Over the last year, we produced quite some pyrolysis oil. Its appearance can vary quite a bit. Some runs produced a white and waxy substance while other runs produced a honey-colored liquid. The appearance is related to the chemical nature of the product. A product is more waxy, when it contains more molecules that have a long chain-length and are larger and thus less flexible or fluid-like. We verified this with gas chromatography coupled to mass spectrometry (GC-MS). This analysis can then be related to the temperature profiles of these runs, which we measured with our thermocouple.
The decisive part of a gas chromatograph is the chromatography column, which is a very thin and long tube, the inside of which is coated with a ‘sticky’ material. The different molecules adsorb on this material more or less strongly depending mostly on their boiling point in the case of the column that we used. After all molecules are adsorbed, the column is heated slowly and a flow is sent through. This way the small molecules, that have a low boiling point come off first followed by slightly larger molecules and so forth. Each peak in the below graph shows one of those molecules. The later they come out, the higher their ‘retention time’, which is the x-axis of this graph.
The mass spectrometer is used to identify the molecules that we make. The molecules are fragmented into smaller pieces, the mass of which is measured. Every molecule fractures in a very unique way forming a specific collection of masses with specific amounts. This ‘fingerprint’ can be identified using a library with all the fingerprints that researchers have measured over the years. The chemical structures of these molecules are drawn into the above graph as well. However, because polypropylene has a lot of little branches, it forms a lot of different and very complicated molecules, a lot of which were not found in our library. With a calibration standard, we could however determine how big the molecules are that are formed. As indicated in black above, molecules with up to 27 carbons are found.
Now, let’s compare the different runs:
Mostly the same types of peaks are found, but there are some differences in their relative distribution. To visualize this better, we integrated the areas of the peaks to get an idea of their relative amounts. The whole chromatogram is normalized by the total area of all peaks. The chromatogram is categorized into different regions corresponding to molecules consisting of a certain number of carbons. This is just a rough division. For completely linear hydrocarbons, the boiling point or retention time does directly correspond to their size. For the more complicated molecules that are formed from the pyrolysis of polypropylene, this is just an approximation. The contribution of each carbon number to the product mix is plotted below.
Comparing this with the temperature profiles of the runs plotted below (colors are the same in the two graphs), it seems that the heating rate is the factor that influences the product distribution most. A faster heating rate (purple, blue, orange, green) lead to a higher content of larger molecules. At the same time a slower heating rate (red and brown) causes the formation of more medium range molecules. This could also be related to the residence time that we discussed in an earlier post. A very fast heating rate (>35 C/min) might cause the formation of a lot of gas that expands very fast and propels less volatile compounds out of the reactor, which would be volatile enough to leave the reactor under normal circumstances. This is also supported by the observation that the amount of gas formed increases at higher heating rates. We do not currently capture these non-condensable products, but just measure their amount by the size and vigor of the flame, when burning it.
Overall, the highest variation in products is observed in the C9 and in the C22-27 region. Since the type of molecules stay the same from run to run, apart from controlling the heating rate, a distillation step after our reactor could ensure a more stable product distribution.