Coherent Captain Mills Detector

• Coherent Captain Mills Detector is a large-volume liquid-argon optical detector that uses TPB wavelength shifting and split PMT coatings to distinguish prompt Cherenkov from delayed scintillation signals.
• Its dual-mode operation enables precise event reconstruction and calibration, yielding high Cherenkov purity and directional sensitivity for sub-MeV electron detections.
• The upgraded CCM200 configuration expands fiducial volume and photocathode coverage, making it a versatile tool for coherent scattering, dark matter, and axionlike particle research.

The Coherent CAPTAIN-Mills (CCM) detector is a large-volume liquid-argon light-collection detector at Los Alamos National Laboratory, operated at the LANSCE Lujan source for coherent-scattering, light dark matter, axionlike particle, and related low-energy rare-event searches. Across its CCM120 engineering configuration and the upgraded CCM200 system, it has been developed as a strictly light-based detector rather than a charge-collection TPC. In its upgraded form, CCM combines high photocathode coverage, TPB wavelength shifting, fast waveform digitization, and a deliberate split between coated and uncoated photomultipliers, enabling operation as a hybrid Cherenkov-plus-scintillation detector in liquid argon (Aguilar-Arevalo et al., 2021, Aguilar-Arevalo et al., 10 Jul 2025, Aguilar-Arevalo et al., 10 Jul 2025).

1. Experimental setting and detector evolution

CCM is deployed at the Los Alamos Neutron Science Center, in connection with the Lujan tungsten target struck by an 800 MeV proton beam. Multiple papers describe the detector as being about 20 m from the target, while later CCM200 descriptions place it 23 m from the proton target and roughly $90^\circ$ off the beam direction. That geometry is central to the program: it favors isotropic decays-at-rest and creates an early beam-synchronous window before slower neutron backgrounds become dominant (Aguilar-Arevalo et al., 2021, Aguilar-Arevalo et al., 10 Jul 2025, Newmark, 8 Jul 2026).

The detector program developed in two principal stages. The 2019 engineering run used the CCM120 configuration, with 120 8-inch photomultipliers and a fiducial mass of about 5 tons, plus an optically isolated veto region. That run established the detector’s basic operating characteristics, produced the first dark-matter and ALP searches with CCM, and exposed limitations associated with contaminated liquid argon and non-optimized shielding (Aguilar-Arevalo et al., 2021, Aguilar-Arevalo et al., 2021, Aguilar-Arevalo et al., 2021). The subsequent CCM200 upgrade expanded the inner instrumentation to 200 PMTs, increased the fiducial volume to 7 tons inside a 10-ton liquid-argon detector, and retained a 3-ton optically isolated veto region (Aguilar-Arevalo et al., 10 Jul 2025).

The surrounding infrastructure is described with substantial passive shielding. Depending on configuration and analysis, papers report about 5 m of steel, 2 m of concrete, and 10 cm of borated polyethylene, or, for CCM200, 6 m steel, 3.5 m concrete, and 5 cm borated polyethylene (Aguilar-Arevalo et al., 2021, Aguilar-Arevalo et al., 10 Jul 2025). This shielding is not ancillary: it defines the usable timing structure for beam-related analyses, suppresses neutron backgrounds, and is repeatedly invoked in sensitivity projections for dark-sector and neutrino measurements (Shoemaker et al., 2021).

2. Cryostat, photosensors, and optical architecture

CCM200 is an upright cylindrical cryostat about 2.58 m in diameter and 2.25 m high, reused from CAPTAIN and filled with 10 tons of liquid argon. Its 7-ton fiducial volume is instrumented by 200 8-inch Hamamatsu R5912-Y002 cryogenic PMTs, giving about 50% photocathode coverage. The arrangement is specified in the thesis literature as five rows of 24 PMTs around the barrel and 40 PMTs on each endcap (Aguilar-Arevalo et al., 10 Jul 2025, Newmark, 8 Jul 2026).

A defining architectural choice is the split optical population. Of the 200 PMTs, 160 are coated with TPB and 40 are left uncoated. The coated PMTs convert the 128 nm vacuum-ultraviolet scintillation light of liquid argon into visible wavelengths, while the uncoated PMTs are comparatively more sensitive to the earliest visible Cherenkov photons. Detector walls are lined with Mylar reflective foils coated in TPB; the optical characterization paper specifies 2.8 $\mu$m TPB on the foils and 2.0 $\mu$m TPB on 80% of the PMTs. The PMTs protrude 6.2 mm into the fiducial volume, giving full photocathode exposure (Aguilar-Arevalo et al., 10 Jul 2025, Aguilar-Arevalo et al., 10 Jul 2025).

Waveforms are digitized with CAEN V1730/V1730S 500 MHz boards, providing 2 ns bins. The PMT timing spread is $\mathcal{O}(1\,\mathrm{ns})$, and a typical single-photoelectron pulse has FWHM $\sim 6$ ns (Aguilar-Arevalo et al., 10 Jul 2025). In the thesis description, the PMTs have an intrinsic electron transit-time spread of about 1.7 ns; this, together with 2 ns timing granularity, is presented as the enabling condition for separating prompt Cherenkov light from delayed scintillation in a high-light-yield medium (Newmark, 8 Jul 2026).

The treatment of argon purity illustrates the evolution of the instrument concept. The 2021 dark-matter paper presents filtered and recirculated liquid argon as a key planned upgrade for CCM200 after the CCM120 engineering run inferred an attenuation length of only about 55.95 cm and contamination at the 10 ppm level (Aguilar-Arevalo et al., 2021). By contrast, the 2025 optical characterization reports CCM operation with commercially supplied argon without additional filtration, with roughly ppm-level impurities of up to 1.95 ppm O$_2$, 2.50 ppm N$_2$, and 0.01 ppm H$_2$O, and argues that a light-only detector can tolerate such conditions and may even benefit from them for Cherenkov/scintillation separation because impurities preferentially absorb VUV/UV scintillation more than visible Cherenkov light (Aguilar-Arevalo et al., 10 Jul 2025). This suggests that CCM’s architecture is unusually robust against purity conditions that would be unacceptable for charge-drift detectors.

3. Hybrid scintillation and Cherenkov operation

CCM’s hybrid character rests on the distinct time and wavelength structure of liquid-argon light. The detector exploits the fact that Cherenkov emission is essentially prompt, while liquid-argon scintillation has a fast singlet component of about 5 ns and a slow triplet component of about 1000 ns. In the CCM optical fit, the preferred scintillation parameters were $R_s = 0.367^{+0.017}_{-0.015}$, $R_t = 0.633^{+0.017}_{-0.015}$, $\mu$0, and $\mu$1, values interpreted as consistent with impurity quenching (Aguilar-Arevalo et al., 10 Jul 2025, Aguilar-Arevalo et al., 10 Jul 2025).

The timing logic is detector-specific rather than generic. In the prompt region of an event, defined as

$\mu$2

with $\mu$3 the reconstructed event start time, uncoated PMTs are primarily sensitive to visible Cherenkov photons. This interval is explicitly named the “Cherenkov enhanced” region, while $\mu$4 is the “scintillation enhanced” region (Aguilar-Arevalo et al., 10 Jul 2025). The separation works because uncoated PMTs do not immediately detect wavelength-shifted scintillation light; most 128 nm scintillation photons must first reach TPB-coated surfaces, undergo wavelength shifting, and then propagate back to a PMT, introducing delay.

The optical model incorporates Cherenkov production through the Frank–Tamm relation,

$\mu$5

and emphasizes that accurate prediction requires a realistic wavelength-dependent refractive index, particularly near the ultraviolet resonance where a divergent Sellmeier form is inadequate. In the fitted CCM model, Cherenkov light contributes more than 10% of the total predicted light up to about 5 ns after event start time, and more than 50% for $\mu$6 ns in the full detector (Aguilar-Arevalo et al., 10 Jul 2025). The later 2026 beam-dump thesis therefore presents CCM as the first demonstration of a hybrid Cherenkov and scintillation optical detector at such a facility (Newmark, 8 Jul 2026).

4. Calibration, differentiable simulation, and event reconstruction

The central calibration source for CCM200 is $\mu$7Na. The detector papers describe a 3 $\mu$8Ci source encapsulated in about 1 mm of stainless steel and inserted at the detector midline or origin while the accelerator is off. The source provides a 1.275 MeV gamma ray and, in the $\mu$9 branch, two back-to-back 0.511 MeV annihilation gamma rays. These gamma rays Compton scatter in liquid argon to produce sub-MeV electrons, which are the actual emitters of scintillation and Cherenkov light. Because Cherenkov emission in liquid argon turns on only above a kinetic-energy threshold of roughly 0.2 MeV, the resulting Compton electrons are well matched to studies of the onset of detectable Cherenkov light (Aguilar-Arevalo et al., 10 Jul 2025, Newmark, 8 Jul 2026).

Event selection is timing-driven. The Cherenkov paper reconstructs event start times by requiring at least 3 PE in a 2 ns sliding window, rejects high-charge cosmic muons in the 16 $\mu$0s acquisition window, removes events with less than 2.2 $\mu$1s separation from neighbors, and imposes a reconstructed radial-position cut of 25 cm from the detector center. The total charge in the first 90 ns shows peaks around 50 PE and 100 PE; the analysis selects events within $\mu$2 PE of the higher-charge peak to isolate the $\mu$3 decay region, with expected random background contamination of $\mu$4 after cuts (Aguilar-Arevalo et al., 10 Jul 2025).

A major technical contribution of CCM200 is the differentiable, re-weightable optical photon Monte Carlo built on Geant4. The model spans more than 20 parameters, including scintillation timing, absorption, index of refraction, scattering, TPB response, and PMT timing. Rather than rerunning full photon transport at each parameter point, the simulation stores photon history information such as whether the photon originated as scintillation or Cherenkov light, its wavelength, path lengths before and after wavelength shifting, delay from scintillation physics, and PMT post-pulsing response. Fits used 191 of 200 PMTs over the interval from $\mu$5 ns to $\mu$6s, optimized with L-BFGS-B and uncertainties estimated from highest posterior density spreads across individually fitted PMTs (Aguilar-Arevalo et al., 10 Jul 2025).

The fitted optical parameters include a 17.42 cm absorption length at 128 nm, a preferred 98.25 cm absorption length in the 300–400 nm region, a Rayleigh scattering length of $\mu$7 cm at 128 nm, and a Mie scattering length of $\mu$8 cm at 200 nm. The study presents this as the first characterization of Mie scattering in liquid argon (Aguilar-Arevalo et al., 10 Jul 2025). In later reconstruction work, a GraphNeT-based transformer graph neural network used charge, time, PMT geometry, and coating information to infer event vertices with roughly 5 cm resolution in each spatial coordinate and an overall three-dimensional resolution of about 7.9 cm at $\mu$9. Energy reconstruction based on position-dependent charge scaling yielded about 12.5% resolution at 1 MeV and roughly 7.5% by 10 MeV; the thesis summary characterizes the low-energy performance as approximately 10% at 1 MeV (Newmark, 8 Jul 2026).

5. Event-by-event Cherenkov identification

The most distinctive experimental achievement associated with CCM200 is the first event-by-event observation of Cherenkov radiation from sub-MeV electrons in a high-yield scintillator detector. Using $\mathcal{O}(1\,\mathrm{ns})$0Na calibration data and focusing on the uncoated PMTs, the collaboration observed a small peak at $\mathcal{O}(1\,\mathrm{ns})$1 ns before the reconstructed event start time. A representative summed pulse series was shown for 43,522 selected sodium decay events, and the early peak was interpreted as prompt Cherenkov light (Aguilar-Arevalo et al., 10 Jul 2025).

Within the interval $\mathcal{O}(1\,\mathrm{ns})$2 ns, the detector attains very high Cherenkov purity. The paper reports 94.8% Cherenkov purity in this region if at least one hit is required, with efficiencies of 9.78% in data and 9.10% in simulation. The result is notable because the analysis uses only the uncoated 20% of PMTs and still isolates a high-purity Cherenkov sample (Aguilar-Arevalo et al., 10 Jul 2025). The thesis independently describes the same requirement as yielding a sample that is about 94.8% Cherenkov-pure and emphasizes that the separation is achieved event by event rather than only statistically (Newmark, 8 Jul 2026).

Directionality provides a second line of evidence. For events with at least two hits in the Cherenkov-enhanced window, CCM computes the angle between hit-PMT directions and the source location. The angular distribution shows a preference for

$\mathcal{O}(1\,\mathrm{ns})$3

consistent with visible Cherenkov photons from electrons with kinetic energies around 0.7–1.0 MeV using a liquid-argon refractive index of about 1.22 for visible wavelengths. The comparison to a scintillation-plus-random-background hypothesis yields $\mathcal{O}(1\,\mathrm{ns})$4 for 20 degrees of freedom and rejection of the background-only hypothesis at greater than $\mathcal{O}(1\,\mathrm{ns})$5 confidence (Aguilar-Arevalo et al., 10 Jul 2025). The thesis adds that a $\mathcal{O}(1\,\mathrm{ns})$6Co control dataset, which should not produce above-threshold Cherenkov electrons, showed only background-level early hits, supporting the method’s validity (Newmark, 8 Jul 2026).

In detector-development terms, the result establishes that liquid argon can operate as a hybrid Cherenkov-plus-scintillation medium even in a high-light-yield environment. The papers argue that scintillation provides energy information, while Cherenkov light contributes directionality and particle-identification content; a plausible implication is that the CCM strategy is relevant well beyond this specific calibration measurement (Aguilar-Arevalo et al., 10 Jul 2025).

6. Physics program and broader role

CCM was originally designed for coherent elastic neutrino-nucleus scattering, sterile-neutrino disappearance, and light dark matter searches. In the dark-matter program, the CCM120 engineering run accumulated $\mathcal{O}(1\,\mathrm{ns})$7 POT over roughly 1.5 months of analyzable data, used a prompt beam window and recoil-energy selection, and found a background-subtracted result of $\mathcal{O}(1\,\mathrm{ns})$8 events, consistent with zero. In a leptophobic model, CCM120 excluded new parameter space for dark-matter masses between $\mathcal{O}(1\,\mathrm{ns})$9 and 30 MeV, and the first dedicated leptophobic dark-matter analysis reported exclusion of $\sim 6$0 at 90% C.L. for the benchmark choice $\sim 6$1 and $\sim 6$2 (Aguilar-Arevalo et al., 2021, Aguilar-Arevalo et al., 2021).

The detector has also been treated as a rare-event platform for axionlike particles. One 2021 study used the 2019 engineering run to begin constraining ALP parameter space and projected that, with a three-year run, CCM200 could improve sensitivity by up to an order of magnitude for both ALP-photon and ALP-electron couplings (Aguilar-Arevalo et al., 2021). A later proof-of-principle beam-dump search used the hybrid optical response directly: four observables exploiting Cherenkov timing, wavelength sensitivity, directionality, pulse shape, and event topology were combined into a likelihood-ratio test statistic to suppress steady-state backgrounds for $\sim 6$3 MeV ALP-induced events. No statistically significant excess was observed, but the analysis excluded new regions of ALP mass-coupling parameter space at the 90% confidence level despite using only about 70% of the POT of the previous CCM120 study (Newmark, 8 Jul 2026).

In neutrino phenomenology, CCM is repeatedly used as a benchmark argon CE$\sim 6$4NS facility. A 2021 study argued that, with a 7-ton fiducial detector, a 25 keV threshold, and the Lujan pulse structure, CCM could place stringent new bounds on NSI and likely rule out LMA-D solutions for a broad class of MeV-scale mediators (Shoemaker et al., 2021). A later comparative analysis treated CCM as a mature stopped-pion CEvNS platform with a 10-ton target, approximately 10 keVnr threshold, and 10% systematics, quoting a projected 90% C.L. interval $\sim 6$5, a magnetic-moment limit $\sim 6$6, and competitive bounds on flavor-diagonal and off-diagonal quark NSI (Carey et al., 15 Oct 2025).

The detector has also inspired unconventional proposals. In a study of the ATOMKI anomaly, the oxygen nuclei in CCM’s borosilicate PMT glass were proposed as the active production medium for an $\sim 6$7-like boson generated by neutron-induced $\sim 6$8O excitations, with the subsequent $\sim 6$9 decay searched for through prompt electromagnetic activity. The proposal concluded that CCM can probe a large fraction of the ATOMKI-allowed parameter space (Dutta et al., 2024).

Taken together, these programs define CCM less as a single-purpose detector than as a versatile optical liquid-argon instrument. The recurring theme across the literature is that high photocoverage, pulsed-beam timing, optical-only reconstruction, and hybrid Cherenkov/scintillation discrimination allow CCM to function simultaneously as a calibration testbed, a low-energy particle detector, and a beam-dump search instrument. The detector’s broader significance lies in demonstrating that liquid argon can support event-by-event Cherenkov identification in a high-scintillation environment and that this capability can be translated into concrete background rejection for rare-event physics (Aguilar-Arevalo et al., 10 Jul 2025, Aguilar-Arevalo et al., 10 Jul 2025, Newmark, 8 Jul 2026).