Batteries included
Quantum plus AI will yield sustainable results
Originally published January 5 2024
You might remember that a few posts ago, I wrote about never having been to Denmark. Well, we can cross that trip off my bucket list—in early October 2023, I had the opportunity to spend a few days in Copenhagen meeting with members of IBM’s global quantum computing team. Flying in to Kastrup International Airport, the airplane made its approach over The Sound, a stretch of water that connects the Baltic Sea to the North Sea and separates Denmark from Sweden.
Looking out the window, I was immediately struck by a large array of windmills floating on the water. And, indeed, wind farms are pervasive—according to the official web site of the Danish government, more than 50 per cent of Denmark’s electricity is generated from wind and solar. The taxi that took me from the airport to my hotel, naturally, was electric. As I walked through Copenhagen’s beautiful downtown later that afternoon, I realized that the city’s most pervasive form of transportation (arguably, after bicycles) is EVs. I felt a twinge of guilt as I thought about driving my SUV along Toronto’s Gardiner Expressway, where I often pass our single windmill on the waterfront—which, half the time, isn’t even turning.
Yes, Canada (not to mention North America in general) has some catching up to do when it comes to clean energy and sustainable technology. But geography hasn’t done us any favours—we’re a vast country with a lot of open space, so distances, both for driving and power transmission, are longer and create more problems. An EV’s driving range is limited, of course, by its battery life. Also, sustainable electricity production is obviously not constant. You can’t generate solar power at night, and many days the wind just doesn’t blow—so when you do produce electricity, you want to store it for later use. Both the consumption and the generation of clean power are driving the need for better batteries. And this, as we’ll see, is where both quantum computing and artificial intelligence can make a difference.
Plug and play
We don’t often think about batteries in our daily lives, except maybe when our TV remote control stops working, our smart phone dies in the middle of a call or our car won’t start on a cold day. Batteries are pervasive and we like to think we know how they work, but do we really? I remember a high school chemistry project to make a battery out of simple substances like graphite powder, zinc and a mild acid such as ammonium chloride. It worked, and we were able to turn on a small light bulb for a short period of time. However, I don’t recall precisely why it worked—what chemical reactions took place to generate that electrical current. It turns out, not only is battery chemistry very complex, but batteries also have other drawbacks—they’re heavy, their ability to hold a charge deteriorates over time, they’re difficult to dispose of in an environmentally friendly way and they can even pose a fire hazard.
Rechargeable batteries have been commercially available for decades, and the industry seems to have standardized on lithium-ion as the technology of choice for everything from cell phones to e-bikes to EVs. First introduced in the early 1990s, lithium-ion has since evolved considerably and arguably provides the best value proposition for rechargeable batteries. There’s a pretty good balance between weight, scalability and its ability to hold a charge for an extended time.
And yet, one of the largest drawbacks of EVs is the time they take to recharge—anywhere from half an hour or more at a roadside charging station to overnight at home—compared to mere minutes to fill up a gas tank at the pump. Combine that with a limited driving range and sparsely located charging stations, and it’s not surprising that EVs are not yet pervasive in North America.
Clearly, rechargeable batteries have room for improvement—and production also needs to scale up dramatically. A recent report covering Canada’s role in the global mobility revolution notes that although two per cent of vehicles on Canadian roads are EVs, it’s expected that by 2030 60 per cent of new cars will be electric and that the federal government is calling for this number to reach 100 per cent by 2035. If other countries follow suit, demand for lithium-ion or other mobile battery technologies will skyrocket, and production will need to increase by 500 per cent by 2050. Fortunately for Canada, the report also notes that we rank first in North America and fourth globally in production of lithium, cobalt and other materials needed for battery manufacturing. Nonetheless, priority should also be placed on looking for alternative battery chemicals, to mitigate our dependence on a single resource.
Another consequence of the anticipated increased demand for EVs is its impact on electricity generation and transmission. The aforementioned report estimates that, in Canada alone, an additional 50,000 charging stations will need to be installed in our cities and along our highways in the next five years. Moreover, all this charging of EVs will contribute roughly a 25 per cent increase in Canadian electricity demand by 2050, putting pressure on power utilities to deliver additional generating capacity. It’s safe to say that there will not be any political will to use fossil-fuel-based generators, and nuclear may also be politically, if not technically, questionable—so this capacity will very likely have to come from renewable sources.
And speaking of renewable power generation, well, as I’ve noted, it’s variable. When the wind blows or the sun shines, it can contribute to the power grid and reduce in real time the need for fossil fuel or nuclear sources of electricity. But when renewable electricity wanes, other sources must be increased. Given that currently just over 13 per cent of the United States’ electricity comes from renewables (just under 19 per cent in Canada), this variability can for now be managed in the grid.
Uphill climb
However, as our reliance on renewable power sources increases, the need for large-scale and long-term storage of electricity becomes obvious. A recent article in Toronto’s Globe and Mail newspaper suggests that using energy storage to better match supply and demand in the grid will significantly reduce its carbon footprint—250 megawatts of storage could remove anywhere from 2.2 to 4.1 million tonnes of emissions. But Canada’s power grid currently has only about one gigawatt of total storage capacity, a number that will have to increase tenfold to reach our national goal of a net-zero grid by 2035.
Intermediate-scale rechargeable batteries that can connect to residential rooftop solar panels, storing power for overnight use are already available—a good start. At industrial scale, large-scale lithium-ion batteries can be effective but often only hold their charge for a few hours. A promising innovation is the electrochemical flow cell which uses liquid electrolytes to store hundreds of megawatt-hours of electricity—enough to power thousands of homes for many hours at a stretch. But flow cell capacity still deteriorates over time due to some inevitable crossover of the liquids during the charging and discharging processes, so more work needs to be done before the technology can be widely adopted. Flow cells based on the element vanadium currently provide the best longevity—but vanadium is not a plentiful resource and is difficult to extract, which in turn limits the technology’s broad deployment.
Auto manufacturers and power utilities are both investing heavily in quantum computing, for essentially the same reasons. The volatility of lithium and the scarcity of vanadium is driving the search for new, better battery chemistry. This research requires the modelling of complex chemical interactions between different elements and compounds at the molecular level—and modelling the energy configuration of a single relatively simple molecule, caffeine, has been calculated by IBM to require on the order of 1048 classical bits, which is far beyond the capacity of the best high-performance computers today. For an idea of the size, 1048 represents a rough approximation of the number of atoms in the earth. Running extremely simplified simulations of chemical interactions in a battery would take days for even the biggest high-performance computers, so it’s estimated that bringing new batteries to market could take years or even decades.
The same molecular model could be done with about 160 logical qubits. Now, logical qubits are assembled from hundreds or thousands of physical qubits as a means of building fault tolerance into quantum computing, and 160 logical qubits is beyond the scope of today’s quantum hardware—but the hardware is improving quickly. Using techniques like IBM’s quantum utility facilitated by mathematical error mitigation, meaningful experiments can already be done in preparation for more robust hardware in the near future. Quantum computing is bringing computational chemistry into the mainstream, enabling computerized simulation of experiments that were previously only possible in the laboratory. Better simulated experiments allow researchers to more quickly eliminate non-promising results, so that expensive lab resources can be directed toward better candidate materials.
And artificial intelligence is being deployed to help quantum computing along the way. Computational chemistry requires the processing of enormous data sets, so machine learning helps in building data models and identifying correlations that can lead to faster discovery of materials—for the batteries themselves and the surrounding infrastructure and insulation. Early research results are showing that quantum combined with machine learning can lead to a tenfold improvement in both the time and the cost required to identify potential new battery materials. The result, hopefully, will be better and safer batteries—that weigh less, operate cooler, charge faster and drain slower.
The long and winding lode
The stakes are high. Better batteries will extend the driving range for EVs, in turn enabling better optimization for placement of charging stations. The first auto manufacturer to leverage quantum computing and bring a new battery to market will literally win the race for widespread EV adoption. Similarly, the first utility operators using quantum computing to develop and deploy stable flow cells will transform the power grid and gain an enduring advantage in delivering cost-effective renewable energy.
Quantum and machine learning together may well drive us over the horizon and into the long-anticipated green future.