1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3: \newcommand{\myemail}{schnee@astro.caltech.edu}
4:
5: \shorttitle{Dust in Perseus}
6: \shortauthors{Schnee, S. et al.}
7:
8: \begin{document}
9:
10: \newcommand{\cso}{C$^{17}$O}
11: \newcommand{\ceo}{C$^{18}$O}
12: \newcommand{\cs}{C$^{34}$S}
13: \newcommand{\dcop}{DCO$^+$}
14: \newcommand{\nthp}{N$_2$H$^+$}
15: \newcommand{\ntdp}{N$_2$D$^+$}
16: \newcommand{\kms}{km s$^{-1}$}
17:
18: \title{Dust Emission from the Perseus Molecular Cloud}
19: \author{S. Schnee\altaffilmark{1}, J. Li\altaffilmark{2}, A. A. Goodman\altaffilmark{2} \& A. I. Sargent\altaffilmark{1}}
20:
21: \email{schnee@astro.caltech.edu}
22:
23: \altaffiltext{1}{Division of Physics, Mathematics and Astronomy,
24: California Institute of Technology, 770 South Wilson Avenue,
25: Pasadena, CA 91125}
26: \altaffiltext{2}{Harvard-Smithsonian Center for Astrophysics, 60 Garden
27: Street, Cambridge, MA 02138}
28:
29:
30: \begin{abstract}
31: Using far-infrared emission maps taken by IRAS and Spitzer and a
32: near-infrared extinction map derived from 2MASS data, we have made
33: dust temperature and column density maps of the Perseus molecular
34: cloud. We show that the emission from transiently heated very small
35: grains and the big grain dust emissivity vary as a function of
36: extinction and dust temperature, with higher dust emissivities for
37: colder grains. This variable emissivity can not be explained by
38: temperature gradients along the line of sight or by noise in the
39: emission maps, but is consistent with grain growth in the higher
40: density and lower temperature regions. By accounting for the
41: variations in the dust emissivity and VSG emission, we are able to map
42: the temperature and column density of a nearby molecular cloud with
43: better accuracy than has previously been possible.
44: \end{abstract}
45:
46: \keywords{ISM: clouds --- dust, extinction --- surveys}
47:
48: \section{Introduction}
49:
50: Far-infrared and sub-millimeter continuum radiation from interstellar
51: clouds is typically attributed mostly to ``dust.'' In truth, however,
52: even the contributions from ``dust'' come from multiple populations of
53: grain types, which are each at a mixture of temperatures. Along any
54: given line of sight, there may be random or systematic variations both
55: in the admixture of grain types and in their temperature, and those
56: variations may or may not be correlated. Historically, researchers
57: have used more and more complicated procedures to account for dust
58: emission and estimate column density.
59:
60: The simplest way to infer column density from ``thermal'' dust
61: emission is to assume a single grain type, radiating with a single
62: efficiency (opacity), at a single temperature. In that case, flux at
63: any wavelength can be converted to a column density, assuming a dust
64: temperature. The next-most-sophisticated approach is to use
65: observations at two wavelengths, and to infer a color temperature, by
66: assuming that the dust emission comes from a modified blackbody, which
67: has a wavelength-dependent emissivity one takes as given
68: \citep[e.g.,][]{Wood94}.
69:
70: With observations at three wavelengths, it should be possible to also
71: constrain the dust emissivity, assuming its functional dependence on
72: wavelength is known and is described by just one parameter
73: ($\kappa_\lambda \propto \lambda^{-\beta}$). As \citet{Schnee07} have
74: shown, however, a three-wavelength constraint on $T_d$, $N_d$ and
75: $\beta$ independently is only possible for the case of very (to date
76: unrealistically) high signal-to-noise data. Nevertheless, with three
77: wavelengths of very high S/N, or measurements at four or more
78: wavelengths, one could constrain the temperature, column density and
79: emissivity independently -- if only all the mixtures of grains along
80: lines of sight were similar, and the temperature distribution along
81: the line of sight were isothermal.
82:
83: In an earlier paper, \citep[hereafter SBG]{Schnee06}, we showed that
84: line-of-sight variations in the dust temperature account for nearly
85: all the scatter in a plot of dust-emission vs. dust-extinction derived
86: column density in the Perseus molecular cloud. In \citet[hereafter
87: SRGL]{Schnee05}, we found that Very Small Grains (VSG's) are likely
88: to account for a large fraction ($> 50$\%) of the dust emission at
89: wavelengths shorter that 100 \micron. If that fraction is relatively
90: constant, then one can say that Big Grains (BG's) are a good proxy for
91: column density, modulo the uncertainty imposed by the line-of-sight
92: variations in the dust temperature. However, if the VSG contribution
93: is not constant, but depends on either $N_d$ or $T_d$, then the
94: situation is further confounded. In this paper, we explore all of
95: these possibilities, and we find, in agreement with earlier works
96: \citep[e.g.,][]{Stepnik03}, that the VSG contribution does seem to
97: vary systematically with $T_d$ and $N_d$, and that it is important to
98: account for this effect in estimates of dust column density and dust
99: temperature.
100:
101: \section{Observations}
102:
103: SRGL showed that far-infrared emission maps can be used to
104: determine the dust temperature and column density in molecular clouds
105: once the conversion between FIR optical depth and $V$-band extinction
106: has been determined. This analysis took advantage of the IRIS
107: (Improved Reprocessing of the IRAS Survey) maps at 60 and 100
108: \micron\ \citep{Miville05} and near-infrared extinction map created
109: from 2MASS data as part of the COMPLETE (COordinated Molecular Probe
110: Line Extinction and Thermal Emission) Survey of Star Forming Regions
111: \citep{Ridge06}. Subsequently, SBG demonstrated that the column
112: density determined from pairs of FIR emission maps should become more
113: accurate and less biased when the emission is measured at longer
114: wavelengths due to the decreased influence of line-of-sight
115: temperature variations.
116:
117: With the recent release of 70 and 160 \micron\ maps of Perseus taken
118: with the Spitzer Space Telescope \citep{Evans03, Rebull07}, we can
119: improve upon the SRGL analysis. In this short paper, we combine the
120: 100 \micron\ IRIS and 160 \micron\ Spitzer maps to derive dust
121: temperature and column density maps that are not strongly biased by
122: the emission from transiently heated VSG's. In addition we are able
123: to estimate the variation of the VSG emission and BG emissivity with
124: $T_d$ and $A_V$.
125:
126: \subsection{Spitzer} \label{SPITZER}
127:
128: We used the 70 and 160 \micron\ emission maps of the Perseus molecular
129: cloud from the Spitzer Science Archive that were taken as part of the
130: Spitzer Legacy Program ``From Molecular Cores to Planet-forming
131: Disks'' (c2d) \citep{Evans03, Rebull07}. Map resolutions are
132: 18\arcsec\ and 40\arcsec\ at 70 and 160 \micron, respectively. The
133: c2d delivery includes error maps, which account for systematic and
134: random errors. The overall gain calibration uncertainty is $\sim$5\%
135: at 70 \micron\ \citep{Gordon07} and $\sim$12\% at 160 \micron\
136: \citep{Stansberry07}. The c2d error maps show that the random error
137: for maps smoothed to 5\arcmin\ resolution (the resolution of the
138: extinction map) is $<$1\% at both 70 and 160 \micron.
139:
140: We applied color corrections to these archival fluxes since they were
141: determined assuming that the emission is from a blackbody at 10,000 K
142: while we have assumed that the emission comes from a modified
143: blackbody with emissivity spectral index $\beta = 2$ \citep{Draine84}
144: and equilibrium temperature to be determined, but certainly less than
145: 10,000 K \citep{Schlegel98}. The color correction, which is combined
146: with the dust temperature calculation, is futher described in
147: Sect.~\ref{TD}.
148:
149: \subsection{IRAS/IRIS} \label{IRIS}
150:
151: IRIS flux density maps at 60 and 100 \micron, in contrast with earlier
152: releases of the IRAS all-sky maps, have been corrected for the effects
153: of zodiacal dust and striping, and have the proper gain and offset
154: calibration \citep{Miville05}. The IRIS maps of Perseus have a
155: resolution of approximately 4\farcm 3 (though we smooth them to the
156: 5\arcmin\ resolution of our NIR extinction map for all analyses). The
157: noise in the IRAS maps is approximately 0.03 MJy sr$^{-1}$ and 0.06
158: MJy sr$^{-1}$ at 60 and 100 \micron\ \citep{Miville05}, which is
159: $\lesssim$1\% of the flux in the portion of Perseus included in the
160: c2d survey. Like the Spitzer maps, these were color-corrected by us
161: (following the IRAS Explanatory Supplement VI C.3). For consistency,
162: we consider only the region of Perseus covered by the c2d MIPS maps at
163: 70 and 160 \micron\ in this analysis.
164:
165: \subsection{NIR Extinction Map} \label{2MASS}
166:
167: An extinction map of Perseus based on the Two Micron All Sky Survey
168: (2MASS) point source catalog was created as part of the COMPLETE
169: survey \citep{Ridge06} using the NICER algorithm, devised by
170: \citet{Lombardi01}. This algorithm uses the near-infrared color excess
171: of background stars to estimate the column density of foreground dust.
172: When directly comparing NIR and FIR-derived column densities (e.g.,
173: Figs.~\ref{TAUAVTEMP}, \ref{AVSCATTER} and \ref{PLOTHIST}), all maps
174: have been smoothed to a common resolution of 5\arcmin, since that is
175: the resolution of the NICER extinction map of Perseus.
176:
177: \section{Analysis} \label{ANALYSIS}
178:
179: The flux density emitted by dust in thermal equilibrium is given by
180: \begin{equation} \label{DUSTFLUX}
181: S_{\nu} = \Omega B_{\nu}(T_d) N_d \alpha \nu^{\beta},
182: \end{equation}
183: where
184: \begin{equation} \label{BLACKBODY}
185: B_\nu(T_d) = \frac{2h\nu^3}{c^2} \frac{1}{\exp(h\nu /kT_d)-1}
186: \end{equation}
187:
188: In Equation \ref{DUSTFLUX}, $S_{\nu}$ is the observed flux density at
189: frequency $\nu$; $\Omega$ is the solid angle of the beam;
190: $B_{\nu}(T_d)$ is the blackbody emission from the dust at temperature
191: $T_d$; $N_d$ is the column density of dust, $\alpha$ is a constant
192: that relates the flux density to the optical depth of the dust at
193: frequency $\nu$, and $\beta$ is the emissivity spectral index of the
194: dust, which we assume to be 2 throughout this paper \citep{Draine84}.
195: In SRGL we showed that $\beta = 2$ is a reasonable value for the cold
196: dust in Perseus.
197:
198: Equation \ref{DUSTFLUX} is only true for BG's in thermal equilibrium,
199: and does not describe the emission from stochastically heated VSG's,
200: which cannot be characterized by an equilibrium temperature
201: \citep{Draine01}. The VSG's are expected to be a significant
202: component of the observed emission in the IRAS 60 \micron\ and Spitzer
203: 70 \micron\ bands, but are not expected to contribute much in the IRAS
204: 100 \micron\ and Spitzer 160 \micron\ bands \citep{Li01}. If we were
205: to ignore the VSG emission, we would systematically overestimate the
206: dust temperature and underestimate the column density in Perseus. We
207: therefore account for it using a procedure that improves upon that
208: adopted by SRGL, as described below.
209:
210: \subsection{Dust Temperature} \label{TD}
211:
212: The fluxes quoted in the IRAS and Spitzer emission maps are derived
213: from the total energy detected convolved with the wavelength-dependent
214: instrumental response and an assumed SED of the emitting source. The
215: quoted flux density at wavelength $\lambda_0$ is related to the
216: intrinsic flux density by:
217:
218: \begin{equation} \label{FQUOTE}
219: F_{\lambda_0}^{quoted} = \frac{\int (F_\lambda^{intrinsic}/
220: F_{\lambda_0}^{intrinsic}) \lambda R_\lambda d\lambda} {\int
221: (F_\lambda^{default} / F_{\lambda_0}^{default}) \lambda R_\lambda
222: d\lambda} F_{\lambda_0}^{intrinsic}
223: \end{equation}
224: \citep{Stansberry07} where $F_\lambda^{intrinsic}$ is the true source
225: SED, $F_\lambda^{default}$ is the default SED and $R_\lambda$ is the
226: instrumental response function (the sensivity of the detector as a
227: function of wavelength). The default SED for the Spitzer data is that
228: of a 10,000 K blackbody, while the default SED for the IRIS data is
229: $F_\lambda \propto 1/ \lambda$. We assume that the intrinsic SED is a
230: modified blackbody with $\beta$ = 2 \citep{Draine84}, so the color
231: correction is temperature dependent.
232:
233: To derive the color correction and dust temperature, we create a
234: lookup table of the intrinsic flux density at $\lambda = $ 60, 70, 100
235: and 160 \micron\ for dust at temperatures between 5 and 100 K, at 0.1
236: K intervals. We then calculate what flux density would have been
237: quoted at those wavelengths using Eq.~\ref{FQUOTE}. Using this table,
238: we are able to derive the color-corrected dust temperature from the
239: ratio of the {\it quoted} flux densities, making the assumption that
240: the dust emission is optically thin and isothermal along each line of
241: sight. The IRAS 100 \micron\ and Spitzer 160 \micron\ maps are not
242: significantly corrupted by VSG emission, so we derive a dust
243: temperature map using our lookup table and the quoted
244: $F_{100}/F_{160}$ after smoothing the emission maps to a common
245: (5\arcmin) resolution.
246:
247: The IRAS 60 and Spitzer 70 \micron\ flux densities come from a
248: combination of BG and VSG emission, but Equations \ref{DUSTFLUX} and
249: \ref{FQUOTE} are only accurate for BG emission. In SRGL we removed the
250: VSG emission by assuming that a constant fraction of all 60 \micron\
251: emission can be attributed to VSG's. This proportion was found by
252: minimizing the difference between the temperature maps derived from
253: IRIS data and the dust temperature map derived by \citet{Schlegel98},
254: which was derived from DIRBE 100 and 240 \micron\ emission (and
255: therefore free from VSG contamination).
256:
257: However, the proportion of 60 and 70 \micron\ flux emitted by VSG's is
258: expected to vary with environment. In regions with high extinction,
259: the VSG's will be shielded from the ISRF and will therefore not be
260: effectively heated. In cold, dense regions dust grains may stick
261: together, removing the VSG population. For instance,
262: \citet{Stepnik03} have shown that the VSG emission is greatly reduced
263: in a dense filament in Taurus where $A_V > 2.1$ and found a
264: corresponding increase in the dust emissivity. A similar transition
265: region of altered dust composition has been reported using IRAS 60 and
266: 100 \micron\ emission maps \citep[e.g.,][]{Laureijs91, Laureijs89,
267: Boulanger90}.
268:
269: Here we take advantage of the higher resolution of the IRAS 100 and
270: Spizter 160 \micron\ images, as compared with the DIRBE maps, to
271: remove the emission from VSG's accounting for spatial variations in
272: their contribution. We divide our 100/160 \micron\ derived
273: temperature maps into ten equally populated temperature and extinction
274: subsets (five temperature subsets for $A_V$ less than the median
275: NIR-derived extinction and five temperature subsets for higher $A_V$)
276: and calculate the VSG contribution to the 60 and 70 \micron\ emission
277: independently for each bin using the dust temperature derived from the
278: 100/160 \micron\ pair of emission maps. In each bin, we multiply the
279: 60 and 70 \micron\ emission (smoothed to 5\arcmin\ resolution) by the
280: scaling factors $f_{60}$ and $f_{70}$ which are chosen such that the
281: temperatures derived from $f_{60}F_{60}/F_{100}$ and
282: $f_{70}F_{70}/F_{160}$ are best matched to the dust temperature
283: derived from $F_{100}/F_{160}$. The comparatively low resolution of
284: the DIRBE-derived temperature map prevented us from performing this
285: more sophisticated correction in SRGL. The scale factors ($f_{60}$
286: and $f_{70}$) used to account for the VSG emission are shown in
287: Fig.~\ref{BGCONVERSION}. Our assumptions that dust is isothermal
288: along each line of sight and that the dust emissivity (characterized
289: by $\alpha$ and $\beta$ in Eq.~\ref{DUSTFLUX}) is constant introduce
290: uncertainties into our calculations. The effect of assuming
291: isothermality is covered in SBG, and we discuss the effects of
292: variable dust emissivity in Section \ref{VARDUST}.
293:
294: \subsection{Column Density: Constant Dust Emissivity} \label{NH}
295:
296: Given the flux density at one wavelength and the dust temperature, the
297: optical depth is given by:
298:
299: \begin{equation} \label{TAUEQ}
300: \tau_\lambda = \frac{S_\lambda}{B_\lambda (T_d)}
301: \end{equation}
302: where $B_\lambda (T_d)$ is the Planck function and $S_\lambda$ is the
303: observed flux density at wavelength $\lambda$.
304:
305: The optical depth at wavelength $\lambda$ can be converted to the $V$-band
306: extinction (A$_V$) using the equation
307:
308: \begin{equation} \label{AVEQ}
309: A_V = X_{\lambda} \tau_\lambda
310: \end{equation}
311: where $X_\lambda$ is the scale factor, depending on the wavelength
312: ($\lambda$) of the emission map, that relates the thermal emission
313: properties of the dust grains to their $V$-band absorption properties
314: (Eq.~5 from SRGL). We solve for $X_\lambda$ for each pair of emission
315: maps by minimizing the difference between the emission-derived column
316: density (smoothed to the resolution of the extinction map) and the
317: 2MASS absorption-derived column density, as in SRGL and SBG. For the
318: 60/100 and 100/160 pair of emission maps we derive the optical depth
319: and conversion factor ($\tau_{100}$ and $X_{100}$) from the 100
320: \micron\ emission map. For the 70/160 pair of emission maps we derive
321: $\tau_{160}$ and $X_{160}$, and scale both values in Figures
322: \ref{BGCONVERSION} and \ref{TAUAVTEMP} to the expected values at 100
323: \micron\ assuming $\beta = 2$ using the scale factor
324: $(\lambda_{100}/\lambda_{160})^\beta = 0.39$.
325:
326: The optimization sets the overall calibration for the FIR data, but
327: does not force the slope of NIR-derived column density plotted against
328: FIR-derived column density to be exactly unity, though it will be
329: close to this value. Variations in temperature along each line of
330: sight will affect the emission-derived $A_V$ at high column density
331: more than at low column density, and the extinction-derived $A_V$ is
332: unaffected by temperature fluctuations. This difference in the way
333: the dust temperature is coupled to the derived column density, along
334: with possible variations in the FIR dust emissivity, can change the
335: slope of the emission vs. absorption-derived column density to be
336: slightly different from unity. Note that Eq.~\ref{AVEQ} assumes that
337: the dust emission and absorption properties are constant. The linear
338: relationship between $\tau_{FIR}$ and $A_V$ is shown in
339: Fig.~\ref{TAUAVTEMP}.
340:
341: \subsection{Column Density: Variable Dust Emissivity} \label{VARDUST}
342:
343: Just as the emission from VSG's is expected to vary with environment,
344: so should the dust emissivity, due to such factors as grain growth
345: through sticking and the formation of icy mantles. As a result of
346: these processes, the dust emissivity in regions with higher column
347: density and lower dust temperature is seen to be higher. This effect
348: has been observed in many environments (e.g. Kiss et al. 2006; del
349: Burgo \& Laureijs 2005; Bernard et al. 1999), where it is shown that
350: the FIR emissivity ($\alpha$ in Eq.~\ref{DUSTFLUX} or $X_\lambda$ in
351: Eq.~\ref{AVEQ}) anti-correlates with dust temperature and varies by a
352: factor of a few in interstellar clouds similar to Perseus. In
353: Fig.~\ref{TAUAVTEMP} we show the FIR dust emissivity (expressed as the
354: ratio of the 100 \micron\ optical depth and NIR-derived extinction)
355: plotted against the color temperature of the dust. We also find this
356: dependence of the emissivity on temperature and column density, which
357: suggests that the linear fit between $\tau_{FIR}$ and A$_V$ (see
358: Eq.~\ref{AVEQ} and Fig.~\ref{TAUAVTEMP}) does not fully describe the
359: relationship between dust emission and absorption.
360:
361: It is important to note that temperature fluctuations along the line
362: of sight can give the appearance of a temperature-dependent emissivity
363: when the dust is assumed to be isothermal, even if the true dust
364: emissivity is constant. Because the line-of-sight averaged dust
365: temperature is emission weighted, the dust temperature is
366: overestimated and the far-infrared optical depth is underestimated.
367: This effect is larger along lines of sight with greater temperature
368: variation (i.e., those passing through denser regions), which are
369: preferentially colder than the more diffuse (and nearly isothermal)
370: portions of the molecular cloud. As a consequence, the derived ratio
371: of the far-infrared optical depth to extinction will be underestimated
372: in the colder, denser regions of the molecular cloud. However, the
373: observed temperature-dependent dust emissivity varies in the {\it
374: opposite direction} of the correlation created by line-of-sight
375: temperature fluctuations. The observed dust emissivity is {\it
376: larger} at low temperatures, supporting the claim that the true dust
377: emissivity is variable.
378:
379: Noise in the FIR emission maps can also give the appearance of a
380: variable dust emissivity. If noise changes the ratio of the fluxes,
381: then the dust temperature will be either overestimated or
382: underestimated. A high dust temperature will result in a low
383: FIR-derived optical depth, and a low dust temperature will result in a
384: high FIR optical depth. To illustrate the effect of noise on the
385: derived dust emissivity, we analyze a simple model of a molecular
386: cloud with constant dust emissivity in which the temperature is
387: anti-correlated with column density and with both quantities are
388: within a typical range for molecular clouds. We then derive the flux
389: at 100 and 160 \micron\ flux that would be emitted and add Gaussian
390: random noise to the maps. We re-derive the dust temperature and 100
391: \micron\ optical depth from the noisy maps and show the dust
392: emissivity that would be estimated from the ``observed'' maps plotted
393: against the temperature. The results are shown in
394: Fig.~\ref{NOISEFIG}. The dust emissivity that we derive in Perseus
395: (see Fig.~\ref{AVSCATTER}) looks similar to what one would expect from
396: maps of dust that have a constant emissivity, but that have noise on
397: the level of $\sim$15\%. However, the random noise in all of our maps
398: is much lower than this ($\lesssim$1\%), so we can rule out the
399: possibility that noise in the Perseus emission maps is tricking us
400: into thinking that the dust emissivity is variable.
401:
402: To account for the variable dust emissivity in the determination of
403: the dust column density, we divide our data into the same ten equally
404: populated bins as we did in Section \ref{TD}, based on their
405: NIR-derived extinction and temperature. Using Eq.~\ref{AVEQ}, we
406: derive the conversion ($X_\lambda$) separately for each bin and each
407: pair of emission maps. The derived dust emissivity, as a function of
408: $T_d$, is plotted in Fig.~\ref{BGCONVERSION}.
409:
410: \section{Results} \label{RESULTS}
411:
412: \subsection{VSG Emission and Dust Emissivity vs. $T_d$ and $A_V$} \label{ENVIRONMENT}
413:
414: As shown in Fig.~\ref{BGCONVERSION}, the 60 and 70 \micron\ emission
415: is dominated by the emission from VSG's. The fractional contribution
416: of BG emission rises with increasing $T_d$ at both 60 and 70 \micron.
417: This is no surprise, given that warmer dust will have larger 60/100
418: and 70/160 \micron\ flux ratios. At 70 \micron, the BG contribution
419: is a larger percentage of the total flux for $A_V > 2.6$ than $A_V <
420: 2.6$, while at 60 \micron\ the difference is smaller and changes sign.
421: This difference may be caused by the 60 and 70 \micron\ passbands,
422: since IRAS is sensitive (sensitiviy $\ge$ 10\%) from 37 to 82
423: \micron\, while Spitzer is sensitive from 55 to 86 \micron.
424:
425: The parameter used to derive the dust column density from the FIR
426: optical depth of the dust, as given by $X_\lambda$ in Eq.~\ref{AVEQ},
427: varies nearly identically with $T_d$ for all three pairs of emission
428: maps. The evolution in $X_\lambda$ is primarily caused by the
429: variable emissivity of the dust (shown in Fig.~\ref{TAUAVTEMP}), but
430: also by line-of-sight temperature fluctuations and noise in the
431: emission maps. Taking into account variations in the VSG contribution
432: to emission from dust with $\lambda < 100$ \micron\ and variations in
433: the emissivity of the BG's represents an increase in the complexity of
434: the analysis performed in SRGL.
435:
436: \subsection{Emission vs Absorption-derived Column Density} \label{SCATTERS}
437:
438: As discussed in SRGL, the point to point scatter between emission and
439: absorption-derived column density can be quite large (a few tens of
440: percent). This scatter can be attributed to line-of-sight temperature
441: fluctions, non-constant dust emissivity and variations in the fraction
442: of emission coming from VSG's (though this last source of error should
443: not affect the column density derived from emission maps at or above
444: $\sim$100 \micron).
445:
446: %In SBG we quantified the effect of temperature fluctuations along the
447: %line of sight on the emission-derived column density. We showed that,
448: %in the absence of variations in the VSG emission and for a constant
449: %dust emissivity, the slope of the best fit line to the emission
450: %vs. absorption-derived column density is always shallower than unity
451: %(so that FIR-derived column density is underestimated at high $A_V$
452: %and overestimated at low $A_V$). The scatter about the best fit line
453: %decreases and the bias becomes smaller (i.e., the slope moves closer
454: %to unity) as the wavelength of the observed emission maps gets longer.
455:
456: In this paper we calculate the emission-derived column density in two
457: ways. In the first method we assume that the dust emissivity is
458: constant in Perseus, and that $A_V$ scales with the FIR optical depth
459: (Eq.~\ref{AVEQ}). Our second method uses a dust emissivity which
460: varies with $A_V$ and $T_d$, as shown in Fig.~\ref{BGCONVERSION}. In
461: both cases we correct for variations in the relative VSG/BG
462: contribution to the 60 and 70 \micron\ emission, as a function of
463: $T_d$ and $A_V$, as described in Section \ref{TD}. The values for the
464: median column density ratio, the scatter around that median, and the
465: slope of the best fit line between the emission vs absorption-derived
466: column density for Perseus are shown in Table \ref{COLUMNTAB} and the
467: corresponding plots are shown in Fig.~\ref{AVSCATTER}.
468:
469: In SBG, we use a simplified model of emission from an externally
470: heated molecular cloud to study the effects of line-of-sight
471: temperature variations on column density estimates without having to
472: worry about the effects of VSG emission or variations in the dust
473: emissivity. Using that model, we predict that the scatter and bias
474: should be largest for the 60/100 \micron\ pair of maps and smallest
475: for the 100/160 \micron\ pair. However, in this paper, when we use a
476: single $\tau - A_V$ conversion, we find that the slope and scatter
477: between our emission and absorption-derived column densities are very
478: similar for all three pairs of emission maps, in conflict with SBG.
479: The amount of scatter in the NIR vs. 60/100 \micron-derived column
480: density is consistent with that predicted by the model in SBG. The
481: larger than expected scatter seen in the 70/160 is probably not caused
482: by an improper VSG correction, since we also find excess scatter in
483: the NIR vs. 100/160 \micron-derived column density, and VSG emission
484: should be insignificant in the 100 and 160 \micron\ maps.
485:
486: When we divide the emission maps into ten subsets and derive the $\tau
487: - A_V$ conversion for each bin (shown in Fig.~\ref{BGCONVERSION}),
488: there is less bias in the emission-derived column density. This is
489: shown by the fit between the FIR and NIR-derived column density in
490: Fig.~\ref{AVSCATTER}, which has a slope $\sim$1. However, the scatter
491: in the relation (see Table \ref{COLUMNTAB}) is only significantly
492: improved for the 100/160 \micron\ derived column density compared to
493: the constant $\tau - A_V$ derivation. This suggests that the emission
494: properties of dust are more complicated than assumed in this paper.
495: For instance, the emissivity (or spectral index) of the dust could be
496: density-dependent or there could be variations in grain properties as
497: a function of position in Perseus, and not merely as a function of
498: dust temperature and column density.
499:
500: We conclude that the 100/160 \micron\ pair of maps is the best for
501: deriving dust temperature and column density because it is least
502: affected by the emission from VSG's and has the least NIR vs. FIR
503: column density scatter. However, since the 70/160 pair of emission
504: maps has much better resolution, the column density and dust
505: temperature maps produced from it are also useful if one is willing to
506: accept the additional uncertainty of using maps presented at
507: $\sim40$\arcsec\ that have been calibrated on a $\sim5$\arcmin\ scale.
508:
509: \subsection{Temperature and Column Density Maps} \label{MAPS}
510:
511: Here we present dust temperature and column density maps derived from
512: the IRAS and Spitzer fluxes. Figures \ref{MAPS160100},
513: \ref{MAPS10060} and \ref{MAPS16070} show the temperature and column
514: density maps derived from the IRAS and Spitzer far-infrared data and
515: are presented at $\sim$5\arcmin\ resolution (which is the resolution
516: of the NIR-derived extinction map and is approximately the resolution
517: of the IRIS 60 and 100 \micron\ maps). Figure \ref{MAPS16070HIGH} is
518: derived from the Spitzer 70 and 160 \micron\ data, and is presented at
519: the 40\arcsec\ resolution of the 160 \micron\ map (the 70 \micron\
520: emission map was smoothed to the resolution of the 160 \micron\ map).
521: The column densities are determined by assuming a variable dust
522: emissivity, as described in Section \ref{VARDUST}. Note that the VSG
523: subtraction and variable dust emissivity are calibrated at 5\arcmin\
524: resolution, so the calibration of the dust temperature and column
525: density using the Spitzer 70 and 160 \micron\ maps present at
526: 40\arcsec\ resolution is more uncertain than the maps presented at
527: 5\arcmin\ resolution.
528:
529: \subsection{Temperature and Column Density Histograms} \label{HISTOGRAMS}
530:
531: Figure \ref{PLOTHIST} shows histograms of the dust temperature and
532: column density distributions derived from from pairs of emission maps
533: at 60/100, 70/160 and 100/160 \micron. The 60 and 70 \micron\ fluxes
534: have been scaled to remove the VSG emission (as explained in Section
535: \ref{TD}). The median temperature derived for this region of the
536: Perseus molecular cloud by \citet{Schlegel98} is 16.5 K, whereas we
537: derive a median temperature of 15.0 K, so our derived temperature is
538: about 10\% lower. The emission-derived column density is scaled to
539: match that derived from the NIR reddening of background stars.
540: Therefore, it is no surprise that the $A_V$ distributions in all three
541: panels of Fig.~\ref{PLOTHIST} are similar to that of the extinction
542: map.
543:
544: \section{Conclusion}
545:
546: Using Spitzer 70 and 160 \micron\ maps released by the c2d Legacy
547: Project, we have created high-resolution dust temperature and column
548: density maps of the Perseus molecular cloud. Although smaller in area
549: than similar maps presented in SRGL using IRAS and 2MASS data, the
550: maps in this paper (Fig.~\ref{MAPS16070}) are better calibrated to
551: account for variations in the dust emissivity and removal of VSG
552: emission. Combining the Spitzer 160 \micron\ map with the IRAS 100
553: \micron\ map we show that the dust temperature in Perseus is
554: significantly colder (by $\sim$10\%) than reported in the dust
555: temperature map derived by \citet{Schlegel98}.
556:
557: We find that the dust emissivity varies with $A_V$ and $T_d$, in
558: agreement with the results of \citet{delBurgo05} and \citet{Kiss06}.
559: By fitting the absorption-derived column density to the FIR opacity in
560: each of ten subsets of the data, we demonstrate an improved method for
561: calculating the emission-derived dust column density. The column
562: density derived from the 60/100 \micron\ pair of emission maps, when
563: compared with the NIR $A_V$, has slope and scatter consistent with
564: those expected from variations in temperature along the line of sight.
565: However, the scatter between the absorption-derived and 70/160 and
566: 100/160 \micron\ emission-derived column density is larger than would
567: be expected from line-of-sight variations in the dust temperature
568: alone. One possible explanation for the large scatter is that the 160
569: \micron\ map is more sensitive to the emission from cold dust grains
570: than are the 60 and 100 \micron\ maps (SBG), and the dust emissivity
571: varies more strongly at colder (T$_d < 15$ K) temperatures. Although
572: the origin of the observed temperature dependence is uncertain, recent
573: theoretical models of amorphous dust grains by \citet{Meny07} and
574: \citet{Boudet05} also exhibit an anti-correlation between dust
575: temperature and emissivity. Another possible cause for the higher
576: dust emissivity at lower temperatures is that the dust grains could be
577: accreting icy mantles in the dense and colder regions within the
578: molecular cloud. Regardless of the reason, it is not surprising that
579: dust grains with a higher emissivity, given their ability to cool more
580: efficiently, will be found at lower equilibrium temperatures.
581:
582: A temperature-dependent dust emissivity presents a challenge for those
583: attempting to derive dust temperatures and column densities from FIR
584: emission maps. In SBG, we suggest that longer-wavelength ($\lambda >
585: 100$ \micron) emission maps should be used to calculate the
586: line-of-sight dust properties because of their decreased sensitivity
587: to fluctuations in the dust temperature and the lack of confusing
588: emission from VSG's. In this paper, we show that the dust emissivity
589: anti-correlates with dust temperature, and that the variability is
590: likely to increase at lower temperatures, an effect which is also seen
591: in other regions \citep[e.g.,][]{delBurgo05, Kiss06}. This
592: complicates the interpretation of longer wavelength emission maps,
593: which are more sensitive to cold dust. The deleterious effects of VSG
594: emission, temperature gradients, and variable dust emission properties
595: add complexity to the determination of the mass and temperature
596: distributions of dust in a molecular cloud. Resolution of these
597: problems may require coverage from tens to hundreds of microns, finer
598: spectral resolution than the broad IRAS and Spitzer filters provide,
599: and comparison with dust emission models of ``realistic'' molecular
600: clouds.
601:
602: Given the complexity of determining column density from dust emission,
603: one may wonder if using extinction might be a superior method. For
604: instance, the extinction map of Perseus made with 2MASS data and the
605: NICER algorithm \citep{Ridge06, Lombardi01} is used as a template with
606: which we calibrate the emission-derived column density in this paper,
607: and we trust extinction-derived column density more than that derived
608: from dust emission in general \citep{Goodman08, Pineda08}. The
609: resolution of extinction maps is limited by the number of background
610: tracers with known color, which provide 1\arcmin\ resolution for
611: clouds nearby the galactic center in projection \citep{Lombardi06},
612: but is coarser towards clouds further away from the galactic plane.
613: However, emission maps are limited only by the size of the detector
614: and also provide information on the dust temperature.
615:
616: \acknowledgments
617:
618: We would like to thank our anonymous referee for substantially
619: improving the scope and clarity of this paper. Scott Schnee
620: acknowledges support from the Owens Valley Radio Observatory, which is
621: supported by the National Science Foundation through grant AST
622: 05-40399. Alyssa Goodman acknowledges support from the National
623: Science Foundation under Grant No. AST-0407172.
624:
625: \begin{thebibliography}{}
626: \bibitem[Bernard et al.(1999)]{Bernard99} Bernard, J.~P., et
627: al.\ 1999, \aap, 347, 640
628: \bibitem[Bethell et al.(2004)]{Bethell04} Bethell, T.~J., Zweibel,
629: E.~G., Heitsch, F., \& Mathis, J.~S.\ 2004, \apj, 610, 801
630: \bibitem[Boudet et al.(2005)]{Boudet05} Boudet, N., Mutschke, H.,
631: Nayral, C., J{\"a}ger, C., Bernard, J.-P., Henning, T., \& Meny,
632: C.\ 2005, \apj, 633, 272
633: \bibitem[Boulanger et al.(1990)]{Boulanger90} Boulanger, F.,
634: Falgarone, E., Puget, J.~L., \& Helou, G.\ 1990, \apj, 364, 136
635: \bibitem[del Burgo \& Laureijs(2005)]{delBurgo05} del Burgo, C., \&
636: Laureijs, R.~J.\ 2005, \mnras, 360, 901
637: \bibitem[Draine \& Lee(1984)]{Draine84} Draine, B.~T., \& Lee,
638: H.~M.\ 1984, \apj, 285, 89
639: \bibitem[Draine \& Li(2001)]{Draine01} Draine, B.~T., \& Li, A.\ 2001,
640: \apj, 551, 807
641: \bibitem[Evans et al.(2003)]{Evans03} Evans, N.~J., II, et al.\ 2003,
642: \pasp, 115, 965
643: \bibitem[Goodman et al.(2008)]{Goodman08} Goodman, A.~A., Pineda,
644: J. \& Schnee, S.\ \apj, in prep.
645: \bibitem[Gordon et al.(2007)]{Gordon07} Gordon, K.~D., et al.\ 2007,
646: \pasp, 119, 1019
647: \bibitem[Heitsch et al.(2001)]{Heitsch01} Heitsch, F., Zweibel, E.~G.,
648: Mac Low, M.-M., Li, P., \& Norman, M.~L.\ 2001, \apj, 561, 800
649: \bibitem[Kiss et al.(2006)]{Kiss06} Kiss, C., {\'A}brah{\'a}m, P.,
650: Laureijs, R.~J., Mo{\'o}r, A., \& Birkmann, S.~M.\ 2006, \mnras,
651: 373, 1213
652: \bibitem[Laureijs et al.(1989)]{Laureijs89} Laureijs, R.~J.,
653: Chlewicki, G., Wesselius, P.~R., \& Clark, F.~O.\ 1989, \aap, 220,
654: 226
655: \bibitem[Laureijs et al.(1991)]{Laureijs91} Laureijs, R.~J., Clark,
656: F.~O., \& Prusti, T.\ 1991, \apj, 372, 185
657: \bibitem[Li \& Draine(2001)]{Li01} Li, A., \& Draine, B.~T.\ 2001,
658: \apj, 554, 778
659: \bibitem[Lombardi et al.(2006)]{Lombardi06} Lombardi, M., Alves, J.,
660: \& Lada, C.~J.\ 2006, \aap, 454, 781
661: \bibitem[Lombardi \& Alves(2001)]{Lombardi01} Lombardi, M., \& Alves,
662: J.\ 2001, \aap, 377, 1023
663: \bibitem[Mathis et al.(1977)]{Mathis77} Mathis, J.~S., Rumpl, W., \&
664: Nordsieck, K.~H.\ 1977, \apj, 217, 425
665: \bibitem[Mathis et al.(1983)]{Mathis83} Mathis, J.~S., Mezger, P.~G.,
666: \& Panagia, N.\ 1983, \aap, 128, 212
667: \bibitem[Meny et al.(2007)]{Meny07} Meny, C., Gromov, V., Boudet, N.,
668: Bernard, J.~-., Paradis, D., \& Nayral, C.\ 2007, ArXiv Astrophysics
669: e-prints, arXiv:astro-ph/0701226
670: \bibitem[Miville-Desch{\^e}nes \& Lagache(2005)]{Miville05}
671: Miville-Desch{\^e}nes, M.-A., \& Lagache, G.\ 2005, \apjs, 157, 302
672: \bibitem[Pineda et al.(2008)]{Pineda08} Pineda, J.~E., Caselli,
673: P., \& Goodman, A.~A.\ 2008, ArXiv e-prints, 802, arXiv:0802.0708
674: \bibitem[Rebull et al.(2007)]{Rebull07} Rebull, L.~M., et al.\ 2007,
675: \apjs, 171, 447
676: \bibitem[Ridge et al.(2006)]{Ridge06} Ridge, N.~A., et al.\ 2006, \aj,
677: 131, 2921
678: \bibitem[Schlegel et al.(1998)]{Schlegel98} Schlegel, D.~J.,
679: Finkbeiner, D.~P., \& Davis, M.\ 1998, \apj, 500, 525
680: \bibitem[Schnee et al.(2007)]{Schnee07} Schnee, S., Kauffmann, J.,
681: Goodman, A., \& Bertoldi, F.\ 2007, \apj, 657, 838
682: \bibitem[Schnee et al.(2006)]{Schnee06} Schnee, S., Bethell, T., \&
683: Goodman, A.\ 2006, \apjl, 640, L47
684: \bibitem[Schnee et al.(2005)]{Schnee05} Schnee, S.~L., Ridge, N.~A.,
685: Goodman, A.~A., \& Li, J.~G.\ 2005, \apj, 634, 442
686: \bibitem[Stansberry et al.(2007)]{Stansberry07} Stansberry, J.~A., et
687: al.\ 2007, \pasp, 119, 1038
688: \bibitem[Stepnik et al.(2003)]{Stepnik03} Stepnik, B., et al.\ 2003,
689: \aap, 398, 551
690: \bibitem[Wood et al.(1994)]{Wood94} Wood, D.~O.~S., Myers, P.~C., \&
691: Daugherty, D.~A.\ 1994, \apjs, 95, 457
692: \end{thebibliography}
693:
694: \clearpage
695:
696: \begin{deluxetable}{lrrrrrrrrr}
697: \tablewidth{0pt}
698: \tabletypesize{\scriptsize}
699: \tablecaption{Emission vs. Absorption-Derived Column Density
700: \label{COLUMNTAB}}
701: \tablehead{
702: \colhead{Emission Maps} & \colhead{$\mu$\tablenotemark{a}} &
703: \colhead{$\mu$\tablenotemark{b}} & \colhead{$\sigma$\tablenotemark{c}} &
704: \colhead{$\sigma$\tablenotemark{d}} & \colhead{slope\tablenotemark{e}} &
705: \colhead{slope\tablenotemark{f}} \\
706: \colhead{} & \colhead{constant} &
707: \colhead{variable} & \colhead{constant} &
708: \colhead{variable} & \colhead{constant} &
709: \colhead{variable}}
710: \startdata
711: 60/100 \micron & 1.01 & 1.00 & 0.32 & 0.31 & 0.80 & 0.88 \\
712: 70/160 \micron & 1.00 & 1.00 & 0.34 & 0.32 & 0.90 & 1.03 \\
713: 100/160 \micron & 1.00 & 1.00 & 0.30 & 0.25 & 0.77 & 0.91
714: \enddata
715: \tablenotetext{a}{Median ratio of $A_{V,emission}/A_{V,extinction}$ assuming a constant dust emissivity}
716: \tablenotetext{b}{Median ratio of $A_{V,emission}/A_{V,extinction}$ assuming a variable dust emissivity}
717: \tablenotetext{c}{1 $\sigma$ scatter in $A_{V,emission}/A_{V,extinction}$ assuming a constant dust emissivity}
718: \tablenotetext{d}{1 $\sigma$ scatter in $A_{V,emission}/A_{V,extinction}$ assuming a variable dust emissivity}
719: \tablenotetext{e}{Best fit slope from Figure \ref{AVSCATTER} (left)}
720: \tablenotetext{f}{Best fit slope from Figure \ref{AVSCATTER} (right)}
721: \end{deluxetable}
722:
723: \clearpage
724:
725: \begin{figure}
726: \epsscale{0.8}
727: \plotone{f1.eps}
728: \caption{({\it top}) The percentage of the observed 60 and 70 \micron\
729: flux emitted by BG's in thermal equilibrium as a function of dust
730: temperature. The remaining portion is emitted by stochastically
731: heated VSG's. The error bars show the standard deviation in each
732: bin. ({\it bottom}) The ratio of the observed extinction to the 100
733: \micron\ optical depth as a function of dust temperature. For the
734: 70/160 \micron\ pair of emission maps, we derive $\tau_{100}$ assuming
735: that $\beta = 2$ and scaling $\tau_{160}$ by $(100/160)^\beta$. In
736: both the top and bottom plots there are $151 \pm 15$ independent
737: 5\arcmin\ data values in each bin.
738: \label{BGCONVERSION}}
739: \end{figure}
740:
741: \begin{figure}
742: \epsscale{1.0}
743: \plotone{f2.eps}
744: \caption{({\it left}) The 100 \micron\ optical depth plotted against
745: the column density (expressed in terms of $V$-band extinction) derived
746: from the 2MASS map. ({\it right}) The ratio of the NIR-derived column
747: density divided by the FIR-derived optical depth (at 100 \micron)
748: plotted against the FIR-derived dust temperature. \label{TAUAVTEMP}}
749: \end{figure}
750:
751: \begin{figure}
752: \epsscale{1.1}
753: \plotone{f3.eps}
754: \caption{({\it Top}) A model temperature and column density
755: distribution, which will be used to generate synthetic emission maps
756: and test the effects of noise on the derived dust emissivity. ({\it
757: Middle and Bottom}) The dust emissivity, expressed as the ratio of the
758: 100 \micron\ optical depth and the column density, plotted against
759: dust temperature. The optical depth and dust temperature are derived
760: from synethic flux maps at 100 and 160 \micron, with relative noise
761: levels ranging from 1\% to 25\%. The ``true'' flux is calculated from
762: the column density and temperature distibution in the top panel, and
763: is then adjusted by Gaussian random noise at the level shown in each
764: middle and lower panel. \label{NOISEFIG}}
765: \end{figure}
766:
767: \begin{figure}
768: \epsscale{1.0}
769: \plotone{f4_color.eps}
770: %\plotone{f4_bw.eps}
771: \caption{({\it left}) The FIR-derived column density plotted against
772: the NIR-derived column density. Here we assume a single value for the
773: dust emissivity for each pair of emission maps. ({\it right}) The
774: FIR-derived column density plotted against the NIR-derived column
775: density. Here we assume a variable value for the dust emissivity for
776: each pair of emission maps, as plotted in Fig.~\ref{BGCONVERSION}.
777: \label{AVSCATTER}}
778: \end{figure}
779:
780: \begin{figure}
781: \epsscale{1.1}
782: \plotone{f5_color.eps}
783: %\plotone{f5_bw.eps}
784: \caption{({\it top}) The dust temperature map and ({\it bottom})
785: column density (expressed in terms of $V$-band extinction) derived
786: from the IRAS 100 and Spitzer 160 \micron\ flux density maps, at the
787: 5\arcmin\ resolution of our NIR-derived extinction map. The column
788: density is derived assuming a variable dust emissivity. Red contours
789: in the bottom plot show the regions with extinction of $A_V = 3, 5, 7,
790: 9$.
791: \label{MAPS160100}}
792: \end{figure}
793:
794: \begin{figure}
795: \epsscale{1.1}
796: \plotone{f6_color.eps}
797: %\plotone{f6_bw.eps}
798: \caption{({\it top}) The dust temperature map and ({\it bottom})
799: column density (expressed in terms of $V$-band extinction) derived
800: from the IRAS 60 and 100 \micron\ flux density maps, at the 5\arcmin\
801: resolution of our NIR-derived extinction map. The column density is
802: derived assuming a variable dust emissivity. Red contours in the
803: bottom plot show the regions with extinction of $A_V = 3, 5, 7,
804: 9$. \label{MAPS10060}}
805: \end{figure}
806:
807: \begin{figure}
808: \epsscale{1.1}
809: \plotone{f7_color.eps}
810: %\plotone{f7_bw.eps}
811: \caption{({\it top}) The dust temperature map and ({\it bottom})
812: column density (expressed in terms of $V$-band extinction) derived
813: from the Spitzer 70 and 160 \micron\ flux density maps, at the
814: 5\arcmin\ resolution of our NIR-derived extinction map. The column
815: density is derived assuming a variable dust emissivity. Red contours
816: in the bottom plot show the regions with extinction of $A_V = 3, 5, 7,
817: 9$.\label{MAPS16070}}
818: \end{figure}
819:
820: \begin{figure}
821: %\epsscale{1.1}
822: \plotone{f8_color_xv.eps}
823: %\plotone{f8_bw.eps}
824: \caption{({\it top}) The dust temperature map and ({\it bottom})
825: column density (expressed in terms of $V$-band extinction) derived
826: from the Spitzer 70 and 160 \micron\ flux density maps, at the
827: 40\arcsec\ resolution of the 160 \micron\ map. The column density is
828: derived assuming a variable dust emissivity. Red contours in the
829: bottom plot show the regions with extinction of $A_V = 3, 5, 7,
830: 9$.\label{MAPS16070HIGH}}
831: \end{figure}
832:
833: \begin{figure}
834: \epsscale{1.0}
835: \plotone{f9_color.eps}
836: %\plotone{f9_bw.eps}
837: \caption{Histograms of the dust color temperature and column density
838: derived from the IRAS 60 and IRAS 100 \micron\ flux density maps ({\it
839: bottom}), Spitzer 70 and 160 \micron\ flux density maps ({\it middle})
840: and IRAS 100 and Spitzer 160 \micron\ flux density maps ({\it top}).
841: The column density is derived assuming a variable dust emissivity.
842: The dust temperature is derived assuming a variable fraction of 60 and
843: 70 \micron\ flux emitted by BG's. The filled histograms show the
844: column density and temperature derived from FIR emission, while the
845: open histograms show the column density derived from NIR absorption.
846: The vertical dashed lines show the median $A_V$ in the 2MASS map ({\it
847: left}) and the median dust temperature in the 100/160 \micron\ $T_d$
848: map ({\it right}). \label{PLOTHIST}}
849: \end{figure}
850:
851:
852: \end{document}
853: