Many electric car advocates are heralding the advent of Tesla’s enormous battery factory, known as the “Gigafactory,” and its new Model 3 electric sedan as great advances for the environment. What they are overlooking are the large quantities of energy and resources that are consumed in lithium-ion battery manufacturing and how these quantities might increase in the future as the production of electric vehicles (EVs) and battery storage ramps up.
Most of the credible life cycle assessment (LCA) studies for different lithium-ion chemistries find large greenhouse gas emissions per kWh of battery. Here are the CO2-eq emissions per kWh with the battery chemistry listed in parentheses:
Hao et al. (2017): 110 kg (LFP), 104 kg (NMC), 97 kg (LMO)
Ellingsen et al. (2014): 170 kg (NMC)
Dunn et al. (2012): 40 kg (LMO)
Majeau-Bettez et al. (2011): 200 kg (NMC), 240 kg (LFP)
Ou et al (2010): 290 kg (NMC)
Zackrisson et al (2010): 440 kg (LFP)
Dunn et al. and Hao et al. are based on the GREET model developed by Argonne National Laboratory, which sums up the steps in the process and is based on the estimated energy consumption for each step. In contrast, Ellingsen et al. and Zackrisson et al. are based on the total energy consumption used by a working battery factory, which better captures all the energy in the processing steps, but the data is old and the battery factory was not very energy efficient, nor was it operating at full capacity. Battery manufacturing is getting more energy efficient over time and the energy density of the batteries is increasing by roughly 7% a year, so less materials are needed per kWh of battery. It is also worth noting that no LCA studies have been conducted on the NCA chemistry used by Tesla. NCA has very high emissions per kg due to the large amount of nickel in the cathode, but is very energy dense, so less total material is needed per kWh, so it is probably similar in emissions to NMC.
The big debate in the LCA studies of battery manufacturing is how much energy is consumed per kWh of battery in the battery factory. In terms of MJ per kWh of battery, Ellingsen et al. estimate 586 MJ, Zachrisson et al. estimate 451 MJ and Majeu-Bettez et al. estimate 371-473 MJ. However, the energy for the drying rooms and factory equipment is generally fixed, regardless of the throughput. Ellingsen et al (2014) found that the energy expended to manufacture a kWh of battery could vary as much as 4 times, depending on whether the factory is operating at full capacity or partial capacity. Since the Gigafactory will probably be operating a full capacity and energy efficiency is improving, let’s assume between 100 MJ and 150 MJ per kWh of battery in the Gigafactory (which converts to 28 – 42 kWh per kWh of battery).
If producing 105 GWh of batteries per year at 100 – 150 MJ per kWh, plus another 45 GWh of packs with batteries from other factories at 25 MJ per kWh, the Gigafactory will consume between 3,229 and 4,688 GWh per year, which is between 8.3% and 12.0% of the total electrical generation in Nevada in 2016.
Telsa announced in January that it plans on installing a 70 MW solar array on the roof of the Gigafactory, which will be the largest rooftop solar array in the world. Presumably the Nevada Gigafactory will use solar panels produced by the Buffalo Gigafactory which are based on Panasonic HIT cells, which are 19.7% efficient in 96 cell modules. The 1.3 million square meter roof of the Gigafactory has the space for a solar array which is 4 times larger, so Tesla presumably chose to install fewer panels with widely spaced rows to avoid shading and use single or double axis tracking to get optimal sun all day. Assuming a yearly average of 7.16 kWh/m2/day of solar radiation, 19% efficiency in the panels in a dusty environment and a 10% system loss, the Gigafactory’s solar array will generate 158 GWh of electricity per year.
Solar panels in dusty locations such as Nevada loose roughly 25% of their output if they are not regularly cleaned. Although robots have been developed to clean panels with brushes, water will most likely be used to clean the Gigafactory’s panels. A study by Sandia National Laboratory found that photovoltaic energy plants in Nevada consume 0.0520 acre-feet of water per MW of nameplate capacity per year. The solar panels at the Gigafactory will probably have 25% less area per MW than the solar panels in the Sandia study, so we can guesstimate that the 70 MW of solar panels on the Gigafactory roof will consume 2.73 acre-feet or 3.37 million liters of water per year.
Solar panels can also be placed on the ground around the factory, but consider the fact that the Gigafactory will only receive 4.23 kWh/m2/day in December, compared to 9.81 kWh/m2/day in July. With less than half the energy from the panels during the winter, the Gigafactory will need other sources of energy during the times when it is cloudy and the sun’s rays are more indirect. Even during the summer, the Gigafactory will probably have to use temporary battery storage to smooth out the solar output or get additional energy with electric utilities which use gas peaking, battery storage or buy energy from the regional grid to give the Gigafactory a stable supply of electricity.
The original mockup of the Gigafactory showed wind turbines on the hillsides around the plant, but wind energy will not work onsite, because the area has such low wind speed. A weather station in the Truckee River valley along I-80, near the Gigafactory, measures an average wind speed of 3.3 m/s at a height of 6 meters, although the wind speed is probably higher at the site of the Gigafactory. Between 4 to 5 m/s is the minimum wind speed to start generating any energy, and between 5 and 6 m/s is generally considered the minimum for wind turbines to be economically viable. It might be possible to erect viable wind turbines onsite with 150 m towers to capture better wind, but the high costs make it likely that Tesla will forgo that option.
The region has good geothermal energy at depths of 4000 to 6000 feet and this energy is not variable like solar and wind. However, there is a great deal of risk in geothermal exploration which costs $10 million to drill a test well. It is more likely that Tesla will try to buy geothermal energy from nearby producers, but geothermal energy in the region is already in heavy demand, due to the clean energy mandates from California, so it won’t be cheap.
Tesla and Panasonic must have figured out how to consume much less energy per kWh of battery in the Gigafactory, because there is no way that between 3,229 and 4,688 GWh per year can be produced on site and buying that much energy from external sources would strain the production capability of the Nevada grid. During a tour of the Gigafactory (at 8:40 in the video), a Tesla engineer commented that they have redesigned their drying room ovens and the recapture of solvent to consume 80% less energy. If Tesla and Panasonic have managed to reduce the energy consumption in the Gigafactory to that degree, then an estimate of 10 to 30 MJ of energy per kWh is a reasonable guesstimate for cell manufacturing, with another 1 to 2 MJ per kWh for pack assembly. If the Gigafactory produces 105 GWh of cells and 150 GWh of packs per year, then the factory would consume between 38 and 109 MW, or between 333 and 958 GWh per year. (I arrived at these assumptions after a discussion on the Tesla Users Forum.)
That level of energy would mean building a solar array of between 137 or 395 MW in nameplate capacity, which at the current prices of around $1.1 million per MW would cost Tesla between $137 and $395 million to build. In addition, the Gigafactory would need battery storage of between 0.9 and 2.6 GWh if it wants to use only onsite energy, during times when there isn’t sun. At a cost of $150 per kWh for batteries, the energy storage system would cost Tesla between $137 and $394 million. Those totals would be doubled if wanting to operate at 100% renewable energy in the winter with onsite solar, but presumably Tesla means net zero energy, meaning that it will sell excess energy to the grid during the sunny months and be buying energy from the Nevada grid during the winter months when there is less sun.
Despite Musk’s rhetoric about producing 100% of the Gigafactory’s energy onsite from renewable sources, Tesla knows that it is highly unrealistic in the near future, which is why it negotiated to get $8 million in electricity rebates from the state of Nevada over an 8 year period. It is possible that the Gigafactory will buy hydroelectric energy from Washington or Oregon, but California already competes for that electricity. If Tesla wants a diversified supply of renewable energy to balance out the variability of its solar panels, it will probably have to provide guaranteed returns for third parties to build new geothermal plants or wind farms in the region.
Given the fact that wind is not a viable option onsite and a geothermal plant will take years to build, solar seems to be the only viable option for onsite energy generation in the near term. If the energy consumption of the Gigafactory only requires a solar array of 137 MW in size, then it is likely that 100% of the energy will be generated on site. If the Gigafactory requires a solar array around 395 MW in size, it will be more challenging, but still possible given the cost of roughly $800 million to build a solar array and battery storage of of that size. If Tesla doesn’t manage to cut its energy consumption down to less than 30 MJ per kWh of battery cells, then it is highly unlikely that the company will be able to generate all its energy on the 2864 acres occupied by the Gigafactory. Telsa can buy more land, but the cost of such a large a large solar array and battery system is the big question for the company.
Tesla is likely to need to buy to buy external electricity from the Nevada grid in the short term. Let’s assume a worst-case scenario were 2/3rds of the electricity consumed by the Gigafactory will come from the standard Nevada grid and 1/3 will be generated onsite or be bought from clean sources. In 2016, utility-scale electricity generation in Nevada was 72.8% natural gas, 5.5% coal, 4.5% hydroelectric, 0.9% wind, 5.7% PV solar, 0.6% concentrated solar, 9.8% geothermal, 0.14% biomass and 0.03% petroleum coke. If we use the grams of CO2-eq per kWh estimated by IPCC AR5 WGIII and Bruckner et al (2014), then natural gas emits 595 g, coal emits 1027 g, petroleum emits 880 g, hydroelectric emits 24 g, terrestrial wind emits 11 g, utility PV solar emits 48 g, residential PV solar emits 41 g, concentrated solar emits 27 g, geothermal emits 38 g and biomass emits 230 g. Based on those emission rates, grid electricity in Nevada emits 499 g CO2-eq per kWh. If 2/3rds comes from the grid and 1/3 comes from rooftop PV solar or a similar clean source, then the electricity used in the Gigafactory will emit 346 g CO2 per kWh.
If consuming between 3,229 and 4,688 GWh per year, the Gigafactory will emit between 1.12 and 1.62 megatonnes of CO2-eq per year, which represents between 3.1% and 4.5% of the greenhouse gas emissions that the state of Nevada produced in 2014 according to the World Resources Institute. On the other hand, if the Gigafactory manages to achieve massive improvements in energy efficiency compared to conventional battery production and manages to generate all its energy onsite from its own solar array, then the greenhouse gas emissions will be dramatically reduced. If the energy consumption is between 333 and 958 GWh per year and is 100% solar energy, then the Gigafactory’s annual greenhouse gas emissions will be between 0.016 and 0.046 megatonnes of CO2-eq. Clearly, the ecological impact of the Gigafactory depends on the level of energy efficiency the plant achieves and the amount of onsite energy generation from renewable sources. It remains to be seen whether the plant will be as carbon neutral as Elon Musk claims.
Aside from the GHG emissions from the Gigafactory, it is necessary to consider the greenhouse gas emissions from mining, refining and processing the materials used in the Gigafactory. The materials used in batteries consume a tremendous amount of energy and resources to produce. The various estimates of the energy to produce the materials in batteries and their greenhouse gas emissions shows the high impact that battery manufacturing has on the planet.
To get some idea of how much materials will be used in the NCA cells produced by the Gigafactory, I attempted to do a rough calculation of the weight of materials in 1 kWh of cells. Taking the weight breakdown of an NMC battery cell in Olofsson and Romare (2013), I used the same weight percentages for the cathode, electrolyte, anode and packaging, but scaled the energy density up from 233 kW per kg in the NCA cells in 2014 to 263 kW per kg, which is a 13% increase, since Telsa claims a 10% to 15% increase in energy density in the Gigafactory’s cells. Then, I estimated the weight of the components in the cathode, using 76% nickel, 14% cobalt, and 10% aluminum and some stochiometry to calculate the lithium and oxygen compared to the rest of the cathode materials. The 2170 cells produced by the Gigafactory will probably have different weight ratios between their components, and they will have more packaging materials than the pouch cells studied by Olofsson and Romare, but this provides a basic idea how much material will be consumed in the Tesla cells.
The estimates of the energy, the emissions of carbon dioxide equivalent, sulfur dioxide equivalent, phosphorous equivalent and human toxicity to produce the metals are taken from Nuss and Eckelman (2014), which are process-sum estimates based on the EcoInvent database. These are estimates to produce generic metals, not the highly purified metals used in batteries, and the process-sum methodology generally underestimates the emissions, so the estimates should be taken with a grain of salt but they do give some idea about the relative impact of the different components in battery cells since they use the same methodology in their calculations.
At this point we still don’t know how large the battery will be in the forthcoming Model 3, but it has been estimated to have a capacity of 55 kWh based on a range of 215 miles for the base model and a 20% reduction in the size of the car compared to the Model S. At that battery size, the cells in the Model 3 will contain 6.3 kg of lithium, 26.4 kg of nickel, 4.9 kg of cobalt, 27.9 kg of aluminum, 56.6 kg of copper and 21.0 kg of graphite.
Even more concerning is the total impact of the Gigafactory when it ramps up to its planned capacity of 150 GWh per year. Originally, the Gigafactory was scheduled to produce 35 GWh of lithium ion batteries by 2020, plus package an additional 15 GWh of cells produced in other factories. After Tesla received 325,000 preorders for the Model 3 within a week of being announced on March 31, 2016, the company ambitiously announced that it would triple its planned battery production and be able to produce 500,000 cars a year by 2018–two years earlier than initially planned. Now Elon Musk is talking about building 2 to 4 additional Gigafactories and one is rumored to have signed a deal to build one of them in Shanghai.
If the components for 1 kWh of Gigafactory batteries is correct and the Nevada plant manages to produce as much as Musk predicts, then the Gigafactory and the cells it packages from other battery factories will consume 17,119 tonnes of lithium, 71,860 tonnes of nickel, 13,292 tonnes of cobalt, 154,468 tonnes of copper and 75,961 tonnes of aluminum. All of these metals except aluminum have limited global reserves, and North America doesn’t have enough production capacity to hope to supply all the demand of the Gigafactory, except in the case of aluminum and possibly copper.
When the Gigafactory was originally announced, Telsa made statements about sourcing the battery materials from North America which would both reduce its costs and lower the environmental impact of its batteries. These claims should be treated with skepticism. The Gigafactory will reduce the transportation emissions in battery manufacturing, since it will be shipping directly from the refineries and processors, but the transportation emissions will still be very high because North America simply doesn’t produce enough of the metals needed by the Gigafactory. If the Gigafactory manufacturers 150 GWh of batteries per year, then it will consume almost 200 times more lithium than North America produced in 2013. In addition, it will also consume 166% of the cobalt, 133% of the natural graphite, 25.7% of the nickel, and 5.6% of the copper produced by North American mines in 2016. Presumably synthetic graphite will be used instead of natural graphite because it has a higher purity level of carbon and more uniform spheroid flakes which allow for the easier flow of electrons in the cathode, but most synthetic graphite comes from Asia. Only in the case of aluminum does it seem likely that the metal will come entirely from North America, since Gigafactory will consume 1.9% of North American mine production and the US has excess aluminum refining capacity and no shortage of bauxite. Even when considering that roughly 45 GWh of the battery cells will come from external battery factories which are presumably located in Asia, the Gigafactory will overwhelm the lithium and cobalt markets in North America, and strain the local supplies of nickel and copper.
Shipping from overseas contributes to greenhouse gases, but shipping over water is very energy efficient. The Gigafactory is located at a nexus of railroad lines, so it can efficiently ship the battery materials coming from Asia through the port of Oakland. The bigger problem is that most ships on international waters use dirty bunker fuels that contain 2.7% sulfur on average, so they release large quantities of sulfur dioxide into the atmosphere that cause acid rain and respiratory diseases.
A larger concern than the emissions from shipping is the fact that the production of most of these battery materials is an energy intensive process that consumes between 100 and 200 mejajoules per kg. The aluminum, copper, nickel and cobalt produced by North America is likely to come from places powered by hydroelectric dams in Canada and natural gas in the US, so they are comparatively cleaner. Most of the metal refining and graphite production in Asia and Australia, however, is done by burning coal. Most of the places that produce battery materials either lack strong pollution controls, as is the case in Russia, the Democratic Republic of Congo (DRC), Zambia, Philippines or New Caledonia, or they use dirty sources of energy, as is the case in China, India, Australia, the DRC, Zambia, Brazil and Madagascar.
Most of the world’s lithium traditionally came from pumping lithium rich subsurface water out of the salt flats of Tibet, northeast Chile, northwest Argentina and Nevada, but the places with concentrated lithium brines are rapidly being exhausted. The US Geological Survey estimates that China’s annual production of lithium which mostly comes from salt flats in Tibet has fallen from 4500 tonnes in 2012 to just 2000 tonnes in 2016. Silver Peak, Nevada, which is the only place in North America where lithium is currently extracted, may be experiencing similar production problems due to the exhaustion of its lithium, but its annual production numbers are confidential.
Since 1966 when brine extraction began in Silver Peak, the concentration of lithium in the water has fallen from 360 to 230 ppm (parts per million), and it is probably around 200 ppm today. At that concentration of lithium, 14,300 liters of water need to be extracted to produce 1 kg of battery-grade lithium metal. This subsurface water is critical in a state that only receives an average of 9 inches of rain per year. Parts of Nevada are already suffering from water rationing, so a massive expansion of lithium extraction is an added stress, but the biggest risk is that brine operations may contaminate the ground water. 30% of Nevada’s water is pumped from underground aquifers, so protecting this resource is vitally important. Lithium-rich water is passed through a series of 4 or 5 evaporation pools over a series of 12 to 18 months, where it is converted to lithium chloride, which is toxic to plants and aquatic life and can contaminate the ground water. Adams-Kszos and Stewart (2003) measured the effect of lithium chloride contamination in aquatic species 150 miles away from brine operations in Nevada.
As the lithium concentrations fall in the water, more energy is expended in pumping water and evaporating it to concentrate the lithium for processing. Argonne National Laboratory estimates that it takes 3 times as much energy to extract a tonne of lithium in Silver Peak, Nevada as in the Atacama Salt Flats of Chile, where the lithium is 7 times more concentrated. Most of the lithium in Chile and Argentina is produced with electricity from diesel generators, but in China and Australia it comes from burning coal, which is even worse.
For every kg of battery-grade lithium, 4.4 kg of slaked lime is consumed to remove magnesium and calcium from the brine in Silver Peak. The process of producing this lime from limestone releases 0.713 kg of CO2 for every kg of lime. In addition, 5 kg of soda ash (Na2CO3) is added for each kilo of battery-grade lithium to precipitate it as lithium carbonate. Production of soda ash is also an energy intensive process which produces greenhouse gases.
Although lithium is an abundant element and can be found in ocean water and salty lakes, there are only 4 places on the planet where it is concentrated enough without contaminants to be economically extracted from the water and the few places with concentrated lithium water are rapidly being exploited. In 2008, Meridian International estimated that 2 decades of mining had extracted 20% of the lithium from the epicenter of the Atacama Salt Flats where lithium concentrations are above 3000 ppm. According to Meridian’s calculations, the world only had 4 million tonnes of high-concentration lithium brine reserves remaining in 2008.
As the best concentrations of lithium brine are being exhausted, extraction is increasingly moving to mining pegmatites, such as spodumene. North Carolina, Russia and Canada shut down their pegmatite operations because they couldn’t compete with the cheap cost of lithium from the salt flats of Chile and Argentine, but Australia and Zimbabwe have dramatically increased their production of lithium from pegmatites in recent years. Between 2004 and 2016, the percentage of global lithium from pegmatites increased from 39% to 44%.
In 2016, Australia produced 40.9% of the global lithium supply by processing spodumene, which an extremely energy-intensive process. It takes 125 MJ of energy to extract a kilo of lithium from Chile’s salt flats, whereas 850 MJ is consumed to extract the same amount of lithium from spodumene in Australia. The spodumene is crushed, so it can be passed through a flotation beneficiation process to produce a concentrate. That concentrate is then heated to 1100ºC to change the crystal structure of the mineral. Then, the spodumene is ground and mixed with sulfuric acid and heated to 250ºC to form lithium sulfate. Water is added to dissolve the lithium sulfate and it is filtered before adding soda ash which causes it to precipitate as lithium carbonate. As lithium extraction increasingly moves to pegmatites and salt flats with lower lithium concentrations, the energy consumption will dramatically increase to produce lithium in the future.
Likewise, the energy to extract nickel and cobalt will also increase in future. The nickel and cobalt from Canada and the copper from the United States, generally comes from sulfide ores, which require much less energy to refine, but these sulfide reserves are limited. The majority of nickel and cobalt, and a sizable proportion of the copper used by the Gigafactory will likely come from places which present ethical challenges. Nickel from sulfide ores generally consumes less than 100 MJ of energy per kg, whereas nickel produced from laterite ores consumes between 252 and 572 MJ per kg. All the sulfide sources emit less than 10 kg of CO2 per kg of nickel, whereas the greenhouse gas emissions from laterite sources range from 25 to 46 kg CO2 per kg of nickel. It is generally better to acquire metals from sulfide ores, since they emit fewer greenhouse gases and they generally come from deeper in the ground, whereas laterite ores generally are produced by open pit and strip mining which causes greater disruption of the local ecology. Between 2004 and 2016, the percentage of global primary production of nickel from laterite ores increased from 40% to 60% and that percentage will continue to grow in the future, since 72% of global nickel “resources” are laterites according to the US Geological Survey.
Cobalt is a byproduct of copper or nickel mining. The majority of the sulfide ores containing copper/cobalt are located in places like Norilsk, Russia, Zambia and the Katanga Province of the Democratic Republic of Congo, where there are no pollution controls to capture the large amounts of sulfur dioxide and heavy metals released by smelting. The refineries in Norilsk, Russia, which produce 11% of the world’s nickel and 5% of its cobalt, are so polluting, that nothing grows within a 20 kilometer radius of the refineries and it is reported that Norilsk has the highest rates of lung cancer in the world.
The Democratic Republic of Congo currently produces 54% of the world’s cobalt and 5% of its copper. Buying cobalt from the DRC helps fuel a civil war in the Katanga Province where the use of children soldiers and systematic rape are commonplace. Zambia, which is located right over the border from Katanga Province, produces 4% of the world’s cobalt and copper and it also has very lax pollution controls for metal refining.
Most of the cobalt and nickel produced by the DRC and Zambia is shipped to China for refining by burning coal. China has cracked down on sulfur dioxide and heavy metal emissions in recent years, and now the DRC is attempting to do more of the refining within its own borders. The problem is that the DRC produces most of its energy from hydroelectric dams in tropical rainforests, which is the dirtiest energy on the planet. According to the IPCC (AR5 WGIII 2014), hydroelectric dams typically emit a medium of 24 g of CO2-eq per kWh, but tropical dams accumulate large amounts of vegetation which collect at the bottom of the dam where bacteria feeding on the decaying matter release methane (CH4) in the absence of oxygen. There have been no measurements of the methane released by dams in the DRC, but studies of 3 Amazonian hydroelectric dams found that they emit an average of 2556 g CO2-eq per kWh. Presumably the CO2 from these dams would have been emitted regardless of whether the vegetation falls on the forest floor or in a dam, but rainforest dams are unique environments without oxygen that produces methane. If we only count the methane emissions, then Amazonian hydroelectric dams emit an average of 2044 g CO2-eq per kWh. Any refining of copper/cobalt in the DRC and Zambia or nickel/cobalt in Brazil will likely use this type of energy which emits twice as much greenhouse gases as coal.
To avoid the ethical problems with obtaining nickel and cobalt from Russia and cobalt and copper from the DRC and Zambia, the Gigafactory will have to consume metals from laterite ores in places like Cuba, New Caledonia, Philippines, Indonesia and Madagascar, which dramatically increases the greenhouse gas emissions of these metals. The nickel/cobalt ore from Moa, Cuba is shipped to Sherritts’ refineries in Canada, so presumably it will produced with pollution controls in Cuba and Canada and relatively clean sources of energy. In contrast, the nickel/cobalt mining in the Philippines and New Caledonia has generated protracted protests by the local population who are effected by the contamination of their water, soil and air. When Vale’s $6 billion high pressure acid leaching plant in Goro, New Caledonia leaked 100,000 liters of acid-tainted effluent leaked into a local river in May 2014, protesters frustrated by the unaccountability of the mining giant burned a third of its trucks and one of its buildings, causing between $20 and $30 million in damages. The mining companies extracting nickel and cobalt in the Philippines have shown so little regard for the health of the local people, that the public outcry induced the Duterte administration to recently announce that it will prohibit all open pit mining of nickel. If this pronouncement is enforced, the operations of 28 of the 41 companies mining nickel/cobalt in the country will be shut down and the global supply of nickel will be reduced between 8% and 10%.
Most refining of laterite ores in the world is done with dirty energy, which is problematic because these ores require so much more energy than sulfide ores. Much of the copper/cobalt from the DRC and Zambia and the nickel/cobalt from the Philippines is shipped to China where it is refined with coal. The largest nickel/cobalt laterite mine and refinery in the world is the Ambatovy Project in Madagascar. Although the majority of the electricity on the island comes from hydroelectric dams, the supply is so limited that Ambatovy constructed three 30 MW of coal-powered generators, plus 30 MW of diesel powered generators.
It is highly likely that many of the LCA studies of lithium-ion batteries have underestimated the energy and greenhouse gas emissions to produce their metals, because they assume that the lithium comes from brine operations and the copper, nickel and cobalt come from sulfide ores with high metal concentrations. As lithium extraction increasingly shifts to spodumene mining and nickel and cobalt mining shifts to laterite ores, the greenhouse gas emissions to produce these metals will dramatically increase.
As the global production of lithium-ion batteries ramps up, the most concentrated ores for these metals will become exhausted, so that mining will move to less-concentrated sources, which require more energy and resources in the extraction and processing. In 1910, copper ore in the US contained 1.9% copper. By 1950, this percentage had fallen to 0.9% copper, and by 1980 it was at 0.5% copper. As the concentration of copper in the ore has fallen, the environmental impact of extraction has risen. In a study of the smelting and refining of copper and nickel, Norgate and Rankin (2000) found that the energy consumption, greenhouse gas emissions and sulfur dioxide emissions per kg of metal rose gradually when changing from ore with 3% or 2% metal to 1% metal, but below 1% the environmental impacts increased dramatically. MJ/kg, CO2/kg and SO2/kg doubled when moving from ore with 1% metal to ore with 0.5% metal, and they doubled again when moving to 0.25% metal. Producing a kilo of copper today in the US has double the environmental impact of a kg of copper half a century ago and it will probably have 4 times the impact in the future.
The enormous demand for metals by battery manufacturers will force the mining companies to switch to less and less concentrated ores and consume more energy in their extraction. If the Nevada Gigafactory produces 150 GWh of batteries per year, then it will dramatically reduce the current global reserves listed by the US Geological Survey. The Nevada Gigafatory will cut the current global lithium reserves from 400 to 270 years, assuming that current global consumption in other sectors does not change (which is highly unlikely). If the Gigafactory consumes metals whose recycled content is the US average recycling rate, then the current global copper reserves will be reduced from 37.1 to 36.9 years, the nickel reserves from 34.7 to 33.9 years, and the cobalt reserves from 56.9 to 52.5 years.
Recycling at the Gigafactory will not dramatically reduce its demand for metals. If we assume that 80% of the metal consumed by the Gigafactory will come from recycled content starting in 15 years when batteries start to be returned for recycling, then current global reserves will be extended 0.04 years for copper, 0.09 years for nickel, 0.9 years for cobalt. Only in the case of lithium will recycling make a dramatic difference, extending the current reserves 82 years for lithium.
The prospects for global shortages of these metals will become even more dire if the 95.0 million vehicles that the world produced in 2016 were all long-range electrics as Elon Musk advocates for “sustainable transport.” If the average vehicle (including all trucks and buses) has a 50 kWh battery, then the world would need to produce 4750 GWh of batteries per year just for electric vehicles, which would be the output of 32 Gigafactories. With energy storage for the electrical grid, that total will probably double, so 64 Gigafactories will be needed. Even that might not enough. In Leonardo de Caprio’s documentary Before the Flood, Elon Musk states, “We actually did the calculations to figure out what it would take to transition the whole world to sustainable energy… and you’d need 100 Gigafactories.”
Lithium-ion batteries will get more energy dense in the future, but they are unlikely to reach the high energy density of the NCA cells produced in the Gigafactory, if using the LMO or LFP chemistries. For that kind of energy density, they will probably need either an NCA or an altered NMC chemistry which is 70%-80% nickel, so the proportion of lithium, nickel, cobalt and copper in most future EV batteries is likely to be similar to the Gigafactory’s NCA cells. If 4750 GWh of these batteries are produced every year at an energy density of 263 Wh/kg, then the current global reserves will be used up in 24.5 years for lithium, 31.2 years for copper, 20.2 years for nickel, and 15.4 years for cobalt. Even if those batteries are produced with 80% recycled metals, starting in 15 years time, the current global lithium reserves would be extended 6.6 years, or 7.4 years if all sectors switch to using 80% recycled lithium. Using 80% recycled metal in the batteries would extend current copper, nickel and cobalt reserves by 0.7, 0.5 and 0.1 years, respectively. An 80% recycling rate in all sectors would make a difference for copper, extending its reserves by 11.5 years, but only 2.8 years for nickel and 0.2 years for cobalt. In other words, recycling will not significantly reduce the enormous stresses that lithium-ion batteries will place on global metal supplies, because they represent so much new demand for metals.
As the demand for these metals increases, the prices will increase and new sources of these metals will be found, but they will either be in places like the DRC with ethical challenges or in places with lower quality ores which require more energy and resources to extract and refine. We can expect more energy-intensive mining of spodumene and more strip mining of laterite ores which cause more ecological disruption. The ocean floor has enormous quantities of manganese, nickel, copper and cobalt, but the energy and resources to scrap the bottom of the ocean will dramatically increase the economic and ecological costs. If battery manufacturing dramatically raises the prices of lithium, nickel, cobalt, copper (and manganese for NMC cells), then it will be doubly difficult to transition to a sustainable civilization in other areas. For example, nickel and cobalt are essential to making carbide blades, tool dies and high-temperature turbine blades and copper is a vital for wiring, electronics and electrical motors. It is hard to imagine how the whole world will transition to a low-carbon economy if these metals are made prohibitively expensive by manufacturing over a billion lithium-ion batteries for EVs.
Future batteries will probably be able to half their weight by switching to a solid electrolyte and using an anode made of lithium metal, lithiated silicon or carbon nanotubes (graphene), but that will only eliminate the copper, while doing little to reduce the demand for the other metals. Switching the anode to spongy silicon or graphene will allow batteries to hold more charge per kilogram, but those materials also dramatically increase the cost and the energy and resources that are consumed in battery manufacturing.
In the near future, lithium-ion batteries are likely to continue to follow their historical trend of using 7% less materials each year to hold the same amount of charge. That rate of improvement, however, is unlikely to last. An NCA cathode currently holds a maximum of 200 mAh of energy per gram, but its theoretical maximum is 279 mAh/g. It has already achieved 72% of what is theoretically possible, so there is little scope to keep improving. NMC at 170 mAh/g is currently farther from its theoretical limit of 280 mAh/g, but the rate of improvement is likely to slow as these battery chemistries bump against their theoretical limits.
Clearly the planet doesn’t have the resources to build 95 million long-range electric vehicles each year that run on lithium-ion batteries. Possibly a new type of battery will be invented that only uses common materials, such as aluminum, zinc, sodium and sulfur, but all the batteries that have been conceived with these sorts of material still have significant drawbacks. Maybe a new type of battery will be invented that is suitable for vehicles or the membranes in fuel cells will become cheap enough to make hydrogen a viable competitor, but at this point, lithium-ion batteries appear likely to dominate electric vehicles for the foreseeable future. The only way EVs based on lithium-ion can become a sustainable solution for transport is if the world learns to live with far fewer vehicles.
Currently 3% more vehicles are being built each year, and there is huge demand for vehicles in the developing world. While demand for cars has plateaued in the developed world, vehicle manufacturing since 1999 has grown 17.4% and 10.5% per year in China and India, respectively. If the developing world follows the unsustainable model of vehicle ownership found in the developed world, then the transition to electrified transport will cause severe metal shortages. Based on current trends, Navigant Research predicts that 129.9 million vehicles will be built in the year 2035, when there will be 2 billion vehicles on the road.
On the other hand, James Arbib and Tony Seba believe that anonymous vehicles and Transport as a Service (TaaS) such as Uber and Lyft will dramatically reduce demand for vehicles, lowering the number of passenger vehicles on American roads from 247 to 44 million by 2030. If 95% of passenger miles are autonomous TaaS by 2030 and the lifespan of electric vehicles grows to 500,000 miles as Arbib and Seba predict, then far fewer vehicles will be needed. Manufacturing fewer electric vehicles reduces the pressure to extract metals from laterite ores, pegmatites, the ocean floor, and lower-grade ores in general with higher ecological costs.
Ellingsen et al (2016) estimate that the energy consumed by battery factories per kWh of batteries has halved since 2012, however, that has to be balanced by the growing use lithium from spodumene and nickel and cobalt from laterite ores, and ores with lower metal concentrations that require more energy and produce more pollution. Given the increased energy efficiency in battery manufacturing plants and the growing efficiencies of scale, I would guesstimate that lithium-ion battery emissions are currently at roughly 150 kg CO2-eq per kWh of battery and that the Gigafactory will lower those emissions by a third to roughly 100 kg CO2-eq / kWh. If the Model 3, uses a 55 kWh battery, than its battery emissions would be roughly 5500 kg CO2-eq.
Manufacturing a medium-sized EV without the battery emits 6.5 tonnes of CO2-eq according to Ellingsen et al (2016). Electric cars don’t have the huge engine block of an ICE car, but they have large amounts of copper in the motor’s rotor and the windings and the Model 3 will have far more electronics than a standard EV. The Model S has 23 kg of electronics and I would guesstimate that the Model 3 will have roughly 15 lbs of electronics if it contains nVidia’s Drive PX or a custom processor based on the K-1 graphics processor. If the GHG emissions are roughly 150 kg CO2-eq per kg of electronics, we can guesstimate that 2.2 tonnes of CO2-eq will be emitted to manufacture the electronics in the Model 3. Given the large amount of copper, electronics and sensors in the Model 3, add an additional tonne, plus 5.5 tonnes for its 50 kWh battery, so a total of 13 tonnes of CO2-eq will be emitted to manufacture the entire car.
Manufacturing a medium-sized ICE car emits between 5 and 6 tonnes, so there is roughly a 7.5 tonne difference in GHG emissions between manufacturing the Model 3 and a comparable ICE car. A new ICE car the size of the Model 3 will get roughly 30 mpg. In the US, a gallon of gasoline emits 19.64 lbs of CO2, but it emits 24.3 lbs of CO2e when the methane and nitrous oxide are included, plus the emissions from extraction, refining and transportation, according to the Argonne National Laboratory. Therefore, we will need to burn 680 gallons of gasoline or drive 20,413 miles at 30 mpg to equal those 7.5 extra tonnes in manufacturing the Model 3.
At this point, the decision whether the Model 3 makes ecological sense depends on where the electricity is coming from. Let’s assume that the Model 3 will consume 0.30 kWh of electricity per mile, which is what the EPA estimates the Nissan Leaf to consume. The Model S will be a smaller and more aerodynamic car than the Leaf, but it will also weigh significantly more due to its larger battery. If we also include the US national average of 4.7% transmission losses in the grid, then the Model 3 will consume 0.315 kWh per mile. After driving the Model 3 100,000 miles, the total greenhouse gas emissions (including the production emissions) will range between 14.1 and 45.3 tonnes, depending on its energy source to charge the battery.
In comparison, driving a 30 mpg ICE car (with 5.5 tonnes in production emissions) will emit 42.2 tonnes of CO2-eq after 100,000 miles. If we guesstimate that manufacturing a Toyota Prius will emit 7 tonnes, then driving it 100,000 miles at 52 mpg will emit 28.2 tonnes. Only in places like Kentucky which get almost all their electricity from coal is an ICE car the better environmental choice. The Model 3, however, will have worse emissions than most of its competitors in the green car market, if it is running on average US electricity, which emits 528 grams of CO2-eq per kWh. It will emit slightly more than a plugin hybrid like the Chevy Volt and an efficient hybrid like the Toyota Prius and substantially more than a short-range electric, like the Nissan Leaf.
Most previous comparisons between electric cars and ICE cars were based on short-range electrics with smaller batteries, such as the Nissan Leaf, which is why environmental advocates are so enthusiastic about EVs. However, comparing the Model S and Model 3 to the Nissan Leaf, Chevy Volt and Toyota Prius hybrid shows that the environmental benefits of long-range EVs are questionable when compared to short-range EVs, plugin hybrids and hybrids. Only when running the Model 3 on cleaner sources of electricity does it emit less greenhouse gases than hybrids and plugin hybrids, but in the majority of the United States it will emit slightly more. Many of the early adopters of EVs also owned solar panels, so buying a Model 3 will reduce their carbon footprint, but the proportion of EV owners with solar panels on their roofs is falling. According to CleanTechnica’s PlugInsights annual survey, 25% of EV buyers before 2012 had solar panels on their roofs, compared to just 12% in 2014-2015. Most people who own solar panels do not have a home battery system so they can not use their clean energy all day, and most EV charging will happen at night using dirtier grid electricity.
Another factor to consider is the effect of methane leakage in the extraction and transport of natural gas. There is a raging scientific debate about what percentage of natural gas leaks into the atmosphere without being burned. A number of studies have concluded that the leakage of methane causes electricity from natural gas to have GHG emissions similar to coal, but there is still no consensus on the matter. If the leakage rate is as high as some researchers believe, then EVs will emit more greenhouse gases than hybrids and efficient ICE cars in places like California which burn large amounts of natural gas.
On the other hand, many people believe that EVs will last 300,000 miles or even 500,000 miles since they have so few moving parts, so their high emissions in manufacturing will be justified. However, the EV battery will probably have to be replaced, and the manufacturing emissions for a long range EV battery can be as high as building a whole new ICE car. Another factor that could inhibit the long life of Telsa’s cars is the fact that the company builds cars described as “computers on wheels,” which are extremely difficult for third parties to fix and upgrade over time. Telsa only sells its parts to authorized repair shops and much of the functionality of car is locked up with proprietary code and secret security measures, as many do-it-yourselfers have discovered to their chagrin. When Tesla cars are damaged and sold as salvage, Tesla remotely disables its cars, so that they will no longer work even if repaired. The $600 inspection fee to reactivate the car plus the towing fees discourage Teslas from being fixed by third parties. These policies make it less likely that old Teslas will be fixed and their lifespans extended to counterbalance the high environmental costs of producing the cars.
Although the Model 3 has high greenhouse gas emissions in its production and driving it is also problematic in parts of the world that currently use dirty energy, those emissions could be significantly reduced in the future if they are accompanied by a shift to renewable energy, more recycling and the electrification of mining equipment, refining and transport. The car’s ecological benefits will increase if the emissions can be decreased in producing battery materials and the greater energy density of batteries is used to decrease the total materials in batteries rather than keep extending the range of EVs. Producing millions of Model 3s will strain the supply of vital metals and shift extraction to reserves which have higher ecological costs. However, the Model 3 could become a more sustainable option if millions of them are deployed in autonomous Transport as a Service fleets, which Arbib and Seba predict will be widespread by 2030, since TaaS will cost a tenth of the price of owning a private vehicle. If the Model 3 and future autonomous EVs become a means to drop the global demand for private vehicles and that helps reduce the demand for lithium, nickel, cobalt and copper down to sustainable levels, then the high environmental costs of manufacturing the Model 3 would be justified.
Nonetheless, the Model 3 and the NCA 2170 batteries currently being produced by Tesla offer few of those possible future ecological benefits. Most of the metal and graphite in the battery is being produced with energy from fossil fuels. In the short term at least, Telsa batteries will keep growing in capacity to offer more range, rather than reducing the total consumption of metals per battery. The extra sensors, processing power and electronics in the current Model 3 will increase its ecological costs without providing the Level 4 or 5 autonomy that would make it possible to convince people to give up their private vehicles. In the here and now, the Model 3 is generally not the best ecological choice, but it might become a better choice in the future.
The Model 3 promises to transform the market not only for EVs, but cars in general. If the unprecedented 500,000 pre-orders for the Model 3 are any indication of future demand, then long-range electrics with some degree of autonomous driving like the Model 3 will capture most of the EV market. Telsa’s stunning success will induce the rest of auto-makers to also start making long-range EVs with large batteries, advanced sensors, powerful image processors, advanced AI, cellular networking, driving data collection and large multimedia touchscreens. These features will dramatically increase the environmental costs of car manufacturing. Whether these features will be balanced by other factors which reduce their environmental costs remains to be seen.
Much of this analysis is guess work, so it should be taken with a grain of salt, but it points out the problems with automatically assuming that EVs are always better for the environment. If we consider sulfate emissions, EVs are significantly worse for the environment. Also, when we consider the depletion of critical metal reserves, EVs are significantly worse than ICE vehicles.
The conclusion should be that switching to long-range EVs with large batteries and advanced electronics bears significant environmental challenges. The high manufacturing emissions of these types of EVs make their ecological benefits questionable for private vehicles which are only used on average 4% of the time. However, they are a very good option for vehicles which are used a higher percentage of the time such as taxis, buses and heavy trucks, because they will be driven many miles to counterbalance their high manufacturing emissions. Companies such as BYD and Proterra provide a model of the kinds of electric vehicles that Tesla should be designing to promote “sustainable transport.” Tesla has a few ideas on the drawing board that are promising from an ecological perspective, such as its long-haul semi, the renting out of Teslas to an autonomous TaaS fleet, and a new vehicle that sounds like a crossover between a sedan and a minibus for public transport. The current Model 3, however, is still a vehicle which promotes private vehicle ownership and bears the high ecological costs of long-range lithium batteries and contributes to the growing shortage of critical metals.
Clearly, EVs alone are not enough to reduce greenhouse gas emissions or attain sustainable transport in general. The first step is to work on switching the electric grid to cleaner renewable energy and installing more residential solar, so that driving an EV emits less CO2. However, another important step is redesigning cities and changing policies so that people aren’t induced to drive so many private vehicles. Instead of millions of private vehicles on the road, we should be aiming for walkable cities and millions of bikes and electric buses, which are far better not only for human health, but also for the environment.
A further step where future Model 3s may help is in providing autonomous TaaS that helps convince people to give up their private vehicles. However, autonomous EVs need to be matched by public policies that disincentivize the kind of needless driving that will likely occur in the future. The total number of miles will likely increase in the future due to autonomous electric cars driving around looking for passengers to pick up and people who spend more time in the car because they can surf the web, watch movies, and enjoy the scenery without doing the steering. Plus, the cost of the electricity to charge the battery is so cheap compared to burning gasoline that people will be induced to drive more, not less.