On the experimental determination of channel back-scattering in nanoMOSFETs

Abstract

By using accurate Multi-Subband-Monte-Carlo simulations of quasi-ballistic transport we carry out a detailed re-examination of the experimental extraction procedure for the ballistic-ratio BR=I D /I BAL in nanoMOSFETs proposed in [1]. It is found that the ballistic-ratio extracted applying this procedure to the simulated drain current severely underestimates the BR extracted by directly compare the simulated I D and I BAL . This is mainly due to inaccurate determination of the temperature dependence of the inversion charge. It is suggested that this limitation of the extraction procedure may explain the apparent lack of improvement in the ballisticity with the geometrical scaling [1,5,13,14]. IntroductionThe back-scattering coefficient (r) [2,3] where (see Fig.1) Q inv and v inj are the inversion charge density and the average velocity of the injected carriers at the virtual source (VS), L kT is the extension of the kT-layer and µ 0 is the low-field mobility. Extraction procedure for BR.As described in [1], by taking the derivative with respect to the temperature T of the expressions in Eq.1, and assuming power-laws with exponent γ µ , γ vinj and γ lkt for the T dependence of µ 0 , v inj and L kT respectively, one gets:that becomes (using Eq.1):where α=∆I D /(I D ∆T ) and β=∆Q inv /(Q inv ∆T ). The parameter α can be measured directly, whereas β is more difficult to estimate, since Q inv needs to be sampled at the virtual source. In [1] it is a priori assumed that: γ µ =−1.5, γ vinj =0.5, γ lkt =1 and β=−η/(V GS −V T ), where η=∆V T /∆T . This set of parameters is denoted as set A in the following. 105 1-4244-0439-X/07/$25.00

ANALYTICAL MODELS AND ELECTRICAL CHARACTERISATION OF ADVANCED MOSFETs IN THE QUASI BALLISTIC REGIME

“…This dependency has been confirmed by Monte Carlo simulations 38 . Note that this apparent mobility corresponds to the mobility extracted from experiments, using the usual Drift Diffusion formula.…”

Criticisms on and comparison of experimental channel backscattering extraction methods

“…We want to point out that the most recent papers [11][12][13][14][15] use the Chen method to extract the backscattering coefficient and that the obtained results can be biased according to our previous considerations. Moreover, Zilli [17] attributed to A1 the major impact on the extracted backscattering, while here we found that is A5 the poorer approximation.…”

Scaling of TG-FinFETs: 3-D Monte Carlo Simulations in the Ballistic and Quasi-Ballistic Regimes

“…For example, one of the most widely used techniques is the one that is based on the temperature dependence of I SAT to extract the ratio of the mean free path to the critical length of the KT layer, hence BR [25]. However, it turned out to be quite controversial, as discussed in [24] and [26]. On the other side, being experimental does not guarantee the absolute validity of the extracted values, since the whole problem lies in the extraction of I BAL value, which remains a theoretical term.…”