ucrl-53868-00 engineering technology reports

FY 00

UCRL-53868-00Volume 1

EngineeringTechnology Reports

Volume 1:Laboratory DirectedResearch and Development

September 2001

Acknowledgments

Scientific EditingAndrea L. BaronRobert T. LanglandCamille Minichino

Graphic DesignIrene J. Chan

Art Production/LayoutPamela A. AllenIrene J. ChanLucy C. Dobson

CompositionAndrea L. Baron

Document Approvaland Report ServicesCynthia Tinoco

Cover:Stylized graphic of an image from a CCD camera.

ENGINEERING

Pushing engineering science to the Xtreme

FY 00

UCRL-53868-00Volume 1

EngineeringTechnology Reports

Volume 1:Laboratory DirectedResearch and Development

September 2001

IntroductionSpiros Dimolitsas, Associate Director for Engineering

600-eV Falcon-Linac Thomson X-Ray SourceJ. K. Crane, G. P. Le Sage, T. Ditmire, R. Cross, K. Wharton, K. Moffitt, T. E. Cowan, G. Hays, V. Tsai, G. Anderson, R. Shuttlesworth, P. Springer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

RF Photoinjector Development for a Short-Pulse, Hard X-Ray Thomson Scattering SourceG. P. LeSage, S. G. Anderson, T. E. Cowan, J. K. Crane, T. Ditmire, J. B. Rosenzweig. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Numerical Technology for Large-Scale Computational ElectromagneticsR. M. Sharpe, D. A. White, N. J. Champagne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Automated Hexahedral Partitioning of 3-D SpaceD. J. Steich, J. S. Kallman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Lattice Boltzmann Simulation of Microfluidic DevicesD. S. Clague, E. K. Wheeler, D. Hilken . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Improved Implicit Finite-Element DynamicsM. A. Puso, E. Zywicz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Quantitative Tomography Simulations and Reconstruction AlgorithmsH. E. Martz, Jr., M. B. Aufderheide, III, D. M. Goodman, A. Schach von Wittenau, C. M. Logan, J. M. Hall, J. A. Jackson, D. M. Slone. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Photothermal Microscopy: Next Generation Nanoscale ImagingD. Chinn, C. Stolz, P.-K. Kuo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Predicting and Effecting Precise DeformationD. Swift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Updating FEM Using the Extended Kalman Filter and the Gauss-Newton SearchR. R. Leach, Jr., L. Ng, D. B. McCallen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

A Cooperative Control Architecture for Mobile Robotic AgentsR. Hills, R. S. Roberts, C. T. Cunningham . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Low-Threshold Oxide-Confined GaInNAs Long-Wavelength Vertical-Cavity LasersM. C. Larson, C. W. Coldren, S. G. Spruytte, H. E. Petersen, J. S. Harris. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

LambdaConnect: Multi-Wavelength Technologies for Ultrascale ComputingS. W. Bond, M. D. Pocha, R. R. Patel, E. M. Behymer, G. A. Meyer, P. L. Stephan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

A MEMS-Based Fuel Cell for Micropower GenerationJ. Morse, A. Jankowski. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Advanced Imaging Catheter P. Krulevitch, D. Hilken, J.-U. Kluiwstra, R. Miles, D. Schumann, K. Seward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Speckle Reduction for LIDAR Using Optical Phase ConjugationM. Bowers, C. Kecy, L. Little, J. Cooke, J. Bentereau, R. Boyd, T. Birks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Dynamic Focusing of Acoustic Energy for Nondestructive EvaluationJ. V. Candy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

ETR • LDRD • FY00 i

Selected Engineering Publications

Selected Engineering Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Author Index

Author Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

FY00 • ETR • LDRDii

IntroductionSpiros Dimolitsas, Associate Director for Engineering

ETR • LDRD • FY00 iii

In FY-2000, Engineering at Lawrence LivermoreNational Laboratory faced significant pressures tomeet critical project milestones, and immediatedemands to facilitate the reassignment of employeesas the National Ignition Facility (the 600-TW laserfacility being designed and built at Livermore, andone of the largest R&D construction projects in theworld) was in the process of re-baselining its planwhile executing full-speed its technology develop-ment efforts. This drive for change occurred as anunprecedented level of management and programchanges were occurring within LLNL. I am pleased toreport that we met many key milestones and achievednumerous technological breakthroughs.

This report summarizes our technical R&D objectives,methods and key results—as structured through ourtechnology centers.

Whether using computational engineering topredict how giant structures like suspension bridgeswill respond to massive earthquakes or devising asuitcase-sized microtool to detect chemical and biolog-ical agents used by terrorists, we have made solid tech-nical progress.

Five Centers focus and guide longer-term invest-ments within Engineering, as well as impact all ofLLNL. Each Center is responsible for the vitality andgrowth of the core technologies it represents. My goalis that each Center will be recognized on an interna-tional scale for solving compelling national problemsrequiring breakthrough innovation. The Centers andtheir leaders are as follows:

• Center for Complex Distributed Systems: David B. McCallen

• Center for Computational Engineering: Kyran D. Mish

• Center for Microtechnology: Raymond P. Mariella, Jr.

• Center for Nondestructive Characterization: Harry E. Martz, Jr.

• Center for Precision Engineering: Keith Carlisle

FY-2000 highlights

The Center for Complex Distributed Systemsexploits emerging information technologies to developintegrated systems for data gathering, processing, andcommunication, and new methodologies for assimilat-ing measured data with computational models in data-driven simulations. Effective combination of data andsimulation leads to enhanced understanding and char-acterization of complex systems ranging from largeapplied physics experiments to complex, highly hetero-geneous geologic systems associated with under-ground defense facilities and nuclear materials geologicrepositories. The Center’s LDRD activities includedevelopment of ultra wideband communication tech-nologies for enabling robust and stealthy communica-tions in a cluttered environment, and the developmentof model-based signal processing algorithms for rigor-ous updating of finite element models based onmeasured system response.

The Center for Computational Engineeringprovides for the development and deployment of soft-ware tools that aid in the LLNL engineering mission.Highlights of the Center’s LDRD projects include 1) Alinear-algebra solver effort, among the first in the worldto address the full scope of equation-solution efforts inscience and engineering. 2) A methodology forhandling smooth-surface contact-impact problems inmechanics using faceted low-order approximationtechnology commonly used in the national laboratorycommunity. This project represents a successfulattempt to combine high-order geometric response andlow-order approximation. 3) A lattice-based microflu-idics simulation tool that permits a wide range of fluidmotions to be simulated on parallel computers. Thisproject facilitates the extension of computational fluidmechanics to micro- and nano-technology systems,where the continuum approximations found in classi-cal fluid mechanics no longer apply. Also included isthe development of one of the first microfluidics toolscapable of using ASCI-class supercomputers.

The mission of the Center for Microtechnology is toinvent, develop, and apply microtechnologies forLLNL programs in global security, global ecology, andbioscience. Its capabilities cover materials, fabrication,devices, instruments, or systems that require microfab-ricated components, including microelectromechanical

systems (MEMS), electronics, photonics, microstruc-tures, and microactuators. Center staff have achievedconsiderable national recognition for the successesdemonstrated in Chem-Bio National Security Programinstrumentation, supported by the DOE and theDefense Intelligence Agency. Over the last year, nineprototypes of the handheld nucleic acid analyzer, theHANAA, were provided for beta testing by evaluatorswith a variety of applications, generally focused onbioterrorism response. Work on a microfabricatedsolid-oxide fuel-cell system, and the Lattice-Boltzmannmethod for modeling of particles in microfluidics haveled to DARPA grants. Work in mesoscale “telerobotic”manipulators has led to a spin-off company.

The Center for Nondestructive Characterizationadvances, develops and applies nondestructive charac-terization (NDC) measurement science to significantlyimpact the manner in which LLNL inspects–and,through this, designs and refurbishes systems andcomponents. The Center plays a strategic and vital rolein LDRD by researching and developing NDC technolo-gies, such as acoustic, infrared, microwave, ultrasonic,visible and x-ray imaging, to allow future incorporationof these new capabilities in LLNL and DOE programs.The near-term strategic mission objectives are toadvance Engineering’s core competencies and technolo-gies in quantitative NDC and in fast nanoscale 3-Dimaging. This year’s LDRD research contributions arein 1) quantitative tomograph simulations and recon-struction algorithms; 2) speckle reduction for LIDARusing optical phase conjugation; and 3) photothermalmicroscopy: next generation nanoscale imaging.

The Center for Precision Engineering is dedicatedto the advancement of high accuracy engineering,metrology and manufacturing. The scope of work

FY00 • ETR • LDRDiv

covers the full range of length scale, from atom-basednanotechnology and advanced lithographic technologyto large-scale systems, including optical telescopes andhigh energy laser systems. A new focus is the manufac-turing and characterization of “meso-scale devices” forLLNL’s NIF. Millimeter-scale physics experiments willprovide data about shock physics, equation of state,opacity, and other essential measurements of weaponsphysics. One of the highlights of this year’s LDRDwork is a project to devise and prove a method ofdimensional metrology for non-rigid components, e.g.,those that can deform during the inspection process.The most noteworthy results are that three dominantFEA uncertainties were identified and quantified,providing a means of improving the credibility ofanalysis results and also increasing the quantitativeusefulness of numerical simulations.

Leveraging our work

In a sense, our Centers serve as the internal venturecapitalists for our programs. They provide the mecha-nism by which Engineering can help LLNL’s programsattract funding, while pioneering the technologies thatwill sustain long-term investment.

Engineering must continually work to buildLLNL’s competitive advantage; we must continue tocreate things that are technically one-of-a-kind. OurCenters do this by fusing the best of mechanical andelectronics engineering, creating a synergy that mostorganizations cannot. Our future depends on how wefind innovative but cost-effective engineering solu-tions to emerging technical problems that lead tosolutions on a national scale.

600-eV Falcon-Linac Thomson X-Ray SourceJ. K. Crane, G. P. LeSage, T. Ditmire, R. Cross, K. Wharton, K. Moffitt, T. E. Cowan, G. Hays, V. Tsai,G. Anderson, R. Shuttlesworth, P. Springer

There is serious need for an ultrashort-pulse, high-brightness, hard x-ray source capable of probing deepinto high-Z solid materials to measure dynamic effectsthat occur on picosecond time scales. This techniquecan be used to look at phase transitions, melting andrecrystallization, and the propagation of defects anddislocations well below the surface in solid materials,phenomena that are of general interest in solid statephysics, materials science, and metallurgy, and havespecific relevance to stockpile stewardship.

We have undertaken the development of a Thomsonx-ray source, producing x rays by scattering a

short-pulse, high-energy laser pulse off a 100-MeV elec-tron bunch from the LLNL linac (Fig. 1). To enable theproduction of a suitably short pulse of x rays, the linacmust be seeded by a short pulse, high-peak-currentphotoinjector that is synchronized to the Falcon lasersystem that provides the femtosecond pulse.

In the past year we demonstrated the high-peak-current photoinjector, designed and built the photoinjec-tor laser system (PLS), and synchronized the electronbunch with the Falcon laser system. We plan to upgradethe linac, adding computer controls, and replacing olderRF drivers with newer solid-state devices; place thephotoinjector on the linac beamline and demonstrateoperation of the linac with the photoinjector as the elec-tron seed; and complete the transportation of the full

energy beam from the Falcon laser system to theThomson interaction region.

The PLS, used to generate photoelectrons for thephotoinjector, is seeded by stretched pulses from theFalcon laser system so that the linac can be synchro-nized to the Falcon master oscillator (Fig. 2).

Our ultimate goal is to scatter a 4-J, 100-fs pulsefrom a 100-MeV, 1-nC electron bunch to produce~109 photons at ~100 keV energy. We have made impor-tant progress toward this goal by completing construc-tion and testing of the 5-MeV photoinjector gun andthe UV photoinjector laser system. We attempted anexperiment to produce and diagnose soft Thomson xrays generated by scattering a weak 800 nm laser pulsefrom the 5-MeV photoinjector electron bunch. Thisexperiment has given us valuable experience inconstructing the interaction chamber, diagnosingspatial and temporal overlap between the electron andlaser pulses, and detecting Thomson x rays against alarge background of hard x rays produced from thescattered 5-MeV electrons.

In the coming year we will upgrade the Falconlaser system to 10 TW. In the following year we willconstruct the interaction chamber at the linac outputand bring the 10-TW laser pulse to collide with the100-MeV electron bunch to produce an ultrafast, hardx-ray light source for dynamic, laser-matter interac-tion experiments at solid densities.

ETR • LDRD • FY00 3

Inverse Compton scatteringof laser photons byrelativistic electrons

Relativistic electron bunchγ = Ee/mec2 = 10 to 200

Scattered photons areupshifted in energyhνscat = 2 γ2 hνlaser

X-ray photons emitted in aforward cone with angle ~1/γ

Scattered pulse durationdetermined by transit timeof laser across electrons

Ultrafastlaserpulse

Figure 1. The principle of short-pulse x-ray production via Thomsonscattering of visible photons from highly relativistic electrons.

Master oscillator

Grating stretcher

Fiber transport

Regenerative amp 4-pass

Grating compressorAutocorrelator

Figure 2. Block layout of the Falcon/Photoinjector laser showingmajor subsystems.

RF Photoinjector Development for a Short-Pulse, Hard X-Ray ThomsonScattering SourceG. P. Le Sage, S. G. Anderson, T. E. Cowan, J. K. Crane, T. Ditmire, J. B. Rosenzweig

An important motivation in the development of thenext generation x-ray light sources is to achieve ps andsub-ps pulses of hard x rays for dynamic studies of avariety of physical, chemical and biological processes.Present hard x-ray sources are either pulse-width orintensity limited, which allows ps-scale temporal reso-lution only for signal averaging of highly repetitiveprocesses. A much faster and brighter hard x-ray sourceis being developed at LLNL, based on Thomson scat-tering of fs-laser pulses by a relativistic electron beam,which will enable x-ray characterization of the transientstructure of a sample in a single shot. This source willcombine an RF photoinjector with a 100-MeV S-bandlinac. We have planned a program of beam dynamicsand diagnostic experiments in parallel with Thomsonsource development.

The LLNL Thomson scattering x-ray sourcecombines three subsystems: 1) a 35-fs “FALCON”

laser system, currently producing 0.3 J at a 1 Hz repeti-tion rate, with a planned upgrade to 4 J with a 0.1 Hzrepetition rate; 2) a photoinjector capable of producing5-MeV electron pulses with charge up to 1 nC, pulselength of 0.2 to 10 ps, and normalized, rms emittance of5 π mm-mrad; and 3) an RF linac with output energyadjustable from 30 to 100 MeV. The LLNL system willprovide a means of performing pump-probe experi-ments on a sub-ps time scale, with flux suitable forsingle-shot measurements. Single-shot measurementenables experiments on samples undergoing irreversibledamage: shocks, plasma ablation, or ultrafast melting.

The LLNL photoinjector uses the BNL/SLAC/UCLA1.6 cell, standing wave accelerator geometry. Thephotocathode surface has a measured flatness of 79 nm

peak-to-valley over a 2-in. diameter (λ/8 at 633 nm).The photoinjector is currently operated separately fromthe RF linac, and has been fully characterized using adiagnostic beamline. Characterization of the photoin-jector has included measurement of cavity Q and fillingtime, dark current, charge and quantum efficiency,Schottky scan, energy, and emittance. Emittance hasalso been characterized as a function of charge andpulse length.

The first demonstration of Thomson scattering isplanned using the 5-MeV beam directly from thephotoinjector interacting with a 30-mJ IR laser pulseproduced by the residual energy of the photocathodelaser system. The interaction with the 5-MeV beam isexpected to produce 104 to 105 600-eV photons per shotwith an interaction angle of 135° between the electronand laser beams. A drawing of the 5-MeV Thomsonscattering experiment beamline is shown in the figure.

In addition to the Thomson scattering experiment,the photoinjector and 100-MeV RF linac also providean ideal test-bed for examining space-charge inducedemittance growth effects. A new emittance measure-ment technique has been demonstrated on the RFlinac to characterize the beam produced by thethermionic injector.

The first demonstration of Thomson scattered lightfrom the linac beam and FALCON laser pulses will bethe key milestone of the project. After the Thomsonscattering demonstration at 600 eV is complete, thenext major steps are the installation of the photoinjectoron the linac, the propagation of the electron beamthrough the linac, and the characterization of the 100-MeV beam. Compression of the FALCON beam, anddelivery of the laser beam to the linac interaction regionwill proceed in parallel.

FY00 • ETR • LDRD4

Beamline for Thomson scattering at 5 MeV.

Alignment scintillator

UV laser input port Laser input port

Final focus solenoidPhotoinjector Beam dump tank

X-ray detector

Numerical Technology for Large-Scale Computational ElectromagneticsR. M. Sharpe, D. A. White, N. J. Champagne

The key bottleneck of implicit computational electro-magnetics (CEM) tools for large complex geometries isthe solution of the resulting linear system of equations.This encompasses virtually all frequency domain solu-tions of Maxwell’s equations. The mathematical opera-tors and numerical formulations used in this arena ofCEM yield linear equations that are complex-valued,unstructured, and indefinite. Also, simultaneouslyapplying multiple mathematical formulations todifferent portions of a complex problem (hybrid formu-lations) results in a mixed-structure linear system (seefigure), further increasing the computational difficulty.Direct solution methods, which are currently the stateof the art for such hybrid formulations will not scaleacceptably for ASCI-class architectures. Approachesthat combine domain decomposition (or matrix parti-tioning) with general purpose iterative methods andspecial purpose preconditioners are being investi-gated. Special purpose preconditioners that takeadvantage of the structure of the matrix will beadapted and developed based on intimate knowledgeof the matrix properties.

Initially in FY00, we established a test suite of repre-sentative linear systems to provide a test bed for algo-

rithm development and to facilitate collaboration withother research groups. The URL for this effort ishttp://cce.llnl.gov/solver.

During FY00, we began the development of a solverframework to 1) provide applications with a commoninterface to a variety of solvers and preconditionersand 2) provide developers with a flexible and extensi-ble system for incorporating new algorithms. Ratherthan start entirely from scratch, we chose to build uponthe ISIS++ Library from Sandia National Laboratories.We re-wrote the ISIS++ software so it could be used forcomplex-valued linear systems, using generic program-ming techniques to make the basic abstractions such asvectors, matrices, iterative solvers, and preconditionersindependent of data type.

We have generic (type-independent) MPI-basedparallel versions of the following Krylov methods:Conjugate Gradient, Conjugate Gradient Squared, Bi-Conjugate Gradient Stabilized, Conjugate GradientNormal Equations, Conjugate Gradient NormalResidual, Quasi Minimal Residual, GeneralizedMinimal Residual, and Flexible Generalized MinimalResidual. The modified library is known as ISIS++2.0.This framework is written in the C++ language. Tosupport legacy C and Fortran codes we developed a setof tools that automatically write C and Fortran interfaces

to the ISIS++2.0 library. These tools are written in thePerl scripting language. This way, new algorithms canbe developed in a modern object-oriented program-ming language, yet legacy codes can still take advan-tage of the algorithms. We are in the process of publish-ing these Perl scripts as an independent by-product ofour research efforts.

The next step was to incorporate solution strategiesfor the hybrid matrices. Initially, our algorithms will bebased upon applying the “best” solution procedure toeach partition of the overall hybrid system. Using thedirect solver, ScaLAPACK, only on the dense partitionof the matrix (e.g., Schur Complement method) yields asignificant memory savings. For example, for acomputer with 8 GB of memory, the direct LU approachrestricts us to problems of order 30,000 unknowns,whereas with the Schur Complement approach we cansolve systems of order 5,000,000 unknowns.

Even though this is a dramatic memory saving, theill-conditioning of the overall hybrid systems willrequire advanced preconditioners to be developed toyield a corresponding reduction in overall solutiontime. Next year’s work will focus on identifying andevaluating advanced preconditioners for the hybridsystem and demonstrating the resulting tools on ASCI-class demonstrations.


Finite element

Boundary elementCoupling terms

Coupling terms

400

500

600

200

300

100

700

800

900

200100 300 400 500 600 700 800 900

Hybrid matrix example resulting from a coupled finite element /boundary element solution of Maxwell’s equations. Blue representszero matrix values. The finite element (upper left – sparse) repre-sents a discretization of the Helmholtz equation while the boundaryelement (lower right – dense) represents a discretization of the elec-tric field integral.

Automated Hexahedral Partitioning of 3-D SpaceD. J. Steich, J. S. Kallman

Raposdy (Rapid Problem Setup for Mesh-BasedSimulation) was a joint LDRD between the Engineeringdirectorate and the Center for Advanced ScientificComputing (CASC). Our project explored automaticpartitioning of 3-D space. Once a 3-D space is parti-tioned into hexahedrals, it is straightforward to obtaina 3-D mesh using transfinite interpolation (TFI) or asimilar technique. Currently large amounts of user timeare required to generate all-hex meshes using existingsoftware mesh generation packages. The Voronoitessellation procedure to automate the partitioning of avolume has promise. Considerable progress was madeon the problem considering this was just under a one-man-year effort.

Projects like CUBIT are attempting to automate thedifficult partitioning process and are proceeding with

incremental successes. However, the search for a break-through technology to solve the problem continues.

Our accomplishments in FY00 include both 2-D and3-D explorations of a Voronoi tessellation of space thatis subsequently modified to yield an all-hex (or quad)partition of space. The partitioning approach beginswith a randomized Voronoi tessellation of space. Theresultant mesh at this point is very poor and consists ofm-sided, n-faced cells in 3-D, and m-sided polygons in2-D, with huge edge, area, and volume aspect ratios.

The next step is to successively move each tessella-tion site to its center of mass and retessellate untilconvergence is achieved. At this stage, the aspect ratiosare much improved but the resulting mesh still consistsof m-sided, n-faced cells. The last, most difficult, stageof the automatic partitioning approach is to apply a setof topological changes to the mesh using a thresholdannealing optimizer to transform the partitions into all-hexes. A big advantage of the approach is that a validmesh, which continues to become more hex-centric, is

maintained throughout the partitioning process. Thedisadvantage is that if the optimization procedure getsstuck, the resulting mesh will not be all-hex.

We implemented and tested a full 2-D implementa-tion of the above procedure and were able to automati-cally partition semi-complex shapes, although theapproach was disappointingly CPU-intensive even in2-D. Numerous variations on the optimization proce-dure were attempted.

Embedded polygons in arbitrary 2-D shapes posed adifficult challenge for the optimization. The resultantmesh did not always converge to all quadrilaterals.However, the results were encouraging and we beganworking on a 3-D implementation.

We implemented and tested a 3-D Voronoi tessella-tion/retessellation procedure as in the 2-D case. Theprocess was very CPU-intensive and required the devel-opment of complex oct- and binary-space-partitioningtrees along with a vector quantization process to reducethe required CPU loads. We did not pursue the thresholdannealing process in 3-D because we wanted thethreshold annealing optimization moves to keep facesplanar as in the Voronoi process. This constraint limitedpotential moves available to the annealing process.

We arrived at a way to modify the mesh that inmany circumstances immediately yielded an almostall-hex partitioning of the volume. The procedureinvolved adding nodes to each face and locally cuttingeach Voronoi cell into a set of local hexahedrals. Theprocedure we developed is guaranteed to produce anall-hex mesh provided each node of the Voronoi cell isattached to three of the cell's faces, which is often, butnot always, the case.

It may be possible to add additional tessellation sitesat all nodes, violating the above restriction to guaranteean all-hex mesh in all situations, or to use the thresholdannealing process as in 2-D but removing the planarityrestriction. These are areas for future research.


Lattice Boltzmann Simulation of Microfluidic DevicesD. S. Clague, E. K. Wheeler, D. Hilken

The design and manufacturing of microfluidics andbiological microfabrication and microelectromechanical(Bio-MEMS) devices at LLNL have application to anumber of important programs of national interest. Forexample, biomedical and biotechnology applicationsinclude subsystems designed for preparing miniatur-ized samples and for detecting target species. However,before such devices can achieve desired functionalityand ultimately lead to potentially new technologies,many challenging scientific issues must be addressed.

Our project focuses on developing simulation toolsto augment and guide microfluidic device designs.

Specifically, we are developing lattice Boltzmann (LB)simulation capabilities to study the transport propertiesof particulate species in microflows. At micrometer-length scales, short-range forces become important. Asa result, we are developing simulation capabilities tostudy coupled physical effects.

During FY00, we focused on the coupling betweenhydrodynamic and dielectrophoretic (DEP) forces. Weperformed LB simulations to explore inertial lift effectsas a function of Reynolds number, and coupled inertialeffects with DEP forces for systems with length scalesof actual interest.

Figure 1 shows the extent of lift as a function ofReynolds number; Fig. 2 shows a DEP capture. In Fig. 1,the lift height is made dimensionless with the sphereradius, and the Reynolds number is based on themaximum fluid velocity. According to Saffman’s

theory (1964), the lift force is proportional to themagnitude of gradient in the velocity profile of thefluid. As the Reynolds number increases, the gradientin the fluid velocity becomes steeper, and the theorypredicts an increase in the lift force. Typical Reynoldsnumbers range from much less than 1 to about 5 inmicrofluidic systems. As expected, our LB simulationscapture the weak inertial forces and predict an increasein lift with an increase in the Reynolds number.

As depicted in Fig. 2, we built a time-averaged elec-tric field into the LB capability to study coupled hydro-dynamic and DEP forces on suspended species. AsFig. 2 shows, the sphere initially experiences inertial liftforces; however, when it encounters the electrode array,it is pulled to the electrode surface and captured. Often,device designers want to position target species in thecentral region of microchannels for detection purposes.Our results in this project now enable the rapid predic-tion of optimal flow rates for either suppressing orusing inertial lift effects to cause the desired reposition-ing of the particles.

In FY01, we will 1) continue investigating the forcesinfluencing DEP manipulation in flow conditions thatare relevant to LLNL’s device designers, and 2) explorethe inclusion of representative particle geometries inthe LB capability. In particular, we will extend our abil-ity to study the transport behavior of macromoleculesusing ellipsoidal and bead-and-spring models, whichare representative of proteins and long-chain polymers(e.g., DNA fragments), respectively.


0 0.2 0.4 0.6 0.8 1.0

Hei

ght

(sph

ere

radi

i)

Reynolds number

10–4

10–3

10–2

10–1

10

1

Figure 1. Inertial lift as function of Reynolds number.

0 50 100 150 200 250 300 350

Vert

ical

pos

itio

n (m

m)

X-position (mm)

5

6

7

Figure 2. Coupled inertial lift and DEP capture. The rectanglesand the sphere represent an electrode array and a target species,respectively.

Improved Implicit Finite-Element DynamicsM. A. Puso, E. Zywicz

Numerical difficulties encountered with implicit analy-sis have forced the use of explicit finite-element codesfor many problems in mildly nonlinear, static structuralmechanics. This is unfortunate, because implicit meth-ods promise better accuracy for nonlinear statics andfor long-time, nonlinear dynamics problems such asseismic events, reentry vehicles and transportationcontainer vibrations.

The goal of this project was to develop implicit,finite-element algorithms that solve these problems

reliably by attacking the two main obstacles inherent tothe implicit method: 1) for nonlinear problems, the clas-sical implicit-integration schemes are numericallyunstable, and 2) convergence of the implicit, nonlinear-

solution algorithm is severely affected by unilateralconstraints on contact surfaces.

During FY00, we developed time-integration schemesthat are unconditionally stable for the general setting ofcoupled (flexible) finite-element and rigid-body analysis.For example, we used a random input to excite a vibra-tion in the FL transportation container shown in Fig. 1.Fig. 2 shows a comparison of the simulated kineticenergy using the old method with that using our newmethod. Gaps between the foam and cylinder wallscause contact-impact during the motion. With the classi-cal trapezoidal simulation, this contact nonlinearitycauses the energy to grow. However, with our newmethod the energy is controlled. Our work is the first forthe general 3-D multibody implicit-dynamics setting andhas been the subject of three journal publications.

Contact surfaces are used to interface disjointedobjects that impact during analysis. This contact inter-action is highly nonlinear and is the major cause ofconvergence problems. In particular, the non-smoothcontact surface causes jumps in contact forces at facetedges and vertices when objects slide past each other.To eliminate this problem, 3-D Gregory patches wereused to smoothly interpolate the bilinear surface meshso that objects will now slide smoothly past each other.

The major technical hurdle overcome in FY00 wasthe ability to smooth arbitrary, irregular meshes.Smoothing allows the solution to many problems thatwould have diverged in the past, and is the topic of arecently submitted journal publication.


Payload

Inner cylinderand caps

Outer cylinderand caps

Foam

Figure 1. FL transportation container.

0

50

100

150

200

0 0.4 0.8 1.2 1.6

New method

Old method

Kin

etic

en

erg

y

Time

Figure 2. Simulated kinetic energy comparison.

Quantitative Tomography Simulations and Reconstruction AlgorithmsH. E. Martz, Jr., M. B. Aufderheide, III, D. M. Goodman, A. Schach von Wittenau, C. M. Logan,J. M. Hall, J. A. Jackson, D. M. Slone

X-ray, neutron and proton transmission radiographyand computed tomography (CT) are important diag-nostic tools that are at the heart of LLNL’s effort tomeet the goals of the DOE’s Advanced RadiographyCampaign. This campaign seeks to improve radi-ographic simulation and analysis so that radiographycan be a useful quantitative diagnostic tool for stock-pile stewardship.

Current radiographic accuracy does not allow satis-factory separation of experimental effects from the

true features of an object’s tomographically recon-structed image. This can lead to difficult and some-times incorrect interpretation of the results. By improv-ing our ability to simulate the whole radiographic andCT system, it will be possible to examine the contribu-tion of system components to various experimentaleffects, with the goal of removing or reducing them.

In this project, we are merging this simulation capa-bility with a maximum-likelihood (constrained-conjugate-gradient (CCG)) optimization techniqueyielding a physics-based, forward-model image-optmization code. In addition, we seek to improvethe accuracy of CT from transmission radiographs bystudying what physics is needed in the forward model.

During FY00, an improved version of the LLNL ray-tracing code called HADES has been coupled with arecently developed LLNL optimization algorithmknown as CCG. The problem of image reconstruction isexpressed as a large matrix equation relating a modelfor the object being reconstructed to its projections(radiographs). Using a CCG search algorithm, a maxi-mum likelihood solution is sought. This search contin-ues until the difference between the input measuredradiographs or projections and the simulated or calcu-lated projections is satisfactorily small.

We developed a 2-D HADES-CCG CT code that usesfull ray-tracing simulations from HADES as the projec-tor. Often an object has axial symmetry and it is desir-able to reconstruct into a 2-D r-z mesh with a limitednumber of projections. The physics (e.g., scattering anddetector response) required in the HADES code isdetermined from Monte Carlo simulations. The currentversion of HADES-CCG reconstructs into a volume-density mesh made of one material and assumes amonochromatic source.

Both 10-MeV neutrons and 9-MV x rays were usedto image the British Test Object (BTO). This objectconsists of a set of nested C, W, and polyethyleneshells and is used for tomographic algorithm verifica-tion. The neutron and x-ray radiographs (figure, leftside) of the BTO show some of the joints and differentmaterial details. However, it is difficult to use the radi-ographic projection data to obtain detailed quantita-tive measurements of features such as joint thicknessand material boundaries. To extract more details fromthe BTO radiographs, a single radiographic projectionwas used to obtain a 2-D CT cross-section of the BTO(figure, right side). These cross-sections were obtainedusing the current HADES-CCG code. Comparing thex-ray to the neutron CT reconstructions on the right ofthe figure, it is apparent that spatial resolution is supe-rior in the x-ray image. This is expected and is largelydue to source unsharpness.

Our current implementation of HADES-CCG hasshown interesting potential, but more work is neededto make a fully viable reconstruction code. In FY01 wewill study the treatment of multiple materials andpolychromatic sources in a reconstruction. We willexpand the HADES-CCG code to include 3-D recon-structions in FY01, broadening its applicability.


Top left: 10-MeV neutron radiograph of the BTO. Top right: 2-D CTreconstruction of the BTO using the single radiograph shown to theleft. Bottom left: 9-MV x-ray radiograph of the BTO. Bottom right:2-D CT reconstruction of the BTO using the single radiographshown to the left. The dotted box highlights the portion of the radi-ograph used to obtain the 2-D cross-sectional CT data in each case. Both CT images were reconstructed into the same voxel size of0.5 mm x 0.5 mm.

Photothermal Microscopy: Next Generation Nanoscale ImagingD. Chinn, C. Stolz, P.-K. Kuo

As structures move to smaller scales, higher resolutioncharacterization techniques are needed. Photothermalmicroscopy (PTM) is a powerful tool for thermal non-destructive characterization of surface and subsurfacefeatures with micrometer spatial resolution. PTM offersa higher spatial resolution alternative to infrared imag-ing. PTM was originally developed as a tool for detect-ing optical absorption and micrometer-sized thermalinhomogeneities in laser optics. The goal in FY00 wasto increase spatial resolution of PTM to 100 nm.

Aphotothermal tool with nanometer resolution is ofgreat importance in the study of 1) energy deposi-

tion and transfer of nanoscale structures such as high-power laser diodes and laser optics; 2) defect detectionin coatings such as silicon multilayers; and 3) materialsscience and manufacturing research. Although thehigh-power laser projects at LLNL are the direct moti-vation for this development, applications of this nonde-structive characterization tool extend far beyond laserdamage to NIF optics. Photothermal and thermo-elasticproperties at micrometer and nanometer scale are ofgreat importance to general material science and manu-facturing research, for large aperture laser optics aswell as nanometer microelectronic devices.

PTM uses scanning pump and probe lasers with aphotodetector to measure thermal absorption of mate-rials. The pump laser heats the test object, while theprobe laser and photodetector measure the diffractionof light resulting from surface deformation as the heatis absorbed. The spatial resolution of PTM is limitedby the pump laser beam size, which can be focuseddown to the diffraction limit (~1 µm), as described byclassical optics.

To increase spatial resolution, implementation ofnear-field optical techniques to PTM has been

investigated. This year, we installed and evaluateda near-field scanning optical microscope (NSOM)for use in nanoscale photothermal experiments.NSOM produces sub-wavelength resolution opticalimages by scanning an optical fiber over materials.PTM at nanoscale resolution entails using theNSOM fiber probe as the PTM pump laser. AnNSOM probe will be used in the PTM pump-probeconfiguration to achieve nanometer PTM.

We purchased an Aurora NSOM manufactured byThermoMicroscopes. This NSOM features a scanningstage supported at three points by piezoelectric tubes.The three-point support configuration enables collectionof light through a sample in transmission mode. Havingaccess to the sample from beneath the scanning stage isalso beneficial for positioning a PTM probe laser.

The Aurora NSOM stage has a tuning fork oscillatorattached to the fiber tip that serves as a proximity indi-cator to sense the distance between the fiber tip andthe sample surface. When the tuning fork is driven atits resonant frequency by a lock-in amplifier, the pres-ence of the sample surface within nanometers of thefiber tip increases damping. A feedback loop from thelock-in amplifier controls the vibration of the tuningfork and keeps the fiber tip at a constant distance fromthe sample.

In evaluating the NSOM, we found that the feed-back control system can be further optimized by vary-ing the phase of the lock-in amplifier. Currently, thelock-in amplifier uses the in-phase response. Thisimprovement to the feedback control system will givemore precise fiber-part offset distances.

In FY01 we plan to complete the necessary modifica-tions to the NSOM so that nanometer resolution PTMcan be tested. Additionally, we will develop a camera-based imaging system that will increase the speed ofmicrometer resolution PTM.


Predicting and Effecting Precise DeformationD. Swift

In traditional dimensional inspection, objects areassumed to be essentially rigid. Measurements aretaken under a set of constraints that have minimaleffect on the overall uncertainty of the measurement.The results, along with a statement of uncertainty, arecompared to a design specification. When the object tobe measured is not rigid, the contribution of the bound-ary conditions to the uncertainty in the measurementresult becomes significant. Additionally, the dimen-sional characteristics of the fixtures and supportsbecome confounded with the characteristics of theobject being inspected and thus the true shape of theobject can not be determined. Finally, if the non-rigidobject is put into use under a set of boundary condi-tions that differ from those present during inspection,the in-use shape is unknown.

Hemi-shells, KDP crystals, EUV Lithography masks,and automobile body parts are all examples where

non-rigid components are used under a different set ofboundary conditions than those present during inspec-tion. To extend the precision to which these objects canbe specified, inspected, and assembled, a new tech-nique in dimensional inspection must be developedand used.

The goal of this project is first to understand methodsof restraining or actuating non-rigid objects so that theirresulting deformation is repeatable and predictable.While this may appear to be a straightforward analysisproblem, it is subject to many sources of error that mustbe reduced to apply such analysis with high precision.The second goal is to develop a methodology fordesigning restraints or actuators that produce apredictable deformation on the surface of a non-rigidobject and can be modeled analytically with minimumuncertainty. This technique will be applied to a non-rigid cylinder first (see figure) and then extended toinclude other shapes in the future.

The most noteworthy results from the FY00 effort arethat three dominant finite-element analysis uncertain-ties were identified and quantified. The first was the useof an idealized geometry versus the exact geometry forthe analysis. It was found that using the idealizedgeometry introduced about an 8% error in the result.The second dominant uncertainty was the material

properties. We found that uncertainties in the materialproperties (e.g., elastic modulus, Poisson's ratio,orthotropy) were directly proportional to uncertaintiesin the analytical result. This uncertainty amounted toabout 0.4% in the analytical result. Finally, it was deter-mined that contact at the restraints was difficult tomodel and characterize and thus contributed to about a14% error in the result. Because the development ofcontact algorithms is beyond the scope of the work, wewill confine our measurements and analytical compar-isons to the regions of the cylinder that are outside thearea affected by localized contact.

The FY01 goals of this project are to complete themeasurements on the thin cylinder and compare themto the analytical results under various displacementand force boundary conditions. From there, the tech-nique will be extended to a thin hemispherical shell.Additionally, the methodology for designing restraintsand actuators will be formalized into a set of designguidelines. These guidelines will serve as the founda-tion for designing fixtures, mounts, and otherconstraining devices for non-rigid objects. Finally, a setof metrics will be developed that are function-basedrather than geometry-based, to aid in the specificationand acceptance of non-rigid objects.


Copper cylinder in measurement fixture.

Updating FEM Using the Extended Kalman Filter and the Gauss-Newton SearchR. R. Leach, Jr., L. Ng, D. B. McCallen

The motivation for this work is to provide a method toupdate Finite Element Models (FEM) of large struc-tures using state-space-based signal processing tech-niques. Two different methods, extended Kalmanfilter, and Gauss-Newton gradient search have beenexplored. We have made progress in each type of algo-rithm. We have achieved extremely accurate stiffness,displacement, and velocity estimates for most of thesensor placement scenarios in a five-story buildingwith a simple 10 degrees of freedom (DOF) model.

The primary goal of this project is to develop andapply state-space-based signal processing tech-

niques to locate the existence and type of modelmismatches common in FEM.

A simple, 10 DOF FEM of a five-story building isshown in Figure 1. Each floor is represented as a nodewith mass M, and the nodes are connected by singlecolumn elements of stiffness K and damping C. Thestiffness elements, K, are estimated by the Kalman filter,the mass and damping values are considered constant.Measurement occurs at the nodes, while the stiffnessesof the columns are the quantities to be identified.

A force was applied to Node M1 consisting ofGaussian noise (0 to 50 Hz). Initial stiffness parame-ters were set to 7/8, 3/4, 9/10, 1/3, and 1/2 of K,

representing a perturbed or “damaged” structure. Inthe first experiment, the model was fully measured andthe stiffness identification errors were essentially zero.Other experiments were then completed (27 total) vary-ing the number of sensors and their locations.

The Kalman filter converged very near to the correctstiffness in most experiments with two or more sensorsor measurement points. This is particularly promisingin that estimates in many cases converged to the correctvalue even with floors having poorly modeled stiffnessparameters. The largest error tended to occur when theuppermost floors were measured (farthest from thevibration source).

The Kalman filter state-space-based parameter identifi-cation method works relatively well in this simple 10 DOFidentification problem, does not require excessive compu-tational time, and performs consistently for situationswhere only partial measurement of the system is available.

In the second approach, an iterative Gauss-Newtongradient search routine is used to minimize the residu-als in the least squares sense.

Results for both methods are shown in Figure 2.Each column of five boxes represents one uniqueexperiment. The numbers at the top of the boxes corre-sponds to the average error of the five floors for boththe Gauss-Newton method and the Extended KalmanFilter method. A Gaussian noise input was applied tothe first floor in all cases.

These results indicate that these methods can beused to update FEMs. They can also be used to identifyoptimum placement and density of sensors. Finally, wehave provided a technique to detect damage in previ-ously tested structures.

This project is completed, allowing us to movetoward practical application of this method, e.g.,conducting experiments using a shaker table andaluminum tower fixed to the shaker table for controlledtesting. There is also potential to extend this work tohigh-DOF (400 or more) simulated systems.


C 5

C4

< Element< NodeC 3

C 2

C 1

M5

K5

M4 < NodeK4

M3 K3

M2

K2

M1K1

Figure 1. Ten DOF FEM of a five-story building.

Mean error for extended Kalman filter (recursive)Mean error for Gauss-Newton (iterative)

EKF

EKF

GN

GN

Measured

Estimated

16.8

17.6 14.11.33 2.63 1.97 2.95 1.84 2.03

0.00 0.49 0.25 0.22 0.14 0.12 32.00.21 0.160.02

4.83 8.53 3.28

7.47

0.21 0.36 0.46 0.29 0.67 0.19 80.5 0.28 4.69 26.8

5.307.46

3.65 2.92 3.32 2.89 2.64 3.90 9.4215.9 8.08 11.2

13.7 8.29

43.247.188.818.3 179 6.6 11.1

Figure 2. Mean identification error for 5-floor 10-DOF structure using Gauss-Newton method (red bars) and the extended Kalman Filter (blue bars).

A Cooperative Control Architecture for Mobile Robotic AgentsR. Hills, R. S. Roberts, C. T. Cunningham

Large networks of land-based sensors are becomingincreasingly important for sensing natural and man-made phenomena. The deployment and operation oflarge, land-based sensor networks can pose difficultproblems, particularly in time-critical situations or inrugged terrains. An example scenario would be one inwhich ground conditions must be monitored in a forestto enable modeling and predicting the advance of aforest fire. Autonomous deployment and operation ofland-based sensor networks in such a scenario are highlydesirable for maximizing sensing efficiency while mini-mizing human risk. Other national security issuesinvolve remote monitoring of sites and facilities (forexample, chemical and biological manufacturing facili-ties), using signatures to determine the manufacturingstate of the facilities for counterproliferation operations.

Our approach to the deployment and operation of aland-based sensor network is to use cooperating,

unmanned air vehicles (UAVs) that deploy the sensorsand then serve as communication hubs for the sensors.In this approach, a group of cooperating UAVs deployssensors over a region of interest. After the sensors havebeen deployed, the sensor network is logically parti-tioned into subnetworks (subnets), with one UAVassigned per subnet. Partitioning the network intosubnets allows the UAVs to service sensors in parallelwhile minimizing interference or duplication of effort.A UAV services the sensors in its subnet by flying aroute (path) through the subnet, uplinking datacollected by the sensors, and forwarding the data to acentral point. Cooperation among UAVs is maintainedthrough the exchange of state information amongUAVs, information that includes not only the status ofthe UAVs but also the status of sensors in the network.In this manner, exceptional events (such as a UAV leav-ing the network for fuel replenishment or the failure ofa group of sensors) can be detected and properlymanaged by the UAVs.

During FY00, we focused on a key component of thearchitecture—the adaptive path-planning algorithm.This algorithm initializes the paths and then adapts thesubnets (and paths through the subnets) in response toone of four basic exceptions: 1) one UAV leaves the

network, 2) one UAV enters the network, 3) severalsensors leave the network, and 4) several sensors enterthe network. More complex situations can be derivedas combinations of these basic exceptions.

The algorithm models the sensor network as a set ofdirected graphs. In this model, the sensors in a subnetare nodes of a graph, and the UAV path through thesubnet is a set of directed links between the nodes.Several approaches have been proposed to find pathsthrough a graph, but these algorithms find the shortestpath between two nodes, not a closed path. For the caseof one UAV, the problem reduces to the “travelingsalesman problem.” However, we are interested inapplications where multiple UAVs are required toproperly service the network.

In the algorithm developed this year, subnet andpath adaptation is driven by a global cost function thatshifts sensors into and out of subnets to reach a mini-mum cost. Our heuristic approach—in which we usegood approximate solutions and make only minoralterations to them—is quite different from conven-tional techniques, which consider much more generaldistortions of existing paths. For example, we foundthat our method performs significantly better thansynthetic-annealing methods used in the past.

In FY01, we will focus on several aspects of thecooperative control architecture, including excep-tion detection, exception handling, and communica-tion networking.


0

7

3

5

91

84

6

2

Paths and subnets for the case of 124 sensors and 10 UAVs. Thesensor locations were determined using a digital terrain map.

Low-Threshold Oxide-Confined GaInNAs Long-Wavelength Vertical-Cavity LasersM. C. Larson, C. W. Coldren, S. G. Spruytte, H. E. Petersen, J. S. Harris

For the first time, we have reported room tempera-ture CW operation of GaInNAs vertical-cavitysurface-emitting laser (VCSEL) diodes, emitting at1200 nm, and grown all-epitaxially in a single step ona GaAs substrate.

Because of its low-cost wafer-scale fabrication andease of packaging and coupling to optical fiber, the

VCSEL (Fig. 1) is an important optical source technol-ogy for short-haul optical fiber transmission systems.Dramatic enhancements in bandwidth and distancecan be achieved in conventional single- and multi-mode fiber by extending the VCSEL emission wave-length to the 1300 to 1550 nm range. We have takenadvantage of the properties of GaAs-based materials—epitaxially-grown thermally-conductive high-contrastmirrors and AlAs-oxide current apertures—to demon-strate low-threshold VCSELS with CW room tempera-ture operation.

Measurements of the output power and voltage vs.injection current for a device with a 5-µm-x-5-µmaperture show the threshold current at approximately1.3 mA, and the slope efficiency at 0.045 W/A. Anextremely high threshold voltage (10.3 V) resultedfrom unoptimized doping and composition-gradingprofiles at the heterointerfaces of the p-DBR (distrib-uted Bragg reflector), creating a large degree of self-heating, which limited the maximum output power to0.080 mW at 3.8 mA. The fact that CW operation waspossible in spite of this excess power dissipation isindicative of the high gain and low temperature sensi-tivity of the GaInNAs multiple quantum well (MQW)active region.

The emission spectrum (Fig. 2) at threshold, indicatedthat the device lased in a single transverse and longitudi-nal mode. Above threshold the spectrum exhibitedmultiple peaks with approximately 0.4-nm spacing andan envelope that broadened with current, which waslikely a result of thermally-induced chirp during thecurrent pulse, in conjunction with optical feedbackinduced by the reflection from the substrate backside.

CW laser operation also occurred for device sizesranging from 3.6 µm to 6.4 µm, with threshold currentsfrom 0.94 to 2.3 mA and slope efficiency as high as0.049 W/A.

Devices were also tested under pulsed operation toprobe characteristics as a function of aperture sizeindependent of thermal effects. Pulsed thresholdcurrent varied from 0.89 mA for 3.6-µm devices, toapproximately 21 mA for 29-µm devices. Slope effi-ciency dropped for aperture sizes below 7 µm, likelydue to optical scattering, and current density rose tonearly 7 kA/cm2 for the smallest devices.

Future improvements in performance should bereadily achieved by further adapting well-established850-nm VCSEL technology. Higher output power willbe possible by reducing the resistance of the p-DBRthrough finer control of composition grading anddoping, and lower threshold current by reducing scat-tering in small-aperture devices. This will also improvehigh-temperature performance through reduced self-heating. Finally, 1300-nm emission should be achievedby increasing the In and/or N content of theGaInNAs/GaAs MQW active layer.


Ti-Au

Substrate emission

n+ GaAs substrate

20 pairp-GaAs/p-AIAsDBR

22.5 pairn-GaAs/n-AIAsDBR

GalnNAs/GaAstriple quantum well active region

50 µm

Figure 2. Emission spectrum near lasing threshold.

Inte

nsi

ty (

a.u.

)

Wavelength (nm)1190 1195 1200 1205 1210

Figure 1. Schematic of GaInNAs VCSEL.

LambdaConnect: Multi-Wavelength Technologies for Ultrascale ComputingS. W. Bond, M. D. Pocha, R. R. Patel, E. M. Behymer, G. A. Meyer, P. L. Stephan

Ultrascale computing—the integration of large numbersof processors into a single, highly capable multiprocessorsystem—is currently of great interest for several nationalsecurity missions. Scalability of these systems, withupwards of hundreds of CPUs, is hampered by thecommunication bottlenecks imposed by latency, band-width, and congestion in electrical interconnection andswitching networks. To alleviate these bottlenecks, we areleveraging the recent emergence of low-cost, byte-wideoptical components containing linear arrays of multi-mode optical fibers in ribbon cable assemblies. We haveshown that the addition of multiple wavelengths overthis byte-wide medium can enable source-routed opticalswitching, which greatly relieves communication conges-tion and improves bandwidth.

Implementation of optical source-routed communica-tions fabrics requires the development of novel

components, including transmitters capable of fastwavelength tuning and fixed-wavelength opticalfilters, all compatible with multimode fiber ribboncables. The system uses an N-x-N optical star coupleras a broadcast element, with the byte-wide wavelengthselectable transmitters as inputs, and fixed-wavelengthoptical receivers as outputs. Switching is achieved bypredetermining the output port by appropriate selec-tion of the transmitter wavelength.

This project has focused on the development of proto-types of these components.

We have developed passive alignment techniquesusing a silicon microbench. Precisely diced VCSELdevices are aligned to multimode fibers using

mechanical stops and the ribbon cable alignment pins.We have also investigated post-growth, processtunable VCSELs and multi-cavity filters.

In addition, we have explored the scalability of oursystem to an ASCI-scale machine, designed for100 TFLOP/s computation power. An advanced, fullyfunctional LambdaConnect system, with fiber paral-lelism of 36, and 19 wavelengths per fiber each, operatingat 5 Gb/s, functions as a fully non-blocking crossbarswitch (no internal contention) with a potential bisectionbandwidth of 10.2 Tb/s. Although this is impressive,ASCI performance requirements for balanced computingindicate 1 B of communication for every 10 FLOP, a bisec-tion bandwidth of 80 Tb/s for a 100-TeraFLOP/s system.

We studied the strictly non-blocking five-stage“Clos” switching architecture shown in the figure. Thesolution we investigated assumed that the 100-TFLOPmachine would be assembled from 500 SMP nodes(N = 500), each with 200-GFLOP performance. An opti-mal solution for a five-stage Clos network with N = 500has the key solution parameters n1 = 10, n2 =5, andr = 10. These parameters were used to calculate thedimensions of the building blocks of the network, todetermine the required number of elements per stage,and to estimate the full system cost.

Although we found the cost of an ASCI-scale solutionusing the Clos network prohibitive, the components andpackaging techniques we developed are useful forsmaller switching networks and for “fat-pipe” applica-tions, where all wavelengths are used concurrently forbyte-wide high bandwidth communication. Potentialcommercial partners are now reviewing the componentsand techniques studied throughout this research.

ETR • LDRD • FY00

rxr

Wavelength selectioncontroller

Number ofswitching elements

Number ofinterconnecting lines

(N/n1n2)(2n1–1)N/n1 N/n1

(N/n1)(2n1–1)(N/n1)(2n1–1)

(N/n1n2)(2n1–1)N(2n1–1)(2n2–1)

(2n2–1)xn2 (2n1–1)xn1n1x(2n1–1) n2x(2n2–1)

n1n2r

Element Dimension

N inputs N outputs

N(2n1–1)(2n2–1)

n1n2

N(2n1–1)(2n2–1)

n1n2

The Clos switching architecture.The interface to each stage of thenetwork (dashed vertical lines)uses an optical-to-electrical andelectrical-to-optical conversion,allowing wavelength “re-use” ofthe coarse wavelength space ofLambdaConnect, which is funda-mental to allow scaling.

15

A MEMS-Based Fuel Cell for Micropower GenerationJ. Morse, A. Jankowski

New power sources are required for all aspects of themilitary, weapons testing, and intelligence community,with many of these having specific performance criteriafor the direct application. A light-weight, long-lastingpower source provides new functionality to missions ofall kinds, promising long-term cost benefits to allgovernment agencies, enabling new levels of safetyand security for personnel in the field, along withgeneral national security.

The goal of this project is to create a working inte-grated MEMS-based (combining micromachining,

microfabrication, and thin film technologies) fuel cellthat delivers power in the 0.5 to 3.0 W range. This powersource will exhibit longer mission durations than exist-ing battery technologies, and be designed for specificimplementation in wireless remote sensing applications.

We will further incorporate a microfluidic catalyticfuel processor with the fuel cell to reform hydrocarbon-based fuels having high specific energy content,thereby providing an integrated power solution target-ing the 0.5 to 20 W range. Through the combination ofthin films materials, microfabrication, and siliconmicromachining at LLNL, a materials and fuel flexiblepower source is realized, as illustrated in the figure.

This represents a completely new approach to fuelcell technologies, and opens a new range of applicationsfor fuel cells providing power for autonomous sensornetworks being developed for national security require-ments. The MEMS-based fuel cell offers advantages interms of manufacturability, fuel flexibility, reducedtemperature of operation, and higher specific energy incomparison to other fuel cell and battery technologies.A further advantage is the direct scalability of thisapproach through stacking of individual fuel cellmodules, achieved through micromachined packagingmethods. This enables a wide range of power require-ments to be addressed with a common module design.

Results to date have demonstrated the lowesttemperature of operation anywhere of a solid oxidefuel cell, exhibiting very competitive power densitiesfor both solid oxide and proton conducting electrolytematerials. Having made these key demonstrations,present efforts are directed towards increasing thepower output density of the thin film fuel cells. We

have done this by extending the total surface area towhich fuel is manifolded through novel integrationprocesses that produce porous host/manifold struc-tures and electrodes, thereby enabling an atmosphericpressure, air breathing fuel cell device.

Efforts during FY00 have continued to optimize theprocess integration of the electrode/electrolyte materialand structure for both PEM and solid oxide fuel cells.Gas diffusion electrode support structures fabricated insilicon substrates using deep anisotropic etch tech-niques will enable a highly porous host structure forgas diffusion to the anode. At the same time they willretain the necessary mechanical strength to withstandhigh temperature swings and pressure gradients.

Subsequently, thin film anode-electrolyte-cathodelayers are deposited on the gas diffusion electrodesupports to form the actual fuel cell structure. At thisstage the cells are ready for test. High temperaturepackage and fixturing are presently being designedand developed.

The next phase of this work is to form completepackaged systems for the MEMS-based fuel cells.Subsequent design and testing will explore integratedheaters, and direct use of methanol fuel.


Electrode

Electrolyte

Resistive heater

Heater isolation

Electrode

Micromachinedmanifold system

Schematic illustration of MEMS-based fuel cell concept.

Advanced Imaging CatheterP. Krulevitch, D. Hilken, J.-U. Kluiwstra, R. Miles, D. Schumann, K. Seward

Catheter-based, minimally invasive surgery isperformed by making a small incision in a mainartery, inserting a long, hollow tube (catheter), andnavigating through the artery to the treatment area.Once positioned, the catheter is used to deliverdevices such as angioplasty balloons and stents foropening occluded arteries. Over 700,000 catheterprocedures are performed annually, and the advan-tages of this technique—reduced patient trauma andfast recovery—make it one of the fastest-growingsurgical procedures. In most cases, catheters are posi-tioned using radiography for visualization, andmanual manipulation for navigation and positioningof the device. We are addressing the difficulties associ-ated with current procedures.

The objective of this project is to enhance the physi-cian’s navigational abilities by producing a compact

catheter that offers imaging and active control to guideand position the distal end (the end inside the body).Clinicians have emphasized that such a device wouldbe a tremendous breakthrough. In addition, this tech-nology has the potential for noninvasive or minimallyinvasive imaging of explosives or weapons systems.

Over the past three years, we have applied LLNL’sexpertise in optical and ultrasonic imaging, microfabri-cation, and modeling toward the development of anadvanced catheter. The emphasis for FY00 was oncatheter articulation, mesoscale polymer molding andextrusion, and forward-looking ultrasound. In collabo-ration with physicians at the University of California,San Francisco and the University of California, Davis,we observed several catheter-based procedures anddiscussed how an articulating, imaging catheter couldimprove their patients’ outcomes. Prototype devicesincorporating shape-memory alloy (SMA), shape-memory polymer (SMP), and hydraulic actuation werefabricated and tested. The figure shows an SMPcatheter tip with an embedded SMA spring. Heatingthe composite device just above the SMP transforma-tion temperature causes it to expand and lengthen;

heating above the SMA transformation temperaturecauses it to contract, enabling advancement of the tipwhen combined in series and actuated peristaltically.By fabricating and testing a 1-in.-long, 0.16-in. insidediameter 0.08-in. outside diameter, hydraulically actu-ated, molded silicone catheter tip, we demonstratedrepeatable 90° bending. In addition, a fluid-filled,0.25-inch-diameter polymer bellows was shown toincrease in length 0.5 in. when heated by a laser.

During FY00, we also investigated using forward-looking ultrasound to aid in guiding a needle duringan intravascular liver-bypass procedure known astransjugular intrahepatic portosystemic shunting. Inthis procedure, a single-element piezoelectric trans-ducer is placed at the end of the articulating catheterand periodically pulsed. Harder tissues reflect moreacoustic energy compared to blood-filled veins, thusenabling navigation through a diseased liver. Todemonstrate the principle, we placed a single-element,circular transducer in a water bath 40 mm from avessel-mimicking tube. The first interface was the frontwall of the tube, and a large amount of energy wasreflected back to the transducer. A second, muchsmaller reflection was observed coming from the backwall of the tube. This demonstrates the ability to useultrasound to measure both the distance to the vesseland the size of the vessel. Numerical simulations opti-mized the size and shape of the transducer.

Several patents have been filed from this project,and two have been issued.

ETR • LDRD • FY00

Our advanced, composite catheter tip, showing a prototype shape-memory polymer tip with an embedded shape-memory alloy spring.

17

Speckle Reduction for LIDAR Using Optical Phase ConjugationM. Bowers, C. Kecy, L. Little, J. Cooke, J. Bentereau, R. Boyd, T. Birks

Remote detection of chemicals with LIDAR (LightDetection and Ranging) using DIAL (DifferentialAbsorption LIDAR) is now a standard detection tech-nique for both military and civilian activities. We havedeveloped a novel nonlinear optical phase conjugationsystem that can reduce the effects of speckle noise andatmospheric turbulence on DIAL remote detectionsystems. We have shown numerically and experimen-tally that it is possible to increase the signal-to-noise(S/N) ratio for LIDAR systems under certain condi-tions using optical phase conjugation. This increase inS/N can result in more accurate detection of chemicaleffluents while simultaneously reducing the timenecessary to acquire this information.

ALIDAR system is one in which a laser beam istransmitted out toward an area to be probed. The

return scatter from the aerosol itself or the topographicbackground is detected and measured. In the DIALtechnique two or more laser pulses of different frequen-cies are transmitted sequentially toward a target area.The differences in return signal power for each of thesefrequencies is measured. Since it is assumed that thereturned power would be identical for all frequenciesin the absence of absorption, this differencing tech-nique can determine the relative absorption at thetransmitted frequencies.

In practical, field operational systems, this simplisticview is complicated by noise sources, such as speckle,which results from the light scattering off a diffusesurface, resulting in a random, but deterministic, lightirradiance at each point in the detector aperture.

Optical phase conjugation is a nonlinear opticalmethod of removing the phase differences of thediffuse rays of light. Thus, light can be made to sumconstructively over the entire detector aperture, remov-ing the presence of a speckle pattern and the associatednoise. DIAL systems require optical pulses of severalhundred microjoules or more over many tens ofnanoseconds. Because of this, and the high-spatial-frequency content of the DIAL return signal beam,stimulated Brillouin scattering (SBS) was selected as thephase-conjugation method.

For SBS to compensate for the speckle, the opticalbeam must pass twice through the distorting regions(see figure).

During the course of this project we were able toshow that optical phase conjugation can, under certainconditions, improve the S/N of DIAL LIDAR signals.We were able to demonstrate this in the lab as well incomputer simulations. We were able to phase conjugate1530-nm light in optical fibers, for what we believe tobe the first time anyone has done this with ns-scalepulses. We also demonstrated the large numerical aper-ture fibers with appropriate tapers necessary for theoptical phase conjugation system to be compact and yetstill acquire the DIAL signal.

However, due to the problems with the sapphirefiber crystalline structure and the low numerical aper-ture of the silica fibers, we were not able to demon-strate this system at the 1-km distances that we hadanticipated. We do believe, however, that we havepaved the way for this technology to be successful inthe near future. We are planning to expand the scope ofthe work to find appropriate fibers that will facilitatethe transfer of this technology to DIAL LIDAR systemsin the field.


Many kilometers

Atmosphericdistortion

Transmitter/receivertelescope

Conjugatetelescope

Amplifier fiber

Diffusetarget

2 π ster scattered signaland return conjugate rays

Lasersource

Oscillator fiber

The two-round-trip optical phase conjugation system. The initialprobe beam is reflected into 2 π ster. A second telescope capturespart of this beam, attenuating it 80 to 100 dB. The beam is trans-formed into pulse format by a BEFWM process, timed appropriatelyfor SBS amplification. The next time the now pulsed conjugatesignal beam is transmitted back to the target, a much larger frac-tion of the light will return to the original transmitting telescopethan would be expected from linear light propagation. This is dueto the phase conjugate properties of the light. The light can now bedetected with little or no speckle noise.

Dynamic Focusing of Acoustic Energy for Nondestructive EvaluationJ. V. Candy

The inspection of parts, whether lenses for opticalsystems or components of sensitive weapons systems,must be accomplished as part of a regularly scheduledmaintenance and monitoring program, especiallyduring the assembly process. The impact of this projectwill be considerable for inspecting optics at theNational Ignition Facility, and component parts forstockpile stewardship, for noninvasive treatments inthe medical area, and for underground target location.

This project is concerned with the R&D required for atechnique to dynamically focus acoustic energy for

both detection and characterization of flaws for the

nondestructive evaluation (NDE) of parts under ultra-sonic testing. We have proposed a systematic approachincorporating detailed simulations, algorithm develop-ment, hardware, proof-of-principle NDE experimentsand prototype design of a viable flaw detection/local-ization/imaging system.

Dynamic focusing is based on the concept of time-reversal (T/R). T/R processing of noisy ultrasonicsensor array measurements is a technique for focusingenergy in various media. The T/R processor can effec-tively be used to detect flaws (or scatterers) by using itsprimary attribute—the ability to iteratively focus on thestrongest flaw. A T/R processor simply receives themultichannel time series radiated from the regionunder investigation, collects the array data, digitizes,time-reverses the sensor array signals and re-transmitsthem through the medium to focus.

Using T/R, we have developed algorithms to itera-tively decompose the received field into its constituentscatterers, even when they overlap temporally. Theoverall structure of the approach is shown in Fig. 1.Typical results are shown in Fig. 2 for a homogeneousaluminum part with flaws.

During FY00 we have developed temporal tech-niques that exploit the T/R focusing property andconstructed new processing algorithms.

In FY01, we plan to continue the theoretical investi-gation of the T/R operator and time domain decompo-sition algorithms, investigate imaging techniques usingthe estimated number and location of the flaws,complete the development of a prototype system, andperform proof-of-principle NDE experiments on mate-rials and parts of interest to LLNL.

Flow map

Localizer

Converge?

Yes

Yes

No

Yes

Imager

Refine grid

Next flaw?

No

Next flaw

FY99 TASKS

FY01 TASKS

FY00 TASKS

T/Rdecompose

T/Rfocus

Iterativefocus

Flaws

Figure 1. Flaw localization and mapping algorithm.

0.5

1

1.5

2

2.5

x 10-3

8 9 10 11 12 13 14 15

2

3

4

5

6

7

8

9

0 50 100 150 200 2505

6

7

8

9

10

11

12

13

14

15

(c) Wavefront (Data(b) – Estimate(r)):

(a) Ultrasonic source localization (Power)

X-position (mm) Iteration no.

(b) Weighting function estimation (Polytope method)

Mea

n-s

qua

red

res

idua

l err

or

Y-p

osi

tio

n (

mm

)

Local (X = 12.01, Y = 6.00)

(d) Flaw map: (X = 12.01 mm, Y = 6.0 mm)

3

3.5Global (X = 11.92, Y = 6.08)

Figure 2. Application of flaw localization and mapping algorithms using T/R processing. Shown are (a) global (ultrasonic source localization),and (b) local (weighting-function estimation) search results for a synthesized aluminum slab. The wavefront match (c) and the final flaw mapfor a single flaw (d) are also shown.


SelectedEngineeringPublications

Berryman, J., L. Barcka, and G. Papanicolaou(2000), “Time reversal acoustics for multiple targets,”J. Acoust. Soc. Amer. (UCRL-140465).

Buettner, H. M., W. Labiak, and A. Spiridon(2000), “Remote instrumentation and safeguardsmonitoring for the Star project,“ Nuclear PlantInstrumentation, Control and Human-MachineInterface Technologies Meeting, American NuclearSociety, Nov. 13–16.

Buettner, H. M., E. Didwall, F. Followill,D. McCallen, and R. Morey (2000), “GPR characteri-zation of underground facilities at NTS,” Proc. of theUnattended MASINT Sensor and Geophysical Conf.,Quantico, Virginia, Nov. 7–8.

Burke, G. J., and D. J. Steich (2000), “Numericalmodeling of shielding by a wire mesh box,” 16thAnnual Review of Progress in Applied ComputationalElectromagnetics, pp. 452–459, Monterey, California,March 20–24.

Candy, J., and D. Chambers (2000), “The role ofthe time reversal processor in acoustic signal process-ing,” J. Acoust. Soc. Amer. (UCRL-JC-141160).

Candy, J., D. Chambers, R. Huber, andG. Thomas (1999), “Matched-field imaging of lasergenerated ultrasound for nondestructive evaluation,”J. Acoust. Soc. Amer. (UCRL-133397).

Candy, J., D. Chambers, R. Huber, andG. Thomas (2000), “Ultrasonic matched-field imagingfor nondestructive evaluation,” Ocean Imaging Conf.(UCRL-136369).

Candy, J., (2000), “The role of the time reversalin signal processing,” C.A.S.I.S. Workshop at LLNL(UCRL-VG-141331).

Caporaso, G., and J. McCarrick (2000), “Ion-Hose instability in long pulse induction accelerators,”Proc. of the 20th Int. LINAC Conf., pp. 500–02,Monterey, California, Aug. 21–25.

Chambers, D., and A. Gautesen (2000),“Multiple eigenvalues of the time reversal operatorfor a single hard scatterer,” J. Acoust. Soc. Amer.(UCRL-139914).

Chambers, D., and A. Gautesen (2000), “Timereversal for a single spherical scatterer,” J. Acoust. Soc.Amer. (UCRL-JC-141165).

Chen, Y.-J., D. Ho, J. McCarrick, A. C. Paul,B. Poole, S. Sampayan, L. Wang, and J. Weir (2000),“Physics design of the ETA-II/Snowtron doublepulse target experiments,” Proc. of the 20th Int. LINACConf., pp. 482–4, Monterey, California, Aug. 21–25.

ETR • LDRD • FY00 • Selected Publications 23

Conrad, D. C. (2000), “Underground explosionsare music to their ears,” Science and Technology Review,pp. 4–11, July/Aug. (UCRL-52000-00-7/8).

Conrad, D. C., P. Egan, and the LLNL/BN SCETeam (2000), presented by Ray Heinle, “U1a vesselexperiments: OBOE and beyond,” JOWOG 32 PDT,Aldermaston, England, Nov 29–Dec. 1.

Cooperstein, G., W. DeHope, et al. (2000),“Progress in rod pinch electron beam diodes asintense x-ray radiographic sources,” 13th Int. Conf.on High-Power Particle Beams (BEAMS 2000),Nagaoka, Japan, Jun. 25–30.

Cunningham, C. T., and R. S. Roberts (2001),“An adaptive path planning algorithm for cooperat-ing unmanned air vehicles,” Proc. of the IEEE Int.Conf. on Robotics and Automation, Seoul, Korea, May(UCRL-JC-140415).

Dudley, D. G., H.-Y. Pao, and D. A. Hill (2000),“Introduction to special issue,” IEEE Transactions onAntenna and Propagations, 48, No. 9, Sept.

Eppler, W. G., D. W. Paglieroni, M. Petersen,and M. J. Louie (2000), “Fast normalized cross-correlation of complex gradients for autoregistra-tion of multi-source imagery,” ASPRS DC 2000Conf., May 22–27.

Falabella, S., Y.-J. Chen, T. Houck, J. McCarrick,S. Sampayan, and J. Weir (2000), “Effect of backscat-tered electrons on electron beam focus,” Proc. of the20th Int. LINAC Conf., pp. 458–60, Monterey,California, Aug. 21–25.

Ge, J., D. Ciarlo, P. Kuzmenko, B. Macintosh,C. Alcock, and K. Cook (2000), “Etched silicon grat-ings for NGST,” Next Generation Space Telescope Scienceand Technology, ASP Conf. Series, 207, p. 457.

Ge, J., D. Ciarlo, P. Kuzmenko, C. Alcock,B. Macintosh, R. Angel, N. Woolf, M. Lloyd-Hart,R. Q. Fugate, and J. Najita (2000), “Adaptive opticshigh resolution spectroscopy: present status andfuture direction,” Proc. Imaging the Universe in ThreeDimensions, ASP Conf., 195, p. 568.

Ge, J., J. P. Lloyd, D. Gavel, B. Macintosh, C. E.Max, D. Ciarlo, P. Kuzmenko, and J. R. Graham(2000), “High spectral and spatial resolution spec-troscopy of YSOs with a silicon prism and adaptiveoptics,” BAAS, 197, p. 5201.

Guethlein, G., T. Houck, J. McCarrick, andS. Sampayan (2000), “Faraday cup measurements ofions backstreaming into an electron beam impingingon a plasma plume,” Proc. of the 20th Int. LINAC Conf.,pp.467–9, Monterey, California, Aug. 21–25.

Handler, F. A., and L. C. Ng (2000), “Signaldetection and feature extraction for target/decoydiscrimination from IFT-1A flight experimental data:a technical response to Dr. Ted Postal’s allegations(U),” published as a POET document.

Holzrichter, J. F., and L. C. Ng (2000), “Speecharticulator and user gesture measurements usingmicropower, interferometric EM-sensors,” IEEE-Instrumentation and Measurement Technology Conf.(IMTC), Budapest, Hungary, May 21–23 (UCRL-JC-140496).

Huber, R., J. Candy, D. Chambers, andG. Thomas (1999), “Processing of laser-based ultra-sound for matched-field imaging,” Quantitative NDEConf. (UCRL-134055).

Le Sage, G. P., S. G. Anderson, T. E. Cowan,J. K. Crane, T. Ditmire, and J. B. Rosenzweig (2000),“RF photoinjector development for a short-pulse,hard x-ray Thomson scattering source,” AIP Conf.Proc. for the Advanced Accelerator Conf., Santa Fe,New Mexico, (UCRL-JC-140148).

McCarrick, J. (2000), “The effect of asymmetricplasma plumes on the transport of high-current elec-tron beams,” Proc. of the 20th Int. LINAC Conf,pp. 461–3, Monterey, California, Aug. 21–25.

Merrill, R. D., et al. (2001), “Material transfersystem in support of the plutonium immobilizationprogram,” Proc. of the American Nuclear Society NinthTopical Int. Meeting on Robotics and Remote Systems,Seattle, Washington, March 4–8.

Merrill, R. D., et al., (2001), “Remote equipmentdevelopment for the plutonium ceramification testfacility at LLNL,” Proc. of the American Nuclear SocietyNinth Topical Int. Meeting on Robotics and RemoteSystems, Seattle, Washington, March 4–8.

Ng, L. C., G. C. Burnett, J. F. Holzrichter, andT. J. Gable (2000), “Denoising of human speech usingcombined acoustic and EM sensor signal processing,”IEEE ICASSP-2000, Istanbul, Turkey, Jun. 6 (UCRL-JC-136631).

Paglieroni, D. W., and R. S. Roberts (2000),“Optimal segmentation strategy for compact repre-sentation of hyperspectral image cubes,” Proc. ASPRSDC 2000 Conf., May 22–27.

Pao, H.-Y. (2000), “In memoriam of James R.Wait,” Guest Editor, IEEE Transactions on Antenna andPropagations, Special Issue, 48, No. 9, Sept.

Selected Publications • FY00 • ETR • LDRD 24

Pao, H.-Y., and J. R. Wait (2000),“Electromagnetic induction and surface impedance ina half-space from an overhead moving currentsystem, IEEE Transactions on Antenna and Propagations,48, No. 9, Sept.

Poole, B. R., Y. J. Chen, A. C. Paul, and L.-F.Wang (2000), “Particle simulation of DARHT-IIdownstream transport,” Proc. of the 20th Int. LINACConf, Monterey, California, August 21–25.

Puso, M. A., and T. A. Laursen (2000),“A method for 3D contact surface smoothing,” Int. J.for Numerical Methods.

Puso, M. A. (2000), “An energy and momentumconserving algorithm for rigid-flexible body dynam-ics,” Int. J. for Numerical Methods.

Richardson, R., S. Sampayan, and J. Wier (2000),“Elliptical x-ray spot measurement,” Proc. of the 20thInt. LINAC Conf., Monterey, California, August 21–25.

Roberts, R. S. (2000), “Characterization ofhyperspectral data using a genetic algorithm, IEEEThirty-Fourth Asilomar Conf. on Signal, Systems andComputers, Pacific Grove, California, Oct. 29–Nov. 1.

Sampayan, S., G. Caporaso, Y.-J. Chen,S. Falabella, D. Ho, T. Houck, E. Lauer, J. McCarrick,R. Richardson, D. Sanders, and J. Weir (2000), “Beam-target interaction experiments for Bremsstrahlungconvertor applications,” Proc. of the 20th Int. LINACConf., pp. 464–6, Monterey, California, Aug. 21–25.

Tomascik-Cheeseman, L. M., R. Raja, L. M.Kegelmeyer, S. L. Mabery, B. J. Marsh, F. Marchetti,J. Nath, and A. J.Wyrobek (2000), “Parallel expressionanalyses of DNA repair genes in adult mouse tissuesusing cDNA microarrays,” Proc. of the EnvironmentalMutagen Society, New Orleans, Louisiana, inEnvironmental and Molecular Mutagenesis, 35, No. 31,p. 61, April.

Wang, L., G. J. Caporaso, and E. G. Cook,“Modeling of an inductive adder kicker pulser forDARHT-2,” Proc. of the 20th Int. LINAC Conf.,Monterey, California, August 21–25.

Wang, L., S. M. Lund, and B. R. Poole (2000),“Dipole septum magnet in the fast kicker system formulti-axis advanced radiography,” Proc. of the 20thInt. LINAC Conf., Monterey, California, August 21–25.

AuthorIndex

Anderson, G. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Anderson, S. G. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Aufderheide, III, M. B. . . . . . . . . . . . . . . . . . . . . . . . . . . 9Behymer, E. M.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Bentereau, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Birks, T. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Bond, S. W. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Bowers, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Boyd, R.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Candy, J. V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Champagne, N. J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Chinn, D.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10Clague, D. S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Coldren, C. W. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Cooke, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Cowan, T. E. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 4Crane, J. K. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 4Cross, R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Cunningham, C. T. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Ditmire, T. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 4Goodman, D. M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Hall, J. M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Harris, J. S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Hays, G. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Hilken, D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7, 17Hills, R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Jackson, J. A.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Jankowski, A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Kallman, J. S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Kecy, C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Kluiwstra, J.-U. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Krulevitch, P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Kuo, P.-K. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10Larson, M. C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

ETR • LDRD • FY00 • Author Index 27

Leach, Jr., R. R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Le Sage, G. P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 4Little, L. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Logan, C. M.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Martz, Jr., H. E.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9McCallen, D. B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Meyer, G. A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Miles, R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Moffitt, K. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Morse, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Ng, L.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Patel, R. R.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Petersen, H. E. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Pocha, M. D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Puso, M. A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Roberts, R. S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Rosenzweig, J. B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Schach von Wittenau, A. . . . . . . . . . . . . . . . . . . . . . . . . 9Schumann, D.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Seward, K. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Sharpe, R. M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Shuttlesworth, R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Slone, D. M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Springer, P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Spruytte, S. G. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Steich, D. J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Stephan, P. L. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Stolz, C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10Swift, D.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Tsai, V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Wharton, K. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Wheeler, E. K.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7White, D. A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Zywicz, E. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Manuscript Date September 2001Distribution Category UC-42

This report has been reproduced directly from thebest available copy.

Available to DOE and DOE contractors from theOffice of Scientific and Technical InformaitonP.O. Box 62, Oak Ridge, TN 37831Prices available from (423) 576-8401http://www.osti.gov/bridge/

Available to the public from theNational Technical Information ServiceU.S. Department of Commerce5285 Port Royal Rd.Springfield, VA 22161http://www.ntis.gov/

OR

Lawrence Livermore National LaboratoryTechnical Information Department’s Digital Libraryhttp://www.llnl.gov/tid/Library.html

This document was prepared as an account of work sponsored by an agency of the

United States Government. Neither the United States Government nor the University

of California nor any of their employees, makes any warranty, express or implied, or

assumes any legal liability or responsibility for the accuracy, completeness, or useful-

ness of any information, apparatus, product, or process disclosed, or represents that its

use would not infringe privately owned rights. Reference herein to any specific com-

mercial products, process, or service by trade name, trademark, manufacturer, or other-

wise, does not necessarily constitute or imply its endorsement, recommendation, or

favoring by the United States Government or the University of California. The views

and opinions of authors expressed herein do not necessarily state or reflect those of the

United States Government or the University of California, and shall not be used for

advertising or product endorsement purposes.

Work performed under the auspices of the U.S. Department of Energy by the University

of California, Lawrence Livermore National Laboratory under Contract W–7405–Eng–48.

ENG-01-0004a-AD

Engineering DirectorateLawrence Livermore National LaboratoryUniversity of CaliforniaP.O. Box 808, L-124Livermore, California 94551

http://www-eng.llnl.gov/

http://www-eng.llnl.gov/

ucrl-53868-00 engineering technology reports

Documents