ComplexSystems

Sciencefor the

21st Century

A new world awaits our exploration. Things act differently in thisworld, but that is its attraction. It is a world of small things and of

complex things. When we understand them and cancontrol them, they will have an enormous impact in our lives.

1

Science for the 21st CenturyA U.S. Department of Energy, Office of Science Workshop

Executive Summary2

Introduction4

Collective Phenomena6

Materials by Design8

Functional Systems10

Nature’s Mastery12

New Tools14

Complex Systems

2

As we look to the next century, we find science and technology at yet an-other threshold: the study of simplicity will give way to the study of “com-

plexity” as the unifying theme.

The triumphs of science in the past century, which improved our lives immea-surably, can be described as elegant solutions to problems reduced to theirultimate simplicity. We discovered and characterized the fundamental par-ticles and the elementary excitations in matter and used them to form the foun-dation for interpreting the world around us and for building devices to workfor us. We learned to design, synthesize, and characterize small, simple mol-ecules and to use them as components of, for example, materials, catalysts, andpharmaceuticals. We developed tools to examine and describe these “simple”phenomena and structures.

The new millennium will take us into the world of complexity. Here, simplestructures interact to create new phenomena and assemble themselves intodevices. Here also, large complicated structures can be designed atom by atomfor desired characteristics. With new tools, new understanding, and a develop-ing convergence of the disciplines of physics, chemistry, materials science, andbiology, we will build on our 20th century successes and begin to ask and solvequestions that were, until the 21st century, the stuff of science fiction.

Complexity takes several forms. The workshop participants identified five emergingthemes around which research could be organized.

Collective Phenomena—Can we achieve an understanding of collective phe-nomena to create materials with novel, useful properties? We already see thefirst examples of materials with properties dominated by collective phenom-ena—phenomena that emerge from the interactions of the components of thematerial and whose behavior thus differs significantly from the behavior ofthose individual components. In some cases collective phenomena can bringabout a large response to a small stimulus—as seen with colossal magnetore-sistance, the basis of a new generation of recording memory media. Collectivephenomena are also at the core of the mysteries of such materials as the high-temperature superconductors.

Materials by Design—Can we design materials having predictable, and yetoften unusual properties? In the past century we discovered materials, fre-quently by chance, determined their properties, and then discarded thosematerials that did not meet our needs. Now we will see the advent of struc-tural and compositional freedoms that will allow the design of materials hav-ing specific desired characteristics directly from our knowledge of atomicstructure. Of particular interest are “nanostructured” materials, with lengthscales between 1 and 100 nanometers. In this regime, dimensions “disap-pear,” with zero-dimensional dots or nanocrystals, one-dimensional wires,and two-dimensional films, each with unusual properties distinctly differentfrom those of the same material with “bulk” dimensions. We could designmaterials for lightweight batteries with high storage densities, for turbine bladesthat can operate at 2500˚C, and perhaps even for quantum computing.

Complex SystemsScience for the 21st Century

Executive Summary

Materials by Design: Supramolecular assembly. Reprinted from cover with permission from Science, vol. 276, 4/18/97; ©1997, AmericanAssociation for the Advancement of Science, New York, NY.

Functional Systems —Can wedesign and constructmulticomponent moleculardevices and machines?

Materials by Design —Canwe design materials havingpredictable, and yet oftenunusual properties?

Collective Phenomena— Canwe achieve an understanding ofcollective phenomena to creatematerials with novel usefulproperties?

3

Functional Systems—Can we design and construct multicomponent moleculardevices and machines? We have already begun to use designed building blocksto create self-organized structures of previously unimagined complexity. Thesewill form the basis of systems such as nanometer-scale chemical factories,molecular pumps, and sensors. We might even stretch and think of self-assembling electronic/photonic devices.

Nature’s Mastery—Can we harness, control, or mimic the exquisite com-plexity of Nature to create new materials that repair themselves, respond totheir environment, and perhaps even evolve? This is, perhaps, the ultimategoal. Nature tells us it can be done and provides us with examples to serve asour models. We learn about Nature’s design rules and try to mimic green plantswhich capture solar energy, or genetic variation as a route to “self-improve-ment” and optimized function. These concepts may seem fanciful, but withthe revolution now taking place in biology, progressing from DNA sequenceto structure and function, the possibilities seem endless. Nature has done it.Why can’t we?

New Tools—Can we develop the characterization instruments and the theoryto help us probe and exploit this world of complexity? Radical enhancementof existing techniques and the development of new ones will be required forthe characterization and visualization of structures, properties, and functions—from the atomic, to the molecular, to the nanoscale, to the macroscale. Terascalecomputing will be necessary for the modeling of these complex systems.

Now is the time. We can now do this research, make these breakthroughs, andenhance our lives as never before imagined. The work of the past few decades hastaken us to this point, solving many of the problems that underlie these challenges,teaching us how to approach problems of complexity, giving us the confidenceneeded to achieve these goals. This work also gave us the ability to compute onour laps with more power than available to the Apollo astronauts on their missionsto the moon. It taught us to engineer genes, “superconduct” electricity, visualizeindividual atoms, build “plastics” ten times stronger than steel, and put lasers onchips for portable CD players. We are ready to take the next steps.

Complexity pays dividends. We think of simple silicon for semiconductors, butour CD players depend on dozens of layers of semiconductors made of aluminum,gallium, and arsenic. Copper conducts electricity and iron is magnetic. Supercon-ductors and giant magnetoresistive materials have eight or more elements, all ofwhich are essential and interact with one another to produce the required proper-ties. Nature, too, shows us the value of complexity. Hemoglobin, the protein thattransports oxygen from the lungs to, for example, the brain, is made up of fourprotein subunits which interact to vastly increase the efficiency of delivery. Asindividual subunits, these proteins cannot do the job.

The new program. The very nature of research on complexity makes it a “newmillennium” program. Its foundations rest on four pillars: physics, chemistry, ma-terials science, and biology. Success will require an unprecedented level of inter-disciplinary collaboration. Universities will need to break down barriers betweenestablished departments and encourage the development of teams across disciplin-ary lines. Interactions between universities and national laboratories will need tobe increased, both in the use of the major facilities at the laboratories and alsothrough collaborations among research programs. Finally, understanding theinteractions among components depends on understanding the components them-selves. Although a great deal has been accomplished in this area in the past fewdecades, far more remains to be done. A complexity program will complement theexisting programs and will ensure the success of both. The benefits are, as theyhave been at the start of all previous scientific “revolutions,” beyond anything wecan now foresee.

Nature’s Mastery —Can weharness, control, or mimic theexquisite complexity of Nature tocreate new materials that repairthemselves, respond to theirenvironment, and perhaps evenevolve?

New Tools: Instruments —Canwe develop the characterizationinstruments to help us probe andexploit this world of complexity?

New Tools: Computer-BasedTheory and Simulation —Can wedevelop the theoretical tools todescribe and predict the proper-ties of increasingly complexphenomena?

4

Nanocrystals of thesemiconductor cadmiumselenide with a radiusbetween 1 and 10 nanometerscan take two crystal forms(above), depending on appliedpressure. Because of their“low dimensionality,” theirproperties, such as meltingpoint and color, depend ontheir size. An electronmicrograph, below, showscolumns of individual atomsand facets of nanocrystalsurfaces.

ComplexSystemsThe goal of science is to understand and appreciate Nature and to use

that understanding to create materials and devices that enhanceour lives.

We can learn much from Nature. Certainly we can learn how she ap-proaches problems. We can also learn how she solves specific prob-lems—sensing, repairing damage, creating mechanical strength. Beyondthat, our study of organisms tells us what can be achieved, what prob-lems can be solved, where we can set our goals in exploiting this under-standing to benefit society.

Nature’s approach to building complex structures and functions is hier-archical. This guides us in two ways.

On a practical level, we can mimic this scheme by building our ownstructures hierarchially: combining simple molecules to fashion morecomplex ones, then combining those, repeatedly, until very complexstructures with greatly enhanced properties emerge. Collagen, for ex-ample, has a tensile strength comparable to that of steel wire, yet it isfundamentally just a single strand of amino acids linked through singlecovalent “peptide bonds.” It gains its strength, however, through theassociation of three single polymeric chains to make triple strands, whichassociate to make stronger microfibrils, which then associate to makestill stronger fibrils, which aggregate to form the collagen fiber.

At the conceptual level, Nature can be our guide in designing a multi-component research program to mirror this hierarchy. At the base ofthis program is the continuing study of simple structures and phenom-ena—continuing because there is still much to learn, and because itprovides the fundamental building blocks for the study of complexity.

While it is intuitive that large, more complicated, multicomponent struc-tures can provide enhanced properties and functions, recent discoverieshave shown that exploration of the “nano” world—structures with lengthscales between one and 100 billionths of a meter, 1000 times smallerthan a human hair—is equally important. Such structures can be madeeither through self-assembly processes or through controlled deposi-tion of the components of the material, for example, by lithography ormolecular beam epitaxy. Crystals of these sizes contain as few as athousand atoms and have remarkable and potentially valuable charac-teristics—most notably, properties that vary with the size and pattern ofthe structure. Surface effects dominate: while an ice cube melts at thesame temperature as an iceberg, a 2-nm-diameter crystal of the semi-conductor cadmium selenide melts at a lower temperature than a 4-nm-diameter crystal of the same material. These nanocrystals can be madeto emit light, and again, in sharp contrast to larger structures, the quan-tum confinement resulting from their reduced dimensions causes the

5

Collagen is constructedhierarchically by joiningamino acids into peptidestrands and bundling andcross linking the peptidestrands into molecules, themolecules into microfibrils,the microfibrils into fibrils,and the fibrils into fiberswhose strength iscomparable to steel wire.

color of the light they emit to depend on their size.Single nanocrystals have been incorporated intostructures patterned using the same lithographic tech-niques employed by chip designers and are able toact as nano-transistors, allowing the passage of onlyone electron at a time. The impact of these materialsand devices on the development of computers, whichowe their extraordinary advances to continual min-iaturization, is incalculable, although not achievablewithout much additional research.

The discovery of “bucky” balls and tubes is anothermanifestation of Richard Feynman’s 1959 predic-tion that there “is room at the bottom.” Spheres of 60carbon atoms have already been incorporated into pho-tovoltaic devices, which capture solar energy as electric-ity. Even smaller balls, containing only 36 atoms, have beenmade and shown to be superconducting when “doped” withmetal atoms. Carbon tubes of nanoscale dimensions have alsobeen made. These are the strongest known fibers and canabsorb large amounts of mechanical energy. Proposed appli-cations range from lightweight armor to hydrogen storage tobattery components.

Other one-dimensional structures include molecular wires of10-to-120-nm diameter, for example, bismuth threads embeddedin channels in an alumina film. With charge carriers confined tothis narrow “wire,” novel thermoelectric properties have been ob-served, and there is evidence that the bismuth has been converted froma semimetal to a semiconductor.

Finally, layering of nanometer-thick films of semiconductors such asindium arsenide and aluminum indium arsenide can create quantum wellsin which electrons are again confined, although this time in a two-di-mensional plane. New properties arise—electron transport is enhanced,and laser light is emitted with the generation of far less heat than inconventional lasers.

More complex are Collective Phenomena, individual, well-understoodcomponents interacting with each other to exhibit characteristics uniqueto the combination. Beyond this is the rational design and constructionof far more complicated structures. Defined function grows out of de-fined structure, the blueprint governing the location of each constituentatom. In Materials by Design, we formulate a structure atom-by-atomin order to allow it to perform a specific desired function. We then pro-ceed to synthesize it, according to that plan. These components can then,through self-assembly or directed fabrication, link in specific ways toproduce Functional Systems, motors, pumps and others that performuseful tasks. At the top of the pyramid is Nature’s Mastery, exploitingall of these stages to produce extraordinarily complex organisms, whosevery existence provides us with goals—and the knowledge that they canbe achieved.

6

Collective Phenomena

The interaction of objects obeying simple rules can produceremarkably complex and yet organized behavior. Electrons

interacting with each other and their host lattice in a solid can giverise to magnetism and superconductivity. Chemical constituentsinteracting in solution can give rise to complex pattern formationand growth. Birds flying in the sky give rise to flocking behavior.Some of these behaviors, and their relation to the underlying mi-croscopic laws, are now understood, but many of the most inter-esting and important cases remain mysteries. Learning how tounderstand and control these emergent behavior patterns will pro-vide the foundation upon which to build complex systems.

Strongly interacting electronic systemsThe origin of high-temperature superconductivity remains a mys-tery, as do many aspects of the behavior of magnetoresistive ma-terials. These materials have been extensively studied, and manyof the important parameters governing their behavior have beenexperimentally identified. However, models based on the weaklyinteracting particles have failed to describe their behavior.

—New concepts and techniques based on a radically differentpoint of view are necessary to describe these materials. Nu-merical studies on ever-larger systems will be used to supportand guide these investigations. New experimental probes willbe required to investigate both the microscopic and emergentorder in these systems.

Phase transitions and responsive systemsSystems that are close to a classical or quantum phase transitioncan exhibit large responses to small external signals. These exter-nal signals can drive the system from one phase to the other, dra-matically changing the properties of the material.

—To utilize the potential of responsive materials, a dramaticincrease in our understanding of phase transitions is neces-sary, particularly ways to harness these transitions to performuseful activities. Examples include percolative networks thatcan sense changes in temperature, humidity, or chemical en-vironment and materials that convert one type of signal to an-other, e.g., electrical to mechanical.

High-Tc superconductorHgBa2Ca2Cu3O8-δ.This materialconducts electricity without lossat temperatures up to –40˚C,higher than any other knownmaterial. The goal: room-temperature superconductors.

7

Coherence in complex systemsCoherence is central to many emergent prop-erties of collective systems, including Bose–Einstein condensation, quantum computing,and quantum interference devices.

— While coherence in a single object suchas an atom, quantum dot, or photon is nowwell understood, the behavior of collectionsof coherent interacting objects remainslargely unexplored. Exploring the uses ofcoherence in such a complex environmentis a tremendous challenge. New techniquesand approaches must be developed to cre-ate and interact with these coherent sys-tems. Improved computational techniquesmust be developed to model the effects oncoherence.

Emergent properties of active objectsInteractive, mobile objects obeying simple rulesfor interacting can generate complexbehaviors that range from flocking to spatialpattern formation. These emergent patterns in-teract strongly with their external environment.

— To understand, classify, and make use of these phenom-ena, the principles of statistical physics must be expanded toinclude systems composed of active elements. Such systemscan be the basis of smart materials that respond collectivelyto external stimuli in a coherent and coordinated fashion.

Simulation of particles in flowing liquids will aid our understanding of complex processes occurring in sedimentationcolumns, fluidized beds, slurries, and hydraulic fracturing.

Bose–Einstein condensate ofsupercooled magnetically trappedatoms. The atoms move at identicalspeeds and in identical directions,defying the laws of classical physics.It has been suggested that this couldbe the basis for lasers for use in thenanoscale sculpture of circuitry.Bose–Einstein condensate. Reprinted cover with permission fromScience, vol. 270, 12/22/95; © 1995, American Association for theAdvancement of Science, New York, NY.

Drafting Kissing Tumbling Equilibrium

8

Materials by Design

One of the most immediate and noticeable impacts of the de-velopment of complex materials will be the availability of

new materials with properties that are greatly enhanced comparedto those of materials available today. The complexity that givesrise to those enhanced properties will, at the same time, presentsignificant design and synthesis challenges.

Compositional ControlDesign and synthesis of complex materials will be radicallychanged. Design will be at the atomic level, starting with the elec-tronic structure of the elements. This fundamental knowledge willbe coupled with precise synthesis and fabrication techniques totranslate that specific design into a material. The result will be thefabrication of materials with precisely defined, predeterminedproperties.

—A thorough understanding of the properties of the elementsand, more importantly, their interactions over short and longdistances in structures, will be required for the design of ma-terials with specified properties. Targets for complex materi-als such as these include extremely lightweight batteries foruse in electric automobiles and other applications. Turbineblades that can operate at extremely high temperatures with

minimal failure rates would greatly increase theefficiency of fuel-burning engines. Quantum co-herent materials; advanced photonic and elec-tronic materials; and superhard, strong, fatigue-resistant, and corrosion-resistant materialswould also have a great impact on energy effi-ciency, pollution prevention, and informationstorage and handling.

—Development of novel systems will require so-phisticated synthetic tools. In some cases thiswill require the development of new chemistries,involving novel reactions and reaction condi-tions to synthesize structures of defined design.In other cases physical techniques such as depo-sition or lithography will be required.

Custom designed molecules (blue and pink)self-assemble to form a structure thatspecifically binds to a target molecule(green). Structures designed to be capable ofbinding each other are the basis for thedevelopment of large functional assemblies.Self-assembling custom-designed molecules.Reprinted cover with permissionfrom Science, vol. 270, 12/1/95; © 1995, American Association for theAdvancement of Science, New York, NY.

Model of a supramolecularassembly of 100 triblock polymericmolecules of defined structure.Such supra-molecular materialscould affect the technologies ofsensors, biomaterials,waveguides, lubricants, catalysts.

9

Unusual conditionsThe properties of materials are very much a func-tion not only of their compositions but also of theconditions under which they have been synthesizedand the conditions under which they are used.

—Greatly increased exploration of materialsmade or used under unusual conditions is re-quired. Techniques to create and maintainthose conditions during synthesis and tomodel and characterize the resulting materi-als must be devised and implemented. Con-ditions in question include nonequilibrium,high pressure, high magnetic field, and highenergy density. In addition, materials withlarge numbers of component elements, withcontrolled doping, chirality, and complex lay-ering will be required to achieve the enhancedproperties envisioned. Low-dimensional ma-terials—“nanostructures”—whose propertiesdo not scale with size will be synthesized andcharacterized. The extraordinary dependenceof the properties of these materials on their size provides thematerials designer with an entirely new palette of “colors” withwhich to work.

Parallel fabrication/characterizationWhile modeling techniques are being developed to allow theatomic-level design of materials with optimized properties, mas-sively parallel combinatorial approaches will be applied to iden-tify new materials.

—Systems need to be designed to synthesize huge numbers ofmaterials with slightly different amounts of each of the constitu-ent elements. These systems must also be able to achieve parallelsynthesis under a wide variety of conditions, for example, vari-able oxygen pressure in superconductor synthesis. Characteriza-tion tools capable of rapid analysis of large numbers of smallsamples are needed. The combinatorial approach is not “hit ormiss” but rather requires the same type of theoretical understand-ing as does conventional synthesis—each group of compositionsused for synthesis must be selected through modeling and theory.The breakthrough is that thousands of compounds are synthe-sized at once rather than one by one. As modeling and predictionimprove, the combinatorial approach will continue to be of value,allowing analysis of samples in narrower composition and con-dition regimes to achieve successively greater degrees of optimi-zation of properties. Theory development will benefit from theenhanced database of structures with characterized properties.

Parallel synthesis of membersof a spatially addressable libraryof high-temperature cupric oxidesuperconductors allows rapidscreening of thousands ofmaterials to identify thecompositions with optimizedproperties.Library of high-temperature cupric oxide superconductors.Reprinted cover with permission from Science, vol. 260,6/23/95; © 1995, American Association for theAdvancement of Science, New York, NY.

10

Functional Systems

Nature assembles atoms to produce materials with requiredproperties—optical, mechanical, catalytic, electrical, tribo-

logical. This entails the precise construction of those materials,with the selection of appropriate atoms and their arrangement inthree-dimensional space. Nature has also learned how to combinematerials and structures to build molecular-level machines. Someserve as pumps, moving material across barriers, often against aconcentration gradient. Others serve in locomotion, moving mol-ecules, structures, or whole cells. Others control processes actingas regulatory systems. Still others produce or transduce energy.We can learn from Nature to develop the tools to design and buildour own complex materials and machines.

Complex multicomponent structuresComplex structures will be multicomponent, yet nano- or meso-scale in dimension. Hybrid materials will combine organic andinorganic components or living and nonliving components.

—Reliable joining of dissimilar materials would allow syn-thesis of hybrid structures, with properties unachievable inless complex architectures. These could provide the criticallink between many of the new materials to be developed inthese programs and the silicon-based world of electronics andcomputing. New tools will be required to characterize thesematerials.

Building blocks/assembliesComplex, functional structures or assemblies are constructedfrom building blocks at both the nano- and the meso-scale.Structure development is driven by self-organization.

— Theoretical tools must be developed and used to de-sign the minimal set of building blocks. These buildingblocks would incorporate into their structure not onlythe required functionality, but also the interfaces thatwill allow their assembly. Again, synthesis will requirethe development of new techniques, whether chemical,involving reactions and reaction conditions, or physi-cal, involving deposition or lithographic techniques.Controlled self-assembly will require theory develop-ment and new tools to study interactions among mol-ecules. This in turn will provide the basis for designingmore sophisticated building blocks that, once as-sembled, have the required features and properties.

Multicomponent fabricatedmolecular rotor. Filament of astructural protein, actin, has beenattached to the rotating γ subunitof the enzyme ATP synthase,which has been immobilized onthe glass substrate. The filamentrotates counterclockwise with the γsubunit in increments of 120˚, onestep per molecule of ATPhydrolyzed by the enzyme.Molecular rotor. Reprinted image with permission fromScience, vol. 282, 12/4/98, page 1844; © 1998, AmericanAssociation for the Advancement of Science, New York, NY

Nanotube formed by joining an“even” rolled graphite sheet(above joint) which is predicted tobe semiconducting, to a “spiral”rolled sheet (below joint) which ispredicted to show metallicbehavior. Theory predicts thatthis would act as “nano-diode.”

11

Bacteria “swim,” propelled by theirflagella, which are spun by a motor

whose rotation is powered by theelectrochemical gradient

maintained across the bacterialplasma membrane.

Propeller-shaped single molecule“rotor” of hexa-tert-butyldecacyclene (above) within asupramolecular “bearing.” Themolecule is locked in position(left-hand images), but oncemoved to a more “open” space,using a scanning tunnelingmicroscope (STM) tip, it can freelyrotate (right).Molecular Rotor. Reprinted image with permission fromScience, vol. 282, 12/4/98, page 1844; © 1998, AmericanAssociation for the Advancement of Science,New York, NY.

MachinesFunctional “machines” are constructed as complex assemblieswhose individually inactive components work together to performa defined function. For example, the individually inactive proteinsubunits of ATP synthase spontaneously assemble to create a multi-subunit structure with channels, binding sites, and catalytic activ-ity, which functions to link the exergonic passage of hydrogen ionsthrough the channel with the endergonic catalytic synthesis of ATPat the binding sites.

— Design and synthesis of building blocks that can assembleto achieve system function require exquisite control of bothfunction and surface interaction of component structures. Tar-get functions include controlled transport across membranes,sensing of analytes, photosynthetic devices, and coatings, aswell as displays that respond to external stimuli. In the longerrun, chemical factories and pumps, actuators, and multifunc-tional electro/photonic devices would be fabricated.

12

Nature’s Mastery

L iving organisms represent the most sophisticated use of theelements to create materials, materials systems, and func-

tional complexes. They operate under the same thermodynamicconstraints that govern all materials; thus the study of Nature’sachievements can allow us to set achievable goals for artificialsystems. Enormous energy savings, pollution reduction, and pro-ductivity increases would be achieved by the development of ma-terials and systems with the enhanced properties seen in Nature,including the ability to self-assemble, self-repair, sense, respond,and evolve. In many cases, accomplishment of these goals mayappear to be beyond our reach, but no more so perhaps than opti-cal tweezers, cloning, polymerase chain reaction, or genetherapy appeared a few years ago. We may or may not be ableto reproduce all the characteristics of natural systems and tai-lor them to nonbiological needs. However, the study of Naturecan clearly guide us and lead to achievements of great valuethat would not otherwise be made. Nature’s achievements aretherefore benchmarks for our increasing control of materialsand materials systems.

Nature responds to its environmentCellular components sense their surroundings and report to othercomponents. Nature’s structures sense damage and repair them-selves. They respond to the environment, alternating among avariety of states in response to short-term changes in conditions.They evolve, altering properties to suit long-term changing envi-ronments.

—The mechanisms by which living things sense their envi-ronment and respond to it must be explored as models forartificial systems. Sensors linked to feedback systems that con-trol functions must be developed.

Nature creates complexity via self-assemblyComponents of systems are synthesized independentlyand then assemble themselves into precisely definedstructures following the laws of thermodynamics.

—The surface science of molecules and structuresneeds to be studied so that the interactions amongthem can be understood, allowing controlled de-sign of surfaces for spontaneous interaction andbinding.

Model of HSP16.5, a proteinthat helps other proteinsmaintain their proper 3-Dstructure. Its 24 identical proteinchains form a hollow spherewith eight triangular and sixsquare “windows.” Theseserve, it is thought, asattachment points for theproteins to be protected.

The intricate, controlledassembly of an inorganic shellover a precisely defined organicscaffold creates a structure thatprotects its inhabitant whileallowing it to move about andcontinue all essential lifefunctions.

13

Nature fabricates hierarchicallyNature creates building blocks designed for link-age into polymers, which join in assemblies, andeventually into functioning organs, and then intoorganisms.

—Control of self-assembly is required to al-low stepwise construction of increasinglycomplex structures. Components at each levelof the hierarchy must be designed such thatthey join together in a structure with the re-quired functions. At the same time, unwantedinteractions between incomplete structuresare blocked.

Nature discovers new structuresThrough combinatorial synthesis and selection,billions of antibody molecules are constructed si-multaneously by rearranging their buildingblocks. The “correct” structure is selected andthen copied on the basis of how well it performsin its required function.

—Theory is required to direct the selection ofcomponents to use in a combinatorial mate-rials discovery effort. Robotic mixing tech-niques and characterization tools for rapidscreening of small samples are required.

Nature uses templatesTemplates are used to build molecules for which there is no ther-modynamic or kinetic selection. DNA directs the construction ofproteins so that only the desired, functional sequences will be pro-duced.

—Templates, whether surfaces or three-dimensional structures,must be developed to control the construction of complex ma-terials or systems.

Nature uses thermodynamicsEntropy disfavors the ordering of components in a complex struc-ture, but it is, for example, the entropy of the water in which theyreside that drives the organization and the stability of proteins,membranes, and functional biological structures.

—A more complete understanding of the fundamental laws ofstructure development is required, whether it be the still in-completely understood structure of water or the folding of pro-teins. These would be used as guides for the design of artifi-cial structures that can form stable structures.

Photosynthetic reaction centeris a self-assembled complex offour proteins containing 1190amino acids, four molecules ofchlorophyll, two molecules of amodified chlorophyll, onemolecule of ubiquinone, one ofmenaquinone, and numerousiron and magnesium atoms. Itharvests light to producenutrients and energy for plants.

14

New Tools

New tools drive scientific revolutions. They allow the discov-ery of phenomena not previously seen and the study of known

phenomena on shorter time scales, at shorter distances, and withgreater sensitivity. The telescope, the microscope, the laser, x-raydiffraction, and recombinant DNA are a few of the many examples.

The study of complex systems will place extraordinary demandson both theory and modeling, and also on experimental character-ization. We need to radically improve the capabilities of existinginstruments and develop those that allow novel observations orthe manipulation of structures in new ways. Prediction of struc-tures and phenomena can direct the experimentalist. The abilityto explain the experimentalist’s data in terms of theoretical mod-els greatly accelerates the rate of discovery.

Experimental characterizationThe novel and diverse phenomena associated with strongly coupledmaterials involve a complex interplay among magnetic, charge,and structural properties, and influence spatial scales from the mac-roscopic to the atomic. Required instrumentation will necessarilyinvolve the enhancement of conventional techniques (steady-stateand time-resolved optical spectroscopies, high field NMR, ESR,mass spectrometry, scanning-probe microscopies, table-top x-ray)for detailed characterization of the diverse properties of these ma-terials with regard to shorter spatial scales, faster time scales, andproperties that emerge under extreme conditions of temperature,pressure, and electric and magnetic fields. Characterization of thesematerials will also depend heavily on revolutionary experimentaltools, including techniques for the active control of growth withparallel analysis using multiple probes and the capability to samplevery small volumes.

The high brightness, and spatial and energy resolution of synchro-tron sources provide a means for the simultaneous measurementof the macro- and microstructure of materials. Neutron scattering,reactor or spallation-based, provides a means for the study of thestructure as well the energetics of systems. Advances in x-ray andneutron detector technology should provide the means for the gath-ering of measurements on fast time scales. Advances in electronmicroscopy will shed light on dynamic structural characteristics.

Among the newly developed tools, atomic force and optical twee-zer techniques promise the investigation of the nanoforces of in-dividual atomic bonds. Spectroscopic information on single mol-ecules is becoming feasible and should also benefit from the useof high-brightness synchrotron and free-electron laser sources. We

Advanced Photon Source,Argonne National Laboratory.The APS provides highbrightness x-rays for theelucidation of detailed atomicstructure and dynamics.

The National Center forElectron Microscopy, LawrenceBerkeley National Laboratory.Its newest microscope canresolve atoms that are lessthan 0.1nm apart.

15

need to develop instrumentation and tech-niques capable of triggering, isolating, oractivating single molecules or independentlyaddressing multiple molecules in parallelprocessing fashion. We must be able totransfer one or more photons or other con-trolled forms of energy to a single molecule,as well as to trigger a molecular activator.New tools must be developed to harvest theenergy output of a molecular machine.Novel synthetic tools will include stereo-and site-specific isotopic labeling of com-plex structures for spectroscopic studies.

Theory and modelingNew generations of theory and computationaltools are required. At present, we have notheory for complex systems. We need aframework for understanding multilevel sys-tems of many interacting objects with indi-vidual states.

Theoretical and computational advances will be required to studyself-assembling systems, to predict interactions of known surfaces,and to predict surfaces for defined interactions. To design novelmaterials capable of performing interesting and well-determinedfunctions, it is critical to have a theoretical framework that definesthe minimal building blocks necessary to generate particular struc-tures. The problems encountered in predicting protein structuresfrom amino acid sequences are examples of the challenges await-ing attempts to analyze motifs with more complex sets of buildingblocks and perhaps even “active” synthetic steps.

This same type of theory and modeling will be critical for the de-sign and synthesis of functional synthetic structures. A combina-tion of simple models and detailed atomistic simulators couldjointly be used to determine the dominant energetic and entropiccomponents that control these processes.

Larger and faster computers would expand the time and lengthscales that can be achieved by first-principles simulations. Futurecomputers may also integrate hardware and software to allow cou-pling of very different computational tasks within a single frame-work (for example, allow reconfiguration of the computational ar-chitecture as a function of the dynamics in a simulation). This mayopen new avenues for simulation of multilevel dynamics. Ulti-mately, some of the complex materials to be synthesized may beused as the very hardware of future generations of computers.

Simulation, on a Cray T3Esupercomputer, of theresponse of a crystal of theamino acid glycine to themagnetic field applied duringnuclear magnetic resonancespectroscopy (NMR) studies.The field induces a current(arrows) and a charge density(colored and gray surfaces) thatdepend on the environment ofeach atom and which affect theNMR signal from that atom. Thisprovides information about thestructure of the material.

16

Workshop ChairCharles V. Shank, Lawrence Berkeley National Laboratory

Panel ChairsDavid Awschalom, University of California, Santa BarbaraMoungi Bawendi, Massachusetts Institute of TechnologyJean Fréchet, Lawrence Berkeley National Laboratory

Donald Murphy, Lucent TechnologiesSam Stupp, Northwestern UniversityPeter Wolynes, University of Illinois

Image Credits

p. 2. Top: Superconductor. ISIS. Andrew Taylor, Rutherford AppletonLaboratory, UK. Center: Supramolecular assembly. Science, vol. 276,4/18/97, cover image; © 1997, AAAS. Bottom: Nanotube. Steven Louieand Marvin Cohen, Lawrence Berkeley National Laboratory (LBNL);Vincent H. Crespi, Pennsylvania State University.

p. 3. Top: Fossilized ammonite shells. Martin Dohrn/Science PhotoLibrary/Photo Researchers, New York, NY. Center: Advanced PhotonSource, Argonne National Laboratory. Bottom: Simulation of glycinecrystal response to magnetic field in NMR spectroscopy. National EnergyResearch Scientific Computing Center (NERSC).

p. 4. Top: CdSe nanocrystals. Bottom: Electron micrograph of nanocrystal.Both images, Paul Alivisatos, LBNL.

p. 5. Collagen. Margaret C. Holm, LBNL.

p. 6. Superconductor (see page 2 reference).

p. 7. Top: Bose-Einstein condensate. Science, vol. 270, 12/22/95, coverimage; © 1995, AAAS. Bottom: Particle motion simulation. Daniel Joseph,Univ. of Minnesota, and NERSC.

p. 8. Top: Supramolecular assembly (see page 2 reference). Bottom:Science, vol. 270, 12/1/95, cover image; © 1995, AAAS.

p. 9. Library of Superconductors. Science, vol. 260, 6/23/95, cover image;© 1995, AAAS.

p. 10. Top: Nanotube (see page 2 reference). Bottom. Molecular Rotor.Science, vol. 282, 12/4/98, page 1844; © 1998, AAAS.

p. 11. Top: Flagella motor, p. 1139, Fig. 34-74, Biochemistry, DonaldVoet and Judith Voet; © 1990 by John Wiley & Sons, Inc., reprinted bypermission of John Wiley & Sons, Inc., New York, NY. Bottom: Molecularrotor. Science, vol. 281, 7/24/98, pages 531-532; © 1998, AAAS.

p. 12. Top: Fossilized ammonite shells (see page 3 reference). Bottom:Model of HSP 16.5, Sung-Hou Kim, LBNL.

p. 13. Photosynthetic reaction center. Johann Deisenhofer and HartmutMichel, Les Prix Nobel, 1989; © 1989 The Nobel Foundation, reprintedby permission of The Nobel Foundation, Stockholm, Sweden.

p. 14. Top: Advanced Photon Source, Argonne National Laboratory.Bottom. National Center for Electron Microscopy, LBNL.

p. 15. Simulation of glycine crystal response to magnetic field in NMRspectroscopy (see page 3 reference).

Front cover. Superconductor (see page 2 reference). Inside front cover.View down center of boron nitride nanotube. Steven Louie and MarvinCohen, LBNL; Vincent H. Crespi, Pennsylvania State University.

Ernest Orlando Lawrence Berkeley National Laboratory is operated forthe U.S. Department of Energy under Contract No. DE-AC03-76SF00098.PUB-826.

A Workshop on Complex SystemsLawrence Berkeley National Laboratory

Berkeley, CaliforniaMarch 5 and 6, 1999

(Above) With new tools, new understanding, and a developing convergence of the disciplines of physics, chemistry, materialsscience, and biology, we will build on our 20th century successes and begin to ask and solve questions that were, until the 21stcentury, the stuff of science fiction.(Back Cover) Dynamics of Discovery: Through computational technologies, scientists from a range of disciplines tame manypathways to discovery. Here, the streamline of fluid dynamics swirl into the smooth mathematical surface of the Mobius strip, whilearound them float molecular structures from chemistry and the double helices of biology. Robert H. Russ, an SDSC animator andartist now at Pixar, used SGI workstations and Alias/Wavefront PowerAnimator Software to create this 3-D image. Image providedcourtesy of the San Diego Supercomputer Center.