Untangling the signals: Investigating accretion and photometric variability in young stars
Sergison, Darryl James
Thesis or dissertation
University of Exeter
In this thesis, an assessment is made of the value of optical CMDs as a useful diagnostic of the accretion properties of young stars. An analysis has been made of the phenomena that we observe and their effect on the position of stars in the CMD. Limitations and potential biases have been identified and evaluated. Variability causes some luminosity spread at a given colour in optical CMDs. A detailed characterisation of variability has been performed which places strong constraints on the magnitudes and the timescales on which the variability is seen. On timescales 15 minutes, almost no variability is detected (at levels greater than ≈ 0.2%) in the i band for a sample of ≈ 700 disc-bearing young stellar objects (YSOs). This suggests that the variability predicted by some accretion shock models is either very weak or not present. On hours to days timescales the optical variability in most stars is well described by a simple power law. The amplitude of the variability, a ∝ f−k, where f is the frequency of the variability in days. Disc-bearing and discless YSOs exhibit median values of k of 0.85 ± 0.02 and 0.95 ± 0.03 respectively, the uncertainity being the error on the median. The power law is valid up to a certain timescale (tmax) at which point the variability amplitude does not increase any further. tmax is found to be 1.50 ± 0.07 days and 1.41 ± 0.10 days for disc-bearing and discless stars respectively. Disc-bearing stars show greater variability amplitudes than the discless stars. However, it is notable that the variability timescale and power spectrum exponent are remarkably similar. This implies that the amplitude of the variability is driven by the physics of the underlying process, but that the timescales are instead driven by geometric effects. For disc-bearing stars, the highest amplitude variables are the accreting stars, which often appear to vary in the CMD along lines that correspond to changes in accretion luminosity. Four disc-bearing stars (approximately 0.5% of the disc-bearing sample) in Cep OB3b show extreme variability on timescales of years. Three (possible EXor candidates), show long-timescale changes that have a dramatic effect on their CMD position. However their small numbers mean that the overall impact on the CMDs of young associations is small. Variability on timescales of the rotational period and shorter adds uncertainty to age estimates of individual stars that are calculated by comparison with PMS models. Having provided a detailed description of variability and its impact on the CMD, it is clear that there are further significant mechanisms that affect the positions of YSOs in the CMD. I show that the spread in luminosity seen in the Orion Nebula Cluster and NGC 2264 could not be explained by accretion at rates of M ̇ ≥ 5 × 10−4 M⊙ yr−1 occurring within the protostellar phase of YSO evolution. Thus it appears that CMDs are not a useful diagnostic for study of the accretion histories of YSOs. The wavelength dependence of the extinction by dust within the inner regions of YSO discs is shown to differ from that seen in the ISM. Typically the wavelength dependence of the extinction is given by RV ≈5-8, compared with the value of RV ≈3.1 typical of the ISM. The interpretation is that grain growth has occurred. The location of this material within the ‘snow line’ implies that grains have coalesced rather than simply gaining an ice mantle. This is evidence for the beginning of planet formation. The effect of the high value of RV on the CMD is to add additional uncertainty of 0.1 mag to photometric measurements that have been corrected for the effects of extinction. Accretion luminosity is shown to be the dominant signal in the luminosity spread seen in CMDs of young associations. Stars which exhibit excess flux in the U band or Hα are displaced in CMD space. The accretion vector is shown to be a significant blueward shift in colour accompa- nied by a modest brightening in the g, g − i CMD. Accretion results in a luminosity spread as stars are displaced blueward below the PMS locus. This effect is not seen in non-accreting disc-bearing stars. Examination of the underlying excess luminosity spectrum for 15 accreting stars shows that the colour of the emission excess is not consistent across the sample. Thus to quantify the effect of accretion luminosity on CMD positions for individual stars, moderate resolution spectra are required with a large range in wavelength. This accretion luminosity may systematically bias estimates of PMS ages. A simple mitigation is to exclude accreting stars from age analysis. U band and Hα flux excesses are shown to vary independently by ≈ 1 dex on timescales shorter than the rotation period of the star. The relation between U band flux excess and veiling at 7000Å also appears to be variable. This implies that single epoch measurements of these parameters will add an uncertainty of ≈ 1 dex on accretion rates derived from them. Accretion rates derived from either U or Hα excess should be calculated from a mean of several photometric measurements, separated by significant fractions of the rotation period of the star. In most stars, the veiling at 7000Å is shown not to be a good measure for the calculation of the accretion rate. Despite providing a detailed characterisation of phenomena that influence the positions of YSOs in the CMD, there exists some residual luminosity spread at a given Teff that cannot be explained by variability on any timescale, extinction uncertainties or accretion luminosity. This residual spread should provide an opportunity to study an as-yet uncharacterised aspect of young stars.
Science and Technology Facilities Council
PhD in Physics