Extended reliability of gold and copper ball bonds in microelectronic packaging
- 4.4k Downloads
Wire bonding is the predominant mode of interconnection in microelectronic packaging. Gold wire bonding has been refined again and again to retain control of interconnect technology due to its ease of workability and years of reliability data. Copper (Cu) wire bonding is well known for its advantages such as cost-effectiveness and better electrical conductivity in microelectronic packaging. However, extended reliabilities of Cu wire bonding are still unknown as of now. Extended reliabilities of Au and Pd-coated Cu (Cu) ball bonds are useful technical information for Au and Cu wire deployment in microelectronic packaging. This paper discusses the influence of wire type and mold compound effect on the package reliability and after several component reliability stress tests. Failure analysis has been conducted to identify its associated failure mechanisms after the package conditions for Au and Cu ball bonds. Extended reliabilities of both wire types are investigated after unbiased HAST (UHAST), temperature cycling (TC), and high-temperature storage life test (HTSL) at 150, 175, and 200 °C aging temperatures. Weibull plots have been plotted for each reliability stress. Obviously, Au ball bond is found with longer time to failure in unbiased HAST stress compared to Cu ball bonds for both mold compounds. Cu wire exhibits equivalent package and or better reliability margin compared to Au ball bonds in TC and HTSL tests. Failure mechanisms of UHAST and TC have been proposed, and its mean time to failure (t50), characteristic life (t63.2, η), and shape parameter (ß) have been discussed in this paper. Feasibility of silver (Ag) wire bonding deployment in microelectronic packaging is discussed at the last section in this paper.
KeywordsGold bonding wire Copper bonding wire Silver wire Extended reliability Microelectronic packaging
Gold bonding wire has been extensively used for the fabrication of integrated circuits because of its good electrical conductivity and mechanical stability with a diameter of 20 μm or less. With significant increases in gold price, gold ball bonding has become a more costly process that has a considerable economic effect on the assembly of packages used in consumer electronics. An alternative wire material to gold is copper, which is much cheaper, has several technical benefits including better electrical conductivity, and has been widely used in discrete and power devices with wire diameters typically larger than 30 μm in diameter for many years. The potentials and cost considerations of finding an alternative to replace gold wire bonding in microelectronic packaging are driven by new technologies coming to the market . Copper wire bonding appears to be the alternate materials, and various engineering studies on copper wire deployment have been reported [2, 3]. The Au–Al intermetallic compound (IMC) growth is widely characterized and analyzed [4, 5, 6]. Zulkifli MN et al.  suggested new approaches: examining the effect of individual phase and surroundings on the strengthening produced by the Au–Al intermetallic compound, combining FEA based on friction and wire-bonding parameters, and correlating TEM results with results obtained from other techniques should enable a more detailed understanding of the bondability and strength of thermosonic gold wire bonds. Key technical barriers such as intermetal dielectric cracking due to excessive bonding, copper ball bond corrosion under moist environments, and extended reliability of copper ball bonds are identified accordingly [8, 9, 10]. Copper ball bond is more susceptible to moisture corrosion compared to gold ball bonds and undergoes different corrosion mechanisms in microelectronic packaging [11, 12]. Our previous studies indicate that Pd-coated copper ball bond outperforms gold ball bonds in biased HAST wearout reliability . Extended reliability of high-temperature storage life (HTSL) of copper ball bonds in TSOP package is found with apparent activation energy (Eaa) of ~0.70 eV compared to gold ball bonds . Blish et al.  investigated Eaa of typical Au–Al IMC of 1.0 ~ 1.5 eV, which is Al thickness dependent. Hence, extended reliability is crucial to determine the lifetime of gold and copper ball bonds in microelectronic packaging. The Cu–Al and Au–Al IMC growth kinetics were studied, and it was found that Cu–Al growth is at least 5× slower than Au–Al IMC . However, copper wire bonding still pose reliability challenges and complex failure mechanisms which could be the main barriers to entirely replace gold wire bonding . In this study, we have prepared FBGA 64 package assembled with gold and Pd-coated copper wire and load for unbiased HAST (UHAST), temperature cycling (TC), and HTSL tests.
Will Cu or Ag wires entirely replace Au wire bonding?
In general, Cu wire is not an ultimate bonding wire solution in semiconductor packaging. Cu wire bonding is more suitable to be deployed in low-pin-count semiconductor packaging, flash memory packaging, or high-power devices which utilize a larger diameter for bonding wire. The various considerations such as its long-term extended reliability performance and bondpad cratering challenges still pose a showstopper for full sweep of copper wire bonding in semiconductor packaging. Undeniably, the improved N2 kit (which is installed on wire bonder) will improve the wire-bonding process with an inert environment since Cu wire is vulnerable to corrosion and oxidation in production floor. Cu wire will not entirely replace conventional Au wire bonding in semiconductor packaging but rather another option of packaging methods other than Au, Al, and possibly Ag wire bonding.
Au, Cu, or Ag wire alloys for semiconductor packaging?
Material properties of bare Ag, Au, and Cu
Au wire exhibits excellent UHAST extended reliability and more stable assembly processes (in terms of shear strength and wire pull strength in the as-bonded stage). This is the most pivotal deciding factor for keeping Au wire bonding in some of the customer-end field applications such as medical, automotive, and military market segments. Silver (Ag) wire bonding is still a new interconnect method in semiconductor packaging and yet to be widely adopted by major semiconductor companies due to lack of reliability data and further engineering evaluations. Many claims on its advantages of moderate AgAl IMC formation and growth rate, and easier pluck-and-play for mass production are key success factors for Ag wire to replace Au or Cu wire bonding. Cho et al.  reported that Pd alloying of the Ag wire was effective in improving the reliability of Ag ball bond. The lifetime in PCT increased with increasing Pd concentration in the Ag wire. Free air ball formation is found better in Ag–Au–Pd compared to 2 N Ag wire alloy . The bonding process of Ag wire bonding is pretty similar to Au wire bonding . Another bondability study is conducted on Au–Ag wire alloy, and caution should be given to bonding temperature and first ball bond parameter setting . This observation convinces the great opportunity of using Ag–Au–Pd alloy instead of bare 2 N Ag wire in microelectronic packaging. Another reliable Ag–8Au–3Pd wire alloy is found with high reliability and low electrical resistivity, which is processed with annealing twins [21, 23, 24]. Bare Ag wire or tertiary Ag alloy (such as Ag–8Au–3Pd or Ag–Au–Pd) are identified as next potential candidates of microelectronic packaging. In a nutshell, Au, Ag, and Cu (bare or Pd-coated Cu) wire alloys will exist as three alternatives of wire-bonding techniques in semiconductor packaging based on its customer-end field applications and packaging cost considerations. The packaging cost of Ag wire bonding is moderate and in between of Au and Cu wire bonding.
Future of Au wire bonding in semiconductor packaging
Au wire bonding will still exist in microelectronic packaging in view of its process stability and higher moisture reliability margin compared to recent penetration of Cu wire bonding. The primary motivation for uptake of copper wire bonding is, however, more strongly cost driven rather than motivated by very clear and distinct process, performance, or reliability advantages . In our extended reliability study, Au ball bond exhibits higher UHAST wearout reliability margin compared to Cu ball bonds for epoxy mold compound (EMC) A. This is one of the most important factors to keep Au wire bonding in microelectronic packaging, especially for high-pin-count semiconductor packaging. Cu is known for more susceptible corrosion activity under moist environments. End customers such as automotive, military, and medical industries had expressed concerns over massive transition from Au to Cu wire bonding. Ag wire bonding, however, is not a mature assembly process, and it requires further engineering studies on IMC formation, extended reliability, and test yield analysis on flawless high-volume manufacturing process to validate the feasibility of Ag wire bonding in microelectronic packaging. There is clearly a place for copper (bare copper or Pd-coated copper wire) or silver (bare silver or Ag–Au–Pd alloy) wire bonding in microelectronic packaging, but it is likely that rather than replacing gold wire entirely, copper or silver wire bonding will become another option alongside gold wire bonding which microelectronic package designers can consider for package assembly.
Summary of experimental matrix (for Au and Cu wires)
Mold compound type
Extended reliability test
85%RH, 130 °C
-40 to 150 °C
85%RH, 130 °C
-40 to 150 °C
Another set of materials was used to estimate the apparent activation energies (Eaa) of Au and Cu ball bonds assembled with EMC A and EMC B. The key materials used include 0.8 mil Pd-coated Cu wire and 4 N (99.99 % purity) Au wire, fine pitch BGA package, and 110-nm device which is packed in fortified fineline BGA package, green (<20 ppm chloride in content) in molding compound and substrate. All direct material used in this evaluation study for the 110 nm, flash device (with top Al metallization bondpad) for packaging purpose. A total of six legs of 45 units of Au and Pd-coated Cu wire bonded on fine pitch 64-ball BGA packages are subjected to 150, 175, and 200 °C aging temperatures. Electrical testing was conducted after each hour and cycle of stress to check Au and Cu ball bond integrity in terms of its ball bond HTSL reliability with various aging conditions.
Result and Discussion
Summary of extended reliability results (mold compound types A and B of Au and Cu wires)
85%RH, 130 °C
-40 to 150 °C
85%RH, 130 °C
-40 to 150 °C
85%RH, 130 °C
-40 to 150 °C
85%RH, 130 °C
-40 to 150 °C
Key material characteristics of EMCs A and B
Linear coefficient of thermal expansion 1 (CTE, α1)
1.1 ± 0.3
0.7 ± 0.3
Linear coefficient of thermal expansion 2 (CTE, α2)
4.5 ± 1.0
3.0 ± 1.0
Glass transition temperature (Tg)
125 ± 15
125 ± 15
Failure analysis and mechanisms of Au and Cu ball bonds
Microstructural analysis of failed ball bonds
EDX analysis of failed Cu and Au ball bonds after the UHAST test
Element (atomic %)
High-temperature storage life
Summary of Eaa and HTSL failure mechanisms from previous studies
Ball bond type
HTSL aging test conditions (°C)
150, 175, 200
150, 175, 200
0.72 ~ 0.83
150, 175, 200
1.00 ~ 1.50
1.00 ~ 1.26
150, 175, 200
0.92 ~ 1.10
Ball bond lifetime analysis by using Eaa (apparent activation energy)
Lifetime estimations for various market segments (Cu wire bond FBGA package)
Typical lifetime (years)
# of lives (years) Cu EMC A
# of lives (years) Cu EMC B
Effects of molding compound on extended reliability
The EMCs used in this evaluation are from suppliers A and B. The important material characteristics of the mold compound datasheets are given in Table 4. The only difference between the mold compounds is that they are from different mold compound manufacturers. EMC A exhibits higher hours to failures in UHAST extended reliability for EMC A (Fig. 1) but lower cycles to failures in TC extended reliability tests (Fig. 2) compared to EMC B. In HTSL tests, EMC B shows no significant difference in the apparent activation energy (Eaa) value in Cu ball bonds while a much lower Eaa value in Au ball bonds (Fig. 13). This proves that the different types of epoxy mold compounds have a significant influence in the HTSL test of Au wire bonding. Au ball bonds are well known for its higher IMC growth rate and increased susceptibility to Kirkendall microvoiding after the HTSL test compared to slower CuAl IMC growth rate. Hence, we observed no significant Eaa values obtained in Cu ball bonds for EMC A and EMC B (Figs. 14 and 13). Both EMCs show promising extended reliability results which far exceed the typical 96 h of UHAST and 1,000 cycles of TC according to JEDEC standards. Hence, both EMC A and B are used in our flash memory BGA laminate.
Effects of wire types on extended reliability
Many previous studies reported Au wire bonding with higher reliability margins compared to Cu wire bonding. However, there are very few published data on extended reliability of Au and Cu ball bonds. In our study, Au ball bonds show higher UHAST reliability compared to Cu ball bond in FBGA package. This is notable as Au is much more stable and has higher corrosive resistance compared to Cu. Cu is easily oxidized and corroded under moist environments, especially in the UHAST or biased HAST tests. Our extended reliability study (Fig. 1) shows similar findings with Au wires in EMC A. Another factor affecting the first ball bond strength is Au wire bond shear or wire pull strength shows less variation compared to Cu ball bonds. This as-bonded stage strength value would influence the reliability of ball bonds in semiconductor packaging.
However, Cu ball bonds exhibit higher cycle to failure in TC test compared to Au ball bonds (Fig. 2) regardless of EMC types. The CTEs of Cu and Au are pretty similar in this case. This is an interesting finding since the reliability performance of TC stress is pretty much material CTE dependent with regard to Al bondpad. The mismatch in CTE between the Cu (17 ppm/°C) and Au ball bonds (14 ppm/°C) to the silicon die (3.0 ppm/°C) induced different thermal expansions and contraction rates in the temperature cycling test. The CTE mismatch between the Au and Cu ball bonds with Al bondpad of silicon die will impose different thermal expansion rates during hot cycles (150 °C) and contraction rates during cold cycles (-40 °C).
Future works and recommendation
Cu wire will be continuously developed to replace Au wire in higher pin counts of semiconductor packages, but transition is predicted to be less on power-device-based packages. Future engineering work should be focused on knowledge-based reliability testing and prediction to understand the initial failure point in semiconductor device packaging. Extended reliability concept would be used in this type of reliability studies. Further characterization should be carried out for Pd–Ag–Au or bare Ag wire bonding in nanoscale device packaging, especially for 45, 28, 22, or subnanoscale 10 nm below technology nodes.
In this research, we analyzed the effects of wire alloy on extended reliability of UHAST, TC, and HTSL stresses. Au ball bonds show a significant higher UHAST reliability compared to Cu ball bond in FBGA package with EMC A. This is notable as Au is much more stable and has higher corrosive resistance compared to Cu. Contrary results occur in TC, where Cu ball bond is more superior compared to Au ball bond. EMC B exhibits higher TC reliability margins compared to EMC A assembled with Au or Cu wires. However, both EMCs are far exceeding the minimum required 96 h of UHAST and 1,000 cycles of TC according to JEDEC standards. The Eaa values obtained for Au ball bonds range from 0.92 to 1.10 eV and 0.72 to 0.83 eV for Cu ball bonds. These values are close to previous HTSL studies conducted on Au and Cu ball bonds. Au wire bonding will still remain as a mainstay in microelectronic packaging, especially for more complicated semiconductor packages (with higher pin counts), while Cu wire bonding will equally gain some market shares in low-pin-count and power device packaging. Ag wire bonding would probably become an emerging technology as an option in microelectronic packaging. However, more engineering works should be carried out to understand the extended reliability performance as well as assembly yield monitoring before deployment for high-volume manufacturing. Future engineering work should be focused on knowledge-based reliability testing and prediction to understand the initial failure point in semiconductor device packaging. Extended reliability concept would be used in this type of reliability studies.
The authors would like to take this opportunity to thank Spansion management (Gene Daszko, Tony Reyes, and Chong HL) for their management support for the paper publication.
- 5.Muller T, Schraplerl L, Altmann F et al (2006) Influence of intermetallic phases on reliability in thermosonic Au–Al wire bonding. IEEE Electronics System Integration Technology Conference, Dresden, pp 1266–1273Google Scholar
- 8.Gan CL, Toong TT, Lim CP, Ng CY (2010) Environmental friendly package development by using copper wire bonding, In Proceedings of 34th IEEE CPMT IEMT. Malacca 2010:1–5Google Scholar
- 10.Gan CL, Ng EK, Chan BL, Hashim U (2012) Reliability challenges of Cu wire deployment in flash memory packaging. IEEE Proceedings of International Microsystems, Packaging, Assembly and Circuit Technology Conference, Taipei, pp 498–501Google Scholar
- 11.Breach CD, Shen NH, Lee TK, Holliday R (2011) Failure of gold and copper ball bonds due to intermetallic oxidation and corrosion. 18th IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits (IPFA) 1–6Google Scholar
- 12.Breach CD, Lee TK (2011) Conjecture on the chemical stability and corrosion resistance of Cu–Al and Au–Al intermetallics in ball bonds. IEEE International Conference on Electronics Packaging Technology and High Density Packaging. 275–283Google Scholar
- 13.Gan CL, Ng EK, Chan BL, Kwuanjai T, Jakarin S, Hashim U (2012) Wearout reliability study of Cu and Au wires used in flash memory fine line BGA package. IEEE Proceedings of International Microsystems, Packaging, Assembly and Circuit Technology Conference, Taipei, pp 494–497Google Scholar
- 14.Classe F, Gaddamraja S (2011) Long term isothermal reliability of copper wire bonded to thin 6.5 μm aluminum. IEEE International Reliability Physics Symposium 685–689.Google Scholar
- 17.Cho J, Yoo K, Hong S, et al. (2010) Pd effects on the reliability in the low cost Ag bonding wire. IEEE International Electronic Components and Technology Conference 1541–1546Google Scholar
- 18.Kai LJ, Hung LY, Wu LW, et al. (2012) Silver Alloy Wire Bonding. pp 1163–1168Google Scholar
- 19.Pagba A, Reynoso D, Thomas S, Toc HJ (2010) Cu wire and beyond—Ag wire an alternative to Cu? 2010 12th Electronics Packaging Technology Conference 591–596.Google Scholar
- 20.Tsai HH, Lee JD, Tsai CH, Wang HC, Chang CC, Chuang TH (2012) An innovative annealing twinned Au–Ag–Pd bonding wire for IC and LED packaging, IEEE International Microsystems Packaging Technology Conference 505–508Google Scholar
- 22.Chuang TH, Chang CC, Chuang CH, Lee JD, Tsai HH (2012) Formation and growth of intermetallics in an. IEEE Transactions on Components, Packaging and Manufacturing Technology 1–7Google Scholar
- 24.JEDEC JEP 122 (2010) Failure mechanisms and models for semiconductor devicesGoogle Scholar
- 25.JEDEC JESD94 (2008) Application specific qualification using knowledge based test methodologyGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.