FPAs for WFC3-IR: a trade-off study

Recent tests indicate that the flight detectors selected for WFC3-IR may be subject to higher than expected background due to radiation effects. In this document I compare the performance of the FWC IR channel assuming FPA#64 with higher background versus the new substrate removed detectors currently under development at Rockwell Science Center. I also use as current benchmark the performance of NICMOS in the F110W, F160W and F126N filters, and of ACS in the F850LP filter. The analysis is based on the model and metrics (speed and discovery efficiency) of Stiavelli and Robberto (2003), with a newly introduced efficiency factor, the ultra-deep sensitivity, that represents the limiting magnitude reached in a HUDF-like project. The main results are: 1) the worst case scenario of high extra background (~1e/s/pix) on FPA#64 has a major impact on the sensitivity of WFC3-IR; 2) the factor of ~2 improvement in quantum efficiency at short wavelengths provided by the substrate removed parts offers a substantial increase of sensitivity in the F110W filter; 3) also with substrate removal, a rather pessimistic increase of readout noise up to ~40e for correlated double sampling does not significantly affect the sensitivity of deep surveys performed in broad band filters; 4) WFC3 with substrate removed detector in the F098M band is approximately 2 times more efficient than ACS in F850LP for deep imaging, a gain that comes at the price of coarser sampling (130mas/pixel vs. 50mas/pixel).