Applied Energy 307 (2022) 118150 Contents lists available at ScienceDirect Applied Energy journal homepage: www.elsevier.com/locate/apenergy Integration of raw materials indicators of energy technologies into energy system models Laura Talens Peiró a, *, Nick Martin a, Gara Villalba Méndez a, b, Cristina Madrid-López a a Sostenipra (2017SGR1683), Institut de Ciència i Tecnologia Ambientals (ICTA-UAB), Universitat Autònoma de Barcelona (Unitat Excelencia Maria de Maetzu MDM CEX2019-000940-M), 08193 Cerdanyola del Vallès, Barcelona, Spain b Department of Chemical, Biological and Environmental Engineering, Catalan Biotechnology Reference Network - XRB, Universitat Autònoma de Barcelona (UAB), Campus UAB, 08193 Bellaterra, Barcelona, Spain H I G H L I G H T S • Four environmental indicators suitable for use in energy system models are proposed (85) • The indicators allow direct comparison of decarbonisation potential of renewables (84) • Gearbox double-fed induction generators are the preferred form of wind turbine (80) • Cadmium telluride photovoltaics exhibit best overall raw materials outcomes (78) A R T I C L E I N F O A B S T R A C T Keywords: Raw materials and their related environmental impacts will play a key role in the implementation of renewable Renewable energy energy infrastructures for decarbonization. Despite the growing amount of data quantifying raw materials for Life cycle assessment energy production technologies, few examples of these data sources are being included in current energy system Material metabolism models. Accordingly, this paper introduces possible pathways for integrating material-specific life cycle assess­ Energy transition Decarbonisation ment outputs and material metabolism indicators into energy system models so that raw material requirements, Material supply and their associated impacts, can be accounted for. The paper discusses the availability of life cycle inventories, impact assessment methods and important output indicators. The material metabolism indicators most relevant to the current policy debate surrounding the European Green Deal–namely, material supply risk and contribution of recycled materials to total supply–are also discussed alongside the value of adding this information to energy system models. A methodology for using data from both approaches is offered and operationalised using four sub-technologies of both wind turbines and solar photovoltaic panels as case studies. The results show that considerable variation exists between and within the two groups for all indicators. The technologies with the lowest global warming potential, cumulative energy demand and supply risk are turbines with gearbox double- fed induction generators and cadmium telluride photovoltaics. Furthermore, wind turbines exhibit significantly higher recycling rates than photovoltaics. Ultimately, the integration of such methodologies into energy system models could greatly increase the awareness of raw material issues and guide policies that maximise compati­ bilities between resource availability and cleaner energy systems. and for Europe to become the first carbon neutral economy by 2050. As stated in the plan, around 75% of greenhouse gas (GHG) emissions in the 1. Introduction European Union (EU) are generated by the production and use of energy [2] and, in order to decarbonise the EU, it is crucial to increase the share The European Green Deal is the latest response by the European of low carbon technologies in the generation and use of this energy. Commission (EC) to climate and other environmental related challenges Emissions relating to energy generation can be systematically [1]. Its key objective is to decouple economic growth from resource use, * Corresponding author at: Sostenipra (2017SGR1683), Institut de Ciència i Tecnologia Ambientals (ICTA-UAB), Universitat Autònoma de Barcelona (Unitat Excelencia Maria de Maetzu MDM CEX2019-000940-M), 08193 Cerdanyola del Vallès, Barcelona, Spain. E-mail address: laura.talens@uab.cat (L. Talens Peiró). https://doi.org/10.1016/j.apenergy.2021.118150 Received 14 May 2021; Received in revised form 27 October 2021; Accepted 29 October 2021 Available online 9 November 2021 0306-2619/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). L. Talens Peiró et al. Applied Energy 307 (2022) 118150 Nomenclature IR import reliance kg kilogram Abbreviations kWh kilowatt-hour a-Si amorphous silicon LCA life cycle assessment C-Si crystalline silicon LCI life cycle inventory CdTe cadmium telluride LCIA life cycle impact assessment CED cumulative energy demand LDS low demand decarbonisation scenario CIGS copper indium gallium diselenide MDS medium demand decarbonisation scenario CO2 carbon dioxide MFA material flow analysis CRM critical raw material MJ megajoule CTUh comparative toxicity unit for humans MSA material system analysis DD direct drive MW megawatt DFIG double-fed induction generator PMSG permanent magnet synchronous generator EC European Commission PV photovoltaic EESG electrically excited synchronous generator SO2 sulphur dioxide ELCD European Platform on Life Cycle Assessment SR supply risk EOL-RIR end-of-life recycling input rate UNEP United Nations Environment Programme EROI energy return on investment WGI world governance indicator ESM energy system model yr year EU European Union EV electric vehicle Notation GB gearbox ci annual consumption level in EU of material i [kg/year] GHG greenhouse gas EOL-RIRi end-of-life recycling input rate of material i [%] GLAD Global Life Cycle Assessment Data Access EOL-RIRtechnology net end-of-life recycling input rate of the GLAM Global Life Cycle Assessment Method technology under study [%] GWP global warming potential mi mass of material i contained in the technology under study HDS high demand decarbonisation scenario [kg/MW] IAM integrated assessment model n number of individual materials in the technology under ICEV internal combustion engine vehicles study IEDC Industrial Ecology Data Commons SRi Supply risk of material i [dimensionless] IMAGE Integrated Model to Assess the Global Environment SRtechnology net SR of the technology under study [year/MW] assessed by the methodology known as life cycle assessment (LCA). LCA securing the supply of raw materials within its borders [15]. The report evaluates the environmental burdens stemming from a process by listed borates (batteries), lithium (batteries), natural graphite (batte­ considering the entire life cycle of the process under study using a ho­ ries), niobium (magnets), silicon metal (PV), and a mix of diverse rare listic perspective [3]. Results from an LCA are given as environmental earth elements (batteries and magnets) as materials with 100% import impact categories, among them the global warming potential (GWP) reliance [15]. measured in terms of greenhouse gas generation in carbon dioxide As the dependencies on raw materials for the development of low equivalent (CO2-eq) units. Blanco et al. point out that most studies that carbon energy technologies become more evident, the need to include have used LCA in combination with ESM to date have been done ex-post them as a variable in energy system models is being acknowledged. to account for the potential environmental impacts of specific technol­ However, to date, only a small number of energy system models (ESMs) ogies, specific sectors or at a global level under different policy scenarios consider environmental impacts. One of the most renowned integrated [4]. These studies have tended to use the environmental impact cate­ assessment models (IAMs) used to model energy systems is IMAGE gories described in the ReCiPe [5] or Impact 2002+ [6] assessment [16,17], a large-scale, ecological-environmental model framework that methods to account for potential environmental and human health im­ simulates the environmental consequences of human activities world­ pacts. However, mineral resource depletion is not included in some cases wide. It addresses some of the most prominent environmental issues and due to data uncertainty on recycling rates and material balances [7]. sustainability challenges such as climate change, land-use change, Energy supply is addressed from a holistic perspective in the Euro­ biodiversity loss, modified nutrient cycles and water scarcity. However, pean Green Deal, thus potential dependencies on resources key to in the latest version (IMAGE 3.0), raw material use is not included. reaching the EU goals also need to be addressed. Indeed, one of the EU’s Raw materials have been addressed in the MEDEAS model created major fears appears to be the shift from a fossil fuel to a materials- within the framework of the EU H2020 project MEDEAS (https://www. dependent economy. To avoid this situation, and to identify materials medeas.eu/model/medeas-model). The model contains seven sub­ that may potentially become problematic in coming decades, the EC has modules including one that models material requirements [18]. This produced several reports addressing the use of so-called critical raw submodule accounts for the materials needed for energy infrastructure materials (CRMs). The EU considers CRMs to be materials with high and the energy related to its manufacturing, expressed as an energy importance to the union’s economy (e.g., lithium for electric mobility) return on investment (EROI). Using this approach, the model assesses and with a potentially high risk regarding their supply [8]. Since 2010, the implications that mineral depletion may exert on energy transitions the EC has reviewed and updated the list of CRMs for the EU every three in relation to potential mineral supply constraints. The demand of years [9-12]. The methodology relating to CRMs has been also revised minerals is compared with their currently estimated level of geological and formally presented together with an extensive guideline document availability (reserves and resources) for the qualitative detection of risks [13]. The reports published by the EC highlight the material needs for of material supply from a global perspective. These concepts are now growing technologies, especially for renewables and electric mobility being further developed within the EU H2020 project LOCOMOTION [14]. In 2020, the EC presented the European Raw Materials Alliance for (https://www.locomotion-h2020.eu) which aims to develop further 2 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 capabilities for the materials module based on geological supply. In any natural resources. Impact categories can refer to a single-issue such as case, neither project is assessing material metabolism factors beyond the the cumulative energy demand, or to multiple issues as in the commonly physical quantities that exist; geopolitical and other risk factors relating used ReCiPe method [22] which includes 21 indicators. In most multiple to material supply between countries are not considered. issue LCIAs, the emissions and consumption of resources are attributable So, although some efforts have been made to assess raw materials to three main areas of protection (ecosystem quality, human health and using a holistic perspective, and several new indicators have been natural resources), which are preceded by several impact indicators that defined [1411], their use in ESMs remains limited. This is partially express the impact on the environment as midpoint and/or endpoint because a systematic process for collecting and providing such infor­ indicators [23]. Midpoint indicators represent the actual environmental mation in an ESM-usable format is yet to be developed. phenomena caused by the life cycle system, such as ‘global warming Accordingly, the present paper proposes the use of LCA data along­ potential’ (CO2-eq) [24] and ‘ozone depletion potential’ (kg CFC-11-eq) side material metabolism approaches as a way of providing a more [25], whereas endpoint indicators are composites that result from a complete picture of the raw materials use and associated environmental combination of midpoint indicators that reflect the damage on so-called impacts within ESM processes. The paper begins by briefly explaining areas of protection [26]. For example, in the ReCiPe method the the LCA methodology and how it could be used to integrate the potential midpoint indicators for ‘global warming potential’ and ‘ozone depletion’ environmental impacts of energy production into ESMs. It also discusses are combined into the endpoint indicator ‘damage to human health’ existing LCA data and outlines a simple methodology for creating [22]. environmental impact indicators for energy infrastructures per unit of Fig. 1 provides a simple conceptualisation for a potential integration power capacity. Section 3 then investigates existing material meta­ of LCA and ESMs by illustrating where inputs and outputs to these bolism information and how this information can complement LCA re­ models are situated in relation to the four stages of the LCA framework. sults. It expands the methodology further to include material supply Again, the most relevant stages for ESMs within this framework involve parameters for supply risk and recycling rates. The article provides, for the LCI and LCIA calculations. Assessing the LCI includes the back­ the first time, the results of a set of four environmental indicators readily ground (green) and the foreground (blue) systems. As always, the usable in ESM to support the assessment of wind turbines and solar background system refers to all processes needed to supply the raw photovoltaic cells. It concludes by confirming the potential of such ap­ materials and energy to the processes of the foreground system. In the proaches, the need for further integration of environmental and meta­ case of ESMs, the foreground system refers to the processes needed to bolic data into ESMs and, to aid future policy decision-making, the manufacture a specific energy technology. For example, for a solar ongoing need for good quality life cycle and material supply data. photovoltaic panel the foreground system would include the processes for manufacturing the cells and the frame and the balance of system 2. Potential contribution of LCA methodology to ESM (wiring, switches, mounting system, solar inverter, battery bank and charger), transport and assembly of all components, operation and The LCA methodology is used for evaluating the environmental maintenance, use, dismantling and transport and all waste disposal burden of a process by accounting for the inflow and outflow of mate­ operations. rials and energies alongside the wastes released to the environment [3]. Fig. 1 also displays the most common multiple issue methods used in Such evaluations are undertaken using a holistic perspective that con­ LCIAs (in dark blue). According to Jungbluth [27], the most frequently siders the entire life cycle of the process under study. LCA accounts for used LCIA methods are CML2002 [3], ILCD2010 [26], ReCiPe 2016 the inflows and outflows of the system from ‘cradle to grave’; that is, [22], the EU Product Environmental Footprint 2018 [28] and from the extraction, manufacturing, consumption and recycling to the ImpactWorld+ [29]. Each of these methods includes a list of impact final disposal. The methodology can be divided into four steps: goal and categories that differ in scope and procedure for characterisation and scope definition, inventory analysis, life cycle impact assessment (LCIA) weighting. In 2016, the United Nations Environment Programme and interpretation. Once the objective and functional unit are defined as (UNEP) launched a consultation for the creation of a Global Life Cycle part of the goal and scope stage, an inventory analysis is done to quantify Assessment Method (GLAM) with the objective of identifying scientifi­ the raw material and energy inputs, and to account for the atmospheric cally robust and applicable methods [30]. Discussions within the scope emissions, waterborne emissions, solid wastes and other releases over of GLAM have led to a prolific number of papers discussing LCA in­ the entire life cycle of a product, process or activity. dicators, especially regarding resource use and availability indicators Each product system inventoried in this stage can be divided into [31,32]. Another important aspect highlighted in the figure is the po­ both foreground and background systems [19]. The foreground system tential contribution of background systems to the overall environmental refers to the main process steps and infrastructure related to the focused impacts of renewable energy supply processes [33]. For example, the product or system of the study. Meanwhile, the background system is potential environmental impacts of wind generation are largely influ­ comprised of the processes needed for the supply of raw materials and enced by the mix of electricity supplied to the production of raw ma­ energy to the foreground system. This generally includes the more terials, mostly steel, concrete and aluminium, which are highly energy dominant processes outside of the study’s focus and are typically out of intensive [34]. the direct control of those undertaking the assessment [20]. Commonly, the background system’s infrastructure (e.g., the manufacturing of the 2.1. Inputs from LCI power plant or the fossil fuels production infrastructure) is included in the secondary data sets used for modelling the background system. The 2.1.1. The background system background system deals with almost all material and energy flows Two types of inputs are considered for background systems: raw going to and coming from the foreground system. Data for the back­ materials and energy. The most common source of data for the assess­ ground system is typically taken from existing databases (e.g., ecoinvent ment of raw materials is the fee-paid ecoinvent database [33]; version v3.7.1 [20], GaBi [21]) while the foreground system can often be 3.7.1 of ecoinvent includes LCIs for the production of about 30 metals, quantified using primary data from case studies, peer-review papers and 20 types of industrial minerals and seven forms of primary solid technical reports. biomass. An extensive list of inventories is available for base metals, In the stage that follows, an LCIA procedure attempts to establish a especially aluminium, iron and steel, copper, zinc, and nickel, as well as link between the materials and energy compiled by the LCI inventories some precious metals like gold, silver and the platinum group metals and their potential environmental impacts. Potential environmental (platinum, palladium, rhodium), alongside specialty metals such as ti­ burdens are given in the form of impact categories defined and selected tanium, tungsten, and uranium. An effort has also been made to include to describe the impacts caused by the emissions and the consumption of LCIs for other materials that are produced in lower quantities but have 3 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 Fig. 1. Possible linkages between energy system models (ESMs) and the life cycle assessment (LCA) framework according to ISO14040 standard. The life cycle inventory (LCI) includes the background system (in green) and foreground system (in light blue). Diverse LCIA methods are represented for illustrative purposes as dark blue rectangles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) high economic importance such as rare earth elements, indium and linked with the GLAD database. Lastly, Paulik and Hasan have created gallium. Additional inventory data for the extraction and purification of the Industrial Ecology Data Commons (IEDC) prototype which contains metals is highly scattered in publications and reports, and funding for around 180 industrial ecology-related datasets from the literature. generating bespoke data inventories for LCA purposes is often not Datasets are not limited to LCIs and include stocks, flows, process de­ available. As such, LCA practitioners tend to use existing LCI informa­ scriptions, input–output tables, material composition of products and tion within databases such as ecoinvent. other factors [41]. At present this collection only contains inventories Several previous studies have attempted to compile life cycle data for for a small number of unit processes for aluminium and steel. metals. Nuss and Eckelman compiled proprietary data from ecoinvent As with raw materials, the major source of LCI information for en­ version 2.2 in conjunction with data from various reports and scientific ergy systems is ecoinvent. In an LCA context, energy inputs generally publications [33]. Meanwhile, Van der Voet et al. used ecoinvent refer to electricity consumption and, thus, are assessed based on the so- version 2.2 to complete an inventory of the background system and called electricity production mix, the share of individual electricity datasets from previous studies [35,36] to assess the environmental im­ sources, generated from a diverse group of technologies, within a local plications of future demand scenarios of seven major metals: aluminium, electricity supply. The composition of this mix changes according to the copper, iron and steel, lead, nickel, manganese and zinc [37]. The LCIs geographical location of the system under study (i.e., region, country generated by both studies have not been made available and, therefore, and continent). It also varies from year to year due to changing energy it is not possible to use them directly in future studies. Although some policy measures, economic growth, energy intensity, technology recognised LCI formats do exist (e.g., ecoSPOLD), the general lack of LCI changes, meteorological conditions, and so on [42]. Accordingly, po­ data in formalised formats impedes their use in LCA software tools and tential environmental impacts can be considered to vary over time and restricts their widespread use by LCA practitioners. Furthermore, not between location. disclosing inventory data in published articles and other reports hinders The number of studies analysing the effects of changing electricity the reproducibility and replicability of the assessment. background systems on LCA results is limited. Mendoza Beltrán et al. Although some initiatives exist for generating fee-free LCIs for raw combined the Integrated Model to Assess the Global Environment materials, these tend to only partially cover raw materials and rarely (IMAGE) with the ecoinvent database to perform prospective LCAs for focus specifically on CRMs. The UNEP Global Life Cycle Assessment Data electric vehicles (EV) and internal combustion engine vehicles (ICEV) Access network (GLAD), launched in April 2018, aims to address this [43]. Changes were mainly focused on the electricity sector as electricity issue by providing a platform for hosting independent LCA databases, is the largest potential source of variability in the environmental impact so-called nodes, that are made available in various data formats. One of results [44]. The development of electricity scenarios firstly included the main functions offered is the conversion of LCI data from the native scenario generation using IMAGE; scenarios could then be evaluated format to a format that allows their use in common LCA softwares. In using LCIA. The adaptation of inventory parameters within LCA con­ GLAD, LCA practitioners can check the availability of datasets from sisted of using and adapting the emission factors of the GHG emissions, diverse providers such as IDEA, ecoinvent, USA LCA Digital Commons, mostly from the EU EDGAR [45], and replacing the shares of electricity- SICV Brazil and ELCD [38]. Another less ambitious initiative is the producing technologies, both using IMAGE. LCI inventories were updated database of the DoSE-LCADB [39,40]. This includes current LCI adapted to IMAGE by the development of the Wurst software platform, a data for agriculture (mainly vegetables), bioenergy (biomass from python-based application that enables the systematic importing, Populus spp. and soybean biofuel), and manufacturing (cement, natural filtering and modification of LCI data (https://github.com/IndEcol/wu cork, rubber mix and fertiliser). LCIs for individual raw materials are not rst). yet available, although this looks likely to change in coming years as the database becomes more widely known and more nodes begin to be 4 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 2.1.2. The foreground system based on the supply disruption probability and vulnerability, respec­ Conversely, foreground systems focus on the processes needed to tively. Berger et al. [32] built on the analysis of Sonderegger at al. [31] generate a certain amount of energy using a particular energy technol­ to give recommendations for the application of such methods by ogy. For example, the LCA for generating one kilowatt-hour (kWh) of formulating seven questions that can be further classified into ‘inside- electricity from a solar photovoltaic (PV) panel includes all processes out’ (i.e., current resource use changing the opportunities for future from the extraction of raw materials for the manufacturing of the panel users to use resources) and ‘outside-in’ (i.e., potential resource avail­ to its end-of-life. As many current decarbonisation targets are based on ability issues for current resource use). The study concluded that there is increased electrification of the energy sector by increasing the share of a need for methodological enhancement across method categories. renewable energy technologies, the number of LCIs for these technolo­ Additionally, the authors suggest that future methods increase the gies has increased considerably in the past decade. Appendix A includes number of abiotic resources considered, including secondary resources a list of LCIs available in version v3.7.1 of ecoinvent and the 2020 and anthropogenic stocks, and include the concept of dissipative re­ edition of GaBi, considering different energy sources and energy car­ sources in future developments. riers. While these values suggest that a significant amount of data is In terms of natural resources, critical raw materials–and metals in already available, new technologies continue to be developed and data particular–have attracted most of the attention in this regard as many of for the newest iterations of energy technologies is often not available in them look set to play a key role in the development of renewable energy a useable format for several years after its introduction. As such, an technologies [8]. However, at present there is no consensus on a single increased effort is needed to formalise and incorporate such data into method or set of methods for measuring resource availability using LCA databases in a more timely fashion. As an example, over 90% of the LCI methodologies and only a small number of approaches are currently listings for solar PV cells in ecoinvent v3.7.1 relate to first-generation available, as listed in Appendix C. Nevertheless, a number of studies cell technologies–41% single crystalline silicon (C-Si), 55% multi- have published data for individual raw materials. Nuss and Eckelman crystalline and 4% ribbon–while only 10% refer to second-generation performed LCA analyses for 63 metals and reported the results for five technologies–50% amorphous silicon (a-Si), 25% cadmium telluride main environmental impact categories: the global warming potential (kg (CdTe) and 25% copper indium gallium diselenide (CIGS). In the 2020 CO2-eq), the cumulative energy demand (CED) (MJ-eq), terrestrial edition of GaBi, 60% of these datasets are dedicated to first-generation acidification (kg SO2-eq), freshwater eutrophication (kg P) and human cells–38% C-Si, 37% multi-crystalline and 25% ribbon–with 20% toxicity (CTUh) [33]. The investigation yielded several interesting re­ describing second-generation cells–40% a-Si, 24% CdTe and 36% CIGS. sults. The global CED of metal production is estimated to have been 49 The remaining 22% refer to general datasets. PJ in 2008, which represents 9.5% of the global primary energy demand. It is also notable that neither of these databases currently contain LCI Iron and steel (74%) and aluminium (17%) dominate the CED impact data for the third-generation of cells such as metal halide perovskite cells category; the remaining 60 metals collectively represented only 9% of which are believed to hold significant economic and efficiency advan­ the total CED. Globally, the largest environmental impacts are found in tages over the currently commercialised first and second-generation the purification and refining of these metals. An environmental assess­ variants [46]. While not yet in widespread use, these technologies are ment of the seven major metals by Van der Voet et al. [37] referred to the expected to play a significant role in the emergence of solar PV cells CED (MJ-eq) and GHG emissions (kg CO2-eq) as defined by the going forward. This highlights the importance of including data relating CML2002 impact categories [3]. Their results show that the environ­ to burgeoning technologies in prospective energy system assessments mental impacts generated by metal production look set to increase and the shortcomings in the current data. gradually, and designate iron as the metal responsible for most impacts and emissions. Both studies provide valuable information about the 2.2. Inputs from LCIA potential environmental impacts of metals in energy systems. The availability of LCIA estimates for these and other materials al­ Using the collected LCI data, an LCIA attempts to form a connection lows composite values to be calculated for specific energy in­ between the product or system and its potential environmental impacts frastructures. If a breakdown of the main material components of a piece by creating indicator values relating to specific impacts; the results of of infrastructure (kg of each material) with a given power capacity (MW) the assessment are given in the form of environmental impact categories is available, a series of material intensities (kg/MW) can be calculated. to help evaluate outcomes in various areas (e.g., potential human health Values of LCIA indicators for a unit mass of each individual material can and ecological effects). Environmental impact categories can consider then be used to calculate a composite score of the indicator per unit of one selected environmental aspect, such as the cumulative energy de­ power capacity. This methodology is formalised for GWP and CED in mand, water footprint or carbon footprint, or combine several envi­ Appendix D, respectively. Using these two methodologies as examples ronmental impacts to become a ‘method’. Each of the most commonly allows final values–per MW of installed capacity–to be calculated for used methods include a list of impact categories (see Appendix B for an GWP (kg CO2-eq/MW) and CED (MJ-eq/MW). Outputs of this type then exhaustive list). allow side-by-side evaluations to be made between different energy As a result of the ongoing discussions surrounding the development technologies as the values of the chosen LCIA output category can be of the GLAM [30], attention to resource availability indicators has been directly compared in terms of installed capacity. Furthermore, using increasing in recent years and several publications now provide a assumptions for infrastructure lifetime and typical energy outputs al­ comprehensive review of indicators [31,32]. Sonderegger et al. revised lows values to be calculated–and compared–for each unit of energy the 27 different methods suitable for LCIA for mineral resource use and produced. It is noted that using the CED as an input to this methodology grouped these methods into four categories: depletion methods, future would return the total energy requirement per unit of produced energy efforts methods, thermodynamic accounting methods and supply risk (MJ-eq/MJ, say) which is essentially the inverse of the EROI, a common methods [31]. The two former methods consider resource depletion indicator used in MEDEAS, LOCOMOTION and many other projects. from a more ‘traditional’ LCIA perspective, where the availability of In any case, some limitations exist in relation to the indicators used in mineral resources given a certain stock are considered (depletion LCA. Firstly, although LCA uses quantitative material and energy input method) or the potential increase of extraction and refining costs, sur­ data to account for the potential environmental impacts, this informa­ plus energy use and other related aspects are considered under the tion is not generally used when quantifying material requirements. assumption of ore decline (future effort methods). The two latter Similarly, LCA indicators do not provide feedback on the contribution of methods, however, provide complementary information to LCA outputs recycling to the total supply of raw materials. In the EU, both of these regarding the use of cumulative material and energy use for a product issues are progressively gaining more importance [11]. As such, in order (accounted in useful energy or exergy), and the availability of materials to provide more complete environmental assessments in ESM there is a 5 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 need to develop methodologies that use LCA indicators which capture needs for a complete MFA involving 21 materials and groups of mate­ the potential environmental impacts of energy technologies alongside rials in 2012 [55]. The second, from 2015, provided a detailed meth­ material specific indicators which give additional information about the odology for developing MFAs [56]. A 2015 study, also referred to as the raw material supply factors of energy technologies EC MSA study, illustrated the entire life cycle of materials using a list of parameters which describe physical flows (including import and export 3. Potential contribution of material metabolism analysis to flows to each stage of the life cycle) and stocks. The study included a ESM total of 52 parameters divided into three groups: parameters repre­ senting physical flows and stocks of materials, parameters relating to The previous section reveals that LCA indicators provide a useful way policy objectives and criticality, and parameters relating to future sup­ to assess the environmental performance of the life cycle of an energy ply and demand change. In 2020, the EC also published the latest data system [43]. However, the findings also suggest that other characteris­ relating to raw material supply of critical raw materials [11] and non- tics of energy systems, such as the supply of materials, are poorly critical raw materials factsheets [57]. Employing a material meta­ captured within LCA methodologies. One of the most overlooked of bolism perspective, the EU also proposed a method for estimating the these is material metabolism. Material metabolism studies offer com­ criticality of resources [58]. This method is considered to be a snapshot plementary information about the supply of raw materials as they of the current situation in the EU and aims to support the development of consider the whole, integrated collection of physical processes that EU raw materials policy to help monitor supply risk and recycling convert raw materials to processed materials, components and finished aspects. products [54]. Material metabolism studies, including so-called material The 2020 CRM assessment sheds further light on the supply of raw flow analysis (MFA) approaches, help practitioners attain a better un­ materials within the value chain, assuming that raw materials can be derstanding of the flows and stocks associated with materials, the supplied in value chains as raw materials, processed materials, compo­ interconnection between mineral ores and materials, recycling aspects, nents and assemblies [59]. Within value chains, many aspects relating to and shed light on potential future constraints for technology develop­ local supply and demand of materials, the location and characteristics of ment and diffusion. Some of the most relevant issues quantified by MFA external supplies, substitutability and end-of-life recycling rates were are the supply of raw materials from mineral deposits and from recy­ identified as being relevant to the assessment of future supply con­ cling, and the interconnection with other raw materials along the value straints. Tellingly, calculations performed within the study found that chain. the EU depends on non-domestic production for more than 80% of the Mineral deposits are heterogeneously and unequally distributed raw materials demanded by its economy [60]. Many of these materials across the Earth and the availability of resources depends on various are extracted within a small group of countries which increases the factors such as natural occurrence, concentration (if they are sufficiently probability of supply shortages and affects the strength of the supply attractive to be mined) and accessibility. Geological surveys generally chain. provide figures about geological availability as ‘reserves’, ‘reserve base’ As a way of monitoring this dependency on non-domestic produc­ and ‘resources’ [47,48]. However, the lifetimes of many reserves and tion, the CRM methodology analysed the import dependencies of spe­ resources has continually been extended over the last 50 years. Thus, the cific materials in further detail by assuming that local dependence–or published reserve figures do not adequately reflect the total amount of import reliance (IR)–can be calculated as the amount of imports divided mineral potentially available in the long term and should not be used in by the total supply (imports plus domestically-sourced supply). IR can be the evaluation of future material availability [9]. As a result, MFA calculated for diverse stages across the value chain (e.g., as unprocessed processes do not, strictly speaking, focus on resource depletion in­ material at the extraction stage or as refined material at the processing dicators. Rather, they focus on raw material supply indicators, which stage). The results show that 28 of all 80 materials analysed–and 19 of can refer to either primary production (mining) or secondary production the 30 materials marked as ‘critical’–have an IR of 100%. Many others (recycling). Indeed, recycling represents a significant challenge due to have IR values well over 50%. This confirms that the EU is highly the great diversity of applications and end products where materials are dependent on imports for many raw materials which are increasingly embodied, the diversity of products recycled together and the variability affected by growing demand pressure from emerging economies and by of the related processes. Despite such difficulties, a few existing studies an increasing number of national policy measures that disrupt the supply recycling estimates, which allow meaningful indicators for the normal operation of global markets. Moreover, the production of many secondary supply of raw materials to be defined. materials is concentrated in a small number of countries (e.g., more than The production of raw materials is highly interconnected, especially 90% of rare earths and antimony, and more than 75% of germanium and those involving metals. Indeed, the topic of by-product dynamics is often tungsten, are produced in China, 90% of niobium is from Brazil and 77% discussed [33,49,50], although few publications propose a methodology of platinum from South Africa). for providing quantitative estimates [51-53]. Based on the literature The supply of secondary materials via recycling represents an op­ available, three different types of by-product metals are distinguished: portunity to offset overall supply risks, particularly for materials with metals derived from ores of major metals (e.g., germanium, indium), high dependencies on non-domestic production. Recycling can occur at metals that occur without a major metal (e.g., platinum group metals) each of the stages considered along the life cycle of a material or a and metals that can be mined when found in high concentrations (e.g., product (e.g., materials can be recycled at either the extraction stage or cobalt, gold). The availability of all three types is largely determined by the assembly stage). As such, when defining recycling indicators, it is the availability of the main ore as mine production cannot adapt quickly important to define the system boundaries in detail alongside the ma­ to meet structural changes in demand patterns. As a result, the supply terial flows included in the calculations. The 2020 CRM assessment [60] risk of these metals is high when the volume mined does not match with considers recycling to be a ‘risk-reducing factor’ and quantifies the market demand. Talens Peiró et al. gave one of the first figures illus­ supply of secondary materials using the so-called end-of-life recycling trating the metabolism of scarce materials and provided production input rate (EOL-RIR) indicator. EOL-RIR reflects the total material input shares between them [50]. Nuss and Eckelman subsequently provided a into the production stage that comes from recycling of post-consumer more complete and detailed illustration of the interlinkages between scrap and is regarded as a robust measure of the contribution of recy­ metals along the supply chain [33]. Obtaining more detailed informa­ cling to meeting materials demand. The results for EOL-RIR suggest that tion about the linkages between raw materials helps identify potential 47 of the 80 materials assessed currently play an insignificant role in the supply restrictions across the value chain that cannot be predicted based overall EU supply (less than 10% EOL-RIR); the results are starker for the on LCA studies. In the EU, the European Commission itself has per­ group of more ‘critical’ materials, where 26 of the 30 materials have formed several studies. The first of these identified information and data EOL-RIR scores less than 10%. 6 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 The key parameter presented in the 2020 CRM assessment [60] was a disassembly along with a more detailed analysis of the material recovery supply risk (SR) factor that used aspects of supply concentration, world from these technologies would need to be further developed. governance indicators (WGIs), IR (as above), trade restrictions and agreements, supply chain and bottleneck issues, EOL-RIR (as above) and 3.2. The supply risk of energy technologies in the EU criticality of substitutes to capture a dimensionless composite measure of EU supply risk for each material. Calculations were made for both the Again, the key output from the 2020 CRM assessment [60] was the mining/extracting and processing/refining stages, and the greater of the SR factor that quantifies the overall supply risk for each material as a two chosen as the final indicator. dimensionless constant based on a number of physical and geopolitical A report by Bobba et al. [10] provides data for EU domestic pro­ factors. Initial attempts to define a methodology for creating a com­ duction at the extraction stage (materials in the form of mineral ore) and posite SR score were based on the same pro-rata approach used for the the processing stage (materials considered refined material). In the EU, LCIA outputs (see section S4 of the supplementary information). How­ most of the materials domestically produced are generated at the pro­ ever, in order to capture the importance of materials that exist in much cessing stage, whereas materials obtained from the extraction stage smaller quantities, an additional parameter was required to normalise represent around 20%. In other words, the greatest supply risks are the amounts of required materials using some measure of overall located at the extraction stage of resources. For example, in the wind abundance of supply. Consequently, in the final formula, each material power supply chain, the risk is reduced along the supply chain from 99% intensity value, m, is normalised by dividing it by the annual con­ at the extraction stage to 88% at the refining stage, 80% at the sumption level within the EU, c. This provides a more useful measure of component stage to a final 42% at the assembly stage. For solar PV, the the significance of using the given amount in relation to the overall supply risk does not vary considerably from the extraction stage (94%) supply. Accordingly, the composite supply risk factor for a given tech­ to the assembly stage (99%). nology using inputs from n materials is as follows: At the larger scale, materials demand can be seen to be driven by technological changes as well as the continual growth of emerging ∑ n mi SRi economies. In the EU, raw materials demand is likely to continue to SRtechnology = (2) ci increase as a result of a commitment to becoming a climate neutral i=1 economy by 2050. Several studies exist that assess the demand for CRMs It is noted that, although the final value of SR is essentially dimen­ coming from several strategic technologies, including wind energy and sionless, the final units are actually the timeframe of the consumption solar PV technologies. The results are given for low-demand (LDS), data divided by the unit that the material intensity is based upon. In this medium-demand (MDS) and high-demand (HDS) decarbonisation sce­ example, the final units are, in fact, the relatively meaningless year per narios [14]. Information relating to the value chains can also help un­ MW. Other measures of material intensity, such as kg of materials per ravel the potential to decarbonise raw material supplies. As 17 of the 24 MJ of energy or kg of fuel supplied could also be used–highlighting the key materials used in these technologies are supplied as refined mate­ flexibility of this methodology to different datasets–but would result in rials to the EU, higher GHG emissions will inevitably be associated with different final units of the composite SR score. However, while many the transport of these materials. Accordingly, less opportunity exists to types of units can be used, one cannot directly compare final scores that reduce the overall carbon footprint of these technologies. use different units of material intensity and/or consumption data. Expanding upon the methodology proposed for calculating com­ posite LCIA indicators, further methodologies are proposed here for 4. Case studies: Wind turbines and solar PV panels in the EU using the values of EOL-RIR and SR for individual raw materials in the EC’s 2020 CRM assessment [60] to calculate composite scores for A better understanding of both the typologies and quantities of raw different energy production processes. As with the LCIA indicator materials used by energy technologies is required to identify technolo­ values, it is hoped that these new scores can be integrated into ESMs as a gies that may introduce more significant resource use issues in terms of way of including material metabolism aspects into the assessment pro­ both environmental impacts and material availability. Inclusion of such cesses relating to a variety of future energy systems. To the best of our factors within ESM projects has the potential to contribute to the current knowledge, this is the first time this has been attempted in such a way. research by allowing for more complete assessments of future energy scenarios to be generated. Most projected future energy scenarios predict significant increases 3.1. The circularity of energy technologies in the EU in the share of renewable energies in the EU energy mix, predominantly via wind turbine and solar PV technologies [11]. And, while utilising Eurostat uses the EOL-RIR parameter as an indicator for monitoring such technologies results in far lower day-to-day emissions once in the EU’s progress towards a circular economy on the thematic area of operation, the production of the infrastructure required to implement ’secondary raw materials’. The current paper proposes the use of EOL- more renewable energy regimes is often overlooked [61]. In the section RIR as a way of monitoring circularity aspects of energy technologies that follows, the methodologies for the two LCIA indicators–GWP and within ESM practices. The EOL-RIR for a technology can be calculated CED, as outlined in section S4 of the supplementary materials–are by considering the EOL-RIR values for individual materials in relation to operationalised alongside the methodologies for EOL-RIR and SR out­ the overall mass of materials in the item of infrastructure under study, in lined in the previous section using material intensity information for the this case expressed as the material intensity m. As rates are expressed as most common wind and solar photovoltaic technologies. a percentage, the pro-rata EOL-RIR values for each material must be Carrara et al. studied the raw material demands relating to four key divided by the total mass of all materials to provide the final EOL-RIR infrastructure types for both wind turbine and solar PV technologies value. Accordingly, the composite EOL-RIR for a given technology [62]. Data from the study provides inputs to case studies using the using inputs from n materials is as follows: current methodology. Firstly, inputs are provided for four types of wind ∑n turbine: two direct-drive (DD)–electrically excited synchronous gener­ mi EOL − RIRi EOL − RIRtechnology = i=1 ∑n (1) ator (EESG) and permanent magnet synchronous generator (PMSG)–and i=1 mi two gearbox (GB) driven–PMSG and double-fed induction generator The results indicate the overall percentage of recycled materials that (DFIG). Material intensity data is supplied for concrete, glass/carbon occur within end product and provide a better understanding of the composites, cast iron, epoxy resin polymers and steel alongside 12 circularity of a technology from a material perspective. To assess the critical metals. These values are provided alongside the corresponding circularity of the technology itself, further analysis that assesses the LCIA indicator data–GWP and CED–and material supply data–EOL-RIR, 7 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 overall annual consumption and SR–for the EU in Table 1. Meanwhile, four times higher than the other cell types. This high score is almost the study also reports data for installations based on four types of solar exclusively the result of a requirement for germanium. Lastly, the results PV cell: the first-generation crystalline silicon (C-Si) and three of the for EOL-RIR are almost identical for all turbine types as they are over­ newer, second-generation ‘thin-film’ cells–cadmium telluride (CdTe), whelmingly dominated by concrete and steel which are assumed to be copper indium gallium diselenide (CIGS) and amorphous silicon (a-Si). identical in plants for all four cell types. Selecting the most ‘desirable’ of Material intensity data is supplied for concrete, glass, plastic, and steel the four solar PV technologies is less straightforward. CdTe cells return alongside 10 critical metals. Input data values are provided in Table 2. the lowest scores for GWP and CED and the second lowest for SR, making All sources of data used in the analysis are summarised in Appendix E. it a strong performer. However, the low SR score for C-Si give it a very Results for both groups are summarised in Table 3. The results for strong advantage if this category is prioritised. wind turbines indicate that a significant amount of variation exists be­ It is noted that the scores for GWP, CED and SR are all significantly tween the four types analysed. The results for GWP and CED are all higher in wind turbines when compared to solar PV facilities on a per dominated by steel, which contributes around half of the total for all MW basis. For GWP and CED, this is explained by the far higher levels of turbine types. The higher level of steel in DD-EESG turbines means that concrete and steel required in wind turbine structures, while the dif­ it scores considerably higher than other turbines in both of these cate­ ferences for supply risk are predominantly due to the presence of rare gories. The four other non-critical materials, alongside zinc, provide earth materials in wind turbine generator systems. Conversely, high most of the remaining contributions to GWP and CED values. Variation levels of recovery and/or recycling for concrete and steel mean that is far higher within the supply risk category. For all turbines, this factor overall EOL-RIR rates are higher for wind turbines. Nevertheless, the is dominated by amounts of the rare earths dysprosium and neodymium results strongly indicate that production of a single MW of electricity and, to a lesser extent, praseodymium and terbium. Accordingly, the generation capacity via new wind turbine installations is considerably final score for the DD-PMSG turbine is substantially higher than the less desirable than solar PV panels in terms of GWP, CED and SR. other turbines owing to higher levels of all four of these metals. The Although many other aspects ultimately affect the adoption of different lowest levels of variation occur in the results for EOL-RIR, which are technologies, these simple findings suggest that certain elements of wind dominated by amounts of concrete and steel. Overall, the most ‘desir­ turbine designs would need to be improved if they were to become able’ of the four wind turbines analysed is the GB-DFIG turbines which comparable to solar PV panels in the aspects investigated. For example, return the lowest scores for GWP, CED and SR and the second highest new wind turbine generators should be designed to include features that EOL-RIR. facilitate their repairability by allowing access, disassembly and the A significant amount of variation also exists between the four types replacement of specific parts. Extending the service life using these of PV cells analysed. The results for GWP and CED are again dominated design features would enhance the remanufacturing and reuse of these by steel, which contributes between 30 and 40% of the observed levels parts and reduce their dependency on imports. Extending the lifespans in both categories. Glass and plastic also influence the final scores in of foundations, blades and other components may also improve their these categories, as does aluminium. Nevertheless, the levels of all ma­ desirability in relation to other renewable energy technology options. terials are assumed to be identical in the material intensities given in Collectively, these measures would result in lower SR and higher EOL- [62], so the intensities of the critical metals other than aluminium are RIR values. ultimately responsible for variations in GWP and CED. Hence, the levels The given case studies confirm the effectiveness of employing a of germanium and silicon in a-Si cells give them the highest scores in relatively simple methodology for obtaining useful information these categories. Variation is again significantly higher in the supply risk regarding emissions, embedded energy, supply risk factors and recycling category. By far the lowest score here is for the first-generation c-Si cells, rates that allow robust comparisons to be made between technologies with only small contributions from silver, aluminium and copper (silicon using readily available data. Furthermore, the exercise demonstrates itself does not make a significant impact). The final SR factor scores for that including raw materials assessments in ESM can help to visualise CdTe and CIGS cells are both moderately high, predominantly via the the relevance of certain materials in achieving energy targets and, presence of tellurium and indium, respectively. However, by far the therefore, to urge the development of new resource management mea­ highest score in this category was returned for a-Si cells, which is over sures directed to ensure the supply of key raw materials and/or Table 1 LCIA indicators, EU material supply data and specific material inputs for four wind turbine sub-technology case studies. Material LCIA indicatorsa EU material supply dataa Case study material inputsb GWP CED Consumption SR EOL-RIR Material intensity DD-EESG DD-PMSG GB-PMSG GB-DFIG [kg CO2-eq/MW] [MJ/MW] [kg/yr] [%] [kg/MW] [kg/MW] [kg/MW] [kg/MW] Concrete 0.12 0.9 90.0% 369,000 243,000 413,000 355,000 Glass/carbon composites 2.45 37.9 19.0% 8,100 8,100 8,400 7,700 Cast iron 1.91 20.9 85.0% 20,100 20,100 20,800 18,000 Polymers (epoxy resins) 4.70 97.3 1.0% 4,600 4,600 4,600 4,600 Steel 1.45 17.3 85.0% 132,000 119,500 107,000 113,000 Aluminium (Al) 9.36 107.7 5,252,000,000 0.59 12.4% 700 500 1,600 1,400 Boron (B) 1.42 22.4 62,850,000 3.19 1.0% 6 1 Chromium (Cr) 0.04 0.7 1,200,000,000 0.86 21.0% 525 525 580 470 Copper (Cu) 1.23 19.6 4,000,000,000 0.32 17.0% 5,000 3,000 950 1,400 Dysprosium (Dy) 59.60 1,170.0 14,000 6.20 0.0% 6 17 6 2 Manganese (Mn) 2.95 36.9 800,000,000 0.93 8.0% 790 790 800 780 Molybdenum (Mo) 16.93 232.1 60,000,000 0.94 30.0% 109 109 119 99 Neodymium (Nd) 49.60 733.7 100,000 6.07 1.3% 28 180 51 12 Nickel (Ni) 6.50 111.0 460,000,000 0.49 17.0% 340 240 440 430 Praseodymium (Pr) 78.43 1,158.4 41,000 5.49 10.0% 9 35 4 Terbium (Tb) 297.00 5,820.0 24,000 5.51 6.0% 1 7 1 Zinc (Zn) 2.76 49.4 4,000,000,000 0.34 31.0% 5,500 5,500 5,500 5,500 a See Table E1 in Appendix E for detailed description of data sources, b Material intensities taken from [11]. 8 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 Table 2 LCIA indicators, EU material supply data and specific material inputs for four solar photovoltaic sub-technology case studies. Material LCIA indicatorsa EU material supply dataa Case study material inputsb GWP CED Consumption SR EOL-RIR Material intensity c-Si CdTe CIGS a-Si [kg CO2-eq/MW] [MJ/MW] [kg/yr] [%] [kg/MW] [kg/MW] [kg/MW] [kg/MW] Concrete 0.12 0.9 90.0% 60,700 60,700 60,700 60,700 Glass 0.97 12.3 40.0% 46,400 46,400 46,400 46,400 Plastic 3.62 90.8 32.5% 8,600 8,600 8,600 8,600 Steel 1.45 17.3 85.0% 67,900 67,900 67,900 67,900 Aluminium (Al) 9.36 107.7 5,252,000,000 0.59 12.4% 7,500 7,500 7,500 7,500 Cadmium (Cd) 5.52 93.4 700,000 0.34 30.0% 50 Copper (Cu) 1.23 19.6 4,000,000,000 0.32 17.0% 4,600 4,600 4,622 4,600 Gallium (Ga) 169.31 2,605.6 27,000 1.26 0.0% 4 Germanium (Ge) 170.00 2,890.0 39,000 3.89 2.0% 48 Indium (In) 119.37 2,101.3 30,000 1.79 0.0% 15 Selenium (Se) 3.44 60.2 1,000,000 0.41 1.0% 35 Silicon (Si) 49.42 964.9 433,000,000 1.18 0.0% 4 150 Silver (Ag) 512.52 7,858.7 3,800,000 0.68 19.0% 20 Tellurium (Te) 6.94 125.4 30,000 0.51 1.0% 52 a See Table E2 in Appendix E for detailed description of data sources, b Material intensities taken from [11]. Table 3 Final indicator results for wind turbine and solar PV case studies (per MW of installed capacity). Lowest values of GWP, CED and supply risk and highest value of EOL- RIR are shaded in green. components for renewable energy technologies. disruptions of raw material supply. Such information will lead to the identification of renewable energy technologies with lower environ­ 5. Discussion and conclusions mental footprints in terms of greenhouse gas emissions while allowing more sustainable and realistic energy system options to be pinpointed Renewable energy technologies are evolving as a promising way of using a range of material supply indicators. reducing global warming potential and the effects of climate change. The proposed methodology for calculating composite indicators for Meanwhile, energy system models represent a powerful tool for fore­ energy supply technologies demonstrates that useful and informative casting possible low carbon future energy scenarios. Although, from a information can be calculated relatively simply from material intensity system perspective, environmental implications aside from greenhouse data in conjunction with life cycle impact assessment outputs and ma­ gas emissions need to be addressed to ensure the implementation of the terial supply data. In that sense, the methodology proposed offers, on most sustainable energy systems, most present-day energy system one hand, a clear definition of a set of indicators that support a more models cannot provide information about the other potential environ­ complete assessment of energy technologies alongside existing life cycle mental and raw materials implications of the systems they replicate. assessment studies. On the other hand, the use of established and reli­ This paper proposes a methodology that combines indicators based on able data sources (ecoinvent and official EU data) allows bespoke data in life cycle assessment and material metabolism studies with the objective a readily usable format to be easily elaborated by energy system of providing complete and valuable new information for exploring po­ modellers. tential climate policy pathways for reducing greenhouse gas emissions Additionally, while the given examples use data inputs for the Eu­ from a holistic perspective. This includes additional information about ropean Union based on a single megawatt of installed capacity, the the potential reduction of greenhouse gas emissions, total energy de­ methodology could easily be adapted to data sources from other regions, mand and, more importantly, a better understanding of the possible for smaller or larger scales and for net energy units. The simplicity of the limitations on obtaining projected installed capacities based on approach also means that any number of other life cycle impact 9 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 assessment or material supply data sources could be adapted and CRediT authorship contribution statement applied. The study has demonstrated that a variety of life cycle assess­ ment and material metabolism data is already available that can be used Laura Talens Peiró: Conceptualization, Methodology, Writing – to assess many forms of fossil-based and renewable energy technologies original draft. Nick Martin: Methodology, Investigation, Writing – using the proposed methodology and, ultimately, to include the derived original draft. Gara Villalba Méndez: Supervision. Cristina Madrid- indicators in energy system models or similar investigations. For some López: Writing – review & editing. technologies, greater effort is needed to improve the availability of life cycle inventory data in a useable and formalised format (e.g., the eco­ SPOLD standard), particularly for newer solar, geothermal and fuel cell Declaration of Competing Interest technologies and the myriad electricity storage options available, to name a few. For biomass-derived products used for energy purposes, the The authors declare that they have no known competing financial data currently available regarding material metabolism studies is interests or personal relationships that could have appeared to influence limited. For instance, the 2020 European Commission Non-Critical Raw the work reported in this paper. Materials report included supply risk and end-of-life recycling input rate data for three biomass materials (natural cork, natural teakwood and Appendix A. Supplementary material sapele wood) mainly used for construction material and high-end furniture and, thus, of little relevance for energy system models. With Supplementary data to this article can be found online at https://doi. the increasing importance of the circular bioeconomy in the European org/10.1016/j.apenergy.2021.118150. Union, more material metabolism studies of biomass-derived products are likely to be available shortly. As such, applying the indicator References calculation methodologies proposed in this paper will soon become feasible for a range of biomass applications. In that sense, it is thought [1] European Commission. The European Green Deal. European Commission 2019;53: 24. https://doi.org/10.1017/CBO9781107415324.004. that the potential of the methodology for comparing competing sub- [2] IEA. European Union 2020: Energy Policy Review. Iea 2020. technologies within a field could be especially useful. [3] Guinée J. Handbook on Life Cycle Assessment. Springer 2002. https://doi.org/ Although the present study is limited in its investigation of material 10.1007/0-306-48055-7. [4] Blanco H, Codina V, Laurent A, Nijs W, Maréchal F, Faaij A. Life cycle assessment metabolism indicators to supply risk and end-of-life recycling input rate, integration into energy system models: An application for power-to-methane in the it is thought that import reliance–included in the European Commis­ EU. Appl Energy 2020;259:114160. https://doi.org/10.1016/j. sion’s calculations of supply risk scores for individual materials–could apenergy.2019.114160. also be used to provide critical information as a standalone indicator. [5] Igos E, Rugani B, Rege S, Benetto E, Drouet L, Zachary DS. Combination of equilibrium models and hybrid life cycle-input-output analysis to predict the While the European Commission leans heavily on the supply risk factor environmental impacts of energy policy scenarios. Appl Energy 2015;145:234–45. for quantifying the overall criticality of materials, it is recognised that https://doi.org/10.1016/j.apenergy.2015.02.007. import reliance is more relevant in terms of greenhouse gas emissions as [6] García-Gusano D, Iribarren D, Martín-Gamboa M, Dufour J, Espegren K, Lind A. Integration of life-cycle indicators into energy optimisation models: The case study it essentially provides information about the transport requirements for of power generation in Norway. J Cleaner Prod 2016;112:2693–6. https://doi.org/ obtaining the raw materials for producing energy infrastructure. As 10.1016/j.jclepro.2015.10.075. such, it could be considered to be a proxy environmental impact indi­ [7] Volkart K, Mutel CL, Panos E. Integrating life cycle assessment and energy system modelling: Methodology and application to the world energy scenarios. cator and worthy of further investigation using similar analysis tech­ Sustainable Production Consumption 2018;16:121–33. https://doi.org/10.1016/j. niques to the current study, particularly as a readily available dataset spc.2018.07.001. already exists–at least for the European Union–for the set of most critical [8] Nuss P, Blengini GA. Towards better monitoring of technology critical elements in Europe: Coupling of natural and anthropogenic cycles. Sci Total Environ 2018; raw materials. For now, such approaches could be used in conjunction 613–614:569–78. https://doi.org/10.1016/j.scitotenv.2017.09.117. with the many energy system models and datasets already in existence [9] Commission E. Critical Raw Materials for the European Union. Brussels: Belgium; for European Union and other global and local energy systems to obtain 2010. [10] Hollins O, Faunhofer ISI. Study on Critical Raw Materials at EU level 2014:1–166. more accurate information regarding materials metabolism. This would [11] European Commission. Study on the review of the list of Critical Raw Materials - enable more informed strategy decisions to be made by climate and Final Report. 2020. https://doi.org/10.2873/11619. resource management policymakers. As wind and solar energy look [12] Deloitte Sustainability, Bristish Geological Survey, Bureau de Recherches likely to remain a policy priority in many European countries in coming Géologiques et Minières, Netherlands Organisation for Applied Scientific Research. Study on the review of the list of Critical Raw Materials - Criticality Assessments. years, the European Union will need to emphasise the importance of Luxenbourg: 2017. https://doi.org/10.2873/876644. better wind turbine and solar photovoltaic designs, including the [13] Commission E. Methodology for establishing the EU list of Critical Raw Materials. implementation of circular economy strategies such as repair and Publications Office of the European Union 2017:1–25. https://doi.org/10.2873/ 769526. remanufacture, as they strive to meet decarbonisation goals. Demand for [14] Bobba S, Carrara S, Huisman J, Mathieux F, Pavel C. Critical Raw Materials for such indicators looks set to increase, particularly as new regulations Strategic Technologies and Sectors in the EU - a Foresight Study. Ispra 2020. continue to include them as requirements. Consideration of the in­ https://doi.org/10.2873/58081. [15] European Commission. Communication from the Commission to the European dicators proposed in this paper represents a vital first step in progressing Parliament, the Council, the European Economic and Social Committee and the towards a more complete methodology for the modelling and identifi­ Committe of the regions on Critical Raw Materials resilience: charting a path cation of more sustainable energy systems. towards greater Security and sustainability. Brussels: 2020. https://doi.org/ 10.1007/978-3-030-40268-6_9. [16] Stehfest E, van Vuuren D, Kram T, Bouwman L, editors. Integrated Assessment of Funding information Global Environmental Change with IMAGE 3.0. The Hague, The Netherlands: PBL publishers; 2014. [17] PBL Netherlands Environmental Assessment Agency. IMAGE 3.0 2020. https:// This research has been made possible by the Sustainable Energy models.pbl.nl/image/index.php/Welcome_to_IMAGE_3.0_Documentation. Transitions Laboratory (SENTINEL) project (sentinel.energy.eu). This [18] Capellán-Pérez I, de Blas I, Nieto J, de Castro C, Miguel LJ, Carpintero Ó, et al. project has received funding from the European Union’s Horizon 2020 MEDEAS: A new modeling framework integrating global biophysical and research and innovation programme under grant agreement number socioeconomic constraints. Energy Environ Sci 2020;13(3):986–1017. https://doi. org/10.1039/C9EE02627D. 837089, and the Spanish Ministry of Science, Innovation and Uni­ [19] Guinée JB, Heijungs R, Huppes G, Zamagni A, Masoni P, Buonamici R, et al. Life versities through the “María de Maeztu” program for Units of Excellence Cycle Assessment: Past, Present, and Future. Environ Sci Technol 2011;45(1):90–6. (CEX2019-000940-M). https://doi.org/10.1021/es101316v. [20] Wernet G, Bauer C, Steubing B, Reinhard J, Moreno-Ruiz E, Weidema B. The ecoinvent database version 3 (part I): Overview and methodology. Int J Life Cycle Assess 2016;21(9):1218–30. https://doi.org/10.1007/s11367-016-1087-8. 10 L. Talens Peiró et al. Applied Energy 307 (2022) 118150 [21] Kupfer T, Baitz M, Colodel CM, Kokborg M, Schöll S, Rudolf M, et al. GaBi [42] Díaz A, Marrero GA, Puch LA, Rodríguez J. Economic growth, energy intensity and databases & modeling principles 2020. Sphera; 2020. the energy mix. Energy Econ 2019;81:1056–77. https://doi.org/10.1016/j. [22] Huijbregts MAJ, Steinmann ZJN, Elshout PMF, Stam G, Verones F, Vieira M, et al. eneco.2019.05.022. ReCiPe2016: a harmonised life cycle impact assessment method at midpoint and [43] Mendoza Beltran A, Cox B, Mutel C, Vuuren DP, Font Vivanco D, Deetman S, et al. endpoint level. Int. J. Life Cycle Assessment 2017;22:138–47. https://doi.org/ When the background matters: Using scenarios from integrated assessment models 10.1007/s11367-016-1246-y. in prospective life cycle assessment. J Ind Ecol 2020;24(1):64–79. https://doi.org/ [23] Dewulf J, Benini L, Mancini L, Sala S, Blengini GA, Ardente F, et al. Rethinking the 10.1111/jiec.v24.110.1111/jiec.12825. area of protection “natural resources” in life cycle assessment. Environ Sci Technol [44] Cox B, Mutel CL, Bauer C, Mendoza Beltran A, Van Vuuren DP. Uncertain 2015;49(9):5310–7. https://doi.org/10.1021/acs.est.5b00734. Environmental Footprint of Current and Future Battery Electric Vehicles. Environ [24] IPPC. Climate Change 2013: The Physical Science Basis. Contribution of Working Sci Technol 2018;52:4989–95. https://doi.org/10.1021/acs.est.8b00261. Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate [45] DG JRC. Emissions Database for Global Atmospheric Research (EDGAR) n.d. Change. Cambridge, United Kingdom and New York, NY, USA,: 2013. [46] Wilson GM, Al-Jassim M, Metzger WK, Glunz SW, Verlinden P, Xiong G, et al. [25] World Meteorological Organisation. Scientific Assessment of Ozone Depletion: photovoltaic technologies roadmap. J Phys D Appl Phys 2020;53(49):493001. 1998. Geneva: Switzerland; 1999. https://doi.org/10.1088/1361-6463/ab9c6a. [26] Tobergte DR, Curtis S. ILCD Handbook. vol. 53. 2013. https://doi.org/10.278/ [47] British Geological Survey. What is the difference between resources and reserves? 33030. n.d. https://www2.bgs.ac.uk/mineralsuk/mineralsYou/resourcesReserves.html. [27] Jungbluth N. Description of life cycle impact assessment methods. ESU-Services [48] U.S. Geological Survey. Mineral commodity summaries 2020. Reston, VA: U.S. Ltd 2020;39. Geological Survey; 2020. https://doi.org/10.3133/mcs2020. [28] Sala S., Benini L., V. C, B. VL, V. DL, R. P. Suggestions for the update of the [49] Verhoef EV, Dijkema GPJ, Reuter MA. Process knowledge, system dynamics, and Environmental Footprint Life Cycle Impact Assessment: Impacts due to resource metal ecology. J Ind Ecol 2004;8:23–43. use, water use, land use, and particulate matter - EUR 28636 EN. 2019. https://doi. [50] Peiró LT, Méndez GV, Ayres RU. Material flow analysis of scarce metals: Sources, org/10.2760/78072. functions, end-uses and aspects for future supply. Environ Sci Technol 2013;47(6): [29] Bulle C, Margni M, Patouillard L, Boulay AM, Bourgault G, De Bruille V, et al. 2939–47. https://doi.org/10.1021/es301519c. IMPACT World+: a globally regionalized life cycle impact assessment method. Int J [51] Graedel TE, Barr R, Chandler C, Chase T, Choi J, Christoffersen L, et al. Life Cycle Assess 2019;24(9):1653–74. https://doi.org/10.1007/s11367-019- Methodology of Metal Criticality Determination. Environ Sci Technol 2012;46(2): 01583-0. 1063–70. https://doi.org/10.1021/es203534z. [30] Frischknecht R, Jolliet O. Global Guidance for Life Cycle Impact Assessment [52] Nassar NT, Graedel TE, Harper EM. By-product metals are technologically essential Indicators - Volume 1. vol. 1. Paris, France: 2016. but have problematic supply. Sci Adv 2015;1:1–10. https://doi.org/10.1126/ [31] Sonderegger T, Berger M, Alvarenga R, Bach V, Cimprich A, Dewulf J, et al. sciadv.1400180. Mineral resources in life cycle impact assessment: Part I - A critical review of [53] Roelich K, Dawson DA, Purnell P, Knoeri C, Revell R, Busch J, et al. Assessing the existing methods. Int J Life Cycle Assess 2020;25(4):784–97. https://doi.org/ dynamic material criticality of infrastructure transitions: A case of low carbon 10.1007/s11367-020-01736-6. electricity. Appl Energy 2014;123:378–86. https://doi.org/10.1016/j. [32] Berger M, Sonderegger T, Alvarenga R, Bach V, Cimprich A, Dewulf J, et al. apenergy.2014.01.052. Mineral resources in life cycle impact assessment: Part II - Recommendations on [54] Ayres RU. Industrial metabolism. In: Ausubel J, Sladovich HE, editors. Technology application-dependent use of existing methods and on future method development and environment. Washington DC: National Academy Press; 1989. needs. Int J Life Cycle Assess 2020;25(4):798–813. https://doi.org/10.1007/ [55] European Commission. Data needs for a full raw materials flow analysis. Norfolk: s11367-020-01737-5. 2012. [33] Nuss P, Eckelman MJ, Janssen PJ. Life cycle assessment of metals: A scientific [56] Petiot C, Planchon M, De Prado Trigo A, Devauze C. Study on Data for a Raw synthesis. PLoS ONE 2014;9(7):e101298. https://doi.org/10.1371/journal. Material System Analysis: Roadmap and Test of the Fully Operational MSA for Raw pone.0101298. Materials. BIO by Deloitte: IFEU, Charles University Environment Center, ERA [34] Garrett P, Rønde K. Life cycle assessment of wind power: Comprehensive results Technology; 2015. from a state-of-the-art approach. Int J Life Cycle Assess 2013;18(1):37–48. https:// [57] European Commission. Study on the EU’s list of Critical Raw Materials (2020), doi.org/10.1007/s11367-012-0445-4. Factsheets non-Critical Raw Materials. Brussels, Belgium: 2020. https://doi.org/ [35] Verboon M. Environmental impacts of nickel production, 2010 –2050. An 10.2873/11619. assessment of the environmental impacts of metal demand and supply scenarios [58] Blengini GA, Nuss P, Dewulf J, Nita V, Peirò LT, Vidal-Legaz B, et al. EU using life cycle assessment. Leiden University & Delft University of Technology, methodology for critical raw materials assessment: Policy needs and proposed Den Haag, Netherlands., 2016. solutions for incremental improvements. Resour Policy 2017;53:12–9. https://doi. [36] Kuipers K. Environmental implications of alternative global copper supply org/10.1016/j.resourpol.2017.05.008. scenarios and the supply criticality of selected copper by-product metals. [59] European Commission. Study on the EU’s list of Critical Raw Materials (2020), Trondheim, Norway: Leiden University & Delft University of Technology; 2016. Factsheets on Critical Raw Materials. 2020. https://doi.org/10.2873/867993. [37] Van der Voet E, Van Oers L, Verboon M, Kuipers K. Environmental Implications of [60] European Commission. Study on the review of the list of Critical Raw Materials – Future Demand Scenarios for Metals: Methodology and Application to the Case of Non Critical Raw Materials Factsheets. 2020. https://doi.org/10.2873/92480. Seven Major Metals. J Ind Ecol 2019;23(1):141–55. https://doi.org/10.1111/ [61] Amponsah NY, Troldborg M, Kington B, Aalders I, Hough RL. Greenhouse gas jiec.2019.23.issue-110.1111/jiec.12722. emissions from renewable energy sources: A review of lifecycle considerations. [38] Ciroth A, Arbuckle P, Cherubi E, Ugaya C, Edelen A. Task 3: Core meta-data Renew Sustain Energy Rev 2014;39:461–75. https://doi.org/10.1016/j. descriptors and guidance on populating descriptors. 2017. rser.2014.07.087. [39] Talens Peiró L, Gabarrell i Durany X, Martinez Gasol C, Rieradevall Pons J. LCADB [62] Carrara S, Alves Dias P, Plazzotta B, Pavel C. Raw materials demand for wind and database (i-depot 120008) 2019. solar PV technologies in the transition towards a decarbonised energy system, EUR [40] Talens Peiro L, Gabarrell i Durany X. DoSE database (i-depot:120009), 2019. 30095 EN. Luxembourg: Publication Office of the European Union; 2020. https:// [41] Paulik S, Hasan MM. Industrial Ecology Data Commons prototype 2020. http:// doi.org/10.2760/160859. www.database.industrialecology.uni-freiburg.de. 11