技術レポートレーザードップラー振動計によるワイヤボンダ超音波振動測定について 2

The vibrator has the characteristic of generating large vibrations at a specific frequency (resonant frequency), but the frequency can change depending on the conditions. Therefore, a feedback circuit is established in the oscillator driving the vibrator to control the frequency. Additionally, vibrations propagated through a metal horn are amplified, resulting in an amplitude of several micrometers at the tool tip under no load. To achieve stable bonding, it is necessary to stabilize the vibration amplitude at the tool tip. As a result, the demand for direct measurement of vibrations at the tool tip has been increasing year by year.

In wire bonding, there is a function to test input drive signals and vibrate the tool. In wire bonders that have been used for a long time, fatigue can accumulate in the vibrator and horn, causing them to start vibrating differently from their initial state. Therefore, when replacing with a new vibrator or horn, recording the vibration values at the horn tip during the initial use allows for quantitative maintenance of the vibrator and horn based on subsequent changes in the horn tip's vibration values. In particular, measuring whether the actual vibration amplitude at the horn tip or tool increases or decreases proportionally to changes in the strength of the drive signal provides useful data for future maintenance. Additionally, when considering switching from original manufacturer parts to specialized manufacturer parts during the introduction or maintenance of the wire bonder, this can serve as a powerful verification method when comparing the performance of each device.

As mentioned earlier, the vibrations of the tool during bonding were traditionally inferred from changes in the resistance (impedance) of ultrasonic transducers or from echo measurement results. However, by using a laser Doppler vibrometer, it becomes possible to verify bondability from the actual vibrations of the tool during bonding. However, it is difficult to direct the laser light to follow the bonding process, so measurements cannot be taken during continuous bonding operations (auto bonding). By using a function that allows for a single, one-time bonding action (manual bonding), it is possible to measure the behavior of the tool at the moment of bonding.

By actually crimping the tool onto the chips or lead frames and measuring the vibrations, amplitude modulation due to differences in contact pressure and US can be observed, as well as frequency modulation (sideband frequencies next to the fundamental frequency) arising from the relationship with the resonance frequency of the lead frame. Furthermore, since the tool is a replaceable part, there are replacement operations, and depending on the installation position of the tool relative to the horn and the tightening torque of the fixing screws, the amplitude of the tool during bonding may change. It has also been found that the vibration amplitude can vary depending on the material and shape of the tool.

In any case, it can be said that obtaining large measurement data on the rate of change is unique to the laser Doppler method. However, the creation of 'bondability evaluation criteria' associated with data acquisition must be individually set for each bonding machine model and location, so it is necessary to regularly collect data and accumulate it during events.

As long as tools are structures, they naturally have vibration modes due to their resonance frequencies. Particularly, when tools vibrate at frequencies of 60 kHz or above, it becomes possible to measure data at multiple points along their length, allowing for the assessment of how subtle differences in the material and shape of the tool affect the vibration modes. It is generally well known that slight changes in the material and shape of the tool can alter bondability, and it has been found that the surface finish of the tool, along with its resonance frequency or vibration mode, influences bondability. Therefore, the Laser Doppler Vibrometer will provide valuable information when developing and verifying tools with new shapes and materials.

With the improvement in integration density per chip, lead frames have also been designed with more pins or finer pitches. During wire bonding, as mentioned earlier, vibrations are induced by ultrasound, and as the pins of the lead frame become narrower, cases have increased where the vibrations of the pins during bonding adversely affect other pins due to resonance phenomena, drawing attention as a cause of defects such as wire disconnection or breakage. Two methods can be considered to suppress vibrations: one is to reduce the amount of vibration from the excitation source or to vibrate at a frequency that is outside the resonance frequency of the lead frame, and the other is to modify the structure of the lead frame or use damping materials to prevent the propagation of vibrations.

However, in the case of wire bonding, vibration during the bonding process is essential, making it practically impossible to reduce the amount of vibration or change its frequency. Therefore, as a method to suppress vibration, it is necessary to consider redesigning the lead frame to minimize the propagation of vibration. When making structural changes or selecting damping materials, it is important to note that the optimal methods may vary depending on the materials and forms of the lead frame, making vibration testing essential during this consideration.

The ball formed at the end of the wire at the tip of the capillary is pressed against the pad on the chip, which serves as the bonding surface, as the capillary descends. Next, the capillary moves over the lead while extending the wire and bonds in a similar manner. After that, the capillary rises, and during this process, the clamp closes, cutting the wire. The cut wire end is melted by high-pressure discharge from a torch, forming a new ball. The bonding to the IC pad at this time is referred to as the first bond, and the bonding to the lead is referred to as the second bond. The bonding parameters during wire bonding include the following, and they can be set individually for the first and second bonds.

Regarding ultrasonic bonding in the wire bonding process, academic analysis and verification have not been extensively conducted until now, except for a few cases. In actual manufacturing settings, the mainstream approach continues to rely on past empirical rules for bonding, and it is not uncommon for operations to proceed based on these rules. However, it is clear that the future of wire bonding will involve increasing uncertainties due to the use of new materials in ICs, multi-pin and narrow pitch designs, and the high-speed and high-frequency capabilities of wire bonders. Therefore, academic analysis and verification concerning ultrasonic bonding are expected to be required not only by wire bonder manufacturers but also by various device manufacturers. As a first step towards the 'intelligentization of the wire bonding process,' the use of a Laser Doppler Vibrometer is recommended.