In powder SHG measurements we are giving an input of 1064 nm (Nd:YAG laser) and getting an output of 532 nm. What actually happens at the molecule level?
first, for SHG the molecules in the powder must have an non-centrosymmetric anharmonic potential. The electric field of the incident light "moves" the electrons in a way that the repsonding electric field is of the incident frequency and double the frequency. Usually a single molecule is not producing lots of SHG signal, so usually a decent signal comes from a bundle of molecules. The problem there is that due to the coherent nature of SHG the phase of SHG and incident light are related. So if the potentials of molecules producing SHG are directed the opposite way, SHG is destroyed by interference. In a powder one might assume all powder particles are arranged in diffrent directions. If you still get SHG than probably the molecules in the powderparticles are aligned so SHG is not destroyed by interference. There is so much more to say concerning SHG and the influence of the material structure on SHG, probably best you get a book. I liekd Nonlinear Optics by Robert Boyd and the book on SHG imaging by Pavone an Campagnola.
The problem can be viewed as a driven anharmonic oscillator in which frequency-doubled radiation arises from second-order nonlinear response of electrons to incident light (1064nm in your case): Here you can neglect any ionic response because ions are much heavier. When driven by the "intense" electric field of the incident light, the restoring force of an electron can have higher order terms -F = kx+ax^2+... You can readily see that your perturbative solution to such problem has higher order expansions in terms of Taylor expansion. The second term corresponds to SHG. Of course, there exist even higher order harmonic generations.