Our Universe is filled by Cosmic Microwave Background (CMB) radiation with black body spectrum having a temperature of 2.7K. In the expanding Universe the temperature earlier was higher than today: Tr = To (1+z), where z is the redshift. Number density of photons in our Universe billion times exceeds the number density of electorns. The most important process of interaction of photons and electrons is the Thomson scattering. Hydrogen recombination at redshifts z ~ 1100 - 1300 leads to rapid decrease of the optical depth of the Universe due to Thomson scattering. Photons coming to us from the epoch of recombination carry information about velocities of electrons and density perturbations of matter and radiation. WMAP and PLANCK spacecraft measured with enormous accuracy these traces in the angular distribution of CMB, originated during the epoch of hydrogen recombination due to existence of the last scattering surface. There is a question: how and at which redshift the observed practically ideal black body spectrum of CMB was formed. Why we dream that proposed space missions like PIXIE will be able to detect traces of any significant energy release in our Universe at redshifts smaller than 2 million? I plan to describe the nature of the black body photosphere of the Universe and why we will never observe in the spectrum of CMB the traces of electron-positron annihilation in the early Universe or due to decay of tritium and Be to Li conversion at z ~ 30 000. Spectral features in the CMB energy spectrum contain a wealth of information about the physical processes in the early Universe, at redshifts z ≲ 2 × 106 (including possible decay and annihilation of unknown types of dark matter particles). The CMB spectral distortions are complementary to all other probes of cosmology. In fact, most of the information contained in the CMB spectrum is inaccessible by any other means. This talk outlines the main physics behind the spectral features in the CMB throughout the history of the Universe, concentrating on the distortions which are inevitable and must be present at a level observable by the next generation of proposed CMB experiments. The spectral distortions considered here include spectral features from cosmological recombination, resonant scattering of CMB by metals during reionization which allows us to measure their abundances, and two fundamental types of spectral distortions typical for local Universe and for very high redshift Universe. First of them is observed already in thousands of massive clusters of galaxies by Planck spacecraft and groundbased South Pole Telescope and Atacama Cosmology Telescope. Second one (Bose-Einstein type) spectral dsitortion was created at high redshifts due to viscous dissipation of the primordial sound waves. It carries the unique information about the spectrum of small scale primordial density fluctuations which originated during inflation stage of the Universe expansion and completely dissipated leaving no other observable traces.