nEXO: Five Tonnes of Liquid Xenon as a Double-Beta Telescope

When EXO-200 began operating in 2010 at the Waste Isolation Pilot Plant in New Mexico, it was the first time anyone had built a liquid-xenon time projection chamber specifically to look for neutrinoless double beta decay. The 200 kg of enriched xenon-136 served as both the source and the detector, with ionisation electrons drifted to an anode plane and scintillation light read out by avalanche photodiodes. The experiment delivered the first competitive limit on the $0\nu \beta \beta$ half-life of ${}^{136}$Xe in 2014, refined it through subsequent runs, and established the liquid-xenon-TPC approach as one of the leading technologies for the next generation of $0\nu \beta \beta$ search.

nEXO — “next EXO” — is the successor experiment now in advanced design. Its target is 5 tonnes of enriched xenon-136 in a single homogeneous TPC, with a projected half-life sensitivity of approximately $10^{28}$ years and an effective Majorana mass reach around 5-20 meV. Together with LEGEND and KamLAND-Zen, nEXO forms a three-pronged attack on the inverted-mass-ordering parameter space, with each experiment using a fundamentally different detector technology and offering different dominant systematics.

This post is about how nEXO works, why a liquid-xenon TPC turns out to be such a good $0\nu \beta \beta$ detector, and what makes it complementary to the germanium and xenon-in-scintillator approaches.

Why liquid xenon

Xenon-136 has the most attractive single combination of properties of any $0\nu \beta \beta$ candidate isotope. Its $2\nu \beta \beta$ half-life of $2.2\times 10^{21}$ years is the longest known, meaning the background from the normal-mode decay in the $0\nu \beta \beta$ signal region is among the lowest of any candidate. Its Q-value at 2,458 keV sits above most natural radioactivity gamma lines, giving a relatively clean spectral region. Natural xenon contains 8.9% of ${}^{136}$Xe — comparable to or better than germanium’s 7.8% of ${}^{76}$Ge — and gas centrifugation can enrich it to over 90% at industrial scale and modest cost.

Liquid xenon adds a distinctive physical advantage: it is simultaneously a heavy, dense, transparent target. The density of 3.06 g/cm³ at the liquid temperature of 165 K provides high source mass per detector volume. Xenon is transparent to its own 175 nm scintillation light over meter-scale distances. It supports drift of free electrons over multi-metre baselines with the same kind of ultra-pure conditions LArTPCs require — the impurity level of oxygen and water has to be kept below parts per billion. And the resulting detector medium delivers excellent energy resolution and particle identification simultaneously.

How a Xe-136 TPC sees a 0νββ event

A $0\nu \beta \beta$ event in the liquid xenon produces two electrons whose combined kinetic energy equals the Q-value of 2,458 keV. Both electrons stop within a few millimeters of their production point, depositing their full energy and producing two distinct signals.

The first signal is prompt scintillation, emitted within a few nanoseconds of the energy deposit. The xenon scintillation comes from de-excitation of singlet and triplet excimer states formed by the ionising electrons; the emission wavelength is 175 nm, in the far ultraviolet. Silicon photomultipliers (SiPMs) sensitive to vacuum-ultraviolet light, arrayed on the cathode and barrel of the TPC, read out the scintillation light to provide the absolute event time $t_0$.

The second signal is ionisation. Electrons freed by the same energy deposit drift through a uniform electric field of about 400 V/cm toward an anode plane many tens of centimetres away. The drift velocity in liquid xenon is approximately 1.7 mm/μs at this field, so multi-metre drift takes a few milliseconds. The anode plane is segmented into closely spaced pads that record the spatial pattern of arriving electrons; combined with the drift time and the $t_0$ from the scintillation, the result is a full three-dimensional image of the event.

The two readouts together provide three crucial capabilities:

• Energy reconstruction combining ionisation and scintillation, with energy resolution around 1% FWHM at the Q-value — better than KamLAND-Zen’s scintillator-based resolution and comparable to LEGEND’s germanium semiconductor resolution.
• Three-dimensional position of every event, allowing fiducial volume cuts to reject events near the detector boundaries where surface backgrounds concentrate.
• Particle identification through the ionisation-to-scintillation ratio. Alpha particles produce a different ratio than electrons, distinguishing alpha-induced backgrounds from $\beta \beta$-like signals.

What EXO-200 demonstrated

EXO-200 ran from 2011 to 2018 with 200 kg of enriched Xe-136. Its $0\nu \beta \beta$ search reached a half-life lower limit of $T_{1/2}^{0\nu }>3.5\times 10^{25}$ years at 90% confidence — competitive with the first KamLAND-Zen results and one of the leading limits at the time.

More importantly for nEXO, EXO-200 demonstrated several technical capabilities that the larger experiment would need:

The xenon purification system maintained electronegative-impurity concentrations below the level required for multi-metre drift over years of continuous operation. The integrated purifier circulated the xenon between the active volume and a getter system that removed oxygen, water and nitrogen contaminants.

The light readout with VUV-sensitive avalanche photodiodes at the cathode and barrel was the prototype for the SiPM arrays planned in nEXO. The detection efficiency reached ~7% of emitted photons across the active volume.

The anode segmentation with crossed wire planes provided the spatial reconstruction. nEXO replaces this with a pad-based readout for higher channel density and better resolution.

The energy reconstruction combining ionisation and scintillation achieved 1.4% resolution at the Q-value — establishing that the combined-readout approach worked in practice.

EXO-200 also produced the most precise measurement of the ${}^{136}$Xe two-neutrino half-life, $T_{1/2}^{2\nu }=2.165\times 10^{21}$ years — directly relevant to the background prediction for any larger experiment.

Scaling to nEXO

nEXO is approximately 25 times the mass of EXO-200 but is not simply a scaled-up version. Several design choices reflect lessons from EXO-200 and from the broader large-TPC experience:

Single-phase rather than dual-phase. EXO-200 used a single-phase liquid xenon TPC, and nEXO retains the choice rather than going to a gas-phase amplification region as some other large xenon experiments do. The single-phase configuration eliminates the gas-liquid interface and its associated instabilities, at the cost of lower gain on the ionisation signal — a trade-off that works for $0\nu \beta \beta$ where the deposited energies are MeV-scale and intrinsic ionisation is already abundant.

SiPM rather than APD readout. Modern silicon photomultipliers tuned for VUV operation provide higher photon detection efficiency than EXO-200’s APDs, and at lower cost per channel. nEXO’s projected design uses tile-mounted SiPM arrays covering both the cathode and the side walls.

Cylindrical geometry. EXO-200 was rectangular; nEXO is cylindrical, providing more uniform drift field and easier fiducialisation. The active volume is roughly 1.3 m in diameter and 1.3 m tall.

Cryostat and shielding. The cryostat is a double-walled vacuum-insulated copper vessel, surrounded by an active water Cherenkov shield similar to LEGEND’s outer veto. The combination provides background rejection at the relevant gamma-ray and muon-induced rates.

The projected sensitivity is approximately $10^{28}$ years half-life after 10 years of operation, corresponding to an effective Majorana mass reach of 5-20 meV depending on the nuclear matrix element used. The full inverted-mass-ordering parameter region is within reach.

Site selection and timeline

The leading candidate site is SNOLAB in Canada, which provides the deepest available muon flux suppression of any major underground laboratory. The depth — 6,000 meters water-equivalent overburden — reduces cosmogenic backgrounds substantially relative to Gran Sasso or even WIPP. Detailed site evaluation is ongoing, with construction expected to begin in the late 2020s.

The schedule envisions detector commissioning in the early 2030s and first physics results within a few years thereafter. A staged approach is being considered, with initial operation at lower xenon mass (potentially 1-2 tonnes) before full 5-tonne loading.

Barium tagging — the path to background-free operation

The most ambitious proposed extension of nEXO is barium tagging: directly identifying the ${}^{136}$Ba daughter atom that every ${}^{136}$Xe double beta decay produces. The barium-tagging idea is that if you can confirm the presence of a barium atom at the location of every candidate event, you can eliminate every background that does not produce a barium atom — and there are essentially no such backgrounds at the relevant rates.

The technique exploits barium’s strong fluorescence at 535 nm under 553 nm laser excitation. A single barium atom extracted from the xenon and trapped electromagnetically can be identified definitively from its fluorescence signal. The proof-of-principle has been demonstrated in cryogenic and warm-trap test setups, with single-atom sensitivity established.

Integrating barium tagging into a multi-tonne TPC is not yet solved. The candidate atom must be located in the active volume, extracted through a thin transfer probe, transported to the trapping region, and held there for fluorescence identification — all while maintaining the TPC’s ultra-pure operating conditions and without producing a re-contamination of the xenon. Multiple groups are pursuing different technical implementations: cryogenic ion trapping, freeze-and-laser-extraction approaches, and selective ionisation followed by mass spectrometry.

If any of these matures, the resulting background suppression would push nEXO’s sensitivity well beyond the $10^{28}$ year scale and into a regime where the relevant systematic becomes the nuclear matrix element rather than the experimental background.

How nEXO fits the program

Three major next-generation $0\nu \beta \beta$ programs are converging on the inverted-mass-ordering parameter space:

• LEGEND-1000 with 1,000 kg of ${}^{76}$Ge in semiconductor crystals immersed in liquid argon, targeting $10^{28}$ years.
• KamLAND-Zen 800 → 8000 with growing amounts of ${}^{136}$Xe dissolved in scintillator at KamLAND, currently at 745 kg, targeting beyond $10^{27}$ years.
• nEXO with 5 tonnes of ${}^{136}$Xe in a TPC, targeting $10^{28}$ years.

The three approaches probe overlapping but distinct parameter regions. LEGEND has the best energy resolution; KamLAND-Zen has the largest isotope mass and lowest absolute background; nEXO offers 3D event reconstruction and the barium-tagging route. The same nuclear physics — the ${}^{136}$Xe and ${}^{76}$Ge nuclear matrix elements — sets the conversion from half-life to effective Majorana mass, and they are correlated through underlying theory. A signal in any one experiment would benefit from independent confirmation in another isotope, where the nuclear physics is independent.

The three programs are designed to be complementary, not competitors. The collective community goal is to reach $|m_{\beta \beta }|\sim 15$ meV — the bottom of the inverted ordering band — across at least two independent isotopes before the end of the 2030s.

Summary

nEXO is the proposed next-generation $0\nu \beta \beta$ experiment using 5 tonnes of liquid xenon enriched in ${}^{136}$Xe as a single-phase time projection chamber. The detector simultaneously reads out ionisation electrons drifted to a segmented anode and prompt scintillation light from the same event, combining them for 1% energy resolution at the 2,458 keV Q-value, 3D event reconstruction, and event-by-event particle identification. Building on EXO-200’s demonstration of the technique at the 200 kg scale, nEXO scales to a half-life sensitivity of $10^{28}$ years and an effective Majorana mass reach of 5-20 meV — into the inverted-mass-ordering region. The proposed extension to barium tagging, in which the ${}^{136}$Ba daughter atom is directly identified by laser fluorescence after extraction from the TPC, could push the experiment to essentially background-free operation regardless of energy resolution. Together with LEGEND-1000 and KamLAND-Zen’s continued expansion, nEXO defines one of three parallel paths into the inverted-mass-ordering parameter space and the answer to whether neutrinos are Majorana particles.