Solar and Space Physics: Space Science in the Twenty-First Century -- Imperatives for the Decades 1995 to 2015 (1988)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Appendix D Probing Fundamental Astrophysical Scales with High-Resolution Observations of the Sun: Prospects for the Twenty-first Century INTRODUCTION The past decouple has seen a great increase in the sophistica- tion with which we are able to confront the physics of the Sun. Physical theories have progressed from those that assume a simply stratified, equilibrium atmosphere overlying a classical convection zone, to those that recognize intermittent magnetic fields in the convection zone and dynamical structures on Al spatial scales throughout the atmosphere. In such situations, it Is possible that there is no static equilibrium structure at ah. Furthermore, it is believed that all of the observed structures in the Sun, even the largest, are ultimately governed by small-scale processes associ- ated with intermittent magnetic fields or turbulent stresses. For example, electric currents on scales of order 10 km or less may well be the fundamental entity giving rise to coronal heating. Understanding the physics of the creation and decay of such small-scale currents and their effects on mass and energy transport is thus essential to a proper description of large-scale structures such as coronal active regions, loops, flares, or mass loss in the solar wind. The interplay between processes occurring on vastly differ- ent spatial scales is ubiquitous in astrophysics and heliospheric 112

113 physics. Small-scale magnetohydrodynam~c turbulence ~ thought to govern the accretion rate of accretion disks feeding compact galactic x-ray sources and black holes in active galactic nuclei; local instabilities such as Keivm-Helmholtz modes are thought to control the coupling between the magnetospheres of neutron stars and the surrounding matter. MagnetoLydrodynam~c turbulence is also believed to suppress efficient heat conduction in the hot halos of galaxies and galaxy clusters, thus controlling the rate of accretion in these halos. Further, the same small-scale processes that heat the solar corona are undoubtedly at work in the coronae that are now known to surround many stars. In all these cases, observations using even the most advanced technology currently conceivable will not allow us to observe di- rectly the controlling small-scale phenomena. In the case of the Sun, however, we can indeed contemplate direct observations. The Sun is therefore a unique too} for understanding a wide range of astrophysical objects, by virtue of the opportunity it affords to observe the underlying physical processes in some detail. In this study, the workshop participants examine the scientific rationale and technological bash for pushing the study of solar magnetohydrodynamic processes well beyond the regime antici- pated from solar space missions planned for the coming decade (~100 km on the Sun, or ~0.! arceec at ~ AU). First, the scientific issues involved are addressed, and it is concluded that observations of solar structure on a spatial scale in some cases as small as 1 km would provide an enormous increase of our knowledge of basic astrophysics. Next, practical limits imposed by the emitted flux and the superposition of separate structures along the line of sight are discussed. Finally, the workshop participants addre" some of the technological challenges to be met if the objectives are to be attainable early in the twenty-first century. In this study, the workshop participants have adopted as a baseline assumption and prerequisite the existence of a program (broadly speaking, the Advanced Solar Observatory) that will, over the next decade, result In the observational study of the structure and dynamics of the solar atmosphere at a resolution of about 100 km, in visible, ultraviolet, and x-ray wavelengths. Thus, the welI-establ~shed rationale for observations at these spa- tial scales is not discussed in any detail.

114 TlIE ROLE OF SMALI-SCA[E PROCESSES IN THE SO[AIt ATMOSPHERE The solar atmosphere exhibits a vast range of spatial and temporal scales, from coronal eruptions as large as the Sun itself, to flare-related instabilities that involve scales of centimeters and micro-seconds. Moreover, there is mounting evidence that the large and small scales are inseparably linked, such that neither is purely cause or effect; it is the combination that produces what we observe. For example, very small-scale reconnection events in the corona may suffice to destabilize a small portion of a highly stressed magnetic arcade. The dynamic restructuring of the field, and the fluid flows that go with it, may in turn drive further reconnection and further instability, until the entire arcade blows oE as a coronal mass ejection. In order to unravel such a synergistic connection between large and small scales, we need to explore the intermediate range of scales. Some of the basic spatial scales of the solar plasma, such as the ion gyroradius or the Debye length, typically fall in a range (on the order of centimeters) that is not amenable to remote sensing and Is unlikely to be explored in the forseeable future. However, important physical processes can be studied on larger scales that are still far below the current Innits of observation. Solar activity and flares are prime examples. Present models of the site of primary energy release and particle acceleration all involve mechanisms—e.g., magnetic reconnection, plasma double layers, strong shock ~raves, anomalous current dissipation that require or generate gradients in magnetic field, temperature, pres- sure, and velocity on scales in the range 0.1 to 10 km. We already know, from indirect arguments, that flare kernels are inhomoge- neous on scales of <100 km. Observations with spatial resolution in the range 1 to 100 km thus offer our best hope for penetrating to the heart of solar flares. There is now good evidence that much of the flare energy ~ channeled into the production of beams of nonthermal electrons and that the deposition of such beams in the chromosphere is the principal means for creating the thermal flare plasma. The detection of linearly polarized bremsstrahinng from the impact of an electron beam would constitute a direct signature of this process. Past attempts have been frustrated by the low efficiency of soft x-ray polarimeters and especially by a complete lack of

115 spatial resolution, which dilutes the signal and makes it difficult to disentangle it from instrumental effects. If technological advances make possible an efficient polarimeter with high spatial resolution, it would be possible not only to demonstrate unequivocally the existence of electron beams, but also to trace the beam (or beamed to the site of primary energy release and to flare kerned in the lower atmosphere. Coronal loops, a fundamental building block of the solar atmo- sphere, are now recognized to be fundamental to the understanding of stellar coronae and stellar atmospheric activity In general. A decade of intensive work has demonstrated that the gross proper- ties of a loop (e.g., temperature, pressure, magnetic field strength) do not allow us to decide whether it will be stable, let alone how it is heated. We need to know the internal structure of the loop. For example, what are the temperature and density profiles transverse to the major axis of the loop? If the magnetic field ~ smooth, add if only classical cornfield transport processes are at work, flux tubes separated by only ~ to 10 km can have widely different temperatures and pressures. If snowfield transport ~ enhanced (due, for example, to a drift-wave instability), the characteristic scale of the gradients may expand to 10 to 100 km. This first of Al affects a directly observable quantity, the differential emission measure, but it also bears on the structure of the loop as a whole. Transport processes in a loop are only one aspect of an even more basic question, the organization of the magnetic field. For example, some theories of coronal heating postulate that intense electric currents flow along coronal loops, and intense currents are associated with smaD-scale twists in the magnetic field. To study the internal magnetic structure of a loop, either by the use of tracers or by direct measurement, we will require at least leant perhaps as many as 10~resolution elements across the minor axe, which extends a few arceeconde. The structure of magnet+fluid turbulence in a gravitationally bound system ~ a basic astrophysical problem that can probably be investigated nowhere but in the Sun. Aside from its intrinsic interest, turbulence in the photosphere concentrates and disperses the magnetic field according to the (unknown) power spectrum of the turbulence, and the photospheric magnetic field is the starting point for creating or maintaining fine structure at higher leveb in the atmosphere. Existing observations and theory already sug- gest that most of the magnetic field in the photosphere exists as

116 discrete flux tubes with characteristic diameters of <500 km. Cur- rently planned observations with lO~km spatial resolution should confirm or deny the existence of such flux tubes beyond doubt. However, as in the case of coronal loops, we win almost certainly need to resolve the internal structure of the tubes before we can understand the physical basis for their formation and persistence. The value of studying small-scale processes in the solar atmm sphere can be illustrated by considering the development of our understanding of the large-scale structure and dynamics of the terrestrial magnetosphere, where we already know much about the linkage between large- and small-scale processes. The early observational and theoretical studies of the magnetosphere dealt primarily with large-scale phenomena, and these studies led to a good understanding of the gross, average properties of the mag- netosphere. It was found, however, that a sound understanding of the large-scale mass structure, electric current structure, and tem- poral evolution of the magnetosphere required the detailed study of microscopic plasma processes. Several examples come readily to mund. The onset of magnetospheric substorms is related to the initi- ation of fast field-line reconnection in the geomagnetic tail, which depends on microscopic plasma processes operating on spatial scales comparable to the ion gyroradius. Magnetospheric current systems, particularly during substorms, are controlled in part by the field-aligned potential drop that ~ associated with double lay- ers, anomalous resistivity, and other microscopic plasma processes occurring on small spatial scales. The large-scale mass and energy structure of the plasma sheet in the geomagnetic tad] is determined in part by particle acceleration on microscopic scales and by parti- cle precipitation into the ionosphere associated with wave-particle interactions in the magnetosphere. Both the current structure and the particle precipitation problems are intimately related to the large-scale structure and evolution of the aurora. EXAMPLES ON TEE SUN OF SMA[I~SCALE PROCESSES OF ASTROPHYSICAL IMPORTANCE Atmospheric Dynamics and Magnetic Fields on Slrmll Spatial Scales Several of the space experiments planned for the 1990s have been designed to study dynamical processes and magnetic fields in

117 the photosphere, chromosphere, and corona on scales as small as 100 km. The experiments are expected to provide data that will help to unprove our understanding of fundamentally important problems such as (~) the nature of granulation, (2) the interaction between small magnetic elements and photospheric turbulence, (3) the physics of spicules, and (4) the flow of matter and energy between the chromosphere, corona, and solar wind. However, many aspects of these problems will eventually require observations on scales even smaller than 100 km. For example, the present theory of granular convection envi- sions a turbulent cascade involving a spectrum of sizes ranging down to scales as small as 100 m. While the smallest scales of the inertial range are probably not observable, it would be important to observe at least one decade (esy, 100 km to 10 km) of the inertial range to characterize the spectral distribution and hence to provide a firmer observational basis for the theory of turbulent transport. It ~ equally probable that such observations vnI} instead demand a physical explanation outside the scope of present theory. The superb whit - light observations of granulation obtained by the SOUP instrument on Spacelab 2 have already shown, even with 0.4~arcsec (300 km) resolution, how oversunplified our conception Of granular convection has been. The upward extensions of fine-scale photospheric magnetic fields are associated with the supergranular network, spicules, and other structures that play a role in the transfer of mass and en- ergy between the chromosphere, the corona, and the solar wind. For example, theory medicates that the interaction between pho- tospheric turbulence and magnetic fields on scales of the order of 10 to 100 km may induce electric currents capable of heating the corona. Experunents planned for the 199Os promise to elucidate aspects of these important processes, in particular by observing dynamical phenomena in the cooler parts of the upper atmosphere (T < 5 x lOsK) on scales as small as 100 km. The results of these experiments will undoubtedly raise new questions regard- ing the interactions between flows and magnetic fields on scales significantly smaller than 100 km. For the hotter atmosphere ~ T > 106K), current observation and theory imply that observations of magnetic field structures and their interactions with flows of matter on scales in the range of 100 to 1000 km are required in order to study the processes involved in coronal heating and the acceleration of the solar wind.

118 Fme Structure and Dyn~m~ce of Flares There ~ good observational evidence that the physical pro- cesses that control solar flares manifest themselves on scales of <50 km. For optically thin radiation, the observed emission mea- sured Tom a single resolution element may be combined with an independent estimate of the gas density and the known radiative efficiency to yield the effective volume (V) of emitting material and thereby a conservative estunate (Vi/3) of the dimensions of the radiating structure. If the structure is filamentary (rather than roughly spherical), its narrow dunension will be smaller than the est~rnate. This technique was applied to many extreme ultraviolet (EUV) slit spectra from Skylab. For example, a flare-associated surge was studied in which the Doppler shift of the surge material allowed it to be isolated from other material in the field of view or along the line of sight. Density-sensitive line ratios were used to estimate an electron density greater than 10~3 cm~3 and a characteristic scale of at most 60 km. The scale lengths derived for this and other flares are consist tent with optical spectra and with ultraviolet and x-ray emission measure data from instruments on board the Solar Maximum Mis- sion (SMM). In short, flare structures with characteristic scales extending significantly below 100 km have been inferred at all wavelengths (and temperatures) thus far observed. The dynamics of flares is linked to their small-scale spatial structure. SMM data have shown that lines such as Ca XIX ~ ~ ~ 107 K) show significant nonthermal broadening during the impul- sive phase of flares. The degree of broadening is nearly indepen- dent of the location of the flare on the solar dek. The broadening may be the result of spatial integration over many unresolved flare loops, of time integration over highly transient, small-scale flows, or of locally isotropic turbulence. It is essential to distinguish be- tween these possibilities by achieving higher spatial (~10 km) and temporal (~l s) resolution while maintaining spectral resolution. One of the key advances that emerged from SMM was confir- mation of the importance of high-energy electron beams in solar flares. A substantial fraction and perhaps the majority—of the flare energy may be channeled through beams. When high-energy electrons interact with the ambient plasma, they produce hard x-rays via bremsstrahiung and ultraviolet emission via collisional

119 heating. However, it is unlikely that electron beams are the only mechanism responsible for exciting flare emission at high densities. High-spatial-resolution studies of flares, simultaneously in visible and ultraviolet light, should clarify the role of electron beams and elucidate the other mechanisms. For example, since an electron beam moves along magnetic field lines, a measurement of the area of the ultraviolet surface region in the chromosphere yields an up- per limit to the cros+sectional area of the beam at that height (an upper limit because scattering and absorption might produce ultraviolet emission beyond the confines of the beam). Arguments based on the observed relative intensities of ultraviolet and x-ray bursts imply that the diameter of the beams is of order 10 km. Plasma Heating and Microflares The fundamental processes leading to the heating of extended stellar atmospheres continue to puzzle us. Previous space exper- iments have provided observational constraints on some plasma heating mechanisms; for example, OS0-8 data showed that scour tic waves alone cannot heat the transition region and corona. Further progress has been impeded by the fact that the diagnose tics necessary to differentiate between competing heating modem cannot be applied using present instrumentation: there ~ simply not enough spatial and temporal resolution. An example of a basic, unanswered point is whether the heat- ing process is steady or transient. Thus, one possible heating process involves the relief of stresses built up In coronal magnetic fields by the motion of the photospheric footpoint of magnetic field lines, leading to steady flaring over a wide range of flare energies; at the low-energy end of this flare spectrum, the flares are referred to as "r~iicroflares." Observational evidence for the existence of such m~croflares dates back to OS0-7 data of hard x-ray events, and more recently to hard x-ray data obtained with balloon payloads. These data show that the total heating rate by ~rflcroflares may be comparable to the coronal luminosity if the energy spectrum of electrons responsible for these transients extends down to ~5 keV and the power law connecting the cumulative number of events with 2~keV photon flux above a given threshold extends below the present threshold of 10-2 photons/cm2/~/keV. Thus, the su- perposition of such events could account for the steady coronal

120 radiative output. Clear observational evidence that heating in so- lar active regions ~ largely a transient process would exclude most of the proposed coronal heating processes; the implications for plasma heating in other astrophysical domains (such as accretion disk coronae) would be similarly of major consequence. An astrophysical situation in which the difference between steady and transient heating ~ of great current interest ~ the formation of coronae on very late-type stars (dMe stars) and very young stars (T Tauri stars). For example, high-speed photometry of T Tauri stars h" revealed short-term fluctuations sirn~lar to solar flares, having power law cumulative spectra similar to those mentioned above for solar hard x-ray transients. Such power law spectra have also been detected In optical observations of ultravi- olet Ceti flare stare. Moreover, it has now been shown that there is a good correlation between the mean level of chromospheric and x-ray emission and flare frequency for these low-mass stars. These stellar data support the idea that the observed emission from the hot outer layers of late-type stars is the result of a temporal super- position of transient energy release events. Thus, a very important question is whether quiescent EUV and x-ray emission from so- lar active regions (and even the quiet corona) ~ due entirely to superunposed transients. ~ order to answer this questions one is faced with the difficulty that, as the energy of m~ividual events decreases, their visibility above background at any given angular resolution decreases as well. This problem ~ not acute at high photon energies (>20 keV) because the slowly varying (background) component at these energies ~ ununportant even for full-disk (completely unresolved) observations, at least at current sensitivity levels. However, at lower photon energies (<5 keV) the slowly varying component begins to affect our ability to detect individual events, so that imaging becomes essential. Since the observed duration of the rnicroflares at high photon energies is a few seconds, it also becomes necessary to consider tune-resolution constraints. For example, because of the thermal inertia and the cooling rate of the transiently heated gas, it is difficult to detect low-level microflaring at soft energies because the coronal gas cooling time (103 s) far exceeds the time scale of the transient heating event itself In contrast, ultraviolet and EUV emission from gas at lower temperatures (~1 to 3 x 105K) can vary at the transient time scale because the corresponding

121 cooling tunes are sufficiently small. However, the smallest events will require high spatial resolution in order to avoid averaging within a resolution element. An estunate of the size of the elemental ultraviolet-em~tt~ng transient gas at the lower end of the energy spectrum is given by the estimated extent of flare kerned, <60 km (see above). It Is necessary to extend observations to even smaller scales in order to resolve these events and to establish the cumulative contribution of the power law m~croflare spectrum to the total ultraviolet/EUV luminosity of active regions. OBSERVATIONAL CONSIDERATIONS The above discussion has shown that much would be gained by observing physical conditions in the solar atmosphere on spatial scales considerably finer than the best anticipated from current program. Later in this appendix the technological considerations that enter into observing small-ecale structure are discussed. How- ever, intrinsic limitations that arise from the available photon flux and the geometry of the source are first discussed. The surface brightness of a feature limits the flux that can be collected from a single resolution element in a time shorter than the characteristic evolution time scale of the element. This tune scale could be as short as the sound or the Alfven crossing time often one second or less for subarcsecond structures. At what rate will photons arrive at ~ AU from an element of EUV-em~tting material? If the element ~ a cube of side it, the number of photons detected per second at ~ AU will be 1.0(n2Ama,,/10~2)(d/10 km)3~/1001) (D/1 m)2 (~/10~2)photon s- ~ where Amiss is the cooling coefficient, nomad is the volumetric radiative power, D ~ the aperture of the telescope, and ~ ~ the overall efficiency of the instrument (photons counted divided by photons incident). It ~ amumed that the pa~band try at wave- length ~ includes all the radiation from the line or lines, and that there is one element in the field of view. If the telescope resolves distances smaller than it, the counts are assumed to be integrated over ~ x d. In the visual and near ultraviolet, the expected flux is encour- aging for a meter-cIam telescope. For example, the C IV doublet near 1548 ~ produces nc2Am'~= ~ 1.0 at a pressure ncT—10~6

122 (roughly appropriate for an active region loop), leading to a pho- ton flux ~~03 photon/s from a I~km element. However, for higher temperature lines (T > 1055) in the EUV, typically ne2Ama': < lo-2 at the same pressure, and, at ~ AU, only a few photons would be detected per second. There could be little confidence that the element would not evolve substantially or even disappear during the time necessary to build up an image with adequate signal-to-nome ratio. These figures refer to structures of roughly average brightness. In flares, pressures 10 or more tunes greater have been inferred. The density enhancement in m~croflares ~ not known, but could be comparable. Thus, there is good reason to expect that some structures, come of the time, will have 100 or more times the average brightness Nevertheless, if we wish to observe typical EUV structures of a size of <10 km on the quiet Sun, there is a flux problem. This could be addressed either by larger collecting areas (of order 100 m2) or by meter-class telescopes in near-Sun orbit (say, at a distance of 0.1 AU). The example given above treats a single emitting volume 10 km on a side. Some lines of sight may include several structures. Although the flux at Earth increases accordingly, the flux prom lem is alleviated only at the expense of source confusion, since the individual structures may have distinct physical conditions. Source confusion could be addressed by stereoscopic observations, in which two or more spacecraft simultaneously exanune the same region of the solar surface from different vantage points. A third problem ~ the reduction in contrast of very fine struc- tures embedded in an optically thick medium such as the pho- tosphere. In this case, image sharpness is limited by the photon mean-free path (~100 km in the photosphere for visible light). This problem is less important for observations of Doppler or Zee- man line shifts to the extent that scattering causes diffusion in position but not In wavelength. Moreover, the problem disap- pears for optically thin media such as the chromosphere (for many spectral lines), transition zone, and corona. It may be that the twin problems of flux limitation and source confusion will, more than purely technical considerations, limit the remote sensing of solar features to scales of ~ km.

123 HIGH-RESO[UTION DAMAGING IN THE EUV Because of the absence of atmospheric Retortions and the ac- cessibility of all wavelengths, high-spatial-resolution observations of the Sun are most advantageously carried out from space. How- ever, in the past, few space-based telescopes approached their theoretical resolution. Instead, resolution was limited by focal length restrictions, the quality of the optical systems, and the properties of the detector. For example, the High Resolution Telescope-Spectrograph (HRTS) ultraviolet telescope is limited to 1 arcsec (rather than 0.! arcsec) because of overall size and detec- tor limitations. The Solar Optical Universal Polarimeter (SOUP) visible-light telescope (3~cm aperture) ~ so far the only telescope to operate in space (Spacelab 2, 1985) at its theoretical resolu- tion of 0.4 arcsec. The projected resolution of the Hubble Space Telescope tHST) approaches its diffraction limut of 0.05 arcsec in the visible and near ultraviolet, but not in the ultraviolet. The 1-m optical telescope of the High Resolution Solar Observatory (HRSO) will have a diffraction-limited resolution of 0.13 arcsec across the visible spectrum. Thus, in the visible wavelength region, small space-based tele- scopes have achieved their diffraction limit, and large telescopes will soon do so. If the diffraction limit could be achieved at EUV or soft x-ray wavelengths, we would gain an order of magnitude or more in angular resolution. This appears feasible if progress made in several areas: 1. Mirrors of higher figure quality must be polished, corrected, or actively controlled. 2. At wavelengths below 1000 A, nearly diffraction limited resolution probably requires normal incidence optics, and this in turn entails the development of multilayer, high-reflectance coatings. The use of phase surface correction and adaptive optics techniques aIreacly explored in the visible may also be required. 3. The linear resolution of available detectors (film or two- dimensional solid state) is inadequate. For example, a diffraction- limited I-m telescope would require an f/3000 focal ratio in order to match a detector with 1~,um pixels. Therefore, without enIarg- ing optics (see the subsection below on detectors), detectors that have submicron resolution elements must be developed. Although high spectral resolution does not appear in itself to pose technical difficulties at this stage, a combination of high

124 spatial, spectral, and temporal resolution may strain the photon flux limits discussed In the section above on observational consid- erations. Interferometry is another technique for achieving extremely high resolution and is considered below. Also addressed below are the problems of pointing systems and the question of where in space these instruments would best be operated. The workshop participants conclude that, in all the above- mentioned technical areas, rapid developments are being made today and that ultrahigh-resolution telescopes or interferometers will be feasible within the time frame 1995 to 2015. Detectors Currently available detectors (including charge coupled de- vices (CCDs), television-type systems, and electronographic cam- eras) are capable of pixel size near 10 Am and a format of about 2000 x 2000 resolution elements. TV-type systems have flown on many my - ions, and recent improvements in these older tech- nologies have shown that 2-pm pixels and 4000 x 4000 format can now be achieved, while maintaining high quantum efficiency throughout the EUV and soft x-ray region. It appears likely that a pixel size of less that 1 Am can be achieved with very fine grain phosphors or thin layers of evaporated scintiliator applied to the aplanatic focus of a high-quality lens (such as a Burch objective) coupled to a high-resolution TV or TV-type detector. Even higher resolution, of order 0.2 ~m, can be achieved with a semitransparent photocathode followed by rennaging of the photoelectrons onto a suitable detector, with acceleration of the electrons and expansion of the plate scale achieved simultaneously. Some increase in plate scale can be accomplished by suit- able design of the optical system, such as using relay optics in a grazing incidence configuration or using a Camegra~n design with normal incidence optics and x-ray multilayers. However, at the highest magnifications, the mechanical positional and alignment tolerances will make such telescopes very expensive. In order to achieve the highest resolution in an instrument of practical size and cost, it wiD be necessary to build detectors with m~cron-size pixels. For solar observations, large image format is also needed, which can be obtained with a very high resolution television tube, a large format CCD or MAMA (multi-anode microchanne} array),

125 or by a mosaic of detectors. None of these techniques has yet been shown to work in a large enough format for high-resolution solar applications, and only systems that enlarge the focal plane image onto the detector have the required pixel size. Mistily Coat~ge for Normal Incidence Optics Recent progress in techniques for the deposition of high- quality, thin-fihn multilayere onto optical surfaces allows us to begin planning a new generation of instrumentation for the EUV and soft c-ray region. We can now consider instruments that are qualitatively different from those traditionally employed in this wavelength region ones that will have the spatial and spec- tral characteristics of the highest quality ground-based optics and some spaceborne instruments such as the HST and the HRSO. High-Resointion bnagmg All conventional coatings have low normal-incidence reflectiv- ity in the EUV. This imposes severe constraints on EUV instru- ment designs for space astronomy. Below 400 i, normal incidence designs have led to unacceptably low throughput, especially in applications involving multiple reflections. However, further de- velopment of a new technology multilayer, thin film coatings that have high reflectivity at normal incidence at wavelengths down to 20 ~ will allow the use of EUV instrumentation that incorpo- rates normal incidence components. In addition, such coatings can be applied to glancing incidence mirrors in order to enhance their short-wavelength performance. The principal advantage of normal incidence optics in compar- ison with glancing incidence is the ability to achieve substantially higher image quality with a given level of effort. In addition to this simplified manufacture and testing, normal incidence tele- scopes will generally be lighter and more compact, and can provide high-resolution images over a broader field. Multilayer coatings increase x-ray reflectivity by exploiting the fact that a discontinuity in the complex refractive index causes reflection of an incident wave at the boundary. In the EUV, this reflection is small because the material has a refractive index near unity; however, by adding more boundaries such that all reflections add in phase, the reflected intensity increases as the square of the

126 number of boundaries N. ~ spacer layers with zero absorption could be found, 100 percent reflectivity could be achieved. In practice, N and therefore the maximum reflectivity- ~ limited by absorption in the spacer layers. To achieve multilayer coatings with high reflectivities, uni- forrnly thick layers and sharp, smooth interfaces are required. These requirements severely limit multilayer performance at short wavelengths. It appears that the fundamental limit is the diffuse nature of the atoms thernseIves, which limits us to wavelengths longer than about 10 to 20 A. In theory, reflectivities of 30 to 50 percent should be obtainable throughout the 2~ to 30~A range, and such values have already been achieved at the longer wave- lengths. In general the substrate must be smooth to better than one- tenth of the multilayer period, because any roughness will be replicated in the multilayer and degrade its performance. Such surface quality appears to be within the state of the art, particu- larly on small substrates, even in the 2~1 wavelength region for which 1-~ root-mean-square surface smoothness is required. The deposition of uniform layers onto meter-cIass substrates should be readily achievable by methods similar to those used for longer wavelength optical coatings. Spectroscopy and Spectroscopic Imaging By an effect analogous to Bragg reflection, normal incidence multilayer optics provides an inherent spectral filtering that allows simultaneous high-angular-resolution imaging and spectroscopy. The multilayer spectral resolution A/~) is roughly equal to the number of layer pairs and varies from ~10 at 200 AL to ~103 at 201. Imaging in a narrow bandpass that reflects a single line, or lines, of a single ionization state, formed over a narrow temperature range, may provide the best opportunity to study fine-sc~e solar features at EUV wavelengths. In such nearly isothermal images, structures are more sharply defined and are more readily interpreted than images averaged over a broad temperature range. Regions of the solar spectrum have been identified that con- tain several emission lines of comparable strength spanning the 105K to 107K temperature range for which multilayers with high- reflectivity can be produced. In such cases, dispersive imaging (similar to that obtained in the EUV on Skylab) will yield useful

127 scientific diagnostics. To accomplish this, a transmission grating introduced into the optical path of a normal incidence telescope produces a high-resolution image in each of the emission lines. Multilayer coatings have been applied to reflection gratings in order to enhance their efficiencies in the EUV by about a factor of 3, demonstrating the feasibility of their use in spectrographs. A simple focal plane, stigmatic spectrometer using a normal inci- dence, toroidal grating can be constructed with a moderate size mirror acting as a feed to the grating. Resolution of A/~> ~ 3 x 104 can be achieved, thereby permitting separation of nearby spec- tral lines, measurement of line broadenings and asymmetries, and deterrn~nation of radial velocities to a few kilometers per second. Glancing Incidence Optics Although the multilayer technology just discussed opens up the exciting prospect of applying the fine-scale imaging capabilities of normal incidence optical systems to solar observations at soft x-ray and EUV wavelengths (30 to 400 Hi, there will still be a need for glancing incidence optical systems at shorter wavelengths. At present, the inherent size of the atoms used to manufacture mul- tilayers (W. Ir, C, Ag, Si, Ma, and so on) appears to restrict their use to wavelengths above about 30 A. In addition, even though substantial improvement in the pixel size of photoelectric detectors is expected in the very near term, it has already been noted that the pixel sizes required for m~liarcsec imaging at these shorter wavelengths become remarkably small. One way to handle this problem is to use multielement telescope systems with substantial magnification. However, the magnifying elements in these sys- tems are then subject to high concentrations of solar flux and the multilayer coatings on them are subject to photopolymer~zation of hydrocarbon contaminants. This buildup of molecular contamina- tion on the mirror surfaces not only decreases the reflectivity in the bandpass of interest (somewhere between 30 and 400 \) but also increases the absorptance of the intense visible light flux falling on them (about 50 percent of this flux is absorbed by a clean, uncon- taminated multilayer surface) and could lead to thermal runaway and destruction of the coatings. Glancing incidence optical systems, at ~ < 30 A, are insensitive to such contamination and can be used to image a very broad wavelength band, including wavelengths below 30 1. However,

128 just as multilayer performance is limited by the surface roughness of their substrates, the performance of glancing incidence optics at 200 to 400 ~ is currently limited by "m-frequency" surface ripple with a spatial scale of a few millimeters. Until now, the optical transfer function of glancing incidence telescopes has been characterized by very narrow cores and broach scattering wings that contain the great bulb of the energy in the point source image. Recently, computer-controlled polishing technology has been applied to the production of glancing incidence optics, and the surface smoothness achieved indicates that good imaging at the arcsec level can now be achieved. The amplitude of the m~- frequency ripple on these surfaces is about 300 to 350 ~ root- mean-square, and within the next few decades it should be possible to produce surfaces with nud-frequency ripple of 30 ~ root-mean- square or better. This improvement in surface smoothness by an order of magnitude Anti reflect itself in an unprovement in attainable resolution by an order of magnitude, the limiting factors then being the ability to achieve and maintain relative alignment of the primary and secondary mirrors and the diffraction effects caused by the large central obscurations of glancing incidence telescopes. Primary Figure Confection and Control The prunary resolution of the telescope is affected by the alignment accuracy of the optical train and the consequences of variations ~ element position having temporal periods of hours (thermal) to teethe of a second (vibrational). In addition, there exist residual figure errors on the primary and subsequent m~r- rors, and smaller-scale errors induced by thermal loading, polish- ing imperfections, and phase distortion due to residual coating nonuniform~ties (which may be wavelength-dependent). The goal for reduction and compensation of these errors will be determined by the desired linear resolution, the incident flux available, and engineering feasibility, not necessarily the diffraction limit at all wavelengths. The diffraction limit at 1-m aperture and = 1000 ~ (~0.02 arcsec) may be realistic, while for ~ = 30 (~0.007 arcsec), it may not. Correction may be accomplished by a hierarchy of techniques. Correction of constant residual figure errors can be done to poten- tially high precision by fabrication of a fixed compensation plate

129 generated by photores~tive techniques. Because multilayer dielec- tric coatings conform very accurately to the substrate on which they are deposited, resolution is critically dependent on substrate figure. After polishing, current meter-cIa" mirrors show residual deviations of 300 ~ peak-to-peak, which would severely degrade short-wavelength resolution. Because the defects are fixed, a cor- rector with conjugate phase errors can be employed to compensate the errors in a subsequent pupil plane. The completed instrument can be up as one arm of an interferometer, with an accurate reference surface in the other arm and the photoresist coated cor- rector In the fringe plane. For small enough residuals (<800 A), the exposure creates the appropriate phase conjugate, and subsequent development of a multilayer coating of the corrector completes the process. Current experiments suggest that residual surface errors of 10 ~ may be achievable. Tune-varying figure errors of low spatial frequency (generally thermally induced) can be measured by wavefront sensing or in- terferometric techniques and used to correct the primary mirror figure (by means of electromechanical actuation), perhaps supply mented by an active mirror operating in a subsequent conjugate plane. The technologies of wave front sensing have already shown accuracies of better than )/500 in the visible. Many forms of ac- tive or adaptive corrector mirrors employing piezoelectric, electro- static, and other principles have been built, and high-performance stabilization of large space mirrors is currently under vigorous development. Furthermore, wave front sensors can easily resolve 10,000 or more pupil elements, which can be used to drive tower resolution correctors. The residual error ~ a dynamic measure of the instrumental optical transfer function, which can be used for post facto resolution enhancement (see following subsection). These technologies are applicable to both normal and glancing incidence configurations, and they are already sufficiently mature to include in the basic design philosophy of any instrument. Resolution Enhancement Techniques The angular resolution of the proposed optical systems may not in practice be diffraction limited at EUV wavelengths, even with correction and control. Several techniques can be applied to improve the spatial resolution after detection, with the degree of unprovement dependent on wavelength, flux levels, and feature

130 contrast. All of these techniques should yield some improvement even at very short wavelengths. Aperture Synthesis (AS) In AS, a mask with a set of transmitting holes is inserted across the prunary mirror, or ~ a reduced image of the primary. The resultant subapertures are positioned so that their pairwise autocorrelations sample the (u,v) (spatial frequency) plane to the greatest degree possible. Conventional images are recorded with the array In place. Subsequent analysis of these coded images allows estimation and correction of the phase errors across the subaperture, and eventually, reconstruction of enhanced images. The major advantage of this technique ~ its simplicity of implementation, with full image reconstruction from one or two exposures. Its drawback is that perhaps only 10 to 20 percent of the total flux Is transrn~tted by the masks. Speckle Nonaging Speckle imaging techniques, which have been used to recover diffraction-lim~ted images with large-aperture telescopes through atmospheric turbulence, may be used to overcome the effects of the uncorrectable errors in EUV optics. If pairs of small aper- tures are scanned across an optical pupil plane, and images are recorded for each position of the subapertures, then averaging of the complex autocorrelations of the images would allow high- resolution recovery. Each position of the pair of subapertures provides a statistically independent realization analogous to ran- dom atmospheric fluctuations between short exposure frames in ground-based speckle. The major problem ~ that the tane required for this serial procedure ~ significantly longer than a single-frame technique such as AS, so its time resolution is Innited. Its advan- tage is the potential to correct to much finer scales. Deconvolution A by-product of an actively controlled optical system is that an accurate measure of the optical transfer function is obtained by the system wavefront sensor. Because the sensor uses its own source, good signal-to-nome and accuracies better than A/1000 (visible)

131 can be achieved. An accurate deconvolution of each image with the system point spread function may therefore be applied as a post- processing step. Only the presence of zeroes in the optical transfer function limits this technique, and these may be eliminated by adequate compensation with active optics. INTER1?EROMETRY One way to obtain very high angular resolution is through spatial interferometers, using either arrays of telescopes or arrays of flat murrors feeding a central beam-combining telescope. The goal of such an interferometer would be very high angular rem olution unaging (m~Biarcseconds) from visible wavelengths (7000 1) to the EUV (~500 1~. Although the precise form of such an interferometer remains to be defined, at least two options are available. Both are so-called monolithic arrays, which means that the interferometer system is mechanically coupled in such a way that the pathlengthe between the Sun and the focal plane are approxanately equal in all legs of the interferometer. Both are also designed to have a wide field-of-,riew, which means that the pupil configuration at the entrance and exit of the interferometer · ~ is preserved. In the first, a "thin monolithic interferometric "ray," HRSO is used as a beam-comb~ner. As in the early Michelson and Pease interferometer, pars of flat mirrors are used to feed the light into the telescope from points outside the telescope. In one version, five of these 4() cm-diameter interferometer beams are arranged in a pentagon with a 12.~m diameter, feeding the lam diameter optical telescope, giving l~iilliarcsec resolution in the visible or 3-milliarcsec resolution in the C IV lines near 15501. It could be assembled in space to the required dimension. Because of the large coherence lengths (~2/^ ~ l/~) that char~terme high-spectral- resolution solar observations, the tolerances on the lengths of the interferometer legs are rather loose for narrow bandpass observa- tions. Such an interferometer could be implemented at an early date at a relatively low cost. For unaging, it will be necessary to determine the relative pathIengths of the interferometer legs. This requires the development of the equivalent of phase closure techniques used in radio image synthesis telescopes like the VLA but for solar observations, and image synthesis by rotation of the array and by changing the length of the arms. Because of the flux

132 limitations in the ultraviolet, phase closure has to be done in the visible wavelengths. The second type of interferometer is a Thick monolithic arrays in which the interferometer elements are coupled by an optical metering system to achieve cophas~g. It gives complete (u,v) plane coverage in one dimension. Rotation of the array will fill the (u,v) plane. Phase closure ~ not needed in this case. Both types of arrays are best used in conjunction with HRSO on the Space Station or on co-orbit~ng platforms. Before the implementation of a free-flying solar space interferometer, it is im- portant to determine the visibility of different scales of solar fine structures by means of two-element interferometers, using ground- based interferometers for the visible and, e.g., a EURECA-ciass experiment for the far ultraviolet, allowing milliarcsec resolution. Phase closure techniques for solar observations can also be pio- neered from the ground in the visible region of the solar spectrum. POINTING SYSTEMS . The SOUP telescope on Spacelab 2 has demonstrated that it is possible to point a white light solar telescope with internal Image stabilization to a stability of 3 m~liarcsec using a 5~Hz control bandwidth. Similar image control techniques have locked two telescopes to an accuracy of a few milliarcseconds, which essential for ultraviolet observations because it is often necessary to use detectors that we insensitive to visible light or require that visible light be excluded from the telescope. The new generation of solar experiments should provide for multiple-aspect Stereo studies of solar structures. Since the structures to be studied may be ~ the I~m~liarcsec size range, it would be desirable to be able to powt several widely separated telescopes at the same feature on the Sun with an absolute accu- racy of a few m~liarcseconds. Since the average solar diameter is known to be stable at that level or better for periods longer than a month, a pointing system based on metrology of the average solar limb should yield the required accuracy. If the solar diameter and the pointing position are measured with a laser interferometer, which should yield )/10 accuracy, the location on a Cm solar image can be measured to 1.6 parts in lo6. Since the solar diameter ~8 about 2000 arcsec, this represents

133 a positional accuracy of 1.3 m~liarceec. Co-powting of two inde- pendent telescope systems also requires that their unage planes be homogeneous to this level and that their roll orientation be measured to 0.25 arcsec. SITING From the preceding discussion, it is clear that spatial res- olutions substantially below 0.! arcsec, at wavelengths from 30 ~ to visible light, can be achieved with techniques that will be perfected in the twentieth century. The question of where to sit- uate the resulting high-resolution solar instruments must now be addressed. Near-Sun Orbit Close approach to the Sun offers a conceptually simple way of achieving ultra-high angular resolution with instrumentation of modest aperture, and it may be the only means of obtaining I~ km resolution of the solar surface with sufficient EUV flux. The basic technology required for near-Sun orbits (e.g., heat shielding, telemetry through the corona, injection into the correct orbit) will be developed for the Starprobe progrmn. It would seem profitable to use this technology in follow-on missions carrying solar unaging experiments. Some of the additional studies for shielding the instruments were done in the preliminary stages of the Starprobe program and indicated that this concept ~ quite feasible. Heliosynibronom Orbit A spacecraft at ~30 solar radii approximately 0.! AU—will have a Today orbit around the Sun, and in the ecliptic plane will hover over a fixed point on the Sun. In addition to anportant long- term studies of the evolution of solar structures, and stereoscopic measurements of fine features when combined with another space- craft in similar or near-Earth orbit, such a vantage point offers an increase in spatial resolution by a factor of 10 and flux by a factor of 100 compared to an orbit at ~ AU. Lunar Basing A solar observatory on the Moon would operate under almost

134 ideal conditions, on a highly stable platform, free from atmospheric turbulence and contamination, and able to view the Sun continu- ously for Malay periods. Such an observatory could operate either manned or unmanned, and should be considered In the context of any future Moon landings. Two lunar observatories separated by 180° would give nearly 100 percent coverage. Solar Orbit at 1 AU Stereoscopic observations of the Sun can be achieved with two observatories in orbit at ~ AU widely separated ahead of and behind the Earth. Alternatively, one of the observatories could be placed at a Lagrangian point. It Is felt that the simultaneous pointing accuracies needed to achieve 0.01-arcsec resolution can be achieved. Earth Orbit Pree Flyer The ideal near-Earth orbit for a solar observatory would be a high-inclination, Sun-synchronous orbit, i.e., one perpendicular to the Earth-Sun line. The first mission of the HRSO (High Rem olution Solar Observatory) on the Space Shuttle is planned for a high-inclination orbit, which will allow long periods of un~nter- rupted observation. Manned Vehicles Space Shuttle It Is expected that the Space Shuttle in some version will still be flying at the century's end. Spacelab has proven itself a valu- able vehicle for astronomical observations. Advantages of Spacelab include significant real-time scientific interaction with the exper- iment as well as the ability to use either solid-state detectors or recoverable fihn. While solar observations can make effective use of week-Ion" flights, longer missions could significantly increase the scientific return. Disadvantages include relatively short observing periods for low-inclination orbits and the potential for contami- nation of highly sensitive ultraviolet and EUV normal-incidence optics. The principal contaminant is expected to be thin (10 to 20 R) layers of polymerized hydrocarbons, which are opaque to light between 300 and 13001. Solar telescopes are particularly prone to

135 organic contamination because the solar ultraviolet light polymer- izes hydrocarbons rapidly. Any site in space where such telescopes are deployed requires rigid material selection standards if the tele- scope is to operate for an extended period. Skylab experience demonstrates that this can be done. Space Station The Space Station shares many of the advantages and disad- vantages of the Shuttle. An additional concern is the expected broader spectrum of vibrations and disturbances, which make pointing difficult; isolation platforms will probably '~e required. Contamination Is also of great concern, and it is unperative to pay careful attention to materials selection for the Space Station from the outset. Of course, a co-orbiting platform circumvents many of these difficulties and would provide a very attractive site for solar observations. PROSPECTS AND [IMITATIONS OF GROUN~BASE:D OBSERVATIONS Radio l~terferometr~r The usefulness of high-angular-resolution observations for so- lar physics has been demonstrated at radio wavelengths since the early 1970s. The use of larg~synthesis radiotele~copes (such as the Westerbork Array and the Very Large Array) at centuneter and decimeter wavelengths made possible observations of a few arcseconds resolution. Source sizes of 3 arcsec have commonly been observed at 15 GHz during flares (somewhat larger sizes are seen at lower frequencies, presumably from loops located higher in the atmosphere). Radio observations with high spatial resolu- tion have been used to pinpoint the site of initial energy release in some flares, and such observations were the first to show the expansion and thermalization of high-temperature flare plasmas following the impulsive phase of the flare. Further evidence for small-scale structure in the solar atmo- sphere comes from observations with high temporal resolution. Microwave spikes of millisecond duration are sometimes observed during flares, implying source sizes of the order of a few hundred

136 kilometers. Even smaller structures are suggested by decimeter- wavelength observations of short-lived, narrow-band bursts. Di- rect observations of such structures at radio wavelengths may become possible during the next decade with the advent of the Very Long Baseline Array. More limited information might be obtained with existing very long baseline interferometers. While high-angular and high-temporal resolution radio obser- vations have yielded unportant new results, they are limited in one sense. All radio emission mechanisms are continuum in na- ture, and represent the signature of either thermal or energetic electrons. Therefore, radio observations do not yield information on the variety of ion species present in the solar atmosphere. Eigh-Angndar-Resoh~tion Optical Nags Tom the Ground There are a number of techniques that are currently yielding high-angular-resolution information at visible wavelengths on sm lar features. Adaptive optics, which use real-tune measurement of atmospheric aberrations and phase correction of those aberrations, has demonstrated a capability for substantive unage enhancement of small solar features (pores) and the surrounding granulation. Speckle interferometry and imaging have been applied to broad- band sunspot and granulation data, yielding detail approaching the diffraction limit of a l.~m-aperture (0.07 arcsec). Aperture synthesis has shown promise of recovering enhanced imagery frown a single short exposure, allowing very high time resolution. As these and other techniques are improved and used on large ground-based telescopes, the requirements for future space-based systems will be greatly clarified. Of course, ground-based systems have the severe limitation of the atmospheric transmission window. [ong-Baselme Optical Interferometry There has been recent, worldwide development of long-baseline (up to 1 km) interferometers. All such instruments have been de- signed for stellar astronomy. However, these or similar instruments could be applied to high-angular-resolution solar measurements. Special techniques that use field stops to isolate small features would be necessary to actively control the interferometric base- line, but these techniques are quite feasible. While no such solar observations have yet been proposed, such a system would have the potential of providing angular resolution approaching ~ milliarcsec

137 at visible wavelengths. However, because of their sparse coverage of the (u,v) plane, such observations are more difficult to interpret than images proper, particularly for complex, low-contrast sources like the Sun. SUGARY AND CONCLUSIONS There is now considerable evidence that all scales of structure on the Sun, as well as other astrophysically interesting objects, are strongly coupled to small-scale processes associated with interniit- tent magnetic fields and turbulent stresses. Understanding the physics of these dynamical structures and their interaction with their surroundings ~ essential for a proper description of large- scale structures (such as coronal active regions, flares, or the solar wind) and their effects on interplanetary space and the near-Earth environment. The interplay between processes occurring on vastly different spatial scales is ubiquitous in astrophysics. Whether in accretion disks feeding black holes at the center of active galaxies or quasars, in the magnetospheres of neutron stars, or in the x-ray coronae now known to surround a wide range of stars, smaD-scale magne- tohydrodynarn~c processes are thought to influence and sometunes control the behavior of the object. In these astrophysical situations, observations using even the most advanced technology currently conceivable will not allow us to directly observe the controlling smaD-scale processes. Using the Sun, however, we can indeed imagine direct observations. The Sun is therefore a unique too! for advancing our understanding of a broad class of important astrophysical phenomena, if we can penetrate to the domain of underlying processes that often operate on spatial scales of 1 to 100 km. An orderly progression of goals that could realize much of this promise would include: 1. Implementation of the High Resolution Solar Observatory on Spacelab, followed by the transfer of HRSO to the Advance Solar Observatory on the Space Station, together with ultraviolet and x-ray solar instruments capable of 0.1-arcsec angular resolu- tion. 2. Development of interferometric experiments at visible and ultraviolet wavelengths anned at preliminary reconnaissance of solar features at angular scales much less than 0.1 arcsec.

138 3. Development of meter-cIa" facilities exploiting the emerg- ing multilayer coating technologies, designed to obtain resolution in the 0.01-arcsec regime at extreme ultraviolet or soft x-ray wave- lengths. Ideally, several spacecraft would be located in near-Sun orbits to provide high flux, high-~near-resolution, and stereoscopic · ~ Imaging. 4. Achievement of ultrahigh-resolution imaging ~ ultraviolet and visible wavelengths, using baselines of order 10 m. Two concepts related to these goals are presented in separate recommendations of the task group: stereoscopic unaging using meter-cIass telescopes at four locations around the ecliptic, and an x-ray/ultraviolet/optical telescope in heliostationary orbit at 30 solar radii. APPE:ND~: LIST OF PA1tTICIPANTS S. Antiochos, Naval Research Laboratory R. G. Athay, High Altitude Observatory, NCAR J. Beckers, Advanced Development Program, NOAO R. Bonnet, European Space Agency G. Brueckner, Naval Research Laboratory R. Catura, Lockheed Palo Alto Research Laboratory L. Cram, Commonwealth Scientific and Industrial Research Organization I`. Dame, I,aboratoire de Physique StelIaire et Planetaire J. -P. Delaboudiniere, I,aboratoire de Physique StelIaire et Planetaire G. Doschek, Naval Research Laboratory G. Epstein, NASA, Goddard Space Flight Center T. Gergely, NASA, Headquarters M. Giampapa, National Solar Observatory, NOAO L. Golub, Harvard-Sm~thsonian Center for Astrophysics J. Harvey, National Solar Observatory, NOAO J. Heyvaerts, Observatoire de Meudon T. Holder, High Altitude Observatory, NCAR R. Howard, National Solar Observatory, NOAO R. Keski-Kuha, NASA, Goddard Space Flight Center and University of Maryland J. Leibacher, National Solar Observatory, NOAO B. Lites, High Altitude Observatory, NCAR

139 R. MacQueen, High Altitude Observatory, NCAR P. Nmenson, Harvard-Sm~thsonian Center for Astrophysics R. Noyes, Harvard-Sm~theonian Center for Astrophysics D. Rabin, National Solar Observatory, NOAO F. Roddier, Advanced Development Program, NOAO R. Corner, Harvard-Sm~theonian Center for Astrophysics L. Schmutz, Adaptive Optics, Inc. E. Shoub, University of Colorado D. Spicer, NASA, Goddard Space Flight Center A. Title, Lockheed Palo Alto Research Laboratory J. Toomre, University of Colorado J. Underwood, Lawrence Berkeley Laboratory