Current Research
We are designing molecules that will make novel materials with improved electro-optic performance for use in next generation computer chips, revolutionizing information technology.
Background
Computers and many other modern devices require interconnections for data sharing and storage. All interconnection is optical: information is encoded in an optical signal (like radio or TV). Materials that make the direct conversion from computer signals to encoded optical signals enable this process. Therefore, designing better materials will reduce the time to send signals (i.e., bandwidth) as well as the expense of data storage (i.e., the energy consumption). The DOE, DOD and the NRO have outlined the need for improved processor interconnects because the existing ones are too slow and require far too much energy. Rapid, power-efficient communication is needed. At present most interconnects run at about 50 GHz (0.05 THz). We have already demonstrated using on-chip prototype devices (using either silicon or plasmon organic-hybrid devices) that Si-based computing components can transfer information at rates up to 5 THz, with a power consumption less than 1 fj/bit.
What the devices require
We are working to achieve a five-fold improvement in EO activity over existing materials. This will allow engineers to design components that will improve rate and power efficiency by over an order of magnitude from what is presently available.
The design process
One uses the principles of organic chemistry to design such molecules and to place pendant groups around the cores to improve the material's plastic quality. Several general principles have already been well established: The core of any EO active molecule must be acentric, with an electron donating group on one end and an electron accepting group on the other. The two groups must be coupled by a bridge that is rigid and conjugated to allow electron flow. We have also shown that in general straight molecules with subtle pattern modulation give rise to the best packing and order. We also now know that the less space donated to non-essential components (e.g. a polymer host) the better will be the EO activity.1
A Molecular System of an "infinite number" of electro-active chromophore-containing molecules, which are condensing and ordering under an applied external electric field. This uses re-entrant boundary conditions (the box repeats in x, y and z) so the molecules on the left are touching those on the right and top and bottom and front and back. In this way the organization of the chromophores simulates the bulk properties. The simulation is done at constant pressure and temperature and naturally condenses to a density consistent with experimental values. The property of interest (for us) is the average net orientation along the poling field. This property is also consistent with experimental estimates.
Tools we use
At present there are about five different core structures which have proven reasonably successful and all have about the same EO performance. Initially, we used our experience and judgment to guess a number of alternative donors, bridges and acceptors to those currently in use. By guessing about a hundred different structures we have found three that we predict will have over five-fold improvement. It is now up to the organic synthesis chemists, who work with us, to actually make the molecules. We will then test these molecules using prototype devices to compare whether there is improved performance. Clearly AI and machine learning are ideal methods to automate the design process and more effectively combine known molecular fragments to predict molecules with even greater activity. This is an exciting area of our research that will combine our work with cutting-edge AI advances.
Quantum mechanical methods can predict (quite accurately) the non-linear polarizability of molecules both as individual molecules, molecules in a general solvent, and molecules within a specific environment. These fundamental theories, through statistical mechanics, and the theory of electricity and magnetism are melded to understand how the materials generate the electro-optic properties.
The theoretical tools to understand the molecular systems include:
- Quantum Mechanical simulations using DFT and TDDFT methods with hybrid functions applied to large organic molecules.
- Statistical Mechanical methods to simulate the organization of nano- and meso-scale molecular systems.
- Maxwell's equations to evaluating the interaction of material and light to predict linear and non-linear polarizability and the interaction of light and matter.
- Relaxation theory to include the effects of non-linear statistical mechanical system response to orienting fields.
We have used molecular dynamics and Monte Carlo techniques to understand the organization of electro-active organic molecules into condensed matter materials. In these systems there is an asymmetric, poling electric field that tends to order the molecules to make a monolithic material with an overall nonsymmetrical order. The simulations require the application of statistical mechanics to molecular dynamics, using classical force fields to characterize the molecules.
We have adapted the theories of electricity and magnetism to understand how the molecules generate the electro-optic properties. Quantum mechanical methods can predict (quite accurately) the non-linear polarizability of molecules both as individual molecules, molecules in a general solvent, and molecules within a specific environment.