<p>This paper investigates spectrum sensing in cognitive radio networks assisted by Reconfigurable Intelligent Surfaces (RIS), where the Primary User (PU) operates under severe energy constraints and is powered by thermal energy harvesting. The PU harvests energy from a temperature gradient, enabling sporadic transmission of its signal. A RIS is employed to enhance signal propagation from the PU to the Secondary User (SU), which performs spectrum sensing using an energy detector. We model the energy harvesting dynamics of the PU as a function of the ambient temperature difference and analyze the impact of RIS parameters, including the number of reflecting elements and phase alignment, on the detection probability at the SU. Simulation results confirm that RIS significantly enhances the sensing performance under low Signal-to-Noise Ratio (SNR) conditions, and demonstrate how the harvested energy and RIS configuration jointly influence the spectrum sensing accuracy. The results validate the effectiveness of RIS in enabling energy-aware spectrum access in thermally powered cognitive radio systems.</p>