AVS 53rd International Symposium
    Electronic Materials and Processing Thursday Sessions
       Session EM+AS-ThM

Paper EM+AS-ThM2
Channel Drift Mobility in High-k Transistors from Galvanomagnetic Measurements

Thursday, November 16, 2006, 8:20 am, Room 2003

Session: High-k Dielectric Characterization
Presenter: R.T. Bate, University of Texas at Arlington
Authors: R.T. Bate, University of Texas at Arlington
W.P. Kirk, University of Texas at Arlington
R. Agrawal, University of Texas at Arlington
R.M. Wallace, University of Texas at Dallas
B.E. Gnade, University of Texas at Dallas
G. Pant, University of Texas at Dallas
Correspondent: Click to Email
Effective mobility µ@sub eff@ in high-k transistors can be degraded by enhanced scattering of carriers and by trapping. The drain current is proportional to the product of the channel carrier density n and the drift mobility µ, and the capacitance measurement used to determine channel carrier density cannot distinguish between free and trapped charge. The result is that trapping can cause µ@sub eff@ to underestimate µ. We combine Hall effect and magnetoresistance (MR) measurements to determine µ of carriers in the channel. Then effective mobility reductions due to increased scattering can be separated from loss of carriers due to trapping. In the past, Hall effect measurements have been used to measure Hall mobility in MOSFET and MISFET channels, and µ has been estimated by making assumptions about the magnitude of the Hall factor r@sub H@ which is the dimensionless ratio of the Hall mobility to the drift mobility, (i.e. r@sub H@ = µ@sub H@/µ). r@sub H@ is usually assumed to be close to unity. Theory indicates that r@sub H@ is strongly dependent on the type of carrier scattering in the channel, which is usually not known a priori. This uncertainty is currently the main drawback of the Hall effect as a means of determining channel drift mobility. We propose to use the MR to estimate r@sub H@ . Analysis based on the silicon 100 inversion layer shows that the MR is strongly correlated with r@sub H@, regardless of the scattering mechanism. This is being experimentally verified on MOSFET structures where r@sub H@ and µ can be independently determined, and the results are being carried over to high-k MISFETs. Knowing r@sub H@ permits µ to be estimated, not only from Hall measurements, but also, for routine diagnosis, from MR measurements on conventional transistors (not Hall devices), even in the presence of significant trapping. @FootnoteText@ Partial support by Texas Advanced Technology Program of THECB #003656-0029-2003.