The impact of GC peak width on detection results
Thread Content
I would like to ask fellow users: for the same product sample, the peak width of the parameters we measured was 0.05 s, and the test result was 99.3%, which was deemed unacceptable (the acceptable standard is 99.7%). When the sample was sent to the customer for testing, the peak width was 0.1 s and the test result was 99.8%, which was considered acceptable. Why does a difference in peak width have such a big impact? With two peak heights appearing in such a short time, could it be the same substance? I’m constantly arguing with the sales team over this issue, which is quite annoying. Is there any way to silence them?
Give it a push; I hope there’s still a solution.
Compare and discuss this with the customer; for example, use the conditions set by the customer to generate results and check whether they are consistent. Also, see if your company’s analysis can still achieve 99.8% accuracy with a condition peak width of 0.1s. If the customer is satisfied with this, the product specifications can be relaxed accordingly. You can also take it to a third party for testing and make a comparison. The peak width probably doesn’t have such a big impact on your results; you need to look for the cause elsewhere
I carried out a comparison as instructed by the team leader, and found that the three GC units gave different results using the same testing method: A (Quality Control) showed 99.6%, with two peaks； B (central control) 99.8%, one peak ； C (domestic customers) 99.93%! One peak. Ah! I wonder what foreign customers will do? Through comparison, only the A quality control GC showed two peaks; we plan to try using a different column.
I’m very curious about the methods you use – to achieve 99.8% and 99.93% accuracy with a single peak, is it the external standard method? I feel that the method you’ve chosen is not the right one; for high-purity analysis, impurities are generally analyzed in order to calculate the purity level
I agree with what was said on floor 5: for high-purity samples, the focus is on detecting impurities and then calculating purity. The area normalization method has its drawbacks; it can only be used if all substances present in the sample can be detected. Also, when there is one peak as mentioned by the original poster, the result obtained using the area normalization method should be 100%, right? Why is it 99.8%? I didn’t understand
I didn’t express myself clearly; I’m referring to the small peak behind the main peak, not the peak that is present at all times.
If it can be confirmed that the impurity in the product appears as only this single peak, you can use this substance for calibration and determine the concentration using the external standard method; ignore the larger peak. This will make the calculations easier and ensure more accurate results
This post was last edited by qhyhxjl on 2017-9-27 at 21:23. The problem is that it’s not possible to identify the substance due to the impurity peaks; several possible substances were suspected and tested, but none matched, making it entirely impossible to provide guidance for production.
The original poster must have made guesses based on the production processes and accompanying products, right? Assuming the substance you are guessing is Substance B and your product is Substance A, would preparing a low-concentration sample of Substance B directly, using Substance C as the diluent, result in a retention time for Substance B that differs significantly from the retention time when testing the product itself, which is Substance B? If that’s the case, things will definitely be different. It is recommended that the original poster add substance B to the products that have already been tested, and see whether the peak area of the small peak increases compared to that of the peak in the product under test. If it does increase, an additional amount of substance B can be added. If the increase in peak area shows a roughly linear relationship with the amount of substance B added, it can be concluded that the impurity is substance B – of course, this is assuming that the impurity peak in question is a mixed peak. If the original poster has already carried out work in this area, it is still recommended that they send sample products to a third party for testing for impurities in order to determine their types. This will be very helpful for analyzing the impurity content in samples in the future, and it will also help avoid disputes with sales parties. This is just my personal opinion, for reference only