American Association for Aerosol Research - Abstract Submission

AAAR 37th Annual Conference
October 14 - October 18, 2019
Oregon Convention Center
Portland, Oregon, USA

Abstract View


Novel Approaches to DMA, CPMA, and APM Transfer Function Evaluation and Inversion to Determine Two-Dimensional Aerosol Mass-Mobility Distributions

TIMOTHY SIPKENS, Jason S. Olfert, Steven Rogak, University of British Columbia

     Abstract Number: 592
     Working Group: Instrumentation and Methods

Abstract
Researchers are increasingly using tandem measurements of the mass and mobility to better characterize aerosols. Most commonly, this has involved using a differential mobility analyzer (DMA) in series with some kind of particle mass analyzer, such as the aerosol particle mass analyzer (APM) or centrifugal particle mass analyzer (CPMA). While most studies calculate some kind of summary parameter from this data, such as effective density or dynamic shape factor, more detailed information of the distribution of particle properties can be determined by instead determining the two-dimensional mass-mobility distribution. This presents unique challenges in terms of deconvolving the instrument functions, which mask the true distribution of properties. The present work expands on this problem in two significant ways. First, we consider novel approaches to evaluating APM and CPMA transfer functions, including particle tracking techniques that allow for closed-form expressions for these transfer functions under a wide range of flow and particle migration conditions. This has the capacity to greatly speed up the computations required to deconvolve tandem measurements. Second, we examine an array of inversion methods available to deconvolve the instrument functions to determine the two-dimensional mass-mobility distribution, including Tikhonov regularization, Twomey approaches, maximum entropy techniques (in the form of the multiplicative algebraic reconstruction technique or MART), and Bayesian or statistical approaches. This is to be demonstrated on simulated data and reveals that Bayesian approaches have the capacity to greatly improve reconstruction accuracy.