A multi-messenger search for the origin of cosmic-rayscta.scphys.kyoto-u.ac.jp/workshop/CTA-J/2015/presentations/22.ctaws2015_sano_20160114...H.E.S.S. Collaboration et al.: Detailed

2015 "The extreme Universe viewed in very-high-energy gamma-rays 2015"

A multi-messenger search for the origin of cosmic-raysHidetoshi SANO (Nagoya Univ.)

haronian+07Broersen+14

Radio (CO, HI) Infrared X-rays Gamma-rays

Williams+11HS+165in5prep.

The extreme Universe viewed in very-high-energy gamma-rays 2015, January 13–14, 2015, ICRR

!  Cosmic rays: relativistic protons (90%), He (10%), electrons + nuclei (<1%) " They have a significant influence on the interstellar space

" Revealing the origin of cosmic-rays is essential to understand the universe

Revealing the origin of cosmic rays

Heating the interstellar gas Galaxy evolution© NASA/J. Hester, P. Scowen © NSF/J.Yang© Kagaya


Earth environment

SNRs are thought to be the primary sources for the bulk of Galactic cosmic-raysHowever, the acceleration site of protons is still being debated…

[1] Origin of gamma-rays (hadronic and/or leptonic)" Spatial comparison between the ISM and gamma-rays

[2] Acceleration models including more realistic conditions" Shock-cloud interaction model " Other acceleration mechanisms ?

(non-liner, 2nd order Fermi, magnetic reconnection etc…)

[3] Gamma-ray obs. with high sensitivity + angular resolution " Cherenkov Telescope Array (CTA)

[4] To reveal an universality " Extend to the distant sources including the LMC SNRs

Outstanding issues solutions


H.E.S.S. Collaboration et al.: Detailed analysis of the shell type SNR RCW 86

E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Leptonic ModelIC

IC CMB

IC dustIC stars

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.

Fig. 4. Spectral energy distribution of RCW 86 for a leptonic scenario. The red solid lines denote the total broadband emission from the one-zonemodeling discussed in Sect. 4 The dotted lines show the IC-contributions and the dashed line is that of the ⇡0-decay contribution. The radio datapoints are from Molonglo at 408 MHz and Parkes at 5 GHz (Caswell et al. 1975; Lemoine-Goumard et al. 2012). X-ray data are from ASCA andRXTE from Lemoine-Goumard et al. (2012). The Fermi-LAT data points are taken from Yuan et al. (2014) and the H.E.S.S. data are from thisanalysis.

E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Hadronic ModelIC

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.

= -2pΓ

Fig. 5. Spectral energy distribution of RCW 86 for a hadronic scenario. The solid blue lines denote the total broadband emission from the one-zonemodeling discussed in Sect. 4 The dotted lines show the IC-contributions and the dashed line is that of the ⇡0-decay contribution. The black dasheddotted line shows the results for a proton spectrum with �

p

= 2. The radio data points are from Molonglo at 408 MHz and Parkes at 5 GHz (Caswellet al. 1975; Lemoine-Goumard et al. 2012). X-ray data are from ASCA and RXTE from Lemoine-Goumard et al. (2012). The Fermi-LAT datapoints are taken from Yuan et al. (2014) and the H.E.S.S. data are from this analysis.

ReferencesAcero, F., Aharonian, F., Akhperjanian, a. G., et al. 2010, Astronomy &

Astrophysics, 516, 7Aharonian, F., Akhperjanian, A. G., de Almeida, U. B., et al. 2009, The

Astrophysical Journal, 692, 1500Aharonian, F. A. 2013, Astroparticle Physics, 43, 71Arbutina, B., Urosevic, D., Andjelic, M. M., Pavlovic, M. Z., & Vukotic, B.

2012, The Astrophysical Journal, 746, 79Bamba, A., Koyama, K., & Tomida, H. 2000, Publications of the Astronomical

Society of Japan, 52, 1157Berezhko, E. G. & Ellison, D. C. 1999, The Astrophysical Journal, 526, 385Berge, D., Funk, S., & Hinton, J. 2007, Astronomy and Astrophysics, 466, 1219Bernlohr, K., Carrol, O., Cornils, R., et al. 2003, Astroparticle Physics, 20, 111

Broersen, S., Chiotellis, A., Vink, J., & Bamba, A. 2014, Monthly Notices of theRoyal Astronomical Society, 441, 3040

Caswell, J. L., Clark, D. H., & Crawford, D. F. 1975, Aust. J. Phys., 37, 39de Los Reyes Lopez, R. 2011, in Proceedings of the 32nd International Cosmic

Ray Conference (ICRC 2011), 216de Naurois, M. & Rolland, L. 2009, Astroparticle Physics, 32, 231Dickel, J. R., Strom, R. G., & Milne, D. K. 2001, The Astrophysical Journal,

546, 447Ghavamian, P., Raymond, J., Smith, R. C., & Hartigan, P. 2001, The

Astrophysical Journal, 547, 995Helder, E. A., Vink, J., Bamba, A., et al. 2013, Monthly Notices of the Royal

Astronomical Society, 435, 910Helder, E. A., Vink, J., Bassa, C. G., et al. 2009, Science, 325, 719

6


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Leptonic ModelIC

IC CMB

IC dustIC stars

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Hadronic ModelIC

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.

= -2pΓ


p













6


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Leptonic ModelIC

IC CMB

IC dustIC stars

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Hadronic ModelIC

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.

= -2pΓ


p













6


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Leptonic ModelIC

IC CMB

IC dustIC stars

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Hadronic ModelIC

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.

= -2pΓ


p













6


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Leptonic ModelIC

IC CMB

IC dustIC stars

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Hadronic ModelIC

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.

= -2pΓ


p













6


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Leptonic ModelIC

IC CMB

IC dustIC stars

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.


E [eV]-610 -410 -210 1 210 410 610 810 1010 1210 1410

eV]

-1 s

-2 F

lux

[cm

2 E

-110

1

10

210Hadronic ModelIC

decay0π

MOST

ATCA

ASCA

RXTE

Fermi/LAT

H.E.S.S.

= -2pΓ


p













6

CR protons

Interstellar protons π0

π-π+ γ-ray

(π0 decay)

γ-ray

Hadronic process CR electrons

Low energy photon CMB, infrared etc.

γ-ray

IC scattering

Leptonic process

B~100 µG B~22 µG

RCW$86$(H.E.S.S.$Col.+15)

[1] Origin of gamma-rays (hadronic and/or leptonic)

CR protons

Interstellar protons π0

π-π+ γ-ray

(π0 decay)

γ-ray

Hadronic process CR electrons

Low energy photon CMB, infrared etc.

γ-ray

IC scattering

Leptonic process

F∝Wpnd 2

Wp: total energy in accelerated proton n: ISM gas density d: distance to the SNR

Gamma-rays distribution ISM gas distribution


[1] Origin of gamma-rays (hadronic and/or leptonic)

Evidence for the CR proton accelerations (Fukui+)

(upper5panels)5Fukui,5HS+12,5(lower5panels)5Fukui,5HS+165in5prep.

AcceleraBon5efficiency5~0.15%TeV5GammaJrays555555555Total5ISM


[2] Acceleration models including more realistic conditions

!  Synchrotron X-rays are enhanced around CO clumps in 0.1 pc scale

Image Suzaku5XJrays5(RXJ1713)5Contours:5NANTEN25CO5

HS+10, 13, 15a

15pc


Shock cloud interaction model: schematic view

ISM clump (density ncloud)

shock front

vcloud = v0 × n0 ncloud

density n0 shock speed v0

vcloud

shock front

Eddy (turbulence)

Before After

Vcloud$will$be$decreased…

# Shock cloud interaction generates turbulence around the ISM clumps ![1] rotational current + dynamo effect " magnetic field amplification ![2] Synchrotron X-rays are enhanced around the ISM clumps


Shock cloud interaction model: HMD simulation (Inoue+12)

ISM5density MagneBc5field5strength

Inoue+12


Shock cloud interaction model: HMD simulation (Inoue+12)

Gas5clumps5are5rimJbrightened5in5synchrotron5XJrays.5"5strong5turbulence5+5B5field5may5accelerate5the5CRs5(e.g.,5HS+15)

Inoue+12

ISM5density MagneBc5field5strength


Shock cloud interaction model: Gamma-ray spectrum


Leptonic5model5(Abdo+11)L72 S. Gabici and F. A. Aharonian

Figure 1. Spectrum of CRs in the SNR shell (dotted line) and inside aclump that entered the shock at tc = 1400, 1500 and 1550 yr (solid lines 1,2 and 3, respectively).

where B = 100 B−4 µG is the magnetic field in the turbulent layer,Ltr = 0.05 Ltr,−1.3 pc its thickness and E = E12 TeV the particleenergy. For a given clump, the minimum energy of the particles thatcan penetrate is given by the equation τd = tage − tc. A significantlyfaster CR diffusion is expected outside of the transition region,both inside the clump, where ion–neutral friction is expected toheavily damp magnetic turbulence, and in the SNR shell, where themagnetic field strength and turbulent level are significantly smaller.

The equation that regulates the time evolution of the total numberof CRs inside a clump Ncl(E) is then

∂Ncl(E)∂t

= (Vcl/Vsh)NCR(E) − Ncl(E)τd

, (6)

where Vcl = (4π/3)L3c and Vsh are the volumes of the clump and

of the SNR shell, respectively. The total volume of the clumps istaken to be much smaller than Vsh, to ensure the validity of equation(4). Moreover, Vcl is assumed to be constant in time (i.e. no CRadiabatic energy losses) and proton–proton interaction energy lossesare neglected since they operate on a time tpp ∼ 5 × 104n−1

c,3 yr,longer than the age of the SNR. Finally, the volume Vsh filled byCRs is taken to be the shell encompassed between the SNR forwardshock and the contact discontinuity. The exact position of the contactdiscontinuity depends on several physical parameters (e.g. Orlandoet al. 2012), and is typically of the order of ≈0.9Rs.

The dotted line in Fig. 1 represents the current CR density in theSNR shell as a function of the particle energy. It has been computedfrom equation (4) after assuming a CR acceleration efficiency ofη = 0.1 and a magnetic field in the turbulent layer of B−4 = 1.2.An exponential cutoff at Emax = 150 TeV has been multiplied to thesolution of equation (4) to mimic the escape of the highest energyCRs from the shock. The CR density inside clumps is derived fromequation (6) and plotted with solid lines. Lines 1, 2 and 3 referto a clump that entered the SNR shock 1400, 1500 and 1550 yrafter the supernova explosion, respectively. Clumps that entered theSNR at tc ≈ 1400 yr are encountering now the contact discontinuity.

Figure 2. Gamma-rays from RX J1713.7−3946. The emission from theclumps is shown as a solid line, while the dashed line refers to the emissionfrom the diffuse gas in the shell. Data points refer to Fermi and HESSobservations.

Clumps that entered the SNR earlier either are disrupted by plasmainstabilities at the contact discontinuity or, if they survive, enter aregion characterized by a low density of CRs and quickly becomethemselves devoid of CRs due to their diffusive escape.

The spectrum of CRs inside the clumps has a characteristic peakat energies of ≈10 TeV. At energies larger than that of the peak, thespectra of the CRs in the clumps and in the SNR shell coincide. Thisis because at large energies diffusion becomes important over timessmaller than the residence time of clumps in the shell, allowing fora rapid equilibration of CR densities. On the other hand, CRs withenergies smaller than that of the peak diffuse too slowly to effec-tively penetrate the clumps. This explains the deficit of CRs withenergies below ≈10 TeV in the clumps. The very hard spectral slopefound below the peak is an effect of the steep energy dependenceof the Bohm diffusion coefficient. The position of the peak movestowards larger energies for clumps that enter later the SNR shock,as can be inferred by equation (5) and the discussion that follows it.

The hadronic gamma-ray emission from all the dense clumps inthe shell is plotted as a solid line in Fig. 2. The gas density withinclumps is nc,3 = 1 and the density of clumps is 3 pc−3, whichimplies a total mass in the clumps within the SNR shell of 550 M⊙and a clump volume filling factor of ≈0.01. The distance to the SNRis 1 kpc. The prediction is in agreement with Fermi and HESS data.The gamma-ray emission from CR interactions in the low-densitydiffuse gas swept up by the SNR is plotted as a dashed line, andshown to be subdominant. The contribution from inverse Comptonscattering from electrons accelerated at the SNR is expected to benegligible, if the magnetic field is !10 µG.

Secondary electrons are also produced in proton–proton interac-tions in the dense clumps. Their production spectrum is similar inshape to that of gamma-rays (Fig. 2), with a normalization smallerby a factor of ≈2 and particle energies larger by the same fac-tor. Thus, the peak of electron production happens at an energyof ≈2 TeV. Such electrons escape the clump in a time ∼200 yr(equation 5), which is shorter than both synchrotron and

MNRASL 445, L70–L73 (2014)

at Nagoya U

niversity on Novem

ber 3, 2014http://m

nrasl.oxfordjournals.org/D

ownloaded from

Hadronic5model5(Gabici5&5Aharonian514)

B ~100 µGB ~10 µG

The Astrophysical Journal, 744:71 (15pp), 2012 January 1 Inoue et al.

scenario. However, in our shock–cloud interaction model, aswe discuss in the following, the spectrum may deviate fromthis conventional spectrum. Because of the heavy shock stallin the clouds, particle acceleration at the transmitted shockwave would be inefficient and most of the particles could beaccelerated at the primary shock in the diffuse gas. The high-energy nuclei accelerated at the primary shock interpenetratediffusively to nearby shocked dense clumps. The amount of theneutral pions generated through the collisions of acceleratednuclei and matters in the clouds, which eventually decay andemit gamma rays (Issa & Wolfendale 1981; Naito & Takahara1994; Aharonian et al. 1994), is proportional to the mass of thedense clumps illuminated by the high-energy nuclei. Since thepenetration depth into the clouds would depend on the energy ofaccelerated particles, the hadronic gamma-ray spectrum can bedeviated from that of the conventional model supposing uniformISM (see also Zirakashvili & Aharonian 2010).

In our simulations, the clouds formed by the thermal insta-bility have a sheet-like structure. Using the penetration depthlpd of the accelerated protons into the cloud, the mass of thecloud illuminated by the high-energy protons is M ∼ R2 lpd ρ,where ρ ≃ 40 cm−3 is the density of clouds and R is the scaleof the cloud sheet ∼1 pc. This indicates that the mass of theinterpenetrated region of the cloud is proportional to the pene-tration depth as long as the penetration depth is smaller than thethickness of the cloud sheet ≃0.1 pc. However, as discussed inSection 4.3, low-density cloud envelope would be wiped out bythe stellar wind. In the case of RX J1713.7−3946, the cloudsthat remain in the SNR would be dense molecular cloud cores/clumps as suggested in Equation (8) and observationally in Sanoet al. (2010). If the dense molecular cloud cores are gravitation-ally bound, collapsing objects, the density structure becomesρ(r) = ρ0 (r/r0)−2 (Larson 1969; Penston 1969; Masunaga &Inutsuka 2000; Andre et al. 2000). This is indeed the case ofthe “peak C” of CO core embedded in RX J1713.7−3946 (Sanoet al. 2010). In this case the mass illuminated by the high-energyprotons can be written as

M =! R

R−lpd

4π r2 ρ0 (r/r0)−2 dr = 4π ρ0 r20 lpd. (11)

If the cores are gravitationally unbound, pressure-confinedobjects, the density would be approximately constant (Bonner1956; Ebert 1955; Alves et al. 2001), and the illuminated masscan be written as

M =! R

R−lpd

4π r2 ρ dr = 4π ρ

3

"3 R2 lpd − 3 R l2

pd + l3pd

#

≃ 4π ρ R2 lpd for R ≫ lpd. (12)

The above examples indicate M ∝ lpd as long as lpd ≪ R.The radius of the molecular cloud cores is typically 0.1–1 pc,and the scale of dense clumps in molecular clouds is usuallylarger than 0.1 pc because 0.1 pc is the minimum scale ofthe thermal instability (Field 1965; Koyama & Inutsuka 2004;Inoue et al. 2006) and the turbulent flows that form clumps byshock compressions are expected to be subsonic below 0.1 pc(Larson 1981; Heyer & Brunt 2004; Mac Low & Klessen 2004).Note that when the density of clouds is as high as 103 cm−3,the cooling timescale of clouds becomes much smaller thanthe age of RX J1713.7−3946 ∼ 1000 years that substantiallydiminish the scale of shocked region of clouds due to coolingcontraction. However, owing to the slowdown of the shock

speed, the penetration length of the shock wave into the cloud in1000 years is evaluated to be ∼0.01 pc (where we have assumedthe density ratio between the clouds and diffuse gas to be 105

as we discussed in Section 4.3) that is much smaller than thescale of the molecular cloud cores !0.1 pc. Thus, we can stillassume the minimum scale of clouds to be 0.1 pc, even if cloudsare apparently embedded in SNR.

Since the diffusion coefficient for the high-energy particlescan be written as κd = 4 η lg c/3π (Skilling 1975), where lg isthe gyroradius of relativistic particles, the penetration depth dueto the random walk of the particles can be written as

lpd ≃ (κd t)1/2

= 0.1 η1/2$

E

10 TeV

%1/2 $B

100 µG

%−1/2 $tage

103 yr

%1/2

pc,

(13)

where E is the particle energy and tage is the age of the SNR.For the magnetic field strength in the above expression, wehave used the observed typical strength in the dense regionsof molecular clouds with density ∼103 to 104 cm−3 (Crutcher1999). The parameter η = B2/δB2 has large ambiguity. As wehave discussed in Sections 3.3 and 4.3, the Bohm-limit diffusioncan be expected not only around the shocked clouds but also inshocked clouds through the scattering by turbulent magneticfield fluctuations (Beresnyak et al. 2011) and fluctuationsgenerated by the effect of a cold ion beam (Ohira et al. 2009).Also, the observations by Uchiyama et al. (2007) suggest thatη ∼ 1 is realized at least around the clouds. Thus, if we assumeη ∼ 1, the penetration depth for particles with E ! 10 TeV canbe smaller than the minimum scale of clouds ∼0.1 pc and theinterpenetrated mass can be proportional to the square root ofparticle energy M(E) ∝ lpd(E) ∝ E1/2.

In the conventional one-zone model that assumes the amountof target matter creating π0 is independent of the energy ofaccelerated protons, the spectral energy distribution of gammarays becomes N (E) dE ∝ E−p dE above the critical energy forthe π0 creation and below the maximum energy of acceleratednuclei. Here p is the spectral index of the distribution of high-energy nuclei with p = 2 in the conventional DSA theory.While in the shock–cloud interaction model, the amount ofmatter creating π0 depends on the square root of the energy ofaccelerated protons, so that the gamma-ray distribution becomesNγ (E) dE ∝ M(E) E−p dE ∝ lpd(E) E−p dE ∝ E−p+1/2 dE.This yields Nγ (E) ∝ E−1.5 for p = 2: that is consistentwith the recent observation of RX J1713.7−3946 (Abdo et al.2011). This photon index of the hadronic gamma-ray emission(p − 1/2) is the same as that of the inverse Compton emission(p + 1)/2 when p = 2. Thus, the spectra in the two scenariosare indistinguishable. Fortunately, as we discuss below, thesetwo emissions can be distinguished if we focus on the spatialdistribution of gamma rays.

4.5. Spatial Inhomogeneity of Nonthermal Emissions

In the previous section, we have shown that the emissionmechanism of gamma rays cannot be distinguished from thegamma-ray observation alone. In the following, we discuss howwe can clarify the emission mechanism. A significant differ-ence between our shock–cloud interaction model and the con-ventional uniform ISM model is the spatial inhomogeneity. Theuniform ISM model predicts that emissions are spatially corre-lated from microwave to gamma ray irrespective of the emission

11

The Astrophysical Journal, 744:71 (15pp), 2012 January 1 Inoue et al.

scenario. However, in our shock–cloud interaction model, aswe discuss in the following, the spectrum may deviate fromthis conventional spectrum. Because of the heavy shock stallin the clouds, particle acceleration at the transmitted shockwave would be inefficient and most of the particles could beaccelerated at the primary shock in the diffuse gas. The high-energy nuclei accelerated at the primary shock interpenetratediffusively to nearby shocked dense clumps. The amount of theneutral pions generated through the collisions of acceleratednuclei and matters in the clouds, which eventually decay andemit gamma rays (Issa & Wolfendale 1981; Naito & Takahara1994; Aharonian et al. 1994), is proportional to the mass of thedense clumps illuminated by the high-energy nuclei. Since thepenetration depth into the clouds would depend on the energy ofaccelerated particles, the hadronic gamma-ray spectrum can bedeviated from that of the conventional model supposing uniformISM (see also Zirakashvili & Aharonian 2010).

In our simulations, the clouds formed by the thermal insta-bility have a sheet-like structure. Using the penetration depthlpd of the accelerated protons into the cloud, the mass of thecloud illuminated by the high-energy protons is M ∼ R2 lpd ρ,where ρ ≃ 40 cm−3 is the density of clouds and R is the scaleof the cloud sheet ∼1 pc. This indicates that the mass of theinterpenetrated region of the cloud is proportional to the pene-tration depth as long as the penetration depth is smaller than thethickness of the cloud sheet ≃0.1 pc. However, as discussed inSection 4.3, low-density cloud envelope would be wiped out bythe stellar wind. In the case of RX J1713.7−3946, the cloudsthat remain in the SNR would be dense molecular cloud cores/clumps as suggested in Equation (8) and observationally in Sanoet al. (2010). If the dense molecular cloud cores are gravitation-ally bound, collapsing objects, the density structure becomesρ(r) = ρ0 (r/r0)−2 (Larson 1969; Penston 1969; Masunaga &Inutsuka 2000; Andre et al. 2000). This is indeed the case ofthe “peak C” of CO core embedded in RX J1713.7−3946 (Sanoet al. 2010). In this case the mass illuminated by the high-energyprotons can be written as

M =! R

R−lpd

4π r2 ρ0 (r/r0)−2 dr = 4π ρ0 r20 lpd. (11)

If the cores are gravitationally unbound, pressure-confinedobjects, the density would be approximately constant (Bonner1956; Ebert 1955; Alves et al. 2001), and the illuminated masscan be written as

M =! R

R−lpd

4π r2 ρ dr = 4π ρ

3

"3 R2 lpd − 3 R l2

pd + l3pd

#

≃ 4π ρ R2 lpd for R ≫ lpd. (12)

The above examples indicate M ∝ lpd as long as lpd ≪ R.The radius of the molecular cloud cores is typically 0.1–1 pc,and the scale of dense clumps in molecular clouds is usuallylarger than 0.1 pc because 0.1 pc is the minimum scale ofthe thermal instability (Field 1965; Koyama & Inutsuka 2004;Inoue et al. 2006) and the turbulent flows that form clumps byshock compressions are expected to be subsonic below 0.1 pc(Larson 1981; Heyer & Brunt 2004; Mac Low & Klessen 2004).Note that when the density of clouds is as high as 103 cm−3,the cooling timescale of clouds becomes much smaller thanthe age of RX J1713.7−3946 ∼ 1000 years that substantiallydiminish the scale of shocked region of clouds due to coolingcontraction. However, owing to the slowdown of the shock

speed, the penetration length of the shock wave into the cloud in1000 years is evaluated to be ∼0.01 pc (where we have assumedthe density ratio between the clouds and diffuse gas to be 105

as we discussed in Section 4.3) that is much smaller than thescale of the molecular cloud cores !0.1 pc. Thus, we can stillassume the minimum scale of clouds to be 0.1 pc, even if cloudsare apparently embedded in SNR.

Since the diffusion coefficient for the high-energy particlescan be written as κd = 4 η lg c/3π (Skilling 1975), where lg isthe gyroradius of relativistic particles, the penetration depth dueto the random walk of the particles can be written as

lpd ≃ (κd t)1/2

= 0.1 η1/2$

E

10 TeV

%1/2 $B

100 µG

%−1/2 $tage

103 yr

%1/2

pc,

(13)

where E is the particle energy and tage is the age of the SNR.For the magnetic field strength in the above expression, wehave used the observed typical strength in the dense regionsof molecular clouds with density ∼103 to 104 cm−3 (Crutcher1999). The parameter η = B2/δB2 has large ambiguity. As wehave discussed in Sections 3.3 and 4.3, the Bohm-limit diffusioncan be expected not only around the shocked clouds but also inshocked clouds through the scattering by turbulent magneticfield fluctuations (Beresnyak et al. 2011) and fluctuationsgenerated by the effect of a cold ion beam (Ohira et al. 2009).Also, the observations by Uchiyama et al. (2007) suggest thatη ∼ 1 is realized at least around the clouds. Thus, if we assumeη ∼ 1, the penetration depth for particles with E ! 10 TeV canbe smaller than the minimum scale of clouds ∼0.1 pc and theinterpenetrated mass can be proportional to the square root ofparticle energy M(E) ∝ lpd(E) ∝ E1/2.

In the conventional one-zone model that assumes the amountof target matter creating π0 is independent of the energy ofaccelerated protons, the spectral energy distribution of gammarays becomes N (E) dE ∝ E−p dE above the critical energy forthe π0 creation and below the maximum energy of acceleratednuclei. Here p is the spectral index of the distribution of high-energy nuclei with p = 2 in the conventional DSA theory.While in the shock–cloud interaction model, the amount ofmatter creating π0 depends on the square root of the energy ofaccelerated protons, so that the gamma-ray distribution becomesNγ (E) dE ∝ M(E) E−p dE ∝ lpd(E) E−p dE ∝ E−p+1/2 dE.This yields Nγ (E) ∝ E−1.5 for p = 2: that is consistentwith the recent observation of RX J1713.7−3946 (Abdo et al.2011). This photon index of the hadronic gamma-ray emission(p − 1/2) is the same as that of the inverse Compton emission(p + 1)/2 when p = 2. Thus, the spectra in the two scenariosare indistinguishable. Fortunately, as we discuss below, thesetwo emissions can be distinguished if we focus on the spatialdistribution of gamma rays.

4.5. Spatial Inhomogeneity of Nonthermal Emissions

In the previous section, we have shown that the emissionmechanism of gamma rays cannot be distinguished from thegamma-ray observation alone. In the following, we discuss howwe can clarify the emission mechanism. A significant differ-ence between our shock–cloud interaction model and the con-ventional uniform ISM model is the spatial inhomogeneity. Theuniform ISM model predicts that emissions are spatially corre-lated from microwave to gamma ray irrespective of the emission

11

Gas5clumps

Ambient5gas

Inoue+12

[3] Gamma-ray obs. with high sensitivity + angular resolutions


15

By S. Funk (modified)

Young SNRs (age ~2000 yrs)

Current5observaBons5could5not5determine5the5spectra5shape5E5>51005TeV5

?

Cherenkov Telescope Array (CTA)


e5will5be5able5to5determine the spectral shape ��> 100 TeV by using CTA!

現行望遠鏡に比べ� 感度 10倍� エネルギー帯域 10 倍

Cherenkov Telescope Array (CTA)計画

20 GeV-300 TeV

� 角度分解能 2 倍（2分角@1TeV）

14

XJray5photon5index5map5(RXJ1713)HS+15

Cherenkov Telescope Array (CTA): imaging simulations


The origin of gamma-rays can be distinguished by morphological studies

Leptonic origin Hadronic origin

Nakamori,$Katagiri,$HS+15$(ICRC)

!(Galac'c!Ring!Survey!by!Boston/FCRAO;!Jackson!et!al.!2006)

NRO$GPS(2013 ((IR(1)

[4] To reveal an universality


We are extending our comparative studies (ISM, X-rays, Gamma-rays) for the distant SNRs by using high-angular resolution CO data (Δθ~20”)

The Astrophysical Journal Supplement Series, 209:8 (13pp), 2013 November Dempsey, Thomas, & Currie

(a)

Figure 4. Integrated intensity map (!

T ∗Adv) from COHRS (+10◦ < l < 22◦ (a); +22◦ < l < 43◦ (b); +43◦ < l < 55◦ (c)). The color bar shows the scaling in units

of K km s−1.(A color version of this figure is available in the online journal.)

also shows the data on a linear scale where multiple peaks can beseen. These peaks reflect the wide range of observing conditionsin which these tiles were taken. The noisiest tiles, responsiblefor the peak around 1 K, are typically the result of observing inpoor weather or at low elevations. A further factor contributingto high noise is the integration time per pixel. This is dependenton the number of HARP receptors included in the reduction. Alow integration time may be the result of dead receptors on theinstrument (which has varied over the duration of this survey),or entire receptors (or time slices) that were later excluded bythe reduction software.

The mean rms for each re-binned tile is given in Table 3. Thisvalue is a mean of all the pixel values calculated for Figure 2.In some cases the final tiles include maps observed on differentnights and in different conditions resulting in mean rms valuesthat are not representative of the individual maps. The spatialnoise distribution across the survey, and within each tile, can bebest visualized in Figure 3, a two-dimensional map of the noisetemperature. This is useful for highlighting the relative qualityof the individual maps that contribute to each tile.

Some regions are seen to display a faint residual tartan pattern(for example between l = 17◦ and l = 18◦ in Figure 4). This

6

NRO545Jm5GalacBc5plane5survey5(FUGIN):512CO,513CO,5C18O5J$=51–05

JCMT5GalacBc5plane5survey:512CO5J$=53–2Coverage:510°5<5l$<565°,5|b|5<50.5°5

Coverage:510°5<5l$<550°,|b|5<51°5+5outer5rim55

R51.25

SNR distributions toward the survey area


!  67 SNRs are located within the FUGIN survey area!" 40 % of SNRs are emitted by diffuse X-rays

Image:5NANTEN5GPS5(Mizuno5&5Fukui52004),5Red:5SNR5boundary,5Green:5FUGIN5area

Preliminary results of 12CO J = 3–2


!  These SNRs are ~10 times smaller than RX J1713 (size ~ 60 arcmin), !while we could identify the interacting molecular clouds

Image JCMT512CO5J5=53–25integrated5intensity,55Contours:5Chandra5broad5band5XJrays5

Preliminary5Result… Preliminary5Result…

of N 157B of 100′′ (19) is of the order of the H.E.S.S.angular resolution. Further significant g-ray emis-sion is detected to the southwest of N 157B.A likelihood fit of a model of two g-ray sources

to the on-source and background sky maps es-tablishes the detection of a second source at anangular distance of 9′ (corresponding to 130 pc ata distance of 50 kpc) from N 157B. The modelconsisting of two sources is preferred by 8.8 SDover the model of one single source. Figure 1Cshows an x-ray image with overlaid contours ofconfidence of the source position. The position ofthe second source [right ascension = 5h35m(55 T 5)s,declination = −69°11′(10 T 20)′′, equinox J2000,1 SD errors] coincides with the superbubble

30 Dor C, the first such source detected in VHE grays, and thus represents an additional source classin this energy regime. A g-ray signal around theenergetic pulsar PSR J0540-6919 is not detected,despite the presence of an x-ray luminous PWN(20). A flux upper limit (99% confidence level) isderived at Fg (>1 TeV) < 4.8 × 10−14 ph cm−2 s−1.Along with the clear detection of N 157B and

30 Dor C, evidence for VHE g-ray emission isobserved from the prominent SNR N 132D (Fig.1D). The emission peaks at a significance of about5 SD above a background that is estimated from aring around each sky bin. At the nominal positionof the SNR, 43 g rays with a statistical signifi-cance of 4.7 SD are recorded.

The g-ray spectra of all three objects are welldescribed by a power law in energy, F(E) = d3N/(dE dt dA) =F0 (E/1 TeV)

−G (where E is energy, tis time, and A is detector area) (Fig. 2). The best-fit spectral indices and integral g-ray luminosi-ties are summarized in Table 1.Even with a deep exposure of 210 hours, sig-

nificant emission from SN 1987A is not detected,and we derive an upper limit on the integral g-rayflux of Fg(>1TeV) < 5.6 × 10−14 ph cm−2 s−1 at a99% confidence level.

Discussion of individual sources

The three VHE emitters belong to different sourceclasses, and their energy output exceeds or at least

SCIENCE sciencemag.org 23 JANUARY 2015 • VOL 347 ISSUE 6220 407

Fig. 1. Sky maps of the LMC. (A) Optical image of the entire LMC (55). Theboxes denote the regions of interest discussed in this paper. Colors denotelevels of 3, 5, 10, and 20 SD statistical significance of the g-ray signal. (B) VHEg-ray emission in the region around N 157B.The green lines represent contoursof 5, 10, and 15 SD statistical significance of the g-ray signal. (C) X-Ray Multi-Mirror Mission (XMM-Newton) x-ray flux image of the region of 30 Dor C.Thesuperimposed cyan lines represent contours of 68, 95, and 99% confidence

level of the position of the g-ray source. Diamonds denote the positions of thestar clusters of the LH 90 association. See the supplementary materials fordetails on the x-ray analysis. (D) VHE g-ray emission in the region around N132D. The green lines represent contours of 3, 4, and 5 SD statistical sig-nificance.The background of the g-ray emission [in (B) and (D)] was obtainedusing the ring background method (56). The resulting excess sky map issmoothed to the angular resolution of the instrument.

RESEARCH | REPORTS

N132D

N157B

305Dor5C

For donor MT-01, input cell numbers were TH1*, 1.5 × 106; TH17,1.5 × 106; CXCR3+CCR4+ TH, 1.2 × 106. For donor MT-02,input cell numbers were TH1*, 1.5 × 106; TH17, 1.4 × 106;CXCR3+CCR4+ TH, 1.1 × 106. Recovered CFSElo cells on day6 for donor MT-01 were TH1*, 2.2 × 106; TH17, 0.95 × 106;CXCR3+CCR4+ TH, 2.0 × 106. For donor MT-02, these numberswere TH1*, 2.4 × 106; TH17, 0.7 × 106; CXCR3+CCR4+ TH, 1.7 × 106.

28. C. S. Lindestam Arlehamn et al., PLOS Pathog. 9, e1003130 (2013).29. T cell subsets from four donors were isolated and stimulated with

TT. Input cell number for each subset was 2.5 × 106 (donor TT-01),1.5 × 106 (donors TT-02 and TT-03), and 1.1 × 106 (donor TT-04).

30. Naïve T cells were isolated from three donors. Input cellnumbers were 3.2 × 106 (donor N-01), 6.6 × 106 (donor N-02),and 8.8 × 106 (donor N-03).

31. In control experiments, the purity of the sorted populationswas confirmed by measuring IFNG, IL4, and IL17 mRNAs byquantitative PCR immediately after sorting.

32. B. Pulendran, J. Immunol. 174, 2457–2465 (2005).33. H. Sigmundsdottir, E. C. Butcher, Nat. Immunol. 9, 981–987

(2008).

34. C. R. Plumlee, B. S. Sheridan, B. B. Cicek, L. Lefrançois,Immunity 39, 347–356 (2013).

35. M. E. Birnbaum et al., Cell 157, 1073–1087 (2014).36. R. M. Welsh, L. K. Selin, Nat. Rev. Immunol. 2, 417–426 (2002).37. D. Mason, Immunol. Today 19, 395–404 (1998).38. L. Rivino et al., J. Exp. Med. 200, 725–735 (2004).39. Production of cytokines was measured on day 6 in the

supernatants of 5 × 104 naïve CD4 T cells stimulated withplate-bound CD3 and CD28 antibodies. Mean values and SEM(n = 3) were as follows: IFN-g, 414 T 270.2; IL-17, 30.7 T 3.3; IL-22,391.6 T 129.2; IL-4, 3.6 T 3.6; IL-5, 9.6 T 5.4; IL-13, 91 T 20.5.

ACKNOWLEDGMENTS

We thank D. Jarrossay for cell sorting; O. Petrini, C. Fragoso,and A. Sette for providing valuable reagents; the AdaptiveBiotechnologies team for technical support; and S. Monticelli forcritical reading of the manuscript. The data presented in thismanuscript are tabulated in the main paper and in the supplementarymaterials. All TCR sequences are available in the supplementarymaterials as .txt files. FASTA format files are available at downloads.

adaptivebiotech.com (username: SallustoSciDec2014data; password:di2PhaiT). The sequences have been deposited in the GeneExpression Omnibus (accession numbers xxxxxx). Supported byEuropean Research Council grant 323183 PREDICT (F.S.), SwissNational Science Foundation grants 149475 (F.S.) and 147662 (A.L.),and European Commission grants FP7-HEALTH-2011-280873,ADITEC (A.L. and F.S.) and FP7-HEALTH-2013-601958, SUPERSIST(T.N.S.). The Institute for Research in Biomedicine is supported bythe Helmut Horten Foundation.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/347/6220/400/suppl/DC1Materials and MethodsFigs. S1 to S7Tables S1 and S2References (40–44)Data files

1 September 2014; accepted 25 November 2014Published online 4 December 2014;10.1126/science.1260668

REPORTS◥

ASTROPHYSICS

The exceptionally powerfulTeV g-ray emitters in the LargeMagellanic CloudThe H.E.S.S. Collaboration*†

The Large Magellanic Cloud, a satellite galaxy of the Milky Way, has been observed with theHigh Energy Stereoscopic System (H.E.S.S.) above an energy of 100 billion electron voltsfor a deep exposure of 210 hours. Three sources of different types were detected: thepulsar wind nebula of the most energetic pulsar known, N 157B; the radio-loud supernovaremnant N 132D; and the largest nonthermal x-ray shell, the superbubble 30 Dor C. Theunique object SN 1987A is, unexpectedly, not detected, which constrains the theoreticalframework of particle acceleration in very young supernova remnants. These detectionsreveal the most energetic tip of a g-ray source population in an external galaxy and providevia 30 Dor C the unambiguous detection of g-ray emission from a superbubble.

The origin of cosmic rays (CRs), the veryhigh energy (VHE) (≳1011 eV) and ultrahighenergy (≳1018 eV) particles that bombardEarth, has puzzled us for more than a cen-tury. Much progress has been made during

the past decade due to the advent of VHE g-raytelescopes. These telescopes detect ∼1011 to1014 eV g rays from atomic nuclei (hadronic CRs)collisions with local gas or from ultrarelativisticelectrons/positrons (leptonic CRs), which produceg-ray emission by upscattering low-energy back-ground photons (1). Indeed, a survey of the innerpart of the Milky Way with the High Energy Ste-reoscopic System (H.E.S.S.), an array of imagingatmospheric Cherenkov telescopes (2), revealed apopulation of supernova remnants (SNRs) and

pulsar wind nebulae (PWNe) emitting g rays withenergies in excess of 100 GeV (3).Here,we report on VHE g-ray sources detected

outside theMilkyWay, namely in the LargeMag-ellanic Cloud (LMC). This satellite galaxy of theMilky Way has a stellar mass of about 4% of theMilkyWay (4, 5). Located at a distance of ~50 kpc(6), it is an irregular galaxy seen almost face-on(7). Consequently, source confusion is much lessof a problem than for the inner Milky Way, andthere is less uncertainty in the distances of thesources. The LMC stands out among nearby gal-axies for its high star formation rate per unitmass, which is about a factor of five higher thanin the Milky Way (8, 9), and contains the bestexample of a local starburst, the Tarantula Nebula.The LMC also harbors numerous massive stellarclusters and SNRs. Among the SNRs is a uniquesource, SN 1987A, the remnant of the nearestsupernova observed in modern times (10).High-energy g-ray emission from the LMCwas

detected by the Energetic Gamma Ray Experi-

ment Telescope (EGRET) (11) and, more recently,by the Fermi Large Area Telescope (LAT) (12),which revealed diffuse emission with an exten-sion of several degrees in diameter, tracingmassive star-forming regions. VHE g-ray tele-scopes, like H.E.S.S., besides providing infor-mation on much higher energy CRs, have anangular resolution of a few arcminutes, whichis substantially better than Fermi-LAT’s reso-lution at g-ray energies <10 GeV. The goodangular resolution allows H.E.S.S. to identifyindividual sources in the LMC. As we will detailbelow, a deep H.E.S.S. observation revealed threeluminous sources in the LMC: the superbubble30 Dor C, the energetic PWN N 157B, and theradio-loud SNR N 132D. Of these sources, onlyN 157B was detected previously in a 47-hourexposure (13). The observations extend the scopeof VHE g-ray astronomy by providing examplesof sources from a population outside the MilkyWay. N 157B and N 132D belong to known g-raysource classes, but both have distinguishing char-acteristics, N 157B being powered by the mostenergetic young pulsar, and N 132D being one ofthe oldest VHE g-ray emitting SNRs. The super-bubble 30 Dor C, however, provides an unam-biguous detection of a superbubble in VHE grays. Conspicuously absent from our list of threesources is SN 1987A, despite predictions that itshould be a bright g-ray source (14, 15).

H.E.S.S. observations

We report on a deep, 210-hour H.E.S.S. exposure,targeted at the region of the Tarantula nebula—corresponding to 30Doradus (30 Dor)—the largeststar-forming region in the Local Group of galaxies.We reconstructed g-ray showers with an image-fitting analysis (16) and cross-checkedwithamulti-variate analysis based on image parameterization(17, 18), with consistent results. In both analyses,a cut on the uncertainty of the reconstructed g-raydirection indicated an angular resolution of ~0.05°.Figure 1A shows an optical image of the LMC

overlaid with TeV g-ray point-source significancecontours. In this data set, 613 g rays with a statis-tical significance of 33 SD are detected from thePWNN 157B. Figure 1B provides a close-up view ofthe g-ray emission from N 157B. The diameter

406 23 JANUARY 2015 • VOL 347 ISSUE 6220 sciencemag.org SCIENCE

*The full list of authors and affiliations is at the end of the paper.†Corresponding author. E-mail: [email protected] (N.K.);[email protected] (C.-C.L.); [email protected] (M.M.); [email protected] (S.O.); [email protected] (J.V.)

RESEARCH

on

Janu

ary

23, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on

Janu

ary

23, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on

Janu

ary

23, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on

Janu

ary

23, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on

Janu

ary

23, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on

Janu

ary

23, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on

Janu

ary

23, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m Next5targets5

Magellanic5SNRs5&5Superbubbles

Chandra5XJray535color5Images

Images:5Mopra5CO51J055Contours:5Chandra5XJray5

Preliminary5Results…

ano+15b5

Image:5(a)5Chandra5XJrays,5(b)5Mopra5CO51J05(MAGMA:5Wong+11)5Contours:5Chandra5XJrays5(0.5J7.05keV)

Magellanic SNR N132D Mopra CO1-0, HS+15b


30 Dor C

J5Superbubble5in5305Dor55

J5NonJthermal5XJrays5

J5TeV5GammaJrays5

J5Containing5young5SNR555(Age:52.2−4.95kyr,5Kavanagh+14)5

HS+16 in prep.

30 Dor C



J5TeV5GammaJrays5


Image:5CO5Contours5:5XJrays5(0.5J7.05keV)

HS+16 in prep.

30 Dor C



J5TeV5GammaJrays5


Image:5HI5(Kim+03)5Contours5:5XJrays5(0.5J7.05keV)

HS+16 in prep.

ISM5associated5superbubbles

Preliminary5Result… Preliminary5Result…

[1] Origin of gamma-rays (hadronic and/or leptonic)" Spatial comparison between the ISM and gamma-rays

[2] Acceleration models including more realistic conditions" Shock-cloud interaction model " Other acceleration mechanisms ?

(non-liner, 2nd order Fermi, magnetic reconnection etc…)

[3] Gamma-ray obs. with high sensitivity + angular resolutions" Cherenkov Telescope Array (CTA)

[4] To reveal an universality " Extend to the distant sources including the LMC SNRs

Summary outstanding issues solutions

MulBJmessenger5search5plays5an5essenBal5role5in5understanding5the5origin5of5galacBc5cosmicJrays!

Documents

A multi-messenger search for the origin of cosmic-rayscta.scphys.kyoto-u.ac.jp/workshop/CTA-J/2015/presentations/22.ctaws2015_sano_20160114...H.E.S.S. Collaboration et al.: Detailed