Trench Width

Institute of Technology (Reese et al., 2003). Using optical lithography to etch away photolithographic resists on stainless steel foils, french pens were designed with a rectangular geometry of 6 p in depfh, 30 p in width with 30 p sidewalls at the tips. To add structural support, the features anterior to the tip were expanded out so that the width of the sidewalls increased to 120 p with a trench width set at 90 p. 

Figure 9 PhotocuiTent from LED-photodiode integrated device (active-layer thickness is —500 nm. trench width is —20 p-m) for three cycles of N2 SO2 Nj at 1 atm each. A reproducible and stable enhancement of —30% was observed with respect to the N2 ambient. (From Ref. 14.)...

Infrared radiation is incident on and passes through a substrate 12. The radiation is absorbed within an n-type absorbing layer 14a. The n-type layer is overlied by a p-type layer. The n-type layer and the p-type layer are divided into mesa structures 16, which have sub-mesa structures 16a and 16b each containing a portion of the p-type layer as a p-type cap layer 14b. Trenches 30 are etched or milled to a depth that extends completely through the p-type cap layer and partially into the n-type layer. The trench width is approximately 10 microns. Trench walls 32... 

Finally, in the limit of very narrow trench widths, the trench bottom will see contact only from asperities that are both high enough and narrow enough to fit into the trench. All other asperities will simply span the gap between the sides of the feature. This is what might be termed the asperity filtering regime. 

For example, if there is no constraint on the allowable width at x (wmax = oo), then from (6.49) the proportion of asperities that may contact is 1. If x is at the bottom of a trench of width W, then Wmax = W and the proportion increases with the trench width. When this modification is incorporated into the Greenwood and Williamson model, the local contact pressure becomes... 

After the pattern is generated and the trench depth is measured by profilometry, Si02 is deposited and subsequently planarized by CMP. After CMP, the surface is again measured by profilometry and the amplitude of the square wave subsequent to polish is compared to die amplitude prior to oxide deposition. Figure 5.26 shows the ratio of the post-polish CMP amplitude, to the predeposition amplitude. A , vs. oxide removed for several trench widths. Note that the plot of log (A/A ) vs. oxide removed is a straight line. Renteln et al. use this fact to define the metric planarization rate, P, which is equal to the slope of the lines in Figure 5.26. 

Square wave patterns with trench depths of 1.0 pm are etched into silicon substrates to determine the planarization rate, P, of two oxide CMP processes. After patterning, 2.0 pm of Si02 is deposited onto the substrates. The wafers are then polished to remove oxide in 0.2 pm increments. The surface amplitude is measured after each polish and the resultant data tabulated in the table below. Plot log(A Aj) vs. oxide removed and determine P for each process and each trench width. Which process is more effective at planarizing ... 

Profilometry was performed on a wafer which was polished in 40 second time increments under baseline conditions using an experimental grade copper slurry at 4 psi on a perforated IClOOO/Suba rv pad on an IPEC 372MU. Figure 3 displays the extracted step-height-reduction as a function of trench width. Several important conclusions are immediately evident. 

Figure 3 Trench Depth Reduction vs. Trench Width for equal time polish increments...

Figure 5 Metal Thickness Difference vs. Trench width for different pad-process combinations Figure 6 Metal Thickness Difference vs. Trench Width for two different slurries...

Figure 2 shows a 21.5mm-long profile extending over four test structure patterns within a single die. Each test pattern has a different combination of trench, width and space. All have the same line/space ratio of land the line/space ratio increases in successive test structures from left to right in both profiler scans. The die-level dishing was 0.5p.m. 

Figure 4. Graph of dishing versus trench width for measured Cu sample with a line/space ratio of 1 1 (A) and 1 2 (B)...

For conventional sizes (1 = 1 pm, 1 = 0.5 qm, d = 2 qm), the required anisotropy coefficient should be at least 10. Small trench width can obviously be achieved by using a... 

Pattern Density [%] (trench width/pitch size)... 

In the case of device applications, the filling material is some form of dielectric material, such as oxides, polymers, or polycrystalline silicon. (Metal filled trenches could serve as buried interconnecting device lines). As suggested previously, generally two types of trenches may have to be filled up - narrow ones for device isolation, and wider ones to benefit other circuit functions. The process requirements are somewhat different for each. However, independent of the trench width, the initial processing is the same. 

Isolation width becomes twice that of the starting trench width. 

The above observations apply to trenches with relatively wide dimensions. When the aspect ratio between the trench width to depth becomes smaller, the deposition can possibly result in physical overgrowth. This will leave a narrow void in the center of the filled trench, as is shown in several examples in Figure 14. The effect occurs more readily for trenches with vertical walls (Figure 13), and with certainty for trenches with convexed sidewalls. 

The planarity of the surface above a narrow CVD-filled trench may be quite good, provided a sufficient amount of CVD material has been deposited. The specific planarity will depend directly on the trench width and the CVD film thickness. This relationship is shown in Figure 20. Surface planarity will not be achieved for wider trenches. Such major surface irregularities cannot be tolerated as a final surface structure, and a post-CVD surface planarization technique must be employed to assure a sufficiently planar surface prior to a subsequent back-etching process step. 

Figure 20 Relationship between surface dip and trench width for different thicknesses.

The aspect ratio of the microtrenches is defined as the ratio of the trench depth to the trench width. DRIE can easily reach an aspect ratio of 30. With special recipe and high plasma power, the highest aspect ratio can reach as high as 100 [7]. The smallest feature size is dependent of the patterning capability. Using the normal optical lithogra-... 

---