Fast Thermal Cycling - Enhanced Electromigratton Failures in Si/TiSi 2 W-Ti/Al-Si-Cu Multilayer Contacts

  • PDF / 1,064,229 Bytes
  • 6 Pages / 414.72 x 648 pts Page_size
  • 93 Downloads / 220 Views

DOWNLOAD

REPORT


Mat. Res. Soc. Symp. Proc. Vol. 391 01995 Materials Research Society

a contact region) was passed through the contacts of each group. The substrate temperature for the devices of the first group was cycled and that of the second one was fixed. The cyclical heating-cooling was provided by the pulsed current passed through the p-Si region of the test structure'. The test condition were chosen so as to ensure a temperature cycling of adjacent contact region with 300 ms thermal cycle period with varying of the thermal cycle amplitude. During the test the main value of the fixed temperature was 150'C, although the test was carried out at 1000C and 2000 C to extract the electromigration activation energy from the Arrhenius plot. To monitor the change in temperature with time we employed the scheme shown in Fig. 1. As follows from the scheme, if a pulsed current was passed throug the p-Si region and a constant current was passed through the thin film resistor, a change in voltage with time, that was monitored by oscillograph, could be recounted easily in the temperature change with time. The time to failure was defined as the time taken to produce both an 1 j.A current leakage of n÷/p diodes at 10 V and 20% rise in contact voltage (contact resistance). The contact sizes used in this study vary from 1.0* 1.0 gim to 1.4* 1.4 jim , although most of the experimental data are obtained from 1.2* 1.2 gm contact structures. EXPERIMENTAL RESULTS The test conditions and a summary of electromigration results are shown in Table 1. The experimental data gathered in Fig.2 and Fig.3 demonstrate the effect of fast temperature cycling on the electromigration results. During the tests the low range of thermal cycling was fixed at 500C. Fig 2 shows that the time to failure for both the failure modes corresponds to a lognormal distribution if the temperature cycling amplitude (AT) is 1250 C and lower. If it rises to 175 0C the lognormal distribution of the time to failure is broken and it tends to shorten the lognormal law. As it is shown in Fig.3 there is a severe degradation of median time to failure (MTF) for both the failure modes if AT increases. However, the cracks in the metallic film were observed mainly in adjacent contact window ranges if the AT reaches 2000C. Change in time to failure distribution as well as the cracks occurring imply that the new degraded processes resulting in failures were consistent under these accelerated test conditions. Taking into account this circumstance we can recommend the thermal cycling conditions with AT =125°C and lower amplitude to carry out an accelerated electromigration test. Table 1. The test conditions and summary of the electromigration results Contact type

Current value/density

Substrate tempera-

Adjacent contact

2 (mA/1O'Acm )

ture (°C)

region temperature

(OC)

Test duration

Fail/ss

t30 Rc/IR

Sigma RIR

(hours)

(hours)

Si/TiSi/ Al-Si-Cu

20/3.0 25/3.8 30/4.5 25/3.8 25/3.8 25/3.8

150 240 2500/2000 20/20 150 270 1650/750 20/20 150 310 1200/550 20/20 100 210 12000/6500 18/20 20