Sharing what went wrong helps fellow researchers

Missed target

Source: © Richard Drury/Getty Images

Most reactions don’t hit the target first time

Before I started my first research project, I was told the brutal truth that most new experiments do not lead to desired results; four pages of a research article might result from several years of hard work. We can precisely plan a synthetic pathway based on similar examples we’ve read, only to find that a seemingly innocent atom we’ve added has drastically changed the behaviour of our molecule, sending our work straight into the metaphorical rubbish bin. Since we spend so much time performing experiments that eventually fail, why don’t we include them in publications?

Recently, I was part of a team that published a paper in which we demonstrate an efficient synthesis of a new porphyrin tape–cycloparaphenylene hybrid.1 At first, this was supposed to be a short side project. Reactions potentially leading to the desired product were already well established and have been described to work well on a variety of building blocks. Nevertheless, things do not always go as smoothly as they appear on paper. In fact, the small project transformed into a long period of struggling with the forces of nature.

My first problem appeared when forming cyclic structures from a smaller porphyrin dimer building block. They were formed, but as we found out later, the presence of acid coming from an oxidising agent resulted in an unwanted rearrangement, modifying their structures. Decreasing the temperature of the reaction to –40°C supressed this side reactivity.

Negative results can be a great source of knowledge

The most challenging aspect was reductive aromatisation of cyclohexadiene to benzene within the structures. Although many conditions for this type of transformation have been reported, it was difficult to perform it on the nanorings, even though the reaction worked well on a linear model substrate in commonly used conditions involving sodium naphthalenide, which indicated that this particular reagent is compatible with porphyrins. Moreover, all attempted acidic conditions led to an undesired rearrangement. Finally, after a few months of struggle, I formed the desired product by using lithium metal as a reducing agent. However, there was one issue: I could not reproduce this reaction. Frustration was running high, but eventually after another month of excluding every possibility of contamination with oxygen or water, I realised that the one time when the reaction had worked, I used a PTFE stirrer bar whose surface appeared black. Its outer coating had been destroyed by a previous reaction with sodium naphthalenide, and so the solution was not in contact with PTFE. Apparently, the PTFE reacts with alkali metals and gets defluorinated, generating radicals that could interfere with the reaction pathway. Armed with this knowledge, I used a glass stir bar instead; this made the reaction reproducible.

When we published our synthesis, we decided to include observations from our unsuccessful synthetic attempts and describe the unwanted behaviour of the compounds in the supplementary file. We wanted to give other researchers who work with similar systems full information about the frustrations they might encounter, rather than leaving only small juicy bits that lead perfectly to our final molecules but might not work well for similar systems. Apparently, referees of our paper loved that and their reports were filled with compliments.

In my opinion, sharing negative results and observations that normally remain secret behind-the-scenes facts is beneficial to science. Omitting unsuccessful attempts might produce a bias towards using a particular chemical transformation – for example, people might prefer a particular method because they noticed that it works on three substrates, but nobody mentioned that it doesn’t work on 20 others. We also save our fellow researchers precious time and grant money by telling them about what have we tried along the way. Similarly, whenever an ‘exotic’ and enormously expensive reagent is used, noting whether a more conventional one was first found to be ineffective could help to justify the cost.

Opponents of this view might argue that negative results are never as solid as positive ones – a chemical reaction could have failed due to the accidental presence of oxygen or water, non-optimal laboratory temperature, intense light, human mistake, or just the wrong phase of the moon. However, negative results can be a great source of knowledge when properly described and supported by data, especially when we can suggest an explanation for any observed side reactivity.

I encourage everyone to share their negative results whenever possible. This will let researchers focus on new projects rather than doing things that someone knows will fail, but has recorded this only in a notebook lost in the depths of a lab drawer.