The Case for a UV/Optical Cosmic Survey Mission

C. Martin & D. Schiminovich, Caltech, May 1999

Summary

We propose a major new mission to map and characterize the dominant baryonic component of the Universe, the intergalactic medium (IGM). With it, we will recover the first true map of the dark matter which dominates the Universe. We will discover and analyze intergalactic gas which is dissipating gravitational energy and collapsing to form galaxies. We will map the galaxies that form from the collapsing IGM with profound sensitivity and efficiency, and relate the distribution, properties, and evolutionary histories of these galaxies to the large gas and dark matter structures in which they formed and evolved. The mission will bring no less than a synthesis of the major elements of cosmogenic theories: the dark matter, the gas which falls into it, cools, and collapses, and the galaxies and galaxy histories which result.  Our ultimate goal is to understand the origins and evolution of structure in the Universe.

Our approach is to use the most sensitive tracer and diagnostic of gaseous matter and star formation in the Universe, ultraviolet spectroscopy. We propose a mission that provides high and low resolution spectroscopy of the faintest possible sources with unprecedented angular coverage and efficiency. We will survey the IGM in absorption using faint quasars and a dense grid of sightlines, and in emission using resonance lines of abundant elements. We will survey galaxies in high resolution to probe their mass, stellar populations, and star formation history. We will map galaxies at low resolution ten times fainter than the Hubble Deep Field, over regions of the sky a thousand times larger, acquiring redshifts, star formation rates and spectral energy distributions for essential all of the galaxies in the last ~99% of the history of the Universe.

The Problem

Mapping the Cosmos

In 1986, the CFA redshift survey group published “A Slice of the Universe”. The result, a deeper look at the distribution of galaxies in the Local Universe, electrified the astronomical community, the community of science, and the public as a whole. It remains for astronomers one of the seminal results of the century. Why?  There are three reasons. First, the survey was the first to sample a volume large enough to contain the largest structures. Secondly, the number density of galaxies had reached a critical threshold that delineated those structures in space as coherent and vast. Finally, it tapped a deep and basic human need: to map the world within which we find ourselves.

Mapping the cosmos remains one of Astronomy's premier goals, and continues to capture the public imagination. Unfortunately, Nature has placed a formidable obstacle in our path. Galaxies, our traditional signpost for constructing the map, represent at best a tiny fraction of the mass and volume of the Universe, and are the end result of processes and structures that we as yet cannot see or understand. The vast majority of the mass is dark. Ninety percent of the “conventional” mass-energy (as opposed to the source of any cosmological constant)  in the Universe is in dark matter, exotic particles that produce no visible emission. Ten percent of the mass we believe is in normal baryons, but only 10% of these reside in luminous, observable galaxies. The remaining 90% of baryons fill the intergalactic void in an intergalactic medium (IGM).  Thus only 1% of the Universe can be mapped using galaxies. We now know that galaxies are not unbiased tracers of the mass. Galaxies form where the density is high, but not where it is low.  Because we do not understand how galaxies are formed, we cannot predict where they will form. We thus cannot recover from the galaxy maps alone a map of the remaining 99% of the Universe.  This is a profound missing piece of the puzzle: the distribution of dark matter drives the formation and evolution of the structure we see, most importantly galaxies.

Mapping the Dark Matter

Theoretical models of cosmogenic evolution now simulate the growth of structure in dark matter, and can produce convincing simulations of the large scale structure of galaxies. For many, these models represent our principle contact between what we observe and what we know. But very little of what they predict has ever been tested. Once the cosmology is known (and we expect to know it in five years), the models rely on at least one bias parameter to normalize their distributions with those observed for galaxies. We have no idea how to derive the bias parameter from first principals. We do not know that it is constant in time or space.  Models  for the distribution and evolution of dark matter on large scales must be observationally tested independent of the bias parameter.

Over the past few years, we have come to realize that the 90% of baryons that fill intergalactic space, the IGM, may hold the key to mapping the Universe. Simulations show that the low density IGM traces dark matter with astounding fidelity (Figure 1). The IGM is perhaps the only way to map the dark matter without bias in regions where highly non-linear growth and gas dissipation have yet to occur. In regions of intermediate scale and overdensity, weak lensing mapping offers a method to cross-check maps derived from the IGM distribution.

Figure 1 --- Left: slice of N-body hydrocode showing distribution of gas density. Right: identical slice showing distribution of dark matter. (Miralde-Escude 1996)

Understanding Galaxy Formation and Evolution

Numerical cosmology predicts that dark matter will grow into non-linear structures, collecting gas until it is dense enough to cool, dissipate potential energy, and form galaxies. Dark matter cannot dissipate gravitational energy, and therefore the higher density IGM begins to decouple from the dark matter.  At this stage, cosmological models run squarely into a wall of complex and unexplored physics. Semi-analytic models that make comprensive predictions for the history of star formation, the luminosity function of galaxies, even the distribution of galaxy types, are at their core heuristic and unphysical. They must rely on simplistic, parameter-driven representations of  processes like halo gas infall and cooling, disk formation and evolution, global star formation, and feedback. Observations which unravel and independently constrain these processes are essential. But the physical regimes connecting diffuse dark matter, dark matter halos, and galaxies we can observe through starlight are almost invisible. How do we probe them? Is there any hope of actually watching a galaxy form?

Galaxies form from high density regions of the intergalactic medium corresponding to overdense regions of dark matter (Figure 2). The high density IGM relates directly to the story of galaxy formation and evolution, and to the uncharted relationship between the dark matter and the galaxies we see. We therefore give it a distinct name, the Pre-galactic medium (PGM). 

In order to understand how galaxies form and evolve, we need to analyze the physical state of the PGM in regions where galaxies and groups have formed and where the might form. We must understand the distribution, density, ionization, metallicity, temperature, kinematics, dynamics, and thermodynamics of the gas. We must some how relate these to the galaxies that actually form, to their global properties: luminosity, star formation history, clustering, angular momentum, size and velocity distribution. It is crucial to understand how galaxies feed back energy into the PGM to delay and modify gas collapse and star formation. Finally we need to connect these properties to the underlying structure of the dark matter. This formidable, critical observational program requires a bold new mission approach to make progress.

Figure 2 -- Left: slice of N-body hydrocode showing distribution of gas temperature. Right: identical slice showing

distribution of column density. Darkest regions have cooling/collapse times less than a Hubble time, and many should form galaxies (Miralde-Escude 1996)

The Method

We combine five observational approaches to achieve a synthesis of cosmogenic models, galaxy formation physics, and the universe we observe in starlight:

1)       Absorption-line spectroscopic mapping of the IGM and PGM over six orders of magnitude of density;

2)       Emission-line mapping of the IGM and PGM over six orders of magnitude of density;

3)       Weak lensing surveys of the dark matter distribution

4)       Redshift & star formation rate surveys combining extraordinary depths and survey volumes.

5)       An absorption-line spectroscopic survey of galaxies

1) Absorption Line Spectroscopic Mapping in 3 Dimensions

The most powerful and sensitive technique for analyzing intergalactic gas is absorption line spectroscopy of fundamental atomic transitions. These transitions are all in the ultraviolet in the rest frame of the absorbing gas.  High resolution UV and optical spectroscopy yields absorption features (Figure 3) which give the distance (redshift), the total amount of gas in each distinct region along the sight (column density), and the random velocity of the gas (line width).  The simulations show that there is a close correspondence between the column density of gas in a feature and  the associated over-density of dark matter d=r/<r> (Figure 4).  The local over-density largely determines the history and fate of a particular region of the cosmos.

High-resolution absorption-line spectroscopy requires the brightest and most distant background objects in the universe, quasi-stellar objects (QSOs).  The objects are among the rarest as well, and even the brightest few dozen require 8-meter class telescopes (e.g. Keck) to analyze their complex spectra. A single QSO spectrum provides a one-dimensional core sample of the Universe, and studies over the past two decades have been hugely productive with just a handful of objects scattered across the sky. A 1D core sample, however, cannot be used in isolation to map a 3D universe. QSOs are so rare that few of these sight-lines are close together. Imagine trying to reconstruct the physiology of a human body or the complex geology of a mountain range with a few 1D ``core samples''. 

Figure 3 LEFT: QSO absorption line spectrum from Keck. RIGHT: Overdensity d corresponds well with Ly a  column density (Miralde-Escude 1996).

How can we make a 3D map from these 1D core samples of the Universe obtained toward QSOs? We must obtain a (random) grid dense enough and over a broad enough region to put multiple lines of sight through coherent structures over an enormous range of scales: from 10-100 kpc (the scales of galaxy halos and the PGM) to greater than 100 Mpc (the largest known structures in the Universe).  In order to produce this both dense and expansive grid, we must measure high resolution spectra from the faintest and most numerous QSOs with superb efficiency, wide angular coverage, over a major fraction of a multi-year mission.

Figure 4 – IGM simulation of  a 100 x 100 x 100 Mpc³ cube (Cen 1998) with representative QSO sight-lines.

We require a mission design that samples the full range of relevant physical scales: from those which result in galaxies to those of the largest structures in the universe. The latter is given simply by the field of view of the survey W---to sample a region 100 x 100 Mpc², we must survey a field of view of 3x3 deg², or W =10deg². The former requirement is analyzed in Figure 5. The abscissa in this figure is the overdensity d, which relates to density, column density, and cooling time. Given a mass, it also relates to scale size. The ordinate is the parameter F=N²/W, or the total number of QSOs observed squared divided by the field of view. This parameter is derived from the requirement that we have a minimum of 10 cases where two lines of sight pass through the same overdense region. As d increases, the cross section decreases and the probability for double intercepts falls steeply. Thus using this criterion, it is the highest overdensity regimes that drive the mission design. Designing the mission to multiply sample individual overdense regions also guarantees that the 3D structure of large, filamentary structures will be delineated clearly with may contiguous lines of sight. This figure of merit (F) is proportional to the effective area (aperture squared times system throughput), the total mission time, and the multiplex advantage.

This small-scale driving requirement is crucial for reconstructing the physics of galaxy formation and evolution from the properties of the PGM. Over the last two orders of magnitude of density cooling time falls from approximately a Hubble time to about 1-3% of a Hubble time (depending on z). Collapse into galaxies, and sustained star formation in existing galaxies requires cooling times less than the age of the Universe. Only by measuring multiple sight lines through these regions can we measure size, derive density and cooling time, and measure rotation velocity. Only by comparing these site-line pairs at differing redshifts (cosmic times) can we directly observe the engine of galaxy formation and sustained fueling at work. As we see below, we will be able connect these absorption line measurements with emission line observations that provide crucial missing information.

Figure 5 --- Figure of merit for absorption line spectroscopy mission F=N² /W, vs. overdensity, for M=10¹¹M_sol (baryonic) condensations. N is the number of QSO spectra obtained, W is the total field of view. The requirement is derived from the stipulation that in at least 10 cases two site-lines will pass through the same overdense region. Three different redshift intervals shown. Also shown are F for HST/COS, a 4 m single object mission, and an 8 m spectroscopic detector mission with multiplexing advantage.  Overdense regions with  3000<d<100,000 have cooling times 100% to 3% of a Hubble time, and can only be sampled with an 8-m mission with spectroscopic detectors and multiplex advantage.

Based on the above considerations, the mission must have an effective area of 100,000 cm² for high resolution spectroscopy, be able to simultaneously observe at least 10 objects over a field of view of 10 arcminutes, and dedicate a full year to surveying an (ideally) contiguous cosmic volume. The proposed mission obtains an F that is 1000 to 10,000 times improvement over HST/COS. A factor of 30-100 comes from effective area, 10-30 from multiplex advantage, and 10 from total mission time dedicated to the survey.

Another possible background object for probing the IGM with absorption lines on a much finer grid is star forming galaxies. While these have much more complex intrinsic spectra, the vastly higher surface density (10-100 /sq. arcmin) may permit statistical combination of neighboring galaxies to average out the intrinsic absorption features. Also, many absorption lines in star forming galaxies are broad and thus distinct from the narrow IGM lines. High sensitivity and multiplexing advantage are essential to making this possible.

2) Emission Line Measurements--A New View of the IGM and PGM

Imagine how our picture of the Milky Way and nearby galaxies would be modified if 21 cm, CO, and molecular emission line maps did not exist. These maps provide the framework for our understanding of star formation in a spiral galaxy. We can understand the physical relationship between diffuse gas, dense molecular gas that forms stars, and the shocks and ionization fronts that feed back energy into the interstellar medium. The formation of galaxies may be far more complex. Unfortunately, the low densities, high ionization, and large distances of the gas that is forms and fuels galaxies makes analogous maps remote.  How can we hope to construct a framework for galaxy formation without such images?

While the dark, non-baryonic matter will almost certainly never be seen, the IGM and PGM does release an extraordinarily faint emission that our proposed mission will be able to measure for the first time. More than simply detecting a completely new component of the Universe, these measurements will provide critical missing information about the nature and distribution of the gas which traces dark matter and forms galaxies. Obviously, the capability to survey emission gives us a true ability to map. These 3D spectroscopic images, obtained over the same regions observed in absorption, will “bind” the limited absorption site-lines into a coherent, contiguous volumetric map. But they will do far more.

Figure 6 – Emission line mapping will survey IGM and PGM structures over the wide range of scales and densities relative to galaxy formation from the diffuse intergalactic medium. Figure shows schematically the collapse of gas into structure (bold arrows), and the sensitivity of emission lines measurements vs. the minimum density and scale which gas clouds must subtend in order to be detectable. Larger condensations are easier to detect because the emission can be binned up. HI clouds that emit resonantly scattered Lyman alpha (blue line) can be detected even in the diffuse IGM. Collisionally excited CIV (orange line), NV, and OVI (orange dashed line)  emission, and Ly alpha recombination (black lines) emission are detectable in denser structures of the PGM. Also shown is the regime in which x-ray emission is produced.

Spectroscopic emission line maps could provide, size, density, cooling rate, abundance, line kinematics and halo dynamics. Observations of a collapsing cloud in a dark halo could in principle provide the mass of the halo (from velocity and scale), and the mass and cooling rate of the gas.  The rate at which stars form may be closely related to the cooling mass flux. Abundances will trace the influence of early generations of star formation and mass outflow on the collapsing PGM. Because most of the background QSOs for absorption lines observations will be quite faint, we can combine absorption and emission line spectra in many cases, tightly constraining the physical conditions and calibrating both techniques. We will be able to probe the relationship between the properties of the collapsing gas, any star formation or galaxy already underway (see below), and the dark matter/IGM matrix in which it is formed.

Emission from the IGM and PGM is extremely faint and low surface brightness. We expect three principle components: collisionally excited C,N,O, recombining HI, and resonance scattering.  Collisionally excited line emission is produced in gas that is collapsing and cooling. Some collisionally excited HI Ly alpha is produced, but most of the hydrogen is rapidly ionized. More likely is the production of collisionally excited CIV 1550, NV 1210 and OVI 1033 in dissipating gas at 10^4.5-10⁶K. Gas moving at several 100 km/s will dissipate strongly in these lines even if CNO are present in only trace amounts (e.g., 1% solar). These are the velocities that correspond to galaxies and galaxy groups. Galaxy clusters dissipate at higher velocities and produce soft x-ray emitting gas. Recombination radiation is produced by ionized hydrogen, and will appear in gas at 10⁴-10^5.5 K. Recombination radiation should be easier to detect, but may be masked in some cases by resonance scattered radiation. The latter is produced when the global metagalactic (non-ionizing) radiation field which impinges upon the IGM is  scattered into the line of sight. This emission can only be seen if the majority of the galaxies producing the metagalactic background are resolved out. Since the comoving UV luminosity density of the universe appears to increase significantly for z>1, the proper luminosity density is much higher, and we expect the resonance scattering emission to be detectable even at high redshift.

Our goal is to detect emission over the full range of density, size, and redshift that will allow us to follow (statistically) the progress of gas as it collapses from the diffuse IGM (r~1) to galaxy sized clouds (r>10⁶). In Figure 6 we summarize the detectability of IGM/PGM emission vs. overdensity and scale size for the proposed mission.

The emission can only be detected using high resolution spectroscopy, since the signal will be a very small fraction of the more local background signal (zodiacal and diffuse galactic light), and velocity information will be crucial. Long slits or Fabry-Perot spectrographs will be required to image sufficient solid angle to detect the lowest surface brightness signals and to map efficiently. The combination of these almost certainly will require order-sorting, spectrally resolving imaging detectors.  Figure 5 shows that with sensitivities that can be obtained in the proposed mission, structures over the wide range of scale and overdensity can be detected and even mapped in emission.

3) Weak Lensing surveys of the Dark Matter distribution

Dark matter concentrations produce subtle shearing of faint background galaxies known “weak lensing”. If enough background galaxies can be surveyed, this shear can be detected over Poisson noise of random alignments. This represents an important tool for determining the dark matter distribution of the universe, and several groups are actively pursuing weak lensing surveys from the ground.  Ground-based surveys are limited to determining the power spectrum of dark matter fluctuations averaged over  the line of sight determined by the redshift distribution of galaxies observed (and correspondly attenuated).  Faint galaxies, which are small, are not resolved so that alignments are difficult to measure.  Thus the redshift range may be limited. 

Space-based weak lensing surveys however offer several powerful advantages. First of all, a panoramic spectroscopic camera operating over 100-1000 nm can determine galaxy redshifts over 0<z<10 to an accuracy of s_z= 0.001. Coupled to a large space telescope, redshifts for galaxies as faint as 0.1 nJy can be obtained. With a large format detector and a dedicated survey mission, large areas of the sky can be surveyed. This combined with the large surface density of galaxies at 0.1 nJy (greater than 10^6.5 deg^-2), and the ability to resolve small objects makes it possible to map the dark matter in 4 dimensions (3 space, 1 time) by differencing alignments measured for galaxies in a redshift bin from the cumulative effects of the foreground matter. In particular, detectable signals can be obtained from volumes as small as 10⁴ Mpc^-3, or 20 Mpc (comoving) on the side.  By mapping, the underlying distribution can be recovered rather than a line of sight statistical moment of the distribution.  This offers the possibility of directly correlating the dark matter with the baryonic matter in the IGM mapped through absorption and emission line spectroscopy, and thus cross-correlating both mapping techniques. It will provide an answer to the question of the randomness of fluctuation phase (the Gaussian approximation). It will allow precision tests of dark matter structure models over a wide range of redshifts.

4) Complete Stellar Census

In order to make the connection between the dark, dim and luminous matter, the mission will perform an essential complete census of star formation in the universe over 0<z<5. The low resolution, panoramic survey discussed above will provide such a census.  The survey can cover 10,000 times the area of the HDF, to a depth 10 times fainter, providing redshifts and low resolution spectra for a hundred million galaxies. Star formation rates as low as a few 0.1 M_sol/yr will be detectable at z=10,  and 0.03 M_sol /yr at z=3. A survey of 10 deg² will yield ~10⁸ galaxies with redshifts accurate to 0.001, star formation rates and extinctions obtained from spectral slopes, and starburst history information from low resolution spectral shapes, in addition to spatially resolved spectral information about the star forming populations, their history and metallicity, and satellites which may drive starbursts and trace the history of mergers. Using a few percent subset of these, spatially resolved spectral cross-correlation will provide rotation velocities and sometimes rotation curves from UV and optical ISM lines. It will also be possible to measure relative velocities of companions and group members. Thus dynamical information can be added to determine dark matter halo masses. With this enormous sample, we can make a robust statistical connection between the properties of the dark matter,  IGM/PGM, and the galaxies and galaxy histories that result, over the redshift range that most star formation and galaxy evolution occurred.

5) High resolution analyses of the ISM and stellar content of star-forming galaxies.

Galaxies are complex systems. High mass stars are likely to play a major role in determining how PGM gas clouds collapse and cool, how galaxies evolve, and how metals are recycled into the IGM/PGM. The ISM in galaxies may provide information about the injection of feedback energy and metals. The highest mass stars have disproportionate impact on the energetics of the ISM, but  the initial mass function may be a function of time, of metallicity, or of local pressure.

High resolution UV/optical spectra provide critical additional diagnostics that can be obtained in no other spectral band. In particular, the upper end of the IMF can only be analyzed using high resolution UV spectroscopy of stellar wind and photospheric lines. These spectra can give definitive information about the size and duration of starbursts along with the distribution of high mass stars.  ISM absorption lines fall predominantly in the UV as well. Only with high resolution UV/optical spectroscopy can the stellar and interstellar lines be separated, and the line profiles, velocity distributions, metallicities measured. A high resolution spectrograph designed for detecting IGM/PGM emission can also be used to survey large numbers of galaxies in the UV/optical. 

Table 1 – Summary of Primary Surveys

The Mission

The mission is accomplished with a 4-8 m class telescope. It is hoped that the optics,  mount design, and mission design will be based entirely on that of NGST.  Deformable mirror or mirrors, located either upstream of the focal plane or in the instruments, produce good  (0.1”)  imaging over the wavelength range from 90-1000 nm.  A focal plane array of energy-resolving, photon-counting, superconducting detectors will provide simultaneous imaging spectrophotometry (R=30-300) over the entire passband over a large field of view (~10 x 10 arcmin²).  At least one additional focal plane instrument will allow multi-object, high resolution spectroscopy (R=5000-20000) of point sources and extended objects over a wide field of view (~10 arcmin minimum diameter). We summarize mission parameters in Table 2.

This mission will only be possible if significant technology funds are invested in development of panaromic spectroscopic detectors such as superconducting tunnel junction detectors. 

Table 2 – Mission Summary

Summary

A UV/O Cosmic Survey Mission will provide the first images of the intergalactic and pre-galactic medium, will map the cosmos over 99% of its history with astounding physical insight, and will unify dark matter, intergalactic gas, galaxy formation, and galaxy evolution in a single, sweeping observational framework (Table 3). Panoramic spectroscopic detectors are required to make the major leap in observational efficiency required to achieve these goals. Aggressive technology development over the next decade will make this mission possible in the middle of the following decade.

Table 3 – Summary of the Cosmic Constituents Surveyed (Objective)                                                                    1%