You'll need a long-pass filter, which filters out the whole wavelength range shorter than the known prominent peaks of PL emission. You install this filter between the sample and the detector. If your sample does not emit, you'll have zero signal in the stopband of the filter. If there is any luminescence, you'll see peaks exactly corresponding to your excitation bands. I did it on translucent thin films in the transmission mode (http://www.sciencedirect.com/science/article/pii/S0921510712003765). For the reflection mode arrangment, it would be not that straightforward but still feasible. The interpretation is another question. All I can say is that I don't expect much difference for the polycrystalline and monocrystalline samples.