Analytical and compact models of the ONO capacitance in embedded non-volatile flash devices Davide Garetto* †, Erwan Dornel*, Denis Rideau §, William F. Clark*, Alexandre Schmid †, Saadia Hniki ‡§, Clement Tavernier §, Hervé Jaouen § and Yusuf Leblebici † * IBM France, 850 rue Jean Monnet, Crolles, France † Microelectronic Systems Laboratory (LSM) - EPFL, Lausanne, Switzerland § STMicroelectronics, 850 rue Jean Monnet, Crolles, France ‡ LAAS / CNRS, Université de Toulouse, 7 Avenue du C. Roche, Toulouse, France PRINCIPLES éé THE STRUCTURE MODEL DESCRIPTION The FG potential V FG in a non–volatile flash memory (NVM) device is the main parameter controlling the behavior of the cell; its calculation is required for compact modeling purposes Common modeling technique for V FG based on the calculation of the coupling coefficients between all the terminals C CF A model for the capacitance C CF between the control gate (CG) and the FG, separated by an oxide-nitride-oxide (ONO) dielectric layer, is required Most compact models used in industry consider fringing or corner capacitances as fitting parameters  not appropriate when technology scalability must be taken into account C CF Develop a physical–level model for C CF supporting accurate modeling of cell layout scalability as well as process variations Integrate the model into an advanced compact model for flash devices 2D cross sections 3D TCAD process simulations using 65 nm node process flow D S CG FG L W fg W/2 L = channel length of the cell W = channel width of the cell W fg = floating gate wing: extension of the FG over the Shallow Trench isolation (STI) region D ONO layer sandwiched between control gate (CG) and floating gate (FG) polysilicon layers 1. Structure analysis and identification of the ONO capacitance components 2. Model definition Apply the structure decomposition approach to the 3 cross-sections Working principles of a floating gate memory cell: 1)Information = charge on a floating gate (FG) node 2)Read : quantify the charge on FG measuring V TH 3)Program: inject electrons in FG using channel hot electron injection (CHE) 4)Erase: discharge electrons on the substrate by Fowler-Nordheim tunneling PROBLEMATICS / OBJECTIVES 3. Model validation Scaling the ONO capacitance C CF and its different components with respect to the width W and the length L of the device. Scaling C CF with respect to the active area of the cell (W * L) Excellent matching with the model has been found (average error < 3%) Model is also scalable with respect to the FG wing W fg For each capacitance component, integrate the Gauss law on the electrical field lines Fringing capacitances Analytical model : field lines approximated as semi-ellipses Compact model : field lines approximated as straight lines with curvature correction CONCLUSIONS Parallel plate capacitances Corner capacitance Fitting parameter representing the curvature of the field lines in the compact model (c) Cross-section BB’ C crn is the capacitance in the edge corner separating C pt from C pl Assumption on the electrical field lines Total ONO capacitance D D = maximum extension of the field lines of C fl in the spacer region L D = maximum extension of the field lines of C ft in the spacer region C pl = lateral parallel plate capacitance between FG and CG in the STI region C fl = fringing capacitances of C pl C pt = top parallel plate capacitance between FG and CG C ft = fringing capacitances of C pt physicalscalableWe have developed a fully physical and scalable model to accurately estimate the FG–CG coupling ONO capacitance in embedded high density memory devices 3D TCAD AC simulationsWe have validated our compact model with 3D TCAD AC simulations studying the dependence of C CF on critical layout dimensions coupling factor α CWe extracted the FG–CG coupling factor α C from DC TCAD simulations and demonstrated the importance of modeling 3D effects compact model for eNVM devicesThe proposed ONO capacitance model has been included into an accurate PSP–based compact model for eNVM devices, comparing the DC characteristics of the flash device with the ones of a “dummy” device, where FG and CG are short–circuited y x z Doping concentration [cm -3 ] COUPLING EXTRACTION FG – CG coupling coefficient α C extracted from DC TCAD simulations on the 3D structure vs. W, L and W fg 3D effects (parasitic hump effects and W dependence) strongly influencing coupling and cell performance 2009