Semiconductor Process Control Research Articles

The Use of Artificial Intelligence and Machine Learning in Soft X-ray Technology for Advanced Semiconductor Process Development and Control

The Use of Artificial Intelligence and Machine Learning in Soft X-ray Technology for Advanced Semiconductor Process Development and Control

Deep reinforcement learning framework for end-to-end semiconductor process control

Deep reinforcement learning framework for end-to-end semiconductor process control

FEI Helios NanoLab 400S FIB-SEM

The FEI Helios NanoLab400S FIB-SEM is one of the world's most advanced DualBeamTM focused ion beam (FIB) platforms for transmission electron microscopy (TEM) sample preparation, scanning electron microscopy (SEM) imaging and analysis in semiconductor failure analysis, process development and process control. The FEI Helios NanoLab400S FIB-SEM combines an ElstarTM electron column for high-resolution and high-contrast imaging with a high-performance SidewinderM ion column for fast and precise cross sectioning. The FEI Helios NanoLab M 400S is optimised for high throughput high-resolution S/TEM sample preparation, SEM imaging and energy dispersive X-ray analysis. Its exclusive FlipStageTM and in situ STEM detector can flip from sample preparation to STEM imaging in seconds without breaking vacuum or exposing the sample to the environment. Platinum gas chemistry is the preferred metal deposition when a high deposition rate and precision of the deposition are required. Carbon deposition can be chosen as well. The system additionally allows for spatially resolved compositional analysis using the attached EDAX Genesis XM 4i X-ray microanalysis system.

반도체 공정에서의 APC 기법 및 이상감지 및 분류 시스템

Traditional semiconductor process control has been performed through statistical process control techniques in a constant process-recipe conditions. However, the complexity of the interior of the etching apparatus plasma physics, quantitative modeling of process conditions due to the many difficult features constraints apply simple SISO control scheme. The introduction of the Advanced Process Control (APC) as a way to overcome the limits has been using the APC process control methodology run-to-run, wafer-to-wafer, or the yield of the semiconductor manufacturing process to the real-time process control, performance, it is possible to improve production. In addition, it is possible to establish a hierarchical structure of the process control made by the process control unit and associated algorithms and etching apparatus, the process unit, the overall process. In this study, the research focused on the methodology and monitoring improvements in performance needed to consider the process management of future developments in the semiconductor manufacturing process in accordance with the age of the APC analysis in real applications of the semiconductor manufacturing process and process fault diagnosis and control techniques in progress.

Integrating the Fault Detection Method and Run-to-Run Control for Improving Semiconductor Process Control

Abstract Recently, detailed investigations of various ‘run-to-run’ (R2R) control schemes for semiconductor manufacturing have been conducted. However, the R2R control scheme has a major problem that needs to be solved, namely, how to detect and differentiate the faults in the control system. In view of this consideration, the objective of this research was to construct a novel module for fault diagnosis so that the R2R control system can operate smoothly. In this study, a fault diagnostic module is added to the multiple-input multiple-output (MIMO) R2R self-tuning control system. The proposed system integrates the R2R selftuning control, weighted sum of squared residuals (WSSR), and joint angle analysis (JAA) methods to detect the system faults, and executes the process control. First, the WSSR is applied to detect whether the system has generated the fault characteristics. If the process is judged to exhibit an obvious fault, the system then uses JAA to define the faulty controllable variable. Finally, a critical step, the chemical mechanical planarization (CMP) in semiconductor manufacturing, is used to illustrate the procedure of fault diagnosis in order to verify the feasibility in practical application.

Properties of EWMA Controllers With Gain Adaptation

Exponentially weighted moving average (EWMA) controllers are the most commonly used run-to-run controllers in the semiconductor industry. Using a linear model, an controller can be implemented in two different ways: either process gain or process intercept can be updated using statistics at each run. The most commonly used controller formulation is to keep the process gain as its off-line estimate and update the intercept term at each run. We term this formulation as EWMA controller with intercept adaptation (EWMA-I), and its properties have been extensively studied and well understood. We term the other implementation, i.e., keeping the intercept term as its off-line estimate and updating the process gain at each run, as EWMA with gain adaptation (EWMA-G). Despite the fact that gain variation and adaptation is typical in the semiconductor industry, little research has been done to investigate the properties of an EWMA-G controller, such as its stability and sensitivity. In this work, the stability and sensitivity properties of an EWMA-G controller are investigated and compared with those of an EWMA-I controller. Both stationary and drifting processes are considered. In addition, the expressions of the process outputs are derived and the output variances for stochastic processes are evaluated. Simulation examples are given to illustrate these properties and the relevance of these properties to semiconductor process control is discussed. The analysis results will provide some useful guidance on the industrial applications of the controllers.

Through-focus scanning-optical-microscope imaging method for nanoscale dimensional analysis

We present a novel optical technique that produces nanometer dimensional measurement sensitivity using a conventional bright-field optical microscope, by analyzing through-focus scanning-optical-microscope images obtained at different focus positions. In principle, this technique can be used to identify which dimension is changing between two nanosized targets and to determine the dimension using a library-matching method. This methodology has potential utility for a wide range of target geometries and application areas, including nanometrology, nanomanufacturing, semiconductor process control, and biotechnology.

Investigations of semiconductor devices using SIMS; diffusion, contamination, process control

We have surveyed 22,155 analyses issues to know the portion of surface analysis at the total analyses activities. According to the survey result, the contribution of SIMS in the total analyses issues was about 7%. The portions of semiconductor process control, composition and contamination in the SIMS analyses issues are 25%, 29% and 16%, respectively. In this article, some examples of the semiconductor device process control, identification of contaminants, and failure analyses have been reviewed. The behavior of H, O, and Ti at the Pt/Ti/GaInZnO interfaces and their influences on the electrical property of thin film transistor are demonstrated. Also discolor issues including organic material contamination problem on Au pad are discussed in detail.

Development of a novel tool for semiconductor process control

A collaborative project between the Surrey Ion Beam Centre and the positron beam group at the University of Bath is developing a bench-top positron tool suitable for use in the semiconductor industry. The technique's non-destructive nature, coupled with its high sensitivity to defects, makes it a potentially ideal method for detecting process problems at an early stage. Measurements on the existing laboratory-based system have shown, for example, a high sensitivity to variations in parameters such as temperature in the growth of epitaxial layers. An instrument suitable for use in the fabrication environment is described, together with diagnostic studies of PECVD-deposited SiN layers, of thicknesses in the range 21–133 nm, performed in conjunction with ellipsometric and RBS measurements. The technique is sensitive to the atomic composition of the SiN epilayers and, in conjunction with ellipsometry, is able to measure the density of the layers and to quantify the densification on annealing at 900°C.

A 1024-element bulk-micromachined thermopile infrared imaging array

This paper reports on a bulk-micromachined silicon uncooled thermal imager intended for use in automated semiconductor process control. The device has a responsivity of 15 V/W, a thermal time constant of 1 ms, and a D* of 1.6×10 7 cm Hz 1/2/W. It is organized as a 32×32-element area array with on-chip address generation, multiplexing, and self-test circuitry. The overall chip size is 16 mm×16 mm, with a 12 mm×12 mm imaging area. Each detector consists of 32 n–p polysilicon thermocouples supported on a dielectric window and is micromachined using a novel combined front-undercut and back-etched process. An infrared camera system using the imager is also described.

The use of inductively coupled plasma mass spectrometry to provide an absolute measurement of surface coverage, and comparison with the quartz crystal microbalance

It is often difficult to obtain an absolute measure of the quantity of an ultrathin layer of an element deposited onto a dissimilar solid surface, even though this is an important parameter for many surface studies. Auger and X-ray photoelectron spectroscopies provide only a relative measure of the quantity of a deposit, and nuclear or ion-scattering spectroscopies are limited in the adsorbate/substrate pairs that can be measured and often require expensive accelerators that are not readily available. Inductively-coupled-plasma mass spectroscopy (ICPMS) has been used as an ex-situ method to determine the absolute coverage at sub-monolayer levels in semiconductor process control. However, it does not seem to have been used to date for determining surface coverage of adsorbates for other surface-science applications, even though it can do this routinely with an accuracy of few percent. This paper uses ICPMS to determine the coverage of a sub-monolayer of antimony on gold, and shows that the results are consistent with high-precision measurements taken with a quartz crystal microbalance on the same sample.

Philips analytical — a one stop characterization shop

Phillips Analytical is a leading supplier of characterization and metrology tools for industrial process control. The company offers a wide range of products from high volume automated equipment, to low end mini systems, to be used in silicon semiconductor process control and research and development environments. The company has truly become a one stop shop for industrial characterization needs.

Noncontacting measurement of thickness of thin titanium silicide films using spectroscopic ellipsometer

Spectroscopic ellipsometry has gained increasing attention in semiconductor process control because the technique is nondestructive and noncontacting. This paper demonstrates the capability of spectroscopic ellipsometer to measure the thickness of conducting thin films of titanium silicide. Unlike cross section TEM measurement, this technique does not involve elaborate process of sample preparation. This technique does not require calibration and is used to determine thickness of silicide films from few tens of angstrom up to tens of nanometer. The thickness of titanium silicide film measured at a single point, using spectroscopic ellipsometer and TEM analysis differs by only 4%.

Thermal Radiation Absorption in Doped Semiconductors Due to Direct Intersubband Transitions

This work proposes an engineering model for thermal radiation absorption due to direct intersubband transitions in doped semiconductors, which are excitations of bound electrons in the infrared spectral region. An existing quantum-mechanical approach is to model these transitions as a continuum of damped harmonic oscillators. This study modifies this approach to yield a more effective method for determining the optical constants of doped semiconductors. The room-temperature absorptance spectra of p-type GaAs samples with dopant concentrations from 8 × 1017 to 1 × 1020 cm-3 are measured using a Fourier transform–infrared spectrometer in the spectral region from 1.5 to 25 μm. The fitting of these spectra using the proposed model combined with the Drude and Lorentz models provides the optical constants and the dependence of the adjustable parameters on dopant concentration. The measured and fitted spectra agree closely. Potential applications of the model are semiconductor process control and infrared detector design.

Application of a Neural Manufacturing Concept to Process Modeling, Monitoring and Control

AbstractThe cost of a fabrication line, such as one in a semiconductor house, has increased dramatically over the years, and it is possibly already past the point that some new start-up company can have sufficient capital to build a new fabrication line. Such capital-intensive manufacturing needs better utilization of resources and management of equipment to maximize its productivity. In order to maximize the return from such a capital-intensive manufacturing line, we need to work on the following: 1) increasing the yield, 2) enhancing the flexibility of the fabrication line, 3) improving quality, and finally 4) minimizing the down time of the processing equipment. Because of the significant advances now made in the fields of artificial neural networks, fuzzy logic, machine learning and genetic algorithms, we advocate the use of these new tools in manufacturing. We term the applications to manufacturing of these and other such tools that mimic human intelligence neural manufacturing. This paper describes the effort at the Lawrence Livermore National Laboratory (LLNL) [1] to use artificial neural networks to address certain semiconductor process modeling, monitoring and control questions.

Application of a Neural Manufacturing Concept to Process Modeling, Monitoring and Control

ABSTRACTThe cost of a fabrication line, such as one in a semiconductor house, has increased dramatically over the years, and it is possibly already past the point that some new start-up company can have sufficient capital to build a new fabrication line. Such capital-intensive manufacturing needs better utilization of resources and management of equipment to maximize its productivity. In order to maximize the return from such a capital-intensive manufacturing line, we need to work on the following: 1) increasing the yield, 2) enhancing the flexibility of the fabrication line, 3) improving quality, and finally 4) minimizing the down time of the processing equipment. Because of the significant advances now made in the fields of artificial neural networks, fuzzy logic, machine learning and genetic algorithms, we advocate the use of these new tools in manufacturing. We term the applications to manufacturing of these and other such tools that mimic human intelligence neural manufacturing. This paper describes the effort at the Lawrence Livermore National Laboratory (LLNL) [1] to use artificial neural networks to address certain semiconductor process modeling, monitoring and control questions.*

Application of ETV-ICPMS in semiconductor process control

Submicron semiconductor manufacturing requires ultra-clean processes and materials to achieve high product yields. It is demonstrated that electrothermal evaporation (ETV) in a graphite furnace coupled with ICPMS offers a new possibility for a fast simultaneous analysis of eight elements with detection limits below 0.2 ng/g in conc. hydrofluoric acid and buffered oxide etch (ammonium fluoride/hydrogen fluoride mixture). ETV-ICPMS also comprises significant improvements in the analysis of metal contamination on silicon wafer surfaces with respect to currently used methods. The contaminants on the surface are usually analyzed by total reflexion X-ray fluorescence spectrometry (TXRF) or dissolved by HF vapour (vapour phase decomposition; VPD) or a mixture of hydrofluoric acid and hydrogen peroxide (droplet surface etching, DSE) and analyzed by GFAA or TXRF. ETV-ICPMS combines the advantages of both analytical methods: the multielemental advantage of TXRF and the possibility to analyze light elements like Al, Mg, Na which may not be analyzed by TXRF. With VPD/DSE-ETV-ICPMS detection limits between 0.2 and 2×109 atoms cm−2 on a 6″ wafer have been achieved in a simultaneous analysis of eight elements. The main advantage of ETV-ICPMS versus conventional ICPMS in both applications — chemical and surface analysis — is its capability to analyze Fe in the sub-ng/g range. As Fe is one of the most important impurities in semiconductor manufacturing ETV-ICPMS is much more useful for semiconductor applications than low-resolution ICPMS. For the present application potassium iodide was used as a modifier. It enhances the sensitivity by a factor of 3–4 and improves the reproducibility significantly.

Ultratrace analysis of volatile organic compounds in semiconductor industry

Recently it has been demonstrated that beside particles or metal impurities also organic contaminants have detrimental effects in semiconductor production. Ion Mobility Spectrometry (IMS) is a rather new, ultrasensitive method for analysis of organic compounds. After a short description of the method typical sources of organic contaminants are mentioned. Examples for IMS applications to semiconductor production and process control are given.

An organization and interface for sensor-driven semiconductor process control systems

An architecture and sensor bus interface for use in generic bus-organized sensor-driven process control systems where high accuracy and high reliability are important are presented. The organization permits 12-b digital sensor data to be communicated to the host processor over bidirectional parallel or serial data buses which include parity checking. Remote testing of the sensor can be implemented along with PROM-based digital compensation of primary and cross-parameter sensitivities. The sensor bus interface is microprocessor controlled, has been implemented using discrete commercial components, and has been designed and simulated in monolithic form. Designs using 3- mu m single-metal double-poly and 1- mu m double-metal single-poly CMOS technologies are contrasted. Monolithic node integration appears well within the range of current technology, with power levels of less than 75 mW and die areas below 40 mm/sup 2/. The organization makes sensor interrogations appear much like the memory accesses, yielding very fast response times to the host.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Spatially Resolved Ellipsometry for Semiconductor Process Control: Application to GaInAs MIS Structures

A new assessment technique, a spectroscopic ellipsometry system with high lateral resolution (10×10 µm spot), is introduced. Its high surface sensitivity is used at each step of a typical technological process, the fabrication of GaInAs metal-insulator-semiconductor structures, namely oxide removal, etching, and dielectric deposition. The measured non-uniformity of the Si3N4/GaInAs interface quality over the wafer is directly correlated to the electrical characteristics of the devices. This is a direct proof of the influence of the native oxide at this interface upon the electrical drift of MIS structures.