In January 2003, issue number 45 of the AGS Newsletter contained an article by Bruno Guillaume, of Arup Geotechnics, who outlined the MCERTS performance standard for the chemical testing of soils. The following is an update, and a view from an analytical chemist`s perspective.
On the Environment Agency website reference has been made to the fact that the Agency is aware that it will take time for laboratories to gain approval through the appropriate accreditation process. An eighteen month period, starting from March 2003, has been given for laboratories to bring their soil testing methods up to the MCERTS standard.
During this period laboratories reporting data to the Agency have as a minimum to be accredited to the ISO 17025 standard for the soil test methods. It is also recommended that tests should have a brief method description together with estimates of bias and precision. From September 2004 only data from laboratories that have been accredited to ISO 17025 for MCERTS will be accepted.
Since the last article in the newsletter, Version 2 of the MCERTS standard has been published, and this was available from February 2003. The standard highlights particularly important areas, namely contract review, bias and precision targets, quality control( both internal and external), method validation, and uncertainty of measurement. Important differences from the first version are the exclusion of expected limits of detection for methods, and the inclusion of an improved protocol for validation.
The issues can be confusing but the standard simply aims to establish a level playing field in a competitive market, based on the Agency`s requirements, and to set a minimum acceptable performance. In short the data received by the laboratory`s customers must be accurate, reliable and comparable.
The analysis of soil is complex in terms of the chemistry involved. It aims to determine both macro and trace components in a matrix that is, quite often, dirty in both a physical and chemical context. There is a need to analyse for trace organic and metallic contaminants in soils that contain large quantities of other industrial materials, such as oil or tar, in a background that also contains high concentrations of naturally occurring, or artificially polluted, inorganic compounds.
We all use “parts per million” as routine terminology, but the significance is commonly ignored. 1 part per million is more easily visualised as 1 grain of salt in a swimming pool. When we talk of the concentrations of polynuclear aromatic hydrocarbons (PAH), an important environmental parameter, we often refer to micrograms per kilogram, which is three orders of magnitude lower.
The contaminated land testing industry has grown very quickly, and methodologies have been borrowed from other more well established areas of analytical chemistry, such as food or potable water. The only industry standard for analysis of soils in the UK are the robust and technically sound ” British Gas Methods “, but even these were not designed to tackle the lower end of detection, and do not take advantage of some of the more modern developments in analytical chemistry.
MCERTS effectively defines a standard for the performance of analytical methods, and includes the requirements of ISO 17025 in terms of certification of instrument performance, approved competency of personnel and the accreditation of laboratory procedures and organisation. It means that it is no longer sufficient that the laboratories follow a rigorous UKAS quality system in line with the international standard, but that the methodologies must also be demonstrated as fit for purpose.
The Environment Agency has not, in its standard, adopted the principle of prescriptive methods, as has been the example in the USA, through the so-called EPA procedures. This approach can commit the industry to inappropriate analytical techniques, a long time in their reform once committed to paper, and takes away the flexibility of developing new improvements for the industry as a whole.
It cannot be relied upon that environmental specialists, requiring the services of an analytical laboratory, will have the depth of technical to knowledge to understand the concepts of analytical bias or precision. MCERTS is designed to take away the need for such expertise.
Another variable that stops a customer from being able to compare “apples with apples” is the limit of detection (LOD) quoted. This can vary widely depending on how the laboratory defines it. A sound statistical principle is to use three times the standard deviation associated with a blank, or a sample with a very low concentration of the determinand of interest. This is all carried out interspersed with other standards and samples over eleven separate days. Other lesser definitions than this one seem to describe a “better” LOD, but mislead the customer into thinking they are getting an improved service, and can give false positive concentrations on soils where none of the contaminant actually exists.
All of these concerns are addressed by the MCERTS standard. Precision and bias must be of an acceptable standard, as must LOD. “Recoveries”, namely what happens when a soil is spiked with known amounts of the material of interest and is reanalysed, are examined in the standard to ensure acceptable performance. The validation must be carried out on three completely different soil types with two spiking levels, and include the use of certified reference materials wherever possible. Detailed methodologies, together with a prescribed uncertainty of measurement must also be given.
The Contract Review is the point at which the client`s needs must be understood, and the point at which the laboratory must document them. What does ” Total PAH” mean, or “Total TPH”, and what does the client consider to be the critical level of interest? This is an area quite often poorly addressed, and to which the standard lends some priority.
The laboratory`s quality control also comes under close scrutiny. At least 5% of the resources allocated to a test must be used to ensure validation. In addition the laboratory must participate in as many of the acknowledged external proficiency tests as is appropriate, such as Contest, Aquacheck, and the SPH test scheme. The results of these must be readily available for inspection by the client.
It is generally recognised amongst the community of analytical laboratories that there is a real challenge in order to be able to comply with the new version. The standards relating to bias and precision and, in particular, the guideline that “the limit of detection usually regarded as being fit for purpose is 10% of the concentration regarded as the critical level of interest” are extremely demanding. There are some method improvements required within the industry before these levels of performance can be achieved. Most laboratories, however, will feel a relief that any ambiguity is now removed, so that everyone can compete to provide a well defined product, and be able to market its expertise without confusion.
Whilst addressing the vagaries of analytical results the Environment Agency has also acknowledged the uncertainty associated with other areas, and is considering certification schemes to address field aspects, including sampling. Other subjects, for example the suitability of leachability tests, toxicity assessments and the bioavailability of metals need to be topics for guidance by the regulator.