INVERSE BREMSSTRAHLUNG IN SHOCKED ASTROPHYSICAL PLASMAS Matthew G. Baring and Frank C. Jones Code 661, Laboratory for High Energy Astrophysics NASA Goddard Space Flight Center Greenbelt, MD 20771 National Aeronautics and Space Administration Goddard Space Flight Center Greenbelt, Maryland 20771 t INVERSE BREMSSTRAHLUNG IN SHOCKED ASTROPHYSICAL PLASMAS Matthew G. Baring and Frank C. Jones Laboratory for High Energy Astrophysics NASA Goddard Space Flight Center Greenbelt, MD 20771, U.S.A. Donald C. Ellison Department of Physics North Carolina State University Raleigh, NC 27695, U.S.A. Accepted for publication in the Astrophysical Journal, Vol 528, January 2000. _ 2 ASTROPHYSICAL JOURNAL_ TO APPEAR ]NJANUARY 10_ _000 ISSUE, VOL. 528 Preprint typeset using I_'I_X style apjgalley INVERSE BREMSSTRAHLUNG IN SHOCKED ASTROPHYSICAL PLASMAS MATTHEW G. BARING t AND FRANK C. JONES Laboratory for High Energy Astrophysics, Code 661, NASA Goddard Space Flight Center, Greenbelt, MD 20771, U.S.A. baringOIheavx.gs.fc.nasa.gov, ]rank.c.jonesOgs]c.nasa.gov AND DONALD C. ELLISON Department of Physics, North Carolina State University, Box 8202, Raleigh NC 27695, U.S.A. don_ellisonOncsu.edu Astrophysical Journal, to appear in January 10, 2000 issue, Vol. 5_8 ABSTRACT There has recently been interest in the role of inverse bremsstrahlung, the emission of photons by fast suprathermal ions in collisions with ambient electrons possessing relatively low velocities, in tenuous plasmas in various astrophysical contexts. This follows a long hiatus in the application of suprathermal ion bremsstrahlung to astrophysical models since the early 1970s. The potential importance of inverse bremsstrahlung relative to normal bremsstrahlung, i.e. where ions are at rest, hinges upon the under- lying velocity distributions of the interacting species. In this paper, we identify the conditions under which the inverse bremsstrahlung emissivity is significant relative to that for normal bremsstrahlung in shocked astrophysical plasmas. We determine that, since both observational and theoretical evidence favors electron temperatures almost comparable to, and certainly not very deficient relative to proton temperatures in shocked plasmas, these environments generally render inverse bremsstrahlung at best a minor contributor to the overall emission. Hence inverse bremsstrahlung can be safely neglected in most models invoking shock acceleration in discrete sources such as supernova remnants. However, on scales _ 100 pc distant from these sources, Coulomb collisional losses can deplete the cosmic ray elec- trons, rendering inverse bremsstrahlung, and perhaps bremsstrahlung from knock-on electrons, possibly detectable. Subject headings: acceleration of particles -- cosmic rays -- supernova remnants -- radiation mechanisms: non-thermal- gamma-rays: theory 1. INTRODUCTION The process of inverse bremsstrahlung has received attention from time to time over the last three decades, in various astrophysical settings. Inverse bremsstrahlung is defined here to be the emission of asingle photon when a high speed ion collides with an electron that is effectively at rest, and has often been referred to as suprathermal proton bremsstrahlung (e.g. Boldt & Serlemitsos 1969; Brown 1970; Jones 1971; Haug 1972). Being a kinematic inverse (an analogy can be drawn with inverse Compton scattering) arising purely via Lorentz transformations between reference frames (i.e. it is still the electron that radiates), it is not to be confused with the true quantum electrodynamical inverse of conventional bremsstrahlung, namely the absorption of photons in 3-body collisions with electrons and ions, a process that is highly improbable in tenuous (i.e. optically thin) astrophysical plasmas. The first application of suprathermal proton bremsstrahiung in astrophysical models dates to the work of Hayakawa & Matsuoka (1964), which considered such radiation in collisions between cosmic rays and ambient electrons in the intergalactic medium as a source of the cosmic X-ray background (XRB) at energies _ 10keV. This concept was developed in the papers by Hayakawa (1969), Boldt & Serlemitsos (1969), and Brown (1970), with Boldt & Serlemitsos proposing that suprathermal proton bremsstrahlung could account for the flat spectral index of the XRB below around 15 keV, while noting that an unusually high column density of ambient electrons would be required to match the observed flux. The interest in the relevance of inverse bremsstrahlung to the cosmic XRB has since diminished, and more recent opinion favours the accumulation of unresolved discrete sources as the dominant contributions (see Fabian & Barcons 1992, for a comprehensive review). Although thermal bremsstrahlung at around 40 keV can fit the 2-100 keV spectrum fairly well (Marshall, et al. 1980; see also Boldt 1987, for a review), recent critical evidence limits the contribution of truly diffuse emission from the intergalactic medium to less than 0.01% of the observed flux. This bound is derived from upper limits to the mean cosmological electron density obtained from the measurements of COBE to Compton scattering-induced distortion of the cosmic microwave background (Wright, et al. 1994). Renewed interest in inverse bremsstrahlung has arisen in the last year. Tatischeff, Ramaty & Kozlovsky (1998) and Dogiel et al. (1998) have discussed expectations for suprathermal proton bremsstrahlung in the X-ray band, assuming a significant enhancement of the density of low energy cosmic ray ions in the Orion region. Inferences of such an abundance of ions were drawn from the reported (e.g. Bloemen et al. 1994), but recently retracted (Bloemen et al. 1999), detection with COMPTEL/CGRO of emission lines from the Orion molecular cloud complex; these were attributed to Carbon and Oxygen nuclear line emission in collisions with ambient ions. In a different context, Valinia and Marshall (1998) conjectured that inverse bremsstrahlung from cosmic rays could be responsible for much of the diffuse X-ray emission seen in the galactic ridge. Pohi (1998) has responded to this proposition by lUniversities Space Research Association 1 2 arguintghatitisdifficulttoproductehediffusceomponeinnttheridgeviainversberemsstrahluwnitghouvtiolatinogbservational limitstonucleaerxcitatiolnineandpiondecacyontinuumemissionF.urthermorTea,tischefRf,amat&y Valinia(1999a)rgue thatbynormalizinthgelowenergcyosmircayionpopulationussingthegalactiBceproductiornate,thecontributioonfinverse bremsstrahtutontghethindiskcomponeonftunres01ved/di1ff0u-s6ek0eVemissiodnetectebdyRXTEisatmosatfewpercent. Thefocusofthepresentatiohnereisoncollisionlesshsockeadstrophysicpalalsmasw,hichdistinguisthhemselvferosmthe diffuseemissiosncenarioasddressbeydValiniaandMarsha(ll1998)P,ohl(1998a)ndTatischeRff,amat&y Valinia(1999b)y theinsignificanocfeCoulomcbollisioncsomparetodinteractionbsetweethnemagnetfiiceldandchargepdarticlesT.heprincipal exampleinsmindhereareyoungsupernorveamnantis.e,.thoseperhapmsatureenougthobeintheSedoevpochb,utwellbefore theradiativpehaseT.hebremsstrahluenmgissiopnropertieosfshockeednvironmeanrtsedictatebdythedissipatiobnetweeionns andelectroninstheshoclkayerW. eshowinSectio2nthat only when electron and ion thermal speeds are comparable (i.e., when the temperature ratio is on the order of, or less than the mass ratio, Te/Tp _ m_/mv )can inverse bremsstrahlung contribute significantly relative to normal electron bremsstrahlung in the optical, X-ray and gamma-ray wavebands. We argue in Subsection 2.4 that such conditions are not easily realized in the majority of shocked astrophysical environments, thereby rendering suprathermal proton bremsstrahlung unimportant for most discrete sources. By extension, the importance of inverse bremsstrahlung as a continuum emission mechanism in extended regions is contingent upon there being a relative paucity of primary cosmic ray electrons in the interstellar medium on appropriate length scales (_ 100 pc). This point, and the role of knock-on electron bremsstrahlung, are discussed in Section 3. 2. INVERSE BREMSSTRAHLUNG VERSUS ELECTRON BREMSSTRAHLUNG The key question of interest to astrophysicists that we address here is when the inverse bremsstrahlung process is significant relative to the normal bremsstrahlung mechanism in shocked astrophysical plasmas. Clearly, the relative numbers of projectile and target particles will be central to answering this question. Here we aim to determine how such numbers and the associated emissivities for the two processes depend on standard plasma parameters. The emissivities, or photon production rates for either process, can be written in the form = nt dpnca(p) c13_ , (i) for shock-accelerated cosmic ray (projectile) particles with momentum distribution ncR(p) colliding with targets of density nt that are effectively at rest in the observer's frame. Here dcr/dw is the cross-section, differential in the photon energy, where we adopt the convention of using dimensionless units throughout this paper: photon energies E_ are expressed in units of the electron rest mass rn, c2, i.e., via w = ET/(m_c _) . Hence &r/dw has c.g.s, units of cm 2. Also, cfl is the speed of the projectile of momentum p. The distribution functions of accelerated particles in shock environs need to be described numerically, in general, as will be evident from the Monte Carlo-generated distributions illustrated later in the paper. However, we can provide a good estimate of the relative importance of inverse bremsstrahlung and classical bremsstrahlung using simple but representative analytic forms for the particle distribution functions. For these purposes, let us approximate the cosmic ray (shock-accelerated) proton and electron distributions in a discrete source by power'laws in momentum, broken at thermal energies. Defining c_3r,v and c_3r,, to be the proton and electron thermal speeds, respectively, their momentum distributions can be cast in the approximate form n, 3(r.- 1) pT,, ' - _ n'(P)-PT'-7 r'-l+3e" e, (p__,)-r,, p> P_., (2) where pT,, = m,c_T., is the thermal (non-relativistic) momentum for species s (= e, p). Here, n, represents the total density and F, is the power-law index of a given projectile species. This discontinuous "schematic" form is chosen to mimic the slopes of the low momentum portion of a MaxwelltBoltzmann distribution and the non-thermal suprathermal portion of the cosmic ray distribution, and hence does not accurately describe the spectral structure near the thermal peak. Consideration of such spectral details, which depend on the nature of dissipation in the shock layer, would introduce only modest corrections to the rates derived below, and are immaterial to the qualitative conclusions of this paper. The factor e, is introduced to account for the fact that the non-thermal tail does not extrapolate directly from the thermal peak in shock acceleration-generated populations, and its values _ 1 define the efficiency of the acceleration mechanism, which can be a strong function of the field obliquity (e.g. Baring, Ellison & Jones 1993) and the strength of particle scattering (e.g. Ellison, Baring & Jones 1996) in the shock environs. For ions, acceleration is generally quite efficient, implying ep _ 0.1. For electrons, the situation is less clear, pertaining to the well-known electron injection problem in non-relativistic shocks. However, as will be mentioned in Subsection 2.4 below, the observed electron-to-proton cosmic ray abundance ratio provides a canonical lower bound to the injection efficiency of electrons, and generally implies that e, cannot be dramatically less than ep. Finally, note that the target density of cool particles for the two processes in question will be written as nt, and will represent a subset of either of the accelerated populations that are formed from the shock heating of the interstellar medium. 2.1. The Bremsstrahlung Ratio in a Nutshell The ratio 7%o-x of inverse bremsstrahlung to bremsstrahlung emission in the optical to X-ray and soft gamma-ray bands can be quickly written down without the encumbrance of the mathematical complexity of the differential cross-sections involved. Here weenunciattheis,theprinciparlesulot fthepapeirnasimpleandenlighteninmganners,othatthedetaileedxpositioonfsthe nexttwosubsectiocnasnbebypassebdythereadeirf,desiredA.tenergiebselowaroundafewhundrekdeV,thewavebanodf interestot thediscussioonfBsoldt&Serlemits(o1s969H),ayakaw(1a969B),rown(1970)T,atischeRff,amat&yKozlovsk(1y998), Dogieeltal.(1998a)ndValiniaandMarsha(l1l 998t)h,enon-relativisdtiicfferenticarloss-sectiinoEnq.(4)belowisoperablaen, d applietsobothbremsstrahluanngdinversberemsstrahluTnhgi.scross-sectiisoanfunctiononlyofthespeeodftheballistipcarticle involvedfo,ragivenphotonenergyH.encea,nydisparityintheemissivitifeosrthetwoprocessceasnonlybeduetodifferences inthenumbeorsfbMlisticprotonasndelectronastagivenspeeda,properttyhatdoesnotextendtorelativistidcomainsS.ince bremsstrahluXn-rgaysinsupernorveamnansthockasregeneralplyroducebdyslightlysuperthermealelctrontsh,isnumbeisrjust nee_//3.r,,, reflecting the efficiency of injection into the shock acceleration process. For protons, whose thermal speeds are much lower, only a small fraction can participate in inverse bremsstrahlung collisions with shocked electrons, namely those with speeds A_rp- l_-Fp exceeding the electron thermal speed /3T,,. Using the suprathermal portion of Eq. (2), this constitutes a "density" npep ,-r,p _r,e , so that it immediately follows that the ratio of the two processes is of the order of T/o-x = _ (_r-_2-_ rp-1 e, \/3T,_/ ' (3) for n_ _ np, as required by charge neutrality. Therein lies the principal result, borne out in the derivations of the next two subsections, which are expounded because of their usefulness in astrophysical problems. Clearly, the ratio R.o-x depends principally on the degree of temperature equilibration, or otherwise, in the shocked plasma. 2.2. Radiation from Non-relativistic Particles A more-detailed estimate of the relative importance of inverse bremsstrahlung and classical bremsstrahlung at optical to X-ray energies can be obtained by considering only non-relativistic accelerated ions, i.e. those with /3 << 1. In this regime, the differential cross-section de/dw for either bremsstrahlung or its inverse can be obtained from textbooks such as Jauch and Rohrlich (1980), and is the non-relativistic specialization of the Bethe-Heitler formula: I 16z2 1 /3+ - .Bn= -'3 w _'log*/3_X//32-2w , 2rrafZ_<<l , (4) where ar = e2/(hc) is the fine structure constant, +Ze is the nuclear charge, and r0 = e2/(m,c 2) is the classical electron radius. Here /3 is the relative speed (in units of c) between the projectile and target, namely the electron speed for classical bremsstrahlung, and the proton (or ion) speed for the inverse process. The applicability of this formula, which is integrated over final electron and photon angles, for both bremsstrahlung and its inverse follows from the Lorentz invariance of the total cross-section _ and the effective invariance of photon angles and energies in non-relativistic transformations between the proton and electron rest frames. With this interpretation, cxf/32 - 2w represents the final electron speed in the proton rest frame for either process. Note that Eq. (4) is derived in the Born or plane-wave approximation, and is strictly not applicable to regimes where /3 _ 2rrafZ or w <</32 << 1 where Coulomb perturbations to the electron wave-functions become important (see the end of this section). Appropriate results for tow frequencies and classical regimes can be found in Gould (1970), and references therein, and amount to modifications of the argument of the logarithm. Using Equation (1), it is straightforward to write down production rates of photons for the two processes in the limit of non- relativistic projectile speeds. For regular bremsstrahlung, the vast majority of shock-accelerated protons are effectively cold due to their generally low thermal speed (except for unusually hot proton components in tenuous plasmas), so that the proton density np represents the target density nt. Hence the photon production rate is t brems _" T /3.,, w I', - 1+ 3e, where F, is the electron power-law spectral index, and 1 f:_ /3+ V"/32- 2w r) = dZ: log,P_ 2,,, = 1 { C/3- )log, /3¢+ __- 2w ---_- (6) /2(_,/3_, r) = for /3c = max {/3T, x'/2-ww} • (7) Here we use the argument r to represent either re or rp, as the case may be, in the work below. The first integral in Equation (6) specializes to [log_(2/3_/w) - 1]/2 in the limit of /_ >> w. The second integral in Equation (6) is generally expressible in terms ofhypergeomefturnicctionsth,oughsuchamanipulatioisnnotenlighteninSgp.ecializattioonthe f_ >> _v limit yields tractable integrals and the result [log_ (2f_/w) + 2/r]/r. However, remembering that modest corrections to the cross-section in Equation (4) are required (Gould 1970) in this low frequency limit, we restrict use of this specialization to the range f_ _ w. The other interesting limit is when f_ << w << 1, for which only the second integral contributes; the result in this case is f2(o;, fiT, F) _ 2F-2 B(r, I'), where B(x, x) is the beta function. The substitutions fl = _ cosh 8, and identity 3.512.1 in Gradshteyn and Ryzhik (1980) facilitate these developments. The limiting cases can then be summarized as 1 f/l_ e_ 2flC,e _ 2ce} dn_(_) _ 16 r_-i _-_.t'_+r_j'°g, _ I+F_ [ , ,_£#_,_<<1, dt brems fT--_eF¢---1"¥3¢_ 2r_/2-1 hr, _ (8) 2 where F(x) is the Gamma function, and A = Z2npn,afr_c (9) defines the fundamental scale for the bremsstrahlung emissivity (i.e. at w -_ 1 and fT _ 1). The w-1 behaviour at low energies reflects the convolution of the bremsstrahlung infra-red divergence with a distribution possessing a finite number of particles. The w-(1+r'/2) dependence above electron thermal energies arises because the bremsstrahlung flux spectrum traces the electron energy distribution: Note that Eq. (8) is strictly valid only when F_ > 1, which wiI1 always be the case for convergent electron distributions. A simplified derivation of the form of Eq. (8) is presented in the Appendix. The inverse bremsstrahlung spectrum can be evaluated in a similar manner, but noting that zero contribution arises whenever the proton speed drops below the electron thermal speed; such a domain of phase space corresponds to normal bremsstrahlung. Hence, thermal protons are never sampled unless the proton thermal speed is comparable to that of the electrons. It follows that the emissivity resembles the form in Equation (5) with just an ]'2-type term; extracting the appropriate power-law factor, the inverse bremsstrahlung emissivity can be written drip(w) 16 _ Z 2afr_¢ rp - 1+3% \flT,_/ epf2(w, _,_, Fp) , (10) where Fv is the proton power-law index. The limiting forms of the f2 function are just as for the bremsstrahlung case, so that we quickly arrive at the limits lc{ (ii) drip(w) ,_ 16 2Fp/2-1 rp F2 r 2 dt bins _ rp - 1+ 3ep \_T,_/ % _ f_,e 2_ , flT,e << w << I. The two photon ranges of interest here, namely _vg f_,e << 1 and f_,_ << w << 1, are the same as those for bremsstrahlung. This is due in part to the fact that only protons with speeds above electron thermal speeds will participate in inverse bremsstrahlung interactions. Note that relaxing the _p _ flT,e requirement to bounds on _p greater than _T,e would introduce the appropriate numerical factor in Equation (11). Clearly, the factor [ep/e_](f_.m/fT,_) r_-I roughly defines the ratio of inverse bremsstrahlung to bremsstrahlung emissivities; expectations for its value in shocked plasmas will form the center of the discussion in Section 2.4 below. It is salient to remark at this point that the differential cross-section in Equation (4) that we are using (and also the ultra- relativistic limit in Eq. (14) below of the Bethe-Heitler result) was obtained from QED calculations in the Born approximation. At non-relativistic speeds, the Coulomb potential of the target electron or ion perturbs the projectile electron wave-function sufficiently that the plane-wave approximation breaks down. This occurs when pc x drops below h, where Pe = fmec is the momentum of the energetic electron. Here x is the spatial scale appropriate to the interaction, which is the classical electron radius r0. Hence, so-called Coulomb corrections become necessary when fl _ af = e2/(hc). In such cases, the matrix element for scattering must be determined using Coulomb wave functions. This has been done exactly for non-relativistic electrons colliding with nuclei by Sommerfeld (1931), leading to the application of a simple corrective multiplicative factor (Elwert 1939) = 1-expf-2_rafZ/f: , s= #,/T- (12) 1- exp [- 21rafZ /fl' known as the Sommerfeld-Elwert factor, to the Bethe-Heitler cross-section: (13) Here ]3' is the electron's speed (in units of c) in the nuclear rest frame after collision, and Z is the charge number of the nucleus. Note that a similar factor can be employed for electron-electron collisions (Maxon and Corman 1967, see also Haug 1975). Such factors become important for projectile electron speeds below around c/10 (for proton targets). When flT,e << 27rarZ, the Coulomb correction factor is simply B/_ 2w, and it propagates through all the integrands of the above developments. It is then quickly ascertained that in the w _/?_,_ << 1 limiting cases, the correction factor is close to unity and therefore can be neglected (modest low frequency corrections for w <<fl_ are discussed in Gould 1970). For higher photon energies, the Coulomb corrections should be included, with the appropriate modification to the algebraic manipulations. While the resulting integral is not tractable except in terms of higher order hypergeometric functions, we determined numerically that the corrective factor that should be applied to the fl_, <<w << 1 cases in Equations (8) and (11) increases monotonically from 1.17 to 1.94 as F, or Fp vary between 1to 5, the rep_:esentative range of power-law distribution indices. Since such factors are of the order of unity, we can assert that Coulomb correction factors are only marginally important for bremsstrahlung computations, and will prove immaterial to the considerations and conclusions of this paper; we neglect them in the algebraic expressions hereafter. 2.3. Emission from Ultra-relativistic Species Hard X-rays and gamma-rays from bremsstrahlung and inverse bremsstrahlung are generated by ultra-relativistic electrons and ions. In this regime, a single cross-section cannot be used for the two mechanisms due to the introduction of aberrations in photon angles and modifications of energies in transforming between the two rest frames of the interacting particles. For normal e - p bremsstrahlung, the differential cross-section is the ultra-relativistic specialization of the Bethe-Heitler formula (e.g. see Jauch and Rohrlich 1980), obtained in the Born approximation: da = 4Z 2aft°2 1+ -- loge - 7>>1 (14) 3-5 SH _ 3 ' ' and is not subject to significant Coulomb corrections. Here, 7 is the electron Lorentz factor. This formula can be used to compute contributions to the bremsstrahlung spectrum from mildly relativistic energies upwards. The review paper of Blumenthal and Gould (1970) provides a useful and enlightening discussion of various features and issues of bremsstrahlung from ultra-relativistic particles. From Equation (2), assuming a non-relativistic thermal speed cBT,s for a species s, the Lorentz factor distribution is easily deduced to be n,e, 3(Fs - 1) 7_r. (15) n,(_) = (Z_,,)I_,.' r, - i + 3_, This can be readily folded with the differential cross-section in Equation (14) via an adaption of Equation (I) to yield an emissivity appropriate to the gamma-ray band: dn (w) _, 12Ae_ Fe- 1 t¢(F¢, w) w-r" 1 _w (16) br_r.s (Z_,.)1-'" re :'f¥-3_. ' ' where A is given by equation (9), and _(_, _) - _o0 _"_l,/'34-'3"t+d4t tl__)[loge[2wt( t- 1)]- 1] 3a2+a+4 V 1 1 3ot4+ 2a3"4-15a 2-- 4 Ll°g 2 - -cJ + 1t (17) 3(_+ 1)_(_-1) -\_ a-1 -3 a + where ¢(z) = d[log¢r(x)]/dx isthe logarithmic derivative of the Gammafunction, and C = tb(1) = 0.577215... isEuler's constant. The evaluation of the integral is facilitated by identity 4.253.6 of Gradshteyn and Ryzhik (1980). The _ ---1 values of Equations (8) and (16) are clearly of comparable magnitude. Note also that the bremsstrahlung emissivity traces the electron energy distribution in the relativistic limit. A simple derivation of the form of Eq. (16) is provided in the Appendix. Determining the emissivity for inverse bremsstrahlung in the limit of ultra-relativistic ions is, in principal, more involved, largely because there are no published expressions for the differential cross-section, integrated over the various interaction angles, that are general enough to take the place of the Bethe-Heitler formula. Exact numerical computations in the Born approximation are presented by Haug (1972). Simple analytic formulae for the cross-section were obtained by Jones (1971), who used the well- known Weizs_icker-Williams method (Weizsiicker 1934; Williams 1935) where the process is treated as stationary electrons Compton scattering the virtual photons carried by the proton's electromagnetic field. Such a technique is quite applicable to hard X-ray and soft gamma-ray energies, but becomes erroneous when w >>1 since then the approximation of the Coulomb potential of the proton by a Fourier decomposition into plane waves (i.e. mimicking free photons) breaks down. Notwithstanding, the approximate formulae of Jones (1971) suffice for the purposes of this analysis (as will become evident shortly); the w <<1 limit of his Equation (4) is d_ar_t,s.ww = -1_6-Z: __ log_ 0'6_87 ' 7 >>1 ' w <<i ' (18) where we have included an extra factor of Z 2 to include consideration of ions other than protons. At photon energies 1 _ w << 7, the numerical computations in Haug (1972) indicate that the differential cross-section approximates a w-2 power-law (exceeding the analytic approximation of Jones 1971), before rolling over near the kinematic maximum energy of w _ 3'. Equation (18) is readily integrated over the power-law Lorentz factor distribution for ions that can be obtained from Equation (15): dt breirnnvs _ (fiT,1p6)Ale-Fv, Fp F-V1-+13ev p(Fp) ¢0-r. w _ 1, (19) where _- ,-logo(o.6-8-t) + j_ 1 (20) This approximation to the contribution to the inverse bremsstrahlung emissivity from relativistic ions is still correct, up to a factor of order unity, when w _ 1, and hence the specification of the range _ _ 1 in Equation (19). When w _ 1, the spectrum breaks to assume a _-(i+rp) form, so that the inverse bremsstrahlung spectrum is steeper than its bremsstrahlung counterpart by an index of roughly unity. We note that while the emissivity in Equation (19) can formally exceed that in Equation (11), in practice this never arises due to the distribution index Fv exceeding 2 considerably at near-thermal energies, while Fp _ 2 at relativistic energies. Hence, Equation (11) should be regarded as the appropriate form for optical and X-ray energies. A comparison of Equations (16) and (19) again yields the result that the factor [ep/e,] (/_T,V/f_T,,) rP-1 roughly defines the ratio of inverse bremsstrahlung to bremsstrahlung emissivities at soft gamma-ray energies. It is also pertinent to briefly mention electron-electron bremsstrahlung. Since this process is a quadrupole interaction in aclassical description, it is strongly suppressed for non-relativistic impact speeds relative to electron-ion bremsstrahlung. This is borne out in the differential cross-sections derived by Fedyushin (1952) and Garibyan (1953), which are of the order of d_r/da; ,.. 4afr_/(15¢_) (see, for example, the exposition in the Appendix of Baring et al. 1999), a factor of f12 smaller than the result in Equation (4). However, the e - e bremsstrahlung cross-section for ultra-relativistic impact speeds is necessarily comparable to that for e - p bremsstrahlung (i.e. Equation [14]), as is indicated in the work Of Baier, Fadin _: Khoze (1967i see also the Appendix of Baring et al. 1999, and the discussion in Blumenthal & Gould 1970), who derived limiting forms using the Weizs_icker-Williams method. The similarity of cross-sections for 3' >> 1 implies that the ratio of electron-electron bremsstrahlung to electron-proton bremsstrahlung is of the order of unity for plasmas with fairly normal ionic abundances, a fact that was noted by Baring et al. (1999). Hence, e-e bremsstrahlung can play an influential role in the determination of the importance of inverse bremsstrahlung. 2.4. Expectations for Shock-Heated Environments Collecting all the rates so far assembled, it is now a simple enterprise to assert when inverse bremsstrahlung is significant in astrophysical scenarios. Setting 7Zo_x to be the ratio of Equations (11) to (8) at w ,-_ (flr,_) 2, so as to represent the ratio of inverse bremsstrahlung to bremsstrahlung at optical to X-ray energies, and RMeV to be the ratio of Equations (19) to (16) at w _ 1, representing the inverse bremsstrahlung/bremsstrahlung ratio at soft gamma-ray energies, we arrive at 7eo_x ~ tz (rod _r,-w_ (21) e, _ T,) "_ 7_MeV , expressing the thermal speeds of the two species in terms of the electron and proton temperatures Te and Tp. Similar order-of- magnitude estimates for such ratios can be obtained for other ionic species such as He2+. In Equation (21), it must be emphasized that for the optical/X-ray band ratio, the index Yv refers to the mean index ranging from thermal proton speeds up to thermal electron ones, while for 7_MeV, Yp represents the mean index of protons with 7 >> 1. Hence, given the general upward curvature of spectra for cosmic rays accelerated in non-linear shocks (e.g. Eichler 1984; Jones _ Ellison, 1991), we expect Fp > 2 for the optical/X-ray band considerations and Fp £ 2 for the gamma-ray band. Since the gamma-ray spectrum for inverse bremsstrahlung (see the discussion after Equation [19]) is steeper than that for bremsstrahlung in Equation (16), _mev defines an upper bound to the ratio of emissivities for the two processes in hard gamma-rays, a bound that is actually conservative given the expectation that e - e bremsstrahlung will amplify the bremsstrahlung signal by a factor of order 2 in the gamma-ray band. The ratio ep/ee of efficiency factors is a free parameter for a shock acceleration model. As noted at the beginning of Section 2, while efficient acceleration of protons is generally expected due to strong turbulence local to the shock (e.g. Ellison, Baring gc Jones 1996), so that ep ,,, 0.1, to order of magnitude (which is borne out in observations of interplanetary shocks: see Baring et al. 1997), much less is known about the value of ee, essentially due to a paucity of in situ observations of electron acceleration in the heliosphere. However, this difficult)' can be circumvented in part (at least in astrophysical scenarios) by arguing that cosmic ray data are a strong indicator of the acceleration efficiency in typical cosmic ray sources, principally supernova remnant shocks. The electron-to-proton abundance ratio above 1 GeV (where both species are relativistic) is well-known to be of the order of a few percent (e.g. see Miiller et al. 1995), which constrains the typical acceleration efficiency, assuming that differences in electron and proton propagation in the interstellar magnetic field at these energies are not great. From the forms in Eq. (2), this ratio can be estimated as a function of the temperatures, masses and spectral indices of the two species. For an isothermal plasma with F, - Yp _ 2,one quickly determines that de(1 GeV/c)/np( 1 GeV/c) -.- [ep/ee] (me(cid:0)rap) 1/2, so that the observed abundance ratio at momenta 1 GeV/c is reproduced only if ee is of the same order of magnitude as ep, i.e. injection of electrons into the Fermi mechanism is indeed efficient. Relaxing the isothermal and equal index stipulations yields a range of ep/c_ on either side of unity.