Merits of Batch ALD

An overview of the merits and applications of batch ALD in a vertical furnace will be presented. We address new material and process developments and throughput enhancement which are key factors for future high-volume manufacturing applications. We present ALD SbOx as a new material on the batch platform. For the workhorse ALD Al2O3 and TiN materials, experimental and simulation results demonstrate that a reduction in cycle time to <21s does not significantly compromise uniformity, resistivity and step coverage. Moreover, batch ALD offers process flexibility in the mitigation of non-idealities, such as growth inhibition, as will be discussed for thermal ALD AlN. The very low non-uniformities < 1% (1σ) for the latter process demonstrate the competitive film properties that can be achieved by batch ALD. With applications as diverse as metal gates in logic, trench capacitor electrodes, capacitor dielectrics, barrier layers and passivation films, batch ALD has firmly established itself at device manufacturers and foundry sites with significant prospects for emerging markets.

Shortening cycle time and maximising output are the major concerns of highly competitive industry. In this paper, an Adjusted X-Factor Contribution (AXFC) measurement is developed, which considers batching process, un-batching process and machine failure. A general model is established to determine the X-factor contribution for all types of machines. In this model, GI/G/m queuing theory is applied to estimate the aggregated cycle time. The machine downtime variability, lot arrival variability, batching and un-batching processing are considered. Finally, the effects of system performances by improving the workstations with high utilisation and high AXFC are explored. The results showed that the cycle time and cycle time variability of products could be affected by the relative locations of high utilisation and high AXFC workstations. Furthermore, the results also revealed that reducing failure frequency of high AXFC workstation will perform as good as high utilisation workstation on cycle time improving.

The impact of lot-sizing on net profits and cycle times in the n-job, m-machine job shop with both discrete and batch processing

Concomitant consumption of glucose and lactate: A novel batch production process for CHO cells

In semiconductor manufacturing, hot lots are often introduced into the system out of marketing and business considerations for extra flexibility regarding delivery lead times. The objective of this paper is to study the scheduling policies to minimize the cycle time of hot lots in batch processes. We present three scheduling policies based on look ahead batching (LAB) and study their effect on the performance of cycle time of hot lots and other priority lots in the system. In LAB, the scheduling decisions are based on the arrival epochs of incoming lots, which are easily predictable in a computer integrated manufacturing environment. The steady state simulation results show that exploiting the knowledge of future arrivals leads to 10% reduction in the cycle time of hot lots without affecting their standard deviation and also without affecting the cycle time performance of other priority lots in batch processes at lower values of utilization level.

Determining the electric-field strength in a passive film via photo-induced electric fields

Bismuth telluride (Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> ) and its alloys are the most commonly used materials for thermoelectric devices. In this paper, the fluxless bonding process was developed to bond Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> chips to alumina substrates for high temperature applications. The silver-indium (Ag-In) system was chosen for the process development. To work with this system, the Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> chips were coated with 100 nm titanium (Ti) and 100 nm gold (Au) as barrier layer and plated with 10 μm Ag layer. The Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> samples were annealed at 250 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">°</sup> C for 200 h. No interdiffusion between Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> and Ag was detected. The Ti/Au barrier layer was not affected either. It showed exceptional step coverage on the rough Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> surface even after the annealing process. To prepare for bonding, alumina substrates with 40 nm TiW and 2.5 μm Au were plated with 60 μm Ag, followed by 5 μm In and thin Ag cap layer for oxidation prevention. The Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> chips were bonded to alumina substrates at 180 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">°</sup> C. No flux was used. The resulting void-free joint consists of five regions: Ag, (Ag), Ag <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> In, (Ag) and Ag. (Ag) is Ag-rich solid solution. The joint has a melting temperature higher than 660 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">°</sup> C. Due to the thick ductile Ag layer on alumina, the Bi <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> Te <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> chip did not break after bonding despite its significant coefficient of thermal expansion mismatch with alumina.

Copper-based alloys are widely used in the consumer electronics industry for components such as connector contacts, shielding gaskets, and terminals. Nickel silver is a general term for alloys that contain copper, nickel, and zinc. Nickel silver first became popular as a base metal for silver-plated cutlery and other silverware, notably electro-plated nickel silver, jewelery and coins. In this article, the electrochemical stability of Cu–10Ni–10Zn alloy was compared with that of pure copper and two of copper ternary alloys, namely Cu–10Al–10Zn and Cu–10Al–10Ni in synthetic sweat solution, Hank’s solution and in Ringer physiological solution. Corrosion behavior of the different electrodes was investigated using cyclic voltammetry, potentiodynamic polarization, and impedance spectroscopy techniques. The CVs showed peaks concerning the redox reactions for copper. These peaks were assigned to the formation and reduction of copper species. Furthermore, they showed that the copper oxidation was suppressed to a large extent by the presence of Ni and zinc in the alloy composition than the presence of Al and Zn or the presence of Al and Ni elements. Also, the results showed that the Cu–10Ni–10Zn alloy exhibited good corrosion resistance due to the formation of a protective passive film. The electrochemical impedance spectroscopy results indicated that either a duplex film with an inner barrier layer and an outer porous layer or a single passive layer was formed on the surface. The formation of such passive film was discussed, and the different alloy surfaces examined by scanning electron microscopy and subjected to EDAX analysis. The surface analysis has shown the participation of the different alloying elements in the passive film according to the alloy constituents. Addition of Ni decreases the rate of corrosion of the Cu, and the Cu–10Ni–10Zn was found to be the most corrosion-resistant alloy.

The stability of the barrier layers of bilayer passive films that form on metal and alloy surfaces, when in contact with oxidizing aqueous environments, is explored within the framework of the Point Defect Model (PDM) using phase-space analysis (PSA), in which the rate of growth of the barrier layer into the metal, (dL{sup +}/dt), and the barrier layer dissolution rate, (dL{sup -}/dt), are plotted simultaneously against the barrier layer thickness. A point of intersection of dL{sup -}/dt with dL{sup +}/dt indicates the existence of a metastable barrier layer with a steady state thickness greater than zero. If dL{sup -}/dt > (dL{sup +}/dt){sub L=0}, where the latter quantity is the barrier layer growth rate at zero barrier layer thickness, the barrier layer cannot exist, even as a metastable phase, as the resulting thickness would be negative. Under these conditions, the surface is depassivated and the metal may corrode at a rapid rate. Depassivation may result from a change in the oxidation state of the cation upon dissolution of the barrier layer, such that the dissolution rate becomes highly potential dependent (as in the case of transpassive dissolution of chromium-containing alloys, for example, in which the reaction Cr{sub 2}O{sub 3} + 5H{sub 2}O {yields} 2CrO{sub 4}{sup 2-} + 10H {sup +} + 6e{sup -} results in the destruction of the film), or by the action of some solution-phase species (e.g., H{sup +}, Cl{sup -}) that enhances the dissolution rate to the extent that dL{sup -}/dt > (dL{sup +}/dt){sub L=0}. The boundaries for depassivation may be plotted in potential-pH space to develop Kinetic Stability Diagrams (KSDs) as alternatives to the classical Pourbaix diagrams for describing the conditions under which metals or alloys exist in contact with an aqueous environment. The advantage of KSDs is that they provide kinetic descriptions of the state of a metal or alloy that is in much closer concert with the kinetic phenomenon of passivity and depassivation than are equilibrium thermodynamic diagrams. Thus, KSDs more accurately account for the limits of passivity in highly acidic systems, where acid depassivation occurs, and at high potentials, where transition to the transpassive state may occur in some systems. In any event, phase space analysis of the PDM permits specification of the conditions over which reactive metals will remain passive in contact with aqueous systems and hence of the conditions that must be met for the existence of our metals-based civilization.

Batch processes are relevant to a wide variety of industries in chemical processes. In batch operations, water sources are almost not directly reused/recycled in process sinks without considering time constraints and storage tanks. However, storage tanks are usually expensive and thus a cost-effective water system has to be synthesized. Rescheduling the water network can contribute to reducing the cost of storage tanks by reducing their number and capacity. In the current research work, a goal programming is used to reschedule the water network in batch processes considering the time and storage tanks. A Mixed Integer Non-Linear Program model is introduced using the Lingo optimization program. This model is used to optimize multiple objectives, which are freshwater usage, wastewater discharge, the number and capacity of tanks, the degree of shifting streams, and the total cost of the water network. Three case studies are presented in this study to demonstrate the effectiveness of the proposed procedure, considering both single and multi-contaminants problems. The results of the first case study show a reduction in the network cost and the freshwater flowrate by 26.4% and 42%, respectively. Regarding the rescheduled water network results of the second case study, the cost is reduced by 24.6% and the freshwater flowrate is decreased by 21.8% with no requirement of storage tanks. The third case study highlights the model’s applicability to multi-contaminants problem, revealing a 15.1% cost reduction and a 25.7% decrease in freshwater flow.

The Czochralski (CZ) crystallization process is used to produce monocrystalline silicon. Monocrystalline silicon is used in solar cell wafers and in computers and electronics. The CZ process is a batch process, where multicrystalline silicon is melted in a crucible and later solidifies on a monocrystalline seed crystal. The crucible is heated using a heating element where the power is manipulated using a triode for alternating current (TRIAC). As the electric resistance of the heating element increases by increased temperature, there are significant dynamics from the TRIAC input signal (control system output) to the actual (measured) heating element power. The present paper focuses on empirical modeling of these dynamics. The modeling is based on a dataset logged from a real-life CZ process. Initially the dataset is preprocessed by detrending and handling outliers. Next, linear ARX, ARMAX, and output error (OE) models are identfied. As the linear models do not fully explain the process' behavior, nonlinear system identification is applied. The Hammerstein-Wiener (HW) model structure is chosen. The final model identified is a Hammerstein model, i.e. a HW model with nonlinearity at the input, but not at the output. This model has only one more identified parameter than the linear OE model, but still improves the optimization criterion (mean squared ballistic simulation errors) by a factor of six. As there is no nonlinearity at the output, the dynamics from the prediction error to the model output are linear, which allows a noise model to be added. Comparison of a Hammerstein model with noise model and the linear ARMAX model, both optimized for mean squared one-step-ahead prediction errors, shows that this optimization criterion is 42% lower for the Hammerstein model. Minimizing the number of parameters to be identified has been an important consideration throughout the modeling work.

Detailed potential/pH stability diagrams for the lithium‐water system at 25°C have been derived as part of a program to describe the electrochemistry of the lithium anode in aqueous alkaline environments. The diagrams include lithium hydride as a stable solid phase at sufficiently negative potentials, and it is postulated that the passive film on Li in contact with concentrated alkaline solutions, under open‐circuit conditions, comprises a LiH barrier layer covered with a hydrated LiOH outer layer. The postulated existence of as a barrier layer, under open‐circuit conditions, is inconsistent with the thermodynamic data for metallic lithium in contact with concentrated LiOH and LiOH + KOH solutions. © 1999 The Electrochemical Society. All rights reserved.

The growth kinetics of the surface film formed upon the multi-principal element alloy AlTiVCr under anodic polarisation in 0.6 M NaCl was investigated using atomic emission spectroelectrochemistry (AESEC). The AESEC charge balance analysis revealed that thickness of the barrier layer of the passive film upon the alloy: (1) increases linearly with the increase in anodic potential during potentiodynamic polarisation, and (2) increases logarithmically with exposure time during potentiostatic polarisation. This is consistent with the assumptions of the point defect model, despite the film being a multi-oxide film with transpassive dissolution of V and Cr. The X-ray photoelectron spectroscopy (XPS) analysis suggested that the growth of the film was predominantly due to TiO2 during anodic polarisation. The electric field was found to decrease with enrichment of TiO2 in the barrier layer. The Mott-Schottky analysis revealed that the diffusivity of oxygen vacancies increased with the increase in fraction of TiO2 in the film, which subsequently led to the increase in the growth rate of the barrier layer during transpassive dissolution. The present work is a discrete effort towards understanding the growth behaviour of the passive film experiencing complex and competing interfacial electrochemical processes, upon a multi-principal element alloy.

Passivity breakdown is the precursor to most forms of localized corrosion, including pitting, stress corrosion cracking, corrosion fatigue, and crevice corrosion, and also results in the phenomenon of depassivation, including electrochemical polishing and trans-passive dissolution, corresponding to passivity breakdown over macroscopic dimensions. In this OLIN Palladium Medal address, I will review my work in this area in terms of developing the Point Defect Model (PDM) for passivity and passivity breakdown and will identify principal challenges, as I see them. Starting with the rate equation for barrier layer growth, as predicted by the PDM, and by using phase-space analysis, I derive the condition that must be met for passivity to exist; the rate of growth of the barrier layer at the metal/barrier layer interface, due to the generation of oxygen vacancies at zero barrier layer thickness, must exceed the dissolution rate of the barrier layer. If this condition is met, the barrier layer may exist as a meta-stable phase and passivate the surface. In this regard, it is important to note that passivity is a kinetic state, not a equilibrium thermodynamic phenomenon, as has been so often assumed in the past. This fact has been used to resolve Faraday’s famous paradox arising from his iron-in-nitric acid experiments of 1833. Violation of this condition implies that the barrier layer cannot exist on the surface, even as a meta-stable phase and, hence, that the surface depassivates (loss of the barrier layer). If depassivation occurs over a macroscopic area, this phenomenon accounts for acid depassivation, transpassive dissolution, resistive depassivation, and electrochemical polishing. However, if depassivation occurs on a microscopic area, localized corrosion phenomena, including pitting, stress corrosion cracking, corrosion fatigue, and crevice corrosion occur, resulting in rapid penetration of a cavity into the metal. This unified theory of corrosion (UTC) accounts for all known forms of corrosion, except for hydrogen embrittlement, which lies outside of the realm of the PDM. After having established the basis of the UTC, I will then proceed to discuss the mechanism by which both general and localized depassivation occurs, which has been identified as being due to cation vacancy condensation at the m/bl interface, such that the barrier layer separates from the metal and film growth into the metal via oxygen vacancy generation cannot occur. This condition assures the depassivation inequality stated above, because the rate of growth of the barrier layer into the metal is reduced to zero and because all oxides exhibit a finite dissolution rate. The lecture will then proceed to show how the UTC predicts the critical breakdown voltage and induction time for a single breakdown site, as a function of the chloride activity, pH, and potential. It will then be shown that, by postulating that the breakdown sites are normally distributed with respect to the cation vacancy diffusivity, it is possible to derive the distribution functions for the breakdown voltage and the induction time for a large population of potential breakdown sites on the surface and to identify the critical characteristic of a breakdown site (high cation vacancy diffusivity). From there, I will demonstrate a variety of phenomena, including photo-inhibition of passivity breakdown, the impact of other halides (F-, Br-, I-), anion inhibition via competitive absorption into surface oxygen vacancies, the role of the precipitated outer layer, and the predicted impact of alloying elements on passivity breakdown. Where possible and appropriate, these predictions will be compared with experiment. As expected the PDM contains many model parameters that are poorly-known. However, values for all of the parameters may be obtained by optimizing the expressions for the impedance of the barrier layer as a function of frequency and film formation potential, as derived from the PDM, on experimental electrochemical impedance spectroscopic (EIS) data and optimization of the expressions for passivity breakdown (potential and induction time) on the corresponding experimental data. Finally, I will demonstrate how the PDM has led to the development of a deterministic protocol, Damage Function Analysis (DFA), for predicting the evolution of localized corrosion damage on metal and alloy surfaces, as a system evolves along a pre-determined corrosion evolutionary path, which is defined as the path taken by the system in terms of those properties of the system (temperature, potential, pH, [Cl-], etc) that have a significant impact of the damage accumulation rate. Time permitting, I will demonstrate how DFA may be used to predict the evolution of general and localized corrosion damage in practical systems.

Thermal ALD deposition of ZrO2 and HfO2 films has been performed in the ASM A412(TM) 300mm vertical furnace using bis-cyclopentadienyl precursors. Due to the long precursor residence time and high surface area to be covered in a batch reactor, the thermal stability and volatility of the precursor is of critical importance. Evaluation of the thermal stability of the precursor was done by depositing on high surface area silica powder and then NMR studies were employed to evaluate precursor reactivity. The first deposition studies were in single wafer reactors to assess the viability for scale-up to batch processing. The deposition studies performed have shown that the family of bis-cyclopentadienyl precursors has demonstrated sufficient thermal stability to successfully achieve ALD films which meet deposition requirements in batch reactors for capacitor dielectrics in memory devices.