Symposium F.SM05—Brain-Inspired Information Processing—From Novel Material Concepts for Neuromorphic Computing to Sensing, Manipulation and Local Processing of Biological Signals

As artificial intelligence (AI) becomes a popular feature in industry, social networks, smart homes and transportation, there is an increasing need for AI accelerators in edge computing, to solve for excessive traffic to the cloud and ensure privacy and security of the processed data. In this frame, in-memory computing (IMC) has risen as low-power processing approach for data intensive problems. Matrix-vector multiplication (MVM) can be accelerated in crosspoint arrays by leveraging the Kirchhoff’s law for current summation and the Ohm’s law for voltage-conductance multiplication, as well as taking advantage of the parallelism in the memory array. Connecting the crosspoint array in a feedback loop also allows to accelerate inverse operations, i.e., one-step solutions of linear systems such as Ax = b where the NxN matrix A is stored as conductance values in the array, vector b represents applied currents and solution vector x is the output voltage. O(1) complexity of eigenvector extraction and O(N) complexity of matrix inversion have been demonstrated, which outperforms classical digital computing and even quantum computing.
In this work, we show that the same circuit concept can serve to accelerate regression and classification problems, which are two key problems of AI. Linear regression is solved within two coupled crosspoint arrays by computing the pseudo-inverse matrix in one step. M-dimensional linear regression allows to make predictions based on an existing data set or temporal sequence, thus enabling fast, energy efficient machine learning on the edge. The same circuit enables logistic regression, hence one-step classification within the crosspoint array, which allows for one-shot training of multiple layer perceptrons (MLP). The implications of these novel concepts for AI acceleration and the best technological implementation for data-intensive computing will be discussed.

This talk will present our latest research adventures on 2D atomic nanomaterials towards greater scientific understanding and computing applications. In particular, the talk will highlight our work on non-volatile memory (atomristors) that can be employed in various applications including information storage, brain-inspired computing, and zero-power RF switches. Non-volatile memory device based on atomically-thin materials is an application of defects, and is a rapidly advancing field with rich physics that can be attributed to metal ion binding to native defects. The promising 2D materials for neuromorphic computing include MoS2 and hBN. In particular, hBN memory devices feature very low switching energy, a relatively wide range of stable programmable resistance states, and good endurance and retention. Recent progress has also demonstrated wafer-scale integration of 2D memory devices with a prospect to leverage silicon electronics for fully integrated systems.

Artificial electrochemical synapses based on the charge-driven proton intercalation demonstrate much promise toward brain-inspired neural computing [1]. Proton transport plays a critical role in these devices determining switching speed and reliability. Despite great attention and intensive experimental investigations, even the cutting-edge proton conductors still suffer from sluggish ion transport kinetics at low temperatures or are not compatible with current CMOS fabrication processes. Therefore, there is an urgent need for fast solid-state proton conductors operating below 100 °C, much required for rapidly growing proton-based neuromorphic computing applications.

In this work, we consider a large class of solid-state materials having hydrogen as a part of their nominal structure such as solid acids and inorganic hydrates. By considering different combinations of common acids and metal ions, it is possible to synthesize thousands of materials with widely varying structures and physicochemical properties, and yet many of them remain relatively unexplored for the ability to conduct protons. We performed a high-throughput computational screening of the available materials datasets to identify chemical compositions and structure types that are likely to have high proton mobility. To estimate proton conductivity, we used ab initio molecular dynamics simulations. We further correlated this data with simple structural features and built a classification model to preselect a small set of materials for more detailed investigations. Based on our simulations, we identified promising materials and suggested effective strategies for improving proton conductivity.

References
[1] Xiahui Yao, Konstantin Klyukin, Wenjie Lu, Murat Onen, Seungchan Ryu, Dongha Kim, Nicolas Emond, Iradwikanari Waluyo, Adrain Hunt, Jesus A. del Alamo, Ju Li, Bilge Yildiz, “Protonic Solid-State Electrochemical Synapse for Physical Neural Networks”, Nature Communications, in press, 2020.

Strongly correlated metal oxide (SCMO), such as manganite, nickelate, and so on show first-order metal-insulator phase transition (MIT) with a change in conductivity by several orders magnitude. Since a discovery of the phase separation during MIT process, many researchers have been trying to capture a nanoscale electric domain and investigate its exotic properties. Typical SCMO material, rare earth Ni oxides (ReNiO₃, Re = Nd, Sm, and Eu) having a perovskite structure, show giant resistance change due to MIT at a wide temperature range of 150 K to 500 K. In this family, bulk neodymium nickelate: NdNiO₃ (NNO) shows 1-2 orders resistance change through MIT at around 200 K [1]. Very recently, the coexistence of insulator and metal domains with a few hundred nm size was observed during the MIT process for NNO system [2]. This phase-separated phenomenon is also observed in typical SCMOs, such as VO₂ and manganite. The overall observed MIT property, i.e., the resistance change for a sample, is considered to be dominated by competing nanoscale electronic phases. A step resistance change due to the electronic phase confinement effect: increase in resistance change rate has been reported in nanoscale samples comparable to domain size [3]. Therefore, it is expected that the nanoscale NNO sample shows a step resistance change due to MIT property of single nanoscale electronic phases. In this study, we fabricated NNO nanowire structures with 50-200 nm widths and measured their transport properties.
The NNO thin films were fabricated on NdGaO₃(110)(a_pc = 0.3858 nm), LSAT(100)(a = 0.3868 nm), and SrTiO₃(100)(a = 0.3858 nm) substrates by pulsed laser deposition method. NNO nanowires with different wire widths from 50 nm to 500 nm were fabricated by electron beam (EB) lithography and Ar ion milling process. All films and nanowire samples showed the resistance changes corresponding to MIT, but the RT curves shape were quite different. While the thin film sample showed gradual resistance change, the nanowire sample showed some step resistance changes both in cooling and heating processes. This step changes indicate that the nano electronic phases are confined in the NNO nanowire and a single electric domain shows first-order MIT with a single-step change in electrical transport property. For example, we found that the 50 nm width nanowire on SrTiO₃ substrate showed the intermittent step-resistance change, implying the further confine electronic phases to a nanometer space. Interestingly, the nanowires on NdGaO₃(110) and LSAT(100) substrates showed more prominent step resistance changes than that on SrTiO₃(100) substrate, which imply the nano-confinement condition were affected by the strain. The dR/dT vs. T plot showed that the transition temperature distribution of each electronic phase in the nanowires were different. This is probably because the strains depending on the three different substrate effect where on the Ni-O-Ni angular distribution, which affected the MIT properties of NdNiO₃ nano-electronic phase. In the presentation, we will present the quantitative correlation between nano-spatial size and nano-confinement MIT properties. We will also discuss the revealed the MIT property for a single domain in a NNO nanowire and the strain effect for the nano-confined domains behaviors.
[1] G. Catalan et al., Phys. Rev. B, 62, 7892 (2000), [2] G. Mattoni et al., Nat. Commun., 7, 13141 (2016), [3] A. N. Hattori et al., Nano Lett., 15, 4322 (2015).

Much progress has been made over the past decade using conductive bridge random access memory (CBRAM) to emulate both artificial synapses and artificial neurons for neuromorphic applications. However, the intrinsically stochastic switching process of metal filament formation and dissolution results in unreliable analog synaptic weight updating, while the non-volatile characteristic of CBRAM prevents the leaky integrate-and-fire function—a hallmark characteristic of an artificial neuron. We have previously reported that silver nanofilament growth dynamics in a polymer electrolyte can be regulated by the addition of ionic liquid (IL) by: (1) reducing stochasticity of the filament switching (shifting its formation time distribution from power-law to Gaussian) and (2) creating a competition between the formation of IL electrical double layers (EDL) and silver electrodeposition. The competition enables the tuning of filament growth dynamics from abrupt to gradual, with gradual growth introducing multiple well-defined resistance states—a feature necessary for emulating the communications between neurons in an artificial neural network. We also observed a linear correspondence (R² = 0.95) between nanofilament resistance states and program pulses, which is essential to improve learning accuracy in Deep Neural Networks (DNNs). Here, we report on silver filament formation and dissolution in response to presynaptic programming biases and focus on benchmarking the performance against the requirements for artificial synapses. The temporal stability of “abrupt growth” silver filaments is also studied. Lastly, we demonstrate a system integrating both gradual and abrupt silver filaments to implement some key artificial neuron functionalities.

Neuromorphic computing with analog memory can accelerate deep neural networks (DNNs) by enabling multiply-accumulate (MAC) operations to occur within memory. Analog memory, however, presents a number of device-level challenges having macro-implications on the achievable accuracy and reliability of these artificial neural networks. This talk focuses on the adverse effects of conductance drift in phase-change memory (PCM) on network reliability. It is shown that conductance drift can be effectively compensated in a variety of networks by applying a ‘slope correction’ technique to the squashing functions to maintain accuracy/reliability for a period of ∼1 year. In addition to conductance drift, PCM poses considerable variability challenges, which impact the accuracy of the initial weights. This talk summarizes recent advances in optimizing initial weight programming, and provides evidence suggesting that the combination of ‘slope correction’ and programming optimization techniques may allow DNN acceleration using analog memory while maintaining software-equivalent accuracy with reasonable reliability.

Memristive devices have become a promising candidate for unconventional computing(1) due to their attractive properties(2, 3). Computing can be implemented on a Resistive Neural Network (ResNN) with memristor synapses and neurons or a Capacitive Neural Network (CapNN) with memcapacitor synapses and neurons. For ResNNs as computing accelerators, we have built a dot-product engine based on a 128 x 64 1T1R crossbar array using traditional non-volatile memristors with 64 stable analog resistance levels(4). Accurate image compression and filtering have been demonstrated with such analog computing accelerator(4). In addition, we have demonstrated efficient and self-adaptive in-situ learning in a two-layer neural networks using such memristive arrays(5), which is expected to significantly improve the speed and energy efficiency of machine learning. A variety of neural networks based on such accelerators have been demonstrated, including reinforcement learning(6), LSTM(7), recurrent CNN(8) memristive networks. For ResNNs beyond accelerator applications, we developed diffusive memristors(9) with diffusion dynamics that is critical for neuromorphic functions. Based on the diffusive memristors, we have further developed artificial synapses(10) and neurons(11) to more faithfully emulate their bio-counterparts. We then integrated these artificial synapses and neurons into a small neural network, with which pattern classification and unsupervised learning have been demonstrated(11). For CapNNs, we have developed pseudo-memcapacitive devices based on the diffusive memristors. Capacitive synapses and neurons enabled by these memcapacitive devices have been developed and used to form a fully integrated CapNN(12), which has shown spiking signal classification and Hebbian-like learning functions. Moreover, the diffusive memristors can be used for cybersecurity(13, 14) and robotics applications(15). The requirements on device properties, and thus the corresponding challenges and solutions, are different for ResNNs and CapNNs, which will be discussed in detail in this talk. J. J. Yang, D. B. Strukov, D. R. Stewart, Memristive devices for computing. Nature Nanotechnology 8, 13-24 (2013). S. Pi et al., Memristor crossbar arrays with 6-nm half-pitch and 2-nm critical dimension. Nature nanotechnology 14, 35 (2019). Q. Xia, J. J. Yang, Memristive crossbar arrays for brain-inspired computing. Nature materials 18, 309 (2019). C. Li et al., Analogue signal and image processing with large memristor crossbars. Nature Electronics 1, 52 (2017). C. Li et al., Efficient and self-adaptive in-situ learning in multilayer memristive neural networks. Nature communications 9, 2385 (2018). Z. Wang et al., Reinforcement learning with analogue memristor arrays. Nature Electronics 2, 115 (2019). C. Li et al., Long short-term memory networks in memristor crossbar arrays. Nature Machine Intelligence 1, 49 (2019). Z. Wang et al., In situ training of feed-forward and recurrent convolutional memristor networks. Nature Machine Intelligence 1, 434-442 (2019). E. J. Fuller et al., Parallel programming of an ionic floating-gate memory array for scalable neuromorphic computing. Science 364, 570-574 (2019). Z. Wang et al., Memristors with diffusive dynamics as synaptic emulators for neuromorphic computing. Nature Materials 16, 101-108 (2017). Z. Wang et al., Fully memristive neural networks for pattern classification with unsupervised learning. Nature Electronics 1, 137-145 (2018). Z. Wang et al., Capacitive neural network with neuro-transistors. Nature Communications 9, 3208 (2018). H. Jiang et al., A Novel True Random Number Generator Based on a Stochastic Diffusive Memristor. Nature communications 8, 882 (2017). H. Jiang et al., Provable Key Destruction with Large Memristor Crossbars. Nature Electronics 1, 548 (2018). J. H. Yoon et al., An artificial nociceptor based on a diffusive memristor. Nature Communications 9, 417 (2018).

Memristive devices¹ have become a promising candidate for energy-efficient and high-throughput unconventional computing², which is a key enabler for artificial intelligent systems in the big data and IoT era³. The computing can be implemented on a Resistive Neural Network⁴ with memristive synapses⁵ and neurons⁶ or a Capacitive Neural Network^7,8 with memcapacitive synapses and neurons. In this talk, I will first briefly introduce the promises and challenges of memristive materials and devices for such applications. I will then discuss a few examples selected from our recent experimental demonstrations of unconventional computing using memristive networks with different levels of bio-inspiration: first, deep learning accelerators⁹ with supervised online learning¹⁰; second, neuromorphic computing for pattern classification with unsupervised learning⁶; last, other computing applications, such as reinforcement learning¹¹ for decision making, artificial nociceptors for robotics¹², provable key destruction¹³ and true random number generators¹⁴ for cybersecurity.
1. Yang, J. J., Strukov, D. B. & Stewart, D. R. Memristive devices for computing. Nature Nanotechnology 8, 13-24 (2013).
2. Xia, Q. & Yang, J. J. Memristive crossbar arrays for brain-inspired computing. Nature materials 18, 309-323 (2019).
3. Wang, Z. et al. Resistive switching materials for information processing. Nature Reviews Materials, 1-23 (2020).
4. Wang, Z. et al. In situ training of feed-forward and recurrent convolutional memristor networks. Nature Machine Intelligence 1, 434-442 (2019).
5. Wang, Z. et al. Memristors with diffusive dynamics as synaptic emulators for neuromorphic computing. Nature Materials 16, 101-108 (2017).
6. Wang, Z. et al. Fully memristive neural networks for pattern classification with unsupervised learning. Nature Electronics 1, 137-145 (2018).
7. Wang, Z. et al. Capacitive neural network with neuro-transistors. Nature Communications 9, 3208 (2018).
8. Li, C. et al. Long short-term memory networks in memristor crossbar arrays. Nature Machine Intelligence 1, 49-57 (2019).
9. Li, C. et al. Analogue signal and image processing with large memristor crossbars. Nature Electronics 1, 52 (2018).
10. Li, C. et al. Efficient and self-adaptive in-situ learning in multilayer memristive neural networks. Nature communications 9, 2385 (2018).
11. Wang, Z. et al. Reinforcement learning with analogue memristor arrays. Nature Electronics 2, 115 (2019).
12. Yoon, J. H. et al. An artificial nociceptor based on a diffusive memristor. Nature communications 9, 417 (2018).
13. Jiang, H. et al. A provable key destruction scheme based on memristive crossbar arrays. Nature Electronics 1, 548-554 (2018).
14. iang, H. et al. A Novel True Random Number Generator Based on a Stochastic Diffusive Memristor. Nature communications 8, 882 (2017).

The Quantum Materials for Energy Efficient Neuromorphic Computing (QMEENC) Energy Frontiers Research Center (EFRC) is a highly collaborative effort dedicated to the development of quantum materials based implementation of energy efficient devices that will form the basis for neuromorphic computing. This project is currently two parallel approaches based on charge, using highly correlated oxides, and spin, using spin torque oscillators. We are also developing hybrid concepts, which take advantage of the functional properties of both approaches.
In addition to the general description of the QMEENC approach, I will describe the first implementation of energy efficient neuristors (that emulates neuronal functionalities) and a synaptors (that emulates synaptic functionalities), based on heterostructures, which incorporate strongly correlated oxides.

This work was supports as part of the Quantum Materials for Energy Efficient Neuromorphic Computing an Energy Frontiers Research Center funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under Award # DE-SC0019273.

Artificial intelligence (AI) has been heralded as the flagbearer of the fourth industrial revolution. But it comes with a cost and that is computing power. It is projected that by 2040, we will need more computing energy than the total energy we can produce now. So, we need devices that can offer higher computing/ storage density with low energy consumption like neuronal computation. We are addressing these challenges using a molecular-electronic route. Historically, organic electronic devices have stimulated scientific excitements in OLEDs but are yet to make any other significant technological impact. The reasons behind their limited success are their poor robustness, stability, endurance and most importantly, the lack of mechanistic understanding that restricts the emergence of approaches to solve these problems. We have overcome each of these difficulties in our memristors based on transition metal complexes of azo-aromatic ligands that exhibit high reproducibility (~350 devices), fast switching (≤30 ns), excellent endurance (~10¹² cycles), stability (>10⁶ s) and scalability (down to ~60nm²)^1-5. Using in-situ Raman spectroscopy we are able to track the electronic changes in molecules in-operando at every point of our voltage sweep providing a clear picture of our molecular mechanism that enables us to do different molecular and device engineering to achieve targeted functionalities. Using devices of this genre we are addressing the existing computing challenges via three routes,
By designing devices with ultra-low power: We can design memristors with switching voltage as low as 70mV, with energy ~36aJ/ 60nm². The current and voltage levels of these devices meet the requirements specified in ITRS road map.
By designing memristors and memcapacitors with multiple discrete plateaus³: We have developed memristors with 3- 4 distinct conducting plateaus which also shows mem-capacitance. Their concomitant occurrence is enabled by symmetry breaking of our film-molecules driven by voltage, a new paradigm in condensed matter physics.
Brain inspired computing: Using devices that exhibit concomitant memristive and memcapacitive functions we can simulate biological actions such as neuronal action potential and even cardiac myocyte pulsing.
References
1. Goswami S, Matula AJ, Rath SP, Hedström S, Saha S, Annamalai M, et al. Robust resistive memory devices using solution-processable metal-coordinated azo aromatics, Nature materials 2017, 16(12): 1216.
2. Valov I, Kozicki M. Non-volatile memories: Organic memristors come of age, Nature materials 2017, 16(12): 1170.
3. Sreetosh Goswami, Santi P. Rath, Damien Thompson, Svante Hedström, Meenakshi Annamalai, Rajib Pramanick, B. Robert Ilic, Soumya Sarkar, Christian A. Nijhuis, Jens Martin, Sreebrata Goswami, and T. Venkatesan, A Ternary Resistive Memory Device Based on Charge Disproportionate Molecular Redox, Nature Nanotechnology, (2020)
4. Sreetosh Goswami, Sreebrata Goswami and T. Venkatesan, An Organic approach to low energy memory and Brain Inspired Electronics, 7(2), 021303 Applied Physics Reviews- Invited (2020)
5. Sreetosh Goswami, R. Stanley Williams, Damien Thompson, Sreebrata Goswami and T. Venkatesan, Colossal current and voltage tunability in an organic memristors via electrode engineering, Applied Materials Today (2020)

In biological neural networks communication relies on electric signals in the form of action potentials (APs) that trigger the release of neurotransmitters at the synapses, thus influencing the mechanisms of synaptic plasticity that regulates over time the transmission efficiency across the synaptic cleft. Indeed, progress over the last few years designing artificial neuromorphic systems (ANS)¹ has paved the way to a rapid improvement of the current machine learning (MC) processes but, focusing mainly on replicating electrical signal transmission, has limited a direct communication with biological systems². Electrical to chemical signal transduction is essential to fully replicate brain functions and establish in future an active interaction with biological tissues³. Biohybrid interfaces are emerging since they promise to couple the activity of biological neural cells with neuromorphic artificial devices, enabling sensing and stimulation through two ways communication.
In this context, the conductive polymer mixture PEDOT: PSS can be exploited for the electrochemical detection of electroactive neurotransmitters⁴ (i.e. dopamine, norepinephrine, ascorbic acid) through redox reactions while providing a suitable biocompatible interface with cells and neurons⁵. Here, we present the design of a novel electrochemical neuromorphic organic device (ENODe) capable of identifying multiple neurotransmitters simultaneously present. Additionally, through modular microfluidic channels, different flow regimes are regulated in order to replicate the condition of exocytosis displayed by a real synapse. Hence, the platform allows to transduce the neurotransmitters-mediated chemical signal, eliciting an output that is stored and read as a change in the polymer’s conductance, a neuromorphic phenomenon that replicates neuronal’s short-term and long-term plasticity⁶. Moreover, by exploiting PEDOT:PSS electrochromism, the system’s readout is alternatively monitored as a variation of the polymer’s transmittance and since this is intimately related to its conductance modulation, a novel dual output system to refine feedback regulation in closed-loop operations is presented.
Hence, this platform allows to fully recapitulate synaptic plasticity and being conceivable as a bio-hybrid interface through direct coupling with neural tissues, it promises to move from a mere recording of biological functions to a dynamic interactive control.⁷

1. Furber, S. Large-scale neuromorphic computing systems. J. Neural Eng. 13, 051001 (2016).
2. van Doremaele, E. R. W., Gkoupidenis, P. & van de Burgt, Y. Towards organic neuromorphic devices for adaptive sensing and novel computing paradigms in bioelectronics. J. Mater. Chem. C 7, 12754–12760 (2019).
3. Markram, H. A history of spike-timing-dependent plasticity. Front. Synaptic Neurosci. 3, (2011).
4. Gualandi, I. et al. Textile Organic Electrochemical Transistors as a Platform for Wearable Biosensors. Sci. Rep. 6, 33637 (2016).
5. Asplund, M. et al. Toxicity evaluation of PEDOT/biomolecular composites intended for neural communication electrodes. Biomed. Mater. 4, 045009 (2009).
6. Abbott, L. F. & Regehr, W. G. Synaptic computation. Nature 431, 796–803 (2004).
7. Keene, S. T., Lubrano, C., Kazemzadeh, S., Melianas, A., Tuchman, Y., Polino, G., Scognamiglio, P., Cinà, L., Salleo, A., van de Burgt, Y., Santoro, F. A biohybrid synapse with neurotransmitter-mediated plasticity. Nat. Mater. (2020).

Organic synaptic transistors using intrinsic organic semiconductors have emerged to overcome the limitation of conventional electronics by demonstrating various synaptic functions. Although organic synaptic transistors have succeeded in mimic biological synapses, especially short-term plasticity, devices emulated only limited long-term plasticity with short retention behaviors. The long-term retention behavior is essential for non-volatile and long-term memory characteristic in neuromorphic computing. Correlating the synaptic responses with the microstructures of polymer semiconductor is a crucial step to achieve long-term retention. The fundamental study on synaptic responses related to microstructures of organic materials has not been discovered yet. Here, we show that the long-term retention in ion-gel gated organic synaptic transistors (IGOSTs) can be achieved by controlling the microstructure of organic semiconductors. We controlled the crystallinity of poly(3-hexylthiophene-2,5-diyl) (P3HT) in films spun-cast on bare and self-assembled monolayer, before and after thermal treatments. The long-term retention has been significantly prolonged in the IGOSTs, which tends to be elongated as the crystallinity of the semiconductor increased. The microstructure and crystallinity are determined by grazing-incidence X-ray diffraction, optical analyses, and atomic force microscopy. We also evaluate the synaptic current decay behaviors according to a de-doping mechanism of the polymer semiconductor over time. We simulated the recognition of handwritten digits of IGOSTs based on highly-crystalline P3HT, thus achieved higher classification accuracy (>92%) compared with IGOSTs based on low-crystalline P3HT. Our study provides fundamental information about the effects of microstructure on the synaptic responses and strategy to design the IGOSTs for neuromorphic electronics.

The brain can perform massively parallel information processing while consuming only ~1- 100 fJ per synaptic event. I will describe a novel electrochemical neuromorphic device (ENODe) that switches at record-low energy (<0.1 fJ projected, <10 pJ measured) as well as voltage (< 1mV, measured), and displays a large number of distinct, non-volatile conductance states within a ~1 V operating range. The tunable resistance behaves very linearly, allowing blind updates in a neural network when operated with the proper access device. ENODes also display outstanding endurance achieving over 10⁹ switching events with very little degradation. I will describe our recent efforts at scaling and materials selection, allowing us to reach 20 ns write pulses and operation at high temperature (up to 120°C). These properties are very promising in terms of the ability to integrate with Si electronics to demonstrate online learning and inference. ENODes are electrochemical devices where gated proton drift induces changes in the electronic states of a semiconductor channel. The peculiarities of the physics of these devices will be discussed along with their consequences on device design and performance.

2D networks of interconnected nano-objets (nanowires, nanoparticles, molecules) are experimentally and theoretically explored to implement unconventional computing machines, and especially reservoir computing [1-9]. 2D networks of molecularly functionalized nanoparticles (NPs) (hereafter called NMN : nanoparticle molecule network) have emerged as an interesting approach in molecular electronics to understand fundamental electron transport mechanisms [10].
I will describe the study of NMN approaches using functional molecules self-assembled in high density 2D networks with topological structures that are intrinsically similar to the structure of a "reservoir computing". NMNs equipped with molecules which can change their electronic properties upon a given excitation (optical, chemical, ionic) are interesting to study multifunctional systems at the nanoscale (<100 nm).
I will discuss several key features of the electronic properties of these 2D networks of molecularly functionalized nanoparticles to assess their possible use for reservoir computing: highly non-linear electron transport, variability, complex/rich dynamics (harmonic, interharmonic, intermodulation distortions), co-tunneling, noise and plasmonic response [9-12]. I will also discuss the electron transport properties of some molecules of interest for the chemical and biochemical sensing with these NMNs reservoir computing approaches, combining sensing and computing in a single nanoscale device.
These approaches, without direct analogs in semiconductor nanoelectronics, would open new perspectives to molecular electronics in unconventional computing.

1. Tour, J. M.; van Zandt, W. L.; Husband, C. P.; Husband, S. M.; Wilson, L. S.; Franzon, P. D.; Nackashi, D. P. IEEE Trans. Nanotechnol. 2002, 1, (2), 100-109.
2. Sköldberg, J.; Wendin, G. Nanotechnology 2007, 18, (48), 485201.
3. Bose, S. K.; Lawrence, C. P.; Liu, Z.; Makarenko, K. S.; van Damme, R. M. J.; Broersma, H. J.; van der Wiel, W. G. Nature Nanotech 2015, 10, 1048-1053.
4. Roychowdhury, V. P.; Janes, D. B.; Bandyopadhyay, S.; Wang, X. IEEE Transactions on Electron Devices 1996, 43, 1688-1699.
5. Sillin, H. O.; Aguilera, R.; Shieh, H.-H.; Avizienis, A. V.; Aono, M.; Stieg, A. Z.; Gimzewski, J. K. Nanotechnology 2013, 24, (38), 384004.
6. Maass, W.; Natschläger, T.; Markram, H. Neural Computation 2002, 14, 2531-2560.
7. Jeager, H.; Haas, H. Science (New York, NY) 2004, 304, (5667), 78-80.
8. Tanaka, H.; Akai-Kasaya, M.; TermehYousefi, A.; Hong, L.; Fu, L.; Tamukoh, H.; Tanaka, D.; Asai, T.; Ogawa, T. Nature Communications 2018, 9, (1), 2693.
9. Viero, Y.; Guerin, D.; Alibart, F.; Lenfant, S.; Vuillaume, D. Adv. Func. Mater. 2018, 28, 1801506.
10. Liao, J.; Blok, S.; van der Molen, S. J.; Diefenbach, S.; Holleitner, A. W.; Schönenberger, C.; Vladyka, A.; Calame, M. Chem. Soc. Rev. 2015, 44, 999-1014.
11. Viero, Y.; Copie, G.; Guerin, D.; Krzeminski, C.; Vuillaume, D.; Lenfant, S.; Cleri, F. J. Phys. Chem. C 2015, 119, 21173-21183.
12. Stievenard, D.; Guerin, D.; Lenfant, S.; Leveque, G.; Nijhuis, C. A.; Vuillaume, D. Nanoscale 2018, 10, 23122–23130.

Organic electrochemical transistors (OECTs) are in a stage of rapid development as novel applications that make use of these versatile devices continue to emerge. OECTs are characterized by the coupling of both ionic and electronic inputs to modulate transistor channel conductance. This attribute renders OECTs ideal for interfacing electronics with biological systems, which make use of ionic and biochemical currents and gradients for signaling. Here we will present a new concept of OECT that evolves with use. The transistor channel can be formed, modulated, and obliterated under operation. The strength of the transistor response to a given stimulus can be modulated within a range that spans several orders of magnitude, introducing behaviors analogous to neuroplasticity into electronic systems. This evolvable transistor can be incorporated into a simple circuit that mimics classical conditioning. Also, we will present several biomimetic logic circuits that link sensory inputs to the growth of the transistor channel, thus to achieve higher order processes like self-regulation and coincidence detection. OECTs that physically and electronically evolve under operation will bring about a new paradigm of machine learning based on evolvable organic electronics.

Electropolymerization is an interesting bottom-up strategy to structure conducting materials at the micro/nano-scale in liquid phase that offers a wide morphological versatility. Since the geometry of these structures governs their electrochemical properties, it is fundamental to decipher the mechanisms that rule the polymer assembling upon the electrically-programmed growth to use this phenomenon as a neuro-inspired building block for unconventional information processing.
Herein, we investigate various electrical parameters of electropolymerization affecting the conducting polymer network geometry. We find that various structures such as dendrites, trees, fractals as well as low-fractality cables can be obtained between the two-wire electrodes, based on applied voltage amplitude, biasing symmetry, bias frequency, the concentration of monomers and electrode configurations. We qualitatively and quantitatively study the relationship between the electrical parameters affecting geometrical parameters of the conducting polymer network as well as electropolymerization dynamics through video and image processing. The systematic analysis shows that an increase in applied voltage leads to higher growth rate, more branches, and lower surface to bulk ratio. At the other hand, an increase in bias frequency leads to higher growth rate, smaller number of branches, and higher surface to volume ratio.
In order to model the experimental phenomena, we simulate a simplified version of the problem with two bipolar metal electrodes and an electrolyte filled with dilute moving particles at random positions, reflecting the monomers in the liquid phase. The wire electrodes are biased with AC frequency and the spacio-temporal potential map is evaluated by solving the Laplace equation. The motion of the monomer particles is controlled by the electric field and by random Brownian motion. The particles that happen to touch the electrodes are stuck to the electrodes, with a certain probability of sticking. The stuck particles are incorporated into the electrode, and the potential is recalculated for the motion of the particles;. The simulations are tested for different AC applied voltages, frequency, duty cycles, and voltage offsets. The increase in applied voltage leads to higher sticking probability; resulting in a higher growth rate, multiple branches, and higher density, which goes well with the experiments. Further, the effect of frequency, which is not so intuitive, shows that higher AC frequency, favors linear cable like growth, while lower frequency leads to more isotropic growth, while voltage offset and non-equal duty cycle lead to asymmetrical growth, in accordance with experiments. In addition, the effect of electrode spacing, different electrode designs, electrical pulse shapes and the concentration of particles, are also studied. The study helps in visualizing the motion of particles in different electrical conditions, which is not possible to probe experimentally. Some subtle experimental features, such as the effect of preferential growth on the tips, and broadening of the electrode before touching, are observed in the modeling studies. Thus, we find that the different network architectures are associated with different Laplace end diffusion fields governing the monomers motion and in turn electropolymerized network geometry. Such unconventional engineering route could have a variety of applications from neuromorphic engineering to bottom-up computing strategies.

Spiking Neural Networks (SNN) provide distributed computational capability and emulate the brain processing principles based on spike frequency and pulse width [1]. Because of being intrinsically compatible with brain signaling, one of the target applications of SNN hardware implementation, known as neuromorphic systems [2], are Brain-Computer Interfaces (BCI) [3]. With a handful of exceptions [4], such implementations have typically been implemented using silicon technology, which is physically hard and rigid, and non-biocompatible, resulting in a permanent damage to the brain's soft tissue. Organic electronics offer physical softness and flexibility, biocompatibility, and low-cost, large-area fabrication, to directly interface with biological tissues [5].

We report on the fabrication of a physically flexible, organic electronics, spiking integrate-and-fire Axon-Hillock (AH) neural circuit. Organic Field-Effect Transistors (OFETs) are fabricated on top of a flexible, 50 μm Polyimide substrate. OFETs’ architecture is bottom-gate top-contact, and fabricated through thermal deposition of Ti (3 nm)/Au (50 nm) as the gate electrode, 60 nm of Parylene as the dielectric, 30 nm of dinaphtho[2,3b:2’,3’-f]thieno[3,2-b]thiophene (DNTT) as the p-type and N,N’-bis(n-octyl)-x:y,dicyanoperylene3,4:9,10-bis(dicarboximide) (PDI8-CN2) as the n-type semiconductor. A 50 nm layer of Au serves as the source and the drain electrodes. A layer of Parylene encapsulates the structure. Non-circuit-integrated commercial resistors and capacitors are used to accelerate the characterization and the fabrication process. All integrated all-organic circuit is beyond the scope of this work and would increase the complexity of the process.

The results show that the AH circuit with 500 nA input current can produce spikes with a range of frequencies. Given the membrane capacitance (C_mem) of 15 nF and feedback capacitance (C_feedback) of 47 nF, the spiking frequency reaches 670 mHz. The frequency can be lowered by increasing capacitance values to C_mem = 33 nF and C_feedback = 100 nF, respectively, to 370 mHz. Decreasing capacitance to C_mem = 2.2 nF and C_feedback = 6.8 nF, will result in a higher spiking rate (5 Hz), but will also increase the influence of power line noise and parasitic capacitance. The frequency-current (F-I) transfer function shows the dynamics of the neuron. By increasing the input current, the F-I curve will saturate indicating an upper bound of frequency spikes. The computational ability of the neuron is also demonstrated with a basic summation of two unitary excitatory synaptic events. The static and dynamic power dissipation is recorded and is limited to 0.4 µW and 40 µW, respectively.

We believe that this work is the first demonstration of a functioning organic electronics integrate-and-fire spiking neuron, paving the way for future implantable BMIs. Beyond neuroscientific applications, this technology can also be used in slow processes, such as temperature measurement, which are intrinsically difficult for ultra-fast silicon devices to achieve, as well as physically flexible large-area applications.

[1] F. Rieke and D. Warland, “Spikes: Exploring the Neural Code,” MIT Press, 1999.
[2] R. Nawrocki, et al., “A mini review of neuromorphic architectures and implementations,” IEEE Transactions on Electron Devices, vol. 63 (10), 2016.
[3] F. Boi, et al., “A Bidirectional Brain-Machine Interface Featuring a Neuromorphic Hardware Decoder,” Frontiers in Neuroscience, vol. 10 (563), 2016.
[4] R. Nawrocki, et al., “Neurons in polymer: Hardware neural units based on polymer memristive devices and polymer transistors,” IEEE Transactions on Electron Devices, vol. 61 (10), 2014.
[5] G. Schwartz, et al., ”Flexible polymer transistors with high-pressure sensitivity for application in electronic skin and health monitoring,” Nature Communications, vol. 4 (1), 2013.

Neuromorphic electronics have handled complex real-world problems such as visual information, speech recognition, and body motion control with compressibility, fault tolerance, and high energy efficiency. These neuromorphic electronics have been investigated to develop artificial nervous system systems that mimic biological nervous systems. Mimicking the learning and memory functions of the human brain has been a major focus so far in neurological electronics. Further, the functional integration of artificial synapses with sensing/motor elements would be a core technology for applications of human-like robotics and neuroprosthetics, enabling the emulation of sensing and responding behaviors in biological systems. In biology, sensations such as vision, hearing, touch, taste, and smell and responding with muscles are related with peripheral nerves.
Herein, bio-inspired neuromorphic electronics for artificial nervous systems were covered. Artificial synapse-based sensory and motor electronic nerves are developed by integrating with sensory and motor organs. Firstly, pressure sensors (artificial mechanoreceptors), organic ring oscillators (artificial nerve fibers), and synaptic transistors were integrated to emulate the functions and operating principles of biological counterparts. By connecting this artificial pressure-sensory nerve to biological motor nerves in a detached cockroach leg, a hybrid reflex arc system was realized. In addition, a stretchable photo-sensitive sensorimotor nervous system composed of a photo-sensitive detector (an artificial photoreceptor) and a stretchable artificial synapse with a polymer actuator (an artificial muscle). In addition, a retina-inspired photo-sensitive synaptic transistor was developed by using ultraviolet-responsive 2-dimensional carbon nitride nanodot layers as a floating gate to selectively detect and process ultraviolet light exposure information. Lastly, for an artificial auditory system, organic artificial synapse was connected to a triboelectric sensor. Our bio-inspired neuromorphic systems can be used for neuromorphic computing, humanoid soft robots, and neural prostheses.

Neuromorphic elements have been predominantly solid-state devices which simulate the resistive and capacitive behaviors needed for neural networks and brain-inspired computing, but in non-brain-like ways. This talk will describe ways we are integrating lipid and polymer bilayer membranes with micro- and nanofabrication to develop fundamentally new types of neuromorphic elements that have the composition (biomolecules), structure (biomembranes), and switching mechanism (voltage-sensitive ion channels) of real biological synapses, and operate at lower power than the current state-of-the-art. Our devices consist of insulating, nm-thick lipid or polymer-based bilayer membranes that assemble at the interfaces of two or more aqueous droplets in oil, and that have demonstrated both memristive and memcapacitive behaviors, including memory resistance and capacitance, synaptic functions such as paired-pulse facilitation and depression, spike rate dependent plasticity, voltage-dependent inactivation and recovery, and charging hysteresis. These behaviors are linked to electrostriction, an electromechanical phenomenon that encompasses both electrowetting and electrocompression in the membrane, which are changes in membrane area and thickness due to charging in the presence of electric fields. Electrostriction results in a voltage-dependent capacitive susceptibility that replaces the more familiar concept of static capacitance, which, up to now, has dominated electrophysiological descriptions and characterizations of biomembranes. In this picture, biomembranes are not just equivalent RC circuits dependent only on ionic currents controlled by the conductance (resistance) of ion channels. Large capacitive currents from small voltage inputs can be generated as well for the development of neuromorphic computing elements exhibiting both short-term and long-term synaptic plasticity.

Self-assembled monolayers (SAMs) are widely used to engineer the surface properties of metals. The relatively simple and versatile chemistry of metal-thiolate bonds makes thiolate SAMs the preferred option in a range of applications, yet fragility and a tendency to oxidize in air limit their long-term use. In our recently published paper in Nature Materials, we report the formation of thiol-free self-assembled mono- and bilayers of glycol ethers, which bind to the surface of coinage metals through the spontaneous chemisorption of glycol ether-functionalized fullerenes. As-prepared assemblies are bilayers presenting fullerene cages at both the substrate and ambient interface. Subsequent exposure to functionalized glycol ethers displaces the topmost layer of glycol ether-functionalized fullerenes, and the resulting assemblies expose functional groups to the ambient interface. These layers exhibit the key properties of thiolate SAMs, yet they are stable to ambient conditions for several weeks, as shown by the performance of tunneling junctions formed from SAMs of alkyl-functionalized glycol ethers. Glycol ether-functionalized spiropyrans incorporated into mixed monolayers lead to reversible, light-driven conductance switching. Self-assemblies of glycol ethers are drop-in replacements for thiolate SAMs that retain all of their useful properties while avoiding the drawbacks of metal-thiolate bonds.^[1]
In a follow-up work, we demonstrate the reconfiguration of molecular tunneling junctions while in operation via the self-assembly of bilayers of glycol ethers. We use well-established functional groups to modulate the magnitude and direction of rectification in assembled tunneling junctions by exposing them to solutions containing different glycol ethers. Variable temperature measurements reveal that rectification occurs by a bias-dependent tunneling-hopping mechanism and that glycol ethers, besides being unusually efficient tunneling medium, behave identically to alkanes. We fabricated memory bits from crossbar junctions prepared by injecting eutectic Ga-In in microfluidic channels. Two 8-bit registers were able to perform logical AND operations on bit strings encoded into chemical packets as microfluidic droplets that alter the composition of the crossbar junctions through self-assembly to effect memristor-like properties. This proof-of-concept work demonstrates the potential for fieldable molecular-electronic devices based on tunneling junctions of self-assembled monolayers and bilayers.^[2]

[1] Qiu, X., Ivasyshyn, V., Qiu, L., Enache, M., Dong, J., Rousseva, S., Portale, G., Stöhr, M., Hummelen, J.C. and Chiechi, R.C. “Thiol-free self-assembled oligoethylene glycols enable robust air-stable molecular electronics.” Nat. Mater. 19, 330-337 (2020).
[2] Qiu, X., Rousseva, S., Ye, G., Hummelen, J.C. and Chiechi, R.C. “Stochastic computing via in operando modulation of rectification in molecular tunneling junctions.” ChemRxiv, 2020.

The interface between biological cells and non-biological materials has profound influences on cellular activities, chronic tissue responses, and ultimately the success of medical implants and bioelectronic devices. The optimal coupling between cells, i.e. neurons, and materials is mainly based on surface interaction, electrical communication and sensing.
In the last years, many efforts have been devoted to the engineering of materials to recapitulate both the environment (i.e. dimensionality, curvature, dinamicity) and the functionalities (i.e. long and short term plasticity) of the neuronal tissue to ensure a better integration of the bioelectronic platform and cells.
On the one hand, here we explore how the transition from planar to pseudo-3D nanopatterned inorganic and organic materials have introduced a new strategy of integrating bioelectronic platforms with biological cells under static and dynamic conditions. Although a spontaneous penetration does not occur, adhesion processes are such that a very intimate contact can be achieved. On the other hand, we investigate how organic semiconductors can be exploited for recapitulating electrical neuronal functions such as long term and short term potentiation. In this way, both the topology and the material functionalities can be exploited for achieving in vitro biohybrid platforms for neuronal network interfacing.

Most of today’s strategies to interface biology with electronic hardware are based on layered architectures where the front-end of sensing is optimized separately from the back-end for processing/computing signals. Alternatively, biological systems are capitalizing on distributed architecture where both sensing and computing are mix together and co-optimized. In this talk, we will present our strategy to implement bio-sensing of electroactive cells in a neuromorphic perspective. We will present how organic electrochemical transistors can be used to record electrical signals from neural cells. We will show various strategies capitalizing on the versatility of organic materials synthesis and organic device fabrication to tune and adapt the functionalities of such bio-sensors. We will then present how these strategies can be efficiently used to realize computing functions directly at the interface with biology. Notably, we will illustrate how a network of ionic sensors can implement the reservoir computing concept, a powerful neuromorphic computing approach of particular interest for dynamical signal processing.

Organic memristive devices contain a heterojunction of conducting polymer and solid electrolyte. Similarly to nervous systems, these elements combine memory and processing functions. Their working principle is based on the significant difference of the polyaniline resistance in reduced and oxidized states [1, 2]. These devices have been successfully applied for the construction of artificial neuron networks at the hardware level [3,4]. The device can be realized on flexible supports and its properties are similar to synapse behavior, what has allowed realization of electrical circuits, mimicking learning of simple animals (pond snail) [5] and classical conditioning (Pavlov dog learning) according to STDP-like algorithm [6,7].
Its biocompatibility was demonstrated by the realization of hybrid systems, where slim mould Physarum polycephalum was interfaced with organic memristive devices [8].
Long- and short-term potentiation and depression of these devices depends not only on the number of passed pulses, but also on the frequency of their application [9].
Finally, we will consider a first step towards the realization of synapse prosthesis, describing experiments on the coupling of two live nervous cells from the rat cortex via organic memristive device [10]. Nervous cells were contacted by patch clamp electrodes, allowing to avoid artifacts. It has been demonstrated that the plasticity of the device electrical properties is similar to that of chemical synapses in living beings.

1. V. Erokhin, T. Berzina, and M.P. Fontana. Hybrid electronic device based on polyaniline-polyethylenoxide junction. J. Appl. Phys., 97, 064501 (2005).
2. V. Erokhin and M.P. Fontana. Thin film electrochemical memristive systems for bio-inspired computation. J. Computational Theor. Nanosci., 8, 313-330 (2011).
3. V.A. Demin. , A.V. Emelyanov, S. Battistoni, G. Baldi, S. Iannotta, P.K. Kashkarov, and M.V. Kovalchuk. Hardware elementary perceptron based on polyaniline memristive devices. Org. Electronics, 25, 16-20 (2015).
4. A.V. Emelyanov, D.A. Lapkin, V.A. Demin, , S. Battistoni, G. Baldi, A. Dimonte, A.N. Korovin, S. Iannotta, P.K. Kashkarov, and M.V. Kovalchuk. First step towards the realization of a double layer perceptron based on organic memristive devices. AIP Adv., 6, 111301 (2016).
5. V. Erokhin, T. Berzina, P. Camorani, A. Smerieri, D. Vavoulis, J. Feng, and M.P. Fontana. Material memristive device circuits with synaptic plasticity: Learning and memory. BioNanoScience, 1, 24-30 (2011).
6. A.A. Minnekhanov, A.V. Emelyanov, D.A. Lapkin, K.E. Nikiruy, B.S. Shvetsov, A.A. Nesmelov, V.V. Rylkov, V.A. Demin, and V.V. Erokhin. Parylene based memristive devices with multilevel resistive switching for neuromorphic applications. Sci. Rep., 9, 10800 (2019).
7. N.V. Prudnikov, D.A. Lapkin, A.V. Emelyanov, A.A. Minnekhanov, Y.N. Malakhova, S.N. Chvalun, V.A. Demin, and V.V. Erokhin. Associative STDP-like learning of neuromorphic circuits based on polyaniline memristive microdevices. J. Phys. D: Appl. Phys. in press https://doi.org/10.1088/1361-6463/ab9262 (2020).
8. A. Cifarelli, T. Berzina, and , V. Erokhin. Bio-organic memristive device: Polyaniline – Physarum polycephalum interface. Physica Status Solidi C, 12, 218-221 (2015).
9. S. Battistoni, V. Erokhin, and S. Iannotta. Frequency driven organic memristive devices for neuromorphic short and long term plasticity. Org. Electronics, 65, 434-438 (2019).
10. E. Juzekaeva, A. Nasretdinov, S. Battistoni, T. Berzina, S. Iannotta, R. Khazipov, V. Erokhin, and M. Mukhtarov. Coupling cortical neurons through electronic memristive synapse. Adv. Mater. Technol., 4, 1800350 (2019).

Unlike the inorganic neuromorphic electronics which can only implement the long-term plasticity, the organic neuromorphic electronics have paid attention due to promising applications from next-generation computing to neuroprosthetics by implementing not only the long-term potentiation (LTP) but also the short-term potentiation (STP). Recently, conjugated polymers such as poly(3-hexylthiophene-2,5-diyl) (P3HT) with electrolyte dielectrics have been studied to implement high-density computing and memory systems using less energy. Parallelly, diketopyrrolopyrroles (DPPs) based artificial synaptic devices have shown promising applications from neuroprosthetics to soft robotics. However, recently studied organic-based artificial synapses only focused on emulating the single function of the biological counterparts; either memory or signal transitions.
In this study, we introduce the engineering of synaptic properties by modulating morphological characteristics of organics semiconductors (OSCs) with fixed presynaptic spike forms in artificial synaptic devices without changing the OSCs. By achieving this, the relation between the morphology of the film and the synaptic property has been revealed. Changing the morphological property of the OSCs, resulted in the clear transition from STP dominant synaptic property to LTP dominant synaptic properties. Furthermore, artificial neural networks (ANNs) simulation and artificial auditory sensory nervous system were successfully demonstrated. This achievement is an imperative step toward the development of versatile neuromorphic electronics and promising for various neuromorphic electronics applications from neuromorphic computing to neural prosthetics, bio-interface devices, and soft robotics which can immensely expand the field of neuromorphic electronics by providing an understanding of the mechanisms of synaptic properties.
[1] D.-G. Seo, et al., Nano Energy 2019, 65, 104035 [2] Y. Kim, et al., Science. 2018, 360, 998 [3] Y. Lee, et al., Sci. Adv. 2018, 4, eaat7387 [4] W. Xu, et al., Sci. Adv. 2016, 2, e1501326

Human-machine interfaces will continue to be critical to advance applications in healthcare, robotics and consumer electronics. Having a cross-disciplinary approach to materials innovation, especially with regards to soft materials¹ will effectively address the mechanical constraints when interfacing with humans. In this talk, I will discuss some of the new material platforms we are developing for more robust human-machine interactions, such as self-healing stretchable optoelectronics devices². Such optoelectronic devices might be employed in optogenetic neural interfaces . At the same time, I will also present our efforts in brain inspired scalable neuromorphic architectures for electronic skins³. Such system architectures can be broadly applied in various applications with high impact potential for wearables, brain-machine interfaces, prosthetics and robotics.

1. Soft Electronically Functional Polymeric Composite Materials for a Flexible and Stretchable Digital Future, B. C-K. Tee*, J. Ouyang, Advanced Materials,1802560, 2018.
2. A transparent, self-healing and high-k dielectric for low-field-emission stretchable optoelectronics, Y. J. Tan, H. Godaba, G. Chen, S. T. M. Tan, G. Wan, G. Li, P. M. Lee, Y. Cai, S. Li, R. F. Shepherd, J. S. Ho & B. C. K. Tee*, Nature Materials, 2019.
3. A neuro-inspired artificial peripheral nervous system for scalable electronic skins , WW Lee, Y. J. Tan, H. Yao, S. Li, H. H. See, M. Hon, K. A. Ng, B. Xiong, J. S. Ho and B. C. K. Tee*, Science Robotics, 2019.

Organic electrochemical transistors (OECTs) have gained considerable interest for applications in bioelectronics, power electronics, circuits and neuromorphic computing. Their defining characteristic is the bulk-modulation of channel conductance owing to the facile penetration of ions into the (semi)conducting polymeric channel. For this reason, their device scaling relies on film thickness, enabling finer size scale devices, and unique form factors. Despite recent progress and a rapidly expanding library of new materials, the understanding of stability and transport/coupling of ionic and electronic carriers remain largely unexplored. We highlight recent synthetic and processing approaches used to tailor device properties and stability, as well as new opportunities enabled by such advances. We demonstrate how the reliance on bulk transport allows for relaxed fabrication demands and production of simple multi-device circuit concepts for analog signal processing. Finally, by studying the role of additives on charging, trapping, and structural re-arrangement, we can gain insight into the next generation of non-volatile organic mixed conductors for neuromorphic hardware. The implementation of OECT-based sensing and circuits holds particular promise in brain-inspired bioelectronics – the role of materials design and development is critical for their success.

Nanosensor arrays have proven to be powerful tools to measure single biological cells and organisms, achieving spatial and temporal precision even at the single molecule level. However, there has not been a way of extending this approach to statically relevant numbers of living cells and organisms. Herein, we design and fabricate a high throughput nanosensor array in a microfluidic channel that addresses this limitation, creating a Nanosensor Chemical Cytometry (NCC) for the first time. We show that one can utilize the cell itself as a lens and optical element to extract much more information from the nanosensor array on a per cell basis at high throughout. An array of nIR fluorescent single walled carbon nanotube (SWNT) sensors is deposited along a microfluidic channel through which a population of flowing cells is guided. As an example, the array can detect, in real time, the H₂O₂ efflux of monocytes at attomole sensitivity at a rate of 10 cells/min. We show that the cells themselves can function as Gaussian lenses, projecting informative excitation profiles onto the nanosensor array. This allows for cross-correlation of the cellular diameter, effective refractive index, and eccentricity with chemical efflux and creates a chemical cytometer for the measurement of cellular heterogeneity with unprecedented precision. For example, we study distinct H₂O₂ efflux rates between 330 and 624 atmole/cell per min in cell size ranges of 271 and 263 µm², eccentricity values between 0.405 and 0.363 and RI values between 1.383 and 1.377 for pristine and activated monocytes, respectively. Hence, we show that this new nanotechnology based tool has significant potential to answer important questions in cellular immunology, cell manufacturing and biopharmaceutical research.

To detect chemical or biological molecules, it is crucial that sensor devices can differentiate various target molecules. However, in general, each sensing methods are geared towards one aspect for sensing while sacrificing others due to their particular limitations. For example, gas sensors based on metal oxide have high sensitivity while having low identification ability. On the other hand, surface-enhanced Raman spectroscopy is a representative label-free and selective detection method that can identify target molecules but often lack quantitative sensing capability, especially for gaseous targets.
Herein, we introduce a label-free electrical/optical multimodal sensor based on systematically assembled 3D nanoarchitecture with multifunctional materials for optimizing the structure effect. It combines several sensing modalities and sensing elements via the integration of semiconducting nanowire frameworks and dual-functioning metallic nanoparticles. Our multimodal sensor can successfully estimate the mixed gas compositions selectively and quantitatively in the sub-100 ppm level, even in a mixture of gaseous aromatic compounds with similar molecular structures by providing complementary information. Our approach not only provides the sensitivity but also has the ability to differentiate the target molecules. Potentially, a multimodal sensor could be an application to early diagnosis of many diseases and the rapid detection of toxic chemicals and explosives

Fluidity, mobility, and resistivity are key characteristics of native membranes in our cells where cation channels regulate numerous molecular mechanisms (e.g., the excitation of neurons). Interfacing with biomimetic membranes via electronic devices promises to expand our understanding of the function of these membranes in disease states. The challenge is to form supported lipid bilayer (SLBs) on (semi)conducting films while sustaining high sensitivity and operational stability. In this talk, I will show the traits of n-type organic semiconductors for such biological interfaces. I will show the very first ion channel sensor based on an n-type, accumulation mode, microscale organic electrochemical transistor (OECT), which interfaces an SLB embedded with a pore-forming protein. The polymer is particularly sensitive to divalent cations, endowing the system with biomimetic properties. The side-chain engineering of semiconducting polymers prompts the development of next-generation bioelectronic hybrids where electron-transporting polymers establish bilateral electronic communication with living systems.

A great demand exists for minimally-invasive neuromodulation technologies to enable next-generation bioelectronic medicine. We report on our developments of ultrathin organic (opto)electronic devices for neurostimulation. All of these devices rely on far red/near infrared irradiation in the tissue transparency window to actuate nanoscale organic semiconductor components. Our flagship technology is the organic electrolytic photocapacitor (OEPC) – a device that mimics biphasic current-pulse neurostimulation and thus transduces an optical signal into directly-evoked action potentials in neurons. These devices are not only wireless, but also 100-1000 times thinner than most existing technologies. We will discuss chronic implants capable of stimulating peripheral nerves (sciatic and vagus) when actuated from outside of the body using diode lasers. We have observed stable operation in rodent models for at least 100 days. As a different approach, we will report on favorable harnessing of DC electrochemistry effects, normally considered spurious and undesired. Specifically, we show how electrochemical generation of low doses of hydrogen peroxide can be used to reversibly stimulate M-type potassium channels, providing a novel method of controlling neural excitability.

Bioelectronic devices must have high speed and high performance to interact with the often rapid, low amplitude signals generated by neural tissue. They should also be capable of forming complementary logic and analog circuits to allow acquisition, processing and manipulation of a biological environment. However, the lack of readily available p-type polymers for use in enhancement mode transistors and their complicated synthesis significantly limits the investigation and application of such devices in bioelectronics. Here, we develop an enhancement mode ion-gated transistor (e-IGT) that functions based on a reversible redox reaction and hydrated ion reservoirs within the conducting polymer channel to enable long-term stable operation and shortened ion transit time. We combined a highly conductive poly(3, 4-ethylenedioxythiophene) doped with polystyrene sulfonate (PEDOT:PSS) and a reducing agent polyethylenimine (PEI) to enable doping of the polymer in an aqueous environment. The transistor operates as following: (1) the protonated PEI bonds with the PSS- and de-dopes the PEDOT:PSS chain resulting in an initial off-state of the transistor, (2) a subsequent application of negative gate voltage dopes the PEDOT:PSS switching the channel on. To determine an optimal transistor configuration and material composition, we microfabricated transistor arrays of varying geometrical parameters. The high-transconductance and high-speed of the e-IGTs resulted in a gain bandwidth product that is several orders of magnitude above other ion-based transistors. We used these transistors to acquire a wide range of electrophysiological signals, including in vivo recording of neural action potentials. Furthermore, we combined E-IGTs with d-IGTs to create a real-time non-linear rectifier to accurately detect epileptic discharges in vivo. In conclusion, E-IGTs offer a safe, reliable, and high performance building block for chronically implanted bioelectronics, with spatiotemporal resolution to the scale of individual neurons.

One of the neuromorphic engineering aims is using nanoelectronics’ materials and devices to reproduce key features that are used by the brain for computing. Currently, neuromorphic engineering has explored standard silicon-based technologies (i.e. such as complementary metal–oxide–semiconductor ) or more emerging material and devices (iono-electronic materials and resistive memory devices, for example). Most of these technologies are still bounded to a top down approach. However, brain computing largely rely on bottom-up processes. For instance, interconnectivity between cells and formation of communication pathway in neural networks result principaly from bottom-up organization. Here, we show how dendritic growth of organic conductive polymers (PEDOT) can be used to mimic structural branching observed in neural network. Conducting-polymer based dendritic structures with different morphology are synthesized in a two-electrode setup by pulsed voltage-driven electropolymerization derived from state-of-the-art bipolar AC-electrochemical synthetic methods. We show how various AC signals can lead to a large variety of dendritic structures and PEDOT morphologies. In a second part, such dendritic structures are used to implement functionnal OECTs. More importantly, we focus on the transconductance and memory effects that can be obtained in such dendritic OECTs such as short tem plasticity. We report on the relationship between dendrites morphologies and STP time constant.
This work paves the way to new approaches for neuromorphic engineering, such as structural plasticity and neural network topology exploration.

We focus on developing a new class of nanoscale materials and novel strategies for the investigation of biological entities at multiple length scales, from the molecular level to complex cellular networks. Our highly flexible bottom-up nanomaterials synthesis capabilities allow us to form unique hybrid-nanomaterials that can be used in various input/output bioelectrical interfaces. For example, we have developed several bioelectrical platforms based on graphene, a two-dimensional (2D) atomically thin carbon allotrope. We have demonstrated recording of the electrical activity of excitable cells with graphene-based ultra-microelectrodes as small as the size as an axon ca. 2µm in size. Using graphene-based hybrid-nanomaterials, we have formed remote, non-genetic bioelectrical interfaces with excitable cells and modulated cellular and network activity with high precision and low needed power. We have also developed a breakthrough bioelectrical interface, a 3D self-rolled biosensor arrays (3D-SR-BAs) of either active field effect transistors or passive microelectrodes to measure both cardiac and neural spheroids electrophysiology in 3D. Our approach enables electrophysiological investigation and monitoring of the complex signal transduction in 3D cellular assemblies toward an organ-on-an-electronic-chip (organ-on-e-chip) platform for tissue maturation investigations and development of drugs for disease treatment. In summary, the exceptional synthetic control and flexible assembly of nanomaterials provide powerful tools for fundamental studies and applications in life science and open up the potential to seamlessly merge either nanomaterials-based platforms or unique nanosensor geometries and topologies with cells, fusing nonliving and living systems together.

The rapid rise of bioelectronics has enabled a number of innovative approaches to treat neurological disorders, restore and repair lost functions, and modulate neural circuitry to control mood and behavior. The vast majority of bioelectronic interfaces, however, still rely on traditional noble metal and silicon materials, which are expensive to source and process and are intrinsically inadequate to address the mechanical, chemical, and electrical properties of excitable circuits in the body. To establish bioelectronic technologies as the clinical standard, significant innovations in materials and fabrication strategies are still required to achieve safety, softness, biocompatibility, and long-term stability. Nanostructured materials are uniquely positioned to address these challenges because they combine high electrical conductivity with intrinsically high mass-specific surface area and mechanical flexibility. Furthermore, they can interact with biological systems on a molecular scale and can be easily integrated within scalable solution-based processing, thus allowing easy modulation of their electronic and optical properties.
In this talk I will present the ongoing work in my lab to leverage the unique properties of nanostructured materials and engineer novel bioelectronic interfaces for sensing and stimulating neural and neuromuscular circuits. Specifically, I will describe the fundamental electronic, electrochemical, and mechanical properties of 2D transition metal carbides nanomaterials (a.k.a. MXenes) and how these translate into significant impedance and noise reduction when MXenes are integrated bioelectronic interfaces. Then, I will present ad hoc, scalable, rapid manufacturing processes developed in my lab to translate the exceptional material properties at the molecular scale into high-resolution, low impedance bioelectronic interfaces that are also compatible with clinical neuroimaging modalities, such a magnetic resonance imaging (MRI) and computerized tomography (CT). Finally, to illustrate the potential of MXene-based bioelectronics, I will present different example of applications in both implantable and wearable interfaces.

Organic electrochemical transistors (OECTs) have recently attracted tremendous attention in the scientific community due to their unique set of features. Especially, the lack of an insulating layer that separates gate from channel, like in conventional field effect transistors, and their feature to support electrolyte gating open new pathways in the field of bioelectronics. In particular, they render the device ideal for measurements in the aqueous environments that dominate the biological world. Additionally, they provide extra possibilities in terms of the device architecture as the gating electrode can be patterned on the same plane as the conducting channel. In a set up like that, what is of extreme interest is the effect of the gate position with respect to the channel. In this work, the spatiotemporal response of a poly(3,4 ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) OECT is examined and its device physics is studied. This work opens new routes for exploiting the physics of multi terminal OECT devices in bioelectronics, sensing and neuromorphic electronics.

Research into mood and emotion has often been dependent on self-report, highlighting the need for a more objective measure of mood state. However, until recently, it has been nearly impossible for machines to match the speed and accuracy of the average person in interpreting faces. In this study, we used digital image speckle correlation (DISC), a technique which was originally used to study the mechanical properties of materials by tracking their deformations, to assess an individual’s mood state based on the analysis of facial microexpressions. Ten healthy volunteers were shown a set of ten happy and ten sad images while being videotaped. Upon measuring facial movements in response to happy and sad images, we conducted detailed Shannon entropy and Principal Component Analysis (PCA) to identify facial maps for different mood states. Within and across subjects, happy and sad images elicited spatially distinct changes in facial expression. PCA and entropy calculations on facial movement reliably distinguished the subject’s underlying mood. Furthermore, we characterized key changes in facial expression which reliably signal changes in mood state across all individuals. DISC-based analysis of the face was capable of determining an individual’s mood with high specificity across multiple test subjects, suggesting that automated camera-based methods may be both robust and economical in clinical monitoring settings. It also showed promising potentials in brain study together with electroencephalogram as subtle facial movements were also found in epilepsy and comatose patients using DISC.

The brain can perform massively parallel information processing while consuming only ~1- 100 fJ per synaptic event. I will describe a novel electrochemical neuromorphic device (ENODe) that switches at record-low energy (<0.1 fJ projected, <10 pJ measured) as well as voltage (< 1mV, measured), and displays a large number of distinct, non-volatile conductance states within a ~1 V operating range. The tunable resistance behaves very linearly, allowing blind updates in a neural network when operated with the proper access device. ENODes also display outstanding endurance achieving over 10⁹ switching events with very little degradation. I will describe our recent efforts at scaling and materials selection, allowing us to reach 20 ns write pulses and operation at high temperature (up to 120°C). These properties are very promising in terms of the ability to integrate with Si electronics to demonstrate online learning and inference. ENODes are electrochemical devices where gated proton drift induces changes in the electronic states of a semiconductor channel. The peculiarities of the physics of these devices will be discussed along with their consequences on device design and performance.

2D networks of interconnected nano-objets (nanowires, nanoparticles, molecules) are experimentally and theoretically explored to implement unconventional computing machines, and especially reservoir computing [1-9]. 2D networks of molecularly functionalized nanoparticles (NPs) (hereafter called NMN : nanoparticle molecule network) have emerged as an interesting approach in molecular electronics to understand fundamental electron transport mechanisms [10].
I will describe the study of NMN approaches using functional molecules self-assembled in high density 2D networks with topological structures that are intrinsically similar to the structure of a "reservoir computing". NMNs equipped with molecules which can change their electronic properties upon a given excitation (optical, chemical, ionic) are interesting to study multifunctional systems at the nanoscale (<100 nm).
I will discuss several key features of the electronic properties of these 2D networks of molecularly functionalized nanoparticles to assess their possible use for reservoir computing: highly non-linear electron transport, variability, complex/rich dynamics (harmonic, interharmonic, intermodulation distortions), co-tunneling, noise and plasmonic response [9-12]. I will also discuss the electron transport properties of some molecules of interest for the chemical and biochemical sensing with these NMNs reservoir computing approaches, combining sensing and computing in a single nanoscale device.
These approaches, without direct analogs in semiconductor nanoelectronics, would open new perspectives to molecular electronics in unconventional computing.

1. Tour, J. M.; van Zandt, W. L.; Husband, C. P.; Husband, S. M.; Wilson, L. S.; Franzon, P. D.; Nackashi, D. P. IEEE Trans. Nanotechnol. 2002, 1, (2), 100-109.
2. Sköldberg, J.; Wendin, G. Nanotechnology 2007, 18, (48), 485201.
3. Bose, S. K.; Lawrence, C. P.; Liu, Z.; Makarenko, K. S.; van Damme, R. M. J.; Broersma, H. J.; van der Wiel, W. G. Nature Nanotech 2015, 10, 1048-1053.
4. Roychowdhury, V. P.; Janes, D. B.; Bandyopadhyay, S.; Wang, X. IEEE Transactions on Electron Devices 1996, 43, 1688-1699.
5. Sillin, H. O.; Aguilera, R.; Shieh, H.-H.; Avizienis, A. V.; Aono, M.; Stieg, A. Z.; Gimzewski, J. K. Nanotechnology 2013, 24, (38), 384004.
6. Maass, W.; Natschläger, T.; Markram, H. Neural Computation 2002, 14, 2531-2560.
7. Jeager, H.; Haas, H. Science (New York, NY) 2004, 304, (5667), 78-80.
8. Tanaka, H.; Akai-Kasaya, M.; TermehYousefi, A.; Hong, L.; Fu, L.; Tamukoh, H.; Tanaka, D.; Asai, T.; Ogawa, T. Nature Communications 2018, 9, (1), 2693.
9. Viero, Y.; Guerin, D.; Alibart, F.; Lenfant, S.; Vuillaume, D. Adv. Func. Mater. 2018, 28, 1801506.
10. Liao, J.; Blok, S.; van der Molen, S. J.; Diefenbach, S.; Holleitner, A. W.; Schönenberger, C.; Vladyka, A.; Calame, M. Chem. Soc. Rev. 2015, 44, 999-1014.
11. Viero, Y.; Copie, G.; Guerin, D.; Krzeminski, C.; Vuillaume, D.; Lenfant, S.; Cleri, F. J. Phys. Chem. C 2015, 119, 21173-21183.
12. Stievenard, D.; Guerin, D.; Lenfant, S.; Leveque, G.; Nijhuis, C. A.; Vuillaume, D. Nanoscale 2018, 10, 23122–23130.

Spiking Neural Networks (SNN) provide distributed computational capability and emulate the brain processing principles based on spike frequency and pulse width [1]. Because of being intrinsically compatible with brain signaling, one of the target applications of SNN hardware implementation, known as neuromorphic systems [2], are Brain-Computer Interfaces (BCI) [3]. With a handful of exceptions [4], such implementations have typically been implemented using silicon technology, which is physically hard and rigid, and non-biocompatible, resulting in a permanent damage to the brain's soft tissue. Organic electronics offer physical softness and flexibility, biocompatibility, and low-cost, large-area fabrication, to directly interface with biological tissues [5].

We report on the fabrication of a physically flexible, organic electronics, spiking integrate-and-fire Axon-Hillock (AH) neural circuit. Organic Field-Effect Transistors (OFETs) are fabricated on top of a flexible, 50 μm Polyimide substrate. OFETs’ architecture is bottom-gate top-contact, and fabricated through thermal deposition of Ti (3 nm)/Au (50 nm) as the gate electrode, 60 nm of Parylene as the dielectric, 30 nm of dinaphtho[2,3b:2’,3’-f]thieno[3,2-b]thiophene (DNTT) as the p-type and N,N’-bis(n-octyl)-x:y,dicyanoperylene3,4:9,10-bis(dicarboximide) (PDI8-CN2) as the n-type semiconductor. A 50 nm layer of Au serves as the source and the drain electrodes. A layer of Parylene encapsulates the structure. Non-circuit-integrated commercial resistors and capacitors are used to accelerate the characterization and the fabrication process. All integrated all-organic circuit is beyond the scope of this work and would increase the complexity of the process.

The results show that the AH circuit with 500 nA input current can produce spikes with a range of frequencies. Given the membrane capacitance (C_mem) of 15 nF and feedback capacitance (C_feedback) of 47 nF, the spiking frequency reaches 670 mHz. The frequency can be lowered by increasing capacitance values to C_mem = 33 nF and C_feedback = 100 nF, respectively, to 370 mHz. Decreasing capacitance to C_mem = 2.2 nF and C_feedback = 6.8 nF, will result in a higher spiking rate (5 Hz), but will also increase the influence of power line noise and parasitic capacitance. The frequency-current (F-I) transfer function shows the dynamics of the neuron. By increasing the input current, the F-I curve will saturate indicating an upper bound of frequency spikes. The computational ability of the neuron is also demonstrated with a basic summation of two unitary excitatory synaptic events. The static and dynamic power dissipation is recorded and is limited to 0.4 µW and 40 µW, respectively.

We believe that this work is the first demonstration of a functioning organic electronics integrate-and-fire spiking neuron, paving the way for future implantable BMIs. Beyond neuroscientific applications, this technology can also be used in slow processes, such as temperature measurement, which are intrinsically difficult for ultra-fast silicon devices to achieve, as well as physically flexible large-area applications.

[1] F. Rieke and D. Warland, “Spikes: Exploring the Neural Code,” MIT Press, 1999.
[2] R. Nawrocki, et al., “A mini review of neuromorphic architectures and implementations,” IEEE Transactions on Electron Devices, vol. 63 (10), 2016.
[3] F. Boi, et al., “A Bidirectional Brain-Machine Interface Featuring a Neuromorphic Hardware Decoder,” Frontiers in Neuroscience, vol. 10 (563), 2016.
[4] R. Nawrocki, et al., “Neurons in polymer: Hardware neural units based on polymer memristive devices and polymer transistors,” IEEE Transactions on Electron Devices, vol. 61 (10), 2014.
[5] G. Schwartz, et al., ”Flexible polymer transistors with high-pressure sensitivity for application in electronic skin and health monitoring,” Nature Communications, vol. 4 (1), 2013.

Neuromorphic elements have been predominantly solid-state devices which simulate the resistive and capacitive behaviors needed for neural networks and brain-inspired computing, but in non-brain-like ways. This talk will describe ways we are integrating lipid and polymer bilayer membranes with micro- and nanofabrication to develop fundamentally new types of neuromorphic elements that have the composition (biomolecules), structure (biomembranes), and switching mechanism (voltage-sensitive ion channels) of real biological synapses, and operate at lower power than the current state-of-the-art. Our devices consist of insulating, nm-thick lipid or polymer-based bilayer membranes that assemble at the interfaces of two or more aqueous droplets in oil, and that have demonstrated both memristive and memcapacitive behaviors, including memory resistance and capacitance, synaptic functions such as paired-pulse facilitation and depression, spike rate dependent plasticity, voltage-dependent inactivation and recovery, and charging hysteresis. These behaviors are linked to electrostriction, an electromechanical phenomenon that encompasses both electrowetting and electrocompression in the membrane, which are changes in membrane area and thickness due to charging in the presence of electric fields. Electrostriction results in a voltage-dependent capacitive susceptibility that replaces the more familiar concept of static capacitance, which, up to now, has dominated electrophysiological descriptions and characterizations of biomembranes. In this picture, biomembranes are not just equivalent RC circuits dependent only on ionic currents controlled by the conductance (resistance) of ion channels. Large capacitive currents from small voltage inputs can be generated as well for the development of neuromorphic computing elements exhibiting both short-term and long-term synaptic plasticity.

Most of today’s strategies to interface biology with electronic hardware are based on layered architectures where the front-end of sensing is optimized separately from the back-end for processing/computing signals. Alternatively, biological systems are capitalizing on distributed architecture where both sensing and computing are mix together and co-optimized. In this talk, we will present our strategy to implement bio-sensing of electroactive cells in a neuromorphic perspective. We will present how organic electrochemical transistors can be used to record electrical signals from neural cells. We will show various strategies capitalizing on the versatility of organic materials synthesis and organic device fabrication to tune and adapt the functionalities of such bio-sensors. We will then present how these strategies can be efficiently used to realize computing functions directly at the interface with biology. Notably, we will illustrate how a network of ionic sensors can implement the reservoir computing concept, a powerful neuromorphic computing approach of particular interest for dynamical signal processing.

Human-machine interfaces will continue to be critical to advance applications in healthcare, robotics and consumer electronics. Having a cross-disciplinary approach to materials innovation, especially with regards to soft materials¹ will effectively address the mechanical constraints when interfacing with humans. In this talk, I will discuss some of the new material platforms we are developing for more robust human-machine interactions, such as self-healing stretchable optoelectronics devices². Such optoelectronic devices might be employed in optogenetic neural interfaces . At the same time, I will also present our efforts in brain inspired scalable neuromorphic architectures for electronic skins³. Such system architectures can be broadly applied in various applications with high impact potential for wearables, brain-machine interfaces, prosthetics and robotics.

1. Soft Electronically Functional Polymeric Composite Materials for a Flexible and Stretchable Digital Future, B. C-K. Tee*, J. Ouyang, Advanced Materials,1802560, 2018.
2. A transparent, self-healing and high-k dielectric for low-field-emission stretchable optoelectronics, Y. J. Tan, H. Godaba, G. Chen, S. T. M. Tan, G. Wan, G. Li, P. M. Lee, Y. Cai, S. Li, R. F. Shepherd, J. S. Ho & B. C. K. Tee*, Nature Materials, 2019.
3. A neuro-inspired artificial peripheral nervous system for scalable electronic skins , WW Lee, Y. J. Tan, H. Yao, S. Li, H. H. See, M. Hon, K. A. Ng, B. Xiong, J. S. Ho and B. C. K. Tee*, Science Robotics, 2019.

The interface between biological cells and non-biological materials has profound influences on cellular activities, chronic tissue responses, and ultimately the success of medical implants and bioelectronic devices. The optimal coupling between cells, i.e. neurons, and materials is mainly based on surface interaction, electrical communication and sensing.
In the last years, many efforts have been devoted to the engineering of materials to recapitulate both the environment (i.e. dimensionality, curvature, dinamicity) and the functionalities (i.e. long and short term plasticity) of the neuronal tissue to ensure a better integration of the bioelectronic platform and cells.
On the one hand, here we explore how the transition from planar to pseudo-3D nanopatterned inorganic and organic materials have introduced a new strategy of integrating bioelectronic platforms with biological cells under static and dynamic conditions. Although a spontaneous penetration does not occur, adhesion processes are such that a very intimate contact can be achieved. On the other hand, we investigate how organic semiconductors can be exploited for recapitulating electrical neuronal functions such as long term and short term potentiation. In this way, both the topology and the material functionalities can be exploited for achieving in vitro biohybrid platforms for neuronal network interfacing.

Organic electrochemical transistors (OECTs) have gained considerable interest for applications in bioelectronics, power electronics, circuits and neuromorphic computing. Their defining characteristic is the bulk-modulation of channel conductance owing to the facile penetration of ions into the (semi)conducting polymeric channel. For this reason, their device scaling relies on film thickness, enabling finer size scale devices, and unique form factors. Despite recent progress and a rapidly expanding library of new materials, the understanding of stability and transport/coupling of ionic and electronic carriers remain largely unexplored. We highlight recent synthetic and processing approaches used to tailor device properties and stability, as well as new opportunities enabled by such advances. We demonstrate how the reliance on bulk transport allows for relaxed fabrication demands and production of simple multi-device circuit concepts for analog signal processing. Finally, by studying the role of additives on charging, trapping, and structural re-arrangement, we can gain insight into the next generation of non-volatile organic mixed conductors for neuromorphic hardware. The implementation of OECT-based sensing and circuits holds particular promise in brain-inspired bioelectronics – the role of materials design and development is critical for their success.