Gels are materials with both liquid-like and solid-like properties, widely used in applications ranging from food to medical products. In this project, we studied a special type of gel formed by particles that both attract and repel each other. These competing interactions create a unique network of rigid chains with a characteristic spacing, connected by flexible joints. Surprisingly, these connections can break over time, causing the network to fall apart again into ordered clusters. When deformed, the network shows a stepwise structural breakdown marked by two yield points. These insights support the design of soft materials with tailored mechanical properties.

The purpose of this seminar is to investigate a fundamentally new framework for controlling manufacturing machines and scientific instruments by exploiting and learning from data. This will lead to unparalleled performance for such high-tech systems, in which the state- of-the-art systems are presently still being controlled by traditional control philosophies that do not exploit the major opportunities of the abundance of data. Manufacturing machines and scientific instruments have a key role in our society. Wafer scanner technology is arguably the most important example in this respect, since Integrated Circuits (ICs) have led to ubiquitous computing power, leading to major developments in communication, medical equipment, transportation, etc. In fact, Moores law dictates a doubling of IC complexity every two years, which is enabled by progress in wafer scanner technology. This wafer scanner technology is developed primarily by industries in The Netherlands and Japan, and it is crucial that this advantage is reinforced. Positioning systems, or motion systems, are the key subsystems in wafer scanner technology that enable the manufacturing machines to function by positioning the ICs within the machine. Future machines must achieve high accuracy of 0.1 nm to allow for a doubling of IC complexity through miniaturization. At the same time, speed and acceleration up to 1 m/s and 100 m/s2 must be achieved to achieve high throughput and hence market viability of the machine and low cost of IC production for the end-user. Although major achievements have been made to follow Moores law already for decades, a major breakthrough in the control paradigm is foreseen to be essential to continue the exponential growth of Moores law. The aim of this seminar is to exploit the huge amount of sensors, actuators, and data in controlling high-tech mechatronic systems, such as wafer scanners, to the limits of performance. Indeed, the working hypothesis of this seminar is that everything in the systems behaviour that can be predicted can also be compensated for. However, this is by no means possible through traditional design philosophies that are still common in the current state-of-the-art systems. The goal is to bring together researchers to develop a new fundamental design framework for learning from data in complex mechatronic systems in view of new generations of future data-intensive mechatronic systems with unparalleled performance. On a longer horizon, the research will have a major impact on the development of radically new data-intensive mechatronic systems, where the use of data and control will be used to design radically different and lightweight systems. Indeed, a radically new view on mechatronic design, automatic control, and machine learning is foreseen, where new system designs will be combined with spatially distributed actuators that control spatiotemporal deformations, leading to a huge potential in speed and accuracy. These systems will be continuously monitored in real time through data and models, which constitute digital twins, to monitor their performance, identify faults, and predictive maintenance.