The explosive deaths of stars, known as a supernovae (SNe), have been critical to our understanding of the Universe for centuries. From the first evidence of a changing Universe beyond the Moon (Brahe1573) to the first evidence of the accelerating expansion of the Universe (Riess et al. 1998; Perlmutter et al. 1999), SNe - and often a specific subclass of SNe called Type Ia SNe (SNe Ia) - have been integral to astronomical research. An introduction to SNe, their importance in astronomy, and how we observe them is given in Chapter 1. How SNe Ia explode, what progenitor systems give rise to them, and how different initial conditions affect the observed outcomes of these objects are understood only at a relatively basic level. In other words, a detailed understanding of the physics behind SNe Ia is still lacking. One way astronomers can begin to solve these problems, and others involving SNe Ia, is to obtain and analyze a large, self-consistent dataset of SN Ia observations. This is the goal of the Berkeley SN Ia Program (BSNIP) which comprises the majority of this Thesis.
In the second Chapter, I present the full BSNIP sample which consists of 1298 low-redshift (z ≤ 0.2) optical spectra of 582 SNe Ia observed from 1989 through the end of 2008. Many of the SNe have well-calibrated light curves with measured distances as well as spectra which have been corrected for host-galaxy contamination. Most of the data were obtained using the Kast double spectrograph mounted on the Shane 3 m telescope at Lick Observatory with typical wavelength coverage of 3300-10400 Å, which is significantly larger than that of most previously published SN Ia spectral datasets. I also present the BSNIP observing and reduction procedures used during the two decades over which the data were collected. In addition, I describe our spectral classification scheme (using the SuperNova IDentification code, SNID; Blondin & Tonry 2007), utilizing my newly constructed set of SNID spectral templates. These templates allow me to accurately spectroscopically classify the entire BSNIP dataset, and by doing so I am able to reclassify a handful of objects as bona fide SNe Ia and a few other objects as members of some of the peculiar SN Ia subtypes. In fact, the BSNIP dataset includes spectra of nearly 90 spectroscopically peculiar SNe Ia. I also present spectroscopic host-galaxy redshifts of some SNe Ia where these values were previously unknown. The sheer size of the BSNIP dataset and the consistency of the observation and reduction methods makes this sample unique among all other published SN Ia datasets and is complementary in many ways to the large, low-redshift SN Ia spectra presented by Matheson et al. 2008 and Blondin et al. 2011.
I present measurements of spectral features of 432 low-redshift (z < 0.1) optical spectra within 20 d of maximum brightness of 261 SNe Ia from the BSNIP sample in the third Chapter. I describe in detail my method of automated, robust spectral feature definition and measurement which expands upon similar previous studies. Using this procedure, I attempt to measure expansion velocities, (pseudo-)equivalent widths (pEWs), spectral feature depths, and fluxes at the center and endpoints of each of nine major spectral feature complexes. A sanity check of the consistency of the measurements is performed using the BSNIP data (as well as a separate spectral dataset). I investigate how velocity and pEW evolve with time and how they correlate with each other. Various spectral classification schemes are employed and quantitative spectral differences among the subclasses are investigated. Several ratios of pEW values are calculated and studied. Furthermore, SNe Ia that show strong evidence for interaction with circumstellar material or an aspherical explosion are found to have the largest near-maximum expansion velocities and pEWs, possibly linking extreme values of spectral observables with specific progenitor or explosion scenarios. A discussion of the relative merits of various classification schemes is presented and I find that purely spectroscopic classification schemes are useful in identifying the most peculiar SNe Ia. However, in almost all spectral parameters investigated the full sample of objects spans a nearly continuous range of values. Comparisons to previously published theoretical models of SNe Ia are made and some of the predictions of these models match the observations presented here. I conclude with a brief discussion of how these measurements and the possible correlations presented will be crucial to future SN surveys.
The fourth Chapter of this Thesis presents comparisons of spectral feature measurements to photometric properties of 115 low-redshift (z < 0.1) SNe Ia with optical spectra within 5 d of maximum brightness. The spectral data come from the BSNIP sample described in Chapter 2, and the photometric data come mainly from the Lick Observatory Supernova Search (LOSS) and are published by Ganeshalingam et al. (2010). The spectral measurements come from BSNIP II (Chapter 3 of this Thesis) and the light-curve fits and photometric parameters can be found in Ganeshalingam et al. (in preparation). A variety of previously proposed correlations between spectral and photometric parameters are investigated using the large and self-consistent BSNIP dataset. We also use a combination of light-curve parameters (specifically the SALT2 stretch and color parameters x1 and c) and spectral measurements to calculate distance moduli. The residuals from these models is then compared to the standard model which only uses light-curve stretch and color. The pEW of Si II λ4000 is found to be a good indicator of light-curve width and the pEWs of the Mg II and Fe II complexes are relatively good proxies for color. However, a distance model only using these spectroscopic measurements performs worse than the standard model which uses only light-curve parameters. When using a distance model which combines the ratio of fluxes near ~3600 Å and ~4300 Å with both x1 and c, the Hubble residuals are decreased by 12%, which is found to be significant at the 2.4σ level. The weighted root-mean square of the residuals using this model is 0.130 ± 0.019 mag (as compared to 0.146 ± 0.019 mag when using the same sample with the standard model). This Hubble diagram fit has one of the smallest scatters ever published and at the highest significance ever seen in such a study. Finally, these results are discussed with regard to how they can improve the cosmological accuracy of future, large-scale SN Ia surveys.
Finally, I conclude this Thesis with an in-depth study of a quite peculiar SN Ia, not included in the BSNIP sample. Chapter 5 presents and analyzes optical photometry and spectra of the extremely luminous and slowly evolving Type Ia SN 2009dc, and offers evidence that it is a super-Chandrasekhar mass (SC) SN Ia and thus had a SC white dwarf (WD) progenitor. Optical spectra of SN 2007if, a similar object, are also shown. SN 2009dc had one of the most slowly evolving light curves ever observed for a SN Ia, with a rise time of ~23 d and Δm15(B) = 0.72 mag. I calculate a lower limit to the peak bolometric luminosity of ~2.4×1043 erg s-1, though the actual value is likely almost 40% larger. Optical spectra of SNe 2009dc and 2007if obtained near maximum brightness exhibit strong C II features (indicative of a significant amount of unburned material), and the post-maximum spectra are dominated by iron-group elements. All of the spectra of SNe 2009dc and 2007if also show low expansion velocities. However, I see no strong evidence in SN 2009dc for a velocity "plateau" near maximum light like the one seen in SN 2007if (Scalzo et al. 2010). The high luminosity and low expansion velocities of SN 2009dc lead to a derived WD progenitor mass of more than 2 MSun and a 56Ni mass of about 1.4-1.7 MSun. I propose that the host galaxy of SN 2009dc underwent a gravitational interaction with a neighboring galaxy in the relatively recent past. This may have led to a sudden burst of star formation which could have produced the SC WD progenitor of SN 2009dc and likely turned the neighboring galaxy into a "post-starburst galaxy." No published model seems to match the extreme values observed in SN 2009dc, but simulations do show that such massive progenitors can exist (likely as a result of the merger of two WDs) and can possibly explode as SC SNe Ia.