ASML (United States)

Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.

ASML’s multi-electron beam inspection (MBI) tool with 25 beamlets offers a viable solution for detecting voltage contrast defects as its primary use case and demonstrates the capability to detect smaller physical defects. This paper evaluates the performance of this MBI tool on Micron Technology’s DRAM and 3D-NAND devices, addressing both electrical and physical defects. The findings illustrate the tool’s benefit for use in technology development and high-volume manufacturing environments to enhance yield and monitor processes.

Abstract Recently, a number of state-of-the-art surrogate machine learning (ML) models have been designed for global weather and climate prediction, which have been trained using reanalysis data products. Reanalysis data products are constructed using numerical model simulations that combine numerical integration of partial differential equations and parameterization schemes. These products are typically only archived and made available using coarsened spatial and temporal resolutions. This study explores the impact of the numerical generation methods used to produce the training datasets and the temporal resolution of those datasets on machine learning surrogate models. Using the nonlinear vector autoregression (NVAR) machine as an explainable ML technique, simple dynamical systems are emulated with ML models trained on data produced by three classical numerical integration schemes. NVAR is validated as a skillful ML method, capable of producing accurate predictions and, more importantly, reconstructing both the underlying dynamics and the numerical integration scheme used to generate the training data. However, the machine fails to generalize predictions on unseen test data generated by different numerical integration schemes, despite the underlying dynamical system being the same. This result provides a word of caution for the growing field of machine learning emulation of weather and climate dynamics. Furthermore, we illustrate using NVAR that training on temporally coarsened data may increase the required complexity of ML models and potentially introduce new numerical challenges. Finally, we discover that empirical integration schemes with arbitrary time-stepping sizes can be constructed directly from the data, which implies a potential for the development of empirical numerical integration schemes.

This paper presents the design and analysis of a novel high-speed, multi-layer ferrite-based interior permanent magnet motor (HSFR) incorporating a reinforced carbon composite (RCC) rotor. While ferrite-based IPMs offer a cost-effective alternative to rare-earth (RE) magnet motors for traction applications, their mechanical limitations at high speeds necessitate innovative design approaches. Traditional central ribs, often employed to enhance rotor stability, compromise electromagnetic performance. Consequently, removing the ribs and introducing a retention sleeve around the rotor is a feasible choice until it offers a good-fit between electromagnetic and mechanical performances. However, this technique has been limitedly studied for ferrite magnets, which is mechanically weaker than RE ones. To address this, the RCC is strategically integrated in circumferential (sleeve) and radial (rib) directions to optimize rotor strength. Additionally, a six-phase winding configuration is implemented to augment torque output, compensating for the inherent lower magnetic energy density of ferrite magnets. A 4kW rated, maximum 30,000rpm, six-phase ferrite-based IPM with an RCC-wrapped internal permanent magnet motor (6PF-CCW-IPM) is developed through analytical optimization and finite element analysis (FEA). Performance comparisons with equivalent reference design validate the effectiveness of the proposed design in achieving high-speed operation while maintaining competitive power density

Extreme ultraviolet (EUV) lithography sources generate energetic tin ions. Accurate modeling of ion transport relies on knowledge of elastic scattering cross-section of tin ions against molecular hydrogen. This work uses available theoretical knowledge of elastic scattering found in literature, quantum chemistry modeling of the interaction potential between tin and hydrogen, and experimental scattering measurements of a tin ion beam to calculate the cross-section values. Furthermore, semi-classical collision calculations were used to calculate collision induced dissociation cross sections of tin ions and molecular hydrogen.

In this contribution we describe a Tachyon based flow to measure mask error systematics and simulate their impact to wafer level, as well as determining the contribution of these errors to OPC model calibration error. The dataset consists of 1000 OPC model calibration patterns for an N5 via layer for which we have a quadruplet of post-OPC GDS, simulated mask contour, mask CDSEM image and wafer CDSEM image as shown in Figure 2. First we examine the error at the mask using a combination of MXP (for metrology) and Tachyon LMC (for contour comparison) software products. Again using LMC we can simulate to wafer level using the simulated mask contour, and the measured mask contour as input. This allows us to correlate mask error to expected wafer error. Finally we calibrate OPC models to the measured wafer data using the various mask representations (postOPC Manhattan, simulated and measured) as input and observe the effect on OPC model error. This is done using Tachyon FEM+ with term based resist models of varying complexity.

Dual monopole exposure has been proposed1 as a way to improve imaging performance in EUV lithography by reducing the loss of contrast from image fading caused by the image shifts for the two poles of a dipole source. Simulations showed that the imaging advantages can be significant, with more than 15% image contrast improvements predicted. This paper presents experimental wafer data to demonstrate and verify the predicted advantages. The observed imaging enhancements include: • 18% better NILS (Normalized Image Log-Slope) for 28nm pitch patterns. • Better tip-to-tip patterns enabling gaps as much as 3nm smaller than normal patterning. Tip-to-tip LCDU and exposure latitude were improved at the same time, for better overall T2T capability. • Best focus offsets between three pitches: P28, P56 and P96 was reduced from 30nm range to nearly zero. • Smaller LWR (Line-Width Roughness), as much as 20% depending on pitch. • Reduced micro-bridging defects in the “stochastic cliffs” of narrow trenches, as large as a 40X defect density reduction for narrow trenches with 82nm pitch. Our experimental results validate the substantial imaging advantages seen by initial simulations. In addition to verifying dual monopole imaging advantages, our experiments have measured the pole-to-pole image shift δx_P2P, an important parameter relating to image contrast which has never been measured before. This important parameter depends on the detailed mask structure as well as the specific shape and location of the source poles. Our measured δx_P2P was consistent with simulated expectations.

With the adoption of extreme ultraviolet (EUV) lithography for high-volume production of advanced nodes, stochastic variability and resulting failures, both post litho and post etch, have drawn increasing attention. There is a strong need for accurate models for stochastic edge placement error (SEPE) with a direct link to the induced stochastic failure probability (FP). Additionally, to prevent stochastic failure from occurring on wafers, a holistic stochastic-aware computational lithography suite of products is needed, such as stochastic-aware mask source optimization (SMO), stochastic-aware optical proximity correction (OPC), stochastic-aware lithography manufacturability check (LMC), and stochastic-aware process optimization and characterization. In this paper, we will present a framework to model both SEPE and FP. This approach allows us to study the correlation between SEPE and FP systematically and paves the way to directly correlate SEPE and FP. Additionally, this paper will demonstrate that such a stochastic model can be used to optimize source and mask to significantly reduce SEPE, minimize FP, and improve stochastic-aware process window. The paper will also propose a flow to integrate the stochastic model in OPC to enhance the stochastic-aware process window and EUV manufacturability.

We study the following lesser-known low rank (LR) recovery problem: recover an <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$n \times q$ </tex-math></inline-formula> rank- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$r$ </tex-math></inline-formula> matrix, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${ \boldsymbol {X}}^{\ast}=[\boldsymbol {x}^{\ast}_{1}, \boldsymbol {x}^{\ast}_{2}, \ldots, \boldsymbol {x}^{\ast}_{q}]$ </tex-math></inline-formula> , with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$r \ll \min (n,q)$ </tex-math></inline-formula> , from <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$m$ </tex-math></inline-formula> independent linear projections of each of its <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$q$ </tex-math></inline-formula> columns, i.e., from <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {y}_{k}:= \boldsymbol {A}_{k} \boldsymbol {x}^{\ast}_{k}, k \in [q]$ </tex-math></inline-formula> , when <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {y}_{k}$ </tex-math></inline-formula> is an <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$m$ </tex-math></inline-formula> -length vector with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$m &lt; n$ </tex-math></inline-formula> . The matrices <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {A}_{k}$ </tex-math></inline-formula> are known and mutually independent for different <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k$ </tex-math></inline-formula> . We introduce a novel gradient descent (GD) based solution called AltGD-Min. We show that, if the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {A}_{k}\text{s}$ </tex-math></inline-formula> are i.i.d. with i.i.d. Gaussian entries, and if the right singular vectors of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${ \boldsymbol {X}}^{\ast}$ </tex-math></inline-formula> satisfy the incoherence assumption, then <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\epsilon $ </tex-math></inline-formula> -accurate recovery of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${ \boldsymbol {X}}^{\ast}$ </tex-math></inline-formula> is possible with order <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$(n+q) r^{2} \log (1/\epsilon)$ </tex-math></inline-formula> total samples and order <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$mq nr \log (1/\epsilon)$ </tex-math></inline-formula> time. Compared with existing work, this is the fastest solution. For <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\epsilon &lt; r^{1/4}$ </tex-math></inline-formula> , it also has the best sample complexity. A simple extension of AltGD-Min also provably solves LR Phase Retrieval, which is a magnitude-only generalization of the above problem. AltGD-Min factorizes the unknown <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${ \boldsymbol {X}}$ </tex-math></inline-formula> as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${ \boldsymbol {X}}= { \boldsymbol {U}} \boldsymbol {B} $ </tex-math></inline-formula> where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${ \boldsymbol {U}}$ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {B}$ </tex-math></inline-formula> are matrices with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$r$ </tex-math></inline-formula> columns and rows respectively. It alternates between a (projected) GD step for updating <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${ \boldsymbol {U}}$ </tex-math></inline-formula> , and a minimization step for updating <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {B}$ </tex-math></inline-formula> . Its each iteration is as fast as that of regular projected GD because the minimization over <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {B}$ </tex-math></inline-formula> decouples column-wise. At the same time, we can prove exponential error decay for it, which we are unable to for projected GD. Finally, it can also be efficiently federated with a communication cost of only <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$nr$ </tex-math></inline-formula> per node, instead of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$nq$ </tex-math></inline-formula> for projected GD.

We developed a low reflective beam splitter (BS) for hot-spot monitoring of argon-fluoride (ArF) laser, and detected hot-spot in a laser beam profile only at high discharge current and high output power mode.