What QC Specs Must a Lab Meet?

The following criteria represent five separate QC specifications that a lab should be provided to each and every client who requests analytical services.

Minimum Demonstration of Capability

Before samples are prepared and analytes quantitatively determined by instrumental analysis, the laboratory must demonstrate that the instrumentation is in sound working order and that targeted analytes can be separated and quantitated. For example, GC operating conditions must be established and reproducibility in the GC retention times, tR, for the targeted analytes must be achieved. Competent analytical scientists coupled with laboratory resources of high quality will greatly help a lab meet this minimum specification.

Laboratory Background Contamination

Before samples are prepared and analytes quantitatively determined by instrumental analysis, the laboratory must demonstrate that the sample preparation bench-top area instrument itself and all reagents and solvents used are essentially free of traces of the targeted analyte of interest. This is accomplished by preparing method blanks using either distilled or deionized water or solvents of ultrahigh purity. This is particularly important for ultratrace analysis (i.e., concentration levels that are down to the low ppb or high ppt range). The use of pesticide residue analysis-grade solvents for conducting trace organics analysis and the use of ultratrace nitric and hydrochloric acids for trace metals analysis are strongly recommended. This author believes that there is a considerable number of organics in organic solvents at concentrations in the low ppt level! Nonzero blanks prevent true IDLs from ever being obtained. Nonzero blanks even affect analyte percent recoveries!

Assessing Targeted and Surrogate Analyte Recovery

Before samples are prepared and quantitatively determined by instrumental analysis, the laboratory must demonstrate that the surrogate and targeted analytes can be isolated and recovered to a degree by the method selected. A rugged analytical method will have an established range of percent recoveries. In TEQA, the value of the percent recovery is said to be analyte and matrix dependent. For example, if EPA Method 625 is used to isolate and recover phenanthracene from wastewater, the percent recovery is acceptable anywhere between 54 and 120%, whereas that for phenol is between 5 and 112%. These percent recoveries are seen to depend on the chemical nature of the analyte (phenanthracene vs. phenol) and on the matrix, wastewater. Provost and Elder have put forth a mathematics-based argument stating that the range of percent recoveries that can be expected also depends on the ratio of the amount of added spike to the background concentration present in the environmental sample.30 Table 2.5 applies several mathematical equations discussed by Provost and Elder and clearly shows the impact of spike- to-background ratios on the variability in the percent recovery. Using a spike-to-background ratio of 100 does not change the range of percent recoveries that can be expected. This changes as the ratio is reduced and begins to significantly increase this range when the ratio is reduced to 1 and lower.

This author has conducted more percent recovery studies than he cares to remember. A valid percent recovery study incorporates Equation (2.49) and enlarges upon it for / = 1, 2, ..., up to the maximum number of peaks in a multicomponent separation such as that which can be accomplished by applying capillary GC techniques. Using summation notation, Equation (2.53) is enlarged to encompass j replicate injections of an extract that contains the /th analyte from a spiked blank or spiked sample. A control reference standard that contains the /th analyte for к replicate injections of this standard is also prepared such that the percent recovery is found according to

TABLE 2.5

Influence of the Spike-to-Background Ratio on Percent Recoveries

Spike/Background

Variance in Mean Percent Recovery

Expected Range in Percent Recoveries

p = 1.0,

RSD = 0.1

p = 1.0, RSD = 0.2

p = 0.5, RSD = 0.2

Zero background

(lOOp) 00*(RSD)

80. 120

60. 140

30, 70

100

1.02 (100p)*(RSD)

80. 120

60. 140

30, 70

50

1.04 (100p)*(RSD)

80. 120

59. 141

30, 70

10

1.22 (100p)*(RSD)

78. 122

56. 144

28, 72

5

1.48 (100p)*(RSD)

76. 124

51, 149

26, 74

1

5.00 (100p)*(RSD)

55. 145

10. 190

5,95

0.5

13.0 (100p)*(RSD)

28. 170

-44, 240

-22, 122

0.1

221 (100p)*(RSD)

-200.400

-500. 700

-247, 347

a 95% tolerance interval for percent recoveries with assumed values for p and RSD; tolerance limits

where L is the total number of replicate injections for the /th analyte in the spiked blank or spiked sample whose peak area is Aj and M is the total number of replicate injections for the /th analyte in the control reference standard (this defines a 100% recovery) whose peak area is A

The nature of the analytical method largely determines whether a percent recovery study can be conducted in the first place. If sample prep is directly interfaced to the determinative technique, i.e., the analytical instrument, without any opportunity for analyst intervention, a control reference standard from which a 100% recovered analyte can be measured cannot be prepared. This is the case for VOCs since sample preparation is most commonly done by purge-and-trap or static headspace sampling. These sample preparation devices are directly interfaced to the injection port of gas chromatographs. Semivolatile organic compounds (SVOCs) must be extracted via application of phase distribution equilibria (to be introduced in considerable detail in Chapter 3). Some analyte is invariably lost despite the application of excellent lab technique with minimization of systematic error. Percent recoveries for SVOCs are found to be less than 100%. The analyst can prepare a control reference standard in the case of SVOCs by taking a precise aliquot of a more concentrated reference standard and dissolving this aliquot in extracting solvent and adjusting to a precise final volume. This final volume should be identical to the final extract volume used to isolate and recover all spiked blanks, spiked samples, and unspiked samples, and hence, can be compared in Equation (2.61) (i.e., comparing apples to apples) since the same amount of the /th analyte is added to both sample and control. Both sample and control are brought to the same final extract volume. To illustrate, consider spiking a serum sample with 335 ng of the organochlorine pesticide dieldrin and taking it through an appropriate sample preparation method. This sample preparation approach yields a 1 mL hexane extract that would contain dieldrin. Consider also adding 335 ng of dieldrin to 1.0 mL of a hexane extract. This important technique, used repeatedly by this author over the years to conduct precise and accurate percent recovery studies, is illustrated below:

A calibration of the analytical instrument is not necessary to conduct a percent recovery study.

A second example, this time drawing on the author’s own research, involves the calculation of the percent recovery of Aroclor (AR) from rat plasma.31 AR 1248 consists of 20 to 30 PCB congeners and was manufactured for many years in the U.S. by the Monsanto Chemical Company. Aroclors had numerous uses, ranging from an insulating medium for large electrical capacitors to newspaper ink. This extensive use was due to the stable and nonflammable properties of PCBs. Knowledge was minimal years ago about the environmental persistence of PCBs coupled to their lipophilic physico-chemical properties. Three spiking scenarios are presented below with the corresponding instrument outputs in terms of a number of counts. The number of counts is proportional to the area under a chromatographically resolved peak. In this study, the integrated area beneath each peak was summed over the entire 20 to 30 fully or partially resolved peaks for AR 1248.

Equation (2.61) addresses only one extraction of one sample. In this case of replicate extractions, the reader should refer back to Equation (2.52). This equation addresses j replicate GC injections for each of g extractions performed.

Once the %R. is found, assuming that the IDL, xD, is known, a method detection limit xMDL can be calculated for the /th analyte. This equation requires a percent recovery expressed as a fraction, R., and a knowledge of the phase ratio, (3. (3 is the ratio of extract volume to sample volume. All of this is summarized below:

Equation (2.62) suggests that in order to reach the lowest MDL, analysts should first achieve as low an IDL as possible. We already discussed how to minimize xIDL [Equation (2.34)] or xD [Equation (2.45)] from a mathematical point of view. A most efficient extraction also serves to maximize R.. However, minimizing the phase ratio, p, is only possible within the physical constraints of the experiment. Consider the difference in p between enviro-chemical vs. enviro- health TEQA. One liter of groundwater that winds up as 1 mL of an organic extract yields p = 0.001, while a 2-mL serum sample that winds up as 1 mL of an organic extract yields P = 0.5!

 
Source
< Prev   CONTENTS   Source   Next >