Backup Filter Research Articles

Elimination of positive and negative sampling artifacts in particulate organic carbon and PAHs using multi-sorbent coated and uncoated denuders

Within the scope of this study, two equivalent PM2.5 samplers were designed and developed to eliminate sampling artifacts in the results of atmospheric particulate organic carbon (OC) and particulate polycyclic aromatic hydrocarbons (PAH) caused by volatile organic compounds (VOCs) and gas phase PAH compounds, respectively. A mass loss of less than 10% due to the denuders was observed. Study results showed that if an impregnated denuder is not used, the results of atmospheric particle OC concentrations will be reported with higher values due to positive errors of 53.2 ± 7.23% (median: 52.00%) on average. It was observed that the total error (net error) was still positive, but decreased to an average of 35.1 ± 16.8% (median: 31.0%) after including the negative errors quantified from the backup filter into the calculation. In cases where denuders were not used in the sampling, it was observed that the results with positive errors of 41.0 ± 14.6% (median: 33.8%) on average would be obtained for the total PAHs. Ozone-induced negative interference was the highest in Acenapthylene (28%), followed by Fluoranthane (20%), Phenanthrene (18%), and 15% for Np and Benzo[g,h,i]perylene compounds, relative to their medians. Negative errors of 10% or less were found in all other individual PAH compounds.

Long term measurements of aerosol mass concentration with optical particle counters: Discrepancies with plausible reasons

Gravimetry-based direct measurements of mass concentration require offline analysis which is not suited for field campaigns. Hence such campaigns rely on the estimation of mass concentration by indirect methods mostly calibrated in controlled laboratory conditions. Optical particle counter (OPC) employs algorithms converting the measured number concentration to mass concentration using appropriate conversion factors. The accuracy of such conversion has not been validated for widely varying atmospheric conditions. This study compares the mass concentration estimated by OPC with those directly obtained from gravimetry-based instruments for outdoor samples collected in Bathinda City, Punjab, India from January 2022 to November 2023. The difference in the gravimetrically measured and OPC predicted values quantified in terms of ratios (gravimetric to optically estimated mass concentration), came out to be 1.42 ± 0.77, 0.99 ± 0.51, and 1.17 ± 0.58 for PM10, PM2.5 and PM1, respectively. This difference when estimated with the back-up filter of OPC itself (C Factor), was 1.37 ± 0.66. More than half of the samples showed ratios outside the 0.8–1.2 range thus indicating under or over-estimation in the OPC predicted values. The probable role of variation in density, shape, and refractive index of atmospheric aerosol particles towards the observed inaccuracy of estimated mass concentration has been highlighted. In the absence of clear guidelines and protocols, the study suggests ways to improve the accuracy via periodic measurement of the C Factor and/or incorporating calibration factors in such measurements.

Two-layer partitioned and deletable deep bloom filter for large-scale membership query

The recently proposed Learned Bloom Filter (LBF) provides a new perspective on large-scale membership queries by using machine learning to replace the traditional bloom filter. However, reducing the false positive rate (FPR) of the learned model with small memory usage, and supporting deletion efficiently become the new issues. In this paper, we propose a novel Two-layer Partitioned and Deletable Deep Bloom Filter (PDDBF) for large-scale membership query, which can reduce the FPR with small memory usage and support deletion efficiently. The proposed PDDBF consists of three main parts: (1) Data partition. To improve the classification accuracy of the learned model, the K-means cluster with the elbow method is used for the data partition. (2) Deep Bloom Filter. To reduce the FPR, deep learning models are used to construct multiple independent learning mechanisms, which correspond to the clusters obtained by part1. (3) Partitioned backup filter. To support deletion under the premise of ensuring low FPR and reducing query time consumption, combine the perfect hash (PH) table and counting bloom filters (CBFs) on the basis of the partition bloom filter. Experiments show that the proposed PDDBF reduces the FPR 87.13% with the same memory usage compared with the state-of-the-art PLBF on real-world URLs data set. Moreover, the PDDBF reduces the FPR 99.68% with the same memory usage and reduces the query time consumption to 2.61x that of the PLBF after data deletion, respectively.

PA-LBF: Prefix-Based and Adaptive Learned Bloom Filter for Spatial Data

The recently proposed learned bloom filter (LBF) opens a new perspective on how to reconstruct bloom filters with machine learning. However, the LBF has a massive time cost and does not apply to multidimensional spatial data. In this paper, we propose a prefix-based and adaptive learned bloom filter (PA-LBF) for spatial data, which efficiently supports the insertion and deletion. The proposed PA-LBF is divided into three parts: (1) the prefix-based classification. The Z-order space-filling curve is used to extract data, prefix it, and classify it. (2) The adaptive learning process. The multiple independent adaptive sub-LBFs are designed to train the suffixes of data, combined with part 1, to reduce the false positive rate (FPR), query, and learning process time consumption. (3) The backup filter uses CBF. Two kinds of backup CBF are constructed to meet the situation of different insertion and deletion frequencies. Experimental results prove the validity of the theory and show that the PA-LBF reduces the FPR by 84.87%, 79.53%, and 43.01% with the same memory usage compared with the LBF on three real-world spatial datasets. Moreover, the time consumption of PA-LBF can be reduced to 5 × and 2.05 × that of the LBF on the query and learning process, respectively.

Measurements, Gas/Particle Partitioning, and Sources of Nonpolar Organic Molecular Markers at a Suburban Site in the West Yangtze River Delta, China

AbstractThis study seeks to understand the ambient levels, gas/particle partitioning, and sources of nonpolar organic molecular makers (OMMs) at a suburban site in Nanjing, China. Particle‐ and gas‐phase OMMs were collected using a medium volume sampler equipped with two quartz filters in series and a polyurethane foam (PUF)/XAD‐4/PUF (PXP) adsorbent sandwich. More than 100 pairs of filter and PXP samples were collected diurnally from September 2018 to September 2019. The concentration profiles of total n‐alkanes and polycyclic aromatic hydrocarbons were peaking at n‐C12–n‐C17 homologs (9.98 ± 5.43–38.4 ± 17.7 ng m−3) and naphthalene (92.5 ± 56.7 ng m−3), respectively, followed by decreasing concentrations with further increases in molecular weight. Three artifact correction methods with different pre‐assumptions on relative contributions of evaporation loss to backup filter measurements were used to separate particle‐ and gas‐phase OMMs. The resulting particle‐phase fractions showed significant differences for OMMs with vapor pressure <10−9 atm at 298.15 K. Absorptive gas/particle partitioning coefficients of OMMs were empirically calculated based on measurements and theoretically predicted for comparison. The results obtained under the three artifact correction methods suggested growing importance of evaporation loss with decreased vapor pressure. Positive matrix factorization (PMF) was performed for source apportionment using total (PMFtotal) and particle‐phase (PMFparticle) OMMs data, respectively. Besides eliminating the influence of gas/particle partitioning, the involvements of more volatile species and gaseous OMMs data improved the identification of specific emission sources, supporting the usage of total OMMs data over a wide vapor pressure range for PMF source apportionment.

Seasonal changes in stable carbon isotopic composition in the bulk aerosol and gas phases at a suburban site in Prague

Isotope fractionation between the gas and aerosol phases is an important phenomenon for studying atmospheric processes. Here, for the first time, seasonally resolved stable carbon isotope ratio (δ13C) values are systematically used to study phase interactions in bulk aerosol and gaseous carbonaceous samples. Seasonal variations in the δ13C of total carbon (TC; δ13CTC) and water-soluble organic carbon (WSOC; δ13CWSOC) in fine aerosol particles (PM2.5) as well as in the total carbon of part of the gas phase (TCgas; δ13CTCgas) were studied at a suburban site in Prague, Czech Republic, Central Europe. Year-round samples were collected for the main and backup filters from 14 April 2016 to 1 May 2017 every 6 days with a 48 h sampling period (n = 66). During all seasons, the highest 13C enrichment was found in WSOC, followed by particulate TC, whereas the highest 13C depletion was found in gaseous TC. We observed a clear seasonal pattern for all δ13C, with the highest values in winter (avg. δ13CTC = −25.5 ± 0.8‰, δ13CWSOC = −25.0 ± 0.7‰, δ13CTCgas = −27.7 ± 0.5‰) and the lowest values in summer (avg. δ13CTC = −27.2 ± 0.5‰, δ13CWSOC = −26.4 ± 0.3‰, δ13CTCgas = −28.9 ± 0.3‰). This study supports the existence of different aerosol sources at the site during the year. Despite the different seasonal compositions of carbonaceous aerosols, the isotope difference (Δδ13C) between δ13CTC (aerosol) and δ13CTCgas (gas phase) was similar during the seasons (year avg. 1.97 ± 0.50‰). Moreover, Δδ13C between WSOC and TC in PM2.5 showed a difference between spring and winter, but in general, these values were also similar year-round (year avg. 0.71 ± 0.37‰). During the entire period, TCgas and WSOC were the most 13C-depleted and most 13C-enriched fractions, respectively, and although the resulting difference Δ(δ13CWSOC − δ13CTCgas) was significant, it was almost invariant throughout the year (2.67 ± 0.44‰). The present study suggests that the stable carbon isotopic fractionation between the bulk aerosol and gas phases is probably not entirely dependent on the chemical composition of individual carbonaceous compounds from different sources.

A simple technique for measuring the activity size distribution of radon and thoron progeny aerosols

In this study, a portable cascade impactor was developed to more efficiently determine the activity size distribution of attached radon and thoron progeny in a natural environment. The developed impactor consisted of four stages with a backup filter stage for collection of the aerosol samples. The aerosol cut points were set for 10, 2.5, 1, and 0.5 μm at a sampling rate of 4 L min−1. Five CR-39 chips were used as alpha detectors for each stage. To separate the alpha particles emitted from radon and thoron progeny, the CR-39 detectors were covered with aluminium-vaporized Mylar films. The thickness of each film was adjusted to allow alpha particles emitted from radon and thoron progeny to reach the surface of the CR-39 detectors. The particle cut-off characteristics of each stage were determined by mono-dispersive aerosols with particle sizes ranging from 0.1 to 1.23 μm from the collection efficiency curve. The test results showed that the respective cut-off size of stages 3 and 4 were close to the designed cut-points. Validation of the technique by comparison with two commercial devices confirmed that the developed technique could provide the necessary information to estimate the activity size distribution of attached radon and thoron progeny for dose assessment, especially, in a field survey where direct electric power is not available.

Magnetic Fault–Tolerant Navigation Filter for a UAV

The heading state of an unmanned aerial vehicle (UAV) often diverges due to magnetic field disturbances, although the magnetic field can provide information about the heading. In this paper, we propose a novel magnetic fault-tolerant navigation filter (MFTNF) for UAVs, which combines a primary filter and a magnetometer-free attitude and heading reference system (AHRS) as a secondary backup filter to obtain an accurate and robust heading under an unknown magnetic disturbance. In the MFTNF framework, the primary filter that uses the magnetic field is the main filter used in normal situations for estimating the attitude, position, and velocity states. When an anomaly in the magnetic field is detected, the magnetometer-free AHRS replaces the heading of the primary filter. To detect magnetic disturbances, a novel decision-making algorithm is proposed in this paper. Flight experiments with UAVs demonstrate that the MFTNF is accurate and reliable under an unknown magnetic disturbance and that in such circumstances, the conventional approaches typically exhibit degraded accuracy in estimating the heading and the position. In addition, the advantage of the proposed MFTNF is that it always produces a robust heading estimate, regardless of the degree of magnetic interference, due to the magnetometer-free AHRS.

Stable learned bloom filters for data streams

Bloom filter and its variants are elegant space-efficient probabilistic data structures for approximate set membership queries. It has been recently shown that the space cost of Bloom filters can be significantly reduced via a combination with pre-trained machine learning models, named Learned Bloom filters (LBF). LBF eases the space requirement of a Bloom filter by undertaking part of the queries using a classifier. However, current LBF structures generally target a static member set. Their performances would inevitably decay when there is a member update on the set, while this update requirement is not uncommon for real-world data streaming applications such as duplicate item detection, malicious URL checking, and web caching. To adapt LBF to data streams, we propose the Stable Learned Bloom Filters (SLBF) which addresses the performance decay issue on intensive insertion workloads by combining classifier with updatable backup filters. Specifically, we propose two SLBF structures, Single SLBF (s-SLBF) and Grouping SLBF (g-SLBF). The theoretical analysis on these two structures shows that the expected false positive rate (FPR) of SLBF is asymptotically a constant over the insertion of new members. Extensive experiments on real-world datasets show that SLBF introduces a similar level of false negative rate (FNR) but yields a better FPR/storage trade-off compared with the state-of-the-art (non-learned) Bloom filters optimized on data streams.

Airborne particulate matter emissions from vehicle brakes in micro- and nano-scales: Morphology and chemistry by electron microscopy

Brake wear particles constitute a significant part of the non-exhaust related particulate matter (PM) associated to traffic. These particles derive from various components of the vehicles’ brake system and were recognized as an important pollution source only recently. In the present study, electron microscopy was applied to elucidate brake wear particle morphology and associated chemical composition. Sampling was carried out on a brake test bench. Particles were collected on 13 successive polycarbonate foil bags and a backup filter by means of a Dekati Low Pressure Impactor (DLPI), based on the equivalent aerodynamic particle diameter. The results revealed a broad size range from the micro-to the nano-scale, as measured on electron microscope images, including coarse (2.5–10 μm), fine (0.1–2.5 μm) and ultrafine ones (<0.1 μm), down to a few nanometers. It cannot be implied with certainty to what extent the particle size spectrum identified under the experimental conditions of the present study strictly applies to real-world conditions. The particles occur typically in form of aggregates, also in the ultrafine scale; single, i.e. non-agglomerated particles are more common in the ultrafine fraction than at larger sizes. Imaged particles have commonly rounded outlines probably due to friction and associated abrasion during braking; melting/evaporation at high temperature braking phases, associated with the rate of the generated frictional energy of braking, and subsequent rapid solidification can be invoked to interpret the nearly spherical shape of nanometer size particles. The individual constituents of the aggregates have various sizes and chemical composition. Energy dispersive X-ray analysis of the imaged aggregates and of their constituents showed that Fe is prevalent in all size fractions but is less pronounced in the smaller sizes, where Ca, occasionally in combination with S and/or P becomes more frequent. Other frequently analysed elements are Al, Sn, Mg, Si, Cr, Ti, K and W; less frequent are Ni, Zn, Zr, Ba, S, C, P, F, Mn and rarely Bi and Sb.

Concept of sediment filtration intake design for raw water drinking water

The drinking water intake capacity of drinking water is very susceptible to sedimentation, and is one of the main obstacles in drinking water services. The existing intake building is not yet equipped with a filtration system that takes into account turbidity. The level of turbidity of water entering the intake and flowing to the water treatment plant (WTP) through the transmission pipe will greatly affect the workload of the PAPs. Sedimentation problems and turbidity of raw water need special handling, starting from the design of the intake building that has the ability to dodge sediment so that the level of turbidity of the water can be minimized. Sediment evasion filtration intake designed with a 0.2 mm strainer nozzle filter system so that the sediment with a size above 0.2 mm will not enter the intake. In the cleaning process the filters installed at the intake door can be done by replacing with a backup filter, so that filter cleaning can be done easily and does not stop the intake operation for a long time. With sediment evasion filtration intakes, the raw water entering the water treatment plant (WTP) will also have a lower turbidity level so that the WTP workload is lower

Evaluation of a sample preparation procedure for total-reflection X-ray fluorescence analysis of directly collected airborne particulate matter samples

This paper investigates in detail a sample preparation procedure for the total-reflection X-ray fluorescence (TXRF) analysis of airborne particles using a three-stage Dekati™ PM10 cascade impactor. The impactor was modified in such a way, that 30 mm diameter quartz reflectors suitable for TXRF spectrometers can be inserted as collection plates below the impactor's nozzle arrays with cut-off diameters of 10, 2.5 and 1 μm. Smaller particles are collected in a backup filter, which can be analyzed using classical energy-dispersive XRF. As an internal standard 5 ng Y were pipetted centrally onto each reflector after collection. The reflector surface has to be coated in order to reduce particle bounce-off. Two surface coatings, silicone oil and petroleum jelly (“Vaseline”), were compared in this study. Before TXRF analysis this coating layer was removed by low-temperature oxygen plasma ashing. A new approach was introduced, which involves masking of the quartz reflectors in order to remove the outer ring of the spot pattern, which consists of two concentric circles, and to produce a sample, which lies entirely inside the detectable area. Results of this approach were compared to those obtained from unmasked samples and discussed in detail. Angle scans show residue-like behavior of the samples in each stage. Samples were taken on the roof of the Atominstitut building in January 2017 on 10 days during working hours.

Simultaneous Measurements of Nanoaerosols and Radioactive Aerosols Containing the Short-lived Radon Isotopes.

The activity size distribution of the Equilibrium-Equivalent Concentration (EER) of 222Rn is one of the most important parameters for the estimation of radiation dose by inhalation of radon decay products. A series of measurements of the EER activity size distribution were performed by the screen diffusion battery in Radon-Aerosol chamber (10 m3) at the National Institute for Nuclear, Chemical, and Biological Protection (SUJCHBO). These measurements were performed at different levels of radon concentration. For this study, the Graded Screen Array Diffusion Battery (GSA DB), developed by the SUJCHBO (based on Earl Knutson and Robert F Holub design), consists of 10 screens and backup filter used to collect all particles that penetrated the screens. The measuring range of this GSA DB allows measuring the radioactive nanoaerosols in the size range from 0.5 to 100 nm. The Earl Knutson algorithm was used for EER activity size distribution evaluation. The results of EER activity size distribution were subsequently compared with the aerosol particle size distribution measured by Scanning Mobility Particle Sizer Spectrometer (SMPS 3936 N, TSI Inc., MN, USA).

A comparative analysis of chemical components and cell toxicity properties of solid and semi-volatile PM from diesel and biodiesel blend

Alternative fuel has raised increasing attention in recent years for better fuel economy and emission reduction. The biodiesel fuel is natively produced from vegetable and animal oils with large number of organic compounds with varying groups and characteristics. Particulate Matter (PM) emissions from the biodiesel blend are often complicated by the volatility of the different organic components. We here report the chemical and toxicological characterization of solid-PM and semi-volatile PM (SV-PM) from diesel and biodiesel blend (BD30) exhausts, using front and backup filter, respectively. While total PM emission was lower for BD30 in general, it contained 19% higher organic compounds (OC) in SV-PM compared to diesel, which on the other hand noted 24% more solid-PM. These organic compounds were further analyzed with OCEC thermograph to confirm the dominant existence of highly volatile OC in SV-PM. We further extracted water-soluble components from the solid- and SV-PM samples, and subjected to RAW 264.7 macrophage cell exposure. More than 2-fold increase of oxidative potential was observed in SV-PM samples over solid-PM with elevated levels of cellular reactive oxygen species (ROS) and a significant correlation with OC concentration. Instead, non-cellular assay for ROS quantification showed the change in oxidative potential is greatly influenced by solid-PM. And non-cellular ROS due to SV-PM is 2-fold less than cell-based assay. Hence, the semi-volatiles, though produced in small quantities, need great attention in assessing health hazards and could well be a scope for further research.

The thin and medium filters of the EPIC camera on-board XMM-Newton: measured performance after more than 15 years of operation

After more than 15 years of operation of the EPIC camera on board the XMM-Newton X-ray observatory, we have reviewed the status of its Thin and Medium filters. We have selected a set of Thin and Medium back-up filters among those still available in the EPIC consortium and have started a program to investigate their status by different laboratory measurements including: UV/VIS transmission, Raman scattering, X-Ray Photoelectron Spectroscopy, and Atomic Force Microscopy. Furthermore, we have investigated the status of the EPIC flight filters by performing an analysis of the optical loading in the PN offset maps to gauge variations in the optical and UV transmission. We both investigated repeated observations of single optically bright targets and performed a statistical analysis of the extent of loading versus visual magnitude at different epochs. We report the results of the measurements conducted up to now. Most notably, we find no evidence for change in the UV/VIS transmission of the back-up filters in ground tests spanning a 2 year period and we find no evidence for change in the optical transmission of the thin filter of the EPIC-pn camera from 2002 to 2012. We point out some lessons learned for the development and calibration programs of filters for X-ray detectors in future Astronomy missions.

Characterization of sampling artifacts in the measurement of carbonaceous particles using high- and low-volume samplers in Daejeon, Korea

This study investigates the influence of positive sampling artifacts on the determination of atmospheric organic carbon (OC) and elemental carbon (EC) by undenuded high-volume (HV) and low-volume (LV) samplers. The samples, which were collected in Daejeon, Korea during the spring and fall of 2014, were analyzed for OC and EC using a thermal optical transmittance (TOT) technique. Positive artifacts for OC using the undenuded LV and HV samplers were determined to be ∼12.2% and ∼7.2%, respectively. The adsorption of organic vapors on quartz filters and the analytical artifacts of thermal-optical method contributed equally to the positive artifact for the HV sampler, whereas that of the undenuded LV sampler was caused primarily by the adsorption of organic vapors on quartz filters with a relatively small contribution from the analytical artifacts of thermal-optical method. The analytical artifacts of thermal-optical method was closely related to the additional formation of pyrolyzed OC (POC) during the analysis, resulting in an underestimation of EC (HV sampler: 15.6 ± 15.6%; undenuded LV sampler: 9.1 ± 7.6%). The POC that formed during the analysis was related to the adsorption of organic vapors (in both the HV and undenuded LV samplers) and high particle loading on the quartz filters (HV sampler only). OC concentrations determined using the undenuded LV sampler and corrected using backup filter measurements correlated well with those of the denuded LV sampler (R2 = 0.98), suggesting that the backup filter correction method can be applied to particulate OC measurements using an undenuded LV sampler under conditions typical of those in Daejeon, Korea.

Speciation of the major inorganic salts in atmospheric aerosols of Beijing, China: Measurements and comparison with model

In the winter and summer of 2013–2014, we used a sampling system, which consists of annular denuder, back-up filter and thermal desorption set-up, to measure the speciation of major inorganic salts in aerosols and the associated trace gases in Beijing. This sampling system can separate volatile ammonium salts (NH4NO3 and NH4Cl) from non-volatile ammonium salts ((NH4)2SO4), as well as the non-volatile nitrate and chloride. The measurement data was used as input of a thermodynamic equilibrium model (ISORROPIA II) to investigate the gas–aerosol equilibrium characteristics. Results show that (NH4)2SO4, NH4NO3 and NH4Cl were the major inorganic salts in aerosols and mainly existed in the fine particles. The sulfate, nitrate and chloride associated with crustal ions were also important in Beijing where mineral dust concentrations were high. About 19% of sulfate in winter and 11% of sulfate in summer were associated with crustal ions and originated from heterogeneous reactions or direct emissions. The non-volatile nitrate contributed about 33% and 15% of nitrate in winter and summer, respectively. Theoretical thermodynamic equilibrium calculations for NH4NO3 and NH4Cl suggest that the gaseous precursors were sufficient to form stable volatile ammonium salts in winter, whereas the internal mixing with sulfate and crustal species were important for the formation of volatile ammonium salts in summer. The results of the thermodynamic equilibrium model reasonably agreed with the measurements of aerosols and gases, but large discrepancy existed in predicting the speciation of inorganic ammonium salts. This indicates that the assumption on crustal species in the model was important for obtaining better understanding on gas–aerosol partitioning and improving the model prediction.

Characterization of Size-Fractionated Particulate Matter and Deposition Fractions in Human Respiratory System in a Typical African City: Nairobi, Kenya

Information from elemental and mass composition of size-fractionated airborne particle matter (PM) provides insightful knowledge about their impact on human health, meteorology and climate. To attain insight into the nature of sizefractionated PM from a typical African city, samples were collected from an urban background site in Nairobi, Kenya, during the months of August and September in 2007. PM samples ranging in size from 0.06 to 16 µm aerodynamic diameter were collected on pre-weighed polycarbonate filters with 0.4 µm pore size using a nine-stage cascade impactor. Particles less than 0.06 µm were collected on a backup filter. A total of 170 samples were collected and analysed for trace elements using the Proton Induced X-Ray Emission (PIXE) technique. The analysis showed that Si, Fe and S dominated in all size ranges and displayed unimodal mass-size distribution whereas K, Cu, Zn and Pb, depicted bimodal mass-size distribution highlighting the multiplicity of their sources. To estimate human exposure to PM, deposition fractions of both the coarse and fine PM in the human respiratory system were calculated. The deposited concentration was found to be highest in the head airways region compared to the tracheobronchial and pulmonary regions.

Modeling and analysis of sampling artifacts in measurements of gas-particle partitioning of semivolatile organic contaminants using filter-sorbent samplers

Measurements of gas/particle partition coefficients for semivolatile organic compounds (SVOCs) using filter-sorbent samplers can be biased if a fraction of gas-phase mass is measured erroneously as particle-phase due to sorption of SVOC gases to the filter, or, if a fraction of particle-phase mass is measured erroneously as gas-phase due to penetration of particles into the sorbent. A fundamental mechanistic model to characterize the air sampling process with filter-sorbent samplers for SVOCs was developed and partially validated. The potential sampling artifacts associated with measurements of gas-particle partitioning were examined for 19 SVOCs. Positive sampling bias (i.e., overestimation of gas/particle partition coefficients) was observed for almost all the SVOCs. For certain compounds, the measured partition coefficient was several orders of magnitude greater than the presumed value. It was found that the sampling artifacts can be ignored when the value of log[Kf/(Kp⋅Cp,a)] is less than 7. By normalizing the model, general factors that influence the sampling artifacts were investigated. Correlations were obtained between the dimensionless time required for the gas-phase SVOCs within the filter to reach steady state (Ts,s∗) and the chemical Vp values, which can be used to estimate appropriate sampling time. The potential errors between measured and actual gas/particle partition coefficients of SVOCs as a function of sampling velocity and time were calculated and plotted for a range of SVOCs (vapor pressures: 10−8 ∼ 10−3 Pa). These plots were useful in identifying bias from the sampling in previously-completed field measurements. Penetration of particles into the sorbent may result in significant underestimation of the partition coefficient for particles in the size range between 10 nm and 2 μm. For most of the selected compounds, backup filters can be used to correct artifacts effectively. However, for some compounds with very low vapor pressure, the artifacts remained or became even larger than they were without the backup filter. Thus, the option of backup filters must be considered carefully in field measurements of the gas/particle partitioning of SVOCs. The results of this work will allow researchers to predict potential artifacts associated with SVOC gas/particle partitioning as functions of compounds, the concentration of particles, the distribution of particle sizes, sampling velocity, and sampling time.

Characterization of Volatilization of Filter-Sampled PM2.5 Semi-Volatile Inorganic Ions Using a Backup Filter and Denuders

The characteristics of the volatilization of semi-volatile inorganic ions sampled on a Teflon filter were investigated using a backup nylon filter and annular denuders. The volatilization ratio (VR) was defined as the fraction of the concentration at the backup filter and denuder to the sum of the concentrations collected at the Teflon filter, backup filter, and denuder. Particles whose aerodynamic diameters are less than or equal to 2.5 μm (PM_(2.5)) were sampled for 24 h each season from summer 2011 to spring 2012 at a site, about 35 km southeast of downtown Seoul. Increase in VR with increasing temperature was evident, but the effects of relative humidity were not clear, because seasonal variation in the two variables was correlated, and the variation in relative humidity was smaller than that in temperature. While VRs generally decreased with the amounts sampled on the Teflon filter, details of the volatilization behaviors varied among ions according to concentrations at the Teflon filter and VR. The removal of gaseous species at the upstream denuders affected VRs of semi-volatile ions by shifting the equilibrium among NH_4^+, NO_3^-, and SO_4^(2-).