Frontiers in Crystalline Matter: From Discovery to Technology (2009)

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 Synthesis Techniques Bulk Crystals A single crystal is called “bulk” when it can be physically separated from the growth medium and is large enough to be handled and measured independently from other crystals. Bulk crystals range in size from submillimeter to the two-foot- long crystal shown in Figure 1.1.1 in Box 1.1 in this report. Not surprisingly, a range of techniques are used to grow bulk crystals. In this appendix, the principal crystal growth techniques used in research and industry are reviewed. Floating-Zone Growth of Oxide Crystals Oxide crystals are grown by a wide spectrum of techniques, including crys- tal pulling by Czochralski (CZ) or Bridgman methods, flux growth, top-seeded solution growth, and others. Common to all of these techniques is the need for a crucible or other holder to contain the molten oxide. Ideally, this crucible will not react with the melt; in practice this is rarely true, as high-temperature oxide melts are typically quite aggressive. The optical-image floating-zone (FZ) technique offers a powerful alternative to these approaches, allowing the growth of congruently and incongruently melting materials that are both line compounds and solid solutions. FZ operates by focusing the images of halogen or xenon filaments to provide an infrared heat source that melts the feedstock polycrystalline oxide. FZ obviates the need for a container, as surface tension suspends the molten liquid between the feedstock rod and the growing crystal (see Figure D.1). Control of the processing 157 

158 Frontiers in C rys ta l l i n e M at t e r FIGURE D.1  Molten zone under stable growth conditions in a floating-zone furnace. From top to b ­ ottom: polycrystalline feed rod, molten zone, growing crystal. The image of the filament is clearly seen. SOURCE: Courtesy of Robert J. Cava, Princeton University. figure d-1.eps bitmap atmosphere is straightforward. While flux methods are more efficient for explor- ing a complex compositional phase space, FZ crystal growth is the best and most generally applicable growth method when large, high-purity crystals of known materials are required. In many cases, crystals of cubic centimeter volumes can be routinely grown. While such crystals impact the full spectrum of physical measurements, the greatest beneficiary of the FZ technique has been the neutron-scattering commu- nity. The relatively low flux of current-generation neutron sources demands large single-crystal samples, and the FZ approach has been able to deliver materials that meet this demand. Indeed, the increasing availability of large, high-quality specimens has led to breakthrough advances in fundamental neutron-scattering measurements in high-temperature superconductivity, colossal magnetoresistance and related magnetic oxide physics, ferroelectricity, multiferroics, geometric frus- tration, and many other areas. The success of future facilities, such as the ­Spallation Neutron Source, will be intimately tied to the broad availability of FZ-grown c ­ rystals. It is no wonder that Princeton University’s Robert Cava has declared the FZ technique “arguably the best thing to happen to single-crystal growth in the past 25 years” and called for a 10-fold increase in the number of floating-zone crystal growth furnaces in the United States. While the success of the optical-image FZ technique for the growth of oxide crystals speaks for itself, there are opportunities for improving this already power- ful technique to further enhance its impact. These opportunities call for research and development on FZ furnace design and operation. The areas for continuing development are as follows: 

A pp e n d i x D 159 • Homogeneity control via feedback. Owing to the extreme sensitivity of many electronic and magnetic states to composition, crystals of high com- positional uniformity are becoming increasingly important for detailed physical measurements on complex systems. Recent work in colossal magneto­resistive oxides, with magnetic phases very closely spaced in composition, attests to the importance of even tighter control on com- positional homogeneity than is at present available from commercial FZ furnaces. One potential solution is to adopt the image-processing-based feedback techniques used in the CZ growth of silicon. Sophisticated image processing of the zone shape can be used to generate an error function that feeds back to process variables. In this way, the zone volume and inter- face shape—two critical determinants of compositional ­uniformity—will be controlled with far greater precision and stability than can be done manually, which is the current state of the art. This is particularly critical in the case of incongruently melting compounds (which are the rule in complex oxides), where slow growth rates are essential and stability over long periods (days to weeks) may be necessary. • Controlled “composition spread.” The feedback approach could alterna- tively be used to deliberately stratify dopant ions in a controlled fashion along the length of the growing crystal. Proof of principle that feedback systems can be used in this mode to control dopant distribution effec- tively has already been demonstrated for laser-heated pedestal growth of Cr-doped LaAlO3. Extension of this approach to more complex oxides utilizing the FZ technique will effectively allow for phase-diagram ­studies in a single growth experiment. As oxides become more widely used as active semiconductors, composition spread samples will aid in funda- mental understanding of the doping mechanisms. • Extreme pressures. Many important materials can only be synthesized at high pressure, and the accessibility of new phases is directly related to the range of available pressures. Currently available commercial FZ furnaces can attain total growth overpressures of only about 10 bar. This over­ pressure is commonly used to suppress component volatility or to pro- mote oxidation. While this rather modest pressure has enabled growth of crystals that do not form at ambient pressure (e.g., La2-xBaxCuO4), much higher pressures would allow access to even more exotic low-dimensional and strongly correlated materials. The ultimate overpressures possible will be determined through the development of suitable transparent envelopes to contain the pressure while allowing transmission of the infrared light to the growing crystal. 

160 Frontiers in C rys ta l l i n e M at t e r Solution Growth of Crystals Solution growth is perhaps the most versatile method for growing single crys- tals. It combines simplicity, versatility, high throughput, and economy. It allows for the growth of single crystals well below their melting points, or, in some cases, decomposition temperatures. This can be critically important for the production of crystals of materials that are either very refractory, or crystals that decompose before their melting points. Further, lower-temperature solution growth methods often offer a practical route to reducing the imperfections in crystals associated with the entropic disorder or loss of stoichiometry that accompany high crystal growth temperatures. In the same way, for example, that NaCl can be grown out of water below 100°C even though it melts near 800°C, CeSb, with a melting point near 1800°C, can be grown out of Sn at temperatures below 1000°C. The solution growth technique has been used to grow single crystals of virtually every class of materials, including elements, intermetallics, oxides, chalcogenides, and organic compounds (Figure D.2). In every case, the key to using this technique lies in discovering the appropriate solution from which to grow the crystals without introducing unwanted impurities, the temperature range for cooling to promote crystallization, and the optimal cooling rate. Determining each of these ­parameters can be time-consuming and is not always possible. Given its ease and power, together with difficulties that often are solved by a combination of experience and intuition, solution growth is considered by many to be a combination of a science and an art. Solution growth of novel materials is currently practiced by numerous research groups in the United States and elsewhere. Established groups are teaching this method to a growing number of their students, and thus it is becoming even more commonly used as a research tool. Solution growth is not yet widely used in the materials physics community; owing to its increasing use, however, improvements FIGURE D.2  (Left) Crystals of Ni2SiO4 olivine grown from an Na2MoO4-Li2MoO4-NaCl solution after cooling for a month. SOURCE: Courtesy of Robert J. Cava, Princeton University. (Right) Methyl pyr- idium tetracyanoethenide grown from acetonitrile after cooling for 4 days. SOURCE: From J.S. Miller, P.M.B. Piccoli, A.J. Schultz et al., CrystEngComm, 11, 686-690 (2009). Reproduced by permission of the Royal Society of Chemistry. figure d-2.eps consists of 3 bitmap slices 

A pp e n d i x D 161 and extensions of the technique will be leveraged by its expanding number of practitioners. Advances in solution growth that are needed in the short term include broader and more practical control of crystal nucleation and of the initial growth process. This can be accomplished by the more general development and implementation of techniques that have been used for specialized crystal growth of highly optimized, technological materials such as silicon and quartz. Such techniques include the automated control of complex cooling profiles, control of thermal gradients, active mixing of the growth solution, and careful control of impurity content. Applying these techniques to the diverse range of novel and complex materials that can be grown from solution will be a substantial challenge. Longer term, the new frontiers for solution growth will be the challenges of extremes: growth of extremely high temperature materials; growth of materials with extremely volatile, toxic, or reactive constituents; and growth of materials under extremes of pressure, for example. Another frontier of solution growth is the need to grow single crystals of complex, organic crystals for use in the determination of pro- tein structure (and function). Many of the problems that exist for complex, ­inorganic compounds also are present for large organic molecules, over even ­narrower tem- perature and composition ranges. With an expanded number of practitioners, with improved, simplified, and automated methods of growth and nucleation control, and with the availability of expanded chemical and physical phases, solution growth will be one of the frontline methods for growing crystals of new materials that will be the creative engines that drive basic and applied research for the next century. Crystal Growth by Vapor Transport For materials with volatile components, the growth of crystals from a vapor phase is often an important technique that is frequently employed to make crystals for fundamental property studies. In these methods, materials in polycrystalline form are either sublimed or placed in furnaces with thermal gradients. Molecules are transported or formed from the components at the hot end of the growth system. The molecules in the gas phase travel down the thermal gradient, where, at the cold end of the system, they then condense to form crystals (in some cases, the mass transport is up the thermal gradient rather than down the thermal g ­ radient). The crystals typically grow over a period of days or weeks. Both closed (e.g., sealed, evacuated quartz tube) and open (e.g., open tube) growth systems can be employed. The same technique can be employed to grow crystals with relatively low vola- tility components, if a “transport agent” is added to the crystal growth system. This transport agent, commonly a halogen such as Cl2, Br2, or I2, reacts with the non- 

162 Frontiers in C rys ta l l i n e M at t e r volatile components, forming complex molecules that are stable in the gas phase at high temperatures. These molecules travel down the thermal gradient where they become unstable in the low-temperature part of the system, decomposing to deposit the normally nonvolatile components at the cold end, where the crystals grow. The vapor growth method has been particularly successful in growing crystals of transition-metal chalcogenides and some oxides as well as in purifying metals (see Figure D.3). The volumes of crystals grown by this method are typically much smaller than those made in the solution, flux, CZ, or FZ methods. The crystals are, however, often sufficiently large for fundamental crystal structure, spectroscopic, magnetic, and transport characterization studies. This crystal growth method therefore is especially useful in solid-state chemistry research programs whose primary goal is the discovery of new materials with interesting magnetic and electrical properties, as fundamental property studies do not typically require large-volume crystals. As in the flux methods, the optimization of vapor-phase crystal growth c ­ onditions requires substantial experience and skill on the part of the crystal grower. The magnitude of the thermal gradient, the type of transport agent employed, its concentration, the growth zone temperature, the transport gas flow rate in open systems, and the diameter of the growth tube are all variables that must be optimized. Although establishing the basic parameters of the growth process and identi- fying the best transport agents have constituted a field of significant study in past years, the technique itself has not recently been subject to concentrated study. As with other crystal growth methods, further development of the vapor growth FIGURE D.3  A crystal of cobalt chloride boracite grown by vapor transport. SOURCE: Courtesy of Robert J. Cava, Princeton University. 

A pp e n d i x D 163 technique might be made in the areas of extreme conditions of temperature and pressure. Further, for materials where vapor-phase growth is of particular interest from either a technological or a fundamental science perspective, further devel- opment of the understanding and functionality of the method could also involve increased spectroscopic analysis of the complex molecules formed in the gas phase during the transport process. Hydrothermal and Solvothermal Crystal Growth Hydrothermal growth methods have in large part been born from scientists’ attempts to emulate geological crystal growth conditions in which crystals pre- cipitate and grow from solutions based on high-temperature, high-pressure water. Temperatures of 400°C-700°C are often employed, at pressures up to a few thou- sand atmospheres, requiring high-strength steel autoclaves as containment vessels. There are relatively few practitioners of this growth technique in the fundamental materials science community at present, likely because of the specialized equipment and facilities (e.g., explosion-safe rooms) required for the growth. This method has, however, been an important part of industry-based crystal growth programs of both a fundamental and applied nature and has been exploited in very impor- tant, large-scale technological processes such as the growth of large single crystals of quartz from alkali-containing water (see Figure D.4). In the basic science arena, researchers at corporate research laboratories, particularly at the DuPont Company, have for many years employed the hydrothermal crystal growth of materials sealed in precious metal tubes such as gold and platinum as part of their exploratory new- materials programs. Small crystals of the “heavy fermion oxide” LiV2O4 have been grown hydrothermally in recent years in a Japanese academic research laboratory, but the use of this method to grow crystals for the study of basic condensed-matter physics in an academic setting is very rare. Although the hydrothermal growth method has seen relatively limited use outside industrial science programs, a related method, called solvothermal growth, does see wide application in academic solid-state chemistry laboratories. This method grows crystals from aqueous and nonaqueous solvents at typically much more modest temperatures and pressures than those needed for hydrothermal methods. Inexpensive Teflon liners simply constrained by ordinary steel vessels with simple overpressure relief systems are very often employed. Many exploratory solid-state chemistry programs use this method to grow small crystals suitable for single-crystal structure determination, and in many cases the solvent molecules are part of the crystal structure of the new compounds obtained. Solvothermal growth is also widely used to grow semiconducting nanoparticles. There is no question that the extension of solvothermal and hydrothermal methods to the growth of larger-volume crystals of a broader range of materials whose study would address 

164 Frontiers in C rys ta l l i n e M at t e r FIGURE D.4  A single hydrothermally grown quartz crystal. SOURCE: Courtesy of Robert J. Cava, Princeton University. forefront issues in condensed-matter physics would be both extremely challenging and very rewarding. Crystalline Thin-Film Materials Thin-film materials are ubiquitous in today’s world. Key examples include the semiconductor laser, the high-electron-mobility transistors found in most cellular telephones, and magnetic multilayer materials that are of pivotal importance to the data storage industry. A large variety of techniques exist for growing crystalline thin films, including molecular-beam epitaxy (MBE), evaporation, sputtering, chemi- cal vapor deposition, and laser ablation. Complex structures composed of stacked thin films with varying functionalities (electronic, optical, magnetic, and so on) are routinely synthesized. The enormous technological and economic importance of crystalline thin films has ensured that tremendous effort is applied to their growth, in the United States and around the world. 

A pp e n d i x D 165 The discussion in Chapter 1 of the fascinating states of matter that emerge in two-dimensional electron systems when subjected to a magnetic field illustrates that crystalline thin-film materials also provide great opportunities for fundamen- tal scientific research. The success of GaAs heterostructures for both science and technology provides a rationale for a similar program in MBE-grown transition metal oxide thin-film heterostructures. Because of the strong hybridization of oxygen p orbitals and transition metal d orbitals, oxides display a wider variety of strongly correlated behavior at room temperature than do covalently bonded semiconductors. The strongly coupled charge, spin, orbital, electronic, and lattice degrees of freedom offer the promise of novel functionality in thin-film devices, provided growth processes in oxides can be mastered. Recent work in Japan has shown that combining oxide functionality with the precision band-gap engineering commonly associated with covalent systems is feasible. Pursuing these goals is a national challenge of the highest order, both for science and for technology. As with bulk-crystal growth, the growth of thin films should be tightly coupled with the measurement and characterization community. In contrast to bulk growth, however, in situ studies of growth processes are common. Ellipsometry, time-of- flight mass spectroscopy, synchrotron radiation, neutron reflectometry, and other related surface-sensitive probes are critical in enabling a fundamental understanding of how the surfaces and interfaces develop in such complex systems. The integra- tion of state-of-the-art scanned probes (scanning tunneling microscopy, atomic force microscopy, magnetic force microscopy, and so on) with growth techniques to explore nanoscale phenomena in complex materials is in its early stages and severely limited by the availability of high-quality model materials systems. Today there are only a few thin-film crystal growers in the world capable of producing the kinds of heterostructure samples needed by condensed-­matter experimentalists for ­cutting- edge research on collective behavior of two- and lower-dimensional electronic s ­ ystems. The situation is made more difficult by the fact that the ultraclean samples of interest to the fundamental physics community cannot generally be grown in the same machines used for device fabrication. The high doping levels and multisource needs of the device developers lead to impurity incorporation and unacceptably low mobilities in the simpler structures needed for fundamental research. In addition, some experimental techniques such as neutron scattering require a large number of repetitions of a given multilayer structure to achieve sufficient sensitivity. Therefore, parallel tracks in growth efforts that push the limits of purity and overall sample size and those that explore new functionality are required. While the United States continues to lead in high-mobility GaAs, the situation today has grown perilously fragile, with a large number of experimental and theoretical condensed-matter physicists dependent on a very small number of growers. In the growing field of MBE oxide films, the United States will have to put forward a concerted effort to remain competitive owing to intense activity abroad. 

166 Frontiers in C rys ta l l i n e M at t e r Heterostructure Synthesis and Growth Diagnostics Thin-film approaches to crystalline matter discovery constitute a rapidly evolv- ing field. Future directions should continue to exploit the collective mastery of design, synthesis, and characterization of novel bulk and thin-film materials. Areas of continued emphasis should include both mature and emerging areas. The growth of semiconductor thin-film heterostructures, in which single-crystal mate- rial growth can be controlled at the atomic level, is at least a decade ahead of the complex oxide thin-film field in terms of the degree of sophistication (for example, materials purity resulting from the implementation of a host of in situ diagnostic tools, as well as the ability to scale to large wafer sizes). In correlated complex oxides, there has been dramatic progress in the development of thin-film deposi- tion tools such as MBE and chemical vapor deposition. Past work on thin-film h ­ eterostructures (manganites, for example) has clearly demonstrated that interface chemistry, structure, and electronic structure are acutely sensitive to lattice mis- match and surface phenomena. Advances in the growth of ­heterostructures, espe- cially the complex oxides, demands the immediate development of new approaches to in situ probes of the growth processes including the surface/interface structure, chemistry, and electronic structure. A deposition system that allows in situ p­robing of growth processes is in place at the Spring-8 beam line in Japan, where pre- liminary experiments have already been reported; some progress toward this has been made in the United States at the Advanced Photon Source (see Figure D.5), but it is clear that the U.S. researchers lag behind colleagues in Japan. Using the penetrating capabilities of x-rays along with their ability to determine surface/­ interface structure with very high precision provides an ideal way to understand the growth processes. Future deposition systems should be capable of reflection high energy electron diffraction (RHEED), ellipsometry, and surface chemical analyses FIGURE D.5  In situ studies on a synchrotron beam line. SOURCE: Courtesy of Ramamoorthy Ramesh, University of California, Berkeley. 

A pp e n d i x D 167 (Auger/x-ray photon spectroscopy [XPS], time-of-flight ion scattering and recoil spectroscopy [TOF-ISARS], and related techniques). Atomic-Layer Deposition of Complex Materials Atomic-scale control over the deposition process, namely, through an MBE-like deposition approach, has evolved considerably; however, a commercially manufac- turable process could prove to be difficult, especially if conformal deposition over nonplanar surfaces is required, as is the case for memory devices. A chemical vapor deposition process that has within it the inherent Key Scientificorder parameters abilityQuestions:atomic-layer control � Coupling of for orbital lattice (as manifested in an atomic-layer deposition process) isconversion Ferroelectricity and magnetism desirable—such � Energy extremely Spiral � Perovskite a process does not currently exist. It will require significantly better understand- Thermoelectrics and photovoltaics � Probing dynamics Bottom electrode Substrate � I. Functional Interfaces spin charge Lattice vs. electron vs. spin � ing of surface chemistry and the development of newparameters at the atomic scale?? can enable � Order precursors that the atomic-level chemical selectivity and control characteristic of an atomic-layer Energy Conversion/ Key Technical Challenges: Transduction Field-Tunable Photonic � Combination of probes deposition process. The development of such a tool would benefit greatly from Bandgap Structures Neutron, x-ray, electrons Information Storage � Thermodynamic environments Radiation Sensing � the collaborative participation of scientists and engineers fromfields Energy Storage Temperature, pressure, � Making materials and probing academia, national Bottom electrode Substrate Perovskite � II. Interface-mediated functionality Nanopillars Crystals, thin films, nanostructures laboratories, and industry. Nanodots matrix � Topological Control of Heteroepitaxial Interfaces Figure D-6 Figure D.6 illustrates two of the possible routes by which heteroepitaxy can be used to create novel interfaces. The firstres bitmap which is more common, involves low approach, with type replaced the layer-by-layer deposition of the relevant layer components (for ­ example, Key Scientific Questions orbital lattice � Coupling of order parameters � Ferroelectricity and magnetism Spiral � Energy conversion Perovskite � Thermoelectrics and photovoltaics Bottom electrode Substrate � Probing dynamics I. Functional interfaces spin charge � Lattice vs. electron vs. spin � Order parameters at the atomic scale?? Energy Conversion/ Key Technical Challenges Transduction Field-Tunable Photonic � Combination of probes Bandgap Structures � Neutron, x-ray, electrons Information Storage � Thermodynamic environments Radiation Sensing � Temperature, pressure, fields Energy Storage Bottom electrode Substrate � Making materials and probing II. Interface-mediated functionality Nanopillars Perovskite � Crystals, thin films, nanostructures Nanodots matrix FIGURE D.6  Quantum materials design algorithms. SOURCE: Courtesy of Ramamoorthy Ramesh, University of California, Berkeley. 

168 Frontiers in C rys ta l l i n e M at t e r GaAs-AlGaAs, LAO/STO, and so on) to create a sandwich-like heterointerface. The second approach consists of using self-assembly as a tool to create verti- cally epitaxial nanostructures (for example, perovskite-spinel nanocomposites). E ­ pitaxial constraints in three dimensions play a key role in the formation of such nanostructures, as well as enabling strong magnetoelectric coupling between the two phases. Understanding of the interface properties in such three-dimensional heterostructures is still in its infancy. From the point of view of the discovery of crystalline matter, it would be valuable to explore such topologies in greater detail: Can one control the architecture of these nanostructures at will? Can one create hierarchies of topologies and length scales at such interfaces? Combinatorial approaches to new materials discovery have been used to great scientific and commercial advantage in the pharmaceutical industry as a rapid means to the discovery of new drugs. Since the mid-1990s, similar approaches have been explored for the discovery of inorganic materials, with limited success, mainly in the discovery of new phosphors and catalysts.