Speaker
Description
The Circular Electron-Positron Collider (CEPC) requires efficient particle identification (PID) in the forward region, where the performance of the baseline $dN/dx+\mathrm{ToF}$ system degrades for high-momentum kaons. In particular, approximately $3\sigma$ $K/\pi$ separation up to $20~\mathrm{GeV}/c$ motivates the development of an additional Endcap Ring Imaging Cherenkov (RICH) detector.
We present the simulation and performance study of an Endcap RICH implemented in the CEPC software framework CEPCSW. The detector concept consists of a low-index aerogel radiator, a SiPM photon-detection plane, a lightweight carbon-fiber structure, and reflective inner surfaces. The current geometry employs a $6~\mathrm{cm}$ aerogel radiator and a $30~\mathrm{cm}$ drift distance. The material budget is also investigated, since additional material upstream of the tracking system can increase multiple Coulomb scattering and degrade momentum resolution.
A focusing aerogel with a gradient refractive index is studied to mitigate the emission-point uncertainty associated with a thick radiator. The refractive-index gradient focuses photons emitted at different depths toward similar positions on the photon-detection plane, allowing improved photon statistics without a proportional increase in the effective emission-point contribution to the Cherenkov-angle resolution. Optical photon transport in Geant4 includes wavelength-dependent Cherenkov emission, absorption, Rayleigh scattering, and refraction/reflection at material boundaries. SiPM response is further modeled with wavelength-dependent photon detection efficiency and dark noise.
The simulation demonstrates the impact of these effects on the observed photon pattern. The focusing aerogel produces a narrower Cherenkov ring, while Rayleigh scattering introduces additional background hits and broadens the photon distribution. At larger incident angles, reflected photons generate more complicated multi-arc topologies, presenting an important challenge for realistic reconstruction.
As a first performance benchmark, a simplified reconstruction based on circle-fit preselection and truth-level track information is used to evaluate the intrinsic detector capability. For a $15^\circ$ incident angle, which means direct, non-reflected rings, the simulated detector achieves approximately $3.2\sigma$ $K/\pi$ separation at $20~\mathrm{GeV}/c$, satisfying the target at the hardware-performance level. This benchmark assumes perfect discrimination between signal photons and dark-noise hits and therefore does not represent the final PID performance.
Future work will focus on quantifying the impact of dark noise, developing robust reconstruction methods for noisy and reflected photon patterns, and performing a complete end-to-end PID evaluation with realistic detector response and track reconstruction.
