On this page
Most particle detectors record collisions through a chain of conversions: scintillation light bouncing off photomultipliers, Cherenkov rings on a wall of phototubes, energy depositions in calorimeter cells. Each step trades information for practicality. The original bubble chamber was different — it recorded the actual three-dimensional trajectories of charged particles in photographic detail, every kink, vertex and curl visible. The cost was equally clear: bubble chambers had to be photographed, the film developed and scanned by humans, and the duty cycle was inevitably terrible.
A liquid argon time projection chamber, or LArTPC, is the bubble chamber reborn in electronics. It records full three-dimensional images of every event with sub-centimetre resolution, but instead of bubbles in superheated liquid it tracks free electrons drifting through cold argon, read out at thousands of frames per second by wires or pixels. The result is the high information content of a bubble chamber image with the speed and triggerability of a modern electronic detector. The technology now underpins the present and future of accelerator-based neutrino physics: MicroBooNE, ICARUS, SBND in the Short-Baseline Neutrino programme at Fermilab, and DUNE, the long-baseline flagship being built deep underground in South Dakota.
This post is about how a LArTPC works, why argon, and what makes the technology so well matched to neutrino physics.
The basic principle
When a charged particle crosses a volume of liquid argon, it does two things at once. It ionises argon atoms along its path, knocking electrons free. And it excites argon atoms that subsequently emit ultraviolet scintillation photons at a wavelength of 128 nanometres, predominantly through the de-excitation of excimer states. Both signals can be read out, and combining them gives the detector its power.
The scintillation light arrives essentially instantaneously — within tens of nanoseconds — and is detected by photon detectors mounted around the active volume. This light serves as the timing signal, fixing the moment at which the interaction took place.
The ionisation electrons, meanwhile, drift slowly through a uniform electric field of about 500 volts per centimetre toward an anode plane several metres away. The drift speed in liquid argon at that field strength is roughly 1.6 millimetres per microsecond, so an electron originating several metres from the anode takes a few milliseconds to arrive. The anode is instrumented with closely spaced wires — typically at 3 millimetre pitch — or, in modern designs, with pixel pads. As the cloud of drifted electrons crosses the anode, it induces signals on the wires that record the projection of the track perpendicular to the drift direction. Combined with the precisely known drift velocity and the from the scintillation light, the third dimension along the drift axis is recovered.
The result is a complete three-dimensional picture: every track, vertex and shower in the event, with millimetre-scale spatial resolution and the calorimetric energy measurement that comes from counting drifted electrons. A muon’s straight track, an electron’s electromagnetic shower with its characteristic conversion radius, a proton’s stubby ionising path, a pion’s hard interaction in flight — all are visible directly, the way a physicist would draw them on a blackboard.
Why argon, and what makes it hard
The choice of liquid argon as both target and medium was not obvious in advance. Several properties combine to make it work.
Argon is chemically inert, which means it does not attack the materials of the cryostat or readout, and it does not form long-lived radicals or compounds that would interfere with electron drift. It is abundant and cheap, because it makes up nearly one percent of the atmosphere and is produced as a byproduct of industrial air separation; a 10-kiloton detector becomes economically feasible. It is dense enough as a liquid — 1.4 g/cm³ — that thousand-ton volumes fit in reasonably sized cryostats. And it is transparent to its own 128 nanometre scintillation light over metre-scale distances.
The technical challenge is purity. Free electrons drifting through liquid argon are easily captured by oxygen, water, nitrogen and other impurities. A typical contamination of one part per billion of oxygen reduces the electron-lifetime to a few hundred microseconds — far too short to allow drift across a multi-metre detector. The required purity is below 100 parts per trillion of oxygen-equivalent contamination. Reaching that level requires elaborate filtration through molecular sieves and copper getters, and maintaining it requires meticulous control over outgassing of every component immersed in the argon. This was the central engineering obstacle that took two decades to solve at the scale of a multi-kiloton detector; ICARUS at LNGS in Italy was the first to demonstrate it, and ProtoDUNE at CERN later confirmed it at full DUNE drift length.
A second challenge is high voltage. To establish the uniform drift field over several metres, the cathode must sit at hundreds of kilovolts. Sustaining that voltage cleanly in the cold liquid, without discharges that would damage the detector, requires careful engineering of feedthroughs and field-shaping rings. The 3.5-metre drift in DUNE corresponds to a cathode voltage of about 180 kilovolts; the longer-drift dual-phase concept, before DUNE retired it, aimed at twice that.
What the technology is good for
The detector’s three-dimensional imaging makes it especially well matched to neutrino physics, where event topologies carry most of the information.
The first and most direct benefit is particle identification. An electron neutrino interacting in liquid argon produces an electron that initiates an electromagnetic shower, distinguishable from a single isolated photon by the conversion radius at the start of the track. A muon neutrino produces a clean muon track that can extend tens of metres. A neutral-current event with a single neutral pion, which would mimic an electron-neutrino signal in a Cherenkov detector, is in LArTPC visible as a separated photon shower starting some millimetres from the vertex. This rejection of the most dangerous backgrounds is what made LArTPC the technology of choice for DUNE’s CP-violation measurement, where electron-neutrino appearance must be cleanly separated from neutral-current pion backgrounds.
The second benefit is low energy threshold. Calorimetric reconstruction by counting drifted electrons works down to thresholds of tens of MeV, where Cherenkov detectors lose efficiency. This makes LArTPC well-suited to supernova neutrino bursts, where the typical energies are around 10–20 MeV and most events would be electron-neutrino charged-current scattering on argon producing a few-MeV electron plus a recoiling potassium nucleus.
The third benefit is the purely electronic readout — no film, no scanning, no humans in the loop. The data come out as digitised waveforms on each wire, processed through reconstruction pipelines that increasingly use deep learning to identify and classify events. The Pandora and SparseConvNet frameworks developed for MicroBooNE and SBND have become the de facto standard for LArTPC reconstruction.
The road from ICARUS to DUNE
The history is short, by particle-physics standards. ICARUS proposed the concept in 1977 and ran its 600-ton detector at LNGS from 2010 to 2014, demonstrating that the technology worked at the kiloton scale and producing the first oscillation results with the CNGS neutrino beam from CERN. MicroBooNE took data from 2015 to 2021 at Fermilab and produced the definitive tests of the MiniBooNE excess, ruling out the simplest sterile-neutrino explanations. SBND came online in 2024 as the near detector of the Short-Baseline Neutrino programme. ICARUS was refurbished and moved to Fermilab to serve as the SBN far detector.
In parallel, the ProtoDUNE detectors at CERN — two 720-ton modules with single- and dual-phase readout — validated the engineering for DUNE’s full 10-kiloton modules. The first DUNE far-detector module is under construction in the Sanford Underground Research Facility, with first beam from the LBNF beam at Fermilab expected later this decade. DUNE will host four 10-kiloton modules in total, all using the LArTPC concept, addressing CP violation in the lepton sector, supernova bursts, and proton decay.
Summary
A liquid argon time projection chamber records every charged-particle trajectory in three dimensions by drifting ionisation electrons through a multi-metre volume of ultra-pure argon, with scintillation light providing the start-of-event time. The technology combines the bubble chamber’s full event imaging with the speed and triggerability of electronic readout, and resolves event topologies — separating electrons from photons, identifying neutral pions, calorimetrically measuring energy — at a level no other neutrino target medium achieves at comparable cost. From the ICARUS pioneer in Gran Sasso to MicroBooNE and SBND at Fermilab and the four 10-kiloton DUNE modules being built in South Dakota, LArTPC has become the workhorse technology for the next generation of accelerator-based neutrino physics, and the central tool for precision oscillation measurements and a future supernova burst.