\n",
"\n",
" Notebook Version 2: This notebook has been updated to reflect changes made to the package DifferentialMobilityAnalyzers.jl to work with the Julia v1 series (tested with Julia 1.1.0). To read the original supplement published with the paper please switch to v1.0.0 of the package DifferentialMobilityAnalyzers.jl and/or download the virtual machine on zenodo.org which contains a complete installation that works with Julia 0.6.4 \n",
"\n",
"This notebook demonstrates calculations related to the configuration where two DMAs are used to size select particles of opposite charge. The two populations are merged and allowed to coagulate. Coagulated dimers are isolated using an electrostatic filter. The dimers are charge neutralized and the size distribution is measured using a DMA operated in stepping or scanning mode. The notebook is a supplement to the manuscript \n",
"\n",
"Petters, M. D. (2018) A language to simplify computation of differential mobility analyzer response functions, Aerosol Science & Technology. \n",
"\n",
"The dimer preparation method was first described by Maisels et al. (2000). Rothfuss and Petters (2016) extended the method to study aerosol viscosity. Since then, the extended technique has been applied in published (Rothfuss and Petters, 2017, Marsh et al., 2018) and unpublished work. As noted in Rothfuss and Petters (2016), the response function of the setup is complex. Only the main peak of the response distribution has been used for interpretation. This notebook is to demonstrate how the response function for the dimer coagulation and isolation system can be modeled. Figure 1 summarizes a typical instrument configuration for the setup. \n",
"\n",
"\n",
" Figure 1. Schematic of the dual tandem DMA setup. \n",
"\n",
"Dried, charge equilibrated particles are classified in DMA1 and DMA2. One DMA is operated with a positive polarity and one with negative polarity power supply, transmitting negatively charged (black) and postively charged (red) particles, respectively. Monodisperse particle concentrations are monitored for both DMAs. The remaining sample flows are merged into a continous flow coagulation chamber with plug flow residence time 10 s < $t$ < 180 s. Particles that underwent coagulation with +1/-1 charge or +2/-2 charges are charge neutral (grey). Also particles that collide with ions in the chamber may revert from being charged to charge neutral monomers. At the exit of the chamber all charged particles are removed with an electrostatic filter. The remaining charge neutral particles are passed through a bipolar charger to reinstate charge equilibrium. Then the size distribution is measured using the third DMA that is operated in scanning or stepping mode, resulting in the function for DMA 3.\n",
"\n",
"To model the response function for DMA 3, the coagulation is conceptualized as follows. Two aerosol populations with identities \"1\" and \"2\" are introduced into in the chamber. Each population has unique and defined chemical composition, charge state (0, Β±1, or Β±2) and size distribution (πβ and πβ). Both distributions have the same mode diameter. Chemical composition and charge state are identical for particles within a population. It is assumed that there are no wall losses in the chamber. With these assumptions, the size distribution of +k/-k dimers is then given by (Notebook S11)
\n",
"\n",
"where $t$ is time elapsed, $\\beta_{1,2}^{z(k_1,k_2)}$ is the Zebel-corrected Brownian coagulatation rate (Zebel, 1958, Jacobson, 2005, Notebook S11), $f = \\sqrt[3]{2}$, $β $ is the operator that shifts the distribution along the diameter axis, $*$ is the operator that scales the concentration of a distribution, $.*$ is the operator that computes square of two size distributions (Notebook S3). The factor $f$ accounts for the increase in sphere equivalent volume of the formed dimers. The Zebel-corrected Brownian coagulatation rate can be computed if temperature, pressure, and particle densities are known or assumed. The above expression models the dimer size distribution to first approximation, as is demonstrated via comparison with predictions from the particle-resolved Monte Carlo code for atmospheric aerosol simulation (PartMC) (Riemer et al., 2009, Tian 2017, Notebook S11). \n",
"\n",
"Production of neutral particles due to ion collisions is assumed to follow a first order process\n",
"
$\\frac{dN}{dt} = \\beta_d N$
\n",
"\n",
"where $\\beta_d$ is the decharge rate. If it is further assumed that the collision rate is independent with size (within the narrow bounds of quasi-monodisperse population), that positive and negative ion concentrations are the same, and that the number of decharged particles remains small relative to the total population, the decharge size distribution can be written as\n",
"\n",
"