TODAY S electronic systems are becoming more complex

Transcription

1 Nondestructive Analysis of Interconnection in Two-Die BGA Using TDR Ming-Kun Chen, Cheng-Chi Tai, Member, IEEE, and Yu-Jung Huang, Senior Member, IEEE Abstract Nondestructive analysis (NDA) is one of the most important tasks that is performed during the industrial characterization of integrated circuits (ICs) because even a tiny defect or failure in the IC packages could be disastrous from the standpoint of quality control. To detect an interconnection failure in IC packages, a time-domain reflectometry (TDR) analysis system was developed. An open-end fixture (OEF) was employed to detect the rapid rise of edge signals from the package and to monitor them under the two parameters of time interval and reflection voltage. We developed a simple and effective electrical NDA system based on the TDR technology that can evaluate the interconnection of ball grid array (BGA) packages. The TDR-measurement results can determine both the failure location and type based on the aforementioned parameters for a two-die BGA package. Index Terms Ball grid array (BGA), failure location, nondestructive analysis (NDA), open defect, open-end fixture (OEF), short defect, time-domain reflectometry (TDR). I. INTRODUCTION TODAY S electronic systems are becoming more complex and compact for computer and telecommunication products. Consequently, the integrated circuit (IC) packages require many input/output pins to be able to adapt to higher signal densities. The ball grid array (BGA) packages have been used for these types of applications due to their improved thermal and electrical performances at relatively low production costs [1]. The structure of a BGA package is complex, especially when it has more than two copper layers in the substrate or when the assembly contains several dies. Since the requirements of product quality and production capability have become more demanding, the failure analysis of BGA packages using the X-ray equipment and software has been rapidly advancing in recent years [2] [5]. In fact, X-ray analysis is the first nondestructive analytical technique of most types of packages. X-ray image reading is a nondestructive graphic process of capturing twoor three-dimensional information about the internal structure of an object [4]. However, the interconnection structure of a BGA package is far more complex, especially when there are more than two copper layers in the substrate and with adhering heat sinks, i.e., the BGA is a high-density assembly. In recent years, the difficulty of using the X-ray method for the nondestructive analysis (NDA) of BGA packages with multilayer substrates Manuscript received August 12, 2004; revised December 22, M.-K. Chen and C.-C. Tai are with the Department of Electrical Engineering, National Cheng Kung University, Tainan 70101, Taiwan, R.O.C. ( n @ccmail.ncku.edu.tw; ctai@mail.ncku.edu.tw). Y.-J. Huang is with the Department of Electronic Engineering, I-Shou University, Ta-Hsu Hsiang 84008, Taiwan, R.O.C. ( yjhuang@isu.edu.tw). has increased tremendously. As a consequence, some of the defective packages could not be detected because of the limitations of X-ray inspection techniques. The time-domain reflectometry (TDR) technique was initially developed for locating faults in long electrical systems such as telephone wires, network lines, and optical transmission paths [6], [7]. Since then, commercial systems have been developed to verify the signal integrity of printed circuit boards, cables, and connectors [8]. In this paper, TDR is applied as a powerful measuring technique for detecting the location of open and shorted signal lines on IC packages. The TDR technique was developed as an attempt to apply NDA for packages and includes the use of the TDR technology and software as presented in [9] [12]. In our previous work [11], the studies were aimed at package failure analysis when the 3M electrically conductive tape attached to the solder balls of the BGA was used. However, they are not accurate enough for detecting faults at very small dimensions, especially for contact interfaces and automatic detections. In contrast to conventional contact methods, an open-end fixture (OEF) was proposed and used for increasing the spatial resolution of the interconnection analysis. The OEF consists of a copper film and a pogo-pin set with a semirigid coaxial probe that connects to a solder ball. This allows the OEF to easily puncture the oxide of the solder ball. Also, the internal mechanical workings have been optimized for highest repeatability. In this paper, a TDR was used to analyze the location of the open or short-circuit defects in three main regions of the BGA package, namely, the substrate, the bond wire, and the solder ball. This paper begins with the presentation of the proposed materials and methodology and is followed by a discussion on failure analysis and verification. The experimental results are analyzed, and the conclusions are presented at the end. II. PROPOSED MATERIALS AND METHODS A. Package and Substrate Design Fig. 1 shows the cross section of a BGA package consisting of a dual die, molding compound, bismaleimide triazine (BT) resin, bond wires, and solder balls. The custom-built 22 mm 14 mm BGA package with a 119-solder-ball count is the failure device of the assembly vehicle for this study. The dimensions of the two dies are 6 5 and mm, with both having a single perimeter row of Al terminal pads. The two-die BGA package has a much higher interconnection density than a single-die BGA package. The reason for this choice is the fact that the package can calibrate a large

2 Fig. 1. Cross-sectional drawing of a two-die BGA package. Fig. 3. Configuration of the OEF. Fig. 2. Schematic drawing of the measurement setup. die-to-package ratio impact during NDA of the interconnection. Thus, the results obtained from a certain die-to-package area ratio can be further extended and serve as a point of reference for the manufacture of the next-generation high-density modules without significantly increasing their dimensions. B. NDA Measuring System The configuration of the TDR NDA system is shown in Fig. 2. The analyzing structure consists of an OEF connected to a Tektronix 11801B digital sampling oscilloscope with SD24 dual-channel TDR/sampling head. The system includes a sampling oscilloscope, a step generator with fast rise time, an OEF, a personal computer, and the Labview software [13]. All components are commercially available except for the OEF. A computer is employed to process the data waveform from the TDR. The sampling head has a rectangular pulse with a fast transition between its baseline (nominally 500 mv) and topline (nominally 0 V). The rise time of 35 ps was propagated through the OEF and into the package interconnection. The NDA system is a digital oscilloscope with a very high sampling rate, which simultaneously records the magnitude and time interval of the reflected voltage. C. Open-End Fixture (OEF) Since the general probe tip cannot be directly connected to the solder ball of the BGA, there was a need make an OEF for probing purposes. The design of the OEF consists of four main parts: 1) open-end semirigid coaxial probe; 2) patterned pogo-pin array; 3) copper film; and 4) pogo-pin house. This OEF combined the pogo-pin array with the copper film, as shown in Fig. 3, wherein the probe has inner and outer radii of Fig. 4. TDR response of a BGA interconnection. a =0.5 and b =1.6 mm, respectively. The coaxial line was filled with a lossless homogeneous dielectric with a relative dielectric constant of ε r =2.1. The other balls were grounded in order to unequivocally isolate the short between two interconnections and enhance the accuracy of failure location in BGA packages. D. System Calibration Multiple reflections can make it difficult to obtain the correct impedances and delays in a voltage curve. The calibration of a TDR NDA system involves the use of a standard impedance and an sub-miniature type-a (SMA)-connector calibration kit with three selected standards: the short, the open, and the reference of 50 Ω. The calibration corrects the error caused by the response of the measurement system. For a 35-ps TDR system of Tektronix 11801B, the bandwidth would be at 10 GHz. During the analysis, the pulse generator launches a square pulse to the OEF with a rising time of less than 35 ps. It is a voltage step pulse with a rise time of approximately 50 ps after being connected to the OEF. However, the reference planes will depend on the SMA connector calibration, and hence, OEF calibration is not required. For the OEF calibration of the TDR

3 Fig. 5. Waveforms for an open-failure curve. (a) Open in the solder ball, (b) failure curve of substrate, and (c) failure curve of Au wire. These plots correspond to TDR results in Fig. 6. NDA, an accurate calibration is required to adjust the delay of the open-standard response. After adjusting the delay time, the system records the reflected waveform and passes it to the computer for further analysis. An averaging function of the TDR system is used to remove the measurement noises as much as possible. III. FAULT ANALYSIS AND VERIFICATION The TDR sends a fast-rising edge of the step voltage down the signal and ground paths and reflects at the extreme impedance discontinuity such as an open or short-circuit defect in the metal wire. The reflected signal is detected at the source end. The magnitude and polarity of the reflected signal specify the type of fault, while the delay interval between the incident and the reflected signals indicate the distance to the fault in the interconnection. The fault location L f from the monitoring point at the scope to the point of fault is given by L f = n i=1 C T p,i 2 ε r,i (1) where n is the section of the package, C is the speed of light ( m/s), T p,i is the time interval from monitoring point to the fault and back again, and ε r,i is the effective dielectric constant of the ith segment. Fig. 4 shows a typical capture of a TDR signal for the BGA package. The time T p,i can be easily determined from the interconnection of the reflected signal recorded by the TDR. A fault occurring in the BGA can be determined from the discontinuity distance and reflected voltage. The time and wave shape will indicate how far into the package is the defect located and whether it is an open or a short-circuit defect. A. Complete Open Defect An interconnection containing a defect that results to a complete break of an interconnection is called a complete open failure. The TDR measurement of a defect-free interconnection and the three signal traces with an open defect are shown as the curves (a) (c) in Fig. 5. The incident voltage is approximately 250 mv in amplitude and the reflected voltage bumps the voltage up to 0 mv at t = 250, 400, and 470 ps, respectively, and agrees with the top-down X-ray inspection micrographs shown in Fig. 6(a) (c). These X-ray micrographs of a failure site are as follows: (a) open in the solder ball; (b) broken copper interconnection traces after the package molding; and (c) the separated Au wire in the substrate after the removal of the solder ball. The arrows and circles in Fig. 6 mark the failure points. B. Short Defect An interconnection containing a defect that results to a connection of two individual interconnections is called a short defect. Fig. 7 shows the top-down X-ray micrographs of a failure site as follows: (a) open in the solder ball; (b) broken copper interconnection traces after the package molding; and (c) the separated Au wire in the substrate after the removal of the solder ball. The arrows and circles in Fig. 7 mark the failure points. Fig. 8 indicates a short fault on the substrate because the TDR waveform has a negative polarity. The decrease in voltage occurred at 260, 390, and 480 ps, which correspond to Fig. 7(a), (b), and (c), respectively. Comparing the curves in Fig. 5 with those of Fig. 8, we note that the short defect shifts the response to the left with respect to the defect-free response and is clearly opposed to the case of the open defect. This agrees with the findings of the developed

4 Fig. 6. Top-down X-ray micrograph of a failing site. (a) Open in the solder ball, (b) broken copper interconnection traces after the package molding, and (c) the separated Au wire in the substrate after the removal of the solder ball. Arrows and circles from the design mark the failure points. X-ray method for separating the open from the short-circuit defects. IV. EXPERIMENTAL RESULTS AND DISCUSSION The X-ray images of the samples, which were obtained from the BGA package production line and contain open and short-circuit failures are shown in Figs. 6 and 7. The experimental results show that the X-ray images have confirmed the analytical findings of the TDR analysis method. The TDR Fig. 7. Top-down X-ray micrograph of a failing site. (a) Bridged wire in the bond-wire region, (b) the shorted wire in the substrate region, and (c) shorts between ball positions show up as elongated blobs. Arrows and circles from the design mark the failure points. curves display the locations of the manufacturing defects of the BGA interconnection and identify them as either open or short failures. In view of the good performance, this simple approach is preferred more in actual practice. Thus, the TDR analysis can be used as a nondestructive method to determine the exact location of failures in BGA assemblies. The most challenging aspect of the comparative TDR analysis is to distinguish between the bonding finger of the substrate or the die pad because the TDR measurements in these two regions have similar waveforms. To facilitate the failure analysis in those areas, we can build up the database of

5 Fig. 8. Waveforms for a short-failure curve. (a) Shorts between two solder balls, (b) short-failure curve of substrate via pad, and (c) short-failure curve of bond wire. These plots correspond to TDR results in Fig. 7. well-known waveforms associated with each failure site for a specific type of interconnection impedance. Fig. 4 shows the final reflection of a good package with a relatively large capacitance to ground at the end of the interconnection for the I/O pad region, thus, resulting in a V-shaped dip trailing at the end of the TDR response. By comparing the waveforms with that of the defect-free curve, the package containing a defect can be easily distinguished. It should be noted that since the TDR is reflected at the end of the open/short failure, it cannot be used to detect a second fault in the same interconnection. The faults located in the same interconnection can only be determined by the first fault when it closes the OEF tip as either an open or a short-circuit defect. In other words, the same trace of the interconnection cannot be used to detect short- and open-circuit faults at the same time. However, it can be used to measure the open or short-circuit fault when there is more than one physical fault. The TDR NDA is unable to distinguish the shape and the size of the defects, but it can detect their locations. From experimentations with the OEF probes for the TDR NDA system, it was found that the cable length greatly affects the response of a given probe as a result of the degraded rise time in the reflected pulse through the long cables. The OEF tip is kept by easily puncturing the oxide on solder balls. Therefore, the internal mechanical operation plays an important role for higher repeatability. A qualitative comparison between TDR and X-ray analyses techniques is provided in Table I. The main advantage of the TDR analysis technique is that the time to detect the failure is faster than using the X-ray analysis technique. In addition, TDR NDA can be set up to automatically analyze the defects. V. C ONCLUSION A novel and practical method for the NDA of BGA packages based on the TDR technology is proposed in this paper. The TABLE I COMPARISON OF TDR AND X-RAY ANALYSES TECHNIQUES potential for improving the resolution of the TDR nondestructive testing using the OEF has been confirmed in the present study. To double check for data consistency, the defects were further inspected by the X-ray method. It is important to obtain a good-quality TDR-measurement waveform using the OEF. The advanced technique provides a new experimental method for the OEF contact system of BGA packages and can be further extended as a point of reference for the manufacture of the next package generation. REFERENCES [1] R. C. Marrs et al., Ball Grid Array Technology. New York: McGraw- Hill, [2] T. D. Moore, D. Vanderstraeten, and P. Forsell, Determination of BGA structural defects and solder joint defects by 3D X-Ray laminography, in Proc. IEEE 8th Int. Symp. Physical and Failure Analysis Integrated Circuits (IPFA), Singapore, 2001, pp [3] M. Ohring, Reliability and Failure of Electronic Material and Device. San Diego, CA: Academic, [4] D. M. Thomas and L. J. John, Failure analysis and stress simulation in small multichip BGAs, IEEE Trans. Adv. Packag., vol. 24, no. 2, pp , May [5] T. D. Moore, D. Vanderstraeten, and P. M. Forssell, Three-dimensional x-ray laminography as a tool for detection and characterization of BGA package defects, IEEE Trans. Compon. Packag. Technol., vol. 25, no. 2, pp , Jun

6 [6] P. I. Somlo and D. L. Hollway, Microwave locating reflectometer, Electron. Lett., vol. 5, no. 20, pp , Oct [7] L. Philen, I. A. White, J. F. Kuhl, and S. C. Mettler, Single-mode fiber OTDR: Experiment and theory, IEEE Trans. Microw. Theory Tech., vol. MTT-30, no. 10, pp , Oct [8] Time Domain Reflectometry Theory. Agilent Application Note , Agilent Technologies. [Online]. Available: [9] Electronic Package Failure Analysis Using TDR. TDR System Application Note. [Online]. Available: [10] O. Charles, and L. Craig. (2002). Reflectometry techniques aid IC failure analysis, Test & Measurement World. [Online]. Available: reed-electronics.com/tmworld [11] M. K. Chen, C. C. Tai, Y. J. Huang, and I. C. Wu, Failure analysis of BGA package by a TDR approach, in Proc. IEEE Int. Symp. Electronic Materials and Packaging, Kaohsiung, Taiwan, R.O.C., 2002, pp [12] C. Lihong, H. B. Chong, J. M. Chin, and R. N. Master, Non-destructive analysis on flip chip package with TDR and SQUID, in Proc. Conf. Electronics Packaging Technologies, Singapore, 2002, pp [13] LabView. [Online]. Available: Ming-Kun Chen was born in Taiwan, R.O.C., in He received the B.S. degree in electronic engineering from National Yunlin University of Science and Technology, Yunlin, Taiwan, and the M.S. degree in electronic engineering from I-Shou University, Kaohsiung, Taiwan, in 1995 and 2002, respectively. He is currently working toward the Ph.D. degree in the nondestructive evaluation (NDE) Laboratory of the Electrical Engineering Department, Cheng Kung University, Tainan, Taiwan. From 1996 to 2002, he worked at Advanced Semiconductor Engineering Test (ASET) Ltd., Kaohsiung, where he performed IC testing interface and tester technology improvement. His research interests are in the areas of high-speed measurement, modeling, simulation methodology of packaging, failure analysis of IC packaging/testing, analysis of signal integrity, and nondestructive testing. Mr. Chen is a Student Member of the Institute of Electrical, Information, and Communication Engineers (IEICE) Society. Yu-Jung Huang (S 86 M 87 SM 02) received the B.S. degree in material science and engineering from National Tsing-Hua University, Taiwan, R.O.C., in 1981 and the M.S. and Ph.D. degrees in electrical engineering from the University of Maryland, College Park, in 1985 and 1988, respectively. From 1988 to 1990, he was with the Brimrose of America, Baltimore, MD, as a Staff Scientist for developing an infrared system. From 1990 to 1992, he was a System Engineer at Integrated Microcomputer System Inc., Dayton, OH, working in the field of system design automation. He joined the Department of Electronic Engineering, I-Shou University, Taiwan, in August 1992, where he is now a Professor and the Director of Library. In addition, he had served as Department Head of the Electronic Engineering Department and Director of Computing Center at the same university. He is also the executive secretary of Surface Mounting Technology Association (SMTA), Taiwan Chapter. He has been involved in organizing many international symposia, conferences, and workshops sponsored by organizations such as SMTA and the IEEE. His present research interests are mainly in the area of system-in-package (SIP)/system-on-chip (SOC) design and very large scale integration computer-aided design (VLSI CAD). Dr. Huang is a member of the SMTA, the Integrated Microelectronics and Packaging Society (IMAPS), the Institute of Electrical, Information, and Communication Engineers (IEICE) Society, the Chinese Institute of Electrical Engineers (CIEE), and the Taiwan Integrated Circuit Design Society (TICD). Cheng-Chi Tai (S 94 M 97) was born in Tainan, Taiwan, R.O.C., on November 10, He received the B.S. degree in electronic engineering from Chung Yuan Christian University, ChungLi, Taiwan, the M.S. degree in electrical engineering from National Cheng Kung University (NCKU), Tainan, and the Ph.D. degree in electrical engineering from Iowa State University, Ames, in 1986, 1988, and 1997, respectively. He is now an Associate Professor with the Department of Electrical Engineering, NCKU. His research interests include bioelectronic instrumentation, medical signal and image processing, and nondestructive evaluation using eddy currents, ultrasonic, and acoustic emission techniques. He also has interest in the application of adaptive filtering technique for active noise control. Dr. Tai has been a member of the American Society for Nondestructive Testing (ASNT) since 1994.