HIGH-VELOCITY LINE FORMING REGIONS IN THE TYPE Ia SUPERNOVA 2009ig

G. H. Marion; Jozsef Vinko; J. Craig Wheeler; Ryan J. Foley; Eric Y. Hsiao; Peter J. Brown; Peter Challis; Alexei V. Filippenko; Peter Garnavich; Robert P. Kirshner; Wayne B. Landsman; Jerod T. Parrent; Tyler A. Pritchard; Peter W. A. Roming; Jeffrey M. Silverman; Xiaofeng Wang

doi:10.1088/0004-637X/777/1/40

1. INTRODUCTION

Research on Type Ia supernovae (SNe Ia) has been guided for many years by the general agreement that the progenitors are carbon–oxygen (C/O) white dwarf (WD) stars as first predicted by Hoyle & Fowler (1960). This hypothesis is based on observational and theoretical evidence that thermonuclear burning of a C/O WD can produce light curves and spectra that are very similar to those observed for SNe Ia. Beyond that general statement, we lack many of the details concerning the composition of the progenitors and the physics of the explosions. In order to increase the effectiveness of SNe Ia as distance indicators we must improve our ability to predict the intrinsic brightness of individual events. Observations that provide new information about the chemical structure deepen our understanding of the explosion physics and may contribute constraints to theoretical models.

Here we identify, measure, and compare high-velocity (HVF) (>20,000 km s⁻¹) and photospheric-velocity (PVF) absorption features from lines of Si ii, Si iii, S ii, Ca ii, and Fe ii in a sequence of optical spectra of SN Ia 2009ig. The spectra of SN 2009ig at phases earlier than +3 days were previously published by Foley et al. (2012). The phase of each observation is expressed in days with respect to the time of B-band maximum light, B-max (September 6.0 UT = JD 2,455,080.5; see Section 2.1).

High-velocity (HVF) absorption features from Ca ii lines are frequently identified in spectra of SNe Ia obtained a week or more before B-max, but reliable identifications of HVF from other elements are rare. The presence of HVF from other lines is inferred from unusual PVF line profiles in very early-time spectra. HVF components of blended Si ii λ6355 features are reported by Mattila et al. (2005), Quimby et al. (2006), Stanishev et al. (2007), Garavini et al. (2007), Tanaka et al. (2008), and Wang et al. (2009b). Evidence for HVF in Fe ii features is discussed by Branch et al. (2003), Branch et al. (2005), Branch et al. (2007), Hatano et al. (1999), and Mazzali et al. (2005a).

The two earliest spectra of SN 2009ig (−14.5 days, −13.5 days) clearly resolve HVF Si ii λ6355 as a separate absorption feature produced in a detached line forming region. All features identified in the earliest spectra of SN 2009ig are HVF. The first unambiguous detection of PVF occurs at −12 days, and we measure both HVF and PVF velocities from −12 days until HVF are no longer detectable at about −6 days. All absorption features in spectra obtained after −5 days are PVF. The early start and dense coverage of our sample allows us to map a complete transition from HVF to PVF in SN 2009ig.

The strongest absorption features in early-time spectra of SN Ia are the Ca ii H&K lines and the Ca ii near-infrared triplet (IR3). Most of the pre-maximum spectra in our SN 2009ig sample include both of these lines, presenting an opportunity to study the development of HVF and PVF in Ca ii. Wang et al. (2006) and Patat et al. (2009) also reported simultaneous observations of HVF and PVF from both Ca ii lines.

HVF from Ca ii are usually observed as very broad features that extend 15,000–25,000 km s⁻¹ beyond typical PVFs. They are most often detected in spectra obtained earlier than −5 days, but it is not unusual for Ca ii HVF to persist as late as maximum light. Detections of HVF Ca ii in early spectra of SNe Ia are reported and discussed by Fisher et al. (1997), Hatano et al. (1999), Wang et al. (2003), Gerardy et al. (2004), Thomas et al. (2004), Branch et al. (2008), Tanaka et al. (2008), Branch et al. (2009), Marion et al. (2009), and others. Childress et al. (2013a) compare and discuss Ca ii HVF in SNe Ia observed near B-max. Mazzali et al. (2005a) assert that all spectra of SNe Ia observed more than one week before maximum will exhibit HVF Ca ii.

Foley et al. (2012) use the same pre-maximum data of SN 2009ig as we do to identify and discuss differences in line profiles of Ca ii H&K, Ca ii IR3, and Si ii λ6355. They reveal the two-component nature of early Si ii λ6355 features and propose that HVF may be ubiquitous in SN Ia. Here we focus on the location and composition of HVF line forming regions. In particular, we examine the characteristics of HVF from lines that are not Ca ii. We directly compare the development of Si ii λ6355 features in the spectra of SN 2009ig to sequences of early-time spectra of five other SN Ia.

Velocity measurements show that the HVF and PVF line forming regions remain separated by about 8000 km s⁻¹ in SN 2009ig. The expansion of SNe Ia is assumed to be homologous, so that measured velocities are proportional to distances from the center of the SN. Consequently, a line forming region that produces HVF absorption features is physically outside of a region that produces lower velocity features. Homologous expansion ensures that they will remain in the same relative positions.

The formation of detached HVF requires a region with a localized enhancement of abundance or density. The HVF layer must also be farther from the center of the SN than the photospheric layer. We are a long way from a consensus for how HVF line forming regions come into existence. A few of the proposed explanations for HVF are uneven density profiles in the ejecta (Mazzali et al. 2005a), a clumpy structure or a thick torus (Tanaka et al. 2006), and interaction between the SN ejecta and surrounding material (Gerardy et al. 2004). Evidence for polarization of the HVF in spectra of SNe Ia was first presented by Wang et al. (2003) and discussed by Kasen et al. (2003). If the HVF regions are not spherically symmetric, then viewing angles will influence the observed velocities.

In Section 2 we describe the observations of SN 2009ig and the reduction of the data. A discussion of the methods used for identification and measurement of absorption features is presented in Section 3. The characteristics of HVF from specific lines are presented in Section 4. A discussion of PVF and their measurements appears in Section 5. Analysis of the composition and location of HVF line forming regions is found in Section 6. In Section 7 we compare the results for SN 2009ig to early-time spectra of other SNe Ia. Possible clues to the origins of HVF regions in SNe Ia are discussed in Section 8. Section 9 presents a summary and conclusions.

2. OBSERVATIONS

SN 2009ig in NGC 1015 (redshift z = 0.008770; NED) was discovered (Kleiser et al. 2009) by the Lick Observatory Supernova Search (LOSS; Filippenko 2001) with the 0.76 m Katzman Automatic Imaging Telescope (KAIT) at a magnitude of 17.5 on 2009 August 20.5 (UT dates are used throughout this paper). The last known nondetection was also by LOSS/KAIT on August 16 to a limiting magnitude of 18.7. On August 21.1 the first spectrum was obtained with the Asiago 1.82 m telescope by Navasardyan et al. (2009). It revealed characteristics of an SN Ia very soon after the explosion with a high expansion velocity for Si ii λ6355 (24,500 km s⁻¹) and no detection of Si ii λ5971 or S ii λ5641. Because SN 2009ig was nearby and the detection was very early, we were presented with a unique opportunity to conduct extensive observations of a young SNe Ia.

2.1. Photometry

Photometric data were obtained in the UBVI bands with the 0.8 m Tsinghua-NAOC Telescope (TNT), located at Xinglong Station of the National Astronomical Observatory of China, 180 km from Beijing. Details of the TNT, detectors, and observing conditions are described by Wang et al. (2008). All photometric data were reduced using standard IRAF packages.

The light curves were fit to MLCS2k2 templates (Jha et al. 2007). The light curves and the best-fit templates are displayed in Figure 1. We adopted A_V(Gal) = 0.089 mag for the Milky Way extinction (Schlafly & Finkbeiner 2011) and a time of maximum B-band brightness of T(B_max) = 55080.50 MJD (2009 September 6.0). MLCS2k2 finds Δ = −0.24 ± 0.08 for the best-fit light-curve parameter, μ₀ = 32.82 ± 0.09 mag for the distance modulus (assuming H₀ = 73 km s⁻¹ Mpc⁻¹), and A_V(host) = 0.01 ± 0.01 mag for the host-galaxy extinction.

From the MLCS2k2 results we find M_B = −19.46 ± 0.12 and M_V = −19.42 ± 0.12 mag on 2009 Sepember 6.0. The Δ = −0.24 parameter corresponds to a decline-rate value of Δm₁₅(B) = 0.90 ± 0.07 mag. These values suggest that SN 2009ig is slightly more luminous and has a slower decline rate than a "normal" SN Ia, but the M_B and Δm₁₅(B) values are not exceptional.

These parameters are consistent with those of Foley et al. (2012), who used KAIT and Swift photometry of SN 2009ig to derive values of μ₀ = 32.96 ± 0.02 mag and A_V(host) = 0.01 ± 0.01 mag. Small differences in the photometric parameters have no effect on the spectroscopic results presented in this paper.

2.2. Spectroscopy

Our complete sample includes 32 optical spectra of SN 2009ig obtained from −14.5 days to +12.5 days with respect to the time of B-max. Spectra obtained before Sepember 9 (+3 days) were previously published by Foley et al. (2012). In most cases where the phase is discussed in the text, we round the value to whole days for simplicity. The Swift U-grism spectra are included in this list of "optical" spectra since we only display and analyze the portion of these spectra in the range 3200–4400 Å.

Figure 2 displays spectra obtained on 25 of the 28 nights between −14 days and +13 days. The maximum wavelength range of 3200–9000 Å reaches both the Ca ii H&K lines (λ3945) and the Ca ii IR3 (λ8579). Full wavelength coverage is achieved on every night from −14 days to −5 days.

For some of the dates, observations of SN 2009ig that cover slightly different wavelengths occurred at two different telescopes within a few hours of each other. We combine these data to create a single continuous spectrum by trimming the data from one source at a specific wavelength and continuing from that wavelength with data from the other source without overlap. The transition points are selected to optimize the signal-to-noise ratio for the combined spectrum. Spectral features in SNe Ia change rapidly at early times, but a daily cadence is sufficient to trace the development of individual features. The use of a single spectrum per day makes it easier to measure and display the data. A complete list of the observational details can be found in Table 1, including the wavelengths at which the spectra were joined.

Table 1. Spectroscopic Observations

Date	Phase^a	Telescope/Instrument	Range^b	λ Merge^c
Date	Phase^a	Telescope/Instrument	(Å)	(Å)
Aug 22.5	−14.5	Lick/Kast	3400–9000	6700
Aug 22.6	−14.4	Keck I/LRIS	3100–7860	6700
Aug 23.4	−13.6	HET/LRS	4210–9000	4225
Aug 23.6	−13.4	Swift/Ugrism	3200–4100	4070
Aug 24.5	−12.5	Lick/Kast	3500–9000	⋅⋅⋅
Aug 25.5	−11.5	Lick/Kast	3500–9000	4000
Aug 25.6	−11.4	Swift/Ugrism	3200–4400	4000
Aug 26.5	−10.5	MMT/Blue Channel	3190–8330	⋅⋅⋅
Aug 27.5	−9.5	Lick/Kast	3400–9000	6700
Aug 27.5	−9.5	MMT/Blue Channel	3180–8270	6700
Aug 28.5	−8.5	Lick/Kast	3400–9000	6700
Aug 28.5	−8.5	MMT/Blue Channel	3180–8320	6700
Aug 29.4	−7.6	HET/LRS	7170–9000	7210
Aug 29.5	−7.5	MMT/Blue Channel	3190–8270	7210
Aug 30.4	−6.6	HET/LRS	4710–9000	6850
Aug 30.4	−6.6	MMT/Blue channel	3180–8250	6850
Aug 31.4	−5.6	HET/LRS	4170–9000	⋅⋅⋅
Sep 1.6	−4.4	Swift/Ugrism	3200–4900	⋅⋅⋅
Sep 2.4	−3.6	HET/LRS	4160–9000	⋅⋅⋅
Sep 3.6	−2.4	Swift/Ugrism	3200–4400	4200
Sep 3.5	−2.5	HET/LRS	4190–9000	4200
Sep 4.4	−1.6	HET/LRS	4180–9000	⋅⋅⋅
Sep 5.4	−0.6	HET/LRS	4180–9000	⋅⋅⋅
Sep 6.7	+0.7	Swift/Ugrism	3200–4900	⋅⋅⋅
Sep 9.4	+3.4	HET/LRS	4180–9000	⋅⋅⋅
Sep 10.4	+4.4	HET/LRS	4180–9000	⋅⋅⋅
Sep 11.5	+5.5	FLWO/FAST	3300–7200	⋅⋅⋅
Sep 12.5	+6.5	FLWO/FAST	3300–7200	⋅⋅⋅
Sep 14.4	+8.4	HET/LRS	4180–9000	⋅⋅⋅
Sep 15.5	+9.5	FLWO/FAST	3300–7200	⋅⋅⋅
Sep 16.5	+10.5	FLWO/FAST	3300–7200	⋅⋅⋅
Sep 17.4	+11.5	FLWO/FAST	3300–7200	⋅⋅⋅
Sep 18.5	+12.5	FLWO/FAST	3300–7200	⋅⋅⋅

Notes. The spectra obtained before September 9 were previously published by Foley et al. (2012). ^aPhase in days with respect to the time of B-max (UT 2009 September 6.0 = JD 2,455,080.5). ^bWavelength range used for this paper. The complete spectra may cover a larger range. ^cWavelength at which this spectrum was trimmed and combined with a contemporaneous spectrum to form a single spectrum for this date.

Download table as: ASCII Typeset image

For the ground-based spectroscopy, the slit was generally oriented along the parallactic angle to minimize differential slit losses caused by atmospheric dispersion (Filippenko 1982). SN 2009ig was observed using the Shane 3 m telescope at Lick Observatory with the Kast spectrograph (Miller & Stone 1993) on August 22, 24, 25, 27, and 28. An additional spectrum was obtained on August 22 at the 10 m Keck I telescope with the Low Resolution Imaging Spectrometer (LRIS; Oke et al. 1995) equipped with an atmospheric dispersion corrector. The Lick/Kast spectra used a 600/4310 grism on the blue side and a 300/7500 grating on the red side with a 2'' wide slit. The Keck/LRIS spectrum was obtained with a 400/3400 grism on the blue side and a 600/7500 grating on the red side using a 1'' wide slit.

The 9.2 m Hobby-Eberly Telescope (HET; Ramsey et al. 1998) with the Marcario Low-Resolution Spectrograph (LRS; Hill et al. 1998) was used to observe SN 2009ig on August 23 (−13 days). Additional HET observations were made on August 29, 30, and 31, as well as on September 2, 3, 4, 5, 9, 10, and 14. The HET/LRS spectra have an effective wavelength range of 4400–9200 Å. Reduction of the HET/LRS data was also conducted with standard IRAF procedures.

Low-dispersion spectra of SN 2009ig were obtained on August 26, 27, 28, 29, and 30 using the 6.5 m MMT with the Blue Channel spectrograph (Schmidt et al. 1989), covering the wavelength range 3200–8200 Å. CCD processing and spectrum extraction for these observations were completed with IRAF and the data were extracted with the optimal algorithm of Horne (1986). After the wavelength calibration was derived from low-order polynomial fits to calibration-lamp spectra, additional small adjustments were applied by cross-correlating a template sky spectrum to the night-sky lines that were extracted along with the SN. IDL routines were used to flux calibrate the data and to remove telluric lines (Wade & Horne 1988; Matheson et al. 2000; Foley et al. 2003).

The Swift satellite observed SN 2009ig with the UV grism mode of the Ultraviolet/Optical Telescope (Roming et al. 2005) onboard the spacecraft (Gehrels et al. 2004). Swift data were obtained on August 23, 25, and 27, and on September 1, 3, and 6. We do not use all of the Swift data for our analysis, but the complete sequence of Swift spectra and photometric measurements was published by Foley et al. (2012). Swift data were extracted and reduced using the "uvotimgrism" package of FTOOLS. The effective wavelength coverage is 1700–4900 Å, but in this analysis we use only a small portion of the Swift spectra covering the wavelength region 3200–4400 Å that contains the Ca ii H&K feature.

3. LINE IDENTIFICATIONS AND VELOCITY MEASUREMENTS

Eight lines have been identified that produce both HVF and PVF absorptions in spectra of SN 2009ig. The features associated with these lines are displayed by phase and velocity space in four figures: Si iii λ4560 and Si ii λ6355 (Figure 3), S ii λλ 5453, 5641 (Figure 4), Ca ii H&K λ3945 and Ca ii IR3 λ8579 (Figure 5), and Fe ii λλ 5018, 5169 (Figure 6). The velocity measurements at each phase are indicated by a blue hashmark for HVF and a red hashmark for PVF. To facilitate comparison of the components between features, all of the figures also have a dotted line plotted at 21,500 km s⁻¹ and a dashed line at 13,000 km s⁻¹. These velocities were selected to be representative of the HVF at −10 days and the PVF near B-max. A detailed discussion of individual features is found in Section 4 (HVF) and Section 5 (PVF).

**Figure 3.** HVF and PVF for Si iii λ4560 (left) and Si ii λ6355 (right) in SN 2009ig are displayed in velocity space (blueshifted velocities are expressed with positive values). Rest wavelengths are indicated on the top abscissa. The spectra were obtained between −14 days and B-max, and the phase of each spectrum is marked near the red end. The line profiles for Si ii λ6355 in the earliest spectra clearly require separate HVF and PVF components. The phases from −12 days to −8 days reveal contributions from both HVF and PVF. Narrow Na i D features are visible in the Si ii λ6355 profiles. HVF Si iii λ4560 is blended with PVF Mg ii λ4481 beginning about −10 days. The velocities measured for each feature are indicated by a blue hashmark for HVF and a red hashmark for PVF. The dotted line at 21,500 km s⁻¹ and the dashed line at 13,000 km s⁻¹ are the same in all figures.
Download figure:
Standard image High-resolution image

**Figure 4.** HVF and PVF for S ii λ5453 (left) and S ii λ5641 (right) are marked on a sequence of spectra of SN 2009ig obtained between −14 days and +0 days. Rest wavelengths are indicated on the top abscissa. The phase of each spectrum is marked near the red end. HVF are strongest before −10 days and PVF are stronger after −8 days. The measured velocities are indicated by a blue mark for HVF and a red mark for PVF. The PVF in S ii λ5453 are compromised by blending after −5 days and we do not measure the velocities (see Section 5). The dotted line at 21,500 km s⁻¹ and the dashed line at 13,000 km s⁻¹ are the same in all figures.
Download figure:
Standard image High-resolution image

**Figure 5.** HVF and PVF for Ca ii H&K λ3945 (left) and Ca ii IR3 λ8579 (right) in SN 2009ig are displayed in velocity space (blueshifted velocities are expressed with positive values). Rest wavelengths are indicated on the top abscissa. The spectra were obtained between −14 days and +1 days and the phase of each spectrum is marked near the red end. HVF velocities for Ca ii are higher than for other lines in the early spectra (compare offsets from the dotted line), but they eventually decline to the mean HVF velocity for all lines (dotted line). Ca ii HVF are detected for a few days longer than HVF for other lines, and the PVF appear later. Comparing the features by phase suggests that the HVF component of Ca ii H&K λ3945 is exaggerated by a strengthening contribution from PVF Si ii λ3858 beginning about −10 days. The measured velocities are indicated by a blue mark for HVF and a red mark for PVF. The dotted line at 21,500 km s⁻¹ and the dashed line at 13,000 km s⁻¹ are the same in all figures.
Download figure:
Standard image High-resolution image

**Figure 6.** HVF and PVF for Fe ii λ5018 (left) and Fe ii λ5169 (right) are marked on a sequence of spectra of SN 2009ig obtained between −14 days and +0 days. Rest wavelengths are indicated on the top abscissa. The phase of each spectrum is marked at the red end. The dashed line in the velocity space of Fe ii λ5018 and the dotted line in velocity space for HVF Fe ii λ5169 are in the same physical location. The feature at that position is produced by HVF only in the first two spectra, by a blend of PVF and HVF from about 12 days to −6 days, and by PVF only after −6 days. Comparing offsets from the dotted lines shows that HVF of Fe ii are deeper, wider, and at higher velocities than HVF of Si ii (Figure 3) and S ii (Figure 4). The measured velocities are indicated by a blue mark for HVF and a red mark for PVF. The dotted line at 21,500 km s⁻¹ and the dashed line at 13,000 km s⁻¹ are the same in all figures.
Download figure:
Standard image High-resolution image

The HVF and PVF of Si ii λ6355 are easily identified and relatively free from blending. These characteristics make them useful templates that guide efforts to identify absorption features associated with other lines. We compare the velocities, the phases of initial and final detections for HVF and PVF, the relative strengths of the HVF and PVF components at a given phase, and the velocity separation between the HVF and PVF. The presence of simultaneous HVF and PVF from several lines increases the number of blended features in the early-time spectra of SN 2009ig. Velocities and line profiles of individual features are measured directly from the flux-calibrated spectra rather than a flattened continuum because the shapes and slopes of the local continua vary widely in different regions of each spectrum.

In the velocity figures, the flux is scaled to make it easier to see details of the absorption features. A representative flux ratio is maintained between the two panels in each figure in order to compare the relative strengths of features from similar lines, but it is not possible to maintain a consistent scale between the figures. The velocity range for the Ca ii and Fe ii plots 48,000 to −8000 km s⁻¹. For Si ii and S ii velocity plots the range is 36,000 to −8000 km s⁻¹. Throughout this paper we reference blueshifted velocities using positive values.

At many phases one of the components may lack a distinct minimum but still makes an obvious contribution to the overall line profile. For example, the right panel of Figure 3 shows that the PVF component from Si ii λ6355 at −12 days produces only a small indentation on the red side of the larger absorption feature dominated by HV. As time passes the strength of the HVF signal diminishes and the PVF component becomes stronger. At −9 days the PVF absorption is dominant and the HVF component forms a small but distinct indentation on the blue side of the PVF absorption feature. We use this spectrum to illustrate a method for measuring velocities from absorption components that form incomplete features.

Figure 7 shows the Si ii λ6355 feature from the August 27 spectrum (−9 days). The photospheric component is simply fit by a Gaussian (solid green line) having a minimum at 14,400 km s⁻¹. The data in the region of the indentation created by the HVF component are fit with a Gaussian profile having a minimum at 20,700 km s⁻¹ (solid blue line). The dotted red lines show alternative Gaussians with minima that have been shifted by ±1000 km s⁻¹ from the minimum of the blue line. Neither of the line profiles plotted in red fit the data as well as the blue line. The optimal position for the Gaussian is determined by changing the minimum in steps of 100 km s⁻¹ and inspecting the results by eye. It is nearly always easy to determine when the Gaussian profile deviates from the actual data by more than a few hundred km s⁻¹. Uncertainties due to this measurement technique are 100–1000 km s⁻¹ depending on the quality of the data and how well the minima are defined. The measurements in Tables 2 (HVF) and 3 (PVF) were obtained by this method. Figure 3 illustrates that for Si ii λ6355, uncertainties will be higher for HVF measured after −9 days and for PVF measured before −11 days.

Table 2. High-velocity Features in SN 2009ig (10³ km s⁻¹)^a

Date	Phase^b	Ca ii	Si iii	Fe ii	Fe ii	S ii	S ii	Si ii	Ca ii
Date	Phase^b	λ3945	λ4560	λ5018	λ5169	λ5453	λ5641	λ6355	λ8579
Aug 22	−14	32.2	25.7	26.6	26.4	⋅⋅⋅	23.9	22.8	31.4
Aug 23	−13	30.7	⋅⋅⋅	25.2	24.5	23.0	23.1	22.0	30.5
Aug 24	−12	29.2	22.8	24.0	23.1	22.1	22.4	21.1	29.2
Aug 25	−11	27.9	22.2	23.2	22.8	21.9	21.8	21.0	28.4
Aug 26	−10	26.5	22.1	22.7	22.4	21.7	21.0	20.8	27.5
Aug 27	−9	25.1	21.7	22.5	22.0	21.6	21.1	20.7	26.2
Aug 28	−8	24.0	21.7	22.2	21.9	21.5	20.8	20.8	25.0
Aug 29	−7	23.4	21.5	22.0	21.5	21.6	20.6	20.6	24.1
Aug 30	−6	22.6	21.3	⋅⋅⋅	21.2	21.4	20.8	20.8	23.2
Aug 31	−5	22.1	21.1	⋅⋅⋅	21.1	⋅⋅⋅	20.4	⋅⋅⋅	22.9
Sep 1	−4	21.3	21.0	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
Sep 2	−3	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	21.8
Sep 3	−2	21.1	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	21.4
Sep 4	−1	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	21.4
Sep 5	0	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	21.2

Notes. Measurement uncertainties are higher for HVF after −7 days (see discussion in text). ^aThroughout this paper we represent velocities for blueshifted absorption features with positive values. ^bRise time τ_r = 17.1 days (Foley et al. 2012), so the first spectrum was obtained 2.6 days after the explosion.

Download table as: ASCII Typeset image

The SYNOW code (Branch et al. 2003) is another tool for line identification. SYNOW creates model spectra that can compare relative line strengths and positions by wavelength. SYNOW makes it possible to explore the contributions of individual ions to a complete spectrum, and it is particularly helpful for distinguishing between the relative contributions of different lines in a blended feature.

4. HIGH-VELOCITY ABSORPTION FEATURES (>20,000 km s⁻¹)

In most SNe Ia, Ca ii produces the only detectable HVF. In SN 2009ig, eight lines from five ions produce distinct HVF in the early spectra: Si iii λ4560 and Si ii λ6355 (Figure 3), S ii λλ5453, 5641 (Figure 4), and Fe ii λλ5018, 5169 (Figure 6). There are also strong suggestions of HVF from several other lines that are associated with blended absorption features for which we are unable to confirm the contributions of individual lines.

HVF are clearly identified from all lines in the −14 days spectrum of SN 2009ig, and HVF remain strong through −12 days. The strength of the HVF absorptions weaken with time, and in most lines HVF are no longer detectable after −6 days. Beginning at about −10 days, many of the HVF are affected by blends with PVF that are increasing in strength.

The highest velocities for features from each line are measured in the earliest spectrum at −14 days. For Ca ii, v_max ≈ 32,600 km s⁻¹, followed by Fe ii at 26,600 km s⁻¹, Si iii at 25,700 km s⁻¹, S ii at 23,900 km s⁻¹, and Si ii at 22,800 km s⁻¹. This range of velocities for the HVF is apparently caused by differences in the effective opacities of the lines at that phase. By −12 days, the HVF velocities for all lines except Ca ii are grouped closely together with a mean of 22,600 km s⁻¹. HVF velocities for all lines converge near 21,500 km s⁻¹ by −6 days.

Table 3. Photospheric Features in SN 2009ig (10³ km s⁻¹)^a

Date	Phase^b	Ca ii	Si iii	Fe ii	Fe ii	S ii	S ii	Si ii	Ca ii
Date	Phase^b	λ3945	λ4560	λ5018	λ5169	λ5453	λ5641	λ6355	λ8579
Aug 22	−14	⋅⋅⋅	16.7	18.4	18.0	15.9	17.0	⋅⋅⋅	⋅⋅⋅
Aug 23	−13	⋅⋅⋅	15.2	16.3	17.6	15.5	16.2	16.5	⋅⋅⋅
Aug 24	−12	⋅⋅⋅	14.0	15.1	16.3	13.9	15.3	15.9	⋅⋅⋅
Aug 25	−11	⋅⋅⋅	13.6	14.5	15.9	13.7	15.2	15.6	⋅⋅⋅
Aug 26	−10	15.8	13.1	14.3	15.1	13.5	15.0	15.2	⋅⋅⋅
Aug 27	−9	15.4	12.9	13.8	14.8	13.2	14.6	14.5	16.0
Aug 28	−8	15.4	12.8	13.6	14.6	13.0	14.5	14.2	⋅⋅⋅
Aug 29	−7	15.2	12.4	13.3	14.4	12.9	14.2	14.1	15.0
Aug 30	−6	14.9	12.3	13.1	14.2	12.8	14.1	13.9	14.4
Aug 31	−5	15.1	11.9	12.9	14.0	12.4	13.7	13.6	13.9
Sep 1	−4	14.9	12.1	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
Sep 2	−3	⋅⋅⋅	⋅⋅⋅	12.0	13.5	⋅⋅⋅	13.2	13.4	13.8
Sep 3	−2	14.8	⋅⋅⋅	12.1	13.4	⋅⋅⋅	13.0	13.3	13.9
Sep 4	−1	⋅⋅⋅	⋅⋅⋅	12.3	12.7	⋅⋅⋅	12.8	13.4	14.2
Sep 5	0	⋅⋅⋅	⋅⋅⋅	12.4	12.5	⋅⋅⋅	12.9	13.4	14.6
Sep 6	1	14.6	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
Sep 7	2	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
Sep 8	3	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
Sep 9	4	⋅⋅⋅	⋅⋅⋅	12.4	11.9	⋅⋅⋅	12.8	13.3	14.2
Sep 10	5	⋅⋅⋅	⋅⋅⋅	12.6	11.8	⋅⋅⋅	13.4	13.3	14.0
Sep 11	6	14.3	⋅⋅⋅	12.7	11.9	⋅⋅⋅	13.7	13.2	⋅⋅⋅
Sep 12	7	14.2	⋅⋅⋅	12.8	12.0	⋅⋅⋅	13.7	13.1	⋅⋅⋅
Sep 13	8	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	13.2	⋅⋅⋅	⋅⋅⋅
Sep 14	9	⋅⋅⋅	⋅⋅⋅	12.7	11.9	⋅⋅⋅	13.2	13.2	14.0
Sep 15	10	13.9	⋅⋅⋅	12.8	12.0	⋅⋅⋅	13.5	13.0	⋅⋅⋅
Sep 16	11	13.8	⋅⋅⋅	12.8	11.9	⋅⋅⋅	13.0	13.0	⋅⋅⋅
Sep 17	12	13.5	⋅⋅⋅	12.6	11.8	⋅⋅⋅	⋅⋅⋅	12.9	⋅⋅⋅
Sep 18	13	13.5	⋅⋅⋅	12.8	11.7	⋅⋅⋅	⋅⋅⋅	13.0	⋅⋅⋅

Notes. Measurement uncertainties are higher for PVF before −12 days (see discussion in text). ^aThroughout this paper we represent velocities for blueshifted absorption features with positive values. ^bRise time τ_r = 17.1 days (Foley et al. 2012), so the first spectrum was obtained 2.6 days after the explosion.

Download table as: ASCII Typeset image

4.1. HVF from Si ii, Si iii, and S ii

The HVF of Si ii λ6355 provides a useful guide for the behavior of HVF in general because it is essentially unblended. The HVF component dominates the Si ii profile from −14 days to −12 days at velocities that are about 1000 km s⁻¹ lower than S ii and about 2000 km s⁻¹ lower than Fe ii. Both HVF and PVF contribute to the Si ii λ6355 line profile between −12 days and −7 days, and HVF are no longer detected from Si ii after −6 days. Narrow Na i D features are visible in the spectra at about 23,500 km s⁻¹, which is zero velocity for Na D, and at 24,200 km s⁻¹, which is the velocity of the host galaxy, NGC 1015.

Si ii has many other lines that often produce detectable features in spectra of SNe Ia. Possible evidence exists for HVF components from Si ii λλ3858, 4130, 5051, and 5972, but the features are compromised by blending.

The data do not include sufficiently short wavelengths to cover the HVF of Si iii λ4560 at −13 days, but this feature is strong at −14 days and −12 days. Contributions from PVF Mg ii λ4481 and PVF Fe iii λ4407 begin to affect HVF Si iii λ4560 by −11 days. We measure HVF Si iii until −4 days, when the absorption minimum shifts abruptly to the red, indicating that the HVF contribution has ended. SYNOW models suggest that the PVF Fe iii component is stronger than PVF Mg ii from −10 days until about −3 days.

Absorption features in the spectra can possibly be associated with several other doubly ionized lines. Possible HVF are found for Si iii λ5740, S iii λ4264, and Fe iii λλ4407, 5129, but as for many other tentative associations, blending prevents definitive identifications.

Figure 4 shows that HVF for both S ii λ5453 and S ii λ5641 are stronger at −12 days than they are at −14 days. HVF velocities for S ii are between the velocities of HVF Fe ii and HVF Si ii. The figure also shows that the S ii λ5641 HVF and the S ii λ5453 PVF form a blended feature. We identify the red component of this blend as HVF S ii λ5641. The HVF is strongest from −14 days to −12 days, contributions from both HVF and PVF are evident from −10 days to −5 days, and weak absorptions remain near the locations of both S ii HVF as late as −3 days. We stop measuring HVF S ii λ5453 after −6 days and HVF S ii λ5641 after −5 days. There is possible evidence in the spectra for HVF from S ii λ5032 and λ5429.

HVF are not detected from carbon, oxygen, or magnesium, so there is no evidence for unburned fuel or carbon burning products in the HVF region of SN 2009ig. If the Mg ii λ4481 line produced a HVF at −14 days, it would have a velocity of about 24,000 km s⁻¹, and it would appear at about 29,000 km s⁻¹ in the velocity space of Si iii λ4560 (left panel of Figure 3). At −10 days, the Mg ii HVF velocity would be about 21,500 km s⁻¹, corresponding to 26,500 km s⁻¹ in the figure. The spectra are steeply sloped toward the red in this region, which makes it more difficult to discern small absorptions, but careful inspection reveals no evidence of absorption features that can be associated with HVF from Mg ii.

4.2. HVF from Ca ii and Fe ii

The velocity plots and tables show that HVF of Ca ii and Fe ii are wider and deeper, with absorption minima at higher velocities than the HVF of other lines. Foley et al. (2012) observed that HVF of both Ca ii lines are stronger and have higher velocities than HVF of any other lines in the spectra of SN 2009ig.

The Fe ii λ5169 HVF produces a very strong absorption from −14 days to −11 days that is nearly as broad and deep as the Ca ii HVF. The feature is subsequently affected by PVF Fe ii λ5018, PVF Si ii λ5051, and PVF S ii λ5032 beginning about −10 days. We stop measuring HVF Fe ii λ5169 after the PVF blend becomes dominant at −5 days.

HVF Fe ii λ5018 is the strongest part of a blend with the HVF components of S ii λ5032 and Si ii λ5051 from −14 days to about −10 days. This identification is confirmed by the close agreement of the measured velocities of HVF Fe ii λ5169 and HVF Fe ii λ5018. This feature loses a distinct minimum after −7 days and we stop measuring HVF Fe ii λ5018 at that point. Possible evidence exists for HVF components from Fe ii λλ4924, 5129.

Figure 5 displays the Ca ii H&K blend λ3935 (left) and the Ca ii IR3 λ8579 (right). At −14 days, the absorption minima for both Ca ii blends are near 32,000 km s⁻¹ and the features have blue wings that extend beyond 40,000 km s⁻¹. Simultaneous observations of Ca ii H&K and IR3 are of particular interest as they provide a rare opportunity to compare the evolution of these features. As shown in Figure 5 and Table 2, the measured velocities and rates of change are very similar for HVF from both of the strong Ca ii blends. Although Ca ii HVF velocities are about 7000 km s⁻¹ higher than Fe ii HVF and nearly 10,000 km s⁻¹ higher than Si ii HVF at −14 days, they have a significantly faster decline rate, and the HVF velocities for all lines converge near 21,500 km s⁻¹ at −6 days.

HVF from the Ca ii IR3 are unblended, and they are weak but detectable until B-max. That is about 5 days after the last HVF detection from other lines. The HVF component of Ca ii H&K λ3945 is blended with PVF Si ii λ3858. Side by side comparison of the Ca ii H&K and IR3 line profiles suggests that the HVF from Ca ii H&K are broadened by Si ii PVF beginning about −10 days and the PVF influence gets stronger with time. We find that −2 days is the final phase at which the absorption minimum of the Ca/Si blend is consistent with the HVF of Ca ii IR3, so we stop measuring Ca ii H&K HVF after that phase.

Foley et al. (2013) demonstrated that there are conditions under which the absorption feature attributed to HVF Ca ii H&K can be produced by a combination of Si ii and a particular density profile for Ca ii. Wang et al. (2006) used polarization data in a −6 days spectrum from SN 2004dt to suggest that Si ii provides the dominant contribution to this blend at −6 days.

Childress et al. (2013a) compare the average strength of the Ca ii HVF component in spectra from SNe Ia at B-max. They find that slowly declining SNe Ia produce either high PVFs or strong HVF Ca ii close to B-max, but not both. SN 2009ig fits this pattern with v_Si = 13,400 km s⁻¹, while the Ca ii IR3 series in Figure 5 shows that HVF Ca ii is very weak at B-max.

5. PHOTOSPHERIC-VELOCITY ABSORPTION FEATURES

We use velocity offsets from PVF and their relative strengths by phase to confirm the identifications of HVF in spectra of SN 2009ig. PVF are identified for each of the same lines that produce HVF detections: Si iii λ4560 and Si ii λ6355 (Figure 3), S ii λλ5453, 5641 (Figure 4), and Fe ii λλ5018, 5169 (Figure 6). PVF velocity measurements are in Table 3.

There are weak suggestions of PVF for some lines in the −14 days spectrum of SN 2009ig, but the first unambiguous detections of PVF are at −12 days. Foley et al. (2012) note that −12 days is the phase at which the rate of change slows down for B−V colors. PVF become progressively stronger with time, and the PVF of most lines are stronger than the HVF by about −10 days. At −6 days the PVF components dominate all blended absorption features with little or no influence from the HVF.

PVF velocities in SN 2009ig decline steeply at first and then more gradually. They become essentially constant after −4 days. This behavior is consistent with the typical pattern for PVF velocities in SNe Ia (Foley et al. 2011; Silverman et al. 2012). The PVF velocities of SN 2009ig are about 2000 km s⁻¹ greater than velocities in a "normal" SN Ia, but they are not exceptional. We measure v_Si = 13,400 km s⁻¹, which places SN 2009ig in the HV group defined by Wang et al. (2009a). Foley et al. (2012) show that v_Si < 13,500 km s⁻¹ for ∼85% of SNe Ia having Δm₁₅(B) ⩽ 1.5 mag. The change in Si ii λ6355 velocity between +0 days and +10 days is ∼40 km s⁻¹ per day. That puts SN 2009ig in the Low Velocity Gradient group defined by Benetti et al. (2005).

5.1. Individual PVF

PVF Si ii λ3858 is blended with HVF Ca ii H&K. The Si ii contribution is significant as early as −10 days, but it is not clear when the Si ii PVF become strong enough to define the minimum in this feature. We do not take velocity measurements from this line. PVF Si ii λ4130 is the probable source of the absorption visible as a small notch near 4000 Å in the P Cygni emission peak from Ca ii H&K. HVF S iii λ4264 is a likely contributor to this feature. PVF of Si ii λλ5051, 5972 are possibly identified.

Most of the Fe ii PVF are strongly affected by blending with other PVF. PVF Fe ii λ5018 (left panel Figure 6) is blended with PVF Si ii λ5051 and PVF S ii λ5032. The Si ii and S ii lines may pull the observed minimum of this feature to the red since the PVF velocities for Fe ii λ5018 before B-max are 1000–1500 km s⁻¹ lower than PVF Fe ii λ5169.

The strong absorption feature near 4950 Å is formed by PVF Fe ii λ5169 with contributions from PVF Fe ii λ5266 and PVF Fe iii λ5129. The blend of PVF Fe ii λ5018, Si ii λ5051, and S ii λ5032 is observed near 4800 Å, and this feature continues to grow stronger after B-max. The P Cygni emission component of this feature pushes the minimum for PVF Fe ii λ5169 to velocities that are about 1000 km s⁻¹ lower than PVF Fe ii λ5018 and other PVF lines after B-max.

PVF S ii λ5453 is blended with HVF S ii λ5641 from −12 days to −9 days. We stop measuring PVF S ii λ5453 after −5 days when the velocity abruptly drops from 12,400 km s⁻¹ at −5 days to 10,400 km s⁻¹ at −3 days. That is not consistent with the other PVF lines at this phase including the other S ii line at λ5641. The displacement of the PVF λ5453 feature suggests that it is also moved to the red by P Cygni emission from the strong blend of Fe-group lines observed at 4950 Å.

PVF from lines of doubly ionized ions are weak at −14 days and −13 days, become significantly stronger at −12 days, and disappear at about −3 days. Figure 3 shows this pattern for PVF Si iii λ4560. Other possible but unconfirmed identifications of PVF from twice-ionized lines show similar behavior: Si iii λ5740, S iii λ4264, and Fe iii λλ4407, 5129. The HVF components of S ii lines also appear to be stronger at −12 days than at −14 days.

PVF Mg ii λ4481 is a major contributor to the very strong absorption found near 4220 Å in a blend with PVF Fe iii λ4407 and HVF Si iii λ4560. The left panel of Figure 3 shows the minimum of this feature near the dotted line at 21,500 km s⁻¹ until −3 days. A Mg ii PVF with a velocity of 14,000 km s⁻¹ would appear at about 19,000 km s⁻¹ on the velocity scale of Si iii λ4560. SYNOW models show a better fit to the data when the PVF Fe iii component is stronger than Mg ii from −10 days until about −3 days. The SYNOW result is consistent with the abrupt shift of this feature to the red at −3 days that aligns the minimum with characteristic PVF velocities for Mg ii λ4481.

The evidence for PVF from unburned C and O in SN 2009ig is inconclusive. C ii λ4743 makes a reasonable fit to absorption features for both HVF near 4370 Å and PVF near 4480 Å (Figure 3, wavelength axis on top). If, however, C ii λ4743 is the source of these features, then C ii λ6580 should also be easily detected. The PVF component of C ii λ6580 would appear near 3000 km s⁻¹ in the velocity space of Si ii λ6355. Foley et al. (2012) cite work with SYNOW models (Parrent et al. 2011) that finds C ii λ6580 to be present in the early-time spectra of SN 2009ig. C ii is purported to cause a flattening of the red wing in the line profile of Si ii λ6355 from −14 days to −12 days. It may be that including C ii improves the SYNOW fit to the line profile, but we do not find any absorption features at this location; the tiny notch at 6200 Å is telluric.

O i λ7773 develops a broad PVF line beginning about −3 days, but it is much weaker than PVF O i found in many SNe Ia. The features near 7400 Å in the spectra from −3 days to +9 days correspond to PVF of O i λ7773 at 11,000–12,000 km s⁻¹, consistent with other PVF velocities those phases.

6. THE LOCATION OF HVF LINE FORMING REGIONS

Figure 8 can be used to trace the location of the line forming regions for HVF of Ca ii H&K, Ca ii IR3, Fe ii λ5169, and Si ii λ6355 at phases −14 days, −12 days, −10 days, −6 days, and +0 days (top to bottom). Velocity serves as a proxy for radial distance, with the center of the SN on the right at 0 km s⁻¹ and the radius increasing with velocity from right to left.

The spectra are normalized to 1.0 at 6500 Å and the features are normalized to a flat continuum. The flux scale is the same in each panel. At −14 days, all features have the same flux calibration. At later phases, all other features retain the same flux calibration, but Ca ii H&K HVF is scaled down to fit in the figure.

At −14 days, both Ca ii HVF (purple and red) occupy the same velocity space, and a blue wing is present in both features that extends from 40,000 to ∼55,000 km s⁻¹. The minima for the Ca ii HVF are at velocities greater than 30,000 km s⁻¹ (Table 2 and Figure 9). The HVF velocity from Fe ii λ5169 is near 26,000 km s⁻¹. The line profile for Fe ii at this phase is similar to that of Ca ii H&K in both width and depth. A velocity difference of about 5000 km s⁻¹ is evident between the Ca ii line forming region (purple line) and Fe ii line forming region (blue line) on the red sides of the features. The blue sides, however, are closely aligned. This discrepancy is caused by the HVF of Fe ii λ5018 that creates a notch observed near 35,000 km s⁻¹ in the velocity space of Fe ii λ5169 (blue line). Between the red edge of this notch and the continuum, the side of the Fe ii λ5169 HVF is shifted by about 7000 km s⁻¹ to the blue.

**Figure 9.** Velocities of HVF and PVF are plotted by phase for eight lines found in the spectra of SN 2009ig between −14 days and +13 days. The measurements are listed in Tables 2 and 3. The lines and their identifying symbols are listed in the top-right corner. Separate HVF and PVF line forming regions are evident with a gap of about 8000 km s⁻¹ at all phases. No features are detected at velocities between the layers. Measurement uncertainties are higher for HVF after −7 days and for PVF before −12 days (see Figure 7 and discussion in the text). To reduce overplotting, data have been shifted slightly for Si iii λ4560 (+0.3 days), S ii λ5641 (−0.3 days; HVF only), and Fe ii λ5169 (−0.3 days).
Download figure:
Standard image High-resolution image

The Si ii HVF at −14 days has a minimum at about 23,000 km s⁻¹, and the line forming region is confined to a velocity space that is almost entirely in the lower velocity half of the Fe ii HVF. The Si ii line profile at −14 days (green) is clearly asymmetric, with a steeper blue side and an abrupt transition at the blue edge from the absorption to the continuum. The blue limit of the Si ii HVF has nearly the same velocity as the absorption minimum for Ca ii. In contrast, the PVF of Si ii (−6 days, +0 days) are nearly symmetric.

At −12 days, HVF are strong from all lines, with velocities for Ca ii about 29,000 km s⁻¹, for Fe ii near 23,000 km s⁻¹, and for Si ii near 21,000 km s⁻¹. The Si ii HVF is again formed in a smaller region than for other lines. The blue limit of the Si ii HVF is close to the same velocity as it was at −14 days. PVF from Si ii and Fe ii can be detected near 16,000 km s⁻¹ at this phase, with the Si ii PVF significantly distorting the line profile.

At −10 days, the HVF of Si ii is noticeably weaker, but it is clearly present as it forms the distortion in the blue side of the Si ii PVF. Using the methods described in Section 3, the velocity of HVF from Si ii is measured to be near 21,000 km s⁻¹. The HVF velocity of Fe ii is near 22,000 km s⁻¹ and the Ca ii HVF remains strong at about 27,000 km s⁻¹. At this phase, the HVF of Fe ii λ5169 has a significant contribution from PVF Fe ii λ5018 that narrows the observed profile. The Ca ii H&K HVF also receives a strong contribution from PVF Si ii λ3858.

At −6 days, the most obvious evidence for HVF is a very weak absorption from the Ca ii IR3 HVF. The blue line absorption near 21,000 km s⁻¹ is primarily due to the PVF of Fe ii λ5018. The Ca ii H&K HVF absorption (purple) near 23,000 km s⁻¹ is a blend with PVF Si ii λ3858. The magnitude of the contribution from Si ii can be estimated by comparing the HVF of Ca ii H&K to the HVF of Ca ii IR3. The broad Ca/Si profile does not provide a HVF detection. At this phase, Si ii has a PVF at 14,000 km s⁻¹, and the line profile is nearly symmetric in velocity space with no evidence of a HVF.

Figure 8 shows that the absorption minima for HVF are always observed above 20,000 km s⁻¹. There is no evidence at any phase for HVF with lower velocities. By +0 days, all absorptions are easily explained by associations with PVF at velocities near 13,000 km s⁻¹. The blue wing of the Ca ii IR3 PVF at this phase is slightly flattened by the remnant of the HVF.

Figure 9 plots HVF and PVF velocity measurements by phase for Si ii λ6355, Si iii λ4560, S ii λλ5453, 5641, Ca ii λλ3945, 8579, and Fe ii λλ5018, 5169. All lines contribute both HVF and PVF measurements to the figure. The plotted data are found in Tables 2 and 3. This figure is used to visually compare the relative velocities of both HVF and PVF from different lines at the same phases. Variations in the decline rates are easy to discern. HVF and PVF from the same lines or the same ions can also be compared.

HVF velocities converge to a common value near 21,500 km s⁻¹ and PVF velocities are grouped together at all phases. A significant gap in velocity space is evident between the HVF and PVF regions for all lines at all phases. The separation is about 8,000 km s⁻¹, except Ca ii which displays a greater separation before −5 days and Si ii λ6355 for which the separation at all phases is about 7000 km s⁻¹. No features are found at velocities that would place them between these layers.

Figure 9 shows that from −14 days to −10 days, the HVF velocities of both Fe ii lines are intermediate between the higher Ca ii HVF velocities and the lower HVF velocities for other lines. By −9 days, Fe ii HVF velocities are essentially the same as the velocities of other HVF that are not Ca ii.

The patterns of velocity by phase can be compared the behavior of the HVF line profiles by phase that are displayed in Figure 8. Si ii and Fe ii both have their maximum velocities in the earliest spectrum at −14 days. At that phase, the Fe ii HVF line profile is wide, deep and more like the HVF of Ca ii then the HVF of Si ii. The velocity difference between the HVF of Fe ii and Si ii is 3600 km s⁻¹. At −10 days, the HVF line profiles are less distinct due to blending with PVF. The Fe ii line profile is narrower and weaker and the velocities of Si ii and Fe ii are separated by 1600 km s⁻¹. The HVF of Ca ii at this phase remain broad and at higher velocities. At −6 days, the HVF of Si ii and Fe ii are very weak and their measured velocities are within 400 km s⁻¹.

The velocity dispersion of the HVF components of Si ii, Si iii, S ii, and Fe ii is confined within a narrow velocity region of less than 2000 km s⁻¹ from −12 days until HVF are no longer detected. By −9 days, the mean HVF velocity (excluding Ca ii) is about 21,500 km s⁻¹, and Ca ii velocities reach this value a few days later. After the HVF velocities reach ∼21,500 km s⁻¹, the velocities remain constant within the range of scatter introduced by measurement uncertainties. The mean PVF velocity is observed to be about 13,000 km s⁻¹ at −4 days, and it declines less than 1000 km s⁻¹ through the end of our observations at +13 days. (Most of the decline is caused by Fe ii λ5169; see Section 5.) The relative velocities of the HVF and PVF regions imply that, with homologous expansion, the radius of the HVF line forming region is ∼1.6 times greater than the radius of the PVF line forming region.

Figure 10 shows SYNOW model spectra plotted with data from SN 2009ig obtained at −11 days and −8 days. These phases are chosen because they include HVF and PVF, and the relative strengths of the components change in the time interval. The data are plotted in the figure with red lines. SYNOW spectra from models with only PVF velocities are plotted in blue with dotted lines, and model spectra with both HVF and PVF are solid black. At both phases, the addition of HVF noticeably improves the models in the regions of the Ca ii and Si ii features. More subtle improvements are evident around the Fe ii and Si iii features.

The SYNOW models using only PVF velocities produce a Ca ii IR3 feature that is weak and shifted to longer wavelengths because the Ca ii lines at −11 days and −8 days are dominated by the HVF component. The relative weakness of the PVF Ca ii IR3 feature in the model is a consequence of SYNOW setting the optical depth for each ion to represent the strongest line (in this case Ca ii H&K) and then calculating the strength of other lines from that ion assuming local thermodynamic equilibrium. The SYNOW parameters for the models are given in Table 4. In these simple models, the same HVF and PVF velocities are assigned to all ions except for HVF Ca ii and Fe ii.

Table 4. SYNOW Parameters for Model Spectra

Ion	τ	v_min	v_max	v_e	τ	v_min	v_max	v_e
	−11 day				−8 day
	v_phot = 16,000 km s⁻¹				v_phot = 13,800 km s⁻¹
Si ii	3.0	16	30	2.0	1.2	16	30	2.0
Si ii(HVF)	1.0	23	50	2.0	0.1	23	50	2.0
Si iii	0.8	16	50	2.0	0.8	13	50	2.0
Si iii(HVF)	0.5	23	50	2.0	0.5	23	50	2.0
Mg ii(HVF)	0.5	22	50	4.0	0.3	22	50	4.0
S ii	0.7	16	50	2.0	0.5	13	50	2.0
S ii(HVF)	0.1	23	50	2.0	0.1	23	50	2.0
Ca ii	2.0	16	50	8.0	2.0	13	50	8.0
Ca ii(HVF)	25	28	50	4.0	6.0	26	50	4.0
Fe ii(HVF)	0.5	28	50	2.0	0.05	23	50	2.0
Fe iii	0.5	16	50	2.0	0.5	16	50	2.0

Notes. τ: reference line optical depth, v_min: minimum velocity, v_max: maximum velocity, v_e: e-folding velocity of optical depth profile. All velocities are in units of 10³ km s⁻¹. Both models assume T_eff = 12, 000 K.

Download table as: ASCII Typeset image

Figure 10 can be compared to Figure 4 of Branch et al. (2005), which uses HVF Fe ii to improve the SYNOW fit to a −12 days spectrum from SN 1994D. Additional HVF contributions from Mg ii and Fe iii can be added to further improve the agreement between SYNOW model spectra and the data, but HVF from these lines are not clearly identified in SN 2009ig. Mg ii and Fe iii are not included in the velocity tables or plots.

7. COMPARISON TO HVF IN OTHER SNe Ia

Figure 11 shows the development of Si ii λ6355 absorption features by phase in six SNe Ia for which there are high-cadence spectroscopic observations that begin at least 12 days before B-max.¹⁵ We note that none of these SNe Ia are rapid decliners; their Δm₁₅(B) values (in SN order of discovery) are 1.02, 1.05, 0.90, 1.11, 0.86, and 0.80.

The spectra obtained before −10 days provide strong evidence for detached HVF regions from both Ca ii and Si ii in five of the six SN. With one exception, the earliest spectra require separate HVF and PVF components to fit the line profiles. Only SN 2011fe at −16 days has a broad absorption feature that can be fit with a single Gaussian.

The line profiles for every spectrum in Figure 11 can be correlated with a profile from one of the phases in the spectral sequence of SN 2009ig. In some cases, the phase must be offset slightly to establish the best match. For example, the sequence of line profiles from −10 days to −5 days in SN 2005cf is nearly identical to the −12 days to −7 days profiles in SN 2009ig. The features of SN 2003du from −12 days to −8 days are similar to the features of SN 2009ig at the same phases, but the HVF diminish more rapidly in SN 2003du.

The most extreme phase offset relative to SN 2009ig is found in SN 2011fe, for which the sequence from −16 days to −10 days corresponds to SN 2009ig from −12 days to −6 days. Although the first spectrum of SN 2011fe was obtained earlier than the first spectrum of SN 2009ig with respect to B-max, the spectrum of SN 2011fe corresponds to a phase in the sequence that is after the detached HVF are observed. There is only a It is possible that even earlier observations of SN 2011fe could have revealed HVF.

Previous authors have studied the eccentricities of the Si ii λ6355 line profile in SNe Ia without directly detecting HVF. The presence of a HVF component from Si ii λ6355 is inferred from unusual line profiles that are wide, square, or triangular. None of the spectra available before SN 2009ig display separate and distinct HVF Si ii features such as found at −14 and −13 days in SN 2009ig and from −14 days to −11 days in SN 2012fr.

Garavini et al. (2007) use SYNOW to model early-time spectra of eight SNe Ia. They show that fitting the Si ii λ6355 line profile requires a "detached" Si ii region at 20,000–22,000 km s⁻¹ for seven of the eight SNe Ia. The other SNe in their sample requires a detached region at 24,500 km s⁻¹.

Wang et al. (2009b) compare the early spectra of six SNe Ia and report that each of the Si ii λ6355 line profiles are well fit by a double-Gaussian function with separate central wavelengths. They show that two separate absorption features can be combined to produce flat-bottomed and triangular line profiles.

Figure 11 shows that for different SNe Ia the characteristic velocities of the HVF and PVF regions can vary by up to 3000 km s⁻¹, but in all cases a gap of at least 5000 km s⁻¹ is maintained between the HVF and PVF. Garavini et al. (2007), Stanishev et al. (2007), Wang et al. (2009b), and Childress et al. (2013b) describe the separation between HVF and PVF components to be about 8000 km s⁻¹, the same value as the velocity gap in SN 2009ig.

The correlation of line profiles from pre-maximum spectra of other SNe Ia to the spectral sequence of SN 2009ig confirms a suggestion made by Stanishev et al. (2007) that the "peculiar" profiles of Si ii λ6355 in early-time spectra of SN Ia can be explained as part of a common evolutionary sequence that includes separate HVF and PVF components. Stanishev et al. (2007) also propose that the strength of the HVF components from Si ii and Ca ii lines is correlated in SNe Ia. Thus, if one of these lines produces strong HVF, then the other will also have a strong HVF component. The spectra of SN 2009ig confirm this relationship, and they also demonstrate that the presence of strong HVF from Si ii and Ca ii may be an indicator for HVF of Si iii, S ii, and Fe ii.

The agreement between the SN 2009ig results and these well-studied SN Ia demonstrates that the behavior of HVF line forming regions in spectra of SN 2009ig is not unique, and in fact is quite common. The papers discussed in this section identify fifteen SNe Ia with spectra obtained at −10 days or earlier. All fifteen SN display HVF Si ii at velocities between 17,000 and 22,000 km s⁻¹, and all but SN 2011fe show evidence for separate HVF and PVF absorption components.

8. POSSIBLE SOURCES OF HVF

The observations of detached HVF in SN 2009ig suggest the presence of a region at 20,000–23,000 km s⁻¹ in the velocity space of the SN with a localized enhancement of element abundance or density, or both. The proposed line forming region would be the source of HVF for Si ii, S ii, and Fe ii from −12 days to −5 days.

This simple model is complicated by the fact that HVF are initially detected at a range of even higher velocities. In addition, the HVF velocities of Si ii, Ca ii, and Fe ii decline at different rates. This suggests that the lines may sample different density structures. Although the velocity ranges overlap, the HVF of each line are not even detected in the same region. Figure 8 shows that at −14 days, Si ii λ6355 has the narrowest detached HVF with a velocity range from 10,000 to 30,000 km s⁻¹. At the same phase, the HVF of Ca ii and Fe ii HVF extend from 10,000 to 50,000 km s⁻¹, but they are offset from each other.

Ca ii absorptions are exceptionally broad in velocity space, with absorption minima that are several thousand km s⁻¹ higher than the HVF components of other lines. Due to the very low excitation potentials of the Ca ii lines (⩽1.7 eV), they are capable of producing detectable absorptions at much lower abundances than most other lines found in SNe Ia. Fe ii lines also have low excitation potentials (⩽2.9 eV). In SN 2009ig, the HVF from Fe ii are nearly as wide and deep as HVF of Ca ii, with velocities between Ca ii and Si ii.

The simplest explanation for both the HVF Ca and Fe observations is that primordial abundances of these elements are sufficient to produce the observed absorptions. The line profiles can be produced by light from the SN passing through a line forming region that extends from about 10,000 to 50,000 km s⁻¹ in velocity space. The observed minima for these features would move to lower velocities with time, as expansion of the SN reduces opacity for HVFs more rapidly than it does at lower velocities. As the column depth is reduced, the observed minima become nearly constant at velocities near to the inner edge of the line forming region.

If the HVF of Ca ii and Fe ii are formed in regions of unburned material from the progenitor, it is still a challenge to explain the simultaneous presence of detached HVF for Si ii λ6355 (8.1 eV) and S ii (13.7 eV). An explanation is also required for the fact that the HVF velocities for all lines, as measured at the absorption minima, eventually converge near 21,500 km s⁻¹. Since Figure 8 shows that the line forming materials are distributed through a wide range of velocities, there must be some motivation for the line opacities of Si ii, Si iii, S ii, Ca ii, and Fe ii to have approximately equal values at the same phases and in a relatively narrow velocity range.

Attempts have been made to explain HVF without invoking a separate line forming region. The flat-bottomed and triangular shapes observed in the Si ii λ6355 line profile at early times receive particular attention. None of the physical models used to generate unusual line profiles survive comparisons with early observations of multiple SN Ia. HVF from Ca ii have also been attributed to opacity effects from recombination, but the time scale of recombination is a day or two and HVF from Ca ii are observed from 8 to 15 days in individual SNe Ia.

Explosion models for SNe Ia make predictions for structures that range from spherical to highly asymmetric. The resulting chemical distributions may be simply stratified by atomic mass, violently mixed between regions of burned material, or contain bubbles of material moving between burning regions. Model results can be tuned to produce specific layering compositions and velocity distributions that may mimic HVF of specific atoms at specific locations. We are not aware of any explosion models that predict the relative uniformity across a range of absorbing materials that is observed for HVF in SN 2009ig.

One way to increase the local density in HVF regions is by interaction between the expanding ejecta and nearby circumstellar material (CSM; Gerardy et al. 2004). The shock front sweeps up CSM and a shell is "locked in" after the shock moves on. The shell will continue in homologous expansion with the rest of the SN ejecta. This model satisfies the constraint that the HVF line forming region exist in a narrow, unchanging velocity window and explains the asymmetric line profiles of the Si ii λ6355 HVF.

Three of the recent models have been able to produce variable density structures: symmetric ignitions that generate strong mixing (Maeda et al. 2010), gravitationally confined detonations with burning regions on the surface of the WD (Meakin et al. 2009), and double-detonation models that produce an HVF burning region with a density gap between it and the center of the SN (Shen et al. 2013).

9. SUMMARY AND CONCLUSIONS

We present a time series of spectra of SN 2009ig obtained between −14.5 days and +12.5 days with respect to the time of B-max. The two earliest observations in our sample (−14.5 days and −13.5 days) are the first spectra of an SN Ia to resolve the HVF component of Si ii λ6355 as a distinctly separate absorption feature. HVF are identified in the spectra from Si ii, Si iii, S ii, Ca ii, and Fe ii, and we show that the line profiles require a detached HVF line forming region.

Simultaneous detections are made for HVF and PVF of the same lines from −12 days to −5 days. Velocity measurements demonstrate that the HVF and PVF line forming regions are separated by ∼8000 km s⁻¹ as long as HVF are detected. Using a mean velocity of 21,500 km s⁻¹ for the HVF layer and 13,000 km s⁻¹ for the PVF layer, we show that R_HVF ≈ 1.6 R_PVF.

The initial HVF velocities and the velocity decline rates are different for each line. The fact that the HVF velocities eventually converge suggests the presence of a line forming region in SN 20019ig at velocities between 20,000 and 23,000 km s⁻¹. A comparison of the HVF from Si ii, Ca ii, and Fe ii by phase in velocity space shows that HVF from different lines are formed in overlapping but not identical regions. HVF of Ca ii and Fe ii have strong features that range from 10,000 to 50,000 km s⁻¹ in the earliest spectra, but the locations of their absorption minima differ by ∼7000 km s⁻¹. At the same phases, Si ii HVF are confined to a much narrower region that extends from 10,000 to only 30,000 km s⁻¹. We explore the physical conditions that can produce detached HVF, and find that it is challenging for any model to reproduce the observations of SN 2009ig.

The spectra of SN 2009ig form a complete map of the transition from the earliest observations of detached HVF to features with contributions from both HVF and PVF, and finally to phases when all of the absorptions are PVF. HVF of both Ca ii and Si ii are frequent characteristics of SNe Ia observed before −10 days, and that allows us to compare our results to observations of several other SNe Ia. All of the line profiles found in early-time spectra of SNe Ia can easily be correlated with individual profiles that are part of the development sequence defined by SN 2009ig. We interpret this as evidence for a common evolutionary sequence in SNe Ia that requires both HVF and PVF line forming regions.

The authors wish to thank the Chairmen of the TACs from the University of Texas and Penn State University for providing Director's discretionary time for the HET/LRS observations. G.H.M. thanks Nick Suntzeff and Rob Robinson for insightful comments. G.H.M. also thanks Mark Phillips for providing perspective, as well as Michael Childress for helpful discussions and for sharing an advance copy of his paper on SN 2012fr. We thank Mark Sullivan, Isobel Hook, Peter Nugent, Andy Howell, and Bill Vacca for sharing their own data on line profiles in young SNe Ia. The authors make frequent use of David Bishop's excellent Web site listing recent supernovae and valuable references associated with them: http://www.rochesterastronomy.org/snimages/. J.V. is supported by Hungarian OTKA Grants K-76816 and NN-107637, NSF grant AST-0707769, and Texas Advanced Research Project ARP-009. The research of J.C.W. in supported in part by NSF grant AST-1109801. The CfA Supernova Program is supported by NSF grant AST-1211196 to the Harvard College Observatory. X. Wang is supported by the Natural Science Foundation of China (NSFC 11178003, 11073013), the China-973 Program 2009CB824800, and NSF grant AST-0708873 (through L. Wang). E.Y.H. is supported by NSF grant AST-1008343. A.V.F. is grateful for financial assistance from NSF grant AST-1211916, the TABASGO Foundation, and the Christopher R. Redlich Fund. We greatly appreciate the valuable assistance provided by staff members at the observatories where SN 2009ig was observed.

Facilities: FLWO:1.5m (FAST) - Fred Lawrence Whipple Observatory's 1.5 meter Telescope, HET (LRS) - McDonald Observatory's Hobby-Eberly Telescope, Keck:I (LRIS) - KECK I Telescope, MMT (Blue Channel) - MMT at Fred Lawrence Whipple Observatory, Shane (Kast) - Lick Observatory's 3m Shane Telescope, Swift (UVOT; UV grism) - Swift Gamma-Ray Burst Mission

HIGH-VELOCITY LINE FORMING REGIONS IN THE TYPE Ia SUPERNOVA 2009ig

Article metrics

Permissions

Author e-mails

Author affiliations

Author notes

Dates

ABSTRACT

1. INTRODUCTION