Development, Evaluation, and Implementation of Thermally Conductive Component Attach Adhesives in Inline Curing SMT Assembly Processes*
Leonard R. Enlow and Dale W. Swanson, Boeing Electronic Systems and Missile Defense Information and Communication Systems, 3370 Miraloma Avenue, Anaheim, CA 92803, (714) 762-3075 Fax (714) 762-6222, eMail firstname.lastname@example.org
Thermally conductive adhesives are extensively used for component attachment in Surface Mount Technology (SMT) assembly operations. Heat sinking adhesives are required in high-power dissipating devices to maintain °JA and °JC for packaged parts. Adhesive curing during inline Infrared (IR) soldering or vapor-phase mass soldering operations can result in misaligned or lifted parts due to “shift and swell” mechanisms. Shift and swell of parts during inline operations result from solvent in formulations and possibly catalysts that decompose in soldering processes. Volatilization of solvents or decomposition of catalysts induces voids. Misalignment, excessive rework, and degraded thermal conductivity of the cured material result. Process modifications that may be required limit single-process initiative and throughput in rate-manufacturing operations with IR and vapor-phase soldering. Excessive rework or scrap rates for assembly, failure to meet visual inspection, or variable thermal dissipation performance are other consequences of improper adhesive selection or application. Finally, adhesive materials must meet the NASA outgassing standards and be reworkable at or below 80°C.
A series of evaluations on off-the-shelf materials that are 100-percent solids materials or solvent-containing adhesives was performed. The commercial materials evaluated were not suitable for a specific, mature program and required rework under excessive conditions or off-line vacuum bake cycles that reduced production throughput. As a result of this initial investigation, a custom formulation was developed by Aptek Laboratories (Valencia, Calif.) and qualified at Boeing/Anaheim.
The new formulation, a 100-percent solids epoxy, was evaluated for inline cure characteristics, dispensability through a Camalot dispenser over work life, reworkability, thermal conductivity, outgassing per the NASA test, and insulation resistance at ambient and after MIL-STD-202 humidity cycling. Other process evaluations for qualification included resistance to cleaning solvents, cracking of glass-body diodes, and adhesion to nickel plated (electrolytic/electroless) and gold-plated surfaces. The material developed has been qualified to an existing specification and implemented in low-to-medium volume, high-mix prototype operations at ES&MD-Anaheim and rate-manufacturing operations at the El Paso manufacturing facility.
Key words: Surface-mount assembly, adhesives, thermal transfer
Printed-wiring boards assembled with screen-printed solder paste and components attached with adhesive in the “green” state are run through an inline cure prior to soldering at high temperatures. Before the reflow zone, adhesive curing occurs in the preheat zones.
Adhesives for the assembly of printed-circuit boards for space-and-satellite programs must meet a variety of demanding requirements. Ideal properties of such an adhesive for an established program are:
- Thermal conductivity > 0.87 W/m-°K or at least within 5 percent of this limit to avoid a thermal redesign
- Autodispensible and maintain orientation of the part with adhesive in the green state prior to cure
- Compatible with inline curing in the preheat section of IR soldering equipment (snap cure, < 3 minutes at 150°C, or equivalent)
- Survive IR soldering temperatures
- Allow easy component removability (lap shear < 500 PSI) at moderate temperatures
- Resist common solvents
- Meet the NASA outgassing standards
Surface mount assembly of printed circuit boards in programs such as Guidance Replacement Program (GRP) and Global Positioning Systems (GPS) with Flexobond 442 have experienced shift and swell during inline curing operations. The Flexobond material was chosen based on long-term use in many programs over the past 20 years. Conventional inline cure profiles with rapid temperature increases result in degradation and volatilization of a catalyst that leads to shift and swell, degraded thermal conductivity and rejection of parts or excessive rework results.
A literature search and industry survey was performed and off-the-shelf adhesives from Ablestik, Furane Products, AI Technologies, and Epo Tek were considered as replacements for Flexobond 442. Adhesives evaluated did not meet all criteria. Epo Tek H70 E-4, for example, met the thermal conductivity requirement and is reworkable, but is not inline cure compatible. Another promising candidate was AI Technologies ME 7155-SC4 that could be cured at two minutes at 180°C, but a “prebake” to drive off solvent prior to bonding limits throughput. Amicon D124F and Amicon E 6001 cure in 1.5 minutes or less at 150°C, but would require a thermal redesign analysis.
Because no off-the-shelf material met all requirements, a custom adhesive was prepared by Aptek Laboratories (Valencia, Calif.) and evaluated for process compatibility and material characterization. The results and discussions that follow are a case history for qualification of the material and implementation in production.
Results and Discussion
The shift and swell of parts during inline operations result from solvent in formulations and possibly catalysts that decompose in soldering processes. Decomposition and volatilization of catalysts induce voids during inline and box-oven cures. Misalignment, excessive rework, and degraded thermal conductivity of the cured material result. Excessive rework or scrap-out of parts is required, or thermal dissipation requirements are not met.
A previous effort was conducted to select off-the-shelf materials that are 100-percent solid materials or solvent-containing adhesives that may be prebaked prior to curing in mass soldering operations. Epi Bond 7275, for example, a 100-percent solid material required rework under excessive conditions and other solvent-containing adhesives resulted in voiding of the bond line. Process modifications such as vacuum baking or air drying were not acceptable because of throughput limitations.
The only solution to this problem is the development of a custom adhesive. Uralane 7760 was also evaluated because of potential compatibility with inline curing, and Flexobond 442 was used as a control.
The following requirements were established:
- Firm gel at three minutes at 150°C
- Thermal conductivity > 0.864 W/m-°K or higher than Flexobond 442
- Reworkable at or below 80°C
- No outgassing or voiding leading to shift and swell during cure
- Pass NASA outgassing standards
Five samples were submitted by Aptek Labs. These materials had work-lives ranging from
2 to 24 hours [10 cc syringes], viscosities from 280K to 800K cps, and shelf lives of 3 to 12 months
(-40°C). Hardness values were in the range of 80A to 80D. The following evaluation procedure was established:
- Run differential scanning calorimetry (DSC) scans to verify cure reaction at 150°C
- Evaluate dispensing pressures over pot life of the materials and schedules for uniformly applying material
- Run parts down a belt furnace with Flexobond 442 as a control and monitor shift and swell of the part. Measure height of the lead above the pad with the focusing microscope before and after inline cure.
The schedule in Figure 1 was used. Verify offline cure at 100 to 150°C or postcure need
- Evaluate reworkability at or below 80°C
- Rank materials and select the most promising candidates
- Measure thermal conductivity
- Determine outgassing based on ASTM E 595 or similar internal procedure
- Develop specification criteria
- Evaluate process requirements (primer, etc.)
DSC scans were run at 1°C/second from 50° to 200°C to identify the range of curing temperatures. Figure 2 shows a typical DSC scan for 95154-4. All materials should cure with inline processes. Figure 3 is the DSC for Flexobond 442 (control).
Next, materials were dispensed using the Camalot and a single part placed on a circuit board (Figure 4). Dispensability of materials, without changing dispensing pressures, over the pot life was verified. The initial -3 sample was not evaluated further due to poor dispensability. The height of the lead above the board was measured, orientation verified, and then parts were run down a belt furnace with the profile in Figure 1. After the cure parts were rechecked for alignment/orientation, the lead height was remeasured. Flexobond 442 was run as a control and Uralane 7760 was evaluated for box-oven curing (100°C for 30 minutes). The cure schedule for parts bonded with Flexobond 442 was two hours at 80°C (180°F). The height of the leads before and after cure were measured for all materials. Average values for pins on eight locations were evaluated for each adhesive and percent change determined (Figure 5). No shifts in x, y alignment were found for any material. The ranking of materials after this initial evaluation is:
- 95154-4 (11%)
- Uralane 7760 (19%)
- Uralane 7760 (inline followed by post cure, 23%)
- 95154-5 (30%)
- 95154-2 (41%)
- Flexobond 442 (64%)
- 95154-1 (88%)
Next, bonded parts were reworked at 80°C. The Airvac was used and a hot-nitrogen jet and a heated stage were required to heat the parts. After heating to approximately 80°C, the jet was turned off and all parts twisted off. All parts were easily reworked, and the mode of failure for the 95154-series materials was adhesive failure with material remaining on the board. Uralane 7760 and Flexobond 442 failure modes were cohesive with material on the package and the board. Appearances of adhesive surfaces for the 95154 series were the most uniform and void free. The 95154-5 material separated from the board in a uniform sheet. Typical photographs of reworked surfaces are in Figure 6. Cotton swabs immersed in isopropanol applied to board surfaces was successfully used for cleanup.
Tensile shear specimens (all panels) were prepared for box-oven cures and inline cures to test for cure completeness and shear strength of the materials. Flexobond 442 was used as a control, and Uralane 7760 (box-oven cure only and inline cure followed by box-oven cure) and the Aptek 95154-4 and 95154-5 materials were tested. Oven-cured parts were exposed to extended cure cycles of 20 to 30 minutes at 150°C for the -4 and -5 because of the higher thermal mass of fixtures used to hold the parts.
Tensile strength results are in Table 1. The data suggests that an offline cure is required to improve shear strength for the Aptek material 95154-4. Failure mechanism was adhesive at the metal surface. Another set of tensile coupons was assembled with 95154-4, 95154-5, and Uralane 7760. The Aptek materials were exposed to inline cure followed by a post cure of 30 minutes at 150°C and the Uralane 7760 post cured for 30 minutes at 100°C after inline cure. Tensile test results were 640 psi for the 95154-4, 408 psi for the 95154-5, and 320 psi for the Uralane 7760. Average increases for the -4, -5, and Uralane 7760 were 74, 15, and 17 percent, respectively. Temperatures of 150°C, however, for 30 minute cures may be too long and flux residue might not be removable. As an alternative, standard inline cure followed by a 30-minute offline cure at 100°C was evaluated. Two boards and parts, bonded with 95154-4 and 95154-5, exposed to an inline cure and then a box-oven post-cure, showed swell values of 4-percent after inline cure (-4) and 24-percent (-5). The results were generally consistent with previous data. After post-cure, however, both materials showed bond-line contractions of approximately 59 and 51 percent for the -4 and -5, respectively. Tensile panels with 95154-4 and 95154-5 were cured under these conditions and pull strengths were 606 and 424 psi, respectively. Control values for Flexobond 442 were 472 psi. Test panel preparation for the Aptek material was changed based on vendor recommendation to use scuff sanding followed by oxygen bleach and an isopropyl alcohol clean.
Rework of parts was verified and then the boards with residual material were exposed to a standard cleaning cycle, five-minute immersion in Kyzen Ionox FR at 145°C, followed by deionized water at 159°C, and a drying cycle. Visual examination of parts after cleaning showed no evidence of attack by the solvent. Next, parts through inline cure only with the -4 and -5 were prepared and run through the cleaning cycle. The parts were reworked and visually examined and no evidence of solvent attack found.
Samples of 95154-6 and 95154-7, (replacements for the 95154-3) were screened by DSC and parts for shift/swell prepared. The -6 and -7 adhesives lack sufficient green strength to maintain orientation and were not evaluated further. The -4 and -5 are the leading inline curable candidates in this work.
Samples of 95154-4 and -5 and the Uralane 7760 were prepared for thermal conductivity and outgassing analysis. Thermal conductivity of the 95154-4 was 0.705 W/m-°K and values for 95154-5 and Uralane 7760 were 0.485 and 0.633 W/m-°K. Published data for Uralane 7760 is 0.691W/m-°K.  Previous test data for Flexobond 442 was 0.484 W/m-°K compared to data sheet value of 0.864
W/m-°K. ,. The 95154-4 material was chosen as the leading candidate. Testing for the material specification is summarized in Table 2, including modifications to requirements for the new material. Because extrusion testing was not a standard vendor test, however, ASTM D 1824 was used as an alternate.
Pot life test method for Dis-A-Paste 2311 PMF was revised to use a Brookfield Viscometer, Spindle 7, Speed 5 in accordance with ASTM D 1824. Pot life is the point at which the viscosity exceeds 700,000 cps. Extrusion time was replaced with viscosity values based on thixotropic index. Viscosity limits are 470 + 125K (Spindle 7, 5 rpm) and 1200 + 300K cps (Spindle 7, 0.5 rpm) for the 95154-4 only. Insulation resistance or ten-day humidity cycling was performed by Delsen Test Laboratories. The 95154-4 material specification was accepted and assigned a production name of Dis-A-Paste 2311 PMF.
Next, a series of process evaluations were made on gold-plated and nickel-plated tensile panels to evaluate the need for primer in El Paso production operations. Flexobond 442 was used as a control and EC 2290 was the primer used on previous GPS programs at Boeing, and two new sets of samples were prepared on electroless (serialized 17) and electrolytic (serialized 814) nickel-plated aluminum tensile panels. All parts with the Dis-A-Paste 2311 PMF were abraded with 240-grit paper, cleaned with an isopropanol wipe followed by an IPA rinse, and then blow-dried with nitrogen. One-half of all parts had EC 2290 primer (13, 810) and one-half of all parts were used as controls without primer (46, 1112). Finally, parts 7 and 14 were bonded with Flexobond 442 without abrasion and primer. The primer was air dried for 45 minutes and then cured at 121°C (250°F) for 1.3 hours (78 minutes). Flexobond 442 was cured at 82°C for two hours and the Dis-A-Paste 2311 PMF was cured at 100°C for 30 minutes.
Tensile adhesion of panels with abrasion and primer were highest (410, 415, and 432 PSI) on electrolytic nickel and electroless nickel (456, 486, and 588 PSI). Samples that were abraded (no primer) were best on electroless nickel (336, 235, 292 PSI) compared to electrolytic (208, 177 PSI), and Flexobond 442 (no primer) were the lowest (125 and 170 PSI).
Based on this data, abrasion of the surfaces and the use of primer with the Dis-A-Paste 2311 PMF is advised. Adhesion of the Dis-A-Paste 2311 PMF without primer was higher than Flexobond 442, but the effect of abrasion on mechanical interlocking only of the Flexobond adhesive and surfaces (no primer) was not evaluated. Optimum adhesion to nickel-plated surfaces requires both abrasion and priming of the surface. A third set of samples was prepared with gold-plated coupons.
Finally, glass-body diodes were bonded to boards with the 95154-4 and then exposed to IR and vapor-phase reflow. No cracking was observed.
Thermomechanical analysis on 95154-4 and Uralane 7760 were also run and results are 146.1 ppm/°C and 157.1 ppm/°C, respectively, from 10° to +70°C.
Thermal vacuum stability (TVS) testing of the Aptek 95154-4 was performed. The testing was performed as specified in Materials and Processes Procedure 3340-006 (Boeing/Downey, Calif.). A total of three samples was tested in triplicate for a period of 24 hours at 125°C and <10-5 torr. Although this test is not identical to the “NASA” test, results show good correlation. Results of the TVS test for Aptek 75154-4 were 0.20, 0.18, and 0.22 for sample 1 and 0.20, 0.20, and 0.16 for sample 2. Volatile condensable materials for sample 1 were 0.06, 0.02, and 0.06 percent and 0.03, 0.01, and 0.01 percent. Water-vapor recovery was not performed because of the low mass loss.
Test data for Uralane 7760 and Flexobond 442 were not taken because literature references and published data.[1, 2, 4]
The experimental Aptek material designated 95154-4 and Uralane 7760 were the most promising materials based on minimizing bond line swell. Thermal conductivity measurements showed that 95154-4 was the highest (0.705 W/m-°K), followed by Uralane 7760 (0.633 W/m-°K) and 95154-5 (0.485 W/m-°K). The bond strength of the tensile coupons after inline cure followed by post cure at 150°C and 100°C were comparable to or higher than Flexobond 442.
The most promising materials were unaffected by Kyzen Ionox FC at 145°C for a standard five-minute cleaning cycle. Panels prepared with 95154-4 had the highest tensile strength under cure conditions, is the material of choice, and has been added to item identification drawing 567-0073. Firm gel at three minutes at 150°C was met only with the 95154-4. Uralane 7760 and the Aptek material meet the requirements of this work:
- Maximum thermal conductivity
- Reworkable at or below 80°C
- No outgassing or voiding leading to shift and swell during cure
- Pass NASA outgassing standards
Uralane 7760 is a vendor-established replacement for Uralane 7764 and Aptek 95154-4 was selected as a direct replacement for Flexobond 442. The experimental designation Aptek 95154-4 has been replaced with the production name of Dis-A-Paste 2311 PMF.
The authors wish to acknowledge the efforts of Laura Lanning and Ralph Garcia for epoxy dispensing, parts rework, and parts cleaning; Guadalupe Vargas, Billy Howard, and Ruthie Arasco for parts placement and alignment; Dan Stiegler and Ed Timms for thickness measurement and solder stenciling; Dottie Rushton for tensile testing; Stella Perez for thermal conductivity testing; and George Gerard and Cynthia Wittman for NASA outgassing analysis. The support, comments, and recommendations of Joe Vaccaro and Mary Nolan of Aptek Laboratories are also appreciated.
Leonard R. Enlow retired in January, 1999, and was a team leader for M&P Laboratories at Electronic Systems and Missile Defense of Boeing Information and Communications Systems in Anaheim, CA. He received his BSEE degree from Pennsylvania State University and his MSEE from California State University at Long Beach. Len had over 31 years experience with Electronic Systems and Missile Defense in the design and application of hybrid microcircuits, and was actively involved in the design of hybrids for space, missile, aircraft, and medical applications. He was responsible for all engineering and manufacturing productivity projects for microelectronics materials and processes. Current projects included implementing best commercial plastics for microelectronics systems. Len lead a team of twelve engineers and technicians evaluating metallic and nonmetallic materials and processes for electronic systems at Boeing/Anaheim. Mr. Enlow is co-author of the book Hybrid Microcircuit Technology Handbook, Noyes Publications 1988, now in its first revision. He is also a frequent lecturer at UCLA in the Department of Electrical Engineering on Hybrid Microcircuits and Multichip Module Packaging technologies.
Dale W. Swanson received his BS degree in Chemistry in 1977 from the University of Minnesota at Duluth and his MS degree (Chemistry) from the University of California, Irvine, in 1982. Mr. Swanson is an Engineering Specialist for organic materials in the M&P Laboratories at Electronic Systems and Missile Defense of Boeing Information and Communications Systems in Anaheim, CA. He is currently responsible for engineering and manufacturing productivity projects including adhesives for fine-pitch surface-mount technology and nonhermetic packaging materials and processes. Previous project responsibilities have been in the area of tamper detection in microelectronic assemblies and reverse-engineering resistant die coatings for single-chip devices and testing services for advanced die coatings. Dale’s responsibilities include research, production support, applied development on coatings and encapsulants, and business development and marketing of plastics and polymers for microelectronics.
- Chow, D. T. and Hermansen, R.D., “Novel Thermally Conductive Elastomeric Adhesive For Electronic Components Assembly,” Proceedings of the Second International SAMPE Electronics Conference, June 1416, 1988, p. 332342.
- Furane Products Data Sheet, Uralane 7760.
- Boeing Test Report, “Adhesive Development for Mass Soldering of Fine-Pitch Surface-Mount Components.”
* An earlier version of this paper was published in the Proceedings of the International Symposium of Microelectronics, San Diego, October 31-November 4, 1998.
| Table of Contents | Previous Page | Next Page |