Keith R. Coffee

Autonomous, broad-spectrum detection of hazardous aerosols in seconds.

Actual or surrogate chemical, biological, radiological, nuclear, and explosive materials and illicit drug precursors can be rapidly detected and identified when in aerosol form by a Single-Particle Aerosol Mass Spectrometry (SPAMS) system. This entails not only the sampling of such particles but also the physical analysis and subsequent data analysis leading to a highly reliable alarm state. SPAMS hardware is briefly reviewed. SPAMS software algorithms are discussed in greater detail. A laboratory experiment involving actual threat and surrogate releases mixed with ambient background aerosols demonstrates broad-spectrum detection within seconds. Data from a field test at the San Francisco International Airport demonstrate extended field operation with an ultralow false alarm rate. Together these data sets demonstrate a significant and important advance in rapid aerosol threat detection.

Detection of biological particles in ambient air using Bio-Aerosol Mass Spectrometry

The Bio-Aerosol Mass Spectrometry (BAMS) system is an instrument used for the real time detection and identification of biological aerosols. Particles are drawn from the atmosphere directly into vacuum and tracked as they scatter light from several continuous wave lasers. After tracking, the fluorescence of individual particles is excited by a pulsed 266nm or 355nm laser. Molecules from those particles with appropriate fluorescence properties are subsequently desorbed and ionized using a pulsed 266nm laser. Resulting ions are analyzed in a dual polarity mass spectrometer. During two field deployments at the San Francisco International Airport, millions of ambient particles were analyzed and a small but significant fraction were found to have fluorescent properties similar to Bacillus spores and vegetative cells. Further separation of non-biological background particles from potential biological particles was accomplished using laser desorption/ionization mass spectrometry. This has been shown to enable some level of species differentiation in specific cases, but the creation and observation of higher mass ions is needed to enable a higher level of specificity across more species. A soft ionization technique, matrix-assisted laser desorption/ionization (MALDI) is being investigated for this purpose. MALDI is particularly well suited for mass analysis of biomolecules since it allows for the generation of molecular ions from large mass compounds that would fragment under normal irradiation. Some of the initial results from a modified BAMS system utilizing this technique are described.

Characterization of ambient aerosols at the San Francisco International Airport using bioaerosol mass spectrometry

The BioAerosol Mass Spectrometry (BAMS) system is a rapidly fieldable, fully autonomous instrument that can perform correlated measurements of multiple orthogonal properties of individual aerosol particles. The BAMS front end uses optical techniques to nondestructively measure a particles aerodynamic diameter and fluorescence properties. Fluorescence can be excited at 266nm or 355nm and is detected in two broad wavelength bands. Individual particles with appropriate size and fluorescence properties can then be analyzed more thoroughly in a dual-polarity time-of-flight mass spectrometer. Over the course of two deployments to the San Francisco International Airport, more than 6.5 million individual aerosol particles were fully analyzed by the system. Analysis of the resulting data has provided a number of important insights relevant to rapid bioaerosol detection, which are described here.

DSP-Based Dual-Polarity Mass Spectrum Pattern Recognition for Bio-Detection

The bio-aerosol mass spectrometry (BAMS) instrument analyzes single aerosol particles using a dual-polarity time-of-flight mass spectrometer recording simultaneously spectra of thirty to a hundred thousand points on each polarity. We describe here a real-time pattern recognition algorithm developed at Lawrence Livermore National Laboratory that has been implemented on a nine digital signal processor (DSP) system from Signatec Incorporated. The algorithm first pre-processes independently the raw time-of-flight data through an adaptive baseline removal routine. The next step consists of a polarity dependent calibration to a mass-to-charge representation, reducing the data to about five hundred to a thousand channels per polarity. The last step is the identification step using a pattern recognition algorithm based on a library of known particle signatures including threat agents and background particles. The identification step includes integrating the two polarities for a final identification determination using a score-based rule tree. This algorithm, operating on multiple channels per-polarity and multiple polarities, is well suited for parallel real-time processing. It has been implemented on the PMP8A from Signatec Incorporated, which is a computer based board that can interface directly to the two one-Giga-sample digitizers (PDA1000 from Signatec Incorporated) used to record the two polarities of time-of-flight data. By using optimized data separation, pipelining, and parallel processing across the nine DSPs it is possible to achieve a processing speed of up to a thousand particles per seconds, while maintaining the recognition rate observed on a non-real time implementation. This embedded system has allowed the BAMS technology to improve its throughput and therefore its sensitivity while maintaining a large dynamic range (number of channels and two polarities) thus maintaining the systems specificity for bio-detection