Polarimetry—the measurement of the degree and direction of oscillation of electromagnetic waves—has long been recognized as a powerful diagnostic tool in X-ray astronomy. Unlike conventional spectroscopy or imaging, polarization measurements provide two additional parameters: the polarization degree, which relates to the asymmetry of the emitting region, and the polarization angle, which reveals its orientation. Together, these quantities can distinguish between competing models of how radiation is produced near compact objects, from accretion disks around black holes to the magnetospheres of neutron stars.

Despite this promise, X-ray polarimetry remained largely unexplored for decades. Early attempts in the 1970s, aboard sounding rockets and early satellites, yielded only upper limits for most sources, with notable exceptions such as the Crab Nebula and Cygnus X-1. It was not until 2021, with the launch of NASA’s Imaging X-ray Polarimetry Explorer (IXPE), that sensitive polarization measurements became routine—but only in the soft X-ray band of 2 to 8 kiloelectronvolts (keV). The hard X-ray regime, above roughly 20 keV, remains poorly charted, with only a handful of measurements from instruments such as PoGO+, X-Calibur, INTEGRAL, and AstroSat’s CZT Imager, most limited by poor photon statistics.

The fundamental challenge is twofold. Hard X-ray photons are scarce—fluxes fall steeply with energy—and the physical process by which polarization can be measured, Compton scattering, requires the reconstruction of the azimuthal distribution of scattered photons. When a polarized photon scatters, it preferentially scatters perpendicular to its electric field vector; by measuring many such scattering events, the degree and angle of polarization can be inferred. But each measurement demands two simultaneous interactions—a scatter followed by absorption—and each introduces noise, thresholds, and systematic uncertainties.

The PRL team’s answer is the Compton X-ray Polarimeter (CXPOL), a compact focal plane instrument designed to sit at the focus of hard X-ray mirror optics of the kind flown on NuSTAR and Hitomi. The original prototype, first reported in 2015, used a plastic scintillator as the active scatterer, read out by a photomultiplier tube, surrounded by a cylindrical array of sixteen cesium iodide—CsI(Tl)—scintillator bars, each coupled at one end to a silicon photomultiplier (SiPM). The concept worked, but testing revealed a critical weakness: light collection in the long CsI(Tl) bars was so poor that the detectors were effectively sensitive only within a few centimeters of the SiPM end, leaving much of each 15-centimeter bar blind.

The new work tackles that limitation head-on with a redesigned absorber module built around a sodium iodide—NaI(Tl)—scintillator bar measuring 100 by 20 by 5 cubic millimeters, read out by SiPM arrays at both ends. NaI(Tl) offers roughly a fourfold improvement in scintillation decay time compared with CsI(Tl) while maintaining comparable light yield, meaning faster signal pulses and less accumulation of dark counts from the photosensors. The team also replaced the earlier KETEK SiPMs with SensL (now Onsemi) MicroJ-series devices, which provide a background rate about five times lower, improved photon detection efficiency, faster microcell recovery, and a spectral response well matched to NaI(Tl)’s emission peak.

The dual-end readout is the heart of the innovation. When an X-ray photon is absorbed somewhere along the length of the scintillator, the scintillation light divides between the two ends in a predictable way: the closer the interaction is to one SiPM, the larger the signal that detector registers. By taking the ratio of the signals from the two ends, the interaction position along the detector can be reconstructed. This achieves two things at once. First, sensitivity extends across the full length of the crystal rather than being concentrated near a single readout face. Second, the position information enables spectroscopy through Compton kinematics: knowing where the photon scattered in the absorber, combined with its interaction position in the scatterer, allows the polar scattering angle to be computed geometrically, and the incident photon energy can then be derived from the energy deposited in the absorber and the known scattering angle. This opens the door to simultaneous spectro-polarimetry, a mode of observation in which energy-dependent polarization measurements can disentangle the contributions of different emission components in complex sources.

Position sensitivity has a second, subtler benefit. For wide-field instruments using collimators rather than focusing optics, photons arriving from off-axis directions distort the azimuthal scattering response and introduce systematic errors in polarization measurements. With position-sensitive absorbers and scatterers, the true scattering angle can be reconstructed and events can be corrected—or weighted in favor of scatter angles near 90 degrees, where the response to polarization is strongest—substantially mitigating these systematics.

Before committing to a design, the team performed detailed optical Monte Carlo simulations using the Geant4 toolkit. A key input was the reflectivity of the PTFE (Teflon) wrapping surrounding the crystal. Although PTFE’s intrinsic reflectance exceeds 95 percent across the visible range, packaging imperfections reduce the effective value; calibrating their simulations against measurements of the earlier CsI(Tl) detectors, the team settled on a realistic effective reflectivity of 80 percent. Simulations of the 10-centimeter NaI(Tl) bar, irradiated at nine positions with photons at 20, 60, and 122 keV, indicated that the detector should remain sensitive along its entire length down to incident energies of 20 keV, provided the equivalent noise background stays below roughly 100 optical photons.

Because NaI(Tl) is hygroscopic, the researchers procured a hermetically sealed module from Advatech UK: the crystal, wrapped in PTFE on all faces except the ends, sits inside an aluminum housing with a 0.2-millimeter carbon entrance window. Each end couples through a quartz window—part of the moisture-proof sealing—to an array of three 6.07-by-6.07-millimeter SiPMs connected in series, with all optical interfaces coupled by a thin layer of optically clear silicone gel chosen for its intermediate refractive index. Aluminum caps shield the sensors from stray light.

The custom readout electronics consist of two boards. A front-end board houses charge-sensitive preamplifiers, built around LM6172IM operational amplifiers, followed by CR-RC-RC shapers that produce Gaussian-shaped pulses with a peaking time of about 4 microseconds. A second FPGA-based board handles triggering, peak detection using AMPTEK PH300 devices, and digitization with a 12-bit ADC. Events from the two ends are recorded in coincidence within a 1-microsecond time window, a scheme that suppresses uncorrelated background. Total electronics dead time per event is 13.2 microseconds, corresponding to a count rate capability of roughly 76 kilohertz—far above what even the brightest on-axis sources would deliver, ensuring negligible dead time losses.

For characterization, the team illuminated the detector with a collimated americium-241 X-ray source emitting 59.54-keV photons, stepped in millimeter increments along the crystal’s length inside a light-tight dark box. The results are encouraging. With two-end coincidence readout, the detector achieves uniform sensitivity along at least 60 percent of its length at 60 keV, with the interaction position recovered to an average resolution of about 1.5 centimeters and a mean energy resolution of 35 percent at 59.54 keV. Critically, the coincidence requirement reduces the effective SiPM background per absorber by an order of magnitude compared with single-ended readout, directly enhancing the polarimetric sensitivity of the full instrument.

The team identifies several paths for improvement before the detector flies. Extending detection efficiency down to 20 keV will require lowering the trigger threshold through improved optical coupling and faster front-end electronics—possibly custom application-specific integrated circuits designed for SiPM readout, which would also allow simultaneous readout of many detectors and tighter coincidence windows for better background rejection. Faster scintillators such as GAGG or CeBr3, with comparable or higher light yield, are also under consideration. For space applications, the modules will face thermal cycling and high-humidity tests to validate hermeticity.

The broader payoff could be substantial. With hard X-ray focusing optics now mature, compact focal plane polarimeters can exploit the narrow field of view of a telescope to slash instrumental background while concentrating scarce photons onto an optimized detection geometry. The XL-Calibur balloon experiment recently demonstrated the power of this approach, measuring the polarization of the Crab nebula at an unprecedented 8.6-sigma significance. A flight-ready CXPOL-class detector, with position-sensitive absorbers providing simultaneous spectroscopy, would give future Indian astronomical missions—and potentially small-satellite demonstration flights—a sensitive new window onto the most violent processes in the universe, from the accretion flows around stellar-mass black holes to the engines powering relativistic jets far across the cosmos.