Skip to main content

High efficiency GaInP/GaAs Tandem wafer bonded solar cell on silicon

Periodic Reporting for period 1 - HISTORIC (High efficiency GaInP/GaAs Tandem wafer bonded solar cell on silicon)

Reporting period: 2015-06-01 to 2017-05-31

The direct conversion of sunlight into electricity is a promising clean energy solution to answer the multiple challenges of energy supply security, competitiveness, electricity prices and climate change. Crystalline silicon, the dominant technology on the photovoltaic market, benefits from a well-established industry and record sunlight to electricity power conversion efficiency around 26%. However, mainly limited by its inherent material properties, the silicon technology has very little room for further power conversion efficiency improvement. On the other hand, much higher solar cell efficiencies have been achieved by stacking several layers of different materials, namely III-V semiconductor crystals, to create a so-called multi-junction solar cell which utilized a larger fraction of the solar energyspectrum. At present, however the III-V solar cells are restricted to specific markets (e.g. space) for cost and availability reasons. A hybrid solution combining the advantages of III-V multi-junction solar cells with the benefits of silicon, the most wide-spread photovoltaic material, offers great opportunities. Indeed, power conversion efficiencies above 46% (under 1-sun AM1.5G conditions) can be theoretically expected (see fig. 1) for conventional silicon based solar cell when combined with two additional III-V active layers to form a triple-junction solar cell (Gallium Indium Phosphide - GaInP - and Gallium Arsenide - GaAs - on top of silicon).

In the practice, however, there are still hurdles to be overcome. Combining III-V semiconductor materials with silicon is highly challenging. III-V materials have some fundamental crystallographic differences with silicon (lattice and thermal mismatch, polarity difference), and thus it is challenging to grow III-V layers with sufficient electrical quality directly on silicon. The innovative approach proposed in this project bypasses this issue and enables to have high quality III-V crystals combined with silicon, using a microelectronic technique called surface activated wafer bonding. With this approach, the III-V solar cells and the wafer based silicon solar cell are prepared separately, and combined using the surface activated wafer bonding technique. Once permanently joined, the III-V substrate is removed, resulting in 3-5 μm thin III-V solar cells on top of the silicon wafer (See fig. 2). The objective of this research project was to use the surface activated wafer bonding technique to produce such hybrid III-V/silicon triple-junction solar cells (GaInP/GaAs//Si), and achieve large area (4 cm2) devices with an efficiency exceeding 30% in 2-terminal configuration (under 1-sun AM1.5G conditions). These solar cells may serve as a drop in replacement into standard flat plate photovoltaic modules which facilitates market introduction.
To achieve the goal of this project, the work was structured around 4 main tasks: i) Optical modelling ii) Solar cell processing iii) Bonding interface investigation and iv) Triple-junction device characterization. The key results obtained in the framework of this project are listed below:

- i) An optical model for the triple-junction GaInP/GaAs//Si solar cell, including about 30 layers, has been developed (anti-reflection coatings, p-n junctions, tunnel junctions, etc.) using the transfer matrix method. This optical model was crucial to identify experimental losses and design III-V top cells suitable for current matching with silicon.

- ii) A processing route compatible with both III-V and silicon material specifications has been established. Among the decisive steps for high processing quality, we can list: - A minimum number of handling steps while meeting the requirements of high performance silicon cell manufacturing (ion implantation) - The implementation of a stable (AlOx/SiNx) silicon back side passivation after bonding - Suitable front grid design ensuring minimum area metal coverage and resistance loss.

- iii) Investigations of GaAs/Si bond interface revealed: - The link between surface particle contamination (optical scan) and voids after bonding (acoustic scans) - The importance of limiting the sample handling steps and working in a controlled environment to minimize surface particle contamination. Excellent bond yield (4 inch, quasi void-free) has been achieved with bond resistance below 5 mΩ.cm2

- iv) Based on the above-listed findings, wafer bonded GaInP/GaAs//Si triple-junction cells were processed and their performances assessed (EQE, IV-characteristics, reflection and electroluminescence). As a result of iterative improvement with simulation and experience, a 2-terminal record 4cm2 solar cell with 1-sun power conversion efficiency of 30.2% was measured and certified in Fraunhofer ISE Calibration Laboratory (see fig. 3). Toward the end of the project, pursuing the same approach, the device efficiency was pushed even further: a certified value of 31.3% was achieved.

Those results were shared with the scientific community in oral presentations and posters during international conferences, in peer reviewed scientific articles (e.g. DOI: 10.1109/JPHOTOV.2016.2629840) as well as to the general public with press releases and special highlight article in PHOTON International – The Solar Power Magazine. Further details can be found on the project webpage ( which summarizes the objectives and main results.
At the beginning of the project, the state of the art for silicon based multi-junction solar cell efficiency under 1-sun conditions was around 25%. We have clearly shown that those III-V//Si tandem cell devices can go beyond 30% power conversion efficiency, and even be pushed beyond 31%. The results of this research project have demonstrated the high efficiency potential of the approach, and attracted attention from the scientific community as well as from industry. However, the development and commercialization of an industrial product will require further work to mature processes and bring the cost down. This will be the topic of future research projects with both academic and industrial partners.
Fig2 Process flow: fabrication of GaInP/GaAs//Si triple-junction solar cells by wafer bonding
Fig1 Theoretical fraction of AM1.5G spectrum which can be converted by a GaInP/GaAs/Si solar cell
Fig3 Wafer-bonded III-V / Si multi-junction solar cell device with an area of 4 cm^2