1: %\documentclass{aastex}
2: \documentstyle[emulateapj]{article}
3: %\documentclass[aaspp4,tighten]{aastex}
4: %\documentclass[12pt,preprint]{aastex}
5:
6: \def\kms{km\,s$^{-1}$\,}
7: \def\vs{$V_{S}$\,\,}
8: \def\vb{$V_{B}$\,\,}
9: \def\etal{ et~al.\rm\,}
10: \def\sb{ergs cm$^{-2}$ s$^{-1}$ arcsec$^{-2}$}
11: \def\fl{ergs cm$^{-2}$ s$^{-1}$}
12: \def\tetp{$(T_e/T_p)_0$}
13: \def\totp{$(T_O/T_p)_0$}
14: \def\fuse{{\it FUSE}}
15: \def\chan{{\it Chandra}}
16: \def\eins{{\it Einstein}}
17: \def\rosat{{\it ROSAT}}
18: \def\dem71{DEM\,L\,71}
19: \def\onine{SNR\,0509$-$67.5}
20: \def\lybovi{F(Ly $\beta$)/F$_{O\,VI}$(1032)}
21: \def\onenine{SNR\,0519$-$69.0}
22: \def\foureight{SNR\,0548$-$70.4}
23:
24: %\tighten
25: \begin{document}
26:
27: \title{The Detection of Far Ultraviolet Line Emission from Balmer-Dominated Supernova Remnants
28: in the Large Magellanic Cloud\altaffilmark{1} }
29:
30:
31: \author{Parviz Ghavamian\altaffilmark{2}, William P. Blair\altaffilmark{2}, Ravi
32: Sankrit\altaffilmark{3}, John C. Raymond\altaffilmark{4} and John P. Hughes\altaffilmark{5}}
33:
34: \altaffiltext{1}{This work is based on data obtained for the Guaranteed Time Team by the NASA-CNES-CSA
35: FUSE mission operated by the Johns Hopkins University. Financial support to U.S.
36: participants has been provided by NASA contract NAS5-32985.}
37: \altaffiltext{2}{Department of Physics and Astronomy, Johns Hopkins University, 3400 North
38: Charles Street, Baltimore, MD, 21218-2686}
39: \altaffiltext{3}{Space Science Laboratory, University of California, Berkeley, Berkeley, CA 94270-7450}
40: \altaffiltext{4}{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA, 02138}
41: \altaffiltext{5}{Department of Physics and Astronomy, Rutgers University, Piscataway, NJ, 08854-8019}
42:
43: \begin{abstract}
44:
45: We present the first far ultraviolet (FUV) spectra of the four known Balmer-dominated
46: supernova remnants (SNRs) in the Large Magellanic Cloud, acquired with the Far Ultraviolet Spectroscopic
47: Explorer (\fuse). The remnants \dem71 (SNR 0505$-$67.9), \onine, \onenine\, and \foureight\,
48: are all in the non-radiative stages
49: of evolution and exhibit expansion speeds ranging from $\sim$ 500 \kms to $\sim$ 5000 \kms.
50: We have detected broad emission
51: lines of Ly $\beta$, Ly $\gamma$, C~III and O~VI in \dem71\, ($V_{FWHM}$ $\sim$\,1000 \kms) and
52: have detected broad Ly $\beta$ and O~VI emission in \onenine\, ($V_{FWHM}$ $\sim$\,3000 \kms).
53: In addition, broad Ly $\beta$ emission ($V_{FWHM}$ $\sim$\,3700 \kms) has been observed in \onine,
54: the first detection of broad line emission from this SNR.
55: No emission was detected in our \fuse\, spectrum of \foureight, allowing us to place only upper limits on
56: the FUV line fluxes. The spectra of these SNRs are unaffected by postshock cooling, and provide
57: valuable probes of collisionless heating efficiency in high Mach number shocks. We have used the
58: \lybovi\, flux ratio and relative widths of the
59: broad Ly $\beta$ and O~VI lines to estimate the degree of electron-proton and
60: proton-oxygen ion equilibration in \dem71, \onine\, and \onenine. Although
61: our equilibration estimates are subject to considerable uncertainty due to
62: the faintness of the FUV lines and contributions from bulk Doppler broadening, our
63: results are consistent with a declining efficiency
64: of electron-proton and proton-oxygen ion equilibration with increasing shock speed. From
65: our shock velocity estimates we obtain ages of 295-585 years for \onine\, and 520-900 years
66: for \onenine, respectively, in good agreement with the ages obtained from
67: SN light echo studies.
68:
69: \end{abstract}
70: \keywords{ ISM: supernova remnants: individual (0505$-$67.9 (\dem71), \onine, \onenine, \foureight)--
71: ISM: kinematics and dynamics, shock waves}
72:
73: \section{INTRODUCTION}
74:
75: The proximity of the Large Magellanic Cloud
76: (LMC) (50 kpc; Feast 1991, 1999) and the relatively small extinction along
77: the line of sight to this galaxy (E(B-V) $\sim$ 0.1) make it an excellent laboratory
78: for study in the X-ray and ultraviolet. In particular, the well calibrated distance of
79: the LMC allows us to study both the kinematics of individual supernova
80: remnants (SNRs) (Blair \etal\, 2006) and properties of the global SNR distribution within this galaxy.
81:
82: The first systematic X-ray survey of the LMC was performed with the Imaging Proportional
83: Counter of the \eins\, observatory (Long, Helfand \& Grabelsky 1981) in the 0.15-4.5 keV
84: range. The observations revealed that nearly half of the 97 detected sources exhibited
85: an extended morphology characteristic of SNRs, making these objects the largest class
86: of X-ray sources revealed in the survey. Four of the SNRs identified with \eins\, were
87: shown by Tuohy \etal\, (1982) to exhibit optical spectra consisting
88: almost entirely of faint hydrogen Balmer line emission (i.e., they are `Balmer-dominated'). The four
89: remnants $-$ 0505$-$67.9 (DEM L 71 from the Davies, Elliott \& Meaburn (1971) catalogue), \onine,
90: \onenine\, and \foureight\, vary widely in age, ranging from a mere few hundred years
91: (\onine; Rest \etal\, 2005) to $\sim$10,000 years (\foureight, Smith \etal\, 1991).
92:
93: Balmer-dominated line emission is produced when the expanding SNR
94: encounters low density ($\sim\,$0.1-1 cm$^{-3}$), partially neutral interstellar
95: ISM (Chevalier \& Raymond 1978; Bychkov \& Lebedev 1979). Due to the high temperature and
96: low density behind the blast wave, the cooling time scale of the gas greatly exceeds
97: the dynamical time scale of these SNRs (they are non-radiative). The hydrogen overrun by the shock is collisionally
98: excited in a thin ($\lesssim$10$^{15}$ cm) ionization zone, producing optical spectra dominated
99: by the Balmer lines of hydrogen (Chevalier \& Raymond 1978; Chevalier, Kirshner \& Raymond
100: 1980). In the ultraviolet, the spectra are
101: 'Lyman-dominated'. Balmer-dominated shocks are mostly found in young remnants (e.g., SNRs in the
102: Sedov-Taylor stage or younger) of Type Ia explosions. In these objects the survival
103: of significantly neutral, low density ISM gas is likely aided by the lack of
104: strong ionizing radiation and stellar wind from the progenitor star (a white dwarf).
105: X-ray spectroscopy of the reverse-shocked ejecta
106: in the four LMC Balmer-dominated remnants has revealed strong Fe L shell emission
107: (Hughes \etal\, 2003; Hendrick, Borkowski \& Reynolds 2003; Warren \& Hughes 2004),
108: confirming the Type Ia origin of these SNRs.
109: These overall characteristics place the four Balmer-dominated LMC remnants into
110: the same class as the well-known Galactic Type Ia SNRs Tycho and SN 1006 (Chevalier, Kirshner \& Raymond
111: 1980; Kirshner, Winkler \& Chevalier 1987; Smith \etal\, 1991; Ghavamian \etal\, 2001),
112: making them the first examples of such remnants identified outside our galaxy.
113:
114: The hydrogen line emission in Balmer-dominated shocks consists of two components:
115: a narrow component produced by collisional excitation of overrun ambient H~I, and
116: a broad component generated when the hot postshock H~I produced by charge exchange is collisionally
117: excited. Both broad and narrow component Balmer lines can often be detected
118: in the optical, but interstellar absorption in the ultraviolet completely removes the
119: narrow component Lyman line emission. However, if the broad component Lyman lines are both
120: bright and wide, the profile wings can be detected.
121:
122: The diagnostic utility of the broad H~I lines is twofold. First, the width of
123: a broad H~I line gives the postshock proton temperature, thus yielding an estimate of
124: the shock speed when combined with the Rankine-Hugoniot jump conditions (Chevalier, Kirshner
125: \& Raymond 1980; Smith \etal\, 1991; Ghavamian \etal\, 2001). Second, comparing
126: the width and flux of the Lyman lines with those of ultraviolet resonance lines such as
127: O~VI $\lambda\lambda$1032, 1038 and C~III $\lambda$977 yields an estimate of the ion-ion temperature
128: equilibration at the shock front, well before the equipartition of energy between the
129: different ionic species by Coulomb collisions (Hester, Raymond \& Blair 1994; Raymond, Blair \& Long 1995).
130: This is key information for probing the nature of collisionless (plasma) heating processes at the
131: shock front, one of the central problems in understanding high Mach number shocks in
132: many astrophysical settings.
133:
134: The availability of broad hydrogen, carbon and oxygen emission lines,
135: along with high quality X-ray spectra from \chan\,
136: provides us with an invaluable opportunity to simultaneously study both the electron-proton and ion-ion
137: equilibration in fast collisionless shocks and the evolution of remnants from Type Ia SNe. In this
138: paper we present FUV spectra of the four Balmer-dominated LMC SNRs. We utilize numerical shock
139: models to interpret the spectra, constrain the kinematics of the four SNRs and investigate collisionless
140: heating processes in fast astrophysical shock waves.
141:
142: \section{OBSERVATIONS AND DATA REDUCTION}
143:
144: \subsection{THE RAW DATA}
145:
146: We observed the four Balmer-dominated LMC SNRs with the Far Ultraviolet Spectroscopic Explorer
147: (\fuse) in 2001 September 23 and 24 (see Moos \etal\, 2000 and Sahnow \etal\, 2000 for a detailed
148: description of the spectrograph). The data were acquired as part of a Cycle 3 Guaranteed Time Observation
149: program (P214; W. P. Blair, PI), with each remnant observed through the low resolution (LWRS)
150: square aperture of \fuse\, (30\arcsec\,$\times$\,30\arcsec). In each case the LWRS aperture
151: was centered on the geometric center of each SNR as measured in the optical by Tuohy \etal\, (1982).
152: The roll angles of these observations were set by the nominal spacecraft roll at the time of data
153: acquisition. The total integration time and the time obtained during orbital night
154: is listed for each object in Table~1.
155: The Balmer-dominated LMC SNRs vary in size from approximately 30\arcsec\,
156: across (\onine) to 1\farcm1 across (\foureight), resulting in significant variations
157: in the fraction of the shell covered in each observation (see Figure~1). While the LWRS aperture
158: covered nearly the entire shell of \onine\, and \onenine, it sampled primarily
159: the approaching and receding faces of the shell in \dem71 and \foureight. As we show
160: in the following sections, the fraction of each SNR covered by the \fuse\, aperture can significantly
161: affect the shapes and centroids of the FUV lines in the final extracted spectra.
162:
163: We acquired a separate \fuse\, observation of DEM L71 during Cycle 3 (Guest Observer program
164: C072; PI: P. Ghavamian). In this observation we placed the MDRS aperture (4\arcsec\,$\times$20\arcsec)
165: parallel to the southern edge of the blast wave at a position angle of 113$^{\circ}$ (see Figure~1). The
166: aperture orientations were nearly constant during these observations. A total
167: integration time of 121.6 ks was accumulated during five separate pointings
168: on 2003 January 4-7 (C0720202-C0720205) and 2003 December 29 (C0720206). However, the target
169: acquisition in observation C0720205 failed, leaving 100 ks of useful data.
170:
171: Our data reduction was performed using version 3.1.3 of the CalFUSE pipeline. The expected
172: FUV line emission in the four LMC remnants is expected to be faint ($\lesssim$10$^{-13}$
173: ergs cm$^{-2}$ s$^{-1}$), with line widths ($\gtrsim$1000 \kms) greatly exceeding the resolution
174: of the filled LWRS aperture (approximately 106 \kms). These properties make the detection of the FUV lines
175: quite sensitive to the quality of the background subtraction. Since the CalFUSE pipeline derives an
176: empirical background for each spectrum using non-illuminated portions of the detector, we can
177: improve the signal-to-noise of the measured background by co-adding the individual raw exposures for
178: each of the four FUSE detectors (1A,1B,2A,2B), then extracting 1-D spectra from the combined raw frames.
179: We used the IDL routine {\sf ttag\_combine} to combine the raw exposures
180: (Figure~2), then used the CalFUSE pipeline to extract the 1-D spectra. We retained only
181: the orbital night portions from each dataset (Table~1) to minimize the contribution of
182: airglow emission lines and scattered airglow continuum to the object spectra.
183:
184: We combined the raw exposures from the MDRS data in the same manner described above for the LWRS
185: data. Since the C0720206 data were acquired nearly a year after the other
186: observations, the FUV spectrum of this observation is significantly shifted in
187: detector pixels relative to the earlier observations. Therefore, we first co-added
188: the C0720202-C0720204 datasets into one raw image, then co-added
189: C0720206 into a separate raw image. We extracted spectra from each image,
190: cross-correlated each one and finally co-added them to obtain the
191: 1-D MDRS spectrum of the DEM L71 blast wave.
192:
193: \subsection{SPECTRAL EXTRACTION}
194:
195: The FUSE spectra are recorded on four channels: LiF 1, LiF2, SiC 1 and SiC 2. Each channel
196: is divided into A and B segments, with the two segments covering the approximate wavelength range
197: 905 \AA\,-\,1100 \AA\, for each of the LiF 1 and LiF 2 channels, and the wavelength range
198: 987 \AA\,-\, 1180 \AA\, for the SiC 1 and SiC 2 channels (see the FUSE Observer's Guide\footnote{
199: http://fuse.pha.jhu.edu/support/guide/guide.html}).
200: In each of our LWRS observations, the LiF 1A and SiC 2A (917 \AA\,-\,1006 \AA)
201: segments offer the highest effective area at the positions of the expected FUV emission lines.
202: Therefore, we searched the spectra from these segments for SNR emission first. Although the
203: extracted \fuse\, spectra are meant to be summed into one high signal-to-noise
204: data product, inter-channel thermal drifts during each exposure can result in
205: misalignments as large as 10\arcsec\, between LiF 1 and the other channels.
206: While the LiF 1 channel was held fixed on the SNR during an observation, the others
207: sampled emission from slightly different portions of the SNR. The resulting
208: spectra can differ from channel to channel. Therefore,
209: for each observed SNR we combined spectra from multiple channels only when they appeared consistent with
210: one another to within the flux uncertainties.
211:
212: In the case of \dem71, prominent, broad emission lines of Ly $\beta$, Ly $\gamma$, O~VI
213: and C~III are detected in the LWRS observations. The spectral features are observed in all
214: segments of the \fuse\, detector save for the LiF 1B and LiF 2A segments, whose spectral range
215: excludes these emission lines. Although inter-channel drift is expected to
216: have occurred during the 28 ks observation, the spectra are all consistent with one another
217: to within the errors. We generated the spectra in Figure~3 by cross-correlating and
218: combining data from SiC 2A with SiC 1B (905 \AA\,$-$\,1005 \AA) and the data from
219: LiF 1A with those of LiF 2A and SiC 1A (988 \AA\,$-$\,1090 \AA). Although
220: broad Ly $\beta$ and O~VI emission were seen in the SiC 2B segment, we did not include
221: this data in our co-addition because the background was unevenly
222: over-subtracted for this segment by the CalFUSE pipeline.
223:
224: Unlike \dem71, \onine\, and \onenine\, fit almost completely within
225: the LWRS aperture (Figure~1), increasing the likelihood that inter-channel drifts
226: would adversely affect the detection of FUV emission from these objects. Since \onine\, exhibits
227: the faintest FUV emission of all the Balmer-dominated SNRs (Figure~4), we
228: restricted our analysis of this object to data from the LiF 1A segment.
229:
230: In the case of SNR 0519$-$69.0, on the other hand, FUV line emission was detected
231: in all segments shortward of 1100 \AA. One factor complicating our interpretation
232: is the presence of a very faint continuum under the Ly $\beta$
233: and O~VI features (clearly seen in Figure~2). The continuum is absent on the
234: short wavelength side of the broad Ly $\beta$ line, but rises to a nearly constant level
235: beneath the broad lines and continues to the long wavelength end of the spectrum.
236: The continuum is absent in the other channel spectra near 1030 \AA, and appears
237: to be stellar contamination. After comparing the emission in all channels,
238: we cross-correlated and combined the LiF 1A, LiF 2B, SiC 1A and SiC 2B
239: data to obtain spectra in the 940 \AA\,$-$1050 \AA\, range. Although broad
240: Ly $\gamma$ was also detected, it was present solely in the SiC 2A
241: spectrum of \onenine\, (the channel with the highest effective area
242: in this wavelength region).
243:
244: In our MDRS observation of \dem71, FUV emission was detected only
245: in the spectrum of the LiF 1A segment. The lack of detection in the other segments is likely
246: due to a combination of their lower effective areas and the
247: the strong impact of inter-channel thermal drifts.
248: The latter effect is especially important here, where the
249: smaller size of the MDRS aperture makes it easier for non-primary channels to drift
250: off the targeted filament.
251:
252: One of the more surprising results of our observations is the lack of any
253: discernible FUV emission from the largest of the Balmer-dominated SNRs,
254: \foureight\, (see Figure~6). No emission is detected in either the raw combined data,
255: the individual channel spectra or the co-added spectra from all channels.
256: In the following section we will describe a possible explanation for this
257: non-detection, and discuss its implications.
258:
259: \section{FAR ULTRAVIOLET SPECTRA}
260:
261: Of the observed Balmer-dominated remnants, two objects $-$ \dem71\, and \onenine\, $-$
262: exhibit readily identifiable, broad FUV line emission consistent with Balmer-dominated
263: shocks: Ly $\beta$
264: ($\lambda$1025.7), O~VI $\lambda\lambda$1032, 1038, Ly $\gamma$ ($\lambda$972.5) and
265: C~III $\lambda$977. In a third remnant, \onine, only the broad Ly $\beta$ line
266: is detected with reasonable certainty. However, this is the first detection of broad
267: hydrogen line emission from this remnant. Surprisingly, no discernible FUV emission is
268: seen from the fourth remnant, \foureight, despite the known detection
269: of broad Balmer line emission in this remnant by Smith \etal\, (1991). Here, we describe
270: the \fuse\, spectra of each remnant in detail. All spectral fits have been performed using
271: the {\sf SPECFIT} task (Kriss 1994) from the {\sf STSDAS} contributed
272: package in IRAF\footnote{IRAF is distributed by the National Optical Astronomy
273: Observatories, which is operated by the AURA, Inc. under cooperative agreement with the National Science Foundation.}
274: unless noted otherwise.
275:
276: \subsection{\dem71}
277:
278: \subsubsection{LWRS Spectra}
279:
280: The combined LWRS spectrum of \dem71\, (Figure~3) provides an especially impressive example of
281: the FUV emission from a Balmer-dominated SNR. The emission lines are very broad
282: ($\sim$1000 \kms), bright and clearly impacted by Galactic and LMC halo absorption. The
283: presence of strong, broad Ly $\beta$ indicates that a significant fraction of the gas
284: entering the blast wave is neutral, while the relative faintness of the C~III line relative
285: to O~VI indicates that postshock cooling is negligible (Raymond 2001). The
286: wavelengths of prominent absorption features from both the Milky Way and LMC are marked in Figure~3.
287: The broad Ly $\beta$, O~VI $\lambda$1032 and O~VI $\lambda$1038 lines are moderately
288: blended, as are the Ly $\gamma$ and C~III $\lambda$977 lines.
289:
290: The LWRS aperture samples emission near the projected center of \dem71\,
291: (Figure~2). Therefore, each emission line should appear
292: doubled by the bulk motion of the postshock gas ($\pm\,\frac{3}{4}\,v_{s}$) on near
293: and far sides of the SNR. The shock velocities in \dem71\, have been
294: estimated to lie between 600 \kms\, and 1100 \kms\, (Smith \etal\, 1991; Ghavamian \etal\,
295: 2003; Rakowski, Ghavamian \& Hughes 2003), so the emission lines in \dem71\, should
296: appear blue/red-shifted by 400$-$800 \kms\, from the rest velocity of the LMC (which
297: we take to be +275 \kms). However, the presence of strong interstellar absorption near the center
298: of each emission line in Figure~3 makes it difficult to distinguish whether each
299: emission line is doubled. As we show in Section~4, the line profile shapes
300: observed in \dem71\, are in fact consistent with the presence of two Doppler-shifted
301: components. We have devised a spherical shell model of the blast wave to
302: address the significant interplay between bulk and thermal Doppler
303: motions in this spectrum. We present this model in the next section.
304:
305: Although a spherical shell model is required for a kinematic interpretation
306: of the spectrum in Figure~3, we can still estimate
307: the observed flux in each of the broad emission features by fitting each with a
308: single Gaussian line. Prior to fitting each spectrum we
309: estimated the baseline level by fitting the data on the blue
310: and red ends of the spectrum (devoid of emission features) with a straight line
311: of zero slope. After obtaining the best fit value to the baseline, we held
312: this quantity fixed during our subsequent fits to the emission lines. We chose a 9-component fit
313: for the \dem71\, spectrum, fitting the emission components with
314: a flat baseline and three Gaussian emission lines for broad Ly $\beta$ and O~VI $\lambda\lambda$1032, 1038.
315: We included three absorption components for broad Ly $\beta$ (Galactic and LMC halo H~I absorption, Galactic
316: O~I absorption at 1027 \AA) while excluding the wavelength range covering the Ly $\beta$ airglow
317: line. We also included one O~VI LMC absorption component for each of the
318: broad O~VI lines. Although there is possible evidence for more absorption components along
319: the line of sight, we found that adding more of these components did not substantially
320: improve the statistical quality of the overall fit.
321:
322: We improved the constraining power
323: of our fits by tying together the widths and velocity centroids of the O~VI
324: $\lambda\lambda$1032, 1038 emission lines. We applied the same constraint to
325: the O~VI absorption features, with the additional requirement $\tau_{1032}$\,:\,$\tau_{1038}$ = 2:1. In fitting the emission
326: near 972 \AA\, we used a 5 component model consisting of a linear baseline, one Gaussian each for
327: Ly $\gamma$ and C~III $\lambda$977, and one Galactic and one LMC H~I absorption component.
328: We held the velocity width and line centroid of the Ly $\gamma$ line equal to the values measured
329: from our Ly $\beta$ profile fit.
330: We included the H~I absorption components solely as an aid for better fitting of the broad Ly $\gamma$ emission
331: line. The fitted H~I absorption lines do not actually provide reliable information on the H~I column
332: density because the data quality near the cores of the absorption profiles are heavily compromised
333: by the bright airglow Ly $\beta$ line. Furthermore, we used a Gaussian profile shape
334: to describe what are actually saturated H~I absorption lines, further invalidating a physical interpretation
335: of the fitted parameters for these components.
336:
337: The resulting emission line fit for \dem71\, is marked in Figure~3 and the predicted emission
338: line parameters are listed in Table~2. The Ly $\beta$ line width (1135$\pm$30 \kms) is significantly
339: larger than that of the O~VI $\lambda$ 1032 line (740$\pm$45 \kms), already suggesting
340: that the protons and oxygen ions have undergone significant temperature equilibration at the
341: shock front. However, we defer the task of estimating the specific shock velocities and
342: comparing the results to existing optical measurements to Section~4. We note the slight
343: velocity shift for the fitted emission line centroids: taking a systemic velocity of
344: +275 \kms\, for the LMC, the Ly $\beta$ and O~VI $\lambda$1032 lines are redshifted
345: by 100$-$150 \kms\, (Table~2). This feature suggests that the emission from the far side of the
346: \dem71\, shell is slightly brighter than the near side. In Section~4 we quantify this
347: result with more detailed. models of the blast wave shell.
348:
349: The prominent absorption features present in the O~VI $\lambda\lambda$1032, 1038 lines
350: are centered at 250$\pm$50 \kms, indicating that they arise in the LMC halo. The Galactic
351: absorption component is considerably weaker. Our model fits indicate $\tau_{1032}$(Gal)\,=\,
352: 1.7$\pm$0.6, with $\sigma\,=\,$50$\pm$22 \kms\, after deconvolving the measured FWHM with the filled slit
353: spectral response of 106 \kms. The implied O~VI column density is approximately
354: (3.5$\pm$2.2)$\times$10$^{14}$ cm$^{-2}$. Both the line width and column density lie at
355: the upper end of, but are fully
356: consistent with, values measured by Howk \etal\, (2002) in their analysis of \fuse\,
357: spectra of 12 early-type stars in the LMC.
358:
359: Aside from the O~VI doublet, C~III $\lambda$977 in \dem71\, is the only other heavy ion line detected
360: in our LWRS spectra of the Balmer-dominated LMC remnants. Our single Gaussian fit (Table~2)
361: indicates that the C~III line is less than half as broad as the Ly $\beta$ line, indicating
362: that significant temperature equilibration has taken place between protons and carbon ions
363: at the shock front.
364:
365: \subsubsection{MDRS Spectrum}
366:
367: The night-only MDRS spectrum of \dem71\, (Figure~7) reveals prominent, broad Ly $\beta$ emission
368: moderately blended with faint O~VI lines. As with the face-on LWRS spectrum, we fit the
369: three lines together while including a fixed baseline level and both Galactic and LMC H~I
370: absorption. The O~VI lines are faint enough to make the Galactic and LMC halo
371: absorption features undiscernible, so we did not include absorption in our O~VI profile
372: fits. The inclusion of absorption components had minimal impact on the quality of the fit.
373: The final results, quoted in Table~2, indicate a broad Ly $\beta$ width
374: of 1365$\pm$75 \kms\, FWHM. The O~VI line is significantly narrower, 935$\pm$125 \kms,
375: an indication that partial ion-ion equilibration has taken place at the shock transition.
376:
377: \subsection{\onine}
378:
379: In their optical spectroscopy of this SNR Tuohy \etal\, (1982) and Smith \etal\,
380: (1991) were unable to detect broad Balmer line emission and concluded that the line was
381: too faint and broad to be detected. Our LWRS spectrum of \onine\, reveals a faint,
382: strongly blueshifted broad Ly $\beta$ line (Figure~4), the first broad emission line detected
383: from this SNR to date. The detected line emission is indeed faint,
384: amounting to only 20\% and 35\% of the broad Ly $\beta$ flux
385: in \dem71\, and \onenine, respectively. We fit the broad Ly $\beta$ profile
386: with a single Gaussian and ignored the regions containing the Ly $\beta$
387: airglow and H~I absorption features. The result of this fit (Table~2) is a broad
388: Ly $\beta$ width of 3710$\pm$400 \kms, where the uncertainty includes both the
389: random statistical error ($\pm$325 \kms) and the baseline uncertainty ($\pm$240
390: \kms; this quantity is the variation in line width when the background level
391: is fixed to the upper or lower limits allowed by the initial baseline fit). This is the largest
392: H~I line width measured for a broad component in any of the Balmer-dominated SNRs
393: thus far.
394:
395: Aside from being very broad, the Ly $\beta$ line is blueshifted to $-$263$\pm$125 \kms,
396: indicating a large velocity shift ($-$540 \kms) relative to the LMC frame. The
397: velocity shift of the broad line center is proportional to the projected component
398: of the bulk velocity along the line of sight. The blueshift of the line indicates
399: that the Ly $\beta$ emission arises on the near side of \onine. As seen
400: in Figure~1, there is a prominent patch of enhanced H$\alpha$ emission stretching
401: inward from the southwest corner of the shell. It is likely that this patch lies
402: on the near side of \onine\, and that \fuse\, has detected the FUV emission from this
403: region. As we will show in Section~4, the size of the Ly $\beta$ broad component shift
404: is fully consistent with the position of the patchy enhancement relative to the
405: geometric center of \onine. Note that the region of enhanced emission
406: is isolated close to the edge of the shell, so the range of viewing angles to
407: this section of the blast wave is likely small (i.e., the emitting region
408: can be approximated as a plane-parallel shock tilted slightly toward
409: the observer).
410:
411: \subsection{\onenine}
412:
413: This Balmer-dominated SNR is similar in size to \onine\, (and likely of comparable
414: age as measured from light echoes from the SN explosion, Rest \etal\, 2005). The
415: FUV spectrum (Figure~5) exhibits broad Ly $\beta$ emission line at 1026 \AA, blended with
416: broad O~VI $\lambda\lambda$1032, 1038 line emission. A broad Ly $\gamma$
417: line can clearly be discerned near 972 \AA\, although the C~III line falls below the
418: detection threshold.
419:
420: We can draw a direct comparison between the broad Ly $\beta$ widths measured in
421: \onenine\, and the H$\alpha$ widths measured by Tuohy \etal\, (1982) and
422: Smith \etal\, (1991). After deconvolving with the filled slit resolution (106 \kms)
423: we find good agreement between the broad Ly $\beta$ width of 3130$\pm$155 \kms\,
424: and the H$\alpha$ value of 2800$\pm$300 \kms\, measured along the rim by
425: Tuohy \etal\, (1982). However, our line width is over twice the 1300$\pm$200 \kms\, width
426: measured for the broad H$\alpha$ line by Smith \etal\, (1991). The reason for the difference
427: is that the spectroscopic slit used by Smith \etal\, (1991) was placed E-W across
428: \onenine, intersecting the bright knot on the eastern rim
429: of the remnant (clearly seen in Figure~2). The spectra of Smith \etal\, (1991)
430: were only deep enough to reveal broad H$\alpha$ emission from this bright knot,
431: where the blast wave has encountered a relatively dense region of the ISM and slowed.
432: On the other hand, Tuohy \etal\, (1982) observed the fainter, more uniform
433: portions of the \onenine\, rim, where considerably less deceleration has
434: occurred. As shown in Figure~2, the position and orientation of the LWRS aperture
435: placed the brightened emission knots entirely outside the slit during the \fuse\, observation.
436: Therefore, our FUV spectrum samples the less decelerated portions of \onenine\, and
437: should give Ly $\beta$ widths closer to the results of Tuohy \etal\, (1982).
438:
439: \subsection{\foureight}
440:
441: There are no emission lines detected in the \fuse\, spectrum of this object (Figure~6).
442: The broad H$\alpha$ width measured by Smith \etal\, (1991) along the eastern rim of \foureight\,
443: is 760$\pm$140 \kms, corresponding to shock speeds of 700$-$950 \kms. Since we easily detected shocks of
444: similar speed in our LWRS \fuse\, spectrum of \dem71 in both Ly $\beta$ and O~VI, it appears that the
445: blast wave region covered by our LWRS slit (Figure~1) is propagating into a very low density, mostly ionized
446: medium. In that case, pressure conservation would require very high shock speeds at this position, meaning
447: that aside from being faint, the lines are probably also very broad.
448:
449: Interestingly, our \fuse\, aperture
450: includes several bright clumps of shocked interstellar material near the projected center
451: of \foureight\, (Figure~1). The lack of C~III and O~VI emission from the knots in our \fuse\, spectrum
452: suggests that they are excited by slow radiative shocks ($\lesssim$90 \kms; Hartigan, Raymond
453: \& Hartmann 1987). In their narrowband
454: imagery of \foureight\, Tuohy \etal\, (1982) detected all of these clumps in H$\alpha$, but
455: only detected [O~III] emission from the clumps lying outside the \fuse\, aperture. In addition,
456: the \chan\, ACIS S image of \foureight\, (Hendrick, Borkowski \& Reynolds 2003) does not show any
457: obvious X-ray emission from clumps within the \fuse\, aperture. These features are
458: all characteristic of slow radiative shocks and suggest that dense, localized material
459: has been overrun by the blast wave on the near and/or far side of \foureight.
460:
461: \section{TEMPERATURE EQUILIBRATION AND KINEMATICS}
462:
463: The \fuse\, spectra of the Balmer-dominated LMC SNRs are particularly valuable in that they feature
464: prominent emission lines of hydrogen, carbon and oxygen, allowing us to directly
465: measure the degree of ion-ion equilibration in fast non-radiative shocks. The diagnostic
466: information available to us includes (1) the relative widths of the C~III $\lambda$977, Ly $\beta$
467: and O~VI $\lambda\lambda$1032, 1038 lines, which reflect both the relative temperatures
468: of these ions and the shock speed (assuming we know the contribution of bulk Doppler
469: broadening to the line profiles), and (2) the Ly $\beta$ / O~VI flux ratio, which
470: reflects the electron-proton equilibration, shock speed and preshock fraction of neutral
471: hydrogen.
472:
473: In the case of \onenine, the good agreement between our measured broad Ly $\beta$
474: width and the H$\alpha$ width measured by Tuohy \etal\, (1982) from the edge of the SNR
475: suggests that the FUV profiles are primarily broadened by thermal motions, i.e.,
476: that the emission is dominated by the limb-brightened edges where the
477: bulk motion along the line of sight is small. Since
478: the LWRS slit also covers most of \onine, we can expect the broad Ly $\beta$
479: profile from this SNR to also be shaped by thermal broadening. Finally, we can
480: expect our MDRS spectrum of the \dem71\, blast wave to also exhibit lines
481: dominated by thermal broadening, since the edge of the blast wave was targeted
482: in this observation. In all these cases, we have used the FUV line widths as
483: a direct measure of the ion temperatures and have utilized the non-radiative
484: shock code of Ghavamian \etal\, (2001, 2002) to predict the emission line
485: fluxes.
486:
487: In contrast to the spectra mentioned above, the interpretation of our LWRS
488: line profiles for \dem71\, is less straightforward. In this case most of the FUV emission
489: is sampled from the projected center of the SNR, where emission from a range of viewing angles
490: (and hence bulk velocities) is sampled in the \fuse\, spectrum.
491: Furthermore, the fraction of the \dem71\, interior
492: covered by the LWRS slit is large enough to make a simple plane-parallel
493: approximation inadequate for describing the two sides
494: of the shell. In the next section we describe a more realistic model
495: where we approximate the SNR as an expanding spherical shell.
496: Having used these models to disentangle the bulk and thermal
497: contributions to the line broadening, we then compute the proper shock speed(s)
498: and $T_{p}$/$T_{O}$ ratio(s) required to match the FUV line profiles for
499: \dem71. Finally, we model the
500: flux ratios for \dem71\, using our numerical shock code and constrain
501: the electron-ion equilibration and the preshock neutral fraction.
502:
503: \subsection{NUMERICAL MODELS}
504:
505: We begin by describing our plane parallel shock models, then describe
506: the spherical shell model for \dem71. We use the former models to predict
507: flux ratios for the broad Ly $\beta$ and O~VI emission lines, and use the
508: latter models to match the shapes of these lines in the LWRS spectrum of
509: \dem71.
510:
511: Before presenting our models, we note that Heng \& McCray (2006; hereafter HM06)
512: have recently presented new calculations of Balmer-dominated shock structures.
513: In their models HM06 found that the fast neutral distribution evolves
514: in shape over multiple charge exchanges, with the overall broad component
515: profile being the sum of multiple profiles (this effect was ignored
516: in our earlier works). They find that for \vs$\,\gtrsim$\,2000 \kms\,
517: the fast neutrals become ionized before they have undergone enough
518: charge exchanges to fully sample the bulk-shifted velocity distribution of the incident
519: hot protons. The resulting fast neutral distribution is shifted by a
520: bulk velocity substantially less than $\frac{3}{4}$\,\vs, and is narrower
521: than predicted from the models of Chevalier, Kirshner \& Raymond (1980) and our
522: own models (Ghavamian \etal\, 2001, 2002). HM06 find
523: this effect to be far less important at the shock speeds relevant to \dem71, so
524: the shell model results presented for that SNR in Section 5.1.1 should remain valid.
525: However, in the cases of \onine\, and \onenine\, where shock speeds are expected
526: to exceed 2000 \kms, our interpretation of the FUV line profiles changes significantly
527: if we adopt the HM06 mapping between shock speed and broad
528: hydrogen line width. In the HM06 calculations the range of shock speeds
529: implied by a given broad H~I line width is significantly narrower in the
530: limits of minimal to full electron-proton equilibration (for \vs\,$\gtrsim$2000 \kms)
531: than in our line profile calculations (Ghavamian
532: \etal\, 2001). We describe these results in more detail in Sections 5.2 and 5.3.
533:
534:
535: \subsubsection{SHOCK MODEL}
536:
537: The numerical code described by Ghavamian \etal\, (2001; 2002) computes
538: the ionization and temperature structure behind a non-radiative shock with
539: arbitrary electron-ion and ion-ion equilibration. The original purpose of the
540: code was to compute the broad-to-narrow flux ratio of the Balmer H$\alpha$ and
541: H$\beta$ lines, including the enhancement of the narrow Balmer lines
542: by Lyman line radiative transfer. We have now extended the ionization structure
543: calculation of this code to include heavier elements up to and including silicon. In addition
544: to the electrons and protons, our code now includes Coulomb equilibration
545: of the He, C and O ion temperatures behind the shock.
546:
547: At the large shock velocities expected in \onenine\,
548: and \onine\, ($V_{S}\,\gtrsim$4000 \kms), collisional ionization of neutral
549: hydrogen by He$^{+}$ ions and alpha particles becomes important (Laming \etal\, 1996), so we have included
550: these processes in the code via rates computed with the atomic close-coupling cross sections
551: of Toshima (1994). We have also updated the ionization balance to
552: include charge exchange between neutral hydrogen
553: and both He$^{+}$ ions and alpha particles, again utilizing the cross sections
554: computed by Toshima (1994). All our rate calculations assume Maxwellian particle distributions
555: and are computed in the center of mass frame of the incident and target particles in
556: the manner described by Ghavamian (1999).
557: The other update included in this shock code is the calculation of emission line
558: fluxes in broad Ly $\gamma$, C~III $\lambda$977, broad Ly $\beta$ and O~VI
559: $\lambda\lambda$1032, 1038 out the front of the shock. We computed the collisional excitation of these
560: lines by electron, proton, He$^{+}$ and alpha particle impact. The atomic
561: data used in our calculation utilize fits to collision strengths for O~VI
562: excitation by electron impact from the compilation of Merts \etal\, (1980).
563: To compute heavy ion excitation rates we have integrated the proton and alpha
564: particle excitation cross sections of O~VI from Laming \etal\, (1996), again under
565: the assumption of Maxwellian particle distributions.
566: We have also computed Ly $\beta$ and Ly $\gamma$ rates
567: from He$^{+}$ and alpha particle impact from the scaled cross sections of
568: Janev (1996).
569:
570: Although all of the observed emission lines arise from resonant
571: transitions, we have ignored the radiative transfer of the FUV photons
572: within the shock in our calculations. The integrated resonance scattering optical depth perpendicular
573: to the shock front in the broad Ly $\beta$, Ly $\gamma$, C~III and O~VI lines is generally
574: small, $\lesssim$0.03 for a preshock density of 1 cm$^{-3}$, preshock hydrogen neutral
575: fraction of 0.5 and shock speeds in the range 600$-$5000 \kms. In addition, the very
576: large Doppler widths of the FUV lines ($\sim$600$-$3000 \kms\, for the SNRs considered
577: here) ensure that most photons propagating perpendicular to the shock front escape
578: before being absorbed. The prominent absorption features observed at the centers of
579: the emission lines are almost entirely due to LMC and Galactic halo absorption.
580:
581: Before describing the spherical shell model for \dem71\, we consider the question of
582: whether the ion-ion equilibration can be measured separately from the electron-proton
583: equilibration in the LMC Balmer-dominated SNRs.
584: At shock speeds greater than 3000 \kms, FUV line emission from collisional excitation
585: by protons, He$^{+}$ ions and alpha particles becomes increasingly important. This implies that
586: the flux ratios can be sensitive not just to the electron-proton equilibration, but the proton-helium
587: ion equilibration as well. Since there are no He lines in our \fuse\, spectra of the Balmer-dominated SNRs,
588: we have no a priori knowledge of the degree of proton-He ion equilibration in these objects.
589: This raises the question of whether flux ratios uniquely determine the
590: electron-proton equilibration. The \lybovi\, ratio is most sensitive
591: to $T_{He}/T_{p}$ in shocks with minimal electron-proton equilibration. For
592: \tetp\,$\lesssim$\,0.05, the ratio \lybovi\,
593: for shocks in the range 3500$-$8000 \kms\, (appropriate for \onine\, and \onenine) increases
594: by only 10\% as $T_{He}/T_{p}$ is varied from 4 (i.e., $m_{He}/m_{p}$) to 1. Therefore, while
595: the FUV line fluxes are sensitive to \vs\, and \tetp\, in \onine\, and \onenine\, (\vs\,$\gtrsim$3000 \kms),
596: they do not vary strongly with ion-ion equilibration over the range of shock speeds
597: set by the width of the broad Ly $\beta$ profile.
598:
599: \subsubsection{SPHERICAL SHELL MODEL}
600:
601: To predict the shapes of the FUV line profiles in our \fuse\, spectrum of \dem71\, we have
602: computed the emergent intensity in broad Ly $\beta$ and O~VI from two limb-brightened,
603: hemispherical shells joined together in the plane of the sky. For each shell
604: the shock speed, preshock density and degree of temperature equilibration between
605: the protons and oxygen ions are free parameters. From Figure~1 we estimate that
606: approximately half the \dem71\, shell is subtended by the LWRS slit.
607:
608: The first step in computing the line profile within the aperture is to compute the emission
609: at a velocity $v$ and an angle $\theta$ from the center of a given shell via the well known radiative
610: transfer relation
611:
612: \begin{equation}
613: I_{v}(\theta)\,\,=\,\,S\,(1\,-\,e^{-\tau_{v}(\theta)})
614: \end{equation}
615:
616: \noindent where $S$ is the source function of the shell and $\tau$ is
617: the optical depth of the shell at the azimuthal angle $\theta$. Under the assumption of a
618: Gaussian line profile, the optical depth for the front or back side of the shell is
619:
620: \begin{equation}
621: \tau_v(\theta)\,=\,\,\tau_0\,\,exp\,(-(V\,\mp\,V_B\,cos\,\theta)^2/b^2)\,sec\,\,\theta
622: \end{equation}
623:
624: {\noindent where} \vb\,$\equiv$\,$\frac{3}{4}$\,\vs is the bulk velocity of the postshock gas,
625: and $b\,\equiv\,(2 k T / M)^{1/2}$ is the Doppler parameter of the emission line.
626: The $\mp$ symbol is negative for the approaching shell and positive
627: for the receding shell. The face-on optical depth $\tau_0$ is evaluated at the center of the line
628: profile,
629:
630: \begin{equation}
631: \tau_0\,\equiv\,\frac{\sqrt{\pi} e^2}{m_e c}\,\frac{f_{ij} \lambda_{ij}}{b}\,N
632: \end{equation}
633:
634: \noindent where $N$ is the column density through the shock of the ion and
635: $f_{ij}$ are the oscillator strength and $\lambda_{ij}$ the rest wavelength
636: of the ionic resonance line. In terms of these quantities, the source function $S$
637: is
638:
639: \begin{equation}
640: S\,=\,\frac{I_0}{\sqrt{\pi} \tau_0\,b}
641: \end{equation}
642:
643: \noindent In the limit $\tau_{0}\,sec\,\theta\,\ll\,$1, Equation (1) reduces
644: to the usual expression for limb-brightened emission $I = I_{0}\,sec\,\theta$,
645: where $I_{0}$ is the face-on emission line intensity.
646: If the shock speed, preshock density and postshock equilibration
647: are allowed to differ between the approaching and receding sides of the shell then
648: $S$, \vb, $b$ and $\tau_{0}$ will also differ between the two sides.
649: Although photons from the far side of the shell must traverse the
650: near side of the shell as they exit the SNR, we assume that there is negligible absorption
651: of these photons on the near side due to large velocity offsets. Despite the moderate blending of the Ly $\beta$
652: and O~VI lines in the LWRS spectrum of \dem71, the optical depth for absorption of highly
653: Doppler shifted photons from one profile into another is negligible within the shock,
654: so we neglect this effect in our models.
655:
656: The flux per velocity interval from each shell $F(v)$
657: is obtained by inserting Equation (2) into Equation (1) for each of the
658: two sides, then performing the integral
659:
660: \begin{equation}
661: F(v) \,\, \propto \,\, \int_{0}^{\Theta}\,I_{v}(\theta)\,cos\,\theta\,\,d\,(cos\,\theta)
662: \end{equation}
663:
664: \noindent for a sample of velocities over the line profile (we assume azimuthal
665: symmetry). The total observed profile is then the summed fluxes from the front and back sides
666: of the shell.
667:
668: In a spherical shell of radius $R$ and thickness $d$, the maximum limb brightening
669: is $sec\,\,\theta_{max}\,\approx\,\sqrt{R / 2 d}$, where $R$ is the radius
670: of the SNR and $d$ the thickness of the emitting layer. The larger the radius
671: and the thinner the emitting layer, the greater the limb brightening.
672: In our models we relate the upper limit on the angular integral in Equation~5,
673: $\Theta$, to the limb brightening factor via $\Theta\,=\,\alpha\,\theta_{max}$,
674: where 0$\,\leq\,\alpha\,\leq\,$1 is determined by the fraction of the SNR shell
675: covered by the spectrograph slit. For a slit covering the full SNR,
676: $\alpha\,=\,$1. For a preshock density of 1 cm$^{-3}$, for example,
677: $d_{O~VI}\,\sim\,$(3-60)$\times$10$^{-3}$ pc between shock speeds of
678: 500 and 5000 \kms. Taking $R\,\approx\,$10 pc for \dem71, the limb brightening
679: factor lies in the range 9\,\,$\lesssim$\,$sec\,\,\theta_{max}\,\,\lesssim$\,40,
680: so that the integrated emission line profile $F(v)$ is dominated by
681: the contribution near the edge of the shell.
682:
683: To predict the line profile shapes we have used our
684: numerical shock code to compute a grid of values for $\tau_{0}$ over a
685: wide range of shock speeds and equilibrations.
686: When computing the line profile for a given set of shock parameters,
687: we interpolated within the pre-calculated grid to compute the appropriate
688: value of $\tau_{0}$. For a given shock speed between 500 \kms\, and 5000 \kms\,
689: the O~VI column density behind the shock is relatively insensitive to the
690: shock speed, but declines with increasing electron-proton equilibration,
691: dropping by a factor of 3 between the limits of minimal and full electron-proton
692: equilibration.
693:
694: For the range of shock parameters quoted above, the line center optical depths
695: of broad Ly $\beta$, O~VI $\lambda$1032 and C~III $\lambda$977 in our models can reach values
696: $\sim$0.5-1 at the edge of the SNR shell. At these moderate optical depths the resonance
697: scattering removes photons from the line center and leaves emission line profiles
698: with slightly flattened, broadened peaks. In the case of the O~VI doublet the
699: $F(\lambda1032)$:$F(\lambda1038)$ ratio drops below the optically thin ratio of 2:1,
700: an effect that has already been observed in non-radiative shocks in the
701: Cygnus Loop (Long \etal\, 1992; Sankrit \& Blair 2002; Raymond \etal\, 2003).
702: Although consideration of resonance scattering is
703: important for interpreting the spectra of a localized filament in
704: a nearby object such as the Cygnus Loop, the net effect on the spectra of the
705: LMC Balmer-dominated SNRs is rather small. In cases where the \fuse\, aperture
706: includes edge-on emission from the SNR,
707: enough O~VI $\lambda$1032 photons are contributed from
708: positions just inside the edge of the shell to compensate for photons scattered
709: out of the line of sight from positions at the edge of the shell. Our simple
710: shell models do not include these effects, so they may over-estimate the
711: importance of resonance scattering on the FUV profiles of the LMC remnants.
712: In comparing line fluxes from these objects with predictions from our plane parallel
713: shock models we have neglected the effects of resonance scattering.
714:
715: The line profiles calculated with the spherical
716: shell model are flatter on top and exhibit slightly more extended wings than
717: Gaussian profiles. However, given the strong interstellar absorption near the centers of the
718: FUV lines and the limited signal-to-noise of most FUV line profiles, both
719: the single Gaussian profile and the spherical shell model adequately match
720: the line shapes in the \onenine\, and \onine\, spectra.
721:
722: \section{MODEL RESULTS}
723:
724: In previous modeling of Balmer-dominated SNRs (Ghavamian \etal\, 2001; 2002),
725: we estimated the degree of electron-proton temperature equilibration by
726: first measuring the width of the broad H$\alpha$ component from edge-on
727: shocks to constrain the range of shock speeds, then modeling the broad to narrow flux ratio
728: to constrain the combination of shock speed and \tetp\, consistent with all the data.
729: Unfortunately, the narrow Ly $\beta$ emission is completely absorbed by the ISM in our \fuse\,
730: spectra, as is the core of the broad Ly $\beta$ line. In addition, the widths of the FUV emission
731: lines in all our \fuse\, spectra reflect both thermal and bulk Doppler broadening.
732: Without taking into account the latter contribution to the line profile, the shock
733: velocities tend to be overpredicted.
734:
735: In the ensuing analysis we take the following approach to constraining the shock kinematics:
736: (1) We use existing measurements of the broad component H$\alpha$ width (where available) from
737: the rim of each SNR to constrain the range of shock speeds. The line widths
738: from these measurements should primarily reflect thermal broadening and
739: give the best constraint on the appropriate range of \vs. (2) Using the range
740: of constrained shock speeds from the plane parallel models, we calculate
741: fluxes in broad Ly $\beta$ and O~VI.
742: Comparing these model fluxes with the observed values, we attempt to further refine
743: our estimates of \vs\, and \tetp, while also obtaining constraints on the preshock neutral fraction of hydrogen.
744: (3) Again utilizing the shock speed constraints from the plane parallel models, we
745: use our spherical shell models to predict the shapes of the integrated Ly $\beta$
746: (and when possible, O~VI) profiles. Comparing these predictions
747: with observed profiles in the LWRS spectra, we gauge the contribution
748: of bulk Doppler broadening to the profile width. By subtracting the
749: bulk Doppler component in quadrature from the total profile width
750: in Table~2, we obtain limits on $T_p$ and $T_O$, thereby constraining \totp.
751:
752: In the following we compare the observed \lybovi\, with predictions from our plane parallel shock
753: models to constrain \tetp\, for the \dem71\, blast wave, \onine\, and \onenine. In all calculations
754: we adopt the LMC abundances of Russell \& Dopita (1992)
755: (in this case by 12 + log(O/H) = 8.35).
756:
757: \subsection{\dem71}
758:
759: \subsubsection{SOUTH RIM OF \dem71}
760:
761: The FWHM of the broad Ly $\beta$ line in our MDRS spectrum is 1365$\pm$75 \kms, nearly 40\% larger than the
762: H$\alpha$ width of 805$^{+140}_{-115}$ \kms\, measured by Ghavamian \etal\, (2003) from
763: Rutgers Fabry-Perot (RFP) observations of \dem71. There are several potential explanations for this
764: discrepancy. First, there was a lack of extensive spectral
765: coverage on either side of the H$\alpha$ profiles in the RFP spectra of Ghavamian
766: \etal\, (2003). Since the H$\alpha$ profiles were not sampled beyond $\pm$800 \kms,
767: fitting a line with wings extending well beyond that range may have resulted in an
768: anomalously low line width. On the other hand it is possible that contrary to our
769: expectation, the MDRS observation of \dem71\, did not completely isolate the limb emission.
770: In recent optical spectra of \dem71\, acquired with the Magellan 6.5m telescope (C. Rakowski
771: 2006; private communication) several broad H$\alpha$ components are observed along portions of the
772: eastern rim. The two components are of comparable width ($\sim$600 \kms), but their
773: line centroids are shifted in opposite directions in velocity. This indicates that even
774: near the limb of the SNR the slightly off-edge
775: emission can be bright enough to significantly affect the
776: observed broad component profile. The multiple nested filaments observed at the southern
777: limb of \dem71\, (Figure~2) suggest multiple tangencies between the blast wave and line of sight,
778: indicating that the emission between these tangencies is likely to include partially face-on
779: emission. Unlike the RFP spectral extraction, which cleanly isolated the H$\alpha$ emission from the southern
780: rim of \dem71, the MDRS slit is wide enough to include some of this off-tangent emission, so the
781: MDRS spectrum is more likely to be affected by this problem. If this explanation is correct, then the line widths
782: measured in our MDRS spectra would overpredict the shock velocity.
783:
784: With the above caveats in mind, we modeled
785: the \dem71\, MDRS spectrum by first using the H$\alpha$ line width of 805$^{+140}_{-115}$
786: \kms\, from Ghavamian \etal\, (2003) to set the range of shock speeds. Since the
787: shock speed is likely to remain relatively constant near the rim of \dem71, we
788: assume that the FUV line ratios reflect excitation from a single shock, even though
789: the lines exhibit some bulk broadening. Under these assumptions, we then compared
790: the predicted \lybovi\, ratios with the observed values.
791:
792: Using the relationship between hydrogen line FWHM and shock speed
793: calculated by Ghavamian \etal\, (2001), the implied range of shock speeds from the
794: H$\alpha$ broad component width (Ghavamian \etal\, 2003) is 775$-$1005 \kms\,
795: over the range \tetp\,=\,$m_{e}/m_{p}$ to 1. We calculated a grid of \lybovi\, values
796: over this range of shock speeds, combining each \tetp\, with the corresponding velocity needed
797: to match the width of the broad Ly $\beta$ line. A plot of our results is shown in the top panel of
798: Figure~8.
799:
800: For the range of shock speeds here, the electrons in the H~I neutral zone behind the shock are close to the
801: optimum temperature ($\sim$10$^{6}$~K) for exciting Ly $\beta$ close to the shock front, producing a peak in the
802: \lybovi\, curve near \tetp\,=\,0. At higher equilibrations and shock speeds, however, the charge exchange
803: rate begins to dominate over the collisional ionization rate, resulting in higher overall density of
804: fast neutral hydrogen behind the shock. The broad Ly $\beta$ flux increases accordingly, producing
805: the rise in \lybovi\, seen in Figure~8 at high equilibrations.
806:
807: Unfortunately, due to the linear dependence of \lybovi\, on
808: preshock neutral fraction, the double valued behavior of the predicted \lybovi\, at
809: low and high \tetp, and the uncertainty in the measured flux ratio,
810: we are unable to obtain a unique combination of \tetp\, and \vs\, from the plot. Although we cannot
811: unambiguously determine these two quantities, we can at least constrain the range of
812: preshock hydrogen neutral fractions in the southern rim of \dem71. If we require
813: at least one point of intersection between the range of measured and predicted \lybovi,
814: the maximum and minimum allowed preshock neutral fractions are 40\% and 20\%, respectively.
815:
816: Assuming a uniform shock speed along the southern rim of \dem71,
817: we have attempted to estimate the degree of proton-oxygen equilibration at the shock transition.
818: At the range of shock speeds considered here (800$-$1000 \kms), charge exchange behind the shock produces
819: a fast neutral distribution with a velocity width nearly equal to the width of the proton distribution.
820: Therefore, the proton temperature is just the temperature obtained from the broad
821: component H$\alpha$ width (805 \kms). If we assume that the difference between this value
822: and our measured Ly $\beta$ width is due to the effect of bulk motion of emitting material
823: within the \fuse\, aperture, then we can estimate the Doppler broadening as
824: $\sim$ (1365$^2$ - 805$^2$)$^{1/2}$ \, = \, 1100 \kms. This value, however, is larger than
825: the measured O~VI line width of 935 \kms. Therefore, $T_O$ is not well constrained
826: from our fit to the O~VI profile, and we can only estimate an upper limit on \totp under the assumption
827: that the O~VI width represents pure thermal broadening. Under these assumptions,
828: \totp\,=\,$\frac{m_{O}}{m_{p}}\,\frac{\sigma_{O~VI}^{2}}{\sigma_{p}^{2}}$, giving
829: \totp\,$\lesssim$\,8. Note that the oxygen temperature derived from the
830: O~VI line width represents an average value weighted by the postshock O~VI emissivity.
831: The O~VI zone behind the shock is extended enough for Coulomb collisions to
832: have transferred some energy from the oxygen ions to the lighter ions. Comparing our measured
833: values of \totp\, with predictions from shock models, we find the emission-weighted
834: ratio should be $\sim$10\% smaller than the immediate postshock value, so that
835: \totp\,$\lesssim\,$9 is a better estimate of the temperature
836: ratio. Clearly this is a rough limit, but it suggests that
837: a moderate amount of proton-oxygen temperature equilibration has
838: taken place at the shock transition.
839:
840: \subsubsection{LWRS SPECTRUM OF \dem71}
841:
842: As described earlier, the LWRS slit in the \dem71\, observation is filled mostly
843: with emission from the interior of \dem71. Therefore,
844: we expect each of the emission lines in Figure~3 to consist of two
845: components, one red-shifted and one blue-shifted from the rest frame of the SNR. Utilizing
846: our spherical shell model for \dem71\, we have attempted to match the shapes of the broad Ly $\beta$
847: and O~VI $\lambda$1032 lines. The faintness of the C~III $\lambda$977 feature prevents us from obtaining
848: meaningful constraints on its shape, so we have not attempted to model the C~III $\lambda$977 profile.
849:
850: We focused first on separating the thermal and geometric contributions to the width
851: of the broad Ly $\beta$ line. We attempted to match the profile shape with our spherical shell models
852: by varying \vs\, and \tetp\, together for each of the two sides of the shell contained in the aperture.
853: %Normalizing the LWRS spectrum to the peak of the broad Ly $\beta$ line,
854: The amplitudes of the frontside and backside profiles are not allowed to vary independently, but are
855: tied together via the requirement of ram pressure conservation: $n_{b}\,v_{b}^{2}\,=\,n_{f}\,v_{f}^{2}$.
856: We do not model the LMC and Galactic Halo absorption
857: in these models, but focus upon matching the wings of the observed Ly $\beta$ profile.
858:
859: Our best estimate of the Ly $\beta$ profile shape from the shell models is shown in
860: Figure~9. As a first estimate, we assumed
861: the LWRS slit was aimed at the center of the SNR
862: and covered half the projected area of that shell (approximately correct for our observational
863: setup). Interestingly, the line profiles predicted by this model (dashed line in Figure~9) overpredict
864: the emission near the center of the
865: observed profile, and give only slightly better agreement in the far wings of the profile. Varying
866: the amplitude or width of the model profile for
867: any individual component (approaching or receding side of the shell) can produce better agreement
868: toward the center of the broad Ly $\beta$ line, but the agreement then worsens in the wings of the profile.
869: Since the profiles from the front and back sides of the shell
870: are centered on the bulk velocities (=$\frac{3}{4}$\vs) of these two components, the blueshifted
871: and redshifted peaks are spread far apart, with too
872: little emission predicted at the center of the profile. We found that the
873: only solution to this problem was to include emission from edge-on shocks in the profile calculation
874: (i.e., by setting $\alpha\,=\,$1 in Equation~5),
875: despite the fact that the LWRS observation does not appear to include filaments near the edge
876: of \dem71\, (Figure~1). Under that assumption the wings of the calculated profile draw inward, with emission from
877: zero radial velocity filling in the center of the broad Ly $\beta$ line.
878: The necessary inclusion of an edge-on component
879: in our models indicates that some of the O~VI emission may be distributed differently
880: from the H$\alpha$ emission (particularly if parts of the \dem71\, blast wave are propagating
881: into fully ionized gas).
882:
883: While the addition of the edge-on component significantly improves the ability of our models to match the observed profile,
884: asymmetric flux contributions from the front and back sides of the shell are also required to match
885: the residual redshift of the broad Ly $\beta$ centroid. To reproduce the +235 \kms\, redshift of the
886: broad Ly $\beta$ feature from the LMC rest velocity (Table~2), the central peak of the Ly $\beta$ profile
887: from the far side of the shell must be higher that of the near side.
888: Keeping this constraint in mind, we adjusted the shock velocity, total preshock density
889: and electron-proton equilibration for the two sides of the shell while conserving
890: ram pressure as described earlier. The final
891: result of our Ly $\beta$ profile calculation for \dem71\, is marked
892: with the solid line in Figure~9.
893:
894: Using the calculated relationship between the velocity width of the proton distribution and
895: the width of the H~I profile transmitted by charge exchange (Chevalier, Kirshner \& Raymond 1980;
896: Ghavamian \etal\, 2001), the best combination of Ly $\beta$ profile
897: widths from our spherical shell model are 920 \kms\, for the approaching side of the shell and
898: 585 \kms\, FWHM for the receding side of the shell. The corresponding combination of
899: shock parameters (\vs, $n_{H^{0}}$, \tetp) are (900, 0.5, 0.05) for the
900: approaching side of the shell and (650, 1.0, 0.6) for the receding side of the shell. Note
901: that the absolute quoted values of $n_{H^{0}}$ are not the meaningful quantities here, but rather
902: their ratio, as required by ram pressure conservation.
903:
904: Although there is some degeneracy between the various shock parameters estimated above,
905: they are fully consistent with prior measurements of the broad component
906: H$\alpha$ width, shock speed and equilibration estimated for \dem71\, by Smith \etal\, (1991), Ghavamian \etal\, (2003) and Rakowski
907: \etal\, (2003). The analyses in these papers indicated that the blast wave of \dem71\,
908: has decelerated to $\sim$600 \kms\, in the northeastern corner of the SNR, precisely where the
909: \fuse\, LWRS slit grazes the edge. On the other hand, the RFP H$\alpha$ spectra and \chan\, X-ray analyses
910: of Ghavamian \etal\, (2003) and Rakowski \etal\, (2003) indicated a nearly 50\% variation
911: in shock speed around the rim of of \dem71, with the highest shock speeds ($\sim$800-1000 \kms)
912: measured along the southern rim. The results of our broad Ly $\beta$ profile models
913: indicate that the far side of \dem71\, may be propagating into a higher density medium
914: than the near side, with a correspondingly lower shock speed on the far side of the SNR.
915:
916: After applying our shell models to the broad Ly $\beta$ profile, we modeled the O~VI $\lambda$1032
917: profile. As a first guess we set the shock speeds of the front and
918: back sides of the shell equal to the values derived from the broad Ly $\beta$ profile model, with
919: the O~VI line width a free parameter. Our best estimates for the O~VI line widths are
920: 475 \kms\, FWHM for the receding shell and 900 \kms\, FWHM for the approaching shell. The centroid of the
921: observed O~VI $\lambda$1032 feature is shifted by the same velocity as the centroid of the Ly $\beta$
922: feature (Table~2), and can be matched with a 2:1 ratio of preshock density between the back and
923: front sides of the shell. Taken together, the Ly $\beta$ and
924: O~VI $\lambda$1032 results suggest that while the front and back sides of \dem71\, are
925: propagating into an ISM with a 2:1 ratio in total density, the hydrogen neutral fractions
926: are similar on both sides of the shell.
927:
928: From our above analysis, we can estimate \totp\, for the approaching and receding sides
929: of the shell in \dem71. Given \vs\,=\,650 \kms, \tetp\,=\,0.6 and an estimated O~VI
930: $\lambda$1032 width of 475 \kms\, on the back side, we obtain \totp\,$\approx\,$6.5.
931: On the near side, for shock speed of 900 \kms, \tetp\,=\,0.05, we obtain \totp\,$\approx$\,16.
932: Although these estimates are subject to the uncertainty of deconvolving the thermal
933: broadening from the bulk Doppler broadening in our spectra, both the magnitude and
934: declining trend of \totp\, with shock speed are consistent with results
935: established for the Balmer-dominated shocks in other SNRs such as the Cygnus Loop
936: (Raymond \etal\, 2003) and SN 1006 (Laming \etal\, 1996; Korreck \etal\, 2004).
937:
938: \subsection{\onine}
939:
940: The sole emission line positively detected in the LWRS spectrum of this Balmer-dominated
941: SNR is broad Ly $\beta$ (Figure~4). The fitted FWHM (3710$\pm$400 \kms) is the
942: largest hydrogen line width measured in a Balmer-dominated SNR to date.
943: The centroid of the broad Ly $\beta$ line is substantially blueshifted ($-$540 \kms)
944: relative to LMC frame, suggesting that the FUV emission from \onine\, primarily arises
945: from the near side of the SNR shell. It is likely that the broad Ly $\beta$ profile
946: is affected by bulk Doppler motions. However, unlike the other three Balmer-dominated SNRs,
947: there has been no prior detection of broad ionic line emission from \onine, so we have no
948: clear constraint on the appropriate range of shock speeds for modeling the profile in Figure~4.
949:
950: Since we do not have enough information for disentangling the bulk Doppler
951: contribution of broad Ly $\beta$ from the thermal Doppler contribution, we
952: proceed with a plane parallel shock analysis under the crude assumption that the broad
953: Ly $\beta$ width in \onine\, reflects purely thermal broadening.
954: Using the calculations of HM06, the measured FWHM corresponds
955: to shock speeds of 5200$-$6300 \kms\, over the range $m_{e}/m_{p}\,\leq\,$\tetp$\,\leq\,$1
956: (for comparison, the FWHM calculations of Ghavamian \etal\, (2001, 2002) give
957: a range of 5610$-$7780 \kms).
958:
959: Although the O~VI lines are not detected in the \fuse\, spectrum
960: of \onine, we can still place a lower limit on both \lybovi\,
961: and the widths of the O~VI lines. First, we estimate an upper limit on the O~VI
962: $\lambda$1032 flux. The upper limit on the flux per wavelength interval from the
963: \fuse\, spectrum of \onine\, is $\approx$ 2$\times$10$^{-15}$ ergs cm$^{-2}$ s$^{-1}$
964: \AA$^{-1}$ at the center of the O~VI $\lambda$1032 line. To convert this number to
965: a flux, we require a limit on the O~VI line width. For each shock speed in the range quoted
966: earlier, the FWHM of the O~VI line is related to the oxygen temperature via
967: $V_{FWHM}(O~VI)\,=\,\,\sqrt{8\,ln\,2}\,\sqrt{k\,T_{O}/m_{O}}$, so that full
968: proton-oxygen equilibration would produce an O~VI line width $\frac{1}{4}$
969: as broad as the proton velocity distribution, while no equilibration would
970: yield equal widths. Over the full range of allowed equilibrations,
971: the predicted O~VI line width ranges from 1245 \kms\, (for \tetp\,=\,\totp\,=\,1)
972: to 5300 \kms\, (for \tetp\,=\,1 and \totp\,=\,16).
973: \kms\, (equilibrated). The corresponding limit on the O~VI $\lambda$1032 flux
974: is then F(1032)\,=\,(0.9-3.9)$\times$10$^{-14}$ \fl\, in the limits of full and minimal
975: proton-oxygen equilibration respectively.
976:
977: In Figure~8 we present the predicted \lybovi\, ratio for \onine\, for the range
978: of shock speeds implied by the width of the broad Ly $\beta$ profile.
979: The lower limit on the \lybovi\, ratio corresponds to the largest O~VI
980: line width ($\sim$5300 \kms) allowed from our analysis above.
981: For a given broad Ly $\beta$ width and corresponding combination of shock parameters (\tetp, \vs),
982: the O~VI emission rises more rapidly with increasing \tetp\, (and shock speed) than
983: the broad Ly $\beta$ emission, causing the modest decline in the predicted \lybovi\,
984: (Figure~8). Our models are shown for two limiting preshock neutral fractions:
985: $f_{H^{0}}\,=\,$0.4 and 0.8. The lower neutral fraction is the smallest value allowing
986: an overlap between our model curves and the observed range. The higher fraction
987: is the largest value likely for the warm ISM, given the minimum photoionization
988: expected from both the SNR and the ambient UV and soft X-ray backgrounds.
989:
990: We can draw several conclusions from the middle panel of Figure~8.
991: First, efficient proton-oxygen ion equilibration is ruled out. In such a case,
992: the narrow O~VI line width (a minimum of 1245 \kms\, FWHM for \totp\,=\,1) would require
993: a very low total O~VI $\lambda$1032 flux to stay consistent with the non-detection
994: of O~VI in the LWRS spectrum. This would raise the observed \lybovi\, ratio
995: (i.e., the lower edge of the shaded region) to $\sim$ 9, which cannot be matched
996: by any model regardless of equilibration or preshock neutral fraction. The
997: prediction from our models is that \lybovi\,$\lesssim$4.5, a constraint which is
998: satisfied for $V_{FWHM}(O~VI)\,\gtrsim\,$2400 \kms. This places the proton-oxygen
999: ion equilibration in the range 4\,$\lesssim\,T_{O}/T_{p}\,\leq\,$16 (the assumed upper limit corresponds to the
1000: oxygen to proton mass ratio, though \totp\,$>$\,16 has been observed
1001: on occasion in solar wind shocks). We cannot place a strong constraint on \tetp\,
1002: from Figure~8, but we can deduce that the preshock gas should be significantly
1003: neutral, 0.4\,$\lesssim\,f_{H^{0}}\,\lesssim$\,0.8.
1004:
1005: We can use our derived shock velocity for \onine\, to estimate its age.
1006: Assuming pure thermal broadening for the detected Ly $\beta$ line and
1007: taking into account the measurement uncertainty in the broad Ly $\beta$
1008: width, the predicted shock speed is 3600$-$7100 \kms\, over the range
1009: $m_e/m_p\,\leq$\,\tetp\,$\leq$\,1. We compute the age $\tau$ by assuming
1010: a radius of 3.6 pc (Warren \& Hughes 2004) and using $\tau\,\equiv\,\delta\,\frac{R}{V}$,
1011: where $\delta$ is the expansion parameter. Between the limits of free expansion and
1012: the asymptotic Sedov-Taylor limit, $\delta$ ranges from around 1.0 down to 0.4.
1013: Our derived shock speeds are significantly
1014: lower than the 10,000-20,000 \kms\, expansion velocities predicted for Type Ia SNe,
1015: indicating that some deceleration of the blast wave has occurred. Assuming standard
1016: Type Ia parameters and a constant ISM density, Dwarkadas \&
1017: Chevalier (1998) predicted that $\delta$ declines to around 0.6 during the first few hundred
1018: years after the SN explosion. Using this expansion parameter, we find $\tau\,\approx\,$295-585
1019: years for \onine. This value compares favorably with the light echo age of 400$\pm$120
1020: years found for SNR by Rest \etal\, 2005).
1021: Given the neglect of bulk Doppler broadening in our analysis, we are at the very least encouraged
1022: by the broad agreement between our derived age for \onine\, and the independent age estimate
1023: of Rest \etal\, (2005).
1024:
1025: \subsection{\onenine}
1026:
1027: \subsubsection{PLANE PARALLEL MODEL RESULTS}
1028:
1029: The broad component H$\alpha$ width measured along the rim of \onenine\,
1030: by Tuohy \etal\, (1982) was 2800$\pm$300 \kms, comparable to the value
1031: measured in the Galactic Balmer-dominated SNR
1032: SN 1006 (Ghavamian \etal\, 2002; Korreck \etal\, 2004). From the
1033: line profile calculations of HM06 this width
1034: corresponds to shock speeds in the range 3200$-$3900 \kms\, in the
1035: limits $m_{e}/m_{p}\,\leq\,$\tetp$\,\leq\,$1 (again these are
1036: considerably narrower values than the range of 4115$-$5930 \kms\,
1037: predicted by our own models).
1038:
1039: The predicted \lybovi\, ratios for \onenine are shown in the bottom panel of Figure~8.
1040: It is clear that in this case the flux ratios do not vary strongly with electron-proton
1041: equilibration, partly a result of the narrow range of shock speeds implied by the
1042: width of the broad Ly $\beta$ line. The \lybovi\, curves are shown
1043: for two limiting preshock neutral fractions: the lower fraction ($f_{H^{0}}\,=\,$0.4) being the smallest
1044: value allowing overlap between our model curves and the observed range, with the
1045: higher fraction ($f_{H^{0}}\,=\,$0.8) being the largest value likely for the warm ISM
1046: surrounding the SNR. We cannot place a strong constraint on \tetp\, from
1047: Figure~8. However, the preshock neutral fraction is well constrained,
1048: 0.4\,$\lesssim\,f_{H^{0}}\,\lesssim$0.5.
1049:
1050: Despite the considerable uncertainty ($\sim$40\%; Table~2) in the O~VI line width,
1051: we can still place reasonable limits on the degree proton-oxygen ion equilibration in the
1052: blast wave of \onenine. The FWHM of the broad Ly $\beta$ line is
1053: 3130$\pm$155 \kms, while that of the O~VI $\lambda$1032 is 4975$\pm$1830 \kms.
1054: The contribution to the FUV profiles from bulk Doppler broadening, $V_{B}$, can be estimated
1055: by subtracting the Tuohy \etal\, (1982) H$\alpha$ width from the observed
1056: Ly $\beta$ width, $V_{B}\,\approx\,\sqrt{3130^2 - 2800^2}$\,=\,1400 \kms.
1057: Assuming the same bulk broadening for the O~VI line, the thermal width
1058: of the oxygen ion distribution is then 4800$\pm$1750 \kms. Since the O~VI line
1059: width cannot significantly exceed that of broad Ly $\beta$, our result can only be
1060: consistent with equal line widths, or \totp\,$\approx\,$16.
1061:
1062: The age obtained by Rest \etal\, (2005) from their light echo study of \onenine\,\,
1063: is 600$\pm$200 years. Taking into account the measurement uncertainty in the broad
1064: H$\alpha$ width, the full range of shock speeds for \onenine\, between the limits
1065: of minimal and full electron-proton equilibration is 2600$-$4500 \kms. Taking
1066: $\delta\,\approx\,$0.6, this results in an age of 520$-$900 yrs, in good overall
1067: agreement with the age predicted by the light echo study.
1068:
1069: \subsubsection{SPHERICAL SHELL MODEL RESULTS}
1070:
1071: Using the above constraints on the shock speed and equilibration, we
1072: have computed the shape of the broad Ly $\beta$ line produced by a spherical shell.
1073: Using Figure~7 of HM06 we find that for an H$\alpha$ line width
1074: of 2800 \kms\, (the edge-on shock value of Tuohy \etal\, 1982) the predicted bulk velocity of
1075: the fast neutrals is $\sim\,\frac{1}{2}$\,\vs, rather than the usually
1076: assumed value of $\frac{3}{4}$\,\vs. In that case
1077: 2000\,$\lesssim\,V_{B}\,\lesssim$\,3000 \kms\, for \onenine\, in the range
1078: $m_e/m_p$\,$\leq$\,\tetp\,$\leq$\,1. In addition, since the centroids of the FUV emission lines in \onenine\,
1079: (Table~2) are consistent with the LMC rest velocity, we assume equal shock speeds and
1080: preshock densities for the front and back hemispheres. Using all of these parameters,
1081: we were able to match the observed broad Ly $\beta$ profile with a thermal broadening
1082: FWHM of 2500 \kms, consistent with the lower bound on the H$\alpha$
1083: width of Tuohy \etal\, (1982). The corresponding shock speed is approximately
1084: 2600 \kms\, from Figure~7 of HM06, with a correspondingly low
1085: electron-proton equilibration (\tetp\,$\lesssim$0.2). This result is consistent
1086: with shock speeds at the lower end of the range predicted in the previous section.
1087: It worsens the disagreement between the light echo age of Rest \etal\, (2005)
1088: and our dynamical age estimate, and may provide further evidence of blast
1089: wave deceleration.
1090:
1091: \section{THE EFFECT OF ASSUMED ABUNDANCES}
1092:
1093: It is clear that our interpretations above are significantly affected by our
1094: choice of oxygen abundance via the linear dependence of the \lybovi\, ratio on the O/H abundance ratio.
1095: Hughes, Hayashi \& Koyama (1998)
1096: obtained a lower average oxygen abundance than Russell \& Dopita (1992)
1097: from their analyses of {\it ASCA} spectra of LMC SNRs: 8.21$\pm$0.07 dex.
1098: %On the other hand,
1099: %oxygen abundances as high as 8.54 have been measured in the N49 SNR (Russell \& Dopita 1992).
1100: It is instructive to consider the effect of such lower O/H ratios on the comparison of our shock
1101: models with the observations. If ISM oxygen abundances around all four Balmer-dominated SNRs were
1102: set to the values derived by Hughes, Hayashi \& Koyama (1998), the modeled O~VI $\lambda$1032
1103: flux would decrease by 28\%, raising the predicted \lybovi
1104: ratios in Figure~8 and lowering the range of acceptable preshock neutral fractions.
1105: In the case of \dem71, the predicted range of acceptable preshock neutral fractions
1106: becomes 15\%-30\%, while a fraction above 30\% is allowed for \onine. The neutral
1107: fraction for \onenine\, is most tightly constrained, with a value close to 30\% required
1108: to match the new \lybovi ratio.
1109:
1110: \section{REDDENING CONSTRAINTS FROM THE FUV SPECTRA OF \dem71\, AND \onenine }
1111:
1112: Aside from the kinematic diagnostics available in the FUV spectra of the Balmer-dominated
1113: SNRs, the ratio of Ly $\beta$ to Ly $\gamma$ flux is sensitive to the interstellar reddening,
1114: providing further useful information on these objects. Over the range of \vs\, and
1115: \tetp\, considered in this paper, the impacting electrons and ions have kinetic energies
1116: greatly exceeding the energy difference between the n\,=\,3 and n\,=\,4 levels of hydrogen,
1117: and the ratio of these two lines reduces to the ratio of their collision strengths. In
1118: addition, the Ly $\beta$ and Ly $\gamma$ lines are optically thin due to their
1119: large Doppler widths and the low density of the postshock gas. This allows us
1120: to ignore radiative transfer of Ly $\beta$ and Ly $\gamma$ photons when computing
1121: their flux ratio. The ratio is then nearly constant, ranging from approximately
1122: 3.3 for minimal equilibration to 3.1 for maximal equilibration.
1123:
1124: In the LWRS spectra of \dem71\, and
1125: \onenine, where emission from both lines has been detected, the Ly $\beta$ / Ly $\gamma$ ratios
1126: are 3.7$\pm$0.4 and 3.4$\pm$0.5, respectively. Comparing these numbers with the predicted
1127: values, these numbers are consistent with low to minimal reddening. Koornneef \& Code (1981) noted that the slopes
1128: of the Galactic and LMC extinction curves are very similar at $\lambda$\,$<$\,1400 \AA. This implies
1129: that for the ratio of two line fluxes in this wavelength range, the correction will be
1130: nearly identical whether we use an LMC or Galactic extinction law. Indeed, the
1131: range of extinctions we derive for \dem71\, and \onenine\, using a Galactic
1132: reddening law (Cardelli, Clayton \& Mathis 1989) are very similar to the
1133: range derived for the non-30 Doradus regions of the LMC (Fitzpatrick 1985).
1134: The color excess predicted for \dem71\,
1135: is then E(B\,$-$\,V)\,=\,0.05-0.13, while for \onenine\, we find E(B\,$-$\,V)\,$\leq\,$0.11.
1136:
1137: \section{DISCUSSION AND CONCLUDING REMARKS}
1138:
1139: We have presented the first far ultraviolet spectra of the four known Balmer-dominated SNRs in
1140: the LMC. These objects - \dem71, \onine, \onenine\, and \foureight, are the non-radiative (adiabatic) remnants
1141: of Type Ia SN explosions. They feature optical spectra dominated by Balmer line collisional excitation of ambient
1142: H~I gas overrun by the blast wave. Our \fuse\, observations resulted in the detection
1143: of three of the four SNRs, \dem71, \onine\, and \onenine.
1144: The detected emission lines $-$ C~III $\lambda$977,
1145: O~VI $\lambda\lambda$1032, 1038, Ly $\beta$ and Ly $\gamma$ $-$ are very broad (600 $-$ 3700 \kms) and are generated
1146: in the thin ionization
1147: zones behind the non-radiative shocks. In the case of \onine, the only detected feature
1148: is faint, broad Ly $\beta$ (FWHM of 3710 \kms); however,
1149: this is the first and only broad emission line detected in this SNR since its discovery by Tuohy \etal\, (1982).
1150:
1151: The relative fluxes of the FUV emission lines are sensitive to the shock velocity, preshock neutral fraction
1152: and the degree of electron-proton equilibration at the shock front. On the other hand,
1153: the widths of the FUV emission lines in Balmer-dominated SNRs are proportional to the relative
1154: temperatures of the heavy ions, making them useful diagnostic probes of the degree of ion-ion temperature
1155: equilibration in collisionless shocks. These properties make FUV spectra of these SNRs
1156: valuable tools for probing the collisionless heating in high Mach number
1157: collisionless shocks.
1158:
1159: Comparing the \lybovi\, flux ratios
1160: with predictions from plane parallel and spherical shell models of non-radiative shocks, we find that the observed
1161: flux ratios are consistent to within the errors with low to minimal electron-proton equilibration
1162: in shocks faster than 2500 \kms\, (observed in \onine\, and \onenine). On the other hand, spectra
1163: from the interior and limb of \dem71\, (\vs\,$\sim$600$-$1000 \kms) show evidence for small to moderate electron-proton
1164: temperature equilibration. Our results are generally
1165: consistent with the inverse trend between electron-proton equilibration and Mach number derived
1166: from optical (Ghavamian \etal\, 2001, 2002; Rakowski 2005, Ghavamian, Laming \& Rakowski 2007) and joint optical
1167: and X-ray (Ghavamian, Laming \& Rakowski 2007; Rakowski, Ghavamian \& Hughes 2003) observations of
1168: Balmer-dominated shocks in SNRs.
1169:
1170: We have estimated \totp\, at the shock transition
1171: by comparing the broad Ly $\beta$ and O~VI emission line widths in our \fuse\, spectra of the
1172: Balmer-dominated LMC remnants. Over the range of shock speeds
1173: sampled by our \fuse\, observations ($\sim$600\,$\lesssim$\,\vs\,$\lesssim$\,6000 \kms) our observations
1174: indicate that \totp\, increases from $\sim$ 7 for shock speeds near 600 \kms\, to $\sim$ 16 (mass-proportional
1175: heating) for shock speeds exceeding 1000 \kms. A summary of our
1176: modeling results is shown in Table~3. From the shock velocities listed in Table~3
1177: we obtain ages of 295-585 years for \onine\, and 520-900 years
1178: for \onenine, respectively, in good agreement with the ages estimated for these SNRs
1179: by SN light echo studies (Rest \etal\, 2005).
1180:
1181: The main source
1182: of uncertainty in our estimates of \totp\, is our uncertainty in the O~VI emission line widths.
1183: In the case of \onenine, the faintness of the O~VI lines and their heavy blending with broad
1184: Ly $\beta$ makes a precise measurement of their widths difficult. In the case of \onine,
1185: the O~VI emission falls below the detection threshold, allowing only an indirect estimate
1186: of the O~VI line widths. Although line blending is far less severe in our FUV spectrum of
1187: the southern rim of \dem71\, all the emission lines are intrinsically very faint, again making
1188: the measurement of the O~VI line widths (and hence \totp) uncertain. Line blending is also
1189: less severe in our \fuse\, spectrum of the interior of \dem71. However, disentangling the frontside
1190: from backside emission in this case is the dominant source of uncertainty in measuring \totp.
1191:
1192: Modulo the bulk Doppler broadening, our results are consistent
1193: with an inverse relationship between the degree of ion-ion equilibration and shock speed, with slower shocks
1194: exhibiting more prompt equilibration at the shock front than faster shocks. This trend is similar
1195: to that found for the electron-proton temperature equilibration, suggesting that the
1196: collisionless processes responsible for
1197: prompt electron-proton equilibration at low shock speeds may also cause prompt
1198: ion-ion equilibration at low shock speeds. A stronger verification of this trend will require
1199: further measurements of \totp\, in other non-radiative shocks at intermediate velocities
1200: (500$-$2500 \kms). Obtaining these observations will remain the subject of future work.
1201:
1202: P.G. would like to thank R. C. Smith for providing H$\alpha$ images of the
1203: Balmer-dominated LMC remnants and J. M. Laming for helpful discussions on the atomic
1204: physics of heavy ion excitation. P. G. acknowledges support from NASA grants NNG04GL79G
1205: and NAS5-32985, and W.P.B. acknowledges NASA grants NNG04GD150 and
1206: NNG05GD75G, all to Johns Hopkins University.
1207:
1208: \begin{references}
1209:
1210: \reference{B05}
1211: Blair, W. P., Ghavamian, P., Sankrit, R. \& Danforth, C. 2006, \apjs\, 165, 480
1212:
1213: \reference{B91}
1214: Blair, W. P., Long, K. S. \& Vancura, O. 1991, \apj, 366, 484
1215:
1216: \reference{C89}
1217: Cardelli, J. A., Clayton, G. C. \& Mathis, J. S. 1989, \apj, 345, 245
1218:
1219: \reference{CR}
1220: Chevalier, R. A. \& Raymond, J. C. 1978, \apj, 225, L27
1221:
1222: \reference{CKR80}
1223: Chevalier, R. A., Kirshner, R. P. \& Raymond, J. C. 1980, \apj, 235, 186
1224:
1225: \reference{D76}
1226: Davies, R. D., Elliott, K. H. \& Meaburn, J. 1976, \memras, 81, 89
1227:
1228: \reference{D93}
1229: Draine, B. T. \& McKee, C. F. 1993, \araa, 31, 373 (p 410)
1230:
1231: \reference{DC98}
1232: Dwarkadas, V. V. \& Chevalier, R. A. 1998, \apj, 497, 807
1233:
1234: \reference{F85}
1235: Fitzpatrick, E. L. 1985, \apj, 299, 219
1236:
1237: \reference{G99}
1238: Ghavamian, P., 1999, PhD thesis, Rice University
1239:
1240: \reference{G01}
1241: Ghavamian, P., Raymond, J., Smith, R. C. \& Hartigan, P. 2001, \apj, 547, 995
1242:
1243: \reference{GRHW03}
1244: Ghavamian, P., Rakowski, C. E., Hughes, J. P. \& Williams, T. B. 2003, \apj, 590, 833
1245:
1246: \reference{G07}
1247: Ghavamian, P., Laming, J. M. \& Rakowski, C. E. 2007, \apjl, 654, L69
1248:
1249: \reference{HRH87}
1250: Hartigan, P., Raymond, J. \& Hartmann, L., 1987, \apj, 316, 323
1251:
1252: \reference{H03}
1253: Hendrick, S. P., Borkowski, K. J. \& Reynolds, S. P. 2003, \apj, 593, 370
1254:
1255: \reference{H01}
1256: Hendrick, S. P. \& Reynolds, S. P. 2001, \apj, 559, 903
1257:
1258: \reference{HM06}
1259: Heng, K., \& McCray, R. 2006, \apj, in press (astroph/0609331) (HM06)
1260:
1261: \reference{HRB}
1262: Hester, J. J., Raymond, J. C. \& Blair, W. P. 1994, \apj, 420, 721
1263:
1264: \reference{H02}
1265: Howk, J. C., Sembach, K. R., Savage, B. D., Massa, D., Friedman, S. D. \& Fullerton, A. W., 2002, \apj, 569, 233
1266:
1267: \reference{H98}
1268: Hughes, J. P., Hayashi, I. \& Koyama, K. 1998, \apj, 505, 732
1269:
1270: \reference{HGRS03}
1271: Hughes, J. P., Ghavamian, P., Rakowski, C. E. \& Slane, P. O. 2003, \apj, 582, L95
1272:
1273: \reference{J96}
1274: Janev, R. K., 1996, \pra, 53, 219
1275:
1276: \reference{KWC87}
1277: Kirshner, R. P., Winkler, P. F. \& Chevalier, R. A. 1987, \apj, 315, L135
1278:
1279: \reference{K81}
1280: Koornneef, J., \& Code, A. D. 1981, \apj, 247, 860
1281:
1282: \reference{KRZG04}
1283: Korreck, K. E., Raymond, J. C., Zurbuchen, T. H. \& Ghavamian, P. 2004, \apj, 615, 280
1284:
1285: \reference{K94}
1286: Kriss, G., 1994, {\it Astronomical Data Analysis Software and Systems III}, ASP Conf.
1287: Series, Vol. 61, p 437
1288:
1289: \reference{LRMB96}
1290: Laming, J. M., Raymond, J. C., McLaughlin, B. M. \& Blair, W. P. 1996, \apj, 472, 267
1291:
1292: \reference{L81}
1293: Long, K. S., Helfand, D. J. \& Grabelsky, D. A. 1981, \apj, 248, 925
1294:
1295: \reference{LB90}
1296: Long, K. S. \& Blair, W. P. 1990, \apj, 358, L13
1297:
1298: \reference{L92}
1299: Long, K. S., Blair, W. P., Vancura, O., Bowers, C. W., Davidsen, A. F. \& Raymond, J. C.
1300: 1992, \apj, 400, 214L
1301:
1302: \reference{M80}
1303: Merts, A. L., Mann, J. B., Robb, W.D. \& Magee, N. H., Jr., 1980, Los Alamos Informal
1304: Report LA-8267-MS
1305:
1306: \reference{M00}
1307: Moos, H. W. \etal\, 2000, \apj, 538, L1
1308:
1309: \reference{RGH03}
1310: Rakowski, C. E., Ghavamian, P. \& Hughes, J. P. 2003, \apj, 590, 846
1311:
1312: \reference{R05}
1313: Rakowski, C. E. 2005, AdSpR, 35, 1017
1314:
1315: \reference{R095}
1316: Raymond, J. C., Blair, W. P. \& Long, K. S. 1995, \apj, 454, L31
1317:
1318: \reference{R01}
1319: Raymond, J. C. 2001, AdSpR 99, 209
1320:
1321: \reference{R03}
1322: Raymond, J. C., \etal\, 2003, \apj, 584, 770
1323:
1324: \reference{R05}
1325: Rest, A., \etal\, 2005, \nat, 438, 1132
1326:
1327: \reference{R90}
1328: Russell, S. C. \& Dopita, M. A. 1990, \apjs, 74, 93
1329:
1330: \reference{R92}
1331: Russell, S. C. \& Dopita, M. A. 1992, \apj, 384, 508
1332:
1333: \reference{S00}
1334: Sahnow, D. J. \etal\, 2000, \apj, 538, L7
1335:
1336: \reference{SB02}
1337: Sankrit, R., \& Blair, W. P., 2002, \apj, 565, 297
1338:
1339: \reference{SKBW91}
1340: Smith, R. C., Kirshner, R. P., Blair, W. P. \& Winkler, P. F. 1991, \apj, 375, 652
1341:
1342: \reference{SRL}
1343: Smith, R. C., Raymond. J. C. \& Laming, J. M. 1994, \apj, 420, 286
1344:
1345: \reference{T84}
1346: Toshima, N., 1994, \pra, 50, 3940
1347:
1348: \reference{T82}
1349: Tuohy, I. R., Dopita, M. S., Mathewson, D. S., Long, K. S. \& Helfand, D. J. 1982, \apj, 261, 473
1350:
1351: \reference{T81}
1352: Treffers, R. R. 1981, \apj, 250, 213
1353:
1354: \reference{W04}
1355: Warren, J. S. \& Hughes, J. P. 2004, \apj, 608, 261
1356:
1357: \end{references}
1358:
1359: \clearpage
1360:
1361: \input{table1.tex}
1362: \input{table2.tex}
1363: \input{table3.tex}
1364:
1365: \begin{figure}
1366: \plotone{f1.ps}
1367: \caption{Continuum-subtracted narrow band H$\alpha$ images of the four Balmer-dominated SNRs (CTIO 4m images
1368: courtesy R. C. Smith). The positions of the \fuse\, apertures (LWRS: 30\arcsec\,$\times$\,30\arcsec\, squares;
1369: MDRS: 4\arcsec\,$\times$\,20\arcsec\, rectangle) are marked.
1370: The wavelength dispersion direction for each of the LWRS spectra shown in Figure~2 are marked in the upper
1371: left corner of each panel. The dispersion direction for the MDRS pointing in \dem71\, is perpendicular
1372: to the long dimension of the slit. }
1373: \end{figure}
1374:
1375:
1376: \begin{figure}
1377: \plotone{f2.ps}
1378: \caption{Combined raw \fuse\, spectra of the four Balmer-dominated SNRs. The
1379: 1A channel data (1013 \AA\, $-$ 1067 \AA) are shown in the figure. Positions of
1380: major emission line features at the LMC systemic velocity (+275 \kms) are marked.
1381: The broad Ly $\beta$ and O~VI emission
1382: lines are clearly seen in the spectra of \dem71, \onine\, and \onenine. A particularly strong
1383: C~II absorption feature can be seen in the spectrum of \onenine\, near 1037 \AA, arising from
1384: LMC halo absorption. There
1385: are two sets of spectra imaged on each detector (marked LiF 1A and SiC 1A). Each
1386: set consists of spectra from the medium resolution (MDRS, at top), high
1387: resolution (HIRS, in middle) and low resolution (LWRS, bottom) apertures. The
1388: object spectra are acquired through the LWRS aperture, with the remaining
1389: apertures containing only residual airglow. }
1390: \end{figure}
1391:
1392: \begin{figure}
1393: \plotone{f3.eps}
1394: \caption{Extracted LWRS spectrum of the face-on shock position in \dem71. The
1395: best fit line profiles are marked, along with the positions of major interstellar
1396: features from LMC (top tick marks) and Galactic (bottom tick marks) absorption. The model components
1397: included in the fit are shown in Table~2. }
1398: \end{figure}
1399:
1400: \begin{figure}
1401: \plotone{f4.eps}
1402: \caption{Extracted LWRS spectrum of \onine. The
1403: best fit line profiles are marked, along with the positions of major interstellar
1404: features from LMC (top) and Galactic (bottom) absorption. The ISM absorption
1405: features have been excluded from the fit. The model components
1406: included in the fit are shown in Table~2. }
1407: \end{figure}
1408:
1409: \begin{figure}
1410: \plotone{f5.eps}
1411: \caption{Extracted LWRS spectrum of \onenine. The
1412: best fit line profiles are marked, along with the positions of major interstellar
1413: features from LMC (top) and Galactic (bottom) absorption. The ISM
1414: absorption features have been excluded from the fit. The model components
1415: included in the fit are shown in Table~2. }
1416: \end{figure}
1417:
1418: \begin{figure}
1419: \plotone{f6.eps}
1420: \caption{Extracted LWRS spectrum of \foureight. No FUV emission is detected
1421: down to the S/N = 3 sensitivity limit of \fuse\, ($\sim$10$^{-15}$ ergs cm$^{-2}$ s$^{-1}$ \AA$^{-1}$.
1422: at 1032 \AA\, for night only exposures. }
1423: \end{figure}
1424:
1425: \begin{figure}
1426: \plotone{f7.eps}
1427: \caption{The MDRS spectrum of \dem71, acquired from the southern
1428: rim of the SNR. The fitted emission line components (broad Ly $\beta$,
1429: O~VI $\lambda\lambda$1032, 1038 and Galactic + LMC H~I absorption are
1430: marked. The Ly $\beta$ airglow line (marked by the $\oplus$) is excluded
1431: from the fit. }
1432: \end{figure}
1433:
1434: \begin{figure}
1435: \plotone{f8.eps}
1436: \caption{The predicted broad \lybovi\, flux ratios, shown as a function of the
1437: assumed electron-proton temperature equilibration for the three FUV spectra
1438: dominated by edge-on blast wave emission. The measured ranges of the \lybovi\,
1439: ratios are marked by the shaded regions. The solid and dotted lines show
1440: the predicted flux ratio curves for the maximum and minimum preshock neutral
1441: fractions allowing agreement between the observed and predicted flux ratios.
1442: In the case of \onine\, the maximum preshock neutral fraction (80\%) is
1443: the largest plausible value given ionization conditions in the warm neutral ISM.
1444: }
1445: \end{figure}
1446:
1447: \begin{figure}
1448: \plotone{f9.eps}
1449: \caption{The observed and modeled Ly $\beta$ profile from the face-on spectrum
1450: of \dem71\, (Figure~3). The solid line shows our best approximation to the profile
1451: from our spherical shell models (see text for details). The dashed line shows our
1452: best approximation when the LWRS aperture (Figure~1) extends only halfway to
1453: the edge of the shell (i.e., no edge-on component is included). The prominent
1454: absorption feature and Ly $\beta$ airglow line between $-$700 \kms\, and
1455: +500 \kms\, are excluded from the profile modeling.
1456: }
1457: \end{figure}
1458:
1459: \end{document}
1460:
1461:
1462: