6-1. Conclusions
A novel wrong wire bonding prevention (WWBP) system was proposed and implemented that can automatically inspect the correctness of wire bonding before the bonding is physically executed.
The proposed CAD-based vision approach of WWBP system has been implemented on several different models of bonding machines such as the KNS MaxumPlus, KNS MaxumUltra, Shinkawa UTC370, and Shinkawa UTC1000. It is the first system that can automatically check the correctness of wire bonding positions with respect to the designated CAD drawing. It can completely solve the mal-detection and lost detection problems that may occur in other available methods and save the manual effort, material cost and production time in bonding a sample for inspection. The experimental results showed that the proposed WWBP system is very efficient and effective for both single-layered wire ICs and multi-layered wire ICs. It also shows that the proposed WWBP system is better for leadframe-based material and ground bond products than other available methods. The proposed WWBP system can fully prevent wrong wire bonding in the entire wire bonding process.
Sometimes IC packing foundries might not have the chance to take part in the design of CAD drawing and cannot get the CAD drawing. When the original CAD drawing is not available, the proposed WWBP system plus SBP will be applied.
6-2 Contributions
1. A set of efficient and effective algorithms for wire bonding position inspection are proposed.
The proposed WWBP system can be applied to inspect the correctness of the wire bonding position on a multi-layered wire IC and fully solve the mal-detection and lost detection
proposed system is particularly good for the leadframe-based material ICs and the product that has ground bond.
2. A complete system solution was proposed and implemented that can automatically inspect the correctness of wire bonding and fully prevent wrong wire bonding throughout the entire wire bonding process.
3. The proposed WWBP system can inspect the correctness of the wire bonding position before the bonding is physically executed. It does not need to bond actual samples for testing and can save material cost and save material transfer time. In mass production environment, the proposed system is fast enough to work synchronously with the wire bonding process.
4. A new direction for applying CAD-based vision techniques for complicated IC parts inspection.
6-3 Further Studies
When production engineers set up the first bonding machine or duplicate the bonding program from the first setup machine into other bonding machines, engineers must calibrate the bias of the wire bonding positions for each of the bonding machines manually. To improve the efficiency and prevent human errors, a method that can automatically calibrate the bias of the wire bonding positions, in stead of manual operation, is helpful and worth pursuing.
References
[1] AUTOOLP2006. Kulicke & Soffa Industries Inc., Available from http://www.kns.com/KNS07/Templates/ShowPage.asp?DBID=1&LNGID=1&TMID=108&FI D=507&PID=1312&IID=1507 [Accessed 16 February 2009]
[2] Ahmed, M., Cole, C.E., Jain R.C. and Rao, A.R., “INSPAD: A System for Automatic Bond Pad Inspection”, IEEE Transactions on Semiconductor Manufacturing, pp. 145-147, 1990.
[3] Khotanzad, A., Banerjee, H. and Srinath, M.D., “A Vision System for Inspection of Ball Bonds and 2-D Profile of Bonding Wires in Integrated Circuits”, IEEE Transactions on Semiconductor Manufacturing, 7(4), pp. 413-42, 1994.
[4] Khotanzad, A., Banerjee, H. and Srinath, M.D., “A Vision System for Inspection of Ball Bonds in Integrated Circuits”, Proc. IEEE Workshop on App. of Comp. Vision, 290-297, November 1992.
[5] Sreenivasan, K.K., Srinath, M.D. and Khotanzad, A., “Automated Vision System for Inspection of IC Pads and Bonds”, IEEE Transactions on Components, Hybrids, and Manufacturing Technology, 16(3), pp. 333-338, 1993.
[6] Tsukahara, H., Nakashima, M. and Sugawara T., “Automated Visual Inspection System for IC Bonding Wires Using Morphological Processing”, Proceedings of SPIE, 1384, pp. 15-26, November 1991.
[7] Imagen, Inc. Available from: http://www.imagen-inc.com [Accessed 16 February 2009].
[8] Machine Vision Products Inc. Available from: http://www.visionpro.com [Accessed 16 February 2009].
[9] Tsai, D.M. and Su, Y.J., “Non-referential, Self-compared Shape Defect Inspection for Bond Pads with Deformed Shapes”, International Journal of Production Research, 47(5), pp.
1225-1244, 2009.
[10] Ayoub, G.T., “Inspecting Post Wire-bond Interconnects”, Advanced Packaging magazine, April
2006. Available from:
&VERSION_NUM=2&p=36 [Accessed 16 February 2009].
[11] Wang, M.J., Wu, W.Y. and Hsu, C.C., “Automated Post Bonding Inspection by Using Machine Vision Techniques”, International Journal of Production Research, 40(12), pp. 2835-2848, 2002.
[12] Ngan, K.N. and Kang, S.B., “Automated Inspection of IC Bonding Wires Using Hough Transform”, 14th Conference of Industrial Electronics Society, 4, pp. 938-942, October 1988.
[13] Ye, Q.Z., Ong, S.H. and Han, X., “A Stereo Vision System for the Inspection of IC Bonding Wires”, International Journal of Imaging System Technology, 11(4), pp. 254-262, 2000.
[14] Perng, D.B., Chou, C.C. and Lee, S.M., “A New Wire Bonding Inspection System by Machine Vision”, 17th International Conference of Production Research, August 2003.
[15] Perng, D.B., Chou, C.C. and Lee, S.M., “Design and Development of a New Machine Vision Wire Bonding Inspection System”, The International Journal of Advanced Manufacturing Technology, 34(3-4), pp. 323-334, 2007.
[16] Perng, D.B., Chou, C.C. and Lee, S.M., “Illumination System for Wire Bonding Inspection”, Applied Optics, 46(6), pp. 845-854, 2007.
[17] Sanderson, A. C., Weiss, L. E. and Nayar, S. K., “Structured Highlight Inspection of Specular Surfaces”, IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(1), pp. 44-55, 1988.
[18] Chia, T. L., Chen, Z. and Yueh, C. J., “Curved Surface Reconstruction Using a Simple Structured Light Method”, IEEE Proceeding of the 13th International Conference on Pattern Recognition, pp. 844-848, August 1996.
[19] Tian, Y. L. and Tsui, H. T., “3D Shape Recovery from Two-Color Image Sequences Using a Genetic Algorithm”. Proceedings of the 13th International Conference on Pattern Recognition, 4, pp. 674-678, August 1996.
[20] Kim, M. Y. and Koh, K. -I., “Shadow-Free Moire Interferometer with Dual Projection for In-Line Inspection of Light Emitting Diodes”, International Journal of Optomechatronics, 1(4), pp. 404-424, 2007.
[21] Lim, H.O., Zhang, W. and Koh, L.M., “Automated Visual Inspection for IC Wire-bond using Auto-focusing Technique”, IEEE/CHMT International Electronics Manufacturing Technology Symposium, pp. 31-36, October 1992.
[22] KAIJO Corporation. Available from: http://www.kaijo.co.jp [Accessed 16 February 2009].
[23] WI-110. Sanlien Technology Corp. Available from:
http://www.sanlien.com.tw/web/homepage.nsf/foundationview/B2A893A7D1C934794825718 C0011F969 [Accessed 16 February 2009].
[24] Pacheco, M., Wang, Z., Skoglund, L., Liu, Y., Medina, A., Raman, A., Dias, R., Goyal, D. and Ramanathan, S., “Advanced Fault Isolation and Failure Analysis Techniques for Future Package Technologies”, Intel Technology Journal, 9(4), pp. 337-352, 2005.
[25] Bonder Connectivity. Kulicke & Soffa Industries Inc. Available from:
http://www.kns.com/KNS07/Templates/showpage.asp?DBID=1&LNGID=1&TMID=87&FID
=507&PID=1312 [Accessed 16 February 2009].
[26] Ng, H.F., “Automatic Thresholding for Defect Detection”, IEEE Proceedings of the Third International Conference on Image and Graphics, pp. 532-535, December 2004.
[27] Maxum Ultra Datasheet. Kulicke & Soffa Industries Inc. Available from:
http://www.kns.com/UPLOADFILES/DGALLERY/WB2.PDF [Accessed 16 February 2009].
Appendix A: The leadframe design and manufacturing process
The leadframe manufacturing process is comprised of 3 steps, as illustrated in Figure A1, and is described below.
Figure A1. The process flow of leadframe manufacturing.
(1) Stamping/Etching process:
Stamping is an automated, high-speed process suitable for large production rates that justify the high initial tooling expense. The process is typically accomplished as a series of stamping operations that progressively approach the final leadframe geometry—the number of steps dependent on the geometrical complexity of the leadframe. Tooling development requires long lead-time and large investment in equipment.
Photochemical etching is one of the most widely used methods for manufacturing leadframes. It can be tooled with low costs and minimal time requirements. Etched leadframes are manufactured in flat sheets, made of either copper or Alloy 42, on which both sides are coated with photoresist film. Using the photofabrication process, a thin metal plate is processed into a leadframe.
(2) Plating process: Leadframes are plated with full plating or partial plating at the required location to improve the surface properties. Partial plating is one of the most widely used methods for manufacturing leadframes with silver, nickel or gold at the inner leads to facilitate wire bonding.
(3) Taping/Downset/Cut process:
Taping process: To maintain inner leads position, the inner lead is locked with polyimide tape, as illustrated in Figure A2.
Figure A2. The inner lead is locked with polyimide tape (From : http://www.dnp.co.jp).
Downset process: The die pad section onto which the die will be bonded is depressed by downset tool, as illustrated in Figure A3.
Figure A3. The cross-section view of downset processed section (From : http://www.dnp.co.jp).
Lead tip cut process: For preventing the lead from changing its shape during the plating process, and for accurately maintaining the inner lead position achieved by the photo fabrication process. The connected portions are then cut off in the final process, as illustrated in Figure A4.
Figure A4. A single leadframe unit before and after the cutting process.
Before the stamping process, the etching process, and the partial plating process, leadframe vendors need to design and manufacturing the tooling or mask. The tooling and mask design will be based on the original CAD drawing from package foundry and make appropriate adjustment.
The main adjustments will consider the following rules:
1. Reserve require thickness for etching and electroplating, so that the size of the final product is as much as possible in line with the specifications
2. Precision varies subjected to each vendor’s equipment and manufacturing capabilities. It needs to set aside space to make the final product size can be as much as possible in line with the specifications.
3. Etching rate at different locations are different. Figure A(b) shows that there are different etching rates at different regions. It requires to appropriately adjust the thickness at different locations.
4. Consider the location and width of the angle and see if it will affect the etching rate
Figure A5. Three are different etching rates at a, b, and c three regions.
The adjustment of tooling and mask design will make the final product is not exactly same as original CAD drawing. At the time of acceptance for leadframe, packaging plant does not require to be exactly the same as original CAD drawing. Packaging foundry sets accuracy specification as
“Acceptable＂approximately between ± 0.8 mil ~ ± 2 mils. But it must meet the following rules:
1. Lead pitch (see Figure A6) must be greater than the specification.
2. Lead width must be within the specification (width ± tolerance).
3. Edge lead position must be within the specification (edge distance to reference point ± tolerance) or the axial of the middle lead must be within the specification (see Figure A7).
Figure A6. An enlarged image of CAD drawing.
Figure A7. The axial of the middle lead and the specification.
Beside the adjustment of tooling/mask design, the leadframe in the production process would have different variation in size and appearance for many reasons, for example:
1. Corner will be over-etching, , as Figure A8
2. Change of level of chemical concentration would cause different level of etching
3. Testing and packaging in the manufacturing process would cause the lead position shift away.
Figure A8. Over-etching of the corner.
Appendix B: Variation issues during wire bonding process
A number of products would experience variation issues during wire bonding process. The major variations include variation from machine to machine, platen issue, lead position variation, bad or contaminated electroplating, and thermal-induced deformation or shrinkage. The detail is as follow:
1. Variation from machine to machine. All wire bond machines have a certain degree of variation. Among them, the difference of light sources would have the greatest impact on wire bonding positions. Therefore, it is necessary to re-adjust the reference points and re-adjust all coordinates of wire bonding positions. As shown in Figure B1, strength of different lights will result in change of lead width and deviation of wire bonding positions from the axial position.
(a) Strong light (b) Weak light
Figure B1. The wire bonding position will shift away the axial from the strong light to weak light.
2. Platen issue. Bonding machine depends on platen and back-plate to fix each lead’s position of the leadframe and to provide sufficient stiffness during wire bond process (see Figure B2). But if platen and back-plate have poor flatness, it will make the leads shift away the original positions during lamination process.
Figure B2. The platen and leadframe in the wire bond working area (From http://www.kns.com).
3: Lead position variation. As with the lead position variation, the leadframe-based material is fabricated by punching or etching and has the problem of high variation and low accuracy for each lead. This implies that in the punching or etching process a lead will invariably be shifted and the engineer must adjust the wire bonding positions from time to time in order to fit the variation.
Figure B3(b) shows that one lead of a leadframe-based material is slightly shifted to the right side.
Figure B3. (a) An enlarged image of a normal leadframe-based material. (b) One lead is shifted to the right side.
4. Bad or contaminated electroplating. If there is bad electroplating or contamination in the actual bonding position, it will result in not being able to perform wire bonding. At this time, it should adjust wire bonding positions to avoid the problematic regions. Figure B4(a) shows that a contamination in the lead and figure B4(b) shows a bad electroplating example.
(a) Contamination (b) Bad electroplating Figure B4. (a) An enlarged image of a contamination in the lead. (b) An enlarged image of bad
electroplating in the lead.
5. Thermal-induced deformation or shrinkage. Wire bond regions (see Figure B5) will be heated to approximately 200 degree to provide better bonding of the molecules. In general, bonding machine would use several heaters. The temperature difference throughout the wire bond region is about 10 degree. During entire heating process, leadframe would experience deformation due to thermal. The magnitudes of the deformation will depend on how long the leadframe stays in the wire bond working region.
Figure B5. Image of wire bond working region.
Before the actual bonding, the bonding machine will use relevant matching method to inspection all leads. If there are any lead position can not be correctly compared due to lead position variation, poor electroplating, or contamination, the bonding machine will stop and wait for adjustment. Until complete successful comparison, automatic wire bonding could then proceed.
This could help to avoid the bonding ball is outside the lead. For lower-order products (larger lead width), functions of lead position comparison is usually not activated in order to increase the throughput.
Appendix C: The rotation angle calibration method for IC chip rotation bias
Figure C1 (a) shows the rotation angle of the chip in the original design (CAD file) is θ1-Φ1.
P1’ and P2’ are the two calibration marks in the pad side. Figure C1 (b) shows the actual rotation angle of the chip in the WVM is θ2-Φ2. P1 and P2 are the two calibration marks on the pad side.
The θ1 should be equal to θ2. Then the rotation bias (∆θ) between CAD file and WVM is Φ2-Φ1.
And the value of Φ1 and Φ2 can be calculated as equation (1) and (2).
Φ1 = tan−1( Figure C1. Chip rotation angle of original design and WVM.
Assumed there is a bonding position P (C Xs,Ys ) in the pad side (see Figure C2). After rotating to the same angle as in the CAD file, the new coordinate values Xs’,Ys ’ can be calculated as below :
φS = tan−1(
Figure C2. The angle between any one bonding point P and the reference point P1. C