Hamamatsu R12699-406-M4 PMTs

• The Hamamatsu R12699-406-M4 PMT is a 2-inch square multianode photomultiplier tube designed for ultra-low radioactive conditions in liquid xenon detectors.
• It achieves high single photoelectron sensitivity, stable gains up to 5.5×10^6, low dark count rates, and reduced after-pulse probabilities under cryogenic conditions.
• Its advanced base circuit and custom, low-radioactivity materials ensure extensive dynamic range and minimal noise, key for rare-event search experiments.

The Hamamatsu R12699-406-M4 photomultiplier tube (PMT), herein "R12699," is a 2-inch square multianode PMT jointly developed by Hamamatsu Photonics K.K. and the PandaX collaboration to address the stringent demands for ultra-low radioactive backgrounds in next-generation liquid xenon (LXe) time projection chambers aimed at rare-event searches, including dark matter and neutrinoless double-beta decay. Distinguished by its four-channel anode design, optimized materials, and robust cryogenic performance, the R12699 achieves significant reductions in radioactivity and operational noise while maintaining high single photoelectron (SPE) sensitivity and dynamic range, features critical for the photon detection systems in experiments such as PandaX-xT, XLZD, and DARWIN (Yun et al., 2024, Huang et al., 18 Jan 2026).

1. Design Principles and Physical Architecture

The R12699 PMT has a 56 mm × 56 mm × 14.8 mm square glass body, incorporating a 48.5 mm × 48.5 mm bialkali photocathode coupled directly to a UV-grade quartz window, providing sensitivity from 160–650 nm, including efficient transmission of 178 nm xenon scintillation light (Yun et al., 2024). Its four independent anode channels share a common 10-stage dynode chain, enabling increased photodetector packing density and reduced dead space. The device operates with nominal voltages applied to the photocathode (–1000 V) and anode (0 V).

Key low-background engineering features include:

• Use of a custom low-cobalt, Fe–Ni alloy for body and electrode parts, supplanting standard Kovar (Fe–Co–Ni), mitigating $^{60}$Co background.
• Sealing glass replaced by a low-thorium/uranium-content quartz glass, synthesized in platinum crucibles to further suppress contamination.
• Internal insulation employs thoroughly screened ceramics, stainless steel (304/316), and an external aluminum guard ring, targeting minimum radioactivity contribution.

This design enables the R12699 to satisfy compactness, mechanical robustness, and cleanliness requirements for dense LXe detector arrays (Yun et al., 2024).

2. Radioactivity Suppression and Measurement

Background minimization is achieved through stringent material selection and process controls. Quantitative radioassay performed with a high-purity germanium detector at the China Jinping Underground Laboratory (CJPL) yielded the following activity levels (all per PMT, 90% C.L. upper limits) (Yun et al., 2024):

Isotope	Activity (mBq/PMT)
$^{60}$Co	0.08
$^{238}$U (late)	0.06
$^{232}$Th (late)	0.09
$^{40}$K	35 (from photocathode)

The $^{60}$Co, $^{238}$U(late), and $^{232}$Th(late) rates represent a reduction of over 15 times compared to the 3-inch R11410 PMT employed in PandaX-4T. Additionally, $^{222}$Rn emanation is constrained to $<3.2$ μBq/PMT (90% C.L.), and surface $^{210}$Po is $<18.4$ μBq/cm$^2$. This comprehensive radioactivity suppression is essential for minimizing detector backgrounds and enabling the desired physics sensitivities (Yun et al., 2024).

3. Electrical and Cryogenic Performance

All operational characterizations of the R12699 were conducted at liquid xenon-relevant cryogenic conditions (down to –105 °C) (Yun et al., 2024). The PMT achieves stable high gain, low noise, and robust multi-cycle endurance at these temperatures.

• Single-Photon Gain and Resolution: Gains range from $2.7 \times 10^6$ to $5.5 \times 10^6$, with SPE charge at $G = 2.7 \times 10^6$ yielding $Q_{\rm SPE} \approx 0.43$ pC. The SPE peak exhibits $<10\%$ FWHM/mean variation over all bench runs, and an average SPE resolution ($R = \sigma_1/\mu_1$) of approximately 42% (Yun et al., 2024, Huang et al., 18 Jan 2026).
• Readout-Base Optimization: An optimized negative-HV readout base, with modified voltage-divider ratios (K–Dy1–Dy2…: 1.5–1.1–1.4–1–…–0.5), produces a gain enhancement of approximately 30% at –1000 V and –100 °C, reaching $\langle G\rangle = (4.23 \pm 1.37) \times 10^6$ (Yun et al., 2024).
• Dark Count Rate (DCR): At –100 °C, $\langle\rm DCR\rangle = (2.5 \pm 0.9)$ Hz/channel or $0.32$ Hz/cm$^2$, reflecting a suppression factor $>90$ compared to ambient temperature operation ($\sim$230 Hz/channel) (Yun et al., 2024).
• After-Pulse Probability (AP): The after-pulse rate is 0.5%, substantially below typical R11410 values (1–1.4%), and no measurable gain shift ($\Delta G / G < 1\%$) is observed across thermal cycling from room temperature to –100 °C (Yun et al., 2024).

These metrics confirm the R12699’s suitability for stable operation in large-scale xenon TPCs.

4. Dynamic Range, Saturation, and Base Circuit Engineering

The R12699’s base circuit employs a 10-stage resistor chain augmented with desaturation capacitors (C1–C4) and a filter capacitor (C5), specifically tuned for PandaX-xT's physics requirements (Huang et al., 18 Jan 2026). Three base variants—differing by the presence of C1, C2, and C3—were systematically characterized:

Base Design	Desaturation Capacitors	Linear Range $Q^{\rm limit}$ (nC)	Approx. $k$PE
BASE-1	C1–C4	286	662k
BASE-2	C2–C4	83	192k
BASE-3	C3–C4	25	58k

BASE-2, with three desaturation capacitors, provides $>60$ nC linear dynamic range at gain $2.7\text{–}5.5 \times 10^6$, thus exceeding the 30 nC requirement for covering the $2.5$ MeV $^{136}$Xe NLDBD Q-value signal (Huang et al., 18 Jan 2026).

The base circuit achieves this by shunting transient dynode charge that would otherwise induce local voltage drops and degrade gain, expressed as $\Delta V_n = \Delta Q_n/(C_{\mathrm{eq}}+C_i)$. By increasing $(C_{\mathrm{eq}}+C_i)$ near the last dynodes, saturation is postponed until higher charges are delivered.

Saturation and suppression behaviors were analyzed using both bench measurements and a dedicated LTSpice simulation, showing agreement within 10% over the dynamic range (Huang et al., 18 Jan 2026). For BASE-2, recovered charge after saturation follows a broken power law with $Q^{\rm sat}\simeq83$ nC and $\alpha\simeq3$. Temporal suppression and recovery of subsequent pulses exhibit dual-exponential ($\tau_1 = 125\,\mu$s, $\tau_2 =12$ ms) dependence, related to C_eq and C_desat relaxation (Huang et al., 18 Jan 2026).

5. Comparative Performance and Suitability for Rare-Event Detectors

The R12699’s performance, compared directly to the R11410 model used in PandaX-4T, is summarized below (Yun et al., 2024):

Parameter	R11410	R12699-406-M4
Gain	$5\times10^6$ @–1.5 kV	$4.23\times10^6$ @–1 kV
AP probability	1%	0.5%
DCR/channel	20 Hz	2.5 Hz
DCR/cm$^2$	0.44 Hz/cm$^2$	0.32 Hz/cm$^2$
$^{60}$Co	<0.12 mBq	0.08 mBq
$^{238}$U(late)	<0.12 mBq	0.06 mBq
$^{222}$Rn	–	<3.2 μBq
$^{210}$Po surface	–	<18.4 μBq/cm$^2$

The R12699’s compact form factor and multi-channel design facilitate high-density assembly and uniform field coverage, rendering it suitable for deployment in tonne- and multi-tonne-scale xenon detectors (e.g., PandaX-xT, XLZD, and DARWIN), where increased granularity and minimized radioactivity are pivotal (Yun et al., 2024).

6. Circuit Simulation and Systematic Correction Strategies

A detailed electrical simulation framework, constructed in LTSpice, models the R12699’s dynode chain incorporating resistor/capacitor networks and controlled-current sources for secondary emission. Tunable parameters—such as equivalent inter-dynode capacitance ($C_{\mathrm{eq}}=0.1$ nF), multiplication coefficients $\{\delta_n\}$, and non-linearity parameters ($A=2.3\times10^4, B=1.7$ for BASE-2)—are validated against saturation and suppression data (Huang et al., 18 Jan 2026). The simulation clarifies that saturation is driven chiefly by C_eq depletion in the last dynodes and quantifies a suppression dead time of ≈20 ms. Empirical correction functions, restoring measured signals to within 5% for separation times $\delta_t\gtrsim 50$ μs, are established, facilitating integration of saturation/suppression corrections into future full-TPC data analyses (Huang et al., 18 Jan 2026).

Future directions include the evaluation of low-radioactivity dielectric materials for desaturation capacitors and extending the simulation framework to other PMT models for cross-validation.

7. Implications for Next-Generation Xenon Detectors

The R12699-406-M4 unites ultra-low intrinsic radioactivity ($^{60}$Co = 0.08 mBq/PMT, $^{238}$U(late) = 0.06 mBq/PMT), minimal radon emanation (<3.2 μBq/PMT), and robust operational characteristics at cryogenic temperatures—high gain (up to $5.5\times10^6$), low DCR (2.5 Hz/channel @ –100 °C), low after-pulse rate (0.5%), and extended dynamic range for high-energy depositions (Yun et al., 2024, Huang et al., 18 Jan 2026). These features make R12699 a leading candidate for photon detection in next-generation, background-constrained LXe rare-event searches. Further reduction of residual backgrounds and advances in base materials are anticipated as the collaboration continues optimizing for both scaling and sensitivity.