i n n o v a t i v e r e s e a r c h

f a l l 2 0 0 8

T h e U n i v e r s i T y o f n e w M e x i c o s c h o o l o f e n g i n e e r i n g

accelerating impact

Y E A R S O F E X C E L L E N C E

1 9 06 – 200 6

0207

1217

From the DeansContents 2 nano-engineering naturally When it comes to building biomedical sensors, sorters and

infection-resistant materials, Gabriel López says nature is

the mother of invention.

7 The laser Trailblazer Luke Lester’s bright ideas set records in quantum dot lasers

and other optoelectronic devices.

12 robots to the rescue Birds and bees do it. Now robots can, too. Bert Tanner explores

how simple rules between individual robots can lead to dynamically

rich group behaviors, such as flocking and swarming, and how one

day this kind of collective action could save lives.

17 Maestro of Microwaves For two decades, Edl Schamiloglu has conducted high power

research on high power microwave sources.

on the coverAt the frontier of medicine, nature’s inspiring nano-machines meet bioengineer Gabriel López’s ingenuity. See page 2.

UnM engineering innovative researchFall 2008Volume III

editorTamara Williams

writerStefi Weisburd

DesignCisneros Design

innovative research is published annually by the University of New Mexico School of Engineering. Subscriptions are free; requests should be submitted to the address below. Material may not be reproduced without permission.

The University of New Mexico has been fully accredited by the North Central Association of Colleges and Secondary Schools since 1922. UNM is an Affirmative Action/Equal Opportunity institution. In accordance with the Americans with Disabilities Act, this material is available in alternate formats upon request.

School of Engineering Deans Office, Centennial Engineering Center, Suite 3071 MSC01-11401 University of New Mexico, Albuquerque, NM 87131-0001505-277-5521

www.soe.unm.edu

a c c e l e r a t i n g i m p a c t 1

Dear Colleagues,

One of the exciting aspects of leading an engineering school is the discovery of connections between faculty whose work, at first glance, would not seem to have anything in common. This issue of the School of Engineering Research Magazine highlights researchers who have made in-depth contributions in the seemingly unrelated fields of robotics, biosensors, semiconductor lasers diodes, and high power microwaves. However, given both the increasingly interdisciplinary nature of research and the accelerating impact engineers have on society, a closer look uncovers plenty of connections.

Just a few weeks ago, the School of Engineering dedicated the new 147,500 square foot, $42 million Centennial Engineering Center (CEC). Naturally there was a ribbon cutting ceremony, but as befits a modern engineering school, a robot did the actual cutting. The robot was designed and programmed by the students of Electrical and Computer Engineering Professor Rafael Fierro, a new collaborator of Professor Bert Tanner whose work on cooperative robot swarms that mimic flocks of birds is presented in this issue. Centennial Engineering Center is the home for the new Center for Biomedical Engineering, whose Director Professor Gabriel López is also featured here. As described in the article, Professor López’s work in biosensors and biodiagnostics has also been greatly inspired by nature and has resulted in several UNM patents now licensed by companies around the world.

Professor Luke Lester is also an inventor with several licensed patents based on his quantum dot laser diode technology described in this issue. He was a founder of Zia Laser, Inc., a startup company that raised nearly $23 million in venture capital. Professor Lester is exploring possible quantum dot laser applications ranging from medical imaging to homeland security. Indeed, all of the featured faculty have worked to enhance the safety and security of the United States in research sponsored by numerous federal agencies. With his long history of support from the Air Force, Professor Edl Schamiloglu has been one of the nation’s key researchers developing approaches to high power microwaves for military missions including disabling improvised explosive devices in Iraq. Finally, to come full circle, Professor Schamiloglu and Professor Tanner were founders of the Institute for Infrastructure Surety at the UNM School of Engineering where one of the research projects investigated the susceptibility of networks, including robotic networks, to damage from electromagnetic pulses originating from high power microwave sources.

Research in today’s world cannot afford to be solitary or disengaged from other research and from society. Connections based on both scientific collaborations and personal relationships are the cornerstone of research in the School of Engineering at UNM.

Our very best wishes,

Joseph L. Cecchi Kevin J. Malloy Dean of Engineering Associate Dean of Research

The school of engineering at the University of new Mexico

Departments Chemical and Nuclear Engineering

Civil Engineering

Computer Science

Electrical and Computer Engineering

Mechanical Engineering

research centers and institutes Advanced Materials Laboratory

Center for Biomedical Engineering

Center for Emerging Energy Technologies

Center for High Performance Computing

Center for High Technology Materials

Center for Micro-Engineered Materials

Center for Nuclear Non-Proliferation

Science and Technology

Institute for Space and Nuclear Power Studies

Manufacturing Training and Technology Center

2 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 3

fa l l 2 0 0 8g a b r i e l l ó p e z

sulfur atoms (see Figure 1). At the upper, unbound ends of the trees, researchers can attach a multitude of functional molecules such as sugars, nucleic acids or peptides. Making SAMs is a relatively simple lab procedure, much easier than constructing molecules on surfaces with techniques such as chemical vapor deposition.

Two other post-docs started at the same time as López, and Whitesides gave them each an assignment. He wanted them to explore different ways of patterning SAMs in two dimensions to see if they could precisely place different functional groups in different areas on the same substrate. Whitesides told one post-doc to develop a pen to “write” different functional configurations; another to scratch away areas in a SAM that could be replaced with other molecules; and López was asked to make a mask that would create various SAM patterns. Whitesides suggested he use transparent tape.

The three post-docs succeeded in their tasks and together developed several simple but pioneering techniques (utilizing microstamping, microwriting and micro-machining, and UV microlithography) that can control dimensions down to the micron-scale. In one 1993 paper, for example, they controlled the concentration and spatial dis-tribution of proteins adsorbed on a synthetic surface by terminating some SAM “trees” with short chains of polyethylene glycols to resist proteins and others with nonpolar and ionic groups to adsorb them. Understanding protein adsorption is key to a host of biotechnology ventures including pharma-ceutical production, protein purification, the design of biosensors and prosthetics, and the

creation of supports for tissue cultures. The groups’ f lexible techniques can also control wetability, interfacial energy, and surface charge, among other characteristics.

For López, one of the more interesting consequences was the group’s ability to di-rect not only where a living cell would stick to a substrate, but what its shape would be. If the shape of a cell can be controlled, so can its function. Inside the body, cells are constrained by a kind of 3-dimensional scaffolding that orchestrates their growth and behavior. Cells cultured in a Petri dish, on the other hand, spread out and grow uncontrollably so that cell shape and behavior can vary appreciably. Also, in a Petri dish, cells can be oriented in different directions, and it can be difficult to keep track of individual cells. So for research-ers hoping to analyze functional changes within cells in their search for new medica-tions, genetic engineering and toxicology studies, conventional cell cultures present significant limitations.

As described in a 1994 Science paper that has been cited over 600 times, the group made a rubbery polydimethylsiloxane (PDMS) stamp to print on gold a grid of rectangular islands to which cells adhere. The cells, in this case, rat liver cells, stuck to island SAMs coated with laminin (a protein that forms sheets around internal organs) and assumed the same regular rectangular shapes of the underlying islands. (See Figure 2).

By changing the size of the rectangles, the researchers could control the cells’ growth and life cycle stage as well as regulate their protein expression. This was the first time a link between cell size and life cycle could be

unequivocally shown in culture without the influence of other variables. The technique represented a huge step in versatility and robustness, allowing researchers to orient, index, and maintain every cell in a large culture for relatively long periods of time. It also permitted the patterning of a wide range of complex, delicate, or reactive organic functional groups or ligands (substances that bind to cell surface receptors) necessary to elucidate the molecular interplay at biologi-cal surfaces.

Since arriving at UNM, López has continued to use SAMs as model systems for unraveling and manipulating complex biomolecular interactions. For example, in one study of the protein streptavidin and biotin, a vitamin that binds very tightly to it, López’s group described how differently molecular recognition progresses when the biotin is attached to a SAM as opposed to being in a solution.

It also found that interactions between the adsorbed streptavidin molecules affected the system kinetics and how seemingly small changes in the chemicals used to make SAMs can have an enormous impact on the system. Since an ultimate goal of biomedical engineering is to make sensors, diagnostic assays, and drug delivery vehicles and other biotechnologies, rooting out these nuances is critical to making accurate and sensitive devices.

In another well-cited study, López’s group used SAMs to develop a convenient new method for detecting protein-ligand interactions with a new protein-sensitive, f luorescent dye that can be excited by cur-rently available laser diodes. One of the key achievements of this study was to make SAMs with f luorescent and ligand probes that will bind to one protein specifically and are resistant to others. Unlike other fluores-cence-based methods, López’s approach does not require a second labeling agent, can be expanded to a wide range of proteins and ligands, and is reversible, meaning that a real-time monitoring device is possible.

Making Devices from Protein Nano-MachinesMaking use of natural biochemical reactions is an essential part of developing biomedi-cal devices. But nature has more to offer, including molecular “machinery.” Most of

SCIeNTISTS Were excited and divided. It was the 1990s, and nano was hot. New microscopy techniques were revealing un-precedented views of the nanoscale. With the discovery of fullerenes and the fervor over carbon nanotubes, people were predicting the fabrication of materials and devices with undreamed of properties. K. eric Drexler published his blueprint for molecular manufacturing, with factories of nanoscale “assemblers” that could build mechanical machines atom by atom. While he envisioned cheap materials and nanorobots that would open clogged arteries, arguments about the physical plausibility of his vision were begin-ning to stir. Could the surface forces that dominate the nanoscale make nanotools too “sticky” to easily manipulate atoms?

But to Gabriel López, an assistant professor at the time, all the hoopla and arguments seemed a little beside the point. Nanomachines of the most elegant design already existed and were in fact ubiquitous.

“Biology is replete with complex, functional nanoscale structures formed by directed synthesis and subsequent self-assembly of the component molecules,” López wrote in the Encyclopedia of Chemical Technology.

Looking to the fruits of millions of years of evolution provides a cornucopia of ideas for navigating the nanoscale. Scientists can ei-ther synthesize assemblies that mimic nature (biomimetics) or steal nature’s “machines” outright (what López calls biokleptics). López practices both, making ultra-thin organic films that lead to prototype devices such as innovative molecular sieves, sensitive biosensors, microbe-expelling materials, and microfluidic assays that analyze miniscule amounts of blood. Underlying this technol-ogy is a disciplined study of the interfaces between biological entities and their interac-tion with engineering materials.

In the process he has mentored 29 Ph.D. and M.S. students, 27 post-docs and 55 un-dergraduate students. Author of 160 papers, book chapters and proceedings, he has given 130 presentations (64 of which were invited) since coming to UNM’s Chemical and Nuclear engineering and Chemistry departments in 1993. recipient of numerous awards, López has 13 patents and a high h number of 31. He is the founding director of the Center for Biomedical engineering and the director of the W.M. Keck Nanofluidics Laboratory. López has been awarded more than $20 million in research funding, including a $2.5 million NSF PreM grant with Harvard for leadership in biomaterials.

“Gabriel’s work that I’m familiar with is unique in that he uses his detailed and deep knowledge of what is rather sophisticated materials science to address important is-sues in public heath and other life sciences areas,” notes David A. Weitz, Mallinckrodt

Professor of Physics and of Applied Physics at Harvard University. “This is clearly going to be an area of very active research and has already become a topic of considerable effort. Gabriel is clearly a leader in the field, both at UNM, and more generally in the broader national and international communities.”

The idea of borrowing from biology was planted early in López’s mind. Growing up on a small ranch and farm in northern New Mexico, López learned how to work with nature—how to use a real grasshopper or a fabricated one to catch fish, for example, or how to coax a cow to feed another dying cow’s calf. This knowledge translated to more abstract, less visible notions as well. López remembers his older brother illustrating chemical bonds by making loops of sharp grass while the two were watching sheep. It’s not surprising then that as an adult scientist hunting for solutions to bioengineering problems, nature was the first place López would look for inspiration.

Foundations of Nanotechnology: Self-Assembled MonolayersWhen López arrived at Harvard as an NIH post-doc in 1991, his advisor, George Whitesides, and his group were investigating self-assembled monolayers (SAMs). SAMs promised to be a versatile and valuable tool for exploring a wide variety of chemistry, but what intrigued López most was their potential for studying and controlling interactions between proteins, cells, and synthetic surfaces. SAMs are ideal model systems that can simulate aspects of biologi-cal membranes and can serve as substrates for cell cultures. A SAM looks a bit like an ordered kelp forest. each “tree” consists of a long alkyl molecule with a thiol head group anchored to a gold film at specific sites via

nano-Bioengineering naturallyWhen it comes to building biomedical sensors, sorters and infection-resistant materials, Gabriel López says nature is the mother of invention.

Figure 1. Self-assembled monolayer (SAM).

X denotes a functional group.

~30˚

X X X X X X X X

SS S S

S S S S

Figure 2. Cells adhere to rectangular

islands on a patterned SAM (a), whereas

they spread out uncontrollably on an

unpatterned surface (b).

“This is clearly going to be an

area of very active research and

has already become a topic of

considerable effort. Gabriel is

clearly a leader in the field, both

at UNM, and more generally in the

broader national and international

communities.”

David A. Weitz, Mallinckrodt Professor

of Physics and of Applied Physics at

Harvard University

4 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 5

fa l l 2 0 0 8g a b r i e l l ó p e z

López’s team successfully grafted PNIPAAM inside 5–10 nanometer wide pores in silica. This polymer is of interest medically because its transition temperature is right between body temperature and room temperature, an especially convenient range for medical ap-plications. Again, the researchers found that the polymer occludes the pores below the transition temperature and then collapses, opening the pores at higher temperature and letting molecules through. They’ve also shown their switchable filter withstands repeated temperature cycling.

Lipid Bilayer Membranes: Nature’s Nano-BaggiesAnother of nature’s nifty inventions are phospholipids, the self-aligned, ampiphilic molecules that comprise the semi-permeable, bilayer membrane of the cell. López calls phospholipid membranes nature’s baggies. Their formation on pre-biotic earth was an essential step to the creation of cells and the beginning of life. They keep internal struc-tures and chemical reactions in, and they regulate the movement of other materials in and out. But they’re not rigid structures. The forces holding the bilayer together are weak, noncovalent interactions, so lipids and the proteins embedded within them diffuse f luidly along the membrane plane in a manner similar to the f luidity of soap bubbles. In thinking about incorporating lipids and membrane proteins into micro- or macro-scale devices, researchers have sought to support lipids on solid glass or other synthetic beads because lipid vesicles by themselves are too fragile to survive in a functional device.

López and his collaborators were the first to comprehensively study the stability of lipids and transmembrane proteins on porous silica microspheres. (To see how these spheres are made see the 2006 SOe research magazine.) The group demonstrated that lipids on silica spheres stay on the outside for the most part and, as shown in Figure 5, form a fluid lipid coating similar to lipids on glass beads. In various studies, the lipid bilayers remained stable and fluid for months, while unsupported lipid vesicles substantially leaked by the end of a week. The supported lipids easily incorporate ligands, f luorescent dyes and transmembrane proteins in con-

centrations controlled by their amounts in the precursor solutions.

The pores in the silica beads appear to minimize some of the unwanted interac-tions between membrane proteins and synthetic substrates experienced by other kinds of beads. More importantly, the pores present the opportunity for encapsulating and subsequently releasing ions, drugs, dyes or biological molecules, greatly expanding the spheres’ possible uses. experiments show that dyes encapsulated in silica spheres disperse uniformly throughout the pores and don’t leak out through the lipid bilayers. Moreover, when the bilayers take up a protein called bacteriorhodopsin, which is a light-activated proton pump, an encapsulated pH-sensitive dye changes its f luorescent properties allowing the researchers to quantitatively monitor the protein’s progress.

López envisions a portfolio of sphere-based sensors for screening assays and diagnostics. Some could be triggered by substances that disrupt the integrity of lipid membranes. In a proof-of-concept paper to appear in Biointerphases, López’s group veri-fied that the action of a non-ionic detergent, a protein produced by bacteria that punches holes in cell membranes and an anti-micro-bial peptide (the active ingredient in bee venom) can each be measured by detecting the release of dyes from lipid membrane– encapsulated microspheres.

To explore how these biochemical sensors might be engineered into a micro-f luidic device, the researchers packed a 2 millimeter-long microchannel with hundreds of micron-sized, lipid-coated spheres. The lipid layers survived the packing process and when exposed to mem-brane disruptors, observable amounts of dye fluoresced.

The researchers also investigated a two-step, serial sensor column. The first active segment contains spheres containing biotin as a model for other non-f luorescent, but bio-active cofactors, catalysts and ligands. For example, when toxins or detergent flow through this section, biotin can be released. The biotin then f lows downstream to a second section packed with spheres coated with a dye-biotin-streptavidin complex. When the free biotin interacts with these spheres, the dye fluoresces.

Soft Petrification: Harnessing Nature’s NanotechnologyBeyond lipid-coated microspheres that incorporate rudimentary cellular func-tions that make them good candidates as sensors, López has recently discovered a way to stabilize and support lipids in lipo-somes (lipid vesicles) and f lat layers. Other researchers have integrated liposomes with transmembrane proteins in silica by sol gel processes but there has been no way to get inside these vesicles and exploit their membrane properties to make practical devices that operate outside the controlled laboratory environment. Instead, López and his colleagues have invented a process they call “soft petrification,” which builds a cross-linked silica matrix around the lipid membranes, creating a rigid, robust mem-brane made of alternating layers of silica and lipids aligned parallel to the thin film sur-face, as shown in Figure 6. The process takes place in solution, with the silica molecules replacing much of the water as the structure of the lipids is maintained. Unlike previous sol-gel processes in which lipids and silica self-assemble at the same time, soft petrifac-tion polymerizes the silica around the lipid assemblies after they have formed.

X-ray diffraction reveals a lamellar lipid structure over all of the sample. remarkably, photobleaching indicates that the lipid bilayer is still f luid within the solid structure. The spacing between layers can be controlled by varying the unsaturated lipid concentration.

nature’s molecular machines are proteins, the long chains of amino acids held together by peptide bonds that participate in every cellular process from maintaining cell shape to catalyzing reactions. In fact, the cell can be seen as a factory filled with interlinked assembly lines of nanoscale protein pumps, motors, actuators, and transducers. Learning how to incorporate these mechanisms into stable, durable devices is a central theme in López’s recent work.

Stimuli responsive polymers (SrPs) are a class of particularly useful bio-tools. Also known as “smart polymers,” they change their conformations when the temperature, pH, salt content, electrical potential, or amount of light in their environment chang-es, even if the change is relatively subtle. For example, elastin, one of the extracellular proteins in skin and blood vessels, contracts when the temperature is raised above a certain level. At lower temperatures elastin is soluble in water (hydrophilic), but above the transition temperature its contraction makes it precipitate out and become more hydrophobic. This is an example of an inverse phase transition; in contrast, most materi-als get more water soluble as temperature is increased.

López’s group has exploited this be-havior, attaching elastin-like polypeptides (eLPs) to surfaces and using the transition temperature to make the surface more hydrophobic or hydrophilic, in effect, changing the wetability of the surface in real time. Over the last decade, López’s group has taken advantage of these properties in its bio-fouling studies. Because most com-mercially available anti-biofilm coatings are toxic, the researchers have been looking for environmentally responsible ways to encourage aquatic organisms to vacate boats and marine structures as well as surfaces at water treatment plants and those on pros-thetics implants. Using a synthetic analog of eLP called poly(N-isopropyl acrylamide), or PNIPAAM, as a model compound, the researchers introduced a new class of foul-ing release agents. In one study they found that both organisms partial to hydrophobic surfaces and those favoring hydrophilic ma-terials sloughed off the PNIPAAM–grafted surface when it was either heated or cooled through its transition temperature.

Besides changing surface wetability, the group has made use of eLP contraction and expansion to fabricate molecular valves or gates. The eLP is mixed with silica to form a hybrid organic/inorganic gel, a matrix of cross-linked silica forming a network of nanosized pores interspersed with eLPs. The researchers demonstrated that even though the gel is tough, enough water remains in it for the eLPs to be soluble and extended at lower temperatures, occluding the pores. But at the transition temperature, the eLPs contract, opening them up. (See Figure 3). The transition temperature depends on a polymer’s molecular weight and which amino acids are used to make it, factors that can be precisely specified through genetic engineering.

The temperature-controlled, aperture-like behavior of the eLPs suggests their use as a differential sieve for separating different sized molecules. Figure 4 illustrates a per-meability test of two hybrid membranes of different molecular weight, and hence tran-sition temperatures. At lower temperatures, the hybrid eLP/silica film is impermeable, while higher temperatures open the porous network. Combining eLP-60 and eLP-30 networks into one membrane could selec-tively filter larger and smaller molecules by changing the temperature.

The group received several patents for this idea. Unlike existing molecular weight cut-off filters, this new class of valves are switchable and can be tuned from the nano to micro-scales. The researchers think their inexpensive sieves (and other eLP-actuated devices) will be particularly attractive for use in microfluidic circuits.

One advantage of working with synthetic, biomimetic materials as opposed to natural ones is that they can be grown or polymer-ized inside an existing porous structure.

25 30 35 40 45

0.0

0.2

0.4

0.6

0.8

1.0

Silica-ELP-60 Silica-ELP-13membrane membrane

0 : No Permeation1 : Permeation

O

Pe

rme

atio

n

O

Figure 4. Hybrid membranes containing

elastin-like proteins (ELPs) of lower

molecular weight become permeable

at higher transition temperatures than

hybrids containing heavier ELPs.

(b)

T = 0 T = 5 min

Figure 5. Lipids supported on water-filled porous

beads (a) remain fluid. After a laser bleaches out

the dye in one part of the membrane, fluorescence

returns (b), indicating that other dye molecules

have diffused into that area.

T = 0 T = 5 min

(a)

Figure 6. Transmission elec tron micrograph

of lipid layers stabilized by soft petrification.

∆T

Figure 3. At low temperatures, elastin-like

proteins expand and close the pores. The

membrane becomes permeable at higher

temperatures as the proteins collapse.

6 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 7

fa l l 2 0 0 8

López’s group has also petrified lipo-somes containing bacteriorhodopsin protein and the antibiotic peptide gramicidin to demonstrate that not only structure, but function is also preserved (See Figure 7). The researchers hope that these thin films will serve as highly selective and perhaps

active transport membranes in devices that approach the fidelity of biological mem-brane function.

Currently López and coworkers are working to stabilize chlorosomes, the highly-efficient, light-harvesting “antennas” found in green sulfur bacteria living in low-light hotsprings and near sea vents. Chlorosomes differ from other antenna complexes because of their large size and their lack of a protein matrix supporting the photosynthetic pig-ments. If a chlorosome is put in water or a solvent, it falls apart as evidenced by the fad-ing of its characteristic green color. Using the soft petrifaction technique, López’ student, Gautam Gupta, succeeded in encapsulating chlorosomes in a hard material that stayed stable for at least 30 days. His collabora-tors at Sandia National Laboratories and Washington University are interested in this material for building photovoltaic or photoconversion devices.

Science and the HumanitiesContrary to romantic visions of a scientist shouting “eureka” in the lab, the working life of an academic engineer is filled with meet-ings, budgets, proposal writing, teaching, and research group management. Moreover, in both the administrative and basic research milieu, it’s easy to lose sight of the signifi-cance of one’s work, of who one is ultimately working for.

López is working for the diabetic patient needing a smart polymer implant that moni-tors glucose and metes out insulin just like a healthy pancreas. He’s working to help a pharmaceutical company sorting through millions of compounds for targeted cancer drugs. For the diagnostician trying to quickly narrow down the causes of an epidemic. For the aid worker needing a simple blood test that is stable without refrigeration. For solu-tions to these problems and more, we have Mother Nature to thank. And, of course, Gabriel López’s inspired biokleptomania.

g a b r i e l l ó p e z

First. Fastest. Best.DUrING HIS 16 year career in semicon-ductors and photonics, Luke Lester has consistently pushed technology to set new records:

• In 1988 he fabricated the world’s fastest transistor, a record that stood for over a decade as noted by the Guinness Book of World Records. This device is found today in many cell phone receivers.

• Two years later he made the first quantum well (QW) laser diode whose operation speed was greater than any previous semiconduc-tor laser. Today’s industry standard is based on his design. In 1993, Lester created the fastest long-wavelength photodetector, with a 1.3 picosecond response time that has yet to be beat.

• In 2000, Lester and his colleagues at UNM’s Center for High Technology Materials (CHTM) produced the first quantum dot (QD) laser whose properties surpassed those of QW devices.

• A year later he spearheaded the first commercial foray into QD laser diodes, co-founding Zia Laser, Inc., which was recently acquired by Germany’s NL Nanosemiconductor and then merged with Innolume.

In spite of these pioneering efforts, however, QD lasers have yet to become commercially viable. So this is also a story of why even the first, fastest, and best in the lab doesn’t always make that quantum jump into marketplace success.

Laser diodes are ever-present in our world. Watch a DVD or scan a barcode, and there is a laser diode to thank. Lester, a pro-fessor of electrical and computer engineering, has focused on lasers whose qualities make them especially attractive for telecommu-nications—lasers whose brightness can be modulated to transmit data across vast fiber optic networks. In particular, Lester was searching for a “tunable” laser that offers a wide range of distinct wavelengths. This kind of device would enable simultaneous transmission of many different data streams in the same optical fiber.

In 1998 Lester was handed a new gallium arsenide (GaAs) wafer on which a layer con-taining millions of nanometer (nm)-sized dots of Indium Arsenide (InAs) had been grown. None of Lester’s students seemed interested in making it into lasers and testing them, so it languished for weeks. But when Lester finally got a look at the spectrum it produced, he knew in an instant this was something big. He had been searching for it for the past five years.

ABCs of LASERSThe active region of the earliest lasers was a gas. Gas atoms have discrete energy states, so when an electron is first excited and then drops down to a lower energy state, a photon of a very specific wavelength is emitted. Gas lasers can produce high quality, stable optical beams of essentially one wavelength. Unfortunately they were big and expensive, couldn’t be modulated quickly, and operated

at dangerous high voltages. When semicon-ductor lasers debuted in the early 1960s, they offered the potential of much smaller, more efficient, and cheaper devices.

Semiconductor lasers are basically a sandwich of two materials: an “n-type” layer that contains an abundance of electrons, and a “p-type” that is missing electrons and is conceptualized as containing positively charged “holes.” This p-n junction is called a diode. When a current is injected across the diode, the electrons and holes f low towards one another, and if they meet and recombine they generate a photon. If that photon stimulates another excited electron-hole pair to combine and emit light at the same wavelength, phase, and direction, a cascading process is begun, with specific wavelengths getting built up and amplified as the photons bounce back and forth in the cavity. This is lasing.

One disadvantage of these “bulk” lasers is that they require a lot of current to make them lase. The active region of a bulk laser is so large, with a plethora of atoms that can absorb light, it is necessary to pump in a lot of electrons and holes to overcome this photonic reabsorption and generate light in a self-sustaining way. More current means greater power and more performance problems due to heat. A large active region also means a greater number of uncontrolled spontaneous (as opposed to stimulated) emissions of photons, which create optical noise.

Another big problem was a large “line-width enhancement factor.” As the current and gain increased, the index of refraction varied as well, which altered the wavelength of the standing waves inside the laser cavity. The end result was that the output wave-length and intensity danced around.

For Further Reading

Davis, R. W., Flores, A., Barrick, T. A., Cox, J.

M., Brozik, S. M., López, G. P., and J. A. Brozik.

2007. Nanoporous Microbead Supported

Bilayers: Stability, Physical Characterization,

and Incorporation of Functional Transmembrane

Proteins. Langmuir 23: 3864-72.

Gupta, G., Atanassov, P., and G. P. López.

2006. Robust Hybrid Thin Films that

Incorporate Lamellar Phospholipid Bilayer

Assemblies and Transmembrane Proteins. Biointerphases 1:6-10.

Ista, L. K., Perez-Luna, V. H., and G. P López.

1999. Surface-Grafted, Environmentally

Sensitive Polymers for Biofilm Release. Appl. and Environ. Microbio. 65: 1603-1609.

Rama Rao, G. V., Krug, M. E., Balamurugan, S.,

Xu, H., Xu, Q., and G. P. López. 2002 Synthesis

and Characterization of Silica-Poly(N-

isopropylacrylamide) Hybride Membranes:

Switchable Molecular Filters. Chem. Mater. 14: 5075-5080.

Sekar, M. M. A., Hampton, P. D., Buranda, T.,

and G. P. López. 1999. Multifunctional

Monolayer Assemblies for Reversible Direct

Fluorescence Transduction of Protein-Ligand

Interactions at Surfaces. J. Amer. Chem. Soc. 131(22): 5135-5141.

Singhvi, R., Kumar, A., López, G. P.,

Stephanopoulos, G. N., Wang, D. I. C.,

Whitesides, G. M., and D. E. Ingber. 1994.

Engineering Cell Shape and Function. Science 264: 696-698.

Vaidya, R., López, G. P., and J. A. López.

“Nanotechnology.” Kirk-Othmer Encyclopedia of Chemical Technology, Supplement to 4th Ed., pp. 397-437. John Wiley and Sons, 1998.

The laser TrailblazerLuke Lester’s bright ideas set records in quantum dot lasers and other optoelectronic devices.

Luke Lester’s Career Brief $11.5 M venture capital funding raised

>$10 M research funding attracted

26 h number

182+ papers

14 invited talks (most in Europe)

21 Ph.D. and M.S. students graduated

8 current students

Gautam Gupta, Plamen Atanassov, and Gabriel P. Lópeza

Department of Chemical and Nuclear Engineering, The University of New Mexico, Albuquerque,New Mexico 87131

Received 20 January 2006; accepted 14 February 2006; published 4 April 2006

This study describes facile methods based on sol-gel processing for the formation of robust thinfilms that incorporate phospholipid bilayer membranes and transmembrane proteins as multilamellarassemblies in cross-linked silica matrices. Transmission electron microscopy and x-ray diffractionwere used to examine the lamellar structure of the hybrid thin films containing1, 2-dioleyl-sn-glycero-3-phospoethanolamine DOPE , an unsaturated lipid, and1,2-dimyristoyl-sn-glycero-3-phosphocholine DMPC , a saturated lipid. While the d spacingmeasured for DOPE containing films varied from 35 to 48 Å depending on the amount of DOPEadded to the coating solution 10–1 wt % , similar changes were not observed for the filmscontaining saturated lipid, DMPC d spacing 43 Å . Addition of purple membrane containingbacteriorhodopsin to the DOPE/silica coating solution led to the formation of multilamellarvesicle-like structures within the thin films. Mild sonication of these solutions containing the purplemembrane prior to coating led to the formation thin films with planar multilamellar structures thatexhibit uniform d spacing. The study further investigates the effects of incorporation of gramicidinand sonication on the structure of hybrid films and speculates on the eventual application of thinfilms prepared in this manner. Reprinted with permission from Biointerphases. Copyright 2006,

DOI: 10.1116/1.2185654

Robust hybrid thin films that incorporate lamellar phospholipidbilayer assemblies and transmembrane proteins

Biointerphases 1(1), March 20066 61559-4106/2006/1(1)/6/5/$23.00 ©2006 American Vacuum Society

American Vacuum Society.

Figure 7. Even though these liposomes

are entrapped in silica (a), light-activated

proteins in the membranes still pump

protons out of the vesicles (lowering the

pH) when the light is on (b).

Bacteriorhodopsin-Liposome-Gel

Light ON

100 sec

Time

pH

br

6.80

6.78

6.76

6.74

6.72

6.70

6.68

6.66

6.64

6.62

blank

(b)

(a)

8 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 9

fa l l 2 0 0 8

it turned out that they were elongated, more like dashes. The lasers still provided a broad spectrum for tuning, but the gain spectrum wasn’t symmetric and the threshold current was higher than that of QDs. Lester wanted to investigate growing the InAs dots dif-ferently. However, it was market forces that ultimately determined the laser’s fate. By the winter of 2001, there were indications that the telecommunication boom was starting to go bust.

Zia’s second product, a single wavelength (1.3 micron) DWeLL laser, was aimed at the relatively short distances (less than 10 km) of data communication for which lasers had to be inexpensive and insensitive to tempera-ture fluctuations.

It’s with respect to the latter that DWeLLs, with their discrete, isolated energy states, are expected to have an edge. But this market supersaturated as well. Plus there was resistance to the 1.3 micron wavelength since many data systems were designed around 1.55 microns. QWs were so entrenched, it’s questionable whether companies understood the benefits of QD well enough to make the switch. But again, this issue was never resolved as decreased demand kept shelves of QW lasers sitting idle. By early 2003 it was clear Zia had to find another niche.

Unlocking Mode-LockingLester turned to mode-locking, a task QD lasers excel at better than any other semi-conductor laser. Mode-locking produces very short light pulses that are ideal for clock applications. Lester was hoping mode-lock lasers could replace electronic clocks in silicon microprocessors.

In addition to the atomic structure of the semiconductor and dot size, the length of the resonance cavity determines the pulse inter-val of a laser’s emitted light. Mode-locking puts all the standing waves in the cavity in phase with one another so that periodically they constructively interfere, producing an intense pulse of light. Quantum dots are ideal for what’s known as passive mode-locking, because all they require is the placement of a saturable absorber in the cavity. This non-linear element blocks low intensity waves, while letting high intensity ones pass through, and since the highest intensity light is at the center of the pulse, it gets ampli-fied while other frequencies are filtered out.

The saturable absorber in a DWeLL is exactly the same material as the gain section, only reversed biased, so the dots absorb light rather than emit it. A quantum dot’s emission and absorption wavelengths are the same.

In a 2001 Applied Physics Letters article, Lester, Malloy, Stintz, and others reported the first mode-locking QD laser, fabricated from the same wafer as their earlier work. They were two years ahead of any other group.

Their mode-locked device had two parts: a 4.73 millimeter (mm)-long active gain region and a 0.85 mm absorber. The device produced fully mode-locked pulses 17 picoseconds long with a repetition rate of 7.4 GHz at 1.3 microns. (A similar device is shown in Figure 2.) However Zia ran out of money before it could develop a viable product focused on optical clocking, and by the 2003–2004 time frame the technol-ogy lost out to “dual-core” and “quad-core” microprocessor technology. To some extent, QD mode-locked lasers remain a technical solution looking for a problem to solve.

QDPICLester returned to UNM full time in 2003. Shortly thereafter he became the Associate Director of CHTM and recently was given an endowed Chair in Microelectronics. But he didn’t give up on QDs or mode-locking. In fact, he designed a novel system for studying and developing QD passively mode-locked lasers. Called a quantum dot photonic inte-grated circuit (QDPIC), the device consists of up to 30 linked, independently controlled 0.5 mm-long optical waveguide sections made of GaAs embedded with InAs QDs. Depending on how a segment is biased or injected with current, it can be made into an absorber, a saturable absorber (absorption stops over a threshold), a passive wave guide, an active laser medium, or a spontaneous LeD emitter.

In a 2007 Optics Express paper, Lester’s group demonstrated how it optimized an 11-segment device by measuring the output of different geometric configurations of absorber, gain and passive sections. (See Figure 3.) For example, in going from a two-part laser to a three-part device, the researchers narrowed the pulse width from 9.7 to 6.4 picoseconds and increased the peak pulsed power by 49 percent. The group re-ports a record peak power of 224 mW for QD mode-locked lasers operating over 40 GHz.

They also systematically explored how moving the absorber section to different positions produces harmonics. For instance,

l u k e l e s t e r

Calculations in the 1980s indicated that reducing the size of the active region would lower the threshold current density and stabilize the output. The hope was to make a laser that combined the best of both worlds: the inexpensive production and small size of bulk lasers and the cleaner, more atomic-like properties of gas lasers. Indeed, the next technological step was to thin the active layer so much that the electrons and holes are confined to two dimensions and become trapped in energy potential wells where they have a greater chance of meeting and combining. As a result, quantum wells (QW) concentrate more electrons in energy states that contribute to laser action, making them more efficient and faster than bulk devices.

Lester, while still in graduate school at Cornell, was the first person to demonstrate the predicted increase in speed by making the first high-frequency QW laser diode that performed better than bulk.

The efficiency and linewidth could be improved even more by further restricting the active area to a sea of small islands called quantum dots (QD). These dots, measuring 20 nm wide and 10 nm high, are so small that they have quantized energy levels somewhat like that of an individual atom. A QD laser has about 10 million dots in its active region.

And that’s what Lester was looking at in his UNM lab in 1998 when he got so excited. It was the first QD laser with properties

superior to QW devices. It could operate at room temperature and had a low threshold current density of 26 Amperes/centimeter2 (A/cm2), 10 times lower than most QWs. (The current record, still held by UNM, is 10 A/cm2.) Its spread of distinct wavelengths spanned 190 nm, twice as broad as QW lasers. (The UNM team ultimately upped the span to 201 nm.) This wide range is due to the range of dot sizes (bigger dots produce longer wavelengths). early tests also indicated the device exhibited the lowest linewidth enhancement factor measured to date. Figure 1 is a schematic of the device.

The Zenith of ZiaThe laser wafer was designed and grown at CHTM by research Assistant Professor Andreas Stinz and electrical and Computer engineering Professor Kevin J. Malloy, who came up with a new molecular beam epitaxy recipe for assembling their quantum dot structure through a judicious choice of materials. They surrounded the InAs dots with GaInAs, which exhibited a greater bandgap energy than InAs but less than the GaAs substrate. This meant that electrons and holes were more attracted to the dots than the surrounding materials, and photons emitted from the dots were less likely to be absorbed or scattered by the GaAs. Lester coined the now trademarked term DWeLL (dots-in-a-well) to describe this structure. The group’s first three papers announcing

the DWeLL laser and a later paper explain-ing the physics have been cited over 521 times and their work resulted in 5 patents (4 US and 1 Japanese), which are the only patents issued to date for this laser technology. Other groups subsequently demonstrated that DWeLLs afford high speed, high efficiency and temperature-insensitive operation. Almost everyone in the laser community now makes QD devices this way.

But it wasn’t only scientists who were interested in the new devices. They piqued the interest of venture capitalists who were visiting Zia Laser, the start-up incorporated in May 2000 by Lester, Malloy, Stinz, and former UNM graduate student Petros Varangis. The VCs nixed Zia’s original plan for producing eye-safe QW lasers in favor of the much trendier and unique QDs. Lester, who was Zia’s Chief Technology Officer and then Chief research Officer, eventually raised $11.5 million (out of a total of $22.5 Zia was to receive) from investors while at the company. Zia’s work also attracted the press. Its r&D was covered by Laser Focus World, Photonics Spectra, The Economist, and other popular and trade periodicals.

Zia’s first device was a specialty laser, tunable around 1.55 microns for long-haul communications, which typically require a pure signal at each wavelength. This laser was made of InAs deposited on indium phosphide. Under an atomic force micro-scope, the InAs initially looked like dots, but

GaAs Be: 3E19 60 nm

2000 nm

110 nm

8.6 nm

3 nm

110 nm

2000 nm

300 nm

SUBSTRATE

Be: 1E17

InAs quantum dots

Si: 1E17

Si: 1E18

N+

GaAs

GaAs

GaAs

GaAs

Al0.7Ga0.3As

Al0.7Ga0.3As

In0.15Ga0.85As

In0.15Ga0.85As

Figure 1. Schematic of the quantum dot

laser structure

Figure 2. Grown on wafer (a), this mode-locked laser (b) has an 8 millimeter active gain length and contains

four layers of indium arsenide dots-in-a-well (c). The dots, which self-assemble during molecular beam epitaxy

growth are about 15 nanometers wide and 7 nanometers tall (d). Depending on how a section is biased, it can

act as either an absorber or emitter. The calculated quantum mechanical wavefunction for an electron inside a

quantum dot is shown in (e). The time domain is seen in (f). This mode-locked laser has an 5 Ghz repetition rate.

A G A P G A P G

A G P A G G A G

Figure 3. This quantum dot photonic integrated circuit consists of eleven 0.5 millimeter (mm)

sections, each of which can be made into an Absorber (blue), Gain region (red), or Passive

waveguide (cyan) depending on how that section is biased. With this flexibility, many kinds of

devices can be configured. For example, laser (a) has a 0.5 mm absorber and a 5 mm gain

region. With an absorber in the 5th section and a 2 mm and a 3 mm gain region on either side,

the device (f) produces higher order harmonics.

(a)

(d)

(b)

(e)

(c)

(f)

Gain SectionWafer

Absorber

Isolation Gap

Quantum Wells InAs“Dots-in-a-well” or DWELL Technology

Quantum Dot

00

0.5

1

1.5

2

20 40

Time Delay (ps)

Absorber bias: 5vGain current: 98mA

Inte

nsi

ty (

a.u

.)

60 80

4.4 ps

(a)

(b) (e)

(c) (d)

(f)

10 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 11

fa l l 2 0 0 8l u k e l e s t e r

when the 6th section is the absorber and the rest is gain, the frequency between pulses changes from 7.2 GHz to 14.4 GHz, the sec-ond harmonic. They observed higher-order harmonics at frequencies up to 115 GHz. Lester, who plays the cello, says this is analo-gous to harmonic generation on a stringed instrument. He is developing a paper for Nature Photonics studying very high laser harmonics in more depth and comparing them to the work of Jean-Louis Duport, a renowned cellist who wrote a treatise on very high “false harmonics” in the early 19th century.

The ability to produce different frequencies and rapidly switch between them— in effect, creating an arbitrary waveform—is of special interest for military laser sensing and ranging (LADAr) where high resolu-tion is needed to identify often obscured targets. Seeing how a precisely modulated laser signal is reflected back provides more detailed information than that from a single waveform.

Pulsed lasers are also useful as diagnostic tools for characterizing very high speed op-tical devices and for “optical time division multiplexing” that could enable communi-cation at very high speeds up to 100 gigabits per second. With Dan Kane at Southwest Sciences in Santa Fe, NM, Lester’s group is exploring other applications including two-photon high-resolution microscopy, laser cutting that eliminates the bead produced by current laser cutters, and terahertz sensing for homeland security. For all applications, QD lasers promise more compact designs than existing technologies.

In its studies, Lester’s lab has employed Frequency-resolved Optical Gating or FrOG (co-invented by Kane) to characterize laser pulse width. Conventional autocorrelation techniques suffer from ambiguities in pulse

shape. The researchers think that with FrOG, they can very accurately measure the non-linear chirp that broadens the pulse and compensate for it with an external grating or pulse compressor. They hope to compress the pulse down to the theoretical limit of 200 femtoseconds in a hand-held device.

Another possible application of a QDPIC is optical coherence tomography (OCT), which images tissues in medicine and paintings in art conservation. OCT is a non-invasive technique that penetrates 2–3 millimeters and produces images in-terferometrically. With the introduction of wide bandwidth light sources emitting wave-lengths over a 100 nm range, researchers have achieved micrometer resolution, better than ultrasound or magnetic resonance imaging. However, these original sources lack power.

QDs are a good candidate to broaden the bandwidth further due to their range of sizes and to boost power. With its QDPIC, Lester’s group is the first to simultaneously achieve a bandwidth of greater than 150 nm at a power greater than 1 milliwatt. In the April 1, 2007 IEEE Photonics Technology Letters, they report that their design can be adjusted to independently change the power and bandwidth, and unlike other methods, does not require a complex growth regime. (See figure 4.) One gain section is biased to saturate the low energy states in favor of high energy emission, while the other segment encourages low energy emission. By adjusting the length of each section, the researchers can saturate the lower energies sooner and use proportionately more dots to generate higher energy emission, produc-ing more power than other technologies which continue to dump more energy into lower transition states even as they pump higher ones.

Rerouted to TelecomIt turns out this interplay between the filling of lower energy transitions and higher ones is also important to the line width enhancement factor, , or chirp parameter. For ultrashort laser pulses, chirp means the pulses spread out due to the dispersion of the material through which they propagate, with some wavelengths moving faster than others. This is a big problem for telecommunications.

Lester and colleagues developed a theory to explain why: as the injection current increases, balloons from 4 to 60 as the lower energy states of the QDs are saturated. Frederic Grillot of the National Institute of Applied Sciences in rennes, France and now visiting CHTM, saw Lester’s paper and extended it, showing that after balloon-ing, the factor plummets to negative 30. The researchers think this happens when higher excited energy states begin “stealing” electrons and holes from the filled lower states. In a paper accepted by the Journal of Quantum Electronics, they elucidate the factors that control chirp, holding out the prospect for chirpless optical transmission over much greater distances. A laser with a negative value could counteract the posi-tive chirp of optical fibers, allowing signals to travel much farther without degrading.

even if the telecommunications market doesn’t rebound, Lester is sure to be inves-tigating some relevant QD or optoelectronic device. With a high h number of 26, over 182 papers under his belt (14 of which were invited especially in europe where QD research is thriving), Lester will continue to make an impact.

“Luke is an extremely respected member of the optoelectronics community,” says Pallab Bhattacharya, Charles M. Vest Distinguished University Professor of electrical engineering and Computer Science and James r. Mellor Professor of engineering at the University of Michigan. “I have been aware of Luke’s work since he was a graduate student at Cornell working on high-speed quantum well lasers. At UNM, Luke is considered as one of the pioneers in the development of QD lasers. In fact, he is credited for demonstrating the lowest threshold current (16 A/cm2—nearly zero!) in a QD laser. Later he went on to demonstrate extremely low chirp and line-width enhancement factor in these lasers. Luke has been a true pioneer.”

For Further Reading

Lester, L. F., Hwang, K. C., Ho, P., Mazurowski,

J., Ballingall, J. M., Sutliff, J., Gupta, S.,

Whitaker, J., and S.L. Williamson. 1993.

Ultrafast Long-Wavelength Photodetectors

Fabricated on Low-Temperature InGaAs on

GaAs. IEEE Photonics Technology Letters

5(5):511-514.

Liu, G. T., Stintz, A., Li, H., Malloy, K. J., and

L. F. Lester. 1999. Extremely Low Room-

Temperature Threshold Current Density Diode

Lasers Using InAs Dots in an In0.15Ga0.85As

Quantum Well. Electronics Letters 35(14):

1163-65.

Newell, T. C., Bossert, D. J., Stintz, A., Fuchs,

B., Malloy, K. J., and L. F. Lester. 1999. Gain

and Linewidth Enhancement Factor in InAs

Quantum Dot Laser Diodes. IEEE Photonics Technology Letters 11(12):1527-1529.

Lester, L. F., Stintz, A., Li, H., Newell, T. C.,

Pease, E. A., Fuchs, B. A., and K. J. Malloy.

1999. Optical Characteristics of 1.24 µm

Quantum Dot Lasers. IEEE Photonics Technology Letters 11(8): 931-933.

Stintz, A., Liu, G. T., Li, H., Lester, L. F., and

K. J. Malloy. 2000. Low-Threshold Current

Density 1.3 µm InAs Quantum-Dot Lasers With

the Dots-in-a-Well (DWELL) Structure. IEEE Photonics Technology Letters 12(6): 591-593.

Wang, R. H., Stintz, A., Varangis, P. M., Newell,

T. C., Lester, L. F., and K. J. Malloy. 2001. Room

Temperature Operation of InAs Quantum-

Dash Lasers on InP (001). IEEE Photonics Technology Letters 13(8): 767-769.

Su, H., Zhang, L., Gray, A. L., Wang, R.,

Varangis, P. M., and L. F. Lester. 2005. Gain

Compression Coefficient and Above-Threshold

Linewidth Enhancement Factor in InAs/GaAs

Quantum Dot DFB lasers. Proceedings of Physics and Simulation of Optoelectronic Devices (SPIE) 5722:72-79.

Xin, Y. C., Li, Y., Kovanis, V., Gray, A. L., Zhang,

L., and L. F. Lester. 2007. Reconfigurable

Quantum Dot Monolithic Multi-Section

Passive Mode-Locked Lasers. Optics Express

15(12): 7623-7633.

Xin, Y. C., Martinez, A., Saiz, T., Moscho, A. J.,

Li, Y., Nilsen, T. A., Gray, A. L., and L. F. Lester.

2007. 1.3-µm Quantum-Dot Multisection

Superluminescent Diodes With Extremely

Broad Bandwidth. IEEE Photonics Technology Letters 19(5-8): 501-503.

Wavelength, �m

Inte

nsi

ty, d

Bm

1.0

-25

-30

-35

-40

-45

-50

-551.1 1.2 1.3

Figure 4. Emission spectra of a three

segment quantum dot photonic integrated

circuit consisting of an absorber, a front

gain section biased to emphasize high

energy emissions, and one that favors the

lowest emissions. As the bias of the back

gain section is increased in steps of 10

milliamperes, the bandwidth broadens.

Room-Temperature Operation of InAsQuantum-Dash Lasers on InP (001)

R. H. Wang, A. Stintz, P.M. Varangis, T.C. Newell, H. Li, K.J. Malloy, and L.F. Lester

Abstract—The first self-assembled InAs quantum dash lasers grown by molecular beam epitaxy on InP (001) substrates are reported. Pulsed room-temperature operation demonstrates wave-lengths from 1.60 to 1.66 µm for one-, three-, and five-stack designs, a threshold current density as low as 410 A/cm2 for single-stack uncoated lasers, and a distinctly quantum-wire-like dependence of the threshold current on the laser cavity orienta-tion. The maximal modal gains for lasing in the ground-state with the cavity perpendicular to the dash direction are determined to be 15 cm-1 for single-stack and 22 cm-1 for five-stack lasers.

The authors are with the Center for High Technology Materials, University of

New Mexico, Albuquerque, NM 87106 USA (e-mail: rwang@chtm.unm.edu).

© 2001 IEEE. Reprinted, with permission, from IEEE Photonics Technology Letters.

IEEE PHOTONICS TECHNOLOGY LETTERS. VOL.13, NO.8, AUGUST 2001 767

Reconfigurable quantum dot monolithic multi-section passive mode-locked lasers

Y.-C. Xin1, Y. Li1, Vassilios Kovanis2, A. L. Gray3, L. Zhang3 and L. F. Lester1

1University of New Mexico, Center for High Technology Materials, 1313 Goddard SE, Albuquerque, NM 87106, USA 2AFRL/SNDP 2241 Avionics Circle, Wright Patterson AFB Ohio 45433-7320 USA

3Zia Laser, Inc. 801 University Blvd., Albuquerque, NM 87106 USA ycxin@chtm.unm.edu

Abstract: We investigate the dynamical response of a quantum dot photonic integrated circuit formed with a combination of eleven passive and active gain cells operating when these cells are appropriately biased as a multi-section quantum dot passively mode-locked laser. When the absorber section is judiciously positioned in the laser cavity then fundamental frequency and harmonic mode-locking at repetition rates from 7.2GHz to 51GHz are recorded. These carefully engineered multi-section configurations that include a passive wave-guide section significantly lower the pulse width up to 34% from 9.7 to 6.4 picoseconds, as well increase by 49% the peak pulsed power from 150 to 224 mW, in comparison to conventional two-section configurations that are formed on the identical device under the same average power. In addition an ultra broad operation range with pulse width below ten picoseconds is obtained with the 3rd-harmonic mode-locking configuration. A record peak power of 234 mW for quantum dot mode-locked lasers operating over 40 GHz is reported for the first time. ©2007 Optical Society of America

12 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 13

fa l l 2 0 0 8B e r t Ta n n e r

AS reSCUe crews and dogs dug through the carnage of 9/11, a few little-known helpers squeezed through small channels in the debris and searched for survivors deeper into the rubble than anyone else could go. They looked for hazardous materials, structural damage, and the easiest paths to excavate. It was the first time that robots, albeit with human operators, had been used for search and rescue in an urban setting.

Fast forward seven years to Herbert Tanner’s computer, which is running simula-tions that one day may evolve into a regiment of robots scouring a disaster zone. Some search for survivors, some do triage, others put out fires, or sniff for toxins, bombs, or gas. Perhaps aerial robots could perform reconnaissance and, when required, move an excavating or triage robot to where it can do the most good. Not only aren’t there any human operators in this scene, but the robots themselves are quickly figuring out how to get a job done together, and in some cases, performing tasks together that none of them could do alone.

Tanner, an Assistant Professor in the Department of Mechanical engineering, works from the bottom up. He first defines the elementary movements and actions each robot can perform. Then he writes algo-rithms that blend and sequence the behaviors of different robots into a problem-solving plan, all of which emerges from local interac-tions between the robots themselves. Tanner is working out the theoretical underpinnings of these decentralized control strategies and elucidating the conditions under which a multi-robot system will work before a lot of time and resources are committed to it.

“Bert Tanner’s highly imaginative work has focused on the issues of control and coordination of multi-agent robot systems,”

says Magnus egerstedt, Associate Professor of electrical and Computer engineering at Georgia Tech. “This is an area that has re-ceived a lot of attention recently. A lot of this work is based on Bert’s outstanding papers on Input-to-State stability and stable f lock-ing, and he has been one of the main players involved in defining the area worldwide. In fact, the whole area of graph-based Lyapunov functions for multi-agent systems can prob-ably be traced back to Bert’s two seminal papers on stable f locking in the 2003 Ieee Conference on Decision and Control, which more or less started an entire field.”

Tanner, who received the NSF CAreer Award in 2005, has given 12 invited talks over the last five years, published 62 papers, and written four book chapters.

Flocking to NatureLike many fields of science, inspiration for robotic control systems comes from nature, in particular the astonishingly fluid flocking or swarming behavior of animals, from fish to gnus. Imagine a flock of swallows swoop-ing, soaring, and gliding through the air with no apparent leader, moving in unison as if they were a single organism. remarkably, this group behavior emerges from the col-lective actions of individuals following a few simple local rules:

• separation (avoid colliding into neighbors),

• alignment (steer toward the average heading of local f lockmates) and,

• cohesion (move toward the average position of local neighbors).

These were first enunciated by Craig reynolds (then of Symbolics Inc., now of Sony Computer entertainment) who in 1986

created a computer model and 3-D anima-tion of f locking creatures he called “boids” (see www.red3d.com/cwr/boids). These boids evolved into the simulated bat swarms and penguin f locks in the 1992 movie Batman Returns and spurred the creation of a new area of computer graphics called “artificial life.”

Flocking and schooling have also starred in the studies of systems and control theorists who are interested in decentralized, autonomous, and self-coordinated “herds” of robots or unmanned vehicles for a variety of applications ranging from military surveil-lance to environmental cleanup to space construction. This approach represents a paradigm shift in robotics, away from centralized control and single multi-tasking robots to networks of many smaller, redun-dant, expendable, and autonomous ones.

This change is due to advances in com-puting, sensors, and communications, which allow for less expensive, more energy-thrifty robots that have greater computing power and sensitivity, and can send and receive signals from much longer distances. This makes for more robust robotics; if one robot fails, others can fill in. Tanner and others are focusing on local control because it scales up nicely. Trying to coordinate the movement of tens, hundreds, or thousands of robots at once from one central computer would be prohibitively complex.

This is also why securing a strong theoretical foundation before an algorithm is implemented is important. Formal proofs are valuable because they give scientists con-fidence that as long as they meet the required conditions of the proof, simulations and real life robot trials will behave as expected. And, because solving the dynamics of a highly complex, multi-bodied, and interactive

system is extremely difficult (NP Hard in computer science parlance), a formal proof also tells scientists whether a solution is achievable and worth pursuing.

Control and ConnectionIn one of the 2003 papers alluded to by egerstedt above, Tanner was the first to prove formally that the general case of f locking will converge, that is, reach a steady state in which flocking behavior will always emerge from local rules, and boids or robots won’t be flying off in different directions or colliding into one another. His co-workers were Ali Jadbabaie and George J. Pappas at the University of Pennsylvania.

Whether a squad of mobile robots or schools of fish, group members are tracked and described mathematically by dynami-cal systems using Newtonian equations of motion that determine the position and velocity of each robot. A control input drives the robots by steering them (via acceleration) in a desired direction. It consists of two parts. The first is responsible for cohesion and col-lision avoidance by keeping the robot a set distance from its neighbors. This is achieved by introducing an artificial potential, func-tion V, that depends on the relative distances between robots. V goes to infinity as the robot spacing approaches 0 and it attains a unique minimum when neighboring robots are separated by a desired distance. The second part of the control input makes all the robots have the same heading by aligning a robot’s velocity vector with a weighted aver-age of the velocities of its nearest neighbors. Its neighbors do the same thing with their neighbors and so forth as the control law propagates through the entire group.

By combining these results of mechanics and classical control theory with algebraic graph theory, Tanner and his colleagues were able to prove that a stable, tight forma-tion results and that all the robots’ velocity vectors asymptotically approach the same value and direction. This stability is achieved only if the information network among the robots (represented by a graph with vertices at each robot) stays connected, so that each robot is continually aware of the positions and speeds of other nearby robots. Studying the relationship between stability and the connectivity of the network is a major focus of Tanner’s work.

The first part of the 2003 paper assumes that once a network topology is established, it never changes; e.g., if robot A is connected with robot B and robot C, it will forever have that configuration. robots A, B, and C will continue to be neighbors. The second part keeps the overall network going, but instead allows individual interconnects to change, e.g., after a time robot A might connect to B and D because C moved out of range, while D drifted into A’s neighbor-hood. robot A may not be directly linked to the farthest robot Z, but may still be con-nected by a route that meanders from B all the way to Z. Again, Tanner proved stability, but he had to add the mathematical tools of non-smooth Lyapunov stability analysis in the mix because the control law in this case becomes discontinuous. From 0 to some value r, the new potential U is the same as V above, but when the distance between ro-bots exceeds a certain value r, the potential becomes flat, indicating that no interaction takes place between robots that are separated by a distance greater than r.

Tanner’s group conducted a 10- and a 50-robot simulation based on this control law. Initial positions and velocities were as-signed randomly, as was the configuration of

the communication network graph. Figure 1 shows how the robots (represented as points) in the 10-robot simulation avoid collisions while aligning in a tight formation moving in the same heading.

After these papers, other researchers began exploring the connectivity issue in more detail. What if there is noise or inter-ference in the information network? What if information f low is only in one direction? Tanner’s group has examined what happens if there is an obstacle in the environment that forces the group to split up and inter-rupt communication. In a 2004 paper, Tanner extends the control law so that it repels robots from obstacles and drives them toward a single rendezvous point. Again ex-ploiting algebraic graph theory properties, he demonstrates that arbitrary communication interruptions due to the group splitting up do not affect stability. When the robots do split into smaller groups, they maintain the internal synchronicity of velocities within their subgroups, and when they rejoin other groups near the rendezvous point, they again align. In effect, he reports, the common objective “binds” the group together so that connectivity is no longer required. In a recently submitted paper, Tanner further

Figure 1. This computer simulation of ten robots (represented as small dots) driven by local control

laws shows how a traveling assembly of robots evolves. Velocity vectors are depicted as arrows,

dotted lines are the trajectory trails for each robot, and the number at the base of each figure

corresponds to the elapsed time in seconds. The robots start at random positions and velocities,

but soon coalesce into a tight formation moving towards one heading. The shape of this formation is

determined by the artificial potential function written into the control law. (a) Initial configuration.

(b) Cohesion forces increase connectivity. (c) A tight formation is created. (d) The group moves in

the same direction. (e) Steady state.

(c)

(e)

(d)

robots to the rescueBirds and bees do it. Now robots can, too. Bert Tanner explores how simple rules between individual robots can lead to dynamically rich group behaviors, such as flocking and swarming, and how one day this kind of collective action could save lives.

5.11199 50.0985 100

(a) (b)

0 1.08636

14 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 15

fa l l 2 0 0 8B e r t Ta n n e r

explores other ways of maintaining con-nectivity by negotiating links in the f lock through auction algorithms.

Robotic DiversityTaking another cue from nature, Tanner also envisions groups of different sorts of robots working together, just as bees perform differ-ent jobs to keep the hive running smoothly. He is one of the first researchers to design a methodology for coordinating a team of het-erogeneous robots, and his specific symbolic approach toward the problem is unique.

As a first step, he considers a party of unmanned ground and aerial vehicles (UGVs and UAVs respectively) that are sent out to scout a safe corridor through hostile terrain. The advantage of adding air support is the UAVs’ ability to provide a greater swath of surveillance by scanning the ground for threats (Figure 2). The scenario calls for the ground vehicles to spread out while keep-ing within an ideal distance of neighbors, to move in unison, and to estimate the center of the group using local rather than global information.

The scenario also demands that the location of the ground forces’ moving center is broadcast to the air vehicles, so they can f ly in an orbit at a set tangential velocity around the centroid. each orbiting plane is assigned a different radius for orbit. Since the centroid is moving, the planes’ trajectories will be spirals. The control design, naturally, requires the planes to do this while avoiding midair collisions.

Tanner has added the additional, but more realistic, constraint that communi-cation is not instantaneous. Instead, the ground vehicles take turns broadcasting their position and velocity to their neighbors so as to minimize communication bandwidth and avoid interference between broadcasts. He uses a time-division-multiple-access protocol. Since each car broadcasts its posi-tion and velocity at a particular time step, its neighbors will only have this latest informa-tion about it until it transmits again and the data is updated. This means that each UGV (and UAV) controller makes calculations based on delayed information.

Through proofs and adept simplifica-tions, Tanner and his graduate student, D.K. Christodoulakis (now at IBM), succeeded in devising a model that satisfies all the scenario’s requirements and stabilizes to the steady state.

Because the UGVs take turns broadcast-ing, discrete time dynamics describe the positions and velocities of the vehicles. Flocking behavior is once again achieved by having each controller steer its car towards the average velocity of its neighbors. Critical to this result, however, is the requirement that a robot and its neighbors use the same state values. even if a robot itself has updated values for its own position and velocity, it cannot use this data until its turn to broad-cast to its neighbors comes around again. Interestingly, the model’s convergence is independent of the length of the time delay. In a 2007 paper appearing in Robots and Autonomous Systems, the researchers also

demonstrate stability even if not all the UGVs are connected to every other vehicle.

At steady state, all the UGVs estimate the same centroid of the ground group, which is broadcasted to the UAVs. The UAVs settle into their designated positions with the help of a controller that changes the angular velocity of the planes if collision seems pos-sible. Some collision avoidance is illustrated in Figure 3, which shows simulations of the model in action.

Whole > ∑PartsTanner wants to go one step beyond coor-dination to cooperation. For example, the ground/aerial scenario above would be en-riched considerably if the aerial vehicles were helicopters capable of picking up and moving ground vehicles over inaccessible terrain or to another spot more important to an overall objective. Is it possible for robots with differ-ent sets of skills to decide to come together and perform tasks none of them could do on their own? Would the total robotic system conclude that working together is a way to solve a problem and, given a suite of possible movements, would the robots figure out how to orchestrate those actions in order to work together harmoniously?

It’s as if Tanner gives the robots an al-phabet of actions with rules stipulating what letters can and cannot go together. Without providing a dictionary, he’d like the robots to come up with words that are real and also necessary for solving a particular problem. So far, the research team requires that the elements of cooperative behavior be specified

a priori; the programmer has to encode ex-plicitly the fact that under some conditions a helicopter can pick up a UGV to move it from point A to point B. Otherwise, the systems will not know on their own that there is a way in which they can work together.

Tanner and student Wenqi Zhang have been investigating this approach by playing with a game invented in the 19th century called the Sliding Tile Puzzle. It consists of a 4 x 4 grid comprised of one unoccupied space and 15 sequentially numbered tiles that can slide into the open space. The goal of the game is to put the 15 scrambled tiles in order by progressively moving different tiles into the empty space.

In the researchers’ formulation, there are two types of “robots”: the tiles that can do nothing by themselves, and one agent robot that can move between the tiles and push adjacent tiles into an empty square. It’s

notable that traditional ways of combining or “composing” different sets of behaviors for this problem would produce no tile movement at all.

Instead, Tanner and Zhang have in-troduced a new formalism to def ine comp ositions of the sliding puzzle system using what are known as “extended motion description languages” to translate these control algorithms into code. In essence they provide the agent and tiles with additional actions that only become active when they are composed with the other type of robot. The results of a small simulation are shown in Figure 4 where a robot successfully moves a tile from position 1 to 6. Tanner hopes to test their methodology with real robots and styrofoam blocks soon.

One thing that is not certain is, once individual elementary actions are spelled out, will the system be able to come up with a plan and a sequence of movements to solve a problem, especially in a reasonable amount of time and with a reasonable amount of en-ergy? For rescue applications, these solutions don’t necessarily have to be optimal or the very best, just reasonable and fast. Searching multidimensional space of all possible solu-tions for the best one is an NP-hard problem, while the general case of the sliding puzzle problem is P-SPACe complete. But by using abstraction to group actions or behaviors together that share the same outcome and to transfer the harder problem of solving the continuous dynamics of the robots into a purely discrete (time step) domain, Tanner’s system can reduce the solution space and hopefully come with a “meta-plan” quickly. Operating in the discrete regime enables Tanner to circumvent some of the difficulties encumbering alternative abstraction approaches.

Back to the Futurerobots have captured the human imagina-tion since ancient Greece, where mythology held that Hephaestus, the god of metalwork, forged mechanical helpmates. Multi-agent robotic systems in particular have found their way into contemporary culture in science fiction and in films such as Minority Report, with its swarms of ID-checking robots. Making real-life ensembles of actual robots with behaviors as nuanced and fluid as a f lock of swallows, however, remains a

challenge. With his elegant proofs and forays into heterogeneous robot teams, Bert Tanner is on the road toward making his vision of rescue robots a reality.

Further Reading

Cortez, R. A., and H. G. Tanner. 2008.

Radiation Mapping Using Multiple Robots. 2nd ANS International Joint Topical Meeting on Emergency Preparedness & Response and Robotic & Remote Systems.

Tanner, H. G. 2004. Flocking with Obstacle

Avoidance in Switching Networks of

Interconnected Vehicles. Proceedings of the ICRA IEEE International Conference on Robotics and Automation 3:3006-3011.

Tanner, H. G. and D. Christodoulakis. 2006.

Cooperation between Aerial and Ground

Vehicle Groups for Reconnaissance Missions. 45th IEEE Conference on Decision and Control 5918-5923.

—. 2007. Decentralized Cooperative Control

of Heterogeneous Vehicle Groups. Robotics and Autonomous Systems 55(11): 811-823.

Tanner, H. G., Jadbabaie, A., and G. J. Pappas.

2003. Stable Flocking of Mobile Agents,

Part I: Fixed Topology. Proceedings of the 42nd IEEE Conference on Decision and Control 2:2010-2015.

—. 2003. Stable Flocking of Mobile Agents,

Part II: Dynamic Topology. Proceedings of the 42nd IEEE Conference on Decision and Control 2:2016-2021.

Tanner, H. G., Pappas, G. J., and V. Kumar.

2002. Input-to-State Stability on Formation

Graphs. Proceedings of the 41st IEEE International Conference on Decision and Control 3(10-13):2439-2444.

Zhang, W. and H. G. Tanner. 2008. Composition

of Motion Description Languages. Hybrid Systems: Computation and Control: Proceedings of the 11th International Workshop, HSCC 2008, Lecture Notes in Computer Science Series, ed. by Egerstedt, M. and B.

Misrha. Springer-Verlag.

Figure 4. Successive time shots of the

sliding puzzle problem being solved. By

moving around the numbered gridlines

and pushing tiles into empty squares, the

robot ( ) succeeds in moving one tile ( )

from position 1 to position 6.

Figure 3. This series of snapshots

shows the time evolution of seven

ground and four air vehicles, all of

which start at random velocities

and positions. The lines connecting

ground vehicles correspond to

information linkages; the airplanes

do not share information. Dotted

lines mark trajectories. The second

and third snapshots are taken

just after planes 2 and 4 avoided

colliding. By the third snapshot

(3000th time step), the ground

vehicles are moving in unison in a

symmetric polygon formation and

the aerial vehicles are spiraling

around the ground centroid.

Step 1 Step 2

Step 3 Step 4

Step 5 Step 6

Step 7 Step 8

Figure 2. An intelligence-surveillance-reconnaissance

mission of a group of ground vehicles (dots) and aerial

vehicles. The latter provide early warning by surveying

the ground in a wider sweep (cones) than the ground

vehicles could do.

xx x

Initial configuration

−3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5−2

−1.5

−1

−0.5

0

0.5

1

1.5

12

3

4

1

2

3

4

5

6

7

x

yy

−4 −2 0 2 4 6

−4

−3

−2

−1

0

1

2

3

41

2

3

4

1

2

34

5

67

x

y

y

40 42 44 46 48 50 52 54

−12

−10

−8

−6

−4

−2

1

2

3

4

12

34

56

7

x

y

Final configurationFinal configuration

y

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

17 18 19 20 21

31 32 33 34 35

45 46 47 48 49

59 60 61 62 63

73

26 27 28 29 30

40 41 42 43 44

54 55 56 57 58

68 69 70 71 72

74 75 76 77

22 23 24 25

36 37 38 39

50 51 52 53

64 65 66 67

78 79 80 81

16 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 17

fa l l 2 0 0 8e d l s c h a m i l o g l u

protecting the U.S. communications and energy grids. He also chaired the High Power Sub-Panel of the National Academies Panel on Directed energy Testing. During his 20 years at UNM, he has attracted over $16 million in funding, published nearly 200 papers, received three patents, won numer-ous awards, and even served as a technical consultant to an MGM Studio production of “The Outer Limits.”

Microwave BasicsThe object of a high power microwave tube is to transfer power from the electron beam passing through it to the eM wave coming out of it. That’s hard to do because two con-ditions must be met simultaneously. First, the beam has to be bunched into clumps of electrons that will radiate and induce eM fields at the desired frequency. This means reinforcing some of the beam’s natural f luc-tuations and quenching others.

Second, the phase velocity of the bunched electron beam has to match the phase velocity of the electromagnetic mode in the device for resonance to occur. As shown in the disper-sion diagram in Figure 2a, the phase velocity of the electromagnetic wave (the slope of the dispersion curve) is greater than the phase velocity of the electron beam. Hence, there is no energy exchange. But, as shown in Figure 2b, the introduction of a periodic spatial perturbation to the waveguide wall causes the dispersion relation for the electro-magnetic wave to be periodic and intersect

the dispersion relation for the beam. In ef-fect, this slows down the phase velocity (/k) of the eM wave, so the periodic structure is known as a Slow Wave Structure (SWS).

Note from the dispersion curve that at the point of intersection, the group veloc-ity of these waves (∂/∂k, or the slope of the dispersion curve) is negative, meaning that the excited wave travels backwards against the beam. That’s why the tube in Figure 3 is called a Backward Wave Oscillator (BWO). In practice this backward wave is ref lected forward from the cutoff neck so that it exits from the horn antenna in Figure 3.

The SWS in this device induces the elec-tron bunches to radiate coherently. When electrons radiate in phase, the resultant power goes up as the number of electrons squared, and since there can typically be 1012 electrons passing through the waveguide every nanosecond, that can be considerable. Peak power of greater than 1 gigawatt is achievable in a BWO, but because of pulse shortening limitations (see below) this can be sustained for no longer than 100 nano-seconds. The BWO was the first microwave device to be driven by a pulsed, high-current relativistic electron beam, an accomplish-ment that ushered in the era of high power microwave generation in the late 1960’s and early 1970’s.

BWO: Tuning into the Smart Tubeearly in his career, Schamiloglu collabo-rated with russian experts to perform what

Gilgenbach calls “pioneering” experiments on Sinus-6, a gigawatt relativistic BWO at UNM. Schamiloglu acquired the Sinus-6 in 1991 on a trip to the Soviet Union as the regime was nearing its end and scientific exchanges with the west were opening up. During its initial year of operation at UNM, however, the Sinus-6 output power fell from the 600 megawatt (MW) that it had been producing when it first arrived. The russian team came to UNM, and got the BWO up to power by hammering the Slow Wave Structure and ramming it into the device. Schamiloglu was puzzled. Why should the power output be so sensitive to the SWS position inside the BWO?

Schamiloglu and his Ph.D. student Larry Moreland decided to explore the position dependence by inserting two smooth-walled spacers at the ends of the SWS (L1 and L2 in Figure 3). The spacers change the length of the cavity and the phase between the forward and backward traveling waves, allowing the researchers to see the effects of constructive and destructive interference between the two kinds of waves. They found that incremental quarter-wavelength changes in the spacer length significantly affects both the relative power efficiency and microwave frequency, producing a series of minima and maxima in each (see Figure 4.) From these laboratory experiments and simulations, Schamiloglu’s team, together with its russian colleagues, concluded that the forward wave harmonics interact significantly with the electron beam.

IN THe late 1880s, Heinrich Hertz used Leyden jars, two iron point sparks, and zinc gutter sheets as antennas to broadcast and detect microwaves. He was the first person to experimentally verify Maxwell’s equa-tions and support his contention that visible light was but a small portion of the electro-magnetic spectrum.

Compared to Hertz’s experimental setup, today’s microwave sources are considerably more sophisticated, powerful, and ubiqui-tous. They cook dinners, carry cell phone signals, and accelerate particles to tremen-dous speeds in colliders, giving us unique glimpses into the subatomic world. Since the invention of radar during World War II, they have also played a decisive role in military ventures. Today, high power microwaves (HPM) can disrupt enemy electronics and disperse unruly crowds.

Instrumental to many of the advances in the understanding of HPM sources is UNM’s edl Schamiloglu, Professor in the Department of electrical and Computer engineering. Schamiloglu’s insights into how the kinetic energy of a beam of electrons is transferred to the electromagnetic (eM) energy of microwaves are paving the way for devices operating with higher efficiency, greater power, longer lifetime, and more compact size. (Figure 1 shows the power and frequencies of typical HPM applications.) Some of Schamiloglu’s specific contributions include the first tunable Backward Wave Oscillator, insight into “pulse shortening” phenomena that limit all HPM devices, a novel magnetron design, and improved computer simulations.

“edl Schamiloglu is one of the world leaders in the field of HPM generation and

applications, says ronald M. Gilgenbach, Professor of Nuclear engineering and radiological Sciences at the University of Michigan. “He is considered the top HPM researcher of his mid-career generation.” Gilgenbach adds that Schamiloglu is the coauthor and editor of two of the most important books on HPM: the textbook High-Power Microwaves and High Power Microwave Sources and Technologies, a state-of-the-art compilation for the f ive-year Multidisciplinary University research Initiative.

An Ieee Fellow and Senior editor of IEEE Transactions on Plasma Science, as well as Associate editor of the Journal of Electromagnectic Waves and Applications, Schamiloglu is also the Director of the Institute for Infrastructure Surety, which he co-founded in 2004 to study ways of

Maestro of MicrowavesFor two decades, Edl Schamiloglu has conducted high power research on high power microwave sources.

Figure 1. Range of applications for high power microwave sources.

Figure 2. Dispersion

diagrams. In (a) the

dispersion line for the

electromagnetic wave

(EM) does not intersect

that of the electron beam

so no energy is exchanged

between them. Adding a

corrugated structure (with

a period of length d) into

the tube undulates the EM

dispersion curve and slows

down its phase velocity

so that it does cross the

beam line (b). The EM phase

velocity is negative at the

intersection point, meaning

that the wave travels

against the beam. Hence

the name backward wave

oscillator (BWO).

1

2

3

4

5 8

9

6

L1 L2

711 10

Figure 4. The power

efficiency and frequ-

ency of a Backward

Wave Oscillator vary

periodically as the length

of the smooth circular

waveguide (as controlled

by L1 and L2 in Figure 3)

changes causing forward

shifting. Results from

both experiment and a

computer simulation called

TWOQUICK are shown.

10 GW

1 GW

100 MW

10 MW

1 MW

100 KW

10 KW

1 KW

100 W

10 W

1 W0.3 31 10 30

Frequency (GHz)

Pow

er

100 300

Communications

Radar

Ultrawideband Linear Colliders

Fusion Heating

Power Beaming

Directed Energy Weapons

TWOQUICK Experimental

1.2

1.0

0.8

0.6

0.4

0.2

0.00 8 16

Shifting (mm)

Re

lati

ve E

ffic

ien

cy

24 32 40

9.75

9.70

9.65

9.60

9.55

9.50

9.40

9.45RF

Fre

qu

en

cy (

GH

z)

0 8 16

Shifting (mm)

24 32 40

Figure 3. Backward Wave Oscillator setup with forward and

backward shifting: (1) capacitive voltage divider, (2) Rogowski

coil, (3) cutoff neck, (4) cathode, (5) A-K gap, (6) magnetic field

coils, (7) slow wave structure, (8) smooth circular waveguide

and shifting lengths L1 and L2, (9) electron beam, (10) output

horn antenna, and (11) reflection ring.

(b)

(a)

18 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u a c c e l e r a t i n g i m p a c t 19

fa l l 2 0 0 8e d l s c h a m i l o g l u

This design has other benefits as well. Current running along each emitter creates circular magnetic field lines around it. Thus, electrons coming out one side of an emitter will see a magnetic field pointing in the direction opposite to that seen by electrons coming out on the other side. This creates a “wiggler” effect, providing magnetic prim-ing or modulation that further induces the electrons to radiate eM energy.

When a magnetron starts up, circulating electrons bunch up into spokes that rotate around the device. In a 2005 Physical Review Letters article, Fuks and Schamiloglu reported on simulations showing that electrons emit-ted by the transparent cathode bunch up and form rotating spokes faster than electrons in the conventional A6 magnetron (Figure 7). What’s more, as shown in Figure 8, the power ramps up twice as fast, is significantly greater in magnitude, and has a more con-stant profile than the other magnetron. This

means the transparent device delivers more energy overall. Furthermore, they found that they could control the mode of operation by adjusting the magnetic field, thereby elimi-nating mode competition. The simulations also suggest that efficiencies of 25.5 percent are attainable by the A6 driven by a transpar-ent cathode as compared to 15 percent when driven by a solid cathode.

Schamiloglu’s group plans to test these simulations in the laboratory. If the results are favorable, a commercial magnetron manufacturer intends to incorporate the transparent cathode design into its product line. It is not only the improved efficiency of the transparent cathode that is attrac-tive to industry. By replacing conventional thermionic emitters, with their heaters and ancillary components, the manufacturer could slim down the total package consider-ably. Small size and portability are crucial, especially with the keen interest in countering

improvised explosive devices (IeD), which have killed more than 1,800 people in Iraq since the war began. Schamiloglu’s group is part of an Office of Naval research basic research program developing neutraliza-tion technologies against IeD’s. In this capacity, another practical test for the transparent cathodes is to see whether they could be phase-locked together to increase output power.

Better SimulationsOne paradigm shift in the HPM field over the last two decades has been a phenom-enal advancement in numerical modeling. Because there is no closed form analytic theory to describe HPM sources, computer simulations are the only means of testing and optimizing designs. But before the mid-1990s, computer simulations matched experimental results only qualitatively at best. Schamiloglu and co-workers helped

They discovered that the BWO’s finite length makes the reflected wave and its harmonics more important to BWO operation than anyone had previously realized.

exploiting the periodic effects afforded by the spacers, the group also constructed the first BWO that can be tuned over a bandwidth of 500 MHz (centered around 9.5 GHz) without sacrificing power. The BWO is now one of only a few pulsed, tunable HPM sources.

Schamiloglu teamed up with Professor of electrical and Computer engineering Chaouki Abdallah and M.S. student Vatche Soualian to go one step further. They computerized an iterative learning feedback system that regulates the output frequency and peak power by moving the SWS and controlling the beam current between pulses. This was the first practical demonstration of what is known as a “smart tube” in the gigawatt range, a device capable of auto-matically learning and adjusting its output to achieve power and/or frequency specified by the user.

The first application of the smart tube was to explore photonic crystals. Schamiloglu’s group demonstrated for the first time that the bandgap of a photonic crystal could be

exploited as a HPM filter, beam shaper, and quasi-optical reflector. The researchers also used the smart tube to study the disper-sion diagram of a photonic crystal (a 1-D, 2-D, or 3-D spatially periodic structure). Previously, textbooks had declared that a curiosity in the dispersion curve predicting group velocities faster than the speed of light were experimentally meaningless because the attenuation of the wave propagating through the crystal would prevent observa-tion. But the group demonstrated otherwise and produced the first direct observation of superluminal group velocities. The BWO’s ability to produce high-power, short pulses at convenient frequencies is what made the experiment possible.

Pulse ShorteningSchamiloglu’s group has also addressed the pulse shortening problem. In the early 1990s as the race to develop HPM sources with ever higher peak powers and longer pulses progressed, it became apparent that all devices were limited by a phenomenon called pulse shortening. Peak power can be increased, but at some point the pulse will be cut off even though energy is still being fed to the system. A maximum pulse energy of about 1 kilojoule seems to be the limit of delivered energy. Pulse shorten-ing changed the focus of research towards testing the idea that unwanted surface plasmas were somehow affecting the HPM source operation.

Looking for these plasmas with real time, in situ measurements in these inaccessible devices and their hostile environments is a challenge. But Schamiloglu’s post-doc, Frank Hegeler, (now at the Naval research Laboratory) succeeded in setting up a laser interferometer to conduct the first measure-ments of plasma electron density inside an HPM BWO. The researchers discovered two episodes of plasma production. The first, they believe, is due to the electron beam scraping the cutoff neck wall, and they have in fact reduced this initial phase by replac-ing the neck with a non-intercepting Bragg ref lector. The second is caused by intense electric fields breaking down the SWS walls. Particle-in-cell computer simulations show that the resultant plasma ions effectively shut down the growth of oscillations, terminating the pulse.

A New Idea for the MagnetronSchamiloglu’s contributions are not lim-ited to the BWO. He and research Professor Mikhail Fuks have also improved the operat-ing features of the magnetron. In a sense, a magnetron is a BWO bent into a closed ring, with the periodic SWS turning into the anode resonance cavities whose sizes and shapes determine the microwave frequency (see Figure 5). In a magnetron, however, a voltage between the electron-emitting cath-ode and the outer anode propel electrons out radially. Moreover, a magnetic field applied axially (into the page in Figure 5) def lects the electrons in the e x B (circular) direction around the cathode.

Invented by the late George Bekefi at the Massachusetts Institute of Technology in 1976, the relativistic A6 magnetron is a high-voltage, high current, pulsed power version of the conventional device, which maxed out at about 10 MW of power. The relativ-istic magnetron is one of the most compact, powerful, robust, and mechanically tunable HPM sources, able to produce frequen-cies from about 1-10 GHz with single-shot powers up to 4 GW.

This is not to say there is no room for improvement. relativistic magnetrons suffer from pulse shortening like all other devices. While conventional magnetrons operate at efficiencies up to 90 percent, the efficiency of their relativistic cousins has yet to match half that value.

Fuks and Schamiloglu came up with a novel approach that takes on both these problems. In place of the large solid cath-ode, they proposed a transparent cathode consisting of a series of axial, emitter strips or rods spread out evenly around the original cathode outer boundary, as shown in Figure 6. Its name is due to the fact that the cathode emitters are transparent to the azimuthal wave electric field inside the magnetron. This is the first novel design ap-proach to increasing magnetron power and efficiency proposed in many years.

In traditional magnetrons, Maxwell’s equations require that the wave electric field tangential to the conducting cathode surface go to zero. Fuk’s and Schamiloglu’s idea instead makes the field go to zero at the axis, allowing it to penetrate the emitters (Figure 6c) and immediately act on the elec-trons as they emerge from their surface.

Figure 5. Cross-sectional view of a

magnetron. Electrons emitted from the

cathode travel toward the more positively

charged anode, but because of the

magnetic field (applied into the page),

they move in the E x B direction, spiraling

around the cathode. At steady state,

the electrons radiate at the magnetron

frequency, which is determined by the size

and shape of the resonance cavities.

Bz

Er

e—

vane (anode)

cavity

solid cathode

Transparent

E�(r)

r0Rcathode Ranode

Solid

gap

Figure 6. To make a transparent cathode (a and b), the solid cathode of an A6 Magnetron (Figure 5) is replaced with 6 thin cathodes. The advantage of this

configuration is that the wave azimuthal electric field no longer goes to zero at the surface of the cathode and so is significantly larger in the gap between

the anode and cathode (6c). This makes oscillations start faster.

Pow

er

[MW

]

600

400

200

0

Time [ns]

600

200

400

010 20 30 40 50

Pow

er

[MW

]

3.8 ns 4.4 ns 5.9 ns 15.5 ns

3.8 ns 4.4 ns 5.9 ns 15.5 ns

(a) (b) (c)

Figure 7. 3D simulations show how electrons start to form spokes (and hence set up resonance

modes) faster in the transparent cathode (bottom) than in the solid cathode magnetron (top).

Figure 8. Simulations of the radiated power from (a) solid cathode and (b) transparent

cathode show that the transparent pulse ramps up faster, while the solid cathode suffers from

competition between the and 2 modes.

Pow

er

[MW

]

600

400

200

0

Time [ns]

600

200

400

010 20 30 40 50

Pow

er

[MW

]

(a)

(b)

20 i n n o v a t i v e r e s e a r c h w w w . s o e . u n m . e d u

fa l l 2 0 0 8

a c c e l e r a t i n g i m p a c t 21

to improve the fidelity of computations by spending a year meticulously tracking down the source of a discrepancy between experi-mental observations and simulations.

Today there is so much faith in these simulations that the Air Force research Laboratory, as an example of one of many HPM research centers, will not begin to build an HPM source until scientists are entirely happy with the predicted device characteristics, a turn of events which, of course, saves time and money.

Sometimes the computer simulations come up with behavior that is not expected. Conventional wisdom holds that electron spokes form in magnetrons because of a periodic physical structure like the anode vanes or due to an additional magnetic field ripple. George Bekefi introduced the latter when in 1982 he proposed the relativistic smooth bore magnetron or free electron laser, a cylindrical cathode surrounded by a featureless cylindrical anode. In addition to the crossed e and B fields, Bekefi sandwiched the two electrodes between an annular magnetic wiggler or a series of magnets with north and south poles alternating around the ring. It had been assumed that this wiggler is needed to get the electrons to bunch and form spokes, but in simulations Fuks and

Schamiloglu have discovered this isn’t neces-sary. Given a smooth bore magnetron of a set circumference, they can form electron spoke patterns with different symmetries, depend-ing on the value of the applied magnetic field. The researchers call this structure the “simplest crossed field ubitron” (ubitron was the ersatz name first given to the original free electron laser).

So 120 years after Hertz, there are still eM mysteries to be solved. even the conven-tional magnetron is not fully understood; it’s assumed, for example, that oscillations grow from noise, but how this process works has yet to be f leshed out — lots of interesting questions to be passed along to the next gen-eration. Here, Schamiloglu has contributed as well, serving the community by chairing many committees and conference sessions and reviewing manuscripts and proposals for numerous journals. In June 2007, for instance, he organized and chaired the joint Ieee International Conference on Pulsed Power and Plasma Science. He has gradu-ated 18 Ph.D. and M.S. students, most of whom work in national, DoD, and commer- cial labs.

His two-decade focus has also been a boon to UNM. “The vast majority of ee de-partments have shortsightedly limited their

coursework strictly to solid state, low power electronics,” says robert J. Barker, Program Manager at the Air Force Office of Scientific research. “It’s noteworthy that thanks to edl, UNM has one of the very few ee cur-ricula in the U.S. in which students can learn to master high power electronics.”

e d l s c h a m i l o g l u

Further Reading

Fuks, M. I. and E. Schamiloglu. 2005. Rapid

Start of Oscillations in a Magnetron with a

Transparent Cathode. Physical Review Letters

95: 205101.

Mojahedi, M., Schamiloglu, E., Hegeler, F., and

K. J. Malloy. 2000. Time Domain Detection

of Superluminal Group Velocity Using Single

Microwave Pulses. Physical Review E 62:

5758-5766.

Abdallah, C. T., Soualian, V. S., and

E. Schamiloglu. 1998. Toward Smart Tubes

Using Iterative Learning Control. IEEE Transactions on Plasma Science 26 (3):

905-911.

Hegeler, F., Grabowski, C., and E. Schamiloglu.

1998. Electron Density Measurements During

Microwave Generation in a High Power

Backward Wave Oscillator. IEEE Transactions on Plasma Science 26(3): 275-281.

Agi, K., Moreland, L. D., Schamiloglu, E.,

Mojahedi, M., Malloy, K. J., and E. R. Brown.

1996. Photonic Crystals: A New Quasi-optical

Component for High Power Microwaves. IEEE Transactions on Plasma Science 24(3):

1067-1071.

Moreland, L. D., Schamiloglu, E., Lemke, R. W.,

Roitman, A. M., Korovin, S. D., and V. V. Rostov.

1996. Enhanced Frequency Agility of High

Power Relativistic Backward Wave Oscillators.

IEEE Transactions on Plasma Science 24(3):

852-858.

Benford, J., Swegle, J. and E. Schamiloglu, High-Power Microwaves, 2nd Ed. (Taylor

and Francis, Boca Raton, FL, 2007).

Barker, R. J. and E. Schamiloglu, High Power Microwave Sources and Technologies (IEEE

Press, Piscataway, NJ, 2001).

soe research by the numbers

30 Ph.D. degrees awarded

55 research and visiting faculty

108 Full-time faculty

106 M.S. degrees awarded

183 B.S. degrees awarded

30% Percentage of degrees awarded to underrepresented groups

20% Percentage of degrees awarded to women

601 Graduate student enrollment

1016 Undergraduate student enrollment

$27.74 M Total sponsored research expenditures

147,500 Square footage of the new Centennial engineering Center (CeC)

$42 M Approximate cost of the CeC

9/14/2008 Grand opening of the CeC

For 2007 Recent Ph.D. dissertation titles

Centennial Engineering Center

The new Centennial Engineering Center is designed

to increase the School’s capacity for innovation,

multidisciplinary research, developing patents,

creating new businesses, and economic growth.

Boian Alexandrov, ChNe “Branching Transport Model of Nal(TI)Alkalihalide Scintillator”

Serhat Altunc, eCe “Focal Waveform of a Prolate-Spheroidal Impulse Radiating Antenna (IRA)”

Andrey Andreev, eCe “Methods to Produce Short-Pulse, High-Power Microwaves"

ram Saran Attaluri, eCe “Growth and Optimization of Quantum Dots-in-a-Well Infrared Photodetectors”

Stephen Bayliss, ChNe “The Sagnac-Configured Fiber Optic Calorimeter: An Investigation of an Improved Non-Destructive Methodology for Determining the Thermal Power Output of Heat- Producing Nuclear Materials”

Gregory Chavez, Ce “On Fusing Linguistic and Assignment Uncertainty in Damage Assessment of Structures”

Xiaoyang Cheng, eCe “Minimally Invasive Capacitive Micromachined Ultrasonic Transducers Array for Biomedical Applications”

Travis Conant, ChNe “Pd-Zn Bimetallic Catalysts for the Steam Reforming of Methanol”

Hugo enrique Farfan Torrez, Ce “Estimating Soil Water Evaporation Using Nonlinear Inverse Theory”

Frederick Gleicher, ChNe “The Subelement Sweeping Method for Radiation Transport Modeling on Polygonal Meshes”

Matthew Higgins, eCe “Models for Electromagnetic Coupling of Lightning onto Multiconductor Cables in Underground Cavities”

Shenghong Huang, eCe “Microscopy Study of Extreme Lattice Mismatched Heteroepitaxy Using Interfacial Misfit Arrays”

Scott Lovald, Me “A Computational Framework for Solving Biomechanics Problems”

Jeremy Martin, eCe “Precision Electron Flow Measurements in a Disk Transmission Line”

robert Mcentire, Me “Atomistic Analyses of Plastic Deformation in FCC Metal Nanostructures”

Brian Nease, ChNe “Time Series Analysis of Monte Carlo Neutron Transport Calculations”

Frederick Newman, eCe “Study of Techniques to Obtain Metamorphic Low Bandgap Devices on GaAs Substrates”

Carrie Noren, Me “Quantitative Mixing Measurements of a Supersonic Injection Coil Nozzle with Trip Jets”

Jose Otazo Torres, eCe “Advanced Parallel Magnetic Resonance Imagining Methods with Applications to MR Spectroscopic Imaging”

Jack Parker, ChNe “Pool Boiling of Dielectric Liquids on Porous Graphite and Extended Copper Surfaces”

Jorge e. Parra, eCe “A Reconfigurable Multiprocessor Architecture for Space Missions: The AFRL-UNM HERC”

elena Plis, eCe “Mid-IR Type-II InAs/GaSb Nanoscale Superlattice Sensors”

Thomas Quirk, ChNe “The Development of a Continuous Energy Adjoint Transport Model”

Shailendra rathod, ChNe “Understanding the Synthesis of Mesoporous Silica Particles by Evaporation Induced Self Assembly”

Nicolas Sau, Ce “Peridynamic Modeling of Quasibrittle Structures”

Prasanna Sridhar, eCe “Hierarchical Aggregation and Intelligent Monitoring and Control in Fault-Tolerant Wireless Sensor Networks”

Peng Sun, eCe “Mathematical Theory of Modern Avalanche Photodiodes and Its Application to Ultrafast Communications”

elise Switzer, ChNe “Nanostructured Electrocatalysts for Fuel Cell Based on Aerosol Templating Synthesis Techniques”

David Tipton, Me “Coupled Fluid-Solid Interactions Under Shock Wave Loading”

Guillermo A. Vera, eCe “A Dynamic Arithmetic Architecture: Precision, Power and Performance Considerations”

Xuefei Wang, eCe “Hybrid Neuro-Fuzzy Inference Models for Outcome Prediction in Acute Leukemia Using Gene Expression and Covariate Data”

David Wood, ChNe “Fundamental Material Degradation Studies During Long-Term Operation of Hydrogen/Air PEMFCS”

Zhen Yuan, ChNe “Biomolecule Derived Nanostructured Arrays”

Shu Zhou, eCe “Improvement of Surveillance Detection with the Aid of a Terrain-Aware Mobile Sensor Network”

School of Engineering

MSC01-1140

1 University of New Mexico

Albuquerque, NM 87131-0001