Semiconductor Manufacturing Process Research Articles

Clustering partially masked images with modified Radon transformation

We construct a new procedure for clustering partially masked images from wafer bin maps (WBMs). A WBM is an image of bin codes from circuit probe tests on a wafer after a semiconductor manufacturing process. Certain defect patterns of WBMs provide engineers with information to isolate manufacturing problems. The engineers prefer to see the identified defect patterns in clusters. Then they may match each cluster to a well-known defect type or they could easily see if there are new defect patterns. However, the WBMs may be masked at a large area due to the sequential circuit probe tests for bin codes. WBMs may also have various proportions of noise. There are two major contributions of this research. First, our method of locally iterative interpolation for the image reconstruction of masked WBMs will recover certain parts of the original patterns. Second, we proposed a new modified Radon transformation to extract features for better clustering of partly masked WBMs. We demonstrated our procedure on synthetic WBMs with various defect pattern types and levels of masking through a simulation study. The image patterns of WBMs during simulation were observed from a specific kind of real wafer product.

Photonic contact thermometry using silicon ring resonators and tuneable laser-based spectroscopy

Abstract Photonic sensors offer the possibility of purely optical measurement in contact thermometry. In this work, silicon-based ring resonators were used for this purpose. These can be manufactured with a high degree of reproducibility and uniformity due to the established semiconductor manufacturing process. For the precise characterisation of these photonic sensors, a measurement setup was developed which allows laser-based spectroscopy around 1550 nm and stable temperature control from 5 °C to 95 °C. This was characterised in detail and the resulting uncertainty influences of both the measuring set-up and the data processing were quantified. The determined temperature stability at 20 °C is better than 0.51 mK for the typical acquisition time of 10 s for a 100 nm spectrum. For a measurement of >24 h at 30 °C a standard deviation of 2.6 mK could be achieved. A hydrogen cyanide reference gas cell was used for traceable in-situ correction of the wavelength. The determined correction function has a typical uncertainty of 0.6 pm. The resonance peaks of the ring resonators showed a high optical quality of 157 000 in the average with a filter depth of up to 20 dB in the wavelength range from 1525 nm to 1565 nm. When comparing different methods for the determination of the central wavelength of the resonance peaks, an uncertainty of 0.3 pm could be identified. A temperature-dependent shift of the resonance peaks of approx. 72 pm/K was determined. This temperature sensitivity leads together with the analysed uncertainty contributions to a repeatability of better than 10 mK in the analysed temperature range from 10 °C to 90 °C.

Dielectrophoretic Trapping for Nanoparticles, High-Molecule-Weight DNA, and SYBR Gold Using Polyimide-Based Printed Circuit Board

Microelectrodes for electrokinetic detection have been employed to manipulate and collect nanomaterials and biomaterials. However, the process of microelectrode patterning has been researched in the semiconductor fabrication process, which is expensive and time-consuming. Here, we demonstrate dielectrophoretic trapping of 200 nm fluorescent polystyrene nanoparticles and λ-deoxyribonucleic acid with SYBR gold using dielectrophoretic trapping onto an interdigitated circular electrode (ICE). For dielectrophoretic trapping, an ICE is designed and an ordered flexible printed circuit board (FPCB), possessing a non-porous polyimide substrate, is constructed. The ICE FPCB could identify the dielectrophoretic trapping conditions and simulate them by the finite element method. The fluorescent molecules in the ICE FPCB are dielectrophoretically trapped at the finger edges of the ICE when a peak-to-peak voltage of 12 V is applied at 6 kHz for 15-20 min. Moreover, monitoring of the dielectrophoretic trapping effect of SYBR gold eliminated the false negative detection error.

Wafer map defect recognition based on deep transfer learning-based densely connected convolutional network and deep forest

Due to the complexity and dynamics of the semiconductor manufacturing processes, wafer maps will present various defect patterns caused by various process faults. Identification of those defect patterns on wafer maps can help operators in finding out root-causes of abnormal processes, and then ensures that the manufacturing process is restored to the normal state as soon as possible. This paper proposes a wafer map defect recognition (WMDR) model based on integration of deep transfer learning and deep forest. Firstly, we transfer the network weight parameters of ImageNet to the convolutional neural network (CNN) (i.e., densely connected convolutional network (DenseNet)) and redesign the classification layer. This reduces the training time and then improves feature learning performance of DenseNet. Moreover, the transfer learning-based feature learning is able to solve class imbalance of wafer defect patterns. Finally, deep forest is utilized to identify the wafer defect pattern based on the abstract features from the wafer maps extracted by DenseNet. The experimental results on an industrial case show that the method can effectively improve WMDR performance and outperforms those well-known CNNs and other typical classifiers.

Adversarial Defect Detection in Semiconductor Manufacturing Process

Detecting defects in the inspection stage of semiconductor manufacturing process is a crucial task to improve yield and productivity as well as wafer quality. Recent Advances in semiconductor process technology have greatly increased the transistor density. As a result, an increasingly high number of defects inevitably emerge and we need a more accurate and efficient detection method to manage them. In this paper, we propose a deep-learning-based defect detection model to expedite the process. It adopts an adversarial network architecture of conditional GAN. The discriminator of an adversarial network architecture helps the detection model learn to detect and classify defects accurately. The high performance is achieved by using Focal Loss, PixelGAN and multi-scale level features, which is shown to be better than the baseline model, CenterNet, when tested for a real industrial dataset.

Adaptive Cautious Regularized Run-to-Run Controller for Lithography Process

Photolithography is the bottleneck for quality improvement in semiconductor manufacturing processes. The decreasing critical dimensions of the semiconductor product requires more effective run-to-run control technology. Currently, Exponential Weighted Moving Average (EWMA) control scheme is widely used in the overlay control of lithography processes. In this article, we address three shortcomings of the current EWMA run-to-run control scheme: (i) the weight parameter λ used in the EWMA control scheme is set as a fixed value instead of adjusted adaptively according to the process changes; (ii) the conventional EWMA control scheme fails to take the model and parameter uncertainties into consideration; and (iii) the adjustable range of the control variables is not considered in the conventional EWMA control scheme. To meet tighter overlay specifications, we propose a new adaptive run-to-run control scheme to address these three limitations. The effectiveness of the new controller is validated through simulation studies.

Improved Ensemble-Learning Algorithm for Predictive Maintenance in the Manufacturing Process

Industrial Internet of Things (IIoT) technologies comprise sensors, devices, networks, and applications from the edge to the cloud. Recent advances in data communication and application using IIoT have streamlined predictive maintenance (PdM) for equipment maintenance and quality management in manufacturing processes. PdM is useful in fields such as device, facility, and total quality management. PdM based on cloud or edge computing has revolutionized smart manufacturing processes. To address quality management problems, herein, we develop a new calculation method that improves ensemble-learning algorithms with adaptive learning to make a boosted decision tree more intelligent. The algorithm predicts main PdM issues, such as product failure or unqualified manufacturing equipment, in advance, thus improving the machine-learning performance. Herein, semiconductor and blister packing machine data are used separately in manufacturing data analytics. The former data help in predicting yield failure in a semiconductor manufacturing process. The blister packing machine data are used to predict the product packaging quality. Experimental results indicate that the proposed method is accurate, with an area under a receiver operating characteristic curve exceeding 96%. Thus, the proposed method provides a practical approach for PDM in semiconductor manufacturing processes and blister packing machines.

Large‐Area Transient Conductive Films Obtained through Photonic Sintering of 2D Materials

AbstractTransient electronics that can chemically or physically disappear after a period of stable operation have been achieved through various metal and semiconductor materials. Despite excellent properties and wide availability of 2D materials and van der Waals (vdW) thin‐film electronics, transient films and devices based on them have rarely been reported, not to mention large‐area production with high throughput. Here, a photonic sintering approach is developed to achieve large‐area transient vdW films that may be used for bioresorbable devices. The approach can process 2D materials such as graphite, graphene, and transition metal sulfide, resulting in continuous films that are 10 × 10 cm2 in area and ≈60 to ≈440 nm in thickness on transient substrates. The influence of interfacial adhesion and sintering conditions to the thickness and the electrical properties of transient graphite films is investigated. In addition, the transient films can yield conductive patterns such as biopotential electrodes, interdigital electrodes, and resistor arrays using complementary metal–oxide–semiconductor (CMOS) fabrication processes or a two‐step sintering method. The photonic sintering method may eventually lead to large‐area transient electronics for innovative applications in healthcare, data security, and consumer electronics, and enrich the category of transient electronics through achievement of more printable transient vdW electronics.

Optimizing the energy efficiency of chiller systems in the semiconductor industry through big data analytics and an empirical study

Big data analytics is playing a more and more prominent role in the manufacturing industry as corporations attempt to utilize vast amounts of data to optimize the operation of plants and factories to gain a competitive advantage. Since the advent of Industry 4.0, also known as smart manufacturing, big data analytics, combined with expert domain knowledge, is facilitating ever-greater levels of speed and automaticity in manufacturing processes. The semiconductor industry is a fundamental driver of this transformation; moreover, due to the highly complex and energy-consuming nature of the semiconductor manufacturing process, semiconductor fabrication facilities (fabs) can also benefit greatly from incorporating big data analytics to improve production and energy efficiency. This paper developed a big data analytics framework, along with an empirical study conducted in collaboration with a semiconductor manufacturer in Taiwan, to optimize the energy efficiency of chiller systems in semiconductor fabs. Chiller systems are one of the most energy-consuming systems within a typical modern fab. The developed big data analytics framework allows production managers to ensure that chiller systems operate at an optimized level of energy efficiency under dynamically changing conditions, while fulfilling the chilling demands. Compared to the commonly-used heuristics previously employed at the fab to tune chiller system parameters, by the utilization of big data analytics, it is shown that fabs can achieve substantial energy savings, greater than 12%. The developed framework and the lessons learned from the empirical study are not only generalizable but also useful for practitioners who are interested in applying big data analytics to optimize the performance of other equipment systems in fabs.

Manufacturable 32-Channel Cochlear Electrode Array and Preliminary Assessment of Its Feasibility for Clinical Use.

(1) Background: In this study, we introduce a manufacturable 32-channel cochlear electrode array. In contrast to conventional cochlear electrode arrays manufactured by manual processes that consist of electrode-wire welding, the placement of each electrode, and silicone molding over wired structures, the proposed cochlear electrode array is manufactured by semi-automated laser micro-structuring and a mass-produced layer-by-layer silicone deposition scheme similar to the semiconductor fabrication process. (2) Methods: The proposed 32-channel electrode array has 32 electrode contacts with a length of 24 mm and 0.75 mm spacing between contacts. The width of the electrode array is 0.45 mm at its apex and 0.8 mm at its base, and it has a three-layered arrangement consisting of a 32-channel electrode layer and two 16-lead wire layers. To assess its feasibility, we conducted an electrochemical evaluation, stiffness measurements, and insertion force measurements. (3) Results: The electrochemical impedance and charge storage capacity are 3.11 ± 0.89 kOhm at 1 kHz and 5.09 mC/cm, respectively. The V/H ratio, which indicates how large the vertical stiffness is compared to the horizontal stiffness, is 1.26. The insertion force is 17.4 mN at 8 mm from the round window, and the maximum extraction force is 61.4 mN. (4) Conclusions: The results of the preliminary feasibility assessment of the proposed 32-channel cochlear electrode array are presented. After further assessments are performed, a 32-channel cochlear implant system consisting of the proposed 32-channel electrode array, 32-channel neural stimulation and recording IC, titanium-based hermetic package, and sound processor with wireless power and signal transmission coil will be completed.

Treatment of Azole-Containing Industrial Wastewater by the Fenton Process

Some semiconductor fabrication processes generate high-strength wastewater that may contain high concentrations of azoles, amines, hydrogen peroxide, organic, and inorganic co-contaminants, making the treatment of this wastewater challenging. In this study, the Fenton process was utilized for the treatment of 53 mM pyrazole and 34 mM 2-(2-aminoethoxy) ethanol (known as diglycolamine, DGA) in a lab-prepared aqueous mixture containing 3.5 M hydrogen peroxide and 16 mM inorganic fluoride. The effects of operational variables for the Fenton process, such as temperature (10, 18, or 25 °C), iron dosing (32.3, 37.3, or 74.5 mM), and pH (2.5, 3.0, or 3.5), on the degradation rates were investigated. The chosen variables were then used to treat wastewater from a semiconductor fabrication facility. The Fenton process was effective in treating both the lab-prepared mixture and semiconductor industrial wastewater. The degradation of pyrazole and DGA yielded a range of byproducts including inorganic ions and organic acids.

Fluorocarbon Rubber-Based Inert Dry Adhesive for Applications under Harsh Conditions

Artificial dry adhesives have been developed based on materials and production methods by many researchers. Nowadays, extensive research has attempted to apply dry adhesives to various industrial fields, such as glass transportation systems, with admirable performance. However, most artificial dry adhesives consist of polymers such as thermosetting polymers, ultraviolet curable polymers, and water-based polymers. These polymers exhibit low thermal resistance and outgassing under high-vacuum conditions. Hence, the usage of dry adhesives under high-temperature and vacuum conditions, such as semiconductor manufacturing processes, is critically limited. In this study, based on the contact mechanism, a dry adhesive for use under high-temperature and vacuum conditions is developed using fluorocarbon rubber with excellent thermal resistance and low outgassing properties. The fluorocarbon rubber dry adhesive (FDA) material shows a high pull-off strength of up to 24.1 N cm–2 and its performance is verified through various experiments. Glass transportation experiments using a robot arm and custom-built equipment are also conducted and the results confirm the feasibility of using FDA in industrial applications. The FDA is expected to be a useful alternative under high-temperature and vacuum conditions, such as those in the aerospace industry and display manufacturing processes.

Quality-Aware Resource Model Discovery

Context-aware process mining aims at extending a contemporary approach with process contexts for realistic process modeling. Regarding this discipline, there have been several attempts to combine process discovery and predictive process modeling and context information, e.g., time and cost. The focus of this paper is to develop a new method for deriving a quality-aware resource model. It first generates a resource-oriented transition system and identifies the quality-based superior and inferior cases. The quality-aware resource model is constructed by integrating these two results, and we also propose a model simplification method based on statistical analyses for better resource model visualization. This paper includes tooling support for our method, and one of the case studies on a semiconductor manufacturing process is presented to validate the usefulness of the proposed approach. We expect our work is practically applicable to a range of fields, including manufacturing and healthcare systems.

Quality-Aware Resource Model Discovery

Context-aware process mining aims at extending a contemporary approach with process contexts for realistic process modeling. Regarding this discipline, there have been several attempts to combine process discovery and predictive process modeling and context information, e.g., time and cost. The focus of this paper is to develop a new method for deriving a quality-aware resource model. It first generates a resource-oriented transition system and identifies the quality-based superior and inferior cases. The quality-aware resource model is constructed by integrating these two results, and we also propose a model simplification method based on statistical analyses for better resource model visualization. This paper includes tooling support for our method, and one of the case studies on a semiconductor manufacturing process is presented to validate the usefulness of the proposed approach. We expect our work is practically applicable to a range of fields, including manufacturing and healthcare systems.

Fault detection based on ICA-GLR for non-Gaussian industrial processes

ABSTRACT As the growth of Industry 4.0, online fault detection plays a crucial role in ensuring the manufacturing quality. Generally, the fault detection methods can be classified into model-based and data-driven methods. There are advantages/disadvantages between two methods. In this study, we integrated both methods in order to develop an efficient fault detection method for non-Gaussian industrial processes. The data-driven method, independent component analysis (ICA) is used to extract non-Gaussian information and dimensionality reduction. Meanwhile, the model-based method, generalized likelihood ratio (GLR) test is adopted as the charting statistic. The proposed ICA-GLR method has advantages of 1) detecting a wide range of process changes, 2) estimating the change points and 3) needless prior parameters to be specified by practitioner. The efficiency of the proposed ICA-GLR fault detection method will be verified via implementing one simulated non-Gaussian process and two real manufacturing processes: Tennessee Eastman process and semiconductor manufacturing process. Results demonstrate that the proposed ICA-GLR method has superior fault detectability when compared to traditional methods, such as principal component analysis and ICA.

반도체 제조공정용 약액 유량 제어장치의 제어 반응속도 개선

Equipment used in the semiconductor manufacturing process generally have a flow rate control system for each nozzle to regulate the flow rate of chemical solution fed to the wafer. In existing flow rate control systems, flow rate overshoots occur because of excess pressure and the control rates of the overshoots are less because additional operation time is required for the electric valves. In this study, to address the shortcomings of existing flow rate control systems, we proposed a method to improve the speed of flow rate control by introducing a constant pressure valve. The constant pressure valve controls the flow path via gas pressure, thereby facilitating prompt control and efficiently improving the flow rate overshoot caused by the pressure overshoot. To improve the control speed and control stability of the constant pressure valve, a three-step automatic control speed application function was developed, and the measured valve, control amount mapping function, and pre-open function were defined to reduce the initial control speed. The experimental results showed a measurement precision within 1% of the target flow rate and stable control performance as well as control speed reduction from 3 seconds in existing systems to 2 seconds or less for the proposed system.

Analytic Design of Segmented Phase Grating for Optical Sensing in High-Precision Alignment System.

Ultra-precision measurement systems are important for semiconductor manufacturing processes. In a phase grating sensing alignment (PGA) system, the measurement accuracy largely depends on the intensity of the diffraction signal and its signal-to-noise ratio (SNR), both of which are associated with the grating structure. Although an equally segmented grating structure could increase the signal of a high odd order, it could also strengthen the signals at the zeroth and even orders which are the main contributors of stray light. This paper focuses on the practical problem of differently responding diffraction orders but in one grating structure. An analytical relationship has been established between the diffraction efficiency and the segment structure of phase grating. According to this analytic model, we then propose a design method to increase the diffraction signal at high odd orders and, meanwhile, to decrease it at the zeroth and even orders. The proposed method provides a fast and effective way to obtain the globally optimal grating structure in the valid scope. Furthermore, the design examples are also verified by means of numerical simulation tool–rigorous coupled-wave analysis (RCWA) software. As a result, the proposed method gives insight into the diffraction theory of segmented grating and the practical value to greatly improve the design efficiency.

Post Chemical Mechanical Planarization (CMP) Cleaning Using Hydrogen Dissolved Water

In current semiconductor manufacturing process, especially chemical mechanical manufacturing (CMP) process, ceria nanoparticle dispersed slurry is well known consumable for various CMP processes such as shallow trench isolation (STI), inter layer dielectric (ILD) CMP process, and etc. After CMP process, the ceria nanoparticles which are the sources of defects must be removed surely before performing the other processes. Typically, Standard clean solutions such as SC-1, SC-2, and etc. are applied to the post CMP cleaning process. These cleaning solutions are still effective to remove the residual ceria particles but there are several disadvantages such as quite harmful for health and need of additional treatments for wastewater. In this study, cleaning efficiency of the H2 gas dissolved water was evaluated comparing with DIW and other standard cleaning solutions. 4x4 cm2 of silicon dioxide film wafers were polished supplying ceria slurry and then immersed into the cleaning solutions such as DIW, standard clean solutions, and H2 dissolved water, respectively. Finally, we quantitively analyzed the residual cerium dioxide on the cleaned wafer surface by using inductively coupled plasma mass spectrometer (ICP-MS). H2 dissolved water removed residual ceria nano particles from the silicon dioxide film considerably comparing to DIW. Figure 1. show residual ceria ion concentration after cleaning process and cleaning efficiency calculated. Figure 1

A novel quality prediction method based on feature selection considering high dimensional product quality data

<p style='text-indent:20px;'>Product quality is the lifeline of enterprise survival and development. With the rapid development of information technology, the semiconductor manufacturing process produces multitude of quality features. Due to the increasing quality features, the requirement on the training time and classification accuracy of quality prediction methods becomes increasingly higher. Aiming at realizing the quality prediction for semiconductor manufacturing process, this paper proposes a modified support vector machine (SVM) model based on feature selection, considering the high dimensional and nonlinear characteristics of data. The model first improves the Radial Basis Function (RBF) in SVM, and then combines the Duelist algorithm (DA) and variable neighborhood search algorithm (VNS) for feature selection and parameters optimization. Compared with some other SVM models that are based on DA, genetic algorithm (GA), and Information Gain algorithm (IG), the experiment results show that our DA-VNS-SVM can obtain higher classification accuracy rate with a smaller feature subset. In addition, we compare the DA-VNS-SVM with some common machine learning algorithms such as logistic regression, naive Bayes, decision tree, random forest, and artificial neural network. The results indicate that our model outperform these machine learning algorithms for the quality prediction of semiconductor.</p>

Artificial Immune System for Fault Detection and Classification of Semiconductor Equipment

Semiconductor manufacturing comprises hundreds of consecutive unit processes. A single misprocess could jeopardize the whole manufacturing process. In current manufacturing environments, data monitoring of equipment condition, wafer metrology, and inspection, etc., are used to probe any anomaly during the manufacturing process that could affect the final chip performance and quality. The purpose of investigation is fault detection and classification (FDC). Various methods, such as statistical or data mining methods with machine learning algorithms, have been employed for FDC. In this paper, we propose an artificial immune system (AIS), which is a biologically inspired computing algorithm, for FDC regarding semiconductor equipment. Process shifts caused by parts and modules aging over time are main processes of failure cause. We employ state variable identification (SVID) data, which contain current equipment operating condition, and optical emission spectroscopy (OES) data, which represent plasma process information obtained from faulty process scenario with intentional modification of the gas flow rate in a semiconductor fabrication process. We achieved a modeling prediction accuracy of modeling of 94.69% with selected SVID and OES and an accuracy of 93.68% with OES data alone. To conclude, the possibility of using an AIS in the field of semiconductor process decision making is proposed.