The correlation function for a two-level atom in contact with a reservoir of photons at temperature T is (I'm using natural units):
C_R(\tau)\propto PolyGamma[4, 1-i (\tau/\beta)],
where \beta=1/T.
A plot for T=300K shows that the reservoir decay time is \tau_R~0.25*10^{-13}s, whereas for T=10K, \tau_R~1.5*10^{-12}s. Both times are much less that the decay of an atom t_s~10^{-8}s, therefore, it looks like the Markov approximation works for low-temperatures. Nevertheless, in the literature, I find that such approximation fails when temperature decreases.
What is wrong with my logic?
Thanks.