Towards market commercialization: Lifecycle economic and environmental evaluation of scalable perovskite solar cells

Many economic and environmental studies on novel perovskite solar cells (PSCs), published ex post the development stage to investigate the market competitiveness, have focused on laboratory‐scale PSC architectures that are not amenable for upscaling. In this paper, we evaluate the market potential and environmental sustainability of a scalable carbon‐electrode‐based PSC by benchmarking it to the market dominating c‐Si photovoltaics and CIGS thin film photovoltaics. The analysis covers the PSCs full lifecycle, at the module and system levels (residential and utility scale), and is based on realistic annual energy output data derived from energy yield calculations. We find that this PSC can produce electricity at low cost (3–6 €cents/kWh), with lowest energy payback (0.6–0.8 years) and greenhouse gas emissions (15–25g CO2 eq./kWh) compared with grid‐connected PV market alternatives, assuming 25years of lifetime, expected PV system cost reductions, and PSC module recycling and refurbishment.

The rapid increase in PCE from less than 4% to 25.5% in the last 10 years 4 has further induced economic interest in the perovskite technology, now considered one of the most promising technologies for next generation photovoltaics. 5 The successful commercialization of innovative PV technologies predominantly depends on the "solar cell golden triangle," which comprises three crucial performance indicators: Efficiency, lifetime, and cost; besides, also environmental sustainability and manufacturability have considerable importance. 6 Different perovskite solar cells (PSCs) configurations have been developed to address the technical and commercialization requirements.
Given the heterogeneity of PSCs configurations and production techniques, it is essential to assess the available options in technical, economic, and environmental terms to evaluate the PSCs commercial potential and identify critical factors to be optimized throughout the technology development. Few research groups have examined PSCs' economic and environmental aspects through cost analyses or techno-economic analysis (TEA) [7][8][9][10][11][12][13][14][15] and life-cycle assessments (LCAs). [16][17][18][19][20][21][22][23][24] In general, cost and environmental impact assessment studies have highlighted the potential of perovskite PV to be competitive with already-established PV technologies, such as c-Si and copper-indium-gallium-selenide (CIGS). In both types of studies, the results nevertheless vary widely depending on the type of configuration considered, the module components included, the chosen process sequence, and the methodological assumptions. A review of LCA literature on perovskite PV showed that the wide divergence of LCA studies' results is mainly due to the significant differences in process energy consumption. The resulting cumulative energy demand (CED) and the global warming potential (GWP) of single-junction perovskite PV have respectively been found to be in the range of 265-13,000 MJ/m 2 and 16-1700 kg CO 2 eq./m 2 . 25 Similarly, a wide range of values has been found for economic indicators, with module cost in the range of 0.17-0.73 US$/W and levelized cost of electricity (LCOE) in the range of 3.5-18.6 US$cents/kWh. [7][8][9][10][11][12][13] Existing assessments have two fundamental limitations. First, most studies on the economic and environmental performance have dealt with perovskite configurations that are not amenable for commercialization. Some studies have focused on the configurations that only meet the technical performance requirements (such as high PCE), and a consequence are composed of expensive materials and are using deposition techniques that are not suitable for industrial applications. 7,11 Consequently, the debate on the best PSC candidate configurations and its processing route for commercialization and deposition techniques is still ongoing. In other cases, the analyzed configuration was optimized exclusively for environmental or financial sustainability. [8][9][10] A recent study by Leccisi and Fthenakis analyzed the environmental impact of perovskite PVs by focusing on scalable configurations; The CED and GWP resulted in low values in the range 265-548 MJ/m 2 and 16-40 kg CO 2 eq./m 2 . 21 Nevertheless, this study focused on environmental aspects, and the cost aspects have been considered only qualitatively. The study by Rao et al. focused only on techno-economic aspects of another potential perovskite PV candidate for large-scale manufacturing. This study, however, resulted in higher module costs (0.53-0.9 US/W) than previously assessed perovskite configurations. 13 A more exhaustive evaluation of emerging PV technologies' large-scale market deployment potential should incorporate cost, efficiency, lifetime, scalability, and environmental performance.
Second, environmental and economic assessments in existing studies were carried out separately, which does not allow a comprehensive evaluation of the financial and environmental aspects of the technology developed and the identification, where present, of environmental and economic trade-offs.
In particular, scope and system boundary assumptions represent relevant limitations of the previous studies. Cost studies were limited to assessing the manufacturing and electricity generation costs (such as LCOE) during the use-phase of PV systems and neglected the endof-life. Similarly, few LCAs have considered the end-of-life implications. 26 Moreover, such analysis has been limited to the device-level, even when considering scalable architectures, such as the carbonbased one. 27 In particular, the performance of PSCs was assessed without including PV system analysis, thus excluding the eventual deployment of perovskite modules for electricity generation of PV systems. Moreover, the method employed to estimate energy generated by perovskite PVs among economic and environmental studies was inconsistent.
In this paper, we focus on a scalable perovskite single-junction PV configuration, produced on a large scale, and evaluate the economic and environmental sustainability by deriving a set of performance indicators. To this end we integrate TEA and LCA methodology, and include the end-of-life stage in the analysis. We present findings both at the PV module level and the PV system level and investigate the impact that recycling has on said outcomes. Finally, we benchmark the chosen PSC architecture with commercially available PV technologies such as c-Si (mono-Si) and CIGS to examine its large-scale production and commercialization potential. The analysis is supported by an energy yield (EY) model that simulates the annual power output under realistic irradiation conditions at different climatic conditions to reproduce the technology performance more accurately compared with estimates based only on local irradiation data.

| Device configuration, and environmental and economic indicators
We selected a monolithic PSC device with a carbon electrode. One of the most critical obstacles for the commercialization of PSCs is the long-term stability; PSCs should guarantee a stable electricity production for over a lifetime that is comparable to outdoor installations of Si PV modules. 28 In this regard, carbon-based perovskites have shown significant stability improvements compared with alternative PSC configurations. Replacing expensive metal electrodes such as gold with carbon has proven to be a successful strategy for achieving high PCEs and improving cell stability and duration. 29 The carbon-based PSC showed the longest stability measurements, around 9000-10,000 h under AM1.5 spectrum at 55 C. 30,31 A particular issue for bringing perovskite PV production to the market is the rapid degradation under reverse-bias conditions. 32,33 Recently, the carbon-based perovskite configuration has demonstrated long-term stability under reversebias-induced degradation and the tests on carbon-based perovskite modules (56.8 cm 2 aperture area) demonstrated prolonged outdoors endurance (IEC 61215-2:2016 standard test procedures). 34 Hence, this cell architecture can be considered a promising candidate of perovskite configuration that might satisfy the most relevant commercialization requirements of efficiency, cost, stability, and scalability. To evaluate the economic and environmental sustainability of the PSCs, we introduce a method for emerging PV technology assessments; we integrate the LCA with the TEA to quantify financial and environmental key indicators through the environmental-technoeconomic assessment (ETEA). With this method, we compute environmental and economic indicators based on the same system boundaries and functional unit. Then, we compare the indicators for the carbonbased perovskite PV to c-Si and CIGS and previous perovskite environmental and economic assessments. First, we evaluate the environmental and economic performance of the carbon-based perovskite at the PV module level by assuming a manufacturing plant based in Europe. Second, the environmental and economic competitiveness of perovskite devices is evaluated at a PV system level. The process environmental profile, the manufacturing costs (MCs) and minimum sustainable price (MSP) are provided at the module level. Moreover, energy demand and carbon footprint values are compared with previous perovskite PV studies as well as established PV technologies, such as c-Si and CIGS. All indicators are based on 1 m 2 , which is the functional unit of this study. Consequently, we assume that the produced modules are installed at optimal tilt in a PV system with the balance of system (BOS) components. Here, the indicators include LCOE, energy payback time (EPBT), and greenhouse gas emission factor (GEF). The assumed areas are 30 m 2 and 0.5 km 2 for the residential and utility scale, respectively. For the LCOE, we employ the MSP to compute the costs of the modules composing the PV system; besides, we F I G U R E 1 (A) Structure of monolithic perovskite solar cell with a carbon electrode; (B) system boundary of manufacturing a carbon-based perovskite PV module, PV system installation and use, and PV modules recycling/refurbishment consider the current (2020) and future (2030) system cost scenario, divided into power dependent, area dependent costs and soft costs.
We compute the economic and environmental indicators at residential and utility-scale PV systems over 25 years of lifetime. In this case, we assume that the selected PSC technology will further improve its stability by achieving a lifetime comparable to competing technology. Nevertheless, we assume that perovskite modules will degrade faster than traditional PVs, as described in Section 2.4.
Regarding the BOS, the inverter is assumed to have a lifetime of 15 years. All the other components are considered to have the same life expectancy as PV modules. These assumptions follow the methodology guidelines on LCA of PV. 35 EY calculations are performed considering outdoor conditions of the device under study, and the results (Table 1) are used as input for the indicators.
The recycling and refurbishment of the perovskite PV modules are also included, and the effect of this process on the module and system-level indicators is studied. The PSC architecture considered in this study was proven to be easily recyclable once end-of-life is reached. The recycling process was demonstrated on laboratory scale devices where the main composing layers were processed and raw precursor materials could be successfully obtained from used solar cells. These materials were then used to produce new PSCs without compromising performance. 36 We assume the perovskite PV manufacturer implements such a refurbishment process, and we evaluate the economic and environmental indicators for different levels of recovery rate and performance of the refurbished PV modules. and Seattle (temperate oceanic). The EY results are given in Table 1 for the three selected locations, and the performance of the carbonbased perovskite device is compared with the c-Si and CIGS performance.

| PV module-level indicators
We construct a bottom-up ETEA model based on the perovskite device manufacturing shown in Figure 1. The environmental performance of the carbon-based perovskite modules is evaluated by assessing the environmental impacts of the inventory input. In this study, the impact assessment is conducted using the Environmental Footprint (EF) 3.0 (adapted). This method is the one recommended by the European Commission and uses the ILCD recommended method as the default basis for the EF method. 41 43 The database used for this study is ecoinvent 3.8 within T A B L E 1 EY results for three selected locations (desert, oceanic, tropical) and related residential and utility-scale system capacities of carbonbased perovskite, c-Si, and CIGS The manufacturing phase is modeled using this method, and consequently, the environmental profile of the processing steps is provided.
Furthermore, regarding the environmental analysis, the values of energy requirements (MJ/m 2 ) and carbon footprint (kg CO 2 /m 2 ) of the carbon-based perovskite module are derived and benchmarked with previous LCA studies results on perovskite, CIGS and c-Si.
The economic performance of the carbon-based perovskite modules is first assessed by calculating the cost per unit area (€/m 2 ), defined as MC, and it includes the sum of the costs incurred in each processing as in Equation (1).
Here, the MC includes cost per unit area for material (M i ), equip-

| PV-System level analysis
The environmental techno-economic performance of carbon-based perovskite PV is also evaluated at the system level, assuming the deployment of perovskite modules for PV systems. We analyze two different PV system scales, residential and utility, to reflect the differences in costs, energy requirements, and GHG emissions. We then compare the environmental-techno-economic performance of the perovskite PV systems with traditional alternatives such as c-Si and CIGS. We assume a constant area of 30 m 2 for residential-scale systems and 0.5 km 2 for utility-scale systems to perform the calculations.
Moreover, EY calculations for the three PV technologies estimate the energy generated at three different climatic locations, which allows a realistic computation of the ETEA indicators at the system level.
Besides, we include losses due to inverter and wiring, assuming these account for 10% reduction in energy generation.
The PV systems' environmental performance is assessed by computing two indicators: EPBT and GEF. The EPBT indicates the time required for the system assessed to generate the same quantity of energy needed to produce the system itself. Hence, for PV systems, it is defined as the ratio between the primary energy demand and the annual electricity generated by the system (E agen ). 46 Besides, the primary energy demand for manufacturing the materials and the PV modules (E mod ), the energy demand for BOS (including inverter) (E bos ), and operation and maintenance (E O&M ) are considered.
Another relevant environmental sustainability indicator is the GEF, which estimates the lifecycle GHG emissions (g CO 2 equivalent) per kWh of electricity generated (E i ) by the PV system throughout its lifetime (N). d is a mutual parameter for the system indicators and represents the annual degradation rate. We consider that c-Si and CIGS PV systems degrade at the same rate annually by 0.2%, as reported for currently installed PV systems 47 and assume that carbonperovskite PVs degrade faster at a rate of 0.50% per year.
To evaluate the potential cost of generating electricity through a PV system of carbon-based perovskite PV modules, we compute the LCOE. The LCOE is generally defined as the ratio between the cost of a PV system throughout its lifetime to the total energy that that system can generate during the lifespan, as in Equation (5) 48 : where C i represents the system cost (€) and E i the electricity generated in the ith year, D is the discount rate, and N is the total lifetime of the PV system.
A mutual parameter for system-level indicators is the system lifetime. We consider that the PV system lifespan is equal for the three technologies (25 years), so we assume that the perovskite technology will be able to overcome the stability and duration issues that currently represent a significant barrier towards commercialization. Furthermore, besides the modules, it is critical to consider PV systems elements such as inverters and BOS components and the yearly operation and maintenance (O&M) activities that contribute to the energy requirements, GHG emissions, and cost, thus having a considerable effect on the EPBT, GEF, and LCOE indicators. The energy requirements and GHG emissions data for inverters and BOS are extracted from the available EcoInvent database. As for the cost, we assume power, area dependent, and soft system costs. As our focus is a European PV system application, we adopt the up-to-date estimates and assume system costs reduction by 2030. The system costs considered in this study represent the average among 21 EU countries. 49

| End-of-life analysis
The PSC architecture addressed in this study was proven to be easily recyclable once end-of-life is reached. The recycling process was demonstrated on laboratory scale devices where the main composing layers were processed, and raw precursor materials could be successfully obtained from degraded solar cells. These materials were then used to produce new PSCs without compromising performance. 36 Ideally, by implementing this recycling procedure, refurbished solar PV devices can be produced by employing raw materials extracted and re-processed from used PV modules. In that case, the operational costs related to materials, environmental impacts, and energy demand may be considerably affected.
For this reason, we address the end-of-life phase and, in particular, the effects of the implementation of a technically feasible recycling process. We then show the impact of perovskite PV modules end-of-life processing on the economic and environmental indicators.
We assume that the PV manufacturer is also responsible for collecting the dismissed PV modules. Recycling allows the manufacturer to recover materials that can be used for the production of successive modules. In this regard, we define the materials and process-related costs and environmental impacts of recycling and then consider the benefits of the recycling process in terms of avoided material purchase and manufacturing steps to calculate the economic and environmental indicators. We include the collection costs of dismissed PV modules which account for logistic operations to transport the PV waste to the recycling facility. 52,53 Hashmi et al. 36 claimed that the refurbishment procedure for carbon-based perovskite PVs could fully recover the materials without compromising performance. However, we adopt a more conservative approach in modeling the recycling performance. Besides the best-case scenario of full recovery (100%), our analysis also considers three levels of recovery for the materials that can be recycled: a low (50%), middle (70%), and high-rate (90%).
Moreover, we consider three cases with the resulting refurbished PV modules that do not retain the technical performance of new modules. In this sense, we consider a high (À30%), medium (À20%), and low (À10%) reduction in the technical performance of the refurbished PV modules. We assume the PCE to represent the performance, and we then consider its relative decrease when computing the MSP. For the system level indicators, thus for LCOE, EPBT, and GEF, the factor influencing the performance is the EY, and its relative reduction is assumed. We compute the relevant indicators for each case and analyze the recycling effects on the perovskite PV environmental and techno-economic performance. However, not all recycled material components can be employed to produce new modules; for these components, we assume that once recycled, they are sold in the market for different applications at a reduced price compared with their initial purchase. The details of the recycling process modeled can be found in the Supporting Information. Equipment and maintenance costs do not significantly impact total module MC, which account for 7% and 3%, respectively. In almost all impact categories, the glass presents the highest environmental burden, accounting for approximately 50-60% in the majority. The substrate influence is also evident from an economic perspective because the dominant input affects the materials cost, and thus, the manufacturing step involving the glass processing is the most costly.

| PV module-level indicators
The silver contacts are primarily responsible for the impact on resource use (minerals and metals). Considerable influence comes from the junction box and cabling of the module in several impact categories (such as human toxicity and ecotoxicity-freshwater). Regarding the other layers, carbon electrode deposition has the highest impact on ozone depletion. The use of lead is seen as an environmental concern to be considered during the commercialization of PSCs because of its harmful effects on the human body. 54  Moreover, it is reasonable to assume that the fast and lowtemperature solution deposition techniques could allow higher production throughput and thus contribute to further reducing the MC and MSP. These processing techniques also ensure low process electricity consumption, resulting in low energy-related-costs and emissions and thus facilitating large-scale production.

| PV system-level indicators
We extend the analysis to the system-level and estimate the relevant indicators for the eventual deployment of this technology for residential and utility-scale electricity production over 25 years of lifetime. Similarly, the GEF of the carbon-based perovskite produces the lowest value of g CO 2 eq. for kWh of electricity generated in contrast F I G U R E 3 EPBT (bars) and GEF (scatter) results for carbon-based perovskite, c-Si, and CIGS PV at three climatic locations for residential scale and utility scale systems to the c-Si and CIGS PV systems at the residential and utility scales.
This is due mainly due to the lower climate change impact of 1 m 2 of the perovskite module (28 kg CO 2 eq.) compared with the CIGS module (94 kg CO 2 eq.) and the c-Si module (143 kg CO 2 eq.). Although the perovskite device produces considerably lower EY, the larger decrease in emissions counterbalances the poorer electricity generation per unit area compared with the other two PV technologies. The significant reduction seen for the EPBT is more pronounced for the GEF because, on average, the perovskite systems have 23% and 19% lower GEF than c-Si and CIGS, respectively. These results show the considerable advantage for electricity generation in terms of energy demand and carbon emissions of carbon-based perovskite PVs.
The economic feasibility at the system level is measured through the LCOE indicator. This is calculated at current and future forecast system cost (2030), as in Figure 4. For 2030, we assume a reduction in selling prices of c-Si and CIGS PV modules due to learning effects.
On the other hand, we do not consider price reductions for the perovskite PV modules because this technology is not yet mature, and it is reasonable to expect a few years before it could be deployed on a large scale. printing, slot-die coating, blade coating, inkjet printing) that would provide higher throughput values than this study's conservative assumption (industry expert validated), and thus lower module prices.
The influence of lifetime and EY on the LCOE and GEF indicators is illustrated in Figure 5 for residential and utility scale systems aver-   We then consider the total effect of relative PCE improvements, EY increase, and at least a 90% recovery rate of the recycling process for the system-level indicators (Figure 7). We assume a relative PCE increase of 20% (≈17.3%), a relative EY increase of 20%, and an end- PVs. A smaller reduction is shown for the GEF, which could decrease below 20 g CO 2 eq./kWh at utility-scale (approximately 24% reduction) and thus perform considerably better than c-Si (29 g CO 2 eq./ kWh) and CIGS (26 g CO 2 eq./kWh).

| CONCLUSIONS
This study quantified the economic and environmental performance of a carbon-electrode-based perovskite PV configuration throughout its full lifecycle ex ante. This configuration is considered promising as it has shown encouraging results for both efficiency and stability in laboratory-scale devices. In addition, it is composed of low-cost materials that can be deposited by employing fast and low-energy demanding manufacturing techniques. Hence, it is a suitable candidate to evaluate the competitiveness of large-scale perovskite manufacturing (100 MW). The analysis focused on quantifying economic and environmental indicators at module and system level, while for the latter assuming a lifetime of 25 years.
With regards to economic indicators, at module level, the carbonbased perovskite configuration was found to offer promising results because its MSP was estimated to be 0.27 €/Wp. This is in line with previously assessed perovskite PV configurations that are not amenable to upscaling. It indicates that carbon-based perovskite PV modules can be sold at a comparable price of c-Si and CIGS modules currently available in the market (0.20-40 €/Wp). Nevertheless, the LCOE computed at system-level and across three climatic zones revealed that perovskite PVs generate electricity at on average a 7 and 20% higher cost compared with c-Si and CIGS both at residential and utility-scale given current system costs. This is mainly the due to the higher rated PCEs of CIGS and c-Si, and thus better yield, which increases the system's output and consequently decreases the overall system cost. In locations with higher insolation (e.g., desert) perovskite's LCOE was closest to that of a CIGS system. This gap between LCOEs across PV technologies would further increase by projecting the LCOE values to 2030 by considering reduced module and system costs. Yet, we demonstrate that the successful implementation of recycling processes, as well as technical improvements (e.g., PCE and EY), can close this gap by generating an LCOE in the range of 3.5-5.5 €cents/kWh, depending on the system scale. This implies that this perovskite architecture can generate electricity at a comparable price to CIGS (3-5.5 €cents/kWh) given further system cost reductions and learning effects that will significantly reduce the LCOE of conventional technologies. With regards to environmental indicators, both at module and system-level, a comparative advantage of carbon-based perovskite PVs was established.
The EPBT and GEF were on average across the three climatic zones found to be 18% and 8% and 23% and 19% lower than respectively c-Si and CIGS. The carbon-based perovskite energy payback period was estimated to amount to less than 1 year with room for further improvement. These results demonstrate the climatic benefit of a large-scale market deployment of carbon-based perovskite photovoltaics.
In order to limit global warming to 1.5 C by 2050, electricity generation from renewable energy sources would need to be significantly

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article. ORCID