Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~bradw/cv/papers/MBM12.pdf Äàòà èçìåíåíèÿ: Thu Apr 7 18:39:55 2005 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 06:55:30 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï

The Astrophysical Journal, 623:225 ­ 234, 2005 April 10
# 2005. The American Astronomical Society. All rights reserved. Printed in U.S.A.

A

CHANDRA OBSERVATIONS OF MBM 12 AND MODELS OF THE LOCAL BUBBLE
R. K. Smith
NASA Goddard Space Flight Center, Code 662, Greenbelt, MD 20771; and Department of Physics and Astronomy, The Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218; rsmith@milkyway.gsfc.nasa.gov

R. J. Edgar, P. P. Plucinsky, and B. J. Wargelin
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138

P. E. Freeman
Department of Statistics, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213

and B. A. Biller
Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721 Received 2004 September 24; accepted 2005 January 3

ABSTRACT Chandra observations toward the nearby molecular cloud MBM 12 show unexpectedly strong and nearly equal foreground O viii and O vii emission. As the observed portion of MBM 12 is optically thick at these energies, the emission lines must be formed nearby, coming from either the Local Bubble ( LB) or charge exchange with ions from the Sun. Equilibrium models for the LB predict stronger O vii than O viii, so these results suggest that the LB is far from equilibrium or that a substantial portion of O viii is from another source, such as charge exchange within the solar system. Despite the likely contamination, we can combine our results with other EUV and X-ray observations to reject LB models that posit a cool recombining plasma as the source of LB X-rays. Subject headingg ISM: bubbles -- plasmas -- supernova remnants -- X-rays: ISM s: Online material: color figure

1. INTRODUCTION After decades of observation, the nature of the diffuse soft ($1 keV ) X-ray background is still mysterious. Early observa4 tions of soft X-ray emission toward many sight lines ( Bowyer et al. 1968; Bunner et al. 1969; Davidsen et al. 1972) showed that there is substantial diffuse emission (see also reviews by Tanaka & Bleeker 1977; McCammon & Sanders 1990). Broadband spectral data from proportional counter observations fit a threecomponent model: an unabsorbed 106 K thermal component, an absorbed 2 ; 106 K thermal component, and an absorbed power law. The latter two components contribute mostly at higher energies, while most of the 1 keV band comes from the local (i.e., 4 unabsorbed) 106 K thermal component. Williamson et al. (1974) excluded on physical grounds all then-proposed sources for the 1 keV band emission other than a 4 hot ($106 K) ionized plasma. Current theories to explain the origin of this emission still require an ionized plasma, and include (1) a local young supernova explosion in a cavity (Cox & Anderson 1982; Edgar & Cox 1993), (2) a series of supernovae ( Innes & Hartquist 1984; Smith & Cox 2001), or (3) an overionized plasma slowly recombining after substantial adiabatic cooling ( Breitschwerdt & Schmutzler 1994; Frisch 1995; Breitschwerdt et al. 1996). One additional possibility was first suggested by Cox (1998), who pointed out that charge exchange from the solar wind might create at least part of the 1 keV band emission. 4 Independent of the observed 1 keV band emission, absorption4 line measurements to many nearby stars (e.g., Welsh et al. 1990, 1991) show that we are surrounded by a irregularly shaped ``cavity'' with very low density. The standard model for the Local Bubble ( LB) combines these two observations into the ``displacement'' model (Sanders et al. 1977; Snowden et al. 1990). In this picture, the diffuse X-rays come from an elongated 225

bubble of hot gas with average radius $100 pc, with the Sun near the center. If the LB is filled with hot ($106 K ) gas in collisional ionization equilibrium (CIE), then the models of the resulting emission fit the observed 1 keV band soft X-ray spectrum. How4 ever, this phenomenological model explains neither the origin of the hot gas nor the low-density region. Early LB models that attempted to explain both the hot gas and the low-density region (Cox & Anderson 1982; Edgar & Cox 1993) modeled it as an $105 yr old supernova remnant. However, this model predicts a large column density of O vi that is simply not observed along many sight lines (Shelton & Cox 1994; Oegerle et al. 2005). Since every oxygen atom passing through the blast wave needs to pass through this ionization state, the model predictions are quite robust and are simply not observed. In addition, the total thermal energy required by the phenomenological models is between 0:37 ; 1051 and 1:1 ; 1051 ergs-- nearly the entire kinetic energy of a supernova. Smith & Cox (2001) considered models with multiple supernovae that are allowed by the O vi data and energy considerations and that roughly fit the observed broadband emission. The recombining plasma model, described in detail in Breitschwerdt et al. (1996), has also been able to qualitatively fit existing observations. However, as discussed below, we are now able to reject the basic recombining plasma model by combining our results with other EUV and X-ray observations. 2. X-RAY EMISSION FROM THE LOCAL BUBBLE The difficulty in measuring the soft X-ray spectrum has limited further analysis. A 106 K plasma with solar abundances, in equilibrium or not, is line dominated in the range 0.1 ­ 1.0 keV, but the first observation able to even partially resolve these lines was done only recently with the Diffuse X-Ray Spectrometer


226

SMITH ET AL.

Vol. 623

( DXS; Sanders et al. 2001). Most of the lines are from L-shell ions of elements from the third row of the periodic chart : Si, S, Mg (and their neighbor Ne), and M-shell ions of Fe around 72 eV. There should also be some X-ray emission from the K-shell lines of carbon, nitrogen, and oxygen. Fitting the spectrum requires a spectral emission code for a collisional plasma, such as described in Smith et al. (2001) or Kaastra & Jansen (1993). These codes model the thousands of lines emitted by the cosmically abundant elements because of collisional excitation. However, the lines in the observed spectrum are numerous, and the atomic data for the line emission are incomplete and sometimes inaccurate. In any event, no calculated spectrum matches the high-resolution observations of the soft X-ray spectrum, such as those from DXS, and it is not clear whether the problem lies in the physical models or in the atomic data (Sanders et al. 2001). To avoid these problems, unambiguous and strong emission lines are needed. In a 106 K plasma, the most abundant line-emitting element is oxygen. In equilibrium at 106 K the dominant stage is helium-like O+6, with trace amounts of O+7 and O+5. However, the hot gas in the LB need not be in equilibrium. If (somehow) a recent (P105 yr) supernova created the LB, then the gas would still be ionizing. Conversely, if the LB is old (k106 yr), then it should be recombining. In either case, the ratios of the O+6 and O+7 ion populations will not be at their equilibrium values. The strongest O vii and O viii emission lines are in the soft X-ray range, between 0.5 and 0.8 keV. O vii has a triplet of lines from n ¼ 2 ! 1, the so-called forbidden ( F ) line at 0.561 keV, the resonance ( R) line at 0.574 keV, and the intercombination ( I ) line (actually two lines) at 0.569 keV, while O viii has a Ly transition at 0.654 keV. These lines are regularly used as plasma diagnostics in other situations: Acton & Brown (1978) discussed the nonequilibrium ionization effects on O vii lines in solar flares. Vedder et al. (1986), using data from the Einstein Focal Plane Crystal Spectrometer ( FPCS), measured the ionization state in the Cygnus Loop using these lines. Gabriel et al. (1991) discussed the general use of O vii and O viii emission lines in hot plasmas, specifically applied to Einstein FPCS observations of the Puppis remnant. The goal of our work is to apply these methods, already used to model supernova remnants, to the specific case of the LB. Earlier observations suggested that oxygen emission lines are strong in the LB. Inoue et al. (1979) detected the O vii F+ I+R emission lines with a gas scintillation proportional counter. Schnopper et al. (1982) and Rocchia et al. (1984), using solidstate Si( Li) detectors, detected O vii as well as lines from other ions. More recently, a sounding rocket flight of the X-Ray Quantum Calorimeter ( XQC) observed a 1 sr region of the sky at high Galactic latitude between 0.1 and 1 keV with an energy resolution of 5 ­ 12 eV, and detected both O vii and O viii. The O vii flux was 4:8 ô 0:8 photons cmþ2 sþ1 srþ1 ( hereafter line units, LU ), and the O viii flux was 1:6 ô 0:4 LU ( McCammon et al. 2002). XQC's large field of view means that the source of the photons cannot be determined, but this does represent a useful upper limit to the LB emission in this direction. Measuring the emission coming solely from the LB requires observations of clouds that shadow the external emission. Snowden et al. (1993, hereafter SMV93) used ROSAT observations of the cloud MBM 12 to measure the 3 keV emission in the 4 LB and found a 2 upper limit of 270 counts sþ1 srþ1 in the 1 3 ROSAT 4 keV band. This band includes both the O vii triplet
For consistency we present all surface brightnesses in units of steradians; note that 1 sr ¼ 1:18 ; 107 arcmin2.
1

and the O viii Ly line. They also fitted a ``standard'' 106 K CIE LB model assuming a path length of $65 pc and found a good match to the observed 1 keV emission with an emission measure 4 of 0.0024 cmþ6 pc. This would generate only $47 counts sþ1 srþ1 in the 3 keV band, well within the 2 limit. For compari4 son, this model predicts 0.28 LU of O vii line emission, which is the dominant contribution to the 3 keV band emission. 4 The ROSAT PSPC had insufficient spectral resolution to separate the O vii and O viii emission lines from each other, the background continuum, and possible Fe L line emission. We therefore used the Chandra ACIS instrument to redo the SMV93 observations with higher spectral and spatial resolution. Our results were unfortunately affected by a large solar flare during part of the observation , the charge transfer inefficiency degradation experienced by the ACIS detectors early in the Chandra mission, and the somewhat higher than expected background. Despite these issues, we were able to clearly detect strong O vii and O viii emission lines. 3. OBSERVATIONS AND DATA ANALYSIS MBM 12 is a nearby high-latitude molecular cloud (l ; b ¼ 159N2; þ34 ). The distance to MBM 12 is somewhat controversial. Hobbs et al. (1986) placed it at $65 pc based on absorptionline studies; this is revised to 60 ô 30 pc using the new Hipparcos distances for their stars. However, Luhman (2001), using infrared photometry and extinction techniques, found a substantially higher distance of 275 ô 65 pc. Andersson et al. (2002) found a similar distance (360 ô 30 pc) for most of the extincting material, but also found evidence for some material at $80 pc. As part of a much larger survey of Na i absorption toward nearby stars, Lallement et al. (2003) found foreground dense gas toward stars 90 ­ 150 pc distant in the direction of MBM 12. They also suggested that the distance discrepancy could be resolved if the molecular cloud MBM 12 is itself behind a nearby dense H i cloud. However, since we use it as an optically thick shadowing target, so long as MBM 12 is nearer than the nonlocal Galactic sources of 3 keV band X-rays that contribute to the diffuse back4 ground, the precise distance is unimportant. Indeed, the greater distance is interesting because it would constrain possible sources of diffuse 3 keV band X-rays. 4 Our Chandra observations of MBM 12 were separated into two nearly equal sections. The first observation ( hereafter Obi0) was performed on 2000 July 9 ­ 10. This observation was cut short by a severe solar flare that led to an automatic shutdown. Preshutdown, the flare apparently also created a systematically higher background in the ACIS during the entire observation, which unfortunately meant the entire observation had to be excluded (see x 3.1). The second observation ( hereafter Obi1), was performed on 2000 August 17 for $56 ks during a time of lower solar activity. The pointing direction for both observations was s 02h55m50.2, 19 300 14B0 (see Fig. 1). The primary CCD, ACISS3, was placed on the peak of the 100 m IRAS emission, which we assume corresponds to the densest part of the cloud. MBM 12 is a relatively thin molecular cloud, and so we must consider whether it is optically thick to background X-rays even at the relatively soft energies of O vii ($0.57 keV ) and O viii (0.654 keV ). Assuming solar abundances, the cross sections at these energies are 9:4 ; 10þ22 and 6:7 ; 10þ22 cmþ2 per H atom, respectively ( Balucinska-Church & McCammon 1992). Measuring the column density over the face of the entire cloud is difficult. Absorption-line measurements show the column density at one position; Luhman (2001) found that most stars in MBM 12 have AV < 2, with background stars in the range AV ¼ 3 8. Converting this to an equivalent hydrogen column density via


No. 1, 2005

OBSERVATIONS OF MBM 12 AND LOCAL BUBBLE MODELS

227

Fig. 1.--IRAS 100 m image of MBM 12, with the Chandra observation coordinates overlaid. [See the electronic edition of the Journal for a color version of this figure.]

NH /AV ¼ 1:9 ; 1021 cmþ2 magþ1 (Seward 2000) gives NH < 3:8 ; 1021 cmþ2 for much of the cloud, with a maximum value in the range NH ¼ (6 15) ; 1021 cmþ2. SMV93 also measured the column density by cross-correlating the ROSAT 3 keV and 4 IRAS 100 m flux, and derived a value of NH /F (100 m) ¼ 1:3×1::1 ; 1020 cmþ2 ( MJy srþ1)þ1. For these observations, the þ0 1 central ACIS-S3 detector was positioned on the brightest infrared position of MBM 12, where F (100 m) ¼ 30 35 MJy srþ1, so using the SMV93 ratio, NH ¼ (3:6 8:4) ; 1021 cmþ2. Since we are observing in the direction of the densest part of MBM 12, we believe that NH ¼ 4 ; 1021 cmþ2 is a conservative value. The opacity of MBM 12 to background line emission is then (O vii) ¼ 3:76 and (O viii) ¼ 2:68, and any distant emission will be reduced by 98% at O vii and 93% at O viii. Only an extremely bright background source could affect our results, and this is unlikely. The ROSAT PSPC had a large 2 field of view, and although SMV93 found some 3 keV emission off-cloud, it 4 was only $3.3 times brighter than their 2 upper limit for the 3 4 keV toward the cloud. 3.1. Data Processing The data reduction process was somewhat unusual, since emission from the LB completely fills the field of view, and the features of interest are extremely weak. Our original plan was to use the front-illuminated ( FI ) CCDs to measure the individual lines, and the back-illuminated ( BI ) CCDs (which have more effective area but lower resolution and higher background ) to

confirm the result. Based on the Gendreau et al. (1995) measurement of the soft X-ray background , we expected prelaunch to obtain at most $0.004 counts sþ1 from O vii and $0.001 counts sþ1 from O viii lines from both the four ( FI ) ACIS-I CCDs and the ( BI ) ACIS S3 CCD. However, after the proton damage to the FI CCDs early in the mission, the highly row dependent response of the FI CCDs means that these lines would be very difficult to extract robustly. Therefore, we focused on data from the BI CCD ACIS-S3. We used CIAO, version 3.1, tools to process the observations, along with CALDB, version 2.27. Our result depends crucially on the background measurement, which must be done indirectly since the source fills the field of view. The two most important steps are removing flares and point sources. We followed a procedure similar to that described in Markevitch et al. (2003), although as the data were taken in FAINT mode, VFAINT filtering was not possible. For the purposes of source-finding only, we merged the two observations using the ­>gn _ evt routine.2 We ali then ran celldetect, requiring a signal-to-noise ratio (S / N) > 3, ­> which found 16 sources (including the well-known source XY Ari). We excluded all these sources, using circles of 1500 radius (except XY Ari, where we used a 3000 radius circle). This procedure eliminated 15% of all events, with XY Ari alone accounting for 11%, and removed 5.0% of the area, leaving a total field of view of 67 arcmin2. We then made a light curve of the ACIS-S3 events between 2.5 and 7 keV for each observation, as shown in Figure 2. We constructed a histogram of the observed rates and fitted it with a Gaussian, representing the quiescent rate, plus a constant to roughly model the flares. We obtained good fits in both cases, which showed that the quiescent level in the first observation was 0.22 counts sþ1, while in the second it was 0.16 counts sþ1. The quiescent rate in the first observation is significantly above those shown in Markevitch et al. (2003), which ranged from 0.1 to 0.16 counts sþ1. These fluctuations in the background counting rate are likely due to protons that are trapped in the Earth's radiation belts or directly from solar flares. Following Markevitch et al. (2003), we removed all times when the count rate was not within 20% of the average rate for the second observation. The maximum allowed rate is therefore 0.192 counts sþ1, which limits the first observation to 2.2 ks of ``good time,'' too little data to be useful. This exclusion was not capricious. Although the first MBM 12 observation was $40% brighter than the average background rate for ACIS-S3, we expended substantial effort attempting to model it. All our attempts required adding what amounted to arbitrary terms at the same
2

See http://asc.harvard.edu /cont-soft /software/align _ evt.1.6.html.

Fig. 2.--Left: Light curve from the first MBM 12 observation between 2.5 and 7 keV, with a dashed line showing our best-fit quiescent rate of 0.22 counts sþ1. Right: Same, from the second observation, with a quiescent rate of 0.16 counts sþ1.


228

SMITH ET AL.
TABLE 1 Total C ounts and Ra tios in Selected Bands ObsID 62850 Band PHA (2500-3000) ......... 0.5 ­ 2.0a ......................... 2.0 ­ 7.0a ......................... 5.0 ­ 10.0a .......................
a

Vol. 623

MBM 12 Obi1 Counts 21,735 4015 7492 14,877 Ratio 1.000 0.185 0.344 0.684

``Good'' Obi0 Counts 1407 314 513 978 Ratio 1.000 0.223 0.365 0.695

``All'' Obi0 Counts 75,925 46,387 63,693 96,311 Ratio 1.000 0.611 0.839 1.269

Counts 36,197 4762 8545 22,742

Ratio 1.000 0.132 0.236 0.628

In units of keV.

energies as our emission lines, so no conclusive results were possible. Therefore, we reluctantly excluded it from the remaining analysis. After filtering the second observation, we were left with 38.2 ks of usable data. To cross-check these results, we followed Markevitch et al. (2003) and compared the total counts in selected energy bands to the total counts seen with pulse-height amplitude ( PHA) values between 2500 and 3000. PHA values in this range correspond to energies above 10 keV, which are outside Chandra's range and thus effectively measure the ``particle background'' rate. Table 1 shows the results for our MBM 12 observations and one of our background data sets, ObsID 62850 (see x 3.2). In Table 1 we show the results from the filtered Obi1 data as well as the 2.2 ks of ``good'' data from Obi0 and values from all of Obi0. Ratios of the number of events in these bands to the number of events with PHA values between 2500 and 3000 are given as well. The unfiltered Obi0 data show very high ratios, suggesting contamination from some source besides the normal particle background. Conversely, the Obi1 observations of MBM 12 show slightly higher ratios compared to the ObsID 62850 ratios, consistent with a weak source such as the diffuse X-ray background. The ratios seen for ObsID 62850 are similar to the average values observed by Markevitch et al. (2003) for the EHM observations. 3.2. Background Since the LB fills the field of view in all normal Chandra observations, determining the true background is nontrivial. There are three types of Chandra observations that contain only instrumental or near-Earth background: the Dark Moon observation described in Markevitch et al. (2003), the event histogram mode ( EHM ) data taken by ACIS during HRC-I observations and also described in Markevitch et al. (2003), and the ``stowed ACIS'' observations first described in Wargelin et al. (2004). These latter observations (to date, ObsID 4286, 62846, 62848, and 62850) were done with ACIS clocking the CCDs in ``timed exposure'' mode and reporting event data in the VFAINT telemetry format, and with ACIS in a stowed position at which ACIS receives negligible flux from the telescope and the onboard calibration source. We used data from ObsIDs 62848, 62848, and 62850 as our background data sets. Together these data sets represent 144 ks of observations, although they were treated independently in our fits. ObsID 4286 is less than 10 ks and was not used. Between our observations and the last of these observations (in 2003 December) there was little change in the overall background rate.3 We fitted the MBM 12 data from Obi1 simultaneously with stowed ACIS data using both a foreground and background model and ­> Sherpa's CSTAT statistic. The fits were restricted to
See http://cxc.harvard.edu /ccw/proceedings/index.html /presentations/ markevitch.
3

the range 0.4 ­ 6 keV, since above 6 keV the background appears to rise because of the near-constant cosmic-ray background combined with the falling mirror response, while below 0.4 keV the background rises and falls in a highly variable fashion. Our foreground model consisted of two unabsorbed delta functions for the O vii and O viii lines, plus an absorbed power law and thermal component to represent the cosmic X-ray background and the distant hot Galactic component. We did not include a thermal component with T $ 106 K to represent the LB emission, as nearly all of that emission would be below 0.4 keV, except for the oxygen lines. The absorption was allowed to vary between NH ¼ 3:6 and 8:4 ; 1021 cm2 for both components; the best-fit value was 6 ; 1021 cm2, although our results were not particularly sensitive to this value. We used the values from Lumb et al. (2002) for the power-law component ( þ ¼ 1:42 ô 0:03, normalized to 8:44×2::55 photons cmþ2 sþ1 keVþ1 at 1 keV ), þ0 23 as well as for the temperature of the thermal component (0:2 ô 0:01 keV ). The normalization on the thermal component was allowed to vary, since significant variation has been seen for this absorbed hot gas ( Kuntz & Snowden 2000). The O viii Ly line position was set at 0.654 keV. However, since the O vii emission is from a triplet of lines whose dominant member is unknown, we allowed the centroid of the line complex to float within a range of O vii line positions between the forbidden line at 0.561 keV and the resonance line at 0.574 keV. We found that the sharp rise seen below 0.5 keV in both the source and background could be fitted using a low-energy Lorentzian, and we also included delta functions at 1.78 keV (Si-K ) and 2.15 keV (Au-M ) to fit the particle-induced fluorescence seen at these energies. Finally, the particle-induced continuum background was modeled as a line with slope and offset, which was not folded through the instrumental response. 4. RESULTS Since we used the standard model of the diffuse X-ray background, we were not surprised to find a good fit to the data. The source model had only three significant parameters: the O vii and O viii line fluxes, and the normalization on the absorbed Galactic thermal component. The total absorbed flux from this last component was only FX (0:4 6 keV ) ¼ 0:066 photons cmþ2 sþ1 srþ1, significantly less flux than from either of the two oxygen lines. Our results for the line emission are shown in Figure 3 and Table 2. Table 2 also lists O vii and O viii fluxes measured at high Galactic latitude with ASCA (Gendreau et al. 1995) and XQC ( McCammon et al. 2002). In addition, line fluxes due to heliospheric charge exchange (Snowden et al. 2004, hereafter SCK04) and geocoronal charge exchange ( Wargelin et al. 2004, labeled ``Dark Moon''; see x 5.1) are listed, along with the 2 upper limits from the ROSAT observations of MBM 12 (SMV93). The listed errors are 1 , except for the Wargelin et al. (2004) data, where the 90% likelihood range is shown. The SMV93 limits


No. 1, 2005

OBSERVATIONS OF MBM 12 AND LOCAL BUBBLE MODELS

229

Fig. 3.--Top left: Best-fit spectrum for MBM 12 observation. Bottom left: Best-fit background spectrum from ObsID 62850; note the fluorescence lines at 1.78 and 2.15 keV. Top right: Best-fit spectrum from MBM 12 between 0.4 and 1 keV, showing the strength of the oxygen lines. Bottom right: Same, showing the background fit.

assume all the emission is from that one line, since ROSAT could not resolve O vii from O viii. When we allowed the O vii line to float between the forbidden and resonance line energies, we found that any value would lead to acceptable fits. The statistics and CCD resolution could not distinguish between either the low- or high-energy end, so our results are shown for an assumed ``average'' line position of 0.57 keV. Our best-fit O vii flux is consistent with the high-latitude Gendreau et al. (1995) results. However, the O viii emission is significantly larger than the results of Gendreau et al. (1995) or McCammon et al. (2002). The strength of the O viii is not strongly dependent on the details of background subtraction, since it is far from the rapidly rising background found at low energies on ACIS-S3. We conclude that this measurement is correct, but contaminated with solar system emission from charge exchange. Other measurements in Table 2 show that charge exchange can easily swamp any LB emission. For example, the Wargelin et al. (2004) ``bright'' results from the dark Moon ( from their Table 7) are significantly larger than either our results or those of Gendreau et al. (1995), and the same is true of the SCK04 results. 5. DISCUSSION Our original goal was to determine the age of the LB by measuring the ratio of the O vii and O viii emission lines. In equilibrium, the O viii emission would be negligible. A number of nonequilibrium models, however, predict detectable O viii.Smith & Cox (2001) considered a range of models involving a series of supernova explosions. If the LB is still evolving 3 ­ 7 Myr after the last explosion, then their Figure 13b suggests that the O viii would still be detectable relative to the O vii emission.4 Alternatively, Breitschwerdt & Schmutzler (1994) and Breitschwerdt (2001) would also imply a detectable amount of O viii emission
4 Note that in Smith & Cox (2001) Figs. 11, 12, 19, and 20 are incorrect; they should be reduced by a factor of 4 because of an error by the author.

(see x 5.3). Solar flares, the variable response of the ACIS-I, and complicated background of the ACIS-S3 CCD have proved significant but not insurmountable obstacles. However, the strong O viii line is a much larger problem, since it is significantly larger than has been measured at high latitudes that include both the LB and Galactic halo emission. As stated above, it seems likely that our results have been contaminated by charge exchange with the solar wind. 5.1. Charge Exchange Charge exchange, as first noted by Cox (1998), is a serious complication for diffuse soft X-ray background studies. Charge exchange creates X-rays when electrons jump from neutral material (usually hydrogen or helium atoms) to excited levels of highly ionized atoms. Charge exchange between neutral H and O+7, for example, can create O vii emission. The highly charged ions come from the solar wind and coronal mass ejections (CMEs). The neutral material can come from either the geocorona or the interstellar medium as it flows into the heliosphere; see Cravens (2000) for an overview of this mechanism. Cravens et al. (2001) show that the so-called long-term enhancements ( LTEs) observed during the ROSAT sky survey, which were apparent brightenings of the diffuse X-ray sky lasting days to weeks, can be explained by X-ray emission from charge exchange between the solar wind and either the geocorona or the interstellar medium, or both. Wargelin et al. (2004) detected a strongly time-dependent O vii line along with weaker O viii emission from Chandra observations of the dark Moon. The source of these X-rays is almost certainly charge exchange between solar wind ions and the geocorona. The time variability of the solar wind makes the time dependence of the X-ray emission understandable. Detailed spectral models of charge exchange emission have been developed (e.g., Kharchenko & Dalgarno 2001). Wargelin et al. (2004) used these calculations to model the O vii and O viii geocoronal emission as a function of the solar wind oxygen ion flux (which can be

TABLE 2 Oxygen L ine E mission This Work (LU) 1.79 ô 0.55 2.34 ô 0.36 ASCA (LU) 2.3 ô 0.3 0.6 ô 0.15 XQC (LU) 4.8 ô 0.8 1.6 ô 0.4 SCK04 (LU) 7.39 ô 0.79 6.54 ô 0.34 Dark Moon (LU) 6.5 ­ 13.6 2.7 ­ 6.1 SMV93 (LU) <7.1 <3.6

Ion O vii .................. O viii..................


230

SMITH ET AL.

Vol. 623

estimated from ACE measurements). Their model estimates the surface brightness in each line as vc np fO yil i nn0 10RE 2 5RE ; Ï 1÷ LS ¼ 4 rmin where vc is the solar wind velocity, np is the proton density, fO is the relative abundance of oxygen, yil is the line yield per charge exchange, i is the charge exchange cross section, and nn0 is the neutral particle density at 10 Earth radii (RE ); rmin is the geocentric distance to the edge of the magnetosphere or the spacecraft position, whichever is farther. Some of these parameters are measured by ACE.5 This model predicts that the O viii /O vii ratio due solely to charge exchange is 1:36(nO×8 /nO×7 ) × 0:14, where nI is the density of ion I. This result uses values from Tables 5 and 6 of Wargelin et al. (2004) and including the O vii K line in the O viii emissivity, since they would be blended. In a typical slow solar wind, nO×8 /nO×7 % 0:35, implying a ratio of 0.616 (SCK04). Unfortunately, ACE does not yet routinely provide measurements of nO×8 /nO×7 , as it does with nO×7 /nO×6 . During Obi1, nO×7 /nO×6 was $0.7, approximately midway between the average value for the slow solar wind ($0.3) and the value of $1.4 measured during the brightest dark Moon observations. Going beyond the Earth-Moon system, the time-variable effects of charge exchange in the broader heliosphere were measured by SCK04. They used a series of four XMM-Newton observation