Room Temperature ACF Bonding Process using Ultrasonic Vibration For Chip-On-Board and Flex-On-Board Applications Kiwon Lee, Hyoung-Joon Kim, and Kyung-Wook Paik Nano Packaging and Interconnect Lab. (NPIL) Department of Materials Science and Engineering Korea Advanced Institute of Science and Technology (KAIST) 373-1, Guseong-dong, Yuseong-gu, Daejeon, 305-701, Korea phone: +82-42-869-3386, fax:+82-42-869-3310 e-mail:
[email protected] Abstract In this study, a novel anisotropic conductive film (ACF) bonding process using ultrasonic vibration was investigated in chip-on-board (COB) and flex-on-board (FOB) applications. The ACF temperature increased as the U/S power increased and the bonding pressure decreased. The ACF temperature was successfully controlled by adjusting both U/S power and bonding pressure. The optimized U/S bonding time was 3 sec at room temperature. The significant meaning of this result is that the ACF bonding process can be remarkably improved by U/S bonding compared with conventional 15 sec T/C bonding at 190 oC. Using the optimized U/S bonding parameters, the ACF interconnects showed similar bonding performances as T/C bonding in terms of the daisy-chain resistance and the adhesion strength. The FTIR (Fourier Transformation InfraRed spectroscopy) analysis showed that the cure degree of adhesive resin was achieved 90 % at 3 sec. In the reliability tests, the U/S bonded ACF interconnects showed no significant change in electrical resistances during 85 oC / 85 % RH test and 125 oC high temperature storage test for 1000 hours and -55 oC ~125 oC thermal cycling test for 1000 cycles. 1. Introduction Anisotropic conductive films (ACFs) are well known adhesive interconnect materials which consists of conductive particles and adhesive polymer resins in a film format. And they have been widely used as interconnect materials in flat panel display applications [1] and also in flip chip semiconductor packaging applications [2]. ACF interconnects are simple and lead-free process as well as cost effective packaging method compared with solder interconnects. For ACF interconnection, thermo-compression (T/C) bonding is the most common method, however it is necessary to reduce the bonding temperature, time and pressure, because T/C bonding is often limited by high bonding temperature, slow thermal cure, uneven cure degree of adhesive, large thermal deformation of the assembly. Therefore, there are constant needs of lower bonding temperature and faster cure ACF bonding to replace the conventional T/C bonding. Ultrasonic (U/S) bonding is one of alternative processes for ACF interconnection which have been suggested by Lee et al. [3][4]. In U/S bonding, ACFs can be rapidly heated by certain ultrasonic vibration, and it can be described by materials’ complex young’s modulus which consists of storage modulus and loss modulus under cyclic stress conditions. Storage modulus relates to elastically stored
energy, and loss modulus relates to energy loss which converts to heat. In general, it is well known that visco-elastic materials such as polymers have large loss modulus. Therefore, it is expected that highly visco-elastic B-stage ACFs may generate a large amount of heat by ultrasonic vibration. As a result, the ACF layer can be rapidly heated and cured without additional chip/substrate heating. In this study, a novel ACF bonding process using ultrasonic vibration was investigated and characterized in terms of adhesion strength, electrical continuity in comparison with those of T/C bonding, and its reliability was evaluated. 2. Experiments 2.1 Materials preparation The test specimens were 680 um-thick Si chips and 1 mmthick FR-4 PCB boards for COB bonding, and 25 um-thick polyimide based flexible substrates and 1 mm-thick FR-4 PCB boards for FOB bonding. ACFs were epoxy based adhesive films with 40 ~ 50 um thickness, and they contained 5 um diameter Ni/Au coated polymer balls for COB bonding and 8 um diameter Au coated Ni balls for FOB bonding, respectively. Table 1 summarizes the specifications of test specimens. Table 1. The specifications of test specimens. Chip-on-board bonding Materials Thickness 3 mm X 3 mm Si test chip 680 um Si chip Bare FR-4 1 mm PCB board Epoxy based adhesive 45 um Base film ACF Conductive 5 um NI/Au coated polymer balls diameter particle Flex-on-board bonding Materials Thickness Polyimide film 25 um Flexible substrate FR-4 1 mm PCB board Epoxy based adhesive 40 um Base film ACF Conductive 8 um Au coated Ni particles diameter particle
2.2 Equipments Figure 1 shows the ultrasonic bonder for COB bonding which consists of an ultrasonic transducer, a horn, a hot-plate and a load cell. Longitudinal vibration frequency of ultrasonic bonder was 40 kHz. The output power of the ultrasonic bonder was constant 200 W. The bonding force was also constant 35 N which was the load of the ultrasonic horn itself.
In order to prevent the damage of Si chips, such as cracking and fracture, a teflon cap of 500 um thickness was applied at the end of the ultrasonic horn.
In FOB bonding, the position ACF bonding area was on the edge of the flexible substrate and the PCB board. Therefore U/S vibration could be easily applied on the ACF interconnects while each flexible substrate and PCB board in figure 3 were fixed on the stage as shown in Figure 4.
U/S horn
Flexible substrate ACF PCB board
Figure 4. An ultrasonic bonder set-up showing an U/S horn and a FOB test specimen. Figure 1. An ultrasonic bonder with longitudinal vibration.
Test Si chips were ACF flip chip assembled using a jig to prevent misalignment of chips during U/S bonding as shown in figure 2. Ultrasonic vibration
Si chip ACF
Jig
PCB board
Figure 2. A schematic diagram of the flip chip assembly structure using a jig for U/S bonding.
In COB bonding, a lab scale U/S ACF bonder showed several limitations due to fixed U/S bonding parameters and inaccurate chip alignment using a jig. Therefore, we had setup a new 28 kHz U/S ACF bonder which can adjust its output power from 160 W to 240 W and bonding force from 3.5 kgf to 6.5 kgf for FOB bonding.
2.3 Feasibility test of U/S ACFs bonding The feasibility test of room temperature U/S ACFs bonding was performed using COB bonding assemblies. In order to investigate the ACF heating behavior by ultrasonic vibration, the ACF temperature was monitored during U/S bonding at various PCB board temperatures with fixed U/S bonding parameters. After U/S bonding, the adhesion strength and the degree of ACF cure of U/S bonded ACF interconnects were measured in comparison of those of T/C bonding. And the ACF thickness change during U/S bonding was measured by cross-sectioning and SEM. 2.4 Decomposition temperature of materials During U/S bonding, test specimens or the ACF can be decomposed due to rapid increase of ACF temperature. Therefore, decomposition temperatures of test specimens and the ACF were measured to prevent decomposition of materials during U/S bonding. Decomposition temperatures were measured by TGA (Thermo-Gravimetric Analysis) with 10℃/min heating rate. 2.5 ACF temperature vs. U/S bonding parameters
PCB board
Figure 5. A schematic diagram of in-situ ACF temperature measurement Figure 1. The design of (a) a PCB board and (b) a flexible substrate
In-situ ACF temperature during U/S bonding was measured to investigate the effect of U/S bonding parameters
2.6 Adhesion strength of U/S bonded ACF interconnects To optimize the maximum adhesion strength of U/S bonded ACF interconnects, adhesion strengths of U/S bonded ACF interconnects were measured after U/S bonding using a 90° peel tester as shown in figure 6. Peel test was performed with a peel rate of 10 mm/min, and adhesion strengths of ACF interconnects were monitored during peel test using a load cell.
400 o
ACF temperature ( C)
such as U/S power and bonding force on ACF temperatures. Figure 5 shows the in-situ ACF temperature measurement setup with 40 um-thick k-type thermocouples and a thermometer with 150 ms sampling period.
350 300 250 200 150 o
80 C
100
o
50 C o 20 C
50 0
0
2
4
6
8
10
12
14
Bonding time (sec) Figure 8. In-situ temperatures of ACF layer on various PCB board temperatures during U/S bonding.
ACF PCB board
Figure 6. A schematic diagram of 90° peel test
2.7 Daisy-chain contact resistance of U/S bonded ACF interconnects Daisy-chain contact resistances of U/S bonded ACF interconnects were measured at various U/S bonding times to examine the electrical continuity. Figure 7 shows the daisychain structure of test specimens.
I V
Figure 7. The daisy-chain structure of test specimens showing a flexible substrate bonded on a PCB board using ACFs
2.8 Reliability evaluation of U/S bonded ACF interconnects Reliability tests were performed with the optimized U/S bonding parameters in terms of the adhesion strength and the daisy-chain contact resistance. Reliability requirements were 85 oC / 85 % RH test and 125 oC high temperature storage test for 1000 hours, and -55 oC ~ 125 oC thermal cycling test for 1000 cycles. Using 10 test specimens for each test, total daisy-chain contact resistances were measured for every 200 hours and 200 cycles. 3. Results and discussion 3.1 Feasibility test of U/S ACFs bonding Figure 8 shows the changes of the ACF temperature in COB bonding assemblies at various PCB board temperatures.
During U/S bonding at room temperature, the temperature of the ACF layer reached to 300 oC within 2 seconds with an extremely high initial heating rate of 340 oC/s. This result shows that the ACF layer can be rapidly heated by ultrasonic vibration. In general, the temperatures of the ACF layer reached steady-states above 300 oC within 3 seconds, and were not significantly affected by PCB board temperatures, because ACFs were heated not by heat conduction from PCB boards but by self heat generation by ultrasonic vibration. Figure 9 shows the degrees of cure of isothermally cured ACFs at various temperatures. 1.0
Degree of ACF cure
Flexible substrate
o
300 C o 230 C o 150 C
0.8
0.6
0.4
0.2
0.0
0
5
10
15
20
Heating time (sec) Figure 9. Isothermal cures of ACFs at various curing temperatures.
As the temperature increased, curing time of ACFs significantly decreased. At 300 oC, ACFs were fully cured within 3 seconds. To examine the degree of cure of ACFs by U/S bonding, degree of cure of ACFs cured by U/S bonding for 0, 1, and 3 seconds at room temperature were analyzed as shown in Figure 10.
during U/S bonding, decrease in die shear strength might be due to degradation of both FR-4 and ACFs. To measure the degradation temperatures of FR-4 and ACFs, DSC and TGA analyses were performed.
Heat flow (arb. unit)
3s 1s 0s
80
100
120
140
o
Temperature ( C)
Heat flow (arb. unit)
ACF
Decomposition
Curing
Figure 10. DSC curing peaks of ACF cured by U/S bonding for 0, 1, and 3 seconds.
300
200
250
300
350
400
Temperature ('C) (a) 1.0 FR-4
Relative mass
0.9
0.8
Decomposition 0.7
0.6
0.5 150
2
Die adhesion strength (kgf/cm )
The area of curing peak which represents the non-cured part of the ACF was significantly reduced, as the U/S bonding time increased. And the curing peak was completely disappeared in the ACF cured by U/S bonding for 3 seconds. This result shows that the ACF was fully cured within 3 seconds during U/S bonding at room temperature. In the previous result, the temperature of the ACF layer during U/S bonding at room temperature reached to 300 oC within 2 seconds and showed a stead-state above 300 oC. Therefore, rapid U/S cure of ACFs by U/S bonding can be explained by the temperature effect. Figure 11 shows die adhesion strengths of U/S bonded COB bonding assemblies at room temperature.
150
200
250
300
350
400
o
Temperature ( C)
250
(b)
T/C bonding
200
Figure 12. Decompositions of the FR-4 PCB board (TGA) and the ACF(DSC).
150 100 50 0
0
1
2
3
4
5
6
7
Bonding time (sec) Figure 11. Die adhesion strengths of U/S bonded COB bonding assemblies.
Die adhesion strengths of U/S bonded COB bonding assemblies were about 80 % of T/C bonded specimens. However, die adhesion strengths rapidly increased by curing of ACFs, and reached the maximum values within 3 seconds. After the maximum values, die adhesion strengths slightly decreased despite the ACFs were fully cured. Considering that the temperature of the ACF layer reached above 300 oC
As shown in figure 12, decomposition of the FR-4 PCB board occurred at about 300 oC which was earlier then the ACF, and that of the ACF occurred at above 340 oC. And a few FR-4 PCB boards showed glass fibers exposed on the surface due to decomposition of epoxy resin after U/S bonding for several seconds as shown in figure 13. Therefore, decrease in die adhesion strengths might be due to the decomposition of epoxy resin of the FR-4 PCB board during U/S bonding. In order to solve the decrease in die adhesion strength caused by the decomposition of the PCB board, U/S bonding was performed by a pulse vibration mode. The pulse vibration mode was controlled by bonding times of 1 second and delay times of 0.5 second to lower the PCB board temperature during U/S bonding. For example, 3 second of pulse vibration consisted of 1 second bonding, 0.5 second delay, 1 second bonding, 0.5 second delay, and 1 second bonding to apply
totally 3 second of bonding time. Figure 14 shows the result of pulse vibration.
60
ACF thickness (μm)
ACF thickness 50 40 30 20 10 0
0
1
2
3
4
5
Bonding time (sec) Figure 15. Thickness of the ACF layer during U/S bonding at room temperature.
300 250
T/C bonding
200 150 100
In feasibility test of U/S bonding using COB bonding assemblies, it was experimentally shown that we can do ACF bonding at room temperature rather than elevated temperature higher than typical 190 oC bonding temperature. And at the same time, the bonding times can be also significantly reduced to several seconds by U/S bonding compared with tens of seconds bonding times of T/C bonding. Subsequently, the effects of U/S bonding parameters will be investigated and optimized for FOB bonding. And the reliability of U/S bonded ACF interconnects will be evaluated. 3.2 Decomposition temperature of materials
50 0
0
1
2
3
4
5
6
7
1.0
Bonding time (sec)
Figure 14. Improved die adhesion strength by pulse vibration
The result shows die adhesion strengths increased by pulse vibration. Therefore, it was clearly shown that pulse vibration could solve the degradation of the PCB board caused by the high temperature during U/S bonding, and die adhesion strengths as high as those of T/C bonding were successfully achieved by U/S bonding. Figure 15 shows the thickness changes of the ACF layer during U/S bonding at room temperature. The thickness of the ACF layer rapidly decreased to 8 um within 1 second during U/S bonding. Considering the heights of PCB board pads and chip bumps, flip chip interconnections by U/S bonding using ACFs is possible at room temperature, because the thickness of the ACF layer effectively decreased before the ACF is fully cured.
Relative weight
2
Die adhesion strength (kgf/cm )
Figure 13. Surface decomposition of FR-4 PCB boards after U/S bonding.
0.8
0.6
0.4
Flexible substrate PCB board ACF
0.2 0
100
200
300
400
500
600
o
Temperature ( C) Figure 16. TGA results of the flexible substrates, the FR-4 PCB board, and the ACF Figure 16 shows TGA results of the flexible substrate, the FR-4 PCB board, and the ACF. As shown in the graph, the weight of the FR-4 PCB board rapidly decreased at 300 oC. However, the flexible substrate and the ACF showed negligible amount of decomposition up to 300 oC. Therefore,
200W 180W 160W
o
300
o
4.6 MPa 6.7 MPa 8.6 MPa
300
200
100
0
U/S on
0
2
6
8
10
12
Bonding time (sec)
200
Figure 18. ACF temperatures vs. bonding pressures at 180 W power U/S off
4
6
8
10
12
Bonding time (sec) Figure 17. ACF temperatures vs. U/S powers at 4.6 MPa bonding pressure The increase of ACF temperature can be explained with U/S vibration amplitudes. U/S vibration amplitudes increase with larger U/S powers at constant pressures, because the work by U/S vibration increases as the U/S power increases. According to the well-known equation which explains heat generation under cyclic deformation [2],
dQ =
f ( Δε ) 2 E ' ' 2
heat generation (dQ) is proportional to cyclic strain (Δε). And cyclic strain increases as U/S vibration amplitude increases. Therefore, the increase of U/S power causes the increase of U/S vibration amplitude, and cyclic strain of the ACFs resulting in more heat generation. Figure 18 shows the ACF temperature during U/S bonding at 4.6 MPa, 6.7 MPa and 8.6 MPa bonding pressures at constant 180 W power. As shown in the graph, the ACF temperatures decrease as bonding pressures increase. The decrease of ACF temperatures can be explained with U/S vibration amplitudes. U/S vibration amplitudes decrease as bonding pressures increase at constant U/S powers. Because the work by U/S vibration is constant at a certain U/S power, it means that a smaller vibration amplitude can be obtained at larger pressures. Therefore, the increase of U/S powers causes the decreases of U/S vibration amplitude, and cyclic strain of the ACF resulting in a less amount of heat generation.
As explained above, ACF temperatures were dependent on both U/S powers and bonding pressures. Therefore, in order to maintain the ACF temperature below 300 oC, various U/S powers were selected at 4.6 MPa, 6.7 MPa and 8.6 MPa bonding pressures. Figure 9 shows the ACF temperatures during U/S bonding with various bonding parameters. 400
ACF temperature ( C)
2
4.6MPa 300
200
100
0
0
2
4
6
8
10
12
Bonding time (sec)
(a) 400
ACF temperature ( C)
U/S on
0
o
100
0
U/S off
4
6.7MPa
o
ACF temperature ( C)
400
400
ACF temperature ( C)
ACF heating temperature during U/S bonding was maintained below 300 oC to prevent FR-4 decomposition. 3.3 ACF temperature vs. U/S bonding parameters Figure 17 shows the ACF temperature during U/S bonding with 160 W, 180 W and 200 W powers at constant 4.6 MPa bonding pressure. As shown in the graph, the ACF temperature increased as the U/S power increased.
300
200
100
0
0
2
4
6
8
Bonding time (sec)
(b)
10
12
300
200
100
0
0
2
4
6
8
10
12
Bonding time (sec)
(c) Figure 19. ACF temperatures during U/S bonding with 160W, 180W, and 200W U/S powers (bonding pressure (a) 4.6 MPa, (b) 6.7 MPa, (c) 8.6 MPa) With 4.6 MPa bonding pressure and 180 W U/S power, the ACF temperature increased up to 290 oC. However, test specimens were decomposed with 6.7 MPa and 200 W parameters due to over-heating above 400 oC. With other U/S bonding parameters, ACF temperatures were maintained about 200 oC which was relatively lower than that of 4.6 MPa and 180 W parameters. Therefore, the U/S bonding parameters were optimized with 4.6 MPa and 180W parameters in terms of the ACF temperature.
Absorbance (arb. unit)
8.6MPa
o
ACF temperature ( C)
400
However, with other U/S bonding parameters, adhesion strength showed relatively low 400 gf/cm adhesion strengths. The adhesion strength behavior is well matched with the previous ACF temperature behavior at figure 19. The ACF temperature during U/S bonding with 4.6 MPa and 180 W parameters was about 290 oC, and those with other U/S bonding parameters were about 200 oC. Lower adhesion strengths were mainly due to lower degree of cure of ACF at lower temperatures. As shown in figure 21, U/S bonded ACF interconnects showed 90 % decrease in the peak area of epoxy groups in FTIR analysis which represents the amount of non-cured epoxy monomers. This result indicates that 90 % degree of ACF cure was achieved by 3 sec U/S bonding with 4.6 MPa and 180 W parameters.
Before bonding
3 sec U/S bonding
880
800 3 sec 5 sec 7 sec 9 sec
600
400
200
900
910
920
930
940
950
-1
Wave number (cm ) Figure 21. Epoxy group peaks of the ACF before and after U/S bonding. 3.5 Daisy-chain contact resistances of U/S bonded ACF interconnects
Daisy-chain resistance (Ohm)
Peel adhesion strength (gf/cm)
3.4 Adhesion strengths of U/S bonded ACF interconnects Figure 20 shows adhesion strengths of U/S bonded ACF interconnects with the optimized U/S bonding parameters. As shown in the graph, the maximum adhesion strength was 630 gf/cm at 3 sec bonding time with 4.6 MPa and 180 W parameters. And 630 gf/cm maximum adhesion strength obtained by the optimized U/S bonding was similar to the typical 620 gf/cm of T/C bonding with 15 sec bonding time and 190 oC bonding temperature.
890
2.5
2.0
1.5
1.0
3
4
5
6
7
8
9
Bonding time (sec) 0
0
1
2
3
4
5
6
7
Displacement (mm) Figure 20. Adhesion strengths of U/S bonded ACF interconnects with the optimized U/S bonding parameters.
Figure 22. Total daisy-chain resistance vs. U/S bonding time with 4.6 MPa and 180 W parameters. Daisy-chain contact resistances of ACF interconnects were measured after U/S bonding with the optimized
parameters of 4.6 MPa and 180 W. As shown in figure 22, stable daisy-chain contact resistances were obtained regardless of bonding times after 3 sec. The average daisychain contact resistance was 1.13 Ohm at 3 sec U/S bonding time. And it was similar to 1.08 Ohm which was obtained by the typical 15 sec T/C bonding at 190 oC. These results show that not only similar adhesion strengths but also similar daisychain contact resistances as the typical T/C bonding were obtained using optimized U/S bonding at room temperature and less than 5 seconds bonding time.
Total daisy-chain resistance (Ohm)
3.6 Reliability evaluation of U/S bonded ACF interconnects Figure 23 shows daisy-chain contact resistances of U/S bonded ACF interconnects during reliability tests. U/S bonded ACF interconnects showed no significant change of daisy-chain contact resistance during the 125 oC storage test, 85 oC / 85 % RH test, and -55 oC ~ 125 oC thermal cycling test.
stable electrical resistances during 125 ℃ high temperature storage test, 85 ℃/85 % RH test, and -55 ℃~125 ℃ thermal cycling test. Table 2 summarizes results of the optimized U/S ACF bonding in comparison with those of typical T/C bonding. Table 2. The optimized U/S ACF bonding condition and their results in comparison with those of typical T/C bonding. T/C ACFs bonding
U/S ACFs bonding
at 190℃ for 15 sec
at room temperature for 3 sec
Peel strength (gf/cm)
622.47 (±28.59)
633.41 (±52.14)
Daisy-chain contact resistance (Ohm)
1.08 (±0.02)
1.13 (±0.04)
10 o
These results indicate that the process temperature of ACF bonding can be significantly reduced from a typical 190 oC to room temperature, and bonding time can be also significantly reduced from typical 15 seconds to less than 5 seconds by utilizing ultrasonic vibration.
o
-55 C / 125 C thermal cycling test o 125 C high temperature storage test o 85 C / 85 % RH test
8
6
4
2
0
0
200
400
600
800
1000
Cycles/Hours
Figure 23. Daisy-chain contact resistances during reliability tests As the result, U/S bonded ACF interconnects showed similar adhesion strengths and daisy-chain contact resistances, and showed stable daisy-chain contact resistance during 125 o C high temperature storage test, 85 oC / 85 % RH test and -55 o C ~ 125 oC thermal cycling test compared with the typical T/C bonded ACF interconnects. 4. Conclusions In this study, a novel anisotropic conductive film (ACF) bonding process using ultrasonic vibration was investigated in chip-on-board (COB) and flex-on-board (FOB) applications. In COB bonding, it was experimentally shown that ACF bonding can be performed with bonding time within 5 sec even at room temperature by utilizing ultrasonic vibration. In FOB bonding, the effects of U/S bonding parameters were investigated and optimized. The optimized U/S bonding time was 3 sec at room temperature with 4.6 MPa bonding pressure and 180 W U/S power. Using the optimized U/S bonding parameters, the ACF interconnects showed similar bonding characteristics as T/C bonding in terms of the adhesion strength, the daisy-chain contact resistance, and
References 1. I. Watanabe et al., “Packaging Technologies using Anisotropic Conductive Adhesive Films in FPDs”, Proc. Asia Display/IDW, pp. 553~556, 2001 2. J. Liu et al., “A Reliable and Environmentally Friendly Packaging Technology-Flip Chip Joining Using Anisotropically Conductive Adhesive”, IEEE Trans. Comp. Packag., Manufact. Technol., Vol. 22, No. 2, pp.186~190, 1999 3. Kiwon Lee et al., “Curing and Bonding Behaviors of Anisotropic Conductive Films (ACFs) by Ultrasonic Vibration for Flip Chip Interconnection”, 56th Electronic Components and Technology Conference, San Diego, California, USA, May 30 – June 2, 2006 4. Kiwon Lee et al., “Ultrasonic Anisotropic Conductive Films (ACFs) Bonding of Flexible Substrates on Organic Rigid Boards at Room Temperature”, 57th Electronic Components and Technology Conference, Reno, Nevada, USA, May 29 – June 1, 2007 5. M. N. Tolunay, P. R. Dawson, and K. K. Wang, “Heating and bonding mechanisms in ultrasonic welding of thermoplastics”, Polymer engineering and science, Vol. 23, No. 13, 1983
