In the mid-2010s, a series of large agriscience companies introduced a range of technologies to help tackle counterfeit agricultural products. Among them was Supply Chain Track and Trace (SCTT) and Innovative Product Label Technology, both from BASF, together with DuPont’s Izon® holographic label and Bayer CropScience’s mobile app.
Although the technologies varied from supplier to supplier, the approaches were all at the macro-level, with track and trace solutions linked to codes carried by a protected substrate.
Now, researchers at MIT (Massachusetts Institute of Technology) are taking crop protection down a level, to the scale of the individual seed.
According to a recent article in MIT News, average crop yields in Africa are consistently far below those expected, and one significant reason for this is the prevalence of counterfeit seeds, whose germination rates are far lower than those of genuine seeds.
In reference to the earlier work of the agriscience companies, the article acknowledges that there have been many attempts to prevent this type of counterfeiting through label tracking, but such labels have been vulnerable to hacking because of the deterministic nature of their encoding systems.
The article goes on to say that a team of MIT researchers has come up with a kind of tiny, biodegradable tag that can be applied directly to the seeds themselves, and that provides a unique, randomly created code that cannot be duplicated.
The new system, which uses minuscule dots of silk-based material, each containing a unique combination of different chemical signatures, is described in the journal Science Advances, in a paper by MIT’s Dean of Engineering Anantha Chandrakasan, Professor of Civil and Environmental Engineering Benedetto Marelli, postdoc Hui Sun, and graduate student Saurav Maji.
Marelli explains that a key to the new system is creating a randomly produced physical object whose exact composition is virtually impossible to duplicate. The labels they create ‘leverage randomness and uncertainty in the process of application, to generate unique signature features that can be read, and that cannot be replicated,’ he says.
The idea of an ‘unclonable’ code was originally developed as a way of protecting the authenticity of computer chips, explains Chandrakasan, who is the Vannevar Bush Professor of Electrical Engineering and Computer Science.
‘In integrated circuits, individual transistors have slightly different properties, coined device variations,’ he explains, ‘and you could then combine that variability with higher level circuits to create a unique ID for the device. And once you have that, then you can use that unique ID as a part of a security protocol.’ ‘Something like transistor variability is hard to replicate from device to device, so that’s what gives it its uniqueness, versus storing a particular fixed ID’.
The concept is based on what are known as physically unclonable functions, or PUFs (see ABN May 2022).
The team decided to try to apply the PUF principle to the problem of fake seeds, and the use of silk proteins was a natural choice because the material is not only harmless to the environment but also classified by the Food and Drug Administration in the ‘generally recognised as safe’ category, so it requires no special approval for use on food products.
‘You can coat it on top of seeds,’ Maji said, ‘and if you synthesise silk in a certain way, it will also have natural random variations. So that’s the idea, that every seed or every bag could have a unique signature.’
The challenge that the team came up against was to devise a form factor for the silk which could be fabricated easily. They developed a simple drop-casting approach that produces tags that are less than 2.5mm in diameter. The second challenge was to develop a method to read the uniqueness, in a high throughput and in an easy way.
For the unique silk-based codes, Marelli eventually found a way to add a colour to the microparticles so that they assemble in random structures. The resulting unique patterns can be read out not only by a spectrograph or a portable microscope, but even by an ordinary smartphone camera with a macro-lens. This image can be processed locally to generate the PUF code and then sent to the cloud and compared with a secure database to ensure the authenticity of the product.
The number of possible permutations that could result from the way the researchers mix four basic types of coloured silk nanoparticles is vast. ‘We were able to show that with a minimal amount of silk, we could generate 128 random bits of security,’ Maji said. ‘So, this gives rise to 2 to the power 128 possible combinations, which is extremely difficult to crack given the computational capabilities of the state-of-the-art computing systems.’
Some additional work will be needed to make this a practical commercial product, Chandrakasan said. ‘There will have to be a development for at-scale reading via smartphones. But the principle now shows a clear path to the day when a farmer could at least, maybe not every seed, but could maybe take some random seeds in a particular batch and verify them.’
