The Lyman alpha (Lyα) emission line of atomic hydrogen (Hi) is intrinsically the most luminous spectral emission feature in astrophysical nebulae. It is produced by the spontaneous decay from the first excited state to the ground state (n = 2 → 1, where n is the principle quantum number), where the energy difference between the levels dictates a photon energy of 10.2 eV, or wavelength λ = 1215.67 Å. After the capture of an electron in ionized gas, the transition probabilities of the following radiative cascade are such that 68% of recombinations involve the production of a Lyα photon (Dijkstra, 2014). Thus if all ionizing photons had an energy slightly above the ionization edge of Hi, ≈ 50% of the total ionizing energy would be reprocessed into the Lyα line. Because the line may be intrinsically so luminous, Lyα was initially proposed as the spectral beacon by which to identify the first generations of primeval galaxies almost five decades ago (Partridge & Peebles, 1967).
Lyα plays a pivotal role in contemporary astrophysics, where it is used to identify high-redshift (z) star-forming galaxies by either narrowband filter observations or spectroscopic techniques, and frequently to confirm the redshift of candidate galaxies selected by other methods. Indeed the recognition that Lyα observations have attained makes the detection of Lyα a primary science goal for many new instruments on large telescopes: first light has recently been seen by HyperSuprime Cam (HSC, Takada, 2010) at the Subaru Telescope, the Multi Unit Spectroscopic Explorer (MUSE, Bacon et al., 2010) at ESO's Very Large Telescope, and the Cosmic Web Imager (CWI, Martin et al., 2010) at Palomar. Furthermore, the Hobby Eberly Dark Energy EXperiment (HETDEX, Hill et al., 2008) will find on the order of 106 Lyα galaxies, and Lyα detection is among the science goals for spectroscopic instruments on the James Webb Space Telescope (JWST) and all plans for Extremely Large Telescopes (ELTs).
The above introductory paragraphs on Lyα production are incomplete. While Lyα forms in astrophysical nebulae at the stated intensities, the transition is a resonant one, and Lyα is also absorbed by Hi in the same transition. After absorption to the 2P level (in the absence of electron collisions), there is is no alternative but for the electron to de-excite through Lyα. The optical depth of Hi, as seen in the core of the absorption line, is given by τ0 ≈ 3 × 10−14 √104 / T · NHI, where T is the temperature of the gas and NHi the column density in cm−2 (Verhamme et al., 2006). Thus at the limiting temperatures to which hydrogen can remain neutral, Hi becomes optically thick to Lyα at NHI ≈ 3 × 1013 cm−2. Assuming a number density of 1 atom per cubic cm, a cloud will exceed τ = 1 when its diameter exceeds 10−5 pc, or just 2 AU. Taking the Milky Way as an example, there are very few sightlines through which NHi drops below 1020 cm−2 (Kalberla et al., 2005), implying that Lyα would almost always see upwards of 106 optical depths.
The upshot is that in most galaxies, Lyα undergoes a radiative transfer process: photons scatter until they either escape from the galaxy or are absorbed by a dust grain, and dust extinction is also strongest in the far UV. This transfer may be thought of as a diffusion-like process, where photons take random walks in both physical and frequency space (Osterbrock, 1962). The path taken by Lyα is entirely regulated by the distribution of Hi that it encounters and must traverse, which in turn determines the likelihood that Lyα will encounter dust. Fortunately Lyα may see a significantly lower optical depth if it is shifted in frequency or the Hi is moving; the former can occur either after many scattering events as it diffuses in frequency through the redistribution profile, or by scattering in Hi that is itself kinematically offset from the Hii media where the Lyα formed. Ultimately the emitted Lyα luminosity (also its EW and departure from intrinsic Lyα / Hα ratio of 8.7) will be a function of Hi distribution, gas kinematics, dust content, and galaxy viewing angle.
1.1. Key Applications of Lyman alpha
Lyα transfer makes it hard to interpret the observed Lyα flux and EW from an individual galaxy, because the escape fraction, fescLyα, is difficult to predict for given configuration. However the transfer process, and the sensitivity of Lyα to different ISM properties, is also one of the major advantages of the transition. I now outline some key applications.
• The Evolution of Galaxies. The fraction of galaxies with WLyα ≳ 20Å (the canonical definition of a Lyα-emitter, LAE), at absolute UV magnitudes brighter than –18, is just 5% in the nearby universe (Cowie et al., 2010), where LAEs are rare. However this fraction increases strongly with increasing redshift, to ∼25% at z ≈ 3 (Shapley et al., 2003) to over 50% at z = 6 − 6.5 (Stark et al., 2010, Curtis-Lake et al., 2012). fescLyα evolves even more strongly over the same redshift range (Hayes et al., 2011, Blanc et al., 2011). This monotonic evolution, that spans a factor of 100 in fescLyα, is a key result of many Lyα surveys but has no conclusive explanation. Dust and Hi covering have both been suggested, and the answer must indeed lie among the quantities mentioned above, or combinations thereof, in the co-evolving properties of stars, gas and dust.
• The Epoch of Reionization. Lyα emission offers a unique opportunity to study the ionization state of intergalactic medium (IGM) at an epoch where other methods – e.g. the Gunn-Peterson trough in QSO spectra and Thompson scattering of the cosmic microwave background (CMB) – are insensitive. The Lyα emitter fraction and fescLyα evolution discussed above reverses after z ∼ 6.5, decreasing to ≲ 20% (e.g. Pentericci et al., 2014) at z ∼ 7. Possible interpretations include an increase in the ionizing photon escape fraction (Dijkstra & Jeeson-Daniel, 2013, which is anyway needed for reionization), but also that and increasingly neutral IGM starts to absorb the Lyα produced by the galaxies themselves. Disentangling the scenarios requires more information and solid constraints on the reionization history and topology require much larger samples (e.g. Jensen et al., 2014), but these will become available in the coming years with HSC. Moreover, as the IGM becomes neutral, the damping wing of Hi Lyα absorption may begin to affect the profile shape of the Lyα line that is transmitted, giving Lyα another unique application.
• Galaxy Kinematics. Lyα scatters coherently in the restframe of the Hi atom, and at scattering events is shifted in frequency by the velocity of the scattering medium. Thus as Lyα may escape from galaxies because of frequency shifts, the kinematic structure of the atomic gas becomes imprinted onto the line. This manifests as both a redshift (for outflowing gas) of the centroid of the main emission peak (e.g. Hashimoto et al., 2013) and also as characteristic features that modify the shape of the line profile (e.g. Verhamme et al., 2008). While there are many probes of kinematics in astrophysics, nebular line kinematics exclusively traces the warm ionized medium. Lyα kinematics on the other hand is shaped by kinematic differences between Hii (production) and Hi (scattering) media. Moreover Lyα is intrinsically very bright, and can be seen redshifted from the most distant galaxies. This again provides unique insights into the evolution of the ISM of galaxies, which can only feasibly be done with Lyα.
• Atomic Gas Surrounding Galaxies. As well as modifying the line profile, scattering also changes the surface brightness profile of emitted Lyα (Steidel et al., 2011, Hayes et al., 2013). The mechanisms by which galaxies obtain the gas they need to fuel star formation is one of the most pressing issues in extragalactic science (e.g. Kereš et al., 2005, Dekel et al., 2009), and necessitates a knowledge of the Hi distribution outside of star-forming regions and into the circumgalactic medium (CGM). Probing this circumgalactic Hi is observationally very challenging. In principle it can be done by λ = 21 cm observations of Hi directly, but current telescopes cannot push such techniques beyond the very local universe. An alternative is to use absorption spectroscopy of background QSOs that pierce galaxy halos at different impact factors, enabling us to measure Hi temperatures, densities and kinematics (e.g. Lanzetta et al., 1995, Tumlinson et al., 2013, Danforth et al., 2014). Unfortunately appropriately bright QSOs are rare and thus studies, while rich with information, are limited to statistical studies of the average galaxy. A promising third method is to illuminate the circumgalactic Hi with Lyα produced in the central star-forming regions. Indeed Lyα is perfect for such an application, being both the brightest intrinsic emission line, and being resonant in precisely the medium we need in order to image the CGM.
1.2. Lyman alpha Observations of the Nearby Universe – This Review
Section 1.1 presents the main astrophysical applications of Lyα emission, both as a diagnostic of the galaxies themselves and the IGM. The key difficulties of observing high-z galaxies are that fluxes are low and high signal-to-noise data are hard to obtain, that spatial information is minimal or absent, and that important features are redshifted away from atmospheric transmission windows. In the local universe surface brightness is higher by a factor of (1 + z)3, and spatial sampling can become almost arbitrarily high. Moreover, only in the local universe can we assemble the complete set of multi-wavelength observations, including but not limited to, all the continuum bands that probe both hot and cold stars, emission lines that provide a wealth of intrinsic diagnostics of the nebulae in which Lyα forms, direct measurements of far infrared continuum for both hot and cold dust, direct Hi measurements at λ = 21 cm, X-ray observations of coronal gas, and many more.
Indeed the science objectives discussed in Section 1.1 can, with the exception of reionization, all be undertaken in the low-z universe. Here Lyα provides a unique suite of information about the ISM of galaxies that still cannot be extracted using other techniques. This makes Lyα an import observable to obtain in any thorough study of (particularly star-forming) local galaxies. Moreover the question can be inverted: when as much information on the dust and gas content (distribution, kinematics, etc) has been assembled, observations of Lyα can then be used to calibrate our understanding of the Lyα transport mechanisms, and the effects of dust, gas, star-formation evolutionary stage may all be disentangled. This is the way, for example, we will assemble the relevant knowledge to interpret the evolution of the Lyα fraction with redshift. In turn we will be able to calibrate Lyα for high-z galaxy surveys, which will soon deliver ∼ 1 million objects, by using local Lyα emitters as analogues – laboratories in which to dissect in detail the processes ongoing in high-z systems. Local Lyα observations will allow us to read kinematic information off the line profile and conversely to predict the flux, EW, and line profile shapes, precisely as needed to address topics such as cosmic reionization that really hinge upon knowing the spectral profile. Indeed this is one of the major legacies established by our ultraviolet satellites; the only difficulties are that such satellites are both expensive and competitive.
This review focuses mainly upon empirical studies of Lyα emission and absorption in star-forming galaxies in the local universe. Somewhat arbitrarily, I have defined the ‘local universe' to mean redshifts where space-based facilities are needed to observe Lyα. In principle this means z ≲ 1.7 or so, but the most distant samples discussed are at z ∼ 1, and thus we are considering roughly the latter half of cosmic time. Where appropriate I may concentrate upon what about galaxies teach us about Lyα, or about what Lyα teaches us about galaxies. The layout of the remainder is as follows:
• In Section 3 I present a brief history of Lyα observations in the local universe, which were ongoing at a time when the first generations of high-z searches were also beginning. This concerns the first vacuum UV observations of active galactic nuclei (AGN) and star-forming galaxies using low dispersion spectrographs.
• In Section 4 we discuss how the Hubble Space Telescope changed the landscape by providing high-resolution spectra that can resolve the Lyα feature and interstellar absorption lines, thereby probing atomic gas kinematics and covering.
• Section 5 is concerned with Lyα imaging observations, also from HST, that simultaneously resolve very fine structures and reveal large-scale, diffuse Lyα halos.
• In Section 6 I present a large number of key results from survey data, that aim to answer questions about how various globally measured properties influence Lyα emission and under what conditions Lyα can be expected to be bright.
These Sections 3–6 aim to establish empirically how we have arrived at the current state-of-the-art.
• In Section 7 I then synthesize all the observational data from the previous Sections, and introduce some more speculative discussion about how various processes fit together.
• Finally I do not present explicit conclusions, but close the review in Section 8 with a number of perspectives and pressing open questions. These are concerned future observations and uses of Lyα at both low- and intermediate-z, with a view to understanding galaxy formation.