Deconvolution of Partitioning Delays from Time-Resolved Trace Gas Measurements

Time-resolved
measurements of low-volatility gas-phase compounds
are limited by partitioning of the analyte to instrument surfaces,
resulting in what are known as partitioning delays. These delays slow
instrument responses and affect the accuracy of subsequent analyses.
In this work, we introduce a deconvolution algorithm that corrects
measurements affected by partitioning delays. We evaluate the performance
of this algorithm using synthetic data and also demonstrate its utility
in correcting partitioning delays in airborne nitric acid measurements.
We compare the effectiveness of deconvolution to the current best
practice for partitioning delays: frequent subtraction of instrument
background. Frequent background measurements are outperformed by the
deconvolution algorithm when sample concentrations are changing faster
than the instrument response time. The deconvolution algorithm can
be applied to time series that include frequent measurement of instrument
backgrounds, enabling reanalysis of past data. Furthermore, the algorithm
does not rely on any coincident data; it is effective without any
external information about the true time series of an analyte. When
applied to nitric acid measurements from a wildfire smoke plume, deconvolution
increases the calculated normalized excess mixing ratio (ΔHNO3/ΔCO) by 72%. We conclude that the deconvolution algorithm
is applicable to ground, airborne, and eddy covariance measurements
of “sticky” compounds.