Before KATRIN: The Mainz and Troitsk Tritium Spectrometers

By 1990, the question of the neutrino’s absolute mass was urgent. Pauli had postulated the particle as nearly massless in 1930. Decades of beta-decay endpoint experiments had pushed upper limits on the electron-neutrino mass from keV to tens of eV. The 1980 ITEP claim of a positive 30-eV mass had created controversy that cast doubt on the entire field. The community needed an experiment with higher precision and better systematic control.

Two groups stepped up. The first, at the University of Mainz in Germany, led by Christian Weinheimer and Ernst Otten, began constructing a tritium beta-decay experiment using a new electrostatic-spectrometer architecture. The second, at the Institute for Nuclear Research in Troitsk, near Moscow, led by Vladimir Lobashev, was building a similar apparatus independently.

By the late 1990s, both experiments were running. By the early 2000s, both had pushed the upper bound on the electron-neutrino mass to approximately 2 eV — an order of magnitude tighter than any previous result. Their measurements held the world record on direct neutrino mass for nearly twenty years, until the KATRIN experiment took over with its 2019 data release.

The Mainz and Troitsk programmes are now historical, but their methodology and equipment are the direct ancestors of every modern direct-mass measurement. KATRIN inherited the MAC-E architecture, much of the analysis machinery, and many of the personnel. Project 8 builds on the same conceptual foundation. This post is the prequel: how the Mainz and Troitsk experiments worked, what they measured, and why their successors have moved on.

Why tritium

Tritium is not the only beta emitter, but it has a combination of properties that no other isotope matches.

Low endpoint energy (18.574 keV). The fractional distortion of the spectrum from a finite neutrino mass scales inversely with endpoint energy. For tritium, a 1-eV neutrino mass produces a 1/18,574 = 0.005% distortion at the endpoint — small but measurable. For higher-Q-value beta emitters, the fractional distortion is smaller and harder to extract.

Super-allowed nuclear transition. The decay ${}^3\text{H}\to {}^3\text{He}$ has a simple nuclear matrix element that is calculated to better than 1% precision. The shape of the spectrum can therefore be predicted theoretically and compared against measurement. With more complex isotopes, nuclear-structure uncertainty would dominate.

Reasonable half-life (12.3 years). Long enough that you can store the source for years; short enough that the decay rate is high enough to give substantial statistics.

Available in molecular and atomic forms. Molecular T₂ can be condensed and used as a thin film, or kept as a gas in a windowless tube. Atomic tritium is harder to produce stably but offers fewer systematic complications (no molecular final-state distribution).

The Mainz and Troitsk experiments both used molecular T₂. KATRIN does too. The transition to atomic tritium awaits Project 8.

The MAC-E filter principle

The dominant systematic in tritium endpoint measurements is the energy resolution of the spectrometer. To resolve a sub-eV mass effect against an 18.6-keV endpoint, you need an energy resolution of order 1 eV — a relative precision of $5\times 10^{-5}$.

A conventional electrostatic spectrometer would require very small angular acceptance to achieve this resolution, which would mean very low signal rates. The trick that made Mainz and Troitsk work was the MAC-E filter — Magnetic Adiabatic Collimation with Electrostatic filter.

The principle: an electron leaving the tritium source enters a strong magnetic field (~5 T) that constrains it to spiral along the field lines. As it travels into a region of much weaker field (~0.001 T), the spiral expands while conserving the magnetic moment $\mu =E_{\perp }/B$. This means the transverse energy decreases as the field drops, while the longitudinal energy is preserved. By the time the electron reaches the central spectrometer, its momentum is almost entirely along the magnetic field axis.

A retarding electrostatic potential is then applied. Only electrons with longitudinal energy greater than the retarding potential can pass through to the detector. By varying the retarding voltage in small steps, you scan the integrated electron spectrum.

The MAC-E filter combines high acceptance (large solid angle at the source) with high resolution (excellent collimation in the spectrometer). It was originally proposed in the 1970s for nuclear-physics applications and adapted for tritium endpoint measurement in the 1980s.

The Mainz experiment

The Mainz tritium experiment, in the city of Mainz on the Rhine, used a frozen molecular T₂ source (a thin film of solid tritium at 4 K) coupled to a 4-metre-long MAC-E spectrometer. Construction began in the late 1980s; first data taking was in 1991.

The detector was an open multichannel plate at the spectrometer exit, counting transmitted electrons. The integrated spectrum was scanned by changing the retarding voltage in steps of a few hundred millivolts.

The Mainz programme reported a series of upper limits over the 1990s:

• 1992: $m_{\nu }<7.2$ eV (95% C.L.)
• 1995: $m_{\nu }<5.6$ eV
• 2000: $m_{\nu }<2.2$ eV
• 2005 (final): $m_{\nu }<2.3$ eV (95% C.L., combined with Troitsk)

The bound was systematic-limited — the dominant uncertainty came from the molecular final-state distribution of the daughter ${}^3$He–T molecule, which had to be calculated rather than measured.

The experiment shut down operations in 2001, with the final analysis published over the following years. The team distributed to the KATRIN project, which was being designed in Karlsruhe with substantially larger tanks and tighter systematic control.

The Troitsk experiment

Troitsk, the Institute for Nuclear Research site near Moscow, ran a similar but distinct apparatus. The Troitsk MAC-E was about 7 metres long, slightly larger than Mainz, and used a windowless gaseous tritium source (a tube of T₂ gas with continuous pumping) rather than a frozen film. The choice gave somewhat different systematic uncertainties — gaseous-source backgrounds rather than condensed-film thermal artifacts.

The Troitsk results, reported in parallel with Mainz:

• 1995: $m_{\nu }<4.4$ eV (95% C.L.)
• 1999: $m_{\nu }<2.5$ eV
• 2011 (final): $m_{\nu }<2.05$ eV (95% C.L.)

Like Mainz, Troitsk reached the systematic floor and recognized that further improvement required a fundamentally larger and better-controlled experiment. Operations effectively ended in 2008. Some Troitsk physicists joined KATRIN; others remained at the Russian institute working on related projects.

A particular Troitsk subtlety: in early data, the experiment occasionally reported a “step” in the electron spectrum near the endpoint that could not be cleanly explained by mass effects. The “Troitsk anomaly” generated theoretical interest for several years. By the late 1990s, refined analyses attributed it to a systematic effect related to the tritium gas pressure, and the anomaly was reclassified as instrumental.

Combined Mainz-Troitsk legacy

The two experiments were independent but used essentially the same MAC-E architecture, the same target isotope, and the same analysis approach. Their consistent results — both reporting $m_{\nu }<2$ eV by the early 2000s — provided a robust cross-check. The combined Mainz-Troitsk bound of approximately 2 eV held the world record from 2003 until KATRIN published its 2019 result of $m_{\nu }<1.1$ eV.

The MAC-E architecture itself was vindicated. KATRIN, using a 23-metre spectrometer and an order-of-magnitude larger tritium source, has pushed the bound to the current 0.45 eV (90% C.L., 2024 data release). It will reach approximately 0.2 eV with full data and improved systematic control.

The technology has limits. KATRIN’s design is essentially the limiting case of MAC-E: any larger and the practical engineering becomes prohibitive. To go below 0.2 eV requires fundamentally different architecture. Project 8’s cyclotron radiation emission spectroscopy is the leading candidate, replacing the electrostatic filter with frequency measurement of trapped electrons.

What ended at Mainz

Mainz shut down its experimental tritium program in 2001 to consolidate effort into KATRIN, which the Karlsruhe Institute of Technology was beginning to design. The Mainz spectrometer itself was preserved and is now part of the KATRIN test infrastructure, used for quick calibration tests rather than headline measurements.

Christian Weinheimer, who led Mainz’s experimental programme, transitioned to KATRIN leadership. The Mainz expertise — particularly in solid-tritium-source preparation, molecular final-state calculations, and MAC-E-filter performance optimisation — fed directly into KATRIN’s design. KATRIN inherits Mainz’s institutional knowledge.

What ended at Troitsk

Troitsk continued running until 2008 and published its final analysis in 2011. The Troitsk team, like Mainz, joined KATRIN. The Troitsk facility itself was repurposed for other physics: tritium handling expertise was used in calibration source development for other experiments, and parts of the spectrometer were dismantled.

A subset of Troitsk physicists continued working on alternative neutrino mass approaches. The Troitsk-Nu-Mass programme proposed using gaseous tritium with much higher source pressures, accepting the resulting energy resolution loss in exchange for higher statistical power. The programme has not been competitive with KATRIN but continues as a complementary effort.

A final reflection

The Mainz and Troitsk programmes occupy a specific historical role in neutrino physics. They were not Nobel-worthy in the way Super-Kamiokande’s atmospheric oscillation discovery was. They didn’t make a positive detection. Their primary published results are upper limits — useful, refining, but not paradigm-shifting.

But they prepared the ground. Without the careful demonstration that MAC-E filters can deliver sub-eV mass sensitivity, without the systematic-uncertainty work that exposed the molecular final-state problem, without the personnel pipeline that fed into KATRIN, the modern direct-mass measurement programme would have looked very different. KATRIN’s 0.45 eV result of 2024 rests directly on the foundation that Mainz and Troitsk built.

The pattern is common in fundamental physics: a generation of experiments approaches a frontier, demonstrates the technique, identifies the limiting systematics, and hands the work to a successor that breaks through. Mainz and Troitsk were that generation for direct neutrino mass measurement. KATRIN is the successor. Project 8 will be the next.

When the absolute neutrino mass is eventually measured — whether at 0.04 eV at Project 8, at sub-meV precision through cosmology, or through some other approach not yet developed — the lineage will run back through KATRIN to Mainz and Troitsk. The 2-eV bound that the two German and Russian groups established in the 1990s and 2000s will be the textbook entry under “first generation of high-precision direct-mass experiments”. A modest entry, but a foundational one.