The gas between stars is a major baryonic component of galaxies, and exists in a variety of physical states at any one time. TheISMis too rarefied to achieve thermal equilibrium with the stellar radiation which pervades it, yet it is this radiation - particularly high energy photons from massive stars - that dictates the energy of the gas. Temperatures in the ISM vary from ∼ 10–20 K in dense molecular clouds (∼1000 cm−3) shielded from stellar UV radiation, to 104−6K in the diffuse ISM (∼10–0.01 cm−3) surrounding massive
stars. Regions of the ISM where Lyman continuum photons have ionised hydrogen - and partially ionised other elements such as carbon, nitrogen and oxygen - are known as Hii–regions. The most famous example is the Orion nebula, which is powered almost entirely by the 4 O-stars of the Trapezium cluster. The resultant emission nebulae display permitted lines of neutral hydrogen and neutral and ionised helium, as well as forbidden emission lines of C, N and O. The radii of Hii-regions are well approximated by the Str¨omgren radius, RS, given by
RS = ( 3 4π S? n2β) 1/3, (1.5)
where n is the density of hydrogen atoms (nuclei), S? is the ionising flux (λ < 912˚A) and
β the recombination rate.
The large bolometric luminosities of massive stars, log(LBol/L ) = 5–6, provide large
radiative momenta, LBol/c. The momentum flux of stellar winds M /v˙ ∞ amounts to
∼30% of this for typical O-stars (Conti et al., 2008); in terms of energy, the mechani- cal luminosity, 1/2 ˙M v2
∞, is only a few percent of the radiative luminosity. However, as
stellar winds intensify during post-MS phases, particularly WR, mechanical luminosi- ties may rise to comparable values, with significant implications for the kinematics of their circumstellar environments. In Figure 1.1 I show the evolution of radiative and mechanical luminosity for a model of a 60 M star (Langer et al.,1994). The integrated
kinetic energy transferred to the ISM over the course of a typical massive star lifetime (107yr) is 1051erg - comparable to that of a typical supernova explosion (Garcia-Segura et al. 1996a, Garcia-Segura et al. 1996b). Therefore, O-stars tend to evacuate their surroundings of gas, before evolving away from theMS. Weaver et al. (1977) provide a
Figure 1.1: Evolution of mechanical (solid) and radiative luminosity (dotted) in a 60 M
stellar model (Langer et al., 1994). Taken from Garcia-Segura et al. (1996a).
mathematical description of these so-called wind-blown bubbles, and predict a structure - driven by gas pressure - consisting of four main parts: the cool expanding wind, hot shocked wind, cool swept-up ISM, and ambient ISM. Typical bubble radii are ∼10 pc, with expansion velocities ∼10 kms−3.
Such wind blown bubbles around O-stars are often unobservable, as they cannot produce enough shock compression to sufficiently raise the density of swept-up mate- rial. This changes however when the star evolves into a supergiant phase. Higher mass (& 60 M ) O-stars are expected to remain on the blue side of the H-R diagram, and
experience episodic mass-loss as they approach their Eddington luminosities, resembling a LBV phase (Langer et al. 1994, see Section 1.2.1). Lower mass O-stars will become
RSG stars, characterised by mass-loss rates ∼10−4M /yr (de Jager et al.,1988) several
orders of magnitude higher than O-stars, ejected at much lower speeds (∼ 10 kms−1). As stars in both mass ranges enter aWRphase, wind speeds recover (103kms−1) and carry
with them large amounts of mass (∼10−5M /yr) (Nugis & Lamers,2000). The complex
interplay between WR star winds and slower moving material ejected during previous phases produces a plethora of nebula morphologies. Garcia-Segura et al.(1996b) present hydrodynamical simulations of a circumstellar environment of a 60 M star, predicting
visible nebula during late LBV and early WR times, containing ∼10 M of material
with CNO abundances consistent with nuclear burning (N enriched, C & O depleted). Similarly, Garcia-Segura et al. (1996a) predict a visible nebula with quite a different morphology for a post-RSG WR star, containing similar amounts of mass, albeit with
Figure 1.2: Image of the Crescent nebula, NGC 6888, taken with the Wide Field Camera on the INT. Colour composite of narrow-band filters selecting emission from H-α and Oiii, coded in the image as red, green (25% H-α and 75% Oiii) and blue. The central
WR star (WR 136), can be seen as the bright point-source near the centre. Credit: Daniel Lopez.
lower N enrichment. Johnson & Hogg(1965) were the first to detect nebulae aroundWR
stars, and many have been associated with a variety of massive stars since (e.g.,Wachter et al. 2010). MostWR nebulae consist of ejected rather than swept-up material. Stock & Barlow (2010) estimate 5–6% of Galactic WR stars have such nebulae, a prominent example of which is NGC 6888, shown in Figure 1.2.
It is therefore clear that massive stars are capable of imparting enormous amounts of radiative and kinetic energy into theISM, even before their death as SN. Massive stars are thought to be important sources of α-elements (CNO) in the early universe (Henry et al.,2000), and pollution of natal material by massive stars has been invoked to explain recently discovered multiple populations in globular clusters (Salaris & Cassisi, 2014). The role of WR stars as significant polluters of the ISM with nitrogen is supported by observations of N-enriched ejecta nebulae around some GalacticWR stars (Stock et al.,
2011), and enhanced N and He abundances measured in star-forming galaxies showing
WR features in their integrated spectra (Pagel et al., 1986; Brinchmann et al., 2008;
L´opez-S´anchez & Esteban, 2010). It is less clear whether evolved massive stars are dominant producers of C and O, as these are more challenging to measure in the gas
phase of ejecta nebulae (Stock et al., 2011) or by analysis of stellar photospheres (see Section1.3.1). Competing sources of carbon are Type iiSNe (Woosley & Weaver,1995) and dredge-up in Asymptotic Giant Branch (AGB) stars (Frost & Lattanzio,1996). From synthetic populations of high and intermediate-mass stars, Dray et al. (2003) conclude that massive stars are comparable to intermediate-mass stars in their C production at solar metallicity, whereas O enrichment is dominated by SNe. Efforts are underway to detect the chemical imprints of the first generation of massive stars (Population III) in low-mass stars with ages comparable to that of the Universe (e.g.,Chiappini et al. 2011). As well as enriching theISM, energetic feedback from massive stars may potentially play a role in regulating star formation. Given the clustered nature of star formation, and apparent universality of the IMF, most clusters will contain at least a few reasonably massive stars. Most stars will therefore form in the vicinity of a massive star, and hence be subject to their strong feedback effects. Hii-regions act as significant sources of energy for turbulence in Giant Molecular Clouds (GMCs) (Matzner, 2002), preventing further collapse of the cloud into protostars. O-stars are the principal agents of GMCdispersal (Williams & McKee, 1997), and it has been shown theoretically that radiation pressure from massive stars can evacuate young clusters of natal gas (Boily & Kroupa,2003;Dale et al., 2005), bringing accretion onto protostars to a halt. Not all massive star feedback is necessarily negative. Momentum injected into the ISM through stellar winds and SN
explosions is a key component of the proposed ‘collect and collapse’ scenario (Elmegreen,
1998), in which further massive star formation is induced in swept-up material. Recent evidence for triggered star formation in the Milky Way was provided by the Milky Way Project (Kendrew et al.,2012).
Any process which affects star formation will also have consequences for galaxy for- mation and evolution, as galaxies build up their baryonic mass by converting gas into stars. The Kennicutt-Schmidt (KS) relation relates the surface gas density of a Galaxy to its star formation rate density (Kennicutt, 1998), and implies that star formation occurs on timescales of ∼50 dynamical times. However, it has been shown that without stellar feedback, gas inside galaxies is predicted to cool too efficiently, and forms stars on shorter timescales than this (Bournaud et al., 2010; Hopkins et al., 2011; Tasker, 2011;
Dobbs et al., 2011). Consequently, in cosmological simulations, star formation peaks at higher redshifts where galaxies are gas-rich (e.g., Springel & Hernquist 2003), and modern-day stellar masses become higher than observed (White & Frenk, 1991; Katz et al.,1996;Cole et al.,2000), providing especially poor fits to the low end of the stellar mass function. Furthermore, observational evidence of galactic-scale gaseous outflows
in starburst galaxies is abundant (Heckman et al., 2000; Martin, 2005; Veilleux et al.,
2005), showing that massive stars play a key role in regulating star formation, both through altering the physical conditions and outright quantity of gas available. Hopkins et al. (2014) show that it is necessary to account for massive star feedback in all forms (radiation pressure, wind momentum, SNe) if the KS relation and galactic winds are to be reproduced in simulations of galaxies over a broad range in halo mass. Another ob- servational canon of galaxy evolution, the Tully-Fischer relation, is also better predicted whenSNe feedback is accounted for (de Rossi et al., 2010).