New Project Proposal - PowerPoint by fkm75091

VIEWS: 41 PAGES: 17

									New Project Proposal


 Thermo-electromigration in
 WL-CSP Pb-free solder joints
         Phase II

HDPUG Quarterly Meeting, Feb 2009

    Proposed by K. N. Tu, UCLA

                                    1
             Purpose
• Reduce thermo-electromigration induced failure in WL-CSP Pb-
   free flip solder joints
• Owing to the ban of Pb-based solder in consumer electronic
   products and the trend of miniaturization in wireless and portable
   devices, the reliability of Pb-free solder interconnection is one of
   the most challenging problems in electronic manufacturing
   industry, especially the failure causes by thermo-electromigration.
   Joule heating due to the on-chip Al interconnect has generated
   thermomigration to accompany electromigration in solder joints.
   For a joint of 200 micron in size, a temperature difference of 20
   C across it will cause a temperature gradient of 1000 C/cm,
   which can lead to thermomigration. We define the combined
   effect as thermo-electromigration.
• This proposal is to continue the previous project supported by
   HDP in 2008. There are two parts; the first one is for six months,
   and the second one is for one year.                              2
       Objective
• To determine mean-time-to-failure (MTTF) of thermo-electromigration
  in SnAgCu solders of low Ag with different dopants and different UBM
  at two current densities of 5 x 103 and 1 x 104 A/cm2 and at two
  temperatures of 125 and 150 °C
• The figure on next foil shows the experimental set-up at UCLA for
  electromigration testing of flip chip SAC sample. We can apply two
  current densities and at two temperatures.
• In the set-up for the combination of two current densities and at two
  temperatures, we have measured in each set 4 boards together, each
  board has 4 chip, and each chip has 36 flip chip solder joints (of which
  2 can be electrically tested for thermo-electromigration and 1 can be
  electrically tested for thermomigration effects only). So we have
  obtained statistical distribution of failure.
• Weibull distribution has been used to analyze the data. We
  determined the activation energy as well as the current density
  exponential factor. MTTF equation has been obtained. (see the Final
  Report submitted for the 2008 project).                             3
Experimental Setup (I)

                     Power supply:
                     I1,   I2,   I1,    I2

                                                    USB data
                                                    collection



                                       Furnace 1:
                  Furnace 2:           150 C
                  125 C



       I1: 3.14A (104A/cm2), I2: 1.57A (5 x 103A/cm2)            4
Technical discussion
      •   In this project, so far we have studied mainly the failure due to
          thermo-electromigration. We have obtain MTTF equation on
          the basis of two current densities and two temperatures.
      •   To have a better fit to Black’s equation, we need to test at the
          least one more temperature and one more current density.
      •   We have not studied thermomigration. While we have studied
          the combined effect of electromigration and thermomigration on
          failure. We need to decouple them.
      •   In electromigration tests of flip chip solder joints, there are two
          factors which affects the MTTF analysis; one is the effect of
          current crowding due to the line-to-bump configuration, and the
          other is joule heating induced thermomigration which
          accompanies electromigration. We have studied the former,
          but not the latter.
      •   It is proposed here to have a one-year project so that we can
          have more data for better fitting of Black’s equation and also to
          study thermomigration.
      •   Alternatively, a six-month proposal is also presented to
          complete the on-going work.

                                                                                5
On-going Work


     • We are running the rest of the tests at
       current density of 5 x 103 A/cm2 and at
       125 and 150 °C. It will take at the least 2
       to 3 months to finish them.
     • We are also running AlNVCu/SAC1205
       samples at 100 °C with current density at
       1 x 104 A/cm2. It may take 3 to 6 months
       to finish it.



                                                     6
Black’s Equation Fit to the AlNiVCu/SAC1205 data
(180°C to 155°C at 5,000A/cm2 to 10,000A/cm2)


  Material Composition              500 A/cm2                        1,000 A/cm2
    UBM           Solder     125° C Oven 150° C Oven 100° C Oven 125° C Oven 150° C Oven
                  SN100C       Phase 2      Phase 2          --        complete     complete
  Ti/Cu/Cu
                 SAC1205       Phase 2      Phase 2          --        complete     complete
                  SN100C       Phase 2      Phase 2          --        complete     complete
 e-less Ni/Au
                 SAC1205       Phase 2      Phase 2           --       complete     complete
                  SN100C       Phase 2      Phase 2          --        complete     complete
  Al/Ni/V/Cu
                 SAC1205       Phase 2        8/16        Phase 2      complete     complete




                        UBM               Solder       Ea (eV)         n            A
                     Al/Ni/V/Cu          SAC1205         1.15          2       7.87 x10-4
                • SN100C: Sn (R), Cu, 0.65%, Ni, 500ppm, Ge 60ppm, Bi 110ppm, Pb 140 ppm
                                                                                               7
                • SAC1205: Sn(R), Ag 1.2%, Cu 0.5%
Proposed work for Jan to June 2009

        On-going work (to be finished in January to June, 2009)

        ● Electromigration Tests of the 6 experimental Legs:
           1. 5,000A/cm2@150C ambient testing
           2. 5,000A/cm2@125C ambient testing
           3. 10,000A/cm2@100C ambient for Leg 8, testing.
        ● Ball Impact Toughness after Annealing at 180C for
           24 hrs and 72 hrs.
        ● FIB/SEM examination of failed samples in the 6
           experimental Legs (examples are shown in the next two
           foils for the TiCuCu/SN100C experimental leg).



                                                                   8
Failure analysis by FIB




         Pancake void         Kirkendall voids




         Sn             IMC      Cu,              Al

                                     Void in Al trace




                                                        9
Scanning Ion Tunnelling Image




                     Kirkendall voids      Cu-2
      Pancake void

                                    Cu-1
                                              Al


                Cu6Sn5

                           Cu3Sn



                                                   10
Six-month Project Plan
      Cost: $20,000
     01/01/09 – 03/31/09 (Y2Q1)
     • Electromigration tests as a function of low current density of 5 x
        103 A/cm2, time and temperature. Cross-section of failed
        samples at higher current density prepared by FIB and SEM.
     04/01/09 – 06/30/09 (Y2Q2)
     • Electromigration experiment tests continue as a function of current
        density, time and temperature. To confirm the mode of failure so
        that we can confirm the test parameters for MTTF.
     • Complete the failure analysis, complete project report, and
        prepare journal publications.
     • Do additional experimental testing to quantify the thermomigration
        and stressmigration effects. If these effects are deemed to be
        significant, consider whether the project can be extended for
        another 6 months to run 3D FEM modeling quantifying the
        individual effects of electromigration, thermomigration and
        stressmigration. (The cost to extend the project for this additional
        6 months will be $25,000 in order to hire a post-doc to assist with
        the computer modeling and to do the additional experimental
        verification work.)
                                                                               11
Proposed work for Jan. to Dec., 2009

            Finish the on-going work ($10,000 per Quarter for Q1 and Q2)
        •   Electromigration Tests of all 6 experimental Legs
              1.Two current densities of 5 x 103 A/cm2 and 1 x 104 A/cm2
              2.Two temperatures of 125 °C and 150 °C
        •   Ball Impact Toughness after Annealing
            (24 hrs and 72 hrs@180C)
        •   FIB of cross-sections of the all 6 experimental Legs.
            New proposed work ($10,000 per Quarter for Q3 and Q4)
        •   Electromigration Testing of all 6 experimental Leg at
            2 x 104 A/cm2 and 100 °C
        •   FIB/SEM examination of all failed samples
        •   MTTF analysis over these wider experimental conditions.
        •   Decoupled Electromigration, Thermomigration, Stressmigration Testing
        •   If Thermomigration and/or Stressmigration are deemed to be
            significant, hire a post DOC to do 3D finite element modeling to
            quantify the magnitudes these effects ($12,500 per Quarter for Q3 and
            Q4 if 3D modeling is added).                                          12
One-year Project Plan
     01/01/09 – 03/31/09 (Y2Q1)
     ● Continue electromigration tests as a function of low current density of 5 x 103
         A/cm2, time and temperature. Cross-section of failed samples at higher current
         density prepared by FIB and SEM. Start electromigration tests at 100 °C of high
         current density of 2 x 104 A/cm2.
     04/01/09 – 06/30/09 (Y2Q2)
     ● Electromigration experiment tests continue as a function of current density, time
         and temperature. Start thermomigration tests to examine un-powered neighboring
         solder joint. FIB and SEM cross-section samples prepared.
     Project Costs for Q1 and Q2: $10,000 per quarter (to finish the current work).

     07/01/09 to 09/31/09 (Y2Q3)
     ● Confirm the mode of failure of electromigration, so that we can confirm the test
         parameters for MTTF. Decouple the effect of electromigration and thermomigration
         on failure. To determine the failure mode of thermomigration.
     10/01/09 to 12/31/09 (Y2Q4)
     ● To finalize MTTF equation and to determine the activation energy and exponential
         factor of current density for Leg 1 to Leg 8 samples. Complete the failure analysis,
         complete project report, and prepare journal publications.

     Project Cost for Q3 and Q4: If significant thermo-migration or stress-migration effects
         can be detected in Q1 and Q2 of this project, Prof. Tu would like to hire a post-doc
         to assist with 3d FEM modeling to quantify the magnitudes of the
        1) electro-migration, 2) thermo-migration and 2) stress-migration effects. This
         would increase the costs for Q3 and Q4 to $12,500 per quarter.                       13
Resources and Requests

     • UCLA has been supported by NSF, SRC, NSC, and Intel
       over the last ten years to study electromigration in flip chip
       solder joints. We have built two sets of test stages with
       multi-channel control to perform large scale
       electromigration tests, so we can obtain statistical
       distribution of failure. Also we have FIB and SEM for
       imaging and e-beam probe for composition analysis.
     • We have access, in Taiwan, to 3-D FEA simulation of
       current distribution, joule heating, and temperature
       distribution in flip chip solder joints, including Al
       interconnect and under-bump-metallization.
     • Project Updates to be preformed on a bi-monthly basis.
     • To perform the proposed work, the support of one M. Sc.
       student for six months at $20,000 or for a year at $40,000 is
       requested (or $45,000 if a post-doc for 3D FEA is hired).



                                                                        14
“Lead-Free and PbSn Bump Electromigration Testing.” Stephen Gee,
Nikhil Kelkar, Joanne Huang and King-Ning Tu, IPAC2005


                                     National Semiconductor & UCLA
                                     Experimental Results:

                                     LEFT: Chart showing the improved performance of
                                           SnAgCu over eutectic PbSn at 1.7 Amps and
                                           150°C.

                                                       RIGHT: Cross-Section of SnAgCu
                                                              after 1475 Hours at 1.8A
                                                              and 150°C (Right)




                                                                                         15
“Modeling of electromigration combined with thermal–mechanical effect
for IC device and package”, Yong Liu, Lihua Liang, Scott Irving, and
Timwah Luk, Microelectronics Reliability, 2008.



                 Fairchild & Zhejiang University 3D modeling of maximum atomic flux
                 divergence due to electro-migration, thermo-migration and stress-
                 migration indicate that SnAgCu should perform better than eutectic
                 PbSn.




                 Predicted Void Formation for SnAgCu after 1074.6 hr at 1.8A and 150C.




                                                                                      16
Team Liaison

     NSC – Stephen Gee
     HDP – Marshall Andrews
     Others are welcome to join.




                                   17

								
To top