, most of these studies focused on optimizing parameters of the H2 plasma,14−16 while little attention was paid to reveal the physical origins that enhance the electric performances. In this work, we studied the microstructure at the surface of the c-Si wafer and the a-Si:H/c-Si interface and the relevant changes caused by H2 plasma treatments. Based on the structural evolution, the physical mechanism responsible for Eff enhancement of an SHJ solar cell is preliminarily delineated.
■
EXPERIMENTAL SECTION
Float-zone (FZ) (100)-oriented Si wafers with a one-sided mirror surface (thickness, 525 ± 25 μm; resistivity, >3000 Ω·cm) were used as substrates for the measurements of spectroscopic ellipsometry (SE; J. A. Woollam, M-2000XI) and Fourier transform infrared spectroscopy (FTIR; PerkinElmer, Spectrum 100). We first studied the surface microstructure of the c-Si wafer under exposure to air in a clean room. Received: August 19, 2015 Accepted: November 13, 2015
A
DOI: 10.1021/acsami.5b07709 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Table 1. Treatments of Specimens [Pre H2 Plasma Treatment (Pre-H), a-Si:H Deposition (a-Si:H), and Post H2 Plasma Treatment (Post-H)] and Measurements of Specimens [Spectroscopic Ellipsometry (SE), Fourier Transform Infrared Spectroscopy (FTIR), X-ray Photoelectron Spectroscopy (XPS), High-Resolution Electron Transmission Microscopy (TEM), Sinton WCT-120 (WCT-120), and Solar Simulator (Simulator)]
etching manner, where the etching time between two measurements was set to be 10 s. In addition, solar-grade Czochralski (CZ) Si wafers (thickness, 180 μm; resistivity, 1−6 Ω·cm) were used to fabricate SHJ solar cells. The CZ wafers were textured with alkali solution and were successively cleaned by the standard RCA process and etched with 2% HF solution for SHJ solar cell fabrication. They were halved into two groups and denoted by Cell-A and Cell-B, respectively. Intrinsic a-Si:H layers of Cell-A were prepared without H2 plasma treatments (the same as that of Hetero-ref), while those of Cell-B were prepared with post H2 plasma treatments (the same as that of Hetero-H2-Post). After deposition of the a-Si:H layers (both intrinsic and doped), Cell-A and Cell-B were covered with transparent conductive oxide (TCO) thin films on both of their sides by reactive plasma deposition (RPD) technique. The TCO films used in the present work are tungstendoped indium oxide (IWO) layers. Then the samples were forwarded to a screen printing process to form silver electrodes with conventional finger and busbar patterns, followed by an annealing at a temperature about 250 °C. Afterward, the electric parameters of the SHJ solar cells were measured with a solar simulator under the standard conditions (AM 1.5G, 25 °C). Finally, to evaluate the electrical influence of the post H2 plasma on the a-Si:H/c-Si interface, two symmetric heterostructures composed of P-doped a-Si:H layer/intrinsic a-Si:H passivation layer/c-Si wafer/intrinsic a-Si:H passivation layer/P-doped a-Si:H layer were fabricated. They are referred to as Symmetric A and Symmetric B, whose intrinsic and P-doped layers are the same as those of Cell-A and Cell-B, respectively. Effective minority carrier lifetime (τeff) of both Symmetric A and Symmetric B were then measured by Sinton WCT-120 after they cooled to room temperature. All of the aforementioned specimens are listed in Table 1.
The c-Si wafer, referred to as Bare, was etched with 2% HF solution for 2 min to remove its native oxide layers (NOLs), and then it was directly forwarded to the SE and FTIR measurements. The SE measurements were continuously taken at a time sequence of ∼0, 30, 100, and 800 min in the clean atmosphere to clarify the change of surface microstructure. The (ψ, Δ) curves at the wavelength region from 246 to 1689 nm were fitted with a simple optical model (NOL/cSi substrate) to obtain the thickness of the NOL. Another FZ Si wafer was measured by FTIR using the background mode in N2 atmosphere right after it was etched with the 2% HF solution. After Bare was exposed to the air in a clean room for 800 min to form an ultrathin fresh NOL on the surface, it was measured by FTIR using the sample mode in N2 atmosphere. The FTIR spectrum was obtained by subtracting the background mode from the sample mode. Five more FZ Si wafers were deposited with a-Si:H layers on the mirror surfaces by PECVD after being etched with 2% HF solution. Before the deposition of the a-Si:H films, they were exposed to a clean atmosphere for more than 300 min so that they were covered with fresh ultrathin NOL. The parameters of PECVD were tentatively changed at a wide region to obtain different a-Si:H layers with different microstructures. The first one was deposited with a low power density and low [H2]/[SiH4] gas flow-rate ratio of less than 20 mW/cm2 and 20, respectively. On these conditions, the etching effect of plasma on the Si substrate was weak. This sample was referred to as LPLR. On the contrary, the second sample was deposited with a high power density and high [H2]/[SiH4] gas flow-rate ratio of 170 mW/cm2 and 75, respectively. In this case, the etching effect of plasma on the Si substrate was strong. The sample was referred to as HPHR. The other three FZ Si wafers were deposited with the same a-Si:H layers but combined with different H2 plasma treatments, which were termed as Hetero-ref, Hetero-H2-Pre, and Hetero-H2-Post. Hetero-ref was prepared without H2 plasma treatment, whereas, Hetero-H2-Pre and Hetero-H2-Post were treated by a brief H2 plasma with a power density of 25 mW/cm2 and a pressure of 1.50 Torr before and after the a-Si:H depositions, respectively. The source gas for the a-Si:H film deposition is pure SiH4 with a flow rate of 100 cm3(STP)·min−1, where the substrate temperature, power density, and total gas pressure were fixed at 200 °C, 25 mW/cm2, and 0.53 Torr, respectively. Before turning on the RF power, it took 5 min to stabilize the chamber ambient. FTIR spectra of LPLR, HPHR, Hetero-ref, and Hetero-H2Post were measured in N2 atmosphere with the same background wafer. Microstructures of Hetero-ref and Hetero-H2-Pre at the a-Si:H/ c-Si interfaces were evaluated by high-resolution electron transmission microscopy (HR-TEM; FEI Tecnai TF-20), which was operated at a voltage of 200 kV. Finally, the a-Si:H/c-Si interfaces of Hetero-ref and Hetero-H2-Post were characterized by X-ray photoelectron spectroscopy (XPS; ESCALAB 250Xi, TMO) with Al Kα radiation in the
■
RESULTS AND DISCUSSION Analysis of the NOL. The surface of the c-Si wafer is usually terminated by hydrogen atoms after being cleaned by HF solution.17 When the Si is exposed to air, the weak silicon− hydrogen bonds are gradually replaced by oxygen−silicon bonds and thus an ultrathin NOL is formed as time passes by. Figure 1a shows the FTIR transmission spectrum of Bare after exposure in the air for ∼800 min and coverage with an ultrathin NOL. The absorption peak at 890 cm−1 is well fitted by a broad Gaussian function with a half-width (Γ) of 79 cm−1. This peak can be assigned as SiH2 bending mode. Another two peaks at 1095 and 1232 cm−1 are also observed; they are originated from interstitial oxygen and small-size (
