Measurement of ‘heavy metals’ in particulate matter in outdoor air

Abstract

Air pollution remains a matter of concern for governments, regulators and the public. Particulate matter (PM) is an important pollutant in this respect, and its composition is thought to be the most critical aspect with regard to its effect on human health. In particular, the presence of ‘heavy metals’ in PM is known to have toxic effects. This Technical Brief describes the theory and practice of making measurements of metals in particulate matter in outdoor air. It also presents examples of the real-world benefits of these measurements and reviews some of the latest developments in this field.

---

Introduction

Pollutants in air remain a concern to governments, regulators and the public because of the detrimental effect they can have on human health and environmental sustainability. Of particular interest for human health is airborne particulate matter (PM), for two reasons: firstly, because particles with an aerodynamic diameter of 10 µm or less (PM₁₀) can enter the lungs via inhalation, and secondly, because of the range of toxic compounds the PM₁₀ can contain. A priority is the metallic content of PM₁₀ because metals and metal compounds may cause cardiovascular, respiratory and neurodevelopmental damage. The main sources of these metals in PM₁₀ in outdoor air are anthropogenic and include industrial activities, non-exhaust vehicle emissions, fuel burning and agricultural activity. Within the air monitoring community, these analytes of interest are often referred to as ‘heavy metals’ (heavy implying toxicity) although this term has no agreed technical meaning.¹ Nonetheless, the most important metals are those whose concentrations have limits or maximum target values in current legislation: Ni, As, Cd and Pb. Other transition metals, for example, Fe, Cu, V, Zn, Co and Cr, are also of interest because they are thought to also cause disease through inhalation by increasing oxidative stress in the body. The monitoring and measurement of metals in PM₁₀ are driven mostly by the requirements of national and regional legislation, which will usually specify reference methods that must be used to make the measurements. Around the world, the process is similar and operates using three stages: sampling, sample processing and sample analysis.

Theory of measurement

Particulate matter is collected using size-selective samplers that draw air at a calibrated flow rate through a series of inlets and nozzles that are carefully designed to sample as closely as possible the size fraction of PM required. Samplers for PM₁₀ have a standardised design that, when used at a specified flow rate, ensures that the collection efficiency is 50% at an aerodynamic diameter of 10 µm. Differences in the sharpness of the sampling cutoff at 10 µm (which can vary slightly between samplers) is not especially important for the measurement of metals since most of the metal content in PM is found in smaller size fractions, such as PM_2.5 (particles with an aerodynamic diameter of 2.5 µm) and below. Once past the size-selective inlet, the air is drawn through filters (often made from a cellulose ester material) where the PM₁₀ is collected. Many samplers draw air at a flow rate of 1 m⁻³ h⁻¹, although this can vary as the filter size and composition change. The actual parameter of importance is the volume of air sampled per area of filter per unit time, often called the ‘filter face velocity’ since the units m³ m⁻² s⁻¹ simplify to m s⁻¹. Sampling periods must be long enough to collect sufficient particulate so that the metal of interest will be quantifiable within the target uncertainty, but not so long that the filter becomes clogged with sample, affecting the flow rate and ability to collect PM appropriately. One week is a common sampling time to balance these competing requirements. Because regulations are based on measured annual average limit values, there is generally no requirement for higher-resolution data. Sampling on regulatory air quality monitoring networks usually occurs at monitoring stations with fixed locations. These locations are designed to meet requirements to properly assess population exposure (in urban areas), the effect of industrial emissions (around industrial locations), and background concentrations away from significant emissions sources (in rural or remote areas; see Fig. 1). The site type where the sampling occurs will naturally affect the quantity of PM collected – less in rural areas and more in urban and industrial areas (see Fig. 2). More information on monitoring locations for metals in PM₁₀ in the UK is available elsewhere.^2,3

Fig. 1 Two PM sampling instruments at Auchencorth Moss (Scotland, UK) sampling PM₁₀ for subsequent analysis of metal content (right), and PM_2.5 for subsequent analysis of organic and elemental carbon (left).

Fig. 2 From left to right (top row then bottom row), filters showing increasing mass loading of PM₁₀: a blank unsampled filter, then filters having sampled approximately 168 m³ of outdoor air over one week at the following sites (with site types in brackets): Cwmystwyth and Eskdalemuir (rural background), Swansea (urban background), Sheffield (urban industrial) and London Marylebone Road (urban traffic).

Sampled filters are returned to a central laboratory facility where they are prepared for analysis, usually according to validated documentary standard methods, and measurements taken in support of air quality regulation usually demand this (in Europe, EN 14902 is specified). The sampled filters are not weighed since the measurand of interest is the mass of metal per volume of sampled air, and not the mass of metal per mass of PM₁₀ collected (although the mass of all metals constitutes up to 1% of the total mass of PM₁₀ collected). The filter is then broken down and dissolved in acid in a sealed vessel within a microwave digester under elevated temperatures (usually around 200 °C for about 30 min). An 80 : 20 mixture by volume of nitric acid and hydrogen peroxide is typically sufficient for digestion of particulate matter, although harsher media such as nitric acid with either hydrochloric acid or hydrofluoric acid are sometimes used. Filters may be batched and digested together for expediency if it is acceptable to provide lower-time-resolution data. Equally, it may be necessary to sub-sample from a filter prior to digestion if there is a limit to the quantity of filter that may be digested in the sealed vessels. Portions of relevant certified reference material – comprising well-characterised atmospheric particulate of similar composition to the samples – are digested at the same time as the samples (but in different vessels) to ensure that satisfactory recoveries are achieved. Unsampled filters are also digested to provide a blank concentration to be subtracted from the final analytical results. If the filter media is not fully digested in the microwave, then an additional filtration step is required before analysis.

The digested samples are diluted with deionised water as appropriate to provide a suitable matrix for the instruments used for the analysis. This also brings the mass fractions within an appropriate calibration range. The elemental content of the digested sample solution is then analysed, normally using the well-established techniques of inductively coupled plasma mass spectrometry (ICP-MS) or graphite furnace atomic absorption spectrometry (GF-AAS), and quantified by comparison against a matrix-matched series of calibration standards of varying mass fractions containing the metals of interest. Once a blank subtraction is performed to account for any metal content in the filter media, this yields the total elemental mass fraction of each metal in each sample. Analysis performed in support of air quality regulation is usually required to be accredited to ISO 17025. The mass concentration of metal i in outdoor air, γ_i (in ng m⁻³), is then given by the following equation:

where w_i is the mass fraction of metal i in the analysed sample solution (in ng g⁻¹), m is the mass of the analysed sample solution (in g), and V is the volume of sampled air (in m³). The relative expanded uncertainty of the overall method for average mass concentrations is usually between 10 and 20%, well below the maximum specified in UK and European legislation of 40% (or 25% for Pb), with method detection limits of less than 1% of the average mass concentrations.

Measured mass concentrations depend on the metal of interest and monitoring location. Averaged across the UK, the most abundant metal in PM₁₀ measured by the UK air quality network is Fe at around 500 ng m⁻³, and Cd is the scarcest at approximately 0.1 ng m⁻³; see ref. 4 for more information. The value of long-term measurement of these pollutants in air is to demonstrate trends in concentration, estimate pollution exposure (for use in health studies), and to inform the development of regulation and subsequently evaluate its effectiveness. This is demonstrated clearly in Fig. 3, which shows how the average mass concentration of lead in PM₁₀, averaged across all monitoring sites in the UK, has decreased over the last 45 years to less than 1% of its maximum value in the 1980s – note that a logarithmic scale has been used on the y-axis. This demonstrates the remarkable effectiveness of a variety of policy interventions to limit the Pb content of petrol and then finally ban leaded petrol altogether on 1 January 2000. There are similar cases demonstrating large decreases in metal concentrations in outdoor air over many years, close to industrial facilities that have implemented progressively more stringent abatement measurements, for example, Ni in South Wales.⁵

Fig. 3 The Pb mass concentration in PM₁₀ averaged across all monitoring sites in the UK Heavy Metals Air Quality Monitoring Network between 1980 and 2024.

The ability of air quality monitoring networks to provide useful trend information depends on the locations of monitoring stations remaining the same over time. Whilst this is not always possible – because sites are redeveloped, rental permissions revoked, funding changed or assessment of new emission sources required – these changes are gradual so rarely adversely affect UK-wide trend information. Changes in individual monitoring locations clearly have a more direct effect on locally measured trends.

The future

Recent technological advancements have enabled the use of high time-resolution X-ray fluorescence (XRF) for monitoring the metal composition of PM, generally for research-based studies aimed at source apportionment.⁶ These transportable instruments can be sited at locations of interest and detect a wide range of metals at hourly or even sub-hourly intervals, offering valuable insights into transient pollution events, emission patterns and atmospheric transformation processes. The instrument works by collecting particulate matter on a continuous reel-to-reel filter tape, which is then analysed using XRF. This integrated sampling and analysis system removes the need for separate sampling and analysis facilities and is transportable, so can be moved to better target areas of high pollution or for campaign-based measurements. XRF is particularly effective for measuring major elements in PM, such as Cl, Ca and S, which are often difficult to quantify using ICP-MS due to interferences. However, XRF has much higher detection limits than ICP-MS, which, when coupled with the typically shorter sampling periods, makes it less suited for detecting metals at very low concentrations. The calibration of XRF with matrix-appropriate calibrants is also challenging, and further development is needed before this technique would be able to replace current reference methods for regulatory measurements.

The analytical techniques discussed so far provide only the total elemental content of the PM. Whilst this is useful information, toxicity to humans and the environment from these metals depends on their chemical form, oxidation state and bioavailability. Therefore, it would be desirable to perform full speciation analysis on the metals in PM to better correlate particulate metal content with toxicological endpoints. Some simple techniques, such as determining the proportion of metals extractable in different solvents, provide an assessment of the different chemical fractions present and can be implemented with existing PM samples; however, these only provide approximate information for understanding toxic effects. More detailed speciation analyses – for instance, using hyphenated techniques such as high-performance liquid chromatography–ICP-MS to enable discrimination between redox-active species (e.g., Fe(II)/Fe(III), Cu(I)/Cu(II)) and inert forms – are already quite well known, although they are not currently robust enough to use as part of the routine operation of air quality monitoring networks. One major consideration is that current sampling techniques do not effectively preserve the chemical state of the metals. This is because of possible reactions with both the filter materials and the oxidant-rich air drawn through the sample. To preserve the samples in their original chemical states, it is necessary to implement alternative strategies, such as sampling directly into liquid impingers. Currently, these methods are costly and complicated to use, and not suitable for sampling over extended periods.

This Technical Brief was prepared by Richard J. C. Brown, Emma C. Braysher, David M. Butterfield and Andrew S. Brown (National Physical Laboratory, UK) on behalf of the Analytical Methods Committee and was approved by the AMC on 20 November 2025.

References

1. J. H. Duffus, Pure Appl. Chem., 2002, 74, 793–807.
2. The UK Heavy Metals Monitoring Network: Annual Report 2024, NPL Report ENV 60, NPL, Teddington, 2025,  DOI:10.47120/npl.ENV60.
3. UK AIR, Interactive monitoring networks map, https://uk-air.defra.gov.uk/interactive-map?network=metals, accessed 20 November 2025.
4. S. L. Goddard, K. R. Williams, C. Robins, D. M. Butterfield and R. J. C. Brown, Environ. Monit. Assess., 2019, 191, 683.
5. R. J. C. Brown, S. L. Goddard, K. R. Williams, C. Robins, D. M. Butterfield and A. S. Brown, Environ. Sci.: Processes Impacts, 2022, 24, 1821–1829.
6. A. H. Tremper, A. Font and M. Priestman, et al., Atmos. Meas. Tech., 2018, 11, 3541–3557.

---