MCC website: ©Board of Trustees University of Illinois Research Objectives: Advance multiscale spatial and time simulations via machine-learning. Approach: Use Genetic Programming – a Genetic Algorithm that evolves a program – to regress all fine- scale information from only a few direct calculations. Significant Results: Regressed all 8196 vacancy- assisted diffusion barriers at alloy surface (due to local environments) with ~0.1% error using < 3% of the barriers, regardless of type of potential! Found that less info needed with increasing complexity. Allows Kinetic MC simulation of real time via in-line function “table”, rather than standard look-up table 2,1. – 100x faster than table method during simulation. – 4-8 orders faster than “on-the-fly” type simulations. Broader Impact: Allows addressing more complexity with less information; e.g, find constitutive law in alloys 1 ; obtain accurate excited-state chemistry reactions by regressed semi-empirical potentials that rival ab initio CASSCF, for (on-going with T. Martinez, Chemistry). Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR Multiscaling using Genetic Programming D.D. Johnson, D.E. Goldberg., P. Bellon, and student K. Sastry (MatSE) 1. K. Sastry, et al., Int. J. of Multiscale Comput. Eng. (accepted). 2. K. Sastry, et al., Phys. Rev. Lett. (submitted). Potentials: Additive and Non-Additive Barrier Prediction: GP-predicted (red) vs. calculated (blue) using 3% of all 8192 barriers. Kinetic Simulation: Surface of a binary alloy with two vacancies showing first and second nearest neighbor (n.n.) diffusion paths with first (green box) and second (red box) n.n. chemical arrangements.

MCC website: ©Board of Trustees University of Illinois Objectives: Control chemically ordered phases that are commonly used in engineered alloys so as to optimize mechanical and magnetic properties. Significant Results: Our atomistic simulations 1 and analytical model 2 predict that ion beam processing, with large enough displacement cascades, can trigger the spontaneous formation of patterns of ordered domains with tunable sizes at the nanometer scale. Broader Impact: This generic phenomenon offers a new route to synthesize chemically ordered nanocomposites, e.g., for exchange-spring magnets and recording media. {111} maps of B atoms from kinetic Monte Carlo at steady state with large dense cascades at increasing disordering frequency: (a)  b = 1 s –1 ; (b)  b = 10 s –1 ; (c)  b = 100 s –1. The four colors represent the four L1 2 ordered variants that form in a A 75 B 25 alloy. Dynamical phase diagram for L1 2 ordering in A 75 B 25 alloy. Threshold for patterning of order is for a cascade size L ≈ 2nm (for Ni 3 Al atomic density).  b (s -1 ) 1 Phys. Rev. B 70, (2004); 2 Phys. Rev. B 70, (2004). (a) (b) (c) Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR NanoScale Patterning of Chemical Order Induced by Ion Beams Pascal Bellon and Student Jai Ye (MatSE)

MCC website: ©Board of Trustees University of Illinois Objectives: Research and educational networking for young scientists. Approach: The Travel Program supports US-based students, postdocs, and faculty to travel to CECAM and PSI-K activities in Europe. Younger scientists from smaller institutions are particularly encouraged to apply. The yearly budget is $23,000; the typical award is $800. Significant results: In the first 6 months, more than 40 travel applications were received and MCC supported 18 trips for scientists from 10 different institutions. Broader impact: Enhances professional development of young researchers at all levels. Participants are able to profit from relevant European activities and develop international contacts. Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR Travel Award Program for Young Scientists D. Ceperley (MCC-UIUC) Alan Aspuru-Guzik, 2004 Travel Program Awardee, at the Quantum Monte Carlo Winter School in Trieste, Italy. Alan calls his trip “an invaluable experience for my graduate education.” The Travel Program is administered by David Ceperley, along with the advisory board and the NSF.

MCC website: ©Board of Trustees University of Illinois Research Approach: Earthquakes, magnets, superconductors, and shape memory alloys (Martensites) all respond to an external driving force or field with crackling noise (avalanches). We study universal (i.e.detail independent) effects of disorder on the event size distribution. Significant results: We have found surprisingly similar behavior for magnets and earthquake faults. An extension of an existing earthquake model to include aftershocks leads to the same phase diagram for earthquakes as found earlier for a model for Barkhausen Noise in magnets (see figure to the right). This work also supported by DMR With special thanks to Mike Weissman and Jim Sethna. Phase Diagram: from Mehta, Dahmen, Ben-Zion, 2004, submitted for publication (“epsilon” is related to the amount of geometrical irregularities in the fault structure). Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR Dynamics of Disordered Nonequilibrium Systems: Hysteresis, Noise, and Domain Wall Dynamics in Systems Ranging from Magnets to Earthquakes Karin Dahmen (UIUC, MCC-UIUC)

MCC website: ©Board of Trustees University of Illinois Invited Total RegisteredNo. SpeakersTalksParticipantsDays Outreach This MCC-supported symposium, originated by Alfred Hubler in 2001, brings together researchers from many academic disciplines and industry to stimulate cross-disciplinary research activities involving complex systems. This event has continued to grow steadily each year and this fourth year we had 400 participants over 4 days. The organizers provide information about funding opportunities for complex systems research and promote linkages for interdisciplinary proposals. Broader Impact Speakers introduce key complex systems concepts in the context of their discipline. Invited plenary talks are on a 'Scientific American' level. Three hands-on tutorials are in parallel with technical sessions, covering the most recent research findings. Lectures are on-line with audio accompaniment. Integrates a diverse group of researchers and students. Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR Understanding Complex Systems Symposium – UCS 2004 Organized by Alfred Hubler (Physics) and others

MCC website: ©Board of Trustees University of Illinois Research Objectives: Improve Fe-Mn (Hatfield) steels by understanding the effects of N interstitials on crystalline fault energies and yield strength, as observed experimentally (grant DMR ). Approach: We use an electronic-structure method allowing both substitutional and interstitial disorder concomitant with stacking faults to study the effect of alloying and nitrogen addition on fault energies and associated shear barriers that determine yield strength. Combine with NSF supported experimental work to improve design. Significant Results: We find that small percentages of nitrogen improve bonding at stacking fault interfaces and stabilize the fcc phase relative to hcp, which is counter intuitive. However, while stacking fault energies decrease the shear barrier is increased – giving the observed strengthening. Broader Impact: Determined critical parameters that control increased yield-strength to improve technological Hatfield steels. Also the REU student obtained valuable research experience, while the graduate student learned to help direct projects. To be published Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR Computational Design and Understanding of Nitrogen-Steels Students: M. Curtis (REU: Olins College), D. Biava (RA: Physics) D.D. Johnson (MatSE) with H. Sehitoglu (MIE) DMR Disordered Fe 87 Mn 13 Stacking Fault Energies, , for intrinsic, extrinsic and twin faults become zero at ~4%N (above), but barriers to shear deformation increase, as is being quantitatively determining now (below), and explain the stress-strain experiments.

MCC website: ©Board of Trustees University of Illinois Objectives: Provide a resource for computational materials science (CMS) community and to foster and encourage interaction and reduce redundancy. Approach: Develop/maintain Software Archive©. Significant Results: Software repository continues development and upgrades under MCC/ITR: Authors contribute documented code and/or links. Authors retain ownership and, if desired, control over dissemination. Authors can submit reference information about publications which use their codes. Authors can view download stats over web. Archive allows specific search of type or area. User offer tips/comments/ratings (like Amazon.com®). Use continues to increase, with more that 50% of visitors requesting Archive Newsletter. Currently, the Archive has 1,070 registered users, who have downloaded more than 2,600 times. Broader Impact: (1) Encourages co-development of codes, fosters interactions and reduces redundancy. (2) Offers Summer School CMS labs for general use. (3) Offers useful tools for simulation and analysis in wide categories, e.g. DFT, QMC, instructional. Software Archive has expanded functionality. or see Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR The MCC Software Archive (additions and improvements) Amy Young, PIs: J. Kim, D. D. Johnson, D.M.Ceperley, and R. M. Martin

MCC website: ©Board of Trustees University of Illinois Research Objectives: Prediction of time-dependent optical gaps in clusters using DFT methods and excited state dynamics in molecules using Quantum Chemistry DFT methods (e.g, B3LYP) compared to ab initio CASSCF. Approach: Direct TD-DFT applications for optical gaps, with good agreement to experiment (see right). For dynamics, which is expensive, use multi-reference re-parameterized semi-empirical methods (see below). – conical intersection well-predicted. – comparable to Multi-reference-perturbation theory – more accurate than CASSCF; 1/10 th the cost! Future: For direct solution of TF-DFT in physics use discontinuous Galerkin methods developed (Haber). For reparameterize quantum chemistry potentials use Genetic Programming and avoid unphysical pathways but match ab initio database. (Johnson and Goldberg). Broader Impact: Controlled device design through science, and cross-disciplinary trained students for industry. Produce reliable integrated methods at the appropriate length/time scales. Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR New Applications/Tools: Methods for TD-DFT Applications R. M. Martin, J. R. Chelikowsky (U. Minn.), T. Martinez (Chem), D. Golberg, R. Haber, D.D. Johnson Si 123 H 100 cluster e.g., Predicted Optical Gap (in eV) versus cluster diameter (in A), I. Vasiliev, R. Martin, J. Chelikowsky

MCC website: ©Board of Trustees University of Illinois Research Objectives: Understand complex dynamics of defects in ion-implanted silicon. Approach: We combine empirical and ab inito DFT methods to reveal the complex dynamics of formations and annihilations of interstitial defects. A novel real-time multi-resolution analysis (RTMRA) technique using Wavelet compression extracts the transient structures during the thermal annealing, Significant Results: We identify new structures that are probable nucleation sites for larger extended defects. Broader Impact: ohmms (multi-scale simulation framework) incorporates new algorithms to handle long-time and large-scale simulations with a wide range of model Hamiltonians. Several materials simulation and analysis tools are developed based on ohmms class library and released under UIUC open source license. This work is done in collaborations with D. Richie (HTPi) and J. W. Wilkins (Physics, OSU). Diffusion path a tri-interstitial defect through crystalline silicon. D. Riche and J. Kim et al., Phys. Rev. Lett. 92, (2004) Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR Complexity of Small Silicon Self-Interstitial Defects Jeongnim Kim (NCSA,MCC) Data compression of RTMRA reduces the file storage by a factor 100 and reproduces the trajectory without loss. Red (black) lines are compressed (physical) trajectories.

MCC website: ©Board of Trustees University of Illinois Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR Portable Materials Simulation Toolkits PIs: J. Kim, D. D. Johnson, D.M.Ceperley, and R. M. Martin Objectives: As no standard environment exists, produce a powerful materials simulation tool to reduce researcher’s “learning curve” for useful applications of techniques/codes. Availability via Software Archive©. Approach: Combine ever-better hardware with new software engineering technology: Use of standard, open-source software for dynamic, maintainable and adaptable scientific code. Use of standard IO for communication between diverse applications. Use of standard tools ( Compilers: C/C++, OpenMP; Documentation: deoxygen; Cmake and GNU automake/libtool). Significant Results: Initial writing and testing of toolkit underway with status – OHMMS, QMC++, atomicHF, TBPW. Analysis and viz.: DataSpork© and MatSimViz© Broader Impact: Rapid integration of multiple application codes, start-up and analysis of results. Unrestricted and large use: Dataspork downloaded over 500 times. Preparing simulations Analyzing and mining data Visualizing

MCC website: ©Board of Trustees University of Illinois Research Objectives: Understand many-body effects in semiconductor quantum dots (QDs) for applications in quantum information processing. Approach: We concentrate on material and design parameters that influence the exchange interaction between conduction electrons in realistic double QDs. For this purpose, we use a combined approach based on density functional theory (DFT) to model the QD potential, and diffusion quantum Monte Carlo to simulate accurately exchange and correlation of electrons in the QD. Significant Results: We investigate quantum structures designed for three linearly coupled vertical QDs made by lithography techniques (artificial “molecules” in analogy with CO 2 or NH 2 linear molecules). DFT calculations show the design of gates is critical for electron confinement and could result in more than three dots depending of the surface covered by metal gates. Broader Impact: Controlled device design through science, and cross-disciplinary trained students for industry. Left: STM image of three linearly coupled QDs device. Right: Contour plot of the first 8 wave functions in a double dot. Localization is shown in the first four states in the center of the structure, while the fifth and sixth are localized at structure’s edge indicating presence of two other dots. R. Ravishankar. P. Matagne, J.P. Leburton, R.M Martin and S. Tarucha, Phys. Rev. B 69, (2004) Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR Spintronics in Quantum Dots J.P. Leburton (ECE) and R.M. Martin (Physics)

MCC website: ©Board of Trustees University of Illinois Potential Profile along 2D dot interface GFMC Finite Element Research Objectives: Simulation of many-body electronic effects in semiconductor quantum dots by quantum Monte Carlo (QMC) with all interactions and applied gate potentials simulated by classical Green Function Monte Carlo (GFMC). Approach: Create hybrid approach that uses QMC to solve the many-body Schrodinger equation for the electrons while at the same time using GFMC to solve the Poisson equation for electrostatic potentials and electron-electron interactions in a realistic device structure with applied gate voltages. Significant Results: We have investigated device structures designed for coupled quantum dots made by lithography techniques. The potential calculated by GFMC agrees with well-established Finite-Element method calculations (see graph at the right). QMC studies of interacting electrons have been done in simple cases and full calculations are in progress. Broader Impact: Controlled device design through science, and cross-disciplinary trained students for industry. Produce reliable integrated methods at the appropriate length/time scales. Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR New Applications/Tools: New Monte Carlo Method for Device Simulation J.P. Leburton (ECE) and R.M. Martin (Physics) Top View Side view Actual device Potential V (in meV) vs. distance (in A) along 2-D electron gas interface by GFMC and FEM solution. QMC in progress.

MCC website: ©Board of Trustees University of Illinois Research Objectives: Simulation of soft-condensed matter systems and complex fluids, mostly driven by electrostatics (colloids, polyelectrolytes, hydrogels, …). Approach: Monte Carlo and molecular dynamics techniques of particle-based, coarse-grained models. Significant Results: Triblock copolymer solutions 1 have been shown to form thermoreversible gels at remarkably low concentrations. Their dynamical properties exhibit a strong similarity to high-density glass-forming materials. Broader Impact: The model calculations lead to an improved understanding of hydrogel formation. Hydrogels have important biomedical applications, including drug-delivery devices. Outreach: Luijten organized an MCC/CECAM sponsored workshop on “Novel Simulation Methods for Soft Condensed Matter Systems” (Lyon, June 2004). Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR Multiscale Methods: Polymeric and Polyelectrolytic Materials Erik Luijten (MatSE), student Lei Guo 1 L. Guo and E. Luijten, J. Polym. Sci. B. (2004; invited)

MCC website: ©Board of Trustees University of Illinois Objectives: Education and outreach to young scientists. Approach: The 16th Annual Workshop on Recent Developments in Electronic Structure Methods was held at Rutgers, the State University of New Jersey, May 27–30, The MCC, among other groups, was a major financial sponsor of the event, and three members of MCC served on the program committee. ES04 attracted active researchers in electronic structure theory from universities, colleges, and government and industrial labs around the world. The program consisted of invited talks and contributed posters describing new methods for computing previously inaccessible properties, breakthroughs in computational efficiency and accuracy, and novel applications of these approaches to the study of molecules, liquids, and solids. MCC funds, along with other support, was used to discount registration fees and other expenses, providing travel opportunities for students and postdocs. Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR Workshop on Recent Developments in Electronic Structure Methods D.Ceperley, D. D. Johnson, R. Martin (MCC-UIUC) Significant results: More than 160 participants, % more than the previous largest meeting of the series (Princeton 2001). 50 were graduate students, who received a discounted registration fee. Presentations included 22 talks and ~70 posters, given primarily by students and postdocs. Co-supported by 1) the Office of Naval Research (ONR) and Defense Advanced Research Projects Agency (DARPA) N , and The Institute for Advanced Materials and Devices and 3) the Department of Chemistry and Chemical Biology, (Rutgers,The State University of New Jersey). Local organizing committee: Kieron Burke, Karin Rabe, and David Vanderbilt (Rutgers University) Broader impact: Enhances professional development of young researchers at all levels. Scientific discussions spark numerous new collaborations and research directions.

MCC website: ©Board of Trustees University of Illinois Objectives: Education and outreach to the computational materials community. Approach: Computational applications in nanotechnology requires working knowledge of interdisciplinary approaches, involving physics, chemistry, engineering, and computer science. The two-week Summer School on Introduction to Computational Nano-technology (organized by Umberto Ravaioli with eleven other lecturers) provided theoretical instruction and practical computational experience on a range of topics, including density functional theory and band structure calculations, numerical methods, carbon nanotubes, nanoelectronic and molecular devices, transport with non- equilibrium Green’s functions, nanofluidics and Nano-Electro- Mechanical Systems, and charge transport in ionic channels. After morning lectures, afternoons were devoted to computational laboratories, working on a variety of problems and approaches. Several computer sessions were based on software residing of the nanoHUB portal of the NSF Network for Computational Nanotechnology (NCN) at Web-published lectures (including audio), notes, and labs, from contributing Lecturers will be posted at MCC website. Co-support obtained from NSF NCN by U. Ravaioli and CRCD by D. Ceperley EE Learning by doing: Participants simulated gas diffusion in a carbon nanotubes using molecular dynamics in the lab taught by Susan Sinnott Materials Computation Center, University of Illinois Duane D. Johnson and Richard Martin, NSF DMR Summer School on Introduction to Computational Nanotechnology Umberto Ravaioli (MCC-UIUC) Significant results: The two-week school was attended by 41 US-based and 24 international participants. There were 10 women and 55 men from 30 institutions (56 graduate and 1 undergraduate students, 3 post-docs, 5 faculty).

MCC website: ©Board of Trustees University of Illinois Objectives: Identify new molecular solids analogous to the alkali-doped C 60 solids with similar electronic structure and enhanced electron-phonon coupling. Approach: Our calculations were performed using the ab initio pseudopotential density functional method as implemented in the SIESTA code: iac/siesta/ Results: Solid C 28 H 4 is found to bind weakly and exhibits many of the electronic structure features of solid C 60 with an enhanced electron-phonon interaction potential. We show that chemical doping of this structure is feasible, albeit more restrictive than its C 60 counter part, with an estimated superconducting transition temperature exceeding those of the alkali-doped C 60 solids. N. A. Romero, J. Kim and R.M Martin, Phys. Rev. B, ®, Materials Computation Center, University of Illinois Duane Johnson and Richard Martin, NSF DMR Electron Phonon Interactions in C 28 -derived Molecular Solids N. A. Romero (Physics), J. Kim (NCSA/MCC) and R. M. Martin (Physics) Isocharge density surface of face-bonded C 28 H 4 solid. Notice that the solid is held by weak bonds as evidence by the lack of bonds between the constituent molecular units. (isosurface 0.2 a.u.) Comparison of the band structure and DOS between solid fcc-Fm3 C 60 and face-bonded C 28 H 4 hyperdiamond. For the C 28 H 4 solid, the set of six bands above and below the gap are derived respectively from the three-fold degenerate LUMO and HOMO for each molecule in the two-molecule cell.