Tandem aerodynamic aerosol classifier – differential mobility analyzer as a single charged aerosol source for diameters up to one micrometer

Abstract This work describes a method to generate monodisperse, spherical, single-charged aerosols for instrument calibration in the size range of 32 – 1000 nm using an aerodynamic aerosol classifier (AAC) and differential mobility analyzer (DMA) in series. The AAC is used to input a monodisperse aerosol into the DMA to reduce the fraction of larger multiple charged particles exiting the DMA. The method produces test aerosols over a wide size range with less than 1% multiple charging and in concentrations between 4,000 – 10,000 particles/cm3 at more than 2 liters/minute flow rate, sufficient to be measured with the aerosol electrometer and condensation particle counter (CPC). At diameters between 32 nm and 1000 nm, the multiple charging was less than 0.1%. Operating parameters for the AAC and DMA are provided, along with CPC detection efficiencies up to 1000 nm measured with traceability to the international system of units. Use of the AAC allows for calibrations to be performed over almost the full particle diameter range of the DMA with an easily reproduced experimental setup using commercially available equipment.


Introduction
Primary calibration of a condensation particle counter (CPC) (Agarwal and Sem 1980) with an aerosol electrometer (AE) is well known, first described in 1974 (Liu and Pui 1974). In the past ten years the AE has been recognized internationally as a primary standard through laboratory inter-comparisons (H€ ogstr€ om et al. 2014) and the publication of an international standard for aerosol number concentration calibration, ISO 27891 (ISO 2015). A key requirement of ISO 27891 is generating a charged monodisperse test aerosol with all particles having the same polarity using a differential mobility analyzer (DMA) (Knutson and Whitby 1975), creating a test aerosol that is appropriate for measurement with the AE. Ideally each particle should have a single net charge when using an AE as a measurement standard for aerosol number concentration. This is possible, but in general there is a fraction of larger multiple charged particles that exit the DMA, creating a bias in the measurement. Much of the effort in the calibration is focused on reducing or eliminating the multiple charged particles.
Several methods are used to reduce the fraction of larger particles exiting the DMA. Some examples are the use of electrospray as a primary aerosol source for particles below a few hundred nanometers (Fletcher et al. 2009), impaction of larger particles downstream of the DMA for diameters from 100 to 1000 nm (Romay-Novas and Pui 1988), and the specialized single charged aerosol reference (SCAR) system for diameters from 10 to 1000 nm with uncertainties due to multiple charging of 0.16% (Yli-Ojanpera et al. 2010, 2012H€ ogstr€ om et al. 2011). Operating a DMA and particle mass analyzer (Ehara, Hagwood, and Coakley 1996;Olfert and Collings 2005) in series is also effective and can be useful for diameters up to 500 nm when using the aerosol particle mass analyzer (Kuwata 2015;Radney and Zangmeister 2016). Each of these methods can reduce the fraction of multiple charged particles to less than 10% as specified in ISO 27891. In this work an aerodynamic aerosol classifier (AAC) and DMA in series are used to reduce the fraction of multiple charged particles exiting the DMA. This method uses the AAC to first classify a monodisperse aerosol of spherical particles from a broad, easily generated, polydisperse primary aerosol. The monodisperse aerosol exits the AAC with an undefined charge state and is passed through a bi-polar charger and input into the DMA, enabling the DMA to produce a nearly single charged test aerosol. Test aerosols are generated in the size range of 32-1000 nm, with less than 1% multiple charging in the worst cases and less than 0.1% over most of the size range, and in sufficient concentration to be measured with an AE. This method is appropriate for calibrating a CPC over the plateau range, while other aerosol generation methods such as electrospray or the SCAR are necessary for calibration with smaller particles near the CPC cutoff diameter. The AAC classifies by particle relaxation time using an applied centrifugal force and aerodynamic drag and does not require a known aerosol charge state (Tavakoli and Olfert 2013). The AAC alone has been used to generate a monodisperse calibration aerosol when there are no particle charging requirements (Horender, Auderset, and Vasilatou 2019;Tran et al. 2020). Tandem AAC -DMA has been previously used to separate particles from a polydisperse aerosol for analysis with additional instrumentation, where mass, density, diameter, mass-mobility exponent, dynamic shape factor, and light absorption properties, among others, can be measured (Tavakoli and Olfert 2014;Yao et al. 2020;Kazemimanesh et al. 2022). Multiple charging from the DMA can introduce error into AAC -DMA measurements and has been addressed in the literature. The tandem AAC -DMA operating conditions to eliminate multiple charged particles has been described theoretically, with application to measuring nonspherical particle size distributions and light absorption by soot (Song et al. 2022). Eliminating the multiple charging is dependent on particle morphology and resolution of the AAC and DMA. Multiple charging from a tandem AAC -DMA configuration can be effectively eliminated for spherical particles. However this is generally not the case for aspherical particles, with the possible exception being for high AAC and DMA resolutions using small sample flow rates that results in low output concentration for downstream instruments (Song et al. 2022). This work describes experimentally the tandem AAC -DMA as part of a single charged test aerosol generator for spherical particles, with an additional requirement to generate aerosols in the range of 4,000 -10,000 particles/cm 3 at flow rates above 2 liters/min. Operating conditions of the system are provided that reduce or effectively eliminate multiple charging. The sections below describe the test setup for different size ranges, the test results, and conclusions.

Materials and methods
Two configurations are required to generate a calibration aerosol at high enough concentrations and with small multiple charging to perform the calibration from 32 to 1000 nm. For diameters below 500 nm, a Collison nebulizer (3076, TSI, Inc.) spraying a solution of polyalphaolefin (PAO) oil and ethanol produces sufficient concentration to generate tens of femtoamps measured by the AE (3068B, TSI Inc). Figure 1 shows the test setup using the Collison, AAC (Cambustion), DMA (3082 Electrostatic Classifier with 3081 Long DMA, TSI, Inc.), scanning mobility particle sizer (SMPS) (3082 electrostatic classifier with 3772 CPC, TSI, Inc.) (Wang and Flagan 1990), and the laser aerosol spectrometer (LAS) (3340 A, TSI, Inc.). The primary aerosol is pressurized through the AAC and DMA into the mixing bulb, where it combines with a pressurized dilution flow. The diluted aerosol is then passed into a vented volume where it is sampled by the AE, CPC, and SMPS for diameters up to 200 nm. For diameters above 200 nm, the SMPS is replaced by the LAS, a 100 channel optical spectrometer that measures size distributions from 90 nm to 7500 nm. Since spherical PAO oil particles are used, in each case the AAC is set to the aerodynamic diameter corresponding to the mobility equivalent diameter for a particle density of 0.8 g/cm 3 by Equation (1) (Hinds 1998), where d a is the aerodynamic diameter, d p is the spherical particle diameter, q p is the density of the spherical particle, and q o is the standard particle density of 1 g/cm 3 . It is assumed that the mobility equivalent diameter is equal to the spherical particle diameter, and the slip corrections are neglected. For diameters above 500 nm, a Laskin nozzle (Air Techniques International) spraying pure PAO oil through the AAC and two DMAs in parallel is required to generate high enough concentration. The relatively large diameters require a low sheath and sample flow through the DMA, so two DMAs with the same settings are used in parallel to double the aerosol flow into the mixing bulb and raise the concentration. The LAS is used to measure the test aerosol size distribution. The test setup is shown in Figure 2. The main components of each setup are listed in Table 1 for the particle size ranges described.
The calibration process is similar to ISO 27891, with particle counter zero checks, inlet flow rate measurement, cycling of the DMA voltage between zero volts and the voltage corresponding to the particle size of interest to measure net charge concentration with the AE, and measurement of splitter bias between the AE and CPC. Calibration of flow meters, the DMAs, and the AE electronic current are performed in-house at the U.S. Army Primary Standards Laboratory, with traceability to the international system of units (SI) through the National Institute of Standards and Technology (NIST) or internal quantum standards. The AE electronic current is calibrated over the range of þ/À 10 fA to þ/À 100 fA (Sakurai and Ehara 2011) using a 100 gigaohm resistor (model 111, Ohm-Labs) and millivolt source (Fluke 5502 A), with uncertainties of 0.14%. Typical noise for this AE is at or below 0.5 fA standard deviation. The flow meter (4140, TSI, Inc.) is calibrated to within 2% of reading for flow at standard conditions of 21.1 C and 101.325 kPa. The DMA is calibrated using the flow meter and 203 nm polystyrene latex spheres (PSL) (3200 A, Thermoscientific) (Wiedensohler et al. 2018), and the LAS is calibrated similarly with PSL for sizing and inlet flow calibration using a mass flow meter (Alicat Scientific). The AAC is treated as an aerosol conditioner, used to input a monodisperse aerosol into the DMA for classification, and is not calibrated for particle sizing. Combining these uncertainty components, along with repeatability of the ratio of the CPC to AE inlet flow rate, five detection efficiency measurements at each particle diameter, and curve fitting errors over the range of particle diameters enables calibration from 32 to 1000 nm   with 2.6% uncertainty. A detailed uncertainty analysis is included in the online Supplemental Information. The main challenge with the calibration process is generating high enough concentrations while keeping the multiple charging low. Increasing sample flow through the AAC and DMA raises concentration and reduces instrument resolution, which widens the transfer functions and can lead to increased multiple charging. Particularly demanding was generating enough 562 nm and 1000 nm diameter particles, requiring the use of two DMAs in parallel. Despite these challenges, the system is believed to be easily reproduced due to the commercial availability of the equipment, ease of primary aerosol generation, and only minor custom plumbing assembly required for the aerosol mixer and splitter.

Test results
The test aerosol size distribution generated using the setup in Figure 1 is shown below with and without classification by the AAC. With the DMA set to 178 nm and no AAC, Figure 3 shows a large fraction of multiple charged particles measured by the SMPS. When operating the AAC upstream of the DMA, Figure 3 shows that all multiple charging is effectively eliminated, resulting in a single peak with geometric standard deviation of 1.04. To estimate the fraction of multiple charged particles in this case, particle counts were summed for 60 s with the SMPS voltage fixed first at the single charge peak and second at the double charged peak. The resulting ratio of the double to single charged particle counts was 1.8e-6. Table 2 shows the average particle net charge and particle concentration for diameters spaced at fixed intervals on a logarithmic scale. In all cases the multiple charging is less than 0.1% except at the largest and smallest diameters where it is less than 1%, and concentrations are high enough for an accurate measurement with the AE. This level of multiple charging is comparable to the SCAR, which has an uncertainty contribution of 0.16% due to multiple charged particles (Yli-Ojanpera et al. 2010, 2012H€ ogstr€ om et al. 2011).
Operating parameters of the AAC and DMA are shown in Table 3. Note that the AAC particle diameter is set consistently lower than the DMA diameter, since the AAC classifies by relaxation time for a 1 g/cm 3 density sphere and the oil particles have density near 0.8 g/cm 3 .
The objective of this work is to generate a single charged aerosol to perform a CPC calibration with an AE. To this end, CPC detection efficiencies in the plateau range were measured over three days using the AE and test setups of Figures 1 and 2. Results are   Table 3. Operating parameters of DMA and AAC during the calibration. The theoretical DMA resolution is calculated for nondiffusive particles (Knutson and Whitby 1975), and the theoretical AAC resolution is in the time domain (Tavakoli and Olfert 2013 Figure 4 is calculated by Equation (2) (Stolzenburg and McMurry 1991), with detection efficiency g, particle diameter D p , plateau efficiency A ¼ 0.965, 10% efficiency diameter B ¼ 6.5 nm, and 50% efficiency diameter C ¼ 10 nm, À Á : (2)

Conclusions
In this work, the tandem AAC -DMA is used in an aerosol number concentration calibration process to generate a monodisperse, spherical, single-charged test aerosol for comparison of a CPC with an AE. The system was able to meet concentration requirements of 4,000 -10,000 particles/cm 3 at more than 2 liters/min, with less than 1% multiple charging for diameters between 32 and 1000 nm. This size range is appropriate for calibrating a CPC over the plateau range. The concentration limits are driven by the femtoamp level AE noise at the lower end and the CPC high concentration specification at the high end. Operating parameters that minimize or effectively eliminate multiple charging are provided. The main tradeoff in the system is between increasing concentration and decreasing multiple charging. Increasing the aerosol sample flow through the AAC and DMA will raise the concentration, but will also reduce instrument resolution, which can lead to an increase in multiple charging. Calibration of a CPC was performed with 2.6% uncertainty and traceability to the SI, and with the use of the AAC, testing over almost the full range of the DMA is possible with an easily reproducible experimental setup using commercially available equipment.
Disclosure statement Figure 4. CPC detection efficiencies. Error bars are the total combined uncertainty at 95% confidence interval (coverage factor k ¼ 2).