NEAR-INFRARED SPECTROSCOPY OF THE TYPE IIn SN 2010jl: EVIDENCE FOR HIGH VELOCITY EJECTA

Type IIn supernovae (SNe IIn) are characterized by narrow optical emission lines and a blue continuum (Schlegel 1990). The presence of a narrow line component at early times is taken to indicate a dense surrounding circumstellar medium that is heated and ionized by radiation from the explosion and the continuing interaction. The presence of wings on the narrow lines, as observed for the Type IIn SN 1998S (Leonard et al. 2000), can at least in some cases be explained by electron scattering in a mildly optically thick medium (Chugai 2001). The optical luminosity of a Type IIn event can also be explained by shock interaction with a dense circumstellar medium (Chugai & Danziger 1994). The velocity of the circumstellar mass loss can be deduced from the velocities indicated by P Cygni line profiles observed in the H Balmer lines; the typical line velocities are 100–1400 km s⁻¹ (Kiewe et al. 2012). The mass loss rates implied by the interaction luminosities are ∼0.01–0.1 M_☉ yr⁻¹. High mass loss rates were also derived from infrared (IR) emission attributed to circumstellar dust (Fox et al. 2011). Roughly 6%–11% of core–collapse supernovae are of Type IIn (Smith et al. 2011a).

These rates of mass loss are much higher than those found in normal stars, and the place of Type IIn events in stellar evolution is unclear. The progenitor stars have been linked to luminous blue variables (LBVs) because these have the requisite mass loss density and outflow velocity during their eruptive phases (e.g., Smith et al. 2010; Fox et al. 2011; Kiewe et al. 2012). In addition, the progenitor of SN 2005gl is consistent with an LBV (Gal-Yam & Leonard 2009). A problem is that stars in this evolutionary phase are not expected to undergo supernova explosions (although see the rotating models of Groh et al. 2013). Suggestions for the proximity in time of mass loss and explosion include wave-driven mass loss (Quataert & Shiode 2012) and binary-driven mass loss (Chevalier 2012), but these studies are very speculative. Soker (2013) also attributes the dense circumstellar medium around SNe IIn to binary interaction.

The bright Type IIn supernova SN 2010jl was discovered on UT 2010 November 3.5 by Newton & Puckett (2010). It was subsequently confirmed at an unfiltered magnitude of 12.9. Pre-discovery images showed that the supernova was present on 2010 October 9.6 (Stoll et al. 2011), which we use as our zero point in time. The supernova is located at R.A. $=9^{\rm h}42^{\rm m}53\buildrel{\mathrm{s}}\over{.}33$, decl. = +9°29'418 (equinox 2000.0). It lies 24 east and 77 north of the center of the host galaxy, UGC 5189A. We take the distance to the supernova to be 49 Mpc (Smith et al. 2011b) and the reddening to be E(B − V) = 0.058 (Fransson et al. 2014). Smith et al. (2011b) used pre-explosion Hubble Space Telescope imaging to identify a bright, blue point source coincident with the position of the supernova, which they took to imply that the progenitor of SN 2010jl had a mass above 30 M_☉. Andrews et al. (2011) observed the supernova in the Spitzer 3.6 and 4.5 μみゅーm bands and the JHK bands during days 90–108, finding evidence for 7500 K and 750 K components. However, Fransson et al. (2014) have examined the IR data for this age and find the temperature of the cool component to be ∼1800 K; this temperature characterizes the cool component for 100's of days. Fransson et al. (2014) presented one 0.3–2.4 μみゅーm spectrum of SN 2010jl (on day 460), as well as extensive optical and ultraviolet observations and near-infrared (NIR) photometry. Spitzer observations were also presented by Fox et al. (2013), who find evidence for circumstellar emission and a high mass loss rate. Maeda et al. (2013) presented a spectrum (optical through NIR) on day 563, finding evidence for dust formation in the supernova ejecta.

SN 2010jl has also been detected as an X-ray source (Immler et al. 2010; Chandra et al. 2012; Ofek et al. 2014). Observations on 2010 December 7–8 (day 58 since discovery of the supernova) showed a hot (kT > 8 keV) thermal spectrum with an absorbing column ∼10²⁴ cm⁻², assuming incomplete ionization of the circumstellar gas and a metallicity 0.3 of solar (Chandra et al. 2012). This column density corresponds to an electron scattering optical depth ∼1. By 2011 October 17–18 (day 372), the absorbing column declined by a factor of three, but the X-ray emission remained steady and hot (kT > 12 keV). However, Ofek et al. (2014) re-examined the Chandra data and found them difficult to model. Hard X-ray observations with NuSTAR on 2012 October 6 yielded a temperature of ${\sim } 18_{-4}^{+6}$ keV (Ofek et al. 2014), which corresponds to a shock velocity of ∼4000 km s⁻¹.

Here we present a set of 10 NIR spectra of SN 2010jl spanning 529 days (Section 2). The various components contributing to the line and continuous emission are described in Section 3. The physical picture implied by the observations is discussed in Section 4 and the conclusions are in Section 5.

Data presented here were obtained using TripleSpec, an 0.9–2.4 μみゅーm, R  ∼  3000 spectrograph, on the ARC 3.5 m telescope at Apache Point Observatory (Wilson et al. 2004; Herter et al. 2008). The supernova was observed 10 times between 2010 November 15 and 2012 April 26. Table 1 shows the details of each observation. Day numbers listed in Table 1 refer to time elapsed since the first detection of the supernova on 2010 October 9. Integration time refers to the total exposure time for each date. Individual exposures were five minutes in length and nodded in an ABBA pattern along the slit. We also observed the spectrophotometric standard HD 85377 for the telluric correction and photometric calibration. The airmass at the time of each observation is given in Table 1 for both SN 2010jl and HD 85377. Sky subtraction was achieved by differencing each pair of A and B exposures. Sky subtraction, spectral extraction, telluric correction, and photometric calibration were carried out using a modified version of the IDL utility SpexTool (Cushing et al. 2004).

Figures 1–3 show the spectra in the J, H, and K bands, respectively. The spectra are normalized to have the same continuum intensity at 1.55 μみゅーm. Identifications of the dominant lines are given.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Variations in seeing and cloud cover are problematic for photometric calibration of any long-slit spectrum. Therefore, the photometric calibrations produced by SpexTool are only roughly accurate. Much of the following discussion deals with descriptions of the line or continuum shapes, rendering a precise photometric calibration unimportant. Figures 5, 9, 15, and 16 show the true flux density as revealed by photometric observations collected by Fransson et al. (2014). To estimate H-band magnitudes at the exact dates of our observations we fit separate second degree polynomials to the early time (before 250 days) and late time (after 403 days) H-band light curves published by Fransson et al. (2014). Because Fransson et al. (2014) does not publish photometry nearby in time to day 403, we used cubic spline interpolation over the full light curve to estimate the magnitude at 403 days. As a result, the calibration to the photometry for day 403 is somewhat more uncertain than that at other dates. Nevertheless, the behavior of the flux density at day 403 seems to fall in line with the other observations. Once H magnitudes had been determined for each TripleSpec observation, we converted each magnitude to a flux density using

$$\begin{equation} H = -2.5 \log (f_{1.63\, \mu {\rm m}}) - 24.860, \end{equation} \tag{ 1 }$$

where 1.63 μみゅーm and −24.860 are the effective wavelength and magnitude zero point of the NIR H band as given by Bessell et al. (1998). We then scaled the spectra output by SpexTool to the level of the calculated flux density at 1.63 μみゅーm.

3.1. Continuum Emission

At early times the continuum emission has a slope that is slightly flatter than Rayleigh–Jeans. The flatness of the continuum relative to the blackbody shape indicates the combination of warm and cool components. The reason for this can be seen in the decomposition of the continuum emission by Fransson et al. (2014), who model the optical through IR spectrum and find that a relatively cool (1800–1900 K) component is present from early times as well as a warmer (∼7000 K) component. Because the blackbody peak is in the optical range, temperatures determined from optical observations should be more reliable than those determined from the IR through day 219. At times in excess of one year, the continuum emission has a local maximum within the wavelength range of our observations. On 2011 November 17 (403 days), the continuum peak is found at about 1.5 μみゅーm. By 2012 February 9 (488 days), it has moved to about 1.7 μみゅーm. Figure 4 presents the spectrum at 488 days overlaid with five blackbodies ranging in temperature from 1500 to 1900 K, showing that the observed continuum is flatter than a blackbody. Modifying the blackbody curve to account for the emissivities of various dust grains does not produce a shape which fits the continuum emission because the efficiency of dust emission drops with increasing wavelength.

Zoom In Zoom Out Reset image size

Assuming a cooler component contributes to the emission at wavelengths longer than 1.5 μみゅーm, we identified selected areas of continuum emission and fit to a blackbody using the IDL fitting package MPFIT (Markwardt 2009; Moré 1978) for each observation later than 1 yr. The temperatures of these fits are ∼2000 K and are shown in Table 2. In Figure 4, the best fits to the continuum short of 1.9 μみゅーm lie well within 100 K of 1800 K. We therefore cite an approximate uncertainty of ±75 K for these temperatures. This is the uncertainty in the shortest wavelength emission observed in the NIR and does not take into account the evidence for cooler emission at wavelengths longer than 1.9 μみゅーm. By fitting the shortest wavelength emission we estimate the temperature of the warmest dust. Although the shape of the continuum on days 488 and 565 does not fit a pure blackbody, a temperature of ∼1900–2000 K is consistent with emission from hot dust and with the observations of Fransson et al. (2014). Such emission has previously been observed in a number of SNe IIn (Fox et al. 2011).

Table 2. Blackbody Fits to the Near Infrared

Observation	Supernova	Blackbody
Date	Age (days)	Temperature (K)
2011 Nov 17	403	2090
2012 Feb 9	488	1890
2012 Apr 26	565	1910

Note. Uncertainty in each temperature is approximately ±75 K.

Download table as:  ASCIITypeset image

3.2. Hydrogen Lines

As can be seen in Figures 1–3, many emission lines of the Paschen and Brackett series of H are detected. Each line shows two distinct components: a narrow component that is close to zero velocity in the host galaxy and a broad component. Although narrow lines could have their origin in the interstellar medium of the host galaxy, we attribute the emission primarily to the unshocked circumstellar medium. One reason is that the fluxes of narrow lines at later times (t ≳ 380 days) decline to about 1/10 of the fluxes at earlier times, implying we can comfortably ignore the contribution of host galaxy emission lines at early times. Another reason is that interstellar emission lines would also be present in the Balmer series. Observations of the early Hαあるふぁ line show a narrow P Cygni feature (Smith et al. 2011b; Fransson et al. 2014), indicating that the circumstellar medium is optically thick in the H Balmer lines. The P Cygni features observed in the Hαあるふぁ line indicate an expansion velocity of 100 km s⁻¹ (Fransson et al. 2014). The Paschen and Brackett series lines do not show absorption, suggesting they are optically thin. The narrow lines are plausibly formed in a region of circumstellar mass loss.

Broadened emission from hydrogen is visible in the data throughout the accessible lines of the Paschen and Brackett series. The best detected lines are Paschen βべーた, Brackett γがんま, and Paschen δでるた. The shape and evolution of the Paschen βべーた line are shown in Figure 5. Paschen γがんま is also strong, but is heavily blended with the red side of the broad He i λらむだ10830 line. Paschen is sometimes observed, but it lies on the edge of an atmospheric absorption region, making the determination of its shape and strength strongly dependent on changing atmospheric conditions. Many of the weaker Brackett series lines are identifiable in the data, and corroborate the common line shape of broad H emission.

Zoom In Zoom Out Reset image size

At early times, the broad component is fairly symmetric about zero velocity. In other SNe IIn, a Lorentzian profile gives a reasonable fit to broad Balmer lines (Leonard et al. 2000; Smith et al. 2010). Here, the broad Paschen βべーた line is moderately well fit by a Lorentzian profile (see also Smith et al. 2011b, for Hαあるふぁ). However, the broad line profile is presumably the result of the thermal particle velocities and scattering; this situation does not naturally produce a Lorentzian profile. The broad line profile is naturally produced by electron scattering (Chugai 2001; Fransson et al. 2014), without the need for Doppler effects due to bulk motion. The narrow line component is made up of photons that escape without scattering, so its strength relative to the broad component places constraints on the scattering optical depth to the line photon source. At early times, the broad component is symmetric, implying that the electron thermal velocities are much larger than systematic velocities. At late times, the broad lines are seen to become asymmetric with regard to zero velocity, although the Hαあるふぁ line remains fairly symmetric relative to the line peak (Fransson et al. 2014); the peak of the broad component shifts to the blue by ∼700 km s⁻¹, as also occurs in the optical lines. Figure 5 shows that the extended wings in the Paschen βべーた line have diminished by day 400. We take this to indicate that the electron scattering optical depth is decreasing, which could occur because the shock wave overruns the scattering region or there is a change in the ionization of the gas.

In addition to finding that the emission lines shift to the blue with time, Fransson et al. (2014) found that the lines remain symmetric about a blueshifted wavelength during the shift. They used this property to argue that the line emission and scattering occur in a comoving slab of material. In Figure 6 we show the Paschen βべーた line on day 219 as well as its reflection about a particular velocity, using a procedure essentially the same as that used by Fransson et al. (2014). The velocity of the shift is chosen so that the wings of the lines over 1000–5000 km s⁻¹ match up. It can be seen that there is an asymmetry near the peak of the line. Although the asymmetry is small, it shows that the hypothesis of electron scattering in a comoving slab is not completely accurate. Fransson et al. (2014) presented NIR spectroscopy of the supernova on day 465, finding symmetry of the lines at that time. Our late time spectra also show that the degree of asymmetry becomes small.

Zoom In Zoom Out Reset image size

Figure 7 compares the broad components of the Paschen βべーた, Brackett γがんま, and Paschen δでるた lines at one time, 164 days. It can be seen that the FWHM of the lines are similar. There are deviations in the line wings, but this aspect is sensitive to the continuum subtraction. We conclude that there is no strong evidence for differences between the lines. In the case of SN 1994W, Dessart et al. (2009) discussed differences between these lines caused by differences in optical depth. However, those authors were considering a case where the emission was from shocked, and thus very high density, gas. In our case, we believe the broad lines are formed in the unshocked circumstellar medium, where the densities are lower.

Zoom In Zoom Out Reset image size

Figure 8 and Table 3 shows the evolution of the flux in the narrow component to that in the broad component of Paschen βべーた, Paschen δでるた, Brackett γがんま, and O i λらむだ11287. Figure 8 and Table 4 show the evolution of the equivalent widths of the prominent lines. In order to measure both narrow component strength to broad component strength ratios and equivalent widths, a linear fit to the continuum in the vicinity of each emission line was found. The ratios of narrow component strength to broad component strength were measured by direct integration over each component after subtraction of the continuum fit. The quoted uncertainties were found by allowing the full width of the narrow component to vary by one pixel in each direction while also taking into account the noise in the continuum about the line. It can be seen that the ratio of narrow to broad line flux decreases with time. Equivalent widths were measured by directly integrating over the ratio of line to continuum. The quoted uncertainties were found by allowing the constant parameter in the linear continuum fit to vary by 1σしぐま in both directions. There is an initial approximately linear increase of the equivalent width with time, followed by a decline between days 220 and 400. Zhang et al. (2012) and Fransson et al. (2014) found a similar evolution for the Hαあるふぁ line.

Zoom In Zoom Out Reset image size

Table 3. Narrow to Broad Line Ratio

Date	Day	Paschen βべーた	Paschen δでるた	Brackett γがんま
2010 Nov 15	36	0.20 ± 0.028	0.12 ± 0.021	0.16 ± 0.029
2010 Dec 2	53	0.15 ± 0.023	0.08 ± 0.019	0.17 ± 0.028
2011 Jan 26	108	0.07 ± 0.017	0.05 ± 0.017	0.08 ± 0.023
2011 Feb 22	135	0.05 ± 0.017	0.03 ± 0.016	0.06 ± 0.021
2011 Mar 23	164	0.05 ± 0.017	0.03 ± 0.016	0.05 ± 0.020
2011 Apr 6	178	0.04 ± 0.017	0.03 ± 0.017	0.05 ± 0.020
2011 May 17	219	0.03 ± 0.017	0.03 ± 0.016	0.04 ± 0.020
2011 Nov 17	403	0.02 ± 0.023	0.02 ± 0.026	0.03 ± 0.026
2012 Feb 9	488	0.04 ± 0.027	0.04 ± 0.033	0.06 ± 0.029
2012 Apr 26	565	0.11 ± 0.032	⋅⋅⋅	⋅⋅⋅

Download table as:  ASCIITypeset image

Table 4. Equivalent Widths (Angstroms)

Date	Day	Paschen βべーた	Paschen δでるた	Brackett γがんま
2010 Nov 15	36	84 ± 20	25 ± 8	50 ± 17
2010 Dec 2	53	114 ± 16	36 ± 7	59 ± 12
2011 Jan 26	108	291 ± 38	75 ± 18	119 ± 32
2011 Feb 22	135	371 ± 34	93 ± 17	160 ± 24
2011 Mar 23	164	441 ± 38	111 ± 20	207 ± 37
2011 Apr 6	178	480 ± 48	118 ± 22	219 ± 39
2011 May 17	219	573 ± 58	148 ± 26	261 ± 60
2011 Nov 17	403	156 ± 11	64 ± 20	29 ± 5
2012 Feb 9	488	93 ± 11	41 ± 15	12 ± 4
2012 Apr 26	565	71 ± 29	⋅⋅⋅	⋅⋅⋅

Note. Broad and narrow line fluxes are included.

Download table as:  ASCIITypeset image

3.3. He Lines

Broad He emission features are detected at 1.0830 μみゅーm and 2.0587 μみゅーm. The evolution of the He lines, especially He i λらむだ10830, shows some similarities to the H lines, although there are significant differences (Figure 9). The broad component shows an increasing equivalent width, in the same way as the H lines (Figure 8). A comparison of the He i λらむだ10830 line with the Paschen βべーた line on day 164 is shown in Figure 10. It can be seen that both lines show extended wings going out ∼8000–9000 km s⁻¹. On the red side of the He i λらむだ10830 line, there is the superposition of the Paschen γがんま line, which has both broad and narrow emission components like the Paschen βべーた line. Assuming the line profile of Paschen γがんま is identical to Paschen βべーた, in Figure 10 we use Paschen βべーた as a template to obtain a Paschen γがんま subtracted He profile. The shift and scale of the Paschen βべーた template is chosen in order to best fit the template narrow peak to the narrow peak of Paschen γがんま. The peak of the broad component shows a shift to the blue by ∼1000 km s⁻¹ in contrast to the ∼700 km s⁻¹ shift in the peak of the H lines (Figure 10 inset). As described below, we attribute this difference to an additional component in the He line.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

On the blue side on day 164, there is a shoulder feature in the He i line that is not present in the H line. In Figure 10, the Paschen γがんま-subtracted He i line is overplotted with the Paschen βべーた line so that the line profiles match at positive velocities. We have used a similar procedure to search for excess emission on the other dates of observation. On days 36 and 53, the feature is not discernable; on days 402 and 488, there is not a well defined shoulder feature, but there is weak excess He emission that peaks at ∼ − 2500 km s⁻¹. The feature grows and fades in relative strength roughly in the same way as the broad H lines. One possibility for the origin of this feature is that it is a separate line with only a broad component and no narrow component. However, a comparison of the He i λらむだ10830 line to the He i λらむだ20587 line (Figure 11) shows that the shoulder feature is present in both lines. The broad wings are likely due to electron scattering; components with different temperatures are possible, but this would not produce a shoulder structure on one side of the line profile. The implication is that the feature is due to material with systematic velocities up to ∼5000 km s⁻¹ toward the observer. Figure 12 shows the excess blueshifted emission in the He i λらむだ10830 line relative to the Paschen βべーた line; the emission extends to at least −6000 km s⁻¹. In addition, the shoulder feature of excess emission moves to the red with time (Figure 13), which is the opposite of the evolution of the H broad features. The deceleration of the break in the shoulder feature follows a velocity evolution t^{−0.39 ± 0.08} (Figure 14). The error bars in the shoulder-break velocities shown in Figure 14 are estimated by measuring the overlap of the 1σしぐま uncertainty regions for each linear fit component shown in Figure 12.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In addition to the broad line feature, there is a narrow P Cygni feature in the He i λらむだ10830 line. Figure 15 shows that initially (day 36) the emission component dominates over the narrow absorption. By day 53, the two components are at nearly equal strength and at later times, the absorption component dominates. Since the resolution of the observations is about 100 km s⁻¹, we place an upper limit of 100 km s⁻¹ on the wind velocity of the circumstellar medium. The width of the absorption line does not change with the time of our observations. The same is true for the Hαあるふぁ line (Fransson et al. 2014). This is indicative of the formation of the circumstellar medium in an approximately steady flow. If the material were launched in an eruptive event, the circumstellar medium velocity would increase roughly linearly with radius, and over time the observed emission-line widths should increase. This is not observed, at least within the spectral resolution. The narrow component in the He i λらむだ20587 line is not as well defined as in the λらむだ10830 line; it appears only in absorption.

Zoom In Zoom Out Reset image size

3.4. Other Lines

The O i line at 1.129 μみゅーm is in a noisy part of the spectrum where there is confusion with night sky features. In addition, it overlaps the red wing of the He i λらむだ10830 line, but it is clearly present (Figure 16). The line shows a broad emission component with properties and an evolution that are comparable to those of the H lines. A narrow line may be detected but is uncertain because of the contamination by night sky lines. The observed emission is compatible with the view that the O i line is the result of pumping by the Lyβべーた line (Fransson et al. 2014).

Zoom In Zoom Out Reset image size

In addition, there is a line present at ∼1.2 μみゅーm (Figure 16). It has a broad, but no narrow, component. We tentatively identify it as Si i 1.1991–1.2270 μみゅーm. We also considered Mg i λらむだ11828, but it gave a poor fit to the line profile, under the assumption that the line profile is comparable to that of He i λらむだ10830. The line profile, when taken to be the Si i line, shows that there is emission associated with the blue shoulder emission observed in the He lines (Figure 17).

Zoom In Zoom Out Reset image size

Our interpretation of the NIR spectra builds on previous discussions of SNe IIn (Chugai & Danziger 1994; Leonard et al. 2000; Fransson et al. 2002, 2014; Hoffman et al. 2008). One component is the pre-supernova mass loss region with which the supernova is interacting. The mass loss has an outward velocity of 100 km s⁻¹ and is fairly steady over the region of observation. The region is photoionized by X-ray radiation coming from the shock interactions.

In our first epoch of NIR spectra, the H line profiles show a narrow component due to the slow wind and a broad symmetric component that we attribute to electron scattering in the slow wind. The line width reflects a combination of the electron scattering optical depth, τたう_e, and the gas temperature (Chugai 2001; C. Huang et al. 2015, in preparation). We initially assume that the emitting gas is the same as the scattering gas, and both are associated with the slow wind which has a density profile n∝r⁻² over an extended region. The scattering process was modeled by a Monte Carlo code to follow the scattering and wavelength shifts of line photons. As in Fransson et al. (2014), the broad line component is well reproduced by the electron scattering line profile (Figure 18). The model assumed spherical symmetry, with an absorbing sphere for an inner boundary R_i. As shown in C. Huang et al. (2015, in preparation), the FWHM of the scattered line, which has an approximately exponential shape, depends on the optical depth to R_i, τたう_e, and the temperature T. Provided that the ratio of outer radius to R_i is >10, the results are not sensitive to the exact value of the ratio. The line width in velocity space is proportional to the thermal velocity and is thus ∝T^1/2. For the Paschen βべーた line on day 36 we find the observed FWHM to be ∼2030 km s⁻¹, which is consistent with the following pairs of values of (τたう_e, T): (1, 58,600 K), (5, 16,400 K), (10, 7760 K), and (20, 3700 K).

Zoom In Zoom Out Reset image size

After interpolating over the narrow component of the day 36 Paschen βべーた line, we perform least-squares fits of the electron scattering models to the broad component. We find the reduced χかい² to lie in the range 1.39–1.55 for τたう_e ranging between 20 and 1 (Figure 18). The data are best fit to the τたう_e = 20 profile, with a reduced χかい² of 1.39. Such a high optical depth should be viewed with skepticism, especially in view of the small range of χかい² when τたう_e varies between 1 and 20.

If the emitting and scattering gas are the same, another constraint on the emission comes from the ratio of the narrow line flux to the flux in the broad wings, which is found to be 0.17 on day 36 for the Paschen βべーた line. Using the model results from C. Huang et al. (2015, in preparation), we find that the ratio observed for SN 2010jl corresponds to τたう_e ∼ 4, which suggests a lower limit of the scattering gas optical depth by assuming there is no other source of narrow line emission. The observed temperature is then ∼20, 000 K, which is reasonable for the ionized gas. Another measure of τたう_e is provided by the outer wings of the line. By adjusting the temperature, we can generate line profiles with the same FWHM at different τたう_e, the line wings become slightly stronger for higher values of τたう_e. However, our data for SN 2010jl do not have sufficient signal to noise in the wings to be able to use this method (Figure 18), as is indicated by the small range of reduced χかい² over the range of τたう_e from 1 to 20.

Although a consistent picture for mixed emission and scattering can be developed for the earliest time, it has problems at later times, indicating that application of the model to the narrow line at early times may be misleading. Over 100's of days, the broad component in the lines shifts to the blue, as also observed in optical lines (Fransson et al. 2014). One suggestion is that the shift is due to dust formation in the expanding medium (Smith et al. 2012; Maeda et al. 2013; Gall et al. 2014). In that picture, the redshifted photons emitted by material expanding away from the observer on the far side of the supernova are preferentially absorbed as they travel through a longer column of dust than their blueshifted near side counterparts. This would lead to a line profile that is less strong on the red side, an effect that would be more pronounced in the shorter wavelength lines as dust absorption within the ultraviolet/optical/NIR bands is more effective at shorter wavelengths. Smith et al. (2012) made this argument based on spectra at an age ∼100 days and Maeda et al. (2013) based on a spectrum at age 563 days. At the earlier times, our spectra show a shift in the Brackett γがんま line that is similar to that at shorter wavelengths (Figure 7), although there is some uncertainty in the continuum level. However, our observations cover a relatively small, long wavelength range and do not provide a good test of wavelength dependent line profiles.

Fransson et al. (2014) discussed a number of problems with the shift in the broad component being due to dust, and proposed a model in which there is radiative acceleration of the gas that gives rise to the broad component. Since the narrow component does not shift to the blue along with the broad component, the two components must originate in different places. The equivalent width in the narrow component of the H lines remains roughly constant over the first 200 days, so the ionized circumstellar medium appears to evolve slowly and the growth in the equivalent width of the H lines is primarily due to the growth of the broad component. Over the first 200 days, the NIR luminosity of SN 2010jl dropped by a factor of three (Fransson et al. 2014), so the narrow lines dropped by this factor and the flux in the broad components approximately doubled from our earliest observation to an age of 200 days. Formation of the narrow and broad lines in the same place would imply that the optical depth increases with time because fewer photons escape without scattering at higher optical depth. This is unexpected, giving further evidence that the line components form in separate regions. The properties of the H emitting regions are distinct from those of the X-ray emitting region, where the shock front has a velocity ∼4000 km s⁻¹ and a preshock τたう_e ∼ 1 on day 58 if there is incomplete ionization of the gas (Chandra et al. 2012; Ofek et al. 2014). The preshock column density subsequently declines. As discussed by Fransson et al. (2014), the fact that the broad component shows a systematic shift indicates that there is a large-scale asymmetry in the object and that the optical emission is perhaps from a polar flow.

As discussed in Section 2.3, the NIR He lines show a difference with the H lines. In addition to the growth of a broad component shifted by −1000 km s⁻¹, the He lines develop a shoulder out to −4000 to −5500 km s⁻¹. The velocity indicates that this component may be associated with the higher velocities inferred from the X-ray emission. The fact that this feature does not appear in the H lines indicates a difference in composition; the high velocity He feature may be associated with supernova ejecta in which H is underabundant.

A surprisingly similar situation has been observed in the Type IIn SN 1997eg (Hoffman et al. 2008). SN 1997eg had a maximum absolute V magnitude ∼1.5 mag fainter than SN 2010jl, so a lower density interaction is indicated. Hoffman et al. (2008) find that the H lines do not have Lorentzian profiles, implying that electron scattering related to the thermal velocities of electrons is not important and consistent with a low density. In addition, SN 1997eg was detected as a radio source at an age of 7 months (Lacey et al. 1998). No detections of SN 2010jl have yet been reported, suggesting higher absorption in this case. Despite the apparent difference in circumstellar density, there are interesting similarities in the H and He lines. In SN 1997eg, the H Balmer lines showed a shift in the peak of the line to the blue by ∼700–800 km s⁻¹. At the same time, the He i λらむだ7065 and λらむだ5876 lines showed a strong shoulder of emission to the blue by ∼5000 km s⁻¹. At later times, a peak like the one extending to the blue developed to the red in the H lines. Hoffman et al. (2008) interpret the line emission as coming from an asymmetric region with different geometries for the H and He dominated emission regions. The H lines are attributed to shocks driven into a dense circumstellar disk or torus, while the He emission is from H depleted gas that is expanding more rapidly into a lower density circumstellar medium. This view is supported by polarization observations of the supernova.

Our observations of H and He lines in SN 2010jl show similar shifts. The H lines show only a blueshifted component, without the redshifted side. We attribute the difference to the higher density and optical depths in the SN 2010jl case. There is also evidence for blueshifted He rich gas moving at 5000 km s⁻¹. However, the He emission is also strong in the lower velocity component, as opposed to the case of SN 1997eg.

Our spectral line observations indicate three kinds of emitting regions: a narrow line region of presupernova mass loss with velocities ∼100 km s⁻¹, an intermediate line region involving ∼700 km s⁻¹ gas in an asymmetric structure, and a He dominant region of supernova ejecta with velocities ∼5000 km s⁻¹. The lines are affected by electron scattering in the early phases of evolution and electron scattering may be responsible for the extended line wings through day ∼400. After day 400, the lines no longer show the characteristic electron scattering profiles and the line widths probably reflect gas velocities to ∼2000 km s⁻¹. However, the bulk of H rich material is moving at ∼700 km s⁻¹. The broad feature observed in the He lines could be produced by a uniformly expanding shell of gas with velocity ∼5000 km s⁻¹; the absence of the redshifted part of the shell may be due to the supernova being opaque. Similar regions have been observed in other SNe IIn, although the high velocity ejecta region is sometimes observed to be O rich, e.g., SN 1986J (Milisavljevic et al. 2008) and SN 1995N (Fransson et al. 2002).

Our study shows that NIR spectral observations of SNe IIn can provide an interesting window on these events. For the H lines in SN 2010jl at early times, the NIR lines (Brackett and Paschen series) have broad components with profiles that are similar to the broad components of the optical Balmer lines (Zhang et al. 2012; Fransson et al. 2014). The line profiles are consistent with the lines being formed by electron scattering in the circumstellar medium. The narrow line components in the Balmer lines show absorption features, i.e., they are optically thick, while the Paschen and Brackett lines are not. The NIR narrow emission thus gives a measure of the flux of unscattered line photons emitted by the slow circumstellar medium.

In SN 2010jl, the optical He i lines are relatively weak so there is not much information on their profiles. However, the NIR λらむだ10830 line is strong and the profile is well defined in our observations. We find a clear difference with the H lines in that there is a shoulder to the blue that implies gas approaching at 4000–5500 km s⁻¹. The emission can be interpreted as H poor supernova ejecta that are expanding into a region of relatively small circumstellar density. A similar situation is observed in SN 1997eg (Hoffman et al. 2008).

The NIR continuum can provide important information on the emission from warm dust, which appears to be frequently present around SNe IIn (Fox et al. 2011). However, at early times the hotter photospheric emission is present and optical observations are needed to separate out the photospheric emission from the dust emission (Fransson et al. 2014). At later times, when the photospheric emission has faded, the NIR observations can provide crucial information on dust emission.

We are grateful to Ori Fox for help with the observations as well as discussion of the results. We thank Claes Fransson for fruitful discussions and correspondence on SN 2010jl, and the referee who provided detailed comments that led to significant improvement of the paper. Thanks are also due to Meredith Drosback, Sarah Schmidt, and Yue Shen, who very graciously donated telescope time for our observations of SN 2010jl. We also wish to thank Mike Skrutskie for his help with gathering, reducing, and interpreting the spectra. This research was supported in part by NSF grant AST-0807727 and NASA grant NNX12AF90G. The research has made use of the SIMBAD database, operated at CDS, Strasbourg, France, and the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

Facility: ARC - Astrophysical Research Consortium 3.5m Telescope at Apache Point Observatory