Long Imaging ModulE (LIME) Detector

• LIME is a 50 L gaseous TPC module featuring a 50 cm drift, triple-GEM amplification, and high-resolution optical readout for low-threshold event detection.
• It enables 3D reconstruction of low-energy (<100 keV) nuclear recoils through combined x-y imaging and PMT-timing, critical for dark matter and solar neutrino studies.
• Its modular design within the CYGNO program validates key technologies and provides insights into overcoming gain saturation, environmental effects, and background rejection challenges.

LIME, the Long Imaging ModulE, is a 50 L active-volume gaseous time projection chamber (TPC) developed within the CYGNO program as the largest prototype of its R&D phase and as the reference module for subsequent demonstrator-scale detectors. It combines a 50 cm drift length, a $33 \times 33~\mathrm{cm^2}$ triple-GEM amplification stage, and optical readout with one Hamamatsu ORCA Fusion sCMOS camera and 4 PMTs, operated with a $\mathrm{He:CF_4}=60{:}40$ gas mixture at atmospheric pressure. Its scientific purpose is the directional detection of rare events, especially low-mass Dark Matter and solar neutrino interactions, through low-threshold imaging, 3D reconstruction, and fiducial background rejection (Amaro et al., 2023, collaboration, 2023, Antonietti, 1 Oct 2025).

1. Role within the CYGNO program

Within CYGNO, LIME is explicitly the unit module from which the future demonstrator is intended to be assembled, and its underground results are described as paramount in the optimization of the CYGNO demonstrator, which is foreseen to use multiple modules with the same LIME dimensions and characteristics (Amaro et al., 2023). The thesis literature places LIME at the end of Phase-0 R&D, after the smaller ORANGE and LEMOn detectors, and as the experimental bridge toward the approved $0.4~\mathrm{m^3}$ CYGNO-04 demonstrator and the longer-term $\mathcal O(30~\mathrm{m^3})$ CYGNO-30 detector (Antonietti, 1 Oct 2025).

The physics motivation is directional rare-event detection in gas. CYGNO targets nuclear recoils below about 100 keV, where directional information is especially valuable because, in a WIMP scenario, the recoil angular distribution is expected to point roughly toward the Cygnus constellation. The choice of a gaseous detector follows from the need to spatially resolve low-energy recoil tracks; the TPC architecture provides drift-time information and fiducialization, while optical readout supplies fine-grained imaging over large areas with relatively low channel count (Amaro et al., 2023).

The overground characterization paper is explicit that LIME was not yet a low-background dark-matter detector. It used non-radiopure materials and was operated overground at Laboratori Nazionali di Frascati (LNF) with no shielding against environmental radioactivity; in that phase its function was technological validation, response characterization, and background diagnosis rather than a competitive search (collaboration, 2023).

2. Detector architecture and readout chain

LIME is a room-temperature, atmospheric-pressure gaseous TPC housed in a 10 mm thick transparent PMMA vessel kept at an internal overpressure of about 3 mbar over atmosphere. The active drift region is bounded by a 0.5 mm copper cathode and a triple-GEM amplification stage. The field cage is formed by 34 copper rings, each 10 mm wide and spaced by 4 mm, connected by $100~\mathrm{M\Omega}$ resistors to establish a uniform drift field orthogonal to the cathode (Antonietti, 1 Oct 2025).

The readout architecture is optical. The main imaging device is a Hamamatsu ORCA-Fusion sCMOS with $2304\times 2304$ pixels of $6.5\times 6.5~\mu\mathrm m^2$, coupled to a 25 mm Schneider Xenon lens with $f/\# \approx 0.95$. The camera is placed 623 mm from the GEM plane and images about $34.9\times 34.9~\mathrm{cm^2}$, corresponding to an effective granularity of $152\times 152~\mu\mathrm m^2$ per pixel. Four Hamamatsu R7378 PMTs, each with a 22 mm diameter bialkali photocathode, are symmetrically arranged around the camera (Antonietti, 1 Oct 2025, collaboration, 2023).

Subsystem	Reported specification	Function
Active volume	50 L	Target and drift region
Drift length	50 cm	Electron transport and fiducial depth
Amplification area	$\mathrm{He:CF_4}=60{:}40$0	Triple-GEM avalanche stage
Gas mixture	$\mathrm{He:CF_4}=60{:}40$1	Low threshold and scintillation yield
Optical imager	ORCA Fusion sCMOS	High-resolution 2D topology
Timing sensors	4 PMTs	Waveform timing and $\mathrm{He:CF_4}=60{:}40$2-information

The amplification stage consists of three 50 $\mathrm{He:CF_4}=60{:}40$3m GEM foils. The overground operating summary reports $\mathrm{He:CF_4}=60{:}40$4, $\mathrm{He:CF_4}=60{:}40$5, and $\mathrm{He:CF_4}=60{:}40$6 across each GEM, while the underground thesis reports typical drift fields of $\mathrm{He:CF_4}=60{:}40$7–$\mathrm{He:CF_4}=60{:}40$8 and notes that Run1 underground was limited to $\mathrm{He:CF_4}=60{:}40$9 per GEM because of discharges (collaboration, 2023, Antonietti, 1 Oct 2025).

The chosen gas is motivated by CYGNO’s low-energy directional program. The papers emphasize the combination of low energy threshold and high scintillation yield, together with optical compatibility between the visible CF$0.4~\mathrm{m^3}$0 emission near 620 nm and the camera’s high-QE band (Amaro et al., 2023, Antonietti, 1 Oct 2025).

3. Operating principle and reconstruction workflow

LIME follows the standard TPC chain, extended with optical avalanche readout. Ionization electrons drift toward the triple-GEM stack, where avalanche multiplication produces both amplified charge and secondary scintillation light. The sCMOS records the spatial distribution of this light on the GEM plane, providing the transverse image; the PMTs register its time development, which contains information on the coordinate along the drift direction. The prototype papers therefore describe LIME operationally as a detector with $0.4~\mathrm{m^3}$1-$0.4~\mathrm{m^3}$2 information from the optical image plane and $0.4~\mathrm{m^3}$3 information from timing, enabling 3D track reconstruction and fiducialization (Amaro et al., 2023).

In the underground thesis, PMT-based waveform analysis is described as still under refinement, but the detector concept and dedicated studies support both absolute and relative $0.4~\mathrm{m^3}$4 reconstruction. The same thesis also reports an absolute-$0.4~\mathrm{m^3}$5 estimator built from camera images alone for 5.9 keV spots, yielding an absolute $0.4~\mathrm{m^3}$6 resolution from about 4 cm at short drift to about 8 cm at large drift distances (Antonietti, 1 Oct 2025).

The image-analysis chain became increasingly elaborate in the underground campaign. Raw $0.4~\mathrm{m^3}$7 sCMOS frames undergo pedestal subtraction, zero suppression, $0.4~\mathrm{m^3}$8 rebinning into macro-pixels, median filtering, and vignetting correction. Clustering is then performed with iDBSCAN, an intensity-based version of DBSCAN; dense seeds are merged into longer structures by superclustering, first with geodesic active contours and later with Chan–Vese segmentation, while iDDBSCAN adds repeated linear or polynomial RANSAC fits to merge long directional tracks such as cosmic rays (Antonietti, 1 Oct 2025).

The principal reconstructed observables include $0.4~\mathrm{m^3}$9 as the total light integral, $\mathcal O(30~\mathrm{m^3})$0 as the number of nonzero pixels, bounding-box coordinates, $\mathcal O(30~\mathrm{m^3})$1, $\mathcal O(30~\mathrm{m^3})$2, $\mathcal O(30~\mathrm{m^3})$3 as the Gaussian width along the minor axis, and $\mathcal O(30~\mathrm{m^3})$4 as the RMS of pixel intensities inside the cluster. Calibration spot selection uses the slimness variable $\mathcal O(30~\mathrm{m^3})$5, and fiducial and quality selections include

$\mathcal O(30~\mathrm{m^3})$6

together with

$\mathcal O(30~\mathrm{m^3})$7

and

$\mathcal O(30~\mathrm{m^3})$8

For preliminary dark-matter counting analyses, the additional variable

$\mathcal O(30~\mathrm{m^3})$9

is used with the selection $100~\mathrm{M\Omega}$0 to reject MIP-like backgrounds (Antonietti, 1 Oct 2025).

4. Calibration, stability, and measured response

The central calibration source throughout the LIME program is $100~\mathrm{M\Omega}$1, which provides 5.9 keV X-rays. Overground, additional calibration lines were obtained from Ca, Ti, Cu, Rb, Mo, Ag, Ba, and in the large-prototype study also from Tb, spanning from 3.7 keV up to 47 keV (Amaro et al., 2023, collaboration, 2023). The overground characterization reports that the detector achieved a few-keV threshold and an energy resolution of 10–20% over the studied range while running for several weeks continuously with very high operational efficiency (collaboration, 2023).

Several complementary threshold statements appear in the literature. The prototype performance paper states that a threshold of 0.5 keV was set such that the expected rate of fake $100~\mathrm{M\Omega}$2-like events due to noise would be at most 10 per year (Amaro et al., 2023). The later thesis reports that fake clusters become negligible above 400 counts, about 0.5 keV, and that a threshold of 1 keV corresponds to roughly 10 false events/year (Antonietti, 1 Oct 2025). The overground characterization also distinguishes between a low cluster-counting threshold corresponding to about 300 eV and the practical spectroscopy range demonstrated down to the Ca and Ti lines near 3.7–4.9 keV (collaboration, 2023).

At 5.9 keV, the commissioning study reports an energy resolution of around 14% across the full 50 cm drift length, and notes that a multivariate regression analysis using track position and various shape parameters was under development with preliminary indications of better than 10% energy resolution at 5.9 keV (Amaro et al., 2023). The overground characterization resolves this more finely: after multivariate correction, the full-sample RMS at 5.9 keV is about 12% for favorable drift regions, while the best-cluster $100~\mathrm{M\Omega}$3 from the Crystal Ball fit is smaller than 10% for $100~\mathrm{M\Omega}$4 cm (collaboration, 2023). The underground thesis then reports that, after equalisation across runs, the energy resolution of the $100~\mathrm{M\Omega}$5 peak is about 12% in every phase (Antonietti, 1 Oct 2025).

A persistent detector effect is the dependence of light yield on drift distance. Rather than decreasing with $100~\mathrm{M\Omega}$6, the average $100~\mathrm{M\Omega}$7 light yield increases away from the GEM plane because diffusion spreads the primary charge cloud, reducing local charge density in the GEM holes and mitigating gain saturation. The thesis quantifies this strongly near the amplification stage: at 4.75 cm the $100~\mathrm{M\Omega}$8 peak is reduced by about 35%, at 14.75 cm by about 15%, and from 24.75 to 45.75 cm it is near the plateau around $100~\mathrm{M\Omega}$9 counts (Antonietti, 1 Oct 2025).

Environmental dependence is another central result. Overground at LNF, the normalized $2304\times 2304$0 response versus pressure was fitted with $2304\times 2304$1, showing a decrease of about 0.6% per mbar as pressure increased (collaboration, 2023). Underground, this behavior was confirmed: Run1 showed $2304\times 2304$2/mbar at 3 l/h and $2304\times 2304$3/mbar at 20 l/h, while Run2 showed $2304\times 2304$4/mbar at 20 l/h (Antonietti, 1 Oct 2025). Once gas recirculation started, humidity became a dominant variable: the normalized $2304\times 2304$5 light yield decreased by about 20% per unit increase in RH in Run3, and by $2304\times 2304$6 per RH unit in Run4 (Antonietti, 1 Oct 2025).

To merge long data periods taken under changing environmental conditions, the thesis introduces a run-by-run equalisation procedure based on the new variable $2304\times 2304$7, defined as the mean light integral of reconstructed clusters between 30 kcounts and 300 kcounts, using only fiducial cuts. The measured $2304\times 2304$8 peak and $2304\times 2304$9 are reported as linearly correlated, and the relative difference between measured and inferred iron-peak positions has a Gaussian width of 13%, which is then adopted as the equalisation uncertainty (Antonietti, 1 Oct 2025).

5. Underground deployment and background characterization

LIME was installed underground at Laboratori Nazionali del Gran Sasso (LNGS) in February 2022, in the TIR gallery between Hall A and Hall B (Antonietti, 1 Oct 2025). The underground campaign was not merely a relocation. It was intended to validate detector operation in a low radioactivity and low pile-up environment, to test the gas system and the full 3D reconstruction and background-rejection chain, and to confront detector simulations with real underground data (Amaro et al., 2023).

The shielding configuration evolved in stages. The thesis defines Run1 as unshielded, Run2 with 4 cm Cu, Run3 with 10 cm Cu, and Run4 with 10 cm Cu + 40 cm water (Antonietti, 1 Oct 2025). The broader prototype study had already emphasized the role of dedicated MC optimization for a shielding concept based on copper against gammas and water tanks against neutrons (Amaro et al., 2023).

Background simulation and measurement became a major component of LIME’s scientific role. The underground MC includes both intrinsic radioactivity of detector materials and natural ambient gamma and neutron flux, and the 50 L prototype paper reports that fiducial cuts can reduce radioactivity-induced backgrounds by 96% (Amaro et al., 2023). The thesis provides more detailed rates for the unshielded and shielded phases: about $6.5\times 6.5~\mu\mathrm m^2$0 ER events/year and $6.5\times 6.5~\mu\mathrm m^2$1 NR events/year without shielding; $6.5\times 6.5~\mu\mathrm m^2$2 ER events/year with 4 cm copper; $6.5\times 6.5~\mu\mathrm m^2$3 ER events/year with 10 cm copper; and $6.5\times 6.5~\mu\mathrm m^2$4 ER events/year with 10 cm Cu + 40 cm water, at which point internal backgrounds dominate. With full water shielding, the environmental neutron recoil contribution is reduced to about 2/year (Antonietti, 1 Oct 2025).

The overground campaign provided the reference point for why underground operation was necessary. In source-free LNF data, with 50 ms exposure and a threshold corresponding to about 300 eV, the average detected interaction rate was about 250 Hz, and the average thresholded energy deposition rate was about $6.5\times 6.5~\mu\mathrm m^2$5 (collaboration, 2023). For cosmic-ray-like events, the overground paper reports a measured rate of about 15 Hz, to be compared with a predicted maximum active-volume interaction rate of about 24 Hz from standard sea-level flux estimates (collaboration, 2023). The underground move was therefore directly tied to reducing pile-up and isolating internal detector backgrounds.

Data/MC comparison in the underground thesis shows good agreement in Run1 above 10 keV and a clear Cu fluorescence peak near 8 keV. In Run2 the shapes remain similar. In Run3, however, the MC underestimates the low-energy data, which is interpreted as evidence for internal background sources not fully simulated (Antonietti, 1 Oct 2025).

6. Dark-matter relevance, limitations, and legacy

LIME’s first dark-matter study is described as explicitly preliminary and as a proof of analysis chain rather than a competitive search. The thesis uses 17 days of Run4 exposure with 10 cm Cu + 40 cm water shielding, approximately constant pressure around 0.907 bar, and thresholds of 1 and 1.5 keVee. A Bayesian treatment with Poisson likelihood is then used to derive 90% credible upper limits; the author states explicitly that the resulting SI and SD exclusions are not competitive with world-leading experiments, although the SD result is compared qualitatively to DRIFT as encouraging given the modest exposure and prototype status (Antonietti, 1 Oct 2025).

The detector nevertheless establishes several features that are directly relevant to future directional searches. It demonstrates atmospheric-pressure operation of a large optical TPC, low-keV calibration, long-term gain monitoring, run equalisation, and underground background characterization (Antonietti, 1 Oct 2025). The large-prototype paper also notes that the detector-response MC already reproduces key observables such as light integral and track dimension for 5.9 keV ERs with preliminary agreement within 10% in the linearity study, and that ER/NR discrimination at LIME scale is under active study (Amaro et al., 2023).

The limitations are equally clear. LIME was not built with radiopure components, so once external backgrounds are shielded, internal radioactivity dominates (Antonietti, 1 Oct 2025). Gain saturation near the amplification plane produces a strong $6.5\times 6.5~\mu\mathrm m^2$6-dependent response below about 15 cm from the GEMs (Antonietti, 1 Oct 2025). PMT-based 3D reconstruction was still under refinement in the underground analysis chain (Antonietti, 1 Oct 2025). Overground, the detector suffered from unshielded environmental radioactivity, cosmic-ray pile-up, and the fact that the 50 ms exposure made overlap of tracks unavoidable in a high-rate environment (collaboration, 2023).

In spite of those limitations, LIME functions as a technically mature module-scale demonstrator. The collected literature supports a consistent interpretation: it is the first CYGNO prototype that matches the scale and architecture of a demonstrator module; it validates stable operation of a 50-liter, 50-cm-drift, triple-GEM optical TPC; it quantifies pressure, humidity, impurity, and saturation effects in detail; and it provides the experimental basis for scaling to CYGNO-04 and, ultimately, to the modular CYGNO-30 concept (Amaro et al., 2023, Antonietti, 1 Oct 2025).