Astroparticle Physics xxx (2013) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Astroparticle Physics

journal homepage: www.elsevier .com/ locate/ast ropart

Understanding galactic cosmic rays

0927-6505/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.astropartphys.2013.01.009

⇑ Corresponding author. Tel.: +44 191 334 35 80; fax: +44 191 334 58 23.E-mail addresses: erlykin@sci.lebedev.ru, a.d.erlykin@durham.ac.uk (A. Erlykin).

Please cite this article in press as: A. Wolfendale, A. Erlykin, Understanding galactic cosmic rays, Astropart. Phys. (2013), http://dx.doi.org/1j.astropartphys.2013.01.009

Arnold Wolfendale a, Anatoly Erlykin b,⇑a Durham University, Durham, UKb Lebedev Institute, Moscow, Russia

a r t i c l e i n f o

Article history:Available online xxxx

Keywords:Cosmic raysGalaxySupernova remnantsFine Structure

a b s t r a c t

The case is made for most cosmic rays having come from galactic sources. ‘Structure’, i.e. a lack ofsmoothness in the energy spectrum, is apparent, strengthening the view that most cosmic rays comefrom discrete sources, supernova remnants being most likely.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction determining the mechanism by which CR are accelerated. Concern-

Insofar as the present paper is a contribution to a Conference in-volved in the history of science it is appropriate to start by discuss-ing the origin of the concept of ‘the galaxy’, i.e. the disc-likedistribution of stars, with the solar system near its central plane(and 2/3 of the way from the centre). A further reason is that theinitiator of the idea, Thomas Wright, was born near the home ofone of the authors (AWW).

Fig. 1 shows the famous self-explanatory diagram from thebook of Thomas Wright (1750). It seems that Immanuel Kant, thegreat German philosopher, saw a summary of the book (but notthe book itself) and, since it accorded with his own ideas, he pub-lished it and Wright’s fame was assured. It is perhaps as well thatKant did not dig more deeply because Wright believed the disc tobe part of a spherical shell with God in the middle!

The scope of the article is to examine the role of the galaxy inexplaining the origin of the Cosmic Rays (CRs) detected at Earth.Inevitably, attention is somewhat focused on the contributions ofthe authors, some of which are somewhat speculative. However,it is through speculation (e.g. Hess’s idea that an extra-terrestrialradiation was responsible for his results) that eventual under-standing arises.

2. Energy densities in the galaxy

The energy of any physical system is important, and ‘cosmicrays’ is no exception. Table 1 shows the relevant energy densitieslocally (the ‘old-fashioned ‘units’, electron volts per cm3, are used).

It is interesting to note their near equality and to appreciate thehazards involved in using ‘equipartition’ arguments as an aid to

ing the first basic question regarding CR origin: Galactic (G) orExtragalactic (EG) there is no guidance, in that there is a near-equality between the various galactic energy densities and theimportant EG density, that of the Cosmic Microwave Background(CMB), which is, of course, Universal; a factor two difference ishardly important, not least because the galactic values dependsomewhat on Galactrocentric distance, and, of course, on distancefrom the galactic plane (z). Having said that, it is physically reason-able that the CR energy density should be close to that in magneticfields, as it is, the field acting as a trapping mechanism and as avalve for CR escape.

In order to complete this discussion of energy densities, thosefor the Universe as a whole are given in Fig. 2. A wide spread isto be noted. The case for ‘UHECR’ being so low will be given later.

3. CR: galactic or extragalactic

An early controversy arose as to the origin of cosmic rays de-tected at Earth. Were they largely of galactic origin, or extragalac-tic? (see [23] for a detailed discussion). The answer came fromgamma-ray astronomy, where most cosmic gamma rays resultfrom the interactions between CR and gas (the inter-stellar med-ium, ISM, and the intergalactic medium, IGM). The history of theearly studies has been given by Wolfendale [31] and will be sum-marised here.

1. Lack of EG gamma rays. If the CR were of EG (Universal) origin, asin Millikan’s ‘birth cries of the elements’ and the CR intensitywere the same everywhere in the Universe, the contribution tothe universal gamma ray flux from CR interacting with the IGMwould be considerable. As shown by Said et al. [28] expectationwould exceed observation by a factor approaching 100.

0.1016/

Fig. 1. Thomas Wright’s ‘galaxy’, Wright (1750). The circles represent stars withthose showing radiating lines being the brighter ones. The Sun is in the centre. Thereason for the appearance of the ‘Milky Way’ is apparent: more stars will be seen onlooking in the direction of the ‘Plane’, i.e. upwards and downwards.

Table 1Energy densities in the galaxy. A comparison of that for cosmic rays with the othersmay have relevance to the origin of CR.

eV cm�3

Magnetic field (B2/8p) ’0.5Gas motion (h1/2 Mm2i) ’0.5Cosmic rays ’0.5Starlight ’0.5(Cosmic microwave background, 2.7 K) ’0.5But, most – f(R) ’0.24

Fig. 2. Energy densities in the Universe. Most are self-explanatory. ‘PE of Galaxies’is the potential energy of the material in Galaxies, the energy realised as kineticenergy when galaxies formed could, in principle, contribute to the universal CRbudget. It is fascinating to note the wide range of energy densities involved inCosmology – a range of a factor of 1010 from that of the Dark Energy to that in theultrahigh energy (extra-galactic) cosmic rays.

2 A. Wolfendale, A. Erlykin / Astroparticle Physics xxx (2013) xxx–xxx

2. Gradient of the CR Intensity. The work of Dodds et al. [14]showed the presence of a gradient of CR intensity in the galaxysuch that the intensity in the outer galaxy is lower than thatlocally, in opposition to expectation for a Universal origin,where the CR intensity would be the same everywhere. Recentwork by Erlykin and Wolfendale [18] describes the gradientproblem, and its solution, in detail.

3. The Ginzburg test, viz an examination of the observed andexpected gamma ray fluxes from the Magellanic Clouds, byChi and Wolfendale [13] gives the same result, viz that the CRintensity in the Clouds is less than that locally.

The conclusion at this stage is that the bulk of the CR detectedas the Earth come from galactic sources. EG sources provide only�10�5–10�6 of the energy content, mainly confined to particlesabove the ‘ankle’ at E � 2 EeV.

Please cite this article in press as: A. Wolfendale, A. Erlykin, Understandingj.astropartphys.2013.01.009

4. Galactic sources: SNR

4.1. Evidence for supernova remnants as sources

The idea that supernovae are involved in CR acceleration stemsfrom the work of Baade and Zwicky [8], Baade and Zwicky [9]. Ini-tially it was thought that it was the SN themselves were the sourcebut later work identified the supernova remnant (SNR) shocks asbeing responsible.

It is apparent [31] that other systems besides SN have enoughtotal energy to be useful CR sources but SNR seem to be uniquein providing individual particles with energy up into the PeV re-gion, a key energy because this is where there is the ‘knee’, a sud-den steepening in the energy spectrum, a feature that suggests thecessation of particles from a particular type of source.

The role of gamma ray astronomy in identifying SNR as CR sourceshas been described by Fazio [22], Bignami and Hermson [12] andRamana Murthy and Wolfendale [27]. Gamma rays from specificSN and pulsars were described in the first two reviews and the evi-dence for extended SNR in the last-mentioned. An early example inthe last-mentioned review is the work of Bhat et al. [11] in whichan excess gamma ray intensity was claimed for the Loop 1 SNR.

Detailed studies of gamma rays from specific SNR is now thestaple diet of modern satellite studies (e.g. Fermi-LAT) and thelarge Cerenkov detectors (H.E.S.S). It is possible that SNR providethe bulk of CR up to PeV energies, but it must be said that oftenthere is ambiguity as to whether protons (nuclei) or electrons areresponsible for the observed gamma rays.

4.2. Fine structure in CR spectra

That the CR energy spectrum is not smooth has been known formany decades, specifically, there is the ‘knee’ at about 3 PeV and

galactic cosmic rays, Astropart. Phys. (2013), http://dx.doi.org/10.1016/

Fig. 3. The initial plot of the energy spectra from the nearby source (comprisingCNO, M-nuclei and Fe and their sum, SNR) from the background and the total, fromErlykin and Wolfendale [15]. More recently, ‘CNO’ have been replaced by He (e.g.,[18]. The intensity has been multiplied by the energy-cubed so as to flatten thespectrum and make the features easier to see. This is a standard procedure.

A. Wolfendale, A. Erlykin / Astroparticle Physics xxx (2013) xxx–xxx 3

the ‘ankle’ at about 2 EeV. The number of papers reiterating thepresence of the knee is legion but those dealing with the quantita-tive shape of the knee (specifically its sharpness) and its signifi-cance are very few. Much of what follows is concerned with our[15] explanation. This, and the evidence for other structure in thegalactic spectrum, is examined in the following sections. It is in-cluded because it is an illustration of the way in which the subjectis going.

5. The single source model

Erlykin and Wolfendale [15] have put forward a model in whichthe knee is explained in terms of a nearby, recent SNR ‘pokingthrough’ a ‘background’ caused by many, more distant SNR. The‘background’ has a slowly increasing slope due to the increasinginefficiency of galactic trapping as the energy increases – that isthe leakage from the galaxy increases. Fig. 3 shows the initial sug-gestion; in it, CNO were thought to be the most prominent parti-cles in the knee region. More recently, He have been proposed asthe particles responsible for the ‘bump’ at the knee position [17].

It can be remarked that if SNR are largely responsible for thegalactic CR, or, indeed, if any discrete sources are the cause thenthere must be ‘structure’ at some level. Whether or not it has beendetected already is a quantitative question; we believe that it has.The classical explanation for the knee is that it is due to galacticmodulation, that is the increasing inability of the galactic magneticfield to adequately trap the CR so that, for a continuing power lawinjection spectrum, the measured spectrum will slowly steepen.Taking this ‘galactic modulation’ spectrum as the datum we have

Fig. 4. Fine structure in the CR energy spectrum. The datum from which the differences inwhich the spectrum steepens slowly.

Please cite this article in press as: A. Wolfendale, A. Erlykin, Understandingj.astropartphys.2013.01.009

prepared Fig. 4. The figure shows not only the Helium ‘peak’ butalso a now accepted further peak in the region where Fe should ap-pear from the same source as provides the He peak if the acceler-ation mechanism depends on the CR rigidity (rigidity = pc/Ze,where p is the momentum and Z the charge. A small peak in be-tween the two may come from CNO, although its magnitude issmaller than expected.

The identification of the single source is not definite but theMONOGEM Ring SNR and pulsar provide a distinct possibility.The association is of age �105y and distance �300 pc. Another pos-sibility is the Vela pulsar but it must be assumed that the SNR is‘leaky’ so that CR can leak out long before its usual 8 � 104y life-time before merging into the general ISM.

6. The kink

The PAMELA and ATIC CR detectors have presented evidence fora kink in the energy spectra of most nuclei at about 100 GeV/nu-cleon (Fig. 5). The figure gives the references; it also shows the re-sults from Erlykin and Wolfendale [20] in which we decomposedthe spectra into two components. The higher energy componenthas a constant exponent beyond the low energy region, where so-lar modulation is important. Subtraction of this component fromthe total yields a ‘‘New Component’’. This new component appearsto be present for all nucleus masses. A possible origin is the LocalBubble but this is not yet certain.

A point in favour of the interpretation (of a NC) is the fact thatthe sharpness of the kink is greater than would be expected for therandom SN model of Erlykin and Wolfendale [16]. This point, andthe similarity of the feature for the different nuclear masses, isshown in Fig. 6.

The ATIC experiment shows a similar feature for electrons(Fig. 7). Again, the Local Bubble is a prime suspect.

There is also evidence for a NC in the gamma ray componentfrom Fermi-LAT [29] as analysed by Erlykin and Wolfendale [18],Erlykin and Wolfendale [19]. The kink here is at a lower energythan for the nuclei and the electrons, as would be expected if thegamma rays are secondaries. Fig. 8 shows the results.

7. Further structure in the energy spectra

It is inevitable that, as the experimental precision increases, fur-ther ‘structure’ should become apparent if, as seems to be the case,the sources are predominantly discrete. As an indication of thelikely structure to come, Fig. 9 shows a collection of six spectrafrom the Erlykin and Wolfendale [16] model involving the diffu-sion of particles from randomly distributed SNR. If such bumpsare not seen, a CR acceleration model involving distributed mech-anisms will need to be involved.

Turning to energies beyond the ‘iron peak’ the origin of the CR isstill obscure although acceleration by rare super-SNR (with magni-fied, compressed magnetic fields) is a distinct possibility, indeed,

‘intensity’ (D-values) are determined comes from the galactic modulation model, in

galactic cosmic rays, Astropart. Phys. (2013), http://dx.doi.org/10.1016/

Fig. 5. A new component ‘NC’ from the analysis by Erlykin and Wolfendale [20] ofthe measurements from PAMELA [4] ATIC [26] and CREAM (Ahn et al., 2009, 2010and Yoon et al. 2011). The abscissa is the energy divided by the atomic mass. Theordinate refers to the intensity multiplied by the energy of the power 2.5, as givenby the original authors.

Fig. 6. The rigidity at the kink for the various nuclear masses and estimates (bydifferent techniques) of the sharpness (2nd differential of the intensity with respectto energy, in logarithmic units). The near consistency of the values of rigidity fromone element to another is taken as positive sign of the applicability of the model.The dispersion in the sharpness values (lower plot) but all the values are muchbigger than expected for the conventional galactic modulation model.

Fig. 7. The electron energy spectrum from the ATIC experiment [26] as analysed byErlykin and Wolfendale [20]. The new component is indicated by the line meetingthe axis at logE = 2.

Fig. 8. Fermi-LAT gamma ray spectrum as measured by Strong et al. [29] and fittedby two components by Erlykin and Wolfendale [18], Erlykin and Wolfendale [19].‘NCc’ is the proposed new component of gamma rays.

4 A. Wolfendale, A. Erlykin / Astroparticle Physics xxx (2013) xxx–xxx

there is already evidence for a spectral flattening above 20 PeV andsteepening above 100 PeV. This energy region is the subject of in-tense contemporary study [7,10,24,25].

The ankle at about 2 EeV may well represent the transactionfrom galactic to extragalactic sources although proof is not yetforthcoming, largely because of experimental ambiguities in themass composition of the primaries in the ankle region. If, indeed,rather rare Super-SNR are responsible in the region a decade orso below the 2 EeV ankle then rather considerable structure shouldexist here.

Fig. 9. Typical CR energy spectra from the SNR model of Erlykin and Wolfendale[16]. As in Fig. 3, 4 and 7 the intensity is multiplied by energy-cubed. Rather obviousstructure is apparent, albeit with intensity variations over a decade or so of energy.

8. Cosmic rays in the inner galaxy

The preceding sections have dealt mainly with CR at Earth, byway of direct measurements. However, insofar as our scope is toexamine ‘galactic CR’, it is appropriate to study CR elsewhere inthe galaxy. We concentrate on the inner galaxy.

Gamma ray astronomy gives guidance as to the average CRintensity along a line of sight, if the column density of gas is

Please cite this article in press as: A. Wolfendale, A. Erlykin, Understanding galactic cosmic rays, Astropart. Phys. (2013), http://dx.doi.org/10.1016/j.astropartphys.2013.01.009

Fig. 10. The integral energy spectrum of diffuse gamma rays in the inner galaxy. Comparison of the Fermi-LAT measurements (open circles) with two versions of spectracalculated for anomalous diffusion with the parameter a equal to 1.0 (dashed line) or 0.8 (full line). The agreement of the experimental measurements with the spectrumcalculated for the more turbulent ISM in the inner galaxy (with a = 0.8) is quite good. l is the galactic longitude and b the galactic latitude. ‘IC’ denotes the Inverse Comptoncontribution of gamma rays from CR electrons on photons.

A. Wolfendale, A. Erlykin / Astroparticle Physics xxx (2013) xxx–xxx 5

known, the gamma rays in question being the diffuse component.We, Erlykin and Wolfendale [21], have made a preliminary analysisof the Fermi-LAT data [3] for gamma ray energies below 30 GeVand MILAGRO data [1,2] for Ec � 10 TeV and confirmed that theCR spectrum in the inner galaxy is flatter than that locally.Fig. 10 illustrates the situation for the ‘, b range indicated. In calcu-lating the expected p� contribution the mode of CR diffusion isneeded, and this is not self-evident. Many workers have used thestandard ‘Gaussian diffusion’ but some (see [30], for a summary)have drawn attention to the importance of ‘anomalous diffusion’in view of the high degree of turbulence in certain parts of the gal-axy. In the work described, anomalous diffusion was assumed, tothe extent of the coefficient, a = 1.0 (a = 2 for Gaussian diffusion).It will be noted that even with this degree of anomalous diffusionthere is a flatter spectrum than expected from local data. In ourview there are three reasons for this:

(a) There is a higher density of SNR in the inner galaxy and thusa bigger fraction of gamma rays coming from near-SNR loca-tions where the CR spectrum of CR will be flatter and nearerthe spectrum of CR on emergence from the shell itself(exponent � 2.15).

(b) The fraction of lines of sight passing actually through (unrec-ognised) SNR will be higher.

(c) The degree of turbulence in the ISM in the inner galaxy is sohigh that a is less than unity.

Of the above we consider (c) to be most important and it isfound that a = 0.8 gives a reasonable fit to the spectrum, as evidentin Fig. 10.

It can be remarked that the ensuing dependence of CR diffusionon the degree of turbulence is still the subject of analysis; in fact,very recent work, allowing for the effect of discrete sources, indi-cates that ‘final’ value of a may be nearer 0.9 than 0.8. Such achange does not invalidate the claim that the spectral shape inthe inner galaxy is flatter than locally.

9. Conclusions

There is no doubt that the bulk of the cosmic radiation is pro-vided by discrete galactic sources, supernova remnants being themost likely candidates. The discreteness of the sources leads tothe expectation of ‘fine structure’ in the energy spectra and this ap-

Please cite this article in press as: A. Wolfendale, A. Erlykin, Understandingj.astropartphys.2013.01.009

pears to have been detected. It is confidently asserted that furtherstructure will be seen in the energy region above the ‘iron peak’and that this will become stronger as the energy at which the EGparticles start to predominate (several EeV?) is reached.

Acknowledgement

The authors are grateful to the Kohn Foundation for financialsupport.

References

[1] A.A. Abdo et al., Astrophys. J. 658 (2007) L33.[2] A.A. Abdo et al., Astrophys. J. 688 (2008) 1078.[3] M. Ackermann et al., Astrophys. J. 750 (2012) 3.[4] O. Adriani et al., Science 332 (2011) 69.[5] H.S. Ahn et al., Astrophys. J. 707 (2009) 593.[6] H.S. Ahn et al., Astrophys. J. 714 (2010) L89.[7] J.C. Arteaga-Velazquez et al., 16 ISVHECRI, Batavia, arxiv: 1009.4716, 2010.[8] W. Baade, F. Zwicky, in: Proceedings of National Acadamic Science, USA, 20

(1934a) 254.[9] W. Baade, F. Zwicky, in: Proceedings of National Acadamic Science, USA, 20

(1934b) 259.[10] S.F. Berezhnev et al., in Proceedings of 2nd International Cosmic Ray

Workshop, Armenia, 48; arxiv: 1201 (2011) 2122.[11] C.L. Bhat et al., Nature 314 (1985) 511.[12] G.F. Bignami, W. Hermson, Ann. Rev. Astrom. Astrophys. 211 (1983) 67.[13] X. Chi, A.W. Wolfendale, J. Phys. G. 19 (1993) 795.[14] D. Dodds et al., Mon. Not. Roy. Astr. Soc. 1971 (1975) 569.[15] A.D. Erlykin, A.W. Wolfendale, J. Phys. G. 23 (1997) 979.[16] A.D. Erlykin, A.W. Wolfendale, J. Phys. G. 29 (2003) 709.[17] A.D. Erlykin, A.W. Wolfendale, J. Phys. G. 32 (2006) 1.[18] A.D. Erlykin, A.W. Wolfendale, in: Proceedings of 2nd. International C. R.

Workshop, Armenia, 16, 2011a.[19] A.D. Erlykin, A.W. Wolfendale, ibid, 106, 2011b.[20] A.D. Erlykin, A.W. Wolfendale, Astro. Part. Phys. 35 (2012) 449.[21] A.D. Erlykin, A.W. Wolfendale, Astro. Part. Phys. 42 (2013) 72.[22] G.G. Fazio, ‘X-Ray and c-Ray Astronomy. In: H. Brady, R. Giacconi (Eds.),

Proceedings of AIU Symposium 55, Madrid, Reidel, Dordrecht, 1973[23] V.L. Ginzburg, S.I. Syrovatsky, The Origin of Cosmic Rays, Pergamon Press,

Oxford, 1964.[24] R.M. Martirosov et al., 16 ISVHECRI, Batavia, arxiv: 1010.6260, 2010.[25] R.M. Martirosov et al., 32 International Cosmic Rays Conference, Beijing, 1/11,

178, 2011.[26] A.D. Panov et al., Bull. Russ. Acad. Sci. Phys. 73 (5) (2009) 564.[27] P.V. Ramana Murthy, A.W. Wolfendale, Gamma Ray Astronomy, CUP, 1986,

1993.[28] S.S. Said et al., J. Phys. G. 8 (1982) 383.[29] A. W. Strong et al., in: Proceedings 12th ICTPP, arxiv:1101. 1381, 2011.[30] V.V. Uchaikin, Phys. Uspekhi 46 (2003) 821.[31] A.W. Wolfendale, Q. Jl. Astron. Soc. 24 (1983) 122.[32] Y.S. Yoon et al., Astrophys. J. 728 (2011) 122.

galactic cosmic rays, Astropart. Phys. (2013), http://dx.doi.org/10.1016/