- Christian Baker analyzed Herschel PACS spectroscopy data of dwarf galaxy NGC 5195 to determine characteristics of its cold gas and dust using a PDR model.
- Key emission lines were detected including [C II], [N II], [O I]63, and [O I]145. [O III] was below the detection threshold.
- Ratios of emission lines to total infrared flux suggested a heating efficiency about an order of magnitude lower than NGC 5195's companion galaxy M51.
- Analysis of line ratios indicated gas densities of log(n/cm-3) between 2.0-3.0 and radiation intensities of logG0 between 2.5-3.3.
NGC5195 PDR Modelling cold gas and dust in dwarf galaxy
1. Draft version April 17, 2015
Preprint typeset using LATEX style emulateapj v. 01/23/15
NGC5195 PDR MODELLING OF COLD GAS AND DUST USING HERSCHEL PACS SPECTROSCOPY
Christian Baker and Supervisor: Christine Wilson
McMaster University
Draft version April 17, 2015
ABSTRACT
Data from the Herschel Space Observatory Photodetector Array Camera and Spectrometer (PACS)
spectroscopy as part of the Very Nearby Galaxies Survey (VNGS) has revealed ISM gas characteristics
for dwarf galaxy NGC 5195, the companion to M51, using spectral lines [C II]158 [N II]122, [O I]63, [O
I]145, and [O III]88. With these observed flux values we compare to a predicted flux from a photon
dominated region (PDR) model to determine the characteristics of the cold gas and then compare
to other recently studied galaxies. We find an averaged [C II]/FTIR value of 4.4×10−4
which falls
an order of magnitude under comparable galaxies. We find a logG0 of 2.5-3.3 and a log(n/cm−3
) of
2.0-3.0.
1. INTRODUCTION
We study the S0 galaxy NGC 5195, the companion
to M51, for its cooling properties. NGC 5195 is lo-
cated at a distance of 7.7±1.0 Mpc away (Tonry et al.
2001). By observing the cooling properties we aim to use
a PDR model to determine gas characteristics of NGC
5195 which will allow us to compare with other galaxies.
Of particular interest is M51 as NGC 5195 is a compan-
ion galaxy and has interacted with M51. This follows the
general goals of the (VGNS) which are to probe the prop-
erties of gas and dust in the interstellar medium (ISM)
of 13 galaxies. NGC 5195 is not one of these galaxies but
data was taken for it at the same time as M51. Thanks
spectroscopy from the PACS instrument offers us fine
structure lines of [C II]158 [N II]122, [O I]63, [O I]145, and
[O III]88 at resolutions better than 12” (see Table 1).
In order to investigate the cooling properties of the in-
terstellar medium (ISM), fine structure lines such as [C
II]158 [N II]122, [O I]63, [O I]145, and [O III]88 can be used.
These lines show gas cooling by de-excitation via photon
emission. In particular the [C II] line traces both neutral
and ionized gas, the [O I] lines trace neutral gas, and
the [N II] and [O III] lines trace ionized gas. Due to the
various emission lines tracing different parts of the gas,
we can use the [C II] emission, a tracer of photon domi-
nated regions (PDRs), even though it also traces ionized
gas (with an ionization potential of only 11.26 eV). In
comparison the [N II] and [O III] lines have ionization
potentials of 14.5 eV and 35 eV and thus require a hard
radiation field.
We probe these lines using the Herschel PACS
(Poglitsch et al. 2010) in order to investigate the ISM in
NGC 5195 using a PDR model, as well as comparing it
to recent studies on M51,and Centaurus A.(Parkin et al.
2013; Parkin et al. 2014). We use the spectra obtained
to investigate the gas component of NGC 5195, and the
PDR model of Kaufman et al. (1999, 2006) to estimate
the properties of the ISM.
A PDR model predicts the physical characteristics
through the hydrogen nucleus density, n, and the
strength of the far-ultraviolet (FUV) radiation field in
units of the Habing Field, G0 = 1.6×10−3
erg cm−2
s−1
bakerca2@mcmaster.ca
(Habing 1968). Kaufman’s model (Kaufman 1999, 2006)
gives a method to use the line ratios and total in-
frared luminosity to determine n and G0. Our data
uses the readily available analysis tool the PDR Toolbox
(PDRT) (Pound & Wolfire 2008); which may be found
at http://dustem.astro.umd.edu/pdrt. PDR models as-
sume a plane-parallel, semi-infinite slab geometry and
include a complex chemical network, thermal balance,
and radiative transfer.
Our comparisons to Cen A and M51 will include find-
ing an average [C II]/FT IR value to compare to 4 × 10−3
of M51 and 8.4 × 10−3
of Cen A (Parkin et al. 2013;
Parkin et al. 2014). We also will compare heating effi-
ciency which is determined from ([C II]+[O I]63)/FT IR.
This same value when used in conjunction with [C II]/[O
I]63 will give values to compare to n and G0.
Organization of this paper is as follows: Section 2 de-
scribes the data and the processing of said data, Section
3 describes the gas characteristics, and Section 4 com-
pares these observations to theoretical models as well as
other recently studied galaxies.
2. OBSERVATIONS
2.1. PACS spectroscopy
We obtained the PACS spectroscopy of NGC 5195 from
the Guaranteed Time Project of the VGNS (PI; C. D.
Wilson). Maps were obtained for [C II]158 [N II]122, [O
I]63, [O I]145, and [O III]88. Properties of the observations
can be seen in Table 1. Images were handed to us already
reduced and ready for analysis (Parkin T. J.). The [C II]
and [N II] lines were mapped in a 3 × 3 set of overlap-
ping footprints that cover a square area of 47” on a side.
The [O I]63 line was mapped in a 5 × 5 grid. The [O
I]145 and [O III]88 lines were mapped in a 2 × 2 grid. T.
J. Parkin (private communication) describes the data re-
duction as follows: ”The raw data was processed by T. J.
Parkin up to Level 2 using the Herschel Interactive Pro-
cessing Environment (HIPE; Ott 2010) version 9.0.2649
with calibration files FM,41. The standard pipeline for
unchopped gratings scans was used. The Level 2 PACS
spectral cubes were then passed to the line fitting and
map making program PACSman 3.52 (Lebouteiller et al.
2012), where the spectral lines in each raster were fit with
a Gaussian line profile and a second order baseline. The
2. 2
[C II]
1E-08 2E-08 3E-08 4E-08 5E-08 6E-08 7E-08 8E-08
[N II]
1E-09 2E-09 3E-09 4E-09 5E-09 6E-09 7E-09 8E-09
[O I]63
1E-08 2E-08 3E-08 4E-08 5E-08 6E-08 7E-08 1E-09 2E-09 3E-09 4E-09 5E-09 6E-09 7E-09
[O III]
[O I]145
1E-09 2E-09 3E-09 4E-09 5E-09 6E-09
Fig. 1.— Maps of NGC 5195 displaying Herschel PACS spectroscopic observations of fine structure lines at native resolution and pixel
scale. North is up and east is to the right. Units are W m−2.
3. 3
resulting line fits were then integrated to obtain fluxes,
then maps were produced by projecting fluxes from each
raster on an oversampled common grid with final pixel
scale of 3.133”.” The reduced images can be seen in Fig-
ure 1.
2.2. Additional Data
A total infrared flux map was also obtained from Tara
Parkin which was made using reprocessed photometry
(Bendo et al. 2012) at 24 µm from the Multiband Imag-
ing Photometer for Spitzer (Rieke et al. 2004) instrument
on the Spitzer Space Telescope (Werner et al. 2004) in
combination with published PACS photometry data at
70 and 160 µm (Mentuch Cooper et al. 2012). Parkin
was then able to estimate the the total far-infrared inten-
sity using the empirical equation (Galametz et al. 2013):
IT IR = (2.133 ± 0.095)v24I24 + (0.681 ± 0.028)v70I70
+(1.125 ± 0.010)v160I160
(1)
The map created from this process can be seen in Figure
2.
2.3. Data treatment for analysis
Before any data analysis on the maps we convolved the
maps to share a full width half maximum (FWHM) of
the [C II] map ( 11.5”) using a Gaussian function using
the STARLINK software’s gausmooth function. Once
convolved the images were also aligned and then tested
with large apertures covering the entire map to see if flux
was lost in the convolving process. We lose under 1% of
the total flux and consider flux to be conserved under the
transformations. Calibration uncertainty for PACS data
is at ±30% which is primarily small offsets in pointing
and drifting of the detector response (PACS OM). Un-
certainties listed in tables are from the uncertainty maps
in the raw data however to give a more precise look at
the data as all of the maps are equally affected by the
calibration uncertainty. [O III] and [O I]145 were below
a mean signal-to-noise ratio (SNR) of 9 across the a 30”
aperture and have been discounted from most analysis.
[O I]145 has a peak SNR of 27 but falls off rapidly as you
can see in Table 2, so we use the peak flux in Section 3.2
but not the mean value.
3. PHYSICAL CHARACTERISTICS OF THE GAS
3.1. Line Emission Morphology
As can be seen in Figure 1 the distributions for most
lines is similar, with peaks in the centre. As previously
discussed we reject the [O III]. The low signal-to-noise
is evident in the uneven distribution of the map where
we would expect a peaked distribution like the other de-
tected lines. Using starlink’s beamfit command we find
it fails to determine a beam width for [O I]145 and [O
III]88. It does provide fits for the full width at half max-
imum (FWHM) of 26”, 21” and 15” for [C II], [N II]122
and [O I]63 which agrees with what we can see visually.
As seen in Table 3 our peak flux is 2-3 times greater
than an average over a 30” aperture. Overall morphology
looks elliptical as there is no visible spiral arm structure.
In Table 4 we observe our well detected lines in a ratio
with FT IR. Here we have used the mean and peak values
again. [C II] contributes the most of the lines studied,
with [O I]63 coming in close behind. Emission from [C II]
and [O I]63 account for up to 0.004% of total emission.
This is a factor of 10 lower than what was observed in
NGC 5195’s companion, M51 (Parkin et al. 2013).
3.2. Heating Efficiency
We can use ([C II]+[O I]63)/FT IR as a proxy to the
heating efficiency (Tielens & Hollenbach 1985). This
value is a measure of the amount of interstellar FUV
radiation that gets converted via the photoelectric effect
into gas heating divided by the fraction of its energythat
is deposited into the dust grains. We will be using it with
the ratio of [C II]/[O I]63 to use the heating efficiency to
predict characteristics of the gas. A map of ([C II]+[O
I]63)/FT IR can be seen in Figure 3. With the ratios of
([C II]+[O I]63)/FT IR and [C II]/[O I]63 we can start
modelling our PDR’s as seen in Section 4.1. There is a
deficit in the center of Figure 3 which suggests reduced
heating efficiency in this region. This is consistent with
what was found in M51 by Parkin et al (2013).
The [O I] lines can give an estimate of the temperature
as displayed in Figure 4 of Liseau et al. (2006). Given
that our [O I]145 is poorly detected we attempt to get a
rough estimate of the temperature if the gas is optically
thick or optically thin by using the peak values where the
signal-to-noise is somewhat decent. We find [O I]63/[O
I]145 to be 10±1 and as the lines intersect close to there
we can estimate the temperature as being T ≥200 K and
n ≥ 103
cm−3
.
With the ratios of ([C II]+[O I]63)/FT IR and [C II]/[O
I]63 we can start modelling our PDR’s as seen in Section
4.1.
3.3. Ionized Gas Fraction
The fraction of emission originating in HII regions is of
particular interest as opposed to emission from ionized
gas. We estimated the fraction of [C II] emission origi-
nates from ionized gas versus neutral gas. The [N II]205
line is commonly used to determine the [N II]122/[N II]205
which can be used to determine the ionized gas density.
We lack the [N II]205 line which we need. Often we es-
timate the [N II] ratio when this happens, such as using
the Galactic value when appropriate as Malhotra et al.
(2001) did. Parkin found however that this disagreed
with other methods (Parkin et al. 2013; Kramer et al.
2005).
We instead estimate the [N II]122/[N II]205 at various
ionized gas densities constrained using the [S III]18.71/[S
III]33.48 which was provided by Tara Parkin (private com-
munication) as n ≤ 102
cm−3
. Parkin obtained this value
using Spitzer low resolution IRS spectrum with the line
fitting program PAHfit (Smith et al. 2007) to obtain line
fluxes for the various lines giving a [S III]18.71/[S III]33.48
ratio of 0.35±0.02 which compared to a theoretical curve
(Snijders et al. 2007) indicates that n ≤ 102
cm−3
. By
taking 4 different estimates all within the range speci-
fied by the silicon ratio we can account for how different
approaches may lead to different estimates.
3.4. Adjustments and Corrections
Kaufman’s PDR model (Kaufman et al. 1999) requires
two adjustments to our observed line fluxes. We must re-
move the fraction of [C II] flux that comes from ionized
4. 4
TABLE 1
Properties of our Herschel Observations
Line Wavelength (µm) FWHMa (”) Integration Time (s) OBSID
[O I] 63.184 9.3 5578 1342223763
145.525 11 1968 1342223766
[O III] 88.356 9.3 1216 1342223765
[N II] 121.898 10 3974 1342223767
[C II] 157.741 11.5 2288 1342223764
aValues are from the PACS Observer’s Manual and the SPIRE Observer’s Manual
All data taken on 2011-07-07.
5E-05 0.0001 0.00015 0.0002 0.00025 0.0003
Fig. 2.— The calculated total infrared intensity using Equation 1 at a resolution of 12”. Units are 10−4 W m−2. This map was aligned
with the previous maps but not convolved to the same resolution as 11.5” is close enough to 12” that it will not cause problems for the
analysis.
TABLE 2
Peak and mean signal-to-noise ratio of all observed lines
Line Mean Peak
[C II] (158 µm) 89.2 178.5
[N II] (122 µm) 11.8 26.6
[O I] (63 µm) 25.1 72.6
[O I] (145 µm) 8.1 26.7
[O III] (188 µm) 5.0 8.7
gas. We also must take into account that Kaufman’s
model is a plane-parallel slab with incident radiation
from only one side, the side we observe emission from
TABLE 3
Observed Flux in a 30” aperture and peak flux
Line Mean (10−9 W m−2) Peak (10−9 W m−2
[C II] (158 µm) 36.6 ±0.2 83.2 ±0.5
[N II] (122 µm) 3.1±0.1 7.2 ±0.3
[O I] (63 µm) 21.5±0.4 61 ±1
in the far-infrared cooling lines. As we cannot guarantee
the cloud’s orientation faces us we use Kaufman’s advice
that the velocity dispersion for many clouds combined
with an assumption that the [O I]63 will become opti-
cally thick much faster than either [C II] or the total
5. 5
0.001 0.002 0.003 0.004 0.005 0.006
Fig. 3.— The heating efficiency map of ([C II]+[O I]63)/FT IR.
TABLE 4
Line to total infrared flux ratio in NGC5195
Line Mean 10−4 Line/FT IR Peak 10−4 Line/FT IR
[C II] (158 µm) 4.4±0.2 2.44 ±0.01
[N II] (122 µm) 0.37±0.01 0.211±0.009
[O I] (63 µm) 2.6±0.5 1.79±0.03
infrared flux means that we only see half the total [O
I]63 flux. Thus we double this value for use in our PDR
model.
Table 5 indicates 4 guessed ratios starting with the the-
oretical lower limit for the [N II]122/[N II]205 ratio(Wright
et al. 1991; Bennett et al. 1994) and going to the theoret-
ical ratio at n = 100 cm−3
. By subtracting the predicted
ionized portion from the total flux we can divide the flux
by the total to obtain the fraction of neutral gas and thus
apply this fraction to our [C II] flux. As a check that our
guessed ratios are reasonably we can quickly compare to
observed ratios for the Milky Way, which fall between 1.0
and 1.6 (Wright et al. 1991).
The correction uses predicted [C II]158/[N II]205 val-
ues as a function of electron density in her study of
M51 using Solar Gas abundances of C/H=1.4 × 10−4
and N/H=7.9 × 10−5
(Parkin et al. 2013). It varies
from 3.2 ±0.3 at an assumed [N II]122/[N II]205=0.7 to
3.1±0.3N II]122/[N II]205=1 and 2 and finally 3.0±0.3 at
N II]122/[N II]205=3.
Kaufman’s paper (Kaufman et al. 1999) recommends
that the total infrared flux be reduced by a factor of two
to account for the optically thin infrared continuum flux
coming from both the front and back sides of the gas
cloud. We have applied this correction to our infrared
flux to ensure our PDR model is accurate.
4. RESULTS AND COMPARISONS
4.1. PDR modelling
We use PDRT for our modelling (Pound & Wolfire
2008; Kaufman et al. 2006). The model assumes
the PDR is a plane-parallel semi-infinite slab and is
parametrized by two free variables, the hydrogen gas
density, n, and the strength of the impinging FUV
radiation field normalized by the Habing field 1.6 ×
10−3
erg cm−2
s−1
(Habing 1968). This model includes
thermal balance, chemical network, and radiative trans-
fer, and produces grids of predicted structure in terms of
two line ratios, [C II]/[O I]63 and ([C II]+[O I]63)/FT IR
along axes of G0 and n. Figure 4 shows the line ra-
tio maps for [C II]/[O I]63 on the left and([C II]+[O
I]63)/FT IR on the right. First we look at Figure 5 left
is the uncorrected mean values. We are now using the
30% calibration uncertainty as it will be important in
determining the potential values. Figure 5 right shows
the corrected peak values being used. Figure 6 repeats
this but for mean corrected values.
We can ignore the low–G0 high–n solutions by consid-
ering the number of clouds emitting within our beam.
If you compare the model predicted [C II] emission for
low–G0 and high–n solutions to our observed [C II] emis-
sion we find that we would require multiple PDR regions
of order of magnitude 103
, which is a very high number
of clouds along our line of sight (Kramer et al. 2005).
Therefore, we ignore solutions in the bottom right of our
plots. We also switch to the mean value for the final
plots as our uncorrected mean at least had a crossover
region of allowed G0 and n values.
Prior to applying the corrections, our plots were sug-
gesting somewhat low–G0 and low–n solutions. This was
fixed once our corrections were made. Surprisingly, the
varying of neutral [C II] compared to ionized [C II] as
based on the estimated ionization density did not make
6. 6
TABLE 5
Estimated N II ratios predict neutral C II emission fraction
Assumed [N II]122/[N II]205 Predicted [C II]158/[N II]205 Assumed ionized Neutral C II
Gas Density (fraction of total)
0.7 8.1 ±0.3 1 0.63 ±0.03
1 11.6 ±0.5 9.2 0.75 ±0.04
2 23 ±1 45 0.88 ±0.05
3 34.8 ±1.5 100 0.92 ±0.05
Fig. 4.— [C II]/[O I]63 is on the left and([C II]+[O I]63)/FT IR is on the right. These are colour maps of constant value for the line
ratios. As can be seen in Figures 5 and 6 the lines trace along paths of the same colour.
a large difference after it left the theoretical lower limit.
With this and the removal of the lower right solution, we
can determine a logG0 of 2.5-3.3 and a log(n/cm−3
) of
2.0-3.0.
If we use Figure 1 from Kaufman et al. (1999) we can
determine the surface temperature of the gas is between
200K and 300K. Prior to corrections we would have esti-
mated the temperature as anywhere from 200K to 1000K.
This agrees with our oxygen ratio temperature that we
determined earlier, despite using a line that was poorly
detected.
4.2. M51 and Cen A
Unlike M51 or Cen A, we did not model arms or disk
regions separately as the object is not extended. Table 6
shows the logG0 and log(n/cm−3
) values for the different
regions of M51, Cen A, and the single region of NGC
5195.
NGC 5195 has similar values to previously studied
galaxies and resembles the average Cen A values as well
as M51’s center values. Both Cen A and NGC 5195 have
TABLE 6
Properties of the gas derived from the PDR model
Object log(n/c−3) logG0 T(K)
NGC 5195 2.0-3.0 2.5-3.3 200-300
M51 nucleus 3.5-4.25 3.25-4.0 240-475
M51 center 2.5-4.0 2.5-3.5 170-680
M51 arms 2.0-3.75 1.75-3.0 100-760
M51 interarm 2.25-3.75 1.5-3.0 80-550
Cen A 2.75-3.75 1.75-2.75 110-260
References: T. J. Parkin (2013; 2014)
smaller ranges on temperature as compared to M51. Per-
haps spiral arm structure contributes to the increase in
surface temperature by allowing new stars to form more
easily which in turn would heat the gas.
5. CONCLUSIONS
Using Herschel PACS observations of the important
fine-structure lines [C II]158 [N II]122, [O I]63, [O I]145,
and [O III]88, we measure several diagnostic ratios to
compare NGC 5195 with M51 and Cen A. With the the-
oretical lower limit for [N II]122/[N II]205 of 0.7 producing
7. 7
Fig. 5.— Uncorrected mean on the left, uncorrected peak on the right
a PDR model with no lower limits, we can assume this
is not likely the physical characteristics of the galaxy.
Therefore, we know that between 8% and 25% of the
observed [C II] emission originates in ionized gas.
We determined a logG0 of 2.5-3.3 and a log(n/cm−3
)
of 2.0-3.0. When compared to PDR models this gives a
temperature range of between 200K and 300K. This is
similar to Cen A and certain regions of M51, however
M51 exhibits much higher max temperatures, perhaps
due to its spiral arms. This temperature range and hy-
drogen density agree with the potentially unreliable re-
sults from using the [O I]63/[O I]145 ratio.
I would like to extend my thanks to my supervisor
Christine Wilson for her extensive help on this project, as
well as Maximilien Schirm for his help in understanding
PDR models.
APPENDIX
REFERENCES
Bendo, G. J., Galliano, F., & Madden, S. C. 2012, MNRAS, 423, 197
Galametz, M., Kennicutt, R. C., Calzetti, D., et al. 2013, MNRAS, 431, 1956
Habing, H. J. 1968, BAN, 19, 421
Kaufman, M. J., Wolfire, M. G., & Hollenbach, D. J. 2006, ApJ, 644, 283
Kaufman, M. J., Wolfire, M. G., Hollenbach, D. J., & Luhman, M. L. 1999, ApJ, 527, 795
Kramer, C., Mookerjea, B., Bayet, E., et al. 2005, A&A, 441, 961
Lebouteiller, V., Cormier, D., Madden, S. C., et al. 2012, A&A, 548, A91
Liseau, R., Justtanont, K., & Tielens, A. G. G. M. 2006, A&A, 446, 561
Malhotra, S., Kaufman, M. J., Hollenbach, D., et al. 2001, ApJ, 561, 766
8. 8
[N II] ratio=0.7 [N II] ratio=1.0
[N II] ratio=2.0 [N II] ratio=3.0
Fig. 6.— Varying nitrogen line ratio to determine effects on n and G0. This changes the amount of [C II] emission the originates from
ionized gas versus neutral gas.
9. 9
Mentuch Cooper, E., Wilson, C. D., Foyle, K., et al. 2012, ApJ, 755, 165
Ott, S. 2010, in ASP Conf. Ser. 434, Astronomical Data Analysis Software and Systems XIX, ed. Y. Mizumoto,
K.-I. Morita, & M. Ohishi (San Francisco, CA: ASP), 139
Parkin, T. J., Wilson, C. D., Schirm, M. R. P., et al. 2013, ApJ, 776, 65
Parkin, T. J., Wilson, C. D., Schirm, M. R. P., et al. 2014, ApJ, 787, 16
Poglitsch, A., Waelkens, C., Geis, N., et al. 2010, A& A, 518, L2
Pound, M. W., & Wolfire, M. G. 2008, in ASP Conf. Ser. 394, Astronomical Data Analysis Software and Systems
XVII, ed. R. W. Argyle, P. S. Bunclark, & J. R. Lewis (San Francisco, CA: ASP), 654
Rieke, G. H., Young, E. T., Engelbracht, C. W., et al. 2004, ApJS, 154, 2
Snijders, L., Kewley, L. J., & van der Werf, P. P. 2007, ApJ, 669, 269
Tielens, A. G. G. M., & Hollenbach, D. 1985, ApJ, 291, 722
Tonry J. L., Dressler A., Blakeslee J. P., Ajhar E. A., Fletcher A. B., Luppino G. A., Metzger M. R., Moore C. B.,
2001, ApJ, 546, 681
Werner, M. W., Roellig, T. L., Low, F. J., et al. 2004, ApJS, 154, 1
Wright, E. L., Mather, J. C., Bennett, C. L., et al. 1991, ApJ, 381, 200