Can Nondestructive Techniques Identify BGA Defects?

Estimated reading time: 9 minutes

By Zhen (Jane) Feng, Ph.D.; Juan Carlos Gonzalez; Sea Tang; and Murad Kurwa,FLEXTRONICS International Inc. and Evstatin Krastev, Ph.D., Dage

The industry needs nondestructive techniques to identify BGA opens and cracks. Currently, X-ray and TDR are most widely used. We compare TDR, AXI, and 2DX to the destructive techniques cross-sectioning/SEM and dye & pry.

In comparing results from three nondestructive test methods with two destructive techniques, we looked for a correlation in finding opens and cracks in BGA joints. Our experiment included testing 30 pins on a BGA from 10 boards using time domain reflectometry (TDR), automatic X-ray inspection (AXI), and transmission X-ray (2DX). Then, researchers examined 8 of those boards with cross-section/scanning electron microscopy (SEM), and the remaining 2 boards with dye & pry. These studies helped us to gain a good understanding of TDR, AXI, 2DX, cross-section/SEM, and dye & pry techniques. TDR identifies BGA opens and larger cracks while 2DX detects opens and smaller-sized BGA cracks easily. Limitations of TDR, AXI, 2DX, cross-section/SEM, and dye & pry techniques also became apparent.

More BGA and area array devices are appearing on PCBAs as product/functional complexity increases. To achieve good signal integrity, more I/Os are packed in smaller areas within the available real estate. Engineers need nondestructive techniques to identify BGA defects when in-circuit test (ICT) and functional test (FT) call for a faulty device.1-2 Identifying BGA cracks is not an easy task with the available tools. AXI systems are used for identifying BGA opens in electronics manufacturing; however, it is challenging for AXI laminography to detect BGA defects smaller than 4 mils. TDR can identify BGA full cracks (opens) using impedance measurement data. The 2DX process has been used to identify BGA defects because of its clear image. AXI has more than 90% coverage for PCBA, and is an effective tool for collecting real-time data for SMT process improvement.3-4 Engineers usually use 2DX to verify critical defective BGA pins found with the AXI. They often use cross-section/SEM or dye & pry to identify defective BGA pins; however, the boards are destroyed during these tests. We studied the nondestructive options to look for a correlation with cross-section/SEM and dye & pry data.

The Experiment

The BGA dual-encoder processor tested has defects based on ICT and FT results. The BGA is on a board that is set-top-box bundled with user-controlled broadcast, frame enhancement, and network access. The fab has six layers and 388 pins with pitch size of 0.75 mm. Solder joint problems have been reported previously for pins A24 and C26 using SEM and dye & pry.

TDR is a measure of the reflections on an applied step pulse from the device under test (DUT), and is a powerful tool for analyzing the change in impedance through a device. TDR measures the reflections that result from a signal traveling through a transmission environment of some kind ? a circuit board trace, cable, connector, etc. The TDR instrument sends a pulse through the medium and compares the reflections from the unknown transmission environment to those produced by standard impedance. This helps examine poor connections, mismatched traces, and other circuit discontinuities in transmission systems. Resolution means the minimum distance allowed between two impedance mismatch points. If the two impedance mismatch points are so close that the distance between them is smaller than the TDR resolution, TDR will take it as one impedance mismatch point. If the trace is broken totally, even if the gap is lower than 1.25 mm, TDR can detect it.

The TDR test procedure was to find a test point connected to the I/O we need to measure; this test point must be easy to contact with the probe and also close to the ground point. We then measure the length of the trace between the test point and the solder ball of the BGA. TDR equipment can then be adjusted, tuned to impedance/time tests, and connected via probes to the test and ground on the golden board. When testing is complete on a golden board, repeat the tests on your variable board. Compare both curves to identify if there is a crack or open on the solder ball. All pins’ graphics of DUT should be recorded and compared to corresponding pins on a golden board.

Ten boards were tested with AXI after the TDR test. The defective pins were recorded. The goal was to find out what size of BGA open can be detected with AXI.

After TDR and AXI testing, 2DX techniques were put to the test. The X-ray absorbency of a particular material depends on its atomic number and density. The transmissive X-ray image generated from a BGA ball and pads shows grey-level variations corresponding with higher X-ray absorption (darker) for thicker materials, or materials that absorb X-rays to a greater extent. For instance, metals absorb X-rays more than organic material of the same thickness. Voids and cracks are lighter greys, as fewer X-rays have been absorbed. We might not be able to detect the crack if the difference in absorption between rays is very small. Different configurations can reveal cracks that were hidden by other, overlapping materials. Looking at the board at 90° is impractical; however, oblique angle viewing, up to 70°, uses open transmissive X-ray tube technology to avoid these obstacles.

The oblique and rotation angles of the X-ray detector are key factors for identifying small cracks.5 The images were collected for all 30 pins of the 10 boards, and measurements were done for some joints with cracks. It is noted that 2DX measurement data for cracks (below 4 mils) is just for reference as the most accurate measurement is accomplished using top view.

After completing nondestructive examinations, eight boards were sent for cross-section/SEM. We chose the best-fitting mounting cup, uniformly applied resin, and ground carefully at the location of interest, which was precisely aligned. No one knows in advance where the “perfect” location is. Therefore, grinding must be done extremely carefully using different grinding grades from 200 to 22 µm.

Two other boards were used for dye & pry testing after nondestructive inspection. The dye sample was inspected using a high-magnification microscope (>25×) to identify dye penetration and failure mode presented.

Results

In total, we acquired 1,200 data points using the different test methods. We summarize that 2DX is the more effective nondestructive technique for identifying BGA joint defects. There is a good correlation between 2DX and SEM. Setting the right testing conditions is the key for optimizing the benefit of the 2DX technology.

Figure 1. TDR shows an open defect.

For example, pin A1 is missing a ball; TDR, AXI, 2DX, and SEM all called it as a defect. Based on the TDR data, it is easy to tell that the pin has an open defect (Figure 1). Pin A26 on the same board (M1 = 41.2Ω, and M2 = 59.2Ω) shows open defect as well. This pin was found defective by 2DX and SEM. Pin C26 on the same board (M1 = 60.7Ω, M2 = 60.6 Ω), is different from A26 by about 0.1Ω; it is not identified as a defective ball by TDR (Figure 2A). In actuality, the BGA ball has 1.5-mil cracks (Figures 2B & C).

Figure 2A. The TDR results for pin C26 do not show defects.

For 300 pins on 10 boards, a total of 18 defects were found based on TDR measurement data. Five defective pins were detected using AXI; only three of them were found with TDR. However, all five defective pins were found with 2DX, SEM, and dye & pry. In another example, only 2DX and SEM found a defect at pin B25, a crack at the FR-4 side of the BGA ball. Of the18 pins TDR identified as defective solder joints, 3 pins have correlation with AXI, 2DX, and SEM; 12 pins show correlation with 2DX and SEM/dye & pry, 2 pins agree only with 2DX. Seventeen of the 18 defective pins from TDR have good correlation with 2DX; only one pin does not show correlation with either 2DX and SEM; 2DX and dye & pry did not find any short issue for the particular BGA ball location.

A total of 70 out of 300 pins of interest were identified as defective by 2DX. Here, 78.6% of the 2DX defective calls show correlation with SEM/dye & pry (41 pins for SEM; 14 pins for dye & pry). This was excellent correlation.

We have 240 cross-section/SEM data points. SEM indicates 72 pins total with defects and 41 pins have correlation with 2DX. In this study, 51.3% data points have defect agreement between SEM and 2DX.

From the two boards used in dye & pry, 60 data points were collected. Twenty pins were identified as defected. At 14 of 20 (70%), dye & pry data shows good correlation with 2DX. The TDR has 15?24% agreement with 2DX, SEM, and dye & pry, taking into account all defective pins found. The 2DX had high correlation of 51?52% with SEM and dye & pry. The correlation percentage of 2DX, SEM, and dye & pry with other testers is 78.6%, 56.9%, and 70.0% respectively.

null

null

null

null

null

null

null

null

null

null

null

null

Figures 2B and C. 2DX and SEM results do identify 1.5-mil cracks at C26.

null

Conclusion

Our overall conclusion is that 2DX is a more effective tool for identifying BGA defects. TDR found fewer defective balls than 2DX and SEM because of its resolution. TDR has the capability to identify large-sized BGA defects like opens (whole crack), or crack sizes above 50 µm. TDR test results also show good correlation with other testers. Current TDR technology has limitation finding small crack defects, which may be obvious on 2DX and SEM.

There is good correlation between 2DX and SEM/dye & pry results: agreement for 55 defective pins is found for 2DX and SEM/dye & pry data. The defect agreement is 51% for 80 pins called defective by 2DX and/or SEM. These results are based on eight SEM boards’ data (including cracks in FR-4 material).

The defect agreement is 52% for 27 pins called defective by 2DX and/or dye & pry based on two boards. 2DX is an effective tool to detect BGA defects including opens and cracks down to 30 µm. Because FR-4 material is very transparent to X-rays, cracks in FR-4 are not found easily by 2DX. For AXI, it is challenging to detect BGA open or crack defects that are smaller than 100 µm in size. SMT

REFERENCES

Contact the authors for a complete list of references. Equipment used for the work were a LeCroy WE100H mainframe with ST-20 TDR module, an Agilent Laminography 5DX, a Dage XD7500 transmissive X-ray, and a JEOL Scanning Electron Microscope. These are TDR, AXI, 2DX and SEM, respectively. The dye used in dye & pry was a Dykem product.

Zhen (Jane) Feng, Ph.D.; Juan Carlos Gonzalez; Sea Tang; and Murad Kurwa, Flextronics International Inc., San Jose, Calif., may be contacted at Jane.Feng@flextronics.com. Evstatin Krastev, Ph.D., applications manager, Dage Precision Industries, a Nordson Company, may be contacted at ekrastev@dage?group.com. Other acknowledgements include Flextronics Technology Laboratory Zhuhai, China and Guadalajara, Mexico; Flextronics engineering team at Zhuhai, China. Dage and LeCroy support teams. The full experiment was presented at SMTAI 2008.