April 4, 2014 Subject: Characterization Summary – Copper Bond Wire at Amkor Philippines SUMMARY Per PCN# 03A-14, this document summarizes the electrical characterization that supports an alternate qualified assembly and test site (Amkor Philippines) and alternate qualified material sets. METHODOLOGY The characterization focused on five items: 1) 2) 3) 4) 5) Assessment of critical Bill of Materials Comparison of production yields Assessment of Critical Parameters SSO (Simultaneous Switching Output) Characteristics SERDES performance (ECP3-150 only) BILL OF MATERIALS Product/Package combinations for characterization were chosen to represent a cross-section of the BOM (Bill of Material) changes specified in the PCN. The product/packages and the critical BOM components are: Product/Pkg LFE3-150EA/ 1156fpBGA ASEM Copper (Control) Mold Wire/ Die Attach Compound Diameter EMEG750SE 0.8mil Pd ABLEBOND Coated Cu 2100A AMKOR Copper (New) Mold Wire/ Die Attach Compound Diameter Hitachi 0.8mil Pd Ablebond GE-110 Coated Cu 2300 LFXP2-17E/ 256ftBGA EMEG750SE 0.8mil Pd Coated Cu ABLEBOND 2100A Hitachi GE-110 0.8mil Pd Coated Cu Ablebond 2300 LFXP2-5E/ 144TQFP EMEG700Y 0.8mil Pure Cu Yizbond 8143 Sumitomo G700SY 0.8mil Pd Coated Cu Ablebond 3230 Multiple lots of in various product/package combinations were built as part of the reliability qualification process for the AMKOR Copper BOM. Samples from the qual lots were characterized and compared to comparable lots processed with the released ASE Malaysia (ASEM) Copper BOM. 5555 Northeast Moore Court Hillsboro, Oregon 97124-6421 • T: +1.503.268.8000 • F: +1.503.268.8347 • www.latticesemi.com From an electrical viewpoint, the same wire diameter and same base material implies that electrical performance should be constant. Characterization was performed to confirm that assumption. ASSEMBLY/ELECTRICAL TEST YIELDS The first step in the characterization process is an analysis of process yields. Yield information is critical to gauge the manufacturability of a new package. As Lattice considers yield information proprietary, the yield information below is normalized with respect to the control material, which in this case is the existing ASEM Copper wire BOM. LFE3-150 Normalized Yield LFXP2-17E Normalized Yield LFXP2-5E Normalized Yield Assembly Yield ASEM Copper (Control) 1.00 Electrical Test Yield Amkor ASEM Copper (Control) Amkor 1.00 1.00 1.00 Assembly Yield Electrical Test Yield ASEM Copper (Control) Amkor ASEM Copper (Control) Amkor 1.00 0.98 1.00 0.97 Assembly Yield ASEM Copper (Control) 1.00 Electrical Test Yield Amkor ASEM Copper (Control) Amkor 0.98 1.00 0.94 Lower electrical test yields on the LFXP2-5E were attributable to a test error. No other discernible differences in either assembly yield or electrical final test yields between ASEM and AMKOR copper assembly processes were noted. CRITICAL PARAMETERS For the purposes of this characterization, critical parameters are defined as speed and power. Samples of the qualification lots from AMKOR were tested at the same time as comparative product from ASEM. The tabulated statistics, Cpk values and histograms of the actual distributions are shown below. Note that Tpdcounter is a Built-in Self Test (BIST) routine that is correlated to datasheet parameters. Higher counts equate to faster devices. Note that in most cases the Cpk of the Amkor material is actually better than the control units but the change is not significant and more a function of the sample sizes used. The critical parameters look normally distributed and do not point to any significant parametric difference between ASEM and AMKOR copper bond wire. 5555 Northeast Moore Court Hillsboro, Oregon 97124-6421 • T: +1.503.268.8000 • F: +1.503.268.8347 • www.latticesemi.com LFE3-150EA AMKOR Copper lot ASEM Copper lot (control) N 704 590 LFXP2-17E AMKOR Copper lot ASEM Copper lot (control) 1083 7349 54.75 53.1 11.01 20.71 395 395 1.66 0.9 LFXP2-5E AMKOR Copper lot ASEM Copper lot (control) 1921 28435 39.3 23.5 8.93 8.08 172 172 1.5 1 FE3-150E SICC_VCC ASEASEM CuCopper Lot Mean 391.73 612.7 Icc(mA) Std Spec 49.04 2693 140.6 2693 ASEM Copper Lot XP2-17E XP2-5E SICC_VCC SICC_VCC Cpk 2.7 1.5 ASEM Copper Lot ASET Copper lot ASET Copper lot ASET Copper lot Amkor Cu 5555 Northeast Moore Court Hillsboro, Oregon 97124-6421 • T: +1.503.268.8000 • F: +1.503.268.8347 • www.latticesemi.com Product Group Amkor Copper LFE3-150 ASEM Copper (control) TPDCOUNT N Mean 698 31398 1161 31875 Std 1043.9 754.3 Spec 26561 26561 Cpk 1.54 2.35 LFXP2-17E Amkor Copper ASEM Copper (control) 949 2711 42040 41702 1525.4 1733.3 31630 31630 2.27 1.94 LFXP2-5E Amkor Copper ASEM Copper (control) 1686 6112 40180 39331 1158.3 1331.0 32000 32000 2.35 1.84 LFE3-150EA LFXP2-17E LFXP2-5E SIMULTANEOUS OUTPUT SWITCHING PERFORMANCE Electrical characterization of a new assembly facility includes a check of Simultaneous Switching Output (SSO) performance. This characteristic is also referred to as Ground Bounce although it can affect both power and ground supply rails. Different assembly site may have different process or tooling which can affect SSO performance. Since copper build in ASE Malaysia and Amkor have the same bond wire geometry (length and diameter), SSO results are expected to be comparable. A delta greater than 10% is considered significant. Ground Bounce Product Material Amkor Cu FE3-150EA-BFN1156 ASEM Cu Control Max (mV) Min (mV) Max (V) 258 -130 1.93 256 -126 1.942 Delta %Delta LFXP2-17E-FTN256 Amkor Cu ASEM Cu Control Delta %Delta LFXP2-5E-TN144 Output Disturb Amkor Cu ASEM Cu Control Delta %Delta Min (V) 1.448 1.486 2 0.8% -4 3.2% -0.012 -0.6% -0.038 -2.6% 222 244 -144 -154 1.376 1.348 0.904 0.956 -22 -9.0% 10 -6.5% 0.028 2.1% -0.052 -5.4% 332 362 -188 -200 1.728 1.708 0.796 0.758 -30 -8.3% 12 -6.0% 0.02 1.2% 0.038 5.0% As expected, SSO measurements of output high and low disturbs vary by less than 10%. This is within experimental variation. 5555 Northeast Moore Court Hillsboro, Oregon 97124-6421 • T: +1.503.268.8000 • F: +1.503.268.8347 • www.latticesemi.com SERDES PERFORMANCE Similar to SSO performance, increased inductance due to bond wire geometry could affect highspeed operation. The LFE3-150EA was chosen as a characterization vehicle so that SERDES performance could be quantified. Three units of LFE3-150EA were programmed with an actual customer pattern that internally generates a PN7 pattern that is then transmitted over the SERDES channel. Eye diagrams and jitter measurements were collected at nominal temperature and voltage to compare relative performance. Pre-emphasis is off. LFE3-150EA 1156fpBGA Eye Diagrams (3.07Gbps) AMKOR Amkor ASEM (Control) Delta % Delta ASEM (Control) Duty Cycle Dist (%) 2.00 2.10 Eye Jitter (ps) 8.07 8.02 Eye Width (ps) 603.35 605.37 Eye Height (mV) 624.95 634.25 -0.10 -4.8% 0.06 0.7% -2.02 -0.3% -9.30 -1.5% SERDES Jitter Statistics As can be seen by the eye diagrams above, SERDES performance has not been measurably affected by the BOM change. SUMMARY There are no significant electrical performance issues related to the new alternate qualified material from AMKOR. Lattice recommends the production release of products from Amkor Philippines.