NOVEL C-SI METALLIZATION ADHESION TESTING USING MODULE ASSEMBLY MATERIALS
NOVEL C-SI METALLIZATION ADHESION TESTING USING MODULE ASSEMBLY MATERIALS Rick Lathrop and Eduardo Paz Franklin Advanced Materials 320 Circle of Progress Drive, Suite 102 Pottstown, PA 19464 ABSTRACT: In the crystalline PV module, the integrity of EVA to cell metallization adhesion is imperative for long term reliability. In traditional module layups the c-Si wafers are encapsulated in EVA polymer for environmental protection, optical coupling, shock absorption, and dielectric properties. In addition to glass and back sheet adhesion, the adhesion of the EVA to BSF aluminum and front contact silver metallizations must be robust. With the introduction of very low bow BSF formulations, the BSF to wafer adhesion may be compromised. Further complicating this reliability issue is the lack of an industry standard to test BSF-wafer adhesion. This paper describes and discusses several methods for testing the adhesion of these interfaces using EVA as the adhesive. Measurements of the EVA to BSF adhesion, which can be significantly reduced by excessive micro-pilling or surface dusting, are discussed. This paper will detail the test setups for both nail head type tensile testing and peel strength testing using FPE backsheet for BSF adhesion. Steam aging effects are also explored and reported. Lastly adhesion-bow tradeoffs are discused and best-of-class bow data presented. Keywords: Back-Surface-Field, Metallization, c-Si, EVA, Bow1 INTRODUCTION Failure mechanisms varied from epoxy-BSF surface toThe PV industry lacks established adhesion test methods wafer break. Since the BSF microstructure is fairlyfor BSF metallizations, similar to front and back contact porous under normal circumstances as can be seen insolderability adhesion testing. This presents a challenge Figure 1, the test became questionable as to whether thefor both developers and users of c-Si BSF aluminum heated epoxy was influencing the BSF adhesion bypastes to perform their own due diligence to ensure that actually strengthening the film during cure. The Quadthe BSF has a robust and reliable bond to the wafer Group Inc. states that their epoxy coating goes fromsurface. At the end of the module value chain the bond of enamel-like to water consistency just prior toboth front and back metallizations to the Ethylene Vinyl polymerizing . Further work involved screeningAcetate (EVA) encapsulant is of equal importance. numerous non-heat cure adhesives to replace the epoxySince the sintered aluminum BSF surface is unsolderable, and small nail and stud materials. The best alternateribbon peel tests used to test both front and back contact adhesive found was the Loctite 454 surface insensitivesilver metallizations are not applicable. For other gel cyanoacrylate. However, there were still drawbacksunsolderable thick films such as thick film dielectrics and with this method. The first drawback was that thisglazes the Quad Group Inc. devised a tensile pull test adhesive was sensitive to the nail or stud surfaceusing b-staged epoxy coated studs. The epoxy is coated metallurgy. The Loctite 454 adhesive worked well withonly on the head of a precision nail shaped stud. This zinc plated nails but not plain steel. This limited ourstud is then clamped perpendicular to the surface under choice to a nail type with an irregular shaped head, whichtest and cured at 125°C for 10 minutes. When the epoxy produced non-perpendicular bonds and quite a bit ofis cooled, a very high tensile strength bond is formed to variability in the pull data. The second drawback was thethe test surface. For delicate substrates like a silicon question of applicability to steam aged BSF films.wafer, epoxy coated alumina coupons can be adhered to Cyanoacrylate glues cure with humidity and we saw anthe back of the wafer to strengthen the assembly. The increase in adhesion after steam testing the BSF. Thestud is then pulled until the weakest bond is broken and third drawback was the question of how “real world” thispeak tensile force is recorded. This was the first test was. Although there are tensile forces on the BSFgeneration adhesion test developed in-house for BSF. film in the PV module, there are no cyanoacrylates. 2 BSF MEASUREMENT With the beginning of the in-house development of the EVA peel test, a way of implementing EVA as the adhesive in the pull test was the next logical move. EVA samples were acquired from STR Inc. Solar Division. Specifically we are using their Photocap fast cure 15295P/UF EVA formulation. EVA is one of the most popular thermosets for encapsulating the wafer in the module. EVA is also the “real world” material that needs to reliably bond with both the BSF aluminum and the front contact silver. 2.1 EVA Pull Test The EVA is in sheet form, and a small disc would be the obvious form factor for the pull test. To achieve this, a Figure 1: Porous Fired BSF Microstructure hole punch was utilized to punch a 3mm diameter disc
from the EVA film. Figure 2 illustrates this pre-cure the wafer width with the backsheet strip placed directlyassembly. The clip not only provides positive alignment on top of the EVA, justified to one end of the wafer. Aof the nail and the EVA disc but also puts pressure on the semi-rigid piece of Teflon sheet is placed on top of theEVA, similar to a laminator. The assembly is placed in an backsheet followed by a section of stainless steel U-oven and allowed to reach 150°C for 10 minutes. After channel to provide compression during the cure, similarcooling, the clip is removed and the coupon is clamped to to a laminator. The tray with this assembly is placed in a box oven and allowed to reach 150°C for 10 minutes to cure the EVA. The actual time that this assembly is in the oven is about 30 minutes. After cooling, the wafer with the laminated EVA-backsheet strip is placed in the peel test fixture. This fixture is mounted on two linear bearings to ensure a true 180 peel test. The backsheet is peeled at 200mm/min until the end of travel on the test EVA stand is achieved as can be seen in Figure 4. The entire peel test is recorded and plotted (force vs position). The failure mechanisms are recorded. Figure 2: Nail-EVA-Wafer Coupon Assemblythe upper surface of a pair of linear bearings mounted tothe test stand stage. These bearings ensure a pure tensilepull with no shear components to the test. A flat washerwith a 4mm center hole is placed on the upper surface tolimit the pull forces to the immediate area surroundingthe nail head as shown in Figure 3. Use of this washerelliminates the need to strenghten the wafer prior to test.The pull test is performed at a pull rate of 5mm/min untila peak force is recorded. Failure mechanisms are alsonoted. This method has been used to test front contactand both pre and post steam aged BSF adhesion. Backsheet Wafer Figure 4: 180 Degree EVA Peel Test 2.3 BSF Tape Peel Test The third BSF adhesion test is a scotch tape test that is performed similarly to the EVA peel test. Industrial clear 3M adhesive tape (#600) is cut to about 2 1/2 times the wafer width to enable a 180 degree peel. This test Figure 3: EVA Pull Test Fixturing originated from the printed circuit board industry and can show gross failures of the sintered aluminum BSF2.2 EVA BSF Peel Test coating. The tape width is 0.75”. The entire length of theThe EVA peel test has been fashioned after our in-house wafer is covered with this tape. The tape is pressed onfront and back contact silver solder adhesion peel test . with finger pressure, folded back on itself, placed in theSince the cured EVA is highly elastic, it needed to be peel test fixture and peeled at 180 degrees at a rate ofstrengthened with a more rigid material. To accomplish 500m/min until the end of travel on the test stand isthis, Dun-Solar FPE backsheet was acquired from the achieved. The section of the tape that was peeled fromDunmore Corporation. A 4mm wide strip of backsheet the BSF surface is placed on white paper and examinedand EVA are cut with the outer Fluoro layer of the for any lifted aluminum.backsheet marked to indicate that the PE inner layer on The tape test can be considered, for the most part,the other side needs to be mated with the EVA prior to redundant to the EVA peel test but is prefered by some tocure. The length of the backsheet strip is about 2 1/2 be a simple go no-go test. The EVA peel test is a muchtimes the wafer width to enable a 180 degree peel, while more severe adhesion test in that peel force per mm hasthe EVA strip length is cut about 1cm short of the wafer been found to be more than 20 times that of the tape test.width to avoid any EVA flowing onto the tray during thecure process. The wafer under test is placed on a metal 2.4 Bow Measurementtray with the BSF side up; the EVA strip is centered on The aluminum BSF backplane dominates the finished cell
bow behavior. This is largely due to the glass frit binderused in BSF pastes . This glass frit is necessary to An interesting comparison can be seen in Figure 6. Thisprovide adhesion to the wafer surface. The thinner the comparison shows a slightly higher median adhesion ofwafer the more pronounced the bow behavior is. The the LunAl 988-F BSF paste over the bare wafer backnominal wafer thickness is 180 microns. Severe bow surface and a significantly higher median adhesion of acauses handling issues at both cell manufacture and large pad of front contact formulation SunAg 898-L2ribbon attachment during module assembly. over the bare wafer front surface. The bare wafer frontThe in-house procedure developed for measuring bow is surface adhesion is higher than the back presumably duea relatively simple non-contact method. First, a finished to texturing of the front surface. When testing a 1.8mmcell is placed with the front contact side up on a granite wide front contact buss bar, the adhesion is similar to theslab and a laser triangulation sensor is zeroed to the bare wafer surface. This is likely due to the overhang ofcenter of the relaxed cell. Second a large “C” washer is the nail head and EVA onto the bare wafer. Equal orplaced around the center to compress the bow as in better EVA adhesion to the wafer surfaces of theseFigure 5. The weight of this washer is 105 grams. The metallization films is considered excellent.distance reading on the sensor is the bow in millimeters. 3.2 EVA and Tape Peel Results The plot below (Figure 7) shows a consistent peel in the 25 to 35 Newton range with no removed BSF in the picture overlay. The strip of backsheet is approximately 4mm wide. The tape test (Figure 8) shows a very consistent peel in the 10 Newton range. The tape is 19mm wide. Laser Spot “C” Washer Figure 5: Cell Bow Measurements3 RESULTSThe results reported in this section are in box plot form Figure 7: BSF EVA Peel Testfor pull tests and force-displacement curves for peel tests.BSF data is from the LunAl aluminum paste series andfront contact data from the SunAg paste series. Waferswere 156mm multicrystalline with a nominal thickness of180 microns. Peak firing temperatures were in the 775°C-790°C range.3.1 EVA Pull Results Figure 8: BSF Tape Peel Test 3.3 Steam Aging Effects Although cells are encapsulated with EVA, moisture intrusion into the module and its negative effects on reliability are of concern. The porous nature of the BSF microstructure, and thus the potential vulnerability to moisture, has given rise to various methods of accelerating the long-term effects of moisture. Typically the accelerant is temperature in the form of damp heat , steam or boiling DI water. As mentioned earlier, the cyanoacrylate adhesive was discounted for steam aged BSF adhesion testing due to its curing method (humidity) and the trend of increasing BSF adhesion after steam testing. Figure 9 shows that the EVA adhesion testing also increases in both steam and boiling DI water. The strengthening of the BSF Figure 6: EVA Pull Adhesion film seems to be independent of both the adhesive used and the accelerant method.
small sections of BSF ripped off the wafer when peel forces exceeded the materials adhesion to the wafer. Once a section was removed, peel forces were relieved then would build up to a failure level and a new section would be removed. This cycle continued throughout the test as can be seen in Figure 11. The glass containing paste retained all of the material on the wafer as can be seen at the bottom in Figure 12. Figure 11: BSF Adhesion Failure Figure 9: Effects of Moisture on BSF Adhesion4 BALANCING BOW AND ADHESIONCommon to all thick film formulations is balancing keyproperties to optimize the overall performance of theproduct. Key properties of BSF aluminum paste are waferbow and adhesion. Figure 10 illustrates this point well. Figure 12: Glass-Free Formula on TopGlass A is driving adhesion up at the expense of bow.Glass B is driving bow down at the expense of adhesion.A blend of these glasses provides optimization of both of 4.2 Low Bow Formulationthese key properties. By moving to an optimized blend of glass from a single glass formula, and careful selection of the aluminum source and powder specifications the current generation II BSF metallization paste (LunAl 988-F) demonstrates best-in-class low bow. Figure 13 shows this performance against generation I and a leading competitor at three different firing temperatures. Figure 10: BSF Glass Formulation Optimization4.1 Enough AdhesionSince there is a tradeoff between adhesion and bow andthere are no industry standards for adhesion, how muchadhesion is enough?Two experimental BSF pastes were created, one with noglass and one with glass in the normal range of 2-5%.The tape test was performed with little contrast betweenthe two formulas. The peeled tape was slightly darkerwith the glass-free formula but too similar to the glasscontaining formula to be photographed for this paper. Figure 13: Best-in-Class BowThe EVA peel test, however, showed a significantcontrast in failure modes. With the glass-free material,
6 CONCLUSIONS • EVA is an ideal real world adhesive for both tensile pull and peel testing. For peel testing it must be strengthened with backsheet. • Steam or boiling water conditioning increases BSF adhesion using either cyanoacrylate or EVA adhesives. • Glass frit is required to give adequate adhesion but can have a negative impact on wafer bow. • Careful selection and characterization of the glass or glass blends can optimize bow and adhesion BSF properties. • Best-in-class sub half millimeter total bow can be achieved on 125 monocrystalline wafers with diligent formulation and material selection.7 REFERENCES Quad Group Inc. website, “Stud Pull Tests”, http://www.quadgroupinc.com/studpull.html Lathrop et al: “Novel Approaches to Benchmarking Solar Cell Tabbing Solderability”, Proceedings 26th EU PVSEC, 2011 Hamburg Carroll et al: “Advances in PV Metallisation Technology”, Proceedings 20th EU PVSEC, 2005 Barcelona Ketola et al: “Degradation Mechanism Investigation of Extended Damp Heat Aged PV Modules”, Proceedings 26th EU PVSEC, 2011 Hamburg