A white dwarf with an oxygen atmosphere, Science, 1 April 2016, Vol. 352, No. 6281

S.O. Kepler¹, Detlev Koester^1,2, Gustavo Ourique¹

¹Departamento de Astronomia, Instituto de Física, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS – Brazil.
²Institut für Theoretische Physik und Astrophysik, Universität Kiel, 24098 Kiel, Germany.

Image of the blue Dox SDSS J124043.01+671034.68 obtained by the Sloan Digital Sky Survey.

Model of the spectra, calculated by Dr. Detlev Koester.

Reports on the discovery of the first star, a white dwarf, with an atmosphere dominated by oxygen, SDSS J124043.01+671034.68, a g=18.3 magnitude star with 0.2 arcsec/yr proper motion. The second most abundant element is Magnesium.

White dwarf stars are the end product of single stars that do not explode as supernova, because they do not have enough mass to increase the central pressure and raise their core temperature to 1 billion Kelvin, necessary to fuse carbon and oxygen to more massive elements, which eventually cause the supernova explosion. More than 95% of all stars end their lives as white dwarfs. They have masses below 1,4 the mass of the Sun, because at this limiting mass the speed of the electrons in their nucleus approach the speed of light, due to the small size, similar to that of the Earth.

We can only study the light coming stars, and the most detailed study is how the flux depends on wavelength, a spectrum. Analyzing the spectra of white dwarfs, we can see only lines of hydrogen in 80% of them. The remainder have spectra with helium lines. These stars have such huge gravity that the heavier elements sediment to the nucleus in only a few million years, and in general these stars are older than one billion years. This sedimentation, just like sand sediments in a glass of water, brings the lightest element to the surface. As hydrogen is lighter than helium and any other element, the spectrum of the large majority, 80%, shows only hydrogen lines. For those stars that we see only He lines, the hydrogen that existed in the star must have been lost, otherwise it would have floated to the surface.

In the paper Kepler, Koester & Ourique published in Science in April 2016, they report the result of a search through the 4.5 million spectra in the Data Release 12 the Sloan Digital Sky Survey (1), distributed along one third of the whole visible sky, for new white dwarf spectra. This Survey uses a 2.5 meter telescope in New Mexico to obtain photometry and spectra of most objects in the sky. They found the first star with an atmosphere dominated by oxygen, with traces of neon and magnesium. If the star had any lighter element than oxygen, i.e., H, He, Li, Be, C or N, they should have floated to the surface. If they do not, they must not be present in the star. It is the first star ever to be found any of the lighter elements.

Even more surprising was the discovery that the mass of the star was only 0.56 of the mass of the Sun. Single star stellar evolution predicts only white dwarf stars with masses above 1.05 times the mass of the Sun should have cores of O-Ne-Mg, but still with an atmosphere of He and H. The lower limit of 1.05 Msun comes from the temperature necessary to allow thermonuclear reactions burning carbon, 1 billion Kelvin. Stars born with masses below 7 to 8 solar masses will not reach this temperature and should have a nucleus of carbon and oxygen.

Libert et al. (2) discovered the first white dwarf with strong oxygen lines, Gänsicke et al. (3) found one star with oxygen abundance stronger than carbon, but much lower than He, and Kepler et al. (4) found two other similar stars. Theoretically, stars with initial masses higher than 7 solar masses, but smaller than 10.6 solar masses, will reach sufficiently high core temperatures to proceed to carbon burning, and produce either oxygen-neon (ONe) core white dwarfs, or undergo a core-collapse supernova (SNII) via electron capture on the products of carbon burning (5, 6, 7). The exact outcome of stellar evolution in this mass range depends critically on the detailed understanding of the nuclear reaction rates involved, mass--loss, on the efficiency of convective mixing in the stellar core (e.g. 8), and metallicity (9). SDSS J124043.01+671034.68, on spectra with Plate-Modified Julian Date-Fiber 7120-56720-0894, is a white dwarf with spectra dominated by oxygen lines, but the first to have oxygen abundance larger than helium. The SDSS spectrum does not show any helium lines or hydrogen lines. Visible at first sight is a large number of OI and MgI, MgII lines, but nothing obviously else. It was therefore plausible to assume a helium-rich atmosphere with Teff < 13000 K, where the He lines would be weak or absent. A few experiments showed that this could not be the solution. The higher excited OI lines could not be fit well, but more importantly, the MgI/II ionization equilibrium was completely wrong. It indicated a substantially higher effective temperature of at least 20000K. This range was also supported by a fit to the SDSS and GALEX photometry, which demanded Teff ~ 20000K, even if no reddening was assumed. At this Teff, however, the absence of helium lines demanded that log (N(He)/N(O)) := [He/O] had to be smaller than -1. That is, this is the first star in the Universe, where oxygen is the most abundant element in the atmosphere. More detailed calculations resulted in the following: Teff < 23000K, from the OII lines at 3947A and 4170-4177A, which may be marginally visible in the observations, but are much too strong predicted at higher Teff. Teff > 22000K, from the MgI/II ionization equilibrium. Note that log g cannot be determined and was assumed first to be 8.00, as the large majority of white dwarfs have this surface gravity. Teff = 23000K, log g = 8.0 was used for further refining the abundances: [H/O] to be determined, [He/O] < -1.5, [C/O] < -2.3 , [Ne/O] = -1.7 (0.5) [Mg/O] = -1.8 (0.2), [Si/O] = -3.5 (0.3) , [Ca/O] < -5.3, and [Fe/O] < -3.2 The detections of Mg, O, Si, are secure. Ne detection is marginal; individual lines are not overwhelming, but the fairly large number coinciding with observed features in the spectrum gives confidence. We also calculated a sequence with log g=9 from 20000 to 28000K. Nothing significantly changes regarding the abundances or temperatures. However, the lines get too broad and we can definitely exclude such a large mass. It can be something between 8 and 9, but it looks more like the best fit would be below 8. Therefore, even though from theory we expected a high mass, the current models indicate a normal white dwarf mass, but it is an oxygen atmosphere with mainly Mg traces.

Technical details
Our understanding of stellar evolution is good for stars that evolve without interacting with other stars, i.e, when any companion, if it exists, is farther away then several time the maximum radius the star will reach when it becomes a supergiant. The evolution is hampered, or modified, by any other star at any distance below that, and the evolution will depend on the real distance between the stars.

All stars loose some mass from the outer layers because they are so hot the photons (light) drags matter with them, i.e. by radiation pressure. In white dwarf stars, the lightest elements diffuse outwards and the heavier elements sediments to the core due to the high gravity (10 000x higher than in the Sun). At the same time, the lightest the element, the higher its velocity for the same energy. Even in the planets, Earth has no free hydrogen or helium in the atmosphere because it does not have a mass large enough to prevent these light gasses from flying away.

There are two types of supernova, The ones from stars born with masses above 10 solar masses, that do not become white dwarfs but explode as supernova type II when their nucleus reach Fe (iron). The other are supernova type Ia, when a companion star throws mass into a white dwarf, when this close-by companion star expands to a giant or supergiant in its normal evolution. The accreted mass is heated by the white dwarf and if reaches high enough temperature it starts a nuclear reaction transforming H into He or He into C explosively, and throws the envelope out.

Te surprise was that O atmosphere stars do exist, when we thought before only H or He could dominate, and that the star has much less than 1 solar mass.

What's next for this kind of work? Search for a binary companion, get a more accurate distance and mass determination, and look for similar stars in the new data that will be available from SDSS and other surveys.

Even the Sun looses 1/(100 trillion) of its mass per year because of radiation pressure, in what is called the solar wind. The whole Universe is made of 90% H and 8% He, with at most 2% of all the other elements combined, starting with O and C. Therefore this star was made initially of H and He, which float to the surface of the star in about 1 million year when the star becomes a white dwarf and does not have nuclear reactions anymore. The fact that we do not see H, He or C means that when they floated to the surface they were lost. The light elements float to the surface in every star. What the companion might do is throw mass onto the white dwarf, which gets heated and explodes, and then all the envelope might be lost.

As all white dwarfs do not have any further nuclear reaction, it just cools down with time. So it will take several billion years for it to be a cool ball of oxygen. How did you make this discovery? Looking by eye through 300 000 spectra that we preselected from the 4,5 million spectra taken by Sloan.

First of all, one has to look over large data sets to find unexpected things. We cannot teach a computer to look for things we do know about. Second, binary stellar evolution is important, because at least half of the stars in the Universe are in binary or multiple systems. Third, binary star evolution depends not only on the masses of the two stars, but on the separation between them, and this can have any value.

Could a small asteroid colliding with a wd make such spectra? O, Mg, Si are the typical elements, but not Ne. And would an asteroid totally hide de H or He underlying atmosphere? No.

Em português: Essa foi a primeira anã branca observada com uma atmosfera dominada por oxigênio. A primeira e única até o momento. O que mais nos surpreendeu foi a massa muito abaixo de 1 massa solar, que é onde a teoria prevê que o núcleo deveria ser de O/Ne/Mg. Para as anãs brancas, o elemento mais leve vai para as camadas mais altas, por sedimentação. Portanto o fato da atmosfera ser de O, indica que não pode ter nem H, nem He, nem C, que são mais leves que o O. Então se nem o núcleo deveria ser de O para massas menores que 1 massa solar, muito menos a atmosfera. Provalmente esta estrela veio de um sistema binário (duas estrelas), que seus envelopes (atmosferas) interagiram e no final perderam massa. Precisamos calcular modelos que resultem em uma estrela de baixa massa e com envelope de O, o que nenhum modelo atual preve. A descoberta foi feita no meio do ano passado (2015), quando estavamos analisando os 4,5 milhões de espectros do Sloan Digital Sky Survey, procurando por novas anãs brancas. A importancia é que os modelos atuais não são adequados, pois não prevem um objeto como o observado. O núcleo desta estrela deve ser de oxigênio, com um pouco de Neônio e Magnésio, como observamos na atmosfera, mas pode ser de carbono e oxigenio.

Pelo fato de não termos encontrado nenhuma evidência de uma estrela normal próxima nas observações (no espectro), sim as duas devem ter se fundido. Mas em geral a interação pode ocorrer com as duas estrelas dentro de um envelope (atmosfera) comum (uma atmosfera grande com os dois núcleos dentro) e quando esta atmosfera é dissipada (perdida), sobrem ainda duas estrelas, com menor massa do que iniciaram.

Aqui na UFRGS também calculamos modelos de evolução de estrelas, com a Prof. Alejandra Daniela Romero. Até agora não calculamos modelos de estrelas que interagem, mas se nossos colaboradores internacionais não o fizerem (provavelmente farão), nós teremos que iniciar estes cálculos.

A white dwarf with an oxygen atmosphere, Science, 1 April 2016, Vol. 352, No. 6281

S.O. Kepler1, Detlev Koester1,2, Gustavo Ourique1

1Departamento de Astronomia, Instituto de Física, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS – Brazil. 2Institut für Theoretische Physik und Astrophysik, Universität Kiel, 24098 Kiel, Germany.

S.O. Kepler¹, Detlev Koester^1,2, Gustavo Ourique¹

¹Departamento de Astronomia, Instituto de Física, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS – Brazil.
²Institut für Theoretische Physik und Astrophysik, Universität Kiel, 24098 Kiel, Germany.