In the US, there are five facilities that carry out nuclear experiments for space research: the Radiation Effects Facility at Texas A&M University (TAMU) (Texas A&M University, 2023); the Berkeley Accelerator Space Effects (BASE) facility at Lawrence Berkeley National Laboratory (LBNL) (University of California Berkeley, 2023); the NASA Space Radiation Laboratory (NSRL) at Brookhaven National Laboratory (BNL) (Brookhaven National Laboratory, 2023a); the Single Event Upset Test Facility (SEUTF) at BNL (Brookhaven National Laboratory, 2023b); and FSEE, the new Single Event Effects test facility at the Facility for Rare Isotope Beams (FRIB) (Michigan State University, 2023a) at Michigan State University (MSU).

The TAMU facility is a part of the Association for Research of University Nuclear Accelerators (ARUNA) (ARUNA, 2023), which also includes the Fox Accelerator Laboratory at Florida State University (Florida State University, 2023), the Ion Beam Analysis Laboratory at Hope College (Hope College, 2023), the Edwards Accelerator Laboratory at Ohio University (Ohio University, 2023), the Triangle Universities Nuclear Laboratory (TUNL) at Duke University (Duke University, 2023), the Ion Beam Analysis Laboratory at Union College (Union College, 2023), the Radiation Laboratory at the University of Massachusetts-Lowell (University of Massachusetts Lowell, 2023), the Accelerator Laboratory at the University of Kentucky (University of Kentucky, 2023), the Institute for Structure and Nuclear Astrophysics (ISNAP) (University of Notre Dame, 2023) at the University of Notre Dame, and the Center for Experimental Nuclear Physics and Astrophysics (CENPA) (University of Washington, 2023) at the University of Washington. Since the other ARUNA facilities produce beams with maximum energies significantly lower than those at TAMU, they have not been used for space research to date.

There are other accelerators in the US that also have not been utilized for space research, including: the ATLAS facility (Argonne National Laboratory, 2023) at Argonne National Laboratory (ANL); the Relativistic Heavy Ion Collider (RHIC) (Brookhaven National Laboratory, 2023c) at BNL; the Continuous Electron Beam Accelerator Facility (CEBAF) (Jefferson Laboratory, 2023) at the Thomas Jefferson National Laboratory (JLab); the Spallation Neutron Source (SNS) (Oak Ridge National Laboratory, 2023) at Oak Ridge National Laboratory (ORNL); the Los Alamos Neutron Science Center (LANSCE) (Los Alamos National Laboratory, 2023a) and Weapons Neutron Research (WNR) facility (Los Alamos National Laboratory, 2023b) at Los Alamos National Laboratory (LANL); and the Tevatron (Fermilab, 2023) at Fermilab National Accelerator Laboratory (FNAL). Finally, there are some accelerators that have been decommissioned, including the K500 and K1200 cyclotrons at the National Superconducting Cyclotron Center (Michigan State University, 2023b) at MSU.

In Canada, the Proton Irradiation Facility (TRIUMF, 2023a) and Neutron Irradiation Facility (TRIUMF, 2023b) at Canada’s National Particle Accelerator Center TRIUMF in Vancouver have been used for irradiation studies of electronic systems.

In Europe, nine accelerators in seven countries have been used for space research. These include: the Heavy Ion Facility (HIF) and Light Ion Facility (LIF) at the Universite Catholique de Louvain (UCL) (Universite Catholique de Louvain, 2023) in Louvain-la-Neuve, Belgium; the RADiation Effects Facility (RADEF) (University of Jyväskylä, 2023) at the Accelerator Laboratory at the University of Jyväskylä, Finland; the G4 facility at the Grand Accelerateur National d’Ions Lourds (GANIL) (GANIL, 2023) in Caen, France; the Heavy Ion Synchrotron SIS18 at the Gesellschaft fur ScherwIonenforschung (GSI) Helmholtz Centre for Heavy Ion Research (Durante et al., 2010) in Darmstadt, Germany; the INFN Southern National Laboratory (LNS) (INFN, 2023) in Catania, Italy; the Holland Proton Therapy Center (Fleury et al., 2021) in Delft, the Netherlands; the Keuring voor Ingebruikname (KVI) Center for Advanced Radiation Technology (KVI-CART) facility (Univ of Groningen, 2023) at the University of Groningen in the Netherlands; the CERN High energy AcceleRator Mixed field facility (CERN CHARM) (Alía et al., 2018) at CERN in Geneva, Switzerland; and the Proton Irradiation Facility (PIF) (Hajdas et al., 2001) at the Paul Scherrer Institute (PSI) in Switzerland.

In Asia, there are 28 accelerator facilities (Tanaka, 2020). While most of these are low-energy facilities, a number can produce beams of sufficiently high energy for space studies. These include: the Research Center for Nuclear Physics (RCNP) (Osaka Univ., 2015) at Osaka University in Osaka, Japan; the Japan Proton Accelerator Research Complex (JPARC) (KEK, 2018) in Tokai-Mura, Japan; the RIKEN Nishina Center (RIKEN, 2023) in Wako, Japan; the Heavy Ion Research Facility (HIRFL) (Chinese Academy of Sciences, 2023a) in Lanzhou, China; the Chinese Neutron Spallation Source (CSNS) (Chinese Academy of Sciences, 2023b) in Dongguan, China; the High Intensity Heavy-ion Accelerator Facility (HIAF) (Chinese Academy of Sciences, 2023c) in Huizhou, China; the Beijing RI beam Facility (BRIF) (Chinese Institute of Atomic Energy, 2023) in Beijing, China; and the RAON facility (Institute for Basic Science, 2023) in Daejeon, Korea.

Each of the facilities listed above have different maximum beam energies and intensities, detector stations, number of hours available for users, and specializations. What they have in common, however, is that the requests for beam time from users for space research significantly outstrips the availability. The space electronics effects community, for example, has requested twice the available beam across the relevant US facilities in 2022; that factor is expected to grow even larger in the coming decade. A portion of this demand could be met by repurposing existing accelerators that have become dormant, such as the aforementioned K500 and K1200 cyclotrons at MSU.

Understanding the effects of GCRs with the highest (well over GeV/u) energies is important to the space radiation protection community. However, with no measurements at projectile energies over 3 GeV/u, simulations of these effects lack an empirical foundation. Higher-energy measurements are also required to understand electronic effects in the latest circuits whose size is greater than the range of ions available at nearly all accelerators used in these studies so far. There is, however, a possibility to fill some of these critical nuclear data gaps by using the RHIC beams at BNL. A beam time proposal (Cebra, 2022) was recently made to bombard C, Al, and Fe targets with He, C, Si, and Fe ions at energies from 3–50 GeV/u, and to detect the light particle production (the “secondaries”) with the STAR detector. This measurement, however, would have to be completed before the conversion of RHIC to the Electron-Ion Collider (EIC) (Khalek et al., 2021) project begins in ∼2025.

An additional capability needed for space research accelerator measurements is a larger beam diameter. Traditional accelerators have beam spot diameters that are capable of irradiating one chip at a time. Rastering and defocusing techniques now enable some of the facilities mentioned above to irradiate batches of chips. However, laboratory measurements can more realistically reproduce the GCR damage inflicted in space by irradiating large, complex subsystems all at once. The special approaches needed to obtain such wide beam diameters while keeping uniform beam densities should be implemented at accelerators carrying out space-based research.

At present, the primary means of disseminating nuclear reaction cross sections measurements relevant for space research is via journal publications. There are, however, two compilations of this data. The first is an online GSI—ESA—NASA database (Luoni et al., 2021) that contains 1786 data points from 110 peer-reviewed publications. The second is NUCDAT (Norbury and Miller, 2012; Norbury et al., 2020), a data collection that contains 50,000 entries. The coverage of relevant ions, bombarding energies, and targets in these databases is, however, a small fraction of the data needed for effective modeling of nuclear reactions for space effects.

It should be noted that neither of these collections were compiled in connection with the major nuclear data organizations, the IAEA Nuclear Data Service (Forrest, 2011; IAEA, 2023) and the US National Nuclear Data Center (NNDC, 2023). The international standard reaction databases for nuclear reaction data, EXFOR (Otuka et al., 2014) (for compilations) and ENDF (Brown et al., 2018) (for evaluations), have limited applicability for space research as they are focused on neutron-induced reactions at energies less than 14 MeV. They do, however, contain some cross sections up to 200 MeV and a few charged-particle induced reactions. Finally, the OECD NEA Shielding Integral Benchmark Archive and Database (SINBAD) (Kodeli et al., 2014) has some integral cross section information that is useful, such as a set containing measurements of 100–800 MeV/u He, C, Ne, Ar, Fe, Xe, and Si ions bombarding C, Al, Cu, and Pb targets.

There are also a number of specialized databases within the space electronics effects community, including (in the US) collections at NASA Goddard Space Flight Center (GSFC) (NASA GSFC, 2023) and NASA Jet Propulsion Laboratory (JPL) (NASA JPL, 2023). A number of factors have, unfortunately, impeded the coordination between these different collections, including the different systems measured, proprietary data, security, formats, and funding to initiate and (especially) maintain a more unified data library.

The current disjoint collections of nuclear and electronic effects datasets for space research could benefit tremendously from the decades of experience in the nuclear data community in establishing, curating, combining, and disseminating datasets. By linking these space-related datasets together and creating new customized collections, the nuclear data community could greatly improve access to the existing data that are so critical for simulations in space research studies. This effort would also be invaluable for guiding future experimental work as gaps in the measurement data could be more easily accessed.

Because all the nuclear cross sections needed for space research will never be measured, nuclear reaction models are critical for simulations that transport GCR primaries through spacecraft materials and predict the flux and composition of secondaries. Results of such transport simulations are subsequently utilized in studies of electronic effects (Höeffgen et al., 2020), human effects (Rajaraman et al., 2018; Walsh et al., 2019), and spacecraft design (Fukunaga et al., 1997). Nuclear reaction models are also essential to predict the yields, and therefore determine the viability of, accelerator-based measurements of reactions important for space research.

The space research community has had some successes with phenomenological nuclear reaction models. One notable example is the Double Differential Fragmentation model (DDFRG) (Norbury, 2021) which has been fine-tuned to agree with measurements in the NUCDAT collection (Norbury and Miller, 2012; Norbury et al., 2020) mentioned above. Other models include NUClear FRaGmentation (NUCFRG) (Wilson et al., 1987), which uses an abrasion–ablation formalism (Hüfner et al., 1975), and the self-consistent Relativistic Abrasion–Ablation FRaGmentation (RAADFRG) code (Werneth et al., 2021). Semi-empirical parameterizations (e.g., Hybrid Kurotama (Sihver et al., 2014), Kox-Shen (Kox et al., 1987)) have also shown reasonable agreement with some datasets, and a number of such formulations have been collected and put online in the GSI-ESA-NASA database (Luoni et al., 2021) mentioned above.

Even though most of the terms in reactions models at energies ranging from 0.1–1 GeV/u (including pre-equilibrium, de-excitation, evaporation, intra-nuclear cascade, fission, and more) are reasonably well understood, more data are needed to fine tune these models. For example, De Napoli et al. (2014) shows that the predictions of two microscopic models—Quantum Molecular Dynamics (QMD) (Aichelin, 1991) and Binary Light Ions Cascade (BLIC) (Folger et al., 2004)—for ⁴He production via the ¹²C + ¹²C reaction at 62 MeV/u vary by a factor of 2. The phenomenological reaction parameterizations discussed in (Luoni et al., 2021) show that the variations in cross section predictions at 200 MeV/u can be as large as a factor of 2 for certain beam-target combinations (see Figure 15 in Luoni et al., 2021). For most systems the variations are, however, 25% or less. Figure 17 in Luoni et al. (2021) shows that variations of parameterization predictions at 10 GeV/u are typically 10%–25% when averaged over many models and systems. It is important to note, however, that these parameterizations are primarily benchmarked on data at a few hundred MeV/u, with only a few data points at 1 GeV/u, so these higher-energy extrapolations need measurements for validation (Luoni et al., 2021).

However, the state-of-the-art models of the highest energy reactions in the space community lag recent theoretical work being done in the RHIC community. While the former conceptualizes heavy ion reactions in terms of abrasion, ablation, and coalescence, the latter embraces models in which hadronic reaction products are constructed from quarks and gluons. A good example of such a RHIC-related model is the Ultrarelativistic Quantum Molecular Dynamics (UrQMD) model (Bleicher et al., 1999). This model has demonstrated the capability to predict the yields of protons and deuterons measured at the BNL Alternating Gradient Synchrotron (AGS) from the bombardment of Be and Au targets with 15 GeV protons (Sombun et al., 2019). The utilization of this, and other, RHIC codes for predicting the yield of secondaries resulting from GCR transport through spacecraft materials would, when combined with additional higher-energy data, significantly advance the simulations needed to make space exploration safer.

The focus of this study has been on the need for a better understanding of interactions of GCR particles of energies above 0.2 GeV/u with spacecraft materials, electronics, and astronauts. Reactions at energies below 200 MeV/u are, however, relevant for the lowest energy GCRs as well as the end of cascades induced by higher-energy GCRs. To develop a comprehensive treatment of nuclear reactions for all GCRs, some improvements are also needed for these low energy interactions. For example, De Napoli et al. (2014) shows disagreements of up to a factor of 5 between measurements of ⁴He production from ¹²C + ¹²C reaction at 62 A MeV and predictions from the popular QMD (Aichelin, 1991) and BLIC (Folger et al., 2004) models. Clearly more work is needed to improve the models at these energies, and to explain other effects that can contribute to light-ion production via fragmentation such as evaporation after incomplete fusion, deep-inelastic processes, and target fragmentation (De Napoli et al., 2014).

Some of the most critical end-user applications in space research include radiation transport simulations, electronics and human effects studies, and spacecraft design. Below we detail the current the state-of-the-art and relevant nuclear data gaps for each of these, with an emphasis on cross-disciplinary collaborative research efforts that can have a strong impact on space exploration.

To identify the most critical nuclear data gaps in space research, it is necessary to examine the quantitative variation in space model predictions that arise from variations in nuclear physics inputs. One approach is to utilize different nuclear model inputs in the same transport code. An example of this is shown in Figure 2 of Norbury et al. (2020), where QMD (Aichelin, 1991) and INCLXX (CEA, 2014) models, used as input to GEANT4 transport calculations, produce differences in light-ion production of up to a factor of 3 at energies below 10 GeV/u. Another approach, also shown in the same figure from Norbury et al. (2020), is to utilize different transport codes with different nuclear inputs; this produces light-ion production variations of up to a factor of 20 over the same energy range. These approaches can be used to quantify the overall contribution of cross section libraries, or that of the combination of cross sections and transport codes, to the uncertainty of the predictions of these models.

With such approaches, it is not possible to identify specific reactions giving the largest impact on the transport calculation predictions because entire libraries of input cross sections are changed. An alternate approach is a “sensitivity study” where only one reaction cross section is systematically changed in each calculation. While sensitivity studies are not commonly used in space research, one study has shown the importance of cross sections for p, n, and α particle production under bombardment by specific ions (e.g., O, Mg, Si, Fe) (Lin, 2007). In another study, the impact of cross section changes on dose equivalent was made as a function of shielding depth, see Lin and Adams (2007), using the HZETRN transport code. Changes of 25% in the input cross sections at 0.3 GeV/u were found in that study to produce 10% changes in the output dose equivalent predictions. An energy survey in the same study found that cross section changes in the 0.2–1.5 GeV/u energy range were the most impactful on the dose equivalent predictions. It is important to note that the 25% cross section variations used in Lin and Adams (2007) are representative of cross section uncertainties at 0.3 GeV/u (Luoni et al., 2021) but could underestimate the uncertainties at higher energies where there are few measurements.

The effects of GCRs on integrated circuits were estimated in Ref. Dodd et al. (2007) using ground-based tests at typical energies of 10 MeV/u. It was shown that the rate of single-event effects that occur in space at the higher energies (∼1 GeV/u) of GCRs may be significantly underestimated because higher energy GCRs generate secondary ions within the integrated circuits that penetrate more deeply into the circuits. The number of single-event effects per unit time can, in general, vary by up to five orders of magnitude depending on the radiation hardening of the circuit, the interaction cross sections, and other effects.

The decades of experience with sensitivity studies in the nuclear data community suggest that critical advances can be made in space science through collaborative, cross-disciplinary efforts. Some of the existing sensitivity tools in the nuclear data community include: TSUNAMI, TSAR, SEN3, and SAMPLER (ORNL) (Perfetti and Rearden, 2013); Whisper and Crater (LANL) (Kiedrowski et al., 2015); Nuclear Data Sensitivity Tool (NDaST) (OECD NEA) (Dyrda et al., 2017); SUSD3D (JAEA/NEA) (Kos et al., 2021); NUSS-RF (PSI) (Zhu et al., 2015); FICST (McMaster University) (Mostofian, 2014); RMC (Tsinghua University) (Wang et al., 2014); and GPT-free in OpenMC (MIT/Purdue/Virginia Commonwealth University) (Wu et al., 2019). While many of these codes were designed specifically to determine the sensitivities of input nuclear cross sections on nuclear reactor performance (such as the effective neutron multiplication factor k _eff) and for nuclear criticality safety studies, some are more general and could potentially be adopted to address sensitivities critical for space research.

Sensitivity studies can be used to translate the uncertainties on the model inputs into uncertainties on the model predictions. While uncertainty quantification (UQ) approaches in general have been a high priority in the nuclear data community, they are not widely adopted in the space research community. The combination of Bayesian approaches with ML tools have recently begun to set the standard for assigning uncertainties to model predictions in the nuclear science community (see, for example, Neufcourt et al., 2018). Collaborative efforts between the two communities could, for example, enable nuclear cross section uncertainties to be propagated through transport models to uncertainties in the secondary flux characteristics, and subsequently propagated through specialized codes to assess uncertainties in electronics and tissue damage.

Combining such UQ approaches with sensitivity studies can then guide future work in measurements, nuclear reaction theory, transport models, and stopping power codes. Currently under development, EUCLID (LANL) will be the first of a new generation of tools in the nuclear data community that uses this combined approach for reactors or critical systems. EUCLID is an extension of work using ML to improve nuclear data validation (Neudecker et al., 2020). By combining sensitivity studies, uncertainty quantification, nuclear dataset adjustment, and ML-driven design of experimental measurements, this tool will be able to identify which new measurements would provide the tightest constraints on predictions for an end-user application. A EUCLID-like tool for the space research community could be used to assign uncertainties to model predictions, identify critical data and theory gaps, and recommend approaches to best fill those gaps.