Final Report Summary - BHMASS (Quasar black hole mass estimates: identifying systematic biases for improved mass scaling laws)
Quasars are powered by super-massive black holes in their centres, which gravitationally attracts and consumes material from a disk in its close vicinity. This process produces high-energy radiation. Away from this disk is gas that also orbits the black hole. This gas absorbs the central high-energy radiation and re-emits light at particular energies determined by the specific atoms in the gas, also known as emission lines. This line emission is Doppler broadened to several million km per hour by the fast orbital motions of the gas. The gas velocities and distance from the black hole are used to measure the black hole mass. These mass estimates are currently limited by the unknown velocity structure of the line gas. We have several projects to investigate and constrain better this velocity structure and to quantify the accuracies in measuring the pertinent velocities for mass estimates. Accurate black hole mass measurements have profound implications on our understanding of the early Universe plus galaxy formation and evolution.
Part 1: Improve mass estimates of actively feeding supermassive black holes (also called 'quasars')
- Improve the gas velocity characterisation for mass estimates in nearby and distant sources
In one study, with a PhD student I investigated if and how the Hbeta 4863A and CIV 1549A emission lines are affected by and depend on the rate at which the central black hole feeds on matter. We extracted a large amount of data from astronomical multiwavelength data archives for nearby galaxies with actively feeding black holes to address this issue. Using these data, we find that the line shapes of Hbeta and especially CIV do indeed change as the black hole feeding rate increases. The profile changes are most sensitive to the ultraviolet (UV) continuum emission. (A paper is in preparation.) This work motivated further investigations into which part of the line profile is most susceptible to changes in the radiation field. For the CIV line, we find the line peak is much less responsive to such changes. This is causing deviations and scatter in the mass estimates based on CIV and Hbeta, respectively. (Another paper is being prepared). In extension thereof, I now investigate ways to correct for this effect so to improve on the current black hole estimates based on the CIV line.
- Identify and quantify the systematics and biases in individual mass estimates
Based on analysis of observational data and of extensive simulations, my Master student and I found that in low-quality data the velocities measured from emission lines actually detected in distant sources may be strongly underestimated. This systematic bias is enhanced when the data contain absorption that cannot be detected due to the poor data quality. This is a very important finding because recent work based on low S/N data suggest that the very first quasars in the Universe tend to have light-weight black holes to power them. But this may be an artifact of the data quality. These particular black hole masses have strong implications for the predictions of how the first black holes and galaxies form and evolve early on and thus are important tracers of the physics that took place at the earliest epochs. We are in the process of finalising our analyses where-after we will prepare the results for publication.
In addition, I launched an extensive systematic study on how spectral quality affects our ability to measure the emission line widths, used for black hole mass estimates. A large database of survey data was used as well as high-quality spectral data of both nearby and distant sources with actively feeding super-massive black holes. From the survey data, with collaborators we also made composite spectra of quasars that are representative of the quasar population, yet span the full range of spectral properties of this populations. With my Master's student I analysed these data before and after degrading the original high quality data to several lower quality levels to investigate the effects thereof. Several factors are at play. There are different ways to characterise the line width and each are sensitive to different issues. The simple FWHM is sensitive to the noise in the peak. This can be compensated by a careful selection of the underlying continuum setting. But the FWHM is poorly defined for some line shapes, such as double-peaked profiles. The line dispersion is well defined in that case but is, sadly, very sensitive to noise or even weak contaminating line emission in the line wings. We find this is a problem even at high data quality levels and for these reasons we do not recommend use of this width parameter. The IPV measure is robust to noise and is less sensitive to line emission contamination. We are currently investigating how the IPV line width can most robustly be measured for spectra of different quality levels. As there is no consensus in the science community on how to measure the velocities used for mass estimates, our work to map the specific systematics and biases associated with each line width measure is very important. But it is equally important that we provide the community with a robust measurement approach. We plan to publish these results together later this year, such that we also provide a solution to a significant problem.
These results and the complex issues that have emerged of this programme have motivated us to host a workshop in Copenhagen in July 2012 on the topic of improving mass estimates of actively feeding super-massive black holes in external galaxies. Experts in various related fields are invited to help identify the issues and solutions to promote significant advances on this topic. All participants will work towards a solution at the workshop and beyond.
Part 2: Modelling the SDSS quasar catalog
The aim is to produce a large catalog of more accurate mass estimates (than currently possible) of the super-massive actively feeding distant black holes based on the Sloan Digital Sky Survey catalog of over 106 000 sources (called 'quasars'). This will be of tremendous use to the community for studies of black hole growth and accretion physics and of black hole feedback onto its surroundings.
To reach this goal requires in part that a more accurate mass estimator has been identified and calibrated (part 1) and, in part, that the entire catalog of over 106 000 quasar spectra has been successfully analysed through spectral decomposition analysis. The latter analysis requires a significant amount of data processing which takes months to perform once the correct analysis approach has been decided and implemented (i.e. once the software code has been written and successfully tested). From the start of this project, I have worked on developing further the software related to this spectral decomposition. However, approximately mid-way through the course of this project new developments related to the accuracy of spectral decompositions and the characterisations of the emission line profiles emerged out of, in part, the research in part 1 which made it necessary to temporarily place this part of the project on hold for now. Otherwise, this work would have to be redone once these issues are resolved. Some of the findings of part 1 are that the spectral decompositions, as previously employed, may not provide emission profile characterisations that are sufficiently robust when applied to low signal-to-noise data. Since survey data such as those contained in the Sloan Digital Sky Survey is often of lower quality, this will have an impact on the very data that we are processing. The current project focus is devoted to investigating this issue with the aim of finding a spectral decomposition approach that is the least affected by this effect. Once this goal is in sight, the new spectral decomposition approach will be implemented, the SDSS quasar catalog processing will resume, and the catalog will be published.
Other significant results:
- Quasars as cosmic distance indicators
This work was done with colleagues in Copenhagen and Australia. We found that the empirical relationship between the distance of the line emitting gas to the black hole and the nuclear power of quasars, used for mass estimates, is tight enough to be used to infer cosmological distances independent of other distance measures. This method has a very large potential for use by the science community as an independent distance check but also for constraints on models of dark energy that cannot be obtained by any other means. For this reason, the publication on this work has sparked some news media attention. While the journal paper presents a good case for the significant potential of this method, follow-up investigations are needed to settle and improve upon some observational details as very high precision and accuracy is needed for competitiveness. For example, using the available database on nearby galaxies with actively feeding black holes, we are establishing the detailed relationship between the luminosity of the UV radiation, key to generating the line emission, and the more commonly observed optical luminosity. As this is not a one-to-one relation it explains most of the current scatter in the radius - luminosity relation that is key to cosmic distance determinations. A scientific paper on the results is in preparation.
- Dust free quasars in the early universe
In collabouration with colleagues in the United States of America (USA) and Germany, we studied about 20 quasars that reside at an epoch where the Universe was only about 800 million years old when the light from these objects were emitted. These quasars are important as we get a glimpse of these special black hole systems when the Universe was quite young; these systems could be some of the first generation of accreting supermassive black holes appearing. We found a positive relation between the amount of hot dust detected in these systems and the mass of the central black hole that power these quasars. This suggests that the strength of the starformation in the host galaxy will increase with the build-up of the black hole at that early epoch. The quasars with little or no hot dust may be in a very early evolutionary phase of the host galaxy. These results were published in Nature and resulted in press releases both in Denmark and by National Aeronautics and Space Agency (NASA).
Part 1: Improve mass estimates of actively feeding supermassive black holes (also called 'quasars')
- Improve the gas velocity characterisation for mass estimates in nearby and distant sources
In one study, with a PhD student I investigated if and how the Hbeta 4863A and CIV 1549A emission lines are affected by and depend on the rate at which the central black hole feeds on matter. We extracted a large amount of data from astronomical multiwavelength data archives for nearby galaxies with actively feeding black holes to address this issue. Using these data, we find that the line shapes of Hbeta and especially CIV do indeed change as the black hole feeding rate increases. The profile changes are most sensitive to the ultraviolet (UV) continuum emission. (A paper is in preparation.) This work motivated further investigations into which part of the line profile is most susceptible to changes in the radiation field. For the CIV line, we find the line peak is much less responsive to such changes. This is causing deviations and scatter in the mass estimates based on CIV and Hbeta, respectively. (Another paper is being prepared). In extension thereof, I now investigate ways to correct for this effect so to improve on the current black hole estimates based on the CIV line.
- Identify and quantify the systematics and biases in individual mass estimates
Based on analysis of observational data and of extensive simulations, my Master student and I found that in low-quality data the velocities measured from emission lines actually detected in distant sources may be strongly underestimated. This systematic bias is enhanced when the data contain absorption that cannot be detected due to the poor data quality. This is a very important finding because recent work based on low S/N data suggest that the very first quasars in the Universe tend to have light-weight black holes to power them. But this may be an artifact of the data quality. These particular black hole masses have strong implications for the predictions of how the first black holes and galaxies form and evolve early on and thus are important tracers of the physics that took place at the earliest epochs. We are in the process of finalising our analyses where-after we will prepare the results for publication.
In addition, I launched an extensive systematic study on how spectral quality affects our ability to measure the emission line widths, used for black hole mass estimates. A large database of survey data was used as well as high-quality spectral data of both nearby and distant sources with actively feeding super-massive black holes. From the survey data, with collaborators we also made composite spectra of quasars that are representative of the quasar population, yet span the full range of spectral properties of this populations. With my Master's student I analysed these data before and after degrading the original high quality data to several lower quality levels to investigate the effects thereof. Several factors are at play. There are different ways to characterise the line width and each are sensitive to different issues. The simple FWHM is sensitive to the noise in the peak. This can be compensated by a careful selection of the underlying continuum setting. But the FWHM is poorly defined for some line shapes, such as double-peaked profiles. The line dispersion is well defined in that case but is, sadly, very sensitive to noise or even weak contaminating line emission in the line wings. We find this is a problem even at high data quality levels and for these reasons we do not recommend use of this width parameter. The IPV measure is robust to noise and is less sensitive to line emission contamination. We are currently investigating how the IPV line width can most robustly be measured for spectra of different quality levels. As there is no consensus in the science community on how to measure the velocities used for mass estimates, our work to map the specific systematics and biases associated with each line width measure is very important. But it is equally important that we provide the community with a robust measurement approach. We plan to publish these results together later this year, such that we also provide a solution to a significant problem.
These results and the complex issues that have emerged of this programme have motivated us to host a workshop in Copenhagen in July 2012 on the topic of improving mass estimates of actively feeding super-massive black holes in external galaxies. Experts in various related fields are invited to help identify the issues and solutions to promote significant advances on this topic. All participants will work towards a solution at the workshop and beyond.
Part 2: Modelling the SDSS quasar catalog
The aim is to produce a large catalog of more accurate mass estimates (than currently possible) of the super-massive actively feeding distant black holes based on the Sloan Digital Sky Survey catalog of over 106 000 sources (called 'quasars'). This will be of tremendous use to the community for studies of black hole growth and accretion physics and of black hole feedback onto its surroundings.
To reach this goal requires in part that a more accurate mass estimator has been identified and calibrated (part 1) and, in part, that the entire catalog of over 106 000 quasar spectra has been successfully analysed through spectral decomposition analysis. The latter analysis requires a significant amount of data processing which takes months to perform once the correct analysis approach has been decided and implemented (i.e. once the software code has been written and successfully tested). From the start of this project, I have worked on developing further the software related to this spectral decomposition. However, approximately mid-way through the course of this project new developments related to the accuracy of spectral decompositions and the characterisations of the emission line profiles emerged out of, in part, the research in part 1 which made it necessary to temporarily place this part of the project on hold for now. Otherwise, this work would have to be redone once these issues are resolved. Some of the findings of part 1 are that the spectral decompositions, as previously employed, may not provide emission profile characterisations that are sufficiently robust when applied to low signal-to-noise data. Since survey data such as those contained in the Sloan Digital Sky Survey is often of lower quality, this will have an impact on the very data that we are processing. The current project focus is devoted to investigating this issue with the aim of finding a spectral decomposition approach that is the least affected by this effect. Once this goal is in sight, the new spectral decomposition approach will be implemented, the SDSS quasar catalog processing will resume, and the catalog will be published.
Other significant results:
- Quasars as cosmic distance indicators
This work was done with colleagues in Copenhagen and Australia. We found that the empirical relationship between the distance of the line emitting gas to the black hole and the nuclear power of quasars, used for mass estimates, is tight enough to be used to infer cosmological distances independent of other distance measures. This method has a very large potential for use by the science community as an independent distance check but also for constraints on models of dark energy that cannot be obtained by any other means. For this reason, the publication on this work has sparked some news media attention. While the journal paper presents a good case for the significant potential of this method, follow-up investigations are needed to settle and improve upon some observational details as very high precision and accuracy is needed for competitiveness. For example, using the available database on nearby galaxies with actively feeding black holes, we are establishing the detailed relationship between the luminosity of the UV radiation, key to generating the line emission, and the more commonly observed optical luminosity. As this is not a one-to-one relation it explains most of the current scatter in the radius - luminosity relation that is key to cosmic distance determinations. A scientific paper on the results is in preparation.
- Dust free quasars in the early universe
In collabouration with colleagues in the United States of America (USA) and Germany, we studied about 20 quasars that reside at an epoch where the Universe was only about 800 million years old when the light from these objects were emitted. These quasars are important as we get a glimpse of these special black hole systems when the Universe was quite young; these systems could be some of the first generation of accreting supermassive black holes appearing. We found a positive relation between the amount of hot dust detected in these systems and the mass of the central black hole that power these quasars. This suggests that the strength of the starformation in the host galaxy will increase with the build-up of the black hole at that early epoch. The quasars with little or no hot dust may be in a very early evolutionary phase of the host galaxy. These results were published in Nature and resulted in press releases both in Denmark and by National Aeronautics and Space Agency (NASA).