If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Present embodied energy, embodied carbon, and energy payback times of CdTe and Si PV
•
Examine the drivers for embodied energy/carbon for state-of-the-art Si and CdTe PV
•
Contextualize the cost of decarbonizing with Si versus CdTe PV and remaining carbon budget
•
Manufacturing location and thin-film PV can markedly reduce the carbon intensity of PV
Context & scale
How does one “green” an already green technology? The scale of decarbonizing the global economy will require a massive investment not only in terms of money but also of our remaining carbon budget to limit the temperature rise. This work examines the embodied energy and embodied carbon (the amount of energy and greenhouse gas emissions required for manufacturing) of the two dominant types of photovoltaics, silicon (Si) and cadmium telluride (CdTe), and drivers that can reduce their carbon intensity. If the highest carbon-intensity scenarios are realized, 2%–14% of the remaining estimated carbon budget might be consumed for manufacturing PV modules.
Most PV are presently manufactured on a coal-rich grid; changing manufacturing location leads to ∼2× (present-day) swings in embodied carbon. Further reductions are possible with increased renewable grid penetration. Furthermore, thin-film technologies such as CdTe present an opportunity to reduce embodied carbon relative to Si PV by another ∼2×.
Summary
Looking beyond the traditional cost and efficiency metrics of photovoltaics (PV), this work evaluates the impact of embodied energy, embodied carbon, and energy payback time of two dominant technologies (CdTe and Si) on global decarbonization goals. The relative effects of PV technology type, technological advances, energy grid mix, and recycling are evaluated in terms of fostering decarbonization goals. If the highest carbon-intensity scenarios are realized, 2%–14% of the remaining estimated global carbon budget might be consumed to manufacture modules without including their balance of systems. Applying a carbon cost indicates that CdTe might have an additional value of $0.02–$0.04/W relative to Si PV manufactured with the same energy mix. Due to the scale of the challenge, any actions leading to an increased deployment of thin-film PV and/or a significant decrease in the deployed PV’s embodied carbon through changing the manufacturing grid mix have demonstrable value in helping the world stay within its remaining estimated carbon budget.
The international community is pursuing ways to curtail anthropogenic greenhouse gas (GHG) emissions to minimize the impacts of climate change. Notably, the 2015 Paris Climate Accords target a temperature rise to well below 2°C, with efforts to limit the rise to 1.5°C.
Although strategies for achieving this have largely been left to individual countries, the Intergovernmental Panel on Climate Change (IPCC) has indicated that to limit the rise to 1.5°C, GHG (i.e., equivalent CO2 or CO2e) emissions must be curtailed with an estimated total remaining carbon budget of 300–900 GtCO2e (17%–83% confidence).
The physical science basis. IPCC, 2021: summary for policymakers.
in: Masson-Delmotte V. Zhai P. Pirani A. Connors S.L. Péan C. Berger S. Caud N. Chen Y. Goldfarb L. Gomis M.I. Climate Change 2021: The Physical Science Basis. Cambridge University Press,
2021
One of the major impediments to reaching these climate goals is transitioning to carbon neutral sources of energy in a cost-effective manner. A significant fraction of GHG emissions occur during the burning of fossil fuels for transportation, heat, and electricity. To benefit from industrial process electrification and the growth of electric vehicles, it is critical to accelerate the implementation of low-carbon energy sources into the electrical grid mix. Increasingly, renewable energy technologies, specifically photovoltaics (PV), have become more cost-effective and are produced at a sufficiently large scale to be plausibly considered as a part of the solution as the era of global Terawatt (TW) scale PV is reached, with projections of 1 TW of installed PV capacity by 2023.
The present PV market is dominated by two technologies. The most prevalent technology, silicon (Si) PV, has greater than 90% of the global market share.
Cadmium telluride (CdTe) PV makes up ∼90% of the balance, with the vast majority of the rest made up by copper indium gallium selenide (CIGS). CdTe notably comprised 40% of the US axis-based tracking market, according to the 2019 US Energy Information Administration (EIA) Annual Electric Generator Report.
Based on analysis of EIA data, CdTe is found to account for ∼25% of cumulative US installations >1 MW. This work focuses on these two technologies; however, many of the conclusions for CdTe could be extrapolated to other thin-film PV technologies (e.g., perovskite PV) that emerge at scale.
While clearly both Si and CdTe PV provide significant benefits over traditional fossil fuel energy sources, it is instructive to look beyond the traditional cost and efficiency metrics for which these technologies are typically evaluated. A life cycle assessment (LCA) is performed that examines the impact in terms of embodied energy, energy payback time (EPBT), and embodied carbon to better understand their costs and contributions in the transition to carbon neutral electricity. Here, the term “embodied” is used to indicate the sunk costs (i.e., energy and GHG emissions) to complete the manufacture of a PV module. First, the new analysis is compared with a survey of previous analyses (details in supplemental information) to support the validity of our analysis and illustrate the improvements in the present state of the art. Next, we examine the sources of the largest contributors to embodied energy and GHG emissions and assess the potential for recovery and reuse of the critical materials of the CdTe semiconductor (i.e., cadmium and tellurium) and their effect on these metrics. Then, the GHG emissions for the manufacturing of Si and CdTe PV are calculated for four electricity generation energy mixes representative of the European Union, United States, China, and India. Notably, electricity GHG emissions go to zero as the energy mix becomes carbon neutral. To frame the effects of the variations between technology and energy mix, the GHG emissions that would be generated in manufacturing 1 TW are evaluated, a value that the US Department of Energy (DOE) recently explored as a 2035 scenario in its Solar Futures Study.
This is then contextualized in terms of the remaining global estimated carbon budget—namely, how to deploy PV while keeping within the budget. If the highest carbon-intensity scenarios are realized, we find that as much as 2%–14% of the remaining budget might be consumed to manufacture modules before accounting for their balance-of-system costs (e.g., racking and inverter). Finally, we ask through a technical lens, “How do we reduce the carbon impact of an already green technology?”
Results
Embodied energy, EPBT, and embodied carbon
Over the past decade, PV has seen numerous advances in both Si and CdTe technologies as they have matured to their present-day forms. Si has undergone a series of cell, module, and processing optimizations, including rear surface passivation with aluminum (Al) oxide, bulk passivation through hydrogenation reducing light-induced degradation, introduction of a selective emitter, and the cost reductions of monocrystalline Czochralski Si wafers through ingot processing and the use of diamond wire saws.
All these contributed to cost reductions as well as efficiency improvements in the transition from Al back surface field (Al-BSF) architecture devices to the now commonplace passivated emitter rear contact (PERC) architecture.
These technological changes, together with high-level mechanisms (e.g., economies of scale), have driven the module price per watt from 1.9 USD/Wp in 2010 down to 0.2 USD/Wp in 2021; average efficiencies have gone from 15% in 2010 to 22% in 2020.
CdTe has also undergone numerous cell and module advances in this period. These include the elimination of parasitic absorption from a CdS emitter, improved stability from a doped ZnTe interlayer in the rear contact, substitution of a graded Cd(Se,Te) alloy for a fixed composition CdTe absorber, increased module size, and the transition of the defect chemistry from copper to group-V dopants.
With these changes, CdTe has continued to directly compete with Si modules at the utility scale, as its efficiency has improved from 10% to 19% and costs have dropped from 1.4 USD/Wp to 0.22 USD/Wp.
Figure 1 provides a survey of LCAs for Si and CdTe PV over the past couple of decades together with the results from our new analysis that takes into account the advances to the state of the art, described in the previous two paragraphs, for both technologies (details regarding our analysis are provided at the end in experimental procedures). Several features are important to note. First, when performing LCAs, it is necessary to establish a scope, differentiating between “cradle-to-gate” and “cradle-to-grave.” For cradle-to-gate, the embodied energy and embodied GHG emissions are ascribed to each step necessary to complete the manufacture of a module, including the mining and transportation of the raw materials, the manufacture of processed components (e.g., glass, extruded Al, and ingots), and, finally, the fabrication of cells and module assembly. In the case of cradle-to-grave, further considerations include installation, use, and end-of-life activities. Cradle-to-grave analyses only increase energy expenditures/emissions, and they do not reduce these values by any energy generated over the life of a module; hence, it is expected that cradle-to-grave should always have higher values of embodied energy and GHG emissions. Scope definition across studies accounts for some of the scatter along with several other factors, including differences in recipe (e.g., glass thickness and single crystal versus multi-crystalline Si), grid mix, integration of transport/indirect energy and CO2, and process scope (e.g., recycling credits).
For the purposes of our analysis, all manufacturing was initially assumed to occur on the US electrical grid mix, with zero degradation for either module type over their lifetime.
Figure 1Comparing CdTe and Si PV architecture, embodied energy, and embodied carbon
Bottom: display of published literature values in open circles for cradle-to-grave analyses, in filled circles for cradle-to-gate analyses, in comparison with the results of this work in triangles. As displayed, the results show high parity with literature, reaffirming the concept that CdTe modules, in blue, have a lower energetic impact than Si counterparts, in orange.
While the literature shows a significant amount of variation, there is a clear downward trend. The improvements are most dramatic for Si, as might be expected, because the embodied energy is normalized to area, and over time, the Si wafer thicknesses decreased together with kerf losses from dicing ingots.
Less change is expected for CdTe since little of the embodied energy comes from the cell, but rather comes from its component materials and the energy required to deposit films and assemble a module (see Figures 2 and S1–S3 for additional component comparisons). Although the embodied energy per kWh for CdTe calculated in this work is lower than that in other studies due to increased module efficiency, it is worth noting that the introduction of the Al frame partially negates these gains. This is because the transition from a frameless module to a larger framed module uses thinner glass, which reduces some embodied energy that is more than balanced by the embodied energy of the new Al frame. For both Si and CdTe, the lower embodied energy coupled with the efficiency improvements mentioned above explain the reduced embodied CO2 per kWh.
Top: energy payback time calculated from the MFI values of embodied energy for both CdTe solar modules (left) and Si solar modules (right).
Bottom: energy payback time as a function of manufacturing energy and manufacturing grid efficiency for both CdTe solar modules (left) and Si solar modules (right). Historically, grids dominated with fossil fuels have grid efficiencies ∼0.3. With the penetration of renewable energy sources that are direct electricity producers (e.g., PV) to replace sources with significant waste heat, the grid efficiency will approach unity. There will still be losses through elements such as inverters. This analysis considers CdTe and Si solar module technologies with efficiencies of 19% and 21.13%, respectively. Irradiation was assumed to be 1,700 kWh/m2/year, consistent with the average irradiations in the US and southern Europe. CdTe annual power production is expected to be at maximum 323 kWh/m2/year, while Si is expected to be 359.2 kWh/m2/year. The analysis did not consider installation or balance of system (BOS). Degradation was not considered, as the EPBT was less than 1 year. The embodied energy was calculated cradle-to-gate, assuming US manufacturing.
Using the embodied energy values produced by the Materials Flows through Industry tool (MFI—more details are provided in experimental procedures) for Figure 1, new EPBTs were calculated for both Si and CdTe solar modules (Figure 2, top). The overall embodied energy of the CdTe module was 794 MJ/m2 compared with 2,297 MJ/m2 for Si, and EPBTs were 2.5 months for CdTe compared with 6.4 months for Si, assuming a US grid with a grid efficiency of 0.3. In this paper, EPBTs are a calculation of how long it takes to produce enough electricity to offset the cumulative primary energy required to manufacture the module. It should be noted that the EPBT is typically calculated for a complete system, as a module does not operate on its own; however, for the purpose of this paper and to compare with the EPBTs reported by manufacturers and in some literature, we use a simplified equation that only describes the module without its balance-of-system embodied energy costs. This leads to lower EPBTs relative to complete systems. Although the present grid efficiency is ∼0.3, as the grid composition trends toward direct electricity generators as opposed to historical generation sources that waste heat, the energy efficiency of the grid will go toward unity, and this will make the EPBT for PV modules longer (Figure 2, bottom). Further details can be found in the supplemental information.
For CdTe modules, aluminum and glass are shown to be significant contributors to overall embodied energy, contributing 32% and 30%, respectively. Outside of these components, assembly energy was the largest contributor to module embodied energy, contributing 33%. Looking further at the critical material, CdTe contributes only 0.4% of the overall embodied energy of a CdTe solar module. This is due largely to its small overall weight contribution (0.12%) to the module and its status as a byproduct during the mining of other materials. Si wafers, on the other hand, contribute 76% of the overall embodied energy of a Si module. MFI values for embodied CO2e for CdTe and Si fall below the range of literature values, likely due to improvements to the state-of-the-art modules and increased penetration of renewable energy in the energy grid mix, as discussed in the discussion section. CdTe modules have reduced impact when compared with Si in both embodied energy and embodied carbon.
Recycling potential
As PV becomes more prevalent, there has been a growing concern associated with the sheer mass of electronic waste that will be produced, with global estimates of up to 80 Mt of waste by 2050.
In the 2012 revision of its waste electrical and electronic equipment (WEEE) directive, the European Union (EU) made PV recycling mandatory, setting responsibilities for producers as well as targets for collection and recovery (80% by mass).
European Commission Directive 2012/19/EU of the European Parliament and of the Council of 4 July 2012 on waste electrical and electronic equipment (WEEE) (recast). Official Journal of the European Commission.
Although Si PV module architectures are remarkably similar across the industry, Si PV manufacturers are not directly involved in recycling and a standardized recycling process has not been established.
While Al and Cu are both high value and readily recycled, solar-quality glass is presently down-cycled into applications such as foam and fiberglass production or road aggregate.
Life Cycle Inventory of Current Photovoltaic Module Recycling Processes in Europe. IEA PVPS Task 12, International Energy Agency Power Systems Programme, Report IEA-PVPS T12-12:2017.
Although Si PV makes use of the critical material, silver, recent trends have been to reduce the amount of silver used in modules, with targets to reduce it by 75%.
This makes efforts to recover Ag less economically attractive such that a large amount of Ag might ultimately be lost if Si modules are not recycled, as there are no toxicity concerns driving Ag recovery. Although there is interest in recovering the Si, especially since a significant fraction of embodied energy stems from it, the sensitivity to impurities is well established,
This means that any recovered Si will require repurification, limiting energy savings, and advances are needed to make this more attractive than down-cycling to metallurgical-grade Si, which has significantly decreased in value relative to solar-grade Si.
First Solar, the largest manufacturer of CdTe, has already established a collection and recycling program with reported recovery rates of 90% for the glass cullet and semiconductor, in addition to metal recovery (e.g., Cu cables). The glass and Al make up the majority of the module weight and are used to meet the EU’s WEEE Directive.
It has a side benefit of securing a future source of CdTe. This is of note because tellurium is classified as a critical material and has low known reserves. CdTe PV uses about 40% of the global annual production of Te.
Typically, Te is a byproduct of other mining streams, but reports indicate that in Cu anodes alone, there is 4–5× the world’s annual Te production, with an estimated potential of 2,300 tonnes/year.
However, global availability may be underestimated due to Te’s nature as a byproduct and its limited use in other applications (for example, some reports suggest entire continents have no Te, although they produce Cu and Zn).
For the analysis, each recycling step’s energy was quantified and attributed to applicable components by the weight of the final recovered material. Their analysis finds the energy required for the initial recycling steps of a CdTe module to be 4.4 kWh/m2, which produces used semiconductor material sludge and glass cullet.
Combining these numbers results in the energy required for recycling the CdTe to be used in another module. Although the analysis of Ménard et al. was performed on 2012 vintage modules, their work can be adapted to examine the modern CdTe module architecture with a larger form factor and Al frame.
The recycling of glass cullet is well studied, so an embodied energy value of 6.63 MJ/kg was used.
The Al frame is a new addition to CdTe module technology; however, these frames are removed mechanically before the process described by Ménard et al. begins. The recycling of this product was also evaluated using an industry-average recycled Al embodied energy value of 42.9 MJ/kg.
—the majority of energy in the mining and refining process is ascribed to Cu recovery. When recovering the semiconductor from the module, it is dilute by mass (CdTe is 0.1% of module weight). If all the energy in the extensive recycling process described by Ménard et al. is ascribed to the CdTe recovery (rather than some to glass and Al recovery), the resulting semiconductor’s embodied energy is 2,657 MJ/kg, an increase of 15 times when compared with the embodied energy of primary (mined) CdTe at 173 MJ/kg.
However, in most recycling processes, including the one performed by First Solar, multiple materials are recovered in tandem with the recovery of the CdTe. The Al frame of the module is stripped off at the plant and reused or recycled, the glass is crushed and cleaned in tandem with the CdTe.
A separation occurs, splitting the CdTe and the cullet glass, and recycling processes occur for these independently. When taking the shared steps into account, the embodied energy of the CdTe semiconductor only increases to 1,577 MJ/kg, nine times the primary (mined and refined) value, while embodied energy of the glass and Al are reduced by 65% and 71.6%, respectively. Reusing the Al and glass reduces the overall module embodied energy significantly, as seen in Table 1. This supposes that the glass is recycled rather than down-cycled, a point that is not clear for solar-quality glass. It is worth noting that if a material is down-cycled, then in standard LCA accounting, the credit is attributed to the final product where the mass will be incorporated in its next life cycle—there is no holistic account that credits the product from which the mass was recovered. Thus, the case of down-cycled glass would be consistent with any of the scenarios except for recycled glass.
Table 1Embodied energy values for CdTe solar modules made of primary materials; recycled CdTe and primary materials; and recycled glass, aluminum, and CdTe with the rest of the primary materials
Composition
Embodied energy
CdTe (primary material)
762 MJ/
55.0 MJ/kg panel
CdTe (recycled CdTe)
805 MJ/
58.1 MJ/kg panel
CdTe (recycled CdTe, recycled aluminum)
610 MJ/
44.1 MJ/kg panel
CdTe (recycled glass, recycled aluminum, and recycled CdTe)
461 MJ/
33.3 MJ/kg panel
It should be noted that it is unlikely that recycled CdTe will provide a large enough resource for a completely circular process in the near term. Similarly, at present, recycling is actually down-cycling as there is no process to recycle solar-quality glass to solar-quality glass.
The electrical grid mix associated with the manufacturing location of these technologies can have a large impact on their embodied CO2. The manufacturing locations for CdTe and Si solar modules are shown in Figure 3. The locations were determined using data from BloombergNEF (BNEF), a provider of strategic research on the pathways for the power, transport, industry, building, and agriculture sectors to adapt to energy transition, and supplemented by primary research.
As shown in Figure 3, Si solar modules are manufactured primarily in China, which has a coal-rich energy grid as represented by energy mix 2 (Table 2), and a significant portion of CdTe is manufactured in the United States on a natural gas-rich grid (mix 1). The largest difference in terms of GHG emissions can be thought of as a substitution of natural gas for a large portion of coal.
Comparing the two solar technologies and their environmental impacts given different energy mixes provides insight on the energy and carbon impact of manufacturing PV. Additionally, we can understand the potential for reducing manufacturing impacts by examining the effect of transitioning a portion of the grid toward clean energy. Although a company might manufacture specifically with renewable energy (e.g., PV), making the embodied carbon of the purchased electricity close to zero, the intermittency of renewables currently requires fossil fuel sources, making the total grid mix a fair comparison. Table 2 shows the analysis for both current energy mixes and projected 2050 mixes.
The resulting kgCO2e/kWh for both CdTe and Si were evaluated for the 8 scenarios. Figure 4 shows the contrast when the energy mix shifts from a cleaner to a majority coal energy mix.
Table 2Energy mixes (%) representative of leading solar manufacturing economies
The embodied carbon is remarkably dependent on energy mix. Comparing the two lowest carbon mixes (mixes 1 and 4) still leads to a 30%–40% difference in emissions generated to manufacture PV; comparing mix 1 with mix 2 leads to an additional ∼25%–30% increase. This is consistent with other analyses of CO2e emissions.
On the scale of a single module, the relative difference in environmental impact between mixes with significant renewable penetration and coal skewed grid mixes is small. However, on the scale required to meet Paris Accord goals, the picture is far more dramatic. Evaluation of the two major solar technologies, CdTe and Si, with different (static) energy mixes to reach 1 TW can be seen in Figure 5.
Figure 5Emissions to manufacture 1 TW of PV modules
Energy input and subsequent CO2e emissions associated with the implementation of the two major solar technologies to manufacture 1 TW of solar capacity.
The physical science basis. IPCC, 2021: summary for policymakers.
in: Masson-Delmotte V. Zhai P. Pirani A. Connors S.L. Péan C. Berger S. Caud N. Chen Y. Goldfarb L. Gomis M.I. Climate Change 2021: The Physical Science Basis. Cambridge University Press,
2021
Clearly, technology choice can significantly reduce carbon impacts, but the effect of the manufacturing grid’s energy mix is also critical. In addition to technology, the geographical location and its influence on energy grid mix plays a smaller but non-negligible role in the scale of environmental impact of solar modules. By implementing CdTe manufactured with energy mix #1 versus Si with energy mix #2, over 786 million metric tons of CO2e would be mitigated. Clearly, the purpose of producing these modules is to make a cleaner energy mix. Thus the 2050 energy mixes illustrate the effect of a direct replacement of the dirtiest energy. However, if massive amounts of PV are manufactured with exponential growth, a direct substitution is overly optimistic since more energy will be needed to facilitate their manufacture. The simplification of 100% deployment of Si or CdTe with a single static energy mix is meant to illustrate the effects of energy mix and technology. It is understood that a mix of the currently available solar technologies, CdTe and Si, as well as future technologies, such as perovskites, will be used to reach these goals and that energy mixes will get cleaner with time. However, this can be used to bound how much CO2 might be emitted to reach 1 TW installed.
Discussion
The IPCC has estimated the world’s remaining carbon budgets to keep the temperature rise at 1.5°C and 2°C. Figure 5 compares the embodied carbon of 1 TW of Si and CdTe modules made by different energy mixes to the 50% confidence levels of this remaining carbon budget. Although 0.1% of these budgets seems small, estimates of PV required to reach decarbonization goals range from 7 to 55 TW.
This translates to 2%–14% of the estimated remaining carbon budget if the highest embodied carbon cases are realized. Although this is likely a high estimate, even if it is within a factor of two, there is a clear benefit to understanding the “knobs” available to reduce the carbon impact of PV deployment. These knobs can be assigned to three general categories: energy grid mix, technology/technology advances, and recycling.
The effect of energy grid mix is the best understood knob. Unfortunately, manufacturing has typically been done using the most carbon-intensive mixes. The present-day and projected 2050 mixes illustrate the importance of this as it steps from a less-carbon-intensive (nuclear and renewable) to natural-gas-rich to more-coal-rich mix (Figure 5). Comparing the 2020 mixes, there is a factor of almost two between the cleanest and dirtiest—creating near-term realizable incentive to shift the manufacturing location. Replacing the dirtiest sources in the projected 2050 scenarios leads to reductions of another factor of 2–3.
The comparison between the two dominant PV technologies highlights the importance of semiconductor choice and its technological advances. Some advances are nominally achievable by any technology. Advances may optimistically enable 26% efficiencies for both Si and CdTe in the next 15 years.
Increased energy yield per nameplate efficiency through advances such as bifaciality is another potential area for improvement. Although, at present, there is no accepted way to quantify the increased energy yield, with many factors such as the ground’s albedo playing a role, reports indicate gains from a few percent to as much as 25%.
Presently, bifacial designs are becoming prevalent in Si, but have not been achieved with thin films. Similarly, reduced degradation rates/increased module life can increase energy yield. The recent transition to a new defect chemistry in CdTe has led to performance warranties that now surpass Si (0.2%/year warranted degradation for CdTe compared with 0.55%/year for Si).
Improvements in efficiency and energy yield will ultimately only be adopted if they result in lower cost. Each of these advances has the potential to further reduce the environmental impact of PV, but at the tens of percent level. Reductions in glass usage, through thinning of glass or introduction of non-glass transparent backsheets for bifacial modules have the potential to further reduce the environmental impact.
Other differences are inherent to technology differences. As shown above, using the same energy mix (Mix 1), Si PV is inherently more energy intensive than CdTe (thin-film) PV, with 2.9× more embodied energy, 2.6× higher EPBTs, and 2.0× higher embodied carbon. Si had 76% of its energy embodied in the semiconductor compared with 0.4% for CdTe. The much larger embodied energy associated with Si relative to CdTe is due to several factors, including the ∼75-fold more material by volume required for the wafers plus kerf losses compared with 3 μm thick films, the increased purity requirement for Si, and the greater thermodynamic stability of SiO2 (hence more energy required to strip the oxygen).
Although Si wafers can be made somewhat thinner, there is limited headroom. Since the vast majority of embodied energy and carbon in a CdTe module stems from its assembly energy and package, emerging thin-film (e.g., perovskite) modules could readily be estimated to have similar values.
Scaling PV beyond a TW can raise concerns associated from perceived scarcity (i.e., Te and Ag) and toxicity (i.e., heavy metals). As mentioned previously there are 4–5× more Te resources than is currently recovered from Cu byproducts alone, but there has been limited economic incentive to extract Te since the market is so small. Furthermore, in the present-market there is a disincentive to target Te-rich Cu deposits since miners are charged penalties by smelters when high Te concentrations are present.
Conservatively, these resources provide sufficient Te resource to manufacture 1–2 TW of CdTe by 2050; in contrast, with appropriate incentives to develop non-Cu Te resources and thinning of the absorber layer could increase CdTe potential by an order of magnitude or more.
For instance, there are underutilized Te resources outside of Cu deposits including Pb, Au, Ag, and Bi as well as decades of tailings which should facilitate relatively low-cost Te sources. If Te-rich ferromanganese resources on the ocean floor were to come into play (conservatively estimated at 9 million tonnes), there is sufficient Te for an additional 200 TW.
Present Te resources could enable at least 25% annual CdTe growth through the end of this decade. For Si PV, similarly Ag scarcity/cost has been raised as a concern for large-scale deployment. As Ag is not a part of the Si absorber, it has been viewed as an important but less integral and limiting concern. There is active research in reducing the amount of Ag in contacts, and although previous targets to eliminate its use have been missed, it is likely that if scarcity limits growth, price signals will push the industry to an alternative.
Silver as a constraint for a large-scale development of solar photovoltaics? Scenario-making to the year 2050 supported by expert engagement and global sensitivity analysis.
Interestingly, although direct Cd releases during manufacturing and fires for CdTe might raise toxicity concerns, a thorough life cycle analysis by Fthenakis et al. indicates that the amount of Cd emissions is two orders of magnitude lower than that for fossil fuels per GWh, which also emit As, Pb, Hg, and Ni.
Due to the higher electricity requirements for manufacturing Si PV, an equivalent amount of CdTe PV results in lower Cd and other heavy metal emissions, but adding each to the grid will significantly reduce the heavy metal emissions relative to the present, fossil-fuel-powered grid. All of this being said, as manufacturing processes scale it is fair to scrutinize their environmental impacts and identify places for improvement.
The last knob is increasing the use of recycled materials. For Si, the biggest single contributor to embodied energy and carbon is the Si wafer, but it is unclear when there might be an established process, thus further research is warrantied. For CdTe, recycling is already occurring. The absorber recycling is motivated by criticality of materials rather than embodied carbon or energy. In both Si and CdTe, glass is presently down-cycled, presenting a potential opportunity for improvements. Al is readily recycled, although recent analysis has indicated that in addition to module materials racking presents another significant source of embodied energy and carbon.
Although clearly important, due to the desire to exponentially increase our deployment of PV, direct recycling is unlikely to strongly impact the spend rate of our carbon budget in the next couple of decades due to the warranted lifetime of PV.
The relative value of generating lower-embodied-carbon PV can be estimated in a couple of ways. The value associated with not generating carbon emissions can either be realized as a tax or as a cost to capture it post generation. First, direct air capture of carbon costs have been estimated to be ∼$500–$600/tCO2 presently, with targets of $94–$232/tCO2.
Alternatively, while present carbon taxes vary widely ($0.08–$127/tCO2e), estimates of pricing requirements to limit warming to less than 2°C suggest $50–$100/tCO2e for 2030.
Together, these indicate $50–$100/tCO2e is a reasonable estimate of the value. For instance, this translates to an overlooked ∼$0.02–$0.04/W difference for Si versus CdTe when manufactured with energy mix 1.
Conclusion
This work provided an updated evaluation of the embodied energy, EPBT, and embodied energy of the dominant PV technologies (Si and CdTe) and put these in terms of the IPCC’s estimated remaining carbon budget. Specifically, state-of-the-art Si and CdTe PV modules were modeled to consider the effects of technological advances, PV type, energy grid mix, and recycling. It was found that as much as 2%–14% of the carbon budget might be consumed in producing the modules to achieve global decarbonization—this does not include balance of systems (e.g., framing and inverters).
While it is critical to replace carbon-emitting energy sources in as short a timeframe as possible, it is worth examining the embodied carbon and energy for the clean energy sources and establish key considerations to reduce their overall impact.
It was found that the biggest opportunities come from manufacturing on less-carbon-intensive energy mixes today (∼2×) with another factor of two to be gained as the grid decarbonizes. Due to the high embodied energy intrinsic to Si, CdTe was shown to have 2–3× lower impact than Si. This was because glass and Al contributed the most to environmental impacts, whereas the impact of CdTe itself was negligible. Other thin-film technologies, such as perovskite PV, produced at scale with a similar architecture (e.g., glass/glass construction with Al frames) would be expected to follow similar trends. Additional technological advances such as efficiency improvements and increased energy yield (bifaciality, improved lifetime) could provide double-digit percent gains. Although recycling was evaluated, given the expected recycling streams and growing demand for PV, recycling will not be a substantial factor in achieving decarbonization goals. Presently, recycling appears to be driven by criticality/perceived toxicity of materials and policy, but in next generation architectures and technologies, embodied energy may begin to drive circularity.
While in the United States, 25% of cumulative PV installations >1 MW are CdTe, the 2020 worldwide PV manufacturing capacity is very different, with >90% Si.
Waiting for CdTe or an emerging thin-film technology to scale up rather than implementing Si would be an incorrect lesson to take away from this analysis, as that would perpetuate existing emissions from fossil fuels. However, due to the scale of the challenge, any actions that lead to an increasing deployment of thin film and/or significant decreases in the embodied CO2e emissions of globally deployed PV by changing the overall grid mix used to manufacture PV have demonstrable value in helping the world stay within its remaining estimated carbon budget.
Most PV is presently manufactured on a coal-rich grid; changing manufacturing location leads to ∼2× (present-day) swings in embodied carbon. Further reductions are possible with increased renewable grid penetration. Furthermore, thin-film technologies such as CdTe present an opportunity to reduce embodied carbon relative to Si PV by another ∼2×.
Experimental procedures
Resource availability
Lead contact
Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Samantha Reese ([email protected]).
Materials availability
This study did not generate new unique materials.
Data and code availability
This study leverages the Materials Flows through Industry (MFI) tool,
which is based on the US industrial sector supply chains and uses supply networks to aggregate associated energy and material requirements in a cradle-to-gate analysis. MFI also estimates energy related CO2e (CO2 equivalent) emissions, allowing for the determination of embodied carbon for a desired product. The reported emissions value includes methane and nitrous oxide in addition to CO2. For the purpose of this paper, this is referred to as embodied carbon. Process data (referred to as recipes) in the MFI database are obtained from sources including the US Life Cycle Inventory database, the Ecoinvent v2.2 life cycle inventory database, and scientific literature.
In this work, updated CdTe and Si solar module recipes based on 2021 state-of-art were created in MFI. Although there are numerous reports of Si, CdTe reports are more limited and older. The goal was to enable a fair comparison by using the same analysis tool and process. For the Al frame and the glass, a new recipe was added to MFI. The specifics used in MFI can be found in the supplemental information (LCA methodology). MFI outputs for kgCO2e per manufactured mass was converted to kgCO2e/kWh using an insolation of 1,700 kWh/m2/year and module nameplate efficiencies, assuming no degradation over the rated module life. The highest performance module specifications were selected from the top manufacturers by size (First Solar and Longi).
For CdTe, the new recipe—adapted from an internal bottoms-up cost model recipe—includes recent updates to module size, a transition from a frameless to Al-frame design, and changes in layer compositions. In a preliminary analysis, the embodied energy approximations for a full device stack utilizing recipe component weights were analyzed to predict high-impact materials. This allowed further refinement and more in-depth evaluation of the high-impact components in the MFI database to ensure accuracy. Al, glass, and CdTe were determined to be the high-impact materials. The module efficiency was 19% and weighed 34.9 kg.
It should be noted that the manufacturing assembly energy for the MFI recipe (Table S2) is conservative when compared with the reported environmental production declaration (see supplemental information for more details).
The Si module recipe followed Jia et al., modeling a bifacial module with 72 PERC cells/144 half-cells with 166 mm sides.
Due to the large number of reports on Si modules, a sensitivity analysis on the component weights was not performed. The module efficiency was 21.13% efficient and weighed 29 kg.
Acknowledgments
This work was authored by the National Renewable Energy Laboratory, operated by Alliance for Sustainable Energy, LLC, for the US Department of Energy (DOE) under contract no. DE-AC36-08GO28308. Funding was provided by the United States Department of Energy Office of Energy Efficiency and Renewable Energy's Advanced Manufacturing Office and Solar Energy Technologies Office.
The authors wish to thank Joe Cresko with the Advanced Manufacturing Office for his significant support and discussions conceptualizing the project; Billy Roberts for his contributions making the maps; Al Hicks for his contributions making graphics; Scott Nicholson and Shubhankar Upasani for their assistance with MFI; and Nancy Haegel, Alberta Carpenter, Joseph Berry, Mark Ruth, and James Burst for their useful discussions.
Author contributions
Conceptualization, S.B.R. and M.O.R.; methodology, S.B.R. and M.O.R.; investigation, H.M.W., S.B.R., and M.O.R.; writing – original draft, H.M.W., S.B.R., and M.O.R.; writing – review & editing, H.M.W., S.B.R., and M.O.R.; resources, M.O.R. and S.B.R.; supervision, M.O.R. and S.B.R.; project administration, S.B.R.; funding acquisition, M.O.R. and S.B.R.
The physical science basis. IPCC, 2021: summary for policymakers.
in: Masson-Delmotte V. Zhai P. Pirani A. Connors S.L. Péan C. Berger S. Caud N. Chen Y. Goldfarb L. Gomis M.I. Climate Change 2021: The Physical Science Basis. Cambridge University Press,
2021
Directive 2012/19/EU of the European Parliament and of the Council of 4 July 2012 on waste electrical and electronic equipment (WEEE) (recast). Official Journal of the European Commission.
Life Cycle Inventory of Current Photovoltaic Module Recycling Processes in Europe. IEA PVPS Task 12, International Energy Agency Power Systems Programme, Report IEA-PVPS T12-12:2017.
Silver as a constraint for a large-scale development of solar photovoltaics? Scenario-making to the year 2050 supported by expert engagement and global sensitivity analysis.
00253-7&id=fx1.jpg){kind=link}
00253-7&id=gr1.jpg){kind=link}
00253-7&id=gr2.jpg){kind=link}
00253-7&id=gr3.jpg){kind=link}
00253-7&id=gr4.jpg){kind=link}
00253-7&id=gr5.jpg){kind=link}