Comparative sustainability assessment of lithium-ion, lithium-sulfur, and all-solid-state traction batteries

The most commonly used battery technologies for BEVs are LIBs (Dunn et al. 2021; Melin 2020). Over the past decade, various types of LIBs with different active materials have been deployed in BEVs. The prevalent positive active materials were lithium nickel manganese cobalt oxide (NMC) and lithium nickel cobalt aluminum oxide (NCA), whereas the formerly widespread lithium iron phosphate (LFP), lithium manganese oxide (LMO), and lithium cobalt oxide (LCO) only played a minor role (Dunn et al. 2021). This relies on the advantageous properties (e.g., specific energy, energy density, safety, lifetime, power, and costs) of NMC and NCA (Andre et al. 2015). Nevertheless, it is predicted that the market share of LFP will increase due to technological improvements (Dunn et al. 2021).

Since the aforementioned LIB types have already reached a high level of maturity, substantial further improvements regarding specific energy or energy density are not expected (Duffner et al. 2021b; Placke et al. 2017). This is contrary to the ambition of using lightweight and small traction batteries that enable long ranges over a long lifetime at moderate costs. Additionally, LIBs are associated with safety issues due to the limited operating temperature range and materials used. For example, a typical liquid electrolyte consists of flammable and volatile materials that react with water and oxygen (Zubi et al. 2018). Therefore, new battery technologies that enable a higher specific energy and energy density, as well as cause fewer safety issues, are required. Two promising technologies are LSBs and ASSBs (Duffner et al. 2021b; Placke et al. 2017).

Figure 1 depicts the main differences between LIBs, LSBs, and ASSBs. LSBs use sulfur instead of metal oxides as positive active material and lithium metal instead of graphite as negative active material. The properties of lithium and sulfur enable a higher theoretical specific energy. Furthermore, safety issues are improved due to a higher operating temperature range (Fotouhi et al. 2017; Manthiram et al. 2017; Zhang et al. 2018). However, the practical energy density is restricted due to the composition of the cathode and the lower nominal voltage of LSB cells compared to LIB cells (Cleaver et al. 2018; Fotouhi et al. 2017; Liu et al. 2020; Zhang 2013). Furthermore, the so-called shuttle effect limits the lifetime of LSBs (Cleaver et al. 2018; Fotouhi et al. 2017; Zhang 2013). Both aspects are in conflict with the goal of using small batteries in BEVs over a long lifetime.

The distinctive feature of ASSBs is the solid electrolyte, which also acts as separator. This enables the use of lithium metal as negative active material while using the same positive active materials as in LIBs and LSBs. Furthermore, the solid electrolyte is added to the positive active material to increase conductivity (Duffner et al. 2021b; Randau et al. 2020). Using solid electrolytes can increase the specific energy and energy density and prevent the shuttle effect of LSBs. ASSBs require fewer components and materials to manufacture the battery pack since no additional separator and cooling system are needed. In addition, the solid electrolyte can enhance the safety of the batteries (Gao et al. 2018; Janek and Zeier 2016; Manthiram et al. 2017; Randau et al. 2020).

However, ASSBs are also associated with challenges that must be resolved before commercialization. New production processes and a dry room environment are required because the materials used are susceptible to moisture, making up scaling challenging (Schnell et al. 2018, 2019).

The different characteristics and compositions of the battery technologies influence the environmental, economic, and social impacts (Aichberger and Jungmeier 2020; Barke et al. 2021; Duffner et al. 2020). A common approach to quantify these impacts is LCSA, which comprises individual assessment methods for each sustainability dimension, namely life cycle assessment, environmental life cycle costing, and social life cycle assessment (Finkbeiner et al. 2010; Kloepffer 2008; UNEP/SETAC 2011). Since the LCSA approach is used in this study to assess the three dimensions of traction batteries, the following literature review examines studies that use LCSA methods. Table 1 provides an overview of selected peer-reviewed studies dealing with the environmental and socio-economic impacts of traction batteries. Their contribution to the scientific literature is classified according to three criteria: sustainability dimension, life cycle scope, and investigated battery technology.

Most of the analyzed studies focus on one of the three sustainability dimensions, primarily environmental or economic. Only one study focuses exclusively on the social dimension, and seven studies consider two or more sustainability dimensions simultaneously.

The life cycle phases found within the scope of the analyzed studies comprise raw material extraction, battery production, use phase, and end-of-life. All investigated studies consider the first two phases. However, most studies focus on battery production while the raw material extraction is modeled with background data (e.g., from the ecoinvent database). The remaining phases are assessed only by few authors, and generic data is often used instead of a detailed assessment.

Finally, the right-hand part of Table 1 shows which battery technologies were assessed. Note that LIB-NMC includes all possible compositions of NMC, such as NMC111, NMC622, or NMC811, where the numbers describe the proportion of the active material components. The column other includes batteries usually not used in BEVs, such as nickel–cadmium batteries. LIB-NMC is the most often analyzed technology, often accompanied by another battery technology for comparison. For example, Deng et al. (2017) focus on the environmental assessment of LSBs, but they compare their results to a LIB-NMC111.

Only a few studies assess a wide range of battery technologies on a common basis. For instance, Peters and Weil (2018) use seven studies to create common LCIs and to ensure comparability. Nevertheless, they focus on the environmental dimension and LIBs, while LSBs and ASSBs are not considered. A wide range of battery technologies is also considered by Berg et al. (2015), Schmuch et al. (2018), and Wentker et al. (2019). However, these studies focus on the economic dimension. Furthermore, they do not calculate the cost at pack level so that costly materials and components (e.g., battery management system (BMS)) are not included. Barke et al. (2021) assess a wide range of battery technologies and consider all three sustainability dimensions. However, they omit ASSBs in their assessment, and there is a lack of data transparency, making it difficult to use this study for comparison.

Even if the data are presented transparently, it is challenging to compare battery technologies between different studies for several reasons. One reason is the different goals and scopes of the studies. For example, the definition of the functional unit has a decisive influence on the comparability of results due to incompatibility between functional units. Further influential parameters can be the definition of the scale and location of the production facilities due to the energy demand, country-specific electricity mix, or socio-economic conditions (Chordia et al. 2021; Thies et al. 2021; Wentker et al. 2019). Besides, the publication date influences the comparability. For example, the databases (such as the ecoinvent database or the Argonne National Laboratory’s GREET^® model) or the various impact assessment methods (such as ReCiPe, ILCD, and CML) evolve over time, leading to differences in the results and thus to significant differences in indicator scores (Chordia et al. 2021). Lastly, different impact indicators are reported due to varying classification and characterization approaches across the impact assessment methods, making it challenging to compare the results.

These aspects highlight the need for a transparent, consistent, and holistic sustainability assessment of current and prospective battery technologies, which will be carried out next.

The presentation of the environmental impact assessment results focuses on four impact categories, namely climate change (CC), human toxicity (HT), mineral resource depletion (MRD), and photochemical oxidant formation (POF). These impact categories were selected due to their high relevance in the context of batteries. The indicator scores per kWh battery pack capacity are depicted in Table 3. The results of the remaining impact categories are depicted in supporting information S3.

In terms of HT, MRD, and POF, the impact scores of all ASSBs are lower than those of LIBs and LSB. Only for CC, the ASSBs using NCA, LFP, NMC622, and NMC811 emit more GHG per kWh battery pack than the corresponding LIBs (e.g., the ASSB-NMC811 has a higher indicator score than the LIB-NMC811). This can be explained by the higher electricity consumption for producing these batteries. For the other impact categories, the electricity demand for battery production has a minor impact so that the higher specific energy and the resulting lower mass of the ASSBs show their advantageousness.

The battery with the lowest indicator scores in all four impact categories is the ASSB-LSB for mainly two reasons. First, the ASSB-LSB has the highest specific energy since this battery requires fewer and lighter materials. Consequently, the corresponding emissions related to the production of these materials are lower. Secondly, the materials used in the ASSB-LSB and LSB are associated with lower impacts. This is especially reflected by the bar for the positive active material in Fig. 4, which shows that the environmental impacts seem negligible in the case of sulfur. In contrast, NCA, LFP, NMC622, and NMC811 have a much stronger influence on the indicator scores.

However, the ASSB-LSB shows different hotspots that vary by impact category. For CC, battery production is responsible for 57% of the impact. This depends mainly on the electricity generation in Germany, which has a high share of fossil sources. The electricity generation also has the strongest influence in terms of the unit process negative active material (lithium), which has a share of 18% of the emissions. Considering HT, battery production and the unit process negative active material (lithium) have smaller shares of 28% and 4%, respectively. Instead, the unit processes negative current collector and cell container have a high influence with shares of 37% and 20%, respectively. These high shares are related to the copper production chain. Similar causes are also found for MRD. However, battery production is responsible for only 2% of the emissions in this impact category, while the negative current collector and the cell container account for 45% and 24%, respectively. In addition, the CSC is responsible for 14% of the indicator scores. For this impact category, this is also mainly due to the copper production chain, with additional impacts related to mining other metals for the production of the sensors. Compared with HT and MRD, the negative active material (lithium) has a more decisive role for POF, with a share of 22%. This is related to electricity generation in China. Since electricity generation plays a role regarding the emission of NMVOC and equivalent substances, battery production is responsible for 27% of the emissions. However, since electricity generation in Germany emits fewer NMVOC equivalents per kWh, the effect is smaller compared to CC. The negative current collector and cell container have a share of 15% or 12%, which shows that the copper production chain has a more minor influence.

For the socio-economic assessment, four impact categories are investigated in detail: total battery cost for the economic dimensions, risk of child labor (RoCL), risk of corruption (RoC), and risk of forced labor (RoFL) for the social dimension. These impact categories were selected due to the strong influence of batteries on the cost of a BEV and social concerns associated with the battery supply chain. The calculated indicator scores per kWh battery pack capacity are depicted in Table 4.

The socio-economic assessment results show similarities to the environmental assessment results. Most notable, the indicator scores decrease when a solid electrolyte substitutes the liquid electrolyte. Furthermore, Fig. 5 shows that battery production has a strong influence with shares from 28 to 45% for the total battery cost. In contrast, the shares for the social impact categories related to battery production are neglectable, as the risk of a socially disadvantageous situation concerning the impact categories investigated in Germany is low.

The battery with the lowest indicator values is the ASSB-LSB, with the exception of RoCL, where the ASSB-LFP has a slightly lower indicator score. In general, the difference between the ASSB-LFP and the ASSB-LSB is small in terms of the social impact categories. The relatively low indicator scores of the ASSB-LSB is a consequence of the much lower influence of the positive active material, which is, e.g., responsible for 48% of the medium risk hours of RoC in the case of the ASSB-NMC622. At the same time, there is virtually no risk associated with the production of the positive active material of the ASSB-LSB.

However, there are some hotspots related to the ASSB-LSB. Battery production accounts for 43% of the total battery cost, which is affected particularly by personnel and energy costs as well as the depreciation of machinery and buildings. The remaining costs depend on the material costs. In the case of the ASSB-LSB, especially two components have a strong influence: CSC and the negative active material (lithium). These two components are responsible for 21% and 10% of the total battery cost, respectively. Furthermore, the costs associated with these two components are much higher than the other ASSBs, resulting from the number of battery cells required for the battery pack.

In comparison, the CSC and the negative active material (lithium) are also the two components responsible for the most medium risk hour equivalents regarding the social impact categories. However, the indicator scores of RoCL, RoC, and RoFL are much higher for the CSC, which is reflected by the respective shares. For example, CSC has a share of 36% of the medium risk hour equivalents of RoCL, while the negative active material (lithium) has a share of 15%. This is largely due to the higher risk of child labor in China’s electronic sector compared to Chile’s mining sector.

Sections 4.1 and 4.2 show that the electricity mix and the used materials are important drivers of the LCSA results. Therefore, this section examines two different scenarios. The first scenario investigates how an electricity mix of 100% renewables for battery production affects the indicator scores. The second scenario analyzes the consequences of changing the value added of battery materials.

To modify the electricity mix of the production facility in Germany, a new unit process for the electricity mix is modeled based on a prediction by Jacobson et al. (2017). The exchanges related to the economic and social dimensions remain the same, i.e., there are no changes in the socio-economic indicator scores. Consequently, Fig. 6 illustrates the results of the four environmental impact categories. The figure shows the indicator scores assigned to component and battery production for the base case and the case with an electricity mix of 100% renewables. Overall, the impacts of CC, HT, and POF decrease, while the impacts of MRD increase slightly. The increase for the latter depends on the impacts of building renewable energy plants. However, the increase is smaller than 3%. The observed effect is larger for the former three impact categories. Particularly for CC, the impact decreases between 26 and 48%. Especially the GHG emissions related to ASSBs can be reduced by using electricity from 100% renewables due to the high electricity demand for battery production. Consequently, almost all ASSBs have a lower indicator score than the respective LIBs with a liquid electrolyte, contrary to the base case.

The second scenario analysis is intended to provide insight into how sensitive the total battery cost is if the value added of the key cathode and anode materials changes, which could be the case if, for example, the price of electricity or wages in a particular country increase. For this purpose, the value added of eleven unit processes of the main anode and cathode materials is doubled each. The results and the change compared to the base case are shown in Table 5. For the negative active materials, it can be observed that the change of the value added of lithium metal has a higher influence on the total battery cost than graphite. In particular, in the case of ASSB-LSB and LSB, there could be a significant change, depending on the high percentage of lithium metal in these batteries. In the case of positive active materials, the largest changes are associated with cobalt sulfate and nickel sulfate, affecting the total battery cost with NCA and NMC. Doubling the value added of sulfur, manganese sulfate, or iron sulfate has a negligible effect on the total battery cost, showing a possible advantage of using these materials. In addition, it can be inferred that ASSBs are less sensitive to changes in the value added of aluminum and copper since fewer of these materials are needed in the cases studied.