Development of a Coherent X-ray Diffraction Instrument

I. K. Robinson

Materials Research Laboratory, University of Illinois at Urbana-Champaign

This describes the dedicated Coherent X-ray Diffraction (CXD) facility under construction at the Advanced Photon Source (APS), Argonne National Laboratory. The instrumentation will take advantage of the unique source characteristics of the APS for the purpose of probing the microstructure of a wide variety of materials ranging from metal and semiconductor surfaces, internal interfaces, nanostructured materials, ferroelectric domains, polymers, biological samples and especially materials containing defect structures. The novelty of using the coherent properties of the beam lies in the measurement's sensitivity to the specific arrangement of the microstructure, and therefore, to any subsequent changes, whether spontaneous or induced. The general advantage of an X-ray based technique is that materials can be probed from the inside (penetration property) and the signal obtained is a linear function of the local density (kinematic approximation). X-ray diffraction naturally provides information with atomic resolution, being sensitive to the positions of individual atoms. However, the most important feature of CXD is that internal fluctuations can be observed in the time domain, using photon correlation spectroscopy (PCS).

Once constructed, the facility will first be used to develop the basic science of CXD. It is envisaged that in the region of 5-10 Ph.D. thesis projects and about 2-4 postdoctoral projects would be centered on the CXD facility. The overriding thrust, however, will be the subsequent application of the new science to problems in Materials Science, Condensed Matter Physics and Materials Chemistry. We will extend these studies to problems in the biological domain as our understanding of CXD increases. In the initial stages, interactions between those scientists expert in CXD and those focused on the materials problems will be used to broaden the science base of this facility. Examples of these interactions are described here.

5a RESEARCH ACTIVITIES

In the first section of the project description, we outline the principles of Coherent X-ray Diffraction and then give specific examples of research projects that can be carried out with the planned instrument. Several of these projects serve to introduce new areas of materials research to the capabilities of synchrotron-based X-ray diffraction and so will bring expert scientists from the University of Illinois Materials Research Laboratory (MRL) to become new users of the APS. Rather than describe past accomplishments separately, they have been used as background material for the proposed research.

The scientific potential of Coherent X-ray Diffraction (CXD) is immense. Because it functions by collecting the diffraction pattern from the entire field of view at the sample, typically 2-20µm square, it can be used to reconstruct an image, somewhat in the same way that a hologram does. The ensemble averaging that takes place in a normal X-ray diffraction experiment is absent, so CXD is sensitive to the specific configuration within the illuminated sample volume. Moreover, since any change of that configuration shows up everywhere in the diffraction pattern, the technique is exquisitely sensitive to fluctuations. By autocorrelating the time-dependence of CXD, the dynamical properties of the system are explored as a function of momentum transfer, q, covering a range that extends far beyond that of light scattering. Spontaneous thermal fluctuations on surfaces, in the near surface region, and inside polymers or biological molecules will be accessible. The mechanisms of relaxation involved with growth, dissolution or damage processes will be mapped out in real time. The time resolution is potentially as short as a microsecond, limited by the positron orbit frequency of the APS. It is this new opportunity of probing fluctuations inside matter with large q which makes CXD unique.

The first demonstration of static CXD was made in 1991 on a sample of quenched Cu₃Au alloy [1] and since then on a GaAs/InGaAs multilayer [2] and from a phase separated thin polymer film [3]. The first application of CXD to a dynamical system in the time domain was for the critical fluctuations associated with a second-order disordering transition in the Fe₃Al binary alloy [4]. The sensitivity to fluctuations was utilized to study diffusion of particles in a colloid [5]; it has led to a series of subsequent works in the area of colloids [6,7].

Technique Momentum[A^-1] Frequency[Hz]

The capability is illustrated in Fig. 1 where we demonstrated the measurement of relaxation times using XPCS [5]. In this experiment at NSLS, we observed the small-angle scattering from a concentrated solution of colloidal gold particles with a mean diameter of 600A. The small scattering angle provided a negligible path-length difference (PLD) in the sample, so the broad-band optical configuration was not a limitation. The central maximum of the small-angle scattering pattern was captured in a CCD area detector which was read out once per second. The 9µm pixel size was well-matched to the size of the speckles expected, so each pixel could record an independent fluctuation spectrum. The parallel detection gave an enormous gain of efficiency, as several thousand time correlation functions for each momentum transfer could be measured simultaneously, and averaged to improve the statistics. The result, shown in Fig. 1, is an impressively accurate correlation function.

The time correlation function in this system measured the diffusion relaxation time due to the Brownian motion of the Au particles. These particles were suspended in glycerol at 244K, well above the glass transition of 187K when diffusion would effectively cease. The observed relaxation times of 43s and 24s are close to the values expected for pure glycerol at this temperature of 61s and 22s respectively. It is noteworthy that the relaxation times do not obey the expected q^-2 dependence, and that there is possibly a second relaxation time apparent in the low-q data.

5b RESEARCH PLANNED

The potential of CXD is best illustrated by considering some examples of science that would be carried out with the instrument. This list is not exhaustive, but serves to show that research topics that are active in the University of Illinois MRL can benefit significantly from the new CXD capabilities. The names of the principal users of the planned instrument in each subject area are given in parentheses.

CXD will be a powerful method for examining surface morphology and the fluctuations thereof, both driven and spontaneous. It is complimentary to but has significant advantages over imaging techniques such as STM and Low-Energy Electron Microscopy (LEEM), especially in the ranges of time and temperature that are accessible. Surface CXD has already been demonstrated [10], and an example is given in Fig. 2. A multilayer-monochromated broad-band coherent beam of 10⁷ photons per second at X25 was passed through a 5µm pinhole onto a Si(111) wafer in specular reflection. The reflected beam was analyzed by scanning a 20µm `back' pinhole across it. The momentum transfer perpendicular to the surface, q_z, was varied to trade off surface-sensitivity (best at large q_z) for overall signal (best at small q_z). Fig. 2 shows a pair of back pinhole scans from the wafer at two different q_z values. At q_z=0.05A^-1 the reflectivity is close to ideally specular with little distribution of intensity beyond a single sharp central peak; part of the structure in the beam is the (always present) Fraunhofer diffraction that arises from the finite 5µm beam size. In this case the sample is acting like a perfect mirror.

The results of Fig. 2 can be extrapolated to the coherent flux levels of the planned facility, where sensitivity to atom-height steps will be achievable. Then the thermal fluctuations of these steps will be visible by XPCS. An excellent example of a surface that should show this is the step array that arises from miscut of a crystal face. This might occur either locally, due to inhomogeneities in polishing for example, or globally on a deliberately miscut surface. In such an array, the steps normally repel each other because of elastic interactions mediated by the crystal lattice. This leads to uniform step spacing at T=0. However, as the temperature is raised, the configurational entropy of fluctuations will favor step meandering with increasing amplitude. For the same entropic reasons, steps will also spontaneously appear and can eventually lead to surface roughening at sufficiently high temperature. Currently, there is debate in the literature about the relative importance of elastic vs entropic step interactions in typical materials at room temperature, but as a general rule, elastic interactions are favored on semiconductors and entropic interactions on metals [12].

For a step array, the magnitude of the fluctuations was shown by Villain [13] (among others) to be related to the free energy of formation of kinks in free step edges. It can be described equivalently by the continuum model of surface capillary waves [14-16]. The connection between the two descriptions is the microscopic interpretation of the surface tension in terms of step structure. The amplitude of thermal fluctuations of individual steps was measured on Cu by STM [17] and by reflection electron microscopy [18] and found to follow a random walk with a characteristic step `stiffness'. The time-correlation function of the step positions on Cu and Pt were found to obey a t<sup>1/4</sup>-law [19,20], as predicted theoretically [12]. Elastic repulsive interactions between adjacent steps in a step array lead to correlated fluctuations. Conversely, attractive interactions lead to `step bunching' [21].

It is difficult to predict how sensitive CXD will be to thermal step fluctuations because of large uncertainties in estimating the size of the fluctuating regions. In the limit of very stiff steps, which extend entirely across the field of view, the situation can be considered as a one-dimensional step array with fluctuating disorder in the step locations. The diffraction pattern will be diffuse in the reciprocal space direction(s) perpendicular to the steps but sharp in the direction along the step edges. We can then compare the level of diffuse scattering with that of the Au colloid, discussed above, which gave a cleanly measurable XPCS signal (Fig. 1). In the colloid experiment, a number of order 10⁴ particles, each containing 10⁶ atoms, was measured [5]; in a hypothetical Au step-fluctuation experiment with the same coherent beam size, around 10³ steps, each containing 10⁴ atoms per row, would be involved. Thus a signal about 100 times smaller is expected in a speckle pattern with 10 times fewer peaks. Another important difference is that the step edges are likely to be relatively well-ordered along their length, so the diffuse scattering will be more localized.

The provisional plan is to study surfaces like Au(110), Pt(110), vicinal Au(111) or Pt(111). Steps on vicinal Pt(111) were already studied by STM [20], providing direct information on the step stiffness between 500-800K. The steps apparently bunch at high temperatures (T>1300K) [21]. The structural anisotropy inherent to the (110) surfaces will provide relatively stiff steps along [10], and the marginal stability of the `missing row' reconstructions should lead to a high spontaneous step density [22]. Previous experience with X-ray diffraction on Pt(110) showed that the steps played a central role in its structural phase transition too [23].

Faceting of surfaces and internal interfaces is a poorly understood phenomenon that can profoundly affect growth and film adhesion. It is believed to be a dynamical process driven by a competition between surface diffusion and surface tension, which undergoes coarsening from atomic to macroscopic length scales. Several important theories, which need testing, apply to faceting [24-27]. CXD offers the potential to make controlled experiments accessing the dynamical behavior on the relevant submicron length scales.

The (115) surface of copper is an excellent example because it is marginally stable when clean and because its faceting can be driven reversibly by the addition of oxygen. Many years ago, it was reported to show a thermally driven roughening transition [13], but this has been discounted in subsequent energy-analyzed He-scattering experiments [28]. Under saturation coverages of oxygen, it has been shown to facet into {104} and {113} orientations in recent STM results [29]. This evolution of Cu(115) during oxygen dosing at 570K leads to the final state shown in Fig. 3, showing the presence of three facets which index as (113), (104) and (04). The widths of the peaks indicate the facet dimensions are approximately 200A.

The distribution of facets during dosing was found to follow a complex sequence of events leading up to this state. Significant mass-transport is needed to build up the facets which therefore have important time- and temperature-dependencies. CXD can probe these in the kinetic regime, either by pulse-dosing at temperature or by cold-dosing and up-quenching. The kinetic exponents can then be compared with the many theoretical predictions and simulations that stem from the Kardar-Paresi-Zhang (KPZ) equation [24-27]. Another example of a suitable system for study by CXD is the recent work of Mochrie et. al. [32] on the thermal faceting of Si(113), where an unexpected kinetic exponent of 0.164±0.02 was found.

A second common, and technologically important, growth mode is the `step flow' mode, in which successive steps follow each other uniformly across the growing interface. In the ensemble average of RHEED or conventional X-ray diffraction, this growth mode is not observable, but with CXD it will be: a single step traversing the field of view will give rise to a cycle of intensity variation: CXD is sensitive to the positions of steps within the field of view. The scientific questions that will be addressed concern instabilities in the spacing of steps. This kinetic `step bunching', as opposed to thermodynamic bunching mentioned above [32], can have a profound effect on the equations governing growth and thus on the resulting morphology of grown films, with examples seen in GaAs [35] and GeSi [36]. The step bunching instability on Si(111) has been studied recently with LEEM [37].

The systems that would be studied initially are Si/Si(111) and Si/Si(100), which can be carried out conveniently in a UHV chamber if suitable sources and cryoshrouds are introduced to attain MBE-like conditions. There are interesting alternative ways to pursue analogous questions without requiring UHV. The first is to look at dissolution of silicon with buffered HF/NH₄F solutions that give rise to well-ordered hydrogen-terminated Si surfaces [38]. If the dissolution front remains locally flat, this implies a step flow mechanism, but the LBL mode has been reported [39].

A different kind of growth problem is the aqueous corrosion of copper, which can be reversibly controlled in an electrochemical cell. In the MRL and Chemistry departments, we have developed the necessary techniques to electropolish copper crystals and observe their oxidation with STM. We have also studied this corrosion process with in situ X-ray diffraction at NSLS and found it to involve the formation of epitaxial Cu₂O, with an inhomogeneous distribution [40]. The corrosion commences with a monolayer of Cu₂O which is equally distributed between `aligned' and `reversed' epitaxial orientations. In the next step, the aligned oxide remains as a monolayer while the reversed oxide starts to grow thicker; the aligned oxide meanwhile is apparently acting as a corrosion barrier. Even as the reversed oxide thickens, it does so inhomogeneously, forming 50A tall islands over a very small fraction of the surface at first. These islands then broaden about tenfold while retaining their thickness, until a more uniform film has formed. In the late stages, the entire film thickens uniformly beyond the initial 50A. Afterwards, the entire system can be reset by reversing the potential of the cell and the growth cycle repeated again.

There are many unanswered questions in this example, concerning the spatial distribution of nuclei. An important question is whether the nucleation always takes place at the same sites on the surface or whether the pattern is different each time. Many questions surround the intermediate regime: whether the island broadening involves coarsening or not. CXD would be sensitive to the lateral distribution of oxide regions and would be able to answer these questions. Since the structural evolution takes place at the oxide-metal interface, deep inside the sample, they are invisible to strictly surface probes, such as STM. There is no other technique that can probe the spatial distribution within a growing oxide film in real time under in situ electrochemical conditions.

A related but separate set of problems concerns the anodic dissolution or corrosion of CuO and other oxide surfaces. Corrosion is an intrinsically inhomogeneous process, with initiation occurring at specific defects, and proceeding in a manner that depends upon the depth and structure of the oxide film, its adherence to the underlying metal surface, and the nature of solution constituents. Probe microscope studies of corroding oxide-passiviated materials, such as Al and steel, have largely been unsuccessful because the STM and/or AFM tip interferes with the corrosion process, and acts as an initiator for the corrosion event. It has also proven difficult to observe the initial stages of pit formation and dissolution on these more complex materials with probe microscopy. Thus, there is yet no good way of evaluating the shape evolution of an oxide covered surface as the morphology develops during in situ dissolution.

Proposed research would utilize CXD measurements to obtain the power spectrum of fluctuations on Al and CuO surfaces as a function of corrosion time. As these surfaces corrode, their roughness develops in a manner characteristic of the mechanism of corrosion, and this mechanistic behavior is mirrored in the power spectrum. Questions to be addressed include the following: what is the mechanism of Al and CuO corrosion? Is anodic dissolution or dissolution followed by redeposition the major pathway of corrosion? What is the interplay between the passivating oxide layer, pit formation, surface diffusion of hydrated species, and formation of dissolved species? How do inhibitors, such as chromate, act to retard pit formation and subsequent corrosion activity? The detailed picture obtained using CXD will go a long way toward resolving these questions.

We anticipate that it will be possible to utilize the coherent part of the diffraction from point defects to measure their diffusion in a momentum-resolved way. We can generate defects by a variety of methods, of which one of the most flexible and yet controllable is by ion-beam irradiation: the energy and mass of the implant allows wide control of depth-range and defect structure, while allowing accurate regulation of dose and dose rate. We plan to bring an ion accelerator from UIUC and couple it to the UHV chamber to carry out in-situ transmission diffraction experiments on thin specimens of metals and semiconductors.

In conventional ferroelectric materials, microscopic behavior can be deduced from macroscopic single crystal measurements, and conventional x-ray diffraction has played a significant role in this pursuit. This is not the case for the new class of `relaxor' ferroelectrics (RXF), in which dynamics on variable time scales play an essential role in determining the macroscopic properties as demonstrated by their strong frequency dependence. The Curie-Weiss temperature dependence of RXFs is characteristically damped compared with that of normal ferroelectrics, again because of the slow dynamics. The typical timescale of the relaxation process is between seconds and microseconds and is, therefore, well-matched to the XPCS method. A wide range of ideas has been advanced to explain the mechanism, which is likely related to the inherent structural disorder that is characteristic of RXF materials.

Lead Magnesium Niobate (PMN) and doped Lead Zirconium Titanate (PZT) are two important RXF samples. They are both based on the cubic perovskite structure, but they are separated into ferroelectric nanodomains with different atomic displacements directions relative to the crystal axes. Although the displacements are small, 0.22A for the Pb atoms in PZT [41], there is a significant change in the structure factor observable by CXD because of the relatively large momentum transfer. Electrostrictive lattice distortions are expected to dominate for doped PZT [42] and will affect the structure factor in an analogous way. XPCS using micron size beams will provide microscopic information on the typically 100 nm ferroelectric domain dynamics through speckle pattern changes produced of polarization changes in single domains as a result of thermal fluctuations or electric field changes.

The lattice disorder in PMN produces considerable diffuse scattering at room temperature, notably near (½,½,½), which can be decomposed into different modes of local ordering. We plan to identify those features in the PMN diffraction pattern that couple to the external field driven ferroelectric domains and investigate these features in detail by CXD. In addition, we will study new materials related to PMN. For instance, La doping for Pb is known to alter the Mg:Nb stoichiometry and lead to corresponding enlargement of the (½,½,½) domain size by more than an order-of-magnitude [43]. CXD speckle patterns will be used to investigate the diffuse scattering. Electric fields will be applied to study domain rearrangements in order to determine their role in the relaxor behavior of PMN.

Magnetic X-ray scattering arises from additional contributions to the form factor from the magnetic electrons, which have a characteristic polarization-dependence [44]. The cross-section of the magnetic contributions depends on the magnetic moment, but is typically 10^-6 of the charge scattering cross-section for pure magnetic term [45] and can be as much as 10^-3 for the cross-term [46]. Both terms display resonance in the form of enhancement of the form factor as a function of X-ray energy when transitions into the appropriate magnetic atomic levels are excited [47]. Thus the signal levels of magnetic scattering experiments will certainly be accessible with coherent beams in the instrument we plan to build.

Magnetic systems with microstructures in the interesting and accessible 10-100nm range are widespread. Ferromagnetic domains of this size are common in thin film materials and will fluctuate or can be driven by external fields. In ferromagnets, which have no new spatial frequencies associated with their magnetic structures, the magnetic scattering always occurs on top of a (larger) charge scattering signal. On the other hand, fluctuations due to spontaneous (or driven) flipping of the magnetic moments will result in an intensity change and would give a measurable XPCS signal if it were not for their rather short time-scale. For magnetic domain structures, the magnetic contribution will be broader than the charge scattering peak, so the visibility of the magnetic speckle will be further enhanced. An important example of this kind of system would be Gadolinium, for which the moments are significantly tilted away from the c-axis in domains of lateral dimension 100nm [48]. In c-axis thin films, the moments lie in-plane and display 2D-like behavior [49]. The intermediate thickness range and many details of the origins of these configurations are wide-open questions that can be tackled with CXD and their spin dynamics can be examined with XPCS. The relationship with film thickness will be explored by systematic series of thin epitaxial metal films that will be prepared by MBE in the MRL's `EpiCenter'. Complimentary examination of the same samples by LEEM will also be possible there.

Antiferromagnets can be investigated through their pure magnetic scattering signal which occurs in otherwise empty regions of reciprocal space, though the signal is smaller. Important examples are the spiral antiferromagnets among the rare earth elements, which have temperature-dependent discommensurate `spin-slip' structures [50] and large resonant CXD enhancements [47]. The spin-slip topology will also give rise to a lateral magnetostrictive effect that can be detected by non-magnetic CXD. Critical fluctuations both along and across the spiral are expected near to the Néel temperature, T_N, which is 133K for Ho and 176K for Dy. The lateral correlation length of the magnetic structure of Ho was found to be about 1000A at T_N - 1K [51], so can be probed by CXD to learn about the lateral structure of the spin-slip discommensurations and their mobility, about which little is currently known.

Magnetic CXD can be applied to commercially interesting `spin-valve' structures. These are constructed from nominally thin (several nm) magnetic layers separated by non-magnetic materials, and can be grown by sputtering or using the MBE facilities of the MRL. The small-scale domain structure and switching characteristics are crucial to important applications in which these serve as memory elements or nanoscale sensors.

5c RESEARCH INSTRUMENTATION

The second hutch is the CXD experimental hutch, into which only the minimum necessary amount of radiation will be allowed to enter so that background levels will be manageable. For the eventual µF experiment, a permanent conduit vacuum pipe will pass through the CXD hutch feeding a third µF hutch behind it. In this way, both experiments can operate in tandem, although access to the CXD hutch will not be possible without closing the shutter to both experiments. The offset between the beams at the position of the experiment will be about 100mm. This constrains but does not seriously impede the design of the CXD instrumentation. For example, the X16C beamline at NSLS, designed by the PI several years ago, has a similar 140mm offset and, therefore, uses a `Kappa'-geometry diffractometer [55] that has not encountered any serious space limitations.

The instrument we plan to build will operate in two modes, based on either broad- or narrow-band radiation. This division is based on technical grounds, because it is not easy to vary the bandwidth in a continuous manner. The broad-band mode uses the raw undulator fundamental directly; the narrow-band mode requires the insertion of a movable monochromator in the first hutch of the beamline. The bandwidth determines the longitudinal (temporal) coherence length of the X-ray beam, which must exceed the maximum possible optical path length difference (PLD) inside the sample. Thus, the broad-band mode will be applied to studies of reflectivity from surfaces, for which a UHV analysis chamber will be provided and small-angle scattering of materials such as metals, ceramics, liquids, bio-molecules or polymers. The narrow-band mode will be used for large diffraction angles to probe domain structures in thin films and single crystals.

There is more flexibility over the positioning of the vertical mirror. The best location appears to be about 2m in front of the sample slits, but this depends on the quality of the mirror and the maximum demagnification that it can achieve. The optical characteristics of both mirrors are explained in Table II. The mirror is preceded by a small vertical aperture that controls the vertical coherence (see below) and also reduces the thermal load to a trivial amount so that cooling becomes unnecessary. The mirror can be rather small and is made by bending an optical flat, so it should not be expensive. It is expected that some development of the fabrication, bending and positioning techniques will be needed here. This can take place in the open air on an optical bench during the development stages, but ultimately the mirror will need to be situated in UHV to protect its surface from contamination and to reduce the total number of beryllium windows.

Figure 7. Schematic design of the in-line UHV chamber for surface diffraction experiments.

In CXD, beryllium windows have a greater significance than in other kinds of diffraction experiment because they can also introduce a marked phase structure on the beam [58]. This is understood to be due to refraction by multiple grains of different effective thickness in the sintered Be powder. Since part of the inhomogeneity is at the surface, polishing of the windows may lead to an improvement. Equally important is to use the highest available grade of Be with the smallest number of impurity precipitates. Radical alternative window materials, such as single-crystal Be or diamond, may become practical as experience becomes available. Close coordination with the APS vacuum procedures is essential here, so the developments will most likely come from the APS. Anticipating that progress may be slow, we have planned our design to minimize the number of Be windows between the source and the sample, even to the extent that windowless operation may eventually be possible.

5d. IMPACT OF INFRASTRUCTURE

The short-term impact of building this instrument will be to develop the CXD method into an important tool for the study of properties of crystalline, polymeric and biological systems. A perhaps more important long-term impact will be to introduce the new CXD technique to a broad range of scientific areas, and to involve non-specialists in the use of this technique. After the capabilities of CXD have been mapped out, there will follow the subsequent application of the new science to problems in materials science, condensed matter physics and materials chemistry. We will extend these studies to problems in the biological domain as our understanding of CXD increases. In the initial stages, interactions between those scientists expert in CXD and those focussed on the materials problems will be used to broaden the science base of this facility.

University of Illinois students will benefit in two ways. Practical experience of instrument design techniques will be gained during the developmental stages. This essential element of graduate training in experimental physical science is becoming scarce in the no-growth laboratory environment. Then, once the CXD facility is built, it is envisaged that in the region of 5-10 Ph.D. thesis projects and 2-4 postdoctoral projects would be centered on it.

The instrument will contribute to meeting our nation's desire for world leadership in science. The APS is one of three 3rd generation hard X-ray sources in the world. The other two, in Grenoble (ESRF) and Osaka (Spring8), differ from APS in that their end-user instrumentation is centrally planned. ESRF, for example, does not permit groups of external scientists to build their own beamlines on its undulators however attractive the scientific case is. At APS, all beamlines are built by groups of users, some internal to APS but most external. This opportunity is vital in our case, since the instrument we are building and designing is critically dependent on using an undulator source. The design of the CXD beamline differs in significant ways from other CXD beamlines at the APS and ESRF. It is important that new design approaches be developed in these early days of CXD physics.

We are not building a general purpose beamline to cater to a wide variety of X-ray needs; the MRL already has that capability through UNICAT's sector 33 (see below). Instead, we plan a specialized instrument that will advance the technique of CXD to the limits determined by the source characteristics. We are confident that the technical capabilities outlined here can be achieved. While we cannot predict how wide an impact this capability will have on other fields of science, we have already identified a large number of potentially fruitful areas of interaction with ongoing research in the MRL. The scientific case described above demonstrates the broad-based scientific interests of the MRL users who will benefit directly from the construction of this instrument.

6. References Cited