Chandra and XMM High Resolution and New Surprises

While only few observations were obtained with ASCA, RXTE, and BeppoSAX, another big step forward came with Chandra and XMM, which have higher sensitivity than ROSAT, grating spectrometers, which allow for X-ray observations with a resolution of up to (R/AR) — 1000, and an energy range that extends, at the hard end, far beyond ROSAT's 2.4 keV.

Already the first nova observed with Chandra, V382 Vel (1999), clearly demonstrated the impact of the new high resolution facilities. V382 Vel, a fast (t3 — 10d) ONeMg nova was with a maximum brightness Vmax=2.6 mag one of the brightest novae of the last century and was extensively observed in many different spectral regimes. X-ray observations with RXTE, ASCA, and BeppoSAX started immediately after discovery [18,27]. During the early phases a hard component was found, which Mukai and Ishida attributed to emission from shocks within the nebular ejecta. On day 185, Orio et al. [28] found with BeppoSAX observations a strong super-soft component, which was still present when V382 Vel was observed on day 223 for the first time with Chandra and ACIS [4].

The first high resolution observation (R — 600) of any nova in outburst were carried out by the same team on day 268 with the low energy transmission grating (LETG) onboard of Chandra [20]. Figure 13.3 shows the LETG spectrum. Surprisingly, the strong soft component had disappeared and was replaced by an emission line spectrum. Below 50 A a marginal continuum emission is found consistent with a black-body spectrum with a temperature of T = 2.7 x 105 K. The most prominent emission lines are N VI, O VII, O VIII, Ne IX, NeX, Mg X, Mg XI, and Si XII. No Fe lines were found. Ness et al. conclude that the abundances of all elements showing lines are significantly enhanced in the ejected material and that Fe shows normal abundance (to hydrogen), i.e., no underabundance. The emission lines are broadened and exhibit a complex line profile. From the Gaussian line width, Ness et al. estimate an expansion velocity of — 1200km s^1 which is somewhat lower than the one found from UV spectra [32]. Since the emission lines found in the spectrum are formed under a range of temperature conditions, Ness et al. could get some rough information on the temperature distribution responsible for the formation of the lines.

Fig. 13.3 Chandra LETG spectrum of V382 Vel (background subtracted). A continuum model is overplotted representing a diluted thermal black-body spectrum [20]

Fig. 13.3 Chandra LETG spectrum of V382 Vel (background subtracted). A continuum model is overplotted representing a diluted thermal black-body spectrum [20]

The absence of the soft-component on day 268 indicates that hydrogen burning must have turned off between days 223 and 268, probably shortly after day 223, since even the longest possible cooling time of about 6 weeks is extremely short. The total duration of the hydrogen burning was with only 7.5-8 months very short. This indicates that the white dwarf in V382 Vel has a high mass that is consistent with its ONeMg nature and the short decay time t3.

The next chapter in the story of X-ray observations was written by V1494 Aql (1999), a fast CO nova with a t3 of 13 days. The first two observations with Chandra's low resolution detector ACIS on days 134 and 187 yielded a hard spectrum with emission lines, but no soft component. After a strong soft component had appeared on day 247, two high-resolution observations with LETG+HRC were carried out on days 300 and 303 with exposure times of 8 and 17 ks, respectively. The high-resolution spectrum was the first one even obtained from a nova in its supersoft phase and features never seen before showed up. The spectra are dominated by a strong soft continuum component on which features are superimposed, which look on a first glance like emission lines. However, so far none of these features have been identified with any known emission lines. Therefore, it cannot be excluded that the spectrum is in reality an absorption spectrum where the emission features are only those part with less local absorption. UV spectra with such a character were observed during the early fireball phase [7]. However, during the early fireball phases one observes the initial expansion of the opaque shell that was certainly not the case 10 months after the outburst started. The application of suitable NLTE models to the observed SED would help to clarify this situation; however, this has not been done so far. The soft component had disappeared on day 726. Since no other X-ray observation had taken place in between, the duration of the hydrogen burning phase between 10 months and <2 years is rather badly defined.

0 5000 10000 15000

Elapsed Time (s)

Fig. 13.4 Chandra LETG+HRC-S X-ray light curve of V1494 Aql. Top: light curve obtained on day 300. Middle: Light curve obtained on day 303. Bottom: Same as middle panel but data scaled by the full range of the count rate to highlight the burst [5]

0 5000 10000 15000

Elapsed Time (s)

Fig. 13.4 Chandra LETG+HRC-S X-ray light curve of V1494 Aql. Top: light curve obtained on day 300. Middle: Light curve obtained on day 303. Bottom: Same as middle panel but data scaled by the full range of the count rate to highlight the burst [5]

An analysis of the X-ray lightcurve of V1494 Aql by Drake et al. [5] yielded several unusual features. Figure 13.4 shows the two X-ray lightcurves, which were extracted by Drake et. al from the bright zeroth order of the LETG+HRC-S spectrum.

Three diffent types of variability are found in these lightcurves: (i) A stochastic irregular short term variability is present on timescales of a few minutes. This kind of variability seems to be present in all X-ray lightcurves of novae obtained during the super-soft phase. For instance, from BeppoSAX observations of V382 Vel Orio et al. [28] found irregular flickering and a decrease of the flux in the 0.10.7 keV range by a factor of two within less than 1.5 h. The flux remained faint for some 15min; no significant spectral changes were found. So far no convincing explanation for this variability could be found; Orio et al. exclude absorption due to an ejected clump of matter since in this case the hardness ratio of the spectrum should have changed. (ii) Very obvious in the lower panel is a short time X-ray burst, which lasted about 1 000 s. At its peak, the count rate of the burst was about a factor of six higher than the mean level before and after the burst. During the burst, the X-ray count rate showed a complex rise and fall with several maxima and minima, two main flares with possibly a precursor and a trailer. During the burst, the spectrum is slightly harder than during the rest of the observations. So far the nature of this outburst remains a puzzle. (iii) A timing analysis of the combined 25 ks observations revealed the existence of periodic variations with a period P — 2499 s. This period was found in independent analyses of both the zeroth-order and the dispersed spectrum. To check whether the periodicity found might be an instrumental artifact, i.e., related to spacecraft motion or dithering, Drake et al. performed identical pe-riodogram analyses on HZ 43 and Sirius B. Since in neither of these analyses was any periodicity found, Drake et al. concluded that the observed periods are real and not instrumental. In addition to the 2 499 s period, several other periods were found. This suggests that the periodic variations are not due to rotation of the white dwarf. Drake et al. interpret this result as the discovery of nonradial g+-mode pulsations in the hot, rekindled white dwarf that is driven by the k/j effects in the partial ion-ization zones of C and O near the surface of the white dwarf. The hot, luminous white dwarf in a nova evolving from explosion to quiescence has a structure that resembles that of the central stars of planetary nebulae. The power spectrum and the X-ray lightcurve of V1494 Aql are very similar to the hot central star of several planetary nebulae [3].

V1494 Aql did not remain a unique case. Figure 13.5 shows the lightcurve of a 25 ks observation with Chandra/LETG+HRC of V4743 Sgr (2002), another fast nova with t3 < 15 days [19]. The observation was carried out on day 180 after the nova had entered the super-soft state.

Immediately obvious are large-amplitude variations from —30 to 60 counts s^1 with a period of 1325 s followed by a decline in the total count rate after —13 ks of observations. The count rate dropped from —40 counts s^1 to practically zero within —6 ks and stayed low for the remaining 6 ks of the observations. Besides the period of 1 325 s a timing analysis revealed two harmonic overtones at 668 and 448 s. Orio [29] reports that a lightcurve obtained from XMM-Newton observations 2 months after the Chandra observations yielded a rich power spectrum with two main periods of 1 308 and 1 374 s. As the middle and lower panels of Fig. 13.5 show, the spectral hardness ratio changed from maxima to minima in correlation with the oscillations and became significantly softer during the decay. Neither the interpretation of the periodic oscillations nor of the decline of the soft flux is straightforward. The strong amplitude of the oscillations is difficult to reconcile with pulsations of the white dwarf. Maybe a combination of rotation and pulsation of the white dwarf was the cause of the complicated lightcurve. For the decline of the count rate an eclipse would be the most obvious interpretation. However, the duration of the decline

time (sec) from MJD 52717.410

Fig. 13.5 Light curve of a 25 ks exposure of V4743 Sgr extracted in the designed wavelength intervals. Top: Complete wavelength ranges in zeroth and first order. Middle: Light curve broken into "hard" and "soft" components of the spectrum. Bottom: Time evolution of the hardness ratio [19]

time (sec) from MJD 52717.410

Fig. 13.5 Light curve of a 25 ks exposure of V4743 Sgr extracted in the designed wavelength intervals. Top: Complete wavelength ranges in zeroth and first order. Middle: Light curve broken into "hard" and "soft" components of the spectrum. Bottom: Time evolution of the hardness ratio [19]

(>6ks) is much too long for a typical cataclysmic variable binary period of several hours. If it was an eclipse, V4743 Sgr should have a longer period of ~2days like the old nova GK Per (1918), which is an intermediate polar with hard X-ray variations. An alternate solution could be a third component in the system. The measurement of the orbital binary period will help to clarify this question. The spectrum obtained after the decline of the count rate showed emission lines of CVI, NVI, and NVII, which are broadened by ^800-1200 km s^1. Three months later, at the XMM-Newton observation and at subsequent Chandra observations, the count rate had indreased again and was at about the same level as at the first obserervation on day 180.

The spectrum of V4743 Sgr (Fig. 13.6) exhibited a strong super-soft continuum with absorption features, which could be identified by Ness et al. as Hand He-like lines of CV, CVI, NVI, NVII, and OVII, and are blueshifted by ~2400kms_1. This velocity is twice the expansion velocity of 1200 kms-1 found for the material ejected early in the outburst. Petz et al. [30] carried out model calculations with PHOENIX, a 1D spherical, expanding, line blanketed, full NLTE model atmosphere [6]. A best fit yielded an effective temperature of Tef = 5.8 x 105 K, an expansion velocity of 2 500 km s^1 and a bolometric luminosity of 2x 1038 erg s^1.

Nova V4743 Sgr, LETGS, 2003-03-19

Nova V4743 Sgr, LETGS, 2003-03-19

20 30 40

Fig. 13.6 Chandra LETGS spectrum of V4743 Sgr. The strongest absorption lines are indicated. Probably also weak emission lines are present [19]

20 30 40

Fig. 13.6 Chandra LETGS spectrum of V4743 Sgr. The strongest absorption lines are indicated. Probably also weak emission lines are present [19]

0 0

Post a comment