M. Bonanomi

Statistical Model for Random Telegraph Noise in Flash Memories

This paper presents a new physics-based statistical model for random telegraph noise in Flash memories. From the probabilistic superposition of elementary Markov processes describing the trapping/detrapping events taking place in the cell tunnel oxide, the model can explain the main features of the random telegraph noise threshold-voltage instability. The results on the statistical distribution of the threshold-voltage difference between two subsequent read accesses show good agreement between measurements and model predictions, even considering the time drift of the distribution tails. Moreover, the model gives a detailed spectroscopic analysis of the oxide defects responsible for the random telegraph noise, allowing a spatial and energetic localization of the traps involved in the threshold-voltage instability process.

Transient conductive path induced by a Single ion in 10 nm SiO/sub 2/ Layers

Large charge loss can happen in isolated conductive lines when hit by a single high linear energy transfer (LET) ion. We have demonstrated this phenomenon by using floating gate (FG) memory arrays, which allowed us to study it on the basis of a large statistical set of data. Charge loss is by far larger than that expected from a simple generation-recombination model. FGs hit by ions experience a charge loss linearly dependent on ion LET and on the electric field. We are proposing a semi-empirical model based on the idea that a conductive path assimilable to a resistance connects the FG to the substrate during the time (10/sup -14/ s) needed for electrons to escape the tunnel oxide. The model is fully consistent with a broad range of theoretical and experimental results, and has excellent fitting capabilities.

A model for TID effects on floating Gate Memory cells

Four different technologies of floating gate (FG) memory arrays were subjected to /sup 60/Co gamma-rays and 10 keV X-rays irradiation to evaluate their response to the total ionizing dose. The effect of irradiation was a uniform charge loss across the whole array. Irradiation effects can be modeled as the result of two phenomena, namely, the generation of charge in the dielectric layers surrounding the floating gate and its subsequent recombination and drift, and the photoemission of carriers from the charged FG. The second phenomenon is effective at high doses. As a consequence of these two phenomena, devices featuring a smaller FG are less prone to total ionizing dose effects than devices featuring a larger FG, proper of older technological generations. We propose a model that accurately fits experimental data over a broad series of experimental conditions.

Radiation induced leakage current in floating gate memory cells

Single ions impacting on SiO/sub 2/ layers generate tracks of defects which may result in a Radiation Induced Leakage Current (RILC). This current is usually studied as the cumulative effect of ion-induced defects in capacitors with ultra-thin oxides. We are demonstrating and modeling this phenomenon in 10 nm oxides by using Floating Gate memories. The impact of a single, high-LET ion can result in severe retention problems, due to several electrically active defects, which cooperate to slowly discharge the FG. We are also proposing innovative simulation tools to reproduce this phenomenon. Results from simulations fully explain our results, and also agree with existing data on thinner (4 nm) oxides.

Cycling Effect on the Random Telegraph Noise Instabilities of nor and nand Flash Arrays

The impact of program/erase (P/E) cycling on the random telegraph noise (RTN) threshold voltage instability of NOR and NAND flash memories is studied in detail. RTN is shown to introduce exponential tails in the distribution of the threshold voltage variation between two subsequent read operations on the cells. Tail height is shown to increase as a function of the stress levels, with a larger relative increase for the NAND case. The slope of the distribution instead remains nearly independent of the number of applied P/E cycles. This reveals that trap generation takes place according to the native trap distribution over the active area and means that the tail slope is a basic RTN parameter, depending on the cell process details for a fixed technology. These results are important for the design of the threshold voltage levels in multilevel nor and NAND technologies.

Angular Dependence of Heavy Ion Effects in Floating Gate Memory Arrays

Single, high energy, high LET, ions impacting on a Floating gate array on grazing or near-grazing angles lead to the creation of long traces of FGs with corrupted information. Up to 30 consecutive devices can be involved in the trace left by a single ion. We demonstrate that charge collection at multiple nodes can be expected as the technology advances. One of the major implications is that the widely adopted cosine law should be used with great care when dealing with modern devices, with sizes smaller than 100 nm.

Error Instability in Floating Gate Flash Memories Exposed to TID

We discuss new experimental results on the post-radiation annealing of Floating Gate errors in Flash memories with both NAND and NOR architecture. We investigate the dependence of annealing on the program level, linking the reduction in the number of Floating Gate errors to the evolution of the threshold voltage of each single cell. To understand the underlying physics we also discuss how temperature affects the number of Floating Gate errors.

Physical modeling of single-trap RTS statistical distribution in flash memories

In this paper we present an accurate physical modeling for the statistical distribution of Random Telegraph Signal threshold voltage fluctuations in Flash memories by means of 3D device simulations accounting for realistic cell morphology, random discrete doping and random trap location. The model quantitatively describes the statistical behavior of the fluctuation amplitude, pointing out an exponential distribution which is in quite good agreement to what experimentally observed. The large distribution spread is explained on the basis of inhomogeneous substrate conduction, i.e. percolation effects induced by fixed charge, random discrete doping and current crowding due to local field enhancement. In addition, we investigate also the statistical spread dependence on technological parameters such as substrate doping, deriving design guidelines for technology optimization against Random Telegraph Signal instabilities.

Subpicosecond conduction through thin SiO2 layers triggered by heavy ions

Heavy ions impinging on thin silicon dioxide layers generate a dense plasma of electrons and holes. Under particular conditions, such as in thin oxide layer surrounding isolated conductive lines, this plasma can act as a conductive medium, able to carry current for very short times (10−14s). We studied this phenomenon by using large data set obtained on state-of-the art floating gate memory arrays. Floating gates hit by ions experience a charge loss linearly dependent on ion linear energy transfer and on the electric field across the tunnel oxide. Despite its absolute low value, reaching at most thousands of electrons, charge lost from floating gate exceeds by orders of magnitude that expected from relatively simple models, such as generation and recombination. The model we are proposing is fully consistent with a broad range of theoretical and experimental results, and has excellent fitting capabilities.

Effect of different total ionizing dose sources on charge loss from programmed floating gate cells

We irradiated programmed Floating Gate (FG) memory arrays with different radiation sources, including 10 keV X-rays, /sup 60/Co /spl gamma/-rays, and 27 MeV protons. After irradiation, FGs experience a net charge loss which can degrade the stored information in terms of MOSFET threshold voltage. The charge loss is the result of two different phenomena: charge generation/recombination in the oxides and photoemission from the FG. The threshold voltage shift in irradiated devices depends on the radiation source: strong dose enhancement phenomena were found after X-ray irradiation, whereas proton results closely follow /spl gamma/-ray results.