Advancing perovskite solar cells with biomass-derived solvents: a pathway to sustainability

Abstract

The conventional fabrication of perovskite solar cells (PSCs) relies on toxic solvents such as N,N-dimethylformamide and dimethyl sulfoxide (DMF and DMSO), which are harmful to the environment and health. This study introduces γ-valerolactone (GVL), a biomass-derived green solvent, as a sustainable alternative for perovskite precursor processing. By combining GVL with ethyl acetate (EA), a less toxic antisolvent, PSCs achieve a high efficiency of 23.74% without hazardous chemicals. Beyond laboratory-scale performance, we conducted a system-level evaluation integrating techno-economic analysis and life-cycle assessment to assess the manufacturing cost, environmental impact, and scalability. GVL/EA-based PSCs can lower manufacturing costs by 50% and reduce climate change impact by 80% compared to DMF/DMSO systems. Furthermore, global deployment scenarios identify break-even points—considering module lifetime and recycling strategies—where these green PSCs can effectively compete with other renewable energy sources including silicon photovoltaics. Overall, our findings highlight the potential of the GVL/EA solvent system to enable a safer, more sustainable, and economically viable route for commercialization.

---

Green foundation

1. This research replaces conventional hazardous solvents like DMF and DMSO in perovskite solar cell (PSC) fabrication with biomass-derived γ-valerolactone (GVL), thereby not only eliminating health risks but also boosting device performance associated with PSC manufacturing.

2. We establish an experimental basis for GVL-based PSCs that achieves power conversion efficiencies above 23%, demonstrating that an eco-friendly solvent can sustain industry-leading device metrics. This study provides a data-driven framework integrating response surface methodology, techno-economic assessment, and life-cycle assessment to optimize process variables for GVL-based PSC fabrication. The results highlight biomass-derived GVL as a key enabler for eco-friendly, scalable, and economically viable PSC technology, providing a roadmap for sustainable commercialization.

3. Future studies are necessary to explore solvent recycling techniques that further decrease the material footprint and drive towards a cradle-to-cradle lifecycle for PSC devices.

1 Introduction

Perovskite solar cells (PSCs) have garnered significant attention due to their high power conversion efficiency (PCE), surpassing that of silicon photovoltaics (Si-PV).¹ However, commercialization of PSCs is hindered by challenges in cost, sustainability, and long-term stability.² While past research has predominantly focused on improving device efficiency, a comprehensive evaluation that includes fabrication costs, environmental impact, and module lifetime is necessary to accurately assess their market potential and accelerate their industrial adoption.^3–5 A critical issue in the PSC fabrication is the reliance on toxic solvents, particularly N,N-dimethylformamide (DMF) and benzene-based solvents.⁶ These solvents not only contribute to direct human and environmental toxicity but also incur additional economic burdens associated with purification and waste management.^7,8 While recent efforts have been made to replace these harmful solvents with greener alternatives, there remains a lack of research on bio-derived solvents that offer a fully sustainable and non-toxic fabrication process.⁹ Addressing this gap is essential for the development of a truly green PSC technology.¹⁰ Recent studies have investigated the fabrication of PSCs using environmentally friendly solvents. However, the use of γ-butyrolactone (GBL) has indeed been widely explored as a biomass-derived solvent for perovskite fabrication; its use is increasingly constrained because it is a known precursor of the controlled psychoactive substance γ-hydroxybutyric acid (GHB), leading to strict usage restrictions or outright bans in several countries.¹¹ In contrast, γ-valerolactone (GVL) is a biomass-derived solvent with a comparable molecular structure but without the same legal restrictions.¹² Moreover, GVL has been reported to exhibit a lower overall toxicity burden than GBL, with composite green-solvent scores of 1.11 versus 2.01, respectively.¹³ This study explored the feasibility of eco-friendly PSC fabrication by comparing conventional toxic solvents (DMF/DMSO) with a green solvent system based on GVL. Here, GVL is derived from lignocellulosic biomass, making it a renewable and sustainable alternative to fossil fuel-based solvents. When forming perovskite thin films, it is essential to select co-solvents while considering environmental issues. In this regard, we compared diethyl ether (DEE) and ethyl acetate (EA) solvents to identify greener choices. These anti-solvents play a critical role in the large-area process, particularly in slot-die coating and roll-to-roll (R2R) fabrication, because they promote uniform film crystallization.^14–16

A schematic overview of the integrated research approach is presented in Fig. 1. Device fabrication followed a standard n–i–p architecture, beginning with SnO₂ deposition, KCl surface treatment, and subsequent perovskite layer formation, followed by spiro-OMeTAD and Au electrode deposition.¹⁷ The primary objective was to assess the effect of solvent choice on device efficiency, stability, and overall environmental impact (Fig. 1a). DMF exhibits an oral lethal dose 50% (LD50) of 2200 mg kg⁻¹ in rats and emits 2.59 kg CO₂ eq. per kg when produced, while DEE has an LD50 of 1215 mg kg⁻¹ and emits 6.39 kg CO₂ eq. per kg when produced.^18,19 Meanwhile, biomass-derived GVL has relatively lower toxicity, demonstrating a negative carbon emission value (−0.75 kg CO₂ eq. per kg).²⁰ These values show that GVL can be considered one of the alternative solvents. PSCs with the DMF/DMSO/DEE system achieved 23.97% PCE. Replacing DEE with EA slightly lowered efficiency to 23.08%, confirming EA as a viable green alternative. Using GVL with EA achieved 23.74% PCE, matching DMF/DMSO benchmarks. External quantum efficiency (EQE) spectra supported these results (Fig. 1a). Moreover, the GVL/EA-based system improved film stability, crucial for commercialization. Additionally, the GVL/EA module is optimized using the response surface method (RSM), with precursor concentration, annealing temperature, and annealing time as processing variables (Fig. 1b). The RSM models identified the optimal experimental conditions (precursor concentration: 1.4 M, annealing temperature: 130 °C, and annealing time: 15 min) that minimize the manufacturing cost and climate change impact. Under these optimal conditions, the PCE of PSCs reached 22.46 ± 0.78%, based on 14 real experimental data points. At the system level, the assessment was conducted based on a 1 m² module, estimating the required material quantities for fabrication (Fig. 1b). A techno-economic assessment (TEA) and life cycle assessment (LCA) were conducted specifically for the perovskite layer to analyze the impact of the GVL/EA-based system on economic feasibility and environmental sustainability. Compared to the DMF/DMSO/DEE perovskite, the GVL/EA-based perovskite reduced the manufacturing cost and climate change impact by approximately 50% and 80%, respectively. At the full-stack level, the GVL/EA perovskite layer accounted for only 0.8% of the total manufacturing cost and 0.5% of the total climate change impact. Through a global-scale analysis, module lifetime considerations were integrated to evaluate the economic and environmental viability of PSCs across different regions, identifying key break-even points where they become competitive with other renewable energy sources (Fig. 1c). The global average climate change impact and manufacturing cost of the GVL/EA module became lower than those of Si-PV at module lifetime thresholds of 7.5 years and 15 years, respectively. At this point, the climate change impact was 38.34 ± 1.32 g CO₂-eq. per kWh, while the manufacturing cost was 0.116 ± 0.004 US$ per kWh. With full recycling of Au (electrode) and FTO glass, the global average manufacturing cost threshold was shortened to just 1 year, with 0.086 ± 0.003 US$ per kWh, while the climate change impact remained higher with 194.3 ± 6.7 g CO₂-eq. per kWh, indicating that further advancements are necessary to achieve full environmental competitiveness. To incorporate regional characteristics, six countries with the highest installed solar cell capacities across continents were selected. In Australia, under a module lifetime of 4 years and an average PCE of 22.46%, the climate change impact was 33 g CO₂-eq. per kWh, and the manufacturing cost was 0.015 US$ per kWh, both lower than those of Si-PV (Fig. 1c).

Fig. 1 Overview of the comprehensive approach integrating lab-scale experiments, system-level modelling, and global deployment scenario analysis. (a) Schematic representation of lab-scale analysis on GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE perovskite solar cells. (b) System-level analysis of perovskite solar modules (GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE) through techno-economic assessment (TEA), life-cycle assessment (LCA), and the response surface method (RSM). (c) Global-wide analysis of manufacturing costs and climate change impact considering module lifetime, material recycling, and regional characteristics, with comparisons to various grid energy sources.

This study proposes an innovative approach for fabricating eco-friendly PSCs using bio-derived solvents, thereby eliminating reliance on hazardous chemicals. A comprehensive TEA and LCA were conducted considering production costs, CO₂ emissions, and material recycling. By integrating module lifetime considerations, we evaluated the economic and environmental viability of PSCs in different global regions and identified the key break-even points for competitiveness with other renewable energy sources. This research provides a framework for the sustainable and economically feasible commercialization of PSCs on a global scale.

2 Methods

2.1 Materials

Formamidinium iodide (FAI) and phenethylammonium iodide (PEAI) were purchased from GreatCell Solar. Lead iodide (PbI₂, 99.99%) was purchased from TCI. 2,2′,7,7′-Tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD) was purchased from Luminescence Technology Co. Dimethylformamide (DMF, 99.8%), dimethyl sulfoxide (DMSO, 99.9%), 2-propanol (IPA, 99.5%), diethyl ether (DEE), ethyl acetate (EA), potassium chloride (KCl), methylammonium chloride (MACl, 99.8%), γ-valerolactone (GVL, 99%), hydrochloric acid (HCl), thioglycolic acid, tin(II) chloride dihydrate (SnCl₂·2H₂O), lithium bis(trifluoromethane-sulfonyl)imide (Li-TFSI), and 4-tert-butylpyridine (TBP, 99.8%) were purchased from Sigma-Aldrich.

2.2 Device fabrication

Patterned FTO substrates were cleaned via sonication with detergent, de-ionized water, acetone, and IPA for 10 min each. The compact SnO₂ layer was deposited on the cleaned FTO using the chemical bath deposition (CBD) method. The solution for CBD was prepared by mixing urea (1.25 g), hydrochloric acid (35 wt% in H₂O, 1.25 mL), thioglycolic acid (25 μL), and SnCl₂·2H₂O (275 mg) per 100 mL of deionized water. FTO substrates were immersed in the prepared CBD solution and then left at an elevated temperature (90 °C) for 4 h. After the reaction, the FTO/SnO₂ substrates were cleaned via sonication with deionized water and IPA for 5 min each, followed by annealing at 170 °C for 1 h. The cleaned substrates were subjected to UV-ozone treatment for 15 min. The 4 mg mL⁻¹ KCl in deionized water was deposited on the UV-ozone treated substrates at 3000 rpm for 30 s and then annealed at 100 °C for 1 h. The KCl deposited substrates were subjected to UV-ozone treatment for 15 min. The perovskite precursor solution was prepared by dissolving FAI and PbI₂ in DMF : DMSO = 4 : 1 or GVL at a concentration of 1.4 M. For the perovskite film prepared from DMF/DMSO, the precursor solution was deposited on the substrates at 5000 rpm for 30 s. After 20 s of spinning, 1 mL of diethyl ether or 400 μL of ethyl acetate was dropped on top of the spinning substrates. For the perovskite film from GVL, the precursor solution was deposited on the substrates at 4000 rpm for 30 s. After 25 s of spinning, 400 μL of ethyl acetate was dropped on top of the spinning substrates. The perovskite deposited substrates were annealed at 130 °C for 15 min. The 5 mg mL⁻¹ PEAI dissolved in IPA was deposited on the perovskite layer at 5000 rpm for 30 s. The hole transporting layer (HTL) was spin-coated at 4000 rpm for 30 s on the substrates using a solution of 72.3 mg Spiro-OMeTAD, 28.8 μL TBP and 17.5 μL Li-TFSI stock solution (520 mg of Li-TFSI in 1 mL of acetonitrile) in 1 mL of CB. An Au metal contact with a thickness of 65 nm was deposited by thermal evaporation.

2.3 Characterization

The current density–voltage (J–V) curves were recorded in a glove box with a voltage source meter (Keithley 2450) under AM 1.5G illumination (100 mA cm⁻²) by using a 1 kW Oriel solar simulator (with respect to a reference silicon photodiode calibrated with NREL). The FE-SEM (Gemini500, Carl Zeiss) images were obtained at the Centre for University-wide Research Facilities (CURF) at Jeonbuk National University. Steady-state photoluminescence (PL, JASCO, 510 nm excitation), UV-vis spectroscopy (V-670, JASCO), X-ray diffraction (XRD) measurement, and EQE measurements were conducted at the Future Energy Convergence Core Center (FECC) at Jeonbuk National University.

2.4 Inventory data of GVL/EA module fabrication

Before conducting the TEA and LCA of the module, the required material quantities and utility demands for module fabrication were estimated. The material quantities required for module production are presented in ESI Table S1.† The active area ratio of the module was considered to be 70%.²¹ The mass of each material was determined based on the thickness of the corresponding layers, the active area ratio of the module, the concentrations of the respective solutions, and the material utilization efficiency.²¹ For perovskite PV module fabrication, the material utilization efficiency was assumed to be 30.0% for spin-coating²² and 82.0% for thermal evaporation.²³ The energy consumption for a 1 m² GVL/EA module is presented in ESI Table S2.† As shown, all processes are conducted using electrically powered equipment. Therefore, energy consumption was calculated by multiplying the power rating of each piece of equipment by its respective operating time.²¹

2.5 Response surface method

RSM is a multivariate statistical tool that provides a systematic approach for investigating the relationship between precursor concentration, annealing temperature, and annealing time in the perovskite layer deposition step, as well as their impact on PCE. RSM facilitates the development of predictive models, improving result reproducibility, and enables process optimization. In RSM, response surfaces are graphical representations used to explain the interaction effects between process variables and the resulting effects on the response.^24,25 The Central Composite Design (CCD) and Box–Behnken Design (BBD) are the two primary factorial approaches employed in RSM to analyze quadratic response surfaces and construct second-order polynomial models.²⁶ In this study, CCD was employed to evaluate the effects of the process parameters (precursor concentration, annealing temperature, and annealing time) on the response. The response function is highly dependent on the nature of the relationship between the response and the independent variables and is expressed through the second-order polynomial model in eqn (1):where Y is the predicted response, β₀ is the intercept or regression coefficient, β_i is the linear coefficient, β_ii is the quadratic coefficient, and β_ij is the interaction coefficient. X_i and X_jj are the coded values of the process variables, and E is the experimental/residual error. In this study, a CCD was developed to optimize three independent process variables: precursor concentration, annealing temperature, and annealing time. The experimental design was conducted over a precursor concentration range of 1.26–1.8 M, an annealing temperature range of 83–167 °C, and an annealing time range of 6–74 minutes. Each numerical factor was varied across three levels (low, medium, and high). Experiments were performed at a minimum of three levels for each factor to develop the quadratic model. ESI Table S3† presents the actual values along with their corresponding coded levels used in the experimental design. Furthermore, the validated model served as the foundation for developing the manufacturing cost and climate change impact models (eqn (2) and (3)). These models were employed to assess the influence of experimental conditions on both manufacturing cost and climate change impact. The PCE used in the manufacturing cost and climate change models was established at 20.0%, representing the highest PCE value among the PCE data used in RSM. The manufacturing cost and climate change impact model can be expressed as follows:where MC_R represents the cost per kWh to manufacture a perovskite solar module at RSM, CC_R represents the climate change impact per kWh to manufacture a perovskite solar module at RSM, X₁ is the precursor concentration, X₂ is the annealing temperature, X₃ is the annealing time, and η_cal is the calculated PCE.

2.6 Techno-economic assessment

To evaluate the economic feasibility of the GVL/EA module manufacturing process, the expected material consumption and costs for producing a 1 m² solar module were considered. This study focuses exclusively on material and utility costs, excluding labor costs, minor consumables, and equipment depreciation.²⁷ Given the commercialization potential of the GVL/EA module, material cost estimations were conducted for scaling up solar panel production by a factor of 10 000. Procurement quantities were estimated while maintaining the same material grade as used in the experiments to ensure consistency. A 10% learning rate was applied, and the cost estimation was determined using eqn (4).²⁸ For Au used in the electrode, due to its intrinsic market value, it is not practical to assume a cost reduction by half with every doubling of procurement quantities.²⁷ Detailed cost estimations are provided in ESI Table S11.† The cost estimation equation can be expressed as the following:where P_cal represents the material costs based on the consumption required for the production of 10 000 m² of solar panels, P_init is the material costs based on the consumption required for the production of 1 m² of solar panels, LR is the learning rate, V_req is the quantity of materials required for the production of 10 000 m² of solar panels, and V_pur is the purchase unit of reagents based on the laboratory scale. Additionally, the electricity price was set at 0.08 US$ per kWh for the calculations.²⁹ The manufacturing cost per 1 m² module was calculated by multiplying the unit cost derived from the above method using the required material quantities (ESI Table S12†). The cost per square meter (US$ per m²) of the module was converted to LCOE, which refers to cost per kilowatt-hour (US$ per kWh) over the lifetime of a solar module to assess its competitiveness with various technologies, as described in eqn (5):³⁰

E_T = η_ave × FF × H × A × PR × N (6)

where MC_L represents the cost per kWh to manufacture a perovskite solar module, MC_T represents the cost per m² to manufacture a perovskite solar module, E_T is the energy produced for module lifetime with average PCE, η_ave is the average power conversion efficiency of 14 data points (22.46%), FF is the fill factor (0.7),³¹H is the global average insolation (1510.22 kWh per m² per year),³²A is the perovskite module area (1 m²), PR is the performance ratio (0.84),³³ and N refers to the time period after which efficiency decreases to 80% (1 year).

2.7 Life-cycle assessment

The environmental assessment was conducted in accordance with ISO 14040/14044 standards,^34,35 with a LCA for perovskite module fabrication performed in four steps: (1) goal and scope definition, (2) inventory analysis, (3) impact assessment, and (4) interpretation.³⁶ This study evaluates the potential life-cycle impacts of three perovskite solar cell (PSC) modules (GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE) to assess their environmental feasibility. Environmental impacts were quantified using the ReCiPe 2016 method.³⁷ The functional unit was defined as 1 kWh of electricity produced by a perovskite solar module, and the system boundary was set as cradle-to-gate (Fig. 3). The life cycle consists of (1) component production and (2) module manufacturing stages, while module use and transportation were excluded from the system boundary.²¹ This assumption aligns with several previous LCA studies on PV technologies.^38,39 The balance of the systems, such as inverters and wiring, was also excluded at the system boundary to facilitate direct comparisons with other PV technologies.²¹ Inventory data for the three PSC fabrication systems were analyzed using data from Ecoinvent 3.7 and literature sources¹⁹ and are summarized in ESI Tables S1 and S13.† Four key environmental impact categories were considered in this study related to potential environmental issues of the PSC field,³⁶ specifically global warming potential (GWP), terrestrial ecotoxicity (TETP), freshwater ecotoxicity (FETP), and human toxicity potential (HTP). Definitions of each impact category are provided in previous studies.⁴⁰ Similar to the manufacturing cost estimation, the impact categories were initially calculated per square meter (m²) and subsequently converted to impact per kilowatt-hour (kWh) using eqn (7):where IC_L is the impact category (GWP, TETP, FETP, and HTP) per kWh for a perovskite solar module and IC_T is the impact category (GWP, TETP, FETP, and HTP) per m² for a perovskite solar module. To evaluate the sustainability of the three PSC (GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE) fabrication processes, four impact categories were compared. Additionally, the key environmental impact contributors in each perovskite solar module fabrication process were identified, providing further insights into the future of sustainable PSC manufacturing.

2.8 Global manufacturing cost and climate change impact of the GVL/EA module

The annual average solar irradiance used in this study was obtained from the Global Land Data Assimilation System (GLDAS), utilizing 2023 1° × 1° resolution data.⁴¹ The downward shortwave radiation (DSR) values from the GLDAS dataset were used to adjust the annual solar irradiance values in the manufacturing cost and climate change impact equations (eqn (8) and (9)), enabling the estimation of region-specific manufacturing cost and climate change impact values:

E_glo = η_ave × FF × DSR × CF × A × PR × N (10)

where MC_glo represents the cost per kWh to manufacture a perovskite solar module according to the region, MC_T represents the cost per m² to manufacture a perovskite solar module, CC_glo represents the climate change impact per kWh to manufacture a perovskite solar module considering regional variations, CC_T represents the climate change impact per m² to manufacture a perovskite solar module, DSR is the annual average downward shortwave radiation value according to the region (W per m² per year), and CF is the unit conversion factor (8.760 kWh W⁻¹). Additionally, to analyze the trends in material recycling, the calculations for MC_T and CC_T excluded the material costs (33.79, 67.58, and 295.58 US$ per m²) and climate change impact values (8.01, 16.02, and 16.73 kg CO₂-eq. per m²) associated with recycling scenarios of 50% Au, 100% Au, and FTO glass + Au, respectively.

3 Results and discussion

3.1 Feasibility and characterization of the GVL/EA-based system

The eco-friendliness of solvents was evaluated using multiple indices. DMF exhibited the highest health index (1.33), indicating significant risks (Fig. 2a).^10,13 In contrast, GVL recorded the lowest health index at 0.51. Additionally, DEE, a common anti-solvent, showed a high index (0.88), while EA, a greener alternative, had a safer value (0.65). From an environmental standpoint, DMF, DMSO, and DEE all presented high values exceeding 1, suggesting significant ecological concerns upon exposure or disposal. Conversely, GVL and EA exhibited notably lower values of 0.6 and 0.97, respectively, implying a reduced environmental footprint. Vapor pressure is crucial in solvent-based processing, impacting evaporation rates and worker exposure. DMF and DEE, with high vapor pressures, evaporate quickly, raising toxicity risks and volatile organic compound emissions.¹⁰ In contrast, GVL and EA have lower vapor pressures, reducing exposure and contamination, making them ideal for industrial use under strict safety and environmental regulations.⁴² Considering the cumulative health, environmental, and safety impacts, we propose replacing the conventional DMF/DMSO/DEE solvent system with a greener GVL/EA combination for perovskite fabrication.

Fig. 2 Properties of solvents in perovskite fabrication and characterization of perovskite films from DMF/DMSO- and GVL-based precursors. (a) Health index, environmental index, and vapor pressure of DMF, DMSO, DEE, GVL, and EA. (b)–(d) Comparison between perovskite films from DMF/DMSO- and GVL-based precursors. (b) Scanning electron microscopy (SEM) images, (c) X-ray diffraction (XRD) patterns, and (d) UV-vis absorbance and steady-state photoluminescence (PL) spectra of perovskite films. (e)–(k) Characterization of perovskite films from GVL-based precursors with various precursor concentrations, annealing times, and temperatures. (e) UV-vis absorbance spectra of perovskite films from 1.2 M, 1.5 M, and 1.8 M GVL-based precursors. (f) J_SC and PCE of perovskite solar cells from 1.2 M, 1.4 M, 1.6 M, and 1.8 M GVL-based precursors. (g) XRD patterns of perovskite films from the 1.4 M GVL-based precursor annealed at 150 °C for 5 min, 15 min, and 30 min. (h)–(k) SEM images of perovskite films from the 1.4 M GVL-based precursor annealed at 90 °C, 120 °C, 150 °C, and 180 °C for 15 min.

Furthermore, to bridge the transition from traditional to green processing, an intermediate assessment was conducted to replace DEE with EA and maintain DMF/DMSO. To verify the feasibility of the GVL/EA-based system, characterization studies of perovskite films from DMF/DMSO- and GVL-based precursors were compared. The perovskite films from DMF/DMSO- and GVL-based precursors have similar-sized and compact grains (Fig. 2b). All films had the α-FAPbI₃ structure, but PbI₂, which can have detrimental effects on stability, appeared in DMF/DMSO/EA films (Fig. 2c).⁴³ Additionally, DMF/DMSO- and GVL-based perovskite films have similar optical band gaps and photoluminescence (PL) intensities (Fig. 2d). These results suggest that GVL has the potential to replace DMF/DMSO in perovskite production. Additionally, we compared the PCE of devices fabricated with GBL and DMSO, which are environmentally friendly solvents used in the production of PSCs, with those made with GVL. The PCE was 20.9% for the GVL-based device, 17.0% for the GBL-based device, and 12.5% for the DMSO-based device. This indicates that the GBL and DMSO solvent systems have low potential as a perovskite precursor solvent compared to GVL (Fig. S4 and S6†).

To optimize the GVL/EA-based system, three conditions were controlled during the perovskite film fabrication step: precursor concentration, annealing time, and annealing temperature. The UV-vis absorbance of the perovskite film and short-circuit current density (J_SC) of the device increase with increasing precursor concentration (Fig. 2e and f). However, the PCE does not improve continuously because unwanted charge recombination may be induced as the thickness of the perovskite film increases.^44,45 The XRD patterns of perovskite films were obtained with various annealing times (Fig. 2g). This indicates that a shorter annealing time results in insufficient growth of perovskite crystal, while a longer annealing time causes the loss of FA⁺, an organic cation, which decreases the structural stability of α-FAPbI₃. As seen in the annealing temperature dependent SEM images, the perovskite film annealed at 90 °C has small perovskite grains (Fig. 2h). The perovskite film annealed at 120 °C shows larger grains than the perovskite film annealed at 90 °C. Uniformly sized grains and clear perovskite facets are detected in the film annealed at 150 °C (Fig. 2j).⁴⁶ The perovskite film annealed at 180 °C exhibited cracks and surface degradation due to high temperature.⁴⁷ Overall, these findings indicate that GVL/EA-based systems exhibit morphology, structure, and optical properties comparable to those made with traditional DMF/DMSO-based systems. Systematic optimization of GVL-based precursors revealed that high-quality perovskite layers can be reproducibly fabricated at slightly lower temperatures (<150 °C). This not only confirms the potential of GVL as a safer, environmentally friendly alternative but also suggests that the lower-temperature processing may reduce energy consumption and enhance the scalability of PSC production.

3.2 Manufacturing cost analysis of perovskite solar modules

An RSM model was developed using regression analysis based on experimental data to identify the key factors influencing PCE in the GVL/EA module, considering precursor concentration (1.26–1.8 M), annealing temperature (83–167 °C), and annealing time (6–74 min) as independent variables during perovskite layer deposition, with the model exhibiting a root mean square error (R²) value of 0.87 (ESI Tables S4 and S5†). To further validate the model, experiments were conducted under new conditions (1.28–1.51 M, 122–125 °C, and 36–40 min) suggested by the RSM model for PCE values (18, 19, and 20%). The results showed a minimal error (0.56–4.00%), demonstrating the model's reliability (ESI Table S6†). The validated RSM model was then extended to evaluate manufacturing cost and climate change impact, forming an extended-RSM model that incorporated PCE, TEA, and LCA (eqn (2) and (3), ESI Table S7†). The optimal conditions proposed by the extended-RSM models (1.4 M, 130 °C, and 15 min) were highly consistent with the experimentally determined values, verifying the model's predictive reliability. Under these conditions, an average PCE of 22.46 ± 0.78% was achieved, derived from 14 real experimental data points in the GVL/EA perovskite module.

The manufacturing cost of each perovskite solar module was estimated as the levelized cost of electricity (LCOE),³⁰ which represents the cost per kilowatt-hour (kWh) over the lifespan of a solar module and serves as a key metric for comparing different technologies. Based on reasonable assumptions (see TEA methods for details), the manufacturing costs were estimated to be 1.733, 1.741, and 1.756 US$ per kWh for modules utilizing GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE perovskite layers, respectively. The manufacturing cost per square meter for the GVL/EA module was 311.10 US$ per m², which corresponds to 1.733 US$ per kWh when incorporating the average PCE of 22.46% (Fig. 3a). The perovskite layer manufacturing costs were estimated at 0.0134 US$ per kWh, 0.0147 US$ per kWh, and 0.0292 US$ per kWh for the GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE modules, respectively (Fig. 3b). PbI₂ accounts for a significant portion of the perovskite layer cost, accounting for 53.3%, 48.8%, and 24.6% for the GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE perovskite layers, respectively. Among the module layers, the substrate layer incurs the highest expense, representing 73.01% of the total module cost, followed by the electrode layer at 21.64%. Interestingly, the perovskite layer itself represents only 0.77% of the total module cost for the GVL/EA module (Fig. 3c). Similar to the GVL/EA module, material costs (substrate and Au) dominate the overall costs, accounting for 98.64% and 98.65% of the total cost for the DMF/DMSO/EA and DMF/DMSO/DEE modules, respectively (Fig. 3d and e). As a result, in terms of manufacturing cost as LCOE, the GVL/EA module demonstrated the lowest cost among the three, followed by the DMF/DMSO/EA and DMF/DMSO/DEE modules (Fig. 3c–e). These findings confirm that replacing the conventional high-efficiency solar cell solvent DEE with the less toxic EA, as well as substituting commonly used solvents such as DMF and DMSO with the eco-friendly GVL, remains economically feasible.

Fig. 3 The estimated manufacturing cost and breakdown of each module and perovskite layer. (a)–(d) When calculating the manufacturing cost in US$ per kWh, the average value of 14 PCE data points obtained under the optimized experimental conditions of annealing temperature: 130 °C, annealing time: 15 min, and precursor concentration: 1.4 M during the perovskite layer deposition step is used. (a) Manufacturing cost and thickness of each layer component on the GVL/EA module. (b) Breakdown of the estimated cost of each perovskite layer in $ per kWh. (c)–(e) The cost breakdown of GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE modules; A-1 represents Au; H-1 represents Spiro-OMeTAD; H-4 represents chlorobenzene; El-1 represents electricity; P-1 represents FAI; P-2 represents PbI₂; P-8 represents DEE; E-1 represents SnCl₂-2H₂O; S-1 represents FTO/glass.

3.3 Perovskite solar module manufacturing LCA

The environmental impacts of the three perovskite layers and the GVL/EA module were evaluated (Fig. 4). The climate change impact of the GVL/EA module is the lowest (0.224 g CO₂-eq. per kWh), while the DMF/DMSO/EA and DMF/DMSO/DEE perovskite layers exhibit higher climate change impacts of 0.264 and 1.08 g CO₂-eq. per kWh, respectively (Fig. 4a). In terms of freshwater ecotoxicity, the DMF/DMSO/DEE, GVL/EA, and DMF/DMSO/EA perovskite layers exhibit values of 0.0185, 0.0039, and 0.0036 g 1,4-DCB-eq. per kWh, respectively (Fig. 4b) and for human toxicity, the GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE perovskite layers show impacts of 0.014, 0.0063, and 0.031 kg 1,4-DCB-eq. per kWh, respectively (Fig. 4c). All three perovskite layers exhibit similar values for terrestrial ecotoxicity, with 4.646, 4.671, and 4.672 g 1,4-DCB-eq. per kWh, in that order, where MACl is the primary contributor (Fig. 4d).

Fig. 4 LCA contribution analysis of each perovskite solar module. When calculating the impact categories in g CO₂-eq. per kWh and g 1,4-DCB-eq. per kWh, the average value of 14 PCE data points obtained under the optimized experimental conditions of annealing temperature: 130 °C, annealing time: 15 min, and precursor concentration: 1.4 M during the perovskite layer deposition step is used. (a)–(d) Contribution analysis on (a) climate change impact, (b) freshwater ecotoxicity, (c) human toxicity, and (d) terrestrial ecotoxicity for each perovskite layer. (e)–(h) Breakdown of estimated LCA results for (e) climate change impact, (f) freshwater ecotoxicity, (g) human toxicity, and (h) terrestrial ecotoxicity of the GVL/EA module. A-1 represents Au; H-1 represents Spiro-OMeTAD; H-4 represents chlorobenzene; El-1 represents electricity; P-2 represents PbI₂; P-4 represents GVL; P-5 represents EA; E-1 represents SnCl₂-2H₂O; S-1 represents FTO glass.

The climate change impact for modules incorporating GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE perovskite layers is estimated at 287.6, 288.8, and 289.6 g CO₂-eq. per kWh, respectively (Fig. 4e and Fig. S14†). The electricity required for GVL/EA module production accounts for 67.2% of its climate change impact, while Au used in the electrode contributes 32.8% (Fig. 4e). A similar contribution trend is observed in the DMF/DMSO/EA and DMF/DMSO/DEE modules. Moreover, Au used in the electrode (99.9% of the total) overwhelmingly dominates the freshwater ecotoxicity impact valued at 607.4 g 1,4-DCB-eq. per kWh (Fig. 4f). A similar contribution pattern of Au is observed for human toxicity (99.9% of the 6696 kg 1,4-DCB-eq. per kWh) and terrestrial ecotoxicity (94.5% of the 809.0 g 1,4-DCB-eq. per kWh) (Fig. 4g and h). DMF/DMSO/EA and DMF/DMSO/DEE exhibited similar trends, with detailed information provided in ESI Fig. S14.†

Overall, LCA analysis confirms that the GVL/EA module is environmentally more viable than the other two modules. Despite these improvements, further reductions in environmental impacts could be achieved by replacing Au with electrode materials that have lower environmental burdens, provided that high PCE values are maintained. These findings underscore the importance of integrating environmental considerations into the development of perovskite solar cell technologies, which have primarily focused on improving PCE, to promote sustainable, scalable solutions. This decision-making process is inherently complex and often involves trade-offs between efficiency and sustainability.

3.4 Comparative analysis of perovskite solar modules against various grid energy sources

To assess the comparative performance of the proposed GVL/EA module against different grid energy sources (fossil fuels, Si-PV, nuclear, and wind power), both climate change impact and manufacturing cost were evaluated by considering module lifetime and material recycling. Module lifetimes are defined as the break-even points at which the GVL/EA module's manufacturing cost or climate change impact falls below that of different grid energy sources; the one-year and four-year lifetimes were drawn from lifetime data in the literature.^48,49 The climate change impact of the comparative technologies is reported as 1001, 43, 13, and 13 g CO₂-eq. per kWh for fossil fuels, Si-PV, nuclear, and wind power, respectively.⁵⁰ The key stability threshold is 7.5 years, where the GVL/EA module exhibits a climate change impact comparable to Si-PV (38.34 ± 1.32 g CO₂-eq. per kWh) (Fig. 5a). To achieve a climate change impact comparable to nuclear and wind power (13 g CO₂-eq. per kWh), the GVL/EA module requires 25 years of module lifetime, at which point the estimated value is 11.50 ± 0.40 g CO₂-eq. per kWh.

Fig. 5 Comparative climate change impact and manufacturing cost of perovskite modules with various energy sources considering module lifetime and material recycling. (a)–(d) When calculating the climate change impact in g CO₂-eq. per kWh and the manufacturing cost in US$ per kWh, the average value of 14 PCE data points obtained under the optimized experimental conditions of annealing temperature: 130 °C, annealing time: 15 min, and precursor concentration: 1.4 M during the perovskite layer deposition step is used. (a) and (b) Box plot comparison of climate change impact across different perovskite modules (GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE) as a function of module lifetime (1–25 years) and material recycling (Au and FTO glass) under various grid energy sources (fossil fuel power plants, Si-PV, nuclear power plants, and wind power) with annotations of the 25% (q1), 50% (Med), 75% (q3) quantiles. (c) and (d) Box plot comparison of manufacturing costs across different perovskite modules (GVL/EA, DMF/DMSO/EA, and DMF/DMSO/DEE) as a function of module lifetime (1–25 years) and material recycling (Au and FTO glass) under various grid energy sources (fossil fuel power plants, Si-PV, nuclear power plants, and wind power) with annotations of the 25% (q1), 50% (Med), 75% (q3) quantiles.

The LCOE for nuclear, fossil fuels, Si-PV, and wind power is reported as 0.51, 0.30, 0.15, and 0.09 US$ per kWh, respectively.⁵¹ The first break-even point occurs at 4 years, at which the GVL/EA module achieves a manufacturing cost of 0.434 ± 0.015 US$ per kWh, closely aligning with the cost of nuclear power (0.51 US$ per kWh). To surpass Si-PV in cost competitiveness, a module lifetime of 15 years is required. At this point, the GVL/EA module achieves a cost of 0.115 ± 0.004 US$ per kWh, outperforming the Si-PV benchmark of 0.15 US$ per kWh (Fig. 5c). A module lifetime of 25 years is necessary to achieve cost parity with wind power. Under these conditions, the GVL/EA module attains a manufacturing cost of 0.069 ± 0.002 US$ per kWh, making it more cost-effective than wind energy. The overall comparative analysis reveals that the GVL/EA module requires at least 7.5 years of stability to achieve a lower climate change impact than the Si-PV and 15 years to become cost-competitive. However, this significantly exceeds the longest module lifetime of 4 years reported in the literature,⁵² highlighting the need for further advancements (Fig. 5c).

Implementing material recycling strategies could further reduce both the climate change impact and manufacturing costs.⁵³ Future research should focus on extending device longevity and optimizing recycling processes to improve the sustainability of perovskite solar technology. In this study, the effects of Au and FTO glass substrate recycling on climate change impact and manufacturing cost were evaluated (Fig. 5b and d). When 100% of Au and FTO glass were recycled, the climate change impact of the GVL/EA module was significantly reduced to 194.3 ± 6.7 g CO₂-eq. per kWh (Fig. 5b), which remains higher than that of Si-PV (43 g CO₂-eq. per kWh). In terms of manufacturing cost, when only Au was recycled at 100% efficiency, the GVL/EA module still exhibited a higher cost than nuclear power (0.51 US$ per kWh), with values ranging from 1.358 ± 0.047 US$ per kWh. However, when both Au and FTO glass were fully recycled, the GVL/EA module's manufacturing cost dropped below that of Si-PV, approaching values comparable to wind power (0.086 ± 0.003 US$ per kWh). These findings emphasize that extending module lifetime with material recycling, particularly of Au and FTO glass, plays a crucial role in reducing both climate change impact and manufacturing cost, further reinforcing the need for sustainable module fabrication strategies.

3.5 Global deployment scenarios analysis of the GVL/EA module

The impact of geographical deployment on the manufacturing cost and climate change impact of the GVL/EA module was evaluated, accounting for variations in module lifetime (1 and 4 years) and material recycling (Au and FTO glass) strategies under global solar irradiance data. Among the analyzed regions with a module lifetime of 1 year, Australia, which reported the lowest values, had a manufacturing cost of 1.182 US$ per kWh and a climate change impact of 195.9 g CO₂-eq. per kWh, reflecting a 2.45-fold difference compared to Germany, which exhibited the highest values (Fig. 6a). However, despite variations in solar resource availability, all evaluated regions exhibited higher values than conventional Si-PV modules (0.15 US$ per kWh and 43 g CO₂-eq. per kWh) when the module lifetime was limited to 1 year and 4 years (Fig. 6b). The impact of Au recycling was examined under a 1 year module lifetime (Fig. 6c and e), while the scenario where both Au and FTO glass are fully recycled was evaluated (Fig. 6g). Despite full recycling of both Au and FTO glass (Fig. 6g), Germany's manufacturing cost was reduced to 0.144 US$ per kWh, lower than that of Si-PV (0.15 US$ per kWh). However, the climate change impact remained high at 324.3 g CO₂-eq. per kWh, exceeding the 43 g CO₂-eq. per kWh of Si-PV. A similar pattern was observed in Australia, where manufacturing costs further declined to 0.059 US$ per kWh, yet the climate change impact remained at 132.3 g CO₂-eq. per kWh (Fig. 6g). These results indicate that extending the module lifetime is critical to achieving climate competitiveness with Si-PV.

Fig. 6 Global analysis of manufacturing costs and climate change impact considering module lifetime and material recycling. (a) and (b) Manufacturing cost and climate change impact for six countries (the United States, Germany, China, Brazil, Australia, and South Africa) without material recycling, under module lifetimes of 1 year and 4 years. (c) and (d) Manufacturing cost and climate change impact for six countries (the United States, Germany, China, Brazil, Australia, and South Africa) with 50% Au recycling, under module lifetimes of 1 year and 4 years. (e) and (f) Manufacturing cost and climate change impact for six countries (the United States, Germany, China, Brazil, Australia, and South Africa) with 100% Au recycling, under module lifetimes of 1 year and 4 years. (g) and (h) Manufacturing cost and climate change impact for six countries (the United States, Germany, China, Brazil, Australia, and South Africa) with full recycling of both Au and FTO glass, under module lifetimes of 1 year and 4 years.

The impact of Au and FTO glass recycling on manufacturing cost and climate change impact was evaluated under a 4 year module lifetime (Fig. 6d, f and h). In these four countries (Germany, the United States, China, and Brazil), both manufacturing costs and climate change impact remained higher than those of conventional Si-PV in the case of 50% recycling of Au. In contrast, in South Africa and Australia, the climate change impact fell below that of Si-PV, reaching 41.8 and 41.4 g CO₂-eq. per kWh, respectively, while the manufacturing cost remained higher than that of Si-PV, at 0.266 and 0.263 US$ per kWh (Fig. 6d). When 100% of Au was recycled, the manufacturing cost in Brazil decreased to 0.282 US$ per kWh, and the climate change impact dropped to 36.1 g CO₂-eq. per kWh, making it lower than that of Si-PV (Fig. 6f). However, in Germany, the United States, and China, the manufacturing cost remained at 0.567, 0.360, and 0.350 US$ per kWh, respectively, while the climate change impact was 82.7, 52.5, and 51.1 g CO₂-eq. per kWh, still exceeding that of Si-PV at 43 g CO₂-eq. per kWh. In Brazil, South Africa, and Australia, where both Au and FTO glass were fully recycled, the manufacturing cost further decreased to 0.016, 0.015, and 0.014 US$ per kWh, respectively, while the climate change impact was reduced to 35.4, 33.5, and 33.1 g CO₂-eq. per kWh (Fig. 6h). These results confirm that in these three countries, when the module lifetime is extended to four years and full material recycling is applied, both manufacturing cost and climate change impact can be lower than those of Si-PV. Upon full Au and FTO glass recycling (Fig. 6h), Brazil, South Africa, and Australia achieved both lower manufacturing costs and lower climate change impacts than Si-PV. However, in Germany, the United States, and China, despite achieving lower manufacturing costs, the climate change impact remained higher than Si-PV. These findings suggest that successful commercialization of the GVL/EA module requires not only improvements in module lifetime and material recycling but also consideration of regional solar irradiance conditions. While material recycling enhances cost competitiveness, reducing climate change impact necessitates further advancements in module lifetime and sustainable material strategies.

4 Conclusions

This study demonstrates that replacing hazardous solvents with biomass-derived GVL establishes a greener, more sustainable PSC fabrication process, reducing health and environmental risks by replacing toxic solvents such as DMF and DMSO. Incorporating EA as a low-toxicity anti-solvent further minimizes solvent hazards, highlighting the potential of the GVL/EA combination as a safer alternative. Systematic optimization of critical processing parameters—including precursor concentration, annealing temperature, and annealing time—revealed that the GVL/EA-based system can achieve efficiencies similar to the DMF/DMSO system and produce high-quality perovskites at slightly lower temperatures, reducing energy consumption in large-scale production. This framework facilitates the transition from laboratory-scale to commercial green manufacturing, laying a foundation for advancing sustainable perovskite solar cell technology.

Furthermore, by integrating RSM with TEA and LCA into extended-RSM models, we developed data-driven predictive models that optimize key process variables in perovskite layer deposition affecting the manufacturing cost, climate change impact, and device stability of PSCs, with optimal conditions (precursor concentration: 1.4 M, annealing temperature: 130 °C, and annealing time: 15 min) aligning closely with experimentally derived values, confirming the model's predictive reliability. Under optimal conditions, the GVL/EA PSCs achieve an average value of 22.46 ± 0.78%, based on 14 real experimental data points, and exhibit a 50% reduction in the manufacturing cost and an 80% reduction in climate change impact compared to DMF/DMSO-based systems, demonstrating the strong potential of GVL as a green solvent to improve the techno-economic feasibility and environmental sustainability of PSCs. Moreover, we systematically evaluated the manufacturing costs and climate change impact of GVL/EA-based perovskite solar modules across different module lifetimes, material recycling strategies, and global deployment scenarios. When extending the module lifetime beyond 4 years, even with material recycling, particularly of Au and FTO glass, the cost competitiveness and sustainability of the GVL/EA module improve significantly, making it competitive with Si-PV and other renewable energy sources, including nuclear and wind power. Our global-scale analysis shows that solar irradiance significantly affects regional competitiveness in both cost and environmental impact of perovskite modules, with GVL/EA modules outperforming Si-PV in high-irradiance regions (Australia, Brazil, South Africa) under a four-year lifetime with full recycling, while lower-irradiance regions (Germany, China) require further longevity and recycling improvements for competitiveness.

While these findings establish GVL/EA perovskite technology as a promising alternative, several challenges remain for large-scale commercialization. First, module lifetime remains a critical barrier, as a 4-year lifetime, even with material recycling, still results in a higher manufacturing cost and climate change impact than Si-PV. Second, future research should focus on alternative electrode materials to replace Au with options that have lower environmental and economic burdens. Third, the scalability of biomass-derived GVL production and supply chain feasibility must be investigated to support widespread industrial adoption. Fourth, integrating the GVL/EA process into existing R2R manufacturing workflows requires further optimization to ensure reproducible device performance on a commercial scale. Our findings emphasize that the GVL/EA-based perovskite solar technology, coupled with module lifetime improvements and recycling approaches, presents a viable roadmap for achieving cost-effective, environmentally sustainable perovskite photovoltaics on a global scale.

Author contributions

J Bae and J Cha: methodology, validation, formal analysis, investigation, data curation, visualization, and writing – original draft. M Kim and J Han: conceptualization, methodology, writing – review and editing, supervision, funding acquisition, and project administration.

Conflicts of interest

The authors declare no competing interests.

Data availability

The data underlying this article are available in the article and its ESI.† The code for RSM can be found at https://github.com/ijkll/ijj/releases/tag/2.0.0. These data are also available from the corresponding authors upon request.

Acknowledgements

This work was supported by the Program of Development of Ecofriendly Chemicals as Alternative Raw Materials to Oil through NRF funded by the Korean government (MSIT; no. 2022M3J5A1085257) and the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (no. RS-2024-00354152 and RS-2024-00409270).

References

1. Best Research-Cell Efficiency Chart, https://www.nrel.gov/pv/cell-efficiency.html, (accessed February 27, 2025).
2. Y. Rong, Y. Hu, A. Mei, H. Tan, M. I. Saidaminov, S. I. Seok, M. D. McGehee, E. H. Sargent and H. Han, Science, 2018, 361, eaat8235.
3. I. Mathews, S. Sofia, E. Ma, J. Jean, H. S. Laine, S. C. Siah, T. Buonassisi and I. M. Peters, Joule, 2020, 4, 822–839.
4. R. Vidal, J. Alberola-Borràs, N. Sánchez-Pantoja and I. Mora-Seró, Adv. Energy Sustainability Res., 2021, 2, 2000088.
5. L. Qiu, S. He, L. K. Ono, S. Liu and Y. Qi, ACS Energy Lett., 2019, 4, 2147–2167.
6. Authorisation – ECHA, https://echa.europa.eu/substances-of-very-high-concern-identification-explained, (accessed February 27, 2025).
7. R. Vidal, J.-A. Alberola-Borràs, S. N. Habisreutinger, J.-L. Gimeno-Molina, D. T. Moore, T. H. Schloemer, I. Mora-Seró, J. J. Berry and J. M. Luther, Nat. Sustain., 2021, 4, 277–285.
8. P. Dou, J. Song, S. Zhao, S. Xu, X. Li and T. He, Process Saf. Environ. Prot., 2019, 130, 317–325.
9. S. K. Podapangi, F. Jafarzadeh, S. Mattiello, T. B. Korukonda, A. Singh, L. Beverina and T. M. Brown, RSC Adv., 2023, 13, 18165–18206.
10. D. S. Lee, H. J. Lee, Y. Song, J. K. Park, J. H. Heo and S. H. Im, Chem. Commun., 2025, 61, 2011–2025.
11. N. Kwon, J. Lee, M. J. Ko, Y. Y. Kim and J. Seo, Nano Convergence, 2023, 10, 28.
12. C. Worsley, D. Raptis, S. Meroni, A. Doolin, R. Garcia-Rodriguez, M. Davies and T. Watson, Energy Technol., 2021, 9, 2100312.
13. B. J. Kim, H. Choi, S. Park, M. B. Johansson, G. Boschloo and M. Kim, ACS Sustainable Chem. Eng., 2024, 12, 13371–13381.
14. X. Li and S. D. Dimitrov, Sol. RRL, 2024, 8, 2400293.
15. Y. Y. Kim, T.-Y. Yang, R. Suhonen, A. Kemppainen, K. Hwang, N. J. Jeon and J. Seo, Nat. Commun., 2020, 11, 5146.
16. D. S. Ham, W. J. Choi, H. Yun, M. Kim, D.-H. Yeo, S. Lee, B. J. Kim and J. H. Lee, ACS Appl. Energy Mater., 2021, 4, 7611–7621.
17. H. Bipp and H. Kieczka, in Ullmann's Encyclopedia of Industrial Chemistry, John Wiley & Sons, Ltd, 2011.
18. Y. Shang, W. Meng, F. Lu, H. Li, Z. Yuan and N. Li, Adv. Materi., 2025, 2504602.
19. X. Tian, S. D. Stranks and F. You, Nat. Sustain., 2021, 4, 821–829.
20. C. Worsley, D. Raptis, S. Meroni, A. Doolin, R. Garcia-Rodriguez, M. Davies and T. Watson, Energy Technol., 2021, 9, 2100312.
21. J. Gong, S. B. Darling and F. You, Energy Environ. Sci., 2015, 8, 1953–1968.
22. R. García-Valverde, J. A. Cherni and A. Urbina, Prog. Photovoltaics, 2010, 18, 535–558.
23. A. A. Gentile, C. Rocco, S. Modeo and T. Romano, J. Cleaner Prod., 2014, 83, 473–482.
24. G. Ye, L. Ma, L. Li, J. Liu, S. Yuan and G. Huang, Int. J. Coal Prep. Util., 2020, 40, 131–145.
25. M. Kumari and S. K. Gupta, Sci. Rep., 2019, 9, 18339.
26. T. Damena and T. Alansi, J. Encapsulation Adsorpt. Sci., 2018, 8, 256–279.
27. Z. Wu, M. Sytnyk, J. Zhang, G. Babayeva, C. Kupfer, J. Hu, S. Arnold, J. Hauch, C. Brabec and I. M. Peters, Energy Environ. Sci., 2024, 17, 4248–4262.
28. N. L. Chang, A. W. Y. Ho-Baillie, D. Vak, M. Gao, M. A. Green and R. J. Egan, Sol. Energy Mater. Sol. Cells, 2018, 174, 314–324.
29. Electric Power Monthly – U.S. Energy Information Administration (EIA), https://www.eia.gov/electricity/monthly/epm_table_grapher.php, (accessed February 26, 2025).
30. Z. Li, Y. Zhao, X. Wang, Y. Sun, Z. Zhao, Y. Li, H. Zhou and Q. Chen, Joule, 2018, 2, 1559–1572.
31. H. D. Kim and H. Ohkita, Sol. RRL, 2017, 1, 1700027.
32. M. Rodell, P. R. Houser, U. E. A. Jambor, J. Gottschalck, K. Mitchell, C.-J. Meng, K. Arsenault, B. Cosgrove, J. Radakovich and M. Bosilovich, Bull. Am. Meteorol. Soc., 2004, 85, 381–394.
33. N. H. Reich, B. Mueller, A. Armbruster, W. G. J. H. M. Van Sark, K. Kiefer and C. Reise, Prog. Photovoltaics, 2012, 20, 717–726.
34. ISO 14044:2006(en), Environmental management—Life cycle assessment—Requirements and guidelines, https://www.iso.org/obp/ui/#iso:std:iso:14044:ed-1:v1:en, (accessed February 26, 2025).
35. ISO 14040:2006(en), Environmental management—Life cycle assessment—Principles and framework, https://www.iso.org/obp/ui/#iso:std:iso:14040:ed-2:v1:en, (accessed February 26, 2025).
36. H. Oh, J. Bae and J. Han, J. Cleaner Prod., 2025, 489, 144651.
37. M. A. J. Huijbregts, Z. J. N. Steinmann, P. M. F. Elshout, G. Stam, F. Verones, M. Vieira, M. Zijp, A. Hollander and R. Van Zelm, Int. J. Life Cycle Assess., 2017, 22, 138–147.
38. R. García-Valverde, J. A. Cherni and A. Urbina, Prog. Photovoltaics, 2010, 18, 535–558.
39. N. Espinosa, R. García-Valverde and F. C. Krebs, Energy Environ. Sci., 2011, 4, 1547–1557.
40. M. Ozturk and I. Dincer, Int. J. Hydrogen Energy, 2020, 45, 26536–26544.
41. H. Ziar, Joule, 2024, 8, 1667–1690.
42. D. Liu and A. Farooq, Combust. Flame, 2023, 253, 112771.
43. F. Liu, Q. Dong, M. K. Wong, A. B. Djurišić, A. Ng, Z. Ren, Q. Shen, C. Surya, W. K. Chan, J. Wang, A. M. C. Ng, C. Liao, H. Li, K. Shih, C. Wei, H. Su and J. Dai, Adv. Energy Mater., 2016, 6, 1502206.
44. J. Chen, L. Zuo, Y. Zhang, X. Lian, W. Fu, J. Yan, J. Li, G. Wu, C. Li and H. Chen, Adv. Energy Mater., 2018, 8, 1800438.
45. B. Zhang, M. Zhang, S. Pang, C. Huang, Z. Zhou, D. Wang, N. Wang and G. Cui, Adv. Mater. Interfaces, 2016, 3, 1600327.
46. C. Ma, F. T. Eickemeyer, S.-H. Lee, D.-H. Kang, S. J. Kwon, M. Grätzel and N.-G. Park, Science, 2023, 379, 173–178.
47. S. J. Yang, H. Jin, J. Cha, M. K. Kim, D. Baek, H. Na and M. Kim, Appl. Surf. Sci., 2023, 612, 155805.
48. M. Khenkin, H. Köbler, M. Remec, R. Roy, U. Erdil, J. Li, N. Phung, G. Adwan, G. Paramasivam and Q. Emery, Energy Environ. Sci., 2024, 17, 602–610.
49. X. Zhao, T. Liu, Q. C. Burlingame, T. Liu, R. Holley, G. Cheng, N. Yao, F. Gao and Y.-L. Loo, Science, 2022, 377, 307–310.
50. National Renewable Energy Lab. (NREL), Life Cycle Greenhouse Gas Emissions from Electricity Generation: Update, 2021, https://www.osti.gov/biblio/1820320.
51. C. Kost, J. N. Mayer, J. Thomsen, N. Hartmann, C. Senkpiel, S. Philipps, S. Nold, S. Lude, N. Saad and T. Schlegl, Levelized Cost of Electricity Renewable Energy Technologies, Fraunhofer Institute for Solar Energy Systems ISE, 2024.
52. X. Zhao, T. Liu, Q. C. Burlingame, T. Liu, R. Holley, G. Cheng, N. Yao, F. Gao and Y.-L. Loo, Science, 2022, 377, 307–310.
53. X. Xiao, N. Xu, X. Tian, T. Zhang, B. Wang, X. Wang, Y. Xian, C. Lu, X. Ou and Y. Yan, Nature, 2025, 1–6.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5gc02249e

‡ These authors contributed equally.

---