Volume 6, Issue 3
Daryl Chrzan is a member of the Electronic Materials Program at Lawrence Berkeley National Laboratory. |
It can take three decades before a new alloy makes its way from a glimmer in a scientist's mind to,say, the body of an airplane. That's because the development of alloys requires years of experiments to characterize the materials' mechanical properties. But what if you could model those characteristics in a computer? UC Berkeley engineer Daryl Chrzan is doing just that. He uses computational materials science to predict the properties of materials from the bottom up. His research could impact fields as diverse as nanotechnology and aeronautics.
"The intent of our work is to start with the properties of atoms and predict the larger scale properties that we experience everyday," says Chrzan, a professor in the Department of Materials Science and Engineering.
The outer radius of a telescoping multi-walled carbon nanotube piston, seen in this transmission electron microscope image, is only 12.6 nanometers at its largest point, nearly 10,000 times thinner than a human hair. (courtesy Lawrence Berkeley National Laboratory) |
Already, researchers use computational tools to gain insight into the optical and electronic properties of certain materials. Predicting the mechanical properties, how a piece of metal will bend, for example, is a much harder problem though. That's because the number of degrees of freedom of a typical solid, how many ways the atoms can move, is "enormous," Chrzan says. A typical cubic centimeter of a metal contains 10 to the 22nd (10 followed by 22 zeros) atoms. Storing even the initial conditions of those atoms would require more computer memory than exists in the world, he explains. The difficulty is compounded by the fact that the structure of a material is not symmetric--defects are what allow it to bend in the first place.
However, using supercomputers at Lawrence Berkeley National Laboratory's National Energy Research Scientific Computing Center (NERSC), Chrzan and his colleagues can run day-long calculations on the behavior of 400 or so atoms. Then they can begin to scale up those measurements and make general predictions about the material.
"We're trying to start at the nanoscale and move up," Chrzan says.
Indeed, Chrzan and graduate student Elif Ertekin have recently focused on carbon nanotubes, sheets of carbon atoms that resemble rolls of chicken wire fused at the seam. Many researchers, including UC Berkeley physicist Alex Zettl, are studying carbon nanotubes as key building blocks for nanodevices. For example, Zettl has demonstrated pistons and conveyor belts fashioned from nanotubes. Several years ago, Zettl built an electric nanomotor that's 300 times smaller than the diameter of a human hair.
A series of scanning electron microscope pictures of the spinning rotor of a nanomotor. The entire electric motor is about 500 nanometers across, 300 times smaller than the diameter of a human hair. (courtesy Zettl Research Group, Lawrence Berkeley National Laboratory and University of California at Berkeley) |
"When I saw that motor, the first thing I wanted to know is how much torque you could get out of it," Chrzan says. "if the drive shaft is going to be a carbon nanotube, you need to determine how much torque a carbon nanotube can withstand."
Computational limitations restricted them to running the atomic-scale calculations on nanotubes of just 36 or so atoms around. But thanks to several formulas they've devised, they're now able to predict the torsional stiffness of "every carbon nanotube on the planet" regardless of its size.
"You can now model the rotation behavior of nanotubes that are larger than what's possible to compute directly using atomic scale calculations," Chrzan says. "But that's just the elastic response. It doesn't tell us how the tubes will deform plastically."
The mechanical properties of materials, including nanotubes, are not entirely dictated by the average properties of the material but rather the defects within it, he explains. In the case of nanotubes, these defects are tiny particles that move around on the rolled up sheet. Those defects, called dislocations, lead to plastic deformation of the material.
"Again, we're developing a description of the energetics of the dislocations so we can go beyond the atomic scale in describing the properties of these defects," Chrzan says. "By doing that, we can explore the material's larger scale behaviors."
As Chrzan's techniques are proven out, they may become useful tools for nanoengineers selecting materials and designing devices on the smallest scale. Someday, they could even speed up the invention of brand new macroscale materials like aerospace alloys for high-performance applications.
"Making high quality alloys is something of a black art," Chrzan says. "The hope is computational materials science can help us speed up the process. We can now start to unravel mysteries at the atomic scale at just astounding levels of detail."
Ionel Dragos Hau research interests also include medical applications for nuclear energy. |
When the dentist is preparing to x-ray your mouth, he or she usually drapes your torso in a lead blanket to shield the rest of your body from the radiation. The metal is an excellent radiation shield. Some fear that terrorists might take advantage of that same phenomenon by using lead containers to smuggle radioactive materials for weapons. However, lead is no match for the new radiation detector built by UC Berkeley nuclear engineering graduate student Ionel Dragos Hau and his colleagues. Along with nonproliferation applications, their technology will likely have broader uses in nuclear science, astrophysics, and materials science.
"We're trying to design an instrument that identifies hidden materials that emit radiation, but also characterizes known materials for scientific study," says Hau, who works in Lawrence Livermore National Laboratory's (LLNL) Advanced Detector Group as part of the University of California's Student Employee Graduate Research Fellowship Program (SEGRF). His faculty adviser is Jasmina L. Vujic, chair of Berkeley's Department of Nuclear Engineering, while the detector project's lead scientist at LLNL is Stephan Friedrich.
A photo of the ultrahigh-resolution neutron spectrometer with the various key components labeled. (courtesy LLNL) |
Radiation detectors are at the core of many well-known technologies and scientific fields, from medical imaging machines to surveys of deep space. For example, gamma-ray detectors on satellites have enabled UC Berkeley physicists to study huge explosions on the other side of our galaxy. Meanwhile, physicians use similar technology for early diagnosis and improved treatment of cancer. In recent years, researchers at LLNL and elsewhere have developed gamma-ray imaging spectrometers as part of airport and ocean port security systems to detect the transport of nuclear materials.
"The problem is that lead around a gamma ray source shields much of the radiation," Hau says. "Neutrons aren't absorbed by lead as easily though. So if we can detect neutrons with a very high energy resolution, then we can determine the nature of that neutron source."
In fact, the researchers' instrument, called an ultrahigh-resolution neutron spectrometer, is sensitive enough to detect a single neutron. And not only can it detect whether a sample is radioactive but it determine the composition of the nuclear material.
The Berkeley instrument doesn't look directly for neutrons, but rather measures how the neutrons heat up another material within the machine. The detector on the instrument is installed at the end of a "cold finger" containing a lithium fluoride crystal. Pointed at a radioactive source, the lithium fluoride absorbs the neutrons. That induces a nuclear reaction that produces excess energy released as heat. The temperature increase can be below one millikelvin (less than 1/500 of a degree Fahrenheit).
Detecting such a minute change demands an incredibly sensitive thermometer that operates at temperatures close to absolute zero. The LLNL researchers fabricated such a thermometer from molybdenum and copper that, thanks to an effect called superconductivity, changes its resistance very rapidly with a tiny shift in temperature. As a result, the temperature pulses are converted into electrical pulses that can then be digitized for computer analysis.
According to Hau, the instrument is likely to be ten times more precise than current detectors that use semiconductors or pressurized gas. They're also more compact than neutron spectrometers that conduct measurements based on the time it takes for the particles to travel over a fixed distance.
With improvements in speed and other factors, Hau believes the system could be ready for the field in a few years.
"This unprecedented resolution could lead to a new generation of instrumentation," he says. "However, its field applications will depend on how small and portable the instrument can be made without sacrificing too much energy resolution.
Professor Sanjit Seshia joined UC Berkeley's Department of Electrical Engineering and Computer Sciences last year after receiving his PhD in Computer Science from Carnegie Mellon University. |
One of the biggest challenges of computer security is that the people who write viruses are smarter than the software used to detect the malicious code. In fact, "malware detectors," like virus-scanning software, aren't very intelligent at all. They simply look at whether the pattern in a particular piece of code, an email attachment for example, matches the signature of a known virus. This isn't a logical approach, says UC Berkeley computer scientist Sanjit Seshia. He means that literally. Seshia and his colleagues are using computational logic to detect the behavioral traits of viruses even if their maliciousness is well-hidden by their creators.
"My research is at the intersection of computational logic and dependable computing," says Seshia, who is affiliated with the Center for Information Technology Research in the Interest of Society (CITRIS). "How do we make computer systems much more reliable, secure available, and safe?"
Most computer users count on antivirus software to rid them of viruses and worms that are transmitted via email attachments or dodgy Internet downloads. At the heart of these software packages are systems that attempt to identify malware through a collection of rules and "signatures," code patterns that act as identifiers of new viruses. It's called a syntactic approach—the software looks for particular sequences, or syntax, of bits. As new viruses are identified, the makers of the software offer updates containing the signatures of the latest threats.
"It turns out that current virus scanners are quite easy to fool," says Seshia, who participates in UC Berkeley's Team for Research in Ubiquitous Secure Technology (TRUST) and the Center for Hybrid and Embedded Software Systems (CHESS). "Malware writers obfuscate their code so that a signature that works today won't work tomorrow for what's essentially the same virus. So anti-virus companies are always trying to track the latest variants of viruses and worms and making sure people download the latest signatures to keep their definitions of viruses up to date."
Rather than look for commonalities in syntax, Seshia and his collaborators are developing mathematical algorithms based on semantics, essentially the "meaning" of the code. Thanks to some semantic smarts, the algorithm can detect variations of malware even if the nasty bit of code is well hidden by those who wrote the virus. He's collaborating on the research with professor Somesh Jha and graduate student Mihai Christodorescu of the University of Wisconsin-Madison, and Carnegie Mellon University professors Randal Bryant and Dawn Song.
"We're trying to develop a more behavioral definition of what it means to be malicious," Seshia says. "Perhaps it deletes files on your hard drive or duplicates itself and emails copies to everyone in your address book."
In computer science, this kind of semantic check is an "undecidable problem." There's no algorithm that can solve it every single time. The beauty of the algorithm though is that it's good enough in most cases, and it can crank through the calculations in a reasonable amount of time.
First, the researchers defined the behavioral signature of a virus, the fragment of the program's instruction set that's indicative of malicious behavior. This is used to form a template, Seshia says, "that's general enough to represent a whole family of programs."
"The algorithm translates the malware detection problem into a set of problems expressed in mathematical logic, which are handed off to another algorithm called a 'decision procedure,' says Seshia, whose major contributions were in the design and implementation of this procedure.
The overall algorithm involves a comparison of the unknown set of instruction sequences to the template. The algorithm abstracts away all superfluous information, including any variables that may have been altered to hide the maliciousness. The bare-bones program is then checked against the template to see if any red flags are raised. In recent experiments, the approach correctly identified multiple variants of two worms using a single template.
"It may be a computationally-hard problem to solve, but in practice, a good encoding of the problem enables the algorithm to run reasonably fast so we can think of eventually putting it into commercial tools," Seshia says.
While the computational logic methods are well-suited to fight the battle against computer viruses, Seshia is also applying his methods to hardware. He's developing tools for computer aided design of integrated circuits that will enable engineers to better verify their designs long before the expensive fabrication process begins.
"All of this research is about checking that a computer system will work as expected," he says. "If a piece of code is malicious, you'd like to know that before you run it."
Cool Alums: The most brilliant and practical design idea this side of indoor plumbing
by Rachel Shafer
The Hands-Off Toilet team poses with its design, which won second place in the student design competition at the National Technical and Career Conference. Members are, from left, Alavaro Frausto, William Tovar, Eustaquio Carrillo, Jorge Carapia, and Herman Bravo. (Photo provided by HES) |
Ladies, how many times have you gone to use the bathroom and the toilet seat is left in the unconscionable upright position by your boyfriend/brother/housemate/fill-in-the-blank? Gents, do you bristle every time you're asked/begged/nagged/fill-in-the-blank to put the toilet seat down? For years, this problem has left men and women flush with anger. No longer. A team of ME seniors (all men, most now alumni) promise peace with their new design called the Hands-Off Toilet, a bathroom system that automates the raising and lowering of the toilet seat and flushing process.
"We did it so the women would stop complaining," explains ME senior Eustaquio Alfonso Carrillo, chuckling.
The story began last fall when Carrillo, a member of Cal's Hispanic Engineers and Scientists (HES) club, wanted to enter the student design competition at the National Technical and Career Conference held in early January by the Society of Hispanic Professional Engineers. Carrillo knew he could find a competitive team. He approached friends who had worked together on a preliminary design of the bathroom system for ME 110. (Carrillo's friend, ME senior Herman Bravo, had a girlfriend, and it was from the personal experience of their relationship that the Hands-Off Toilet idea was born.) They agreed to form a team.
But taking a design from paper to competition-ready prototype in just one semester is another matter. First, according to the competition's rules, a design produced in a class must be further enhanced. So the team took surveys of fellow students who suggested adding automatic flushing. After hours and hours of calculations and the frustrating work of machining and testing two prototypes, the team came up with a two-pedal/lifting-bar and tension cord system. Step on the first pedal, the seat automatically raises. Step on the second pedal and it flushes the toilet and lowers the seat.
The team didn't have enough money to ship a real toilet to the competition so they built a plastic model. "When we got there, we discovered the box we had shipped it in was water damaged and the main tank had cracked," says Carrillo, laughing now. "Luckily, we were able to take it apart and salvage it. We used lots and lots of tape." The team worked until 6 a.m. on the day of the competition to get the system and model in working order. Then, instead of going to sleep, they practiced their presentation, which included a marketing plan and video designed by ME senior Alvaro Frausto which illustrated the system's exact functionality. When the team introduced its design, a female judge stood up and applauded them. "About time someone did that," she told the group.
At the closing banquet, the team found out it won second place. "Everyone was screaming with excitement," says Carrillo. "It was really fun."
Today, everyone except Carrillo has graduated and is out working, so there are no future plans for the design. But for the sake of personal relationships everywhere, Engineering News thinks an idea like this shouldn't disappear down the drain.
For more information about HES, go to www.ocf.berkeley.edu/~hes/.