Effects of Wafer Cleaning Reduction on Metals Removal and Ultrathin Gate Oxide Quality
- PDF / 481,033 Bytes
- 7 Pages / 414.72 x 648 pts Page_size
- 9 Downloads / 217 Views
cleans. Initial work focused on lowering the NH OH concentration in SCI to minimize surface roughness. [2] Later. reduced concentrations of both NH4OH and H20 2 were modeled [3] and evaluated [4]. The role of peroxide in SC2 has been studied as well as the cleaning efficiency of dilute HCIIH 20 mixtures in tanks and spray processors.[5,6] There has been much success in showing that more dilute SCI and SC2 cleans provide the same level of surface cleanliness as the standard 1:1:5 dilutions. In this study, the effects of temperature reduction on the SCI clean and dilution of the SCI and SC2 cleans are studied on surface metal removal and resulting gate oxide quality. Metal levels on silicon wafers cleaned by these experimental reduced cleans are monitored by TXRF. The levels of Ca, Fe, Cu, and Zn are compared among these cleans for removal capability. Aluminum gate and Poly-gate capacitors are fabricated using these reduced cleans before gate oxidation. Breakdown voltages and defect densities of these capacitors are then compared as a means to evaluate these reduced cleans. The goal of this work is to show that the standard SCI and SC2 concentrations can be reduced as well as the SCI temperature and still maintain good metals performance and yield high electrical quality ultrathin gate oxides. EXPERIMENTAL
All experiments were performed on 125 umm diameter, p-type silicon wafers. Wafer c!eaning was
performed in an Interlab Microrinse station consisting of 3 chemical and 4 vessels flowing 18 meg-ohm resistivity water. For all of the cleaning and oxidation experiments, wafers were dipped for 2 minutes in 100:1 dilute HF at 231C which was filtered and recirculated, Wafers were then diprped for 5 min. in NH 4OH: H,O2 :H20 (SCI) followed by 5 min. in HCI:H O0:H 20 (SC2). Both SCI and SC2 quartz tanks
are equipped with ultrasonics and are temperature controlled by externally heated water. The wafers were
rinsed for 5 min. in cascading 18 Meg-ohm dionized water following each chemical treatment and dried in a Semitool Rinser/Dryer. For the TRXRF experiments, wafers were p-type, 0.3-0.7 ohm-cm resistivity. All wafers for these experiments were chosen from the same lot and metal levels were checked by TRXRF on these "out of 233 Mat. Res. Soc. Symp. Proc. Vol. 477 1997 Materials; Research Society
the box" wafers on several instances throughout the experiments. These wafers were cleaned, put into single wafer carriers for transport, and measured by TXRF. For each metal detected, 13 points per wafer were measured, sampling approximately 27% of the total wafer surface. To account for errors arising from particulate contamination due to handling, data points outside of 2 standard deviations of the mean were removed from the data set and a new mean was calculated. The TXRF beam spot size was 18 mm diameter and the angle of incidence was 0.09 *,allowing for a detection depth of 10-30 nm into the silicon. The mean detection limit for Fe, Cu and Zn was 5E9 atoms/cm 2, while that for Ca was lElO 2 atoms/cm . Aluminum gate capacit
Data Loading...
