Jie
Zhang
*a,
Junmei
Du
b,
Chuan
Yang
b,
Haotian
Liang
b,
Zelin
Cao
cd,
Xuegang
Duan
cd,
Wentao
Yan
cd,
Yong
Zhao
b and
Bai
Sun
*cd
aCollege of Mathematics and Physics, Chengdu University of Technology, Chengdu, Sichuan 610059, China. E-mail: jiezhang@cdut.edu.cn
bSchool of Physical Science and Technology, Key Laboratory of Advanced Technology of Materials, Southwest Jiaotong University, Chengdu, Sichuan 610031, China
cFrontier Institute of Science and Technology (FIST), Xi’an Jiaotong University, Xi’an, Shaanxi 710049, China. E-mail: baisun@xjtu.edu.cn
dMicro-and Nano-technology Research Center, State Key Laboratory for Manufacturing Systems Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, China
First published on 22nd November 2023
Memristors, as a novel type of electronic device with memory resistance characteristics, are attracting considerable interest in the field of biomedical applications. In the realm of biosensors and diagnostic technologies, memristors can interact with biomolecules, detecting and analyzing disease biomarkers, thus enabling highly sensitive early cancer detection and rapid virus diagnosis. Additionally, in the domain of neuromorphic computing and brain–machine interfaces, memristors’ ability to emulate biological neuron behavior opens up new possibilities for neuromorphic computing systems and neural-controlled prosthetics. Moreover, memristors are also applicable in simulating biological neurons and researching brain disorders, aiding the comprehension of complex brain functions and neurological diseases, and exploring novel treatment methods. This paper provides an overview of the basic principles of memristors, summarizes the latest achievements and advancements in biomaterial-based memristors, and discusses the challenges and prospects of employing biomaterial-based memristors in the biomedical field. Therefore, biomemristors hold tremendous potential for applications in biomedical research and therapy, offering new prospects for medical research and treatment.
A memristor is a non-linear two-terminal device, and its resistance varies as the input current or voltage accumulates.13 In 2008, Strukov et al. at Hewlett-Packard Labs successfully demonstrated a TiO2 nano-film device that functioned as a resistive device, following the predictions.14 A memristor is a circuit device that captures the correlation between charge (q) and magnetic flux (φ).15 These components are characterized by distinctive memory attributes, non-volatility, and non-linearity. Fig. 1 illustrates the existence of capacitors, resistors, inductors, and memristors as the four fundamental passive circuit elements, with voltage, current, charge, and magnetic flux serving as the four fundamental variables that define their interrelationships.16 In the proposed context of amnesia, when considering the fundamental primitive devices, amnesia, along with other passive basic circuit elements like resistors, capacitors, and inductors, forms a comprehensive set of quaternary basic circuit primitives. Since then, a substantial amount of research has been conducted on memristors. Memristors find widespread applications in various fields such as artificial neural networks, biomedicine, circuits and systems, new types of storage, and chaotic systems. However, one of the most intriguing areas of application is the integration of memristors with biomedicine.
Typically, the unit structure of a memristor follows the metal/insulator/metal (MIM) configuration. The symbol “M” generally represents materials that are good electronic conductors, including active and inert metals, among others. On the other hand, “I” denotes insulator materials, which, in most cases, refer to good ionic conductors, such as oxides and organic compounds. Simultaneously, the upper and lower metal layers are referred to as the top electrode and the bottom electrode, respectively. Between these electrodes, there exists a resistive layer with memristive properties. The schematic structure of the memristor is shown in Fig. 2.17
Based on their rheostat switching characteristics, memristors can be categorized as binary or analog. Binary memristors are better suited for storing or processing binary data due to their two distinct resistance states, often representing 0 and 1. On the other hand, analog memristors are more appropriate for multi-resistance storage and analog calculations, as they can maintain a continuous range of resistance values.
Depending on the resistive material used, memristors can be categorized into several types, they are simply classified as oxide memristors,18,19 solid-state electrolyte materials,20 calcium titanium oxide,19 memristors for 2D materials, and organic materials (e.g. egg white, blood serum).21,22 Conventional memristors offer several advantages, including miniaturization and robust read/write capabilities. The classification diagram of the memristor is shown in Fig. 3.
However, in terms of electrical performance, traditional memristors have certain drawbacks, such as discrete switching voltages caused by local conductive filaments and relatively weak durability. Yan et al. have been dedicated to the advancement of memristors and have achieved significant progress in brain-like synaptic bionics.19,23 Notably, they have made strides in areas such as long time-range potentiation (LTP), long time-duration inhibition (LTD), spike-timing-dependent plasticity (STDP), two-pulse facilitation (PPF), and paired-pulse depression (PPD). As a result, memristors have been further developed in the realm of brain neuromimetics. Pulse facilitation (PPF) and paired-pulse depression (PPD) have been further developed in the field of brain neuromimetics, alongside the progress made with amnesia-based memristors.
Since the initial realization of amnesia prototype devices in 2008, significant progress has been made in academia concerning amnesia-based memristors, achieving numerous noteworthy results. Nevertheless, due to the absence of a standardized set of evaluation criteria, there remains no uniform standard answer for selecting the amnesia material system and the mechanism model. As a consequence, research on amnesia is currently constrained to the fundamental research phase. Both the selection of the material system and the establishment of the mechanism model are still in the realm of individual exploration, with researchers striving to achieve a “perfect” amnesia device. The use of biomaterials in amnesia has generated significant research interest among a considerable number of scholars. This interest is primarily driven by the biomaterials' appealing characteristics, such as their lightweight nature, high flexibility, ease of processing, non-polluting properties, and recyclability. Currently, significant progress has been achieved in the biomedical application of amnesia. Fu et al. have made noteworthy advancements in this area. Protein nanowires derived from microbial soil bacteria enable electronic amnesia to function at biological voltages to replicate neuronal activity in the brain.24 This development holds promise for establishing genuine neuronal communication in biological systems. Wu et al. presented a novel implementation approach to create an artificial sensory neural system with habituation characteristics using amnesia.25 They developed a habituation impulse neural network, which adopts habituation as a biological learning rule. This neural network was successfully applied to robotic autonomous cruise obstacle avoidance, demonstrating the potential for integrating biological learning principles into artificial systems. These natural biomaterials hold potential to serve as the “ideal” memory blockers.
In 2013, Mohamad et al. presented the design of an amnesia sensor based on a TiO2 material for sensing applications.30 In 2014, Francesca Puppo et al. introduced an equivalent circuit that replicates the amnesic effect observed in the current–voltage (I–V) properties of silicon nanowires.31 Bequently, a human vascular endothelial growth factor (VEGF) antibody was employed to functionalize the functional layer of memristor device. The I–V characteristics of the nanowire were then investigated both before and after protein functionalization. The binding of biomolecules to the surface of the memristor was evidenced by the observed expansion of voltage intervals in the hysteresis curve. In 2015, Sun et al. introduced a novel ionic memristor that operates based on ionic currents in an aqueous solution with arbitrary ionic concentration.32 The ionic memristor is composed of silicon microelectrodes immersed in an aqueous solution environment, offering rapid and resilient switching with latch-up capabilities suitable for large-scale physiological ionic circuits as shown in Fig. 4a. The utilization of dielectric-immersed fluid-based sensors, actuators, and logic circuits eliminates the need for complex wireless connections with external solid-state integrated circuits. This advancement offers increased versatility and allows for larger signals in the development of future implantable and embedded smart medical devices. In the identical year, Tzouvadaki et al. produced collections of independent two-terminal Schottky barrier silicon nanowires exhibiting amnesic characteristics, aiming to acquire amnesic biosensors.33 The existence of biomolecules adhered to the nanostructures’ surface was assessed through the observation of a voltage gap in the memristive electrical properties. The system demonstrates promising potential for applications in molecular diagnostics, particularly owing to its capability for detection in the millimolar range, enabling early identification of cancerous diseases.
Fig. 4 (a) Schematic and image of a fluid-based ionic memristor.32 copyright 2015, Wiley-VCH. (b) Schematic illustration of AlOOH flexible memristive device and the physical display of the flexible device.39 copyright 2023, BioMed Central Ltd. (c) Schematic diagram of a memory synapse. |
In 2017, Volkov et al. introduced circuit-connected biosensors in plants and trees, exploring the various electrochemical components and devices that nature has developed within plants.34 Amnesia facilitates the electrical signal conduction between plant sensors and actuators. Electrical processes hold significant importance in the physiology of plants, trees, fruits, and seeds. The resistance of a memristor relies on its previous state, and this characteristic can be employed to emulate synaptic connections in the brain. In 2018, Dang et al. created a high-performance physical transient synapse utilizing memory resistors, designed for secure neuromorphic computing applications.35 They constructed physically transient and biodegradable amnesia devices using a W/MgO/ZnO/Mo configuration. The device exhibits exceptional analog switching characteristics and high reliability on silk protein substrates. It can be dissolved in deionized water for 7 minutes at room temperature, leading to synaptic device failure or disappearance, thereby emulating the process of biological neuronal cell death. In 2018, Barlucea et al. achieved the first detection of Ebola matrix proteins using nanoscale electronic sensors operating in memristor mode.36 In 2020, Mohamad et al. demonstrated the utility of non-structural protein 1 (NS1) as a widely adopted biomarker for early dengue diagnosis.37 In their research, they employed a fluid-based memristor sensor with label-free hair detection to identify NS1 protein. This sensor offers an alternative for label-free NS1 protein diagnosis under wet conditions, facilitating early dengue detection. In 2022, Mao et al. developed a magnetic field-regulated memristor to safeguard the well-being of specific populations.38 They achieved this by detecting the intensity of the surrounding magnetic field through the construction of an Ag/Cu/MnO/Ti memristor. This memristor was then incorporated into the design of an implantable detector that could be controlled and regulated by a magnetic field. In 2023, Chen et al. achieved significant potential for utilizing memristors as next-generation implantable multi-level resistive memory devices for long-term human health monitoring, as shown in Fig. 4b, through the construction of Pt/AlOOH/ITO memristor devices.39 The application of implantable electronic devices in synapses, neural systems, and sensory systems, among others. The memristor-based biosensing system is shown in Fig. 4c.
In 2020, Jang et al. presented a one-transistor-two-memristor (1T2M) synaptic device, its array structure, and its operationalization in neurological applications, their proposed array structure was robust to the latent path problem, and superior pattern recognition accuracy was confirmed using artificial neural network simulations.45 In 2021, Pei et al. proposed a completely memristor-based artificial visual perception neural system (AVPNS) and utilized photonic memristors and threshold switching (TS) memristors on nanosheets to achieve synaptic and leaky integrate-and-fire (LIF) neuron functions, respectively. Their work demonstrates that the functionality of the biological visual neural system can be systemically emulated through a hardware system based on memristors.46 In 2022, they further developed an efficient multifunctional artificial vision system capable of recognition, memory, and self-protection through the utilization of a Sb2Se3/CdS-core/shell (SC) nanorod array optoelectronic memristor, a threshold-switching memristor, and an electrochemical actuator. When the photonic memristor is activated, it can initiate the movement of an electrochemical actuator, simulating the contraction of eye muscles and replicating the self-protective response of closing the eyes when exposed to intense light, akin to human eyes. This work provides a potential technological avenue for the application of memristors in biosensor systems.47
In 2023, Chen et al. proposed that CsPbBr3-based amnesia resistors with high on/off ratios, stable durability, and multilevel resistive memories could be used as artificial synapses to realize basic biological synaptic functions and neuromorphic computation based on controllable resistive modulation.48
In the same year, Yan et al. introduced a stable ferroelectric memristor based on the Pd/BaTiO3:Eu2O3/La0.67Sr0.33MnO3 structure grown on a silicon substrate with SrTiO3 serving as the buffer layer. The device exhibits a low coercive field voltage range of −1.3 to 2.1 V and robust endurance characteristics with approximately 1010 cycles achieved through optimization of the growth temperature. By integrating a pressure sensor, a photosensitive sensor, and a robotic arm, they reported, for the first time, a highly stable artificial multimodal sensory memory system with both visual and tactile functionalities.49 This work provides potential applications for memristor sensors in robotic perception systems.
In 2020, Rajasekaran et al. introduced a TaOx/AlN-based flexible and transparent memristor with stability and durability under extreme conditions, and the device showed excellent flexibility under extreme bending conditions, which has excellent potential for wearable applications.52 In the same year, Liu et al. introduced a transparent flexible pressure sensor with microporous dielectrics recruited only by filling them with matching ionomers and demonstrated that the transparent flexible sensor could be used as a skin touch screen and smart interface.53 For cardiovascular medical applications, Chen et al. developed a flexible layered elastomer-tuned self-powered pressure sensor that achieves high sensitivity and a wide pressure range and also achieves fast response, high signal-to-noise ratio, and good stability. Fig. 5c shows a schematic diagram of cardiovascular monitoring.54
Fig. 5 Schematic diagram of smart healthcare applied to wearable devices: (a) schematic of in vivo diabetes diagnosis and treatment with smart contact lenses.56 copyright 2020, AMER ASSOC ADVANCEMENT SCIENCE. (b) The tester rotated his head from side to side at a frequency of 0.75 Hz.57 copyright 2021, Springer Nature. (c) Schematic diagram of the structure of HSPS and its use for real-time continuous cardiovascular monitoring.54 copyright 2020, Elsevier. (d) Structural design of muscle fibre-inspired piezoelectric textiles.58 copyright 2021, Wiley-VCH. (e) 3D hierarchical interleaved PME map based on PVDF/ZnO fibres for muscle behaviour monitoring.55 copyright 2020, Elsevier. (f) Walking sensor schematic.59 copyright 2021, Wiley-VCH. |
At the same time, they demonstrated that self-powered pressure sensors are not only sensitive for monitoring pulse, arterial, and cardiac conditions. In order to accurately monitor biological subtle signals, Yang et al. prepared a 3D hierarchical interlocking piezoelectric sensor based on PVDF/ZnO nanofibres by epitaxially growing ZnO nanorods. The muscle behavior monitoring schematic is shown in Fig. 5e.55 A three-dimensional hierarchically interlocked PVDF/ZnO nanofibre piezoelectric sensor was prepared by epitaxially growing ZnO nanorods on the surface of electrospun PVDF nanofibres, which resulted in a fiber-based physiological monitoring electronic device (PME) with good flexibility and high gas permeability. The PME designed on this basis can accurately monitor complex and subtle physiological signals such as respiration, wrist pulse, and muscle behaviour. In diabetes medicine, Keum et al. developed smart contact lenses that can be used for continuous glucose monitoring and diabetic retinopathy treatment. Fig. 5a illustrates a schematic diagram of their smart contact lens system for diagnosis and treatment.56 The smart contact lens device is made of a biocompatible polymer and contains ultrathin flexible circuitry and a microcontroller chip for real-time electrochemical biosensing, on-demand controlled drug delivery, wireless power management, and data communication. In a diabetes model, it can measure glucose levels in tears, validated by traditional invasive blood glucose tests, and trigger the release of drugs from the reservoir for the treatment of diabetic retinopathy. In 2021, Yu et al. developed an ultra-flexible BTS-GFF/PVDF composite film and highly sensitive sensors and demonstrated that BTS-GFF/PVDF sensors can detect very small forces (e.g., water droplet) and human body movements (e.g., head rotation). As shown in Fig. 5b, it displays a schematic illustration of the test subject rotating their head at different frequencies.57
In the same year, Su et al. developed a muscle fiber-inspired non-woven piezoelectric textile with adjustable mechanical properties for wearable physiological monitoring systems. The structural design and a functional schematic of piezoelectric textiles inspired by the MFP textile are shown in Fig. 5d.58 Meanwhile, Mariello et al. present a novel flexible, ultrathin wearable sensor that detects impulsive forces of sudden movements as well as slower microfriction phenomena with high sensitivity and a wide measurement range, ensuring stable and reproducible recognition of bio-signals. Fig. 5f shows a schematic diagram of gait and walking sensors.59 The versatility of the sensor is demonstrated by the recognition of gait walking and the monitoring of human joint movements. Feng et al. demonstrated that pearl-inspired enhancement of sensor sensitivity through spatially regulated cracking can be a pathway to developing sensors through crack engineering.60 In 2022, Zhou et al. reported flexible piezoelectric sensors based on self-assembled 10 nm BaTiO3 nanocubes on glass fibre fabrics (GFFs), which achieved high sensitivity and short response times and the sensors could intelligently identify note or keyboard users.61
In the same year, Li et al. designed a flexible pressure sensor with engineered microstructures on polydimethylsiloxane (PDMS) film, which has excellent bending and torsional strain detection properties, is mechanically durable, and is expected to be applied to wearable biosensor technology in healthcare.62 In 2023, Xie et al. summarised recent research advances in amnesia for smart healthcare as well as wearable applications.63 The application of wearable amnesia sensors is shown in Fig. 5.
The memristor exhibits significant potential advantages in the field of smart healthcare applications: (1) memristors can store information in wearable electronic devices and retain it even after power interruptions, reducing energy consumption. This is particularly crucial for implantable medical devices such as pacemakers, which require extended operation and minimize the need for battery replacements.64 (2) Memristors are of paramount importance for the storage of extensive patient data and medical images. Their high-density storage capacity can enhance the performance and efficiency of medical devices.65 (3) Memristors can emulate synaptic functions, enabling medical devices to learn and adapt to evolving patient needs and conditions.66 This feature enhances the precision and personalization of healthcare. Overall, memristors have tremendous potential in the field of smart healthcare, as they can enhance the performance, efficiency, and biocompatibility of medical devices, ultimately leading to improvements in medical treatment and patient care.
The applications of memristors in artificial intelligence medicine primarily fall into three major categories: ultra-thin films, degradable memristors, and biodegradable memristors. They represent ultra-thin film based memristors.68,69 Ultra-thin film memristors have the capacity to minimize electronic waste and can be easily applied to curved and dynamic surfaces, akin to cling film. As such, they find applications in mobile electronic devices, artificial intelligence, healthcare, and biomedical systems, as shown in Fig. 6a–c. Fig. 6d and e represent bio-degradable memristors.70–72 With the rapid advancement of technology, there is a massive production scale of electronic components and an extremely fast pace of updates. This inevitably generates a significant amount of electronic waste, leading to severe environmental pollution. The introduction of degradable memristors effectively addresses these issues. Degradable memristors, when applied in implantable medical devices, can degrade or be absorbed by the human body after fulfilling their designated functions. Degradable memristors hold considerable application value in fields such as medicine, implantable equipment, and eco-friendly electronic products. They represent natural biomaterials based-memristors. The schematic diagram is shown in Fig. 6g–i.73–75 Natural biological materials are gifts from nature, and their utilization in the preparation and research of natural bio-memristors, such as soybeans, silk, and leaves, holds immense potential for applications in the manufacturing of implantable electronic devices and medical materials.76–80 The memristor devices hold significant potential applications in the field of biomedicine, simultaneously laying the foundation for the development of implantable neural systems.
Fig. 6 Memristors in artificial intelligence in medicine: (a) schematic of ultrathin plastic electronic foils and feather float.68 Copyright 2013, NPG. (b) The image of the memory foil can be sustained by human hair.69 Copyright 2016, Wiley-VCH. (c) The picture of tactile sensor sheet tightly conforming to a model of the human upper jaw.68 Copyright 2013, NPG. (d) The image of Mg/Ag-doped chitosan/ITO memristor dissolved in DI water.70 Copyright 2023, Wiley-VCH. Copyright 2023, Elsevier. (e) Photograph of the Ag/pectin/ITO memristor. Inset: Optical pictures of pectin powders and suspension extracted from orange peel.71 Copyright 2018, Wiley-VCH. (f) The silk protein resistive switching devices dissolved in deionized water or in phosphate-buffered saline.72 Copyright 2016, Wiley-VCH. (g) Schematic of Al/soybean MWCNT/ITO/glass memristor devices simulating the potentiation and suppression behavior of biological synapses.73 Copyright 2023, Elsevier. (h) Schematic of mesoscopic bioelectronic hybrid materials of silk fibroin (SF)-Ag nanoclusters (AgNCs@BSA; BSA: bovine serum albumin) memristors.74 Copyright 2019, Wiley-VCH. (i) Schematic of a memristor (flexible, transparent, and biocompatible resistive switching random access memory (ReRAM)) applied in wearable electronic devices.75 Copyright 2021, American Chemical Society. |
Currently, numerous challenges remain in realizing high-performance and multifunctional applications based on amnesia, including the following aspects: (1) there is still a lack of reliable technology to ensure the high performance of amnesia; (2) some biological amnesia can be easily denatured or even decomposed under high pressure, high temperature, or other harsh conditions, which impedes the preparation of structurally sophisticated amnesia; and (3) the mechanism for some biological amnesia is still not clear. Currently, the implementation of biological amnesia in smart healthcare and medical electronics remains a challenging issue.
To further enhance the application of amnesia in the biological domain, forthcoming researchers should emphasize the following aspects. Firstly, enhancing the performance of amnesia in the biomedical context can be achieved by developing multiple multilevel resistive storages, thereby significantly augmenting memory capacity. Concurrently, incorporating biocompatible and degradable functional materials can confer new capabilities to amnesia, such as photoelectric properties. Additionally, a combination of experimental and computational simulation methods should be employed to comprehensively analyze the memory resistance mechanism of the memristor. This would render amnesic resistors highly valuable in various fields, including logic operations, bionic synapses, neural networks, image recognition, in vivo diagnosis, and biomedical intelligent systems.81,82
This journal is © The Royal Society of Chemistry 2024 |