Cost‐Performance Analysis of Perovskite Solar Modules

Perovskite solar cells (PSCs) are promising candidates for the next generation of solar cells because they are easy to fabricate and have high power conversion efficiencies. However, there has been no detailed analysis of the cost of PSC modules. We selected two representative examples of PSCs and performed a cost analysis of their productions: one was a moderate‐efficiency module produced from cheap materials, and the other was a high‐efficiency module produced from expensive materials. The costs of both modules were found to be lower than those of other photovoltaic technologies. We used the calculated module costs to estimate the levelized cost of electricity (LCOE) of PSCs. The LCOE was calculated to be 3.5–4.9 US cents/kWh with an efficiency and lifetime of greater than 12% and 15 years respectively, below the cost of traditional energy sources.


Introduction
The photovoltaic (PV) market has increased dramatically during recent decades. In 2014, there were about 40 GW of PV modules installed globally, 92% of which were crystalline silicon solar cells. [1] Although the price of silicon modules has decreased dramatically, the cost of electricity produced by PVs is still higher than that of electricity supplied by conventional fossil fuels. [2,3] Hence, expansion of the PV market has relied on government support to a great extent in the past. For example, as a result of the policy of feed-in tariffs, [4] PV installations in Europe increased greatly at the beginning of this century. [5] However, subsequent political policy adjustments have led to a considerable decline of PV installations in Europe. [6] To consistently promote the PV market, there is an urgent need to establish a cost-effective PV industry that can survive without government support. Figure 1 shows the structures of the designed modules, which were assembled with series connections. The cell in module A based on a mesoporous structure can be fabricated by using a series of simple techniques based mainly on screen printing (denoted as humble process) to produce moderately efficient modules as high as 15% (Figure 1a). [29] Spray pyrolysis can be used to deposit a compact bottom layer such as TiO 2 on a transparent conductor oxide (TCO) glass on which a pattern of rectangles has been etched with a laser, and the scaffold layers, including the back electrode, can be simply formed by multiple screen printing (with a defined mesh size for producing the desired patterns) and sintering at 400-500 ºC. The perovskite material is dip-coated within the mesoporous scaffold by moderate thermal annealing (90-150 ºC). Because of the imprecise boundaries produced by the screen printing technique, the active area in module A is unlikely to be very large. Assuming that the active area covered 80% of a module surface with an area of 1 m 2 , we calculated that the module efficiency would be 12% (cell efficiency of 15%) and that a power output of 120 W could be achieved with 1 m 2 of module A (Table S1). The cells in module B (Figure 1b) based on a precise structure were composed of several layers of high-quality thin films to produce highly efficient modules ≈20%. [32,33] Their fabrication required a series of finely controlled processes, production of patterns with lasers, and vacuum evaporation to produce metal electrodes. The high precision of the fabrication processes may cause the manufacturing cost to increase. The narrowness and precision of the etching produced by lasers is expected to improve the accuracy of the boundary and lead to a relatively large active area (0.95 m 2 in one piece of 1 m 2 module B). The calculated module efficiency is 19% (cell efficiency of 20%), and the resulting power output is 190 W (Table S1). Figure 2 shows the costs of modules of Module A and Module B at 1 st year, 5 th year and amortizing capital cost over 5 years. The module cost can be divided by the cost of materials, overhead cost, and capital cost. The capital costs for Module A and B were calculated based on the capital costs of DSCs fabricated using the printing process and thin-film silicon solar cells, respectively (Table S2 and S3). The cost of materials was estimated based on the amount of the materials that were used. The overhead cost was estimated based on reasonable assumption. The details of the calculation are shown in the Methods section and Supporting Information. The relatively high module cost in the first year was due to the high depreciation rate (50%) of capital investment. The calculated capital costs in the first year were 0.110 and 0.160 US$/W for Module A and B, respectively. The initial capital cost of Module A was lower because the capital investment associated with use of cheap printing facilities was lower than that of the high-vacuum machines used in Module B. However, the capital cost rapidly decreased because of depreciation, the result being a monotonic decrease of the total module cost during the first 5 years (Table S4 and S5). After that time, the contribution of capital cost to total cost became very low, so that, the module cost was determined mainly by overhead and materials costs. Regarding the cost of materials, Figure 3 presents the distribution of the materials cost for PSCs production routes. Active layers represent of perovskite layer, charge extraction layers, charge transport layer and back electrode. The cost of others in Figure 3 mainly included the expense on sealing materials and glass covered on backside of device. The cost of TCO is tallied separately because it takes up most of the materials cost in both structure. Although inexpensive transport layers and printable electrode materials were used in Module A, the total calculated cost of materials for Module A 0.127 US$/W was a little higher than the cost for Module B 0.102 US$/W (Table S6). The higher cost of materials for Module A was due to the fact that the cost of activelayer materials (except for TCO) accounted for only a small proportion of the total cost of materials, and the cost of TCO was high due to the smaller active area and low efficiency of  Table S7 and S8) were estimated to be 0.098 US$/W and 0.075 US$/W based on the report of DSCs and thin-film silicon solar cells production. [34,35] Hence, the conclusion could be drawn that cost of Module B produced by Module B and that of Module A produced by Module A are almost the same.

Estimation of Costs of PSC Modules
To compare the module cost with other PV technologies and calculate the electricity generating cost, amortizing module cost was also calculated by amortizing total capital cost by working lifetime of equipment. As shown in Figure 2, the amortizing module costs were calculated to be 0.250 US$ for Module A and 0.215 US$ for Module B, which are one third of module cost of bulk silicon solar cells (Table S9). These two amortizing module costs will be used for following sensitivity analysis and estimation of levelized cost of electricity which is usually considered as electricity generating cost.

Sensitivity Analysis of Module Cost
It is noteworthy that these cost estimates were based on assumptions about the two kinds of cell structures. However, the assumed parameters may vary when the PSCs are commercialized. Hence, we performed further sensitivity analyses to consider the effect of PCEs on module costs. The module costs increased exponentially as their module efficiency decreased (Figure 4) Figure 4). This result revealed that the module efficiency acted as an important factor for module cost no matter which route was used for manufacturing. Improvement of the cell efficiency and active area by upgrading precision of printing method for further increase of the module efficiency is effective way to reduce the cost of module A.

Levelized Cost of Electricity Produced with PSCs
The LCOE is typically used to compare system costs of electricity produced using different sources of energy. The LCOEs of traditional energy sources were 7.04-11.90 US cents/kWh, and the costs of solar PV technologies were 9.78-19.33 US cents/kWh reported in Levelized Cost and Levelized Avoided Cost of New Generation Resources in the Annual Energy Outlook 2015. [36] The LCOE was calculated according to Equation.
(2) (Method part), it was affected mainly by module cost, efficiency, and lifetime. In our module cost analysis, both Module A and Module B were estimated to produce perovskite solar modules at a cost in the range of 0.21-0.28 US$/W. We calculated the LCOE of a perovskite solar module by assuming a module cost of 0.25 US$/W and a lifetime of 15 years. The LCOEs were 4.9 US cents/kWh, 4.2 US cents/kWh, and 3.5 US cents/kWh corresponding to module efficiencies of 12%, 15%, and 20%, respectively, which were lower than that of traditional energy sources (Figure 5). Details of the calculation are shown in the Methods section and Table S10. This analysis indicates that module efficiency has a significant influence on the LCOE.

www.advancedscience.com
Adv. Sci. 2017, 4, 1600269 www.advancedsciencenews.com  Table S1.  Figure 6 shows the effect of lifetime on the LCOE of perovskite solar cells. The LCOEs estimated by module efficiency of 12%, 15% and 20% decrease exponentially with the extension of the system lifetime in the range 10-30 years. For highefficiency (20%) modules, a lifetime of 10 years can lead to an LCOE of 4.7 US cents/kWh. The low-efficiency (12%) modules require a long lifetime (30-years) to achieve the similar LCOE. A conservative estimate of discount rate 5% is used above.

ESSAY
Based on the analysis above, the module efficiency and lifetime were the most sensitive factors for the LCOE of PSCs. The ultra-low LCOE of PSCs was achieved to be 3.5-4.9 US cents/kWh with 15 years lifetime, surpassing the United States "SunShot Initiative" target of 6.0 US cents/kWh. [37] Although the efficiency of small size PSCs had already beyond 20%, the efficiency record of larger size with 1 cm 2 aperture area was reported to be 15%. Regarding lifetime, studies related to longterm stability of PSCs were still limited. [38,39] Hence, improvement of the efficiency and the lifetime of PSCs are urgent tasks from the perspective view of cost, and more efforts should be devoted to this field.

Conclusions
We designed two representative module structures for PSCs as low materials cost module and a high precision high efficiency module. The costs of the modules were estimated based on an annual capacity of 100 MW and the reported solar cell performances of the two kinds of cell structures. We found that the module costs for both structures could be much lower than those of other solar PV technologies. The results of the sensitivity analysis indicated that an increase of module efficiency could significantly reduce module cost. The fabrication of high-efficiency modules through the high precision processes was the most promising approach for further reducing the cost. The LCOE of perovskite solar cells was also very sensitive to module efficiency and can be expected to be lower than that of traditional energies if the module efficiency and lifetimes can exceed 12% and 15 years, respectively. To achieve these targets, more efforts should be made to improve the lifetime and efficiency of perovskite photovoltaic devices rather than to identify cheaper materials and processes.

Methods
Module cost estimation: To assess the cost of fabricating the modules, we assumed the production capacity of both routes www.advancedscience.com Adv. Sci. 2017, 4, 1600269 www.advancedsciencenews.com   Because the full process of printing module A was based on the fabrication of dye-sensitized solar-cell (DSCs), the capital cost was based on the capacity of DSCs, for which the module efficiency was 6% and the CI was 11 million US$ for a production capacity of 50 MW. Because the module efficiency of 12% via Module A was twice that of DSCs, the capital investment for Module A (CI Module A ) with 100 MW capacity per year was estimated to be 11 million US$ per year (Table S2). The module cost for Module B was estimated in a similar manner based on the production of silicon solar cells with an annual capacity of 60 MW, as shown in Table S3. The capacity for Module B was 1.6 times the capacity of thin-film, silicon solar cells because of the higher efficiency of the PSC module. In addition, the capital investment for a capacity of 100 MW via Module B (CI Module B ) without chemical vapor deposition was 40% of the CI of a thin-film, silicon solar cell manufacturing line with a capacity of 60 MW. Thus, the CI Module B for a capacity of 100 MW was 16 million US$. The details of the estimate are presented in the Supporting Information.
The depreciation of the facility resulted in a decrease of capital investment from year to year according to Equation (1): [37] CI( ) CI n n β = × where n is the number of years after construction and β is the depreciation ratio, which we assumed to be 0.5 based on the PV industry. Depreciation of an investment should cease when β n is less than 0.  (Table S9). These were the baseline values used in the sensitivity analysis.
Estimation of the levelized cost of electricity: The total cost of the solar cell system, including the costs of the module, balance of systems (BOS), land, support structures, wiring, power conditioning, and installation, [37] was calculated with Equation (2): [40,41] LCOE ICC 1000 CRF / CF 8760 O & M ( ) ( ) where ICC is the Installed Capacity Cost ($/W DC) = BOS cost + module cost, CRF is the Capital Recovery Factor = (i × (i + 1) n )/((i + 1)n -1 ), CF = Alternating Current Capacity Factor (0.8 × sunlight/8760 hours, reduced by 20% losses to go from direct current to alternating current), O&M = Operation and Maintenance ($/kWh), i = discount rate, n = lifetime (the lifetime of system).
Assumptions were as follows: BOS was 75 US$/m 2 based on based on the projected long term goal for traditional siliconbased solar cell in 2020. [37] BOS costs for efficiency of 12%, 15% and 20% were 0.625 US$/W, 0.5 US$/W and 0.375 US$/W, respectively, by using BOS cost = 75 US$ × m -2 /output; O&M = $0.001/kWh; i = 5%, and n = 20 (no tax credits and no accelerated depreciation), from these values, CRF (i = 5%, n = 15) = 0.096. To find the energy produced in a year by 1 W of installed PV, we used a CF of 20%. This assumption takes into account that PV cells only operate at a fraction of peak power when averaged over the course of a year with 1700 kWh/m 2 per year. [42] www.advancedscience.com Adv. Sci. 2017, 4, 1600269 www.advancedsciencenews.com Figure 6. The relationship between LCOE and lifetime. A system lifetime <10 years was not considered in our analysis.