The Rosat Hri X-Ray Survey of the Cygnus Loop

Dartmouth College Dartmouth Digital Commons

Open Dartmouth: Published works by Dartmouth faculty Faculty Work

1997

N. A. Levenson University of California, Berkeley

J R. Graham University of California, Berkeley

B. Aschenbach Max–Planck–Institut fur Extraterrestrische Physik

W. P. Blair The Johns Hopkins University

W. Brinkmann Max–Planck–Institut fur Extraterrestrische Physik

See next page for additional authors

Follow this and additional works at: https://digitalcommons.dartmouth.edu/facoa

Part of the Astrophysics and Astronomy Commons

Dartmouth Digital Commons Citation Levenson, N. A.; Graham, J R.; Aschenbach, B.; Blair, W. P.; Brinkmann, W.; Busser, J. U.; Egger, R.; Fesen, R. A.; Hester, J. J.; Kahn, S. M.; Klein, R. I.; McKee, C. F.; Petre, R.; Pisarski, R.; Raymond, J. C.; and Snowden, S. L., "The Rosat Hri X-Ray Survey of the Cygnus Loop" (1997). Open Dartmouth: Published works by Dartmouth faculty. 3450. https://digitalcommons.dartmouth.edu/facoa/3450

This Article is brought to you for free and open access by the Faculty Work at Dartmouth Digital Commons. It has been accepted for inclusion in Open Dartmouth: Published works by Dartmouth faculty by an authorized administrator of Dartmouth Digital Commons. For more information, please contact [email protected]. Authors N. A. Levenson, J R. Graham, B. Aschenbach, W. P. Blair, W. Brinkmann, J. U. Busser, R. Egger, R. A. Fesen, J. J. Hester, S. M. Kahn, R. I. Klein, C. F. McKee, R. Petre, R. Pisarski, J. C. Raymond, and S. L. Snowden

This article is available at Dartmouth Digital Commons: https://digitalcommons.dartmouth.edu/facoa/3450 THE ASTROPHYSICAL JOURNAL, 484:304È312, 1997 July 20 ( 1997. The American Astronomical Society. All rights reserved. Printed in U.S.A.

THE ROSAT HRI X-RAY SURVEY OF THE CYGNUS LOOP N. A.LEVENSON,1 J. R.GRAHAM,1 B.ASCHENBACH,2 W. P.BLAIR,3 W.BRINKMANN,2 J.-U. BUSSER,2 R.EGGER,2 R. A.FESEN,4 J. J.HESTER,5 S. M.KAHN,6 R. I.KLEIN,1,7 C. F. MCKEE,1,8 R.PETRE,9 R.PISARSKI,9 J. C. RAYMOND,10 AND S. L. SNOWDEN9 Received 1996 December 4; accepted 1997 February 11

ABSTRACT We describe and report progress on the joint U.S. and German campaign to map the X-ray emission from the entire Cygnus Loop with the ROSAT High Resolution Imager. The Cygnus Loop is the prototype for a supernova remnant that is dominated by interactions with the interstellar medium and supplies fundamental physical information on this basic mechanism for shaping the interstellar medium. The global view that these high-resolution (FWHM D 10A) observations provide emphasizes the inhomogeneity of the interstellar medium and the pivotal nature of cloudÈblast-wave interactions in determining the X-ray morphology of the supernova remnant. While investigating the details of the evolution of the blast wave, we also describe the interstellar medium in the vicinity of the Cygnus Loop, which the progenitor star has processed. Although we do not expect the X-ray observations to be complete until 1997 September, the incomplete data combined with deep Ha images provide deÐnitive evidence that the Cygnus Loop was formed by an explosion within a preexisting cavity. Subject headings: ISM: individual (Cygnus Loop) È supernova remnants È X-rays: ISM

1. INTRODUCTION wave interactions. The ROSAT HRI combines high 0.1È2.4 keV sensitivity (approximately three times better than the Supernova explosions are one of the most basic processes Einstein HRI;Zombeck et al. 1990), high spatial resolution that deÐne the physical and chemical state of the interstellar (FWHM B 6A on axis), and high contrast imaging (the half- medium (ISM). The interaction between supernova rem- energy radius is 3A on axis, with scattering less than a few nants (SNRs) and the ISM is complex and symbiotic. X-ray percent;Aschenbach 1988) to provide a powerful tool for observations are important in the study of SNRs because investigating cloudÈblast-wave interactions. thermal emission from hot material (kT D 0.1È1 keV) is the The region of the Milky Way toward Cygnus is viewed primary tracer of gas that the supernova remnant blast along the Carina-Cygnus arm and exhibits a complex array wave heats and accelerates. High-resolution X-ray observ- of bright and dark clouds on the Palomar Sky Survey. Most ations are important for studying SNR gas dynamics for remarkable are the arcs and loops of faint Ha nebulosity, three reasons. First, the length scale, , for cooling and l the smallest and brightest of which is the Cygnus Loop recombination behind a radiative shock is small: l B 2.5 (l 74.0, b 8.6). The Cygnus Loop is the prototype 10 v /n cm, wheren is the preshock number density \ \[ ] 17 s4 middle-aged SNR and is a particularly suitable candidate andv is7 the0 shock velocity0 in units of 100 km s (McKee s ~1 for a case study being nearby (770 pc;Minkowski 1958) and 1987).7Second, the preshock medium is not uniform, having bright with low foreground extinction [E(B V ) 0.1], clouds and substructure ranging over scales from tens of [ \ thus ensuring an unrivaled signal-to-noise ratio and spatial parsecs and downward. Third, gas-dynamical instabilities resolution. From an observational perspective, the Cygnus are important, and multidimensional hydrodynamic calcu- Loop is unique in terms of X-ray brightness, shock front lations predict that fully developed turbulence occurs when resolution (1A 4 1.1 10 cm), and interaction with a the blast wave collides with an interstellar cloud (Klein, ] 16 range of interstellar conditions. The Cygnus Loop is also at McKee, & Colella1994; Klein, McKee, & Woods 1995). sufficiently high galactic latitude that confusion with back- Consequently, structure is present down to the smallest ground emission from the plane of the galaxy is minimized. scales that can be resolved computationally. The improved Unlike other well-known younger SNRs such as Cas A, ROSAT HRI sensitivity and spatial resolution compared Tycho, and Kepler, the structure of the Cygnus Loop is with Einstein make it invaluable for exploring cloudÈblast- dominated by its interaction with its environment. The large diameter of the Cygnus Loop (3¡ 4 40 pc) assures that 1 Department of Astronomy, University of California, Berkeley, CA it is interacting with di†erent components of the ISM, from 94720. molecular gas on the northwest edge(Scoville et al. 1977), to 2 Max-Planck-Institut fu r Extraterrestrische Physik, Giessenbachs- trasse, D-85740 Garching, Germany. di†use atomic gas along the northeast and eastern limbs 3 Johns Hopkins University, Department of Physics and Astronomy, (DeNoyer 1975; Hester,Raymond, & Blair 1994), and low- Baltimore, MD 21218. density, hot, ionized gas to the south(Ku et al. 1984). Pre- 4 Department of Physics and Astronomy, Dartmouth College, vious studies of the Cygnus Loop have focused on Hanover, NH 03755. 5 Department of Physics and Astronomy, Arizona State University, individual regions selected for their brightness or morpho- Tempe, AZ 85287. logical peculiarity (e.g.,Hester & Cox 1986; Hester et al. 6 Departments of Astronomy and Physics, Columbia University, New 1994; Fesen,Kwitter, & Downes 1992; Graham et al. 1995; York, New York 10027. Levensonet al. 1996). The goal of this high-resolution inves- 7 Lawrence Livermore National Laboratory, Livermore, CA 94550. tigation(Graham & Aschenbach 1996) is to make an 8 Department of Physics, University of California, Berkeley, CA 94720. 9 Goddard Space Flight Center, Greenbelt, MD 20771. unbiased survey and thereby establish globally the state of 10 Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138. the medium into which the blast wave expands. A coherent 304 X-RAY SURVEY OF THE CYGNUS LOOP 305 picture emerges from our preliminary results(° 2). We therefore adopted the following observing strategy. We use present a uniÐed interpretation of the environment and the full 6A resolution to observe the brightest regions, which development of the Cygnus Loop(° 3), constrain the global restricts the e†ective Ðeld of view to 8@ radius. Each of these models that are relevant(° 4), and summarize the conclu- 24 pointings requires 15È20 ks to obtain 15 counts per pixel. sions in ° 5. In a comparable exposure time, the moderate surface brightness interior regions yield about 10 counts per 2. OBSERVATIONAL STRATEGY AND DATA REDUCTION 12A ] 12A pixel. At this resolution, the e†ective HRI Ðeld of Our primary goal is to map the entire Cygnus Loop view has a 12@ radius. We have observed these nine Ðelds for remnant at the highest spatial resolution that can be 20È25 ks. Observing the low surface brightness regions, we attained with the ROSAT HRI. Achieving this ambition is obtain 10 counts per cell only when using 15A or larger not simple, because the spatial resolution is limited by the pixels, for which the full HRI Ðeld of view is suitable. Each intrinsic instrumental properties and photon counting sta- of these 10 Ðelds requires a 35 ks exposure. Finally, one very tistics. The on-axis resolution is about 6A, but the o†-axis low surface brightness region along the western edge of the imaging performance is dominated by telescope aberra- ““ breakout ÏÏ requires a 50 ks exposure to obtain reliable tions, which increase with increasing Ðeld angle. At 20@, the counting statistics. The locations of the HRI pointings are edge of the useful Ðeld of view of the HRI, the resolution has shown on the Einstein IPC map (Fig. 1). degraded to about 30A. The large range of surface brightness A few Ðelds were observed during the initial operation of allows us the luxury of mapping the brightest regions at full ROSAT , and a concerted e†ort to make a complete map HRI resolution but forces us to seek a compromise between began during AO5, in 1994 October. Because of the large e†ective resolution and observing time in the dimmest total observing time this program requires, it is a collabo- regions. To give a concrete example, the surface brightness rative project between U.S. and German investigators. So observed with the ROSAT HRI yields a count rate of far, 771 ks out of a total 1066 ks of data have been obtained, 3 ] 10~5È1 ] 10~3 s~1 ina5A]5Apixel. For the brightest and we expect completion in AO7, by 1997 September. The regions along the shell (within a factor of 2 of the highest complete mosaic will consist of 47 Ðelds, of which 37 have surface brightness), a 15 ks exposure will net 15 counts and been observed at least in part. We present processed data 0.4 background counts. It would take longer than 450 ks to for 26 Ðelds, of which 15 are complete. The target names, obtain a similar number of counts in the faint interior. We Ðeld centers, and exposure times are listed inTable 1, and

32°

31° DEC (J2000) 30°

29°

20h 55m 50m 45m RA (J2000)

FIG. 1.ÈROSAT HRI pointings superposed on the Einstein IPC image of the Cygnus Loop(Seward 1990). Solid circles indicate Ðelds that are presented in this paper; dashed circles show unprocessed or proposed observations. 306 LEVENSON ET AL. Vol. 484

TABLE 1 ciency. These images have 5A pixels, matching the instru- HRI OBSERVATIONS mentÏs angular resolution. These maps from individual Ðelds are summed on a pixel-by-pixel basis into three TIME (ks) mosaics of the entire Cygnus Loop Ðeld. For the initial ad NAME (J2000) (J2000) Live Remaining analysis we present here, we use a uniform pixel scale and Ðeld of view from each observation. Only the central 17@ of WEST ...... 204541 310223 29.2 0 each observation was used in order to avoid the signiÐcant W RIM4...... 204553 312736 17.6 0 W RIM3...... 204648 314536 13.0 6.9 degradation of the point spread function and decreased effi- SW CLOUD ...... 204750 291800 46.4 0 ciency at large o†-axis angles, except for two Ðelds where a W RIM2...... 204755 320223 10.1 4.9 radius of 20@ was used to avoid a gap in the resulting image. NORTH ...... 205002 321123 16.0 0 The count rate map(Fig. 2 [Pl. 6]) has 15A pixels and is the E RIM1 ...... 205131 321912 16.5 0 di†erence of the total counts and the total background, CENTER ...... 205153 310712 11.9 13.1 SEINT2...... 205231 300636 25.7 9.3 divided by the net exposure(Fig. 3). Background subtrac- E RIM11...... 205250 292359 19.1 0 tion is not a problem for the bright and moderately bright E RIM2 ...... 205258 321836 10.2 4.8 Ðelds, but the background does become comparable with E RIM10...... 205331 294348 21.1 0 the source surface brightness for the lower surface bright- NEINT1...... 205341 315100 17.0 3.0 E RIM3 ...... 205429 321500 11.1 3.9 ness regions. The extent of the observations is demonstrated SEINT1...... 205434 302812 26.8 8.2 inFigure 4, where the individual image boundaries are E RIM9 ...... 205436 295659 14.8 0 drawn on the count rate map. NEINT2...... 205458 313223 14.2 5.8 To display the high-resolution information contained in NEINT4...... 205524 304012 18.6 0 the mosaic, we also show three smaller Ðelds that have been NEINT3...... 205534 310859 18.9 0 E RIM4 ...... 205541 315848 12.6 0 binned with 6A pixels (Figs.5a,6a,7 anda [Plates 7È12]). E RIM8 ...... 205558 300412 14.3 0 For comparison, we present aligned optical images of the SE CLOUD ...... 205622 302359 28.3 0 same regions taken through a narrow Ha Ðlter (Figs. 5b,6b, E RIM7 ...... 205631 304236 11.5 3.5 and7b). These were obtained using the Prime Focus Cor- E RIM5 ...... 205646 314311 11.9 3.1 EAST ...... 205717 310223 30.1 0 rector on the 0.8 m telescope at McDonald Observatory E RIM6 ...... 205729 312512 15.9 0 and are described inLevenson et al. (1997). After basic reduction consisting of bias subtraction and Ñat-Ðelding, W RIM6...... 204607 302548 0 20.0 individual images were remapped and averaged using the W RIM7...... 204702 300900 0 20.0 same technique that was applied to the X-ray observations. W RIM5...... 204723 312700 0 20.0 NWINT ...... 204728 312655 0 15.0 Figures5c,6c,7 andc are color versions of these combined W RIM8...... 204748 295100 0 20.0 data sets to show the relative spatial distributions directly. W INT1 ...... 204755 310000 0 35.0 W INT2 ...... 204805 303148 0 35.0 3. RESULTS W RIM9...... 204850 293748 0 20.0 W INT3 ...... 204855 301131 0 35.0 The X-ray morphology of an SNR is a primary diagnos- W RIM1...... 204904 321406 0 20.0 tic of the medium into which the blast wave propagates. At INT1...... 204934 312201 0 25.0 Ðrst sight, the X-ray morphology depicted in the ROSAT W INT4 ...... 205003 304930 0 35.0 HRI mosaic presents a bewilderingly complex view. This W INT5 ...... 205027 301828 0 35.0 shows prima facie that the Cygnus Loop is dominated by its N INT2...... 205028 314926 0 20.0 BREAK ...... 205041 291348 0 35.0 interaction with an inhomogeneous environment. In con- W INT6 ...... 205049 294431 0 35.0 trast, we note also that the Cygnus Loop appears nearly N INT1...... 205127 315732 0 15.0 circular, except for the breakout region to the south, which N INT3...... 205144 313816 0 20.0 suggests that the SNR expands in a uniform medium. Any SEINT3...... 205207 303448 0 35.0 NEINT5...... 205324 311112 0 25.0 model for the Cygnus Loop must resolve this apparent NEINT6...... 205341 305812 0 25.0 paradox. There is considerable large- and small-scale azimuthal NOTE.ÈUnits of right ascension are hours, minutes, and seconds, and variation around the perimeter of the Cygnus Loop and units of declination are degrees, arcminutes, and arcseconds. signiÐcant structure projected across its face. X-ray emission is bright near the rim, but the brightest peaks all lie the observations that are presented here appear Ðrst. interior to the circle that circumscribes the X-ray edge. The Column (4) contains the live time, which is the useful two most prominent X-ray regions, NGC 6960 on the portion of the observations, and column (5) lists the remain- western edge and NGC 6992 to the northeast, lie at the der of the requested time. largest projected distance inside this circle. These regions The data were processed and combined into a single contain complex networks of X-ray Ðlaments and high image using software designed to analyze extended sources surface brightness clumps. The emission here is strongly and the di†use background(Snowden et al. 1994). Over limb brightened. Figures89 and show radial cuts of the such large spatial scales, each observation is a distinct pro- emission from these regions. The limb brightening (i.e., the jection of the sky onto the plane of the detector, and these ratio of brightness at center to edge) is about a factor of 6 in must be remapped to a common projection. For each the northeast(Fig. 8) and 30 in the west (Fig. 9). The adia- observation, using only the accepted time intervals listed in batic expansion of a supernova blast wave in a uniform the raw data Ðles, three images are created. The Ðrst map medium results in some degree of limb brightened X-ray contains the total counts, the second map is a model of emission but is inadequate to account for these proÐles. We particle background, and the third is an exposure map, have calculated the X-ray surface brightness proÐle (Fig. 10) which includes the variations in the detector quantum effi- using the Sedov density proÐle(Sedov 1993), taking X-ray No. 1, 1997 X-RAY SURVEY OF THE CYGNUS LOOP 307

32°

31° DEC (J2000) 30°

29°

20h 55m 50m 45m RA (J2000)

FIG. 3.ÈExposure map of combined ROSAT HRI observations, including detector variations. The map is scaled linearly from 0 (white) to 82000 (black)s. volume emissivity to be proportional to density squared progress of the blast wave in neutral gas, [O III] emission and assuming that it is independent of temperature, which is reveals where postshock gas Ðrst starts to cool and recom- reasonable for the energy coverage of our observations bine, and [S II] and bright Ha together show where com- (Hamilton,Sarazin, & Chevalier 1983). The maximum limb plete cooling and recombination zones have developed. brightening is 3.3 and occurs near the edge of the SNR. The Utilizing X-ray and optical information together, however, clear message from this comparison is that the observed provides a more complete understanding of the structure X-ray proÐles in these bright regions of the Cygnus Loop and development of supernova remnant shocks. have peaks in the wrong place and are too great in contrast The observed correlation between X-ray and optical line to be the result of an adiabatic blast wave propagating in a emission falls into three broad categories that are distin- uniform medium. Thermal evaporation does not explain the guished by surface brightness and morphology: (1) regions X-ray proÐles either, as the evaporative solutions of White of high surface brightness X-ray emission distributed in & Long(1991) demonstrate (Fig. 10; see also Cowie, long coherent arcs that are aligned tangentially with, and lie McKee, & Ostriker1981). The e†ect of thermal conduction interior to, the bright optical Ðlaments; (2) localized inden- is to provide more mass farther behind the blast wave, so tations in the shell, typically less than 20@ in size, where these solutions always exhibit less extreme limb brightening bright X-ray emission occurs at large distances behind the than the adiabatic case. T he Cygnus L oop is neither a Sedov projected edge of the blast wave; and (3) regions of low remnant nor a Sedov remnant modiÐed by thermal evapo- surface brightness, limb-brightened X-rays bounded on the ration. outside by faint, narrow Balmer line Ðlaments. Examples of type 1 morphology are NGC 6992 in the 3.1. X-Ray and Optical Correlations northeast (a \ 20h56m, d \]32¡, J2000) and NGC 6960 on Optical line emission alone reveals some details of shock the western edge (20h46m, ]31¡). Type 2 regions are found structure. For example, faint Ha Ðlaments trace the on the eastern edge (20h57m, ]31¡), the southeast cloud 308 LEVENSON ET AL. Vol. 484

32°

31° DEC (J2000) 30°

29°

20h 55m 50m 45m RA (J2000)

FIG. 4.ÈThe boundaries of individual pointings are shown on the ROSAT HRI count rate mosaic of the Cygnus Loop. The sharp, nearly circular edge of the supernova remnant is real, not an artifact of the limited extent of the observations.

(20h56m, ]31¡), and the bright funnel-shaped section location of the current projected edge of the blast wave (20h50m, ]32¡) that is located about 30@ northeast of the because Balmer line Ðlaments occur where previously optical emission region known as the ““ carrot ÏÏ (Fesen, undisturbed atomic gas is shocked(Chevalier & Raymond Blair, & Kirshner1982). These regions tend to be more 1978; Chevalier,Kirshner, & Raymond 1980; Hester, turbulent and include the most compact and highest surface Danielson, & Raymond1986). The X-rays do not com- brightness knots within the Cygnus Loop. Categories 1 and pletely overlap the optical emission but are o†set to the 2 occur where the shock has encountered a greater than interior. We measure o†sets of 5AÈ20A between the exterior average column of gas, which has caused substantial decel- edge of X-ray emission and the edge of the Balmer Ðla- eration of the blast wave and associated enhancement of the ments; o†sets of 10A (equivalent to 1017 cm) are typical. The X-ray emissivity by a reÑected shock(Hester & Cox 1986; separation between the X-ray and optical emission is likely Hesteret al. 1994; Graham et al. 1995; Levenson et al. due to the time required to heat the gas behind the shock 1996). The di†erence between regions 1 and 2 may be due to front before X-rays are emitted. For a shock of velocity viewing geometry, the shape of the cloud that is being vs \ 400 km s~1, an o†set of 20A implies a heating time of encountered, and the time since enhanced density was about 200 yr. This is consistent with Coulomb collisions encountered by the shock. The western edge provides a alone equilibrating the postshock ion and electron tem- vivid example of a cloudÈblast-wave interaction because peratures(Spitzer 1962). If the blast wave has decelerated in this region is viewed almost exactly edge-on, minimizing the dense gas where the Balmer Ðlaments are detected, the same complications of projection e†ects(Levenson et al. 1996). separation from the X-ray edge implies that a longer time The best examples of type 3 Ðlaments occur at the extreme has elapsed, so Coulomb equilibration alone is still a pos- north (20h54m, ]32¡), to the south of the southeast knot, sible heating mechanism. Whether Coulomb collisions or and to the north of NGC 6960. These Ðlaments show the some other mechanism heat the postshock electrons, long No. 1, 1997 X-RAY SURVEY OF THE CYGNUS LOOP 309

0.15 3.0 −2

2.5

arcmin 0.10 −1

2.0

0.05 HRI counts s X−ray emission measure 1.5

0.00 1.0 0.80 0.90 1.00 0.0 0.2 0.4 0.6 0.8 1.0

R/Rsnr R/Rsnr

FIG. 8.ÈHRI surface brightness as a function of radius across the FIG. 10.ÈModel X-ray surface brightness proÐles for Sedov SNR (solid bright northeastern rim of the Cygnus Loop. This radial proÐle is an line) and evaporative cloud model ofWhite & Long (1991) (dotted line) azimuthal average over 2@. Very bright emission appears projected to the with their parameters C \ 10 and q \ 10. Both proÐles have been normal- interior of the SNR and extends over more than 12% of its radius. While ized to a uniform level at the center of the supernova remnant. there is no signiÐcant X-ray emission at the edge of the SNR, fainter X-ray limb brightening appears to the interior, atR \ 0.98R . (Balmer-dominated optical Ðlaments deÐne the edge of the SNR that issnr indicated in Figs. the blast wave are isolated and part of a random distribu- 8, 9,and 11.) This X-ray emission is clearly distinct from that due to tion or part of a coherent structure, such as an H II region expansion of a blast wave in a uniform medium (cf. Fig. 10). cavity or wind-blown bubble formed by the supernova progenitor. Kinematic observations(Kirshner & Taylor 1976) equilibration timescales are expected, asLaming et al. show an asymmetry in radial velocity measures, which (1996) demonstrate in analysis of a nonradiative shock motivated global models of the Cygnus Loop(Falle & observed in SN 1006. Garlick1982; Tenorio-Tagle, Rozyczka, & Yorke 1985) that describe an SNR breaking out of a molecular cloud, 4. A GLOBAL VIEW but these ignore the evidence for and e†ects of encounters with a distribution of clouds.Shull & HippeleinÏs (1991) 4.1. Cavity Explosion kinematic observations support a variation on these models Various X-ray studies have successfully discussed individ- in which the supernova explosion occurred in a cavity ual regions of the Cygnus Loop in terms of cloudÈblast- enclosed by a partial, hemispherical shell near a density wave collisions (e.g.,Hester & Cox 1986; Hester et al. 1994; discontinuity in the interstellar medium. On the contrary, Grahamet al. 1995; Levenson et al. 1996), and it is well these X-ray observations emphasize that over large scales established that large-scale collisions fundamentally deter- the blast wave is running into dense material, not expand- mine the morphology of portions of this supernova ing into a contrasting lower density region. remnant. However, the examination of particular regions With the data presented here, it is clear that the Cygnus does not lead to an understanding of the global environ- Loop is the result of a cavity explosion. The observations of ment in which the SNR evolves. SpeciÐcally, it is important the interior are consistent with blast wave propagation to determine whether the clouds that have been struck by through a rather smooth, low-density medium, and at the periphery, the edge of the cavity has been encountered recently. The limb-brightening of the ROSAT X-ray data show that the blast wave is currently decelerating in denser 0.20 materialÈdenser than the medium through which it previously propagatedÈover 80% of the perimeter that we

−2 have observed. Some of these enhancements are narrow, 0.15 tangential Ðlaments, while others are the large structures

arcmin that had been identiÐed in earlier observations. Such exten- −1 0.10 sive clumps of material cannot be characteristic of the pre- supernova ISM within the current remnant and thus distinguish the cavity boundary. We observe the cavity 0.05 walls in these data! Only two extended sites, near HRI counts s a \ 20h48m, d \]32¡10@, and a \ 20h57m, d \]30¡30@, show no evidence for blast wave interaction with the cavity 0.00 wall. The cavity walls are nearly complete, consisting of a 0.70 0.80 0.90 1.00

R/Rsnr neutral shell as well as large clouds. When the blast wave reaches the dense atomic shell, it decelerates. The Balmer- FIG. 9.ÈHRI surface brightness as a function of radius across the dominated Ðlaments directly trace the motion of the blast bright western rim of the Cygnus Loop, azimuthally averaged over 2@. Extremely bright emission appears projected to the interior of the SNR, at wave in the atomic gas of the shell. The X-ray signature of R \ 0.92R . SigniÐcant, though less extreme, limb brightening is also this category 3 morphology (see° 3.1) is enhancement at the observed atsnrR \ 0.98R . SNR edge, where gas is piled up behind the slowing shock. snr 310 LEVENSON ET AL. Vol. 484

Figures8, 9, and 11 provide examples of this characteristic mate, however, because the brightest Balmer Ðlaments were X-ray proÐle in three di†erent locations, where the enhance- analyzed, and these are the ones that have evolved most ment occurs at radii of0.98R , 0.98R , and 0.96R , toward the radiative stage.) The radius of the SNR blast respectively. These enhancementssnr are notsnr due to the globalsnr wave is Ðxed at the location of the Balmer emission. The expansion of the blast wave in a uniform medium or evapo- interior edge of the X-ray enhancement marks the location ration in a uniform distribution of clouds. They are too of the reÑected shock. We measure this to be at R \ slight, too limited in radial extent, and misplaced relative to 0.94R , so the blast wave initially encountered the wall at the Sedov and evaporative model predictions. The Balmer R \ 0.96snr R and is now atR \ 1.0R . Similar to all the Ðlaments and associated X-rays are detected over half the model calculationssnr presented here, thesnr X-ray surface bright- perimeter of the SNR, indicating that blast wave deceler- ness proÐle is computed for the spherically symmetric case ation in atomic material is common. Some regions exhibit (Fig. 11), assuming that X-ray volume emissivity is pro- the type 3 X-ray signature, but the corresponding optical portional to density squared and is independent of tem- Ðlaments are too faint to be detected. Although the blast perature. This simple model reasonably matches the shape wave edge is not directly observed in these cases, the preva- and location of the X-ray enhancement, although this is not lence of category 3 and the observed X-ray morphology a Ðt to the data. The model does not allow for heating time, imply that the blast wave is decelerating in the dense cavity so it predicts emission toward the edge of the SNR blast wall in these locations, as well. wave that we do not detect. Despite these approximations, A radial proÐle across the northwestern edge (20h46m, however, the simple model indicates the scales of the ]31¡40@) emphasizes the type 3 X-ray limb brightening and parameters that cause the X-ray enhancement at the edge avoids confusion with emission due to interactions with where the blast wave has run into the atomic cavity wall. large clumps(Fig. 11). There is no extended bright optical Several observed properties of the Cygnus Loop argue emission due to radiative shocks near the edge, and this is for a large-scale cavity explosion. The Ðrst is the approx- not a particularly bright region of the X-ray map. Two imate spherical symmetry of the emission. This symmetry of components are obvious, corresponding to two overlapping the Cygnus Loop is surprising in view of the rich X-ray projections of the blast wave that are not necessarily physi- morphology, which argues for an inhomogeneous medium. cally related. We neglect the interior component and model If the inhomogeneity were in the form of a random distribu- the exterior component very simply as a shock that strikes a tion of clouds, then the clouds would have to be very small density discontinuity and generates a reÑected shock. The in order to account for the fact that the observed gas pro- Rankine-Hugoniot jump conditions determine the relation- duces few, if any, signiÐcant indentations into the remnant ship among the Ñow variables in di†erent regions. The free (McKee& Cowie 1975). There are no entirely concave sec- parameters of the model are the density contrast and its tions on scales greater than 20@ (4 pc) and no evidence for initial location, from which the density proÐle due the pro- thermal evaporation of large clouds on the interior, for jected and reÑected shocks as a function of radius is calcu- example. However, X-ray studies of individual regions show lated. In the Cygnus Loop, Balmer-dominated Ðlaments are that the clouds are quite extended along the periphery of the typically observed propagating through regions of n D 1 remnant, with sizes of about0d.5 (7 pc). The clouds are cm~3 (e.g.,Hester et al. 1994; Raymond et al. 1983), while in preferentially aligned parallel to the blast wave. In this case, the former H II region cavity n D 0.1 cm~3, so the density a spherical remnant can be produced only if neither the contrast is taken to be 10. (This is likely to be an overesti- shape nor the distribution of the clouds is random. To produce a spherical remnant, the clouds in the vicinity of the Cygnus Loop must be located at the edge of the current 0.06 shell, and they must be Ñattened along the shell. There is also dynamical evidence that these are the Ðrst large clouds that have been encountered by the blast wave: the cloudÈ

−2 0.04 blast-wave interactions to the northeast and the west are recent and approximately the same age(Hester et al. 1994;

arcmin Levensonet al. 1996). This conclusion is consistent with −1 that ofCharles, Kahn, & McKee (1985), who determined 0.02 that the data are inconsistent with an explosion in a uniform distribution of clouds based on Einstein IPC observations adjacent to NGC 6960 byCharles, Kahn, & HRI counts s McKee(1985), who determined that these data are inconsis- 0.00 tent with an explosion in a uniform distribution of clouds. 4.2. Progenitor ModiÐcation of the ISM 0.70 0.80 0.90 1.00 The modiÐcation of the SNR environment by the presu- R/Rsnr pernova star has been discussed byMcKee, Van Buren, & FIG. 11.ÈHRI surface brightness as a function of radius across the Lazare†(1984) and Shull et al. (1985). If the progenitor star northwestern rim of the Cygnus Loop, azimuthally averaged over 2@. Two of the Cygnus Loop SNR were an early B star, then the separate components of the limb-brightened shell are evident, near R \ 0.96R andR \ 0.85R . Both components are the result of interaction of Lyman continuum radiation from this main-sequence star the blastsnr wave with the cavitysnr wall, although the latter appears projected to created an H II region. In the warm neutral interstellar the interior of the SNR. The outer peak is modeled as the result of a density medium with average density distribution n B 0.2 exp contrast and reÑected shock (dotted line). In this model, the blast encoun- ([z/220 pc) cm~3 (Kulkarni& Heiles 1988), a star at height ters a density contrast of 10 atR \ 0.96R , and the reÑected shock has z 120 pc emitting Lyman continuum photons at a rate of progressed toR \ 0.94R when the forwardsnr shock has reached R \ \ 1.0R . snr 1046 s~1 creates an H II region with a density of about 0.1 snr No. 1, 1997 X-RAY SURVEY OF THE CYGNUS LOOP 311 cm~3 and a radius of about 30 pc. As denser clumps of gas the processed interstellar medium is anisotropic: toward the within the H II region are photoevaporated, adding more original location of the progenitor, gas has been exposed to material to the ionized interior, the radius of the H II region ionizing radiation, so it is more homogeneous and less decreases. Small clouds are ultimately destroyed by photoe- dense, while the processing is limited in the direction that vaporation, and larger clouds are driven away by the rocket the star moves, so the ISM is clumpier and less dense there. e†ect(Bertoldi & McKee 1990). The details of the photoe- The progenitor of the Cygnus Loop would have been later vaporation depend on the velocity of the progenitor star than B0, provided that it was not moving too rapidly (v [ * and on the properties of the clouds: temperature, density, a few km s~1, so that surrounding clouds are exposed to the and magnetic Ðeld. UsingBertoldiÏs (1989) results, one can ionizing radiation for a timeR/v Z 6 10 yr). Alterna- * ] 6 show that atomic clouds (T D 102 K) in pressure balance tively, an earlier-type progenitor is possible provided that it with the H II region (P/k D 2000 K cm~3 for n D0.1 was moving predominantly along our line of sight. In this cm~3) will be either destroyed or rocketed out ofHII the H II case, the projected edge of the cavity would still be circular, region. Molecular clouds are more difficult to destroy by yet have a smaller radius than one carved by a stationary photoevaporation(Shull et al. 1985), but at the distance of star. the Cygnus Loop below the plane, very large molecular clouds are unlikely. As a result of photoevaporation and 5. CONCLUSIONS photodissociation, a dense atomic shell that contains most of the mass surrounds the H II region, which is bounded by The density inhomogeneities around the Cygnus Loop a recombination front. A vigorous wind may aid formation not only determine the appearance of the SNR at optical of the shell, although this is unlikely in the case of an early B and X-ray wavelengths, but they also fundamentally alter star we consider here. Once the progenitor leaves the main the blast-wave development. The consequences are signiÐ- sequence and becomes a red giant, the H II region begins to cant beyond this single example. Global models of the inter- recombine, but the density is too low for much recombi- stellar medium that assume the clouds around supernova nation to occur before the star explodes. remnants are insigniÐcant or rely on statistical measures When the progenitor of the Cygnus Loop became a such as the number-diameter relation must be revised. supernova, the blast wave initially expanded adiabatically The high-resolution, global view of the Cygnus Loop through the di†use cavity gas. Eventually it hit the walls of these data provide resolves the apparent paradox of the the cavity, which are comprised of the surrounding atomic Cygnus Loop: a near-circular shape and evidence for inter- shell and the remains of clouds that protrude into the H II action with large-scale inhomogeneities in the interstellar region. NGC 6960 is a clear example where the blast wave medium. The ubiquitous X-ray data make clear that the has encountered dense molecular gas(Levenson et al. 1996). blast wave has recently encountered density enhancements, The eastern edge and the southeast cloud may be additional which we identify as the walls of a cavity. The H II region of examples of elephant-trunk structures, such as those the progenitor star worked to homogenize a spherical observed in M16 (e.g.,Hester et al. 1996), protruding into cavity with ionizing radiation, and the blast wave initially the shell(Graham et al. 1995). The existence of large clouds propagated through this cavity, maintaining its spherical near the cavity edge is expected, for the development of an symmetry. Marking the edge of the cavity is a dense shell of H II region tends to homogenize its interior, while its action atomic gas and large clumps of the processed surrounding on the boundary is limited by intervening clouds along the medium or formerly interior clouds that were rocketed out. same line of sight and decreased ionizing Ñux at larger radii. Portions of the blast wave have been decelerated in the The clouds currently at the edge were either rocketed out of smooth atomic shell, which Balmer-dominated Ðlaments the interior or were originally located at the periphery of the and slightly limb-brightened X-ray proÐles trace. Other sec- H II region. tions of the blast wave have swept up even larger column We have considered a progenitor of spectral type later densities, resulting in very bright X-rays and optical emis- than B0 because a stationary star of earlier spectral type sion. The observed features of the Cygnus Loop are star would homogenize an H II region to a radius of about expected rather than surprising in the context of stellar 50 pc, while there are clearly inhomogeneities within the evolution that preceded the supernova. present 20 pc extent of the SNR(McKee et al. 1984; Charles et al.1985). If a massive star is moving, however, the clouds We thank the USRSDC for providing the ESAS software situated perpendicular to its path are not exposed to the used to reduce the X-ray data. The optical images were Lyman continuum emission for as long. The rocket e†ect obtained using the Prime Focus Corrector on the 0.8 m and photoevaporation are less e†ective on these clouds, and telescope at McDonald Observatory in collaboration with the homogenizing radius across the direction of motion is L. D. Keller and M. J. Richter. The research of C. F. M. is reduced. Along the direction of motion, the distribution of supported in part by NSF grant AST 95-30480.

REFERENCES Aschenbach,B. 1988, Appl. Opt., 27, 1404 Graham, J. R., Levenson, N. A., Hester, J. J., Raymond, J. C., & Petre, R. Bertoldi,F. 1989, ApJ, 346, 735 1995, ApJ, 444, 787 Bertoldi,F., & McKee, C. F. 1990, ApJ, 354, 529 Hamilton,A. J. S., Sarazin, C. L., & Chevalier, R. A. 1983, ApJS, 51, 115 Charles,P. A., Kahn, S. M., & McKee, C. F. 1985, ApJ, 295, 456 Hesteret al. 1996, AJ, 111, 2349 Chevalier,R. A., Kirshner, R. P., & Raymond, J. C. 1980, ApJ, 235, 186 Hester,J. J., & Cox, D. P. 1986, ApJ, 300, 675 Chevalier,R. A., & Raymond, J. C. 1978, ApJ, 225, L27 Hester,J. J., Danielson, G. E., & Raymond, J. C. 1986, ApJ, 303, L17 Cowie,L. L., McKee, C. F., & Ostriker, J. P. 1981, ApJ, 247, 908 Hester,J. J., Raymond, J. C., & Blair, W. P. 1994, ApJ, 420, 721 DeNoyer,L. K. 1975, ApJ, 196, 479 Kirshner,R. P., & Taylor, K. 1976, ApJ, 208, L83 Falle,S. A. E. G., & Garlick, A. R. 1982, MNRAS, 201, 635 Klein,R. I., McKee, C. F., & Colella, P. 1994, ApJ, 420, 213 Fesen,R. A., Blair, W. P., & Kirshner, R. P. 1982, ApJ, 262, 171 Klein, R. I., McKee, C. F., & Woods, D. T. 1995, in The Physics of the Fesen,R. A., Kwitter, K. B., & Downes, R. A. 1992, AJ, 104, 719 Interstellar Medium and Intergalactic Medium, ed. A. Ferrara, C. F. Graham,J. R., & Aschenbach, B. 1996, in ROSAT Newsletter, No. 13, 6 McKee, C. Heiles, & P. R. Shapiro (San Francisco: ASP), 366 312 LEVENSON ET AL.

Ku,W. H.-M., Kahn, S. M., Pisarski, R., & Long, K. S. 1984, ApJ, 278, 615 Scoville, N. Z., Irvine, W. M., Wannier, P. G., & Predmore, C. R. 1977, Kulkarni, S. R., & Heiles, C. 1988, in Galactic and Extragalactic Radio ApJ, 216, 320 Astronomy, ed. G. L. Verschuur & K. I. Kellermann (2d ed.; New York: Sedov, L. I. 1993, Similarity and Dimensional Methods in Mechanics (10th Springer), 95 ed.; Boca Raton: CRC) Laming, J. M., Raymond, J. C., McLaughlin, B. M., & Blair, W. P. 1996, Seward,F. D. 1990, ApJS, 73, 781 ApJ, 472, 267 Shull, P., Jr., Dyson, J. E., Kahn, F. D., & West, K. A. 1985, MNRAS, 212, Levenson, N. A., Graham, J. R., Hester, J. J., & Petre, R. 1996, ApJ, 468, 799 323 Shull,P., Jr., & Hippelein, H. 1991, ApJ, 383, 714 Levenson, N. A., Graham, J. R., Keller, L. D., & Richter, M. J. 1997, in Snowden, S. L., McCammon, D., Burrows, D. N., & Mendenhall, J. A. preparation 1994, ApJ, 424, 714 McKee, C. F. 1987, in Spectroscopy of Astrophysical Plasmas, ed. A. Spitzer, L., Jr. 1962, Physics of Fully Ionized Gases (2d ed.; New York: Dalgarno & D. Layzer (New York: Cambridge Univ. Press), 226 Wiley) McKee,C. F., & Cowie, L. L. 1975, ApJ, 195, 715 Tenorio-Tagle,G., Rozyczka, M., & Yorke, H. W. 1985, A&A, 148, 52 McKee,C. F., Van Buren, D., & Lazare†, B. 1984, ApJ, 278, L115 White,R. L., & Long, K. S. 1991, ApJ, 373, 543 Minkowski,R. 1958, Rev. Mod. Phys., 30, 1048 Zombeck,M. V., et al. 1990, Proc. SPIE, 1344, 267 Raymond,J. C., Blair, W. P., Fesen, R. A., & Gull, T. R. 1983, ApJ, 275, 636 FIG. 2.ÈROSAT HRI count rate map of the Cygnus Loop. This image is a mosaic of 41 pointings in 26 positions. Each observation has been background subtracted and exposure corrected before being combined into this map, which has 15A pixels. It is scaled linearly from 0 to 0.15 counts s~1 arcmin~2.

LEVENSON et al. (see 484, 306)

PLATE 6 30´

32° 0´

DEC (J2000) 30´

31° 0´

20h 58m 56m 54m 52m RA (J2000)

FIG.5a

30´

32° 0´

DEC (J2000) 30´

31° 0´

20h 58m 56m 54m 52m RA (J2000)

FIG.5b FIG. 5.ÈThe northeast rim of the Cygnus Loop at 6A resolution. (a) HRI data, which exhibits three distinct types of X-ray emission: long Ðlaments of bright emission, spots of extremely bright emission accompanied by indentations in the X-ray shell, and fainter limb-brightened Ðlaments. (b) The northeast rim of the Cygnus Loop in Ha, scaled linearly. The optical image contains extended radiative Ðlaments, small spots of bright emission, and long, faint Ðlaments due to Balmer-dominated shocks, corresponding to the classes of X-ray emission. (c) False-color composite of HRI (green) and Ha (red) observations, illustrating the exact relation between the X-ray and optical emission. The optical image has been median Ðltered to remove stars.

LEVENSON et al. (see 484, 306) PLATE 7 FIG.5c

LEVENSON et al. (see 484, 306)

PLATE 8 31° 0´

30´ DEC (J2000)

30° 0´

20h 58m 56m 54m 52m RA (J2000)

FIG.6a

FIG.6b FIG. 6.ÈThe southeast region of the Cygnus Loop at 6A resolution. (a) HRI, (b)Ha, and (c) false-color composite, as in Fig. 5.

LEVENSON et al. (see 484, 306)

PLATE 9 FIG.6c

LEVENSON et al. (see 484, 306)

PLATE 10 FIG.7a

FIG.7b FIG. 7.ÈThe northwest edge of the Cygnus Loop at 6A resolution. (a) HRI, (b)Ha, and (c) false-color composite, as in Fig. 5.

LEVENSON et al. (see 484, 306)

PLATE 11 FIG.7c

LEVENSON et al. (see 484, 306)

PLATE 12