betaFeedback

WorksPhysics and Astronomy

Interpretation of AMS-02 beryllium isotope fluxes using data-driven production cross sections

Abstract

The Be isotopic measurements preliminarily reported by the AMS-02 Collaboration have reached an unprecedented energy of 12 GeV/nn. As secondary cosmic rays (CRs), the Be isotopes include both stable and unstable species, which are crucial for constraining the propagation parameters of Galactic CRs. However, uncertainties in their production cross sections can skew the interpretation of the CR data, especially when cross-section measurements are of significantly lower quality than CR measurements. In this work, we consider the uncertainties of the cross sections to interpret the Be isotopic data by adopting a cross-section parametrization that fully utilizes the available experimental data. Owing to the high-quality measurements of the 7^7Be production cross section, we innovatively employ 7^7Be instead of 9^9Be to constrain propagation parameters. Notably, the diffusion halo thickness is constrained to 5.67±0.765.67\pm0.76~kpc, representing a moderate value compared to previous analogous works. Combining the well-constrained CR propagation model and the precise CR measurements of 9^9Be, we conversely constrain the major production cross section of 9^9Be and find that it ought to be remarkably lower than previously thought. Our analysis also questions the reliability of certain cross sections measured by some experiments, potentially marking the first time CR data has been used to identify dubious nucleon production cross sections. The method presented in this work holds promise for analyzing upcoming isotopic data from other nuclei.

INTRODUCTION

The study of Galactic cosmic rays (CRs) has advanced into an era of precision, owing to the improvements in experiment instruments and analysis methods. Exemplary experiments such as PAMELA [1] and AMS-02 [2, 3, 4, 5, 6] have achieved remarkable precision levels, with uncertainties now reduced to an impressive 1–3%. These collaborations are collecting large numbers of CR events to explore more spectrum features that have not been discovered yet, which could update our knowledge on the CR acceleration and propagation mechanism, the spatial distribution of the CR sources and the interstellar medium (ISM), and the inelastic and production cross sections of nuclei reactions. The calculated background information can conversely provide reliable identification of signatures of the dark matter or new physics.

The measurement of secondary CR species is pivotal in constraining the propagation characteristics of Galactic CRs [7]. Secondary CRs arise from the fragmentation of heavier nuclei upon collision with the ISM gas. Consequently, the flux ratio of secondaries to their primary nuclei (e.g., B/C) informs us about the grammage of CRs, which represent the integrated gas density along their path before they escape from the Galaxy. Additionally, the ratio of unstable-to-stable secondary isotopes (e.g., 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be}) provides insight into the residence time of CRs within the Galaxy. By synthesizing these two ratio types, we can independently determine the average diffusion coefficient for Galactic CRs and the height of the diffusion halo, which are the essential propagation parameters [8, 9].

Another crucial factor affecting the flux of secondary CRs is their production cross section. Regrettably, the precision and energy range of current cross-section measurements falls short of matching the secondary CR data obtained from space experiments. Most of the cross-section data were measured a long time ago, and the uncertainties for important reactions range from 10% to 20% [10, 11], much larger than those found in high-precision CR datasets. Besides, for some astrophysically significant reaction channels, cross-section measurements are limited to low energies or are entirely lacking. In these cases, semi-empirical parametrizations have to be employed to approximate the cross sections [12, 13, 14]. The determination of cross sections is further complicated by the intricate network of multi-step and ghost-nuclei reactions. Nonetheless, the analysis can be simplified by ranking the cross-section production contribution, focusing primarily on the predominant channels [10, 11]. To sum up, the uncertainty in production cross section is an essential consideration when using secondary species measurements to investigate CR propagation.

Li, Be, and B have the highest abundance among secondary CRs and have the most precise measurements [5]. Working under the assumption that these species are purely secondary in origin, the AMS-02 measurements can be understood within the conventional framework of CR propagation [15, 16, 17, 18, 19, 20, 21, 22]. However, to account for certain anomalies, such as the overprediction of Be at low energies, minor adjustments to the production cross-sections within their uncertainties have been proposed [19, 23]. On the other hand, by making use of the preliminary measurement of Be isotopes reported by the AMS-02 Collaboration [24, 25], Refs. [26, 27] argued that nontrivial CR propagation models might be necessary to account for the latest 10Be/9Be^{10}\mathrm{Be}/^{9}\mathrm{Be}, such as a two-zone diffusion model on a Galactic scale [28, 29]. It is evident that when experimental measurements of secondary CRs diverge from model predictions, disentangling whether this inconsistency arises from propagation factors or cross-section factors becomes a challenging problem.

In this study, we conduct the first individual analysis of the Be isotopes (7Be^{7}\mathrm{Be}, 9Be^{9}\mathrm{Be}, and 10Be^{10}\mathrm{Be}) detected by AMS-02 [24, 25], moving beyond the traditional approach of solely interpreting the 10Be/9Be^{10}\mathrm{Be}/^{9}\mathrm{Be} ratio. 7Be^{7}\mathrm{Be} (in the interstellar environment) and 9Be^{9}\mathrm{Be} can be considered stable isotopes. Their propagation processes can be described by the same framework, which does not exhibit significant variability in fragmentation timescales or diffusion coefficients due to differences in atomic mass numbers [30, 31]. Therefore, should there be any discrepancies in the effectiveness of a propagation model in explaining the behavior of 7Be^{7}\mathrm{Be} and 9Be^{9}\mathrm{Be}, these must arise from uncertainties in the production cross sections, thus providing a good opportunity to disentangle propagation factors from production cross-section factors. Intriguingly, the cross-section measurements for 7Be^{7}\mathrm{Be} are of significantly superior quality compared to those for 9Be^{9}\mathrm{Be}. Therefore, we can initially use the CR energy spectrum of 7Be^{7}\mathrm{Be} to constrain the propagation parameters. Then, using the derived propagation parameters, we can inversely constrain the production cross section of 9Be^{9}\mathrm{Be} by interpreting its precise CR energy spectrum. Moreover, since the cross-section measurements for 7Be^{7}\mathrm{Be} are of higher quality, we can break the degeneracy between the diffusion coefficient and the thickness of the diffusion halo by interpreting the CR measurements of 7Be^{7}\mathrm{Be} and 10Be^{10}\mathrm{Be}, superseding the previous method that used the 10Be/9Be^{10}\mathrm{Be}/^{9}\mathrm{Be} ratio. In the process of constraining the propagation parameters, we take into account the uncertainties of the production cross sections using existing cross-section data, also adding some updated data that offer enhanced precision and extend to higher energies (e.g., [32, 33]).

This paper is organized as follows. In Sec. II, we introduce the CR propagation model and the cross-section setup for analysis. As a preliminary step, we estimate the Gaussian-distributed normalization for each reaction channel by fitting the cross-section data. In Sec. III, we perform a combined fit to the CR data and the data-driven normalizations of the cross sections obtained earlier and discuss the results of the combined fit. CR data of 7Be^{7}\mathrm{Be}, 10Be^{10}\mathrm{Be}, B, C, and O are included in the fit. Same as 7Be^{7}\mathrm{Be}, B is also an important stable secondary species, and we will show that its data can be explained within the same propagation model. The inclusion of B data thus enhances the constraints on propagation parameters. C and O, the dominant primary particles for these secondary CRs, must also be included in the fitting process. In Sec. IV, we compare the predicted 9Be^{9}\mathrm{Be} and Be/B with the measured data to constrain the cross section of specific channels. Since the measurement of the Be/B ratio has been extended to 1 TeV, which is well above the energy range of measurements for Be isotopes, the inclusion of this data allows us to better distinguish between different high-energy extrapolation forms of the production cross sections. Finally, Sec. V is the summary of our findings above. For improved readability, we have moved several figures and discussions regarding the updated cross-section data and parametrizations used in this study to Appendix A.

CALCULATION SETUP

Propagation

We adopt the standard CR propagation model with reacceleration, which is frequently used in CR analysis [34, 35, 36]. Generally, the propagation equation of Galactic CRs is expressed as

∂ψ∂t=q(x,p)+∇⋅(Dxx∇ψ−Vcψ)+∂∂p[p2Dpp∂∂p(ψp2)]−∂∂p[p˙ψ−p3(∇⋅Vc)ψ]−ψτf−ψτr,(1)\frac{\partial\psi}{\partial t}=q(x,p)+\nabla\cdot\left(D_{xx}\nabla\psi-\mathbf{V}_{c}\psi\right)+\frac{\partial}{\partial p}\left[p^{2}D_{pp}\frac{\partial}{\partial p}\left(\frac{\psi}{p^{2}}\right)\right]-\frac{\partial}{\partial p}\left[\dot{p}\psi-\frac{p}{3}\left(\nabla\cdot\mathbf{V}_{c}\right)\psi\right]-\frac{\psi}{\tau_{f}}-\frac{\psi}{\tau_{r}}, \tag*{(1)}

where ψ\psi is the density per unit of particle momentum, q(x,p)q(x,p) is the source distribution, DxxD_{xx} is the spatial diffusion coefficient, VcV_c is the convection velocity1, DppD_{pp} is the momentum space diffusion coefficient, p˙≡dp/dt\dot{p}\equiv dp/dt is the ionization and Coulomb losses terms, τf\tau_f is the time scales for particle fragmentation, and τr\tau_r is the time scales for radioactive decay.

The scattering of CR particles on randomly moving magnetohydrodynamics waves leads to stochastic acceleration, which is described in the transport equation as diffusion in momentum space DppD_{pp}. Considering the scenario where the CRs are reaccelerated by colliding with the interstellar random weak hydrodynamic waves, the relation between the spatial diffusion coefficient DxxD_{xx} and the momentum diffusion coefficient DppD_{pp} is expressed as [37]:

DxxDpp=4p2Va23δ(4−δ)(4−δ2)ω.(2)D_{xx}D_{pp}=\frac{4p^{2}V_{a}^{2}}{3\delta(4-\delta)(4-\delta^{2})\omega}. \tag*{(2)}

To solve the propagation equation, we adopt the numerical GALPROP v562 framework [8, 38]. The code of GALPROP includes a nuclear reaction network to calculate the 1-step and multi-step fragmentation of the particles that collide on the interstellar medium gas. The reaction network is a series of repeated procedures, starting at solving the propagation equation of the heaviest nuclei 2864Ni{}^{64}_{28}\mathrm{Ni}, computing all the resulting secondary sources, and then proceeding to the nuclei with A−1A-1, where AA is the mass number of nuclei. This network makes use of the available cross-section measurements and also parametrizations, which are significant for predicting the cross sections of the data-lacking channels. In this work, we have made several updates to the nuclear data and adjusted the parametrizations of some dominant channels based on recent observations. See Appendix § 1.1.1 for more details.

The high-energy spectral hardening has been discovered by several observations [1, 2, 3, 4, 39, 40, 41, 42, 43, 44], but its origin is still under debate [29, 45, 46, 47, 48]. For simplicity, the diffusion coefficient DxxD_{xx} is defined as a broken power-law with a harder index δh\delta_h above the high-energy break rigidity RhR_h. We performed a prior fitting for the B/C and B/O ratios measured by AMS-02 and DAMPE [2, 49] to pin down the break rigidity and the slope index change, which are Rh=280 GVR_h = 280\ \mathrm{GV} and δh=δ−0.226\delta_h = \delta- 0.226. This can reduce the parameters in the following fitting process.

To describe the common bump structure of nuclei fluxes at low rigidities [30, 50], we model the injection spectrum of primary nuclei as a broken power law, with slope indices ν0\nu_0 and ν1\nu_1 below and above the low-energy break rigidity RbrR_{br}:

q(R)={(R/Rbr)ν0,R<Rbr,(R/Rbr)ν1,R≥Rbr.(3)q(R)=\begin{cases} (R/R_{br})^{\nu_0}, & R<R_{br},\\ (R/R_{br})^{\nu_1}, & R\geq R_{br}. \end{cases} \tag*{(3)}

We assume that the carbon, nitrogen, and oxygen nuclei share the same injection parameters (RbrR_{br}, ν0\nu_0 and ν1\nu_1) as suggested by the observation of AMS-02 [4, 5]. The source abundance ACA_C and AOA_O are also free parameters. Other individual isotopic source abundances and slope indices have been adjusted and fixed to match the corresponding primary fluxes [3, 4, 5, 6, 51, 52].

The B/C ratio is constrained by the AMS-02 measurements, and the degeneracy of L/DxxL/D_{xx} is kept as well. To break the degeneracy, the 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} ratio is often adopted. The radioactive decay of unstable 10Be{}^{10}\mathrm{Be} is related to the decay timescale and is sensitive to the Galactic halo height. If the halo size increases, the 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} and 10Be/7Be{}^{10}\mathrm{Be}/{}^{7}\mathrm{Be} ratio become smaller. Note that the measurements of 9Be{}^{9}\mathrm{Be} flux and 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} ratio are not used in the fitting. As shown in Appendix § 1.1.1, the uncertainty of 9Be{}^{9}\mathrm{Be} production is much larger than that of 7Be{}^{7}\mathrm{Be} and 10Be{}^{10}\mathrm{Be}, owing to less observation available. Prior fitting implies that the default cross-section parametrization predicts a systematically higher 9Be{}^{9}\mathrm{Be} flux relative to the CR measurements of AMS-02. This anomaly will be further analyzed and discussed in Sec. IV. The cross-section uncertainty of 7Be{}^{7}\mathrm{Be} is smaller, providing better constraining quality than that of 9Be{}^{9}\mathrm{Be}. Therefore, in the work, we constrain the Galactic halo height by 7Be{}^{7}\mathrm{Be} and 10Be{}^{10}\mathrm{Be} instead.

In Table I, we list all the CR measurements used for constraining the parameters in the work. The precise measurements of C, O, and B fluxes from AMS-02 [53] are used, and other data are included for better parameter constraints. For the C and O fluxes, we also include the CALET [41], NUCLEON [54], and CREAM-II [55] measurements to cover the multi-TeV energy region. The isotope 7Be{}^{7}\mathrm{Be} and 10Be{}^{10}\mathrm{Be} fluxes are constrained by the recently reported preliminary AMS-02 data [25].

To consider the solar modulation effect on the spectrum inside the heliosphere, we adopt the force-field approximation [57]. The strength is described by the solar modulation potential ϕ\phi. The CR measurements used were taken during a similar period (B, C, and O taken during May 2011-May 2021 [53] and beryllium isotopes taken during May 2011-nearly 2022 [25]), hence we assume that they share almost the same ϕ\phi. To estimate the modulation potential, the local interstellar spectra are also needed for determining the unmodulated (ϕ=0\phi=0) fluxes. Hence, we include the Voyager 1 measurements of B, C, and O fluxes [56] in the fitting.

Cross Section

The cross-section uncertainty has been regarded as a subdominant or even negligible factor in CR research for a long time due to the larger uncertainties of CR fluxes. However, as CR measurements become increasingly precise, the uncertainty in the production cross sections of secondary CRs has emerged as a critical factor that must be considered when constraining propagation parameters [15, 58]. Acknowledging its impact, we introduce additional parameters to adjust the cross sections of secondary nuclei during the fitting process.

To begin, it is essential to quantify the uncertainties associated with each production channel. Here we follow a data analysis routine introduced in our previous work [31], which was used to analyze the cross-section uncertainty of the F and B production. Through the implementation of a data-driven least-squares fitting approach, we can get the Gaussian-distributed dispersion values for the renormalization consistent with the selected dataset.

The default parametrization developed in GALPROP code is [GAL12] ([GAL22]), partly based on their fits of a compilation of cross-section measurements and code evaluations, and partly based on the Webber’s [12] (or Silberberg’s [13]) parametrization model with semi-empirical formulas. For isotope Be and B production cross sections, the parametrization follows a direct fit to the available data. The plateaus3 is mostly decided by the highest-energy measurement in the data sheet named eval_iso_cs.dat, which has been updated in Appendix § 1.1.1 by adding more observations. General features

ExperimentEnergy Rangedata pointsReference
C
NUCLEON(7/2015-6/2017)250-17000 GeV/n10[54]
CREAM-II(12/2005-1/2006)85-7500 GeV/n9[55]
CALET(10/2015-10/2019)×1.27\times1.27a10-1700 GeV/n22[41]
AMS-02(5/2011-5/2021)2-2000 GV66[53]
Voyager 1-HET(2012-2015)0.02-0.13 GeV/n8[56]
O
NUCLEON(7/2015-6/2017)300-13000 GeV/n9[54]
CREAM-II(12/2005-1/2006)64-7500 GeV/n9[55]
CALET(10/2015-10/2019)×1.27\times1.2710-1700 GeV/n22[41]
AMS-02(5/2011-5/2021)2-2000 GV66[53]
Voyager 1-HET(2012-2015)0.02-0.15 GeV/n10[56]
B
AMS-02(5/2011-5/2021)2-2000 GV66[53]
Voyager 1-HET(2012-2015)0.02-0.11 GeV/n8[56]
7Be,10Be{}^{7}\mathrm{Be},{}^{10}\mathrm{Be}
AMS-02(preliminary)0.7-11 GeV/n26[25]

Table 1. Data Used in This Analysis.

of nucleon production have been discovered by analyzing available measurements, for example, the cross section becomes constant at above 2 GeV/n [59]. However, there is an energy-dependent rise observed in the total and inelastic nucleon-nucleon cross section [60, 61]. As implied from recent analysis [62, 63], a slow change of the cross section can still appear above 2 GeV/n, making the determination of the plateau according to the highest-energy measurement less reliable. We notice that these production channels lack of high-energy (>10> 10 GeV/n) measurements, which means that the prediction of secondary nuclei relies seriously on the extrapolation of the mid-energy data measured by multiple experiments. Therefore, the uncertainty in the extrapolation region should not be strictly constrained as well as at low energies. Data points above 2 GeV/n should be important for analyzing and constraining the plateau, thus we choose to use high-energy data above that energy to determine the uncertainty in the extrapolation region.

We have made several least-squares fits to the cross-section data with the parametrization formulae given by the GALPROP code (see Appendix § 1.1.1), to determine the Gaussian-distributed dispersion as an estimate of the uncertainty. In Table II we list the fitting result of cross-section uncertainties. These channels are the main production channels for B, 7Be{}^{7}\mathrm{Be} and 10Be{}^{10}\mathrm{Be}, including the contributions from ghost nuclei5 10C{}^{10}\mathrm{C} and 11C{}^{11}\mathrm{C}. The average uncertainty ω0\omega_{0} of the individual channel is determined by a global fitting for all of the available data, while the high-energy uncertainty ω1\omega_{1} is determined by exclusively fitting the data points at ≥2\ge2 GeV/n. We notice that the ghost nuclei 10C{}^{10}\mathrm{C} contributes less than 5% of the total

Channelω0\omega_{0}ω1\omega_{1}
12C⟶10B{}^{12}\mathrm{C}\longrightarrow{}^{10}\mathrm{B}0.0050.113
12C⟶11B{}^{12}\mathrm{C}\longrightarrow{}^{11}\mathrm{B}0.0090.066
12C⟶10C⟶10B{}^{12}\mathrm{C}\longrightarrow{}^{10}\mathrm{C}\longrightarrow{}^{10}\mathrm{B}0.0120.081
12C⟶11C⟶11B{}^{12}\mathrm{C}\longrightarrow{}^{11}\mathrm{C}\longrightarrow{}^{11}\mathrm{B}0.0030.011
16O⟶10B{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{B}0.0240.038
16O⟶11B{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{B}0.0220.033
16O⟶10C⟶10B{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{C}\longrightarrow{}^{10}\mathrm{B}0.0510.111
16O⟶11C⟶11B{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{C}\longrightarrow{}^{11}\mathrm{B}0.0020.063
12C⟶7Be{}^{12}\mathrm{C}\longrightarrow{}^{7}\mathrm{Be}0.0040.020
12C⟶10Be{}^{12}\mathrm{C}\longrightarrow{}^{10}\mathrm{Be}0.0090.084
16O⟶7Be{}^{16}\mathrm{O}\longrightarrow{}^{7}\mathrm{Be}0.0070.021
16O⟶10Be{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{Be}0.0130.079

Table 2. The average and high-energy uncertainties of different cross-section channels for the production of B and Be isotopes.

B flux at all energies, and its uncertainties are 1 order of magnitude smaller, which should not significantly impact the resulting flux (<0.5%< 0.5\%). Thus, we ignore the uncertainty contribution from 12C⟶10C⟶10B{}^{12}\mathrm{C}\longrightarrow{}^{10}\mathrm{C}\longrightarrow{}^{10}\mathrm{B} and 16O⟶10C⟶10B{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{C}\longrightarrow{}^{10}\mathrm{B} channels.

An energy-dependent modification rather than a constant re-scaling is better for describing the cross-section uncertainties as listed in Table II. Following Refs. [63, 64], we construct a two-part modification formula to renormalize the production cross section σ\sigma, which depends on the kinetic energy per nucleon Ekin/nE_{\mathrm{kin}}/n:

σ=σ0[1+μ1+(Eth/Ekin/n)2+μω0/ω11+(Ekin/n/Eth)2](4)\sigma=\sigma^{0}\left[1+\frac{\mu}{1+(E_{\mathrm{th}}/E_{\mathrm{kin}}/n)^{2}}+\frac{\mu\omega_{0}/\omega_{1}}{1+(E_{\mathrm{kin}}/n/E_{\mathrm{th}})^{2}}\right] \tag*{(4)}

where σ0\sigma^{0} is the cross section determined by a data-driven parametrization introduced in Appendix § 1.1.1, and μ\mu is the renormalization factor that will act as a free parameter in the CR data fit. The threshold energy EthE_{\mathrm{th}} is set to 2 GeV/nn, to allow for a slow change of the cross section above that energy [63, 64].

In the high-energy extrapolation region, Eq. (4) approaches the asymptotic value σ→σ0(1+μ)\sigma\to\sigma^{0}(1+\mu). When the energy becomes lower, the asymptotic value tends to be smaller as σ→σ0(1+μω0/ω1)\sigma\to\sigma^{0}(1+\mu\omega_{0}/\omega_{1}). Thus, the factor μ\mu represents the renormalization above 2 GeV/nn. When μ\mu approaches ω1\omega_{1} (−ω1-\omega_{1}), the low-energy renormalization acts as narrower constraints, approaching ω0\omega_{0} (−ω0-\omega_{0}). As an example, we show the 68% confidence interval of the parametrization renormalization for the C + p ⟶10\longrightarrow{}^{10}B channel in Figure 1. Above 2 GeV/nn, the renormalization range approaches [−ω1,ω1][-\omega_{1},\omega_{1}].

Cross section for the C + p $\longrightarrow {}^{10}$B channel

Figure 1. Channels: C + p ⟶10\longrightarrow{}^{10}B. The blue line is the cross section σ0\sigma_{0} obtained by the data-driven parametrization. The blue band is the 1σ1\sigma confidence interval allowed by the cross-section measurements, the definition of which is referred to Eq. (4).

The individual contributions from N, Ne, Mg, Si, and Fe to the concerned secondaries cannot be ignored, as the cumulative contributions are up to 20% of the total fluxes. However, the cross-section measurements of these subdominant channels are relatively scarce, and the uncertainties are much larger than those of the dominant channels listed in Table II. To simply consider the uncertainties among subdominant channels, we renormalize the cross section of them by using Eq. (4), while the renormalization factor μ\mu is decided by the average renormalization of C and O as (μC+μO)/2(\mu_{\mathrm{C}}+\mu_{\mathrm{O}})/2 in the following fitting process.

FITTING RESULTS

For the goodness-of-fit of the model to the data, we use the least-χ2\chi^{2} method. The χ2\chi^{2} statistic is expressed as

χ2=∑qχcr,q2+χcs2(5)\chi^{2}=\sum_{q}\chi_{\mathrm{cr},q}^{2}+\chi_{\mathrm{cs}}^{2} \tag*{(5)}
χcr2=∑i=1nbin(yidata−yimodelσidata)2(6)\chi_{\mathrm{cr}}^{2}=\sum_{i=1}^{n_{\mathrm{bin}}}\left(\frac{y_{i}^{\mathrm{data}}-y_{i}^{\mathrm{model}}}{\sigma_{i}^{\mathrm{data}}}\right)^{2} \tag*{(6)}
χcs2=∑i=1ncs(μiω1i)2(7)\chi_{\mathrm{cs}}^{2}=\sum_{i=1}^{n_{\mathrm{cs}}}\left(\frac{\mu_{i}}{\omega_{1i}}\right)^{2} \tag*{(7)}

where yimodely_{i}^{\mathrm{model}} is the CR flux predicted by the model, and yidatay_{i}^{\mathrm{data}} and σidata\sigma_{i}^{\mathrm{data}} are the flux and error of the CR measurements, respectively. In Eq. (5), qq runs over the CR fluxes of B, C, O, 7{}^{7}Be, and 10{}^{10}Be (in Table I), each energy bin is calculated separately to get the quadratic distance between the data and the model. The constraint from the cross-section data contributes an additional term χcs2\chi_{\mathrm{cs}}^{2}, where ω1i\omega_{1i} is the high-energy uncertainty of the specific channel ii taken from Table II, and μi\mu_{i} is the tested value of renormalization factor in the fit. Note that the CR data contributes most of the degrees of freedom (d.o.f.) and has better precision than the cross-section data. This means that even if χ2/nd.o.f.\chi^{2}/n_{\mathrm{d.o.f.}} is less than 1, we still need to check whether χcs2/ncs\chi_{\mathrm{cs}}^{2}/n_{\mathrm{cs}} is reasonable.

Markov chain Monte Carlo (MCMC) methods are widely used in Bayesian inference and are powerful to sample the multi-dimensional parameter space for CR propagation models [30, 65, 66, 67]. We perform a combined fitting of CR observations together with cross-section uncertainties by using the public code CosmoMC6 as a generic Monte Carlo sampler to explore parameter space. In the previous paper [29], we have introduced the basic settings for GALPROP and CosmoMC. For the resolution of the GALPROP calculation in the work, we set a 2D spatial grid of dr=1dr=1 kpc and dz=0.2dz=0.2 kpc, and an energy grid of Ekin.factor = 1.2, considering both accuracy and speed. The size of the initial time step (start-timestep) is set to be 1.0e8, which is smaller than the default. We have checked that it does not affect the results. Other parameters are kept as the defaults of GALPROP v56.

The group of free parameters in the fitting procedure is

θ={D0,δ,L,Va,η,ν0,ν1,Rbr,AC,AO,ϕ,μC−B10,μC−B11,μC−C11,…}\theta=\{D_{0},\delta,L,V_{a},\eta,\nu_{0},\nu_{1},R_{\mathrm{br}},A_{\mathrm{C}},A_{\mathrm{O}},\phi,\mu_{\mathrm{C}-\mathrm{B}_{10}},\mu_{\mathrm{C}-\mathrm{B}_{11}},\mu_{\mathrm{C}-\mathrm{C}_{11}},\ldots\}

where the D0D_{0}, δ\delta, LL, VaV_{a}, and η\eta are the propagation parameters; the ν0\nu_{0}, ν1\nu_{1}, RbrR_{\mathrm{br}}, ACA_{\mathrm{C}}, AOA_{\mathrm{O}}, and ϕ\phi are the injection parameters and solar modulation potential, respectively; the rest parameters are the renormalization factors for the cross sections of specific channels.

The fitting result is shown in Table III. The goodness of fit is acceptable with χ2/nd.o.f.=232.6/320\chi^{2}/n_{\mathrm{d.o.f.}}=232.6/320, and the χ2\chi^{2} contribution from cross-section renormalization is χcs2/ncs=3.2/10\chi_{\mathrm{cs}}^{2}/n_{\mathrm{cs}}=3.2/10. The best-fit slope index of diffusion in the halo δ=0.433\delta=0.433 prefers the Iroshnikov-Kraichnan type (1/2) [68] hydromagnetic turbulence as the source of diffusion. The posterior distribution of the halo height LL is constrained to be 5.674±0.7585.674\pm0.758 kpc at 68% confidence. The result is smaller, compared with the ∼7\sim7 kpc result from some works [19, 69]. The difference mainly originates from the adopted cross-section parametrizations, since we have updated the cross section datasets and calculated a data-driven parametrization. We refer the reader to Appendix § 1.1.1 for checking the differences. On the other hand, a smaller halo height is constrained as the result of some works [70]. The differences in parametrizations may also explain it, for example, the inclusion of [NA61/SHINE] data [32, 33] has increased the production of secondary B, which allows a slightly larger halo height for nuclei to diffuse out of the Galaxy. Another reason is that the 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} ratio observed by AMS-02 [25] is smaller than the data of ISOMAX [71], indicating the current estimation of the 10Be{}^{10}\mathrm{Be} flux is smaller than that obtained by earlier works based on the ISO-MAX data. As the diffusion should dominate the fluxes of 10Be{}^{10}\mathrm{Be} over the radioactive decay at high energies, a smaller 10Be{}^{10}\mathrm{Be} flux indicates a larger halo size. The quality of the isotope fluxes measured by AMS-02 is better and can give a strict constraint, while the inclusion of cross-section uncertainty enlarges the confidence interval of the halo height conversely. As a result, the constraint of the halo height shown in Table III is slightly larger in comparison with the ±0.4\pm0.4 kpc result based on a simple analysis without considering cross-section uncertainties [27].

ParameterPrior rangeBest-fit valuesPosterior 95% range
D0(1028 cm2 s−1)D_{0}(10^{28}\,\mathrm{cm}^{2}\,\mathrm{s}^{-1})[0,15.0][0,15.0]5.197[4.176,6.396][4.176,6.396]
δ\delta[0.2,1.0][0.2,1.0]0.433[0.424,0.456][0.424,0.456]
L(kpc)L(\mathrm{kpc})[1.0,20.0][1.0,20.0]5.674[4.384,7.443][4.384,7.443]
Va(km/s)V_{a}(\mathrm{km}/\mathrm{s})[0,50][0,50]15.409[12.333,18.068][12.333,18.068]
η\eta[−5,5][-5,5]-0.484[−0.732,−0.161][-0.732,-0.161]
ν0\nu_{0}[0.5,2.4][0.5,2.4]1.249[1.003,1.446][1.003,1.446]
ν1\nu_{1}[2.2,2.5][2.2,2.5]2.390[2.372,2.400][2.372,2.400]
Rbr(GV)R_{\mathrm{br}}(\mathrm{GV})[0.1,15][0.1,15]2.088[1.743,2.551][1.743,2.551]
Ac(103)A_{c}(10^{3})[2.5,4.5][2.5,4.5]3.304[3.257,3.328][3.257,3.328]
Ao(103)A_{o}(10^{3})[3.5,5.5][3.5,5.5]4.114[4.062,4.185][4.062,4.185]
ϕ(GV)\phi(\mathrm{GV})[0.4,1.0][0.4,1.0]0.645[0.619,0.697][0.619,0.697]
μC−B10\mu_{C-B10}[−0.5,0.5][-0.5,0.5]0.099[−0.085,0.316][-0.085,0.316]
μC−B11\mu_{C-B11}[−0.5,0.5][-0.5,0.5]0.075[−0.007,0.211][-0.007,0.211]
μC−C11\mu_{C-C11}[−0.5,0.5][-0.5,0.5]-0.001[−0.018,0.025][-0.018,0.025]
μO−B10\mu_{O-B10}[−0.5,0.5][-0.5,0.5]0.010[−0.062,0.081][-0.062,0.081]
μO−B11\mu_{O-B11}[−0.5,0.5][-0.5,0.5]0.026[−0.058,0.071][-0.058,0.071]
μO−C11\mu_{O-C11}[−0.5,0.5][-0.5,0.5]0.026[−0.095,0.134][-0.095,0.134]
μC−Be7\mu_{C-\mathrm{Be}7}[−0.5,0.5][-0.5,0.5]-0.001[−0.037,0.049][-0.037,0.049]
μC−Be10\mu_{C-\mathrm{Be}10}[−0.5,0.5][-0.5,0.5]-0.039[−0.159,0.122][-0.159,0.122]
μO−Be7\mu_{O-\mathrm{Be}7}[−0.5,0.5][-0.5,0.5]0.001[−0.036,0.045][-0.036,0.045]
μO−Be10\mu_{O-\mathrm{Be}10}[−0.5,0.5][-0.5,0.5]0.031[−0.154,0.128][-0.154,0.128]
χmin⁡2/ndof\chi_{\min}^{2}/n_{\mathrm{dof}}⋯\cdots232.6/320⋯\cdots
χcs2/ncs\chi_{\mathrm{cs}}^{2}/n_{\mathrm{cs}}⋯\cdots3.2/10⋯\cdots

Table III. The prior range, best-fit values, and posterior 95% range of all parameters in the combined fitting.

Compared with the best-fit result of the B spectrum, we find that they are almost the same, implying that the renormalization factors μ\mu of B production channels are no longer required if the [Ba05] measurements are discarded. The renormalization of the cross section can also impact the fluxes below 2 GeV/nn and prefers an increment of B production at low energies. The removal of [Ba05] does not change the spectrum prediction below 2 GeV/nn, since the cross section is constrained by other measurements. As a result, the slightly under-prediction of fluxes around 5 GV can be seen in the top panel, which needs a thorough estimation of the uncertainties from the cross section, solar modulation, and also the systematic errors of AMS-02 data with its covariance matrices [15]. We also illustrate the best-fit 7Be{}^{7}\mathrm{Be} flux in the bottom panel of Figure 3, which matches the AMS-02 experimental data well, without the need for significant correction to the production cross section of 7Be{}^{7}\mathrm{Be}. As 7Be{}^{7}\mathrm{Be} and B nuclei both originated from the fragmentation of primary CR nuclei (mostly C and O) and have similar mass numbers, they should indeed be explained by almost the same diffusion process in the Galaxy. Thus, our results indicate that the production cross-section measurements and CR flux measurements of 7Be{}^{7}\mathrm{Be} and B are self-consistent.

Figure 3. Top panel: The B spectrum calculated with the best-fit parameters (green solid line) compared with the AMS-02 experimental data [53]. The red solid line represents the result without using the [Ba05] data and renormalizing the cross section. Bottom panel: The 7Be{}^{7}\mathrm{Be} spectrum calculated with the best-fit parameters compared with the AMS-02 experimental data [25].

In Figure 2, we illustrate the best-fit parameters of the cross-section renormalization factor, compared with the data expectation ranges from Table II. The first 6 channels are related to the B production. The best-fit result implies a systematical overproduction to fulfill the CR observation, which may not be attributed to the random errors of the cross sections. As we mentioned in Sec. II, the data expectation ω1\omega_1 is obtained by fitting the cross-section data points at ≥2\ge2 GeV/nn. Coincidentally, we noticed that the cross-section measurements of [Ba05] [72] at 3.25 GeV/nn for both the two dominant channels of 16O⟶11B{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{B} and 16O⟶11C{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{C} imply a reduction of about 30% compared with other measurements, which significantly changes the expectation of parametrizations at high energies. By removing those two points and ignoring the renormalization of the first 6 channels (μC−B10\mu_{\mathrm{C-B10}}, μC−B11\mu_{\mathrm{C-B11}}, μC−C11\mu_{\mathrm{C-C11}}, ...), we calculate another B flux illustrated in the top panel of Figure 3.

Figure 2. Comparison between cross-section uncertainties (ω1\omega_1 in Table II) and best-fit renormalization factors (μ\mu in Table III).

The extrapolation of 10Be{}^{10}\mathrm{Be} fluxes is still uncertain since the cross-section data points above 3 GeV/nn ex-

hibit an unexpected suppression. To analyze the impact of the suppression, we assume three parametrizations that determine the position of the plateau differently, named after bestfit, be10low, and be10low2. The bestfit parametrization is obtained without fitting these data points above 3 GeV/n, considering that they are significantly (∼70%\sim70\%) lower than the energy-independent expectation. The calculated fluxes are shown in Figure 4. The be10low2 parametrization (yellow line) significantly under-predicts the 10^{10}Be fluxes compared with the AMS-02 observation and is thus disfavored. The cross-section measurement at 3.25 GeV/n from [Ba05] [72] may not be accurate, which is also disapproved by the above analysis of the B production channels. The green line illustrates the best-fit result of the parametrization we used in the combined fitting, and the purple line illustrates the be10low parametrization that considered the last point from [Yi69] [73]. The latter allows a slower decline, and the cross section becomes constant at above 19 GeV/n. The predicted 10^{10}Be flux is a bit lower than the best-fit result but shows a decreasing trend that might be implied from the final two points of AMS-02 measurement [53]. Thus, the bestfit and be10low parametrizations are both acceptable in terms of interpreting the 10^{10}Be flux.

Figure 4. Top panel: Cross-section measurements of the channel O+p ⟶\longrightarrow 10^{10}Be, compared with three possible parametrizations based on different adoption of data. Bottom panel: The calculated of 10^{10}Be fluxes based on three possible parametrizations.

To distinguish the extrapolation of the 10^{10}Be cross-section parametrization, we use the Be/B ratio provided by AMS-02 [5] as a reference, which is measured up to ∼1\sim1 TV. We simply assume that the cross section of 9^{9}Be follows the parametrization introduced in Appendix § 1.1.1, which becomes constant at above 1 GeV/n that is preferred by the 9Be^{9}\mathrm{Be} flux (see Sec. IV). In Figure 5, the predictions of the three parametrizations are compared with the measurements. The highest result (green line) fits the AMS-02 data best, while those with much smaller cross section may be disfavoured as their overall Be fluxes are not sufficient to reproduce the Be/B ratio. Since the total Be flux is determined by the isotope 7Be^{7}\mathrm{Be}, 9Be^{9}\mathrm{Be}, and 10Be^{10}\mathrm{Be} fluxes, it is necessary to analyze the 9Be^{9}\mathrm{Be} production, to determine a proper extrapolation of the Be/B ratio. In the next section, we will show that the parametrizations of bestfit and be10low can hardly be distinguished, considering the unclear determination of the 9Be^{9}\mathrm{Be} cross section at high energies. Nonetheless, the diffusion halo size is mainly determined by the 10Be^{10}\mathrm{Be} flux at ≤10\leq10 GeV/n, which would not be affected by the assumption of the cross-section extrapolation in higher-energy range.

Prediction of Be/B ratios based on three possible parametrizations

Figure 5. The prediction of Be/B ratios based on three possible parametrizations defined in the Figure 4 and the 2σ2\sigma confidence interval (yellow band) of the bestfit parametrization, compared with the AMS-02 observation [5].

Lessons from the AMS-02 measurements of Be-9 and Be/B

The uncertainty of 9Be^{9}\mathrm{Be} cross-section measurements is much larger than that of 7Be^{7}\mathrm{Be}, and we did not use the 9Be^{9}\mathrm{Be} observation from AMS-02 [25] to constrain the model parameters in Sec. III. In this section, we conversely use the precise 9Be^{9}\mathrm{Be} flux to give a constraint to its cross-section production.

As illustrated in Appendix § 1.1.1, the 9Be^{9}\mathrm{Be} cross-section uncertainty is dominated by the O + p ⟶\longrightarrow 9Be^{9}\mathrm{Be} channel. To transfer the uncertainty of the 9Be^{9}\mathrm{Be} flux to the uncertainty of this dominant cross-section channel, we adopt the routine:

  1. For each energy bin [Ei,Ei+ΔEi][E_i,E_i+\Delta E_i] provided by the AMS-02 measurement [25], we calculate the renormalization factor ki=yidata/yimodelk_i=y_i^{\mathrm{data}}/y_i^{\mathrm{model}} and its uncertainty ri=σidata/yimodelr_i=\sigma_i^{\mathrm{data}}/y_i^{\mathrm{model}}, where yimodely_i^{\mathrm{model}} is the 9Be^{9}\mathrm{Be} flux calculated with the best-fit propagation parameters determined in Sec. III.

  1. To transfer the total 9Be^{9}\mathrm{Be} flux to the contribution of only the O + p ⟶\longrightarrow 9Be^{9}\mathrm{Be} channel, we calculate the fraction of the reaction [10] with

fabc=ψ−ψ(σa+b→c=0)ψ.(8)f_{abc}=\frac{\psi-\psi(\sigma^{a+b\to c}=0)}{\psi}. \tag*{(8)}

The renormalization factor k0k_0 for specific channel can be calculated with fabc∗(k0−1)=k−1f_{abc}*(k_0-1)=k-1, and its error fabc∗r0=rf_{abc}*r_0=r.

  1. As the observed data corresponded to the post-modulated 9Be^{9}\mathrm{Be} flux, the original energy bin shall be [Ei,Ei+ΔEi]+ϕZA[E_i,E_i+\Delta E_i]+\phi\frac{Z}{A}, where ϕ=0.645\phi=0.645 GV, Z=4Z=4, A=9A=9.

  1. The renormalized cross section for each energy bin is calculated with σ=σ0∗(k0±r0)\sigma=\sigma_0*(k_0\pm r_0), where σ0\sigma_0 is the parametrization we adopted in the work, which is equal to 2.4 mb above 1 GeV/n.

In Figure 6, we illustrate the calculated result of the uncertainty band{7} for the O + p ⟶\longrightarrow 9Be^{9}\mathrm{Be} channel. The AMS-02 observation prefers a much lower cross section, which may disfavor some cross-section measurements. The highest and most disfavoured one is the [Bao5] experiments, which measured the cross section at 3.25 GeV/n. In the figure, the default parametrization [GAL12] and the parametrizations of [WE93] (WNEW) and [TS00] (YIELDX) taken from the GALPROP code seem to systematically overpredict the cross section, implying that a significant reduction should be applied to better reconstruct the 9Be^{9}\mathrm{Be} flux. Above 12 GeV/n, the cross section is not constrained due to the maximum observed energy of the AMS-02 isotope. The cross section measurements at 19 GeV/n are from [Yi69] [73] and [RV84] [74], making the extrapolation result of 9Be^{9}\mathrm{Be} unable to be distinguished yet.

Figure 6. Cross section of O + p ⟶\longrightarrow 9Be^{9}\mathrm{Be} constrained by the AMS-02 9Be^{9}\mathrm{Be} data (green band), compared with the available cross-section measurements by different experiments. The reference parametrizations are taken from the GALPROP code.

We notice that the [Yi69] [73] data gives unexpected cross section observation at high energies. In Figure 4, we analyzed that the inclusion of [Yi69] can predict a decline of cross section, and the extrapolation of 10Be{}^{10}\mathrm{Be} flux is lower. While in Figure 6 the inclusion of [Yi69] can predict an increment of cross section, making the extrapolation of 9Be{}^{9}\mathrm{Be} flux larger. Combining these assumptions, we can obtain new results that are different from what we have got without using [Yi69] data.

In Figure 7, we illustrate the 9Be{}^{9}\mathrm{Be}, 10Be{}^{10}\mathrm{Be}, 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be}, and Be/B result with or without using the [Yi69] data to predict the extrapolation of cross section. Generally, these two assumptions fit all the available observations well. In the top panel, the prediction of 9Be{}^{9}\mathrm{Be} flux is higher when adopting the [Yi69] data, but still within the data constraints. The preliminary 10Be{}^{10}\mathrm{Be} flux measured by AMS-02 implies an unexpected decline feature at above 10 GeV/nn, which remains to be confirmed by higher-energy measurements. By including the [Yi69] data, we assume that the cross section at above 2 GeV/nn decrease steadily for 16O⟶10Be{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{Be} channel, whose validity requires more observations to prove. As a result, the predicted 10Be{}^{10}\mathrm{Be} flux follows a decline suggested by observation. For almost the same reason, as shown in the middle panel, the prediction of 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} ratio at about 10 GeV/nn is better when using the [Yi69] data; the assumption of not using [Yi69] data predicts higher 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} ratio as the energy increases. The best-fit result of 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} ratio predicts a bump structure at around 1 GeV/nn, which has been discovered by the preliminary AMS-02 measurement. This structure is related to the 10Be{}^{10}\mathrm{Be} cross-section bump shown in Figure 4, which is constrained by a series of data from Michel’s group [75, 76].

Figure 7. Spectra calculated with or without [Yi69] data. Top panel: The 9Be{}^{9}\mathrm{Be} and 10Be{}^{10}\mathrm{Be} spectra compared with AMS-02 measurements [25]. Middle panel: The 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be} ratio compared with measurements [25, 71, 79, 80]. Bottom panel: The Be/B ratio compared with AMS-02 measurements [5].

As shown in Appendix § 1.1.1, most of the 7Be{}^{7}\mathrm{Be} production cross section channels are strictly constrained by observations, while the contributions from 9Be{}^{9}\mathrm{Be} and 10Be{}^{10}\mathrm{Be} depend on how we extrapolate the cross sections. The AMS-02 Be/B ratio [5] may constrain the total Be flux at high energy and testify the assumption of 9Be{}^{9}\mathrm{Be} and 10Be{}^{10}\mathrm{Be}. In the bottom panel of Figure 7, we predict the Be/B ratio with or without using the [Yi69] data, but the results are so close that cannot be distinguished. The former result (green line) under-predicts the Be/B ratio at above 100 GV, while better reproducing the flux at about 10 GV compared with the latter result. As we have adjusted the cross section of B and Be at low energies according to the available cross-section measurements, the predicted Be/B ratio at several GVs shows similar trends as what AMS-02 measured.

It is important to decide the 10Be{}^{10}\mathrm{Be} cross section since the production of 10Be{}^{10}\mathrm{Be} directly impacts the determination of the halo height LL. In Table III the best-fit value of LL is 5.674 kpc, under the assumption that the 16O⟶10Be{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{Be} cross section is energy-independent above 3 GeV/nn. If using the [Yi69] data, we find that the best-fit value of LL would decrease to ∼5\sim5 kpc. The difference is less significant since the Be isotope fluxes remain unchanged at low energies where the halo height is more sensitive to the flux.

From a global analysis of the CR and cross-section data mentioned above, we prefer discarding the [Ba05] [72] data for channels of 16O⟶11B{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{B}, 16O⟶11C{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{C}, 16O⟶9Be{}^{16}\mathrm{O}\longrightarrow{}^{9}\mathrm{Be} and 16O⟶10Be{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{Be}. In Appendix § 1.1.1, the [KO99] and [KO02] [77, 78] data of several channels were ignored as they significantly deviate from other measurements. Coincidentally, both groups (Bazarov et al. and Korejwo et al.) performed their experiments by using the synchrospahatron’s beam at the Joint Institute for Nuclear Research in Dubna.

We have also calculated the total Li flux and its isotopic composition with the best-fit parameters in Table III. Compared with the AMS-02 measurements [5, 24], we found a 12% overprediction of Li flux, which was also mentioned by Maurin [22]. Our result of the 6Li{}^{6}\mathrm{Li} spectrum significantly overpredicts the flux while the 7Li{}^{7}\mathrm{Li} spectrum fits the preliminary AMS-02 measurements [24] well. Those features are mostly attributed to the over-assumed cross section of the 6Li{}^{6}\mathrm{Li}, as most of its production channels are poorly determined with seldom or no data. The next step of this work would be to calculate and analyze these secondary nuclei to make constraints on their cross sections.

SUMMARY

The preliminary results of the Be isotope flux measurements provided by AMS-02 have reached an unprecedented energy of 12 GeV/n, which is of significant importance for constraining the propagation parameters of Galactic CRs. Moreover, the precision of the AMS-02 measurements is now significantly better than that of the production cross sections for certain isotopes. In light of this, we combine the high-precision cross sections of selected isotopes with the AMS-02 CR measurements to refine the propagation model. This model, in conjunction with the AMS-02 measurements, has been employed to inversely estimate the less constrained production cross sections. This method should also be meaningful for studying the forthcoming isotopic data of other nuclei.

We employ the GALPROP code to calculate CR propagation and conduct a comprehensive Bayesian analysis using a MCMC sampling algorithm to derive posterior distributions of the model parameters. We meticulously account for the uncertainties in the production cross sections of secondary particles to achieve more robust constraints on the parameters. Our approach incorporates a data-driven parametrization of cross sections, fully utilizing the wealth of available experimental data. We have enriched the isotopic cross-section database within GALPROP by including additional data from the EXFOR database, as well as the fresh high-energy B production cross sections from the NA61/SHINE experiment. Moreover, we adopt an energy-dependent form for the uncertainty of the parametrization, thereby reflecting the increased uncertainty at higher energies where the cross sections are less stringent.

In the Bayesian analysis, we use the CR flux measurements of 7Be{}^{7}\mathrm{Be}, 9Be{}^{9}\mathrm{Be}, B, C, and O. Owing to the high-quality measurements of the 7Be{}^{7}\mathrm{Be} production cross section, we have innovatively employed 7Be{}^{7}\mathrm{Be} in place of 9Be{}^{9}\mathrm{Be} to constrain propagation parameters. The resulting parameters are largely consistent with previous studies: the diffusion coefficient at 1 GeV is ≈5×1028 cm2s−1\approx5 \times10^{28}\ \mathrm{cm}^{2}\mathrm{s}^{-1}, and the diffusion index is 0.43, aligning more closely with the expectation of Kraichnan’s turbulence theory. Notably, the thickness of the diffusion halo is constrained to 5.67 ±\pm 0.76 kpc, representing a moderate value compared to previous analogous works. Both as mid-mass stable secondary particles, B and 7Be{}^{7}\mathrm{Be} data are well-fitted simultaneously, yielding acceptable χ2\chi^{2} for both the global fit and the cross-section part. This suggests that the CR measurements and the production cross-section measurements of 7Be{}^{7}\mathrm{Be} and B are self-consistent.

Using the well-constrained CR propagation model and the 9Be{}^{9}\mathrm{Be} cross-section parametrizations embedded in GALPROP, we discover a significant overestimation of the 9Be{}^{9}\mathrm{Be} CR fluxes in comparison with the AMS-02 measurement. In response to the demands of the CR data, we impose a constraint to the cross section of the 16O⟶9Be{}^{16}\mathrm{O}\longrightarrow{}^{9}\mathrm{Be} channel in the energy range of ∼1\sim1 to 10 GeV/n. The result indicates that the production cross section ought to be significantly lower than previously thought. In particular, we observe that the cross section at 3.25 GeV/n measured by the [Ba05] experiment [72] is significantly elevated beyond our expectation and is therefore considered less credible.

Intriguingly, the [Ba05] measurement for 16O⟶10Be{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{Be} is similarly challenged by the 10Be{}^{10}\mathrm{Be} spectrum of AMS-02. Furthermore, while the 7Be{}^{7}\mathrm{Be} and B data can be explained consistently, the required renormalization factors for B production channels are slightly but systematically higher, which may not be ascribed to random fluctuations in cross-section observations. This discrepancy is resolved when the [Ba05] cross-section measurements for 16O⟶11B{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{B} and 16O⟶11C{}^{16}\mathrm{O}\longrightarrow{}^{11}\mathrm{C} channels are excluded. This could be the first instance of picking up unreliable nucleon production cross-section data through the CR study, rather than through inconsistencies among different cross-section measurements.

The possible high-energy extrapolations of the Be isotopic spectrum are discussed. We notice that the cross-section data of [Yi69] [73] for 16O⟶9Be{}^{16}\mathrm{O}\longrightarrow{}^{9}\mathrm{Be} and 16O⟶10Be{}^{16}\mathrm{O}\longrightarrow{}^{10}\mathrm{Be} channels, which are not used in our default calculations, indicate energy-dependent cross sections up to 19 GeV/n. It could influence the high-energy behavior of the extrapolated spectra. By adopting the [Yi69] observations, we find that the outcomes remain consistent with the CR data of Be isotopes and Be/B, making it challenging to distinguish from the default scenario. However, it is noteworthy that the inclusion of [Yi69] data yields a better fit to the high-energy tail of the AMS-02 10Be{}^{10}\mathrm{Be} (or 10Be/9Be{}^{10}\mathrm{Be}/{}^{9}\mathrm{Be}) data. The forthcoming balloon-borne experiment HELIX [81, 82] can provide precise 10Be{}^{10}\mathrm{Be} and 9Be{}^{9}\mathrm{Be} measurements in the energy range of 0.2 ∼\sim 3 GeV/n, and a new magnet spectrometer would be subsequently payload to measure up to 10 GeV/n. This may give a cross-check to the high-energy features observed by AMS-02, thereby offering a critical evaluation of the associated cross-section measurements.

Acknowledgments

This work is supported in part by the National Natural Science Foundation of China under Grants No. 12042507, No. 12105292, and No. 12175248.

References

CROSS SECTION DATA

Here we show the plots of the most important channels needed for analyzing Be and B production in the paper. Secondary CRs arise from the fragmentation of heavier nuclei upon collision with the ISM gas, which is composed of mostly hydrogen and helium gas. In the section, we only present channels of collisions with the hydrogen target for simplicity. As implemented in GALPROP’s fragmentation routine, the collisions with the helium target are calculated using a parametrization by Ferrando [83] where the interstellar gas ratio of helium to hydrogen is set to be 0.11.

Figures below show the comparison between the model and measurements for the relevant channels. The available data are obtained from:

  1. isotope_cs.dat: The isotopic cross-section database file is built in the GALPROP code [84, 85] for normalizing the parametrization formulae, such as WNEW code by Webber [62, 86, 87] or YIELDX code by Tsao and Silberberg [13]. The cross section data assembled in the file were taken from multiple cross-section measurements published before 2003.

  1. EXFOR (Experimental Nuclear Reaction Data): The website8 is an extensive database containing experimental data, as well as bibliographic information, experimental setup, and source of uncertainties. By querying the EXFOR database, we can add most of the measurements published so far.

  1. NA61/SHINE: Additional measurements are reported by the NA61/SHINE Collaboration [32, 33] in the International Cosmic Ray Conference (ICRC). The recent pilot run provided precise high-energies measurements of cross sections from the C

projectile at 13.5 GeV/n, which is valuable for constraining the uncertainties of the dominant channels.

In this work, we analyzed a data-driven parametrization by using the default evaluation routine implemented in the GALPROP code (see nuc_package.cc for details). The routine checks if there exists cross-section data for corresponding channels given in the file eval_iso_cs.dat and if so interpolates it linearly. Otherwise, the routine would use the parametrization formulae of WNEW or YIELDX and re-normalize it according to the data given in isotope_cs.dat. Webber’s parametrization doesn’t define the secondary Li production, and only the contributions of the dominant channels are introduced in eval_iso_cs.dat for calculating the Li flux when adopting the [GAL12] parametrization. It has been pointed out [22] that the Li flux predicted by [GAL12] is significantly lower than that of [GAL22], as the former doesn’t properly calculate the Li production from Fe projectile channels and lacks other missing channels. To improve that, we use the parametrization provided by Silberberg and Tsao instead for specific channels, when Webber’s parametrization cannot provide a non-zero cross section.

We have added thousands of data in eval_iso_cs.dat up to ∼10\sim10 GeVs to interpolate credible parametrization for important production channels of 2H{}^{2}\mathrm{H}, 3He{}^{3}\mathrm{He}, Li, Be, B, F, P, Sc, Ti and V. As shown in Figure 8 and Figure 9, labeled as [Not used], some data points published by different groups could deviate from the general interpolation by a significant factor. We remove these measurements during the estimation of parametrization for a clearer determination. For example, the [KO99] and [KO02] [77, 78] data of several channels (such as 12C⟶11B{}^{12}\mathrm{C} \longrightarrow{}^{11}\mathrm{B}) were omitted as they significantly deviate from other measurements. Sometimes we are not sure whether to discard an important measurement, which decides how the cross section becomes constant and extrapolates to the higher energy. For example, the newly measured data by [Ba05] [72] at 3550 MeV/n indicates a smaller cross section compared with other measurements, and we cannot distinguish if the cross section reduction could happen at that energy.

Figure 8. Channels: C + p →\to B or Be isotopes. The parametrizations [GAL12] shown in the figures were taken from the GALPROP code.

Channels: O + p → B or Be isotopes

Figure 9. Channels: O + p → B or Be isotopes. The parametrizations [GAL12] shown in the figures were taken from the GALPROP code.

The parametrization of the channel 16O⟶11B{}^{16}\mathrm{O} \longrightarrow{}^{11}\mathrm{B} predicts a reduction of nearly 30%, and we choose to test its validity in the combined fitting together with CRs measurements in Sec. III. The channel 16O⟶10Be{}^{16}\mathrm{O} \longrightarrow{}^{10}\mathrm{Be} predicts a reduction of over 70%, hence we consider discarding it. In Sec. IV we find that the channel 16O⟶9Be{}^{16}\mathrm{O} \longrightarrow{}^{9}\mathrm{Be} prefers a much lower cross section, owing to the expectation from the AMS-02 9Be{}^{9}\mathrm{Be} observation. Hence we omit almost all the points at above 1 GeV/n and evaluate an energy-independent parametrization for this channel, which can naturally explain the 9Be{}^{9}\mathrm{Be} measurement.

In these figures, we also illustrate the default parametrization [GAL12] defined in GALPROP. As designed in the evaluation routines, the contributions of ghost nuclei 10C{}^{10}\mathrm{C} and 11C{}^{11}\mathrm{C} are directly counted as the cumulative 10B{}^{10}\mathrm{B} and 11B{}^{11}\mathrm{B} if the projectile is 16O{}^{16}\mathrm{O} or 14N{}^{14}\mathrm{N}. To compare [GAL12] and [GAL22] with other parametrizations and also the data, we subtract ghost nuclei based on Webber’s prediction [62, 87] of the isotopic proportion to the cumulative cross section. This is the same as how Génolini [10] treated ghost nuclei. As a result, the default parametrization illustrated in some specific channels is labeled as [GAL12*] instead, representing the subtracted parametrization based on [GAL12]. The [GAL12] parametrization of the reaction 16O⟶10B{}^{16}\mathrm{O} \longrightarrow{}^{10}\mathrm{B} illustrates the cumulative production 10B+10C{}^{10}\mathrm{B}+{}^{10}\mathrm{C} labeled as [GAL12(X10)], which fits the available measurements better and we didn’t subtract it.

Illustrated in Figure 8 and Figure 9, several channels lack of high-energy (>10> 10 GeV/n) measurements, such as 12C⟶10C{}^{12}\mathrm{C} \longrightarrow{}^{10}\mathrm{C}, 16O⟶10B{}^{16}\mathrm{O} \longrightarrow{}^{10}\mathrm{B}, 16O⟶11B{}^{16}\mathrm{O} \longrightarrow{}^{11}\mathrm{B}, 16O⟶10C{}^{16}\mathrm{O} \longrightarrow{}^{10}\mathrm{C}, 16O⟶11C{}^{16}\mathrm{O} \longrightarrow{}^{11}\mathrm{C}, 16O⟶9Be{}^{16}\mathrm{O} \longrightarrow{}^{9}\mathrm{Be} and 16O⟶10Be{}^{16}\mathrm{O} \longrightarrow{}^{10}\mathrm{Be}. Most of them are less constrained with inadequate datasets, and we hope they could be improved by more cross-section observations in the future.

In addition, we attach ancillary files in the arXiv version that can be used to check the B and Be production results given in this work. These files (eval_iso_cs.dat, isotope_cs.dat and p_cs_fits.dat) can be downloaded and added to the original files of the GALPROP code in the “galtoolslib/nuclei” folder for testing purpose.

References

  1. [1]O. Adriani et al. (PAMELA), Science 332, 69 (2011), arXiv:1103.4055 [astro-ph.HE].DOI
  2. [2]M. Aguilar et al. (AMS), Phys. Rev. Lett. 120, 021101 (2018).
  3. [3]C. Consolandi (AMS), in 25th European Cosmic Ray Symposium (2016) arXiv:1612.08562 [astro-ph.HE].
  4. [4]M. Aguilar et al. (AMS), Phys. Rev. Lett. 119, 251101 (2017).
  5. [5]M. Aguilar et al. (AMS), Phys. Rept. 894, 1 (2021).
  6. [6]M. Aguilar et al. (AMS), Phys. Rev. Lett. 127, 02101 (2021), [Erratum: Phys.Rev.Lett. 127, 159901 (2021)].
  7. [7]A. W. Strong, I. V. Moskalenko, and V. S. Ptuskin, Ann. Rev. Nucl. Part. Sci. 57, 285 (2007), arXiv:astro-ph/0701517.DOI
  8. [8]A. W. Strong and I. V. Moskalenko, Astrophys. J. 509, 212 (1998), arXiv:astro-ph/9807150.DOI
  9. [9]D. Maurin, F. Donato, R. Taillet, and P. Salati, Astrophys. J. 555, 585 (2001), arXiv:astro-ph/0101231.DOI
  10. [10]Y. Genolini, D. Maurin, I. V. Moskalenko, and M. Unger, Phys. Rev. C 98, 034611 (2018), arXiv:1803.04686 [astro-ph.HE].DOI
  11. [11]Y. Génolini, D. Maurin, I. V. Moskalenko, and M. Unger, arXiv e-prints , arXiv:2307.06798 (2023), arXiv:2307.06798 [astro-ph.HE].DOI
  12. [12]W. R. Webber, J. C. Kish, and D. A. Schrier, Phys. Rev. C 41, 566 (1990).
  13. [13]R. Silberberg, C. H. Tsao, and A. F. Barghouty, Astrophys. J. 501, 911 (1998).
  14. [14]K. Summerer and B. Blank, Phys. Rev. C 61, 034607 (2000), arXiv:nucl-ex/9911006.DOI
  15. [15]N. Weinrich, Y. Génolini, M. Boudaud, L. Derome, and D. Maurin, Astron. Astrophys. 639, A131 (2020), arXiv:2002.11406 [astro-ph.HE].DOI
  16. [16]M. J. Boschini et al., Astrophys. J. Suppl. 250, 27 (2020), arXiv:2006.01337 [astro-ph.HE].
  17. [17]Q. Yuan, C.-R. Zhu, X.-J. Bi, and D.-M. Wei, JCAP 11, 027 (2020), arXiv:1810.03141 [astro-ph.HE].DOI
  18. [18]P. De La Torre Luque, M. N. Mazziotta, F. Loparco, F. Gargano, and D. Serini, JCAP 03, 099, arXiv:2101.01547 [astro-ph-HE].
  19. [19]P. de la Torre Luque, M. N. Mazziotta, A. Ferrari, F. Loparco, P. Sala, and D. Serini, JCAP 07 (07), 008, arXiv:2202.03559 [astro-ph-HE].DOI
  20. [20]M. Korsmeier and A. Cuoco, Phys. Rev. D 103, 103016 (2021), arXiv:2103.09824 [astro-ph.HE].DOI
  21. [21]P. D. L. T. Luque, M. N. Mazziotta, F. Loparco, F. Gargano, and D. Serini, JCAP 07, 010, arXiv:2102.13238 [astro-ph.HE].DOI
  22. [22]D. Maurin, E. F. Bueno, Y. Génolini, L. Derome, and M. Vecchi, Astron. Astrophys. 668, A7 (2022), arXiv:2203.00522 [astro-ph.HE].DOI
  23. [23]M. J. Boschini et al., Astrophys. J. 889, 167 (2020), arXiv:1911.03108 [astro-ph.HE].DOI
  24. [24]L. Derome, Cosmic-ray lithium and beryllium isotopes with ams02, https://indico.desy.de/event/27991/contributions/101805/ (2021).
  25. [25]J. Wei, Properties of cosmic beryllium isotopes, https://agenda.infn.it/event/28874/contributions/170166/ (2022).
  26. [26]P. Lipari, arXiv e-prints , arXiv:2204.13085 (2022), arXiv:2204.13085 [astro-ph.HE].
  27. [27]H. Jacobs, P. Mertsch, and V. H. M. Phan, Mon. Not. Roy. Astron. Soc. 526, 160 (2023), arXiv:2305.10337 [astro-ph.HE].arxiv.org/abs/2305.10337
  28. [28]N. Tomassetti, Astrophys. J. Lett. 752, L13 (2012), arXiv:1204.4492 [astro-ph.HE].
  29. [29]M.-J. Zhao, K. Fang, and X.-J. Bi, Phys. Rev. D 104, 123001 (2021).
  30. [30]G. Jóhannesson et al., Astrophys. J. 824, 16 (2016), arXiv:1602.02243 [astro-ph.HE].DOI
  31. [31]M.-J. Zhao, X.-J. Bi, and K. Fang, Phys. Rev. D 107, 063020 (2023), arXiv:2209.03799 [astro-ph.HE].DOI
  32. [32]N. Amin (NA61/SHINE), PoS ICRC2021, 102 (2021), arXiv:2107.12275 [nucl-ex].arxiv.org/abs/2107.12275
  33. [33]N. Amin (NA61/SHINE), PoS ICRC2023, 075 (2023).
  34. [34]R. Trotta, G. Jóhannesson, I. V. Moskalenko, T. A. Porter, R. R. d. l. Austri, and A. W. Strong, Astrophys. J. 729, 106 (2011), arXiv:1011.0037 [astro-ph.HE].DOI
  35. [35]Y. Génolini et al., Phys. Rev. D 99, 123028 (2019), arXiv:1904.08917 [astro-ph.HE].DOI
  36. [36]Q. Yuan, Sci. China Phys. Mech. Astron. 62, 49511 (2019), arXiv:1805.10649 [astro-ph.HE].
  37. [37]E. S. Seo and V. S. Ptuskin, Astrophys. J. 431, 705 (1994).
  38. [38]A. W. Strong, I. V. Moskalenko, and O. Reimer, Astrophys. J. 537, 763 (2000), [Erratum: Astrophys.J. 541, 1109 (2000)], arXiv:astro-ph/9811296.
  39. [39]A. D. Pavon Val et al., Bull. Russ. Acad. Sci. Phys. 73, 564 (2009), arXiv:1101.3246 [astro-ph.HE].
  40. [40]O. Adriani et al. (CALET), Phys. Rev. Lett. 122, 181102 (2019), arXiv:1905.04229 [astro-ph.HE].DOI
  41. [41]O. Adriani et al., Phys. Rev. Lett. 125, 251102 (2020), arXiv:2012.10319 [astro-ph.HE].DOI
  42. [42]Q. An et al. (DAMPE), Sci. Adv. 5, eaax3793 (2019), arXiv:1909.12860 [astro-ph.HE].DOI
  43. [43]F. Alemanno, Q. An, P. Azzarello, F. Barbato, P. Bernardini, X. Bi, M. Cai, E. Catanzani, J. Chang, D. Chen, and et al., Physical Review Letters 126, 10.1103/physrevlett.126.201102 (2021).
  44. [44]Y. S. Yoon et al., Astrophys. J. 839, 5 (2017), arXiv:1704.02512 [astro-ph.HE].DOI
  45. [45]V. Ptuskin, V. Zirakashvili, and E.-S. Seo, Astrophys. J. 763, 47 (2013), arXiv:1212.0381 [astro-ph.HE].DOI
  46. [46]P. Blasi, E. Amato, and P. D. Serpico, Phys. Rev. Lett. 109, 061101 (2012), arXiv:1207.3706 [astro-ph.HE].DOI
  47. [47]S. Thoudam and J. Horandel, Mon. Not. Roy. Astron. Soc. 421, 1209 (2012), arXiv:1112.3020 [astro-ph.HE].DOI
  48. [48]Y. Ohira and K. Ioka, in 32nd International Cosmic Ray Conference (2010) arXiv:1011.4405 [astro-ph.HE].
  49. [49]D. Collaboration (DAMPE), Sci. Bull. 67, 2162 (2022), arXiv:2210.08833 [astro-ph.HE].DOI
  50. [50]V. H. M. Phan, F. Schulze, P. Mertsch, S. Recchia, and S. Gabici, Phys. Rev. Lett. 127, 141101 (2021), arXiv:2105.00311 [astro-ph.HE].DOI
  51. [51]M. Aguilar et al. (AMS), Phys. Rev. Lett. 124, 211102 (2020).
  52. [52]M. Aguilar et al. (AMS), Phys. Rev. Lett. 126, 041104 (2021).
  53. [53]M. Aguilar et al. (AMS), Phys. Rev. Lett. 130, 211002 (2023).
  54. [54]N. Gorbunov et al., Adv. Space Res. 64, 2546 (2019), arXiv:1809.05333 [astro-ph.IM].arxiv.org/abs/1809.05333
  55. [55]H. S. Ahn et al., Astrophys. J. 707, 593 (2009), arXiv:0911.1889 [astro-ph.HE].
  56. [56]A. C. Cummings, E. C. Stone, B. C. Heikkila, N. Lal, W. R. Webber, G. Jóhannesson, I. V. Moskalenko, E. Orlando, and T. A. Porter, Astrophys. J. 831, 18 (2016).
  57. [57]L. J. Gleeson and W. I. Axford, Astrophys. J. 154, 1011 (1968).
  58. [58]S. Weinrich, M. Boudaud, L. Derome, Y. Genolini, J. Lavalle, D. Maurin, P. Salati, P. Serpico, and G. Weymann-Despres, Astron. Astrophys. 639, A74 (2020), arXiv:2004.00441 [astro-ph.HE].DOI
  59. [59]V. V. Glagolev, K. G. Gulamov, M. Y. Kratenko, V. D. Lipin, S. L. Lutpullaev, K. Olimov, S. N. Shpilev, A. A. Yuldashev, I. E. Shokirov, and B. S. Yuldashev, JETP Lett. 58, 497 (1993).
  60. [60]M. M. Block and F. Halzen, Phys. Rev. Lett. 107, 212002 (2011), arXiv:1109.2041 [hep-ph].DOI
  61. [61]M. M. Block, L. Durand, P. Ha, and F. Halzen, Phys. Rev. D 92, 114021 (2015), arXiv:1511.02406 [hep-ph].DOI
  62. [62]W. R. Webber, A. Soutoul, J. C. Kish, and J. M. Rockstroh, The Astrophysical Journal Supplement Series 144, 153 (2003).
  63. [63]A. Reinert and M. Winkler, JCAP 01, 055, arXiv:1712.00002 [astro-ph.HE].
  64. [64]C. Evoli, R. Aloisio, and P. Blasi, Phys. Rev. D 99, 103023 (2019), arXiv:1904.10220 [astro-ph.HE].DOI
  65. [65]N. Masi, Nuovo Cimento C 39, 282 (2016).
  66. [66]A. Putze, L. Derome, and D. Maurin, Astron. Astrophys. 516, A66 (2010), arXiv:1001.0551 [astro-ph.HE].DOI
  67. [67]Q. Yuan, S.-J. Lin, K. K. Fang, and X.-J. Bi, Phys. Rev. D 95, 083007 (2017), arXiv:1701.06149 [astro-ph.HE].
  68. [68]R. H. Kraichnan, Phys. Fluids 8, 1385 (1965).
  69. [69]C. Evoli, G. Morlino, P. Blasi, and R. Aloisio, Phys. Rev. D 101, 023013 (2020), arXiv:1910.04113 [astro-ph.HE].DOI
  70. [70]D. Maurin, E. Ferronato Bueno, and L. Derome, Astron. Astrophys. 667, A25 (2022), arXiv:2203.07265 [astro-ph.HE].DOI
  71. [71]T. Hams et al., Astrophys. J. 611, 892 (2004).
  72. [72]E. K. Bazarov, V. V. Glagolev, V. V. Lugovoi, S. L. Lutpullaev, K. Olimov, V. I. Petrov, A. A. Yuldashev, and B. S. Yuldashev, JETP Lett. 81, 140 (2005).
  73. [73]F. Yiou, M. Baril, J. J. De Cِtres, P. Fontes, E. Gradsztajn, and R. Bernas, Phys. Rev. 166, 968 (1968).
  74. [74]S. Read and V. Viola, Atomic Data and Nuclear Data Tables 31, 359 (1984).
  75. [75]R. Michel et al., Nucl. Instrum. Meth. B 103, 183 (1995).
  76. [76]R. Michel, R. Bodemann, H. Busemann, R. Daunke, M. Gloris, H. J. Lange, B. Klug, A. Krins, I. Leya, M. Lupke, S. Neumann, H. Reinhardt, M. Schnatz-Buttgen, U. Herpers, T. Schiekel, F. Sudbrock, B. Holmqvist, H. Conde, P. Malmborg, M. Suter, B. Dittrich-Hannen, P. W. Kubik, H. A. Synal, and D. Filges, Nucl.Instrum.Methods Phys.Res. B129, 153 (1997).
  77. [77]A. Korejwo, T. Dzikowski, M. Giller, J. Wdowczyk, V. V. Perepelgin, and A. V. Zarubin, J. Phys. G 26, 1171 (2000).
  78. [78]A. Korejwo, M. Giller, T. Dzikowski, V. V. Peryelgin, and A. V. Zarubin, J. Phys. G 28, 1199 (2002).
  79. [79]N. E. Yanasak, M. E. Wiedenbeck, R. A. Mewaldt, A. J. Davis, A. C. Cummings, J. S. George, R. A. Leske, E. C. Stone, E. R. Christian, T. T. von Rosenvinge, W. R. Binns, P. L. Hink, and M. H. Israel, Astrophys. J. 563, 768 (2001).
  80. [80]F. Nozzoli and C. Cernetti, Universe 7, 183 (2021).
  81. [81]N. Park et al., PoS ICRC2021, 091 (2021).
  82. [82]S. P. Wakely et al., PoS ICRC2023, 118 (2023).
  83. [83]P. Ferrando, W. R. Webber, P. Goret, J. C. Kish, D. A. Schrier, A. Soutoul, and O. Testard, Phys. Rev. C 37, 1490 (1988).
  84. [84]I. V. Moskalenko, S. G. Mashnik, and A. W. Strong, in 27th International Cosmic Ray Conference (2001) arXiv:astro-ph/0106502.
  85. [85]I. V. Moskalenko and S. G. Mashnik, in 28th International Cosmic Ray Conference (2003) pp. 1969–1972, arXiv:astro-ph/0306367.
  86. [86]W. R. Webber, J. C. Kish, and D. A. Schrier, Phys. Rev. C 41, 547 (1990).
  87. [87]W. R. Webber, A. Soutoul, J. C. Kish, J. M. Rockstroh, Y. Cassagnou, R. Legrain, and O. Testard, Phys. Rev. C 58, 3539 (1998).

Paper details

vlnc:W_a6u8cp5z
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents