Mold Compound and Copper Wire Selection for Quad-Flat Packages with High Density Leadframe in Automotive Applications

Transcription

1 Mold Compound and Copper Wire Selection for Quad-Flat Packages with High Density Leadframe in Automotive Applications Vanessa Wyn Jean Tan Poh Leng Eu Yin Kheng Au Lan Chu Tan 1 Fusionopolis Walk, #12-01/02 South Tower Solaris Singapore lc.tan@nxp.com Boon Yew Low boon.low@nxp.com Abstract The amount of electronics within vehicles continue to increase with the drive towards autonomous driving and green vehicles through improved engine controls and hybrid or full electric vehicles. Engine control units require electronic components which are located very close to the engine, and thus must operate within a high temperature ambient environment, at least meeting AEC Grade 1 or Grade 0 requirements. This paper reviews the bill of material selection for quad-flat package (QFP), in particularly, mold compound and copper wire, with high density leadframes. Palladium coated copper wire (PCC) is the current preferred wire of choice due to the improved manufacturability. Although PCC wire is susceptible to galvanic corrosion due to the presence of sulfur in the molding compound, the level of corrosion is minimal and impact is cosmetic only. Wire sweep is a major factor in the mold compound selection, especially for high density leadframes. Wire sweep percentage differs greatly from row to row, and mold compound type contributes significantly for a robust process on high density leadframes. Furthermore, on large body sized QFPs, package warpage is another main consideration in the selection of mold compound type as warpage directly impacts the lead co-planarity. It is shown that additional heat treatment post assembly, such as during burn-in, may contribute significantly towards warpage degradation. As such, it is necessary to select the molding compound with sufficiently low warpage through repeated heat treatment on the package. Lastly, the selected mold compound needs to meet the delamination requirements as per AEC-Q006 for copper wire interconnects. Keywords Mold compound, copper wire, quad-flat package, high density leadframe I. INTRODUCTION The use of electronics components within vehicles continue to grow year on year. This is contributed by a multitude of reasons which includes the growth of green vehicles in terms of both hybrid or full electric vehicles. The push towards autonomous driving is another major driver that requires immense computation powers in a vehicle. Autonomous driving indirectly promotes the need for connected vehicles, with vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) actively improving the safety level of Level 5 autonomous driving. This increased connectivity feature in vehicles has opened up the door to a different market segment for automotive electronics. With all these increased electronic features in vehicles, the value of semiconductors in a car is expected to increase from an average of $300 per car to approximately $1000. To ensure competitiveness of various semiconductor manufacturers in this highly competitive and growing market segment, key enablement are high computation power, competitive cost, and highly reliable components. Automotive microcontrollers are now using advanced technology nodes, up to 16nm FinFET technology. To meet the cost competitiveness, copper (Cu) wire bond is able to deliver lower cost and has been a very matured technology for advanced node automotive electronics. However, a major concern with copper wire is the reliability level due to corrosion mechanisms. In order to achieve a reliable package, bill of materials (BOM) selection for copper wire bond devices are extremely important. In addition, for further cost competitiveness, high density leadframe form factors are commonly used in the industry. While the leadframe form factor is transparent to the end users, high density leadframes poses additional challenges to the manufacturing process. In general, automotive electronics need to meet at least AEC Grade 1 requirements, but there is an increasing drive towards the need to meet AEC Grade 0 requirements with the devices from advanced nodes. II. COPPER WIRE SELECTION A. Galvanic corrosion in palladium coated copper wire Bare copper wire provides the lowest cost in terms of wire selection for quad-flat package (QFP) applications. However, for large body sized QFPs with high pin count, second bond with bare Cu wire becomes more challenging in terms of continuous bonding performance. Second bond challenges are further aggravated in high density leadframe applications. As such, palladium coated copper wire (PCC) has in recent years become the preferred wire type, as the best cost versus overall throughput tradeoff. However, in high temperature applications, where the ambient temperature requirement is at 175 o C, galvanic corrosion may occur for PCC wire, as previously reported by Stephen et al. [1]. The level of galvanic corrosion, or dry corrosion, is significantly impacted by the level of sulfur in the molding compound. However, sulfur is generally a form XXX-X-XXXX-XXXX-X/XX/$XX.00 20XX IEEE

2 of adhesion promoter in molding compounds for leadframes. Reducing sulfur content may result in delamination issues on the package. Galvanic corrosion initiates at the exposed Cu area beside an area with Pd, creating a galvanic cell. To improve the level of galvanic corrosion, improving the Pd coverage on the bonded ball can improve the reliability of the ball bond towards galvanic corrosion. B. Experimental results The high temperature storage life (HTSL) performance at 175 o C for 2 different types of PCC wires were evaluated. Wire type A is a conventional PCC wire widely used in the industry, while wire type B is an improved PCC wire with improved Pd plating process control to enhance the uniformity of Pd coating on the surface of the free-air-ball (FAB). Fig. 1 below shows the comparison of the bonded ball of wire type A vs wire type B. The samples were intentionally heated longer, so that the exposed Cu area will be oxidized and can be observed from the brownish color tone on the ball flange. More exposed Cu area can be observed on wire type A as compared to wire type B. Fig. 3. Cross-section of the ball bond for wire type B. Almost no copper voids observed, with minimal interfacial crack at the edge of the ball bond Cross-section was also performed on the second bond to assess the level of galvanic corrosion. In this case, both wire types showed comparable level of copper voids corrosion. This is expected because the improvement on the Pd plating process was designed to improve only the FAB Pd coating consistency. At the second bond stitch surface, the scrubbing of the capillary would still expose the core Cu material during second bond, and as such, galvanic corrosion would still occur. However, the level of copper voids seen on the second bond on both wires are minimal and poses no reliability risk after HTSL. Fig. 1. Bonded ball appearance of wire type A (left) vs wire type B (right) Both wire types were then subjected to HTSL at 175 o C for 2016 hours, which is equivalent to 2 AEC Grade 0 requirements. Samples were then cross-sectioned to check for copper voids due to galvanic corrosion. Wire type B showed significantly better copper voids performance, with almost no voids observed. Conversely, wire type A showed not only Cu voids, but also a large interfacial crack along the Cu-IMC interface. Fig. 4. Cross-section of the stitch bond for wire type A. Small voids observed on the stitch bond Fig. 2. Cross-section of the ball bond for wire type A. Severe copper voids observed, with large interfacial crack seen Fig. 5. Cross-section of the stitch bond for wire type B. Small voids observed on the stitch bond

3 III. MOLD COMPOUND SELECTION A. Wire Sweep Considerations For improved cost, migration towards high density leadframes is one of the main cost reduction opportunity in the industry. However, high density leadframes pose severe challenge in terms of wire sweep, especially for large body sized QFPs due to the long wire lengths and fine inner lead pitches. As such, to select the right mold compound for the widest range of package sizes, the test vehicle chosen should be a combination of the largest possible body size on the largest leadframe size, and with the longest wire length in the device. The test vehicle chosen for this study is a 176LQFP package on a high density leadframe. The leadframe consists of 8 columns 3 rows matrix, with a strip size of 281.1mm 99.44mm. This is one of the largest strip size for large body sized QFPs in the industry, and approach the maximum strip width based on current wire bonders capability. 4 molding compound candidates were selected for this evaluation. The various properties of the molding compound are shown in Table I. TABLE I. MOLDING COMPOUND CANDIDATES PROPERTIES Properties Mold Compound Candidates A B C D Filler content (wt%) Spiral flow (cm) Gelation time (s) T g ( o C) CTE1 (ppm) CTE2 (ppm) Mold shrinkage (%) A CMOS090 technology die was used on this LQFP176 package. In this particular test vehicle, the maximum wire length was 4.5mm. Wire sweep was specifically taken from units at the 3 rd row of the leadframe from the mold chase cull block. The 3 rd row is the row which locates furthest away from the pellet pot. It is expected wire sweep to be the worst at this area, as the further the distance travelled by the mold compound, the weaker the resin flowability. This is due to more heat is being absorbed, resulting to higher viscosity. Wire sweep measurements are summarized in the boxplots in Fig. 6. Fig. 7. Severe wire shorting on compound A and B The improved wire sweep performance of compound C and D are due to the longer spiral flow and gelation time of these compound. Based on these results, for robust manufacturability across different devices and wire layouts, only compound C and D are selected for further evaluations. B. Effect of Compound Staging Life and Storage Temperature It is important to identify the shelf life and storage temperature as they affect greatly on the flow rate. To slow down the curing process, compounds are kept in refrigerators. As mold compound is heat reactive polymeric resin, a study has been conducted to understand the wire sweep performance in response to the compound floor life and the storage temperature. Factors studied are compound staging time in both room temperature ~25 o C and higher temperature, which is around 30~32 o C (pellets staged at machine pellet bowl with no air circulation). The warmer condition can be simulated by turning off either mold machine s exhaust fan or the air-conditioning unit (most mold machines come with built-in air conditioning situated at the pellet chamber). Fig. 8 shows the effect of the staging time and storage temperature on the wire sweep performance for mold compound C and D. The main study is to show the effect of compound staging time within 24 hours and 48 hours, and also the effect of storage temperature. Result shows that compound C is least affected by the staging time, as wire sweep percentage is comparable for both <24hrs and >48hrs staging time. Average wire sweep percentage is slightly increased for compound D from staging time <24hrs to >48hrs. Wire sweep performance is significantly affected by the storage temperature for both compounds. Fig. 6. Wire sweep boxplot comparison of the 4 compound candidates Compound A and B showed significantly higher wire sweep compared to compound C and D. In addition, compound A and B also showed clear wire short on units from the 3 rd row of the leadframe, as shown in Fig. 7. Fig. 8. The effect of mold compound staging time and storage condition on wire sweep performance C. Package Warpage and Co-planarity A major consideration for large body sized QFPs is the package warpage performance. Package warpage has a direct effect on the package co-planarity, which is crucial to the customers for board mounting.

4 The desired mold compound for mass production should have good package warpage performance, not only during assembly, but also after additional heating processes, such as through burn-in and reflow. Only compounds C and D are assessed for package warpage and co-planarity. The test vehicle is again a large body sized 176LQFP package. The device is assembled through a normal QFP assembly process flow and then subjected to package warpage measurements using an Akrometrix Thermoire system. Fig. 9 shows the warpage measurement for compound C and D throughout a simulated reflow process. Lead co-planarity of compound C post assembly has a lower maximum value as compared to compound D, which is consistent with the package warpage measurements. Both compounds showed reduced co-planarity values after subsequent simulated heat treatments. After final heat treatment through the simulated reflow, compound D once again has higher co-planarity. D. Package Delamination Based on the package warpage and lead co-planarity results, compound C is preferable in terms of manufacturability. However, it is necessary to assess the reliability performance of compound C, particularly in terms of package delamination. 176LQFP samples assembled using a standard assembly process flow were subjected to MSL3/260 o C preconditioning, followed by temperature cycling at -55 o C to 150 o C, up to 2000 cycles. The samples were then subjected to scanning acoustic tomography (SAT) to check for any interfacial delamination. Fig. 11 shows the SAT results. (a) (b) Fig. 9. Package warpage measurements for compound C vs compound D Thermoire warpage measurements from room temperature up to 260 o C and back consistently showed compound D having more severe warpage. This is attributed to the higher mold shrinkage % of compound D. Post assembly lead co-planarity was measured using an ICOS T120 system. The samples were then subjected to simulated heating as experienced by units through burn-in, dry-baking before packing, and finally reflow. ICOS scanning were performed after each heat simulation stage. Fig. 10 shows the comparison of lead co-planarity for compound C and D. Fig. 10. Lead co-planarity measurements for compound C vs compound D Fig. 11. (a) Top scan, and (b) through scan results No delamination was found on any critical area, and meets AEC-Q100 and AEC-Q006 requirements. Only minor delamination at the die paddle tie bar was observed, but this is a non-critical area and is acceptable. IV. SUMMARY Automotive semiconductors are becoming increasingly common in new generation of vehicles. These devices require increasingly higher reliability levels for high temperature applications. In order to meet the high level of reliability requirements with competitive costs, it is necessary to select the proper bill of materials for semiconductor packaging. Palladium coated copper wire has been shown to provide the best overall cost of ownership in terms of raw material cost and throughput efficiency. It was shown that good high temperature storage life reliability at 175 o C can be achieved by selecting PCC wire with improved Pd coating uniformity. This reduces the risk of galvanic corrosion on PCC wire due to sulfur content within the molding compound. Proper mold compound selection is crucial to enable the most robust manufacturability performance. Mold compound

5 with longer spiral flow results in significantly better wire sweep performance, particularly on high density leadframes with multiple rows of units. Package warpage is an important consideration to ensure good lead co-planarity of the package. Last but not least, the selected mold compound must exhibit good package delamination performance as per AEC-Q100 and AEC-Q006 requirements. ACKNOWLEDGEMENT The authors would like to thank the members of the Reliability Lab and Failure Analysis Lab staffs of NXP Semiconductors for supporting this study. REFERENCES [1] C. C. Lee, T. Tran, D. Boyne, L. Higgins, and A. Mawer, Copper versus palladium coated copper wire process and reliability differences, 2014 Electronic Components & Technology Conference, 2014, pp [2] F. Quercia, and A. Mancaleoni, Copper wire in automotive: key challenges and robust validation, 6 th Electronic System-Integration Technology Conference, 2016, pp [3] T. Saruwatari, T. Takahashi, A. Ono, Y. Asano, T Iwasaki, M. Ooida, and Y. Hiruta, Reliability improvement of Cu-wire bonded lead frame package for automotive applications, 2017 Pan Pacific Microelectronics Symposium, 2017, pp [4] J. H. Jeon, S. H. Na, S. H. Jeon, M. Mo, D. B. Kang, K. M. Lim, and J. Y. Kim, High reliability challenges with Cu wire bonding for autoomtive devices in the AEC-Q006, IEEE 67 th Electronic Components & Technology Conference, 2017, pp [5] S. K. Teng, R. Ibrahim, and S. H. Teh, Mold compound selection study for CMOS lead LQFP 24 24mm package, 35 th IEEE/CPMT International Electronics Manufacturing Technology Conference, 2012, pp [6] M. M. Fernandez, M. T. Bauca, and R. J. C. Malifer, Leadframe-tomold adhesion performance of different leadframe surface morphologies, 18 th Electronics Packaging Technology Conference, 2016, pp