21st European Photovoltaic Solar Energy Conference, 4-8 September 2006, Dresden, Germany DAMP HEAT DEGRADATION OF CIGS-BASED PV MODULES P-O. Westin, P. Neretnieks and M. Edoff Ångström Solar Center, University of Uppsala, P.O.Box 534, SE-75121 Uppsala, Sweden Corresponding author: per-oskar.westin@angstrom.uu.se, Phone +46-18-4713149, Fax +46-18-555095 ABSTRACT: Modules from the micro-pilot line at Ångström Solar Center were tested under the IEC 61646 conditions for damp heat testing, i.e. 85°C and 85% relative humidity for 1000h. Testing was made on both unencapsulated and encapsulated modules, applying conventional glass/EVA/glass encapsulation. Alternative encapsulation, such as replacing EVA with PVB, or adding an edge seal to the packaging was also evaluated in damp heat testing. Degradation of the modules was monitored and it was shown how conventional encapsulation leads to a reduced degradation that concentrates on FF losses suspected to be related to the window/CIGS interface. PVB lamination was not successful, which may be related to unrefined lamination process. Adding an edge seal to the module packaging effectively reduced the moisture ingress into the modules, showing efficiency losses within the 5% limit for 1000h. Keywords: Cu(InGa)Se2, Encapsulation, Degradation 1 INTRODUCTION Commercial solar cell module manufacturers offer long-term product warranties in the range of 20-30 yrs. To ascertain long-term stability without actual field data, several types of accelerated tests are employed. An important tool in the qualification of solar cell modules is the Damp Heat test, where modules are placed in a humidity and temperature controlled environment, for the duration of 1000h. According to the IEC61646 standard modules must degrade no more than 5% in relative output efficiency during 1000h at 85°C and 85% Relative Humidity to pass the DH test. The CIGS technology is one of the most promising new technologies for achieving low-cost, high-efficiency solar cell modules. A challenge for CIGS modules is to prove long-term stability comparable to existing technologies. Several investigators have shown how conventional ethylene vinyl acetate, or EVA encapsulation schemes fall short, and successful results of damp heat testing have only been achieved with “alternative materials” or “modified encapsulation schemes” [1,2]. Such modified encapsulations are now used, and CIGS or CIS modules that pass IEC 61646 are commercially available [3]. Previous ÅSC work has primarily regarded degradation of single cells [4,5,6,7] and modules without encapsulation [8]. Results have shown how cells degrade mainly by Voc and FF losses [4], and that modules suffer additional FF losses, associated with series resistance increase in the front contact and interconnect zones [8]. In the present work the damp heat behavior of CIGS modules is presented, evaluating stability of modules from our micro pilot line. The CIGS deposition was made using an evaporation system that approaches an in-line industrial process but works on a small scale using 5” glass substrates (for more info see [9]). The impact of encapsulation on module stability is discussed. In order to achieve successful results of the qualification tests, experimentation is underway with regards to materials selection and design of the module packaging. As an alternative material to EVA for encapsulation, poly-vinyl butyral, PVB, has been suggested [10,11]. PVB has been used for a long time in the automotive industry and therefore has a proven UV-stability in the field, with comparable transmission properties and good adhesion to glass [10]. Its main disadvantage is long lamination cycle 2470 times, although this can be improved [11]. An advantage of using PVB would be that it already complies with certain building safety regulations, facilitating market deployment [12]. 2 EXPERIMENTAL Modules were prepared in the ÅSC lab-line, described elsewhere [9]. Edge deletion was performed with a bead-blast setup using SiO2 beads (44-74 m), isolating a ~1.5 cm perimeter around the cells. Modules consist of 16 cells with a 75 cm2 active area, contacted by tab-wires that were soldered or attached using a conductive adhesive. Lamination was performed in a PANAMAC® vacuum laminator. A “standard” module is laminated using an ETIMEX® fast-cure EVA, with a cycle time of approx 20 min and Tcure of 150°C. Along with standard modules, several experimental encapsulations have been tested using alternative materials, e.g. PVB from TROSIFOL®, or by using an edge sealant. Furthermore, interconnect test structures (ITS) were prepared, consisting of front and back contact sections with interconnects that circumvented the CIGS p-n junction (see ref [13]). The purpose of these structures was to isolate, evaluate and monitor the degradation of series resistance in modules. In all aspects except the position of interconnect zones, the ITS were therefore processed as modules, including buffers, edge deletion and wiring. Individual front and back contact films were also prepared by sputtering onto glass substrates. Damp heat testing of modules and ITSs was performed in a Vötsch V2020 climate chamber. The modules were removed at different intervals and cooled in ambient temperature (for at least 15 min) before testing. No light soaking was performed prior to testing. Module I/V-characteristics were examined using a Quicksun® flash solar simulator setup. ITS structures were examined for series resistance, and the contact films for sheet resistance. Unencapsulated modules were also tested at a lower relative humidity, at 85°C and 65% RH, and both encapsulated and unencapsulated modules were tested at 85°C in a standard oven as a “dry” reference. 21st European Photovoltaic Solar Energy Conference, 4-8 September 2006, Dresden, Germany RESULTS 3.1 Unencapsulated modules As in previous investigations, unencapsulated modules (E and F) exhibit a sharp drop in Voc, losing 15% during the initial 60 hrs of the test (see Figure 1). The module voltage then remains stable for the duration of the test. FF losses show a similar behavior, rapidly dropping 20-30% depending on the humidity conditions, the 85%RH conditions imposing more severe losses than 65% RH. After these initial losses a steady but much slower degradation is observed throughout the test (see Figure 2). Higher humidity in the test chamber induces slightly more severe FF degradation. Short circuit current shows no degradation during the initial 100 hrs of testing followed by a slow and steady decrease throughout the testing. As expected [1,7], dry heat testing shows steady performance of the modules (D), underlining the importance of humidity in the degradative processes of CIGS based modules. 3.2 Encapsulated modules Encapsulating the modules (A and C) improves the resistance to damp heat induced degradation. Losses are observed only after 500 hrs of testing, and essentially only as FF losses (see Figure 2). Hence, encapsulation hinders degradation for a period of time, after which a point is reached when the encapsulation no longer withstands the environmental attack. The degradation that follows is not manifested by the same losses as that of bare modules. This is discussed further in the following section. Dry heat testing showed stable performance for over 1000h. Using a modified encapsulation scheme (B), with an edge seal around the perimeter of the module, CIGS modules pass the 1000h limit with less than 5% loss in efficiency at 85°C and 85% RH (Sample B, maintains all I/V-characteristics throughout the test. See Figure 1 and 3, and note that current was maintained for all encapsulated samples). Normalised Voc 1,2 1,1 A 1 B 0,9 C 0,8 D 0,7 E 0,6 F 0,5 0,4 0 200 400 600 Time [h] 800 1000 Figure 1. Voc development during 1000h of exposure to different conditions: A, D – Dry 85°C; B, C, F – 85°C and 85% RH; E – 85°C and 65% RH. Filled symbols are standard encapsulated modules, outlined symbols are unencapsulated, and X are encapsulated with an edge seal. 1,2 1,1 Normalised FF 3 A 1 B 0,9 C 0,8 D 0,7 0,6 E 0,5 F 0,4 0 200 400 600 Time [h] 800 1000 Figure 2. FF development during 1000h of exposure to different conditions: A, D – Dry 85°C; B, C, F – 85°C and 85% RH; E – 85°C and 65% RH. Filled symbols are standard encapsulated modules, outlined symbols are unencapsulated, and X are encapsulated with an edge seal. 3.3 Contacts and ITS The front contact ZnO film showed a moderate but slow increase in sheet resistance, reaching a 50% increase at the end of 1000 hrs, while the Mo back contact sheet remained conductive, even with strong visible evidence of corrosive attack on the Mo surface, for between 100200 hrs after which corrosion had penetrated through the 450nm film and resistance increased catastrophically. Similarly, the ITSs displayed a catastrophic increase in series resistance after 400 hrs of damp heat exposure (see Figure 3). This coincided with P3 isolation scribes becoming completely transparent. Loss of connection in the P3 scribe area occurred also for several of the modules. Investigation by optical microscope of the P3 scribe during the damp heat treatment showed how the surface attack is most pronounced along the center of the scribe line where the mechanical tip has damaged the Mo surface (see Figure 4). Adjacent Mo which has been exposed through chipping of the CIGS/ZnO layers does not exhibit the same level of corrosion. By correlating the observed values for ITS series resistance and contact film sheet resistance, an approximation of the P2 interconnect contact resistance (Rc) can be calculated. These results give an approximation of interconnect resistance degradation during damp heat testing. Plotting these results show how the interconnect contribution to series resistance starts out as modest but goes on to be the dominant factor after a few hundred hours of damp heat exposure (see Figure 3). Even when correcting for increased window resistance caused by underlying buffer layer [4] (dashed lines), the contact resistance dominates after a few hundred hours. During this investigation, ITSs have been subject only to unencapsulated damp heat testing. Testing of encapsulated ITSs is underway. 2471 21st European Photovoltaic Solar Energy Conference, 4-8 September 2006, Dresden, Germany 4 40 Rtot Rc Rc (CdS) ZnO R (CdS) ZnO R Resistances [Ohms] 35 30 25 20 15 10 5 0 0 100 200 300 Time [h 400 500 Figure 3. The development of resistances during damp heat treatment of contact films and ITSs. Rtot is the total resistance, R ZnO is the contribution by the front contact, calculated from sheet resistance measurements of ZnO films. RC is the derived ZnO/Mo connection resistance. Correction for increased ZnO resistance in the presence of underlying buffer layer is shown as dashed lines. Figure 4. Corrosion of the Mo back contact in the P3 isolation scribe is mainly concentrated to the area where the mechanical tip has damaged the surface. 3.4 PVB results I/V-measurements do not show any conclusive difference between the EVA and PVB laminated modules regarding module performance and efficiency. Evidence of FF degradation appears after 500 hrs of DH testing for all modules. However, series resistance measurements show how the PVB laminated modules reach considerably higher values for series resistance towards the end of the test. Visual inspection reveals how the PVB modules exhibit a discoloration phenomenon, with clouding of the films in a circular pattern from the edges of the module toward the middle. PVB modules also exhibit brown-colored substance around the tab wires indicative of corrosion. On the other hand, modules laminated with EVA remained visually clear but exhibited corrosion along the P3 scribe for approximately 5 mm into the module. These differences are discussed briefly below. 2472 DISCUSSION 4.1 Unencapsulated module degradation It has been shown previously that the degradation of unencapsulated CIGS PV modules is more severe than that of single cells [8]. Concerning cells, degradation of the CIGS absorber layer and/or the CIGS/CdS interface has been put forward as the cause of the observed rapid losses in Voc and FF [14,15]. In particular, Schmidt et al. suggested that losses in FF would be principally caused by changes in the front window and CIGS/window interface, while Voc losses are related mainly to degradation in the bulk of the absorber layer [14]. Additional FF losses in the modules are suggested to be caused by increased series resistance in the front contact layer and interconnects [8]. In this work, the modules that were placed in 65% RH and 85°C showed slightly less degradation than those in 85%RH. This difference was shown in the FF losses (see Figure 2) whereas Voc behavior was similar for both conditions. Modules in dry heat did not degrade to a significant degree, and this, yet again, underlines the importance of humidity in CIGS module degradation. 4.2 Encapsulation Performance degradation is observed in all modules that were encapsulated using only standard glass/EVA/ substrate configurations. This degradation set on after around 400-500 hrs of DH testing suggesting either a breakdown of the encapsulant, such as loss of adhesive strength which can follow exposure to damp heat [16], or a saturation phenomenon with humidity ingress reaching a critical level. Encapsulated module degradation was limited to FF losses and increased series resistance. The pronounced drop in Voc that is characteristic for unencapsulated modules and cells is not present which would suggest that the degradation that occurs of encapsulated modules is by a different mechanism. Following the reasoning of Schmidt et al.[14] it is possible that this degradation is limited to the front window layer, causing not only an increased series resistance but a shift in the electric field, severely deteriorating the FF. The relative stability of the Voc would be explained by stable absorber bulk. Doubtlessly, there is also a considerable contribution to FF degradation from increased series resistance in the interconnects. Three different reasons why degradation could differ between encapsulated and unencapsulated modules seem possible: 1) Difference in RH at the module surface, 2) difference in oxygen-to-water ratio, and 3) encapsulant material interactions, for example the degradation of EVA into acetic acid. If difference in humidity is the only factor, then it must be more pronounced than that of lowering RH to 65%, as this investigation has shown. Further investigation into the behavior of CIGS cells and modules in differently composed environmental conditions would help in understanding this. The similarity in the degradation of modules laminated using EVA and PVB would suggest that encapsulant chemistry is not a primary cause for the losses, but may come into play for secondary effects, namely the corrosion of tab wires and Mo surfaces. In any case, this behavior looks promising in that module stability may be improved by alterations in the front contact and interconnects, without manipulating the absorber layer. It should also be mentioned that no 21st European Photovoltaic Solar Energy Conference, 4-8 September 2006, Dresden, Germany particular efforts were made in cleaning or priming the edge deleted regions to improve adhesion of encapsulant. Further investigation will include adhesive properties of the encapsulant in relation to processing and damp heat. 4.3 Alternative encapsulation The initial tests using PVB as a module encapsulant show discouraging results. No particular gain was seen with regards to the electronic performance of the modules, and visual inspection revealed both clouding of the PVB in the aperture area and brown corrosion products precipitated along the tab wires. This in spite of strong opinions that these types of problems were a thing of the past with up-to-date PVB [12]. However, there are PV module manufacturers that use PVB as an encapsulant for commercial modules (e.g. ref [17]) and it is not unlikely that the problems encountered in this investigation will be helped with improved processing and handling of the PVB films prior to lamination. The observed P3 corrosion in EVA laminated modules that was not present in the PVB modules suggest a negative interaction between the EVA and the Mo film that does not affect the performance of the module, or is overshadowed by other degradation mechanisms. Possibly acetic acid from the EVA could be the cause for this. Adding an edge seal to the conventional EVA encapsulation we were able to produce modules that pass the criteria of the EN 61646 standard for damp heat degradation (see Figure 2). This will not hinder our efforts to further investigate the stability of our modules and the influence of the encapsulant on long-term stability. For example improved interconnect stability and optimization of lamination conditions for different laminates. 4.4 Contacts and ITS The molybdenum layer, and consequently the P3 interconnect, retains its low resistivity although severe corrosion is observed visually. Only when the film is completely penetrated by corrosion does the resistivity increase, but then it is a catastrophic rise, and connection between adjacent cells is broken. During the mechanical patterning step the Mo surface was damaged by scratching, increasing its sensitivity to corrosion. This has previously been observed [2,18] and at present the mechanical scribing step has been improved to cause less damage to the underlying Mo contact. The results of series resistance investigation using ITSs show how the interconnect resistance in its initial state contributes only modestly to the resistance of the entire structure. During damp heat treatment this relation shifts and after a few hundred hours the contact resistance ZnO:Al/Mo becomes the dominant factor. 5 characteristic Voc drop when encapsulated modules degrade. This could be explained by encapsulant material interaction or changes in the humidity and oxygen levels. The interconnect contact resistance was shown to start out as inferior to the contribution of the front contact, but increased during damp heat treatment to become dominating after a few hundred hours. 6 ACKNOWLEDGEMENTS This work is funded by the Swedish Energy Agency. Many thanks to M. Ruth for providing the CIGS for the modules and ITSs investigated. The authors would also like to thank U.Malm and Dr. U. Zimmermann for valuable insights, K. Ottoson for test collaboration and the entire ÅSC and Solibro teams for cooperation and support. [1] F. Karg, H. Calwer, J. Rimmasch, V. Probst, W. Riedl, W. Stetter, H. Vogt, and M. Lampert, Inst. Phys. Conf. 152 (1998) p909-913 [2] M. Powalla, B. Dimmler, Thin Solid Films 387 (2001) p251-256. [3] http://www.wuerth-solar.de [4] J. Wennerbeg, J. Kessler, M. Bodegard and L. Stolt, in Proceedings of the 2nd (European) WCPVEC (1998) p1161-1164. [5] U. Malm, L. Stolt, in Proceedings of the 20th EUPVSEC (2005) [6] U. Malm, M. Edoff, L. Stolt, Proceedings of the 19th EUPVSEC (2004), p1890-1893. [7] J. Malmström, J. Wennerberg, L. Stolt, Thin Solid Films 431-432 (2003), p436-442. [8] J. Wennerberg, J. Kessler, L. Stolt, Solar Energy Materials & Solar Cells 67 (2001), p59-65 [9] M. Edoff, S. Woldegiorgis, P Neretnieks, M Ruth, J Kessler and L Stolt, in Proceedings of the 19th European Pfotovoltaics Specialists Conference (2004) p1690-1693. [10] Koll, B, Presented at The Glass Processing Days 2003, http://www.glassfiles.com/library/37/article683.htm, (2003), was available 2006-08-22 [11] H. Schmidhuber, K. Krannich, Proceedings of 17th EUPVSEC (2001) [12] K.H. Diefenbach, Online article @ http://www.photonmagazine.com/news/news_200504%20ww%20feat%20PVB%20foils.htm (2005) was available 2006-08-16 [13] J. Palm, V. Probst, T. Nielsen, R. Tölle, S. Visbeck, H. Vogt, H. Calwer, M. Wendl, W. Stetter, Proceedings of 19th EUPVSEC (2004) p1955-1958 [14] M. Schmidt, D. Braunger, R. Schäffler, H.W. Schock, U. Rau, Thin Solid Films 361-362 (2000) p283-287 [15] M. Igalson, M. Wimbor, J. Wennerberg, Thin Solid Films 403-404 (2002) p320-324. [16] T.J. McMahon, G.J. Jorgensen, NCPV Program Review Meeting (2001) [17] http://www.wulfmeiersolar.de/ [18] J. Wennerberg, J. Kessler, L. Stolt, Solar Energy Materials & Solar Cells 72 (2003) p47-55 CONCLUSION In this investigation encapsulation and damp heat testing of CIGS-based PV modules has been performed. It was shown how encapsulation improves the stability of the modules but that addition of an edge seal was necessary in order to pass 1000h with less than 5% performance loss. Encapsulating modules appears to change the conditions for degradation, reflected by the absence of the 2473