Nanowires and Molecular Dyes for Artificial Light-Driven Synapses and Neurons

Abstract

The development of neuromorphic computing requires devices that can use low power consumption and highly parallel architecture to process and store information, similar to the biological nervous systems. Molecular photoswitches, especially Donor–Acceptor Stenhouse Adducts (DASAs), offer a promising platform for such applications, because they undergo reversible, light-driven structural changes and show wavelength-dependent optical responses. However, to transform their molecular behavior into functional device structures, it is necessary to have a deeper understanding of how these molecules interact with nanostructure substrates, and how their optical and electrical signals evolve under controlled illumination.

This thesis studies the properties of DASA molecules AP-048 and AP-066 on both sapphire substrates and InP nanowire arrays. First, their bleaching and recovery dynamics were examined in solution and thin films to establish baseline behavior and to evaluate how casting conditions affect optical stability. Then, mixed-dye films were studied under dual-wavelength illumination to assess whether independent channels could be defined in a single molecular layer. The results show that although the spectral overlap is relatively small, the two dyes can still maintain a distinguishable response under selective excitation, and a small spectral overlap may introduce a weak additional modulation.

When integrated with vertical InP nanowire arrays, the DASA layer exhibits enhanced signal amplitude and clear photogenic electrical signal reading. By combining green and red Write/Read pulses, four different readout channels can be distinguished. Channels directly exposed to the writing wavelength display exponential decay and recovery, which is consistent with the reversible photochemical kinetics. In contrast, the indirect channels only show very small, almost linear changes. These weak residual changes are more in line with optical scattering, moderate spectral overlap, or measured background drift. According to current data, they cannot be directly attributed to molecular transitions. The comparison highlights that the dual-dye system can tolerate the slight cross-effect inherent in broadband lighting while maintaining the independence of the functional channel.

Overall, these results demonstrate that DASA-nanowire device provides a viable platform for multi-wavelength optical memory and synaptic-like functions. The combination of fast and slow response of AP-066 and AP-048 provides a variety of time dynamics in a hybrid system, suggesting the potential for scalable photonic neuromorphic devices. Future improvements may include optimizing molecular alignment, using a narrower-band excitation source, and designing nanowire geometry to further reduce crosstalk and improve device performance.

Popular Abstract

Before discussing this research, it helps to look at a familiar phenomenon in daily life: why does your laptop’s fan suddenly roar like a jet engine when you open several programs at once? It is not struggling with the tasks, but it has to shuttle data back and forth between the processor and the memory, and this movement is inherently inefficient.

Nature, however, works differently. Even a small creature like a bee can quickly judge whether a flower is worth approaching or if a place is dangerous. Its secret is that processing and memory usually occur in the same physical location. Synapses change their connection strength according to recent events, and the experience is recorded in subtle structural changes. There is no need to move information around; it naturally flows through the system.

This contrast has prompted scientists to ask: could materials also learn from biological synapses, changing their internal state after a stimulus and retaining it for some time? Light is a fascinating candidate for such systems. It is fast, does not interfere much with other signals, and can even “speak different languages” through different wavelengths and polarization.

Photoreversible molecules such as DASA undergo reversible structural changes under light and slowly return to their original state once the light is removed. This “bleach-and-recovery” behavior resembles a short-term memory. Earlier studies on light-responsive synapses often began with single-molecule systems. This research expands on that basis: if two molecules sensitive to different spectral regions are mixed in the same thin film, will two independent optical memory traces appear? Can the material “understand” multiple light inputs at the same time?

To explore this, we chose two types of molecules with different “personalities”, each preferring a different spectral region. After mixing them, we deposited the film on a semiconductor nanowire array. This “forest of nanowire light catchers” enhances the local optical field, amplifies small molecular changes, and converts them into electrical signals so we can read the traces left by light.

When the film is illuminated with different spectral components, different molecular groups are activated, leaving their own optical memories. Although the two responses show some mutual influence, they remain distinguishable within the same material. In other words, a single device now has two different input channels. It is also observed that the polarization of the light affects the response. Light has not only “color” but also “direction”, and if the material can sense this difference, future devices may encode information through both wavelength and polarization.

Although this is still an early prototype, it shows the potential of light-driven synapses in future brain-like computing. As data volumes grow and energy becomes a bottleneck, if materials can perform both “computation + memory” internally, it will help reduce energy consumption and pave the way for new computing architectures.

• author: He, Xinyu
• supervisor: Anders Mikkelsen
• organization: Synchrotron Radiation Research; Department of Physics
• course: PHYM03 20251
• year: 2026
• type: H2 - Master's Degree (Two Years)
• subject: Physics and Astronomy
• keywords: neuromorphic computing, synapses, neurons, photoswitches, nanowires, multichannel
• language: English
• id: 9218720
• date added to LUP: 2026-02-02 15:01:50
• date last changed: 2026-02-02 15:01:50