Green Mold Compound
04/01/2003
Moldability and Reliability in Wire Bonded Packages
BY T. Y. LIN AND C.M. FANG
With the electronics industry driving for cost reduction, miniaturization and functional integration, the conventional integrated circuit (IC) package design is forced to migrate to finer wire bonding pad pitches, from 90 to 60 µm, or even 35 µm staggered pitch. Glob-top dispensing and central gate molding are innovative methods for achieving high-assembly yield.
For these alternative methods to transfer molding, the concurrent fine-tuning of mold compound viscosity and process parameter optimization are required. The ongoing trend for high-density wire bonds dictates not only a smaller bonding pad pitch, but the wire itself also is becoming thinner. Filler contents of the mold compound are increased gradually to reduce the coefficient of thermal expansion (CTE) and improve package strength.
 Figure 1. Wire sweep is defined as the maximum wire deformation expressed as a percentage of the wire loop length. |
null
The green mold compound used for high-density wire bonded packages often is highly moisture resistant, which can prevent the occurrence of "popcorn" failures in the plastic IC packages.
In this study, both moldability and reliability were assessed in terms of wire sweep, mold void, coplanarity performance and temperature cycle results. Wire sweep issues were improved greatly, but coplanarity still needs to be improved due to relatively high warpage problems. Although the traditional popcorn failures were not observed, new failure modes were found after 500 temperature cycles (temperature range from -55° to 150°C) with moisture soaking at MSL3.
Test Vehicle Selection
Transfer molding is an automatic compression molding process where the mold compounds are preheated and then transferred into the mold cavities. Generally, transfer molding is applied for chip-up types of packages, and gold wires at the corner adjacent to the mold gate show maximum sweep depending on force and speed. The bond pitch and wire span as well as the bonding strength could be the critical factors influencing wire sweep under a certain level of flow velocity, viscosity, etc. Wire bond loop height, as well as wire diameter and stiffness also impact the performance of wire sweep. In this study, a 216-lead thin quad flat pack (TQFP) was selected as the test vehicle.
The objective of the moldability studies was to assess the suitability of the evaluated compound for mass production. Additionally the process window was assessed and batch-to-batch, lot-to-lot variations of the mold compound were evaluated. The moldability studies focus on the evaluations of wire sweep, lead frame paddle tilt, mold voids and coplanarity.
Three green mold compounds (GMC-1, -2 and -3) were evaluated. GMC-1 was the original green compound for evaluation, and its wire sweep was 5 to 8 percent, more than the 4 percent that was specified as the internal wire sweep limit. Wire sweep is defined as the ratio of the maximum wire deviation (deformation) to the wire span. The wire sweep for molding temperatures in the range of 170° to 190°C were above 4 percent, which indicates that no process window was found for GMC-1. Following that, GMC-2 was formulated after fine-tuning the viscosity. The performance of wire sweep was improved, but coplanarity could not achieve the target for production. GMC-3 was then formulated by changing the catalyst type and fine-tuning the viscosity. The performance of wire sweep on GMC-3 was almost the same as GMC-2, and the coplanarity of GMC-3 passed the requirements of production. The warpage and coplanarity for GMC-3 were better at a molding temperature of 170°C than 180° or 190°C. This result indicates that the process window was not large.
 |
null
 |
null
 |
null
 |
null
 |
null
 |
null
 |
null
Reliability Assessment
Typically, the major parts of a reliability evaluation include autoclave, temperature shock (liquid-to-liquid -55° to 125°C), temperature cycling (TC), high temperature storage, and temperature and humidity with bias. In this study, only the main TC results based on JEDEC condition C (-65° to 150°C) were reported because several failure modes were observed during the TC tests.
 Figure 2. Mold body warpage measured by co-planarity scanner (molding at 170??C). |
null
Three failure modes — ball bond lift, ball neck cracking and die crazing — were observed after 500 temperature cycles. After removing the encapsulation, it was found that ball bond lift and ball neck cracking induced electrical function failures. The cracks occurred at the intermetallic layer between the bond pad and wire bond. It seems that the mold compound pulled out the wire and created fractures at the ball neck and interface between the bond pad and gold ball.
C-SAM was applied to assess the delamination of the package. However, no delamination examined the die surface before and after wafer backgrinding and sawing. No edge cracks on the die surface were seen after checking. Therefore, die crazing was identified as a result of temperature cycling. No evidence was shown that die crack failures occurred in the front of line during the assembly process. Therefore, the failures indicated that the stress level of the mold compound could be higher.
Conclusion
It was demonstrated that green mold compound has some issues in the moldability studies, such as a relatively low process capability in the coplanarity measurement. The moldability was improved during development, but further improvements would be beneficial. Additionally, the reliability assessment had identified the presence of temperature cycling failures, such as ball bond lift and ball neck cracking that induced electrical function failures.
For a complete list of references, please contact the authors.
T.Y. Lin may be contacted at Motorola Innovation Centre, (65) 6486 3160; Fax: (65) 6481-5129; E-mail: [email protected]. C.M. Fang may be contacted at Delphi Automotive Systems Singapore Pte Ltd., (65) 6450 8515; Fax: (65) 6454 8391; E-mail:[email protected].