Result of measured N-content in test set of Ge2Sb2Te5:N samples. XRF calibrated using a set of reference materials certified by IBA.

Telluride films are widely applied in data storage devices (advanced resistive memories, DVDs and Blue-rays disks), photovoltaic cells and infrared detectors. The properties of thin telluride alloys are deeply influenced by their chemical composition and compositional depth profile, whereas surface/interface effects may become preponderant in ultrathin films. The combination of X-ray reflectometry (XRR) and grazing-incidence X-ray fluorescence (GIXRF) is particularly adequate to probe these complex thin layered materials. In this paper, we evaluate the performances of Lab-based and synchrotron-based XRR/GIXRF strategies to characterize ultrathin (< 10 nm) amorphous titanium-tellurium films elaborated by Physical Vapor Deposition (PVD), and capped in situ with 5 nm-tantalum passivation layer. We highlighted the impact of the instrumental setup on the qualitative XRR and GIXRF data and on the quantitative information deduced from the combined analysis. Both synchrotron-based and Lab-based strategies were sensitive enough to track the impact of small PVD process changes on the chemical depth-profiles, and to unambiguously reveal undesired tantalum-tellurium inter-diffusion that was confirmed by X-Ray Photoelectron Spectroscopy.

We present in this paper a method to measure the transmission spectra of optical filters composed of Complementary Metal-Oxide-Semiconductor (CMOS) compatible materials thin layers in order to be fully integrated on various types of CMOS image sensors (ambient light sensors, proximity detection, red green blue colour imaging, etc.). As the filters have to be deposited on top of a CMOS device, a good approach in order to evaluate with accuracy their response on chip is (i) to achieve the stacks on Si wafers (as it is the case for the CMOS sensor) (ii) then to perform a direct bonding of the structure on glass wafers (iii) in the end to remove the entire bulk silicon. In this way, we show the measured spectral responses of multilayer interference filters and can check particularly the agreement of the transmission peak with the theoretical calculations and its reproducibility wafer to wafer. It enables to optimize the filters optical designs and to demonstrate that the developed filters fulfill typical CMOS requirements of integration and reliability.

Direct bonding of PECVD SiOx films is of great interest in the field of microtechnologies for applications such as stacked structures, thin film transfers, whose fabrication processes still deserve to be investigated. This work deals with the influence of water diffusion in the SiOx films deposited onto 200 mm Si wafers and its impact on adherence of hydrophilic surfaces. The as-deposited film nature induces various stresses. Stresses evolutions and water penetration are characterized and a correlation is made between the kinetics of both stress variations and water diffusion through the oxide thin films. Impacts of room temperature (RT) storage prior to bonding and thermal treatments applied for strengthening the bonding is shown. As direct bonding operates through surface asperities deformation, specific mechanical properties of such contact asperities can lead to a stronger bonding by increasing bonding area (Fournel et al., ECS J Solid State Sci Technol 4(5):124–130, 2015; Rieutord et al., ECS Trans 3(6):205–215, 2006; Ventosa et al., J Appl Phys 104:123524, 2008, Electrochem Solid State Lett 12(10):H373–H375, 2009). As the mechanical properties of silicon oxide material are greatly influenced by internal water concentration, aging and water diffusion on asperity have a real impact in term of direct bonding energy. Moreover at bonding interfaces, it is observed by X-ray reflectivity (XRR) that aging of deposited SiOx prior to bonding enables shallower bonding gaps, indicating better bonding interface closure which is coherent with the bonding energy enhancement. All these results confirm the link between water diffusion and hydrophilic surface adherence of deposited SiOx films.

In this paper, we use Kinetic Monte Carlo (KMC) simulations to investigate CBRAM variability. A full consistent model able to simulate SET, RESET, retention and endurance characteristics was proposed for the 1st time, allowing to describe experimental data obtained on Al 2 O 3 /CuTex based CBRAM. The role of oxygen vacancy generation during programming is described and its impact on reliability (retention and endurance) is elucidated. The origin of the resistance spread is discussed and linked to the conductive filament shape and operating conditions. The cycle to cycle contribution on resistance variability is uncorrelated from the intrinsic resistance distribution limit. Finally, guidelines are given in order to optimize the memory distribution, reduce tail bits and improve CBRAM reliability.

Conductive filament formation and composition in Oxide-based Conductive Bridge Random Access Memory (CBRAM) are investigated. To this end, Al2O3/Cu-based CBRAM is electrically characterized and studied. Current-voltage characteristics exhibit different forming behaviors depending on device polarization exposing the charged species involved during the forming process. In order to get more insights at the microscopic level, ion diffusion is investigated in depth by first-principles calculations. We study different point defects in Al2O3 which can come either from the post-process of the material itself or after top electrode deposition or during device operation. Since the role of Oxygen Vacancies (VO) and Copper (Cu) ions is core to the switching mechanism, ab initio calculations focus on their displacements. For different charge states in Al2O3, we extract the thermodynamic and activation energies of Cu, Te, Al, and O related point defects. The results reveal that Cu is not the only ion diffusing in the A...

In this paper the memory performances of the TiN/HfO 2 /Ti/TiN and TiN/Ta 2 O 5 /TaOx/TiN memory stacks are compared. First, the bipolar switching parameters and the effect of the compliance current on the memory window and endurance are investigated. Then, the endurance and data retention properties are compared at a given operating current (100μA). Ta 2 O 5 based memory stack exhibits a better memory window (2 decades) and data retention, while the HfO 2 one shows good endurance properties (10 8 cycles). Finally, thanks to ab initio calculations using Density Functional Theory, the stability of the conductive filament is investigated in both HfO x and TaO x dielectrics.

In this paper we investigate the impact of N-doping in optimized Ge-rich Ge2Sb2Te5 materials on device programming and storing performance. We integrate these alloys in state-of-the-art Phase-Change Memory (PCM) cells and we analyze the efficiency of the SET operation in N-doped and undoped memory cells, comparing voltage based programming with current based programming. This aspect is extensively investigated through electrical characterization, physico-chemical analysis and electro-thermal simulations. The thermal stability of these devices is finally evaluated and high temperature data retention is granted enabling PCM for embedded applications.

In this paper, a detailed reliability analysis of metal-oxide conductive bridge memories (CBRAM) is presented. This paper mostly focuses on electrical characterization of metal-oxide CBRAM devices endurance, using optimized program/erase conditions, and data retention at high temperature. The addition of a thin metal-oxide layer (0.5 nm-thick Al2O3) in the bottom of the GdOx memory stack significantly increases the ROFF and the memory window (more than one decade), with improved endurance performance (up to 105 cycles) with respect to the monolayer CBRAM device. Meanwhile, high thermal stability was also achieved (two decades of window margin are constantly maintained beyond 24 h at 250 °C). The bilayer oxide GdOX/Al2O3 CBRAM is a promising technology for potential future high density memory applications.

In this paper, we propose the integration of an Al2O3/CuTe x based Conductive Bridge RAM (CBRAM) device in vertical configuration. The performances of the memory devices are evaluated. 20ns switching time, up to 106 cycles and stable 150°C retention were demonstrated. Functionality is compared with Vertical RRAM integrating an HfO2/Ti OXRAM stack, showing the pros and cons of each configuration. Then 2 potential applications are discussed using design approach. For high density, the Vertical RRAM cell features and circuit are dimensioned to optimize the memory page density. Finally, for neuromorphic applications, selector and array configuration are tuned to reduce the variability in terms of voltage seen by each cell constituting a vertical synapse.

There is a great deal of activity in the development of new memory technologies that can be used to provide the required density and reliability for future generations of data storage [1, 2]. At this time there is a bewildering array of proposed systems each with advantages and disadvantages. One of the problems with the development of these types of materials systems is that it is not clear exactly how these devices function and as a consequence, it is difficult to select the best combinations of materials to provide the best overall performance. In this presentation we will present results that have been obtained on a range of different TaO and Ta2O5 OxRAM structures that have processed using reactive and RF deposition physical vapour deposition (PVD). The focus on this work is the mapping of oxygen and results will be presented that have been obtained by a range of different techniques including aberration‐corrected high‐resolution annular bright‐field (ABF) scanning transmission electron microscopy (STEM) imaging, EDS (Energy dispersive X‐Ray Spectroscopy) or Electron Energy Loss Spectroscopy (EELS) for the measurement of oxygen concentration (atoms) and electron holography and differential phase contrast (DPC) for the distribution of electrostatic potential caused by the distribution of the oxygen. An example is shown in Figure 1 where a (a) high resolution ABF STEM and (b) HAADF STEM image of a TaO/Ta2O5 stack with TiN top and bottom electrodes can be seen. Figures 1(c) and (d) show EELS spectra that have been acquired for the N, Ti, O and Ta regions. Figure 1(e) and (f) show quantitative maps and profiles respectively. Here the variations in the Ta and O concentrations can be observed across the active region of the device. In this presentation we will highlight the best techniques for measuring the variations of oxygen in the devices and discuss which stacks have the best stoichiometry for memory device applications. The key to the performance of these devices is the movement of oxygen during switching. As a consequence it is necessary to perform these observations in situ in the TEM to avoid problems with locating the conducting filaments during specimen preparation and additional issues such as device retention and other modifications that could occur during specimen preparation. In this presentation we will present results on specimens that have been switched using a dedicated in‐situ holder in the electron microscope. Figure 2 shows how a movable probe is placed onto a FIB‐prepared specimen with nm‐scale accuracy by using a Nanofactory biasing system. An electrical pulse is then used to switch the specimen in situ in the TEM. In this presentation we will show how to avoid common problems such as device heating and electrical shorting of the specimen from redeposition during preparation by focused ion beam milling. Finally in this presentation we will show results obtained on our TaO specimens that have been switched in situ in the microscope.

In this paper we clarify for the first time the correlation between endurance, window margin and retention of Resistive RAM. To this aim, various classes of RRAM (OXRAM and CBRAM) are investigated, showing high window margin up to 10 10 cycles or high 300°C retention. From first principle calculations, we analyze the conducting filament composition for the various RRAM technologies, and extract the key filament features. We then propose an analytical model to calculate the dependence between endurance, window margin and retention, linking material parameters to memory characteristics.

In this paper, we investigate in depth Forming, SET, and Retention of conductive-bridge random-access memory (CBRAM). A kinetic Monte Carlo model of the CBRAM has been developed considering ionic hopping and chemical reaction dynamics. Based on inputs from ab initio calculations and the physical properties of the materials, the model offers the simulation of both the Forming/SET and the Data Retention operations. It aims to create a bond between the physics at atomic level and the device behavior. From the model and experimental results obtained on decananometric devices, we propose an understanding of the physical mechanisms involved in the CBRAM operations. Using the consistent Forming/SET and Data Retention model, we obtained good agreement with the experimental data. Finally, the impact of each layer of the CBRAM on the Forming/SET behavior is decorrelated, allowing an optimization of the performance.

In this paper, a novel conformal boron nitride film deposited by PECVD is presented. In view of its low deposition temperature (Tdep=480°C), low dielectric constant (κ=3.8) and low wet etch rates in standard clean chemistries, this film is very attractive for the replacement of conventional silicon nitride as offset spacer in 3D VLSI CoolCube TM integration.

In this paper, a detailed reliability analysis of metal-oxide CBRAM devices is presented. We demonstrated that the addition of a thin metal-oxide layer in the bottom of the memory stack significantly increases the ROFF and the memory window (more than 1 decade), with improved endurance performance. At the same time, high thermal stability was also achieved (window margin constant during more than 24 hours at 250°C). The origin of the window margin degradation during endurance is discussed and interpreted by means of a Trap Assisted Tunneling Model, putting in evidence the role of defect generation and Cu residual atoms in the resistive layer.