Evaluation of the Degradation Dynamics of Thin Silicon Dioxide Films using Model-Independent Procedures
- PDF / 433,686 Bytes
- 6 Pages / 416.88 x 635.4 pts Page_size
- 84 Downloads / 218 Views
Mat. Res. Soc. Symp. Proc. Vol. 592 ©2000 Materials Research Society
2
- --Pre-stressed oxides virgin oxides
.--
001
-2 10-3
QBD
(C)
0-2
Fig. 1. Gumbel plot of the statistical breakdown distributions of +10.6V/15s pre-stressed oxides and virgin oxides (thickness: 8nm, area: 3.24x10 4 cm 2 ). Breakdown was always provoked during a +10.3V CVS. The distribution measured on pre-stressed oxides is shifted towards lower values of QaD. The shift from the distribution measured on virgin oxides, AQ, is a measure of the degradation induced during the pre-stress step. (final breakdown test), breakdown is always provoked during a predetermined DC test (high enough to produce breakdown in a reasonable time), independently of the pre-stress conditions. Due to the degradation induced during the pre-stress, the breakdown statistic al distribution of the pre-stressed oxides is shifted towards lower values of the injected-charge-to-breakdown, as expected for a cumulative process. For instance, figure 1 shows the breakdown statistical distributions measured during a +I 0.3V CVS on virgin capacitors (continuous line) and oxides pre-stressed with a +10.6V CVS during 15 seconds (dots), as a function of the injected chargeto-breakdown (QBD). In the case of the pre-stressed capacitors, QBD refers to the injected chargeto-breakdown during the final +10.3V CVS (second step of the method). Those samples which broke down during the pre-stress are counted at QBD=O. Since breakdown was in both cases provoked during a +10.3V DC test, the shift must be related to the degradation induced during the pre-stress stage. In this way, the shift from the distribution measured on virgin oxides, AQ (see figure 1), is a direct measure of the degradation induced during the pre-stress: the larger AQ, the larger the degradation. Note that the method allows a direct measurement of the degradation without any assumption on the microscopic degradation mechanism, as other methods need. Moreover, since breakdown is always produced under the same stress condition, the comparison between different stress tests is, in this way, straightforward [4].
EXPERIMENT The test structures used in this work were n+ polysilicon gate capacitors with z8nm thick oxide, n type substrate and areas of 3.24x 104 cm2 and 2.3x 10-3 cm2 . The two-step stress method has been used to analyse the degradation of these oxides under constant voltage and constant current stress (CCS) conditions. In the first case, CVS ranging from +9.5V to +1 0.9V were
148
chosen as pre-stress conditions. During the second stage of the test, breakdown was always provoked during a +10.3V CVS. The I-t characteristics were recorded during this final CVS test, to compute the injected charge to breakdown. The degradation of 3.24x 10-4 cm2 area oxides 3was analysed in this case. When current was the stressing magnitude, the degradation of 2.3x10cm2 area oxides under constant-current stresses was studied. Current densities ranging from 87mA/cm 2 to 390 mA/cm2 were the pre-stress conditions and breakdown wa
Data Loading...
