PTF10nvg: AN OUTBURSTING CLASS I PROTOSTAR IN THE PELICAN/NORTH AMERICAN NEBULA

Although PTF10nvg is not known historically as an optical source, it was detected by PTF at m_R ≈ 18 mag in mid-2009 but then faded to ∼19.25 mag on a timescale of roughly 2 months. It brightened again by more than 4 mag in less than 200 days, while the field was too close to the Sun for observations. PTF10nvg reached m_R = 15 early in 2010, faded to 16.3 mag within 40 days of the first peak, and then brightened again to 13.5 mag before fading to ∼15.5 mag in early 2010 November. The two brightness peaks, as well as the intermediate lull, are all ∼50 days in width.

Pre-outburst photometry that we have collated from the literature is presented in Table 4, and the spectral energy distributions (SEDs) assembled from this pre-outburst photometry and our own post-outburst detection data are displayed in Figure 5.

Zoom In Zoom Out Reset image size

From 3 to 10 μm, the SED is rising then becomes roughly flat from 10 to 100 μm. The integrated luminosity of the pre-outburst IR source between 3 and 100 μm is 25 L_☉; extrapolating the SED longward of the IRAS 100 μm measurement as a standard Rayleigh–Jeans function, and integrating to λ = ∞, increases the luminosity by ∼40%. Other Class I sources, however, have SEDs that peak at or slightly longward of 100 μm (e.g., Enoch et al. 2009); if PTF10nvg's SED possesses a similar peak, the total luminosity would be larger than this, which assumes a sharply declining power-law directly from 100 μm. The integrated luminosity of the outbursting source between 0.7 and 2.2 μm is 2.5 L_☉, compared to <0.016 L_☉ before the outburst. There is no post-outburst photometry longward of 2.2 μm.

The optical and NIR colors of PTF10nvg, shown in Figure 6, are extremely red during outburst: r − i ≈ 0.6 mag, i − z ≈ 1.1 mag, J − H ≈ 1.9 mag, and H − K_s ≈ 1.9 mag. These colors are within the distribution but toward the blue end of NIR colors exhibited by Class-I-type stars in Taurus, and significantly redder than can be explained by a range of disk accretion, inner hole, and inclination properties along the Class II "Classical T Tauri locus" (Meyer et al. 1997). In 2010 July, PTF10nvg appeared redder in H − K_s and bluer in J − H than most known "FU Ori-like" objects (e.g., see Figure 4 by Greene et al. 2008). As PTF10nvg began fading in 2010 September, however, its observed JHK_s colors became somewhat bluer, comparable to the colors of V346 Nor, a previously known "FU Ori-like" object. The JHK_s colors of PTF10nvg are also similar to those of V1647 in 2004 February–March, early in the 2004–2005 outburst.

Zoom In Zoom Out Reset image size

High-energy radiation has been detected from other outbursting young stars (e.g., V1647; Kastner et al. 2006), but no X-ray detection can be confidently associated with PTF10nvg at this time. X-ray observations of PTF10nvg were obtained with the Swift X-Ray Telescope (XRT; Burrows et al. 2005) starting on 2010 August 29 for a total exposure time of 3.6 ks. We extract the 0.5–8.0 keV counts from an extraction region of 64 pixels, ∼2.5 arcmin, where we fit the point spread function model (see Butler & Kocevski 2007) at the centroid of the optical position of PTF10nvg. We detect no X-ray emission from PTF10nvg. Assuming a Γ = 2 power-law spectrum and a column density of N_H = 10²² cm⁻² we find a 3σ upper limit of the flux: F_X < 1.2 × 10⁻¹³ erg cm⁻² s⁻¹. Consistent with both the sensitivity of the ROSAT All Sky Survey (Voges et al. 1999) and the Swift upper limit, the closest ROSAT source is approximately 09 SW of PTF10nvg's optical position, suggesting that it is unlikely to be a counterpart to the star.

PTF10nvg has strong H i emission lines including Hβ, Hα, the Paschen series (detectable up to 21), and the Brackett series including Br-α at 4.05 μm. No Balmer jump is apparent, but the pileup of the higher-order Brackett lines is visible in the H band of the SpeX observations. The high-dispersion NIRSPEC spectrum reveals profile widths in the Pa-γ and Pa-δ H i lines of ∼125 km s⁻¹ FWHM. The lines are slightly asymmetric with respect to their peak emission at roughly zero velocity, having somewhat more integrated flux on the blue side of line center than on the red side. The Hα line EWs are −26 to −46 Å among the four optical spectra, moderate but not extreme relative to Taurus Class I or Class II objects (White & Hillenbrand 2004).

We have derived an estimate of the extinction towards PTF10nvg from the strength of its Pa δ and Br γ emission lines. These lines, representing the 7–3 and 7–4 transitions respectively, share the same upper level. Neglecting collisional effects (a questionable assumption that we will return to below), the relative frequency of radiative de-excitations from the same upper level should be determined entirely by atomic physics. That is, the intrinsic flux ratio of these lines can be calculated in the collisionless limit as

$$\begin{equation} \frac{F_{Br \gamma }}{F_{Pa \delta }} = \frac{A_{Br \gamma } \lambda _{Pa \delta }}{A_{Pa \delta } \lambda _{Br \gamma }}, \end{equation} \tag{ 1 }$$

where F, A, and λ give the flux, Einstein coefficient, and wavelength of each line. Using line constants from the NIST atomic database, we calculate an intrinsic Br γ-to-Pa δ line ratio of 0.42 for a collisionless emission region. Densities in magnetospheric accretion columns, however, are sufficiently high that collisions are likely to be an important factor for the emergent fluxes of H i emission lines. Detailed radiative transfer models of relatively cool (T_max= 6000 K) magnetospheric accretion columns nonetheless return a Br γ-to-Pa δ line ratio very close to this naive calculation. Models of significantly warmer (T_max= 9000 K) accretion columns return a larger line ratio of 2.21 (R. Kurosawa 2010, private communication). The line strength ratio in the SpeX spectrum of PTF10nvg is ∼8.2, demonstrating the presence of significant extinction relative to either accretion model. We therefore estimate the extinction to the H i line emission region (which may or may not be coincident with the stellar photosphere) by assuming a standard R_V= 3.1 extinction law, and de-reddening the observed line fluxes to match each of these predicted intrinsic line ratios. These estimates place crude bounds on the extinction to PTF10nvg's emission line region of 6 <A_V< 12.4.

Methods have also been developed to estimate a young star's mass accretion rate from the strength of its NIR H i emission lines (e.g., Muzerolle et al. 1998). These relationships, which link observed H i line luminosities with mass accretion rates inferred from UV continuum excesses, have only been calibrated for observations of T Tauri stars; applying these relationships to more heavily embedded protostars requires the assumption that the dynamics of the underlying accretion process are similar across these evolutionary phases, and that the line luminosity can be measured accurately even when the protostar's photosphere is not detectable. These assumptions are even more problematic for an extreme accretor such as PTF10nvg, but an accretion rate estimate via H i line fluxes can still provide a useful lower limit on the object's accretion rate: changes in the accretion dynamics (i.e., boundary-layer accretion instead of magnetospheric accretion) or the presence of significant veiling flux will both likely result in an underestimate the object's full line flux, and/or an increase in the conversion factor between line luminosities and accretion rates.

We have generated crude mass accretion rate estimates for PTF10nvg from the strengths of its Br-γ and Pa-β emission lines. We have dereddened the Pa-β and Br-γ line fluxes listed in Table 6 assuming an A_V ≈ 9 mag (intermediate to the bounds calculated above), and converted those dereddened fluxes into line luminosities assuming a distance to PTF10nvg of ∼600 pc. Using Equations (1) and (2) from Muzerolle et al. (1998) to estimate Log L_acc/L_☉ from each line, we infer Log L_acc/L_☉ = 0.62 and 0.335 from the Pa-β and Br-γ lines, respectively. Assuming canonical Classical T Tauri Star parameters (M = 0.5 M_☉ and R = 3 R_☉), these accretion luminosities imply mass accretion rates of $\dot{M} \approx 2.5 \times 10^{-7}$ M_☉ yr⁻¹.

The Ca ii triplet lines at 8498, 8542, and 8662 Å are also generally associated with accretion activity in young stars. PTF10nvg exhibits EWs for these lines of −15 to −25 Å, varying among the observations, but in the realm of the rapidly accreting young Class II stars in Taurus and comparable to the emission levels measured by Connelley & Greene (2010) in a large survey of Class I objects. The line ratios seen from PTF10nvg varied, being somewhat typical of Class II sources ($F_{\rm \rm Ca\,{\mathsc {ii}} 8498}/F_{\rm \rm Ca\,{\mathsc {ii}} 8542} = 0.91$, $F_{\rm \rm Ca\,{\mathsc {ii}} 8662}/F_{\rm \rm Ca\,{\mathsc {ii}} 8542} = 0.81$) in the LRIS data taken on 2010 July 8, but roughly equal ($F_{\rm \rm Ca\,{\mathsc {ii}} 8498}/F_{\rm \rm Ca\,{\mathsc {ii}} 8542} = 1.08$, $F_{\rm \rm Ca\,{\mathsc {ii}} 8662}/F_{\rm \rm Ca\,{\mathsc {ii}} 8542} = 0.93$) in the SpeX data from July 14. The latter case is again similar to heavily accreting T Tauri stars at the extreme of the Class II range illustrated by Hamann & Persson (1992) and close to the ratio predicted by their highly optically thick (τ = 1000) slab models. It is also similar the outbursting behavior of V1647 Ori (Walter et al. 2004).

Like the Paschen lines, the Ca ii triplet lines appear slightly asymmetric with respect to their line centers, having somewhat more flux on the blueshifted sides of the lines than the red.

The Ca ii K line is also in emission, though the H line is affected by blueshifted Balmer line absorption.

The only absorption features we detect in these optical spectra are blueshifted lines of H i, Na i D, and K i, along with the O i triplet at 7773 Å, which appears centered at the systemic velocity. The IR spectra, similarly, have only a single blueshifted line in absorption (the He i λ 10830 feature seen in Figure 8 and discussed in Section 3.2.4), so no spectral type can be derived for the outbursting object. Instead of a stellar photosphere, the spectrum is dominated by the circumstellar signatures of accretion and outflow processes.

In addition to the H i and Ca ii lines discussed in Sections 3.2.1 and 3.2.2, the optical and NIR spectra of PTF10nvg show many emission lines characteristic of young accreting stars. The optical spectra show several permitted and forbidden species, including O i, Fe i, Fe ii, Ca i, Ca ii, [Fe ii], [Ca ii], and [S ii]. The high-dispersion Keck NIRSPEC Y-band spectrum shows low-excitation (<1–2 eV) Ti i lines as well as higher excitation (6–8 eV) Si i and C i lines in abundance, identified from the line list presented by Sharon et al. (2010). A Y-band survey including many Class II and some Class I stars in Taurus, as well as other young stars (Edwards et al. 2006; Fischer et al. 2008), suggests that some extreme emission line objects display the higher excitation lines. The Ti i is quite rare, however, and is found in our experience only in the spectra of V1331 Cyg, ZCMa, and SVS 13—all embedded sources driving strong outflows. The lower dispersion SpeX spectrum shows emission lines from many neutral species including Al i, Ca i, Fe i, K i, Mg i, Na i, O i, Si i, and Ti i over the 1–2.5 μm region; many of these features are also visible in the NIR outburst spectrum of V1647 (shown for comparison in Figure 8).

There is significant evidence for a substantial wind associated with the outburst of PTF10nvg. Our optical spectra have limited resolution of ∼150–300 km s⁻¹, but nevertheless the Hα and Hβ lines display clear P Cygni features with sub–continuum absorption out to −400 km s⁻¹ (see kinematic profiles shown in Figure 10). The upper Balmer series lines lack the redshifted emission and display only a broad, blueshifted absorption trough. In the NIR, where our NIRSPEC spectrum provides a higher resolution by a factor of 2–3, none of the Paschen or Brackett lines has P Cygni structure. The Na i D lines, like the upper Balmer lines, are seen in blueshifted absorption out to about −400 km s⁻¹ but are blended at our resolution.

Zoom In Zoom Out Reset image size

The K i doublet at 7664 and 7698 Å is seen in absorption in PTF10nvg's spectrum, blueshifted by ∼175 km s⁻¹; this line is not typically seen in the spectra of Class I or Class II stars. Strong and blueshifted K i and Na i D absorption, as well as P Cygni Balmer profiles, are seen in the optical spectra of the FU Ori stars V1057 Cyg and V1515 Cyg, however, and in the outburst spectrum of V1647 Ori.

The high-dispersion NIRSPEC spectrum shows a somewhat complex He i λ10830 profile (see Figure 11). Not unlike many heavily accreting T Tauri stars, such as DR Tau (Edwards et al. 2003, 2006), there is strong blueshifted absorption out to about −300 km s⁻¹ which reaches 60% of the continuum level. The absorption does not extend continuously inward to zero velocity, however; instead, the continuum level is reached by about −100 km s⁻¹ and there is a second narrow (FWHM ≈17 km s⁻¹) absorption component centered at PTF10nvg's rest velocity. Similar low-velocity and narrow absorption components are seen in the weakly or nonaccreting T Tauri stars V836 Tau and TWA 2. When considering both the broad absorption terminal velocity and the narrow absorption depth, the He i λ10830 profile of PTF10nvg is quite similar to that of V1057 Cyg (unpublished data). The origin of a narrow absorption feature at roughly rest relative to the star is unknown; we note, however, the similar feature of V836 Tau was attributed to a disk wind by Kwan et al. (2007). Also of note is the possible presence of Si i emission at 10830 Å that could be filling in some of the blueshifted He i absorption at intermediate velocities; this can be gauged by comparison to the Si i λ10869 line also shown in Figure 11.

Zoom In Zoom Out Reset image size

Other typical indicators of mass loss (Hartigan et al. 1995) are the [O i] λλ6300, 6364 emission lines, which in PTF10nvg are symmetric and blueshifted by ∼175 km s⁻¹, and the [S ii] λλ6717, 6731 lines, also blueshifted by ∼175 km s⁻¹. The [N ii] λλ6548, 6584 lines are relatively weak and [O i] λ5577 is not apparent in our low spectral resolution data. The forbidden-line EWs are comparable to those seen in the some of the strongest Class II stars in Taurus such as CW Tau, DD Tau, DG Tau, HN Tau, and UZ TauE (Hartigan et al. 1995), but weaker than those typical of Class I and other stars in Taurus driving strong outflows (White & Hillenbrand 2004).

Hartigan et al. (1995) used detailed shock models to derive relations allowing a young star's mass loss rate to be estimated from the luminosities of its [O i] 6300 and [S ii] 6731 forbidden lines. These relations are based upon assumptions about the typical density, velocity, and abundance of the shock region, and also require that observed line fluxes be converted to line luminosities, with corrections to account for reddening. Given the strong outflow signatures PTF10nvg possesses, it is possible, and perhaps likely, that the parameters commonly assumed for T Tauri star jets are not a good description of the physical properties of shocked regions in PTF10nvg's outflow. Nonetheless, we can convert the observed [O i] 6300 and [S ii] 6716 line fluxes to luminosities by adopting the same distance and extinction estimates used above to calculate H i line luminosities, and thus infer crude estimates of PTF10nvg's mass loss rate from Hartigan, Edwards & Ghandour's (1995) equations A8 and A10. These calculations imply estimated mass loss rates of 7 × 10⁻⁷ M_☉ yr⁻¹ (as inferred from [S ii] 6716) and 2 × 10⁻⁶ M_☉ yr⁻¹ (as inferred from [O i] 6300). Given the sizable uncertainties in the adopted extinction correction, and even the appropriateness of the physical properties assumed for the shock regions, these mass loss rates may easily be incorrect by more than an order of magnitude. Nonetheless, they do emphasize the considerable mass loss associated with the PTF10nvg outburst; relative to the ∼10⁻⁷ M_☉ yr⁻¹ accretion rate inferred above from the H i emission lines, these estimates suggest an inflow/outflow ratio ≳1, one to two orders of magnitude larger than the 0.01–0.1 ratios typically inferred for T Tauri stars.

NIR line ratios provide additional evidence that PTF10nvg is driving a strong outflow. Lorenzetti et al. (2009) and Antoniucci et al. (2008) have noted that EXor/jet driving sources typically demonstrate flux ratios between their Br-γ and 2.206 μm Na i emission lines of ∼2–5. Class I sources that lack jet signatures, however, tend to have ratios that favor Br-γ much more strongly (e.g., 6–20; Nisini et al. 2005). The NIR spectra of PTF10nvg presented here possess a ratio of these lines of ∼2, placing PTF10nvg firmly in the company of the jet driving sources, as is consistent with the morphology of the wind-sensitive lines analyzed above.

PTF10nvg shows clear emission in the 2.12 μm H₂(1 − 0)S(1) line, as well as a weaker feature which is likely the 2.24 μm S(2 − 1)S(1) line. Modeling has established that the ratio of these line strengths can diagnose between collisional and radiative excitation mechanisms: pioneering work by Gredel & Dalgarno (1995) demonstrated that UV and X-ray irradiation will produce distinctive H₂(1 − 0)S(1)/S(2 − 1)S(1) line ratios (0.54 and 0.06, respectively), and could also be distinguished from the ratio expected from a T = 2000 K shocked gas (0.13). More recently, Nomura et al. (2007) calculated synthetic H₂ spectra as expected to emerge from protoplanetary disks with realistic gas-to-dust ratios and density/temperature profiles. The Nomura et al. (2007) models predict that H₂ excitation by UV radiation is sensitive to the dust properties in the disk, producing H₂(1 − 0)S(1)/S(2 − 1)S(1) line ratios from 0.025–0.23 for power-law distributions of dust grains with maximum sizes of 10 μm to 10 cm, respectively. X-rays are much less sensitive to dust properties; models where X-rays dominate the excitation produce ratios of ∼0.07, irrespective of the dust properties.

The H₂(1 − 0)S(1) and S(2 − 1)S(1) line strengths recorded in Table 6 correspond to ratios of 0.23 and 0.3 for the SpeX and TSPEC observations; the less secure measurement of the (1-0)S(1) line suggest this ratio may best be treated as a tenuous detection which could be uncertain at the factor of two level. Even accounting for this relatively large uncertainty, models attributing the excitation to X-rays alone are essentially ruled out; models incorporating UV or UV + X-ray heating and dust grains ≲1 mm would also be difficult to reconcile with this ratio. The observed values are most consistent with, though slightly larger than, the Gredel & Dalgarno (1995) model of shock excitation or the Nomura et al. (2007) model of UV + X-ray irradiation of a disk with dust subject to coagulation and settling, which predict H₂(1 − 0)S(1) and S(2 − 1)S(1) line ratios of 0.13 and ∼0.18, respectively. Given the wealth of outflow signatures present in PTF10nvg's spectrum, we consider shocks to be the most likely source, but this conclusion is far from established.

[Fe ii] is another common outflow tracer for young stars: Connelley & Greene (2010) demonstrate that the line strengths of H₂ and [Fe ii] are strongly correlated for a large sample of Class I protostars. The TripleSpec observation of PTF10nvg shows clear [Fe ii] λ1.257 and λ1.644 μm emission; these lines may also be present in the SpeX observation, but the lower spectral resolution leads to a lower line-continuum contrast. In comparison to the Class I stars in the Connelley & Greene (2010) sample, PTF10nvg shows relatively weak H₂ and [Fe ii] emission. Connelley & Greene (2010) note, however, that the stars in their sample with weak H₂ also have low levels of continuum veiling, which is not true for PTF10nvg, where we cannot detect any photospheric absorption features.

As both the 1.257 and 1.644 μm lines share the same upper level, Connelley & Greene (2010) use their observed line fluxes to infer extinction estimates for each star in their sample, as with the H i-based estimates presented in Section 3.2.1. We find a line ratio in PTF10nvg of 1.48, essentially equal to that assumed by Connelley & Greene (2010) for an unextinguished source, in strong disagreement with the extinction estimate of 6–12 which we infer from the H i lines of PTF10nvg. Connelley & Greene (2010) find a similar discrepancy, however, in their analysis of the Class I source 06297+1021W, whose H i lines imply A_V ≈ 18 mag but whose [Fe ii] lines suggest A_V ≈ −1 mag. In many regards, 06297+1021W is quite similar to PTF10nvg: in addition to their similarly incongruous H i and [Fe ii] extinction estimates, both show weak H₂ and [Fe ii] emission relative to other Class I sources (Connelley & Greene 2010 are unable to measure values for these features for 06297+1021W), both objects appear as a K-band point source absent significant nebulosity, and both display a rich, morphologically similar NIR emission-line spectrum.

Connelley & Greene (2010) also explore the correlation between the strengths of the H₂, [Fe ii], and Br-γ emission lines: relative to their sample of Class I objects, PTF10nvg lies on the weak side of these relations, though not significantly separated from the rest of the Class I population.

In addition to the rich atomic emission-line spectrum, the most salient feature of the optical spectrum of PTF10nvg is the broad emission that we attribute to TiO and VO. The PTF10nvg spectrum (Figure 7) looks much like an inverted mid-M spectral type star, suggesting that in addition to the hot and moderate density emission lines, there is a cooler (∼1500–4000 K, Lodders 2002; Ferguson et al. 2005; Sharp & Burrows 2007) and high-density (>10¹⁰ cm⁻³) emitting component, perhaps the outer layers of a disk. Notably, the TiO emission is variable among our three spectra, being strongest just before the object brightened for the second time during the 2010 season.

To our knowledge, this is the first time such prominent hot molecular emission has been reported for a young star. There is, however, literature discussion of emission in the optical TiO band heads, though not the broader full bands as we observe in PTF10nvg. TiO band head emission has been mentioned in the spectra of several luminous B[e] stars (Zickgraf et al. 1989) and several red giant or supergiant dusty objects, notably VY CMa (Phillips & Davis 1987) and U Equ (Barnbaum et al. 1996), as well as in the red nova V4332 Sgr (Goranskii & Barsukova 2007). Most relevant to the present case is the report by Herbig (2009) that V1057 Cyg exhibited emission in 2004 in "the heads of the (0-0) bands of the γ system of TiO at 7054, 7087, and 7125 Å." We have confirmed this narrow emission in our own data on V1057 Cyg but emphasize that the PTF10nvg TiO emission is through the full band region, not just in the band heads.

In addition to the optical TiO and VO emission, there is also molecular emission throughout the 1–5 μm spectra of PTF10nvg. Readily apparent broad features above the likely dust-dominated red continuum are VO at 1.05 μm, several H₂0 bands around 1.4, 1.8, 2.0, and 2.45 μm as well as the Q and R branches at 3.0 and 3.3 μm, and prominent CO first overtone band heads beginning at 2.3 μm. While the VO and H₂O suggest emitting region temperatures of <3500 K, the CO emission could come from warmer (2500–5000 K) dense (>10¹⁰ cm⁻³) gas.

Connelley & Greene (2010) display correlations between the morphology and strengths of the CO, Na, Ca, and Br-γ features for Class I sources, finding that all sources with CO in emission also display Br-γ and Na in emission. PTF10nvg follows this relation and displays CO, Na, and Ca emission-line strengths similar to those of other Class I sources with each set of features in emission. PTF10nvg does show somewhat weaker Br-γ relative to its CO emission strength than most Class I sources, but is not an extreme outlier to that relation.

In contrast to the CO band head emission at 2.3 μm, the 4.7 μm CO fundamental band is in absorption with both the P and R branches evident.