Browse Prior Art Database

A SPECIAL CASE OF USING ELLIPSOMETRY TO MEASURE THE THICKNESS OF OXIDE ON POLYSILICON

IP.com Disclosure Number: IPCOM000005729D
Original Publication Date: 1988-Oct-01
Included in the Prior Art Database: 2001-Oct-31

Publishing Venue

Motorola

Related People

Authors:
Barbara Vasquez Harland G. Tompkins

Abstract

A convenient, nondestructive method forthe measurement of the thickness of oxide on poly silicon (poly- ox) is not commonly available in semiconductor manufacturing environment. Ellipsometry is used routinely to measure the thickness of oxide grown on single crystal silicon, but not of poly-ox. In order to determine the thicknesses of poly-ox with a single measurement from ellipsometry of a three film test structure (oxide/ polylpolyoxide), it is necessary to know the thickness of both the underlying oxide and the poly, and to then use a multiple film program to determine the thickness of the oxide. Even slight uncertainties in the thickness of either film beneath the poly-ox will make single measurement determinations impossible, because the film- free value of Del and Psi could be anywhere on the plot. In addition, the act of oxidizing the poly will change the thickness of the poly. It is not surprising that E. A. Irene' would comment that "the accurate measurement of oxide grown on polycrystalline silicon surfaces is not practical': There is a special case where ellipsometry can be used to estimate polyoxide thickness which requires a test structure (Fig. 1) where the oxide layer which separates the single crystal silicon from the poly is thin, i.e. 1OOA or less. Under these conditions, a simplifying assumption can be made in the calculations. Figure 2 shows the calculated Del/Psi trajectories for increasing poly thickness deposited over 200A, lOOA, and 50A oxide. As the poly thickness changes, the Del/Psi point moves around the slowly closing spiral. The size of the slowly closing spiral is directly affected by the thickness of the underlying oxide with thinner oxides giving tighter curves. For a poly layer of unknown thickness in the range of 3000A to 5000A, the starting point for a Del/Psi curve for oxide deposited on top of the poly will fall somewhere on the slowly closing spiral. Given this uncertainty in the starting value for Del/Psi, estimates of the poly-ox thickness can be obtained using a single film approximation with an accuracy of the order of the thickness of the oxide beneath the policy.