The rapid rise of artificial intelligence (AI) is revolutionizing industries such as healthcare, robotics, and automation, creating a demand for more efficient, adaptive computing systems. In recognition of AI's profound impact, the 2024 Nobel Prize in Physics was awarded to John J. Hopfield and Geoffrey E. Hinton for their pioneering work in machine learning and artificial neural networks.^[1-3] This underscores the growing recognition within the scientific community of AI's transformative role in society. However, the energy-intensive nature of deep learning models and the computational demands of modern AI continue to outpace the capabilities of traditional silicon-based hardware, posing significant sustainability challenges as AI scales.^{[4, 5]}

To address these growing challenges, there is an increasing need for hardware that can meet the demands of next-generation AI. The concept of “more-than-more” calls for breakthroughs beyond traditional improvements, such as new materials and innovative architectures. Neuromorphic computing, inspired by the brain, offers a promising direction, providing biological-level energy efficiency and adaptability, making it ideal for next-generation AI hardware.^[6-9]

While organic materials may not yet offer a complete solution for all AI hardware needs, they hold significant promise in addressing critical challenges in neuromorphic computing. Their flexibility, biocompatibility, and low power consumption make them ideal for applications in wearable systems, bioelectronics, and bio-interfacing technologies—areas where traditional silicon-based devices face limitations.

These properties enable the development of Neuromorphic Organic Devices (NODs) that mimic synaptic functions like plasticity, learning, and memory storage. The intrinsic flexibility of organic materials, due to Van der Waals bonding, allows for time-dependent interactions, which are crucial for neuromorphic applications. Moreover, chemical modification can optimize properties, such as ionic conduction in materials like mixed ionic-electronic conductors, further enhancing their dynamic behavior.^{[10, 11]} Low-cost fabrication techniques, such as inkjet printing, also make organic materials a scalable solution for producing integrated neuromorphic devices. Despite the promise of organic materials, they face challenges such as lower carrier mobility and stability issues compared to traditional silicon-based devices. These limitations can hinder their use in high-speed, high-performance neuromorphic devices, and long-term stability remains a critical concern, especially in environments with exposure to oxygen, moisture, and light. Despite these challenges, ongoing research is exploring ways to improve their electrical performance, durability, and scalability, making them a promising area of development for the future of neuromorphic computing.

In recent years, the growing interest in organic materials for neuromorphic devices has gained significant momentum, attracting attention from the academic community across diverse fields, ranging from theoretical research to applied device development and material science. As these materials show promise in providing scalable solutions for neuromorphic computing, there is a pressing need to address both the challenges and opportunities they present. This growing demand for deeper exploration has highlighted the need for dedicated forums to discuss how organic neuromorphic devices can advance and the obstacles that remain in their development.

To unite the growing community of researchers and experts in the field of NODs, Professor Fabio Biscarini and collaborators launched the first NOD Workshop in 2019 in Ferrara, Italy. The inaugural event featured strong contributions from both device applications and biosensing, particularly in interfacing with biological systems, establishing the workshop's interdisciplinary foundation. Building on this success, the second edition, held in 2022 in Chania, Greece, led by Professor Paschalis Gkoupidenis, shifted its focus to the development and fabrication of neuromorphic devices, highlighting the role of organic materials and showcasing progress in material science.

In 2023, the third edition, held in Bad Schandau, Germany, under Dr. Hans Kleemann, emphasized theoretical modeling and simulation of neuromorphic devices. Sponsored by Advanced Electronic Materials, this special collection presents key works from the 2023 NOD workshop.

The most recent edition of the workshop, held in 2024 in Paris, was organized by Professor Laurie Calvet. It expanded discussions to advanced applications and systems integration, continuing to emphasize organic neuromorphic hardware, biosensors, and computational models, reflecting the growing impact of NODs in neuromorphic computing.

A strong theoretical foundation is essential for advancing neuromorphic computing, addressing challenges like ionic behavior, memory stability, and material variability. These studies explore scalability and efficiency, providing critical insights into neuromorphic systems.

Barbosa et al. (article number 202400481) investigate the ionic trapping mechanism in Electrochemical Neuromorphic Organic Devices (ENODes). They propose a 2D drift-diffusion model that describes the progressive filling and emptying of trapping sites as a key factor for achieving non-volatility. The authors validate their model experimentally, demonstrating its ability to predict memory stability across various device geometries. This work addresses a fundamental challenge in neuromorphic computing: achieving reliable, energy-efficient memory retention in complex environments.

Dreamer et al. (article number 202400373) present a novel approach to modeling Organic Electrochemical Transistors (OECTs). Their work highlights the impact of non-monolithic capacitance on device performance and proposes a method to accurately estimate parasitic resistance. By bridging theoretical predictions with experimental validation, the authors provide a reliable tool for optimizing OECTs in biosensing and neuromorphic applications. This study emphasizes the importance of refining device-level models to enhance system-level performance.

Calvet et al. (article number 202400515) employ machine learning to tackle variability in Organic Thin Film Transistors (OTFTs). Using multilayer perceptron (MLP) regression models, they simulate OTFT behavior with high accuracy, addressing one of the most persistent challenges in organic electronics. This work demonstrates the power of AI-driven modeling in improving device predictability and optimizing circuit design for neuromorphic systems. By applying advanced computational techniques, the study expands the theoretical toolbox available for organic neuromorphic devices.

Collectively, these works address critical aspects of neuromorphic computing by elucidating device behavior, offering solutions to challenges like memory stability and device variability. By bridging theoretical models with practical applications, they pave the way for more efficient and scalable neuromorphic systems.

The integration of neuromorphic devices into biosensing opens new possibilities for healthcare, environmental monitoring, and bioelectronics. A key future application of NOD devices is the development of tailored technologies to interface with biological systems. These efforts connect biosensing detection with neuromorphic organic devices (NODs), demonstrating their ability to enable precise sensing and signal processing. The following studies showcase significant advancements in this field.

Vyshniakova et al. (article number 202400509) introduce a hybrid ZnO-organic neuromorphic sensor for ammonia detection. By integrating a ZnO thin-film sensor with organic neuromorphic circuits, the device achieves high sensitivity and signal resilience in noisy environments. This innovation demonstrates the potential of hybrid systems for environmental monitoring, addressing challenges in signal processing and stability.

Rondelli et al. (article number 202400467) present a single-electrode organic neuromorphic device designed for dopamine sensing. The device emulates synaptic plasticity and offers real-time mapping of neurotransmitter levels in vivo. This study underscores the importance of bio-compatible neuromorphic sensors for advanced brain-machine interfaces, bridging bioelectronics with neuromorphic technology.

Rühl et al. (article number 202400748) focus on MoS₂ Electrolyte-Gated Field-Effect Transistors (EGFETs) for biosensing. MoS₂, a Transition Metal Dichalcogenide (TMDC), can be exfoliated into single atomic layers, enabling highly modifiable interfaces for sensing applications. Since conduction occurs only on the atomically thin layer, these devices hold significant potential for biosensing. However, their impact depends on addressing stability challenges in aqueous environments. This work advances performance optimization through protective designs and material improvements, positioning MoS₂ EGFETs as reliable tools for biosensing in challenging conditions.

These contributions emphasize the role of neuromorphic devices in bridging biological and electronic systems. By enhancing sensitivity, stability, and biocompatibility, they demonstrate the potential for transformative applications in healthcare and environmental monitoring.

Energy efficiency and scalable fabrication are critical for advancing neuromorphic computing to meet the demands of next-generation AI systems. Addressing these challenges requires innovations in device architecture, materials, and manufacturing processes, as highlighted by the following studies.

Fedorova et al. (article number 202400258) provide an in-depth review of Spiking Neural Networks (SNNs), emphasizing their energy efficiency compared to traditional Artificial Neural Networks (ANNs). They detail specific mechanisms by which SNNs leverage event-driven computation to drastically reduce energy consumption. The study highlights the role of memristive hardware in emulating synaptic plasticity and achieving real-time adaptive learning. Additionally, it discusses the potential for integrating biologically inspired feedback mechanisms into neuromorphic architectures, offering a pathway to sustainable AI systems. This review underscores the need to develop energy-efficient designs to meet the growing demands of AI while reducing environmental impact.

Gärisch et al. (article number 202400479) explore inkjet printing as a scalable fabrication method for neuromorphic devices using Organic Mixed Ionic-Electronic Conductors (OMIECs). This approach demonstrates the feasibility of scalable fabrication while reducing material waste and achieving performance comparable to traditional spin-coated devices. The study also highlights two-terminal devices, such as synaptic diodes, as innovative alternatives to three-terminal devices, broadening the possibilities for neuromorphic system design. Inkjet printing thus offers a pathway to large-scale, sustainable, and cost-effective production of neuromorphic devices.

In conclusion, the works presented here represent only a selection of the outstanding contributions from the NOD Workshop 2023. These studies underscore the inherently multidisciplinary nature of the field, where material science, device engineering, and theoretical modeling converge to push the boundaries of innovation. This dynamic interplay offers an excellent playground for the development and fabrication of novel technologies, paving the way for transformative advancements in neuromorphic devices and their applications.

We invite readers to join us at the next NOD Workshop, taking place from October 9-11, 2025 in Berlin, Germany. This event will continue the vibrant discussions and groundbreaking research that define the NOD workshop series.

人工智能 （AI） 的迅速崛起正在彻底改变医疗保健、机器人和自动化等行业，从而产生了对更高效、适应性更强的计算系统的需求。为了表彰人工智能的深远影响，John J. Hopfield 和 Geoffrey E. Hinton 获得了 2024 年诺贝尔物理学奖，以表彰他们在机器学习和人工神经网络方面的开创性工作。^[1-3]这突显了科学界越来越认识到人工智能在社会中的变革性作用。然而，深度学习模型的能源密集型性质和现代 AI 的计算需求继续超过传统硅基硬件的能力，随着 AI 的扩展，这将带来重大的可持续性挑战。^{[4， 5]}

为了应对这些日益增长的挑战，对能够满足下一代 AI 需求的硬件的需求越来越大。“more-than-more”的概念要求超越传统改进的突破，例如新材料和创新架构。受大脑启发的神经形态计算提供了一个有前途的方向，提供生物级的能源效率和适应性，使其成为下一代 AI 硬件的理想选择。^[6-9]

虽然有机材料可能尚未为所有 AI 硬件需求提供完整的解决方案，但它们在解决神经形态计算中的关键挑战方面具有重大前景。它们的灵活性、生物相容性和低功耗使其成为可穿戴系统、生物电子和生物接口技术应用的理想选择，而传统硅基设备在这些领域面临限制。

这些特性使神经形态有机设备 （NOD） 的开发成为可能，这些设备可以模拟突触功能，如可塑性、学习和记忆存储。由于范德华键合，有机材料固有的柔韧性允许时间依赖性的相互作用，这对于神经形态应用至关重要。此外，化学改性可以优化性能，例如混合离子电子导体等材料的离子传导，从而进一步增强其动态行为。^{[10， 11]}喷墨打印等低成本制造技术也使有机材料成为生产集成神经形态设备的可扩展解决方案。尽管有机材料前景广阔，但与传统的硅基器件相比，它们面临着载流子迁移率低和稳定性问题等挑战。这些限制会阻碍它们在高速、高性能神经形态器件中的使用，长期稳定性仍然是一个关键问题，尤其是在暴露于氧气、湿气和光线的环境中。尽管存在这些挑战，但正在进行的研究正在探索提高其电气性能、耐用性和可扩展性的方法，使其成为神经形态计算未来有前途的发展领域。

近年来，人们对用于神经形态器件的有机材料的兴趣日益浓厚，吸引了从理论研究到应用器件开发和材料科学等各个领域的学术界的关注。由于这些材料在为神经形态计算提供可扩展解决方案方面显示出前景，因此迫切需要解决它们带来的挑战和机遇。这种对更深入探索的日益增长的需求凸显了建立专门论坛的必要性，以讨论有机神经形态器件如何发展以及其发展中仍然存在的障碍。

为了团结 NOD 领域不断增长的研究人员和专家社区，Fabio Biscarini 教授和合作者于 2019 年在意大利费拉拉发起了第一届 NOD 研讨会。首届活动突出了设备应用和生物传感的巨大贡献，特别是在与生物系统的接口方面，为研讨会奠定了跨学科基础。在这一成功的基础上，2022 年在希腊哈尼亚举行的第二届大会由 Paschalis Gkoupidenis 教授领导，将重点转移到神经形态器件的开发和制造上，突出了有机材料的作用并展示了材料科学的进展。

2023 年，在德国巴特尚道举行的第三届研讨会由 Hans Kleemann 博士主持，强调神经形态设备的理论建模和仿真。该特别馆藏由 Advanced Electronic Materials 赞助，展示了 2023 年 NOD 研讨会的主要作品。

最近一届研讨会于 2024 年在巴黎举行，由 Laurie Calvet 教授组织。它将讨论扩展到高级应用程序和系统集成，继续强调有机神经形态硬件、生物传感器和计算模型，反映了 NOD 在神经形态计算中日益增长的影响。

坚实的理论基础对于推进神经形态计算、解决离子行为、内存稳定性和材料可变性等挑战至关重要。这些研究探讨了可扩展性和效率，为神经形态系统提供了关键见解。

Barbosa 等人（文章编号 202400481）研究了电化学神经形态有机器件 （ENODes） 中的离子捕获机制。他们提出了一个 2D 漂移-扩散模型，该模型将捕获位点的逐渐填充和排空描述为实现非挥发性的关键因素。作者通过实验验证了他们的模型，证明了它能够预测各种设备几何形状的内存稳定性。这项工作解决了神经形态计算中的一个基本挑战：在复杂环境中实现可靠、节能的内存保留。

Dreamer 等人（第 202400373 篇文章）提出了一种模拟有机电化学晶体管 （OECT） 的新方法。他们的工作强调了非单片电容对器件性能的影响，并提出了一种准确估计寄生电阻的方法。通过将理论预测与实验验证联系起来，作者为优化生物传感和神经形态应用中的 OECT 提供了可靠的工具。本研究强调了改进设备级模型以提高系统级性能的重要性。

Calvet 等人（文章编号 202400515）采用机器学习来解决有机薄膜晶体管 （OTFT） 的可变性问题。他们使用多层感知器 （MLP） 回归模型，以高精度模拟 OTFT 行为，解决了有机电子学中最持久的挑战之一。这项工作展示了 AI 驱动建模在提高设备可预测性和优化神经形态系统电路设计方面的力量。通过应用先进的计算技术，该研究扩展了可用于有机神经形态器件的理论工具箱。

总的来说，这些工作通过阐明设备行为来解决神经形态计算的关键方面，为内存稳定性和设备可变性等挑战提供解决方案。通过将理论模型与实际应用联系起来，它们为更高效和可扩展的神经形态系统铺平了道路。

将神经形态设备集成到生物传感中，为医疗保健、环境监测和生物电子学开辟了新的可能性。NOD 设备未来的一个关键应用是开发与生物系统接口的定制技术。这些努力将生物传感检测与神经形态有机器件 （NOD） 联系起来，展示了它们实现精确传感和信号处理的能力。以下研究展示了该领域的重大进展。

Vyshniakova 等人（文章编号 202400509）介绍了一种用于氨检测的 ZnO 有机杂化神经形态传感器。通过将 ZnO 薄膜传感器与有机神经形态电路集成，该器件在嘈杂环境中实现了高灵敏度和信号弹性。这项创新展示了混合系统在环境监测方面的潜力，解决了信号处理和稳定性方面的挑战。

Rondelli 等人（文章编号 202400467）提出了一种设计用于多巴胺感应的单电极有机神经形态装置。该设备模拟突触可塑性，并提供体内神经递质水平的实时映射。这项研究强调了生物相容性神经形态传感器对先进脑机接口的重要性，将生物电子学与神经形态技术联系起来。

Rühl 等人（文章编号 202400748）专注于用于生物传感的 MoS₂ 电解质门控场效应晶体管 （EGFET）。MoS₂ 是一种过渡金属二硫属化物 （TMDC），可以剥离成单个原子层，为传感应用提供高度可修改的接口。由于传导仅发生在原子薄层上，因此这些器件具有巨大的生物传感潜力。然而，它们的影响取决于解决水性环境中的稳定性挑战。这项工作通过保护设计和材料改进来推进性能优化，将 MoS₂ EGFET 定位为在具有挑战性的条件下进行生物传感的可靠工具。

这些贡献强调了神经形态器件在桥接生物和电子系统中的作用。通过提高灵敏度、稳定性和生物相容性，它们展示了在医疗保健和环境监测中变革性应用的潜力。

能源效率和可扩展的制造对于推进神经形态计算以满足下一代 AI 系统的需求至关重要。应对这些挑战需要在设备架构、材料和制造工艺方面进行创新，正如以下研究所强调的那样。

Fedorova 等人（文章编号 202400258）对脉冲神经网络 （SNN） 进行了深入的回顾，强调了与传统人工神经网络 （ANN） 相比，它们的能源效率。它们详细介绍了 SNN 利用事件驱动计算来大幅降低能耗的特定机制。该研究强调了忆阻硬件在模拟突触可塑性和实现实时自适应学习方面的作用。此外，它还讨论了将生物启发的反馈机制集成到神经形态架构中的潜力，为可持续的 AI 系统提供了一条途径。本综述强调了开发节能设计的必要性，以满足 AI 不断增长的需求，同时减少对环境的影响。

Gärisch 等人（文章编号 202400479）探索了喷墨打印作为使用有机混合离子电子导体 （OMIECs） 的神经形态器件的可扩展制造方法。这种方法证明了可扩展制造的可行性，同时减少了材料浪费并实现了与传统旋涂设备相当的性能。该研究还强调双端器件，例如突触二极管，作为三端器件的创新替代品，拓宽了神经形态系统设计的可能性。因此，喷墨打印为大规模、可持续且具有成本效益的神经形态设备生产提供了一条途径。

总之，这里展示的作品仅代表 2023 年 NOD 研讨会的杰出贡献的一部分。这些研究强调了该领域固有的多学科性质，材料科学、器件工程和理论建模融合在一起，以突破创新的界限。这种动态的相互作用为新技术的开发和制造提供了绝佳的平台，为神经形态设备及其应用的变革性进步铺平了道路。

我们邀请读者参加我们于 2025 年 10 月 9 日至 11 日在德国柏林举行的下一届 NOD 研讨会。本次活动将继续定义 NOD 研讨会系列的热烈讨论和开创性研究。