Mon. Not. R. Astron. Soc. 000, 000-000 (0000) Printed 14 May 2009 (MN WFe^ style file v2.2) Thermal instability with the effect of cosmic-ray diffusion Mohsen Shadmehri* Department of Mathematical Physics, National University Ireland, Co Kildare, Maynooth, Ireland 14 May 2009 ABSTRACT We study dynamical effects of cosmics rays (CRs) on the thermal instability in the linear regime. CRs and the thermal plasma are treated as two different interacting fluids, in which CRs can diffuse along the magnetic field lines. We show that growth rate of the magnetothermal condensation mode is reduced because of the existence of CRs and this stabilizing effect depends on the diffusion coefficient and the ratio of the CRs pressure to gas pressure. Thus, a slower rate of structure formation via thermal instability is predicted when CRs are considered. Key words: ISM: cosmic rays - instabilities - stars: formation 1 INTRODUCTION Birtli of the stars and the complex observable patterns within the interstellar medium (ISM) are mainly understood by the instabilities that may start with small amplitude per- turbations le ading to highly nonlinear processes, i.e. tur- bulen ce (e.g.. Sc alo fc Elmegreenll2004l : iMac Low fc KlessenI |2004| ). In fact, among the most important physical ingredi- ents in any theory of star formation one can mention self- gravity, magnetic field and net cooling function. However, ' there is another significant physical factor that its dynam- ical role has not been considered in most scenarios of star formation until rather recently: Cosmic Rays (CRs). Equipartition exists between energy densi ties of thermal , gas, magnetic field, turbulence and CRs (e.g.. lGaisserlll990l : . lFerrierg|200ll ). The energy equipartition encourages one to ' expect significant dynamical effects of CRs on the structure formation i n ISM as has been realized by Parker in a simpli- ■ fied model (|Parkeil 19661 ). It is believed that giant cloud com- plexes in the sp iral arms of galaxies are forming via Parker instability (e.g.. lMouschoviaall974l : lElmegreen fc ElmegreenI Ll986). More recent studies show that the growth rate of the Parker instability is very sensi tive to the CR pressure and the diffusion coefficient Hanasz 1997: Ha nasz fc Leschllloooi n Kuwabara et al.l Kuznetsov fc Ptuskin 198a 2004 kuwabara fc Kal2"006l ). CRs have also a vital role in propa- gation of MHD waves. ILo fc Kol (|2007l ) studied the the sta- bility of a cosmic-ray plasma system. This stability analysis has been generalized to a four-fluid cosmic-ray-MHD system which comprises magnetized thermal plasma, cos mic rays, forwa rd and backward propagating Alfvn waves (|Ko fc Lol l2009l ). * E-mail: mshadmehri@thphys.nuim.ie (MS); © 0000 RAS Interactions of CRs may al so operate as a global heat- ng m ec hanism in ISM (e.g.. [ Field. Goldsmith fc Habing 19691: iGoldsmith. Habing fc Fieldl Il969l ). Recently, Yusef-Zadeh. Wardle fc Rovl (|2007l ) proposed that due to heating of the gas clouds in the central regions of our Galaxy by an enhanced flux of CR electrons, the rate of star formation in these star forming sites decreases and a similar mechanism may operate i n the nuclei of the other galaxies. In another related study, iFatuzzo. Adams fc Melial (2006) studied the interplay between molecular clouds and the irradiation by CRs produced by supernova remnants. This increased CRs flux has important consequences for star formation. In particular, a higher ionization level due to CRs lead to a longer ambipolar diffusi on time scale an d therefore slower star formation rate (Fa tuzzo et al.l [20061 ). However, none of these studies considered the possible dynamical effects of CRs on the star forming regions. A linear analysis of non-magnetized thermal gas and CR has revealed that CRs suppress the growth of small amplitude pertur b ations jKuznetsov fc PtuskinI 19831: iBegelman fc Zweibel Il994l : IWagner et al.l |2005| ) Kuznetsov fc PtuskinI l|l983D (hereafter KP83) studied sta- bility of ISM consisting of regular and turbulent magnetic flelds, thermal gas and CRs. Although stabilizing effect due to the existence of CRs has been shown, heating-cooling of the system has not been considered by KP83 and a poly- tropic equation of state is used instead of the energy equa- tion for the gas component. Moreover, their analysis resem- bles to the classical Parker instability, in which the initial unperturbed state is determined by the balance of the forces rather than thermal equilibrium stat es as we will consider in our study. Therefore, analysis of iKuznetsov fc PtuskinI 1 19831 ) can not address stability of a system with CRs when the net cooling function is playing a signiflcant role like 2 M. Shadmehri structure forma tion in HI reRions due t o the thermal con- densation (e. IWagner et al .Fukue fc Kamavall2007f ). In another study, derived stabihty condition of a non- magnetized thermal system for a few simplified cases. How- ev er, growth rate of the perturbations has not been studied bv lWagner et al] l|2005l ). In this study, we follow a similar approach but in a magnetized case considering the net cool- ing function of the system and the dynamical effects of CRs. In this regard, our study is different from KP83 who did not take into account thermal behavior of the system and as we will show growth rate of the perturbations and the criteria of stability are completely different in comparison to KP83 . Also, our study is complementary to IWagner et alj (|2005l ) who did not consider magnetic field and the growth rate of the perturbations has not been calculated in detail. It is assumed that CRs particles can diffuse along the magnetic field lines (see also. ISchlickeiser fc Lerch3ll985l ). In the next section basic equations of a magnetothermal system includ- ing CRs are presented. A dispersion relation is obtained for the linear perturbations in section [31 Analysis of the the growth rate of the condensation modes is done in section |4l We conclude by a summary of the results and possible implications in the final section. 2 GENERAL FORMULATION There are three different approaches to study the dynam- ics of CRs. In the particle-particle approach, the plasma and CRs are considered as particles that may interact with each other via complicated processes. In a simpler approach, known as fluid-particle, the plasma is treated as a fluid, though CRs are still described as particles. The simplest approach is the fluid-fluid approach in which CRs and the thermal gas are described by different interacting fluids. The hydrodynamic approach can not provide the spectrum of CRs, however, it is a good approxima tion to use to analyze dynamics o f a plasma with CRs (e.g., iDrurv fc Voelk|[l98ll : iDrurvl 19831 ). We adopt the hydrodynamic approach to study effects of CRs on the thermal instability; We also follow the same steps as in lFieldl l| 19651 ) , except that CRs are considered as a fluid, and diffusion is consid- ered only along magnetic field lines. For simplicity we neglect ionization and heating by CRs, since their effects in the ab- sence of the dynamical role of CRs are well understood (e.g., [Goldsmith et al.lll969l ). The basic equations are dp „ It = P^ = ^(VxB)xB-V(p + p„), dB 'dt V.B = (1) (2) (3) (4) 1 ^^^^^^ J — 1 at 7 — 1 p di and = V X (v X B) where d/dt — d/dt + v.'V is the Lagrangian time derivative, r = -K||b(b.Vpcr), (7) is the diffusive fiux of cosmic-ray energy and K|| is diffusion coefficient along magnetic field lines. All the variables have their usual meaning. Also, b is a unit vector along the mag- netic field lines, i.e. b — B/_B. The adiabatic indices of the thermal gas and cosmic rays are denoted by 7 and 7cr, re- spectively. Also, Q represents the energy losses minus energy gains per unit mass. The coefficient of thermal conductivity K has the values K\\ and K± in directions parallel to and perpendicular to the magnetic field B. Finally, we can write equation of state as R —I -pT, (8) 1 dpcr 7cr Per dp 7cr — 1 dt 7cr — 1 P rff v.r = 0, (6) where R is the gas constant and fi represents the molecular weight. The energy equation of CRs in its complete form in- cludes also an extra term correspo nding to an effective CRs e nergy loss (see equation (13) in ISchlickeiser fc Lerchd (1985')). But this term is neglected in our analysis simply based on a time scale argument. We neglect the ionization energy losses of the cosmic rays because these are surely on a much longer time scale than we are interested. In fact, CR pressure is mainly from mildly relativistic protons, and that these have very low ionization energy loss rates and also low nuclear collision rates. Therefore, typically the ion- ization and nuclear loss time scales are around 10* or more years for normal interstellar values. But typical cooling time scale is around 10® years. Also, there is an effect due to second-order Fermi ac celeration of cosmic r a ys ( s ee equa- tions (15) and (13) in ISchlickeiser fc Lerchel |l985l )). which in fact would turn the cosmic ray loss time into a cosmic ray gain time. This effect is not so easy to discard, as this acceleration is caused by the same MHD waves that con- trol the spatial di ffusion of cosmic rays. In fact it is known l|Schlickeiseilll989l ) that the product of spatial diffusion time To and the acceleration time scale rp equals a constant that depends on the spectral index of the turbulence spectrum and (L/va)'^, where L denotes the size of the system and is the Alfven speed (see equation (86) in lSchlickeiseij (|l989l )). The time scales ar e related approximately totf = (L/va)^ (jSchlickeisej Il989l ) , where the spatial diffusion time scale is td ~ L^/ii. For a typical cooling ISM we can assume L = 10^" cm. Bo = lO"'' G and the ion density is n, = 10"^ cm~^. So, we obtain td = 10^^ s and tf = 7.2 x lO^'' s and so, rp S> td which means we are indeed in a parameter range to neglect the term corresponding to an effective CRs energy loss. Thus, it seems our approach is fine. We actually have no source term either, but we are interested in the effect of a pre-existing and stable CR population on the magne- tothermal condensation modes. To introduce additional time scales related to sources and sinks is a nonsense unless they are comparable to the time scales we are studying. Note that in the hydrodynamic approach of CR, the energy spectrum of the particles of CR is not considered in detail. Moreover, type of the CR particles is not specified in this analysis and they can consist of electrons or protons in any combination. In comp arison to th e classical magne- tothermal instability analysis (|Fieldlll965h . there is an extra © 0000 RAS, MNRAS 000, 000-000 Thermal instability with the effect of cosmic-ray diffusion 3 pressure term due to the CRs in the equation of motion ([2]) . So, one may expect an stabilization effect of CRs. But equa- tion ^ describes advection of CRs by flowing gas and the diffusion of CRs along the magnetic field lines through the thermal gas. Little is known about the diffusion coefficient K|l and its possible depend ence on the physical variabl es of the system. As explained in lSchlickeiser fc Lerchd (|l985l ) the parallel spatial cosmic ray diffusion coefficient k\\ in a cos- mic ray fiuid theory is an effective diffusion coefficient aver- aged over all cosmic ray momenta. Depending on the actual momentum distribution function of cosmic rays and the mo- mentum dependent cosmic ray diffusion coefficient, its value can be ve ry different from the standard value 10'^^ cm^ s~^ (see also, teerezinskn et al.iri990l : |Ptuskinll200ll '). 3 LINEAR PERTURBATIONS We completely neglecting gradients in the background medium and the velocity is zero in equilibrium state po, To, Po, PcrQ. Also, we assume Q,{po,To) — 0. By perturbing of the form 5X{r,t) = (5Xexp(aji -I- ik ■ r), the equations ([T])-® become ujSp + jpok.<5v = 0, (9) a;po5v + ik5p + ik5pcr + i(Bo.5B)^ -i(k.Bo)^ = 0, (10) 47r 47r ujS'B + iBo(k.5v) - i(k.Bo)5v = 0, -Sp ^ 7-1 (7 - l)po Sp + poVlpSp + poVItST (11) (12) LU I^7crPcr0 c- , (k.Bo)^ , , "''Per ^ 77 lT—"P+'^\\ OPcr = 0, (13) (7cr - 1)P0 5p 5p ST (14) po po To Note that the derivative flp = {dQ/dp)T and fir = {dil/dT)p are evaluated for the equilibrium state. Then, we introduce the coordinate system e^, Cy, and fiz specified by Bo Bo X k Gx — Gy X Gz (15) Bo ' " |Bo X k| Also, we introduce the following wavenumbers fcp = A*(7 " l)po^p{RcsToy^ , kr = p.{-y - l)flT{Rcs)~^ , '^ify = [m(7 - '^)Kl\^'^{RCsPo), kii^ = [p(7 - l)Ki]-\RcsPo). (16) Now, we can write the dispersion equation using the follow- ing non-dimensional quantities, kr k k V r = — (Jp = Also, possible effects of CRs diffusion are appeared in our final dispersion relation through the non-dimensional parameters (j) a^nd tp as , _ Per-O po (17) / / ^ rvn (18) where Vo = (icr ~ i)kp{cs/ k^^)^^ . Defining a new wavenum- ber fcc — (7cr — 1)~^{cs/k\]), then we have tpo = kp/kc- Also, this parameter can be re-written as i/)o ~ f^i'Jcr — l)ft:||/CsTcooi, where TcooI is the cooling time-scale. Obviously, the non-dimensional parameter i/jq shows the level of diffu- sion of CRs. When there is no diffusion and the thermal gas and CRs are well coupled together, we have i/) = 0. But as this parameter increases CRs are diffusing more along the magnetic field lines, the level of the coupling becomes weaker. In our analysis, these non-dimensional parameters are the key input parameters to explore possible effects of CRs on the thermal instability. Therefore, the characteristic equation becomes + (o-T +(TK + IpO^'" + blKicTT + CTif ) + 1 + a +lcr 1 or varies as 0.1, 0.4 and 0.9 and also we assume a = 1, 7 = 5/3 and 7c,- = 4/3. © 0000 RAS, MNRAS 000, 000-000 4 M. Shadmehri © 0000 RAS, MNRAS 000, 000-000 Thermal instability with the effect of cosmic-ray diffusion 5 1 2 3 4 5 6 7 k/k 1 2 3 4 5 6 7 k/k Figure 2. Growth rate of thermal condensation mode versus wavenumber of the perturbations when a = 1, 7 = 5/3, 7cr = 4/3, tpo = 0.1, = 0.6, (7t/<^p = 1/2 and rjpaK = 0.01. Each curve is labeled by ipo- © 0000 RAS, MNRAS 000, 000-000 6 M. Shadmehri Top, middle and bottom plots of Fig.[l]are corresponding to ^ = 0.1, 0.01 and 1, respectively. Light gray regions in this Figure are domains of the stability. In order to make easier comparison, the area between two dashed lines corresponds to the domain of the stability in a case without CRs. We see that domains of the stability extend due to the existence of CRs, in particular, for nearly transverse perturbations this extension is very significant. There are stable regions for nearly perpendicular perturbations, for which parallel perturbations are not stable. But in the absence of CRs, domains of stability are the same for all the cases. There- fore, existence of CRs may lead to the formation of highly elongated or even filamentary structures. As we mentioned for transverse perturbations, i.e. = 0, we can present a simple analytical criteria for stability of the system. In this case, the characteristic equation p9p reduces to a third degree algebraic equation for which one can simply do Hurwitz analysis. Criteria of the stability is the following inequality 1 + 7cr0 + (20) Obviously, in the absence of CRs and magnetic field the above c ondition red uces to the classical thermal stability criteria (|Fieldlll965h . Now, we study condensation modes numerically. We take parameters a = 1, ar l from 0.1 to 0.6. In the case of no CRs, the growth rate is represented by a dashed curve. Top plot is for perturbations that are nearly perpendicular to the initial magnetic field line, i.e. C, — 0.01. But the bottom plot shows growth rate for the cases with the perturbations along the initial magnetic field line, i.e. C = 1- Note that each curve is labelled by corresponding ratio (j>. Reduced growth rates are seen in Fig.[2]due to the existence of CRs, irrespective of the direction of the perturbations. Also, reduction of the growth rate is larger as the ratio of the unperturbed CRs pressure to gas pressure increases. This stabilizing effect of CRs can be easily understood by noting to the fact that CRs pressure is acting as an extra support to help the gas pressure against the unstable perturbations in the momentum equation. Comparing to a case without CRs, for a fixed ratio (j>, the reduction to the growth rate of the nearly transverse perturbations is more than a case with perturbations along the magnetic field line according to Fig. [2] In our analysis, CRs particles are diffusing along the magnetic field lines. Thus, diffusion of CRs is more or less negligible for nearly transverse perturbations which implies a maximum pressure support by CRs in the momentum equation. The profile of the growth rate of condensation mode reaches to its largest value for a wavenumber fcmax which depends on the input parameters. Fig. [2] shows that in the absence of CRs as the direction of the perturbations tends to be along the magnetic field line, the maximum wavenum- ber fcniax decreases. But when dynamics of CRs is included, the wavenumber fcmax depends on the ratio and the non- dimensional diffusion coefficient ■0o. For nearly transverse perturbations. Fig. [2] shows that the wavenumber fcmax de- creases when the ratio (p increases. But this behavior is re- versed when perturbations are along the magnetic field lines, i.e. fcniax increases with the ratio (j>- Fig. [3] shows the effect of CRs diffusion along the mag- netic field lines. All the input parameter are the same as Fig. [2I except for ipo which varies from to 100 and the ratio is kept to be fixed, i.e. (j) ~ 0.6. Growth rate of the condensation mode increases with the diffusion of CRs. This destabilizing effect of diffusion is enhanced for the per- turbations along the magnetic field lines. For this type of perturbation the maximum wavenumber fcmax is decreasing as more CRs particles are diffusing. Note that when diffu- sion of CRs is efficient, the growth rate tends to the profile of a case without CRs. This behavior is more evident for the perturbations along the magnetic field lines when ^/jq is large according to Fig. [3] In other words, compression along the magnetic field lines due to the magnetothermal instability is stronger when diffusion of CRs are considered. 5 DISCUSSION Interstellar CRs can impinge on the structure formation in interstellar medium either by contributing to ionization or constituting as an extra source of heating or even through dynamical coupling to the plasma. In this study, we ana- lyzed dynamical effects of CRs on the unstable modes in magnetothermal systems. CRs and the plasma are consid- ered as two different fluids and our linear analysis implies a stabilizing effect due to the existence of CRs. Since CRs are diffusing along the magnetic field line, the level of stabiliza- tion of the condensation modes decreases, in particular for the perturbations along the magnetic field lines. In fact, in the classical analysis of magnetothermal instability one can show that a purely tr ansverse m agnetic field can prevent thermal condensation (jFieldl l 1965*1 . On the other hand, CRs can diffuse along the magnetic field line and not perpendicu- lar to it. Therefore, the magnetic and CRs pressures add up to the thermal pressure for transverse perturbations where diffusion of CRs is negligible. This implies a more magne- tothermally stable system. However, CRs pressure reduces for the perturbations along the magnetic field line because of the diffusion of CRs particles. Since the stabilization ef- fect of CRs is stronger for the transverse perturbations, we may expect formation of elongated clouds via thermal in- stability in the presence of CRs. Effectiveness of CRs on the reduction of the growth rate is anisotropic and CRs diffusion can be a key factor in the final alignment of the elongated clouds. We can also compare our analysis with similar previous studies like KP83. Although the main goal of KP83 is about possible role of CRs in ISM, in comparison to our study there are significant differences that we summarize them here: (a) The present paper studies thermal instability, and so, the net cooling function is considered. But cooling function is completely neglected in KP83 and the same for the ther- mal conduction. Adiabatic variations of the gas component is considered in KP83 instead of the complete form of the energy equation (see, eqs. 22 and 23 of KP83). That is a significant difference. Because in our analysis thermal insta- © 0000 RAS, MNRAS 000, 000-000 Thermal instability with the effect of cosmic-ray diffusion 7 bility is occurring on cooling time scale, but in KP83 there is not thermal effects due to the cooling of the system. So, KP83 can not address thermal condensation modes with CRs in a typical ISM (such as HI regions) which may have a vital role in theory of star formation. But our analysis discusses about thermal condensation modes including dynamical role of CRs. Therefore, KP83 and the current paper are address- ing somewhat different systems, but leading to a consistent picture for the role of CRs. KP83 showed that CRs oper- ate as an stabilizing factor in Parker instability. Our paper shows that CRs play the same role but within the context of thermal instability in ISM even at scales comparable to HI regions. (b) In the light of the above point, our initial state is also different. While we start with a homogenous initial config- uration, KP83 starts with an initial configuration in which the gas is supported in the vertical direction by both grav- ity and magnetic and thermal pressures. Then, their initial configuration depends on the vertical spatial coordinate (see Figure 1 of KP83). This initial set up is similar to the classi- cal Parker analysis. In our analysis, self-gravity is neglected. Because we are interested in formation of structures due to thermal instability in systems where self-gravity has a neg- ligible role (like HI regions). (c) Our initial state is actually corresponding to a thermal equilibrium configuration for which the net cooling function is zero. But initial state of KP83 is defined by the balance of different forces in the vertical direction. (d) Considering the above points, we determined domains of the stability in Fig. [T] numerically (see also equation (|20p '). Since we have considered the complete form of the energy equation including the net cooling function, the stability re- gions are mainly determined by the net cooling function and the parameters relating to the CRs. We show that when the net cooling of the system is included the stability criteria is not a trivial issue. But since KP83 did not take into account the energy equation for the gas component, their stability criteria has been written in terms of a critical polytropic ex- ponent. Therefore, one can not use their stability condition for a system with a given net cooling function as we have in our study. We showed that the level of diffusion of CRs particles is determined by the non-dimensional coefficient t/jq- But de- pendence of this parameter on the physical parameters of the system is a complicated function of the net cooling function. Having a fixe d value for the diffusion coefficient k \\ around 10^* cm^ s"^ l|Berezinskn et al.lll990l : |Ptuskinll200 j) . our key parameter ^/jq depends on the physical variables that control the rate of cooling such as temperature and ionization frac- tion. For simplified cooling systems, we can derive depen- dence of -00 on the param e ters o f the system. For instance, ISchwarz. McCrav fc SteinI (|l972l ) studied formation forma- tion of clouds via thermal instability in ISM. They showed that the cooling time scale of a low-density plasma of cosmic abundance can be approximated as TcooI ~ T/nx for tem- peratures within the range of 100 K to 10* K. Here, n is the number density of particles and x is the fraction density of the electrons, i.e. x — Ue/n. Then, we have ijjo oc Kypx/T^. Therefore, in cold systems, one may expect an efficient dif- fusion of CRs along the magnetic field lines which implies negligible dynamical effect of CRs on the condensation mode (see Fig. [3]). But as the temperature of the system increases, the level of diffusion decreases. Note that K|| is assumed to be fixed and independent of the properties of the system in the above argument. However, there are some points that make the problem more complicated. An unstable magne- tothermal system may become turbulent at nonlinear regime of the evolution. Then, diffusion of CRs particle may be af- fected by the level of turbulence within the system. In the light of our results, we think, the formation of structures in the protogalactic halo environment can be re- analyzed by considering dynamical role of CRs. For exam- ple, B^k et aL (2006;) found that dense clumps first form out of hot background gas by thermal instability in the pro- togalactic environment. For such systems, the cooling time scale and the sound spee d are estimated a s t^ooI = 2 x lO'^ yr and Cs = 198 km s"^ (|Baek et all 120061 ). Then, the non- dimensional parameter diffusion parameter becomes Vo ~ 0.04. This implies that the diffusion of CRs along the mag- netic field line is going to be negligible and the growth rate of condensation mode is significantly reduced depending on the ratio parameter 0. Therefore, the rate of the structure formation in the protogalactic halo environment via thermal instability is reduced because of the existence of CRs. This encouraging result can be studied further via numerical sim- ulations of thermal instability with CRs for the protogalactic systems. In another study, iKim. Kim fc Ostrikeij l|2008t ) studied galactic spiral shock waves with thermal instability. They showed that initially uniform gas rapidly separates into warm and cold phases as a result of thermal instability and also for ms a quasi - stead y shock that prompts phase tran- sitions (|Kim et al.l |2008| ). Although existence of CRs and th eir stabilizing effe ct may not change qualitative scenario of iKim et al.] (|2008| ). we can expect the initial gaseous disc evolves into warm and cold phases with much slower rate comparing to a similar system without CRs. This is an in- teresting research topic for future. ACKNOWLEDGMENTS I thank the anonymous referee for comments leading to a greater clarity of the paper. I am grateful to Luke Drury and Tom Hartquist for their useful comments and suggestions. REFERENCES Back C. H., Ryu D., Kang H., Kim J., 2006, ApJ, 643, L83 Begelman M. C, Zweibel E. C, 1994, ApJ, 431, 689 Berezinskii V. S., Bulanov S. V., Dogiel V. A., Ptuskin V. S., 1990, Astrophysics of cosmic rays. Amsterdam: North-Holland, 1990, edited by Ginzburg, V.L. Drury L. O., 1983, Reports on Progress in Physics, 46, 973 Drury L. O., Voelk J. H., 1981, ApJ, 248, 344 Elmegreen B. G, Elmegreen D. M., 1986, ApJ, 311, 554 Fatuzzo M., Adams F. C, Melia F., 2006, ApJ, 653, L49 Ferriere K. M., 2001, Reviews of Modern Physics, 73, 1031 Field G. B., 1965, ApJ, 142, 531 © 0000 RAS, MNRAS 000, 000-000 8 M. Shadmehri Field G. B., Goldsmith D. W., Habing H. J., 1969, ApJ, 155, L149 Fukue T., Kamaya H., 2007, ApJ, 669, 363 Gaisser T. K., 1990, Cosmic rays and particle physics. Cam- bridge and New York, Cambridge University Press, 1990, 292 p. Goldsmith D. W., Habing H. J., Field G. B., 1969, ApJ, 158, 173 Hanasz M., 1997, A&A, 327, 813 Hanasz M., Lesch H., 2000, ApJ, 543, 235 Kim C.-G., Kim W.-T., Ostriker E. C, 2008, ApJ, 681, 1148 Ko C.-M., Lo Y.-Y., 2009, ApJ, 691, 1587 Kuwabara T., Ko C.-M., 2006, ApJ, 636, 290 Kuwabara T., Nakamura K., Ko C. M., 2004, ApJ, 607, 828 Kuznetsov V. D., Ptuskin V. S., 1983, A&SS, 94, 5 Lo Y.-Y., Ko C.-M., 2007, A&A, 469, 829 Mac Low M.-M., Klessen R. S., 2004, Reviews of Modern Physics, 76, 125 Mouschovias T. C, 1974, ApJ, 192, 37 Parker E. N., 1966, ApJ, 145, 811 Ptuskin V. S., 2001, Space Science Reviews, 99, 281 Scalo J., Elmcgrccn B. G., 2004, ARA&A, 42, 275 Schlickeiser R., 1989, ApJ, 336, 243 Schlickeiser R., Lerche I., 1985, A&A, 151, 151 Schwarz J., McCray R., Stein R. F., 1972, ApJ, 175, 673 Wagner A. Y., Falle S. A. E. G., Hartquist T. W., Pittard J. M., 2005, A&A, 430, 567 Yusef-Zadeh F., Wardle M., Roy S., 2007, ApJ, 665, L123 © 0000 RAS, MNRAS 000, 000-000