SuperKEKB: Next-Gen e⁺e⁻ Collider

• SuperKEKB is an upgraded e⁺e⁻ collider featuring asymmetric beam energies and a nano-beam collision scheme that achieves ultra-high luminosity in the ϒ-resonance region.
• It integrates advanced RF, injector, and beam diagnostics systems to enable precise measurements in heavy-flavor decays, CP violation, electroweak, and dark-sector physics.
• The collider’s innovative design, including ultra-low emittance and crab-waist optics, marks a shift from discovery-driven to precision frontier experiments.

SuperKEKB is the upgraded asymmetric-energy $e^+e^-$ collider at KEK in Tsukuba, built as the successor to KEKB and serving as the accelerator foundation of the Belle II experiment. It is an intensity-frontier machine rather than an energy-frontier collider: its central function is to deliver extremely high luminosity in the $\Upsilon$-resonance region, especially at $\sqrt{s}=10.58~\mathrm{GeV}$, so that Belle II can perform precision studies of heavy-flavor decays, time-dependent $CP$ violation, rare processes, $\tau$ physics, electroweak measurements, and dark-sector searches in a clean $e^+e^-$ environment with known initial-state kinematics (Yuan, 2012, Collaboration, 31 Mar 2025).

1. Scientific role and historical rationale

SuperKEKB emerged from the $B$-factory program established by KEKB/Belle and PEP-II/BaBar. Those facilities accumulated about $1.5~\mathrm{ab}^{-1}$ near the $\Upsilon(4S)$, established the CKM description of quark mixing and $CP$ violation, and produced broad programs in $\Upsilon$0, charm, and $\Upsilon$1 physics, spectroscopy, and rare searches. The upgrade rationale was that several measurements remained statistically limited, and some tensions with Standard Model expectations were discussed at roughly the $\Upsilon$2 level; one explicitly cited example was the unexpectedly large branching fraction for $\Upsilon$3 (Yuan, 2012).

The machine was therefore conceived as a “super flavor factory” able to move flavor physics from the scale of discovery-era consistency tests to much more stringent precision tests. In the Belle II program, the intended integrated luminosity target is $\Upsilon$4, corresponding in one 2015 projection to about 55 billion $\Upsilon$5 pairs, 47 billion $\Upsilon$6 pairs, and 65 billion $\Upsilon$7 states (Wang, 2015). Later strategy documents continued to frame SuperKEKB as the collider that enables Belle II’s precision frontier across flavor, $\Upsilon$8, electroweak, and dark-sector physics, in a regime where low backgrounds, no fragmentation at the $\Upsilon$9, and known initial-state four-momentum are decisive (Collaboration, 31 Mar 2025).

The collider’s asymmetric beam energies are not incidental to that program. At the $\sqrt{s}=10.58~\mathrm{GeV}$0, the 7 GeV electron beam in the High Energy Ring and the 4 GeV positron beam in the Low Energy Ring generate a center-of-mass boost $\sqrt{s}=10.58~\mathrm{GeV}$1, which is specifically used for time-dependent $\sqrt{s}=10.58~\mathrm{GeV}$2-violation measurements (Collaboration, 31 Mar 2025). In Belle II vertexing studies, the relevant decay-time differences for $\sqrt{s}=10.58~\mathrm{GeV}$3 pairs are stated to be of order $\sqrt{s}=10.58~\mathrm{GeV}$4, corresponding to flight distances around $\sqrt{s}=10.58~\mathrm{GeV}$5, which directly couples the collider design to the inner-detector geometry and vertexing strategy (Simon et al., 2010).

2. Luminosity concept and beam-parameter redesign

The defining machine objective of SuperKEKB is luminosity. Early design documents state a target luminosity of $\sqrt{s}=10.58~\mathrm{GeV}$6, about 40 times the KEKB record of $\sqrt{s}=10.58~\mathrm{GeV}$7, with the increase driven primarily by the nano-beam collision scheme rather than by current alone (Abe et al., 2010, Akai et al., 2018). Later background and strategy papers discussed staged peak-luminosity targets of $\sqrt{s}=10.58~\mathrm{GeV}$8 or $\sqrt{s}=10.58~\mathrm{GeV}$9 in the context of long-term operation and upgrades (Natochii et al., 2022, Collaboration, 31 Mar 2025).

The nano-beam concept reduces the effective overlap region of the bunches by colliding very small beams under a large crossing angle. In one formulation, the large-Piwinski-angle regime is written as

$CP$0

so that the ordinary hourglass limitation of head-on operation is replaced by a much shorter effective interaction region (Zhou et al., 2023). The Belle II Technical Design Report expressed the same idea through the effective overlap length

$CP$1

with the nano-beam condition

$CP$2

rather than the conventional $CP$3 requirement (Abe et al., 2010).

This strategy drove a wholesale re-optimization of machine parameters. Relative to KEKB, the interaction-point beta functions were reduced from $CP$4 horizontally and $CP$5 vertically to design values of $CP$6 and $CP$7 or $CP$8, depending on the design document cited (Yuan, 2012, Akai et al., 2018). The crossing angle was increased to $CP$9, and the beam energies were changed from the KEKB configuration to $\tau$0 for the positron ring and $\tau$1 for the electron ring (Wang, 2015, Akai et al., 2018).

Because the beam-beam parameter scales with $\tau$2, the drastic beta squeeze required a corresponding emittance reduction. Upgrade papers describe reducing emittance from $\tau$3 to about $\tau$4, together with beam-current increases from KEKB values to $\tau$5 in the LER/HER design configuration (Yuan, 2012, Akai et al., 2018). This combination—nano-beams at the IP, lower emittance, and higher stored currents—is the central technical identity of SuperKEKB.

3. Rings, interaction region, injector, and RF complex

SuperKEKB is a double-ring collider consisting of the 4 GeV positron LER and the 7 GeV electron HER, reusing the KEKB tunnel but rebuilding major subsystems. The rings collide in the Tsukuba straight section, and the interaction region is dominated by the final-focus superconducting magnet system QCS. One accelerator review describes QCS as comprising 8 main quadrupole magnets, 43 corrector magnets, and 4 compensation solenoid coils, with local chromaticity correction sections in both planes because the tiny $\tau$6 values generate very large chromaticity (Akai et al., 2018).

The injector complex is integral to the collider concept because the low-$\tau$7, low-emittance regime shortens beam lifetime and makes top-up injection essential. SuperKEKB uses a 600 m linac with 60 accelerating units and a new $\tau$8 positron damping ring. The damping ring is described as indispensable for LER injection and is specified with a circumference of $\tau$9, damping times of $e^+e^-$0, and extracted emittances of $e^+e^-$1 in the horizontal and vertical planes (Akai et al., 2018). This indicates that SuperKEKB was designed as an accelerator–injector system rather than a standalone collider upgrade.

The RF system is correspondingly large. Operational papers describe 38 cavities and 30 klystron stations, with 22 ARES normal-conducting cavities in the LER and 8 ARES plus 8 superconducting cavities in the HER, all at an RF frequency of $e^+e^-$2 (Ogasawara et al., 2022). High beam loading at the design currents made longitudinal coupled-bunch instability a central issue. For SuperKEKB, the lowest dangerous modes were identified as $e^+e^-$3, and a new LCBI damper was developed to suppress them. That system combines an analog single-sideband filter with three parallel digital bandpass filters implemented in an FPGA so that the three unstable modes can be damped simultaneously and phase-adjusted independently (Hirosawa et al., 2018).

Later RF-operation work addressed the distribution of beam loading among the heterogeneous station types. In that context, beam power is written as

$e^+e^-$4

and the ring synchronous condition as

$e^+e^-$5

which made beam-loading optimization a station-phase-allocation problem rather than a simple equal-sharing problem (Ogasawara et al., 2022). This suggests that, in routine operation, SuperKEKB luminosity performance depends not only on optics and beam-beam dynamics but also on detailed RF phase management across many stations.

4. Commissioning history and luminosity performance

SuperKEKB commissioning was staged. Phase 1, from February to June 2016, was carried out without Belle II and without the final-focus system; its goals were basic machine tuning, low-emittance tuning, and vacuum scrubbing. By the end of that phase, stored currents had reached $e^+e^-$6 in the LER and $e^+e^-$7 in the HER, with integrated beam doses of 780 A·h and 660 A·h, respectively (Akai et al., 2018). Phase 2, from 19 March to 17 July 2018, was the first stage with the QCS final focus and Belle II installed, and it was explicitly intended to verify the nano-beam concept in collision (Ohnishi, 2019).

Phase 2 established the basic collider optics and produced the first hadronic event on 26 April 2018. In luminosity operation, the machine reached $e^+e^-$8 in both rings, with test squeezes to $e^+e^-$9 in the HER and $B$0 in the LER, and peak currents of 860 mA and 800 mA. The central commissioning result was that, after correcting local coupling and waist errors near the IP, the luminosity behaved consistently with the expected $B$1 scaling even with $B$2, which was taken as verification of the nano-beam scheme (Ohnishi, 2019).

From April 2020 onward, SuperKEKB entered its crab-waist operating era. Luminosity-performance studies describe the crab-waist optics as a means of suppressing beam-beam-driven resonances that had caused severe vertical blowup without crab waist; its direct geometric luminosity gain was only a few percent, but its dynamical effect on beam-size blowup was decisive (Zhou et al., 2023). In this regime the luminosity record reached $B$3 on 22 June 2022, overtaking KEKB’s best by more than a factor of two (Zhou et al., 2023).

Later status reviews report further progress. A 2025 strategy document states that SuperKEKB reached a peak luminosity of $B$4 with stored beam currents of 1354 mA in the HER and 1699 mA in the LER, and planned to reach $B$5 by 2026 through optics squeezing to $B$6 and increased stored currents of 2.6 A in the LER and 1.8 A in the HER (Collaboration, 31 Mar 2025). A 2025 comparative note emphasized that, although this was still far below the nominal $B$7 design point, SuperKEKB had already achieved specific luminosities substantially above KEKB and had demonstrated sustained operation with $B$8, occasional running at $B$9, and routine use of the virtual crab waist (Zimmermann, 9 Sep 2025).

5. Beam backgrounds, collimation, and machine protection

SuperKEKB’s luminosity strategy produces a correspondingly harsh detector and machine environment. The dominant background classes identified in Belle II background studies are beam-gas scattering, Touschek scattering, radiative Bhabha processes, two-photon $1.5~\mathrm{ab}^{-1}$0 pair production, synchrotron radiation, and injection background (Natochii et al., 2022, Natochii et al., 2023). Early Belle II design papers had already anticipated background levels 10–20 times higher than Belle and about a factor-10 increase in event rate, which is why the detector had to be rebuilt or heavily upgraded for SuperKEKB conditions (Yuan, 2012).

The background-forecast literature treats the path to higher luminosity as conditional on continued control of those sources. One 2022 projection concluded that backgrounds would remain “high but acceptable” until at least $1.5~\mathrm{ab}^{-1}$1 at $1.5~\mathrm{ab}^{-1}$2, while emphasizing that predictions beyond that point were highly uncertain because of the planned redesign of the interaction region (Natochii et al., 2022). A 2023 data-driven analysis, after refinements to the accelerator-plus-Geant4 simulation chain, reached a similar operational benchmark: at $1.5~\mathrm{ab}^{-1}$3 for $1.5~\mathrm{ab}^{-1}$4, the TOP and CDC were predicted to be at approximately half of their maximum acceptable hit rates once both single-beam and luminosity-dependent components were included (Natochii et al., 2023).

Collimation is therefore a central subsystem rather than an accessory. SuperKEKB/Belle II simulation work showed that long-standing discrepancies between measured and predicted backgrounds were largely due to unrealistic collimator modeling. Once jaw geometry, tip scattering, and sequential tracking were improved, the agreement with measurements improved substantially, and collimator aperture scans became informative enough to support a new alignment method (Natochii et al., 2021). This indicates that background control at SuperKEKB is not reducible to a global vacuum or luminosity parameter; it depends sensitively on the exact collimator geometry, aperture, and impedance trade-offs around the ring.

A separate protection problem is Sudden Beam Loss. Operational studies describe SBL as a major luminosity obstacle because it can cause substantial losses within a few beam cycles, damage collimators and Belle II components, and threaten the QCS. To address it, SuperKEKB developed a new protection architecture based on high-speed loss monitors, White Rabbit time synchronization, upgraded abort logic, and acoustic sensors. Those studies report a White Rabbit TDC time resolution of 8 ns, 15 high-speed loss monitors installed by the time of the study, and a new upstream CLAWS-based abort path at D05V1 that was typically $1.5~\mathrm{ab}^{-1}$5–$1.5~\mathrm{ab}^{-1}$6 faster than the second-fastest abort source (Yoshihara et al., 2024). This suggests that machine protection at SuperKEKB is now a time-critical diagnostic discipline in its own right, tightly coupled to luminosity performance.

6. Belle II interface, diagnostics, and planned evolution

SuperKEKB and Belle II were designed as a coupled system. The detector side of that coupling is evident in the Belle II pixel vertex detector: the PXD uses two layers of DEPFET sensors at radii of $1.5~\mathrm{ab}^{-1}$7 and $1.5~\mathrm{ab}^{-1}$8, with $1.5~\mathrm{ab}^{-1}$9-thick active silicon, rolling-shutter readout with a $\Upsilon(4S)$0 readout time, and an expected zero-suppressed data rate on the order of $\Upsilon(4S)$1 in a background-dominated environment (Simon et al., 2010). Those parameters were chosen specifically because the SuperKEKB program combines small vertex separations, higher backgrounds, and far larger event rates than Belle.

The physics program enabled by the collider has also expanded beyond the original heavy-flavor brief. Belle II upgrade-status papers explicitly include searches for direct $\Upsilon(4S)$2 violation, lepton-flavor violation, and dark matter or dark photons, while later strategy documents add electroweak precision measurements and future accelerator upgrades as part of the long-term role of SuperKEKB (Wang, 2015, Collaboration, 31 Mar 2025). One optional direction is the polarized-electron “Chiral Belle” upgrade, which would introduce about 70% longitudinal polarization at the interaction point and add a program of left-right asymmetry measurements in $\Upsilon(4S)$3 at $\Upsilon(4S)$4 (Roney, 2019).

The accelerator itself remains under staged evolution. A 2025 strategy paper describes a near-term plan to reach $\Upsilon(4S)$5 and $\Upsilon(4S)$6 by tightening optics to $\Upsilon(4S)$7 and increasing currents, followed by a step to $\Upsilon(4S)$8 with 3.0 A in the LER and 2.0 A in the HER before the next long shutdown. Longer-term machine-side upgrades under evaluation include injector improvements, RF-system upgrades, and interaction-region work requiring superconducting final-focus magnets based on Nb$\Upsilon(4S)$9Sn cables (Collaboration, 31 Mar 2025).

Recent diagnostics development shows the same trajectory toward finer-grained machine control. A 2026 instrumentation paper reports a silicon-based ultra-fast X-ray beam size monitor that, for the first time at SuperKEKB, provides bunch-by-bunch measurements of the vertical beam size using synchrotron radiation. In that system, the reconstructed bunch structure and single-bunch coded-aperture X-ray images agreed with the existing CMOS-based XRM, and the measurement precision was estimated to be better than $CP$0 (Andrew et al., 9 Jul 2026). A plausible implication is that the future SuperKEKB operating model will depend increasingly on high-bandwidth, bunch-resolved diagnostics as much as on further hardware upgrades.