Influence of air exposure duration and a-Si capping layer thickness on the performance of p-BaSi2/n-Si heterojunction solar cells

Fabrication of p-BaSi2(20nm)/n-Si heterojunction solar cells was performed with different a-Si capping layer thicknesses (da-Si) and varying air exposure durations (tair) prior to the formation of a 70-nm-thick indium-tin-oxide electrode. The conversion efficiencies (η) reached approximately 4.7% regardless of tair (varying from 12–150 h) for solar cells with da-Si = 5 nm. In contrast, η increased from 5.3 to 6.6% with increasing tair for those with da-Si = 2 nm, in contrast to our prediction. For this sample, the reverse saturation current density (J0) and diode ideality factor decreased with tair, resulting in the enhancement of η. The effects of the variation of da-Si (0.7, 2, 3, and 5 nm) upon the solar cell performance were examined while keeping tair = 150 h. The η reached a maximum of 9.0% when da-Si was 3 nm, wherein the open-circuit voltage and fill factor also reached a maximum. The series resistance, shunt resistance, and J0 exhibited a tendency to decrease as da-Si increased. These results dem...


I. INTRODUCTION
For future deployment of terawatt-scale solar cells, extensive research has been conducted on thin-film solar cell materials such as chalcopyrite and cadmium telluride as well as Si-based materials because of their high energy conversion efficiency (η) and low cost. [1][2][3][4][5][6] Perovskite-based solar cells have also gained increasing attention owing to their astonishing increase in efficiency. 7,8 However, these materials contain critical raw materials such as In, Cd, and Pb. There has also been growing interest in Si thin-film solar cells that employ an efficient light-trapping system, [9][10][11][12][13][14][15] but with this system it is not easy to achieve an η as high as 20%. Hence, it is necessary to explore alternative materials for thin-film solar cell applications. Among such materials, much attention has been given to the semiconductor BaSi 2 , which consists of safe and earth-abundant elements and has a band gap (E g ) of 1.3 eV, matching the solar spectrum. 16 One of the most striking features of this material is that both its large absorption coefficient (α) and large minority-carrier diffusion length (L) can be used. Because of its indirect band gap, undoped BaSi 2 can attain a minority-carrier lifetime (τ) and an L value as large as approximately 10 µs and 10 µm, respectively. [17][18][19] These values are sufficiently large for thin-film solar cell applications. Furthermore, α exceeds 3 ×10 4 cm −1 for photon energies greater than 1.5 eV because the direct transition occurs at energies slightly larger than E g . [20][21][22] For these reasons, we expect η to be larger than 25% in a 2-µm-thick BaSi 2  An a-Si capping layer plays an important role in BaSi 2 solar cells, whereby an undoped n-BaSi 2 surface with a few nm-thick a-Si layer can exhibit a τ = ∼10 µs with excellent repeatability. 24 Measurements of the valence band offset at the a-Si/BaSi 2 interface by hard x-ray photoelectron spectroscopy have shown that the barrier height of the a-Si layer for the minority carrier (i.e., holes) in the n-BaSi 2 is −0.2 eV, 25 whereas that of the native oxide layer is 3.9 eV. 26 Therefore, the a-Si capping layer works as a good electrical contact for hole transport as well as a passivation layer. Very recently, we have attained η = 9.0%, a short-circuit current density J SC = 31.9 mA/cm 2 and an open-circuit voltage V OC = 0.46 V for B-doped p-BaSi 2 /n-Si heterojunction solar cells using an a-Si capping layer. 27 The a-Si layer thickness (d a-Si ) and the duration of air exposure (t air ) after a-Si layer deposition are important parameters that may influence the performance of BaSi 2 solar cells. Previous studies have shown that t air and d a-Si have a significant effect upon the performance of a-Si:H(n)/a-Si:H(i)/c-Si(p) heterojunctions with intrinsic thin layer (HIT) solar cells. 28 This sensitivity arises from the significant influence the t air and i-layer thickness have upon the minority-carrier transport and recombination at the interfaces in HIT solar cells. 28 However, there is limited information regarding their effects upon BaSi 2 solar cells. In this work, we fabricated a-Si/p-BaSi 2 (20 nm)/n-Si heterojunction solar cells via molecular beam epitaxy (MBE) prepared with various values of t air and d a-Si , and attempted to clarify the influence of these parameters upon the properties of p-BaSi 2 /n-Si solar cells.

II. EXPERIMENTS
For the growth of the BaSi 2 layers, we used an ion-pumped MBE system (AVC Co., Ltd.) equipped with an electron-beam evaporation source for Si and with standard Knudsen cells for Ba and B. The deposition rates of Si (R Si ) and Ba (R Ba ) were controlled using an electron impact emission spectroscopy (EIES) feedback system (INFICON CO., Ltd). We first deposited Ba on a heated Czochralski n-Si(111) (resistivity ρ = 1-4 Ω·cm) substrate at 500 • C by reactive deposition epitaxy to form a 5-nm-thick BaSi 2 template layer, 29 where R Ba was set at 1.0 nm/min. This template layer acted as seed crystals for the subsequent BaSi 2 layer. We next co-deposited Ba, Si and B on the templates at 600 • C by MBE to form a 20-nm-thick B-doped p-BaSi 2 layer, [30][31][32] where R Si and R Ba were fixed at 0.9 and 2.3 nm/min, respectively. The B concentration was set to 2×10 18 cm −3 to be comparable to the hole concentration. 27 The epitaxial growth of the BaSi 2 layers in all of the samples was confirmed by reflection high-energy electron diffraction and x-ray diffraction (data not shown). We first prepared four samples (samples A-D) to examine the influence of t air upon the solar cell performance. For this purpose, we deposited a 2 or 5-nm-thick a-Si layer on the BaSi 2 surface in situ at 180 • C with R Si = 0.9 nm/min, followed by air exposure for t air = 12 or 150 h. In the previous study, 25 we found that the oxidation of BaSi 2 do not progress for sample capped with 5-nm-thick a-Si layers even for t air = 24 h. Therefore, we anticipated that the oxidation of BaSi 2 would be suppressed much further for t air = 12 h. On the other hand, we expected that the oxidation would progress for t air = 150 h. This is the reason why we chose these two air exposure durations. As described below, the η increased with t air for samples with d a-Si = 2 nm. We next prepared samples in which d a-Si was varied from 0.7 to 5 nm while keeping t air = 150 h to examine the influence of d a-Si . Please note that the a-Si layers in this study were not hydrogenated. They were just evaporated from the solid source of Si by electron beam irradiation. After keeping samples in air for t air , each sample was introduced into a radio-frequency (RF) sputtering chamber, and 1-mm-diameter and 70-nm-thick indium-tin-oxide (ITO) electrodes were sputtered on the front and Al electrodes on back surfaces at room temperature. The RF power was set to 100 W. The solar cell properties of samples A-D are summarized in Table I.
The plots of the current density versus voltage (J-V ) were measured for as many electrodes as possible in an area of 1 × 1 cm 2 on the sample wafer under AM1.5, 100 mW/cm 2 illumination at approximately 25 • C using a mask with holes 1 mm in diameter. To accurately obtain the series resistance, R S , diode ideality factor, γ, and reverse saturation current density, J 0 , of a diode, we adopted a technique described in Ref. 33  R S and γ can be given as Here, T is the absolute temperature, q is the elemental charge, k B is the Boltzmann constant, S is the area of the electrode, R SH is the shunt resistance, and J SC is the photocurrent density. Using the plot of dV/dJ versus the term in brackets in Eq. (1), we can directly deduce γ from the slope and R S from the intercept. The external quantum efficiency (EQE) spectra were evaluated at 25 • C using a lock-in technique with a xenon lamp and a 25-cm focal-length single monochromator (Bunko Keiki, SM-1700A).

III. RESULTS AND DISCUSSION
First we discuss the influence of t air on the solar cell performance. Figure 1(a) and 1(b) show typical examples of the J-V characteristics under AM1.5 illumination on samples with d a-Si = 2 nm (samples A and B, Fig. 1(a)) and 5 nm (samples C and D, Fig. 1(b)), for t air of 12 or 150 h. As shown in Fig. 1(a), V OC drastically increases with increasing t air for samples with d a-Si = 2 nm, and η was improved from 5.3 to 6.6%, as shown in Table I. Meanwhile, V OC decreases with increasing t air for samples with d a-Si = 5 nm, while J SC increases and η remains approximately constant at 4.7%. As shown in Table I, R S increases with t air regardless of d a-Si , meaning that part of the a-Si and/or BaSi 2 layer became oxidized during the exposure of the samples to air. In our previous study, we found that τ was improved when the oxygen composition in the region close to the BaSi 2 surface became large. 22 We thus speculate that the oxygen concentration became higher in the BaSi 2 region close to the a-Si/BaSi 2 interface for samples with d a-Si = 2 nm than for those with d a-Si = 5 nm when t air was increased. This increase in oxygen concentration may lead to a reduction of the surface recombination and thereby a decrease in J 0 . It was indeed found that J 0 decreased by approximately 1/50 in sample B (d a-Si = 2 nm) compared with sample A (d a-Si = 2 nm) after 150 h, as shown in Table I. In an ideal case, V OC is given by It is therefore reasonable that V OC becomes larger by increasing t air . Meanwhile, J SC increased but the fill factor (FF) and V OC decreased with increasing t air for samples at d a-Si = 5 nm because J 0 increased by more than 20 times and R S increased. The increase of J SC likely arises from the decrease of absorption in the a-Si layer owing to its partial oxidation, as will be discussed later.
Although the mechanism behind the large increase of J 0 at t air = 150 h is not clear at present, it is safe to state that η is improved with increasing t air when the a-Si layer thickness is small. We next discuss the influence of d a-Si upon the solar cell performance when t air = 150 h. Figure 2 shows typical examples of the J-V characteristics under AM1.5 illumination where d a-Si varies as 0.7, 2, 3, and 5 nm. It can be seen that the J-V curves significantly depend upon d a-Si and, in particular, the sample with d a-Si = 0.7 nm exhibits J-V characteristics similar to those obtained without an a-Si capping layer (results in Ref. 27). To understand this result, we compared the solar cell properties of p-BaSi 2 /n-Si heterojunction solar cells fabricated with varying d a-Si , as shown in Fig. 3. Some variation can be expected in the solar cell parameters, but it can be seen in Fig. 3 that there is a definite dependence of the solar cell parameters upon d a-Si . When the a-Si layer was thinnest (0.7 nm), the R S value was relatively large. This large R S value suggests that it is possible to oxidize BaSi 2 with a sufficiently thin a-Si layer, 25  for samples with d a-Si = 0.7 nm and those without a-Si capping layers. In addition, the η value reached a maximum of 9.0% when d a-Si was 3 nm and exhibited a decrease at d a-Si = 5 nm, which was a trend echoed by the V OC value. This is likely caused by a reduction of J SC for d a-Si = 5 nm because other parameters such as J 0 , R S , R SH and γ were seen to decrease with increasing d a-Si , and to almost saturate at d a-Si = 3 nm. To clarify this reason, we compared the external quantum efficiency (EQE) spectra for the samples with varying d a-Si . Figure 4 shows the EQE spectra for typical samples with d a-Si = 2 and 5 nm, where the broken line shows the ratio of EQE at d a-Si = 5 nm to that at d a-Si = 2 nm. The EQE spectra are seen to decrease as d a-Si increases, especially in the wavelength range shorter than ∼730 nm, which is equivalent to the bandgap energy of a-Si (1.7 eV). Therefore, it is reasonable to conclude that the reduction in J SC arises from the light absorption within the a-Si layer. This result is similar to that reported in HIT solar cells, 28 and we thus attribute the reduction of J SC to the same optical loss caused by the absorption in the a-Si layer. Therefore, a further increase in d a-Si may be found to reduce η.

IV. CONCLUSION
We fabricated a-Si/B-doped p-BaSi 2 (20 nm)/n-Si heterojunction solar cells with various t air and d a-Si values, and investigated the influence of these varying parameters upon the solar cell performance. Solar cell parameters such as η, V OC , J SC , J 0 , R S , γ, and R SH were found to depend upon t air and d a-Si . The η value increased from 5.3 to 6.6% as t air increased from 12 to 150 h for samples with d a-Si = 2 nm. When t air was fixed at 150 h, the η value reached a maximum of 9.0% at d a-Si = 3 nm. These results reveal that the precise control of BaSi 2 oxidation can enhance η much further.