Reworkable component attach adhesives
10/01/2001
Evaluation, testing and selection for large MCM applications
BY PATRICK GLYNN
For large multi-chip modules (MCMs) populated with many active and passive components, it is important to use a reliable, reworkable die attach adhesive. The adhesive must be reworkable at a low enough temperature to avoid damage to any part of the MCM, yet be able to withstand environmental stresses imposed during reliability testing. Adhesive deposition techniques - either liquid dispensing for epoxy materials, or direct pick-and-place transfer for thermoplastic and silver-filled films - must be compatible with the production environment and process equipment. Properties such as adhesive strength, thermal conductivity, electrical resistivity and cure temperatures also need to be considered. This article summarizes results of an effort to evaluate, select and qualify a reworkable adhesive suitable for high reliability MCM assembly.
Background and Approach
Harris Advanced Manufacturing Technology (AMT) needed to find a substitute for its standard component adhesive to meet MCM design ground rules that called for an adhesive with a rework temperature of less than 150°C. Several rework approaches were taken using a standard component attach adhesive, but this material was determined to be inappropriate for rework, primarily because a very high (greater than 175°C) rework temperature was needed to break the adhesive bonds to a point where rework techniques could be used successfully.
Industry, supplier and peer searches yielded a list of 12 potential reworkable adhesive candidates. The various component attach adhesives were considered and tested, a low-temperature rework process was defined and standardized, and test methods were devised to simulate standard manufacturing flows as well as to include extended environmental stresses.
 Figure 1. Test vehicle layout. |
Adhesive strength data was gathered during component rework and at several environmental pre-stress and post-stress stages, and these strength results helped further reduce the adhesive candidate list. A ranking system was employed to aid in the selection process and ultimately a material was selected.
Because the selected component adhesive was to be used in an application requiring high reliability, hermetic packaged samples were manufactured and submitted for a short-term residual gas analysis (RGA) test. In addition, because there may be applications where components require electrical connection through the adhesive and substrate, sheet resistivity of the selected adhesive was compared with that of the control material.
Candidate Adhesives
The list of candidate reworkable die-attach adhesives included thermoplastics in film and paste form, as well as silver filled epoxies (Table 1). Harris AMT's standard epoxy adhesive was included as a control.
 Table 1. Reworkable component attach adhesives. |
Initial epoxy testing focused on dispense-related material properties. Many of the proposed materials were eliminated because of poor performance, such as excessive post-dispense tailing, epoxy non-wetting to either substrate or component, and excessive resin bleedout. (Bleedout is migration of resin from the epoxy bondline after some time, often carrying silver flake with it, that can contaminate adjacent areas on the substrate.) Ceramic substrates and Au-plated Kovar were used in the evaluations.
 Table 2. Die force values after initial die attach, heating to rework temperatures, and after 1 and 5 rework cycles. |
Other undesirable properties, such as complex thermal cure schedules, high adhesive strength at the target rework temperature, and low overall adhesive strength, also eliminated other materials. Although several film adhesives had desirable properties, they were eliminated because Harris AMT's process equipment is better suited for liquid adhesives, and the cost of modifying existing equipment or purchasing new equipment to accommodate the film adhesives could not be justified for this project. The reworkable epoxies chosen for further study are highlighted in Table 1.
Rework Process and Issues
Harris AMT developed a proprietary low-temperature rework process whereby components can be removed intact from a substrate at an applied force. After removal of the component, residual adhesive is manually removed and the site is ready for component re-attachment.
 Table 3. Summary of die force values for four adhesives and a control material. |
Initial component attach rework evaluations were carried out on each candidate material using 0.25 x 0.25 inch silicon die and ceramic package substrates. Component failure values after initial die attach, at the 150°C rework temperature, and after one and five rework cycles were recorded along with their respective failure modes. No significant difference in failure force values between reworked and non-reworked die was observed for any of the materials. Data are summarized in Table 2.
Dispense properties, including resin bleedout, excessive tailing and wetting to the substrate, were summarized as overall dispense properties and included in Table 2.
Test Vehicles and Test Results
Large ceramic substrates were assembled as test vehicles for each candidate epoxy from Table 2 using 0.35 x 0.35 inch and 0.14 x 0.14 inch silicon die. Each test vehicle was populated with non-reworked as well as reworked die to simulate actual manufacturing conditions. Figures 1 and 2 show the design layout and a photograph of the actual test vehicle.
 Figure 2. Test vehicle. |
To simulate worst-case environmental conditions, an evaluation process was defined involving a series of extended environmental and thermal stresses and corresponding die force measurements (Figure 3). Test vehicles were populated with 24 die each, four of which were reworked and replaced. The samples were subjected to a simulated manufacturing flow consisting of several cure cycles and a vacuum bake.
Environmental testing included 168 hours of storage at 150°C followed by 350 temperature cycles (-55 to 150°C). Die force measurements were made on both reworked and non-reworked die at various points in the process flow, as shown in Figure 3. Post screen flow simulates solder reflow temperature excursion followed by possible die attach rework.
 Table 4. Ranking of overall properties. |
Table 3 summarizes the die force values and failure modes from the extended environmental and thermal stress tests. Die forces are the average of the three die tested (both reworked and non-reworked) at each point. All adhesive candidate materials maintained excellent adhesion even after 350 temperature cycles. As seen in Table 3, some of the materials (epoxies E and J) gained adhesive strength as a result of extended temperature cycles, and other materials (epoxies L and Control) experienced degradation in adhesion. No significant difference in force values between reworked and non-reworked die was observed.
All candidates were tested to and passed Mil-Std-883, Test Method 2019 requirements for die shear strength in excess of three times the minimum 2.5kg (5.5 lbs) requirement.
Final Selection
Final material selection was based on results from the environmental and mechanical tests, compliance to Mil-Std-883 requirements, and overall dispense properties (pot life, material homogeneity, bleedout, tailing and wetting to the substrate).
 Table 5. Short-term outgassing RGA data for epoxy E. |
In addition to other properties, an ideal die attach adhesive should have relatively high adhesive strength at room temperature, but yield at low strength during the rework and residue removal processes. Results of the low-temperature rework evaluation show that epoxy J is the easiest material to rework, followed by epoxies E and H, respectively. Because of their high adhesive strengths at rework temperature, epoxies L and Control were not considered rework-friendly using the low temperature die force method and were eliminated. Taking into account the relatively poor dispense properties of epoxy J (non-wetting to the substrate and moderate tailing), epoxies E and H best fit the criteria for a reworkable epoxy.
Because of high reliability requirements, the material candidate must meet most of the criteria in Mil-Std-883, Test Method 5011. Test Method 5011 provides a good outline for testing adhesives and is widely used in the electronics industry. It outlines material acceptance testing, adhesive strength through environmental stress and various important material properties. According to supplier inputs, epoxy H does not pass the Test Method 5011 Ionic Impurities levels, which could be an important factor in terms of long-term reliability. Epoxy E does meet all the requirements of Mil-Std-883, Test Method 5011, according to data provided by the supplier. Final and overall ranking of the reworkable epoxy candidates is outlined in Table 4.
 Figure 3. Test vehicle - process flow. |
Based on rework properties, successful environmental and thermal stress results, compliance to all requirements of Mil-Std-883, Test Method 5011, and good overall dispense properties, epoxy E was recommended for use as the reworkable component attach epoxy for the large MCM.
RGA and Resistivity Testing
Sample hermetic thin outline style packages using epoxy E for die attach were sent for verification of the short-term outgassing requirements of Test Method 5011, Section 3.8.6, Outgassed Materials. Short-term outgassing requirements call for a 24 hour/150°C pre-bake before RGA testing of at least three hermetic packages (verified by fine and gross leak testing). RGA results were excellent and are summarized in Table 5.
Sheet electrical resistivity of epoxies E and Control were determined and compared with data sheet values. Material E had a measured value of 2.7 (data sheet specified <4) and the control epoxy had a measured value of 1.1 (data sheet specified 0.7).
Conclusion
A method has been developed to evaluate and select low-temperature reworkable adhesives to be used for component attach in MCM applications. This method involves standardization of a low-temperature component rework process and collection of die force data from test vehicles populated with reworked and non-reworked die sites. Die force data were collected before and after environmental stresses including high-temperature storage and temperature cycling.
After passing environmental screening, ranking was made based on rework and dispense properties and compliance to Mil-Std-883, Test Method 5011, and a material was selected. Results of RGA testing and sheet resistivity measurement of the selected adhesive candidate proved acceptable for use in high reliability MCM applications. AP
Patrick Glynn, mechanical engineer, can be contacted at Harris Corp., P.O. Box 37, Melbourne, FL 32902; 321-729-2497; Fax: 321-729-7348; E-mail: [email protected].