Indications for a line of continuous phase transitions at finite temperatures connected with the apparent metal-insulator transition in 2d disordered systems

In a recent experiment, Lai et al. [Phys. Rev. B 75 (2007) 033314] studied the apparent metal-insulator transition (MIT) of a Si quantum well structure. Tuning the charge carrier concentration n, they measured the conductivity sigma(T,n) for a very dense set of n values. They observed linear T dependences of sigma around the Fermi temperature and found that the corresponding T -> 0 extrapolation sigma_0(n) exhibits a sharp bend just at the MIT. Reconsidering the data by Lai et al., it is shown here that this sharp bend is related to a peculiarity of sigma(T=const.,n), which is clearly detectable in the whole T range up to 4 K, the highest measuring temperature in that work. It may indicate a sharp continuous phase transition between the regions of apparent metallic and activated conduction to be present at finite temperature. This interpretation is confirmed by a scaling analysis without fit, which illuminates similarities to previous experiments and provides understanding of the shape of the peculiarity and of sharp peaks found in dlog_{10}sigma/dn(n). Simultaneously, the scaling analysis uncovers a strange feature of the apparent metallic state.