Cu-Al intermetallic growth behaviour study under high temperature thermal aging

Transcription

1 Cu-Al intermetallic growth behaviour study under high temperature thermal aging C.L Cha, H.J Chong, Yaw HG, Chong MY, Teo CH Infineon Technologies, Melaka, Malaysia Abstract Copper (Cu) wire always gains the population in semiconductor industry for its superior thermal and electrical performance with increase of gold wire price in the market. The expanding use of electronics components in automotive electronics and rising in high reliability requirement are stimulating vigorous research and development on intermetallic (IMC) growth at the interface between Cu wire bonded ball and Aluminum (Al) bond pad metallization. In microelectronics packaging, IMC is an essential for interconnect formation between bonding wire and bond pad metallization. It grows during other assembly processes. Cu-Al intermetallic grows slowly and perceived higher reliability performance compare to Au-Al system especially during high temperature storage stress test. However, Cu wire with Al pad metallization formed type of IMC with no or very narrow range of solubility and this type of compound is very strong but very brittle. This study evaluates bare Cu wire bonding on Al pad metallization. As Cu wire IMC is known hardly observed at T0 after wire bonded, evaluation samples were subject to thermal aging to promote the IMC growth in un-molded strip form. Preliminary responses e.g wire pull, ball shear strength were collected. IMC coverage were analyzed and its growth thickness were examined from cross section sample by scanning electron microscopy (SEM) method. The stability of Cu-Al IMC in molded package form was studied under different high temperature storage stress test condition include both 150 and 175 SEM based analysis technique - energy dispersive x-ray spectroscopy (EDX) is used to identify the IMC phases. Aside, transmission electron microscopy (TEM) analysis is adopted to understand IMC growth behavior and phase in depth. IMC thermal driven degradation mechanism was analyzed and discussed. (literature study, bench mark other study) Keywords Copper wire, Aluminum pad, Intermetallic I. INTRODUCTION Copper wire had been introduce in semiconductor industry more than 10 years to replace gold wire mainly due to cost pressure, many research has shown beside cost advantages, the copper wire also has many advantages over gold wire especially in material properties and also reliability performance. Gold wire on Aluminum pad reveal intrinsic weakness on the fast IMC growth lead to Kirkendall void, Cu wire had proven to be the solution as CuAl IMC growth is much more slower if compare to AuAl IMC growth.[1,2,3] Copper wire device application started mainly on industry product where the quality requirement is according to JEDEC standard, the acceptance of Cu wire to replace the Au wire are also well receive by the market. However, in recent years, Electronics system has transform the automotive industry trends, the advancements of semiconductors technology has contributes more than 75% of the innovation made in the automotive industry. The application of Cu wires in automotive electronic system start to get attention from automotive customer. According to a recent market research publish by IC Insights, the automotive semiconductor market demand is increasing rapidly and will continue to growth in recent years, it is forecasts that the automotive semiconductor market growth rate will be the strongest from 2016 to See Figure 1 [4] Figure 1: Worldwide Electronics System forecast compound annual growth rate (CAGR) from 2016 through 2021 In June 2015, the Automotive Electronics Council (AEC) had released the first revision of AEC-Q006 (Qualification Requirements For Components Using Copper (Cu) Wire Interconnect) which specifies the requirements for qualification of Cu wire for automotive device. The stringent requirement on HTS stress condition (2x of AECQ100) has impose a lot of challenge to the CuAl IMC stability. Accordingly many study on IMC reliability risk [3,4,5], one of the critical influencing factor to ensure CuAl interface robustness is the IMC coverage at time zero before stress. However, CuAl IMC growth is slow and it is hard to detect after wire bonding, this study aim to evaluate the IMC quality using different thermal aging condition, correlate the IMC coverage using SEM method vs EDX analysis. The CuAl IMC growth behavior is also study under different thermal aging condition by cross section view.

2 II. OBJECTIVE Intermetallic (IMC) coverage is one of the key wire bond response measurement during bonding process characterization. It is essential in such a way that we are having sufficient coverage as a safety factor due to bonding degradation during HAST and HTSL stress environment. Having a large intermetallic coverage at T0 minimize the risk of having lifted ball during ball pull test as the ball to pad strength is greater than the wire breaking load. In additional to it, the ball will take longer time to lift compare to the bonded ball with low intermetallic coverage in the case of corrosion occurred. SEM and Optical method are two common way used for intermetallic coverage measurement. See Figure 2 4. Repeat step 1 to 3 until the Copper bond are completely etched After Copper ball etching, the image of bonding area were captured using both SEM and Optical method. The contrast between Al and IMC was optimized for SEM imaging. The intermetallic coverage were then calculated with selected software tool. See Figure 3 Figure 2: SEM and Optical IMC imaging Each of this methodology has its pro and con tabulated in Table1. SEM 1. Sample Slower preparation 2. Ease of use Technical training needed 3. Resolution High Low 4. Identification of true coverage Easy to distinguish Optical Faster Quick briefing with simple operation mode Difficult to differentiate and misinterpret with presence of contamination, reflection due to rough surface and poor alignment of sample. Table 1: SEM and Optical methodology s pro and con Identify and analyse for Cu-Al IMC is always be a challenge in the semiconductor industry. Copper bonding is known with its slow IMC growth at T0 and the IMC phases are not visible after wire bonding. The wire bonded samples are required to bake under certain thermal aging condition to accelerate the growth of intermetallic. With such, we are able to measure and analyse the IMC coverage. In this study, wire bonded samples in un-molded strip form were subject to two different thermal aging condition in N2 purged oven. 200C@5hrs and 250@8hrs. After baking, the Copper bond were etching using nitric acid: 1. Drop nitric acid on the die surface 2. Rinse the sample with DI water 3. Inspect under microscope Figure 3: IMC measurement methodology using SEM imaging III. RESULT & DISCUSSION 3.1 Identify the optimum IMC measurement methodology (optical vs SEM) Bonding area after chemical etching were subject for intermetallic coverage calculation using selected software tool. Optical imaging result revealed larger IMC coverage 87.78% compare to SEM imaging 70.31%. See Figure 4 Figure 4: IMC coverage for SEM and Optical imaging

3 To identify the validity of IMC coverage area, the same bonding area sample was sent for EDX analysis. Cu element from EDX mapping as part of Cu-Al IMC phases was captured. The Cu element mapping was analysed and show its coverage of 71.58%. See Figure 5 Baking condition: 200C@5hrs Figure 5: IMC analysis on EDX mapping SEM measurement shows comparable intermetallic coverage percentage to EDX element mapping. This analysis result concluded that SEM methodology has better accuracy for IMC coverage measurement and least influenced by the other factor e.g. contamination, rough surface due to plastic deformation during wire bond scrubbing and etc. as observed using optical method. 3.2 IMC coverage vs HTS condition With the optimum IMC coverage calculation method defined, two different thermal aging condition 200C@5hrs and 250C@8hrs sample were analysed. In average, baking condition at 250C@8hrs having 20% more IMC coverage compare to 200C@5hrs. See Figure 6 Baking condition: 250C@8hrs Figure 6: IMC analysis result comparing 200C@5hrs and Ball shear test result shows the shear strength was increased significantly after bake. In additional to it, baking sample at 250C@8hrs exhibited higher ball shear strength compare 200C@5hrs. The increase in shear strength is mainly due to the thicker IMC growth during higher thermal aging temperature. See Figure 7 Figure 7: Ball shear test comparison T0, 200C@5hrs and No occurrence of lifted ball bond found after thermal aging up to 250C@8hrs. On the contrary, observed drop in pull strength after baking. Literately, this can be explained by annealing of wire under high temperature environment in unmolded strip form sample. Wire grain size become larger and this is lead to lower breaking load during pull test. See Figure 8

4 Figure 8: Ball pull test comparison T0, and IV. THE NEXT STEP Based on the above finding & learning, an optimize wire bond parameter had been developed, the output response after wire bond are showing good result including IMC coverage is >75% after thermal annealing at This sample are now subjecting to HTS stress according to AEC Q100 requirement. The HTS stress condition define in AEC Q100 can either be or the next step is to study the IMC growth rate / behaviour under these 2 condition (HTS150c vs HTS175c), does HTS150c@2000hrs is equivalent to HTS175c@1000hrs in term of IMC growth rate? 3.3 IMC thickness vs HTS condition Baking sample were cross section to understand the growth of IMC thickness under two different thermal aging condition. The IMC thickness growth was found in the incremental of 4 to 5 times from 200C@5hrs to 250C@8hrs. See Figure 9 200C@5hrs- IMC thickness range ~0.2um to ~0.5um. 250C@5hrs- IMC thickness range ~1.0um to ~2.0um. V. CONCLUSION Several conclusions can be drawn from this study: 1. SEM method are a more reliable and stable measurement for IMC after wire bond if compare to Optical method. Optical inspection method tend to overestimate the IMC coverage as the light deflection can influence by many factor. 2. CuAl IMC will continue to growth under thermal aging condition, it wills growth vertically and laterally. It seem the IMC growth is more sensitive to temperature rather to duration. 3. Ball shear strength will increase proportional with IMC growth, it is align with the IMC coverage will improve with longer thermal aging duration. Figure 9: Cross section on IMC growth after 200C@5hrs and

5 VI. REFERENCES [1] Wulff, F. W. et al, "Characterisation of intermetallic growth in copper and gold ball bonds on aluminium metallization," Proc. of 6th Electronics Packaging Technology Conference, 2004, pp [2] Saraswati et al, "High temperature storage (HTS) performance of copper ball bonding wires," Proc. of 7th Electronic Packaging Technology Conference, 2005, pp [3] Singh, Inderjit, Sr. et al, "Enhancing Fine Pitch, High I/O Devices with Copper Ball Bonding," Proc. of 55th Electronic Components and Technology Conference,2005, pp [4] 2018 edition of the IC Insights IC Market Drivers A Study of Key System Applications Fueling Demand for Integrated Circuits [5] C.J. Hang, C.Q. Wang, M. Maye, Y.H. Tian, Y. Zhou, H.H. Wang, Growth behavior of Cu/Al intermetallic compounds and cracks in copper ball bonds during isothermal aging, Microelectronics Reliability 48(2008), pp [6] Tracy Jia Lin Yap, Yin Kheng Au, Poh Leng Eu, Importance of Cu/Al Intermetallic Coverage in Copper Wire Bonding with Sensitive Pad Structure IEEE 2012 [7] Wulff FW, Breach CD, Stephan D, et al. Further characterization of intermetallic growth in copper and gold ball bonds on aluminium metallization. In: Proceedings of SEMICON;