By Kent Griffith
July 14, 2022 | Portable electronics and lithium-ion batteries have grown up together over the past thirty years. While EV battery capacity is dwarfing that used in consumer electronics, the number of lithium-ion batteries you carry and the profound ways in which they impact your life are as large as ever. How often do you adjust your plans to take into consideration when and where you can charge your phone or laptop? It may even affect which device you purchase and, for the DIYers, whether to buy a corded vs. cordless power tool. Researchers from leading electronic device companies gathered at the International Battery Seminar this spring to discuss future directions in a field that is led as much by consumer behavior as it is by technology constraints.
Smart charging was the dominant theme for Intel, Microsoft, and Lenovo. Naoki Matsumura, Intel Corporation, introduced us to context-based battery charging and the ways in which artificial intelligence can influence the field. Matsumura emphasized that fast charging, full charging, or holding the battery at high states of charge may unnecessarily degrade the battery cycle life while providing little or no benefit to the user. These conditions also increase cell impedance and reduce battery power and performance limits. For example, many laptops are left plugged in at 100% state of charge while in the office or even semi-permanently for those working from home. In an ideal scenario, the device should be at an intermediate state of charge until just before mobility is required. In the words of Matsumura, “Charge as needed, fast charge when needed.” The challenge is in implementing such a charging solution.
Toward this goal, Matsumura conducted a deep learning study with field data from 120 randomly selected PC users. Note that in the hierarchy of Intel, artificial intelligence (AI) is a broad term that encompasses machine learning (ML) with specified features which then encompasses deep learning with hidden matrix layers. Taking the first 30 days of usage as a training set, the model was able to identify patterns and predict smart charging for the rest of a year that covered more than 90% of users from running out of battery while achieving a 1.3–1.7x longevity extension by avoiding superfluous fast charging or full charging. The model was applied to LiCoO2//graphite cells but could be done for others. According to Matsumura, the main errors were associated with travel, which is something that the user likely knows in advance and could input into the computer. It could even be read from the calendar. He also noted that the battery usage data was taken from a pre-pandemic period because prediction would be too easy for a work-from-home environment.
Ryan Klee, Microsoft Corporation, echoed many of previous ideas as he discussed his company’s approach to AI-enabled smart charging. According to Klee, laptop users can be broken down into three main categories: mobile, constantly plugged, and office. The goal is to reduce the amount of time spent at 80–100% state of charge, i.e., above 4.2 V. In the concept presented by Klee, smart charging warns the user and then allows the battery to discharge to 80%. A heart appears over the battery indicating that it is staying at 80% and this branding will be available to any Windows platform. At this time, Klee said that AI-informed smart charging is not applicable to mobile users, easy for constantly plugged users, and hard but worthwhile for office users. The charging model uses battery data from tens of thousands of users who opted-in to data sharing. Users also have the option to opt-out of smart charging entirely. Among the challenges with user adoption are the perception that AI/ML is random or simply a lack of concern about battery life. In the future, additional data could inform their usage models such as calendar inputs, battery health data, and time zone changes.
Over at Lenovo, Jeremy Carlson focused specifically on fast charging with the thesis that fast charging in personal computers is possible, but that there are many trade-offs. The industry generally targets 1000 cycles and three years of battery longevity and then takes the highest energy density that will meet these requirements with maximum charge rate being somewhat of an afterthought but in the range of 1C to 1.5C. He also noted that fast charging above 5C is difficult to imagine because it would require a portable charger capable of delivering more than 500 W. Carlson identified three trends that he believes are counter to fast charging: (i) smaller, thinner designs, (ii) mobility, and (iii) sustainability. In the context of smaller and thinner devices, fast charging is hard because higher rate batteries are typically lower in energy density. Furthermore, higher power electronics mean more printed circuit board space, larger connectors, and more thermal management. With respect to mobility, fast charging means a bigger and heavier adapter (as well as a larger battery if runtime is kept constant). The enhanced degradation under fast charging conditions with graphite-based batteries may mean more frequent replacement, which is not environmentally friendly. On the other hand, there are a range of technologies that may help mitigate the trade-offs including new anode materials, multi-tab manufacturing, extended power range with USB-C charging, programmable power supply mode, gallium nitride power devices for smaller AC adapters and DC/DC converters, advanced charging algorithms, and contextually aware charging. He also noted that some applications have stronger motivation for high power and fast charging such as medical, industrial, gaming, and education.
In the world of power tools, input and output power are critical metrics. Lisa King of Stanley Black and Decker operates in a very different world from laptops and most other consumer electronics. In cordless power tools, power requirements can be 10–20C continuous operation with 30C pulses. The big change at Stanley Black and Decker has been the recent introduction of the pouch cell power pack they call PowerStack. To the questions “Why pouch, why now?”, King answered that they looked into pouch cells in the early 2010s but were only seeing 10% higher power and twice the cost vs. other form factors. With advances in nickel-rich cathodes, electrolytes, and tab design, they are now working with pouch cells that deliver 8 kW/l at 450 kWh/l, more than twice the power compared to an energy-equivalent cylindrical cell. The charge rates are now pushing 2C, though they would still like to charge faster. King noted that pouch cells also operate better than cylindrical cells under the extreme thermal conditions experienced in power tool packs (in excess of 130 °C).
Batteries are a game of trade-offs. In a world without a holy grail, consumer patterns and preferences drive manufacturer decisions regarding battery materials, cell design, and charging protocols.