The Psyche Gamma-Ray and Neutron Spectrometer

Abstract

A Gamma-Ray and Neutron Spectrometer (GRNS) instrument has been developed as part of the science payload for NASA’s Discovery Program Psyche mission to the M-class asteroid (16) Psyche. The GRNS instrument is designed to measure the elemental composition of Psyche with the goal to understand the origin of this mysterious, potentially metal-rich planetary body. The GRNS will measure the near-surface abundances for the elements Ni, Fe, Si, K, S, Al, and Ca, as well as the spatial distribution of Psyche’s metal-to-silicate fraction (or metal fraction). These measurements address three of the five Psyche mission science objectives: determine if Psyche is a core; determine whether small metal bodies incorporate light elements into the metal phase; and determine whether Psyche was formed under reducing conditions. The Gamma-Ray Spectrometer (GRS) uses a cryocooled, high-purity Ge (HPGe) sensor to detect cosmic-ray generated gamma rays in the 60 to 9000-keV energy range. The HPGe sensor is surrounded by a borated plastic anticoincidence shield that provides three functions: active background rejection from charged particle interactions in the HPGe sensor; fast neutron measurements; and direct measurements of the incident galactic cosmic ray flux. The Neutron Spectrometer (NS) uses three ³He gas proportional sensors, each with different material wraps to measure thermal (<0.4 eV), low-energy epithermal (0.4 eV to 1 keV), and high-energy epithermal (up to 100 keV) neutrons. This paper provides an overview of the Psyche GRNS, including: its science and measurement objectives; the design of the instrument hardware, software, and operation; pre-launch performance measurements and its initial performance in space; and an overview of its data products and expected operation for different Psyche mission phases.

Introduction

NASA’s Psyche mission will carry out orbital reconnaissance of main-belt asteroid (16) Psyche (hereafter called Psyche), which belongs to the M spectral class, a group of solar system objects whose red-sloped, featureless reflectance spectra are consistent with laboratory measurements of metal-dominated surfaces (Vernazza and Beck 2017). Psyche was chosen for detailed study with the expectation that it might represent the exposed core of a proto-planetary body (Elkins-Tanton et al. 2017, 2020). To accomplish its science objectives, the Psyche spacecraft carries a payload of three instruments: a magnetometer (Weiss et al. 2023, this collection), redundant multispectral imagers (Bell et al. 2025, this collection), and a gamma-ray and neutron spectrometer (GRNS) (Fig. 1). The mission will also carry out gravity science using its radio system (Zuber et al. 2022, this collection).

Fig. 1.

View of the Psyche spacecraft highlighting the locations and relative dimensions of the GRNS and the spacecraft coordinate system. The GRS and NS are located on -Y boom, with the GRS at the end of the boom and NS at the boom midpoint. The two GRNS DPUs are located on inside the spacecraft on the -Y panel, which is illustrated by the DPU label pointing to their location. Additional components of the Psyche spacecraft are described by other papers in this collection (e.g., Polanskey et al. 2025, this collection)

The goal of the Psyche GRNS is to quantify Psyche’s surface elemental composition using the established technique of planetary nuclear spectroscopy, where cosmic-ray generated gamma rays and neutrons are passively measured from orbit in close proximity to an airless body (Reedy 1978; Evans et al. 1993; Feldman et al. 1993). Each measured element produces a unique “fingerprint” of gamma-ray energy lines that identifies the element that produced them. The number of gamma rays at these characteristic energies is used to quantify the abundance of that element. Neutrons are measured in broad energy ranges, i.e., not specific to a given element, but where each energy range is sensitive to a different mix of elements such that compositional information can be inferred.

Planetary nuclear spectroscopy is a mature technique, having made compositional measurements of multiple planetary bodies such as the Moon, Mars, Mercury, and the asteroids Eros, Vesta, and Ceres (e.g., Goldsten et al. 1997; Boynton et al. 2004; Feldman et al. 2004; Goldsten et al. 2007; Prettyman et al. 2011, 2019). On the Psyche mission, the gamma-ray spectrometer (GRS) will quantify the elemental abundances for the following elements: Fe, Ni, Si, K, S, Al, and Ca. These abundance measurements provide key information that address three of the five Psyche mission objectives (see Sect. 2). Of particular importance is Psyche’s Ni abundance, which is an element that has yet to be remotely measured on a planetary surface with gamma-ray spectroscopy but will be key to determining Psyche’s evolutionary history. Because of the importance (and challenge) of obtaining a robust Ni composition measurement, this Ni composition requirement drives much of the GRS design, particularly the need to use a high-purity Ge gamma-ray sensor that has the excellent energy resolution needed for separating the key Ni gamma-ray line from a nearby K line.

The neutron spectrometer (NS), in conjunction with the multispectral imager, will quantify the metal-to-silicate fraction. For Psyche, Ni, C and S are expected to partition with the metallic phase, as would the majority of Fe. Sulfur would be expected to occur in sulfides, primarily troilite (FeS), and carbon as graphite. In contrast, major rock-forming elements (Si, O, Al, Mg, Ca) are expected to partition into the silicates. The relative amount of reduced (metallic) to oxidized Fe can be deduced from the total Fe measured by NS and the composition of FeO-bearing mafic silicates constrained by depth of the absorption bands of olivine and pyroxene from the multispectral imager. Thus, Fe, Ni, and S present in the metallic portion can be distinguished from Si, Mg, Fe, Al and Ca present as oxides in the silicate portion. Psyche might also have spatial variations of H content on its surface (200 to 300 ppm) (Reddy et al. 2018), which can be quantified by the NS. In addition to its own scientific interest, H abundance variations can affect the neutron composition parameters as well as the gamma-ray data, and thus such variations need to be understood. More generally, neutron flux information is useful for understanding and carrying out corrections for neutron capture and inelastic gamma-ray measurements (e.g., Lawrence et al. 2002; Prettyman et al. 2006).

Planetary nuclear spectroscopy measurements have a few foundational top-level requirements that drive the mission and spacecraft design. Most importantly, gamma-ray and neutron signals have inherently low counting rates – fractions of counts per minute per elemental line for gamma rays, and up to a few counts per second for neutrons – and therefore require many days of accumulation to achieve statistically meaningful results. Second, since the sensors are omnidirectional with a full hemispherical view of the planetary body, low altitudes of roughly one planetary radius (or lower) are needed to detect sufficient quantities of planet-originating gamma rays and neutrons (Peplowski 2016). Finally, it is advantageous to place the gamma-ray and neutron sensors on an extension (i.e., fixed boom) away from the primary spacecraft material. The reason is that spacecraft materials generate background gamma rays and neutrons from the same processes that generate planetary gamma rays and neutrons. While some missions have successfully used body mounted instruments due to accommodation and resource constraints (e.g., Goldsten et al. 2007; Prettyman et al. 2011), the high precision Ni measurement needed at Psyche drove the design to require a boom (see Fig. 1). Based on the same “one-body radius” principle, the spacecraft-generated background can be substantially reduced when the sensor is located at least one spacecraft radius away from the spacecraft. The Psyche spacecraft is roughly a cube, 2.5 m on a side, giving an effective “radius” of ∼2 m, which sets the 2-m boom length.

The GRNS contains four primary components: the GRS, the NS, and two Data Processing Units (DPUs), one dedicated to each of the two sensors. The GRS inherits its basic functional design from the MESSENGER GRS (Goldsten et al. 2007) and, similarly, uses a cryocooled high-purity germanium (HPGe) gamma-ray sensor to measure gamma rays with optimum energy resolution. A new type of pulse-tube cryocooler will provide longer operational life compared to the rotary cryocooler used for MESSENGER. The gamma-ray sensor is surrounded by a plastic scintillator anticoincidence (AC) shield that serves three primary functions: active background rejection, fast neutron detection, and galactic cosmic-ray (GCR) monitoring. The plastic scintillator also provides an auxiliary function as a gamma-ray burst monitor. Since the GRS is most sensitive to spacecraft backgrounds, it is located at the end of the boom. The NS inherits its basic functional design from the Lunar Prospector NS (Feldman et al. 2004) and uses three ³He gas proportional counters to measure neutrons in three separate energy ranges. The NS is placed halfway up the boom at ∼1 m from the spacecraft deck. The DPUs are located within the spacecraft body and are connected to each of the sensor components via harnesses attached along the boom.

The flight model GRNS was delivered to the NASA Jet Propulsion Laboratory (JPL) in August 2021 for integration on the Psyche spacecraft, and the spacecraft was launched on 13 October 2023. Subsequent sections describe the design and performance of the delivered GRNS, including its comprehensive performance test (CPT) on the Psyche spacecraft, as well as its initial performance in space. All portions of the GRNS are presented, including the mechanical and electronic aspects of the sensors, as well as an overview of its firmware and software. We also present the end-to-end functional performance of the GRNS components, a description of the GRNS data products, and a summary of the initial cruise operation as well as the expected operation of the GRNS both during the remaining portion of cruise and while in orbit at Psyche.

Overview of Elemental Composition Measurements on the Psyche Mission

It has been proposed that Psyche could be the stripped iron core of a differentiated protoplanet, in which case the exploration of Psyche offers an unique opportunity to provide valuable insights into planetary formation processes and test our theories for core formation for terrestrial planets (Elkins-Tanton et al. 2022, this collection). The Psyche mission has five science objectives: A) Determine whether Psyche is a core, or if it is primordial unmelted material; B) determine the relative ages of regions of Psyche’s surface; C) determine whether small metal bodies incorporate the same light elements into the metal phase as are expected in the Earth’s high pressure core; D) determine whether Psyche was formed under conditions more oxidizing or more reducing than Earth’s core; and E) characterize Psyche’s morphology (Elkins-Tanton et al. 2022).

The GRS is required to measure the abundances for the elements Fe, Ni, Si, K, S, Al, and Ca. GRNS data will also measure the bulk composition parameters of macroscopic neutron absorption (${\mathrm{\Sigma }}_{\mathrm{a}}$) (Feldman et al. 2000) and average atomic mass (<A>) (Gasnault et al. 2001) from which Psyche’s metal-to-silicate fraction can be determined. Table 1 lists the elements and composition parameters that will be measured with the GRNS, along with the required detection limits, precisions, and expected accumulation times needed to make each measurement. The GRNS composition measurements directly address three of the five Psyche mission science objectives (A, C, and D). For objective A, all of the compositional measurements contribute towards the determination if Psyche is a core. Ni concentrations in the metal phase below ∼4 wt.% are consistent with Psyche not being a product of core formation processes (Elkins-Tanton et al. 2022). If the Ni concentration is ∼4 wt.% or higher, then the measured value provides information about how core-like material may have formed (Elkins-Tanton et al. 2022), as well as the oxidation or reduction state of its material (McCoy et al. 2022). If the Ni concentrations are very low (i.e., ≪4 wt.%) along with high metal (e.g., Fe) abundances, such a measurement would indicate that Psyche could be a metal-rich body that may never have melted, and thus represent highly reduced, primordial metal. Alternatively, high Ni concentrations (>12 wt.%) are uncommon in iron meteorites and occur in irons that originated from oxidized cores, where metallic Fe is oxidized to FeO and incorporated in the mantle, leaving a Ni-rich core. Such a high Ni concentration would also indicate a low or very low silicate concentration, as Ni concentrations in silicates are typically at the part per million levels. A primary means for measuring Ni abundances is via a 1454-keV gamma ray produced by an inelastic scatter reaction between a fast neutron and a Ni nucleus. Figure 2 shows simulated gamma-ray spectra, which provides an example of the type of data that can be acquired with the GRS in order to quantify Ni abundances on Psyche’s surface.

Table 1.

List of elements and composition parameters measured with the GRNS, along with required detection thresholds, precision, and expected accumulation time in Orbit D to meet requirements

Element	Gamma-ray line or neutron energy band	Required detection threshold	Required relative precision		Expected nadir-pointing time at altitudes <1 Psyche radius to meet global requirement (days)
			Global	Mapped on 200 km2 area
Alb	2211 keV	1.3 wt.%	20%	-	22
Ca	4438 keV	n/a	-	-	-
Cab	1943 or 3736 keV	0.3 wt.%	20%	-	52
Fe	846 keV	4 wt.%	20%	20%	0.8
Ha	Low-energy/high-energy epithermal	n/a	-	-	-
K	1460 keV	200 ppm	20%	20%	4
Mga	1369 keV	n/a	-	-	-
Ni	1454 keV	4 wt.%	20%	-	36
Pa	1266 keV	n/a	-	-	-
S	2232 keV	3 wt.%	20%		46
Si	1778 keV	2.5 wt.%	20%	20%	16
Tha	2615 keV	n/a	-	-	-
Ua	609 keV	n/a	-	-	-
Metal-to-silicate fraction	Thermal, fast via ${\mathrm{\Sigma }}_{\mathrm{a}}$ and <A>	n/a	-	0.02	Few days

^aNot required to meet mission science objectives but detectable if present at greater than ∼1 wt.% (C, Mg, P), ∼10 ppm (H), ∼1 ppm (Th), and ∼25 ppb (U). ^bRequirements are met either with the Al or Ca measurement.

Fig. 2.

Simulated gamma-ray spectra for a 100-day-long Orbit D at Psyche. For this orbit, we assumed an altitude of one Psyche radius with a 70% duty cycle; see Sect. 7 for more discussion of duty cycle when in Psyche orbit. Panel A shows combined Psyche-originating (black) and background (red) gamma rays. Psyche-originating gamma rays are calculated for an assumed Psyche composition that includes 85 wt.% Fe, 4 wt.% Ni with the remaining material being Mg-rich orthopyroxene. The Psyche-originating gamma rays are propagated through the GRS sensor using estimated performance parameters for the Psyche HPGe crystal size and an energy resolution of 3.5 keV full-width, half maximum at 1332.5 keV. This energy resolution was set as the pre-launch energy resolution requirement based on the best pre-radiation damage performance of the MESSENGER GRS (see Sect. 4.1). The gamma-ray and continuum background was estimated using the measured background from the MESSENGER GRS (Evans et al. 2012). Panel B shows the Psyche-originating spectra minus the background for different amounts of assumed Ni abundances. Gamma rays for different elemental lines including background (BG) are labeled

The elements Si, Al, K, and Fe help discriminate whether any non-metal portions of Psyche’s surface are chondritic or achondritic (Elkins-Tanton et al. 2020). Earth-based measurements show evidence that Psyche may have large-scale (hemispherical) metal-to-silicate heterogeneity (Sanchez et al. 2017; Takir et al. 2017; Shepard et al. 2021; Cambioni et al. 2022; Hasegawa et al. 2024), suggesting that GRNS measurements may likewise reveal compositional heterogeneity. Due to better statistical precision, the neutron-based composition parameters will provide the highest spatial-resolution GRNS maps, particularly for Psyche’s metal-to-silicate fraction. For objective B, composition measurements of Si, K, and S will provide information about light-element partitioning in the core, if present, or possibly of unmelted (i.e., primordial) metal-rich material. Finally, for objective D, Ni composition measurements will provide information about oxidizing and reducing conditions such that oxidizing conditions are likely required to produce Ni abundances greater than 12 wt.%, and reducing conditions are likely required to produce low Ni abundances (McCoy et al. 2022, this collection).

Recent Earth-based measurements and new interpretations of existing measurements have updated our understanding of the nature of Psyche (Elkins-Tanton et al. 2020). The currently accepted density of Psyche is bounded to be 3.4–4.1 g/cm³ (Elkins-Tanton et al. 2020) with a recent value of 4.172 ± 0.145 g/cm³ having been reported (Farnocchia et al. 2024). It is therefore now understood that Psyche can have a broad range of possible compositions. Psyche’s Fe abundances could plausibly range from 25 wt.% to upwards of 90 wt.%, and by inference, the silicate fraction could range from less than 0.1 to greater than 0.7. Thus, instead of assuming Psyche may be an analog of iron meteorites, a wider range of possible compositions can be considered that match the density of Psyche, including analogs such as CB and enstatite chondrites, pallasites, and mesosiderites (Peplowski et al. 2019b), although spectral reflectance considerations suggest that enstatite chondrites and mesosiderites may be a less likely match to Psyche (Dibb et al. 2023).

For meteorites, the metal-to-silicate fraction is generally defined as the total amount of metal (e.g. metallic iron, nickel) versus total silicates (elements bound as oxides). In this scenario, iron in silicates (e.g. FeO) is “book kept” with silicates. In contrast, GRNS measures elements without regard to their mineralogical state. Since the largest concentration of “metal” elements being considered consist of Fe and Ni, “metal” content is defined as the total concentration of Fe and Ni, and “silicates” are defined as other major rock-forming elements (Si, O, Al, Mg, Ca, etc.)

The lower density value for Psyche detailed in (Elkins-Tanton et al. 2020) requires either less metal than originally thought, and/or significantly higher porosity than is typical for >100-km-diameter asteroids (e.g., Zhang et al. 2022). For the low metal case, measurements of additional elements beyond that originally planned could be important for fully characterizing Psyche’s bulk elemental composition. For example, for the meteorite types listed above, Mg concentrations can provide important discrimination between different composition types (Peplowski et al. 2019b). While Mg is not a required element for the baseline Psyche mission science plan, the GRS sensor was fabricated without magnesium materials to enable Mg measurements to be accomplished with no changes to the instrument design or operations plan. Similarly, other elements that could be detected include C and P, if present at high enough concentrations (e.g., greater than 1 wt.%).

Another element that has gained increased importance at Psyche is H. Ground-based measurements using the NASA Infrared Telescope Facility have detected a 3-μm absorption on Psyche that is consistent with hydration features attributed to OH and H₂O bearing phases (Takir et al. 2017). Because Psyche likely has no extensive permanently shaded regions, and thermal conditions do not allow for water ice to be stable at its surface, this detection suggests that, as with the asteroid Vesta (Prettyman et al. 2012), Psyche may also have significant amounts of water-rich carbonaceous-chondritic material accumulated on its surface. Based on a cross correlation of Earth-based data from both Vesta and Psyche with orbital H measurements at Vesta, Reddy et al. (2018) estimated that the bulk hydrogen concentrations could be in the range of 200 – 300 ppm, which is comparable to the range of H variations (0 – 400 ppm) seen at Vesta (Prettyman et al. 2012; Lawrence et al. 2013b). While H measurements are not needed to directly address the mission’s science objectives, H abundances in this range can affect the measured neutron composition parameters (Lawrence et al. 2013b; Prettyman et al. 2013), and possibly gamma-ray data when making composition measurements using both capture and inelastic gamma-ray lines (Peplowski et al. 2015a; Wilson et al. 2019). Thus, effects from bulk H concentrations and its possible spatial variability need to be taken into account with the measured GRNS data. The Psyche NS was specifically designed to be sensitive to a large range of possible metal content on Psyche (i.e., Fe and Ni) as well as possible H abundances greater than a few tens of ppm. To illustrate, Fig. 3 shows simulated neutron spectra for different metal-to-silicate content and H abundances. In typical silicate dominated compositions, variations of thermal (neutron energy $E_{\mathrm{n}}<0.4\text{ eV}$) and low-energy epithermal (0.4 eV < $E_{\mathrm{n}}<\sim$100 eV) neutrons are most sensitive to neutron absorption and H concentrations, respectively (Feldman et al. 2000; Lawrence et al. 2006). In contrast, when there are large concentrations of the neutron absorbers Fe and Ni (e.g., greater than Fe + Ni ∼30 wt.%), these elements will strongly suppress thermal neutrons and shift H-dependent variations to higher energies (Fig. 3B). Thus, in addition to sensors that measure thermal and low-energy epithermal neutrons, the NS has a third sensor that is sensitive to higher energy epithermal neutrons (Sect. 3.3).

Fig. 3.

Calculated neutron lethargy spectra for different assumed Psyche compositions. Lethargy (flux times energy) flattens out the otherwise steep power law spectra. Panel A shows the fluxes for compositions ranging from 10 volume % metal to 90 volume % metal. Panel B shows fluxes for a 90 volume % metal and variable hydrogen abundances from 0 to 500 ppm H. The vertical dashed lines are located at the lower energy boundaries for epithermal (0.4 eV) and fast neutrons (0.7 MeV)

Finally, Psyche’s lower-than-anticipated density means that a wider range of metal and silicate abundances need to be considered. Of primary importance is the relative proportions of Ni and K concentrations, which have closely spaced gamma-ray lines at 1454 keV and 1461 keV, respectively (see Fig. 2). For example, if the metal fraction of Psyche is significantly lower than originally expected, then its Ni concentration will likely be correspondingly lower, and K (which can be enhanced in silicate materials) could be correspondingly higher, which can make separating the two lines more challenging. This possible condition reinforces the need for the excellent energy resolution provided by the HPGe sensor.

GRNS Instrument Design

GRNS Overview

The GRNS was designed and built by the Johns Hopkins University Applied Physics Laboratory (APL) in collaboration with the Lawrence Livermore National Laboratory (LLNL). LLNL designed and built the GRS detector and cryostat assembly. APL designed and built the AC shield, the NS sensor assembly, the GRS and NS front-end electronics, and the GRS and NS DPUs. The cryocooler for the GRS detector was designed and built by the Lockheed Martin Advanced Technology Center. The overall design and functionality of the GRNS follows closely the MESSENGER GRNS, which successfully made compositional measurements of planet Mercury (Peplowski et al. 2011; Evans et al. 2012; Peplowski et al. 2012; Lawrence et al. 2013a; Peplowski et al. 2014; Evans et al. 2015; Peplowski et al. 2015b,c, 2016; Lawrence et al. 2017). The similarity with MESSENGER includes an HPGe sensor with plastic scintillator anticoincidence shield, separate neutron spectrometer, and separate DPUs for the two sensor subsystems, where APL “slice” heritage (i.e., modular boards that couple together to form a compact electronics stack) is used for the DPU boards. The AC shield fast-neutron signal processing is largely inherited from the MESSENGER NS fast-neutron measurements.

Despite these similarities, a number of key changes were made to the GRNS design. These changes were made to accommodate different requirements unique to the Psyche mission design as well as incorporate lessons learned from the MESSENGER GRNS. Also notable is that when the Psyche GRNS started its conceptual design phase in 2016, it had been almost 15 years since the initial design work on the MESSENGER GRNS. As a consequence, significant redesign of the electronics and software was needed to bring the instrument up to date (e.g., due to parts obsolesce) and to accommodate new mission assurance requirements.

For the GRS, there were important lessons learned from MESSENGER that resulted in improvements to the Psyche design. In particular, six months into the primary mission, the MESSENGER HPGe detector started to develop leakage currents that degraded its energy resolution performance in the final months of operation (Evans et al. 2017). In addition, the radiation damage that was generated by being exposed to 6.5 years of GCRs during cruise was never fully reversed despite extended annealing at 85 °C, resulting in an unanticipated degradation in energy resolution performance. There were attempts made to mitigate both these effects in flight but, ultimately, the bearings in the rotary-style cryocooler reached end of life and gave out during MESSENGER’s first extended mission ending its HPGe sensor operation.

To address the unexpected leakage currents in the MESSENGER HPGe sensor that scaled with total anneal time (Evans et al. 2017), we incorporated two changes to the Psyche cryostat: a passivation layer was added to the HPGe crystal and the detector encapsulation was vented. Passivation is now a standard process for the manufacture of HPGe crystals as a way to stabilize its highly reactive surface, thereby reducing its susceptibility to contamination that can lead to leakage currents. The MESSENGER HPGe crystal was not passivated. Instead, it was hermetically sealed in an aluminum encapsulation filled with two atmospheres of clean dry nitrogen, where the nitrogen gas was intended to occupy surface sites and therefore act as a barrier to contamination. However, the repeated high-temperature annealing attempts of the MESSENGER GRS likely generated an excessive build-up of internal outgassing products within the sealed enclosure itself that became trapped, ultimately overwhelming the intrinsic detector surface to the point of producing significantly elevated detector leakage currents that, in turn, degraded system energy resolution. As a way to avoid this catastrophic scenario, the Psyche design switched to a “vent-to-space” encapsulation scheme. Because the HPGe surface is now passivated, it is relatively insensitive to contaminants, making a hermetically sealed enclosure no longer necessary. Long-term exposure to air has not been shown to be a problem, as was previously demonstrated on INTEGRAL, where their detectors were vented approximately two years prior to launch (Vedrenne et al. 2003). Testing of the Psyche GRS vented design has also shown stable performance after long exposures to air for periods up to 9 months.

To address the incomplete annealing observed on MESSENGER, we carried out a comprehensive study to determine quantitatively the most effective annealing strategy for recovering full performance from radiation-damaged HPGe detectors. The results of this study, reported by Peplowski et al. (2019a), showed that higher temperatures (105 °C) than used by MESSENGER were required to achieve full recovery following irradiation for selected mission scenarios, a result consistent with previous investigations (Albernhe et al. 2002) and verified with our particular detector geometry and predicted radiation dose. We have fully incorporated this revised annealing strategy into our operational plans.

Compared to MESSENGER, the Psyche GRS uses a fully redesigned cryostat that includes several new thermal and mechanical features based on many years of improvements since the MESSENGER design (Burks et al. 2020). The new design is completely separable from the AC shield, which streamlines assembly and facilitates stand-alone testing of each subsystem. Improved energy resolution was achieved by mounting the preamplifier front-end circuit board directly to the detector encapsulation inside the cryostat for better noise performance and less susceptibility to microphonic pick-up of vibrations from the cryocooler.

Psyche also took advantage of advances in electronic parts available for spaceflight missions to improve performance. MESSENGER was limited in the number of digital logic gates available for signal processing, and so relied on a hybrid analog/digital signal processing scheme for pulse height analysis. The Psyche GRS implements a fully digital design, which eliminates analog filtering stages and their associated temperature dependence and adds the capability to adjust filter shaping parameters in flight to optimize energy resolution. The improvement in pre-launch performance – less than 2.0 keV full-width, half-maximum (FWHM) for Psyche (Sect. 4.1) compared to 3.5 keV FWHM for MESSENGER (Goldsten et al. 2007) – is largely due to these improvements.

Another significant change from the MESSENGER design is the replacement of its limited-life rotary cryocooler with a long-life pulse-tube cryocooler. While a pulse-tube cryocooler is less efficient than a rotary cryocooler, its lower level of exported vibration allows the signal processing chain to reach the system noise floor even under ambient thermal conditions, achieving laboratory-grade energy resolution performance in a fully qualified spaceflight instrument. The virtually unlimited life of a pulse-tube cooler ensures the GRS will be ready for any exciting extended mission opportunities the Psyche mission may bring.

For the NS, the Psyche design returns to the use of ³He neutron sensors like those used on Lunar Prospector (Feldman et al. 2004) rather than the scintillator-based sensors used for MESSENGER. The MESSENGER scintillator sensors were designed to take advantage of the high orbital velocity of the MESSENGER spacecraft around Mercury (∼few km/s) and use a Doppler filter technique (Feldman and Drake 1986) to discriminate neutron energies. The spacecraft velocity around Psyche is much lower (<200 m/s) and thus the Doppler filter technique is not appropriate at Psyche. Further, ³He neutron sensors are robust and reliable sensors and were easily incorporated into the Psyche design by adopting a straightforward mounting approach.

Separate GRS and NS harness bundles connect the sensor assemblies at their respective boom locations to their DPUs. The DPUs mount to the −Y spacecraft panel on the inside of the spacecraft. Harness cables from each DPU are feed-through bulkhead connectors on the + Z deck to maintain the spacecraft Faraday cage.

Taken together, the GRNS employs five separate sensors to make its measurements: a HPGe gamma-ray sensor, a plastic scintillator AC shield, and three ³He gas proportional counter neutron sensors. These five sensors combine to produce 10 different types of measurements, summarized in Table 2. The rest of Sect. 3 describes further details of each sensor assembly, the DPUs, the software and firmware that controls the GRNS, and a summary of the environmental tests that were carried out on the hardware prior to delivery to the spacecraft.

Table 2.

Types and characteristics of measurements from the five GRNS sensors

Measurement Type	Sensor	Characteristics
High-gain gamma-ray spectra (separate accepted & rejected)	HPGe sensor	60 – ∼3000 keV gamma-rays
Low-gain gamma-ray spectra (separate accepted & rejected)	HPGe sensor	60 – ∼9000 keV gamma-rays
Fast neutrons	AC shield, high gain	0.7 – 5 MeV neutrons; double pulse interaction
Broad-band epithermal neutrons	AC shield, high gain	Few hundred to 0.7 MeV neutrons; single pulse interactions within the ∼90 keV 10B(n,α) peak
Galactic cosmic-rays	AC shield, low gain	0.5 MeV to ∼35 MeV, mostly protons
Gamma-ray burst detection	AC shield, high gain; after triggering a set threshold	1024-sample, 50-ms resolution profile of integrated counts of gamma-rays falling within an energy window ∼20 keV to ∼100 keV (commandable upper limit)
Slow neutrons	Bare 3He sensor	0 – few hundred keV neutrons
Low-energy epithermal neutrons	Cd-covered 3He sensor	0.4 eV – few hundred keV neutrons
High-energy epithermal neutrons	Polyethylene-covered 3He sensor	∼1 keV – few hundred keV neutrons
Thermal neutrons	Bare and Cd-covered 3He sensor	Count rate difference of bare and Cd-covered sensors

Gamma-Ray Spectrometer (GRS)

A cutaway view of the GRS is shown in Fig. 4; photographs of the GRS are shown in Fig. 5; a functional block diagram of the GRS is shown in Fig. 6; detector characteristics and resource use are given in Table 3. The GRS consists of multiple components located on a single mounting deck. At the inside of the GRS is the HPGe sensor, which is housed within the cryostat. The cryocooler (Sect. 3.2.3) and three sensor electronics boxes (Sect. 3.2.2) are connected to the cryostat and perform various functions. The cryostat is surrounded by the AC shield, which contains the plastic scintillator sensor and photomultiplier tube (PMT). The combined GRS/AC shield housing is attached to the mounting deck that then attaches to the spacecraft boom structure via three titanium “nodes” with integral flexures (Fig. 5B).

Fig. 4.

Cutaway (A) and exterior (B) views of the GRS. Different sensor components as described in the text are labeled. Figure adapted from Peplowski et al. (2025)

Fig. 5.

Photographs of the FM GRS on a test stand. Panel A and B show the front and back of the GRS, respectively

Fig. 6.

GRS functional block diagram. All instrument components that are located on the boom are shown within the large dashed rectangle. The components for the five-slice DPU are shown on the right side of the block diagram. Individual components within each DPU slice are labeled. Additional details are given in the text

Table 3.

GRS characteristics and resource use

Primary sensor	High-purity germanium (HPGe) coaxial detector, 5 cm × 5 cm beveled cylinder; inner hole is 8 mm diameter; 3.8 cm deep
Anticoincidence shield	Borated plastic scintillator (BC454) cup coupled to a 7.6-cm PMT;
	Measured scintillator mass: 1.63 kg annulus 12.9 cm inner diameter; 16.5 cm outer diameter; 1.8 cm thick puck: 4.92 cm tall; diameter is 7.7 cm at the PMT and 12.9 cm at the annulus
Photomultiplier tube (PMT)	Hamamatsu R6233-01; 76-mm diameter
Effective field-of-view	4π omnidirectional; full view of Psyche asteroid
Cryocooler	Lockheed Martin pulsed-tube cryocooler, ∼500 mW heat lift
HPGe operating voltage	<3000 V, leakage current <10 pA
Gamma-ray energy range	60 keV to 9000 keV
Neutron energy range	Broad-band epithermal neutron: <0.7 MeV
	Fast neutron: 0.7 MeV to ∼10 MeV
Number spectral channels	16,384 HPGe; 512 AC high gain; 1024 AC low gain
Energy resolution (FWHM)	1.92–2.09 keV @ 1332 keV
Pulser resolution (FWHM)	Hardware high gain: 1.12 keV; rate: 1 to 10,000 Hz
	Hardware low gain: 1.32 keV; rate: 1 to 10,000 Hz
	Digital high gain: 1.12 keV; rate: any
	Digital low gain: 1.32 keV; rate: any
Intrinsic efficiency	0.212 @ 511 keV, 0.102 @ 1332 keV, 0.020 @ 6130 keV
Maximum throughput	>5,000 events per second
Background rejection ratio	Varies from 2:1 @ ∼1000 keV to >10:1 @ ∼8000 keV
Differential nonlinearity (DNL)	<1%
Integral nonlinearity (INL)	<3 channels (high gain); <2 channels (low gain)
Electronics gain drift	0.2 channels per °C
Preamplifier gain drift	0.1 channels per °C
Sensor thermal environment	−25 °C < Tradiator < 30 °C
Cool-down time to 90 K	<12 hrs, Tradiator @ 0 °C
HPGe crystal temperature regulation	<0.25 °C, −25 °C < Tradiator < 65 °C (transient conditions)
HPGe crystal annealing temperature	105 °C, regulated
GRS sensor mass	9.06 kg (including mounting deck and passive radiator)
GRS DPU mass	2.64 kg
GRS harness mass	4.44 kg (includes harness along boom and DPU to + z deck harness)
GRS total mass	16.14 kg
Power use	57.7 W peak (DPU: 10.6 W; cryocooler, sensor dissipation: 21.0 W; cryocooler, heater dissipation: 25.4 W; harness dissipation: 0.7 W)
Data rate	6000 bits per second Orbit D allocation (combined with NS)

The GRS mounting deck serves multiple purposes. First, it provides the mounting structure for the GRS/AC shield housing along with the other GRS components. These components include: a split-configuration pulse-tube cryocooler that is attached both on, and extends through the deck (Fig. 5B); a dual-armed copper heat strap that thermally connects the cryocooler warm flange to the deck; and a high-voltage filter box, which receives up to −3000 V from the DPU and attenuates any voltage ripple before feeding it to the sensitive HPGe detector (Fig. 4B). Second, the upper end of the deck contains a field joint bracket (labeled in Fig. 4B), which groups together several bulk-head connectors, simplifying installation and connection of the GRS sensor assembly to the boom harness carrying signal and power cables to the DPU. The backside of the field joint bracket contains bulk-head connectors into which the boom harness cables attach. This arrangement allowed the entire GRS and most of its sensor component connectors to be assembled at APL prior to delivery to the spacecraft. Thus, for spacecraft installation, the only harness mates that needed to be completed were those on the backside of the GRS and those to a spacecraft field joint bracket near the base of the boom that fed signals through to the DPU. Finally, the deck provides the mounting structure for two opposing bolt-on radiator wings that help reject waste heat from the cryocooler to space (exposed portions of the AC-shield housing also provide significant radiator area). The deck, the radiator wings, and the AC-shield housing are all made of AlBeMet, a metal matrix composite material comprised of aluminum and beryllium. AlBeMet has better thermal properties than aluminum and its increased stiffness improves mechanical response under launch vibration, as well as reduces the conducted vibration from the cryocooler to the HPGe sensor. The lower density and effective atomic number of AlBeMet also helps to minimize attenuation of gamma rays that pass through the AC shield. The next subsections describe details for each GRS component.

GRS Sensor and Cryostat

The central gamma-ray detector is a 5-cm-diameter, 5-cm-long coaxial HPGe crystal (Fig. 4A) that was manufactured by Mirion Technologies. The crystal is N-type germanium with reverse-electrode configuration, which offers better resistance to radiation damage (Pehl et al. 1979), and is specified with an impurity level that targets an initial depletion voltage of 2000 V. The bare Ge crystal is encapsulated within a gold-plated aluminum encapsulation that is highly polished to achieve low emissivity in the infrared. The encapsulated detector is then suspended using a Kevlar suspension inside a similarly plated and polished cryostat, forming a separable unit that can then be lowered into the well of the AC shield. The GRS cryostat (Fig. 7) is optimized for three goals: thermal isolation from the environment, mechanical protection of the germanium crystal against launch vibration loads, and high-resolution readout of the gamma-ray signal. More details regarding the development of the cryostat are given in (Burks et al. 2020).

Fig. 7.

Photograph of the GRS cryostat prior to delivery from LLNL to APL. The bottom lid (on left) shows initials of LLNL persons who contributed to the build of the cryostat. The iron meteorite Odessa is shown on the right side of the photograph

Thermal isolation is required to operate the germanium crystal at cryogenic temperature, typically 85 K to 95 K, while protecting against the ambient environment. In space, where liquid nitrogen is not an option, sufficient cooling can be achieved using a mechanical cryocooler. For terrestrial applications these coolers can have a mass of several kg and consume considerable power. The Psyche GRS uses a compact, low-mass, low-power, pulse-tube cryocooler (Sect. 3.2.3). The tradeoff is that its heat lift capacity can be as low as ∼500 mW at cryogenic temperatures. The design of the cryostat must therefore constrain the heat loss to within the capacity of the cooler. This requirement is achieved by combining two methods. First, the conductive heat loss is minimized by mounting the encapsulation on a low-conductivity Kevlar-29 suspension system. Second, infrared heat transfer is minimized by employing infrared (IR) shielding comprised of highly reflective, low-emissivity surfaces. The cryostat is vented directly to space so conductive heat transfer is negligible. The total heat load on the detector was measured at 380 mW ± 20 mW at 23 °C ambient temperature, well within the capacity of the cryocooler.

Mounted onto the detector encapsulation are two power Zener diodes serving as heaters. These are used to bake the system at ∼50 °C to desorb water from the low-emissivity surfaces. They are also used to anneal the detector at 105 °C to reverse accumulated radiation damage (Peplowski et al. 2019a). Zener diodes were selected as heaters because they operate at high voltage and low current (∼80 V × ∼20 mA), allowing for the use of low-conductivity, 0.25-mm diameter, NbTi wires, with minimal conductive heat loss. Two temperature-sensing diodes are mounted onto the detector encapsulation and cryocooler cold-tip, respectively, to allow for temperature feedback and control. By having temperature sensors on each end of the short, braided copper heat strap that makes the thermal connection between the cold finger and the encapsulation, one can directly measure and monitor the heat flow in or out of the detector once the conductance of the cold braid is calibrated.

The Kevlar suspension also serves an important secondary purpose: it holds the detector encapsulation firmly in place during launch. A BeCu wave spring maintains a preload on the encapsulation/Kevlar suspension system. The detector encapsulation contains an internal BeCu wave spring as well to hold the germanium crystal in place within the encapsulation. BeCu springs were selected over stainless steel to minimize the amount of Fe and Ni close to the detector, so as to reduce background gamma rays specifically from Ni. The elimination of Ni in these springs was part of a more general effort to reduce Ni in the Psyche spacecraft with the goal of minimizing background gamma rays from Ni (Bradford et al. 2022). Initial results of this effort are described in Sect. 5.1.

Of particular importance for the Psyche cryostat is optimizing the gamma-ray energy resolution to distinguish the important ⁵⁸Ni gamma-ray line at 1454 keV from potential interference from ⁴⁰K at 1461 keV. In the MESSENGER GRS, the three biggest limitations to resolution were caused by radiation damage, microphonic pickup from the cryocooler, and the pulse-processing electronics. Radiation damage is addressed in the Psyche GRS through an improved annealing procedure (Sect. 3.1). Microphonic pickup arises when vibrations in the system induce electrical noise in the sensitive front-end electronics. The pulse tube cryocooler was chosen (Sect. 3.2.3) primarily for its long life but also benefits from its reduced exported vibration. The most sensitive components of the preamplifier – the front-end junction-gate field-effect transistor (JFET) and feedback components – were mounted directly onto the detector encapsulation. This approach reduces the JFET gate-lead length and thus its susceptibility to microphonics. There is a slight noise penalty for operating the JFET at a temperature below optimum of ∼115 K, but the improved end-to-end system energy resolution with the cooler running justifies this trade.

The cryostat has several vacuum feed throughs that allow for the connection of the cryocooler, high voltage (HV), Zener and temperature sensors, and the preamplifier signal. Another port is used to mount a light shield (Hines et al. 2022). This light shield has been designed to allow the cryostat to vent to space while blocking sunlight from penetrating into the cryostat. Direct sunlight is a concern because the photons cause thermal excitation in the crystal resulting in leakage current. The light filter uses a torturous path, painted with a layer of low-reflectance Stycast (Loctite 2850FT), to reduce the ambient light by several orders of magnitude.

A low-activity ¹³⁷Cs source (5 nCi) is attached to the cryostat lid (Fig. 4B). This source enables energy and efficiency calibration of the detector in flight, which allows flight monitoring of the detector efficiency as it changes with each radiation damage/annealing cycle (Peplowski et al. 2019a). The low activity of the source ensures that it does not interfere with the measured signal from Psyche.

GRS Electronics Modules

The GRS preamplifier module attaches to one side port of the cryostat. Its overall topology and functionality are similar to that flown on MESSENGER (Goldsten et al. 2007), but with a few significant improvements. The front-end JFET and feedback components now reside inside the cryostat and operate at cryogenic temperature, improving noise performance. A BF862 JFET provides improved noise performance at lower power. Two low-noise amplifier stages were added to split the signal into low- and high-gain outputs, where the high-gain channel provides better energy resolution and finer bin spacing for the low-energy region spanning ∼100 keV to ∼3000 keV, a region that includes the critical Fe and Ni lines, specific to the Psyche mission. Since the charge-amplifier is direct-current-coupled to the detector, a buffered version of its output allows monitoring the detector leakage current, which is nominally <10 pA but can measure up to ∼4,000 pA, if necessary.

The Sensor Control Unit Module (SCUM) attaches to the opposite side port of the cryostat. It contains the readout circuitry for the cryogenic diode temperature sensors along with two mechanical relays that short the anneal heater lines to chassis when not in use to prevent external interference from entering the cryostat. During annealing operations, the relays are opened, allowing up to ∼2 W at ∼100 V to power the Zener diodes used as heaters. The SCUM also contains a digitally activated analog pulser that injects charge onto the high-voltage side of the detector, providing a way to measure the depletion voltage of the detector during voltage ramp-up or ramp-down. The detector depletion voltage generally moves down with increasing radiation damage and is then restored following high-temperature annealing, so provides an additional measure of detector health.

The High Voltage Filter Box (HVFB) contains a two-stage ripple filter with ∼120 dB rejection. It was too large to mount directly to a port on the cryostat, so it is attached to the mounting deck. It is located opposite the cryocooler to distance it from any exported vibrations generated by the cryocooler compressor, as high voltage components can be microphonic, injecting unwanted interference into the signal chain. Only capacitors with Negative-Positive-Zero (NPO) dielectric were used as added precaution. Shielded coaxial cables carry the high voltage up the boom from the DPU to the HVFB and also for the short distance from the HVFB to the sensor. Versions of the obsolete Reynolds Century Plus series were resurrected by special build. For the final mating of the connectors, a thin layer of Braycote high vacuum grease was applied to the mating surfaces to fill any trapped volumes, avoiding the possibility of a latent high-voltage breakdown after many months in space.

Cryocooler

The cryocooler for the GRS sensor (Fig. 8) is a pulse tube microcryocooler developed and manufactured by Lockheed Martin’s Advanced Technology Center (Olson et al. 2015). Their Micro-1 model was chosen for its low mass (∼450 g), cooling performance (∼500 mW at 85K with ∼20 W input and 300 K reject temperature), and long life. Oxford-style pulse tube coolers of this type are expected to last 10 + years, far exceeding the ∼100-day science requirement. Two engineering model units were built prior to the flight unit, with one designated as a Life Cycle Test Unit (LCTU) that underwent accelerated life testing. The LCTU exceeded the equivalent of three times the total expected number of operating hours, including extensive thermal cycling and characterization, with no sign of degradation.

Fig. 8.

Photograph of the GRS cryocooler on its delivery test stand. The compressor (left side) is connected to the cold head with cold finger (right side) through the transfer tube

The cryocooler consists of three separate components: a compressor module, a cold head with cold finger, and a transfer line between them. The cold head mounts directly to the cryostat with a flange and O-ring interface. A short, flexible copper braid thermally links a tab on the end of the cold finger to the detector encapsulation. The copper braid has also been shown to provide significant damping of radial loads acting on the cold finger. A cutout in the mounting deck allows the bulbous end of the cold head to protrude through it so as to minimize the length of a two-armed copper heat strap that conducts heat from the cold head to the deck (Fig. 5B). The compressor mounts directly to the radiator deck via a thermal gasket for efficient transfer of heat. Cooler power dissipation is split roughly evenly between the cold head and the compressor. The connecting transfer line between them is looped to provide strain relief on the pressure seals, given the separate mounting surfaces for the two components. The fill pressure of the system is 625 PSIA at room temperature.

The cryocooler has been tested regularly to ∼40 W input power but will be limited in software to a maximum of ∼30 W in flight to prevent overheating under hot conditions or excessive piston stroke under cold conditions. Once in space, the temperature of the entire sensor is tightly regulated to −10 °C using operational heaters, where the expected input power to the cryocooler will be ∼20 W or less. See Sect. 5.1 for details regarding the flight performance of the cryocooler.

Pulse tube coolers contain no moving parts in the cold head and no contact bearings in the compressor, so they produce low exported vibrations. The GRS HPGe detector system is “microphonic” and therefore sensitive to cooler vibrations, which can degrade overall system energy resolution slightly when operating at high power. Figure 9 shows GRS energy resolution versus cryocooler (CC) commanded power. Cryocooler power is shown in digital numbers (DN), where the calibration to watts, W, is: W = 47(DN/186)^2.2. Data are shown for detector temperatures of 90 K (closed circles) and 95 K (open circles). These data show that for power levels above ∼150 DN (∼29 W) higher power contributes to poorer energy resolution. These data further show that better energy resolution performance is achieved for the warmer detector temperature of 95 K, due to the detector JFET, which performs better at slightly warmer temperatures (see Sect. 3.2.1). In the context of full system performance, when radiation damage is being accumulated (as is the case during space-based operation), performance degradation manifests more quickly at higher detector temperatures, which offsets the improvement from a warmer JFET operating temperature. In flight, the cryocooler operates at much lower power levels (∼110 DN or ∼14 W; see Sect. 5.1), which is well below the range where energy resolution is impacted by cryocooler power level.

Fig. 9.

Energy resolution of the 1332.5-keV gamma-ray peak, as measured via the peak full-width at half maximum (FWHM), as a function of cryocooler power level as measured in digital units (DN). These measurements were taken during GRS ground calibration

GRS Thermal Subsystem

The GRS thermal subsystem is designed to accommodate the heat produced by the cryocooler. While the cryostat isolates the detector, the AC shield provides additional isolation from the thermal environment. In order to reject the heat produced by the cryocooler, the instrument not only utilizes existing areas on the cryostat and the AC shield housing, but also has dedicated radiator wings bolted onto the instrument deck (Figs. 4, 5, and 10). Thermal management on instrument surfaces is achieved with the use of fluorinated ethylene propylene (FEP) tapes backed with silver lining. Due to its exceptional optical properties (Fig. 10), silvered FEP tapes can efficiently emit heat in the IR spectrum while minimizing solar absorptance. Additionally, these silvered FEP tapes are adhered to existing surfaces, reducing the size of the dedicated radiators and added mass impact.

Fig. 10.

Photograph of the GRS illustrating the high degree of reflectivity of one of the two radiators. The reflection of technician Mark Hoff (who carried out much of the mechanical assembly of the GRS) is clearly seen

The cryocooler distributes its dissipated heat approximately equally in two places: the compressor and the cold head interface to the cryostat (the “warm flange”). The cryocooler compressor is directly coupled to the instrument deck thermally, but the warm flange is thermally coupled to the cryostat, so a copper thermal strap was added to provide a heat pathway from the warm flange to the radiator wings as well (Fig. 4B). Other key components, such as the PMT, on-sensor electronics boxes, and AC shield are all thermally tied to the sensor housing and generally have a similar temperature, with the PMT temperature running slightly cooler being the furthest away from the operational heaters.

The GRS has operational heaters located on the mounting deck, while the survival heaters are distributed on the AC shield housing and radiator wings. The operational heaters are controlled by the DPU, while the survival heaters are controlled by the spacecraft. Surfaces not being used as radiative area were covered in multilayer insulation (MLI) blankets. These blankets limit heat loss from undesired areas, as well as provide shielding from solar ultraviolet (UV) radiation. The vast majority of the on-instrument harnessing was also wrapped in MLI. This was especially important as the high-voltage harnessing had more restrictive temperature requirements.

GRS Anticoincidence Shield

A gamma-ray spectrometer system operating in space measures a significant background from charged-particle interactions in the gamma-ray sensor. For the Psyche GRS, GCR protons can deposit sufficiently low energy in the HPGe sensor to create a continuum that is in the same energy range as the gamma-ray signals of interest. This increased background lowers the sensitivity of the HPGe measurements. Following NEAR, Lunar Prospector, and MESSENGER flight heritage (Goldsten et al. 1997; Feldman et al. 2004; Goldsten et al. 2007), the Psyche GRS therefore includes an anti-coincidence (AC) shield.

The Psyche AC Shield is a cup-shaped, monolithic piece of borated plastic scintillator (EJ-254), read out by a 7.6-cm-diameter photomultiplier tube (Hamamatsu R6233-01) operating at approximately 400 V. The geometry of the scintillator (Figs. 4A, 11) was chosen to surround the HPGe sensor such that any highly penetrating charged particle reaching the interior gamma-ray detector must also interact with the AC shield. The scintillator is wrapped with several layers of polytetrafluoroethylene (PTFE) reflective tape that in combination with the conical sides, guides the scintillation light produced in the detector toward the bottom where it is read out by the PMT via an optical coupling pad. A charge-sensitive preamplifier contained in the end cap of the PMT housing converts the PMT charge pulses to voltage pulses, suitable for transmission down the boom to the DPU. The preamplifier contains two signal outputs that transmit the PMT signal with two different gains (see below). Because the AC shield scintillator (low density, low Z) is essentially transparent to gamma rays, AC shield signals that are in coincidence with HPGe signals are classified as background events and electronically removed (vetoed) from the HPGe AC spectral products.

Fig. 11.

Photographs of the of the AC shield scintillator. Panels A and B show the top and bottom, respectively, of the scintillator. Prior to installation, the scintillator is wrapped in reflective tape to internally reflect the scintillator light, and foil to provide light tight cover

In addition to the HPGe anticoincidence function, the AC shield also measures neutrons from two different energy bands, as well as GCR protons through energy deposition spectra that are recorded by the GRS DPU. Two spectral products (low and high gain) are recorded. The low-gain “GCR” spectrum extends to a maximum energy of 35 MeV, sufficiently high to record the characteristic peaks resulting from minimum-ionizing particle (MIP) interactions in the scintillator. Cosmic-ray protons produce characteristic MIP peaks at 4 and 17 MeV, resulting from interactions in the annular and base portions of the AC shield, respectively (see Sect. 5.1). The GCR spectrum therefore provides a means to measure and monitor the GCR proton environment in flight. The high-gain spectrum extends to a maximum energy of approximately 0.5 MeV, and includes the ∼90 keV peak resulting from the ¹⁰B + n Li + $\alpha$ neutron capture reaction. Section 3.5.2 describes how the neutron and energetic proton measurements are carried out within the GRS DPU. Sections 4 and 5 provide examples of pre-launch and in-flight AC shield measurements, respectively.

Neutron Spectrometer (NS)

A graphic view of the NS is shown in Fig. 12; photographs of the NS are shown in Fig. 13; a functional block diagram of the NS is shown in Fig. 14; detector characteristics and resource use for the NS are given in Table 4. The NS consists of three ³He neutron sensors that are sensitive to three different neutron energy ranges via different surrounding materials. Each cylindrical sensor is identical and is filled with 10 atm of ³He gas. A thin wire attached at each end runs through the center of each sensor and is set to 1360 V relative to the grounded sensor housing. Neutrons are detected via the ³He neutron-capture reaction:

$${}^3\text{He}+\text{n}\to \text{H}+\text{t}+764\text{ keV},$$ 1

where the combined kinetic energy (764 keV) from the proton (H) and triton (t) is detected as a charge pulse on the high-voltage line. Each sensor has dimensions of 5.08-cm diameter by 30.6-cm-long, and “dead volume” lengths of 2.07 cm (non-connector end) and 3.13 cm (connector end) that are not sensitive to neutron detection. The sensor length sensitive to neutron detection is 25.41 cm. Further details of the sensors and their calibration are given in Peplowski et al. (2020) and (Peplowski et al. 2025).

Fig. 12.

Graphical view of the NS. The different sensor components described in the text are labeled. Figure adapted from Peplowski et al. (2025)

Fig. 13.

Photographs of the NS on a test stand. Panels A and B show the front and back of the NS, respectively

Fig. 14.

NS functional block diagram. All instrument components that are located on the boom are shown within the large dashed rectangle. The three-slice DPU is shown on the right side of the block diagram. Individual components within each DPU slice are labeled. The 40 MHz ADC, which is used for the high-gain HPGe channel in the GRS DPU, is capped off and not used in the NS DPU. Additional details are given in the text

Table 4.

NS characteristics and resource use

Neutron sensors	3He gas proportional sensors; 5.08-cm diameter; 25.4-cm active length; aluminum housing; all sensors identical;
	Thermal: bare (no covering)
	Low-energy epithermal: 0.5 mm Cd covered; 3–12 μm Ni plating
	High-energy epithermal: 1 cm polyethylene covered
Energy range	Thermal: ∼0.025 eV to 0.4 eV
	Low-energy epithermal: > 0.4 eV
	High-energy epithermal: ∼1 keV to ∼100 keV
Energy resolution (FWHM/mean)	<4% for 764 keV neutron capture peak
Effective field-of-view	4π omnidirectional; full view of Psyche asteroid
Number of spectral channels	256 channels for each 3He sensor
Lower-level discriminators	Separate adjustable discriminator settings for each sensor
Maximum throughput	>5000 events per second
Operating voltage	1360 V
High-voltage stability	≪1 V
Electronics gain drift	None
Sensor gain drift	0.05 channels per °C
NS sensor mass	4.54 kg (including mounting plate)
NS DPU mass	1.56 kg
NS harness mass	1.94 kg (includes harness along boom and DPU to + z deck harness)
NS total mass	8.04 kg
Power use	11.4 W (DPU: 7.9 W; operational heater: 3.2 W; harness dissipation: 0.3 W)
Data rate	6000 bits per second Orbit D allocation (combined with GRS)

Neutron energy discrimination is achieved by surrounding each sensor with different types of materials. The bare sensor (no surrounding material) is sensitive to thermal and low-energy epithermal neutrons. The Cd-covered sensor is surrounded by a 0.5-mm layer of Cd, which absorbs neutrons with energies less than 0.4 eV, and is therefore sensitive to low-energy epithermal neutrons. The count-rate difference between the bare and Cd-covered sensor provides a measure of thermal ($E_{\mathrm{n}}<0.4\text{ eV}$) neutrons. For applications in the vacuum environment of space, Cd has the property that it is easily evaporative and fragments of Cd can migrate to other mechanical and electronics parts, potentially affecting their performance. To mitigate this effect, the Cd layer on the low-energy epithermal neutron sensor was plated with a 3–12 μm layer of Ni to encapsulate the Cd to ensure there would be no Cd migration. The third sensor is covered in a 1-cm-thick layer of high-density polyethylene (C₂H₄, density 0.9 g/cm³). The polyethylene downscatters neutrons to lower energies where they can be more efficiently detected by the ³He gas. The polyethylene thus provides higher sensitivity detection for high-energy epithermal (∼1 keV < $E_{\mathrm{n}}<\sim$100 keV) neutrons. It is noted that while the AC shield is not formally part of the NS, its broad-band epithermal and fast-neutron measurements ($E_{\mathrm{n}}>500\text{ keV}$), which are similar to measurements carried out on the Lunar Prospector GRS (Gasnault et al. 2001; Genetay et al. 2003), complete the full GRNS five energy-band neutron detection capability. The modeled effective area for all five neutron energy bands is shown in Fig. 15.

Fig. 15.

Effective area of all GRNS neutron sensors and derived thermal neutron measurement as a function of incident neutron energy. Effective area is the sensor efficiency times its cross-sectional area, and were derived as part of the GRNS calibration activities (Peplowski et al. 2025). ACS singles is the AC shield broad-band epithermal neutrons; ACS FN is the AC shield fast neutrons

All three neutron sensors are mounted to a single aluminum plate (1 mm thick with 8 mm thick reinforced ribs) that is attached to the boom at two attachment points (Fig. 13B). In contrast to the end-to-end sensor arrangement for the Lunar Prospector NS, the Psyche bare and Cd-covered sensors are oriented parallel to each other. Sensor “self-shielding” effects are minimized by separating the sensors by multiple sensor diameters. The polyethylene-covered sensor is oriented perpendicular to the bare and Cd-covered sensors and is located at the “dead volume” ends of the bare and Cd sensors, thus minimizing its shielding effects on the other two.

The backside of the NS mounting plate contains three separate but identical preamplifier boxes (Figs. 12 and 13) that feed high-voltage lines to each sensor, and receive and condition each signal line before sending the signals through the boom harness to the NS DPU. The fourth backside corner of the mounting plate contains a high-voltage junction box and field-joint bracket (lower left in Fig. 13B). The high-voltage junction box receives the two high-voltage lines from the DPU and distributes them to the three sensors, with a single high-voltage line going to the bare sensor, and two lines (from a single high-voltage supply) going to the Cd- and polyethylene-covered sensors, respectively (see Fig. 14). The field-joint bracket combines all the individual cables from the sensors into a single harness that runs from the sensor plate to the NS DPU. In its final configuration on the spacecraft, the sensors and plate are covered with a thermal blanket of MLI, and are kept thermally stable with thermal straps on the front side of the plate. The NS is enclosed by another layer of thermal blanketing that encompasses most of the GRS/NS boom structure.

Data Processing Units (DPUs)

GRS DPU

The GRS DPU consists of five boards (Figs. 6 and 16). Three of the boards are common to the GRS and NS DPUs: a Low Voltage Power Supply (LVPS), Processor, and High Voltage Power Supply (HVPS). The other two boards, the Cryocooler Power Conditioning Board (CCPCB) and Cryocooler Driver Board (CCDB) are part of the GRS DPU only. Each board in the DPU is a slice about 10.2 cm × 16.5 cm with an internal stacking connector providing the connections required between boards. They are each housed in an aluminum frame that is stacked together to form a complete housing (the end slices receive end covers). External connections to the spacecraft and to the sensor assemblies are provided by connectors on the top of each slice and on the side on the Processor and HVPS slices.

Fig. 16.

Photographs for two views of the GRS DPU show the top connectors and slice names (A) and side connectors (B)

The Processor board contains several Analog-to-Digital Converters (ADCs), spacecraft communications interfaces, memories, and two Field Programmable Gate Arrays (FPGAs). The Pulse-Process FPGAs perform pulse processing algorithms on the incoming signals via the science ADCs and provides the processed signals to the Event-Processor FPGA. The Event-Processor FPGA contains an embedded LEON3-FT processor along with custom logic for instrument control. The LEON3-FT processor core runs instrument flight software, which is responsible for processing commands, generating telemetry, operating the instrument, and handling science data from the Pulse Process FPGA.

The LVPS is a Printed Circuit Board (PCB) that powers the DPU electronics. Both the GRS and NS DPUs have nearly identical LVPS boards. The LVPS converts spacecraft input voltage, which ranges from 27.2 VDC to 33.6 VDC, to isolated conditioned voltages for the electronics in the DPU and for the NS and AC shield preamplifier boards which are external electrical PCBs to the DPU stack. The LVPS includes telemetry circuits to monitor all voltages and currents to other DPU slices, primary current on the LVPS, NS operational heater current, temperatures on the AC shield preamplifier board and board temperature on the LVPS. ADCs convert these analog-sensed housekeeping signals into serial digital signals that are sent via a Serial Peripheral Interface bus to the Processor board. The LVPS also has circuity to turn on and off the NS operational heaters, which are external to the DPU. The LVPS has an inrush current limiter and an electromagnetic interference (EMI) filter to properly handle power-on transients and prevent noise from going back onto the spacecraft payload power bus. The LVPS includes a DC/DC converter that steps down the input voltage from the spacecraft to ±12 VDC and + 3.3 VDC. Other voltages required for the instrument such as ±6 VDC and ±5 VDC are generated by linear regulators and sent to the stacking connector, the sensor electronics, or utilized on the LVPS itself.

The CCDB is a low-distortion 50-W Class-D-type amplifier that supplies the cryocooler with sinusoidal power of variable amplitude and frequency. The CCDB circuits are powered directly from spacecraft primary power to maximize efficiency, made possible because the cryocooler motor coils are completely isolated from chassis. The CCDB draws filtered input power via the CCPCB, which interfaces to a dedicated spacecraft “dirty” bus to ensure that any conducted emissions emanating from the cryocooler electronics will not reach and potentially interfere with other sensitive instruments or spacecraft subsystems. Communication with secondary-referenced electronics is achieved via high-speed digital isolators. Several protection circuits are incorporated into the design including an overcurrent trip, shoot-through prevention, and “launch lock” motor coil shorting relays.

All CCDB drive signals are digitally synthesized. The Event Processor FPGA generates a numerical sinusoidal waveform of variable frequency and amplitude using a Coordinate Rotation Digital Computer (CORDIC) algorithm. The frequency is selected to match the optimum drive frequency of the cryocooler, a value that varies linearly with temperature over the range 117-132 Hz. The resulting sinusoid is then compared against a digitally-generated sawtooth carrier that converts the signal to pulse-width modulation (PWM) format, where the carrier frequency is also selectable. This feature is used to set the carrier to a value not harmonically related to the magnetometer signal processor. The PWM signals are sent to the CCDB over the stacking connector. “3-level” modulation is used to minimize total harmonic distortion (THD), a scheme that requires independent control of each half of the Class-D H-bridge. Dead time can also introduce a significant source of distortion and can limit maximum output power capability. Gate drivers that offer independent high and low side control were chosen to allow selectable dead time. After extensive testing and timing analysis, a dead time of 50 ns was chosen, yielding a THD of ∼1%. The output is low-pass filtered with balanced two-stage Butterworth LC filters that provide >60 dB differential-mode attenuation at switching frequencies above 100 kHz. The PWM frequency is nominally set to ∼107 kHz.

The primary bus voltage is nominally ∼31 V but can range ±5% so is regulated down further to produce a “stiff” 27-V cryocooler power bus based on a bank of eight paralleled linear regulators mounted to the frame to dissipate heat. Each regulator is capable of delivering up to 1 A, providing ample margin and a degree of fault tolerance. Small ballast resistors associated with each regulator ensure equal distribution of the load current to within ∼50 mA. An additional two linear regulators create a 12.5-V bus for the H-bridge gate drivers and a 5-V bus that powers local digital circuits, housekeeping monitors, and the Hall Effect position sensors. The Hall Effect sensors provide piston stroke displacement information, an important cryocooler health diagnostic.

The CCDB incorporates several protection features. An overcurrent-monitor circuit trips if the peak current exceeds 8 A, a value chosen to be nearly twice the maximum operating current of the cryocooler, so as not to trip unnecessarily, but sufficient to interrupt a serious fault condition at the output. When the associated comparator circuit triggers, the circuit quickly disables the gate drivers (<10 μs), sets a flag to the Event Processor FPGA, and then remains latched in a disabled state until cleared by ground command or power-on reset. Another protection feature is an H-bridge shoot-through prevention circuit that rejects erroneous digital PWM signals that could otherwise short the cryocooler bus to ground. Lastly, magnetic latching relays are included to short the cryocooler motor coils during launch. Shorting the coils heavily dampens vibration-induced motion of the cryocooler pistons with an opposing back electromotive force. Without this feature, the pistons could potentially impact the end-stops with sufficient force to cause permanent damage. A grouping of four relays provides full series-parallel redundancy, i.e., if one relay opens during launch, the short is maintained by a parallel branch, and if one relay fails closed during launch, then a series relay ensures that the short can be opened to allow the cryocooler to operate.

The HVPS slice consists of two boards: the digital interface board and the analog high voltage board. The digital interface board is the low-voltage portion of the slice that communicates between the processor board and the analog high voltage board. The analog high voltage board amplifies the command voltage (0-5 V) to the voltage levels required by the GRS and NS. The digital interface board was designed and manufactured at APL, while the analog high voltage board was designed and manufactured at University of California, Berkeley. APL integrated both boards into one slice and tested the integrated slice. The digital interface board consists of signal chains that convert and scale commanding and housekeeping signals between the processor board and the analog high voltage board, a safing circuit that prevents unintended high voltage turn-on during ground testing, and a sequence circuit that ensures that the digital board is powered before the analog board to prevent board damage. The analog high voltage board consists of self-resonant power converters that amplify the commanding voltages, and housekeeping circuits that measure voltages and currents.

NS DPU

The NS DPU (Fig. 17) is a similar design to the GRS DPU with the exception of the cryocooler board set. The Processor board runs an NS-specific flight software application and the HVPS outputs are tailored to provide the correct high voltages for the NS ³He sensors. Both DPUs have identical LVPS boards.

Fig. 17.

Photographs for two views of the NS DPU show the top connectors and slice names (A) and side connectors (B)

GRNS Flight Software and Operation

Software Overview

The GRNS flight software is based heavily on the APL-generated “common core” library used for APL space instruments. Most software tasks utilize routines from this common library, which is actively used and maintained for multiple on-going missions. The common core library includes modules to handle the following types of functionality: command buffering, parsing, and execution; telemetry packet generation, compression, formatting, and output buffer management; alarm monitoring, response execution, and reporting; on-board definition, execution, and maintenance of stored sequences (macros); memory management including load, dump, copy, and checksum; stored parameter management; housekeeping, command echo, and alarm packet generation; and variable telemetry collection intervals.

The common core library provides hooks to include custom functionality where necessary. The GRS and NS flight software applications include hardware device drivers for the GRS and NS Pulse Process FPGAs, respectively. There is a device driver for each major hardware function in the FPGA design. In the GRS flight software, there is a driver for each of the following functions: housekeeping ADCs; cryocooler functionality; anneal heater functionality; high voltage functionality; pulse processing functionality; miscellaneous instrument interfaces. In the NS flight software, there is a driver for each of the following functions: housekeeping ADCs; high voltage functionality; pulse processing functionality; miscellaneous instrument interfaces. Each flight software application also includes a science processing task which is responsible for controlling the cadence of science data collection.

GRS and NS Data Processing

The underlying framework for retrieving and histogramming GRS and NS products is similar for the two different software applications on each DPU. Each DPU processor treats the Pulse Process FPGA as an input/output memory interface, with registers to read and write. The software configures the pulse processing channels by writing to control registers and retrieves pulse processing event data and status by reading data and status registers. The Pulse Process FPGA multiplexes event data from all channels into a single first-in, first-out (FIFO) buffer and presents it to software as a single data source. The processor is interrupted at a “half full” mark of the FIFO to trigger the driver to read out all available events in the FIFO. Based on which data products are enabled and the histogram configuration, the software will histogram valid events into the appropriate data product or drop invalid events. All histogram data produced by the GRS and NS are compressed within the DPU using a lossless fast compression algorithm. Non-histogrammed data from both the GRS and NS are uncompressed.

HPGe spectra

The HPGe spectra are the primary data product of the GRS. The GRS sensor has two parallel signal processing chains from the HPGe pre-amplifier. One is the low-gain channel that characterizes gamma-ray energies up to 9000 keV and the other is the high-resolution high-gain channel that characterizes energies up to 3000 keV. The high-gain channel digitizes the analog signals from the sensor with a 40 mega-samples per second (MSPS) ADC (AD9246S) on the processor board (Fig. 6), which is integrated down to 10 MSPS in the Pulse Process FPGA for the signal processing chain. The low-gain channel digitizes incoming signals with a 20 MSPS ADC (RHF1401) (Fig. 6) integrated down to 10 MSPS by the Pulse Process FPGA. After the integration stage, both channels have a three stage (unipolar, bipolar, tripolar) shaping algorithm for detection of HPGe events. The processor bins tripolar events into appropriate histograms based on the amplitude reported by the Pulse Process FPGA. Each channel has two 16,384-bin histograms where the highest 5 V signal goes into the highest bin. The “raw” histogram contains every valid energy event detected and the “anticoincident” histogram only contains events that also saw no coincident activity on the AC shield during a 0.5 μs analysis window of the event (see below).

AC shield neutron and GCR spectra

Fig. 18 illustrates how different AC shield pulse combinations are categorized into different event types. When an AC shield high-gain pulse is seen in coincidence with an HPGe pulse (Fig. 18A), the event is categorized as an HPGe coincidence event, and the HPGe pulse is removed from the HPGe AC histogram; the AC shield pulse for this event type is not placed in any histogram. When an AC shield high-gain pulse is detected without a coincident HPGe pulse, this opens a 25-μs analysis window. If no second pulse is detected with this window (Fig. 18B), the high- and low-gain pulse amplitudes are added to separate high- and low-gain histograms, respectively. The high-gain histogram (512 bins) provides a measure of broad-band epithermal neutrons by detecting the energy deposition from a single ¹⁰B(n,$\alpha$) capture in the scintillator. The low-gain histogram (1024 bins) measures the energy deposition from energetic GCRs. If a second pulse is detected within the 25-μs-long window, then this double-pulse event is categorized as a fast-neutron candidate event (Fig. 18C). For these events, the first pulse in the analysis window is the “prompt” pulse, and the second pulse is the “capture” pulse. The time difference between the two pulses is the time-to-second-pulse (TTSP).

Fig. 18.

This schematic diagram illustrates the pulse processing associated with the AC shield. A) When an AC shield high-gain pulse (solid line) arrives in coincidence (within a 0.5 μs window; window size not shown) with a HPGe pulse (dashed line) this event is kept in the raw spectrum but removed from the anticoincidence spectrum. B) When an AC shield high-gain pulse is the only pulse within a 25 μs window, the pulse amplitude is recorded in histograms for both the high- and low-gain channels. C) When two AC shield high-gain pulses arrive within a 25-μs window, this event is categorized as a fast-neutron candidate event. The first (prompt) and second (capture) events are recorded in early and late histograms based on the time between the two pulses (i.e., the TTSP). More details for fast-neutron events are provided in the text

The physical process of the fast-neutron detection is the following. When a fast neutron enters the AC scintillator with an energy above the nominal threshold of ∼0.7 MeV (Peplowski et al. 2025), it quickly slows down through interactions with the hydrogen atoms in the scintillator. The energy deposition from this slowing down process produces the prompt pulse; the amplitude of the prompt pulse provides a measure of the initial neutron energy. After the neutron has sufficiently slowed down in the scintillator, it can be captured by a ¹⁰B atom in the scintillator with an exponential TTSP distribution. The mean time, $\tau$, for capture is ∼2 μs, which is determined by the 5% ¹⁰B in the scintillator (Feldman et al. 1991). The capture pulse identifies the 90 keV energy deposition from the ¹⁰B(n,$\alpha$) reaction. The size of the analysis window is set to cover multiple e-folding lengths such that most fast neutrons are dominantly detected in the early portion of the window (0–5 μs) and accidental coincidences dominate the late portion of the window (20.6 – 25 μs). The nominal fast-neutron data products are four separate 256-bin histograms of early and late histograms of the prompt and capture pulses, respectively. The net fast neutron histograms are derived by taking the differences between the early and late histograms for the respective prompt and capture pulses. More details about the AC shield fast-neutron energy threshold and detection calibration are provided by (Peplowski et al. 2025). Fast neutron data acquired in flight are discussed in Sect. 5.1.

Gamma-ray burst

The AC shield scintillator is sensitive to the energy deposition of gamma rays of energies above ∼20 keV. A gamma-ray burst (GRB) product is generated using the full energy range of the AC shield high-gain channel. While this product is not part of the science requirements, it is included for additional science investigation to help triangulate galactic gamma-ray burst events as part of an ongoing interplanetary burst detection network (Hurley et al. 2010); the MESSENGER NS included a similar capability (Goldsten et al. 2007). Qualified above-threshold signals are counted (not histogrammed) at a 20 Hz cadence (i.e., 50 ms time resolution) and stored in a 1024-sample ring buffer. A trigger is detected if the count rate exceeds a threshold above a moving average across the time samples, indicating the detection of a gamma-ray burst. The collection of count rates will continue until the ring buffer has been filled post-trigger, leaving 256 samples of pre-trigger count rates for analysis. Once the buffer has been filled, it will be queued for downlink and the system will continue to look for the next trigger. The trigger threshold is nominally set at 20 keV, but will be tuned in flight to minimize the number of false positive triggers.

³He neutron sensor spectra

Each NS sensor generates a 256-bin histogram using a dedicated signal processing chain that run signals in parallel. The processor board digitizes signals with a 20 MSPS ADC (Fig. 14) and the Pulse Process FPGA integrates the samples down to 10 MSPS for shaping using a bipolar shaper for pulse detection. The amplitudes of the pulses from each neutron sensor are histogrammed in the flight software. Additionally, the time between the positive and negative peaks of each bipolar shaped pulse in each sensor is reported on an event-by event basis. This value, called ‘$dt$’ (i.e., ‘delta time’) can be used for background rejection. More information about the use of $dt$ for background rejection is given by Peplowski et al. (2025) and in Sect. 5.1.

Additional data products

The flight software has other data products that are produced either nominally in flight and/or upon command. During nominal operation, both the GRS and NS produce raw engineering and counter data packets. The engineering packets contain various instrument health and safety parameters (e.g., voltages, currents, temperatures), as well as instrument operational settings. Each of the counter packets contains a collection of various sensor and software count rates such as trigger rates, out-of-range rates, and other values. The accumulation period for these engineering and counter packets is commandable. The flight software also provides for the collection of GRS and NS raw event data where individual pulse parameter values (e.g., pulse amplitudes, time tags) are provided. For raw GRS and NS events, these products contain up to 50 events per second. The fast-neutron raw-event packet provides up to 255 events per second. The generation of these raw event packets is enabled and disabled via command. Detailed descriptions of these and all other GRS and NS data packets are given by Espiritu et al. (2022a) and Espiritu et al. (2022b).

Other Subsystem Controls

Operational heaters

Both the GRS and the NS have two operational heaters under software control. On each sensor, the two operational heater controls are independent. Parameters define a temperature range for the operational heaters to control with hysteresis. If the temperature sensor dips below the lower temperature bound, the operational heater will turn on. Likewise, if the temperature goes above the upper bound, the operational heater turns off. The purpose of the operational heaters is to more tightly regulate the temperature of the sensors as a way to minimize temperature variations of the sensor that could result in gain variations producing “smear” as well as to improve long-term reliability of the sensor electronics. The nominal setpoint is −10 °C ± 1 °C.

HVPS control

Both the GRS and the NS have two high voltage power supplies (Figs. 6 and 14) under software control. On each sensor, the two high voltage supplies are independently controlled. High voltage is ramped slowly to a commandable goal using a parameterized three-stage ramping pattern. While high voltage is ramping, the software monitors the housekeeping for unsafe conditions. If any such conditions are seen during a ramp, the high voltage will be brought back to a safe level and the safing level will be updated as appropriate. If at safing level 1, the high voltage can automatically re-attempt to ramp for a commandable number of retries after a waiting period. If the high voltage cannot be safely ramped after exhausting all retries, that high voltage supply will be set to safing level 2 and requires ground intervention before attempting to ramp again.

Sensor safing

The flight software employs a number of autonomous safing mechanisms to keep the instrument safe in case of anomalous behavior due to either internal instrument and/or environmental conditions.

On the GRS HPGe high voltage supply, the detector leakage current is monitored during different stages of the ramp. The safing behavior is dependent on which stage of the ramp and the leakage current seen. Stage 1 of the HPGe high voltage ramp is a blind ramp and the leakage current is not monitored. While ramping above the bias voltage (Segment 1 Hold Voltage) in stages 2 and 3, the bias current is checked against non-extreme leakage current threshold. If the leakage current is persistently above this threshold, the software will trigger a safing level 1 event. Persistence in the leakage current check is required to avoid instantaneous spikes in the leakage current due to GCR hits. If this safing level 1 event occurs below the depletion voltage (Segment 2 Hold Voltage), the high voltage is held until the retry interval expires or the safing level is reset. If this safing level 1 event occurs above the depletion voltage, the high voltage is ramped back down to the depletion voltage and held there until the retry interval expires or the safing level is reset. At any time that the high voltage is above the depletion voltage, including when at goal, if the leakage current is persistently above an extreme leakage current threshold, the software will automatically trigger a safing level 2 event and ramp back to the depletion voltage. The thresholds for non-extreme and extreme leakage current, as well as the persistence of both checks, are configurable by command. A second safing condition for the HPGe sensor is to bring high voltage back to 0 V and disable the high voltage control if the temperature at the detector is detected to be out of range persistently for three seconds. The leakage current thresholds for the HPGe detector are set well above nominal expected values but below those that would indicate a more serious condition, i.e., ∼1000 pA.

The NS sensors and AC shield PMT have a count-rate safing feature where the flight software monitors the count rates in each of these sensors. If the count rates on any of these sensor channels exceeds unique thresholds for each sensor, the software triggers as safing level 1 event and brings the high voltage back to 0 V. Typical scenarios for triggering this type of safing are when large solar particle events hit the spacecraft and induce count rates that are significantly above nominal background levels. When a level 1 safing event occurs, the software waits a specified amount of time (e.g., a few hours) and then autonomously raises the high voltage. If the count rate has gone down, then the sensors resume their normal operation. However, if the count rate remains high after a set number of attempts to raise the high voltage, then the flight software enters safing level 2 and will keep the high voltage at 0 V and wait for ground intervention. The count rate thresholds were chosen to be above the expected nominal science counting rates for the sensors yet comfortably below any rate that could accelerate aging (even for an extended period of time). Current values are 3 kHz for the NS and 20 kHz for the ACS. Autonomous recovery times are based on typical profiles for large solar particle events. Attempts at recovery are typically made every 12 hours for up to a few days.

Temperature safing and control for all subsystems use a pair of redundant temperature sensors. If the primary sensor were to fail, the software automatically switches control to the secondary sensor to main instrument operation and safety without intervention. If both sensors fail, the software configures the subsystem in a safe state until the sensors can be recovered.

Cryocooler

Using a three-level PWM scheme (see Sect. 3.4.1), the event processor controls the power to the cryocooler to manage the temperature. The power output is nominally set automatically by a proportional integral controller implemented in software but can also be set to fixed values manually via command. Control loop parameters are configurable, as is the temperature set point of the detector and the maximum output power allowed. To avoid damage to the sensor, the cryocooler will be disabled automatically if the temperature at the compressor or the warm flange is detected to be out of limit persistently for three seconds.

Anneal heaters

The GRS sensor has two anneal heaters to heat the Ge sensor during bakeout or annealing operations. The processor generates a pulse train to the anneal heater to provide power. The software uses “on-off” control to reach its temperature set point, with parameterized upper and lower dead band limits to reduce switching. A power switch must be enabled for each anneal heater for the pulse train to generate any power. Only one anneal heater may be powered at a time. The anneal heater control will be disabled if the Zener diode is detected to be in an “open” or “shorted” condition persistently for three seconds. The thresholds for open and short are parameterized. The anneal heater control will also be disabled if the temperature at the detector is over 115.5 °C persistently for three seconds. The anneal heaters use the same redundant sensors as the high voltage and cryocooler for detector temperature sensing. If both primary and secondary sensors have failed, anneal heater control is disabled.

GRNS Commanding

The GRNS has an extensive set of available commands for instrument turn on, configuration, maintenance, debugging, and data collection. The set of GRS commands is more extensive than the NS, and the NS commands are similar or identical to subsets of the GRS commands. The GRNS is able to generate and execute command macros, which are a sequence of commands that can be stored and then executed later.

The GRNS commands can be divided into the following categories: instrument power on/off, software core commands, GRS bake out/anneal, GRS cryocooler, operational heaters, GRS pulser, and data accumulation. The power on/off commands are formally spacecraft commands that provide the spacecraft voltage to each of the DPUs, which then for power-on initiates the boot process for each of the DPUs. Software core commands are an extensive set of commands that carry out various flight software tasks, such as memory check loads, and clears, and setting up of macros. The GRS bake out/anneal commands carry out the processes for operating the GRS anneal heaters that are used both for the initial GRS bake out following launch, and subsequent GRS anneal activities. The GRS cryocooler has a set of commands used to turn on and configure the cryocooler for initial and nominal operation. Operational heater commands are used to configure the autonomy for the operational thermal heating as well as set various temperature limits and thresholds. Pulser commands set up and configure parameters (e.g., rate, width) of the SCUM and digital pulsers. Finally, data accumulation commands enable and disable the various data products, as well as set up various parameters such as thresholds, accumulation times, and pulse processing parameters.

Summary of Instrument Builds and Environmental Testing

As part of the GRNS development program, a full set of engineering model (EM) hardware was built for development, debugging, and instrument characterization. A second complete EM DPU was built and delivered to JPL for ongoing development and testing in the Psyche mission testbed. Environmental tests carried out on the EM included vibration, shock, and electromagnetic interference and compatibility (EMI/EMC) qualification tests. Following the EM, the full flight model was built, tested, qualified, and delivered to JPL and installed on the Psyche spacecraft. Qualification testing included standard tests of vibration, shock, and thermal vacuum testing.

Photographs of the final installed GRS and NS are shown in Fig. 19. Figures 19A and 19B show the Psyche spacecraft following final GRS instrument closeout in August 2023, which included the installation of the light shield and calibration source. Figure 19C shows each sensor component with only the NS thermal blanket installed. The NS is covered with a secondary blanket that is a portion of the boom thermal blanketing (Fig. 19A). The radiator surfaces on the GRS are exposed, whereas all other portions of the GRS (e.g., cryocooler, harness, PMT, etc.) are covered in thermal blankets.

Fig. 19.

(A) Photograph of the GRS and NS on the Psyche spacecraft with final thermal blanketing installed after the GRNS closeout. (B) Focused picture of the GRS after its closeout. (C) Photograph of installed GRS and NS on the Psyche spacecraft prior to the installation of the final thermal blankets. The thermal blanketing for the NS is seen

A full set of characterization and calibration measurements were carried out on both the NS and GRS. Initial NS calibration measurements are described by Peplowski et al. (2020). Comprehensive GRS and additional NS performance and calibration measurements were carried out prior to delivery and are described by Peplowski et al. (2025).

Ground Performance Measurements

GRS Performance Measurements

The performance of the GRS was characterized during development, final instrument calibration, and post-spacecraft installation. Here we present instrument checkout measurements performed on the flight model GRS prior to launch; Sect. 5 describes the GRNS performance post launch.

Figure 20 shows the full gamma-ray spectrum measured with an experimental setup designed to illuminate the GRS with gamma rays up to ∼9000 keV (Peplowski et al. 2025). This setup used an AmBe neutron source surrounded by various types of materials (polyethylene, stainless steel, potassium chloride) to generate neutron capture and inelastic scatter gamma rays from elements such as K, H, Ni, Cl, and Fe. Carbon gamma rays from the AmBe source were also seen. The full low-gain spectrum (Fig. 20A) shows gamma rays from ∼90 keV to >9000 keV with 16,384 energy channels. Figure 20B shows the high-energy range from 6000 keV to 9500 keV, including capture lines from Cl and Ni. Figures 20C and 20D show the energy range around the key 1454 keV ⁵⁸Ni gamma-ray line for both the high- and low-gain spectra, respectively. The high-gain spectrum provides measurements of gamma rays with energies of <3000 keV, again with 16,384 energy channels. The oversampling of the high-gain data (Fig. 20C) provides better energy resolution than the low-gain spectrum (Fig. 20D). For this experiment, the count rate of K gamma rays at 1461 keV was sufficiently large to dwarf the measured count rate of Ni gamma rays from the stainless steel. Even with this significant count-rate difference, the Ni line is cleanly resolved and thus demonstrates that even with a similar count-rate difference at Psyche between these two gamma-ray lines (see Fig. 2), the GRS will be able to robustly quantify the flux of 1454-keV gamma rays.

Fig. 20.

Measured laboratory gamma-ray spectra taken with the flight model GRS as described by Peplowski et al. (2025). Part A shows the energy range from 0 to 9000 keV for the low-gain channel, with some prominent gamma-ray peaks labeled. Part B shows the high-energy portion of the low-gain channel with Cl and Ni gamma-ray peaks labeled. Part C shows the high-gain spectra in the energy range that highlights the 1454 and 1461 keV peaks from Ni and K, respectively. Fits to the peaks in this energy range are shown in red. Part D shows the same energy range shown in Part C but with the low-gain channel

The performance of the GRS was further verified by checking the energy resolution using a ⁶⁰Co gamma-ray source. Following ANSI standard N42.14, which characterizes the FWHM energy resolution of gamma-ray spectrometers at 1332.5 keV, we performed spectral analysis of the HPGe measurements. Figures 21 and 22 show gamma-ray spectra collected during the first CPT in September 2021 after the GRS was installed on the Psyche spacecraft in August 2021. These spectra were obtained with full processing of the data through the spacecraft flight software and ground data system. The measured spectra include gamma-ray peaks from room backgrounds (primordial radionuclides ⁴⁰K, ²³²Th, and U isotopes). A fit to the 1332.5-keV peak from the ⁶⁰Co source is shown in Fig. 22. The optimum high-gain measurement demonstrates 1.93 ± 0.01 keV FWHM energy resolution (Fig. 22A). This value is typical of laboratory-quality measurements and exceeds the 3.5 keV FWHM pre-launch requirement. As expected, the energy-resolution performance of the low-gain channel is about 8% poorer than the high-gain channel with a resolution of 2.08 ± 0.01 keV (Fig. 22B). The GRS performance was characterized through a range of operating environments during four separate pre-launch tests (e.g., thermal vacuum, spin up of all reaction wheels, etc.), and showed consistent performance during all tests.

Fig. 21.

Measured gamma-ray spectra taken from the high-gain channel during the September 2021 CPT after the GRS was installed on the Psyche spacecraft. Black shows data taken using the ⁶⁰Co calibration source, and red shows room background data

Fig. 22.

High-gain (A) and low-gain (B) data taken using the ⁶⁰Co calibration source during the September 2021 GRS CPT while on the Psyche spacecraft. Gaussian fits to the peaks are shown in red, and the fit parameters are shown in the plots. We note that the best-fit peak centroid in both cases has a higher energy than the standard 1332.5 keV; however, this fit value is consistent with the standard value when the uncertainties in the pre-launch calibration parameters are taken into account (Peplowski et al. 2025)

The neutron detection performance of the AC shield was characterized with a moderated and unmoderated AmBe neutron source as well as a ¹³⁷Cs source. The high-gain spectrum is shown in Fig. 23. The 60-keV X-ray and ¹³⁷Cs Compton edges were used to establish the energy scale of the spectrum. The moderated AmBe measurements (red) show a clear spectral peak at ∼100 keV that is not present when a moderated neutron source is absent. This spectral feature is the ¹⁰B(n,$\alpha$)⁷Li neutron capture peak. Its presence is indicative of neutrons, and its amplitude provides a measure of the count rate for broad-band neutron energies less than 0.7 MeV. Note that double-pulse neutron events are not classified as single-pulsed events and therefore do not appear in this spectrum.

Fig. 23.

Neutron events observed in the high-gain spectrum of the GRS AC shield. An estimated channel-to-energy calibration is included as the second horizontal axis. The 60-keV X-ray from AmBe (blue), neutron capture peak (red), and 661-keV Compton edge (477 keV; green) features are marked with vertical dashed lines

Ground-based fast neutron data and calibration are described in detail by Peplowski et al. (2025). Flight-based fast neutron data are presented in Sect. 5.

The low-gain AC shield spectrum provides a means for monitoring the flux of GCR protons. Laboratory testing of cosmic-ray monitoring is challenging, as cosmic-ray protons do not reach Earth’s surface due to the presence of the thick atmosphere. GCR-generated muons do reach Earth’s surface, however, and can serve as a proxy for testing the GCR measurement capability. Peplowski et al. (2025) describe how this ground-based muon measurement was carried out with the AC shield.

The performance of the GRB mode was demonstrated by manually moving the ⁶⁰Co source near the GRS. The source was moved toward the GRS twice (Fig. 24A). The first time, the peak rate in the AC shield was insufficient to trigger the GRB mode. The second time, the threshold was met and a 50-ms-time-resolution spectrum showing the counting rates before, during, and after the “burst” was produced as seen in Fig. 24B.

Fig. 24.

Example gamma-ray burst data from pre-launch ground-based testing of the AC shield taken on 23 September 2021. A radioactive gamma-ray source was twice moved rapidly towards the AC shield to induce a count-rate enhancement. The high-gain count rate with a time resolution of one second is shown in (A) as a function of time relative to a fiducial time at 0 seconds. The first time the source was moved, the source did not get too close to the AC shield and so did not initiate a burst trigger (left peak at 120 seconds); the second time the source was moved, it came closer to the AC shield and initiated a burst trigger (right peak at a time of 290 seconds). The burst trigger with 50-ms time resolution (B) is shown at the time of 290 seconds

NS Performance Measurements

The performance of the NS was characterized during development, final instrument calibration, and post-spacecraft installation. Detailed NS calibration data were presented by Peplowski et al. (2020) based on measurements taken at the National Institute of Standards and Technology Center for Neutron Research.

Figure 25 shows measured pre-launch neutron spectra taken with the NS and DPU using an AmBe neutron source close to the sensors. The data in Fig. 25A were taken with the AmBe source surrounded by polyethylene in order to moderate the neutrons to include a range of lower energy neutrons. In contrast, the data in Fig. 25B used a bare AmBe source, which consequently had a higher fraction of higher energy neutrons. This difference in neutron energies is reflected in the data such that the bare sensor shows the highest count rate for the moderated source whereas the polyethylene covered sensor – which has the highest sensitivity to higher energy neutrons – shows the highest count rate with the bare source. Additional quantitative characterization of the NS (e.g., temperature dependence, gain characteristics, flight performance) is presented by Peplowski et al. (2025).

Fig. 25.

Pre-launch energy deposition spectra measured by the three NS sensors. Relative energy deposition is given in channel numbers. (A) Data shown for an AmBe neutron source surrounded by a polyethylene shield that moderates the neutron energies to lower energies than the few MeV neutrons naturally emitted from the AmBe source. Compared to the bare sensor, the polyethylene covered sensor (green) shows a lower relative count rate indicating fewer higher energy neutrons compared to thermal neutrons as measured by the bare sensor (blue). (B) Data shown for a bare AmBe neutron source that has a high proportion of higher energy neutrons than the shielded source. Here, the polyethylene covered sensor shows the highest count rate of the three sensors indicating the detection of a larger proportion of higher energy neutrons than for the case of the shielded source

GRNS in-Flight Performance

Initial Checkout

The Psyche spacecraft launched on 13 October 2023 from Cape Canaveral, Florida on a Falcon Heavy launch vehicle. As part of the multi-month spacecraft initial checkout (ICO) activities, the GRS ICO took place over three weeks from 6 November 2023 to 27 November 2023. The NS ICO took place on 11–12 December 2023. This section describes the results from these first in-space operations of all GRNS components.

The first turn on of the GRS DPU occurred on 6 November. There was a one-time operation to disable launch locks on the cryocooler. This was followed by a two-hour cryocooler operation to ensure that the cryocooler survived launch and operated correctly. GRS detector temperature and command power profiles for this and the subsequent full-cooldown operation are shown in Fig. 26. The initial two-hour cool down is shown as arrow ‘1’ in Fig. 26A. During this quick cooldown, the cryocooler demonstrated good operation, and the GRS reached 163 K (- 110 °C). Following this cooldown, the GRS anneal heater was turned on for a bake out to 50 °C (323 K) in order to remove any volatiles that may have accumulated inside the cryostat prior to launch. The heater took ∼12 hours to reach the bake out temperature, and the total time at bake out temperature (arrow ‘2’ to arrow ‘3’ in Fig. 26A) was 36 hours.

Fig. 26.

GRS cryocooler temperature (A) and power (B) profiles during the initial turn on and first full cooldown of the cryocooler. Arrow labels indicate the following: 1): initial two-hour cryocooler operation; 2) start of 36-hour cryostat bakeout at 50 °C; 3) start of passive cool down of GRS cryostat; 4) start of full GRS cooldown. The insets in each figure show the temperature and power profiles around the approach to and regulation at the equilibrium temperature

At the end of the GRS bake out (arrow ‘3’ in Fig. 26A), the GRS was passively cooled, and the full cooldown was initiated on 10 November 2023 (arrow ‘4’ in Fig. 26A). The cooler took roughly 12 hours to reach its commanded temperature of 95 K. The temperature of 95 K was chosen to ensure sufficient thermal margin when operating relatively close to the Sun. While in orbit at Psyche, the cryocooler will operate at a temperature of 85 K to 90 K. It is preferred to operate at the lower 85 K temperature, as this temperature slows down the process of radiation damage, thus enabling longer accumulation periods without the need to anneal the detector (Jourdain 2011). The commanded cryocooler power for these operations is shown in Fig. 26B; the equilibrium power draw after reaching temperature was 14.5 W. The insets in each figure show the detector temperature and cryocooler power for the approach to, and at equilibrium. The temperature control loop held the temperature constant to less than 0.5 K.

Soon after the bake out command was sent on the first day of operation, the AC shield high voltage was raised for the first time. Figure 27 shows AC-shield data from the low-gain (Fig. 27A) and high-gain (Fig. 27B) spectra, respectively. The low-gain spectrum shows various features indicative of charged particle energy deposition in the AC shield. Specifically, the peaks around channels 70, 120, and 400 indicate energy deposition in different geometrical portions of the AC shield. Details of the low-gain response to charged particles are given by Peplowski et al. (2025). The high-gain spectrum shows a featureless spectrum with no indication of lower energy neutrons. This lack of lower energy neutrons while in cruise is consistent with prior space-based measurements (e.g., Lawrence et al. 2013a), and indicates that the amount of material in typical spacecraft is insufficient to downscatter energetic neutrons to lower energies.

Fig. 27.

Initial checkout data from the GRS AC shield. (A) Energy deposition spectrum of low-gain, AC shield channel. (B) Energy deposition spectrum of the high-gain AC shield channel. Both spectra are averaged from the dates of 12 November to 27 November 2023

Figures 28A and 28B show the four fast-neutron spectral products (Sect. 3.5.2) from data acquired on November 12–27, 2023. The two early spectra (black lines in Figs. 28A and 28B) were accumulated from early TTSP values (0 – 5 μs) using onboard processing; similarly, the two late spectra (red lines in Figs. 28A and 28B) were accumulated from late TTSP values (20.6 – 25 μs), also using onboard processing. The net fast-neutron energy spectrum (blue line in Fig. 28A) is consistent with similar feature-less profiles seen in prior space-based fast-neutron measurements (e.g., Lawrence et al. 2014). The ¹⁰B(n,$\alpha$) neutron capture peak is seen in Fig. 28B near channel 70, and indicates a clean and unambiguous detection of neutrons. The average fast neutron count rate (e.g., the sum of the net peak counts in Fig. 28B) is ∼2 cps, which is broadly consistent with that measured with the MESSENGER NS (1.8 to 2.5 cps for variable GCR conditions) (Rodgers et al. 2015) using a similar sized borated plastic scintillator.

Fig. 28.

Fast neutron data collected during ICO. (A) and (B) show prompt and capture spectra as generated by onboard DPU processing as described in Sect. 3.5.2. (C) shows TTSP spectra acquired during four-hours of fast-neutron event mode collected on November 6, 2023. The left shaded region shows early TTSP times (0 to 5 μs); the right shaded region shows late TTSP times (20.6 – 25 μs). Additional details are described in the text

Figure 28C shows TTSP data that were acquired during a four-hour accumulation of fast-neutron event-mode data early in ICO on November 6, 2023. This data mode is not usually turned on during nominal operation, but was turned on during ICO to demonstrate that the fast-neutron processing is working as expected. The spectrum of TTSP values (black line in Fig. 28C) is a combination of real neutron detections at early times of 0 to 5 μs (left shaded region in Fig. 28C), and accidental coincidences at late times of 20.6 to 25 μs (right shaded region in Fig. 28C). The net TTSP histogram (gray line in Fig. 28C) is derived by subtracting an exponential fit to the late TTSP values (red line in Fig. 28C) from the measured TTSP spectrum. This net TTSP spectrum has a mean decay time of 1.95 μs, which is consistent with that expected for the 5% ¹⁰B in the type of scintillator material used in the AC shield (Feldman et al. 1991).

On 13 November 2023, the HPGe high voltage was raised in a stepwise fashion from 0 V to 2500 V with ground-in-the-loop monitoring for each step. The HPGe sensor was then operated for nearly 14 days (333.5 hours, or 13.9 days). There was no significant solar energetic particle activity during this time period. In addition, there were no activation lines apparent in the data that would indicate significant solar activity in the weeks or months prior to operation. Therefore, this dataset provides a clean accumulation of GCR-generated background gamma rays. In addition, prior to its initial power on, the GRS was only in space for 30 days of GCR radiation. Thus, the accumulated radiation damage was relatively low for the initial GRS operation.

Summed high-gain and low-gain spectra for the 13.9-day period are shown in Fig. 29. Raw spectra are shown in red, and anticoincidence spectra are shown in black. The ratio of the raw to anticoincidence spectra (see Peplowski et al. 2025) shows that the background rejection ranges from a factor of three for the high-gain channel, to greater than a factor of 10 for the low-gain channel. Over 100 gamma-ray peaks are identified in the high- and low-gain spectra, and many of these are listed in Table 7. The Appendix describes the channel-to-energy calibration derived for this analysis.

Fig. 29.

High-gain (A) and low-gain (B) gamma-ray spectra for the 333 hours of accumulation time during the GRS initial checkout. Raw spectra are shown in red, and anticoincidence spectra are shown in black. In the high-gain spectra, the location of the 662 keV ¹³⁷Cs calibration peak is labeled

Table 7.

List of prominent gamma-ray lines in the ICO dataset. The reference energies (‘Energy’ column) are taken from Evans et al. (2017) and Chadwick et al. (2006). The ‘Centroid channel’ is the fit centroid channel value when a single gaussian plus second-order polynomial function is fit to each peak. The FWHM value is determined from a fit using a gaussian plus a second-order polynomial function using the calibrated energy scale. The FWHM is a factor of 2.35 times each peak width. The * symbol denotes a gamma ray that can be created by more than one nuclear process.

Reference Energy (keV)	Source	Centroid channel	Centroid energy (keV)	FWHM (keV)	${\chi }^2/\nu$	Counts	Counts per second	Centroid channel	Centroid energy (keV)	FWHM (keV)	${\chi }^2/\nu$	Counts	Counts per second
		High Gain						Low Gain
139.69	75mGe	809.0	139.8 ± 0.004	1.150 ± 0.0094	11.11	125170.6 ± 909.4	0.1048	247.9	139.8 ± 0.005	1.324 ± 0.0124	32.17	121670.9 ± 966.5	0.1019
185.72	235U	1076.1	185.9 ± 0.005	1.319 ± 0.0108	1.17	94959.2 ± 68.9	0.0795	329.3	185.9 ± 0.005	1.497 ± 0.0130	2.34	95695.6 ± 739.1	0.0801
198.40	71mGe	1148.4	198.3 ± 0.001	1.280 ± 0.0029	134.48	399256.8 ± 857.3	0.3342	351.4	198.3 ± 0.002	1.455 ± 0.0035	445.58	397447.6 ± 862.1	0.3327
238.63	232Th	1382.7	238.7 ± 0.017	1.275 ± 0.0409	1.20	16400.1 ± 466.2	0.0137	422.8	238.8 ± 0.021	1.425 ± 0.0511	0.72	15698.1 ± 508.2	0.0131
271.25	44Sc	1571.5	271.2 ± 0.017	1.385 ± 0.0400	1.18	16583.8 ± 415.5	0.0139	480.3	271.4 ± 0.019	1.561 ± 0.0472	1.38	16754.8 ± 422.4	0.0140
403.20	67Ga + K	2336.8	403.1 ± 0.020	1.403 ± 0.0485	1.15	9597.0 ± 292.0	0.0080	713.1	403.1 ± 0.025	1.675 ± 0.0604	0.96	10033.6 ± 304.2	0.0084
472.20	24mNa	2737.5	472.2 ± 0.012	1.546 ± 0.0275	3.13	19734.5 ± 304.7	0.0165	834.9	472.1 ± 0.013	1.727 ± 0.0319	7.10	20045.0 ± 310.4	0.0168
511.00	Annih.	2962.3	510.9 ± 0.003	2.698 ± 0.0057	6.19	312183.1 ± 638.0	0.2613	903.3	510.8 ± 0.003	2.802 ± 0.0060	15.73	312249.2 ± 643.6	0.2614
661.66	137Cs	3836.7	661.6 ± 0.001	1.610 ± 0.0018	6.54	597944.4 ± 795.4	0.5005	1169.6	661.6 ± 0.001	1.768 ± 0.0020	14.46	599456.6 ± 798.3	0.5018
817.90	58Co + K	4743.0	817.8 ± 0.025	1.627 ± 0.0606	2.39	5289.9 ± 175.4	0.0044	1445.5	817.8 ± 0.028	1.740 ± 0.0668	6.17	5309.1 ± 179.3	0.0044
843.70	27Al	4893.8	843.8 ± 0.019	1.714 ± 0.0458	5.57	7938.5 ± 185.5	0.0066	1491.4	843.8 ± 0.021	1.800 ± 0.0504	17.72	7762.1 ± 188.9	0.0065
882.50	69Ge + K	5118.3	882.4 ± 0.022	1.756 ± 0.0533	2.70	6629.7 ± 176.3	0.0055	1559.8	882.5 ± 0.025	1.869 ± 0.0589	5.79	6557.2 ± 185.8	0.0055
911.20	232Th	5284.6	911.1 ± 0.046	1.743 ± 0.1117	1.04	2878.2 ± 160.2	0.0024	1610.6	911.2 ± 0.054	1.945 ± 0.1333	0.58	2860.8 ± 161.5	0.0024
983.50	48Ti	5705.0	983.5 ± 0.030	1.875 ± 0.0732	1.37	4884.8 ± 163.4	0.0041	1738.4	983.6 ± 0.035	2.002 ± 0.0853	2.55	4754.4 ± 174.2	0.0040
1001.03	238U	5806.6	1001.0 ± 0.025	1.898 ± 0.0611	0.95	5978.9 ± 166.6	0.0050	1769.2	1001.1 ± 0.029	2.049 ± 0.0705	0.68	5918.1 ± 174.1	0.0050
1014.50	27Al	5884.4	1014.4 ± 0.023	1.879 ± 0.0546	0.97	6572.4 ± 167.9	0.0055	1792.9	1014.5 ± 0.027	2.073 ± 0.0646	1.13	6578.7 ± 174.9	0.0055
1048.90	66Ga + K	6083.8	1048.8 ± 0.040	1.949 ± 0.0969	0.93	3670.4 ± 158.0	0.0031	1853.6	1048.9 ± 0.043	2.055 ± 0.1062	1.10	3682.0 ± 169.0	0.0031
1117.10	69Ge + K	6480.6	1117.1 ± 0.013	1.924 ± 0.0291	2.29	13778.7 ± 185.1	0.0115	1974.2	1117.1 ± 0.014	2.071 ± 0.0327	4.64	13776.8 ± 190.0	0.0115
1312.10	48Ti	7613.6	1312.3 ± 0.046	1.917 ± 0.1130	1.20	2613.0 ± 132.2	0.0022	2318.8	1312.2 ± 0.049	2.007 ± 0.1192	1.70	2661.7 ± 135.4	0.0022
1368.67	24Mg	7940.6	1368.7 ± 0.013	2.137 ± 0.0292	1.23	13922.4 ± 170.9	0.0117	2418.1	1368.5 ± 0.014	2.238 ± 0.0316	1.73	13720.0 ± 171.8	0.0115
1460.82	40K	8475.7	1460.8 ± 0.053	2.250 ± 0.1305	0.99	2642.2 ± 128.7	0.0022	2580.8	1460.6 ± 0.058	2.451 ± 0.1429	2.05	2720.3 ± 133.8	0.0023
1633.60	20Ne*	9479.0	1633.7 ± 0.035	2.287 ± 0.0856	1.09	3924.2 ± 124.3	0.0033	2886.0	1633.5 ± 0.039	2.519 ± 0.0953	1.42	4061.0 ± 128.9	0.0034
1808.74	26Mg*	10495.9	1808.8 ± 0.046	3.066 ± 0.1175	1.02	4253.3 ± 132.2	0.0036	3195.3	1808.6 ± 0.049	3.154 ± 0.1249	1.95	4215.1 ± 133.2	0.0035
2614.51	232Th	15175.4	2614.6 ± 0.069	2.587 ± 0.1745	1.00	1840.5 ± 103.6	0.0015	4618.7	2614.7 ± 0.055	2.436 ± 0.1333	0.66	1672.9 ± 78.4	0.0014
2754.02	24Na, 24Mg	15984.9	2753.9 ± 0.031	2.755 ± 0.0749	1.27	5643.7 ± 129.7	0.0047	4864.8	2754.1 ± 0.026	3.001 ± 0.0616	2.38	5914.2 ± 103.8	0.0050
5107.30	16O, DE							9019.0	5107.6 ± 0.300	4.312 ± 0.8318	0.93	345.4 ± 45.3	0.0003
5618.30	16O, SE							9919.4	5617.9 ± 0.346	4.997 ± 0.9594	0.99	371.5 ± 46.2	0.0003
6129.30	16O							10821.9	6129.4 ± 0.201	4.518 ± 0.6140	0.68	456.3 ± 43.4	0.0004

Figure 30 shows the high-gain spectrum in an energy range that contains the key 1454 keV Ni gamma-ray peak. The six strongest peaks in the region are labeled along with the location of the ⁵⁸Ni peak. A fit to all peaks – including the putative ⁵⁸Ni peak – is shown by the solid red line. To obtain a good fit for the ⁵⁸Ni peak region, the peak centroid and width needed to be fixed, with its magnitude the only variable. In the other peaks, all peak parameters were allowed to converge on their best-fit values. The net total counts for the ⁵⁸Ni peak, and its neighboring ⁴⁰K peak at 1460.8 keV are 141.9 ± 170.7 and 2629.0 ± 195.2, respectively (or equivalent count rates of $1.19\times {10}^{-4}\pm 1.43\times {10}^{-4}$ cps and $2.20\times {10}^{-3}\pm 1.63\times {10}^{-4}$ cps). The derived uncertainties are one standard deviation, and assume Poisson statistics in the peak and background. The measured counts in the ⁵⁸Ni peak are consistent with zero. This result demonstrates that the effort to minimize nickel-bearing materials on the spacecraft was largely effective (Bradford et al. 2022). Further, when the low nickel background is combined with the excellent energy resolution of the GRS (see below), we conclude the GRS should meet its nickel measurement requirement with substantial margin, perhaps lowering our detection threshold, which may prove useful if Psyche turns out to have less exposed metal-rich surface than currently expected.

Fig. 30.

High-gain anticoincidence spectra for the energy region from 1300 to 1500 keV. The locations of prominent peaks are labeled, and the locations of the 1454 keV ⁵⁸Ni and 1460 ⁴⁰K peaks are identified by vertical dashed lines. The red line shows a fit using a gaussian plus a second order polynomial where all clear peaks are fit along with the region around the 1454 keV peak. In the peak fits for six identified peaks (plus an unidentified peak close to the ⁴⁸Ti peak), all peak parameters (energy position, width, height) are allowed to vary and converge to a best fit. For the 1454 keV region, the energy position and width were fixed, with the height allowed to vary. The net counts in the 1454 keV region are 141.9 ± 170.7, which is consistent with zero at a one standard deviation uncertainty

For the high-energy portion of the low-gain spectrum (Fig. 31), the 6129 keV oxygen gamma-ray peak is seen in both the raw and anticoincidence spectra. Here, the anticoincidence spectrum shows roughly a factor of eight background reduction compared to the raw spectrum. Fits to both peaks are shown in the plot. The net counts for the raw and anticoincidence peaks are 709.5 ± 336.0 and 403.5 ± 113.3, respectively (or $5.94\times {10}^{-4}\pm 2.8\times {10}^{-4}$ cps and $3.38\times {10}^{-4}\pm 9.49\times {10}^{-5}$ cps). The count rates are consistent to within the one-standard-deviation uncertainty. The uncertainty for the anticoincidence measurement is nearly a factor of three improved compared to the raw spectrum measurement. Since Poisson counting uncertainties scale as the square root of time, this implies that in this energy range, the anticoincidence shield provides a signal-to-background improvement roughly equivalent to almost a factor of nine increase in accumulation time. Further, when assessing the degree to which Psyche is a metal or silicate body, oxygen could be a key measurement since oxygen will only be present in silicate materials and not in pure iron-nickel metal.

Fig. 31.

Low-gain spectra in the energy region from 6050 to 6250 keV region. The raw and anticoincidence spectra are labeled. The 6129 keV oxygen peak is clearly seen in both spectra, and peak fits for each are shown by the blue trace for the raw spectrum, and the red trace for the anticoincidence spectrum. The net counts for the raw and anticoincidence peaks are 709.5 ± 336.0 and 403.5 ± 113.3, respectively

Comparing the GRS energy resolution, ${\sigma }_{E,\mathit{FWHM}}$, as measured in space, with the pre-launch measurements is not straightforward. The reason is that the statistical precision of the canonical 1332.5 keV ⁶⁰Co peak is significantly worse in the space measurements compared to the pre-launch data with the ⁶⁰Co source (Fig. 22). The flight data show a small 1332.5 keV peak (Fig. 30) from GCR-induced spallation reactions on the copper heat strap that is part of the GRS housing. The energy resolution of this peak is 2.41 ± 0.40 keV. This large uncertainty allows a range of ${\sigma }_{E,\mathit{FWHM}}$ values from 2.0 to 2.8 keV. Thus, we use an alternate way of establishing the energy resolution at 1332.5 keV. To achieve a higher precision estimate of energy resolution, we measure the energy resolution for a number of peaks across the full-energy range of both the high- and low-gain spectra, fit a functional form to the ${\sigma }_{E,\mathit{FWHM}}$ versus energy ($E$) data, and then use this functional form to infer the energy resolution at 1332.5 keV. These data are shown in Fig. 32. For the functional form, we use a general relation given by Knoll (2000):

$${\sigma }_{E,FWHM}=\sqrt{b_0+b_{1}E+b_2{E}^2},$$ 2

where $b_0$ represents the intrinsic electronic noise, $b_1$ represents the statistical fluctuations inherent in the physics of detection in the HPGe sensor, and $b_2$ is a term that broadly accounts for effects such as incomplete charge collection. Initially, all peaks in Table 7 were available as candidate data points for this analysis. However, prior to fitting the ${\sigma }_E$ versus $E$ values, some selections were made to the data to obtain a more robust fit. First, some peaks (e.g., 511 keV, 1808 keV) have an intrinsic broadening that is larger than the optimum width of the HPGe detector. If included in the fit to Eq. (2), these peaks would bias the fit to high ${\sigma }_E$ values. Second, other peaks have sufficiently large counts that non-Gaussian effects are present in the specific peak fits and can affect the measured peak width and uncertainty. This kind of effect is identified via a large reduced chi-squared value (${\chi }^2/\nu$) from each of the peak fits (see Table 7 for the ${\chi }^2/\nu$ values). Finally, to ensure sufficiently small ${\sigma }_{E,\mathit{FWHM}}$ uncertainties, a lower limit was placed on the total counts in a given peak. For the high-gain spectra, the ${\chi }^2$/$\nu$ and peak count limits were ≤1.5 and ≥2500 counts, respectively. For the low-gain spectra, these limits were relaxed to ensure that the highest energy peaks were used, and were ≤2.5 and ≥300 counts, respectively. The red data points in Fig. 32 are those that passed these selections, and are used in the high- and low-gain fits to Eq. (2).

Fig. 32.

GRS FWHM energy resolution, ${\sigma }_{E,\mathit{FWHM}}$, versus gamma-ray energy, $E$, for the high-gain (A) and low-gain (B) channels. The data points are those given in Table 7 from 28 prominent peaks seen in the GRS ICO measurements. The ${\sigma }_{E,\mathit{FWHM}}$ values are those obtained by carrying out the peak fits described in the Appendix. The error bars represent the full propagated one-standard-deviation uncertainties from each peak fit. Red data points indicate those values that passed total counts and ${\chi }^2$/$\nu$ selection criteria. For the high-gain spectra, the criteria were >2500 counts and ${\chi }^2$/$\nu <1.5$; for the low-gain spectra, the criteria were >300 counts and ${\chi }^2$/$\nu <2.5$. The selected data points were fit to Eq. (2). The fit parameters and associated uncertainties are shown in the plot along with the goodness-of-fit value (${\chi }^2$/$\nu$) for each fit. The best-fit functions are shown by the solid red lines. The best-fit function for the high-gain spectra is shown by the dashed red line in panel B

For the final high-gain fit, the parameter $b_2$ was set to zero, for the reason that when it was included, it resulted in a poorer ${\chi }^2$/$\nu$ value (1.11 versus 1.04), the overall uncertainty for ${\sigma }_{E,\mathit{FWHM}}$ was significantly larger (0.13 keV versus 0.03 keV), and the $b_2$ value was quite small ($5.9\times {10}^{-8}$). In short, including the $b_2$ parameter resulted in a poorer fit. When $b_2$ is set to zero, the resulting fit for the high-gain spectra shows a reduced ${\chi }^2$/$\nu$ of 1.04 indicating a robust fit. Using the fitted parameters and their associated uncertainties, the inferred energy resolution at 1332.5 keV is 2.09 ± 0.03 keV. This performance is slightly poorer than what was measured just after spacecraft delivery (Fig. 22) and may reflect multiple factors. For example, the ¹³⁷Cs peak shows a drift of 0.038 keV from the beginning to the end of the measurement period (Fig. 45), so correcting this variation could reduce the measured energy resolution to as low as 2.05 keV. There may also be a modestly different electronic noise environment that could affect the energy resolution. Finally, there may also be a slight amount of radiation damage due to the 30 days that the GRS spent in space prior to its initial operation (see Sect. 5.2).

Even so, this performance is significantly better than previous space-based HPGe detectors, which generally have energy resolution values at 1332.5 keV in the 3 to 4 keV range (Smith et al. 2002; Goldsten et al. 2007; Diehl et al. 2018; Roques et al. 2022). The prior best-in-class energy resolution for a space-based HPGe detector was 2.27 keV at 1117 keV for one of the nine detectors on the RHESSI solar gamma-ray spectrometer (Smith et al. 2002). For the low-gain spectra, the inferred energy resolution is 2.21 ± 0.03 keV, which is about 6% wider than the high-gain measurement, and is broadly consistent with the increase seen in the pre-launch data for the low-gain spectra (Fig. 22).

The fit parameters for Eq. (2) can provide information about the detector performance characteristics. The parameter $b_0$ provides a measure of the baseline energy resolution due to energy-independent factors, such as electronics noise and broadening due to different energy per channel factors from the high- and low-gain spectra. The $b_0$ value of 1.29 keV for the high-gain spectra is slightly larger than the width of the electronic pulser peak of 1.11 keV (which occurs at an equivalent energy of 107 keV; the spectral fit for this peak is not shown). For the low-gain spectra, $b_0$ has a larger value than can be attributed, at least in part, to the larger energy-per-channel covered by the low-gain spectra. The $b_1$ parameter provides information about the statistical fluctuations in the Ge sensor (Knoll 2000), where

$$b_1^2={\left(2.35\right)}^{2}FϵE.$$ 3

Here, $ϵ=2.96\text{ eV}$ is the energy necessary to create an electron-hole pair in the Ge sensor, and $F$ is a factor known as the Fano factor, which quantifies the reduction in statistical fluctuations from Poisson statistics seen with various types of detectors. For both the high- and low-gain spectra, it is inferred that $F=0.14$. According to Knoll (2000) and other studies (e.g., Pehl and Goulding 1970; Croft and Bond 1991; Devanathan et al. 2006), for Ge sensors $F$ is typically in the range of 0.06 to 0.15, with indications that values in the lower range are preferred. Thus, while 0.14 might be considered “acceptable”, the fact that it is in the upper range indicates that for ICO there are some residual broadening effects that have not been fully quantified. Finally, since this analysis shows that $b_2=0$ is a best-fit value, it indicates that incomplete charge trapping is, at worst, a small effect for the ICO operation of the GRS. As cruise and orbital operations continue, and radiation damage and subsequent annealing operations occur, it will be instructive to observe how the Ge sensor performance changes from this well characterized in-space baseline performance.

On 11 December 2023, the high voltages on the NS were raised and the three ³He sensors started to collect data. Figure 33 shows space-based energy deposition spectra for the three sensors compared to the ground-based measurements with the polyethylene sensor. A significant difference between the space- and ground-based data is that the background at the lower channels is dramatically higher for the space-based measurements compared to the ground-based data. These higher count rates are due to charged particle energy deposition present in space from GCRs that is not present with the ground-based data. The highest count rate of the three sensors is seen in the polyethylene sensor. After subtracting a non-neutron continuum, the net neutron background count rates are 0.23 cps, 0.14 cps, and 1.61 cps for the bare, Cd-covered, and polyethylene-covered sensors, respectively. This observation confirms that the dominant neutron energy range at the Psyche spacecraft is higher energy neutrons, and that the flux of lower energy epithermal and thermal neutrons is low.

Fig. 33.

Energy deposition spectra from the NS ³He sensors. Data from the bare, Cd-covered, and polyethylene-covered sensors are shown in the black, red, and blue traces, respectively. The orange trace shows ground-based data from the polyethylene-covered sensor, where it has been multiplied by a factor of 10 to match the peak values of the flight polyethylene-covered sensor data

Figure 34 shows a summary of NS event-mode data that were collected during the initial turn on operation. Here, $dt$ versus peak-to-peak data show that the population of charged particle energy deposition extends over a wider range of $dt$ values than the true neutron counts (the peak-to-peak value is proportional to the energy deposition in the sensor). When a $dt$ cut is applied to the data (which can be done via a commandable parameter in flight software), this cut can reduce the background by nearly an order of magnitude, thus increasing the signal-to-background in a manner that is similar to the improvement seen in the GRS anticoincidence spectra shown in Fig. 31. Additional analyses of these event-mode data are given by Peplowski et al. (2025) where further background reduction can be achieved.

Fig. 34.

NS in-space event-mode data from the Cd-covered sensor. (A) shows a two-dimensional plot of $dt$ values versus peak-to-peak values. The scale bar is given in log units of 10^counts. (B) shows the $dt$-summed peak-to-peak values when accepting all $dt$ values (black), and when only allowing $dt$ values in the range from 250 to 275

Periodic Maintenance Calibration #1

The HPGe sensor was operated a second time during a spacecraft operations period known as the Periodic Maintenance Calibration (PMC) #1. During this time, the HPGe sensor was cooled to 90 K, the high voltage was raised to 2000 V, and data were collected for four days. Figure 35 compares data from the ¹³⁷Cs calibration source taken during three time periods: during the initial pre-launch CPT in September 2021 (Fig. 35A); during ICO in November 2023, 35 days after being in space (Fig. 35B); and during PMC #1, 277 days after being in space (Fig. 35C). The trending of peak widths (both FWHM and full-width, tenth maximum, FWTM) show an increasing broadening due to space-based radiation damage (Fig. 35D). The source of radiation damage is due not only to nominal GCRs hitting the HPGe sensor, but there were a number of high intensity solar energetic particle events that hit the Psyche spacecraft during the time between ICO and PMC #1 (see Sect. 5.3) that possibly contributed to the radiation damage. Based on prior operation with space-based HPGe detectors, the amount of energy resolution degradation is consistent with prior experience (e.g., Evans et al. 2017). Subsequent PMCs (roughly every six months) will be used characterize the energy resolution performance versus time (see Sect. 7.1). This information will then be used as input, in part, for determining the annealing duration needed prior to the science measurements at Psyche.

Fig. 35.

Comparison of energy resolution performance for three different time periods using the 661 keV ¹³⁷Cs peak. (A) This shows the performance during the September 2021 pre-launch CPT at JPL where the energy resolution is 1.52 keV. (B) The energy resolution of the 661 keV peak was 1.61 keV during ICO. (C) The energy resolution of the 661 keV peak is significantly broadened at 2.25 keV during the first PMC due to radiation damage from GCRs and solar particle events. (D) Time dependence of the FWHM and full-width, tenth-maximum (FWTM) for the three time periods

Early Cruise Operation

Figures 36 and 37 provide a summary of the operation of the GRS AC shield and NS during the first year after initial checkout. A number of different space environmental and operational aspects are illustrated with these data.

Fig. 36.

Time variation of the AC shield high-gain count rate during the first seven months of the orbital cruise mission. Various environmental and operational events that took place during this time period are labeled

Fig. 37.

Time variation of the three NS sensor count rates during the first seven months of the orbital cruise mission. Blue, red, and green show data from the bare, Cd-covered, and polyethylene-covered sensors, respectively. Various environmental and operational events that took place during this time period are labeled

Most prominently, these data show numerous instances of solar particle events (SPEs) hitting the spacecraft. These events, which are mostly made up of protons with energies greater than tens of MeV, typically last many hours to many days. These events occur more frequently during solar maximum, where the current solar maximum is expected to occur in the 2024 to 2025 time period. Figure 38 shows GRNS count-rate measurements along with GOES Earth-orbiting energetic proton data (University of Colorado Boulder and SWx TREC 2019). These data, which were acquired soon after launch when the Psyche spacecraft and Earth were relatively closely aligned in solar longitude, indicate that the same SEP events detected by GOES were also detected by the GRNS. As the Psyche spacecraft travels away from Earth alignment, this agreement will lessen, since Psyche will be located in a different space environment than at Earth. Also notable from these events is that the GRNS data not only carries information about the existence of SPE events, but likely has information about the proton energy spectra, where lower energy protons show higher count rates than higher energy protons. This is particularly seen with the NS count rates, where the three different ³He sensor coverings (bare, Cd, polyethylene) stop increasingly higher proton energies. Short, few-minute-long count-rate spikes are seen in the high-gain AC shield data, but not in the low-gain, fast neutron or ³He sensor data. Because these enhancements are only seen in the high-gain AC shield data, which has a photon energy threshold of ∼20 keV, it is likely they are due to X-ray energy deposition in the AC shield. A number of these events align in time with known X-ray flares as identified by GOES X-ray data (University of Colorado Boulder and SWx TREC 2019), an example of which is shown in Fig. 39. We therefore conclude that most, if not all, of these spikes are due to short duration (few tens of minutes) X-ray flares.

Fig. 38.

Time variation of various GRNS count rate values are shown in the top four panels (A, B, C, D) during the first two months of orbital operation. The bottom panel (E) shows the time variation of three proton energy bands from the Earth-orbiting GOES spacecraft (University of Colorado Boulder and SWx TREC 2019)

Fig. 39.

Time variation of the AC shield high-gain count rate as acquired from the engineering data packet with a time resolution of 60 seconds (black trace) that shows a likely detection of a solar X-ray flare. GOES B X-ray data is scaled to AC-shield count rate data (red trace)

Operational aspects of the AC shield and NS cruise data include the stochastically timed safing of these sensors. As described in Sect. 3.5.3, all the GRNS sensors have a safing feature such that when the count rates rise above a threshold count rate, the HV is lowered for a set period of time, and then re-raised to determine if the count rate is reduced below that same threshold. As seen, this safing has been triggered a number of times during the first year of operation. Interestingly, there are some events that cause the AC shield to safe, but not the NS sensors (e.g., see event between 5/31 and 6/22/2024), which indicates that the AC shield is likely more sensitive to charged particle deposition than the NS.

Figure 40 shows a gamma-ray burst (GRB) that was detected on 25 August 2024, and corresponds to GRB240825A (https://user-web.icecube.wisc.edu/~grbweb_public/Individual_GRB_webpages/GRB240825A.html). While a full analysis of this GRB is yet to be carried out, we can check the consistency of this measurement with known parameters for this GRB. As viewed from the Earth, this burst was located at right ascension (RA) 344.5°, declination (DEC) 1.03°. For this date, the Psyche spacecraft as viewed from the Earth was at RA 174.2°, DEC 3.0°, and at a distance of $4.98\times {10}^8\text{ km}$, which is almost directly opposite in the sky compared to the GRB. The light travel time difference between Psyche and Earth was 1662.1 seconds. At Earth, the GRB was detected at a time of 15:52:59 UTC; at Psyche the detection time was 16:20:15.656 UTC, for a measured time difference of 1636.7 seconds. The difference between the light travel and detected time differences is 25.5 seconds, which is consistent with the GRB occurring at a location roughly opposite on the sky from the Psyche spacecraft location. Resolving the remaining time difference will require an analysis of the exact position locations of the spacecraft, Earth, and the GRB, and their relative timings.

Fig. 40.

Time profile of a gamma-ray burst detected with the AC shield burst mode on August 25, 2024. The time of detection is given in the figure

Data Products and Processing

The GRNS data are distributed to the science community via NASA’s Planetary Data System (PDS) through the Psyche mission Science Data Center at Arizona State University. Per PDS convention, the GRNS data categories include raw, calibrated, and derived data. There are 12 GRS raw data product types (Table 5): high- and low-gain HPGe detector raw spectra, high- and low-gain HPGe AC spectra, high- and low-gain AC shield spectra, shield fast-neutron spectrum, gamma-ray burst counts, gamma-ray counters, gamma-ray events, fast neutron events, and engineering data.

Table 5.

Summary of GRS raw data products

GRS Product	Product Type	Description
HPGe Low Gain	16,384 bin spectra	Raw spectra collected from the high purity germanium (HPGe) detector. Covers 0 to 9200 keV energy range.
HPGe High Gain	16,384 bin spectra	Raw spectra collected from the high purity germanium (HPGe) detector. Covers 0 to 2800 keV energy range.
HPGe AC Low Gain	16,384 bin spectra	Anticoincidence (AC) spectra collected from the high purity germanium (HPGe) detector. Covers 0 to 9200 keV energy range.
HPGe AC High Gain	16,384 bin spectra	Anticoincidence (AC) spectra collected from the high purity germanium (HPGe) detector. Covers 0 to 2800 keV energy range.
AC Shield Low Gain	1024 bin spectra	AC spectra collected from the shield detector. Used for cosmic-ray proton monitoring. Sensitive to energy deposition events up to ∼35 MeV.
AC Shield High gain	512 bin spectra	AC spectra collected from the shield detector. Used for neutron science. Sensitive to energy-deposition events up to ∼500 keV.
AC Shield Fast Neutron	4 × 256 bin spectra	AC Shield spectra capturing double-pulse events, grouped into four separate categories, based on their time to second pulse values.
Gamma-Ray Burst	1024 time-binned counts	Counts from a triggered gamma-ray burst event.
Gamma-Ray Counters	binary table	Counters of events measured across the integration period. Used for diagnostic purposes.
Gamma-Ray Events	binary table	Events read from the hardware. Includes events not considered valid to be included in a spectrum. Used for diagnostic purposes.
Fast Neutron Events	binary table	Fast neutron events and metadata read from the hardware. Includes events not considered valid to be included in a spectrum. Used for diagnostic purposes.
GRS Engineering	binary table	Measurements of the health and status of the GRS instrument.

The NS raw product pipeline generates four raw data product types (Table 6): NS spectra, NS counter, raw event, and engineering data products. The raw NS spectra product consists of three 256 bin histograms, from the polyethylene, cadmium, and bare wrapped detectors. As with the GRS, the NS histogram data are compressed using a fast lossless compression. More details for the raw and calibrated data are described by Espiritu et al. (2022a,b).

Table 6.

Summary of NS raw data products

NS Product	Product Type	Description
NS Spectra	3 × 256 bin spectra	Spectra histogram arrays collected from the Bare, Cadmium-, and Polyethylene-wrapped sensors.
NS Events	binary table	Event time series data collected by the NS instrument. Events are collected even if they are not considered valid to be included in an NS spectra packet.
NS Counters	binary table	Contains counters of events across the integration period.
NS Engineering	binary table	Measurements of the health and status of the NS instrument

Expected Measurements During Cruise, Mars Flyby, and at Psyche

Summary of Expected Measurements for Each Mission Phase

The major GRNS operational phases throughout the Psyche mission include:

• Instrument Checkout Operations (ICO): November – December 2023

• Interplanetary Cruise 1: January 2024 – March/April 2026

• Mars gravity assist: March/May 2026

• Interplanetary Cruise 2: May 2026 – May 2029

• Psyche Approach: May 2029 – August 2029

• Psyche Orbit Operations (Orbits A through B1): August 2029 – February 2030

• Psyche Orbit Transfer (Orbit B1-to-D transfer): February 2030 – May 2030

• Psyche Orbit D: May 2030 – August 2030

• Psyche Orbit D margin: August 2030 – October 2030

• Psyche Orbit Transfer (Orbit D-to-C transfer): October 2030 – January 2031

• Psyche Orbit C: January 2031 – April 2031

• Psyche Orbit Transfer (Orbit C-to-B2 transfer): April/May 2031

• Psyche Orbit B2: May 2031 – August 2031

Full GRNS operations are performed during ICO, Mars gravity assist, Psyche Orbit B1-to-D transfer, Orbit D, and Psyche Orbit D-to-C transfer. During interplanetary cruise and early Psyche orbit operations (Orbits A and B), the GRNS operates in a reduced science mode where neutron and AC shield measurements are made but the HPGe sensor is not generally operated.

The cruise measurements, during interplanetary cruise periods 1 and 2, include the near-continuous operation of the AC shield and all three NS sensors, as well as GRS HPGe operation roughly every six months during spacecraft PMCs. The entire GRNS will operate during the Mars gravity assist in the spring of 2026, where the Psyche spacecraft will approach the surface of Mars at an altitude of ∼3500 km. There are multiple purposes for these cruise operations. Most importantly, there are aspects of GRNS operation that cannot simulated or replicated on the ground. For example, the GCR flux was measured at 1 AU, contemporaneously with Earth-orbiting space weather assets (see Fig. 38). The continued operation of the AC shield and NS sensors enables the GCR flux variation to be monitored throughout the mission, which facilitates precise knowledge of the asteroid-incident GCR flux once reaching (16) Psyche. Knowledge of this flux is an important input for interpreting the GRNS measurements. Continuous monitoring of the energetic particle environment is important not only to fully understand the GCR flux while at Psyche, but provides information about the energetic particle radiation encountered by the HPGe sensor. This information is important for optimizing the HPGe annealing procedure(s) prior to science data acquisition while in orbit about Psyche (Peplowski et al. 2019a). We finally note that the data acquired during cruise is critical not only for achieving Psyche science objectives, but can be useful for other studies of the space environment, such as understanding the time and spatial dependent GCR flux (Rodgers et al. 2015; Lawrence et al. 2016), as well as carrying out space weather and heliophysics studies with multi-spacecraft correlation analyses (Battarbee et al. 2018; Winslow et al. 2018). Additional studies could combine the GRNS data with cruise data from other instruments (e.g., interplanetary magnetic field data from the magnetometers) and from engineering subsystems (e.g., single event effect data from onboard computer memories), which in part can help inform radiation effects on the Psyche spacecraft.

During inbound and outbound Orbit D transfer phases, the spacecraft is at >3.5-Psyche-radii altitude, and neutron and gamma-ray measurements provide a measure of background rates and spectra. During Orbit D, the spacecraft is within 1-Psyche-radius altitude and gamma-ray and neutron measurements are sensitive to the composition of Psyche’s surface. The average planned altitude during this period is 76 km (0.68 Psyche radius), with a range of 59 to 91 km (0.53 to 0.83 Psyche radius) (Fig. 41). These are the measurements that contribute to Psyche science requirement closure. Orbit D is planned to last for 100 days, approximately 75% of which will be nadir-pointed and thus contribute to GRNS science observations. The remaining 25% of time includes off-nadir pointing to downlink data to Earth. An additional ∼60 days of margin are included in the Orbit D phase. Based on the current orbit-plan estimates and including the 75% duty cycle for nadir-pointed attitudes, the total Orbit D times including margin for altitudes less than 0.6, 0.8, and 1.0 body radius are ∼19 days, ∼119 days, and ∼122 days, respectively. The total times less than 1.0 body radius for the B1-to-D and D-to-C transfer orbits are ∼53 and ∼2.5 days, respectively. It is unknown how much nadir-pointing time will be available during these transfer orbits.

Fig. 41.

(Left) Histograms of altitude occurrences for the Psyche mission Orbits D and C for the current orbital mission plan (black and red, respectively), along with the transfer orbits B1 to D (green) and D to C (blue). The altitude bins are 1-km wide, and the occurrences indicate the number of days per 1-km-wide altitude bin. For Orbit D, altitudes range from ∼0.53 to 0.83 Psyche radius, and for Orbit C, altitudes range from 1.3 to 2.1 Psyche radii. (Right) For reference, graphical representation of the four Psyche orbits during the primary orbital mission (Polanskey et al. 2025, this collection). The average altitudes for Orbits A and B (not shown at left) are 6.4 and 2.7 Psyche radii, respectively

Due to Psyche’s irregular shape, a polar orbit is not expected to be stable at an orbit radius of <1-Psyche-radius for a sufficiently long time to meet mission safety guidelines. Thus, during Orbit D the spacecraft will be in an equatorial orbit that samples the surface composition equatorial of ±15° latitude. The total surface coverage is 86%, which is enabled by the broad size of the NS and GRS spatial footprints. The polar regions will not be measured or mapped by Psyche GRNS during Orbit D. However, during Orbit C, which is currently planned to take place after Orbit D, the spacecraft will have full latitude coverage. The altitude range during Orbit C is 1.3 to 2.1 body radii (Fig. 41). Neutron data from the Lunar Prospector mission can be used to estimate the order-of-magnitude NS count rate for Psyche at this altitude range. Prior to going into a circular orbit, the Lunar Prospector spacecraft obtained data during multiple elliptical orbits about the Moon (Maurice et al. 2004). The Lunar Prospector and Psyche neutron measurements are similar such that both use the same sized neutron sensors, and the Lunar Prospector measurements were made around solar minimum conditions, as will be Psyche during Orbit C in 2031. The Lunar Prospector count rate of its Cd-covered neutron sensor for 1.3-to-2.1 body radii altitude range at the Moon is 1.91 ± 0.54 cps. The measured Psyche cruise count rate is 0.14 cps. Because of this low background level, there is therefore a good opportunity to obtain additional neutron data to characterize global compositional variabilities during Orbit C, albeit with broader spatial resolution than for the lower altitude Orbit D.

Science Closure

During the early development of the Psyche mission, the asteroid (16) Psyche was thought to be the exposed core of a protoplanetary object. Initial theories suggested that the asteroid would be ∼90% metal (≥4 wt% Ni) and ∼10% Mg-rich silicates. On the basis of these expectations, sensitivity simulations and calculations were carried out to determine what elements would be measured by the GRNS, including required measurement times, detection limits, and statistical plus systematic uncertainties. Figure 2 shows an example of simulated spectra in the region around the 1454 keV Ni gamma-ray line. While our expectations of (16) Psyche’s composition have broadened significantly since the original studies, our sensitivity studies and mission requirements are still based on this assumed composition. These sensitivity calculations, which were finalized after the GRNS design was completed, included the following details. For GCR-generated gamma-ray lines, gamma-ray production was calculated using the particle transport code MCNPX (Pelowitz 2005). The gamma-rays were transported to the assumed spacecraft altitude with standard geometric relations (e.g., Prettyman et al. 2006; Chabot et al. 2021). The GRS efficiency was determined using particle transport calculations (Agostinelli et al. 2003), as was done for the MESSENGER (Peplowski et al. 2012), and used a full accounting of the Psyche GRS assembly. Gamma-ray background (line and continuum) was taken into account using the measured MESSENGER GRS background. A number of factors were given conservative estimates to provide margin in the sensitivity calculation. These margins guard against unanticipated performance degradations – either in the instrument, mission operations, or both – that would prevent the accomplishment of the primary measurement requirements. These factors include: 1) an assumed energy resolution of 3.5 keV (MESSENGER GRS optimum performance); 2) a constant altitude of one Psyche radius; 3) a lower limit value of the incident GCR flux at the start of the original orbit phase; 4) a partial background reduction to account for the expected improvement from the boom. Given the detection thresholds and required precisions listed in Table 1, the output values of the sensitivity calculations are the times listed in the rightmost column of Table 1. For some elements (e.g., Fe, K), the time needed to meet the requirements is fairly short (less than a week). The elements that take the longest to meet the required threshold and precision are S and Ca. Post-calibration measurements of the Psyche GRNS showed that the instrument meets or exceeds all performance requirements (Peplowski et al. 2025), and the current Psyche mission operations plan includes at least 75 days of nadir pointing (not including time margin) at an altitude well below 1-Psyche-radius. We therefore conclude that the GRNS will exceed the required performance, providing margin against the now less-constrained range of compositions expected for the asteroid.

List of acronyms

AC

Anticoincidence

ADC

Analog-to-Digital Converter

ANSI

American National Standards Institute

APL

Applied Physics Laboratory

BG

Background

CC

Cryocooler

CCDB

Cryocooler Driver Board

CCPCB

Cryocooler Power Conditioning Board

CORDIC

Coordinate Rotation Digital Computer

CPT

Comprehensive performance test

DN

Digital number

DNL

Differential nonlinearity

DPU

Data Processing Unit

EM

Engineering model

EMC

Electromagnetic compatibility

EMI

Electromagnetic interference

FEP

fluorinated ethylene propylene

FIFO

First-in, first-out

FPGA

Field Programmable Gate Array

FWHM

Full-width, half-maximum

FWTM

Full-width, tenth-maximum

GCR

Galactic cosmic ray

GOES

Geostationary Operational Environmental Satellite

GRB

Gamma-ray burst

GRNS

Gamma-Ray and Neutron Spectrometer

GRS

Gamma-Ray Spectrometer

HPGe

High-purity Germanium

HV

High voltage

HVFB

High Voltage Filter Box

HVPS

High Voltage Power Supply

ICO

Initial checkout

INL

Integral nonlinearity

INTEGRAL

INTErnational Gamma-Ray Astrophysics Laboratory

IR

Infrared

JFET

Junction-gate field-effect transistor

JHU/APL

Johns Hopkins University Applied Physics Laboratory

JPL

Jet Propulsion Laboratory

LCTU

Life Cycle Test Unit

LLNL

Lawrence Livermore National Laboratory

LVPS

Low Voltage Power Supply

MCNPX

Monte Carlo N-Particle eXtended

MESSENGER

Mercury Surface, Space Environment, Geochemistry, and Ranging

MIP

Minimum-ionizing particle

MLI

Multilayer insulation

MSPS

Mega-samples per second

NEAR

Near Earth Asteroid Rendezvous

NPO

Negative-Positive-Zero

NS

Neutron Spectrometer

PCB

Printed Circuit Board

PDS

Planetary Data System

PMC

Periodic Maintenance Calibration

PMT

Photomultiplier tube

PTFE

Polytetrafluoroethylene

PWM

Pulse-width modulation

SCUM

Sensor Control Unit Module

SPE

Solar particle event

THD

Total harmonic distortion

TTSP

Time-to-second pulse

UV

Ultraviolet

VDC

Volt direct current

Appendix:  GRS Energy, Total Counts Check, and Gain Temperature Dependence

This first part of the appendix describes the channel-to-energy calibration used for the GRS ICO dataset. The gain of the system is dependent on the temperature, $T$_Ge, of the Ge crystal. Thus, the energy calibration derived here for the crystal at $T$_Ge = 95 K will be different than that derived for either the pre-flight calibration data ($T$_Ge = 90 K) and future flight data, for which $T$_Ge is expected to be in the range of 85 K to 90 K.

Table 7 lists 28 of the most prominent peaks that are used for the channel-to-energy calibration. Each peak in Table 7 was fit with a gaussian plus a second-order polynomial function, $f(x)$, within a ∼7-keV window around each peak:

$$f\left(x\right)=A_0{e}^{\frac{-{[z(x)]}^2}{2}}+A_3+A_{4}x+A_5{x}^2,$$ 4

where,

$$z\left(x\right)=\frac{x-A_1}{A_2}.$$ 5

These initial fits were carried out using spectra channels (i.e., $x=C$) that ranged in value from 0 to 16,303 (i.e., a total of 16,384 channels). The returned parameters (including parameter uncertainties) for each fit are the peak height ($A_0$), centroid ($A_1$), width ($A_2$) and three polynomial parameters ($A_3$, $A_4$, $A_5$). Note that for the high-gain data, only peaks with energies up to the 2754-keV peak are included. Figure 42 shows the peak centroid values (‘centroid channel’ column in Table 7) plotted versus the reference peak energy values (‘energy’ column in Table 7) for both the high- and low-gain spectra. A second-order polynomial was fit to the values in Fig. 42 with the fit parameters shown in the figure. To demonstrate the robustness of the channel-to-energy calibration, a calibrated energy scale, $E$_cal, was generated using the fit parameters of Fig. 42, and each peak was re-fit using these energy values (i.e., x = $E$_cal). The residual between the reference peak energies, $E$_ref, and newly fit energy centroids is shown in Fig. 43. The error bars in Fig. 43 are the one-sigma uncertainties for the fitted energy centroid values. As seen, within the uncertainties, the residuals show no obvious bias or energy dependent pattern. This indicates that the energy calibration is robust across all energies for both the high- and low-gain spectra.

Fig. 42.

Channel-to-energy calibration of high-gain (A) and low-gain (B) spectra. The data points show the data values given in Table 7; the red solid line shows the fit of a second-order polynomial function. The best-fit parameters for each fit are shown in each figure

Fig. 43.

Residuals from the channel-to-energy calibration where the differences are shown between the reference and calibrated energy values versus the calibrated energies. The error bars show the one-standard-deviation statistical uncertainty in the calibrated energy peak centroids. Panel A shows the high-gain data; panel B show the low-gain data

Another data robustness test is to compare the total counts within each peak for the high- and low-gain data. The derived total counts should be the same for each type of spectra since they both originate from the same detector signal. The total counts for each peak were determined by subtracting the fitted polynomial background and summing the net peak counts. The count uncertainty is assumed to be based on Poisson counting statistics, and is the square root of the sum of peak and background counts within a window of three FWHM around the peak centroid. Here, the FWHM value is 2.35 times the peak width $A_2$. Figure 44 shows the difference between the high- and low-gain peak counts as a function of the high-gain peak counts. As expected, the count difference is consistent with zero within the one-standard-deviation uncertainty. This demonstrates that both high- and low-gain spectra individually return the same counts within the limits of the available statistics.

Fig. 44.

Total count difference between the high- and low-gain data for all the peaks given in Table 7. The errors bars show the combined one-standard-deviation statistical uncertainties for both peak fits

Fig. 45.

Time-dependent variation of the ¹³⁷Cs peak centroid throughout the GRS ICO data collection. A linear fit to the data is shown by the solid red line

Finally, to understand the effects of possible temperature variations on the gain of the GRS, Fig. 45 shows the variation of the peak centroid of the ¹³⁷Cs peak at ∼662 keV. As seen, there is a roughly monotonic variation of the peak centroid throughout the 13.9-day GRS background measurement. A linear fit to this relation is shown by the solid red line. The energy difference between the beginning and end times is 0.038 keV. Additional detail about the temperature dependent gain variation of the GRS is given by Peplowski et al. (2025).

Declarations

Competing Interests

The authors declare no conflict of interest.