ESO Band 5 Workshop Carbon in the ISM of z=4 dusty starburst galaxies Matt Bothwell (Cambridge) + the SPT SMG team
Observing gas at high redshift 1. Traditional approach use a visible tracer molecule ( 12 CO) 2. Alternative use dust mass and gas/dust ratio 3. Potential alternate tracer atomic carbon, [CI]
Problems with 12 CO, (I) H2
12 Problems with CO, (I) H2 CO
Problems with 12 CO, (I)
log αco 12 Problems with CO, (I) Metallicity
12 Problems with CO, (I) is a very inefficient tracer of molecular gas at low metallicity log αco 12CO Metallicity
Problems with 12 CO, (II) CO Even at fixed metallicity, can vary based on the structure of the ISM
Problems with 12 CO, (II) CO Even at fixed metallicity, can vary based on the structure of the ISM The classic Milky Way value ( CO = 4.6 M (K km s -1 pc 2 ) -1 ) is derived assuming molecular clouds are separate and virialised
Problems with 12 CO, (II) CO Even at fixed metallicity, can vary based on the structure of the ISM The classic Milky Way value ( CO = 4.6 M (K km s -1 pc 2 ) -1 ) is derived assuming molecular clouds are separate and virialised In the early 1990s, it was realised that using this value for ULIRGs produced M(gas) > M(dyn) (Scoville et al. 1991) Led to a new ULIRG value for α, of ~0.8 alongside the normal value of ~4.5
Problems with 12 CO, (III) 12 CO is also dissociated by cosmic rays. In high SFR environments, CO may be a poor gas tracer (Bisbas+15)
line ) between at redshift z and at z = 0. The ratios are obtained Figure 5. Predicted brightness ratios of line emission (RB line when using LVG modelling which assumes a molecular density of 103 cm 3 and virialised conditions. Left: RB z=0 = T z=0 = 20 K. Right: Rline when assuming T z=0 = T z=0 = 50 K. assuming Tkin B d kin d 12 Problems with CO, (IV) CO J=(1-0) Zhang+16 12CO also becomes difficult to see against the CMB at high redshifts TCMB goes like (1+z) at high-z, observations in Raleigh-Jeans domain lack contrast against CMB
Observing gas with dust emission Eales+12 Also possible to measure a total dust mass (via SED fitting), and combine with a gas-to-dust ratio, to measure Mgas Advantage: observations are cheap compared to mm lines. Disadvantage: no kinematic information, dust-to-gas ratio not well understood
Tracing gas with atomic carbon
Tracing gas with atomic carbon Atomic carbon, [CI], is closely associated with low-j CO emission across a wide range of environments (e.g., Papadopoulos+04; Walter+11; Israel+15)
Tracing gas with atomic carbon Atomic carbon, [CI], is closely associated with low-j CO emission across a wide range of environments (e.g., Papadopoulos+04; Walter+11; Israel+15) Carbon is simple compared to CO, so physical parameters (excitation temperature, carbon mass) can be easily calculated
Tracing gas with atomic carbon SPT0243 49 z=5.698 SPT0346 52 z=5.656 SPT0459 58 z=4.856 SPT0459 59 z=4.798 SPT2146 55 z=4.567 SPT0441 47 z=4.477 SPT2103 60 z=4.435 SPT0345 47 z=4.296 SPT0113 46 z=4.232 SPT0418 47 z=4.224 SPT0125 50 z=3.955 SPT2147 50 z=3.761 SPT2132 58 z=3.615 SPT0300 46 z=3.594 SPT0532 50 z=3.399 SPT0529 54 z=3.369 SPT0103 45 z=3.090 SPT2134 50 z=2.780 SPT0125 47 z=2.515 SPT0512 59 z=2.234 SPT0551 50 z=2.123 SPT0550 53 z=2.096 SPT0452 50 z=2.010 SPT0128 51 SPT0319 47 SPT0457 49 13 CO (3 2) 12 CO (3 2) 13 CO (4 3) 12 CO (4 3) [C I] 3 P 1 3 P 0 200 300 400 500 600 700 800 Rest Frequency (GHz) 13 CO (5 4) o H 2 O 1 10 1 01 12 CO (5 4) 13 CO (6 5) 12 CO (6 5) o H 2 O + 2 02 1 11 p H 2 O 2 11 2 02 Vieira, MB+2013, Weiß + 2013
[CI](1-0) Tracing gas with atomic carbon Flux (mjy) Flux (mjy) Flux (mjy) Flux (mjy) Redshift [CI](1-0) 4.198 3.56 4.221 3.59 4.244 3.61 4.268 3.63 8 8 [CI](1-0) SPT0113-46 SPT0300-46 6 CO(2-1) 6 4 4 2 0-2 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 4.190 3.56 4.213 3.59 4.236 3.61 4.260 3.63 15 8 [CI](1-0) CO(2-1) SPT0300-46 SPT0418-47 6 10 4 25 0 0-2 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 4.190 4.761 4.213 4.786 4.236 4.812 4.260 4.838 15 8 [CI](1-0) [CI](1-0) CO(2-1) SPT0459-59 CO(2-1) SPT0418-47 6 10 4 52 0 0-2 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 4.761 3.37 4.786 3.39 4.812 3.41 4.838 3.43 8 10 6 4 5 2 0-2 [CI](1-0) CO(2-1) SPT0459-59 SPT0532-50 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 3.37 3.39 3.41 3.43 Flux (mjy) Flux (mjy) Flux (mjy) Flux (mjy) 3.926 3.948 3.970 3.992 6 4 2 0-2 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 4.260 4.441 4.284 4.465 4.308 4.489 4.331 4.514 8 6 6 4 4 2 2 0-2 -2 [CI](1-0) [CI](1-0) CO(2-1) CO(2-1) SPT0345-47 SPT0441-46 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 4.441 3.34 4.465 3.36 4.489 3.38 4.514 3.40 15 6 [CI](1-0) CO(2-1) SPT0441-46 SPT0529-54 10 4 25 0 0-2 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 4.399 3.34 4.424 3.36 4.448 3.38 4.472 3.40 15 [CI](1-0) 6 SPT2103-60 CO(2-1) SPT0529-54 104 2 5 0-2 0 Redshift [CI](1-0) SPT0125-50 -2000-1000 0 1000 2000 Velocity (km/s) Redshift [CI](1-0) 4.399 4.424 4.448 4.472 Some of our [CI] spectra; Grey = [CI] Blue = CO(2-1), normalised 13 sources in total (12 with good detections)
Deriving gas mass from [CI] M(H 2 ) = 1375.8 D 2 L (1 + z) 1 X[CI] 10 5 1 A10 10 7 s 1 1 Q 1 10 S [CI] v, Papadopoulos & Greve 04
Deriving gas mass from [CI] [CI] abundance Einstein A coefficient M(H 2 ) = 1375.8 D 2 L (1 + z) 1 X[CI] 10 5 1 A10 10 7 s 1 1 Q 1 10 S [CI] v, Excitation factor (=0.5) Papadopoulos & Greve 04
Deriving gas mass from [CI] Three approaches (I) M(H 2 )= CO L 0 CO
Deriving gas mass from [CI] Three approaches (I) (II) M(H 2 )= CO L 0 CO M(H 2 )=X [CI] I [CI] ( C)
Deriving gas mass from [CI] Three approaches (I) (II) M(H 2 )= CO L 0 CO M(H 2 )=X [CI] I [CI] ( C) (III) M(H 2 )= GDR M(dust)
Deriving gas mass from [CI] Three approaches (I) (II) M(H 2 )= CO L 0 CO M(H 2 )=X [CI] I [CI] ( C) (III) M(H 2 )= GDR M(dust) All methods use an observable, multiplied by an unknown `conversion factor So why the claim that CI is a superior diagnostic?
Deriving gas mass from [CI] CO and GDR are ~quadratically dependent on metallicity X [CI] is ~linearly dependent on metallicity So given an unknown metallicity, derived M(H2) via [CI] involves smaller uncertainty and [CI] isn t destroyed by cosmic rays; and [CI] can be seen against the CMB at high-z; and [CI](1-0) is easily accessible at high-z and [CI](1-0) provides kinematic information
Gas masses: [CI] vs CO MNRAS 457, 4406 4420 (2016) Advance Access publication 2016 February 9 doi:10.1093/mnras/stw275 A survey of the cold molecular gas in gravitationally lensed star-forming galaxies at z > 2 M. Aravena, 1 J. S. Spilker, 2 M. Bethermin, 3 M. Bothwell, 4 S. C. Chapman, 5 C. de Breuck, 3 R. M. Furstenau, 6 J. Gónzalez-López, 7 T. R. Greve, 8 K. Litke, 2 J. Ma, 9 M. Malkan, 10 D. P. Marrone, 2 E. J. Murphy, 11 A. Stark, 12 M. Strandet, 13 J. D. Vieira, 6 A. Weiss, 13 N. Welikala, 14 G. F. Wong 15,16 and J. D. Collier 15,16 1 Aravena, MB et al. (2016) low-j CO survey of SPT DSFGs Allows CO-based gas masses to be calculated for same sample
Gas masses: [CI] vs CO Aravena, MB et al. (2016) low-j CO survey of SPT DSFGs Allows CO-based gas masses to be calculated for same sample
Gas masses: [CI] vs CO M(H2), CO ( CO = 0.8) 10 11 10 10 SPT DSFGs, low-j CO SPT DSFGs, high-j CO Literature DSFGs, low-j CO Literature DSFGs, high-j CO Literature AGN 10 10 10 11 M(H2), [CI]
Gas masses: [CI] vs CO H2 masses from [CI] are systematically larger than from CO Have high-z gas masses (derived using CO) been systematically underestimated? Culprit incorrect CO/H2 conversion factor M(H2), CO ( CO = 0.8) 10 11 10 10 SPT DSFGs, low-j CO SPT DSFGs, high-j CO Literature DSFGs, low-j CO Literature DSFGs, high-j CO Literature AGN 10 10 10 11 M(H2), [CI]
Gas masses: [CI] vs CO M(H 2) (CI) / L CO(1-0) [M O (K km s -1 pc 2 ) -1 ] 6 5 4 3 2 1 0 SPT DSFGs, low J CO SPT DSFGs, high J CO Literature DSFGs, low J CO Literature DSFGs, high J CO Literature AGN MW Value ULIRG value 6 5 4 3 2 1 0 Implied α CO -1 10 12 10 13 L(FIR) [ L O ] 1
Gas masses: [CI] vs CO M(H 2) (CI) / L CO(1-0) [M O (K km s -1 pc 2 ) -1 ] 6 5 4 3 1 0-1 SPT DSFGs, low J CO SPT DSFGs, high J CO Literature DSFGs, low J CO Literature DSFGs, high J CO Literature AGN MW Value 2Sample average α CO = 2.0-2.5 ULIRG value 10 12 10 13 L(FIR) [ L O ] 6 5 4 3 2 1 0 1 Implied α CO
What about dust-based M(H 2 )? Aravena et al. (2016) also calculate dust masses for our sample of DSFGs Combined with a dust-to-gas ratio, this gives us another measure of M(H2). We can compare our three independently-derived gas masses
Three measures of gas mass CO (+ alpha_co, assume alpha_co ~ 1) Dust (+ GDR, assume GDR ~ 100) [CI] (+ [CI] abundance, assume X CI ~ 3e-5) These disagree!
Three measures of gas mass ~7e10 XCI ~ 3e-5 M(H2) [M_sun] ~3e10 GDR~100 alpha_co~1
Three measures of gas mass ~7e10 XCI ~ 3e-5 M(H2) [M_sun] ~3e10 GDR~100 alpha_co~1 Solution (a) GDR~240, alpha_co~2.5, X CI ~3e-5 Solution (b) GDR~100, alpha_co~1, X CI ~7e-5
Three measures of gas mass Solution (a) GDR~240, alpha_co~2.5, X CI ~3e-5 Solution (b) GDR~100, alpha_co~1, X CI ~7e-5 Either alpha_co and DGR are 2-3 higher than normal, or X CI is 2-3 higher than normal NB Strandet (in prep) modelling points towards solution (a), implying we have been underestimating high-z gas masses
Gas mass from [CI] Future need more [CI] observations!!
Gas mass from [CI] [CI] 610µm in Band 5 = 1.3 < z < 2.1 [CI] 371µm in Band 5 = 3.8 < z < 5.1
[CI] with SEPIA band 5 Range of ratios [CI] 371µm CO(7-6) Béthermin et al. in prep.