Technical Report No. 3
Environment Australia, May 2002
ISBN 0 6425 4781 5
5. Discussion of methods used in Australian studies
Determination of lead in ambient air in Australia is normally based on the Australian Standard method AS2800 (1985) which requires the collection of TSP or PM10 onto filter paper, gravimetric determination of particle loading, dissolution of the lead and analysis by Atomic Absorption Spectroscopy for lead concentration. Australian studies involving the determination of several metal elements were based on the standard method for determining lead. However, there were several differences in the procedures applied and instruments chosen for sample collection and analyses, which will be discussed in this section.
The sampling techniques applied in each study depended upon the choice of location for monitoring, the particle sizes excluded, the types of metals/elements targeted, the number of samples collected and the frequency of sampling.
The siting of the samplers used for lead determination normally conforms to Australian Standard AS2922 (1987), a guide for siting of sampling units. Further, the majority of the metal studies were conducted at sites that formed part of the State or Territory network of air quality monitoring sites. Since all of the data reported here were acquired after AS2922 was adopted, in 1987, it would be expected that a majority of samples were collected in accordance with the Australian Standard for siting samplers.
There is no clear pattern on how the number of sites was chosen for each study, and it is possible the number of sites was determined by consideration for costs and not based on the population size. For example, the Perth Haze Study collected samples from 3 sites in Perth, which has population of 1.3 million, but the Melbourne Aerosol Study collected data from only two sites (Alphington and Footscray) in Melbourne, which has a population of 3 million.
The high volume sampler has been used consistently for lead monitoring. Some jurisdictions monitor lead in PM10 and others monitor lead in TSP or in both types of particulate matter. In the case of the elemental studies, the table in Appendix G shows that a wide range of samplers have been used for sample collection. These samplers have varying sample collection efficiencies for each metal type, and different filters were used in the samplers. The table also reveals that in most of the studies, fine particulate matter, PM2.5, was collected; in other studies PM10 was collected. To facilitate comparison with data from other studies, some studies collected three size fractions; PM10, PM2.5 and the coarse fractions (2.5µm<PM<10µm).
A requirement of the NEPM for Ambient Air Quality is for the operators of performance monitoring stations to be accredited by the National Association of Testing Authorities (NATA) or have in place an equivalent system of quality assurance and data validation procedures (NEPC, 1998). The high volume samplers used to collect particulate matter for determination of lead are expected to be operated according to the procedures set down in the Australian Standard method AS2800 (1985), which has quality assurance procedures for calibration, monitoring and reporting. However, a review of PM10 sampling methods in Australia by Richardson and Crerar (1999) listed some of the potential sources of errors that could occur when operating these samplers. The errors identified as being relevant to the determination of heavy metals were:
- Loss of volatile particles collected on the filters during sampling and during pre-weighing conditioning of the filters.
- Improper filter handling, which may lead to the loss of particles or contamination of the sample.
- Variations in the ambient temperature and humidity could vary the flow rates hence affect the efficiency of the size selective inlet (SSI).
- If the flow rate is not determined accurately, then any variations in the airflow would introduce errors in the calculated mass concentration.
- Since sample collection and weighing are labour-intensive, there is significant potential for operator error to occur.
The errors arising from loss of volatile particles and the effect of ambient parameters on the efficiency of the size selective inlet cannot be controlled and are limitations of the technique used for sampling. However, these errors can be quantified if the effects of the extreme variations in ambient parameters were determined during the sampling period and reported with the results.
Accurate flow rate measurement requires continuous flow measurement during the full period of sampling. Some studies measured flow rates only at the beginning and at the end of the sampling period to obtain an average value. The sampling volume determined for this type of sampling could be in error if there were large variations in the prevailing ambient parameters during the sampling period.
Since the method is labour-intensive, human errors could be minimised by using highly trained personnel for collecting samples.
The recommended filter paper used in high volume samplers for compliance monitoring of particle loading is not appropriate for chemical analysis. This is because the filter normally has very high background levels of the targeted elements, and may interact with the chemical elements in the particles. The other samplers such as the dichotomous or sequential samplers use Teflon-membrane filters, which provide better samples for chemical analysis.
There are very limited data available on cadmium and mercury, which are very toxic pollutants, and on other heavy metals such as arsenic, selenium and gallium, which form volatile compounds in air. Some of these metals and their compounds have appreciable vapour pressures at ambient conditions (Fishbein, 1991). The very low ambient concentrations, or the absence of data, for some of these metals could be attributed to the fact that they may not have been effectively retained on the filter papers, when hivols or even low volume samplers were used for sampling. Cold Vapour Atomic Fluorescence Spectrometry (CVAFS) is the recommended method for the determination of mercury in ambient air. However, none of the Australian studies reported using this method. Lead also forms volatile compounds, and there was a potential for losses to arise when lead compounds were collected by high volume samplers. There were no reports of the efficiency of lead collection associated with reported ambient concentration.
The Australian Standard method for determining lead in particulate matter requires a 24-hour sampling duration on a 6-day cycle, with sampling starting at midnight (AS2800, 1985). The 24-hour sampling duration accounts for any diurnal variation in ambient particulate matter concentration. The 6-day cycle is a compromise, since it would be impractical and costly to collect samples on a daily basis using hivols. However, there is a probability that 5 out of 6 pollution events may not be captured by the sample collection system if this sampling frequency is applied (Richardson and Crerar, 1999). Further, since pollution events tend to peak around midnight during winter, the start of sample collection at midnight would result in the collection of only a fraction of the particles responsible for the pollution event.
Although most of the multiple metals studies collected samples for analysis on the six-day cycle, a few had very different sampling schedules. The ASP and Queensland study collected samples on two fixed days in a week-Wednesdays and Sundays. During the Perth Haze Study, sample collection was synchronised with the high volume samplers, which were collecting samples for gravimetric analysis of particulate matter and/or determination of lead on a six-day cycle. However, other samples were collected during days of reduced visibility for a 12-hour duration. The Melbourne Aerosol Study collected samples for a duration of 8 hours, during periods of reduced visibility. The Australian Fine Particle Study collected 24-hour samples on a six-day cycle, but samples were collected for only one to two months, consecutively, from the six sites. The collection of samples using different sampling frequencies and duration adversely effects the reconciliation of the data for comparability, and is therefore difficult to consolidate the data into consistent formats. Further, none of the multiple metal studies spanned four full calendar years, making statistical analysis for the long-term trends in concentrations unfeasible.
The dissolution of metals is a wet chemistry method; aside from being labour-intensive there is the potential for the sample to be contaminated by the chemical reagents used for analysis. The analytical techniques that require sample dissolution are FAAS, GFAAS, ICP-MS and ICP-AES. Most State or Territory lead measurements involved FAAS/GFAAS, with only one jurisdiction reporting use of XRF, which does not require dissolution of lead during chemical analysis.
The dissolution of lead for chemical analysis is only applicable to samples collected in urban areas where motor vehicle emissions dominate the emitted pollutants. It is not applicable to samples collected in the vicinity of specific sources, which may give rise to high levels of insoluble particulate lead sulphate or lead sulphide (AS2800, 1985). This would also apply to other insoluble metal compounds collected on the filters. However, it appears from the reported studies that the dissolution method has been used on filter samples collected from all types of air environments.
Most of the multiple metals studies used PIXE for analysis and did not require metal dissolution. For such techniques, sometimes only a fraction of the filter sample is analysed for its metal composition. Thus it is important that there is a uniform deposition of the particles on the filter paper during sampling. However, none of the studies using PIXE analysis reported cutting off fractions of the filters for analysis.
There were several techniques available for determining the elemental composition of suspended particulate matter for their heavy metals content, and '… no one analytical method can address all data quality objectives for a particular ambient air monitoring program. Each method has its own attributes, specificities, advantages and disadvantages, …' (US EPA, 1999b). Sensitive methods that can measure the different isotopes of the metals, and hence provide the isotopic ratios, should be considered superior since they would provide more information on the sources of the metal compound. Neutron Activation Analysis (NAA) appears to be the most promising technique, but it cannot determine nickel, cobalt or lead, and requires a nuclear reactor. Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) would be an excellent choice, but it is expensive. A summary of the merits of the instrumental methods used routinely for determining heavy metals composition in particulate matter is provided in Table 11 (adapted from US EPA, 1999b).
|Minimal sample preparation||x||x||x|
|Large linear working range||x|
|High sample throughput||x||x|
|Single element analysis||x||x||x|
|Matrix mismatch problems||x||x||x|
|Require nuclear reactor||x|
|Limited working range||x|
|Low sample throughput||x|
|Limited number of metals||x||x|
Note: * Abbreviations are explained in the Glossary of terms
Appendix G shows that most of the elemental studies in Australia involved analysis by the PIXE technique at ANSTO. Several thousand samples have been analysed at the ANSTO laboratory using Ion Beam Analysis methods since 1992. The PIXE technique was used to determine the presence and concentration of up to fifteen metals/metalloids in Australian samples. These data sets unfortunately have little or no data on magnesium, scandium, arsenic, selenium, rubidium, strontium and zirconium, which have been reported in similar samples from the Charles Point Study, and analysed by PIXE in Brazil (Ayers et al., 2000).
There has been very limited application of ICP-MS and ICP-AES, and none using NAA techniques in the Australian studies, although these techniques are more sensitive for measuring metals in particulate matter. Studies involving lead-only analysis employed AAS or XRF techniques, since these were not expensive (though less sensitive) when only one metal type was to be analysed.
There are no Australian standard methods for determining the concentrations of several metals simultaneously by these techniques. The US Standard Methods (US EPA, 1999b) were compiled after the Australian studies commenced. The laboratories that utilised the PIXE, ICP-MS, ICP-AES and XRF techniques would have followed the different procedures recommended by the instrument manufacturers or other laboratories.
Fortunately, the PIXE technique, which was used for most of the metals analysis in the Australian studies, is an absolute analysis method. The efficiency curve that establishes the conversion of X-rays counted to some more meaningful number such as micrograms/cm2 of element present on the filter paper is due to the physics of the atom (US EPA, 1999b). This curve is established by running standards of known concentration for a given amount of time and measuring both the X-rays generated and protons used for generation. At ANSTO, computational spectrum fitting codes were used to obtain quantitatively accurate and precise element concentration data from PIXE spectra (Cohen, 1992).
The ANSTO laboratory has reported good precision and accuracy in PIXE analysis data for six elements: aluminium, silicon, chlorine (non-metal), calcium, iron and strontium over a six-year period (1991-1996). A value of 1.00 + 0.03 (where the error was one standard deviation) was obtained from averaging 60 measurements of the ratio of measured concentration to nominal concentrations of MicroMatter thin foil standards (Cohen, 1998).