Chemometric Methods for Assessing Exposures to Indoor Air Contaminants with a Microanalytical System

Research Trainee: Chunguang (Jerry) Jin, PhD Student, Dept. of Environmental Health Sciences at the University of Michigan

Faculty Researcher: Edward T. Zellers, PhD, Professor of Environmental Health Sciences and Chemistry at the University of Michigan

The PI is currently engaged in research projects aimed at developing miniature instrumentation that couples tunable gas-chromatographic (GC) separations with microsensor array detection. The separation module consists of two series-coupled micro-columns whose retention characteristics can be tuned by use of pressure and temperature modulation. The sensor array comprises a set of chemiresistors (CR) with different gold-thiolate nanoclusters as sorptive interface layers whose output provides a pattern or “spectrum” that is characteristic of the vapor being detected. The hardware required to realize both “meso-scale” and “micro-scale” prototypes of such systems is well along toward development, however, the software has lagged behind. This project sought to develop multivariate statistical methods and other chemometric methods, to address critical modeling and data analysis functions needed to guide the development and allow the implementation of these novel instruments specifically for application in assessing exposures to indoor air contaminants. First, software routines was developed that combine algorithms for pattern recognition with retention time values to assign identities to chromatographically resolved and partially unresolved response signals (i.e., peaks) from the analyzer. The software routine has been successfully tested on sensitivity data set collected on a 3-sensor CR array detector. Next, a class model based on Mahalanobis distance was developed for evaluating the fidelity of the response patterns obtained from chromatographically resolved target vapors to the patterns for such vapors stored in a calibration pattern library, thereby establishing a means of assigning a statistical confidence level to the assignments of vapor identities when possible (uncalibrated) co-eluting interferences may be present. Then, in order to develop a method was developed for combining these sophisticated statistical methods with physico-chemical models of interactions between analyte vapors and both the column stationary phases and the microsensor interface layers for the purpose of assigning identities to unknown, previously uncalibrated analytes, retention models of a dual column was established, and response pattern data set was collected for more than 30 compounds to help with exploration of response model for CR sensors. Linear-solvation-energy relationship (LSER) models were employed as a very powerful and versatile tool for predicting responses and improving the reliability of analyses performed with such microsystems. The possibility of construction of a hybrid sensor array from TSMR sensors and capacitive sensors was also studied by math modeling and multivariate statistic analysis. This research has addressed on-going needs identified in the NIOSH National Occupational Research Agenda (NORA) in the areas of emerging technologies and exposure assessment methods.

 

Publications resulting from this project:
Lu CJ, Jin C, Zellers ET. Chamber evaluation of a portable GC with tunable retention and microsensor-array detection for indoor air quality monitoring. J Environ Monit. 2006;8(2):270-278. doi:10.1039/b515696c.

Grants resulting from this project:
Transportation Security Administration-Deptartment Of Homeland Security. 06-G-024. Au-Thiolate Nanoparticles as Interfacial Layers on Microsensor Arrays for Trace Explosive Vapor Detection (PI: Zellers). 09/29/06–09/28/09.
 

Research trainee’s current position:
Chunguang (Jerry) Jin received his PhD in 2008 and is currently a Process Analytics Scientist at Actavis.