ePIC Detector: Electron-Proton/Ion Collider

Updated 12 November 2025

ePIC Detector is a high-precision, general-purpose spectrometer that integrates advanced silicon vertexing, fast timing, fine-grained calorimetry, and robust particle identification.
It leverages cutting-edge MAPS, AC-LGAD, and MPGD technologies to achieve superior momentum resolution and effective synchrotron radiation mitigation across a broad kinematic range.
Its comprehensive coverage and low occupancy performance are essential for precise measurements in deep inelastic scattering, nucleon structure, and QCD dynamics.

The Electron-Proton/Ion Collider (ePIC) Detector is the general-purpose spectrometer under development at the Electron-Ion Collider (EIC), designed for high-precision measurements of deep inelastic scattering, heavy flavor, and three-dimensional partonic imaging across a broad kinematic regime. ePIC integrates state-of-the-art silicon vertexing, fast timing, fine-grained calorimetry, advanced particle identification, and low-background infrastructure to enable the full EIC physics program. The following sections detail the technical principles, subsystem architectures, performance validation, SR background mitigation, and R&D underlying the ePIC detector.

1. Detector Architecture and Subsystem Overview

The ePIC detector provides nearly hermetic coverage from the interaction point (IP) over the pseudorapidity range $-3.5 < \eta < 3.5$ , with dedicated far-forward (hadron-going, $\eta > 4$ ) and far-backward (electron-going, $\eta < -4$ ) arrays further extending acceptance for small-angle physics (Pitt, 4 Sep 2024). The main subsystems comprise:

Silicon Vertex and Tracking: Five barrel layers of Monolithic Active Pixel Sensors (MAPS, 10 μm × 10 μm pitch, bent for inner layers) covering $r=3.6$ –$42$ cm, with 0.05–0.55% $X_0$ per layer. Endcap MAPS disks at $z=\pm25$ to $\pm135$ cm provide forward and backward coverage to $|\eta|<3.5$ (Li, 2023, Li, 2023).
Timing and Outer Tracking: AC-LGAD silicon layers (0.5 mm granularity, $\sim$ 30 μm $\sigma_x$ , $<30$ ps $\sigma_t$ ) in barrel and endcaps. Depleted MAPS (MALTA2) forward modules (40–45 μm pitch, $\sim$ 2 ns $\sigma_t$ ) further enhance timing and hit discrimination (Li et al., 1 Oct 2024).
Micro-Pattern Gas Detectors (MPGD): Central $\mu$ RWELL gas trackers (1 mm × 10 mm pads, $\sim$ 100 μm $\sigma_{r\phi}$ ), arranged in three cylinders at $r=33.1$ , 51.0, 77.0 cm.
Electromagnetic Calorimetry: PbWO $_4$ crystals ( $\eta<-1.4$ ), Pb/SciFi barrel calorimeter with embedded AstroPix imaging MAPS ( $-1.7<\eta<1.3$ ), and tungsten/SciFi forward calorimeters ( $1.4<\eta<4$ ) achieve $\sigma_E/E = 2$ – $10\%/\sqrt{E}$ $\oplus$ const (Klest, 19 Aug 2024, Kim et al., 7 Nov 2025).
Hadronic Calorimetry: Steel/scintillator sampling calorimeters (barrel and endcaps), 2.4–7 $\lambda_I$ deep, with fine transverse segmentation for jet and missing energy reconstruction.
Particle Identification: Barrel hpDIRC (3–8 cm, fused-silica bars, 3-layer lens, %%%%27 $r=3.6$ 28%%%% $\pi/K$ at $p<6$ GeV/ $c$ ) (Kalicy, 2022); forward dual-radiator RICH (aerogel + gas, $0.3$–$50$ GeV/ $c$ , SiPM readout); backward pfRICH (aerogel, MCP-PMT, up to $7$ GeV/ $c$ ), barrel and forward AC-LGAD TOF ( $\sim$ 30 ps resolution) (Chatterjee, 27 Oct 2024).
Far-forward/Backward Detectors: AC-LGAD trackers, PbWO $_4$ /LYSO calorimeters, Roman pots, and OMDs for small-angle proton, neutron, and scattered electron tagging (Pitt, 4 Sep 2024).

A 1.7 T solenoidal magnet provides bending power, with the overall envelope $\sim$ 9.5 m (z) $\times$ 3.3 m (transverse) (Li, 29 Jan 2025).

2. Synchrotron Radiation Backgrounds and Mitigation

ePIC is sited at the high-luminosity IP-6, subjecting it to intense synchrotron radiation (SR) produced as beam electrons pass through strong dipole and quadrupole fields. Accurate modeling and mitigation of this background is critical for detector longevity and stable operation (Natochii, 21 Aug 2024).

Monte Carlo SR Simulation Framework:

The custom SynradG4 package, written in C++ atop Geant4 (v11.2.0), implements:

Importation of beamline geometry from Bmad XML or DD4hep detector files.
Transport of primary beams through field maps using G4SteppingManager.
On-the-fly sampling of SR photon emission at each curved path segment, using the theoretical emission spectrum $\varepsilon_{c} = \frac{3}{2}\,\hbar c\,\frac{\gamma^{3}}{\rho}$ $\frac{d^{2}N}{dE\,d\Omega} = \frac{3\alpha}{16\pi^{2}}\,\frac{E}{\varepsilon_{c}}\,\left[ \int_{E/\varepsilon_{c}}^\infty K_{5/3}(x)\,dx \right]$ with $K_{5/3}$ the modified Bessel function of the second kind, and $\alpha$ the fine-structure constant.
Photon propagation through the full detector and beamline geometry.
Surface interactions (absorption, specular and diffuse reflection) using G4XraySurfaceProperty with user-defined optical constants from Henke-Gullikson-Davis tables and the Nevot–Croce roughness factor, $\exp\left[-4k^2\sigma^2\sin^2\theta\right]$ ( $k=2\pi/\lambda$ , $\sigma =$ r.m.s. roughness).
Energy deposits in sensitive detector elements recorded to ROOT outputs.

This SR model is validated against analytic formulas (photon yield in test dipoles, agreement to $<2\%$ ), and X-ray reflectivity vs. grazing angle (better than 5% for Al, Cu) (Natochii, 21 Aug 2024).

SR Flux and Impact:

Simulations yield:

At backward tracker ( $z=-1.2$ m, $r$ =12 cm): SR spectrum peaks at $\varepsilon_c \approx 8$ keV, rate above 1 keV is $1.5\times10^8$ photons/cm $^2$ /s.
Inner pixel layer: SR-induced hit rate of 0.02 hits/mm $^2$ per 10-ns readout, corresponding to an annual dose $\sim$ 200 Gy on silicon sensors.
Geant4 studies indicate no significant SR contribution to calorimeter or PID system occupancy.

Countermeasures:

Measure	Location	Effectiveness
5 mm thick W-alloy SR masks	$\pm$ 0.5 m from IP	$\times$ 25 reduction in forward SR flux
2 mm copper beam-pipe liner	Beam pipe	Additional factor of 5 for $E\lesssim$ 5 keV
Al $_2$ O $_3$ ceramic feedthroughs	Electronics interfaces	Cuts X-ray leakage into readout environment
In-situ bake-out (every 6 months)	Vacuum chamber, beam pipe	Retains low surface roughness, matches reflectivity models

Combined, these measures suppress inner-tracker occupancy from 0.02 to 0.0004 hits/mm $^2$ /10 ns with no measurable tracking or calorimetry degradation (Natochii, 21 Aug 2024).

3. Silicon Vertexing, Timing, and Tracking Subsystems

Vertex and Tracking Geometry:

MAPS Barrel: 5 layers, radii 3.6–42 cm, pitch 10 μm, material budget 0.05–0.55% $X_0$ per layer.
Endcaps: 5 MAPS disks per side, z = $\pm$ 25 to $\pm$ 135 cm, radial coverage up to 43 cm.
Outer AC-LGAD Barrel: single layer at $r\approx64.6$ cm, 0.5 mm × 1.0 mm strips, $\sim$ 30 μm spatial, $<30$ ps timing resolution (Li, 2023, Li, 2023).
Hadron endcap AC-LGAD: disk at $z\approx192$ cm, 0.5 mm × 0.5 mm, $\sim$ 7% $X_0$ per layer including supports.

Timing and Forward Enhancement:

MALTA2 DMAPS FMT: 45 μm pitch, 2.1 ns timing, 4.1 μm spatial precision per hit, $\sim$ 0.7% $X_0$ /disk; deployed in forward/backward regions for enhanced background rejection (Li et al., 1 Oct 2024).
Occupancy: $R_\mathrm{occ} \sim 1.1 \times 10^{-4}$ hits/BC per pixel at design luminosity.
Integration: FMT disks mount outside inner MAPS, reusing mechanical and service interfaces.

Performance Metrics:

Momentum resolution: Central $|\eta|<1$ : $\sigma_{p}/p \lesssim 0.3\%$ at 1 GeV/ $c$ ; $\lesssim 1\%$ at 10 GeV/ $c$ (Li, 2023).
Impact parameter:

$\sigma_{d_0}(p_T) = a \oplus \frac{b}{p_T}$ with $a\approx5$ μm, $b\approx20$ μm GeV, yielding $\lesssim10$ μm for $p_T>1$ GeV.

Efficiency: $\epsilon_\mathrm{trk}>98\%$ for $p_T>0.2$ GeV/ $c$ ( $|\eta|<2.5$ ), $\sim$ 95% to $|\eta|=3.5$ .
Occupancy: $<10^{-4}$ hits/pixel/10 ns (MAPS) at $10^{34}$ cm $^{-2}$ s $^{-1}$ ; LGAD $<1\%$ /channel/BX.

Radiation Tolerance:

Demonstrated $<$ 4 μm resolution post- $10^{15}$ n $_\mathrm{eq}$ /cm $^2$ for MAPS (ALICE ITS3).
LGADs tolerate up to $10^{16}$ n $_\mathrm{eq}$ /cm $^2$ with maintained gain and $<$ 30 ps timing.

4. Calorimetry and Imaging Layers

Electromagnetic Calorimeters:

Region	Type	Resolution	Segmentation
Backward	PbWO $_4$ crystal	$2\%/\sqrt{E}\oplus1$ – $3\%$	$2\times2$ cm $^2$
Barrel	Pb/SciFi+AstroPix	$5\%/\sqrt{E}\oplus1\%$	$<$ 1 mm (AstroPix), 2.5 mm (SciFi)
Forward	W/SciFi blocks ("SpaCal")	$10\%/\sqrt{E}\oplus1$ – $3\%$	$2.5\times2.5$ cm $^2$

Imaging and Particle Separation (Kim et al., 7 Nov 2025):

Embedded AstroPix MAPS (500 µm × 500 µm pitch) layers in the barrel calorimeter enable 3D shower profiling. Layered centroiding yields per-layer spatial resolution $\sim$ 150 μm ( $x, y$ ), and shower-axis resolution $\sim$ 1 mm.
e/π suppression factor $S_\pi\sim100$ at 90% electron efficiency; $\gamma/\pi^0$ separation via cluster imaging enables neutral-pion background reduction in DIS analyses below 5%.

Hadronic Calorimeters:

Steel–scintillator tile modules, $44$– $75\%/\sqrt{E}\oplus6$ – $15\%$ resolution, with depth up to $7\lambda_I$ for full jet containment.

Calibration and Alignment:

In-situ light pulser, source, and survey systems maintain $\lesssim0.5\%$ channel equalization and $<100\,\mu$ m–$1$ mm alignment.

5. Particle Identification Subsystems

Barrel hpDIRC (Kalicy, 2022, Chatterjee, 27 Oct 2024):

Fused-silica bars coupled to a 3-layer spherical lens and prism; MCP-PMT or SiPM readout with 3 mm × 3 mm pixels.
Achieves $>3\sigma$ $\pi/K$ separation up to $6$ GeV/ $c$ , $e/\pi$ to $1.8$ GeV/ $c$ , $p/K$ to $10$ GeV/ $c$ (1.6 mrad single-track $\sigma$ ).

Forward Dual-Radiator RICH:

Aerogel ( $n\approx1.02$ , 4 cm) for $p$ below $5$ GeV/ $c$ , C $_2$ F $_6$ gas for $p$ up to $50$ GeV/ $c$ .
Track $\sigma_\theta =$ $0.5$–$1$ mrad, $\sim$ 22–27 photoelectrons/track, $>3\sigma$ $\pi/K$ to $50$ GeV/ $c$ at 95% $\pi$ efficiency.

Backward pfRICH:

Aerogel ( $n\approx1.04$ ), MCP-PMT, $1.4$ mrad track resolution, $>3\sigma$ $\pi/K$ to $7$ GeV/ $c$ .

Barrel/Forward AC-LGAD TOF:

30 ps time resolution, $3\sigma$ $\pi/K$ at $<3$ GeV/ $c$ (TOF equation $N_\sigma=|t_\pi-t_K|/\sigma_t$ ).

Key RICH/DIRC Equations:

Cherenkov angle: $\theta_C = \arccos(1/(n\beta))$
Momentum threshold: $p_{\text{th}} = m/\sqrt{n^2-1}$
Photostatistics: $N_\sigma = (\theta_{C,\pi} - \theta_{C,K})/(\sigma_{\rm track})$

6. Far-Forward/Far-Backward Taggers and Integration

The FF/FB systems extend ePIC’s kinematic reach for exclusive, low- $x$ , and spectator-tagged processes (Pitt, 4 Sep 2024). Key elements:

Roman pots (z $\sim$ 18 m): AC-LGAD planes track protons with $x_L>60\%$ , $\theta<5$ mrad.
Zero-degree calorimeter (ZDC, z $\sim$ 30 m): LYSO or PbWO $_4$ (EM) plus hadronic section, photon resolution $5\%/\sqrt{E}\oplus3\%$ , neutron $50\%/\sqrt{E}\oplus5\%$ .
Luminosity monitoring via e p $\to$ e $\gamma$ p bremsstrahlung: pair-conversion spectrometer with AC-LGAD layers ( $\sigma_x\sim20\,\mu$ m), $\Delta L/L<1\%$ .

Subdetectors are synchronized via AC-LGAD timing ( $\sigma_t<30$ ps), ensuring event matching between central and forward/backward regions. Geometric and timing integration with barrel and endcap tracking is maintained.

7. Physics Performance and R&D Directions

Full-chain GEANT4 simulations and physics analyses demonstrate:

Track and vertex resolution: primary vertex $\sigma_\text{vertex}\sim20\,\mu$ m, secondary vertex $\lesssim50\,\mu$ m.
Momentum resolution: central $\Delta p/p\sim0.4\%\oplus0.1\%p$ /[GeV/ $c$ ].
PID separation: $>3\sigma$ for $\pi/K$ to $50$ GeV/ $c$ (forward), $6$ GeV/ $c$ (barrel).
EMCal energy resolution: $2$– $10\%/\sqrt{E}$ , hadronic $44$– $75\%/\sqrt{E}$ .
Occupancy: All subsystems designed for $<10^{-4}$ – $1\%$ /channel/BX at $10^{34}$ cm $^{-2}$ s $^{-1}$ (Li, 2023, Kim et al., 7 Nov 2025).
SR background: Controlled to negligible operational impact via layered masking and surface engineering (Natochii, 21 Aug 2024).

The ongoing R&D program targets finalization of large-area MAPS, mass production of AC-LGADs, multi-layer RICH/barrel DIRC prototypes, beam and irradiation tests, and engineering of full-module staves and services for reliable integration and sustained performance at EIC luminosity.

The ePIC detector combines cutting-edge silicon technologies, robust SR mitigation, highly segmented calorimetry, and layered PID to realize a fully hermetic, high-resolution collider spectrometer. This architecture underpins the EIC’s ability to access 3D parton imaging, spin structure studies, small- $x$ QCD, and heavy-flavor observables with precision and efficiency (Yano, 9 May 2025, Higinbotham, 2022).