This LDRD project has involved the development and application of Sandia's massively parallel materials modeling software to several significant biophysical systems. They have been successful in applying the molecular dynamics code LAMMPS to modeling DNA, unstructured proteins, and lipid membranes. They have developed and applied a coupled transport-molecular theory code (Tramonto) to study ion channel proteins with gramicidin A as a prototype. they have used the Towhee configurational bias Monte-Carlo code to perform rigorous tests of biological force fields. they have also applied the MP-Sala reacting-diffusion code to model cellular systems. Electroporation of cell membranes has also been studied, and detailed quantum mechanical studies of ion solvation have been performed. In addition, new molecular theory algorithms have been developed (in FasTram) that may ultimately make protein solvation calculations feasible on workstations. Finally, they have begun implementation of a combined molecular theory and configurational bias Monte-Carlo code. They note that this LDRD has provided a basis for several new internal (e.g. several new LDRD) and external (e.g. 4 NIH proposals and a DOE/Genomes to Life) proposals.
Collective impact ionization has been used to explain lock-on in semi-insulating GaAs under high-voltage bias. We have used this theory to study some of the steady-state properties of lock-on current filaments. In steady state, the heat gained from the field is exactly compensated by the cooling due to phonon scattering. In the simplest approximation, the carrier distribution approaches a quasi-equilibrium Maxwell-Boltzmann distribution. In this report, we examine the validity of this approximation. We find that this approximation leads to a filament carrier density that is much lower than the high density needed to achieve a quasi-equilibrium distribution. Further work on this subject is in progress.
A new type of GaAs laser is based on the electron-hole plasma in a current filament and is not limited in size by p-n junctions. High energy, electrically controlled, compact, short-pulse lasers are useful for: active optical sensors (LADAR, range imaging, imaging through clouds, dust, smoke, or turbid water), direct optical ignition of fuels and explosives, optical recording, and micro-machining. The authors present a new class of semiconductor laser that can potentially produce much more short pulse energy than conventional (injection-pumped) semiconductor lasers (CSL) because this new laser is not limited in volume or aspect ratio by the depth of a p-n junction. They have tested current filament semiconductor lasers (CFSL) that have produced 75nJ of 890nm radiation in 1.5ns (50W peak), approximately ten times more energy than ISL. These lasers are created from current filaments in semi-insulating GaAs and, in contrast to CSL, are not based on current injection. Instead, low-field avalanche carrier generation produces a high-density, charge-neutral plasma channel with the required carrier density distribution for lasing. They have observed filaments as long as 3.4cm and several hundred microns in diameter in the high gain GaAs photoconductive switches. Their smallest dimension can be more than 100 times the carrier diffusion length in GaAs. This paper will report spectral narrowing, lasing thresholds, beam divergence, temporal narrowing, and energies which imply lasing for several configurations of CFSL. It will also discuss active volume scaling based on recent high current tests.
To optimally design circuits for operation at high intensities of ionizing radiation, and to accurately predict their a behavior under radiation, precise device models are needed that include both stationary and dynamic effects of such radiation. Depending on the type and intensity of the ionizing radiation, different degradation mechanisms, such as photoelectric effect, total dose effect, or single even upset might be dominant. In this paper, the authors consider the photoelectric effect associated with the generation of electron-hole pairs in the semiconductor. The effects of low radiation intensity on p-II diodes and bipolar junction transistors (BJTs) were described by low-injection theory in the classical paper by Wirth and Rogers. However, in BJTs compatible with modem integrated circuit technology, high-resistivity regions are often used to enhance device performance, either as a substrate or as an epitaxial layer such as the low-doped n-type collector region of the device. Using low-injection theory, the transient response of epitaxial BJTs was discussed by Florian et al., who mainly concentrated on the effects of the Hi-Lo (high doping - low doping) epilayer/substrate junction of the collector, and on geometrical effects of realistic devices. For devices with highly resistive regions, the assumption of low-level injection is often inappropriate, even at moderate radiation intensities, and a more complete theory for high-injection levels was needed. In the dynamic photocurrent model by Enlow and Alexander. p-n junctions exposed to high-intensity radiation were considered. In their work, the variation of the minority carrier lifetime with excess carrier density, and the effects of the ohmic electric field in the quasi-neutral (q-n) regions were included in a simplified manner. Later, Wunsch and Axness presented a more comprehensive model for the transient radiation response of p-n and p-i-n diode geometries. A stationary model for high-level injection in p-n junctions was developed by Isaque et al. They used a more complete ambipolar transport equation, which included the dependencies of the transport parameters (ambipolar diffusion constant, mobility, and recombination rate) on the excess minority carrier concentration. The expression used for the recombination rate was that of Shockley-Reed-Hall (SRH) recombination which is dominant for low to mid-level radiation intensities. However, at higher intensities, Auger recombination becomes important eventually dominant. The complete ambipolar transport equation including the complicated dependence of transport parameters on the radiation intensity, cannot be solved analytically. This solution is obtained for each of the regimes where a given recombination mechanism dominates, and then by joining these solutions using appropriate smoothing functions. This approach allows them to develop a BJT model accounting for the photoelectric effect of the ionizing radiation that can be implemented in SPICE.
The longevity of high gain GaAs photoconductive semiconductor switches (PCSS) has been extended to over 100 million pulses at 23A, and over 100 pulses at 1kA. This is achieved by improving the ohmic contacts by doping the semi-insulating GaAs underneath the metal, and by achieving a more uniform distribution of contact wear across the entire switch by distributing the trigger light to form multiple filaments. This paper will compare various approaches to doping the contacts, including ion implantation, thermal diffusion, and epitaxial growth. The device characterization also includes examination of the filament behavior using open-shutter, infra-red imaging during high gain switching. These techniques provide information on the filament carrier densities as well as the influence that the different contact structures and trigger light distributions have on the distribution of the current in the devices. This information is guiding the continuing refinement of contact structures and geometries for further improvements in switch longevity.
The longevity of high gain GaAs photoconductive semiconductor switches (PCSS) has been extended to over 100 million pulses. This was achieved by improving the ohmic contacts through the incorporation of a doped layer that is very effective in the suppression of filament formation, alleviating current crowding. Damage-free operation is now possible at much higher current levels than before. The inherent damage-free current capacity of the bulk GaAs itself depends on the thickness of the doped layers and is at least 100 A for a dopant diffusion depth of 4 μm. This current could be increased by connecting and triggering parallel switches. The contact metal has a different damage mechanism and the threshold for damage (approximately 40 A) is not further improved beyond a dopant diffusion depth of about 2 μm. In a diffusion-doped contact switch, the switching performance is not degraded at the onset of contact metal erosion, unlike a switch with conventional contacts. For fireset applications operating at 1 kV/1 kA levels and higher, doped contacts have not yet resulted in improved longevity. We employ multi-filament operation and InPb solder/Au ribbon wirebonding to demonstrate >100 shot lifetime at 1 kV/1 kA.