A DETAILED PHOTOMETRIC AND SPECTROSCOPIC STUDY OF THE 2011 OUTBURST OF THE RECURRENT NOVA T Pyxidis FROM 0.8 TO 250 DAYS AFTER DISCOVERY

The 1.5 m telescope of the SMARTS II Consortium located at CTIO Chile is equipped with a long-slit Ritchey-Chrétien spectrograph. An ultraviolet transmitting grating is used at the f/7.5 focus with a plate scale 18.1 arcsec mm⁻¹ and a 1200 × 800 CCD. We obtained 99 low- to moderate-resolution (300 < R < 3400) optical spectra of T Pyx. These spectra include 49 moderate-resolution spectra in the blue region (3655–5424 Å and 3870–4544 Å), 37 moderate-resolution spectra in the red region (5630–6950 Å), and 13 low-resolution spectra in a wide wavelength region (3250–9520 Å). We obtained 56 spectra of particular interest to us from 2011 April 14 (JD 2455666.55056, t = 0.757 days, given hereafter 0.8 days), which was the earliest epoch T Pyx was observed spectroscopically, to 2011 July 3 (JD 2455746.49198, t = 80 days). Another 43 spectra were obtained from 2011 September 15 (JD 2455820.92030, t = 155.127) to 2011 December 19 (JD 2455915.72555, t = 249.9 days). All data are available through the SMARTS atlas; see Walter et al. (2012).⁴

The 2 m robotic LT (see Steele et al. 2004) is at the Observatorio del Roque de Los Muchachos on the Canary Island of La Palma, Spain. T Pyx's spectra on the nights of 2011 April 22–25 were secured using the Fibre-fed RObotic Dual-beam Optical Spectrograph (see Barnsley et al. 2012) on the LT over 3900–5700 Å in the blue (R ∼ 2200) and 5800–9400 Å in the red arm (R ∼ 2600), with exposures of 60 s. Data reduction was performed through a pipeline that initially performs bias, dark frame, and flat-field subtraction. A spectroscopic standard star HD289002 (α = 06^h45^m13371, δ = +02°08'1470) was observed at a similar airmass and used to remove instrumental and atmospheric response. We used the onedspec package in IRAF⁵ to analyze all LT spectra.

We divide the SMEI light curve into four parts based on the idealized nova optical light curve given in Warner (2008); see Figure 3.

• 1.  
  The initial rise (2011 April 15–16, t = 1.5–3.3 days). The first reliable detection of the nova outburst by SMEI occurred at the end of the rapid rise seen in AAVSO data, at m_SMEI = 8.80 ± 0.09 on 2011 April 15.84 UT (JD 2455667.3422, t = 1.5 days). After that time, SMEI observed T Pyx approximately every 102 minutes. The light curve rose to m_SMEI = 8.26 on JD 2455668.0476 (t = 2.2 days) and began quasi-periodic variations (Hounsell et al. 2011) as shown in the top left panel of Figure 3. This phase ended at t ∼ 3.3 days when the light curve began to flatten.
• 2.  
  Pre-maximum halt (2011 April 16–27, t = 3.3–13.3 days). Following the initial rise, the SMEI light curve rose very slowly (noted as "almost plateau" by Hounsell et al. 2011) with clear quasi-periodic variations as shown in Figure 3(b). The peak-to-peak times of the variations vary between 1.1–2.8 days. The magnitudes ranged between 7.7–8.5 mag. The first peak and dip of this variation was also noted by Schaefer et al. (2013). However, the AAVSO light curve did not clearly show the quasi-periodic variations detectable in the SMEI light curve. This phase may coincide with the pre-maximum halt defined as a pause at about 2 mag below maximum for a few days for slow novae (Warner 2008; Hounsell et al. 2010). There was then a gap in the SMEI data (due to instrumental problems) lasting 1.4 days (t = 13.3–14.7 days).
• 3.  
  Final rise (2011 April 28 to May 11, t = 14.7–27.9 days). After the pre-maximum halt phase, the light curve was seen to rise more steeply to m_SMEI = 6.88 ± 0.04 at t = 20.9 days. Meanwhile, the apparent quasi-periodic variations still persisted with approximately half the amplitude of those seen in the previous phase.The light curve then suddenly dropped at t = 22.1 days to m_SMEI ∼ 7.4 for about two days. This event appeared as a major dip in the light curve before it reached maximum light. After this dip, the light curve rose again to m_SMEI = 6.79 ± 0.04 at t = 25.5 days and stayed there for roughly two days before it reached visual maximum at m_SMEI = 6.33 ± 0.03 on 2011 May 12.22 (JD 2455693.7251, t = 27.9 days) as seen in the bottom left panel of Figure 3.
• 4.  
  Early decline (2011 May 11 to [∼October 3], t = 27.9 days to [∼90 days]). After optical maximum, the nova declined with variations (amplitude ranges roughly between 0.1–0.3 mag) to m_SMEI ∼ 7.2–7.4 from 2011 May 19.42–20.76 UT (JD 2455700.9203–2455702.2605, t = 35.1–36.5 days). The light curve experienced a dip again at around t ∼ 36 days and followed this with a broad hump lasting from t ∼ 44–47 days as seen in Figure 3(d). The last reliable detection of T Pyx by SMEI was at m_SMEI = 7.26 ± 0.05 on 2011 June 2.37 UT (JD 2455714.8165, t = 49.02 days).Further light curve observations provided by the AAVSO in Figure 1 show that the early decline phase ends at around t ∼ 90 days with the brightness having declined by approximately 2.5–3.5 magnitudes from maximum. The final decline (∼6 magnitudes from maximum) began around t ∼ 260 days.

We may compare the early stages of the outburst captured by SMEI to the light curves presented by Hillman et al. (2014) from TNR modeling with a range of parameters. There is certainly some resemblance to some of these model light curves in terms of the pre-maximum halt and subsequent smaller reversal just before optical peak. The former is attributed to a temporary drop in energy flux as convection in the expanding, thinning envelope ceases to be efficient near the surface of the envelope. The later dip resembles the shorter timescale feature noticed in, e.g., RS Oph and KT Eri by Hounsell et al. (2010) just before maximum light.

When directly comparing the SMEI to the AAVSO light curves, we see that the nova seems to be bluer and similar to the magnitude in the B filter at the pre-maximum halt phase, then exhibits the same brightness in the B and V filters at visual maximum where it is expected to behave like an early type star, usually in the range from B5–F5 (Warner 2008). After visual maximum, it tends to be redder, approaching the magnitude in the R filter as shown in Figure 2.

When comparing the SMEI data to the previous five outbursts in B and V and their resulting eruption templates provided in Schaefer (2010), we find that the shape and the magnitude of the visual peak in 2011 are compatible with previous outbursts. However, the pre-maximum halt phase in 2011 seems to be fainter, while during the decline, after the visual maximum, the nova is brighter in 2011 than shown in the templates (Figure 4). All the maxima, including the 2011 maximum, have exhibited brightness fluctuations near the optical peak and the declines have been slow. This is consistent with observations noted in Catchpole (1969) and Williams (1982) for the 1966 outburst.

Zoom In Zoom Out Reset image size

The SMEI light curve was searched for any periodic modulations. The analysis was undertaken using all the available SMEI data, but we separated it into four cases: (a) from the first observation to the last observation, (b) from the first observation to visual maximum, (c) from visual maximum to the last observation, and (d) from the first observation to the end of the pre-maximum halt phase. The analysis used the PERIOD04⁷ PC code from Lenz & Breger (2004, 2005). PERIOD04 is especially useful for analysis of astronomical time series containing gaps. The program extracts the individual frequencies from the multi-periodic content and provides the frequencies, semi-amplitude, and phase of the harmonic signals of the light curve using a combination of least-squares fitting and the discrete Fourier transform algorithm. The uncertainty of the estimated periods was derived from Monte Carlo simulations.

The resulting periodograms for all five cases are displayed in Figures 5–9. The top panels show how closely the calculations agree with the observations, and the most prominent peaks are presented in Table 2. A prominent period of 1.8 days is apparent after the pre-maximum halt. However, by far the most strongly detected period, P = 1.44 ± 0.05 days, with the highest signal ratio ∼10⁶ as shown in Figure 6, is found up to this time. This period is close to the weak signal of 1.24 days found by Patterson et al. (1998) who suggested it might originate from precession in an accretion disk. However, it is doubtful whether the disk would re-form so soon after the outburst, and at this early time, the central system is expected to lie well within the pseudo-photosphere (see below) and therefore not be directly observed.

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

Zoom In Zoom Out Reset image size

Some of the TNR models by Hillman et al. (2014) produced marked oscillations prior to or during extensive mass loss, caused by the restructuring and rebalancing of the envelope as it expands. However, we found no significant changes in the Hα line shape with the 1.44 day period when we compared the four red spectra (t = 5.7, 8.6, 9.6, and 11.6 days) taken during the optical pre-maximum halt. Similarly, we found no significant differences in the overall ionization during a cycle from consideration of the appearance of lines across the full spectral range. These two aspects are also found to be the same in the blue spectra, where the Hγ line was of course investigated rather than Hα.

The ejection velocities (V_ej) were measured from the P Cygni profiles. The measured lines include Hα, Hβ, and Hγ, which have measurable P Cygni profiles from 2011 April 16 to June 1 (t = 2.7–48.6 days); Fe ii recombination lines at 5169, 5018, 4233, 4178, and 4173 Å; O i 7775 Å; and the Ca ii K line at 3934 Å. Figure 11 shows the P Cygni profiles of Hα lines from the initial rise to the end of the pre-maximum halt phase.

Zoom In Zoom Out Reset image size

Figure 12 shows all low-resolution spectra taken from t = 1.7–247 days, which cover the initial rise phase through the transition phase of the light curve. We discuss the spectral evolution based on the idealized nova optical light curve given in Warner (2008) together with a recognition of a common pattern of line development described by McLaughlin (1942, 1944). We also note the physical interpretation of various stages. For example, from the earliest moment, the nuclear explosion on the surface of the WD leads to the ejection of a hot, luminous, and massive shell that expands radially with time. Ney & Hatfield (1978) called this stage the "pseudo-photospheric expansion" for the spectral energy distribution (SED) and spectroscopic features are characteristic of the photosphere of a star with spectral type A to F (Payne-Gaposchkin 1957). Gehrz (1988) called this stage the "fireball" because it has been used to describe the early development of man-made atomic explosions and therefore he used it to describe the expanding pseudo-photosphere of the nova.

Zoom In Zoom Out Reset image size

The fireball's envelope is initially small and dense with the radiation peak at X-ray wavelengths. Then, the envelope cools adiabatically as it increases in size. As a result, the opacity increases, the ejecta become optically thick after the outburst, and the radiation peak shifts toward longer wavelengths (Shore et al. 1994). Therefore, this cooling of the ejecta envelope together with the rapid expansion of the photosphere (Ennis et al. 1977; Gehrz et al. 1980) provide the rapid initial rise in the optical light curve.

Following optical maximum, the effective temperature (T_eff) of a nova, which is radiating at constant bolometric luminosity, rises as the pseudo-photosphere shrinks back onto the WD and as the mass loss rate from the WD decreases (Bath & Harkness 1989). As a result, the peak emission now shifts toward shorter wavelengths. The effective temperature of the pseudo-photosphere changes as the visual flux declines according to

$$\begin{equation} T_{{\rm eff}} = T_{0}\cdot 10^{\Delta {V}/2.5}, \end{equation} \tag{ 1 }$$

where ΔV is the decline in magnitude from visual maximum, and T₀ is the photospheric temperature at optical peak (Bath & Harkness 1989). We take T₀ = 8000 K (Evans et al. 2005) and use the equation above to estimate T_eff. According to Bath & Harkness (1989), the physical state of the photosphere (i.e., temperature, radius, density, and pressure) is approximately the same at equal magnitudes below the optical maximum in all novae. Consequently, one might anticipate that the spectral appearance of a nova should be the same at the same magnitude below peak in all novae if this simple model is correct. It can then be applied before and after peak if the mass loss rate from the WD is the governing factor.

• 1.  
  The initial rise (2011 April 14–16, t = 0.8–3.3 days). Accordingly, the spectral evolution is divided into the following four stages. The first red spectra were obtained on 2011 April 15.05 UT (JD 2455666.55056, t = 0.8 days), about 27 days before the visual maximum. At this time, the AAVSO visual magnitude was V ∼ 8.7 (about 6.8 mag brighter than quiescence at V ∼ 15.5 and about 2.4 mag below visual maximum; see Figure 1). The top panel of Figure 13 reveals that the lines were broad and diffuse. Hα and Hβ emissions were strong while the P Cygni profiles were not yet seen clearly when compared to other later spectra, shown in Figure 11. Moreover, the spectrum at t = 1.7 days in Figure 12 shows that the brightness of the nova comes almost entirely from the continuum at this early stage of the outburst. This is consistent with the characteristics of the pre-maximum spectrum stage given by Payne-Gaposchkin (1957). This is expected as the expanding pseudo-photosphere is of similar extent to the maximum radius initially reached by the ejecta.The pre-maximum spectrum is defined as the earliest spectrum at which any given nova has been observed on the rise until at least one or two days after maximum light (McLaughlin 1942). It usually contains lines that are broad and diffuse with negative velocity displacement (Payne-Gaposchkin 1957).Our spectra showed P Cygni profiles in the Balmer, Fe ii, and O i lines quite early in the first spectra, during this fireball phase. This aspect is similar to that observed at this time in DQ Her and LMC 91 as mentioned in Schwarz et al. (2001). The absorption components of P Cygni profiles became broader and shallower from t = 0.8 to 2.7 days.The initial rise ended just after 2011 April 16.95 UT (JD 2455668.44853, t = 2.7 days), where the first medium-resolution blue spectra were obtained. At t = 0.8 days, the H i Balmer lines were present in emission and becoming stronger with blueshifted absorptions and the presence of He i. By t = 2.7 days, the rise of ionized iron emission lines was evident. There were emissions of high-excitation lines (see Table 4), i.e., C iii, N iii, Ne ii, O ii, N ii, He i, and Ne i, present during this phase (see Figure 13) which disappeared later around 2011 April 22–23 (t = 8.6–9.7 days). These O ii and N ii lines are also expected to be the ionized elements that should be found at ∼1.5 mag below peak (Bath & Harkness 1989); see Table 3. Another expected emission line at 2.9 mag below peak is O iii at 4435 Å, which is also seen at t = 1.7–2.7 days.A marked drop in the derived expansion velocity (V_ej) during this phase is also noted (see Figure 10). We note that Imamura & Tanabe (2012) and Shore et al. (2011) also report a decline in derived expansion velocities at early times. The first observations show a derived expansion velocity of ∼4000 km s⁻¹ at t = 0.8 days, which then drops to ∼2000 km s⁻¹ at t = 2.7 days for Balmer lines. This should not be interpreted as a deceleration. This aspect of the pre-maximum spectrum with the dramatic decrease in V_ej during the initial rise is also found in the slow nova DQ Her (McLaughlin 1937) for Balmer lines. Meanwhile, the fast nova V603 Aql (Wyse 1940) also showed a decrease in V_ej for Balmer and metal lines from a few observations during the final rise.If the initial ejection is a Hubble flow, then one will a see high V_ej initially, which declines as one sees into deeper layers. We also note, as an aside, a comparison to the high-velocity features found in all Type Ia SNe and believed to be the result of the interaction of initial highest velocity ejecta with a circumstellar envelope (Benetti et al. 2005). The subsequent change in behavior of the derived V_ej during the pre-maximum halt phase suggests two different stages of mass loss: a short-lived phase first occurring immediately after outburst, followed by a more steadily evolving and higher mass loss phase.
• 2.  
  The pre-maximum halt (2011 April 17–27, t = 3.6–13.7 days). The SMEI light curve at this phase has m_SMEI ∼ 8 mag (with 0.1–0.3 mag variations), which is about 2 mag below maximum (6.33 mag). This phase lasts for about 10 days. As Warner (2008) notes, the pre-maximum phase is generally much longer lasting in slow novae than in fast novae.All of the C iii lines disappeared by t = 3.6 days, while other lower ionization lines such as N iii, Ne ii, N ii, and Ne i lasted a little longer until disappearing approximately around t = 5.7–10.7 days as shown in Table 4, consistent with behavior reported in Shore et al. (2011). The spectra at t = 3.6–5.7 days show broad and strong emission lines present between λ ∼ 6480–6530 Å and then these fade away. Lines that were present all the time during this phase are O i, O ii, He i, and Fe i, which are again consistent with the behavior proposed by Bath & Harkness (1989) at this magnitude below peak. We note that He i 5875 Å and 6678 Å were present from t = 0.8 days then became weaker and totally disappear by t = 14.8 days. They appeared again at t = 80.7 days.The end of the fireball stage occurs around t = 8.6 days which is also the beginning of the next optically thick phase called the "iron curtain." The iron curtain (Hauschildt et al. 1992; Schwarz et al. 2001) occurs when the pseudo-photosphere reaches its minimum temperature (∼10⁴ K). The iron-peak elements recombine at this temperature. The overlapping of absorptions of Fe ii lines in the near UV region results in line blanketing in the UV which dominates the UV SED and redistributes most of the emitted light into the optical and IR.At this stage, we found three characteristics of the iron curtain, which are mentioned in Schwarz et al. (2001). First is the increase in the width of the emission lines, which is clearly occurring in the final rise (see Figure 10); second is the increase in the derived expansion velocity which Schwarz et al. (2001) suggested could be due to the gradually accelerating optically thick wind that is proposed to begin after the initial outburst (Kovetz 1998); third is the increase in the intensity of Fe ii multiplets and O i 7775 Å and 8447 Å. Figure 14 shows that the low-excitation lines such as Fe–peak transitions, particularly Fe ii recombination lines at 5169, 5018, 4233, 4178, and 4173 Å, begin to rise from t = 8.6 days.There is also a tendency for the absorption components, especially those of the Fe ii lines, to become sharper and stronger as maximum is approached. We note that during this phase (for particular ions) the spectrum tends to develop such that the shorter wavelength lines are evident first, followed by the next lines toward the longer wavelength. For example, Fe i 7443 Å develops first, followed by Fe i 7446 Å. This also happens with Fe i 7469–7473 Å, O ii 7895–7898 Å, and Fe i 8468–8471 Å.The variation of the light curve during the pre-maximum halt phase seems to be consistent with variations in the strength of the Hα, Hβ, Hγ, and Fe ii lines during this halt as shown in Figure 10. There is an obvious sharp variation in the light curve during t = 8.6–11.6 days where spectra show that some Fe i lines were present at t = 9.6–10.6 days but not before or after that (bottom panel of Figure 15). For example, the Fe i 7854 Å line was not detected at t = 8.6 days but was present on days 9.6 and 10.6 before disappearing again at t = 11.6 days. Meanwhile, Fe i 8838 Å was present only at t = 10.6 days.Imamura & Tanabe (2012) and Ederoclite (2013) suggested that T Pyx evolved from a He/N to an Fe ii type nova (Williams 1992) by the time it reached visual maximum. Here, we can point out that the exact time that this process began was during this pre-maximum halt phase and it was completed by the final rise. We note, however, that this spectral classification is normally applied to novae from maximum light onward. The evolution of T Pyx from a He/N nova to an Fe ii type nova suggests that there was sufficient mass loss for the optical thickness in the outer layers to increase during the rise to maximum so that we were seeing less deep into the expanding layers. This type of evolution could be typical for many novae in outburst but has not been seen before because of insufficient observations early in the outburst.The maximum V_ej of approximately 2200 km s⁻¹ during this phase derived from the Balmer lines agrees very well with that obtained by Imamura & Tanabe (2012). Although we do not have information from the P Cyg profiles in Fe ii lines until t = 3.6 days to verify the very high V_ej at the same initial rise phase as the Balmer lines, the Fe ii lines seem to show a similar trend subsequently to that of the Balmer lines in that V_ej stabilizes at ∼1500 km s⁻¹ during the pre-maximum halt and tends to increase afterward.
• 3.  
  Final rise (2011 April 28 to May 11, t = 14.7–27.9 days). Here, the light curve rises more steeply toward the maximum. As noted above, the iron curtain stage is expected to peak around this time. However, Figures 16 and 17 shows the Fe ii flux rose to a sharp peak just before t ∼ 23 days when there is a major dip in the SMEI light curve with a minimum at t = 23.7 days. The visual maximum then occurred four days later at t = 27.9 days.The "principal spectrum," which dominates CNe spectra at visual maximum (Payne-Gaposchkin 1957; Warner 2008), displays strong lines of O i (i.e., the "O i flash"). In T Pyx, these became apparent around t ∼ 17 days and grew in intensity at about the same rate as the Fe ii lines, as shown in Figure 18. At the O i flash, the V magnitude was ∼1.5 mag below peak, which agrees with that expected by Bath & Harkness (1989) from the ΔB at this time. Strong bright lines of Fe ii and Ca ii are always present. The emission of [N ii] 5755 Å begins to grow stronger at t = 26.7 days, about ten days after the O i flash, since N lines, i.e., [N ii] 5755 Å, together with emission lines of N iii, N iv, and N v in the UV, are expected to be seen in the principal spectrum phase of novae (Jaschek & Jaschek 2009).⁸Again the Balmer lines and other lines show a similar trend of V_ej, i.e., gradually increasing (after the initial decrease at early times). This may imply that here the innermost layers of material now move faster than the outer layers (i.e., the relative radius of the pseudo-photosphere to that of the ejecta is shrinking significantly and therefore revealing higher velocity material again). As noted above, this was proposed by Schwarz et al. (2001) in terms of a gradually accelerating wind in nova LMC 1991 and has been used to model early hard X-ray emission in some CNe (e.g., O'Brien et al. 1994). The visual maximum at t = 27.9 days seems to exhibit the lowest ionization lines, again as predicted in the simple Bath & Harkness (1989) models.The three characteristics of the iron curtain, which are first mentioned in the pre-maximum halt phase, persist in this final rise phase. Returning to the major dip in the light curve at t = 23.7 days, we find that the normalized flux of the blue spectra with respect to the Hα line on the day before, at, and after the dip (at t = 21.7, 23.7, and 25.7 days, respectively) all looked exactly the same in shape, and the only difference was in the Hα flux, which dropped very significantly at the dip. Thus, the mechanism that causes the very marked drop in the emission line strength of Hα, and other lines at longer wavelengths, must be responsible for the major dip in the total flux in the final rise. Meanwhile, there are almost no other emission lines apart from Hα in the red spectra at t = 24.7 days.
• 4.  
  Early decline (2011 May 11 to October 3, t = 27.9–90 days).The early decline phase of the CN light curve is defined as the beginning of the decrease in brightness to ∼3.5 magnitudes below peak (Warner 2008). The SMEI light curve declines in brightness from m_SMEI = 6.5 (t = 27.9 days) to m_SMEI = 7.26 at the last SMEI detection (t = 49.7 days). The multi-color light curves from AAVSO subsequently show sharp drops in all colors at t ∼ 90 days, where ΔV ∼ 2.5–3.5 mag from maximum.Hα reached its maximum observed flux at t = 31.7 days, which is after the visual maximum was reached at t = 27.9 days. The strong emission of Fe ii and Ca II, which is always present in the conventional principal spectrum (Payne-Gaposchkin 1957), persists until sometime before t ∼ 48 days. About five days after the optical maximum, the nebular lines of [O i] 6300 and 6363 Å appeared. The O ii lines, which are expected to became apparent at ΔB ∼ 1.5 mag below peak (Bath & Harkness 1989), are always present from t = 33.7–61.7 days.The Balmer lines began to show a double-peaked structure from the spectrum taken at t = 42.7 days. About twenty days after the maximum (t = 45.6 days), the forbidden line of [O iii] 5007 Å appeared. The flux of Balmer lines began to fade around t = 70 days. The "4640 emission" (Payne-Gaposchkin 1957), produced by a blend of N iii and N ii lines and known as a characteristic of the "Orion spectrum," becomes apparent at t ∼ 70–80 days as shown in Figure 19. The emission line of N v at 4603 Å, which is expected to appear during the Orion spectrum phase of a typical nova (Jaschek & Jaschek 2009), also begins to emerge at t = 73.7 days. The emergence of these lines is designated as "nitrogen flaring." Thus, we find that the Orion spectrum stage (Payne-Gaposchkin 1957; Warner 2008) started around t = 70 days. We note that the N iii and N v lines appeared earlier than would have been expected from the simple Bath & Harkness model.This evolution is in line with the progression toward the "nebular spectrum" stage of CN (Warner 2008). Meanwhile, the pseudo-photosphere is continuing to shrink in radius and the effective temperature is increasing.The He i 5876 Å line and the [Fe x] 6375 Å coronal line were marginally detected at t = 80.7 days (see Figure 20), which was the last spectrum observed before the seasonal gap. When spectroscopic observations started again at t = 155.1 days, the nova displayed the expected nebular spectrum.
• 5.  
  Transition to the nebular phase (2011 October 3 to December 19, t = 90–250 days). After the seasonal gap, T Pyx had declined to about 5 mag below maximum. By this time, as indicated in the previous phase, the [O iii] 5007 Å nebular and [Fe x] 6375 Å coronal lines had clearly developed. It is then of interest to note that the Swift satellite detected the rise in the X-ray light curve at t = 111 days (see below).The first blue spectrum (t = 155.1 days) already showed [Ne iii] 3869 Å, [C iii] 4364 Å, and N iii 4640 Å, which were increasing in intensity, especially the [C iii] line as shown in Figure 21. The appearance of N iii at t ∼ 90 days is consistent with the expectations of Bath & Harkness (1989) for ΔB = 2.9 mag. H, Hδ, and Hγ have already faded. Moreover, the coronal lines [Fe x] 6375 Å and [Fe vii] 6087 Å were clearly present in the first red spectrum at t = 161.1 days as shown in Figure 22. The Balmer lines are still stronger than [O iii] 5007 Å, however, and they present a multi-peak structure. [O iii] 5007 was the strongest rival to the Balmer lines at t = 165.1 days. The emission of [O iii] and N iii was strongest again at t = 221.9 days. The last spectroscopic observation was made at t = 249.9 days where the nova was clearly in the nebular spectrum stage and still exhibited [N ii], He i, [Fe vii], and [Fe x] lines.

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

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

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size