The accuracy of any analytical result depends on numerous factors. Quality control procedures are designed to minimise the variability and potential errors caused by these factors.
The QC issues that cause the greatest errors during analysis are: Calibration, Instrument Function or Operator. The greatest cause of error by far is from poor sampling practice; Analysis error magnifies the total error. Correct sampling procedures are described in Guidance on the classification and assessment of waste (1st Edition v1.1.GB ) Technical Guidance WM3;
All analysis methods require a calibration process of some sort to work out the relationship between the raw signal output and concentration of the chemical compound being analysed. Laboratories use certified reference calibrators and assume that if the expiry date of the calibrator has not expired and the storage of the calibrator has met the manufacturers guidance and the label on the calibrator indicates the contents of the calibrator is correct, the calibrator is good to use. No other checks are routinely made. However calibration solutions can easily evaporate, decompose, become contaminated or over-diluted through incorrect storage and handling and will therefore give incorrect results. They should be checked before each use, but this does not happen for the calibrators used for hydrocarbon, heavy metals and anion analysis used by the majority of laboratory analysers, therefore the QC in these cases is not sufficiently robust.
The QED on site hydrocarbon analyser also requires calibration and uses a single calibrator called Scanset. To minimise the potential for calibration errors, the QED uniquely checks the concentration and the composition of the Scanset calibrator used, both before, during and after analysis. If the Scanset fails this QC check, it will be rejected and no analysis can be carried out until an acceptable Scanset is used. The Scanset is used to accurately measure the energy output of the QED analysis system. The QED uses the principles of Quantum Electro Dynamics where molecules that absorb photons of a given energy will re-emit lower energy photons at a constant and repeatable ratio to the input energy. This is called fluorescence. This is measured by using the Scanset.
Different hydrocarbon types create different patterns of energy output and a unique fingerprint can be derived for each hydrocarbon type. When a QED is manufactured, a reference library of hydrocarbon fingerprints is created using certified reference standards. The 20 distinct calibrators contained in the library cover almost all sources of petroleum hydrocarbon, coal tar and bitumen and can factor in degradation of the hydrocarbon. The QED system can unravel the sample raw energy output and determine what proportions are from which reference library hydrocarbon. Up to 4 individual hydrocarbon types can be resolved. The energy output for each hydrocarbon type is then compared to the energy input and a concentration calculated. The combined concentrations give the total hydrocarbon concentration value. This procedure ensures the correct hydrocarbon calibrators are selected for the hydrocarbon in the sample and the most accurate results generated. No other analyser does this. Standard GC methods use 2 calibrator types at most.
There is a caveat to this procedure. The energy calibration only works when the concentration of molecules/atoms in the sample is below a critical concentration. The QED minimises this problem by using another spectroscopy technique to estimate the total concentration of target molecules in the sample. If this concentration is exceeded, the QED software alerts the user and suggests a sample dilution that will bring the QED into the correct operating range. The QED is unique in this ability to identify if a sample is outside the concentration range and to then provide a dilution to get the sample within range. The QED also checks that the hydrocarbons identified in the sample are not above or below the concentration range of the library calibrations. If they are the software suggests an appropriate sample dilution to use to ensure the analysis is within the correct calibration range.
The conventional laboratory method for hydrocarbons uses a Gas Chromatography analyser, and rely on a QC manager to check if the sample concentration is within the calibration range. If the sample is outside the range, the sample is diluted or concentrated and reanalysed. If a sample contains a higher concentration of target compounds than the calibration used, the GC system cannot estimate the sample dilution required. To get around this problem, most labs analyse several dilutions of the same sample to ensure at least one dilution is within range. A typical GC run takes between 5 and 15 minutes. Running 3 dilutions to make sure one is in range can take 45 minutes for just 1 sample. The QED by providing the dilution needed and only taking a few seconds per analysis, is considerably faster than the GC and also much more cost effective because it is not consuming expensive high purity gases for a long time.The use of fluorescence techniques allows detection and measurement of petroleum hydrocarbons, polyaromatic hydrocarbons, coal tars, BTEX, heavy metals (Cd, As, Pb, Cu, Zn, Hg etc) and other elements such as sulphur, chlorine and lighter metals such as potassium and aluminium. Fluorescence analysers when designed to minimise the effects of interference from other elements or too high or low concentrations of the target compounds are very accurate. Fluorescence is also very sensitive. The QED can detect Benzo Pyrene at concentrations below 5 microgram/kg without using any sample concentration. A standard lab GC has detection limits in the 1 – 10 milligram/kg range and requires samples to be concentrated. This means that effectively the lower limit for GC is 200 times higher than that of the QED.
A good QC system will confirm if the analyser is working correctly. In the laboratory a lot of time is spent running reference samples of known concentration and type through the analyser to confirm it is giving the expected results. This is a time consuming and expensive activity. For on site analysis this is not practical, so alternative monitoring procedures are built in to check the analyser operation automatically.
The QED runs multiple self diagnostic checks on initial warmup that confirms the excitation source is working and producing sufficient energy. If deterioration in performance or a failure is detected, a message will alert the user to change the excitation source. A spare is always included with a QED. The QED continuously monitors the excitation source and will alert the user if it has failed. The most common cause of failure is the electrical supply has been interrupted or if the baseline check has not been carried correctly. The system also checks that the sensor is working correctly, the baseline signal is linear and has no bad segments, the wavelengths detected correspond to their expected segments, the condition of the cuvette and cuvette holder is acceptable with no damage or contamination and that the solvent used to run the analysis is contamination free.
One of the key factors in instrument performance is baseline drift. All analysers use some form of sensor to measure the output signal. For the GC, a Flame Ionisation Detector (FID) is a common sensor. The QED uses a charged coupled device (CCD.) It measures baseline drift during each analysis and automatically forces a simple 15 second procedure to reset the baseline if drift is detected. Each baseline reset checks the sensor for linearity and bad segments
The GC analyser uses the time a molecule takes to travel the length of the column to identify molecules. This is called retention time. This can vary as temperature changes, the column ages, high concentration samples go through the column and injection times vary. To counter this reference standards are analysed interspersed within the samples to measure and compensate for this drift.
Analysers such as the QED that measure energy output at specific wavelengths must ensure the sensor is correctly aligned and the CCD segment assigned to photon energy X is only detecting energy X photons. The QED is able to use the excitation energy wavelengths which are of a known energy and frequency to check the sensor is correctly aligned. It does this on initial instrument start up and regularly during analysis. The sensitivity of each segment is also checked.
Contamination of the instrument will create false results. For GC, running blanks between the samples identifies if carry over from the previous sample could be measured by the next sample. With fully automated GC systems, any carry over detected should trigger injector/column/detector cleaning cycles before any more samples are analysed. These cleaning cycles are time and reagent consuming. In severe cases the entire injector/column/detector set must be replaced and a new calibration created. This can add a day or two to the analysis process. On manual systems, the QC manager should be reviewing the data to check for potential carry over and if detected repeats of the analysis should be carried out.
For the QED, running a blank shows the solvent, the analysis cuvette, cuvette holder and detector lens are contamination free. If a highly concentrated sample is put into the QED cuvette an over range warning shows and the QED software forces the user to rinse out the cuvette and run a blank. This is repeated until the blank shows no contamination. A blank run takes less than 15 seconds and is easy to do. During the initial start up the QED also checks the cuvette/cuvette holder/lens/solvent for contamination. At any point if contamination is detected and the cleaning procedure fails to remove the contamination, the QED will not generate any more results. All components can be deep cleaned on site in a few minutes and instrument contamination is rarely a problem. Inadvertent contamination of the solvent used to run the analysis is the most likely cause of failure. Users are advised to keep a good supply of solvent.
For any analyser, having a stable power supply is essential. The QED uses battery power or an AC110/240V wall outlet. During analysis, the QED monitors the excitation source energy output and energy detected, baseline drift and sensor temperature. If the expected ratios differ, a warning will show asking the user to check the power supply. Any error message that shows that is critical to analyser performance prevents any more analysis from being carried out until the error is resolved.b As a final QC check of the instrument performance, a procedure called “Bookending” can be used. This is where the calibration procedure is repeated after all samples in the set have been analysed and the raw signal from the initial calibration is compared to the raw signal of the final.
The QED automatically recommends this bookending procedure is followed before saving any data. The Scanset solution is analysed after analysing a set of samples and the raw data compared to the original data from the initial calibration. A value showing the drift is displayed in the results. A drift of +/- 5% is normal. Drift above 15% triggers a QC fail and it is recommended all samples in that sequence are re-analysed. The maximum number of samples that can be analysed before bookending is 10 and just 1 sample can be bookended. The QC also triggers a bookend procedure if more than 2 hours have elapsed since the last Scanset check or the sensor temperature has changed more than the acceptable value or that a baseline has been reset more than 5 times since the last Scanset check. Once a sample set has been bookended, a new set must be started.
Results can be saved without the bookend, but the results will be missing the important drift value and OK stamp that verifies the bookend has been completed and the drift has been acceptable. The Scanset drift is not equivalent to results variance, but relates to energy output. As a guide 10% variance in Scanset gives an approximate 7% variance in the results
The analyser operator can be a significant cause of errors. Any manual method is dependent on operator ability. Laboratory technicians are highly trained yet have constant supervision and the data generated by the analyser is checked by the QC manager. On site analysers use in built QC that checks on the user input and instrument performance and flags potential errors.
The QED uses messages on a screen to guide the operator through the analytical process. Any errors such as putting the cuvette holder into the QED without a sample in, detecting an empty or partially filled cuvette, not putting the cuvette holder in at all, leaving a blank in the cuvette instead of a sample or putting the baseline check block in the cuvette, not the sample, are checked and an error message shows with the message stating the correct procedure to follow.
The software also checks for unlikely data entry parameters (125g of soil extracted and not the actual 12.5), incorrect extract volumes or incomplete data entry. It also prevents the user from over writing previous results with new results.
The QED software also monitors the sample for excessive particulate/turbidity and identifies if background organics are likely to be present, all of which can affect the results. The QED is unique in being able to identify background organics and guide the user through a procedure to remove them from the results. Background organics are a common cause of very high TPH results when using a GC method. The results also show any errors that may influence the results.
The software also checks when the last calibration was made and forces a new calibration if the time interval is more than 2 hours since the last calibration. It also forces a new calibration if the sensor temperature varies by more than a certain amount or the baseline drift has been detected more than 5 times since the last calibration event.
Current Environment Agency guidance suggests that in some cases it will be a requirement to verify the on site method results using MCERTS accredited laboratory results, as MCERTS is currently unavailable for on-site methods. There is no current guidance as to the rate of confirmatory testing required, but as a rule of thumb, if 15% of samples that have a contaminant concentration up to +/-50% of the site/regulatory limit and 5% of samples that are more than +/-50% are sent to the lab this should be adequate. For small sites, at least 5 samples up to 50% of the site limit and one each above and below 50% should be confirmed.
Sending samples for confirmation will build up a database of results showing the correlation of the on site method compared to the lab method for different soil matrixes and hydrocarbon types. It should be possible to use this database as a “lines of evidence” approach, to demonstrate the on site method correlates sufficiently well to the lab method for the target contaminant in the matrix type. This will reduce or eliminate entirely the need to send samples for confirmatory analysis. The Environment Agency accepts a “lines of evidence” approach even on regulated sites.
QC and Data Quality Objectives
The new Environment Agency acceptance of on site methods will, as stated on their website, potentially improve overall data quality and improve on the site conceptual model. This leads to a more cost effective and sustainable SI or remediation.
All analysis, whether from an MCERTS accredited laboratory or an on site analyser, has error associated with it. Current MCERTS accredited laboratory performance requirements for TPH and PAH analysis allow lab methods to have +/30% variance in the reported concentration for a performance reference sample. These are highly homogenised samples containing stable hydrocarbon or metal containing compounds. Real samples from sites that consist of non homogeneous soils will have a much wider real world variability. It is well known that if 2 samples from the same location on a site are sent to 2 different labs for analysis, 2 different results will be returned. The same laboratory will often produce different results for duplicate samples. This is not a consequence of poor lab QC, but of sample homogeneity, sample collection and transport to the lab and lab methods not being the same.
A single 500g sample taken from a point source (trial pit, borehole, window sampler) on a 20m x 20m grid is unlikely to be representative of the average contaminant concentration in a 20m x 20m x 0.5m volume of soil. This volume is 200m3, the equivalent of 15- 18 lorry loads of soil. For a similar sample taken just 1 metre away from the first, it is unrealistic to expect identical results for both samples. A solution to this problem is to analyse more samples from a site using on site analysers. The more samples that are analysed, the greater the statistical confidence in the result. When using analytical data, it is important to assess the relevance of the result. For a site where the site limit is 250 mg/kg of TPH (as diesel) any result from any method that can show an intrinsic variability of +/-30% that is below 175 mg/kg indicates the site limit has not been breached. A value of 2,300 mg/kg shows the site limit has definitely been breached. A value of 280 mg/kg should trigger more analysis of more samples in the area to determine if that 280 mg/kg is a low or high outlier, or if 280 mg/kg is a realistic average. Only by using on site analysis can an answer be derived in a timely manner. As experience with an on site method grows, the likely variance in each method will become known and this can be used to assess the relevance of the on site analysis result to the overall project data quality objectives.
The latest generation of on site analysers such as the QED and XRF readily meet the Environment Agency requirements for QC. They often exceed the QC protocols used in conventional laboratories. Users can easily obtain training on these easy to use analysers and removes the need for users to be trained analysts. The actual data from the latest generation on site analysers is also as accurate as any data from an accredited laboratory method. For volatile compounds, compounds that decompose readily, where samples are shipped on a Friday in hot weather or for samples containing more than 5% background organics, on site analysis results are likely to be more accurate and reflect the true concentration of contamination in the soil on the site. The purpose of any analysis is to identify what contaminant and how much of it is present in the soil/water on the site or in the waste soil being sent for treatment or disposal. A poorly representative single sample sent to a lab may not meet these data quality objectives.
On site analysers allow for much higher sampling densities at a lower cost and in a time frame that allows real time management decisions to be made. For site investigations, identifying where the contaminant plume is during the first intrusive investigation is cost effective and faster. The on site results inform the SI team which samples are best sent to a lab for high resolution analysis and the sample preservation procedures they should use to ensure sample integrity. Improving this type of data also improves the site conceptual model and reduces liability for incorrect assessments.
Remediation management, which is not regulated by the Environment Agency, does not require confirmatory testing. Final sign off of the remediation is the only analysis required to be of MCERTS standard. Waste classification of soil is under WM3 guidance that only requires UKAS accredited analysis where possible. The benefits of real time analysis in waste classification and handling are well documented. As most UKAS methods are lab based and take days to weeks to generate results, UKAS methods can be deemed not possible. Provided a waste producer can show the on site method is suitable for correctly classifying waste, a waste receiver and the EA will accept the data. Evidence of correlation with any site investigation results and on site results can be used to confirm the suitability of the on site method. Several major soil recyclers, muck away companies and waste management companies already use QED and XRF analysers to assess incoming soil to minimise their liability against incorrectly classified soil entering their facility.
Another Environment Agency document states that if an MCERTS accredited laboratory uses an on site analyser method that complies with the EA QC criteria for an on site analyser, the results do not require confirmatory testing. It is also possible for organisations that currently have UKAS accreditation for other analysis to include suitable on site analysers in their scope and put UKAS on the results from the on site analysers.
On site analysis can now be used for the majority of the day to day analysis needs of environmental professionals. Good quality accredited laboratory analysis is also needed, but predominantly to identify and quantify individual compounds of interest for risk assessment and toxicology purposes.