Summary and conclusions

Milky Way, the inter-stellar polarization (ISP) measured along the sightlines to highly reddened stars shows a characteristic wavelength dependency, known as the Serkowski law (Serkowski et al., 1975).

The wavelength of the ISP peak (λmax) depends on the dust grain size distribution, in the sense that for an enhanced abundance of small dust grains,λ_max moves to shorter wavelengths, whereas for an enhanced abundance of large dust grains it shifts longer wavelengths. In general, Milky Way stars have λmax ∼0.55µ. As shown by Patat et al. (2015), highly reddened SNe Ia with low total-to-selective extinction ratiosR_V, display peculiar continuum polarization wavelength dependencies, steeply rising towards the blue, with polarization peaks at short wavelengths (λ_max ≤0.4µm), definitely different from what is observed in the Galaxy. It is not clear why SNe Ia sight lines display such peculiar polarization profiles. Possible explanations are that the composition of interstellar dust in SNe Ia host galaxies is different from the dust in our Galaxy, or that there is circumstellar dust, ejected from the progenitor system before the explosion, characterised by an enhanced abundance of small grains.

Scattering from CSM dust is also a possibility. This conundrum is addressed in the thesis with two different approaches, discussed in Chapters 3 and 4.

In Chapter 3, we investigated the linear polarization of 17 sightlines to Galactic stars with anomalous extinction and lowRV values, selected from the Mazzei & Barbaro (2011) sample, in order to identify possible similarities to SNe Ia. We found that, despite their anomalous extinction curves, they obey to normal polarization laws (with a meanλ_max ∼0.53µm). This can be explained in the light of the following considerations. Not all dust that contributes to extinction also contributes to polarization:

polarization mainly depends on the dust grain size distribution of silicates, because grain alignment is more efficient for silicates than, for instance, for carbonaceous dust grains (Somerville et al., 1994).

On the contrary,RV is strongly dependent on carbonaceous grains too. We also found that there is no significantR_V−λ_maxrelation in our sample (Fig. 3.8). Theλ_max values in our data set are higher than what is typical for normal stars that follow the empiricalR_V−λ_max relationship described, for instance, by Whittet & van Breda (1978). However, by comparing theRV values of our sample with those published by Wegner (2002) for a subset of our stars, we find some differences. These are probably due to a different spectral classification and/or luminosity class adopted to derive their extinction curves (see Sect. 3.7.2). Theλmax value that we measure and the deviation from the empiricalRV −λmax

relationship may also suggest a spectral misclassification of some stars by Savage et al. (1985). The Serkowski parametersK andλ_max are known to be correlated. However, using our sample we find a slope that is steeper than in the classical empirical relationship presented by Whittet et al. (1992).

For better interpreting our results, we adopted a simple dust model that can reproduce the observed sightlines with lowR_V values and normal polarization curves. The simulations show that, to reproduce a polarization curve with the normalλ_max and lowRV, there must be a population of large interstellar silicate grains of sizea ≥0.1µm. Moreover, variations in grain alignment and size distribution together are required to reproduce the variation inλ_max for a fixed, low,R_V value. However, a change in grain alignment has a greater effect. Interestingly, theK−λ_max relation appears to be an intrinsic property of dust polarization. The numerical calculations show that for a fixedRV , the grain alignment function becomes narrower (broader) for a lower (higher) value ofλ_max andK (see Sect.. 3.7.5). An increase in the Serkowski parameterK and a deviation from the standard value in theK −λ_max plane can be reproduced by a decreasing contribution of large Si grains (Fig. 3.10).

On the other hand, in Chapter 4 we show that some post-AGB stars (proto-planetary nebulae, PPNe), which also may play an important role in the evolutionary path of some SNe Ia (Jones & Boffin, 2017), have polarization curves that are remarkably similar to those observed in highly reddened SNe Ia. It is a well known fact that, in PPNe, the curves are produced by CS scattering (Oppenheimer et al., 2005).

Thus, we suggest that the polarization curves seen in some SN Ia may be produced by CSM dust

129 scattering (Cikota et al., 2017c). No other sight lines to any normal, known Galactic stars have similar polarization curves: this opens the intriguing possibility that also in the case of SNe Ia scattering may be playing an important role. Furthermore, we speculate that those SNe Ia might have exploded within a PPN, and may provide observational support for the core-degenerate progenitor scenario (Kashi &

Soker, 2011), in which a white dwarf merges with the core of a companion AGB star. However, as first pointed out by Patat et al. (2015), the observed alignment along the local magnetic field which characterizes the polarization angle of SNe Ia (see Zelaya et al., 2017b) still needs to be reconciled with the random alignment expected for PPN.

We aim to undertake imaging polarimetry of∼100 SNe Ia using FORS2 (Proposal ID: 0101.D-0190, PI: Cikota) within the next∼2-3 years to create a statistical sample, with the aim to investigate whether there is a relation between the degree of polarization with galactocentric distance, and therewith answer the question whether the steeply rising polarization curves are produced by host galaxy interstellar dust, or by CSM dust. For instance, high polarization degrees along sightlines towards SNe Ia located far away from the centers of their host galaxies, where we expect low interstellar dust amounts, or in elliptical galaxies, which are known to be dust poor (see Cikota et al., 2016), would provide strong supportive evidence for our hypothesis. On the other hand, if we do not observe high polarization values of SNe Ia in dust poor regions, and elliptical galaxies, that would be evidence against our PPNe hypothesis.

In Chapter 5, we study the ejecta asymetries by examining the polarization of the Si II lines in SNe Ia. We reduced and examined archival spectropolarimetric data of a sample of 35 SNe Ia in a homogeneous way. All the spectra were obtained with FORS1 and FORS2 between 2001 and 2015, at 128 epochs in total. The linear polarization of the prominent Si IIλ6355Å line displays an evolution in time with a variety of peak polarization degrees that range from 0.1% to 1.7%. Maximum polarization is attained at different epochs relative to peak brightness, ranging from−10 to 0 days (Fig. 5.9), at variance with what was assumed in previous studies based on smaller samples (Wang et al., 2007). We populated the∆m15-PSiII plane and analyzed the relationship first determined by Wang et al. (2007).

Although the overall behaviour is confirmed by our larger sample, the scatter around the best-fit relation is larger. We show that subluminous and transitional objects display lower polarization values, and are located below the∆m15-PSiII relationship (Fig. 5.10), likely indicating a different explosion mechanism.

We found a statistically significant linear relationship (ρ ∼0.8) between the degree of linear polarization of Si II line before maximum with the Si II line velocity (Fig. 5.11). We suggest that this relationship, presented here for the first time, along with the∆m15-PSiIIrelationship (Wang et al., 2007) is consistent with the delayed-detonation model. Furthermore, we investigate the evolution of the Si II line in theQ−U plane for a subsample of SNe, which have been observed at multiple epochs, and run simple simulations to explore the effect of clumps on the polarization spectra. In the cases of SN 2005df, SN 2006X, and SN 2002bo, we observe the formation of loops, growth, and shrinking of the loops (see all plots in Appendix C.2), which may be explained by the evolution of the projected silicon ejecta size, from large axisymmetric structures, to large clumps, and finally to small clumps, as the time evolves and the photosphere recedes into the ejecta. Finally, we compared our Si II polarization measurements to numerical predictions for the double-detonation, delayed detonation and violent merger models calculated by Bulla et al. (2016a,b). Our observations are consistent with the predictions for the delayed-detonation and double-detonation models, which have a comparable degree of polarization, while only SN 2004dt has a degree of polarization that is compatible with the predictions for the violent-merger model. We also try to reproduce the Si II polarization-velocity relationship using the simulations (Bulla et al., 2016a,b). Although the calculations show a

velocity-polarization trend, the velocity range produced by the simulations is not sufficient to reliably match the observed relationship. This will be an important test for future, more sophisticated spectral modelling calculations.

Superluminous supernovae (SLSN) are an important class of objects that are∼10 times more luminous than common SNe. SLSN are divided into H-rich (type-II) and H-poor (type-I) subtypes.

While it is believed that SLSNe-II are explosions of massive stars that occur within a thick hydrogen envelope (Gal-Yam, 2012), the hydrogen-poor SLSNe-I that have a quite featureless early spectrum and are poorly understood. Understanding their nature is fundamental, because they may represent rare examples of dying supermassive stars in low-metallicity environments, similar to the conditions that most likely characterized the early Universe. A possible scenario that might explain the observed luminosities is one in which SLSNe-I are powered by an exotic internal engine, such as a magnetar or an accreting black hole. Since strong magnetic fields or collimated jets can circularly polarize light, polarimetry may provide the diagnostics required to distinguish between the two proposed mechanisms.

For this purpose, in Chapter 6 we investigate for the first time circular polarization of two H-poor superluminous supernovae, with the final aim of testing the magnetar scenario. The two objects, OGLE16dmu and PS17bek, were discovered during the course of the thesis work. OGLE16dmu is a slowly evolving H-poor SLSN, for which we obtained circular imaging polarimetry with FORS2 at 101.3 days past peak brightness, whilst PS17bek is a fast evolving SLSN-I, which we observed at 4.0 days before maximum light. In both cases, we did not find any evidence (within the noise) of a circular polarization signal. Nevertheless, we can not rule out the magnetar scenario purely based on the non-detection of circular polarization, because the strong magnetic field drops as 1/r³(wherer is the distance), and thus the light will be significantly circularly polarized by magneto-emissivity only very close (∼5000 km) to the surface of the magnetar. In other words, while the positive detection of a magnetic field would have strongly supported the magnetar scenario, the non-detection of a polarization signal is not sufficient to rule it out.

The main difficulty of investigating SLSNe with polarimetry is that they are typically very distant, and thus very faint (which makes spectropolarimetry prohibitively expensive in terms of telescope time). This is the reason why we considered an alternative approach to investigate SLSN progenitors, by studying their host galaxy environments. One hallmark of very massive progenitors would be a tendency to explode in very dense, UV-bright and blue regions of the parent galaxy. In Chapter 7 we investigate the spatially resolved host galaxy properties for two nearby H-poor SLSNe, PTF 12dam and PTF 11hrq, usingHubble Space Telescope(HST) and VLT/MUSE data. For this purpose, we studied the host galaxies of these two objects using imaging data obtained with HST in theF225W,F336Wand F625Wpassband filters, and compared the environment at the projected explosion position with the rest of the galaxy. Additionally, we obtained integral field spectroscopy with VLT/MUSE of the PTF 11hrq host galaxy. Both hosts show some evidence for interaction, and for PTF 11hrq this is possibly related to the SN explosion site. We speculate that the SLSN explosions may originate from stars generated in regions of recent or ongoing interaction. We demonstrate that the combination of high-resolution imaging and integral-field spectroscopy is very powerful for characterizing the explosion environment.

However, larger samples are needed to extract robust constraints on the progenitor population and on how their galaxy environments affect the star formation process.