How clean is clean enough? You test for a contaminant. The results indicate contamination level is zero, but the product still fails – it’s as if the contaminant is still there. What happened? Even with a clean manufacturing facility, you can be caught in a detection dust storm, one that doesn’t let you see where there’s a product contamination problem. Understanding three common reasons for these dust storms allows you to choose the right analytical and monitoring tests for your critical cleaning process. The right test design helps avoid costly production shutdowns and keeps customers happy.
Detection dust storms result from
• too little of the stuff you are trying to detect
• too much of the stuff you are trying to detect
• too much other stuff
Too little stuff
How low can you go? Typically, you can’t detect a single molecule of contaminant; and you don’t need to. Analytical and monitoring tests are geared to specific requirements; some are more sensitive than others. Let’s look at Figure 1.
Figure 1 – Meaningful Detection
At low levels, the inherent variability of the test, the analytical “noise,” masks the analyte. Ok, translation from lab speak to English: analyte is a fancy word for the chemical or group of chemicals being tested. Ask the lab analyst about the Limit of Detection (LOD) for the method. The LOD is the lowest concentration that produces a response that is detectable above the noise. This is typically an S/N (signal to noise) of 3. However, being just above the LOD is not good enough to reliably indicate the quantity of contaminant. For this, you need to understand the LOQ or Limit of Quantification. The LOQ is the lowest level of analyte concentration that can be determined with acceptable precision and accuracy. The LOQ is often (not always) a S/N of 10.
Beware of a test report that reads zero. What that probably means is “below X,” where X is the LOQ or maybe the LOD. Sometimes, design agencies or customers or even regulatory agencies expect you to see low levels of contaminants in the dust storm of noise. If there’s a chronic problem, investigate the method. It might be possible to get the requirements changed (we can always dream).
Just the right amount of stuff
Ah, the comfort zone of the dynamic range!
Let’s get back to Figure 1. Moving left to right from the LOQ, we see an indication of the dynamic range. This is where you want to be in the world of analysis. All other things being equal, a technique with a higher slope – where the line is steeper is going to be more sensitive than one with a lower slope. Of course, there needs to be a balance; a dynamic range that is a vertical line is like an on/off switch. It is very sensitive at one point; but there’s no range to the dynamic range.
Too much stuff
We leave the comfort zone at LOL. No, this does not mean Laugh out Loud (although the results may become laughable). LOL stands for the Limit of Linearity. It’s where you are entering another dust storm, a storm where there are too many molecules of the stuff you are trying to detect. An increase in level of the contaminant doesn’t change the observed signal; the detector is saturated. The detector behaves somewhat the same way your eye does during a dust storm – you can’t see dust that’s behind other dust. Quantifying at higher levels may be as simple as diluting the sample.
Too much other stuff
Counting diamonds in a compost heap is harder than counting diamonds in the window of “Cartier.” That reliable analytical detector has the same problem. Complicated mixtures act like a dust storm, or a pile of compost. This can mean problems at low levels, or even in the middle of the dynamic range.
Changes in the manufacturing process can compromise detection results due to too much other stuff. You may have problem contaminants but are no longer able to detect them reliably. Let’s say you change suppliers but don’t specify metalworking fluids or the cleaning process – this can mean an increase in other “stuff.” Even if you have control of the processes, beware when there’s a new or modified product line.
Avoiding stuff
Good test design helps prevent detection dust storms. At the stage of test design, things like the use of standards, controls, internal controls, and recovery studies can improve accuracy, especially near the lower limits. Good test design might be called out in standards or requirements – or not! It may be productive to set up standards and controls specific to your facility and requirements.
Ask the analyst about test design and test details. Some analysts alert the manufacturer to problems; others simply supply a number. It’s better for your bottom line to find an analyst who understands what you want to achieve, who worries a little, and who communicates problems right away.
References:
D. Williams, E. Kanegsberg, and B. Kanegsberg, Lecture Notes, Product Quality Cleaning Workshop, Sam Houston State University, May 2018.
B. Kanegsberg and E. Kanegsberg, “How Low?” Controlled Environments Magazine, Jan, 2008
https://www.cemag.us/article/2008/01/how-low
Figure 1 by kkmurray on Wikipedia
https://commons.wikimedia.org/wiki/File:Calibration_curve.png
