Plated gold concentration from e-waste by machine-vision sorting and cyanide-free chemical recovery

Recovering gold from electronic waste plays a primary role in making recycling economically attractive. Shredded PCB fragments were sorted as “gold-positive” or “gold-negative” depending on the presence or absence of gold using a colour-based image analysis in the HSV (hue–saturation–value) colour space. Optimising parameters on three training batches yielded 81% recall and 86% accuracy, reducing the mass of material subjected to chemical treatment by up to a factor of five. When applied to a held-out batch with a different appearance, accuracy decreased to 59%, while recall was 100%. Consequently, no gold was lost and the usage of chemical treatment was still reduced by 45% on the held-out batch. The gold-rich fraction was subsequently treated in a single dissolution–filtration step using an acidic ferric sulfate solution (13 g ·L⁻¹ of Fe 3+ , pH ≈ 1.2, 35 °C), enabling selective dissolution of the underlying copper and nickel and recovery of gold flakes by filtration.