Title: Probing the last scattering surface through the recent and future CMB observations

Abstract: We have constrained the extended (delayed and accelerated) models of hydrogen recombination, by investigating associated changes of the position and the width of the last scattering surface. Using the recent CMB and SDSS data, we find that the recent data constraints favor the accelerated recombination model, though the other models (standard, delayed recombination) are not ruled out at 1-$\sigma$ confidence level. If the accelerated recombination had actually occurred in our early Universe, baryonic clustering on small-scales is likely to be the cause of it. By comparing the ionization history of baryonic cloud models with that of the best-fit accelerated recombination model, we find that some portion of our early Universe has baryonic underdensity. We have made the forecast on the PLANCK data constraint, which shows that we will be able to rule out the standard or delayed recombination models, if the recombination in our early Universe had proceeded with $\epsilon_\alpha\sim-0.01$ or lower, and residual foregrounds and systematic effects are negligible.