1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{graphics}
3: \input epsf
4: \begin{document}
5:
6: \title{Dusty Acoustic Turbulence in the Nuclear Disks
7: of two LINER Galaxies NGC 4450 and NGC 4736}
8:
9: \author{Debra Meloy Elmegreen\affil{Vassar College, Dept. of Physics \& Astronomy,
10: Box 745, Poughkeepsie, NY 12604; elmegreen@vassar.edu}}
11:
12: \author{Bruce G. Elmegreen \affil{IBM Research Division, T.J. Watson
13: Research Center, P.O. Box 218, Yorktown Heights, NY 10598, USA,
14: bge@watson.ibm.com} }
15:
16: \author{Kate S. Eberwein \affil{Vassar College, Dept. of Physics \& Astronomy,
17: Box 745, Poughkeepsie, NY 12604, kaeberwein@yahoo.com } }
18:
19: \begin{abstract}
20: The structure of dust spirals in the nuclei of the SAab-type Liner
21: galaxies NGC 4450 and NGC 4736 is studied using archival HST PC images.
22: The spirals are typically only several hundredths of a magnitude fainter
23: than the neighboring disks, so unsharp mask techniques are used to
24: highlight them. The ambient extinction is estimated to be less than
25: 0.1 mag from the intensity decrements of the dust features and from the
26: spiral surface filling factor, which is about constant for all radii and
27: sizes. The nuclear dust spirals differ from main-disk spirals in several
28: respects: the nuclear spirals have no associated star formation, they are
29: very irregular with both trailing and leading components that often cross,
30: they become darker as they approach the center, they completely fill the
31: inner disks with a constant areal density, making the number of distinct
32: spirals (the azimuthal wavenumber $m$) increase linearly with radius,
33: and their number decreases with increasing arm width as a power law.
34: Fourier transform power spectra of the spirals, taken in the azimuthal
35: direction, show a power law behavior with a slope of $-5/3$ over the range
36: of frequencies where the power stands above the pixel noise. This is
37: the same slope as that found for the one-dimensional power spectra of
38: HI emission in the Large Magellanic Cloud, and also the slope expected
39: for a thin turbulent disk. All of these properties suggest that the dust
40: spirals are a manifestation of acoustic turbulence in the inner gas disks
41: of these galaxies. Such turbulence should dissipate orbital energy and
42: transfer angular momentum outward, leading to a steady accretion of gas
43: toward the nucleus. \end{abstract}
44:
45: \keywords{turbulence --- extinction --- ISM: structure --- galaxies: active
46: --- galaxies: nuclei }
47:
48: \section{Introduction}
49:
50: Dust spirals in the inner kpc regions of galaxies reveal a source of
51: compression that can affect the angular momentum distribution of the gas
52: and possibly drive accretion to an AGN. Barred galaxies with an inner
53: Lindblad resonance tend to have two long and symmetric dust spirals near
54: the resonance that are a continuation of the leading-edge dust lanes in
55: the bar (Athanassoula 1992). Many barred
56: galaxies have ILR rings too (Buta \& Crocker 1993; P\'erez-Ram\'irez et
57: al. 2000; Knapen et al. 2000). Non-barred galaxies, galaxies without
58: an ILR (e.g., late Hubble types), and regions of barred galaxies inside
59: their ILRs can have more irregular dust spirals.
60:
61: Whether the nuclear spirals are regular or irregular, their presence
62: in optical images suggests a density variation that
63: is at least a factor of 2 and therefore likely to involve shocks. These
64: shocks are oblique for azimuthal flows, so the gas will experience a
65: torque when it enters a spiral and an opposite torque when it
66: leaves. In cases where the gas moves faster than the spirals, the net
67: torque is negative and the gas loses angular momentum and energy at the
68: shock, causing it to spiral inward. If the spirals move faster than the
69: gas, as might be the case outside a fast nuclear bar, then the gas gains
70: angular momentum and may gradually move out.
71:
72: We are interested in the source of compression for the irregular nuclear
73: dust spirals that appear in non-barred galaxies or inside the ILRs of
74: some barred galaxies. We have proposed that some of these are caused by
75: random sonic noise that amplifies weakly as it propagates toward the
76: center (Elmegreen et al. 1998; Montenegro, Yuan, \& Elmegreen 1999;
77: Englmaier \& Shlosman 2000). A
78: signature of this process is an irregularity of structure, a wide range
79: of pitch angles reflecting different times of origin and different wave
80: propagation directions, a tendency to fill the volume available with a
81: spiral separation comparable to the epicyclic radius for motions at the
82: sound speed, and a general trend of increasing density with decreasing
83: radius for each spiral. Because the velocities inside these spirals
84: cannot yet be measured, there is no way to be sure they are driving
85: inflow. Nevertheless, if any of the spirals come close to
86: the nucleus, then such accretion would seem to be likely.
87:
88: Nuclear dust spirals and clouds have been studied previously using Hubble
89: Space Telescope (HST) data. Van
90: Dokkum \& Franx (1995) found them on WFPC V-band images of early
91: type galaxies. Malkan, Gorjian, \& Tam (1998) did a large
92: snapshot survey of active galaxies and classified the nuclear morphology
93: including the dust.
94: Elmegreen et al. (1998) noted dust spirals in WFPC2 images
95: of the interacting galaxy NGC 2207 and proposed they might drive
96: accretion to the nucleus.
97: Regan \& Mulchaey (1999) found nuclear dust spirals in 6 AGN galaxies
98: using WFPC2 and NICMOS images, and Martini \& Pogge (1999) found them
99: with HST data in 20 Seyferts; both studies concluded that the spirals
100: could drive accretion.
101: Ferruit, Wilson \& Mulchaey (2000) studied 12 early
102: type Seyfert galaxies in more detail and noted the dust spirals too.
103: Tomita et al. (2000) used HST archival images to
104: study dust features in E and S0 galaxies, but did not comment on spirals
105: specifically.
106: Tran et al. (2001) studied nuclear dust in elliptical galaxies.
107:
108:
109: Here we measure the extinction and structural properties of irregular dust
110: spirals in the central kpc of
111: two LINER galaxies, NGC 4450 and NGC 4736. Their Hubble types
112: are about the same, SA(s)ab and RSA(r)ab (de Vaucouleurs et al. 1991),
113: and their distances are taken to be 16.8 and 4.3 Mpc (Tully 1988).
114:
115: \section{Observations}
116:
117: Hubble Space Telescope preview images of the galaxies NGC 4450 and NGC
118: 4736 are shown in Figure 1. These images were obtained from the HST
119: archive web site and are degraded in resolution by a factor of 2 from
120: the full HST data; they are from proposals 5375 by V. Rubin and 5741
121: by J. Westphal, respectively. For our analysis of dust structure, we
122: used the original Planetary Camera (PC) images with
123: $0.0455^{\prime\prime}$ per pixel, corresponding to 3.7 pc and
124: 0.95 pc per pixel, respectively. The PC fields of view cover 3000 pc and
125: 760 pc. Multiple images in the V passband (F555W filter) were combined
126: to remove cosmic ray tracks and bad pixels.
127: Figure 1 shows the full WFPC2 image to illustrate the outer
128: disk morphologies, their position angles,
129: and the orientation of the HST images relative to north.
130:
131: \subsection{Dust Absorption}
132:
133: Intensity decrements for dust cloud absorption were measured for many
134: features using radial and azimuthal profiles on the PC images. Figure
135: 2 shows a sample profile from the V-band image of NGC 4736
136: taken along the strip shown in Figure 3
137: (which uses the G4-G12 enhanced image from Fig. 6 below). Most of the
138: intensity dips deeper than a few hundredths of a magnitude
139: are dust features, as
140: can be determined by comparing Figures 2 and 3. The
141: thinner features may be verified as dust also by comparing Figure 2
142: with the G1-G3 and G2-G6 images in Figure 6.
143:
144: Intensity decrements do
145: not measure the cloud extinction because there is likely to be a
146: smooth intensity contribution from foreground starlight. If dust
147: with opacity $\tau$ in the midplane is next to an intercloud medium
148: with opacity $\tau_0$, then the cloud to intercloud intensity ratio is
149: $\left(1+e^{-\tau}\right)/\left(1+e^{-\tau_0}\right)$. For $\tau >\tau_0$,
150: this ratio is between 0.5 and 1, making the maximum cloud contrast 0.75
151: mag. Thus the magnitude difference is not a good measure of intrinsic
152: extinction unless $\tau$ is small and $\tau_0$ is either independently
153: known or close to zero. An intensity ratio lower than 0.5 corresponds to
154: an opaque dust cloud on the near side of the galaxy.
155:
156: Radiative transfer models of the intensity distributions in several
157: passbands are necessary to determine the intrinsic opacities of these
158: clouds, but the present HST data were not acquired for this purpose and
159: the wavelength coverage and integration times are not adequate for
160: a proper analysis. Dust spirals in another, similar, nuclear region
161: were analyzed by radiative transfer
162: previously using HST data in V and I passbands (Elmegreen et
163: al. 1998). We found that typical extinctions are several magnitudes
164: inside the darkest clouds, and that the average extinction in the
165: inner disk is so low that the gas is not likely to be self-gravitating.
166: We expect the same situation here because of a lack of star formation
167: associated with the dust. Radiative transfer models of dust features
168: for main galaxy disks are more common than for nuclear disks (e.g.,
169: Howk \& Savage 1997; Kuchinski et al. 1998; Regan 2000). Galaxy
170: extinctions have also been measured using background sources
171: (e.g., Pizagno \& Rix 1998;
172: White, Keel, \& Conselice 2000; Keel \& White 2001;
173: Elmegreen et al. 2001)
174:
175: The typical absorption trough for a dust feature in Figure 2 is several
176: hundredths of a magnitude in V band. The deep feature indicated is part
177: of a long dust spiral that covers a factor of two in radius and then
178: branches into two thinner spirals in the inner part (see Fig. 3). Figure 4 shows
179: the magnitude difference for this spiral and several others in the two
180: galaxies as a function of radius. These differences were measured by
181: comparing the intensities of the dust lanes to the adjacent interdust
182: regions at several radii. Estimated measuring uncertainties are 0.002 mag.
183: Each line in Figure 4 is a different spiral; the top dashed line
184: corresponds to the dust feature shown in Figure 2. The darkness of the dust
185: generally increases inward, suggesting that either the density of the
186: spirals or their thickness relative to the stellar disk increases inward.
187:
188: \subsection{Image Enhancement, Fractal Structure, and Dust Opacity}
189:
190: The spiral dust structures are very faint. They require enhancement
191: with unsharp masks or other filtering techniques. Figures 5 and 6 show
192: composites of four filtered images for each galaxy. Each filter isolates
193: structures within a range of scales spanning a factor of 3. The top left
194: panels contain unsharp-masked images made by subtracting a smoothed
195: version from the original image, using a smoothing function that is a
196: Gaussian with a dispersion of $\sigma=3$ pixels. This image shows the
197: smallest structures, from 1 to 3 pixels. In the top right, a smoothed
198: image with a $\sigma=6$ px Gaussian was subtracted from a smoothed image
199: with a $\sigma=2$ px Gaussian (this will be called a G2-G6 image). This
200: shows features with scales between 2 px and 6 px. The bottom left and
201: right images subtract $\sigma=12$ px from $\sigma=4$ px convolutions
202: (G4-G12) and $\sigma=24$ px from $\sigma=8$ px convolutions (G8-G24),
203: respectively. Each image has considerable structure, indicating that
204: clouds span a wide range of scales.
205:
206: The fractal dimension of multiscale structure can be determined from box
207: counting or cloud counting techniques, plotting the number of structures
208: as a function of scale in log-log coordinates (Mandelbrot 1982; Westpfahl
209: et al. 1999). Because we have already isolated certain scales with
210: each filter, the cloud counting results on the subtracted images will
211: give the same slope as the derivative with respect to length of the
212: box counting results on the original images. Recall that box counting
213: includes all structures larger than the scale of the box. The fractal
214: dimension $D$ is the slope of the function $N(>S)$ versus $S$ on log-log
215: coordinates for number of structures larger than $S$ determined by the box
216: counting technique. If we write $N(>S)=\int_S^\infty n(S)dS$ for number of
217: structures $n(S)$ between $S$ and $S+dS$, then $n(S)dS\propto S^{-D-1}dS$
218: and $n(S)d\log S \propto S^{-D}d\log S$ for linear and log intervals
219: of size, respectively. Thus the slope on a log-log plot of the number
220: of clouds in each interval of filter size, for logarithmically spaced
221: intervals like we have in Figures 5 and 6, gives the fractal dimension
222: of the cloud structures. We choose the filter technique rather than the
223: box technique so we can see the nature of the structure on each scale,
224: whether it is filamentary or shell-like, for example. Figures 5 and 6
225: indicate that it is filamentary on all the resolved scales.
226:
227: Figure 7 shows the number of spirals intercepted by deprojected
228: azimuthal scans on the G2-G6 image plotted as a function of radius.
229: The radii used for the measurements are 22, 45, 67, 90, and 112 pixels.
230: A deprojected azimuthal scan of this image showing average counts for the
231: radial interval from 66 to 68 pixels is shown in Figure 8. The number of
232: clouds plotted in Figure 7 at this radius, 22 clouds, is shown again in
233: Figure 8, with each cloud indicated by an arrow. The full circle of the
234: azimuth in Figure 8 corresponds to 421 pixels. Generally the cloud count
235: can be done easily by eye on a computer screen down to the brightness
236: limit of the image. The number of dark features in an azimuthal scan
237: is always about one-third of the radius, which makes the azimuthal
238: spacing between them $\sim6\pi$ pixels. Because the average cloud size
239: on this image is $\sim4$ px, the clouds fill $4/\left(6\pi\right)=21$\%
240: of the circumference.
241:
242: This spiral filling factor makes sense given the extinction contrast.
243: If the spirals formed by the compression of a fraction $f$ of the gas
244: and the spiral spacing is $C\sim5$ times the spiral width ($C$ is the
245: inverse of the filling factor given above), then the column density of
246: each spiral is $(C-1)f+1$ times the ambient average. This column density
247: has a contribution from uncompressed material that was at the spiral
248: position already, which is the fraction $1-f$ of the ambient column
249: density, plus the compressed material from the whole volume out to the
250: next spiral, which is the amount $fC$ times the ambient column density.
251: The sum is $1-f+fC=(C-1)f+1$ times the ambient column density, as above.
252: The assumption here is that not all of the interstellar medium will be
253: compressed by the turbulence. A warm or hot component, with a thermal
254: speed larger than the mean turbulent speed, should flow around the moving
255: clouds and not compress easily. Thus $f$ may be considered the mass
256: fraction of the gas that is compressible by the prevailing turbulence.
257:
258: In this situation,
259: suppose the spiral opacity is $\tau$, the
260: interspiral opacity is $\tau_0$, and the ambient average before the
261: compression was $\tau_a$, all from gas concentrated in the midplane.
262: Then the compression gives $\tau=\left[\left(C-1\right)f+1\right]\tau_a$,
263: the conservation of mass gives $\tau+\left(C-1\right)\tau_0=C\tau_a$,
264: and the density contrast in magnitudes, $\Delta m$, gives \begin{equation}
265: {{1+e^{-\tau}}\over{1+e^{-\tau_0}}}=10^{-0.4\Delta m}
266: \end{equation}
267: These equations can be solved for the three opacities given $C$,
268: $f$, and $\Delta m$. This solution has a simple expression in the
269: limit of small $\Delta m$. In that case, there are two solutions
270: for $\tau$, $\tau_0$ and $\tau_a$, one where the opacities
271: are very large so hardly any background light gets through,
272: and another where the opacities are very small. The relatively
273: smooth continuation of the outer disk starlight into the inner
274: region implies the latter. Then in the limit of small $\Delta m$,
275: \begin{equation} \tau_a\sim 0.8\ln10\Delta m/\left(fC\right).
276: \end{equation}
277: For $C\sim5$ and $f\sim1/2$, this gives $\tau_0\sim0.7\Delta m$ on
278: average. Considering $\Delta m\sim0.04$ from Figure 4, the ambient
279: opacity of $\tau_a\sim0.03$ corresponds to a mass column density of
280: $\sim0.5$ M$_\odot$ pc$^{-2}$. This result is $\sim10$ times smaller
281: than the average extinction in the center of NGC 2207, which was about
282: 0.6 mag (Elmegreen et al. 1998), but the dust features here are ten times
283: fainter than in NGC 2207, where a prominent dust lane used for the opacity
284: measurement had a V-band intensity contrast of 0.3 mag. Evidently, the
285: column densities in NGC 4450 and NGC 4736 are too small for the inner
286: gas disks to be significantly self-gravitating; for reasonable epicyclic
287: frequency and velocity dispersion, the Toomre stability parameter $Q$
288: exceeds 100.
289:
290: We can do this same exercise for spirals of different sizes. The top
291: part of Figure 9 shows the number of dust spirals counted in different
292: azimuthal scans versus the size of the Gaussian filter, plotting the
293: G1-G3 filter combination as a point with an average Gaussian radius
294: of 2, for example. The radii used for the azimuthal scans are 45, 67,
295: 90 and 112 pixels. The number of dust spirals decreases approximately
296: inversely with Gaussian filter radius for all galactic radii, suggesting
297: again that the spirals are packed in tightly, even on a variety of scales.
298: If the spacing between spirals in the azimuthal direction is $\lambda$,
299: and this spacing is the factor $C$ times the spiral width, $W$, then the
300: number $N$ of spirals at radius $R$ is $2\pi R/\lambda$ and the inverse
301: of the filling factor is $C=2\pi R/\left(NW\right)$. This inverse
302: filling factor, $C$, is shown in the bottom of Figure 9. For all but
303: the smallest features, which are probably undercounted, $C\sim3-5$
304: for all sizes and radii. This is the same value of $C$ that was used
305: in the previous discussion of $\tau_a$. Thus the spirals are likely to
306: be made on a wide variety of scales by {\it compressions} of a factor
307: of $C\sim3-5$ from an ambient medium with an opacity slightly less than
308: 0.1 mag. We suggest below that the mean gas
309: density could be $\sim1.5$ cm$^{-3}$.
310:
311: The top part of Figure 9 may also be used to determine a fractal
312: dimension, which equals the slope according to the discussion at the
313: beginning of this subsection. The number of clouds decreases a little
314: less rapidly than the inverse of the smoothing scale $\sigma$, giving
315: an apparent fractal dimension slightly less than 1. If we write the
316: dependence of the number of counted clouds on the cloud size as a power
317: law, $\sigma^{-D}$, for logarithmic intervals of $\sigma$, then $D$ is the
318: fractal dimension. From Figure 9, $D$ equals $0.79\pm0.16$ for NGC 4450
319: and $0.51\pm0.04$ for NGC 4736, considering Gaussian radii from 4 to 32.
320:
321: \subsection{Fourier Transform Power Spectra}
322:
323: Fourier transform power spectra of the azimuthal intensity profiles
324: at ten radii in each galaxy are shown in Figure 10. The radii $R$ were
325: chosen so that $2\pi R$ equals an allowable number of pixels in a Fast
326: Fourier Transform done with the SRCFT subroutine in the IBM ESSL software
327: package. These numbers give radii $R=$ 33, 57, 81, 107, 134, 153, 183,
328: 204, 229, and 244 pixels.
329:
330: The power spectra are the sums of the squared real and imaginary parts
331: of the FFTs. These are one-dimensional FFTs, taken from the azimuthal
332: intensity traces along the deprojected images. Deprojection was made
333: with a position angle of $7^\circ$ and an inclination of $42.2^\circ$
334: for NGC 4450 on the HST image, and with a position angle and inclination
335: of $-2^\circ$ and $35.6^\circ$ for NGC 4736. The inclinations were taken
336: from the Third Reference Catalogue of Bright Galaxies (de Vaucouleurs
337: et al., 1991). The position angles were measured from the HST images
338: of the nuclear regions and differ from the position angles of the outer
339: disks. In terms of the number of degrees counter-clockwise from North,
340: the position angles of the nuclear-disk major axes are $\sim1^\circ$
341: and $\sim21^\circ$ for NGC 4450 and NGC 4736, respectively.
342:
343: The projected ellipses used for the intensity traces are shown on the
344: G4-G12 images in Figure 11. Each intensity trace is an average over 9
345: deprojected azimuthal scans spaced by one pixel and centered on one of
346: the radii $R$ given above. Each pixel used for an azimuthal average is
347: blackened in Figure 11.
348:
349: The power spectra are illustrated in Figure 10 by stacking the different
350: radii on top of each other with the smallest radius at the bottom.
351: The vertical scale corresponds to a factor of 10 in the power spectrum
352: for each tic mark; the absolute scale is arbitrary. The solid,
353: dashed, and dot-dashed lines have slopes of $-5/3$, $-1$, and $-8/3$.
354: Solid lines with a slope of $-5/3$ are drawn through the low frequency
355: portions of each power spectrum to illustrate the general trend.
356:
357: The power spectra all have slopes of about $-5/3$ at low frequency. The
358: slopes flatten at high frequency to a slope of about $-1$ or flatter.
359: The low frequency power spectrum could be steeper than $-5/3$ at the
360: smallest radius, possibly as steep as $-8/3$, as shown by the dot-dashed line.
361:
362: Turbulence gives a one-dimensional power spectrum with a slope of $-5/3$
363: for structures that are thinner on the line-of-sight than the transverse
364: wavelength, and a slope of $-8/3$ for structures that are thicker on
365: the line-of-sight than the wavelength (Lazarian \& Pogosyan 2000).
366: Two-dimensional power spectra have slopes for these two cases that
367: are steeper by one, i.e., $-8/3$ and $-11/3$. The azimuthal scan
368: results here, giving slopes of $-5/3$ at small spatial frequencies,
369: are consistent with turbulent structures that are wider than the galaxy
370: disk scale height.
371:
372: The same slope of $-5/3$ was found for power spectra of azimuthal scans
373: of the HI intensity of the Large Magellanic Cloud (Elmegreen, Kim, \&
374: Staveley-Smith 2001). Thus the origin of the structure is probably the
375: same. Power law power spectra have also been found in other interstellar
376: emission line surveys, using HI (Crovisier \& Dickey 1983; Green 1993)
377: and CO (Stutzki et al. 1998) in the Milky Way and HI in the Small Magellanic
378: Cloud (Stanimirovic et al. 1999). The slopes are always consistent with
379: turbulence as the origin of the structure.
380:
381: The LMC showed a turnover in the power spectrum at high spatial frequency
382: in both the azimuthal and linear scans, going from slopes of $-5/3$
383: to $-8/3$ at high spatial frequency, and it showed a similar turnover
384: in the two-dimensional power spectrum of the whole disk, going from
385: $-8/3$ to $-11/3$ at high frequency. These turnovers were proposed
386: to correspond to the transition from structures that are larger in the
387: transverse direction than the galaxy thickness to structures that are
388: smaller than the galaxy thickness (Elmegreen, Kim, \& Staveley-Smith
389: 2001), as predicted in the general case by Lazarian \& Pogosyan (2000).
390: Elmegreen et al. also found that LMC thickness increased with radius.
391: This result was confirmed by Padoan et al. (2001), who found a similar
392: thickness-dependent feature in the spectral correlation function of HI
393: emission from the LMC, and used this feature to map the inferred disk
394: thickness around the plane. Analogous turnovers in the power spectrum
395: are not seen for the nuclear dust spirals studied here, presumably because the
396: noise level is too high. This problem is suggested by the flattening
397: of the power spectra at intermediate to high frequencies -- before the turnover
398: can occur. Longer integration times on these sources could
399: possibly get the noise down to a level where the power spectrum
400: turns over. Then the line-of-sight disk thicknesses could be measured.
401: All we can say here is that the thickness is less than $\sim50$ pc in
402: NGC 4450 and less than $\sim10$ pc in NGC 4736. These limits are not
403: unreasonable because they can still give the inner gas disks about the
404: same aspect ratios as a typical main galaxy gas disk. When combined
405: with the average extinction estimate given in the previous subsection,
406: $A_V\sim0.03$ mag, a disk thickness of $\sim10$ pc corresponds to an
407: average gas density of $\sim 1.5$ cm$^{-3}$, which is about
408: the same as that in the Solar neighborhood.
409:
410: The morphologies of the dust structures in the nuclear regions of NGC 4450
411: and NGC 4736 differ from the morphologies of the gas emission structures
412: in the Milky Way and Magellanic Clouds surveys. In the nuclear regions,
413: the dust is mostly in the form of spirals, which show a power-law power
414: spectrum in the azimuthal direction and a gradient of intensity in the
415: radial direction (cf. Fig. 4). For the HI and CO emission line surveys,
416: the structures are more blob-like or shell-like, with essentially no
417: spirals, even in the whole LMC and SMC. Evidently, the origin of these
418: gaseous structures is the same, i.e., random and persistent compressions
419: from turbulence, but the nuclear regions of NGC 4450 and NGC 4736 have
420: a lot of shear also, and this shear distorts the compressed regions
421: into spirals. The LMC and SMC have relatively little shear, as do local
422: interstellar clouds.
423:
424: Numerical simulations of this shearing effect for gravitating media were
425: in Toomre \& Kaljnas (1991), Huber \& Pfenniger (1999), Wada \& Norman
426: (1999), and Semelin \& Combes (2000). Pressures from star formation were
427: also considered by Wada, Spaans, \& Kim (2000) and Wada \& Norman (2001).
428: The present case differs from these others slightly because the nuclear
429: dust structures here are mostly without star formation and they are
430: probably also without significant self-gravity. The origin of the
431: turbulence is unknown, although a mild, pressure-driven instability was
432: suggested by Montenegro et al. (1999).
433:
434: \section{Discussion}
435:
436: The nuclear spirals in NGC 4450 and NGC 4736 vaguely resemble the
437: outer spirals, which are somewhat flocculent in each case, but the
438: nuclear spirals do
439: not continue smoothly from the outer spirals and there are important
440: structural
441: differences. The B-band image of NGC 4450 in the Carnegie Atlas of
442: Galaxies (Sandage \& Bedke 1994) shows two long dust spirals in the main
443: disk, along with some flocculent structure; the stellar spiral arms are
444: smooth. In contrast, the nuclear region has no stellar arms and at least
445: 7 prominent dust arms, some with pitch angles as high as $45^\circ$ and some
446: crossing each other. The eastern side of the nuclear region shows more
447: dust than the western side because of the galaxy's inclination. Some small
448: dust feathers extend nearly radially from the center toward the south,
449: reminiscent of jets. HST spectral observations by Ho et al. (2000) reveal
450: double-peaked line profiles with high velocity wings, characteristic of
451: accretion disk activity observed in other LINERs.
452:
453: NGC 4736 is an early-type galaxy with an outer ring and a circumnuclear
454: starburst ring. Its main disk structure is flocculent and defined
455: primarily by the dust. Sandage \& Bedke's B-band print shows the inner
456: disk structure as composed of many tightly-wrapped arms, but the central
457: region is saturated in the reproduction. Waller et al. (2001) present
458: UIT UV and ground-based R-band images of the central regions, including
459: an unsharp-masked image showing the complicated flocculent structure of
460: the main disk. They also show an HST FOC image of the main nuclear dust
461: arms. Ground-based NIR observations by Mollenhoff, Matthias, \& Gerhard
462: (1995) suggested a weak stellar bar with a length of $20^{\prime\prime}$,
463: which was also noted by Maoz et al. (1995) from an HST FOC image.
464: The bar was observed in CO by Sakamoto et al. (1999) and Wong \& Blitz
465: (2000). In the HST image, the region corresponding to the bar shows up
466: as an elongated disk with a position angle nearly perpendicular to the
467: major axis of the galaxy. The structure inside the circumnuclear ring,
468: which is really two tightly wrapped arms, consists of a dozen dust arms
469: within a radius of 50 pc from the center, branching to dozens more dust
470: spirals out to 200 pc. The nuclear dust spirals are not attached to the
471: main inner disk dust spirals.
472:
473: The nuclear dust in NGC 4450 and NGC 4736 has several characteristics
474: that differ from spiral arms and dust clouds in main galaxy disks. These are:
475:
476: \begin{itemize}
477:
478: \item The nuclear dust spirals shown here have no associated star
479: formation. Other nuclear spirals in different galaxies have star
480: formation (e.g., Coma D15 in Caldwell, Rose, \& Dendy 1999), so the
481: gaseous nature here is not universal. The lack of star formation suggests
482: that the inner gas disks in NGC 4450 and NGC 4736 are not strongly
483: self-gravitating. The same was true for NGC 2207 (Elmegreen et al. 1998)
484: and for several other inner disks in the study by Martini \& Pogge (1999).
485: Our opacity estimate in Section 2.2 is also consistent with this.
486:
487: \item The nuclear dust is in the form of spiral arms of various pitch
488: angles, widths, and lengths. Some of the arms are trailing, a few are
489: leading, and many cross each other. This pattern is generally more
490: irregular than main disk flocculent arms (see atlas in Elmegreen 1981).
491: Main disk flocculent arms are rarely leading. They generally do not cross
492: each other; if they branch into spurs, then this branching is toward larger
493: radii (Elmegreen 1980). They also have star formation that gives
494: them a thicker, more patchy quality, rather than a filamentary quality.
495:
496: \item The nuclear dust spirals in NGC 4450 and NGC 4736 have decreasing contrast with
497: increasing radius. Ambient dust extinctions generally decrease with
498: galactocentric distance because of the exponential distribution of gas
499: column density. The flocculent galaxy NGC 5055 has such a decrease, for
500: example, as measured by the extinctions of OB associations (Acarreta et
501: al. 1996). Nuclear dust spirals are not just ambient extinctions,
502: however. They are morphologically more similar to main disk spiral arms
503: than diffuse cloud extinctions because they are organized and most
504: likely formed by compressive processes in the presence of shear. From
505: this point of view, nuclear dust spirals should be compared to main disk
506: spirals, and then the radial decrease in nuclear spiral amplitude is
507: unusual. Density wave spirals in non-barred galaxies tend to get
508: stronger with increasing radius, out to at least the corotation zone
509: (Elmegreen \& Elmegreen 1984; Elmegreen et al. 1996).
510: The unusual result that nuclear dust
511: spirals get weaker with radius is presumably the result of crowding near
512: the center for waves that move inward, as predicted for solutions to
513: Bessel's wave equation (Elmegreen et al. 1998; Montenegro et al. 1999).
514: This is a different dynamical situation than for main disk spiral
515: arms, for which the curvature terms ($\propto 1/kr$ for wavenumber $k$)
516: in the wave equations can usually be ignored (e.g., Bertin et al. 1989).
517:
518: \item The number density of dust features is about constant with radius,
519: indicating that the inner disk is completely filled with structure. This
520: is unlike the situation for main galaxy disks which often have a small
521: number of arms (e.g., 2-5) that get further apart with
522: radius. The nuclear spirals also have some indication of a hierarchical
523: or fractal structure because of a non-integer slope of the size
524: distribution, examined with unsharp masks. This size distribution is
525: approximately a power law with a slope in the range from 0.5 to 0.8.
526: This power law is reminiscent of other properties of interstellar clouds
527: formed by turbulence, but complicated in this case by the effects of
528: shear, which make spirals rather than clumps, and by the Coriolis force,
529: which resists turbulent motions on large scales. The dust features are
530: also very weak, and the smaller clouds, as well as those further from the
531: center, are difficult to see above the pixel noise. Fractal structure in
532: the dust of another galaxy was also found by Keel \& White (2001)
533: using a background illumination source. Fractal structure in galactic
534: clouds is well known (e.g., Falgarone, Phillips, \& Walker 1991).
535:
536: \item Fourier transform power spectra in the azimuthal direction show
537: the characteristic signature of turbulence compression, which is a power
538: law slope of $-5/3$ for one-dimensional structures that are larger than
539: the line-of-sight thickness of the galaxy disk. This makes the nuclear
540: dust features studied here resemble the HI clouds in the LMC, with the
541: important difference that the nuclear clouds are spiral filaments,
542: presumably affected by shear, and the LMC clouds are globular and
543: shell-like in a low shear environment.
544:
545: \end{itemize}
546:
547: There is no direct evidence in our observations for accretion driven
548: by the dust spirals. Radial velocities will have to be measured to
549: determine this. However, the increase in dust opacity with decreasing
550: radius for the main spirals is consistent with the amplification
551: that is expected for inward motions. In that case, the spirals
552: could drive nuclear accretion.
553:
554: \acknowledgments
555: D.M.E. and B.G.E. gratefully acknowledge HST archival grant HST-AR-09197.
556: We also acknowledge useful comments by the referee.
557:
558: \begin{thebibliography}{}
559:
560: \bibitem[]{} Acarreta, J.R., Manteiga, M., Pismis, P., Mampaso, A.,
561: \& Cruz-Gonzalez, G. 1996, AJ, 112, 1894
562:
563: \bibitem[]{} Athanassoula, E. 1992, MNRAS, 259, 345
564:
565: \bibitem[]{} Bertin, G., Lin, C.C., Lowe, S.A., \& Thurstan, R.P. 1989, ApJ, 338, 104
566:
567: \bibitem[]{} Buta, R.J., \& Crocker, D.A. 1993, AJ, 105, 1344
568:
569: \bibitem[]{} Caldwell, N., Rose, J.A., \& Dendy, K. 1999, AJ, 117, 140
570:
571: \bibitem[]{} de Vaucouleurs, G., de Vaucouleurs, A.,
572: Corwin, H. G., Jr., Buta, R.,
573: Paturel, G., \& Fouque, P. 1991, Third Reference Catalogue of Bright Galaxies,
574: (New York: Springer-Verlag)
575:
576: \bibitem[]{} Elmegreen, D.M. 1980, ApJ, 242, 528
577:
578: \bibitem[]{} Elmegreen, D.M. 1981, ApJS, 47, 229
579:
580: \bibitem[]{} Elmegreen, B.G., Elmegreen, D.M., Brinks, E., Yuan, C., Kaufman, M., Klaric,
581: M., Montenegro, L., Struck, C., \& Thomasson, M. 1998, ApJL, 503, L119
582:
583: \bibitem[]{} Elmegreen, D.M., \& Elmegreen, B.G. 1984, ApJS, 54, 127
584:
585: \bibitem[]{} Elmegreen, B.G., Elmegreen, D.M.,
586: Chromey, F.R., Hasselbacher, D.A., \& Bissell, B.A.
587: 1996, AJ, 111, 2233
588:
589: \bibitem[]{} Elmegreen, D.M., Kaufman, M., Elmegreen, B.G., Brinks, E.,
590: Struck, C., Klaric, M., \& Thomasson, M. 2001, AJ, 121, 182
591:
592: \bibitem[]{} Englmaier, P., \& Shlosman, I. 2000, ApJ, 528, 677
593:
594: \bibitem[]{} Falgarone, E., Phillips, T. G., \& Walker,
595: C. K. 1991, ApJ, 378, 186
596:
597: \bibitem[]{} Ferruit, P., Wilson, A. S., \& Mulchaey, J. 2000, ApJS, 128, 139
598:
599: \bibitem[]{} Ho, L.C., Filippenko, A.V., \& Sargent, W.L.W. 1997, ApJ, 487, 591
600:
601: \bibitem[]{} Howk, J.C., \& Savage, B.D. 1997, AJ, 114, 2463
602:
603: \bibitem[]{} Huber, D., \& Pfenniger, D. 1999, in The Evolution of
604: Galaxies on Cosmological Timescale, eds. J.E. Beckman \& T.J. Mahoney,
605: Astrophysics and Space Science, poster paper
606:
607: \bibitem[]{} Keel, W.C., \& White, R.E., III 2001, AJ, 121, 1442
608:
609: \bibitem[]{} Knapen, J., Shlosman, I., Heller, C. H., Rand, R.,
610: Beckman, J., \& Rozas, M. 2000, ApJ, 528, 219
611:
612: \bibitem[]{} Kuchinski, L. E., Terndrup, D. M., Gordon, K. D.,
613: \& Witt, A. 1998, AJ, 115, 1438
614:
615: \bibitem[]{} Lazarian, A., \& Pogosyan, D. 2000, ApJ, 537, 720
616:
617: \bibitem[]{} Mandelbrot, B.B. 1982, The Fractal Geometry of Nature,
618: (San Francisco: Freeman)
619:
620: \bibitem[]{} Maoz, D., Filippenko, A.V., Ho, L.C., Rix, H.-W.,
621: Bahcall, J.N., Schneider, D.P., \& Macchetto, F.D. 1995, ApJ, 440, 91
622:
623: \bibitem[]{} Malkan, M.A., Gorjian, V., \& Tam, R. 1998, ApJS, 117, 25
624:
625: \bibitem[]{} Martini, P., \& Pogge, R.W. 1999, AJ, 118, 2646
626:
627: \bibitem[]{} Mollenhoff, C., Matthias, M., \& Gerhard, O.E. 1995, A\&A, 301, 359
628:
629: \bibitem[]{} Montenegro, L., Yuan, C., \& Elmegreen, B.G. 1999, ApJ, 520, 592
630:
631: \bibitem[]{} P\'erez-Ram\'irez, D., Knapen, J., Peletier, R., Laine,
632: S., Doyon, R., \& Nadeau, D. 2000, MNRAS, 317, 234
633:
634: \bibitem[]{} Pizagno, J., Rix, H.-W. 1998, AJ, 116, 2191
635:
636: \bibitem[]{} Regan, M.W. 2000, ApJ, 541, 142
637:
638: \bibitem[]{} Regan, M. W., \& Mulchaey, J. S. 1999, AJ, 117, 2676
639:
640: \bibitem[]{} Sakamoto, K., Okumua, S.K., Ishizuki, S., \& Scoville, N.Z.
641: 1999, ApJ Suppl, 124, 403
642:
643: \bibitem[]{} Sandage, A., \& Bedke, J. 1988, Atlas of Galaxies,
644: U.S. Govt. Printing Office. Washington, D.C., NASA
645:
646: \bibitem[]{} Tomita, A., Aoki, K., Watanabe, M., Takata, T., \&
647: Ichikawa, S. 2000, AJ, 120, 123
648:
649: \bibitem[]{} Toomre, A., \& Kalnajs, A.J. 1991, in Dynamics of Disk
650: Galaxies, ed. B. Sundelius, University of Chalmers, p. 341
651:
652: \bibitem[]{} Tran, H.D., Tsvetanov, Z., Ford, H.C., Davies, J.,
653: Jaffe, W., van den Bosch, F.C., \& Rest, A. 2001, AJ, 121, 2928
654:
655: \bibitem[]{} Tully, R.B. 1988, Nearby
656: Galaxies Catalog, Cambridge: Cambridge University Press.
657:
658: \bibitem[]{} van Dokkum, P.G., \& Franx, M. 1995, AJ, 110, 2027
659:
660: \bibitem[]{} Wada, K., \& Norman, C. A. 1999, ApJ, 516, L13
661:
662: \bibitem[]{} Wada, K., \& Norman, C. A. 2001, ApJ, 547, 172
663:
664: \bibitem[]{} Wada, K., Spaans, M., \& Kim, S. 2000, ApJ, 540, 797
665:
666: \bibitem[]{} Waller, W. H., Fanelli, M.N., Keel, W.C., Bohlin, R.,
667: Collins, N.R., Madore, B.F., Marcum, P.M., Neff, S.G.,
668: O'Connell, R.W., Offenberg, J.D., Roberts, M.S., Smith, A.M.,
669: \& Stecher, T.P. 2001, AJ, 121, 1395
670:
671: \bibitem[]{} Westpfahl, D.J., Coleman, P.H., Alexander, J.,
672: \& Tongue, T. 1999, AJ, 117, 868
673:
674: \bibitem[]{} White, R.E., III, Keel, William C., \& Conselice, C.J.
675: 2000, ApJ, 542, 761
676:
677: \bibitem[]{} Wong, T., \& Blitz, L. 2000, ApJ, 540, 771
678:
679:
680: \end{thebibliography}
681:
682: \newpage
683:
684:
685: \begin{figure}
686: %\plottwo{f1left.eps}{f1right.eps}
687: \caption{NGC 4450 (left) and NGC 4736 (right)
688: V-band images from HST WFPC2.
689: (See jpeg files in astro-ph.) }
690: \end{figure}
691:
692: \clearpage
693:
694: \begin{figure}
695: \plotone{f2.eps}
696: \caption{An intensity trace across a dust spiral in NGC 4736 showing
697: an absorption dip, which is labeled,
698: and other dust features.
699: The maximum of intensity in the center of the scan is from the exponential disk.}
700: \end{figure}
701: \clearpage
702:
703: \begin{figure}
704: %\plotone{f3.eps}
705: \caption{The 600-pixel long
706: scan used for Fig. 2 is shown superposed on the G4-G12 image from
707: Fig. 6. The dust is dark in this image.
708: (See jpeg file in astro-ph.) }
709: \end{figure}
710:
711: \begin{figure}
712: \plotone{f4.eps}
713: \caption{The magnitude differences between the dust features and
714: the neighboring interdust regions are plotted versus the
715: distance from the galaxy centers for four dust spirals in NGC 4450
716: and two dust spirals in NGC 4736. The dust spirals tend to get darker toward
717: the center. }
718: \end{figure}
719:
720:
721: \clearpage
722:
723: \begin{figure}
724: %\plotone{f5.eps}
725: \caption{Unsharp mask images of NGC 4450, made by subtracting
726: two Gaussian smoothed images, Gn-Gm, for Gaussian dispersions n and m,
727: measured in pixels. The top left
728: panel shows G1-G3; top right: G2-G6; bottom left: G4-G12, and bottom right:
729: G8-G24. The number of pixels along the horizontal edge of each
730: image is 725, corresponding to 2700 pc.
731: (See jpeg file in astro-ph.) }
732: \end{figure}
733:
734: \clearpage
735:
736: \begin{figure}
737: %\plotone{f6.eps}
738: \caption{Unsharp mask images of NGC 4736, as in Fig. 5.
739: The number of pixels along the horizontal edge of each
740: image is 590, corresponding to 560 pc. (See jpeg file in astro-ph.)}
741: \end{figure}
742:
743: \clearpage
744:
745: \begin{figure}
746: \plotone{f7.eps}
747: \caption{The numbers of distinct dust spirals in azimuthal scans
748: on the G2-G6 images are
749: plotted versus the radii. The numbers equal approximately
750: one-third the radii, indicating that the filling factor for the
751: dust is about 0.2 on the scale of $\sim4$ pixels.}
752: \end{figure}
753:
754: \clearpage
755:
756: \begin{figure}
757: \plotone{f8.eps}
758: \caption{A deprojected azimuthal scan at a
759: radius of 67 pixels for the G2-G6 image of
760: NGC 4836 showing dust features as negative counts. The
761: 22 features counted for Figure 7 are indicated.
762: The full azimuth of $360^\circ$ plotted here corresponds to 421 pixels, so
763: the average spacing between these clouds is 421/22=19 pixels,
764: which is close to the value of $6\pi$ discussed in the text.
765: }
766: \end{figure}
767:
768: \clearpage
769:
770: \begin{figure}
771: \plotone{f9.eps}
772: \caption{(top) The number of distinct dust spirals in azimuthal scans
773: of different radii is shown as a function of the Gaussian dispersion
774: of the unsharp mask, which is the characteristic feature size.
775: The radii for the scans are 45, 67, 90, and 112 pixels.
776: The number of spirals decreases approximately as a power law over
777: a factor of ten in size. The slope of this power law is
778: the fractal dimension.
779: (bottom) The ratio of the azimuthal spacing between
780: the spirals to the spiral width given by the average Gaussian smoothing radius.
781: This ratio is the inverse of the areal filling factor of the spirals,
782: and it is about constant, equal to $\sim4$ for all radii and
783: spiral arm widths larger than $\sim4$ pixels. }
784: \end{figure}
785:
786: \begin{figure}
787: \plotone{f10.eps}
788: \caption{Fourier Transform power spectra of total intensity along
789: the azimuthal scans shown in
790: Fig. 10. The scans are stacked in order of
791: increasing radius in the galaxy. The solid lines drawn through each scan
792: have a slope of $-5/3$, the dashed line
793: at the bottom and top has a slope of $-1$, and the dot-dashed line at the
794: bottom has a slope of $-8/3$. The similarity between the one-dimensional power spectra of
795: dust spirals in these galaxies, the 1D power spectrum of HI emission on large scales
796: in the
797: Large Magellanic Cloud, and the 1D power spectrum of two-dimensional turbulence
798: suggests that the spirals result from turbulent compression in a relatively thin disk.}
799: \end{figure}
800:
801: \begin{figure}
802: %\plottwo{f11left.eps}{f11right.eps}
803: \caption{The projected azimuthal scans used for the power spectra are shown superposed on the
804: G4-G12 images.
805: (See jpeg files in astro-ph.) }
806: \end{figure}
807:
808:
809: \end{document}
810: