Electrical, Electromechanical, and Optoelectronic Response of Nanodevices

We are primarily a Theory & Modeling group who work on a variety of problems in nano devices and materials. The focus of our work centers around developing an understanding of the device physics of both semiconductor and bio nanstructres. We use a variety of tools to model these structures. To find the atomic coordinates we use a variety of semiclassical, Monte Carlo and quantum molecular dynamics method. To study the electronic properties, we use both density functional theory and tight binding methods. To study the electrical transport properties of nanostructures,we develop both algorithms and code based on Green’s function methods.

We work on both fundamental and applied problems. Our current directions of focus are:

  1. DNA Nanostructures:

Figure: DNA consists of two single strands in a double helix. The distance between the bases is ~ 3.4 Å. Biomolecules (DNA, peptides and RNA) offer a platform for electronic devices. Electrical methods for sequencing and disease detection, with the biomolecule as channel for charge transport have recently been demonstrated.

Deoxyribo Nucleic Acids (DNA), the building block of life, is a nanoscale material that can be engineered precisely. Its immense potential arises from the ability to form arbitrary sequences with high accuracy and create complex 1D, 2D and 3D structures. From the viewpoint of material science and engineering, the electronic properties of DNA are important for (i) engineering sequences with device properties and (ii) all electrical methods for disease detection. The main challenges to move the field forward are the inter-related needs to develop a robust design methodology that can yield sequences with specific electrical properties and gain the ability to rapidly predict the conductance of a given sequence. This is a huge challenge which needs multiple iterations between computational and experimental efforts because (a) the environment cannot be precisely controlled, (b) of the sheer large number of possible sequences even for a strand that is only ten bases long and (c) there are currently not enough controlled experiments with a variety of base sequences to provide inputs or feedback to design. Along with our experimental collaborators, we are pursuing a multi-pronged approach combining theory & modeling, the efficient use of data, chemical synthesis, and transport experiments to effectively design electrical properties of DNA.

At the mesoscopic level, DNA consists of four distinct building blocks, Guanine (G), Adenine (A), Thymine (T), and Cytosine (C).  The most common form of DNA is quasi one-dimensional, and consists of two complementary single strands in a double helix. Each strand consists of the four building blocks linked together by a sugar-phosphate backbone. The strands are held together by hydrogen bonding between bases on complementary strands. When isolated, the four building blocks have distinct energy levels, highest occupied (HOMO) and lowest unoccupied (LUMO) molecular orbitals, electron affinities, and ionization potentials, all of which are important factors in determining the electronic properties. Synthetic chemistry allows these building blocks to be further expanded by using artificial bases, where the mentioned energies can be designed. The rich and interesting electronic properties of DNA emerge from the interaction between bases, both along and between strands, and with the backbone and environment.

The different energy levels and ionization potentials of bases provide a vehicle to engineer the bandstructure along a DNA strand, which gives rise to the possibility to create wells, barriers and superlattices that form the materials basis for electronic devices such as resonant tunneling diodes, heterostructure junctions, as well as more exotic species such as quantum-interference (QI)-based transistor platforms. Moreover, the recent emergence of 2D and 3D DNA-based nanostructures, which are engineered via techniques like DNA origami provides a platform that is capable of integrating unique DNA electronic structures into a device platform or even into higher dimensional device integration (3D). Achieving this goal has been difficult with conventional semiconductor technology, and any integration has been difficult at the molecular-scale. As such, DNA-based devices provide a unique opportunity to overcome several of the issues with both conventional electronics (bottom-up manufacturing and quantum transport) and nanoscale electronics (reproducibility and integration). As such, these devices may form the basis for developing widely deployable devices that exploit quantum transport effects, or may form a basis for a more conventional vehicle for electronics beyond the limits of lithography. One can think of quantum wells and barriers constructed from DNA as having transmission-resonances akin to double barrier resonant tunneling diodes and superlattices built from conventional semiconductor heterostructures. Additionally, these devices may be biocompatible, thus opening opportunities for utilizing the electronic properties of DNA for sensing, diagnostics, or health-care applications. Given this broad-range of potential applications, the primary goal of this proposal is to iteratively develop these building-blocks, through the intimate interaction between theory, experiment, and synthesis, so that large-scale systems and DNA-based applications can begin being actively designed, built, and utilized.

An additional very exciting field is the link between diseases and DNA sequences, mutations, and chemical modifications. The transport through DNA is inherently quantum mechanical in nature, and as such mutations in the bases cause a change in energy levels, which translates sequence information into transport properties. This fact has given rise to ideas centering on whether these mutations can be electrically detected. Of great interest are recent experiments that show the potential to detect diseases and sequence DNA by measuring electrical conductance of a single DNA molecule. These studies demonstrate the potential for an all-electrical method for disease detection, a technology that has promise to be portable and inexpensive. Even more important is that it can be used in situations where only a small number of mutated DNA strands are available; situations where conventional amplification methods involving polymerase chain reaction are not suitable. Therefore, the development of a predictive capability for DNA transport properties based on the building blocks will enable the exploration of new paradigms for sensing applications and genetic disease detection.

To achieve these goals we are pursuing a a three pronged iterative approach involving (i) modeling and theory involving reduced order models (ROM), (ii) a data driven approach to modeling that connects full order models (FOM) to ROM, and (iii) continuous interplay between theory, transport experiments, and chemical synthesis to verify design predictions and provide feedback for model and performance improvement.


2. Resistive and Phase Change Memory Devices:

Figure: Resistive Memory Device: (left) Shows filament growth and breakage, which leads to a low and high resistance state. (right) A model with three and seven Copper atom filaments that we model using DFT and Green’s function approaches.

Conductive filaments play a role in a number of electronic applications varying from dielectric breakdown, phase change memory and resistive memories. In the case of memory devices, they are a boon, which makes switching between high and low resistance levels possible. The high and low resistance levels in turn are used to represent the 0 and 1 logic states. The conductive filaments in memory devices can be anywhere from a few to tens of nanometers in length. The non equilibrium phenomena of formation and stability of conductive filaments is an open problem that is rich in basic science and of interest to applications.

In the area of electronics, the importance of conductive filaments arises from the need for large amounts of non volatile memory (defined to be memory that retains data even when powered down) in both mobile applications and data centers. Applications require nonvolatile memory to have short time scales for reading and writing of the memory state. Flash memory is the current leading nonvolatile memory technology. It is scaled and at the 25 nm technology node, the data retention time is more than ten years. However, continued scaling results in the degradation of retention time and endurance; therefore, the promise of significant areal density improvement beyond today’s flash devices is limited. In flash devices, electron charge is the state variable for memory storage, and the number of electrons dramatically decreases with scaling. The presence or absence of electrons in a conducting island buried in an oxide determines the memory state. Recent studies show that when the technology node reaches 16 nm, the memory state will be represented by only ~16 electrons stored in the oxide. In this case, tunneling of a few of these electrons will lead to loss of data. As a result, scaled Flash memory devices can easily fail due to space radiation and thermal fluctuations. To overcome the problems with flash memory scaling and performance, there is interest in technologies that do not use electron charge as a state variable. As opposed to charge based memory (where electrons and holes store information), it is possible to store information in filamentary structures, which represent either the presence or absence of atoms. Atoms are heavy and as a result, the memory state can be retained for a long time. In memory devices, the filamentary structures are formed in dielectric materials, where the presence or absence of a filament bridging electrodes can greatly alter the resistance. Changing the state variable from “electron” to “atom location” represents a transformational shift.

Conductive filaments are a wide spread physical phenomenon without sufficient understanding and affecting many important technologies that include computer memory, reliability of thin film devices, and power electronics. We are developing a multi-physics approach to understand filament formation caused by an electric field. The goal is to develop a physics based understanding of filament formation and kinetics. Theoretical and computational models to address: filament formation, high and low resistance values, and switching time between the two resistance values are at the core of our work. The approach involves solving equations underlying both atom movement and electron flow using both Monte Carlo and Green’s function based approaches. Temperature plays a crucial role and the heat equation is an important component.


3. Theory and Algorithms:

Quantum mechanical effects play an important role in determining the characteristics of nanoelectronic devices. These devices can be classified into two different categories (i) devices where operation is classical but quantum corrections are necessary to model them accurately and (ii) devices where quantum mechanics is central to device operation. Classical transistors which can be quite accurately modeled by drift-diffusion and hydrodynamic equations with some quantum corrections that account for the finite width of the inversion layer and tunneling through the oxide, is an example in the first category. Examples that belong to the second category include many emerging device technologies: (a) tunnel transistors (which can have an inverse sub threshold slope of smaller than classical 60 mV per decade of current), (b) superlattice based quantum cascade devices, (c) spintronic devices, (d) device concepts based on topological insulators and (e) quantum interference devices. In addition, quantum mechanical methods are also required to study the properties of devices based on new nanomaterials such as nanotubes, graphene and nanowires. In these structures, there is significant variation in the electronic properties (band structure, effective mass and density of states) depending on the surface passivation, diameter, and crystalline orientation / chirality. The lack of a set of transferable material properties for these variations makes it necessary to develop atomistic models where the location of atoms can be accounted for accurately. The modeling of devices at the nanoscale is fundamentally quantum mechanical, with robust methods being a necessity for devices in category (ii) and those based on new nanomaterials. New methods need to be developed to make a transformative change by improving simulation speed by at least one order of magnitude.

At the simplest level, when the applied voltages and the temperatures are extremely low (typically sub 4K), a nanodevice behaves phase coherently and an approach based on solving for the scattering states and using the Landuaer-Buttiker approach is justified in many cases. To use the Landauer-Buttiker approach, one needs to ensure that the device dimension is significantly smaller than the phase coherence length of the electron wave. However, in almost all situations of interest, devices operate at close to room temperature and away from equilibrium. Momentum and energy relaxation, and the breaking of quantum mechanical phase are important due to excitations such as phonons at finite biases even at low temperatures. The quantum mechanical modeling of devices by including decoherence and scattering has made some progress using a variety of interrelated many-body approaches: non equilibrium Green’s function, Liouville equation, density matrix, Kubo-based theory, Pauli master equation and Wigner function. It should also be mentioned that the Monte-Carlo approach has been specifically successful in probing the device physics of a wide class of nanodevices with complicated scattering mechanisms; the only drawback being that it cannot handle quantum coherence. In modeling the electrical properties of devices, the Green’s function approach has been rather successful in modeling devices using simplified Hamiltonians in the ballistic (phase-coherent limit) and in some cases with scattering. More recently, this approach has also been use to model silicided contacts in an abinitio manner.

We are performing research with our Applied Math collaborator Prof. Ulrich Hetmaniuk to make progress towards the development of mathematical algorithms that can be used in simulation tools to probe device physics and predict the current-voltage characteristics of a broad class of nanodevices.

While our work has been primarily funded by the National Science Foundation, we also acknowledge support from the following industrial sponsors: Dalsa Teledyne, Winbond, SRC, and HRL Inc.