MoO X and WO X based hole-selective contacts for wafer-based Si solar cells

— Highly-transparent carrier-selective front contacts open a pathway towards entirely dopant free Si solar cells. Hole-selective a-Si:H/MoO x /ITO front contact stacks were already successfully applied in such novel devices. However, for optimum device performance, further improvements are required: We evaluate the use of the high-work-function material WO X as a replacement for MoO x in an attempt to reduce optical absorption losses. In addition, we investigate the use of thin hydrogenated SiO X instead of a-Si:H, and the impact of the residual pressure for MoO x evaporation.


I. INTRODUCTION
The development of dopant-free Si solar cells in which carrier extraction is provided by electron and hole transport layers (ETL and HTL), integrated into passivating contacts, is a promising route to reach efficiencies close to 27% [1]. A typical carrier-selective contact employs a broadband optically transparent material which provides both a chemical passivation of Si surface states and a high (low) work function, inducing in dark and at equilibrium an electrical potential at the Si wafer surface, yielding hole (electron) collection.
A fully doping-free cell with efficiency of 19.4% was realized in ref. [2] by employing a transparent MoOX-based hole-selective contact on the front side and a LiF-based electron-selective stack on the rear side of an n-type Si wafer. As MoOX and LiF themselves do not passivate electronic defects at the c-Si surface, thin ~5 nm thick intrinsic hydrogen-rich amorphous silicon (i)a-Si:H interlayers have to be inserted to achieve a high VOC. Our present research aims to further optimize the hole-selective front contact, as already used in ref. [2] and [3]. The motivation for this is that the proposed (i)a-Si:H/MoOX/ITO front contact stack comes with an important drawback: parasitic light absorption [3] in both the MoOX, mainly due to transparent conductive oxide (TCO) sputter-induced damage [4], and the ~5 nm thick (i)a-Si:H layer. Furthermore, MoOX based contact stacks suffer from a fill factor degradation upon annealing at 190°C [5], required for curing of screen printed contacts. This effect could recently be attributed to the release of hydrogen from adjacent layers [6]. In our experiments, we evaluated the possible replacement of MoOX by WOX layer. WOX provides a high work function similar to MoOX [7,8] but has a higher transparency both before and after TCO sputter-deposition [4]. Furthermore we tested the impact of the residual base pressure in the evaporation chamber on the performance of solar cells with MoOX-based front contacts and investigated the use of thermal silicon oxide as an alternative passivation layer with higher transparency and improved thermal resilience.

II. EXPERIMENTAL
Float-zone (100) Si wafers with a thickness of 240 µm and resistance of 3 Ωcm (n-type) were textured and cleaned. For solar cell (Fig. 1a) fabrication, 5-nm-thick (i)a-Si:H layers were applied on the front and rear sides by plasma enhanced chemical vapor deposition (PECVD). On the rear side an about 10 nm thick phosphorous doped a-Si:H layer was deposited in the same tool. After loading in an evaporation chamber, an either ~7-nm-thick MoOX (x<3) or 5-nm-thick WOX (x< 3) layer was deposited on the wafer front side by thermal evaporation using stoichiometric powder (MoO3 or WO3). Then, 2-cm 2 -sized pads of ITO or IZO/ITO were sputter-deposited on the front side and a full-area ITO/Ag rear contact stack was applied. Front metal grids were deposited by Ag screen-printing and cured 20 minutes at 190 °C.   Figure 3 shows the JV-curve of our so far best cell with WOx-based hole-selective contact and its comparison to a cell with an unoptimized MoOx-based contact fabricated in the same evaporation chamber at low base pressures (10 -6 mbar). In both cases, sputtered IZO [9] was used as the front TCO layer, deposited in the same chamber immediately after the MoOX or WOX thermal evaporation without breaking the vacuum. The cells were characterized after annealing the screen-printed Ag front contact at 190°C. Even though the WOX based cell achieves a higher short-circuit current density JSC, its VOC equals only 638 mV (Table 1) compared to 705 mV for the MoOX based cell. Both cells suffer from an S-shaped JV-curve close to the VOC, leading to fill factors of only 70% (WOX) and 73% (MoOX). Increasing the WOX thickness leads to an even stronger S-shape and further reduction in FF (data not shown here). Most likely, the observed low VOC and efficiencies of cells with WOX based front contact are due to an insufficient band bending provided by our thermally evaporated WOX material, as discussed in ref. [10].   Figure 2 compares JV curves of cells with MoOX-based hole-selective front contacts in which the MoOX layers were evaporated in vacuum chambers pumped down to different base pressures. We specifically investigated the influence of the water partial pressure by comparing results using a tool equipped with a glovebox and a transfer chamber as a waterfree deposition tool (labelled N2), and a tool opened to air before pumping down (thus with most residual pressure being water vapor, labelled as H2O). Pumping times were adjusted in both tools to reach a base pressure around 10 -5 mbar (labelled high p) or 10 -6 mbar (labelled low p). ITO was then sputterdeposited on all samples, a silver grid was screen printed, and samples were cured at a low temperature of 130 °C. All JV curves show very similar VOC and JSC. A slight S-shape is observed in all cases, yet much more pronounced for the sample prepared with a high base pressure in the tool vented to atmosphere (high p H2O). This suggests that residual water during evaporation can impact negatively the performance of MoOX-based devices. Fig. 3. JV curves of SHJ cells with (i)a-Si:H/MoOX/ITO holeselective front contact. The MoOX layers were deposited after evacuating the chamber to a high (~10 -5 mbar) or low (10 -6 mbar) base pressure. MoOx layers evaporated in the chamber installed in a N2 glovebox lead to superior cell performance compared to those deposited in the chamber which was vented in air before each deposition (labelled "H2O"). Figure 4 compares the minority carrier lifetimes of symmetrical test structures with different buffer layers underneath the thermally evaporated MoOx. The investigated silicon oxide layers were generated by two different recipes leading to oxide thicknesses of 2.0 nm and 2.4 nm, as summarized in Table 2.

IV. ALTERNATIVE BUFFER LAYERS FOR MOOX BASED HOLE-SELECTIVE CONTACTS
The thin thermal oxide layers result in higher lifetimes than achieved with a reference sample that was HF-dipped prior to the MoOx deposition (no buffer). However their lifetimes of 100 µs to 170 µs at an injection level of 1.10 15 cm -3 are one order of magnitude lower than achieved with an (i)a-Si:H buffer layer (~6 ms). Our data indicate that the surface passivation increases with increasing oxide thickness and hydrogenation. Hydrogenation of the oxide layer 2 (DSP wafer) prior to MoOx deposition resulted in a higher lifetime (Fig. 4) which translates into an increase in implied VOC (iVOC) from 603 mV to 621 mV. The same effect was observed on textured (TXT) wafers which achieved an iVOC up to 650 mV (hydrogenated oxide 2, Fig. 4).
Further analysis is necessary to investigate the carrier transport in SiOX/MoOX/TCO contact stacks and to achieve higher passivation levels comparable to (i)a-Si:H. We expect that there will be a tradeoff between passivation and series resistance for the oxide layer thickness when applied in a real solar cell device.  (Fig 1b) with different buffer layers as described in the text.

V. CONCLUSION AND OUTLOOK
Our experimental results show that both MoOX and WOX can be used in hole-selective contacts. However, so far WOXbased hole-selective contacts resulted in lower VOC. The performance of a-Si:H/MoOx/TCO hole-selective contact stacks depends critically on the base pressure and residual water vapor in the deposition chamber. Symmetrical test structures with stacks of 2.4 nm thick hydrogenated SiOx layers and MoOx resulted in iVOC of 650 mV. Further investigation is required to evaluate its applicability in a real device and its optimal properties for best performance. holds a Marie Skłodowska-Curie Individual Fellowship from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No: 706744, action acronym: COLIBRI). Part of this work was supported by the European Union's Horizon 2020 Programme for research, technological development and demonstration Grant Agreements no. 727529 (project DISC).