The energy of the lowest-lying optical phonons in VO2 is of the order if 10 meV (see the attached diagram for phonon dispersions in VO2), which, converted to frequency ν, amounts to of the order of ν = 2.5 × 1012 Hz. The period T corresponding to this frequency amounts to T = 0.4 × 10-13 s ≡ 40 fs. An integer multiple of this value amounts to the fundamental relaxation time for restoring local equilibrium in VO2. The time 200 fs that you mention seems therefore a reasonable value, as 200/40 = 5. □

http://www.nature.com/nature/journal/v515/n7528/fig_tab/nature13865_F3.html

Thanks Behnam, 

Just to make sure that I understood your point let me ask one more question. Is demetallization (going to semiconductor phase) of VO2 can be as fast as its metallization process?

Mamnoon,

You are welcome.

As for your last question, not necessarily.  The metal-insulator phase transition in VO2 is first order, so that there is a hysteresis in the conductivity curve as a function of temperature (I assume that you induce insulator-metal transition in your work by heating your sample by means of an intensive laser pulse). The dynamics is in general different in going from insulating to the metallic phase than in the reverse direction. It seems almost certain to me that in your case the dynamics of the two transitions are vastly different: while you can induce an insulator-metal transition by overheating your sample (through increasing the intensity of the laser pulse), you cannot do much to get the heat out of your sample in order to get it to a temperature below the metal-insulator transition; the only means available is having the sample in contact with a low-temperature heat bath, maximising the temperature gradient for accelerating the diffusive heat transport from your sample to the heat bath.  □

Then how fast is the metal to insulator phase transition (getting heat out of my sample)? Are 40 fs or 200 fs logical numbers?

To answer the question properly, one should solve a non-equilibrium problem. A hand-waving argument suggests a multiple of 40 fs, since a single collision (a process of momentum transfer) is not sufficient for equilibration. The time 200 fs seems therefore reasonable. Of course, as an experimentalist you will have to find out whether a time considerably shorter than 200 fs is feasible. □