Terry Knowles, European Correspondent08.02.21
Since just before the turn of the century, the pharmaceuticals sector has employed automated, high-throughput systems for screening chemical structures as a way of zeroing in on drugs with special potential and greater efficacy, thereby making savings in experimental costs and reducing time to market. It’s an understandable premise: the identification of potential drugs is a needle in a haystack at best, given the galaxy of chemical structures available.
For the paint and coatings sector and for those companies that supply it, the possibilities soon become apparently infinite too. A selection of say, 12 resins, 12 pigments and the same amounts of single additives and solvents or solvent combinations – regardless of variations in proportions and interactions – soon multiplies up into a matrix of more than 20,000 possible outcomes that might need analyzing. The choices in research and chemical selection become bewildering, if not impossible, prompting more raw materials companies to consider employing similar high-throughput screening technology for better identification of performance potential at the experimental level. In addition to the necessity for automation in this kind of experimental design comes the need for a computational-based analysis of all of the performance results that ensue from such large-scale experiments.
Data in: high-throughput screening and artificial intelligence
Although automation has been around for a long time, these more modern techniques have heralded a new dawn in automation-enhanced chemical R&D, and in its wake, a new level of digital analysis too. A few chemical companies in Europe have already been investing in these sorts of areas, targeting much closer understanding of their clients’ needs, more rapid identification of promising chemicals and, ultimately, greater market shares. Over the last 25 years, we have witnessed what looks like a coatings superleague with the top companies garnering market share through acquisition. In the raw materials sector, massive investments in technology and an understanding of the need for digital advances and artificial intelligence will project major chemical companies into a kind of superleague too. Some companies in Europe that have been working in these areas have been allnex (at the time Nuplex Resins, in co-operation with Van Loon Chemical Innovations in The Netherlands), Evonik, and just recently, BYK Chemie. Whether smaller specialty chemical companies will enter into similar developmental partnerships remains to be seen.
With a digital lab that is one of a kind worldwide, BYK Chemie recently embarked upon a new era of applications technology and so is giving a whole new dimension to its range of tailor-made additive solutions for its customers. A new fully automated high throughput screening (HTS) system allows BYK chemists to pinpoint more closely which of a vast number of products under testing are the two or three that will be of the greatest use and advantage to the customer.
The new system can perform series tests at impressive speeds and with remarkable efficiency. The company claims that 220 samples can be produced within a day, which is the potential equivalent of 80,000 samples annually. A fully digitalized process enables the time required for series testing to be slashed from months to just a few days. The time-saving this entails not only generates greater scope for the creation of new differentiating customer solutions, but also doubles BYK’s capacity in application technology.
The ultra-modern facility that BYK has announced is complex and includes 32 modules with 27 different functionalities all set to check and test BYK additives in paints, plastics and adhesives. The modules are connected to each other via a rail system, with shuttles moving the samples individually from one module to the next.
The examination of liquid samples with respect to pH value, foam or particle size distribution forms part of this seemingly unique HTS facility. In addition, two rheology measuring instruments evaluate the flow behavior.
Depending on the usage concerned, the samples are then applied by pouring out, drawing down or spraying, and cured using UV radiation, or by means of oven or air drying. Finally, they are then subjected to a fully automated test for property characteristics such as gloss, flow and color.
Another recent development coming from Europe has been Evonik’s announcement that it is extending an agreement with IBM to become the first chemical company to participate in the MIT-IBM Watson Artificial Intelligence Lab at the Massachusetts Institute of Technology. Evonik has already been cooperating with IBM for the last eight years; the companies have been developing ways to create a collaborative intelligence between humans and machines through pilot projects. Experts no longer have to search for valuable information like for a needle in a haystack. Instead, decades of knowledge are now easy to search and to analyze intuitively and so new connections between data can now be found. Chemists can intuitively explore and compare formulations and their related properties to come up with ideas for new products.
Evonik and IBM have also succeeded in building artificial intelligence that will further accelerate the research for new materials. In close collaboration with IBM Research, a deep neural network was developed and trained that predicts properties or new formulations for high-performance polymers. These predictions help researchers by indicating promising directions for new products and customer solutions.
Data out: the rise of digital assistants
Staying with Evonik, which is clearly a protagonist in the digital-chemical interface, last year its launch of the COATINO digital assistant brought a new dimension to R&D. This is the outcome of Evonik’s own high-throughput experiment unit held under the company’s coatings additives umbrella. The unit, which is housed at Evonik’s Goldschmidtstraße location in Essen, doses raw materials, formulates them into coatings, and characterizes the finished coatings. All of this runs fully automatically according to a precisely defined program that can be reproduced at any time. As a result, the unit can formulate an average of 120 samples in 24 hours.
Then came the task for the researchers to start feeding all of the results into a huge database, which was followed by the even bigger challenge of allowing it to be interrogated by voice, where a whole new set of linguistic semantics needed to be addressed from speaking to it in languages as far apart as German and Chinese. More precisely, the key point is that customers have very definite ideas about the final properties that a coating should possess and that the assistant should recognize key words. The end result of all of the testing and trials is a voice-activated assistant that not only offers lists of suitable additives but can also rank them in an order of expected suitability. Meanwhile, back in the land of humans, it offers the opportunity to order samples, view technical data sheets and to establish direct contact with Evonik. All of this is a high-tech way of ensuring customer engagement and priority positioning within the laboratory and an ultra-modern way of advancing business in the 21st century.
More recently, another digital assistant has been born online, this time in the resin sector and from allnex. Termed a ‘digital employee’, Nexi operates on the part of the company’s website and is connected to a product finder. Nexi works by questioning the user about the substrate and the properties desired in the final coating. Perhaps unsurprisingly, this also leads to enhanced customer engagement at the human level when problems become insoluble by the online assistant.
As we stand on the edge of greater digitalization and automation in different parts of the coatings chain, it is clear that companies the world over will be sitting on years, if not decades, of experimental and performance data that could ultimately be converted into databases for interrogation and much wider use. How they do that and harness it in ways that are user-friendly, language friendly and laboratory friendly will surely create a whole new set of challenges – and possibly industry ventures – where chemistry and computing will become evermore closely connected in the future.
For the paint and coatings sector and for those companies that supply it, the possibilities soon become apparently infinite too. A selection of say, 12 resins, 12 pigments and the same amounts of single additives and solvents or solvent combinations – regardless of variations in proportions and interactions – soon multiplies up into a matrix of more than 20,000 possible outcomes that might need analyzing. The choices in research and chemical selection become bewildering, if not impossible, prompting more raw materials companies to consider employing similar high-throughput screening technology for better identification of performance potential at the experimental level. In addition to the necessity for automation in this kind of experimental design comes the need for a computational-based analysis of all of the performance results that ensue from such large-scale experiments.
Data in: high-throughput screening and artificial intelligence
Although automation has been around for a long time, these more modern techniques have heralded a new dawn in automation-enhanced chemical R&D, and in its wake, a new level of digital analysis too. A few chemical companies in Europe have already been investing in these sorts of areas, targeting much closer understanding of their clients’ needs, more rapid identification of promising chemicals and, ultimately, greater market shares. Over the last 25 years, we have witnessed what looks like a coatings superleague with the top companies garnering market share through acquisition. In the raw materials sector, massive investments in technology and an understanding of the need for digital advances and artificial intelligence will project major chemical companies into a kind of superleague too. Some companies in Europe that have been working in these areas have been allnex (at the time Nuplex Resins, in co-operation with Van Loon Chemical Innovations in The Netherlands), Evonik, and just recently, BYK Chemie. Whether smaller specialty chemical companies will enter into similar developmental partnerships remains to be seen.
With a digital lab that is one of a kind worldwide, BYK Chemie recently embarked upon a new era of applications technology and so is giving a whole new dimension to its range of tailor-made additive solutions for its customers. A new fully automated high throughput screening (HTS) system allows BYK chemists to pinpoint more closely which of a vast number of products under testing are the two or three that will be of the greatest use and advantage to the customer.
The new system can perform series tests at impressive speeds and with remarkable efficiency. The company claims that 220 samples can be produced within a day, which is the potential equivalent of 80,000 samples annually. A fully digitalized process enables the time required for series testing to be slashed from months to just a few days. The time-saving this entails not only generates greater scope for the creation of new differentiating customer solutions, but also doubles BYK’s capacity in application technology.
The ultra-modern facility that BYK has announced is complex and includes 32 modules with 27 different functionalities all set to check and test BYK additives in paints, plastics and adhesives. The modules are connected to each other via a rail system, with shuttles moving the samples individually from one module to the next.
The examination of liquid samples with respect to pH value, foam or particle size distribution forms part of this seemingly unique HTS facility. In addition, two rheology measuring instruments evaluate the flow behavior.
Depending on the usage concerned, the samples are then applied by pouring out, drawing down or spraying, and cured using UV radiation, or by means of oven or air drying. Finally, they are then subjected to a fully automated test for property characteristics such as gloss, flow and color.
Another recent development coming from Europe has been Evonik’s announcement that it is extending an agreement with IBM to become the first chemical company to participate in the MIT-IBM Watson Artificial Intelligence Lab at the Massachusetts Institute of Technology. Evonik has already been cooperating with IBM for the last eight years; the companies have been developing ways to create a collaborative intelligence between humans and machines through pilot projects. Experts no longer have to search for valuable information like for a needle in a haystack. Instead, decades of knowledge are now easy to search and to analyze intuitively and so new connections between data can now be found. Chemists can intuitively explore and compare formulations and their related properties to come up with ideas for new products.
Evonik and IBM have also succeeded in building artificial intelligence that will further accelerate the research for new materials. In close collaboration with IBM Research, a deep neural network was developed and trained that predicts properties or new formulations for high-performance polymers. These predictions help researchers by indicating promising directions for new products and customer solutions.
Data out: the rise of digital assistants
Staying with Evonik, which is clearly a protagonist in the digital-chemical interface, last year its launch of the COATINO digital assistant brought a new dimension to R&D. This is the outcome of Evonik’s own high-throughput experiment unit held under the company’s coatings additives umbrella. The unit, which is housed at Evonik’s Goldschmidtstraße location in Essen, doses raw materials, formulates them into coatings, and characterizes the finished coatings. All of this runs fully automatically according to a precisely defined program that can be reproduced at any time. As a result, the unit can formulate an average of 120 samples in 24 hours.
Then came the task for the researchers to start feeding all of the results into a huge database, which was followed by the even bigger challenge of allowing it to be interrogated by voice, where a whole new set of linguistic semantics needed to be addressed from speaking to it in languages as far apart as German and Chinese. More precisely, the key point is that customers have very definite ideas about the final properties that a coating should possess and that the assistant should recognize key words. The end result of all of the testing and trials is a voice-activated assistant that not only offers lists of suitable additives but can also rank them in an order of expected suitability. Meanwhile, back in the land of humans, it offers the opportunity to order samples, view technical data sheets and to establish direct contact with Evonik. All of this is a high-tech way of ensuring customer engagement and priority positioning within the laboratory and an ultra-modern way of advancing business in the 21st century.
More recently, another digital assistant has been born online, this time in the resin sector and from allnex. Termed a ‘digital employee’, Nexi operates on the part of the company’s website and is connected to a product finder. Nexi works by questioning the user about the substrate and the properties desired in the final coating. Perhaps unsurprisingly, this also leads to enhanced customer engagement at the human level when problems become insoluble by the online assistant.
As we stand on the edge of greater digitalization and automation in different parts of the coatings chain, it is clear that companies the world over will be sitting on years, if not decades, of experimental and performance data that could ultimately be converted into databases for interrogation and much wider use. How they do that and harness it in ways that are user-friendly, language friendly and laboratory friendly will surely create a whole new set of challenges – and possibly industry ventures – where chemistry and computing will become evermore closely connected in the future.