Planarization efficiency

Viscoelasticity (asperity) Asperity radius Planarization efficiency... 

The CMP output parameters include removal rate, planarization efficiency, surface finish, material removal rate selectivity, wafer-to-wafer uniformity, within-wafer uniformity, dishing and erosion, and defect levels [15]. 

Figure 3.4 shows subsystems and related materials in a CMP system. In addition to a polisher and post-CMP cleaning station, a CMP system encompasses slurry supply, waste treatment, monitors, slurry, and pad. The performance of a CMP system is measured by its output in (a) wafer uniformity, (b) polishing rate, (c) planarization efficiency within chip, and (d) defect count. From a mechanistic point of view, the tool uptime, throughput, and reliability of the system are also very important [16]. In order to satisfy these requirements, all subsystems described in Fig. 3.4 are considered as one total system of CMP and should be upgraded as a whole whenever needed. 

Nguyen VH, Daamen R, Hoofman R. Impact of different slurry and polishing pad choices on the planarization efficiency of a copper CMP process. Microelectron Eng 2004 76 95-99. 

Owing to the fact that free ferric ions are stable only in the acidic regime (Fig. 7.8), most slurries using ferric ions as an oxidizer are formulated at pH substantially below 4. At such a low pH, the copper surface oxidized via the reaction described in Equation 7.7 will not form any native oxide film. Without the protection from such an oxide, the copper surface is prone to corrode, which results in high static etch rate and practically no planarization efficiency. To provide a balance, therefore, the presence of a corrosion inhibitor is a must for copper CMP slurry. 

The formation of an anodic film (ion-concentrated diffusion layer or salt film) depends on electrolyte and processing conditions. The characteristics of the film are critical to the planarization efficiency of ECP and ECMP. Hence, the analysis, especially in situ analysis of anodic films, is highly valuable. A few techniques can be used for the purpose. 

FIGURE 10.12 Influence of electrolyte on ECMP planarization efficiency. 

To achieve better planarization efficiency than ECP, different modified ECP approaches have been explored. These approaches, including ECP-DI water technique [27], sectorial cathode ECP [28], membrane-mediated ECP [29], and contact ECP (i.e., ECMP) [30,31], work by different mechanisms and exhibit different performances. 

These conditions mainly depend on the electrolyte. However, the thickness and surface profile of the anodic film, which strongly affect planarization efficiency, also depend on the ECMP pad. 

Conventional CMP typically requires an overburden of copper of at least 4000 A due to low planarization efficiency. An overburden of 1000 A is sufficient for ECMP to achieve planarization without compromising performance in terms of dishing, erosion, and defectivity. A stress migration test was carried out at 250 °C for 333 h on a structure with metal level 1 and level 2 lines with thickness in the range of 0.14-0.16 pm. An increase in resistance by more than 10% is considered a fail. The reduction in electroplated copper thickness can reduce the number of fails in stress migration test as shown in Table 11.1. 

FIGURE 11.17 ECMP planarization efficiency at removal rate of 6000A/min on 100 X 100 pm lines (from Ref 25 ). 

Unlike in the ECMP process described earlier, the initial conductivity of the liquid between the wafer and the pad is not required. As a matter of fact, a conductive solution actually lowers the planarization efficiency as shown in Fig. 11.24. Similar to ECMP, high planarization efficiency can be obtained at... 

FIGURE 11.25 Correlation between planarization efficiency and downforce used during an ECP-DI polishing (from Ref. 32). 

The key requirements for an acceptable STI slurry include adequate removal rate (> 3000 A/min) on oxide, desirable selectivity of oxide over nitride (between 200 1 and 400 1), high planarization efficiency across the die regardless of the patterned density, low scratch count, and low particle residue. The typical operating downpressure for an STI process is 3-5 psi. The typical platen and carrier speeds of a rotary tool are 100 and 60 rpm, respectively. The... 

If the applied shear to the wafer is less than the threshold value, this mechanism can be used effectively to enhance the planarization efficiency. The localized pressure on the topography will be effectively larger than the threshold, which will selectively remove material at a high rate from the... 

The planarization efficiency is defined as the unitless formula one minus the ratio of removal rate for the bottom over the removal rate for the top of the structure. It is actually difficult to achieve planarization efficiency over 99% for all structure types, all feature sizes, all density variations, and all step heights at the same time. The easiest features to planarize are small individual structures (submicrometers or a few micrometers in size). Over such a structure it is unlikely to see any color variation in the oxide around the structure. A large array of structures adjacent to another large array of features or large field area (size in millimeters) with different relative average height of oxide create an oxide step that is difficult to planarize. Before CMP, there is no color variation within each individual array. After CMP, it is likely to see color variation at the edge of each array or in the adjacent field area. 

First, no planarization is achieved for feature sizes above 4.5 mm. For feature sizes 1.0 mm to 4.5 mm, only partial planarization is possible, and with the exception of the 3, 3.5 and 4 mm structures, its effectiveness increases monotonically with feature size reduction. And finally, for feature sizes 1.0 mm and below, nearly constant planarization efficiency was achieved, as the trench depth was reduced from 0.8 pm to zero. We believe that the irregularity in our experimental data for 3-4 mm feature sizes was caused by an error in the exposure. This caused each of these structures on wafers from this particular lot to be actually composed of two smaller trenches with slight separation, as evidenced from visual inspection of the pre-polished wafers. 

Using a new planarization monitor, the planarization efficiency and planarization distance were computed for a copper CMP process. The influence of polish pad, slurry and process conditions was demonstrated. Results indicate that harder, stiffer pads can extend planarization performance to wider structures given a suitable choice of slurry. Process conditions play a relatively minor role. The planarization of copper during the as-plated copper overburden removal was found to exhibit similar behavior to the planarization of oxide topography in oxide CMP, with a planarization distance of approximately 2mm for a typical stiff-pad copper CMP process. 

The scenario above is certainly dependent upon the compressibility of pad and the down force used, as discussed by Grillaert et al For a perfectly incompressible pad, isolated patterns would be planarized more rapidly. For a perfectly compressible pad, the reverse is true if the side polish rate is insignificant. Also worth noting is the effect of table speed, which is not considered in the present study. A high relative velocity has been found effective in improving the planarization efficiency. This may result from the increasing shear rate and hence side polish rate. This, in conjunction with a lower down force, would improve the planarization efficiency. This point will be elaborated later. 

---