Introduction

The electric mobility transition is a pivotal strategy for achieving global climate targets1,2,3,4,5, as electric vehicles (EVs) emit nearly zero operational emissions compared to internal combustion engine vehicles (ICEVs). However, this shift depends on increased mineral extraction6,7, which also leads to environmental challenges such as soil erosion, deforestation, and ecosystem destruction8,9. During the extraction process, attention is primarily focused on recovering critical minerals (e.g., cobalt for batteries3,10). In contrast, the associated mine wastes—known as “hidden flows”—are often overlooked, despite typically exceeding the useful resource by hundreds of times. Generally, hidden flows include topsoil overburden, waste rock, and tailings11,12,13,14, which contribute substantially to natural degradation. For instance, at least 50 gigatons (Gt) of carbon have been released into the atmosphere due to soil removal since the industrial revolution15. This amount is roughly equivalent to global greenhouse-gas emissions in 202216 and has disrupted the global carbon cycle17,18,19. Moreover, waste rock and tailings cover about 1 million km2 of the Earth’s surface20. They have exposed 23.48 million people and 5.72 million livestock to toxic pollutants20, and have driven 46.5% of tropical forest loss between 2000 and 201921, corresponding to foregone carbon sequestration of approximately 3.3 Gt CO₂ yr−1 based on average tropical forest removals of 7.1 Gt CO₂ yr−122.

In addition to their substantial quantity and environmental burden, hidden flows also rarely cross national borders, unlike critical minerals. Consequently, their environmental impacts remain concentrated in resource-supplying countries. As the electrification of the automotive sector continues3,10,23,24, the volume of hidden flows is projected to rise dramatically, further exacerbating the inequitable distribution of environmental burdens and placing an increasing strain on these nations. Moreover, declining ore grades25,26,27,28,29 and the development of new mines30 in recent decades have generated enormous amounts of mining waste, leading to severe natural degradation and aggravating existing ecological issues. However, the magnitude, spatial distribution, and effective management of hidden flows remain insufficiently examined. This hinders efforts to develop sustainable mitigation strategies. Therefore, to achieve a fair and sustainable e-mobility transition, quantifying hidden flows is an essential prerequisite.

The quantification of hidden flows in the e-mobility transition can be achieved through diverse approaches, including substance flow analysis (SFA)31, economy-wide material flow accounts (EW-MFA)32,33,34,35, and input-output-based material footprint (IO-MF) analysis36,37,38. Each method offers distinct advantages; for example, SFA provides process-by-process resolution, EW-MFA considers multi-resource flows, and IO-MF captures hidden flow footprints (Supplementary Fig. 1). Recent research has integrated these insights to develop a new indicator, the Total Material Requirement (TMR)14,39,40,41, which accounts for both extracted materials and associated hidden flows across the entire product life cycle. It tracks these flows along the supply chain, including raw material extraction, material processing, product manufacturing, and international trade13,24,42. Initial research on TMR primarily focused on specific mined materials (e.g., iron ore and coal)41,43,44. More recently, studies have expanded to cover critical products (e.g., batteries and vehicles)13,42 and industrial sectors (e.g., automobile industry and electricity sector)7,24, from both global7,24 and national perspectives13,42.

Despite such expanding research on TMR for mining and vehicle production, several critical gaps remain. First, the mining database used for TMR quantification is often outdated (i.e., data from the early 2000s) and lacks spatial resolution (i.e., based on a global average)41,43,44. Given recent shifts in mining operations (e.g., mining techniques, ore grades) due to intensive mining activities, there is an urgent need to update these databases with improved spatial and temporal granularity. Second, while raw material suppliers have been extensively studied, the broader international trade network along the e-mobility supply chain has often been overlooked. In particular, the transfer of hidden flows across different stages of the entire supply chain remains significantly neglected13,42,45. Third, existing studies have primarily focused on high-income economies13,42 (as classified by the World Bank46 based on gross national income per capita) or global averages7,24, failing to account for low- and middle-income economies (under the same classification) such as China, the world’s most important TMR contributor33,47. These omissions have resulted in an underestimation of hidden flows and their associated environmental impacts, emphasizing the necessity to re-examine and update TMR assessments of e-mobility transition with high resolution.

China plays an increasingly critical role in the global supply chain as the top producer of vehicles (33% of the global capacity)48 and passenger cars49. To achieve its climate goals, the Chinese government has actively promoted the e-mobility transition through policy support, technological innovation, and capital investment3,50,51. The rapid expansion of China’s passenger car market is expected to further accelerate domestic mining activities52,53, with significant global implications across the supply chain7,13,43. Quantifying these waste flows along China’s e-mobility supply chain is essential for both resource-supplying and e-mobility manufacturing countries, as it provides crucial insights for developing sustainable resource management strategies and mitigating environmental impacts.

This study aims to quantify hidden flows through TMR assessment using China as a case study. We develop a life-cycle-based TMR quantification framework (Fig. 1a) and compile a global mine-site-specific database (Fig. 1b) to support high-resolution TMR assessments across mine sites, minerals, and products. We uncover the multidimensional characteristics of hidden flows in China’s domestic passenger car supply chain and in their global transfer across the world. The results can provide insights into accurate hidden flow assessment, inform strategies to mitigate the impacts of hidden flows, and could contribute to improving resource efficiencies in the global e-mobility transition.

Fig. 1: System definition and mine-site-specific database.
figure 1

a TMR quantification framework for the passenger car supply chain. b Map of the global distribution of selected mineral types adopted in our calculation. ICEV internal combustion engine vehicle, BEV battery electric vehicle, PHEV plug-in hybrid electric vehicle, FCV fuel cell vehicle, LFP lithium iron phosphate, NMC lithium nickel manganese cobalt oxide, NCA lithium nickel cobalt aluminum oxide, Fe iron, Al aluminum, Cu copper, Ni nickel, Li lithium, Pt platinum, Zn zinc, Mg magnesium, Co cobalt, Mn manganese, Pb lead. The locations of 10 types of metallic minerals and coal are shown in Fig. 1b as the four dimensions of data (ore grade, mining waste, production, and location) of those minerals are available at the mine level. Each individual operation is plotted as a single circle on the map, with the circle size indicating the TMR coefficient. The percentages represent the share of the total production of mine sites with China’s total consumption in 2019. TMR coefficients of mines, ranging from 0 to 17 tonnes (t) per t, are plotted as a bar plot. Map generated using ArcGIS Pro 3.2.0 (Esri) with the Monochrome basemap. Source: Esri, HERE, Garmin, FAO, NOAA, USGS, © OpenStreetMap contributors, and the GIS User Community.

Full size image

Results

Figure 1a shows the life cycle TMR quantification framework, which traces both useful and waste flows of various materials across multiple regions54,55. Five transformation processes for vehicle production are included: mining, beneficiation, smelting and refining, fabrication, and manufacturing and assembling. We estimated the TMR coefficients (the material requirement intensity from the lithosphere for a specific mine-site/mineral/product11,33,47,56) for 1784 mines and 1126 products, and the TMR across 25 resource-supplying countries in China’s passenger car supply chain. The TMR coefficients for ore, semi-product, and vehicle indicate the material extracted from lithosphere to produce 1 t of ore or semi-product, or one unit of vehicle. The national TMR reflects the total material extracted to support China’s passenger car production in each resource-supplying country (Supplementary Fig. 3).

An NEV requires over three times the material extraction of an ICEV

For most minerals, the TMR coefficient typically rises along the passenger car supply chain due to ongoing investment in mineral resources, leading to the accumulation of hidden flows. Based on the TMR coefficient of associated global mines (Supplementary Fig. 4) and products (Supplementary Data 7), we estimate the average TMR coefficient along the Chinese passenger car supply chain (Fig. 2a). In the initial stages of the supply chain, TMR coefficients decrease in some cases, such as from bauxite (7 t per t) to Al concentrate (5 t per t), mainly due to lower coefficients for imported Al concentrate. In the mining stage, Pb ore has the largest TMR coefficient (9 t per t) among the 12 metal minerals, followed by bauxite (7 t per t) and Zn ore (7 t per t) due to the significant waste rock generation during mining. The TMR coefficients in the beneficiation, smelting, and refining processes are heavily influenced by ore grade. For instance, producing 1 t of Pt catalyst for China’s passenger car industry requires 501,045 t of material, with Co sulfate (1105 t per t) and Cu wire (651 t per t) also having high TMR values due to low ore grades. Downstream, significant material inputs are required for semi-products: 548 t for 1 t of powertrain system for FCVs due to Pt usage, 347 t for 1 t of NMC111 battery production due to Co sulfate and Cu anode usage, and 148 t for 1 t of transmission system in BEVs, PHEVs, and FCVs due to the high Cu usage (see Supplementary Fig. 8 for TMR coefficients of each product).

Fig. 2: TMR coefficients, material inputs, and material efficiencies along the passenger car supply chain in China.
figure 2

a TMR coefficients of minerals and products by lifecycle process. b Direct material inputs for each vehicle. c Hidden material flows for each vehicle. Material efficiency by material (d), process (e), and vehicle (f) type. CoSO4 cobalt sulfate, Li2CO3 lithium carbonate, MnSO4 manganese sulfate, NiSO4 nickel sulfate. In a, relative position on the scale encodes the TMR coefficient magnitude.

Full size image

NEVs generally have higher TMR coefficients than ICEVs. Among all passenger cars, BEVs with NMC111 batteries require the highest material input from the lithosphere, requiring nearly 159 t per unit, while ICEVs require 48 t per unit. Although one NEV accounts for 93–104% of the direct material flows for one ICEV, it generates 183–330% more hidden flows (Fig. 2b, c). This is due to greater consumption of critical minerals, such as Li, Co, and Ni in Li-ion batteries and Pt in fuel cell stacks. While these minerals make up only 0.4–9.5% of the direct material flows in NEVs, they contribute 6–41% of the hidden flows per vehicle. Consequently, Li-ion batteries significantly increase the TMR coefficient. In BEVs, batteries account for 26–28% of TMR, while in PHEVs, they contribute 39–64%. Among battery types, NMC-series and NCA batteries generate 40–101% more hidden flows than LFP batteries. Fuel cell stacks in FCVs also contribute significantly, accounting for 42–43% of hidden flows. Another major factor is Cu usage. Compared to ICEV, one NEV requires 0.04–0.09 t more Cu per vehicle for battery and electrical systems, resulting in an additional 28–51 t of hidden flows per NEV produced (see Supplementary Fig. 5 for detailed material flows by vehicle type).

High hidden flows inevitably result in low material efficiency. Material efficiency refers to the ratio of material used to the supplied material (see “Methods”). In the passenger car sector, Pt, Co, and Cu exhibit low average material efficiency, with values of 0.0002%, 0.1%, and 0.2%, respectively (Fig. 2d). This is primarily due to their low ore grades, ranging from 0.0001 to 0.0007% for Pt, 0.02 to 0.48% for Co, and 0.21 to 4.84% for Cu among all considered mines. Conversely, Mn, Fe, and other materials (plastic, rubber, and glass) have higher material efficiency (15.5%, 6.8%, and 16.3%) due to the lower mining waste and higher ore grades of Mn and Fe, or the reliance on nonmetal minerals like silicon for glass. BEVs, particularly those with NMC111 batteries, have the lowest material efficiency (0.9%) among all passenger cars owing to the reliance on Li-ion batteries (Fig. 2f). PHEVs exhibit higher material efficiency (1.4–1.6%) than other NEVs (BEVs: 0.9–1.4%, FCVs: 1.0–1.1%) owing to their smaller Li-ion battery weight and absence of fuel cell auxiliaries. Notably, the extensive use of coal contributes to lower material efficiency in the refining and smelting (25%) as well as manufacturing and assembling (31%) processes (Fig. 2e). See Supplementary Fig. 7 for material efficiency across different material types and production processes.

Only 3% of extracted materials enter China’s car sector

Figure 3 shows the total material flows of China’s passenger car supply chain, calculated by multiplying TMR coefficients by car production (see “Methods” for details). In 2019, the whole Chinese passenger car production system prompted the extraction of 1163 Mt of materials from the lithosphere. Of this total, 57 Mt of minerals were mined, of which 31 Mt were ultimately used in passenger car production. This resulted in 1106 Mt of hidden flows, bypassing the passenger car industry and accumulating as mining waste. Notably, these hidden flows exceeded the valuable resources by 35 times, resulting in a meager total material efficiency of 3% for China’s passenger car industry. The TMR for China’s passenger car industry was driven primarily by Cu production (37%), followed by Fe (28%), Al (16%), Pb (7%), and Pt (6%).

Fig. 3: Total material flows along China’s passenger car supply chain in 2019.
figure 3

Units are million metric tonnes (Mt) for all values. The material flows starting from the left represent the total amount of material that needs to be mined from the lithosphere to support China’s passenger car production. The dark-colored flows (represented by black numbers) indicate direct material flows, while the relatively light-colored flows (represented by gray numbers) represent the accompanying hidden flows. Dark gray flows show metal losses in each process, whereas light gray flows represent hidden flows that do not enter the car sector.

Full size image

Material requirements for passenger car production are notably influenced by material efficiency and vehicle demand. For instance, despite Co sulfate’s low material efficiency in BEVs with NMC batteries, Co products contributed only 1% to the total materials mined for China’s passenger car industry, due to the low market share of BEVs with NMC batteries (1.5%)10,57. The cumulative TMR of Li, Ni, Mg, Zn, and Mn used in China’s passenger car industry totaled 9.7 Mt, representing only 1% of this industry’s TMR. This was due to their relatively low direct materials consumption (0.3% of the total materials used in a passenger car) and low TMR coefficients. Fe products, although pivotal in car production and making up 72% of the final product weight, accounted for only 28% of the TMR in this sector because of their low TMR coefficients (13 t per t for steel and 10 t per t for cast Fe). Processes such as beneficiation and refining generated significant waste streams, accounting for 36% and 34% of the total mining waste in the entire lifecycle production process, respectively. This was attributed to beneficiation removing gangue from raw ore to produce a higher-grade concentrate, while the refining process required substantial energy resource input, such as coal, whose production also incurs large mining waste.

Half of hidden flows occur outside China

In 2019, with total material extraction of 605 Mt within China, the TMR for China’s passenger car industry was driven primarily by coal production (49%) for energy generation, followed by Cu (16%), Fe (12%), Pb (9%), and Al (8%) (Fig. 4a). This pattern was attributed to China’s substantial import of raw materials, with the energy-intensive refining process taking place domestically. China shouldered 52% of the material extraction cost for its passenger car production. As a country that both extracts resources and consumes them domestically, China faced a dual reality. While the nation relies on domestic resources, the TMR coefficient of certain ores in China exceeded that in other countries owing to the lower quality of China’s mineral reserves, which are characterized by relatively low ore grades (e.g., Fe, Cu, and Mn) and high strip ratios (e.g., Al, Pb, and Zn). For instance, producing 1 t of bauxite in China requires, on average, the extraction of 6.7 t of materials, compared with 1.2 t in Indonesia, 1.4 t in Guinea, and 1.8 t in Australia. This difference is attributed to the thicker overburden in China’s bauxite mining, which requires the removal of more waste rock. Further details are provided in Supplementary Data 7, which highlights that producing the same weight of minerals in China requires more extensive resource extraction and results in higher mining waste generation compared to other countries.

Fig. 4: Multi-regional TMR associated with China’s passenger car supply chain in 2019.
figure 4

a Twenty-five countries and regions with the largest TMR by mineral type for China’s passenger car production. b TMR induced by China’s passenger car sector and 12 vehicle types by region. ARG Argentina, AUS Australia, BRA Brazil, CAN Canada, CHL Chile, CHN China, COD Democratic Republic of the Congo, GAB Gabon, GHA Ghana, GIN Guinea, IDN Indonesia, KAZ Kazakhstan, MEX Mexico, MMR Myanmar, MNG Mongolia, MYS Malaysia, PAN Panama, PER Peru, PHL Philippines, RUS Russian Federation, TUR Türkiye, USA United States of America, ZAF South Africa, ZMB Zambia, OTH Other regions. Map generated using ArcGIS Pro 3.2.0 (Esri) with the Monochrome basemap. Source: Esri, HERE, Garmin, FAO, NOAA, USGS, © OpenStreetMap contributors, and the GIS User Community.

Full size image

Materials sourced globally contributed to China’s passenger car production through international trade, resulting in substantial hidden flows generated on a global scale (Supplementary Fig. 6). Other countries bore 48% of the TMR for China’s passenger car production, with the primary contributors being Chile (10%), Australia (7%), Peru (6%), and South Africa (5%). China heavily relied on imports, especially Fe from Australia, Cu from Chile and Peru, and Pt from South Africa. The substantial amount of imported Fe and the high TMR coefficients for Cu and Pt contributed to the remarkable hidden flows beyond China’s borders in those regions. Meanwhile, China’s production of NEVs resulted in greater TMR  abroad compared with that for ICEV production (Fig. 4b). Other countries bore 57% of the TMR for one unit of ICEV production in China and 60–63%, 62–64%, and 74% of the TMR for one unit of production of BEVs, PHEVs, and FCVs, respectively. It can be inferred that certain regions experience an increase in hidden flows and heightened nature disturbance in China’s future transition to e-mobility. Examples include Democratic Republic of the Congo supplying Co for NMC-series batteries, Chile and Peru providing Cu for control systems, and South Africa contributing Pt for fuel cell stacks for FCVs.

Discussion

Our study compiles a mine-specific database and developes a life-cycle-based TMR quantification framework to assess hidden flows along China’s passenger car supply chain. Based on a global mining database covering 1784 mines, we find that 1163 Mt of total materials were extracted from the lithosphere for China’s passenger car production in 2019, of which 57 Mt were directly extracted materials and triggered 1106 Mt of hidden flows. These hidden flows were 35 times larger than the minerals ultimately used, with only 3% of extracted resources entering the car sector. Notably, 48% of these hidden flows occurred outside China, primarily in Chile, Australia, Peru, and South Africa, highlighting the unequal distribution of resource extraction impacts. Furthermore, our framework, which covers 1126 products along the vehicle production lifecycle, reveals a disproportionate distribution of hidden flows across minerals, products, and processes. Although critical minerals in NEVs (Li, Co, and Ni in Li-ion batteries and Pt in fuel cell stacks) constituted only 0.4–9.5% of direct material flows, they accounted for 6–41% of hidden flows per vehicle. As a result, NEV production generated over three times the hidden flows of an ICEV. Throughout the entire lifecycle, Cu accounted for only 2% of direct material inputs but contributed 37% of total hidden flows, while energy-intensive processes drove 34% due to their heavy reliance on coal consumption.

These results update the magnitude and spatial patterns of hidden flows along the vehicle supply chain, offering a comprehensive perspective for their efficient management. Compared to previously reported TMR coefficient of 62–70 t per car for NEV by Kosai et al.13, our analysis identifies higher values of TMR coefficient ranging from 95 to 159 t per car, primarily driven by declining global ore grades. For instance, the global average Cu ore grade decreased from 0.84% in the early 2000s41 to 0.64% in 2019, increasing energy inputs and hidden flows during the beneficiation process to produce Cu concentrate of comparable quality. Moreover, substantial generation and transfer of hidden flows between nations are uncovered by the supply chain perspective. Approximately half of the hidden flows associated with China’s passenger car industry occur abroad, revealing an uneven spatial distribution that previous global TMR quantifications (e.g., Watari et al.22,24) did not reveal. Furthermore, in contrast to Kosai et al.11, which considered a single battery type, our results demonstrate significant differences among battery technologies. Specifically, an NMC-series or NCA battery generates 40–101% more hidden flows than an LFP battery. This high-resolution decoupling of hidden flows among different battery technologies provides more possibility for mitigating hidden flow impacts and improving material efficiency during the global transition to e-mobility.

The large scale of hidden flows not only indicates lower material efficiency but also contributes to significant environmental impacts. Extraction exposes deeper soil to oxygen, causing the oxidation of organic carbon stored in the soil and resulting in CO2 release58. Based on an estimated emission of approximately 4.6 t of CO₂ per 100 t of soil59, the extraction of 1163 Mt of materials for China’s passenger car production could release around 53 Mt of CO2 from soil carbon. These previously unreported emissions are comparable to the annual emissions of an upper-middle-income country such as Peru (58 Mt in 2019)60, potentially exacerbating global climate change. However, the overall environmental burden also depends on factors such as mining waste quality61,62, extraction methods63, and remediation efforts64, which also need to be considered in the assessments of mining impacts. Nevertheless, the TMR coefficient can serve as an intermediate indicator that extends beyond resource extraction to inform assessment of factors such as carbon emissions, surface disturbance, and tailings generation. By decomposing the TMR amount across minerals, products, and countries in our life-cycle-based framework, the responsibility of hidden flow generation is tracked like a carbon footprint. Since the current regulations mainly focus on energy use or carbon emissions65, integrating the TMR coefficient into Sustainable Development Goal 12 (responsible consumption and production)66 and into the life cycle assessment (LCA) framework67 as an additional dimension to incorporate “embodied” hidden flows per specific functional unit (e.g., unit ore, battery, or car) could incentivize more materially efficient product design.

The material efficiency of China’s passenger car production system is only 3% based on our estimation, with the remaining extracted materials generated as hidden flows. To enhance material efficiency, we analyze TMR coefficients for 1126 products and all production processes, identifying key contributors to hidden flows generation—including Cu, NMC-series and NCA batteries, and energy-intensive processes—and evaluate targeted mitigation strategies based on our high-resolution framework (Supplementary Fig. 9). By recycling and substituting secondary Cu for primary Cu, which currently contributes to 37% of the sector’s hidden flows, the total hidden flows from China’s passenger car production could be reduced by 35%. Among Li-ion batteries, NMC-series and NCA batteries exhibit lower material efficiency than LFP. Replacing them with LFP batteries reduced hidden flows by 3% in 2019; this strategy is expected to play a significant role in the projected e-mobility transition, potentially achieving a 12% reduction by 2050. Additionally, energy-intensive processes like smelting and refining contribute over one-third of total hidden flows, and transitioning to clean energy could reduce hidden flows by 13–14%. However, challenges remain in improving the Cu recycling system68, enhancing the energy density of LFP batteries69,70, and advancing energy storage technologies71, all of which will require cross-sector collaboration to balance decarbonization goals with mitigating hidden environmental burdens.

About half of these hidden flows generated from China’s passenger car production occur abroad, and this burden is expected to grow in resource-supplying countries like Chile and Peru (Cu), South Africa (Pt), and Democratic Republic of the Congo (Co) as China’s e-mobility transition proceeds. Collaboration along the supply chain is essential for mitigating hidden flows and improving material efficiency. Upstream raw-material suppliers (e.g., China, Chile, Australia, Peru, South Africa, and Democratic Republic of the Congo) should enhance mining data transparency and shift away from high-TMR mines. For example, our mine-level analysis in Chile shows that Centinela Oxide Mine, with a higher TMR coefficient (7.2 t per t), produces 1.5 times more Cu ores than Carmen de Andacollo Mine, which has a lower TMR coefficient (1.49 t per t) but 1.3 times more Cu reserves. This provides evidence for resource-supplying nations to reduce hidden flows by prioritizing extraction from low-TMR, high-reserve mines, enabling more informed decision-making. Along the supply chain, sourcing minerals from countries with higher material efficiency could substantially reduce hidden flows. For instance, bauxite extracted in Indonesia has a TMR coefficient that is only 20% of that for Chinese bauxite. However, decisions on resource sourcing are influenced not only by environmental considerations but also by geopolitical dynamics, strategic concerns over resource security, and broader economic interdependencies. These factors must be carefully balanced when designing sustainable and resilient supply chains. Moreover, implementing ecological taxes72,73 on raw material commodities in international trade could help compensate for the negative environmental externality of resource extraction. Downstream manufacturers (e.g., China, Japan, and Germany) are accountable for responsible sourcing and promoting more efficient product designs. Incorporating embodied hidden flows into products and adopting sustainable practices—such as Cu recycling, battery substitution, and the clean energy transition—are particularly critical in these nations. Only such collaborative initiatives can effectively minimize the globally pervasive degradation of assets caused by resource utilization and inaugurate a new path toward the sustainable transition to e-mobility.

Methods

System definition

Passenger car production (Fig. 1a) includes five transformation processes: mining, beneficiation, smelting and refining, fabrication, manufacturing, and assembling. Losses at each transformation stage are calculated, including the traditional metal losses as well as materials extracted from the lithosphere that do not enter the product itself, such as the topsoil overburden, waste rock, and tailings. This TMR quantification framework considers 12 types of passenger cars, including ICEVs, BEVs with four battery types (LFP, NMC111, NMC811, NCA), PHEVs with three battery types (LFP, NMC111, NMC811), and FCVs with four battery types (LFP, NMC111, NMC811, NCA). International trade related to China’s passenger car production includes 28 countries, consisting of 25 resource-supplying countries and 3 manufacturing countries (Supplementary Data 5), and covers 188 traded commodities (Supplementary Data 6) that are incorporated to quantify hidden flows in a global context. In the end, 1126 products along China’s passenger car supply chain, from mining to assembly, as well as domestic and imported products, are accounted for during the calculation of TMR coefficients.

Mining database

This global mine-site-based database includes 11 metals (Fe, Al, Cu, Ni, Li, Pt, Zn, Mg, Co, Mn, Pb), 3 energy resources (coal, natural gas, oil), and 13 nonmetal minerals (granite, sand, silica stone, limestone, dolomite, feldspar, fluorspar, bentonite, rock, calcite, clay, phosphate rock, other dimension stone). Four types (ore grade, mine waste, production amount, and location) of data for 10 metals and coal are available at the mine level; their locations are illustrated in Fig. 1b. For other minerals (e.g., Mg), we used national average data or typical mineral data to calculate their TMR coefficient at the mineral level (Supplementary Data 1 and 7). Data for 1784 specific mine sites were compiled from the US Geological Survey74, the S&P Global Market Intelligence database75, mining company reports, government yearbooks, and websites. The year 2019 was used as the base production year to calculate the TMR because it was the most recent year with sufficiently complete data.

For minerals with data collected from specific mine sites, we calculated their data coverage using Eq. (1), expressed as a percentage of China’s total consumption of these resources, as illustrated in Fig. 1b.

$${D}_{{{{\rm{i}}}}}=\frac{{\sum }_{m=1}^{n}{o}_{{{i}},{{m}}}}{{T}_{{{i}}}}$$
(1)

where \({D}_{i}\) is the data coverage of mineral \(i\), \({o}_{{{{i}}},{{{m}}}}\) is the quantity of mineral ore \(i\) produced at mine site \(m\), and \({T}_{{{{\rm{i}}}}}\) is the total consumption of mineral ore \(i\) in China.

Total material requirement coefficient

The TMR coefficient can be calculated for a single operation or aggregated to any desired level, including mine, mineral, and product levels. The decision tree used for data selection and calculation is provided in Supplementary Fig. 2.

Total material requirement coefficient at the mine level

TMR coefficients \(x\) for each mine was calculated using the following equation42:

$${x}_{{{{\rm{m}}}},{{{\rm{i}}}},{{{\rm{c}}}}}=\frac{{r}_{{{m}},{{i}},{{c}}}}{{o}_{{{m}},{{i}},{{c}}}}+1$$
(2)

Where, for mineral type \(i\) from mine \(m\) in country \(c\), \(r\) is the quantity of waste rock removed, \(o\) is the quantity of ore mined. A mine usually produces more than one type of mineral as several minerals geologically occur together in inseparable form and hence necessarily be mined and processed together76,77. Based on the data availability, we considered the Pb-Zn symbiosis mine and allocated the waste rock generated by each mine to Pb and Zn.

Total material requirement coefficient at the mineral level

The TMR coefficients were calculated for the 11 minerals in resource-supplying countries that support China’s passenger car production, as shown in Eq. (3). To calculate the TMR coefficient at the mineral level for each country, we categorized the countries into two types based on their trade dependence: resource-supplying and resource-consuming countries (see Supplementary Data 5). We assumed that the ores produced by each mine are uniformly mixed in the market. For the resource-supplying countries (e.g., Australia, South Africa, Brazil), the TMR coefficient was calculated as the weighted average of the TMR coefficients of their domestic mines, as shown in Eq. (3). For the resource-consuming countries (e.g., Japan, Korea, Germany) that mostly rely on imports, we calculated their TMR coefficient by calculating the weight average TMR coefficient of their imported countries.

$${x}_{{{i}},{{c}}}=\sum\limits_{m=1}^{n} \, {x}_{{{m}},{{i}},{{c}}} * \frac{{o}_{{{m}},{{i}},{{c}}}}{{o}_{{{i}},{{c}}}}$$
(3)

where \({x}_{{{{\rm{i}}}},{{{\rm{c}}}}}\) is the TMR coefficient of mineral ore \(i\) in country \(c\), \({o}_{{{{i}}},{{{c}}}}\) is the total amount of mineral ore \(i\) produced in country \(c\).

Total material requirement coefficient at the product level

We employed the iterative life cyle inventory (LCI) method to determine the TMR coefficient of the final products (passenger cars). Ecoinvent67 and MiLCA78 databases, as well as literature reviews providing LCI data, were used to estimate material inputs for each product. First, the input quantities of semi-products (e.g., battery, body parts, powertrain system) for the production of various car types were estimated. Subsequently, the metal/energy/nonmetal product (e.g., steel, Al alloys, electricity, silicon) inputs for the production of these semi-products were also quantified. This sequential process enables us to trace the inputs back to ascertain the quantity of ores required to manufacture the final product, as shown in Fig. 1a. The TMR coefficient calculations for each product is given by Eq. (4). In this study, we adopted the LCA cut-off rule to select material inputs in each inventory and trade flow; auxiliary flows and material input quantities smaller than 5% were excluded79.

$${x}_{{{p}},{{c}}}=\sum\limits_{i=1}^{n}{x}_{{{i}},{{c}}} * {q}_{{{i}}}$$
(4)

where \({x}_{{{{p}}},{{{c}}}}\) is the TMR coefficient of product \(p\) in country \(c\), \({q}_{{{{i}}}}\) is the quantity of mineral ore \(i\) required to manufacture product \(p\).

Total material requirement

Domestic total material requirement

The TMR for China’s passenger car industry is the cumulative amount of material extracted from the lithosphere to produce all passenger cars within the country, as calculated in Eq. (5).

$${{TMR}}_{{car}}(t)=\sum\limits_{l=1}^{12}{x}_{{car},{{{l}}}} * {p}_{{car},{{l}}}(t)$$
(5)

where \({{TMR}}_{{car}}(t)\) is the total amount of material extracted for supporting China’s passenger car production in year \(t\), \({x}_{{{\mathrm{car}}},{{{l}}}}\) is the TMR coefficient of passenger car type \(l\), \({p}_{{car},l}(t)\) is the production amount of passenger car type \(l\) in year \(t\), here \(t=2019\). The key parameters for battery chemistries and passenger cars in China are provided in Supplementary Data 2, 3 and 8.

Transfer total material requirement

In resource-supplying countries, materials mined for supporting China’s passenger car production along the supply chain were quantified using Eq. (6).

$${{TMR}}_{{{c}}}(t)=\sum\limits_{i=1}^{n}{x}_{{{i}},{{c}}} * {e}_{{{i}},{{c}}}(t)$$
(6)

where \({{TMR}}_{{{{\rm{c}}}}}(t)\) is the total amount of material mined in country \(c\) to support China’s passenger car production in year \(t\), \({e}_{{{{\rm{i}}}},{{{\rm{c}}}}}(t)\) is the amount of mineral ore \(i\) exported to China from country \(c\) in year \(t\), here \(t=2019\).

Material efficiency

Material efficiency refers to the ratio of materials used in a process or industry to the amount of material originally extracted from the lithosphere. We calculated the material efficiency of each process (\({e}_{{{{n}}}}\)), product (\({e}_{{{{p}}}}\)), and industry (\({e}_{{{{d}}}}\)) using the following equations.

$${e}_{n}=\frac{{m}_{{out},{{n}}}}{{m}_{{in},{{n}}}}$$
(7)
$${e}_{p}=\frac{{m}_{{{p}}}}{{m}_{{{m}}}}$$
(8)
$${e}_{d}=\frac{{m}_{{out},{{d}}}}{{m}_{{in},{{d}}}}$$
(9)

where \({m}_{{{out}},{{{n}}}}\) and \({m}_{{{in}},{{{n}}}}\) are the total material (including used and unused flows) output and input for process \(n\), respectively. \({m}_{{{{p}}}}\) is the amount of useful material that finally enters product \(p\), \({m}_{{{{m}}}}\) is the initial materials extracted from the lithosphere to produce product \(p\). \({m}_{{{out}},{{{d}}}}\) and \({m}_{{{in}},{{{d}}}}\) are the total material (including used and unused flows) output and input for industry \(d\), respectively.

Reduction potential

Based on the findings of this study, we estimated the TMR reduction potential through the replacement of NMC-series batteries and NCA batteries with LFP batteries, transitioning the current electricity system to clean energy sources (i.e., hydropower, solar, wind, biomass, and nuclear), and substitution of primary Cu with secondary Cu. In each scenario, we assumed that NMC-series and NCA batteries are completely replaced by LFP batteries, fossil fuel energy is entirely replaced by clean energy sources, and primary Cu is fully substituted by secondary Cu. The reduction potential of TMR for the entire passenger car sector was calculated using Eq. (10).

$$r=\frac{{{TMR}}_{{car}}\left(t\right)-{{TMR}}_{{car}}\left(t\right)^{\prime} }{{{TMR}}_{{car}}(t)}$$
(10)

Where \({{TMR}}_{{{\mathrm{car}}}}(t)^{\prime}\) represents the total amount of materials mined to support China’s passenger car production under the changes in the three-parameter mentioned above.

Limitations and sensitivity analysis

Our findings must be interpreted in the light of limitations. For data coverage, despite our best efforts to achieve high coverage for each mineral, some minerals still have low coverage (e.g., Pb at 44%) due to limited data availability. Moreover, in accordance with the LCA ISO standard, we adopted a 5% cut-off line, meaning that materials with a mass share of less than 5% of total weight were excluded from the LCI76. Because the impacts of these excluded flows were not quantified in this study, their effects remain uncertain. This limitation should be addressed in future work. Sensitivity analysis results of TMR and material efficiency of China’s passenger car sector, as well as the TMR of one car, are shown in Supplementary Fig. 10 and 11. We varied five key parameters (battery market share, battery capacity, car market share, car body weight, and car production) at 20% increments above and below the set value; further details and results are provided in Supplementary Data 4.