Abstract Preview

Here is the abstract you requested from the imaps_2018 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.

Intelligent Production of Wire Bonds using Multi-Objective Optimization – Insights, Opportunities and Challenges
Keywords: wire bonding, physical modeling, multi-objective optimization
Ultrasonic wire bonding is an indispensable process in the manufacturing of semiconductor devices. It is used to connect the silicon die to e.g. connectors in the electronic module or to other semiconductors in complex components. Wire bonds in high power applications, such as wind turbines, locomotives or electric vehicles, operate close to the thermal and mechanical limits of aluminum wire. Copper wire is currently replacing the well-established aluminum wire because of its superior electrical, thermal and mechanical properties: Conductivity, maximum operation temperature, and mechanical strength are higher, and the thermo-mechanical mismatch in connection with silicon semiconductors is less pronounced. For the fast-growing new generation of wide bandgap semiconductors based on Silicon Carbide SiC and Gallium Nitride GaN, operating at junction temperatures of 175°C and beyond, copper wires as a top side connection are mandatory. On the downside, using copper increases process forces and tool wear. Copper wire processes also are more sensitive to external disturbances, which reduces the range of parameter values for a stable process, especially on the die. This makes the process design more challenging. Currently, copper wire is not yet used in large volume despite its superior physical properties. In order to produce reliable wire bonds and in particular copper bonds, external disturbances can be compensated by an adaption of process parameters during the process. Finding suitable parameters manually is difficult and time-consuming because of the high number of possible parameter combinations and the complex interaction between the individual parameters. Therefore, a multi-objective optimizing bonding machine (MOBM) was to be developed, which determines pareto-optimal operating points and the associated process parameters and implements these parameters online. (Meyer et al. 2016) For the prototype of such a multi-objective optimizing bonding machine, information processing was structured according to the concept of an operator controller module (OCM, Hestermeyer 2004). The existing machine control is used as a foundation, providing low-level real-time control. The bonding process is influenced by process parameters which are normally manually defined by the operator and typically determined using design of experiments (DoE) methods. In the MOBM, an interface to a manufacturing execution system is implemented instead, which allows external setting of parameters during operation and transmitting of measurement data. Multi-objective optimization is implemented on an external computer system including a suitable process model which maps settable parameters to performance objectives. The four objectives investigated in this prototype are process duration, tool wear, bond strength and the occurrence of tool-substrate contacts. Objectives need not only be modeled, but their current fulfillment must also be obtained online from process data. During setup of the MOBM, process modeling turned out to be the most difficult task, even compared to process parameter adaptation. However, the presented model is able to map all necessary effects and the influences of essential process parameters in detail, enabling a subsequent model-based multi-objective optimization. Modeling of the detailed bonding process including a two-dimensional friction based adhesion model delivered deep insights into the process. Effects of changed process parameters or external perturbations can be simulated. To facilitate modeling, the process was modularized into several individual phenomena, which are the bond formation including touchdown, ultrasonic softening, welding of wire and substrate, and slip between the bonding tool and wire. (Meyer et al. 2015, Brökelmann et al. 2016) In previous publications, the contact between tool and wire has been neglected although this contact limits the transmission of ultrasonic energy into the bonding interface. The presented modelling approach includes this contact and thus makes it possible to predict energy dissipation in this contact and tool lifetime. During development, the isolated effects were modeled in detail and validated individually with process measurements, some of them requiring complex experimental setups. Model validity is crucial for optimization result validity, which in turn is required for parameter selection based on optimization results. Multi-objective optimization techniques, which form the basis of the implemented online parameter selection, require a high number of objective function evaluations and thus process simulations. Model complexity is high due to the required precision. Model reduction is used to balance good model validity and low computing time. The presented model can be used to select the currently best operating point of the bonding machine as a compromise of concurrent objectives, which in turn determine process parameters. The adaptation then ensures that objective values are reached despite process perturbations. One major goal of developing a MOBM was a compensation of perturbations without expensive testing and experience. This way, the complexity of machine operation is reduced. Evaluation of the finished prototype shows that this goal was reached.
Dr. Andreas Unger, Senior Research Engineer
Hesse GmbH
Paderborn, NRW

  • Amkor
  • ASE
  • Canon
  • Corning
  • EMD Performance Materials
  • Honeywell
  • Indium
  • Kester
  • Kyocera America
  • Master Bond
  • Micro Systems Technologies
  • MRSI
  • Palomar
  • Promex
  • Qualcomm
  • Quik-Pak
  • Raytheon
  • Rochester Electronics
  • Specialty Coating Systems
  • Spectrum Semiconductor Materials
  • Technic