nlin0204007/ms.tex
1: \documentstyle[./aasms4,psfig]{article}  
2: %\documentclass{aasms4}  
3: \begin{document}  
4:   
5:   
6:   
7: \title{ROSAT/ASCA OBSERVATIONS OF THE MIXED-MORPHOLOGY SUPERNOVA REMNANT W28}  
8: \author{Jeonghee Rho}  
9: \affil{  
10: SIRTF Science Center, California Institute of Technology, MS 220-6,  
11: Pasadena, CA 91125,   
12: e-mail:rho@ipac.caltech.edu}  
13: \centerline{AND}   
14: \author{Kazimierz J. Borkowski}  
15: \affil{Department of Physics, North Carolina State University, Raleigh,  
16: NC 27695}   
17: 
18: \received{November 2001}
19: \accepted{March 2002}
20: \centerline{To appear in the Astrophysical Journal} 
21: 
22:    
23: 
24: %\centerline{\bf July version 01}  
25: %\centerline{\bf (printed \today)}  
26: %\maketitle  
27: %  
28: \begin{abstract}  
29:   
30: We present three sets of ROSAT PSPC and four sets of ASCA observations  
31: of the supernova remnant (SNR) W28.  The overall shape of X-ray emission in  
32: W28  is elliptical, dominated by a centrally-concentrated interior   
33: emission, sharply peaked at the center. There are also partial northeastern    
34: and southwestern shells, and both central and shell X-ray emission   
35: is highly patchy.   
36:   
37: The ASCA spectra reveal emission lines of  Ne, Mg, Si, and  Fe K$\alpha$  
38: and continuum extending at least up to 7 keV, showing that X-ray emission in   
39: W28 is mostly of thermal origin with a hot thermal component.   
40: We found that spectral variations are present in W28.  
41: The southwestern shell can be fit  
42: well by a plane-shock model with a temperature of 1.5 keV and an ionization  
43: timescale of $1.5 \times 10^{11}$ cm$^{-3}$ s.   
44: The long ionization timescale combined with a low estimated electron density   
45: of $\sim 0.2$ cm$^{-3}$ implies SNR age of several $\times 10^4$ yr.  
46: The low density in the southwest is consistent with the shock breakout   
47: away from molecular clouds in the north and northeast.  
48: The northeastern shell, with a lower temperature of $0.56$ keV and a longer   
49: ionization timescale of $1.7 \times 10^{13}$ cm$^{-3}$ s,   
50: spatially coincides with the radiative shell delineated by radio and optical   
51: filaments. But a relatively high temperature   
52: and a low density of X-ray emitting gas in the northeastern shell indicate  
53: that we are not observing gas cooling from high temperatures.  
54: Unlike for the southwestern and northeastern shells, the central  
55: emission cannot be fit well by a single temperature model, but two  
56: components with temperatures of  0.6 keV and 1.8 keV are required. The  
57: long ionization timescales imply that the  gas is close to the  
58: ionization  equilibrium.   
59: The low temperature component is  
60: similar to those seen in other Mixed-morphology SNRs.  
61: The X-ray luminosity of W28 is $\sim 6 \times 10^{34}$ ergs s$^{-1}$,  
62: and the estimated X-ray mass is only $\sim 20 - 25 M_\odot$.  
63: A comparison of W28 with other typical Mixed-morphology SNRs  
64: reveals significant differences in its X-ray properties; W28 has  
65: a higher temperature and noticeable spectral variations.   
66:   
67: 
68:   
69: W28 belongs to a class of SNRs considered  
70: by Chevalier (1999), with a radiative shell interacting with clumpy  
71: molecular  clouds. X-ray emission at its center is   
72: a ``fossil'' radiation  
73: from gas which was shocked early in the evolution of the remnant, and its  
74: centrally-peaked morphology could have been caused   
75: by processes such as evaporation, electron thermal  
76: conduction, and mixing induced by various hydrodynamical instabilities.  
77: But W28 poses a challenge for existing models of X-ray emission,  
78: because the evaporation model   
79: of White \& Long (1991) is in conflict with observations, while the presence of  
80: temperature variations seems inconsistent with SNR models with efficient   
81: thermal conduction.   
82:   
83: \end{abstract}  
84: \keywords{ISM: individual (W28) -- supernova remnants -- X-rays: ISM}  
85:   
86:   
87:   
88:   
89: \section{Introduction}  
90:   
91: Supernova remnants (SNRs) can be classified into three broad categories  
92: based on their X-ray and radio morphologies. One class is those in  
93: which both the X-ray and radio morphologies are shell-like; most of  
94: shell-like SNRs have thermal   
95: X-rays, from a plasma heated  by a shock wave  
96: (such as in Cas A). Some shell-like SNRs such as SN 1006   
97: emit nonthermal synchrotron X-rays due to radiation from shock-accelerated   
98: electrons. The second  
99: class consists of the Crab-like remnants, also called plerions.  The  
100: X-ray and radio morphologies of these are center-filled. Their energy  
101: source is an active pulsar (Seward 1985),  and the X-ray spectrum is a  
102: featureless power law due to synchrotron processes. The third  
103: morphological category is a composite, %of the former two, which shows  
104: simultaneously central emission and a shell (such as Vela and CTB 80).  
105: Recently we have classified a group of SNRs, which have centrally  
106: peaked X-ray and shell-like radio morphology, as a new, distinct  
107: morphological class which we term  ``mixed-morphology'' (MM) supernova  
108: remnants (Rho \& Petre 1998) or ``thermal composite'' SNRs (Jones et al. 1998).   
109: Members of this class  
110: include 3C391, W28, and W44. Despite their X-ray morphological similarity to  
111: the plerions all X-ray spectral evidence now clearly show that the  
112: X-ray emission is primarily thermal (Rho et al. 1994; Rho 1995;  
113: Rho \& Petre 1998). ASCA observations revealed  
114: the presence of strong thermal line emission in many of these  
115: remnants: W44 (Harrus et al. 1997; Rho et al. 1994),  
116: Kes 27 (Rho et al. 1998), %MSH 11-61A (Rho 1995),  and   
117: 3C391
118: (Chen \& Slane 2001).  The population of these  
119: mixed morphology SNRs is at least 10\% of all known SNRs, and more than  
120: 25\% among X-ray detected SNRs, suggesting that the mechanism for  
121: creating a remnant with the centrally-peaked thermal emission is not rare   
122: (Rho \& Petre 1998).  
123:   
124: The most important physical property of these remnants distinctly  
125: different from others is that their temperature profiles are uniform,  
126: rather than radially dependent. A radial temperature gradient is  
127: generally expected for   
128: a remnant of an SN explosion in a uniform  
129: ambient medium, such as in the classical Sedov solution.  
130: The uniform profile was first detected through  
131: spatially-resolved PSPC spectroscopy of W44 (Rho et al.  
132: 1994), and IC 443 (Asaoka \& Aschenbach 1994; Rho  
133: et al. 1995).  
134: The spectral hardness maps of MSH 11-61A, CTB 1, W63, HB 3 (Rho  
135: 1995) and 3C400.2 (Saken et al. 1995) are smooth, also indicating  
136: uniform temperature profiles.  
137:   
138: There are two competing scenarios which have been proposed to explain  
139: the centrally-filled morphology of mixed-morphology (MM) remnants:  
140: fossil radiation from the interior of a remnant in the  
141: shell-formation (radiative phase) stage (Cox et al. 1999; Chevalier 1999), and evaporation of interior   
142: clouds (White \& Long 1991). The first hypothesis is {\it radiative model}   
143: in which the shell of an expanding  
144: SNR cools below $\sim 10^6$ K, and its X-ray emission becomes undetectable  
145: because of low temperatures and the ISM absorption. But X-rays from the  
146: hotter interior are still detectable as  ``fossil'' thermal radiation  
147: (Seward 1985).   
148: Although dense environments of MM SNRs could hasten the onset of the   
149: shell-formation   
150: evolutionary stage, their uniform temperature profiles still stand in  
151: contrast to the radial temperature gradient expected from ``fossil  
152: radiation.'' Recent work by Cox et al. (1999)  
153: and  Shelton et al. (1999), however, suggests  
154: that thermal conduction significantly reduces the radial thermal  
155: gradient and thus it could explain the centrally-condensed X-ray  
156: morphology and the uniform temperature profiles observed  
157: in the MM SNRs.  
158:   
159: The other hypothesis is the {\it evaporation model}, in which   
160: the enhanced interior X-ray emission could arise from gas evaporated  
161: from clouds.  A model based on cloud evaporation (McKee 1982; White \&  
162: Long 1991) reproduces some of properties of a number of MM remnants  
163: observed by Rho \& Petre (1998).  
164: This model assumes SNR shock propagation through a cloudy ISM.  Clouds  
165: which are too small to affect the blast wave propagation, but of  
166: sufficiently high density to survive passage through a strong shock,  
167: subsequently provide a reservoir of material inside the remnant  
168: cavity.   
169:   
170: A correlation  between molecular interacting SNRs  
171: and X-ray center-filled SNRs   
172: was suggested by Rho \& Petre (1998).  
173: Recently Chevalier (1999) presented a framework for the SNR evolution  
174: in molecular clouds, where the radiative shell is interacting with   
175: clumpy molecular clouds. In his framework X-ray emission is a ``fossil''   
176: radiation from the SNR interior, although he admits that the observed   
177: X-ray morphologies of MM SNRs might pose the greatest challenge   
178: for his models. Just like in the conduction models of Cox et al. (1999) and   
179: Shelton et al. (1999),  
180: thermal conduction might be crucial for explaining uniform temperature  
181: profiles in interiors of these remnants.  
182:   
183: The supernova remnant W28 (G6.4-0.1) is an archetype  
184: MM SNR, showing center-filled X-rays and  
185: a shell-like radio morphology (Long et al. 1991).   
186: There are double radio shells  in the northern part of the SNR, and the  
187: northeastern and eastern shells are apparent;  the magnetic field lines  
188: are tangential over the northern shell (Frail, Kassim, \& Weiler  
189: 1994; Shaver \& Goss 1970;  Kundu \& Velusamy 1972).  The radio  
190: spectral index is $\alpha \sim 0.4$ (Kassim 1992), where $S \propto$  
191: $\nu^{-\alpha}$. The {\it Einstein} IPC observation revealed that the X-ray   
192: emission is thermal with a temperature of $2.3 \times 10^7$ K (2 keV)  
193: and an intervening column density $N_H \sim 2 \times 10^{21}$  
194: cm$^{-2}$ (Long et al. 1991). Analysis of EXOSAT data also led to a high   
195: temperature of $2.5 \times 10^7$ K and  
196: a strong Fe K$\alpha$ line at 6.7 keV was detected (Jones 1991).   
197: Torii et al.\ (1996) from their preliminary analysis of ASCA data  
198: reported that a combination of thermal and nonthermal models  
199: provided a good fit to X-ray spectra of W28.  
200:   
201: Diffuse optical nebulosity was detected throughout most of the  
202: interior of W28 (van den Bergh 1978). W28 lies in a complicated region  
203: of the inner Galaxy which is  confused by several large \ion{H}{2} regions.  
204: %and the northern boundary is close to the Galactic center.   
205: Wootten (1981) suggested an interaction with a  
206: molecular cloud in W28, based on observations of broad CO (1-0) lines with  
207: FWHM of 11 km s$^{-1}$ and the presence of warm dense clouds. A recent  
208: convincing detection of even broader lines (as large as 70 km s$^{-1}$) was  
209: reported by Arikawa et al. (1999) and Frail \&\ Mitchell (1998) using   
210: CO (3-2) line. Arikawa et al. mapped shocked gas at the northern  
211: boundary, finding about 2000 $M_{\odot}$ of shocked molecular gas.  
212: Twenty-six shock-excited OH masers which are signposts of molecular   
213: interactions were also   
214: detected along the northern boundary and along the eastern   
215: shell of the remnant (Frail, Goss, \& Slysh 1994; Claussen et al. 1999).   
216: In addition, the  
217: infrared H$_2$ S(3) and S(9) lines were seen with ISO (Reach \& Rho 2000),  
218: together with strong atomic fine-structure infrared lines (such  
219: as C, N, O, Si, P and Fe), with the [\,O I\,] line ratios implying   
220: a high ($> 10^3$ cm$^{-3}$) medium density. All these observations  
221: provide an unambiguous evidence for interaction with molecular clouds in W28.  
222: The presence of molecular clouds visibly affects the
223: overall shape of the remnant, which is far from uniform and circular, typical
224: for SNRs; it is instead dented in the north and east where molecular clouds are
225: located.
226: 
227:   
228: W28 has long been considered as a highly evolved remnant (e. g., see   
229: Lozinskaya 1992).   
230: In the north, gas behind the blast wave propagating through the interclump   
231: medium of molecular clouds collapsed into a radiative shell, which can  
232: be observed as filamentary emission at optical and radio wavelengths.    
233: Optical spectra typical of radiative shocks, the tangential orientation of the   
234: magnetic field (Milne \& Dickel 1975), and  
235: an excellent correspondence between optical and radio emission (Dubner   
236: et al. 2000) provide irrefutable evidence that the northern half of W28 is   
237: in the late radiative stage of the SNR evolution. The weakness of the observed  
238: X-ray emission at the position of the northern radio shells adds further  
239: support for this conclusion. The blast wave most likely  
240: broke out of the molecular cloud in the south (Dubner et al. 2000).  
241:   
242: Two candidates for a pulsar associated with W28 have been suggested  
243: (Andrews et al. 1983; Kaspi et al. 1993).   
244: A glitching pulsar PSR  
245: B1758-23 was detected at the northern boundary of W28 (Kaspi et  
246: al. 1993; Frail, Kassim, \& Weiler 1994).  
247: This remnant might also be the  
248: gamma-ray source 2CG 006-00 (Ormes 1988; Esposito et al., 1996), and it might harbor a recently   
249: discovered soft gamma ray repeater, SGR 1801-23 (Cline et al. 2000). 
250: 
251: The distance to W28 is uncertain, between 1.6--3 kpc (Clark \& Caswell  
252: 1976; Milne 1979; Frail, Kassim, \& Weiler 1994).   The $\Sigma$-D relation  
253: and optical measurements suggest 1.8 kpc (Goudis 1976) and 2  
254: kpc (Long et al. 1991), respectively. Based on the assumed association of   
255: PSR 1758-23 with W28 (Kaspi et al. 1993), a distance of 3 kpc was suggested  
256: using an H I absorption measurement toward this pulsar (Frail, Kulkarni, \&  
257: Vasisht 1993).   However, this association was not confirmed   
258: by VLBA measurements of interstellar scattering toward the pulsar   
259: (Claussen et al. 1999).  Recent CO observations of the shock interaction  
260: with the molecular cloud imply a distance of 1.8 kpc, adopting the rotation   
261: curve of   
262: Clemens (1985) for a galactocentric distance of the Sun of 8.5 kpc.  
263: Because the evidence for the interaction with molecular clouds is very   
264: convincing, we adopt 1.8 kpc as the distance to W28.  
265:   
266:   
267: In this paper, we present ROSAT and ASCA images and  
268: spectral mapping of W28 using three sets of ROSAT and  
269: four sets of ASCA data, and we  
270: compare them to  
271: other MM SNRs.   
272: We interpret the results in the framework of an SNR evolution in   
273: molecular clouds.   
274: We also examine other nearby X-ray sources in the field of view and   
275: search for X-ray emission from the pulsar PSR B1758-23.  
276:   
277:   
278: \section{Observations}  
279:   
280: W28 was observed with the ROSAT Position Sensitive  
281: Proportional Counter (PSPC). The count rate is 5.78$\pm0.11$ cts s$^{-1}$,  
282: almost twice as much as the IPC count rate of 3.2 cts s$^{-1}$ (Seward 1990).  
283: A total of three pointed observations were performed, with  
284: the observation parameters listed in Table \ref{w28obser}.   
285: The total  
286: exposure time is 23,800 s. The 45$^{\prime}$ remnant size extends  
287: beyond the central circular field of view   of the PSPC, where the PSPC response  
288: is most uniform and sensitive. Therefore, an exposure correction was  
289: necessary in order to obtain the mosaiced image. We used  
290: the analysis techniques for extended objects developed by Snowden et  
291: al. (1994).   
292: The scattered solar X-rays,  the time-dependent flux of short- and  
293: long-term enhancements, and background contamination were evaluated as  
294: three parameters of the total number of contaminating counts in each  
295: band, the gradient of the counts across the image, and the rotation  
296: angle of that gradient, and they were subtracted. Residual  
297: afterpulse events were also removed.  The vignetting  and  
298: detector-artifact corrected exposure maps, and modeled particle  
299: background maps (including the soft background using a background model)  
300: were created and applied to the data.  After correction of three images   
301: for background and  
302: exposure time, the images were merged together,  and smoothed using   
303: adaptive-filtering algorithm. 
304:   
305:   
306: We used archival ASCA data in order to investigate  
307: spectral properties of W28 at moderate  ($E/\Delta E = 60$ at 6 keV) spectral  
308: resolution. ASCA (Tanaka, Inoue, \& Holt 1994) had two  
309: detector pairs: Gas Imaging Spectrometers (GIS2 and GIS3) and  
310: Solid-state Imaging Spectrometers (SIS0 and SIS1).  The SIS covers an  
311: energy band of 0.5--10 keV and the GIS 0.6--10 keV.  The on-axis  
312: angular resolution of the GIS and SIS is about 1--2 arcminutes. Four  
313: sets of ASCA data pointed toward the northeast, center (two sets of data) and   
314: southwestern portion of the remnant were used, and the observational  
315: parameters are summarized in Table~\ref{w28obser}.  
316: We filtered the ASCA data using several criteria such as Cut off Rigidity  
317: (COR) and earth elevation (based on revision 2 processing).     
318:   
319: We generated two ASCA mosaiced images in 0.6--10.0 keV and  
320: hard (3--10 keV) energy bands.    
321: We used 4 sets of  GIS2 and GIS3 images  
322: with exposure correction and background subtraction (the GIS fields of  
323: view are marked in Figure \ref{w28ascafov}).   
324: First, the instrument maps were generated using FTOOLS task ``ASCAEFFMAP",  
325: and the  
326: effective exposure maps were generated by combining the instrument maps  
327: with exposure times (``ASCAEXPO"). To subtract the background, we   
328: used long-exposure blank sky observations from which  
329: all point-like sources brighter than $2 \times 10^{-13}$ ergs s$^{-1}$  
330: cm$^{-2}$ in the 2--10 keV range were removed. These files were  
331: generated using high latitude observations taken between June 1993 and  
332: December 1995 by the ASCA team (Ikebe et al. 1998). The outer detector edges   
333: beyond 17$'$ radius were removed.    
334: Then we mosaiced 4 sets of GIS images using ``FMOSAIC".   
335:   
336: 
337:   
338: \section{Results}  
339:   
340: \subsection{Images and Morphology}  
341:   
342:   
343: The mosaiced ROSAT PSPC image is shown in Figure~\ref{w28pspc}.   
344: The coverage of the GIS ASCA observations is marked on the  
345: ROSAT PSPC image in Figure \ref{w28ascafov}. The mosaiced ASCA image is  
346: shown in Figure \ref{w28ascaall}, with the ROSAT contours superposed.  
347: The ASCA image is consistent with the ROSAT image, considering that  
348: its spatial resolution is much  
349: lower, only about 1\arcmin, and that it also includes photons with energies  
350: higher than 2 keV which could not be detected by ROSAT.  
351: The global shape is elliptical; X-ray emission  
352: is concentrated at the center and  fills  the entire interior of the  
353: remnant. The PSPC image reveals that the central  
354: emission is  sharply peaked and highly  
355: patchy.   
356: Limb-brightened partial shell  
357: structures are present in the  northeast and southwest, and the  
358: northeastern shell is also partially  broken, with patchy structures.   
359: There are a few places  showing highly structured X-ray  
360: emission: there are ear-like segments of limb-brightened northern shell  
361: and a sudden jump (northern flat boundary in Fig.~\ref{w28regionmark}) in surface brightness in the north.  
362: The X-ray surface  
363: brightness  between the two ear-like patches (left patch is the northeastern shell  
364: and right patch is shown in Fig.~\ref{w28regionmark}) stays very flat from the  
365: east to west along Dec -23$^{\circ}$ 15$'$.  The PSPC image is  
366: superposed on contours of a 328 MHz radio continuum image (Dubner et al. 2000)  
367: in Figure \ref{w28xrayradio}. The X-ray surface brightness is  
368: anticorrelated with radio morphology not only on  large scales, but  
369: also on small scales: the eastern prominent radio shell is located outside  
370: the X-ray emission and the northern thin inner radio shell is immediately  
371: outside the northern flat boundary of X-ray emission. In the north,  
372: double radio shells are present, which is probably due to projection,  
373:  and in the southwest both X-ray and radio  
374: emission are faint but show a partial shell. The only places where the radio  
375: and X-ray emission match well are the two ear-like structures, reinforcing  
376: the suggestion that the emission here is from the shell.       The  
377: northeastern shell matches well with the partial radio shell  as shown in the  
378: radio superposed image.   Faint emission is found   in the southwest,  
379: apart from the main body of the remnant.  This part of the SNR is seen  
380: faintly in X-ray and radio  but not at optical wavelengths.  This suggests that  
381: it is either a breakout region of the remnant, or that extinction is higher here.  
382: The X-ray emission may extend beyond the radio boundary in  
383: the west.  
384:   
385:   
386: In the PSPC soft band ($E < 0.5$ keV) most of the emission arises  from  
387: the central part of the remnant and a bright patch in the  northeastern  
388: shell, as shown in Figure~\ref{w28soft}.   
389: We created  
390: a PSPC hardness map between energy band of 0.9--2.2 keV  and  
391: 0.5--0.9 keV (i. e., bands 5--7 and bands 2--4  as defined by Snowden et  
392: al. 1991).  We chose the energy boundary of 0.9 keV because  
393: the spectral peak of the PSPC spectrum appears at this energy (see   
394: Figure~\ref{w28spec}). A spectral hardness ratio (SHR) map, defined here as the ratio of the  
395: high energy band map to the low energy band map, can reveal  
396: absorption and/or temperature variations.  A high value of SHR  
397: indicates a locally high column density or a high temperature.   
398: The hardness ratio map is superposed on the PSPC surface  
399: brightness contours in Figure~\ref{w28shr}.    
400: This hardness map shows that the emission  
401: is harder in the southwest and south than in the north and  
402: northeast. This is consistent with our  spectral results   
403: that the southwestern shell has a higher temperature than  
404: that of northeastern shell, and possibly elsewhere in the remnant    
405: (see \S~3.2).  
406: We generated an ASCA hard (3--10 keV)  
407: image as shown in Figure \ref{w28ascahard};  
408: the southwestern part is brighter  
409: in hard X-rays relative to the total energy ASCA map. The central  
410: bright peak is also bright at hard X-rays, and the hard  
411: energy maps give hint of  
412: hard emission at the patches of both ear structures and the northern  
413: boundary.  A hard source appears in the lower right corner of the image in Figure \ref{w28ascahard},
414: which position coincides with two IRAS sources and an A star (HD 313524).
415:   
416:   
417:   
418: In order to show clearly differences in the radio and X-ray morphologies,  
419: we compare X-ray and radio surface brightness profiles in Figure~\ref{w28sb}.   
420: We chose  
421: the strong X-ray peak ($\alpha$ = 17$^{\rm h}$ 57$^{\rm m}$ 23.2$^{\rm s}$,  
422: $\delta$ = -23$^{\circ}$23$^{\prime}$50.6$''$, B1950) % \ref{w28sb})  
423: as the center for this analysis.   
424: We divided the remnant into 5 sectors based on their  
425: different appearance in radio and X-ray images: NW, covering position angles   
426: between  
427: 285$^{\circ}$ - 360$^{\circ}$ (measured east of north);  
428: N, covering P.A. 0$^{\circ}$ to 45$^{\circ}$ and 250$^{\circ}$ to   
429: 360$^{\circ}$;   
430: SW,   
431: 185$^{\circ}$ to 250$^{\circ}$; SE, 85$^{\circ}$ to 160$^{\circ}$;  
432: and northeast, 65$^{\circ}$ to 70$^{\circ}$.  
433: All X-ray profiles peak at the center, while the  
434: radio profiles show a prominent shell in most directions, but weakest at  
435: the south and southwest. A noticeable feature of the X-ray surface  
436: brightness is that it shows a shell structure in the northeast  
437: which coincides with the radio shell at  
438: 20\arcmin\ distance from the center (see Figure~\ref{w28sb}). %\ref{w28pspc}).  
439:   
440: The surface brightness is composed of 3 components: 1) innermost central part   
441: with centrally peaked brightness ($r_s <  2$\arcmin),  
442: 2) smooth component (2\arcmin\ $< r_s <$ 35\arcmin), and 3)  northeastern and   
443: southwestern shell   
444: (20--30\arcmin\ in radius) components.  
445: In the innermost central region, the spectrum  
446: is rather hard, and this region can be seen in both hard and soft  
447: emission in Figure~\ref{w28ascahard}.   
448: The observed surface  
449: brightness distribution in the interior of W28 implies a radial density   
450: profile described by  
451: $n/n_c = 1 - r/R$, as shown in Fig. \ref{w28sb}a,   
452: where $n(r)$ is plasma density as a function of  
453: distance $r$ from the remnant's center, $n_c$ is the central density,   
454: and $R$ is the remnant's radius. This profile is  
455: steeper than profiles usually found in other mixed-morphology SNRs  
456: (Rho \& Petre 1998). The overall  
457: shape of remnant is far from uniform and circular, typical for supernova  
458: remnants; it is dented in the north and east because of the presence of   
459: molecular clouds. Partial shell structures and X-ray features in  
460: the interior are often adjacent to molecular interaction sites, and appear  
461: particularly bright when projection effects allow for longer   
462: line-of-sight distances through the X-ray emitting gas. This most often   
463: happens in regions least affected by the irregular cloud geometry, such as  
464: the northeastern shell. The projection effects are not  
465: favorable in the north, where molecular clouds are located on the near   
466: (front) side of the remnant (Arikawa et al. 1999). Instead of a shell,  
467: we observe an abrupt drop in the X-ray surface brightness at the location of  
468: molecular clouds (the northern flat boundary).   
469:   
470: \subsection{Spectral Analysis using ASCA/ROSAT spectra}  
471:   
472:   
473: We extracted three sets of ASCA and ROSAT PSPC spectra  
474: for   
475: the northeastern shell, center and southwestern shell (see  
476: Figure~\ref{w28regionmark}).   
477: The GIS and SIS spectra  
478: were extracted from each of the three pointings (we used the long exposure of  
479: sequence 51022000 for central region).  
480: The central region spectra covered most  
481: of GIS and SIS fields of view in the central pointing   
482: (ad51022000; see Figure \ref{w28ascafov}).   
483: The southwestern shell and the northeastern shell extraction regions follow   
484: the shapes of the shells as shown in Figure \ref{w28regionmark}:   
485: for the southwestern  
486: shell, the extracted region is  
487: a 12$' \times 8'$ ellipse inclined at PA 162$^{\circ}$,  
488: and for the northeastern spectra, it is an 8\farcm2 $\times$ 3\farcm2 ellipse at   
489: PA 150$^{\circ}$.  
490: %Smaller region of spectra did not have sufficient  spectral  
491: %signal to noise to obtain meaningful spectral fitting results.   
492: The   
493: central spectra, after background subtraction, have count rates of   
494: 1.004$\pm$0.007 cts s$^{-1}$  for GIS2,  0.734$\pm$0.006 cts s$^{-1}$ for GIS3,   
495: 0.785$\pm$0.007 for SIS0, 0.950$\pm$0.008 cts s$^{-1}$ for SIS1,   
496: and 0.704$\pm$0.015 cts s$^{-1}$ for ROSAT PSPC.   
497: The southwestern shell spectra after background subtraction have count rates of  
498: 0.412$\pm$0.006 cts s$^{-1}$  for GIS2,  0.320$\pm$0.006 cts s$^{-1}$ for GIS3,  
499: 0.222$\pm$0.006 for SIS0, 0.174$\pm$0.005 cts s$^{-1}$ for SIS1,   
500: and 0.765$\pm$0.015 cts s$^{-1}$ for ROSAT PSPC.  
501: Constraints imposed by the internal calibration source led to a reduction  
502: in the effective GIS3 extraction area, resulting in lower GIS3 count  
503: rates relative to GIS2.  
504: The northeastern shell spectra after background subtraction have count rates of  
505: 0.215$\pm$0.003 for SIS0, 0.276$\pm$0.005 cts s$^{-1}$ for SIS1,   
506: and 0.357$\pm$0.007 cts s$^{-1}$ for ROSAT PSPC. We did not use GIS spectra  
507: because the signal-to-noise ratio for this small region was too low  
508: to be useful.  
509:   
510: The central spectra show emission lines of Ne, Mg, Si,  
511: and Fe K$\alpha$, and continuum at least up to 7 keV (Figure \ref{w28spec}a).  
512: The presence of emission lines means that the  
513: emission is  thermal   despite the center-filled morphology,  
514: implying that X-rays are not powered by the pulsar like in the Crab Nebula.   
515: A single power law fit is significantly rejected.  
516: This spectrum is different  
517: from another archetype mixed-morphology SNR, W44, where no emission  
518: above 5--6 keV is reported. We examined the  
519: source-free background region to see if the Fe K$\alpha$ line comes from   
520: the Galactic Ridge, but there is no evidence of Fe K$\alpha$ in source-free  
521: background spectra. In addition, the Fe K$\alpha$ line image confirms that the  
522: emission is from the remnant itself, primarily from the center and  
523: the southwestern shell.   
524:   
525: When we compare the central spectra with the southwestern and northeastern  
526: shell spectra, we find noticeable differences.   
527: The southwestern spectrum (at RA 17$^{\rm h}$56$^{\rm m}$,   
528: Dec -23$^\circ$35$'$, see Fig. \ref{w28regionmark}) is much harder than the   
529: central spectrum as it can be seen in Figure \ref{w28spec}a, while  
530: the northeastern shell spectrum (see Figure \ref{w28spec}b) is softer  
531: than spectra in other regions of the remnant. In order to quantify these  
532: differences, we simultaneously fit the spectra to five sets of  
533: ROSAT/PSPC, ASCA/SIS0 and SIS1, and ASCA/GIS2 and GIS3 spectra.    
534: We used  the collisional ionization equilibrium (CIE)   
535: thermal model (Mewe-Kaastra plasma model; Kaastra 1992), and nonequilibrium  
536: ionization (NEI) thermal models as implemented in XSPEC  
537: (Borkowski, Lyerly, \& Reynolds 2001), which included updated Fe L-shell  
538: atomic data based on calculations by Liedahl, Osterheld, \& Goldstein (1995).   
539: Fit parameters are tabulated in Table \ref{w28ascafit}.  
540: The SW spectrum is fit well by the plane shock model with a high  
541: temperature of $kT_s = 1.2$ keV. A fit with a simpler single ionization  
542: timescale NEI model reproduced the spectra   
543: moderately well ($\Delta \chi^2 \sim 1.16$). The Sedov  
544: model for SW region yielded $kT_{s} = 1.2$ keV, and an   
545: ionization timescale of $n_et = 3.6 \times 10^{11}$  
546: cm$^{-3}$ s. The  
547: northeastern shell is fit best by NEI  
548: model with a lower temperature of $kT_e = 0.56$ keV, and  
549: a longer ionization timescale of $n_et = 1.7 \times 10^{13}$  
550: cm$^{-3}$ s, than those from  the southwestern shell.  
551: The line-of-sight absorption toward the northeastern shell is   
552: $N_H = 0.8 \times 10^{22}$ cm$^{-2}$.  
553:   
554: For the central spectra,  
555: one-temperature CIE or NEI models did not   
556: yield acceptable fits to  
557: the central spectra ($\Delta \chi^2 > 2.3$, and $\Delta  
558: \chi^2 > 1.8$, respectively).  
559: To reproduce  the moderately strong Fe K$\alpha$ line in  
560: the central region  requires a second thermal component with a higher  
561: temperature (see Figure \ref{w28spec}c).  
562: For the two-temperature model fitting, we used three steps in  
563: spectral fitting. 
564: First, we fit the Fe K$\alpha$ line centroid using a Gaussian model,  
565: arriving at 6.6$^{+0.04}_{-0.10}$ keV, which allowed us to estimate  
566: ionization parameter. 
567: As the ionization timescale $\tau$ increases, Fe ions become  
568: more ionized, with electrons gradually removed from the L shell   
569: which results in a higher line emissivity and a line  
570: centroid shift to higher energies.  
571: For $\tau <  10^{10}$ cm$^{-3}$ s, Fe ions are very  
572: underionized as compared to ionization equilibrium, and the Fe  
573: K$\alpha$ line is produced mostly through fluorescence following  
574: K-shell ionization by electrons. %so as $\tau$ increases, Fe ions become  
575: The observed  centroid  implies an ionization time scale for the hot  
576: temperature component of $\sim 5 \times 10^{11}$ cm$^{-3}$ s (e. g., see Fig.  
577: 2 of Borkowski \& Szymkowiak 1997).  As the second step, we  constrained the   
578: temperature and  
579: Fe abundances using only hard energy band (4--10 keV) spectra, with  
580: the bremsstrahlung model for the continuum and a Gaussian line    
581: for Fe K$\alpha$. The constraint  
582: is 1.8 keV for the central  
583: region, and the Fe abundance is  consistent with solar abundances.  
584: A single ionization timescale NEI model fit in the hard energy band  
585: alone also gave $\tau = 5 \times 10^{11}$ cm$^{-3}$ s, 1.8 keV temperature,  
586: and approximately solar Fe abundance.  
587: The fits are summarized in Table \ref{w28ascafit}. As the third step,  
588: we added a second, soft component (CEI  
589: model, single ionization timescale NEI model, or Sedov  
590: model), while the temperature and  
591: ionization timescale for the hot component were fixed to the values   
592: obtained in previous steps.   
593: A  two-temperature  
594: NEI model   
595: reasonably well reproduced the observed  
596: spectra  ($\Delta\chi^2 \sim 1.1$, see Figure \ref{w28spec}d) with a hot   
597: temperature of $kT_{hot}  
598: = 1.8\pm0.5$ keV and ionization timescale of  
599: $\tau_h = 5 \times 10^{11}$ s~cm$^{-3}$, and a low temperature of  
600: $kT_l = 0.6$ keV with $\tau_l = 2 \times 10^{12}$ s~cm$^{-3}$.  
601: (We note that modeling with just 2  
602: temperature  components is merely a convenient way of quantifying  
603: temperature differences present in the remnant. Because  
604: multi-temperature plasmas might be expected in complex remnants such as W28,   
605: a considerable caution is required in  
606: interpreting results of the two-component fits.)   
607: Both NEI and Sedov models provide equally good fits for the soft component,  
608: with a high ionization parameter of $\sim 2 \times 10^{12}$ cm$^{-3}$ s.   
609:   
610: Torii et al. (1996) suggested the presence of nonthermal X-ray emission in   
611: the southwest. With the poor S/N SW spectra, the presence of nonthermal   
612: emission is still an open question, which cannot be completely ruled out   
613: at this time.  
614: But because we see Fe K$\alpha$ line in GIS2 spectra of the SW region, we   
615: favor thermal over nonthermal origin for the high energy emission.   
616: If nonthermal component were present in SW spectra, we would have  
617: a combination of thermal and nonthermal emission, because thermal line   
618: emission from elements such as Mg and Si is certainly present at low  
619: energies.  
620: It is also possible that the hard X-ray continuum and the Fe K$\alpha$   
621: line come from different regions.   
622: We attempted to find differences in spatial distributions between the  
623: hard continuum and the Fe K$\alpha$ line, but this attempt  
624: failed because of low count rates at high energies.  
625: Better data are needed to confirm that the southwestern   
626: emission is of purely thermal origin, and to study the remnant's  
627: morphology in the Fe K$\alpha$ line and in the hard continuum.  
628:   
629: The derived ISM absorption is higher in the SW than in the northeastern  
630: shell and the central part of remnant (see Table 2). Considering that  
631: the remnant is interacting with clouds in north, east, and northeast,  
632: one might expect an enhanced absorption in these regions of the remnant,  
633: just the opposite of what is observed. However, if  
634: there are clouds in front of the remnant in the SW, the line-of-sight  
635: absorption can still be high  
636: without any interaction between the remnant and clouds.  
637: This is the case of IC 443, where $N_H$ is higher across the central part of  
638: remnant because of the presence of  
639: molecular clouds in the front of the remnant, although the interaction  
640: with the clouds is seen in the south (Rho, Petre, \& Hester 1994).  
641: There is  
642: no optical emission detected in the SW part of W28 (van den Bergh 1978;  
643: Long et al. 1991; Winkler, in private communication) which seems to suggest  
644: that absorption is indeed higher in the SW. But an equally plausible  
645: explanation  
646: for the weakness of optical lines is simply the absence of radiative shocks  
647: in the breakout SW region of the remnant. More observations are needed in  
648: order to resolve this issue, because the poor S/N spectrum of the  
649: southwestern shell ultimately limits our ability to understand the origin  
650: of its relatively hard spectrum.    
651:   
652: 
653: \subsection{Other sources in the field of view including the pulsar}  
654:   
655: A glitching pulsar, PSR B1758-23, was detected at the northern  boundary of  
656: W28 (Kaspi et al. 1993; Frail, Kassim, \& Weiler 1994), at  
657: R.A. $18^{\rm h} 01^{\rm m} 19.859^{\rm  
658: s}\pm0.058^{\rm s}$   
659: and Dec. $-23^\circ$ 06$^{\prime} 17^{\prime \prime}$$\pm 102''$.  
660: An association of this pulsar with W28 was  
661: suggested by Kaspi et al. (1993), with   
662: the pulsar distance estimated at  
663: 3 kpc (Frail, Kulkarni, \& Vasisht 1993). We   
664: examined this region in detail and found that the PSPC surface brightness  
665: is 2.5 times higher than the background brightness. 
666: The pulsar position is marked on  
667: the PSPC image in Figure~\ref{w28pulsar} (rp900395 set of data, 9.3 ks in duration).   
668: Within errors, the centroid coincides with the pulsar.  
669: The small number of detected counts was not sufficient to perform a  
670: spectral or a timing analysis.  
671: The X-ray PSPC count rate is 3.05$(\pm1.21)\times$10$^{-3}$  
672: counts s$^{-1}$, which implies a luminosity of  
673: 1.2$\times$10$^{32}$ ergs s$^{-1}$, assuming   
674: a line-of-sight absorption of $8 \times 10^{21}$ cm$^{-2}$  
675: and a power law spectrum with $\Gamma = 2$ (equal to the spectral index of   
676: the  Crab Nebula). The luminosity  
677: is less than 0.5\% of the total X-ray luminosity of W28.  
678: A deeper observation is  
679: required to confirm whether this is true detection or not.  
680:   
681:   
682: There are two X-ray sources to the east of the remnant. The  
683: northern source (R.A.  
684: $18^{\rm h} 02^{\rm m} 10^{\rm s}$ and Dec. $-23^\circ$ 33$^{\prime}  
685: 38^{\prime \prime}$) is very soft, without  
686: a known counterpart. It is  
687: probably a stellar source, which we denote as RX1802-2333.  The  
688: second (southern) source,  
689: located at R.A. $18^{\rm h} 01^{\rm m} 59.7^{\rm s}$ and Dec.  
690: $-23^\circ$ 41$^{\prime} 20.8^{\prime \prime}$, is the HII region BFS 1,  
691: and its O-type star, CPD-23 6751, emits X-rays. % (see Figure~\ref{w28pspc}).   
692: There is another  
693: X-ray source  in the northeast which can be identified with another H\,II region,  
694: the Trifid nebula,  
695: whose observation (R.A. $18^{\rm h} 02^{\rm m} 23.35^{\rm s}$ and Dec.  
696: $-23^\circ$ 01$^{\prime} 47.0^{\prime \prime}$) is reported elsewhere   
697: (Rho et al. 2001).  
698:   
699: 
700: 
701: 
702:   
703:   
704: \section{Discussion}  
705:   
706: Because W28 is a complex remnant in an advanced evolutionary stage, we first   
707: discuss its global kinematics and dynamics based on  
708: observations at wavelength bands other than X-rays. Next, we discuss and   
709: interpret X-ray observations. Finally we compare W28 with other   
710: mixed-morphology remnants.  
711:   
712: \subsection{Kinematics and Dynamics of W28}  
713:   
714: Fabry-Perot observations of the brightest optical emission at the center of  
715: W28 gave expansion velocity of 40 km s$^{-1}$ (Lozinskaya 1974), with  
716: individual filaments moving with radial velocities up to 80 km s$^{-1}$ and  
717: with the average H$\alpha$ width of 40--50 km s$^{-1}$.  
718: This should also be the blast wave velocity if this optical emission   
719: projected near the center of W28   
720: has the same origin as the radiative filaments seen at the northern edge of  
721: the remnant. Optical emission in the interior has a more diffuse, chaotic  
722: appearance than the filamentary emission usually associated with radiative  
723: shocks seen edge on. Such diffuse emission is generally seen in remnants  
724: at late stages of their evolution (Lozinskaya 1992), and might potentially  
725: be associated with other physical processes such as cloud evaporation  
726: (Long et al. 1991) or cooling of X-ray emitting gas in the remnant's  
727: interior. But at least some of this emission must be produced by   
728: the radiative blast wave seen in projection against the center of the remnant,  
729: in view of a generally good correlation between radio, optical, and molecular   
730: emission (Dubner et al. 2000). Kinematic information about the blast  
731: wave velocity obtained from Fabry-Perot observations is in a fair agreement   
732: with estimates of shocks speeds derived from optical spectroscopy:   
733: Bohigas et al. (1983) estimate shock velocities between 60 and 90 km s$^{-1}$  
734: from the ratio of [\,O III\,] $\lambda$5007 /H$\alpha$ line strengths,   
735: while Long et al. (1991) derive velocities larger than 70 km s$^{-1}$  
736: from ratios of [\,N II\,] and [\,S II\,] line strengths to the H$\alpha$ line  
737: strength. All available evidence based on optical observations suggests   
738: that the blast wave velocity $v_s$ is $\le 100$ km s$^{-1}$, with considerable  
739: spatial scatter with velocities varying in the 60 -- 100 km s$^{-1}$ range.  
740:   
741: Only upper limits on the preshock density $n_0$ ahead of the blast   
742: wave can be   
743: derived from optical and infrared observations. Long et al. (1991) obtained  
744: an upper limit for the electron density of 70 cm$^{-3}$ from observations of   
745: the density sensitive  
746: [\,S II\,] $\lambda$6717/$\lambda$6731 line ratio. This is consistent with  
747: the density of $\sim 100$ cm$^{-3}$ derived by Bohigas et al. (1983) from  
748: this line ratio. An upper limit  
749: of $\sim 100$ cm$^{-3}$ from the [\,O III] 52$\mu$/88$\mu$ line ratio  
750: (Reach \&\ Rho 2000) is also consistent with this result. An average preshock   
751: density must be significantly lower than these upper limits, but by   
752: an uncertain factor which depends primarily on the magnitude of the magnetic   
753: field support. In addition, the observed optical emission may originate not   
754: only in the postshock cooling region. For example, there is some evidence for   
755: the presence of a faint H II region from variations in the optical line ratios   
756: across the remnant (Long et al. 1991). Neglecting contribution from  
757: a suspected H II region and in the absence of the  
758: magnetic field support, the inferred preshock density would be below  
759: 1 cm$^{-3}$.   
760:   
761: Magnetic field is likely to be dynamically important in the  
762: shocked gas at temperatures where [\,S II\,]  
763: $\lambda$6717/$\lambda$6731 emission is produced.  
764: Magnetic pressure becomes  
765: equal to the gas pressure in the postshock cooling gas at densities  
766: $n_m = 240 v_s (n_0/10 {\rm cm}^{-3})^{3/2} (B_0/10^{-5}{\rm G})^{-1}$   
767: cm$^{-3}$, where  
768: $B_0$ is the preshock tangential field (Chevalier  
769: 1999). For shock velocity of 80 km s$^{-1}$, preshock density  
770: of 5 cm$^{-3}$, and $B_0 = 10^{-5}$ G, $n_m$ is equal to 70 cm$^{-3}$ which   
771: is just consistent with the upper limit derived from [\,S II\,] lines.   
772: The preshock density of 5 cm$^{-3}$ is at the lower end of the  
773: typical density range of 5--25 cm$^{-3}$ encountered in interclump media  
774: of galactic molecular clouds (Chevalier 1999), through which the blast wave  
775: propagates. This uncertain estimate depends   
776: strongly on the unknown value of $B_0$, and a possible contamination of the   
777: optical and infrared emission by an H II region means that   
778: the preshock interclump density might be higher than 5 cm$^{-3}$. A fiducial   
779: interclump density of 10 cm$^{-3}$ (Chevalier 1999) might be consistent with  
780: observations.  
781:   
782: The postshock cooling gas should be recombining into neutral hydrogen, which  
783: can be observed in emission and absorption in the 21 cm H\,I line. Based on the  
784: preshock density of 10 cm$^{-3}$ and the remnant's size, it is possible to   
785: estimate the total mass of the neutral hydrogen in the swept radiative shell.  
786: The bright hemispherical northern part of W28, where most of the optical  
787: emission is located, is about 30$'$ in diameter (9.5 pc in radius at  
788: distance of 1.8 kpc), while the southern part is much   
789: fainter and about twice as large in the east-west direction. This suggests  
790: that the blast wave broke out of the molecular cloud in the south. For the   
791: bright  
792: northern part we estimate that about 700 $M_{\odot}$ should be present in  
793: the neutral shell. Radio observations in the 21 cm line suggest that there is  
794: neutral hydrogen associated with W28.  
795: From a rather sparse cross-pattern  
796: mapping, Koo \& Heiles (1991) found an excess emission over the galactic  
797: background at the location of W28, some of it with radial velocities in  
798: excess of 100 km s$^{-1}$ with respect to the systemic velocity of W28.  
799: Venger et al. (1982) reported a detection of an expanding H I envelope 82 pc  
800: in diameter, with mass of $6.9 \times 10^4$ $M_{\odot}$ and expansion velocity  
801: of 20 km s$^{-1}$, centered on W28. While they identified this massive   
802: envelope with the expanding shell of W28, this interpretation is not  
803: consistent with information obtained at other wavelength bands.  
804: The emission  and absorption in the 21 cm  
805: line  in this region of the sky is apparently complex, and further  
806: studies of W28  and its neighborhood in the 21 cm line are warranted.  
807: Such studies could provide us with crucial information about the  
808: properties of the radiative shell.  
809:   
810: A large amount of ionized gas is seen at optical wavelengths. Long et al.  
811: (1991) estimated the total H$\alpha$ flux at $1.3 \times 10^{-8}$  
812: ergs cm$^{-2}$ s$^{-1}$, which at $10^4$ K temperature, $\sim 70$ cm$^{-3}$  
813: density, and 1.8 kpc distance implies the presence of 300 $M_{\odot}$ of  
814: ionized gas. This is comparable to the estimated mass of the radiative  
815: shell. For the preshock density of 10 cm$^{-3}$, blast wave velocity of   
816: $\sim 100$ km s$^{-1}$ and radius $R$ of 9.5 pc, we estimate that the ionizing  
817: radiation generated by the radiative blast wave falls short of producing   
818: the ionizing flux necessary to account for the observed H$\alpha$ flux  
819: by at least one order of magnitude. An unknown additional source of ionizing  
820: radiation must be present within the remnant. Possible sources of ionizing  
821: radiation include evaporating clumps in the remnant's interior (Long et al.  
822: 1991) or gas cooling from X-ray emitting temperatures (e. g., Cox et al. 1999).  
823:   
824: The remnant's age is approximately equal to $0.3 R/v_s = 3.6 \times 10^4$ yr,  
825: with $R = 9.5$ pc and $v_s = 80$ km s$^{-1}$, and assuming that  
826: the remnant is in the snow-plow stage of the evolution. The estimated  
827: kinetic energy $E$ of the explosion is $4 \times 10^{50}$ ergs (using   
828: equation  
829: [26] of Chevalier 1974). This might be an underestimate because the blast  
830: wave most likely broke out of the molecular cloud in the south. We also note  
831: that the derived value of $E$ depends strongly on the assumed distance $d$ to  
832: the remnant ($E \propto d^{3.12}$, see \S~1 for discussion on the uncertainty   
833: of the distance). From simple analytical estimates (e. g.,  
834: see Blondin et al. 1998), the transition from the adiabatic (Sedov) stage to  
835: the radiative stage should have occurred at   
836: $t_{tr} \approx 8.6 \times 10^3 E_{51}^{4/17} n_1^{-9/17}$ yr, where   
837: $E_{51}$ is the explosion energy in $10^{51}$ ergs, and preshock density  
838: $n_1$ is measured in units of 10 cm$^{-3}$. The radius and velocity of the  
839: shock at this time were equal to $7.4 E_{51}^{5/17} n_1^{-7/17}$ pc and  
840: $340 E_{51}^{1/17} n_1^{2/17}$ km s$^{-1}$. While cooling is important at  
841: $t_{tr}$, the cool radiative shell forms at about  
842: $1.5 t_{tr}$ (see Table 1 in Blondin et al. 1998), when the shock velocity  
843: dropped to $220 E_{51}^{1/17} n_1^{2/17}$ km s$^{-1}$.  
844:   
845: Significant pressure variations must be   
846: present within the remnant. For $n_0 = 10$ cm$^{-3}$ and   
847: $v_s = 80$ km s$^{-1}$, the average pressure $p/k$ in the radiative shell is  
848: equal to $1.1 \times 10^7$ cm$^{-3}$ K, while Frail \& Mitchell (1998)  
849: obtain  $p/k \sim 4.6 \times 10^7$ cm$^{-3}$ K   
850: from observations of the Zeeman splitting in the OH 1720 MHz line.   
851: At two locations within W28, Claussen et al.  
852: (1999) detected an even stronger magnetic field, which would imply pressures  
853: $p/k \sim 4.6 \times 10^9$ cm$^{-3}$ K. Such high pressures are expected  
854: to be generated by the impact of the radiative shell with the molecular  
855: clumps (Chevalier 1999). Pressures should be lower than average in the   
856: southern and the southwestern parts of W28, but no pressure estimates   
857: are available in this faint breakout region from radio, optical, or   
858: infrared observations.  
859:   
860: \subsection{Interpretation of X-Ray Observations}  
861:   
862: The southwestern region has a shell-like morphology, with a fairly good   
863: correlation between the X-ray and radio emission, and optical emission  
864: from radiative shocks is not present. This suggests that perhaps description of  
865: the SW shell spectra in terms of plane-parallel shock and Sedov   
866: models might be appropriate. This is unlike the central region, where  
867: X-ray emission is of unknown origin and its center-filled morphology is   
868: different from the Sedov solution,  while the northeastern shell is located   
869: in the radiative section of the remnant, close to dense  
870: molecular clouds with OH masers.   
871:   
872: We use our spectral fits (Table \ref{w28ascafit}) to estimate  
873: physical parameters for the SW  
874: region. % summarized in Table~\ref{w28shock}.  
875: The temperature is equal to 1.5 keV, implying a shock velocity of  
876: 1130 km s$^{-1}$. Electron density may be derived from emission   
877: measure in Table 2, equal to $3.5 \times 10^{56} d_{1.8}^2$ cm$^{-3}$, and  
878: from the estimated volume $V$ of the X-ray emitting material. The size of the  
879: shell seen in this location is 8\arcmin\ $\times$ 16\arcmin, or   
880: $14 \times 9$ pc at  
881: 1.8 kpc distance. Assuming a comparable (14 pc) depth of the shell along the   
882: line of sight, we arrive at $V \sim 9 \times 10^{57} d_{1.8}^3$ cm$^3$.   
883: Then, $n_e  = 0.22 f^{-1/2} d_{1.8kpc}^{-1/2}$ cm$^{-3}$,  
884: where f is a filling fraction of the X-ray emitting material within  
885: volume $V$.  
886: With the electron density of 0.22 cm$^{-3}$ and  
887: ionization age of $1.5 \times 10^{11}$ cm$^{-3}$ s, the shock is about  
888: $2.2 \times 10^4$ yr old. This is consistent with the estimated remnant's age   
889: of $3.6 \times 10^4$ yr. The preshock hydrogen density  
890: $n_0$ is equal to 0.05 cm$^{-3}$, where  
891: $n_0 = n_e/4.8$ for cosmic abundance plasma and strong shock jump  
892: conditions. This density is 2 orders of magnitude lower than the  
893: interclump density in the northern radiative section of the remnant,   
894: consistent with the observed breakout morphology of W28. The thermal pressure  
895: $p/k$ of the X-ray emitting plasma, equal to   
896: $\sim 5 \times 10^6$ cm$^{-3}$ K, is as expected somewhat lower than   
897: pressures encountered in the bright northern hemisphere. One   
898: problem is the high shock speed of 1100 km s$^{-1}$,   
899: because with this speed the shock would propagate to a distance of 40 pc  
900: in $3.6 \times 10^4$ yr, which is much larger than the remnant's radius.   
901: In order to investigate this issue, we now discuss the SW shell in the  
902: framework of spherical (Sedov) shock models instead of plane shocks.  
903:   
904: Our Sedov spectral model fit yielded $T_s = 1.2$ keV, and  
905: ionization timescale of $3.6 \times 10^{11}$ cm$^{-3}$ s. The shock  
906: temperature in Sedov model is defined as the temperature immediately behind the  
907: shock. These spectral parameters are consistent with the plane-parallel shock   
908: model fit parameters; for ion-electron  
909: equipartition, we expect $T_{shock} = 1.27T_{Sedov}$   
910: (Borkowski, Lyerly, \& Reynolds 2001).  The ionization timescale   
911: $\tau_{Sedov}$ is defined as the product of  
912: the electron density immediately behind the shock and the remnant's age,  
913: but the mean ionization timescale  
914: of the gas is equal to 0.202 $\tau_{Sedov}$, which   
915: should be equal to half of shock age, i.e., $\tau_{shock}/2$. Therefore,   
916: $\tau_{shock} \approx 0.404\tau_{Sedov}$, which is consistent with   
917: results obtained from shock and Sedov spectral fits. The physical parameters   
918: of the remnant can be estimated from the Sedov spectral fit  
919: parameters.  Using   
920: $T_s = 1.2$ keV and  $n_et = 3.6 \times 10^{11}$ cm$^{-3}$ s in equations of   
921: Hamilton, Chevalier, \& Sarazin (1983), we get shock velocity of   
922: 1000 km s$^{-1}$, age of 5100 yr, explosion energy of  
923: $2.2 \times 10^{51}$ ergs s$^{-1}$, and preshock density of 0.5 cm$^{-3}$.  
924: The remnant's age estimated in this way is much lower than   
925: $3.6 \times 10^4$ yr, which is not surprising in view of the problem  
926: that the shock velocity is too high for the radius of W28,  
927: encountered with the plane shock analysis.  
928: There is an order of magnitude discrepancy between the preshock  
929: densities derived from the Sedov model and estimated from   
930: the emission measure. This discrepancy, coupled with the low estimated age,  
931: suggests that Sedov analysis probably does not apply to the breakout  
932: region of W28, for example, because of large density gradients expected  
933: there. It is then doubtful that the current shock velocity is as high as  
934: 1000 km s$^{-1}$. For example, material in the southwestern  
935: shell section could have been shocked long time  
936: ago, and then adiabatically expanded into the low density region on the  
937: periphery of the molecular cloud complex. In this case, the blast wave itself   
938: would have propagated in the low density medium ahead of the observed shell,  
939: and could be too faint to be detectable with ROSAT or ASCA.   
940: Observations with a higher spectral  
941: and spatial resolution, and with a higher S/N ratio are needed in order to   
942: resolve this issue. The absorbed flux (0.3-10 keV) from this shell section is   
943: 4.5($\pm$0.4)$\times 10^{-12}$ ergs  
944: s$^{-1}$ cm$^{-2}$, and unabsorbed flux is 4.0($\pm$0.4)$\times  
945: 10^{-11}$ ergs s$^{-1}$ cm$^{-2}$, which is equivalent to X-ray  
946: luminosity of L$_{\rm x}$ of  1.45$\pm 0.14 \times  10^{34}$ ergs  
947: s$^{-1}$ at 1.8 kpc distance.  
948:   
949: For central spectra, we estimated electron densities  
950: and filling factors using emission measures. The lower temperature  
951: component with 6.9$\times$10$^6$ K (0.59 keV) has a density of $0.48 f_{cold}^{-1/2}$   
952: cm$^{-3}$ and the higher temperature component with 2.1$\times$10$^7$ K  
953: (1.8 keV) has a density of  $0.19 f_{hot}^{-1/2}$ cm$^{-3}$.   
954: Assuming that the hot and cold gases   
955: are in pressure equilibrium, and that the whole volume is filled with the  
956: emitting gas (i.e., $f_{cold}+f_{hot}=1$), the filling factors of   
957: cold and hot gases are 0.4 and 0.6, respectively. Electron  
958: densities in low and high temperature components are 0.75 and 0.25 cm$^{-3}$.  
959: The thermal pressure $p/k$ of $9.5 \times 10^6$ cm$^{-3}$ K is of the same   
960: order as our estimates of the pressure in the radiative shell, but   
961: significantly less than in  
962: dense regions with OH masers.   
963: Using the best-fit two-temperature NEI model, we estimated the flux and   
964: luminosity of X-rays.  
965: The absorbed flux for the central region (using 2T vnei models) is   
966: 1.7$\pm0.3$$\times 10^{-11}$  
967:  ergs s$^{-1}$ cm$^{-2}$,  
968: and unabsorbed flux is 8($\pm$2)$\times 10^{-11}$ ergs s$^{-1}$ cm$^{-2}$, which  
969: is equivalent to X-ray luminosity of L$_{\rm x}$ of  
970: 2.9($\pm0.7$)$\times 10^{34}$ ergs s$^{-1}$, which is twice as high as that   
971: of the SW region. The northeastern region spectral fit gives an X-ray   
972: luminosity similar to that of central region.  
973:   
974: The limb brightened northeast region (the left ``ear'') has a shell-like  
975: appearance and it appears to be spatially coincident with radiative shell   
976: as indicated by a very good spatial match with bright radio and optical  
977: filaments (Dubner et al. 2000; see also Fig.~\ref{w28sb}). The temperature of   
978: the X-ray emitting gas is equal to 0.56 keV (6.5$\times 10^6$ K).   
979: This temperature is   
980: significantly higher than expected from a spherically symmetric remnant  
981: at the late radiative stage of the evolution with parameters as estimated  
982: above. For example, the postshock temperature at $t_{tr}$ is equal to  
983: $0.14 E_{51}^{1/17} n_1^{2/17}$ keV in the Sedov model. Falle (1981)  
984: demonstrated that the emission-measure weighted temperature of the  
985: X-ray emitting gas at times longer than $t_{tr}$ is no longer related to the  
986: shock speed. It fluctuates instead by about 50\%\ around the value just  
987: quoted until at least $10 t_{tr}$ (see Fig. 8 in Falle 1981). This  
988: discrepancy between the expected and the observed temperature cannot be  
989: solved by simply postulating a much higher interclump density, because our  
990: density estimates based on the emission measure rule out very high   
991: densities of the X-ray emitting gas. Note however that it might be difficult  
992: to detect gas with very low temperatures because of the high   
993: ISM absorption along the line of sight to W28.   
994:   
995: Our best estimate of the total X-ray emitting mass is   
996: $\sim 20-25 d_{1.8}^{5/2} M_{\odot}$:  
997: most of it in the center and 8 $M_{\odot}$ in the northeastern and  
998: southwestern shells. This is less than expected from a radiative remnant  
999: with parameters appropriate for W28. Numerical calculations of   
1000: Hellsten \& Sommer-Larsen (1995) indicate that about 10\%\ of the total   
1001: shocked gas (by mass) is emitting in X-rays at $3.6 \times 10^4$ yr, for  
1002: a model SNR remnant with $E_{51}=1$ and $n_1=1$. We would then expect   
1003: $\sim 100$ $M_{\odot}$ of X-ray emitting gas in W28, mostly at low  
1004: (0.14 keV) temperatures. This is consistent with an analytical estimate  
1005: of 120 $M_{\odot}$, using equation (33) of Cui \& Cox (1992).  
1006: The observed X-ray emission comes from a much smaller  
1007: amount of hotter gas, suggesting a somewhat different SNR evolution than  
1008: implied by standard spherically-symmetric SNR models in a uniform ambient   
1009: medium.   
1010: This is ``fossil'' radiation, a conclusion supported by long cooling   
1011: timescales of the gas.  
1012: We estimated cooling timescales using temperatures and densities derived  
1013: above; for isobaric cooling this timescale is   
1014: $6 \times 10^{6}$ yr in the SW, and $3.7 \times 10^{5}$ yr for northeastern   
1015: shell gas. This is much longer than the age of the remnant, so that the  
1016: X-ray emitting gas cannot be associated with    
1017: fast radiative shocks or with gas cooling from high temperatures.  
1018: Instead, this material was shocked much earlier   
1019: in the evolution of the remnant, and its temperature and density may have  
1020: been modified since then by processes such as evaporation, electron thermal  
1021: conduction, and mixing induced by various hydrodynamical instabilities.   
1022: We consider next whether the evaporation model of White \& Long (1991) or  
1023: the conduction model of Cox et al. (1999) can explain the observed X-ray   
1024: properties of W28.  
1025:   
1026: The evaporation model of White \& Long (1991) is in conflict with    
1027: ASCA spectra. In this model, because of the short timescale   
1028: of evaporative flows induced by saturated thermal conduction, heavy-element  
1029: ions in gas evaporated from small clouds are strongly underionized with respect  
1030: to CEI plasmas when they are mixed with the  
1031: hot diffuse gas occupying most of the SNR interior. Other processes such as  
1032: hydrodynamical instabilities would further hasten mixing with the hot gas, so  
1033: that ionization should occur mostly in the hot diffuse gas. Long et al. (1991)   
1034: obtained an age of only 2400 yr for W28 using the evaporation model, which in   
1035: combination with electron densities estimated from  
1036: emission measures implies the presence of strongly underionized  
1037: gas in W28. But ASCA spectra reveal gas nearly in ionization equilibrium,  
1038: with ionization timescales at least one or two orders of magnitude longer  
1039: than expected. X-ray spectra alone are sufficient   
1040: to reject evaporation model for W28. But perhaps a more convincing argument  
1041: against this model is provided by the presence of radiative shocks in the   
1042: northern half of the remnant, which are not expected in the nonradiative  
1043: evaporative model considered by Long et al. (1991). The failure of this  
1044: model for W28 further supports conclusions of Chevalier (1999)   
1045: who argued against the White \& Long (1991) model as the explanation for the   
1046: MM SNRs.  
1047:   
1048: The radiative shell model of Cox et al. (1999) and Shelton et al. (1999), in  
1049: which electron thermal conduction modifies the interior of an SNR evolving in  
1050: a uniform ambient medium, provides reasonable values for the central density  
1051: and temperature. Using equations of (3) and (10) of Cox et al. (1999), with  
1052: $E_{51} =1$, $n_1 = 1$, and the Spitzer's conductivity coefficient, we arrive  
1053: at the central electron density of 0.93 cm$^{-3}$ at the present   
1054: ($3.6 \times 10^4$ yr) time. The central temperature of 0.52 keV is then  
1055: obtained by  
1056: equating the central pressure to the pressure within the radiative shell.  
1057: There is obviously a good agreement with the densities and temperatures  
1058: of the cool temperature component found in the central regions of W28. But   
1059: the conduction model predicts an X-ray emitting mass which is even larger   
1060: than in   
1061: the standard radiative shell model without conduction considered by  
1062: us previously, in apparent disagreement with observations. In addition, the   
1063: surface brightness of W28 is too sharply   
1064: peaked at the center in comparison with the conduction model (it is sharper   
1065: than that of W44; see Rho \& Petre 1998), and the clumpiness seen in X-ray   
1066: images is in contrast to a smooth gas distribution expected in this   
1067: model. But the most serious challenge for  
1068: any models in which conduction plays a major role might be the presence of  
1069: temperature gradients in W28. The observed temperature differences must  
1070: have originated early in the remnant's evolution, before the radiative shell  
1071: formed, when efficient conduction   
1072: would have obliterated temperature gradients in the remnant's interior.  
1073: Using a self-similar conduction model of Cui \& Cox (1992), we find that at the  
1074: time of the shell formation at $1.3 \times 10^4 $ yr, temperature dropped  
1075: from its central value by a factor of 2 at a sphere enclosing the innermost  
1076: 75 $M_\odot$ of X-ray emitting gas. In  
1077: apparent conflict with the model predictions, we observe larger  
1078: temperature differences within much smaller amounts of X-ray emitting gas.   
1079: Efficient thermal conduction might be inconsistent with the observed   
1080: X-ray properties of W28.  
1081:   
1082: The X-ray properties of W28, in particular the simultaneous presence of hot   
1083: (1.5 keV) and cold (0.5 keV) components with long ionization timescales,   
1084: are in conflict with predictions of the evaporation model of White \& Long   
1085: (1991) and perhaps are inconsistent with predictions of the  
1086: radiative shell model with conduction (Cox et al. 1999). Apparently   
1087: evaporation is less important than postulated by White \& Long (1991), and  
1088: conduction might be less efficient than postulated in both models.  
1089: The radiative shell interacting with clumpy molecular clouds, as  
1090: considered by Chevalier (1999), seems to provide a suitable framework for  
1091: W28. In this model, X-ray emission at the center is a ``fossil'' radiation,   
1092: whose properties   
1093: are not fully understood at this time. Uniform temperature gradients   
1094: seen in MM SNRs led Chevalier (1999) to argue that conduction  
1095: is generally important in their interiors. This argument cannot be used  
1096: for W28 because of the presence of temperature variations, including  
1097: a large-scale temperature gradient. This temperature gradient may   
1098: be attributed to different preshock densities (and hence shock speeds) in the   
1099: northeast and southwest, due to interaction with molecular clouds in the north and  
1100: northeast, and the breakout in southwest. In absence of efficient conduction  
1101: on spatial scales comparable to the radius of the remnant, appreciable   
1102: temperature gradients of the opposite sign to density gradients are generally  
1103: expected. Their presence in W28 suggests that conduction might have been at   
1104: least partially inhibited by magnetic fields, perhaps preferentially in the  
1105: direction perpendicular to the field lines of large-scale magnetic fields.   
1106: Another possibility involves poorly understood mass and energy exchange  
1107: processes between the hot and tenuous interclump medium and dense cool clumps  
1108: of gas embedded in this medium. While idealized models dominated by either  
1109: evaporation or by conduction failed to explain observations, conduction,  
1110: evaporation and mixing induced by various hydrodynamical   
1111: instabilities must still be present in view of the clumpy nature of molecular  
1112: clouds. Their relative importance remains to be determined by  
1113: comparing future model calculations with X-ray spectra and morphology of W28.   
1114: W28 should provide a stringent test for future modeling because of the  
1115: presence of the appreciable temperature variations.  
1116:   
1117:   
1118: \subsection{Comparison of W28 and other mixed-morphology SNRs}  
1119:   
1120: Understanding of MM SNRs requires a careful assessment of their X-ray   
1121: properties, so that competing models can be tested on the basis of their   
1122: ability to explain these properties. For example, uniform temperature profiles   
1123: characteristic of MM remnants are well explained by models with efficient  
1124: thermal conduction, but the presence of large temperature gradients in even   
1125: a small subset of these SNRs might pose a problem for these models. Because    
1126: X-ray spectra revealed multitemperature plasma in W28,   
1127: we contrast X-ray properties of W28 with other MM SNRs.  
1128:   
1129: W28 is known to be an archetype MM SNR along with W44 (Long et al. 1991;  
1130: Rho \& Petre 1998), because both show typical mixed morphology  
1131: with center-filled X-rays and the shell-like radio morphology. However,   
1132: our images and spectral analysis show that X-ray properties of W28 are   
1133: somewhat different. First, unlike in other known MM SNRs, a high  
1134: temperature plasma is certainly present, as evidenced by the Fe K$\alpha$ line   
1135: and the thermal continuum extending up to 10 keV. The southwestern shell has a   
1136: high temperature of 1.5 keV, and a high temperature component of 1.8 keV is  
1137: also present at the center.   
1138:   
1139: X-ray emitting gas in W44 has temperature of 0.4--0.9 keV without any  
1140: hard  emission above 6 keV present (Harrus et al. 1997; Rho et al.  
1141: 1994). Other MM  SNRs (e.g., see Rho \& Petre 1998) such as MSH 11-61A  
1142: (Rho  1995), HB21 (Lee et al. 2001), 3C391 (Rho \& Petre 1996; Chen \& Slane  
1143: 2001), and 3C 400.2 (Saken et al. 1995; Yoshita et al. 2001), have gas in the  
1144: temperature range of 0.4--0.9 keV.    
1145: ASCA data of HB21 also imply a low temperature of 0.6 keV (Lee et al.  
1146: 2001), similar to results based on ROSAT data (Rho \& Petre 1998).  
1147: Plasma temperatures in these MM SNRs are between 0.4--0.9 keV. Other  
1148: remnants also have gas in this range of temperatures (Rho \& Petre  
1149: 1998). In contrast, there is gas with a much higher temperature of 1.5  
1150: keV in the southwestern shell and 1.8 keV in the central region of  
1151: W28.   
1152:   
1153:   
1154: The second difference is the presence of spectral variations in W28,   
1155: in contrast to other MM SNRs which have  
1156: uniform temperatures as revealed by spectral mapping  
1157: (Rho et al. 1994; Rho \& Petre 1998).  
1158: The spectrum of the southwestern region   
1159: is different  
1160: from the center. The southwestern spectrum  
1161: seems to show a high (1.5 keV) temperature  
1162: component alone without the soft component.  
1163: This may be partially accounted for by a higher absorption along the line of  
1164: sight in the SW, but the poor S/N spectrum of the SW region ultimately limits  
1165: our ability to understand the origin of these spectral differences.  
1166: The spectrum of the northeastern region is also different from  
1167: the center. The northeastern spectrum  
1168: seems to show a low (0.56 keV) temperature  
1169: component alone without requiring a second component.  
1170: The central spectrum  
1171: is noticeably different from the southwestern and northeastern spectra.    
1172: Our results show the presence of pronounced spectral  
1173: variations within W28, quite unlike what is seen in more  
1174: typical MM SNRs.  
1175: Another difference is that X-ray mass of W28 is much  
1176: less than those of other MM SNRs such as W44 and 3C391 (Rho et al. 1994;  
1177: Rho \& Petre 1996).  
1178:   
1179:   
1180: W28 has still common characteristics with other MM SNRs; X-ray  
1181: emission is thermal  with rich line emission as shown by the ASCA  
1182: spectra, and the spectral fitting implies that the abundances are close to  
1183: solar (ISM) abundances, without any evidence for ejecta material. There is  
1184: also the prominent central X-ray emission within the well-defined radio shell.  
1185: Therefore, W28 is still an MM SNR, but   
1186: with a number of different X-ray properties;  
1187: 1) partial shells are present, 2) X-ray emission is extremely clumpy,  
1188: particularly in the northeastern shell and at the center,  
1189: 3) temperature is higher than in other MM SNRs, and   
1190: 4) spectral variations are present due to  
1191: temperature differences. The difference between the  
1192: northeastern and southwestern shells appears to be related to differences in  
1193: preshock densities, while both hot and cool components are present in the  
1194: central region.  
1195:   
1196: Recently discovered MM SNRs  
1197: such as G359.1-0.5 (Bamba et al. 2000),  
1198: Sgr A East (Maeda et al. 2001) and Kes 27 (Rho et al. 1998)   
1199: also share some of these characteristics  
1200: with W28. They show evidence for very hot temperatures  (1.5 -- 4 keV),  
1201: and mixed morphology between X-rays and radio, and they are located    
1202: in dense environments, possibly interacting with nearby clouds.  
1203: Several of them even show evidence of SN ejecta,   
1204: which strongly suggests that they are younger than more well-known  
1205: MM SNRs W44 and 3C 391.   
1206:   
1207: \section*{Summary}  
1208:   
1209: 1. The overall shape of X-ray emission in W28 is elliptical, with the  
1210: emission concentrated at the center of the remnant. The interior emission  
1211: is sharply peaked at the center and its morphology is highly patchy.  
1212: Partial shells are present in the northeast and southwest, and the  
1213: northeastern shell also has a patchy, partially broken appearance.   
1214: There are ear-like segments of limb-brightened shells in the northeast  
1215: and northwest. The X-ray surface brightness between the two ear-like  
1216: patches stays very flat from  east to west along Dec -23$^{\circ}$  
1217: 15$'$. The surface brightness is composed of 3 components: 1) the  
1218: innermost central region with centrally peaked brightness, 2) smooth  
1219: component, and 3)  northeastern and  southwestern shell  components.  
1220: The observed surface brightness distribution in the SNR interior  
1221: implies a radial density profile  described by $n/n_c = 1 - r/R$, where  
1222: $n(r)$ is plasma density as a function of distance $r$ from the  
1223: remnant's center, $n_c$ is the central density,  and $R$ is the  
1224: remnant's radius. This profile is steeper than profiles usually found  
1225: in other mixed-morphology SNRs. The overall shape of the remnant is far  
1226: from uniform and circular, typical for supernova remnants; it is dented  
1227: in the north and east because of the presence of  molecular clouds.  
1228:   
1229: 
1230:   
1231: 2.   
1232: The ASCA spectra show emission lines of  
1233: Ne, Mg, Si, and Fe K$\alpha$, and continuum at least up to 7 keV.  
1234: ASCA/ROSAT spectra reveal spectral variations across the remnant:  
1235: while the southwestern and northeastern shells are fit well by  
1236: a one-temperature model, the central emission requires a two-temperature  
1237: model with hot and cold components.  The southwestern shell can be fit well  
1238: by the plane shock model with a temperature of 1.5 keV,  
1239: an ionization timescale of $1.5 \times 10^{11}$ cm$^{-3}$ s,    
1240: and a line-of-sight absorption of   
1241: $1.2 \times 10^{22}$ cm$^{-2}$. The  
1242: northeastern shell has a lower temperature of $0.56$ keV,  
1243: a longer ionization timescale of $1.7 \times 10^{13}$  
1244: cm$^{-3}$ s, and $N_H = 0.8 \times 10^{22}$ cm$^{-2}$.  Unlike for the  
1245: southwestern and northeastern shells, the central emission cannot be  
1246: fit well by a single temperature model. Two components are required,  
1247: a hot temperature component with  
1248: $kT_{h} = 1.8 \pm0.5$ keV and a long ionization timescale  
1249: of $5 \times 10^{11}$ cm$^{-3}$ s, and a low  
1250: temperature component with $kT_{l} = 0.6$ keV, $n_et = 2 \times 10^{12}$  
1251: cm$^{-3}$ s. Note that with the possible exception of the SW shell, the long  
1252: derived ionization timescales imply that the gas is close to the ionization   
1253: equilibrium.  
1254: The X-ray luminosity of W28 is $6 \times 10^{34}$ ergs s$^{-1}$  
1255: (this includes only regions observed by ASCA/SIS),    
1256: and the estimated total X-ray mass is rather small,   
1257: only $\sim 20 - 25 M_\odot$, much less than expected from standard SNR models.  
1258: 
1259:   
1260:   
1261: 3. Observations at wavelength bands other than X-rays strongly suggest  
1262: that W28 is in a radiative stage of evolution, at least in the northern half  
1263: of the remnant where the interaction with the molecular clouds has been  
1264: detected. This remnant then belongs to a class of SNRs considered by  
1265: Chevalier (1999), with a radiative shell interacting with clumpy molecular   
1266: clouds, in which X-ray emission in the center is a ``fossil'' radiation.   
1267: The properties of X-ray emitting gas are broadly consistent with this scenario,  
1268: but when compared with standard spherically-symmetric SNR models   
1269: in a uniform ambient  
1270: medium, the gas is hotter, there is less X-ray emitting gas than expected,  
1271: and of course the centrally-filled morphology cannot be explained by such  
1272: models. The material which we are observing at the center  
1273: was likely shocked early  
1274: in the evolution of the remnant, and its temperature and density may have  
1275: been modified since then by processes such as evaporation, electron thermal  
1276: conduction, and mixing induced by various hydrodynamical instabilities.  
1277: But the evaporation model of White \& Long (1991) is in conflict with  
1278: observations, and efficient thermal conduction might not be consistent  
1279: with large temperature variations detected in W28. Both evaporation and  
1280: thermal conduction might be less efficient than assumed in these idealized  
1281: models, while the clumpy nature of molecular clouds plays an uncertain  
1282: but perhaps even a crucial role in determining X-ray morphology and spectra.  
1283:   
1284: 4. W28 differs from other known MM SNRs by: 1) the presence of hot (1.5 keV)  
1285: gas, 2) spectral variations are clearly present, the SW shell is much hotter  
1286: than the NE shell, and a mixture of hot and cold gas is present at the  
1287: center, 3) X-ray emission is more   
1288: knotty and clumpy than in other MM SNRs, and partial shell-like structures  
1289: are present. The large scale temperature gradient seems to to be related  
1290: to the ambient ISM density, with the cooler gas associated   
1291: with the molecular clouds and the hotter gas with the fainter   
1292: breakout region.   
1293: The separation between X-ray emitting gas and radiative shocks emitting in  
1294: infrared and optical is smaller than in other MM SNRs, more  
1295: similar to shell-like SNRs.    
1296:    
1297: 5. We report a marginal excess of X-ray emission in the ROSAT image   
1298: at the position of   
1299: the  pulsar, PSR B1758-23,  which is located at the northern  boundary  
1300: of W28. The count rate implies a luminosity of $1.2 \times 10^{32}$ ergs  
1301: s$^{-1}$, assuming  a line-of-sight absorption of $8 \times 10^{21}$  
1302: cm$^{-2}$ and a power law spectrum with $\Gamma = 2$ (equal to the  
1303: spectral index of  the  Crab Nebula). In addition to the pulsar, two  
1304: isolated X-ray sources appear a few arcminutes away from the  
1305: eastern boundary. The northern source is a point source, likely a  
1306: foreground star with no counterpart at other wavelengths. The  
1307: southern source which appears to be extended, can be identified  
1308: with an HII region.     
1309:   
1310:         
1311:   
1312:   
1313: \acknowledgements  
1314: We thank Robert Petre for helpful discussion about  the earlier version  
1315: of the paper, and Ho-Gyu Lee for help in generating ASCA mosaiced images.   
1316: We also acknowledge very helpful comments and suggestions by an anonymous   
1317: referee.  
1318: This research has made use of data obtained from the High Energy   
1319: Astrophysics Science Archive  
1320: Research Center (HEASARC), provided by NASA's Goddard Space Flight Center.   
1321: J. R. acknowledge the support  
1322: of the Jet Propulsion Laboratory, California Institute of Technology,  
1323: which is operated under contract with NASA.  
1324:   
1325:   
1326:   
1327:   
1328:   
1329:    
1330: \def\Mdot{\dot M}  
1331:   
1332:   
1333: \begin{table}  
1334: \caption{ROSAT and ASCA Observations of W28 (G6.4-0.1).}  
1335: \label{w28obser}  
1336: \begin{center}  
1337: \begin{tabular}{lccccccc}  
1338: \\  
1339: \hline \hline  
1340:  &pointing &Sequence No. &observation& exposure &\multicolumn{2}{c}{center} \\  
1341:  &         &       & date      &time (s)     &R.A. &Dec.\\  
1342: \hline  
1343:  ROSAT  \span\omit  \span\omit  \span\omit && \\  
1344: %   &   G & common name & source&date& exptime& center& quality \\  
1345: &center&wp500129 &Mar5, 92& 4224 &18:00:49 & -23:22:33  \\  
1346:  &NE& rp500236&Sep9-10, 93&10,476&18:01:12 & -23:25:60 \\  
1347:  &SW& rp500237&Sep 9, 93&9100 &17:59:59 & -23:31:60 \\  
1348: \hline  
1349: ASCA \span\omit  \span\omit  \span\omit &&  \\  
1350:  & NE  & ad51014000  & Mar4, 94 & 28,000 & 18:01:06.00 & -23:13:51.6 \\   
1351:  & center1 & ad51022000 &Mar4-Apr4, 94,&21,000 & 18:00:41.76 & -23:26:36.6 \\   
1352:  & SW     & ad54003000 & Mar31, 96 &11,500 & 17:59:18.0   & -23:45:56.5 \\  
1353:  & center2 & ad54003010 & Mar31, 96 &10,500 & 18:00:22.8 &-23:20:18.0 \\  
1354: \hline  
1355: \end{tabular}  
1356: \end{center}  
1357: \end{table}  
1358:   
1359: \begin{table*}  
1360: \caption{Spectral Fits and Inferred Physical Parameters}  
1361: \label{w28ascafit}  
1362: \begin{center}  
1363: \begin{tabular}{lcccccccccc}  
1364: \tableline \tableline \tableline  
1365: Region     &   
1366: \multicolumn{3}{c|}{Center} &  \multicolumn{2}{c|}{SW shell} & \multicolumn{1}{c|}{NE shell} \\  
1367: \hline  
1368: Model     & A\tablenotemark{a} & B\tablenotemark{b} &C\tablenotemark{c} &D\tablenotemark{d} & E\tablenotemark{e} & F\tablenotemark{f} \\  
1369: \tableline  
1370: $\Delta\chi^2$$^g$ & 1.4 &1.1 & 1.1 &  0.96 & 0.93 & 1.01\\  
1371: $N_H$ ($10^{22}$cm$^{-2}$) &  0.73$^{+0.04}_{-0.06}$  & 0.66$\pm$0.06 & 0.79$\pm$0.15 & 1.28$\pm$0.25 & 1.26$^{+0.24}_{-0.16}$ & 0.77$\pm$0.08 \\  
1372: \hline \hline  
1373: \multicolumn{7}{l}{Low temperature component }  \\  
1374: \hline  
1375: $kT_l$ (keV)  & 0.58$^{+0.04}_{-0.07}$ & 0.59$\pm$0.05 & 0.35$\pm$0.15 & 1.2 ($>$0.8) & 1.5$\pm$0.2 & 0.56$\pm$0.06\\  
1376: %$kT_s$ (keV) &      &      & 0.38 &   1.2 ($>$0.8)    &  \\  
1377: $\tau_l$ (10$^{12}$ cm$^{-3}$ s) & -- & 2.0 ($>$1) & 1.8 ($^{+2.7}_{-0.8}$) & 0.36$^{+0.20}_{-0.16}$ & 0.15$^{+0.17}_{-0.09}$  & 17 ($>$7) \\  
1378: Abundance (fixed) &\multicolumn{3}{c}{2/3} & \multicolumn{2}{c}{2/3} & \multicolumn{1}{c}{2/3}\\  
1379: $EM_l/(4\pi d^2)$ (10$^{12}$ cm$^{-5}$)  & 4-5.3 & 3.5-4.6 & 6.3-8.5 &0.8-1 & 0.9-1 & 1.9-2.3\\  
1380: \hline \hline  
1381: \multicolumn{7}{l}{High temperature component }\\  
1382: \hline  
1383: $kT_h$ (keV) (fixed$^h$)  &\multicolumn{3}{c}{1.8($^{+0.7}_{-0.5}$)} &\multicolumn{2}{c}{--}& \multicolumn{1}{c}{--}\\  
1384: $\tau_h$ (10$^{12}$ cm$^{-3}$s) (fixed$^h$)  &  \multicolumn{3}{c}{0.5($^{+0.50}_{-0.43}$)}&\multicolumn{2}{c}{--} & \multicolumn{1}{c}{--}\\  
1385: %Abundance (fixed) & \multicolumn{3}{c|}{1} & \multicolumn{3}{c|}{1}&\multicolumn{2}{c|}{1} \\  
1386: Fe (fixed) & \multicolumn{3}{c}{1.1($^{+0.9}_{-0.6}$)}  &\multicolumn{2}{c}{--} &\multicolumn{1}{c}{--}\\  
1387: $EM_h/(4\pi d^2)$ (10$^{12}$ cm$^{-5}$)  & \multicolumn{3}{c}{0.65-0.8} &\multicolumn{2}{c}{--} &\multicolumn{1}{c}{--}\\  
1388: \hline \hline  
1389: \multicolumn{2}{l}{ Angular Diameter} & \multicolumn{4}{c}{50$'$$\times$45$'$}\\  
1390: \multicolumn{1}{l}{ Shell Radius} & \multicolumn{3}{c}{--} &  
1391: \multicolumn{2}{c}{28$'$ (14.5$\times$d$_{1.8kpc}$ pc)$^i$}  
1392: & \multicolumn{2}{c}{18$'$ (9.5$\times$d$_{1.8kpc}$ pc)$^j$ } \\  
1393: \hline \hline  
1394: $n_e$ (cm$^{-3}$, from EM) &  
1395: \multicolumn{3}{c}{$0.48(d_{1.8kpc} f)^{-1/2}$, $0.19(d_{1.8kpc} f)^{-1/2}$} &   
1396: \multicolumn{2}{c}{$0.22(d_{1.8kpc} f)^{-1/2}$} & \multicolumn{1}{c}{$0.65(d_{1.8kpc} f)^{-1/2}$ }\\  
1397: Flux (ergs s$^{-1}$ cm$^{-2}$)  &  
1398: \multicolumn{3}{c}{8$\times 10^{-11}$}  &  
1399:  \multicolumn{2}{c}{4$\times 10^{-11}$} & \multicolumn{1}{c}{4.5$\times 10^{-11}$  }\\  
1400: Luminosity (ergs s$^{-1}$) &  
1401: \multicolumn{3}{c}{2.9$\times10^{34}$$d_{1.8kpc}^2$ }  &   
1402: \multicolumn{2}{c}{1.45$\times 10^{34 }$$d_{1.8kpc}^2$ } &  
1403: \multicolumn{1}{c}{1.63$\times 10^{34}$$d_{1.8kpc}^2$}     \\  
1404: \hline \hline  
1405: \tableline  
1406: \end{tabular}  
1407: \end{center}  
1408: \tablenotetext{a}{$mekal+nei$}  
1409: \tablenotetext{b}{$nei+nei$}  
1410: \tablenotetext{c}{$sedov+nei$}  
1411: \tablenotetext{d}{$sedov$}  
1412: \tablenotetext{e}{$pshock$}  
1413: \tablenotetext{f}{$nei$}  
1414: \tablenotetext{g}{$\Delta\chi^2$ is reduced $\chi^2$, and typical degree of freedom is $\sim$650. The errors
1415: of the best fits for each region with 99\% confidence are given.}  
1416: \tablenotetext{h}{$kT_h$  and the Fe abundance were determined from fits to the high energy   
1417:   (4-10 keV) spectra alone, and $\tau_h$ is determined using the Fe K$\alpha$ line centroid.}  
1418: \tablenotetext{i,j}{Measured from the center to the SW and NW shells, respectively.}  
1419: %\tablenotetext{j}{The radius is from the center to the NE shell}  
1420: \end{table*}  
1421: \noindent  
1422:   
1423: \small  
1424:   
1425:   
1426: 
1427: \begin{thebibliography}{}  
1428:   
1429: \def\pp{\parshape 2 0truecm 23truecm 0.5truecm 22.5truecm}  
1430: \def\jref#1;#2;#3;#4;#5 {\par\noindent \pp#1 #2, {\it #3} {\bf #4}, #5. \par}  
1431: % author (date); title; journal; volume; page  
1432: %\tenrm  
1433:   
1434:   
1435:   
1436: \bibitem[]{} Andrews, M. D., Basart, J. P., Lamb, R. C., \& Becker,  
1437: R. H. 1983, ApJ, 266, 684  
1438:   
1439: \bibitem[]{} Arikawa, Y., Tatematsu, K., Sekimoto, Y., \& Takahashi, T.   
1440: 1999, PASJ, 51, L7  
1441:   
1442: \bibitem[]{}  Asaoka, I., \& Aschenbach, B. 1994, A\&A, 284, 573  
1443: %\bibitem[]{}Blitz, L., 1987,  
1444: %{\it Physical processes in interstellar clouds},  
1445: % Morfill and Scholer (ed.), p437  
1446:   
1447: \bibitem[]{} Bamba, A., Yokogawa, J., Sakano, M., \& Koyama, K. 2000, PASJ,  
1448: 52, 259  
1449:   
1450: \bibitem[]{} Blondin, J. M., Wright, E. B., Borkowski, K. J., \& Reynolds, S.  
1451: P. 1998, ApJ, 500, 342  
1452:   
1453: \bibitem[]{} Borkowski, K., Lyerly, W. J., \& Reynolds, S. P. 2001, ApJ, 548, 820  
1454:   
1455: \bibitem[Borkowski \& Szymkowiak 1997]{bz97} Borkowski, K. J., \& Szymkowiak,   
1456: A. E. 1997, ApJ, 477, L49  
1457:   
1458: \bibitem[]{} Bohigas, J., Ruiz, M. T., Carrasco, L., Salas, L., \& Herrera,   
1459: M. A. 1983, RMxAA, 8, 155  
1460:   
1461: \bibitem[]{} Chen, Y., \& Slane, P. O. 2001, ApJ, 563, 202  
1462:   
1463: \bibitem[]{} Chevalier, R. A. 1974, ApJ, 188, 501  
1464:   
1465: \bibitem[]{} Chevalier, R. A. 1999, ApJ, 511, 798  
1466:   
1467: \bibitem[]{} Clark, D. H., \& Caswell, J. L. 1976, MNRAS, 174, 267  
1468:   
1469: \bibitem[]{} Claussen, M. J., Goss, W. M., Frail, D. A., \& Desai, K. 1999,   
1470: ApJ, 522, 349   
1471:   
1472: \bibitem[]{} Clemens, D. P. 1985, ApJ, 295, 422  
1473:   
1474: \bibitem[]{} Cline, T., et al. 2000, ApJ, 531, 407  
1475:   
1476: \bibitem[]{} Cox, D. P., et al. 1999, ApJ, 524, 179   
1477:   
1478: \bibitem[]{} Cui, W., \& Cox, D. P. 1992, ApJ, 401, 206  
1479:   
1480: %\bibitem[]{} DeNoyer, L. K. 1983, ApJ, 264, 141  
1481:   
1482: \bibitem[]{} Dubner, G. M., Vel\'{a}zquez, Goss, W. M., \& Holdaway, M. A.   
1483: 2000, AJ, 120, 1933  
1484:   
1485: \bibitem[]{}Esposito, J.A., Hunter, S.D., Kanbach, G., \& Sreekumar, P., 1996, ApJ, 461, 820
1486: 
1487: \bibitem[]{} Falle, S. A. E. G. 1981, MNRAS, 195, 1011  
1488:   
1489: \bibitem[]{}Frail, D. A., Goss, W. M., \& Slysh, V. I. 1994, ApJ, 424, L111  
1490:   
1491: \bibitem[]{}Frail, D. A., Kassim, N. E., \& Weiler, K. W. 1994, AJ, 107, 111  
1492:   
1493: \bibitem[]{}Frail, D. A., Kulkarni, S. R., \& Vasisht, G. 1993, Nature,   
1494: 365, 136  
1495:   
1496: \bibitem[]{}Frail, D. A., \& Mitchell, G. F. 1998, ApJ, 508, 690  
1497:   
1498: \bibitem[]{} Goudis, C. 1976, Ap\&SS, 45, 133  
1499:   
1500: \bibitem[]{} Hamilton, A. E. S., Chevalier, R. A., \& Sarazin, C. L. 1983,  
1501: ApJS, 51, 115  
1502:   
1503: \bibitem[]{}Harrus, I. M, Hughes, J. P., Singh, K. P., Koyama, K., \&   
1504: Asaoka, I.  
1505: 1997, ApJ, 488, 781  
1506:   
1507: \bibitem[]{} Hellsten, U., \& Sommer-Larsen, J. 1995, ApJ, 453, 264  
1508:   
1509: \bibitem[]{} Ikebe, Y., et al. 1998, ASCA Newsletter No. 3   
1510:   
1511: \bibitem[]{} Jones, L. R. 1991, in Iron Line Diagnostics in X-ray Sources, eds.  
1512: A.Treves, G.C.Perola, \& L.Stella, (Springer-Verlag: Berlin: Heidelberg:   
1513: New York), 45  
1514:   
1515: \bibitem[]{} Jones, T. W., et al., 1998, PASP, 110, 125  
1516:   
1517: \bibitem[]{}Kaastra, J. S. 1992, An X-ray Spectral Code for Optically  
1518: Thin Plasmas (SRON-Leiden Report)  
1519:   
1520: \bibitem[]{} Kaspi, V. M., Lyne, A. G., Manchester, R. N., Johnston, S.,  
1521: D'Amico, N., \& Shemar, S. L. 1993, ApJ, 409, L57  
1522:   
1523: \bibitem[]{} Kassim, N. E. 1992, AJ, 103, 943  
1524:   
1525: \bibitem[]{} Koo, B.-C., \& Heiles, C. 1991, ApJ, 382, 204  
1526:   
1527: \bibitem[]{}  Kundu, M. R., \& Velusamy, T. 1972, A\&A, 20, 237  
1528:   
1529: \bibitem[]{} Liedahl, D. A., Osterheld, A. L., \& Goldstein, W. H.   
1530: 1995, ApJ, 438, L115  
1531:   
1532: \bibitem[]{} Lee, H.-G., Rho, J., Koo, B.-C., Petre, R., \& Decourchelle, A.,   
1533: 2001, BAAS, 198, 4002L  
1534:   
1535: \bibitem[]{} Long, K., Blair, W. P., Matsui, Y., \& White, R. L. 1991, ApJ,   
1536: 373, 567  
1537:   
1538: \bibitem[]{} Lozinskaya, T. A. 1974, SvA, 17, 603  
1539:   
1540: \bibitem[]{} Lozinskaya, T. A. 1992, Supernovae and Stellar Wind   
1541: in the Interstellar Medium, (American Institute of Physics: New York)  
1542:   
1543: \bibitem[]{}Maeda, Y., et al., 2001, ApJ, in press (astro-ph/0102183)  
1544: \bibitem[]{} McKee, C. F. 1982, in Supernovae: A Survey of Current Research,   
1545: (Reidel: Dordrecht), 433  
1546:   
1547: \bibitem[]{} Milne, D. K. 1979, Austr. J. Phys., 32, 83  
1548:   
1549: \bibitem[]{} Milne, D. K. \& Dickel., J. R. 1975, Austr. J. Phys., 28, 209   
1550:   
1551: \bibitem[]{} Ormes, J. F., Ozel, M. E., \& Morris, D.J. 1988, ApJ, 334, 722  
1552:   
1553: \bibitem[]{}Rho, J.-H., 1995, Ph.D. thesis, University of Maryland  
1554:   
1555: \bibitem[]{} Rho, J., Corcoran, M. F., Chu, Y.-H., \& Reach, W. T. 2001,  
1556: ApJ, in press  
1557:   
1558: \bibitem[]{} Rho, J., \& Petre, R. 1996, ApJ, 467, 698  
1559:   
1560: \bibitem[]{} Rho, J., \& Petre, R. 1998, ApJ, 503, L167  
1561: \bibitem[]{}Rho, J., Petre, R., \& Hester, J. J., 1994, The soft X-ray Cosmos, eds. E.M.Schlegel \& R.Petre, (American Institute of Physics: New York), 318  
1562: \bibitem[]{} Rho, J.-H., Petre, R., Schlegel, E. M., \& Hester, J. J. 1994,  
1563: ApJ, 430, 757  
1564: \bibitem[]{} Rho, J., et al. 1998, BAAS, 193, 740   
1565: \bibitem[]{} Reach, W. T., \& Rho, J. 1999, ApJ, 511, 836  
1566:   
1567: \bibitem[]{} Reach, W. T., \& Rho, J. 2000, ApJ, 544, 843  
1568: \bibitem[]{} Shaver, P. A. \& Goss, W. M. 1970, Aust. J. Phy. Astr. Sup.,  
1569: 14, 77  
1570:   
1571: \bibitem[]{} Saken, J. M., Long, K. S., Blair, W. P., \& Winkler, P. F.   
1572: 1995, ApJ, 443, 231  
1573: \bibitem[]{}Seward, F. D. 1985, Comm. Astroph. XI, 1, 15  
1574:   
1575: \bibitem[]{} Seward, F. D. 1990, ApJS, 73, 781   
1576:   
1577: \bibitem[]{} Shelton, R., et al.  1999, ApJ, 524, 192   
1578: %\bibitem[]{} Tatematsu, K., Nakano, M., Yoshida, S., Wiramihardja, S.D. and   
1579: %Kogure, T.;1985;PASJ;37;345  
1580:   
1581: \bibitem[]{} Snowden, S. L., McCammon, D., Burrows, D. N., \& Mendenhall, J.A.  
1582: 1994, ApJ, 424, 714  
1583:    
1584: %\jref Tatematsu, K., Fukui, Y., Nakano, M., Kogure, T, Ogawa, H., and  
1585: %Kawabata, K.;1987;A\&A;184;279   
1586:   
1587:   
1588:   
1589: \bibitem[Tanaka et al. 1994]{Tanaka} Tanaka, Y., Inoue, H., \& Holt,  
1590: S. S. 1994, PASJ, 46, L37  
1591:   
1592: \bibitem[]{} Torii, K., Tsunemi, H., Kaneda, H., \& Slane, P.  1993, in  
1593: UV and X-ray spectroscopy of Astrophysical and Laboratory Plasmas,  
1594: eds. K. Yamashita \& T. Watanabe, (Universal Academy Press: Tokyo), 291  
1595:   
1596: \bibitem[]{} van den Bergh, S.  1978, ApJS, 38, 119  
1597:   
1598: \bibitem[]{}Venger, A. P., et al., 1982, SvA, 26, 12  
1599:   
1600: \bibitem[]{} White, R. L., \& Long, K. S. 1991,  ApJ, 373, 543  
1601:   
1602: \bibitem[]{} Wootten, A. 1981, ApJ, 245, 105  
1603:   
1604: \bibitem[]{} Yoshita, K., Tsunemi, H., Miyata, E., \& Mori, K. 2001, PASJ,  
1605: 53, 93  
1606:   
1607: %\end{references}  
1608: \end{thebibliography}  
1609:   
1610:   
1611: \clearpage  
1612:   
1613: \begin{figure}  
1614: %Fig.   
1615: \caption{The mosaiced ROSAT PSPC image (0.5-2.4 keV) of W28.  
1616: The grey-scale ranges from 0.0008 to 0.029   
1617: counts s$^{-1}$ arcmin$^{-2}$.}  
1618:   
1619: \label{w28pspc}  
1620: %\clearpage  
1621: \end{figure}  
1622:   
1623: %Fig. 2  
1624: \begin{figure}  
1625: \caption{ASCA GIS fields of view (four pointings)  
1626: superposed on the ROSAT image.}  
1627: \label{w28ascafov}  
1628: \end{figure}  
1629:   
1630: %fig. 3  
1631: \begin{figure}  
1632: \caption{ASCA greyscale image (0.5-10 keV) and PSPC contours. The grey-scale image  
1633: range is  (1.0 - 5.5) $\times 10^{-5}$ counts s$^{-1}$ arcmin$^{-2}$.}  
1634: \label{w28ascaall}  
1635: \end{figure}  
1636:   
1637:   
1638: %Fig. 4   
1639: \begin{figure}  
1640: \caption{ Spectral extraction regions (solid circle and ellipses) superposed on the PSPC image.  
1641: Right "ear-like" structure and northern flat boundary are also marked (dashed lines). }  
1642: \label{w28regionmark}  
1643: \end{figure}  
1644:   
1645: \begin{figure}  
1646: %Fig. 5   
1647: \caption{The PSPC greyscale image superposed on  
1648: contours of the 328 MHz radio continuum image (from Dubner et al. 2000): the radio  
1649: morphology is shell-like and X-rays are centrally peaked.}  
1650: \label{w28xrayradio}  
1651: \end{figure}  
1652:   
1653: %Fig. 6  
1654: \begin{figure}  
1655: \caption{Soft ($E < 0.5$ keV) PSPC X-ray image superposed on PSPC total  
1656: brightness contours. The greyscale  
1657: is from 0.0008 to 0.0043 counts s$^{-1}$ arcmin$^{-2}$.}   
1658: \label{w28soft}  
1659: \end{figure}  
1660:    
1661:   
1662: %Fig. 7  
1663: \begin{figure}  
1664: \caption{The PSPC spectral hardness ratio (0.9--2.2 keV/0.3--0.9 keV) image   
1665: with a greyscale range from 1.1 to 1.4,   
1666: superposed on PSPC total brightness contours.   
1667: The southern half of the  
1668: remnant is harder than the northern half.}  
1669: \label{w28shr}  
1670: \end{figure}  
1671:   
1672: %Fig. 8  
1673: \begin{figure}  
1674: \caption{  
1675:  Hard (3--10 keV) ASCA image superposed on PSPC contours.  
1676: The southwestern shell and center are harder than the northeast.  
1677: The grey-scale image  
1678: range is  (1 - 3) $\times 10^{-5}$ counts s$^{-1}$ arcmin$^{-2}$.}  
1679: \label{w28ascahard}  
1680: \end{figure}  
1681:        
1682:   
1683:   
1684: %Fig. 9  
1685: \begin{figure}  
1686: \caption{X-ray ({\it solid line}) and radio ({\it dotted line}) surface   
1687: brightness profiles of W28 in the north (a), southwest (b),  
1688: southeast (c), and northeast (d).  Position angles are marked in each panel.   
1689: The X-ray surface brightness is centrally peaked, but there are also shell  
1690: structures at about 20$'$ from the center in the southwest and  
1691: northeast.  
1692: The dashed line in the northern sector corresponds to a radial density   
1693: profile described in the text. }  
1694: \label{w28sb}  
1695: \end{figure}   
1696:   
1697:   
1698:   
1699: %fig  10.  
1700: \begin{figure}  
1701: \caption{Comparison of W28 spectra for various regions. The best fits
1702: are shown, which are derived using 5 spectra from SIS0, SIS1, GIS2, GIS2
1703: and ROSAT PSPC.
1704: (a) The central ({\it crosses}) and SW ({\it squares}) SIS0 spectra. Soft component  
1705: is not present in the SW.  (b) ASCA SIS spectra of the  
1706: northeastern shell.  The best-fit NEI model with  
1707: $kT_e = 0.56$ keV, $N_H = 0.8 \times 10^{22}$ cm$^{-2}$,  and  
1708: the residuals are  shown.  
1709: (c) The ASCA SIS and ROSAT/PSPC spectra of W28 for the  
1710: central region showing hard emission and the Fe K$\alpha$ line.   
1711: A two-temperature NEI model with $kT_{h} = 1.8\pm0.5$ keV and  
1712: $kT_{l} = 0.6$ keV is also shown.  
1713: (d) Two-temperature model of the central spectrum (only SIS0 spectrum is shown) with the hot and cold   
1714: components indicated (total -- {\it solid line}, cold -- {\it dash-dotted},  
1715: hot -- {\it dotted}).  
1716: }   
1717: \label{w28spec}  
1718: \end{figure}  
1719:   
1720:   
1721:   
1722: %Fig. 11   
1723: \begin{figure}  
1724: \caption{ A portion of the PSPC image with the pulsar position marked. Its  
1725: position is at least 30$''$ away from a weak X-ray peak. The ellipse is  
1726: an error box of the pulsar position. The diffuse emission is  northern
1727: part of W28.}  
1728: \label{w28pulsar}  
1729: \end{figure}  
1730:   
1731: 
1732:   
1733:   
1734: 
1735:   
1736: \clearpage  
1737: %\plotone{f1color.ps}  
1738: {Fig. 1,2,3 are  jpeg files.} 
1739: 
1740: %\clearpage
1741:   
1742: %\psfig{figure=f2.ps,width=10truecm}  
1743: %{Fig. 2}  
1744: 
1745: %\clearpage
1746: 
1747: %\psfig{figure=f3.ps}  
1748: %{Fig. 3}  
1749:   
1750:   
1751: \psfig{figure=f4.ps}  
1752: \vskip 13truept  
1753: {Fig. 4}  
1754:   
1755: \clearpage  
1756: {Fig. 5,6,7, and 8 are jpg files.}
1757: %\psfig{figure=f5color.ps}  
1758: %\vskip 13truept  
1759: %{Fig. 5}  
1760: %\clearpage  
1761: %  
1762: %\plotone{f6.ps}  
1763: %\vskip 13truept  
1764: %{Fig. 6}  
1765: %  
1766: %
1767: %\psfig{figure=f7.ps,height=15truecm}  
1768: %\vskip 13truept  
1769: %{Fig. 7}  
1770: %  
1771: %  
1772: %\psfig{figure=f8.ps,height=10truecm}  
1773: %{Fig. 8}  
1774: %  
1775: \clearpage  
1776:   
1777: \vbox{  
1778:   
1779: \centerline{\hbox{  
1780: \psfig{figure=f9a.ps,width=8truecm,height=7truecm}  
1781: \psfig{figure=f9b.ps,width=8truecm,height=7truecm}  
1782: }}  
1783: \centerline{\hbox{  
1784: \psfig{figure=f9c.ps,width=8truecm,height=7truecm}  
1785: \psfig{figure=f9d.ps,width=8truecm,height=7truecm}  
1786: }}   
1787: }  
1788: \vskip 13truept  
1789: {Fig. 9}  
1790:   
1791:   
1792: \psfig{figure=f10a.ps,height=7truecm,width=13truecm,angle=270}  
1793: {Fig. 10a}  
1794:   
1795: \psfig{figure=f10b.ps,height=7truecm,width=13truecm,angle=270}  
1796: {Fig. 10b}  
1797:   
1798:   
1799: \clearpage  
1800:   
1801: \psfig{figure=f10c.ps,height=8truecm,width=13truecm,angle=270}  
1802: {Fig. 10c}  
1803: \vskip 1.5truecm  
1804: \psfig{figure=f10d.ps,width=15truecm,height=8truecm}  
1805: {Fig. 10d}  
1806:   
1807:   
1808: \clearpage  
1809: {Fig. 11 is a jpeg file.}   
1810: %\psfig{figure=f11.ps,width=10truecm}  
1811: %\vskip 13truept  
1812: %{Fig. 11}  
1813: %\clearpage  
1814:   
1815:   
1816:   
1817: \end{document}  
1818:   
1819: