Triangle NoTES - March 2012

This newsletter is to explain the applicability of blank subtraction for the USEPA Method 25 analysis and problems associated with the use of blank subtraction.

The use of blank analysis is common in laboratory procedures, but there are many different types of blanks which may be analyzed and there are many different uses for those blanks depending on what they represent and how consistent they are. The blanks may be used to correct data or they may only be used to establish a level of confidence in the data set with which they are related.

The most basic is the analytical blank, which involves the analysis of a clean analyte to show contamination or background levels in the analytical system. This is the most common and easiest to use. If the purpose is to show background, and that background is consistent, that blank can be justifiably subtracted to improve the quality of the data. If the purpose is to show lack of contamination, there is generally no subtraction although it may be done in cases where the samples cannot be re-analyzed. If the blank analyses are not consistent, there is little justification for subtraction, which only leaves the impact on the quality of the data.

The next most basic type of blank is the sampling media or equipment blank. This is primarily the analysis of unused and clean media involved with sampling to ensure there is no contamination. This type of blank may also be subtracted from the sample analysis in certain circumstances, but is often simply a “pass/fail” determination for quality control. There can be several additional layers for this type of blank. The analysis of media/equipment shipped into the field, but not opened, is called a trip blank. The analysis of a trip blank which is also opened and exposed to the ambient field conditions is called a field blank. These additional levels are usually only used for quality assurance rather than any subtraction from the sample.

The most in-depth type of blank involves the sampling of a clean analyte, which includes all of the sampling equipment and protocols in addition to the analytical procedures. These may be collected in the laboratory or in the field, depending on the goal. In our case, this is often called a method blank as it involves all aspects of the method.

There has been some discussion of the use of blank subtraction for the new audit program, which if approached, would have to entail the application of similar subtractions from the samples as well. Historically, the EMC has not looked favorably on the use of blank subtraction, but has given some guidance on the preferences in collecting blank samples. The preferred methodology has been a complete method blank where the sample console, sampling procedures, tank, and trap are challenged with a clean gas collected as the audits have historically been collected. This creates a blank sample which is representative of the equipment used, the sampling crew, and the analytical procedures. The method blank is still not capable of being shown to represent other individual samples, in even the associated field sample set, because there are differences in the collection.

However, the type of blank sample which is more popular with the field crews is the equipment or trip blank. This is the recovery and analysis of a tank and trap pair which accompanied the equipment used for the field sample set. There is no collection of a sample, so the sample crew and sample console portion are mostly by-passed. As a result, this gives an equipment, recovery, and analytical blank concentration. Since there has to be a “volume sampled” in order to calculate the condensible fraction from the trap, we use the average pre-test and post-test data from the field samples for this calculation. This allows for the determination of how the blank level for this particular pairing of tank and trap would have been affected by the average volume sampled of the sample set. This gives some confidence concerning the degree of bias, but still does not rise to the comfort level the regulatory community generally has for blank subtraction. The lack of truly comparable data for these blanks makes it more difficult to justify the subtraction in many opinions.

The problem with determining an appropriate blank concentration is the number of variables which can change from sample to sample. To give a better understanding of those variables and their impact on the reported blank concentration, I would like to go over the formula used for calculating the condensible fraction and the effects on the results from changing just one of the many variables while the rest remain unchanged.

The method formula for calculation of concentration for the condensible fraction is:

(ICV Volume/Tank Volume) * (ICV P/T) / ({post-test tank P/T}-{pre-test tank P/T})* ICV conc.

(Pressure = mmHg & Temperature = K)

The ICV volume in our lab is generally ~8.3 L, but in some cases may be ~6 L or even ~4 L.

The tank volume generally may be from ~4 L to ~8.3 L depending on the equipment.

The pre-test pressure is assumed to be 0.0 rather than <10 mmHg for ease in calculation.

The post-test pressure is assumed to be a maximum of 728 mmHg, which is ~8 L in an ~8.3 L tank

The temperature is assumed to be 300 K for ease in calculation.

The ICV concentration is assumed to be measured at 10 ppmC for ease in calculation.

With this set of assumptions, and all other things being equal, the reported blank for the ~8 L sample volume in this set would be ~16.5 ppmC, which is pretty reasonable.

If the post-test pressure is changed to 637 mmHg to mimic a ~7 L sample volume, the reported blank jumps up to a ~19ppmC level, which is not bad.

If the pressure is changed to 546 mmHg for a ~6 L sample, we see the concentration changing to ~22 ppmC, which is only slightly worse.

If the pressure is changed again to 456 mmHg for a ~5 L sample, the concentration climbs to ~26.5 ppmC , which is not as good.

If the pressure is lowered to 364 mmHg for a ~4 L sample volume, the concentration again climbs to ~33 ppmC, which is questionable at best. This should be near the lowest sample volume taken, but the comparison will continue.

Dropping the pressure to 273 mmHg for a ~3 L sample volume gives a reported blank to ~44 ppmC. The sample volume can also be changed by changing the volume of the canister used.

The final pressure in the ICV will also have an impact on the reported blank concentration. For a ~4 L sample, the concentration would increase ~3 ppmC on average per each 100 mmHg increase from the minimum pressure required for analysis.

These changes only illustrate the effects of measurement variables and not other sampling conditions, which may also impact the individual sample results and can make a blank subtraction even less appropriate. Factors specific to an individual sample and the collection thereof cannot always be compared to other samples individually. A blank can currently only give a level of confidence in the quality of the equipment, personnel, and procedures, but should not be used to change reported sample concentrations.

We, as a laboratory, have collected a large number of modified method blanks and internal audits over the years that give us an expected range and an average of those blanks. This, of course, excludes the blanks and audits used to test equipment suspected of contamination or other possible issues we would not want to have impact a client sample. The problem with using this blank data for subtraction from field samples is that we take a conservative stance on everything we do. This means we use the minimal volume sample for our blanks, which is 3.6 liters. With all things equal, a larger sample volume should correspond to a lower reported blank concentration. Thus, the blank for our 3.6 L sample should be twice that of a 7.2 L sample. This proportional relationship makes it possible to achieve a level of comfort concerning bias due to the blank backgrounds, but there is still enough variation to make it less of a general concentration for all samples and more specific to a few particular samples.

The typical blank sample generally has a reported concentration for only the trap or condensible fraction. However, there are some which also have reported concentrations in the tank or non-condensible fraction. This means there may be cases where the tank fraction of the blank sample should not be considered representative of the actual sample, especially if the blank tank sample concentration is higher than the sample tank fraction concentration. Generally, the blank sample concentration being discussed is just the portion in the trap and not the potential for the tank fraction. For example, the true blank samples where there were no reported concentrations for either fraction we have collected and analyzed in the lab have a total of <9 ppmC or <10 ppmC

depending on the pressurization of the tank and ICV. A blank sample with only 10 ppmC in the trap and <5 ppmC in the tank is generally called a 10 ppmC blank rather than a <15 ppmC blank.

We have also seen cases where sets of field samples were reported below the average level of any blank analysis we had performed. This anomaly is related to the previously mentioned effect of sampling variables on the reported blank concentrations, but also may indicate the possibility of the presence of an organic compound affecting the factors going into the recovery of blank samples.

To examine the possible impact of blank subtraction on the historical data, I used the data provided for the last phase of the US EPA audit program for only the samples containing less than 150 ppmC. This was taken from the data provided by the SSAS audit committee from the EPA audit program. This is the most recent data for which the field collection and laboratory analytical procedures are combined with verifiable concentrations for the sample stream. Also, any blank subtraction would have the greatest impact on the lower level audits, which is the reason for the cut off at 150 ppmC. The fail rate (>+20% reported variation from the known concentration) for samples in that range was 41.4%. If a 10 ppmC blank were subtracted across the board from the reported concentrations the fail rate increases to 42.0%. The increase of the blank subtraction to 15 ppmC results in a fail rate increase to 50.3%. Unless there had already been some type of blank subtraction performed on the reported audit concentrations, this indicates such a process would not have improved the average accuracy of the analysis of these audits.

Wayne Stollings

Triangle Environmental Services, Inc.