3.3 Methodology used:
3.3.1 Experimental preparation
HARP SA-USG was deposited by using Applied Materials producer chamber (fig-3.2), equipped with the precision liquid injection system (PLIS) for TEOS delivery and the high temperature ceramic heater. Remote microwave technology was used for cleaning to reduce metal contamination and to improve cleaning efficiency. All film properties were evaluated on 300mm p-type Silicon (100) substrates. The film thickness and uniformity were monitored by KLA Fx-100
KLA Fx-100 is very popular methodology tool for film property monitoring in semiconductor manufacturing, it is accurately in stability and repeatability for widely film applications include film thickness, reflected index and stress.
Figure 3.9 Contour map of HARP film by using KLA Fx-100.
- 81 -
The wet etch rate ratio (WERR) which is a good indicator for film integrity was performed in a BOE 300:1(300 NH4F/ 1 HF/ 540 H2O) referring thermally grown oxide film. We do not recommend dry etch rate (low process pressure, high RF>1000W with C2H2 gas) as a film monitoring because its faster plasma etch rate (>300nm/ min) results in unstable repeatability and reliability monitoring.
All SEM samples were decorated using 6:1 BOE for 10 seconds and a film coating, the pictures were taken on Hitachi 5200 SEM.
3.3.2 Experimental procedure
The effects of vary of temperature, pressure, spacing, and TEOS flowing on the deposition rate and wet etch rate ratio (WERR) have been compared for these two processes. The experiments were done with a nominal set points of 600Torr, 540 ℃, 1400mgm TEOS, 330 mils spacing, and 27slm ozone at 12.5 wt%. The carrier gas flows were optimized for best thickness uniformity with 8slm for Nitrogen.
Since both low shrinkage and WERR are the indicators of good gap filling, a set of experiment was created to explore and to optimize these film properties with higher deposition rate prior to gap fill testing, the detailed process experiments are outlined in Table-4.1.
Base on last experience, a well gap filling extendibility should composition of high process pressure, high O3/TEOS ratio and low deposition rate. Thus we will approach a high O3/TEOS ratio and a low deposition rate in
- 82 -
the initial deposition to achieve trench gap filling, throughput enhancement will be improved in bulk film by increasing of TEOS flow with low concentration of ozone. It is needed to explore basic film properties by the experiments. From the experiments, optimization of recipes was tested on production wafers to monitor gap filling performance.
Deposition rate improvement was clearly for all recipes. Increases in pressure, heater temperature, and Ozone concentration will prohibit its deposition rate, but it still need to provide significantly higher deposition rate to meet throughput. The results are shown in chapter 4.
3.3.3 Recipe optimization:
In order to retain the fastest deposit, ion rate while providing lowest WERR and shrinkage, a recipe with high O3 concentration was chosen along with an additional recipe using O2 to improve uniformity. Multiple steps recipe required to satisfy not only the void free of STI trench, but also does the throughput compatible as HDPCVD.
To achieve STI gap filling, we approach O3/TEOS ratio variation in deposition steps. We found that high O3/TEOS ratio during deposition is more favorable for gap filling in SA-BPSG. There were three-steps recipe was set up to meet throughput and gap filling capability. During the process, the initial step aims at having better nucleation of linear surface with very low TEOS flow rate and very high ozone concentration to achieve gap filling, TEOS ramps process that introduces very low amounts of TEOS in a Ozone rich environment to get a high quality initial oxide layer followed by a continuous
- 83 -
Table 3.1 Experimental design table.
but slow increase in TEOS concentration, so that the subsequent layers are deposited faster without adversely affecting the gap filling. Bulk film, which is the major portion of the film stack, could be deposited at a much higher rate. In a typically requested 700nm thick film, only the first 200nm
Recipe# Chamber
- 84 -
is needed for completely trench filling, the rest of 500nm does not require filling any narrow trenches and this is where the attention is focused in order to improve throughput.
Deposition rate could be increased by simply increasing the totally amount of TEOS used in the final step of recipe. Decrease in process spacing between wafer and gas distribution plate (show-head) to minimize the mean free path is another dominant for faster deposition rate. Changing of TEOS flow rate and heater spacing versus time is as shown in figures 3.10.
Figure 3.10 TEOS flow rate Vs time during 3-step HARP USG for gap filling.
Three steps recipe is required to achieve STI trench without throughput suffering. The initial step deposition approach with a high O3/TEOS ratio
- 85 -
DOE_1 168 1000 2700 5 2700 142
DOE_2 336 1000 2700 15 2700 42
DOE_3 68 1000 2000 5 2000 330
DOE_4 68 1000 1500 15 1500 540
DOE_5 240 300 2700 10 2700 140
ramp up
results in a more homogeneous nucleation layer onto trench surface, while the second-step is to fix trench gap filling with TEOS ramp process that which introduces very low amounts of TEOS in a rich Ozone environment.
Step monitoring is required to safe STI gap filling to prevent of some parameters drifting
Table 3.3 Split conditions for STI gap fill.
- 86 -
3.3.4 HARP USG anneal by furnace
The annealing conditions are very important for the gap-fill. As reference to SA-BPSG for pre-metal gap filling, steam annealing is much helpfully to avoid seams or voids in the trench. Thus, two steps annealing will be approached in our experiment, The priority steam annealing at low temperature favors in gap filling , and then followed by 1000℃ dry annealing to densify the whole film.
Table 3.4 Post HARP film annealing condition splits.
- 87 -