Thursday, March 5th, 2020

Time	Event
3:05p	3 Questions: Emre Gençer on the evolving role of hydrogen in the energy system As the world increasingly recognizes the need to develop more sustainable and renewable energy sources, low-carbon hydrogen has reemerged as an energy carrier with the potential to play a key role in sectors from transportation to power. At MITEI’s 2019 Spring Symposium, MIT Energy Initiative Research Scientist Emre Gençer gave a presentation titled “Hydrogen towards Deep Decarbonization,” in which he elaborated on how hydrogen can be used across all energy sectors. Other themes discussed by experts at the symposium included industry’s role in promoting hydrogen, public safety concerns surrounding the hydrogen infrastructure, and the policy landscape required to scale hydrogen around the world. Here, Gençer shares his thoughts on the history of hydrogen and how it could be incorporated into our energy system as a tool for deep decarbonization to address climate change. Q: How has public perception of hydrogen changed over time? A: Hydrogen has been in the public imagination since the 1870s. Jules Verne wrote that “water will be the coal of the future” in his novel “The Mysterious Island.” The concept of hydrogen has persisted in the public imagination for over a century, though interest in hydrogen has changed over time. Initial conversations about hydrogen focused on using it to supplement depleting fuel sources on Earth, but the role of hydrogen is evolving. Now we know that there is enough fuel on Earth, especially with the support of renewable energy sources, and that we can consider hydrogen as a tool for decarbonization. The first “hydrogen economy” concept was introduced in the 1970s. The term “hydrogen economy” refers to using hydrogen as an energy carrier, mostly for the transportation sector. In this context, hydrogen can be compared to electricity. Electricity requires a primary energy source and transmission lines to transmit electrons. In the case of hydrogen, energy sources and transmission infrastructure are required to transport protons. In 2004, there was a big initiative in the U.S. to involve hydrogen in all energy sectors to ensure access to reliable and safe energy sources. That year, the National Research Council and National Academy of Engineering released a report titled “The Hydrogen Economy: Opportunities, Costs, Barriers, and R&D Needs.” This report described how hydrogen could be used to increase energy security and reduce environmental impacts. Because its combustion yields only water vapor, hydrogen does not produce carbon dioxide (CO2) emissions. As a result, we can really benefit from eliminating CO2 emissions in many of its end-use applications. Today, hydrogen is primarily used in industry to remove contaminants from diesel fuel and to produce ammonia. Hydrogen is also used in consumer vehicles with hydrogen fuel cells, and countries such as Japan are exploring its use in public transportation. In the future, there is ample room for hydrogen in the energy space. Some of the work I completed for my PhD in 2015 involved researching efficient hydrogen production via solar thermal and other renewable sources. This application of renewable energy is now coming back to the fore as we think about “deep decarbonization.” Q: How can hydrogen be incorporated into our energy system? A: When we consider deep decarbonization, or economy-wide decarbonization, there are some sectors that are hard to decarbonize with electricity alone. They include heavy industries that require high temperatures, heavy-duty transportation, and long-term energy storage. We are now thinking about the role hydrogen can play in decarbonizing these sectors. Hydrogen has a number of properties that make it safer to handle and use than the conventional fuels used in our energy system today. Hydrogen is nontoxic and much lighter than air. In the case of a leak, its lightness allows for relatively rapid dispersal. All fuels have some degree of danger associated with them, but we can design fuel systems with engineering controls and establish standards to ensure their safe handling and use. As the number of successful hydrogen projects grows, the public will become increasingly confident that hydrogen can be as safe as the fuels we use today. To expand hydrogen’s uses, we first need to explore ways of integrating it into as many energy sectors as possible. This presents a challenge because the entry points can vary for different regions. For example, in colder regions like the northeastern U.S., hydrogen can help provide heating. In California, it can be used for energy storage and light-duty transportation. And in the southern U.S., hydrogen can be used in industry as a feedstock or energy source. Once the most strategic entry points for hydrogen are identified for each region, the supporting infrastructure can be built and used for additional purposes. For example, if the northeastern U.S. implements hydrogen as its primary source of residential heating, other uses for hydrogen will follow, such as for transportation or energy storage. At that point, we hope that the market will shift so that it is profitable to use hydrogen across all energy sectors. Q: What challenges need to be overcome so that hydrogen can be used to support decarbonization, and what are some solutions to these challenges? A: The first challenge involves addressing the large capital investment that needs to be made, especially in infrastructure. Once industry and policymakers are convinced that hydrogen will be a critical component for decarbonization, investing in that infrastructure is the next step. Currently, we have many hydrogen plants — we know how to produce hydrogen. But in order to move toward a semi-hydrogen economy, we need to identify the sectors or end users that really require or could benefit from using hydrogen. The way I see it, we need two energy vectors for decarbonization. One is electricity; we are sure about that. But it's not enough. The second vector can be, and should be, hydrogen. Another key issue is the nature of hydrogen production itself. Though hydrogen does not generate any emissions directly when used, hydrogen production can have a huge environmental impact. Today, close to 95 percent of its production is from fossil resources. As a result, the CO2 emissions from hydrogen production are quite high. There are two ways to move toward cleaner hydrogen production. One is applying carbon capture and storage to the fossil fuel-based hydrogen production processes. In this case, usually a CO2 emissions reduction of around 90 percent is feasible. The second way to produce cleaner hydrogen is by using electricity to produce hydrogen via electrolysis. Here, the source of electricity is very important. Our source of hydrogen needs to produce very low levels of CO2 emissions, if not zero. Otherwise, there will not be any environmental benefit. If we start with clean, low-carbon electricity sources such as renewables, our CO2 emissions will be quite low. (Comment on this)
11:59p	Showing robots how to do your chores Training interactive robots may one day be an easy job for everyone, even those without programming expertise. Roboticists are developing automated robots that can learn new tasks solely by observing humans. At home, you might someday show a domestic robot how to do routine chores. In the workplace, you could train robots like new employees, showing them how to perform many duties. Making progress on that vision, MIT researchers have designed a system that lets these types of robots learn complicated tasks that would otherwise stymie them with too many confusing rules. One such task is setting a dinner table under certain conditions.   At its core, the researchers’ “Planning with Uncertain Specifications” (PUnS) system gives robots the humanlike planning ability to simultaneously weigh many ambiguous — and potentially contradictory — requirements to reach an end goal. In doing so, the system always chooses the most likely action to take, based on a “belief” about some probable specifications for the task it is supposed to perform. In their work, the researchers compiled a dataset with information about how eight objects — a mug, glass, spoon, fork, knife, dinner plate, small plate, and bowl — could be placed on a table in various configurations. A robotic arm first observed randomly selected human demonstrations of setting the table with the objects. Then, the researchers tasked the arm with automatically setting a table in a specific configuration, in real-world experiments and in simulation, based on what it had seen. To succeed, the robot had to weigh many possible placement orderings, even when items were purposely removed, stacked, or hidden. Normally, all of that would confuse robots too much. But the researchers’ robot made no mistakes over several real-world experiments, and only a handful of mistakes over tens of thousands of simulated test runs.   “The vision is to put programming in the hands of domain experts, who can program robots through intuitive ways, rather than describing orders to an engineer to add to their code,” says first author Ankit Shah, a graduate student in the Department of Aeronautics and Astronautics (AeroAstro) and the Interactive Robotics Group, who emphasizes that their work is just one step in fulfilling that vision. “That way, robots won’t have to perform preprogrammed tasks anymore. Factory workers can teach a robot to do multiple complex assembly tasks. Domestic robots can learn how to stack cabinets, load the dishwasher, or set the table from people at home.” Joining Shah on the paper are AeroAstro and Interactive Robotics Group graduate student Shen Li and Interactive Robotics Group leader Julie Shah, an associate professor in AeroAstro and the Computer Science and Artificial Intelligence Laboratory. Bots hedging bets Robots are fine planners in tasks with clear “specifications,” which help describe the task the robot needs to fulfill, considering its actions, environment, and end goal. Learning to set a table by observing demonstrations, is full of uncertain specifications. Items must be placed in certain spots, depending on the menu and where guests are seated, and in certain orders, depending on an item’s immediate availability or social conventions. Present approaches to planning are not capable of dealing with such uncertain specifications. A popular approach to planning is “reinforcement learning,” a trial-and-error machine-learning technique that rewards and penalizes them for actions as they work to complete a task. But for tasks with uncertain specifications, it’s difficult to define clear rewards and penalties. In short, robots never fully learn right from wrong. The researchers’ system, called PUnS (for Planning with Uncertain Specifications), enables a robot to hold a “belief” over a range of possible specifications. The belief itself can then be used to dish out rewards and penalties. “The robot is essentially hedging its bets in terms of what’s intended in a task, and takes actions that satisfy its belief, instead of us giving it a clear specification,” Ankit Shah says. The system is built on “linear temporal logic” (LTL), an expressive language that enables robotic reasoning about current and future outcomes. The researchers defined templates in LTL that model various time-based conditions, such as what must happen now, must eventually happen, and must happen until something else occurs. The robot’s observations of 30 human demonstrations for setting the table yielded a probability distribution over 25 different LTL formulas. Each formula encoded a slightly different preference — or specification — for setting the table. That probability distribution becomes its belief. “Each formula encodes something different, but when the robot considers various combinations of all the templates, and tries to satisfy everything together, it ends up doing the right thing eventually,” Ankit Shah says. Following criteria The researchers also developed several criteria that guide the robot toward satisfying the entire belief over those candidate formulas. One, for instance, satisfies the most likely formula, which discards everything else apart from the template with the highest probability. Others satisfy the largest number of unique formulas, without considering their overall probability, or they satisfy several formulas that represent highest total probability. Another simply minimizes error, so the system ignores formulas with high probability of failure. Designers can choose any one of the four criteria to preset before training and testing. Each has its own tradeoff between flexibility and risk aversion. The choice of criteria depends entirely on the task. In safety critical situations, for instance, a designer may choose to limit possibility of failure. But where consequences of failure are not as severe, designers can choose to give robots greater flexibility to try different approaches. With the criteria in place, the researchers developed an algorithm to convert the robot’s belief — the probability distribution pointing to the desired formula — into an equivalent reinforcement learning problem. This model will ping the robot with a reward or penalty for an action it takes, based on the specification it’s decided to follow. In simulations asking the robot to set the table in different configurations, it only made six mistakes out of 20,000 tries. In real-world demonstrations, it showed behavior similar to how a human would perform the task. If an item wasn’t initially visible, for instance, the robot would finish setting the rest of the table without the item. Then, when the fork was revealed, it would set the fork in the proper place. “That’s where flexibility is very important,” Ankit Shah says. “Otherwise it would get stuck when it expects to place a fork and not finish the rest of table setup.” Next, the researchers hope to modify the system to help robots change their behavior based on verbal instructions, corrections, or a user’s assessment of the robot’s performance. “Say a person demonstrates to a robot how to set a table at only one spot. The person may say, ‘do the same thing for all other spots,’ or, ‘place the knife before the fork here instead,’” Ankit Shah says. “We want to develop methods for the system to naturally adapt to handle those verbal commands, without needing additional demonstrations.”   (Comment on this)
11:59p	Novel method for easier scaling of quantum devices In an advance that may help researchers scale up quantum devices, an MIT team has developed a method to “recruit” neighboring quantum bits made of nanoscale defects in diamond, so that instead of causing disruptions they help carry out quantum operations. Quantum devices perform operations using quantum bits, called “qubits,” that can represent the two states corresponding to classic binary bits — a 0 or 1 — or a “quantum superposition” of both states simultaneously. The unique superposition state can enable quantum computers to solve problems that are practically impossible for classical computers, potentially spurring breakthroughs in biosensing, neuroimaging, machine learning, and other applications. One promising qubit candidate is a defect in diamond, called a nitrogen-vacancy (NV) center, which holds electrons that can be manipulated by light and microwaves. In response, the defect emits photons that can carry quantum information. Because of their solid-state environments, however, NV centers are always surrounded by many other unknown defects with different spin properties, called “spin defects.” When the measurable NV-center qubit interacts with those spin defects, the qubit loses its coherent quantum state — “decoheres”— and operations fall apart. Traditional solutions try to identify these disrupting defects to protect the qubit from them. In a paper published Feb. 25 in Physical Letters Review, the researchers describe a method that uses an NV center to probe its environment and uncover the existence of several nearby spin defects. Then, the researchers can pinpoint the defects’ locations and control them to achieve a coherent quantum state — essentially leveraging them as additional qubits. In experiments, the team generated and detected quantum coherence among three electronic spins — scaling up the size of the quantum system from a single qubit (the NV center) to three qubits (adding two nearby spin defects). The findings demonstrate a step forward in scaling up quantum devices using NV centers, the researchers say.   “You always have unknown spin defects in the environment that interact with an NV center. We say, ‘Let’s not ignore these spin defects, which [if left alone] could cause faster decoherence. Let’s learn about them, characterize their spins, learn to control them, and ‘recruit’ them to be part of the quantum system,’” says the lead co-author Won Kyu Calvin Sun, a graduate student in the Department of Nuclear Science and Engineering and a member of the Quantum Engineering group. “Then, instead of using a single NV center [or just] one qubit, we can then use two, three, or four qubits.” Joining Sun on the paper are lead author Alexandre Cooper ’16 of Caltech; Jean-Christophe Jaskula, a research scientist in the MIT Research Laboratory of Electronics (RLE) and member of the Quantum Engineering group at MIT; and Paola Cappellaro, a professor in the Department of Nuclear Science and Engineering, a member of RLE, and head of the Quantum Engineering group at MIT. Characterizing defects NV centers occur where carbon atoms in two adjacent places in a diamond’s lattice structure are missing — one atom is replaced by a nitrogen atom, and the other space is an empty “vacancy.” The NV center essentially functions as an atom, with a nucleus and surrounding electrons that are extremely sensitive to tiny variations in surrounding electrical, magnetic, and optical fields. Sweeping microwaves across the center, for instance, makes it change, and thus control, the spin states of the nucleus and electrons. Spins are measured using a type of magnetic resonance spectroscopy. This method plots the frequencies of electron and nucleus spins in megahertz as a “resonance spectrum” that can dip and spike, like a heart monitor. Spins of an NV center under certain conditions are well-known. But the surrounding spin defects are unknown and difficult to characterize. In their work, the researchers identified, located, and controlled two electron-nuclear spin defects near an NV center. They first sent microwave pulses at specific frequencies to control the NV center. Simultaneously, they pulse another microwave that probes the surrounding environment for other spins. They then observed the resonance spectrum of the spin defects interacting with the NV center. The spectrum dipped in several spots when the probing pulse interacted with nearby electron-nuclear spins, indicating their presence. The researchers then swept a magnetic field across the area at different orientations. For each orientation, the defect would “spin” at different energies, causing different dips in the spectrum. Basically, this allowed them to measure each defect’s spin in relation to each magnetic orientation. They then plugged the energy measurements into a model equation with unknown parameters. This equation is used to describe the quantum interactions of an electron-nuclear spin defect under a magnetic field. Then, they could solve the equation to successfully characterize each defect. Locating and controlling After characterizing the defects, the next step was to characterize the interaction between the defects and the NV, which would simultaneously pinpoint their locations. To do so, they again swept the magnetic field at different orientations, but this time looked for changes in energies describing the interactions between the two defects and the NV center. The stronger the interaction, the closer they were to one another. They then used those interaction strengths to determine where the defects were located, in relation to the NV center and to each other. That generated a good map of the locations of all three defects in the diamond. Characterizing the defects and their interaction with the NV center allow for full control, which involves a few more steps to demonstrate. First, they pump the NV center and surrounding environment with a sequence of pulses of green light and microwaves that help put the three qubits in a well-known quantum state. Then, they use another sequence of pulses that ideally entangles the three qubits briefly, and then disentangles them, which enables them to detect the three-spin coherence of the qubits. The researchers verified the three-spin coherence by measuring a major spike in the resonance spectrum. The measurement of the spike recorded was essentially the sum of the frequencies of the three qubits. If the three qubits for instance had little or no entanglement, there would have been four separate spikes of smaller height. “We come into a black box [environment with each NV center]. But when we probe the NV environment, we start seeing dips and wonder which types of spins give us those dips. Once we [figure out] the spin of the unknown defects, and their interactions with the NV center, we can start controlling their coherence,” Sun says. “Then, we have full universal control of our quantum system.” Next, the researchers hope to better understand other environmental noise surrounding qubits. That will help them develop more robust error-correcting codes for quantum circuits. Furthermore, because on average the process of NV center creation in diamond creates numerous other spin defects, the researchers say they could potentially scale up the system to control even more qubits. “It gets more complex with scale. But if we can start finding NV centers with more resonance spikes, you can imagine starting to control larger and larger quantum systems,” Sun says. (Comment on this)

<< Previous Day

2020/03/05 [Calendar]

Next Day >>