AEROSOL SAMPLINGMETHODS FOR WIDEAREA ENVIRONMENTALSAMPLING (WAES)Finnish support to IAEA

Valmari T1, Tarvainen M1, Lehtinen J2, Rosenberg R3,Honkamaa T1, Ossintsev A4, Lehtimäki M5, Taipale A5,Ylätalo S1, Zilliacus R3

1 Radiation and Nuclear Safety Authority (STUK)

2 SENYA Ltd., Rekitie 7 A, 00950 Helsinki, Finland

3 VTT Processes, P.O. Box 1404, FIN-02044 VTT, Finland

4 Institute of Radiation Safety and Ecology of the National NuclearCenter of the Republic of Kazakhstan (IRSE), Kurchatov, Kazakhstan

5 VTT Industrial Systems, P.O. Box 1307, FIN-33101 Tampere, Finland

STUK-YTO-TR 183 / J U N E 2 0 0 2

STUK • SÄTEILYTURVAKESKUSSTRÅLSÄKERHETSCENTRALEN

RADIATION AND NUCLEAR SAFETY AUTHORITY

Osoite/Address • Laippatie 4, 00880 HelsinkiPostiosoite / Postal address • PL / P.O.Box 14, FIN-00881 Helsinki, FINLANDPuh./Tel. (09) 759 881, +358 9 759 881 • Fax (09) 759 88 500, +358 9 759 88 500 • www.stuk.fi

ISBN 951-712 -560 -7 (p r in t )ISBN 951-712 -561 -5 (pdf )ISSN 0785-9325

Tummavuoren K i r j apa ino Oy, Vantaa 2002

S T U K - Y T O - T R 1 8 3

3

Abstract

Enrichment of uranium or reprocessing of nuclear fuel are expected to produce releases ofnon-natural radionuclides, that are carried away in air attached in airborne particles.Wide Area Environmental Sampling (WAES)-method utilises air samplers distributed inthe monitored area to detect the possible releases. For reduction of expenses, the samplerslocated in remote areas should be able to operate long periods of time unattended. In thiswork, a high-volume (flow rate 150 m3/h) air sampler with an automatic filter changingsystem, Hunter MKII, was developed for WAES. The sampler can collect six one-weeksamples until it needs to be visited for unloading the used filters and loading the newones. The device sends real-time state-of-health information to headquarters so that long-time loss of sampling can be avoided in case of malfunction. The state-of-health data alsoincludes indication to prevent inconspicuous tampering of the unattended samplingprocess.

Organic filter materials are used to collect particles due to their applicability to radio-chemical analysis. Four filter materials were tested for collection efficiency and pressuredrop. The material selected for current use (Petrianov FPP-15-1.5 used as two-layers oneupon the other) can collect more than 90% of the 0.2 µm particles throughout the samplingperiod. If there is a large concentration of coarse particles in air (as is typically the case indesert conditions), the filter clogging rate can be significantly decreased by preceding itwith a low-pressure-drop pre-filter that collects the coarse particles. The filter pressuredrop is low enough to easily allow one-week sampling time in typical sampling conditions(a pre-filter may be needed in heavily dust laden desert air).

VALMARI Tuomas, TARVAINEN Matti, LEHTINEN Jukka, ROSENBERG Rolf, HONKAMAA Tapani,OSSINTSEV Alexandr, LEHTIMÄKI Matti, TAIPALE Aimo, YLÄTALO Sampo, ZILLIACUS Riitta.Aerosol sampling methods for wide area environmental sampling (WAES). Finnish support to IAEA.STUK-YTO-TR 183. Helsinki 2002. 19 pp + Appendix 1 pp.

Keywords: WAES, safeguards, aerosol sampling, high-volume air sampler

4

S T U K - Y T O - T R 1 8 3

Contents

ABSTRACT 3

1 INTRODUCTION 5

2 DESCRIPTION OF HUNTER AUTOMATIC MKII 62.1 Air sampling 62.2 State-of-health data transmission 7

3 CHOOSING THE FILTER MATERIAL 83.1 Criteria 83.2 Concentration and size distribution of the atmospheric aerosol 93.3 Filter tests 9

4 PRE-FILTRATION OF COARSE PARTICLES 134.1 Motivation 134.2 Methods 134.3 Tests 134.4 Simulation of air sampler flow rate 144.5 Practical experience from field trial in Kazakhstan 16

5 CONCLUSIONS 17

ACKNOWLEDGEMENTS 18

REFERENCES 19

APPENDIX EXAMPLE OF THE SAMPLER OPERATION PROCEDURE 20

S T U K - Y T O - T R 1 8 3

5

The international Wide Area Environmental Sam-pling (WAES) working group considers aerosolsampling as a potential method for detecting un-declared nuclear activity [1]. Enrichment of urani-um and reprocessing of nuclear fuel are expectedto produce releases of non-natural radionuclides,that are carried away by air flows. The radionu-clides, with an exception of noble gases and thegaseous fraction of iodine, are attached in airborneparticles.

Airborne particles can be collected by an airsampler that aspirates air through a filter. Parti-cles are retained on the filter. The air flow ratemust be large in nuclear applications, because thedetection limit with many analytical methods,such as gammaspectrometry, is the better thelarger the sample is. Usually the air flow ratewhen detecting radionuclides in air is of order of100–1000 m3/h. The period between filter chang-ing typically varies from 1 to 7 days.

WAES-method utilises a large number of sam-plers located in an area of hundreds of squarekilometers. Fine aerosol particles (diameter small-er than one micrometer) can be carried in airthousands of kilometers from the source. However,it is important to locate samplers close to thesource, since the concentration of the non-naturalnuclides is decreased as the distance from thesource is increased. A sufficiently dense networkof samplers is important especially for the locali-sation of the source. For reduction of expences, thesamplers located in remote areas should be oper-ated long periods of time unattended. Therefore,devices that carry out filter changing automatical-ly are preferred.

An automatic air sampler, Hunter MKII, devel-oped for WAES is described in this report. The

1 Introduction

device is equipped with 6 filters. Each filter col-lects a one-week sample, thus the sampler needsto be visited once every six weeks. The devicesends real-time state-of-health data to the head-quarters. Long-time losses of sampling due tomalfunction or interruption in sampler use can beavoided by real-time information of the samplerstatus, as an inspector can be sent to the site whenneeded. Also, the state-of-health data includesindication to prevent inconspicuous tampering ofthe unattended sampling process.

The Finnish Nuclear Verification organisation(FINUVE) performed an aerosol sampling cam-paign in Iraq during the autumn 1998. One of thefindings was that the large concentration of air-borne dust in the Baghdad area resulted in filterclogging in a few days. During that time, some10000 m3 of air was aspirated through a 500 cm2

filter, or 20 m3 of air through 1 cm2 of filter. Indevelopment of the new device, a special attentionhas been given to the filter sampling process toenable longer sampling periods. The filter materi-al has been changed, filter collection area has beendoubled, and a pre-filtration technique has beenimplemented to reduce the filter clogging-rate bycollecting the coarsest particles (that are usuallynot interesting) separately.

The device is described in section 2. The crite-ria and the tests for filter material selection arepresented in section 3. The optional pre-filtrationtechnique for extending the filter collection-timein desert conditions is discussed and test resultsshown in section 4. Conclusions from the testingare drawn in section 5. Development of filtersample analysis methods has been carried out inconnection with the work described here. Resultsare reported in reference [2].

6

S T U K - Y T O - T R 1 8 3

The new air sampler, Hunter Automatic MKII, wasdeveloped based on the experience from using astandard-model Hunter and Hunter MKI (SenyaLtd.) in Iraq 1998. The design specifications forthe Hunter Automatic MKII were:• Six filter samples can be collected automatical-

ly. Automatic data storing and transmission.• Air flow rate at least 150 m3/h (expressed in

Standard Temperature and Pressure, STP),when the filter is clean.

• Sampling time 3 or 7 days per filter, totalunattended operation time 18 or 42 days.

• Filter composition adequate for chemical anal-ysis. The filter must be easily commerciallyavailable.

• Particle collection efficiency: Sampler shouldcollect ≥ 80% of particles with a diameter0.2 µm, and ≥ 60% of particles with a diameter10 µm. The former requirement constricts theproperties of the filter, and the latter the prop-erties of the rain shield upstream of the filter.

• Automatic data handling: Data logger with asufficient capacity to calculate and store data.

• Can be operated in a temperature range of–30°C…+50°C.

• Power supply 230V / 50Hz, power less than1.5 kW, employs single phase electricity.

2.1Air samplingAir is aspirated by a gas ring vacuum pump. Theair flow rate, Q, depends on the pressure dropcaused by the filter, ∆p, approximately as

Q = 203 – 0.008 ∆p, (1)

where Q is in m3/h, and ∆p is in Pa. A flow ratelarger than 150 m3/h is maintained as far as thefilter pressure drop does not exceed 6600 Pa. Fil-ter blocking results in blower overheating. The pri-mary reason for sampler stoppage as the filter is

2 Description of Hunter Automatic MKII

clogging up is an intentional stoppage for overheatprotection. If the ambient temperature is veryhigh, then the overheat protection stoppage maytake place at a lower filter dust loading level thanin cool climate.Hunter Automatic MKII is shown in Figure 1. Therotating drum has positions for the six filters. Dur-ing operation, the drum is automatically turned60° once a week, exposing a new filter to the airflow. The filter change position is on the front. Thefilter collection area is 42 × 24 cm, or 0.1 m2. Theface velocity, that is the flow rate divided by thefilter collection area, or the velocity of the air justbefore being aspirated through the filter, is 0.4 m/swhen the flow rate is 150 m3/h. The total filterdimensions, including the margins for attachment,are 46 × 27.5 cm. Filter is put in its place by afixing frame, that is closed without a need for ascrewdriver or any other tool. In case of flexiblefilter material, the filter is attached in advance toa cardboard frame to obtain the firmness neededfor easy filter placement and handling.

The filter air collection position is at the bot-tom of the device, thus the air is aspirated up-wards through the filter. Particles with a diameterlarger than 80 µm (assuming particle density of2.5 g/cm3) are not sampled, as they are settled inair faster than 0.4 m/s. On the other hand, 10 µmparticles easily follow the flow, as their down-wards settling velocity is only 0.003 m/s [3]. Therejected coarse particles include sand, occasionalgarbage, and rain droplets. Upwards samplingcarries out the duty of coarse-dust-rejection con-veniently, without a need for a leak-tight jointbetween a special immobile sampling inlet and amoving filter drum.

Appendix presents the tasks that an inspectorneeds to carry out when visiting the sampler aftersix weeks of sampling.

S T U K - Y T O - T R 1 8 3

7

2.2State-of-health data transmissionTo avoid major interrupts in sampling, and to con-trol the status of sampling network, the followingstate-of-health (S-O-H) signals should be trans-mitted:a) External power supply (on/off).b) Sampler blower (air flow rate).c) Temperature inside the sampler.d) Tamper indication. It can include data from a

door switch, motion sensor and electronic seal.

The sampler is equipped with a datalogger, whichstores sampling data from various sensors. The S-O-H data is also collected by the datalogger. Thesampler uses the data for air flow calculation. Thedatalogger will also be used for storing data fromoptional meteorological station. The program inthe datalogger automatically sends the data outvia RS-232 serial port. The data is received andstored by an industrial PC, which is run under NToperating system. The data sent by the dataloggeris read by the PC, encrypted and sent to the head-quarters. The data transmission will be carriedout via Inmarsat-C due to its global and reliable

coverage. The data can be e-mailed or faxed tospecific address.

Data reading and transmission is controlled bythe industrial PC. It performs the following tasks:a) The S-O-H data is monitored in very short

intervals.b) All received data is stored in the PC.c) In a case of alarm or failure the alarm message

is sent in real time.d) The PC sends the status information to the

headquarters at predefined intervals

Encryption method is One-time pad (OTP), whichcan not be decrypted without knowing the originalencryption key [4]. OTP is useful in low bandwidthapplications. Each key is used only once. The keysare produced in the headquarters and the inspec-tor loads them into the sampler during his visit, sothe risk for unauthorised access to the key is mini-mal. All information is encrypted before sending.Tampering the equipment is not possible withoutleaving any evidence behind. In case of tampering,the PC destroys the original encryption keys.

Figure 1. Hunter Automatic MKII air sampler. Air is aspirated upwards through the filter located on thebottom position.

OPERATION

1. CONNECT TO POWER SUPPLY

2. LOAD NEW FILTERS 6*1

3. START THE SAMPLER

X

X

X

4. STOP THE SAMPLER

5. UNLOAD DATA

6. CHANGE FILTERS

7 START THE SAMPLER

1700mm

800mm

METEOROLOGICAL SENSORS

MEASURING UNIT

CONTROL UNIT

FILTERING UNIT

HUNTER MK2 AUTOMATIC

INTAKE

12

3

TO CHANGE THE FILTERS:

1. OPEN THE COVER

2. OPEN THE FILTER FRAME

3. CHANGE THE FILTERS

STORAGE BOXES 80 x 350 x 500 mm.

8

S T U K - Y T O - T R 1 8 3

3.1CriteriaUsual criteria for filter material include capabilityto collect particles with a sufficient efficiency, re-sistance to clogging, convenient to handle, and me-chanical strength (does not break easily). The ma-terial should have a low concentration of U andPu, if these elements are to be analysed. Mostlythe problem here is uranium, as it is, unlike pluto-nium, present in nature to a significant extent.The filter material should be appropriate for pre-conditioning methods used in radiochemistry. Or-ganic filter materials, such as polypropylene orpolyvinylchloride, are suitable as they are ashableat low temperatures, leachable to weak acids suchas HNO3, and do not contain significant amountsof uranium.

Collection efficiency is defined as a fraction ofparticles collected on the filter of the particles inthe air entering the filter. Collection efficiencydepends on the filter structure, the particle sizeand face velocity [3]. It is typically the lowest forparticles of about 0.1–0.3 µm in diameter. Parti-cles in this size range are most easily able tofollow the air flow through the porous filter matrix(an example of filter structure is shown in Fig-ure 2). Super-micrometer particles have too muchinertia to follow the curved air flow, and aredeposited on the filter fibers due to impaction.Particles finer than 0.1 µm experience substantialBrownian motion, as gas molecules collide withthem. The wiggling Brownian motion increasesthe probability of particles to collide on filterfibers. The collection efficiency in the particle sizerange below 0.1 µm is therefore the higher thesmaller the particle is. Generally, the filters with agood collection efficiency also produce a high flowresistance. Collection efficiency E can be increasedby using two consecutive filters, but that alsodoubles the pressure drop caused by the filter, ∆p.A quantity called quality factor, Qf, can be used to

compare the collection characteristics of filterswith different pressure drops [3],

( )( )ln 1 1

pf

EQ

−=

∆ .ln 1 1

pf

EQ

−=

∆ . (2)

Quality factor describes how large pressure dropis needed to collect the particles with a given effi-ciency. Here it is assumed that a two-stage filterconsisting of two similar layers placed one uponthe other have the same quality factor than onefilter alone. Thus if the first layer collects 70% ofthe 0.2 µm particles, then the second layer is as-sumed to collect 70% of those 0.2 µm particles thatwere not collected by the first layer (that is 21% ofall the 0.2 µm particles). The overall efficiency ofthis hypothetical two-stage filter would be 91% for0.2 µm particles.

Collection efficiency is likely to change duringsample collection, as the particles collected changethe structure of the filter matrix. The phenomenais the most dominant for electret-type filters. Theyutilise fibers made of charged and insulating ma-terial so that one side is positively, and the other

3 Choosing the filter material

Figure 2. A scanning electron microscope photo-graph of a fibrous filter. The scale is shown underthe photograph.

S T U K - Y T O - T R 1 8 3

9

negatively charged, resulting a zero net charge.Electrostatic attraction enhances collection effi-ciency, especially of 0.1–1 µm particles that arethe most difficult to collect otherwise. However,particles collected on the filter may mask thecharge and reduce collection efficiency [3,5]. Also,the fiber charging can be lost when electret filtersare exposed to, e.g., high humidity or organicliquid aerosols.

The filter with the lowest initial pressure dropis not necessarily the most resistant to clogging.The pressure drop increase rate during air sam-pling depends on filter properties, such as fiberdiameter and the fraction of the filter volume thatis covered by fibers. The larger the empty spacebetween the fibers, the lower the clogging rateusually is. The properties of the collected particles,such as particle size distribution and particlesurface properties, also affect the clogging rate.

3.2Concentration and size distributionof the atmospheric aerosolThe mass concentration and size distribution ofairborne particles, and therefore also the filterclogging rate, varies depending on the environ-ment. There is two airborne particle modes with adifferent formation mechanism, and consequentlya different particle size. Fine particles (diameterbelow 1 µm) are formed as gases and vapours arecondensed. Atmospheric fine particles are mainlyformed in various combustion processes, such aspower plants and motor engines, where the sub-stances are volatilised in high temperature. In ad-dition, gases released in combustion processesform fine particles in atmosphere, as they are con-verted to condensed species by chemical reactions(such as conversion of gaseous SO2 to condensedH2SO4). Coarse particles (diameter larger than1 µm) include wind-blown soil, dust and other par-ticles released to air mechanically, that is withoutvolatilisation. In marine conditions coarse parti-cles are predominantly sea salt that are releasedto air mechanically as sea-water droplets, followedby evaporation of water.

Urban aerosol is dominated by the fine particlemode. Particle concentration, especially that of thefine mode, is lower in rural continental areas ascompared to urban areas (Figure 3.a). There islarge spatial variation in aerosol mass concentra-tion. It is typically more than 0.1 mg/m3 in large

cities, and around 0.04 mg/m3 in rural areas.Desert aerosol is characterised by a high concen-tration (may exceed 1 mg/m3) of coarse wind-blown crustal material (Figure 3.b). Wind velocitystrongly affects the mass concentration and parti-cle size in desert, as high velocity wind raises verycoarse (> 10 µm) particles from the ground. Parti-cles in the size range of 10–100 µm have beenfound to dominate the total airborne mass concen-tration during sandstorm in Sahara [6].

3.3Filter testsFour filters were tested for collection efficiencyand pressure drop during exposure to aerosol thatcontained 25 weight-% of diesel fume generated bya motor boat engine, and 75 weight-% of SAE finedust (Figure 3.c). Diesel fume simulates combus-tion-originated sub-micrometer airborne particles,

Figure 3. a) and b) Typical aerosol particle masssize distributions (based on references [3,6 and 7].Note the different scales in ordinate axis. c) The masssize distribution of the aerosol used for filter expo-sure in our tests.

0

0,5

1

1,5

2

2,5

0,01 0,1 1 10 100Particle diameter, µm

dM/d

log(

D),

mg/

m³

UrbanDesertDesert (sandstorm)

0

0,02

0,04

0,06

0,08

0,1

0,01 0,1 1 10 100Particle diameter, µm

dM/d

log(

D),

mg/

m³ Urban

Rural

0,01 0,1 1 10 100Particle diameter, µm

a)

b)

c)

10

S T U K - Y T O - T R 1 8 3

and SAE fine dust is a typical test dust for coarsewind-blown dust. Our test aerosol resembles conti-nental rural area aerosol, whereas in urban areasfine particle mode is more dominant. If marineaerosol was to be simulated, the coarse particlecontribution to total mass concentration should beabout 95%, and the coarse particles should be saltinstead of crustal dust used here.

The experimental set-up is shown in Figure 4.Collection efficiency was determined by measur-ing size distribution of di-ethyl–hexyl sebacate(DEHS) aerosol with an optical particle size ana-lyser (PMS LAS X) alternately upstream anddownstream of the filter.

Collection efficiency was measured for unexposedfilters and after exposure to 2.5 mg/cm2 and 5 mg/cm2 of aerosol (Figure 5). Here exposure is definedas a mass of particles in air volume that is drawnthrough a unit area of filter. The values 2.5 mg/cm2 and 5 mg/cm2 correspond to a one-week opera-tion of Hunter MKII (flow rate 150 m3/h and filtercollection area 0.1 m2) in an environment wherethe particle mass concentration is 0.1 mg/m3 and0.2 mg/m3, respectively. The measured collectionefficiencies are shown in Figure 6, the pressuredrops in Figure 7 and the quality factors calculat-ed from equation (2) to 0.2 µm particles in Fig-ure 8.

Figure 4. The experimental set-up for collection efficiency measurements. 1 = Mixing chamber, 2 = Airoutlets downstream of the filters (four filters can be measured simultaneously, one is shown in the draw-ing), 3 = Particle size analyzer and computer.

21

3

1

3

2

S T U K - Y T O - T R 1 8 3

11

0

20

40

60

80

100

0,1 1 10Diameter, µm

Colle

ctio

n ef

ficie

ncy,

% FPP-15 (Double), dp =1196 PaSBMF-40VF, dp = 909 Pa

Non-electret, dp = 515 Pa

Electret, dp = 132 Pa

Electret isopr., dp = 136 Pa

Exposure 0 mg/cm²

0

20

40

60

80

100

0,1 1 10Diameter, µm

Colle

ctio

n ef

ficie

ncy,

%

FPP-15 (Double), dp = 2187 Pa

SBMF-40VF, dp = 1361 Pa

Non-electret, dp = 731 Pa

Electret, dp = 163 Pa

Electret, isopr., dp = 167 Pa

Exposure 2.5 mg/cm²

0

20

40

60

80

100

0,1 1 10Diameter, µm

Colle

ctio

n ef

ficie

ncy,

%

FPP-15 (Double), dp = 3880 Pa

SBMF-40VF, dp = 2206 Pa

Non-electret, dp = 1280 Pa

Electret, dp = 213 Pa

Electret, isopr., dp = 218 Pa

Exposure 5 mg/cm²

Figure 5. A filter exposed to 5 mg/cm2 of aerosolcontaining 25 wt-% of diesel fume particles and75 wt-% of SAE fine dust. Diameter of the darkenedarea, where the particles are collected, is 15 cm.

Figure 6. The dependence of collection efficiency on filter exposure. Pressure drops are shown in legend.Isopr. = 24 h treatment in isopropyl alcohol.

One of the filters tested was of electret-type. Ithad a superior quality factor as unexposed. Thegood initial performance was shown to be based onfilter charging, as 24 hours treatment with isopro-pyl alcohol damaged its capability to collect sub-micron particles. Even an untreated piece had lostthe gain from fiber charging by the time exposurewas 2.5 mg/cm2, when it collected 0.2 µm particlesonly with an efficiency of 19%. Further exposureto 5 mg/cm2 slightly increased the collection effi-ciency, apparently due to the enhancement of non-electrical collection mechanisms. There was no sig-nificant difference between the isopropyl-treatedand untreated sample at the exposures of 2.5 mg/cm2 and 5 mg/cm2.

12

S T U K - Y T O - T R 1 8 3

Two of the filters, Petrianov FPP-15 (Double) andSBMF-40VF, were able to collect 0.2 µm particleswith an efficiency better than 90% at all the threeexposure levels. Petrianov FPP-15-1.5 was testedas double-layer configuration, two filters one uponthe other. The advantage as compared to the nor-mal single configuration is the increased collectionefficiency, and the disadvantage is the higher pres-sure drop. For easy handling, the nominal collec-tion sides of the two FPP-15 filters were againsteach other (as delivered) so that the filter on thetop was upside down. Thus the collection on FPP-15 took place on the side other than the one rec-ommended by manufacturer.

The decision between FPP-15 (Double) andSBMF-40VF depends on the application. FPP-15(Double) is better if the collection efficiency is tobe maximised, but SBMF-40VF produces a lowerpressure drop. Apparently FPP-15 (Double) is bet-

Figure 7. Pressure drops as functions of exposure (the same test runs as in Figure 6).

Figure 8. Quality factors for 0.2 µm particles when filters were exposed to 0–5 mg/cm2 of aerosol (thesame test runs as in Figure 6).

0

1000

2000

3000

4000

0 1 2 3 4 5

Exposure, mg/cm²

Pres

sure

dro

p, P

a

FPP-15 (Double)

SBMF-40VF

Non-electret

Electret

Electret isopr.

0

2

4

6

8

10

12

14

16

FPP-15 (Double) SBMF-40VF Non-electret Electret Electret isopr.

Qual

ity fa

ctor

, 1/k

Pa

0 mg/cm²

2.5 mg/cm²

5 mg/cm²

ter if sampling time is so short, that clogging is nota problem. The difference in collection efficienciesbecomes less significant when the exposure isincreased, as the collection efficiency of both fil-ters approaches 100%. Thus the collection efficien-cy becomes less significant selection criteria, if thecollection time is extended. SBMF-40VF appearsto be better if the collection time should be as longas possible, because the pressure drop of FPP-15(Double) is increased faster than that of SBMF-40VF. It should be noted, that the conclusionsdrawn here are limited to the filters when used asconfigurations similar to those tested here. Forinstance, single-layer FPP-15 produces a lowerpressure drop, and may therefore be more prefera-ble to some applications than the FPP-15 (Double)tested here. The fact that we collected particles onthe bottom side of FPP-15 may also have affectedthe results.

S T U K - Y T O - T R 1 8 3

13

4 Pre-filtration of coarse particles

4.1MotivationNatural uranium in air is present in coarse parti-cle mode associated with wind-blown dust. Pre-filtration of coarse particles may increase the sig-nal to background ratio, since the freshly releaseduranium and plutonium are presumably enrichedin the fine particle mode. This is because the vola-tilised species, such as UF6, are more easily acci-dentally released than the condensed ones. Spe-cies that have volatilised, followed by condensa-tion, are enriched in the fine mode [3]. Besides, thespecies attached to coarse particles can be foundonly if the sampler is located near the source, un-less the species were released to a significantheight (Table I).

Collection of coarse particles from air separate-ly upstream of the sampling filter serves alsoanother purpose. The filter clogging rate is de-creased, if clogging is caused predominantly bycoarse particles. The benefit from pre-filtration ispresumably largest in desert aerosol conditions(especially during sandstorm), where airborne par-ticles are dominated by coarse crustal particles.

4.2MethodsPre-separation of coarse particle is usually carriedout with devices utilising impaction. The sampledair flow makes abrupt curves that the coarse par-

ticles are not able to follow due to their inertia,but are deposited on the collection surface. Forinstance, sampling inlets to meet PM-10 criteriacollect particles with aerodynamic diameter largerthan 10 µm [3,8]. This kind of sampling inlets aremeant to be permanently leak-tightly assembledupstream of the filter cassette. A sampler withautomatic filter changing can not easily apply apermanently assembled inlet, because there is aneed for a sealing between two parts (inlet andfilter cassette) that move relative to each other.Instead, an alternative system consisting of twoconsecutive filters, with the first one acting as apre-filter, was applied here. The pre-filter is signif-icantly more porous than the collection filter, andthus both the clogging rate and the collection effi-ciency are lower for the pre-filter. In our configu-ration, the pre-filter and the collection filter areplaced directly one upon the other.

4.3TestsFour different filter configurations were tested;i) a single collection filter (Petrianov RFM-1.7),ii) two consecutive collection filters,iii) a collection filter preceded by a non-electret

pre-filter andiv) a collection filter preceded by an electret pre-

filter.

Table I. Distance that a particle travels with the air flow during the time it settles down from a givenheight. Values are for spherical particles with a density of 1 g/cm3. Gravitational settling is assumed to bethe only mechanism causing vertical particle movement (no vertical advection).

Time to settle down from a height of

Horizontal distance travelled during settling (wind velocity 1–10 m/s)

Diameter, µm Deposition velocity, m/s 10 m 100 m 10 m 100 m

10 0.003 54 min 9 h 3–30 km 30–300 km

100 0.25 40 s 7 min 40–400 m 0.4–4 km

14

S T U K - Y T O - T R 1 8 3

Configuration i) is the conventional case, with nopre-filtration. Configuration ii) is similar to i), ex-cept that it has a back-up filter to increase thecollection efficiency. A separate analysis of theback-up filter gives a sample with a very highenrichment of sub-micron particles.

Configurations iii) and iv) provide a pre-filter.The electret pre-filter collects fine particles moreefficiently than the non-electret one, at least dur-ing the early stage of sampling. The electret pre-filter is meant to be used when the filter cloggingrate reduction is the main motivation for pre-filtering, and the non-electret pre-filter when asharp size-classification between the fine andcoarse modes is more important than the cloggingprevention.

Filters were exposed to three different kinds ofaerosol. The aerosol used in collection efficiencytests (mixture of diesel fume and SAE test dust,Figure 3.c) was also used here. In addition, dieselfume alone was used to represent sub-micronparticles, and SAE test dust alone to representcoarse particles.

The pressure drop during exposure is shown inFigure 9. Pre-filter decreases clogging rate, espe-cially when particles are coarse. If particles arefine, only a small fraction of them is collected onthe pre-filter, and consequently the clogging rate

is not remarkably decreased. Figure 10 shows,that the collection filter located downstream of thepre-filter is almost clean when particles werecoarse, but heavily loaded when particles werefine.

The pressure drop of the two-stage filter con-figuration ii) is initially two times that of thesingle filter i), but the difference is not increasedduring exposure. In the two-stage system, most ofthe mass is collected on the first layer, and conse-quently the additional pressure drop developedduring exposure is that developed on the firstlayer alone. Thus, additional filter layers do notincrease clogging rate significantly, as far as theinitial pressure drop with the additional layersremains low as compared to the maximum accept-able pressure drop.

4.4Simulation of air sampler flow rateThe air flow rate maintained by a sampling pumpis the lower the higher the pressure drop over thefilter is. Filter clogging finally results in a situa-tion where the flow rate is too low to be accepta-ble, or the pump has to be stopped to avoid over-heating. Figure 11 shows the simulated flow ratefor Hunter MKII during sample collection withthe following assumptions:• Flow rate maintained by the pump depends on

Figure 9. Pressure drop during exposure at face velocity of 0.4 m/s.

Coarse particles

0

2000

4000

6000

8000

10000

0 2 4 6 8 10Exposure (mg/cm²)

Pres

sure

dro

p, P

a

Coarse+fine particles

0

2000

4000

6000

8000

10000

0 2 4 6 8 10 12 14Exposure, mg/cm²

Pres

sure

dro

p, P

a Prefilter (electret) + filter

Prefilter + filter

2 filters

Filter

Fine particles

0

2000

4000

6000

8000

10000

0 2 4 6 8 10Exposure (mg/cm²)

Pres

sure

dro

p, P

a

S T U K - Y T O - T R 1 8 3

15

the pressure drop as described by equation (1)• Pressure drop at a face velocity of 0.4 m/s

depends on exposure as in Figure 9(coarse+fine particles). In other words, theproperties of the airborne particles and of thefilter are similar to those during our exposuretests.

• Pressure drop is directly proportional to theface velocity.

• Mass concentration in air is 0.2 mg/m3, whichis a typical value in urban areas. Having otherparameters unchanged, the time for the flow

rate to drop to a given value is directly propor-tional to the mass concentration.

Figure 11 shows, that doubling the filter area from500 to 1000 cm2 increased collection time signifi-cantly. Also pre-filtration extends the collectiontime in this example, because there is a high con-centration of coarse particles in air. If there is anetwork of air samplers being built, simulationcan be helpful in the process of applying practicalexperience from the first samplers to design modi-fications to the later samplers.

Figure 10. The pre-filter + collection filter system after exposure to 5 mg/cm2 of a) coarse particles andb) fine diesel fume particles. Pre-filter is turned away, and the upside of the collection filter is shown.

Figure 11. Simulated flow rate of Hunter Automatic MKII (filter collection area 1000 cm2) assuming thatthe mass concentration of airborne particles is 0.2 mg/m3 and that they are similar to the mixture ofdiesel fume and coarse dust used in our tests. Flow rate is also shown for a sampler with a half-sizecollection area.

b)a)

0

50

100

150

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Time, days

Flow

rate

, m³/

h

Hunter MKII, with pre-filter

Hunter MKII, no pre-filter

Filter collection area 500 cm², no pre-filter

16

S T U K - Y T O - T R 1 8 3

4.5Practical experience from fieldtrial in KazakhstanPre-filtration was tested during field trial in Kaza-khstan [9,10]. Two Hunter MKII samplers wereutilised next to each other. The electret-type pre-filter was used in one of the samplers during sixseparate weeks, otherwise there was a single col-lection filter only. The other sampler was equippedwith a permanently assembled impactor-type fil-tering unit to remove particles larger than 13 µm[11]. The sampling period was usually one week,longer sampling periods up to 4 weeks were alsotested. Mass concentration was 0.1–0.25 mg/m3.Filter dust loading did not cause clogging-up prob-lems to any of the filter configurations. The analy-sis results are presented in [12,13]. The followingpractical observations were made during the fieldtrial:

• The preparation of two-stage filters before fieldcampaign is an additional work task. However,working time can be spared if the coarse parti-cle fraction is also of interest, as it can be

conveniently recovered with the two-stage fil-ter.

• The upside of the two-stage filter must beclearly indicated to avoid upside-down loadingduring sample collection. This is not so acutewith single filters. Collection on the wrong sideof a single filter does not create a seriousproblem with most filter types, even if alsosingle filters usually have a preferred upside.

• The physical separation of the pre-filter fromthe collection filter at laboratory did not in-volve difficulties. Particles were collected in-side the pre-filter matrix, and were not de-tached to a significant extent during the filterhandling process. Detachment could be a prob-lem, if particle loading was so high, that theparticle deposit layer begins to grow on thesurface of the pre-filter. If there is loose partic-ulate matter present in the transport package,it presumably originates predominantly fromthe pre-filter.

S T U K - Y T O - T R 1 8 3

17

5 Conclusions

An automatic air sampler for WAES, Hunter Auto-matic MKII, was designed and built. The deviceenables unattended operation for 6 weeks, duringthe time six weekly samples are collected. Severalmodifications were made to the original samplers.To avoid filter clogging, the filter collection areawas doubled, filter material was changed, and anoptional pre-filtration method was designed to re-move coarse particles. Due to their negligible ura-nium concentration and ashability in low temper-atures, the presently used organic filter materialsare more suitable for chemical analysis methodsthan the previously used glass-fiber ones.

The filter exposure tests and simulation resultsshow, that with an applicable filter (possibly pre-ceded by a pre-filter), the sampler is easily able tooperate for one week without clogging, if theairborne particle mass concentration is that typi-cal in desert conditions. This can be done withoutcompromising the specification for collection effi-ciency (must collect more than 80% of the 0.2 µmparticles). However, the functioning in real-condi-tions can be verified only on the spot, as we do notknow beforehand if the conditions on site can beconsidered “typical”. Especially sand storms set a

challenge to any air sampler. Very high ambienttemperature may also decrease the operation timebefore the sampler is stopped for overheat protec-tion. Simulation method demonstrated in this re-port provides a tool for sampler modificationsduring sampler-network build-up, if the first sam-plers taken into operation show clogging, or otherneed for modifications.

The pre-filtration of coarse particles has twoadvantages. It decreases the filter clogging rate, ifthe airborne particles contain large fraction ofcoarse particles, as is the case in desert conditions.In typical urban aerosol conditions pre-filtration isless beneficial. Secondly, the absence of coarseparticles is expected to improve detection limit inchemical analysis, as the artificial radionuclidesare most likely enriched in the fine particles, andthe natural uranium in the coarse particles. Thedownside is, that there is no absolute certainty onthe absence of artificial nuclides in the coarsefraction, and separate analysis of both fractionsdoubles the number of samples. The decision,whether pre-filtering is used or not, is to be madebased on the precise objectives of the campaign inquestion.

18

S T U K - Y T O - T R 1 8 3

Acknowledgements

The Kazakhstan aerosol sampling field trial wascarried out in close co-operation between the Stateauthorities, Atomic Energy Committee of the Re-public of Kazakhstan (KAEC) and Radiation andNuclear Safety Authority of Finland (STUK). Theactive organizational and logistical role of DirectorTimur Zantikin and Head of Division Gulnara El-igbaeva of KAEC is greatly acknowledged. Imple-mentation of the field trial, first in Kurchatov andlater on in Astana, was successfully carried out bythe National Nuclear Center of the Republic ofKazakhstan (NNC RK), Institute of RadiationSafety and Ecology (IRSE) and the Astana Center

of Hydrometeorology Monitoring (KAZHYDRO-MET). The authors are thankful for the excellenttechnical support and seamless co-operation withthe staff of IRSE, Director Murat Akhmetov, Dep-uty Director Larissa Ptitskaya, Interpreter YuliyaLogvinova and many others. The authors arethankful also for the support and advice of Direc-tor Ludmila Chuntonova of KAZHYDROMET dur-ing the field trial in Astana.

We thank Mr. Jyrki Juhanoja from the Insti-tute of Biotechnology, University of Helsinki forthe electron microscopy facilities (Figure 2).

S T U K - Y T O - T R 1 8 3

19

References

[1] R. Rosenberg, G. Brachet, W.D. Lauppe, I.Vintersved, K. Nicholson, P. Krey, D. Swindle,N. Wogman, R. Hanlen and several addition-al contributors. IAEA use of Wide Area Envi-ronmental Sampling In the detection of Un-declared Nuclear Activities, Member StateSupport Programmes to the IAEA STR-321,August 27, 1999.

[2] I. Riekkinen, T. Jaakkola, S. Pulli, S. Salmin-en, S. Ristonmaa, R. Rosenberg, R. Zilliacus.Analytical Methods for Wide Area Environ-mental Sampling (WAES) for Air Filters.STUK-YTO-TR 184, 2002.

[3] W.C. Hinds. Aerosol Technology. Properties,Behavior, and Measurement of Airborne Par-ticles. 2. ed. Wiley-Interscience, New York.483 p., 1999.

[4] B. Schneier. Applied Cryptography, Protocols,Algorithms and Source Code in C, 2nd Ed,pp.15–17. John Wiley & Sons, 1996.

[5] M. Lehtimäki. Development of Test Methodfor Electret Filters. J. Aerosol Sci., Vol 26.Suppl 1, pp. S737–S738, 1995.

[6] G.A. d’Almeida, L. Schütz. Number, mass andvolume distributions of mineral aerosol andsoils of the Sahara. J. Climate Appl. Meteor-ol., 22, 233–243, 1983.

[7] J.H. Seinfeld, S.N. Pandis. Atmosphericchemistry and physics from air pollution toclimate change. New York: Wiley-Inter-science Publication, 1998.

[8] J.G. Watson, J.C. Chow. Ambient Air Sam-pling. In: Willeke, K. and Baron, P.A. (Eds.)Aerosol Measurement: Principles, Tech-niques, and Applications. Van Nostrand Re-inhold, New York. pp. 622–639, 1993.

[9] M. Tarvainen, T. Valmari, T. Honkamaa, S.Ylätalo, J. Lehtinen, R. Rosenberg, R. Zillia-cus, M.. Lehtimäki, A. Taipale. OptimizingAerosol Sampling for Environmental Moni-toring of Nuclear Signatures. Institute ofNuclear Materials Management, 42nd Annu-al Meeting, Indian Wells, California, 15–19July 2001. 8 p.

[10] M. Tarvainen, T. Valmari, S. Ylätalo, R. Pöl-länen, R. Rosenberg, R. Zilliacus, J. Lehtinen,T. Jaakkola, I. Riekkinen, S. Pulli, J. Lehto.Kazakhstan Aerosol Sampling Field Trial.ESARDA, 23rd Annual Meeting, Symposiumon Safeguards and Nuclear Material Man-agement, Brugge, Belgium 8–10 May, 2001.

[11] S. Ylätalo, T. Valmari, J. Lehtinen. Construc-tion of impactor-based pre-filtration unit forhigh-volume aerosol sampling. Environmen-tal and Chemical Physics Vol. 24, No 1/ 2002,pp 2–6.

[12] R. Rosenberg, R. Zilliacus, M. Tarvainen, T.Valmari, S. Ristonmaa, T. Jaakkola, I. Riek-kinen, S. Pulli, J. Lehto. Aerosols in WideArea Environmental Sampling; AnalyticalChallenges. Proceedings of the Institute ofNuclear Materials Management, 42nd Annu-al Meeting, Indian Wells, California, 15–19July 2001. 7 p.

[13] R. Rosenberg, R. Zilliacus, M. Tarvainen, T.Valmari, S., Ylätalo, S. Ristonmaa, T. Jaakko-la, I. Riekkinen, S. Pulli, J. Lehtinen. Pre-filtration in Aerosol Sampling for Wide AreaEnvironmental Sampling. ESARDA, 23rdAnnual Meeting, Symposium on Safeguardsand Nuclear Material Management, Brugge,Belgium 8–10 May, 2001.

20

S T U K - Y T O - T R 1 8 3

Each sampling station collects weekly samples. Aninspector visits each of the samplers every sixweeks, unless the state-of-health data indicatesproblems.

During the normal filter-changing visit to asampler, the inspector should carry the followingitems with him/her:• Six new filters stored in plastic bags.• Two pairs of clean disposable rubber gloves.• Watch and ball pen.• Six empty large plastic bags.• Logbook.• Organic solvent, e.g. alcohol, to clean filter

holder frames.• Empty bag for waste.• A data sheet for each of the six new filters,

filled in with the following information; Filternumber, Sampler ID, Scheduled samplingweek, Scheduled filter position at sampler (po-sition 1 for first filter of the six weeks cycle,etc), Filter loading date, Name of the inspector.

During the visit to a sampler, the inspector carriesout the following operations:

Removal of used filters from sampler• Open sampler front cover and lay down new

filters on left top shelf of sampler.• Write down in logbook the following informa-

tion; Is power on?, Which filter is in samplingposition (number six down indicates that auto-matic filter changing is completed)?; Conditionof sampler based on visual observation (e.g.indication of tampering, dust, power is on, etc.);Number of records in the data terminal; Dateand time when the sampler was stopped.

APPENDIX

EXAMPLE OF THE SAMPLER OPERATION PROCEDURE

• Stop sampler, Disconnect data terminal, take itinto laboratory and start data unloading tocomputer (unloading takes time!).

• For the filter in front of you, select the properdata sheet (check filter number) and fill inCondition of filter prior to removing it (is itwhole, well in place, evenly loaded, etc), Actualfilter position at sampler, Date and Your name.

• Remove used filter from sampler using cleandisposable rubber gloves, insert the used filterstraight into large plastic bag and seal. Checkfilter number against number on the datasheet.

• Lay used filter in plastic bag on right top shelfof sampler.

• Repeat the previous 3 steps to remove otherfilters one by one.

• Clean air sampler frames from dust, usingorganic solvent, if needed.

Loading new filters into sampler• Put on new clean disposal rubber gloves• Take a new filter (from left top shelf), and the

relevant pre-filled data sheet. Check filternumber against the number in the data sheet.

• Confirm that the filter position of sampler(position 1–6) is correct.

• Take a new filter from plastic bag, set it inplace in sampler and close the fixing frame.

• Repeat the previous 3 steps to load other newfilters.

• Rotate sampler holder, as needed, and leaveposition number 1 down (that is, on samplingposition).

• Connect data terminal to sampler after com-pleting data unloading and deleting old files.

• Start sampling and close the sampler cover.• Fill in Logbook.