Despite the tremendous success, it appears a bit surprising that a heavily disordered material like amorphous silicon should contain only one single well-defined type of defect. In fact, experimental evidence against this simple picture was reported by several autors, but apparently not much acknowledged by the promoters of the equilibrium model.
One of the first such observations came from Fritzsche's lab in Chicago. Han and Fritzsche observed that that illumination at 160 and 300 K produces essentially the same degradation of the conductivity, but very different amounts of defect absorption at 1.0 eV [Han-1983jncs]. The authors conclude the presence of at least two different defect states. During illumination, conductivity and defect absorption increase, but during anneal the conductivity (evidenced by the mobility-lifetime product) reovers faster than the absorption signature. Eventually both can be annealed out, but the latter requires higher temperatures.
Another insight came from solar cell characterization. The accepted condition for presenting stabilized results is after exposure to 1000 hours of AM1.5 illumination at 50°C. For development this procedure is a bit cumbersome because after producing a device, you have to wait for more than a month before getting a definite feedback on the process modifications you might have done. Therefore, people tried to speed up the stabilization time by exposing to higher light intensities or at different temperatures. With surprising results!
The figure below shows the behaviour of efficiencies during a degradation experiment [Yang-1993apl]. The efficiency does not directly reflect the defect density, but it is nevertheless a sensitive gauge. Of two identical cells, one was kept in the initial state, the other was degraded by high intensity light of 50 suns at 50°C. Then, both cells were exposed to the standard conditions of 1 sun. If the single defect model were correct, the additonal illumination would just result in further degradation of the degraded cell. However, this cell is actually observed to recover to some extent, and only after prolonged illumination it reaches the same stabilized state as the one that started from its initial state.
Normalized efficiencies of two cells during exposure to 1 sun; one of the devices started from the initial (annealed) state, the other one starts from a degraded state (2 hours of 50 suns high intensity exposure) [Yang-1993apl].