Definitions and system boundary
As shown in Fig. 1, the stocks and flows of PEAMs in the automotive market, which contains the critical metals Li, Co, Ni and Mn, were considered totally circulated domestically in Japan through an assumption of a 100% recycling rate. These critical metals were thought to only come from overseas mining and extraction (primary supply) and domestic recycling and recovery (secondary supply). The spent LIBs were directly sent to the recycling system or used as stationary batteries for energy storage, with no export or other diversion. Here, the vehicle fleets defined in this study included passenger cars but excluded buses and trucks because of the strong competition from the fuel cell vehicle industry. Also, the LIBs included the ones served in battery electric vehicles (BEVs) and plug-in hybrid electric vehicles (PHEVs). In the boundary of LCA, the remaining parts of LIBs were considered to be newly manufactured in all scenarios, and the material losses (not caused by technical defects) in mining, extraction, manufacturing and recycling were not considered. The energy storage market was assumed to unconditionally accept the spent LIBs if the policy promoted this, of which the capacity would not exceed the market projection of stationary energy storage.
Historical changes in the stocks and flows of critical metals
Assuming the precise separation using a high-voltage pulsed discharge method will be applied from 2025, a dynamic MFA approach47 was developed to quantify the cycle of critical metals from 2025 to 2050. The approach was annually iterated to simulate the changes in the passenger car market, lifetime use of vehicles and LIBs, secondary use, and recycling of spent LIBs, as usually applied48. First, the passenger car ownership per household \(C\) was calculated as shown below:
$$C=\frac{N·\gamma ·(1-\eta ·\delta )}{1+\alpha ·\text{e}\text{x}\text{p}(-\beta ·\frac{\stackrel{-}{I}}{\stackrel{-}{P}})}$$
1
where \(N·\gamma ·(1-\eta ·\delta )\) represents the potential passenger car market, negatively affected by the multiple of average annual income per household \(\stackrel{-}{I}\) and average purchase price per vehicle \(\stackrel{-}{P}\). \(N\) is the number of households in Japan, \(\eta\) is the rate of membership of car sharing, and \(\delta\) is the decrease rate of car ownership by car sharing49. Parameters \(\alpha\) and \(\beta\) were set as 10.5 and 0.79150, respectively. Parameter \(\gamma\) was estimated as 0.000757 by the average fitted value because of the passenger car ownership rate during 2015–2020. The results showed that the ownership of passenger cars would gradually decrease from 61 million to 46 million during 2020–2050 mainly because of the decrease in population (Supplementary Fig. S3, S5).
Then, according to the ambitious targets carried out by the Japanese government and automotive associations, the sales of internal combustion engine vehicles will be stopped domestically by the early 2030s, while the transport sector should be decarbonized by 205051. Thus, we assumed that the ICEs will gradually quit the new sales market by 2035 and that the HEVs will gradually quit the new sales market by 2050, as shown in Supplementary Fig. S6. The scenario result (Supplementary Fig. S7) based on the assumption was close to the scenario ‘BEV75’ estimated by the Japan Automobile Manufacturers Association52. Regarding the projection of battery model changes, we referred to the high-probability scenario (named ‘NCX’) established by Xu et al.3 (Supplementary Fig. S8). The battery producers were assumed to replace Co with Ni-rich PEAMs to reduce costs. Thus, the battery model NMC111 will gradually change to NMC523, NMC622, NMC811 and NMC955. The NCX scenario proposed that although lithium iron phosphate (LFP) has advantages in technology readiness, costs and resource availability and lithium–sulphur (Li–S) and lithium–air (Li–Air) batteries have specific energy densities three times greater than NMC LIBs, they are more difficult to use than NMC LIBs because of the low specific energy of LFP and low technology readiness of Li–S/Li–Air batteries.
In 2023, the sales of BEVs and PHEVs were only 3.6% of the passenger car market53. The timing of collecting spent LIBs can affect both the lifespan of EVs and the service life of automotive LIBs in EVs (Fig. 1). To estimate the lifespan of vehicles, we conducted the following Weibull distribution equation:
$$F\left(t\right)=1-{e}^{-{\left(\frac{t}{}\right)}^{m}}$$
2
where \(F\left(t\right)\) is the survival rate of a vehicle when it services \(t\) years. Parameters \(m\) and µ are calibrated as 2.69 and 15.49, respectively, according to the age distribution of existing passenger cars surveyed in 201954 (Supplementary Fig. S9, S10). For easier dynamic simulation, we assumed all types of passenger cars have the same survival rate and that the lifespan does not extend yearly because of the launch of new cars. The lifespan of automotive LIBs was set as 8 years, and the lifespan of spent LIBs for secondary usage in stationary energy storage was set as 10 years, referred to in many studies such as those by Hara55 and Zeng et al.5.
Prospective LCA and scenarios setting
Based on the scaled-up experiment of the high-voltage pulsed discharge method applied to separate the PEAMs56, we conducted a prospective LCA approach57 to quantify the possible environmental impact of applying precise separation to recycle PEAMs while considering the uncertainty in evaluating such an emerging technology58. We set eight scenarios to simulate the different situations considering four technology combinations (landfill after incineration, pyrometallurgy and hydrometallurgy combination, hydrometallurgy after precise separation, or direct recycling after precise separation) and two choices in the timing of recycling spent LIBs (immediate recycling, or recycling after secondary use) (see Table 1).
Table 2 summarizes the details of processes considered in the cases defined in Table 1. In Cases a-1 and b-1, spent LIBs are incinerated and landfilled without metal resources recycling. In Cases a-2 and b-2, spent LIBs are first disassembled into cells that are then roasted and smelted for metal recovery. In Cases a-3 and b-3, spent LIBs are disassembled into cells in which the PEAMs are then separated by high-voltage pulsed discharge for metal recovery. In Cases a-4 and b-4, the PEAMs are separated by high-voltage pulsed discharge and directly incorporated for new automotive LIBs production.
Table 2
Production and recycling processes defined in cases
Processes in cases in Table 1 | Case a-1, b-1 | Case a-2, b-2 | Case a-3, b-3 | Case a-4, b-4 |
Mining and extraction | 〇 | 〇 | 〇 | 〇 |
M-SO4 production | 〇 | 〇 | 〇 | 〇 |
Raw PEAM production | 〇 | 〇 | 〇 | 〇 |
PEAM production | 〇 | 〇 | 〇 | 〇 |
Battery production | 〇 | 〇 | 〇 | 〇 |
Use in vehicle | 〇 | 〇 | 〇 | 〇 |
Collection/dismantling | 〇 | 〇 | 〇 | 〇 |
Incineration | 〇 | | | |
Landfilling | 〇 | | | |
Battery pack dismantling | | 〇 | 〇 | 〇 |
Roasting | | 〇 | | |
Smelting | | 〇 | 〇 | |
Battery cell dismantling | | | 〇 | 〇 |
Pulsed discharging | | | 〇 | 〇 |
Recycling of metal resources | | 〇 | 〇 | 〇 |
Based on the scaled-up analysis in the previous study56, we further considered the annual changes in impact assessment in the target of realizing a carbon neutral society by 205059. In the fiscal year 2022, grid electricity used in Japan emitted 0.435 kg-CO2/kWh60 on average, which will be reduced to 0.25 kg-CO2/kWh by 2030 and net zero by 2050 because of the official projection of energy demand and supply61. Excluding incineration and landfilling, pyrometallurgy, hydrometallurgy, dismantling, separation, and other processes will be gradually decarbonized by 2050. However, while decarbonization may bring about more resource consumption62, optimistically, we assumed the following technological innovation can suppress this additional resource consumption (Supplementary Table S8). Regarding the lack of data for LCA on different types/sizes of automotive LIBs, we assumed the intensity of GHG emissions and resource consumption potential is proportional to the weight of PEAMs in a standard LIB pack (per vehicle). Accordingly, the LCA on LFP, Li–S, and Li–Air battery production and recycling as well as the secondary use of spent LIBs was out of the scope of this work.
Limitations and uncertainties
Applying an integrated model, our simulation and scenario analysis build on various assumptions and parameters that cannot avoid correlated limitations and uncertainties (Table 3). First, the EV market and stationary energy storage market in Japan may not optimistically follow the government’s ambition and institutional projections. A lower popularization rate of EVs and LIBs will de-level the simulation and analysis results in our study. Here we tested the influence on critical metals’ circulation if the energy storage market does not accept spent LIBs or is infinite to accept them (Supplementary Fig. S13). Second, the lifespan of passenger cars and automotive LIBs was fixed, but it may gradually extend in the future. Of course, compared with our simulation, the recycling of spent LIBs will slow down and the annual production of LIBs will increase in proportion. Third, the battery capacity required for EVs was assumed to be fixed in the future as the reference set3 (Supplementary Table S5). Although the capacity for one EV will increase in the future, the increment of energy intensity in the new battery model will increase to offset the capacity requirement. Fourth, the collection rate of spent LIBs and the actual recycling rate were set as 100% for easier simulation. In the near future, all the automotive LIBs may have their own codes in the tracking system, while the recycling rate may approach 100%63. As a sensitivity analysis, the collection rate was found to mainly affect Cases a-1 and b-1 in terms of life-cycle GHG emissions and affect Cases a-4 and b-4 in terms of potential resource consumption because of the obvious differences in parameter on circulation and resource consumption potential (Supplementary Figs S14, S15; Supplementary Tables S9, S10). Fifth, the battery models such as LFP, Li–S and Li–Air were not considered because of the scenario adopted from Xu et al.3. In fact, all-solid-state batteries have advantages in application to light-duty EVs, while LFP will have a large share in heavy-duty EVs and stationary energy storage64. In the long term, various batteries will share the market, for which we need to verify the performance of precise separation considering different structures in the cathode active materials65. Sixth, in LCA, we set NMC111 as the reference standard56 and conducted a sensitivity analysis accordingly56. An LCA study in China showed that the LFP battery, NMC battery and LMO battery (28 kWh) would produce GHG emissions in the production of 3,061 kg CO2-eq, 2,912 kg CO2-eq and 2,705 kg CO2-eq, respectively66; in addition, these results vary across countries67. Seventh, transport distance and volume were fixed in LCA, but it could be quite reduced because of the application of precise separation.
Table 3
Description and assumption of key model parameters
Key parameters | Description/assumptions | Details |
Passenger car ownership | Passenger car ownership is assumed to be affected by population68, the number of households69, annual income per household, consumer price of vehicles62, and the participation of car sharing49. Road infrastructure, public traffic systems, and urban shape are considered fixed parameters. The import and export of passenger cars are out of the scope of this paper. | Supplementary Tables S1, S2 Supplementary Figs S3–5, S9–11 |
EV market share | The EV market share is assumed to increase according to the Carbon Neutral Target declared in Japan59. The sales of internal-combustion engine vehicles will be stopped from the year 2035, and the proportion of EVs (almost BEVs) in new car sales linearly increases to 90% by 2050, while the proportion of Fuel cell vehicles reaches 10%52. The service life of vehicles is assumed to remain the same as in 201970. | Supplementary Tables S3, S4 Supplementary Figs S6, S7 |
Battery capacity of BEV/PHEV | The battery capacity for BEVs and PHEVs is set as the sales-weighted average based on the projection of small, mid-size, and large car fleets3. | Supplementary Table S5 |
Stationary energy storage market | The capacity of the stationary energy storage market is set not to surpass 25% of the capacity of automotive LIBs in use in EVs37. | Supplementary Table S13 Supplementary Fig. S12 |
Battery lifetime | The lifetimes of automotive LIBs in use in EVs and stationary energy storage are assumed to be 8 and 10 years, respectively3,5. | Supplementary Table S6 |
Battery model changes | The model change of the automotive batteries in Japan is assumed to keep the same path of the worldwide market, following the ‘NCX’ scenario of Xu et al.3. | Supplementary Fig. S8 |
Recycling rate and residual loss | The collection rate of spent automotive LIBs is assumed to be always 100% during 2025–2050. The weight of critical metals included in the PEAMs are set as the same as those set by Xu et al.3. Residues are not in consideration. | Supplementary Table S7 |
Life-cycle GHG emissions and RCP factors | Except for the emissions from waste combustion, life-cycle GHG emissions are assumed to decrease in line with the decarbonization of the power generation sector in Japan61. Life-cycle RCP is assumed not to change in the future. | Supplementary Table S8 |