Figure - uploaded by Gokhan Bakan
Content may be subject to copyright.
(a) Cartoon illustration of the simulated phase-change memory cell (cross-section view) in SET (low resistance) state showing the layers forming the cell and electrical/thermal boundary conditions. An external 100 kΩ resistor is connected in series with the cell to limit the current. Close-up shows the active region with the recessed heater (1 nm) and current directions for the positive/negative voltage polarity configurations. (b) RESET cell resistance after pulses with indicated pulse conditions: polarity, current amplitude, pulse duration. SET resistance level is marked with a red dashed line. (c) Resistivity map of the cell around the active region after the RESET pulse for the indicated pulse conditions.
Source publication
We have observed how thermoelectric effects that result in asymmetric melting of silicon wires are suppressed for increasing electric current density (J). The experimental results are investigated using numerical modeling of the self-heating process, which elucidates the relative contributions of the asymmetric thermoelectric Thomson heat (∼J) and...
Contexts in source publication
Context 1
... and low resistance crystalline phase using self-heat- ing. 19 The most common PCM structure, mushroom cell ( Figure 6(a)), is asymmetric along the current path in con- trast to symmetric Si wire structures. As a result of this asymmetry, the phase-change material on top of the narrow bottom electrode (heater) heats up the most due to the cur- rent confinement around the heater. ...Context 2
... method, however, yields slightly different current ampli- tudes for opposite voltage polarities. Hence, here, a 100 kX load resistor is used to limit the current, while varying the pulse amplitude (Figure 6(a)). For the positive polarity con- figuration, the electrical pulse amplitudes are chosen such that the same RESET resistance is achieved for various pulse durations (0.1 to 100 ns) (see supplementary material for applied voltage pulse amplitudes and resulting currents). ...Context 3
... the positive polarity con- figuration, the electrical pulse amplitudes are chosen such that the same RESET resistance is achieved for various pulse durations (0.1 to 100 ns) (see supplementary material for applied voltage pulse amplitudes and resulting currents). The same electrical pulse configurations (amplitude and du- ration) are then used for the negative polarity resulting in smaller amorphous volumes, hence RESET resistances (Figures 6(b) and 6(c)). The difference between the RESET resistances for positive and negative polarities decreases with increasing voltage amplitudes, due to dominant Joule heating, suggesting that symmetric operation of PCM cells is possible when large amplitude, short duration pulses are used (Figures 6(b) and 6(c)). ...Context 4
... same electrical pulse configurations (amplitude and du- ration) are then used for the negative polarity resulting in smaller amorphous volumes, hence RESET resistances (Figures 6(b) and 6(c)). The difference between the RESET resistances for positive and negative polarities decreases with increasing voltage amplitudes, due to dominant Joule heating, suggesting that symmetric operation of PCM cells is possible when large amplitude, short duration pulses are used (Figures 6(b) and 6(c)). ...Similar publications
Cored wire injection is used to perform alloying additions during
secondary steelmaking, presenting advantages such as higher
yield over bulk additions. The wire consists of a casing wrapped
around a core of powdered material, this setup delays the release
of material into the melt. This work presents a 1-D finite volume
numerical model aimed for t...
Citations
... En effet, l'effet Thomson a un impact sur la puissance des TEGs [57,58] et donc des refroidisseurs Peltier. De manière plus spectaculaire, il est responsable du fait que la fusion de micro-fils en sillicium chauffés par des courants ne se fasse pas en leur milieu [59]. Il a aussi un homologue magnétique où le courant électrique est remplacé par un courant de spins causé par un champ magnétique appliqué au matériau [60]. ...
Dans le domaine industriel, la thèse présente l'étude, la conception et l'implémentation d'un dispositif autonome en énergie fonctionnant avec des générateurs thermoélectriques (TEGs), qui alimente des capteurs qui effectuent des mesures de température et d'humidité relative qu'il envoie à un hôte passerelle. Ce dispositif fonctionne grâce à la chaleur causée par la différence de température entre la surface du sol et une certaine profondeur souterraine. Il est question de déterminer la longueur optimale des tiges thermoconductrices qui permettent le passage de la chaleur à travers les TEGs, leur forme, leur nombre, le choix du matériau les constituant, le choix du type, du nombre et de l'arrangement thermique et électrique des TEGs. Ces TEGs alimentent une carte électronique qui possède une partie gestion de puissance, dont les caractéristiques ont été choisies pour que le dispositif puisse fonctionner en permanence. Un prototype a été fabriqué, puis implémenté dans un sol terreux en France et a fonctionné environ 80% du temps durant plus d'un mois, à partir du mois de mai 2019. Des pistes d'améliorations sont évoquées pour l'aboutissement d'une version commerciale.Dans le domaine de la métrologie en thermoélectricité, la thèse présente la description d'une nouvelle méthode de mesure de l'effet Thomson, adaptée pour des échantillons filaires métalliques. Elle est distincte des méthodes préexistantes, non seulement par le fait que l'échantillon n'est pas perturbé par une mesure de température avec un thermomètre qui ferait contact, mais aussi par sa simplicité et sa rapidité. Seule la mesure de la résistance électrique de l'échantillon doit être effectuée préalablement, et elle s'apparente à la méthode 3ω avec pour différence l'ajout d'un gradient thermique le long de l'échantillon. La méthode a été vérifiée de manière numérique, ainsi que par comparaison avec une mesure du coefficient de Seebeck de manière classique. Nous montrons qu'elle peut s'étendre à des températures à plus de 1000 K car les résultats sont peu affectés par les effets du rayonnement. Cette nouvelle méthode de mesure peut servir pour établir une calibration avec les matériaux à disposition, en permettant l'obtention du coefficient de Seebeck de manière absolue.La thèse comporte aussi une étude de faisabilité d'une nouvelle méthode de mesure des propriétés de transport composant le facteur de mérite thermoélectrique, pour les matériaux anisotropes. La méthode est une sorte d'extension de la méthode présentée pour mesurer l'effet Thomson. Cependant, pour les matériaux anisotropes, elle se heurte aux problèmes de l'élaboration de bons contacts électriques avec des échantillons minces monocristallins, et à la mesure précise de l'emissivité de ces petits échantillons, entre autres. Elle a toutefois permis l'obtention du tenseur de résistivité électrique pour un échantillon de Au0.1TiS2 où l'or est un intercalant dans la structure cristallographique du TiS2.
... Every submodels have their own tasks to fulfill such as; an electrical model including temperature and phase dependent electrical conductivity changing, a thermal model to solve heat diffusion equation to obtain the joule heating from the electrical current with temperature and phase dependent thermal conductivity, and last one, phase change model to determine temperature dependent homogeneous and heterogeneous nucleation and growth kinetics of crystallites (Reifenberg et al., 2006;Peng et al., 1997;Fiflis et al., 2013). Table 1 shows used thermal and electrical conductivities, Seebeck coefficient, as well as heat capacity values at room temperature for the materials used (Reifenberg et al., 2006;Peng et al., 1997;Fiflis et al., 2013;Bakan et al., 2014;Kim et al., 2007;. Table 1. ...
... Table 1. The electrical conductivity, thermal conductivity, Seebeck coefficient and heat capacity used in simulations (Reifenberg et al., 2006;Peng et al., 1997;Fiflis et al., 2013;Bakan et al., 2014;Kim et al., 2007;. [ ∇ ] = 0 is solved iteratively (with 10 ps time steps) for each mesh element in conjunction with the thermal submodel to obtain the spatial electrical potential distribution F(x,y,z). ...
Generally, a semiconductor device using chalcogenide elements as a fundamental components is considered as a potentially revelation technology for future ultra-high density data storage technology. These kind of device having high contrast between 0 and 1 logic states brought out the possible application of the idea of multiple logic levels in a single bit in an effort to boost data storage density. The potential stabilization of resistance levels in between the logic states enables storage of several data in a single cell (such as 00, 01, 10, 11 levels). I report on investigation of the role of the current injection and material selection in stabilizing middle resistance states within a nanoscale semiconductor cell for fabrication of a multiple-bit-per-cell through 3D finite element modeling. First, to visualize the complex nature of the switching dynamics, 3D finite element simulations were carried out in cell with two active layers Ge2Sb2Te5/Ge2Sb2Te5 (GST/GST) alloys incorporating phase change kinetics, electrical, thermal and percolation. Simulation was constructed by using an iterative approach with coupled differential equations, which are all as a function of temperature, as well as Seebeck coefficient to account for thermoelectric effect. The complex nature of switching dynamics appears highly sensitive to the exact programming voltage and material properties. The model suggests that the physical origin of the formation of stable middle states unexpectedly in circular top contact devices is mainly due to anisotropic heating during the application of a programming current pulse. The model successfully predicts the required programing conditions and the importance of material selection for such mixed-phase levels, which can be used to optimize memory cells for future ultra-high-density data storage applications.
... If the solid material is unipolar, direction In this chapter we describe a non-equilibrium non-isothermal semiconductor treatment that captures heat exchanges associated with free carrier generation-transport-recombination (GTR) as well as local charging and changes in the band structures as a function of temperature. We use this model to analyze the contributions that result in the significant asymmetry we have observed in pulsed self-heating experiments on nanocrystalline silicon (nc-Si) micro-wires, where ~10 7 A/cm 2 was forced through the wires with application of 1 µs, ~30 V pulses, generating ~1 K/nm thermal gradients [12], [19], [27]- [29], [111]. These micro-wires were fabricated by Ali Gokirmak and Nathan Henry by (i) patterning a highly doped (~10 19 cm -3 ) nc-Si thin film deposited on silicon dioxide, using conventional photolithography and etching, (ii) releasing the wires from the substrate using a buffered oxide etch, while the two ends remain anchored to the substrate from the larger contact regions [12], [28]. ...
Under extreme thermal gradients (~> 1 K/nm) significant asymmetric melting of current-carrying highly-doped silicon micro-wires was observed. This bizarre phenomenon was explained by the generation-transport-recombination (GTR) process of the minority carriers in semiconductors at elevated temperatures. The finite element simulation model, using effective media approximation, supported this explanation; however, the asymmetry observed in the experiments is significantly larger than that obtained from the computational results. As the calculation based on conventional non-isothermal semiconductor physics underestimates the exponential growth of carrier generation near melting, the carrier concentration, hence the shift of the hottest spot because of the thermoelectric Thomson effect, was underestimated.
In this work, a high temperature semiconductor model is developed where the continuity equations for electrons and holes are solved and the energy exchange associated with generation-recombination processes are duly accounted for. Modified generation-recombination mechanisms included in the model sharply increase the carrier concentration to ~1022 cm-3 at melting for silicon, and the shift of hottest spot is closer to that observed in the experiments. This approach allows us to calculate a more accurate melt profile in non-equilibrium conditions, which is also critically important for designing phase change memory (PCM) cells. PCM is an emerging non-volatile memory technology where electrical pulses are used to transition between high and low resistance states and where the thermoelectric effects are significant due to extreme thermal gradients (~> 50 K/nm) at high operating temperatures.
To adopt this model for PCM, characterization of material properties at elevated temperature is required. In this work, the effective activation energy of metastable Ge2Sb2Te5 (GST) is calculated from device level high speed resistivity measurements data. The extracted temperature dependent activation energy shows parabolic shape with maximum at ~450 K, and is expected to correspond to a distribution of trap levels that contribute to conduction. Also, from the simultaneous Seebeck-temperature and resistivity-temperature measurements of amorphous-fcc (face centered cubic) mixed-phase GST thin-films, the Fermi energy of single crystal fcc GST was calculated, which is 0.16 eV bellow the valence band edge.
... Next, we consider a graphene channel that is Jouleheated by a current injected through thermally anchored contacts. Due to a combination of the Seebeck and Peltier effects (explained below), the temperature profile is skewed in the direction of the particle flow [50][51][52]. Commonly referred to as the "Thomson effect", this phenomenon strongly depends on the local value of zT and is typically weak in conventional systems. In the hydrodynamic regime, however, we show that the temperature dependence of κ Hyd and the increased Q drastically amplify the Thomson effect. ...
... These terms, reintroduced in the high-bias regime, are responsible for skewing the temperature profile in the direction of particle flow. Such behavior is commonly referred to as the "Thomson effect" [50][51][52], and can be understood as follows. Via the Seebeck effect, the non-uniform temperature profile produces an electric force that pushes particles in the direction of increasing temperature, independently of the sign of their charge. ...
The thermoelectric (TE) properties of a material are dramatically altered when electron-electron interactions become the dominant scattering mechanism. In the degenerate hydrodynamic regime, the thermal conductivity is reduced and becomes a {\it decreasing} function of the electronic temperature, due to a violation of the Wiedemann-Franz (WF) law. We here show how this peculiar temperature dependence gives rise to new striking TE phenomena. These include an 80-fold increase in TE efficiency compared to the WF regime, dramatic qualitative changes in the steady state temperature profile, and an anomalously large Thomson effect. In graphene, which we pay special attention to here, these effects are further amplified due to a doubling of the thermopower.
... The reset pulse duration is chosen to be short to suppress the asymmetry in the molten volume due to the thermoelectric effects. 20 The short pulse duration, however, necessitates a large pulse amplitude (9 V) for melting. Both reset and set operations use the same setup: an external load resistor (R load ) is connected in series with the PCM cell (R GST þ R TiN ) which is grounded through a 50 X termination resistor (R termination ) (Figure 2(a)). ...
Phase-change memory (PCM) devices are enabled by amorphization- and crystallization-induced changes in the devices' electrical resistances. Amorphization is achieved by melting and quenching the active volume using short duration electrical pulses (∼ns). The crystallization (set) pulse duration, however, is much longer and depends on the cell temperature reached during the pulse. Hence, the temperature-dependent crystallization process of the phase-change materials at the device level has to be well characterized to achieve fast PCM operations. A main challenge is determining the cell temperature during crystallization. Here, we report extraction of the temperature distribution on a lateral PCM cell during a set pulse using measured voltage-current characteristics and thermal modelling. The effect of the thermal properties of materials on the extracted cell temperature is also studied, and a better cell design is proposed for more accurate temperature extraction. The demonstrated study provides promising results for characterization of the temperature-dependent crystallization process within a cell.
... The reset pulse duration is chosen to be short to suppress the asymmetry in the molten volume due to the thermoelectric effects. 20 The short pulse duration, however, necessitates a large pulse amplitude (9 V) for melting. Both reset and set operations use the same setup: an external load resistor (R load ) is connected in series with the PCM cell (R GST þ R TiN ) which is grounded through a 50 X termination resistor (R termination ) (Figure 2(a)). ...
Phase-change memory (PCM) devices are enabled by amorphization- and crystallization-induced changes in the devices' electrical resistances. Amorphization is achieved by melting and quenching the active volume using short duration electrical pulses (∼ns). The crystallization (set) pulse duration, however, is much longer and depends on the cell temperature reached during the pulse. Hence, the temperature-dependent crystallization process of the phase-change materials at the device level has to be well characterized to achieve fast PCM operations. A main challenge is determining the cell temperature during crystallization. Here, we report extraction of the temperature distribution on a lateral PCM cell during a set pulse using measured voltage-current characteristics and thermal modelling. The effect of the thermal properties of materials on the extracted cell temperature is also studied, and a better cell design is proposed for more accurate temperature extraction. The demonstrated study provides promising results for characterization of the temperature-dependent crystallization process within a cell.
With the emergence of phase change memory, where the devices experience extreme thermal gradients (∼100 K/nm) during transitions between low and high resistive states, the study of thermoelectric effects at small scales becomes particularly relevant. We had earlier observed asymmetric melting of self-heated nano-crystalline silicon micro-wires, where current densities of ∼10⁷ A/cm² were forced through the wires by 1 μs, ∼30 V pulses. The extreme asymmetry can be explained by the generation of considerable amount of minority carriers, transport under the electric field, and recombination downstream, a heat transfer process we termed as generation–transport–recombination, which is in opposite direction of the electronic-convective heat carried by the majority carriers. Here, we present a full semiconductor physics treatment of this carrier-lattice heat transport mechanism and the contribution of the minority carriers on the evolution of the melt–solid interface, which can be applied to various high-temperature electronic devices.
The thermoelectric (TE) properties of a material are dramatically altered when electron-electron interactions become the dominant scattering mechanism. In the degenerate hydrodynamic regime, the thermal conductivity is reduced and becomes a decreasing function of the electronic temperature, due to a violation of the Wiedemann-Franz law. We here show how this peculiar temperature dependence gives rise to new striking TE phenomena. These include an 80-fold increase in TE efficiency compared to the Wiedemann-Franz regime, dramatic qualitative changes in the steady state temperature profile, and an anomalously large Thomson effect. In graphene, which we pay special attention to here, these effects are further amplified due to a doubling of the thermopower.
