<p>The work focuses on designing intelligent reconfigurable surface (IRS) non-orthogonal multiple access (NOMA) energy harvesting (EH) cognitive radio networks (CRNs) with the goal of enhancing reliability, spectral efficiency, and energy efficiency. These networks leverage advanced technologies but are affected by various imperfect factors that degrade performance such as nonlinear energy harvesting, imperfect successive interference cancellation, IRS’s phase adjustment error, and fading severity. Moreover, CRNs, due to their open access mechanisms, are vulnerable to information security threats. The study proposes an analytical framework to quickly evaluate the secrecy performance of IRS NOMA EH CRNs while accounting for all the above imperfections. Analytical results demonstrate that secrecy performance is considerably reduced by the presence of these imperfect factors. Despite these challenges, IRS NOMA EH CRNs outperform baseline networks (e.g., IRS orthogonal multiple access EH CRNs) under all parameter conditions.</p>