Characterization of 0.6mils Ag Alloy Wire in BGA Package

Transcription

1 Characterization of 0.6mils Ag Alloy Wire in BGA Package Toh Lee Chew, Alan Lumapar Visarra, *Fabien Quercia, *Eric Perriaud STMicroelectronics Muar, Tanjung Agas Industrial, P.O.Box 28, 84007, Muar, Johor *STMicroelectronics Grenoble, 12 rue Jules Horowitz, B.P. 217, Cedex, Grenoble, France Phone: (+606) Abstract With cost erosion in semi-conductor business, manufacturing cost is becoming a real driving force for new trend. For Back End (BE), package cost is then one of the key to control the cost. For high I/O Wire Bond (WB) packages e.g. Quad Flat Packages (QFP), Ball Grid Arrays (BGA) etc., interconnect wires is one of the dominant portion on package cost, conventionally with gold (Au) wire. Despite Au wire is superior in bond-ability to different materials, soft mechanical properties for process workability and loop forming flexibility, the material cost and its market price fluctuation which giving big impact to product sales margin, is forcing the industry to look for alternatives. To respond this scenario, Bare Copper (Cu) and Palladium (Pd) Coated Copper are two of the successive substitutes which feasibility already proven and already industrialize. However, concerns and drawbacks have been raised up in terms of bonding quality, bonding process time, package limitations and reliability performance. A newly alternative material, Ag Alloy wire has been developed recently with its advantages of material and mechanical properties that similar to gold wire. Ag Alloy wire is now becoming latest technology trend that fully study as a potential alternative solution for gold wire bonding. Generally, characteristics and technical challenges of Ag Alloy wire must be well understanding and examining. This paper, intended to summarize the characterization of Ag Alloy wire in BGA package. The characterization covers (i) Capillary selection that focus on tools design based on fine pitch application requirement; (ii) Enhancement of forming gas control on free air ball formation; (iii) Process aspects (EFO; Base parameters) that fully optimized based on bondability and workability; (iv) Intermetallic (IMC) growth vs time on unmolded units to study behavior of Ag Alloy wire; (v) Reliability aspects on molded units at high temperature storage (HTS), thermal humidity storage (THS) and thermal cycle (TC) to verify and justify Ag Alloy wire reliability performance. 1. Introduction With the excessive rising of gold price in recent years, a lots of semiconductor manufacturers have looking alternative materials to replace gold wire bonding. Bare copper (Cu) and palladium (Pd) coated copper wires are two of the materials that have been successfully introduced into semiconductor packages recently. However, bonding reliability and bonding quality like lifted ball, pad peeling, poor assembly yield had proven the limitations of copper wire bonding development especially for fine pitch applications in BGA packages. Therefore, Ag and Ag Alloy wires have been taking in place to become newest bonding materials that substitute for gold wire [1]. Ag and Ag Alloy wires have excellent electrical and mechanical properties that similar to gold wire. Table I List of Material Properties Material Properties Au PdCu Ag Alloy Young Modulus (GPa) Resistivity,20 C (µωcm) Hardness (Hv) Thermal Conductivity (W/mK) Melting Point ( C) Based from recent research, Ag Alloy wire which is more reliable had been selected for this study [2,3]. This study was carried out from aspect of bondability, workability and reliability performances on BGA package. Study had been conducted firstly on direct and indirect materials selection such as wire and capillary. Emphasis is placed on output performance with the selected silver alloy composition while capillary focused on tool designs that depend on package requirement. For advance fine pitch application, combinations of materialization are a significant factor that being contribute on successful bonding. As off-centered ball is well-known criteria for 0.6mils silver alloy bonding, optimum electronic flame off (EFO) parameters with enhancement of forming gas must well analyze in order to create perfect FAB formation. Furthermore, main bonding parameters for 1st and 2nd bond such as force, power, time and speed are optimized. Qualities checking on output responses are examined by standard wire pull, ball shear, cratering and intermetallic test. Bonded ball profile had been analyzed as well by using conventional way like cross section, together with advance surface profiler technology too. Unmolded bonded package with accelerate baking condition was study to have fast understanding on silver alloy wire interaction and reliability performance. Also, behavior of intermetallic growth based on time over temperature had been input for investigation and observation. To further understanding workability of silver alloy wire bonding, common industrial reliability test have been performed on molded units. Full bonded package has been conducted on standard reliability test such as high temperature storage (HTS), thermal humidity storage (THS) and thermal cycle (TC). 2. Experiments The test vehicle used on this study is LFBGA 15x15 package on CMOS40 wafer technology with 40um bond pad opening. The study was conducted on aluminum bond pad with 0.6mils wire diameter. The details of the test vehicle are shown in below table.

2 Table II Test Vehicle Details Package Type LFBGA 15x15 Wire Diameter 0.6mils Wire Composition Ag Alloy with 90%Ag content Wafer Technology CMOS40 Bond Pad Material Aluminum Pad BPO/BPP 41um/45um The bonding study was conducted on ASM bonder with installment of nozzle kit. SPT capillary was designed based on package application as mainly focus on chamfer diameter, hole diameter, tip diameter and bottle neck height [4]. on few reliability tests such as high temperature storage (HTS), thermal humidity storage (THS) and thermal cycle (TC). 3. Results and Discussions A. FAB formation of 0.6mil Ag Alloy Wire As EFO firing is very important during FAB forming, the EFO parameters must be optimize. EFO parameters like EFO current and time will need to fine tuning in order to get desired FAB size. As mentioned earlier, forming gas is essential for good ball formation, thus it is a need to input on the study. Justification was done as trial without forming gas also carried out on the matrix. Based on the results as summarize on below table, few forming gas settings can formed good ball formation. Table III Matrix of Forming Gas Flow Test Forming Gas Flow Off-Centered Ball Good Ball No Gas Apply 30 0 Gas Flow Setting Gas Flow Setting Gas Flow Setting Maximum Gas 8 22 Figure 1 Capillary Drawing Design As free air ball (FAB) formation is very important for the bonding, the FAB of Ag Alloy wire was first investigated. The forming gas, mixture of nitrogen and hydrogen was used on this study in order to prevent oxidation and to create better FAB formation. Optimum electronic flame off (EFO) parameters and forming gas control must well analyze for this fine wire diameter bonding. The FAB formation inspection was carried out to detect abnormal ball shape and any oxidation during firing process. After obtained the good FAB formation, the bonding tests was performed with process parameters reference from Au wire bonding. The output responses were examined and measured such as ball size, ball height and aluminum splash. Standard qualities checking on bonded ball such as wire pull, stitch pull and ball shear were performed as well on Dage 5000 series. To further investigate workability of Ag Alloy wire, the unmolded bonded samples were sent for high temperature storage on various time intervals. Also, the qualities evaluation were carried out to have fast understanding of Ag Alloy wire reliability performance. Cross section had been performed to analyze the IMC growth after high storage temperature on various times. To justify the Ag Alloy wire reliability performance, molded samples had been conducted Figure 2 Off-Centered Ball and Good Ball Formation Further validation was carried out on optical inspection for bonded balls. Nine full bonded units with around 10k bonded balls were inspected for ball bond geometry. Final forming gas flow setting shall base from performance on bonded balls. Optimum combination of EFO parameters together with forming gas can achieve 100% good bonded ball geometry. Off-centered balls had observed on nonoptimum forming gas settings. Figure 3 Off-Centered Ball and Good Ball Geometry

3 B. Bonded Ball Response and Quality Checking Next, standard in process control criteria such as ball size, ball height, aluminum splash were measured. In order to achieve the target process control, the bond parameters must be optimized which are quite similar to normal Au wire bonding parameters settings. From the results, the aluminum splash can be observed although was lesser compare to copper bonding. Due to the fine pitch bonding on 0.6mils wire, the bonded ball size was target on 32um with the aluminum splash less than 37um. Standard qualities checking criteria such as neck pull, stitch pull and ball shear were performed on 60 bonded balls. The pull test and ball shear tests were important to confirm the bonded balls strength on time zero. Ball lift and peeling is the rejected criteria during these tests. While stitch pull test was important to check the 2 nd bond adhesion on lead finger. Optical checking must be done in order to check and analyze all the wire pull and ball shear failure modes. IMC checking was performed to analyze the IMC coverage on optimized parameters on bonded ball. The normal target on IMC coverage for bonded balls must achieve above 75% which same target as gold wire bonding. Also, the cratering test was performed on bonded pads in order to ensure no pad damage issue on the Ag Alloy bonding. Conventional cross section on bonded ball was performed to analyze the bonded ball shape and aluminum remnant. In addition, modern surface profiler was performed in order to have more details analysis on bonded ball shape and aluminum remnant. Table IV Time 0 Bonded Balls Results Criteria Results Stitch Pull Ball Shear Cratering Test IMC Coverage Average 3g; with stitch remain Average 10g; 100% alu shear No Cratering Found on 5 Units Ball Size Ball Height 30 to 34um 6 to 10um Above 75% Coverage Bonded Ball Geometry T0 Cross section Good Bonded Ball More than 30% aluminum remnant Aluminum Splash Below 37um Surface Profiler Pull Test More than 30% aluminum remnant Average 4g; 100% neck break

4 C. Unmolded Parts Reliability Results To have fast understanding of 0.6mils Ag Alloy on reliability performance, the unmolded bonded samples were sent to high temperature storage. The unmolded HTS test was carried out on 175degree with various time intervals. The pull test and ball shear data were collect on each time interval. The output responses were recorded and failure modes were checked to ensure no ball lift and peeling happened. Based on the responses, ball shear had shown increasing shear value which correlates to intermetallic growth of Ag Alloy, while constant pull test readings were observed with 100% neck break. Overall, the ball shear and pull test passed the criteria during unmolded parts reliability. Table V Unmolded Parts HTS Results S/Size : 60 data Pull Test Ball Shear 0 Hour 100% Neck Break 100% Alu Shear 168 Hours 100% Neck Break 100% Ag Shear 240 Hours 100% Neck Break 100% Ag Shear 336 Hours 100% Neck Break 100% Ag Shear Figure 6 Unmolded HTS Pull Test Results To further analyze and understanding of intermetallic growth between Ag Alloy with Al pads, cross section and SEM were performed on the samples. After 168 hours of baking condition, the intermetallic starting to grow and form. The IMC layers continuously growing during 240 hours and full saturated observed on 336 hours. From the figure 7, clear and thick IMC layers can be seen after 336hrs HTS test. There is no crack line or void observed on the layers as well. (a) Figure 4 Example of Ag Shear after HTS test (b) (c) Figure 5 Unmolded HTS Ball Shear Results (d) Figure 7 IMC layers after 175degree on (a) 0 hour, (b) 168 hours, (c) 240 hours, (d) 336hours

5 D. Full Package Reliability For the full package reliability tests, three tests such as high temperature storage (HTS), thermal humidity storage (THS) and thermal cycle (TC) were selected. The assembly good units were built to perform the reliability tests, 77 units for each test conditions. The electrical test was performed after on each reliability test conditions to check for O/S failures, as per below reliability matrix. All reliability sample units were successfully passed. Table VI Package Reliability Results Reliability Test Conditions Results (O/S) JL3 + TC 100 & 500hrs Pass JL3 + HTS 168 & 500hrs Pass JL3 + THS 168 & 500hrs Pass 4. Conclusions In overall, the 0.6mils Ag Alloy wire workability is comparable as Au wire for BGA package on this study. The challenge on off-centered ball was resolved thru series of forming gas characterization study. Ag Alloy wire also had shown positive results on all the quality checking tests like pull test, ball shear, stitch pull, cratering, intermetallic check and cross section analysis. Various reliability tests have also shown that Ag alloy wire was able to stand and perform as Au wire. Generally, the 0.6mils Ag Alloy can become alternative source for Au wire on fine pitch application bonding. Acknowledgments The authors would like to thank ASM team for the support on the 0.6mils Ag Alloy Wire study. References 1. Jie Wu et al., Study of Ag-Alloy Wire in Thermosonic Wire Bonding, Electronics Packaging Technology Conference (EPTC), IEEE 14 th, 2012, pp Liao J. K. et al., Silver Alloy Wire Bonding, Electronic Components and Technology Conference (ECTC), IEEE 62 nd, 2012, pp Cheng C. H. et al., Low Cost Silver Alloy Wire Bonding With Excellent Reliability Performance, Electronic Components and Technology Conference (ECTC), IEEE 63 rd, 2013, pp Jason Tan et al., Breaking Barriers in Silver Alloy Wire Bonding For Ultra Fine Pitch Application, SEMICON SINGAPORE 2014, 2014.