Oxide erosion

Local oxide erosion is associated with dishing in damascene process, as discussed in the following. 

---

Ablative mastics absorb heat as they lose mass through oxidative erosion... 

Fig. 26. Degree of oxide erosion as a function of polishing time in tungsten CMP [12]. Reproduced by permission of the Electrochemical Society, Inc.

The H202-based slurry performed the best in this evaluation. However, it has problems with plug recess and field oxide erosion. 

Plug recess and oxide erosion between the H2O2 (silica as abrasives) and the HjIOg (mixture of different abrasives) based slurries are also evaluated on patterned wafers. The same polisher (rotating), recipe, and pad (ICIOOO/... 

The plug size is 0.5 gm in diameter, and the oxide erosion measurement is... 

High hardness top pad IC1400 Oxide erosion 220 A Defect density 700 ... 

Medium hardness top pad UR II Oxide erosion 400 A Defect density 180 ... 

While some of the WEB drawbacks can be eliminated by W CMP, W CMP itself generates new problems. Because W CMP uses alumina (AI2O3, one of the hardest materials known) abrasive in a Fe(N03)3-based slurry to polish, defectivity (scratches) and Fe contamination become issues. For the former, oxide buff can be used to reduce defects. However, this may lead to an oxide erosion problem, as discussed in a later section. 

Fig. 25. Probe pad structure (a) without and (b) with oxide studs in the two-dimensional wells. The latter can be used to reduce dishing. However, local oxide erosion may be still a problem.

Dishing can be also reduced by adding oxide studs in the two-dimensional wells (see Fig. 25). Since the two-dimensional well structures are usually probe pads for electrical measurements, the addition of oxide studs will not affect the pad function as long as the test probe can still touch metal materials. Although this method can reduce dishing, local oxide erosion may still exist. The mechanism of local oxide erosion due to dishing is still unclear at this point of time [26]. 

A.A. Haasz, S. Chiu, J.E. Pierre, Y.I. Gudimenko, Thermo-oxidative erosion of amorphous hydrogenated carbon films, J. Vac. Sci. Technol. A 14 (1996) 184-191... 

FIGURE 17.13 W deposition with no overfill (left). W deposition with overfill does not change anything to the void formation. At just the metal clearing of stage W CMP, the void is not opened (center). W overpolishing will result in oxide erosion and oxide loss. This will tend to open the void and make it visible (right). 

CHARACTERIZATION OF SLURRY SYSTEM AND SUPPRESSION OF OXIDE EROSION IN ALUMINUM CMP (CHEMICAL-MECHANICAL PLANARIZATION)... 

Pero.xide is suggested by many vendors as the oxidizer for metal polish and recommended as one component of their slurry system. It is still unclear, however, wliat roles the peroxide plays in the polish process. The original purpose of this work is to evaluate the impact of this variable upon the process performance. The experiment result turns out to be inspiring and suggests a new approach to minimization of the oxide erosion. 

Fig.3. oxide erosion versus feature size obtained from a preliminaiy experiment. The data collected on the same wafer but different dies is represented by the same symbols linked with a line. The erosion obviously increases with the over-polish lime despite of the data scatter. 

Fig.5. 40x10 pm AFM images of a trench array (S/L=l/0.7) after polished witli llte traditional process (a) and the process proposed (b), respectively. Although the total polish lime is the same in both cases, the metal dishing is almost doubled in (b) since the removal rate is much higher due to the use of the slurry in Zone I. Nevertheless, the oxide erosion is much smaller. 

Figure 14. Dishing and oxide erosion rate parameter dependencies. Shown at left is a conjectured dependency between the maximum dishing d, and the metal line width. Shown at right is a schematic acceleration of the oxide removal rate for small oxide space widths.

Figure 1 Trench oxide erosion measured on SIT monitor (100 im x lOOum) for Low Selectivity Slurry (LSS). Its mean and standard deviation were obt ed from S measurement sites, corresponding to each CMP process (DOE).

Chapter 11 presents and discusses porous SiC use in catalysis for which the large porous film surface areas are again clearly useful. SiC s high chemical bond strength allows its use at high temperature and renders it resistant to oxidative erosion. 

---