Abstract
A variety of epoxy resins, filler mixtures, and adhesion additives were examined in semiconductor molding compounds for solder reflow crack resistance (popcorn crack resistance). A limited comparison of molding compound high-temperature strength properties and the extent of popcorn cracking suggests no strong positive correlation. On the other hand, factors that might have an impact on the adhesion properties of the molding compound, such as low melt viscosity resins and putative adhesion additives appear to have the greatest impact. A molding compound developed with these factors showed no internal or external cracks on 84 lead QFPs (29.2 × 29.2 × 3.68 mm) after 72 hours of 85 °C/85% RH and one 10-second solder dip at 260 °C. Nor were any external cracks observed after 168 hours of 85 °C/85% RH and one solder dip. However, minor internal cracks were observed.
Introduction
Surface mount plastic molded semiconductor packages often crack during the solder-reflow process required to mount them on the board. Three common reflow methods include infrared heating, vapor phase soldering, and immersion in a molten solder bath. SMD packages are heated for as long as 2 minutes at 215–260 °C, which is considerably higher than the soldering temperature for insertion mounted packages (~150 °C). The mechanism of cracking has been extensively investigated (1,2,3,4). Unmounted surface mount parts are often exposed to humid environments (e.g. one week at 30 °C/75% RH) before they are mounted on the board. The resulting moisture absorbed by the epoxy encapsulant surrounding the device and accumulated near the mold compound/lead frame interface is rapidly turned to steam during the solder-reflow step leading to package swelling and fracture of the mold compound. As a result, device manufacturers routinely “dry-pack” devices before shipment to PC board mounters. Once the devices are removed from the dry pack, they become perishable items, which must be mounted within a specific time, depending on the ambient relative humidity and the temperature of their storage area.
Among the many factors that can influence popcorn cracking (e.g. die paddle dimples and slots, wing leads to moisture expulsion, die attach type, and molding compound type), possible improvements in the molding compound have generated the most interest. Most of the discussion has centered on improvements in three broad categories: increased high-temperature strength, lower moisture absorption, and better adhesion. A variety of molding compound formulation methods have become widely accepted for improvements in these categories (Table 1). For example, a high Tg epoxy resin often is correlated with high-temperature strength, while a low Tg resin is often correlated with better wettability and better adhesion. Several recent reports (5,6) suggest that the largest benefit results from improving adhesive strength, with other properties contributing only marginal benefit.
The present study examines the popcorn performance of a variety of mold compounds that differ in many of these properties to validate the most important formulation methods for improving popcorn performance.
EXPERIMENTAL
Molding Compounds
Semiconductor grade, flame retardant novolac-hardened epoxy molding compounds were prepared with a variety of formulation methods examined (A-F, Table 2). Compounds C and D were prepared with two different multifunctional resins, with the highest Tg resin used in Compound C (Table 3). Compound F (control compound) was prepared with a standard epoxy cresol novolac resin (ECN) and contains a few of the formulation methods examined. Compounds A, B, and E were prepared with either of two low Tg, low viscosity resins and were highly silica-filled. Compounds A, B, and F also contained hydrophobic resins or hardeners. Compounds A–E contained mixtures of crushed and spherical fused silica while Compound F contained only crushed fused silica. Compounds A and B also contained small amounts of additives which might be expected (based on their chemical structure) to improve adhesion to the copper lead frame.
Crack Analysis
The external surfaces of the parts were visually examined for cracks after solder dip with a 70-powder microscope and through the use of a bubble chamber. The fluid in the bubble chamber operating at 165°C (Galden DO5 fluid) penetrates the package through cracks or delaminations. The air is subsequently replaced by the fluid bubbles out at the positions of the external defects. Any package that bubbles strongly is considered a failure. Internal cracks were determined with a C-mode scanning acoustic microscope (C-SAM). This method also allowed the measurement of the extension of the crack away from the die pad.
Popcorn procedure
The experiment was conducted in two parts with 84 lead QFPs (29.2 × 29.2 × 3.68 mm) using copper frames and no die. Die were not included to obviate the influence of the die attach material. Compounds A, B, and C were examined first, and then Compounds A, C, D, E, and F were examined.
In the first experiment, three strips of five parts each were molded of each of the three compounds. One strip of each molded and post-cured (6 hours at 175 °C) compound was then subjected to each of three moisture conditions (35.5, 72.5, or 168 hours of 85 °C/85% RH) or in the second experiment either 47, 72, or 168 hours of 85 °C/85% RH (Figure 1).
In all cases, no cracks were observed after moisture absorption – before solder dip, but some paddle back delamination was observed. C-SAM analysis of the parts after solder dip indicated that nearly all the cracks extended down and away from the die paddle towards the bottom of the package where the external cracks were found. In an effort to estimate the overall crack length, the average extension of the crack away from the die paddle towards the side of the package was also measured in the second experiment. However, measurement of the downward extension of the crack was not measured. Almost no secondary cracks toward the leads were observed. A typical crack pattern after solder dip is shown in Figure 2. In this case, one edge of the crack extends to the bottom of the package (left side) and the other edge of the crack remains internal (right side) to the package. The amount of paddle back delamination after solder dip for each of the three moisture conditions was also measured and found to be the same for each of the compounds and each of the conditions (100% delamination). Since the extent of popcorn cracking found depended on both the degree of moisture exposure and the molding compound used in this study, there appeared to be no correlation between delamination and cracking. Recent reports have also questioned the direct relationship between delamination and cracking (6).
Results
Examination of the physical properties of molding compounds A–F (Table 3) revealed a wide range of Tg (130–202 °C), CTE (14–22 ppm/°C), and high-temperature strength values (2700–5100 psi). Two low Tg compounds (A and 8) and one high Tg compound (C) were examined in the first experiment. As expected, the cracking frequency increased with moisture exposure (Table 4). For external cracks, the level of scrutiny was more severe when using the bubble chamber since this method showed fine external cracks that were not seen during visual examination under a microscope and also moisture escape paths along the mold compound/lead interface. C-SAM analysis gave the highest level of scrutiny since internal cracks were included as failures even when no external cracks or delamination moisture pathways were seen. Although there is some scatter in the data, it is clear that only Compound A showed no external cracks after 168 hours of 85/85 and solder dip. Compounds B and D showed nearly the same increased degree of cracking. Compound A also gave the lowest crack frequency in the second experiment (Table 5). The close agreement between the frequency of external cracks and the estimate of internal crack extension by C-SAM suggests that crack extension analysis may be one of the best ways to rate molding compounds for popcorn performance since it provides internal crack information while correlating well with the commonly used external crack analysis.
DISCUSSION
A comparison of the popcorn performance of these six compounds suggests that high-temperature strength cannot be a major indicator of performance. For example, Compounds C and D have significantly higher high-temperature flex strengths when compared to Compound A, but performed less well with regards to popcorn cracking. These results are contrary to the sometimes prevailing view that high-temperature strength is the most important factor. In the present case, the low Tg compound with all its attendant properties (low viscosity, good wettability, and low coefficient of thermal expansion) gave good results. These properties are tied together and determined, in large part, by the properties of the epoxy resin used. The overriding role of any one property cannot be explicitly defined, but the better wettability of low-viscosity resins and the presence of proprietary putative adhesion promoters is suggestive. These results are in agreement with a recent report that emphasizes adhesion as the key formulation factor (5). Picking the right pathway (high Tg vs. low Tg) in the development of anti popcorn molding compounds has been a major concern. The present work lends support to the low Tg pathway.
Although the developed compound (Compound A) shows no internal or external cracks after 72 hours of 85/85 and solder dip, minor internal cracks (extending less than 30 mils away from the die pad) were observed when the moisture exposure was increased to 168 hours. The same internal cracks by IPC standards (7), however, are acceptable if they are less than two-thirds (2/3) of the distance from any internal feature to the outside of the package. Notwithstanding delamination and internal cracking concerns, a separate concern is whether accelerated pre-conditioning stress e.g. 168 hours at 85/85 properly predicts a components moisture sensitivity in a workplace environment (e.g. 30 °C/75% RH -see Ref. 8). Prudhomme reports more cracking of thin packages at lower levels of moisture uptake under ambient conditions than would be predicted from accelerated testing. He suggests that the distribution of the moisture in the package is just as important as the total moisture uptake in predicting popcorn cracking.
Notwithstanding all of these concerns, the significance of the present work is shown by the fact that Compound A has been accepted by a major device manufacturer for an automotive application, in which no “dry bagging” will be used. This is the first molding compound that this device manufacturer has tested for this application which would pass these stringent “no dry bagging” requirements.
CONCLUSIONS
Among the three broad categories for improving molding compounds for popcorn performance (increased high-temperature strength, lower moisture absorption, and higher adhesion), improvements in adhesion appear to yield the greatest benefit in the short term. In the long term improvements in moisture absorption will ultimately lead to the widespread use of popcorn-resistant packages that do not need to be dry-packed.
References
(1) S. Golwalkar, P. Boysan, R. Foehringer, and J. Jacobs, “Moisture Sensitivity of Thin Small Outline Packages”, Electronic Components and Technology Conference, pp. 745-749, 1991
(2) M. Harada, S. Tanigawa, S. Ohizumi and K. lkemura, “X-ray Analysis of the Package Cracking during Reflow Soldering”, International Reliability Physics Symposium, pp. 182-189, 1992
(3) A. Tay, G, Tan, and T. Lim, “A Method for Predicting Properties of Suitable Molding Compounds tor IC Packages”, IEEE Singapore IPFA, pp. 276-280, 1993
(4) H. Fujita and N. Mogi, “High Reliability Epoxy Molding Compound tor Surface Mount Devices”, Electronic Components and Technology Conference, pp. 735-741, 1993
(5) G. Gànesan and H. Berg, “Model and Analysis tor Solder Reflow Cracking Phenomenon in SMT Plastic Packages”, Electronic Components and Technology Conference, pp. 653- 660, 1993
(6) T. Leong and L. Beng, “Pop corn Package Cracking: Critica! Review and New Findings”, IEEE Singapore IPFA, pp. 170-176, 1993
(7) Institute for Interconnecting and Packaging Electronic Circuits, Draft IPC-SM-786A, Spring 1993
(8) M. Prudhomme, “Ambient Moisture Characterization of Thin Small Outline Packages (TSOPs)”, International Reliability Physics Symposium, pp. 79-86, 1994
