This chapter describes spectral classification in general, without referring to a particular classification system.

We start by discussing certain specifications of the spectroscopic material. Then we describe the features of the spectra and the factors which may affect them. Next we consider the process of classification itself and finally we deal with certain aids to classification.

The material of spectral classification

A spectrum is the display of stellar radiation as a function of wavelength. The names of the different wavelength regions are given in table 2.1. Astronomers also talk of the ‘ultraviolet spectrum’ of a star; this is a short way of speaking of the ‘ultraviolet region of the spectrum’ of the star.

Once a spectrum has been obtained, certain details about it must be provided. In first place, what and how it was observed must be specified. Among the most important specifications is the wavelength range covered. Note that one should specify what the (usable) spectrum region is, rather than state that ‘103aO plates used’ because in a few years this will not be understandable. Next the plate factor (D) should be specified, which is the number of angstroms of the spectrum entering 1 mm length of receiver. Much confusion originates here because astronomers call D the ‘dispersion’, whereas opticians call it (correctly) the ‘reciprocal dispersion’. A small plate factor D implies highly dispersed spectra (i.e. large scale), and a large value of D, low dispersion spectra.

The second part of this book is devoted to the description of the different groups of stars which have been defined over the years. We shall start with the hottest stars and work toward the cooler ones, each chapter dealing with one type of object. The material is arranged broadly into ‘families’ (chapters), so that stars which are similar appear together.

Stars with peculiar spectra are dealt with in sections added to the relevant chapter. Since many groups are not limited to one spectral class, the place of a given group in the book is usually the one corresponding to the first appearance of the group. So, for example, Am stars are discussed in section 10.1 of the A-type stars, despite the fact that some Am stars are found among F-type objects.

As far as possible, each chapter has the following structure. First, we define each group, taking in as much history as we need, but without trying to write the whole history of the group. Then we describe the spectrum of the type of star, usually in the classical region, but, if information exists, also in the ultraviolet, infrared and radio regions. Rotation is also included here. Next we examine what photometry can tell us about the group; then we consider the absolute magnitude, whether any of the group are binaries and finally statistical properties like presence in clusters, distribution on the sky and the frequency of the stars of the group.

In this chapter we shall study two photometric systems in order to show in detail some of the uses of a photometric system. We have chosen the UBV and the uvby or Strömgren system. For other photometric systems the reader is advised to consult Golay (1974) or Straizys (1977).

The UBV system

The UBV system was developed in the fifties by Johnson (see Johnson and Morgan 1953) for the photometric study of stars classified in the Yerkes system.

It uses three wide passbands, each about a thousand angstroms wide, called U (ultraviolet), B (blue), and V (visual), with λ0 around λ3500, λ4300 and λ5500 respectively. The choice of the passbands was made in part for historical reasons. V corresponds approximately to the ‘visual magnitudes’ handed down essentially from Ptolemy. B corresponds on the average to the ‘photographic magnitudes’ of the end of the nineteenth century. Finally U was chosen so as to get as much ultraviolet light as possible. The functions S(λ) are tabulated in table 5.1, but readers should be warned that different authors use slightly different S(λ). The system is thus not suited for very high precision. Furthermore the S(λ) of the U color goes down to λ3000, whereas the atmosphere usually cuts off the spectrum below λ3300. The ultraviolet limit of the U band is therefore not fixed by the filter system, but differs from place to place (for instance with the elevation of the observatory) and sometimes from night to night.

Introduction

Proof of the existence of interstellar gas was first provided by observations of narrow absorption lines in the visible spectra of distant stars. In 1904, J. Hartmann identified lines of Ca II that did not share the periodic variations in Doppler shift exhibited by the principal stellar absorption lines in a spectroscopic binary star, and attributed these ‘stationary lines’ to foreground material outside the stellar system. Somewhat more than three decades later, the first interstellar molecules, CH, CH+, and CN, were discovered in similar ways. There is a similarly long history of related investigations in laboratory spectroscopy and in theoretical interpretation.

The interstellar absorption lines tend to be very narrow compared with the photospheric absorption features in the spectra of the hot stars used as background light sources. In terms of line broadening by thermal motions and frequent atomic collisions, this suggests low densities and low temperatures for the absorbing material. In most cases, the observed lines arise only in the lowest states of atoms and molecules, indicating also that the densities and temperatures are too low to maintain significant excited state populations. As we will see, it is possible to infer from such observations quite a lot of information about element abundances, temperatures, densities, cosmic-ray fluxes, and intensities of radiation inside particular clouds.

The term ‘diffuse interstellar cloud’ has no precise denotation and distinctions between different kinds of interstellar clouds – e.g., diffuse, dark, and giant molecular – are somewhat poorly defined.

Discovery

In Ptolemy's Almagest, six objects are listed as ‘stella nebulosa’, hazy, luminous spots on the Celestial Sphere. Once viewed telescopically, these six objects were resolved into clusters of stars; however, other nebulous objects were noticed. The first, in Orion, appears to have been discovered by Fabri de Peiresc in 1610. Two years later, Simon Marius noted a nebula in Andromeda that had also been recorded by Al Sufi in the tenth century. The discoveries continued and, in 1781, Charles Messier compiled a list of nebulae and star clusters. The 105 objects he catalogued are still identified by their Messier number; the Orion Nebula is M42 while the Andromeda Nebula is M31.

An all-sky survey carried out by the Herschels at the turn of the nineteenth century resulted in a General Catalogue containing over 5000 nebulous objects. In 1888 a New General Catalogue was published by J. L. E. Dreyer; later editions included Index Catalogues and tabulated more than 13000 objects. The Orion Nebula, M42, is also identified as NGC 1976; the Andromeda Nebula, M31, is NGC 224.

About the middle of the nineteenth century, Lord Rosse constructed a sixfoot reflector in Ireland with which he made numerous visual sketches of nebulae and applied names to them by which they are still referred. M97 (NGC 3587) is called the ‘Owl’, while M51 (NGC 5194) is known as the ‘Whirlpool’, indicative of its spiral form.

During the initial discovery period of nebulae, a debate was ongoing as to whether or not all nebulae could be resolved into stars if a telescope of sufficient light-gathering power and resolution were available.

Introduction

Stellar spectra differ widely. Lines of atomic hydrogen are strong in the spectrum of the bright star Sirius, relatively weak in the solar spectrum, while lines of iron and other metals are strong in the solar spectrum, weak or absent in the spectrum of Sirius. Do such differences in the relative strengths of spectral lines reflect differences in relative abundances? Can we conclude that the atmosphere of Sirius is made up largely of hydrogen while the solar atmosphere consists largely of metal atoms?

Since the early 1920s astronomers have understood that the most conspicuous differences between stellar spectra arise from differences in the temperature of the atmospheric layers where the spectral lines are formed, rather than from differences in the relative abundances of the chemical elements. To absorb light at the frequency of one of the Balmer lines, a hydrogen atom must be in its first excited state. The fraction of hydrogen atoms in this state depends on the temperature (and, much more weakly, on the pressure). At the temperature of the solar photosphere (the visible layer of the Sun's atmosphere), nearly all of the hydrogen atoms are in the ground state, where they can absorb lines of the ultraviolet Lyman series. The spectrum of Sirius is formed at a temperature of around 10000 K. A much larger (though still numerically small) fraction of the hydrogen atoms is in the first excited state at this temperature. At still higher temperatures hydrogen is almost fully ionized, so the relative population of the first excited state is again very low.

Introduction

The solar corona is the most thoroughly explored example of a hot, dilute, extended stellar atmosphere. Because we are so near the Sun, we can resolve and observe individual structures. Using atomic physics, we can develop and test spectroscopic methods and apply them to construct coronal models that accurately represent the run of temperature, density, and velocity.

The solar corona turns out to be a fascinating astrophysical environment. Beyond its own intrinsic interest, it is also important because it is typical of many other stellar atmospheres. X-ray detectors aboard the HEAO-2 satellite (‘Einstein’) have revealed soft X-ray emission from stars of every spectral type and luminosity class. Thus, the diagnostic techniques developed with solar observations may help us to understand the nature of stellar coronae in general.

Despite our advantage of being near the Sun, we have as yet very incomplete ideas about some of the basic coronal processes. For example, we are not yet certain how the corona is heated, or how it replenishes the material it loses to the solar wind. Better observations and improved diagnostic techniques will help us to answer these large questions.

In this chapter, I will focus on the spectroscopic diagnostic techniques that have been devised for the study of the solar corona (and which are applicable, in principle, to other coronae) and mention some of the results that have come to light. After briefly discussing monochromatic imaging of the corona, I will consider how coronal temperature density, velocity and magnetic field have been measured by spectroscopic techniques.

Introduction

The neutral interstellar gas in the Milky Way is largely confined to the galactic plane with a density stratification away from the plane that is approximately given by n(HI)≈n(HI)0e-|z|/h, with n(HI)0≈1.2 atoms cm-3 and h≈0.12 kpc. However, this simple exponential distribution does not adequately describe a very extended and highly ionized component of the interstellar gas that is commonly referred to as galactic halo gas or galactic corona gas. The density stratification of the halo gas is very uncertain but may have a scale height that exceeds by a factor of 30 that of the neutral disk gas.

The observational study of galactic halo gas is a very young field – the youngest of those subjects included in this volume. That a hot (104–106 K) and extended (z ≈10–30 kpc) gaseous halo surrounds the Milky Way was suggested by Shklovsky (1952) based on measurements of non-thermal radio emission from the galaxy and by Spitzer (1956) based on the apparent stability of high-latitude interstellar clouds. Spitzer noted that stars at z distances exceeding 0.5 kpc more frequently exhibit high velocity interstellar Ca II optical absorption lines than do stars at smaller z. He concluded that an appreciable fraction of the high-velocity clouds producing these absorption features must exist more than 0.5 kpc from the plane. The basic problem associated with a cloud at z > 0.5 kpc is its instability to outward expansion unless the cloud is in near pressure equilibrium with an external medium. The external medium was postulated to be a high-temperature low-density gas – the galactic corona.

Introduction

These words describe the enduring relationship between astronomy and atomic physics and, in particular, the wonderful unity of laboratory and astrophysical plasmas. In this chapter we will explore the laboratory part of this unity and will discuss the measurements of some parameters and processes which are important to astronomy and atomic physics.

The determination of chemical abundances and the calculation of model atmospheres for the Sun and stars have become increasingly sophisticated since the landmark work of Goldberg and colleagues. However, despite the impressive progress, calculations based on the best existing atomic data, chemical abundances, and model atmospheres do not reproduce the measured, high-resolution, ultraviolet, solar spectrum nor do they match the center-to-limb variations. When high-resolution, ultraviolet, solar spectra (see Fig. 12.1) are examined it is apparent that the complex of overlapping and nearby lines (‘line blanketing’) could be responsible for much of the discrepancy between observations and calculations.

Overview

A normal main-sequence or red giant star possesses a photosphere, an outer layer of the stellar atmosphere that generates the optical photons that we observe and which is gravitationally bound to the star. Physical conditions in the layer above the photosphere vary greatly among mainsequence stars of different mass and among giant stars, many of which undergo large amplitude photospheric pulsations. Many stars possess chromospheres and, perhaps, also coronae. Mass motions are complicated with some matter falling back onto the photosphere and some being ejected to infinity. It is somewhere in this unsettled region that the outflowing circumstellar shell (CS) begins. For most stars, the circumstellar matter is at least partially transparent and the photospheric spectrum may be observed at wavelengths characteristic of the photospheric temperature. For a few stars with very large mass loss rates, such as IRC + 10216 and CRL 3068, the CS is sufficiently opaque that the photosphere is largely invisible from near infrared to ultraviolet wavelengths. For these stars we see a false photosphere at infrared wavelengths. That is, dust grains that have formed in the outflowing circumstellar gas have sufficient opacity to absorb essentially all of the true photospheric radiation. We then see a cool, roughly blackbody, emission spectrum characteristic of the temperature of dust grains in the inner portions of the CS.

The nature of astrophysical shocks

Energetic phenomena are common in the Universe: Stars inject matter into the interstellar medium via high-velocity winds at their birth and during their lifetime, and massive stars explode as supernovae when they die. Observations of active galaxies reveal rapidly expanding radio sources and jets which appear to be blasting into the ambient medium. On yet larger scales, over-dense regions of the Universe draw matter in at high velocity by gravitational attraction, whereas under-dense regions can expand at high velocity into their surroundings. In many of these cases, the sound speed of the ambient medium is less than the velocity of the gas expanding into it, especially if radiative cooling is efficient. In this case there is no way for the ambient medium to respond to the energy injection in a smooth way. Instead a near discontinuity is produced, a shock, which suddenly accelerates, heats, and compresses the ambient gas. Shocks often govern the dynamics of astrophysical plasmas: they transmit energy from stars to the interstellar medium, they can compress gas past the point of gravitational instability so that stars form, they may terminate the growth of protostars by driving the ambient gas away, they efficiently destroy dust grains and thereby determine interstellar gas phase abundances, and they are efficient accelerators of highenergy particles. Shocks are particularly important in astronomy because they heat gas and cause it to radiate, providing valuable diagnostics for the underlying energetic phenomena and for the ambient medium%.

Introduction

The study of coronal (T≳ 106 K) interstellar gas is a relatively new branch of astronomy. Before the 1970s, there was little direct evidence for such gas, although theoretical models predicted that it should be found in the interiors of supernova shells. In 1956, Spitzer made the prescient suggestion that the galaxy would likely possess a hot corona much like the solar corona. By the early 1970s, a series of rocket experiments had shown that the Milky Way was glowing in soft X-rays, indicating that coronal gas was pervasive in the interstellar medium; this interpretation was supported by observations by the Copernicus satellite of the interstellar absorption line O VI λ1035, showing that this tracer of high-temperature gas was extensively distributed throughout the galaxy.

We now have good maps of the brightness and temperature distribution of the soft X-ray emission from the Milky Way. With X-ray telescopes we have seen emission from coronal gas in elliptical galaxies and between the galaxies in clusters. As a result of these observations, the theory of coronal interstellar gas has advanced rapidly. The atomic processes that determine the local temperature, ionization, and spectral emissivity of the gas have been studied in detail. We have also learned much about the energy sources and macroscopic processes that control the global properties of the interstellar gas. It is now clear that the coronal gas in the Milky Way is produced mainly by the blast waves from supernova explosions, although stellar winds and compact X-ray sources may dominate in specific locales.

Introduction

Quasistellar radio sources were first recognized as objects at very large distances in 1963, when Schmidt identified several nebular emission lines in the spectrum of the stellar appearing, thirteenth magnitude object 3C273, and measured its redshift as z = 0.158. Greenstein and Matthews soon identified similar lines in 3C 48, with V = 16.2, giving z = 0.367. It was immediately clear that these highly luminous objects are beacons that can be observed out to the distant reaches of the universe. For many years OQ 172, with z = 3.53 and V= 17.8, was the most distant object known, until in 1982 an even larger redshift, z = 3.78, was measured for PKS 2000-330, a quasar with red magnitude 17.3. However, observations also quickly showed that all quasars and QSOs do not have the same absolute magnitude; like stars, they are spread over an enormous range in luminosity. Hence we can only hope that ‘understanding’ quasars, through the study of their spectra, will mean not only understanding the strongest energy sources we know in the universe, but also recognizing their absolute magnitudes from their spectra and thus determining their distances.

Long before the first quasars were discovered, galaxies with generally similar emission-line spectra were known. In 1908, Fath, working with a small slitless spectrograph on the Crossley reflector of Lick Observatory, recognized five nebular emission lines (in addition to weak Hβ) in the spectrum of NGC 1068, lines we now know as [O II] λ3727, [Ne III] λ3869, and [O III] λλ4363, 4959, 5007.