A gap between the data of VB measured with our back-scatterometer and the data provided by SurreyNanoSystems made me curious about a possible offset in our data (our data were constantly 10 x higher). So I searched for systematical errors and ended up on the beam-splitter. It seems that by illuminating it with the laser, the PD gets a signal of 0.252nA even when the backscattering is cut of.

My interpretation is that this additional signal is actually scattering from the beam-splitter.

I am now thinking how we can get rid of this influence without violating the SNR too much severely.