Carbon Nanotube-Based Devices
This project explores the development and capabilities of carbon nanotube (CNT)-based nanoelectromechanical systems (NEMS). Previously, we focused on nanofabrication challenges and pervasive failure modes which currently preclude widespread realization of large-scale arrays of high performance CNT-based NEMS. We plan to build upon this understanding and investigate the ultimate electro-mechanical performance of CNT devices by determining what frequencies they can attain under laser-induced loading schemes. Additionally, we plan to develop methods to establish robust protocols for fabricating large-scale arrays of logic gates.
Ultimately, our goal is to create a metric for design of robust, high-performance CNT-based devices which can be feasibly fabricated in large numbers.
We use a CNT-based switch with closed-loop feedback control (Figure 1), developed by our group, as a platform for our investigations. When the applied voltage U exceeds a critical value (called the pull-in voltage), the CNT accelerates toward the bottom electrode, thus closing the switch. This dynamic pull-in event and the subsequent CNT-electrode impact can result in a number of different failure modes which are common to this class of electrostatically-actuated NEMS. These include CNT fracture, ablation due to rapid charge dissipation, and irreversible stiction between the CNT and electrode. A primary goal in this study is to understand the underlying mechanisms of failure and establish a metric for the design of robust CNT-based NEMS.
By creating arrays of devices with incrementally-varying geometry, we perform parametric studies to identify the various modes of failure and their point of onset within the design space. Multiphysics models then enable us to investigate the underlying mechanisms for the experimentally observed failure modes.
Device Fabrication and Characterization To fabricate the devices, we combine standard microfabrication processes with novel nanopatterning techniques. We leverage the tools of our Probe-Based Nanomanufacturing efforts to pattern catalyst for subsequent growth of CNTs on pre-fabricated electrodes. CNTs are then grown in place on the devices. Here the parallel sub-100-nanometer patterning capabilities enable us to create well-ordered arrays of CNTs. Arrays of devices are constructed in this manner with incrementally-varying geometry. We then characterize the performance and failure modes of these devices in situ the scanning electron microscope (Figure 2). This enables direct visualization of the operation of the device and its failure modes.
Dynamic Modeling To analyze the dynamic performance of the CNT-based devices and to explain the experimental measurements, we employ both analytical and finite element methods. By solving an analytical model with finite difference methods, the dynamic response of the CNT cantilever can be estimated. We also employ multiphysics finite element methods on the entire device to obtain detailed information about stress distribution and wave propagation, and transients in electrical and temperature signatures. Figure 3 gives an example of the evolution of the speed distribution along the NEMS device during the pull-in event. More details are available here.
Failure Mode Identification We explored the failure modes of carbon nanotubes under cyclic voltage-induced loading. Initial tests with gold-coated electrodes demonstrated that CNT switches could fail due to irreversible stiction, incremental shortening (as a result of burning), or a combination of the two. However, subsequent tests with diamond-like carbon (ta-C) electrodes showed a much lower incidence of failure (Figure 4). These results indicate that use of ta-C electrodes significantly broadens the design space for future CNT-based devices.
Next Steps In the future, we aim to investigate the resonant frequencies that CNT-based devices can attain under laser-induced loading. We will also develop and refine methods of fabricating large-scale arrays of these devices for use as logic gates. Given these objectives, understanding the correlations between materials selection, design, and performance will be essential. Our broad expertise in device fabrication, dynamic modeling, and failure characterization will prove invaluable as we pursue new applications of CNT-based devices.