STATISTICAL MODELING OF TRANSMISSION-LINE MODEL TEST STRUCTURES .1. THE EFFECT OF INHOMOGENEITIES ON THE EXTRACTED CONTACT PARAMETERS

Extraction of electrical contact parameters by transmission line model (TLM) test structures is based on the assumption that the electrical and geometrical parameters of the contacts within a structure are identical. Contact parameters measured over a wafer, however, typically display large scatter. This fact suggests that the individual contacts within a TLM structure may not be perfectly identical either (nonideal TLM structure). Although several procedures to correct for two- and three-dimensional parasitic effects have been published in the literature, there is no quantitative assessment of the errors introduced by the nonideality of a TLM structure. Statistical modeling presented in this paper shows that the usual extraction procedure can lead to large errors in the extracted contact parameters in the case of nonideal TLM structures even if there is no error in the measured electrical and geometrical parameters. Errors in the extracted parameters as functions of the nominal contact parameters are presented graphically. These graphs can be used to assess the accuracy of the TLM measurements and to optimize TLM test structure design. In the parameter intervals examined in this work, the TLM method works best (yields the least errors in all extracted parameters) when the geometrical length of a contact is equal to about two times the electrical transfer length. © 1990 IEEE