Nova atmospheres

PHOENIX was initially designed to compute the radiation released by a rapidly expanding nova or supernova envelope during the first weeks and months after the outburst. A lot of work has been done by the PHOENIX group to model early nova atmospheres and spectra in the UV, IR, and visual spectral ranges (see Nova related publications in journals and proceedings). Latest works include the X-ray region of the electromagnetic spectrum. The computed spectra can be compared with observed spectra from the Low Energy Transmission Grating Spectrometer (LETGS) onboard the CHANDRA X-ray Observatory. The results will show the accuracy of the models and test the assumptions that we made.
The LETGS observations of nova V4743 Sagittarii (Ness et. al 2003, Ness et al. in preparation) have been modeled in the PhD Thesis of Alexander Petz. The sources and the evolution of the X-ray emission as well as the chemical composition, the atmosphere structure and the NLTE effects have been determined for this nova. Discrepancies between models and observations were discussed.

Main physical properties of novae

A classical nova occurs on a white dwarf showing a sudden increase in brightness by 7 to 18 mag. Novae are members of the cataclysmic variables (CVs). For a nova outburst to happen it is necessary that the white dwarf is a member of a binary system with a late-type main sequence star as the secondary component. The secondary has to be close enough to fill its Roche lobe. The white dwarf will accrete hydrogen rich material on its surface through an accretion disk (Fig. 1). There is a hot spot at the point where the material from the secondary star hits the accretion disk.

binary system
The white dwarf forms a hydrogen-rich electron-degenerated envelope around its core. After accreting a critical mass of hydrogen rich material, explosive hydrogen burning ignites in the accreted envelope through a non-equilibrium CNO-cycle. The hydrogen burning region is still electron degenerate, therefore, the released energy leads to a huge rise in the temperature with no expansion of the hot material. After reaching a temperature of several 108 K the degeneration breaks down and the gas starts to behave like an ideal gas. The energy generation is so large that the hydrogen rich envelope initially explosively expands and will be ejected as a wind during the later development. The released energy causes an enormous increase in the luminosity and the outer parts of the envelope, where there is no nuclear energy generation, will be blown away by the radiation pressure. This expanding material behaves like an optically thick wind and can be modeled with PHOENIX.
An expanding gas envelope forms around the nova, consisting of material from the accreted envelope, of the processed material during the burning cycle and of core material from the white dwarf. A so far open question is the mechanism of mixing the core material with accreted material. Because the emitting surface gets larger through the expansion, the temperatures will decrease and the maximum of the emitted radiation will be shifted from the UV to the visual spectral region, keeping the bolometric luminosity constant. The maximum visual brightness will be reached at a temperature of about 7000 to 10000 K, where the radius of the envelope will be about 1000 to 10000 times the white dwarf's radius. A steady energy generation can proceed in the outer layers of the white dwarf for a long time (about months) after it has ejected most of the envelope. After exhausting its fuel, the luminosity of the white dwarf decreases and the nova is going into the postnova state.
The emission from the postnova consists of the emission from the white dwarf, from the accretion disk with its hot spot and from the secondary star. Observed variations in the brightness are up to 1 mag, they could be caused by a non-uniform outflow of the material from the secondary star which causes a flickering in the hot spot and the accretion disk.
It is believed that all novae are recurrent during their lifetime, occurring with a period of about 105 years. The time between two outbursts depends on the accretion rate and the mass of the white dwarf, whereas a larger accretion rate and a larger dwarf mass lead to a shorter outburst period.
A classical nova is emitting in the X-ray with an intensity of only a few percent compared to the optical region. An exception are the X-ray novae, which are emitting about 1000 times more X-ray radiation. It is believed that they consist of a more compact object than a white dwarf, maybe a black hole. Causes for the outburst on these objects are widely unknown.


X-ray emission of classical novae

A theory for soft X-ray emission of novae exists. Three phases have to be distinguished:

  1. Early fireball phase: After starting the CNO-cycle the lifetimes for proton captures is much longer than the half life times of the β+ unstable nuclei. With increasing temperature the lifetimes for proton captures will become shorter and will be comparable with the β+ half life times short before the onset of the outburst. Furthermore, during the evolution to the outburst a convective region is formed in the accreted layer near the surface of the white dwarf. This convective region grows outside the accreted material with increasing temperature. Before the outburst the convection reaches the surface and can transport the β+ unstable nuclei to the surface of the accreted material (the convective turnover time is comparable to the β+ half life times). There, the β+ unstable nuclei can decay and support the energy necessary for the outburst. An energy of about L = 50 000 L is released in a shell with a radius of about 108 cm and the maximum of radiation is in the far UV and soft X-ray range. Therefore, a soft X-ray radiation like from a hot stellar atmosphere is expected.
  2. After the outburst: The wind of the nova will become optically thick for X-rays and there will be no soft X-ray radiation. The first phase is very short (in the order of hours) and no soft X-ray emission of a classical nova has been observed before the outburst.
  3. Constant bolometric luminosity phase: During the evolution of the nova wind the bolometric luminosity stays constant. Short after the outburst the maximum of the radiation is in the visual. With time, the layer where radiation is emitted (pseudo-photosphere), moves into deeper layers with higher temperatures and the maximum of radiation will be shifted in the UV. After a few months the hot outer layers of the white dwarf will be visible again. Not all of the accreted material is ejected during the outburst. The remaining material falls back on the WD surface where the hydrogen is still burning in the CNO-cycle. This material is hot enough (about 107-108 K) to emit soft X-ray radiation like in the early fireball phase.
In the second and third phase hard X-ray emission can appear from interaction of the ejected material with circumstellar material or with an older nova shell.

What has been done with PHOENIX on nova atmospheres in the UV, IR, and visual spectral ranges?

compare spectrum
Model atmospheres and synthetic spectra of novae in the UV, IR, and optical spectral range have been computed extensively with PHOENIX since the 1990s. For the calculations the combined radiative transfer, radiative equilibrium, and non-LTE rate equations for an expanding shell with a given density and velocity law have to be solved. Line blanketing is very important in nova models. Therefore, together with the Lagrangian radiation field, the solution of the radiative transfer equation for many wavelength points (about 50,000 to 500,000) is required. It is assumed that the nova photospheres are spherical and expanding, but steady state configurations because the time variations occur on long time scales compared to the time scales for radiation. Therefore, it is assumed that all time dependent terms both in hydrodynamics and in the radiative transfer equation can be neglected and all quantities depend only upon the radial coordinate.
The most important model parameters are the reference radius Rref (defined as the radius where the optical depth is unity), the effective temperature Teff (defined by Teff=(Lbol/4 π Rref2 σ)1/4 where Lbol is the bolometric luminosity and σ is the Stefan-Boltzmann's constant), the density parameter N (defined by ρ(r) ∝ r-N ), the maximum expansion velocity v which describes the mass loss rate, the velocity profile (for example the standard velocity profile v(r) = v (1 - a/r)b, with the free parameters v, a and b), the density at the outher edge of the envelope ρout and the elemental abundances.
In the following the most important results of the computed nova model atmospheres are listed:

  • The atmospheres are very extended, up to 1011cm.
  • The ejected material can reach the high velocities observed in novae (up to 4000 km s-1 ) by radiation pressure alone.
  • A sensitivity to the velocity profile can be seen in the spectra, which makes it possible to determine the real velocity profile.
  • The effects of line blanketing on the structure of the atmospheres and on the spectra are very large. For example the "Fe II forest" forms a quasi-continuum at UV wavelengths, while the observed emission lines in this wavelength range are merely "holes in the Fe iron curtain".
  • One can see very deep into the atmospheres, with a wide range of electron temperatures. Thus, the ionization conditions will vary widely and there are a large number of ionization stages of the same element simultaneously present in the atmospheres, as observed.
  • The structure of the atmospheres and the calculated spectra are very sensitive to changes of Fe, C, N and O abundances, which will make abundance determinations possible.
  • Non-LTE effects are very important in the nova atmospheres. A detailed self-consistent treatment of thousands of non-LTE levels of many ions is important to include the majority of the line blanketing opacity in non-LTE and to include the non-LTE effects on the structure of the atmospheres. Non-LTE may greatly affect the chemical concentration of ions that are important to the total opacity of the models, the flux distribution, the individual line strengths (in both directions) and to a lesser extend, the atmospheric structure (for example Pgas ). It has to be noted here that it is important to include non-LTE effects of several ionization stages of the same element because there are effects, for example overionization, from one ionization stage on the others. So, more than 300 ions of 39 elements with up to 26 ionization stages have been included in the calculations, yet.
Computed spectra with PHOENIX agree very good with observed spectra (for example by the IUE satellite) in the UV, IR, and optical spectral ranges. A comparison in the case of Nova Cygni 1992 during the optical thick wind phase can be seen in Fig. 2.

Nova model atmospheres in the X-ray spectral range computed with PHOENIX

In order to compute hot nova model atmospheres (with effective temperatures in the order of 105-106 K) in X-rays which are in the constant bolometric luminosity phase (third phase, see above) and to generate synthetic spectra in X-rays, data for energy electron transitions with high energies are necessary. Due to the relatively low densities in the nova wind scattering is an important opacity source and detailed non-LTE treatment is necessary. Many atomic lines of highly ionized atoms as well as electron and proton collision rates must be taken into account in the models. Therefore, new atomic databases with non-LTE lines in the X-ray spectral range have to be included. Current work has extended the interface to the CHIANTI Version 3 database to the CHIANTI Version 4 and 5 and the APED (ATOMDB) databases. The number of atomic lines which can be used in PHOENIX have now been enhanced from about 45 000 in the CHIANTI Version 3 database to over 100 000. In addition, collision rates for proton-proton collisions, relativistic and non-relativistic thermal bremsstrahlung for all elements up to z = 30, and two-photon continuum emission for H and He were included into the code. In future work dielectronic processes with data from the CHIANTI Version 5 database will be implemented.
The X-ray spectra of nova V4743 Sagittarii (2002) taken with the LETGS on March 2003, July 2003, September 2003, and February 2004 have been modeled. All four observations show a bright continuum between 19 and 55 Å with strong absorption lines from the highest ionization stages of carbon, nitrogen, and oxygen. It is not possible to fit the observations with a simple blackbody. Spherical, expanding, NLTE model atmosphere computed with PHOENIX are necessary. The results of a comparison between models and observations can be used to check the reliability of the model assumptions, to develop a model for the emission and evolution of X-ray radiation, to determine the chemical composition of the nova wind, and to examine the atmosphere structure and NLTE effects. The fits for the four observations can be found here and the fit parameters are in table 1. To fit the observations the abundances of helium, carbon, nitrogen, oxygen, and iron had to be changed and the continuum and lines had to be fitted simultaneously. A very large grid of abundances and effective temperatures were necessary to obtain the best fits.

Table 1: Fit parameters for the four LETGS observations of nova V4743 Sgr.
Fit-Parameter03 / 200307 / 200309 / 200302 / 2004
Teff in 105 K6.005.805.805.40
nH in 1021 cm-24.604.404.203.30
[He] / [He]7.53.03.03.0
[C] / [C]2.53.03.03.0
[N] / [N]17.027.026.026.0
[O] / [O]12.02.52.52.5
[Fe] / [Fe]0.10.10.10.1
The effective temperature was nearly constant between March and September 2003 and lower in February 2004, indicating that the CNO-cycle weakened. Between March and July 2003 the determined abundances were different but stayed the same in the other observations. A change in the abundances may be due to a further release of material with evolved abundances from the white dwarf after the outburst. This material has different abundances because it is from deeper layers of the accreted material or because the CNO-cycle change the abundances in the accreted material. If the change in the abundances between March and July 2003 was due to this mechanism, no more matter was ejected from the WD after July 2003 because the abundances did no more change.
An examination of the continuum processes in the atmosphere has shown that the X-ray emission is from thermal bremsstrahlung near the surface of the white dwarf where it is hot due to the hydrogen burning through the CNO-cycle. This emission is absorbed in continuum and lines in the optically thick wind. Below ∼ 25 Å the continuum absorption is very strong due to carbon, nitrogen, oxygen, and iron.
As in the case of the model atmospheres in the UV, IR, and optical spectral range the atmosphere is geometrically extended with a large range of temperatures. The highest ionization stages are present in the atmosphere. NLTE does change the ionization balances and show effects on the synthetic spectrum where the continuum becomes stronger and the spectral lines become weaker if treated in NLTE.
At the moment, the observation of nova V1494 Aql in figure 3 is modeled. The spectrum does not only show a strong continuum but also very strong forbidden emission lines. Therefore, far more extended model atmospheres including the optically thin outer nebula are necessary to fit the observation.

chandra spectrum
The computation of detailed non-LTE spectra requires large amounts of CPU time. Therefore, fast computers and the MPI-parallel mode of the PHOENIX-code are necessary. Supercomputers which can be used in parallel mode like the HLRN are in use for this work.

Nova related publications

Publications in journals
Publications in proceedings

Modeling atmospheres of classical novae in X-rays with PHOENIX
Alexander Petz, PhD-Thesis (2005)