Tech Today Fall07

Embed Size (px)

Citation preview

  • 7/27/2019 Tech Today Fall07

    1/5

    10

    ENGINEERING

    NE

    W

    S

    T E C H T O D A Y

    Put an ice cube on your kitchen counter and it will melt. But how does it

    melt? What is happening at the microscopic level? When it is melting,

    is it liquid or solid or something in between? Figuring out how atomsbehave in a chunk of ice may not seem the makings of high science, but it is. Especially if

    you think about ice and basically all materials in a broader, fundamental way: What hap-

    pens to a materials structure as it changes its form? And what conditions bring about these

    changes? These are the same principles that govern everything from catalyst reactivity to

    superconductivity.

    Fundamental questions intrigue Mohammad Islam, an engineer who turns to science

    for inspiration. This year, the young researchers penchant for wanting to understand how

    things work earned him both an NSF CAREER Award and an Alfred P. Sloan Award.

    Islam, a professor in chemical engineering and materials science and engineering,

    explains that when alloying elements are added to iron to make steel, making it very strong,

    changes have occurred to the materials structure at the sub-micron level.

    There must be reasons why the structure changes and gives rise to new properties

    that we want. Wouldnt it be great if we could understand this phenomena and tailor it?

    asks Islam. To this end, he and his research group synthesize colloids, which can be thoughtof as micron-sized atoms, and use them as building blocks to make structures generally

    formed by atoms. By working with colloids instead of fast-moving Angstrom-sized atoms, he

    can view the colloids behavior within a three-dimensional material in real time with an optical

    microscope. These experiments are not possible in structures formed by atoms because

    visualizing the atoms would be extremely difficult.

    From Nanomaterials to Macro-Applications

    Presently, Islam is working with carbon nanotubes, which

    are a micron in length and a nanometer in diameter. They

    are rigid and strong, much stronger than steel. They con-

    duct heat better than diamond, yet their density is close

    to that of air. Impressed with these attributes, Islam has

    set out to make macroscopic composite materials that

    incorporate the same properties of tiny carbon nanotubes.

    Then the question becomes: Can we put carbon nano-

    tubes into polymers and make the composite material

    very strong with high heat conductivity but maintain the

    polymeric malleability? he said. Polymers are generally

    nonconducting and lack strength.

    The applications go on: If carbon nanotubes can

    increase the conductivity of polymers, can they dissipate

    heat from electronic chips? This would allow a more

    compact circuitry with higher performance for smaller,

    more powerful computers. Also, if Islam and his team can

    quickly and cheaply produce exceptionally strong and light

    THE NSFRECOGNIZESGREAT TALENT

    THE BEST OF

    BOTH WORLDS

    Mohammad Islam and his research

    group synthesize colloids and use them

    as building blocks to make structures

    generally formed by atoms.

    This year, three CIT faculty membersFred Higgs, Mohammad Islam, and Ken Maiearnedthe National Science Foundations Faculty Early Career Development (CAREER) Award. These

    important awards, along with sizable funding, are given to young professors who are doing an

    exemplary job integrating education and research. We have no doubt that these individuals

    will make significant contributions to engineering and inspire students to follow their leads.

    B Y S H E R R Y S T O K E S

  • 7/27/2019 Tech Today Fall07

    2/5

    Whether you are fabricating arti-

    ficial hip joints or computer chips,

    you have particles in between rub-

    bing surfaces that cause wear. We

    are developing computer models to

    predict this general problem, and

    were doing state-of-the-art experi-

    ments to validate the models.

    carbon nanotube composites, they could be used in aircrafts or cars. He says, If we can

    make a car that weighs a hundred pounds and its body is strong like steel, wouldnt that be

    better for the environment?

    These materials can be used for biological applications, too. He explains that carbon

    nanotubes can be coated with biological polymers and used to deliver drugs or genes inside

    human cells. So, then the question becomes, what is the effect of having this foreign

    material inside your cell? Is it toxic? Can you use these materials for drug delivery without

    having a negative impact on your health? These are some of the questions we are asking,

    and we are doing experiments that explore this, concludes Islam.

    Tribology. You may not know

    the word, but engineers are

    certainly familiar with the

    concepts that comprise this highly specialized area of

    study: friction, lubrication, and wear.Many engineers do not understand what tribol-

    ogy is because it is such a broad and diffuse topic. It

    usually shows up as friction-related problems. Or with

    mechanical engineering and materials science devices,

    the great showstopper is adhesion, where surfaces

    stick together, and thats a tribology problem, too,

    says C. Fred Higgs III.

    Joining Carnegie Mellon in 2003, Higgs is

    the first professor to teach tribology in mechanical

    engineering, and he is earning impressive accolades,

    including the NSFs Early Career Development Award.

    For the next five years, Higgs will receive a

    total of $400,000 from the NSF to develop computer

    models that will predict how surfaces will wear whenthey are under a load and rub together. Complicating

    the matter is the fact that debris or foreign particles

    will affect wear, too. Higgs says that many industries,

    ranging from data storage to biotechnology, deal with

    friction and surface-wear problems caused by abrasive

    nanoparticles that are sandwiched between rubbing

    surfaces.

    Illustrating his point, Higgs explains how artificial hips deteriorate over time. In the

    human hip, the femur (the top of the leg bone) moves around in the cup-like acetabulum,

    forming the hip joint. In the hip joint, a thick liquid called synovial fluid keeps the femur

    lubricated, reducing friction and easing movement.

    But in an artificial hip, we are not able to create the same conditions. While there

    are some fluids in the joint, they are unable to give complete separation of the surfaces.

    Because of that contact, you get wear on the artificial hip, and nanoparticles begin toaccelerate the wear. After 10 or 15 years, the hip needs replaced again, says Higgs.

    Whether you are fabricating artificial hip joints or computer chips, you have particles in

    between rubbing surfaces that cause wear. We are developing computer models to predict

    this general problem, and were doing state-of-the-art experiments to validate the models,

    he says. For example, Higgs developed a sophisticated algorithm on particle augmented

    mixed lubrication (PAML), which can predict wear in hip joints and in fabricating computer

    chips. Consequently, Carnegie Mellon, along with Higgs and his recently graduated Ph.D.

    student, Elon Terrell, filed for a patent on the PAML algorithm, which is the engine behind

    these computer models. The algorithm enables Higgs to develop what he calls in silico

    modeling simulations, where the actual engineering process, such as the polishing of

    computer chips or the wear of artificial hip joints, is simulated on a computer without

    TRIBOLOGY GAINS

    FOOTHOLD IN CIT

  • 7/27/2019 Tech Today Fall07

    3/5

    12

    ENGINEERING

    NE

    W

    S

    omitting the complex physics involved. By doing in silicomodeling (a termed coined by Higgs

    colleague at the University of Florida), he can run computer experiments that very closely

    mimic the actual physical experiments that his group is conducting in the lab to validate their

    models. Some of the companies interested in testing the PAML model on their devices are

    Hitachi, Seagate, and the Data Storage Institute in Singapore.

    The work that Higgs and his students are involved in will affect a variety of technolo-

    gies, including integrated circuits and data storage nanotechnology, coal flow energy systems,

    dental tribology, and, of course, total joint replacements.

    If you look at the way computing machines are built, they fall

    into two broad groups: hardwired systems that have fixed

    hardware functionality and reconfigurable systems that enable

    people to define what their systems will do at the hardware level.

    You really havent seen very much of the reconfigurable world creeping into the hard-

    wired microprocessor world, says

    Ken Mai, a professor in electrical and

    computer engineering. He explains

    that reconfigurable logic has lower

    performance and efficiency than logic

    hardwired for the same function.

    But, then again, reconfigurability has

    its prosuser-defined functionality,

    fast time to market, and the ability to

    fix bugs and upgrade hardware if

    necessary.

    Early in his career, Mai began

    exploring the notion of transferring

    some aspects of reconfigurability onto

    the hardwired side, and he focused

    his attention on the memory system.

    The memory system is fairly ame-

    nable to adding configurability to it be-

    cause there arent that many different

    things you want to do with memory,

    says Mai. He believes that making the

    memory reconfigurable will have small

    impact on performance and that users

    could gain a lot in performance and

    efficiency.

    This year, Mai received an NSF

    Early Career Development Award,

    along with $400,000, to explore

    options for adding reconfigurability to memory systems at various levels of their design,

    including the circuit, microarchitecture, and architecture levels.

    One of the more compelling reasons for enhancing the memory system, says Mai, is

    that if you look at the way computers are designed, we are at an inflection pointwe are not

    really sure what we need to do next.

    He explains that in terms of microprocessor design we have hit two walls. One prob-

    lem that limits how applications perform is that we cant pull data from the memory system as

    quickly as we run calculations. The memory has not scaled the same way as microprocessor

    cores have, says Mai.The second wall deals with processor speed. For a number of years, people have been

    increasing the clock frequency or the processor rate, says Mai. Today, processing speed

    isnt growing as it once had. If you look now, companies arent trying to sell their processors

    based on clock frequency, he says. At this point, trying to increase performance by scaling up

    the clock rate unacceptably increases the power. Eventually we get to the point where the

    microprocessors can no longer be cooled sufficiently in a normal desktop system. With these

    fundamental constraints identified, the question becomes, how do we build computing sys-

    tems that are robust, easy to operate, reliable, and economically feasible? This question fuels

    a large portion of Mais research, and his NSF-funded project will reveal the role reconfigurable

    memory systems may play in advancing microprocessor design.

    A TALE OF POWER AND

    PERFORMANCE

    Ken Mai

    You really havent seen very much of

    the reconfigurable world creeping into

    the hardwired microprocessor world.

  • 7/27/2019 Tech Today Fall07

    4/5

  • 7/27/2019 Tech Today Fall07

    5/5