Are new reactive molecules present in flames?

tl;dr

Which reactive molecules lead to the formation of soot? Recently we showed that adding hydrogen to the edges of soot molecules makes the edges reactive. In this paper, we showed that these sites are very common in flames, making them likely to be important for soot formation.

In my previous blog post, I talked about our systematic comparison of possible crosslinks between reactive aromatic molecules that had recently been detected using atomic force microscopy. This gave the bond energies and allowed us to work out which bonds could be stable at flame temperature. It showed that adding hydrogen to a pentagonal ring gives a reactive localised π-radical that allows crosslinking and stacking.  

In our recent paper that just got accepted in the Proceedings of the Combustion Institute, "Reactive localised π-radicals on rim-based pentagonal rings: properties and concentration in flames" (see preprint here), we showed that these reactive sites are present in the flame in significant concentrations. 

To show what we mean by localisation the electron's spin density is shown below. The spin density shows where the reactive electron is likely to be able to form a bond with higher values indicating higher reactivity. We find two classes of π-radicals, those that 1) delocalise in 6-membered ring aromatic molecules and 2) reactive localised π-radical for pentagonal rings or methylene (–CH2) that does not delocalise across the molecule. 

While these reactive sites have been seen in molecules sampled from the flame it was not clear whether they also exist in the flame. For example, sampling these molecules from the flame could lead to hydrogen being added while the molecules in the flame could actually be lacking this hydrogen i.e. it could be an artifact of sampling. 

In order to determine if these species are present in the flame, we calculated all of the reactions that could allow hydrogen to be added or removed (thanks to Angiras and Dingyu for this). We could then consider, given the concentrations of hydrogen species in the flame, what sort of concentration we would expect. The reactions are shown below (those barrierless reactions do not make computing the rates very easy but it can be done with sufficient approximations).

We found that between 1-10% of the molecules contained a localised π-radical on their rim-based pentagon (for between 1400-1500K which are temperatures within a flame where soot begins to form). Comparing these results with the HR-AFM structures recently imaged we found a consistent frequency of rim-based pentagonal sites with a ratio of 27:12:4 for the unsaturated, saturated and partially saturated rim-based pentagonal rings, showing that these species are present in the flame in significant concentrations.

We explored another exciting possibility - that multiple localised π-radicals are present on a single molecule. These species are also likely to be present in reasonable fractions (thanks to Gustavo and Angiras for developing the KMC simulations). Of the molecules that were recently imaged using HR-AFM, over half contained one rim-based pentagon and roughly a quarter had two rim-based pentagons suggesting that the formation of multiradicals in flames is likely.

These results suggest a new mechanism for soot formation where molecules with two or more localised π-radicals can polymerise (a rapid chain reaction) - what we called the aromatic rim-linked hydrocarbon mechanism (ARLH).

There is a nice historical connection with New Zealander Prof. John Abrahamson, Canterbury University. During a sabbatical in the '70s at the Chemical Engineering department at the University of Cambridge (where I did my PhD) he wrote a paper proposing that the partial saturation of aromatic platelets forms soot (see structure below). I spoke with him recently during the lockdown in New Zealand about the HR-AFM results that show partial saturation of aromatic platelets and our results showing the localised π-radicals and he was happy to hear about the recent insights and how close he got in 1977.

Abrahamson, John. "Saturated platelets are new intermediates in hydrocarbon pyrolysis and carbon formation." Nature 266.24 (1977): 323-327.

Finding the links between reactive molecules involved in soot formation

tl:dr
We still don't know how soot forms and this is stopping us from eliminating it from internal combustion engines and furnaces. Recently, the molecules present, just before soot formation, were directly imaged. For the first time, many of the reactive edges could be seen. In this work, we computationally screened these reactive edges. We then considered all possible crosslinks between these edges. We discovered a new crosslink that allows the molecules to be stabilised by physically stacking on top of each other and then becoming bonded at their rim. This could help explain the rapid growth of soot particles in the flames and lead to new ways to clean up combusiton.

We have just published a new paper in the Journal of Physical Chemistry C. Here is the infographic/abstract figure.

Reactive molecules involved in soot formation

At the 37^th International Symposium on Combustion, an extraordinary paper was presented directly imaging the molecules present just prior to soot formation. In the case of most of these aromatic molecules it is the edge that is the most reactive and over the years many suggestions have been made but never directly observed. So here they are.

Here are some of the most exciting findings. 

Firstly, some were found to be crosslinked suggesting reactions between radicals and molecules during soot formation. This contradicted a commonly held view that only physical interactions and not chemical reactions were involved. 

Secondly, there were lots and lots of pentagonal rings. Out of the 49 molecules (above 4 rings) imaged 28 contained at least one pentagonal ring and 12 contained two pentagonal rings on their rim. Previously only six-membered rings were thought to be stable at flame temperature.

Thirdly, species very close to curvature integration were found. While curved 3D were unable to be imaged using this technique at present, the presence of the almost curved molecules was encouraging for our suggestion of curved aromatic molecules being important in soot formation as I have previously discussed in this blog. 

Finally, some of these pentagonal rings were found to have hydrogen added to them. This forms a completely new radical type (–CH=CH– + H → –C^•HCH₂–  which we found formed a localised π-radical).

Given the wide range of interesting new molecules that were found we considered how their reactivities compared.

We made use of computational chemistry to compute the energy needed to remove an electron from a particular spot on the "surface" of the molecular surface (average local ionisation energy). This told us how likely it was to form a bond with another molecule and therefore allowed us to compare their reactivities.

One significant surprise was the reactivity of pentagonal rings and a new localised π-radical on pentagonal rings B).

Many reactions are important in the flame

Now that the reactive sites were characterised we considered which crosslinks between them could be important in the flame. Below is a figure of the crosslink energies. The green indicates bonds that are strong enough to persist at the high temperatures within a flame.

Most crosslinks are well-known mechanisms, however, the reactions with the localised π-radicals B) were completely novel.

A new type of bonding is possible - rim-bonding

Most of the ideas for how the molecules in flames come together to form soot particles have been either stacked physically interacting interactions or chemical bonds in a long polymer that did not stack. However, the localised π-radicals B) allows for stacked and bonded structures that are strongly bound.

This could allow molecules to rapidly condense and then crosslink which could explain the rapid growth of soot. Below is a drawing of how such a cluster could form we are calling an aromatic rim-linked hydrocarbon.

We need to figure out the concentration of this reactive site in the flame. We also need to compare how all of the possible crosslinks contribute to soot formation. Once this is achieved we can consider how to stop particular reactive sites from being made and reduce soot emissions.