Juerg Leuthold

We present a predictive multiscale modeling workflow for Nb ₃ Sn Rutherford cables that bridges strand and cable scales to support the design of accelerator magnets in the 14 T range. This experimentally validated framework enables prediction of critical current reduction under load of Nb ₃ Sn Rutherford cables. The workflow is parameterized in strand and cable architecture, allowing it to be applied across cable designs; as done here on two 21-strand cables of different design (CD1 and SMACC-HF), differing most notably in strand diameter and Cu/non-Cu ratio. The workflow integrates a cabling model capturing strand transposition and compaction-induced deformation, a Python tool for generating FEM-ready 2D cross-sections, homogenization of the Nb ₃ Sn interfilamentary region with inter-strand contact modeling, and a linkage connecting simulated strain to critical current reduction. The workflow was mechanically validated against 10-stack Rutherford cable measurements at 77 K, with the simulated macroscopic stress–strain slope matching the measurement to within 6% in virgin loading and 0.1% in reloading. The predicted critical current reduction was validated against the compression BOX experiment. Across the 10–250 MPa transverse-pressure range covered by the experiments, the predicted cable critical current agrees with the measurement to a mean absolute error of ⩽ 3.2% for CD1, where I c was taken from an independent strand measurement and ⩽ 0.9% for SMACC-HF, where the initial I c is taken from the virgin cable measurement. Both measurement and simulation show a correlation between a lower Cu/non-Cu ratio and retaining a higher fraction of the virgin critical current under transverse pressure.

The brain embeds learning algorithms within its physical architecture, enabling on-site learning processes with remarkable advantages such as energy efficiency, learning autonomy and continual learning capabilities. Current memristive neuromorphic implementations, despite being inspired by the brain, are predominantly trained via offline, software-based methods. Consequently, this paradigm restricts their potential to realize on-chip learning. To address this limitation, we implement a simple memristive-based electronic circuit with an op-amp that emulates a neuron with biologically inspired learning rules, characterized by recent electrophysiological observations. To demonstrate its potential as a fundamental computational primitive, we integrate it into a network with one hidden layer capable of learning all fundamental binary operations. As a proof-of-concept, we build a physical circuit that successfully learns and performs XOR, AND, and OR logic operations. To validate the scalability of our approach beyond simple logic, we also perform SPICE simulations on a compressed MNIST classification task. Therefore, the proposed circuit could be leveraged to build memristive neuromorphic systems capable of doing on-chip learning.

High-speed, integrated photodetectors are crucial for addressing the bandwidth bottlenecks in next-generation data centers and high-performance computing. While graphene offers exceptional intrinsic speed and broadband absorption, realizing its potential in scalable, high-responsivity integrated devices has proven challenging. Here, we demonstrate a waveguide-integrated plasmonic graphene photodetector architecture capable of operating in high-capacity data links. By utilizing plasmonic resonators to enhance light-matter interaction and facilitate photocarrier extraction, the device exhibits a 3 dB bandwidth of 155 GHz and a responsivity exceeding 200 mA/W in the L-band under bias. We validated the device’s capabilities for high-speed data communications through direct detection of intensity-modulated signals, achieving a single-lane line rate of 192 Gbit/s. To further illustrate the platform’s scalability, we cointegrated eight such photodetectors with a low-crosstalk silicon photonic 1 × 8 arrayed waveguide grating to form a wavelength-division multiplexing receiver. This integrated system yielded an aggregated line rate of 816 Gbit/s. These results, representing the fastest waveguide-coupled plasmonic graphene photodetectors reported to date, pave the way for a new generation of dense, high-capacity optical interconnects essential for future high-performance computing and AI infrastructure.

Silicon–graphene hybrid photodetectors represent a promising platform by combining graphene’s high carrier mobility and pronounced electrostatic sensitivity with silicon’s superior light absorption capability. However, systematic comparisons of different detection mechanisms in such hybrids remain limited, making it challenging to identify the optimal architecture for different imaging applications. Here, we present a systematic comparison of three photodetection mechanisms in silicon–graphene hybrid photodetectors: interfacial photogating, Schottky photodiodes, and Schottky-assisted photogating. These devices have been implemented on the same chip and evaluated under identical conditions. By analyzing responsivity, noise, dynamic range, speed, and array-level imaging, including crosstalk and pixel variation in 5 × 5 arrays, we identify the key performance trade-offs. Interfacial photogating offers high gain and strong low light sensitivity, Schottky-assisted photogating provides higher gain but larger variability and slower response, while Schottky photodiodes enable fast, low-cross-talk imaging. These insights establish practical guidelines for selecting hybrid silicon–graphene detector architectures based on application-specific performance priorities.

Emerging cryogenic technologies suffer from scaling discrepancies due to limitations stemming from the electrical signal lines connecting the ambient to the cryogenic environment. Currently employed radio frequency cabling offers limited operation bandwidths, therefore requiring large amounts of valuable space and inherently also conducting large amounts of heat. Realizing the ambient-to-cryogenic signal transfer by optical fibers could resolve all of these issues. However, for this purpose, cryogenic-compatible, high-speed, zero-bias-operated photodetectors are required. In this work, we demonstrate a metamaterial-graphene photodetector able to generate high-speed electrical signals in a cryostat without any need for electrical feed lines. This is enabled by a nanostructured metamaterial-graphene architecture that utilizes nanoscale optical hot spots to simultaneously enhance absorption and enable zero-bias carrier extraction. Beyond that, the devices offer high-frequency operations above 100 GHz, a small footprint, and compatibility with various substrates. The signal transfer is further demonstrated by cryogenic optical communication with bit rates of >100 Gbit/s. These results demonstrate a robust platform for high-bandwidth optical-to-electrical conversion, providing a scalable pathway for low-power signal interfacing in next-generation cryogenic systems and quantum technologies.

Integrated photonics has enabled the miniaturization and development of classical as well as quantum-optical technologies. However, traditional lithographic techniques limit the required optical elements to binary height profiles. By relaxing these constraints, grayscale fabrication methods have the potential to deliver more efficient and compact devices. In contrast to binary profiles, wavy surfaces (also known as optical Fourier surfaces, OFSs) only introduce spatial frequencies that are required for their functionality, yielding more control over the optical response. In this work, we demonstrate photonic integrated circuits with grayscale OFS elements in state-of-the-art material platforms, including silicon-on-insulator and thin-film lithium niobate. Using thermal scanning-probe lithography and dry etching, the OFSs are written and transferred with high fidelity. We employ an intuitive and straightforward design scheme to create a series of devices that exploit the capabilities of wavy height profiles. First, we fabricate sinusoidal single- and multiband Bragg reflectors in silicon waveguides with an extinction ratio of up to 44 dB at 1550 nm. Second, cavities in lithium-niobate waveguides are fabricated with a high quality factor of 1.6 × 10⁵ and a theoretical modal volume of 2.4(λ/n)³. Finally, we exploit this high-quality-factor cavity and the large optical nonlinearity of lithium niobate to produce frequency-doubled light via second harmonic generation.

Optical digital-to-analog converters (DAC) offer power-efficient amplitude multiplexing in the optical domain. They allow simplification to complex electronic circuits as the binary signals are directly fed to the Mach-Zehnder modulators (MZM). Drive circuits with a large effective number of bits (ENOB) and complex power-hungry electronic predistortion of the MZMs nonlinear transfer characteristic are eliminated. This results in a power advantage and smaller electronics, ultimately delivering an improved signal-to-noise and distortion ratio (SNDR). As oDACs can be operated by means of direct digital drives (DDD) with minimal signal processing, they are ideal for highest speed operation. When implemented with terahertz-bandwidth plasmonic modulators, they may provide highest symbol rates on a small footprint with the least power consumption. We demonstrate the first plasmonic oDAC implementation for IM/DD and coherent transmission, operating at up to 420 Gb/s line rate without SNR penalty. We show energy-efficient amplitude multiplexing in the optical domain at symbol rates up to 180 GBd.

Three-dimensional integrated circuits (3D ICs) have emerged as a key technology to sustain scaling trends in the microelectronics industry. This advancement calls for a fundamental shift in how electrical interconnects are implemented, with through-silicon vias (TSVs) playing a pivotal role in enabling vertical connectivity between stacked chips. However, the metallization of TSVs traditionally involves elaborate and demanding processes, which can limit the speed and flexibility of prototyping and design modifications. In this paper, we investigate the use of Ultra-Precise Dispensing (UPD) technology of novel silver nanoparticle-based pastes as a simple and adaptable alternative to the metallization of TSVs process. The TSV filling process is outlined, followed by a detailed analysis of their morphology, filling quality, and electrical performance. We successfully achieve filled vias through a 280 μm thick silicon substrate with diameters down to 20 μm, resulting in an aspect ratio of up to 14:1, exhibiting favorable electrical properties. This work contributes to the achievement of dense, high-aspect ratio TSV fabrication using additive manufacturing, demonstrating a path towards reduced complexity of standard technology processes cycle, lower cost potential, and increased design flexibility.

Future radio access networks must accommodate growing mobile data traffic. Capacity between mobile devices and remote radio units (RRUs) must increase by using higher wireless carrier frequencies. Higher frequencies reduce reach, requiring RRU densification and high-capacity front- and backhaul connections. Optical fiber offers high throughput, but deployment can be expensive or unfeasible. Here we show an all-photonic sub-THz wireless link at 226 GHz over 1400 m, achieving a record-high net-rate-distance product of 214.2 Gbit s ⁻¹ km. Broadband photonic and plasmonic components enabled flat frequency response at high speeds. A novel dual-sideband receiver increased signal-to-noise ratio by 2 dB. We assessed power variations due to atmospheric turbulence; the scintillation index remained below 0.019 under strong turbulence, confirming sub-THz link resilience. A theoretical comparison with free-space optical links highlights the turbulence resistance. In this work, we show sub-THz links offer high capacity, resilience to weather and turbulence, and cost-effective deployment for wireless front- and backhaul.