Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The complications of learning from Super Soft Source X-ray spectra

The complications of learning from Super Soft Source X-ray spectra Dr. Jan-Uwe Ness (Researcher) XMM-Newton and Integral SOC, European Space Astronomy Centre, Camino Bajo del Castillo s/n, Urb. Villafranca del Castillo, 28692 Villanueva de la Canada, Madrid, Spain A R T I C L E I N F O A B S T R A C T Keywords: Super Soft X-ray Sources (SSS) are powered by nuclear burning on the surface of an accreting white Cataclysmic Variables dwarf, they are seen around 0.1-1 keV (thus in the soft X-ray regime), depending on effective tem- Super Soft Sources perature and the amount of intervening interstellar neutral hydrogen (N ). The most realistic model Nuclear burning to derive physical parameters from observed SSS spectra would be an atmosphere model that simu- lates the radiation transport processes. However, observed SSS high-resolution grating spectra reveal highly complex details that cast doubts on the feasibility of achieving unique results from atmosphere modeling. In this article, I discuss two independent atmosphere model analyses of the same data set, leading to different results. I then show some of the details that complicate the analysis and conclude that we need to approach the interpretation of high-resolution SSS spectra differently. We need to fo- cus more on the data than the models and to use more phenomenological approaches as is traditionally done with optical spectra. 1. Introduction In order to see a permanent SSS spectrum, special fine Cataclysmic Variables are characterised by mass trans- tuning of several parameters such as accretion rate and burn- fer from a companion star onto the surface of a white dwarf. ing rate is needed, and the class is thus very small. Further, If there was only accretion, the white dwarf would grow in the softness of the spectra allows us to only see them along mass until reaching the Chandrasekhar mass limit, when it lines of sight with low N , limiting the sample to either either collapses into a neutron star or explodes as a super- nearby systems or extragalactic ones such as Cal 83 in the nova Ia (if the composition is rich in carbon and oxygen). LMC. However, life is more complicated. Since the accreted ma- terial is rich in hydrogen, it can fuse to helium, an energy Transient SSS emission can be produced during nova source that leads to mass loss which can be more than the outbursts which have been observed with much higher count amount of mass gained by accretion. This may happen af- rates. It is conceivable that novae may also be intrinsically ter a long episode of accretion leading to a successive in- more luminous than permanent SSS because they rely on a crease in temperature and pressure. If ignition conditions are much larger reservoir of previously accreted and accumu- reached, the accreted hydrogen explodes in a thermonuclear lated material. During the early phases of a nova outburst, runaway known as a nova. The radiation pressure drives an the high-energy radiation produced on the surface of the white initially optically thick wind that obscures any high-energy dwarf is blocked by higher layers of optically thick material radiation from the nuclear burning zones. If the mass lost in that has been ejected by radiation pressure. Generally, after the wind is higher than the mass previously accreted, then several weeks to months, the ejecta become optically thin, the white dwarf cannot reach the Chandrasekhar mass limit. and the soft X-ray emission can be observed. If, however, When the wind becomes optically thin, the central burning nuclear burning switches off before the ejecta have cleared regions can be seen. Radiation temperatures then reach sev- to allow SSS emission to escape, the nova may never be seen eral 10 K, and a blackbody of this temperature has it’s Wien as a transient SSS. Some examples are discussed by Schwarz tail in the soft X-ray regime, around 1 keV (í 10 Å). A small et al. (2011). fraction of white dwarfs permanently host conditions of tem- perature and pressure to allow the hydrogen-rich accreted Transient SSS emission can be observed in other galax- material to undergo nuclear fusion at the same rate as the ies, owing to their brightness and the low amount of interstel- accretion rate. The observational class of these Super Soft lar absorption. As example is the remarkable short-period Sources (SSS) was discovered by the Einstein satellite and recurrent nova in M31N2008-12a (Henze et al., 2014) with significantly expanded with large-area ROSAT observations. multiple outbursts having bee observed every year since 2013. Initially it was believed that they were powered by accretion Observations of SSS emission from other galaxies may be in neutron stars, but van den Heuvel et al. (1992) proposed easier than for Galactic novae because of the lower amount nuclear burning on white dwarfs which was later confirmed of interstellar absorption. by Kahabka and van den Heuvel (1997). An early example of transient SSS emission from a nova Corresponding author juness@sciops.esa.int (J. Ness) outburst was described by Krautter et al. (1996) based on januweness.eu,juness@sciops.esa.int (J. Ness) ROSAT data (0.1-2.4 keV). The spectral resolution of R = ORCID(s): 0000-0003-0440-7193 (J. Ness) E/E= 0:5 at 1 keV is low enough that a blackbody fit al- J.-U. Ness: Preprint accepted by Elsevier Page 1 of 7 arXiv:1909.09711v1 [astro-ph.HE] 20 Sep 2019 Complex SSS spectra ready reproduced the data. However, the resulting bolomet- ric luminosity, derived from the Stefan-Boltzmann law as- suming spherical symmetry, was unrealistically high (far above the Eddington limit), and Krautter et al. (1996) emphasised that blackbody fits do not lead to reliable parameter esti- mates. A followup paper by Balman et al. (1998) analysed the same data set with LTE atmosphere models, also yielding good fits to the data but lower luminosities. The more real- istic physical assumptions led the authors to conclude that more realistic results have been found. A similar analysis was performed by Parmar et al. (1998, 1997a) using Bep- poSAX data (0.1-10 keV, R = 0:18 at 1 keV using equation 10 in Parmar et al. 1997b) of the two prototype SSSs Cal 83 and Cal 87, demonstrating that atmosphere models lead to lower luminosities, however, not to better values of  . The spectral resolution of BeppoSAX was designed to resolve absorption edges which are the most striking features at the temperatures and energies of SSS spectra. Abundances esti- mates are thus possible with the non-dispersive spectrome- ters. 2. High-resolution X-ray spectra of Super Soft Sources With the advent of the high-resolution X-ray gratings, operated on board the XMM-Newton (R = 0:004 at 1 keV) and Chandra (R = 0:001 at 1 keV) missions, it became pos- sible to resolve important details that are needed to better constrain atmosphere models, such as elemental abundances from the depth of absorption lines. From low-resolution spec- tra, only very rough abundance patterns can be estimated from the depth of absorption edges. The first X-ray grat- ing spectra of SSSs were taken of the prototype SSSs Cal 83 Figure 1: Comparison of high-resolution X-ray spectra of and Cal 87, and a comparison of these spectra is shown in the prototype SSSs Cal 83 (Chandra ObsID 1900) and Cal 87 Fig. 1. Even though, both sources are at the same distance, (XMM-Newton ObsID 0153250101), both located in the LMC. they differ in flux by a factor two (although Cal 87 is slightly In the bottom panel, blackbody fits are used to illustrate that brighter at higher energies), but more strikingly, the spec- Cal 83 is an atmosphere spectrum while Cal 87 hosts strong trum of Cal 87 is an emission line spectrum while Cal 83 re- emission lines on top of blackbody-like continuum. sembles more an atmosphere spectrum which was success- fully modeled with a non-LTE atmosphere model by Lanz et al. (2005). The low-resolution spectrum of Cal 87, shown the observed effective temperature (Wien tail) requires a ra- by Parmar et al. (1997a), is more structured, but the data dius of the pseudo photosphere to be small enough that it quality is low enough to consider the structure as Poisson must be fully eclipsed by the companion around phase 0. noise. A blackbody fit and an atmosphere model fit still yield The fact that we see continuum emission requires the contin- good fits to the poor data while the high-resolution spec- uum emission to have undergone some scattering processes. tra now show that both models are wrong. A very similar The fact that the continuum is not spectrally distorted indi- grating spectrum to that of Cal 87 was found by (Ness et al., cates that the scattering mechanism is independent of photon 2012) with the recurrent nova U Sco that is an eclipsing sys- energy, thus the conclusion by Ness et al. (2012) that we are tem, leading Ness et al. (2013) to conclude that SSS spectra dealing with Thomson scattering. like that of Cal 87 (which is also eclipsing) are strongly af- fected by obscuration effects, e.g., by an accretion disc in a high-inclination system. In the bottom panel of Fig. 1, 3. Analysing high-resolution SSS spectra with blackbody fits illustrate that for Cal 87, there is weak un- derlying atmospheric continuum emission, possibly origi- atmosphere models nating from the surface, a fraction of which getting to the After Krautter et al. (1996) made a strong case that black- observer via Thomson Scattering (Ness et al., 2012). As- body fits are not reliable, several approaches were attempted suming the bolometric luminosity within reasonable limits, to use atmosphere models. In the most central core of an J.-U. Ness: Preprint accepted by Elsevier Page 2 of 7 Complex SSS spectra Figure 3: Comparison of atmosphere model parameters of the two models by Rauch et al. (2010) and by van Rossum (2012). Figure 2: Two independent approaches for fitting atmosphere The observed (absorbed) fluxes shown in the top have been models to the same data set (Nova V4743 Sgr, day 180 after derived by integration of the fluxes in each wavelength bin. outburst, Chandra ObsID 3775 (day 180)) by Rauch et al. (2010) (top) and by van Rossum (2012) (bottom). N (bottom) derived from the two approaches and a simple atmosphere model, there is still a blackbody model (source blackbody fit. The effective temperature systematically dif- function), and the Stefan-Boltzmann law also applies to at- fers by í 25%. van Rossum (2012) has assumed a constant value of 5:5  10 K while Rauch et al. (2010) have iterated mosphere models - at least on a qualitative level. While the this parameter, detecting small variations above 7  10 K. blackbody model makes the extremely simplifying assump- A blackbody fit yields the lowest effective temperature val- tion of thermal equilibrium (TE), the next step foreward is to ues. The trends in temperature evolution differ only slightly. assume local thermal equilibrium (LTE, e.g., Balman et al. The TMAP model shows a small increase in temperature un- (1998)). More modern codes compute non-LTE (NLTE) at- til day 300 and a continuous decline thereafter. Meanwhile, mospheres. One such code developed for white dwarfs is the the wind-type model can explain the changes that the TMAP plane-parallel hydrostatic atmosphere code TMAP (Tübin- model attributes to temperature changes to changes in other gen NLTE Model-Atmosphere Package). This code has proven parameters and thus concludes the observations to be con- particularly useful for compact objects such as isolated white sistent with constant effective temperature until day 370 and dwarfs, e.g. Werner et al. (2012) or neutron stars, e.g. Rauch only then a decrease. The blackbody fits yield the largest et al. (2008). In the limit of highly compact objects, a plane changes in best-fit effective temperature which is likely a re- parallel geometry yields similar results as a spherically sym- metric geometry which is computationally more expensive. sult of ignoring any absorption that may be variable. A second important atmosphere code is the PHOENIX code by P. Hauschildt that solves the NLTE radiative transport Shortly after the outburst, the amount of interstellar ab- equations in a co-moving frame, thus allowing expanding sorption by neutral hydrogen was determined by Lyke et al. 21 *2 ejecta to be modelled. Based on PHOENIX, a wind-type (2002) with 1:4  10 cm , based on a method by Gehrz model for SSS spectra has been developed by van Rossum et al. (1974) that was also employed by Gehrz et al. (2015) (2012). for the nova V339 Del, also giving some more details. Inter- estingly, exactly the same value was found by van Rossum The first nova observed as a transient SSS with an X- (2012) who argues that N requires special attention prior ray grating was V4743 Sgr (Ness et al., 2003), and five SSS to any adjustment of atmosphere model parameters. Small spectra where taken at different times during the evolution of variations in N imply large differences in assumed fluxes at long (EUV) wavelengths at which hardly any flux is ob- this outburst. This dataset allows detailed conclusions about served. Therefore, poor values of N will hardly be pe- the evolution of the ejecta during the SSS phase. Two inde- nalised by  fitting even though strong assumptions are pendent analyses with atmosphere models were performed made about the invisible part of the spectrum. After careful by Rauch et al. (2010) (based on TMAP) and by van Rossum pre-adjustment of N , the values determined by van Rossum (2012) with examples of their models compared to data shown (2012) are all spot on with those derived by Lyke et al. (2002). in Fig. 2. One can see that agreement with the overall shape I emphasise here that Lyke et al. (2002) was not quoted by has been achieved in both cases while there are substantial van Rossum (2012), indicating that van Rossum (2012) was differences in the details. No statistical goodness criterion not biased by the literature value. Meanwhile, the N values like  was given, and from a statistical point of view, both derived by Rauch et al. (2010), determined by fitting N si- fits are unacceptable. In Fig. 3, I compare the evolution multaneously with other atmosphere parameters, are all well of the parameters effective temperature (middle panel) and J.-U. Ness: Preprint accepted by Elsevier Page 3 of 7 Complex SSS spectra below the value derived by Lyke et al. (2002) (which was also not quoted). This can either mean that Lyke et al. (2002) overestimated N , that the value of N was lower during the H H SSS phase, or that Rauch et al. (2010) underestimate N . Meanwhile, the values of N from the blackbody fits are all much higher which can either mean that local absorption was higher during the SSS phase or that the blackbody fit overestimates N . As all of the spectral fits have unaccept- able goodness of fit, it is also possible that all of the derived parameter values are incorrect. The reason why the effective temperature can be under- estimated while N is overestimated can be explained as fol- lows: Figure 4: Evolution of absorption lines in the two novae • If the ionisation energy of an abundant element such SMC 2016 (top three panels, Chandra ObsIDs 19011/19012 as nitrogen coincides with the Wien tail, a significant (days 38.9 and 86.2) and XMM-Newton ObsID 0794180201 amount of high-energy emission is missing. (day 73.8)) and V4743 Sgr (bottom two panels, Chandra Ob- sIDs 3775 (day 180) and 3776 (day 302)). Dotted lines mark a • A model not accounting for ionisation absorption edges N VI resonance line (28.8 Å), an unidentified line at í 30:5 Å, will see the Wien tail at longer wavelengths, thus un- an interstellar N I 1s-2p line at 31.3 Å, another unidentified derestimating the effective temperature. line at í 32:1 Å, and a C VI resonance line at 33.7 Å. The N VI • A model with underestimated effective temperature and C VI lines are blue shifted by the same amount during the evolution indicating continuous expanding absorbing material will overpredict the emission at soft energies, and rig- (see § 4.1). The unidentified line at 32.1 Å undergoes strange orous parameter fitting will increase N in order to changes. In SMC 2016, it was only present in the first obser- get rid of the excess soft emission. vation while in V4743 Sgr, it has changed from 32 Å to 32.1 Å Inversely, if the effects of absorption edges in the Wien tail (see § 4.2). are overestimated, then effective temperatures are overesti- mated and N underestimated. The large differences be- tween the two atmosphere models in N can thus be a con- sequence of the difference in effective temperature, and it is of high importance that the absorbing behaviour of the outer ejecta is well understood in order to derive reliable principle parameters. 4. The annoying little details In the following subsections, a few effects are described that are seen in the observed spectra that complicate any quantitative analysis based on global models. 4.1. Blue shifts Already the first SSS spectrum of a nova, V4743 Sgr, displayed absorption lines that are blue-shifted in excess of *1 2000 km s (Ness et al., 2003). Ness (2010) showed that line blue shifts are observed in many nova SSS spectra. In Fig. 4, some example spectra are shown where the blue shifts can be seen in two resonance lines of N VI and C VI. We have Figure 5: Comparison of line profiles of four spectra shown in determined the blue shifts with Gauss fits to the respective Fig. 4 (see text). absorption lines and show the results in Fig. 5. In SMC 2016, the expansion velocity has increased from day 38.9 to day 86.2 (47 days) by í 20% in both lines while in V4743 Sgr, The observed blue-shifted lines may either be formed in it has only increased, from day 180 to day 203 (122 days), an optically thin wind or in the ejecta. In SMC 2016, one by í 3%. The increase of observed blue shifts may indicate can see evidence for P Cyg profiles in the first observation, that during the evolution of the SSS phase, the photospheric whereas the emission component is not present in the later radius has continued to shrink into a regime with higher ve- two observations. A similar behaviour was seen in RS Oph locity, and the outflow is thus not constant with radius. (Ness et al., 2007). This may indicate that the mass loss rate J.-U. Ness: Preprint accepted by Elsevier Page 4 of 7 Complex SSS spectra can decrease during the SSS phase. Regardless of whether the absorption lines are formed in an optically thin wind or in the ejecta, both atmosphere model approaches compute the absorption lines as part of the same system. The model by Rauch et al. (2010) is a hydro- static model that cannot reproduce blue shifts self-consistently, and in order to fit the blue shifts in the observed spectra, a negative red-shift parameter was added. All the dynamics are thus not included, and it is unknown what effect (apart from blue-shifted absorption lines) the expansion can have on the observable spectrum. It is also not confirmed whether the photosphere is already on the surface of the white dwarf or somewhere within the outflow. The wind-type model by van Rossum (2012) assumes a hy- drostatic core with an optically thin wind on top of that and models line blue shifts selfconsistently. The absorption be- haviour of the wind can lead to significantly different model spectra compared to pure hydrostatic models, and adjust- Figure 6: Comparison of line profiles of three spectra of nova ment to observed spectra can thus lead to much different pa- V4743 Sgr at different times of the SSS evolution (see la- rameters. bels). The top three panels compare the profiles of the three It may be instructive to compare the Rauch et al. (2010) photospheric lines of N VI, C VI, and N VII while the bottom model with the hydrostatic core of the van Rossum (2012) panel shows the unidentified line assuming a rest wavelength which should lead to consistent results if the different im- of 32.3 Å. Best-fit blue shifts are given in the respective bottom plementations and side assumptions are correct. right legends. 4.2. Unidentified lines The example spectra in Fig. 4 also show two unidentified such as effective temperature, blue shifts of known absorp- lines at í 30:5 Å and í 32:1 Å in both novae. In SMC 2016, tion lines etc. The identification is important to improve the both lines were only present on day 38.9 while in V4743 Sgr, atomic data underlying the atmosphere models. they are seen in both example spectra, albeit with different profiles. Strangely, one line shifted from 32 Å on day 180 to 4.3. Complex line profiles 32.1 Å on day 302. A line shift of 0.1 Å at 32 Å corresponds In a hydrostatic atmosphere, one expects narrow absorp- *1 to almost 1000 km s . Only two spectra are shown as exam- tion lines, but we see not only blue-shifted but also broad- ples, but I looked at all five observation of V4743 Sgr (Chan- ened absorption lines. This is illustrated in Fig. 6 where the dra ObsIDs 3775 (day 180), 3776 (day 302), 4435 (day 371), two photospheric lines from Fig. 5 are shown in comparison and 5292 (day 526) and XMM-Newton ObsID 0127720501 to the N VII line at 24.74 Å. One can see that the profiles dif- (day 196)) and found that this change already took place be- fer dramatically with the N VII line yielding a line width of tween day 180 and day 196 (see bottom panel of Fig. 6) - *1 several thousand km s while the N VI and C VI lines are thus within only two weeks! - and remained at 32.1 Å af- much narrower. The line profile of the N VII line is also ter day 196. We know nothing about this line, it is obvi- complex with at least two sub-components that have similar ously not interstellar and thus unlikely to arise from a neu- widths as the N VI and C VI lines. One of these components, tral atom like the N I line at 31.3 Å. If it arises in the same *1 at í *4000 km s , has converted from an absorption line blue-shifted system as the high-ionisation lines of N VI and on day 180 to an emission line on day 196 but then back to C VI, the wavelength change in V4743 Sgr suggests a rest an absorption line day 302; possibly also the lower-blueshift wavelength of around 32.3 Å. Various databases give incon- *1 line at í *2500 km s . sistent results, e.g., there are several lines in NIST around 32.3 Å, e.g., Mg XI, S XIII while Atomdb The complexity of the N VII line profile indicates that we (http://www.atomdb.org/Webguide/webguide.php) are observing several velocity components simultaneously. lists K XI lines that go to the ground. Such identifications This observation adds to numerous observational evidence need to be carefully checked for consistency, e.g., with other for non-homogeneous or asymmetric outflows from novae lines in the same iso-electronic sequence which is beyond of such as high-amplitude variations during the early SSS phase this work. There are numerous other observed absorption (Osborne et al., 2011). Therefore, even the model by van lines without clear identifications in other novae, see table 5 Rossum (2012) is oversimplified. in Ness et al. (2011). An approach to identify them would be to assemble a data base with observed wavelengths of un- known lines and the environments in which they are formed J.-U. Ness: Preprint accepted by Elsevier Page 5 of 7 Complex SSS spectra Yet, a lot has been learned from optical spectra by classify- 5. Conclusions ing spectra by presence or absence of certain lines, creating The standard interpretation approach in X-ray astronomy classifications based on anomalously strong or weak types of for spectra is to fit spectral models to the observations. When lines etc. The limited spectral resolution of X-ray observa- the spectral resolution is low enough that no details are re- tions has not allowed such approaches so far, but after almost solved, this approach yields acceptable fits with already quite 20 years of having high-resolution grating spectra, it is time simple models. If resulting parameters are unrealistic, more to explore more such methods. An example is the classifi- complex models need to be tested, but it brings about the cation of SSS spectra dominated by emission or absorption problem that such models are overdetermined when fitted to lines by Ness et al. (2013). low-resolution spectra with only a few hundred spectral bins. A way forward would be to study the diverse absorption With the advent of high-resolution X-ray spectra, a paradigm line profiles, identify the corresponding transitions and clas- shift would be necessary, but this has not been realised yet. sify lines behaving similarly in order to determine the dy- namics in the different regions of the outflow. The formation Highly complex atmosphere models provide of course characteristics of each line give physical information such as more physics than blackbody fits, however, they only give temperature, so one can qualitatively localise the emission acceptable fits when fitted to low-resolution spectra. No at- region, assuming a radial temperature profile. mosphere model has so far been found that reproduces an SSS grating spectra in all the details. A quantitative assess- Future high-energy observatories such as XRISM or Athena ment of the goodness of fit appears hopeless at this time, will provide more high-resolution spectra, and I predict the and the best that has been achieved was qualitative agree- days of model fitting to low-resolution spectra to be counted. ment with data. So, are the atmosphere models useless? And what can References actually be learned from the high-quality SSS grating spec- Balman, S., Krautter, J., Ögelman, H., 1998. The X-Ray Spectral Evolution tra? of Classical Nova V1974 Cygni 1992: A Reanalysis of the ROSAT Data. ApJ 499, 395–406. Gehrz, R.D., Evans, A., Helton, L.A., Shenoy, D.P., Banerjee, D.P.K., I propose to use the atmosphere models for the interpre- Woodward, C.E., Vacca, W.D., Dykhoff, D.A., Ashok, N.M., Cass, A.C., tation of the data rather than trying to determine physical, 2015. The Early Infrared Temporal Development of Nova Delphini quantitative parameters. It is of course important to get a 2013 (V339 DEL) Observed with the Stratospheric Observatory for In- handle on the effective temperature - but what does it help frared Astronomy (SOFIA) and from the Ground. ApJ 812, 132–143. doi:10.1088/0004-637X/812/2/132. if we can never be sure whether we have actually determed Gehrz, R.D., Hackwell, J.A., Jones, T.W., 1974. Infrared observations of Be the correct value? Rather than believing we have found the stars from 2.3 to 19.5 microns. ApJ 191, 675–684. doi:10.1086/153008. correct values, we should consider any values found as an Henze, M., Ness, J.U., Darnley, M.J., Bode, M.F., Williams, S.C., Shafter, assumption because they depend on the assumptions behind A.W., Kato, M., Hachisu, I., 2014. A remarkable recurrent nova in M the models. 31: The X-ray observations. A&A 563, L8–L12. doi:10.1051/0004-6361/ 201423410, arXiv:1401.2904. Kahabka, P., van den Heuvel, E.P.J., 1997. Luminous Supersoft X-Ray An example use case for the atmosphere models is the Sources. ARA&A 35, 69–100. identification of the absorption lines which is tricky because Krautter, J., Ögelman, H., Starrfield, S., Wichmann, R., Pfeffermann, E., each identification needs to be consistent with the rest of the 1996. ROSAT X-Ray Observations of Nova V1974 Cygni: The Rise spectrum. It is not enough to consult an atomic database for and Fall of the Brightest Supersoft X-Ray Source. ApJ 456, 788–797. Lanz, T., Telis, G.A., Audard, M., Paerels, F., Rasmussen, A.P., Hubeny, I., the strongest line at a given wavelength. Any candidate iden- 2005. Non-LTE Model Atmosphere Analysis of the Large Magellanic tification needs to be checked for consistency, e.g., which Cloud Supersoft X-Ray Source CAL 83. ApJ 619, 517–526. other lines should then be seen or not seen. Doing this man- Lyke, J.E., Kelley, M.S., Gehrz, R.D., Woodward, C.E., 2002. Free-Free ually is at best tedious because one would have to go through Turnover in Nova V4743 Sgr 2002 #3. Bulletin of the American Astro- the entire multiplet, finding the strongest lines and compare nomical Society 34, 1161. Ness, J., 2010. Observational evidence for expansion in the SSS spectra which ones are seen at the expected wavelengths and with the of novae. Astronomische Nachrichten 331, 179–182. doi:10.1002/asna. expected strengths. It would not be possible to account for 200911322, arXiv:0908.4549. complex processes such as line emission and self absorption Ness, J., Schaefer, B.E., Dobrotka, A., Sadowski, A., Drake, J.J., Barnard, (thus radiation transport), and an atmosphere model would R., Talavera, A., Gonzalez-Riestra, R., Page, K.L., Hernanz, M., Sala, be the only possibility to determine line identifications un- G., Starrfield, S., 2012. From X-Ray Dips to Eclipse: Witnessing Disk Reformation in the Recurrent Nova U Sco. ApJ 745, 43–58. doi:10. der complex conditions in a self-consistent way. 1088/0004-637X/745/1/43, arXiv:1105.2717. Ness, J., Starrfield, S., Beardmore, A., Bode, M.F., Drake, J.J., Evans, When looking beyond X-ray astronomy, one will encounter A., Gehrz, R., Goad, M., Gonzalez-Riestra, R., Hauschildt, P., Kraut- numerous methods of interpretations, e.g., of optical spectra, ter, J., O’Brien, T.J., Osborne, J.P., Page, K.L., Schönrich, R., Wood- not using any models at all. Optical spectra are tradition- ward, C., 2007. The SSS phase of RS Ophiuchi observed with Chan- dra and XMM I. Data and preliminary Models. ApJ 665, 1334–1348. ally of too high quality that it was hopeless from the start to arXiv:astro-ph/0705.1206. find any model that could give quantitative parameter values. J.-U. Ness: Preprint accepted by Elsevier Page 6 of 7 Complex SSS spectra Ness, J., Starrfield, S., Burwitz, V., Wichmann, R., Hauschildt, P., Drake, J.J., Wagner, R.M., Bond, H.E., Krautter, J., Orio, M., Hernanz, M., Gehrz, R.D., Woodward, C.E., Butt, Y., Mukai, K., Balman, S., Truran, J.W., 2003. A Chandra Low Energy Transmission Grating Spectrome- ter Observation of V4743 Sagittarii: A Supersoft X-Ray Source and a Violently Variable Light Curve. ApJL 594, L127–L130. Ness, J.U., Osborne, J.P., Dobrotka, A., Page, K.L., Drake, J.J., Pinto, C., Detmers, R.G., Schwarz, G., Bode, M.F., Beardmore, A.P., Star- rfield, S., Hernanz, M., Sala, G., Krautter, J., Woodward, C.E., 2011. XMM-Newton X-ray and Ultraviolet Observations of the Fast Nova V2491 Cyg during the Supersoft Source Phase. ApJ 733, 70–85. doi:10.1088/0004-637X/733/1/70, arXiv:1103.4543. Ness, J.U., Osborne, J.P., Henze, M., Dobrotka, A., Drake, J.J., Ribeiro, V.A.R.M., Starrfield, S., Kuulkers, E., Behar, E., Hernanz, M., Schwarz, G., Page, K.L., Beardmore, A.P., Bode, M.F., 2013. Obscuration ef- fects in super-soft-source X-ray spectra. A&A 559, 1–15. doi:10.1051/ 0004-6361/201322415, arXiv:1309.2604. Osborne, J.P., Page, K.L., Beardmore, A.P., Bode, M.F., Goad, M.R., O’Brien, T.J., Starrfield, S., Rauch, T., Ness, J., Krautter, J., Schwarz, G., Burrows, D.N., Gehrels, N., Drake, J.J., Evans, A., Eyres, S.P.S., 2011. The Supersoft X-ray Phase of Nova RS Ophiuchi 2006. ApJ 727, 124–123. doi:10.1088/0004-637X/727/2/124, arXiv:1011.5327. Parmar, A.N., Kahabka, P., Hartmann, H.W., Heise, J., Martin, D.D.E., Bavdaz, M., Mineo, T., 1997a. A BeppoSAX obser- vation of the super-soft source CAL87. A&A 323, L33–L36. arXiv:arXiv:astro-ph/9706008. Parmar, A.N., Kahabka, P., Hartmann, H.W., Heise, J., Taylor, B.G., 1998. A BeppoSAX LECS observation of the super-soft X-ray source CAL 83. A&A 332, 199–203. arXiv:arXiv:astro-ph/9712040. Parmar, A.N., Martin, D.D.E., Bavdaz, M., Favata, F., Kuulkers, E., Va- canti, G., Lammers, U., Peacock, A., Taylor, B.G., 1997b. The low- energy concentrator spectrometer on-board the BeppoSAX X-ray astron- omy satellite. AAPS 122, 309–326. doi:10.1051/aas:1997137. Rauch, T., Orio, M., Gonzales-Riestra, R., Nelson, T., Still, M., Werner, K., Wilms, J., 2010. Non-local Thermal Equilibrium Model Atmospheres for the Hottest White Dwarfs: Spectral Analysis of the Compact Com- ponent in Nova V4743 Sgr. ApJ 717, 363–371. doi:10.1088/0004-637X/ 717/1/363, arXiv:1006.2918. Rauch, T., Suleimanov, V., Werner, K., 2008. Absorption features in the spectra of X-ray bursting neutron stars. A&A 490, 1127–1134. doi:10. 1051/0004-6361:200810129, arXiv:0809.2170. Schwarz, G.J., Ness, J., Osborne, J.P., Page, K.L., Evans, P.A., Beard- more, A.P., Walter, F.M., Helton, L.A., Woodward, C.E., Bode, M., Starrfield, S., Drake, J.J., 2011. Swift X-Ray Observations of Clas- sical Novae. II. The Super Soft Source Sample. ApJS 197, 31–56. doi:10.1088/0067-0049/197/2/31, arXiv:1110.6224. van den Heuvel, E.P.J., Bhattacharya, D., Nomoto, K., Rappaport, S.A., 1992. Accreting white dwarf models for CAL 83, CAL 87 and other ultrasoft X-ray sources in the LMC. A&A 262, 97–105. van Rossum, D.R., 2012. A Public Set of Synthetic Spectra from Expanding Atmospheres for X-Ray Novae. I. Solar Abundances. ApJ 756, 43–53. doi:10.1088/0004-637X/756/1/43, arXiv:1205.4267. Werner, K., Rauch, T., Ringat, E., Kruk, J.W., 2012. First Detection of Krypton and Xenon in a White Dwarf. ApJL 753, L7–L11. doi:10.1088/ 2041-8205/753/1/L7. J.-U. Ness: Preprint accepted by Elsevier Page 7 of 7 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Astrophysics arXiv (Cornell University)

The complications of learning from Super Soft Source X-ray spectra

Astrophysics , Volume 2019 (1909) – Sep 20, 2019

Loading next page...
 
/lp/arxiv-cornell-university/the-complications-of-learning-from-super-soft-source-x-ray-spectra-UWuR0WopdJ
ISSN
0273-1177
eISSN
ARCH-3330
DOI
10.1016/j.asr.2019.09.002
Publisher site
See Article on Publisher Site

Abstract

Dr. Jan-Uwe Ness (Researcher) XMM-Newton and Integral SOC, European Space Astronomy Centre, Camino Bajo del Castillo s/n, Urb. Villafranca del Castillo, 28692 Villanueva de la Canada, Madrid, Spain A R T I C L E I N F O A B S T R A C T Keywords: Super Soft X-ray Sources (SSS) are powered by nuclear burning on the surface of an accreting white Cataclysmic Variables dwarf, they are seen around 0.1-1 keV (thus in the soft X-ray regime), depending on effective tem- Super Soft Sources perature and the amount of intervening interstellar neutral hydrogen (N ). The most realistic model Nuclear burning to derive physical parameters from observed SSS spectra would be an atmosphere model that simu- lates the radiation transport processes. However, observed SSS high-resolution grating spectra reveal highly complex details that cast doubts on the feasibility of achieving unique results from atmosphere modeling. In this article, I discuss two independent atmosphere model analyses of the same data set, leading to different results. I then show some of the details that complicate the analysis and conclude that we need to approach the interpretation of high-resolution SSS spectra differently. We need to fo- cus more on the data than the models and to use more phenomenological approaches as is traditionally done with optical spectra. 1. Introduction In order to see a permanent SSS spectrum, special fine Cataclysmic Variables are characterised by mass trans- tuning of several parameters such as accretion rate and burn- fer from a companion star onto the surface of a white dwarf. ing rate is needed, and the class is thus very small. Further, If there was only accretion, the white dwarf would grow in the softness of the spectra allows us to only see them along mass until reaching the Chandrasekhar mass limit, when it lines of sight with low N , limiting the sample to either either collapses into a neutron star or explodes as a super- nearby systems or extragalactic ones such as Cal 83 in the nova Ia (if the composition is rich in carbon and oxygen). LMC. However, life is more complicated. Since the accreted ma- terial is rich in hydrogen, it can fuse to helium, an energy Transient SSS emission can be produced during nova source that leads to mass loss which can be more than the outbursts which have been observed with much higher count amount of mass gained by accretion. This may happen af- rates. It is conceivable that novae may also be intrinsically ter a long episode of accretion leading to a successive in- more luminous than permanent SSS because they rely on a crease in temperature and pressure. If ignition conditions are much larger reservoir of previously accreted and accumu- reached, the accreted hydrogen explodes in a thermonuclear lated material. During the early phases of a nova outburst, runaway known as a nova. The radiation pressure drives an the high-energy radiation produced on the surface of the white initially optically thick wind that obscures any high-energy dwarf is blocked by higher layers of optically thick material radiation from the nuclear burning zones. If the mass lost in that has been ejected by radiation pressure. Generally, after the wind is higher than the mass previously accreted, then several weeks to months, the ejecta become optically thin, the white dwarf cannot reach the Chandrasekhar mass limit. and the soft X-ray emission can be observed. If, however, When the wind becomes optically thin, the central burning nuclear burning switches off before the ejecta have cleared regions can be seen. Radiation temperatures then reach sev- to allow SSS emission to escape, the nova may never be seen eral 10 K, and a blackbody of this temperature has it’s Wien as a transient SSS. Some examples are discussed by Schwarz tail in the soft X-ray regime, around 1 keV (í 10 Å). A small et al. (2011). fraction of white dwarfs permanently host conditions of tem- perature and pressure to allow the hydrogen-rich accreted Transient SSS emission can be observed in other galax- material to undergo nuclear fusion at the same rate as the ies, owing to their brightness and the low amount of interstel- accretion rate. The observational class of these Super Soft lar absorption. As example is the remarkable short-period Sources (SSS) was discovered by the Einstein satellite and recurrent nova in M31N2008-12a (Henze et al., 2014) with significantly expanded with large-area ROSAT observations. multiple outbursts having bee observed every year since 2013. Initially it was believed that they were powered by accretion Observations of SSS emission from other galaxies may be in neutron stars, but van den Heuvel et al. (1992) proposed easier than for Galactic novae because of the lower amount nuclear burning on white dwarfs which was later confirmed of interstellar absorption. by Kahabka and van den Heuvel (1997). An early example of transient SSS emission from a nova Corresponding author juness@sciops.esa.int (J. Ness) outburst was described by Krautter et al. (1996) based on januweness.eu,juness@sciops.esa.int (J. Ness) ROSAT data (0.1-2.4 keV). The spectral resolution of R = ORCID(s): 0000-0003-0440-7193 (J. Ness) E/E= 0:5 at 1 keV is low enough that a blackbody fit al- J.-U. Ness: Preprint accepted by Elsevier Page 1 of 7 arXiv:1909.09711v1 [astro-ph.HE] 20 Sep 2019 Complex SSS spectra ready reproduced the data. However, the resulting bolomet- ric luminosity, derived from the Stefan-Boltzmann law as- suming spherical symmetry, was unrealistically high (far above the Eddington limit), and Krautter et al. (1996) emphasised that blackbody fits do not lead to reliable parameter esti- mates. A followup paper by Balman et al. (1998) analysed the same data set with LTE atmosphere models, also yielding good fits to the data but lower luminosities. The more real- istic physical assumptions led the authors to conclude that more realistic results have been found. A similar analysis was performed by Parmar et al. (1998, 1997a) using Bep- poSAX data (0.1-10 keV, R = 0:18 at 1 keV using equation 10 in Parmar et al. 1997b) of the two prototype SSSs Cal 83 and Cal 87, demonstrating that atmosphere models lead to lower luminosities, however, not to better values of  . The spectral resolution of BeppoSAX was designed to resolve absorption edges which are the most striking features at the temperatures and energies of SSS spectra. Abundances esti- mates are thus possible with the non-dispersive spectrome- ters. 2. High-resolution X-ray spectra of Super Soft Sources With the advent of the high-resolution X-ray gratings, operated on board the XMM-Newton (R = 0:004 at 1 keV) and Chandra (R = 0:001 at 1 keV) missions, it became pos- sible to resolve important details that are needed to better constrain atmosphere models, such as elemental abundances from the depth of absorption lines. From low-resolution spec- tra, only very rough abundance patterns can be estimated from the depth of absorption edges. The first X-ray grat- ing spectra of SSSs were taken of the prototype SSSs Cal 83 Figure 1: Comparison of high-resolution X-ray spectra of and Cal 87, and a comparison of these spectra is shown in the prototype SSSs Cal 83 (Chandra ObsID 1900) and Cal 87 Fig. 1. Even though, both sources are at the same distance, (XMM-Newton ObsID 0153250101), both located in the LMC. they differ in flux by a factor two (although Cal 87 is slightly In the bottom panel, blackbody fits are used to illustrate that brighter at higher energies), but more strikingly, the spec- Cal 83 is an atmosphere spectrum while Cal 87 hosts strong trum of Cal 87 is an emission line spectrum while Cal 83 re- emission lines on top of blackbody-like continuum. sembles more an atmosphere spectrum which was success- fully modeled with a non-LTE atmosphere model by Lanz et al. (2005). The low-resolution spectrum of Cal 87, shown the observed effective temperature (Wien tail) requires a ra- by Parmar et al. (1997a), is more structured, but the data dius of the pseudo photosphere to be small enough that it quality is low enough to consider the structure as Poisson must be fully eclipsed by the companion around phase 0. noise. A blackbody fit and an atmosphere model fit still yield The fact that we see continuum emission requires the contin- good fits to the poor data while the high-resolution spec- uum emission to have undergone some scattering processes. tra now show that both models are wrong. A very similar The fact that the continuum is not spectrally distorted indi- grating spectrum to that of Cal 87 was found by (Ness et al., cates that the scattering mechanism is independent of photon 2012) with the recurrent nova U Sco that is an eclipsing sys- energy, thus the conclusion by Ness et al. (2012) that we are tem, leading Ness et al. (2013) to conclude that SSS spectra dealing with Thomson scattering. like that of Cal 87 (which is also eclipsing) are strongly af- fected by obscuration effects, e.g., by an accretion disc in a high-inclination system. In the bottom panel of Fig. 1, 3. Analysing high-resolution SSS spectra with blackbody fits illustrate that for Cal 87, there is weak un- derlying atmospheric continuum emission, possibly origi- atmosphere models nating from the surface, a fraction of which getting to the After Krautter et al. (1996) made a strong case that black- observer via Thomson Scattering (Ness et al., 2012). As- body fits are not reliable, several approaches were attempted suming the bolometric luminosity within reasonable limits, to use atmosphere models. In the most central core of an J.-U. Ness: Preprint accepted by Elsevier Page 2 of 7 Complex SSS spectra Figure 3: Comparison of atmosphere model parameters of the two models by Rauch et al. (2010) and by van Rossum (2012). Figure 2: Two independent approaches for fitting atmosphere The observed (absorbed) fluxes shown in the top have been models to the same data set (Nova V4743 Sgr, day 180 after derived by integration of the fluxes in each wavelength bin. outburst, Chandra ObsID 3775 (day 180)) by Rauch et al. (2010) (top) and by van Rossum (2012) (bottom). N (bottom) derived from the two approaches and a simple atmosphere model, there is still a blackbody model (source blackbody fit. The effective temperature systematically dif- function), and the Stefan-Boltzmann law also applies to at- fers by í 25%. van Rossum (2012) has assumed a constant value of 5:5  10 K while Rauch et al. (2010) have iterated mosphere models - at least on a qualitative level. While the this parameter, detecting small variations above 7  10 K. blackbody model makes the extremely simplifying assump- A blackbody fit yields the lowest effective temperature val- tion of thermal equilibrium (TE), the next step foreward is to ues. The trends in temperature evolution differ only slightly. assume local thermal equilibrium (LTE, e.g., Balman et al. The TMAP model shows a small increase in temperature un- (1998)). More modern codes compute non-LTE (NLTE) at- til day 300 and a continuous decline thereafter. Meanwhile, mospheres. One such code developed for white dwarfs is the the wind-type model can explain the changes that the TMAP plane-parallel hydrostatic atmosphere code TMAP (Tübin- model attributes to temperature changes to changes in other gen NLTE Model-Atmosphere Package). This code has proven parameters and thus concludes the observations to be con- particularly useful for compact objects such as isolated white sistent with constant effective temperature until day 370 and dwarfs, e.g. Werner et al. (2012) or neutron stars, e.g. Rauch only then a decrease. The blackbody fits yield the largest et al. (2008). In the limit of highly compact objects, a plane changes in best-fit effective temperature which is likely a re- parallel geometry yields similar results as a spherically sym- metric geometry which is computationally more expensive. sult of ignoring any absorption that may be variable. A second important atmosphere code is the PHOENIX code by P. Hauschildt that solves the NLTE radiative transport Shortly after the outburst, the amount of interstellar ab- equations in a co-moving frame, thus allowing expanding sorption by neutral hydrogen was determined by Lyke et al. 21 *2 ejecta to be modelled. Based on PHOENIX, a wind-type (2002) with 1:4  10 cm , based on a method by Gehrz model for SSS spectra has been developed by van Rossum et al. (1974) that was also employed by Gehrz et al. (2015) (2012). for the nova V339 Del, also giving some more details. Inter- estingly, exactly the same value was found by van Rossum The first nova observed as a transient SSS with an X- (2012) who argues that N requires special attention prior ray grating was V4743 Sgr (Ness et al., 2003), and five SSS to any adjustment of atmosphere model parameters. Small spectra where taken at different times during the evolution of variations in N imply large differences in assumed fluxes at long (EUV) wavelengths at which hardly any flux is ob- this outburst. This dataset allows detailed conclusions about served. Therefore, poor values of N will hardly be pe- the evolution of the ejecta during the SSS phase. Two inde- nalised by  fitting even though strong assumptions are pendent analyses with atmosphere models were performed made about the invisible part of the spectrum. After careful by Rauch et al. (2010) (based on TMAP) and by van Rossum pre-adjustment of N , the values determined by van Rossum (2012) with examples of their models compared to data shown (2012) are all spot on with those derived by Lyke et al. (2002). in Fig. 2. One can see that agreement with the overall shape I emphasise here that Lyke et al. (2002) was not quoted by has been achieved in both cases while there are substantial van Rossum (2012), indicating that van Rossum (2012) was differences in the details. No statistical goodness criterion not biased by the literature value. Meanwhile, the N values like  was given, and from a statistical point of view, both derived by Rauch et al. (2010), determined by fitting N si- fits are unacceptable. In Fig. 3, I compare the evolution multaneously with other atmosphere parameters, are all well of the parameters effective temperature (middle panel) and J.-U. Ness: Preprint accepted by Elsevier Page 3 of 7 Complex SSS spectra below the value derived by Lyke et al. (2002) (which was also not quoted). This can either mean that Lyke et al. (2002) overestimated N , that the value of N was lower during the H H SSS phase, or that Rauch et al. (2010) underestimate N . Meanwhile, the values of N from the blackbody fits are all much higher which can either mean that local absorption was higher during the SSS phase or that the blackbody fit overestimates N . As all of the spectral fits have unaccept- able goodness of fit, it is also possible that all of the derived parameter values are incorrect. The reason why the effective temperature can be under- estimated while N is overestimated can be explained as fol- lows: Figure 4: Evolution of absorption lines in the two novae • If the ionisation energy of an abundant element such SMC 2016 (top three panels, Chandra ObsIDs 19011/19012 as nitrogen coincides with the Wien tail, a significant (days 38.9 and 86.2) and XMM-Newton ObsID 0794180201 amount of high-energy emission is missing. (day 73.8)) and V4743 Sgr (bottom two panels, Chandra Ob- sIDs 3775 (day 180) and 3776 (day 302)). Dotted lines mark a • A model not accounting for ionisation absorption edges N VI resonance line (28.8 Å), an unidentified line at í 30:5 Å, will see the Wien tail at longer wavelengths, thus un- an interstellar N I 1s-2p line at 31.3 Å, another unidentified derestimating the effective temperature. line at í 32:1 Å, and a C VI resonance line at 33.7 Å. The N VI • A model with underestimated effective temperature and C VI lines are blue shifted by the same amount during the evolution indicating continuous expanding absorbing material will overpredict the emission at soft energies, and rig- (see § 4.1). The unidentified line at 32.1 Å undergoes strange orous parameter fitting will increase N in order to changes. In SMC 2016, it was only present in the first obser- get rid of the excess soft emission. vation while in V4743 Sgr, it has changed from 32 Å to 32.1 Å Inversely, if the effects of absorption edges in the Wien tail (see § 4.2). are overestimated, then effective temperatures are overesti- mated and N underestimated. The large differences be- tween the two atmosphere models in N can thus be a con- sequence of the difference in effective temperature, and it is of high importance that the absorbing behaviour of the outer ejecta is well understood in order to derive reliable principle parameters. 4. The annoying little details In the following subsections, a few effects are described that are seen in the observed spectra that complicate any quantitative analysis based on global models. 4.1. Blue shifts Already the first SSS spectrum of a nova, V4743 Sgr, displayed absorption lines that are blue-shifted in excess of *1 2000 km s (Ness et al., 2003). Ness (2010) showed that line blue shifts are observed in many nova SSS spectra. In Fig. 4, some example spectra are shown where the blue shifts can be seen in two resonance lines of N VI and C VI. We have Figure 5: Comparison of line profiles of four spectra shown in determined the blue shifts with Gauss fits to the respective Fig. 4 (see text). absorption lines and show the results in Fig. 5. In SMC 2016, the expansion velocity has increased from day 38.9 to day 86.2 (47 days) by í 20% in both lines while in V4743 Sgr, The observed blue-shifted lines may either be formed in it has only increased, from day 180 to day 203 (122 days), an optically thin wind or in the ejecta. In SMC 2016, one by í 3%. The increase of observed blue shifts may indicate can see evidence for P Cyg profiles in the first observation, that during the evolution of the SSS phase, the photospheric whereas the emission component is not present in the later radius has continued to shrink into a regime with higher ve- two observations. A similar behaviour was seen in RS Oph locity, and the outflow is thus not constant with radius. (Ness et al., 2007). This may indicate that the mass loss rate J.-U. Ness: Preprint accepted by Elsevier Page 4 of 7 Complex SSS spectra can decrease during the SSS phase. Regardless of whether the absorption lines are formed in an optically thin wind or in the ejecta, both atmosphere model approaches compute the absorption lines as part of the same system. The model by Rauch et al. (2010) is a hydro- static model that cannot reproduce blue shifts self-consistently, and in order to fit the blue shifts in the observed spectra, a negative red-shift parameter was added. All the dynamics are thus not included, and it is unknown what effect (apart from blue-shifted absorption lines) the expansion can have on the observable spectrum. It is also not confirmed whether the photosphere is already on the surface of the white dwarf or somewhere within the outflow. The wind-type model by van Rossum (2012) assumes a hy- drostatic core with an optically thin wind on top of that and models line blue shifts selfconsistently. The absorption be- haviour of the wind can lead to significantly different model spectra compared to pure hydrostatic models, and adjust- Figure 6: Comparison of line profiles of three spectra of nova ment to observed spectra can thus lead to much different pa- V4743 Sgr at different times of the SSS evolution (see la- rameters. bels). The top three panels compare the profiles of the three It may be instructive to compare the Rauch et al. (2010) photospheric lines of N VI, C VI, and N VII while the bottom model with the hydrostatic core of the van Rossum (2012) panel shows the unidentified line assuming a rest wavelength which should lead to consistent results if the different im- of 32.3 Å. Best-fit blue shifts are given in the respective bottom plementations and side assumptions are correct. right legends. 4.2. Unidentified lines The example spectra in Fig. 4 also show two unidentified such as effective temperature, blue shifts of known absorp- lines at í 30:5 Å and í 32:1 Å in both novae. In SMC 2016, tion lines etc. The identification is important to improve the both lines were only present on day 38.9 while in V4743 Sgr, atomic data underlying the atmosphere models. they are seen in both example spectra, albeit with different profiles. Strangely, one line shifted from 32 Å on day 180 to 4.3. Complex line profiles 32.1 Å on day 302. A line shift of 0.1 Å at 32 Å corresponds In a hydrostatic atmosphere, one expects narrow absorp- *1 to almost 1000 km s . Only two spectra are shown as exam- tion lines, but we see not only blue-shifted but also broad- ples, but I looked at all five observation of V4743 Sgr (Chan- ened absorption lines. This is illustrated in Fig. 6 where the dra ObsIDs 3775 (day 180), 3776 (day 302), 4435 (day 371), two photospheric lines from Fig. 5 are shown in comparison and 5292 (day 526) and XMM-Newton ObsID 0127720501 to the N VII line at 24.74 Å. One can see that the profiles dif- (day 196)) and found that this change already took place be- fer dramatically with the N VII line yielding a line width of tween day 180 and day 196 (see bottom panel of Fig. 6) - *1 several thousand km s while the N VI and C VI lines are thus within only two weeks! - and remained at 32.1 Å af- much narrower. The line profile of the N VII line is also ter day 196. We know nothing about this line, it is obvi- complex with at least two sub-components that have similar ously not interstellar and thus unlikely to arise from a neu- widths as the N VI and C VI lines. One of these components, tral atom like the N I line at 31.3 Å. If it arises in the same *1 at í *4000 km s , has converted from an absorption line blue-shifted system as the high-ionisation lines of N VI and on day 180 to an emission line on day 196 but then back to C VI, the wavelength change in V4743 Sgr suggests a rest an absorption line day 302; possibly also the lower-blueshift wavelength of around 32.3 Å. Various databases give incon- *1 line at í *2500 km s . sistent results, e.g., there are several lines in NIST around 32.3 Å, e.g., Mg XI, S XIII while Atomdb The complexity of the N VII line profile indicates that we (http://www.atomdb.org/Webguide/webguide.php) are observing several velocity components simultaneously. lists K XI lines that go to the ground. Such identifications This observation adds to numerous observational evidence need to be carefully checked for consistency, e.g., with other for non-homogeneous or asymmetric outflows from novae lines in the same iso-electronic sequence which is beyond of such as high-amplitude variations during the early SSS phase this work. There are numerous other observed absorption (Osborne et al., 2011). Therefore, even the model by van lines without clear identifications in other novae, see table 5 Rossum (2012) is oversimplified. in Ness et al. (2011). An approach to identify them would be to assemble a data base with observed wavelengths of un- known lines and the environments in which they are formed J.-U. Ness: Preprint accepted by Elsevier Page 5 of 7 Complex SSS spectra Yet, a lot has been learned from optical spectra by classify- 5. Conclusions ing spectra by presence or absence of certain lines, creating The standard interpretation approach in X-ray astronomy classifications based on anomalously strong or weak types of for spectra is to fit spectral models to the observations. When lines etc. The limited spectral resolution of X-ray observa- the spectral resolution is low enough that no details are re- tions has not allowed such approaches so far, but after almost solved, this approach yields acceptable fits with already quite 20 years of having high-resolution grating spectra, it is time simple models. If resulting parameters are unrealistic, more to explore more such methods. An example is the classifi- complex models need to be tested, but it brings about the cation of SSS spectra dominated by emission or absorption problem that such models are overdetermined when fitted to lines by Ness et al. (2013). low-resolution spectra with only a few hundred spectral bins. A way forward would be to study the diverse absorption With the advent of high-resolution X-ray spectra, a paradigm line profiles, identify the corresponding transitions and clas- shift would be necessary, but this has not been realised yet. sify lines behaving similarly in order to determine the dy- namics in the different regions of the outflow. The formation Highly complex atmosphere models provide of course characteristics of each line give physical information such as more physics than blackbody fits, however, they only give temperature, so one can qualitatively localise the emission acceptable fits when fitted to low-resolution spectra. No at- region, assuming a radial temperature profile. mosphere model has so far been found that reproduces an SSS grating spectra in all the details. A quantitative assess- Future high-energy observatories such as XRISM or Athena ment of the goodness of fit appears hopeless at this time, will provide more high-resolution spectra, and I predict the and the best that has been achieved was qualitative agree- days of model fitting to low-resolution spectra to be counted. ment with data. So, are the atmosphere models useless? And what can References actually be learned from the high-quality SSS grating spec- Balman, S., Krautter, J., Ögelman, H., 1998. The X-Ray Spectral Evolution tra? of Classical Nova V1974 Cygni 1992: A Reanalysis of the ROSAT Data. ApJ 499, 395–406. Gehrz, R.D., Evans, A., Helton, L.A., Shenoy, D.P., Banerjee, D.P.K., I propose to use the atmosphere models for the interpre- Woodward, C.E., Vacca, W.D., Dykhoff, D.A., Ashok, N.M., Cass, A.C., tation of the data rather than trying to determine physical, 2015. The Early Infrared Temporal Development of Nova Delphini quantitative parameters. It is of course important to get a 2013 (V339 DEL) Observed with the Stratospheric Observatory for In- handle on the effective temperature - but what does it help frared Astronomy (SOFIA) and from the Ground. ApJ 812, 132–143. doi:10.1088/0004-637X/812/2/132. if we can never be sure whether we have actually determed Gehrz, R.D., Hackwell, J.A., Jones, T.W., 1974. Infrared observations of Be the correct value? Rather than believing we have found the stars from 2.3 to 19.5 microns. ApJ 191, 675–684. doi:10.1086/153008. correct values, we should consider any values found as an Henze, M., Ness, J.U., Darnley, M.J., Bode, M.F., Williams, S.C., Shafter, assumption because they depend on the assumptions behind A.W., Kato, M., Hachisu, I., 2014. A remarkable recurrent nova in M the models. 31: The X-ray observations. A&A 563, L8–L12. doi:10.1051/0004-6361/ 201423410, arXiv:1401.2904. Kahabka, P., van den Heuvel, E.P.J., 1997. Luminous Supersoft X-Ray An example use case for the atmosphere models is the Sources. ARA&A 35, 69–100. identification of the absorption lines which is tricky because Krautter, J., Ögelman, H., Starrfield, S., Wichmann, R., Pfeffermann, E., each identification needs to be consistent with the rest of the 1996. ROSAT X-Ray Observations of Nova V1974 Cygni: The Rise spectrum. It is not enough to consult an atomic database for and Fall of the Brightest Supersoft X-Ray Source. ApJ 456, 788–797. Lanz, T., Telis, G.A., Audard, M., Paerels, F., Rasmussen, A.P., Hubeny, I., the strongest line at a given wavelength. Any candidate iden- 2005. Non-LTE Model Atmosphere Analysis of the Large Magellanic tification needs to be checked for consistency, e.g., which Cloud Supersoft X-Ray Source CAL 83. ApJ 619, 517–526. other lines should then be seen or not seen. Doing this man- Lyke, J.E., Kelley, M.S., Gehrz, R.D., Woodward, C.E., 2002. Free-Free ually is at best tedious because one would have to go through Turnover in Nova V4743 Sgr 2002 #3. Bulletin of the American Astro- the entire multiplet, finding the strongest lines and compare nomical Society 34, 1161. Ness, J., 2010. Observational evidence for expansion in the SSS spectra which ones are seen at the expected wavelengths and with the of novae. Astronomische Nachrichten 331, 179–182. doi:10.1002/asna. expected strengths. It would not be possible to account for 200911322, arXiv:0908.4549. complex processes such as line emission and self absorption Ness, J., Schaefer, B.E., Dobrotka, A., Sadowski, A., Drake, J.J., Barnard, (thus radiation transport), and an atmosphere model would R., Talavera, A., Gonzalez-Riestra, R., Page, K.L., Hernanz, M., Sala, be the only possibility to determine line identifications un- G., Starrfield, S., 2012. From X-Ray Dips to Eclipse: Witnessing Disk Reformation in the Recurrent Nova U Sco. ApJ 745, 43–58. doi:10. der complex conditions in a self-consistent way. 1088/0004-637X/745/1/43, arXiv:1105.2717. Ness, J., Starrfield, S., Beardmore, A., Bode, M.F., Drake, J.J., Evans, When looking beyond X-ray astronomy, one will encounter A., Gehrz, R., Goad, M., Gonzalez-Riestra, R., Hauschildt, P., Kraut- numerous methods of interpretations, e.g., of optical spectra, ter, J., O’Brien, T.J., Osborne, J.P., Page, K.L., Schönrich, R., Wood- not using any models at all. Optical spectra are tradition- ward, C., 2007. The SSS phase of RS Ophiuchi observed with Chan- dra and XMM I. Data and preliminary Models. ApJ 665, 1334–1348. ally of too high quality that it was hopeless from the start to arXiv:astro-ph/0705.1206. find any model that could give quantitative parameter values. J.-U. Ness: Preprint accepted by Elsevier Page 6 of 7 Complex SSS spectra Ness, J., Starrfield, S., Burwitz, V., Wichmann, R., Hauschildt, P., Drake, J.J., Wagner, R.M., Bond, H.E., Krautter, J., Orio, M., Hernanz, M., Gehrz, R.D., Woodward, C.E., Butt, Y., Mukai, K., Balman, S., Truran, J.W., 2003. A Chandra Low Energy Transmission Grating Spectrome- ter Observation of V4743 Sagittarii: A Supersoft X-Ray Source and a Violently Variable Light Curve. ApJL 594, L127–L130. Ness, J.U., Osborne, J.P., Dobrotka, A., Page, K.L., Drake, J.J., Pinto, C., Detmers, R.G., Schwarz, G., Bode, M.F., Beardmore, A.P., Star- rfield, S., Hernanz, M., Sala, G., Krautter, J., Woodward, C.E., 2011. XMM-Newton X-ray and Ultraviolet Observations of the Fast Nova V2491 Cyg during the Supersoft Source Phase. ApJ 733, 70–85. doi:10.1088/0004-637X/733/1/70, arXiv:1103.4543. Ness, J.U., Osborne, J.P., Henze, M., Dobrotka, A., Drake, J.J., Ribeiro, V.A.R.M., Starrfield, S., Kuulkers, E., Behar, E., Hernanz, M., Schwarz, G., Page, K.L., Beardmore, A.P., Bode, M.F., 2013. Obscuration ef- fects in super-soft-source X-ray spectra. A&A 559, 1–15. doi:10.1051/ 0004-6361/201322415, arXiv:1309.2604. Osborne, J.P., Page, K.L., Beardmore, A.P., Bode, M.F., Goad, M.R., O’Brien, T.J., Starrfield, S., Rauch, T., Ness, J., Krautter, J., Schwarz, G., Burrows, D.N., Gehrels, N., Drake, J.J., Evans, A., Eyres, S.P.S., 2011. The Supersoft X-ray Phase of Nova RS Ophiuchi 2006. ApJ 727, 124–123. doi:10.1088/0004-637X/727/2/124, arXiv:1011.5327. Parmar, A.N., Kahabka, P., Hartmann, H.W., Heise, J., Martin, D.D.E., Bavdaz, M., Mineo, T., 1997a. A BeppoSAX obser- vation of the super-soft source CAL87. A&A 323, L33–L36. arXiv:arXiv:astro-ph/9706008. Parmar, A.N., Kahabka, P., Hartmann, H.W., Heise, J., Taylor, B.G., 1998. A BeppoSAX LECS observation of the super-soft X-ray source CAL 83. A&A 332, 199–203. arXiv:arXiv:astro-ph/9712040. Parmar, A.N., Martin, D.D.E., Bavdaz, M., Favata, F., Kuulkers, E., Va- canti, G., Lammers, U., Peacock, A., Taylor, B.G., 1997b. The low- energy concentrator spectrometer on-board the BeppoSAX X-ray astron- omy satellite. AAPS 122, 309–326. doi:10.1051/aas:1997137. Rauch, T., Orio, M., Gonzales-Riestra, R., Nelson, T., Still, M., Werner, K., Wilms, J., 2010. Non-local Thermal Equilibrium Model Atmospheres for the Hottest White Dwarfs: Spectral Analysis of the Compact Com- ponent in Nova V4743 Sgr. ApJ 717, 363–371. doi:10.1088/0004-637X/ 717/1/363, arXiv:1006.2918. Rauch, T., Suleimanov, V., Werner, K., 2008. Absorption features in the spectra of X-ray bursting neutron stars. A&A 490, 1127–1134. doi:10. 1051/0004-6361:200810129, arXiv:0809.2170. Schwarz, G.J., Ness, J., Osborne, J.P., Page, K.L., Evans, P.A., Beard- more, A.P., Walter, F.M., Helton, L.A., Woodward, C.E., Bode, M., Starrfield, S., Drake, J.J., 2011. Swift X-Ray Observations of Clas- sical Novae. II. The Super Soft Source Sample. ApJS 197, 31–56. doi:10.1088/0067-0049/197/2/31, arXiv:1110.6224. van den Heuvel, E.P.J., Bhattacharya, D., Nomoto, K., Rappaport, S.A., 1992. Accreting white dwarf models for CAL 83, CAL 87 and other ultrasoft X-ray sources in the LMC. A&A 262, 97–105. van Rossum, D.R., 2012. A Public Set of Synthetic Spectra from Expanding Atmospheres for X-Ray Novae. I. Solar Abundances. ApJ 756, 43–53. doi:10.1088/0004-637X/756/1/43, arXiv:1205.4267. Werner, K., Rauch, T., Ringat, E., Kruk, J.W., 2012. First Detection of Krypton and Xenon in a White Dwarf. ApJL 753, L7–L11. doi:10.1088/ 2041-8205/753/1/L7. J.-U. Ness: Preprint accepted by Elsevier Page 7 of 7

Journal

AstrophysicsarXiv (Cornell University)

Published: Sep 20, 2019

References