How standardised is MCerts analysis in stony samples? 

Despite articles such as that by Mark Perrin (Ground Engineering, April 2007), many engineers, consultants and regulators still appear to be largely ignorant of the differences in soil sample preparation methods between laboratories and the effect this can have on the results produced.  Many assume that the MCerts accreditation commonly required by Local Authorities and the Environment Agency has ‘standardised’ laboratory analysis such that the reported contaminant concentration will be that of the soil tested without any significant variance.  Unfortunately, whilst the MCerts scheme focuses on quality procedures to ensure repeatability within an individual laboratory, it does not standardise the test method itself, and thus variations between laboratories remain.

The MCerts scheme requires comparison trials between labs to be carried out on a standard homogeneous matrix, which generally produces a high degree of comparability between laboratories; unfortunately most soils that are sent for analysis tend not to be like this homogeneous matrix.  Typically most analysis is carried out on samples of near surface made ground which is by its very nature largely heterogeneous, and usually with a high stone content.  As an extract can not be produced from the stones, sample preparation including either grinding of the stones to a fine powder or excluding them such that the sub-sampling uncertainty can be reduced or eliminated.  However, the methods adopted to achieve this are by no means standardised between laboratories.

How do laboratory methods vary?
All commercial laboratories with MCerts accreditation for PAHs and toxic metals were asked about their sample preparation methods. Responses were received from 18 of these laboratories with 17 being accredited for metals and 17 for PAH.  A surprising number of laboratory managers were unsure of the sample preparation method and had to check their procedures before being able to reply; however the responses received are summarised below;

Metals	No of labs using this method	Percent of labs that responded
Dry and crush to less than 0.5mm	8	47%
Remove stones greater than 2mm diameter	2	12%
Remove Stones Greater than 10mm diameter	6	35%
Remove all ‘inert’ stones	1	6%

 

For metals the main difference appears to be whether the stones are removed or are crushed and when removed what size stones are removed. However, with PAH a greater range of sample preparation methods were apparent;

 

PAH	No of labs using this method	Percent of labs that responded
Test as received sample but avoiding stones	8	47%
Dry and crush to less than 0.5mm	4	24%
Remove Stones Greater than 2mm diameter	2	12%
Remove Stones Greater than 4.75mm diameter	1	6%
Remove Stones Greater than 10mm diameter	2	12%

 

Of those labs that dried the samples, two did so at 28 ^oC, five at 30 ^oC, one at 35 ^oC and one at 37 ^oC.  Variation in the extraction solvent was also apparent as set out below.

 

Extraction solvent for PAH analysis	No of labs using this method	Percent of labs that responded
Dichloromethane (DCM)	13	76%
DCM & Hexane	1	6%
Hexane:Acetone	2	12%
Pentane	1	6%

 

This variation has been further complicated recently, as during the recent recession, the partial shut down of the car industry reduced the demand for foam rubber, a bi-product of the production of which is DCM.  Therefore the price of DCM rose significantly and several laboratories switched from using pure DCM to a mix of DCM and hexane or acetone or even hexane:acetone:triethylamine.  Once the car industry recovered and the price of DCM fell these laboratories reverted back to using pure DCM.

What are the effects of variation in preparation methods?

In a soil where the contaminant concentration is distributed evenly between the matrix and the stone content, removing or crushing the stones would clearly have no effect on analysis results.  However, where the metallic contaminants are expected to be concentrated in the coarser particles, as in a slag or clinker, crushing of the stones will produce a higher contaminant concentration than would be produced in a laboratory that removes the stones.  Similarly, as the bulk of the made ground which is present in garden areas that have been used for a considerable period will have been subject to bonfires and the active digging in of ash, part burnt fragments of coal or timber are common.  Such relatively coarse fragments could be expected to be a source of PAH (including benzo(a)pyrene) and therefore a sample which was crushed prior to analysis could be expected to yield a higher concentration than one from which the larger particles had been removed. Conversely, where the burning of painted wood has resulted in elevated lead concentrations in a fine ash, or where soot has been dug into the soil, the presence of natural stones in the sample will act to ‘dilute’ the measured concentration if they are crushed during the analysis rather than removed..  However where the weight of the stones removed is back-calculated into the reported result this effect should be eliminated.

In addition to the effect of stone content, where a sample is dried the variation in drying time and temperature will surely have an effect upon the more volatile compounds such as naphthalene which could be lost to some extent, especially where the sample is dried overnight at 37oC.  The reported naphthalene concentration of a dried and crushed sample would thus be expected to be lower than that of a sample tested in the as received condition.

The variation in extraction efficiency of the different solvents and solvent mixes would also be expected to induce variability.

So which is the right method?
Unfortunately, there is no right or wrong method of analysis, as the applicability of each method is dependent on the use to which the results are to be put, which is beyond the control of the laboratory.  The onus therefore ,has to be placed upon the consultant scheduling the testing and interpreting the results to use an appropriate method.

For example; waste classification is based upon a hazard assessment and the analysis is required to be representative of the whole load being disposed of, and thus a dried and crushed approach may be applicable.  However, for human health a risk based assessment is adopted and thus consideration needs to be given to the likely exposure pathways.

For Benzo(a)Pyrene some 56% of the total exposure for the residential land use arises from the ingestion of soil and indoor dust.  As soil Pica has not been included as an exposure route in the CLEA model, the bulk of the ingestion of soil will be from accidental ingestion from hand to mouth contact and as such larger particles are unlikely to be involved.  Similarly, dermal contact, which amounts to 36% of the total exposure for the residential land use is unlikely to be significantly affected by the larger particles which will have a far lower surface area to volume ratio.  Therefore, for a Benzo(a)Pyrene risk assessment, an analysis of the fine particles following sieving and the exclusion of the stone content from the calculated results would appear to be far more appropriate.  This would also be the case with the majority of the PAHs and toxic metals.

Conclusions
Engineers and consultants scheduling laboratory analyses need to be aware of the sample preparation method that is to be adopted by the laboratory and the effect that this will have on the results for the specific soil which is being tested.  It would also be helpful if laboratories published a basic summary of the sample preparation method along with the results to aid those interpreting the results and to assist where comparisons between different sets of data on the same site are being carried out, be it by different laboratories or by the same laboratory over a prolonged period.  Furthermore, when the chemical analyses indicate that the soil in a garden contains contaminant concentrations close to or a little in excess of the adopted threshold values, it would be worth considering the effect of the sample preparation method and whether retesting a sieved sample may produce a more appropriate concentration to compare with the adopted threshold, considering the assumed critical exposure pathways.

Mike Plimmer
GEA Associates