Developing Tools for Nanolithography
Developing Tools for Nanolithography
"When it comes to making faster, smaller computers, the microelectronics industry is reaching a limit," says Joseph Lyding, NCSA principal investigator and UIUC professor of electrical and computer engineering. "Our goal is to use the Scanning Tunneling Electron Microscope (STM) for research-- particularly in the area of circuit miniaturization."
In 1984 Lyding came to the UIUC to study the dynamics of charge-density wave (CDW) transport in quasi-1D metals, under the tutelage of electrical and computer engineering Professor John Tucker and legendary two-time Nobelist in physics John Bardeen. While using STM to observe CDWs, Lyding became fascinated with the technology's endless experimental possibilities.
"John Bardeen encouraged my pursuit of STM," says Lyding. "He was very interested in probing the microscopic aspects of CDW phenomena, and he also encouraged the semiconductor STM work that we are now pursuing."
Key to Lyding is the integrationof the STM experiment into NCSA's metacomputer environment. Since Lyding's STM facility is located in the Beckman Institute, which is intended to intermingle various scientific disciplines and to encourage collaboration, it was inevitable that he would learn of NCSA and its biological imaging group headed by Clint Potter.
"A typical scenario for us is to use AVS [software] in conjunction with NCSA's CONVEX C3880 and CM-5, and our laboratory's HP 700 [Hewlett Packard] and PCs simultaneously. We typically use Rachael's [NCSA research programmer Rachael Brady of the biological imaging group] STM module running on the CONVEX to control the STM, acquire image data, and perform Viewit operations. All of this occurs within a remote module from within AVS, which is running on the HP in our laboratory. AVS on the HP is used to render high-quality light-sourced images of the STM scan area, bringing out details that we cannot see in the raw data.
"Furthermore, we frequently process these images by taking 2D fast Fourier transforms [FFT] and looking at the power spectra for symmetry features in the image. The FFT module in our HP-AVS display actually sends the data to the CM-5 for parallel computation of the power spectrum which is then displayed on the HP.
"All of this interactivity between machines is transparent to the user, thus my students and I can concentrate on the experiment and use the advanced rendering and analysis to steer the course of the experiment."
An SGI workstation in a Showcase display booth was connected to NCSA's CONVEX C3880 in Urbana via the T3 fiber optic link. "People were genuinely impressed when they found out we were controlling a live experiment from Urbana in Chicago and could remotely move the STM scan area around in real time," says Lyding. "The distance had no effect on response time."
Providing Lyding with remote capabilities was a challenge for Brady, who explains how she forced a solution: "All of Joe Lyding's microscopes are controlled by IBM PCs running MS/DOS on either 386 or 486 chips. These are single tasking machines, which means you cannot run something in the background while you do something in the foreground. To hook his computers up to another box, those PCs are going to have to control the experiment and at the same time listen on the network for any commands coming from a remote site.
"Making his machines do two things at once is a total hack. You tie all the network listening stuff to interrupts, which are built into the MS/DOS operating system. When a request comes from the network, it gets shoved into a buffer on the PC and a flag is set within the PC's hardware.... Right now, this is all done synchronously. Next, we would like to do it asynchronously--that is, interrupt Joe's program and say--here's a new request.'"
UIUC members of the research group include: Ilesanmi Adesida, Stephen Bishop, K.-Y. (Norman) Cheng, Karl Hess, and John Tucker, Department of Electrical and Computer Engineering, and Munir Nayfeh, Physics. Members at Minnesota are Stephen Campbell and Ted Higman, Department of Electrical Engineering.
"We are finding that these tiny new devices will not function like today's larger-scale ones. Quantum mechanical effects begin to play a dominant role, so we are exploring ways to harness those effects for our advantage."
Since STMs operate sequentially, a large array of them will be necessary to map circuits with enough speed to be cost effective. Lyding's STM images are currently rendered from 2D datasets, but they will become 3D as soon as he begins to use electron energy as the third dimension.
Lyding says he will need teraflops computing speed to bring detailed simulation of the tunneling probe tip and its interaction with the sample online with the experiment. Understanding the probe tip is especially important in the STM-based Nanolithography University Research Initiative where atomic-scale surface modification schemes are being developed. Voltages required to modify a sample could also damage the tip.
Other collaborations underway with Karl Hess, director of the NSF-funded National Center for Computational Electronics (NCCE), will also require more compute power. "Karl Hess's group has developed the analysis tools to perform a full quantum mechanical molecular dynamics simulation of the tip and surface atoms in response to the fields and forces that exist in the tunneling junction," says Lyding. "Tremendous compute power is needed to extend this calculation to a volume consisting of, say, 50 atoms that might realistically bound the volume of tip/surface interaction. This effort will dovetail smoothly with the existing compute/experiment protocols developed for us by NCSA.
"Hess has proposed a number of intriguing new electronic devices that will only operate in the size regime where quantum effects emerge and dominate," Lyding continues. "Using conventional lithographic technology, these devices will only work at liquid helium temperature (4.2 K) or lower. However, with STM nanolithography, room temperature operation should be possible. Hess is working in close collaboration with Tucker on developing and simulating new device concepts that are amenable to STM nanolithography."
In September, Brady got an opportunity to outreach to other researchers when she copresented with Lyding at the Workshop on Real-Time Applications of High Performance Computing for Biological Imaging that NCSA cosponsored with the Beckman Institute with NSF funding. Their topic was entitled "The Distributed Scanning Tunneling Microscopy Laboratory: Real-time Control, Visualization, and Modification at the Atomic Level." Potter and Bridget Carragher, director of the Optical Visualization Facility at the Beckman Institute, were co-chairs.
"The great thing about working with Joe is that he appreciates computers," says Brady. "He understands what they can do."
