The Electrical Measurement of Pattern Placement Errors of Mask Makers

Abstract

The misregistration between patterns in integrated circuits can be due in part to the masks. Mask errors can be simulated by writing different patterns on the same mask. The patterns can be arranged to form thin film resistors in a bright chrome mask and the resistor value can be used to calculate misregistrations. Normally the two patterns have to be aligned to get useful results. This paper, however, describes patterns that do not require alignment and this extends the use of the measurement technique to the assessment of the long term mask making errors and errors in copying. It is shown that the technique can be accurate to better than -+0.05 ~m and an evaluation of runout effects in mask copying is described.

An NMOS test chip for a course in semiconductor parameter measurements

“…A third misregistration test structure designed to measure misregistration between the same two photomask levels as the previous two structures is the Stickman-type structure seen in figure 6c, The structure is a modification of that of Nicholas et at. [15,18] in which the current taps and voltage taps are separated in order to provide a four-terminal or Kelvin measurement of resistance. For this design, a long blocking bar on a second photomask is aligned with a larger rectangle on the first photomask level to form two parallel and narrow resistors of equal size for no misregistration.…”