SolarCity Photovoltaic Modules with 35 Year Useful Life

Transcription

1 SolarCity Photovoltaic Modules with 35 Year Useful Life Andreas Meisel 1, Alex Mayer 1, Sam Beyene 1, Jon Hewlett 1, Karen Natoli Maxwell 1, Nate Coleman 1, Frederic Dross 2, Chris Bordonaro 3, Jenya Meydbray 2, Elizabeth Mayo 2 1) 2) 3) SolarCity, 161 Mitchell Blvd, Suite 104, San Rafael, CA DNV GL, 1360 Fifth Street, Berkeley, CA DNV GL Renewables Advisory, 155 Grand Ave, Oakland, CA 94612

2 SolarCity Photovoltaic Modules with 35 Year Useful Life Andreas Meisel 1, Alex Mayer 1, Sam Beyene 1, Jon Hewlett 1, Karen Natoli Maxwell 1, Nate Coleman 1, Frederic Dross 2, Chris Bordonaro 3, Jenya Meydbray 2, Elizabeth Mayo 2 1) SolarCity, 161 Mitchell Blvd, Suite 104, San Rafael, CA ) DNV GL, 1360 Fifth Street, Berkeley, CA ) DNV GL Renewables Advisory, 155 Grand Ave, Oakland, CA Table of Contents SolarCity Photovoltaic Modules with 35 Year Useful Life Executive Summary Useful Life Introduction SolarCity Total Quality Control Program Rigorous Supplier Selection and Three-Level Oversight Process Stringent Module Quality Specifications Effective Prevention of Quality Deviations Constant Refinements and Total Integration World-Class Team behind the Scenes Useful Life Extrapolation from Accelerated Testing Ongoing Reliability Testing Overview ORT data Thermal Cycling ORT data Damp Heat, Humidity Freeze, and Dynamic Mechanical Load Testing PQP Testing Testing Beyond Standard Qualification Tests Product Qualification Program Testing Overview Product Qualification Program Testing Extended Thermal Cycling Product Qualification Program Testing Extended Damp Heat Product Qualification Program Testing PID Testing Product Qualification Program Testing Extended Humidity Freeze and UV Test Next Steps Tests with Improved Correlation to Real Life Useful Life Extrapolation Based on Degradation of Fielded Modules Conclusion References... 21

3 3 SolarCity Photovoltaic Modules with 35 Year Useful Life Executive Summary SolarCity believes that the Useful Life of the photovoltaic (PV) modules that are being installed on its residential and commercial systems is 35 years or longer. Per this definition, 95 % of all modules installed are expected to have an annual average degradation rate of less than ~0.5 % and produce at least 80 % of their power after 35 years of service. Experimental data from accelerated stress tests according to the industry standard IEC 61215, which were performed by DNV GL (the leading US certified 3 rd party), demonstrates that the median power degradation of modules supplied by seven key SolarCity approved module manufacturers for all tests and all module suppliers combined is as low as -1.1 % and as much as 35 % lower than for a comparable industry-wide selection of non-solarcity modules measured at DNV GL. Furthermore, data from DNV GL demonstrates that after extending accelerated testing to more than 3x beyond the conditions of IEC 61215, the modules produced for SolarCity show only 1 to 2 % median degradation and outperform non-solarcity modules, which are typically warranted for 25 years. The reason for this advantage is SolarCity s implementation of a stringent and industry-leading Total Quality Program, which adopted its features from the Automotive Industry and was implemented by SolarCity in early Following this program, SolarCity strategically chooses to engage with a select group of Tier-1 suppliers only. In order to be qualified as a SolarCity supplier, manufacturers need to have effective Quality Assurance programs and refined manufacturing processes in place, and steady product and manufacturing quality must be demonstrated. Rigorous tests need to be passed on an ongoing basis, performed by a qualified 3 rd party lab. Furthermore, we require that factory controls and in-line testing are in place to ensure quality is sustained over time and deviations are rapidly detected, so the deployment of faulty products in the field is prevented. Additional work is underway to demonstrate that the degradation rate from SolarCity modules in the field is lower than industry-standard. Lastly, the development and implementation of state-ofthe-art accelerated testing methods will enable SolarCity to probe degradation modes that are not detectable with the current industry-standard suite of testing and to more reliably predict real-life performance in the field. The most comprehensive meta-study of Field Degradation rates to date, where more than 11,000 annual degradation rates have been aggregated and analyzed, observed a near-linear degradation behavior for the majority of crystalline-silicon (Si) modules and established a median degradation rate for Si modules of around 0.5 % per year [1, 2]. The data in this study was analyzed and filtered by DNV GL analysts, and the annual median degradation rate for crystalline-si modules was confirmed to be ~0.50 % per year, while the corresponding value determined for systems is 0.77 % per year [3]. The data presented in the following supports the assumption that SolarCity s PV modules, as a result of its Total Quality Program and advancements in Materials Science, manufacturing, and quality control, perform at least similar, if not better than the median of all crystalline-si modules observed in the study above. Therefore, an annual module degradation rate of % per year is a realistic assumption, which warrants a postulation of Useful Life of 35 years with a power output of 80 to 82.5 % thereafter. 1 Useful Life Introduction SolarCity defines 35 year Useful Life as 95% of modules producing at least 80% of their power after 35 years in their use environment [4]. Use environment is defined as all geographic and meteorological conditions that the PV modules will experience during their lifetime. Site environmental conditions, installation, and handling are included in useenvironment considerations. This definition of Useful Life postulates a higher threshold for the remaining power output than the value of 70 % that has been assumed elsewhere [5]. Industry analysists have been getting more comfortable with the idea of Useful Life beyond 30 years [6]. The Useful Life of a PV module is determined by

4 Cumulative probability Normalized Frequency 4 SolarCity Photovoltaic Modules with 35 Year Useful Life wear-out failures, which occur at the end of the working lifetime of the module. SolarCity defines the end of a PV module s Useful Life if a safety problem occurs or if the module power drops below 80 % of the initial power rating. Long-term studies that have investigated wear-out failures [7] found that the predominant End-of-Life failures led to a median power loss of only 10 % (between 0 % and 20 %), and that nearly all of these PV modules were still functional and met the manufacturer s power warranty. Another literature meta-study summarizing ~400 reports on degradation rates of silicon modules confirms that modules are usually observed to degrade slowly in the field [8]. The degradation most often is dominated by a gradual loss of short-circuit current, which is mostly associated with discoloration and/or delamination of the encapsulant material. In other words, the most critical module failures have been observed to occur relatively fast, whereas modules that do not show early failures are likely to reach the wear-out portion of the bathtub product reliability curve, where the power declines in a gradual and slow manner rather than showing abrupt failure [9, 10]. There are numerous examples of installations that have delivered stable performance for well over 25 years [11]. In 1984, Sweden s first grid-connected photovoltaic system was built in Stockholm. Since its installation, the 2.1 kw system has been continuously and reliably producing energy with less than 3 % change since the system was installed 31 years ago. Another system installed in 1984 is at Kyocera s Sakura Solar Energy Center near Tokyo. The 43 kw array continues to generate a stable amount of electricity today 32 years later [12]. Given the drastic advancements in terms of Materials Science, manufacturing processes, quality control and standards, and theoretical understanding over the last 30+ years, it is considered reasonable to assume that the quality and reliability of modules fabricated over the last few years can have Useful Life well beyond 35 years, provided that adequate quality assurance measures, such as SolarCity s Qualification Program, have indeed been implemented. The most comprehensive study of Field Degradation rates to date, where more than 11,000 annual degradation rates have been aggregated and analyzed, established a median degradation rate for crystalline-silicon (Si) modules of consistently around 0.5 % per year (Figure 1) [1-3]. The data presented in the following supports the assumption that SolarCity s PV systems, as a result of its Total Quality Program and industry-wide advancements in Materials Science, manufacturing, and quality control, perform at least similar, if not better than the median of all crystalline-si systems observed in the study above. Therefore, an annual module degradation rate of 0.5 % per year is a realistic assumption, which warrants a postulation of Useful Life of 35 years with a power output of 82.5 % thereafter. (a) (b) Modules, all (1552) Systems, all (385) Modules, median (61) Systems, median (71) Degradation rate (%/year) Figure 1 [Taken from Ref 1]: Histograms of all data, high quality data and the median per study and system presented as the normalized frequency (a). Cumulative distribution functions for high-quality x-si systems and modules (b). The median is indicated by a dashed horizontal line, 0.5 %/year and 1%/year degradation are indicated as a dashed and dash-dotted line, respectively. The number of data points for the respective subsets is given in parentheses.

5 5 SolarCity Photovoltaic Modules with 35 Year Useful Life 2 SolarCity Total Quality Control Program 2.1 Rigorous Supplier Selection and Three-Level Oversight Process SolarCity has taken on an industry thoughtleadership position in the field of Quality and Reliability (Q&R) by developing and implementing a Total Quality Control Program, which is unique in its depth for the solar industry (Table 1). Its features were adopted from the Automotive Industry, and it was implemented by SolarCity in early Conventional Method Choose suppliers offering lowest cost No supplier qualification Buy off the shelf Implicit Quality/Reliability No feedback loops No learning from past experiences No Quality philosophy SolarCity Method Tier-1 suppliers only Rigorous supplier qualification Design to spec Automotive-based Q&R programs: For all key system components Define/control/validate Gates with sign-off, involvement Constant feedback between Installers, O&M, Engineering Analyze field and warranty data Company-wide Quality philosophy Industry thought-leadership in Q&R Table 1: Comparison of Quality and Reliability practices for typical solar installers and SolarCity, respectively. The Total Quality Program starts with a stringent selection process to establish its product suppliers. SolarCity chooses strategically to only engage with a select group of Tier-1 suppliers that have effective Quality Assurance programs and refined manufacturing processes with well-controlled Bills of Materials (BOM), which are thoroughly tested to rigorous standards. Tier 1 manufacturers are required to invest heavily in R&D, use highly automated manufacturing techniques and have at least five years history of producing solar panels. By exclusively selecting strategic Tier1 suppliers, SolarCity demonstrates its unconditional commitment to not compromise quality and reliability in the pursuit of ever lower cost targets, which is in contrast to numerous competitors who have been plagued by serious quality problems. For example, significant defect rates of PV modules were detected during audits of 50 Chinese factories between 2012 and 2013 [13]. SolarCity has been working relentlessly with suppliers to ensure that SolarCity products are free of such defects. SolarCity s commitment to quality is reflected in the fact that Q&R requirements are directly embedded into and enforceable through the Master Procurement Agreements (MPA) for its product suppliers. In order to be qualified as a SolarCity supplier, suppliers are contractually required to subscribe to a welldocumented three-level process: (1) Initial vendor qualification, which requires demonstrating the capability to manufacture the products according to well defined specifications and quality requirements, to pass an onsite factory audit, and to pass reliability testing through a chosen 3 rd party lab (DNV GL); (2) Continuous Production Oversight, which ensures consistent production of goods of high quality by means of regular BOM inspections and factory audits through SolarCity as well as 3 rd party auditors; (3) Ongoing Quality Assurance and Testing to ensure compliance with the Initial Qualifications and all quality criteria. The Total Quality Program was first implemented for PV modules and has since been extended to all key components of the entire photovoltaic system. It is now also in place for inverters, which often are considered the central part of the PV system, as well as for pre-installed and field-made PV connectors, which are another essential system component. Additionally, the program has been implemented for battery & storage and gateways. This three-pronged Quality Program that is in place for all relevant system components ensures that SolarCity products are designed for long-term reliability and consistent performance over the entire system life exceeding 35 years. 2.2 Stringent Module Quality Specifications SolarCity has developed a rigorous testing program of all the components in the PV system, which goes well beyond the common practices in the solar industry space. The Module Quality Specification, which is part of the initial vendor qualification, was developed with input from over half a dozen Tier-1 module suppliers and standardizes well-defined module quality requirements across all vendors to

6 6 SolarCity Photovoltaic Modules with 35 Year Useful Life ensure consistency and highest quality in SolarCity s products. Product Qualification Program (PQP) Testing and Ongoing Reliability Testing (ORT) are implemented in parallel to ensure that performance parameters achieved on the first set of modules qualified are reliably sustained over time. PQP and ORT testing are performed for every single BOM variation per module supplier. The extended testing conditions of the Product Qualification significantly exceed industry common test standards. Besides extended test durations and exposure conditions for wellestablished tests, SolarCity also requires an additional salt-mist corrosion certification for materials, as well as reliable PID resistance under most aggressive test conditions for up to 600 hours. Furthermore, in order to obtain product certification, module suppliers are required to not only meet UL (Underwriter Laboratories) PV module standards, but at the same time also IEC (International Electrotechnical Commission) standards. This leads to additional quality enhancements, since IEC certifications involve performance testing, whereas UL certification is limited to safety-related requirements only. 2.3 Effective Prevention of Quality Deviations This distinctive Quality and Reliability (Q&R) program has led module suppliers to improve their product quality when they produce SolarCity modules, which over time has raised the quality of products delivered to SolarCity and helped to push excellence in the entire Solar Industry. The clearly defined system of controls and tests guarantees that quality is ensured from the beginning and sustained over time, while new quality deviations are rapidly detected. Monthly ORT testing reveals unforeseen quality problems. Once problems are detected, we have a systematic plan in place to implement corrective actions. Under this program, SolarCity has implemented the requirement to perform an end-of-line (EOL) Electroluminescence (EL) inspection on 100 % of all modules fabricated in order to detect defects such as cracks and micro-cracks across all suppliers. Following input from literature and industry collaborations, guidelines for allowable crack types and size of inactive areas were developed and implemented in order to prevent long-term power loss and risk of hotspot formation. Another feature of the Quality Program is that intended changes to the approved BOM require prior notification and approval by SolarCity, and strict retest requirements are in place for any BOM modification. SolarCity enforces defined change management procedures on each supplier. This feature has proven successful in many instances. In one case, it was detected that a supplier had modified the BOM components without any notification. SolarCity issued a Corrective Action Request (CAR) requiring the supplier to perform adequate qualification testing for this modification and any changes thereafter. Another supplier was found to have modified BOM components without notification, and the consecutive Corrective Action Request and vendor management plan resulted in the implementation of a Global Change Management program at the supplier in order to improve oversight and maintain quality. SolarCity also performs regular factory audits through internal personnel or independent thirdparty auditors to detect issues regarding inconsistent quality management or quality escapes. As an example, inspections at two supplier factories discovered and corrected a decrease in quality standards before the product would have been deployed at large scale in the field. Similarly, factory audits resulted in the request of improvements to four modules suppliers to correct deviations with respect to product and manufacturing quality. In summary, SolarCity has an unparalleled system in place to protect quality and prevent the deployment of faulty modules in the field. 2.4 Constant Refinements and Total Integration The Quality Program is constantly evolving and expanding. Test conditions for initial product qualification and ongoing reliability testing are refined on a regular basis to ensure best possible correlation with real life performance. As a consequence, the tests

7 7 SolarCity Photovoltaic Modules with 35 Year Useful Life are becoming more effective and more efficient at the same time. The latest revision of SolarCity s Module Product Qualification and Ongoing Reliability Test procedures has been refined and optimized in close collaboration with DNV GL, and DNV GL decided to use it as a new test standard and implement the program across all suppliers. In order to guarantee highest possible quality on a system level, the Quality Program has been extended and implemented for key components of the PV system, such as inverters and connectors. Several key features of SolarCity s inverter test program have also been adopted by DNV GL and implemented in their standard test program. 2.5 World-Class Team behind the Scenes SolarCity s Quality team consists of industry-wide recognized experts on technician-, engineer- and PhD-level, with extensive experience in relevant industries such as Solar and Automotive and with background in fields such as Engineering, Physics, Chemistry, Materials Science, and Quality Assurance. The team has repeatedly been recognized by manufacturers for their ability to avoid or detect quality aberrations, as well as to rapidly resolve those by advising and guiding manufacturers in terms of process optimizations and/or materials selection. 3 Useful Life Extrapolation from Accelerated Testing In the following section, it is demonstrated that SolarCity s Total Quality Program is succeeding, and as a result of the strategic supplier engagement, coupled with rigorous quality requirements, SolarCity modules show improved performance and projected lifetime versus other Tier-1 modules tested under identical conditions in similar timeframes. The supporting data has been generated by accelerated testing within the framework of SolarCity s PQP and ORT programs, which was performed by DNV GL PVEL (PV Evolution Labs), a world-renowned, independent certified third-party testing laboratory. 3.1 Ongoing Reliability Testing Overview A central part of SolarCity s Total Quality Program is Ongoing Reliability Testing. Module manufacturers are required to submit a defined number of modules every month, which have randomly been selected from a typical manufacturing line under supervision, for reliability testing by an independent third-party, such as DNV GL. The testing is done according to the IEC standard. The overall test duration is about 16 weeks per batch. An overview of the required test procedures is listed in Table 2. # Test ORT 1 Initial characterization IV, EL, Visual 2 Thermal Cycling TC Damp Heat DH Humidity Freeze TC50/HF10 5 Dynamic Load DML/TC50/HF10 Table 2: Overview of ORT test conditions for monthly quality assurance. Certifications according to standards such as IEC have gained industry-wide acceptance over the last 15 years. The stress tests defined in the standards are short-duration accelerated tests performed at stress levels higher than the operating stress level, so the occurrences of failure modes can be stimulated within reasonable timeframes. The qualification tests constitute a minimum requirement on reliability testing and are a measure for the ability of the module to withstand prolonged exposure in real life use environments. It is widely accepted that these test procedures are appropriate to identify infancy failures and product weaknesses. While the tests prescribed in these standards are not fully adequate to determine the exact working lifetime of a module, the stress conditions prescribed by these standards are, however, derived from real-life outdoor stresses. The climate chamber tests yield an accepted indication of the longevity to be expected, the quality of the materials, and the workmanship of the products. The ORT data shown in the following sections, was obtained from DNV GL PVEL, SolarCity s approved independent third-party testing lab. It summarizes

8 Power degradation [%] 8 SolarCity Photovoltaic Modules with 35 Year Useful Life data from 10 monthly batches each consisting of two modules per test condition from seven SolarCity approved module manufacturers TC DH TC+HF DML+TC+HF ORT Summary IEC Pass Criteria Supplier 1 Supplier 2 Supplier 3 Supplier 4 Supplier 5 Supplier 6 Supplier 7 Figure 2: Summary of ORT data obtained from DNV GL PVEL for accelerated testing on modules from seven Solar- City approved module manufacturers. Each symbol represents the average module power degradation for 10 monthly batches each consisting of two modules per test condition: Thermal Cycling ( TC, 200 cycles; green circles), Damp Heat ( DH, 1000 hours; red triangles), combined Thermal cycling and Humidity Freeze ( TC+HF, 50 thermal cycles followed by 10 humidity freeze cycles; grey diamonds), Dynamic Mechanical Load testing followed by Thermal Cycling and Humidity Freeze ( DML+TC+HF, 1000 mechanical load cycles followed by 50 thermal cycles and 10 Humidity Freeze cycles; orange squares). Figure 2 shows a summary of ORT data for accelerated testing that was performed on modules supplied by seven key SolarCity approved module manufacturers. Each symbol represents the average module power degradation of 10 monthly batches each consisting of two modules per test condition. The median power degradation for all tests and all module suppliers combined is as low as -1.1 % (± 0.1 % standard error) and therefore significantly lower than the pass criteria of -5 % of IEC The modules from all suppliers have a tight distribution across all test conditions, indicating excellent process control and product quality. While still well below the allowed pass criteria, the modules of supplier 7 showed a slightly larger degradation for Thermal Cycling and mechanical load testing than SolarCity considers satisfactory, and the supplier was requested to implement a corrective action plan to improve the reliability performance. 3.2 ORT data Thermal Cycling The industry-standard test to simulate thermal stresses in PV modules as a result of changes of extreme temperatures is Thermal Cycling (TC). PV modules are fabricated using several materials including silicon, metals, polymers, glass, etc. During temperature changes, these materials expand and contract according to their coefficient of thermal expansion (CTE). Therefore, interfaces in the modules are mechanically stressed due to their differences in CTE every time a module heats up during day-night cycles or, what is more, during cycles between cloud coverage and sunlight. For example, copper-based ribbons, which electrically connect neighboring cells in the module, are soldered to the cells made of silicon, and due to a large difference in the CTE of metal and silicon, temperature changes can cause significant mechanical stress to these solder joints. One of the main effects of Thermal Cycling is to simulate the stress on the soldered connections within the module. This may trigger fatigue of the ribbons, interruption of the electric circuitry, cell cracking, and power degradation. The modules are placed in an environmental chamber and subjected to extreme temperature swings from -40 ⁰C to +85 ⁰C for 200 times, while maximum power current is sourced into the panel whenever the temperature exceeds 25 C. Thermal Cycling is considered a key accelerated test. Together with Damp Heat testing, failures due to Thermal Cycling can account for more than 70% of the total failures for c-si modules after accelerated testing. There is no consensus on the acceleration factor of this test due to the dependency on environmental factors, so it is difficult to relate number of cycles to years in the field. However, the interconnection failures seen after TC testing are among the most common failures that are observed in the field. For example, long-term studies of modules in the field of 21 manufacturers have shown that of all failures observed, the highest fraction was due to failed electrical interconnects (as much as 36 %, see Figure 3) [6].

9 9 SolarCity Photovoltaic Modules with 35 Year Useful Life Figure 3: Field study of PV module failures found for various PV modules of 21 manufactures installed in the field for 8 years. The rate is given relative to the total number of failures. Approximately 2% of the entire fleet are predicted to fail after years (do not meet the manufacturer's warranty). [Taken from Ref 5]. Similarly, a study on returns from a fleet of >3 million modules from ~20 manufacturers [14] highlights the significance of Thermal Cycling testing and the fact that this test is suitable to reveal weaknesses of the electrical interconnections. The study found that the majority (~66%) of modules with infancy failures (returned after an average deployment of 5 years), were returned because of problems with electrical interconnections in the laminate (e.g. breaks in the ribbons and solder bonds). Figure 4 shows a statistical overview of Thermal Cycling TC-200 ORT test data that was obtained from DNV GL PVEL, SolarCity s approved independent third-party testing lab. The data is a statistically significant overview of ~350 modules submitted for ORT testing performed at DNV GL PVEL. It compares TC data from ~70 modules fabricated for SolarCity against data of ~280 Non-SolarCity modules. 0.7% better Figure 4: Thermal Cycling data comparing the power change after 200 thermal cycles from -40 C to +85 C of ~70 modules fabricated for SolarCity ( SCTY, green bars and line) against data of ~280 Non-SolarCity modules from an industry mix of module makers ( Non-SCTY, red bars and line). The colored bars represent actual data points, while the lines are Gaussian fits.

10 10 SolarCity Photovoltaic Modules with 35 Year Useful Life The data demonstrates that SolarCity modules show significantly better performance after Thermal Cycling stress testing than their Non-SolarCity counterparts. From the Gaussian fits to the actual data points, it can be seen that the power degradation for SolarCity modules peaks at -1.4 %, which is 0.7 % better than the Non-SolarCity-type modules (-2.1 %), corresponding to an improvement by almost one sigma. A statistical hypothesis test confirmed that the difference in means is statistically significant (p < 0.001). In addition, SolarCity modules show a 35 % tighter distribution than the Non-SolarCity modules, and with a maximum degradation of 4 %, there are no outliers falling beyond the maximum allowable 5 % threshold. In contrast, as much as 6 % of all modules not made for SolarCity show degradation in excess of 5 %, and 3 percent degrade by as much as 7 %. The significantly improved reliability performance of SolarCity modules is attributed to the strict requirements that SolarCity imposes on its modules suppliers with respect to all processes that are related to the electrical interconnections, since this aspect is considered key to reliable long-term performance as explained above. For example, SolarCity successfully imposed 100 % end-of-line electroluminescence (EL) inspection on all of its suppliers to reliably detect problems related to soldering of electrical interconnections. During factory audits, SolarCity s experts place key focus on inspecting all process steps related to the interconnections, and great success has been achieved in detecting and resolving problems with these processes. 3.3 ORT data Damp Heat, Humidity Freeze, and Dynamic Mechanical Load Testing Another industry-standard test is the Damp Heat 1000 (DH-1000) test, which simulates the effects of moisture and humidity effects. In this test, the product is placed in an environmental chamber at 85 C and 85% relative humidity (RH) for 1000 hours. Similar to other tests within the standard certification procedures, there is no consensus as to its acceleration factor and the time of exposure in the field it corresponds to, especially given that there is a strong influence of the climate zones which the modules are deployed in; rather, the test is considered appropriate to exclude short- and near-term problems and indicate a nominal level of safety in the field. The test is useful to evaluate the quality of lamination, which protects the solar cells from humidity ingress. In particular, Damp Heat is a stress test to evaluate the quality of the encapsulant (moisture protection) and test for any degradation due to corrosion. Typically, PV module backsheets and encapsulants do allow water vapor to pass through, which may cause stress on interfacial adhesion and lead to delamination. However, for safe operation, the interfaces in a PV module must remain adhered during the entire product lifetime. The main failure modes triggered by DH testing are backsheet and/or encapsulant adhesion loss resulting in delamination and junction box adhesion loss, both of which can cause safety problems, and other modes are contamination problems, material weaknesses, and electrochemical corrosion. In general, the failure rates for Damp Heat testing appear to have declined during recent years. Nowadays, manufacturers have on-site environmental chambers for the assessment of new products and materials, which is very effective for failure prevention. Additionally, advances in encapsulation materials and the lamination process, as well as better edge sealing methods led to an improved protection against moisture ingress. The graph in Figure 5 shows a statistical overview of Damp Heat ORT test data (1000 hours at 85 C / 85 % relative humidity) that was obtained from DNV GL PVEL. The data is a statistically significant overview of more than 350 modules submitted for ORT testing performed at DNV GL PVEL. It compares DH data from ~70 modules supplied to SolarCity against data of ~280 modules from modules that were not fabricated for SolarCity. For both SCTY- and Non-SCTY-type modules, the degradation after Damp Heat testing is low and only -0.8 % and -0.6 %, respectively. This points to the fact that the encapsulation materials and lamination processes are well controlled. However, for all Solar-

11 Fraction of all modules (%) Fraction of all modules Fraction of all modules (%) Fraction of all modules (%) 11 SolarCity Photovoltaic Modules with 35 Year Useful Life City modules the degradation stays below 3 % after Damp Heat testing, whereas 6 % of Non-SolarCity modules show degradation between 3 and 5 %. At the same time, the SolarCity modules have a significantly tighter distribution. The standard deviation for Non-SolarCity modules is with 1.2 % twice as high as for SolarCity modules (0.6 %), and a statistical hypothesis test confirms that the difference is statistically significant (p < 0.001). These observations indicate a further improvement in process and quality control for the SolarCity modules and an effective prevention of problems due to moisture ingress. 15% As described above, PV modules are not impermeable to water vapor, which can lead to a weakening of interfacial adhesion over time. When moisture present inside of the laminate freezes, ice crystals may cause additional damage to the interfaces in the module and cause delamination. The Humidity Freeze (HF) test is an environmental test designed to determine the module's ability to withstand the effects of high temperatures combined with humidity, Non-SCTY 10% DH1000 SCTY Non-SCTY SCTY followed by extremely low temperatures. PV modules are subjected to temperatures of 85 C and relative humidity of 85 % for 21 hours, which causes 5% partial saturation of the module with water. The modules are then cooled down to -40 C, which causes the moisture to freeze. The modules are subjected to complete -3 cycles -2 in -1the closed 0 1climatic 2 0% Power degradation (%) chamber. 50% 40% Gaussian Fit Non-SCTY SCTY a b c Gaussian Fit Non-SCTY SCTY a b c TC50-HF10 SCTY Non-SCTY SCTY Non-SCTY % Histogram of Delta Pmax, SCTY, Delta Pmax, Non-SCTY Normal 20% Variable SCTY Non-SCTY 10% Mean StDev N % Power degradation (%) Power Degradation (%) 2 45% 40% 35% 30% 25% 20% 15% SCTY Non-SCTY SCTY Non-SCTY Gaussian Fit Non-SCTY SCTY a b c DML Figure 5: Damp Heat data comparing the power change (bottom graph) after 1000 hours of Damp Heat at 85 C and 85 % relative humidity (RH) of ~70 modules fabricated for SolarCity ( SCTY, green bars and line) against data of ~280 Non-SolarCity modules from an industry mix of module makers ( Non-SCTY, red bars and line). The colored bars represent actual data points, while the lines are Gaussian fits. The top graph shows the standard deviations for both types of modules. 10% 5% 0% Power degradation (%) Figure 6: ORT data comparing the power change after combined 50 cycles of Thermal Cycling and 10 Humidity Freeze cycles (top), and Dynamic Mechanical Load testing followed by 50 cycles of Thermal Cycling and 10 Humidity Freeze cycles (bottom) of ~200 modules of SolarCity module suppliers against data of ~200 modules from industry mix of module makers that are not suppliers to SolarCity. Symbols are actual data points, while the lines are Gaussian fits with the following fit parameters: a is the height of the curve's peak, b is the position of the center of the peak, and c (Gaussian RMS width) controls the width of the "bell".

12 12 SolarCity Photovoltaic Modules with 35 Year Useful Life The main failure modes triggered by Humidity Freeze testing are caused by thermo-mechanical stress due to the different thermal coefficients of glass, Silicon, and copper, and consist of delamination, junction box detachment, and/or cell interconnect failures. Further, high-temperature glass corrosion can occur as a result of alkali removal from the glass surface. The freezing of the moisture propagates the corrosion effect deeper into the glass. A porous silica material may form as a result of glass corrosion, which affects the transmission properties of the glass and can cause reduced module power output. From Figure 6 (top), it can be seen that for both SCTY- and Non-SCTY-type modules, the degradation after 10 cycles of Humidity Freeze testing is only -1.3 % and -1.1 %, respectively. This confirms that the encapsulation materials and lamination processes are well controlled, resulting in a high-quality laminate. The degradation in SCTY-type modules is 0.2 % lower than in Non-SCTY-type modules, and similar to Damp Heat testing, the distribution is tighter. The data for SolarCity modules are closer to Gaussian, while the non-scty data deviate from a Gaussian shape and reflect an inhomogeneous distribution, confirming the improved process and quality control for the SolarCity-type modules that was suggested from the Damp Heat results and indicating the effective prevention of problems due to moisture ingress and laminate deficiencies. Dynamic Mechanical Load (DML) testing will be discussed next. Modules in the field are subjected to mechanical stress due to wind and snow loads. The resulting deflections depend on glass thickness, encapsulant and backsheet properties, frame design, as well as temperature and magnitude of the loads. Cycling deflections may result in the formation of cell cracks. In order to simulate the stress caused by wind and snow loads, the modules are first subjected to 1,000 Dynamic Mechanical Load cycles to test for the formation of cracks. However, the output power often stays unaffected unless the crack fully penetrates the metallization on the rear side of the cell. Therefore, the modules are also exposed to 50 Thermal Cycles, which can cause the propagation of cracks that may have formed. Lastly, the modules need to undergo ten Humidity Freeze cycles. The high humidity followed by freezing temperatures causes the cracks to propagate through the cell metallization. Failures that are seen after DML testing are broken glass, cracked cells, and/or damaged electrical interconnect ribbons. The results from Dynamic Mechanical Load testing on the 400 modules submitted for ORT testing to DNV GL PVEL are plotted in the bottom graph of Figure 6. A clear advantage of SolarCity-type modules is evident. After the DML stress testing, the distribution for the degradation of Non-SolarCity-type modules has a maximum at -3.2 %, whereas this value for the SolarCity-type modules is at 0 %. Additionally, the distribution is noticeably tighter for SCTY-type modules with an RMS value of 1.2 compared to 4.0. The excellent mechanical stability confirms the results from above and demonstrates the high quality of SCTY-type modules and their integrity against thermo-mechanical stresses. In Figure 7, electroluminescence (EL) images of a representative SCTY module supplier before DML testing (bottom left), after 1,000 cycles of Dynamic Mechanical Load testing (bottom center), and after the completed DML test sequence DML/TC/HF (bottom right) are shown. The data verifies the mechanical integrity of SCTY-type modules. Even after this aggressive test procedure, the power degradation is as low as -0.8 %, demonstrating the improved mechanical stability of SolarCity s patented Zep Solar Panel Mounting System compared to the conventional mounting system of Non-SolarCity modules.

13 13 SolarCity Photovoltaic Modules with 35 Year Useful Life P max change ( rel. %) Pre stress DML-1000 DML-1000 / TC50 / HF % -0.8% Figure 7: Electroluminescence images and corresponding power degradation of a representative SCTY module supplier before DML testing (left), after 1,000 cycles of Dynamic Mechanical Load testing (center), and after the completed DML test sequence of 1,000 Dynamic Mechanical Load cycles, 50 Thermal Cycles, and 10 Humidity Freeze cycles (right). 3.4 PQP Testing Testing Beyond Standard Qualification Tests The ORT test sequence described above follows the IEC test standard. It is generally understood that these qualification tests are a minimum requirement of reliability tests and appropriate to detect infant mortality failures and anticipate shortterm reliability issues in the field. Passing these standard tests demonstrate the ability of modules to withstand prolonged exposure in general use environments. However, there is a gap with respect to long-term performance prediction, and there is broad consensus in the solar community that the IEC standards allow no conclusions to be made concerning the actual lifetime expectancy for tested products. There is agreement that lifetime depends on the design, the materials, the manufacturing quality, and the use environment under which the product is operated. To address the gap between the standard IEC qualification tests and long-term performance prediction, several global standards development activities are underway, which are primarily based on extending the individual tests of IEC SolarCity follows this methodology and implemented a Product Qualification Program testing sequence that is based on extended IEC tests, with an added electrically biased Damp Heat test to evaluate Potential Induced Degradation (PID) and a sequence of dynamic mechanical load testing followed by Thermal Cycling and humidity freeze. While adding significant confidence that the more demanding test procedures allow to better predict the long-term reliability of the tested modules, there is still no clear understanding of whether the expanded tests trigger realistic failures or instead failures that are not found under realistic operating conditions. Therefore, SolarCity has a program underway to evaluate the validity of the test results of these extended tests and correlate these to the performance in the field under real life conditions. SolarCity has a unique advantage of having direct access to one of the largest networks of installed residential and commercial PV systems in a large variety of use environments. Degradation rates under real-life conditions will be eval-

14 14 SolarCity Photovoltaic Modules with 35 Year Useful Life uated and failure modes observed will be correlated to the ones seen in accelerated testing. 3.5 Product Qualification Program Testing Overview Besides ORT testing discussed above, a second key part of SolarCity s Total Quality Program is the Product Qualification Program Testing. The tests are based on extended IEC tests, with an added electrically biased Damp Heat test to evaluate Potential Induced Degradation (PID) as well as a sequence of dynamic mechanical load testing followed by Thermal Cycling and humidity freeze testing. While not yielding conclusive information about the product life expectancy, the successful passing of such tests still justify the assumption of longer lifetimes. SolarCity believes that the extended upfront testing required for product qualification in conjunction with the continuous Ongoing Reliability Testing in regular intervals, constitute a quality program that is industry-leading and appropriate to justify module Useful Life of 35 years. # Test PQP 1 Thermal Cycling TC Damp Heat DH Humidity Freeze TC50/HF10 (3x) 4 UV Exposure 90 kwh 5 Dynamic Mechanical Load / Thermal Cycling/ Hum. Freeze 6 PID testing 1000 cycles (1440 Pa) / TC50 / HF10 Both Polarities; C/85% Table 3: Overview of PQP test conditions for qualification of initial products or modified BOMs. For initial product qualification and after any changes to the Bill of Materials, module manufacturers are required to submit a defined number of modules, which have randomly been selected from a typical manufacturing line under supervision, for reliability testing by an independent third-party, such as DNV GL PVEL. An overview of the required test procedures is listed in Table Product Qualification Program Testing Extended Thermal Cycling As mentioned above, the tests that show the largest effect on PV module performance and appearance are Temperature Cycling tests as well as Damp Heat testing. It has been shown that Thermal Cycling with injected current is an appropriate test to reveal design weaknesses and identify early failures of cell interconnect ribbons and solder bonds [15, 16]. However, it is generally understood that the 200 thermal cycles from IEC testing are not sufficient to give confidence in a module lifetime of more than 25 years [17, 18]. While there is evidence that longer Thermal Cycling is a more adequate test to ensure long-term reliability and reduce field failures, there is no consensus as to how many cycles correspond to what lifetime, especially given that there may be significant variations with climate and use conditions. Studies on an extended number of modules have shown that with an increasing number of thermal cycles beyond the standard of 200 cycles, problems with electrical interconnects between the cells can occur, as is evident from dark areas in electroluminescence images. In general, the degree of damage will get more severe and the output power will decrease with increasing number of cycles. The increased number of cells with broken busbars can lead to an inhomogeneous current distribution between the cells. This can lead to serious safety problems, since high temperatures or even hot spots and arcing can occur. In Figure 8, the relative output power after a consecutive sequence of 200, 400, 600, and 800 Thermal Cycles is plotted for a representative mix of SolarCity-type modules (green line with filled circles; median and standard error) and compared against a statistically significant mix of Non-SolarCity-type modules (66 %ile: black line; 50 %ile: grey line; 33 %ile: light grey line) that have been measured at DNV GL PVEL over time. The modules discussed here are the collection of modules that were submitted by module suppliers to DNV GL for PQP testing, where SolarCity-type modules were fabricated for SolarCity, while Non-SolarCity-type modules made

15 15 SolarCity Photovoltaic Modules with 35 Year Useful Life up the rest of the collection and are referred to as Industry mix. Additionally, data from a selection of modules from a comprehensive literature study is shown (red dashed lines) [14]. As can be seen from the plot, the median degradation for a representative mix of modules from SolarCity suppliers after this extensive stress test is as low as 2 %. The comparison with a large number of Non-SolarCity-type modules demonstrates that SolarCity-type modules perform at least as well as the very best modules in industry and outperform the vast majority of their industry counterparts. After 800 cycles, DNV GL PVEL s 66 percentile for the average degradation is 1 % higher. Figure 8: Relative output power after a consecutive series of 200, 400, 600, and 800 Thermal Cycles from -40 C to +85 C for a representative mix of SolarCity-type modules (green line with filled circles; median and standard error) and a statistically significant mix of Non-SolarCity-type modules (66 %ile: black line; 50 %ile: grey line; 33 %ile: light grey line) that have been measured at DNV GL PVEL over time. Additionally, data from a comprehensive literature study is shown (red dashed lines) [14]. This important result further strengthens the findings from section 3.2. Even when subjected to one of the most aggressive test procedures currently used in industry, SolarCity-type modules are resilient to metallization-related problems and do not show any gridline interruptions or problems with the electrical interconnects, which have been shown to be one of the most prevalent failure modes that have been observed in modules in the field [6]. Figure 9 visualizes this finding. Electroluminescence images and corresponding power degradation of modules from a representative SCTY supplier before Thermal Cycling testing (left), after 400 Thermal cycles (center), and after 800 Thermal cycles (right). It can be seen that the electrical interconnects and gridlines are intact, and the degradation after 800 cycles is as low as -1.3 %. For comparison, EL images of modules from the literature [14] after 200 thermal cycles (left), 400 thermal cycles (center), and after 600 thermal cycles (right) are shown in Figure 10, and dark areas appearing after >400 cycles indicate disconnection of busbars (red markers). The insignificant degradation after extended TC is credited to the strict requirements that SolarCity

16 16 SolarCity Photovoltaic Modules with 35 Year Useful Life imposes on its modules suppliers with respect to all processes that are related to the electrical interconnections. As mentioned above, SolarCity successfully imposed 100 % end-of-line electroluminescence (EL) inspection on all of its suppliers to reliably detect problems related to soldering of electrical interconnections. During factory audits, SolarCity s experts place key focus on inspecting all process steps related to the interconnections, and great success has been achieved in detecting and helping resolve problems with these processes. However, as discussed above, the correlation of defects seen after such extended TC tests and failures occurring in the field has not unambiguously been proven, and it still a matter of debate whether these extensive stress tests might overstress the modules and trigger issues that would not occur in the same way in the field. Nonetheless, the fact that after accelerated TC testing problems commonly affecting a large fraction of modules after extended amounts of time in the field are not observed, justifies the assumption that the degradation rate of SolarCity modules will be at least as low as the industry average for modules of 0.5 % per year [1] if not better, and a Useful Life of 35 years yielding a power output of 82.5 % thereafter appears realistic. P max change ( rel. %) Pre stress TC-400 TC % -1.3% Figure 9: Electroluminescence images and corresponding power degradation of modules from a representative SCTY supplier before Thermal Cycling testing (left), after 400 Thermal cycles (center), and after 800 Thermal cycles (right).