Analysis/Waste Treatment and Data Analysis ② Aiming to Generalize Error Factors from Analysis Results

Hello, this is Kazunami.

There is a follow-up to the article I posted the other day.
Please read this as a continuation. (__)

Follow-up

Actually, it turned out that the indoor air conditioning was malfunctioning during this measurement, and the room temperature was significantly higher than the specified level (25°C) during the measurement.

Therefore, when I looked at the relationship between the sensitivity of each substance in the MS (mass spectrometer), which was thought to be affected, and the z-value, I found that...

It was found that substances with higher MS sensitivity tend to have larger z-values (degree of deviation).

Furthermore, when I performed multiple regression analysis using the physical property values of each substance used earlier and the MS sensitivity, I was able to regress with high accuracy using the same physical property values as the previous z-value.

In other words, it was found that MS sensitivity, like the z-value, can be explained by the index of "ease of escaping into the air of the substance."

Modeling

Regarding this factor, it was considered that it could be modeled as a monomolecular adsorption line (*) where the horizontal axis is MS sensitivity (vapor pressure of the substance) and the vertical axis is the z-value indicating error (amount of substance adsorbed within the MS system) on the z-value vs. MS sensitivity relationship line.

(*) Assuming that the adsorption characteristics (following the Langmuir equation) depend on the type of substance, and treating the vapor pressure approximated by multivariate analysis as its change under constant temperature, I thought that although there is a deviation from the model, it can be explained to some extent.

In other words, it was inferred that due to the temperature change (temperature rise) during measurement, the adsorption/desorption state of each substance within the MS system, including the interface, changed, and substances with a higher "ease of escaping into the air" were more easily desorbed than those with a lower one, existing in the system at a higher concentration, thus affecting the z-value.

Verification

At a later date, as a verification experiment, after creating a calibration curve, I turned off the air conditioning and performed measurements in an environment where the room temperature increased.

For the measurement, fluorobenzene was used as an internal standard, and the concentration of each substance was calculated from the ratio to its intensity.
Also, the MS sensitivity of this substance was calculated from physical property values in the same way as other substances.

The results are shown in the table below.
It was found that substances with lower MS sensitivity than the internal standard substance had lower measured concentrations due to temperature rise, while those with higher sensitivity (benzene) had higher concentrations.

From the above, I was able to explain the VOC analysis error this time using the "ease of escaping into the air" index.

Summary

Due to the property of "ease of escaping into the air,"
it was found that it affects multiple processes:
① Standard sample preparation process
② MS measurement sensitivity

This is considered applicable not only to analysis but also to resource recovery processes (such as distillation) that occur through similar mechanisms, such as setting appropriate control parameters.

In other words, the more you explain (generalize) a phenomenon withmore fundamental properties,the higher the applicability to a wider range of subjects becomes.

In this case, I believe it was shown that an index generalized by the "ease of a substance escaping into the air" can be specialized for analysis conditions; in other words, VOC analysis is one of the things influenced by the general phenomenon of "ease of a substance escaping into the air."

Through analysis, I would like to aim for a solution by discovering the general laws of the natural worldmost appropriate for the problem and implementing specific measures based on them.

This turned into a long message, but thank you for bearing with me!