LHCb VELO Upgrade: VELI4SBR Hybrid Pixel Detector

• VELI4SBR is the upgraded LHCb vertex detector using a hybrid pixel system tailored for precise heavy-flavor tracking in high-luminosity environments.
• It incorporates the VeloPix ASIC, enabling a 40 MHz trigger-less readout with advanced pattern recognition in dense collision events.
• The design emphasizes reduced material and innovative thermal management, ensuring robust operation under extreme radiation and occupancy conditions.

VELI4SBR most plausibly denotes the upgraded Vertex Locator (VELO) of the LHCb experiment: a replacement of the strip-based silicon vertex detector by a hybrid pixel system designed for LHC Run III and for a trigger-less full-readout architecture at 40 MHz. The VELO surrounds the interaction region and provides precise tracking close to the primary vertex, supporting excellent impact parameter resolution and strong sensitivity to decays of heavy-flavor hadrons. In the upgrade configuration, it is intended to operate at luminosities of $2\times10^{33}\ \mathrm{cm}^{-2}\mathrm{s}^{-1}$ while supplying fast pattern recognition and track reconstruction to a software trigger farm (Hennessy, 2016).

1. Experimental role and motivation

The VELO is the silicon detector surrounding the LHCb interaction region. Its central function is vertexing and near-beam tracking, so its performance directly affects impact parameter resolution and, consequently, heavy-flavor reconstruction. The upgrade is tied to the broader reconfiguration of LHCb into a trigger-less system in which the full detector is read out continuously at the 40 MHz bunch-crossing rate and all data reduction algorithms are executed in a high-level software farm.

The immediate motivation is that by the end of Run II many LHCb measurements were expected to remain statistically limited, while Run III increases luminosity by about a factor of five. Under those conditions, the previous trigger architecture becomes a bottleneck: the hardware trigger saturates for many hadronic decay modes, the 1 MHz limit suppresses efficiency, and removing the hardware trigger allows use of full event information at the first decision stage. The upgrade is therefore driven simultaneously by higher luminosity, higher occupancy, higher radiation dose, trigger-less readout, and the need for improved tracking and vertexing for software-trigger reconstruction.

A common simplification is to describe the change as only a strip-to-pixel replacement. The published design is broader: it couples new sensors and front-end ASICs to a different readout model, a revised mechanical envelope closer to the beam, a lower-mass RF foil, and a new cooling concept.

2. Detector replacement and geometric changes

The upgraded VELO replaces the existing strip sensors with a hybrid pixel detector. In this architecture, the sensor and readout ASIC are separate components that are bump-bonded together. The stated rationale is improved local hit-occupancy handling, finer granularity, better pattern recognition in dense environments, and greater robustness at higher rates.

Several baseline parameters change simultaneously:

Parameter	Current VELO	Upgraded VELO
Modules	42	52
Sensor thickness	300 $\mu$m	200 $\mu$m
Closed position from beam	8.2 mm	5.1 mm
Readout rate	1.1 MHz	40 MHz

The pixel sensors have a pitch of $55\times55\ \mu\mathrm{m}^2$. Moving the detector closer to the beam while also reducing upstream material is integral to the expected gain in impact parameter resolution, since material before the first measured hit increases multiple scattering and degrades tracking precision. The paper also describes the two VELO halves as closing to form a diamond shape around the beam, with the beam passing through the center (Hennessy, 2016).

3. Front-end electronics, bandwidth, and software-trigger integration

The upgraded system is built around the VeloPix ASIC, from the Timepix/Medipix family and based on Timepix3. VeloPix is capable of reading out at 40 MHz, supports bandwidth up to 20.4 Gb/s per ASIC, uses binary readout, and has a $256\times256$ pixel matrix. Data are grouped into SuperPixels of $2\times4$ pixels, a packaging choice reported to reduce output bandwidth by about 30%.

The rate environment is extreme. The hottest region reaches 900 Mhits/s, and the abstract quotes a total data rate of more than 3 Tbit/s for the upgraded VELO. The body description also notes a comparison-table value of about 1.2 Tb/s for the detector; the stated explanation is that these figures correspond to different levels of aggregation or estimation. A useful chip-level comparison is also given: Timepix3 operates at about 80 Mhit/s, whereas VeloPix is intended for 900 Mhits/s in the hottest region.

These electronics are not separable from the trigger model. Because event selection is no longer performed in hardware, the upgraded VELO must deliver high-quality space points and fast pattern-recognition inputs directly into software reconstruction. The detector is therefore specified not only by sensor granularity and radiation tolerance, but also by its ability to sustain dense, low-latency data transport into the software-trigger environment (Hennessy, 2016).

4. Mechanical envelope, vacuum separation, and thermal management

The detector modules operate in a secondary vacuum separated from the primary LHC beam vacuum by a thin RF foil. This arrangement allows the active detector to approach the beam line while preserving vacuum isolation. The foil is reduced from the current 300 $\mu$m aluminium thickness to 200 $\mu$m, is custom manufactured, may be milled from a solid aluminium block, and may be further thinned in selected regions by chemical etching.

Material reduction is closely linked to tracking performance. The foil thickness has a direct effect on impact parameter resolution because multiple scattering before the first measured hit worsens precision. This makes the foil both a vacuum boundary and a tracking-performance constraint.

Low-mass heat extraction is handled by evaporative CO$_2$ cooling through microchannels etched into a 400 $\mu$m-thick silicon substrate. Coolant flows from an inlet pipe into a cooling connector soldered to the silicon substrate and then fans out into parallel microchannels directly below the VeloPix chips. The channel geometry includes a restriction region and a transition region, widening from $\mu$0 to $\mu$1 to stimulate boiling. The design target is to minimize temperature gradients and route coolant directly under the dominant heat sources, which are the VeloPix chips, with full-chip power density estimated at $\mu$2.

The expected maximum cooling power is greater than 36 W. The sensors must be kept at about $\mu$3 to avoid thermal runaway from radiation damage, while the coolant operates around $\mu$4. The module concept combines a carbon-fibre structure, a silicon microchannel substrate, four silicon sensors per module, twelve VeloPix ASICs bump-bonded to the sensors, ASICs glued directly to the substrate, and a Kapton hybrid carrying discrete electronics, power, bias, readout circuitry, and a GBTX chip for slow control (Hennessy, 2016).

5. Radiation environment and validated sensor behavior

The upgraded sensors are specified for a substantially harsher radiation environment than the previous VELO. The maximum fluence is quoted as $\mu$5, compared with $\mu$6 for the current detector. In parallel, the high-voltage tolerance rises from 500 V to 1000 V.

Beam-test and irradiation studies were reported for candidate sensors from Hamamatsu (HPK) and Micron. These devices were tested before and after irradiation to the full expected dose. After irradiation, they are expected to retain 99% hit efficiency at up to 1000 V, and both vendors showed similar performance. The collected charge after irradiation still meets the 6000 electron requirement at 1000 V, although the single-hit resolution worsens by roughly a factor of two in the worst case.

The baseline sensor choice is 200 $\mu$7m n-on-p silicon, with the caveat that thinner sensors remain possible if they perform similarly. Irradiated sensor assemblies were tested with a Timepix3 telescope, used as a proxy because it has the same pixel geometry as VeloPix even though its readout is slower and analog. The reported result is that irradiated sensors up to the full dose still met operational requirements, and the measured charge collection and hit resolution were sufficient to support the upgrade baseline (Hennessy, 2016).

6. Technical significance, interpretation, and possible ambiguity of the identifier

Taken together, the design choices enable the upgraded VELO to deliver much better vertexing and impact parameter resolution, fast high-granularity pattern recognition, robust track reconstruction in high-occupancy conditions, radiation-tolerant operation near the beam line, and efficient operation in the trigger-less Run III software trigger. In that sense, the detector is not a peripheral subsystem but a central enabling element of the full-readout LHCb strategy.

A plausible misconception is that the main novelty lies only in finer segmentation. The documented system-level novelty is broader: 40 MHz readout, software-trigger integration, VeloPix bandwidth management, reduced material in the RF foil, microchannel evaporative CO$\mu$8 cooling, and operation under substantially higher fluence all contribute to the redesign.

Within the literature associated with the identifier, there is also a possible lexical ambiguity. A separate and unrelated instrument, Veloce Rosso, is an ultra-stable fibre-fed R4 echelle spectrograph for the 3.9 m Anglo-Australian Telescope, intended for precision radial-velocity measurements of Sun-like and M-dwarf stars at sub-m/s precision. That system belongs to astronomical spectroscopy rather than collider vertex detection. The available context indicates that VELI4SBR is most likely to be understood as the LHCb VELO Upgrade rather than Veloce Rosso, but the similarity of names can create confusion if the surrounding experimental domain is not specified (Gilbert et al., 2018).