Bondability, Reliability and Yield Benchmarks for High Volume Specialty Gold Fine Bonding Wire

1. Bondability, Reliability and Yield Benchmarks for High Volume Specialty Gold Fine Bonding Wire By Heiner Lichtenberber, Micheal Zasowski, Gery Lovitz & Daniel Ha For Williams Advanced Materials Buffalo, NY and YCMC Taipei, Taiwan Presentation by: Prasad Weerakoon Date: September 19th , 2007

2. Introduction • The demand for smaller diameter wires is increasing • There is a need for processes used to make this wires to be made simpler. • This paper studies the effects of controlling the dopant chemistry, break strength and elongation on production of these fine wires.

3. Background • Low loop and long length characteristics require a short HAZ which is controlled by the additions of certain metals in the parts per million range. • Be and Ca are used to obtain high strength and recrystallization temperature. • Bonding wire is produced by a combination of continuous casting and step drawing processes.

4. Parameters • The effects of deliberate additions of Be, Ca, Pt, Cu, Al and Ag are discussed. • A gold wire was used. • The effects of some 20 elements were determined using a DOE software. • Once the desired chemistry of the dopants were determined, the properties of the wire was compared to the hydrostatically extruded test samples.

5. Correlation of HAZ to Recrystallization Temperature The recrystallization temp. is inversely proportional to the HAZ as a higher temp means that the wire has to get hotter to recrystallize the grain structure. BL- Break load

6. Experimental Procedures-Chemical • To determined the effects of minor additives to high purity gold, binary gold alloys of Be, Al, Ca, Cu, Ag and Pt and multi-component of alloys were prepared. • These were vacuum melted, annealed and hydrostatically extruded to 0.001” diameter. • Individual dopant levels varied from 3 to 30 parts per million. • Analysis were done using a software (ICP-MS).

7. The hydrostatic extrusion equipment used in the experiment has processing pressures from 100,000 psi to 200,000 psi depending on the strength of the ingot. Extrusion speeds varied from 250fpm at large diameters to over 1000fpm at final size.

8. Experimental Procedures-chemical Contd. • Break load and elongation were determined. • Tests were done for wires at their extruded condition and after annealing to 2.5% elongation. This picture shows the large grain structure in the ball of the wire and how it gets finer further away from the ball

9. Experimental Procedures- Mechanical • Continuous casting was used to produce 0.312” wire. • Continuous casting was chosen since it is superior to ingot casting because the integrity and the grain structure of the casting is uniform throughout the length. • This method is done by melting the metal in a graphite crucible with a graphite die located at a bottom.

10. Experimental Procedures- Mechanical Contd. • After the final annealing to the desired properties, the wire was bonded onto ceramic substrates (ICs)using a machine. • The results were tested both visually and electrically to determine the consistency of the wire interconnections. The picture shows the ceramic being connected to the wires.

11. Results-Chemical • Dopants showed some increase in strength in the hard as extruded condition • Be and Ca had the greatest effect while Cu and Ag had the least. • Ca and Be had the most significant strengthening effects after annealing even though most dopants strengthened the wire to some extent during extrusion and drawing.

12. Break Strength Vs. PPM of dopant We can see that Be has an immediate strengthening effect while Ca requires a minimum level before any strengthening can occur.

13. Recrystallization temp Vs. PPM of dopants This graph shows that calcium had the most significant effect on the recrystallization temp while silicon had a negative effect.

14. Results-Mechanical • Wires produced by continuous casting, wire drawing and hydrostatic extrusion produced identical mechanical and physical properties as those processed by hydrostatic extrusion alone. • Wire drawing is best for high volume production as it can handle long wire lengths. • Complete numerical values of the results are discussed in the handout

15. Wire pull tests Wire pull tests were carried out at 24 hour intervals. Ball shear strength showed a slight increase at 48hrs and held steady throughout the test.

16. X-ray images • X-ray images were taken of the ICs to show the sweep of the wires. The worst case deviation was 2.98% which was within the specifications and tolerances.

17. Discussion • Since the atomic size and lattice parameters of Ag, Cu, Al, and Pt are similar to that of Au, they had little effect on the properties of gold. • Since Be is 30% smaller than Au, Be atoms can occupy interstitial sites in the gold lattice, thus hardening and strengthening the gold. • Ca is 30% larger than gold and incorporates itself to the grain boundaries of gold. This increases the recrystallization temp.

18. Summary • By choosing the appropriate dopants in the correct level, a series of alloys that exhibit low loop and long length characteristics can be produced. • Consistency of the bonding wire and other characteristics are due to the chemical formulation of the bonding wire and the manufacturing method.

19. References (1)G. G. Harman, “Wire Bonding in Microelectronics”, McGraw-Hill 1997 (2)H. Lichtenberger, et.al, “Gold Bonding Wire – The Development of Low Loop, Long Length Characteristics” IPMI 1998