Review Application of recent advances in aerosol sampling science towards the development of improved sampling devices: the way ahead† James H. Vincent,*a Gurumurthy Ramachandran,b Yngvar Thomassenc and Gerald J. Keelera aDepartment of Environmental and Industrial Health, School of Public Health, University of Michigan, 109 S. Observatory Street, Ann Arbor, MI 48109, USA bDivision of Environmental and Occupational Health, School of Public Health, University of Minnesota, 420 Delaware Street S.E., Minneapolis, MN 55455, USA cNational Institute of Occupational Health, P.O.Box 8149 Dep, N-0033 Oslo, Norway This paper reviews the framework that underpins the development of a new generation of personal samplers capable of operating at much lower flowrates that those of the current generation and so capable of being used for exposure assessment not only for ‘traditional’ occupational populations (i.e., industrial workers) but also for people exposed to aerosols in the ambient atmosphere (including children). The opportunity for this new generation of samplers stems from the availability of very light and compact low-flowrate pumps.The development and deployment of such instruments presents: (a) physical challenges in terms of how to collect particle size fractions in a manner which is consistent with the new particle size-selective sampling criteria, and (b) analytical challenges in terms of how to quantitate the much smaller amounts of collected material that need to be analysed.The paper lays out the physical and analytical scenarios, and points the way forward to how such challenges can be overcome.Work is already in progress in several countries to develop prototype instruments for applications like those described. the occupational setting, such application of personal sampling 1. Introduction has underpinned most exposure assessments in most developed Aerosol samplers are important to occupational and environ- countries during most of the past two decades.This has been mental hygiene through their role in the measurement (and successful because of the availability of appropriate equipment hence regulation) of exposures of workers and the general (see below) and the willingness of workers to wear it for population to airborne contaminants. In the context of occu- extended periods while going about their daily tasks.pational hygiene, proposed new health-related, particle size- Similar health concerns drive the study of exposures to selective aerosol standards require that such measurement ambient particulate air pollution (widely referred to as ‘particushould reflect the true physical nature of human exposure (i.e., late matter’ or ‘PM’), specially for the more vulnerable memthe manner in which they are inhaled and penetrate into the bers of the population (e.g., children and the elderly).6–8 The respiratory tract).1,2 This, in turn, has stimulated the search measures of PM used in the studies cited were obtained using for new generations of practical sampling devices, and develop- fixed point samplers placed at various locations within the ment of these—now and in the future—should be facilitated region of interest.But it is widely agreed that such studies are by the improved knowledge about their basic performance of limited use because of the fallacy inherent in estimating characteristics derived from recent research from several invesindividual risk from group data; that is, the concentration of tigators.That research has included both mathematical and PM collected from a sampler in a metropolitan area may not semi-empirical studies of particle motion in the complex flows be a reasonable proxy for personal exposure to PM.9–12 Not about bluV bodies where there is aspiration (i.e., sampling by only do particle concentrations vary spatially over a geographi- suction at one or more orifices).3,4 cal region with a central site measurement not adequately The National Occupational Research Agenda (NORA) representing that variability, but such measurements do not currently being promoted by the US National Institute of represent personal exposures of a population that spends a Occupational Safety and Health (NIOSH) recognises the large fraction of its time indoors.Here, it has been shown that continuing need to address chronic lung disease associated most people spend 90% of their time indoors inside their with aerosol exposures.5 Most occupational hygienists now residences, at work, or elsewhere.13–16 As a result, central site agree that the most eVective way to control and regulate monitoring in some areas might be better than others in this exposures to airborne contaminants is to use personal samplers regard.From this experience, it is clear that the problems of which measure the individual exposure of the worker chosen exposure assessment for airborne contaminants in the ambient to wear the sampler (which in turn would be chosen to be living environment become even more complicated than for representative of a group of similarly exposed workers).‡ In workplaces.This paper describes how the research that has been carried †Presented at AIRMON ’99, Geilo, Norway, February 10–14, 1999. out in the field of aerosol sampling science has led us to the ‡On a somewhat philosophical point, some occupational hygiene researchers have commented (including during the Geilo Symposium) point where we can prescribe a new generation of aerosol that, at low exposure concentrations, aerosol concentrations near the sampling devices that meet the needs of occupational hygiene worker might be dominated by particles that are released from the and community air monitoring.Specifically, it sets out to worker’s clothing or body. It then therefore becomes, it is said, a identify how the knowledge gained in previous research may question of the extent to which the personal sampler is measuring the be applied to the development of new small-scale, lightweight appropriate health-related aerosol concentration.This is an interesting point, and one deserving of future consideration. and user-friendly personal sampling systems for the inhalable J.Environ. Monit., 1999, 1, 285–292 285and thoracic fractions of airborne particles, suitable for use in the first time, such small pumps into sampling systems for the routine measurement of personal aerosol exposures. the occupational setting. The same body of knowledge may also be applied to devise sampling systems for measuring During the last two decades there has been great progress towards the setting of scientifically-based criteria for aerosol personal exposures to aerosol fractions such as PM10 and PM2.5 in the general population.measurement in the workplace and elsewhere, led by the American Conference of Governmental Industrial Hygienists (ACGIH) together with the International Standards Organisation (ISO)19 and the Comite� Europe�en Normalisation 2.Background (CEN).20 All three bodies have moved during recent years towards international harmonization of particle size-selective The physical problem of health-related aerosol sampling involves the aspiration of the inhalable fraction through the criteria for health-related aerosol sampling.2 Such criteria are intended specifically as ‘yardsticks’ for the design of new entry of the sampling device and, where appropriate, the aerodynamic pre-selection of the desired subfraction (for sampling instrumentation.They focus attention firmly on the need for sampling to directly reflect the true physical nature example, the thoracic fraction). The same approach can be taken with sampling aerosol fractions such as PM2.5 that are of the human exposure.First, they identify the inhalability of the human head itself, representing the eYciency, as a function of regulatory interest in the US. In either case, the selected health-related fraction is usually collected on a filter that can of particle size, with which particles enter through the nose and/or mouth during breathing. This defines the inhalable be subsequently analyzed after sampling.Such analysis make the form of gravimetric assessment (i.e., weighing) to fraction. Then they identify the thoracic and respirable fractions as subfractions of that inhalable fraction. Here, the thoracic determine the total mass of particulate material collected and/or chemical quantitation to determine the mass contained fraction describes the probability that an inhaled particle may penetrate into the lung below the larynx, and the respirable within specific chemical species considered most relevant. So-called ‘area, or microenvironmental, sampling’ is carried fraction the probability of penetration down to the alveolar region.Each of the three conventions is represented by a single out using instruments which sample air from the general environment in the general vicinity of the persons believed to curve describing it as a function of particle aerodynamic diameter.These are expected to form the basis of future be exposed. This was the approach commonly adopted in the years before the 1970s when sampling required an attendant occupational exposure limits (OELs) for aerosols, and, indeed, are already being applied in some places.Most future OELs operator and the required equipment was inevitably rather massive and required an external power source (and so was for substances occurring as aerosols are expected to be expressed either in terms of the inhalable or the thoracic far from portable). ‘Personal sampling’ only emerged with the advent of samplers and (particularly) pumps that could be fractions. The former will be applied for substances which are carcinogenic or may present a risk to health following systemic miniaturized to the point where they could conveniently be carried by, or worn on the person of, the worker.As related uptake from anywhere in the whole respiratory tract. The latter will be applied when the health eVect is restricted to the in a recent historical account by Sherwood,17 the possibility of small sampling pumps first became apparent in the late respiratory tract, including asthma, chronic bronchitis and chronic obstructive pulmonary disease (COPD).The PM10 1950s when health physicists at the UK Atomic Energy Research Establishment at Harwell realised that small con- curve, defined by the US Environmental Protection Agency (EPA), closely follows the thoracic convention, and is based stant-speed dc motors, like those developed for the early electrical phonographs, might be applied to air pumps much on a similar rationale.With the introduction of these new criteria for aerosol sampling and exposure assessment comes smaller than anything which had previously been conceived. The first practical sampling device based on this new concept the need to revisit existing OELs and, where appropriate, to conduct new exposure assessments for the purpose of new was reported in 1960 (Sherwood and Greenhalgh18).The first commercial personal sampling pumps subsequently appeared epidemiologic investigations. Returning to the question of aerosol sampling, the adoption around 1962, and by the early 1970s were becoming routine occupational hygiene tools.Now, 20 or so years further on, of criteria like those described is stimulating the re-assessment of existing instrumentation as well as the search for new ones. personal sampling is widely regarded as the only satisfactory way to assess the exposures of a workforce. Pumps that can So far, just one personal sampler21 and one area sampler22 have been designed specifically to match the inhalability curve.deliver upwards of 2 Lpm for up to 12 h are now commonplace and widely used by occupational hygienists. Nonetheless, to Of these, only the personal sampler is currently available commercially to occupational hygienists (the so-called ‘IOM this day, such pumps are cumbersome (e.g., typically weighing between about 25 and 40 oz—or about 0.7 to 1.1 kg) and sampler’, from SKC, Eighty-Four, PA, USA and Blandford Forum, Dorset, UK).However a recent large laboratory study bulky, dictated in large measure by the batteries. So, they are considered inconvenient by certain segments of the workforce, of personal aerosol sampler performance, sponsored by the European Community, has identified a number of other sam- including older workers and women.Further, in the wider context, where there is concern about the exposures of people plers which also look promising.23,24 This opens up the real possibility that a range of options may soon be available to to aerosols in the general living environment, current personal sampling pumps are not suitable for many of the groups of industrial hygienists for inhalable aerosol sampling.A key feature of existing aerosol sampling systems that interest, including (again) older people and children. Personal sampling pumps are currently available which makes them less convenient for some users is the relatively high sampling flowrate that is required, typically 2 to 4 L min-1 provide flowrates down to 0.1 Lpm.These are much smaller than the ones referred to above, weighing only a few ounces and upwards. This involves the use of a personal sampling pump which is bulky and heavy—even for the modern equip- and small enough to fit in a breast pocket. Such pumps are widely used for gas and vapor sampling, where sampling itself ment that is currently available. As already mentioned, the smallest commercially-available pump that will deliver 2 Lpm is not influenced or biased by anything like the aerosol mechanical forces which dominate aerosol sampling, and where weighs about 25 oz (about 0.7 kg), and the more commonlyused ones weigh as much 40 oz (1.1 kg).Although some analytical methods are readily available for the determination of small collected samples.However, new knowledge about manufacturers supply much smaller and lighter pumps, these operate at much lower flowrates and have found application the physical performance characteristics of aerosol sampling heads now makes it possible to think about integrating, for primarily in sampling for gases and vapours where sensitive 286 J. Environ. Monit., 1999, 1, 285–292analytical methods have long been available for quantitating plers, a parallel eVort was conducted to extend our knowledge of how to model the aspiration eYciencies of blunt aerosol samples.SKC, for example, supplies a pump that can deliver between 20 to 225 mLpm for periods of at least 8 h against a samplers and relate such models to samplers of the type actually used in industrial hygiene practice. This was achieved pressure drop of 10 in of water.This pump measures 4.5 in×2.2 in×1.4 in and weighs about 5 oz (about 0.14 kg), so by formulating physically-based models with empirical coeYcients which were then estimated (using non-linear is genuinely ‘pocket-sized’ and ‘user-friendly’ to the group of potential wearers identified earlier. The problem with such a regression techniques) by reference to the experimental data that were available from previous research.Such models have pump for aerosol sampling, however, is that it is diYcult to maintain the desired particle size-selective sampling perform- been widely described in the literature,35,36 and so the details will not be elaborated in this paper. SuYce to say that they ance over such a wide range of sampling flowrate.However, the emergence of improved knowledge of the physics of the are based on the good understanding we have for sampler performance at particular orientations with respect to the sampling process, together with the development of more exquisite analytical methods for analysing small samples, wind, namely a=0°, 90° and 180° respectively (i.e., providing A0, A90 and A180). For these, A0 may be calculated from the points the way forward to a new generation of aerosol samplers.model of Vincent,3 and A90 and A180 from that suggested by Tsai and Vincent. To illustrate the types of model that have emerged, the latter has yielded 3. Scaling laws for aerosol samplers and their A90=1/[1+4(2.21St)(R/r)1/2] and A180=1/[1+4(4.5St) application (r2/R)1/3r-0.29] (2) Sampling for inhalable aerosol These equations have since been incorporated into more Performance of an aerosol sampler may be defined in terms complex expressions for the aspiration eYciencies of specific of a number of indices, the primary one of which, the aspiration sampling systems averaged over all possible orientations with eYciency (A), is that which defines the ability of a particle to respect to the wind (which is the most likely scenario in be drawn into the sampler.3 This scenario involves consider- occupational and environmental situations).The most relevant ation of the physics of particle transport in the complex air to the present paper is that which was developed for a personal flow near the sampling device.A may be expressed as the ratio sampler mounted on the body of a worker, where Tsai et al. cS/c0, where cS is the concentration of particles passing through developed the expression37 the plane of the entry orifice of the sampler and c0 is that in Apersonal=(0.4A0+0.2A90+0.4A180)-[0.4(A0-A90)/ the undisturbed air outside the sampler. A number of basic (32St0-0.97R0-0.75r00.69+1)]+[0.1(A0-A90)/(85St017.1R08.55r0 theoretical approaches have been taken to express the physics -1.12+1)] (3) of the aerosol sampling process.Such theoretical models suggest the general relation where St0, R0 and r0 are versions of the previously-used St, R and r adjusted to correspond to the particular situation where A=f (St, R, r, a, B) (1) the sampler is mounted asymmetrically on the worker’s body.which expresses the basic physical dependencies in terms of Such an expression provides a useful starting point for the dimensionless groups or quantities. Such a relation articulates development of scaling laws by which new sampling heads the classical concept that the behaviour of any physical system may be developed. must be independent of the system of units used to describe it With the above in mind, for equivalent personal sampling in the real world.It therefore becomes the starting point for between a new, small-scale sampler (N) and a current reference any discussion about dimensional and dynamic scaling between sampler (R), we can make the identity like systems. In eqn. (1), St=dae2c*U/18gd, R=U/Us and r= NApersonal�RApersonal (5) d/D.Here, St is the Stokes’ number which embodies the eVects of inertia, reflecting the ability (or otherwise) of a particle to From this we can seek the combination of dimensionless quantities, and in turn individual variables, that will describe follow a diverging or converging airflow. This, in most cases, is the dominant physical eVect governing how eVectively a the new sampler having the same performance as the reference sampler.In practice it is found that a very large number of particle can be drawn (or aspirated) into the sampler. In addition, dae is the particle aerodynamic diameter, U the such combinations is theoretically possible. It therefore remains to determine (a) how should the solution be con- freestream air velocity, Us the mean air velocity at the sampling inlet, d the width of the sampling inlet, D the characteristic strained (i.e., dimensionally or dynamically) based on the range of practical or feasible conditions for the desired out- dimension of the sampler, a the sampling inlet orientation with respect to the wind, and B an aerodynamic shape factor come, and (b) which of the possible desirable solutions is the best (e.g., is least sensitive to changes in one or more variables).(or ‘bluntness’). The latter embodies consideration of the shape of the external airflow outside the sampler, determined by the To do this fully involves a complex multi-dimensional minimisation approach,38,39 and we have not yet performed this for sampler body itself. In addition, c* is the density of water (103 kg m-3) and g the viscosity of air.In this description, it the specific problem in hand. However, by way of illustration of this type of scaling is assumed that the air in the general environment where sampling takes place is moving. Although this is generally exercise for the case of the personal sampler of interest here, it is useful to show here an example of how it might be relevant to most workplace situations,25 there are some situations where windspeeds are so low that particle transport in achieved for a specific instrument; say, the IOM inhalable aerosol sampler mentioned earlier, mounted on the lapel of the vicinity of the sampler is controlled not only by inertial eVects but also by gravity.26,27 an exposed person.The original instrument itself is shown in Fig. 1.It has a circular entry orifice with d=15 mm and, in We have carried out extensive field studies of occupational aerosol exposures during the past few years.28–34 Many of order to collect the inhalable fraction, it has been shown that it must operate at a sampling flowrate of 2 Lpm.21 For these studies were concerned with the eVects of the application in aerosol standards of the new particle size-selective sampling modeling purposes, we also need to prescribe D1 and D2 as the width and the thickness respectively of the cross-section criteria, and involved the use of several diVerent types of sampler.To support that eVort, and to enable scientific of the wearer’s body, and we assume that, typically at fullscale (for a ‘real’ person), D1#400 mm and D2#150 mm.The interpretation of the results obtained using the diVerent sam- J. Environ. Monit., 1999, 1, 285–292 287the conditions of these earlier experiments. Further, at very low sampling flowrates, it is likely that not only inertia but also gravity will play a stronger role on particle transport near the sampler entry. In this regard it should be noted that gravitational eVects do not feature in the model outlined above, and this underlines the need to include consideration of these in future such models.That notwithstanding, models like those described can provide useful guidance to the development of new samplers which, in any event, should be tested and validated both in the laboratory and in the field.Ongoing work in our laboratory and elsewhere is addressing the aspects raised here. The latter point is relevant to other related work in which the same scaling laws may be applied to sampling systems which have diVerent dimensions to the reference system of interest. Here, Ramachandran et al.40 have shown how this is particularly relevant to the development of aerosol sampler test procedures which can be carried out in small wind tunnels (thus eliminating the current need for such testing to be carried out at full-scale in very large wind tunnels like the one at the Institute of Occupational Medicine in Edinburgh, Scotland, UK, and others elsewhere).In turn, this opens the door to Fig. 1 The IOM personal inhalable aerosol. more rapid and cost-eVective testing procedures.aspiration eYciency curve for this ‘reference’ system, for up Sampling for thoracic aerosol to dae=90 mm and for a windspeed of 1 m s-1, is shown in Fig. 2 (see ‘R’), and it is seen to be well within the ±10% Previous sampling instruments for collecting finer aerosol tolerance bounds originally defined for the inhalability curve. subfractions have been based on a range of physical processes The goal now is to apply the scaling laws in relation to the by which particles may be separated or classified in terms of desired endpoint of achieving a sampler which can achieve the their aerodynamic diameter.3 The first of these included horisame particle size-selectivity for a much lower sampler flowrate zontal and vertical elutriation (utilizing the force due to (e.g., 0.1 Lpm).This solution too is shown in Fig. 2. It is seen gravity). Later devices were based on impaction (utilizing that the same sampler (i.e., same dimensions) but operated at inertial forces). But the most commonly used approach since the lower flowrate of 0.1 Lpm, provides an aspiration eYciency the 1960s has been the cyclone (utilizing centrifugal body characteristic curve which is very close to that for the higher forces), and this has been applied very successfully in miniatur- flowrate.ized personal samplers for the respible fraction. The advan- At first glance, this result is very surprising. There are two tages of this approach are that (a) the resultant sampler is comments by way of discussion. Firstly, there is some physical very compact and simple, and (b) it has been tried and tested support for the observed tendency.The results shown are for very widely in the practical industrial hygiene setting. Its the averaging of aspiration eYciency uniformly over all poss- disadvantage, however, is that its performance is highly depenible angles with respect to the wind, and this, in turn, is dent on the integrity of maintaining the accuracy of the cyclone considerably influenced by the contributions of rearward- dimensions, and hence the requirement for high manufacturing facing orientations where sampling flowrate eVects are weak- tolerances.est.35 Secondly, however, it is noted that the model on which Porous plastic foam media have received some attention in the scaling exercise was based was derived from the fitting of this regard and, in the first instance, are attractive because of a semi-empirical model to the available experimental data.So their flexibility, versatility, availability and, importantly, low such a model should be applied with caution, acknowledging cost. Fig. 3 shows an electron micrograph of the structure of especially an application which involves extrapolation beyond such material, revealing it as somewhat ‘fibrous-like’ and so having features that enable it, in the context of aerosol science, to be treated as filter media.With this in mind, the penetration characteristics of such media, and the application of such knowledge in industrial hygiene, were first described in the late 1970s.41 The first prototype of a personal sampler based on such foam media as a pre-selector was described in the late 1980s, and the ideas were developed further in the 1990s.42–45 Based on the available cumulative experimental data for the penetration of particles through such media, and the understanding that the physics of deposition inside such media is dominated by a combination of inertial and gravimetric forces, an empirical mathematical model embodying the penetration (P) of foam media was proposed,43 thus -(df/t)lnP=54.86Stf2.382+38.91Ng0.880 (7) in which Fig. 2 Application of the scaling laws towards identifying the design Stf=dae2c*Uf/18gdf (8) parameters for new low-flowrate aerosol samplers (of the form of the IOM inhalable aerosol sampler). Full scale; U=1.0 m s-1; Q=2 Lpm; Ng=dae2c*g/18gUf (9) d=15 mm; D1=400 mm; D2=150 mm; Us=0.188 m s-1.Small scale; where df is the physical dimension of the solid parts of the U=1.0 m s-1; Q=0.1 Lpm; d=15 mm; D1=400 mm; D2=150 mm; Us=0.009 m s-1. foam media structure, Uf is the face velocity of the air 288 J. Environ. Monit., 1999, 1, 285–292the non-occupational setting. Fig. 4 shows penetration curves for two cylindrical plugs of porous foam media of dimensions appropriate for installation in a small personal sampler like that already described above.Both are for foam media grades that are readily available commercially and have external dimensions that can easily be achieved in practical instruments. One of the systems described can provide performance to the desired curve at the flowrate of 2 Lpm, the other for the lower flowrate of 0.1 Lpm.It is interesting to note that the performance, in terms of the proximity of the calculated curve to the desired thoracic aerosol curve, is even better for the lower flowrate. This relates to the fact that, at the lower flowrate, particle penetration is controlled largely by gravitational eVects (as opposed to inertial eVects at the higher flowrate).Fig. 5 extends the preceding to the suggested parameters for a pre-selector which provides the PM2.5 fraction. Here, again, it is seen that—based on the model, at least—the potential for realising a practical instrument is strongly suggested. For both the thoracic/PM10 and the PM2.5 fractions, as for the inhalable sampler mentioned earlier, the model that has been described Fig. 3 Scanning electron microscope picture showing the microstructure of a typical porous foam media sample of the type used as pre- should be applied with caution, and used primarily as guidance selector material for an aerosol sampler. towards the development of actual practical instruments. Such approaching the foam media, and g is the acceleration due to gravity. Here, Stf is now the Stokes inertial parameter for describing the role of inertia in particle collection inside the foam media, and Ng is the gravitational parameter describing the role of gravitational deposition.More recently, Aitken and Donaldson46 examined the robustness of this model by comparing its results with a comprehensive set of new experimental data, and found that it held up quite well.In addition, they demonstrated that the consistency of foam media samples from the manufacturer, in terms of their intrinsic penetration characteristics, was very good, even for foam samples taken from diVerent batches. This was somewhat surprising since such foams are manufactured for applications (e.g., as packing and padding media) where high dimensional tolerances are not necessarily required.In addition, it was found that such foam media have good particulate loading properties, and that the eVects of particle blow-oV or re-entrainment can be eliminated by the use of oil impregnation. Elsewhere, in studies of the filtration characteristics of foam media in relation to its Fig. 4 Penetration characteristics of a pre-selector for a personal sampler for the thoracic fraction based on porous plastic foam filter potential practical applications in certain air cleaning situmedia, suggested by Equation (7). Results of calculations, and the ations, it has been shown47 that the physical characteristics of corresponding foam median characteristics, are shown for both the foam media obtained in the United States—in particular the higher-flowrate of 2 Lpm and the lower-flowrate of 0.1 Lpm.microscopic physical dimensions and bulk porosity for a given nominal ‘pores per inch’ (or ‘ppi’)—are very close to those for foam media from a diVerent manufacturer in Europe. This suggests a high degree of consistency in commerciallyavailable media. It is therefore concluded that porous plastic foam media have excellent potential for use as pre-selectors for particle size-selective aerosol samplers.Further, the semi-empirical model given in eqn. (7) provides a good working, at least starting, basis for the design of practical devices, including samplers for the thoracic fraction. It is particularly appropriate that the combination of physical mechanisms for particle collection in the foam media (inertial impaction and gravitational settling) is the same as that governing penetration into the lung.The first lung penetration fraction of interest is the thoracic fraction since it is expected that, in the relatively near future, OELs for some, perhaps many, substances will be expressed in terms of this fraction. It is, of course, the same in principle, and very similar in reality, to the PM10 fraction which is Fig. 5 Penetration characteristics of a pre-selector for a personal identified by the US EPA as the primary metric for exposures sampler for the PM2.5 fraction based on porous plastic foam filter to ambient airborne particulate matter. So the preceding media, suggested by Equation (7). Results of calculations, and the applies directly to the measurements of personal exposure that corresponding foam median characteristics, are shown for the lower- flowrate of 0.1 Lpm.environmental scientists and hygienists might wish to make in J. Environ. Monit., 1999, 1, 285–292 289prototypes should always be tested in the laboratory and in from about 10 to 50 mg m-3 (annual average). For such concentrations, the mass of particles collected on sample filters the field. for the flowrate of 2 Lpm will have ranged from approximately 20 to 150 mg.For a sampling flowrate of 0.1 Lpm, this means 4. Analysis of collected samples that the PM2.5 equivalent collected mass would be of the order of from 1 to 10 mg. The advantages of the new approach to personal exposure An excellent study of the problems, pointing out the issues assessment are clear in terms of the greater convenience to the associated with such gravimetric assessment, has been reported wearer of the new generation of samplers, opening up the by Vaughan et al.49 Balances are available that have ‘read- possibility of less obtrusive instruments which can be worn by ability’ to as low as 1 mg, so that the standard deviation for a wider range of the exposed population, both occupational low mass measurements will be greater than that.Bearing in and non-occupational. However, there is a price for these mind that the limit of quantitation (LOQ) is generally regarded advantages in that the collected samples will inevitably be as of the order of 10× the standard deviation, it becomes smaller. For example, a reduction in sampling flowrate from apparent that, for some occupational settings and, probably, 2 to 0.1 Lpm, will result in a 20-fold reduction in the amount most non-occupational ones, straight gravimetric assessment of collected particulate matter to be analysed to any given of samples obtained using the new generation of low-flowrate particle size-selected fraction. This in itself presents a considersamplers does not appear to be a promising prospect, especially able challenge to the analyst who will be given the task of for the lower flowrates that are being discussed here.Here, quantitating the samples collected. Analysis falls into two therefore, we have a direct conflict between the desire to categories: gravimetric assessment to determine the total mass develop low-flowrate samplers and the diYculty of performing of particulate matter collected for the particle size fraction of straight gravimetric assessment of samples collected using interest, and chemical analysis to determine the mass of a them.It will not be easy to close that gap. given chemical (or sometimes biological ) fraction. Gravimetric assessment Chemical assessment: bulk analysis In workplace settings the chemical composition of the aerosol Gravimetric assessment of collected aerosol samples is, in some ways, even more problematic than chemical quantitation.is usually characteristic of the processes taking place there. Therefore it becomes feasible that the workplace atmosphere It is highly dependent first on the balance that is used and on the procedures by which the filters or substrates are con- can be ‘fingerprinted’ in terms of the proportions of given chemical species.These in turn may be related to the overall ditioned and prepared prior to weighing (both before and after sampling). For the latter, a major complicating factor is mass concentration (for given particle size fractions).If so, therefore, chemical analysis of a given chemical species—for the moisture uptake by which filters can change weight dramatically, even over periods of a few minutes. One approach example, a metal which can be analysed to much lower levels of quantitation (see below)—can be related to the desired to dealing with this is to condition samples overnight prior to weighing in a controlled balance room environment (e.g., overall mass.But, in any case, individual chemical contaminants may themselves be the direct object of concern (e.g., temperature and relative humidity), as is required by the EPA and is carried out in the Air Quality Laboratory at the lead, nickel, etc.). There is a very wide range of options for the measurement University of Michigan.Another is to desiccate filter samples overnight prior to weighing and to weigh immediately upon of individual chemical species. Metals form one group of interest, both for occupational and non-occupational expo- removal from the desiccator. The latter can be acceptable when appropriate environmental facilities cannot be available. sures. For these, the early history of occupational hygiene measurement was revolutionised by the introduction of atomic For example, we have used the latter for samples taken in many of our occupational hygiene field studies, and found it absorption spectrometry in the late 1950s.But further advances emerged with the development of simultaneous multi-element can be successful provided that a consistent routine is established.analytical techniques. For example, inductively-coupled plasma atomic emission spectrometry (ICP-AES) provides Another approach, of course, is to use filters which are less sensitive to such instability. In this regard, Teflon filters are improved detection limits and is rapidly overtaking atomic absorption spectrophotometry as the method of choice for the much better than membrane filters.Even filters made from polyvinylchloride (PVC) are an improvement. But dealing determination of multiple toxic metals and metalloids in workplace air.50 Similarly, inductively-coupled plasma mass with the instability of the filter itself in this way does not solve the problem of the filter holder or cassette which, in some spectrometry (ICP-MS) has provided even further improvements in terms of lowering detection limits.It too has become cases (e.g., the use of the IOM sampler) must also be weighed. Here, the eVect of filter instability is negligible in comparison a powerful tool for routine multi-element measurements, especially at low concentration levels.51 to that for the rest of the collection system. Here, therefore, it is not possible to obtain detection limits below about 50 mg.By way of illustration of what can be achieved using such current methodologies, the estimated detection power of ICP- Measurements of inhalable aerosol in the primary nickel production industries, made using the IOM sampler, showed AES in occupational hygiene is shown in Table 1, expressed in terms of the ratio between the OELs developed by the concentrations typically ranging from 0.3 to 20 mg m-3.48 Therefore, for an 8 h sampling period with a flowrate of 2 American Conference of Governmental Industrial Hygienists (ACGIH) [commonly known as ‘threshold limit values’ or Lpm, the mass collected therefore would have ranged from about 0.3 to 20 mg.If the sampling flowrate were scaled down (TLVs)1] and practical detection limits.Here, values of the order of 10 or less are problematical. It is evident that, for all to 0.1 Lpm, the masses collected would have ranged from about 14 to 1000 mg. Similarly, in monitoring of ambient metals except, possibly, lead, reliable exposure data can probably be obtained even when the air volume is reduced to 40 L PM2.5 levels, the mass of particles collected would be even smaller. With sampling flowrates of 2 Lpm and nominal (which is approximately equivalent to sampling 0.1 Lpm over a working shift).Similar calculations can be performed for sample collection times of 24 h, the nominal sample volume would be about 3 m3. By way of illustration, scientists at the ICP-MS, reflecting appropriate greater sensitivity. Using a specific industrial setting as an example, we have University of Minnesota have measured PM10 concentrations in Minneapolis ambient air, and found these to range typically recently collected inhalable masses of metals (cobalt, copper, 290 J.Environ. Monit., 1999, 1, 285–292Table 1 Estimated ratios between ACGIH threshold limit values (TLVs)a and detection limits for quantitation by ICE-AESb for a range of substances for diVerent sampled volumes corresponding to using the two sampling systems referred to in this paper over a full working shift.Elements with ratios less than 20 are problematical Sample air volume/L 1000 40 Detection limitb Metal ACGIH-TLVa/mg m-3 /ng ml-1 Ratio Cd 0.01 1 700 30 Co 0.02 1 1300 50 Cr 0.5 2 17000 700 Fe 5 10 34000 1300 Mn 0.2 1 13000 500 Ni(soluble) 0.1 2 3400 140 Pb 0.05 10 340 13 Massc 10d 10 mg 1000 40 a1998 ACGIH threshold limit values for chemical substances.bProcedure routinely used at NIOH, Oslo, for occupational air samples, using Perkin-Elmer Optima 3000 inductively-coupled plasma atomic emission spectrometer (ICP-AES), sample dilution volume 15 mL. cBy microbalance, for samples collected on 25 mm Teflon filters.dParticulates (insoluble) not otherwise classified (PNOC). nickel, and lead) at a copper electrowinning process of a nickel will, of course, have a bearing on the statistical interpretation of results for ensembles of particles). refinery ranging typically from 20 to 200 mg, for samples taken at 2 Lpm over a full working shift. Corresponding thoracic Such methods are of course relevant to the full characterisation of particulate air contaminants.But they can also be masses range from 5 to 50 mg. After scaling down to the lower flowrate, the collected masses may range from 1 to 10 mg. useful to the ‘fingerprinting’ of airsheds for the purpose of assessing very low levels of PM10 or PM2.5 if the purpose of Ambient atmospheric levels are lower still.Metal concentrations have been reported typically of the order of about sampling is to determine the overall mass contained in one or other of those fractions. Scientists at the University of 30 ng m-3 for lead and copper, and 120 ng m-3 for zinc and manganese. For 24 h samples, therefore, these translate Michigan are currently engaged in research to characterise particulate air contaminants in the ambient atmosphere in approximately to 100 to 400 ng of collected mass for each metal.By reference to Table 1, it becomes apparent that this Michigan and elsewhere in the Great Lakes region, involving the application of SEM-EDX to the quantitation of large falls outside what can be realistically achieved by ICP-AES. But such mass can be quantitated using ICP-MS.However, ensembles of particles collected at sampling stations throughout the State. By such methods, the aim is to develop key to reduce interference from background contamination, special facilities are required. Such measurements are possible, for ‘indicators’ by which the measurement of specific chemical constituents might relate to overall mass. example, in facilities like those in the Air Quality Laboratory at the University of Michigan.Here, the use of the ‘Class 100’ clean room enables the handling of samples under ultra-clean conditions, so that the quantitation of metals down to nanog- 5. The way ahead ram levels can be achieved by ICP-MS. This paper has outlined the framework that underpins the The aforementioned methods require preparation proproposed new generation of low-flowrate aerosol samplers, cedures involving dissolution of the filter.However, for driven, of course, by the availability of a new generation of on-filter analysis, X-ray fluorescence spectrometry (XRF) is miniature, lightweight sampling pumps. The science of aerosol an available option for measuring the elemental composition sampling, and the emergence of new analytical techniques that of particulate matter.Here, the high detection power of ‘totalallow the quantitation of very small samples of particulate reflection’ XRF makes this technique especially attractive in matter, provide many new options for the practical develop- the measurement of micro-mass aerosol samples.52 Techniques ment and implementation of such samplers.are currently being developed to eliminate the need to extract New research and development is exploring and validating the filters, and these use the electro-thermal vapourisation the various facets of the scenario that has been described. For (ETV) technique which allows direct injection of the content the physical aspects of sampler development, recently- of filters into ICP-MS apparatus.completed research has provided the scientific basis for the scaling laws and porous foam filtration models described Chemical assessment: single-particle analysis above. Now, new studies in both the United States and Europe are providing the opportunity to develop sampling instruments In addition to bulk analysis, individual particle information can provide detailed information concerning the origin, forma- along the lines described.That eVort will involve application of the models indicated to the design of new practical proto- tion, transport, reactivity, transformation reactions and environmental/health impact of exposure. The most relevant types, and their validation and testing in wind tunnels and in the field.In parallel with this eVort, another new project is and commonly used micro-beam techniques are electron probe X-ray microanalysis (EPXMA) and scanning electron aimed at applying what has been learned about aerosol sampling scaling laws to the development of test methods and microscopy (SEM), coupled either with energy-dispersive or wavelength-dispersive X-ray detection (EDX or WDX).The protocols which are smaller in scale and more cost-eVective than the present methods involving large wind tunnels and wealth of such topochemical methods that can be brought to bear on single particle characterisation has been reviewed by laborious experimental procedures. It is hoped that such an international eVort like that contained in this current body of Jambers et al.,53 Ortner et al.54 and many others.For such techniques, a large reduction in sampling flowrate does not research will enable truly positive advances in aerosol sampling methodology which are demanded by the emergence of the have any direct influence on their analytical performance and applicability as they relate to individual particles (although it various new aerosol standards, both for occupational and J.Environ. Monit., 1999, 1, 285–292 291Testing of Workplace Aerosol Sampling Instruments, Report of non-occupational environments. For the present, we can conwork carried out under EC Contract MAT1-CT92-0047, UK clude that there is a physical basis to guide the development Health and Safety Executive, SheYeld, England, UK, September of the new generation of particle size-selective sampling 1995.devices. 24 L. C. Kenny, R. J. Aitken, C. Chalmers, J. F. Fabries, New sampler physical development must be accompanied E. Gonzales-Fernandez, H. Kromhout, G. Liden, D. Mark, G. Riediger and V. Prodi, Ann. Occup. Hyg., 1997, 41, 135. by advances in quantitation methods, most notably the search 25 T. L. Ogden, in Aerosols in the Mining and Industrial Work for methods to measure yet smaller amounts of sampled Environments, ed.V. A. Marple and B. Y. H. Liu, 1983, Ann particulate material. Studies in both the United States and Arbor Science, Ann Arbor, MI, 1983, pp. 185–204. Europe are exploring the new options that are needed. Again, 26 S. Grinshpun, K. Willeke and S. Kalatoor, Atmos. Environ., 1993, for the present, it is noted that there is available a range of 27A, 1459.analytical methods by which to quantitate samples obtained 27 R. J. Aitken, J. Environ. Monit., in preparation. 28 P.-J. Tsai, J. H. Vincent, G. A. Wahl and G. Maldonado, Occup. using the devices in the new generation of aerosol samplers. Environ. Med., 1995, 52, 793. 29 P.-J. Tsai, M. A. Werner, J. H. Vincent and G.Maldonado, Appl. The authors wish to thank the following for their support of Occup. Environ. Hyg., 1996, 11, 484. the large body of the work described in this paper: the US 30 P.-J. Tsai, J. H. Vincent, G. A. Wahl and G. Maldonado, Ann. National Institute for Occupational Safety and Health Occup. Hyg., 1996, 40, 651. (NIOSH), the Norwegian National Institute of Occupational 31 M.A. Werner, T. M. Spear and J. H. Vincent, Analyst, 1996, 121, 1207. Health, the Nickel Producers Environmental Research 32 M. A. Werner, J. H. Vincent, Y. Thomassen, S. Hetland and Association (NiPERA), and the US Environmental Protection S. Berge, Occup. Hyg., 1999, in the press. Agency (EPA). 33 T. M. Spear, M. A. Werner, J. Bootland, A. Harbour, E. P. Murray, R. Rossi and J.H. Vincent, Am. Ind. Hyg. Assoc. J., 1997, 58, 893. References 34 P. W. Wilsey, J. H. Vincent, M. J. Bishop, L. M. Brosseau and I. A. Greaves, Am. Ind. Hyg. Assoc. J., 1996, 57, 1149. 1 American Conference of Governmental Industrial Hygienists, 35 P. J. Tsai and J. H. Vincent, J. Aerosol Sci., 1993, 24, 919. Threshold Limit Values for Chemical Substances and Physical 36 P. J.Tsai, J. H. Vincent, D. Mark and G. Maldonado, Aerosol Agents, ACGIH, Cincinnati, OH, 1998. Sci. Technol., 1995, 22, 271. 2 Particle Size-Selective Sampling for Particulate Air Contaminant, 37 P.-J. Tsai, J. H. Vincent and D. Mark, Ann. Occup. Hyg., 1996, ed. J. H. Vincent, ACGIH, Cincinnati, OH, 1999, in the press. 40, 93. 3 J. H. Vincent, Aerosol Sampling: Science and Practice, Wiley, 38 W.H. Press, S. A. Teukolsky, W. T. Vetterling and B. P. Flannery, London, 1989. Numerical Recipes in Fortran. The Art of Scientific Computing, 4 J. H. Vincent, J. Aerosol Sci., 1998, 29, 1085, suppl 1. Cambridge University Press, Cambridge, 1993. 5 US National Institute of Occupational Safety and Health 39 G. Ramachandran and M. Kandlikar, J. Aerosol Sci., 1996, 27, (NIOSH), National Occupational Research Agenda (NORA), 1099.Washington, DC, USA, 1998. 40 G. Ramachandran, A. Sreenath and J. H. Vincent, J. Aerosol Sci., 6 D.W. Dockery, C. A. Pope III, X. Xu, J. D. Spengler, J. H.Ware, 1998, 29, 875. M. E. Fay, B. G. Ferris and F. E. Speizer, New England J. Med., 41 H. Gibson and J. H. Vincent, Ann. Occup. Hyg., 1981, 24, 205. 1993, 329, 1753. 42 D.Mark, G. Borzucki, G. Lynch and J. H. Vincent, The develop- 7 D. W. Dockery and C. A. Pope III, Annu. Rev. Public Health, ment of a personal sampler for inspirable, thoracic and respirable 1994, 15, 107. aerosol, presented at the Annual Conference of the UK Aerosol 8 J. Schwartz and D. W. Dockery, Am. Rev. Respir. Dis., 1992, Society, Bournemouth, March 1988. 45, 600. 43 J. H. Vincent, R. J. Aitken and D. Mark, J. Aerosol Sci., 1993, 9 J. Schwartz, Environ. Res., 1994, 64, 36. 24, 929. 10 S. Richardson, I. Stu�cker and D. He�mon, Int. J. Epidemiol., 1987, 44 R. J. Aitken, J. H. Vincent and D. Mark, Appl. Occup. Environ. 16, 111. Hyg., 1993, 8, 363. 11 S. Piantadosi, D. P. Byar and S. B. Green, Am. J. Epidemiol., 45 P. Courbon, R. Wrobel and J. J. Fabries, Ann. Occup. Hyg., 32, 1988, 127, 893. 129. 12 S. Greenland and H. Morgenstern, Int. J. Epidemiol., 1989, 18, 46 R. J. Aitken and R. Donaldson, Development of Health-Related 269. Sampling Instrumentation for Assessing Workers’ Exposures to 13 K. Sexton, R. Letz and J. D. Spengler, Environ. Res., 1983, 32, 151. Aerosols in the Nickel Industry, Institute of Occupational 14 K. Sexton, J. D. Spengler and R. D. Treitman, Atmos. Environ., Medicine, Edinburgh, UK, final report to the Nickel Producers 1984, 18, 1385. Environmental Research Association, 1994. 15 J. D. Spengler, R. D. Treitman, T. D. Tosteson, D. T. Mage and 47 W. A. Poon, PhD Thesis, University of Minnesota, Minneapolis, M. L. Soczek, Environ. Sci. Technol., 1985, 19, 700. MN, USA, 1997. 16 D. W. Dockery, J. D. Spengler, M. P. Reed and J. Ware, Environ. 48 P.-J. Tsai, PhD Thesis, University of Minnesota, Minneapolis, Int., 1981, 5, 101. MN, USA, 1995. 17 R. J. Sherwood, Appl. Occup. Environ. Hyg., 1997, 12, 229. 49 N. P. Vaughan, B. D. Milligan and T. L. Ogden, Ann. Occup. 18 R. J. Sherwood and D. M. S. Greenhalgh, Ann. Occup. Hyg., Hyg., 1989, 33, 331. 1960, 2, 127. 50 O. T. Butler and A. M. Howe, J. Environ. Monit., 1999, 1, 23. 19 International Standards Organisation (ISO), Air Quality— 51 G. M. Hieftje, P. G. Galley, M. Glick and D. S. Hanselman, Particle Size Fraction Definitions for Health-Related Sampling, J. Anal. At. Spectrom., 1991, 7(2), 69. ISO CD7708, ISO, Geneva, 1992. 52 R. Klockenka�mper and A. von Bohlen, J. Anal. At. Spectrom., 20 Comite� Europe�en de Normalisation (CEN), Workplace 1991, 7(2), 273. Atmospheres—Assessment of Performance of Instruments for 53 W. Jambers, L. De Bock and R. Van Grieken, Analyst, 1995, Measurement of Airborne Particle Concentrations, 120(3), 681. TC137/WG3/N192, CEN, Brussels, Belgium, 1997. 54 H. M. Ortner, P. HoVmann, F. J. Stadermann, S. Weinbruch and 21 D. Mark and J. H. Vincent, Ann. Occup. Hyg., 1986, 30, 89. M.Wentzel, Analyst, 1998, 123(5), 833. 22 D. Mark, J. H. Vincent, H. Gibson and G. Lynch, Am. Ind. Hyg. Assoc. J., 1985, 46, 127. 23 L. C. Kenny, Pilot Study of CEN Protocols for the Performance Paper 9/02801C 292 J. Environ. Monit., 1999, 1, 285