The Detection of a Light Echo from the Type Ia Supernova 2006X in M100 Xiaofeng Wang1,2, Weidong Li1, Alexei V. Filippenko1, Ryan J. Foley1, Nathan Smith1, and Lifan Wang3 ABSTRACT 8 0 0 We report the discovery of a light echo (LE) from the Type Ia supernova (SN) 2006X in the 2 nearby galaxy M100. The presence of the LE is supported by analysis of both the Hubble Space n Telescope (HST) Advanced Camera for Surveys (ACS) images and the Keck optical spectrum a thatwe obtainedat∼300d after maximumbrightness. Inthe image procedure,both the radial- J profileanalysisandthepoint-spreadfunction(PSF)subtractionmethodresolvesignificantexcess 3 emissionat2–5ACS pixels (∼0.05′′−0.13′′) fromthe center. In particular,the PSF-subtracted ] ACS images distinctly appear to have an extended, ring-like echo. Due to limitations of the h p imageresolution, we cannotconfirmany structure or flux within 2 ACSpixels fromthe SN. The - late-timespectrumofSN2006Xcanbe reasonablyfitwithtwocomponents: anebularspectrum o r ofanormalSN Iaanda synthetic LEspectrum. Bothimage andspectralanalysisshowa rather t s blue color for the emission of the LE, suggestive of a small average grain size for the scattering a dust. Using the Cepheid distance to M100 of15.2 Mpc, we find that the dust illuminated by the [ resolved LE is ∼27–170 pc from the SN. The echo inferred from the nebular spectrum appears 3 to be more luminous than that resolvedin the images (atthe ∼2σ level), perhaps suggestingthe v 0 presence of an inner echo at <2 ACS pixels (∼ 0.05′′). It is not clear, however, whether this 7 possiblelocalechowasproducedbya distinctdustcomponent(i.e., the localcircumstellardust) 5 or by a continuous, larger distribution of dust as with the outer component. Nevertheless, our 2 . detectionofasignificantechoinSN2006Xconfirmsthatthissupernovawasproducedinadusty 1 environment having small dust particles. 1 7 0 Subject headings: circumstellar matter – dust, extinction – supernovae: general – supernovae: : v individual (SN 2006X) i X r a 1. Introduction Lightechoes(LEs)areproducedwhenlightemittedbytheexplosiveoutburstofsomeobjectsisscattered towardthe observerbythe foregroundorsurroundingdust, withdelayedarrivaltime due to the longerlight path. This phenomenon is rare, having been observed only around a few variable stars in the Galaxy, and around several extragalactic supernovae (SNe). The best-studied events are SN 1987A (Schaefer 1987; Gouiffes et al. 1988; Chevalier & Emmering 1988; Crotts 1988; Crotts, Kunkel, & McCarthy 1989;Bond et al. 1990; Xu et al. 1995) and the peculiar star V838 Mon (Bond et al. 2003). Other SNe with LEs include 1DepartmentofAstronomy,UniversityofCalifornia,Berkeley,CA94720-3411, USA;[email protected]. 2Physics Department and Tsinghua Center for Astrophysics (THCA), Tsinghua University, Beijing, 100084, China; wang [email protected]. 3PhysicsDepartment,Texas A&MUniversity,CollegeStation,TX77843. – 2 – the Type II SNe 1993J (Liu et al. 2002; Sugerman 2003), 2002hh (Meikle et al. 2006; Welch et al. 2007), and 2003gd(Sugerman 2005;Van Dyk et al. 2006), as well as the Type Ia SNe 1991T(Schmidt et al. 1994; Sparks et al. 1999), 1998bu (Cappellaro et al. 2001; Garnavich et al. 2001), and possibly 1995E (Quinn et al. 2006). Besides their spectacular appearance, LEs offer a unique means to diagnose the composition, distribution, andparticle size of the scattering dust. In particular,LEsfromthe circumstellarenvironments might provide constraints on SN progenitors. The Type Ia SN 2006X was discovered on 2006 February 7.10 (UT dates are used throughout this paper) by S. Suzuki and M. Migliardi (IAUC 8667, CBET 393) in the nearby spiral galaxy NGC 4321 (M100). Extensive photometric and spectroscopic coverage is presented by Wang et al. (2007, hereafter W07). They suggestthat SN 2006Xis highly reddened [E(B−V)host =1.42±0.04 mag] by abnormaldust with ℜ = 1.48±0.06. Its early-epoch spectra are characterized by strong, high-velocity features of both V intermediate-mass and iron-group elements. In addition to the anomalous extinction and the very rapid expansion,SN2006XexhibitsacontinuumbluerthanthatofnormalSNeIa. Moreover,itslate-timedecline rate in the B band is slow, β = 0.92±0.05 mag (100 d)−1, significantly below the 1.4 mag (100 d)−1 rate observed in normal SNe Ia and comparable to the decay rate of 1.0 mag (100 d)−1 expected from 56Co → 56Fe decay. This may suggestadditional energy sources besides radioactivedecay, such as the interactionof the supernova ejecta with circumstellar material (CSM) and/or a LE. Attempts to detect the CSM in SNe Ia in different wavebands were unsuccessful before SN 2006X,and onlysomeupperlimitscouldbeplaced(seePatatetal. 2007a,andreferencestherein)exceptforthepeculiar SNe Ia/IIn 2002ic (Hamuy et al. 2003; Deng et al. 2004; Wang et al. 2004; Wood-Vasey et al. 2004) and 2005gj (Aldering et al. 2006; Prieto et al. 2007). Recent progress in this respect was made from high- resolution spectroscopy by Patat et al. (2007b, hereafter P07), who find time-variable Na I D absorption linesinspectraofSN2006X.ThishasbeeninterpretedasthedetectionofCSMwithinafew1016cm(∼0.01 pc) from the explosion site of the supernova. With the inferred velocity, density, and location of the CSM, P07 proposed that the companion star of the progenitor of SN 2006X is most likely to be a red giant (but see Hachisu et al. 2007,who present a main-sequence star model with mass stripping). Note, however,that SN 2006Xexhibited somewhatabnormalfeatures in spectra and photometry;it may not representa typical SN Ia. Multi-epoch, high-resolution spectral observations of SN 2007af, a normal SN Ia, do not reveal any significant signature of CSM absorption (Simon et al. 2007). In this paper we report the discovery of an optical LE around SN 2006X, with evidence from Hubble Space Telescope (HST) Advanced Camera for Surveys (ACS) images and Keck optical spectra. The paper is organizedas follows. In §2 we briefly describe the late-epoch data available for SN 2006X, while the data analysis and the interpretation are presented in §3. We discuss the properties of the light echo and the underlying dust in §4. Our conclusions are given in §5. 2. Observations 2.1. HST Archival Images Several epochs of HST data covering the site of SN 2006X are publicly available in the MAST archive. The pre-discovery images were taken on 1993 December 31 (Proposal ID 5195: PI, Sparks) by the Wide Field Planetary Camera 2 (WFPC2) in the F439W and F555W filters, with the same integration time of 1800 s. The images taken prior to the SN explosion allow us to examine the immediate environment of the progenitor, whereas the post-explosion observations enable us to searchfor a possible LE. The most recent, – 3 – post-discovery images of SN 2006X were obtained with the High Resolution Channel (HRC, with a mean spatial resolution of 0.026′′ pixel−1) of HST/ACS on 2006 May 21 (90 d after B maximum) and on 2006 December 25 (308 d after B maximum), respectively (GO–10991; PI, Arlin Crotts). At t = 90 d, the SN wasimagedinF435W(1480s),F555W(1080s), and F775W(1080s),while att=308d, the SN wasagain observed in the same three bandpasses, with the exposure times of 920 s, 520 s, and 520 s, respectively. The standard HST pipeline was employed to pre-process the images and remove cosmic-ray hits. In Figure 1 we show the pre- and post-explosion HST images of SN 2006X in the F555W filter. This pre- discovery image does not reveal any significant source brighter than 24.0 mag in F555W, excluding the possibility ofa significantstarcluster atthe locationofSN 2006X.Neither ofthe twopost-discoveryimages exhibits any resolvedLE arcs or rings aroundSN 2006X.The magnitudes ofSN 2006Xwere measuredfrom the HST ACS images with both the Dolphot method (Dolphin 2000) and the Sirianni procedure (Sirianni et al. 2005), and the mean photometry is given in Table 1. 2.2. Keck Optical Spectrum Observations of nebular-phase spectra provide an alternative way to explore the possibility of an LE around SNe, as the scattered, early-phase light will leave a noticeable imprint on the nebular spectra (e.g., Schmidt et al. 1994; Cappellaro et al. 2001) when the SN becomes dimmer. Two very late-time spectra of SN2006Xtakenatt≈277and307daysafterB maximumwerepublishedbyW07(seetheirFig. 19),which wereobtainedbytheKecktelescopesattheW.M.KeckObservatory: onewiththeLowResolutionImaging Spectrometer (LRIS; Oke et al. 1995) mounted on the 10 m Keck I telescope, and the other with the Deep Extragalactic Imaging Multi-Object Spectrograph (DEIMOS) mounted on the 10 m Keck II telescope. In the following analysis we focus on the LRIS spectrum taken at t ≈ 277 d because of its wider wavelength coverage. 3. Data Analysis 3.1. Late-Time Light Curves Figure2 showsthe absolute B,V, andI lightcurvesofSN2006XandSNIa 1996X(Salvoetal. 2001). The former were obtained using the Cepheid distance µ = 30.91±0.14 mag (Freedman et al. 2001), and corrected for extinction in the Milky Way (A (MW) = 0.08 mag; Schlegel et al. 1998) and in the host V galaxy (A (host) = 2.10 mag; W07). The distance modulus and host-galaxy extinction for SN 1996X are V derived by Wang et al. (2006) and Jha et al. (2007) using independent methods, and we adopted the mean values µ = 32.11±0.15 mag (H0 = 72 km s−1 Mpc−1 and AV(host) = 0.08 mag are assumed throughout this paper). The absolute magnitudes of these two SNe are similar near maximum, except in the I band where SN 2006X is ∼0.4 mag fainter than SN 1996X. Noticeable differences between the two SNe emerge in B one month after maximum light, when SN 2006X begins to decline slowly at a rate of 0.92±0.05 mag (100 d)−1. The discrepancy reaches about 0.9 mag in B at t=308 d, while it is ∼0.7 mag in V and ∼0.2 mag in I. This suggests that the emission of SN 2006Xis 130%± 60%higher in B, 90% ± 40%higher in V, and 20%± 20%higher in I with respect to SN 1996X. The large error bars primarily reflect the uncertainty in the distances. The apparentlyoverluminousbehaviorseeninSN 2006Xin the tailphaseis possiblylinked to the light – 4 – scattering of the surrounding dust, though the interaction of the SN ejecta with the CSM produced by the progenitor system and/or the excess trapping of photos and positrons (created in 56Co −→ 56Fe decays within the ejecta) cannot be ruled out. The resultant LE, if present, may not be directly resolved even in the HST/ACS images at a distance of ∼15 Mpc due to the limited angular resolution. To examine this conjecture,in§3.2wecomparethePSFsoftheSNandlocalstars,andin§3.3weapplytheimage-subtraction technique to analyze the SN images. 3.2. Radial Brightness Profile The radial brightness profiles of the images of the SN and local stars in the same field (see Fig. 1) are compared in Figure 3. These were obtained by extracting the flux using different apertures, ranging from 0.1to 10pixels witha resolutionof0.1pixel. The fluxesofthe four localstarslabeledinFigure 1arescaled so that the integrated flux within the 10-pixel aperture is the same. Based on the distribution of the radial profiles of these four stars, we derived a mean radial profile with the same integrated flux through Monte Carlosimulations. The radialprofile ofthe four starswasthus normalizedby the peak flux of the simulated radial profile. Onecanseethatthe starprofilesareuniformatlargeradiusbutshownoticeablescatterwithin2pixels from the center. For comparison,the centralflux of SN 2006Xis scaled to be 1.0, with the assumption that the central region of the SN image was not affected by any LE. At t=90 d the SN profile does not show a significant difference from that of the local stars. At t =308 d the SN profiles appear distinctly broader at radii of 2–4 pixels, especially in the F435W and F555W images. Note that the SN data are quite steep at around 1 pixel, probably due to noise. The insetplotofFigure3 showsthe residualofthe radialprofilebetweenSN2006Xandthe localstars. This was obtained by subtracting the simulated radial profile of the stars from that of the SN. Also plotted is the scatter of the simulated profile of the local stars. At t=308 d, the SN shows significant extra flux in the F435W, F555W, and F775W filters at radii of ∼2 to 5 pixels, suggesting the presence of a LE. Such a residual flux was not present at t=90 d. Onecanseesomestructure(peaksandvalleys)at<2pixelsintheinsetresidualplotofFigure3. These alternating negative/positive residuals clearly show that the substructure within the inner 2 pixels cannot be trusted, and could result from the misalignment of the peak surface brightness of the images. Of course, it is possible that part of the LE is so close to the SN, but the above analysis cannot definitively reveal it. Integratingtheoverallresidualemissionintherange2–10pixels,wefindthattheobservedLEbrightness is ∼22.8 mag in F435W, ∼22.0 mag in F555W, and ∼22.1 mag in F775W. Its contribution to the total flux of the SN + LE is ∼29% in F435W, ∼27%in F555W, and ∼11% in F775W.In view of potential additional LE emission at radii <2 pixels, these values are probably lower limits to the true brightness of the LE. AlthoughStar1andSN2006XshowsomediffractionspikesinFigure1,thespikeshavethesameshape and orientationfor all stars in the field. Thus, (a) they should affect the radialsurface brightness profiles of all stars in the same way, and not affect the excess light from an echo, and (b) they should be adequately removed by the image subtraction procedure (§3.3). ThedifferencebetweentheradialprofileofSN2006Xandotherstarscanalsobe demonstratedbytheir measured full width at half-maximum intensity (FWHM). Table 2 lists the FWHM of SN 2006X and the – 5 – average value of several local stars, obtained by running the IRAF1 “imexamine” task in three modes: r (radial profile Gaussian fit), j (line 1D Gaussian fit), and k (column 1D Gaussian fit). At t = 90 d, the PSF of SN 2006X is comparable to that of the average values of the local stars, while at t= 308 d, the SN exhibits asignificantlybroaderprofile. TheFWHMincreasesbyabout0.3pixelinther-profileandby∼1.0 pixel in the j-profile and k-profile with respect to the local stars. The reasonable interpretation is that the PSF is broadened by scattered radiation (that is, the LE). 3.3. Light Echo Images The radial-profile study suggests the presence of a LE in SN 2006X. In this section, we apply image subtraction to provide further evidence for the LE, and study its two-dimensional (2D) structure. We extract a small section (20×20 pixels) centered on SN 2006Xand Star 1 (the brightest star in the field),andaligntheirpeakpixelstohighprecision(0.01pixel). WethenscaleStar1sothatitspeakhasthe same counts as that of SN 2006X,and subtract it from the SN 2006Ximage. The underlying assumption is the same as in our radial-profile study: the central peak of SN 2006X is not affected by any LE. Figure 4 showsthe PSF-subtractedimages ofSN 2006X.The left panelshows the subtractedimagesat the originalHST/ACSresolution. To bring out more details, the middle panel showssubsampled imagesby usingacubicsplinefunctiontointerpolateonepixelinto8×8pixels. Therightpanelhasthreecircles(with radii of 2, 4, and 6 pixels, respectively) overplotted. The residualimages all show an extended, bright, ring- like feature around the supernova, consistent with the general expected appearance of a LE. These features emerge primarily at radii of 2–4 pixels (or 0.05′′–0.11′′) in the images, consistent with those derived above from the radial profiles. The central structure seen within a circle of radius 2 pixels (e.g., the asymmetric feature in F435W, the double features in F775W, and the arc in F555W) are not to be trusted; due to the limited spatial resolution, the images used for the image subtraction may not be perfectly aligned (in terms ofthe geometryand/orthe flux of the centralregions),andsome artifactscouldbe introducedatthe center of the subtracted images. Similarly, the apparent clumps within the echo ring are not reliable, generally being only a few pixels in size. The integrated flux, measured from the PSF-subtracted images at 2–10 pixels from the SN site, con- tributes to the total flux of SN + LE by ∼33% in F435W, ∼29% in F555W, and ∼9% in F775W. This is fully consistentwith the aboveestimate fromthe radial-profileanalysis,taking into accountthe uncertainty inthePSFsubtraction. ThebrightnessoftheLEcomponentisestimatedtobe∼22.7maginF435W,∼21.9 mag in F555W, and ∼22.3 mag in F775W. Aswiththeradial-profileanalysis,thePSF-subtractionmethodmightremovesomefractionoffluxfrom the LE itself; it had been assumedthat none of the flux in the central2-pixel radius is produced by the LE, butthismightbeincorrect. Thus,ourestimateoftheechofluxfromtheimageanalysismaybe onlyalower limit of the true LE emission. In view of the image analysis, we cannot verify or rule out that the LE may be distributed continuously from the SN site to an angular radius of ∼6 pixels (0.15′′). 1IRAF, the Image Reduction and Analysis Facility, is distributed by the National Optical Astronomy Observatory, which isoperated bytheAssociationofUniversitiesforResearchinAstronomy,Inc. (AURA)undercooperative agreementwiththe NationalScienceFoundation(NSF). – 6 – 3.4. Light Echo Spectrum A consistency check for the existence of a LE around a source can also be obtained by comparing the observed supernova spectrum and the synthetic spectrum using an echo model. The observed spectrum should be a combination of the intrinsic late-time SN spectrum and the early-time scattered SN spectrum. Inspection of the late-epoch Keck spectrum (see Fig. 19 of W07) clearly reveals that SN 2006X behaves unlikeanormalSNIa,showingaratherbluecontinuumatshortwavelengthsandabroadabsorptionfeature near 6100 ˚A (probably due to Si II λ6355). Toconstructthecompositespectrumcontainingtheechocomponent,weusethenebular-phasespectrum of SN 1996X to approximate that of SN 2006X. SN 1996X is a normal SN Ia in the elliptical galaxy NGC 5061 (Salvo et al. 2001), with ∆m15 = 1.30±0.05 mag, similar to that of SN 2006X (W07). Late-time optical spectra with wide wavelengthcoverageand high signal-to-noiseratio (S/N) are available on day 298 ∼ forSN1996X(Salvoetal. 2001;http://bruford.nhn.ou.edu/ suspect/)andonday277forSN2006X(W07). Comparing the spectrum of SN 2006X obtained at t = 277 d with that taken at t = 307 d, we found that the overall spectral slope changed little during this period. We thus could extrapolate the original nebular spectrat=308d,aphasewhenbothSNe haverelativelygoodmulticolorphotometry. Tocompletelymatch the spectrum of SN 2006X, the spectral flux of SN 1996X was multiplied by a factor of 3.0 caused by the difference in distances. Extinction corrections have also been applied to the nebular spectra of these two SNe (W07; Wang et al. 2006). We considered the cases of both SN 2006X and SN 1996X as the central pulse source when deriving the echo spectrum. The observed spectra of SN 2006X are available at eleven different epochs from about −1 d to 75 d after B maximum, while 14 spectra of SN 1996X are available from about −4 d to 87 d after B maximum (Salvo et al. 2001). The above spectra were properly dereddened2 and interpolated to achieveuniformphasecoverage. Regardlessoftheoriginalfluxcalibration,alloftheinputspectrahavebeen recalibratedaccordingto their light curvesat comparablephases (W07; Salvo et al. 2001)and correctedfor the effects of scattering using a similar function, S(λ) ∝ λ−α (e.g., Suntzeff et al. 1988; Cappellaro et al. 2001). These corrected spectra were then coadded and scaled, together with the nebular spectrum of SN 1996X, to match the nebular spectrum of SN 2006X. The best-fit α values obtained for the combinations of SN 2006X (near B maximum) + SN 1996X (nebular) andSN 1996X(near B maximum) + SN 1996X(nebular) are3.0±0.3 and3.3±0.5,respectively. One can see that the combination of SN 2006X + SN 1996X gives a somewhat better fit to the observed spectrum of SN 2006X. This is not surprising; the spectrum of SN 2006X differs from that of a normal SN Ia at early times, showing extremely broad and blueshifted absorption minima (W07). The large value of α may indicate a small grain size for the scattering dust. The composite nebular spectrum and the underlying echo spectrum are compared with the observed spectrum of SN 2006X in Figure 5 (upper and middle panels). Given the simple assumption of the scattering function, incomplete spectral coverage, and intrinsic spectral difference between SN 1996X and SN 2006X, the agreement between the observation and the model is satisfactory, with major features in the spectrum well matched. This provides independent, strong evidence for the LE scenario. However, the broad emission peak seen at λ≈4300–4500˚A cannot be reasonably fit by the echo model (see Fig. 5); this mismatch is probably produced by intrinsic features in the nebular spectrum of SN 2006X. 2HereweassumethatthedustsurroundingSN2006Xisaplane-parallelslaband/orshell,sothatboththeSNandtheLE wereaffectedbyroughlythesameamountofextinction. – 7 – 3.5. Light-Echo Luminosity and Color We canconstrainthe propertiesofthe LEandthe underlyingdustthroughthe luminosityandcolorsof the LE.TheLEluminosityofSN2006XhasbeenestimatedbyanalyzingtheHSTSNimages;itcanalsobe obtainedbyintegratingthe echospectrumshowninFigure5. Themagnitudesofthe echogivenbydifferent methods are listed in Table 3. For the image-based measurement, the error accounts only for the scatter of the stellarPSF.Onthe otherhand,forthe spectrum-basedmeasurement,the errorprimarilyconsistsofthe uncertainties in extinction correction (i.e., ∼0.2 mag for SN 2006X and ∼0.1 mag for SN 1996X in the B band) and distance modulus (i.e., ∼0.14 mag for SN 2006X and ∼0.15 mag for SN 1996X). We note that the echoinferredfromthe spectralfitting seemssomewhatbrighterthan thatrevealedby the image analysis: δmF435W = −0.6±0.3 mag. This difference is also demonstrated in the bottom panel of Figure 5, where the flux ratio of the inferred echo spectrum and the observed spectrum of SN 2006X is plottedasafunctionofwavelength. Overplottedaretheratiosyieldedforthephotometryofthe echoimage (circles) and the spectrophotometry of the echo spectrum (squares) in F435W and F555W, respectively. Such a discrepancy, at a confidence level of only ∼2σ, may suggest that there is some echo emission within a radius of 2 pixels (21% ± 12% of the total flux of SN + LE in F435W and 17% ± 11% in F555W) that was not resolvedby the image analysis. Despite this possibility, we must point out that the echo luminosity derivedfromtheechospectrummayhaveanerrorthatisactuallylargerthanourestimate,sincewedidnot consider possible uncertainties associated with the spectrum itself and the simple scattering model adopted in our analysis (see §3.4). Assuming thatallofthe observeddifferences betweenthe lightcurvesandspectra ofSN2006XandSN 1996X at t = 308 d are entirely due to the LE around SN 2006X, we can place an upper limit on the LE brightness as 21.9±0.3 mag in F435W and 21.3±0.3 mag in F555W. The magnitudes and the resulting color are not inconsistent with those presented in Table 3, especially in the case of the spectral fit which likely takes into account most of the echo emission. This leaves little room for other possible mechanisms for the extra emission, suggesting that the echo is the primary cause of the abnormal overluminosity of SN 2006X at t=308 d. We find, from analysis of both the HST images and the nebular spectrum (see Table 3), that the LE has an average color (F435W–F555W)echo = 0.8±0.3 mag (this roughly equals (B −V)echo = 0.8±0.3 mag), which is much bluer than the SN color at maximum brightness. The LE is clearly brighter in bluer passbands than at redder wavelengths (see Fig. 5). Comparing the colors of the echo and the underlying SNlighthelps us interpretthe dust, asthe colorshiftdepends onthe scatteringcoefficientandhence onthe dimensions of the dust grains (Sugerman 2003a). Integrating over the entire SN light curve (W07) from about −11 d to 116 d after B maximum yields (B −V)SN = 1.70 mag for the overall emission of SN 2006X. The observed change in color, ∆(B −V) = −0.9±0.3mag,ismuchlargerthanthe colorshiftderivedforGalacticdustbutiscomparabletothechange derived for Rayleighan dust3, ∆(B−V)max =−0.96 mag (Sugerman et al. 2003b). This is consistent with constraints from the direct spectral fit, which suggests that the dust has a scattering efficiency proportional toλ−3.0. WethusproposethatthedustsurroundingSN2006XisdifferentfromthatoftheGalaxyandmay have small-size grains, perhaps with diameter . 0.01 µm, reflecting the shorter wavelengths of light more effectively. Smaller dust particles are also consistent with the low value of ℜ ≈1.5 derived by W07. V 3The Rayleighan dust consists of only small particles with grain size < 0.01 µm, and hence has a scattering efficiency proportionaltoλ−4 (Sugerman2003a). – 8 – 3.6. Dust Distance Of interest is the distribution of the dust producing the echo; for example, it may be a plane-parallel dust slab or a spherical dust shell. Couderc (1939) was the first to correctly interpret the LE ring observed around Nova Persei 1901. Detailed descriptions of LE geometries can also be found in more recent papers (e.g., Sugerman 2003a; Tylenda et al. 2004; Patat 2005). In general, the analytical treatment shows that bothadustslabandadustshellcouldproduceanechothatisacircularringcontainingthesource. Assuming that the SN light is an instantaneous pulse, then the geometry of an LE is straightforward: the distance of the illuminated dust material lying on the paraboloid can be approximated as D2θ2∓(ct)2 R≈ , (1) 2ct where D is the distance from the SN to the observer, θ is the angular radius of the echo, c is the speed of light, and t is the time since the outburst. The equation with a minus sign corresponds to the single dust slab, while the plus sign represents the case for a dust shell. As suggested by the analysis of the radial profile and the PSF-subtracted image of the SN, there is a confirmed LE ring ∼0.08′′ away from the SN, with a possible width of ∼0.03′′. For this echo of SN 2006X, ct = 0.27 pc, which leads to R ≈ 27–120 pc, consistent with the scale of the ISM dust cloud. As the dust cloudinfrontofSN2006Xseemstobeveryextended,wedonotgivethethicknessofthedustalongtheline ofsight. Consideringthe possible echoemissionwithin 2.0pixel(∼0.05′′) inferredfromthe echoluminosity (see discussion in §4.1) and that extending up to 5 pixels (∼ 0.13′′; see Fig. 3), the actual distribution of the dust may be from <27 pc to ∼170 pc from the SN. In principle, one can also estimate the distance of the dust itself from the SN through a fit to the observed echo luminosity using the light-echo model (e.g., Cappellaro et al. 2001), as the actual echo flux is related to the light emitted by the SN, the physical nature of the dust, and the dust geometry. However, current analytical treatments for the LE model must assume some idealized configuration, which may not apply to the dust surrounding SN 2006X that is found to probably have smaller dust grains with ℜ ≈1.5 V andarelativelyextendeddistribution. Moreover,multiplescatteringprocessesratherthanasinglescattering shouldbe consideredinthe echomodel due to the largeopticaldepth measuredfromthe dust: τV ≈2.0 for d SN 2006X. Detailed modelling of the LE emission seen in SN 2006X is beyond the scope of this paper. 4. Discussion Analysis ofboththe late-time HSTimagesandthe late-time Keckopticalspectrumfavorsthe presence of a LE in SN 2006X, the fourth non-historical SN Ia with a detection of echo emission. Comparison of the SN 2006Xecho with the other three known events, SNe 1991T,1995E,and 1998bu,shows that the Type Ia echoes may have a wide range of dust distances from . 10 pc to ∼ 210 pc. The echo detected in SN 1991T is consistent with being a dust cloud of radius 50 pc (Sparks et al. 1999), while the echo speculated from SN 1995Eprobably correspondsto a dust sheet at a distance of 207±35 pc (Quinn et al. 2006). Garnavich et al. (2001) proposed from the HST WFPC2 imaging that SN 1998bu may have two echoes, caused by dust at 120±15 pc and < 10 pc away from the SN; the outer echo is consistent with an ISM dust sheet, while the inner component is likely from the CSM dust. On the other hand, the resolved echo image of SN 2006Xappearsquite extendedin the directionperpendicular to the line of sight. This yields a dustdistance spanning from ∼ 27 pc to ∼ 170 pc away from the site of the SN, indicating that the dust causing the LE – 9 – may not be a thin dust sheet but could be a cloud or shell distribution of the dust around the progenitor or a more complicated dust system. The echo from SN 2006X is found to be brighter than that of the other three Type Ia echo events. Assuming the echo magnitude listed in Table 3 and the SN peak magnitude derived in W07, one can find that the echo flux with respect to the extinction-corrected peak magnitude of SN 2006X is ∼9.6 mag in V. Quinn et al. (2006) proposed that all of the other three Type Ia echoes (SNe 1991T, 1995E, 1998bu) show a striking similarity in their echo brightness relative to the extinction-corrected peak SN brightness, ∆V ≈10.7 mag. According to the analytical expression of the dust scattering (e.g., Patat 2005), the excess echo brightness from SN 2006X by ∼1 mag perhaps suggests a dust distribution closer to the SN, given the similar optical depth for SNe 2006X and 1995E. The SN 2006X echo emission also shows a prominent wavelengthdependence,withmorelightfromtheshorterwavelengths,suggestiveofsmaller-sizedustaround SN 2006X. This is also demonstrated by the difference of the scattering coefficient α required to fit the observed nebular spectrum, which is ∼3.0 for SN 2006X, ∼2.0 for 1991T (Schmidt et al. 1994), and ∼1.0 for SN 1998bu (Cappellaro et al. 2001). In fitting the nebular spectrum, the echo brightness is found to be ∼ 60% brighter than that from the echo image at the ∼2σ level, likely suggesting the presence of a local echo that was not resolved at the regions close to the SN site. Regarding the location of the echo emission in SN 2006X, one may naturally tie the distribution of the dust underlying the echo to a combination of local CSM dust and distant ISM dust, giventhe quite extendeddust distributionand the smalldust grains that were nottypicalfor the ISM dust. Detection of the CSM dust is of particular importance for understanding SN Ia progenitor models. P07 recently reported the detection of CSM in SN 2006X from variable Na I D lines, and they estimate that the absorbing dust is a few 1016 cm from the SN. It is hence expected that an echo very close to the SN (<0.01 pc away) should be produced, although the SN UV radiation field could destroy or change the distribution of the surrounding dust particles out to a radius of a few 1017 cm (Dwek 1983). However, it is notpossible forustodetectthe emissionofsuchacloseCSMechoatt=308d, sincethe maximumdelayed travel time of the light for this echo is <0.1 yr and the SN radiation decreases with time. As noted by W07, the spectrum of SN 2006X probably showed a UV excess at t ≈ 30 d. This may be a signature of the nearby CSM claimed by P07, but the S/N of the spectrum is quite low below 4000 ˚A. In this case, the possible echo emission at <27pc inferredfrom the nebular spectrum at t=308d couldresult from a dust shell that is farther out than that claimed by P07. This is possible if the CSM dust around SN 2006X has multiple shells, such as the dust ring (or shell) of a planetary nebula (Wang 2005) and nova-like shells. The presence of a local echo helps explain the slow decline of the B-band light curve of SN 2006X at early phases. Nevertheless, the local echo (if present) is not necessarily from the CSM dust, as forward scattering from the distant dust cloud in front of the SN could also produce an echo of very small angular size. To further distinguish between the two possible cases of distant ISM plus local dust and single ISM dust, future HST observations of SN 2006X are necessary. More late-phase HST/ACS images would help constraintheevolutionoftheLE.Usingequation(1),wecanpredicttheevolutionoftheechoringwithtime. If the dust formed as a result of past mass loss from the central source, the echo will be more symmetric and the expansion will slow down after the initially rapid phase; with time, its size will eventually shrink to zero. On the other hand, if the dust is of interstellar origin, the echo should expand continuously with slowlydecreasingbrightnessas more-distantregionsareilluminated. Assuming thatthe inner componentof the echowithin 2pixels is causedby aCSM dustshell∼1 pc fromthe SN, then the emissionwithin 2pixels will finally decrease to zero at t ≈ 6.5 yr. In contrast, the local echo from distant ISM dust should remain – 10 – nearly constant for a longer time. It is worth pointing out that the recent nearby SNe Ia, SN 2007gi (CBET 1017, CBET 1021), SN 2007le (CBET 1100, CBET 1101), and probably SN 2007sr (CBET 1172,1174, ATEL 1343), may exhibit high-velocityfeaturesintheirspectrasimilartothoseofSN2006X.IftheSN2006X-likeeventspreferentially occur in environments with abundant ISM dust or CSM dust (Wang et al. 2008, in prep.), then we might expect to detect late-time echo emission in the above three SNe Ia. Thus, it would be interesting to obtain future high-resolution HST/ACS images of these SNe. 5. Conclusions The emergence of a LE in SN 2006X has been confirmed with PSF-subtracted HST ACS images which show a ring-like, but rather extended, echo 2–5 pixels (0.05′′–0.13′′) from the SN site at t = 308 d past maximum brightness. A Keck nebular spectrum of the SN taken at a similar phase provides additional evidence for the LE scenario; it can be decomposed into a nebular spectrum of a normal SN Ia and a reflection spectrum consisting of the SN light emitted at early phases. From the resolved echo image, we derive that the intervening dust is ∼27–170 pc from the supernova. Based on the quite blue color of the echo, we suggest that the mean grain size of the scattering dust is substantially smaller than Galactic dust. Smaller dust particles are also consistent with the low ℜ value V obtained fromthe SN photometry. Our detection of a LE in SN 2006Xconfirms that this SN Ia occurredin a dusty environment with atypical dust properties, as suggested by the photometry (W07). Analysis of the nebular spectrum might also suggest a local echo at <27 pc (or at <2 pixels) that is not resolved in the PSF-subtracted image. This possible local echo is likely associated with the CSM dust produced by the progenitors,though detailed modeling of the echo spectrum and/or further high-resolution imaging are required to test for the other possibilities, such as very forwardscattering by a distant cloud or CSM-ejecta interaction. Some of the data presented herein were obtained at the W. M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California, and the National Aeronautics and Space Administration (NASA). The Observatory was made possible by the generous financial support of the W. M. Keck Foundation. This research was supported by NASA/HST grants AR–10952 and AR–11248 from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5–26555. We also receivedfinancialassistancefromNSFgrantAST–0607485,theTABASGOFoundation,theNationalNatural Science Foundation of China (grant 10673007), and the Basic Research Funding at Tsinghua University (JCqn2005036). REFERENCES Aldering, G., et al. 2006, ApJ, 650, 510 Bloom, J. S., et al. 2007,ATEL, 1343 Bond, H. E., Gilmozzi, R., Meakes, M. G., & Panagia,N. 1990,ApJ, 354, L49