Considerations for practical dose equivalent assessment of space radiation and exposure risk reduction in deep space

Shielding from space radiation, especially galactic cosmic rays (GCRs), is a significant safety challenge for future human activities in deep space. In this study, the shielding performances of potential materials [aluminum (Al), polyethylene (PE), and carbon fiber reinforced plastic (CFRP)] were investigated using Geant4 Monte Carlo simulation considering two types of biological scale parameters, the International Commission on Radiological Protection (ICRP) quality factor (QFICRP) and the plausible biological effectiveness (RBEγacute), for GCRs. The effective dose equivalent was reduced by 50% for QFICRP and 38% for RBEγacute when shielding using 20 g/cm2 of CFRP. A spacecraft made from CFRP will have a better radiation shielding performance than conventional Al-based spacecraft. The contribution of heavy ions for QFICRP based effective dose equivalent was larger by a factor of ~ 3 compared to that for RBEγacute based effective dose equivalent. The shielding materials efficiently reduced the effective dose equivalent due to ions with QFICRP > 3.36 and RBEγacute > 2.26. QFICRP and RBEγacute have advantages and disadvantages in quantifying the dose equivalent of space radiation, and the establishment of a standard parameter specified for a mixed radiation environment occupied by protons and heavy ions is necessary for practical dose assessment in deep space.

www.nature.com/scientificreports/ The acute gamma-ray model has obtained consistent RBEs by using the linear response to epidemiological data of atomic bomb survivors 19,20 . Meanwhile, Cucinotta et al. 21 suggested that the RBE γacute might be insufficient to assess risk as well as QF due to restricted information on the physics of incident ions; e.g., Z * 2 /β 2 in (Eq. 3). After that, Cacao et al. 22 obtained the RBE γacute response to particle LET and charge number based on experimental datasets of 16 O, 20 Ne, 28 Si, 48 Ti, and 56 Fe exposure. Radiation shielding is a strategy used to reduce radiation exposure risks. Passive shielding is an approach to absorb relatively low energy particles and break up HZE particles into lighter particles in the shielding material, resulting in dose reduction. HZE particle exposure and dose reduction by shielding materials have been studied through calculations and measurements using ground accelerators e.g., Ref. [23][24][25][26][27][28][29][30][31] and in space modules e.g., Refs. 6,[32][33][34][35][36][37][38][39] . Several types of shielding materials, such as aluminum (Al), polyethylene (PE, (C 2 H 4 ) n ), and carbon fiber reinforced plastic (CFRP), have been studied. Al is commonly used in spacecraft construction, PE is known to be an effective shielding material, and CFRP is a potential structural material with a relatively high shielding efficiency for spacecraft owing to its high mechanical strength 40,41 .
In this study, we discuss the effective dose equivalents due to GCR exposure based on two types of biological scale parameters under passive shielding, which will be crucial for evaluating radiation-induced risks during upcoming long-term stays in space. Figure 1 shows the energy spectra of the major GCR elements in free space and behind the shielding materials: Al, PE, and CFRP. Error bars represent the statistical error of the calculation. Several space modules provide shielding of ~ 20 g/cm 2 or more mass thickness on average 42,43 , therefore, we employed 20 g/cm 2 as the typical thickness of spacecraft shielding. The proton flux was increased by shielding because the target and heavier projectile fragmentation reactions produced numerous secondary protons along the pathway of the primary particles in the materials. The number of heavy ions was decreased by the projectile fragmentation reactions, while some light ions at low energies (< 10 MeV/n) were increased by particle energy loss and secondary particle production owing to the target and heavier projectile fragmentations. Secondary particle production in the high energy region was highest in PE by projectile fragmentation, whereas that in the low energy region was highest in Al by target fragmentation. This is explained by the fragmentation cross section per unit mass 41 and the particle production rate by fragmentation. PE has the largest cross section of the employed materials, followed by CFRP and Al. The fragmentation of heavier particles produces a larger number of particles e.g., Ref. 44 .

GCR fluences behind shielding materials.
The energy spectra of the fluence were converted to LET-dependent fluence spectra in the ICRU four-element tissues 45 , as shown in Fig. 2. Summed LET spectra were obtained and are shown in Fig. 3. LET peaks for He, C, O, and Fe appeared at approximately 0.9, 7.5, 15, and 150 keV/µm, respectively, in free space. The mean QF ICRP and RBE γacute were obtained from these fluence spectra using Eqs. (2) and (3), respectively. Note that our results only include the contributions by the charged particles from protons to Fe ions; the contributions of neutrons, photons, pions, and muons were not considered. The shielding materials significantly reduced the flux of primary particles heavier than He owing to projectile fragmentation. The flux reduction rate increased with www.nature.com/scientificreports/ heavier nuclei; this was observed for Fe, which is one of the major contributors to the total dose. The shielding of high LET particles should contribute to the efficient reduction of biological effects. The increase in low energy H appeared as an enhancement at LET = 1-30 keV/µm. This was mainly because of the target fragmentations. Similar enhancements appeared in the heavier particles: He (30-100 keV/µm), C (500-1000 keV/µm), and Ne (1000-1500 keV/µm) (Fig. 2).
Biological scale parameters. The absorbed dose rates and mean QF ICRP and RBE γacute of GCR elements in free space are summarized in Table 1. The total mean QF ICRP and RBE γacute values were obtained as follows:  www.nature.com/scientificreports/ where subscript i is an element in the GCR particles (Z = 1-26). The total absorbed dose rate was 162 mGy/year and the QF ICRP based effective dose equivalent (H E(ICRP) ) was 523 mSv/year. The total mean QF ICRP was larger than the total mean RBE γacute by a factor of ~ 2.5. It should be noted that the effective dose equivalents based on QF ICRP could not be overestimated by a factor of ~ 2.5. QF ICRP is defined based on radiobiological data for conditions of low dose and low dose rate gamma-rays 15 . The difference due to the reference gamma-ray is compensated by the dose and dose rate effectiveness factor (DDREF) 15 . The value of the DDREF is a critical factor for obtaining the absolute value of the dose equivalent. The ICRP recommends a DDREF of 2 8,15 . However, the DDREF ranges from 2 to 5 depending on the targets and radiation quality e.g., Refs. [46][47][48][49][50][51] . The differences between QF ICRP levels and RBE γacute (Table 1) were consistent with the DDREF range. Figure 4 shows elemental contributions to the effective dose equivalents in free space. The contributions of H, He, C, N, and O to H E(RBEγacute) were larger than their contributions to H E(ICRP) . The contribution of Fe to H E(RBEγacute) was smaller than that to H E(ICRP) by a factor of ~ 2.5. The LET dependences of effective dose equivalents in free space and with 20 g/cm 2 Al, PE, and CFRP shielding are given in Fig. 5. The vertical axis of Fig. 5b, i.e., H E(RBEγacute) , does not consider the DDREF value. Therefore, direct comparisons between (a) and (b) cannot be made. The reduction rates of H E(ICRP) and H E(RBEγacute) were similar for the whole LET range. While the enhancement of the low energy and high LET particles generated in the shielding materials made a small variation in flux for LET < 100 keV/µm, the fragmentations from heavier ions than C contributed much to a dose reduction for LET > 8 keV/um. The mean QF ICRP and RBE γacute (Table 1) indicate that the shielding material is efficient at reducing the effective dose equivalent due to the high LET particles of QF ICRP > 3.36 or RBE γacute > 2.26. The dose reduction by CFRP was by a factor of ~ 2 at 10 keV/µm, ~ 5 at 100 keV/µm, and ~ 25 at 1000 keV/µm. The variations of relative absorbed dose, mean QF ICRP and mean RBE γacute for the different shielding materials are summarized in Table 2.

Discussion
The calculated total absorbed dose and H E(ICRP) in this study were 10-15% and ~ 20% lower than previous interplanetary measurements, respectively 6,39 . Here, we note that H E(ICRP) was compared with the measured dose equivalents. The effective dose equivalent, which is a protection value, is determined by normalizing dose equivalent of human body, which is an operational value, with tissue or organ weighting factors. Although the effective dose equivalent is not equal to the dose equivalent, their comparison is reasonable considering the differences in irradiation targets (human body for our calculations and detector mediums for measurements). Considering discrepancies between the simple calculation geometry and actual measurement configuration, measurement uncertainties, the differences in the solar modulation factor, and contribution from secondary particles, which were not included in our calculation, these results are almost consistent. The reduction rates of H E(ICRP) by material shielding in this study were 45-55% at 20 g/cm 2 , which are also similar values to previous calculations considering the above discrepancies 38,52 . There was an important difference between RBE γacute and QF ICRP in the ratio of light and heavy ions (e.g., H vs. Fe): the QF ICRP value of Fe was 11-12 times higher than that of H, compared to 3.5 times higher for RBE γacute . This difference implies that the contribution of heavy ions to H E(ICRP) is ~ 3 times higher than that to H E(RBEγacute) . Comparing Figs. 3 and 5 indicates that the low energy edge of H does not contribute to the H E spectra and a small enhancement due to Mg and Si makes a contribution to LET of ~ 30 keV/µm. Because the kinetic energy of the secondary H around the edge is ~ 1 MeV, its energy deposition is not large enough to increase the dose. Meanwhile, heavy ions, such as C and O, for LET ~ 10 keV/µm are relativistic, which increases the dose. The penetrative ability of ions is an important factor when considering the energy deposition in the human body.  The reduction in the absorbed dose mainly came from the fragmentation of HZE particles. Around the mean energy of the GCR particles (~ 1 GeV/n), the energy loss in the shielding materials, which depended on the stopping power, was not significant. The mean QF ICRP and RBE γacute values decreased as a function of shielding thickness. The reduction rate of HZE particles by fragmentation in the shielding materials was larger than the rate of increase in protons. At a typical mass thickness of 20 g/cm 2 , the mean QF ICRP and RBE γacute decreased to 30% and 15% for Al, 40% and 20% for PE, and 37% and 18% for CFRP, respectively. The reduction rates of mean RBE γacute were lower than those of mean QF ICRP because the heavy ion contributions to mean QF ICRP were larger than those to mean RBE γacute . The relative variations in H E values among the shielding materials were obtained from the relative absorbed doses and relative mean QF ICRP or RBE γacute (Eq. 3). At a thickness of 20 g/cm 2 , the reduction rates of H E(ICRP) and H E(RBEγacute) were 45% and 32% for Al, 55% and 40% for PE, and 52% and 38% for CFRP, respectively. PE achieved ~ 24% and ~ 5% higher reduction rates than Al and CFRP, respectively. An idea for deep space missions is to construct some spacecraft parts made of CFRP, which has a better shielding capability than conventional Al based materials. Our previous evaluation implied that the material switching from Al to CFRP at the same actual thickness provided a similar dose reduction despite that the total mass of CFRP module was much smaller than that of Al by a factor of their density ratio 41 . The actual shielding materials are not only spacecraft materials but also fuel, water and other supplies. The materials to construct a spacecraft will be selected by not only radiation shielding performance but also many requirements such as thermal property, ultraviolet resistance, moisture absorption resistance and so on. If the complete material switching from Al to CFRP is attained by the same mass thickness (g/cm 2 ), CFRP will give a benefit for the protection of crews from GCRs.
The results were discussed for charged particles from protons to Fe ions as mentioned in the results section; the contributions of neutrons, photons, pions, and muons were not considered. In particular, the neutron contribution to the dose equivalents may not be small among secondary radiation particles (e.g., Refs. 4,5 ). The neutron contributions obtained from the ICRP conversion coefficients, H E(ICRP)neutron , are also listed in Table 2. The neutron contribution rate was much lower than the charged particle contributions (5.9% of the H E(ICRP) at most). The fragmentation reactions of the primary particles, which produce neutrons, occur effectively in light shielding materials. The higher neutron contribution in Al indicates a high thermalization with hydrogen atoms in PE and CFRP. RBE γacute for neutrons is one of issues to be addressed in future.
The difference between QF ICRP and RBE γacute , except for the reference gamma-ray, is in the models: QF ICRP depends on the energy deposition (i.e., LET) while RBE γacute depends on the charge and energy of the incident particles. The energy of the secondary particles is dependent on that of the primary particles. One possible explanation for the larger QF ICRP peak than RBE γacute peaks in their LET dependencies (Fig. 6) is the difference in targeted radiation for dose assessment. Heavy ions at their peaks have relativistic energy, which produces relativistic secondary particles. The fact that the RBE γacute peaks are lower than the QF ICRP peak reflects the significant contribution of low energy secondary particles because electrons are biologically effective at low energies (< 10 keV) 10 .
The advantages and disadvantages of QF ICRP and RBE γacute are summarized in Table 3. The RBE γacute evaluates relative biological effectiveness with a smaller uncertainty than RBE max relevant to QF ICRP , as mentioned in the introduction. One of the difficulties in using RBE γacute is the selection of the parameters. The selected parameters, and thus RBE γacute , depend on the targeted radiation field and biological effects. Thus, the obtained dose equivalent was not comparable in different radiation environments. Meanwhile, QF ICRP , which is determined by LET, offers good usability. However, the effective dose equivalents and QF ICRP have been established for general dose assessment on the ground and have been replaced with the effective dose 15 . The radiation environment on the Table 2. Variation of the relative effective dose equivalent, H E , with shielding materials of differing mass thickness. The relative neutron contribution rates (H E(ICRP)neutron ) and contributions to the charged particle contributions (H E(ICRP) ) are given. www.nature.com/scientificreports/ ground is primarily composed of photons from natural radioisotopes. Other contributors to this environment are protons, alpha particles, and neutrons in specific radiation fields, such as medical accelerators and nuclear power plants. HZE particles were not the main target of QF ICRP on the ground. QF ICRP has not been updated since the 1990 ICRP recommendation 15 . The establishment of a standard parameter specified for a mixed radiation environment occupied by protons and heavy ions is necessary for practical dose assessment in deep space.

Conclusion
We investigated the radiation shielding performance and effective dose equivalents based on the ICRP radiation QF and the plausible RBE for three shielding materials, Al, PE, and CFRP, using the Geant4 Monte Carlo simulation. The QF ICRP values of Fe were 11-12 times larger than those of H, compared to ~ 3.5 times larger for RBE γacute . Therefore, the contribution of heavy ions to H E(ICRP) was larger by a factor of ~ 3 compared to that to H E(RBEγacute) . The shielding materials reduced the flux of primary particles heavier than He due to projectile fragmentation. The flux reduction rate increased with successively heavier nuclei and the increase was particularly large for heavier ions. The shielding materials efficiently reduced the effective dose equivalent due to ions with QF ICRP > 3.36 and RBE γacute > 2.26. The reduction rates of H E(ICRP) were higher than those of H E(RBEγacute) because of the large contribution of heavy ions. The expected radiation exposure risk was reduced by 50% for QF ICRP and 38% for RBE γacute when using 20 g/cm 2 CFRP. Therefore, a spacecraft made of CFRP could improve the radiation shielding performance compared to conventional Al-based spacecraft and help mitigate space radiation hazards in future space missions. The discrepancy between QF ICRP and RBE γacute highlighted the necessity of a new standard for mixed radiation environments occupied with protons and heavy ions in deep space.

Methods
We calculated the fluences of GCR protons and heavy ions up to Fe (Z = 1-26) with energies ranging from 1 MeV/n to 100 GeV/n in free space and with shielding materials using Monte Carlo simulation with Geant4 ver. 10.04.02 [53][54][55] . The primary particles in the energy region of 1-10 MeV/n ( Fig. 1 dotted lines) did not contribute to the flux behind the materials because these low energy particles were stopped within a thin shielding layer of only a few micrometers. Primary particles in this energy range were excluded from the projectile to reduce the calculation cost. The mean QF ICRP and RBE γacute values were obtained, including the particles in this energy range. Al, PE, and CFRP with mass thicknesses of 5, 10, 15, and 20 g/cm 2 were employed as shielding materials. The GCR source was derived from the DLR model during the solar minimum phase 56 . The solar minimum assumption provides the worst case for radiation exposure. The number of each primary ion was fixed at 10 6 to obtain the total dose rate by merging all the dose rates. The elemental composition of the CFRP was assumed to be that of a commercial composite material, CF/PEEK (Toray Cetex TC1200, Toray Advanced Composites, Figure 6. LET dependencies of biological scale parameters. RBE γacute (left axis) for simple exchanges to human peripheral blood lymphocytes 22 and QF ICRP (right axis) 15 . www.nature.com/scientificreports/ USA). The dose and LET changes due to nuclear fragmentation simulated by Geant4 have been experimentally validated in our previous studies 40,41 . The absorbed dose (D) for the whole human body, due to each particle in free space and behind the target materials, was obtained from the particle fluences and ICRP conversion coefficients for isotropic exposure 9 . The ICRP defines the QF ICRP as the following 15 : The mean RBE γacute was obtained through the charge and energy dependences for the targeted effects model as follows 21,22 : where Z * and β are the effective charge number of the particle and particle velocity relative to light, respectively; parameters σ 0 , m, and κ are constants based on radiobiological experiments; and α γ is the linear regression coefficient for the acute dose of gamma-rays at the same endpoint. We employed these parameters for simple exchanges with human peripheral blood lymphocytes from Cacao et al. 22 since blood is uniformly distributed in the whole body. Here, the dose rate in space is of the order of hundreds of micro grays per day e.g., Refs. 6,33,34,37,43,57 . Nontargeted effects are expected to be negligible at sufficiently low doses of < 1 mGy e.g., Refs. 58-61 and a large number of DNA double-strand breaks are required for complex exchanges. Therefore, we selected a targeted effects model and assumed a simple exchange with human lymphocytes. Although the parameters studied in previous works 17,21 provided 2-5 times larger RBE γacute values, these studies targeted solid cancer or leukemia, which are also relevant to nontargeted effects and complex exchanges. Other parameters must be used to target these critical radiation hazards. The RBE γacute and QF ICRP values of major GCR particles as a function of LET are presented in Fig. 6 15,22 . The LET range of each particle corresponded to an energy range of 1 MeV/n-10 GeV/n. The dose equivalent H E was obtained by the products of the absorbed dose and the QF ICRP or RBE γacute : where i indicates the GCR particle (Z = 1-26).