Introduction

This is an introduction to quality assurance of chemical measurements. Quality assurance is defined as the records kept on the results of the routine analysis of quality control samples. Many laboratories mistakenly assume that merely running quality control samples constitutes an adequate quality assurance program. This is incorrect. In fact, without proper and ongoing documentation of quality control sample results quality assurance does not even exist.

This is not a lesson in statistics, however, a knowledge of statistics is required. Neither is this an analytical chemistry lesson, but without some knowledge in chemical analysis there is really no need to read further.

Quality Control

Quality Control consists of either the analysis of samples of known quantities for the purpose of verifying a method's accuracy or the repeat analysis of a sample to determine the methods precision. Quality control samples may be relatively clean interference free matrices, or complex matrices that duplicate the sample. Results may either be recorded as absolute or relative percent recovery.

Blanks consist of all reagents used in a test and may contain everything in the sample except the analyte of interest. The purpose of the blank is to assess laboratory contamination. High, or variable blank values indicate a contamination that needs to be located and eliminated.

Blank Spikes are blanks to which a known amount of analyte has been added. Blank spikes largely determine whether significant analyte is lost during sample processing. Since the blank matrix is interference free a high blank spike result is further indication of contamination, or an inadequate calibration.

Blank Spike Duplicates measure the ability of a method to duplicate analytical results in an interference free matrix. Bad precision indicates either loss of analyte (lower than expected recovery) or contamination.

Matrix Spikes are real samples to which a known amount of analyte has been added. Subtracting the amount of analyte determined in an unspiked portion enables calculation of the percent analyte recovered from samples of that matrix.

Matrix Duplicates are repeat analyses of a sample matrix used to evaluate precision. If the amount of analyte is expected to be near or below the Method Detection Limit (MDL), Matrix Spike Duplicates are often run allowing precision to be evaluated.

Method Detection Limit (MDL) is a statistically determined number that represents the lowest concentration of analyte that can be detected with the confidence of not being a false reading. One popular calculation of MDL multiplies the standard deviation of seven replicate tests by 3.14. The replicate tests should be blank spikes with an analyte concentration 3-5 times the calculated MDL.

It is important for all users of this statistically derived MDL to realize the great inaccuracies associated with this number. The MDL that is determined by analysis of replicates made on purified water only applies to the purified water. This number generated also only applies to the analyst that made the determination and the instrument that was used. Also, statistically speaking there is no real accuracy or precision associated with this number, as variability can be as high as 100%.

Minimum Level, or reporting limit is the lowest calibration standard, or a concentration of 3.18 times the MDL. The minimum level is approximately 10 times the standard deviation of the noise and represents the point where data has an accuracy and precision of within about 30 % of its true value.

A more accurate determination of the minimum level is to plot RSD and Recovery of collected multiple laboratory data and determine the lowest point where both accuracy and precision are within 30%.

Calibration is a representation of a response that is in proportion to an amount. In modern instrumentation the calibration is an electronic signal relative to an amount of analyte. A graphical plot of concentration versus signal is represented by a calibration curve, which is hoped to be linear, but may be second or third order depending on the measurement method and concentration range. Calibration could, however, also represent mass measured on a balance or volume measured with a burette.

Conclusion

Measurement techniques are moving more and more towards instrumentation leaving behind chemical methods such as gravimetric precipitations and volumetric titrations. The problem introduced by strictly instrumentation analysis is that instruments require known calibrants that responses of unknowns can be compared to. As the classical volumetric and gravimetric chemical approaches to analysis and measurement are gradually forgotten we are gradually losing the ability to prepare new calibration standards for our instruments. Also, classical techniques are more accurate and precise in high purity chemical assays while instrumentation is best at trace analysis. A laboratory does itself service by maintaining classical methods using instrumentation for trace analyses such as environmental testing, or the verification of the purity of precipitates.

William Lipps
http://www.oico.com
(979) 690-1375 ext. 230
wlipps@oico.com

One Extraction, One Dilution, One Injection, Automation solutions. Segmented Flow, Flow Injection, Discrete Analyzers, TOC Analyzers, Sample Introduction, Gas Chromatography, Gel Permeation, Purge & Trap

Expert Methods Consulting
http://www.williamlipps.com

Classical Wet Chemistry, HPLC, IC, GC, GC/MS, SFA, FIA, Discrete, TOC, TOX, UV-VIS, IR, Osmometry, Petroleum, Proximate Analysis, Extractions, Digestions