Progress in Particle and Nuclear Physics 66 (2011) 681–685

Contents lists available at ScienceDirect

Progress in Particle and Nuclear Physics

journal homepage: www.elsevier.com/locate/ppnp

Review

Gamma-ray emission from molecular clouds: A probe of cosmic-rayorigin and propagationSabrina Casanova ∗

Max Planck Institut für Kernphysik, Saupfercheckweg 1, 69117 Heidelberg, GermanyRuhr Universität Bochum, Universitätsstrasse 150, 44801 Bochum, Germany

a r t i c l e i n f o

Keywords:Cosmic raysGamma-ray emissionCosmic-ray flux

a b s t r a c t

Cosmic rays up to at least 1015 eV (PeV) are believed to be emitted by Galactic sources,such as supernova remnants. However, no conclusive evidence of their acceleration hasbeen found yet. A trace of ongoing cosmic-ray acceleration is the gamma-ray emissionproduced by these highly energetic particles when they scatter off the interstellar mediumgas, mainly atomic and molecular hydrogen. Whereas the atomic hydrogen is uniformlydistributed in the Galaxy, the molecular hydrogen is usually aggregated in dense clouds,and the gamma-ray emission fromsuch clouds is particularly intense and localised. Amulti-frequency approach, which combines the data from the upcoming and future gamma-rayemissions with the data from the submillimeter and millimeter surveys of the molecularhydrogen, is therefore crucial to probe the Galactic cosmic-ray flux. In order to fully exploitthis multi-frequency approach, one needs to develop predictions of the expected emission.Here we will discuss the GeV to TeV emission from runaway CRs penetrating molecularclouds close to the young supernova remnant RX J1713-3946 and in molecular cloudsilluminated by the background cosmic-ray flux.

© 2011 Published by Elsevier B.V.

1. The standard model of cosmic rays

Cosmic rays (CRs) are the highly energetic protons and nuclei which fill the Galaxy and carry, at least in the vicinity of theSun, as much energy per unit volume as the energy density of starlight or of the interstellar magnetic fields or the kineticenergy density of the interstellar gas. One hundred years after their discovery by the Austrian physicist Victor Hess, theorigin of cosmic rays is still unclear. Diffusive shock acceleration in supernova remnants (SNRs) is the most widely invokedparadigm to explain the Galactic cosmic-ray spectrum. Galactic SNRs provide, in fact, the necessary power to sustain theGalactic cosmic-ray population. One expects about one supernova event every 30 years, and, in order to account for theenergy density of cosmic rays (about 1 eV/cm3) and the cosmic-ray confinement time deduced from spallation, the typicalnon-thermal energy release per supernova has to be about 1050 ergs, which is about ten percent of the total energy releasedin supernova explosion [1]. This is in good agreementwith the typical amount of energy predicted to be produced during theacceleration of relativistic particles in SNR shocks. If the bulk of Galactic CRs up to at least PeV energies are indeed acceleratedin SNRs, then TeV γ -rays are expected to be emitted during the acceleration process CRs undergo within SNRs [2]. IndeedTeV γ -rays have been detected from the shells of SNRs, such as RX J1713.7-3946 [3]. However, such observations do notconstitute a definitive proof that CRs are accelerated in SNRs, since the observed emission could be produced by energeticelectrons up scattering low energy photon fields.

The direct observation of cosmic rays from the candidate injection sites such as supernova remnants is not possible sinceCRs escape the acceleration sites and eventually propagate into the Galactic magnetic fields. CR secondary data suggest that

∗ Corresponding address: Max Planck Institut für Kernphysik, Saupfercheckweg 1, 69117 Heidelberg, Germany.E-mail address: Sabrina.Casanova@mpi-hd.mpg.de.

0146-6410/$ – see front matter© 2011 Published by Elsevier B.V.doi:10.1016/j.ppnp.2011.01.029

682 S. Casanova / Progress in Particle and Nuclear Physics 66 (2011) 681–685

cosmic protons and nuclei diffuse in the magnetic fields for timescales of the order of about tescape ≈ 107( <E10GeV )−0.65 years,

E being the particle energy, before escaping the Galaxy. During these timescales the particles from individual sources losememory of their origin, and contribute to the bulk of Galactic cosmic rays known as cosmic-ray background or CR sea,losing the information on the original acceleration locations and spectra. If h is the distance that CRs have to travel beforeescaping the Galaxy, then the diffusion coefficient will be D ≈ h2/tescape and one deduces an energy dependence of thepropagation in the Galactic disk, D(E) = D10(

E10GeV )δ , where the diffusion coefficient at 10 GeV is about D10 = 1028 cm2/s,

and δ = 0.3–0.7. The average cosmic-ray density is thus determined by the contribution of all Galactic sources over a longperiod of time comparable to the CR escape time from the Galaxy.

The information on the locations of individual CR sources, their spectra and their injection rate, which get lost during thediffusion and convection processes CRs undergo, can be traced back through the gamma rayswhich cosmic rays radiatewhenthey interact with the ambient gas in the interstellar medium. In fact, the gamma rays are emitted through decay of neutralpions produced in inelastic collision of CR hadrons and interstellar gas. Contrary to CRs, the gamma rays, being neutral,travel in straight lines from the site where they were emitted to the detector. For this reason γ -ray astronomy has alwaysplayed a key role to probe the Galactic cosmic-ray flux and to solve the longstanding question of the origin of cosmic rays.The high sensitivity, high resolution γ -ray data from current (HESS, Magic, Veritas, Fermi and Agile) and future detectors,such as AGIS, CTA and HAWC [4,5], together with the knowledge of the distribution of the atomic and molecular hydrogenin the Galaxy on sub-degree scales are crucial to explore the flux of high energy CRs close to the candidate CR sourcesand to pinpoint the long searched-for sites of CR acceleration. Besides the origin of CRs other open questions concern themaximumenergy towhich the candidate Galactic astrophysical accelerators of cosmic rays, SNRs and pulsars, can accelerateparticles and the role of the extreme gravitational andmagnetic fields in these powerful sources. To answer these questionswould bring us to understand the conditions of the extreme astrophysical objects, which are able to accelerate particles tovery high energies. Furthermore, the cosmic rays are an energetically important component of the interstellar matter (ISM),which is the content in gas, particles, radiation fields, magnetic fields, and relativistic electrons and nuclei (cosmic rays) ofthe Galaxy. In fact, the energy density of cosmic rays is comparable to the energy density of the Galactic magnetic fields,radiation fields and turbulent motions of the interstellar gas. The different components of the ISM are strongly coupledtogether, and constitute a physical entity which strongly influences the Galactic dynamics and evolution.

In Section 2 we will present the results of the modeling of the γ -ray emission in molecular clouds close to young SNRs.In Section 3 we will describe a methodology to investigate the CR flux and spectrum in distant regions of the Galaxy fromobservations of γ -rays emitted by CRs propagating through molecular clouds. Our conclusions are given in Section 4.

2. On the origin of cosmic rays

Cosmic rays escaping supernova remnants diffuse in the interstellar medium and collide with the ambient atomic andmolecular gas [6–9]. From such collisions gamma rays are created, which can possibly provide the first evidence of a parentpopulation of runaway cosmic rays. Before being isotropised by the Galactic magnetic fields, the injected CRs produce, infact, γ -ray emission, which can significantly differ from the emission of the SNR itself, as well as from the diffuse emissioncontributed by the background CRs and electrons, because of the hardness of the runaway CR spectrum, which is not yetsteepened by diffusion. These diffuse sources are often correlated with dense molecular clouds (MCs), which act as a targetfor the production of γ -rays due to the enhanced local CR injection spectrum. SNRs are located in star forming regions, whichare rich in molecular hydrogen. In other words, CR sources andMCs are often associated and target–accelerator systems arenot unusual within the Milky Way [10,6]. γ -rays in association with dense MCs close to candidate CR sources, have beendetected both at GeV energies [11–14] and TeV energies [15–19].

The γ -ray radiation from hadronic interactions from regions close to CR sources depends not only on the total poweremitted in CRs by the sources and on the distance of the source to us, but also on the ambient interstellar gas density,the local diffusion coefficient and the injection history of the CR source. It is therefore difficult to definitely recognize thesites of CR acceleration from γ -ray observations alone, since very often only qualitative predictions are provided, rather thanrobust quantitative predictions, especially fromamorphological point of view. In order to fully exploit the present and futureexperimental facilities and to test the standard scenario for CR injection in SNRs and propagation,we computed the expectedγ -ray emissivity from hadronic interactions of runaway CRs in a region around the SNR RX J1713.7-3946, by constructingas quantitative a model as possible. In particular, we have conveyed all information concerning the environment, thesource age, the acceleration rate and history, which all play a role in the physical process of CR injection and propagation[20].

Fig. 1 shows the γ -ray spectra fromhadronic interactions of background and runaway CRs in four regions around the SNRRX J1713.7-394. In all four locations the hadronic γ -ray emission is enhancedwith respect to the hadronic emission due onlyto background CRs at energies above few TeVs as a consequence of the fact that the CR fluxes are enhanced above 100 TeV.CRs of about 100 TeV are, in fact, thought to be released now by RX J1713.7-394 [21]. The hadronic γ -ray emission from theregion d is particularly enhanced with respect to the hadronic emission from background CRs. This is due to two factors, theenhanced CR flux nearby the injection source and the particularly high ambient gas density. In the three regions a, b and dthe high energy emission is clearly dominated by the radiation produced along the line of sight distance between 900 and1100 parsecs, plotted in dashed lines in Fig. 1. The γ -ray emission from high latitude regions, such as region c , is instead

S. Casanova / Progress in Particle and Nuclear Physics 66 (2011) 681–685 683

1e-16

1e-15

1e-14

1e-13

1e-12

1e-11

0.001 0.01 0.1 1 10 100

E2 d

N/d

E

( T

eV s

-1 c

m-2

)

Energy (TeV)

a

1e-16

1e-15

1e-14

1e-13

1e-12

1e-11

0.001 0.01 0.1 1 10 100

E2 d

N/d

E

( T

eV s

-1 c

m-2

)

Energy (TeV)

b

1e-16

1e-15

1e-14

1e-13

1e-12

1e-11

0.001 0.01 0.1 1 10 100

E2 d

N/d

E

( T

eV s

-1 c

m-2

)

Energy (TeV)

c

1e-16

1e-15

1e-14

1e-13

1e-12

1e-11

0.001 0.01 0.1 1 10 100

E2 d

N/d

E

( T

eV s

-1 c

m-2

)

Energy (TeV)

d

Fig. 1. The γ -ray energy flux in four different regions of 0.2° × 0.2° around the positions a = (346.8, −0.4), b = (346.9, −1.4), c = (347.1, −3.0)and d = (346.2, 0.2). The emission produced along the line of sight distance between 900 and 1100 parsecs is plotted in dashed lines, while the emissionobtained by summing the radiation contributions over the whole line of sight distance, from 50 parsecs to 30 000 parsecs, is shown in solid lines. Theemission in the panels a, b, c and d, corresponding to the different locations, is plotted for different diffusion coefficientsD0 . The emission from backgroundCRs is also shown in each panel for comparison. The contribution to the emission from inverse Compton scattering of background electrons is indicatedwith a dashed light blue line. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2. Ratio of the emission due to the sum of background CRs and runaway CRs and background CRs only. The SNR is supposed to have exploded at 1kpc distance from the Sun at 347.3° longitude and −0.5° latitude 1600 years ago and to have started injecting the most energetic protons 100 years afterthe explosion. The diffusion coefficient assumed within the region 340° < l < 350° and −5° < b < 5° is 1026 cm2/s in the left panel, 1027 cm2/s in themiddle panel and 1028cm2/s in the right panel. In the three panels the ratio of the emission is shown for different energies from 1 GeV to 10 TeV.

dominated by the contribution from IC scattering of background electrons, almost at all energies. In regions closer to theGalactic Plane the emission from inverse Compton scattering of background electrons is subdominant at TeV energies, whererunaway cosmic rays produce the enhanced emission. Therefore the regions where to look for the emission from runawayparticles are low latitude regions of higher gas density. The γ -ray spectra show a peculiar concave shape, being soft at lowenergies and hard at high energies, which, as discussed in [9],might be important for the studies of the spectral compatibilityof GeV and TeV gamma-ray sources. The peculiar spectral and morphological features of the γ -ray due to runaway CRs canbe therefore revealed by combining the spectra and γ -ray images provided by the Fermi and Agile telescopes at GeV and bypresent and future ground based detectors at TeV energies. As shown by the surveys of the Galaxy, published by Fermi atMeV–GeV energies, by HESS at TeV energies and at very high energies by the Milagro Collaboration, the various extendedGalactic sources differ in spectra, flux and morphology. However, there is growing evidence for the correlation of GeV andTeV energy sources. These sources appear often spectrally and morphologically different at different energies, possibly duenot only to the better angular resolution obtained by the instruments at TeV energies, but also to the energy dependence ofphysical processes, such as CR injection and CR diffusion. For this reason it is important to properly model what we expectto observe at different energies by conveying in a quantitative way all information by recognizing that the environment,the source age, the acceleration rate and history, all play a role in the physical process of injection and all have to be takeninto account for the predictions. Fig. 2 shows the ratio of the hadronic gamma-ray emission due to total CR spectrum tothat of the background CRs for the entire region under consideration. In our modeling only CRs with energies above about100 TeV have left the acceleration site and the morphology of the emission depends upon the energy at which one observesthe hadronic gamma-ray emission. The different spatial distribution of the emission is also due to the different energy-dependent diffusion coefficients, assumed in the three different panels, making it into a useful tool to investigate the highlyunknown CR diffusion coefficient [20,22].

684 S. Casanova / Progress in Particle and Nuclear Physics 66 (2011) 681–685

3. On the level of the CR background

The cosmic-ray flux measured close to the Earth, with its characteristic soft −2.7 spectrum, is usually assumed to berepresentative of the average cosmic-ray flux throughout the Galaxy. By combining the locally observed spectrum and thepropagation effects, one would conclude that the injection spectra from individual sources have a power-law slope close to−2, in accordancewith the predictions of the diffusive shock acceleration theory. However, on the basis of the state of the artknowledge of CR propagation processes, a somehow steeper source spectrum up to −2.4 cannot be excluded assuming thatCRs might re-accelerate in the Galaxy [2]. Also, for protons and nuclei a good working hypothesis to interpret their locallymeasured energy spectrum is to assume that the CR sources are distributed uniformly and continuously in the Galaxy and allcontribute to the locally observed cosmic-ray flux. However, in principle, one cannot exclude that the cosmic-ray spectrumin the immediate vicinity of the Sun is due to a single local source [23], and that the level of the cosmic-ray sea in otherlocations of the Galaxy might be different from the one observed locally.

Direct measurements provide the CR spectrum and flux only in the vicinity of the Solar System and the level of the CRspectrum and flux in other regions of the Galaxy is unknown. Interestingly, based on EGRET data [24] concluded that theγ -ray emission from the inner Galaxy at energies greater than about 1 GeV exceeds, by some 60%, the intensity predicted bymodel calculations, in which the average Galactic CR flux is equal to the one measured locally [25,24]. Recent observations,performed by the LAT instrument on board, the Fermi satellite, show that the spectra of the Galactic diffuse emission atMeV–GeV energies, at least at intermediate latitudes and in the Outer Galaxy [26,11], can be explained by cosmic-raypropagationmodels based on local observations of cosmic-ray electron andnuclei spectra. On the other hand, at TeV energiesthe spectral features of the γ -ray emission detected by HESS from the Galactic centre (GC) region [3], and the emissionmeasured by Milagro from the Cygnus Region [27,28], suggest that the CR flux might significantly vary in the differentlocations of the Galaxy.

The emissivity of molecular clouds located far away from CR sources, so called passivemolecular clouds, i.e. clouds whichare illuminated by the supposedly existing CR background, can be used to probe the level of the CR sea [7]. Given that theγ -ray-emission from themolecular cloud depends only upon the total mass of the cloud,M , and its distance from the Earth,d, the CR flux, ΦCR, in the cloud is uniquely determined as

ΦCR ∝Fγ d2

M(1)

where Fγ is the integral γ -ray flux from the cloud. Under the assumption that the CR flux in the cloud is equal to the locallyobserved CR flux, the calculated γ -ray flux from the cloud can be compared to the observed γ -ray flux in order to probe theCR spectrum in distant regions of the Galaxy. The detection of under-luminous clouds with the respect to predictions basedon the CR flux at Earth would suggest that the local CR density is enhanced with respect to the Galactic average density.This would cast doubts on the assumption that the local CRs are produced only by distant sources, and that the CR flux andspectrum measured locally is representative of the typical CR flux and spectrum present throughout the Galaxy. In whatfollows we will describe a methodology to test the CR flux in distant molecular clouds.

The longitudinal profile of the γ -ray emission due to protons scattering off the atomic [29] and molecular hydrogen[30,31] from the region which spans Galactic longitude 340° < l < 350° and Galactic latitude −5° < b < 5° is shownin Fig. 3 (Left). A peak in the emission at longitude of about 345.7° close to the Galactic Plane is clearly visible, next to adip in the longitudinal profile. While the atomic gas is generally broadly distributed along the Galactic Plane, the molecularhydrogen is less uniformly distributed and the peaks in the γ -ray longitudinal profiles correspond to the locations of highestmolecular gas column density. The peaks in the longitudinal profile reveal the directions in the Galaxywheremassive cloudsassociated with spiral arms are aligned along the line of sight.

Fig. 3(Right) shows that the peak in the γ -ray emission from the direction 345.7° close to the Galactic plane is mostlyproduced within 0.5 and 3 kpc distance from the Sun. The contribution to the emission from the atomic hydrogen in the0.5 kpc < d < 3 kpc amounts to about 10% of the total contribution from atomic and molecular hydrogen. Thus the γ -rayemission from this direction provides a unique probe of the CR spectrum in 0.5–3 kpc [32].

4. Conclusions

Here we have presented amethodology to test the cosmic-ray flux in discrete distant regions of the Galaxy by combiningthe most recent high resolution γ -ray and gas data. Through the knowledge of the mass and distance of distant molecularcloudswehave predicted their γ -ray emissivity under the assumption that the CR flux in the clouds is equal to thatmeasuredat Earth. By comparing the predicted and the measured γ -ray flux one can obtain a test of the level of the CR flux in distantregions of the Galaxy. The emission from the regions surrounding young SNR shells can provide crucial information on thehistory of the SNR acting as a CR source and important constraints on the highly unknown diffusion coefficient. Detailedmodeling of the energy spectra and of the spatial distribution of the gamma-ray emission in the environment surroundingRX J1713.7-3946 have been presented by using the data from atomic and molecular hydrogen surveys. These predictionsshow that the age and acceleration history of the SNR, the particle diffusion regime and the distribution of the ambientgas are all paramount. Also, depending on the time and energy at which one observes the remnant and the surrounds, one

S. Casanova / Progress in Particle and Nuclear Physics 66 (2011) 681–685 685

Galactic Longitude [deg]340 342 344 346 348 350

]-1

]-1

TeV

-1 s

r-1

s

2G

amm

a–R

ay P

rofi

le [

cm

TeV

-1 s

r-1

s

2G

amm

a–R

ay P

rofi

le [

cm

TeV

-1 s

r-1

s

2G

amm

a–R

ay P

rofi

le [

cm

]-1

]-1

TeV

-1 s

r-1

s

2G

amm

a–R

ay P

rofi

le [

cm

0.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

0.055 sum

HI

H2

1 GeV

Galactic Longitude [deg]340 342 344 346 348 350

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

-310×

sum

HI

H2

10 GeV

Galactic Longitude [deg]340 342 344 346 348 350

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45-610×

sum

HI

H2

100 GeV

Galactic Longitude [deg]340 342 344 346 348 350

0.2

0.4

0.6

0.8

1

-910×

sum

HI

H2

1 TeV

Distance (kpc)

10-11 10

)-1

kpc

-1 s

(pho

tons

cm

-2

10-10

10-9

10-8sum

HI

H2

Fig. 3. (Left) The longitudinal profile of the γ -ray emission from the region 340° < l < 350°, integrated over the latitude range −5° < b < 5°. Theemission is convolved with the energy-dependent PSF of Fermi at 1 and 10 GeV, and of a future Cherenkov telescope such as CTA at 100 GeV and 1 TeV. Thedotted green lines are for the emission arising from CRs scattering off molecular hydrogen, the dashed red ones for the emission arising from CRs scatteringoff atomic hydrogen and the solid black ones for the sum. (Right) Flux above 1 GeV as a function of the line of sight distance, produced by CRs interactingwith the atomic andmolecular gas [32]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of thisarticle.)

will observe different spectra and morphologies. This has important implications for the current and future generations ofgamma-ray observatories. The high sensitivity and high resolution, which will be reached by future detectors, such as AGIS,CTA and HAWC (e.g. see reviews [4,5]), makes the detection of the predicted emission possible.

References

[1] V.L. Ginzburg, S.I. Syrovatskii, Origin of Cosmic Rays, Pergamon Press, London, 1964.[2] L.O’C. Drury, F.A. Aharonian, H.J. Völk, Astronomy and Astrophysics 287 (1994) 959.[3] F.A. Aharonian, et al., [H.E.S.S. Collaboration], Nature 432 (2004) 75.[4] F. Aharonian, J. Buckley, T. Kifune, G. Sinnis, Reports on Progress in Physics 71 (2008) 9.[5] J.A. Hinton, W. Hofmann, Annual Review of Astronomy and Astrophysics 47 (2009) 523.[6] T. Montmerle, Astrophysical Journal 231 (1979) 95.[7] M.R. Issa, A.W. Wolfendale, Nature 292 (1981) 430.[8] F.A. Aharonian, A.M. Atoyan, Astronomy and Astrophysics 309 (1996) 917.[9] S. Gabici, F.A. Aharonian, S. Casanova, Monthly Notices of the Royal Astronomical Society 396 (2009) 1629.

[10] J. Paul, M. Casse, C.J. Cesarsky, Astrophysical Journal 207 (1976) 62.[11] A.A. Abdo, et al., [Fermi Collaboration], Astrophysical Journal 703 (2009) 1249A.[12] A.A. Abdo, et al., [Fermi Collaboration], Science 327 (2010) 1103.[13] M. Tavani, et al., Astrophysical Journal 710 (2010) L151.[14] D. Castro, P. Slane, Astrophysical Journal 717 (2010) 372C.[15] Albert, et al., Astrophysical Journal 664 (2007) 87.[16] F.A. Aharonian, et al., Astronomy and Astrophysics 481 (2008) 401.[17] F.A. Aharonian, et al., Astronomy and Astrophysics 483 (2008) 509.[18] F.A. Aharonian, et al., Astronomy and Astrophysics 490 (2008) 685.[19] F.A. Aharonian, et al., [H.E.S.S. Collaboration], Nature 439 (2006) 695.[20] S. Casanova, et al., Publications of the Astronomical Society of Japan 62 (2010) 1127.[21] V.N. Zirakashvili, F.A. Aharonian, Astrophysical Journal 708 (2010) 965Z.[22] S. Gabici, S. Casanova, F.A. Aharonian, G. Rowell, Preprint arXiv:1009.5291.[23] A.D. Erlykin, A.W. Wolfendale, Journnal of Physics G (Nuclear and Particle Physics) 23 (1997) 979.[24] S.D. Hunter, et al., Astrophysical Journal 481 (1997) 205.[25] D.L. Bertsch, et al., Astrophysical Journal 416 (1993) 587.[26] A.A. Abdo, et al., [Fermi Collaboration], Physical Review Letter 103y1101A (2009).[27] A.A. Abdo, et al., [Milagro Collaboration], Astrophysical Journal L658 (2007) 33.[28] A.A. Abdo, et al. [Milagro Collaboration], Astrophysical Journal, 688, 1078A.[29] P.M.W. Kalberla, et al., Astronomy and Astrophysics 440 (2005) 775.[30] Y. Fukui, et al., Publications of the Astronomical Society of Japan 51 (1999) 6.[31] Y. Fukui, et al., Publications of the Astronomical Society of Japan 53 (2001) 6.[32] S. Casanova, et al., Publications of the Astronomical Society of Japan 62 (2010) 769.