Semiconductor Industry Research Articles

Semiconductor production scheduling and industrial economic coordination based on deep learning and heat loss optimization

With the growing global focus on energy efficiency and environmental impact, the semiconductor industry faces the dual challenge of improving production efficiency and reducing energy consumption. Heat loss is an important source of energy consumption in semiconductor production, so optimizing heat loss is of great significance. This paper aims to explore the optimization method of semiconductor production scheduling based on deep learning technology, with special attention to the reduction of heat loss, so as to achieve the coordinated development of industrial economy. The source of thermal energy loss in semiconductor production is analyzed and a deep learning model is constructed to predict thermal energy loss under different production scheduling schemes. Through deep learning training of production data, the model can identify and optimize the thermal energy consumption patterns of each link in the production process. In order to achieve the balance between minimum heat loss and maximum production efficiency, a variety of production scheduling schemes are optimized using simulated annealing algorithms. The experimental results show that the production scheduling scheme based on deep learning can effectively reduce heat loss and optimize production efficiency. Compared with traditional scheduling methods, the optimized system has reduced thermal energy loss and increased production efficiency. These results demonstrate the potential of deep learning in semiconductor production scheduling, significantly optimizing heat loss and promoting the coordinated development of the industrial economy.

From microchips to microneedles: semiconductor shear testers as a universal solution for transverse load analysis of microneedle mechanical performance

Abstract Microneedles are a promising technology for pain-free and efficient pharmaceutical delivery. However, their clinical translation is currently limited by the absence of standardized testing methods for critical quality attributes (CQAs), such as mechanical robustness, which are essential for demonstrating safety and efficacy during regulatory review. A key aspect of mechanical robustness is transverse load capacity, which is currently assessed using diverse, non-standardized methods, which have limited capability to measure transverse failure forces at different heights along a microneedle. This is critical for understanding mechanics of potential failure modes during insertion after skin penetration. In this work we utilize a wire bond shear tester, a piece of test equipment widely used in the semiconductor industry, to measure the transverse load capacities of various microneedle designs. This approach is compatible with diverse microneedle types, geometries, and materials, and offers high-throughput and automated testing capabilities with high precision. We measure transverse failure loads with micron-scale control over the test height and have established comprehensive profiles of mechanical robustness along the length of different microneedle designs, which is a capability not previously demonstrated in literature for polymeric and metal microneedles. Transverse failure forces were 10±0.3 gf to 128±12 gf for wire bonded gold and silver microneedles, 11±0.7 gf to 480±69 gf for conical and pyramidal polymeric microneedles, and 206±80 gf to 381±1 gf for 3D printed conical stainless steel microneedles. Additionally, we present standardized definitions for microneedle structural failure modes resulting from transverse loads, which can facilitate root cause failure analysis and defect detection during design and manufacturing, and aid in risk assessment of microneedle products. This work establishes a standardized approach to evaluating a significant CQA of microneedle products, which is a critical step towards expediting their clinical adoption.

Applying Fuzzy Decision-Making Trial and Evaluation Laboratory and Analytic Network Process Approaches to Explore Green Production in the Semiconductor Industry

As environmental awareness grows, society emphasizes green business. Thus, semiconductor companies encompass energy-saving processes and innovative product development. This study employs the fuzzy decision-making trial and evaluation laboratory (DEMATEL) analysis method to assess the impact of four key dimensions, externalities, market orientation, green technology, and corporate social responsibility, on semiconductor companies’ green production decisions. This study uses the DEMATEL-based analytic network process (DANP) approach to rank the criteria order in green production decision-making. The results from the DEMATEL causality diagram highlight externalities as the most critical dimension influencing other dimensions of green production decisions. This study suggests that regulatory pollution punishment and subsidies emerge as the primary drivers for green production decisions. Companies adopt environmentally friendly production practices to prevent regulatory pollution penalties or reduce carbon trading costs. Additionally, the DANP results reveal that corporate image criteria in the corporate social responsibility dimension hold the utmost priority in semiconductor firms’ green production decision-making. This implies that considerations for improving a company’s image should take precedence as semiconductor companies seek shareholder support and governmental subsidies to ensure sustainable operations. Externalities arise as the secondary priority dimension in green production decision-making, aligning with the impact of externalities identified in DEMATEL findings.

Two-step post-synthetic modification of bipydine-based sp2c-COF for stepwise gold recovery

Recovery of gold element from electronic waste is a promising approach for mitigating the rising gold demand in semiconductor industry. However, the complex and harsh environment in gold recovery restricted the selectivity and stability of absorbents. Herein, taking advantage of the robust sp2-conjugated covalent organic frameworks (COFs), we proposed a two-step post-synthetic modification (PSM) strategy to prepare the isp2c-COFox-NH2 for efficient Au extraction with high selectivity and recyclability. The stepwise electrostatic and chelation interaction of the modified quaternary ammonium and amidoxime groups resulted in an efficient adsorption capacity of 1389 mg/g. Experimental and theoretical studies show that the quaternary ammonium groups ionize the sp2c-COFs and significantly enhance the selectivity of gold adsorption (Ni2+, Co2+ and Cu2+ for negligible influence), and the amidoxime groups further increase the adsorption capacity of the COFs via chemical reduction. Furthermore, the isp2c-COFox-NH2 retained 95.1 % of its adsorption efficiency even after 10 adsorption–desorption cycles. It is worth mentioning that isp2c-COFox-NH2 displays good adsorption performance for trace amounts of gold in complex water samples and practical electronic waste, demonstrating excellent potential for practical applications. This study not only provides novel inspirations for improving the efficiency of gold resource recovery but also paves the way for the widespread application of sp2c-COF materials.

A machine learning-based simulation metamodeling method for dynamic scheduling in smart manufacturing systems

Conventional Digital Twins (DTs) in smart manufacturing rely on complex and time-intensive simulation models, hindering real-time DT-based decision-making. However, the availability of big data in Manufacturing Execution Systems (MES) enables training different Machine Learning (ML) models for fast and accurate predictions and decision assessments. Accordingly, this paper proposes an ML-Based Simulation Metamodeling Method (MLBSM) to facilitate DT-based decision-making for dynamic production scheduling in complex Stochastic Flexible Job Shop (SFJS) environments. The proposed MLBSM integrates three modules: a novel data vectorizing method (SPBM), multi-output Adaptive Boosting Regressor (ABR) models, and a new statistical risk evaluation method. SPBM converts unstructured production log data into numerical vectors for ABR training by calculating numeric penalty scores for each job based on the position of operations in the schedule queue. Each trained ABR predicts mean job completion times for various dynamic scenarios based on shift schedules. The risk evaluation method estimates the standard deviation of job completion times and calculates the delay probability scores for each job, aiding DT in promptly evaluating production schedules. Working seamlessly together, MLBSM modules present a novel way of using production log data for ML training and ultimately bypassing several computationally intensive simulation replications. In this research, a simulation model generates the synthetic MES data, focusing on the machining process at a photolithography workstation in the semiconductor manufacturing. Experiments demonstrate the MLBSM’s accuracy and efficiency, predicting high-risk jobs with over 80% recall and being at least 70 times faster than conventional simulation runs. Sensitivity analyses also confirm the MLBSM’s consistency under different workstation conditions.

Integrating resilience and reliability in semiconductor supply chains during disruptions

The semiconductor industry, a cornerstone of modern technology, has been crucial in driving globalization and supporting various sectors, from consumer electronics to automotive industries. However, in recent years, the industry has faced substantial challenges threatening its ability to meet the surging demand for semiconductor chips. Disruptions at any point in the supply chain, from raw material sourcing to end-product delivery, can substantially influence the semiconductor ecosystem. The intricate nature of such SCs makes them highly vulnerable to various disruptions, emphasizing the critical need for building resilient and reliable supply chain strategies. This article presents comprehensive research aimed at addressing critical gaps in the understanding and management of resilience and reliability within the semiconductor supply chain (SSC). This study proposes a multi-objective mixed-integer non-linear programming (MO-MINLP) model to configure an SSC while considering reliability and resilience measures. It emphasizes and draws attention to the importance of resilience and reliability in managing SSC disruptions during a pandemic and potential future epidemic outbreak. Exploring the precise breakdown of batch transportation between two sites shows how disruption can affect product flow along the SC. The applicability of the proposed method is demonstrated through a numerical example of an SSC, solved using the LINGO solver. Finally, a sensitivity analysis is conducted on the model's parameters to assess the capability and effectiveness of the results from managerial viewpoints.

Stable novel silicon allotropes in space group P2/m with various band gap structures by high-throughput screening

Silicon is the fundamental material for the semiconductor and microelectronics industries, but controlling the electronic band gap structure of diamond-type silicon remains a huge challenge to adapt to growing applications. Here, we have predicated 23 new silicon allotropes in space group P2/m from 279 possible structures by high-throughput screening accompanied by graph and group theory based on random strategy (RG2) code. The mechanical, electronic and optical properties of these structures were studied in detail. These novel silicon allotropes demonstrate various electronic structures, including metal, direct/quasi direct bandgap structure, and indirect bandgap structures. These new silicon allotropes demonstrate various electronic structures, including metal, direct/quasi direct bandgap structure, and indirect bandgap structures. Besides different electronic bandgap structures, all 23 structures exhibit strong absorption in the visible light region and P2/m-15 demonstrates the excellent mechanical properties (Bulk modulus beyond 80 GPa). Based on their nice stability, good mechanical, electronic and optical properties validated by the ab inito molecular dynamics simulation, phonon spectra and density functional theoretical calculations, these predicted silicon allotropes provide not only ideas for the synthesis of new silicon allotropes but also dawn for expanding the application of semiconductor materials.

Preface for Semiconductor Manufacturing Systems and Processes

Preface for Semiconductor Manufacturing Systems and Processes

Development of an ultra-clean sample heating stage for thermal desorption spectroscopy

Control of surface molecular contamination (SMC) for components used in chemical vapor deposition (CVD), atomic layer deposition (ALD) and EUV photolithography is important to maintaining high yield and optimal tool operation at the latest process nodes in leading edge semiconductor manufacturing. High temperature thermal desorption spectroscopy (TDS) is a versatile tool for analyzing the cleanliness of surfaces, simulating thermal vacuum processes and studying the kinetics of desorption processes. A basic analysis of TD spectra allows for full characterization of volatile outgassing from surfaces, while detailed analysis can provide chemical information about the substrate surface.In fundamental studies, TDS is often carried out from low temperatures to room temperature or for small samples. However, for microelectronics applications, high temperature studies of large (100 mm or greater) samples are of greater interest due to direct applications for cleanliness testing and thermal vacuum simulation. A limitation for TDS sensitivity is the outgassing of sample stage materials, particularly when analyzing gases that may be present in the chamber background such as water, CO and CO2. Typical sample stages are often tested only for total pressure or at room temperature.In this study, we present a simple ultra-high vacuum (UHV) compatible sample heating stage for trace outgassing analysis of 100 mm samples at high temperatures. Simulation results are presented to support the feasibility of the concept. Experimental results verify the cleanliness of the stage via room temperature residual gas analysis (RGA) analysis and X-ray photoelectron spectroscopy (XPS) of stage components. Finally, use of this stage in a TDS analysis of a 100 mm Si witness wafer and comparison to room temperature RGA demonstrates operational capability.The sample heating stage is both shown to be clean at high temperature and capable of analyzing 100 mm wafers to higher sensitivity than room temperature RGA for all m/z at the 1 × 10−9 mbar level. Despite its high performance, the heating stage is also easily produced by any laser machining service, greatly improving the accessibility of UHV science for all researchers.

Gel-Based PVA/SiO2/p-Si Heterojunction for Electronic Device Applications.

The current work presents a new structure based on Au/PVA/SiO2/p-Si/Al that has not been studied before. An aqueous solution of polyvinyl alcohol (PVA) polymer gel was deposited on the surface of SiO2/Si using the spin-coating technique. The silicon wafer was left to be oxidized in a furnace at 1170 k for thirty minutes, creating an interdiffusion layer of SiO2. The variations in the dielectric constant (Є'), dielectric loss (Є″), and dielectric tangent (tanδ) with the change in the frequency, voltage, and temperature were analyzed. The results showed an increase in the dielectric constant (Є') and a decrease in the dielectric loss (Є″) and tangent (tanδ); thus, the Au/PVA/SiO2/p-Si/Al heterostructure has opened up new frontiers for the semiconductor industry, especially for capacitor manufacturing. The Cole-Cole diagrams of the Є″ and Є' have been investigated at different temperatures and voltages. The ideality factor (n), barrier height (Φb), series resistance (Rs), shunt resistance (Rsh), and rectification ratio (RR) were also measured at different temperatures.

Wavelength-multiplexed multi-mode EUV reflection ptychography based on automatic differentiation

Ptychographic extreme ultraviolet (EUV) diffractive imaging has emerged as a promising candidate for the next generationmetrology solutions in the semiconductor industry, as it can image wafer samples in reflection geometry at the nanoscale. This technique has surged attention recently, owing to the significant progress in high-harmonic generation (HHG) EUV sources and advancements in both hardware and software for computation. In this study, a novel algorithm is introduced and tested, which enables wavelength-multiplexed reconstruction that enhances the measurement throughput and introduces data diversity, allowing the accurate characterisation of sample structures. To tackle the inherent instabilities of the HHG source, a modal approach was adopted, which represents the cross-density function of the illumination by a series of mutually incoherent and independent spatial modes. The proposed algorithm was implemented on a mainstream machine learning platform, which leverages automatic differentiation to manage the drastic growth in model complexity and expedites the computation using GPU acceleration. By optimising over 200 million parameters, we demonstrate the algorithm's capacity to accommodate experimental uncertainties and achieve a resolution approaching the diffraction limit in reflection geometry. The reconstruction of wafer samples with 20-nm high patterned gold structures on a silicon substrate highlights our ability to handle complex physical interrelations involving a multitude of parameters. These results establish ptychography as an efficient and accurate metrology tool.

Interfacial thermal conductance in 2D WS2/MoSe2 and MoS2/MoSe2 lateral heterostructures

In-plane Transition Metal Dichalcogenides (TMDs) heterostructures hold immense potential for various applications in the modern semiconductor industry, including electronics, optoelectronics, and photovoltaic devices. Different TMD monolayers can be ‘stitched’ together to construct an in-plane (lateral) heterostructure. As different TMD monolayers present different work functions and have their intrinsic shortcomings, a TMD heterostructure is an excellent form to optimize their properties and to achieve the best functionality. This requires a quantitative understanding of the properties of the interfaces in the heterostructures. In this work, we perform nonequilibrium molecular dynamics simulations, based on a parametrized Stillinger-Weber potential, to investigates the thermal conductance of the interfaces in 2D WS2/MoSe2 and MoS2/MoSe2 in-plane heterostructures, as well as in 2D lateral WS2/MoSe2 superlattices. Three distinct types of interfaces, including defect-free coherent interfaces, interfaces with the 5∣7 defects, and the alloy-like incoherent interfaces, are explored. The effects of interphase structure and temperature are quantified. Phonon density of states (PDOS) analysis is used to understand the effect of different interphase structures. The effect of superlattice period on thermal conductance of the superlattices has also been quantified.

Magnetohydrodynamics Marangoni boundary-layer copper/water nanofluid flow driven by surface temperature gradient over a rotating disk

Marangoni driving problem is very important in many practical science engineering, such as semiconductor industry, crystal growth, aerospace, material synthesis. Consider these applications, this work we investigate Marangoni boundary-layer Copper/Water nanofluid driven by the surface temperature gradient over a rotating disk in the presence of magnetic field. Buongiorno model of nanofluids, which contains two important terms, thermophoresis and Brownian motion, are taken into account. Rotating disk model is established, also a suitable Kármán transformation and the multi-shooting technique are applied. Graphical discussion include free stream concentration, rotating speed, magnetic field, Marangoni driving, thermophoresis, Brownian diffusion, Prandtl number, and Schmidt number. The results show that rotating speed inertia and Marangoni driving force tend to reduce temperature/concentration and thickness of thermal/mass boundary layer, while free stream concentration and magnetic field raise up. Meanwhile, Prandtl number tends to reduce temperature, while thermophoresis and Brownian diffusion raise up, and Schmidt number has little influence on temperature, and Schmidt number and Brownian diffusion tend to reduce concentration, while thermophoresis raises up, and Prandtl number has an important influence on concentration. For most selected values of physical parameters, as the nanofluid far away from the disk, the local temperature decreases directly, and the local concentration firstly increases to a peak and then decreases to zero.

Exploring Defect Dynamics and Twin-Layer Interactions in SiC Crystals through Molecular Simulations.

Silicon carbide (SiC), a third-generation semiconductor material, is pivotal for applications in new energy vehicles, aerospace, and high-speed electronics, owing to its superior properties. This study delves into the twin-induced growth behaviors of SiC crystals through molecular dynamics simulations at temperatures ranging from 2700 to 3200 K. It focuses on the wurtzite and zinc blende SiC structures, revealing dynamic defect behavior during growth, including an initial rise and subsequent decrease in vacancies, with particular emphasis on prevalent defects within zinc blende twin layers. A significant finding is the direct correlation between temperature and growth rates across different SiC structures, highlighting temperature control as essential for optimizing crystal quality. Furthermore, this work contributes to the analysis of the interactions of twin layers and their impact on structural stability and defect formation in SiC crystals. The insights gained here have substantial implications for the semiconductor industry, potentially enhancing device performance by better controlling growth conditions and defect management in SiC-based electronic and optoelectronic devices.

Numerical modelling on dispersion behavior of particulate contamination induced by a moving operator in a semiconductor cleanroom: A eulerian-eulerian method

Contamination control in cleanrooms is crucial across various industries. In semiconductor manufacturing, operator-induced contamination presents a significant challenge, as it reduces the critical dimension. In this research, one comprehensive model describing contamination behaviour induced by operator motion was proposed, considering the moving operator as both a flow pattern changer and a contaminant source, combined with the number of particles accumulated (NPA) on the process instruments. To quantitatively characterise the transient contaminant dispersion behaviour caused by the operator, a Eulerian-Eulerian-based particle number density equation, integrated with overset mesh technology was employed. Our findings reveal that contamination dispersion behaviour is governed by convection and diffusion. The NPA on the side instruments at operator moving speeds at 0.5, 1, and 1.5 m/s were 4404.33, 8241.26, and 9384.41, respectively. And those of central instruments are 28463.75, 35848.60, and 43680.74 in corresponding cases. The largest reduction in influence on both the central and side instruments was found between cases with a maximum operator speed of 0.5 and 1.5 m/s, with NPA reaching reductions of 53.08 % and 34.84 % respectively. This indicates that the moving speed significantly affects contaminant dispersion.

Optimum production lot size for a perishable product under exponential demand with partial backordering and rework

The management of inventory significantly impacts an organization's profitability. Numerous researchers have developed the production model tailored for decaying items to boost sales and minimize inventory, thereby increasing turnover ratios. This study focuses on the EPQ model for reworkable items under exponential combined with inventory dynamics. The primary objective is to optimize the production plan to manage maintenance costs effectively, ensuring optimal Leveraging human and technological resources. The research determines the optimal cycle length, backordered quantities, and production quantities to minimize overall costs. Specifically, penalties are incurred for excessive production times during rework and non-rework phases. This model finds applications in industries such as printed circuit board assembly in semiconductor manufacturing, metal parts, and plastic molding. The study concludes with numerical examples comparing total costs under the proposed model, supplemented by sensitivity analyses that explore the impacts of parameter variations on decision-making.

Atomic scale smoothing of nanoscale quartz mold using amorphous carbon films

Abstract Surface roughness control of end products is increasingly becoming significant, especially with the miniaturization trends in the semiconductor industry. Ultra-thin amorphous carbon (a-C) films offer a prime solution to optimize surface roughness due to their outstanding characteristics. In this study, hydrogenated a-C films are deposited on two-dimensional quartz plates and three-dimensional quartz molds to evaluate the growth mechanisms and changes in the surface roughness, which is supported by molecular dynamics simulations. Results reveal that surface roughness encounters multiple variations until it reaches stable values. These fluctuations are categorized into four different stages which provide a concrete understanding of various growing mechanisms at each stage. Different behavior of the atoms in the top layers is recorded in the cases of normal and grazing incidents of carbon atoms. Lower surface roughness values are obtained at low-angle deposition. Interestingly, surface smoothing is attained on the sidewalls of the nanotrench mold where the deposition occurs with high incident ion angles.

Similarity searching for wafer bin maps by measuring shape, location, and size similarities of defect patterns

A wafer bin map (WBM) is a visual representation of the spatial distribution of defective chips on a wafer. WBMs showing specific defect patterns are usually a result of process-assignable causes; thus, it is important to identify them to eliminate assignable causes. With advances in semiconductor manufacturing technology, identifying new defect patterns, and diagnosing their causes have become critical. However, most existing methods for WBM analysis use a supervised learning approach, which only detects previously known defect patterns. The similarity-search approach is a suitable alternative for defining new defect patterns. The proposed method uses an unsupervised approach to search for similar WBMs by measuring the similarities of three spatial features of defect patterns–shape, location, and size–which are useful for defining new defect patterns and diagnosing their causes. These three similarities are achieved using tensor voting and the mountain function for shape similarity, Euclidean distance for location similarity, and a combination of defect count and average radius for size similarity. The overall similarity was assessed using the weighted average of the three similarities. The weights are determined by quantifying the uncertainty of each similarity based on information entropy theory to better distinguish between similar patterns. The experimental results demonstrate the effectiveness of the proposed method compared to existing methods and highlight its capability to identify and describe the spatial features of defect patterns.

Optimising distributed heterogeneous flowshop group scheduling arising from PCB mounting: integrating construction and improvement heuristics

ABSTRACT In real-world manufacturing systems for processing printed circuit boards (PCBs), the workshops integrating various flowline-based cells are common in large-scale enterprises. The scheduling problem within the systems is modelled as the distributed heterogeneous flowshop group scheduling problem (DHFGSP) in this study. Departing from practical requirements and considering the grouping characteristics among PCB components, we account for carryover sequence-dependent setup time (CSDST). Moreover, recognising the critical importance of just-in-time production in semiconductor manufacturing, the total tardiness, a previously unexplored objective within the DHFGSP context, is addressed. To tackle this problem, a mixed-integer linear programming (MILP) model, capable of obtaining optimal solutions for small-scale instances, is proposed. Due to the NP-hard nature of the problem, obtaining high-quality solutions in a reasonable time using the MILP model becomes challenging for large-scale instances. Therefore, a solution algorithm, comprising construction and improvement heuristics, is developed. Capitalising on the problem’s characteristics, the construction heuristic efficiently generates high-quality feasible solutions in a very short time. Built primarily around artificial bee colony (ABC) optimisation, the improvement heuristic can significantly further enhance solutions by integrating the collaborative and restart operators. Comprehensive experiments on instances of varying scales demonstrate the effectiveness of the proposed algorithm for the problem under investigation.

Multi-objective optimization and inverse design of complementary field-effect transistor using combined approach of machine learning and non-dominated sorting genetic algorithms for next-generation semiconductor devices

Complementary field-effect transistors (CFETs), which are structures in which different types of transistors are vertically stacked with a shared control gate, are being focused on for continuing to satisfy Moore's law by overcoming the limitations in pitch scaling. The structures of semiconductor devices become more complex as technology node shrinks, and interrelated multivariate parameters increase. In addition, predicting problems and proposing solutions by identifying complex patterns within extensive data collected for emerging semiconductor designs pose significant computational challenges and are inherently difficult. As a breakthrough in design technology co-optimization for advanced devices, this study developed a novel optimization framework integrating technology computer-aided design simulations, machine learning, and non-dominated sorting genetic algorithms. The developed framework provides unbiased optimal solutions, even in a high-dimensional objective space, while considering the tradeoff relationships between multiple variables. In addition, it enables inverse design to identify the design parameters of devices that satisfy specific electrical performance criteria using only a forward model, while achieving an error rate of less than 2%. Using this framework, we analyzed the operational mechanism of CFETs by comparing the inverse designs of various devices. This novel approach is particularly important when the design space is complex and extensive and is well suited for developing devices that emerge with technological advancements in the semiconductor industry.