Wire bonding characteristics of gold conductors for low temperature co-fired ceramic applications

https://doi.org/10.1016/S0026-2714(03)00195-1Get rights and content

Abstract

Low temperature co-fired ceramic (LTCC) with gold conductors has been used for high reliability applications, such as satellite communications. When the gold metallization is co-fired with the LTCC tape, inorganic adhesion additives in the gold conductor interact with the LTCC glass. This interaction is essential to provide adequate adhesion for the gold metallization. However, this interaction can also reduce the softening point of the LTCC glass. The reaction product can migrate to the surface of the gold conductor and affect the wire bondability of the gold conductor.

Effort has been made to develop surface gold conductors with optimized interaction with LTCC. The new gold conductor shows significant reduction of inorganic particles on the surface and improvement of wire bondability. A systematic wire bonding study has been performed on gold conductors under various wire bonding conditions. It is demonstrated that the wire bonding window can be improved significantly by reducing the interaction between LTCC and gold metallization. This paper reports the results of the wire bonding study.

Introduction

Low temperature co-fired ceramic (LTCC) systems are the next generation of packaging technology for wireless applications. Passive components are incorporated into a single, multi-functional high-density circuit.

LTCC is advantageous because of its wide range of materials options and its lower production and component cost. Two types of LTCC systems are commercially available: crystallizing glass systems and ceramic filled glass systems. In the ceramic filled glass system, the ceramic filler, such as alumina, reacts with glass upon firing. The interaction between ceramic filler and glass, controlled by diffusion, may be continuous in the subsequent firing. In the crystallizing glass system, crystalline phase, such as cordierite and wollastonite, crystallizes upon firing and no further reaction occurs. High conductivity metals like silver and gold can be co-fired with the ceramic in an air (oxidizing) atmosphere, creating a package with high packaging density. The capability to fire in air enables the integration of capacitor and resistor components.

The typical LTCC circuit contains three or more different conductors. The internal and via-fill conductor(s) make up the internal components of the circuit. The surface conductor is used for components and for connecting other surface-mounted components.

Gold conductors are commonly used for high reliability applications, such as aerospace and military. There are two options for using gold as a surface conductor with LTCC. The first is an all gold system (surface, internal conductors and via-fill). The recommended internal gold conductor has excellent electrical characteristics and a glass additive for good shrinkage match to the tape. The recommended via-fill conductor fires without defects and shows no surface posting.

The second alternative is to use gold at the surface, silver as the internal conductor, and a transition via-fill to connect the surface and internal layers. This option calls for an extra via-fill conductor, a transition via-fill between gold and silver, and is considerably less expensive than the all gold system. Typically, a 100% silver paste is recommended for the buried conductor layers, and a 100% silver via-fill is recommended layers 2+ (the second layer from the top and/or bottom of the LTCC). For the top or bottom layer, a Ag/Au transition via-fill is required.

Gold surface conductors are ideal for use in LTCC systems because they are highly resistant to corrosion and migration and have good conductivity. They form an excellent interconnect for surface mounting of active and passive components. There are a few options for bonding chip components to the substrate. The most widely used method is wire bonding, which accounted for over 90% of all chip-to-package interconnections in 1999 [1].

There are four current methods for wire bonding: thermocompression (TC), ultrasonic (US), thermosonic (TS), and microgap bonding [2]. This study used the thermosonic bond method, which combines features of both thermocompression and ultrasonic bonding.

Thermosonic bonding creates wire bonds by incorporating heat and pressure, from the thermocompression bonding technique, with ultrasonic bonding’s scrubbing mechanism. The power level of the sonic energy and the tool force are two important adjustable parameters for this mechanism. TS bonding is becoming very widely used because of its automation capability. TS bonding is less vulnerable to surface impurities and employs lower temperatures than TC bonding (≈150 °C instead of 220–350 °C used for TC bonding).

Wire bonding is a very delicate procedure. The wire is usually gold or aluminum. The machine bonds one end of the wire to the metallization using an ultrasonic pulse, creates a loop with the wire, and makes a second bond, cutting the wire at that bond.

Two types of wire bonders can be used: a wedge bonder or a ball bonder. The difference between the two types of wire bonders is the apparatus: the ball bonder’s capillary tube feeds the wire down, while the wedge resembles a sewing needle and is threaded manually. For ultrasonic bonding, the wedge must have teeth in order to grip the wire (see Fig. 1).

Gold ball bonds comprise 95% of all wire bonds formed [1]. The ball bonder’s advantage is the high voltage spark, called the electronic flame off or EFO, that creates a ball at the end of the wire after the second bond is completed, allowing for much better wire bondability at the first bond of the next cycle. The entire ball is bonded to the surface. The capillary tube acts as a wedge at the second bond, bonding the end of the wire and cutting it at the same time.

The second bond can be formed at any point within 360° around the first bond.

The wedge bonder is not as easy to use as the ball bonder. The second bond must be made along the same axis as the first bond in order to keep the wire under the wedge. The advantage of a wedge bonder is that wedge bonds can be made on smaller pads than ball bonds. The wire deformation is usually 25–30% of the wire diameter compared to 60–80% for ball bonds [3].

Once wire bonding is completed, the bonds are pulled and bond strength and failure mode are determined. Possible failure modes are:

  • (0)

    Wire break at the second bond.

  • (1)

    Wire break at the first bond (ball bonder: wire break at the neck of the ball bond).

  • (2)

    Wire break at any other point on the wire.

  • (3)

    Wire/metallization interface failure at the first bond.

  • (4)

    Wire/metallization interface failure at the second bond.

  • (5)

    Film (metallization) lift from the substrate at either bond.

  • (6)

    Substrate break at either bond.

The failure mode may give insight to wire bondability and bond strength. If film lifting occurs, it may indicate inadequate adhesion strength of gold to the ceramic. The failure mode may also be related to the wire bonding conditions. If interface failures occur, it may suggest that either the gold conductor is difficult to bond or the bonder force and/or power are inadequate. Wire breaks are the preferred failure mode in this experiment because a high wire load with a wire break indicates good conductor adhesion to the substrate and bonding between the wire and the gold conductor, while low load failure within the wire indicates that the wire is defective.

In addition to wire bonding properties, the wire bondable gold conductor for LTCC applications should have a good shrinkage match with the ceramic during firing. After firing, it should have good conductivity and the surface should have no defects. Adhesion is a major concern: for proper connection and for high yield at the production stage, the wire must bond well to the metallization, which in turn must adhere to the substrate.

Section snippets

Experimental procedures

All of the gold conductors consist of gold powder mixed with organic vehicles and inorganic additives, three-roll milled. The paste is then forced through a 400-mesh screen to remove any large flakes. The viscous paste formed has between 70% and 85% solids loading and a viscosity between 1000 and 1500 poise at 9.6 s−1 shear rate using a Brookfield CP51 cone/plate viscometer. Four gold conductors were tested. They will be labeled A, B, C, and D. Conductor A was developed as an internal

Ball bonding

Three gold conductors, B, C, and D were tested under two different conditions. Ball bonds can tolerate various conditions due to the larger area of bonding; therefore the ball bond settings remained the same throughout the study. Only the parameters of the second bond (the wedge/stitch bond) were altered (see Table 2).

Fig. 6 illustrates the results. Conductor B had excellent pull strength for both bonding conditions although the percentage of wire breaks dropped significantly at lower bond

Conclusion

This experiment compared four conductors at many wire bonding conditions. Conductor A, as expected, showed very poor results in all trials. Conductor B had higher average pull strength when higher power and force were applied. The large number of glass and oxide particles that migrated to B’s surface during firing required higher force and power settings in order to bond. The surface microstructure also explains Conductor B’s steadily decreasing wire bondability as the force and power settings

Acknowledgements

The authors would like to thank Bob Nicholson, Tung Pham, and Vicente Geverola for their help in collecting data. The author would also like to thank Jerry Fadner for illustrations.

References (5)

  • R.R Tummala

    Fundamentals of microsystems packaging

    (2001)
  • J.J Licari et al.

    Hybrid microcircuit technology handbook

    (1998)
There are more references available in the full text version of this article.

Cited by (19)

  • Deformation characteristics of Au wire bonding

    2007, Journal of Materials Processing Technology
  • Effect of ultrasonic power on wedge bonding strength and interface microstructure

    2007, Transactions of Nonferrous Metals Society of China (English Edition)
  • Deformation analysis of Au wire bonding

    2006, Journal of Materials Processing Technology
  • Effect of tightening torque on the frequency of the sandwich piezoelectric ceramic transducer vibrator

    2010, Proceedings - 2010 11th International Conference on Electronic Packaging Technology and High Density Packaging, ICEPT-HDP 2010
View all citing articles on Scopus
View full text