ProtoDUNE-SP: LArTPC Prototype

• ProtoDUNE-SP is a full-scale single-phase LArTPC prototype built for DUNE, featuring an active volume of 7×6×7.2 m³ and near-complete electronics operability.
• It integrates full‐size components such as membrane cryostat, high voltage systems, and cold electronics, validated under challenging beam and cosmic-ray conditions.
• Calibration campaigns and beam tests provided essential hadronic, low-energy electron, and photon-detection measurements to refine reconstruction and systematic strategies for DUNE.

ProtoDUNE Single-Phase (ProtoDUNE-SP) is the single-phase liquid argon time projection chamber prototype constructed for the Deep Underground Neutrino Experiment (DUNE) at CERN’s Neutrino Platform. It was built with full-size components of the first DUNE far-detector module, with an active volume of $7\times 6\times 7.2~\mathrm{m}^3$ and about $0.77~\mathrm{kt}$ of liquid argon; the technical design report described it as the largest monolithic single-phase LArTPC detector to be built to date (Collaboration et al., 2021, Abi et al., 2017). Beyond technology validation, ProtoDUNE-SP operated as a charged-particle beam and cosmic-ray experiment from 2018 to 2020, establishing the operating characteristics of large single-phase LArTPCs and producing calibration, reconstruction, photon-detection, and hadron-argon measurements directly relevant to DUNE (Collaboration et al., 2020, Collaboration et al., 14 Nov 2025).

1. Experimental role within DUNE

ProtoDUNE-SP was built and operated in CERN’s North Area under experiment NP-04 as a full-scale prototype of one DUNE single-phase far-detector module (Collaboration et al., 2021). Its function was not limited to validating isolated subsystems: the program covered the membrane cryostat, cryogenics, mechanics, high voltage, cold electronics, photon detection, DAQ, calibration, and integrated operations, while simultaneously exploiting the H4 beam line for controlled exposure to charged pions, kaons, protons, muons, and electrons in the range $0.3$ to $7~\mathrm{GeV}/c$ (Collaboration et al., 2020).

The detector therefore occupied an intermediate position between a technology demonstrator and a physics instrument. It incorporated full-size anode, cathode, field-cage, and cryogenic components as intended for DUNE, but it was operated in a surface test-beam environment with beam-line PID instrumentation and a substantial cosmic-ray rate. This suggests that ProtoDUNE-SP should be understood not as an environment-identical replica of an underground DUNE far detector, but as a full-scale prototype in which detector construction, commissioning, calibration strategy, and reconstruction algorithms could be stressed under experimentally demanding conditions.

Operationally, ProtoDUNE-SP reached beam running in 2018 and continued through 2020. The detector first accumulated cosmic-ray data and then recorded test-beam data in autumn 2018; later campaigns included dedicated studies such as xenon doping of the liquid argon and detailed purity measurements over extended periods (Collaboration et al., 14 Nov 2025, Collaboration et al., 2024).

2. Detector architecture and instrumentation

ProtoDUNE-SP is a single-phase LArTPC split by a central cathode plane assembly into two drift regions of approximately $3.6~\mathrm{m}$ each, with the nominal cathode bias at $-180~\mathrm{kV}$ and a drift field of $500~\mathrm{V/cm}$ (Collaboration et al., 2021). Ionization electrons drift entirely in the liquid phase toward anode plane assemblies (APAs) mounted on the outer walls. Each APA is approximately $2.3~\mathrm{m}\times 5.984~\mathrm{m}$ and carries four wire layers, denoted $G$, $U$, $0.77~\mathrm{kt}$0, and $0.77~\mathrm{kt}$1; the $0.77~\mathrm{kt}$2 and $0.77~\mathrm{kt}$3 induction planes are wrapped at $0.77~\mathrm{kt}$4, while the $0.77~\mathrm{kt}$5 collection plane is vertical (Collaboration et al., 2021). In operation, the drift-time coordinate complements the multiple wire-angle projections to provide three-dimensional reconstruction (Collaboration et al., 14 Nov 2025).

Component	Principal characteristics	Citation
Active TPC	$0.77~\mathrm{kt}$6 active volume; two $0.77~\mathrm{kt}$7 drift regions	(Collaboration et al., 2021)
APAs	Six APAs; four wire layers $0.77~\mathrm{kt}$8; $0.77~\mathrm{kt}$9 at $0.3$0, $0.3$1 vertical	(Collaboration et al., 2021)
Readout channels	$0.3$2 LArTPC wires instrumented with cryogenic front-end electronics	(Adams et al., 2020)
Beam entrance	Nitrogen-filled beam plug to minimize upstream material	(Collaboration et al., 2020)
Photon detection	Photon detectors embedded inside each APA	(Collaboration et al., 2020)

The field cage surrounds the active region and grades the cathode potential to the grounded anodes in discrete steps (Collaboration et al., 2021). In one detailed description, the voltage was graded from $0.3$3 to $0.3$4 in $0.3$5 steps using $0.3$6 per $0.3$7 resistor stage (Collaboration et al., 2021). The beam-facing side included a nitrogen-filled beam plug to reduce the inactive liquid-argon thickness ahead of the TPC and thereby limit multiple scattering and upstream interactions before particles entered the fiducial region (Collaboration et al., 2020).

Photon detection formed an integral part of the instrument. ProtoDUNE-SP embedded photon-detector modules within the APA frames, including ARAPUCA and light-guide technologies, with silicon photomultipliers as sensors (Collaboration et al., 2020, Collaboration et al., 2024). In later dedicated studies, additional X-ARAPUCA devices were installed to separate xenon-shifted light from the total scintillation signal during xenon-doping operation (Collaboration et al., 2024).

3. Cryostat, high voltage, and argon purity

The detector employed a membrane cryostat with inner dimensions of $0.3$8, approximately $0.3$9 of insulation, and an average heat leak of about $7~\mathrm{GeV}/c$0 (Collaboration et al., 2021). It held about $7~\mathrm{GeV}/c$1 of liquid argon, corresponding to roughly $7~\mathrm{GeV}/c$2, at temperatures between $7~\mathrm{GeV}/c$3 and $7~\mathrm{GeV}/c$4 (Collaboration et al., 2021). The liquid was recirculated through molecular-sieve and copper filters, with two $7~\mathrm{GeV}/c$5 pumps giving a full turnover time of $7~\mathrm{GeV}/c$6 days (Collaboration et al., 2021).

For charge transport, the central issue was the electron lifetime $7~\mathrm{GeV}/c$7, with attenuation described by

$7~\mathrm{GeV}/c$8

To keep charge attenuation below $7~\mathrm{GeV}/c$9 over the $3.6~\mathrm{m}$0 maximum drift, ProtoDUNE-SP required $3.6~\mathrm{m}$1, and it achieved substantially longer values in operation (Collaboration et al., 2021). A dedicated later study compared purity monitors with TPC-based measurements using cosmic muons and found that for extended periods on the timescale of weeks the drift electron lifetime was above $3.6~\mathrm{m}$2 using both systems; purity monitors occasionally observed lifetimes above $3.6~\mathrm{m}$3 (Collaboration et al., 11 Jul 2025).

ProtoDUNE-SP instrumented the cryostat with three ICARUS-style purity monitors mounted at heights of about $3.6~\mathrm{m}$4, $3.6~\mathrm{m}$5, and $3.6~\mathrm{m}$6 in the beam-left corner (Collaboration et al., 11 Jul 2025). The purity-monitor method measured the ratio $3.6~\mathrm{m}$7 from photoelectrons released by UV flashes, while TPC-based methods used CRT-matched through-going muons and pure TPC cathode-crossing muons, fitting calibrated $3.6~\mathrm{m}$8 most-probable values versus drift time to extract $3.6~\mathrm{m}$9 (Collaboration et al., 11 Jul 2025). The two methods agreed within uncertainties, with $-180~\mathrm{kV}$0 over common periods (Collaboration et al., 11 Jul 2025).

The purity studies also established spatial structure in the cryogenic performance. Because liquid argon was extracted from the beam-left side and returned beneath beam-right, the beam-right side was on average cleaner than beam-left; a vertical stratification with top $-180~\mathrm{kV}$1 middle $-180~\mathrm{kV}$2 bottom was also observed (Collaboration et al., 11 Jul 2025). This is significant because it shows that electron lifetime in kiloton-scale LArTPCs is not only a global scalar metric but also a spatial field linked to recirculation geometry and thermal transport.

4. Electronics, DAQ, and reconstruction

ProtoDUNE-SP used cold front-end electronics mounted directly on the APAs. Front End Motherboards (FEMBs) digitized $-180~\mathrm{kV}$3 channels each using front-end and ADC ASICs immersed in the liquid argon, with continuous sampling at $-180~\mathrm{kV}$4 (Collaboration et al., 2021). Across the detector, $-180~\mathrm{kV}$5 wires were instrumented with low electronic-noise pre-amplifier and digitization ASICs integrated into cryogenic FEMBs (Adams et al., 2020).

The warm DAQ architecture combined two TPC readout paths: an ATCA-based RCE system for five APAs and a PCIe-based FELIX system for one APA (Sipos, 2018). Fermilab’s artDAQ provided the dataflow software, while custom timing and trigger electronics distributed synchronization and enforced backpressure (Sipos, 2018). The DAQ paper described an aggregate front-end input of about $-180~\mathrm{kV}$6, reduced by compression and triggered readout windows to a sustained output bandwidth of order $-180~\mathrm{kV}$7–$-180~\mathrm{kV}$8 to CERN EOS storage (Sipos, 2018). In the FELIX path specifically, one APA with $-180~\mathrm{kV}$9 wires generated a total link rate of $500~\mathrm{V/cm}$0, with $500~\mathrm{V/cm}$1 trigger windows at $500~\mathrm{V/cm}$2 corresponding to about $500~\mathrm{V/cm}$3 per event before compression (Borga et al., 2018).

Offline signal processing converted raw waveforms into calibrated charge information. The first performance paper described pedestal estimation, correction of “sticky” ADC codes, AC-coupling tail removal, correlated-noise subtraction within FEMB channel groups, and two-dimensional Wiener-filtered FFT deconvolution using detector-response kernels derived from field and electronics response simulations (Collaboration et al., 2020). Gaussian hit finding then provided the inputs for pattern recognition.

Pandora supplied the primary event reconstruction framework (Collaboration et al., 2022). In ProtoDUNE-SP, PandoraCosmic and PandoraTestBeam chains were combined with cosmic-ray tagging, drift-volume stitching, event slicing, and a slice-identification BDT. In simulated data, the reconstruction and identification efficiency for triggered test-beam particles was above $500~\mathrm{V/cm}$4 for the majority of particle-type and beam-momentum combinations; for simulated $500~\mathrm{V/cm}$5 charged pions and protons, the reported efficiencies were $500~\mathrm{V/cm}$6 and $500~\mathrm{V/cm}$7, respectively (Collaboration et al., 2022).

Complementary ML-based reconstruction was also developed. A convolutional-neural-network algorithm classified energy deposits and reconstructed particles as track-like or shower-like and identified Michel electrons, with performance consistent between data and simulation (Collaboration et al., 2022). ProtoDUNE-SP therefore served not only as a detector prototype but also as a large-scale benchmark for LArTPC reconstruction methodologies spanning deterministic pattern recognition and learned classifiers.

5. Commissioning and achieved detector performance

The electronics program required large-scale production quality control, careful integration into the APAs, and detector-wide commissioning. That effort achieved a working electronics channel percentage of $500~\mathrm{V/cm}$8—$500~\mathrm{V/cm}$9 of $2.3~\mathrm{m}\times 5.984~\mathrm{m}$0 channels in total—and the reported operating performance exceeded expectations (Adams et al., 2020). This figure is central to ProtoDUNE-SP’s historical importance, because it established that a full-size single-phase LArTPC could reach near-complete cold-electronics operability at cryogenic temperature.

The first beam-performance paper quantified the achieved noise, calibration stability, and photon-detector performance. After filtering, the equivalent noise charge was reported as $2.3~\mathrm{m}\times 5.984~\mathrm{m}$1 on the collection plane and $2.3~\mathrm{m}\times 5.984~\mathrm{m}$2 on the induction planes, well below the DUNE requirement of less than $2.3~\mathrm{m}\times 5.984~\mathrm{m}$3 ENC (Collaboration et al., 2020). Channel gain variations over eight months were below $2.3~\mathrm{m}\times 5.984~\mathrm{m}$4, and unresponsive channels amounted to $2.3~\mathrm{m}\times 5.984~\mathrm{m}$5 of the total, again below the design limit (Collaboration et al., 2020). For photon detection, single-photoelectron spectra were clearly resolved, timing resolution from double-LED pulses was $2.3~\mathrm{m}\times 5.984~\mathrm{m}$6, and an all-ARAPUCA system was projected to yield $2.3~\mathrm{m}\times 5.984~\mathrm{m}$7 photons/MeV, exceeding the DUNE specification of more than $2.3~\mathrm{m}\times 5.984~\mathrm{m}$8 photons/MeV (Collaboration et al., 2020).

Commissioning also established the stability of large-scale cryogenic and high-voltage operation. ProtoDUNE-SP reported high-voltage uptime of at least $2.3~\mathrm{m}\times 5.984~\mathrm{m}$9, DAQ stability at the beam target of $G$0, and overall data-taking efficiency of at least $G$1 of the beam window (Collaboration et al., 2021). At the same time, the run exposed operational issues that were technically consequential for DUNE design practice: impurity spikes from filter saturation, recirculation pump failures, HV “streamers,” and local field distortions associated with unpowered electron diverters on one side of the detector were all identified and managed during operation (Collaboration et al., 2021, Collaboration et al., 2020).

A frequent oversimplification is to treat ProtoDUNE-SP performance numbers as if they were obtained in an idealized environment. In fact, the detector operated on the surface with substantial space-charge effects; distortions up to $G$2 and field variations up to $G$3 were mapped and corrected using UV-laser data and cathode-crossing muons (Collaboration et al., 2020). The significance of the performance record therefore lies not only in meeting specifications, but in doing so under conditions more complex than those anticipated for the underground far detector.

6. Beam program, calibration campaigns, and physics outputs

ProtoDUNE-SP sat on CERN’s H4-VLE tertiary beam line and was exposed to mixed hadron and lepton beams in the range $G$4 to $G$5, with upstream time-of-flight, Cherenkov, and tracking instrumentation supplying per-particle momentum and PID information (Collaboration et al., 2020). That beam program transformed the detector into a calibration and hadronic-interaction facility for argon, complementing its role as a hardware prototype.

Among the most consequential results were the first measurements of hadron-argon inelastic cross sections in the DUNE-relevant sub-GeV regime. Using selected $G$6 and proton samples from the $G$7 beam data, ProtoDUNE-SP reported the first measurement of the total inelastic cross sections for $G$8-Ar in the $G$9–$U$0 kinetic-energy range and for $U$1-Ar below $U$2 (Collaboration et al., 14 Nov 2025). Representative unfolded values were $U$3 at $U$4 and $U$5 at $U$6 for $U$7-Ar, and $U$8 at $U$9 and $0.77~\mathrm{kt}$00 at $0.77~\mathrm{kt}$01 for $0.77~\mathrm{kt}$02-Ar (Collaboration et al., 14 Nov 2025). A subsequent exclusive-channel analysis reported the first measurements of $0.77~\mathrm{kt}$03-argon absorption and charge exchange in the $0.77~\mathrm{kt}$04–$0.77~\mathrm{kt}$05 range, with total inelastic cross sections of $0.77~\mathrm{kt}$06, $0.77~\mathrm{kt}$07, and $0.77~\mathrm{kt}$08 at $0.77~\mathrm{kt}$09, $0.77~\mathrm{kt}$10, and $0.77~\mathrm{kt}$11, respectively (Collaboration et al., 17 Nov 2025).

ProtoDUNE-SP also supported low-energy calibration and rare-event-reconstruction studies. The Michel-electron analysis selected low-energy electrons from stopping cosmic muons with a purity of $0.77~\mathrm{kt}$12 and showed that, after addition of lost energy using Monte Carlo simulation, the energy resolution improved from about $0.77~\mathrm{kt}$13 to $0.77~\mathrm{kt}$14 at $0.77~\mathrm{kt}$15 (Collaboration et al., 2022). For proton-decay-motivated kaon identification, the detector selected $0.77~\mathrm{kt}$16 kaon candidates from approximately $0.77~\mathrm{kt}$17k beam triggers at $0.77~\mathrm{kt}$18–$0.77~\mathrm{kt}$19, with total selection efficiency $0.77~\mathrm{kt}$20 and sample purity $0.77~\mathrm{kt}$21; the selected candidates covered the expected low-energy range for $0.77~\mathrm{kt}$22 from $0.77~\mathrm{kt}$23 in DUNE (Collaboration et al., 9 Oct 2025).

A separate operational campaign investigated xenon doping of liquid argon on kiloton scale. From February to May 2020, xenon was injected up to a concentration of $0.77~\mathrm{kt}$24 in a detector containing $0.77~\mathrm{kt}$25 of total liquid argon and affected by a $0.77~\mathrm{kt}$26 nitrogen contamination (Collaboration et al., 2024). The ratio of xenon light to total light was measured to be about $0.77~\mathrm{kt}$27 at $0.77~\mathrm{kt}$28 xenon, light-collection uniformity improved for the anode-mounted photon-detection system, and no significant change in collected charge was observed (Collaboration et al., 2024). The result is significant because it demonstrated, at ProtoDUNE-SP scale, that xenon doping can flatten spatial response and recover scintillation light lost to nitrogen contamination without degrading TPC charge readout.

Taken together, these beam, calibration, and detector-response studies show that ProtoDUNE-SP became more than a construction prototype. It provided a modern argon-target dataset for hadronic interactions, validated reconstruction strategies for low-energy electrons and kaons, and tested photon-detection concepts under realistic kiloton-scale conditions. In DUNE terms, its legacy is both infrastructural and phenomenological: it established that the single-phase LArTPC architecture could be built and operated at full component scale, and it supplied measurements and methods that directly constrain the systematic foundations of the future far detector.