James Allen Mullens

NMIS plus gamma spectroscopy for attributes of HEU, PU and HE detection

Abstract A combined nuclear materials identification system–gamma ray spectrometry system can be used passively to obtain the following attributes of Pu: presence, fissile mass, 240/239 ratio and metal versus oxide. This system can also be used with a small, portable, DT neutron generator to measure the attributes of highly enriched uranium (HEU): presence, fissile mass, enrichment, metal versus oxide; and detect the presence of high explosives (HE). For the passive system, time-dependent coincidence distributions can be used for the presence, fissile mass, metal versus oxide for Pu, 240/239 ratio, and gamma ray spectrometry can also be used for 240/239 ratio and presence, allowing presence and 240/239 ratio to be confirmed by two methods. For the active system with a DT neutron generator, all relevant attributes for both Pu and HEU can be determined from various features of the time-dependent coincidence distribution measurements. Active gamma ray spectrometry would determine the presence of HE. The various features of time-dependent coincidence distributions and gamma ray spectrometry that determine these attributes are discussed with some examples from previous determinations.

252Cf-source-correlated transmission measurements for uranyl fluoride deposit in a 24-in-OD process pipe

Characterization of a hydrated uranyl fluoride (UO{sub 2}F{sub 2}{center_dot}nH{sub 2}O) deposit in a 17-ft-long, 24-in.-OD process pipe at the former Oak Ridge Gaseous Diffusion Plant was successfully performed by using {sup 252}Cf-source-correlated time-of-flight (TOF) transmission measurements. These measurements of neutrons and gamma rays through the pipe from an external {sup 2521}Cf fission source were used to measure the deposit profile and its distribution along the pipe, the hydration (or H/U), and the total uranium mass. The measurements were performed with a source in an ionization chamber on one side of the pipe and detectors on the other. Scanning the pipe vertically and horizontally produced a spatial and time-dependent radiograph of the deposit in which transmitted gamma rays and neutrons were separated in time. The cross-correlation function between the source and the detector was measured with the Nuclear Weapons Identification System. After correcting for pipe effects, the deposit thickness was determined from the transmitted neutrons and H/U from the gamma rays. Results were consistent with a later intrusive observation of the shape and the color of the deposit; i.e., the deposit was annular and was on the top of the pipe at some locations, demonstrating the usefulness of this method for deposit characterization.

Autonomous Control Capabilities for Space Reactor Power Systems

The National Aeronautics and Space Administration’s (NASA’s) Project Prometheus, the Nuclear Systems Program, is investigating a possible Jupiter Icy Moons Orbiter (JIMO) mission, which would conduct in‐depth studies of three of the moons of Jupiter by using a space reactor power system (SRPS) to provide energy for propulsion and spacecraft power for more than a decade. Terrestrial nuclear power plants rely upon varying degrees of direct human control and interaction for operations and maintenance over a forty to sixty year lifetime. In contrast, an SRPS is intended to provide continuous, remote, unattended operation for up to fifteen years with no maintenance. Uncertainties, rare events, degradation, and communications delays with Earth are challenges that SRPS control must accommodate. Autonomous control is needed to address these challenges and optimize the reactor control design. In this paper, we describe an autonomous control concept for generic SRPS designs. The formulation of an autonomous control concept, which includes identification of high‐level functional requirements and generation of a research and development plan for enabling technologies, is among the technical activities that are being conducted under the U.S. Department of Energy’s Space Reactor Technology Program in support of the NASA’s Project Prometheus. The findings from this program are intended to contribute to the successful realization of the JIMO mission.

Paradigm for selecting the optimum classifier in semiconductor automatic defect classification applications

The automatic classification of defects found on semiconductor wafers using a scanning electron microscope (SEM) is a complex task that involves many steps. The process includes re- detecting the defect, measuring attributes of the defect, and automatically assigning a classification. In many cases, especially during product ramp-up, and when multiple products are manufactured in the same line, there are few training examples for an automatic defect classification (ADC) system. This condition presents a problem for traditional supervised parametric and nonparametric learning techniques. In this paper we investigate the attributes of several approaches to ADC and compare their performance under a variety of available training data scenarios. We have selected to characterize the attributes and performance of a traditional K-nearest neighbor classifier, probabilistic neural network (PNN), and rule-based classifier in the context of SEM ADC. The PNN classifier is a nonparametric supervised classifier that is built around a radial basis function (RBF) neural network architecture. A basic summary of the PNN will be presented along with the generic strengths and weakness described in the literature and observed with actual semiconductor defect data. The PNN classifier is able to manage conditions such as non-convex class distributions and single class multiple clusters in feature space. A rule-based classifier producing built-in core classes provided by the Applied Materials SEMVision tool will be characterized in the context of both few examples and no examples. An extensive set of fab generated data is used to characterize the performance of these ADC approaches. Typical data sets contain from 30 to greater than 200. The number of classes in the data set range from 4 to more than 12. The conclusions reached from this analysis indicate that the strengths of each method are evident under specific conditions that are related to different stages within the VLSI yield curve, and to the number of different products in the line.