Effect of Mechanical Stress on Electromigration Failure Mode During Accelerated Electromigration Tests

  • PDF / 1,374,892 Bytes
  • 6 Pages / 414.72 x 648 pts Page_size
  • 85 Downloads / 277 Views

DOWNLOAD

REPORT


RESULTS & DISCUSSION The interconnect lines were tested for electromigration under various conditions which are group A (current density J = 6.25MA/cm 2 at 263 °C), group B ( current density J = 4.16MA/cm 2 at 243 °C ), and group C (current density J = 6.25MA/cm 2 at 205 'C). In Fig. 1, percent(%) resistance change (A R%) vs. test time (in hours) of a typical interconnect line has been plotted for all three test conditions. For all three test conditions, AR% vs. time behavior is quite different. For group A, interconnect lines fail by an extrusion which is shown by the sharp resistance drop in Fig. 1a at 722 hours whereas for group B, electromigration failure takes place by the void formation which is depicted by resistance increase (at -1400 hours) in AR% vs. time plot. For group C, no failure was observed and test was stopped at 1600 hours. Note for the all three test conditions, there were sharp spikes, see Fig. 1, depicting -2% to -6% increase in resistance during the initial phase of testing. It is most probable that these spikes represent electromigration induced voiding and rapid healing of these voids soon after their formation. Also, note that for the all test groups, an -4% resistance drop was observed which is due to the precipitation of Cu and Si from the Al alloy. The failure modes for test groups A and B were further verified by scanning electron microscopy of failed parts. The SEM micrograph, see Fig. 2a, of the part tested at high current density (a) J=6.25E6AIcm2 0 T =263oC Failure by extrusion

4 ------------...................................

2

.......

0C at 243 and (b) 4.16MA/cm2 (c) 6.25MA/cm2 at 205

-. 6 -2-4 -10 -. N

Ný0N

N

M N

N N

L0

Time (H)

(b)

J=4.16E6AIcm2 0 T =243oC Failure by voiding 64 2

0 --- ................................. . . . . . -2 .4 -6 0 C;

Fig. I . Plot of percentile(%) resistance change vs. test time for following test (a) 6.25MA/cm 2 at 263 °C,

~ c0

0

N

0

0

N W

0 r-

r-.

Nn N

N

0

0 N -

Time (H)

(c) J=6.25E6Acnm2 0 T =205oC No failure 6 4 -2 .6 .8

Time (H)

508

.... Nl CQ

N)

Nl

Fig. 2 : Scanning electron micrographs of interconnect lines showing electromigration failure by

(a) an extrusion for 6.25MA/cm 2 at 263 TC and (b) voiding for 4.16MA/cm 2 at 243 *C.

(6.25MA/cm 2 ) and high temperature (263 0C) reveals that the electromigration failure is due to extrusion (or hillock) only. For the parts tested at 4.16MA/cm 2 and 243 0C, only voids were observed by SEM, see Fig. 2b, and no extrusions or hillocks were found. The scanning electron microscopy was performed on multiple samples from the both test conditions to further reconfirm their respective failure modes. To understand the electromigration failure modes, the mechanical stress state of these interconnect lines and its dependence upon test conditions warrants close attention. The local mechanical stress state in the interconnect is a function of temperature and current density, the two principal variables in electromigration testing. The origins of mechanical stress in these intercon

Data Loading...