Towards a Monolithic, Substrate-Reusable and an All-Epitaxial Design for III-V-on-Si Solar Cells Nikhil Jain, Michael Clavel, Patrick Goley, and Mantu Hudait Virginia Tech, Blacksburg, VA 24060 Abstract - Integration of III-V multi j unction solar cells on Si substrate can address the future levelized cost of energy by unifying the high-efciency merits of III-V materials with the low-cost and abundance of Si. A Si-compatible monolithically integrated 3J InGaP/GaAs/Ge-Si cell design with a hybrid Ge-Si bottom cell is investigated. Utilizing a combination of comprehensive modeling and experimental material characterization techniques, we present our results for ultra thin epitaxial Ge directly grown on Si substrate using molecular beam epitaxy. Virtual "Ge-on-Si" substrates could provide a large-area, low-cost alternative to expensive Ga A s wafers, a promising step towards realizing monolithic, high efciency and low-cost III-V-on-Si photovoltaics. Inde Terms - III-V-on-Si, Ge-on-Si, heteroepitaxy, photovoltaic cells, solar cell design. I. INTRODUCTION While the efciency of mainstream Si based solar cells has almost saturated at �25%, III-V multijunction solar cells have steadily shown perforance improvement, reaching a recent record efciency of 46%. Integration of such III-V multijunction cells with Si can address the fture levelized cost of energy by unifing the high-effciency merits of III-V materials with low-cost and abundance of Si. To date, effciency of 3J III-V/Si tandem solar cells have merely exceeded 25% [1] even afer employing non-monolithic techniques such as wafer-bonding [2] and areal current matching [3]. Challenges associated with material growth, reliability and reproducibility have limited the success of III V-on-Si technology. Thus, novel approaches are sought for realizing the potential of III-V-on-Si multijunction solar cells. A very promising path for monolithic integration of III-V solar cells on Si would be to utilize a thin intermediate Ge buffer layer. Successfl demonstration of virtual "Ge-on-Si" template could signifcantly reduce the cost per watt attributed to the large area and low cost of Si substrate. Interestingly, utilizing Ge intermediate layer de-couples two critical challenges for GaAs-on-Si growth: (i) polar on non-polar epitaxy and (ii) lattice-mismatch growth. Owing to a small band gap, the Ge layer absorbs a wide spectrum of the incident sunlight beyond the GaAs absorption edge, and therefore the hybrid Ge-Si subcell does not limit the current in 3J InGaP/GaAs/Ge-Si solar cells. The Ge intermediate layer approach for III-V-on-Si integration could be utilized (i) to create virtual "Ge-on-Si" template for subsequent GaAs growth (could potentially involve active Ge subcells), (ii) 978-1-4799-7944-8/15/$31.00 ©2015 IEEE solely as a buffer layer for connecting III-V cells to an active Si bottom subcell, and (iii) as the emitter layer for bottom Si base, forming a hybrid Ge-Si subcell. Utilizing a Si homojunction cell beneath the Ge buffer layer would likely require a difsion process and a thicker Si substrate for current-matching in comparison to the hybrid Ge Si approach. Furthermore, to allow sufcient light penetration to active Si subcell, extremely thin Ge buffer would be essential, rendering the subsequent GaAs growth very challenging. Prior reports on Ge integration on Si substrate for photovoltaics typically employed graded SixGel-x buffers, which were several microns thick [4-6], elevating the thermal mismatch and cost issues. While other approaches utilized a pattered selective area epitaxy [7, 8], typically involving an additional pattering and/or chemical polishing step. Here, we focus on developing a non-selective area Ge-on-Si epitaxial process with a key goal of realizing thin epitaxial Ge layers allowing light penetration to the bottom Si substrate cell. Thus, in this work we focus on the design, modeling and epitaxial growth for hybrid Ge-Si bottom subcell, wherein the epitaxial Ge layer serves as a uniforly doped emitter for bottom Si subcell, thus forming a hybrid Ge-Si subcell. This approach precludes the need for difsed Si junction, allowing an in-situ and an all-epitaxial process for subsequent III-V growth requiring very thin Si « 60Ilm). Such 3J cells with very thin Si would also be very promising for CPV applications and could frther beneft fom additional cost InGaP I (1.86eV) Hybrid I Ge-Si •• Fig. 1 Schematic depiction of tandem 3J InGaP/GaAs/Ge-Si solar cell utilizing a hybrid Ge-Si bottom subcell.