3 ways to attach PMTs:

- direct (effectively with an air gap)
- using optical grease
- using silicon cookies

Direct coupling is attractive by its simplicity. Cricial point - photon loss on the air gap. Simple calculations (assumption of perpendicular incidence) - 7% photon loss.

Simulations were used to study the distribution of the photon incident angle at the photosensors. Track configuration: theta = 4 degree, phi = 40 degrees (gx = 225 mm, gy = 270 mm)

1. John's analytical simulation. The photon incident angle was measured at the entrance into the window:

2. Geant-based simulation. The photon incident angle was measured at the photosensitive layer:

Photon yield and relative photon loss for different materials between the window and the EV (no gap, airgap 0.1 and 1 mm thick, gap filled with epotek material of thickness 0.1, 1, 5 mm). 3.5 GeV/c muons. Some technical info about cookies is here.

- the shape of the photon incident angle distribution is similar for both simulations (analytical and Geant-based);
- photon loss is larger than the expectation based on the assumption of the perpendicular incidence;
- cookies are promising to reduce the photon loss. Further studies of the cookies are needed.

**Next steps:**

- implement cookies into the Geant-based simulation (material, Cherenkov properties)
- update quantum efficiency of the photosensors (H12700)