Breakdown Spots Research Articles

Nondestructive imaging of breakdown process in ferroelectric capacitors using in situ laser-based photoemission electron microscopy

HfO2-based ferroelectrics are one of the most actively developed functional materials for memory devices. However, in HfO2-based ferroelectric devices, dielectric breakdown is a main failure mechanism during repeated polarization switching. Elucidation of the breakdown process may broaden the scope of applications for the ferroelectric HfO2. Here, we report direct observations of a breakdown process in HfO2-based ferroelectric capacitors, by in situ laser-based photoemission electron microscopy. We have not only clearly visualized the hard dielectric breakdown (HDB) spot but also observed the regions responsible for the soft dielectric breakdown (SDB), which is a precursor phenomenon to HDB. It was found that the low-resistance region formed after SDB is wider than the conduction path formed after HDB. Furthermore, our spectromicroscopic analysis revealed that the photoelectron spectrum after SDB shows an enhancement in intensity without spectral-shape modulation, interpreted that the initially existed defects are increased. In the HDB spot, however, an additional shoulder structure was observed. These results provide spectroscopic evidence that the electronic states responsible for the conduction path after SDB are different from those after HDB. Through this work, we propose this microscopic approach as a versatile tool for studying buried materials as they are, accelerating the development of material engineering for advanced electronic devices.

Oxide Breakdown Spot Spatial Patterns as Fingerprints for Optical Physical Unclonable Functions

Oxide Breakdown Spot Spatial Patterns as Fingerprints for Optical Physical Unclonable Functions

Direct atomic layer deposition of ultra-thin Al2O3 and HfO2 films on gold-supported monolayer MoS2

In this paper, the atomic layer deposition (ALD) of ultra-thin films (<4 nm) of Al2O3 and HfO2 on gold-supported monolayer (1L) MoS2 is investigated, providing an insight on the mechanisms ruling the nucleation in the early stages of the ALD process. A preliminary multiscale characterization of large area 1L-MoS2 exfoliated on sputter-grown Au/Ni films demonstrated: (i) a tensile strain (from 0.1 to 0.3%) and p-type doping (from 1 × 1012 to 4 × 1012 cm−2) distribution at micro-scale; (ii) an almost conformal MoS2 membrane to the Au grains topography, with some locally detached regions, indicating the occurrence of strain variations at the nanoscale; (iii) atomic scale variability (from ∼ 4.0 to ∼ 4.5 Å) in the Mo-Au atomic distances was detected, depending on the local configuration of Au nanograins. Ab initio DFT calculations of a free-standing MoS2 layer and a simplified MoS2/Au(111) interface model showed a significant influence of the Au substrate on the MoS2 energy band structure, whereas small differences were accounted for the adsorption of H2O, TMA (co-reactant, and Al-precursor, respectively) molecules, and a slight improved adsorption was predicted for TDMAHf (Hf-precursor). This suggests a crucial role of nanoscale morphological effects, such as the experimentally observed local curvature and strain of the MoS2 membrane, in the enhanced physisorption of the precursors. Afterwards the nucleation and growth of Al2O3 an HfO2 films onto 1L-MoS2/Au was investigated in detail, by monitoring the surface coverage as a function of the number (N) of ALD cycles, with N from 10 to 120. At low N values, a slower growth rate of the initially formed nuclei was observed for HfO2, probably due to the bulky nature of the TDMAHf precursor as compared to TMA. On the other hand, the formation of continuous films was obtained in both cases for N > 80 ALD cycles, corresponding to ∼ 3.6 nm Al2O3 and ∼ 3.1 nm HfO2. Current mapping on these ultra-thin films by conductive-AFM showed, for the same applied bias, a uniform insulating behavior of Al2O3 and the occurrence of few localized breakdown spots in the case of HfO2, associated to a less compact films regions. Finally, an increase of the 1L-MoS2 tensile strain was observed by Raman mapping after encapsulation with both high-κ films, accompanied by a reduction in the PL intensity, explained by the effects of strain and the higher effective dielectric constant of the surrounding environment.

Automated real-time study of the defect-induced breakdown occurring on a film-electrode system under a high electric field.

The energy storage density of a capacitor depends on its relative permittivity and breakdown strength. Breakdown of a thin film always first occurs at weak defect spots of dielectrics under a high electric field. It is of great significance to study the defect-induced breakdown of dielectrics to improve the breakdown strength of the dielectric. The majority of studies about the defect-induced breakdown only determine a certain voltage inducing the breakdown, and the single-hole breakdown spots influence the defect-induced breakdown and the intrinsic breakdown under a high electric field, which is hard to facilitate the in-depth study of improving the breakdown strength. Herein, the self-healing breakdown techniques are applied to avoid the influence of single-hole breakdown. An automated real-time testing system is used to study the defect-induced breakdown of various complex film-electrode systems, which accomplishes the temporal and spatial localization of breakdown events according to the physical chemistry characteristics of breakdowns and intelligently displays breakdown events, and detailed classification methods of the defect-induced breakdown are discussed concisely and efficiently. This real-time testing system is effective in revealing the defect-induced breakdown of various complex film-electrode systems under a high electric field, paving the way for uncovering the breakdown mechanism and studying how to improve the capacitor's breakdown strength and energy density.

Application of artificial neural networks to the identification of weak electrical regions in large area MIM structures

Large area metal-insulator-metal (MIM) structures are prone to exhibit weak electrical regions when they are subjected to severe stress conditions. Although the root cause of this problem is hard to identify, it has been attributed to non-uniform oxide thickness, variable dielectric permittivity, correlated defect generation, and fringe effects in capacitors. In this paper, we explore the application of artificial neural networks (ANNs) to the spatial localization of such weak regions. To this end, HfO2-based MIM structures were electrically stressed with the objective of generating a large number of breakdown spots. These spots are statistically distributed both in size and location over the device area. Two-input/two-output ANNs with different number of neurons and hidden layers were assessed with the purpose of identifying the best and simplest option for detecting where the most severe damage occurs. The obtained results are compatible with previous studies based on spatial statistics techniques. The method can be applied to other systems that exhibit multiple localized failure events.

Electrical breakdown spots in metal-aluminum oxide-metal structures

Metal-aluminum oxide-metal structures are prepared by evaporation in vacuum on silicon wafers using Au, Cu or Pd on top and Al as bottom electrode. The aluminum oxide layer is deposited by electron beam evaporation with a thickness of 200 nm. The diameter of the top electrode is varied between 0.1 and 1.9 mm. Capacitance measurements reveal a relaxational dielectric permittivity as expected for aluminum oxide. Applying ramp voltages results in single breakdown spots on the electrode before the structure finally becomes conductive. The breakdown spots, as characterized with an AFM, are in the form a crater, reaching 500 nm into the silicon. However, for the smallest electrode, the calculated electrostatic energy stored in the capacitance is not high enough to form the craters.

Dual-laser-induced Breakdown Thermometry via Sound Speed Measurement：A New Procedure for Improved Spatiotemporal Resolution.

Measurement of acoustic waves from laser-induced breakdown has been developed as gas thermometry in combustion atmospheres. In the measurement, two laser-induced breakdown spots are generated and the local gas temperature between these two spots is determined through the measurement of the sound speed between them. In the previous study, it was found that the local gas breakdown can introduce notable system uncertainty, about 5% to the measured temperature. To eliminate the interference, in present work, a new measurement procedure was proposed, where two individual laser pulses with optimized firing order and delay time were employed. With the new measurement procedure, the system uncertainty caused by local gas breakdown can be largely avoided and the temporal and spatial resolutions can reach up to 0.5 ms and 10 mm, respectively. The improved thermometry, dual-laser-induced breakdown thermometry (DLIBT), was applied to measure temperatures of hot flue gases provided by a multijet burner. The measured temperatures covering the range between 1000 K and 2000 K were compared with the ones accurately obtained through the two-line atomic fluorescence (TLAF) thermometry with a measurement uncertainty of ~3%, and a very good agreement was obtained.

Impact of bilayered oxide stacks on the breakdown transients of metal–oxide–semiconductor devices: An experimental study

The role of the bilayered structure of the gate oxide on the dynamics of progressive breakdown is systematically studied on Au/Cr/HfO2/Al2O3/InGaAs metal–oxide–semiconductor stacks. Samples with bilayered oxides of 100 Å total thickness were fabricated using different Al2O3 interfacial layer thicknesses to investigate the effects of combining insulator materials with largely different electrical and thermal properties. The breakdown current growth rate dIBD/dt was captured by means of low and high bandwidth measurement setups, and the results were compared in the framework of an electromigration-based progressive breakdown model, originally derived for single-layered oxides. Experimental results show that as the interfacial layer is thicker, a clear increase is observed on the applied voltage required to obtain dIBD/dt values in the same range. However, this effect is not observed for thicknesses above 10 Å for the Al2O3 layer. This is linked to both the electrical stress distribution across the bilayered structure and to the thermal characteristics of Al2O3 that contribute to reduce the temperature of the breakdown spot. The progressive breakdown model is modified to account for these features, showing good agreement with experimental results, behavior that cannot be explained by the model considering one of the layers as already broken during progressive breakdown.

Analysis of OFF-state dynamic avalanche instability in silicon-on-insulator lateral IGBTs at low temperature

The dynamic avalanche instability in the silicon-on-insulator (SOI) lateral insulated-gate bipolar transistors (LIGBT) at low temperature is investigated. The measured results show a time-dependent collector-emitter voltage (VCE) walk event at –40 °C under the OFF-state dynamic avalanche conditions. A charge couple is proposed and TCAD simulations are performed for the mechanism revealing. It is found that the dynamic avalanche instability is closely related to the hole accumulation at the collector-side bottom, the depletion in P-type substrate (P-sub) and the transfer of the breakdown spot. The optimization strategy for the dynamic avalanche stability is drawn based on the revealed mechanism. The VCE walk can be suppressed or eliminated by preventing the dynamic expansion/shrinking behaviour of the depletion layer in substrate or satisfying an equal relationship between the vertical breakdown voltage (BVV) and the lateral breakdown voltage (BVL). With different device types, collector structures, N-drift lengths, BOX thicknesses, substrate biases and substrate types, the VCE walk events at low temperature (–40 °C) are comprehensively discussed in this paper. The VCE walk can be completely eliminated through replacing the P-sub by the N-type substrate (N-sub).

Assessing the Correlation Between Location and Size of Catastrophic Breakdown Events in High-K MIM Capacitors

The connection between the spatial location of catastrophic breakdown spots occurring in metal-insulator-metal capacitors with a high-permittivity dielectric film (HfO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> ) and their respective sizes is investigated. Large area structures (10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sup> -10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sup> μm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) are used for this correlation assessment since, for statistical considerations, a large number of spots in the same device is imperatively required. The application of ramped or constant voltage stress across the capacitor generates defects inside the dielectric that result in the formation of multiple failure sites. High power dissipation takes place locally, leaving a permanent mark on the top electrode of the device. The set of marks constitutes a point pattern with attributes that can be analyzed from a statistical viewpoint. The correlation between the spot locations and their sizes is assessed through the mark correlation function and the method of reverse conditional moments. The study reveals that for severely damaged devices, there exists a link between the spot location and size that leads to a short range departure from a complete spatial randomness (CSR) process. It is shown that the affected region around each failure site is actually larger than the visible area of the spot. A structural modification of the dielectric layer in the vicinity of the spot caused by the huge thermal effects occurring just before the microexplosion might be the reason behind this extension of the damage.

Detection of inhibitory effects in the generation of breakdown spots in HfO2-based MIM devices

In this work, the connection between the generation of catastrophic breakdown (BD) spots in metal-insulator-metal capacitors with a high-permittivity dielectric film (HfO2) and their spatial distribution was investigated. To gain insight into this issue, large area devices (104 μm2) were constant voltage stressed at high electric fields (3.5 MV/cm) with the aim of generating a large number of spots in a single device. The set of BD spots was analysed as a point pattern with attributes (their sizes) using the methods of spatial statistics. Our study reveals that beyond the visible damage on the top metal electrode, the spots exhibit soft inhibitory regions around them where the creation of new spots is less likely. The origin of these inhibitory regions is ascribed to structural modifications of the dielectric layer in the vicinity of the spots caused by the huge thermal effects occurring at the very moment of the BD event.

Characterization of the Failure Site Distribution in MIM Devices Using Zoomed Wavelet Analysis

The angular wavelet analysis is applied to the study of the spatial distribution of breakdown (BD) spots in Pt/HfO2/Pt capacitors with square and circular areas. The method is originally developed for rectangular areas, so a zoomed approach needs to be considered when the observation window does not coincide with the device area. The BD spots appear as a consequence of the application of electrical stress to the device. The stress generates defects within the dielectric film, a process that ends with the formation of a percolation path between the electrodes and the melting of the top metal layer because of the high release of energy. The BD spots have lateral sizes ranging from 1 μm to 3 μm and they appear as a point pattern that can be studied using spatial statistics methods. In this paper, we report the application of the angular wavelet method as a complementary tool for the analysis of the distribution of failure sites in large-area metal–insulator–metal (MIM) devices. The differences between considering a continuous or a discrete wavelet and the role played by the number of BD spots are also investigated.

Exploratory study and application of the angular wavelet analysis for assessing the spatial distribution of breakdown spots in Pt/HfO2/Pt structures

The angular wavelet analysis is applied for assessing the spatial distribution of breakdown spots in Pt/HfO2/Pt capacitors with areas ranging from 104 to 105 μm2. The breakdown spot lateral sizes are in the range from 1 to 3 μm, and they appear distributed on the top metal electrode as a point pattern. The spots are generated by ramped and constant voltage stresses and are the consequence of microexplosions caused by the formation of shorts spanning the dielectric film. This kind of pattern was analyzed in the past using the conventional spatial analysis tools such as intensity plots, distance histograms, pair correlation function, and nearest neighbours. Here, we show that the wavelet analysis offers an alternative and complementary method for testing whether or not the failure site distribution departs from a complete spatial randomness process in the angular domain. The effect of using different wavelet functions, such as the Haar, Sine, French top hat, Mexican hat, and Morlet, as well as the roles played by the process intensity, the location of the voltage probe, and the aspect ratio of the device, are all discussed.

Dielectric Breakdown in Chemical Vapor Deposited Hexagonal Boron Nitride.

Insulating films are essential in multiple electronic devices because they can provide essential functionalities, such as capacitance effects and electrical fields. Two-dimensional (2D) layered materials have superb electronic, physical, chemical, thermal, and optical properties, and they can be effectively used to provide additional performances, such as flexibility and transparency. 2D layered insulators are called to be essential in future electronic devices, but their reliability, degradation kinetics, and dielectric breakdown (BD) process are still not understood. In this work, the dielectric breakdown process of multilayer hexagonal boron nitride (h-BN) is analyzed on the nanoscale and on the device level, and the experimental results are studied via theoretical models. It is found that under electrical stress, local charge accumulation and charge trapping/detrapping are the onset mechanisms for dielectric BD formation. By means of conductive atomic force microscopy, the BD event was triggered at several locations on the surface of different dielectrics (SiO2, HfO2, Al2O3, multilayer h-BN, and monolayer h-BN); BD-induced hillocks rapidly appeared on the surface of all of them when the BD was reached, except in monolayer h-BN. The high thermal conductivity of h-BN combined with the one-atom-thick nature are genuine factors contributing to heat dissipation at the BD spot, which avoids self-accelerated and thermally driven catastrophic BD. These results point to monolayer h-BN as a sublime dielectric in terms of reliability, which may have important implications in future digital electronic devices.

Study of dynamics of thin liquid layer breakdown under conditions of spot heating and formation of a droplet cluster

Breakdown dynamics was studied experimentally for the horizontal layers of various liquids (ethanol, water) with the thickness of 300 μm under the conditions of spot heating from the substrate. The main stages of the process of liquid layer breakdown were determined, and time of dry spot formation was measured. Time of dry spot formation for ethanol at the heat flux of 12.6 W/cm2 was 7.85 s, and for water at the heat flux of 117 W/cm2, it was 0.13 s. It was found that for both working liquids, a residual layer appears in the region of spot heating before liquid layer breakdown. It is shown that together with the thermocapillary effect, evaporation is one of the main factors affecting dynamics of liquid layer breakdown and dry spot formation.

Electron energy dissipation model of gate dielectric progressive breakdown in n- and p-channel field effect transistors

We report the data and a model showing that the energy loss experienced by the carriers flowing through breakdown spots is the primary cause of progressive breakdown spot growth. The experiments are performed in gate dielectrics of metal-oxide-semiconductor (MOS) devices subjected to accelerated high electric field constant voltage stress under inversion conditions. The model is analytical and contains few free parameters of clear physical meaning. This is compared to a large set of data on breakdown transients at various oxide thicknesses, stress voltages, and temperatures, both in cases of n-channel and p-channel transistors and polycrystalline Si/oxynitride/Si and metal gate/high k dielectric/Si gate stacks. The basic idea is that the breakdown transient is due to the growth of one or more filaments in the dielectric promoted by electromigration driven by the energy lost by the electrons traveling through the breakdown spots. Both cases of polycrystalline Si/oxynitride/Si and metal gate/high-k dielectric/Si MOS structures are investigated. The best fit values of the model to the data, reported and discussed in the paper, consistently describe a large set of data. The case of simultaneous growth of multiple progressive breakdown spots in the same device is also discussed in detail.

Spatial analysis of failure sites in large area MIM capacitors using wavelets

The spatial distribution of failure sites in large area (104–105μm2) metal-insulator-metal (MIM) capacitors with high-K dielectric (HfO2) is investigated using angular wavelets. The failure sites are the consequence of constant or ramped electrical stress applied on the capacitors. Because of the important local thermal effects that take place during stress, the failure sites become visible as a point pattern on the top metal electrode. In case of less damaged devices, the results obtained with the wavelet variance method are consistent with an isotropic distribution of breakdown spots as expected for a Poisson point process (complete spatial randomness). On the contrary, for severely damaged devices, the method shows signs of preferred directions of degradation related to the voltage probe location. In this case, the anisotropy is confirmed by alternative spatial statistics methods such as the angular point-to-event distribution and the pair correlation function.

Degradation characteristics of oil-immersed pressboard samples induced by partial discharges under DC voltage

To investigate partial discharge-induced degradation characteristics of oil-immersed pressboard under DC voltage, an experimental platform with needle-plane oil-pressboard insulation model is established. The surface condition and DC breakdown voltage of pressboard samples after PD pretreatment are observed and measured. In addition, a fitting formula to evaluate the degradation extent of pressboard samples is proposed. The results show that under constant PD pretreatments, surface roughness of pressboard samples decreases first and increases afterwards. The changing rate is positively correlated with PD intensity. During PD pretreatment, a “whitish-colored circular region” appears at samples' surface and grows with PD intensity. For PD pretreatments of 500 pC and 1000 pC, DC breakdown voltage of pressboard samples remains unchanged. The breakdown spot has no clear carbonization nor creepage trace. For PD pretreatments of 2000 pC and 3000 pC, DC breakdown voltage decreases. The breakdown spot is obviously carbonized and creepage paths emerge around spots. Further analysis indicates that degradation of oil-immersed pressboard samples due to DC PDs is the result of motion activation process induced by electrical and thermal effects under DC electric field.

Breakdown dynamics of a horizontal evaporating liquid layer when heated locally

Breakdown of liquid layer when heated from a localized hot spot was investigated experimentally. Water and ethanol were used as working liquids with a layer thickness of 300 μm. Basic steps of the breakdown process were found and mean velocities of the dry spot formation were determined. The formation of residual layer over the hot-spot before the breakdown has been found for both liquids. The creation of a droplet cluster near the heating region is observed when using water as a working fluid. It was shown that evaporation is one of the general factors influencing the process of layer breakdown and dry spot formation as well as thermocapillary effect.

The influence of electrode contact on the anodization of Pt/sol–gel derived SrTiO3 thin film/Al capacitor

When the valve metal is used as anode material, the dielectric properties of sol–gel-derived thin film are greatly enhanced. That is because anodized self-healing occurs at the interface of the anode and the thin film. An oxide insulating layer with a good dielectric property is generated. Anodization heals the defects and the breakdown spots of thin film. Other than anodized self-healing, electrode contact is also an important factor, affecting the dielectric characteristics of the thin film capacitor, which cannot be ignored, especially when both factors exist in one film device. In this study, the sol–gel-derived strontium titanate thin film is used as dielectric material in the film capacitor, and platinum and aluminum as electrode materials. We find that the Schottky barrier at the interface of platinum and strontium titanate greatly reduces the leakage current. However the interface barrier restricts the anodizing of aluminum electrode and the further enhancement of the dielectric strength of the thin film capacitor.