Analysing microplastics samples is a long and meticulous work, a bit like that of a coroner, yet less gruesome – even if results sometimes also give you the creeps.
Once sieves, binocular loupes, tweezers and scales have been handled to separate, identify, count and weigh all plastics particles present in a sample, remains a « quality control » with a representative batch as a final step. Indeed, we have to make sure of the types of polymers present in all analysed microplastics.
Surely you have already watched a crime drama where the lab guy, wearing a white coat 99% of the time (scoop: 99% of scientists don’t wear one in real life), throws in, in an obvious and erudite way for the average viewer, “this sample has to be identified with a mass spectrometer”. Here we are. It is a bit similar for Oceaneye, except that we use a FTIR spectrometer (stands for Fourier Transform InfraRed) and not a mass spectrometer, the former being generously provided by the Chimiscope, our University of Geneva partner. We had addressed this topic in a previous news, let’s go further yet in a simpler way.
How does it work?
The equipment needed to carry out an identification consists of the actual spectrometer, a tablet, powerful lighting thanks to a rostrum camera, an analysis sheet, alcoholic solution, a binder with reference samples and their chemical formula, and, of course, a computer. The whole process is actually simpler than it seems.
His name is Cary. Not Grant, but Agilent Cary 630. A little gem.
So-called compostable bags are featured in the reference database since they belong to a bioplastics family.
The particles to be analysed are individually placed on the squared sheet, then photographed with the tablet connected to the computer and saved in the database.
The tablet must be perfectly positioned before starting the analysis process.
No bingo grid, no quine, and a full house is more of a bad sign than a nice prize…
A single particle is then placed on the surface of the spectrometer, cleaned beforehand with an alcoholic solution to avoid any potential contamination by the previous sample, and wedged under the press until a “click” indicates that the particle is correctly in place. The flatter and smoother the contact surface between the sensor and the sample is, the more accurate the analysis will be. Indeed, the sensor sends an infrared spectrum and “reads” the response sent by the polymer; each plastic polymer having a very specific spectrum – a bit like DNA or finger prints – thus allowing its identification. As a result of their age, prolonged immersion in seawater or exposure to sun at the time they were collected, some samples can be very brittle and disintegrate under the pressure of the sensor. However, they remain exploitable in most cases.
After press: with the “click”, came the “scrouitch”. This sample was unfortunately particularly fragile.
The Agilent MicroLab software first checks that the sample is correctly aligned with the sensor. If the blue bar reaches the green area, the particle is ready for the analysis.
Green light: next step.
The software analyses the particle and compares it with over a thousand references – called signatures – in three databases in order to find a match. It only takes a few seconds before a list of matches appears in order of probability. In the below example, it is an almost perfect match: there is an 82% chance that this particle is polyethylene. Only results with a rate of verified polymers superior to 72% are considered reliable.
In blue is the reference curve of the existing sample in the database, to which is compared that of the analysed particle, in red.
However, particles occasionally puzzle us when curves don’t match the database, partly or totally. In the following example, there is a 74% probability of natural latex rubber against a 72% for polyethylene – which does not facilitate our task. That would be too easy. Several explanations are possible.
A little challenge for our lab guys’ neurons and analytical skills.
A first trail could be that the contact surface between the sensor and the particle is neither smooth nor big enough, thus confusing the spectrometer’s effectively reduced reading field. So the analysis must be done all over again.
We can also have a close look at the elastic properties of the sample: is it flexible like rubber or rigid like polyethylene? This approach is nevertheless not easily practicable on a few millimetres’ small sample since it could damage it.
Another possibility can be found in the actual composition of the sample: it could be an aggregate right from the production stage – many items are composed of several types of material – or a fusion – that is to say, when different particles, whether of plastics or not, aggregated in an environment where they ended up. The signatures will logically be multiple.
A more powerful equipment than the ATR FTIR would be needed to carry out further analysis, should a correspondence be of only 50% or 60%. Which, unfortunately, the Chimiscope does not have.
At the end of the day, we proceed by elimination and evaluation of probabilities. If a serious doubt still remains, the sample is labelled as “unidentified” and put aside for a while, then analysed again when the software database is richer. But it needs to be in a workable state.
Fair enough. So what?
Why is it so important since it is a matter of plastics anyway? Oceaneye provides its data to large-scale international bodies for policy and/or economical decision-making, just like the European Commission banned a certain type of single-use plastics on the basis of this data. Hence is it crucial for us to guarantee their total quality and reliability because this has made, and still makes today, our reputation to these institutions. To this day, our polymers identification result is 89% and we aspire to improvement.
A bit like a coroner who would be well advised not to be mistaken when declaring the cause of a death…