Tutorial #9
Based on “Gold to Gold Thermosonic Flip-Chip Bonding” by
L. K. Cheah, Y. M. Tan, J. Wei and C. K. Wong
Proceedings HDI 2001, April, 2001, pp.165-175

Introduction

Flip-chip assembly is an attractive solution for high performance and miniaturized microelectronics packaging. A well-established process of flip-chip assembly is based on lead-tin solder, which requires flux to remove oxide during assembly. While it offers high yield and reliable connections, soldering requires complex processes and sometimes involves materials with potential hazard to the environment.

Alternatives to lead-tin soldering, such as fluxless lead-free assembly processes, have been the focus of much research and development in recent years. These fluxless techniques include pressure contacts or thermocompression bonding enhanced by conductive adhesives, thermosonic bonding, and bonding by fusible metals.

Thermosonic flip-chip bonding is an emerging, solderless technology for area-array connections [1,2]. The thermosonic approach is used to join ICs with gold bumps to gold plated pads on substrate. It offers a range of features superior to solder and adhesive counterparts, e.g., simplifies the processing and assembly steps, reduces the levels of assembly temperature, loading pressure and bonding time, and increases current carrying capacity.

Thermosonic bonding was previously applied mainly for wire bonding. This bonding method is a combination of ultrasonic and thermocompression welding that optimizes the best qualities of each for microelectronics usage. Thermocompression welding usually requires interfacial temperature of the order of >300°C. This temperature can damage some die attach plastics, packaging materials, and laminates, as well as some sensitive chips. However, in thermosonic welding, the interface temperature can be much lower, typically between 100 to 150°C, which avoids such problems. The ultrasonic energy helps disperse contaminates during the early part of the bonding cycle and helps complete the weld in combination with the thermal energy.

Advantages

Thermosonic bonding has the following advantages:

(i) metallurgical joining is more reliable than conductive particles and adhesive joining

(ii) process cycle time can be reduced from several minutes to less than 10 seconds

(iii) lower manufacturing cost per unit.

The aim of the project is to establish a thermosonic flip-chip assembly process. Once fully developed, the thermosonic flip chip bonding process can be implemented with conventional equipment and technology such as gold wire bonder, thin or thick film substrate technology, flip-chip bonder and underfill dispenser.

Process Development

Die Design

Thermosonic flip-chip assembly evaluation was conducted using FA10 full array dice. [3] The FA10 is a daisy chain test die for use in evaluating assembly techniques, board continuity, temperature cycle and life test evaluation. The FA10 is a 4.98 mm per side device which contains a full array of 10 mil pitch pads across the entire surface of the die. Contact pads were made of aluminum.

Figure 1 Gold Bump

The gold bumps (Figure 1) were formed using a Panasonic wire-bonding machine with 1 mil (25 m m) gold wire. This resulted in approximately spherical bump of 75 and 50 m m in diameter and height, respectively. This wire bumping method is ideal for getting bumps onto individual chips. Once the wafer is diced it becomes difficult to bump by any other process. Three types of bumping configurations were designed with 32, 48 and 68 I/Os.

Substrate Design

To investigate the thermosonic bonding process, test substrates were fabricated with dimensions of 100 mm x 100 mm and 0.4 mm thick. The design provides three different bonding configurations with eight bonding sites each to create a continuous daisy chain. By measuring the resistance between the two ends of the daisy chain, any missing bond can be detected.

To investigate the effect of different pads finishing, two types of gold surface were prepared:

i) Printed gold paste fired on 96.5% alumina substrate. The thickness of the gold layer is about 10 m m

ii) Sputtered Au / TiW on 99.5% alumina substrate. The thickness of the gold layer is about 1 m m. A multi-target unbalanced magnetron sputtering system was used to produce the Au / TiW layers continuously without breaking the vacuum to ensure good adhesion of the films.

Thermosonic Assembly

In this project, a Panasonic flip-chip bonder with an ultrasonic tool was used to perform the evaluation. Prior to the thermosonic bonding, co-planarity of the die and the substrate must be carefully aligned to achieve good bonding. The bonding procedure begins with the substrate sitting on a heated stage. A vacuum holds the substrate in place. The temperature of the substrate is maintained at 150 °C. The chip is held by the bonding tool with vacuum and is brought into contact with the substrate. After the bonding force has reached a certain level, ultrasonic vibration is applied through the ultrasonic tool for a predetermined length of time to complete the process. Thermosonic bonding with loading pressure from 40 to 100 g/bump bonded on different pad materials were studied.

Process and Reliability Study

After the flip-chip was bonded to the substrate, the assembled parts were subjected to die shear test to obtain qualitative value for bond strength. The die shear test was performed on the parts assembled without underfilling. To assess the reliability of the thermosonic bonding process, the dice and substrates were assembled with selected process parameters. Underfill was dispensed after the bonding cycle and was cured for 2 hours at 150°C. The parts were subjected to cross-section inspection and open/short test to determine the deformation of the gold bump and the connectivity respectively. Electrical resistance was recorded for each part. These parts were subjected to the following tests for reliability assessment:

i) Thermal cycling test was performed using Heraeus HT7012 S2 2 zone air to air thermal shock chamber with temperature setting from -55 to 125 °C

ii) For moisture sensitivity test, Napco (Model 8100-TD Test Chamber) pressure cooker tester (15 psig in 100% RH water) was used.

Results

Figure 2. Shear strength vs bond pressure

Die shear tests were carried out on the flip-chip dice, bonded with different bonding pressures, on pads with printed gold paste fired on 96.5% alumina substrate. The shear force increases from 8.40 to 32.16 g/bump as the bonding pressure increases from 40 to 75 g/bump and decreases to 23.49 g/bump as the bonding pressure increases to 100 g/bump. An optimum bonding pressure of 75 g/bump was observed with highest shear strength. Figure 2 shows the plot of the die shear strength versus bonding pressure.

Fig. 3 Deformation of gold bump

The assembled dice with different loading pressures were subjected to cross-section inspection. Figure 3 shows the deformation of gold bump under 75 grams of bonding pressure. Standoff height is about 25 m m for samples bonded above 50 g. The gold to gold diffusion was observed on the thermosonic bonded interface.

Gold bumped flip-chip is bonded on two different gold surface finishing boards. It is noted that the shear force for flip-chips bonded on the thin film sputtered gold surface is greater than the thick film printed gold surface finishing for bonding pressure of 75 and 100 g/bump as shown in Table 1. The better performance of the thin film sputtered gold finish substrate can be attributed to smoother surface morphology. The uniformity and the thickness of the sputtered thin film are relatively easy to control compared to the printed thick film technology.

Unlike gold wire bonding, co-planarity is important to achieve good gold to gold thermosonic flip-chip bonding. Parallelism and co-planarity adjustment of the ultrasonic tool with respect to the substrate must be achieved to obtain good gold togold diffusion bonding. Figures 4 and 5 show the effect of typical co-planarity issues that affect assembly yield.

Figure 4. Misalignment may cause diffusion bonding on side A but cold joint on side B due to insufficient pressure. Most of the I/Os on side A will pass the test but those on side B will be opened.

Figure 5. Misalignment may cause short circuit on side A and possible diffusion bonding on side B due to overpressure. One or more I/Os on side A will be shorted and most of the I/Os on side B will pass the open/short test. The assembled boards were subjected to open/short test to determine the connectivity of the daisy chain between the flip-chip and the substrate. All the assembled chips passed the test. Electrical resistances measured for each assembled die ranged from 3.5 to 5.5 ohms.

As a first step of investigating reliability, the assembled boards were passed through the reliability tests described. The thermal cycle test was stopped after 500 cycles and the electrical resistance was unchanged after the cycles. This was followed by pressure cooker test for 24 hours. The electrical resistance remained unchanged. The results demonstrated that good gold to gold diffusion bonding has been achieved.

Conclusions

A thermosonic flip-chip bonding process using conventional equipment has been successfully developed:

i) Gold wire bonder to form pull-off bump on silicon die with gold wire bondable aluminum pads.

ii) Thick or thin film ceramic substrate.

iii) Conventional flip-chip bonder with optional ultrasonic tool to provide alignment, heated stage, thermocompression loading, ultrasonic power and controllable duration to perform the flip-chip thermosonic flip-chip assembly.

iv) Conventional dispensing system for underfill dispensing.

The thermosonic flip-chip bonding process is proven to be useful for die with dimension up to 5 x 5 mm and up to 68 I/Os. The following steps have been taken to verify the reliability of the process:

i) Die shear test was performed to meet the criteria of MIL-STD-883 with 5g/bump.

ii) Cross-section of the die to ascertain gold to gold diffusion interface.

iii) Thermal cycle and pressure cooker tests to verify the bonding.

Acknowledgements

The authors gratefully acknowledge the support of Dr. X. Shi and Mr. W. Fan, both from Gintic Institute of Manufacturing Technology for their help in the die shear test and substrate design respectively.

References

[1] S. Y. Kang, P. M. Williams and Y. C. Lee, “Modeling and experiment studies on thermosonic flip chip bonding”, IEEE Trans. On Components, Packaging, and Manufacturing Technology – Part B, 18, No. 4, pp. 728-733, November, 1995

[2] T. S. McLaren, S. Y. Kang, W. Zhang, T. H. Ju and Y. C. Lee, “Thermosonic bonding of an optical transceiver based on 8 x8 cavity surface emitting laser array”, IEEE Trans. On Components, Packaging, and Manufacturing Technology – Part B, 20, No. 2, pp. 152-160, May, 1997

[3] Flipchip Technologies, Inc.

For More Information

Contact:

Dr. L. K. Cheah

Electronics Packaging Group, Gintic Institute of Manufacturing Technology,

Singapore

Tel: 65-7938540 E-mail: lkcheah@gintic.gov.sg