Original Publication Date: 1999-Apr-01
Included in the Prior Art Database: 2003-Jul-21
Thomas Morgenstern: AUTHOR [+2]
The planarisation of the product is normally done either by CMP or in DPS chambers, due to their higher removal or etch rates compared with MxP Centura chambers. Due to the numerous planarisation steps during the production it is the main topic to have several possibilities for the removal of polysilicon before starting the recess processes. The uniformity of the resulting product have to be better than 3%, before starting the recess. Unfortunately MxP Centura chambers exhibit only 8% of uniformity. The reason seems to be the negative influence of the extended cathode within these chambers compared to the oxide chambers.
- 25 -
Idee: Thomas Morgenstern, Dresden; Michael Schmidt, Ottendorf-Okrilla
The planarisation of the product is normally done either by CMP or in DPS chambers, dueto their higher removal or etch rates compared with MxP Centura chambers. Due to thenumerous planarisation steps during the production it is the main topic to have severalpossibilities for the removal of polysilicon before starting the recess processes.
The uniformity of the resulting product have to be better than 3%, before starting therecess. Unfortunately MxP Centura chambers exhibit only 8% of uniformity. The reasonseems to be the negative influence of the extended cathode within these chambers comparedto the oxide chambers.
During the presented work a new OES endpoint algorithm for the process is developed andtests on product wafers were carried out. During the development the DC-bias has to belower than 800V to omit arcing during the process. The parameters applied for first testswere taken from a matrix determined during optimisation of the IT-etch on AMAT P-5000MxP. In addition the influence of the magnetic field was determined. After the planarisationthe recess processes were realised in MxP chambers in dependence on the applied product.
The experiments have been performed on blanket polysilicon wafers and on product wafers.
Polysilicon: 500 nm undoped poly on 13 nm oxide (KP 016)
The values determined are:
Etch rate polysilicon (ER_poly)
Uniformity polysilicon (Un_poly)
The etch time for the polysilicon was t = 60s. In addition the influence of a magnetic fieldwas checked to omit arcing during the process step. The values therefore were varied in-between B = 0 – 40 G and a decrease of the DC-bias is measured with increasing magneticfield. Both the resulting etch rates and the uniformity were determined by a 9-point-interferometer-measurement. The uniformity have been calculated using the (Max-Min)-formula:
Unif = (Max-Min) / 2 x mean
Siemens Technik Report
Jahrgang 2� Nr. 3� April 1999
- 26 -
All other values were taken from a DOE measured during the development of a new IT etchon the same tool (AMAT P-5000 MxP). The� values taken for first planarisationexperiments are summarised in table 1.
Parameter Power Pressure B-Field N 2� CHF 3� NF 3� Temp. ER_Poly Un_Poly
Value 800 W 60 mT 0 G 53 sccm 5 sccm 18 sccm 20 °C 175 nm /min
Table 1� Parameters used for the� determination of the influence of the b-field on etch-rate and uniformity
After the experiments on blanket wafers the results were tested on product wafer to checkthe recess 1 and on another product wafer for the recess 2 process. Characterisation of etchdepth was measured by SEM and additionally for recess 2 with AFM.
Development of the endpoint algorithm
For the development of the end point algorithm a product wafer was applied and planarisedfor 300 s to get a survey of the shape. An increase of the curve exhibit the increase offluorine radicals in t...