My sample was the copper paste contained PEG, and there were no endothermic peaks or exothermal peaks, so I was wondering why did the heat flow in the 2nd and the 3rd cycle decrease? Thanks!
There is nothing wrong or strange with your results. Usually, the power of the sample heater and reference heater are slightly asymmetric, and the asymmetry isn't constant (or the signal from the thermocouples is slightly asymmetric, depending on the type of your calorimeter). The system's state at starting point of the measurement cycle is always slightly different, and for the given measurement cycle it is just the value of asymmetry at which the cycle starts.
In this particular case, there may be the reason I would take into account: the volatility of the sample. After each cycle, weigh the sample and check whether the mass hasn't changed. If the mass is constant throughout the experiment, the explanation of the gradual shift of the curves is simple: the starting point (before heating) is at certain level of power or signal difference (depending on the type of your DSC). After the end of cooling stage (when the asymmetry is different from the heating stage) you start again the next measurement cycle, so the curve starts at a different level of power or signal difference. And again, and again... As I see, the subsequent curves are equally spaced, so the process is nicely repetitive. Just disregard the gradual shift of the curves.
Finally, let me emphasise that the level of the signal is not important, when you investigate the phase transitions. What matters, is the deviation from the baseline. The baseline manifests the "properties" of the equipment (asymmetry of sensing heads or heaters, crucibles, sample's mass and its heat capacity etc.) The level of the signal is important when you perform the measurement of specific heat (heat capacity).
Also, In DSC analysis, the peaks integrated area as well as their temperature (enthalpy) and transition temperature are more important than heat flow values.