Tandem Photovoltaic Architectures: Terminal Configurations, Loss Management, and the Path from Certified Cells to Deployable Modules

Abstract Tandem photovoltaics is the only experimentally demonstrated route past the 33.7% single-junction detailed-balance limit at one sun. This topical review examines the tandem device itself, rather than individual absorber chemistries, and covers 2015–2026. Its methodology is a structured synthesis of certified results listed in the NREL chart and the Solar Cell Efficiency Tables (Version 68), cross-checked against the primary peer-reviewed literature, with every benchmark reported together with area, illumination condition, certifying body and year. The review first develops the device physics that distinguishes a tandem from two single junctions: bandgap pairing; two-, three- and four-terminal configurations, including how three terminal devices are measured; current matching at the operating point and its spectral sensitivity; tunnel-junction and recombination-layer design; optical management; photon recycling; and the capabilities and limits of the one- to three-dimensional simulation tools used to design tandems. It then surveys the perovskite–silicon, all-perovskite, perovskite–CIGS, perovskite–organic and III–V//Si pathways, and decomposes the certified efficiency gains of the two leading pathways into open-circuit voltage, current and fill-factor contributions. Three findings stand out. First, the 2T perovskite–silicon tandem has reached 35.5% (ESTI-certified), with gains since 2020 coming mainly from fill factor and current rather than voltage. Second, a well-designed 2T device loses only about 2% of its annual energy to spectral mismatch. Third, the decisive barriers are now the cell-to-module gap (35.2% on1 cm 2 versus 29.4% on a 1.7 m 2 module) and an operational-stability record still one to two orders of magnitude short of a 25-year life. The review closes with specific, quantitatively motivated research targets for interconnects, three-terminal devices, validated modelling and stability testing.