AbstractHot charge carriers from plasmonic nanomaterials currently receive increased attention because of their promising potential in important applications such as solar water splitting. While a number of important contributions were made on plasmonic charge carrier generation and their transfer into the metal’s surrounding in the last decades, the local origin of those carriers is still unclear. With our study employing a nanoscaled bicontinuous network of nanoporous gold, we take a comprehensive look at both subtopics in one approach and give unprecedented insights into the physical mechanisms controlling the broadband optical absorption and the generation and injection of hot electrons into an adjacent electrolyte where they enhance electrocatalytic hydrogen evolution. This absorption behavior is very different from the well-known localized surface plasmon resonance effects observed in metallic nanoparticles. For small ligament sizes, the plasmon decay in our network is strongly enhanced via surface collisions of electrons. These surface collisions are responsible for the energy transfer to the carriers and thus the creation of hot electrons from a broad spectrum of photon energies. As we reduce the gold ligament sizes below 30 nm, we demonstrate an occurring transition from absorption that is purely exciting 5d-electrons from deep below the Fermi level to an absorption which significantly excites “free” 6sp-electrons to be emitted. We differentiate these processes via assessing the internal quantum efficiency of the gold network photoelectrode as a function of the feature size providing a size-dependent understanding of the hot electron generation and injection processes in nanoscale plasmonic systems. We demonstrate that the surface effect, compared with the volume effect, becomes dominant and leads to significantly improved efficiencies. The most important fact to recognize is that in the surface photoeffect presented here, absorption and electron transfer are both part of the same quantum mechanical event.