Study of the Interstellar Medium Gas and Cosmic …mizuno/GLAST/...T. Mizuno et al. 1/16 Study of the Interstellar Medium Gas and Cosmic Rays using GeV Rays Oct. 24th, 2017@MW workshop

1T. Mizuno et al. /16

Study of the Interstellar Medium Gas and Cosmic Rays using

GeV RaysOct. 24th, 2017@MW workshop in

KagoshimaT. Mizuno (Hiroshima Univ.) on

behalf of the Fermi-LAT Collaboration

2017-10_KagoshimaMW_ISM.ppt


GeVガンマ線を用いた

天の川銀河の星間ガスと宇宙線の研究

2017年10月24日@鹿児島大学(天の川銀河研究会)

水野恒史 (広島大学) on behalf of the Fermi-LAT Collaboration


Motivation: ISM and CRs

• Interstellar Medium (ISM) gas plays an important role in physical processes in the Milky Way (e.g., star formation)– It has been traced by HI line, CO line, and dust emission or

absorption• If the ISM gas distribution is know accurately, cosmic-rays

(CRs) can be studied through GeV -ray observation

Fermi-LAT 4 year all-sky map (>1 GeV)

Vela

Crab

Geminga

Planck dust map (353)

Cepheus & Polaris

MBM 53,54,55

R CrA

Chamaeleon

OrionTaurus

3C454.3

∝


GeV rays as a Tracer of CRs

• Cross section is know accurately for all processes• In GeV band 0 decay -ray emission is dominant, making

it a powerful probe of CR protons ( ∝ )

rays = CRs x ISM gas (or ISRF)

Fermi-LAT (GeV) or imaging air Cherenkov telescopes (TeV)

CRs

Interstellar Medium


Uncertainty of the ISM: Dark Gas

• A significant amount of ISM gas not well traced by standard 21 cm and 2.6 mm lines (Grenier+05) has been recognized recently

• This “dark gas”(DG) has been traced by dust observation data, but the procedure has not been established yet (e.g., choice of the tracer)

Residual rays in the Chamaeleon clouds when fitted by N(HI) and WCO

Residual gas inferred by dust (fitted by N(HI) and WCO) (mag)

()

Chamaeleon Molecular Cloud:MH2,CO ~ 5x103 MsolarMDG ~ 2x104 Msolar

Ackermann+12, ApJ 755, 22 (CA: Hayashi, TM)


Fermi-LAT Study of the ISM (and CRs)

MBM 53-55

Pegasus Loop

• The procedure to convert dust data into N(Htot) has not been established yet, giving uncertainty of the dark gas distribution

• Under the assumption of a uniform CR density in local, we can use GeV rays as a robust tracer of N(Htot) since ∝

• We can use Planck dust model as a precise tracer of N(Htot) and apply corrections to match to -ray data

• Here we present the study of the MBM53-55 clouds and the Pegasus loop (Mizuno+16)

dust temperature (Td) N(HIthin) (1020 cm-2) WCO (K km s-1)


Fitting Procedure

• Since ISM is optically thin to rays, under the assumption of uniform CR density (valid in high latitude), the -ray intensity is a linear combination of templates

, , ·

, ,

, ,

· . . , ·

Fermi ray Planck dust, LAB HI, WCO, etc.

IC model (e.g., galprop)(Energy dependent terms as free parameters)

Fit quality tells us what tracer is betterCoefficients ( ) tell us as properties


Initial N(Htot) templates ( R, 353)

N(Htot) template (∝ R) (1020 cm-2) N(Htot) template (∝ 353) (1020 cm-2)

• Dust is mixed with gas and has been used as a tracer of N(Htot)– But what quantity should we use?

• We prepared two template N(Htot) maps (∝ radiance (R) and optical depth at 353 GHz (353)) and correlated them with rays– Different contrast in N(Htot) distribution since two tracers show

different and Td-dependent correlations with WHI– The map ∝ R gives better fit to -ray data


Initial N(Htot) templates ( R, 353)


• We prepared two template N(Htot) maps (∝ radiance (R) and optical depth at 353 GHz (353)) and correlated them with rays– Different contrast in N(Htot) distribution since two tracers show

different and Td-dependent correlations with WHI

(Areas with Wco>1.1 K km/s are masked)(Lines are best-fit linear relations in Td>21.5 K to construct N(Htot) templates in the optically thin condition

R, small scatter 353, large scatter

(see also Fukui+14,15, Planck Collab. 2014)


Dust Temperature Dependence

• Even though R-based N(Htot) is preferred by -ray data, true N(Htot) could be appreciably different

• We split N(Htot) template map into four based on Td and fit -ray data with scaling factors (∝ / , ) independently varying– Scale factors should not depend on Td if N(Htot) ∝ D (R, 353) and UCR

is uniform• Fit improves significantly but scale factors depends on Td

– Negative correlation with R implies underestimate of N(Htot) in low Td

Dust temp.

Scale Factor(∝ / , )


Td-Dependence Correction

• Scale factors ( / , ) depends on Td


• We used -ray data to compensate for the dependence and obtained an accurate estimate of N(Htot)

, ..

, (N(Htot, mod)=N(Htot,R) in Td>20.5 K)(coeff. determined to fit to -ray data)

N(Htot, mod) inferred by -rays (1020 cm-2) N(Htot, mod)-N(HIthin) (1020 cm-2)


Td-Dependence Correction

• Scale factors ( / , ) depends on Td


• We used -ray data to compensate for the dependence and obtained an accurate estimate of N(Htot)

, ..

, (N(Htot, mod)=N(Htot,R) in Td>20.5 K)(coeff. determined to fit to -ray data)

WCO (K km s-1)N(Htot, mod) inferred by -rays (1020 cm-2)


Discussion of the ISM (1)

• The correlation between WHI and the corrected N(Htot) map– Moderate scatter due to dark gas– Ts<100 K is inferred in the optically-thick HI scenario

(Areas with Wco>1.1 K km/s are masked)


Discussion of the ISM (2)

• Integral of gas column density (∝ Mgas) as a function of Td for N(Htot), N(HIthin), N(Htot)-N(HIthin) and 2N(H2,CO)– MDG is ~25% of MHI,thin and ~ 5 x MH2,CO, ~10 times larger than model

predictions of CO-dark H2 scenario (Wolfire+10, Smith+14)– MDG differs by a factor of ~4 if we use only R (or 353); The correction

based on -ray data is crucial

1022 cm-2 deg2 corresponds to ~740 Msun for d=150 pc

MDG, = ~4 x MDG, R~1/4 x MDG, 353

N(HDG) = N(Htot)-N(HIthin)-2N(H2,CO)


Discussion (HI emissivity or ICR)

• HI emissivity spectrum is compared with model curves based on the local interstellar spectrum (LIS) and results by relevant LAT studies employing a conventional template-fitting method

• Our spectrum agrees with the model for LIS with m (nuclear enhancement factor)=1.45, while previous LAT studies favor m=1.84

Systematic study of other high-latitude regions is necessary to better understand the ISM and CRs(e.g., Mizuno+17@Fermi Symp.)

(∝


Summary

• An accurate estimate of the ISM distribution is important, but the uncertainty is still large

• We carried out a joint Fermi-LAT & Planck study of MBM 53-55 clouds and the Pegasus loop– We used rays as a robust tracer of N(Htot) and Planck dust

model as a precise tracer– We obtained physical quantities of the ISM and CRs

• Ts for optically-thick HI scenario (<100 K)• Mass of dark gas (~25% of MHI,thin and ~ 5 x MH2,CO)• Local CR density (~20% lower than relevant studies)

• Systematic study of other high-latitude regions is important and underway

Thank you for your Attention


References (Fermi-LAT Studies of Diffuse Emission in MW)

• Abdo+09, ApJ 703, 1249 (CA: TM)• Abdo+09, PRL 103, 251101 (CA: Johanneson, Porter, Strong)• Abdo+10, ApJ 710, 133 (CA: Grenier, Tibaldo)• Abdo+10, PRL 104, 101101 (CA: Ackermann, Porter, Sellerholm)• Ackermann+11, ApJ 726, 81 (CA: Grenier, TM, Tibaldo)• Ackermann+12, ApJ 750, 3 (CA: Johanneson, Porter, Strong)• Ackermann+12, ApJ 755, 22 (CA: Hayashi, TM)• Ackermann+12, ApJ 756, 4 (CA: Kamae, Okumura)• Ackermann+12, A&A 538, 71 (CA: Grenier, Tibaldo)• Ackermann+14, ApJ 793, 64 (CA: Franckowiak, Malyshev, Petrosian)• Casandjian 2015, ApJ 806, 240• Ackermann+15, ApJ 799, 86 (CA: Ackermann, Bechtol)• Tibaldo+15, ApJ 807, 161 (CA: Digel, Tibaldo)• Planck Collaboration 2015, A&A 582, 31 (CA: Grenier)• Ajello+16, ApJ 819, 44 (CA: Porter, Murgia)• Acero+16, ApJS 223, 26 (CA: Casandjian, Grenier)• Mizuno+16, ApJ 833, 278• Remy+17, A&A 601, 78 (CA: Grenier, Remy)


References (others)

• Atwood+09, ApJ 697, 1071• Dame+01, ApJ 547, 792• Fukui+14, ApJ 796, 59• Fukui+15, ApJ 798, 6• Grenier+05, Science 307, 1292• Kalberla+05, A&A 440, 775• Kiss+04, A&A 418, 131• Planck Collaboration XI 2014, A&A 571, 11 • Strong & Moskalenko 98, ApJ 509, 212• Welty+89, ApJ 346, 232• Wolfire+10, ApJ 716, 1191• Smith+14, MNRAS 441, 1628• Yamamoto+03, ApJ 592, 217• Yamamoto+06, ApJ 642, 307• Ysard+15, A&A 577, 110


Backup Slides


Processes to Produce rays

rays = CRs x ISM gas (or ISRF)

a powerful probe to study ISM and CRs

Abdo+09, PRL 103, 251101(CA: Porter, Johanneson, Strong)

(Isotropic)

Inverse Compton,~2.1

Bremsstrahlung, ~3.2 above a few GeV

0 decay, ~2.7 above a few GeV

Pro: optically-thin, “direct” tracer of all gas phasesCon: low-statistics, contamination (isotropic, IC), depend on CR density=> need to be complemented with other gas tracers

-ray data and model(mid-lat. region)

We can distinguish gas-related rays from others based on the spectrum and morphology


Uncertainty of ISM: Dark Gas

• Fermi revealed a component of ISM not measurable by standard tracers (HI 21 cm, CO 2.6 mm), confirming an earlier claim by EGRET (Grenier+05)

• Mass of “dark gas” is comparable to or greater than that of H2traced by WCO

Residual rays in Chamaeleon when fitted by N(HI)+WCO

Molecular cloud H2 mass traced by WCO (Msolar)

“dark gas” (Msolar)

Chamaeleon ~5x103 ~2.0x104

R CrA ~103 ~103

Cepheus & Polaris ~3.3x104 ~1.3x104

Orion A ~5.5x104 ~2.8x104

Ackermann+12, ApJ 755, 22 (CA: Hayashi, TM); Ackermann+12, ApJ 756, 4 (CA: Okumura, Kamae)See also Planck Collaboration 2015, A&A 582, 31 (CA: Grenier)

MDG/MH2,CO

~4~1~0.4~0.5


Uniform CR Density

• Direct observation of CRs (path length of ~5 g/cm2, anisotropy of ~0.1%) indicates that they propagate through diffusion and have a small gradient at a ~100 pc scale

• Smooth CR distribution is supported by -rays

sun

Remy+17

HI e

mis

sivi

ty(∝

/)


WHI-Dust Relation (1)


• We examined correlations btw. WHI and two dust tracers (radiance (R) and optical depth at 353 GHz (353)) (see also Fukui+14,15, Planck Collab. 2014)

– Two tracers show different and Td-dependent correlations with WHI

(Areas with Wco>1.1 K km/s are masked)(Lines show best-fit linear relations in Td>21.5 K used to construct N(Htot) templates in the optically thin condition

R, small scatter 353, large scatter


WHI-Dust Relation (2)

N(Htot) template (∝ R) (1020 cm-2) N(Htot) template (∝ 353) (1020 cm-2)


• We examined correlations btw. WHI and two dust tracers (radiance (R) and optical depth at 353 GHz (353)) (see also Fukui+14,15, Planck Collab. 2014)

– Two tracers show different and Td-dependent correlations with WHI• Different contrast in N(Htot) template maps (∝ R, 353). The map ∝ R gives

better fit to -ray data


Possible Explanation of Td Dependence (1)

• We found, from -ray data analysis, neither the radiance nor 353 are good tracers of N(Htot)– Even though the interstellar radiation field (ISRF) is uniform in

the vicinity of the solar system, the radiance (per H) could decrease as the gas (and dust) density increases, because the ISRF is more strongly absorbed by dust. This will cause a correlated decrease in the Td and the radiance (per H).

Ysard+15, Fig.2(Radiance per H vs. Td for several choices of ISRF hardness. Both radiance and Tddecrease as the ISRF is abosrbed)


Possible Explanation of Td Dependence (2)

• We found, from -ray data analysis, neither the radiance nor 353 are good tracers of N(Htot)– In the optically-thin limit, I = B(Td) = N(Htot) B(Td), where

and are the optical depth and the dust opacity (cross section) per H, respectively. depends on the frequency and is often describes as a power law, giving I = (/0) B(Td) (modified blackbody, ~1.5-2).

– Therefore, IF the dust cross section is uniform, ∝ N(Htot) and we can measure the total gas column density by measuring the dust optical depth at any frequency (e.g., 353).

‒ However, dust opacity is not uniform but rather anti-correlates with Td as reported by Planck Collaboration (2014).

Relation btw. Tdust and in MBM & Pegasus


Results by a Conventional Template-Fitting Method

• We also employed a conventional template-fitting method– Fit gamma-ray data with N(HIthin) map, WCO map, Rres map (template of

dark gas) with isotropic, Inverse Compton and point sources– MDG (shown by red dotted histogram) is ~50% smaller than that we

obtained through Td-corrected modeling

1022 cm-2 deg2 corresponds to ~740 Msun for d=150 pc

1/10

N(HIthin) and the regions studied (HI4PI data)

Study of the Interstellar Medium and Cosmic-rays in Local HI Clouds

Oct. 15-20, 2017, 7th Fermi Symposium, Garmisch-Partenkirchen, GermanyT. Mizuno (Hiroshima Univ.) on behalf of the Fermi-LAT collaboration

• Aims. We aim to study the interstellar medium (ISM) and cosmic-rays (CRs) in local HI clouds in the 3rd

Galactic quadrant

• Methods. We evaluated the total gas column density N(Htot) by investigating the correlations among 21 cm survey data (HI4PI), Planck dust models (optical depth at 353 GHz 353 and radiance R), and Fermi-LAT -ray data

• Results & Prospects. We found N(Htot,)/353 and N(Htot,)/R depend on dust temperature Td in the North region, and N(Htot,)/353 is not constant over 353 in the South region. We will examine the systematic uncertainties and discuss ISM and CRs properties.

(1020 cm-2)

2/10

Objective of the Study

8 years, P8R2_CLEAN_V6, 0.1-25.6 GeV

Several areas (an intermediate velocity cloud, the Orion-Eridanus superbubble, and a peculiar WHI-dust relation) are masked

North South

• An accurate estimate of N(Htot) is crucial to understand the ISM and CRs• Considerable amount of ISM gas is not properly traced by HI and CO line surveys [1].

The distribution of this “dark gas” can be estimated by dust data, but the procedure has not been established yet.

• We studied mid-latitude region of the 3rd quadrant using Fermi-LAT -ray data (as a robust tracer of N(Htot)), HI4PI data [2], and Planck dust models [3], in order to examine the following ISM properties and implications on CRs– (a) Td dependence of dust-emission to gas ratio [4]– (b) Non-linearity of dust-emission to gas ratio [5][6]

3/10

WHI-Dust Relations

• North: Td dependence is seen and is larger in the WHI-353 relation

• South: Td dependence is weak, but a non-linear relation is observed

• We used linear relations which follow trends in high Td & low WHI area to construct initial N(Htot) template maps from 353 and R

: . . .

: . . .

4/10

Results and Prospects

References: [1] Grenier+05, Science 307, 1292 [2] HI4PI Collaboration 2016, A&A 594, 116 [3] Planck Collaboration XI 2014, A&A 571, 11 [4] Mizuno+16, ApJ 833, 278 [5] Roy+13, ApJ 763, 55[6] Planck Collaboration XXVIII 2015, A&A 582, 31 [7] Abdo+09, ApJ 703, 1249

• We fit -ray data with a linear combination of gas template maps and other components (isotropic, inverse Compton, sources etc.)

• Under the assumption of a uniform CR density, emissivity should not depend on Td (North) and should be constant over 353 (South), if N(Htot) ∝ 353 or R

• North: We prepared Td-sorted maps and found a positive Td dependence for 353, likely due to an overestimate of N(Htot)/353 in low Td area (similar trend seen in [4])

• South: We prepared 353-sorted maps and found negative 353 dependence, likely due to an overestimate of N(Htot)/353 in high 353 area (similar trend seen in [5][6])

• Future plan: examine the systematic uncertainties and discuss ISM and CR properties

Scale factors to the model for the local interstellar spectrum [7]

North

South

Documents

Study of the Interstellar Medium Gas and Cosmic …mizuno/GLAST/...T. Mizuno et al. 1/16 Study of the Interstellar Medium Gas and Cosmic Rays using GeV Rays Oct. 24th, 2017@MW workshop