Wafer Temperature

The minienvironment approach to contamination control has been increasing in use. A minienvironment is a localized environment created by an enclosure that isolates the product wafer from contamination and people (48). Another approach is using integrated processing, where consecutive processes are linked in a controlled environment (32). Both requite in situ sensors (qv) to measure internal chamber temperatures, background contamination, gas flow rates, pressure changes, and particularly wafer temperature (4). 

It can be observed that these thermal conductances G(7) are typical of phonon conduction between two solids at very low temperature, as already reported [45], The value of the heat capacity was calculated from equation C = r G, where the thermal time constant r is obtained from the fit to the exponential relaxation of the wafer temperature. 

While the resulting model is not quantitatively predictive, important observations can be made based on parametric simulation studies. It is proposed that changes in viscosity due to wafer temperature may be as large as 30%, and that such viscosity dependencies can have significant impact on fluid film thickness and transitively on removal rate. The importance of other process parameters, such as wafer curvature, is also indicated by the model. 

Consider an atmospheric-pressure process to deposit a silicon film from a silane (SifLj) precursor. The showerhead-to-wafer distance is 3 cm. In this process a helium carrier gas makes up the bulk of the flow, with the active silane accounting for only 0.17% of the inlet mixture. The precursor gases enter the reactor at 300 K, but the wafer temperature and inlet velocity are varied to observe different process characteristics. 

Fig. 17.16 Boundary-layer predictions of temperature and selected species number densities (cm-3). The wafer temperature is 1050°C and pressure is 6 Torr (798 Pa). The nitrous-oxide flow rate is Q = 3 standard liters per minute, with the inlet mixture being 3% H2 in N2O.

A new flame-based process is being considered to oxidize a film on a wafer. As illustrated in Fig. 17.22, a combustible mixture flows downward from a showerhead manifold onto the wafer and the exhaust products are drawn out through an annular channel. A control system is presumed to hold the wafer temperature at a fixed temperature. The objective of the process is to deliver an atomic-oxygen number density of approximately 1015 cm-3 at the wafer surface while the wafer temperature is held at approximately Tw = 320°C. Assume the following nominal process conditions showerhead-to-... 

Explore a range of potential design and operating alternatives. For example, consider the effects of H2 stoichiometry, inlet velocity, process pressure, and wafer temperature. Consider the behavior of radical production in the high-temperature regions (i.e., the flame zone) and the recombination in the relatively cool surface boundary... 

As the strain rate increases (e.g., higher inlet velocities), the flame can be extinguished. Determine how process pressure, wafer temperature, and mixture stoichiometry affect the extinction limits. 

It may still be desirable to have a moderately high wafer temperature for other reasons. For example, at very low temperatures, film density may be low, or temperature may play an important role in determining film structure. Nonetheless, it is possible to operate at lower wafer temperatures than would be allowed by a strictly thermal process. 

The process of "characterizing" a reactor can be illustrated for a parallel-plate cold-wall reactor operated at 50 kHz.8 System power was kept at 500 W, pressure at 200 mTorr, and wafer temperature of 240°C. Wafers are placed on a circular electrode which is rotated to promote uniformity of deposition. Therefore, we are only interested in the radial variation of deposition rate. Reactive gases enter at the center and flow out at the periphery. 

When deposition is controlled by surface chemical reaction conditions in the low temperature regime, uniformity over the wafer surface and repeatability between runs are almost entirely determined by the uniformity and repeatability of wafer temperature. This is the case for BST thin films for DRAM applications, as mentioned previously, so that a very stringent requirement is control of the wafer temperature. Usually, the [(max. - min.)/ 2average] uniformity... 

Therefore, it is quite logical to emphasize low temperature MOCVD (wafer temperature of less than roughly 500°C) of BST tWn films even though the status of the research is still behind that of the high temperature deposition process (wafer temperature of higher than roughly 600°C). Data on films from high temperature CVD is discussed only for reference purposes. 

An interesting observation results from the comparison between the Ti concentrations in Figures 15 and 17. When the Ti precursor input rate is increased by a factor of 2.93 (4.92/1.68) the Ti concentration of the film is increased by a factor of only 1.6 for a wafer temperature of 470°C, whereas it is increased by a factor of 2.73 when it is 420°C. This implies that as the deposition temperature increases, a type of selfregulating mechanism of film composition to the stoichiometric one is present, as in the case of Pb incorporation into PZT thin films. This kind of self-regulating mechanism can not generally be understood from simple thermodynamic calculations which usually predict formation of... 

Figure 20 shows excellent step,coverage of one of the MOCVD BST films deposited by the dome type reactor with single injection nozzle at a wafer temperature of 413°C. The substrate has a linelspace pattern with an aspect ratio of 1 6 made of SiO covered with a very thin Pt film. The step coverage is more than 80%. The featureless surface morphology of the film implies that the film has an amorphous structure which should be crystallized by proper postannealing. As discussed previously, however, the hydrocarbon incorporation problem for low temperature CVD must be considered. 

Fig. 20 SEM micrograph showing exceiient step coverage of the MOCVD BST film deposited in a dome type reactor with singie injection nozzie at a wafer temperature of 413°C.

Aluminum. This is by far the most commonly used interconnect material. It can be doped with elements such as Si [Learn93, Hirashita et al.92] and Cu [D Heurle91] to improve the properties such as contact reliability and electromigration. The maximum wafer temperature allowed once aluminum is present is about 400-430°C. 

Selectivity The most important parameters for selectivity for the SiH4/WF6 chemistry are the temperature and the reactant flow ratio. Although there is some dispute on how to determine exactly the wafer temperature (see section 7.3), there is a general belief that the selective temperature window is rather narrow (270-320°C). Below about 250°C there is no growth at all and above 350°C the selectivity is completely lost, as only blanket depositions are observed. See section 3.5 for more details about loss of selectivity. 

Nevertheless, one substantial advantage of the tube systems is that the tube can be considered more or less an isothermal system. This is very advantageous since now the determination of the real wafer temperature is not a problem. This is, as we will see, in contrast with cold wall systems where the real wafer temperature is very difficult to measure and sometimes difficult to control. 

We see that at lower pressures radiation dominates the heat transport. However, at about 500 mTorr the amount of heat transported by either route is almost equal. At 10 Torr radiation accounts for only ca. 10% in the overall transport. This implies that at pressures of 10 Torr and greater, the wafer temperature becomes independent of the emissivities of the chuck and the back side of the wafer. This is nicely illustrated by the data in table 7.3. 

Figure 7.3. Wafer temperature versus pressure. The chuck is coated with tungsten. The backside of the wafer is silicon, the front side is coated with tungsten. [From ref. 174, reprinted with permission].

Some more insight in the temperature-pressure profile can be gained by the following method. Consider the situation at base pressure (i.e. a few mTorr), a chuck temperature of 450°C, a wafer temperature of 350 °C,... 

Effect of sudden pressure increase on wafer temperature... 

Joshi et al.51, investigated the dependence of the wafer temperature on the pressure for a range up to 55 Torr (see figure 7.4). Note that at pressures above ca. 20 Torr, the difference between the wafer temperature and the hot plate is about 10 degrees. What parameters determine this temperature difference at high (i.e. >20 Torr) pressure To answer this... 

Figure 7.4. Wafer temperature as a function of hydrogen pressure and five hot plate temperatures. [From Joshi et al.51, reprinted with permission].

Since, as shown above in table 7.2, Er(chuck-wafer) is small compared to Ec(chuck-wafer) at higher pressures we neglect this contribution in our solution of the equation. After substituting equations 7.1 and 7.2 and solving for the wafer temperature (Tw) we find ... 

---