Fermi National Accelerator Laboratory FERMILAB-Pub-94/010-A _ f_:_:_i:_: January 1994 COSMIC RAY MODELS FOR EARLY GALACTIC LITHIUM, BERYLLIUM, AND BORON PRODUCTION t Brian D. Fields 1, Keith A. Olive 2, and David N. Schramm 1,3 1The University of Chicago, Chicago, IL 60637-1433 2School of Physics and Astronomy, University of Minnesota, Minneapolis, MN 55455 3NASA/Fermilab Astrophysics Center, Fermi National Accelerator Laboratory, Batavia, IL 60510-0500 ABSTRACT To better understand the early galactic production of Li, Be, and B by cosmic ray spallation and fusion reactions, the dependence of these production rates on cosmic ray models and model parameters is examined. The sensitivity of elemental and isotopic production to the cosmic ray pathlength magnitude and energy dependence, source spectrum, spallation kinematics, and cross section uncertainties is studied. Changes in these model features, particularly those features related to confinement, are shown to alter the Be- and B-versus-Fe slopes from a naive quadratic relation. The implications of our results for the diffuse _,-ray background are examined, and the role of chemical evolution and its relation to our results is noted. It is also noted that the unmeasured high energy behavior of a + a fusion can lead to effects as large as a factor of 2 in the resultant yields. Future data should enable Population II Li, Be, and B abundances to constrain cosmic ray models for the early Galaxy. (NASA-CR-197626) COSMIC RAY MCDEL3 N95-I8965 FOE EARLY GALACTIC LITHIUM, BERYLLIUM, AN9 BORON PRCCUCTIGN (Fermi_Nationa] Acce]erator Lab.) Unclas 58 p G3173 0038394 tsubmitted to The Astrophysical Journal .....".. V Operated by Universities Research Association Inc. under contract with the United States Department of Energy 1 Introduction The Population I abundances of eLi, Be, and B (LiBeB) have been thought for some time to have their origin in spallation and fusion processes between Cosmic ray and interstellar medium (ISM) nuclei (see, e.g., Reeves, Folwer, & Hoyle 1970; Meneguzzi, Audouze, & Reeves 1971; Walker, Mathews, & Viola 1985). In the past few years these elements have been sought extreme Population II dwarfs, stars well known to exhibit the "Spite plateau" in lithium (Spite & Spite 1982), which is understood to indicate the primordial 7Li abundance (Walker, Steigman, Schramm, Olive, &Kang 1991). Recently these same Pop II stars have also been shown to additionally contain Be, (Rebolo et ai. 1988a; Ryan et al. 1990, 1992; Gilmore et al. 1992a, 1992b; Boesgaard & King 1993) B, (Duncan, Lambert, & Lemke 1992) and most recently 6Li (Smith, Lambert, & Nissen 1992). These abundances provide important clues about the early galaxy, and cosmic rays have been considered the most likely production mechanism for LiBeB in these stars as well (Steigman & Walker 1992 (SW); Prantzos, Cass_, & Vangioni-Flam 1992 (PCV); Walker et al. 1993 (WSSOF); Steigman et al. 1993 (SFOSW)). If indeed the extreme population II LiBeB arise from cosmic ray interactions, the study of their isotopic abundances opens an important window on astrophysics. In principle, we may be able to gain insight on early cosmic rays, as well as early star formation rates and chemical evolution (e.g. PCV; Fields, Schramm, &_Truran 1993; Silk gz Schramm 1992). In addition, associated with these early cosmic ray events is an appreciable gamma-ray flux which would contribute (perhaps significantly) to the present diffuse gamma-ray background (Silk & Schramm 1992; Prantzos & Cassd 1993). The early cosmic ray scenario is also of importance to cosmology. One may use the results of a galactic cosmic-ray (GCR) spallation and fusion model to help infer primordial 7Li from the observed Li abundance (Olive & Schramm 1993). One may also use the GCR model's cosmic ray spallation information on the Li isotopes, compared with their observed abundances, to deduce the amount of possible stellar depletion (SFOSW). Our previous work (WS, WSSOF, SFOSW) was an attempt at a relatively model- independent approach. Without assuming a specific model (of cosmic ray or galactic chemical evolution) we concentrated on testing the consistency of standard GCR models with the recent LiBeB observations and with primordial nucleosynthesis where possible by examining elemental and isotopic abundance ratios. In another approach, Gilmore et al. (1992b), Feltzing & Gustaffson (1993), Prantzos (1993), Prantzos & Cassd (1993), Pagel (1993), and particularly PCV have chosen to adopt a particular detailed model of early galactic chemical and cosmic ray evolution to examine its predictions and compare with observations. In this paper we will examine in detail the dependence of the LiBeB abundances produced by GCR nucleosynthesis on the uncertainties of, or allowed variations in, the cosmic-ray model. As such, we will discuss the uncertainties in cosmic ray models. After a brief review of the data, we present some model options in section 3. Our assumptions :ili!i:: :_i_Ii:,I/ regardingthe evolution of the abundancesof a, C, N, and O are discussed in section 4. We address the issue of the Be and B slopes versus [Fe/H] in section 5. The cosmic- ray spectrum and the confinement of cosmic rays will be discussed in section 6. We also explore implications of these models on v-ray production in section 7. We draw conclusions in section 8. 2 LiBeB Abundance Data To make this work as self-contained as possible, we show in table I LiBeB isotopic abundances observed in Pop II halo dwarf stars. We list those stars in which at least two light elements have been observed as we will for the most part be primarily interested in elemental or, even better, isotopic ratios. The notation we employ below and use throughout the paper is that [X/H] represents the log abundance relative to the solar abundance, namely log(X/H) - log(X/H)® and [X] = 12 + log(X/H). In the table, the iron abundance represents an unweighted "world" average. For the other abundances, a weighted average is given. The 6Li abundance was taken from Smith et al. (1993). The 7Li abundances were taken from Spite K: Spite (1982,1986); Spite et al. (1984); Hobbs and Duncan (1987); Rebolo et al. (1988b); Hobbs & Thorburn (1991); and Pilachowski et al. (1993). The 9Be abundances were taken from Rebolo et al. (1988a); Ryan et al. (1992); Gilmore et al. (1992a,b); Molaro, Castelli, & Pasquini (1993); and Boesgaard & King (1993). Finally, the boron abundances were taken from Duncan et al. (1992). The ratios of 6Li to rLi, Li to Be and of B to Be are the observed ratios. Because, for Pop II, the dominant contribution to the rLi abundance comes from primordial nucleosynthesis rather than GCR nucleosynthesis, a certain degree of caution is necessary when comparing the first two of these ratios to the predictions we discuss below. On the other hand because there is no appreciable big bang source for either Be or B (Thomas et al., 1993), the ratio of these two may be compared (unless there is an additional primary source for nB as discussed in Dearborn et al. (1988), Woosley et al. (1990), and Olive et al. (1993)). Because of the importance of the B/Be ratio, we note that the values given in the table for Be and the ratio B/Be, are averages over observations by several groups. These observations themselves show some spread which may be significant. For the star HD19445, beryllium upper limits were obtained by Rebolo et al. (1988a), giving [Be] < 0.3, and B/Be > L3, Ryan et al. (1990) found the upper limit [Be] < -0.3 and hence B/Be > 5 which should be compared with the value of 3.5 in the table which represents the only positive identification of Be in this star by Boesgaard and King (1993) and is slightly discrepant with the upper limit of Ryan et al. For HD140283, we have measurements of Be by three groups: [Be] = -1.25 5=0.4 from Ryan et al. (1992) giving B/Be = 14 5= 14; [Be] = -0.97 5=0.25 from Gilmore et al. (1992) giving B/Be = 7 5=5; [Be] < -0.90 from Molaro, Castelli, & Pasquini (1993), giving B/Be > 6; and iii:: i/ [Be] = -0.78 + 0.14 from Boesgaard and King (1993) giving B/Be = 5 -4-3. Finally for HD201891, [Be] = 0.4 + 0.4 from Rebolo et al. (1988a) giving B/Be = 20 =t=26 and [Be] = 0.67 =t=0.1 from Boesgaard and King (1993) giving B/Be = 11 + 10. As one can see there appears to be a wide range in values (and uncertainties) in this ratio, and the values given my be completely dominated by systematics which are poorly accounted for in the stated error. In particular, ratios are most useful if constant surface tempera- ture and surface gravity assumptions are used in the specific elemental determinations. Unfortunately, this is not yet the case. Indeed it is the uncertainties in these measurements and in the averages of these measurements which are themselves uncertain. Namely, the treatment of systematic errors is far from systematic. The conversion of line strengthes to abundances involves inferences on the surface temperature and surface gravity of the star. Most observational determinations have been made using differents sets of inputs. Though one can ascribe some uncertainty to chosen values of these inputs, it is not always clear to what extent these systematic errors have been incorporated into the quoted so-called statistical error, and different authors make divergent assumptions on the uncertainty of their assumed stellar parameters. Furthermore any average will surely underestimate the true error because of the mistreatment of systematic errors. Thus it is our feeling that B/Be and Li/Be ratios are extremely uncertain. TABLE 1. OBSERVED POP II ABUNDANCES OF LIBEB ISOTOPES STAR [Fe/H] Li [Be] [B] _Li/TLi Li/Be* B/Be* ttD16031 -1.9 2.03 4-0.2 -0.37 4-0.25 251 4- 185 HD19445 -2.1 2.074-0.07 -0.144-0.1 0.44-0.2 1624-46 3.54-1.8 HD84937 -2.2 2.114- 0.07 -0.85 4-0.19 0.05 4-0.02 912 4-425 HD94028 -1.6 2.10 4-0.08 0.44 4-0.1 464- 13 HD132475 -1.6 2.05 4-0.09 0.60 4-0.3 284-20 HD134169 -1.2 2.20 4-0.08 0.71-4- 0.11 31+ 10 HD140283 -2.6 2.08 4-0.06 -0.87 4-0.11 -0.1 4- 0.2 891 4-262 64- 3 HD160617 -1,9 2.22 4- 0.12 -0.47 4-0.18 4904- 244 ttD189558 -1.3 2.014-0.12 0.894-0.33 134-11 HD194598 -1.4 2.00 4-0.2 0.37 4-0.12 434-23 HD201891 -1.3 1.974-0.07 0.654-0.1 1.74-0.4 21+6 114-11 BD23°3912 -1.5 2.374-0.08 0.304-0.4 1174- 110 *Ratios are extremely uncertain due to inadequate treatment of systematics. Formal errors on ratios are underestimates. 3 Cosmic Ray Models for LiBeB Spallation Pro- duction Currently studied models for LiBeB production are based on the work of Reeves, Fowler, and Hoyle (1970) and the more detailed follow-up work of Meneguzzi, Audouze, and Reeves (1971). They describe cosmic ray propagation via the simple leaky box model. This assumes a spatially homogeneous distribution of sources, cosmic rays, and inter- stellar material. Within this model, the propagation equation is ONA O(bANA) 1 NA (1) Ot - JA + OT r_ss Here NA = NA(t, T), is the number density of cosmic ray isotopes A = LiBeB at time t with energy (per nucleon) between T and T + dT. Note that eq. (1) is the general propagation expression for A a primary or secondary species, with different terms being important for each. The first term on the righthand side, JA, includes all sources for A: JA(T)=QA(T)+_.. nj /T:2 dT¢i'(T)--_-_'-(dTa,AT') . (2) s3 Here QA(T) is a possible galactic source of A, given as a number rate per volume and per unit energy; and spallation production of A appears in the sum runs over projectiles i and targets j. Additionally, ¢i = Nivi is the cosmic ray flux spectrum of species i and, (rA is the cross section for the process i + j _ A +.... The second term of propagation equation (1) energy losses to the ISM, with bA -- :(OT/Ot)A allows for ionization from the Galaxy. The third term of the propagation equation accounts for catastrophic cosmic ray losses, 1 v" o.inel v n 1 -z., ia A i+_ (3) Te/$ i Test with a_.__tencoding spallation losses of A in the ISM, and r_¢(T) being the lifetime for cosmic rays against escape. The propagation equation (1) is solved for the case of a steady state, ON/Ot = O, in which cosmic ray production is in equilibrium with the losses. One thus obtains the spectrum N of these elements, propagated from their source J. At present, we will ignore losses due to inelastic nuclear collisions, valid when Tes-c1 _ _nS.O_iiA,_t_ "t)A (but see section 6.1). We assume that the primary cosmic ray species p, a, and CNO, have some homogeneous galactic source J = Q, and negligible spallation production or losses: a -- 0. Writing the solution for these in terms of the cosmic ray flux ¢ = Nv, we have 1 oo ¢/(T) - wi(T) IT dT'qi(T')exp (- [Ri(T') - Ri(T)] /A) (4) wherewi = bi/PlSMV, and qi = Qi/PISM, and i = p, a, CNO. Also, RA(TA) : fTA dT'/(DT'/cOX)xsM is the ionization range which characterizes the average amount of material a particle with energy TA can travel before ionization losses will stop it (expressed in gcm -2, as X = PXSMVt). The ionization ranges were taken from North- cliffe and Schilling (1970) for low LiBeB energies (< 12 MeV/nucleon) at which partial charge screening effects are important, and from Janni (1982) for higher energies, using the scaling law RA(Z;T) = A/Z _ Rv(T ) . (5) Here T the kinetic energy per nucleon and R v is the proton range. In figure 1 we plot the proton and a fluxes ¢i calculated from eq. (4). To show the effect of energy losses on the propagated flux, we also plot Aqi, the solution of eq. (1) for negligible energy losses (bi _ 0). As we will discuss below in sections 5 and 6.1, these losses are important at low energies and negligible at high ones. This is manifest in figure 1, which shows the scaling ¢ = Aq to be followed closely at high energies, while at low energies the ionization energy losses significantly reduce the propagated flux from this scaling. This behavior is qualitatively similar for the two source types we consider; propagation differences at low energies are discussed in section 6.1. In solving the propagation equation for the secondary elements LiBeB, the source term J is assumed to have no primary source component, i.e. Q = 0. The exclusive production of these elements occurs via spallation between the primaries and the ISM: aA # 0. One can write an expression like eq. (4) for LiBeB, giving their cosmic ray spectrum. What we wish to know, however, is not this steady-state spectrum but instead the amount of LiBeB thermalized and added to the ISM. To compute this from the LiBeB spectrum one assumes that all such nuclei below some threshold kinetic energy Tth_rm are thermalized and then one examines the LiBeB current bANA(Ttherm) below this threshold. Below the lowest spallation threshold there is no source term in eq. (1) for LiBeB, and the ionization loss term is much larger than the escape term. Thus to a good approximation the propagation equation (1) reads o-_bANA(T) = 0 ,T <_Tt_"c (6) and so the LiBeB current bANA(Tth_,-,,) is constant for Tth_rm below spallation production thresholds. One may thus choose any Tth_m < T - tThM at which to evaluate this current; we choose ours to be right at threshold. With this method of computing the production rate of LiBeB via evaluation of the subthreshold LiBeB current, we can write the rate of LiBeB accumulation in the ISM as .+ C 13 where A = _'_Li,_Be, roB,YA nA/nS-s,the @ come from eq. (4),and we have ignored the smalltime variationofnil. • 5 The factor S_(TA, t) accounts for the energy loss of nucleus A in the ISM and gives the probability of its capture and thermalization; it is a function of the lab energy TA of the daughter nucleus A, as well as the epoch t, and is given by SA(TA, t) -=-exp[-- {RA(TA) -- nA(Ttherm)} /A(t)] (8) These energy losses in the ISM serve to trap the spallation products and so to allow them to contribute to the LiBeB finally thermalized in the ISM. This process competes: with the the spallation products' escape from the galaxy, quantified by the escape rate -1 7e_c of eq. (1). This quantity appears in eq. (8) through A = PlSMVTesc the (possibly) energy-dependent average pathlength 1, also in g cm -2. Note that as r_s_ sets the scale for the cosmic ray residence time before escape, A is the amount of matter traversed before escape. As any Tth_r,n <_Tt_h"c gives R(Tth_,_) _:: A, we may put SA _--exp(-RA/A). For "forward" kinematics, with a light cosmic ray nucleus impinging on a stationary, heavy ISM nucleus, TA is small for typical spallation energies, and so RA << A and SA _-- 1. For "inverse" kinematics, with a heavy cosmic ray nucleus on a light ISM particle, SA can differ significantly from unity and the LiBeB yields are reduced accordingly. Following Walker et al. (1992), we put qi(T,t) = y_n(t) qp(T,t) (9) with i = p,a, CNO: i.e., we posit the constancy of the cosmic ray isotopic and elemental ratios at the source and over the entire energy spectrum. We then choose to make the more serious assumption that we may express the proton source strength in the separable form q,(T,t) = qp(T) f(t). (10) We will in fact consider various Population II source spectra q (see section 6.2), but we will not allow a fully general energy or time dependence. While this has the immediate advantage of simplifying the calculations, one may view this assumption as a postulation that the present mechanism of cosmic ray'acceleration does_ not differ dramatically over time in its energetics, but only in its net cosmic ray output. In our analysis of cosmic ray model features, and in the accompanying figures, we will concern ourselves with the effect on the LiBeB production rates--or rather their ratios--as calculated by eq. (7). Given a set of cosmic ray and ISM abundances, and a confinement parameter A, these rates may be evaluated numerically. As in previous work, weused empirical values of partial cross sections tabulated in by Read and Viola (1984). To proceed further and integrate the rates to get the LiBeB yields would require a full model for cosmic ray and chemical evolution. We discuss this issue in the next section. 1By writing the argument of the exponential in SA as we have in eq. (8), we are assuming A to be constant in energy. For further discussion see section 6.1. 6 As a cosmic ray model of propagation, the above amounts to the simple leaky box (reviewed by, e.g., Ceasarsky 1980, 1987). This is the simplest model used to describe cosmic rays today, and as such has obvious computational advantages. While the sim- plicity of the leaky box makes it a useful tool, it is physically unrealistic and thus, in some cases, quite inaccurate. A proper treatment of cosmic ray propagation must ex- plicitly include diffusion effects only sketched with the leaky box confinement parameter A. Also, a proper model must abandon the simple leaky box assumption of spatial ho- mogeneity of sources and interstellar material. Despite these shortcomings, however, we will follow previous authors and adopt this model in what follows, both for the above pragmatic reasons and moreover because the epoch we consider is so poorly understood in its relevant details that adoption of a more detailed description is not warranted at the present time. 4 The Role of Chemical Evolution Although we can calculate the rates in eq. (7), for a given set of parameters, using well-understood physics, to integrate these rates requires knowledge of the chemical and dynamical evolution of the cosmic ray and ISM species in the Galaxy's early history. Such knowledge is sketchy at present, but some reasonable assumptions will prove fruit- ful. If we assume that the H and He abundances do not change much from their values as set by the big bang, we have H(t) _.. H BB 0.0s. (11) If we further assume that the CNO nuclides evolve at the same rate, so that their ratios remain constant within the early epoch considered here, we put yo(t) yN(t) yo(t) yoBs- yoBs- (12) assumed to hold for times t less than the galactic age r at the birth of the Be or B star having CNO abundances denoted here as OBS. One should interpret equation 12 with some care, as approximation of constant C:N:O ratios is not strictly true. We expect these elements to have different sources: O is made in type II supernovae but C and N are made primarily in intermediate mass stars. Thus any differences in the evolution of these sites will result in differences in the abundance ratios. Fortunately, large differences between CN and O production are only important at early times, and the observed Pop II CNO abundances are consistent with constant C:N:O over the timescales important here. One should note, however, that the O/CNFe ratio, though the roughly constant, . does exceed the solar ratio. This comes about because of the predominance at this epoch of type II supernovae over type I's. The effect is important for our purposes, as spallation off of 0 gives a lower B/Be ratio than spallation off of C. We again simplify by extrapolating the present relation between abundances of cos- mic ray and ISM species, namely we assume that the relative enhancement of the cosmic ray CNO abundances over interstellar CNO abundances has remain constant in time. We put = -- (13) yi(t) \ Yi / pr_8_,_t where i = CNO and the righthand side denotes the present value. We will in fact adopt the stronger assumption, made in most work, that YC'(t) = ylSM(t) (14) but we note that this assertion is false for the present cosmic rays. In particular, Asaki- mori et al (1993), and references therein, report that found a H/O depletion of about a factor of 2 relative to solar. The proton-to-helium ratio in the cosmic rays is energy de- pendent and varies from ,_ 0.2 at the lowest measured energies to ,,_0.05 at the highest energies which interest us here. Additionally, as shown in Buckeley et al (1993), He/O in the cosmic rays is depleted by a factor of ,-_ 4.6 relative to its local galactic value. In this paper we have normalized to the protons, and thus adopting the observational results would amount to reducing YH_ by 2.3, and enhancing YCNO by a factor of 2. We will instead use eq. (14) to allow comparison of ours with previous results. Bear in mind, however, that the cosmic ray abundance scalings are uncertain by at least a factor of 2. One proposed model which clearly contradicts this assumption is that which explains the Be/O constancy by having the CNO either as targets or projectiles be at supernova production abundances. We may model this behavior by putting yo .(t)=ySp rno oy r 8on> > 0.5 (15) In what follows, however, wherever not explicitly noted otherwise, eq. (13) will be as- sumed to hold. If we allow the confinement pathlength to vary with time, then as we will see below, we may not further simplify the rate equation (7), or the integral that is its solution. If the confinement is taken to be constant in time---a dubious proposition, as we shall see--we may now separate the propagated flux: ¢(T, t)= f(t)v(T ) . (16) This allows one to write the solution to the rate equation as a sum of terms, each of which has a factor that involves an integral over time and a factor involving an integral over energy: yA( )=E (17) ij . .. with the "exposure time" given by Aij(r) = fO T dt yj(t) y_n(t) f(t) (18) with r is the age of the Galaxy at the birth of the star in question, and the "reduced rate" is (¢a A) =/TA_ c dT ¢,(T,t)aA(T) SA(TA, t) . (19) Note that these factors are different for forward and inverse kinematics, that is, ((I)aA)y ((I)aA)r. As mentioned above, this stems from the longer range and greater chance of escape for the fast A-nuclei produced in inverse kinematic collisions. With the assumptions made thus far in this section--most importantly, that of the time constancy of A, the number of independent exposure times may be reduced to three (sw, WSSOF): Aii = { AA_IS,_M fio=rjw=aard kinematics (20) "-*v,o inverse kinematics Of these, we will take A_,_ as the fiducial quantity, as it measures the integrated flux enhancement exclusively (ie with no explicit dependence on chemical evolution) A_,,_ = dt f(t) (21) We then define the quantity (O/H), which is a measure of the average abundance of oxygen relative to hydrogen,: AISM (O/H)IsM -- "-'_,0 /_ Ot_Ot y?,dtygsM(t)y(t) f_ dt f(t) (22) with a similar expression for (O/H)cn. Note that (O/H) is to the weighted average of O/H at time T. This average weighted by the net flux enhancement f(t), and is normalized. With this notation we have ) YBe,B = Aaa _'_aii )l + (O/H)cn (¢a_'B)r (23) ij ij YLi -- Aac_ (Oa_} + (O/H)zsM _j (oaLi)l + (O/H)cn _' (¢aL')_ (24) _3 9