STCF: Next-Gen Tau-Charm Collider

• Super Tau-Charm Facility (STCF) is a proposed next-generation electron-positron collider operating in the 2–7 GeV range for precision charm, tau, and hadron studies.
• It features advanced accelerator technologies like a crab-waist collision scheme and high-luminosity targets up to 50 times that of BEPCII, enhancing quantum-coherent production.
• The integrated detector and simulation systems provide high-resolution tracking, particle identification, and robust offline processing, supporting a diverse physics program.

The Super Tau-Charm Facility (STCF) is a proposed next-generation high-luminosity electron-positron collider and detector complex in China, designed to operate in the center-of-mass range $2$–$7\ \mathrm{GeV}$. In the conceptual design report, accelerator studies, detector R&D papers, and later physics reviews, it is presented as the post-BEPCII tau-charm facility, with a peak luminosity above $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$ or equivalently exceeding $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$, annual integrated luminosity at the inverse-attobarn scale, and a program spanning charm, tau, hyperon, hadron-spectroscopy, nonperturbative-QCD, CP-violation, and beyond-the-Standard-Model measurements (Achasov et al., 2023, Shi et al., 2020, Bao et al., 15 Sep 2025, Petrov et al., 27 Mar 2026).

1. Project definition and scientific positioning

STCF is framed as a new-generation symmetric or separated-ring $e^+e^-$ collider for the tau-charm energy region, intended to extend the role played by BEPCII/BESIII while moving to much larger samples and a broader energy reach. The physics motivation repeatedly emphasizes the transition region between perturbative and non-perturbative QCD, threshold production of $\tau$ pairs and charmed hadrons, quantum-coherent production, and the clean initial state available in low-energy $e^+e^-$ collisions. In the physics-and-detector conceptual design report, the project is described as providing a data sample about a factor of 100 larger than BEPCII; in the accelerator conceptual design report, the luminosity goal is stated as about 50 times BEPCII (Achasov et al., 2023, Bao et al., 15 Sep 2025).

The phrase “tau-charm factory” is narrower than the full scope adopted in the STCF literature. The facility is proposed not only for charm and tau measurements, but also for hyperon physics, hadron structure, time-like form factors, exotic hadrons, glueballs, light-hadron spectroscopy, and new-physics searches. Later review work treats STCF as part of a broader class of “super tau-charm factories,” stressing threshold kinematics, low background, near-full reconstruction efficiency, and strong control of systematics as the distinctive features of this collider class (Petrov et al., 27 Mar 2026).

Some machine descriptors differ across design stages. The 2023 conceptual design report describes a collider circumference of about 600 m, whereas later accelerator lattice documents describe baseline collider-ring configurations with circumference $860.321\ \mathrm{m}$, and the alternative one-fold lattice study quotes $835.12\ \mathrm{m}$ at the pre-optimized stage and $885.232\ \mathrm{m}$ in the fully optimized design. Likewise, some documents describe luminosity optimization around $7\ \mathrm{GeV}$0, while dedicated longitudinal-beam-dynamics studies formulate the target as $7\ \mathrm{GeV}$1 at the optimized beam energy of $7\ \mathrm{GeV}$2 (Achasov et al., 2023, Zou et al., 25 Jul 2025, Liu et al., 18 May 2026, Zhang et al., 2024).

2. Collider architecture and beam-dynamics design

The accelerator is described as a double-ring collider with a crab-waist collision scheme and an injector that provides top-up injections for both electron and positron beams. The design beam-energy range is $7\ \mathrm{GeV}$3–$7\ \mathrm{GeV}$4 per beam, corresponding to $7\ \mathrm{GeV}$5–$7\ \mathrm{GeV}$6 center-of-mass energy. A large crossing angle of $7\ \mathrm{GeV}$7 and a large-Piwinski-angle collision scheme are central to the luminosity strategy, together with very small $7\ \mathrm{GeV}$8 at the interaction point. In the accelerator conceptual design report, the collider is treated as a typical third-generation circular $7\ \mathrm{GeV}$9 collider facing coupled constraints from strong nonlinear optics, short lifetime, injection, collective effects, and detector integration (Bao et al., 15 Sep 2025).

The luminosity logic is formulated in several related ways. A longitudinal-design study summarizes the dependence as

$0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$0

while crab-waist studies write the Piwinski angle as

$0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$1

The design consequence is that luminosity growth requires high current, small $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$2, and acceptable beam-beam tune shift, but these same choices intensify chromatic and geometric aberrations, reduce dynamic aperture, and increase sensitivity to collective effects and beam-beam instabilities (Zhang et al., 2024, Liu et al., 18 May 2026).

A recurrent constraint is the coherent $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$3–$0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$4 instability generated by the large crossing angle and crab-waist configuration. In the longitudinal-beam-dynamics design, the stability criterion is stated as $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$5, with the practical requirement that $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$6 be at least about 3 times larger than $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$7. The same study imposes a detector-driven bunch-length upper bound of $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$8, considers intrabeam scattering, Touschek effect, potential-well distortion, longitudinal microwave instability, and transverse mode-coupling instability, and sets a minimum Touschek-lifetime goal of $0.5\times10^{35}\,\mathrm{cm^{-2}s^{-1}}$9 (Zhang et al., 2024).

Interaction-region and ring-optics studies develop this program into a modular nonlinear design. The crab-waist interaction-region paper uses local chromaticity correction up to third order, exact $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$0 transformations between sextupole pairs, minimization of the dispersion invariant, optimized beta functions at crab-sextupole locations, octupole correction of superconducting-quadrupole fringe fields, and local anti-solenoid compensation of detector-solenoid effects. When integrated into the collider ring, that design achieves a Touschek lifetime exceeding $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$1 at beam energy $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$2 (Zhang et al., 10 Oct 2025). The collider-ring optics paper extends this with a quasi-two-fold symmetric lattice, $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$3, multi-objective optimization of 46 sextupole families using PAMKIT, momentum acceptance beyond $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$4, and a Touschek lifetime of at least $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$5 at $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$6 and luminosity $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$7; with errors and corrections included, the reported lifetime is about $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$8 (Zou et al., 25 Jul 2025). An alternative one-fold lattice study pushes the same design logic to a more ambitious luminosity of $5\times10^{34}\,\mathrm{cm^{-2}s^{-1}}$9 with a Touschek lifetime of about $e^+e^-$0 at $e^+e^-$1 (Liu et al., 18 May 2026).

3. Detector concept and subsystem development

The conceptual detector is a nearly $e^+e^-$2 spectrometer with, from inside out, an inner tracker, a main drift chamber, barrel particle identification, endcap particle identification, an electromagnetic calorimeter, a superconducting solenoid, and a muon detector integrated with the iron yoke. The target tracking acceptance is $e^+e^-$3; the conceptual goals include charged-track position resolution better than $e^+e^-$4, momentum resolution around $e^+e^-$5 at $e^+e^-$6, and $e^+e^-$7 resolution around $e^+e^-$8. For the calorimeter, the design goal is photon reconstruction from $e^+e^-$9 to $\tau$0, with energy resolution about $\tau$1 at $\tau$2 and position resolution around $\tau$3; a dedicated EMC R&D paper states a 6 mm position-resolution requirement and 300 ps time resolution for a 1 GeV photon (Achasov et al., 2023, Jia et al., 2022).

The baseline tracking system in the CDR uses a split architecture: an inner tracker close to the beam pipe and a 48-layer helium-based main drift chamber extending to about 850 mm radius. The 2024 global-tracking study describes the baseline as a 3-layer uRWELL inner tracker plus a 48-layer MDC in a $\tau$4 magnetic field and covering polar angles from $\tau$5 to $\tau$6. A later MAPS study evaluates an alternative three-layer inner tracker based on Monolithic Active Pixel Sensors, called ITKM, with about $\tau$7 per layer. Using a TCAD-plus-Monte-Carlo workflow integrated into OSCAR, that study selects an HR epi sensor with active-connect strip-like pixels as the baseline MAPS design and reports average layer efficiencies of 99.3%, 99.3%, and 99.4%, spatial resolutions of $\tau$8 in $\tau$9 and $e^+e^-$0 in $e^+e^-$1, intrinsic single-cluster time resolution of $e^+e^-$2, and overall RMS timing resolution of $e^+e^-$3 including 50 ns clock discretization (Zhou et al., 2024, Zhang et al., 4 Jun 2025).

Particle identification is intentionally split between barrel and endcap. For the barrel, the conceptual design adopts a RICH detector using a liquid $e^+e^-$4 radiator and CsI-coated THGEM plus Micromegas photon detection. For the endcap, the DTOF study proposes a DIRC-like time-of-flight detector as the endcap PID system. Its baseline geometry consists of two identical endcap discs located about 1400 mm from the interaction point, each segmented into 4 quadrantal sectors with a 15 mm fused-silica radiator, reflective lateral surfaces, absorbers on the outer surface, and MCP-PMT readout. In Geant4 simulation, with an assumed event-start-time precision of about 40 ps and full detector effects included, the overall reconstructed TOF resolution is about 50 ps, and the $e^+e^-$5 separation power at $e^+e^-$6 is better than $e^+e^-$7 over the entire sensitive area; the same paper also reports $e^+e^-$8 separation in direct TOF comparison and $e^+e^-$9 or better under reconstructed mass hypotheses (Qi et al., 2021).

The EMC baseline uses undoped CsI crystals read out by APDs, and one identified limitation is low effective detected light yield for the UV-dominated scintillation of pure CsI. The WLSP EMC study proposes wavelength shifting in propagation using a nanostructured organosilicon luminophore coated on the Tyvek reflector wrapping. In simulation, the light yield increases from $860.321\ \mathrm{m}$0 to $860.321\ \mathrm{m}$1 for cosmic-ray-like events and from $860.321\ \mathrm{m}$2 to $860.321\ \mathrm{m}$3 for a 200 MeV photon entering from the front. In the cosmic-ray test, the measured amplitude ratio is $860.321\ \mathrm{m}$4, corresponding to a 159% improvement in light yield, and no obvious degradation is observed up to 100 krad in practical terms (Jia et al., 2022).

4. Offline software, simulation, and reconstruction algorithms

The common offline platform is OSCAR, the Offline Software of Super Tau Charm Facility. It is built on SNiPER and integrates podio for the event data model, DD4hep for detector description, and Geant4 for detector simulation. The core architecture is organized into a bottom layer of external libraries, a middle layer of core software, and a top layer of STCF-specific applications. The software design is driven by projected data-processing scale: at the $860.321\ \mathrm{m}$5 peak, the event rate is estimated to be about 400 kHz; STCF is expected to collect about $860.321\ \mathrm{m}$6 $860.321\ \mathrm{m}$7 events per year; the average event size is about 45.7 KB; and these numbers imply roughly 100 PB of raw data per year. OSCAR therefore includes MT-SNiPER/TBB parallelism, a podio-based EDM, and a GlobalBuffer mechanism that decouples I/O from event processing in parallel workflows (Huang et al., 2022).

A separate fast-simulation package supports detector optimization and early physics studies. Built on the BESIII Offline Software System, it parameterizes subsystem responses in terms of detection efficiency, spatial resolution, momentum resolution, energy resolution, timing resolution, PID efficiencies and fake rates, helix-fit error matrices, secondary-vertex quantities, and kinematic-fit variables. The package can vary subdetector performance through scale factors, external histograms, input curves, or discrete data points, and is validated against BESIII full simulation for single-object resolutions, event-selection efficiencies, invariant-mass distributions, vertex fits, and kinematic-fit $860.321\ \mathrm{m}$8 distributions (Shi et al., 2020).

Charged-particle reconstruction has been developed around Hough-transform-based global tracking in the STCF software framework. The 2024 tracking paper uses a standard five-parameter helix representation $860.321\ \mathrm{m}$9, conformal mapping in the transverse plane, Hough-space peak finding, stereo-hit association in the $835.12\ \mathrm{m}$0-$835.12\ \mathrm{m}$1 plane, and final fitting with the GENFIT2 Deterministic Annealing Filter. The algorithm is designed for charged particles from $835.12\ \mathrm{m}$2 to $835.12\ \mathrm{m}$3, and for particles exiting through the MDC barrel it reports relative momentum resolution better than $835.12\ \mathrm{m}$4 over most of the momentum range and better than $835.12\ \mathrm{m}$5 around $835.12\ \mathrm{m}$6 for muons (Zhou et al., 2024).

Later work extends this baseline in two directions. A Graph Neural Network-based noise-filtering algorithm for the STCF drift chamber converts hits into graphs, classifies edges with PyTorch Geometric, and applies a tiered threshold strategy with threshold 0.1 in superlayers 0–1 and 0.5 in other layers. In Monte Carlo studies for $835.12\ \mathrm{m}$7, it achieves ROC AUC $835.12\ \mathrm{m}$8, noise rejection rate 86.8%, signal selection efficiency 98.2%, and reduces the fake rate from 8.7% to 1.7% under standard background and from 21.6% to 2.5% under double background (Jia et al., 12 Jul 2025). For online processing, a GPU-accelerated matrix-based Hough transform reformulates conformal Hough tracking as regular matrix operations on CUDA. On five representative channels with nominal background overlay, it retains 93.04% of true signal hits while reducing retained hit volume to 34.92% of the original, processes 1,000 events in approximately 0.14 s, and reaches a speedup of 151.57$835.12\ \mathrm{m}$9 over the CPU baseline (Peng et al., 5 Jul 2026).

5. Physics program

The STCF physics case is organized around precision Standard Model tests and new-physics searches in the $885.232\ \mathrm{m}$0–$885.232\ \mathrm{m}$1 region. The major domains are charm physics, tau physics, CP violation in mesons and baryons, hadron structure, time-like form factors, hadronization, exotic spectroscopy, glueball and hybrid searches, and light hidden-sector particles. The principal methodological advantages are threshold production, quantum coherence, low background, known initial-state quantum numbers, and high full-event reconstruction efficiency (Achasov et al., 2023, Petrov et al., 27 Mar 2026).

In charm physics, the conceptual design report emphasizes quantum-correlated $885.232\ \mathrm{m}$2 production at threshold and projects about $885.232\ \mathrm{m}$3 precision on the mixing parameters $885.232\ \mathrm{m}$4 and $885.232\ \mathrm{m}$5 with $885.232\ \mathrm{m}$6 at 4.009 GeV. The same document states that STCF can produce $885.232\ \mathrm{m}$7 $885.232\ \mathrm{m}$8 pairs per year at $885.232\ \mathrm{m}$9, with sensitivity to CP-violating observables at the $7\ \mathrm{GeV}$00 level in many charm and tau channels, and order-of-magnitude improvements in hyperon CP-violation measurements through quantum-entangled baryon-antibaryon samples (Achasov et al., 2023).

Hadron spectroscopy is a central component of the program. The dedicated spectroscopy paper argues that STCF is especially well matched to hidden-charm and charmonium-like spectroscopy because of its clean environment, broad energy coverage, and fine-scan capability. It proposes a fine scan from 3700 MeV to 7000 MeV in 10 MeV steps with $7\ \mathrm{GeV}$01 at each point initially, followed by dedicated high-statistics running. At $7\ \mathrm{GeV}$02, STCF is described as an “$7\ \mathrm{GeV}$03-meson factory,” with about $7\ \mathrm{GeV}$04 $7\ \mathrm{GeV}$05, $7\ \mathrm{GeV}$06 each of $7\ \mathrm{GeV}$07 and $7\ \mathrm{GeV}$08, and about $7\ \mathrm{GeV}$09 $7\ \mathrm{GeV}$10 events per year. The same paper highlights fully charmed tetraquarks in the 6–7 GeV region, charmonium hybrids such as the $7\ \mathrm{GeV}$11 hybrid around $7\ \mathrm{GeV}$12, and extensive light-hadron and baryon spectroscopy, including $7\ \mathrm{GeV}$13, missing nucleon resonances, and $7\ \mathrm{GeV}$14 states (Guo et al., 2022).

The STCF literature repeatedly frames the machine as complementary to Belle II and LHCb rather than redundant with them. Review work stresses that Belle II can access the tau-charm region through ISR and LHCb has very large heavy-flavor samples, but neither offers the same combination of direct threshold production, coherence, low background, and model-independent absolute measurements for channels with neutrals or missing energy. This complementarity is a recurring argument for STCF’s role in precision charm, tau, hyperon, and nonperturbative-QCD measurements (Petrov et al., 27 Mar 2026).

6. Prospective searches and the state of the R&D program

Beyond the core Standard Model program, STCF has been used as a platform for targeted feasibility studies of light and long-lived new particles. A detector-level study of the protophobic $7\ \mathrm{GeV}$15 boson uses the MDC geometry and displaced $7\ \mathrm{GeV}$16 vertices in the process $7\ \mathrm{GeV}$17, finding that STCF can discover the protophobic boson while tolerating $7\ \mathrm{GeV}$18 background events for specific regions of parameter space around the 17 MeV peak. That study explicitly identifies itself as the first feasibility analysis of displaced light-boson searches at STCF and states that a full Geant-4 simulation is required before any definitive discovery claim can be made (M. et al., 10 Dec 2025).

Other studies extend the same logic to exotic tau decays and heavy neutral leptons. A search for leptophilic axion-like particles in $7\ \mathrm{GeV}$19 considers prompt detection at $7\ \mathrm{GeV}$20 and $7\ \mathrm{GeV}$21 and proposes a cylinder-like far detector with front face 10 m from the interaction point, radius 20 m, and length 30 m. In the tauphilic scenario, the paper identifies the previously untested region around $7\ \mathrm{GeV}$22 and $7\ \mathrm{GeV}$23 as a particularly important STCF target; in the lepton-flavor-universal scenario, it identifies $7\ \mathrm{GeV}$24 and $7\ \mathrm{GeV}$25 as the main new regime accessible at STCF (Jiang et al., 28 Sep 2025). A separate OSCAR-based study of long-lived heavy neutral leptons produced through $7\ \mathrm{GeV}$26, $7\ \mathrm{GeV}$27, at $7\ \mathrm{GeV}$28 concludes that STCF can probe $7\ \mathrm{GeV}$29 about one to two orders of magnitude beyond existing bounds, especially for $7\ \mathrm{GeV}$30, while also stressing the model dependence of the result and the assumption of vanishing background after selection (Li et al., 12 May 2026).

Taken together, the accelerator, detector, software, and physics papers present STCF as an extensive R&D program rather than a frozen design. The 2025 accelerator conceptual design report states that the project aims to secure support from the Chinese central government for construction during the 15th Five-Year Plan (2026–2030), while the 2023 detector-and-physics report emphasizes continuing R&D on the inner tracker, PID, calorimetry, trigger and DAQ, software, radiation tolerance, aging, and benchmark physics studies. This suggests an evolving facility concept in which machine design, detector optimization, and physics reach are being developed in parallel (Bao et al., 15 Sep 2025, Achasov et al., 2023).