Resistive-protection schemes are nowadays ubiquitous in avalanche-based gaseous detectors for which the detection of single-photons (Melai et al., 2011) or single-electrons (Ligtenberg et al., 2018) is required, in harsh environments with an abundance of highly ionizing particles (Chefdeville et al., 2021) or when a very fast response is needed (Fonte et al., 2000; Lisowska et al., 2023). During detector operation, when the regular avalanche growth unnaturally destabilizes into the formation of a streamer channel, resistive-protection effectively helps at quenching further evolution and preventing electrical breakdown. At much lower gains than the ones needed for the aforementioned applications, operation in ultra-pure noble gas can in principle benefit from a similar protection scheme (Leardini et al., 2023), thus mitigating the impact of the feedback processes that eventually lead to discharge in those conditions.

When it comes to spark quenching through resistive protection, three main approaches exist: coatings (Lv et al., 2020), micro-patterned resistive elements (Chefdeville et al., 2021), and plates (Deppner, 2012). Although the first two represent attractive directions, the development of plates remains crucial in some detector configurations, chiefly for resistive plate chambers (RPCs) (Parkhomchuck et al., 1971; Santonico and Cardarelli, 1981) and resistive plate wells (RP-WELLs) (Jash et al., 2022). The scarcity of materials suitable for use as plates was a concern in the early 00s (e.g. (Morales et al., 2013)), however the situation has been greatly alleviated over the last 2 decades with the introduction of ‘chinese’ glass (Wang, 2013), warm-glass (González-Díaz et al., 2005) and Si₃N₄/SiC ceramics (Laso García et al., 2016), an effort largely driven by the R&D of the Compressed Baryonic Matter experiment (CBM) at FAIR (Darmstadt) (Deppner, 2012). More recently, other materials have become available too (see for instance (Liu et al., 2020) and references therein). Stimulated by this progress, a natural follow-up question to us was whether resistive-plate protection of gaseous detectors could be extended to temperatures considerably below standard conditions.

It must be noted that the region of greatest interest to modern gaseous instrumentation starts at 87.5 K (the boiling point of argon) and new material technologies should cover at least the 165 K landmark (the boiling point of xenon), ideally up to room temperature. The difficulty of spanning such a broad temperature range lies in the fact that many insulator materials display an Arrhenius-type conductivity law: a material with the right resistivity for spark quenching at around 77 K can be expected to be a decent conductor at room temperature. Studies performed in the context of resistive plate chambers point to the need of volume resistivities around 10⁹–10¹² Ω⋅cm to achieve good spark-quenching (e.g. (Francke et al., 2003)). Further, the material should ideally display a stable conductivity during operation as a function of time and transported charge, not being this the case for most common insulators (Morales et al., 2013). Therefore, and given that the detector atmosphere should be driven by the detector operational characteristics and not by the need to replenish the ionic charge carriers that migrate through the resistive protection, electron conductivity is an attractive feature.

When targeting a new insulating material focused on this latter aspect, ceramics represent an appealing possibility due to the large margin offered for customization, the possibility of achieving low outgassing and ruggedness, and their potential for scalability to large areas at affordable price. Fe₂O₃ (hematite), in particular, represents an interesting starting point as it displays electron conductivity based on the polaron-hopping mechanism (Liao et al., 2011; Ahart et al., 2022). Despite being too resistive for cryogenic use in gaseous detectors, its conductivity might be enhanced by orders of magnitude through the addition of Zr, Ti, Si or Ge (Morin, 1951; Kennedy et al., 1981; Ce et al., 2006; Kumar et al., 2011). In this case the dopant acts as an electron donor, up to concentrations of 10% Zr atoms per Fe atom (or 16.3%wt, in weight percentage), when a new phase is formed. For larger concentrations, ZrO₂ (zirconia) might be used instead, leading to speculation of additional contributions to the electron conductivity, either through the grain-boundary (Boukamp, 2003), or electron hopping through Fe cations in the zirconia lattice (Kharton, 1999). Given this landscape, a natural possibility was to evaluate ceramics based on Fe₂O₃ and zirconia stabilized with yttria at 8% molar (hereafter yttria-stabilized zirconia, YSZ) for which a large body of experience exists ((Verkerk et al., 1982; Slilaty, 1996; Gao et al., 2008; Molin et al., 2009; Bohnke et al., 2014; Guo, 2021) and references therein). Despite the existence of abundant experimental work, we found ourselves quickly in the range above the solubility limit of Fe₂O₃ in YSZ (about 6% molar (Molin et al., 2009)), that has been much less studied in literature.

In this work we evaluate the viability of using ceramics based on hematite (Fe₂O₃) and yttria-stabilized zirconia (YSZ) as resistive plates for spark-quenching of gaseous detectors, with Fe₂O₃ concentrations in the range of 30%–100% per weight (%wt). This paper is structured as follows: in Section 2 we describe the fabrication procedure, Section 3 introduces the setup and experimental methods; Section 4 includes a detailed characterization as a function of electric field, temperature, frequency, transported charge and Fe₂O₃ concentration; Section 5 presents first results from an RP-WELL operated at LArT, and we end with a short discussion in Section 6 and main conclusions in Section 7.

Ceramic plates were produced through slip casting. YSZ with crystallite sizes of 22 nm, and hematite of grain size below 5 µm (purity above 99%) were employed. Raw materials were weighted on a precision scale and admixed with 25% water and 1% deflocculant (Dolapix PC67). The resulting slurry was stirred in a mill filled with zircon balls for 24 h, after which it was poured into a plaster mold of internal dimensions 50 × 50 × 3 mm³. It was then put on a vibrating table for some minutes and, once the sample was consolidated, it was dried in air.

The sintering process started with the sample heated in an oven up to 900°C, in air atmosphere (pre-sintering). After the sample was further consolidated and contracted down to about the final dimensions of 40 mm × 40 mm, it was thinned with SiC sandpaper down to 1–3 mm to ensure plano-parallelism. The final sintering was accomplished at 1250°C–1300°C for about 2 h, again in air atmosphere, and the sample let cool down at about 2 °C/min. Although magnetite formation (Fe₃O₄) can be expected only above 1345°C, a magnetic hysteresis analysis confirmed its absence from the final samples.

Finally, the external oxide layer stemming from the sintering process was removed by thinning the samples further down by about 100 µm. The criteria employed was that the surface electrical resistance would remain constant upon further polishing. Ceramic pieces are shown for illustration in Figure 1-left (pre-sintered) and right (sintered).

Samples were studied by X-ray dispersion and found to be largely free from contaminating species (Figure 2). The results from the diffractogram were generally found to be consistent with the mass fraction targeted during the production of the slurry. Surface analysis showed good homogeneity, compatible with the grain sizes from the raw materials, without strong indications of clustering (Figure 3). A study of the characteristic X-ray emission was compatible with the absence of species other than Fe, Zr, Y, and O. Further details on the fabrication process and morphological characterization will be given in forthcoming publications.

For plate-based resistive protection of gaseous detectors, the product of the bulk resistivity times the sample thickness represents the main metric driving system performance (e.g. (Deppner, 2012)). Accordingly, a thorough characterization of the bulk resistivity was performed following the recommendations in (IEC 62631-3-1, 2016). Guard-ring was dimmed necessary only for low-temperature measurements of the ceramics with low Fe₂O₃ concentration (30%wt), otherwise measurements would be shunted by surface currents. Figure 4 describes the setup. The characteristic size of the electrodes was chosen to be at least 10 times larger than the plate thickness (1–3 mm, hence electrodes were in the range 1–10 cm²), to keep fringe-field effects at the 10% level or below (Lisowski and Kacprzyk, 2006). Connections were made through electrical leads clamped to the sample, that was immersed directly in a liquid N₂ (LN) bath in the case of measurements performed at LN-temperature (LNT). Different electrodes were studied: Ag-loaded epoxy (CHEMTRONICS, 2023), Ag-loaded paint (SAFETY DATA SHEET, 2023), Pt (deposited via magnetron sputtering), Au and Cr (deposited via physical evaporation at 100 nm thickness). All samples were thoroughly polished prior to adhering the electrodes in order to exclude the presence of oxide skin. The electrical resistivity was determined with a Keithley 6487 device. Measurements were performed up to the limits of the apparatus, namely,: 1) loss of sensitivity at low voltage due to noise, 2) exceeding the current limit for high voltages (20 mA), 3) exceeding the maximum voltage (500 V).

At LNT, measurements were generally found to be compatible within ±25%, however deviations of up to 3 orders of magnitude appeared at room temperature (RT), as shown in Figure 5. In this latter case, Cr-electrodes yielded the lowest resistivity and a good ohmic behavior while the rest exhibited some form of charge blocking that was reduced as the field increased, approaching the Cr values. Therefore, the contact of the Cr-electrode was dimmed good across all temperatures explored in this work, and a systematic uncertainty assigned to the LNT measurements based on the dispersion between different electrode types. ¹ Point-to-point variations of the resistivity were studied in different regions within the sample (near the four corners and at the center). Their relative spread, in the range 8%–30%, was used as an estimator of the sample uniformity.

Measurements as a function of temperature were taken in “drift” mode, i.e., by removing the sample from the LN bath and letting it gradually warm up, unimpeded. A PT-100 was epoxied directly to the sample (Figure 4-bottom). The first 5 s upon removal from the bath were excluded from the measurement to allow the sample to reach equilibrium, and the average of two identical series was taken. The combined uncertainty was estimated from their standard deviation, the point-to-point variations, and the electrode-to-electrode variations, added quadratically.

Last, measurements of the dielectric constant were performed, as its value impacts the signal induction process and the recovery time of resistive-protected detectors (e.g. (González Díaz et al., 2011; Riegler, 2016; Xie et al., 2021)). For practical reasons, a first assessment based on a standard LCR-meter with a limited range of 0.1 kHz–1 MHz was performed.

Prior to this work, an RP-WELL was successfully operated at LXeT (163 K) under both X-rays and UV irradiation, reporting a superior performance compared to standard (unprotected) amplification (Roy et al., 2019). In that study, a ceramic plate made of Fe₂O₃/YSZ (35%wt Fe₂O₃) with a bulk resistivity of ∼ 1 0 11 Ω⋅cm was used. Here we showcase the basic performance results from the recent operation of a cryogenic RP-WELL down to 90 K in purified argon vapor. The detector was assembled by means of a 0.8 mm-thick THGEM coupled to a pick-up anode via a ceramic plate (75%wt and 65%wt Fe₂O₃ were employed, corresponding to ρ ∼ 10¹⁰–10¹¹ Ω⋅cm at 90 K). Irradiation was performed with a collimated and attenuated ²⁴¹Am alpha source installed on a metallic plate (cathode) 15 mm below the amplification structure. A drift/collection field of 0.5 kV/cm was set to the ionization region (Figure 11).

The source releases ionization electrons that are transferred to the RP-WELL holes and multiplied there. Typical charge-collected spectra for RP-WELL structures based on ceramic plates of 75%wt (light blue) and 65%wt (dark blue) are shown in Figure 12. They exhibit a full-energy peak corresponding to perpendicular emission, accompanied by a low-energy tail accounting for partial energy loss in the collimator at high angles. Given the energy of 4 MeV for the attenuated alpha particles in case of perpendicular incidence and the average energy to create an electron-ion pair, W _I = 26.4 eV (Blum et al., 2008), the primary ionization in the gas can be estimated at 0.024 pC. Hence the charge spectra displayed in Figure 12 correspond to charge-multiplication factors in the range of 10. A more detailed analysis shows that the highest stable gain is 16, about × 3 the values obtained for an unprotected structure (details can be found in (Tesi et al., 2023)).

Tunability of the resistivity of hematite ceramics by nine orders of magnitude, through addition of YSZ in different proportions, makes Fe₂O₃/YSZ ceramics suitable for gaseous-detector assemblies at temperatures from liquid nitrogen up to room temperature. Of particular interest are the 75%wt, 35%wt, and 100%wt samples, which show resistivities in the range 10⁹–10¹² Ω⋅cm at the LAr, LXe boiling temperatures and room temperature, respectively. The approximate agreement between the resistivity values of the new ceramic batch and an earlier version developed in (Roy et al., 2019) suggests that the overall characteristics of the final samples do not depend critically on the fabrication details.

Apart from the suitable resistivity range, the high breakdown voltage observed (above 3 kV) together with the strong hints of electronic conductivity and stability over time and transported charge, represent (a priori) essential assets for the intended application. Both literature and our measurements support the electron-conductivity hypothesis.

(i) On the low Fe₂O₃-concentration end (few %wt), Fe- and Fe₂O₃-doped YSZ ceramics have been extensively studied with concentrations up to around 5% mol and at high temperatures. Under these conditions, the conductivity is predominantly ionic. When the content of Fe-based impurities exceeds the solubility limit at around 6% mol (2.84%wt), paths of hematite have been reported to generate across the sample (Verkerk et al., 1982; Kharton, 1999; Boukamp, 2003).

Some (or all) of the above electron-conductivity mechanisms are expected to become dominant for low temperatures in case of medium and high Fe₂O₃-concentrations, given that the activation energy will be too low to enable ion conductivity (Kharton, 1999).

The existence of mixed electron-ion conductivity at room temperature is hinted in our data by indications of charge-blocking at low fields (Figure 5-bottom) as well as the non-monotonous behavior of the relaxation time with Fe₂O₃ concentration (Figure 10-right). While, based purely on the stability of the conductivity with transported charge, ion conductivity seems to be subdominant, additional indications can be found at LNT. In such conditions, charge blocking disappears for low electric fields and the relaxation times evolve monotonically with Fe₂O₃ concentration, facts that can be interpreted as a result of the conduction process becoming ‘simpler’ at low temperatures, and expectedly dominated by a form of electron conduction.

We demonstrate the maturity of Fe₂O₃/YSZ ceramic plates made in-house for use in gaseous detectors as resistive protection, across temperatures, specifically from 77 K (LNT) up to 300 K (RT). Fe₂O₃/YSZ ceramics are robust, display low-outgassing and their resistivity is tunable through adjustment of the Fe₂O₃ content. Optimal Fe₂O₃ concentrations are found to be 30-40%wt (LXeT), 65-75%wt (LArT) and 100%wt (RT). In the 65-75%wt range, for which ‘in-detector’ operation has been presented here for the first time at around LArT, the material has an approximately linear response up to 20 V voltage drop, and displays a breakdown voltage above 3 kV. Very good signal-coupling is expected by virtue of a relatively large dielectric constant around ϵ _r ∼ 20. No electrical-aging effect (e.g., exhaustion of charge carriers) was observed up to 8 mC/cm² (LNT) and 1 C/cm² (RT). According to these results, and existing literature, the dominant conductivity form of this family of ceramics is likely of electron origin, thus guaranteeing stability during run-time.