Non-destructive Electrical Test Detection on Copper Wire Micro-crack Weld Defect in Semiconductor Device

Transcription

1 Non-destructive Electrical Test Detection on Copper Wire Micro-crack Weld Defect in Semiconductor Device Robin Ong 1 and K. Y. Cheong 1, Member, IEEE * 1 Electronic Materials Research Group, School of Materials and Mineral Resources Engineering, Engineering Campus, Universiti Sains Malaysia, Nibong Tebal, Pulau Pinang, Malaysia robinong0702@gmail.com, *srcheong@usm.my Abstract Structural integrity of wire bonding interconnection is having a significant impact on the quality of microelectronic devices. Conventional electrical test methodology is unable to detect 10 to 20 µm of cracks that exists in wire bond stitch weld (wedge bond) in semiconductor device. This type of crack is termed as micro-crack and it becomes prominent in Power MOSFET Molded Leadless Package. If the imperfect bonded electronic package does not screen out, it may create a potential interconnection failure during product lifetime. Typical industrial based testing method was unable to identify and isolate the failure packages. Therefore, this was the aim of this research to investigate another methodology [Time Domain Reflectometry (TDR)] for this purpose. In order to complement with the TDR results, other non-destructive [3D X-ray Computed Tomography (CT) inspection] and destructive [Scanning Electron Microscope (SEM)] test techniques were used. Novelty of this work is the nondestructive electrical test methodology that able to detect micro-crack defect at wedge bond in a Power MOSFET gate wire. This test methodology offers the short test time and provides high accuracy and efficiency test result. TDR has overcome the conventional test limitation and achieved a novel approach through the defined detection resolution for micro-crack weld. Index Terms Micro-crack weld, test detection, Time Domain Reflectometry (TDR). 1. INTRODUCTION W edge bond cracking problem has been a very critical issue that affects microelectronic devices reliability. Crack size in between 10 to 20 µm on Copper wire bond material is not screened out by ATE test method. It is the severing of the wire from its wedge bond due to a fracture in the weld area. Wire bond technology is the interconnections that connect between IC and semiconductor die to its packaging during semiconductor assembly process. The wire bond interconnect technology has been the most flexible and common in the semiconductor packaging industry since the last two decade. It usually used the aluminium, copper and gold material as the bond wire with various types of wire diameters and wire length in between connection from IC to the packaging leadframe. Reliability of the products will be much depends on the quality control of the wire bonding process since the increasing demand for the smaller and thinner form factor package. Time domain reflectometery (TDR) test measurement technique is the main non-destructive test methodology using for detection of micro-crack electrically in this paper. TDR is able to detect a micro-crack weld on Copper wire bonded Gate pin in power MOSFET by using an electrical test detection technique with high frequency TDR probe measurement. Two TDR bandwidths at 20 GHz and 50 GHz will be tested on experimental setup. Existing well-known non-destructive [3D X-ray Computed Tomography (CT) inspection] and destructive [Scanning Electron Microscope (SEM)] test techniques are used for benchmarking and comparing TDR measurement test results. 2. EXPERIMENTAL SETUP METHODOLOGY A. Time Domain Reflectometry TDR experiment is set up to perform the measurement on dual N-channel Power MOSFET to detect the micro-crack weld defect on Copper wire bonding on device s gate pin. Two Tektronix s TDR bandwidth is selected to study the micro-crack wire bonding defect detection resolution on wedge bond connected to NiPdAu metallization pad. Two types of Tektronix TDR module bandwidth were selected for crack weld non-destructive test technique methodology: the conventional type with lower bandwidth and the higher end model in the market with higher bandwidth. The higher bandwidth of TDR is giving a faster rising pulse in time domain. The resolution of pulse generated by TDR will greatly affect the detection of crack wire bond size. Tektronix s 20 GHz (80E04) and 50 GHz (80E10B) TDR module is used for the TDR measurement on Power MOSFET Gate pin. The 20 GHz and 50 GHz TDR has approximately 23 psec and 12 psec of incident rise time respectively. TDR module is attached to Tektronix digital sampling oscilloscope (DSA8300) to display the TDR measurement response. The main purpose to attach TDR module to highspeed digital sampling oscilloscope is to capture the voltage signal and stores into the oscilloscope s internal memory. This voltage signal is retrieved and can be displayed on the oscilloscope screen later. Besides, high time resolution waveforms of AC voltage and transient signal in time domain can be analyzed directly from the oscilloscope display screen. The nature of impedance change and fault location in the transmission line due to micro-crack can be analyzed from this TDR waveform [1]. However, only one single-ended channel is required for this TDR test detection experiment because only single DUT pin that is being measured and investigated [2]. The TDR module is connected to Cascade Microtech s high bandwidth prober with frequency support up to 100 GHz for

2 high accuracy, repeatable and precise measurement. In this experiment setup, a flexible on-wafer probe system from Cascade Microtech is connected to TDR test equipment. It provides fully automated probing on the molded package DUT in ring form. Efficiency increased with capability to maintain probe-to-pad accuracy for probing on small pad and fine pitch at down to 40 µm. TDR measurement setup is connected very close to the DUT to avoid measurement signal loss. B. Non-destructive X-ray Inspection X-ray inspection is performed at MIMOS Berhad, National Research and Development Centre, Malaysia. All DUTs are tested with Yxlon Y.Cheetah micro-focus X-ray system to perform the inspection on the crack weld defect. This X-ray equipment has the capability with resolution of micro-focus capability up to 0.5 µm on three-dimensional CT imaging instrument [3]. With the estimated crack weld defect in few micron ranges, CT is the existing well-known technique available in industry to perform inspection on semiconductor package without destructs it. Besides, CT provides a unique and three-dimensional image for inspection items. Virtual cross section can be done by reconstruct the three-dimensional image in X, Y and Z axis. A two-dimensional image to capture the top view of DUT is performed. However, the resolution is lower than three-dimensional X-ray that takes about 60 sec to reconstruct the image. C. Destructive Scanning Electron Microscope (SEM) Another popular imaging test technique, SEM is used to investigate the micro-crack weld defect sample that showing the abnormality from TDR test measurement study. Known good DUT is also inspected on SEM to compare with the micro-crack weld DUT. SEM equipment from Oxford Instrument in failure analysis lab is used mainly for this experiment. Both good and micro-crack weld DUTs that have been performed TDR test measurement and X-ray inspection sent for laser decapsulation process. Beside, this decapsulation technique can cause a crack on wire bond. By using laser decapsulation technique, the wire bonding is remaining on its original position as shown in. Fig. 1. Fundamental DUT characteristics can be determined from the TDR waveform. The integrity in the DUT's conductive path will cause a change in impedance. An amplitude change in the reflected signal propagation will display in a high-speed digital sampling oscilloscope. Waveform of problematic DUT will be compared with a known good DUT to determine the location and distance of the impedance change (Z). Fig. 2. TDR measured voltage response in time domain. Fig. 3. Functional block diagrams of the TDR. A. Equations TDR measurement is quantified by using the reflection coefficient, rho (ρ). It is the parameter that describes amount of an electromagnetic wave is reflected by an impedance discontinuity in the electrical conduction medium [4]. It is the ratio of reflected voltage amplitude, V reflected and the incident voltage amplitude, V incident. The mathematical relationship between reflected voltage amplitude and incident voltage amplitude is shown below [5]. V reflected r i ρ = = (1) V incident Z Z r Z + Z i Where Z r is the reflected impedance and Z i is the load impedance which is typically controlled at 50 Ω. Fig. 1. Laser Decapsulation Technique. 3. TDR TECHNICAL INFORMATION TDR test technique measuring impedance in time domain of the conductive path from the TDR probe to the DUT. c l = (2) 2 f ε r where l is the resolution factor for discontinuity length, f is the TDR system frequency bandwidth, ε is the dielectric constant

3 for material that being tested. With the equation (2), minimum crack side (l) can be determined with the use of TDR with the identified frequency bandwidth (f) and the material for the DUT (ε). Hence, this explains the crack side (l) is inverse proportional to TDR frequency bandwidth (f) [6]. 4. EXPERIMENTAL RESULT Device under test (DUT) in this paper is Power MOSFET gate wire micro-crack at wedge bond. This is the copper gate wire that has 38 µm in diameter and 800 µm long. Firstly, three DUT is tested on both 20 GHz and 50 GHz TDR module before 2D and 3D X-ray inspection. Lastly, all DUT will be inspected by using SEM to confirm the TDR test result. Three types of micro-crack weld samples (10%, 50% and 90% crack) is used for this study. Sample 1 has 50% crack while sample 2 and 3 is categorized as 90% and 10% crack respectively. A. 20 GHz TDR Test Result Firstly, lower bandwidth TDR module is used to perform the test measurement. The overall TDR voltage reflection waveform on the three micro-crack weld samples compared to known good sample is shown in Fig.4. Incident pulse of 250 mv, as shown in the TDR waveform, is detected. Both microcrack weld samples and know good sample has almost identical TDR waveform. Voltage amplitude on Y-axis and time domain on X-axis has same TDR reflection response. Time domain scale shown in Fig 4 is 100 psec/div. Voltage reflection amplitude scale is set at 50 mv/div. The TDR waveform area circle in red from Figure 4 is re-plotting as shown in Fig 5. X-axis scale is increasing from 100 psec/div to 20 psec/div. No change on the scale for Y-axis. B. 50 GHz TDR Test Result Fig. 6. Tektronix 50 GHz TDR voltage full waveform on micro-crack weld samples compared to the good, bare and open sample. Fig. 4. Tektronix 20 GHz TDR voltage full waveform on micro-crack weld samples compared to the good sample. Fig. 5. Zoom in of TDR waveform from 20 GHz measurement. Second measurement on higher bandwidth of TDR module is to study the effect of TDR bandwidth corresponding to the micro-crack weld failure mode. With the similar test measurement setup, samples are preceding to probe on Tektronix s 50GHz TDR module. In comparison to 20 GHz TDR module, 50 GHz bandwidth is 200% faster with faster and shorter rise time. It has approximately 12 psec of system rise time. Incident voltage step pulse at 250 mv that introduced from 50 GHz TDR Module is exactly same with lower bandwidth 20 GHz TDR module. Fig. 6 shows the overall TDR voltage waveform for tall the three micro-crack weld samples compared to the good sample and bare sample. Open probe without any DUT on the probe is also taken as the reference for the comparative measurement purposes. The open probe is measuring the lifted probe without DUT. Incident voltage pulse from TDR signal source is consistently generating at 250 mv. It is displayed on Y-axis with scale of 50 mv/div. X-axis has the scale of 20 psec/div in time dombain. Both X-axis and Y-axis setting is exactly similar to 20 GHz TDR probe setup.

4 C. X-ray Inspection Result Fig. 10. Three-dimensional X-ray inspection image on 10% crack sample Fig. 7. Two dimensional X-ray inspection image on micro-crack weld sample. 2D X-ray is usually the fundamental test technique before proceed for more detail 3D X-ray scanning and inspection process. This industry s well-known X-ray inspection test methodology is identified as non-destructive technique to confirm the micro-crack weld sample that has been performed TDR measurement. Very low image resolution appears in 2- dimensional X-ray, as shown in Fig. 7. Micro-crack weld zone is circled in white color. The micro-crack weld area is not visible and not clear to be displayed on 2D X-ray image. Fig. 8. Three dimensional X-ray inspection image on 90% crack sample Micro-crack weld sample is scanning by 2D X-ray in 360 degree and constructed into the 3-dimensional computed tomography image. Similar to 2D X-ray image, it only captured leadframe and wire bond inside semiconductor device. However, the captured image is in 3-dimension where image can be rotated and inspected in 360 degree. Full 3D X- ray CT images on samples with micro-crack weld are illustrated in Figs. 8 to 10. Micro-crack weld zone is circled in white. In this magnification, 3D X-ray image on suspected micro-crack weld area is clearly inspected. Higher resolution image at scale of 0.07 mm and 0.08 mm is reconstructed by zoom in the 3D CT scanned image in X-ray software that available in the X-ray equipment system. D. Scanning Electron Microscope (SEM) Inspection Result Last and final investigation on micro-crack weld defect analysis is using the destructive test technique. Industry wellknown inspection test technique by using SEM destructive imaging technique is carrying out on micro-crack weld samples to verify and confirm the non-destructive test methodology on 50 GHz TDR measurement result or both 2D and 3D X-ray imaging analysis result. SEM inspections technique is destructive and micro-crack weld samples will be decapsulated by using the laser technique to open a window on the molding compound area. This window allows the access to the wedge bond area where micro-crack weld defect occurred. Fig. 11 to Fig. 14 is showing SEM image of 90%, 10%, 50% crack samples and good wedge bonding sample. Fig. 9. Three dimensional X-ray inspection image on 50% crack sample Fig. 11. SEM inspection image on 90% crack sample

5 Fig. 12. SEM inspection image on 50% crack sample Y-axis is significant on three defective samples compared to good, open and bare samples. All three micro-crack weld samples have the reflected voltage amplitude in between bare DUT and good DUT. This has proven that micro-crack weld DUTs has reflection signal occurs somewhere that makes the reflected voltage amplitude higher than known good DUT. However, it still connected electrically if compared to bare DUT that has no wire bonding between leadframe and silicon die. Higher voltage level on sample 1, sample 2 and sample 3 compared to the good device is suggesting failure in the package that causing a reflection. From the reflected voltage amplitude, it is showing that these three samples have different level of micro-crack defect. 2D x-ray method is not effective to detect the micro-crack weld defect on semiconductor device. All three types of crack size that is 90%, 50% and 10% crack is not visible on 2D X- ray. However, 3D x-ray has clearly displaying the wedge bonding condition on all samples. Fig. 13. SEM inspection image on 10% crack sample 6. CONCLUSION From this study, we successfully detected the micro-crack weld defect at Copper wire bonding that connected between Gate pin and leadframe inside dual die Power MOSFET that integrated into the 2x2 mm semiconductor package. As a conclusion, the study presented suggests that TDR measurement is an effective non-destructive electrical test technique to detect the micro-crack weld problem that is facing by most of the semiconductor manufacturing facilities. It is an innovative electrical test methodology with fine resolution that able to detect crack defect in wire bond. It also achieved test time reduction to less than one minute per DUT. This is considered very short test time if compared to other non-destructive wire bond package level test technique that available in the industry today. Test equipment setup is simple and easy as TDR is the conventional instrument that used in industry for other testing methodology purposes. Fig. 14. SEM inspection image on good wedge bonding sample 5. DISCUSSION In 20 GHz TDR measurement, signal is travelling equally fast on both known good and defective sample. Reflection voltage amplitude due to impedance mismatch along the electrical conduction path on all samples is identical. Total time taken to the crack line until reflected voltage occurred is interrelated to the rise time of the TDR module. This happens especially on the micro-crack weld defect at wedge bond that is nearer to the TDR probe contact. If the TDR rise time does not have sufficient time to settle down before TDR signal travel to the crack defect location, reflection voltage that causing by crack length is not visible on TDR waveform measurement. In 50 GHz TDR measurement, voltage amplitude change in REFERENCES [1] Time Domain Reflectometry Theory, Agilent Application Note , Agilent Technologies, May, [2] TDR Primer, Application Note TDRP-0402, Time Domain Analysis (TDA) Systems, [3] Yxlon (2015) Y.Cheetah, Microfocus and nanofocus X-ray inspection systems for PCBS and semiconductor industries [Online], Available from World Wide Web: [Accessed 29 June 2015] [4] James A. Strickland, (1970) Time-domain Reflectometry Measurements. US: Tektronix, Inc. P. 23. [5] TDR Impedance Measurements: A Foundation for Signal Integrity Tektronix, [6] Kipp Schoen, Picosecond Pulse Labs (PSPL), Boulder, High Resolution TDR Measurements Using the PSPL Model 4020 and Model ps TDR Source Enhancement Modules, Colorado, USA, May 2004.

6 Robin Ong was born in Sibu Sarawak, Malaysia, in He received the B.Sc. degree in applied physics and M.Sc degree in solid state physics from Universiti Sains Malaysia (USM), Main Campus, Penang, Malaysia, in 2004 and 2006 respectively. He is currently under taking Ph.D. degree from the materials engineering school, Universiti Sains Malaysia, Nibong Tebal, Penang, Malaysia. From 2005 to 2013, he worked as a Test and Product Engineer in semiconductor manufacturing operation in Penang, Malaysia. Then, from 2014 to 2015, he joined the microphone manufacturing company as a Test and Yield Analysis Engineer. Now, he worked in semiconductorfabless company as a Test Engineer, resident in contract manufacturing operation site in Penang, Malaysia. Kuan Yew Cheong (S 01, M 05, SM 16) was born in Ipoh, Malaysia, in He received the B.Eng. (1st Hons.) and M.S. degrees in materials engineering from Universiti Sains Malaysia (USM), Nibong Tebal, Penang, Malaysia, in 1997 and 2001, respectively, and the Ph.D. degree from the School of Microelectronic Engineering, Griffith University, Brisbane, Australia, in 2004, under the supervision of Prof. S. Dimitrijev. He also received the Certificate of Teaching and Learning from USM in From 1997 to 1999, he worked as a Project Engineer and a Quality Assurance Engineer in a project management company and a semiconductor-device manufacturing factory in Malaysia. Then, he joined the School of Materials and Mineral Resources Engineering, USM, as a Fellow under the academic staff training scheme. Currently, he is a full Professor with USM with main research interests include semiconductor device fabrication technology and electronic packaging. The outcomes of his research have been published in more than 170 journals, 5 book chapters, and 1 Malaysian Patent. Due to his expertise in wide-bandgap semiconductor-based device fabrication technology, he has been invited three times as Visiting Scientist to Korea Electrotechnology Research Institute (KERI), Korea. He has also being constantly invited as a Keynote and Invited Speaker at international conferences. Prof. Cheong is serving as a reviewer for numerous referred journals and as theses examiner both local and abroad. He reviewed more than 600 manuscripts and now serving as an Editor (March 2015 Feb 2017) of Materials Science in Semiconductor Processing (Elsevier Science B.V.) and Associate Editor of Physics Express. Now, he is also editing a book entitled Twodimensional Nanostructures for Energy Related Applications that will be published on mid of 2016 by Science Publisher (an imprint of CRC Press/Taylor and Francis Group). He is a Professional Engineer in Material Engineering discipline registered with Board of Engineers, Malaysia. Honouring his excellent contribution in research, Academy of Sciences Malaysia (ASM) has accolade Prof. Cheong with Top Research Scientists Malaysia (TRSM) Outside the university, he is very active in professional and community activities. Prof. Cheong is a member of Materials Research Society (USA). He was the honorary treasurer for Electron Microscopy Society, Malaysia (2005/2006). Since 2005, he has been elected as one of the Ex-comms of the Institute of Electrical, Electronic Engineers (IEEE), Component, Package, and Manufacturing Technology (CPMT) Society and Electron Device Society (EDS), Malaysia Chapter. In addition, he served as the Chairman ( ) and Advisor ( ) of Electronic Engineering Technical Division under The Institution of Engineers Malaysia (IEM) and a Council Member of IEM ( ). For the past years, he organized a number of local and international conferences. Prof. Cheong, a Certified PSMB Trainer registered with Ministry of Human Resource, Malaysia and a Certified Member of Steinbeis Transfer Centre Network (Steinbeis Malaysia Foundation), delivers technical training courses for local and multinational industries and more than 3,500 engineers and researchers have benefited from those courses.