2 Data reduction and observational results

Full grating scan spectra of 43-197 $\mu$m were obtained for 9 galaxies with the LWS01 mode in the open time programs of TONAKA.IRGAL and GALIR. In addition, we extracted LWS01 full grating scan data of 25 galaxies from the ISO archival database for a total of 34 nearby galaxies LWS spectra that were analyzed in the present study. The sample includes various types of galaxies, ranging from active galactic nuclei (AGNs), starburst, to normal galaxies. Table 1 lists the present sample, where the flux density at 60 $\mu$m and the FIR color R(60/100) are derived from the LWS spectra convolved with the IRAS band filters (for the data reduction, see below).

 

 

Table 1: List of the galaxies in the present analysis.

Galaxy	Type	Morphology	ISO(60)a[Jy]	R(60/100)b	visual size [$^\prime$]
Cen A	AGN	S0 pec, Sy2	98.4	0.43	$25.7 \times 20.0$
Circinus	AGN	SA(s)b:, Sy2	334.1	0.71	$6.9 \times 3.0$
IC 2554	normal	SB(s)bc pec:	15.9	0.52	$3.1 \times 1.3$
IRAS 00506+7248	normal		24.1	0.66
IRAS 13242-5713	normal	S...	89.2	0.70	$1.1: \times 0.2$
M 51	normal	SA(s)bc pec, HII, Sy2.5	36.4	0.38	$11.2 \times 6.9$
M 82	starburst	I0, Sbrst, HII	1486.6	0.87	$11.2 \times 4.3$
M 83	starburst	SAB(s)c, HII, Sbrst	138.0	0.60	$12.9 \times 11.5$
Maffei2	normal	SAB(rs)bc:	94.8	0.47	$5.82 \times 1.57$
NGC 1068	AGN	(R)SA(rs)b, Sy1, Sy2	206.3	0.62	$7.1 \times 6.0$
NGC 1097	starburst	(R'1:)SB(r'l)b, Sy1	49.6	0.54	$9.3 \times 6.3$
NGC 1365	AGN	(R')SBb(s)b, Sy1.8	92.0	0.52	$11.2 \times 6.2$
NGC 2146	starburst	SB(s)ab pec, HII	163.3	0.65	$6.0 \times 3.4$
NGC 253	starburst	SAB(s)c, HII, Sbrst	1044.7	0.70	$27.5 \times 6.8$
NGC 3256	starburst	Pec, merger, HII	107.8	0.71	$3.8 \times 2.1$
NGC 3690	starburst	IBm pec, HII	121.7	1.00	$2.0 \times 1.5$
NGC 4038	starburst	SB(s)m pec	21.5	0.52	$5.2 \times 3.1$
NGC 4041	normal	SA(rs)bc:	11.7	0.43	$2.7 \times 2.5$
NGC 4414	normal	SA(rs)c?	24.8	0.38	$3.6 \times 2.0$
NGC 4945	starburst	SB(s)cd: sp, Sy2	577.2	0.50	$20.0 \times 3.8$
NGC 520	starburst		37.9	0.65	$1.9 \times 0.7$
NGC 5430	starburst	SB(s)b, HII, Sbrst	9.2	0.51	$2.2 \times 1.1$
NGC 5937	normal	(R')SAB(rs)b pec	10.0	0.46	$1.9 \times 1.1$
NGC 6156	normal	(R'1)SAB(rs)c	20.9	0.61	$1.6 \times 1.4$
NGC 6240	starburst	I0: pec, LINER, Sy2	25.8	0.87	$2.1 \times 1.1$
NGC 6764	starburst	SB(s)bc, LINER, Sy2	4.6	0.50	$2.3 \times 1.3$
NGC 6810	normal	SA(s)ab:sp, Sy2	17.0	0.48	$3.2 \times 0.9$
NGC 6824	normal	SA(s)b:	6.8	0.43	$1.7 \times 1.2$
NGC 6946	starburst	SAB(rs)cd, HII	58.6	0.50	$11.5 \times 9.8$
NGC 7469	starburst	(R')SAB(rs)a, Sy1.2	24.7	0.69	$1.5 \times 1.1$
NGC 7552	starburst	(R')SB(s)ab, HII, LINER	74.7	0.59	$3.4 \times 2.7$
NGC 7582	AGN	(R'1)SB(s)ab, Sy2	51.9	0.64	$5.0 \times 2.1$
NGC 7714	starburst	SB(s)b:pec, HII, LINER	12.7	0.97	$1.9 \times 1.4$
NGC 891	normal	SA(s)b? sp, HII	24.3	0.31	$13.5 \times 2.5$

^a Flux density at 60 $\mu$m derived from the LWS spectra convolved with the IRAS band filters.

^b Ratio of the 60 $\mu$m and 100 $\mu$m flux densities (in Jy) derived from the LWS spectra with the IRAS band filters.

We excluded Arp220 and II ZW40 from the present analysis, both of which have also been observed with LWS01. The former shows a spectrum optically thick even in the far-infrared (Fischer 2000; Malhotra et al. 2001b), for which the present simple analysis cannot be applied, and the continuum of the latter object is faint and suffered from the uncertainty in the dark current estimate. The spectral resolution ( $\lambda/\Delta\lambda$) of LW01 mode was about 200 and the beam size of LWS was estimated to be $60\hbox{$^{\prime\prime}$ }$- $80\hbox{$^{\prime\prime}$ }$ (Gry 2000). The sample galaxies have optical sizes of $1\hbox{$^\prime$ }$- $30\hbox{$^\prime$ }$ and thus for some galaxies only the central portion was included in the LWS beam. Smith & Harvey (1996) reported observations of far-infrared emitting regions in external galaxies from the Kuiper Airborne Observatory (KAO), indicating that most far-infrared emission is concentrated in the central $30\hbox{$^{\prime\prime}$ }$ regions of the galaxies. LWS observations are supposed to detect most of the far-infrared emission and therefore probe the properties of central part of the galaxies.

In the present study, we used the Standard Processed Data (SPD) of off-line processing (OLP) version 9 products provided by the ISO data center. The dark current and the drift in the detector responsivity were corrected by using the LWS Interactive Analysis software (LIA version 7.3). The ISO Spectral Analysis Package (ISAP version 1.6a) was then used for further data reduction. The continuum spectra were stitched together by shifting each detector signal with the offset method, adjusted to the SW5 channel in most galaxies. In some cases where the SW5 channel is noisy, the adjustment was made referring to the LW3 or LW4 channels. The offsets between the detectors were typically less than 20%. The line flux, the total far-infrared flux, and the dust temperature of the continuum emission were derived by ISAP. The 60 and 100 $\mu$m flux densities from the LWS spectra were found to agree with the IRAS data within about 20%. The uncertainties in LWS spectra were suggested to be about 15-20% in previous works (e.g., Braine & Hughes 1999; Unger et al. 2000) and we adopt 20% errors for the flux uncertainty.

The continuum emission shorter than the 60 $\mu$m region is affected by the contribution from very small grains. To derive a typical temperature of submicron grains $T_{\rm d}$ in each galaxy, we fitted the LWS spectrum for $\lambda \ge 80$ $\mu$m with the following equation:

$$F(\lambda) = \Omega\; \tau_{0.55} \left(\frac{0.55~ \mu{\rm m}}{\lambda} \right) B_{\lambda}(T_{\rm d}),$$ (1)

where $\Omega$, $\tau _{0.55}$, and $B_{\lambda}(T_{\rm d})$ are the solid angle of the object, the optical depth at 0.55 $\mu$m, and the Planck function of temperature $T_{\rm d}$, respectively. The parameters $\Omega$ and $\tau _{\rm0.55}$ cannot be determined independently and only the product of them is a meaningful parameter. We assumed that the dust emissivity is proportional to 1/$\lambda$, which provides reasonably good fits for the present sample of galaxies. A typical example of the fit is shown in Fig. 1.

Figure 1: The LWS spectrum and the fitted curve (Eq. (1)) for M 82.

We then define the total infrared integrated flux from submicron dust grains as

$${\it FIR} = \int_{0}^{\infty} F(\lambda){\rm d}\lambda.$$ (2)

Note that FIR does not include the excess emission shorter than 80 $\mu$m, which is attributed to the emission from very small dust grains. The uncertainties in the flux level of the SW1 detector due to the memory effect and the spectral shape of the excess emission make it difficult to estimate the excess emission accurately. The excess emission may be about 30% of FIR according to the study of the diffuse Galactic emission of Dwek et al. (1997). The present FIR is larger by 30% on average than the far-infrared flux for 42-122 $\mu$m estimated from the IRAS 60 and 100 $\mu$m flux densities (Helou et al. 1988).

The observational results are summarized in Table 2.

  In most of the sample galaxies, far-infrared forbidden lines, such as [O I]63 $\mu$m, [O III]88 $\mu$m, [N II]122 $\mu$m, and [C II]158 $\mu$m, were detected. [O I]145 $\mu$m was detected only in 13 galaxies, while only 8 galaxies show detectable [O III]52 $\mu$m emission. The indicated errors include the relative errors estimated from the uncertainty in the base line fit and 20% flux uncertainty. The error in FIR is estimated from the uncertainty in the temperature determination, which mostly comes from the goodness of the fit.

[C II]158 $\mu$m is one of the most important lines for the diagnosis of physical conditions of PDRs because of its large luminosity and low critical density for collisional excitation. However, the carbon atom has a lower ionization energy (11.26 eV) than hydrogen (13.6 eV), and carbon ions are expected to be present not only in the neutral region, such as PDRs, but also in the ionized regions.

[N II]122 $\mu$m is a good tracer of diffuse low-density ionized gas, such as ELDWIM, because the ionization energy of nitrogen atom nearly equals to that of hydrogen atom and the critical density for collisional excitation by electrons is about 300 electrons cm^-3.

Oxygen atoms have an ionization energy of 13.6 eV almost the same as that of hydrogen. [O I]63 $\mu$m is one of the most luminous lines as well as [C II]158 $\mu$m and it becomes a more efficient cooling line than [C II]158 $\mu$m in high-density gases. Together with the upper level transition at 145 $\mu$m, it is an important probe for neutral gas. [O I]145 $\mu$m was weak and detected only in a limited number of galaxies, and we cannot examine the major fraction of the sample galaxies by using the [O I]145 $\mu$m line. [O III]88 $\mu$m is a luminous line of dense ionized gas. It has an upper level transition at 52 $\mu$m and the line ratio of the 52 $\mu$m to 88 $\mu$m emission can be used to derive the electron density of the ionized gas (e.g., Moorwood et al. 1980). Unfortunately, the spectra in the 52 $\mu$m region do not have a sufficient signal to noise ratio to derive a reliable [O III]52 $\mu$m line intensity for most of the present sample galaxies.

In Table 3 we list the electron density estimated from the ratio of the [O III] lines and the neutral hydrogen density estimated from that of the [O I] lines for the galaxies in which [OIII]52 $\mu$m emission was detected. These are rough estimates and should be taken with caution because of the large errors in the obtained line ratios. For about a half of the galaxies with the detected [O III]52 $\mu$m emission, the line ratio is near the low-density limit and only upper limits of the electron density are given. In the derivation of the neutral hydrogen density we assume that the gas temperature is 1000 K. Even with this temperature the line ratios are in the low-density limit for the galaxies listed in Table 3. For lower temperatures the upper limits will further be decreased.

In Fig. 2a,

Figure 2: a) Ratios of the [C II]158 $\mu$m flux and [N II]122 $\mu$m flux to the total far-infrared flux FIR against the far-infrared color R(60/100). Filled circles indicate [C II] $/{\it FIR}$ and open squares [N II] $/{\it FIR}$. b) Ratios of the [O I]63 $\mu$m flux and [O III]88 $\mu$m flux to FIR against R(60/100). Filled circles indicate [O I] $/{\it FIR}$ and open squares [O III] $/{\it FIR}$.

the ratios of [C II]158 $\mu$m and [N II]122 $\mu$m fluxes to the far-infrared flux, [C II] $/{\it FIR}$ and [N II] $/{\it FIR}$, are plotted against the far-infrared color R(60/100) while the ratio [O I]63 $\mu$m and [O III]88 $\mu$m to FIR, [O I] $/{\it FIR}$ and [O III] $/{\it FIR}$, are plotted against R(60/100) in Fig. 2b. Figure 2a indicates a trend that the ratio [C II] $/{\it FIR}$ decreases with R(60/100). The ratio [N II] $/{\it FIR}$ also seems to decrease with R(60/100). On the other hand, [O I] $/{\it FIR}$ does not show a clear systematic trend with R(60/100). There seems a weak trend that [O III] $/{\it FIR}$ increases with the color, though the scatter is quite large.

Similar trends in the ratios of the line intensities to the far-infrared intensity have been obtained for the normal galaxy sample (Malhotra et al. 1997, 2001b). The present sample includes not only normal galaxies but also starburst galaxies and AGNs. In Fig. 3,

 

 

Table 3: Densities estimated from the line ratios.

object	$n_{\rm e}$ (cm-3)a	n (cm-3)b
Cen A	<140	<56000
Circinus	$170 \pm 130$	<10000
M 82	$200 \pm 120$	<1000
NGC 1068	$200 \pm 120$	<48000
NGC 2146	$125 \pm 100$	<45000
NGC 253	<70	-c
NGC 3690	<180	<38000

^a The electron density derived from the [O III]52 $\mu$m to 88 $\mu$m line ratio.

^b The neutral hydrogen density derived from the [O I]145 $\mu$m to 63 $\mu$m line ratio for the gas temperature of 1000 K.

^c The line ratio of [O I] is too large and no reasonable density can be derived for NGC 253 (see Fig. 7 and next section).

Figure 3: a) Ratio of the [C II]158 $\mu$m line flux to the total far-infrared flux FIR against the far-infrared R(60/100). b) Ratio of the [O I]63 $\mu$m line flux to FIR against R(60/100). Filled circles are taken from Malhotra et al. (2000) and open squares indicate the present sample. The value of R(60/100) is based on IRAS data and FIR is derived by the formula of Helou et al. (1988).

we plot the data for normal galaxies by Malhotra et al. (2001b) (filled circles) together with the present sample (open squares) for comparison. In the figure, the total far-infrared flux FIR[IRAS] is derived by the formula of Helou et al. (1988) based on the IRAS data to make direct comparison. Both show a similar trend to each other except that the data of normal galaxies seem to have a steeper trend with the color. AGNs (e.g., Cen A, Circinus, NGC 1068, and NGC 7582) are located in the same trend as normal and starburst galaxies, suggesting that the far-infrared emission of AGNs is driven mainly by star-formation activities.