Recent developments in passive, or unpowered dust monitoring, utilising ‘sticky pad’ samplers, have facilitated the monitoring of dust in flux on the pathway between source and receptor, and in deposition at the receptor. When combined with established procedures for elemental analysis and data handling, site specific source attribution studies become possible.
‘Dust’ is used in BS 6069 Part 2 (BSI, 1994) to define particulate matter < 75µm in diameter. Smaller particles are associated with impacts on health, through inhalation, and coarser particles with annoyance, through public perception. The size fraction essentially up to 10µm is referred to as PM10. Concentrations of PM10 are used as an indicator of local air quality through the National Air Quality Strategy (NAQS) and associated Air Quality Objectives (AQO) (Defra, 2007).
Coarser fractions of dust are harder to define but are often referred to as ‘nuisance dust’, an expression usually taken to mean generally visible particulate matter. The impacts of coarser fractions of dust occur as both acute and chronic phenomena, from the immediate visible impact of dust clouds to long term surface soiling caused by dust accumulation.
Source apportionment of PM10 dust is highly relevant to pollutant dispersion modelling, studies of local air quality and site specific emission monitoring. Emission inventories, including the National Atmospheric Emissions Inventory (NAEI) focus on PM10 and below. Emissions of coarser dust are not included (AEA, 2008).
Quantification of emissions from industrial processes such as construction, mining and quarrying is especially problematic as emissions are frequently directly into the atmosphere rather than as controlled releases such as stack emissions (AEA, 2008).
Passive, ‘nuisance’, dust sampling methods are considerably less expensive than those targeting a specific fraction, e.g. PM10 (Environment Agency, 2004). Directional and non directional samplers are frequently used at or near industrial sites to assess dust propagation. In some circumstances samples can be characterised by different geochemical techniques depending on the determinant, e.g. gas chromatography – mass spectrometry (GC-MS) for organic compounds such as Polychlorinated biphenyls (PCBs), or inductively coupled plasma-mass spectrometry (ICP-MS) for specific metals such as lead (Pb) and arsenic (As). (Environment Agency, 2004).
Dusty forensics methodology
Passive monitoring of nuisance dust may be achieved by many techniques, one of the more widespread of which is ‘sticky pad’ collection. Such systems use self-adhesive collection slides mounted on cylinders to collect dust in flux, in the vicinity of industrial sites such as quarries, surface coal mines, landfill sites and construction and demolition sites, e.g. Figure 1, Walton and Datson, 2010. A principal advantage of such systems is that the dust is collected on a directional basis, the pattern of accumulation indicating the direction from which it came.
The sticky pads are manufactured by specialist suppliers from stock material and comprise three principal layers: • A transparent PVC film • A hot-melt adhesive • A silicone coated paper liner – removed at the start of monitoring
Samples are normally taken over weekly or fortnightly intervals then sealed, scanned and, as illustrated in Figure 2, are analysed with bespoke computer software to report percentage Absolute Area Coverage (%AAC) and percentage Effective Area Coverage (%EAC) at 15° directional intervals (Datson and Birch, 2006). AAC represents the dust coverage of a surface irrespective of colour; EAC is a measure of dust soiling, or the loss of reflectance of a surface.
Sub-samples of the sealed sticky pads can be taken and the dust may be released from the adhesive and collected for gravimetry. Many geochemical analyses are possible, the best developed being elemental analysis using ICP-MS. Sub-samples are typically subjected to a hydrofluoric/nitric acid (HF/HNO3) attack and analysed for a range of metals (Datson and Fowler, 2007; Datson et al, 2011). The main advantages of ICP-MS are its rapid, multi-element capability and very low instrumental detection limits for most elements – parts per billion and below.
Results
The combined methodology provides several complementary data sets, and for all but the smallest studies, several large data tables are produced. The bespoke computer software quantifies dusting intensity to a 15˚ directional resolution. The gravimetry and elemental analysis provide quantitative estimates of dust chemistry. These may then be manipulated on a site specific basis to constrain the source of any problematic dust propagation, and the proportion of the problem dust among that from other local sources.
Dust coverage and directional data A typical example of the directional data is presented in Figure 3 on a site map using dust rose diagrams centred at the sampling locality. Bar length is proportional to dusting intensity using AAC or EAC measurement. In this way, a rapid overall assessment of dusting direction and intensity on site can be gained for any given sampling interval, and appropriate sub-samples taken for analysis.
Elemental analysis Usually, and as would be expected from crustal element averages, aluminium (Al), sodium (Na), potassium (K), magnesium (Mg), calcium (Ca) and iron (Fe) are the most abundant metals, but silicon (Si) is not analysed since it is volatilised during the hydrofluoric digestion. All others are present at rather lower abundance, sometimes below the method detection limit.
Data handling
It is always difficult, and often impossible, to use basic arithmetical methods to sufficiently interrogate a tabulated dataset that contains a significant number of analyses; some more advanced form of data handling is usually necessary. The methods chosen depend on the nature of the data and the type of information required from the exercise.
Sometimes, simple bar charts or bivariate element plots are sufficient to make the point. Experience has shown, however, that for the purposes of dust source attribution a variety of multi-element diagrams are useful, especially when organised to provide elemental ‘fingerprints’ of specific dust types. These depend on pre-existing knowledge with which to select the samples to differentiate. A less judgemental approach uses multivariate statistics, for which no preconceptions are required.
Crustal enrichment factors Crustal enrichment factors are commonly used in aerosol literature as a way of summarising the overall elemental chemistry of dust samples and emphasising any departures from typical crustal abundances. A sample with the same composition as the Earth’s crust would have an enrichment factor of one.
Figure 4 gives examples from several quarries. It is not surprising that chalk and dolomite associated sites show similar elemental chemistry. Both show significant major element enrichments in Ca (and K), and the dolomite site in Mg. Of the trace metals, both are enriched in chromium (Cr), cobalt (Co), copper (Cu), As and Pb, the dolomite site also in cadmium (Cd). Likewise, the igneous-related sites show similar pattern shapes, although enrichment/depletion might vary. As an example, the major element shape from Na to Ca is similar, yet unsurprisingly the granite site has significant depletion in Mg, whereas the dolerite site does not.
Interestingly, both these sites show a considerable peak at As: both are located in South-West England, an area that has well known regional enhancement in As resulting from the historic mining activity in the area. Such diagrams emphasise the recognisable elemental chemistry of individual sites, which may be manipulated to quantify relative contributions of ‘problem’ and background components.
Multivariate statistics Principal Component Analysis (PCA) allows an objective treatment of the variance inherent in a dataset, rather than its interrogation in the light of expected and previously characterised end members, however successful the latter might be. Figure 5 shows the scores plot of data from a landfill site (Fowler et al, 2010). The top right quadrant (positive pc1 and positive pc2) includes samples enriched in Cu, Pb, Cd and As relative to the average. The bottom right (positive pc1, negative pc2) represents relative enrichment in Al, Fe and Mg. The data define a wide range, occupying parts of all four quadrants.
Three potential vectors are drawn through the data, corresponding (vector 1) to a population enriched in Ca, Cu, Pb, Cd and As; another rich in Al, Fe, Mg, titanium (Ti) and manganese (Mn) (vector 2); and a third depleted in all elements (vector 3). Grab samples of APC residue, Lias clay and local soil clarify the significance of vectors one and two. Vector three can be understood as low dust coverage on the sticky pads.
Quantitation and source attribution
As outlined below, methods of quantitation and source attribution include bivariate ratio-ratio plots and mixing models and Partial Least Squares (PLS) analysis and chemical mass balance (CMB).
Bivariate ratio-ratio plots and mixing models On the basis of sample characterisation as described above, critical element ratios may be selected that stretch the data distribution between two or more end members and arithmetic binary mixing lines superimposed to provide quantitation estimates. Each of the examples in Figure 6 has two potential source compositions (end members of the mixing calculations background and site specific), several sampling locations at variable distance from the site, and a number of sampling intervals each producing a data point. It is clear from both diagrams that the sample localities fall in rational, overlapping groups, from which sensible site conclusions may be drawn.
PLS analysis and CMB Rather more sophisticated approaches are possible with larger data sets. An extension of the multivariate approach, for example, uses Partial Least Squares analysis to seek the elemental signature of a problem dust and the local background within a PCA dataset. A variety of least squares mixing software packages provide variations on a theme of chemical mass balance. In these, iterative solutions to multi-element source mixing are provided, which minimise the residual differences between sample and model. Figure 7 shows the PLS quantitation plotted against CMB data for the hazardous waste landfill dataset, showing a strong correlation between the two.
Dust propagation mapping
Once the analytical data have been transformed into a quantitative dataset by one or more of the methods outlined above, they can be displayed graphically in order to display periodic dust distributions. A number of software packages are available for such purposes, the most robust using variography and kriging procedures. Figure 8 shows a modified version of a fortnightly dust map, generated through a calendar year study at one site. Geostatistical modelling such as this also provides empirical data to complement theoretical dispersion models often used on and around industrial sites.
Discussion
The generation of fugitive dust from industrial processes and sites is a widespread phenomenon and considerable expense is incurred in dust suppression. Even so, disputes arise on occasion between a site operator and the local community, during which it might be assumed that all dust in the vicinity is produced by the site in question. This is not necessarily the case, and directional monitoring of dust emanation, combined with the possibility of quantifying site dust proportion, offers a significant step towards settling such disputes and away from ‘tick box’ surveys that might be more common practise.
The methods described above combine a cost effective directional passive dust sampler with sophisticated laboratory analysis and data handling, which together offer a powerful source apportionment method. The approach is typical of ‘environmental forensics’ – the application of investigative environmental science to real or potential pollution events. As is the intention of the ever tightening legislative and regulatory regimes, it would be preferable to be proactive and to feed data generated in this way back into site planning and operational procedures, thus minimising the actual impact on the environment rather than dealing with it subsequently.
Conclusions
1. The combination of passive, directional monitoring and ICP-MS elemental analysis provides a powerful tool for the source apportionment of fugitive dusts from industrial sites.
2. Elemental fingerprints may be derived that effectively separate problem dust from background.
3. Data manipulation by simple or sophisticated methods allows quantification of relative proportions in individual samples.
4. With suitable data sets, quantitative data thus derived may form the basis for spatial investigation and ‘dust mapping’.
Acknowledgements
We are grateful to Professor Geoffrey Walton for his continued encouragement of this work, and to the various clients and funding agencies who have contributed to its progress.
Published: 05th Sep 2013 in AWE International