Eg. Colgan et al., KINETICS OF AL GRAIN-GROWTH, AL2CU PRECIPITATION, AND DISSOLUTION IN BLANKET THIN-FILMS AND FINE LINES, Journal of applied physics, 76(5), 1994, pp. 2781-2790
The apparent activation energy E(a) for Al grain growth, Al2Cu (THETA-
phase) precipitation, and Al2Cu dissolution were determined by ramped
resistance measurements for both Al(Cu) blanket films and patterned li
nes. The E(a)'s measured for the initial stages of grain growth in 0.5
-, 1-, and 2-mum-thick Al(4 wt % Cu), Al(2 wt % Cu), and Al films rang
ed from 1.19 to 1.46 eV. The E(a)'s for grain growth were higher for 0
.6-0.9-mum-wide Al(Cu) lines than for blanket Al(Cu) films 1.89-3.1 eV
, and the temperature of the peak transformation rate occurred at a mu
ch higher temperature, 310-400 vs 90-155-degrees-C. This is due to the
geometric constraints in patterned lines. The E(a)'s for Al2Cu precip
itation in Al(4 wt % Cu) and Al(2 wt % Cu) films varied from 0.86 to 1
.25 eV. For 0.6-mum-wide Al(4 wt % Cu) lines, the E(a) for Al2Cu preci
pitation was 1.7 eV The E(a)'s for Al2Cu dissolution increased with de
creasing Cu content from 1.62-1.74 eV to 2.23-2.30 eV with Al(4 wt % C
u) and Al(2 wt % Cu) films, respectively. The temperature of the peak
reaction rate T(p) for Al2Cu dissolution increased markedly with incre
asing film thickness at constant ramp rates. These results demonstrate
that the microstructure and Cu distribution in Al(Cu) interconnection
s on microelectronic chips vary as a function of feature size. This im
plies that blanket film data is not necessarily applicable to patterne
d features.