Next-generation laser rework: some applications for high-density and lead-free manufacturing process are explored.
Additionally, the increasing emphasis on tin/silver/copper (SnAgCu, or SAC) lead-free assembly presents new BGA rework challenges. Specifically, BGAs may only be rated to a maximum temperature of 245[degrees]C or 250[degrees]C, whereas the joints must be reflowed to 235[degrees]C to 245[degrees]C. Achieving the desired joint temperatures without exceeding the maximum allowable BGA package temperature is difficult using the current forced convection rework platforms.
On the other hand, a laser rework platform may provide a method of achieving the required joint temperatures without exceeding the maximum allowable package temperature. A laser rework platform also offers the potential of a faster BGA replacement cycle. This paper compares the latest generation laser rework platform with a common forced convection platform.
The Laser Rework Platform
The laser rework platform (Figure 1) used in this study had the following features:
* yttrium aluminum garnet (YAG)-IR 1064 nm radiation
* fixed laser beam diameter (0.118 in.) programmed to cycle on top of the BGA in rectangular patterns
* initial heat transfer into the package that is radiant followed by conduction
* complete board preheat via a forced convection bottom heater.
[FIGURE 1 OMITTED]
The Test Vehicles
Three test vehicles were selected.
Test Vehicle 1: This test vehicle was used to gain experience profiling on the laser platform. It was not profiled on the convection platform. Features included:
* board size: 18 in. x 20 in.
* board thickness: 0.093 in.
* 596 pin BGA package
--glob top BGA (dummy BGA/no die)
--pin count: 596
--size: 45 mm square
--pattern: perimeter + center
--pitch: 0.050 in.
--pad diameter: 0.025 in.
Test Vehicle 2: This test vehicle was used to obtain optimum laser and convection profiles for comparison purposes. Features included:
* board size: 14 in. x 18 in.
* board thickness: 0.098 in.
* 776 pin BGA package
--pin count: 776
--pitch: 0.040 in.
--pad diameter: 0.022 in.
* 1012 Pin BGA package
--pattern: perimeter + island
--pin count: 1012
--pitch: 0.040 in.
--pad diameter: 0.022 in.
Test Vehicle 3: This test vehicle was used for laser profiling and laser BGA replacements. Features included:
* board size: 10 in. x by 16 in.
* board thickness: 0.125 in.
* 1849 pin BGA package
--pin count: 1849
--pitch: 0.040 in.
--pad diameter: 0.022 in.
Profiles were created on the two rework platforms according to these guidelines:
* complete topside board preheat between 100[degrees]C and 125[degrees]C
* ramp rate not to exceed 2[degrees]C/see.
* time above liquidous: 60 to 90 sec.
* maximum die temperature: 225[degrees]C
* peak joint temperature: 205[degrees]C to 220[degrees]C
* maximum joint temperature gradient: 10[degrees]C (5[degrees]C preferred)
* soak between 140[degrees]C and 160[degrees]C: 60 to 120 sec.
* temperature 0.200 in. from BGA: <183[degrees]C.
The results appear in Tables 1-5.
The peak case temperature of the 596 pin BGA (186[degrees]C) was significantly below the peak joint temperatures (207[degrees]C to 214[degrees]C), which would not occur on a forced convection profile. The optimum laser profile required the laser beam to be rotated around the skirt of the BGA without cycling across the glob top. Laser profiles that included cycles across the glob top resulted in excessive case temperatures. The internal construction of the BGA allowed the thermal energy from the perimeter of the package to travel easily into the center spheres, while inhibiting the thermal transfer to the top of the glob top where the thermocouple was.
The laser and convection profiles for the 776 pin BGA as shown in Tables 2 and 3 were similar. The laser profile lowered the die temperature case by 5.5[degrees]C. Although the profile was cooler than a SAC lead-free profile, the reduction in die temperature would have to be greater to have more of an impact on lead-free rework.
The laser and convection profiles for the 1012 pin BGA shown in Tables 4 and 5 were similar. The laser profile lowered the die temperature case by 3[degrees]C. Similar to the 776 pin BGA, the reduction in die temperature would have to be greater to have more of an impact on an SAC lead-free profile.
The 1849 Pin BGA
The laser was only able to create an acceptable profile for the 1849 pin BGA on Test Vehicle 3 by directing the laser beam onto the package and the board due to the construction of the BGA; however, this practice resulted in excessive heating of the board. The heat spreader made intimate contact with the BGA die, but an air gap occurred between the heat spreader and the majority of the remainder of the BGA. Each laser profile resulted in the majority of the heat entering the die and the spheres directly under the die, but insufficient heat was transferred to the outer row of spheres until the laser beam was directed onto the board and the BGA. An 1849 pin BGA was replaced twice.
A number of design improvements have been made to the laser platform since this work was done, and continued evaluation of the platform is recommended. Improvements include an infrared (IR) bottom heater, which includes forced convection cooling and a software update that allows the laser beam to be directed onto the board immediately adjacent to the BGA, without excessive heating of the board. Also, the laser beam can be directed in virtually any pattern versus the previous rectangular pattern only.
Laser rework was not significantly faster than forced convection during the reflow portion of the cycle but did demonstrate an overall reduction in cycle time during the removal process. The improved IR bottom heater is expected to further reduce cycle time. Assessing cycle time reduction during the installation was difficult because a significant portion of the cycle was related to the learning curve associated with the equipment.
Highly accelerated life testing (HALT) was conducted to compare the reliability of the two-time laser-reworked 1849 BGA with the original assembly. The HALT parameters were as follows:
* continual vibration: 4 grams
* thermal ramp: -20[degrees]C to 90[degrees]C
* 10 minute dwells at temperature extremes
* 50 thermal cycles.
The results indicated the failure mode of the two-time laser-reworked BGA was equivalent to the original assembly.
Cross Sectional Analysis
Cross sections of the sphere to PCB pad interface on the two-time laser-reworked 1849 BGA were made. The intermetallic layer appeared normal.
X-Ray Laminography Analysis
X-ray laminography was used to compare the 1849 BGA before and after laser rework; the data appear in Table 6. The diameter of the hall, pad and part were slightly less after laser rework, but the difference was not significant.
The peak temperature 0.250 in. from the skirt of the 1012 pin BGA was measured during two successive runs of the laser profile and appears in Table 7.
Both temperatures were above liquidous but below what would be expected with forced convection rework. The latter temperatures are dependant on the nozzle used, amount of airflow and presence/ absence of shielding. Eliminating the reflow of adjacent components may be required in selected applications. The laser beam was contained to the BGA during this profile, and the heat was traveling to the adjacent components via conductive transfer through the board. How ever, due to the development nature of the work, the beam may have been slightly off the BGA and the above figures should be viewed in this light.
Engineering attention is required to deploy the laser platform in the initial stage. A more extensive reference and profile library incorporated into the machine's software would facilitate the development of thermal profiles.
Profiles for the 596, 776 and 1012 pin count BGAs were acceptable and equivalent to forced convection profiles. The profile for the 1849 pin count BGA needs additional engineering development due to the package construction. This profile indicated internal construction of the BGA is more significant in laser rework as compared with forced convection.
Laser rework reduces the heating of adjacent areas. The average temperature of 188[degrees]C measured 0.250 in. from the skirt of the BGA was below the temperature expected with convection. This feature is important in applications where reflow of adjacent components is not allowed.
Laser rework was not significantly faster during the reflow portion of the profile, but it did demonstrate overall cycle time reduction during the removal process. And, finally, the reduction in die temperatures on the 776 and 1012 pin count BGAs were 5.5[degrees]C and 3.0[degrees]C, respectively. A more significant temperature reduction would be beneficial for lead-free rework.
TABLE 1: Test Vehicle 1; 596 pin BGA package; laser profile. Thermocouple Location Max Rise Time at Peak Ramp Rate 140[degrees]C- Temp ([degrees] 160[degrees]C ([degrees]C) C/sec) (seconds) 5 Case 0.65 65 186 2 Perimeter Joint 0.65 54 208 3 Perimeter Joint 0.65 60 207 4 Perimeter Joint 0.65 58 207 1 Center Joint 0.65 54 214 Thermocouple Location Time Above Max Fall 183[degrees]C Ramp Rate (seconds) ([degrees] C/sec) 5 Case N/A 1.07 2 Perimeter Joint 75 1.07 3 Perimeter Joint 68 1.07 4 Perimeter Joint 73 1.07 1 Center Joint 83 1.07 TABLE 2: Test Vehicle 2; 776 pin BGA; laser profile. Thermocouple Location Max Rise Time at Peak Ramp Rate 140[degrees]C- Temp ([degrees] 160[degrees]C ([degrees]C) C/sec) (seconds) 5 Die 0.38 61 207 1 Perimeter Joint 0.38 69 209 2 Perimeter Joint 0.38 60 206 3 Perimeter Joint 0.38 55 208 4 Perimeter Joint 0.38 61 207 Thermocouple Location Time Above Max Fall 183[degrees]C Ramp Rate (seconds) ([degrees] C/sec) 5 Die 97 0.96 1 Perimeter Joint 102 0.96 2 Perimeter Joint 88 0.96 3 Perimeter Joint 92 0.96 4 Perimeter Joint 96 0.96 TABLE 3: Test Vehicle 2; 776 pin BGA; convection profile. Thermocouple Location Max Rise Time at Peak Ramp Rate 140[degrees]C- Temp ([degrees] 160[degrees]C ([degrees]C) C/sec) (seconds) 1 Die 0.80 34 212.5 2 Perimeter Joint 0.80 46 209.5 3 Perimeter Joint 0.80 46 204 4 Perimeter Joint 0.80 50 205.5 5 Perimeter Joint 0.80 50 204 Thermocouple Location Time Above Max Fall 183[degrees]C Ramp Rate (seconds) ([degrees] C/sec) 1 Die 102 0.92 2 Perimeter Joint 92 0.92 3 Perimeter Joint 87 0.92 4 Perimeter Joint 86 0.92 5 Perimeter Joint 85 0.92 TABLE 4: Test Vehicle 2; 1012 pin BGA; laser profile. Thermocouple Location Max Rise Time at Peak Ramp Rate 140[degrees]C- Temp ([degrees] 160[degrees]C ([degrees]C) C/sec) (seconds) 5 Die 0.5 35 212 2 Perimeter Joint 0.5 35 206 4 Perimeter Joint 0.5 45 206 1 Perimeter Joint 0.5 44 207 3 Center Joint 0.5 38 208 Thermocouple Location Time Above Max Fall 183[degrees]C Ramp Rate (seconds) ([degrees] C/sec) 5 Die 119 0.65 2 Perimeter Joint 96 0.65 4 Perimeter Joint 95 0.65 1 Perimeter Joint 101 0.65 3 Center Joint 102 0.65 TABLE 5: Test Vehicle 2; 1012 pin BGA; convection profile. Thermocouple Location Max Rise Time at Peak Ramp Rate 140[degrees]C- Temp ([degrees] 160[degrees]C ([degrees]C) C/sec) (seconds) 1 Die 0.70 38 215 2 Perimeter Joint 0.70 46 208.5 4 Perimeter Joint 0.70 48 205.5 5 Perimeter Joint 0.70 48 208 Thermocouple Location Time Above Max Fall 183[degrees]C Ramp Rate (seconds) ([degrees] C/sec) 1 Die 110 0.94 2 Perimeter Joint 94 0.94 4 Perimeter Joint 89 0.94 5 Perimeter Joint 82 0.94 TABLE 6: X-ray laminography data before and after laser rework. 1849 Pin BGA Before Laser Rework Slice Estimated Sigma Minimum Maximum Range Diameter (min-max) Sphere 22.00 2.16 15.83 27.67 11.82 Pad 18.76 2.30 12.31 24.63 12.32 Part 17.41 1.77 11.54 22.71 11.17 1849 Pin BGA After Laser Rework Slice Estimated Sigma Minimum Maximum Range Diameter (min-max) Sphere 21.16 1.54 16.44 24.75 8.31 Pad 17.35 1.80 11.12 22.03 10.91 Part 18.04 1.45 13.38 21.90 8.52 TABLE 7: Peak temperature 0.250 in. from the skirt of the 1012 pin BGA; laser profile. Profile Iteration Peak Temp. ([degrees]C) a 184.3 b 192
Acknowledgements: Larry Sirois of VITechnology; Donna Colvard, Richard Garnick, Steve Beck, Joann Newell and Dan Gibbs of Benchmark Electronics Inc.; and Ken Kochi and Margaret Hsu of Sun Microsystems are acknowledged.
(1.) Dr. Paul RE. Wang, Dr. Steven Perng, and Erick Russell, Laser Rework Technology-Energy Source Performance and Process Related CSP Reliability Studies, Proceedings of APEX Electronic Assembly Process Conference, January 2001, pp MP1-2 1-11 and Proceedings of SMTA conference, September 2000, pp. 922-997.
(2.) Erick Russell, "Photonic Soldering for Rework Applications," Proceedings of APEX, Section P-AD/3, pp. 1-5, 2000.
Robert Fatten is a principal engineer with Benchmark Electronics, Hudson, NH; e-mail: Robert. Farrell@bench.com. Dr. Paul P.E. Wang is a sr. process and reliability engineer with Sun Microsystems, Inc., Santa Clara. CA; e-mail: pauchiu.wang @sun.com.
This article was originally presented at SMTA International 2002, Chicago, IL.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Rework & Repair|
|Author:||Wang, Paul P.E.|
|Date:||Aug 1, 2003|
|Previous Article:||Rework with lead-free solders: temperature differences between eutectic and lead-free solders mandate tighter processes, better profiles and precise...|
|Next Article:||20/20 foresight: detecting bridges before they occur.|