| Summary: | 博士 === 國立成功大學 === 微電子工程研究所碩博士班 === 97 === In this dissertation, an automatic RF test system, updated from a manual RF bench tester, has been developed with the advantages of accuracy and time saving. Compared with the commercially automatic RF test systems, the developed system has lower cost with the same performances. In addition, we developed some novel pad structures. These novel pad structures can be laid on the scribe line with the same performances and the advantages of area-saving and effective monitoring of devices in chips. With the automatic RF test system and novel pad structures, many silicon-based active and passive RF devices, including RF CMOS, inductors and transmission lines were investigated.
We firstly illustrate the basic concept about the RF measurements, including the scattering matrix, calibrations and de-embeddings in chapter 2. We then introduce the common RF bench tester comprising of a Cascade S-300 12-inch probe station, Agilent E8363B PNA, Agilent 4156C dc source, and the control program ICCAP. How to update this RF bench tester to the automatic RF test system is described in this chapter. In addition, the effects of pad structures on RF measurements are discussed in this chapter.
Then, a common issue on the BEOL process, i.e., the wire-edge enlargement effect (WEE), which is very serious on the lower metal layer of the deep submicron technology, is introduced. Next, we investigate the RF performances of the inductors and the transmission lines using the developed automatic RF test system. For inductors, the effects of dimension, dc bias and temperature are emphasized with a simple proposed model in detail. For transmission lines, their dimensions and the under shielding lines are also studied with a proposed model in detail.
Moreover, the hot carrier effects on dc and RF parameters of 45-nm and 55-nm NMOSFETs are investigated and compared in detail by the developed automatic RF test system. We find, as devices scale down to the deep submicron region, after hot carrier stress, -∆Cgd is less than +∆Cgs, and results in the fT being dependent on gm, Cgs and Cgd. These new observations are explained comprehensively by the developed surface channel resistance model. Next, the effects of hot carriers on the dc and RF performances of PMOSFETs with various oxide thicknesses prepared by 45 nm CMOS technology were investigated by automatic RF measurements. The devices with a thinner oxide layer suffer more serious damage from hot carriers than those with a thicker oxide layer. The greatest degradation due to hot-carrier injection is shifted from the condition of Vgstr = Vdstr / 2 for peak Ig and Isub in long-channel devices to Vgstr = Vdstr in short-channel devices. The fT degradation is only dependent on the transconductance gm, regardless of the total gate capacitance Cgg (=Cgs+Cgd). Furthermore, both hot carrier and Fowler-Nordheim (FN) tunneling stresses on RF performances of 40-nm PMOSFETs with and without SiGe S/D are compared for the first time. The strained PMOSFET suffers higher degradations in both hot carrier and FN tunneling reliabilities than those of the non-strained device. For both structures, the degradations caused by HCS are more serious than that caused by FN tunneling stress. The results of +�媒gs �l -�媒gd after HCS and �媒gs �l �媒gd �l0 after the FN tunneling stress imply that the degradation of fT depends only on that of gm regardless of the changes of Cgs and Cgd for both PMOSFETs with and without SiGe S/D.
|
|---|
