Fine pitch copper wire bonding in high volume production
Introduction
Wire bonding is still the dominant method of interconnecting die and substrate. Gold (Au) wire bonding technology has been advanced to very fine wire diameters and bond pitches, and thereby, has delayed the wide spread conversion to flip chip bonding. The steady increase of raw materials prices over the past years have driven the electronic packaging cost up to where the cost contribution of the Au wire has become the largest component of packaging materials aside from the substrates. This economic pressure has accelerated the development of finer Au wires, i.e., 0.6 mil Au wire is now available in low volume assembly and 0.5 mil Au wire manufacturing is in development.
Copper (Cu) wire bonding has been practiced for over 20 years using thick wires of about 2 mil diameters for power and automotive applications where superior electrical and thermal conductivity at low cost were required. The challenges of Cu wire bonding: high hardness and propensity for oxidation and corrosion were managed successfully by also developing die design rules for metal stacks, pad thickness and pad hardness. The cost savings potential and manufacturability of fine diameter Cu wire have now paved the way for fine pitch wire bonding to break into the market.
At the same time, new challenges have been introduced even for Au wire bonding by moving die circuitry under the bond pads (CUP) which required new metal stack structures as well as modified bond parameters to avoid cratering. Further, by reducing the dielectric constant of the die dielectrics, the mechanical strength of dice was also weakened which necessitated further stack modifications and more sophisticated controls on the wire bonders. Die design rules are still geared towards fine pitch Au wire bonding at this point in time. Hence, there was and is still a lot of hesitation toward rapid adoption of fine pitch Cu wire bonding despite the economic incentives.
The fundamental issues of Cu wire bonding have been investigated for a number of years [1], [2], [3], [4], [5]. Forming gas (5% H2 and 95% N2) had been proposed as a shielding gas [3] to prevent oxidation of the free air ball (FAB) during electronic flame off (EFO). This approach has been widely adopted and Cu wire bonding kits are offered for all brands of wire bonders to shroud the Cu during FAB formation and to provide the proper controls for longer spark duration and higher current. The hydrogen in the forming gas provides additional thermal conduction during FAB formation [6], [7], [8] and may also reverse some oxidation [9]. Cu is typically harder than Au and wire manufacturers have devoted considerable efforts into providing soft Cu wires by increasing Cu purity (4 N and 5 N) or by adding proprietary dopants [10]. The effective bonding hardness however is different from the wire hardness due to recrystallization as part of the FAB formation [11]. That aside, hardness measurements on ball and wire are rather difficult to perform at best and they are performed at room temperature. The actual bonding takes place at elevated temperatures, i.e., at least at die temperature and probably higher yet, depending on the cooling rate of the FAB. The key factor may really be the work hardening which occurs during the actual bond formation. Initial studies of this phenomenon have been performed already [11] and more will undoubtedly follow.
Cu wire pull strength and ball shear are considerably higher than comparable diameter Au wire values. This is quite remarkable given that intermetallic compound (IMC) coverage is much less than for Au and is difficult to determine at time zero [1], [3], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22]. Typically, coverage is characterized as spotty, limited and thin. Based on the experience with Au, slow IMC growth is considered an advantage in enhancing the bond reliability [1], [3], [9], [10], [12], [16], [21], [23] of Cu. The immediate bond strength has attracted intensive investigations and several bonding mechanisms have been proposed [14], [15], [16], [20], [23]. As usual, wire pull strength and failure mode are used to judge the quality of the bond. Likewise, ball shear and failure mode are used in a similar fashion. However, pull strength and intermetallic coverage carry more weight in the reliability assessment of the Cu wire bond.
The ease of oxidation of Cu remains a concern for the long term durability of Cu wire and bonds under the varying temperature and humidity conditions during the life of the respective components. Further, the ease of corrosion of Cu is another industry concern and has prompted mold compound manufacturers to lower the amount of free ions like chloride and bromide in the compounds. Essentially, the corrosion mechanism is assumed to be similar to that of Au wire [4], [8], [24]. It is assumed that one of the Al–Cu IMCs is attacked leading to the formation of Al2O3 and eventual ball lifting, i.e., an electrical open. The mechanism is not yet entirely proven [1], [15], [25]. The propensity for oxidation and corrosion has fostered the development of alternative wires based on palladium (Pd) coating of Cu wire [8], [26]. Clearly, the floor life of Pd–Cu wire is many times that of Cu wire and a N2 gas shroud is sufficient for FAB formation and bonding. This process and facility simplification may be sufficient in itself to warrant the use of the more expensive Pd–Cu wire even if the enhanced reliability is not fully proven at this point.
Section snippets
Manufacturing process development
All Cu wire bond development work was performed on KnS Maxum Plus and Maxum Ultra wire bonders equipped with Cu wire EFO kits. In the case of Cu wire, a 4 N Cu wire (Maxsoft, Heraeus) was used under forming gas (95/5 N2/H2). Pd–Cu wire (NX1, Nippon Steel) was bonded in a N2 atmosphere. FAB formation was optimized by varying gas flow, spark energy and duration as well as distance of wire to electrode. The optimization focused on obtaining a FAB of reproducible size, spherical shape and the
High volume manufacturing
Given that as an Out-Sourcing Assembly and Test (OSAT) service provider dice from any source must be assembled and that at this stage all dice follow the Au assembly design rules, a basic set of design rules has been distilled out of the many experiences that have been encountered to date (see Table 1).
These rules are the starting point for a rigorous methodology that was implemented for the qualification of each new device (Fig. 11) for all three major package families: QFP, QFN and BGA. This
Conclusion
Cu wire bonding has been successfully implemented in a high volume manufacturing environment. Presently the conversion rate from Au to Cu is 10% and rising rapidly. Extensive process development and characterization has been done to establish a broad data base of experiences which has been the foundation for a rigorous product qualification methodology. Yield and reliability are equivalent to Au wire bonding. Strict clean room and line management as well as meticulous yield improvement are key
Acknowledgements
The authors would like to thank the engineering teams under the leadership of Scott Chen, Louie Huang, Mike Zhao and Sabran Samsuri for sharing their experiences. We also thank Bill Chen and Simon Li for their valuable discussions.
References (28)
- et al.
In situ ultrasonic force signals during low-temperature thermosonic copper wire bonding
Microelectron Eng
(2008) - et al.
Bonding wire characterization using automatic deformability measurement
Microelectron Eng
(2008) - et al.
An analysis of intermetallics formation of gold and copper ball bonding on thermal aging
Mater Res Bull
(2003) - et al.
Effect of wire size on the formation of intermetallics and Kirkendall voids on thermal aging of thermosonic wire bonds
Mater Lett
(2004) - et al.
Mechanical reliability of Au and Cu wire bonds to Al, Ni/Au and Ni/Pd/Au capped Cu bond pads
Microelectron Reliab
(2006) - et al.
Growth behavior of Cu/Al intermetallic compounds and cracks in copper ball bonds during isothermal aging
Microelectron Reliab
(2008) - et al.
A re-examination of the mechanism of thermosonic copper ball bonding on aluminium metallization pads
Scripta Mater
(2009) - et al.
Effect of wire diameter on the thermosonic bond reliability
Microelectron Reliab
(2006) - et al.
Corrosion study at Cu–Al interface in microelectronics packaging
Appl Surf Sci
(2002) - et al.
Grains, deformation substructures, and slip bands observed in thermosonic copper ball bonding
Mater Charact
(2003)
Investigation of the reliability of copper ball bonds to aluminum electrodes
IEEE Trans Compon Hybrids Manuf Technol
Development of copper wire bonding application technology
IEEE Trans Compon Hybrids Manuf Technol
A comparison of copper and gold wire bonding on integrated circuit devices
IEEE Trans Compon Hybrids Manuf Technol
Cited by (51)
Cu-Al interfacial formation and kinetic growth behavior during HTS reliability test
2019, Journal of Materials Processing TechnologyCitation Excerpt :The Cu-Al intermetallic compound is characterized by its hardness that is one or more orders of magnitude higher than the original material of the Au-Al bond. That’s the reason why Appelt et al. (2011) stressed that Cu ball bonding needs more contact power and higher ultrasonic energy that harm the integrity of metal underlayer, initiate silicon fracturing, die cratering, metal splash, and induce bonding pad peeling and cracking. Therefore, this mechanical attribute needs proper parameters of free air ball (FAB) formation under electronic flame off (EFO) process, the reduction of Cu wire stiffness generally could be applied by adopting high purity, thermal annealing, or even microdoped wire.
Deriving lifetime predictions for wire bonds at high temperatures
2018, Microelectronics ReliabilityCitation Excerpt :Copper (Cu) bonding wire has gained widespread acceptance as a replacement for costlier gold wire in microelectronic devices [1–7].
Effects of the wire-bonding technique on the QFN16b's thermal performance. New correlations for the free convective heat transfer coefficient
2015, International Communications in Heat and Mass TransferScanning acoustic GHz-microscopy versus conventional SAM for advanced assessment of ball bond and metal interfaces in microelectronic devices
2015, Microelectronics ReliabilityCitation Excerpt :However, those destructive mechanical test methods only provide information about the required forces and the sites of the interconnect breakdown but do not allow direct access to information about the condition of the bond interfaces themselves. Moreover, for modern (fine pitch) Cu wire bonding [1] and sensitive pad metallizations the pull and shear values strongly depend on the applied decapsulation process which is a prerequisite for providing access to the wire bonds [2,3]. Commonly in failure analysis additional inspection methods like mechanical or focused ion beam (FIB) cross-sectioning combined with subsequent SEM imaging are used.
More uniform Pd distribution in free-air balls of Pd-coated Cu bonding wire using movable flame-off electrode
2015, Microelectronics ReliabilityCitation Excerpt :The demand for more cost effective bonding wire has increased in the microelectronics packaging industry due to the soaring cost of gold [1,2].
Comprehensive transmission electron microscopy study on Cu-Al intermetallic compound formation at wire bond interface
2014, Journal of Materials Research