Now that we’ve looked at the things to consider when preparing for a pilot, we can move on to actually conducting some preliminary tests with our chosen compounds! Here we describe our typical approach to a pilot study. If you missed the first part of this blog, you can read it here.
Our approach to pilot studies is to first measure the compounds in headspace. This way we see what kinds of dilutions and system parameters are needed in order for us to clearly detect the wanted molecules. Over time, we have come up with quite universal starting parameters for the setups, which can then be modified depending on the initial results.
The detection of molecules is dependent on the moisture of the sample and circulating gasses, with lower moisture making the system more sensitive. Usually, we start the measurements in ambient humidity and temperature. In a basic measurement setup (see on the right), we attach a filter unit and a headspace chamber to the inlet of the IonVision. The IonVision has pumps to take in sample air, so no pressurized air is needed.
In the figure above, you can see a typical pilot study setup for IonVision. We have the headspace chamber (1), the filter unit consisting of activated carbon and molecular sieve (2), differently sized petri dishes (3), and a permeation tube (4). The touchscreen on the IonVision combined with a screw-on lid on the headspace chamber make this setup really convenient in iterating concentrations!
Several strategies for diluting the sample gas entering IonVision exist if need be. If the sample is soluble in water, we can proceed with diluting it directly, which also adds to the sample gas moisture. By adjusting the size of the petri dish and the sample dilution, we gain quite good control over the DMS response. Using a dish with a smaller diameter dilutes the sample by restricting the evaporation area in the case of liquid samples.
If the sample is not soluble in water, we can adjust the rate of sample intake and the circulating clean air inside the IonVision. Pressurized air can also be used to increase the airflow through the headspace chamber. The sample is diluted by only passing some of this air to the IonVision and rest vented out.
If/when we want to move onto determining the detection range for the chosen chemicals, a good option is to make a permeation tube. While getting the permeation rate to stabilize in a gas calibrator can take several weeks, the ready tube can give a great estimate of the detection limit in a controlled environment. This makes moving onto more complex background matrices a breeze as well. For example, moisture, unpurified room air or a cocktail of other compounds can be mixed with the sample air from the gas calibrator while keeping a known concentration of the wanted chemical.
The approach with our six fatty acids was quite a straightforward one. We utilized the basic setup described above, and diluted the acids to concentrations in the high ppm range to get non-saturated responses from each sample. We were happy to note that our chosen 6 were all distinguishable from each other with plain eye just from the peak locations.
A small classification task was prepared where three concentrations of each fatty acid were prepared and measured. A simple classification algorithm, linear discriminant analysis, was able tell apart the samples and their concentrations with great accuracy.
Forward feature selection is a technique where features from the spectra are chosen until adding new features no longer improves the classification accuracy. It was used to see which features are the most important in identifying the acids and concentrations from each other. This technique can be applied to limit the number of needed measurement points in the DMS spectra. Of course, adding a number of measurement points close to the chosen features make the system more robust to small changes in humidity, concentrations etc. This approach can allow us much quicker analysis as time is not wasted measuring areas known to only contain noise.
Our interest and previously done groundwork around the VFAs allowed us join the exciting SynbioPro project! Our role in the project is to aim at on-line monitoring of fermentation processes. In the project, volatile end products of the process will be identified, and we’ll then develop ways to follow these products to get information on the process health.
If successful, our technology could become a competitor for the widely-used high-performance liquid chromatography and the combinations of liquid and gas chromatography used with mass spectrometry, allowing on-line measurements.
Follow us here and on our LinkedIn to get updates as the SynbioPro shifts into gear, or contact us about your potential pilot case!