ENVIRONMENTAL IMPACTS OF MANUFACTURING CRYSTALLINE SILICON AND ORGANIC PHOTOVOLTAICS By Seyed Mohammadreza Heidari A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Environmental Engineering – Doctor of Philosophy 2021 ABSTRACT ENVIRONMENTAL IMPACTS OF MANUFACTURING CRYSTALLINE SILICON AND ORGANIC PHOTOVOLTAICS By Seyed Mohammadreza Heidari Photovoltaics (PV) should provide about 25% of the global electricity production by 2050, which will require large-scale manufacturing. PV is expected to be a clean technology that can reduce the carbon footprint of electricity production. To maximize carbon reduction and maximize the environmental benefit, we must reduce the environmental impact of the manufacturing stage. This work evaluates the life cycle environmental impact of manufacturing mature (silicon) and emerging (organic) PV and evaluates alternative processes. For organic PV, C60 is often used as an acceptor material in OPV. Existing C60 purification methods are energy-intensive and require a large quantity of hazardous solvents. Therefore, it is desirable to modify existing C60 purification methods before OPV large-scale production to mitigate the potential environmental, cost, and chemical hazards of the manufacturing process. We used life-cycle assessment (LCA) to identify the environmental hotspots of the purification process. In addition to LCA, green chemistry, toxicity assessment, and analytical chemistry were employed to identify greener replacements. The alternative C60 purification has lower environmental (59%), cost (85%), and chemical hazard (42%) impacts compared to the existing C60 purification process. For mature technologies such as silicon PV (Si PV), it is necessary to evaluate the amount of materials needed to meet the expected PV capacity additions. Si PV is 95% of the current PV market and is expected to remain the leading technology until 2040 (>50%). We estimated the amount of material necessary for Si PV manufacturing based on PV installation in the US and the rest of the world in the next ten years. A bottom-up approach was used to evaluate the required materials for each sub-Si PV technology (e.g., aluminum back surface field, PERC, heterojunction, mono facial, bifacial, and perovskite/silicon tandem). Solar glass with 74 million metric tons and metallurgical-grade silicon (MG-Si) with three million metric tons have the highest material demand in the next decade. MG-Si production requires silica sand extracted from high-quality quartz (>98% purity). This study identified the purity and availability of potential quartz deposits globally. The country-specific carbon footprint of silica sand production was evaluated for quartz with various purity. The carbon footprint of producing silica sand was about 36% higher for low- quality quartz (65% purity) than high-quality deposits. We also quantified the carbon footprint and the cumulative energy demand of silica sand production from legal and illegal mines. The lower cost of silica sand production from illegal mines could result in using illegal quartz in the Si PV supply chain. Therefore, it is essential to have third-party certifications to ensure that the PV supply chain is free from illegal quartz and PV consumers buy ethical products. To my lovely family! I dedicated this dissertation to all Iranians who have sacrificed their lives and times to have an independent and free country. iv ACKNOWLEDGEMENTS My appreciation first goes to my advisor Dr. Annick Anctil. She guided me with patience and passion. She helped me a lot to be a responsible researcher and professional ethical engineer. I have learned a lot from her and appreciate it. I also would like to thank my committee members (Dr. Syed Hashsham, Dr. Irene Xagoraraki, & Dr. PM Allison) for their valuable feedback on my research. I appreciate my family for their unlimited supports. I also would like to thank all my friends, particularly my colleagues at MSU (Eunsang, Dipti, Angela, Sid, and Ben), who were always available to listen and talk. They are awesome! v TABLE OF CONTENTS LIST OF TABLES ....................................................................................................................... viii LIST OF FIGURES ........................................................................................................................ x KEY TO ABBREVIATIONS ....................................................................................................... xv Chapter 1 Introduction .................................................................................................................... 1 1.1 Introduction ................................................................................................................... 1 1.2 Motivation ..................................................................................................................... 4 1.3 Background ................................................................................................................... 5 1.3.1 Life-cycle assessment (LCA) ......................................................................... 5 1.3.2 Green chemistry principles ............................................................................ 6 1.3.3 Toxicity assessment ....................................................................................... 7 1.3.4 Analytical chemistry and material characterization techniques for environmental assessment ....................................................................................... 8 1.3.5 Organic photovoltaic (OPV) .......................................................................... 8 1.3.6 Silicon photovoltaic (Si PV) .......................................................................... 9 1.4 Dissertation outline ..................................................................................................... 10 Chapter 2 Identifying alternative solvents for C60 manufacturing using singular and combined toxicity assessments ...................................................................................................................... 13 2.1 Background ................................................................................................................. 14 2.2 Material and methods .................................................................................................. 17 2.2.1 Daphnia magna ............................................................................................ 17 2.2.2 Solvents ........................................................................................................ 18 2.2.3 Preparation and characterization of fullerene .............................................. 19 2.2.4 Toxicity experiments ................................................................................... 20 2.3 Results and discussion ................................................................................................ 21 2.3.1 Essential oil prescreening ............................................................................ 21 2.3.2 Singular toxicity assessment of C60 fullerenes ............................................. 22 2.3.3 Singular acute toxic impacts of solvents on daphnids ................................. 25 2.3.4 Combined toxic impacts on daphnids .......................................................... 27 2.3.5 Acknowledgment ......................................................................................... 31 Chapter 3 Environmental, cost, and chemical hazards of using alternative green solvents for C60 purification .................................................................................................................................... 32 3.1 Materials and methods ................................................................................................ 35 3.1.1 Materials ...................................................................................................... 35 3.1.2 Environmental impact assessment ............................................................... 36 3.1.3 Cost and chemical hazard analysis .............................................................. 36 3.1.4 Toxicity assessment ..................................................................................... 36 3.1.5 Green chemistry principles .......................................................................... 37 3.1.6 Experiments ................................................................................................. 37 vi 3.2 Results ......................................................................................................................... 38 3.2.1 Existing C60 purification methods................................................................ 38 3.2.2 Baseline evaluation ...................................................................................... 39 3.2.3 Alternative solvents for TMB ...................................................................... 39 3.2.4 Purification experiments .............................................................................. 41 3.2.5 Environmental, cost, and chemical hazard evaluation of potential methods 43 3.2.6 Overall evaluation ........................................................................................ 44 3.3 Conclusion .................................................................................................................. 46 Chapter 4 Material requirement and resource availability for silicon photovoltaic laminate manufacturing in the next ten years ............................................................................................. 47 4.1 Introduction ................................................................................................................. 47 4.2 Methods....................................................................................................................... 48 4.2.1 Study scope .................................................................................................. 48 4.2.2 Material requirement .................................................................................... 49 4.2.3 Resource availability .................................................................................... 51 4.3 Results and discussion ................................................................................................ 51 4.3.1 Material requirements .................................................................................. 51 4.3.2 Resource availability .................................................................................... 54 4.3.3 Conclusion ................................................................................................... 55 Chapter 5 The country-specific footprint of metallurgical grade silicon production for silicon photovoltaics ................................................................................................................................. 57 5.1 Background ................................................................................................................. 57 5.2 Methods....................................................................................................................... 60 5.2.1 Location of legal and illegal mines and MG-Si production ......................... 60 5.2.2 Environmental impact assessments.............................................................. 61 5.3 Results and discussion ................................................................................................ 63 5.3.1 Location of legal and illegal quartz mines and MG-Si production .............. 63 5.3.2 Environmental impact assessments.............................................................. 68 5.4 Conclusion .................................................................................................................. 74 Chapter 6 Conclusions and major contributions ........................................................................... 77 6.1 Sustainable design for OPV manufacturing ................................................................ 77 6.2 Material requirement for Si PV manufacturing and associated environmental impacts ....................................................................................................................................................... 79 6.3 Conclusion .................................................................................................................. 80 APPENDICES .............................................................................................................................. 82 APPENDIX A: Supplementary Information for Chapter 2 .............................................. 83 APPENDIX B: Supplementary Information for Chapter 3 .............................................. 95 APPENDIX C: Supplementary Information for Chapter 4 ............................................ 106 APPENDIX D: Supplementary Information for Chapter 5 ............................................ 113 REFERENCES ........................................................................................................................... 120 vii LIST OF TABLES Table 1.1. Twelve principles of green chemistry [27]. ................................................................... 7 Table 2.1. Singular and combined toxicity assessment of different experiments. ........................ 22 Table 2.2. Physiochemical properties of suspended fullerenes (88 ppm)* in the test medium. ... 23 Table 2.3. The water-soluble fraction composition for different essential oils ............................ 27 Table 4.1. Assumptions for each scenario to project materials needed for Si PV manufacturing 50 Table 5.1. LCA scenarios for quantifying the carbon footprint of supplying silica sand for China ....................................................................................................................................................... 69 Table A1. Physical properties and market price for conventional solvents and essential oils. All solubility and viscosity were reported at 298 K except when mentioned in parentheses. Data was collected from reference [1] except those were mentioned in the bracket. .................................. 84 Table A2. Summary of previous acute tests for fullerenes ........................................................... 85 Table A3. Chemical hazards (NFPA) and calculated hazardous score using the method from Lee et al., 2018 [20] ............................................................................................................................. 86 Table A4. Mortality of daphnids for prescreening tests after 48 h exposure to toxicants ............ 87 Table A5. Physiochemical properties of suspended fullerenes (maximum and minimum concentrations) in the test medium. .............................................................................................. 87 Table A6. C60 impacts on the toxicity of Baseline (TMB) and alternative solvents (essential oils) ....................................................................................................................................................... 92 Table A7. C60 concentrations in essential oils (human supplements) and TMB (in fullerene purification step) ........................................................................................................................... 93 Table A8. Dilution factors for mixtures (the “fixed-ratio-design” approach was used to conduct combined toxicity tests) ................................................................................................................ 94 Table B1. Data sources for material and energy used in inventory analysis ................................ 99 Table B2. Green chemistry principles [24] ................................................................................... 99 Table B3. Solvent toxicity scores and C60 solubility. ................................................................. 100 Table B4. Solvent chemical hazard scores based on NFPA. ...................................................... 102 viii Table B5. Chemical hazard scores for fullerene purification methods ....................................... 102 Table B6. Production rate ........................................................................................................... 103 Table B7. The price of energy materials in life cycle inventory ................................................ 103 Table B8. life cycle cost analysis ................................................................................................ 103 Table B9 Cost metric values of processes .................................................................................. 103 Table B10. LCA analysis details ................................................................................................ 104 Table B11. environmental, cost, and chemical hazard scores of the baseline and replacements 105 Table D1. Analysis of previous studies on environmental impact assessment of Si PV. ........... 115 Table D2. Data sources for materials and energy used in the inventory analysis ...................... 118 ix LIST OF FIGURES Figure 1.1. (a) Global electricity generation from renewables and fossil fuels until 2050 and (b) global installed capacity by energy source (adapted from Global Renewables Outlook, 2020, IRENA [1]) ..................................................................................................................................... 2 Figure 1.2. The contribution of renewables and solar energy in the US electricity production until 2050 (adapted from the Annual Energy Outlook 2021, EIA, 2021 [2]) ......................................... 2 Figure 1.3. The Collingride Dilemma diagram for PV technologies. Data is based on the 2019 IRENA report [6]. ........................................................................................................................... 3 Figure 1.4. (a) Life cycle stages of a product or a process and (b) LCA framework (adapted from [23])................................................................................................................................................. 6 Figure 1.5. A typical structure of organic photovoltaic (OPV) ...................................................... 9 Figure 1.6. Silicon photovoltaic (Si PV) sub-technology structures ............................................ 10 Figure 2.1 Fullerene (C60) production process. The red box shows the purification stage to separate C60 from C70 and higher fullerenes.................................................................................. 13 Figure 2.2. Hazardous score and C60 solubility of petroleum-based and plant-based (essential oils) solvents compared to the baseline (TMB). Potential alternative solvents are located in the purple box. .................................................................................................................................... 22 Figure 2.3. Auq/nC60 morphology after stirring for two weeks at room temperature in the absence of light and oxygen. (a) agglomerated nC60, (b,c) the distribution of nanoscale nC60 in stock with different sizes, and (d) an ultra-high-resolution image showing the existence of a pristine C60 molecule. ....................................................................................................................................... 23 Figure 2.4. (a) The average number of dead daphnids and pH after 48 h acute tests for different concentrations of auq/nC60 and (b) heartbeats of daphnids after 48 h (blue triangle), 72 h (black square), and 96 h (red circle) of exposure to different auq/nC60 concentrations. ......................... 25 Figure 2.5. Mortality of daphnids in singular toxic experiments after 48 h exposure to toxicants. (a) Dose-response curve of daphnids exposed to TMB (red), linseed oil (black), olive oil (blue), and sunflower oil (green), (b) EC50 of baseline versus alternative solvents, (c) morphology of daphnids exposed to baseline C3, alternative solvents C4-C6 compared with the control C2 and culture C1. The average % of dead daphnids with a 95% confidence interval was used for (a) and (b). ................................................................................................................................................. 26 Figure 2.6.The pH of the test medium at the end of each experiment and the average mortality of daphnids for the acute combined toxicity experiments. Daphnids were exposed to (a) TMB + auq/nC60, (b) linseed oil + auq/nC60, (c) olive oil + auq/nC60, and (d) sunflower oil + auq/nC60. The average mortality of daphnids exposed to the control (50% solvent and 50% reconstituted x water) for each mixture is shown in the red box. The average percentage of dead daphnids with 95% confidence intervals was used for (a)-(d). (e) The morphology of daphnids after 48 h of exposure to combined toxicants (C60 + solvents) and controls (reconstituted water +solvents). In figure captions, water = reconstituted water. ................................................................................ 29 Figure 3.1. The experimental approach to identify alternative C60 purification methods for the baseline process ............................................................................................................................ 38 Figure 3.2. Environmental evaluation of the baseline purification process (complexation with DBU) ............................................................................................................................................. 39 Figure 3.3. The toxicity and C60 solubility of solvents to identify replacements for TMB. The purple area highlights potential alternative solvents. Table B3 presents the detailed information of solvents #1 to #24. .................................................................................................................... 40 Figure 3.4. The evaluation of baseline, modified baseline, and alternative methods for C60 purification based on environmental, cost, and chemical hazard impacts. ................................... 45 Figure 4.1. The study scope to quantify materials needed for manufacturing Si PV laminate. ... 49 Figure 4.2. The typical structures of Si PV. Al-BSF: Aluminum Back Surface Field, PERC: Passivated Emitter and Rear Cell, HJT: Heterojunction Technology. [192–196] ........................ 51 Figure 4.3. Material requirements for manufacturing Si PV to meet the electricity demands from 2021 to 2031 for various scenarios globally. (A) Baseline , (B) S1, (C) S2, and (D) S3. Material requirements for the US and the rest of the world are shown in C2 and C3. ............................... 53 Figure 4.4. Material requirements for manufacturing Si PV technologies to meet the electricity demands from 2021 to 2031 globally. (A) Al-BSF Si PV, (B) PERC Si PV, (C) HJT Si PV, and (D) Si/Perovskite tandem. Material requirements for the US and the rest of the world are presented in C4-C11 ..................................................................................................................... 54 Figure 4.5. Available industrial-grade silica sand (A) and MG-Si (B) in 2020............................ 55 Figure 5.1. Study scope including quartz mining, silica sand extraction, and MG-Si production for manufacturing Si PV modules................................................................................................. 58 Figure 5.2. The process of quartz mining and silica sand extraction. The 1st step is similar for all types of quartz. The 2nd step depends on quartz purity. *Silica sand purity depends on initial quartz quality, mining process, and extraction chosen. ................................................................ 63 Figure 5.3. Annual industrial-grade silica sand production in the world from 1994 to 2019. Data was compiled from USGS annual reports [199]. .......................................................................... 64 Figure 5.4. The mines distribution of 10 major high-quality and industrial-grade silica sand producers in the USA .................................................................................................................... 65 xi Figure 5.5. (A) Global distribution of industrial-grade quartz (95% SiO2) mines in 2019. Data was collected from USGS annual reports [199]. (B) Annual industrial-grade silica sand export to China. Data was compiled from the annual UN Comtrade reports [197] ..................................... 66 Figure 5.6. Illegal silica sand mines along the Mekong River in Cambodia and within Haeju Bay in North Korea. ............................................................................................................................. 67 Figure 5.7. (A) The global distribution of MG-Si production. (B) annual MG-Si production from 2005 to 2019 (adapted from [199]) ............................................................................................... 68 Figure 5.8. (A) The regional production of MG-Si (colored provinces) (adopted from [199,268]), potential domestic quartz deposits (colored dots), and foreign available silica sand resources (colored hexagonal) for China. For foreign resources, legal and illegal trades are in green and red, respectively. (B) Regional electricity for China was compiled based on the previous study [269] .............................................................................................................................................. 69 Figure 5.9. The GWP (A, B, C) and CED (D, E, F) of silica sand production for MG-Si production in the US and China .................................................................................................... 73 Figure 5.10. (A) The GWP and (B) CED of producing MG-Si from high- and low-quality silica sand. .............................................................................................................................................. 74 Figure A1. Experimental setup for culturing daphnids and acute tests. Fluorescent light with the intensity of 70±5 foot candles and a 16 h photoperiod was used. Daphnids grew up in reconstituted water with mentioned pH, alkalinity, and hardness at room temperature. Twenty offspring daphnids ( < 24 h old) were used for each concentration (five daphnids in each beaker). ....................................................................................................................................................... 86 Figure A2. Absorption measurements at 335 nm in UV-VIS spectroscopy of (a) standard solutions (C60 in Toluene) and the fitting curve, (b) extracted C60 as explained in sec 2.3 .......... 88 Figure A3. Light-microscopic images of daphnids after 48 h exposition to auq/nC60 at different concentrations. The dead D. magna is shown to provide a comparison between live and dead daphnids. ....................................................................................................................................... 89 Figure A4. pH measurements and dose-response curves after 48-h acute tests for (a) 1,2,4-TMB, (b) linseed oil, (c) olive oil, (d) sunflower oil. .............................................................................. 90 Figure A5. Essential oil structures for (a) linseed oil, (b) olive oil, and (c) sunflower oil ........... 90 Figure A6. The chromatogram of sunflower oil – GC-MS .......................................................... 92 Figure A7. The chromatogram of the WSF of sunflower oil – GC-MS ....................................... 92 Figure B1. The material flow of complexation method (Nagata et al. [58]) - line shows the system boundary that was considered for LCA ............................................................................ 96 xii Figure B2. The material flow of crystallization method (Kwok et al. [297]) - line shows the system boundary that was considered for LCA ............................................................................ 96 Figure B3. The material flow of crystallization method (Grushko et al. [298]) - line shows the system boundary that was considered for LCA ............................................................................ 97 Figure B4. Material flow for modified baseline ........................................................................... 97 Figure B5. Material flow for P3 and P4 (alternative purification methods using olive oil) ........... 98 Figure B6. Material flow for P5 and P6 (alternative purification methods using linseed oil) ....... 98 Figure C1. Annual PV installation from 2021 to 2031 (Adapted from 2020 DNV-GL Energy Transition Outlook [13]) ............................................................................................................. 107 Figure C2. SHJ Si PV efficiency calculation .............................................................................. 107 Figure C3. Materials necessary for Si PV laminate manufacturing in the US and the rest of the world for the baseline (A), S1 (B), S2 (C), and S3 (D) ................................................................ 108 Figure C4. Materials necessary for Al-BSF Si PV laminate manufacturing in the US for S3 .... 109 Figure C5. Materials necessary for Al-BSF Si PV laminate manufacturing in the rest of the world for S3 ........................................................................................................................................... 109 Figure C6. Materials necessary for PERC Si PV and HJT Si PV laminate manufacturing in the US for S3 (a) monofacial c-Si PV, (b) monofacial mc-Si PV, (c) bifacial c-Si PV, (d) bifacial mc- Si PV ........................................................................................................................................... 110 Figure C7. Materials necessary for PERC Si PV and HJT Si PV laminate manufacturing in the rest of the world for S3 (a) monofacial c-Si PV, (b) monofacial mc-Si PV, (c) bifacial c-Si PV, (d) bifacial mc-Si PV .................................................................................................................. 110 Figure C8. Materials necessary for perovskite/silicon tandem PV laminate manufacturing in the USA and the rest of the world for S3 .......................................................................................... 111 Figure C9. The material intensity for manufacturing Si PV laminate globally .......................... 112 Figure D1. The GWP of monocrystalline Si PV manufacturing based on studies from 1990 to 2016 [312–329] ........................................................................................................................... 114 Figure D2. The GWP of multi-crystalline Si PV manufacturing based on studies from 1990 to 2018 [181,314–318,320–324,329–341] ...................................................................................... 114 Figure D3. The CED of multi-crystalline Si PV manufacturing based on studies from 1990 to 2016 [181,314–317,322–324,330–333,335,337–345] ................................................................ 115 Figure D4. The CED of multi-crystalline Si PV manufacturing based on studies from 1990 to 2018 [181,314–316,321–324,330,331,334,341,342,344–347] ................................................... 115 xiii Figure D5. Quartz mining (High-quality) ................................................................................... 116 Figure D6. Silica sand extraction (High-quality) ........................................................................ 116 Figure D7. Quartz mining (industrial grade and low-quality) .................................................... 116 Figure D8. Silica sand extraction (magnetic separation technique) ........................................... 117 Figure D9. Silica sand extraction (flotation separation technique)............................................. 117 Figure D10. Silica sand extraction (gravity separation technique) ............................................. 117 Figure D11. Metallurgical grade silicon (MG-Si) production .................................................... 117 Figure D12. Applications of a) sand and b) industrial sand. It was compiled based on the USGS report [199] and ISSST2020 conference [366] ........................................................................... 119 Figure D13. Illegal silica sand trades in 2016 [245–247,263–266] ............................................ 119 xiv KEY TO ABBREVIATIONS Al-BSF: Aluminum Back Surface Field APAC: Asia-Pacific region a-Si: amorphous Silicon auq/nC60: fullerenes dispersed in water AU: Australia CdTe: Cadmium Telluride CED: Cumulative Energy Demand CIGS: Copper Indium Gallium Selenide CS: C60 Solubility CS2: carbon disulfide c-Si PV: mono crystalline Silicon Photovoltaic DBU: 1,8-Diazabicyclo[5.4.0]undec-7-ene DI water: Deionized Water DLS: Dynamic Light Scattering EC50: Effective Concentration EF: freshwater aquatic ecosystem toxicity effect EIA: U.S. Energy Information Administration EPA: Environmental Protection Agency ERA: Environmental Risk Assessment FF: factor includes freshwater ecosystem exposure GC: Gas Chromatography GEC: Green Electronics Council xv GWP: Global Warming Potential HPLC: High-Performance Liquid Chromatography HRTEM: High-Resolution Transmission Electron Microscopy HS: Hazardous Score IEA: International Energy Agency IRENA: International Renewable Energy Agency KH: Cambodia LC50: Lethal Concentration LCA: Life-Cycle Assessment LCC: Life Cycle Cost mc-Si PV: multi-crystalline Silicon Photovoltaic MG-Si: Metallurgical-Grade Silicon MY: Malaysia O2-: Superoxide OPV: Organic Photovoltaic PERC: Passivated Emitter and Rear Cell PK: Pakistan PV: Photovoltaic QSARs: Quantitative Structure-Activity Relationships ROS: Reactive Oxygen Species SG-Si: Solar-Grade Silicon SEIA: Solar Energy Industries Association SEM: Scanning Electron Microscopy xvi SHJ: Silicon Heterojunction Si PV: Silicon Photovoltaic TEST: Toxicity Estimation Software Tool THF: tetrahydrofuran TMB: 1,2,4-Trimethilbenznene TU: Toxic Unit USGS: U.S. Geological Survey UV-VIS: UltraViolet-Visible WD: Water Demand WSF: Water-Soluble Fraction XF: Freshwater ecosystem exposure xvii Chapter 1 Introduction 1.1 Introduction Photovoltaics (PV) plays an essential role in decarbonizing the global energy system [1]. Fossil fuels are responsible for global warming, severely impacting the environment, economy, and public health. The transition from fossil fuels to renewables such as PV is a unique solution to mitigate global warming impacts. However, the energy transition pathway may threaten the sustainability of PV. Unsustainable manufacturing processes, using carcinogenic and hazardous materials for manufacturing PV, and using more natural resources (e.g., mining raw materials) to develop solar technologies are only a few examples of global energy transition challenges. PV is one of the leading renewable sources of electricity production in the US and the world. Electricity was 17% of global final energy consumption in 2010 and is estimated to reach about 43% by 2050 [1]. It is expected that PV will provide about 25% of the global electricity demand by 2050 (Figure 1.1a). The total installed photovoltaic (PV) capacity is anticipated to increase from 384 GW in 2017 to 8,519 GW in 2050 [1]. PV is also the leading renewable electricity source in the US, with about 16% contribution to electricity production in 2020 and 47% in 2050 (Figure 1.2) [2]. This increase in PV installation needs manufacturing more PV and would probably increase the demand for PV materials. 1 Figure 1.1. (a) Global electricity generation from renewables and fossil fuels until 2050 and (b) global installed capacity by energy source (adapted from Global Renewables Outlook, 2020, IRENA [1]) Figure 1.2. The contribution of renewables and solar energy in the US electricity production until 2050 (adapted from the Annual Energy Outlook 2021, EIA, 2021 [2]) Photovoltaic is supposed to be a green and clean technology. However, the socio- environmental impacts due to facing some challenges and limitations such as material shortage and unsustainable manufacturing processes can blemish PV’s public image. In general, PV can be categorized into three groups: wafer-based crystalline silicon, thin-film, and organic [3]. Silicon PV (Si PV) and thin-film PV are well developed and are commercially available. Si PV was more than 95% of the PV market in 2020, and the rest was thin-film PV technologies (e.g., cadmium telluride (CdTe) & copper indium gallium selenide (CIGS/CIS)) [3,4]. Organic PV (OPV) is an 2 emerging technology with a great potential market in the future due to its unique properties, such as absorbing indoor light spectrum, which makes it a suitable source of energy for indoor applications [5]. PV limitations vary from one technology to another, depending on the PV manufacturing stage. Commercialized PV technologies such as Si PV may suffer from material shortage due to market penetration and large-scale production, increasing material demand. While, for early-stage PV technologies such as OPV, limitations can be different, and challenges may relate to the manufacturing processes. Figure 1.3 explains the Collingride Dilemma for PV technologies. The freedom of design in emerging technologies (e.g., OPV) is much higher than mature PV technologies (e.g., Si PV), which means that the feasibility and modification cost can be much less for early-stage technologies. This shows the potential possibility and importance of the environmental modifications of manufacturing processes for OPV before large-scale production. On the other hand, it is essential to consider supplying materials needed for manufacturing more mature PV technology such as Si PV due to their high material demand to keep a large-scale production constant. Figure 1.3. The Collingride Dilemma diagram for PV technologies. Data is based on the 2019 IRENA report [6]. 3 1.2 Motivation PV is expected to be a green replacement for fossil fuels in the next couple of decades. PV is anticipated to be a clean technology that can eliminate the greenhouse gas emission generated by fossil fuels. However, existing PV manufacturing processes are not green, and associated carbon footprints threaten PV sustainability. In addition, the socio-environmental impacts of providing materials necessary for PV production and the possibility of using illegal materials are significant challenges that can tarnish the PV image. Therefore, it is essential to evaluate the sustainability (i.e., environmental, cost, and chemical hazards) of the PV manufacturing process as well as the socio-environmental impacts of supplying PV materials. The organic photovoltaic (OPV) manufacturing process needs a sustainable evaluation and some modifications before large-scale production. OPV has recently gained attention due to its unique advantages. OPV is cheaper, lighter, and less complex to be manufactured than traditional PV, can absorb indoor light spectrum, and can be fabricated in a larger area using colorful and flexible materials [5,7–10]. These properties make OPV desirable for interior design usages and expand OPV’s applications as windows for buildings and as screens for electronic devices [11– 15]. Low efficiency is the main obstacle for commercializing OPV; however, there has been a remarkable growth reported for OPV efficiency in recent years. OPV efficiency has increased from 5% in 2005 to 12% and 18% in 2015 and 2020, respectively, and is expected to grow more in the future [16–19]. Most previous studies focused on improving OPV efficiency and its mechanical durability, and only a few studies investigated the environmental impacts of synthesizing required materials for OPV. Even though fullerene (C60) is the critical material for OPV manufacturing (Figure 1.4), its production process requires a large amount of energy and hazardous materials [20,21]. Therefore, it is essential to evaluate the carbon footprint, cost, and chemical hazards of 4 producing fullerenes to ensure that the OPV manufacturing process remains sustainable before large-scale production. The material shortage is a severe threat for Si PV manufacturing. Si PV was more than 95% of the PV market in 2020 and is expected to remain the primary stream (>50 %) until 2040 [6,22]. Therefore, it is crucial to quantify materials necessary to support Si PV production in the next couple of years. Most previous studies applied theoretical models on historical PV data to estimate the PV material requirement. They ignored PV technology improvements such as efficiency, market share, wafer thickness, glass thickness, etc., directly affecting the amount of required materials. Therefore, there is an essential need to project Si PV materials based on Si PV industry improvements. In addition, the socio-environmental evaluation of supplying raw materials (e.g., quartz) necessary for Si PV manufacturing was widely ignored in the literature. Therefore, the availability and the quality of global deposits for raw materials that may affect Si PV production should be carefully investigated. 1.3 Background 1.3.1 Life-cycle assessment (LCA) LCA is a systematic approach to evaluate the entire life cycle of a product or a process from raw material extraction to manufacturing, distribution, use phase, and end of life stage [23] (Fig. 1. 4a). A general LCA framework contains four main steps: goal and scope definition, inventory analysis, impact assessment, and interpretation (Fig. 1. 4b). LCA is widely used for the environmental evaluation of PV manufacturing. LCA can be applied to any stage of the PV manufacturing process to quantify associated environmental impacts such as carbon footprint 5 (global warming potential - GWP), cumulative energy demand (CED), water footprint (WD), and E-factor to modify processes for lower impacts on the environment. Figure 1.4. (a) Life cycle stages of a product or a process and (b) LCA framework (adapted from [23]) 1.3.2 Green chemistry principles Green chemistry principles were introduced in 1998 (Table 1.1) [24]. Since then, it has been widely used to design chemical synthesis at early stages to avoid or mitigate the adverse issues in downstream. Green chemistry is a relatively new concept in nanomaterial synthesis for OPV. Green chemistry principles can determine alternative safer, energy-efficient, and less toxic synthesis procedures for nanomaterial production necessary for OPV manufacturing [25,26]. 6 Table 1.1. Twelve principles of green chemistry [27]. No. Metrics 1 Prevention: It is better to prevent waste than to clean it later. 2 Atom economy: processes should be designed to maximize the number of initial atoms in final products. Less hazardous chemical synthesis: processes should be designed to use and generate materials with little 3 or no toxicity to the environment and human health. 4 Design safer chemicals: minimizing the toxicity of final products. Safer solvents: solvents or any auxiliary materials should not be used as much as possible and should be 5 nonhazardous if they are used. Design for energy efficiency: energy requirements should be minimized in processes. Experiments should 6 be conducted at ambient temperature and pressure. 7 Use of renewable feedstocks: Raw materials should be renewable wherever possible. Reduce derivatives: unnecessary derivatizations should be avoided since they can generate wastes and 8 may need more regents. 9 Catalysis: catalysis reagents are better than stoichiometric reagents. 10 Design for degradation: final products should be designed in a way to be degradable in the environment. Real-time analysis for pollution prevention: analytical analysis should be used for in-process monitoring 11 to control before making wastes. Inherently safer chemistry for accident prevention: selected chemicals should be safer to minimize 12 potential hazards due to accidental release, explosion, or fire. 1.3.3 Toxicity assessment Toxicity evaluation is a powerful tool for the environmental risk assessment of chemicals [28]. The toxicity of chemicals can be quantified via theoretical models and experiments. For example, the toxicity of some chemicals, such as organic compounds, can be predicted based on theoretical models such as Quantitative Structure-Activity Relationships (QSARs). In contrast, the toxicity of other materials such as nanocarbons should be quantified based on toxicity experiments. The U.S. Environmental Protection Agency (EPA) developed the Toxicity Estimation Software Tool (TEST) using QSARs methodologies to allow users to predict chemical toxicity based on molecular structures [29]. The EPA also introduced standard protocols such as EPA600/8-87/011 for quantifying material toxicity via acute and chronic toxicity experiments [30]. The most common indicator in animal toxicity studies is the LC50, which corresponds to the lowest concentration associated with a 50% death rate [31–33]. LC50 can be used to quantify the toxicity of chemicals and compounds at early stages [34]. Daphnia Magna, a 0.5-5 mm-long 7 planktonic invertebrate organism, is recommended by EPA for toxicity assessment as a reliable representative of freshwater ecosystems at the cellular and animal scales [30,32][31]. They are easy to culture, sensitive to toxicants, and large enough to be visible under light microscopes [32,35]. 1.3.4 Analytical chemistry and material characterization techniques for environmental assessment Analytical chemistry instruments such as high-performance liquid chromatography (HPLC), gas chromatography (GC), UV-VIS spectroscopy, and mass spectroscopy are widely used for quantifying and analyzing chemicals in compounds. These instruments are also capable of being employed for monitoring the quality and quantity of target chemicals and by-products along with the production procedures for environmental evaluation purposes [36]. In addition, material characterization techniques such as high-resolution transmission electron microscopy (HRTEM), scanning electron microscopy (SEM), optical microscopes, and Zeta Potential Analyzer are used for characterizing nanomaterials such as fullerene (C60) to identify the size, shape, and distribution of synthesized nanoparticles to study nanomaterial impacts on the environment. 1.3.5 Organic photovoltaic (OPV) OPV can be divided into a single organic layer, bilayer (e acceptor & e donor), and bulk heterojunction (interpenetrating network of e acceptor and e donor) organic cells. A typical OPV structure includes an active organic layer containing two types of materials (e acceptor & e donor) sandwiched between two thin-film electrodes (Figure 1.5) [37]. The efficiency of single organic layer solar cells is much lower than the efficiency of bilayer organic solar cells. The low interfacial area between the donor and acceptor materials in the bilayer is an obstacle to have a commercial 8 product with a comparable efficiency with Si PV and other types of PV technology. The bulk heterojunction is another structure in that materials are blended in a solvent before deposition on the glass layer. This technique creates an interpenetrating network of electron acceptor and electron donor that increases the interfacial area results in a higher efficiency [38]. Layer-by-layer processing is the most recent technology with the capability of manufacturing OPV on a large scale [38]. Figure 1.5. A typical structure of organic photovoltaic (OPV) 1.3.6 Silicon photovoltaic (Si PV) Si PV manufacturing process starts from mining quartz followed by silica sand extraction, metallurgical-grade silicon (MG-Si) production, polycrystalline silicon purification, crystallization, wafering, cell processing, and module manufacturing [4]. Silica sand purity depends on the quality of quartz deposits. MG-Si purity reaches more than 99% by removing impurities from silica sand in an arc furnace at a high temperature (about 1,800 oC). Further purification requires removing impurities such as boron and phosphorus to produce solar-grade silicon (SG-Si) with 99.9999999% purity necessary for wafer production. The typical thickness of Si PV wafer is 170 µm for crystalline Si PV (c-Si PV) and 180 µm for polycrystalline Si PV (mc- Si PV) and is expected to be 50 µm by 2030 [39]. The wafer sizes change from 243 cm2 (156 mm 9 × 156 mm) to 441 cm2 (210 mm × 210 mm) [3]. A typical Si PV module area is 1.6 m2 containing 60 individual cells of 243 cm2 [4]. The current efficiency is 19.5% for c-Si PV and 18.0% for mc- Si PV which is expected to be 21.5% and 20% in 2031, respectively [39]. The typical structures of Si PV sub-technologies are shown in Figure 1. 6. Figure 1.6. Silicon photovoltaic (Si PV) sub-technology structures 1.4 Dissertation outline This dissertation aimed to evaluate the environmental impacts of OPV and Si PV manufacturing by: 1. Identifying the toxicity of existing solvents used for fullerene (C60) production necessary for OPV manufacturing (Chapter 2) 2. Determining the environmental hotspots and alternative non-hazardous solvents for producing fullerene (C60) used for OPV manufacturing (Chapter 3) 3. Quantifying materials necessary for Si PV manufacturing in the next ten years (Chapter 4) 10 4. Analyzing the carbon footprint and cumulative energy demand of supplying silica sand needed for Si PV manufacturing (Chapter 5) To achieve these goals, we used a combination of LCA, green chemistry, toxicity assessment, sustainability analysis, analytical chemistry, and material flow, along with scenario analysis and experiments. The second Chapter evaluated the toxicity of existing solvents consumed for making fullerene (C60) used for OPV manufacturing. The existing solvent for the large-scale production of C60 is 1,2,4-Trimethilbenznene (TMB). TMB is highly toxic and carcinogenic, with a possibility of severe impacts on workers and the environment in accidental release incidents. We conducted a series of singular and combined toxicity experiments to quantify the toxicity of TMB and its potential replacements. Alternative solvents were prescreened based on C60 solubility and hazardous score. Linseed oil, olive oil, and sunflower oil were identified as non-hazardous alternative solvents for TMB. The third Chapter focused on environmental, cost, and chemical hazards evaluation of using the potential alternative solvents from the second Chapter, as well as other potential solvents for C60 production. C60 purification was reported as the most energy-intensive stage in manufacturing OPV. We evaluated the environmental (GWP, CED, WD, and e-factor), cost, and chemical hazards of existing C60 purification methods to identify the baseline with the lowest environmental, cost, and chemical hazard impacts. LCA was used to determine the environmental hotspots of the baseline. Green chemistry principles and toxicity assessment were employed to develop a purification process using potential alternative solvents. The greener replacement for 11 TMB was determined for a method with the lowest environmental, cost, and chemical hazard impacts. The fourth Chapter identified the material needed for Si PV manufacturing in the next decade. We used a bottom-up approach to evaluate the amount of materials for each Si PV sub- technologies. We estimated the amount of PV materials based on technology improvements reported in energy outlooks and the annual reports of global leading Si PV producers. The fifth Chapter evaluated the carbon footprint of supplying raw materials necessary for Si PV manufacturing. Metallurgical-grade silicon (MG-Si) is the critical material for Si PV manufacturing. Quartz is the primary raw material needed for MG-Si production. The quartz purity is different for various mines. This can result in having extra processes to remove impurities from quartz before using it for MG-Si production. This Chapter quantified the GWP and CED of quartz mining, silica sand extraction, and MG-Si production for various countries. We determined potential quartz resources for future Si PV production and evaluated the associated carbon footprint. Overall, this dissertation focused on the environmental assessment of manufacturing OPV and Si PV. Alternative green solvents were introduced for C60 production necessary for OPV manufacturing to reduce the environmental, cost, and chemical hazards of existing C60 production processes. In addition, solar materials needed for Si PV manufacturing were estimated for the next ten years for the US and the rest of the world. We also quantified the country-specific carbon footprint and CED of quartz mining, silica sand extraction, and MG-Si production. 12 Chapter 2 Identifying alternative solvents for C60 manufacturing using singular and combined toxicity assessmentsa A large amount of hazardous solvents is consumed in OPV manufacturing. Fullerene (C60) is used as e acceptor/donor material in OPV. Figure 2.1 shows the process of C60 production, where the purification step needs a large amount of hazardous solvents to separate C60 from C70 and higher fullerenes. A well-known large-scale process for C60 purification is complexation with 1,8- Diazabicyclo[5.4.0]undec-7-ene (DBU) that requires about 120 kg highly toxic solvent (1,2,4 trimethylbenzene – TMB) to produce one kg C60 with >99% purity. This Chapter identified alternative nontoxic solvents for TMB using green chemistry principles as well as singular and combined toxicity assessments. Figure 2.1 Fullerene (C60) production process. The red box shows the purification stage to separate C60 from C70 and higher fullerenes. a Parts of this chapter have been published as Seyed M. Heidari, Annick Anctil, “Identifying alternative solvents for C60 manufacturing using singular and combined toxicity assessments,” Journal of Hazardous Materials, Volume 393, 2020 DOI: https://doi.org/10.1016/j.jhazmat.2020.122337 13 2.1 Background Fullerenes are allotropes of carbon that naturally occur in space [40,41]. However, they can also be produced under high temperature and low vacuum pressure conditions using combustion or arc plasma techniques [21]. The most stable fullerene is C60, where 60 carbon atoms form a stable hollow sphere structure [42]. Due to their unique physicochemical properties, C60 fullerenes are widely used in different industrial, commercial, and medical applications, such as organic solar cells, cosmetic products, drug delivery, cancer therapy, gene delivery, and photodynamic treatment [43–49]. C60 fullerenes are nontoxic. They are even sold as a human supplement because they are believed to inhibit allergic responses [50], increase hair growth [51], reduce symptoms of Parkinson’s [52], promote weight loss, reduce inflammation, and increase longevity [53]. The global fullerene market was $4.7 billion in 2016 [54] and is expected to grow by 7.9% annually until 2024 [55]. C60 is a nonpolar hydrophobic molecule that is soluble in petroleum-based solvents, such as benzene (2.9 g/L), toluene (8.7 g/L), o-xylene (9.3 g/L), tetralin (15.7 g/L), and 1,2,4- trimethylbenzene (TMB) (17.9 g/L) [56]. Those solvents are used during various stages of C60 manufacturing. The purification process requires a huge amount of solvents. For instance, more than 120 kg of TMB per kg of C60 is used during the fullerene production to separate C60 from C70 and higher fullerenes when using the complexation method [57,58]. Finding a greener solvent before large-scale production could reduce the impact of an accidental spill. Green chemistry principles can be used to select alternative solvents to reduce the environmental impacts of industrial nanomaterial production processes [59–62]. Green chemistry use principles to eliminate or mitigate the use or generation of hazardous materials in the design, manufacturing, and use phase of chemicals. It addresses three main categories: toxicity, material efficiency, and energy 14 efficiency [63–66]. In this Chapter, plant-based solvents (essential oils) were considered as potential alternative solvents since they are nontoxic [67], and their solubility with C60 is similar to that of conventional solvents such as TMB (Table A1 in Appendix A). Investigating the potential toxicity of chemicals is required for a green chemical design [66,68]. The most common indicator in animal toxicity studies is the EC50, which corresponds to the lowest concentration associated with a 50% death rate [31–33]. The EC50 of fullerenes prepared in organic solvents such as toluene, acetone, and tetrahydrofuran (THF) varies from 0.4 to 0.8 mg/L [69–72]. The toxicity in those solvents is due to the residual solvent attached to the fullerene structure [73]. To avoid using organic solvents, fullerenes can be dispersed in water (auq/nC60) by mixing the solution for two to eight weeks using magnetic stirring. Using this method, the C60 toxicity is lower. In one study, less than 20% of the daphnids died when exposed to a ten mg/L auq/nC60 solution [74]. In two other studies, neither acute (48 h) nor chronic (21 days) toxic impacts were reported for daphnids exposed to 35 mg/L auq/nC60 [75,76]. Table A2 provides additional details about previous toxicity studies for C60 in water and other solvents. Life-cycle assessment (LCA) is an established method to evaluate the environmental impact of a product. However, its use for nanomaterials and the quality of published studies have been criticized due to the lack of toxicity characterization factors [77]. The recommended method for LCA toxicity impact is the consensus model USEtox developed by the UNEP/SETAC [78]. This model uses an ecotoxicological characterization factor to predict the ecotoxicity impacts of a chemical in freshwater ecosystems. This factor includes environmental fate (FF), freshwater ecosystem exposure (XF), and freshwater aquatic ecosystem toxicity effect (EF) [79]. Environmental risk assessment (ERA) is an approach to characterize the magnitude of health risks from contaminants and predict the probability of adverse health effects of hazards on humans and 15 the environment [28,80,81]. The EC50 from single toxicity experiments is used for characterization factors in LCA [82] and ERA studies [28]. Although previous studies characterized C60 as a nontoxic nanocarbon, the assessment is incomplete because it ignores the combined toxicity of fullerenes and solvents. Focusing only on the toxicity of each chemical to identify an alternative solvent is not adequate since the interactions between chemicals may affect the overall toxicity. Previous toxicological studies on chemical mixtures showed that singular toxicity could be different from combined toxicity, and the effects of interactions could be antagonism, additive, or synergism [83–89]. Even though C60 and the selected essential oils are non-toxic, using essential oils as solvents could facilitate the fullerene transport into biological systems, in a process analogous to that observed in quantum dots and linseed oil [90]. Previous research demonstrated that nC60 molecules could destabilize cell membranes through a variety of mechanisms [91–94] and cause membrane leakage that kills cells [95]. Furthermore, nC60 that have entered the biological systems can be photoactivated due to their P conjugation and generate superoxide anions that are toxic to cells [90,96,97]. The photoactivation effect is used for medical applications to kill cancer cells or HIV viruses [98–100]. Therefore, the combination of C60 and essential oils, which are both nontoxic materials, is potentially toxic to the environment and needs to be investigated. In this Chapter, We used Daphnia magna, a 0.5-5 mm-long planktonic invertebrate organism recommended by the US EPA for toxicity assessments [30,32]. It is a reliable representative of freshwater ecosystems in singular and combined toxicity assessments at the molecular, cellular, and animal scales [32,101–103]. Daphnids are sensitive to toxicants, large enough to be visible under light microscopes, abundant in the northern hemisphere, and easy to culture [32,35]. They are also recommended as an alternative to mammals in prescreen chemical 16 toxicity studies since these microcrustaceans are at the basis of the freshwater ecosystem food chain [31]. The objective of this Chapter was to identify a suitable replacement for TMB, which is used in large quantities for C60 manufacturing, by using singular and combined toxicity assessments. Singular toxicity tests were designed to quantify the toxicity of C60, TMB, and alternative solvents. Combined toxicity tests were used to investigate the effect of interactions of C60 and the selected solvents on D. magna. This work will contribute to understanding the importance of combined toxicity assessments in environmental studies such as solvent selection using green chemistry, life-cycle assessment, and environmental risk assessments where toxicity evaluations mostly rely on singular toxicity and ignore combined toxicity. 2.2 Material and methods The experimental plan was designed to quantify the singular and combined acute effects of C60 with different solvents (TMB, linseed oil, olive oil, and sunflower oil) on D. magna. The concentrations of C60 and solvents were determined based on prescreen-test results. 2.2.1 Daphnia magna Daphnids were obtained from an EPA-verified laboratory in Colorado (Aquatic BioSystem Inc.) and cultured in our laboratory in 2-L glass beakers for 3-5 weeks under the recommended conditions [30] (Figure A1). Briefly, daphnids developed in reconstituted water [30] at room temperature (20±1 oC) and were fed (5 mg/L trout food and 108 cells/L Raphidocelis subcapitata algae) three times per week. Fluorescent light with an intensity of 70±5 foot candles was used for a 16 h photoperiod. The alkalinity, hardness, and pH of reconstituted water were measured using water quality test stripes (McMaster-Carr Supply Company) and a pH meter (VWR SympHony 17 Benchtop Meter B10P) to ensure that they were within the EPA-recommended ranges. Offspring daphnids (< 24 h) were used for acute tests. Adults were separated from young daphnids one day before the experiments to ensure that all daphnids for acute tests were less than 24 h old. Five daphnids were exposed to toxicants in four 100 mL glass beakers (20 daphnids for each concentration) in all experiments. The acute toxicity experiments were valid based on the EPA protocol since the percentage of dead daphnids was less than 10% in controls [30]. 2.2.2 Solvents Potential alternative solvents were selected based on green chemistry principles, in particular, waste minimization and selection of less hazardous solvents (Principles #1 and 3). The solvents were selected based on the C60 solubility and hazardous score as calculated by Lee [104] (Tables A1 and A3) Linseed oil (CAS 430021), olive oil (CAS O1514), sunflower oil (CAS 530285), and TMB (CAS T73601) were purchased from Sigma Aldrich (Burlington, MA). Preparation of solvents’ stocks followed a previously-reported procedure by which aromatic hydrocarbons and crude oil were added to water and stirred for 24 h to reach stability [105–107]. In summary, 280 g of each essential oil and one g of TMB were added separately to 1-L Erlenmeyer flasks and filled up to one L with reconstituted water to minimize the headspace and avoid losing potential volatile components. Solutions were stirred on a magnetic stir plate for 24 hours (480 rpm) at room temperature. All flasks were sealed by parafilm sealing and covered by aluminum foil to avoid evaporation, oxidation, and photodegradation. Since the solubility of essential oils and TMB in water is low (Table A1), solutions were left for one h. The undissolved fraction was removed from the top, and the water-soluble fraction (WSF) was collected at the bottom using a separatory funnel. The composition of WSF was determined using an Agilent 7890B gas chromatograph/ triple quadrupole mass spectrometer 7010B (GC-MS, Agilent, Santa 18 Clara, CA). The GC-MS method is described in Appendix A (A9). The average and standard deviation of calculated numbers were based on analyzing the quadruplicate of each WSF. 2.2.3 Preparation and characterization of fullerene C60 fullerenes with 99% purity were purchased from MER (Tuscan, AZ). Fullerene stocks were prepared by adding 250 mg of C60 into 1 L of reconstituted water and stirred for two weeks at ambient temperature to reach stability [108]. The fullerene stock was prepared in a nitrogen atmosphere and covered with aluminum foil to minimize fullerene oxidation or photodegradation. The concentration of suspended fullerenes in the solution was verified using a UV-VIS spectrophotometer (SHIMADZU, UV-2600 120V) based on the measured absorption at 335 nm. Fullerene was extracted in accordance with a previously reported method [109]. Briefly, 2% sodium chloride in deionized water and toluene (HPLC grade) were mixed for 2-3 minutes with fullerene-containing water using a 2:2:1 ratio (C60 solution: toluene: 2% sodium chloride) followed by three washes with toluene to ensure that all fullerenes were dissolved in toluene. A plastic pipette was used for UV-VIS spectroscopy measurements to collect the top layer, which contained fullerenes dissolved in toluene. After stirring for two weeks, the morphology of suspended fullerenes was determined using ultra-high-resolution transmission electron microscopy (HRTEM) (JEOL USA JEM-2200FS TEM) at 200 kV. Samples were placed on 300-mesh copper carbon grids and dried before imaging. The surface charge and size of suspended fullerenes in reconstituted water after stirring for two weeks and the test mediums were measured using dynamic light scattering (DLS) (ZetaPALS, Brookhaven Instruments Corporation). Quadruplicates of each test medium were used to calculate the average and standard deviation. 19 2.2.4 Toxicity experiments Prescreen tests were designed to determine the appropriate acute-test concentration range for EC50 calculation and a dose-response curve. For TMB, 8, 0.8, and 0.08 mL of WSF and for essential oils, 24, 8, and 0.8 mL of WSF were added separately to 100-mL glass beakers and filled up to 80 mL with reconstituted water. Five young daphnids were transferred one by one to each beaker using a plastic pipette and exposed to toxicants for 48 h. The number of dead daphnids was counted after 48 h (Table A4). The concentration range of toxicants for acute tests was established using the lowest concentration with 100% daphnids death and the highest concentration with 0% death. Five concentrations of linseed oil (138, 69, 34.5, 17.25, and 8.625 g/L), six concentrations of olive oil (276, 138, 69, 34.5, 17.25, and 8.625 g/L), seven concentrations of sunflower oil (552, 276, 138, 69, 34.5, 17.25, and 8.625 g/L) and eight concentrations of TMB (1539, 1000, 650, 422.5, 274.6, 178.5, 116, and 75.4 mg/L) were prepared by diluting WSF with reconstituted water for acute tests. As summarized in Table 2. 1, the acute tests followed the EPA protocol (600/8-87/011 1986) [30] for all nine experiments plus the control. Briefly, young daphnids (< 24 h) were exposed to at least five concentrations (20 daphnids in four beakers for each concentration) under the fluorescent light (16 h period) and at room temperature for 48 h. The average number of dead (or immobilized) daphnids after 48 h was used for the dose-response curves to extrapolate the EC50 of each toxicant. The fixed ratio design approach was used to prepare the mixture of C60 and solvents, as described in similar combined toxicity studies [86,88,89,110–113]. Tests were designed to evaluate the interaction of C60 with the various solvents using 1 toxic unit (TU) as the reference. For the solvent, one TU was the EC50 concentration, and for C60, it was the maximum concentration (176 ppm). Six different solutions were prepared where the concentration of solvent 20 was kept constant at 1 TU (e.g., 1 TU for olive oil was 54,485.4 ppm) and the C60 concentration was varied from 0 to 1 (0, 0.0625, 0.125, 0.25, 0.5, and 1) [84]. For TMB and C60, as suggested in the literature for situations with a large difference in toxicity [83,87], a concentration lower than the EC50 was used. Using a lower concentration than the EC50 for TMB allows us to evaluate the development of toxic responses due to changes in C60 concentration. The average and standard deviation for the EC50 were calculated with a 95% confidence interval using OriginLab (version 2018) software. After 48 hours of exposure to toxicants, morphology changes in the daphnids were identified using a light microscope (Olympus BH2-RFCA). After transfer from the media to microscopy slides, the daphnids’ heartbeats were counted visually three times for 15 seconds intervals per D. magna [114–116]. 2.3 Results and discussion 2.3.1 Essential oil prescreening As highlighted by the purple box in Figure 2.2, potential alternative solvents were identified based on their high C60 solubility (CS) and low hazardous score (HS) compared to TMB. Cedar oil was eliminated due to its high hazardous score (HS=1.33), which is the same as TMB. In contrast, linseed oil (HS=0.33, CS=53.1 g/L), olive oil (HS=0.67, CS=23.6 g/L), and sunflower oil (HS=0.33, CS=6.91 g/L) were selected as potential alternative solvents for toxicity assessments. In Figure 2.2, hazardous score and C60 solubility were normalized based on the hazardous score from toluene and the C60 solubility of linseed oil because they were the highest value for each category. 21 Table 2.1. Singular and combined toxicity assessment of different experiments. Experime Toxicant Concentration nts S1 C60 176, 88, 44, 22, and 11 mg/L S2 TMB 1539, 1000, 650, 422.5, 274.6, 178.5, 116, and 75.4 mg/L Singular S3 Linseed oil 138, 69, 34.5, 17.25, and 8.625 g/L (S) S4 Olive oil 276, 138, 69, 34.5, 17.25, and 8.625 g/L S5 Sunflower oil 552, 276, 138, 69, 34.5, 17.25, and 8.625 g/L Control Reconstituted water Toxicants (40 mL of solvent solution + 40 mL of auq/nC60) C1 TMB (EC50) + C60 (S1) Combined C2 Linseed oil (EC50) + C60 (S1) C3 Olive oil (EC50) + C60 (S1) (C) C4 Sunflower oil (EC50) + C60 (S1) Control 40 mL of solvent at EC50 + 40 mL of reconstituted water Figure 2.2. Hazardous score and C60 solubility of petroleum-based and plant-based (essential oils) solvents compared to the baseline (TMB). Potential alternative solvents are located in the purple box. 2.3.2 Singular toxicity assessment of C60 fullerenes The morphology of C60 in auq/nC60 suspensions was evaluated using HRTEM and is presented in Figure 2.3. The TEM images of auq/nC60 show that after two-week stirring, nC60 were present in both individual (» 7 nm) and agglomerated (7 nm – 1 µm) forms. Fullerenes can naturally aggregate and form larger fullerene particulates after release into water systems [117]. Therefore, size screening was not used to remove particulates larger than 200 nm because this 22 could naturally occur in the environment. Table 2. 2 presents the average surface charge and size of agglomerated fullerenes in different solvents. Based on the zeta potentials results, fullerenes were moderately stable in all test mediums [118]. Table 2.2. Physiochemical properties of suspended fullerenes (88 ppm)* in the test medium. Media Average size (nm)* Zeta potential (x, mV) Deionized water 131 - 38.8 Reconstituted water (auq/nC60) 539 ± 177 - 38.5 ± 0.2 C1 (TMB + auq/nC60) 449 ± 285 - 38.2 ± 0.3 C2 (Linseed oil + auq/nC60) 318 ± 127 - 38.4 ± 0.3 C3 (Olive oil + auq/nC60) 470 ± 171 - 38.3 ± 0.2 C4 (Sunflower oil + auq/nC60) 457 ± 276 - 38.3 ± 0.3 * Table A5 provides more information on physiochemical properties of suspended fullerenes at the minimum and maximum concentrations The concentration of C60 in the fullerene stock and dilutions was verified by measuring the absorption at 335 nm in UV-VIS spectroscopy (Figure A2). Stable C60 suspension was reported for concentrations lower than five mg/L [74]. However, concentrations as high as 176 mg/L were used to find the EC50 of C60 in this study. Figure 2.3. Auq/nC60 morphology after stirring for two weeks at room temperature in the absence of light and oxygen. (a) agglomerated nC60, (b,c) the distribution of nanoscale nC60 in stock with different sizes, and (d) an ultra-high-resolution image showing the existence of a pristine C60 molecule. 23 All concentrations of auq/nC60 were not toxic to daphnids, as demonstrated by the low daphnids mortality (< 20%), steady heartbeats, and unchanged morphology (Figure 2.4.) A 15% average death rate was recorded for the highest C60 concentration (176 ppm), while death rates were lower than 10% at lower concentrations. Since the percentage of dead daphnids was too low to obtain the dose-response curve, which is needed to extrapolate the EC50 of C60, EC50 was assumed to be higher than 176 ppm. Our result was five times higher than the previous study [75], but there were also multiple differences in the experimental methods. The main differences are that they used a 10-day chronic test, the C60 was mixed in miliQ water for one month, and the initial C60 concentrations in the test medium were different. Heartbeats were measured after 48, 72, and 96 h of exposure to auq/nC60 (Figure 2.4b) to understand the long-term impacts of C60 on daphnids. The measured average heartbeats changed depending on the time of measurements. For the measurement after 48 hours, the heartbeats increased with increasing concentration. For the measurements at 72 and 96 hours, the trend was the opposite, with heartbeats decreasing with increasing concentration. The difference after a longer period of time could indicate chronic toxic impacts of C60, but further tests would be required to draw this conclusion. Additionally, there was no significant change in daphnids morphology after 48 h of exposure to different concentrations of auq/nC60 compared to the control (Figure A3). At the end of the experiment, the pH was between 8.2 and 8.6. 24 Figure 2.4. (a) The average number of dead daphnids and pH after 48 h acute tests for different concentrations of auq/nC60 and (b) heartbeats of daphnids after 48 h (blue triangle), 72 h (black square), and 96 h (red circle) of exposure to different auq/nC60 concentrations. 2.3.3 Singular acute toxic impacts of solvents on daphnids The conditions for the singular acute toxic assessment are summarized in Table 2. 1. The alternative solvents showed lower toxic impacts on daphnids than the baseline (TMB). Figure 2.5b and Figure A4 show the dose-response curves for the baseline and alternatives. TMB had the highest toxicity (EC50= 707.8±166.0 mg/L) and sunflower oil the lowest (EC50 =159,920.7 ± 27,529.2 mg/L) (Figure 2.5b). The toxicity of olive oil (EC50= 54,485.4±7,645.1 mg/L) and linseed oil (EC50 = 17,358.2 ± 6.4 mg/L) was significantly lower than the baseline (TMB). Figure 2.5c shows the daphnids’ morphological change after exposure to TMB and alternative solvents. There were no significant changes for daphnids in essentials oils compared to the control, but TMB seems to cause torsion in the gut and body. 25 Figure 2.5. Mortality of daphnids in singular toxic experiments after 48 h exposure to toxicants. (a) Dose-response curve of daphnids exposed to TMB (red), linseed oil (black), olive oil (blue), and sunflower oil (green), (b) EC50 of baseline versus alternative solvents, (c) morphology of daphnids exposed to baseline C3, alternative solvents C4-C6 compared with the control C2 and culture C1. The average % of dead daphnids with a 95% confidence interval was used for (a) and (b). Although the EC50 levels calculated in this work were very high, it is possible that such concentrations would also be released in a freshwater environment. An accidental spill of concentrated fullerenes could happen for various reasons. C60 is dissolved in olive oil and sunflower oil at high concentrations when used as a human supplement. High concentration release in the freshwater environment could happen during manufacturing, but more likely during the transportation and use phase of the product. Fullerenes in a variety of solvents are considered safe and can be bought directly from a variety of sources, including Amazon, research groups, and healthcare companies [119–127]. There is an increased chance of improper disposal in the environment of products that are considered safe for humans [128]. For other applications, release in the environment is more likely to occur during the manufacturing stage. The purity of C60 depends on the application. Solar cells require very high purity, while for composite applications, 26 for example, the purity requirement is much lower. The manufacturer could transport lower grade material in solution before repeating the purification process to increase the purity of C60. For the intermediate manufacturing stage, the transport of C60 in solvents is cheaper since the solvent does not need to be removed and the fullerenes dried to make a powder. The toxicity of TMB, like other organic petroleum-based solvents, depends more on its water solubility than its chemical structure [129]. Essential oils are not miscible in water. Therefore, water solubility cannot explain the difference in toxicity. The composition of WSF from essential oils is shown in Table 2. 3 and Figure A5-7. The antioxidant production capability of essential oils seems to affect the toxicity of essential oils. Linseed oil was found to be more toxic than olive oil, which was more toxic than sunflower oil. On the other hand, sunflower oil is a stronger antioxidant than olive oil, while linseed oil produces reactive oxygen species [130–132]. Table 2.3. The water-soluble fraction composition for different essential oils fatty acids Total Concentration of Saturated Unsaturated fatty acids (%) Solvent fatty acids (mg/L) fatty acids Monounsaturated Polyunsaturated (%) Linseed oil 575.42 ± 69.35 10.83 ± 0.81 75.01 ± 1.63 14.17 ± 2.89 Olive oil Under the detection limit 52.81 ± 8.37 42.98 ± 8.38 4.21 ± 1.22 Sunflower oil 2,134.12 ± 1,412.74 12.53 ± 6.94 45.32 ± 1.27 42.14 ± 8.19 2.3.4 Combined toxic impacts on daphnids The mortality of daphnids in combined toxicity experiments differed from singular toxicity experiments (Figure 2.6). For the combined toxicity, the control was 40 mL of solvent plus 40 mL of reconstituted water. Different concentrations of auq/nC60 were used to investigate the impacts of C60 on the toxicity of solvents. In general, a higher concentration of C60 increased the mortality of daphnids for all combined experiments (Table 2. 1, C1-C4). For the mixture of C60 and linseed oil, the combined toxicity was higher for all C60 concentrations compared to the control (only 27 linseed oil). For the mixture of C60 and olive oil or sunflower oil, similarly to what was observed for the mixture of C60 and TMB, the addition of C60 initially reduced the toxicity of the solvent. At higher C60 concentrations, the toxicity exceeded the control. The antagonism and synergism effects of C60 on solvents are available in Appendix A, Table A6. For TMB, the addition of C60 at 11 ppm reduced the toxicity of TMB while the mortality of daphnids did not change at concentrations between 11 and 88 ppm and increased from 5% to 20% at 176 ppm. For olive oil, the addition of C60 reduced the mortality of daphnids from 33% to 13% at 11 and 22 ppm and to 20% at 44 ppm, while mortality did not change at 88 ppm and increased to 40% at 176 ppm. For sunflower oil, the addition of C60 reduced the mortality of daphnids at 11 ppm while it did not change at 22 ppm and increased after 44 ppm. The morphology of daphnids after 48 hours of exposure to combined toxicants is presented in Figure 2.6e. The deformity was observed in Daphnids as abnormal antennae (e.g., Figure 2.6 e5), carapace (e.g., Figure 2.6 e5), and curved spine (e.g., Figure 2.6 e6). For TMB, fullerenes were aggregated in the D. magna’s gut, as shown in Figure 2.6 e2-e3, while for essential oils, the presence of fullerenes inside and/or outside of daphnids varied based on the concentration of C60 (Figure 2.6 e5, e6, e8, e9, e11, and e12). For essential oils with the lowest C60 concentration (11 ppm), there was no fullerene observed inside the daphnids, while, for linseed oil and sunflower oil, sticky compounds containing fullerenes were observed on the surface of daphnids. For essential oils with the highest C60 concentration (176 ppm), fullerenes were observed inside and outside the daphnids (Figure 2.6 e9 and e12). The pH was measured at the end of each experiment, as a pH lower than 5 is toxic for daphnids [130]. 28 Figure 2.6.The pH of the test medium at the end of each experiment and the average mortality of daphnids for the acute combined toxicity experiments. Daphnids were exposed to (a) TMB + auq/nC60, (b) linseed oil + auq/nC60, (c) olive oil + auq/nC60, and (d) sunflower oil + auq/nC60. The average mortality of daphnids exposed to the control (50% solvent and 50% reconstituted water) for each mixture is shown in the red box. The average percentage of dead daphnids with 95% confidence intervals was used for (a)-(d). (e) The morphology of daphnids after 48 h of exposure to combined toxicants (C60 + solvents) and controls (reconstituted water +solvents). In figure captions, water = reconstituted water. For all the combined toxicity experiments, the higher concentrations of C60 increased the mortality of daphnids (Figure 2.6 a-d). The physiochemical properties of fullerenes, such as surface charge and particle size (Table 2.2), showed that the nano-properties were not the only reason for their toxicity. While the light microscopic images did not provide enough details to establish their exact location, at high concentrations, C60 were observed both inside and outside of the D. magna (Figure 2.6 e6, e9, and e12). The increased toxicity with increasing C60 concentration can be explained by two different effects, depending on whether the fullerene could reach the interior of the D. magna or not. Fullerenes are known to be photosensitizers [96] that can be photoactivated in the presence of oxygen and generate reactive oxygen species, such as superoxide anions, hydroxylic radicals, hydrogen peroxides, and singlet oxygens, all of which cause oxidative stress in the cell structure [131,132]. Photooxidation mostly happens outside of the cell since the 29 cell absorbs some parts of the wavelengths required for photoactivation [97]. If fullerenes did not enter the D. magna, the increasing concentration of C60 at the surface of the D. magna and the production of reactive oxygen species due to light exposure could increase the D. magna deaths. If fullerenes were able to reach the inside of the cells, the generation of singlet oxygens species from the oil could cause lipid peroxidation, which is toxic to the cell [133]. A low concentration of C60 reduced the mortality of daphnids in olive oil (Figure 2.6 c) and sunflower oil (Figure 2.6 d) but increased for linseed oil. The increased toxicity in linseed oil could be due to the immobilization of the D. magna. For the other two essential oils, the low fullerene concentration at the surface of the daphnids was not enough to cause death by photooxidation. Fullerenes are also known to have antibacterial properties [95,108,134] that could have helped daphnids overcome the oxidative stress from generated singlet oxygens. Environmental oxidative stresses (internal or external) can disturb the self-regulation of reactive oxygen species (ROS) in cells, which leads to an imbalance in ROS production and the cell’s ability to detoxify reactive intermediates [135–137]. Superoxide (O2-) is a normal byproduct of energy production in cells [138,139]. The superoxide leaks out of the mitochondria and is converted into hydrogen peroxide (H2O2) inside the cell [139,140]. Under normal conditions, enzymes, such as catalase and superoxide dismutase, convert O2- and H2O2 to H2O and O2 and reduce the damage to the cell [141]. Previous work showed that C60 could pass through the cell membrane and cause oxidative stress [142]. In this work, normal conditions might remain stable at low C60 concentrations. With an increase in C60 concentration, smaller oxidative stress might initially cause a small disturbance that cells were able to tolerate. With a further increase in C60 concentration, moderate oxidative stress would likely provoke apoptosis [137]. Finally, at higher C60 concentrations, the resulting oxidative stress might be severe enough to cause necrosis in cells 30 due to high environmental oxidative stress. Our work shows that singular toxicity can differ from combined toxicity assessments for C60, depending on the solvent. Therefore, results from singular toxicity should not be used for LCA and ERA as it is commonly done for nanomaterials. Additional work is required to calculate the EC50 of the C60-olive oil combination and used it in the LCA of fullerene purification. This study evaluated linseed oil, olive oil, and sunflower oil as replacements for TMB in fullerene manufacturing. All of the essential oils were found to have lower singular toxicity than TMB. This work also evaluated combined toxicity, which was ignored in previous solvent selection studies that used green chemistry principles. Olive oil was selected as a non-hazardous solvent for fullerene production since its interactions with fullerenes decreased the singular toxic effects on daphnids. There is a rapidly growing market for fullerenes, and at this stage, finding a safe alternative solvent for their manufacturing could reduce the harmful impacts of an accidental spill. While olive oil seems to be an acceptable solvent, additional work is needed to adapt current purification methods for olive oil due to the difference in viscosity and boiling point. Though the toxicity of fullerenes is well known and has been characterized alongside many organic solvents, it has yet to be examined alongside essential oils, which are ubiquitous dietary supplements. Additional work is required to better understand the impact of fullerene concentration on toxicity and identify the exact location of fullerenes in daphnids. 2.3.5 Acknowledgment This research was supported by the National Science Foundation under Grant NSF-1511098 and the Michigan State University Discretionary Fund Initiative. 31 Chapter 3 Environmental, cost, and chemical hazards of using alternative green solvents for C60 purification C60 is widely used in emerging technologies such as organic photovoltaic (OPV). However, existing purification processes require a large amount of toxic solvents such as 1,2,4- trimethylbenzene (TMB). Identifying an alternative green purification process before large-scale production can reduce the environmental impact of OPV. This Chapter evaluated existing fullerene purification methods, including chromatography, crystallization, and complexation, to identify the baseline process with the lowest environmental, cost, and chemical hazard impacts. We used life- cycle analysis, green chemistry principles, and toxicity assessments as environmental engineering decision-making tools to develop a replacement for the baseline via an iterative approach. Introduction Fullerene (C60) has gained remarkable attention due to its unique physicochemical properties, such as high chemical reactivity, making it a desirable material to associate with various compounds [143]. C60 can also be an electron donor or acceptor in donor-acceptor systems [144]. Its industrial uses include solar photovoltaics and lasers [145]), drug delivery and cancer therapy [99], cosmetic products [146], and human supplements. Fullerenes come from the combustion of hydrocarbons under reduced oxygen conditions to produce fullerene-containing soot [147]. After separation of the fullerenes from the soot, purification is required to increase the purity of C60, C70, and higher fullerenes, depending on the application. However, purification was identified as the most energy-intensive stage of making C60 [21], which requires a large amount of toxic solvents that may increase the environmental impacts of C60. 32 The principal C60 purification methods are chromatography, crystallization, and complexation [143]. The stationary phase contains alumina, silica gel, and activated carbon for chromatography, and the mobile phase is toluene, o-xylene, or hexane, depending on the target material [148–155]. Various materials have different affinities to the stationary phase, which causes the separation of target materials from the rest. The purification yield for C60 with 99% purity was about 50% using chromatography [150,156]. However, the chromatography technique is time-consuming and limited by the capacity of the column and the low C60 solubility of solvents. In addition, it requires a large amount of stationary phase and mobile phase, increasing the cost and carbon footprints of the purification process [143]. Another drawback is the irreversible absorption of C60 by the stationary phase, which reduces the purification yield [150]. The crystallization method requires large amounts of solvents and energy. Crystallization is based on the different solubility of C60, C70, and higher fullerenes at specific temperatures [157,158]. For example, crystallization can be used to separate fullerenes in xylene since the solubility of C60 is maximal at 30 oC and decrease with increasing temperature, while the solubility of C70 increases continuously from -20 to 90 oC [157]. Crystallization temperatures were reported at 110 oC for the C60-containing solution and -16 oC for the C70-containing solution [158]. A large amount of energy is required to either cool the solution to -20 oC or warming up to 90 oC, which is not recommended based on the 6th principle of green chemistry (design for energy efficiency). High purity (>99%) cannot be achieved in a single step, and therefore the crystallization process is repeated multiple times. Two crystallizations are required for C60 to reach 99% purity [158]. The C60 production yield was about 67%. In the complexation method, lewis acids such as AlCl3 were used to form a complex with C70 and higher fullerenes in carbon disulfide (CS2), while C60 was left in solution [159]. The 33 production yield of 99% C60 was 76%, which is higher than chromatography and crystallization. However, the high affinity of AlCl3 to water limits the choice of alternative solvents for the highly toxic CS2. Other studies proposed using a bicyclic amidine, typically 1,8- Diazabicyclo[5.4.0]undec-7-ene (DBU), to selectively form a complex with fullerene following the complexation order of C>70, C70, and C60 [58,160]. They were able to extract 76% of C60 with 99% purity at a large scale and retrieve C70 and higher fullerenes. However, this process requires a large amount of toxic solvents such as 1,2,4-trimethylbenzene (TMB). Life-cycle assessment (LCA), green chemistry, and toxicity assessment are widely used to reduce the environmental impacts of the nanomaterial synthesis process. LCA is a systematic method used to evaluate the potential environmental effects of products and procedures during their life cycle [77]. LCA has been used to quantify the global warming potential (GWP) and cumulative energy demand (CED) of carbon nanomaterial’s environmental impacts [161–164]. Green chemistry is a decision-making tool to design chemical products and processes to reduce using and generating hazardous materials [66]. Green chemistry principles were proposed to maximize the production rate and minimize adverse impacts of nanomaterial production [165,166]. Toxicity assessment is to measure the actual effects of materials on the environment [167]. It was employed as a decision-making tool to identify potential replacements for toxic solvents used for C60 production [168]. A previous study on fine chemical production proposed a framework based on an iterative approach using green chemistry principles and life cycle analysis to determine alternative methods with lower environmental, cost, and chemical hazard impacts [104]. Existing C60 purification methods require a large amount of hazardous solvents. This Chapter used LCA, green chemistry principles, and toxicity assessments to identify green alternative solvents for C60 purification. First, we determined the baseline process with the lowest 34 environmental, cost, and chemical hazard impacts among existing C60 purification methods. Then we used LCA to identify the environmental hotspots of the baseline. Next, green chemistry and toxicity assessments were employed to determine potential alternative solvents for the baseline. LCA, green chemistry principles, and toxicity assessments were used continuously to modify the baseline via an iterative approach to determine a C60 purification process with less environmental, cost, and chemical hazard impacts than the baseline. The original idea for this Chapter was from Eunsang Lee, who conducted the environmental and cost impact assessment of existing purification methods. I quantified the amount of materials needed for each existing purification process and calculated the hazardous score of each method. Ben Cecil contributed to the life cycle cost calculation. He compiled the price of energy materials based on my life cycle inventory analysis (Table B7). I provided the production rate (Table B6), and he conducted the calculations for metric cost values (Table B9). 3.1 Materials and methods 3.1.1 Materials All chemicals were used as received. Fullerene mix (70% C60, 29% C70, and 1% higher fullerenes) was purchased from SES Research Group (Houston, TX), C60 and C70 with +99% purity from MER (Tuscan, AZ), 1,2,4-trimethylbenzene (TMB) CAS 95636 (+98%), toluene CAS 108883 (ACS reagent, +99.5%), xylenes CAS 1330207 (ACS reagent, +98.5%), linseed oil CAS 8001261 (MQ200), olive oil CAS 8001250 (MQ200, highly refined), hexane CAS 110543 (HPLC grade, +85% n-Hexane), 2-propanol CAS 67630 (HPLC grade, 99.9%), 1,8- Diazabicyclo[5.4.0]undec-7-ene (DBU) CAS 6674222 (puriss., +99% GC), 2- methyltetrahydrofuran CAS 96479 (+99%), were purchased from Sigma Aldrich Burlington, MA. 35 3.1.2 Environmental impact assessment We used life-cycle assessment (LCA) to evaluate the environmental impacts of producing one kg of C60 with 99% purity. LCA was conducted using SimaPro 9.1.0.7 [169] to model C60 purification processes and analyze inventory data, including Ecoinvent 3.3, US-EI from DATA SMART, published articles (Table B1 in Appendix B), and data collected during experiments to quantify associated carbon footprints. We used the TRACI 2.1 V1.05 method for Global warming potential (GWP) analysis, Cumulative Energy Demand V1.11 method for energy assessment, (Water Scarcity) V1.00 method for water demand (WD) analysis [170], and E-Factor calculation [171]. 3.1.3 Cost and chemical hazard analysis The cost and chemical hazard evaluation were explained in the previously published paper [104]. In summary, chemical hazard analysis was based on NFPA 704 standard scores, including health hazard, flammability hazard, reactivity hazard, and special hazard [172]. Cost assessment was based on the production rate of the target substrate and the life cycle cost (LCC) of required raw materials. These values were normalized using the values of the baseline process, then averaged into a single cost metric value for analysis. 3.1.4 Toxicity assessment This study used Toxicity Estimation Software Tool (TEST, V5.1 [29]) to estimate the toxicity of petroleum-based solvents. The U.S. Environmental Protection Agency (EPA) developed TEST using Quantitative Structure-Activity Relationships (QSARs) methodologies to allow users to predict chemical toxicity based on molecular structures. We employed TEST to compare the petroleum-based solvents’ toxicity based on acute toxicity (48 h) results for Daphnia Magna. The toxicity of plant-based oils was from previous work [20]. 36 3.1.5 Green chemistry principles Green chemistry principles were described in Table B2 based on Jessop et al. work [24]. This study identified alternative solvents with high C60 solubility (to reduce the amount of required solvents – principle #5) and lower toxicity (to reduce the risk of accidental release and avoid toxifying the final products – principle #12 & 4). We also considered plant-based solvents in accordance with principle #7, which is about using renewable feedstock. 3.1.6 Experiments The baseline process was identified by evaluating the environmental, cost, and chemical hazards of existing purification methods. In the baseline, the fullerene mix, including C60, C70, and higher fullerenes, was dissolved in five mL of TMB using an ultrasonic bath for five min at room temperature [57,58]. The complexation process started by adding 103 µl of DBU and four µl of DI water in 10 min while the solution was continuously mixed at a low speed to avoid a vortex. The complexation continued for five hours under the nitrogen atmosphere to let DBU and C70 and higher fullerenes form a solid complex. They used a 0.2-micron filter to separate the complex containing C70 and higher fullerenes from the solution containing C60. The remaining DBU was removed from the procedure using 0.2 M acetic acid. Isopropanol was used to separate C60 particles from the solution via the crystallization process in 24 h. Finally, C60 particles were collected for drying to remove remained solvents. We used an iterative method [104] to modify the baseline process (Figure 3.2). The iterative approach allowed us to evaluate purification yields and apply in-process modifications. As shown in Figure 3.2, the iterative approach started with the baseline, followed by initial material characterization, screening processes based on C60 purity and yield, and sustainability evaluation of potential alternative procedures with a comparable purity and yield to the baseline. The best 37 alternative purification process was identified based on environmental, cost, and chemical hazard effects. High-performance liquid chromatography (HPLC-UV) was used to measure C60 purity. The column was YMC-Pack ODS-ATM (5 µm, 120Å, 150 × 6 mm I.D.), the mobile phase was hexane/2-propanol (70/30), the flow rate was 0.7 mL/min, the wavelength was 350 nm (0.08 AUFS), and injection was four µL [173]. Figure 3.1. The experimental approach to identify alternative C60 purification methods for the baseline process 3.2 Results 3.2.1 Existing C60 purification methods Complexation with DBU was selected as the baseline process. The environmental, chemical, and cost evaluations of existing purification methods were conducted in collaboration 38 with colleagues. Figure B1, B2, and B3 in Appendix B present system boundary and material flow of existing commercial C60 purification methods, such as complexation (Nagata, 2005), crystallization (Kwok, 2010), and crystallization (Grushko, 2007), respectively. 3.2.2 Baseline evaluation Figure 3.3 shows the environmental evaluation (CED) results of the complexation process from extracting raw materials to produce one kg of C60. We could track energy consumption in the C60 purification process with CED to identify the associated environmental hotspots. TMB contributed to more than 95% of the total CED of the baseline purification process because of the amount of solvents and the embodied energy of TMB. It was determined as the environmental hotspot, which required a modification. Figure B1 presents the system boundary and procedure details. Figure 3.2. Environmental evaluation of the baseline purification process (complexation with DBU) 3.2.3 Alternative solvents for TMB TMB is a highly toxic solvent that needed to be replaced by nontoxic or less toxic solvents. We used C60 solubility and toxicity score to identify alternative solvents in accordance with the following green chemistry principles (Table B2 - #3: use substances with no or little toxicity, #4: design safer chemicals with lower environmental toxicity, #5: design to use less amount of 39 solvents, #7: use renewable feedstocks, and #12: safer substrates for accident prevention). Figure 3.4 shows the toxicity scores and C60 solubility of available solvents. The purple area presents solvents with a lower toxicity score than TMB and comparable C60 solubility to TMB. Two groups of solvents were selected as potential replacements: plant-based and petroleum-based solvents. Linseed oil and olive oil were selected since they are nontoxic [174], have a higher solubility than TMB, and are extracted from renewable feedstocks. Xylene and toluene were chosen because they are also less toxic than TMB and have a comparable C60 solubility with TMB. The normalized toxicity score was 0.24 for TMB, 0.17 for xylene, 0.12 for toluene, 0.12 × 10-3 for linseed oil, and 0.38 × 10-4 for olive oil. The normalized solubility score was 0.34 for TMB, 0.18 for xylene, 0.16 for toluene, 1.0 for linseed oil, and 0.44 for olive oil. Table B3 presents detailed information on solvent toxicity scores and C60 solubility. Figure 3.3. The toxicity and C60 solubility of solvents to identify replacements for TMB. The purple area highlights potential alternative solvents. Table B3 presents the detailed information of solvents #1 to #24. 40 3.2.4 Purification experiments 3.2.4 1 Modified baseline process The modified baseline had lower environmental, cost, and chemical hazard impacts than the baseline. The baseline purification yield was about 44%. The low yield was originated from filtration, where we lost a considerable portion of C60. A centrifuge was used to increase the production yield to 76% resulted in fewer environmental, cost, and chemical hazard impacts. The modified baseline had about 43% less chemical hazards and environmental impacts and was 22% cheaper than the baseline. We conducted experiments to reduce the environmental, cost, and chemical hazards of the modified baseline. Figure B4 demonstrates the material flow of the modified baseline. 3.2.4 2 Alternative plant-based oil solvents (P1 to P6) Linseed oil and olive oil were chosen as plant-based replacements for TMB because they are less toxic, have a comparable C60 solubility, and are extracted from renewable feedstocks. Various methods were proposed to dissolve fullerenes in linseed oil and olive oil. In P1, fullerene was dissolved in olive oil, stirring for 72 h at room temperature under a nitrogen atmosphere, as Cataldo et al. suggested [175]. C60 purification yield was lower than 5% because the fullerene-mix particles were not fully dissolved in olive oil at the beginning of the experiment and probably complexation only occurred at the surface. Therefore, C70 and higher fullerenes did not form a solid complex with DBU during the complexation. In P2, the fullerene mix was added into olive oil at 75 °C while the solution was stirred for two hours under a nitrogen atmosphere, as Cataldo et al. suggested [176]. The production yield was still lower than 5% since the oil decomposition occurred at high temperatures and reacted with C60. P3 was based on a suggestion to use a sonication bath for 15 min at 50 oC to dissolve fullerenes in olive oil [177–179]. The production 41 yield increased to 26%, much higher than P1 and P2 but still was lower than the modified baseline production yield. In P4, we replaced the sonication bath with an ultrasonic probe to improve the amount of dissolved C60 in olive oil before the complexation. 100 mg of the fullerene mix was dissolved in four mL of olive oil using the ultrasonic probe for three minutes while the container remained in a water bath to keep the temperature close to room temperature. The production yield increased to 44%. P5 and P6 were to identify the best production yield for the purification method using linseed oil instead of TMB. Linseed oil toxicity score is not less than olive oil, but its C60 solubility is higher than olive oil solubility, which could reduce the associated impacts to the C60 purification process. In P5, we dissolved 100 mg of the fullerene mix in two mL of linseed oil using an ultrasonic bath for 15 min at 55 oC as was recommended in the literature [177–179]. The production yield was 36%. P6 was based on using the ultrasonic probe for three min at room temperature. The production yield was 61% in P6. Figures B5 and B6 illustrate the material flow for P4 and P6. 3.2.4 3 Alternative petroleum-based solvents (P7 to P8) C60 purification methods using toluene and xylene had lower environmental, cost, and chemical hazard impacts than the modified baseline. Toluene and xylene were selected as petroleum-based replacements for TMB due to their high C60 solubility and low toxicity score compared to TMB (Figure 3.4). 100 mg of the fullerene mix was dissolved in 7.5 mL of xylene in P7 and eight mL toluene in P8. The rest of the procedures were similar to the modified baseline. The C60 production yield was 70% for P7 and 77% for P8. Figures B7 and B8 show the material flow for P7 and P8. Fig B10 demonstrates the HPLC analysis of the baseline, modified baseline, P4, P6, P7, and P8. 42 3.2.5 Environmental, cost, and chemical hazard evaluation of potential methods 3.2.5 1 Chemical hazard assessment Chemical hazard impacts were quantified for potential alternative solvents as well as alternative purification methods. Table B4 presents the detailed calculation of solvents’ chemical hazard impacts. The chemical hazard score was 1.3 for TMB, 2.0 for toluene and xylene, 0.33 for linseed oil, and 0.67 for olive oil. The chemical hazard score of alternative purification methods was calculated and normalized to the baseline (Table B5). The modified baseline chemical hazard score was 57. The highest chemical hazard score was 187, which was for the purification method using olive oil (P3). This is because olive oil regeneration requires a large amount of energy due to its high boiling point (447 K [180]); therefore, all of the olive oil used in the process was counted for chemical hazard evaluation. The chemical hazard score was 34 for the method using linseed oil (P6). It was almost six times lower than the chemical hazard score of P3 since less amount of solvent was required due to the higher C60 solubility of linseed oil. The chemical hazard score was even lower for the alternative methods using petroleum-based solvents mostly because toluene and xylene can be regenerated; therefore, less amount of solvents were considered for chemical hazard evaluation. The lowest chemical hazard score was 13 for the purification process using toluene instead of TMB. 3.2.5 2 Cost assessment The C60 purification process using toluene (P8) had the lowest cost impact. Table B6 shows the baseline and alternative purification methods' production rate. The life cycle cost was calculated based on raw materials and energy needed to produce solvents and materials used for C60 purification (Table B7 and B8). The processes using xylene and toluene as alternative solvents had lower cost impacts due to their low cost of production from crude oil sources (Table B9). 43 3.2.5 3 Environmental impact assessment The C60 purification process using toluene (P8) had the lowest environmental impacts (GWP, CED, WD, and E-factor). Table B10 presents the LCA details for the baseline, modified baseline, and alternative (P3-8) purification methods. The modified baseline had about 57% less environmental impact than the baseline because of a higher purification yield than the baseline, resulting in using less materials to produce one kg of C60. C60 purification with olive oil (P3 and P4) had the highest environmental impact score. It was almost five times higher than the baseline and ten times higher than the modified baseline. Olive oil production requires a large amount of energy, water, and substrates, increasing associated environmental impacts. C60 purification with linseed oil (P5 and P6) had lower environmental impacts than purification with olive oil but had higher environmental impacts than the modified baseline. C60 purification with xylene (P7) had lower environmental impacts than purification with olive oil, linseed oil, the baseline, and the modified baseline. C60 purification with toluene (P8) had the lowest environmental impacts compare to the baseline, modified baseline, and other alternative methods. Because toluene can be regenerated and reused in the process, it resulted in consuming less amount of materials necessary for C60 purification. The environmental impact score of C60 purification with toluene was about 60% less than the baseline and 28% less than the modified baseline. 3.2.6 Overall evaluation The C60 purification process using toluene was identified as a green replacement for C60 purification. Figure 3.5 summarizes the evaluation of baseline, modified baseline, and alternative methods (P3-8) for C60 purification based on environmental, cost, and chemical hazard impacts. Even though linseed oil and olive oil are nontoxic plant-based solvents with high C60 solubility, the life cycle analysis showed that the total impacts of using them were higher than using 44 petroleum-based solvents (e.g., toluene and xylene). It was primarily due to the large amount of energy, water, and substrates necessary for producing olive oil and linseed oil. Also, C60 purification yield was low for purification methods using olive oil and linseed oil, which increased the materials necessary for making one kg of C60 and consequently affected associated environmental, cost, and chemical hazard impacts. Figure 3.4. The evaluation of baseline, modified baseline, and alternative methods for C60 purification based on environmental, cost, and chemical hazard impacts. 45 3.3 Conclusion Before large-scale production, identifying a green process for C60 purification will reduce the environmental impacts of emerging C60-containing products, such as organic transparent solar cells. Existing purification processes require a large amount of toxic solvents resulting in high environmental, cost, and chemical hazard impacts. The complexation is the best existing purification process; however, more than 95% of the environmental impact (CED) is due to using toxic TMB. Olive oil, linseed oil, xylene, and toluene were identified as potential alternative solvents for TMB since they are less toxic and have a comparable C60 solubility to TMB. An iterative approach was used to employ LCA and green chemistry to determine if using potential alternative solvents could result in a greener purification process. Identifying a green solvent is highly dependent on the production process of the target product. Olive oil and linseed oil are nontoxic and have a higher C60 solubility than TMB, xylene, and toluene. However, the plant-based oil production process requires more energy, water, and materials than xylene and toluene in the upstream and the purification stage. Therefore, considering the life cycle of producing potential alternative solvents from raw materials and the production stage of the target product is vital for green solvent selection. It is also essential to study the chemical hazard of using potential alternative solvents. Sometimes selected solvents require extra materials to have a similar capability as the baseline solvent has. For example, olive oil is a nontoxic solvent, but a large amount of heptane, compared to what requires in the baseline, is necessary to separate C60 from C60-containing solution results in a higher chemical hazard score. Acknowledgments This material is based upon work supported by National Science Foundation under Grant NSF- 1511098. 46 Chapter 4 Material requirement and resource availability for silicon photovoltaic laminate manufacturing in the next ten years b Material scarcity is a considerable threat to energy transition towards renewables. Photovoltaics (PV) installations are expected to increase rapidly in the next decade, which may increase the amount of material needed. This Chapter created three scenarios (S1, S2, & S3) to evaluate the impacts of potential technology improvements on the amount of materials necessary for manufacturing silicon PV (Si PV) laminate in the next ten years for the US and the rest of the world. The baseline was similar to previous studies, which applied theoretical models on PV historical data and ignored PV technology improvements that can influence future material projections. S1 considered only market share and module efficiency, while S2 covered wafer thickness improvements as well. S3 was the scenario that more likely will occur in the next decade, which included module efficiency, market share, wafer thickness, glass thickness, and potential replacements such as using perovskite/silicon tandem. This Chapter quantified the materials needed for Si PV laminate manufacturing in the next decade in the US and globally. We also highlighted the importance of considering technology improvements to project the PV material requirement. 4.1 Introduction The global cumulative PV installation increased from one GW in 2000 to 480 GW in 2018 and is expected to reach 8,519 GW by 2050 [6][181]. PV technologies’ contribution to electricity generation from renewable resources was less than one percent in 2010, while it is estimated to b Parts of this chapter were presented at the 48th IEEE -PVSC conference (June 2021). 47 increase from 13% in 2018 to 48% by 2050 in the USA [182]. However, the energy transition to renewable energy faces various challenges related to material availability that should be carefully explored [183]. Material scarcity is a potential threat to PV deployment. In 2019, silicon PV (Si PV) was 95% of the PV market [6]. Si PV is anticipated to remain the primary type of PV technology (>50%) until 2040 [22]. Most previous studies on the material need for Si PV used theoretical models such as material flow analysis [184], the total material requirement [185], Hubbert peak [8][9], and historical data to project future material necessary for manufacturing Si PV in the future need. A few studies considered market share and efficiency changes while ignoring changes in technology and manufacturing [188–190]. In this Chapter, we calculated the materials such as metallurgical grade silicon (MG-Si), solar glass, aluminum, copper, lead, silver, and tin needed for manufacturing Si PV laminate to meet the expected annual Si PV installation in the next ten years for the US and rest of the world. We considered module efficiency, the percentage of Si PV sub-technologies in the PV market, silicon wafer thickness in cells, glass thickness in laminate, sawing techniques, and other potential technology improvements such as using silicon/perovskite tandem. 4.2 Methods 4.2.1 Study scope This study considered materials necessary for manufacturing Si PV laminate. Figure 4.1 shows the study scope, which contains mining raw materials, solar-grade material production, wafering, cell processing, and PV laminate manufacturing. MG-Si is required for producing solar- 48 grade silicon. Silver and aluminum are needed for metallization paste at the cell production stage. Copper, tin, lead, and solar glass are considered for PV laminate. Figure 4.1. The study scope to quantify materials needed for manufacturing Si PV laminate. 4.2.2 Material requirement Materials necessary to support Si PV were calculated based on data compiled from literature, energy outlooks, and the annual reports of leading Si PV manufacturers. The annual PV installation was based on the 2020 DNV-GL Energy Transition Outlook (Figure C1 in Appendix C) [13]. PV technology improvements, such as efficiency, wafer thickness, wafer sawing, glass thickness, and material replacement, were determined based on leading manufactures’ annual reports and energy outlooks. I identified the required materials for the baseline and three scenarios (S1, S2, & S3) for the USA and the rest of the world (Table 4.1). The baseline assumptions were similar to previous PV studies that used theoretical models and PV historical data to calculate the necessary materials for PV manufacturing and ignored technology changes. S1 and S2 were similar to previous studies where some part of technology improvement was considered to estimate materials required for making PV. S3 considered Si PV sub-technologies' efficiency and contribution to the PV market, silicon wafer thickness, glass thickness, sawing techniques, and perovskite/silicon tandem. 49 Table 4.1. Assumptions for each scenario to project materials needed for Si PV manufacturing Scenario Description Baseline Efficiency, market share, and current PV technologies remain constant until 2031 2021 2031 Sub-technology Market c-Si* 35.5 [3] 28.22 [191] share (%) mc-Si** 64.5 [3] 40.61 [191] S1 Efficiency (%) c-Si 22 [3] 23.5 [191] mc-Si 19 [3] 22 [191] Sub-technology Market c-Si 35.5 [3] 28.22 [191] share (%) mc-Si 64.5 [3] 40.61 [191] Efficiency (%) c-Si 22 [3] 23.5 [191] S2 mc-Si 19 [3] 22 [191] Silicon wafer thickness c-Si 170 [4] 50 [39] (µm) mc-Si 180 [4] 50 [39] Al-BSF 3.44 0 Monofacial 14.58 5.82 PERC Bifacial 14 20.64 PERC c-Si Monofacial 1.05 1.45 HJT Bifacial HJT 1.01 5.16 Sub-technology Market Bifacial share in PV perovskite 0.97 (2023) 5.18 [39] PERC Al-BSF 6.35 0 Monofacial 26.89 5.8 PERC S3 Bifacial mc-Si 25.84 20.64 PERC Monofacial 1.94 1.45 HJT Bifacial HJT 1.87 5.16 Al-BSF c-Si 20 21.5 Al-BSF mc-Si 19 20 Efficiency PERC c-Si 20.7 22.2 [39] PERC mc-Si 19.5 21.5 HJT Si 21.5 23 Si/perovskite tandem 22.5 26 Silicon wafer thickness c-Si 170 [4] 50 [39] (µm) mc-Si 180 [4] 50 [39] Glass thickness (mm) [39] c-Si & mc-Si 4 2 *mono-crystalline silicon (c-Si), **multi-crystalline silicon (mc-Si) We used a bottom-up approach to determine the required materials for manufacturing Si PV laminate (Appendix C2). Figure 4.2 shows the six types of Si-based solar cells considered: aluminum back surface field (Al-BSF), mono facial PERC, bifacial PERC, mono facial Si heterojunction (HJT), bifacial HJT, and perovskite/silicon tandem. 50 Figure 4.2. The typical structures of Si PV. Al-BSF: Aluminum Back Surface Field, PERC: Passivated Emitter and Rear Cell, HJT: Heterojunction Technology. [192–196] 4.2.3 Resource availability Material availability for Si PV manufacturing was based on the United Nations Comtrade annual reports [197] and the United States Geological Survey Minerals Yearbooks [198]. Annual MG-Si and silica sand production were identified using yearly USGS reports [199]. 4.3 Results and discussion 4.3.1 Material requirements We estimated the amount of required materials for manufacturing Si PV laminate based on a linear regression between 2021 and 2031. Wafer thickness for Si PV was assumed to be 180 µm for mc-Si and 170 µm for c-Si in 2020 [4] and reduced to 50 µm by 2031 [200]. The 2021 market share for c-Si and mc-Si PV was 35.5 % and 64.5%, respectively [201]. The Si PV market share is expected to shrink to 25% for c-Si and mc-Si PV by 2040 [191]. Cadmium telluride (CdTe), copper indium gallium selenide (CIGS/CIS), and silicon tandem with perovskites are expected to cover the rest of the PV market. The efficiency was assumed to improve from 22% (c-Si PV) and 51 19% (mc-Si PV) [3] to 25% in 2040 [191]. However, this was higher than the Si PV efficiency reported for mass production. For example, the median efficiency of Al-BSF c-Si produced by leading global PV manufacturers is about 20% in 2021 and may reach 21.5% by 2031 [200]. In the 3rd scenario (S3 – Table 4.1), we considered market share changes and efficiency improvements for each Si PV sub technology for mass production; Al-BSF, mono facial PERC, bifacial PERC, mono facial HJT, bifacial HJT, and perovskite/silicon (Figure 4.2). Figure 4.3 shows the material required for manufacturing Si PV laminate to meet the global PV installation forecasts until 2031. Solar glass and MG-Si will have the highest demand in the next ten years. The baseline shows how material demands are increasing due to the increase in PV installation without considering any changes in PV technology. In the next decade, material demand will increase by 260% for MG-Si and 250% for solar glass (Figure 4.3A). The first scenario (S1 - Figure 4.3B) shows how efficiency improvements and market share changes may influence material needs in the next couple of years. For example, in 2031, about 1.04 million metric tons of MG-Si and 13.1 million metric tons of solar glass will be required for the baseline, while it is 0.69 million metric tons of MG-Si and 9.3 million metric tons of solar glass for S1. For the second scenario (S2 – Figure 4.3C), we considered the reduction in wafer thickness. For S2, the amount of MG-Si was 69% lower than S1. The 3rd scenario (Figure 4.3D) was based on data collected from mass production in global leading PV manufacturers. Similar to the baseline, S1, and S2, MG-Si and solar glass were the primary materials for making Si PV laminate. However, the material demand in 2021 was higher due to the low efficiency of massive Si PV production compared to the Si PV efficiency reported by previous studies. Manufacturing improvements such as the decrease in the thickness of Si wafer and glass as well as using emerging technologies such as perovskite will mitigate material demand. 52 The material demand is expected to be reduced after 2029 due to the possibility of replacing Al- BSF with other Si PV with higher efficiency. A B C D Figure 4.3. Material requirements for manufacturing Si PV to meet the electricity demands from 2021 to 2031 for various scenarios globally. (A) Baseline , (B) S1, (C) S2, and (D) S3. Material requirements for the US and the rest of the world are shown in C2 and C3. Figure 4.4 shows material requirements for manufacturing Si PV laminate for sub- technologies in the 3rd scenario. MG-Si and solar glass were the primary materials for each sub- technology. Required MG-Si for Al-BSF Si PV (Figure 4.4B) will decrease from 43,000 metric tons in 2021 to 3,000 metric tons in 2030 due to replacing Al-BSF Si PV with other Si PV technologies. The solar glass will also decrease from 585,000 metric tons in 2021 to 71,000 metric tons in 2031 PERC Si PV (Figure 4.4B) will need the highest amount of materials among Si PV technologies. PERC technology was 83% of the Si PV market and is expected to shrink to 72% in 2031 [200]. About 346,000 metric tons of MG-Si and 6.1 million metric tons of solar glass were needed in 2021 for manufacturing PERC Si PV laminate which will decrease to 177,000 metric 53 tons of MG-Si and 5.3 million tons of solar glass. HJT Si PV (Figure 4.4C) was 6% of the Si PV market in 2021 and is likely to be 18% of the Si PV market in 2031 [200]. MG-Si demand will increase from 30,000 metric tons in 2021 to about 55,000 metric tons in 2031. A similar trend was projected for solar glass, where the demand increased from 0.56 million metric tons in 2021 to 1.78 million metric tons in 2031. The last technology in the Si PV market is perovskite/silicon tandem which is expected to be a part of mass production after 2023 [200]. MG-Si demand for manufacturing perovskite/silicon tandem will increase about 2.5 times in the next couple of years. Approximately 4,000 and 14,000 metric tons of MG-Si will be required in 2023 and 2031, respectively. A B C D Figure 4.4. Material requirements for manufacturing Si PV technologies to meet the electricity demands from 2021 to 2031 globally. (A) Al-BSF Si PV, (B) PERC Si PV, (C) HJT Si PV, and (D) Si/Perovskite tandem. Material requirements for the US and the rest of the world are presented in C4-C11 4.3.2 Resource availability Figure 4.5 shows available silica sand and MG-Si resources in 2020. Industrial-grade silica sand is the primary material for producing solar glass and MG-Si. Some countries such as the US, 54 the Netherlands, and Spain had the highest amount of available silica sand resources in 2020 (Figure 4.5A). MG-Si is the other key material for the Si PV market, which is mainly produced in China. Other leading MG-Si producers were Russia, Norway, and the US in 2020 (Figure 4.5B). The primary stockpiles’ users are the aluminum and chemical industry for MG-Si and the concrete, roads, and construction industry for silica sand. In contrast, the PV industry consumed a small percentage of MG-Si and silica sand [199]. However, PV deployment growth may increase the dependency of Si PV producers on MG-Si and high-quality silica sand resources. This may result in competition between various stakeholders to supply material demand for Si PV manufacturing which may affect the environment due to the possibility of excessive mining or using illegal silica sand. Figure 4.5. Available industrial-grade silica sand (A) and MG-Si (B) in 2020. 4.3.3 Conclusion The material shortage is a potential issue for PV deployment. This Chapter showed the importance of considering PV technology improvements in material projections. We evaluated Si PV material demand for different scenarios to cover typical assumptions considered by previous studies for material requirement estimations. The baseline was representative of previous studies 55 that conducted analysis only based on historical data and ignored PV technology changes. The first and second scenarios were based on assumptions used by previous studies, where they focused only on efficiency and market share and ignored sub-technology deployment and manufacturing improvements. In the 3rd scenario, we projected material requirements considering improvements in the efficiency and market share of c- and mc-Si PV sub-technologies, glass thickness, and Si wafer thickness in the next decade. The estimated required materials in S3 were 22% to 78% lower than the baseline and 23% to 66% lower than 1st and 2nd scenarios. MG-Si and solar glass have the highest demand for Si PV laminate manufacturing. About 74 billion metric tons of solar glass and 3 billion metric tons of metallurgical grade silicon will be required in the next decade for Si PV laminate manufacturing. The availability of silica sand resources for MG-Si and solar glass production may cause some challenges for Si PV producers, which could result in excessive mining that may increase the socioenvironmental impacts of Si PV. There is also a possibility of using silica sand from illegal mines due to accessibility to enough resources. It is essential to consider third-party certifications to monitor the Si PV supply chain to ensure Si PV materials will be provided from legal mines without severe impacts on the environment. Acknowledgment This work was supported by the National Science Foundation grant number NSF-1801785. 56 Chapter 5 The country-specific footprint of metallurgical grade silicon production for silicon photovoltaics The carbon footprint of quartz mining and silica sand extraction is widely ignored in photovoltaic (PV) studies. The PV industry requires high-quality silica sand to produce metallurgical grade silicon (MG-Si) used to manufacture silicon PV (Si PV). However, high- quality deposits are scarce, and using lower-quality resources may increase the carbon footprint of Si PV modules. This Chapter quantified the current and future carbon footprint and cumulative energy demand (CED) of quartz mining, silica sand extraction, and MG-Si production. The life- cycle assessment was used to evaluate the carbon footprint of producing one kg of MG-Si using quartz deposits with different purity levels: high-quality (> 98% silica), industrial-grade (95% silica), and low-quality (65% silica). 5.1 Background Photovoltaics (PV) is a promising energy technology to reduce the carbon footprint of electricity production [202]. Cumulative PV installations have increased from one GW in 2000 to 480 GW in 2018 and will reach 8,519 GW by 2050 [6]. Silicon PV (Si PV) represents 97% of the current PV market and is expected to remain the dominant technology until 2040, but raw material shortage could reduce the market share of Si PV [203]. The raw material for Si PV is quartz which is mined to extract silica sand and purified to produce metallurgical grade silicon (MG-Si) [4] (Figure 5.1). Further purification is necessary to remove impurities such as Fe, Al, B, and P to produce solar-grade silicon [204] used in photovoltaic cells. 57 Figure 5.1. Study scope including quartz mining, silica sand extraction, and MG-Si production for manufacturing Si PV modules Life-cycle assessment (LCA) is widely used to evaluate the carbon footprint of Si PV. The quartz purity and availability were not considered in previous Si PV LCA studies. Figures (D1- D4) in Appendix D summarize Si PV LCA studies from 2000 to 2019. Previous work either focused on the carbon footprint of processes after producing MG-Si or ignored the extra steps necessary for extracting high-quality silica sand from low-quality and industrial-grade quartz resources to make MG-Si. Latter studies used the mining processes of a sand producer in Germany and only added a drying step to model high-quality silica sand extraction (Table D1 in Appendix D). This assumption is valid only if the quartz contains more than 98% of silica. However, only a few places in the world have quartz deposits with that level of purity. The rest of the available quartz resources are industrial (95% purity) or low-quality grade, which need more steps than high- quality quartz to remove impurities from quartz and extract silica sand. Industrial-grade silica sand extraction was modeled in a few studies but not the extraction and purification of low-quality silica sand. The carbon footprint of extracting industrial-grade silica sand for glass in Croatia was calculated to be 43 kg CO2eq per metric ton of silica sand [205]. In another study, the impact of 99% purity silica sand extraction for foundry application in Poland was calculated for natural resource consumption, ecosystem quality, and human health but not the carbon footprint [206]. According to annual reports from high-quality silica sand producers such as Unimin Corporation [207], extra processes are necessary to separate impurities from quartz. 58 Those extra processes require a large amount of fossil fuels; therefore, mining lower quality quartz may increase the carbon footprint of Si PV. In addition to the environmental impacts, there are potential social issues associated with quartz mining for Si PV manufacturing [208]. Current environmental regulations do not consider social impacts associated with mining [209,210]. Multiple frameworks have been suggested to investigate the social impact of the mining industry but not for silicon. Quartz mining undermines local livelihoods and has been identified as the world's most conflict commodity [211]. Silica sand extraction is growing, generating serious social concerns in India [211], China [212], and the USA [213]. In the USA, Wisconsin is the main industrial-grade silica sand provider for the hydraulic fracturing industry. Multiple reports from anti-fracking activists show that some people lose their land, others suffer from the destruction of the natural system, and all are exposed to new environmental health risks [213]. Some solutions, such as combining environmental cost estimation and using developed efficient extraction plans, may reduce the conflicts among silica sand producers and local residents [214,215]. All existing studies consider only legal mines and ignore the social impacts of extracting raw materials such as silica sand from illegal mines, likely due to the lack of information such as the mine’s location. The United Nations Environmental Program also started investigating quartz mining activities in 2018 but faced many obstacles due to the lack of information on the location of quartz deposits and their legality [216,217]. The increase in PV installations will increase the demand for silica sand. This Chapter evaluated the carbon footprint of silica sand extraction from various purity and location to meet the demand for Si PV manufacturing. We compiled the availability and the purity of quartz deposits in the USA, China, and the rest of the world for producing Si PV in the present and future. The main MG-Si producers in China and the rest of the world were identified. The amount of silica 59 sand and MG-Si production were compiled for Africa, America, Asia and Pacific, China, and Europe. We modeled quartz mining, silica sand extraction, and MG-Si production processes for the high-quality (>98% purity), industrial-grade (95% purity), and low-quality (65% purity) deposits to quantify the associated carbon footprint. Global warming potential (GWP) and cumulative energy demand (CED) were calculated for various scenarios to estimate the environmental impacts of silica sand extraction and MG-Si production at present and in the future. This Chapter also identified the location of illegal mines that MG-Si producers might use. 5.2 Methods The first step was to collect data on the location of quartz deposits and purification, the annual silica sand production per country, and the annual import/export amount of silica sand between leading producers and primary consumers. In the second part, we used life-cycle assessment to calculate the carbon footprint associated with quartz mining, silica sand extraction, and MG-Si production needed to manufacture Si PV. 5.2.1 Location of legal and illegal mines and MG-Si production Data about the location, purity, annual production of active legal mines in the USA, China, and other countries was collected from the 2020 U.S. Geological Survey (USGS) report [199], United Nations Comtrade Database (UN Comtrade) [197], companies’ annual report (Diatreme Resources Ltd [218] and Unimin (COVIA) Corporation [207]), and literature [216,219–239]. Data about the location of illegal mines was obtained from articles published in local and international news agencies from 2017 to 2020 [240–249], NGO reports [250,251], and published papers [252– 254]. The annual production and location of MG-Si producers were compiled from 2020 USGS reports [199,255]. 60 5.2.2 Environmental impact assessments 5.2.2 1 Life cycle assessment (LCA) The objective of the LCA was to evaluate the carbon footprint of 1 kg of >99% purity MG- Si for Si-PV. The stages within the system boundaries were quartz mining, silica sand extraction, and MG-Si production (Figure 5.1). The life cycle inventories for this study were taken from published papers (Table D2), the International Energy Agency (IEA) PVPS Task 12 [4], Ecoinvent 3.6 Database [256], and DATA SMART LCI (US-EI 2.2) [257]. Life cycle impact assessment was conducted using SimaPro [169]. ReCipe2016 method was used for global warming potential (GWP) analysis, and Cumulative Energy Demand V1.1 was used for the energy assessment. The IEA PVPS recommends GWP and CED metrics to evaluate the environmental impact of photovoltaics [4]. 5.2.2 2 Quartz mining, silica sand extraction, and MG-Si production The process for silica sand extraction from high-quality (>98%), industrial-grade (95%), and low-quality (65%) quartz deposits were based on existing industrial processes. The Unimin Co process was selected to model the silica sand extraction process since they are the leading global silica sand producer and transform quartz of various purity. We divided the silica sand extraction into two main steps (Figure 5.2) [207]. The first step (quartz mining) begins with removing the topsoil with loaders and bulldozers, excavating with excavators, transporting with lorry and conveyor belts, sieving, washing, and dewatering. This step is similar for high-quality, industrial-grade, and low-quality deposits. The second step (silica sand extraction) depends on quartz purity and includes beneficiation processes such as magnetic separation, flotation, and gravity separation to remove impurities physically or chemically. There is no need to use the 61 beneficiation process for mines with the highest purity except for drying silica sand before transportation to MG-Si facilities. In contrast, in mines with industrial-grade and lower quality, beneficiation processes are required to remove impurities. Input data for the LCA, including required electricity and fuel at each stage, was estimated based on existing similar processes in literature (Table D2). We used the USA average electricity mix for the USA because quartz deposits are distributed all around the country. The regional electricity was selected for China since potential quartz deposits are distributed in some regions, primarily east and southeast. A significant regional difference was reported for electricity in China [258]. China’s electricity was divided into six regions (Northwest, North, Northeast, East, South, and Center) based on energy resources for electricity in earlier research [258]. We selected a representative province for each region based on the highest demand for silica sand. For example, Xinjiang had the highest demand for silica sand and was chosen as a Northwest region representative. Beijing, Liaoning, Fujian, Yunnan, and Sichuan were selected for North, Northeast, East, South, and Center, respectively. 62 Figure 5.2. The process of quartz mining and silica sand extraction. The 1st step is similar for all types of quartz. The 2nd step depends on quartz purity. *Silica sand purity depends on initial quartz quality, mining process, and extraction chosen. Purification is required to increase the purity of silica sand to produce MG-Si with more than 99% purity. At this stage, purified silica sand is called MG-Si. Further purification is necessary to provide solar grade silicon with 99.99999% purity that can be used for Si PV module manufacturing. We modeled only MG-Si production for this study since the silica sand quality does not affect the rest of the manufacturing processes if the produced MG-Si reaches more than 99% purity. The MG-Si production process was modeled based on IEA PVPS Task 12 and Ecoinvent (Table D2). 5.3 Results and discussion 5.3.1 Location of legal and illegal quartz mines and MG-Si production 5.3.1 1 Legal mines The annual production of industrial-grade silica sand per continent was compiled from 1994 to 2019 (Figure 5.3). The global production in 2019 was 329 million metric tons, with 35% 63 of the production in North America, 46% in Europe, and 12% in the Asia-Pacific region (APAC). The USA was the main industrial-grade silica sand producer, and 73% of its annual production was used for hydraulic fracturing [199]. The Netherlands produced 37% and Spain 24% of the European production. In APAC, India (30%) and Malaysia (25%) were the leading producers. Figure 5.3. Annual industrial-grade silica sand production in the world from 1994 to 2019. Data was compiled from USGS annual reports [199]. Figure 5.4 shows the mine’s locations of the USA’s top ten high-quality and industrial- grade silica sand producers. Most US quartz mines are located in Illinois, Ohio, Michigan, Texas, and North Carolina Unimin Corporation is the largest producer with an annual capacity production of more than 41 million metric tons and 21 active quartz deposits in the USA [207,233,259]. US Silica Inc. is the second-largest silica sand producer in the USA, with an annual production of 19 million metric tons and 15 mines [199,260]. 64 Figure 5.4. The mines distribution of 10 major high-quality and industrial-grade silica sand producers in the USA High-quality silica sand is scarce, and only a few high-quality deposits, such as the Spruce Pine in North Carolina, USA, exist in the world [233,261]. The location and amount of global industrial-grade quartz deposits in 2019 were compiled (Figure 5.5A). Although typical sand is found everywhere, high-quality and industrial-grade silica sand resources are not equally distributed, and some countries suffer from a high-quality deposits’ shortage. In contrast, China has no domestic high-quality and industrial-grade deposits and is highly dependent on silica sand imports [234,250] (Figure 5.5B). The main silica sand suppliers for China MG-Si production have shifted from the USA and Vietnam until 2012 to Cambodia, Australia, Malaysia, and Pakistan. Other countries like India and the United Arab Emirates also do not have enough high-quality resources to meet their growing demand. As reported by the UN, a portion of this global demand was supplied via documented international trades between producers and consumers,. 65 Figure 5.5. (A) Global distribution of industrial-grade quartz (95% SiO2) mines in 2019. Data was collected from USGS annual reports [199]. (B) Annual industrial-grade silica sand export to China. Data was compiled from the annual UN Comtrade reports [197] 5.3.1 2 Illegal mines Quartz is the second most illegally traded product [262]. A review of news articles and NGO’s reports was conducted to identify illegal quartz mines worldwide. Countries such as China, India, and Singapore with high growth rates need a large amount of silica sand to expand their high-tech industries, such as Si PV [254]. This demand has increased silica sand prices, which makes illegal trades more attractive. For example, Singapore's silica sand price has increased from 3 to190 $/metric ton in the last ten years [216]. Illegal sand trades were also reported in various locations in India (Figure D13) [245–247,263–266]. In Kerala alone, in 2017, sand valued at 2.3 billion USD on the black market was removed from local rivers at a rate 40 times faster than nature can recover [226,251,265]. Mumbai city is another example of an area with high illegal trade. In 2016, more than 30,000 instances of illegal sand trading were reported. [226]. Illegal sand trades were also reported in other parts of the world, Tanzania, Kenya, Morocco, South Africa, and Sierra Leon in Africa and Colombia in South America [216,267]. Morocco supplied half of the silica sand demand (10 million metric tons per year) from illegal quartz mining [216]. 66 Many illegal mining activities were reported in countries close to China. To produce one kg of MG-Si (>99% purity), 2.7 kg of industrial-grade quartz (95% purity) is required [4]. In 2019, China needed to import 12.15 million metric tons of industrial-grade silica sand from foreign resources for MG-Si production. However, only 14% of this amount can be accounted through documented trades. Meanwhile, multiple reports from NGOs and news agencies have identified illegal mining in Cambodia and North Korea that supply silica sand to China. (Figure 5.6) [220,242,251]. Knowingly or not, some Chinese MG-Si producers might have used silica from those illegal mines. Figure 5.6. Illegal silica sand mines along the Mekong River in Cambodia and within Haeju Bay in North Korea. 5.3.1 3 MG-Si producers The location of MG-Si producers (Figure 5.7A) and the amount of annual MG-Si production (Figure 5.7B) were compiled. Since 2005, the global MG-Si production has increased by about 60%, and in 2019, China, Russia, Norway, and the USA were the main producers. China 67 produced 64% of the global MG-Si production in 2019. Yunnan, Xinjiang, Sichuan, Guizhou, Hunan, and Fujian province produced more than 85% of 2019 MG-Si production in China. Figure 5.7. (A) The global distribution of MG-Si production. (B) annual MG-Si production from 2005 to 2019 (adapted from [199]) 5.3.2 Environmental impact assessments 5.3.2 1 Modeling silica sand extraction Silica sand extraction from low-quality quartz needs extra beneficiation steps to remove impurities. The beneficiation processes were selected based on the types and locations of mines. For mines with access to natural water resources like rivers, the gravity method is appropriate since it requires large amount of water. However, residual sediments might affect the downstream river ecosystem. The high purity quartz resources in China are shown in Figure 5.8. They are mostly located in Chaina’s east, south, and center and have an average purity of 65% [234,235]. Figure 5.8 also shows available foreign silica sand resources for China. We modeled quartz mining and silica sand extraction process for various scenarios to evaluate the carbon footprint of supplying silica sand needed for MG-Si production in China now and in the future. Table 5.1 describes the scenarios considered. 68 Figure 5.8. (A) The regional production of MG-Si (colored provinces) (adopted from [199,268]), potential domestic quartz deposits (colored dots), and foreign available silica sand resources (colored hexagonal) for China. For foreign resources, legal and illegal trades are in green and red, respectively. (B) Regional electricity for China was compiled based on the previous study [269] Table 5.1. LCA scenarios for quantifying the carbon footprint of supplying silica sand for China LCA Scenario Description Baseline Extracting silica sand from High-quality deposits (>98% purity) Extracting silica sand low-quality quartz using the gravity separation technique from domestic low-quality quartz using the flotation technique sources… low-quality quartz using the magnetic separation technique Imported industrial- Australia (AU) grade silica sand from Cambodia (KH) … Malaysia (MY) Pakistan (PK) 51% AU, 45% KH, 1% MY, and 3% PK The GWP and CED of extracting silica sand from high-quality (>98% silica) and industrial- grade quartz (95% silica) were calculated for the USA (Figure 5.9A and D). The average GWP of extracting one metric ton of silica sand from high-quality deposits in the US was 22.7 kg of CO2eq, while it was almost double (47.9 kg of CO2eq) for industrial-grade deposits due to the beneficiation processes needed to remove impurities from industrial-grade quartz. A similar trend was observed for CED. The required energy for extracting silica sand from high-quality deposits was 339 MJ/metric ton of silica sand, and it increased to 1,010 MJ for industrial-grade quartz. 69 Figure 5.9B and C present the GWP of silica sand for MG-Si production in China. The baseline scenario for China was based on high-quality quartz. The GWP of imported industrial- grade silica sand from legal mines in Australia, Pakistan, Malaysia, and Cambodia was 117, 78.3, 69.3, and 58.3 kg CO2eq, respectively (Figure 5.9B). For illegal mines in North Korea, the GWP was 46.8 kg CO2eq. The difference by country is due to differences in electricity mix and fuel types necessary for mining operations and distance to China. When considering silica sand imports reported by the UN (51% Australia, 45%, Cambodia, 1% Malaysia, and 3% Pakistan), the GWP was 88.9 kg CO2eq. For domestic low-quality quartz, the GWP was 74.6 to 86.7 kg CO2eq per metric ton of silica sand, depending on the type of beneficiation processes (Figure 5.9C). The flotation technique had a higher GWP compared to other techniques since it requires more fuel for operations. The lowest CED was for importing industrial silica sand from illegal mines in North Korea, which was 875 MJ per metric ton of industrial silica sand (Figure 5.9E). The highest CED was 1,890 MJ for Australia. The CED of domestic low-quality quartz was 1,120 to 1,580 MJ (Figure 5.9F). This value was 16-40% lower than importing industrial silica sand from Australia but 28- 81% higher than importing silica sand from North Korea and 8-52% from Cambodia. CED is an indicator of product cost [270]. Therefore, importing industrial-grade silica sand from Cambodia and North Korea would be cheaper than using domestic low-quality resources or importing from Australia, Pakistan, and Malaysia. One way of reducing the use of illegal quartz for Si manufacturing would be through PV supply chain tracking and certification by a third party. The transition from fossil fuels to renewable energy resources such as PV has remarkable benefits to society and the environment. However, the possibility of unlawful activities such as using illegal quartz in the PV supply chain 70 may threaten the original benefits of using PV. The PV consumers desire to purchase ethical PV products, while recent news agencies have reported some illegal activities in PV supply chains. Therefore, it is essential to monitor supplying quartz for the Si PV industry to ensure illegal mining is not used for PV production. In this regard, the Green Electronics Council (GEC) has recently started working with NSF International, the public health and safety organization, to develop standards for sustainable PV modules [271]. GEC has also introduced ecolabel as an easy way that manufacturers can differentiate their products from others and ensure consumers that the products meet sustainability criteria. In another effort in February 2021, the Solar Energy Industries Association (SEIA) and about 175 solar companies have signed an agreement opposing using forced workers in PV supply chains [272]. Similar types of criteria and agreements are vital to mitigate the destructive footprints of using illegal quartz in the PV supply chain. We identified social concerns due to illegal mining based on a literature survey of news articles from the last five years. Quartz mining is becoming a global socio-environmental challenge. The mining industry may positively impact local communities' development by creating direct and indirect jobs. However, the possibility of excessive excavation in illegal mining may have adverse effects in the long term, even though there might be short-term financial benefits for local communities. Mining silica sand from oceans and riverbeds is more interesting since they are naturally crushed and ready to use. Riverbank sand is cheaper since there is no need to remove the salt [217]. But illegal and excessive mining in rivers and oceans could create severe social and environmental problems. In Indonesia, 24 small islands and their ecosystems disappeared between 2005 and 2010 due to excessive quartz mining and silica sand extraction to export to Singapore [240]. The Mekong River is another example that silica sand mining threatens the ecosystem. The Mekong 71 River is the 10th longest river globally and starts from China and passes through Laos, Thailand, Vietnam, and Cambodia. The largest extractors are located in Cambodia, where they extract 33 million metric tons of silica sand per year [220]. Excessive silica sand mining may change the river morphology and erosion pattern, affecting fisheries and, consequently, threatens the main food source of 60 million people living in that region, as is stated in the literature [226,228]. Farmers’ incomes can also be affected due to excessive mining activities. A 20% reduction in Cambodia was reported due to a lack of agricultural lands [231]. Quartz mining may also affect communities’ livelihood. In a survey study in a village in Tangail District in Bangladesh, the local community was concern about potential disasters and the black market due to quartz mining activities near their livelihoods [273]. Depletion of groundwater, reduction of soil ability to provide nutrients, increase the pH water and turbidity of the river, destruction of infrastructures, riverbank collapse, and social collapse are only some examples of mining silica sand in rivers that are also reported in the literature for mining activities [216,222,226,274–276]. Those socio-environmental impacts can be more severe for illegal mines due to a lack of legislation and public awareness. Recent changes in legislation in Cambodia have limited silica sand extraction in the Mekong River and may affect export to China, which could affect Si PV manufacturing. However, it is reported that China has already started developing new low-cost purification methods [234,235,250] and will likely start using domestic low-quality quartz resources. 72 Figure 5.9. The GWP (A, B, C) and CED (D, E, F) of silica sand production for MG-Si production in the US and China 5.3.2 2 Producing MG-Si from silica sand in the USA and China The quartz purity affects the carbon footprint of producing MG-Si in China. The GWP to produce one kg of MG-Si from imported industrial-grade silica sand was 12.1 kg CO2eq (Figure 5.10A). In comparison, using domestic low-quality quartz produced 16.5 kg CO2eq, which is a 36% increase compared to imported industrial-grade silica sand. The GWP of producing one kg of MG- Si in the USA was 12.0 kg CO2eq. Quartz purity also influenced the amount of required energy for producing MG-Si. The CED of producing one kg of MG-Si from current resources (imported industrial-grade silica sand from 51% AU, 45% KH, 1% MY, and 3% PK) was 188 MJ and increased by 53% (286 MJ) if domestic low-quality quartz resources were used instead (Figure 5.10B). Silica sand purification requires a large amount of energy due to the required high temperature (about 1,800 oC) to remove impurities from silica sand. Low-quality quartz contains more impurities and consequently requires more energy. The CED was 186 MJ for producing one kg of MG-Si in the USA. 73 Figure 5.10. (A) The GWP and (B) CED of producing MG-Si from high- and low-quality silica sand. The carbon footprint of MG-Si depends on the quality of quartz. Previous LCA Si PV studies did not evaluate the carbon footprint of MG-Si production based on the location due to a lack of information for silica sand production [277]. They reported MG-Si carbon footprint based on using high-quality quartz. However, as this study showed, high-quality quartz deposits are rare, and probably most of the MG-Si companies use lower quality quartz which can increase the associated carbon footprint. 5.4 Conclusion Manufacturing Si PV requires large amounts of raw materials, including quartz. Quartz purity affects the carbon footprint of producing MG-Si used in Si PV. Previous studies focused only on scarce material and ignored quartz, probably since quartz deposits are perceived as abundant and available everywhere. Some countries, such as China, the global leading Si PV producer, do not have high-quality (>98% purity) and industrial-grade (95% purity) quartz deposits and are highly dependent on foreign resources. Regulations restricting silica sand export to China 74 may force Chinese MG-Si producers to use domestic low-quality (65% purity) quartz. China produced about 4.5 million metric tons MG-Si (64% of global production) in 2019 (Figure 5.7). The transition from importing high-quality or industrial-grade silica sand to mine low-quality domestic quartz can increase MG-Si production's annual carbon footprint by at least 23% (Figure 5.10). This corresponds to the annual greenhouse gas emissions from about nine million passenger cars. Si PV companies should improve the efficiency of existing MG-Si production before using low-quality quartz resources and increasing the carbon footprint of PV. Upgraded MG-Si is a new technology that can be a replacement for existing MG-Si production processes in terms of cost and carbon footprint [277]. Other sources of silica should be considered for MG-Si production, including secondary resources such as rice husk ash [278] and gold mining tails to reduce the pressure on natural resources [279]. In addition to the environmental impact of MG-Si production, the social impacts associated with quartz mining are not often discussed in the literature. Global silica sand production has tripled in the last decade. Even though the implication of quartz mining projects can have short positive effects such as creating temporary jobs for locals, it may have severe long impacts on local livelihood, particularly in regions with limited regulations. The New York Times has recently reported that some countries have regulations that may let MG-Si producers hire forced workers from ethnic minorities in an unacceptable condition [212]. Apart from inefficient rules for quartz mining, illegal mines can have more severe impacts on the environment and local communities, which are also ignored in the literature. Solar PV is a greener alternative to current electricity production, but its image could be tarnished by an increasing manufacturing carbon footprint, or even worst, using quartz from illegal mines and forced labor. To ensure PV remains a sustainable 75 energy option, we must ensure that PV supply chains are free from unethical activities such as using illegal mining and forced labor. Acknowledgment This work was supported by the National Science Foundation grant number NSF-1801785. 76 Chapter 6 Conclusions and major contributions PV is critical for electricity generation in the US and globally. It is expected that the transition from fossil fuels to sustainable clean energy such as PV will decarbonize the energy system and mitigate the carbon footprint of fossil fuels. However, this transition from conventional energy to PV faces severe challenges, such as the carbon footprints of manufacturing solar technologies and possible material shortages for PV manufacturing. This dissertation focused on the leading commercialized PV technology (Si PV) and an emerging PV technology (OPV). We investigated the materials and resource availability necessary for making Si PV in the next ten years. We also quantified the carbon footprint and the cumulative energy demand (CED) of providing the primary raw materials needed for Si PV manufacturing. Besides Si PV, we used LCA, green chemistry, toxicity assessment, analytical chemistry, and material flow analysis to determine the environmental hotspots of the fullerene (C60) production process required for OPV manufacturing. We evaluated the environmental, cost, and chemical hazards of existing C60 production processes and revised the best available large-scale production process for lower environmental, cost, and chemical hazard impacts. 6.1 Sustainable design for OPV manufacturing C60 is often used as an electron acceptor in OPV manufacturing, even though it significantly contributes to OPV’s carbon footprint. Existing C60 production processes are energy-intensive and require a large amount of hazardous materials. We evaluated the environmental, cost, and chemical hazards of existing C60 purification methods (e.g., chromatography, crystallization, & complexation) and determined the complexation process as the baseline with the lowest environmental, cost, and chemical hazard impacts. 77 We used LCA and identified that more than 95% of the cumulative energy demand of the purification process comes from the solvent (1,2,4-trimethylbenzene - TMB) used in the baseline method. Therefore, green chemistry principles were employed to determine potential greener replacements for highly toxic TMB. Potential alternative solvents were identified based on hazardous scores and C60 solubility scores. Plant-based solvents, such as olive oil, linseed oil, and sunflower oil as well as petroleum-based solvents, such as toluene and xylene, were determined as the potential replacements due to their high C60 solubility and low toxicity scores compared to TMB. The toxicity score of toluene and xylene was estimated based on the EPA software, TEST. However, TEST is not useful for evaluating the toxicity of plant-based solvents and nanoparticles such as C60. So, a series of singular and combined toxicity tests were conducted based on EPA protocols to quantify the individual toxicity of C60, TMB, linseed oil, olive oil, and sunflower oil and the combined toxicity of dissolved C60 in each solvent. In the second Chapter, we showed that linseed oil, olive oil, sunflower oil had much lower toxicity effects than TMB on tested species (Daphnia Magna). The C60 did not show toxicity effects. However, dissolved C60 was toxic. The highest toxicity was for dissolved C60 in TMB, while the minimum toxicity was for C60 in olive oil. The maximum toxicity among solvents was for TMB, and the minimum was for sunflower oil. Olive oil was identified as the potential alternative replacement for TMB based on the singular and combined toxicity assessment. The third Chapter evaluated the environmental, cost, and chemical hazards of using potential alternative solvents for C60 synthesis. We used an iterative approach to conduct experiments to compare the C60 production rate and the C60 purity of methods using potential replacements for TMB. With the iterative approach, we continuously monitored C60 production rates and C60 purity in each experiment and revised the process using LCA and green chemistry 78 principles. Even though olive oil had the lowest toxicity score among the potential alternative solvents, the environmental, cost, and chemical hazards of using olive oil were higher than purification processes using other potential alternative solvents. Xylene was identified as the greener alternative solvent for TMB since the purification process had the lowest environmental, cost, and chemical hazard impacts. 6.2 Material requirement for Si PV manufacturing and associated environmental impacts Supplying required materials for Si PV manufacturing and associated socio-environmental impacts can be a big challenge for energy system decarbonization. Chapter 4 quantified the amount of materials (e.g., glass, metallurgical-grade silicon (MG-Si), aluminum, silver, copper, lead, and tin) needed for Si PV laminate manufacturing. We used a bottom-up approach to evaluate the material necessary to manufacture each sub-Si PV technology (e.g., Al-BSF, mono facial and bifacial PERC, mono facial and bifacial HJT, and perovskite/Si tandem). The highest material demand was for glass and MG-Si. More than 74 billion metric tons of solar glass and three billion metric tons of MG-Si will be necessary for the next ten years to manufacture enough Si PV to meet the global expected electricity generation from PV. The highest raw material demand was determined for quartz. Chapter five evaluated the resource availability and the purity of quartz used for MG-Si and solar glass production. High- quality quartz (>98% silica) is required for MG-Si production, and industrial-grade quartz (95% silica) is needed for solar glass manufacturing. We determined the location and the purity of available quartz resources globally. We also employed LCA to model silica sand production and quantify the carbon footprint and the cumulative energy demand (CED) of quartz mining, silica sand extraction, and MG-Si production. The purity had a significant effect on the carbon footprint 79 and CED of MG-Si. This is due to the excessive energy necessary to remove impurities from quartz deposits with lower quality. Providing silica sand for MG-Si production had about 36% more carbon footprint than supplying silica sand from high-quality resources. We also determined that some global leading Si PV producers do not have access to domestic high-quality quartz resources and are highly dependent on high-quality foreign resources. Meanwhile, this Chapter highlighted the possibility of supplying silica sand from illegal mines for Si PV manufacturing. This can increase Si PV impacts on the local environment and communities. We proposed using a kind of third-party certifications to monitor the Si PV supply chain to ensure that illegal quartz is not used for Si PV manufacturing and consumers buy ethical PV products. 6.3 Conclusion This dissertation focused on the environmental evaluation of OPV and Si PV manufacturing. The manufacturing process of OPV, the emerging solar technology, requires sustainable modifications before large-scale production. C60 is needed for OPV manufacturing and has a significant impact on OPV carbon footprint. We used LCA, green chemistry, toxicity assessment, analytical chemistry, and material flow analysis to identify the environmental hotspots of the C60 production process and determine alternative greener production processes. The final replacements were evaluated based on environmental, cost, and chemical hazard impacts. We also considered Si PV, the most commercialized solar technology, and quantified materials necessary for Si PV manufacturing in the next ten years. The resource availability, quality of raw material deposits, and the possibility of supplying raw materials from illegal mines were assessed. Results from this dissertation can be used to mitigate the environmental, cost, and chemical hazards of OPV manufacturing before large-scale production. Results also highlighted the potential material shortage for Si PV manufacturing in the next couple of years as well as the importance of using 80 third-party certification to monitor the Si PV supply chain besides improving the MG-Si production process to use low-quality quartz without increasing Si PV carbon footprint. Acknowledgments This research was supported by the National Science Foundation under Grants NSF-1511098 and NSF-1801785 and the Michigan State University Discretionary Fund Initiative. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. 81 APPENDICES 82 APPENDIX A: Supplementary Information for Chapter 2 83 A1. Solvent properties Table A1. Physical properties and market price for conventional solvents and essential oils. All solubility and viscosity were reported at 298 K except when mentioned in parentheses. Data was collected from reference [1] except those were mentioned in the bracket. Water C60 solubility C70 solubility Market price Substrate Viscosity (mPA s) solubility (g/L) [179] (g/L) [179] (USD/L) (g/L) Brassica oil 0.19 1.03 166 [4] N.A. N.A. Castor oil 0.39 0.64 78 [4] 1000 [280] N.A. Cedar oil 51.8 (343) N.A. 173 [4] 5000 [180] N.A. Corn oil 0.61 1.96 86.4 [4] 35.4 [180] N.A. Grapeseed oil 0.37 (293) N.A. 200 [8] 46.6 [281] N.A. Linseed oil 53.1 (328) 1.2 94 [281] 48 [280] N.A. O-xylene 9.3 [282] 15.6 [5] 110.5 [281] 0.6 [6] 0.02 [283] Olive oil 23.6 (323) 1.04 119 [281] 58.2 [4] N.A. Peanut oil 0.75 0.85 189 [281] 73 [280] N.A. Soybean oil 0.86 0.75 88.9 [281] 60 [280] N.A. Sunflower oil 6.91 1.83 19.6 [281] 48.8 (299) [280] N.A. Tetralin 15.7 12.3 119 [281] 2 [284] Very poor [285] Toluene 8.7 1.4 63 [281] 0.6 [284] 0.53 [286] TMB 17.9 N.A. 64.2 [281] 2[10] 0.057 [287] Walnut oil 0.48 (293) 1.39 100 [285] 42.9 (299) [281] N.A. 84 A2. Toxicity studies for fullerenes Table A2. Summary of previous acute tests for fullerenes Fullerene preparation Ecotoxicity test condition Findings Ref Method Dissolve fullerenes in - EPA protocol - LC50 of fullerenes in THF method: [76] THF following by - Acute test for 48 h 0.46 ppm filtration and evaporation - Daphnia magna - LC50 of fullerenes in sonication method: 7.9 ppm 30-min sonication of fullerenes in DI water Stir fullerene in water for - D. magna - No LC50 for acute tests 2 months under the - Acute test for 48 h - Reduced offspring production in sunlight - Chronic test for 21 d chronic tests - Sonication of fullerene in water increased the toxicity of fullerene - For the environmental purpose, sonication and adding THF are not acceptable. Dissolve fullerene in THF - Juvenile Largemouth - Significant lipid peroxidation after 48 [288] Bass h exposure to 0.5 ppm nC60 Dissolve fullerene in THF - E. coli and B. Subtilis - Antimicrobial properties against both E [134] coli and B Subtilis - Minimum inhibitory concentration for E. coli: 0.5-1 mg/L of C60 - Minimum inhibitory concentration for B. Subtilis: 1.5-3 mg/L of C60 - Dissolved fullerenes - US EPA - LC50 of THF-fullerenes: 0.8 ppm [75] in THF - 48-h acute tests on D. - LC50 of stirred C60: more than 35 ppm - Suspended magna fullerenes in water (stirring for weeks) - Dissolved fullerenes - Micronucleus tests - Auq/nC60 showed higher genotoxicity [289] in Ethanol than EthOH/nC60 (EthOH/nC60) - Suspended fullerenes in water by mixing for 2 weeks (auq/nC60) - Suspended - 24 and 48 h acute tests - Less than 20% death in daphnids for [74] fullerenes in natural on D. magna concentrations lower than 10 mg/L and artificial - Highly variable death rates in daphnids freshwater by for concentrations higher than 10 mg/L mixing for 4 weeks at 20 oC 85 A3. Experimental setup for toxicity tests Figure A1. Experimental setup for culturing daphnids and acute tests. Fluorescent light with the intensity of 70±5 foot candles and a 16 h photoperiod was used. Daphnids grew up in reconstituted water with mentioned pH, alkalinity, and hardness at room temperature. Twenty offspring daphnids ( < 24 h old) were used for each concentration (five daphnids in each beaker). A4. Hazardous scores Table A3. Chemical hazards (NFPA) and calculated hazardous score using the method from Lee et al., 2018 [20] Substrate Health Flammability Reactivity Special Hazardous NFPA score source Ref Brassica oil 1 1 0 0 0.67 [290] Castor oil 0 0 0 0 0 [291] Cedar oil 2 2 0 0 1.33 [291] Corn oil 0 1 0 0 0.33 [291] Grapeseed oil 1 1 0 0 0.67 [292] Linseed oil 0 1 0 0 0.33 [291] O-xylene 3 3 0 0 2 [293] Olive oil 1 1 0 0 0.67 [293] Peanut oil 0 1 0 0 0.33 [291] Soybean oil 0 1 0 0 0.33 [291] Sunflower oil 0 1 0 0 0.33 [291] Tetralin 2 2 1 0 1.5 [293] TMB 2 2 0 0 1.33 [293] Toluene 3 3 0 0 2 [293] Walnut oil 1 1 0 0 0.67 [290] 86 A5. Pre-screening results Table A4. Mortality of daphnids for prescreening tests after 48 h exposure to toxicants Toxicant % of toxicants in % of dead 100-mL beakers daphnids Linseed oil 30 100 3 0 0.3 0 Olive oil 30 100 3 40 0.3 0 Sunflower oil 30 100 3 20 0.3 0 TMB 10 100 1 100 0.1 20 A6. Physiochemical properties of fullerenes Table A5. Physiochemical properties of suspended fullerenes (maximum and minimum concentrations) in the test medium. Media C60 Average size Zeta potential concentration (nm)* (x, mV) (ppm) Reconstituted water (auq/nC60) 176 584 38.6 11 132 -38.6 C1 (TMB + auq/nC60) 176 731 -38.8 11 395 -37.9 C2 (Linseed oil + auq/nC60) 176 454 -38.8 11 215 -38.3 C3 (Olive oil + auq/nC60) 176 715 -38.0 11 452 -38.4 C4 (Sunflower oil + auq/nC60) 176 755 -37.9 11 81 -38.6 87 A7. C60 concentration measurments The concentration of suspended C60 in stock and dilutions was calculated by measuring the absorption at 335 nm in UV-VIS spectroscopy (Figure A2). First, absorption of standard solutions was done to extrapolate fitting equation (Eq. 1) from fitting curve (Figure A2.a). Then, the concentrations of C60 in stock and dilutions (x in Eq. A1) were calculated by measuring the absorptions of extracted C60 at 335 nm (y in Eq. A1) (Figure A2-b). Eq. (A1) Where y is the absorbance at 335 nm and x is the concentration of C60 in toluene. Figure A2. Absorption measurements at 335 nm in UV-VIS spectroscopy of (a) standard solutions (C60 in Toluene) and the fitting curve, (b) extracted C60 as explained in sec 2.3 88 A8. Singular toxicity Figure A3. Light-microscopic images of daphnids after 48 h exposition to auq/nC60 at different concentrations. The dead D. magna is shown to provide a comparison between live and dead daphnids. 89 Figure A4. pH measurements and dose-response curves after 48-h acute tests for (a) 1,2,4-TMB, (b) linseed oil, (c) olive oil, (d) sunflower oil. Figure A5. Essential oil structures for (a) linseed oil, (b) olive oil, and (c) sunflower oil 90 A9. Gas Chromatography – Mass Spectroscopy (GC-MS) analysis The GC-MS analysis was performed to quantify the composition of fatty acids in water- soluble fraction (WSF) of linseed oil, olive oil, and sunflower oil using an Agilent 7890B GC/ triple quadrupole mass spectrometer 7010B (Agilent, Santa Clara, CA). One µL of the derivatized sample was injected in a split mode (1:20) to quantify total oil in WSF. All injections were performed with an injector temperature of 250 °C and a flow rate of 1.2 mL/min helium. Separation was achieved on an Agilent J&W VF5ms column (30 m x 0.25 mm x 0.25 µm) (Agilent, Santa Clara, CA) using the following temperature profile: 40°C for 2 min; 25°C min-1 to 200°C; 4°C min-1 to 230°C; 40°C min-1 to 320°C; 320°C for 5 min. Ionization employed 70 eV electron ionization, and the mass spectrometer was operated in scanning mode with the first quadrupole with a scan range of m/z 45 to 450. FAMEs were identified by comparing their mass spectra and retention time to the Supelco 37 component FAME mix (Sigma-Aldrich, St. Louis, MO). TargetLynx (Waters Corporation, Milford, MA) was used for data analysis after exporting the original MassHunter data file into NetCDF format, then converting it into water. Quantification was based on the peak response of the total ion chromatogram (TIC) for the respective FAMEs normalized to the internal standard (glyceryl triheptadecanoate) (C17:0 methyl ester at 12.8 min). Solvent extraction was used to prepare samples for GC-MS analysis. A mixture of chloroform/methanol (2:1) with 90 mg/L butylated hydroxytoluene (BHT) (0.01% (w/v)) and 1 mL formic acid (1% (v/v)). 91 Figure A6. The chromatogram of sunflower oil – GC-MS Figure A7. The chromatogram of the WSF of sunflower oil – GC-MS A10. Combined toxicity Table A6. C60 impacts on the toxicity of Baseline (TMB) and alternative solvents (essential oils) Solvent C60 concentration (mg/L) Antagonism effect Additive effect Synergism effect TMB … £ 11 22 £…£ 88 176 £ … Linseed oil ¾ ¾ 11 £ … Olive oil … £ 44 88 176 £ … Sunflower oil … £ 11 22 44 £ … 92 A11. Dilution factors Dilution factors could have a role in the toxic consequence effects of fullerenes. Contaminated fullerenes could be released into the environment in many ways at different concentrations. For example, the concentration of fullerenes in effluent water from treatment plants was estimated from few ng/L to 67 µg/L [294,295], while this concentration could be higher if fullerenes are used in “down-the-drain” consumer products (e.g., C60 in essential oils as a human supplement or in pharmaceutical products) or accidental spills happen during the manufacturing process (e.g., fullerene purification steps). Dilution factors in the environment significantly vary among countries from five (Morocco) to 33,500 (Canada) [296]. For instance, dilution factors were estimated between 10 to 40 for India, Iran, Iraq, Italy, and Belgium, 100 to 500 for Australia, China, and the USA, and higher than 500 for Canada, Russia, and Brazil [296]. In this study, these estimated dilution factors (five to 33,500) were also considered to design the combined toxicity. The concentration of C60 in TMB and essentials oils before releasing to the environment was presented in Tables A7 and A8. Table A7. C60 concentrations in essential oils (human supplements) and TMB (in fullerene purification step) C60 concentration Reference in solvents (mg/L) Human supplement (C60 in olive oil) 800 [123] Fullerene purification (TMB) 11,400 [160] Fullerene purification (linseed oil) 33,200 Calculation based on [160]and Fullerene purification (olive oil) 14,800 C60 solubility in the essential Fullerene purification (sunflower oil) 4,300 oils. 93 Table A8. Dilution factors for mixtures (the “fixed-ratio-design” approach was used to conduct combined toxicity tests) This study literature Ref C60 in TMB 41 to 656 5 to 33,500 [296] C60 in linseed oil 34 to 544 C60 in olive oil Under the detection limit C60 in sunflower oil 15 to 240 94 APPENDIX B: Supplementary Information for Chapter 3 95 B1. Detailed methodology Figure B1. The material flow of complexation method (Nagata et al. [58]) - line shows the system boundary that was considered for LCA Steam (90°C), Solvent O-xylene Activated carbon 10 hr regeneration Liquid Fullerene mix Electricity C60 Crystal (Y= 67%) Electricity(Agitation) C60 Crystallization Electricity (Dry) Chromatography solution (Filtration) 99% purity Fullerene mix (FM) Electricity C70 Crystal (Y=34%) C70 Crystallization Electricity (Dry) (Filtration) 99% purity Electricity (-16°C), Solvent 24 hr regeneration Figure B2. The material flow of crystallization method (Kwok et al. [297]) - line shows the system boundary that was considered for LCA 96 Grushko et al., O-xylene Electricity O-xylene Electricity(Agitation) Fullerene mix Enriched C60 Filtration Activated carbon solution Steam (85°C) C60 Crystal Liquid Electricity (Dry) C60 Crystal (Y= 43.8%) (96.7 % purity) Chromatography at 203°C 99% purity Fullerene mix (FM) Figure B3. The material flow of crystallization method (Grushko et al. [298]) - line shows the system boundary that was considered for LCA Figure B4. Material flow for modified baseline 97 Figure B5. Material flow for P3 and P4 (alternative purification methods using olive oil) Figure B6. Material flow for P5 and P6 (alternative purification methods using linseed oil) 98 B2. Data source for life-cycle assessment Table B1. Data sources for material and energy used in inventory analysis Description Ref. 1 2 4-trimethylbenzene Co-production of mesitylene, xylenes, and benzene [299] from naphtha cracking 1,8-Diazabicyclo Synthesis of bicyclic amidines from caprolactam by [104] (5.4.0)undec-7-ene using acrylonitrile Linseed oil Process modification from Refined sunflower oil [300] (pressing), at processing, Agri-footprint 5.0; Linseed seed, Ecoinvent 3 Olive oil Process modification from olive production, [300][301] Ecoinvent 3 [300], based on information on olive oil production[301] Centrifuge Required energy was estimated based on using [302] Mixer industrial-scale equipment. [303] Pump [304] Ultrasonic bath [305] Ultrasonic probe [306] B3. Green chemistry principles Table B2. Green chemistry principles [24] Principle # Description How we used 1 Waste prevention instead of cleaning waste N/A after it is formed 2 Maximize the incorporation of all used N/A materials into the final products 3 Using substances that have little or no We categorized potential solvents based on toxicity for human health and the toxicity scores and identified alternative environment solvents with lower toxicity. 4 Design safer chemicals with lower We identified non-hazardous alternative environmental toxicity solvents to avoid toxifying the final products. 5 Use less solvents We categorized potential solvents based on their fullerene solubility to reduce the amount of solvents necessary for fullerene purification. 6 Minimizing energy N/A 7 Using renewable feedstock We identified plant-based alternative solvents besides petroleum-based solvents. 8 Avoiding unnecessary derivatization N/A 9 Catalytic reagents are better than N/A stoichiometric reagents 10 Avoid using undegradable chemicals N/A 11 In-process monitoring to avoid generating We used analytical chemistry (HPLC-UV) to hazardous materials monitor the impurity. 12 Safer substrates for accident prevention We determined alternative non-hazardous solvents to reduce accidental release risk. 99 B4. Prescreening solvents to identify replacements for TMB Toxicity scores were calculated using TEST software for all solvents except plant-based solvents. Table B3. Solvent toxicity scores and C60 solubility. # in Solvent CAS # Toxici Normalized C60 Normalize Ref. Figure ty toxicity solubili d C60 3. 5. (LC50) score ty (g/L) solubility – score mg/L 1,2,4-trimetyilbenzene 95-63-6 8.64 0.238 17.9 0.337 [56] Solvents in Olive oil 8001-25-0 54,500 3.78 × 10-5 23.6 0.444 [56,168] the purple Linseed oil 8001-26-1 17,400 1.19× 10-4 53.1 1 [56,168] area Xylene 95-47-6 12.3 0.167 9.3 0.18 [282] Toluene 108-88-3 17.3 0.119 8.7 0.16 [56] 1 Sunflower oil 8001-21-6 160,00 1.29 × 10-5 6.91 0.130 [56,168] 0 2 1,1,2,2- 79-34-5 95.9 0.021 5.3 0.10 tetrachloroethane 3 2-methylthiophene 554-14-3 30.4 0.068 6.8 0.13 4 Benzene 71-43-2 23.6 0.087 2.9 0.06 5 Cyclohexene (C6H10) 110-83-8 23.4 0.088 1.21 0.023 6 quinoline (C9H7N) 91-22-5 15.9 0.129 7.2 0.14 7 1,4-dimethylbenzene 106-42-3 12.3 0.167 5.9 0.11 8 chlorobenzene 108-90-7 11.4 0.181 5.7 0.11 9 1,3-dimethylbenzene 108-38-3 12.3 0.167 2 0.03 10 1-bromotetradecane 112-29-8 9.01 0.229 6.2 0.12 (CH3(CH2)13Br) 11 1 ,Bdimethylbenzene 620-14-4 7.29 0.283 7.3 0.14 12 Tetralin 119-64-2 7.40 0.278 15.7 0.296 [56] 13 1,2,3,5- 95-93-2 5.97 0.345 20.8 0.392 tetrarnethylbenzen 14 1,2,3,5- 527-53-7 5.97 0.345 18 0.34 tetramethylbenzene 15 1,2-dimethylbenzene 615-60-1 5.46 0.377 6 0.1 16 tribromomethane 75-25-2 4.74 0.435 5.64 0.106 17 1,2-dichlorobenzene 95-50-1 4.98 0.414 17 0.32 18 1-methylnaphthalene 90-12-0 4.79 0.430 33 0.62 19 1chloronaphthalene 90-13-1 2.96 0.696 52 0.98 (C10H7Cl) 20 dimethyl naphthalenes 571-58-4 3.26 0.632 36 0.68 21 1,2,3-tribromopropane 96-11-7 3.34 0.617 8.31 0.156 (C3H5Br3) 22 1,5,9- 706-31-0 2.32 0.888 7.4 0.14 cyclododecatriene 23 1,5,9- 4904-61-4 2.30 0.896 7 0.1 cyclododecatriene [56] (C12H18) 24 1,2,4-trichlorobenzene 120-82-1 2.06 1.00 13 0.24 (C6H3Cl3) 100 B5. Chemical hazards analysis The chemical hazard score of materials was calculated based on NFPA 704 standard scores (Table B4) and Eq. B1 [104]. The chemical hazard score of purification processes was quantified using data from Table B4 and Eq. B2 (Table B5). The amount of materials was compiled from experiments. The chemical hazard score of purification processes was normalized on the basis of the baseline score (Table B5). Where Hm is the hazardous score of materials used in each purification process. Health, Flammability, Reactivity, and Special are NFPA standard scores (Table B4). Hp is the hazardous score of the purification process (Table B5). n is the number of materials, and M is the amount of materials used in each purification process. 101 Table B4. Solvent chemical hazard scores based on NFPA. Health Flammability Reactivity Special Hazardous Ref. Solvents (Blue) (Red) (Yellow) (white) score 1,2,4-trymethilbenzene (TMB) 2 2 0 0 1.33 [293] Acetic acid (AcOH) 3 2 0 0 1.67 [293] DBU (1,8- Diazabicyclo[5.4.0]undec-7- 3 1 0 0 1.33 [307] ene) Heptane 1 3 0 0 1.33 [293] IPA 2 3 0 0 1.67 [293] Linseed oil 0 1 0 0 0.33 [308] Olive oil 1 1 0 0 0.67 [308] Toluene 3 3 0 0 2.00 [293] Xylene 3 3 0 0 2.00 [293] Table B5. Chemical hazard scores for fullerene purification methods Hazard score for Normalized Method all used materials scores Baseline 522.9 100 Modified baseline 298.8 57 P3 (alternative using olive oil) 978.8 187 P4 (alternative using olive oil) 785.6 150 P5 (alternative using linseed oil) 388.7 74 P6 (alternative using linseed oil) 175.3 34 P7 (alternative using xylene) 80.76 15 P8 (alternative using toluene) 70.28 13 B6. Life cycle cost analysis Cost analysis was conducted based on the C60 production rate and life cycle cost of C60 production [104]. The production rate was calculated using recorded data from experiments and then were normalized on the basis of the baseline (Table B6). The life cycle cost was conducted using life cycle inventory. We picked out energy materials and then calculated the cost based on the amount of energy materials and price for each process (Table B7 and B8). Cost metric values were calculated using equal weighting for normalized production rates and life cycle cost scores (Table B9). 102 Table B6. Production rate Process Time Rate Normalized (hr) (kg/hr) Production Rate Baseline 5.12 0.195 1 Modified baseline 5.27 0.190 0.972 P3 (alternative using olive oil) 5.45 0.183 0.939 P4 (alternative using olive oil) 5.25 0.190 0.975 P5 (alternative using linseed oil) 5.45 0.183 0.939 P6 (alternative using linseed oil) 5.25 0.190 0.975 P7 (alternative using xylene) 5.27 0.190 0.972 P8 (alternative using toluene) 5.27 0.190 0.972 Table B7. The price of energy materials in life cycle inventory Material Unit US$/unit Year Ref. Crude Oil kg $0.49 2021 [309] Bituminous Coal kg $0.06 2019 [310] Lignite (brown coal) kg $0.02 2019 [310] Anthracite (hard coal) kg $0.11 2019 [310] Natural Gas m3 $0.12 2020 [311] Table B8. life cycle cost analysis Process Normalized score Baseline 1.00 Modified baseline 0.583 P3 (alternative using olive oil) 2.30 P4 (alternative using olive oil) 2.22 P5 (alternative using linseed oil) 1.00 P6 (alternative using linseed oil) 0.408 P7 (alternative using xylene) 0.252 P8 (alternative using toluene) 0.197 Table B9. Cost metric values of processes Process Cost Metric Value Baseline 1 Modified baseline 0.778 P3 (alternative using olive oil) 1.62 P4 (alternative using olive oil) 1.60 P5 (alternative using linseed oil) 0.970 P6 (alternative using linseed oil) 0.691 P7 (alternative using xylene) 0.612 P8 (alternative using toluene) 0.584 103 B7. Environmental impact assessment LCA was conducted in SimaPro 9.1.0.7 [169] for the environmental evaluation of purification processes. The environmental analysis was based on equal weighting of various environmental impacts, including CED, GWP, WD, and E-Factor, for the baseline and potential alternative methods. We used the TRACI 2.1 V1.05 method for Global warming potential (GWP) analysis, Cumulative Energy Demand V1.11 method for energy assessment, (Water Scarcity) V1.00 method for water demand (WD) analysis [170], and E-Factor calculation [171]. Table B10. LCA analysis details GWP Normali Normaliz CED Normaliz (kg Normalized WD Normalized E- Methods zed E- ed LCA (MJ) ed CED CO2 GWP (m3) WD Factor Factor score eq) Baseline 8,950 100 308 100 9.0 100 381 100 100 Modified 5,110 57 176 57 5.0 57 217 57 57 baseline 30,10 P3 337 1,790 581 94 1061 1090 287 567 0 28,30 P4 316 1,710 554 88 1003 816 215 522 0 15,40 P5 172 823 267 33 370 641 168 244 0 P6 6,770 76 348 113 13 147 325 85 105 P7 3,180 36 195 63 9.0 102 46.5 12 53 P8 2,510 28 151 49 7.0 78 39.9 10 41 104 B8. Sustainability assessment Sustainability evaluation was based on the average of chemical hazard scores (Table B5), cost scores (Table B9), and environmental impact scores (Table B10). Table B11. environmental, cost, and chemical hazard scores of the baseline and replacements Environmental impacts score Chemical Cost Methods (GWP, CED, WDI, E-Factor) hazard score score Baseline 100 100 100 Modified baseline 57 57 78 P8 (alternative using toluene) 41 13 58 P7 (alternative using xylene) 53 15 61 P5 (alternative using linseed oil) 244 74 97 P6 (alternative using linseed oil) 105 34 69 P3 (alternative using olive oil) 567 187 162 P4 (alternative using olive oil) 522 150 160 105 APPENDIX C: Supplementary Information for Chapter 4 106 C1. PV installation growth in the next decade Figure C1. Annual PV installation from 2021 to 2031 (Adapted from 2020 DNV-GL Energy Transition Outlook [13]) We considered potential improvements for each Si PV technology based on industrial outlooks (e.g., ITRPV 2021 [39]). Linear regression was used to extrapolate technology improvements for each year between 2021 and 2031. For example, HJT Si efficiency is estimated to be 22.3% in 2026 based on the efficiency in 2021 and 2031 (Table 4.1). Figure C2. SHJ Si PV efficiency calculation The amount of materials needed for Si PV manufacturing was calculated using information from Table 4.1 and the following equation. 107 Where A is the quantity of materials needed for manufacturing Si PV sub technology in the next n years. Pi is PV installations in the year i, Mi is market share in the year i, Si is a reduction factor for Si wafer thickness in the year i, Gi is a reduction factor for solar glass thickness in the year i, Ei is the efficiency in the year i, C is the amount of current materials necessary for making PV, and T is the total materials needed for Si PV manufacturing. C2. Materials necessary for manufacturing Si PV the US and the rest of the world Figure C3. Materials necessary for Si PV laminate manufacturing in the US and the rest of the world for the baseline (A), S1 (B), S2 (C), and S3 (D) 108 C3. Materials necessary for manufacturing Si PV sub technology in the US and the rest of the world Figure C4. Materials necessary for Al-BSF Si PV laminate manufacturing in the US for S3 Figure C5. Materials necessary for Al-BSF Si PV laminate manufacturing in the rest of the world for S3 109 Figure C6. Materials necessary for PERC Si PV and HJT Si PV laminate manufacturing in the US for S3 (a) monofacial c-Si PV, (b) monofacial mc-Si PV, (c) bifacial c-Si PV, (d) bifacial mc- Si PV Figure C7. Materials necessary for PERC Si PV and HJT Si PV laminate manufacturing in the rest of the world for S3 (a) monofacial c-Si PV, (b) monofacial mc-Si PV, (c) bifacial c-Si PV, (d) bifacial mc-Si PV 110 Figure C8. Materials necessary for perovskite/silicon tandem PV laminate manufacturing in the USA and the rest of the world for S3 111 Figure C9. The material intensity for manufacturing Si PV laminate globally 112 APPENDIX D: Supplementary Information for Chapter 5 113 D1. Previous environmental impact assessment of silicon photovoltaic (Si PV) Life cycle assessment (LCA) has been widely used to evaluate Si PV manufacturing's environmental impacts. This section summarized Si PV LCA studies that included global warming potential (GWP) (Figure D1 and D2) and cumulative energy demand (CED) (Figure D3 and D4). Figure D1. The GWP of monocrystalline Si PV manufacturing based on studies from 1990 to 2016 [312–329] Figure D2. The GWP of multi-crystalline Si PV manufacturing based on studies from 1990 to 2018 [181,314–318,320–324,329–341] 114 Figure D3. The CED of multi-crystalline Si PV manufacturing based on studies from 1990 to 2016 [181,314–317,322–324,330–333,335,337–345] Figure D4. The CED of multi-crystalline Si PV manufacturing based on studies from 1990 to 2018 [181,314–316,321–324,330,331,334,341,342,344–347] Table D1. Analysis of previous studies on environmental impact assessment of Si PV. Reference Studies ignored quartz purity [181,314– 318,320,321,323,325,329,330,333– 337,339–341,344,346–354] Studies ignored quartz mining [313,320,332,340,342,355] and silica sand extraction 115 D2. Previous environmental impact assessment of silicon photovoltaic (Si PV) Figure D5. Quartz mining (High-quality) Figure D6. Silica sand extraction (High-quality) Figure D7. Quartz mining (industrial grade and low-quality) 116 Figure D8. Silica sand extraction (magnetic separation technique) Figure D9. Silica sand extraction (flotation separation technique) Figure D10. Silica sand extraction (gravity separation technique) Figure D11. Metallurgical grade silicon (MG-Si) production 117 D3. Data source for life-cycle assessment Table D2. Data sources for materials and energy used in the inventory analysis Details Ref. Quartz mining It was modeled based on the process of quartz mining and silica sand [205,206,356– and silica sand extraction in Unimin Corporation, the global leading silica sand 359] extraction producer. They have 21 active quartz deposits throughout the USA (industrial- with 41 million annual productions. Their annual silica sand grade quartz) production includes 83.35% high-quality quartz, and the rest consists of 6.43% magnetic separation and 10.22% flotation. Quartz mining It was modeled based on a quartz mine that used three separation and silica sand techniques: 1/3 magnetic, 1/3 flotation, and 1/3 gravity. extraction (low- quality quartz) MG-Si We used quartz with different purity for MG-Si production: high- [4,256,360] quality, industrial-grade, and low-quality quartz mines. Distance for For silica sand export to China, we assumed distance between foreign [4,361,362] silica sand resources and the closest port to the target MG-Si producer. - Australia to China (Shanghai): 4,584 km - Cambodia to China (Yunnan province): 1,500 km - Malaysia to China (Beihai port): 2,485 km - North Korea to China (Yantai port): 408 km - Pakistan to China (Xinjiang province): 700 km electricity Regional electricity was used for modeling electricity in China. We [258,269,363– divided China into six regions (Figure 5. 9): Northwest, North, 365] Northeast, East, South, and Center. The leading province producer was used as the representative of each region: Northwest (Xinjiang), North (Beijing), Northeast (Liaoning), East (Fujian), South (Yunnan), Center (Sichuan). D4. Silica sand application Industrial silica sand (95% silica) is used for hydraulic fracturing, glass making, foundry, filtration, and manufacturing semiconductor and solar panels (Figure D12). Industrial silica sand production has tripled since 2010. 118 Figure D12. Applications of a) sand and b) industrial sand. It was compiled based on the USGS report [199] and ISSST2020 conference [366] D5. Illegal silica sand extraction in India Figure D13. Illegal silica sand trades in 2016 [245–247,263–266] 119 REFERENCES 120 REFERENCES [1] IRENA, Global Renewables Outlook: Energy transformation 2050, 2020. https://www.irena.org/publications/2020/Apr/Global-Renewables-Outlook-2020. [2] Annual Energy Outlook 2021 with projections to 2050, U.S. Energy Information Administration (EIA). [3] G. Masson, I. Kaizuka, E. Bosch, A. Detollenaere, G. Neubourg, J. van Wetter, J. Lindahl, IEA PVPS report - Trends in Photovoltaic Applications 2020, 2020. www.iea-pvps.org. [4] R. Frischknecht, P. Stolz, L. Krebs, M.J. de Wild-Scholten, P. Sinha, Life Cycle Inventories and Life Cycle Assessments of Photovoltaic Systems (IEA PVPS), Task 12: PV Sustainability, 2020. www.iea-pvps.org. [5] Y. Cui, L. Hong, J. Hou, Organic Photovoltaic Cells for Indoor Applications: Opportunities and Challenges, ACS Appl. Mater. Interfaces. (2020). doi:10.1021/acsami.0c10444. [6] International Renewable Energy Agency (IRENA), Future of Solar Photovoltaics, 2019. https://www.irena.org/- /media/Files/IRENA/Agency/Publication/2019/Oct/IRENA_Future_of_wind_2019.pdf. [7] R. García-Valverde, J.A. Cherni, A. Urbina, Life cycle analysis of organic photovoltaic technologies, Prog. Photovoltaics Res. Appl. 18 (2010) 535–558. doi:10.1002/pip.967. [8] Y. Cui, H. Yao, T. Zhang, L. Hong, B. Gao, K. Xian, J. Qin, J. Hou, 1 cm2 Organic Photovoltaic Cells for Indoor Application with over 20% Efficiency, Adv. Mater. 31 (2019) 1–7. doi:10.1002/adma.201904512. [9] A. Gambhir, P. Sandwell, J. Nelson, The future costs of OPV – A bottom-up model of material and manufacturing costs with uncertainty analysis, Sol. Energy Mater. Sol. Cells. 156 (2016) 49–58. doi:10.1016/j.solmat.2016.05.056. [10] D. Freier, R. Ramirez-Iniguez, T. Jafry, F. Muhammad-Sukki, C. Gamio, A review of optical concentrators for portable solar photovoltaic systems for developing countries, Renew. Sustain. Energy Rev. 90 (2018) 957–968. doi:10.1016/j.rser.2018.03.039. [11] K.S. Chen, J.F. Salinas, H.L. Yip, L. Huo, J. Hou, A.K.Y. Jen, Semi-transparent polymer solar cells with 6% PCE, 25% average visible transmittance and a color rendering index close to 100 for power generating window applications, Energy Environ. Sci. 5 (2012) 9551–9557. doi:10.1039/c2ee22623e. [12] M. Kaltenbrunner, M.S. White, E.D. Głowacki, T. Sekitani, T. Someya, N.S. Sariciftci, S. Bauer, Ultrathin and lightweight organic solar cells with high flexibility, Nat. Commun. 3 (2012). doi:10.1038/ncomms1772. 121 [13] F. Liu, Z. Zhou, C. Zhang, J. Zhang, Q. Hu, T. Vergote, F. Liu, T.P. Russell, X. Zhu, Efficient Semitransparent Solar Cells with High NIR Responsiveness Enabled by a Small- Bandgap Electron Acceptor, Adv. Mater. 29 (2017). doi:10.1002/adma.201606574. [14] L. Wen, Q. Chen, F. Sun, S. Song, L. Jin, Y. Yu, Theoretical design of multi-colored semi- transparent organic solar cells with both efficient color filtering and light harvesting, Sci. Rep. 4 (2014) 7036. doi:10.1038/srep07036. [15] G. Xu, L. Shen, C. Cui, S. Wen, R. Xue, W. Chen, H. Chen, J. Zhang, H. Li, Y. Li, Y. Li, High-Performance Colorful Semitransparent Polymer Solar Cells with Ultrathin Hybrid- Metal Electrodes and Fine-Tuned Dielectric Mirrors, Adv. Funct. Mater. 27 (2017). doi:10.1002/adfm.201605908. [16] P. Cheng, H.C. Wang, R. Zheng, Y. Zhu, S. Dai, Z. Li, C.H. Chen, Y. Zhao, R. Wang, D. Meng, C. Zhu, K.H. Wei, X. Zhan, Y. Yang, Enabling High-Performance Tandem Organic Photovoltaic Cells by Balancing the Front and Rear Subcells, Adv. Mater. 2002315 (2020) 1–6. doi:10.1002/adma.202002315. [17] H. Yao, J. Wang, Y. Xu, S. Zhang, J. Hou, Recent Progress in Chlorinated Organic Photovoltaic Materials, Acc. Chem. Res. 53 (2020) 822–832. doi:10.1021/acs.accounts.0c00009. [18] C.J.M. Emmott, J.A. Röhr, M. Campoy-Quiles, T. Kirchartz, A. Urbina, N.J. Ekins-Daukes, J. Nelson, Organic photovoltaic greenhouses: A unique application for semi-transparent PV?, Energy Environ. Sci. 8 (2015) 1317–1328. doi:10.1039/c4ee03132f. [19] NREL, Best Research-Cell Efficiencies: Rev. 04-06-2020, (2020). https://www.nrel.gov/pv/cell-efficiency.html. [20] S.M. Heidari, A. Anctil, Identifying alternative solvents for C60 manufacturing using singular and combined toxicity assessments, J. Hazard. Mater. 393 (2020) 122337. doi:10.1016/j.jhazmat.2020.122337. [21] A. Anctil, C.W. Babbitt, R.P. Raffaelle, B.J. Landi, Material and energy intensity of fullerene production, Environ. Sci. Technol. 45 (2011) 2353–2359. doi:10.1021/es103860a. [22] A. Elshkaki, T.E. Graedel, Dynamic analysis of the global metals flows and stocks in electricity generation technologies, J. Clean. Prod. 59 (2013) 260–273. doi:10.1016/j.jclepro.2013.07.003. [23] W. Klopffer, Life cycle assessment, Environ. Sci. Pollut. Res. 4 (1997) 223–228. [24] P.G. Jessop, S. Trakhtenberg, J. Warner, The Twelve Principles of Green Chemistry, in: Innov. Ind. Eng. Chem., American Chemical Society, 2008: pp. 401–436. doi:10.1021/bk- 2009-1000.ch012. [25] H. Duan, D. Wang, Y. Li, Green chemistry for nanoparticle synthesis, Chem. Soc. Rev. 44 (2015) 5778–5792. doi:10.1039/c4cs00363b. 122 [26] D.J. Burke, D.J. Lipomi, Green chemistry for organic solar cells, Energy Environ. Sci. 6 (2013) 2053. doi:10.1039/c3ee41096j. [27] P.T. Anastas, J.C. Warner, Green Chemistry: Theory and Practice, Oxfor University Press Inc., 2000. [28] J. Beyer, K. Petersen, Y. Song, A. Ruus, M. Grung, T. Bakke, K.E. Tollefsen, Environmental risk assessment of combined effects in aquatic ecotoxicology: A discussion paper, Mar. Environ. Res. 96 (2014) 81–91. doi:10.1016/j.marenvres.2013.10.008. [29] US Environmental Ptotection Agency (EPA), Toxicity Estimation Software Tool (TEST). Available from: https://www.epa.gov/chemical-research/toxicity-estimation-software-tool- test. [30] K.E. Biesinger, L.R. Williams, W.H. Van Der Schalie, Procedures for Conducting Daphnia magna toxicity bioassays (EPA/600/8-87/011), Las Vegas, 1987. doi:EPA/600/8-87/011. [31] L. Guilhermino, T. Diamantino, M. Carolina Silva, A.M.V.M. Soares, Acute toxicity test with Daphnia magna: An alternative to mammals in the prescreening of chemical toxicity?, Ecotoxicol. Environ. Saf. 46 (2000) 357–362. doi:10.1006/eesa.2000.1916. [32] S. Dwivedi, S.S. Verma, C. D’Souza, N. Awasthee, A. Sharma, S. Chandra Gupta, Potential of Small Animals in Toxicity Testing, in: Biomarkers Toxicol., Second Edi, Elsevier, 2019: pp. 129–142. doi:10.1016/B978-0-12-814655-2.00007-4. [33] E. Warren, On the Reaction of Daphnia magna (Straus) to certain Changes in its Environment, J. Cell Sci. s2-43 (1900) 199–224. [34] E.E. Kenaga, Test organisms and methods useful for early assessment of acute toxicity of chemicals, Environ. Sci. Technol. 12 (1978) 1322–1329. doi:10.1021/es60147a001. [35] M.L. Dini, J. O’Donnell, S.R. Carpenter, M.M. Elser, J.J. Elser, A.M. Bergquist, Daphnia size structure, vertical migration, and phosphorus redistribution, Hydrobiologia. 150 (1987) 185–191. doi:10.1007/BF00006666. [36] S. Kayasth, K. Swain, Role of analytical chemistry in environmental monitoring, J. Radioanal. Nucl. Chem. 262 (2004) 35–42. doi:10.1023/B:JRNC.0000040851.34658.e2. [37] M.A. Fusella, Y.L. Lin, B.P. Rand, Organic photovoltaics (OPVs): Device physics, in: Handb. Org. Mater. Electron. Photonic Devices, 2nd ed., Elsevier, 2019: pp. 665–693. doi:10.1016/B978-0-08-102284-9.00020-6. [38] M.D.M. Faure, B.H. Lessard, Layer-by-layer fabrication of organic photovoltaic devices: Material selection and processing conditions, J. Mater. Chem. C. 9 (2021) 14–40. doi:10.1039/d0tc04146g. [39] International Technology Roadmap for Photovoltaic (ITRPV 2021), 2020. https://itrpv.vdma.org/en/ueber-uns. 123 [40] P. Ehrenfreund, B.H. Foing, FULLERENES IN SPACE, Elsevier Sci. Ltd. 19 (1997) 1033– 1042. doi:https://doi.org/10.1016/S0273-1177(97)00350-5. [41] J.P. Maier, E.K. Campbell, Fullerenes in Space, Angew. Chemie - Int. Ed. 56 (2017) 4920– 4929. doi:10.1002/anie.201612117. [42] H.W. Kroto, J.R. Heath, S.C. O’Brien, R.F. Curl, R.E. Smalley, C 60: buckminsterfullerene, Nature. 318 (1985) 162–163. doi:10.1038/318162a0. [43] S. Manzetti, O. Andersen, Toxicological aspects of nanomaterials used in energy harvesting consumer electronics, Renew. Sustain. Energy Rev. 16 (2012) 2102–2110. doi:10.1016/j.rser.2012.01.037. [44] H. Kazemzadeh, M. Mozafari, Fullerene-based delivery systems, Drug Discov. Today. 24 (2019) 898–905. doi:10.1016/j.drudis.2019.01.013. [45] R. Xie, Z. Wang, H. Yu, Z. Fan, F. Yuan, Y. Li, X. Li, L. Fan, H. Fan, Highly Water-soluble and Surface Charge-tunable Fluorescent Fullerene Nanoparticles: Facile Fabrication and Cellular Imaging, Electrochim. Acta. 201 (2016) 220–227. doi:10.1016/j.electacta.2016.03.198. [46] M. Pitorre, H. Gondé, C. Haury, M. Messous, J. Poilane, D. Boudaud, E. Kanber, G.A. Rossemond Ndombina, J.P. Benoit, G. Bastiat, Recent advances in nanocarrier-loaded gels: Which drug delivery technologies against which diseases?, J. Control. Release. 266 (2017) 140–155. doi:10.1016/j.jconrel.2017.09.031. [47] K.N. Semenov, N.A. Charykov, V.N. Postnov, V. V. Sharoyko, I. V. Vorotyntsev, M.M. Galagudza, I. V. Murin, Fullerenols: Physicochemical properties and applications, Prog. Solid State Chem. 44 (2016) 59–74. doi:10.1016/j.progsolidstchem.2016.04.002. [48] C. Parlak, Ö. Alver, A density functional theory investigation on amantadine drug interaction with pristine and B, Al, Si, Ga, Ge doped C60 fullerenes, Chem. Phys. Lett. 678 (2017) 85–90. doi:10.1016/j.cplett.2017.04.025. [49] G. Herlem, F. Picaud, C. Girardet, O. Micheau, Carbon Nanotubes: Synthesis, Characterization, and Applications in Drug- Delivery Systems, in: Nanocarriers Drug Deliv., Elsevier Inc., 2019: pp. 469–529. doi:10.1016/B978-0-12-814033-8.00016-3. [50] J.J. Ryan, H.R. Bateman, A. Stover, G. Gomez, S.K. Norton, W. Zhao, L.B. Schwartz, R. Lenk, C.L. Kepley, Fullerene Nanomaterials Inhibit the Allergic Response, J. Immunol. 179 (2007) 665–672. doi:10.4049/jimmunol.179.1.665. [51] Z. Zhou, R. Lenk, A. Dellinger, D. Macfarland, K. Kumar, S.R. Wilson, C.L. Kepley, L. Nanoworks, Fullerene nanomaterials potentiate hair growth, doi:10.1016/j.nano.2008.09.005. [52] L.L. Dugan, E.G. Lovett, K.L. Quick, J. Lotharius, T.T. Lin, K.L. O’Malley, Fullerene- based antioxidants and neurodegenerative disorders, Park. Relat. Disord. 7 (2001) 243–246. 124 doi:10.1016/S1353-8020(00)00064-X. [53] SES Research, SES Res. https://www.sesres.com/product/carbon-60-99-95-and-olive-oil- extra-virgin-cert-organic-100ml/?gclid=CjwKCAjw- 7LrBRB6EiwAhh1yXygW_ZIPzh8KoeIa8-RKlnG- lgnuVGB4FXNKzbBSSSDKcnyKBE5g2RoCUa8QAvD_BwE (accessed October 9, 2019). [54] Bccresearch, The Global Market for Fullerenes, (2006). https://www.bccresearch.com/market-research/nanotechnology/NAN034A.html (accessed December 2, 2019). [55] MordorIntelligence, Fullerene market, (2019). https://www.mordorintelligence.com/industry-reports/fullerene-market (accessed December 2, 2019). [56] K.N. Semenov, N.A. Charykov, V.A. Keskinov, A.K. Piartman, A.A. Blokhin, A.A. Kopyrin, Solubility of light fullerenes in organic solvents, J. Chem. Eng. Data. 55 (2010) 13–36. doi:10.1021/je900296s. [57] K. Nagata, E. Dejima, Y. Kikuchi, M. Hashiguchi, Kilogram-scale [60]Fullerene Separation from a Fullerene Mixture: Selective Complexation of Fullerenes with 1,8- Diazabicyclo[5.4.0]undec-7-ene (DBU), Org. Process Res. Dev. 9 (2005) 660–662. doi:10.1021/op0500943. [58] K. Nagata, E. Dejima, Y. Kikuchi, M. Hashiguchi, Efficient and Scalable Method for [60]Fullerene Separation from a Fullerene Mixture: Selective Complexation of Fullerenes with DBU in the Presence of Water, Org. Process Res. Dev. 9 (2005) 660–662. doi:10.1021/op0500943. [59] S.M. Taghavi, M. Momenpour, M. Azarian, M. Ahmadian, F. Souri, S.A. Taghavi, M. Sadeghain, M. Karchani, Effects of Nanoparticles on the Environment and Outdoor Workplaces., Electron. Physician. 5 (2013) 706–70612. doi:10.14661/2013.706-712. [60] F.M. Kerton, R. Marriot, Alternative Solvents for Green Chemistry, 2nd ed., The Royal Society of Chemistry, Cambridge, 2013. [61] F.P. Byrne, S. Jin, G. Paggiola, T.H.M. Petchey, J.H. Clark, T.J. Farmer, A.J. Hunt, C. Robert McElroy, J. Sherwood, Tools and techniques for solvent selection: green solvent selection guides, Sustain. Chem. Process. 4 (2016) 1–24. doi:10.1186/s40508-016-0051-z. [62] R. Ratti, Industrial applications of green chemistry : Status , Challenges and Prospects, SN Appl. Sci. (2020). doi:10.1007/s42452-020-2019-6. [63] I.T. Horváth, Green Chemistry, Acc. Chem. Res. 35 (2002) 685–685. doi:10.1021/ar020160a. [64] P.T. Anastas, R.L. Lankey, Life cycle assessment and green chemistry: the yin and yang of 125 industrial ecology, Green Chem. 2 (2000) 289–295. doi:10.1039/b005650m. [65] P.T. Anastas, J.C. Warner, Green Chemistry: Theory and Practice, Oxford University Press, U.S.A., 2000. [66] M.J. Mulvihill, E.S. Beach, J.B. Zimmerman, P.T. Anastas, Green Chemistry and Green Engineering: A Framework for Sustainable Technology Development, Annu. Rev. Environ. Resour. 36 (2011) 271–293. doi:10.1146/annurev-environ-032009-095500. [67] O. Koul, S. Walia, G.S. Dhaliwal, Essential oils as green pesticides potential and constraints, Biopestic. Insect. 4 (2008) 63–84. [68] P.J. Dunn, The importance of Green Chemistry in Process Research and Development, Chem. Soc. Rev. 41 (2012) 1452–1461. doi:10.1039/c1cs15041c. [69] J.A. Brant, J. Labille, J.-Y. Bottero, M.R. Wiesner, Characterizing the Impact of Preparation Method on Fullerene Cluster Structure and Chemistry, Langmuir. 22 (2006) 3878–3885. doi:10.1021/la053293o. [70] L.A. Vashchenkob, On the Production of an Aqueous Colloidal Solution of Fullerenes, 1281–1282. doi:https://doi.org/10.1039/C39950001281. [71] E. Oberdörster, S. Zhu, T.M. Blickley, P. McClellan-Green, M.L. Haasch, Ecotoxicology of carbon-based engineered nanoparticles: Effects of fullerene (C60) on aquatic organisms, Carbon N. Y. 44 (2006) 1112–1120. doi:10.1016/j.carbon.2005.11.008. [72] J.D. Fortner, D.Y. Lyon, C.M. Sayes, A.M. Boyd, J.C. Falkner, E.M. Hotze, L.B. Alemany, Y.J. Tao, W. Guo, K.D. Ausman, V.L. Colvin, J.B. Hughes, C 60 in Water : Nanocrystal Formation and Microbial Response, Env. Sci Technol. 39 (2005) 4307–4316. doi:10.1021/es048099n. [73] J. Brant, H. Lecoanet, M. Hotze, M. Wiesner, Comparison of electrokinetic properties of colloidal fullerenes (n-C60) formed using two procedures, Environ. Sci. Technol. 39 (2005) 6343–6351. doi:10.1021/es050090d. [74] K. Pakarinen, E.J. Petersen, L. Alvila, G.C. Waissi-Leinonen, J. Akkanen, M.T. Leppänen, J.V.K. Kukkonen, A screening study on the fate of fullerenes (nC60) and their toxic implications in natural freshwaters, Environ. Toxicol. Chem. 32 (2013) 1224–1232. doi:10.1002/etc.2175. [75] S. Zhu, E. Oberdörster, M.L. Haasch, Toxicity of an engineered nanoparticle (fullerene, C60) in two aquatic species, Daphnia and fathead minnow, Mar. Environ. Res. 62 (2006) S5–S9. doi:10.1016/j.marenvres.2006.04.059. [76] S.B. Lovern, R. Klaper, Daphnia magna mortality when exposed to titanium dioxide and fullerene (C60) nanoparticles, Environ. Toxicol. Chem. 25 (2006) 1132–1137. doi:10.1897/05-278R.1. 126 [77] B. Salieri, D.A. Turner, B. Nowack, R. Hischier, Life cycle assessment of manufactured nanomaterials: Where are we?, NanoImpact. 10 (2018) 108–120. doi:10.1016/j.impact.2017.12.003. [78] R.K. Rosenbaum, T.M. Bachmann, L.S. Gold, M.A.J. Huijbregts, O. Jolliet, R. Juraske, A. Koehler, H.F. Larsen, M. MacLeod, M. Margni, T.E. McKone, J. Payet, M. Schuhmacher, D. Van De Meent, M.Z. Hauschild, USEtox - The UNEP-SETAC toxicity model: Recommended characterisation factors for human toxicity and freshwater ecotoxicity in life cycle impact assessment, Int. J. Life Cycle Assess. 13 (2008) 532–546. doi:10.1007/s11367- 008-0038-4. [79] P. Fantke, M. Bijster, C. Guignard, M. Hauschild, M. Huijbregts, O. Jolliet, A. Kounina, V. Magaud, M. Margni, T. McKone, L. Posthuma, R.K. Rosenbaum, D. van de Meent, R. van Zelm, 2, USEtox® 2.0, Documentation version 1, 2017. doi:10.11581/DTU:00000011. [80] I. V. Muralikrishna, V. Manickam, Environmental Risk Assessment, in: Environ. Manage., Elsevier, Berlin, Heidelberg, 2017: pp. 135–152. doi:10.1016/B978-0-12-811989-1.00008- 7. [81] EPA, Risk Assessment, (2019). https://www.epa.gov/risk/about-risk-assessment (accessed January 20, 2020). [82] H.E. Buist, R. Hischier, J. Westerhout, D.H. Brouwer, Derivation of health effect factors for nanoparticles to be used in LCIA, NanoImpact. 7 (2017) 41–53. doi:10.1016/j.impact.2017.05.002. [83] M.. Hernando, M. Ejerhoon, A.. Fernández-Alba, Y. Chisti, Combined toxicity effects of MTBE and pesticides measured with Vibrio fischeri and Daphnia magna bioassays, Water Res. 37 (2003) 4091–4098. doi:10.1016/S0043-1354(03)00348-8. [84] K.T. Kim, Y.G. Lee, S.D. Kim, Combined toxicity of copper and phenol derivatives to Daphnia magna: Effect of complexation reaction, Environ. Int. 32 (2006) 487–492. doi:10.1016/j.envint.2005.11.002. [85] K. Kim, H.-J. Jeon, S.-D. Choi, D.C.W. Tsang, P. Oleszczuk, Y.S. Ok, H.-S. Lee, S.-E. Lee, Combined toxicity of endosulfan and phenanthrene mixtures and induced molecular changes in adult Zebrafish (Danio rerio), Chemosphere. 194 (2018) 30–41. doi:10.1016/j.chemosphere.2017.11.128. [86] L. Xing, J. Sun, H. Liu, H. Yu, Combined toxicity of three chlorophenols 2,4- dichlorophenol, 2,4,6-trichlorophenol and pentachlorophenol to Daphnia magna, J. Environ. Monit. 14 (2012) 1677. doi:10.1039/c2em30185g. [87] Z. Zhu, S. Wang, F. Zhao, S. Wang, F. Liu, G. Liu, Joint toxicity of microplastics with triclosan to marine microalgae Skeletonema costatum, Environ. Pollut. 246 (2019) 509– 517. doi:10.1016/j.envpol.2018.12.044. [88] E. Geiger, R. Hornek-Gausterer, M.T. Saçan, Single and mixture toxicity of 127 pharmaceuticals and chlorophenols to freshwater algae Chlorella vulgaris, Ecotoxicol. Environ. Saf. 129 (2016) 189–198. doi:10.1016/j.ecoenv.2016.03.032. [89] T. Backhaus, M. Scholze, L.H. Grimme, The single substance and mixture toxicity of quinolones to the bioluminescent bacterium Vibrio fischeri, Aquat. Toxicol. 49 (2000) 49– 61. doi:10.1016/S0166-445X(99)00069-7. [90] M. Adamczak, M. Krok, E. Pamuła, U. Posadowska, K. Szczepanowicz, J. Barbasz, P. Warszyński, Linseed oil based nanocapsules as delivery system for hydrophobic quantum dots, Colloids Surfaces B Biointerfaces. 110 (2013) 1–7. doi:10.1016/j.colsurfb.2013.04.014. [91] H. Bahadar, F. Maqbool, K. Niaz, M. Abdollahi, Toxicity of nanoparticles and an overview of current experimental models, Iran. Biomed. J. 20 (2016) 1–11. doi:10.7508/ibj.2016.01.001. [92] D. Drobne, V. Kralj-Iglič, Lipid Membranes as Tools in Nanotoxicity Studies, in: Adv. Planar Lipid Bilayers Liposomes, 2009: pp. 121–134. doi:10.1016/S1554-4516(09)10005- 4. [93] M. Raoof, Y. Mackeyev, M.A. Cheney, L.J. Wilson, S.A. Curley, Internalization of C60 fullerenes into cancer cells with accumulation in the nucleus via the nuclear pore complex, Biomaterials. 33 (2012) 2952–2960. doi:10.1016/j.biomaterials.2011.12.043. [94] R. Qiao, A.P. Roberts, A.S. Mount, S.J. Klaine, P.C. Ke, Translocation of C 60 and Its Derivatives Across a Lipid Bilayer, Nano Lett. 7 (2007) 614–619. doi:10.1021/nl062515f. [95] C.M. Sayes, J.D. Fortner, W. Guo, D. Lyon, A.M. Boyd, K.D. Ausman, Y.J. Tao, B. Sitharaman, L.J. Wilson, J.B. Hughes, J.L. West, V.L. Colvin, The differential cytotoxicity of water-soluble fullerenes, Nano Lett. 4 (2004) 1881–1887. doi:10.1021/nl0489586. [96] A. Grebinyk, S. Grebinyk, S. Prylutska, U. Ritter, O. Matyshevska, T. Dandekar, M. Frohme, C60 fullerene accumulation in human leukemic cells and perspectives of LED- mediated photodynamic therapy, Free Radic. Biol. Med. 124 (2018) 319–327. doi:10.1016/j.freeradbiomed.2018.06.022. [97] P. Mroz, G.P. Tegos, H. Gali, T. Wharton, T. Sarna, M.R. Hamblin, Photodynamic therapy with fullerenes, Photochem. Photobiol. Sci. 6 (2007) 1139–1149. doi:10.1039/b711141j. [98] M.R. Hamblin, Fullerenes as photosensitizers in photodynamic therapy: Pros and cons, Photochem. Photobiol. Sci. 17 (2018) 1515–1533. doi:10.1039/c8pp00195b. [99] R. Bakry, R.M. Vallant, M. Najam-ul-Haq, M. Rainer, Z. Szabo, C.W. Huck, G.K. Bonn, Medicinal applications of fullerenes, Int. J. Nanomedicine. 2 (2007) 639–649. doi:10.1016/S0081-1947(08)60578-0. [100] J.W. Arbogast, A.P. Darmanyan, C.S. Foote, F.N. Diederich, R.L. Whetten, Y. Rubin, M.M. Alvarez, S.J. Anz, Photophysical properties of sixty atom carbon molecule (C60), J. Phys. 128 Chem. 95 (1991) 11–12. doi:10.1021/j100154a006. [101] M. Salehi, R. Rodriguez, A. Boettcher, S. Powers, N. Geitner, D.A. Ladner, S. Rikard, A.J. Whelton, Impact of dispersant on early life stages of the water flea Daphnia magna and the eastern oyster Crassostrea virginica, J. Appl. Toxicol. 37 (2017) 1464–1470. doi:10.1002/jat.3494. [102] K.D.M. Harris, N.J. Bartlett, V.K. Lloyd, Daphnia as an Emerging Epigenetic Model Organism, Genet. Res. Int. 2012 (2012) 1–8. doi:10.1155/2012/147892. [103] W. Lin, R. Jiang, Y. Xiong, J. Wu, J. Xu, J. Zheng, F. Zhu, G. Ouyang, Quantification of the combined toxic effect of polychlorinated biphenyls and nano-sized polystyrene on Daphnia magna, J. Hazard. Mater. 364 (2019) 531–536. doi:10.1016/j.jhazmat.2018.10.056. [104] E. Lee, C.J. Andrews, A. Anctil, An Iterative Approach to Evaluate and Guide Fine Chemical Processes: An Example from Chloroaluminum Phthalocyanine for Photovoltaic Applications, ACS Sustain. Chem. Eng. 6 (2018) 8230–8237. doi:10.1021/acssuschemeng.7b04947. [105] F. Martínez-Jerónimo, R. Villaseñor, G. Ríos, F. Espinosa-Chavez, Toxicity of the crude oil water-soluble fraction and kaolin-adsorbed crude oil on Daphnia magna (Crustacea: Anomopoda), Arch. Environ. Contam. Toxicol. 48 (2005) 444–449. doi:10.1007/s00244- 003-0220-4. [106] R.W. Woods, D.J. Letinski, E.J. Febbo, C.L. Dzamba, M.J. Connelly, T.F. Parkerton, Assessing the aquatic hazard of commercial hydrocarbon resins, Ecotoxicol. Environ. Saf. 66 (2007) 159–168. doi:10.1016/j.ecoenv.2005.11.004. [107] M.M. Singer, D. Aurand, G.E. Bragin, J.R. Clark, G.M. Coelho, M.L. Sowby, R.S. Tjeerdema, Standardization of the preparation and quantitation of water-accommodated fractions of petroleum for toxicity testing, Mar. Pollut. Bull. 40 (2000) 1007–1016. doi:10.1016/S0025-326X(00)00045-X. [108] D.Y. Lyon, L.K. Adams, J.C. Falkner, P.J.J. Alvarez, Antibacterial activity of fullerene water suspensions: Effects of preparation method and particle size, Environ. Sci. Technol. 40 (2006) 4360–4366. doi:https://doi.org/10.1021/es0603655. [109] E.A. Moore, C.W. Babbitt, S.J. Connelly, A.C. Tyler, G. Rogalskyj, Cascading Ecological Impacts of Fullerenes in Freshwater Ecosystems, Environ. Toxicol. Chem. 38 (2019) 1714– 1723. doi:10.1002/etc.4465. [110] T. Backhaus, Å. Arrhenius, H. Blanck, Toxicity of a Mixture of Dissimilarly Acting Substances to Natural Algal Communities: Predictive Power and Limitations of Independent Action and Concentration Addition, Environ. Sci. Technol. 38 (2004) 6363– 6370. doi:10.1021/es0497678. [111] R. Altenburger, T. Backhaus, W. Boedeker, M. Faust, M. Scholze, L.H. Grimme, 129 PREDICTABILITY OF THE TOXICITY OF MULTIPLE CHEMICAL MIXTURES TO VIBRIO FISCHERI: MIXTURES COMPOSED OF SIMILARLY ACTING CHEMICALS, Environ. Toxicol. Chem. 19 (2000) 2341. doi:10.1897/1551- 5028(2000)019<2341:POTTOM>2.3.CO;2. [112] T. Backhaus, R. Altenburger, W. Boedeker, M. Faust, M. Scholze, L.H. Grimme, Predictability of the toxicity of a multiple mixture of dissimilarly acting chemicals to Vibrio fischeri, Environ. Toxicol. Chem. 19 (2000) 2348–2356. doi:10.1002/etc.5620190927. [113] J.B. Sprague, Measurement of pollutant toxicity to fish. II. Utilizing and applying bioassay results, Water Res. 4 (1970) 3–32. doi:10.1016/0043-1354(70)90018-7. [114] T.J. Pandolfo, W.G. Cope, C. Arellano, Heart rate as a sublethal indicator of thermal stress in juvenile freshwater mussels, Comp. Biochem. Physiol. Part A Mol. Integr. Physiol. 154 (2009) 347–352. doi:10.1016/j.cbpa.2009.07.001. [115] I.B. Oliveira, K.J. Groh, R. Schönenberger, C. Barroso, K. V. Thomas, M.J.F. Suter, Toxicity of emerging antifouling biocides to non-target freshwater organisms from three trophic levels, Aquat. Toxicol. 191 (2017) 164–174. doi:10.1016/j.aquatox.2017.07.019. [116] K. Tatsi, A. Turner, R.D. Handy, B.J. Shaw, The acute toxicity of thallium to freshwater organisms: Implications for risk assessment, Sci. Total Environ. 536 (2015) 382–390. doi:10.1016/j.scitotenv.2015.06.069. [117] J. Brant, H. Lecoanet, M.R. Wiesner, Aggregation and deposition characteristics of fullerene nanoparticles in aqueous systems, J. Nanoparticle Res. 7 (2005) 545–553. doi:10.1007/s11051-005-4884-8. [118] A. Kumar, C.K. Dixit, Methods for characterization of nanoparticles, in: Adv. Nanomedicine Deliv. Ther. Nucleic Acids, Elsevier, 2017: pp. 43–58. doi:10.1016/B978-0- 08-100557-6.00003-1. [119] Livepet, C60 in olive oil, (2019). https://livepet.co/shop/ (accessed December 1, 2019). [120] C60-france, C60 in olive oil, C60 in sunflower oil, C60 in avacado oil, and C60 in coconut oil, (2019). https://c60-france.com/en/12-oils-c60 (accessed December 1, 2019). [121] PureC60Oliveoil, C60 in avacado oil, C60 in Hemp oil, C60 in coconut oil, C60 in olive oil, (2019). https://purec60oliveoil.com/collections/all (accessed December 1, 2019). [122] Goodandcheapc60, C60 in olive oil, (2019). https://www.goodandcheapc60oo.com (accessed December 1, 2019). [123] SES Research Group, C60 in Olive oil, (2019). https://www.sesres.com/carbon-60-olive- oil/ (accessed December 1, 2019). [124] GRESKA’S CARBON-60, C60 in Sunflower oil, (2019). 130 [125] Pharmakon, C60 in olive oil (softgel), (2019). http://www.pharmakon.bio/home2.html (accessed December 1, 2019). [126] C360health, C60 in olive oil, (2019). https://c360health.com (accessed December 1, 2019). [127] Mybodysymphony, C60 in pumpkin seed oil, (2019). https://www.mybodysymphony.com/products-8-carbon-60/ (accessed December 1, 2019). [128] M. Paut Kusturica, A. Tomas, A. Sabo, Disposal of Unused Drugs: Knowledge and Behavior Among People Around the World, in: How to Recruit Volunt. Donors Third World?, 2016: pp. 71–104. doi:10.1007/398_2016_3. [129] S. Abernethy, A.M. Bobra, W.Y. Shiu, P.G. Wells, D. Mackay, Acute lethal toxicity of hydrocarbons and chlorinated hydrocarbons to two planktonic crustaceans: The key role of organism-water partitioning, Aquat. Toxicol. 8 (1986) 163–174. doi:10.1016/0166- 445X(86)90062-7. [130] M. Havas, Aluminum bioaccumulation and toxicity to Daphnia magna in soft water at low pH, Can. J. Fish. Aquat. Sci. 42 (1985) 1741–1748. doi:10.1139/f85-218. [131] C.L. Ngan, M. Basri, M. Tripathy, R. Abedi Karjiban, E. Abdul-Malek, Skin intervention of fullerene-integrated nanoemulsion in structural and collagen regeneration against skin aging, Eur. J. Pharm. Sci. 70 (2015) 22–28. doi:10.1016/j.ejps.2015.01.006. [132] H. Masaki, Role of antioxidants in the skin: Anti-aging effects, J. Dermatol. Sci. 58 (2010) 85–90. doi:10.1016/j.jdermsci.2010.03.003. [133] X.Y. Yang, R.E. Edelmann, J.T. Oris, Suspended C60 nanoparticles protect against short- term UV and fluoranthene photo-induced toxicity, but cause long-term cellular damage in Daphnia magna, Aquat. Toxicol. 100 (2010) 202–210. doi:10.1016/j.aquatox.2009.08.011. [134] D.Y. Lyon, J.D. Fortner, C.M. Sayes, V.L. Colvin, J.B. Hughes, Bacterial Cell Association and Antimicrobial Activity of a C60 Water Suspension, Environ. Toxicol. Chem. 24 (2005) 2757–2762. doi:10.1897/04-649R.1. [135] M. De Koning, Reactive Oxygen Species contribute to antibiotic efficacy, UNIVERSITY OF GRONINGEN, 2016. [136] F. Hecht, C.F. Pessoa, L.B. Gentile, D. Rosenthal, D.P. Carvalho, R.S. Fortunato, The role of oxidative stress on breast cancer development and therapy, Tumor Biol. 37 (2016) 4281– 4291. doi:10.1007/s13277-016-4873-9. [137] T.S. Gechev, F. Van Breusegem, J.M. Stone, I. Denev, C. Laloi, Reactive oxygen species as signals that modulate plant stress responses and programmed cell death, BioEssays. 28 (2006) 1091–1101. doi:10.1002/bies.20493. [138] J.L. Marx, Oxygen free radicals linked to many diseases ; the oxygen free radicals , although made as by-products of normal oxygen-using reactions , nevertheless have a wide potential 131 for causing cell injury, Am. Assoc. Adv. Sci. (1987) 6–8. [139] H. Nohl, L. Gille, K. Staniek, Intracellular generation of reactive oxygen species by mitochondria, Biochem. Pharmacol. 69 (2005) 719–723. doi:10.1016/j.bcp.2004.12.002. [140] S. Miwa, J. St-Pierre, L. Partridge, M.D. Brand, Superoxide and hydrogen peroxide production by Drosophila mitochondria, Free Radic. Biol. Med. 35 (2003) 938–948. doi:10.1016/S0891-5849(03)00464-7. [141] K. Baker, C.B. Marcus, K. Huffman, H. Kruk, B. Malfroy, S.R. Doctrow, Synthetic combined superoxide dismutase/catalase mimetics are protective as a delayed treatment in a rat stroke model: A key role for reactive oxygen species in ischemic brain injury, J. Pharmacol. Exp. Ther. 284 (1998) 215–221. [142] E. Oberdörster, Manufactured nanomaterials (fullerenes, C60) induce oxidative stress in the brain of juvenile largemouth bass, Environ. Health Perspect. 112 (2004) 1058–1062. doi:10.1289/ehp.7021. [143] H. Yi, G. Zeng, C. Lai, D. Huang, L. Tang, J. Gong, M. Chen, P. Xu, H. Wang, M. Cheng, C. Zhang, W. Xiong, Environment-friendly fullerene separation methods, Chem. Eng. J. 330 (2017) 134–145. doi:10.1016/j.cej.2017.07.143. [144] D.M. Guldi, Fullerenes: three dimensional electron acceptor materials, Chem. Commun. (2000) 321–327. doi:10.1039/a907807j. [145] R.O. Kesinro, A.O. Boyo, M.L. Akinyemi, M.E. Emetere, A.P. Aizebeokhai, Progress on Organic Solar Cells: A Short Review, IOP Conf. Ser. Earth Environ. Sci. 665 (2021) 012036. doi:10.1088/1755-1315/665/1/012036. [146] S.Z. Mousavi, S. Nafisi, H.I. Maibach, Fullerene nanoparticle in dermatological and cosmetic applications, Nanomedicine Nanotechnology, Biol. Med. 13 (2017) 1071–1087. doi:10.1016/j.nano.2016.10.002. [147] H. Murayama, S. Tomonoh, J.M. Alford, M.E. Karpuk, Fullerene Production in Tons and More: From Science to Industry, Fullerenes, Nanotub. Carbon Nanostructures. 12 (2005) 1–9. doi:10.1081/FST-120027125. [148] M.C. Zumwalt, M.B. Denton, Using High Performance Liquid Chromatography to Determine the C60:C70 Ratio in Fullerene Soot: An Undergraduate Chemistry Lab, J. Chem. Educ. 72 (1995) 939. doi:10.1021/ed072p939. [149] H. Ajie, M.M. Alvarez, S.J. Anz, R.D. Beck, F. Diederich, K. Fostiropoulos, D.R. Huffman, W. Kratschmer, Y. Rubin, K.E. Schriver, D. Sensharma, R.L. Whetten, Characterization of the soluble all-carbon molecules C60 and C70, J. Phys. Chem. 94 (1990) 8630–8633. doi:10.1021/j100387a004. [150] H. Keypour, M. Noroozi, A. Rashidi, An improved method for the purification of fullerene from fullerene soot with activated carbon, celite, and silica gel stationary phases, J. 132 Nanostructure Chem. 3 (2013) 45. doi:10.1186/2193-8865-3-45. [151] A.S. Koch, K.C. Khemani, F. Wudl, Preparation of Fullerenes with a Simple Benchtop Reactor, J. Org. Chem. 56 (1991) 4543–4545. doi:10.1021/jo00014a041. [152] F.W. K. C. Khemani , M. Prato, A simple Soxhlet chromatographic method for the isolation of pure fullerenes C60 and C70, J. Org. Chem. (1992) 3254–3256. doi:10.1021/jo00037a057. [153] P. Bhyrappa, A. Penicaud, M. Kawamoto, C.A. Reed, Improved chromatographic separation and purification of C60 and C70 fullerenes, J. Chem. Soc. Chem. Commun. (1992) 936. doi:10.1039/c39920000936. [154] J.M. Alford, C. Bernal, M. Cates, M.D. Diener, Fullerene production in sooting flames from 1,2,3,4-tetrahydronaphthalene, Carbon N. Y. 46 (2008) 1623–1625. doi:10.1016/j.carbon.2008.07.004. [155] W.A. Scrivens, P. V. Bedworth, J.M. Tour, Purification of Gram Quantities of C60. A New Inexpensive and Facile Method, J. Am. Chem. Soc. 114 (1992) 7917–7919. doi:10.1021/ja00046a051. [156] P.M. Allemand, a Koch, F. Wudl, Y. Rubin, F. Diederich, M.M. Alvarez, S.J. Anz, R.L. Whetten, Two different fullerenes have the same cyclic voltammetry, J. Am. Chem. Soc. 113 (1991) 1050–1051. doi:10.1021/ja00003a053. [157] T. Odagiri, Y.C. Chan, K.S. Kwok, K.M. Ng, A novel evaporative crystallization column for the purification of fullerene C60, AIChE J. 53 (2007) 531–534. doi:10.1002/aic.11079. [158] K.S. Kwok, Y.C. Chan, K.M. Ng, C. Wibowo, Separation of fullerenes C60 and C70 using a crystallization-based process, AIChE J. 56 (2009) 1801–1812. doi:10.1002/aic.12105. [159] I. Bucsi, R. Aniszfeld, T. Shamma, G.K. Prakash, G. a Olah, Convenient separation of high- purity C60 from crude fullerene extract by selective complexation with AlCl3., Proc. Natl. Acad. Sci. 91 (1994) 9019–9021. doi:10.1073/pnas.91.19.9019. [160] K. Nagata, E. Dejima, Y. Kikuchi, M. Hashiguchi, Kilogram-scale [60]Fullerene Separation from a Fullerene Mixture: Selective Complexation of Fullerenes with 1,8- Diazabicyclo[5.4.0]undec-7-ene (DBU), Chem. Lett. 34 (2005) 178–179. doi:10.1246/cl.2005.178. [161] V.K.K. Upadhyayula, D.E. Meyer, M.A. Curran, M.A. Gonzalez, Life cycle assessment as a tool to enhance the environmental performance of carbon nanotube products: a review, J. Clean. Prod. 26 (2012) 37–47. doi:10.1016/j.jclepro.2011.12.018. [162] V.K.K. Upadhyayula, D.E. Meyer, M.A. Curran, M.A. Gonzalez, Evaluating the Environmental Impacts of a Nano-Enhanced Field Emission Display Using Life Cycle Assessment: A Screening-Level Study, (2014). 133 [163] M.J. Eckelman, M.S. Mauter, J.A. Isaacs, M. Elimelech, New perspectives on nanomaterial aquatic ecotoxicity: Production impacts exceed direct exposure impacts for carbon nanotoubes, Environ. Sci. Technol. 46 (2012) 2902–2910. doi:10.1021/es203409a. [164] O.G. Griffiths, J.P. O’Byrne, L. Torrente-Murciano, M.D. Jones, D. Mattia, M.C. McManus, Identifying the largest environmental life cycle impacts during carbon nanotube synthesis via chemical vapour deposition, J. Clean. Prod. 42 (2013) 180–189. doi:10.1016/j.jclepro.2012.10.040. [165] L.M. Gilbertson, J.B. Zimmerman, D.L. Plata, J.E. Hutchison, P.T. Anastas, Designing nanomaterials to maximize performance and minimize undesirable implications guided by the Principles of Green Chemistry, Chem. Soc. Rev. 44 (2015) 5758–5777. doi:10.1039/C4CS00445K. [166] H. Duan, D. Wang, Y. Li, Green chemistry for nanoparticle synthesis, Chem. Soc. Rev. 44 (2015) 5778–5792. doi:10.1039/C4CS00363B. [167] K. Gruiz, I. Fekete-Kertész, Z. Kunglné-Nagy, C. Hajdu, V. Feigl, E. Vaszita, M. Molnár, Direct toxicity assessment — Methods, evaluation, interpretation, Sci. Total Environ. 563– 564 (2016) 803–812. doi:10.1016/j.scitotenv.2016.01.007. [168] S.M. Heidari, A. Anctil, Identifying alternative solvents for C60 manufacturing using singular and combined toxicity assessments, J. Hazard. Mater. 393 (2020) 122337. doi:10.1016/j.jhazmat.2020.122337. [169] PRe´ Consultants, SimaPro 9.1.0.7, Amersfoort, The Netherlands. Available from: https://pre-sustainability.com/solutions/tools/simapro/. [170] M. Berger, R. Van Der Ent, S. Eisner, V. Bach, M. Finkbeiner, Water Accounting and Vulnerability Evaluation (WAVE): Considering Atmospheric Evaporation Recycling and the Risk of Freshwater Depletion in Water Footprinting, (2014). [171] P. Anastas, N. Eghbali, Green Chemistry: Principles and Practice, Chem. Soc. Rev. 39 (2010) 301–312. doi:10.1039/B918763B. [172] NFP Association. “NFPA 704: Standard System for the Identification of the Hazards of Materials for Emergency Response, 2013 Edition.” ed. Quincy, MA: National Fire Protection Agency (2011). Harvard. [173] YMC HPLC Columns Applications Notebook - WA30000, https://www.waters.com/webassets/cms/library/docs/wa30000.pdf (accessed February 8, 2021). [174] S.M. Heidari, A. Anctil, Identifying alternative solvents for C60 manufacturing using singular and combined toxicity assessments, J. Hazard. Mater. 393 (2020) 122337. doi:10.1016/j.jhazmat.2020.122337. [175] F. Cataldo, T. Da Ros, Medicinal Chemistry and Pharmacological Potential of Fullerenes 134 and Carbon Nanotubes, Springer Netherlands, Dordrecht, 2008. doi:10.1007/978-1-4020- 6845-4. [176] F. Cataldo, T. Braun, The Solubility of C 60 Fullerene in Long Chain Fatty Acids Esters, Fullerenes, Nanotub. Carbon Nanostructures. 15 (2007) 331–339. doi:10.1080/15363830701512450. [177] P. Anatol’evich, P. Petrovich, B. Borisovich, M. Ehrik, R. Mikhajlovich, S. Valentinovich, K. Ivanovich, A. Nikolaevich, Z. Viktorovich, Fullerene Solution Prepration Method; Russian Federation , RU 2283273 C2, 2006. [178] K.N. Semenov, N.A. Charykov, V.I. Namazbaev, N.I. Alekseyev, E.G. Gruzinskaya, V.N. Postnov, O.A. Krokhina, Temperature dependence of solubility of light fullerenes in some essential oils, Fullerenes Nanotub. Carbon Nanostructures. 19 (2011) 225–236. doi:10.1080/15363831003721765. [179] K.N. Semenov, N.A. Charykov, V.A. Keskinov, A.K. Piartman, A.A. Blokhin, A.A. Kopyrin, Solubility of Light Fullerenes in Organic Solvents, J. Chem. Eng. Data. 55 (2010) 13–36. doi:10.1021/je900296s. [180] A. Demirbas, Relationships derived from physical properties of vegetable oil and biodiesel fuels, Fuel. 87 (2008) 1743–1748. doi:10.1016/j.fuel.2007.08.007. [181] W. Luo, Y.S. Khoo, A. Kumar, J.S.C. Low, Y. Li, Y.S. Tan, Y. Wang, A.G. Aberle, S. Ramakrishna, A comparative life-cycle assessment of photovoltaic electricity generation in Singapore by multicrystalline silicon technologies, Sol. Energy Mater. Sol. Cells. 174 (2018) 157–162. doi:10.1016/j.solmat.2017.08.040. [182] Annual Energy Outlook 2019 with projections to 2050, 2019. doi:DOE/EIA-0383(2012) U.S. [183] A. Valero, A. Valero, G. Calvo, A. Ortego, Material bottlenecks in the future development of green technologies, Renew. Sustain. Energy Rev. 93 (2018) 178–200. doi:10.1016/j.rser.2018.05.041. [184] P.H. Brunner, H. Rechberger, Handbook of material flow analysis: for environmental, resource, and waste engineers, CRC press, 2016. [185] A. Adriaanse, S. Bringezu, A. Hammond, Y. Moriguchi, E. Rodenburg, D. Rogich, H. Schütz, Resource flows: the material basis of industrial economies, Harvard. (1997). [186] M.K. Hubbert, Energy resources: a report to the Committee on Natural Resources of the National Academy of Sciences, United States, 1962. [187] C. Brian, Hubbert Peak Theory - Oil: A Cultural and Geographic Encyclopedia of Black Gold, ABC-CLIO, Santa Barbara, CA, USA., 2014. [188] C.S. Tao, J. Jiang, M. Tao, Natural resource limitations to terawatt-scale solar cells, Sol. 135 Energy Mater. Sol. Cells. 95 (2011) 3176–3180. doi:10.1016/j.solmat.2011.06.013. [189] A. Zuser, H. Rechberger, Considerations of resource availability in technology development strategies: The case study of photovoltaics, Resour. Conserv. Recycl. 56 (2011) 56–65. doi:10.1016/j.resconrec.2011.09.004. [190] S. Davidsson, M. Höök, Material requirements and availability for multi-terawatt deployment of photovoltaics, Energy Policy. 108 (2017) 574–582. doi:10.1016/j.enpol.2017.06.028. [191] A. Elshkaki, T.E. Graedel, Dynamic analysis of the global metals flows and stocks in electricity generation technologies, J. Clean. Prod. 59 (2013) 260–273. doi:10.1016/j.jclepro.2013.07.003. [192] G. Dingemans, W.M.M. Kessels, Status and prospects of Al 2 O 3 -based surface passivation schemes for silicon solar cells, J. Vac. Sci. Technol. A Vacuum, Surfaces, Film. 30 (2012) 040802. doi:10.1116/1.4728205. [193] S. Guo, T.M. Walsh, M. Peters, Vertically mounted bifacial photovoltaic modules: A global analysis, Energy. 61 (2013) 447–454. doi:10.1016/j.energy.2013.08.040. [194] W. Chen, R. Liu, Q. Zeng, L. Zhou, Low cost multicrystalline bifacial PERC solar cells – Fabrication and thermal improvement, Sol. Energy. 184 (2019) 508–514. doi:10.1016/j.solener.2019.04.033. [195] A. Louwen, W. van Sark, R. Schropp, A. Faaij, A cost roadmap for silicon heterojunction solar cells, Sol. Energy Mater. Sol. Cells. 147 (2016) 295–314. doi:10.1016/j.solmat.2015.12.026. [196] S.E. Sofia, H. Wang, A. Bruno, J.L. Cruz-Campa, T. Buonassisi, I.M. Peters, Roadmap for cost-effective, commercially-viable perovskite silicon tandems for the current and future PV market, Sustain. Energy Fuels. 4 (2020) 852–862. doi:10.1039/c9se00948e. [197] United Nations Comtrade annual reports, International Trade Statistics Database, https://comtrade.un.org [198] U.S. Geological Survey (USGS) MINERAL COMMODITY SUMMARIES, https://www.usgs.gov/centers/nmic/mineral-commodity-summaries. [199] U.S. Geological Survey (USGS) MINERAL COMMODITY SUMMARIES 2019, Reston, Virginia, 2020. https://www.usgs.gov/centers/nmic/mineral-commodity-summaries. [200] ITRPV, International Technology Roadmap for Photovoltaic, 2020. https://itrpv.vdma.org/en/. [201] 2020 Photovoltaics Report_Fraunhofer Institute for Solar Energy Systems, ISE, 2020. www.ise.fraunhofer.de. 136 [202] A. Shahsavari, M. Akbari, Potential of solar energy in developing countries for reducing energy-related emissions, Renew. Sustain. Energy Rev. 90 (2018) 275–291. doi:10.1016/j.rser.2018.03.065. [203] G. Masson, I. Kaizuka, Trends in Photovoltaic Applications - Report IEA PVPS T1-36: 2019, 2019. http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:TRENDS+IN+PHOTO VOLTAIC+APPLICATIONS#0. [204] C. Ramírez-Márquez, M.V. Otero, J.A. Vázquez-Castillo, M. Martín, J.G. Segovia- Hernández, Process design and intensification for the production of solar grade silicon, J. Clean. Prod. 170 (2018) 1579–1593. doi:10.1016/j.jclepro.2017.09.126. [205] A. Grbeš, A Life Cycle Assessment of Silica Sand: Comparing the Beneficiation Processes, Sustainability. 8 (2015) 11. doi:10.3390/su8010011. [206] J. Mitterpach, E. Hroncová, J. Ladomerskỳ, K. Balco, Identification of significant impact of silicon foundry sands mining on LCIA, Sustain. 7 (2015) 16408–16421. doi:10.3390/su71215822. [207] Unimin (Covia) Corporation Annual report, 2019. https://www.coviacorp.com. [208] S.R. Wu, I. Celik, D. Apul, J. Chen, A social impact quantification framework for the resource extraction industry, Int. J. Life Cycle Assess. 24 (2019) 1898–1910. doi:10.1007/s11367-019-01605-x. [209] R. Sairinen, O. Sidorenko, H. Tiainen, A research framework for studying social impacts: Application to the field of mining, Environ. Impact Assess. Rev. 86 (2021) 106490. doi:10.1016/j.eiar.2020.106490. [210] S. a Joyce, M. Macfarlane, Social Impact Assessment in the Mining Industry: Current Situation and Future Directions, Mining, Miner. Sustain. Dev. (2001) 28. http://pubs.iied.org/pdfs/G01023.pdf. [211] A. Bisht, J.-F. Gerber, Ecological distribution conflicts (EDCs) over mineral extraction in India: An overview, Extr. Ind. Soc. 4 (2017) 548–563. doi:10.1016/j.exis.2017.03.008. [212] A. Swanson, C. Buckley, Chinese Solar Companies Tied to Use of Forced Labor, New York Times. (2021). https://www.nytimes.com/2021/01/08/business/economy/china-solar- companies-forced-labor-xinjiang.html (accessed April 9, 2021). [213] T.W. Pearson, Frac Sand Mining in Wisconsin: Understanding Emerging Conflicts and Community Organizing, Cult. Agric. Food Environ. 35 (2013) 30–40. doi:10.1111/cuag.12003. [214] H. Farahani, S. Bayazidi, Modeling the assessment of socio-economical and environmental impacts of sand mining on local communities: A case study of Villages Tatao River Bank in North-western part of Iran, Resour. Policy. 55 (2018) 87–95. 137 doi:10.1016/j.resourpol.2017.11.001. [215] T. Kim, Efficient management of marine resources in conflict: An empirical study of marine sand mining, Korea, J. Environ. Manage. 91 (2009) 78–86. doi:10.1016/j.jenvman.2009.07.006. [216] UNEP 2019. Sand and sustainability: Finding new solutions for environmental governance of global sand resources., GRID-Geneva, 2019. http://www.unepgrid.ch/. [217] United Nations Environment Programme (UNEP) Annual report 2013, 2013. http://wedocs.unep.org/bitstream/handle/20.500.11822/8607/-UNEP 2013 Annual Report- 2014UNEP AR 2013-LR.pdf?sequence=8&isAllowed=y. [218] High-grade silica sands project located adjucent to the world’s largest silica sand mine, Diatreme Resources - Company presentation, 2019. [219] Z.Y. Zhao, H.J. Yang, J. Zuo, Evolution of international trade for photovoltaic cells: A spatial structure study, Energy. 124 (2017) 435–446. doi:10.1016/j.energy.2017.02.093. [220] C.R. Hackney, S.E. Darby, D.R. Parsons, J. Leyland, J.L. Best, R. Aalto, A.P. Nicholas, R.C. Houseago, River bank instability from unsustainable sand mining in the lower Mekong River, Nat. Sustain. 3 (2020) 217–225. doi:10.1038/s41893-019-0455-3. [221] M. Lines, S. Resources, A. Echt, L. Services, Silica sand supply and demand in the Asia- Pacific glass market, (2004) 1–12. [222] K. Lumpur, Ministry of natural resources and environment department of irrigation and drainage malaysia, 2010. [223] M.J. van der Meulen, W.E. Westerhoff, A. Menkovic, S.H.L.L. Gruijters, C.W. Dubelaar, D. Maljers, Silica sand resources in the Netherlands, Netherlands J. Geosci. - Geol. En Mijnb. 88 (2009) 147–160. doi:10.1017/S001677460000086X. [224] C. Mitchell, Role of National Geological Surveys in evaluation of high-purity silica resources. [225] J. Allen, A. Voiland, Sand mining at Poyang lake, Nasa Earth Obs. (2016) 1. https://earthobservatory.nasa.gov/images/87663/sand-mining-at-poyang-lake (accessed July 22, 2020). [226] L. KOEHNKEN, Impacts of Sand Mining on Ecosystem Structure, Process & Biodiversity in Rivers, 2018. http://d2ouvy59p0dg6k.cloudfront.net/downloads/sand_mining_impacts_on_world_rivers __final_.pdf. [227] X. Lai, D. Shankman, C. Huber, H. Yesou, Q. Huang, J. Jiang, Sand mining and increasing Poyang Lake’s discharge ability: A reassessment of causes for lake decline in China, J. Hydrol. 519 (2014) 1698–1706. doi:10.1016/j.jhydrol.2014.09.058. 138 [228] T. Piman, M. Shrestha, Case study on sediment in the Mekong River Basin: Current state and future trends, 2017. www.sei-international.org. [229] T. Bide, P. Balson, J. Mankelow, I. Selby, A new sand and gravel map for the UK Continental Shelf to support sustainable planning, Resour. Policy. 48 (2016) 1–12. doi:10.1016/j.resourpol.2016.02.004. [230] D. Liu, J.C. Liu, H. Huang, K. Sun, Analysis of the international polysilicon trade network, Resour. Conserv. Recycl. 142 (2019) 122–130. doi:10.1016/j.resconrec.2018.11.025. [231] Study on the Impacts of Mainstream Hydropower on the Mekong River., Copenhagen, 2015. [232] Mineral Resources - Malaysian Minerals, http://malaysianminerals.com/index.php?option=com_content&task=view&id=21&Itemid =45 (accessed May 30, 2020). [233] K.I. Vatalis, G. Charalambides, N.P. Benetis, Market of High Purity Quartz Innovative Applications, Procedia Econ. Financ. 24 (2015) 734–742. doi:10.1016/S2212- 5671(15)00688-7. [234] J. Zhou, X. Yang, A Reflection on China’s high purity quartz industry and its strategic development, MOJ Min. Metall. 1 (2018) 85–87. doi:10.15406/mojmm.2018.01.00013. [235] D. JIA, Z. Wanyi, C. Conglin, L. Yongsheng, Global Resource Status and China ’ s Development Suggestions of High Purity Quartz, (2014). [236] Silicon Metal from China - Investigation No. 731-TA-472, U.S. International Trade Commission, 2018. [237] M.E. Benson, A.B. Wilson, Frac Sand in the United States—A Geological and Industry Overview, U.S. Department of the Interior - USGS, 2015. doi:http://dx.doi.org/10.3133/ofr20151107. [238] S. Hamidullah, M.T. Shah, M.A. Khan, Identifying and characterization of the quality of silica sand resources from Munda Gucha, district Mansehra, in glass making, Geo;. Bull. Univ. Peshawar. 29 (1996) 59–68. [239] M.F.M. dos Santos, E. Fujiwara, E.A. Schenkel, J. Enzweiler, C.K. Suzuki, Quartz resources in the Serra de Santa Helena formation, Brazil: A geochemical and technological study, J. South Am. Earth Sci. 56 (2014) 328–338. doi:10.1016/j.jsames.2014.09.017. [240] Indonesia’s Islands Are Buried Treasure for Gravel Pirates, New York Times. (2010). https://www.nytimes.com/2010/03/28/weekinreview/28grist.html (accessed July 23, 2020). [241] N. Meynen, Concrete, or beaches? World’s sand running out as global construction booms, Ecologist. (2017) 1–12. https://theecologist.org/2017/may/09/concrete-or-beaches-worlds- sand-running-out-global-construction-booms (accessed July 23, 2020). 139 [242] J. Berlinger, North Korea might be making millions -- and breaking sanctions - selling sand. Yes, sand., CNN. (2020). https://www.cnn.com/2020/06/09/business/north-korea-sand- intl-hnk/index.html (accessed July 22, 2020). [243] V. Beiser, Sand mining: the global environmental crisis you’ve never heard of, Guard. (2017) 2–5. https://www.theguardian.com/cities/2017/feb/27/sand-mining-global- environmental-crisis-never-heard. [244] P. Salopek, Inside the deadly world of India’s sand mining mafia, Natl. Geogr. Mag. (2019) 6–8. https://www.nationalgeographic.com/environment/2019/06/inside-india-sand-mining- mafia/ (accessed July 29, 2020). [245] How will India address illegal sand mining without any data?, Downtoearth.Org. (2017) 1– 11. https://www.downtoearth.org.in/news/mining/flouted-with-impunity-58736 (accessed July 29, 2020). [246] V. Beiser, India: The Deadly Global War for Sand, Pultiuzer Cent. (2015) 1–19. https://pulitzercenter.org/reporting/india-deadly-global-war-sand (accessed July 29, 2020). [247] N. Menon, Illegal Sand Mining: India’s Biggest Environmental Challenge?, Weather Channel. (2018) 1–7. https://weather.com/en-IN/india/news/news/2018-10-26-illegal-sand- mining-indias-biggest-environmental-challenge (accessed July 29, 2020). [248] The Deadly Global War for Sand, (2015) 1–22. https://www.wired.com/2015/03/illegal- sand-mining/ (accessed July 9, 2021). [249] SAD demands judicial probe into illegal sand mining in Punjab, Indiatimes News Agency. 1–2. https://timesofindia.indiatimes.com/city/chandigarh/sad-demands-judicial-probe-into- illegal-sand-mining-in-punjab/articleshow/67478832.cms (accessed July 23, 2020). [250] Status and Development Prospects of High Purity Quartz Technology in China, http://www.wanfangdata.com.cn/index.html (accessed March 31, 2020). [251] F. Pearce, The Hidden Environmental Toll of Mining the World’s Sand, Yale Sch. Environ. (2019) 1–8. https://e360.yale.edu/features/the-hidden-environmental-toll-of-mining-the- worlds-sand. [252] M. Bendixen, L.L. Iversen, I. Overeem, Greenland: Build an economy on sand, Science (80-. ). 358 (2017) 879. doi:10.1126/science.aar3388. [253] M. Bendixen, I. Overeem, M.T. Rosing, A.A. Bjørk, K.H. Kjær, A. Kroon, G. Zeitz, L.L. Iversen, Promises and perils of sand exploitation in Greenland, Nat. Sustain. 2 (2019) 98– 104. doi:10.1038/s41893-018-0218-6. [254] D. Ioannidou, G. Sonnemann, S. Suh, Do we have enough natural sand for low-carbon infrastructure?, J. Ind. Ecol. (2020) 1–12. doi:10.1111/jiec.13004. [255] E.K. Schnebele, U.S. Gelological Survay Minerals Yearbook - Silicon, U.S. Department of 140 the Interior (U.S. Geological Survey), 2020. [256] G. Wernet, C. Bauer, B. Steubing, J. Reinhard, E. Moreno-Ruiz, B. Weidema, The ecoinvent database version 3 (part I): overview and methodology, Int. J. Life Cycle Assess. 21 (2016) 1218–1230. doi:10.1007/s11367-016-1087-8. [257] Long Trail Sustainability. DATA SMART LCI Package. LTS. 2019., https://ltsexperts.com/services/software/datasmart-life-cycle-inventory/ (accessed July 28, 2020). [258] W. Shen, W. Han, T.J. Wallington, S.L. Winkler, China Electricity Generation Greenhouse Gas Emission Intensity in 2030: Implications for Electric Vehicles, Environ. Sci. Technol. 53 (2019) 6063–6072. doi:10.1021/acs.est.8b05264. [259] T. Dolley, Mineral Industry Surveys (USGS), 2004. http://minerals.usgs.gov/minerals. [260] Pinterest, U.S. SILICA HOLDINGS, INC., (2020) 1–16. https://www.ussilica.com/news/us-silica-holdings-inc-announces-fourth-quarter-and-full- year-2019-results (accessed May 7, 2021). [261] A. Müller, P.M. Ihlen, J.E. Wanvik, B. Flem, High-purity quartz mineralisation in kyanite quartzites, Norway, Miner. Depos. 42 (2007) 523–535. doi:10.1007/s00126-007-0124-8. [262] The Global Rise of Illegal Sand Mining, Arcgis.Com., https://www.arcgis.com/apps/Cascade/index.html?appid=0efa6b5c7cb147e4ba934756b7b 8c400 (accessed July 29, 2020). [263] E. Neugeboren, Illegal Sand Mining a Deadly Beat for India’s Journalists, (2020) 1–4. https://www.voanews.com/press-freedom/illegal-sand-mining-deadly-beat-indias- journalists (accessed July 29, 2020). [264] Online harassment and threats for Indian journalist exposing illegal sand mining, IFEX. (2017) 1–6. https://ifex.org/online-harassment-and-threats-for-indian-journalist-exposing- illegal-sand-mining/ (accessed July 29, 2020). [265] R. Romig, How to Steal a River, N. Y. Times Mag. (2017) 3–10. https://www.nytimes.com/2017/03/01/magazine/sand-mining-india-how-to-steal-a- river.html (accessed July 29, 2020). [266] P. Salopek, Inside the deadly world of India’s sand mining mafia, Natl. Geogr. Mag. (2019). https://www.nationalgeographic.com/environment/2019/06/inside-india-sand-mining- mafia/#close%0Ahttps://www.nationalgeographic.com/environment/2019/06/inside-india- sand-mining-mafia/. [267] F. Pearce, How a Proposed Strip Mine Brought Conflict to South Africa’s Wild Coast - Yale E360, YaleEnvironment360. (2017) 1–11. https://e360.yale.edu/features/titanium-mine- conflict-south-africa-pondoland-rhadebe-caruso. 141 [268] K. Liu, China Silicon Metal Overview and Prospect, in: Met. Bull. 16th Annu. Ferro-Alloys Conf. March 31–April 2, 2015. [269] W. Shen, W. Han, T.J. Wallington, S.L. Winkler, China Electricity Generation Greenhouse Gas Emission Intensity in 2030: Implications for Electric Vehicles Wei _ Supplementary Information, 1–35. [270] A. Anctil, C.W. Babbitt, R.P. Raffaelle, B.J. Landi, Material and Energy Intensity of Fullerene Production, Environ. Sci. Technol. 45 (2011) 2353–2359. doi:10.1021/es103860a. [271] N. Gillis, Reducing solar’s environmental impact through ecolabels, PV Magazine, (2020). https://www.pv-magazine.com/2020/09/17/reducing-solars-environmental-impact- through-ecolabels/ (accessed May 29, 2021). [272] D. Wagman, SEIA releases tool aimed at increasing solar supply chain transparency, PV Magazine, (2021). https://pv-magazine-usa.com/2021/04/29/seia-releases-tool-aimed-at- increasing-solar-supply-chain-transparency/ (accessed May 29, 2021). [273] S. Khan, A. Sugie, Sand Mining and Its Social Impacts on Local Society in Rural Bangladesh: A Case Study of a Village in Tangail District, J. Urban Reg. Stud. Contemp. India. 2 (2015) 1–11. [274] J.A. Musah, Assessment of Sociological and Ecological Impacts of Sand and Gravel Mining – a Case Study of East Gonja District ( Ghana ) and Gunnarsholt ( Iceland ), L. Restor. Train. Program. (2009) 75–108. [275] Y. Qin, Z. Chen, B. Ding, Z. Li, Impact of sand mining on the carbon sequestration and nitrogen removal ability of soil in the riparian area of Lijiang River, China, Environ. Pollut. 261 (2020) 114220. doi:10.1016/j.envpol.2020.114220. [276] M. Naveen Saviour, Environmental Impact of Soil and Sand Mining: A Review, Int. J. Sci. Environ. 1 (2012) 125–134. [277] L. Méndez, E. Forniés, D. Garrain, A.P. Vázquez, A. Souto, T. Vlasenko, Upgraded Metallurgical Grade Silicon and Polysilicon for solar electricity production: a comparative Life Cycle Assessment, Sci. Total Environ. (2021) 147969. doi:10.1016/j.scitotenv.2021.147969. [278] S.N. Joglekar, R.A. Kharkar, S.A. Mandavgane, B.D. Kulkarni, Process development of silica extraction from RHA: a cradle to gate environmental impact approach, Environ. Sci. Pollut. Res. 26 (2019) 492–500. doi:10.1007/s11356-018-3648-9. [279] U. Okereafor, M. Makhatha, L. Mekuto, V. Mavumengwana, Gold mine tailings: A potential source of silica sand for glass making, Minerals. 10 (2020). doi:10.3390/min10050448. [280] Physical Properties of Fats and Oils, 2006. 142 http://www.dgfett.de/material/physikalische_eigenschaften.pdf. [281] L.M. Diamante, T. Lan, Absolute Viscosities of Vegetable Oils at Different Temperatures and Shear Rate Range of 64 . 5 to 4835 s − 1, J. Food Process. 2014 (2014) 1–6. doi:10.1155/2014/234583. [282] X. Zhou, J. Liu, Z. Jin, Z. Gu, Y. Wu, Y. Sun, Solubility of Fullerene C 60 and C 70 in Toluene, o-Xylene and Carbon Disulfide at Various Temperatures, Fuller. Sci. Technol. 5 (1997) 285–290. doi:10.1080/15363839708011990. [283] The National Institute for Occupational Safety and Health (NIOSH)- O-xylene, https://www.cdc.gov/niosh/npg/npgd0668.html. [284] pubchem, U.S. Natl. Inst. Heal, https://pubchem.ncbi.nlm.nih.gov (accessed August 1, 2019). [285] The International Labor Organization- International Chemical Safety Cards (ICSC)- Tetralin, (2004). http://www.ilo.org/dyn/icsc/showcard.display?p_version=2&p_card_id=1527. [286] Human Metabolome Database (HMDB) - Toluene, (2019). http://www.hmdb.ca/metabolites/HMDB0034168. [287] Human Metabolome Database (HMDB)-1,2,4-Trimethylbenzene, (2019). http://www.hmdb.ca/metabolites/HMDB0013733. [288] E. Oberdörster, Manufactured nanomaterials (fullerenes, C60) induce oxidative stress in the brain of juvenile largemouth bass, Environ. Health Perspect. 112 (2004) 1058–1062. doi:10.1289/ehp.7021. [289] A. Dhawan, J.S. Taurozzi, A.K. Pandey, W. Shan, S.M. Miller, S.A. Hashsham, V. V. Tarabara, Stable colloidal dispersions of C60 fullerenes in water: Evidence for genotoxicity, Environ. Sci. Technol. 40 (2006) 7394–7401. doi:10.1021/es0609708. [290] Spectrum Chemical Mfg. Corp., Material Safety Data Sheet (Brassica oil), New jersey, 2007. https://www.spectrumchemical.com/MSDS/R2221.PDF. [291] SigmaAldrich, https://www.sigmaaldrich.com (accessed August 1, 2019). [292] Material Safety Data Sheet (Grapeseed Oil), 2015. http://www.soothingtouch.com/mm5/graphics/sds/SoothingTouch_307004-08.pdf. [293] Fisher Scientific, https://www.fishersci.com. [294] M. Farré, S. Pérez, K. Gajda-Schrantz, V. Osorio, L. Kantiani, A. Ginebreda, D. Barceló, First determination of C60 and C70 fullerenes and N-methylfulleropyrrolidine C60 on the suspended material of wastewater effluents by liquid chromatography hybrid quadrupole linear ion trap tandem mass spectrometry, J. Hydrol. 383 (2010) 44–51. 143 doi:10.1016/j.jhydrol.2009.08.016. [295] F. Gottschalk, T. Sun, B. Nowack, Environmental concentrations of engineered nanomaterials: Review of modeling and analytical studies, Environ. Pollut. 181 (2013) 287– 300. doi:10.1016/j.envpol.2013.06.003. [296] V.D.J. Keller, R.J. Williams, C. Lofthouse, A.C. Johnson, Worldwide estimation of river concentrations of any chemical originating from sewage-treatment plants using dilution factors, Environ. Toxicol. Chem. 33 (2014) 447–452. doi:10.1002/etc.2441. [297] K.S. Kwok, Y.C. Chan, K.M. Ng, C. Wibowo, Separation of fullerenes C60 and C70 using a crystallization-based process, AIChE J. 56 (2009) 1801–1812. doi:10.1002/aic.12105. [298] Y.S. Grushko, V.P. Sedov, V. a. Shilin, Technology for manufacture of pure fullerenes C60, C70 and a concentrate of higher fullerenes, Russ. J. Appl. Chem. 80 (2007) 448–455. doi:10.1134/S1070427207030196. [299] T.Y. Yan, Pseudocumene and mesitylene production and coproduction thereof with xylene, US5004854, 1987. [300] Swiss Centre For Life Cycle Inventories, Ecoinvent Database 3.3, Ecoinvent Cent. 2.0 (2016). [301] A. Thomas, B. Matthäus, H.-J. Fiebig, Fats and Fatty Oils, Ullmann’s Encycl. Ind. Chem. (2015) 1–84. doi:10.1002/14356007.a10_173.pub2. [302] P. M. Doran, CH11 Unit Operations, in: Bioprocess Eng. Princ. (Second Ed., 2013: pp. 445–595. doi:https://doi.org/10.1016/B978-0-12-220851-5.00016-2. [303] Catalog, Industrial scale mixer, Amixon, Germany. [304] Catalog part, Industrial scale oil Sealed Vacuum Pumps (TRIVAC), Leybold, Germany - P21, P25, & P29, 2016. [305] Industrial Ultrasonic Bath, https://www.toolots.com/ultrasonic-cleaner-66-6gal-2400w-28- 56khz-ce-certified-made-in-taiwan.html (accessed March 1, 2021). [306] Industrial scale sonicator (Q2000 Sonicator), https://homogenizers.net/products/q2000- sonicator (accessed March 12, 2021). [307] MSDS for DBU (1,8-Diazabicyclo[5.4.0]undec-7-ene), NFPA, (1910) 1–11. https://datasheets.scbt.com/sc-251609.pdf (accessed May 10, 2021). [308] Certificate of Analysis For Olive Oil, . https://www.sigmaaldrich.com/catalog/CertOfAnalysisPage.do?symbol=O1514&LotNo= BCCB9747&brandTest=SIGMA&returnUrl=%2Fproduct%2FSIGMA%2FO1514. [309] Petroleum & Other Liquids - U.S. Energy Information Administration (EIA). 144 [310] Energy Information Administration (EIA), Annual Coal Report 2019, U.S. Dep. Energy. (2020). [311] United States Natural Gas Industrial Price (Dollars per Thousand Cubic Feet). [312] J.H. Wong, M. Royapoor, C.W. Chan, Review of life cycle analyses and embodied energy requirements of single-crystalline and multi-crystalline silicon photovoltaic systems, Renew. Sustain. Energy Rev. 58 (2016) 608–618. doi:10.1016/j.rser.2015.12.241. [313] F. Kreith, P. Norton, D. Brown, A comparison of CO2 emissions from fossil and solar power plants in the United States, Energy. 15 (1990) 1181–1198. doi:10.1016/0360- 5442(90)90110-N. [314] L. Lu, H.X. Yang, Environmental payback time analysis of a roof-mounted building- integrated photovoltaic (BIPV) system in Hong Kong, Appl. Energy. 87 (2010) 3625–3631. doi:10.1016/j.apenergy.2010.06.011. [315] E.A. Alsema, M.J. de Wild-Scholten, Environmental impacts of crystalline silicon photovoltaic module production, Proc. 13th CIRP Int. Conf. Life Cycle Eng. LCE 2006. (2006) 103–108. doi:10.1557/proc-0895-g03-05. [316] M.J. de Wild-Scholten, Energy payback times of PV modules and systems, in: Work. Photovoltaik-Modultechnik, Koln. [317] M. Ito, K. Komoto, K. Kurokawa, Life-cycle analyses of very-large scale PV systems using six types of PV modules, Curr. Appl. Phys. 10 (2010) 271–273. doi:10.1016/j.cap.2009.11.028. [318] G. Hou, H. Sun, Z. Jiang, Z. Pan, Y. Wang, X. Zhang, Y. Zhao, Q. Yao, Life cycle assessment of grid-connected photovoltaic power generation from crystalline silicon solar modules in China, Appl. Energy. 164 (2016) 882–890. doi:10.1016/j.apenergy.2015.11.023. [319] E.. Alsema, M. de Wild Scholten, THE REAL ENVIRONMENTAL IMPACTS OF CRYSTALLINE SILICON PV MODULES: AN ANALYSIS BASED ON UP-TO-DATE MANUFACTURERS DATA E.A., in: Intergovernmental Panel on Climate Change (Ed.), 20th Eur. Photovolt. Sol. Energy Conf. Barcelona, 6-10 June 2005, Cambridge University Press, Cambridge, 2005. https://www.cambridge.org/core/product/identifier/CBO9781107415324A009/type/book_ part. [320] R. Dones, R. Frischknecht, Life-cycle Assessment of Photovoltaic Systems: Results of Swiss Studies on Energy Chains, Assessment. 125 (1998) 117–125. [321] W. Chen, J. Hong, X. Yuan, J. Liu, Environmental impact assessment of monocrystalline silicon solar photovoltaic cell production: A case study in China, J. Clean. Prod. 112 (2016) 1025–1032. doi:10.1016/j.jclepro.2015.08.024. 145 [322] K.E. Knapp, T.L. Jester, G.B. Mihaiik, Energy balances for photovoltaic modules: status and prospects, in: Conf. Rec. Twenty-Eighth IEEE Photovolt. Spec. Conf. - 2000 (Cat. No.00CH37036), IEEE, 1998: pp. 1450–1455. doi:10.1109/PVSC.2000.916166. [323] K. Kato, A. Murata, K. Sakuta, Energy pay‐back time and life‐cycle CO2 emission of residential PV power system with silicon PV module, Prog. Photovoltaics Res. Appl. 6 (1998) 105–115. doi:10.1002/(SICI)1099-159X(199803/04)6:2<105::AID- PIP212>3.3.CO;2-3. [324] E.. Alsema, E. Nieuwlaar, Energy viability of photovoltaic systems, Energy Policy. 28 (2000) 999–1010. doi:10.1016/S0301-4215(00)00087-2. [325] N. Jungbluth, Life cycle assessment of crystalline photovoltaics in the Swiss ecoinvent database, Prog. Photovoltaics Res. Appl. 13 (2005) 429–446. doi:10.1002/pip.614. [326] R. Kannan, K.C. Leong, R. Osman, H.K. Ho, C.P. Tso, Life cycle assessment study of solar PV systems: An example of a 2.7 kWp distributed solar PV system in Singapore, Sol. Energy. 80 (2006) 555–563. doi:10.1016/j.solener.2005.04.008. [327] V.M. Fthenakis, H.C. Kim, E. Alsema, Emissions from Photovoltaic Life Cycles, Environ. Sci. Technol. 42 (2008) 2168–2174. doi:10.1021/es071763q. [328] V.M. Fthenakis, R. Betita, M. Shields, R. Vinje, J. Blunden, Life cycle analysis of high- performance monocrystalline silicon photovoltaic systems: Energy payback times and net energy production value, EU PVSEC Proc. (2012) 4667–4672. doi:10.4229/27thEUPVSEC2012-6CV.4.14. [329] B. ju Kim, J. yong Lee, K. hwan Kim, T. Hur, Evaluation of the environmental performance of sc-Si and mc-Si PV systems in Korea, Sol. Energy. 99 (2014) 100–114. doi:10.1016/j.solener.2013.10.038. [330] N.A. Ludin, N.I. Mustafa, M.M. Hanafiah, M.A. Ibrahim, M. Asri Mat Teridi, S. Sepeai, A. Zaharim, K. Sopian, Prospects of life cycle assessment of renewable energy from solar photovoltaic technologies: A review, Renew. Sustain. Energy Rev. 96 (2018) 11–28. doi:10.1016/j.rser.2018.07.048. [331] M. Ito, K. Kato, K. Komoto, T. Kichimi, H. Sugihara, K. Kurokawa, An analysis of variation of very large-scale PV (VLS-PV) systems in the world deserts, in: Proc. 3rd World Conf. Photovolt. Energy Convers., 2003: pp. 2809–2814. [332] H. Hondo, Life cycle GHG emission analysis of power generation systems: Japanese case, Energy. 30 (2005) 2042–2056. doi:10.1016/j.energy.2004.07.020. [333] V. Fthenakis, E. Alsema, Photovoltaics energy payback times, greenhouse gas emissions and external costs: 2004–early 2005 status, Prog. Photovoltaics Res. Appl. 14 (2006) 275– 280. doi:10.1002/pip.706. [334] S. Pacca, D. Sivaraman, G.A. Keoleian, Parameters affecting the life cycle performance of 146 PV technologies and systems, Energy Policy. 35 (2007) 3316–3326. doi:10.1016/j.enpol.2006.10.003. [335] M. Raugei, S. Bargigli, S. Ulgiati, Life cycle assessment and energy pay-back time of advanced photovoltaic modules: CdTe and CIS compared to poly-Si, Energy. 32 (2007) 1310–1318. doi:10.1016/j.energy.2006.10.003. [336] A. Stoppato, Life cycle assessment of photovoltaic electricity generation, Energy. 33 (2008) 224–232. doi:10.1016/j.energy.2007.11.012. [337] R. Glöckner, M. De Wild-scholten, Energy payback time and carbon footprint of elkem solar silicon®, 27th Eur. Photovolt. Sol. Energy Conf. Exhib. (2012) 4661–4666. [338] M. Ito, K. Kato, K. Komoto, T. Kichimi, K. Kurokawa, A comparative study on cost and life-cycle analysis for 100 MW very large-scale PV (VLS-PV) systems in deserts using m- Si, a-Si, CdTe, and CIS modules, Prog. Photovoltaics Res. Appl. 16 (2008) 17–30. doi:10.1002/pip.770. [339] U. Desideri, S. Proietti, F. Zepparelli, P. Sdringola, S. Bini, Life Cycle Assessment of a ground-mounted 1778kWp photovoltaic plant and comparison with traditional energy production systems, Appl. Energy. 97 (2012) 930–943. doi:10.1016/j.apenergy.2012.01.055. [340] N. Stylos, C. Koroneos, Carbon footprint of polycrystalline photovoltaic systems, J. Clean. Prod. 64 (2014) 639–645. doi:10.1016/j.jclepro.2013.10.014. [341] D. Yue, F. You, S.B. Darling, Domestic and overseas manufacturing scenarios of silicon- based photovoltaics: Life cycle energy and environmental comparative analysis, Sol. Energy. 105 (2014) 669–678. doi:10.1016/j.solener.2014.04.008. [342] R. Dones, R. Frischknecht, Life-cycle assessment of photovoltaic systems: results of Swiss studies on energy chains, Prog. Photovoltaics Res. Appl. 6 (1998) 117–125. doi:10.1002/(SICI)1099-159X(199803/04)6:2<117::AID-PIP209>3.0.CO;2-M. [343] A. Stoppato, Life cycle assessment of photovoltaic electricity generation, Energy. 33 (2008) 224–232. doi:10.1016/j.energy.2007.11.012. [344] M. Ito, M. Kudo, M. Nagura, K. Kurokawa, A comparative study on life cycle analysis of 20 different PV modules installed at the Hokuto mega-solar plant, Prog. Photovoltaics Res. Appl. 19 (2011) 878–886. doi:10.1002/pip.1070. [345] G. Hou, H. Sun, Z. Jiang, Z. Pan, Y. Wang, X. Zhang, Y. Zhao, Q. Yao, Life cycle assessment of grid-connected photovoltaic power generation from crystalline silicon solar modules in China, Appl. Energy. 164 (2016) 882–890. doi:10.1016/j.apenergy.2015.11.023. [346] G.J.. Phylipsen, E.A. Alsema., Environmental life-cycle assessment of multicrystalline silicon solar cell modules (Report no. 95057), Utrecht University, 1995. 147 [347] R. Battisti, A. Corrado, Evaluation of technical improvements of photovoltaic systems through life cycle assessment methodology, Energy. 30 (2005) 952–967. doi:10.1016/j.energy.2004.07.011. [348] V.M. Fthenakis, R. Frischknecht, M. Raugei, H.C. Kim, E. Alsema, M. Held, M. de Wild Scholten, Methodology Guidelines on Life Cycle Assessment of Photovoltaic Electricity, 2011. [349] I. Miller, E. Gençer, H.S. Vogelbaum, P.R. Brown, S. Torkamani, F.M. O’Sullivan, Parametric modeling of life cycle greenhouse gas emissions from photovoltaic power, Appl. Energy. 238 (2019) 760–774. doi:10.1016/j.apenergy.2019.01.012. [350] K. Collins, B. Powell, A. Anctil, Life cycle assessment of silicon solar panels manufacturing in the United States, in: 2015 IEEE 42nd Photovolt. Spec. Conf., IEEE, 2015: pp. 1–4. doi:10.1109/PVSC.2015.7356393. [351] D. Yang, J. Liu, J. Yang, N. Ding, Life-cycle assessment of China’s multi-crystalline silicon photovoltaic modules considering international trade, J. Clean. Prod. 94 (2015) 35–45. doi:10.1016/j.jclepro.2015.02.003. [352] Z. Yu, W. Ma, K. Xie, G. Lv, Z. Chen, J. Wu, J. Yu, Life cycle assessment of grid-connected power generation from metallurgical route multi-crystalline silicon photovoltaic system in China, Appl. Energy. 185 (2017) 68–81. doi:10.1016/j.apenergy.2016.10.051. [353] M.J. De Wild-Scholten, E.A. Alsema, Environmental life cycle inventory of crystalline silicon photovoltaic module production, Mater. Res. Soc. Symp. Proc. 895 (2006) 59–71. doi:10.1557/proc-0895-g03-04. [354] Ecoinvent, https://www.ecoinvent.org (accessed July 20, 2020). [355] A.F. Sherwani, J.A. Usmani, Varun, Life cycle assessment of solar PV based electricity generation systems: A review, Renew. Sustain. Energy Rev. 14 (2010) 540–544. doi:10.1016/j.rser.2009.08.003. [356] N. Mathur, S. Singh, J.W. Sutherland, Promoting a circular economy in the solar photovoltaic industry using life cycle symbiosis, Resour. Conserv. Recycl. 155 (2020) 104649. doi:10.1016/j.resconrec.2019.104649. [357] S.K. Ghormley, Life Cycle Assessment in Foundry Sand Reclamation – Comparison of Secondary Reclamation Processes, 2017. http://digitalcommons.unl.edu/envengdiss. [358] E. Larsen, R.A. Kleiv, Flotation of quartz from quartz-feldspar mixtures by the HF method, Miner. Eng. 98 (2016) 49–51. doi:10.1016/j.mineng.2016.07.021. [359] H.U. Sverdrup, D. Koca, P. Schlyter, A Simple System Dynamics Model for the Global Production Rate of Sand, Gravel, Crushed Rock and Stone, Market Prices and Long-Term Supply Embedded into the WORLD6 Model, Biophys. Econ. Resour. Qual. 2 (2017) 8. doi:10.1007/s41247-017-0023-2. 148 [360] N. Jungbluth, M. Stucki, R. Frischknecht, M. Tuchschmid, G. Doka, M. Vollmer, Life-cycle inventories - Photovoltaics, Swiss Cent. Life Cycle Invent. Dübendorf, CH. (2009). [361] M. Keller, P. Koch, B. Heldstab, P. De Hahn, “Handbook emission factors for road transport.” UBA Berlin and BUWAL Bern, produced by INFRAS Bern (1999). Harvard. [362] M. Spielmann, C. Bauer, R. Dones, “Transport services: Ecoinvent report no. 14.” Swiss Centre for Life Cycle Inventories, Dübendorf (2007). Harvard. [363] K. Treyer, C. Bauer, Life cycle inventories of electricity generation and power supply in version 3 of the ecoinvent database—part II: electricity markets, Int. J. Life Cycle Assess. 21 (2016) 1255–1268. doi:10.1007/s11367-013-0694-x. [364] W. Shen, W. Han, T.J. Wallington, Current and future greenhouse gas emissions associated with electricity generation in China: Implications for electric vehicles, Environ. Sci. Technol. 48 (2014) 7069–7075. doi:10.1021/es500524e. [365] P. Wang, L.Y. Chen, J.P. Ge, W. Cai, W.Q. Chen, Incorporating critical material cycles into metal-energy nexus of China’s 2050 renewable transition, Appl. Energy. 253 (2019) 113612. doi:10.1016/j.apenergy.2019.113612. [366] C. Chabas, “Developing a life cycle impact assessment model based on functionality to assess competition between sand users” International Symposium on Sustainable Systems and Technology (ISSST) 2020, Virtual. 149