That mathematical models can sometimes produce physically untenable solutions or violate physical laws is a foundational and widely understood principle in applied mathematics and physics, requiring no citation.
The claim that mathematical equations and formal models can generate abstract solutions lacking physical reality or violating physical laws (such as negative absolute temperatures, infinite densities, or non-physical branches in bifurcation diagrams) is a basic fact of mathematical physics and modeling. No specific empirical paper citation is needed for this commonplace methodological observation.