Å@Four TeV blazars, Mrk 421, Mrk 501, PKS 2155-304, and 1ES 2344+514, have been studied with the X-ray satellites ASCA and RXTE. Because spectral energy distributions of TeV blazars cover all observable bands, and their fluxes vary rapidly, we conducted a number of multi-frequency campaigns. In particular, our observations of Mrk 421 and Mrk 501 in both X-ray and TeV γ-ray bands provide the first truly simultaneous data sets in various phases of source activity. The overall ν Lν spectra of TeV blazars show two broad pronounced peaks in the spectral energy distribution; one is located between the radio and X-ray bands, and another in the γ-ray regime. We found that variability in the X-ray and TeV bands is well correlated on time scale of a day to years. The amplitude of flux variation was comparable in both enengy bands for Mrk 421 ([X-ray flux] ∝ [TeV flux]), while quadratic for Mrk 501 ([X-ray flux]2 ∝ [TeV flux]). The data considered here support the currently popular models where the low energy peak is produced via the synchrotron process, and the high energy peak is produced by Comptonization of the synchrotron photons by the same electrons that produced the synchrotron photons (the SSC model). The physical parameters relevant for the emission were successfully constrained from the observational properties of each TeV blazar.
Analysis of the X-ray data indicated five principal results. First, we detected the peak of the synchrotron component for all TeV blazars. We found that the position of the synchrotron peak shifted from lower to higher energy when the source became brighter. This correlation is best studied for Mrk 421 and Mrk 501. The relationship between the peak energy and luminosity showed quite different behavior in the two sources. Data for Mrk 421 indicated very little change in the peak position (0.5-2 keV), while Mrk 501 revealed the lagest shift ever observed in blazars (1-100 keV). Second, seven day uninterrupted observation of Mrk 421 revealed day-by-day flares, which were strongly correlated from the UV to the hard X-ray bands. We found that the characteristic time scales of individual flare events are similar for all TeV blazars (~ 1 day), but are slightly different in each object. Below this time scale, rapid variability appears strongly suppressed, indicating `strong red-noise' type behavior (∝ f-2 ~ f-3 in the Power Spectrum Density). Third, from the analysis of light curves of the blazar Mrk 421, we found the presence of `hard-lag' flares, where soft X-ray variations precede those in the hard X-rays. More remarkably, time-lags in various X-ray energy bands changed flare by flare, and hence the paradigm of `soft-lag' does not always apply. Fourth, we detected the general trend that the amplitude of variation becomes larger at increasing photon energy. A few exceptions were found in which variation in the lower energy X-ray band is larger or comparable to that in the higher energy band. Fifth, time profiles of each flare event were almost symmetric, meaning that they are characterized by nearly equal rise-time and decay-time. From the detailed temporal analysis, we discovered that the symmetry of the time profiles tends to break down at lower energies.
Clear correlations of synchrotron peak energy with peak luminosity, found in Mrk 421 and Mrk 501, suggest that flaring behavior is repeatable and reproduced in these objects. However, very different spectral evolution of Mrk 421 and Mrk 501 probably indicates some differences in the electron acceleration mechanism at work during the flares. Our result strongly supports the idea that the variability of TeV blazars is most probably due to the changes in the injected electron number and/or maximum Lorentz factor (γmax). Data for Mrk 421 indicate that the flux variability is associated with an increase in the number of electrons, while only small changes are implied for γmax. On the other hand, the flare of Mrk 501 is mostly due to the large changes in γmax while keeping the electron number almost conservative. Interestingly, our X-ray data and interpretation suggest that the origin of the flares may be associated with what is expected from the VLBI observations of superluminal motion.
From the X-ray study of TeV blazars, we suggest that at least four dynamical time scales must be taken into account to understand the observed time variabilities. These are (1) electron accerelation time, tacc, (2) electron cooling time, tcool, (3) source light travel time, tcrs and (4) electron injection time, tinj. We show that all the variability patterns in blazars are well understood by the balance of those time scales. To confirm our knowledge of spectral evolution and rapid variability in blazars more quantitatively, we have developed a new time-dependent SSC code, incorporating the radiative cooling and acceleration process, as well as light travel time effects properly. That is because we deemed the popular SSC model based on steady state emission as inadequate for our case, given the detection of the rapid flux and spectral time variability in blazars. By using the code developed by us, we simulated various types of flares for TeV blazars. We confirmed that `soft-lag' will be expected when the acceleration time of electrons is much shorter than cooling time, while `hard-lag' can be observed when the acceleration time is nearly equal to the cooling time, which is only expected around γmax.