Defect detection in silicon solar cells using modulated electroluminescence imaging and Eulerian video magnification

Electroluminescence (EL) imaging of crystalline silicon solar cells typically relies on expensive InGaAs cameras or specialized cooled silicon CCD cameras optimized for near-infrared detection, which limits its accessibility in cost-constrained laboratories and photovoltaic (PV) test facilities. While conventional lock-in techniques overcome noise limitations through phase-sensitive detection, they require complex hardware synchronization. Here, we demonstrate that EL detection can be achieved using a standard CMOS camera by applying a sinusoidally modulated electrical current and Eulerian Video Magnification (EVM). We capture short video sequences at 60 fps by modulating the cell current from zero to maximum values ( I max ) of 0.5–4.0 A with frequencies of 0.1–2.0 Hz. Our EVM pipeline employs multiscale spatial decomposition, narrow-band temporal filtering centered on the modulation frequency, and a current-dependent amplification factor, followed by percentile-based averaging of the brightest frames. This EL–EVM technique enables the detection of subtle surface wear, mechanical cracks, and heavily damaged regions in monocrystalline cells. Quantitative comparison with reference InGaAs EL images and LBIC maps reveals that EL–EVM provides enhanced defect visibility, at the cost of reduced spatial resolution, while maintaining computational efficiency ( ∼ 1 min per 3600-frame video on standard hardware). This approach thus enables full-field imaging of modulated EL with readily available CMOS cameras and achieves lock-in-inspired contrast enhancement through purely digital temporal filtering, eliminating the need for phase-sensitive lock-in electronics. This methodology therefore provides an affordable and effective screening tool for PV cell diagnostics across research, educational, and industrial environments. • Electroluminescence detection in solar cells enabled by Eulerian Video Magnification. • Temporal filtering reveals electroluminescence signals invisible in raw video. • Reconstructed images reproduce defect patterns observed in InGaAs imaging. • Radiometric fidelity preserved, enabling quantitative defect analysis. • Accessible PV screening tool with potential for broader diagnostic applications.