Quality Assurance (QA) - Quality Control (QC)
On this page:
- What is the purpose of analyzing the matrix spike (MS) sample versus analyzing the laboratory control sample (LCS) and why should we run both?
- Why do many of the SW-846 quality control (QC) operations have to be run "once for every 20 samples?"
- What is the procedure when matrix interference effects cause elevated TCLP Lower Limit of Quantitation (LLOQ) that are above the TCLP regulatory limits?
- For SW-846 Method 5030 combined with Method 8260, is the Tune Batch equal to the Analytical Batch?
- What is the SW-846 position concerning the use of accuracy measurements determined from the matrix spike in place of the laboratory control sample?
- Can you clarify the quality control (QC) requirements for SW-846 Method 9060A, Total Organic Carbon (TOC)?
- Spiking surrogates for SW-846 Methods 3510C and 3520C.
- For SW-846 Methods 1311 and 1312, is it an acceptable alternative to use internal standards for all leachates?
- Differences in quality control (QC), preservation, and holding time recommendations among SW-846 parts.
- Are Method Detection Limit (MDL) studies required for SW-846 test methods?
What is the purpose of analyzing the matrix spike (MS) sample versus analyzing the laboratory control sample (LCS) and why should we run both?
The MS/MSD results are an important measure of the performance of the method relative to the specific sample matrix of interest. The Office of Resource Conservation and Recovery (ORCR) believes that such a demonstration is an important aspect of an overall quality assurance program, and is particularly important for the RCRA program, which governs a wide range of different matrices.
The primary purpose of these MS/MSD analyses is to establish the applicability of the overall analytical approach (e.g., preparative, cleanup, and determinative methods) to the specific sample matrix from the site of interest.
The primary purpose of the laboratory control sample (LCS) is to demonstrate that the laboratory can perform the overall analytical approach in a matrix free of interferences (e.g., in reagent water, clean sand, or another suitable reference matrix) and its analytical system is in control.
Therefore, the LCS results should be used in conjunction with MS/MSD results to separate issues of laboratory performance and "matrix effects."
ORCR believes that consistent trends in MS/MSD results can be of some use in evaluating laboratory performance, as are trends in surrogate recoveries, LCS recoveries, and other QC data. Appropriate use of a single set of MS/MSD results is to evaluate method performance in the matrix of interest, not to evaluate laboratory performance.
Why do many of the SW-846 quality control (QC) operations have to be run "once for every 20 samples?"
Why do so many of the QC operations (MS, MSD, LCS, blanks, etc.) have to be run "once for every 20 samples?"
The 1 per 20 (5%) frequency is a typical value that has been used in many EPA programs for many years. The Office of Resource Conservation and Recovery (ORCR) also recognizes that other frequencies may be appropriate under certain circumstances. For example, in the case of a long-term monitoring project involving a small number of analyses of a sample matrix that does not change, it may not be necessary to prove that the method applies to the matrix each time that samples are collected and analyzed. Therefore, the MS/MSD analyses may be run less frequently than 1 every 20 samples. To that end, ORCR recommends that, if another frequency for the QC analyses is chosen, it be clearly documented in a sampling and analysis plan that is reviewed and approved by the relevant regulatory authority.
What is the procedure when matrix interference effects cause elevated TCLP Lower Limit of Quantitation (LLOQ) that are above the TCLP regulatory limits?
For example, if you analyze a soil sample for TCLP SVOCs and you have a LLOQ for 2,4‐Dinitrotoluene (D030) of “<2.0 mg/L,” how is this handled if the TCLP regulatory limit for 2,4‐Dinitrotulene is 0.13 mg/L? Based on generator knowledge, you know that there should be no 2,4‐Dinitrotoluene at the site ‐ however you can’t disprove a negative since the LLOQ is higher than the TCLP regulatory limit for this constituent.
According to Table 7‐1 footnote (2) in Chapter Seven of SW‐846, Characteristics—Introduction and Regulatory Definitions, several contaminants, including 2,4‐dinitrotoluene, have quantitation limits greater than the calculated regulatory level. In such cases, the quantitation limit becomes the regulatory level.
However, for this to be valid, the laboratory should take every step possible to keep the reporting limit as low as possible (e.g., avoid unnecessarily high sample dilutions, use a clean‐up method, etc.).
For analytes not covered by the footnote, it is expected that the laboratory will use analytical procedures that achieve quantitation limits at or below the regulatory level.
For SW-846 Method 5030 combined with Method 8260, is the Tune Batch equal to the Analytical Batch?
For method 5030 combined with method 8260, is the Tune Batch equal to the Analytical Batch, i.e., does all of the associated QC analyzed within a Tune Batch equal that which should be included in the Analytical Batch (method blank, laboratory control sample, matrix spike/matrix spike duplicate)?
In the following example, analysis batch is interpreted as being within a tune window, which would make the analytical batch equivalent to the tune:
Analysis batch – A group of up to 20 samples, sample extracts, or sample digestates (including QC aliquots), that are analyzed together on the same instrument. For the analysis of volatiles, there may be no sample preparation equipment other than that attached directly to the determinative instrument (e.g., Methods 5030B and 8260B), so the analysis batch drives the frequency of the method blank and LCS for volatiles, as well as the frequency of calibration verification standards for methods using external standard calibration. The limit of 20 in the analysis batch includes all the analyses, including the method blank, LCS, MS, and MSD, so that an analysis batch for volatiles will include fewer than 20 field samples.
The statement that an analysis batch will have <20 field samples implies that the entire batch must fit within the tune window.
In the end it comes down to how a regulating body, site QAPP, or QA/QC program defines analysis batch. Most labs seem to tend toward the tune window being the equivalent of the analysis batch to avoid problems with auditors and comply with as many programs as possible.
What is the SW-846 position concerning the use of accuracy measurements determined from the matrix spike in place of the laboratory control sample?
Please provide the SW-846 position concerning Section 126.96.36.199.3 in V1M4 of the 2009 TNI standard which allows the use of accuracy measurements determined from the matrix spike to replace that of the laboratory control sample (LCS) in establishing analytical process control. The section states this is acceptable as long as the matrix spike (MS) meets LCS criteria. Is this acceptable practice and when should it apply (e.g., when one or more analytes fail recovery criteria in an LCS or only when an LCS is not performed)?
Under performance-based methodology, the official SW-846 position is the use of MS in place of the LCS as long as the acceptance criteria are as stringent as for the LCS, and it meets the needs of the project, it would be allowed, just as it is in the TNI standard.
There is a downside, however. Oftentimes, especially in metals, matrix spike amounts are not appropriate for the levels seen in the native sample and recoveries are not calculated. Therefore, if no LCS is available or if the LCS failed, there may be no accuracy check information at all on a particular parameter or several parameters from MS data alone. This coupled with the fact that MS is in a real matrix with matrix effects, and there can be sampling differences between the native sample aliquot and the spiked sample aliquot, getting the MS to come in under LCS criteria will be extra challenging, especially for multiple analytes. In addition to meeting the acceptance window criteria for the LCS, we would view the LCS acceptance criteria to also include the minimum number of analytes listed under 188.8.131.52.3.
We would view the note in the TNI standard to be used as an occasional "batch saver" wherein, for whatever reason, the LCS failed or was not available, but you have back-up accuracy data from the matrix spike, and use that instead to be able to report data for that analyte for that batch of samples. You would not want to rely on MS data instead of LCS data for accuracy checks on a routine basis, especially for multi-analyte methods. Both forms of quality control are needed on a routine basis.
Can you clarify the quality control (QC) requirements for SW-846 Method 9060A, Total Organic Carbon (TOC)?
Section 7.6 states, "Quadruplicate analysis is required. Report both the average and range." Is the definition of "quadruplicate" analysis: four complete analyses of the same field sample, with analyses meaning separate "injections/aspirations" of the sample? Is this definition correct?
Section 8.3 states, "Verify calibration with an independently prepared check standard every 15 samples." Does this frequency mean 15 samples analyzed in quadruplicate (i.e., every 60 injections/aspirations – 15 samples times 4 injections/aspirations)? Or does it mean every 15 injections/aspirations (i.e., every 3.75 quadruplicate sample analyses)?
Would the QC samples (i.e., Laboratory control sample (LCS), Matrix spike/spike duplicate (MS/MSD), Calibration verifications (CCV), Instrument blanks (CCB) analyzed in a typical "CCV/CCB" set) also require quadruplicate analyses?
The four replicate analyses are obtained through four injections from a single field sample, and make up a single analysis for that sample. The arithmetic mean concentration is reported as the analyte concentration in the sample, and the calculated standard deviation on the mean from the replicate analyses is used as a QC diagnostic.
Calibration verification frequency should be based on the number of unique samples analyzed (not including duplicates of the same sample). This criterion refers to the analysis of 15 field samples, as well as to the analysis of QC samples; i.e., matrix spike samples, duplicate spike samples, field duplicate samples, and/or equipment blank samples. Thus, if 13 field samples + 2 QC samples are analyzed, the calibration verification standard must be run next.
QC samples should be analyzed in the exact same manner as the field samples. This includes the LCS, MS/MSD, initial or continuing calibration verification (ICV or CCV), and continuing calibration blank (CCB), etc. So, if the field samples are analyzed in quadruplicate, the QC samples should also be analyzed in quadruplicate.
Other Category: 9000 Series
Spiking surrogates for SW-846 Methods 3510C and 3520C.
Methods 3510C and 3520C state to spike surrogates into the graduated cylinder or sample bottle and mix well. Multiple transfers introduces potential for losses and mixing well in a graduated cylinder is difficult.
Would it be acceptable to transfer the sample to a separatory funnel for 3510C then add the surrogates? This minimizes the possibility of cross contamination from the graduated cylinders.
Knowing that it is important to already have solvent in the continuous liquid/liquid extraction (CLLE) body before the addition of the sample or the water to the boiling flask, would it be acceptable to add the surrogates to the upper (aqueous) phase in the CLLE?
It would be acceptable to add the surrogate spiking solution directly to the separatory funnel rather than to a graduated cylinder or the sample container. Even though the cylinder and container are solvent rinsed, there is a remote chance a portion of the surrogates could be retained on the glass walls. Should you choose to spike directly into the sample volume in the separatory funnel, be sure to perform this step prior to pH adjustment and the first solvent addition.
It would also be acceptable to set up the sample and methylene chloride in the CLLE and then spike directly into the upper aqueous phase, provided your spike is in a water miscible solvent.
However, since the final decision on whether this change is acceptable will come from your regulatory agency/accrediting body, it would be advisable to contact them first.
Other Category: 3000 Series
For SW-846 Methods 1311 and 1312, is it an acceptable alternative to use internal standards for all leachates?
Both Method 1311 and Method 1312 have language specifying the use of an internal calibration quantitation method for each metallic component if the spike recovery for the element is < 50% and the concentration of the contaminant falls in the range from 80 to 100% of the regulatory level. Both of these methods then specify that “the method of standard additions shall be employed as the internal calibration quantitation method for each metallic component.”
For Methods 1311 and 1312, is it an acceptable alternative to use internal standards for all leachates to meet the requirement of an internal calibration quantitation method rather than using methods of standard additions?
The purpose of the language in Section 8.4 of the methods is to remove the effects of matrix suppression that might cause a sample extract within 20% of the regulatory limit to appear to be below the regulatory limit. When 1311 and 1312 were written, atomic absorption techniques were most commonly used and could be strongly affected by matrix effects. Since you are free to choose the analytical method for TCLP and SPLP extracts, choosing an inductively coupled plasma (ICP) method, which is far less prone to matrix effects, will make the situation described in Section 8.4 of Methods 1311 and 1312 much less likely.
Use of dilution or internal standards, or both, increases the likelihood of a passing matrix spike recovery above 50%, removing the need for the method of standard additions (MSA). If dilution and internal standards do not work and spike recoveries are still below 50%, then MSA may be the only option left. This would be very unlikely if ICP analysis is used.
Internal standards are not mentioned anywhere in either Method 1311 or 1312. As mentioned in Section 8.4.1 of both methods, "The method of standard additions shall be employed as the internal calibration quantitation method for each metallic contaminant." Use of an analytical method that utilizes internal standards (whether ICP-MS or ICP‐AES) does not eliminate the need for MSA when spike recoveries are below 50%. Section 8.2.3 allows you to use "other internal calibration methods, modification of the analytical methods, or use of alternate analytical methods to accurately measure the analyte concentration in the [TCLP or 1312] extract when the recovery of the matrix spike is below the expected analytical method performance." "Accurately measure" in that statement means acceptable spike recovery.
Other Category: Hazardous Waste Characteristics
Differences in quality control (QC), preservation, and holding time recommendations among SW-846 parts.
When there are differences in the QC recommendations between Chapter One, a general method (e.g. 8000), and a determinative method (e.g. 8082), what is the proper precedence?
Would recommendations for preservation and holding time follow the same precedence as QC?
The precedence for holding time and preservation is the same as that for QC. According to Chapter Two of SW-846, QC guidance specific to a given analytical technique (e.g., extraction, cleanup, sample introduction, or analysis) may be found in general methods 3500, 3600, 5000, 7000, and 8000.
When inconsistencies exist between the information in a chapter, a general method, or a specific method, method-specific QC criteria take precedence over both criteria found in the general method and those criteria given in Chapter One. QC criteria in general methods take precedence over the criteria in Chapter One.
Other Categories: Holding Time & Preservation, General
What is meant by the Lower Limit Of Quantitation (LLOQ) for SW-846 test methods?
SW-846 Chapter One defines the Lower Limit Of Quantitation (LLOQ) as: The lowest point of quantitation which, in most cases, is the lowest concentration in the calibration curve. The LLOQ is initially verified by spiking a clean control material (e.g., reagent water, method blanks, Ottawa sand, diatomaceous earth, etc.) at the LLOQ and processing through all preparation and determinative steps of the method. Laboratory-specific LLOQ recovery limits should be established when sufficient data points exist. See Section 9 of Methods 8000D and 6010D/6020B for additional guidance in establishing, implementing and verifying LLOQs for organic and inorganic analytes. LLOQs should be determined at a frequency established by the method, laboratory’s quality system, or project.
Other Category: Detection & Quantitation
Are Method Detection Limit (MDL) studies required for SW-846 test methods?
Method Detection Limit (MDL) is no longer defined in Chapter One, nor is it referenced in the SW-846 methods. Does that mean that MDL studies are no longer required?
The EPA Office of Resource Conservation and Recovery (ORCR) that publishes the SW‐846 test methods manual no longer uses the MDL, and promotes the use of the Lower Limit of Quantitation (LLOQ) approach for establishing the reporting limit for a respective test method or laboratory SOP. EPA ORCR is in the process of transitioning existing SW‐846 methods and preparing new methods to specifically address the LLOQ approach and procedures for establishing, implementing, and verifying it.
Other EPA programs (e.g., the Clean Water Act (CWA)) and projects still utilize the MDL concept. In those cases when an MDL is needed, method users may follow the procedure stipulated in Appendix B to 40 CFR Part 136 for determining an MDL.
Other category: Detection & Quantitation