Examining ways to capture airborne microorganisms

Examining ways to capture airborne microorganisms

Four methods used to capture airborne microorganisms — sedimentation, impaction, impingement, and filtration — can produce different results. This article will help users select the correct sampling method.

By Derek E. Pendlebury and David Pickard

Current U.S. legislation specifies both the test procedures and test limits for airborne particulate levels in designated areas. No limits currently exist for airborne bacterial contamination. United States Pharmacopeia (USP) chapter <1116> gives recommendations for bacterial levels in

designated areas that are accepted by both industry and regulatory authorities.

Table 1 highlights the current proposals of both the USP and the European authorities for revised limits of airborne contamination. This table highlights the low contamination levels expected within Class 100 and unidirectional air flow areas. The levels of airborne contamination expected inside an isolator is many orders of magnitude cleaner than expected in a Class 100 area. An isolator is designed to produce and maintain a sterile environment with all gas exchange through HEPA (high efficiency particulate air) filters, which are designed to remove particulates and associated bacterial contamination, and then the gas is moved into the isolator.

The available methods for the sampling of air within Class 100 and unidirectional air flow areas include sedimentation, impingement and impaction. All of these sampling techniques were designed for sampling in areas where a higher level of bioburden was expected, for example, hospital wards, operating theaters, etc. and are not accurate enough to be able to sample to the levels desired for Class 100 areas or in cleaner environments.

This article will introduce a new technique for quantitative recovery of airborne microorganisms.


When a petri dish (usually 90mm wide) is filled with a suitable media(15) and exposed to atmosphere, particles and associated bacteria will settle onto the plate and can be recovered. Whyte(16) demonstrated that gravitational settling was the most important of the mechanisms identified by which deposition occurs, but that other mechanisms will also affect the results obtained.

Herman and Morelli(17) reported that air turbulence around an open plate can affect the results and smaller particles may not settle at all(4). It has been reported(19) that the long exposure times required to overcome the effects of turbulence result in desiccation of the agar and subsequent poor bacterial growth, and Kingston(15) has shown that drying of a plate containing stress sensitive Streptococcus salivarious reduced the viable count after 1 hour to 50 percent of the original value.

Whyte and Niven(23) demonstrated that settle plates are not suitable for use in areas where the expected airborne contamination level is below 20cfu/m3, for example a Class 100 area where the limit is <3 organisms/m3 the statistical settling time for two bacteria would be 18 hours, which is too long. Many workers(4, 17, 18, 22) have concluded that the use of settle plates is not a suitable method for the quantitative measurement of airborne microorganisms because the volume of air sampled is not measured, and different sized particles are sampled at different rates. Lu(22) has demonstrated that settle plates are unsuitable for monitoring laminar airflow benches.


In an impinger system a known quantity of air is pulled at high velocities through a fluid in a glass vessel(24, 27). Bacteria in the air are retained in the washing fluid while the air passes through the fluid and out through the pump system. The high, almost sonic air speed used in impinger devices results in high shear forces which break up bacterial/particulate aggregates so that the total count obtained closely reflects the actual number of viable organisms. This shearing action differentiates impingement from other methods, all of which measure only the number of bacteria bearing particulates which give colony forming units (cfu)(24, 25, 26, 29).

Some workers(28) have reported wide variations in the efficiency of impingers depending on the position of the base of the impinger tube within the liquid column. Impaction in the liquid on or near the liquid surface results in less efficient collection of smaller particles, and already collected particles can be regenerated because of splashing at the liquid surface. Macher and First(43) demonstrated that the sampling efficiencies of different impingers is dependent on both the design of the system and the particle sizes being sampled. Lyons,(59) who showed that particles smaller than 5.0 &#181m can pass through the impingers tested without collection in the liquid, also found that accuracy and reproducibility were difficult to achieve with an impinger sampling system.

The air velocity passing through the liquid sometimes requires the use of antifoam agents(25, 26, 27) to prevent frothing and subsequent passage of liquid and bacteria into the pump system, and during longer sampling procedures some fluid loss will occur due to evaporation which must be counteracted by the addition of make-up fluid. Impinger flasks must also be sterilized after each use, they are costly and time consuming to use(27), and they require the use of a vacuum pump which limits mobility. Brachman(42) in 1964 recommended that the AGI-30 (All Glass Impinger) be used as the standard reference method for monitoring of viable airborne contaminants. However advances in the area of impaction type collectors has led to their wide acceptance by both industry and regulatory authorities.


There are a range of sampling devices available which work on the principle of impaction of airborne particles. The common mechanism in all devices is that as air is drawn into the sampling device it is forced to change direction to by-pass a non-porous capture media, particles and associated bacteria are forced out of the air stream due to their mass and velocity and are impacted onto the media. This is cultured and counted to give results expressed as, the number of colony forming units per unit volume of air, usually cfu/ft3 or cfu/m3.

There are three types of impaction mechanism, each of which is represented by commonly available commercial equipment. Each mechanism is different in the way it retains particles and these differences confer different capture patterns for different sized particles.

Slit to agar impactors

Slit to agar systems, or STA, utilize a revolving plate beneath a slit orifice. Air is drawn through the orifice under vacuum and impelled onto the revolving plate(21,30), on which is placed a petri dish containing a suitable capture media. The time taken for the plate to revolve can be varied. STA samplers can be used to show changes in airborne organism concentration directly related to time, and multiple sampling with a single plate is possible(25).

STA samplers have been shown not to be suitable for use when a high concentration of organisms are expected(30, 31) and have been described as cumbersome to use(24, 25). However it is still the method of comparison against which other air sampling devices are assessed(6, 7, 9, 10, 11, 24, 25, 32, 33, 34, 35, 36) and it was the method used to determine proposed limits for microbial contamination in clean rooms and clean zones as proposed by USP. USP currently states that other samplers can be used but results obtained should be able to be correlated to results obtained by STA systems(53), however the recently revised chapter <1116> is expected to drop this requirement.

Fields and co-workers(58) raised objections to the Reyniers STA system. Their work demonstrated a significantly higher percentage of particles sized 0.5-0.8 &#181m, and a significantly lower percentage of particles between 3.0-25.0 &#181m in size, in air that had passed through the slit of the STA, than they found in ambient air. This indicated that larger particles were being fragmented either after passing through the slit, or fragmentation was caused during passage through the slit of the STA.

Tauber(54) concluded that the STA should not be used in clean rooms and critical areas, because of the difficulty of sterilizing all parts of the equipment, however no alternative system was recommended.

Sieve impactors

Sieve impactors are available in single- or multi-stage configurations. Single-stage systems impact microorganisms passing through a perforated plate onto a capture media and all organisms collected are cultured and counted from a single plate.

Lach(36) demonstrated that the surface air system (SAS) was only efficient for collection of particles of 4.0 &#181m or larger, below this level, collection efficiency decreases rapidly. Jensen(11) demonstrated very poor recoveries of both Bacillus subtillis and Escherichia coli by the SAS when compared to other impaction devices and recommended the use of the SAS only when a bioaerosol is well characterized, or where the size distribution of bacteria to be sampled is known to be greater than 4.0 mm.

Buttner(37) described the Andersen six- stage sieve impactor as having greater sensitivity than the SAS and being able to recover a significantly higher number of viable spores of Penicillium chrysogenum than the SAS system. Other reports comparing the SAS to slit to agar systems showed the SAS to be 50 percent(49), and 40 percent(50) less efficient for sampling airborne fungi. Jensen(11) also demonstrated that while a single stage Andersen sieve impactor was significantly more efficient than a single stage SAS impactor in collecting airborne B.subtillis and E.coli, there was no significant difference in retention efficiencies between a single-stage and a six-stage Andersen sampler for the same two organisms. Work by Jones(38) showed a linear relationship between results obtained from a single-stage and a six-stage Andersen sampler.

A multi-stage sieve impactor(39) contains a stack of plates each of which contains holes of decreasing diameter in succeeding plates. Air is pulled through the device and particles passing through each plate are collected on an agar capture dish beneath each plate. Air travels at increasing velocities through each plate, and this allows impaction of particles with successively smaller diameters and lower mass. Use of a multistage impactor allows particles to be separated into size ranges.

Significant differences have been demonstrated however between Andersen two- and six-stage impactors(11, 40), and between Andersen two- and eight-stage impactors(41). Jensen(11) demonstrated that differences between two- and six-stage impactors could be accounted for by the lower collection efficiency of the two-stage impactor for particles with diameters of less than 1.0 mm. It has also been shown that there is no significant difference in results from a six-stage and an eight-stage impactor(41).

Marple(80) discussed areas of loss within multi-stage impactors and classified them as follows:

Inlet losses: The effects of crosswind at the point of particle entry to the sampler which will affect the efficiency of the sampler.

Interstage losses: Particle deposits onto surfaces other than the impaction medium which can account for up to 10 percent of total particles.

Particle reentrainment: Particles being forced back into the airstream due to effects such as particle bounce; dehydration effects of the surface of the impaction medium.

Centrifugal impactors

The principle of centrifugal impaction for sampling of airborne microorganisms is used by the Reuter Centrifugal Samplers. Both the RCS and the RCS+ work on the same principle but there are differences in the design of the two units which may affect the results obtained.

The RCS operates when a propeller with a small blade pitch rotates within an open fronted housing. This causes air movement to be directed radially outward which in turn results in a lower pressure in the area in front of the central point of the propeller, air is drawn into this central area to maintain atmospheric pressure. Air is forced toward the internal wall of the housing before leaving the housing, and a plastic strip coated with agar is placed around the inside wall of the housing to trap bacteria and associated particles which impact at this point. A major problem with this design is that air is discharged as a cone around the periphery of the housing which results in mixing of incoming and outgoing air. Therefore the same air volume can be repeatedly sampled, and it is very difficult to discriminate between incoming and outgoing airstreams to quantify flow rate or to measure the particulate content(45).

Kaye(46) calculates how a stated air flow rate of 280L/min., translates into a separation volume for the instrument of 40L/min. for particles with a diameter of 4.0 &#181m, and that the sampling rate of the RCS depends directly on the square of the particle diameter. Macher and First(45) demonstrated the efficiency of the RCS at retaining polystyrene microspheres of varying sizes by counting both those trapped on the agar surface, and those passing uncollected into the exhaust. They found that the RCS was 100 percent effective only for particles of 16.0 &#181m or larger diameter, but becomes increasingly less effective as particulate diameter reduces.

Clarke(47) generated aerosols of bacterial spores of uniform particle size which was controlled by the addition of varying amounts of potassium iodide. The volume of air sampled by the RCS was calculated and the results compared with a slit to agar sampler. The effective sampling rate of the RCS varied from 1.2L/min. for 0.7 &#181m diameter particles, to 166L/min. for 17.3 &#181m diameter particles. If a flow rate of 40L/min. is used in-place of the actual or measured flow rate when samples are being taken, the concentration of particles <4.0 &#181m will be underestimated, and the concentration of particles >4.0 &#181m will be overestimated(11, 45, 46, 47). This may have led to discrepancies in previous reports of the efficiency of the RCS against slit to agar systems, with some workers concluding that the RCS was more efficient (10, 12, 32, 34), one worker(48) concluding that the two systems were equally efficient, and yet other workers concluding that the RCS discriminated in the size of the particulates retained(35, 45, 47).

Since the accuracy of the results obtained by the RCS depends upon knowing the size of the particles, and therefore the air volume being sampled, and that the particle size distribution in air samples is not routinely determined, Kaye(46) concludes that the RCS is not a quantitative device because the effective air volume sampled by the RCS is not known.

Macher and First(45) conclude “that the RCS is not the sampler of choice for quantitative estimates of microbiological aerosol concentrations, except where the particle size of the aerosol is known and the collection efficiency of that size particle has been measured.” Trudeau(57) in a recent study of agents involved in sick building syndrome concluded “the RCS is not accurate quantitatively, especially for small particles.”

The RCS+ is a modified version of the RCS, the main modification is in the design of the sampling head. Air is drawn into the system through the front of the head and impacted onto the agar strips around the inside of the head. Air is sampled at a flow rate of 50L/min. which then exits through exhaust ports on the rear of the sampling head.

Benbough(6) compared the RCS+ with the Casella slit to agar sampler and found that the RCS+ is substantially more effective at sampling submicron particles than the original RCS. However the RCS+ was shown to be only 54 percent as effective as the STA. for particles of 0.7 &#181m diameter, and only 67 percent as effective for particles of 1.8 &#181m diameter when tested against spores of Bacillus subtillis. Recent work at CAMR, Porton Down, UK,(51) has shown the RCS+ to be significantly less effective than both the slit to agar and gelatin membrane filtration system when sampling airborne log phase Eschericia coli where 91 percent of the viable E.coli generated were present on particles of less than 2.1 &#181m diameter.

Ljungquist and Reinmuller(52) have recently reported on the effects of both RCS and RCS+ samplers on air patterns in unidirectional air flow systems. The RCS causes air to move in a turbulent mixing manner, which means that the unidirectional airflow pattern around the sampler is heavily disturbed. The RCS+ sampler also disturbs unidirectional air flow, the two exhaust ports generate exhaust flow rates of 45 cm/s from the back of the sampling head which creates a vortex mixing pattern, that extends out from the rear of the sampler and disturbs the airflow pattern. Ljungquist concludes that it is necessary to evaluate the location of both the RCS and RCS+ samplers prior to use in unidirectional airflow patterns, that the RCS+ is preferable to the RCS, but that both cause disturbance of the airflow pattern.

Membrane filtration

Many workers have investigated the use of membrane filtration as a method for the capture and detection of airborne microorganisms. Goetz(55) summarized work undertaken in the early 1950s using molecular filter membranes to assess the absolute numbers and viability of airborne bacteria, and concluded that membrane filters were capable of retaining almost 100 percent of the airborne bacteria sampled. Goetz also concluded that membrane filters exhibited better viable recovery of a spore former, (B Subtilis) and equal or better recovery of a stress tolerant organism (Staphylococcus aureus), than an all glass impinger tested under identical conditions.

When tested against a stress sensitive organism (Serratia marcescens)(55), membrane filtration exhibits a lower recovery rate than an impinger, probably due to desiccation effects on organisms captured on the surface of the membrane. Jensen(11) also demonstrated the effects of desiccation on the survival rate of a stress sensitive organism (E. coli), compared with a stress tolerant organism (B Subtilis), when both were captured on a membrane filter. Other authors have also documented the effects and subsequent problems in accuracy caused by desiccation of organisms captured on membrane filters(7, 33, 55, 56, 58, 60, 61) when compared to other air sampling devices.

Early work (62, 63, 64, 65, 66) concluded that the use of a gelatin foam filter gave significantly higher recovery rates than membrane filters over the same sampling period. Sartorius AG, a German filtration company began commercial production of gelatin membrane filters in the mid 1970s. The gelatin membrane is a microporous membrane with a pore rating of 3.0 mm. The membrane has been shown to be 100 percent effective in retaining 0.5 &#181m latex particles(78), 99.9995 percent efficient for the retention of aerosols of Bacillus subtillis var niger(72) and 99.9 percent efficient in the retention of T1 virus aerosols(79).

The higher than expected efficiencies of capture demonstrated by a 3.0 &#181m pore size membrane on sub-micron particles is explained by the types of separation mechanisms that take place within a membrane filter. Particle removal occurs by the following mechanisms: inertial impaction, diffusional interception and direct interception. The role and importance of each mechanism changes according to the filter type and the fluid being filtered, but inertial impaction and diffusional interception mechanisms explain how a filter can remove particles much smaller than the stated pore size.

The gelatin membrane is supplied sterile packed in either a single plastic pack for use in non-critical areas, or triple bagged for use in critical areas. The sampling head is mounted either directly onto the front of the MD8 air sampler, or onto the end of an extension hose which allows sampling to take place at distances of up to 10 meters from the pump system. The MD8 system is self regulating to constantly maintain the correct airflow to ensure that the selected air volume will be sampled, both the time of sampling and the volume of air being sampled can be varied and the system is easily calibrated by use of an external reference system.

Following sampling, the gelatin membrane can be directly plated onto the surface of an agar plate. The gelatin dissolves into the surface of the agar and allows microorganisms to grow directly on the nutrient media, this is referred to as the direct processing method. In the indirect method the gelatin membrane is dissolved in a sterile solution, designed not to cause osmotic shock to the sampled microorganisms, and is subsampled from that solution to allow, for example, removal of inhibitors from the sample, dilution of the sample where high counts are expected or to allow plating onto different nutrient media where species identification is required.

Hambraeus(67) showed in work on anaerobes that gelatin membranes were significantly more efficient at recovering anaerobic bacteria than both a slit to agar sampler, cellulose membrane filters and an N-6 Andersen sampler when tested against the same organisms and under the same conditions. Work in Europe(25, 68, 69, 70) has investigated the suitability of using gelatin membranes for sampling airborne microorganisms and concluded that the use of gelatin as a capture media for microorganisms is beneficial in helping to reduce desiccation effects. Scheuerman(71) investigated the preparation, handling and analysis of gelatin membranes and discussed factors affecting the results obtained.

Macher(43) reported that Bacillus subtillis could be recovered at higher rates using gelatin membrane filters than by using an all glass impinger, but that the gelatin membrane could not detect viable Eschericia coli which could be recovered by the impinger method. The E. coli aerosols generated for Macher`s tests were nebulized from water suspensions, no data was presented to show the survival rate for the organisms under test in the environment. It has been previously demonstrated(51, 75, 76, 77) that E. coli generated from water suspensions have a very short survival time once aerosolized. Recent work (51) has shown the ability of gelatin membranes to collect and maintain the viability of log phase Eschericia coli over prolonged sampling and storage periods, where the E. coli was nebulized from nutrient broth. Work at the Center for Applied Microbiological Research in the UK(72, 73) comparing the efficiency of gelatin membranes with other methods, has shown gelatin membranes to be equally as effective as slit to agar samplers, irrespective of associated particle size, but have conclusively demonstrated that gelatin membrane filtration is significantly more effective than the RCS+ in collecting organisms associated with particulate sizes below 5.0 mm.

The MD8 can be used with gelatin membrane filters to sample airborne microorganisms under both isoaxial and isokinetic conditions. Sampling of non-viable particulates in the air are required to be sampled under isokinetic conditions(2) in unidirectional systems, similarly viable particulates should also be sampled under isoaxial and isokinetic conditions because microorganisms are associated with airborne particles of varying sizes. This is especially true in unidirectional air flow systems as found in laminar flow hoods and in some glovebox and isolator systems.

Isoaxial sampling is defined as, “A condition of sampling in which the direction of the airflow into the sampling probe inlet is the same as the unidirectional airflow being sampled.” Isokinetic sampling is further defined as, “The condition of isoaxial sampling in which the mean velocity of the air entering the probe inlet is the same as the mean velocity of the unidirectional air being sampled.” Under isokinetic sampling conditions all microorganisms and associated particles will be sampled equally, irrespective of size or mass, the results that sampling under these conditions provides are representative of the total viable count in the volume of air being sampled.

Where the air velocity into the sampling probe is lower than the velocity of the air being sampled smaller particles with less mass and inertia will pass around the sampling probe and the air sampled will contain a higher than representative number of larger particles and associated bacteria. Where the air velocity into the sampling probe is higher than the velocity of the air being sampled smaller particles will be selectively sampled and the sample will contain a lower than representative number of larger particles. Under both conditions the sample collected is not representative and is inaccurate.

Recent work in the UK(81) has shown that when different air samplers were tested in laminar air flow systems, and flow patterns were visualized by smoke trains the MD8 and gelatin membrane filter system produced no turbulence in the smoke train generated prior to sampling. Turbulence was caused by the Biotest RCS+ sampler, primarily by the exhaust air generated immediately behind the sampling head. The slit to agar (STA) system tested also demonstrated a significant amount of turbulence, eddies were formed around the sampler, the air exhaust caused turbulence and the whole laminar flow pattern was disturbed.


The gelatin membrane filter method is a new technique for the quantitative recovery of airborne microorganisms, that allows the sampling of areas of low bioburden, under laminar flow conditions without affecting the accuracy of the results due to desiccation of captured organisms, or passage of captured organisms through the membrane.

Recent work has shown it to be accurate for recovery of airborne stress-sensitive as well as stress-tolerant bacteria associated with a range of particulate sizes and contamination levels. Its accuracy has also been demonstrated for sampling of airborne virus particles under a range of temperature, humidity and air velocity conditions.

Derek Pendlebury, Ph.D., is the director of sales and marketing for the Laboratory Separation Products Division of Sartorius Corp. in Edgewood, NY. He received his bachelor of science degree in biological sciences from Coventry University, a master of science degree in pollution control from the University of Leeds, and his doctorate in marine biology from the University of Manchester. He joined Sartorius UK in 1986, and transferred to Sartorius North America in 1994. He can be reached at Sartorius`s U.S. location at (516) 254-4249.

David Pickard, Ph.D., is services manager at Griffiths Analytical UK in Somercotes, Derbyshire, UK. He received his bachelor of science degree in microbiology from the University of London. He is a Fellow of the Institute of Biomedical Sciences, an elected member of the International BSI Committee for New Draft Cleanroom Standards, and is a lead assessor for ISO 9001 Quality Systems. He can be reached at Griffiths Analytical in the UK at 01773 830505.


For a complete listing of the 81 sources referenced in this report, please visit our World Wide Web site at www.cleanrooms.com. If you are unable to access our Web site, please contact our Editorial Assistant Sheri Kovalchek at (603) 891-9429 and ask her to send you a copy, or fax your request to Sheri`s attention at (603) 891-9200.

Click here to enlarge image

Click here to enlarge image


Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.