Image for post
Image for post
Credit: IBM

Quantum Computing and AI to Enable Our Sustainable Future

By Katia Moskvitch

Next-generation plants to efficiently capture carbon dioxide from our overburdened atmosphere, and to store it safely, mitigating climate change. Modelling nature to feed our surging population, without a carbon footprint. Rethinking batteries and gadgets before we have to rethink the world. Learning from our past for a healthier future — with sustainable materials, sustainable products, sustainable planet.

We are not there yet, but it’s not just wishful thinking. Here at IBM Research, we believe that within the next half a decade, we’ll be able to accelerate the process of discovery of new materials — to tackle many, if not most, global challenges facing our world.

For that, we first need to change the current centuries-old slow and inefficient scientific method. It follows a linear approach to science — you start with a question, develop a hypothesis, create a model, test it experimentally, and get the result. If the result fails, you start all over again.

Instead, the Scientific Method 2.0 should be a closed loop where computers upgrade humans. This way, a question is followed by artificial intelligence (AI) combing through all our past knowledge about a specific problem, leading to a much more concrete hypothesis. Then we test it, assess it, create a report — and the result, even if it’s a failure, is likely to lead to a new question, continuing the loop.

This new approach should turbocharge the way we do material design, a hard and complex process — mainly because the chemical space of potential molecular combinations is incredibly vast. There are more possible combinations than there are atoms in the universe, and it typically takes about 10 years and $10 million to create one new material.

We want to cut both years and cost 10 times. The good news is we have the technological means to make it happen. They are the traditional classical computers that are becoming ever mightier in augmenting data. And the emerging quantum computers that will soon propel this data augmentation and simulation to a whole new level. And it’s also AI with its neural networks and their amazing pattern recognition ability, as well as the power to create the so-called generative models — crucial for the prediction of a new material. Finally, it’s also laboratory automation through the open, hybrid cloud.

Working together, these technologies are bound to revolutionize our world. Their convergence will allow us to address the material discovery process in a fundamentally new way. First, AI will consolidate all of humanity’s knowledge on a specific topic — say, a global challenge we want to address. Then traditional and quantum simulations will cover our knowledge gaps: using the past data the AI had obtained, we will create models to generate hypotheses about the new materials necessary to tackle that challenge. Finally, we will automate the making and testing of these materials, with the help of cloud technologies.

To see how we aim to get there, I invite you to join me on a virtual tour of some of the IBM Research’s global labs, where scientists are working tirelessly to make it happen.

Stop one: IBM Research Yorktown. AI, quantum and the quest for antivirals

Nested in a maple tree forest that every fall morphs into a vivid display of every shade of yellow, orange and red, the horseshoe-shaped IBM Research Yorktown is about an hour drive from New York City. And it’s home to some of the most powerful – and promising – quantum computers in the world.

Every time I enter a quantum computing lab, the first thing I notice is an unmistakable hissing, humming sound. I like to call it the sound of the future. The noise is mainly due to the refrigeration equipment essential to cool down the qubits — tiny superconducting circuits made of niobium, silicon, and aluminum that make quantum computers tick — and get them into the quantum realm. For that, we use an open dilution refrigerator, a cryogenic device that operates at a 50 millikelvin (-459F), making it the coldest place in the known Universe, cold enough to make atoms almost completely motionless.

But how can a quantum computer help us with material design? For centuries, material discovery has depended either on serendipity or on a slow and iterative process of mixing different compounds and testing the result. That’s how we came up with Teflon, Velcro, Vaseline and vulcanized rubber. Molecular simulation with the help of computers has given material discovery a boost. Still, discovering and designing materials using the time-consuming traditional method, even assisted by high performance computers, hasn’t been trivial. The bigger and more complex a molecule is, the more difficult computer simulation becomes. The vast design space combined with multiple, time-consuming tests means that creating a new material using this approach is expensive and typically takes years.

A quantum computer should take molecular simulation to a whole new level, accurately and rapidly predicting the outcome of complex chemical reactions. While a classical computer has to sift through all likely outcomes one by one, a quantum machine will go through many possibilities simultaneously, dealing with a much larger range of possible computational states. That’s because it relies on qubits to make computations — the mighty quantum cousins of the ‘bits’ transistors in a standard computer work with.

Qubits behave just like atoms, but in a programmable way. “The behavior of both qubits and molecules is governed by the same quantum laws of nature, so qubits are ideal to simulate the natural behavior of molecules,” Zaira Nazario, a quantum computing researcher at IBM Yorktown, tells me as I make my way around a chiming and whooshing quantum computer. And while quantum computing isn’t yet better than classical computing in molecular simulation, it could reach its full potential over the next several years.

Another vital element of the next-generation material design is artificial intelligence. It’s already been of great help, especially following the explosion of deep learning around 2012. Deep neural networks are great in screening novel high-performance materials and creating models to outline desired properties of unknown ones. To do so, AI sifts through vast materials literature, tapping into the information tucked away in academic papers, articles, books and patents. It enables us to build on the known and then fill data and knowledge gaps with classical and quantum-based simulations and AI-assisted automated experimentation.

Boosting and complementing each other, AI, quantum and high-performance computers (HPC) will pave the way for a supercharged scientific method that should soon become the default way to discover materials.

AI in particular is already positioning medical researchers to more rapidly find efficient antivirals when we face a future pandemic. This is exactly what Wendy Cornell, a chemist here at Yorktown, and her team, have been working on for the past several years. “It can take more than $2 billion and about a decade to get a new drug to market,” she tells me. “And a third of the cost and time has to do with the drug discovery phase.” Drug repurposing allows you to bypass the discovery phase since the drug was already discovered and characterized for its original disease.

“The behavior of both qubits and molecules is governed by the same quantum laws of nature, so qubits are ideal to simulate the natural behavior of molecules.”

The idea is to identify existing approved treatments which are effective (can be repurposed) against new or additional diseases. The repurposing is achieved through a combination of AI and analytics, specifically causal inference, focused on analyzing medical real-world evidence faster than ever before. AI is excellent in spotting patterns in the deidentified medical records of millions of patients — and those patterns can help researchers. “In the future, these tools may reach widespread adoption across the pharmaceutical industry, effectively becoming one of the means of rapidly responding to global, life-threatening viruses,” says Cornell.

The true beauty of the accelerated discovery cycle is that we can apply it to a variety of materials. That’s why material designers probably never get bored — how could they, materials are all around us, just in different phases. And just like Cornell and her colleagues are exploring how AI can be used to repurpose drugs, three other teams at IBM Research lab in Almaden are examining how this new approach can be applied to make greener electronics and more efficient batteries.

To see how they are tackling these two challenges, let’s hop over to sunny California.

Stop two: IBM Research Almaden. Greener gadgets and better batteries

Unlike the Yorktown lab’s horseshoe shape, the bird view of Almaden in San Jose, CA, reminds me of four train carriage-like buildings linked to the main — larger — one. Or perhaps a train station with five train garages.

Let’s walk through the entrance, exchanging masked nods with a friendly receptionist. A few more minutes, and we enter the lab where a Russian émigré, chemist Dmitry Zubarev, is helping design new materials at the heart of our electronics: photoresists.

Semiconductor devices have long been shrinking, with manufacturers cramming ever more transistors on a single chip to give us smaller and more powerful gadgets. We’ve managed to achieve this scaling, in part, thanks to advances in photolithography — a fabrication process where a light-sensitive material, a photoresist, is exposed to ultraviolet light through a mask. Inside the photoresist, chemical compounds called photoacid generators (PAGs) translate this light into a catalyst that defines a pattern that’s then turned into transistors, wires, and other device components.

“We all love our gadgets — so we gotta make sure the semiconductor industry not only survives, but becomes more sustainable.”

IBM was the first company to create and deploy modern photoresists more than three decades ago, and chipmakers have used them ever since. Now, as semiconductor chips are more widely used than ever, we need to ensure that they are made as sustainably as possible and remain sustainable throughout their entire lifecycle. For that, device makers have to discover new compounds that are both efficient and environmentally friendly.

That’s not an easy task, Zubarev tells me, as the possibilities of molecular search are huge. AI, HPC and quantum can help and dramatically accelerate the discovery process for these materials, so critical for the manufacturing of semiconductor computer chips. Zubarev’s team is already using AI to ingest all the information about PAGs and photoresists in patents, academic papers and other literature into data sets, to dramatically increase our knowledge and understanding of what’s been done so far. Another big part of the effort is to capture all the human expertise and creativity — and infuse AI with this critical knowledge.

The next step is to use HPC systems and, in future, quantum computers to accurately simulate possible new molecules and their behaviour and build AI models. These models should help us identify new possible classes of compounds that could serve as PAGs. “And then, our AI-enhanced robotic systems will test the PAGs that we found to be environmentally safe, with minimal human intervention,” says Zubarev. After all, “we all love our gadgets — so we gotta make sure the semiconductor industry not only survives, but becomes more sustainable,” he adds, as he heads to lunch in the lab’s canteen.

Typically buzzing with chatting researchers, the canteen is now empty. In this surreal era of COVID-19, Zubarev himself isn’t in the lab every day. Once in a while, he meets Young-Hye Na, a chemist from another lab here at Almaden. Originally from South Korea, she came to the US in 2003 to do postgraduate research and joined IBM a few years later. Since joining the battery team in 2016, she’s been trying to make our future world a better place — focussing on next-generation batteries to address the surging global electricity demands.

Better batteries are crucial to ensure that our electric cars don’t run out of juice in the middle of nowhere. Better batteries may, one day, give us electric transatlantic passenger jets and perhaps even an electric spacecraft. In the shorter term, they can help wean us off polluting fossil fuels, as instead we’ll rely on batteries to store all (or nearly all) electricity generated by renewable energy plants. After all, “solar panels are useless when the Sun doesn’t shine, and so are wind turbines on a perfectly still day,” laughs Na. At the moment, we can only store about three percent of the energy we produce worldwide.

To keep global warming to below two degrees by 2050, we have to triple the amount of energy we currently store, at the very least. Lithium-ion battery technology is still king when it comes to energy storage, considered one of the most efficient and lightest battery solutions. But heavy metals that are needed to make these batteries — such as cobalt and nickel — are difficult to mine, supplies are dwindling and the batteries are harmful to the environment if they are not disposed of properly.

Still, lithium-ion batteries are gradually getting more efficient, with prototypes sporting relatively low cobalt content. But IBM researchers are also looking at next-generation batteries, betting on artificial intelligence and quantum computing to get us there. Earlier this year, Na’s team developed a new type of cobalt- and nickel-free battery that relies on an iodine-based cathode. The researchers showed that the battery could have higher power density, lower flammability and faster charging time than conventional lithium-ion batteries — for instance, when configured for high power, it reached an 80 percent charge in just five minutes.

The next step is to use AI to search for safer and more efficient materials, in a bid to improve the battery’s performance. The researchers have already partnered with a battery manufacturer for further research and development.

The use of quantum computing will become of critical importance to improve next-generation technologies — such as lithium sulphur batteries that could be more powerful, longer lasting and cheaper than lithium-ion. Next, AI could predict the correct molecular configurations, allowing the researchers to test the best candidates in the lab — and hopefully one day this new battery will make its way to market.

With all these efforts, I’m confident that in the next half a decade, we’ll develop much more efficient energy storage technologies than we have today. We simply have no choice. To tackle climate change effectively, the world must become carbon-neutral by the end of the century — and for that, we need to achieve net zero emissions by 2050.

There are other challenges to overcome though, beyond better batteries and more sustainable gadgets. One of them is rivalling the smallest organism on Earth: bacteria. And with that, I welcome you to picturesque Zurich in Switzerland.

Stop three: IBM Research Europe. Taming bacteria to feed the world

Just half an hour drive from the airport, Zurich is a peculiar mix of traditional ‘swissness’ and the latest tech. Cosy chalets, delicious fondue, luxury ski resorts and plenty of Sauvignon Blanc co-exist seamlessly with a highly developed infrastructure for electric cars. Crowds of techies from the more than 5,000 tech giants and start-ups make this city one of Europe’s major information and communication technology hubs. Since 1956 IBM Research has been producing cutting-edge science from a leafy village outside of Zurich called Rüschlikon.

Facing the crystal-clear Zurich lake and surrounded by fields with myriads of cows jingling their bells, the campus is a collection of eight buildings. Each has a letter as a name, my smiley communications department colleague Chris Sciacca tells me, and the “M” or main building is particularly special. That’s where what Albert Einstein once called ‘spooky action at a distance’ — quantum entanglement — and other quantum weirdness happens. It’s home to quantum computing research.

But this is not what we are here to see today. Instead, let’s take the elevator up to the Z wing of the building, where Italian-born chemist Teodoro Laino and his team have designed an autonomous chemical laboratory, accessible through the cloud and controlled by AI. The idea: to come up with a way that could help researchers design better fertilizers than we have today, rivalling the mightiness of bacteria. After all, by 2050, the world’s population is expected to exceed 10 billion people. Laino and his colleagues aim to create new catalysts to ‘fix’ nitrogen and produce more efficient artificial fertilizers with less environmental impact.

Nitrogen is the most abundant gas in the Earth’s atmosphere. It makes up four fifths of the air we breathe and is the key ingredient of proteins essential to life. But here’s the problem: most living systems, including humans and plants, can only use nitrogen in ‘fixed’ form, when it is turned into ammonia. For that to happen, nitrogen has to be combined with organic compounds that contain hydrogen. Ammonia then gets converted into nitrates that plants use to make proteins for healthy growth.

There are some bacteria on the roots of plants that fix nitrogen — nature’s clever way to make its own fertilizers to feed plants that feed us. Since the 1960s, researchers have been trying to replicate this biological process, but we are not totally there yet. Today’s ammonia plants rely on the Haber-Bosch process that use a very energy intensive iron-based catalyst. Making one metric ton of ammonia requires10 MWh (or 20 Giga Joules) of energy. That’s equivalent to the energy in one ton of fossil fuel, says Laino, accounting for two percent of global energy demand. “A single modern ammonia plant produces more than 750,000 tons of ammonia per year,” he adds. “And some 88 percent of that is used to make fertilizer.”

Over the decades, researchers have been trying to create new, better catalysts to facilitate the reaction between nitrogen and hydrogen and lower the amount of energy needed to sustain it. Once again, AI and quantum computing can help. First, AI would sift through the existing literature about catalysts. In a few years, a quantum computer might be able to precisely simulate different nitrogen fixation catalytic processes, further augmenting our knowledge. Then researchers would use the resulting data to create predictive models and determine new molecules, using only a small amount of energy compared to the today industrial processes. IBM could also help validate those predictions: the candidate materials can be tested in the AI-driven chemical lab and screened to check their effectiveness.

“A single modern ammonia plant produces more than 750,000 tons of ammonia per year.”

The next goal would be to scale the process in a way that meets the world’s agricultural needs. This might be achieved using fuel cells — devices that convert a fuel’s chemical energy into electricity. It would work like a reverse battery — instead of storing energy, fuel cells could use energy from renewable sources to combine nitrogen from the atmosphere and hydrogen from water to produce ammonia. Catalytic molecules would play an important role here, by lowering the amount of energy needed to sustain the nitrogen fixation process.

Food security is vital, and so is the quality of the air we breathe. To see how we are dealing with this global challenge, let’s visit IBM Research Brazil, in Rio de Janeiro.

Stop four: IBM Research Brazil. Atmospheric janitors

Over the past century, we have generated greenhouse gases in the atmosphere, primarily by burning fossil fuels for transportation, power and heat. Carbon dioxide, a byproduct of fossil fuel combustion, is the main culprit contributing to global warming, together with methane, nitrous oxide, and other industrial-process gases. The level of CO2 in the atmosphere is now higher than at any point in the past 800,000 years.

To limit the temperature rise to 1.5C above pre-industrial levels, we’ll have to cut CO2 emissions by 45 percent by 2030, reaching net zero around 2050. That’s tough — and the continued expanding use of fossil-fuel burning power sources jeopardizes those efforts and our chances to meet that target.

We need to capture CO2 at a global scale. With the help of quantum and AI, we should be able to discover better materials for capturing it and converting it into concrete, fuel, or other materials. There are several approaches to CO2 capture. One method relies on fluid mixtures that bind the gas; another technique involves filtering CO2 out of flue gas with plastic membranes. But these membranes are still not efficient enough in separating the gases, so we need better polymers to improve their performance.

That’s where my colleagues from Brazil come in. IBM Research Rio de Janeiro are two magnificent buildings towering over a local sandy beach and a bay full of sailing boats of all sizes. Here, German physicist Mathias Steiner is working with a team of global researchers to create a cloud-based knowledge base of methods and materials for capturing CO2emitted by industrial plants. “In the future, we hope it might be possible to include scaled techniques for capturing CO2from the atmosphere,” Steiner tells me.

The scientists use IBM’s AI technology for annotation and natural language processing to sift through patents and scientific papers. AI then digests the data and presents findings to the researcher, such as a ranking of the best-known materials for CO2 separation. Based on this knowledge, the researchers define desired properties of the best new molecules for CO2 capture and separation processes. Finally, AI predicts the optimal molecule to use as a building block for new, more efficient polymer membranes for CO2 separation.

Once captured, CO2 can be put to use. Back in Almaden, another IBM team is designing a sustainable materials development platform to harness CO2 as a feedstock or raw material for monomers and polymers such as plastic.

These are promising first steps, however, there is still a long way to go. Once we are able to better capture CO2 and even put it to use, we might — just might — reduce its amount in the atmosphere and slow down climate change.

As the sun sets in Rio, it’s time to bring our virtual tour to a close. We’ve seen just four out of IBM Research’s 19 international locations and peeked at the work of scientists tackling five of the most pressing global challenges.

There are many more challenges that we will have to solve — for our kids to enjoy a world not just jam-packed with cutting-edge gadgets but also one that is clean, safe and abundant. Never before have we needed science as much as we need it today.

Written by

This is the official Medium account of IBM Research. It’s managed by IBM Research’s Chief Writer Katia Moskvitch & follows the IBM Social Computing Guidelines.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store