Development of A Low-cost Simultaneous Low Volume Air Sampler Controlled with Sonic Venturi

This study presents the results of designing a low-cost air sampler for multi-component analysis of PM2.5. We developed a multi-channel air sampling device using sonic venturi (critical nozzle) that supplies a constant airflow rate of 16.7 L/min. The constant flow rate is achieved with a backpressure ratio of 0.8 or less with a nozzle diameter of 1.375 mm. The sampler halves the cost of comparable devices and simplifies the hardware by replacing the flow meters with critical nozzles. For quantitative evaluation, we measured the coefficient of variation (CV) of each component simultaneously collected five samples at various particulate matter concentrations. This sampler can use different types of filters such as polytetrafluoroethylene and quartz fiber filter at the same time. Meteorological conditions such as ambient temperature, relative humidity and atmospheric pressure did not seem to affect much the flow rate of the sampler. The CV of multi sample analysis for TSP and PM2.5 concentrations in all sampling periods were less than 10%. CV of multi sample analysis for most of the major inorganic elements being within the range of 15% or less shows that the precision is sufficient for reliable measurement. We believe that the low-cost multiple air sampler will be used world-widely, and it can contribute to many researchers in both developed and developing countries.


INTRODUCTION
In the past few decades, the global community has begun to understand the effects of particulate matter (PM) and its relationship to many common diseases (Bonzini et al., 2010).PM is also generated when gaseous chemical species react or interact in the atmosphere.Mainly, PM is classified as PM 10 (<10 μm; coarse) and PM 2.5 (<2.5 μm; fine) based on their size of particle diameter.PM 2.5 fraction represents more than 50% of the total mass of PM 10 (Kim et al., 2015;Lim et al., 2012).The size of particles is directly linked to health problems, in which fine particles are of greater risk since they can penetrate into lung tissues and ultimately enter the blood stream (Kumar and Gupta, 2015;Huang and Ghio, 2006;Cohen et al., 2005;Kunzli and Tager, 2005;Sharma and Agarwal, 2005).
Numerous studies have shown that inhaled particles cause serious respiratory diseases (Kim et al., 2015;Cadelis et al., 2014;Correia et al., 2013;Fang et al., 2013;Meister et al., 2012;Atkinson et al., 2010).Because of this, air quality research is prioritized by developed countries (Viana et al., 2006) and many of them are trying to reduce air pollutant emissions by using environmentally friendly and efficient practices in industry.In many developing countries, air pollution is a major problem due to the shortage of funds which does not allow the use of environment-friendly techniques for the disposal of waste and the safe handling of hazardous chemicals (Sabyrbekov and Ukueva, 2019;Sapkota and Bastola, 2017;Treesubsuntorn et al., 2017).Because of this, industries in developing countries are releasing thousands of tons of air polluting PM into the atmosphere (Mannucci and Franchini, 2017).Moreover, the major disadvantages of conventional aerosol sampling methods such as filtration are: one kind filter is not always good enough to analyze all elements or components simultaneously (Liu et al., 2015;Hinds, 1998).Multi-component analysis requires sampling with several types of filters at the same time.For example, polytetrafluoroethylene (PTFE) is suitable for elemental analysis whereas quartz fiber filter (QFF) is required to analyze carbonaceous species.In this case, the aerosol sampling needs several devices.Measurement and regular monitoring of fine PM in developing countries is extremely important to reduce emissions in the air so that one can understand the sources and formation mechanisms of PM; however, many of the countries with poor economics would not afford to purchase many of highly accurate but expensive sampling analysis devices.
A major aim of the study is to develop a low-cost air sampler to obtain multiple samples at the same time.The multiple samples would allow researchers to carry out a variety of chemical analysis that are requested to conduct multi-component analysis of air pollutants so that one can understand these sources and formation mechanisms.A multiple air sampler is ideal when sampling under identical conditions of temperature and humidity, and over the same period to conduct multicomponent analysis of air compositions.Multiple air samplers also require a stable air flow rate to obtain the same sample.However, it is difficult to collect samples under the same conditions using multiple air samplers due to the cost of expensive flow meters and flow control devices.Critical nozzles give advantages such as minimal upstream piping, long term accuracy of flow rate, no mechanical parts, and excellent flow rate repeatability due to unaffected by downstream flow disturbances.Various devices with critical nozzles have been developed to measure low gas flow rates in a calibration facility (Nakao et al., 1996), and Wang et al. (1999) presented the multipoint air sampling using a critical nozzle to measure aerosol spatial distribution at multiple points in a room.The sampler, however, is not directed to study outdoor PM 2.5 , instead, it intends to measure indoor aerosol.Besides, there have not been achieved the verification that provided the information of uniformity of the multiple samples collected using those samplers by analyzing the chemical compositions of the obtained samples.
In this paper, we present the design and lab evaluation of a low-cost and simultaneous low volume air sampler controlled with sonic venturi to obtain multiple homogeneous samples at the same time to conduct multi-component analysis of aerosol particles.The sampler allows to obtain multiple homogeneous and heterogeneous (in terms of filter materials) samples at the same time.Also, the device reduces the repairing and operation (running) cost on account of very simple construction.This device will be used extensively in our field researches in various developing locations such as the Central Asian countries.

1 Simultaneous Low Volume Air Sampler
Development Schematic drawing and a picture of the constant flow multiple sampler are shown in Fig. 1a and 1b.It consists of impactors to remove particles greater than 2.5 μm of aerodynamic diameter, a filter for collecting PM 2.5 , a criti cal nozzle for fixing the flow rate, the flow control valve, and a pump.This sampler is named "SilVy-5", a shortening of "Simultaneous low Volume air sampler controlled with Sonic Venturi" and it has five sampling lines.The number of samples is not limited to be five, it can be variable depending on the situation.
Multiple homogeneous and single blank filter holders are connected to one outlet line.Each sampling device includes an impactor for removing coarse particles and a filter for collecting fine particles.The two-stage impactor, which used round jets, was designed for collecting PM 2.5 at an air flow rate of 16.7 L/min, based on the design in ISO (International Organization for Standardization) and JIS ( Japanese Industrial Standards) ( JIS Z 8767:2006;ISO 9300:2005).The five filters are provi ded with an equal airflow rate and the blank filter does not have air flow through it to compare it with the other filters.

1 Theory
Measurement of the airflow rate is an important issue when using a multiple air filter sampler because the concentration of the air pollutant is determined by the ratio of the sampled components quantity to the sampled air volume.Using a venturi sound nozzle to control airflow rate is considered to be a method that can solve this issue and work long term, after calibration and metering of gas flow meters.It is accurate, simply construction, and inexpensive.
According to the flow continuity equation ( 1), the flow rate Q is constant, but, the flow velocity u changes depending on the cross-section area S ( = πd 2 /4) of the venturi sound nozzle portions (Fig. 2): When the pressure ratio between the upstream (P 1 ) and downstream (P 2 ) sides of the nozzle is kept below the critical backpressure ratio (P c ), the flow velocity u at the throat is fixed at the sonic speed a, and the flow rate is constant regardless of the downstream state of the nozzle (Nakamura et al., 2014;Morioka et al., 2011;Wang and Zhang, 1999;Johnson, 1965).P c is the pressure ratio between critical pressure (P cr ) at the throat of the nozzle and downstream (P 2 ) sides of the nozzle.The flow velocity u at the throat can be expressed as (Wang and Zhang, 1999).(2) where P th -the pressure at the throat (Pa); ρ 1 -air density (kg/m 3 ); γ -a specific heat ratio of the gas.Here, γ = 1.4.In this case, the constant flow rate depends on a constant upstream condition.The flow velocity is fixed at the sonic speed at the throat portion of the critical nozzle due to the properties of the gas flow at the throat of the nozzle.
The theoretical flow rate Q th passing through the venturi sound nozzle is expressed by the equation (3) based on the flow state quantity in the throat of the nozzle (Morioka et al., 2011):   (Morioka et al., 2011;JIS Z 8767:2006).
where S th -the throat cross section area (m 2 ); a* -the sound speed at the throat (m/s).
The speed of sound depends on the heat capacity ratio of flowing phase (e.g.Morioka et al., 2011;Younglove and Frederick, 1992;Wong, 1990).The heat capacity ratio expresses the effect of changing temperature, pressure, volume, density, etc. on the properties of the gas or air phases.Younglove and Frederick (1992) measured the speed of sound in the vapor and supercritical phases of air.The results showed that with increasing pressure (in the range 0-4 MPa) the speed of sound decreases, but with increasing temperature the speed of sound increases.The equation (4) holds between the sound speed a 0 at the stagnation point upstream of the nozzle and the sound speed a* at the throat portion.where γ -a specific heat ratio of the gas.Here, γ = 1.4; a 0 -the sound speed in air at atmospheric pressure and room temperature, it is 340 m/s (Amrani, 2013).
The throat portion is in a state of higher pressure, in which the density of compressed air is high, in contrast to the normal state.In a flow path in which the cross-sectional area of the critical nozzle continuously increases or decreases, the flow cannot reach the sonic speed (critical state) in the middle of the flow path.In order for a flow to be sonic in a flow channel, there can only be a subsonic-sonic-supersonic configuration or vice versa.

2 Design New Nozzle to Control Flow Rate for
Environmental Air Sampling In this study, instead of controlling the flow rate with a flow meter, we manufactured a critical nozzle that fixes the flow rate at 16.7 L/min.The shape of the critical nozzle is based on ISO and JIS ( JIS Z 8767:2006;ISO 9300:2005).The standard drafting is as shown in Fig. 2. Since each dimension of the critical nozzle depends on the value of the throat diameter d, the critical nozzle shape can be determined as long as the throat diameter is known.The nozzle was manufactured with a throat diameter between 1.30≤ d≤1.38.The experiments were conducted to determine the impacts of meteorological conditions (temperature and humidity) and filters materials (Polytetrafluoroethylene (Cytiva's Whatman 7592-104, PM 2.5 PTFE membrane filter, 46.2 mm, pore size 2 μm), polycarbonate (Toyo Roshi Kaisha, Ltd, Advantec K020A047A membrane filter, 47 mm, pore size 0.2 μm), quartz fiber (Toyo Roshi Kaisha, Ltd, Advantec QR-100, 47 mm, Collection Efficiency 99.99% for 0.3 μm DOP), glass fiber (Toyo Roshi Kaisha, Ltd, Advantec GA-100, 47 mm, Collection Efficiency 96% for 0.3 μm DOP), and nitrocellulose (MF-Millipore AAWP04700 membrane filter, 47 mm, pore size 0.8 μm), on the flow rate.
The experimental system for setting the fixed flow rate is shown in Fig. 3.An experiment was conducted to confirm the backpressure ratio of all manufactured critical nozzles.When the valve was opened, the suction power of the pump gradually increased, and the pressure on the outlet side of the nozzle decreased to confirm the backpressure coefficient (backpressure coefficient), at which the flow rate became constant.

3. 1 Sampling
Five TSP (total suspended particles) and PM 2.5 samples were simultaneously collected using the SilVy-5.The control (blank) sample was also set into another fil- ter holder without any air flow through the filter.TSP and PM 2.5 were collected on PTFE filters (same product for PM 2.5 sampling).We used impactors by placing a donut filter between the nozzle plate and impaction plate to collect particles larger than 2.5 μm, which let PM 2.5 particles be collected on the sampling filters.
The sampler was located on the laboratory balcony on the 5th floor of a building at Yagami campus of Keio University in Yokohama, Japan.TSP was also collected during 72 hours in the following three sampling periods: 2020/08/21-24, 2020/08/24-27, 2020/08/28-31 and 24 hours in three sampling periods: 2020/09/02-03, 2020/09/03-04, and 2020/09/14-15.PM 2.5 was collected during 72 hours in the following three sampling periods: 2019/11/25-28, 2019/11/29-12/2, and 2019/12/8-11.The air flows rate for TSP and PM 2.5 samplings was fixed at 16.7 L/min through each A, B, C, D, and E impactors and 0 L/min for the control.The flow rate (with impactor, but without filters) was checked before and after sampling, and it was confirmed that the flow rate was always fixed at 16.7±0.1L/min at all the inlets.The backpressure ratio was adjusted using the bypass so that it was always maintained at 0.70±0.04,which was below the critical backpressure ratio.

3. 2 Determine of Mass Concentration by
Weighing Method Mass concentration was determined by following the official method ( JIS Z 8851:2008).Weighing was carried out by electric balance Sartorius ME235S (resolution d = 0.01 mg) in room temperature 24±3°C, and relative humidity <35%.Arithmetic mean of field blank (n = 3-5) was subtracted from sample weight.

3. 3 Elemental Analysis by X-ray Fluorescence
Analysis (XRF) The collected filter samples were placed in a clean polystyrene case, sealed in a polyethylene bag, stored refrigerated (4°C or below), and analyzed within one week after the samples were collected.The samples were subjected to mass concentration measurement and inorganic element analysis using Energy dispersive X-ray fluorescence analysis (EDXRF) (EDXL300, manufactured by Rigaku Corporation) (Okuda et al., 2014).EDXL300 allows the users to achieve standardless analysis that we do not need to prepare standard materials of each target element due to the device has a powerful fundamental parameter (FP) algorithm.In FP method, many parameters are consi-dered in the calculations such as the interaction and absor ption between photons and atoms, the thickness of the atomic layer, the elemental compositions.SRM2783 (Air Particulate on Filter Media, provided by NIST) was analyzed daily to check the instrument condition.

1 Constant Flow Rate Performance of the Critical Nozzles
The critical backpressure coefficient P c changes according to the Reynolds number, which greatly depends on the shape of the critical nozzle, the viscosity of the fluid, and a value unique to the nozzle.Therefore, the value P c is an important variable that must be confirmed experimentally.Before a critical nozzle is used for controlling the flow rate, the critical backpressure coefficient needs to be checked to confirm that the airflow in the nozzle transitions to a critical state.When the air is drawn through a filter, the flow rate will be constant, if the pressure ratio between the upstream (P 1 ) and downstream (P 2 ) sides of the nozzle is kept below the critical backpressure ratio (P c ).When the backpressure ratio (P 2 /P 1 ) becomes 0.8 or less, the flow rate is fixed and steady.
In this study, instead of controlling the flow rate with a flow meter, we used a critical nozzle that fixes the flow rate at 16.7 L/min.Controlling the flow rate by critical nozzle heavily depends on the throat diameter of the critical nozzle and the backpressure of the flow.We conducted experiments to confirm the constant flow rate of air in the throat diameter range of 1.30 mm to 1.38 mm.The results of the experiments are shown in Figs. 4 and 5.In the range of the aforementioned nozzle throat diameters, it was found that a constant flow rate fluctua ted from 14.8 L/min to 16.9 L/min.In this case, when the back pre ssure ratio becomes 0.8 or less, the flow rate is stable.An unstable flow rate was observed when the back-pressure ratio was higher than 0.8 for all nozzle diameters.Fig. 5 shows the relationship between the throat diameter of the critical nozzles and the fixed flow rate.Here, the theoretical value of the fixed flow rate equals 310 m/s at the throat portion at room temperature in the cross-sectional area.From the experiment, it was found that when the throat diameter d was in the range of 1.30≤d≤1.38, the relationship between the throat diameter and the fixed flow rate was approximated by y = 8.35 x 2.17 , which was 0.65 times lower than the theoretical value.The discrepancy between the theoretical value and the experimental value was quite large.It was questioned whether the flow velocity at the throat in the critical nozzles did not reach sonic speed.Therefore, in this case, it is necessary to deepen the discussion quantitatively on the fixed flow rate at the throat.
Assuming that the theoretical value of the speed of sound at a fixed flow rate holds, the boundary layer that develops inside the actual nozzle is considered to be the cause of the difference between the theoretical value and the experimental value.When the throat portion is compressed, the air density reaches its maximum.Therefore, the flow velocity at the throat is fixed at the sonic speed at the throat portion of the critical nozzle due to the nature of the gas flow, and the flow rate remains constant regardless of the downstream state of the nozzle.In this case, the constant flow rate depends on a constant upstream condition.The speed of sound changes depending on the flow density, that is to say, when flow density decreases, the speed of sound increases.
Table 1 summarizes the throat diameter, cross-sectional area, effective area (experimental value of fixed flow rate/sonic speed), thickness of the boundary layer, and ratio of the boundary layer for each throat diameter.The ratio of the boundary layer to the throat diameter is about 26±1%, which agreed well with a simulation result reported by Alam et al. (2016).Therefore, it is determined that the critical nozzle we used has reached the sound speed at the throat portion.
According to experimental results, the constant air flow rate at 16.7 L/min requires a backpressure ratio of 0.8 or less with a nozzle diameter of 1.375 mm.A critical nozzle with the same property as shown in Fig. 6 was prepared and used in the experiment.
Studying the effect of meteorological conditions on  the flow rate through the passed nozzle is most important because, in many regions, the temperature and humidity vary with rapid differences.Fig. 7 represents the relationship between the flow rate and weather conditions.The flow rates of sampling lines under the meteorological conditions of ambient temperature ranged from a minimum of 16.0°C to a maximum of 32.7°C and the humidity ranged from 32% to 88%.The maximum flow rate was observed 17.3 L/min at 16.5°C and the minimum flow rate was 16.7 L/min at atmospheric temperatures ranging from 25°C to 32.7°C.The flow rate was 17.1 L/min at 16.0°C (and 1017.1 hPa) and 17.3 L/min at 16.5°C (and 1021.5 hPa) possibly due to the higher upstream (atmospheric) pressure.Atmospheric pressure varied from 1006 hPa to 1021.5 hPa in measuring the flow rate.When the temperatures were almost equal, the pressure changed the flow rate to 0.1-0.2L/min.During experiments, the exact correlation was not observed between the flow rate and the atmospheric pressure because the variation of the atmospheric pressure was low.The humidity did not significantly affect the flow rate.The experiments showed that high temperature slightly reduces the flow rate, but the relationship is not significant.It means that the flow rate is almost constant at the observed range of ambient air temperature.The impacts of temperature may depend on the compressibility of the air passing passes through the nozzle throat and the pressure increases.In the past study (Johnson, 1965), the compressibility is given as a function of density and temperature for air and steam, it is necessary to develop the mass-flow-rate solution.There is defined the pressure-density-temperature relation is given by ( 5)  The ideal-gas critical-flow factor defined by (6) The equation ( 5) shows that the upstream pressure (P 1 ) changes the flow rate.The flow resistance of filters reduces the upstream pressure between the filter and nozzle.Fig. 8 shows the experimental results for determining the flow rate resistances of different filter materials.The resistance decreased the flow rate passing through nozzles up to 14.5 L/min.The highest value of pressure loss passing through filters belongs to the polycarbonate filter and, the smallest belongs to the PTFE filter.When user want to keep the flow rate in the ±5% range for 16.7 L/min (15.9-17.5 L/min), one can use the filters made of PTFE, quartz or glass fiber.
It is known that the penetration of the impactors changes related to the flow rate (Okuda et al., 2015).Fig. 9 shows the results of experiments to determine the 50% cut-off diameter for the impactor (range, 12 to 20 L/min) and the USEPA WINS PM 2.5 impactor (at 16.7 L/min).The 50% cut-point of the impactor for ambient aerosols at an air flow rate of 14.5-18.0L/min meets the JIS Z8851:2008 (2.5±0.2 μm), and that for WINS PM 2.5 impactor at 16.7 L/min was 2.46 μm.Therefore, PM 2.5 separation was practically achieved well for all types of filters examined in this study.

2 Evaluation of SilVy-5 by TSP and PM 2.5
Sample Collection Performance Evaluation of SilVy-5 has been carried out by calculating mass loading values on the filter during sampling periods and element concentration analyzing of TSP and PM 2.5 samples.TSP samples were collected in the 2020/08/21-24, 2020/08/24-27, and 2020/08/28-31 periods for 72 hours.The concentrations of TSP in the air were 19.4±0.5 μg/m 3 , 21.7±0.3μg/m 3 , and 14.4± 0.2 μg/m 3 , respectively.TSP samples were also collected in the 2020/09/02-03, 2020/09/03-04, and 2020/ 09/14-15 periods for 24 hours.The concentrations of TSP in the air were 8.1±0.4 μg/m 3 , 5.8±0.3μg/m 3 , and 6.2±0.3 μg/m 3 , respectively.Table 2 shows that CV for TSP concentrations are from 1.5 to 2.5% for 72 hours sampling, and from 4.9 to 5.2% for 24 hours sampling, respectively.The CV for TSP concentrations in all sampling periods were less than 15% of the target value,  which recommended by Ministry of the Environment, Government of Japan (MOEGJ, 2019).PM 2.5 samples collected in the 2019/11/25-28, 2019/11/29-12/2, and 2019/12/8-11 sampling periods for 72 hours, the concentrations of PM 2.5 in the air were 6.6±0.4 μg/m 3 , 10.4±0.3 μg/m 3 , 9.8±0.8μg/m 3 , respectively.The images of the blank and sample filters of PM 2.5 are shown in Fig. 10, and element analysis results for PM 2.5 and TSP are shown in Fig. 11.Table 2 shows that the CVs for PM 2.5 concentrations were between 3.1% and 8.3% that  were less than 15%.The reference values in the 2020/08/21-24, 2020/08/24-27, and 2020/08/28-31 periods were 7.7 μg/m 3 , 11.4 μg/m 3 , and 9.5 μg/m 3 , respectively.In the data displayed in Table 2, it can be seen that the difference between the concentrations of PM 2.5 and the reference values were between 3% and 14%.In the sampling periods, the PM 2.5 concentration values were obtained for Nakahara Ward Office, Kawasaki City Government (located about 2.2 km from the sam pling location) (NWOKCG, 2020).Fig. 12 show the coefficient of variation (CV) for each average elemental mass loading values in all filters.For quantitative evaluation of three sampling periods, CV of each component concentration of five samples of TSP and PM 2.5 was calculated and the variation is summarized in Table 3.The inorganic element components analysis of TSP and PM 2.5 showed that the major components are Mg, Al, Si, S, Cl, K, Ca, Fe, and Zn.The mass  loading values on filter and concentrations of the components in the air are higher than 100 ng/cm 2 and 20 ng/ m3, respectively.Also, in the comparison of the samples obtained over the three sampling periods for TSP, and PM 2.5 the composition of inorganic elements were quite similar.
In general, it is said that desirable deviation from arithmetic mean for duplicate analysis is within ±15% (MOEGJ, 2019).In the collected samples of TSP, the CVs of Mg, Si, S, K, Mn, Ni, and Zn were less than 15%, Cl, Ca, Ti, Cr, and Fe were between 15-20% and Al, P, V, and Cu were greater than 20% in 2020/09/02-03 sampling period, all the other elements besides Cl, Cr, and Ni was less than 15% in 2020/09/03-04 and only V, Cr, Ni exceed 15% in 2020/09/14-15.Similarly, the CVs for mass loading value on filter of Mg, Al, S, Cl, K, Ca, Mn, Fe, Ni, Cu and Zn were less than 15% for a concentration larger than 30 ng/cm 2 in the three sampling periods for PM 2.5 , CV for V was 16.1% in 2019/11/25-28 and CV for Ti, V, Cr were between 16.1-29% for a concentration lower than 50 ng/cm 2 in 2019/11/29-12/2.The CV of Si in the PM 2.5 samples collected in the three sampling  periods exceeded 20%.In this case, the CV of Si in the TSP samples in Table 3a confirmed that a sudden particle resuspension of large soil particles including Si might be collected on the filter.By excluding this accidental case, it was possible to collect five filter samples of PM 2.5 at the same quality in each sampling period, with CV of the elemental components with an air concentration of 50 ng/cm 2 or more falling within the range of about 15%  or less in this study.In Table 3, the inorganic element components analysis results of the blank filters were shown.The slight excess of the CVs of Al, P, Ca, Ti, Cr, Fe, Ni, Cu in the TSP and PM 2.5 samples might be induced by large variation of the mass loading values in the blank filters of these elements, which were similar level of the concentration in the collected samples.Table 4 shows certified and measured values for the standard material SRM2783 air particulate matter on filter.CV of repeated analysis of SRM2783 was less than 15% for Mg, Al, Si, P, S, Cl, K, Ca, Ti, Fe, Cr, Mn, Ni, Cu and Zn, and 49% for V.The EDXRF-FP method showed  good repeatability that was less than 15% of the target value except V.The deviation of four measured values from the certified value (Relative standard deviation) for SRM2783 was less than 15% for Al, Si, Cr, Fe and Zn, from 15% to 20% for K, Ca, Ti and Mn and greater than 30% for Mg, V, Ni, Cu and S. The concentrations of all other elements besides V in the SRM2783 sample were greater than 10 ng/cm 2 for the certified and measured values.Fig. 12 shows the relationship between the average mass loading values on the filters and CV of simultaneously corrected PM 2.5 and TSP samples (n = 5) of each component.It is shown that the lower elemental mass loadings, the higher CV.In all EDXRF analyses, the X-ray from its source irradiates a sample, and detector counts (measures) incoming photons (radiation) to detect a peak of an element.A high concentration of an element emits more photons than a low concentration.Collecting more photons and hence increase the precision of detection.Also, the presence and concentrations of other elements influence the precision.The elements attenuate or enhance the number of emitted photons.Heavy elements absorb the greater part of the incoming radiation and the fluorescent radiation than light ones (Brouwer, 2010).However, ICP-MS or other higher sensitivity devices can further improve the measurement precision of each element.

3 Cost Calculation
The most important factors of the sampler for the wide and successful application are high measurement accuracy, low-cost and ease to use.The high costs of sampling devices of PM 2.5 in locations in developing countries limit comprehensive study of local and global air pollu-tion.This has spurred researchers to develop low-cost samplers.We have created a simultaneous low volume air sampler using a critical nozzle.The sampler reduces the cost and simplifies hardware by replacing the flow meters with critical nozzles.The approximate cost of SilVy-5 produced in this study and compares it with a similar sampler using traditional flow meters (Table 5).According to the estimation, the manufacturing cost of SilVy-5 can be halved in compared to a sampler with similar performance that uses ordinary flow meters.Therefore, this is a great step in forwarding the ability to measure air quality in developing countries.The running cost of the sampler estimates pump working costs (electricity) the capital cost is very low due to the simple construction.For example, SilVy-5 never needs fixing or replacing the flowmeters or valves since it has not them at all.Moreover, optional costs, the calibrations and maintenances would not be required at beginning and during working processes.

CONCLUSIONS
We developed a low-cost multiple air sampling device to use in both developed and developing region such as the Central Asian countries.We use sonic venturi air flow devices to collect multiple PM 2.5 samples that can allow researchers to conduct multi-component analysis of aerosol compositions.According to experimental results, the constant airflow rate at 16.7 L/min was set with a backpressure ratio of 0.8 or less with a nozzle diameter of 1.375 mm.The sampler uses fewer mechanical devices and cuts costs by about half and simplifies the design by replacing the flow meters with critical nozzles.Multiple samples were collected for 72 hours for TSP and PM 2.5 , and 24 hours for three times during three sampling periods to evaluate collection performance of sampler.For quantitative evaluation, CV of each component concentration of five samples was calculated and the variation was summarized.The CV of multi sample analysis for TSP and PM 2.5 concentrations in all sampling periods were less than 10%.CV of the major inorganic elements being within the range of 15% or less shows that the precision is sufficient for sampling for practical use.We believe that the low-cost multiple air sampler will be used world-widely, and it can contribute to many researchers in both developed and developing countries.

Fig. 1 .
Fig. 1.Schematic drawing (a) and picture (b) of simultaneous low volume air sampler (SilVy-5).P is the pressure measurement gauge and N is the nozzle.

Fig. 3 .
Fig. 3. Schematic diagram of the experimental setup for confirming the back-pressure ratio of the nozzle.Mass flow meter: AZbil CMS0050 calibrated at 20°C and 1 atm.Pump: ULVAC diaphragm dry vacuum pump DA-120S.P 1 and P 2 are the pressures before (upstream pressure) and after (downstream pressure) the nozzle, respectively.

Fig. 4 .
Fig. 4. Relationship between backpressure ratio and flow rate for the different nozzle diameters.

Fig. 5 .
Fig. 5. Relationship between throat diameter and fixed flow rate.
the ideal-gas critical-flow factor; P o -the stagnation pressure (Pa); R -ideal gas constant; T o -stagnation temperature (K).

Fig. 8 .
Fig. 8.The impacts of filter materials on the flow rate through passing the nozzle.

Fig. 9 .
Fig. 9. Penetration curves for the impactor at different flow rate, and for the USEPA WINS PM 2.5 impactor (@16.7 L/min).

Fig. 12 .
Fig. 12. Relationship between the average mass loading values on the filters and CV of simultaneously corrected TSP and PM 2.5 samples(n = 5) of each component.

Table 1 .
Cross-sectional area, boundary layer thickness and percentage of boundary layer for each throat diameter.

Table 2 .
Measured TSP and PM 2.5 sample mass concentration and reference value (μg/m 3 ).

Table 3a .
The mass loading values on the filter and CV for each elements of TSP.The target of CV value is fixed 15% (MOEGJ, 2019).

Table 3b .
The mass loading values on the filter and CV for each elements of PM 2.5 .The target of CV value is fixed 15% (MOEGJ, 2019).

Table 5 .
The approximate cost of SilVy-5 and similar sampler using flow meters instead of the critical nozzles.