Peter J. Jenks
the Jenks Partnership, Newhaven House, Junction Road, Alderbury, Salisbury, Wiltshire SP5 3AZ, UK
Much of my concern comes from the use of ISO 17025 Accreditation by Government agencies to effectively privatise environmental testing work previously carried out by government laboratories. On the face of it this trend is good: accredit efficient commercial labs to carry out the testing: their accreditation body will ensure that, by virtue of properly validated methods regular within and between batch QA and participation in external PT, the lab continues to produce data that is reliable and fit for purpose.
So what is the problem? It is the demands that the accreditation process makes on CRMs and PT providers. Good practice suggests that method validation should be done using CRMs that are matrix matched to the sort of sample the lab will receive for analysis. Further, the use of reliable CRMs is accepted as the best way to validate a procedure. But real world matrix CRMs contain naturally accrued analytes, not including all the analytes the lab wants to analyse.
Consider the UK MCERTS Standard for Soil Analysis.1 When accrediting a new procedure Para 188.8.131.52 says:
“For the method, parameter and matrix, the performance characteristics shall be determined with a minimum of ten degrees of freedom. This shall be carried out by analysing each certified reference material or spiked samples in duplicate in different analytical batches. 11 batches will guarantee a minimum of ten degrees of freedom.”
It goes on to say that: “Recovery estimates shall be obtained using two significantly different but appropriate concentration levels, for example, at 20% and 80% of the expected range.”
So the lab must find CRMs that contain all the analytes of interest, e.g. the EPA 16 PAHs, one at a high level and one at a low level. So labs need to find, per matrix two CRMs, one at high level and one at low level in an amount to perform a minimum of 22 determinations.
But a real world matrix CRM does not contain 16 PAH at a uniform level. So what does the lab do and what can the producer do to meet the labs needs? I recently reviewed a catalogue offering of PAH in soil CRMs from a CRM supplier.2
Their catalogue lists 24 soil CRMs certified for “BNAs”, a group of more than 30 analytes including PAHs. Their CRMs cover a range of soil types, from sediment, through clay and loam to sand. Their CRMs contain from one to 15 of 16 PAH, at levels ranging from 169 mg kg–1 to 0.03 mg kg–1. They also list four PAH in soil CRMs that contain the full suite of PAH at typical levels.
But even with this large range of CRMs it isn’t possible to select three soil types and then find all 16 PAH at two levels. To do so a lab might have to use more than 12 actual CRMs and still do their own spiking of a clean soil to fill in the gaps. With matrix CRMs at more than £130 per 50–100 g this is expensive.
Looking at the BNA in soil CRMs offered by this supplier it was clear that they evolved from totally naturally sourced matrices to semi-synthetic materials. Why? Because I suspect that only by starting with clean soils of defined type can the producer be reasonably certain of ending up with a soil containing the range of analytes at the levels the customers demand.
Over the last six months I’ve been talking to scientists from UK labs working towards MCERTs or ISO 17025 accreditation for organic analytes in soil. It is clear that the labs would really like a small range of CRMs containing all their accredited determinands in three matrices and at two levels, so they would all need just six CRMs. The CRMs would contain enough material to carry out 66 determinations. The argument was that if everyone used the same CRMs there would be even better comparability of data between labs, and (perhaps more to the point) the process of accreditation would be cheaper.
Whilst it may be technically possible to produce such CRMs they would be far removed from CRMs produced by national metrology institutes (NMIs) 10 or 15 years ago. The original idea of a NMI CRM was material that could show that a determination could analysed and allow the proper demonstration of traceability to the mole with a clear and credible statement of uncertainty. In a past RM column3 I pointed out that the relationship between an analyte and a matrix was complex and that efforts to improve “production efficiency” in commercial environmental laboratories by the use of accelerated extraction systems had made achieving accreditation a difficult challenge. Some labs felt that because a natural matrix, but fortified CRM, would not allow them to achieve the required level of precision and bias there was something “wrong” with the CRM.
All of this is slowly leading me to have a real concern that accreditation of commercial service laboratories to ISO 17025 is actually compromising chemical metrology. This is because as labs struggle to “jump through the hoops” that are essential for their continued commercial viability they are losing sight of what good chemical analysis is supposed to be about. I’d be very interested to know what you think about this issue.
- RT Corporation, www.rt-corp.com
- P.J. Jenks, Spectrosc. Europe 19(1), 30 (2007).