What does it mean to be NIST traceable?
By Janine Arviszu, Certified Quality Auditor
[Editor’s note: Too frequently in forensic science, laboratories or laboratory personnel claim that their process or even their whole laboratory is “NIST traceable” as if to act as a shield to ward off external validation. In this article expert witness Janine Arviszu explains what it means to be truly NIST traceable.]
ONLY measurement results can be traceable.
A laboratory can’t be traceable. An accredited laboratory isn’t traceable, and its results aren’t necessarily traceable.
An instrument can’t be traceable. Even if an instrument was calibrated by NIST, it doesn’t make its results traceable.
A solution can’t be traceable. Only the measurement result for a solution can be traceable, and only if the solution was appropriately prepared, maintained, used, and documented.
In order for the measurement result of this solution to be traceable, it would require the following:
1. What is it? The clearly defined quantity being measured. In this instance, notice that the analyte (i.e., the parameter being measured) is never even defined. They say it is traceable to NIST ethanol standards, but they never even document that they are certifying the concentration of ethanol. A truly bush league mistake.
2. An unbroken chain: Comparisons from the measurement result reported by a laboratory all the way back to a nationally recognized primary standard.
3. Documentation: Every link in the chain must be performed according to documented procedures, and the results of these procedures must be documented. This documents the measurement system. Do such procedures exist?
4. Competence: Laboratories performing steps in the chain must have demonstrated competence (e.g., accreditation to ISO 17025 for testing labs and ISO Guide 34 for reference material producers). This Certificate of Analysis doesn’t even document the name of the responsible laboratory (another bush league mistake, unless of course, Ronald D. Oliver thinks he is the certifying entity), so their accreditation status can’t be evaluated.
5. Measurement assurance: the laboratory must systematically establish the status of reference materials and working standards at all times pertinent to a given result. That is, you can’t simply assume that everything was working properly at the time you did your work, you have to measure it and prove it. What was the calibration status of the analytical balance used? What was the calibration status of the pipettes used? What was the temperature at the time the solution was prepared, and what was the calibration status of the temperature measurement device?
6. Measurement uncertainty: the measurement uncertainty must be determined for each link in the traceability chain, and the measurement uncertainty must be reported for the final measured result.
Under the provisions of the ISO Guide 34, the assigned value of a solution can not be determined based only on statistical analysis of characterization data. That is, when you prepare a reference material, it is typically prepared gravimetrically (using analytical balances to weigh the mass of ethanol) or volumetrically (using Class A glassware or calibrated pipettes to measure a volume of ethanol) from stock materials with certificates of analyses. In either case, the calibration status of the equipment used to make the solution must be documented, as well as the temperature at which the solution was prepared. These values are key to the assigned value of the solution and its associated uncertainty.
Read the Certificates of Analysis provided by Cerilliant and NIST. They identify the laboratory. They describe the preparation and verification of the solution (e.g., gravimetric preparation with chromatographic and titrimetric verification).
This “prepared from scratch” approach does not appear to be the method used by all laboratories or all vendors. Some rely solely on statistical analysis of the analysis results to determine the assigned value of their “reference material.” This is not acceptable under the ISO standard.
This is important. For example, if the ethanol concentration of their simulator solution was traceable, it would not mean that the result from an instrument that used that simulator solution would in turn be traceable. The same is true with Gas Chromatography or any other analytical device.
Janine Arvizu is a chemist (BS biochemistry, Cal Poly State University at San Luis Obispo; ABD chemistry, University of New Mexico)and ASQ Certified Quality Auditor who helps users of forensic results understand laboratories and the reliability of their work. She established and managed an analytical chemistry laboratory for the Department of Energy, and served as Program Manager for the U.S. Navy’s nationwide laboratory Quality Assurance Program. She has audited the work of government and commercial laboratories, and testified as an expert witness in state, federal, and international courts on issues of laboratory quality assurance and data quality assessment.