Modeling the light response of an optically readout GEM based TPC for the CYGNO experiment

The use of gaseous Time Projection Chambers enables the detection and the detailed study of rare events due to particles interactions with the atoms of the gas with energy releases as low as a few keV. Due to this capability, these instruments are being developed for applications in the field of astroparticle physics, such as the study of dark matter and neutrinos. To readout events occurring in the sensitive volume with a high granularity, the CYGNO collaboration is developing a solution where the light generated during the avalanche processes occurring in a multiplication stage based on Gas Electron Multiplier (GEM) is read out by optical sensors with very high sensitivity and spatial resolution. To achieve a high light output, gas gain values of the order of $10^5\text{-}10^6$ are needed. Experimentally, a dependence of the detector response on the spatial density of the charge collected in the GEM holes has been observed, indicating a gain-reduction effect likely caused by space-charge buildup within the multiplication channels. This paper presents data collected with a prototype featuring a sensitive volume of about two liters, together with a model developed by the collaboration to describe and predict the gain dependence on charge density. A comparison with experimental data shows that the model accurately reproduces the gain behaviour over nearly one order of magnitude, with a percent-level precision.

💡 Research Summary

The CYGNO collaboration is developing a cubic‑metre‑scale gaseous Time Projection Chamber (TPC) for rare‑event searches such as low‑mass WIMP dark‑matter interactions and low‑energy neutrino scattering. Conventional electronic readout would require hundreds of thousands of channels to achieve the millimetre‑scale spatial resolution needed for keV‑scale recoil tracks. To overcome this, CYGNO adopts an optical readout concept: the light produced during electron avalanches in a Gas Electron Multiplier (GEM) stack is collected by a high‑sensitivity scientific CMOS (sCMOS) camera and by fast photomultipliers (PMTs).

A dedicated 2‑litre prototype, called GIN, was built at INFN‑LNF. The detector consists of a PMMA vessel, a field cage made of copper rings, and a three‑GEM stack (50 µm inner hole, 70 µm outer hole, 140 µm pitch). The drift region is 23 cm long, giving a total active volume of about 2 L. The gas mixture is He/CF₄ (60/40) at atmospheric pressure, providing an average ionisation energy of 35 eV and an electroluminescence spectrum with a strong component around 620 nm, which matches the peak quantum efficiency of silicon‑based sensors.

The optical system uses a 200 µm PET window, a 25.6 mm focal‑length Schneider Xenon lens (aperture f/0.95) and an sCMOS camera (2304 × 2304 pixels, 6.5 µm pitch). The geometric acceptance (Ω) is ≈9 × 10⁻⁴, giving each pixel a projected area of 50 µm × 50 µm on the GEM plane. Two Hamamatsu R1894 PMTs provide timing information. The camera exposure is set to 150 ms to keep the average number of events per frame low while avoiding excessive empty frames.

The detector was operated with a drift field of 1 kV cm⁻¹, a transfer field of 2.5 kV cm⁻¹, and GEM voltages in the range 420–440 V, delivering effective gains of 10⁵–10⁶. To probe the gain dependence on charge density, a ⁵⁵Fe source (5.9 keV X‑rays) was placed behind a thin beryllium window and moved along the z‑axis. Each X‑ray creates ≈168 primary electrons (35 eV per ion‑pair) within a sub‑millimetre track. After multiplication, the avalanche produces 0.07 visible photons per secondary electron, i.e. a few hundred photons per primary X‑ray, which are then detected by the camera and PMTs.

A clear non‑linear reduction of the GEM gain with increasing charge density was observed. The authors attribute this to space‑charge buildup inside the GEM holes: as many electrons accumulate during a high‑gain avalanche, the local electric field is screened, reducing the probability of further ionisation. To quantify the effect, they introduced a simple phenomenological model

 G(ρ) = G₀ / (1 + α ρ)

where ρ is the charge density (e⁻ mm⁻³) in the GEM holes, G₀ is the low‑density gain, and α is a suppression coefficient. By fitting the model to the measured gain versus charge‑density curve, they obtained α values that reproduce the observed ≈30 % gain drop at ρ ≈ 10⁶ e⁻ mm⁻³. The model predictions agree with the data within 1 % over almost a full decade of charge density, demonstrating that the gain reduction can be accurately predicted and corrected.

The successful validation of this model has several practical implications. First, it enables the design of larger‑scale optical TPCs (up to 1 m³) with reliable gain control, because the required gain (10⁵–10⁶) can be maintained by adjusting GEM voltages or gas composition to keep ρ below the saturation regime. Second, it provides a framework for optimizing GEM geometry (hole diameter, pitch) and electric‑field configuration to minimise space‑charge effects. Third, the model can be incorporated into full detector simulations that combine charge transport, avalanche multiplication, and photon production, facilitating realistic sensitivity studies for dark‑matter and neutrino searches.

The paper also outlines future work: extending the optical system to multi‑camera arrays for larger coverage, exploring alternative gas mixtures to increase photon yield, implementing active feedback on GEM voltages to compensate for instantaneous charge‑density fluctuations, and integrating the gain‑density model into Monte‑Carlo tools for event reconstruction.

In summary, this study presents a comprehensive experimental characterisation of an optically read out GEM‑TPC, identifies the space‑charge induced gain suppression as the dominant non‑linearity at high gains, and delivers a simple yet precise analytical model that reproduces the effect with percent‑level accuracy. The results constitute a solid technical foundation for scaling the CYGNO detector to the cubic‑metre scale and for achieving the low‑energy, high‑resolution performance required for next‑generation rare‑event searches.