High Reliablity of Conductive Adhensives For Thin-film Interconnects

PV PRODUCTION FORUM 2010 HIGH RELIABILITIY OF CONDUCTIVE ADHESIVES FOR THIN-FILM INTERCONNECTS Rich Wells and Mark Francis Engineered Conductive Materials 101 Johnson Dr. Delaware, OH 43015 ABSTRACT: With the solar industry focusing on cost reduction and efficiency improvements, the reliability of the cell interconnect is often taken for granted. The c-Si technology has relied on solder as the cell interconnect, but typically the thin- film industry is precluded from high temperature solders because of material temperature limitations. Electricallyconductive adhesives (ECAs) have emerged as the popular choice for interconnecting solar cells with tin and tin-silver coated ribbon in the thin-film industry. ECAs can be formulated to exhibit rubber-like flexibility which reduces stress in the bond line, improves thermal cycling performance, mitigates the risk of delamination, and increases the overall peel strength of the assembly. Cell interconnects must also exhibit exceptional humidity resistance. Protecting non-noble metals from corrosion has been an elusive task for ECAs, since high humidity conditioning can induce corrosion resulting in electrical resistance increases at the adhesive-ribbon interface. Stable electrical performance after 1000 hours at 85˚C/85%RH on tin and tin-silver coated ribbon has been demonstrated. ECAs with stable contact resistance to nonnoble metals with rubber-like flexibility offer a robust cell interconnect for the thin-film industry for both rigid and flexible solar panels. 1 INTRODUCTION two difficult reliability requirements became apparent. First, the adhesive is required to pass a drop/impact test. In order to pass the test, the ECA needed to possess dampening or low modulus characteristics to dissipate the stress from impact. Rigid, high strength adhesives would typically delaminate upon impact. Second, the adhesive must maintain a stable electrical resistance after damp heat conditioning which is typically conducted at 85˚C/85%RH. Since many electronic applications specified low cost, non-noble metals such as tin and tin alloys, the adhesive-metal interface was susceptible to humidity induced corrosion resulting in unstable, increasing contact resistance [5]. 2.2 ECA as Thin Film Interconnect Historically in the c-Si solar industry, the cell interconnect was formed with high temperature solders. The bus bar on the top side of the cell is soldered to a metal alloy ribbon. The ribbon is then soldered to the back side of the adjacent c-Si cell. Depending on the design, many cells are strung together by soldered ribbons. ECA has become a popular choice for thin-film applications for various reasons. As mentioned because of material temperature limitations, high temperature solder is precluded from use as the interconnect. Stresses created by high temperature solders can lead stress cracking of rigid substrates. ECAs create less stress and provide improved survival of thermal cycling and high temperature conditioning. In addition, many thin-film technologies enable flexible designs that could only be made possible with a flexible interconnect such as a flexible, rubber-like ECA. 3 RELIABILITY CONCERNS 3.1 Potential Corrosion Low stress polymers, that are essential to minimizing stress, intrinsically possess a low glass transition temperature, Tg, and a high polymeric free volume atApproaching grid parity with low cost solar energy solutions is the impetus of established large companies as well as numerous start-up companies. Crystalline silicon (c-Si) technology has been utilized for solar energy for several years. However in the last decade, thin-film technology has emerged as a low cost alternative offering potentially lower material and manufacturing costs. [1]. In addition, thin-film solar technology enables new applications where flexibility, low weight, and lower profiles can be realized. Research in thin-film has focused on cell level aspects such as cell efficiency, manufacturing costs, TCO optimizations, and deposition techniques. Of equal importance is the reliability of the cell interconnect. Because thin-film technology requires lower processing temperatures, the historical soldering approach has been displaced with electrically-conductive adhesive (ECA). For example, the thin-film technology CIGS (Copper Indium, Gallium, and Selenium) should not undergo high temperature exposure to maximize efficiency [2]. Since the solar industry has been assuring customers reliability for up to 25 years, engineers are faced with a challenge of utilizing accelerated aging tests to evaluate failures with materials such as the interconnect. Since thin-film technology is still in its infancy, long term data is unavailable. In addition to the IEC requirements for module testing, additional accelerated testing can be employed to test the limits of an ECA and provide additional certainty for the long term reliability of solar modules. 2 CONDUCTIVE ADHESIVES2.1 Solder Replace History ECAs were identified as potential solder replacements for surface mounting of electronic components in the 1990s. Benefits such as lower process temperatures, improved mechanical fatigue resistance, and elimination of lead content from the tin-lead solder were driving a potential switch to ECAs [3, 4]. However,Dimensional Change (expansion/shrinkage)Sn → Sn2+ + 2e−(1)The silver filler in the ECA typically acts as a cathode and the reduction of water occurs at the cathode forming hyrdroxide ions:2Η 2Ο + Ο2 + 4e − → 4OH −(2)Expansion of liquidsTime Cure shrinkage Cooling shrinkage (<Tg)Finally, the hydroxide ions react with tin, and oxidation of tin to tin oxide occurs resulting in an increase in the interfacial resistance of the ECA-ribbon joint [7, 8].Cooling shrinkage (>Tg)Sn + 2OH → Sn(OH)2 → TinOxide2+−(3)Figure 1. Strain development or dimension change during the cure and cool of the ECA Depending on the materials and geometries, the resultant interfacial stress can contribute to delamination, cracking in the substrate or adhesive bondline, and potentially curling or deflection of the substrates. These effects may not manifest at ambient temperature, but further cooling to lower temperatures will amplify the interfacial stresses present at ambient. As long as the expansion coefficients of adhesive and substrate are different, cooling will increase the net interfacial strain and resulting stress. However, a lower interfacial stress at ambient will certainly deliver lower stress when the assembly is cooled to sub-ambient temperatures. Defining what is meant by “low stress” is always going to be a difficult challenge. Complex stress models that utilize physical property data and geometries of the materials are good predictors of the stresses created by an adhesive. A comprehensive model was developed for a bi-material strip where the adhesive and one interface are isolated and the stress created by the adhesive during cure is measured from the deflection of a bonded strip [9]. The equation is shown below:The table below shows the oxidation potential of several materials. The larger potential difference between the dissimilar metals results in a faster rate of corrosion.Anodic - tendency to oxidize (lose electrons) Lithium Sodium Aluminum Nickel Tin Li Na Al Ni Sn Li+ +E˚(V) 3.04 2.71+e-Na Al Ni+e3+ 2+ 2++ 3e + 2e + 2e1.66 0.25 0.14 E˚(V) 0.00--SnCathodic - tendency to be reduced (gain electrons) Hydrogen Water/Oxygen Copper Silver Carbon Gold H2 4OH Cu Ag C Au-2H + 2e+-2H2O + O2 + 4e Cu + e Ag + e C +e Au3+ + + + --0.40 -0.52 -0.80 -0.80 -1.50-+ 3eTable 1. Standard oxidation potentials of common materials in electronics and solar industry 3.2 Interfacial Stresses With any joining technique, whether solder or adhesive, an expansion coefficient mismatch exists between adhesive and substrate and stress at some level is unavoidable. Interfacial stress is a result of a summation of strains that occur at the interface of adhesive and substrate during the bonding process. The total strain along with an effective modulus of the entire assembly produces a resultant interfacial stress. When the ECA passes through the gel point, the summation of strains begins as the adhesive takes on more elastic than liquid characteristics. Initial strains from the crosslink reactionσ ≡ 3Gr sc + 6Gr (αs − αr )ΔT1 + 4Gg (αs − α g )ΔT2(4)where σ is the interfacial stress between substrate and adhesive, sc is cure shrinkage after gelation, Gr is the rubbery modulus above the glass transition temperature (Tg ), Gg is the glassy modulus below the glass transition temperature, (αs- αr) and (αs- αg) are the coefficient of thermal expansion differences between substrate and adhesive in the rubbery and the glassy phase respectively, and ΔT1 and ΔT2 are the temperature differences between cure temperature to Tg and Tg to final ambient temperature. One important aspect of the model that should be emphasized is the magnitude of the glassy shear modulus, Gg. Fundamentally, polymers undergo a significant increase in modulus when transitioned to temperatures below the Tg. The shear elastic modulus can change byTemperatureambient temperature. Free volume in a polymer is an indicator of the space between molecules and is related to vapor permeability [6]. During damp heat conditioning, the high free volume passes relatively high levels of moisture vapor into the bulk of the adhesive that diffuses to the adhesive-metal interface. With water, electrolyte (ionic species typically in the adhesive), and an electrochemical potential difference between the ECA and the substrate, a galvanic cell is formed. The tin or tin alloy ribbon is the anode with a relatively high oxidation potential or likelihood to liberate electrons. The general reaction is shown below:are followed by cooling strain from the cure temperature since the adhesive will shrink with a different contraction coefficient than the substrates that are bonded. Figure 1 shows the development of strain from the adhesive perspective.Strain DevelopmentCure schedule ends; begin cool1-3 orders of magnitude after passing through the Tg making Gg 1-3 orders of magnitude larger than Gr. Therefore, for a glassy polymer (an ECA having a Tg sufficiently above ambient temperature), the 3rd term will have the largest impact on the overall stress due to the relatively large magnitude of Gg. The 3rd and most significant term in the equation can be minimized by reducing the ΔT2 which is accomplished Figure 2 by simply reducing the Tg of the adhesive. depicts how the stresses begin to increase rapidly once the material passes through the Tg during cooling. Furthermore, when the Tg is reduced to below ambient (~0˚C), the 3rd term disappears because the adhesive is in the “rubbery” phase at ambient; it does not approach the glassy modulus unless the temperature is further reduced to below the glass transition temperature.Stress DevelopmentCure schedule ends; begin cool1010G'Low Modulus, rubber-Like ECA Rigid, High Tg ECA109 ) G' ( [Pa] 108 107 -50.0 -25.0 0.0 25.0 50.0 75.0 Temp [°C] 100.0 125.0 150.0Figure 3. Shear modulus comparison of a rigid ECA with a rubber-like ECA The rubber-like ECA will become glassy and more rigid when exposed to temperatures approaching -40˚C. Some sensitive applications with complex geometries, large expansion mismatches with the adhesive, or fragile non-compliant substrates may require a lower Tg ECA. However, a rubber-like ECA with a Tg near 0˚C exhibits a very low interfacial stress at ambient temperature after curing and cooling, and the only significant stress is during the cooling to lower temperatures below the Tg. Interfacial stresses are cumulative. Relative to the a rigid, high Tg ECA that is already highly stressed at ambient before cool temperature exposure, the rubberlike ECA will typically survive low temperature exposures without delamination or stress cracking. 3.4 Peel Adhesion With an ECA exhibiting rubber-like elasticity as the starting point, adhesion to critical materials such as molybdenum, tin, tin-silver, TCO, and conductive ink are essential. Fundamentally, adhesion is a function of interfacial interactions between the adhesive and the substrate. Molecular forces such as van der Waals forces, hydrogen bonding, electrostatic interactions, and covalent bonds occur to create a strong bond [11]. Often components and additives in an adhesive are matched to the substrate based on the affinity for the substrate. In addition to the interfacial forces, adhesion strength is also highly dependent on the ability of the adhesive to plastically deform and dissipate stresses applied during the adhesion test [12]. Because of the flexibility demands of the thin-film industry and most likely because of the geometry of the ribbons employed, peel testing of the ECA has been adopted by many thin-film manufacturers. The 180˚ peel test was conducted with tin-silver coated ribbon on aluminum substrates comparing a rigid ECA and the ECA with rubber-like flexibility. The test was conducted in accordance with ASTM D903 with a pull rate of 305 mm/s. Tin-silver coated ribbon with a width of 2.5 mm was used. The ECAs were each cured for ½ hr at 150˚C to ensure complete cure. Figure 4 shows a plot of peel strength vs. displacement where the adhesive with rubber-like elasticity exhibits superior peel strength despite having a lower modulus at ambient temperature.Stress = 0 Time Expansion of liquids Cure shrinkage Cooling shrinkage (<Tg)Cooling shrinkage (>Tg)Figure 2. Stress development from an ECA during cure and cooling Another more simplified model interfacial stress is shown below [10]: for adhesiveσ = 2.66Gg (α s − α g )ΔT2(5)Both models emphasize the same physical property data. 3.3 Measuring shear modulus Shear modulus is easily measured by a parallel plate rheometer. In figure 3, an overlay of two significantly different ECAs is shown. The green line represents the typical rigid ECA with a high Tg. The material begins to soften (onset of Tg) at around 50˚C. The Tg would be identified as the inflection of the transition near 85˚C. Also shown is a low modulus, “rubber-like” ECA with a Tg near 0˚C. From the graph it is apparent that the shear modulus of the rubber-like ECA is an entire order of magnitude lower than the rigid ECA. Using interfacial stress models it becomes apparent that the 3rd and most significant term in the model is reduced to nearly zero with the rubber-like ECA.Temperature14 Peel Strength (N/cm) 12 10 8 6 4 2 0 4.0for 1000 hours. Sanding of the ribbon to remove any oxide was performed where the probes were placed to ensure a good connection to the ribbon. The series connection of several interfaces amplifies any corrosion signal.4 pt probex xECARibbon6.0 8.0 10.0 Displacement (cm)12.0Figure 4. Peel adhesion comparison between a rubberlike and rigid ECA In addition to the high peel adhesion values observed, the failure mode was also desirable. The failure mode was characterized as “cohesive”, where adhesive was still attached to each interface, as shown in Figure 5.Figure 6. Series resistance scheme Both the rigid and the rubber-like ECA were chosen for the initial damp heat evaluation in the series resistance scheme. 5 RESULTS AND DISCUSSION 5.1 Initial rubber-like damp heat test The chart in figure 7 shows the rapid escalation of the series resistance with the rigid material, while the rubberlike material exhibits contact resistance that increases steadily over the 1000 hour test period. After the 1000 hours the rigid material exhibited a total resistance change of about 10,000%, while the rubber-like ECA underwent a resistance change of about 400%. The % change in resistance is based on the initial series resistance values. The initial values were in the range of 40 to 60 mohm.5000% 4000%Rubber‐like RigidFigure 5. Cohesive failure mode after peel testing% Change Resistance4EXPERIMENTAL3000% 2000% 1000% 0% 0 ‐1000% 200 400 600 800 1000 1200Although the solar cell is encapsulated, polymeric materials are unable to provide a complete hermetic seal from moisture permeation. Moisture will diffuse into the cell, and certainly the possibility exists that moisture may not easily diffuse out of the cell. In order to exasperate the effect of moisture, the adhesive and ribbons were exposed to 85˚C/85% humidity (damp heat conditioning) without encapsulation. Essentially, the ribbon-adhesive assembly was exposed directly to the damp heat conditions without any barriers. A test scheme was developed in which the ribbons are joined with ECA in series creating 12 ribbonadhesive interfaces. The ribbons were affixed to a sheet of Kapton in the pattern shown in Figure 6, and the adhesive was dispensed and cured. One hour at 150˚C was chosen as the cure condition to ensure complete cure of the ECAs. The resistance was measured by utilizing a Keithley 2750 4-point probe multimeter at the points identified in figure 6. The measurement will take into account resistance of the ribbons, the 6 adhesive beads, and most importantly 12 ribbon-adhesive interfaces. After an initial resistance value was taken, the test pieces were placed in the damp heat chamber at 85˚C/85% RH conditions. The parts were removed about every 150 hours and resistance was tested before placing samples back into the chamber. Testing was performedHours Damp Heat (85/85)Figure 7. Series resistance. Rubber-like vs. rigid The performance disparity is not a function of whether the material is rigid or rubber-like. Permeability of water vapor is related to both the diffusion rate and the solubility of the molecule in the polymer (in this case the rubber-like ECA). The simplified relation is shown below: P=DxS (6)Where D is the diffusion rate and S is the solubility of the penetrating molecule. As mentioned, rubber-like materials typically pass higher levels of moisture vapor because of the high free volume, or molecular space for diffusion. It would be expected that corrosion resistance of the rubber-like ECA would be poor relative to the rigid ECA with much less free volume and slower diffusionrates. The novel approach to the rubber-like ECA elasticity has shown solid progress towards a potential stable electrical resistance joint after exposure to damp heat conditioning. 5.2 Rubber-like Improved ECA After examination of the rubber-like formulation, some modifications were made to the rubber-like ECA, and the damp heat testing was repeated. The improved version is labeled “rubber-like improved”. As shown in Figure 8, not only is the rubber-like ECA better, but the overall series resistance has been reduced by 60% after 1000 hours of damp heat conditioning.1000%Rubber‐likewas observed on tin-silver ribbon from vendor 1. Cleaning appeared to enhance damp heat performance with vendor 2, as well. The test set-up was repeated with both vendors and both conditions with the improved rubber-like ECA, as well. Figure 10 below shows the typical improved performance of the rubber-like improved version.80 1000 hr, % change 40 0 -40 -80 As is 1 Condition within Vendor Clean As is 2 Clean800%Rigid Rubber‐like  improved% Change Resistance600% 400% 200% 0% 0 ‐200% 200 400 60080010001200Hours Damp Heat (85/85)Figure 10. “Rubber-like improved” ECA with two ribbon vendors and two cleaning conditions When testing the improved version, MEK cleaning improved performance on ribbon from vendor 1, but performance was significantly compromised with cleaning with vendor 2 ribbons. After inspecting the samples, it was apparent that the large cross section of adhesive was out of position during the test affecting both “cleaned” trials from vendor 2. Individual joints were tested and results were comparable to other well performing interfaces. Therefore, the higher resistance results are believed to be unrelated to interfacial performance changes on that cleaned ribbon from vendor 2. Ribbons from vendor 1 were soaked in MEK for 16 hours, and FTIR was performed in an effort to identify the potential material removed from the ribbon by the MEK. Subtracting the spectrum from MEK, the scan in Figure 11 identifies the ribbon contaminants as a range of aliphatic carboxylic acids.Figure 8. Improved rubber-like series resistance 5.3 Ribbon variables When manufacturing with tin and tin silver ribbon, manufacturers want the liberty of using ribbons without any cleaning or special handling to ensure performance. Tin-silver ribbons from two different vendors were tested for corrosion resistance under two different conditions. First, the ribbon was removed from the spool and used without cleaning, and was essentially used as received. The condition is labeled “as is”. In addition, the ribbons from each vendor were subjected to a cleaning process with MEK and a lint-free towel. Those samples are labeled “clean”. Both ribbon types had a 96.5/3.5 Sn/Ag composition and a 15-25 micron finish over a copper core. The damp heat conditioning test scheme was performed twice for each condition and the initial rubberlike ECA was used. The results are shown in Figure 9.1000 1000 hr, % change 800 600 400 200 As is 1 Condition within Vendor Clean As is 2 CleanFigure 11. FTIR of residue from ribbon from vendor 1 5.4 Expanding the chemistry Applying knowledge from the “rubber-like improved” material, two additional ECA materials were developed. To provide the fastest application of adhesive in the assembly process, jet dispense technology is anFigure 9. Rubber-like ECA with two ribbon vendors and two cleaning conditions A clear distinction between the “clean” and “as is”attractive option. Jet dispensable ECAs require unique rheology characteristics to ensure repeatable dispensability. A “rubber-like improved” ECA with jet dispensable rheology characteristics was developed and tested for resistance stability with damp heat conditioning. In addition, a rigid higher strength adhesive with a Tg of around 70˚C was developed with the “improved” corrosion resistance technology. Both chemistries along with the “rubber-like improved” ECA were tested on a 3rd vendor’s tin-silver coated (96.5/3.5) ribbon and on a 100% pure tin coated ribbon. The results are shown in Figure 12.REFERENCES [1] M. Powalla, M. Cemernjak, J. Eberhardt, F. Kessler, R. Kniese, H.D. Mohring, International PVSEC-14 (2004) 1-4. [2] B.R. Olaisen, M. Edoff, A. Holt, 24th European Photovoltaic Solar Energy Conference, 21-25 September (2009) 3007-3010. [3] D. Lu, C.P. Wong, Proceedings International Symposium on Advanced Packaging Materials (1999) 288-294. [4] H.K. Kim, F.G. Shi, Microelectronics Journal 32 (2000) 315-321. [5] Q.K. Tong, G. Fredrickson, R. Kuder, D. Lu, Proceedings of the 49th Electronic Components and Technology Conference (1999) 347-352. 1000 hr, % change-40 -45 -50 -55 -60 -65 -70 Jet R-like Sn ECA within Ribbon Surfac e Rigid Jet R-like Sn-Ag Rigid100% Sn Coating96.5/3.5 Sn-Ag[6] J. Bicerano, Prediction of Polymer Properties, 3rd Ed. Marcel Dekker, Inc. NY, 2002. [7] C.P. Wong, Daoqiang Lu, Quinn K Tong, IEEE Transactions of Electronics Packaging Manufacturing, (1999) 228-232. [8] Y.L. Li, C.P. Wong, Materials Science and Engineering R 51 (2006) 1-35. [9] J. Lange, S. Toll. J.E. Manson, Polymer 36 No. 16, (1995) 3135-3141. [10] C. Brahatheeswaran, V.B. Gupta, Polymer 34 No. 2, (1993) 289-293. [11] K.W. Allen, International Journal of Adhesion & Adhesives 23 (2003) 87-93. [12] J.Kim, K.S. Kim, and Y.H. Kim, Mechanical Effects in Peel Adhesion Test, J. Adhesion Sci. Technol. 3 (1989) 175-187.Figure 12. Jettable, rubber-like improved, and rigid ECAs employing same corrosion-resistant technology show stable electrical resistance after 1000 hours on both tin-silver-coated and 100% tin-coated ribbon. Employing the technology that was developed in the “rubber-like improved” version, the jettable and rigid ECAs also demonstrated stable contact resistance through damp heat conditioning. Despite subtle differences, all samples demonstrated at least a 50% drop in the total resistance across the 12 interfaces indicating that the resistance stability is attainable in both a rigid or rubberlike ECA. Furthermore, the strong performance was not affected by the ribbon composition. Tin-coated ribbon alone performed comparable to tin-silver ribbon. 6 CONCLUSIONS A reliable, stable interconnect is essential to assuring long term life of a PV panel. The reliability of the interconnect can improved by mitigating the risk of both stress and corrosion. Stress concerns with the ECA are addressed with a relatively low Tg ECA. An ECA with a low Tg reduces the temperature range in which the significant stresses develop in the glassy region of the polymer. However, low Tg materials have been problematic concerning corrosion resistance on nonnoble substrates when subjected to damp heat conditioning. A new stable rubber-like ECA has been demonstrated that exhibits the damp heat stability on tin and tin-silver coated ribbon With the tandem properties of low interfacial stress and corrosion resistance the new rubber-like ECA enables the thin-film industy with the ability to provide more flexible solutions with long term reliability. Furthermore, potential utility in c-Si applications can be realized when interfacial stress of a solder interconnect is a potential problem.。