Detective work and contamination

Successfully tracking down the source of contamination depends on applying the right analysis tools for the job. Scientist Stephen Mudge explains

As a forensic environmental practitioner, the questions I am commonly asked relate to the state of the environment and typically revolve around: “Who contaminated this soil/sediment/water?” And the next question is often: “How long ago did this occur?”

It is fortunate that there is a wealth of knowledge on how chemicals behave in the environment and their different degradation or transformation rates. Analytical equipment is also available that can quantify whole suites of compounds and elements rapidly, repeatedly and with low-detection levels. Several new instruments push the envelope even further down the concentration range.

The correct analysis

All this is great, but there still needs to be considerable effort invested in performing the correct analysis to answer the question being asked. If an issue regarding the source of a metal such as lead (Pb) comes up, for example, simply measuring its concentration might not be enough to ascertain the source, especially where it has been transported in water and deposited in another location. In some cases, it may be possible to backtrack a concentration gradient to the strongest signal, but this is frequently not possible.

It would be better in such cases to use the stable isotopes of lead to define its geochemical signature. Samples would need to be collected from all potential sources, as well as from the affected area. It may be that the site has received the metal from several sources, some of which may have changed over time. If there is sufficient discrimination between the isotopes, it would be possible to develop a mixing model and quantify the input from each source.

For example, runoff from a former lead mine in North Wales – Parc Mine in Gwydyr Forest – produces considerable quantities of oxidised iron that provides a mechanism by which other metals might be transported further. The question was how much this particular source was contributing to the contamination relative to other known sources of lead in the area. The stable isotopes – 204Pb, 206Pb, 207Pb and 208Pb – clearly showed that this was a minor contributor and another sub-catchment of the river was the major provider. Measuring the lead concentration alone would not have distinguished between these sources.

Oily mess

Similar approaches are necessary when seeking the source of oil hydrocarbons in the environment. From a forensic standpoint, it is fortunate that both crude oil and many products derived from oil have hundreds, if not thousands, of compounds. These compounds have a range of physicochemical properties that can assist in determining the most likely source and the time since deposition.

Short-chain alkanes are readily lost through evaporation and microbial degradation. These processes have comparatively less effect on the long-chain or branched compounds. A simple approach to determining the relative age of the deposition is to look at the ratio of these compounds; a single total petroleum hydrocarbon value cannot be used for this.

Alkanes are common to all these oils and cannot be used on their own to discriminate between sources. In this case, biomarkers, such as the steranes and terpanes that are remnants of the original source organic matter, can often provide a signature and allow source identification. These compounds are stable in the environment for extended periods and have been used to identify the responsible parties when tar balls have washed ashore.

Although these compounds are not difficult to analyse, they are not routinely measured with the other compounds associated with oils. Many commercial laboratories do not normally examine these compounds and may not even have the capability – typically their bread and butter work is monitoring for compliance. Forensic investigations require a much broader range of compounds, as different ones are needed to answer the different questions that may arise (see examples below).

Error prone

Modern commercial laboratories can provide quantitative results for compounds, or elements, with relatively few errors, and today’s analytical instruments can provide very good, repeatable results. However, before these data can be used, consideration should be given to the errors associated with environmental variability; the sampling approach; and the robustness of the proposed post-analytical methods.

Experience has shown that even within a one metre square area of apparently uniform sediment, the concentration of some analytes might vary by 100%. The collection of small sample volumes that do not take into account spatial variability may lead to data that are difficult to interpret and may also mask temporal trends.

Many dioxin, furan, and polychlorinated biphenyl (PCB) congeners may be present in samples, but laboratories and regulators tend to focus only on toxic congeners, of which there are relatively few (seven dioxins, 10 furans, and 12 PCBs). Data for these congeners alone will not necessarily tell you where the compounds may have arisen; only a full suite of congeners can provide such insights.

Separating the anthropogenic input from the natural background might not be an easy task either. Some chemicals, and many elements, are naturally present in soils and sediments and may have been enhanced as a result of ancient activities, such as metal extraction by the Romans.

Dioxins, for example, may be discharged from industrial operations, but they are also formed when burning domestic waste and are naturally present in some clays.

Context and the environmental and industrial history of a site are all important and help to determine the analytical approach to take, as different physical or chemical species may represent each time period.

After analysis, it may be possible to generate precise ratio values or other measures, which can be applied to models. For instance, the ratio between the different BTEX components (benzene, toluene, ethylbenzene and xylenes) has been used to estimate the timing of a petroleum release into the environment. However, the behaviour of the individual chemicals is highly dependent on several environmental factors, such as oxygen content and temperature.

Applying these data to a decay curve without considering the external factors may also lead to erroneous interpretation. What can be said, however, is that samples with more benzene and toluene (the more volatile components) than ethylbenzene and xylenes will have been in the environment for less time than samples from the same area that have proportionately less benzene and toluene.

Robustness

Analytical instrumentation now generates considerable data for each sample, and this has certainly increased the robustness of further interpretation. The use of multivariate statistics or chemometrics enables environment professionals to investigate vast datasets to determine the underlying structure and sources of variation in samples.

This has many benefits, but the adage about “garbage in, garbage out” still applies. There are several risks associated with the way data are treated before they can be submitted to these statistical methods, including collecting signatures not concentrations, normalisation, missing data and the limits of detection. Again, the correct analysis method must be chosen to accurately identify the source of contamination.

Analysing oil and oil products in the environment