Improved the simulations. Instead of calculating the temperature at fixed times, I calculated the real part and the imaginary part of the temperature. In this way, the time dependence is given by the rotation of the phasor on the complex plane.
Then I calculated the propagation of the probe through the real part an through the imaginary part of the temperature distribution. These two separate propagations give two values on the photodetector. One is the real part of the signal and one is the imaginary part. Putting these two values together in the complex plane we get the modulus and the phase of the signal. Doing this for each position of the sample we get the scan of the AC signal and of the phase. See the plot.

In the plot there is the comparison of two scans of the same sample, same thickness, same absorption rate, same incident power, but made of different materials: Silica and Sapphire. The sapphire sample looks shorter because the angle inside the sample is smaller (snell's law). The AC signal is smaller because the sapphire diffusivity is higher. The ratio between the two AC signals is 3.7. The phase difference is 33°.