Effect of CU and Si in Aluminum on Stress Change and on TiAl 3 Formation in Al Alloy/Ti Bilayer Films During Annealing
- PDF / 406,528 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 42 Downloads / 186 Views
471
Mat. Res. Soc. Symp. Proc. Vol. 356 ©1995 Materials Research Society
The composite stress evolution of the Ti/Al bilayers during thermal cycling was measured using a Flexus F2320 (nitrogen ambient). Figure 2 shows the composite stress-temperature behavior for a 500A Ti/1400A AI-I%Cu stack for two different thermal cycles. The samples were heated to 430°C and then either cooled directly to room temperature or held for 2 hours before cooling to room temperature. The heating rate was 6.8 °C/min, and the cooling rate ranged from I 10C/min at 430'C to 2TC/min near room temperature. Figure 3 shows the composite stress-temperature behavior for a 500A Ti/1400A Al-0.5%Cu-l%Si stack for the same two thermal cycles as Figure 2. Ti/1400A Al-I%Cu and Ti/1400A Al-0.5%Cu-l%Si bilayer films were annealed at 430°C for 20 hours and cooled to room temperature prior to the thermal cycle in Figure 4, which has no hold at 430TC. The thermal cycle plotted in Figure 4 is the second cycle to 430'C for each sample. Cross-sectional Transmission Electron Microscopy (TEM) was also used to study the Ti/Al-alloy reaction. TEM revealed a thin interaction layer (-100A) in the as-deposited Ti/AI-I%Cu film stack while the Ti/Al-0.5%Cu-l%Si stack showed no interaction layer (as-deposited). TEM observations of the annealed films confirmed the TiAl3 thicknesses determined from sheet resistance measurements. The TiAl3 layer formed after annealing (in both the Ti/AI-I%Cu and Ti/Al-0.5%Cu- I%Si stacks) was rough and the grains were relatively small. X-ray texture analysis revealed that the TiAl3 layer had a broad (112) texture, formed from initial sharply-textured Ti(0002) and Al(1 11) films.[5] RATE OF TiAI3 FORMATION The sheet resistance measured as a function of anneal time at 430'C was used to calculate the TiAl3 thickness based on the thickness of Al-alloy consumed. The Ti/AI-I%Cu stack reacts faster than the Ti/AI-0.5%Cu-l%Si stack. In fact, the Ti/AI-I%Cu reaction is almost complete after one cycle. When the TiAl3 thickness squared is plotted as a function of time (Figure 1), there is a linear dependence, indicating diffusion-controlled growth. This can be represented by Equation (1) [6], h2 =
4
Dat
(1)
where h is the TiAl3 thickness, t is time, and Da is the effective chemical interdiffusion coefficient in TiAl3. For the Ti/Al-0.5%Cu-l%Si stack, Da is lxl0- 5 [Lm2 /min at 430'C (from Figure 1). It is apparent that the reaction rate is much faster for the Ti/AI-I%Cu stack, with most of the reaction occurring in the first thermal cycle. This fast reaction rate leads to significant TiAI3 formation while heating to, and cooling from, the annealing temperature. This is supported by TEM observations of an intermixing layer after deposition at 250'C. Because of the fast reaction rate, it is difficult to calculate an accurate, quantitative interdiffusion coefficient for the Ti/All%Cu stack using these thermal cycles. However, it is apparent from the data that Da for the Ti/AI-I%Cu stack is larger than Da for the Ti/AI-0.5%Cu-l%Si stack. Experimen
Data Loading...
