Journal of Materials Processing Technology 210 (2010) 1035–1042 Contents lists available at ScienceDirect Journal of Materials Processing Technology journal homepage: www.elsevier.com/locate/jmatprotec Initial bond formation in thermosonic gold ball bonding on aluminium metallization pads Hui Xu a,∗ , Changqing Liu a , Vadim V. Silberschmidt a , Zhong Chen b , Jun Wei c a Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, Epinal Way, Loughborough, Leicestershire LE11 3TU, UK b School of Materials Science and Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore c Singapore Institute of Manufacturing Technology, 71 Nanyang Drive, Singapore 638075, Singapore article info Article history: Received 23 August 2009 Received in revised form 14 February 2010 Accepted 16 February 2010 Keywords: Gold ball bonding Lift-off footprint Ultrasonic power Bonding force Bonding mechanism abstract The morphological features of lift-off footprints on the aluminium metallization pads were investigated to gain an understanding of the effects of bonding parameters on formation of initial bonds during ther- mosonic gold ball bonding. The obtained results showed that metallurgical bonding initiated at the peripheral areas of the contact area situated along the direction of ultrasonic vibration. Those areas extended inwards with an increase in ultrasonic power. Both the external bonded area and central non- bonded area increased with increasing bonding force. Based on the evolution of footprints, the bonding models were proposed, and the effects of the bonding parameters on the formation of initial bonds were discussed. © 2010 Elsevier B.V. All rights reserved. 1. Introduction Wire bonding is the most commonly used technique for micro- electronic interconnects of an integrated circuits to lead frames. Over 90% of the packages of semiconductor, optoelectronic devices and micro-electro-mechanical systems (MEMS) are produced with thermosonic gold wire bonding due to its various benefits such as self-cleaning, a high yield rate, flexibility and reliability (Harman, 1997). Many investigations (Khoury et al., 1990; Tan et al., 2005; Qi et al., 2006; Ji et al., 2006; Shah et al., 2008; Tian et al., 2008) have been carried out to reveal the effects of bonding parameters on bondability. However, optimization of the process parameters still remains a challenging task in thermosonic ball bonding due to the relatively poor understanding of the underlying bonding mech- anism. Such fundamental understanding is becoming even more essential, since the packages increasingly require ultra-fine pitch size bonding. Ultrasound plays a key role in the bonding process. Winchell and Berg (1978) and Xu et al. (2009) highlighted that ultrasonic (US) vibration disrupted the contamination and oxide layer on the sur- face of the ball and the bond pad during wire bonding. In addition, Langenecker (1966) found that ultrasonic energy density required to produce deformation in aluminium was about 10 7 times less ∗ Corresponding author. Tel.: +44 1509 227684; fax: +44 1509 227671. E-mail addresses: H.Xu3@lboro.ac.uk, huixu.hit@gmail.com (H. Xu). than that was required for the same deformation resulted from the thermal energy alone, therefore he pointed out that an ultra- sonic energy could activate dislocations, and then soften and heat materials. Besides, the history of the interfacial temperature dur- ing bonding has been regarded as a good indicator of the bonding process (Panousis et al., 1983). A previous in situ measurement of interfacial temperature using K-type thin film thermocouples (1 m thick, 20 m wide) yielded 320 ◦ C(Ho et al., 2004). Joshi (1971) found that the bonds would be made at liquid N 2 tempera- ture, therefore the bonding is a solid-state process. Many efforts have been made to establish bonding mechanisms to explain the complex thermosonic/ultrasonic bonding process (Hulst, 1978; Hu et al., 2006). A fretting mechanism proposed by Hulst (1978) assumed that interfacial sliding cleaned and heated the contact surface, resulting in robust bonds. But it is not consis- tent with the measurement of the wire movement during bonding using a laser interferometer by Wilson (1972), which indicated that the wire position remained fixed relative to the substrate, notwithstanding the movement of a capillary tool with amplitude of 0.5–2.5 m. Thus, a bond can be successfully achieved without a gross interfacial motion. Therefore, the fretting, friction or sliding model became implausible. In addition, the sliding model predicted preferential bonding at the centre of the interface, while observa- tions for aluminium–aluminium wedge bonds demonstrated that bonding occurred near the interfacial periphery, and there was no bonding near the central region (Harman and Albers, 1977). Chen (1972) used the Mindlin’s microslip theory (Mindlin, 1949) to 0924-0136/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2010.02.012