Correction of Partitioning Delays in Time-Resolved Measurements Using Deconvolution

Abstract:

Partitioning of semivolatile compounds to instrument and inlet surfaces causes delays in response time for in-situ time-resolved measurements. These partitioning delays can slow instrument response time by minutes for some analytes, significantly affecting quantification and time-series analysis of these compounds. A currently-used approach to correcting for these partitioning delays is frequent measurement of instrument background response. Here we present a new approach to correcting for partitioning delays that utilizes deconvolution to separate partitioning delays from the underlying time series. We use synthetic data to compare correction via deconvolution to frequent measurement of instrument background. We also use airborne extractive electrospray ionization mass spectrometry (EESI) measurements during FIREX-AQ to demonstrate the approach’s effectiveness in a field setting where fast time response is essential. We find that the deconvolution correction improves the correlation of the affected measurements with fast-response gas measurements (e.g. CO), both for airborne EESI measurements and for synthetic data. Deconvolution outperforms frequent measurement of background concentrations when partitioning delay timescales are known. Lastly, deconvolution can be applied to time series where frequent measurements of background concentration were made, allowing this approach to be applied to many existing measurements.