No abstract

A b s t r a c tThe high-frequency (HF) behaviour of MOSFETs from different CMOS processes has been characterized. Smallsignal Y-parameters and derived quantities such a s current, voltage, and power gain have been measured and good agreement with calculations using the Philips compact model, MOS MODEL 9, has been obtained. Using HF measurements on MOSFETs and Bipolar devices, it is shown that the figures of merit, related to the gains mentioned above, are not sufficient to evaluate the HF capabilities of CMOS. BiCMOS and Bipolar technologies. I n t r o d u c t i o n Recent CMOS technologies with gate lengths of 0.1 pm have resulted in unity-current-gain frequencies ( f~) as high as 118 GHz [l] and ring-oscillator gate delays as low as 11.8 psec 121. In the literature both quant.ities are used to benchmark the high-frequency (HF) performance of CMOS technologies relative to Si bipolar technologies, for which fr's of 74 GHz and ECL gate delays of 15 psec have been reported [3]. The use of specific test circuits to evaluate the HF performance of different technologies emphasizes that specification of the maximum f~ is insufficient. Moreover, CMOS ring-oscillators and ECL are not quite comparable: they represent each only a limited application area of CMOS and bipolar technologies. Effective evaluation of different technologies should consist of i) HF characterization; iz) compact model verification;iii) (using this compact model) simulation of the performance of any analogue circuit, which may be processed in the technologies under investigation. Here we present a detailed HF characterization of CMOS devices from different process generations and the verification of the Philips compact MOS model MOS MODEL 9 [4] for these processes.It will be demonstrated that a fair comparison between the HF performance of Bipolar and CMOS processes can not be performed using figures of merit on device level. E x p e r i m e n t a l s e t -u p and test devices Small-signal scattering (S) parameters in two-port configuration have been measured with an HP8510 Network ,41ialyzer. To match t hc ground-signal-ground probes of Cascade Microtech, special test structures have been designed with the MOSFETs in common source-bulk configuration. Measured S-parameters have been converted to admittance (Y) parameters and dummy structures have been used to correct for interconnect parasitics. The individual terminal capacitances have been measured at low frequencies with an HP4284A LCR-meter.Small-signal Y -p a r a m e t e r s In fig. 1 measurements of the real and imaginary part of the drain-gate admittance are plotted. In order to get insight in the quantities and parasitic elements determining the Y-parameters, approximate expressions for these parameters have been derived using the equivalent circuit of fig. 2. For the actual simulations we used the Philips compact model MOS MODEL 9 [4] with parameters obtained from DC current measurements. Overlap and depletion capacitances were determined from low-frequency capacitance measurements. The fr...

No abstract

No abstract

No abstract