Introduction

Radionuclide Aerosol Sampler Analyzer (RASA)

A network of Radionuclide Monitoring Stations (RMS) is used to monitor for Comprehensive Nuclear Test Ban Treaty compliance (CTBT) as a part of the International Monitoring System (IMS). The U.S. developed Radionuclide Monitoring Station was designed by Pacific Northwest National Laboratory (PNNL) and is called the Radionuclide Aerosol Sampler Analyzer (RASA). RMS is a general term for technologies utilized by treaty members. RASA is the U.S. system. System design and performance features can be found in Miley et al. [1] and Forrester et al. [2]. The system samples aerosols for 24 hours (h), holds the samples in an intermediate position for 24 h to allow short-lived isotopes to decay, and then counts the radiation spectrum for 24 h with a high purity germanium detector (HPGe). The total flow of the system is approximately 900 cubic meters per hour of air (m3 h−1). Air is filtered by 0.25 m2 of electrostatically charged blown microfiber filter media which we designate here as filter material 1 (FM1). The baseline radionuclide sensitivity for each station is 10 µBq m−3 140Ba in air. An upper limit of 30 µBq m−3 is allowed for regions with high naturally occurring radioactive backgrounds. The power consumption for a radionuclide and noble gas station is on the order of several kilowatts [3]. The United States is responsible for operations and maintenance of 11 radionuclide monitoring aerosol systems and 4 xenon noble gas systems. General Dynamics (GD) operates and maintains U.S. radionuclide stations.

Current filter media

FM1 is the filter media currently utilized in the RASA. Polypropylene fibers are electrostatically charged to enhance the capture of nanoparticles. Electrostatic filter media mitigate particle penetration around the most penetrating particle size of 300 nm. Particles of this size are not captured as efficiently in non-electrostatic filter media since both diffusional capture and inertial capture are at a minimum for this particle size. Thompson et al. [4] originally selected FM1 based on measurements provided by the Radiation and Nuclear Safety Authority (STUK) in Finland [5] as well as radiochemical measurements performed by PNNL [6]. Sandia National Laboratories (SNL) has performed additional characterization of FM1 [7, 8]. SNL found that electrostatic effects were more dominant at lower filter face velocities (0.5 m/s) with some particle losses at 1.5 m/s and 2.5 m/s for supra-micrometer particles due to inertial particle rebound. At higher filter face velocities, electrostatic effects are less dominant with respect to inertial effects. Particles with significant inertia impact on fiber surfaces and sometimes rebound from polymer fibers. The addition of retention agents to the surfaces of polymer fibers was moderately successful for enhancing inertial particle capture, and mitigating particle losses, at lower velocities.

System requirements

Minimum RMS system requirements are specified by the Comprehensive Test Ban Treaty Organization [3]. Requirements pertinent to this study are provided in Table 1. Air flow rates must exceed 500 m3 h−1 with minimum filter collection efficiency and global collection efficiency of 80% and 60%, respectively. The baseline sensitivity for 140Ba is 10–30 µBq m−3 where the upper limit is provided for RMS locations with high backgrounds. The objective of this study was to enhance RMS performance through any combination of the following performance parameters: reduced power consumption (reduced operational cost and field deployment), reduced baseline sensitivity for 140Ba, improved air flow, reduction in collection time for enhanced geolocation of radionuclide sources, and enhanced data availability (reduced downtime).

Table 1 RMS performance requirements taken from CTBTO [9]
Full size table

Many of these performance metrics are coupled. For instance, increasing the volume of air collected decreases the minimum detectable concentration of 140Ba. However, increasing airflow generally requires more power consumption and a larger blower to pull higher air flows at increased system pressure drop. The required minimum aerosol collection efficiency of 80% at 200 nm (nm) is also challenging since increased filter face velocity generally reduces the performance of electrostatically charged filters. The coupling between performance parameters makes system optimization a challenge. For system enhancement, aerosol collection efficiency must be improved or maintained while other performance enhancing measures generally result in less efficient filter collection mechanisms.

Theory

Minimum detectable concentration (MDC)

CTBTO [9] specifies methods for certification of baseline 140Ba sensitivity. The minimum detectable concentration (MDC) is given by

$$MDC = \frac{{L_{\rm D} }}{{T \times V \times \varepsilon_{\rm E} \times \gamma_{i} \times \xi_{i} }} \times K_{\rm S} \times K_{\rm W} \times K_{\rm C}$$
(1)

where symbol definitions are given in Table 2. To compare different filter media in this study, we calculated the relative MDC between different filter media at different air flow rates.

Table 2 Variable definitions for Eq. (1), minimum detectable concentration (MDC)
Full size table

For our study, relative MDC calculations eliminate the need for specific details of the radiation detector. The three factors of interest for this study were (1) sampled air volume, V, (2) lower limit of detection, L D, and (3) filter efficiency E. The global collection efficiency, ξ, is assumed to be the product of the inlet transport efficiency and the filter efficiency. Assuming the inlet transport efficiency is constant (when comparing the performance of two different filters), the global transport efficiency was replaced by the filter efficiency in Eq. (1). To control these parameters, configuration changes could be made to the RASA filter material, air flow rate, or aerosol collection efficiency. The relative change in MDC can be calculated according to the following:

$$\frac{{MDC_{1} }}{{MDC_{0} }} = \frac{{V_{0} }}{{V_{1} }}\left[ {\frac{{2.71 + 4.65\sqrt {\mu_{B1} } }}{{2.71 + 4.65\sqrt {\mu_{B0} } }}} \right]\frac{{E_{0} }}{{E_{1} }} .$$
(2)

Assuming that \(4.65\sqrt {\mu_{B1} } > > 2.71\), E 0 = E 1, and μ B1/μ B0 scales proportionately with V 1/V 0, Eq. (2) simplifies to the following:

$$\frac{{MDC_{1} }}{{MDC_{0} }} = \sqrt {\frac{{V_{0} }}{{V_{1} }}} .$$
(3)

The relative change in MDC from a volume change alone is not fully realized due to the change in lower limits of detection (increased background). To cut the MDC in half, the sampled air volume would need to be quadrupled.

Weighted average collection efficiency

Minimum detectable concentration (MDC) could also be improved through changes to aerosol collection efficiency. The aerosol collection efficiency is dependent on particle size, thus, the collection efficiency curve must be condensed to a single averaged quantity for comparisons between two filter media. In this study, we formulated average weighted collection efficiencies for MDC calculations where the weighting function was derived from the particle size distribution of atmospheric aerosol. Radionuclide emission from underground nuclear testing may occur through the release of volatile gases which subsequently condense on to atmospheric aerosols. Aerosols containing trace radionuclide signatures are then collected by the RASA. The aerosol collection efficiency curves should be weighted by the activity distribution where particles of various sizes will contain more radioactive material depending on the attachment of radionuclides to atmospheric particles. Tokonami studied the attachment of radon progeny to atmospheric aerosols [10] and postulated that there are three separate attachment regimes: (1) for particles much smaller than the mean free path (~ 65 nm) of air the attachment coefficient is proportional to the square of particle diameter (~ surface area), (2) for particles much larger than the mean free path of air the attachment coefficient is proportional to particle diameter, and (3) a transition regime where particle diameter is of the similar order of magnitude as particle diameter. In this study, we analyzed data with weighting functions proportional to particle diameter and particle surface area to explore the impacts of choosing different moments of the atmospheric aerosol particle size distribution on the average weighted aerosol collection efficiency.

The tri-modal lognormal particle size distribution, used to represent atmospheric aerosols, is given by the following:

$$\begin{aligned} \frac{dN}{{d\ln (d_{\rm p} )}} = &\frac{{n_{1} }}{{\sqrt {2\pi } \ln (\sigma_{g1} )}}\exp \left[ {\frac{{ - \left[ {\ln (d_{\rm p} ) - \ln (d_{g1} )} \right]^{2} }}{{2\left[ {\ln (\sigma_{g1} )} \right]^{2} }}} \right] \\ &+ \frac{{n_{2} }}{{\sqrt {2\pi } \ln (\sigma_{g2} )}}\exp \left[ {\frac{{ - \left[ {\ln (d_{\rm p} ) - \ln (d_{g2} )} \right]^{2} }}{{2\left[ {\ln (\sigma_{g2} )} \right]^{2} }}} \right] \\ &+ \frac{{n_{3} }}{{\sqrt {2\pi } \ln (\sigma_{g3} )}}\exp \left[ {\frac{{ - \left[ {\ln (d_{\rm p} ) - \ln (d_{g3} )} \right]^{2} }}{{2\left[ {\ln (\sigma_{g3} )} \right]^{2} }}} \right]. \\ \end{aligned} .$$
(4)

For Eq. (4), variables are defined in Table 3.

Table 3 Variable definitions for Eq. ( 4 ), atmospheric aerosol size distribution
Full size table

Whitby and Sverdrup [11] presented 8 sets of parameters describing common atmospheric aerosols. There are three size modes typical of atmospheric aerosols: the nucleation mode containing particles which have formed from gas phase precursors, the accumulation mode consisting of particles which have formed from agglomerated nucleated particles and persist in the atmosphere for long periods of time, and the coarse mode containing dust and other particles that have been re-entrained from the ground. The parameter sets provided by Whitby and Sverdup represent aerosols ranging from coastal, inland, urban, and pristine conditions. For this work, we selected the average. However, for specific RMS locations, other particle size distributions may be more appropriate and could be analyzed on an ad hoc basis. Average atmospheric aerosol size distribution parameters are given below (Table 4).

Table 4 Tri-modal, lognormal, atmospheric particle size distribution parameters [11]
Full size table

The surface area particle size distribution, dS/d ln (d p), is formed from the moment of the number particle size distribution:

$$\frac{dS}{{d\ln (d_{\rm p} )}} = \frac{dN}{{d\ln (d_{\rm p} )}}\pi d_{\rm p}^{2} .$$
(5)

The units of dS/d ln (d p) are typically expressed as square micrometers of particle surface area per cubic centimeter of air (µm2 cm−3). According to Tokonami, this would be the appropriate distribution for weighting aerosol collection efficiencies in the small particle regime (d p << 65 nm). The first moment distribution (length) is given by

$$\frac{dL}{{d\ln (d_{\rm p} )}} = \frac{dN}{{d\ln (d_{\rm p} )}}d_{\rm p} ,$$
(6)

which is useful for weighting aerosol collection efficiencies in the large particle regime. Selection of the appropriate moment is important to assess what size particles are the most relevant for radionuclide monitoring.

The number particle size distribution, diameter weighted number distribution, and surface-area particle size distribution, for average atmospheric aerosol are shown in Fig. 1. The three modes (nucleation, accumulation, and coarse) are distinct. The shape of the dS/d ln (d p) curve is different from the dN/d ln (d p) curve because the surface area of particles grows with the second power of particle diameter.

Fig. 1
figure 1

Aerosol number, diameter, and surface-area distributions for atmospheric aerosol

Full size image

The cumulative number distribution, F N, can be calculated from the integral of (4):

$$F_{\rm N} (d_{\rm p} ) = \frac{{\int\limits_{d_{\rm p}=0}^{{d_{\rm p}}} {\frac{dN}{{d\ln (d_{\rm p} )}}} d\ln (d_{\rm p} )}}{{\int\limits_{d_{\rm p}=0}^{\infty } {\frac{dN}{{d\ln (d_{\rm p} )}}} d\ln (d_{\rm p} )}} = \int\limits_{d_{\rm p}=0}^{{d_{\rm p}}} {\frac{{d(N/N_{\rm T} )}}{{d\ln (d_{\rm p} )}}} d\ln (d_{\rm p} )$$
(7)

where the total number concentration, integrated over all particle sizes, is given by

$$N_{\rm T} = \int\limits_{d_{\rm p}=0}^{\infty } {\frac{dN}{{d\ln (d_{\rm p} )}}d\ln (d_{\rm p} )} .$$
(8)

Similarly, the cumulative surface-area distribution can be formulated from the appropriate integral of dS/d ln (d p). The cumulative number, diameter weighted, and surface-area weighted distributions for average atmospheric aerosol are shown in Fig. 2. These cumulative distribution curves are interpreted as the percent of total atmospheric particles, based on number, length scale, or surface-area, which reside in particles below size d p. By number, most atmospheric particles are less than about 200 nm. The diameter weighted distribution suggests that most particles lie in the particle size range of 5 nm < d p < 500 nm. Very little particle surface area resides with particles smaller than 10 nm, and nearly all particle surface area resides with particles smaller than 10 µm. Again, the selection of the weighting function determines the influence of aerosol collection efficiency on the average aerosol collection efficiency. Selecting the first moment distribution under-emphasizes the role of micrometer sized particles. The second moment (i.e., surface area) under-emphasizes the role of particles on the order of the mean free path of air (~ 65 nm).

Fig. 2
figure 2

Cumulative aerosol distributions by number and surface-area for atmospheric aerosol where α in the axis title denotes N, L, or S

Full size image

The weighting functions, w l(d p) and w s(d p), needed to calculate an average weighted aerosol collection efficiency can be formulated from dL/d ln (d p) and dS/d ln (d p). The surface area weighting function is given by

$$w_{\rm s} (d_{\rm p} ) = \frac{{d(S/S_{\rm T} )}}{{d\ln (d_{\rm p} )}}$$
(9)

and is simply the normalized surface-area distribution. The integral of Eq. (9) is equal to one. The weighting functions of the number, diameter, and surface area size distributions are shown in Fig. 3. It is apparent that the weighting functions for the surface area and number distributions are significantly different. By number, nucleation mode particles are the most prevalent in the atmosphere. However, due to the influence of the second moment of particle diameter, the surface area weighting function shifts to larger particles. If one were to focus on particle number rather than surface area or particle diameter, the average weighted aerosol collection efficiency would be primarily influenced by the collection efficiency curve between 10 nm and 100 nm. We focus on the first and second moment distributions where the average weighted aerosol collection efficiency is influenced by the range of aerosol collection efficiencies between particle diameters of 5 nm and approximately 10 µm.

Fig. 3
figure 3

Weighting functions for aerosol collection efficiency based on atmospheric aerosol number, length, and surface area distributions

Full size image

Space charge effect

In this study, we utilized particle-pre-charging to enhance aerosol collection efficiency. We also assessed the influence of particle charge on aerosol transport from the inlet of the system to the filter. A cloud of electrostatically charged aerosol particles creates its own electric field. This phenomenon is called the space charge effect. Charged aerosol particles within the electric field experience additional body forces which then influence their trajectories. Within an enclosure, charged aerosol particles will drive themselves to the surrounding walls. In the RASA, this would be an unacceptable consequence of particle-pre-charging as it would reduce the global collection efficiency.

Space charge phenomenon is governed by the Poisson equation where the electric potential, φ, is given by

$$\nabla^{2} \varphi = - \frac{\rho }{{\varepsilon_{0} }}.$$
(10)

In (10), ɛ 0 is the permittivity of free space (8.85 × 10−12 C2 N−1 m−2), and ρ is the space charge density (C m−3). For regions without space charge the Poisson equation simplifies to the Laplace equation where boundary conditions determine the electric potential. Boundary conditions can be specified for the Laplace equation as the first or second kind, φ = V and ∇φ = −σ/ɛ 0, respectively, where V is a defined voltage (often ground) and σ is surface charge density [12].

In cylindrical coordinates, with the boundary at a potential of zero (electrically grounded), the electrical potential created by space charge in the radial direction is given by the following:

$$\varphi (r) = \frac{{\rho R^{2} }}{{4\varepsilon_{0} }}\left[ {1 - \left( {\frac{r}{R}} \right)^{2} } \right].$$
(11)

In (11), r and R are the radial coordinate and maximum duct radius, respectively. The electric field in the radial direction (E r) is then calculated as the gradient of the scalar potential,

$$E_{\rm r} =- \frac{d\varphi }{dr}.$$
(12)

In cylindrical coordinates, the radial electric field strength is given by

$$E(r) = \frac{\rho r}{{2\varepsilon_{0} }}.$$
(13)

The electric field creates an electrostatic body force (F r) on each particle,

$$F_{\rm r} = qE_{\rm r} ,$$
(14)

where q is individual particle charge (Coulombs). The additional body force alters particle trajectories through an additional electrostatic migration velocity in the radial direction,

$$V_{\rm tr} = Z_{\rm e} E_{\rm r} ,$$
(15)

where the electrical mobility, Z e, is given by

$$Z_{\rm e} = \frac{{qC_{\rm c} }}{{3\pi \mu d_{\rm p} }} .$$
(16)

In (16), C c is the Cunningham slip correction factor which accounts for non-continuum drag on small particles and μ is the dynamic gas viscosity.

Electrostatic body forces on particles can be created through Coulombic interactions with electrostatically charged fibers. This enhances aerosol collection efficiency and is the desirable effect of particle-pre-charging experimentally quantified in this study. Electrostatically charged particles also create the space charge effect, which results in the deposition of particles on the walls of aerosol transport ducts. Deposition upstream of the filter due to space charge would be a deleterious effect and should be minimized to ensure global transport efficiencies remain unaffected.

Diffusion charging

Electrostatic particle charging is well documented in the literature [13, 14]. Aerosol particles will acquire charge through the diffusion of ions to particle surfaces. The diffusion charging process can be described by the following:

$$n_{\rm diff} (\tau ) = \frac{{d_{\rm p} kT}}{{2K_{\rm E} e^{2} }} \times \ln \left[ {1 + \frac{{\pi K_{\rm E} d_{\rm p} \bar{c}_{i} e^{2} N_{i} \tau }}{2kT}} \right]$$
(17)

where n diff is the number of particle charges acquired through diffusion charging, k is the Boltzmann constant, 1.38 × 10−23 (N m K−1), τ is the diffusion charging characteristic time (s), T is the gas temperature (K), K E is the constant of proportionality, 1/4πɛ 0 = 9 × 109 (N m2 C−2), e is the charge on an electron, 1.6 × 10−19C, \(\bar{c}_{i}\) is the mean thermal speed of ions, 240 (m s−1), N i is the Ion concentration (m−3).

Field charging

In the presence of an electric field, ions are driven to the surface of particles rather than transported through the diffusive process. Field charging is given by the following equation:

$$n_{\rm field} (\tau ) = \left( {\frac{3\varepsilon }{\varepsilon + 2}} \right) \times \left( {\frac{{Ed_{\rm p}^{2} }}{{4K_{\rm E} e}}} \right) \times \left( {\frac{{\pi K_{\rm E} eZ_{i} N_{i} \tau }}{{1 + \pi K_{\rm E} eZ_{i} N_{i} \tau }}} \right)$$
(18)

where n field is the number of particle charges acquired through electric field charging, ɛ is the relative permittivity of particle, ɛ = 3.9 for silicon dioxide, E is the electric field strength (V m−1), Z i is the ion mobility taken as 1.5 × 10−4 (m2 V−1 s).

Charge limit

Diffusion and field charge mechanisms can act additively but particle charge has a theoretical limit. Charges too closely spaced on the surface of a particle will result in spontaneous emission. The particle charging limit for solid particles is given by

$$n_{\rm lim} = \frac{{d_{\rm p}^{2} E_{\rm L} }}{{4K_{\rm E} e}}$$
(19)

where n lim is the theoretical limit to number of particle charges acquired, and E L is the surface field strength required for spontaneous emission (9 × 108 V m−1).

Experiment

Filter test bed

A filter test system capable of operating at high filter face velocities, low air pressures, and low air temperatures, was constructed during the work of Hubbard et al. [15] to characterize inertial filtration at filter face velocities ranging from 5.0 to 20.0 m/s. The system was modified during the work of Sanchez et al. [8] to characterize the performance of electrostatically charged filter media at moderate filter face velocities ranging from 0.5 to 2.5 m/s at atmospheric pressure representative of the RASA aerosol sampling regime. The performance of FM1 at atypically high filter face velocities was also reported in Hubbard et al. [7].

HEPA filtered air was drawn into a 5.08 cm diameter (2 inch) stainless steel tube where test aerosols were injected and diaphragm valves were used to control the filter face velocity and air pressure in the filter test section. Air flow rates were measured using laminar flow elements (LFE, Merriam Process Technologies, Cleveland, Ohio) with maximum air flow rates of 27.0 and 170.0 cubic meters per hour at a pressure drop of 8 inches of water. Two LFEs were needed to cover the range of flow rates studied here. Two test aerosols were generated: Arizona road dust (ISO 12103-1 A2, 0.5–5.0 µm) dispersed with a fluidized bed, and sodium chloride (NaCl, 30–400 nm) atomized from solution.

Particle densities and shape factors for each test aerosol are given in Table 5 where NaCl data were taken from the work of Spencer et al. [16] and ISO 12103-1 data were taken from Endo et al. [17]. The geometric mean diameter (d g) of NaCl was approximately 40 nm (nm) and the geometric standard deviation (σ g) was approximately 1.76. The geometric mean diameter of ISO test dust is in the range of 1–2 µm.

Table 5 Particle properties for filter challenge aerosols
Full size table

Aerosol measurements were taken upstream and downstream of the filter sample to calculate aerosol collection efficiency. Upstream and downstream measurements of aerosol concentration required approximately 5 min each including aerosol extraction from the filter test system and subsequent measurement. A single experiment consisted of 6 measurements (3 upstream and 3 downstream samples). Upstream and downstream measurements were alternated and the six aerosol concentration measurements constituted a single experiment. Each flow condition (e.g., filter face velocity) was measured in triplicate for both ISO and NaCl test aerosols. A separate filter specimen was used in each experiment.

In the past, we have attempted to generate both aerosols simultaneously. However, using multiple aerosol measurement instruments simultaneously resulted in undesirable dilution of measured aerosol samples (low concentration). Collection efficiency tests using NaCl and ISO aerosols were therefore conducted independently. A TSI Scanning Mobility Particle Sizer (SMPS) was used to measure aerosol size and concentration of NaCl aerosol (10–400 nm) and a TSI Aerodynamic Particle Sizer (APS) was used to measure the aerosol size and concentration of ISO test aerosol. Detailed experimental procedures, data analysis, and aerosol physics governing electrostatic and inertial aerosol filtration can be found elsewhere [7, 8, 15].

The current RASA filtration media, FM1, uses electrostatically charged filter fibers to enhance the collection efficiency of nanoparticles. SNL has performed extensive studies of electrostatic filtration in the past. In the current study, we hypothesized that aerosol collection efficiency could be enhanced by electrostatically pre-charging the aerosol particles. Lee et al. employed particle-pre-charging upstream of a conventional filter sandwiched between aluminum mesh structures held at high potential to attract charged particles [18]. Electrostatic charges on filter fibers can polarize particles creating a particle-fiber attraction force. Coulombic forces can also enhance particle-fiber capture if particles possess some charge. Particle-pre-charging was therefore implemented to determine if aerosol collection efficiencies could be improved substantially through Coulombic attraction between fibers and highly charged particles.

Some discussion of particle charge states is in order since we manipulated particle charge to affect aerosol capture efficiencies. Due to cosmic radiation and the emission of radioactive species from soil, there is a natural bipolar ion concentration in the air of approximately 103 ions cm−3 [14, 19, 20]. Highly charged aerosol particles will come into charge equilibrium, to the Boltzmann charge distribution, within approximately 100 min of exposure to the atmospheric bipolar ion concentration [14]. Thus, particles in the atmosphere generally have some electrostatic charge, but it is small, and the charge distribution is centered about zero. Many studies have addressed the role of radioactive self-charging of aerosols with particular emphasis on nuclear reactor accidents where ionizing radiation inside containment vessels is high [21–23]. More recently, Kim et al. have modeled the influence of radioactive particle self-charging on the transport of radioactive aerosols from atmospheric radionuclide releases from Chernobyl and Fukushima [24]. Kim et al. concluded that radioactive particle self-charging could impact short-range and long-range transport of radionuclide aerosols. Nuclear treaty monitoring scenarios are likely to result in different particle activities and different ion populations in the surrounding gas phase due to lower radioactivity. More work is needed to assess equilibrium particle charge distributions for treaty monitoring scenarios but the result will likely lie between the bounds established for non-radioactive aerosols and radioactive aerosols from nuclear reactor accidents.

The filter test system utilized in previous studies was modified such that corona discharges were included past the point of aerosol injection to pre-charge aerosol particles. Two ¾” bore-thru Swagelok fittings were welded onto a piece of 2” tube. Two particle pre-charging probes (SIMCO Ion Five Point Pinner Charging Applicator, model 4004738) were inserted approximately 2 cm into the air stream. Particle pre-charging probes ionized adjacent air through a corona discharge. The physics of particle charging and corona discharges will be included in the Discussion section. The pre-charging probes were powered and controlled with a SIMCO Ion Chargemaster VCM Power Supply (VCM30-Bipolar) capable of creating positive or negative polarity corona discharges. The potential gradient at the sharp points of the pre-charging probes cause the adjacent air to become ionized. Air ions then attach to aerosol particles passing the probes in the air stream. Each probe was limited to approximately 250 µA output current corresponding to a power supply voltage of approximately 25–30 kV.

Commercial electrostatic precipitators employ similar technology implemented here (particle-pre-charging). A corona discharge is used to pre-charge aerosol particles. In an electrostatic precipitator, conducting plates are used to create an electric field [25]. Charged particles migrate across the electric field and are collected on the plates. In the RASA, if particle-pre-charging were applied to enhance RASA sensitivity, charged particles would experience greater Coulombic attraction forces to electrostatically charged filter fibers thereby increasing particle collection efficiency.

Test parameters

Three filter media were assessed: Filter Material 1 (FM1, current RASA filter), FM2, and FM3. FM1 is sandwiched between two layers of “scrim” which provides structural support. FM1 scrim will be denoted S1. FM2 and FM3 were sandwiched between two layers of an alternate scrim, S2. Materials will be described in more detail below. Aerosol collection efficiencies were measured at three filter face velocities: 1.1 m/s (approximate RASA operating point), 2.75 m/s (2.5 times the approximate RASA operating point), and 5.5 m/s (5 times the approximate RASA operating point). Recall from the theory section that increased sample volume improves the baseline sensitivity. Three electrostatic particle charge states where characterized: the Boltzmann equilibrium charge distribution, particle charge induced by a positive corona discharge, and particle charge induced by a negative corona discharge. The Boltzmann equilibrium charge distribution was considered neutral or uncharged for this study as it is centered about zero and has zero net charge.

Materials

Scrim 1 (S1)

For the RASA, filter media is sandwiched between two layers of S1 scrim material. This scrim protects the filter material from mechanical devices (e.g., rollers) used to automatically move the filter material through the system. Fiber diameter is on the order of 25–35 µm.

Scrim 2 (S2)

SNL has identified and characterized a structural support material with lower pressure drop than S1. The material will be called S2. The fibers have a polyester core with a nylon skin where the relative amounts of materials are approximately 75 and 25%, respectively. Any increase in air permeability obtainable through structural support materials should also be considered in addition to the air permeability of the filter. Individual fibers are approximately 30–40 µm.

Filter material 1 (FM1)

As mentioned previously, FM1 has been studied extensively. Its performance can be found in the other works [7, 8]. The mean fiber diameter for FM1 lies in the range of 1–3 µm.

Filter material 2 (FM2)

A reduction in the total number of fibers in the mat, and increased fiber diameter of FM2, are characteristics that could give it higher air permeability. Thompson et al. [4] provide no data for a single layer of FM2 but state that it did not meet the performance requirement of 80% efficiency at 0.2 micrometers. Thompson et al. [4] and Valmari et al. [5] mention the use of dioctyl-phthalate (DOP) aerosol to challenge filters. DOP is an oil droplet based technique where the challenge aerosol is liquid rather than solid. Romay and Liu were among the first to assess the influence of oil droplet loading on electret filter performance [26]. Romay and Liu concluded that electret sites were masked by oil droplets resulting in performance degradation that exponentially increased with filter loading. They noted the emergence of oil resistant electrets for respirator applications due to the observation of this effect. Martin and Moyer [27] tested the performance of electrostatically charged respirator filters designed to be oil-resistant and oil-proof. Filter grades designed for use against oils showed little performance degradation after exposure to DOP aerosol whereas non-resistant grades showed 20–35% lower efficiencies when tested against DOP. Barret and Rousseau [28], Stevens and Moyer [29], and Martin et al. [30] also observed performance degradation of electrostatically charged filter media when challenged and loaded with DOP but results tended to be specific to certain filters and not a general characteristic of electrostatically charged filter media. Solvent baths can be used to eliminate electrostatic fiber charge, and performance degradation accompanies the loss of fiber charge. Coating fibers with oil aerosol seems to have a similar effect unless the fibers are specifically engineered to possess oil-resistance features. Recent SNL test data showed a single layer FM2 efficiency of approximately 90% at 200 nm with a filter face velocity of 5 m/s and an air pressure of 0.8 atmospheres. SNL data with solid NaCl particles therefore disagreed with data from previous assessments of FM2 with DOP aerosols. We hypothesized that DOP aerosol was less affected by particle-fiber electrostatic forces once fibers became coated with DOP during laboratory tests, and this artifact of testing may not be representative of atmospheric aerosols collected by the RASA. SNL therefore reexamined the performance of FM2 to see if its performance was satisfactory for the RASA using SNL developed test methods.

Filter material 3 (FM3)

Other electrostatic materials have been tested by SNL for high volume aerosol sampling. The air permeability of FM3 is substantially higher than FM1 or FM2. The manufacturer specification sheet for FM3 stated this non-woven electrostatic material has a collection efficiency of 87% at 0.1 µm. The specification sheet also gave a bipolar charge density of 50 nC cm−2. Sanchez et al. [8] calculated the fiber charge density for FM1 as 12 nC cm−2. FM3 fibers were rectangular, approximately 10 micrometers by 40 micrometers, and made of polypropylene similar to FM1 and FM2.

Calculations

Particle diameter conversion

The SMPS and APS particle sizing instruments utilize different fundamental principles to size aerosol particles: electrical mobility, and aerodynamic time of flight, respectively. The SMPS size segregates particles in an electric field and then counts nanoparticles after water has been condensed on to their surfaces. The aerodynamic particle sizer accelerates particles in a nozzle and utilizes time of flight to determine aerodynamic equivalent diameter. Thus, the diameters given by each instrument are not directly comparable to those given by the other instrument. Electrical mobility diameters (d m) given by the SMPS were converted to aerodynamic diameters (d a) to make an accurate comparison of filter efficiency across the entire spectrum of particle sizes (30 nm to 5 μm). DeCarlo et al. review particle sizing principles and particle size equivalents [31]. The volume equivalent diameter, d ve, is given by

$$d_{\rm ve} = d_{\rm m} \frac{{C_{\rm c} (d_{\rm ve} )}}{{C_{\rm c} (d_{\rm m} )}} \times \frac{1}{\chi }$$
(20)

where χ is the particle shape factor and C c is the particle slip correction factor use to incorporate the effects of non-continuum fluid drag on nanoparticles. The slip correction factor is easily calculated and is outlined in other seminal references on aerosol physics [14, 32]. The aerodynamic diameter, d a, can then be calculated with the true particle density, ρ p, and unit density ρ 0 = 1000 kg m−3.

$$d_{\rm a} = d_{\rm ve} \sqrt {\frac{1}{\chi }\frac{{\rho_{\rm p} }}{{\rho_{\rm o} }}\frac{{C_{\rm c} (d_{\rm ve} )}}{{C_{\rm c} (d_{\rm a} )}}}$$
(21)

Equations (20) and (21) were solved in MATLAB to determine the volume equivalent and aerodynamic diameters from electrical mobility diameters given by the SMPS. The APS measures ISO test particle size in aerodynamic diameter, thus, no conversion of APS data was required.

Aerosol collection efficiency

Aerosol collection efficiencies were calculated according to

$$E = 1 - \frac{{\bar{C}_{\rm d} }}{{\bar{C}_{\rm u} }}\frac{{\bar{P}_{\rm u} }}{{\bar{P}_{\rm d} }}$$
(22)

where \(\bar{C}_{\rm u}\) was the average of three measurements of upstream aerosol concentration, \(\bar{C}_{\rm d}\) was the average of three measurements of downstream aerosol concentration. The ratio of upstream and downstream air pressures, \(\bar{P}_{\rm u} /\bar{P}_{\rm d}\), was used to account for downstream flow expansion due to filter pressure drop [33]. A major pressure change across the filter causes the aerosol concentration to change even if no particles are removed from the flow.

The weighted average aerosol collection efficiencies, \(\hat{E}\), for filtration tests were calculated according to the following equation:

$$\hat{E} = \int\limits_{d_{\rm p}=0}^{\infty } {w(\ln (d_{\rm p} )) \times E(\ln (d_{\rm p} )) \times d\ln (d_{\rm p} )} .$$
(23)

Neutralizer dead volume correction

Particle-pre-charging was first applied in SNL filter research during this study. A TSI 3088 Advanced Aerosol Neutralizer was placed on the inlet to the aerosol sampling piston used to extract aerosol samples from the filter test bed. In previous testing this was unnecessary since particles had little excess electrostatic charge. However, upon particle-pre-charging the aerosol upstream of the filter, electrostatic losses within the sample piston were probable for highly charged aerosol particles. For this reason, we charge neutralized the aerosol particles after they had been extracted from the filter test system prior to being sized and counted in the instruments.

The space charge effect is used to describe mutual repulsion of charged aerosol particles and was outlined above. The sample extraction piston was needed for filter characterization since aerosol instrumentation was not capable of pulling against vacuum present inside the filter test system. The aerosol residence time within the piston was approximately two minutes, an appreciable amount of time for particles to migrate across the electric field and deposit on the interior walls of the extraction piston before they were injected into aerosol measurement instruments. This effect will not occur in the full-scale RASA since airborne particles do not reside within any specific portion of the RASA for more than a few seconds.

To mitigate measurement artifacts, we placed the neutralizer on the inlet of the piston to remove electrostatic charge from the particles prior to being measured in the APS or SMPS. By doing so, we introduced a dead volume into the aerosol sampling system which resulted in aerosol carryover between upstream and downstream measurements. Essentially, the neutralizer was full of the upstream aerosol when a downstream aerosol was extracted. That increased the downstream aerosol concentration above its true value. Likewise, the downstream aerosol concentration was not zero, and the subsequent upstream aerosol sample was affected by the previous sample. The actual upstream and downstream aerosol concentrations are given below where variable definitions are given in Table 6.

$$\bar{C}_{u,act} = \left( {\frac{{V_{\rm p} }}{{V_{\rm p} - V_{n} }}} \right)\bar{C}_{u,meas} - \left( {\frac{{V_{n} }}{{V_{\rm p} - V_{n} }}} \right)\bar{C}_{d,act}$$
(24)
$$\bar{C}_{d,act} = \left( {\frac{{V_{\rm p} }}{{V_{\rm p} - V_{n} }}} \right)\left( {\frac{{\bar{P}_{\rm u} }}{{\bar{P}_{\rm d} }}} \right)\bar{C}_{d,meas} - \left( {\frac{{V_{n} }}{{V_{\rm p} - V_{n} }}} \right)\bar{C}_{u,act}$$
(25)
Table 6 Variable definitions for neutralizer dead volume corrections
Full size table

The volume of the neutralizer was approximately 8% of the combined internal volumes of the piston and neutralizer. Without this dead volume correction, aerosol collection efficiencies would be systematically biased 5–10% with respect to the true collection efficiency.

Results

Filter pressure drop

The filter pressure drop is an important performance characteristic of the material. Lower air permeability (high pressure drop) materials require more powerful blowers to pull a given air flow rate through the filter. Powerful blowers consume more energy, have greater physical dimensions, and have higher acquisition cost. Low pressure drop materials are therefore desirable. Figure 4 shows the pressure drop of FM1, FM2, FM3, two layers of S2, and two layers of S1. For all data in this report, FM1 was sandwiched between two layers of S1 as it comes from the supplier. FM2 and FM3 were both sandwiched between two layers of S2. At a filter face velocity of approximately 1.1 m/s (current RASA set point), FM2 had approximately 70% of the pressure drop of FM1. For FM3, the pressure drop at 1.1 m/s was approximately 22% of the pressure drop of FM1. Pressure drop is non-linear at higher filter face velocities but these comparisons are a good illustration of the potential performance enhancements possible through the utilization of FM2 or FM3 in place of FM1

Fig. 4
figure 4

Filter pressure drop (ΔP f) versus filter face velocity (U) for FM1, FM2, FM3 and 2 layers of scrim materials S2 and S1. Filter materials FM2, and FM3 were sandwiched between two layers of S2. FM1 was sandwiched between two layers of S1

Full size image

Aerosol collection efficiency

Although FM1 has been characterized in the past, particle-pre-charging was not employed during previous studies. For this reason, we re-characterized FM1 without particle-pre-charging so its relative performance with particle-pre-charging could be assessed. FM1 data are given in Fig. 5. The aerosol collection efficiency of particles between 10 nm and 100 nm improved 5–20%. There was also modest improvement for micrometer size particles although the aerosol collection efficiencies were already greater than 90–95%. Error bars were included in the figure but cannot be seen because they are smaller than figure symbols. FM1 typically displayed less than 3% standard deviation between three independent aerosol collection efficiency tests. From this figure we concluded particle-pre-charging does offer some improvement to performance with FM1 but these improvements may be marginal in light of engineering changes needed to implement particle-pre-charging in fielded RASAs.

Fig. 5
figure 5

Aerosol collection efficiency (E) plotted against aerodynamic equivalent particle diameter (d a) for FM1 at filter face velocities of 1.1, 2.75, and 5.5 m/s, and electrostatically charged in negative and positive coronas, and to the Boltzmann charge distribution (neutral or uncharged)

Full size image

Figure nomenclature qf implies fibers were electrostatically charged. We only present data for charged fibers. However, the electrostatic fiber charge can easily be discharged in a bath of solvent like isopropanol and we have studied the effects of fiber charge in the past. Particle charge states are designated as follows: qp.boltz implies particles were in a quasi-Boltzmann charge distribution state, qp.neg implies particles were charged in a negative corona, qp.pos implies particles were charged in a positive corona. Filter face velocities are listed in the legend as well as set point currents for the corona discharge probes: 0.00 mA and 0.25 mA.

Data for FM2 are shown in Fig. 6. For uncharged aerosol particles, aerosol collection efficiency decreased for small particles (< 300 nm) when the filter face velocity increased. This was expected since electrostatic capture mechanisms became less dominant with increasing filter face velocity. Performance enhancements obtainable through increased air volume collection could therefore be counteracted by lower collection efficiencies. Particle-pre-charging enhanced nanoparticle aerosol collection efficiencies (< 300 nm) to nearly 100%. Aerosol collection efficiencies were lower than RMS requirements at a filter face velocity of 5.5 m/s for particles above 1–2 micrometers in aerodynamic diameter. With FM2, it is possible to operate at 2.75 m/s (2.5 times the current RASA flow) and maintain acceptable performance with or without particle pre-charging. Particle-pre-charging enhances the collection of particles with aerodynamic diameters below about 300 nm.

Fig. 6
figure 6

Aerosol collection efficiency (E) plotted against aerodynamic equivalent particle diameter (d a) for FM2 at filter face velocities of 1.1, 2.75, and 5.5 m/s, and electrostatically charged negatively, positively, and to the Boltzmann charge distribution (neutral)

Full size image

FM3 had high air permeability but lower aerosol collection efficiency when particle-pre-charging was not employed. Without particle-pre-charging FM3 would not satisfy performance requirements for radionuclide monitoring stations (80% collection efficiency at 200 nm). Figure 7 shows all data collected for FM3. FM3 had much higher variability from filter to filter. We attributed this to its much higher porosity which may exhibit more variance with respect to FM1 and FM2. Radionuclide monitoring station performance requirements could be met at filter face velocities of 1.1 m/s or 2.75 m/s with particle pre-charging from a negative corona. Collection efficiencies were higher when negative coronas were employed. This will be discussed in the Discussion section.

Fig. 7
figure 7

Aerosol collection efficiency (E) plotted against aerodynamic equivalent particle diameter (d a) for FM3 at filter face velocities of 1.1, 2.75, and 5.5 m/s, and electrostatically charged negatively, positively, and to the Boltzmann charge distribution (neutral)

Full size image

Weighted aerosol collection efficiencies were calculated according to (23) and shown in Table 7. Weighted efficiencies based on the first and second moments, \(\hat{E}_{l}\) and \(\hat{E}_{\rm s}\), respectively, are given. Comparatively, the first moment distribution places more emphasis on particles with diameters below 100 nm, and the second moment places more emphasis on supra-micrometer particles. The true weighted efficiency lies between \(\hat{E}_{\rm l}\) and \(\hat{E}_{\rm s}\).

Table 7 Weighted aerosol collection efficiencies for FM1, FM2, and FM3 for filter face velocities of 1.1, 2.75, and 5.5 m/s, and particle charge states of negative, positive, and neutral
Full size table

Discussion

Electrostatic effects

The effects of space charge were analyzed to determine if particle-pre-charging is feasible for full-scale RASA systems. This analysis was prompted by an observation in laboratory-scale testing. Figure 8 shows the aerosol number concentration as a function of electrical mobility diameter (d m) for NaCl aerosol generated with the TSI Constant Output Analyzer with multiple electrostatic charge distributions: (1) Boltzmann equilibrium charge distribution (uncharged), (2) charge distribution acquired in a positive corona, and (3) charge distribution acquired in a negative corona. The concentration of negatively charged particles was roughly 33% of the uncharged distribution. The positively charged aerosol was approximately 60% of the original distribution. We see from laboratory-scale experiments that highly charged particles are more efficiently captured by electrostatic filter media. However, if particle-pre-charging results in particle losses upstream of the filter, baseline sensitivity improvements would be offset by inefficiencies in charged particle transport within the system. For particle-pre-charging to be a viable option for the RASA we must understand if particle losses were an artifact of the particle-pre-charging apparatus in laboratory-scale testing and if we would expect to see similar losses in fielded RASA systems if particle-pre-charging were employed.

Fig. 8
figure 8

Aerosol number concentration for NaCl test aerosol in laboratory scale system

Full size image

Coronas are generated when free electrons are accelerated in a high strength electric field, collide with molecules, and eject additional electrons leaving behind charged molecules [34]. Those molecules may then migrate toward aerosol particles and attach to form charged aerosol particles. In our experiments, we varied the current control on the ion generators to provide the maximum current possible with these ionizers. The ionizer power supply adjusted the set point voltage internally (limit of ± 30 kV) to achieve the desired current (limit of approximately ± 0.25 mA). Negative coronas (negative potential) require the surrounding gas to contain species that can absorb free electrons whereas the positive corona does not [13]. For experiments where a negative corona was used, the SIMCO ionizer power supply had typical values of −20 kV and 0.25 mA. These values were within 20% for the positive corona and exhibited variability from day to day. It is difficult to conclude from these data why the negative corona charged particles would have approximately one half the number concentration of the positive corona charged particles. Electrons, negative air ions, and positive air ions have different electrical mobilities: 6.7e−2, 1.6e−4, and 1.4e−4 m2 V−1 s−1, respectively [14]. Ion-particle attachment factors also vary for different polarities. Thus, it is not clear why the resultant charge state was substantially different for the positive and negative polarity coronas but many factors were likely to contribute to the net result. Laboratory-scale filtration tests showed that aerosol particles charged in a negative corona were captured more efficiently by FM3.

Test data showed there was an effect of particle-pre-charging on aerosol concentration downstream of the ionizer. This effect could be attributed to (1) space charge losses, (2) losses attributable to high electric fields in the vicinity of the corona ionizers, or (3) sampling losses in ¼” tubing used to sample aerosols with TSI instruments (not a part of fielded RASAs). Order of magnitude estimates for parameters which affect particle charging and charged particle transport are given in Table 8 for the 2 inch diameter tubing (filter test system) and 0.25 inch diameter tubing (aerosol sampling tubes).

Table 8 Estimated and measured parameters used to calculate particle charge effects
Full size table

Diffusion and field particle charge states were calculated and are shown in Fig. 9. Field charging is typically dominant for larger particles (> 1 µm). The combined diffusion and field charge levels were compared to the theoretical maximum charge in Fig. 9. The estimated charge levels are 1–2 orders of magnitude less than the theoretical limit.

Fig. 9
figure 9

Particle charge associated with diffusion and field charging mechanisms

Full size image

The electric field strengths due to space charge, and due to the presence of the high voltage ionizer tips, are shown in Fig. 10, for flow inside the 2 inch diameter filter test system. This figure illustrates an important result. The calculated electric field strength due to the space charge effect was nearly three orders of magnitude lower than the electric field strength due to the presence of high voltage ionizer tips within an electrically grounded enclosure. Both electric fields will cause charged aerosol particles to migrate to the walls of the enclosure. However, from this analysis we conclude that we were likely creating significant particle losses in the region close to the ionizer tips where the electric field strength was considerably higher. Engineering solutions may be obtainable to reduce this component of the electric field and mitigate particle losses in the filter test system. Estimated electrostatic migration velocities are shown Fig. 11 along with gravitational settling velocity for particles with a density of 2.3 g cm−3 (silicon dioxide). This assumes the space charge field strength shown in Fig. 10 and particle charge, n comb, shown in Fig. 9. Electrostatic migration velocities are significant for particles between 1 and 10 nm. The electrostatic migration velocities become less significant beyond approximately 50–100 nm.

Fig. 10
figure 10

Radial component of electric field inside pipe due to space charge effect as well as field-strength in region of high voltage corona discharge

Full size image
Fig. 11
figure 11

Estimated migration velocities due to space charge effects and gravitational settling of particles with density of 2300 kg m−3

Full size image

Several approximations were used to estimate the penetration of charged particles through laboratory test systems. As a first-order estimate, the Deutsch-Anderson equation was used to assess the penetration of charged particles:

$$P = \exp \left( {\frac{{ - V_{\rm te} 2\pi RL}}{Q}} \right).$$
(26)

In (26), the electrostatic migration velocity, V te, was calculated with the space charge electric field strength at the wall of the tube (maximum value). This is a conservative estimate since the electric field is smaller closer to the axis of the tube. The Deutsch-Anderson equation was intended for use with wire-and-tube electrostatic precipitators.

A more accurate approximation was derived by Yu [35] where a dimensionless time constant was derived for the flow of charged particles through tubes. Yu’s original derivation was for particles of a single size and single charge level. Particles in our experiments had size and charge distributions. As an approximation, we modified Yu’s dimensionless time variable to separate the space charge of the entire aerosol from the charge and mechanical mobility of individual aerosol particles. The dimensionless time variable from our approximation is

$$t_{\rm j}^{*} = B_{\rm j} q_{\rm j} \frac{\rho }{{\varepsilon_{0} }}$$
(27)

where B j and q j represent the mechanical mobility of particle size j and electrostatic charge of particle size j, respectively. The first factor in (27) comes from the transport of individual charged particles within an electrical field. The second factor arises from the electrical field due to the ensemble of charges. The penetration of particle size j, P j, is approximated by the following:

$$P_{\rm j} = \frac{{t_{\rm j}^{*} }}{{1 + t_{\rm j}^{*} }}.$$
(28)

Numerical solutions are required to solve for the penetration of polydisperse aerosol in a tube. Adachi et al. [36] utilized numerical schemes to solve for two-dimensional concentration fields inside a laminar flow where an electrostatically charged, polydisperse aerosol flows. Hubbard et al. [37] also utilized computational fluid dynamics to analyze the effects of space charge for polydisperse charged aerosols. These schemes should be applied in future analyses of charged particle transport in fielded RASAs but were beyond the scope of this work. We present our approximation to Yu’s derivation with the caveat that more rigorous analysis should be performed. Our goal for this work was to assess the feasibility of scaling the particle-pre-charging approach to fielded RASAs without transport losses upstream of the filter.

Equations (26) through (28) were used to estimate the penetration efficiencies through the laboratory-scale filter test system and the 0.25 inch aerosol sampling tube used to transport aerosols from the test system to aerosol measurement instruments. Estimates are shown in Fig. 12. The cumulative length and surface area distributions are also given in the figure. Below 10 nm, a small fraction of the weighted aerosol size distribution is present. Above 10 nm, the projected penetration of particles is approximately 95%.

Fig. 12
figure 12

Charged particle penetration through laboratory-scale tubing (2 inch) and aerosol sampling tubing (0.25 inch) along with cumulative surface area distribution for atmospheric aerosol

Full size image

Based on these estimates, we believe particle-pre-charging could be achieved with minimal aerosol transport losses in the full-scale RASA. Two approaches would be used to mitigate transport losses: (1) reduce the electric field strength generated by the corona charging device, and (2) minimize the transit time between the particle-pre-charging and particle collection (filter) components of the RASA. These results also support our hypothesis that particle losses in the laboratory-scale filter test system were attributable to losses in the immediate vicinity of the ionizing tips where high electric fields existed.

Alternate flow configurations

Based on our results, ten RASA configurations are given in Table 9. The air flow rate, filter face velocity, and filter pressure drop are given in absolute values as well as relative values compared to the current RASA configuration. Weighted aerosol collection efficiencies are also given from laboratory-scale testing where collection efficiency curves were weighted by the atmospheric aerosol distribution. \(\hat{E}\) in the table is the average of \(\hat{E}_{\rm l}\) and \(\hat{E}_{\rm s}\). Relative MDC values were calculated assuming the lower limit of detection scaled with air flow rate. Engineering requirements and motivations are also given for each configuration.

Table 9 Possible RASA configurations that may result in enhanced performance
Full size table

Configuration 1 is the current RASA configuration, or the baseline from which all alternate configurations were compared. A marginal reduction in MDC could be obtained by pre-charging particles that enter the RASA system using the same filter media (configuration 2). The calculated reduction in MDC is within experimental uncertainty.

In configuration 3, FM2 could be used as the filtration material, at the current RASA air flow rate, to reduce system pressure drop and potentially reduce power consumption. Power savings were not calculated but the ratio of pressure drop, with respect to the baseline, would be approximately 70%. Without particle-pre-charging, the MDC of configuration 3 would be marginally higher than the current RASA configuration. In configuration 4, with particle-pre-charging, the reduction in pressure drop could be achieved while maintaining an equivalent MDC.

In configuration 5, FM2 at 2.5 times the baseline flow, the MDC could be driven down to approximately 70% of its current value. This approach would have twice the pressure drop of the baseline configuration. With particle pre-charging, configuration 6, FM2 could be used at 2.5 times the baseline flow where the benefit of particle-pre-charging would reduce the MDC to approximately 62% of its current value.

In configuration 7, FM3 could be used at the baseline flow at approximately 22% of the baseline pressure drop. This could result in power savings but would require the use of particle pre-charging to maintain the current MDC. FM3 could be used at 2.5 times the baseline flow with particle-pre-charging (configuration 8). With a pressure drop of approximately 73% of the baseline, energy savings may still be achieved while obtaining a MDC of 63% of the baseline value.

Conclusions

Alternate filter materials were studied for use in the U.S. Radionuclide Monitoring Stations (RMS) supporting the Comprehensive Nuclear Test Ban Treaty. Three electrostatically charged filter media were tested at higher filter face velocities where challenge aerosols were electrostatically neutral or highly charged in a corona. Two alternate filter media were found to meet RMS performance requirements for aerosol filtration efficiency with reduced filter pressure drop. Laboratory results were used to calculate the potential reduction in baseline sensitivity from collecting higher volumes of air. Results suggest that baseline sensitivities may be reduced to approximately 60% of their current values by increasing the air volume collected by a factor of 2.5. Alternate RMS configurations were analyzed based on experimental results. RMS system modifications (e.g., particle-pre-charging) were described and analyzed. The scalability of particle-pre-charging from laboratory-scale tests to full-scale fielded systems was also analyzed. Our analysis suggests that particle-pre-charging could be engineered into full-scale RMS systems to enhance particle capture.