Pad temperature

Many endpoint detection systems, based on mechanisms, such as those based on reflected optical light [9], spindle motor current [10], pad temperature [11,12], have been used to resolve this problem, with limited success. Some systems may work with blank wafers or wafers with relatively low pattern density (at the STI level, for example), but for the PMD or ILD levels no useful results have been reported. The presence of a pattern at the PMD or ILD levels adds a great deal of complexity to the signals. Currently, use of an endpoint detection system to control the final post-CMP thickness is still a fertile topic for research and development. 

Pad temperature influences not only pad physical properties but also the chemistry of the slurry on the pad, especially when reactive metal films are polished, since the metal slurries containing oxidizing and complexing agents are more sensitive to temperature change. Pad temperature increases because... 

Figure 3. Pad temperature and coefficient of Mction for cast iron as a function of test time. Test velocity was 6.0 mJs and pressure was 0.39 MPa.

Pad Temperature Elevated pad temperatures result in improved removal rates because of expanded pad pores and better slurry transport and also good planarity and smooth surface morphology of the pad [40, 48]. However, it lengthens the time to achieve steady state, and also plays a major role in generation of wafer defects like dishing and erosion during the copper CMP process [49]. 

Next, in Figure 15.12, we show removal rate and defect results from a tungsten process run on an F-REX 300 Ebara polisher. The process uses an IC pad and W2000 slurry with a normal flow rate of 250 mL/min. For this process, the PA removal rate declines steeply as the flow rate decreases while the SIS removal rate is nearly constant. The SIS defect count is also about half of the PA count and does not change when the flow rate is halved (Figure 15.12). The nonuniformity and carrier trailing edge peak pad temperature (not shown) were also measured and are the same for both apphcation methods. 

The thermal cycle may be considered as beginning at the entrance into the lower pad with lubricant at an intermediate temperature between that of the feed lubricant and that of the fluid exiting from the top pad. Temperature rises through the lower pad. Temperature drops in the upper pad feed groove after miking with feed fluid, and rises again through the upper pad. 

This would give an advantage since turbulent mixing will reduce the temperature of oil in the entry steamtube, even if some of this is in the laminar sub layer. If this is the case, the sudden reduction of pad temperature with increasing high speed (as observed in several laboratory test machines) could be due in part to turbulence and cooling outside the film, rather than viscosity enhancement from super-laminar effects within the film. 

A 3D solution for the film and pad temperature using the control cell technique, with upwinding of the film flows. 

Fig. 3 Calculated maximum. pad temperature vs. circumferential pivot position.

Calculated maximum pad temperature (62% pivot) vs. pad thickness/pad radial length, t/b. 

The calculated data in Figures 2, 3 and 4 indicate improved performance for pivots beyond 60% for all three items film thickness, temperature, and pressure. In the tests reported here, pad temperature was the only item of these three performance characteristics that was measured, and it is thus used as the basis for conclusions regarding performance. For babbitted bearings, a combination of local temperatures and pressures may also be of value in judging bearing capabilities due to the strength-temperature relationship of babbitt. 

The temperature data used in all of the plots presented here is the average value of the two readings from corresponding locations on the two pads. The term "maximum pad temperature" is used, and it is, of course, the maximum recorded pad temperature. The thermocouple locations were chosen to cover the area where the highest pad temperatures have normally been found, both by analysis and test. However, only a limited number of distinct locations on the pad face can be sensed. In the majority of cases, the highest temperature was recorded at 85-85. However, depending on load, support disk diameter, and pivot location, the maximum temperature moved to... 

Fig. 8 Measured maximum pad temperature vs. pivot position, 2000 RPM, 4.14 MPa loading, d 38.1 mm.

Figure 8 shows the effect of pivot position on maximum pad temperature for the three pad thicknesses at a specific operating condition. This plot covers the full range of pivot posi-... 

Figure 9 shows the effect of pivot position on maximum pad temperature for three support disk diameters (d) at a specific operating condition. Here, also, the effect of pivot position is clear, but the same is not true for support disk diameter. This plot is data from tests on the thickest pad, however, where support disk diameter would be expected to have least influence. 

In Figures 11 and 12 (the thin pad), a small diameter support disk provides reduced pad temperatures for center pivot pads, but at the 75% pivot location, a large diameter support disk is better. 

As the pad thickness increases, the 75% pivot location remains optimum (at least as determined by pad temperature), but the optimum support disk diameter decreases. 

Various methods were proposed to control thermal distortion. Basically, all the methods can be divided into three groups. The first group deals with the hot oil carryover. Pad temperature and consequently thermal distortion are reduced if greater amount of cold oil is introduced into the film. This is achieved by spray jets [20-22] or scrapers [23]. Both arrangements are intended to scour off the hot oil carryover that adheres to the rotating collar. Scrapers are more efficient. As shown in [23] pad temperatures can be decreased by 20°C and more but it enacts a penalty in a form of additional power loss. 

The most efficient system is directed lubrication introduced in the end of 1960 s [75]. The oil was sprayed at high speed on to the collar surfoce between the pads of die freely drained bearing. The power and pad temperatures were considerably reduced compared to die flooded design. Thicker oil films were also observed [70]. Various distribution systems for the spray jets were tested. Umbrella nozzle system was studied in [70] while the spray bars were tested in [71]. 

The most commonly utilised and versatile lubricants are mineral oils. Viscosity is the main oil characteristic. The effect of oil viscosity grade on bearing operating limits was considered in [7]. From the point of view of minimal power loss and pad temperature thinner oils are recommended at high speeds. Thicker ones are usually used at low speeds to avoid too thin oil films. 

There are three basic operating characteristics that need to be considered in the bearing design and limited in order to insure satisfactory operation. These are maximum pad temperature, minimum film thickness, and peak oil film pressure [7]. The last two of these are difficult to measure due to the lack of simple and robust sensors and problems with their mounting. Test data available on film thickness and pressure is limited. Film pressures were... 

---