|M.Sc Student||Halevy Ran|
|Subject||A Method for Measuring the Specific Interface Resistivity|
between Two Semiconductor Layers and its
Application to a Heavily Doped N-Type
|Department||Department of Electrical Engineering||Supervisor||Professor Dan Ritter|
|Full Thesis text|
The emitter contact resistivity is one of the most dominant factors holding back the cutoff frequencies for high speed hetero-junction bipolar transistors (HBTs), and therefore accurate characterization and improvement of this parameter become of great importance.
RF characterizations of devices in current use exhibit a measured resistivity higher than what is usually attained in standard Transfer Length Method measurements (TLM), which are commonly used to characterize contacts. This work comes to address this issue by more closely examining and improving upon the existing contact characterization methods for single layer and dual layer contacts. It specifically aims to examine dual layer hetero-structure based contacts, which are of frequent use, and determine the actual role of the inter-layer interface resistivity in the overall effective contact resistivity.
Previous works regarding the single layer TLM characterization method include a first order analytic error analysis, which is brought here as reference. As for the dual layer structure, a simplistic interface resistivity estimation method, which was originally presented for p-type contacts, is brought and evaluated, and also an analytic model of the full two layer TLM structure is brought as reference.
In order to correctly and gradually address the two layer structure in this study, I first examined the single layer TLM method on all its aspects, and specifically the error analysis it entails. In order to also attain a numeric form of error analysis, which would later benefit the dual layer part of the study, I devised a custom designed numeric Monte Carlo simulation, based on measured and estimated error factors, as done in the single layer analytic error analysis work that is brought as reference. In order to confirm the validity of the numeric error analysis, simulations were compared to results of the analytic error analysis. However, since the analysis performed in previous work was of a first order approximation, a second order correction had to be made within this study for greater accuracy, and indeed agreement numeric simulation results was observed.
Focusing on the dual layer structure, the existing method previously mentioned is examined and shown to be inadequate for our needs. At that juncture a new analytic model is presented, one of a two layer structure in which the upper layer has been etched away between the contacts. Combining the two analytic models, an innovative
numeric extraction method was formed, giving the ability to extract the four structure parameters. The numeric simulation algorithm was then fitted for the dual layer structure, and simulations were performed. Measurements results for all structures are also presented and compared to simulations.
Finally, the average specific interface resistivity attained from these measurements is found to be 1.6 ± 0.38 [Ω • µm2], and the specific contact resistivity is 9.8 ± 0.51 [Ω • µm2]. It is concluded that such values of interface resistivity, along with the error margins of both resistivity values, may indeed account for most of the discrepancies in effective resistivity estimations.