Trade-off between critical metal requirement and transportation decarbonization in automotive electrification

Automotive electrification holds the promise of mitigating transportation-related greenhouse gas (GHG) emissions, yet at the expense of growing demand for critical metals. Here, we analyze the trade-off between the decarbonization potential of the road transportation sector and its critical metal requirement from the demand-side perspective in 48 major countries committing to decarbonize their road transportation sectors aided by electric vehicles (EVs). Our results demonstrate that deploying EVs with 40–100% penetration by 2050 can increase lithium, nickel, cobalt, and manganese demands by 2909–7513%, 2127–5426%, 1039–2684%, and 1099–2838%, respectively, and grow platinum group metal requirement by 131–179% in the 48 investigated countries, relative to 2020. Higher EV penetration reduces GHG emissions from fuel use regardless of the transportation energy transition, while those from fuel production are more sensitive to energy-sector decarbonization and could reach nearly “net zero” by 2040.


Model framework and key parameters
The overall model framework of this study is shown in Figure S1. This model consists of a (i) dynamic material flow-stock analysis (dMFA) module that aims to simulate the stock and flow of vehicles and associated material requirements and (ii) a carbon footprint module is used to assess the fuel-induced road transportation greenhouse gas (GHG) emissions.  Key parameters in this study include: (1) geographical scope and territorial division, (2) population, (3) gross domestic product (GDP), (4) vehicle ownership, (5) vehicle category; (6) electric vehicle (EV) and internal combustion engine vehicle (ICEV) market, (7) Battery and catalytic converter material intensity, (8) EV battery market, (9) EV lithium-ion battery (LIB) capacity, (10) Engine power, (11) vehicle and battery lifetime, (12) freight and passenger traffic volume, (13) transportation market, (14) energy transition and fuel emission, (15) material recycling. Key parameters and their assumptions are listed in Table S1.  The population from 2010 to 2050 is from the World Bank 2 .

Gross domestic product
The GDP of each country is obtained from the OECD 3 in Million USD at constant prices and Purchasing Power Parities (PPPs) of 2015, which is converted into 1995 level by the Purchasing power parity conversion factors from OECD 4 . Vehicle ownership Vehicle ownership per 1,000 capita is assumed to grow based on regional historical vehicle ownership levels, population, and GDP. A modified Gompertz model was used to simulate how GDP and population drive the vehicle ownership level 5 . Vehicle category Powertrain-based categorization: (i) Electric vehicle (EVs): battery electric vehicles (BEVs), plug-in electric vehicles (PHEVs), and fuel cell electric vehicles (FCEVs); (ii) internal combustion engine vehicles (ICEVs): gasolinebased ICEVs, diesel-based ICEVs, gasoline-based hybrid electric vehicles (HEVs), diesel-based HEVs, natural gas-based ICEVs. Transport mode-based categorization: (i) light-duty passenger vehicles (LDPVs): cars; (ii) light-duty commercial vehicles (LDCVs): vans and light-duty trucks; (iii) heavy-duty passenger vehicles (HDPV): bus; (iv) heavy-duty commercial vehicles (HDCVs): heavy-duty trucks.

Vehicle market
The EV market share is assumed to increase based on the prediction of four World Energy Outlook-2021 (WEO-2021) scenarios from the International Energy Agency (IEA) [6][7][8] , which considers regional differences in carbon mitigation ambitions and existing EV policies. Based on the four WEO-2021 scenarios (Stated Policies Scenario, Announced Pledges Scenario, Sustainable Development Scenario, Net Zero Emissions by 2050 Scenario) we created four new EV penetration scenarios, namely S40%, S60%, S80%, and S100%, which aims to realize the EV market stock of 40%-100% in 2050. The ICEV market share is assumed based on the BLUE Map scenario from IEA 9 . The share of LDV and HDV was obtained from the ANL 10 and the study 11 .

Material intensity
The material compositions of EV Li-ion batteries are calculated based on the BatPaC model version 3.1 12 and the past study 13 . The data on PGM loading of vehicle catalysts is derived from the study 11 . Battery market EV battery technologies of lithium nickel cobalt aluminum oxide (NCA) batteries, lithium nickel manganese cobalt oxide (NMC) batteries (NCA, NMC111, NMC523, NMC622, NMC811, and NMC955), lithium-sulfur (Li-S) batteries, and lithium-air batteries (Li-air) are considered. The market share of those battery technologies are calculated based on the study 13 . Battery capacity The LIB battery capacity of BEVs and PHEVs is modified based on the study 13,14 . The battery capacity of light-duty FCEVs is not considered 15 . The battery capacity of heavy-duty is assumed based on the GREET model 16 and the study 17 .

Engine power
The engine power (kW) of ICEVs and FCEVs is assumed based on the study 11 . Vehicle and battery lifetime The average lifetime of a vehicle is assumed to be 15 years 13,18,19 . The average lifetime Li-ion batteries for light-duty EVs is assumed to be the same as the EVs, 15 years, considering battery lifetime extension in the furture 13 . For heavy-duty BEVs, PHEVs, and FCEVs, lower battery lifetime leads to battery replacement 16 . The lifetime of use EoL battery as an energy storage system is assumed to be 10 years 20,21 . Weibull distribution is used to simulate the survival distribution of vehicles.

Transportation activities
The data on the freight (t·km) and passenger (p·km) transportation are collected from multiple sources. The historical freight transportation data is from the ITF 22 , Eurostat 23 , and NationMaster 24 . The historic passenger transportation data is from the ITF 22 , Eurostat 25 , and NationMaster 26 . The forecasted freight transportation data was collected from IEA 27 . The data on forecasted passenger transportation activities was estimated based on population growth.

Transportation market
The transportation market is assumed based on the vehicle stock from the dMFA model and the ITF 28 . ITF forecasted the share of passenger public (buses) and private (cars) transportation till 2050 28 . The last mile transportation by vans would account for 28 29 -35% 30 . Therefore, the share of long-haul transportation can be calculated, which is 65%-72%.

Energy transition and fuel emission
The energy transition is simulated by referring to the SSP2 scenario in the IAM IMAGE 3.2 1 . Three pathways under the SSP2 scenario, namely SSP2-RCP60, SSP2-RCP26, and SSP2-RCP19, are selected to model the energy transition. Fuel emissions in each scenario were modeled based on the prospective life cycle assessment modeling platform Brightway2 31 . The Ecoinvent 3.8 (cut-off) database, IMAGE 3.2 database 1 , and IMAGE-based inventory database PREMISE 1.2.6 32 are used.

Material recycling
The recycling rates for lithium, nickel, cobalt, and manganese are assumed based on the World Bank 33 . It is assumed that the recycling rate of lithium, nickel, cobalt, and manganese will reach 80% in 2030 34 . The recycling rate for PGMs is

Vehicle ownership
We represent the relationship between vehicle ownership per 1,000 capita and gross domestic production (GDP) per capita via a modified Gompertz Model 5 that considers temporal lags in the adjustment of the vehicle stock responding to income changes, as shown in Eq. (1). Where: , denotes vehicle ownership per 1,000 capita denotes vehicle saturation level , denotes gross domestic product per capita (PPP constant 1995 international $) denotes rising income adjustment coefficient (0 The vehicle ownership per 1,000 capita based on Eq.(1) is shown in Figure S2. The historic data from 1990-2019 is from the International Historical Statistics 35 and NationMaster 36 . The data from 2020 to 2050 was estimated by the modified Gompertz Model. We select Weibull distribution to demonstrate the lifetime distributions of vehicles. The Weibull random variables t and t ' are characterized by the shape parameter k and a scale parameter λ, as shown in Eq. (4).
where: Lv(t, t') denotes a probability distribution function that presents the probability that a vehicle manufactured in year t' < t will be demolished in year t. k is the shape parameter and λ is the scale parameter λ.

Recycling rate
The recycling rate of each critical metal in this study is shown in Figure S3. Based on the recycling potential of lithium, nickel, cobalt, and manganese 34 , we assumed their recycling rates will stabilize at 80% in 2030. The historic recycling rate of lithium, nickel, cobalt, and manganese was modified based on the World Bank's report 33 . The recycling rate of PGMs was modified based on the past study 11 .

Road transportation emission modeling
The road transportation emissions were estimated based on the transportation emission framework of the IPCC 38 . The total transportation emissions can be calculated through four decomposition factors: (i) system-infrastructure modal choice, (ii) fuel carbon intensity, (iii) energy intensity, and (iv) activity. We modified IPCC's transportation emission model and the GHG emissions of road transportation in a region in year y can be calculated via Eq. (5).
Where: denotes road transportation greenhouse gas emission denotes traffic volume denotes transportation market share denotes fuel intensity denotes fuel carbon intensity denotes passenger transportation denotes freight transportation denotes nation denotes year denotes vehicle category denotes fuel category

Critical metal requirement modeling
The critical metal requirement is estimated based on four decomposition factors: (i) annual sale of vehicle ( | , , , ), (ii) battery capacity/fuel cell power ( , , , ), and (iii) metal loading intensity ( , , , , ). The critical metal requirement of road transportation in a region in year t can be calculated via Eq. (6).
Where: denotes critical metal requirement denotes the amount of vehicle sales denotes battery capacity or fuel cell power denotes metal loading intensity denotes nation denotes year denotes vehicle category denotes fuel category denotes the type of required metal

EV stock projection
The forecasted EV stock is shown in Figure S4. Transport mode-specific EV stock of the S40%-S100% scenarios are shown in Figure S5- Figure S8. Hybrid electric vehicles (HEVs) are categorized as internal combustion engine vehicles (ICEVs) since the capacity of LIBs in HEVs is negligible (approximately 2 kWh).

ICEV stock projection
The forecasted ICEV stock is shown in Figure S9. Transport mode-specific ICEV stock of the S40%-S100% scenarios are shown in Figure S10- Figure S13.

Sankey diagrams
The Sankey diagrams of the S40%-S100% scenarios are shown in Figure S14. Each Sankey diagram starts with a secondary production layer from the left side, followed by a primary production layer, a region layer, a vehicle layer, and a battery layer. The Sankey diagrams below show to which department cumulative (2010-2050) critical metals are used.

PGMs requirement for EVs and ICEVs
The requirement of platinum group metals (PGMs) for internal combustion engine vehicles (ICEVs) and electric vehicles (EVs) is shown in Figure S15.

Sensitivity analysis for the future battery market
We established three extreme battery market scenarios-(i) NMC/NCA scenario, (ii) LFP scenario, and (iii) Li-air/Li-S scenario-to evaluate the sensitivity of the baseline battery scenario. It is assumed that the market shares of the NMC/NCA, LFP, and Li-air/Li-S battery will linearly increase to 100% under NMC/NCA, LFP, and Li-air/Li-S scenarios in 2050, respectively, as shown in Figure S16. The market share of NMC battery technologies in the four battery scenarios is shown in Figure S17.  The annual critical metal requirement under four battery market scenarios is shown in Figure S18. Based on the annual critical metal requirement under different battery market scenarios in Figure S18 and the population in each region, the critical metal use per kWh of battery is shown in Figure S19. The cumulative material requirements of battery-related metals (lithium, nickel, cobalt, and manganese) under the NMC/NCA scenario, LFP scenario, and Li-S/air scenario are shown in Figure S20- Figure  S22.    Figure S23a-e compares the annual demand and global gross production of each critical metal in 2020. Figure S23f-j compares the cumulative production and global reserves of each critical metal in 2020. The global lithium production in 2020 was estimated at 82.5 Gg, approximately 74% of which (61.1 Gg) was used for the battery industry. Australia is the primary lithium supplier and produced 39.7 Gg of lithium in 2020 (48% of the global production). While lithium production in China was around 13.3 Gg in 2020, the USA, Europe, and India are not major producers either. The global nickel production was around 2.5 Tg in 2020, and only 7% of that was used for the battery industry. China will contribute 43.4-47.1% to the gross nickel demand in 2050. Indonesia was the biggest producer, yielding 0.8 Tg (31%) of nickel in 2020. The nickel production in the two biggest nickel consumers, China and the USA, accounted for only 5% and 0.4%, respectively. Manganese production in 2020 was around 11 Tg and is sufficient for manufacturing EV batteries. In contrast, cobalt is more likely to encounter supply risks. The global cobalt production was just 0.1 Tg in 2020, and around 70% of production was from the Democratic Republic of the Congo. Approximately 40% of the global cobalt was supplied to the battery industry. The global PGM industry produced 0.4 Gg of PGMs in 2020, and 50% and 29% are from South Africa and Russia, respectively. Around 39% (approximately 0.1 Gg) of the production in 2020 was used in the vehicle sector. It is worth noting that reserves of the five critical metals have remained relatively stable over the last two decades 39 , while reserves may undergo major volatilities in the future based on new geological discoveries, technological advancements, etc.

Figure S23
Comparison of annual critical metal demands and annual production (a-e), and comparison of cumulative critical metal demands and reserves and resources (f-j). Note: Tg means teragram, and Gg denotes gigagram. "Recycling w/o 2 nd " indicates retired batteries are directly recycled without a second life as energy storage systems (ESS). "Recycling w/2 nd " denotes retired batteries are used with a second life as ESS before recycling. Note: the global resources and reserves data were collected from USGS 39 . The end-usage shares of lithium (74%) and cobalt (40%) production and reserves used for the LIB industry are assumed based on the data from Statista 40,41 . The end-usage share of nickel (11%) production and reserves used for the LIB industry are assumed based on the data from Nickel Institute 42 . The end-usage share of manganese (15%) production and reserves used for the LIB industry is assumed based on the data from Mordor Intelligence via https://www.mordorintelligence.com/industry-reports/manganese-market. The end-usage share of platinum group metals (36-42%) production and reserves used for the automotive sector is assumed based on the data from Mordor Intelligence via CME Group 43 . The recycling rates for lithium, nickel, cobalt, and manganese are assumed based on the World Bank 33 . The recycling rate for PGMs is assumed based on the study 11 . Figure S24 shows the production and reserve share of each critical metal in 2020. The data was collected from the United States Geological Survey (USGS) 39 . Australia (40 Gg), the Democratic Republic of the Congo (Congo) (95 Gg), and Indonesia (760 Gg) were the largest producers of lithium, cobalt, and nickel worldwide in 2020, respectively. South Africa produced the most manganese (5,200 Gg) and PGMs (190 tons) in 2020. Regarding the reserves, Chile and Congo owned the largest lithium and cobalt reserve in 2020, amounting to 9.2 and 3.6 Tg, respectively. For nickel, the largest three reserves in 2020 were located in Indonesia (21 Tg), Argentina (20 Tg), and Brazil (16 Tg). South Africa had the largest manganese and PGMs reserves in 2020, 520 Tg and 63 Gg, respectively.

Fuel emission intensity
Fuel production emissions Figure S28 shows the carbon footprint per kWh of electricity and per kg of fuel in 16

Biofuels blended in gasoline and diesel
The Share (by weight) of the bioethanol blended in gasoline in 16 regions under three energy transition scenarios is shown in Figure S29. The Share (by weight) of the biodiesel blended in diesel in 16 regions under three energy transition scenarios is shown in Figure S30. The premise v1.2.6 database 32 does not include the market share of hydrogen production approaches. The projected global hydrogen production market data from IMAGE 3.2 was used to represent the hydrogen production market in each region, as shown in Figure S31. From 2010 to 2020, over 80% of the hydrogen was produced based on fossil fuels such as coal gasification and methane auto-thermal reforming (ATM). Under the 3.5% scenario, grey hydrogen dominates the global hydrogen production market; and steam methane reforming (SMR) will dominate the hydrogen market since 2025 (80%) and will slowly decrease to 60% in 2050. In the more ambitious 2°C and 1.5°C scenarios, CCS technologies will be employed for coal gasification, ATM, and SMR for blue hydrogen production. The green hydrogen produced from water electrolysis and solar thermal is less than 1%, thus does not consider. Figure S31 Hydrogen production market in the three energy transition scenarios.

Fuel use emissions
The fuel emission factors of fuel are based on the European Union European Environment Agency's (EEA) Air pollutant emission inventory guidebook 2019 44 . Emission factors for CO2, CO, and N2O different road transport fuels are listed in Table S2 to Table S4.  Table S5. For H2 supplied to FCEVs, 1.5% mass loss (0.015 kg lost to air per kg of hydrogen supplied to vehicles) is assumed with 99 % confidence 46 .