Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Mater Today Bio . 2026 Mar 30;38:103088. doi: 10.1016/j.mtbio.2026.103088 Search in PMC Search in PubMed View in NLM Catalog Add to search Engineering halogen-doped carbon dots for enhanced bioinspired synapses toward neuromorphic computing and neural interfaces Haotian Hao Haotian Hao a Institute of Biomedical Engineering, College of Artificial Intelligence, Taiyuan University of Technology, Taiyuan, 030024, China Find articles by Haotian Hao a, ⁎ , Xiaochen Lang Xiaochen Lang b MOE Key Laboratory of Interface Science and Engineering in Advanced Materials, College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan, 030024, China Find articles by Xiaochen Lang b , Yanli Cao Yanli Cao b MOE Key Laboratory of Interface Science and Engineering in Advanced Materials, College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan, 030024, China Find articles by Yanli Cao b , Lin Chen Lin Chen b MOE Key Laboratory of Interface Science and Engineering in Advanced Materials, College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan, 030024, China Find articles by Lin Chen b , Mixue Wang Mixue Wang c Academy of Medical Sciences, Shanxi Medical University, Taiyuan, 030600, China Find articles by Mixue Wang c , Zhibin Ding Zhibin Ding d Department of Neurology, Third Hospital of Shanxi Medical University, Shanxi Bethune Hospital, Shanxi Academy of Medical Sciences, Tongji Shanxi Hospital, Taiyuan, 030032, China Find articles by Zhibin Ding d , Yuhao Peng Yuhao Peng b MOE Key Laboratory of Interface Science and Engineering in Advanced Materials, College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan, 030024, China Find articles by Yuhao Peng b , Bai Sun Bai Sun e Frontier Institute of Science and Technology (FIST), Xi'an Jiaotong University, Xi'an, 710049, China Find articles by Bai Sun e , Meiwen An Meiwen An a Institute of Biomedical Engineering, College of Artificial Intelligence, Taiyuan University of Technology, Taiyuan, 030024, China Find articles by Meiwen An a, ⁎⁎ , Guangdong Zhou Guangdong Zhou f College of Artificial Intelligence, Chongqing Key Laboratory of Brain-inspired Computing and Intelligent Chips, MOE Key Laboratory of Luminescence Analysis and Molecular Sensors, Southwest University, Chongqing, 400715, China Find articles by Guangdong Zhou f, ⁎⁎⁎ , Yongzhen Yang Yongzhen Yang b MOE Key Laboratory of Interface Science and Engineering in Advanced Materials, College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan, 030024, China Find articles by Yongzhen Yang b, ⁎⁎⁎⁎ Author information Article notes Copyright and License information a Institute of Biomedical Engineering, College of Artificial Intelligence, Taiyuan University of Technology, Taiyuan, 030024, China b MOE Key Laboratory of Interface Science and Engineering in Advanced Materials, College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan, 030024, China c Academy of Medical Sciences, Shanxi Medical University, Taiyuan, 030600, China d Department of Neurology, Third Hospital of Shanxi Medical University, Shanxi Bethune Hospital, Shanxi Academy of Medical Sciences, Tongji Shanxi Hospital, Taiyuan, 030032, China e Frontier Institute of Science and Technology (FIST), Xi'an Jiaotong University, Xi'an, 710049, China f College of Artificial Intelligence, Chongqing Key Laboratory of Brain-inspired Computing and Intelligent Chips, MOE Key Laboratory of Luminescence Analysis and Molecular Sensors, Southwest University, Chongqing, 400715, China ⁎ Corresponding author. [email protected] ⁎⁎ Corresponding author. [email protected] ⁎⁎⁎ Corresponding author. [email protected] ⁎⁎⁎⁎ Corresponding author. [email protected] Received 2025 Nov 25; Revised 2026 Mar 27; Accepted 2026 Mar 30; Collection date 2026 Jun. © 2026 The Authors This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). PMC Copyright notice PMCID: PMC13068863 PMID: 41970257 Abstract Neural interfaces demand memristor-based artificial synapses with comprehensive performance for neuromorphic computing and man-machine interaction. This study proposes a halogen doping engineering to enhance carbon dots (CDs)-based memristors performance for artificial synapses. Halogen-doped CDs (FCDs, ClCDs, BrCDs) and undoped CDs (UCDs) were synthesized via a solvothermal method. Systematic characterization confirms successful doping and reveals that Br doping optimally modulates electronic structure. The BrCDs-based memristor demonstrates the best memristor performance among these devices. Experimental and computational results illustrate that appropriate electronegativity of Br atoms can facilitate electron trapping and detrapping process. As a bioinspired synapse, the device successfully mimics key short-term and long-term plasticity, and demonstrates excellent performance in image classification tasks. Furthermore, the BrCDs-based device serves as the core of an artificial neural interface chip, successfully enabling chemical neurotransmitter dopamine release and eliciting Ca 2+ responses in PC12 cells, realizing information communication with the neural system and addressing the mismatch between conventional electronic interfaces and the chemical signaling of biological synapses. This work exhibits the potential of halogen-doped CDs, particularly BrCDs, in advancing artificial synapses integrating the functions of neuromorphic computing and adaptive learning interaction. Keywords: Neural interface, Artificial synapse, Carbon dot, Halogen-doped engineering, Memristor Graphical abstract Neural interfaces require high-performance memristors. This study enhances carbon dots (CDs)-based memristors via halogen doping (F, Cl, Br). Bromine-doped CDs (BrCDs) demonstrate optimal memristive behavior by modulating electronic structures. The BrCDs-based device mimics key synaptic plasticity, achieving high accuracy in image classification. Furthermore, it functions as a core of neural interface, successfully realizing information communication with neural cells. Open in a new tab 1. Introduction Neural interface technology holds the transformative potential to bridge the human nervous system with external devices, offering revolutionary solutions for neuroprosthetics, neurological rehabilitation, and even cognitive control [ [1] , [2] , [3] ]. The development of neural interfaces that mediate the interaction between biological neural signals and external electrical signals is crucial under this technical framework. For the pursuit of efficient, low-power, and adaptive learning interactions, neural interface chips based on artificial synapses represent a highly promising direction, as they can process the complex, non-linear signals from the brain in a more natural way, and the information processing mechanism that is closely similar to biological synapses [ 4 ]. An ideal artificial synapse should perform fundamental functions such as information memory, activation, and transmission, and support brain-inspired neuromorphic computing with in-memory computing capability, massive parallelism, and advanced cognitive features [ [5] , [6] , [7] ]. By providing output channels for brain-like neural computing and completing information exchange, artificial synapses thereby play a key role in developing next-generation advanced neural interfaces [ 8 ]. The memristor is a key physical device for building an artificial synapse, which mimics the pre-synaptic, synaptic cleft, and post-synaptic structure of biological synapses and utilizes ion or electron migration to achieve information memory and signal switching [ 9 , 10 ]. Its switching functionality of the current level can be applied to simulate the release of neurotransmitters [ 11 ]. Thus, the memristor represents a promising artificial synaptic device suitable for neural interfaces. However, biological synapses rely on chemical neurotransmitter release, which is unusual in conventional electronic interfaces that primarily exchange electrical signals. Bridging this electrical-chemical gap is therefore essential for bio-integrated neural interfaces. In this regard, artificial synapses that interface with biological systems through chemical signaling, rather than relying on conventional electrical signals, represent an emerging but critical direction. The memristor performance is fundamentally governed by the properties of its functional materials [ 12 ]. Carbon dots (CDs) represent a promising material for novel memristor-based artificial synapse, owing to their tunable memristance, facile synthesis, and compatibility with implantable applications [ 9 , 13 ]. In particular, the doping strategy of heteroatoms into CDs provides an effective method for precise engineering their electronic structure, allowing for tailored adjustment of memristor performances (switching voltage, stability, endurance, etc .), and thereby facilitates optimized performance in brain-inspired neural applications [ 14 , 15 ]. Halogen doping is anticipated to modulate memristor performance by adjusting energy barriers, regulating electron migration, and tuning ion transport dynamics [ 16 , 17 ]. Halogen atoms, as an n-type dopant [ 18 ], possess electronegativity and can serve as charge trapping centers within the functional layer, thereby enhancing the charge trapping capability of CDs. This makes it crucial for electronic engineering, as it directly governs the conduction mechanism of the device, which in turn fundamentally influences key performance such as threshold voltage, current level, conductance linearity and symmetry [ 19 , 20 ]. Consequently, a systematic study on halogen (e.g., F, Cl, Br) doping in CDs is essential to clarify the structure-property relationships between halogen species and device performance, which will in turn provide material design principles for memristors toward brain-like neural computing and neural interface. In this work, we propose a halogen doping engineering to improve the performance of CDs-based memristors possessing brain-like neuromorphic computing and neural interface functions. Halogen-doped CDs (FCDs, ClCDs, BrCDs) and undoped CDs (UCDs) were synthesized via a solvothermal method, and a systematic comparison reveals that BrCDs-based memristors exhibit the best performance, establishing a clear halogen doping strategy for optimizing device performance. Combined experimental and computational analyses further elucidate the mechanism, demonstrating that the halogen doping, particularly with moderate electronegativity of Br, effectively facilitates balanced electron trapping and detrapping, a key insight that provides a clear structure-property relationship for designing memristors. Building on this optimized material, the BrCDs-based memristor is explored as a bioinspired artificial synapse that integrates both neuromorphic computing and neural interface functions. It successfully mimics synaptic plasticity and achieves high recognition accuracy in image classification, while also serving as the core of a neural interface chip that enables neurotransmitter-mediated communication with neural cells. This work demonstrates the potential of halogen-doped CDs, particularly BrCDs, in advancing artificial synapses toward neural interfaces that integrate neuromorphic computing with bio-interactive functions. 2. Materials and methods 2.1. CDs synthesis A mixture of deionized water and N, N-dimethylformamide (DMF) (volume ratio = 1:1) was stirred uniformly in a beaker. Then, phloroglucinol, urea, and ammonium bromide (molar ratio = 3:1:6) were added under continuous stirring. The mixture was sonicated for 10 min to achieve complete dissolution, which was then transferred into a 30 mL Teflon-lined stainless autoclave and reacted at 180 °C for 6 h in an oven. The resulting solution was filtered through a 0.22 μm membrane, dialyzed (500 Da) for 48 h, and freeze-dried to obtain BrCDs powder. The same procedure can be applied for the synthesis of FCDs and ClCDs by replacing ammonium bromide with ammonium fluoride and ammonium chloride, respectively. UCDs were synthesized without the halogen dopant. 2.2. Fabrication of heat-triggered DA-loaded PVA hydrogel Heat-triggered DA-loaded PVA hydrogel (DA@PVA hydrogel), which can release the neurotransmitter dopamine (DA) upon heating, is an important component of neural interface chips. To fabricate a PVA hydrogel, 3.8 g of low-molecular-weight PVA (13,000∼23,000) and 0.2 g of high-molecular-weight PVA (146,000∼186,000) were first dissolved in deionized water (90 °C) under vigorous stirring. When fully dissolved to form a uniform hydrogel, cool it to room temperature. Add DA (100 mg) and keep stirring for one day under the N 2 atmosphere. The resulting solution was subjected to cyclic freezing and thawing, beginning with one day at −20 °C, followed by 4 h at room temperature and then four additional cycles, each consisting of 4 h at −20 °C and 4 h at room temperature, finally forming the cross-linked hydrogel. 2.3. Device fabrication Indium tin oxide (ITO)-coated glass substrates (19.10 mm × 19.10 mm) were initially cleaned with detergent and flowing deionized water to remove surface contaminants. Subsequently, the ITO substrates were ultrasonicated in acetone, deionized water, and isopropanol for 25 min each, followed by drying in a convection oven. Prior to device fabrication, the substrates were exposed to UV-ozone treatment for 25 min to eliminate ITO residues and enhance surface wettability. FCDs, ClCDs, BrCDs, and UCDs dissolved in anhydrous ethanol were filtered through a polytetrafluoroethylene filter (0.22 μm) and then spin-coated on the pre-cleaned ITO substrates. This solution processing has become more popular due to its simplicity, low cost, and characteristics of organic material [ [21] , [22] , [23] , [24] ]. The films were annealed on a hotplate at 80 °C for 20 min. Finally, top aluminum (Al) electrodes were thermally evaporated through a shadow mask with a 100 μm linewidth in a high-vacuum chamber to complete the fabrication of CDs-based memristors. All the fabrication methods of CDs based memristors in this manuscript use the same parameters. 2.4. Fabrication of the neural interface chip A memristor array monolithically integrated with a series-connected heater was fabricated on a flexible polyethylene terephthalate (PET) substrate. This memristor-heater configuration modulates the release profile of DA-loaded PVA hydrogel. A polydimethylsiloxane (PDMS) microfluidic film, loading the DA-loaded PVA hydrogel, was securely bonded to the heater surface. The PDMS microfluidic film was fabricated by cross-linking and curing its precursor in a custom-designed mold, with its size precisely covering the substrate and allowing the PVA hydrogel to be embedded. Upon activation, the liquid DA@PVA mixture flows through microchannels into the target neural cells. The CDs-based memristor array, PDMS microfluidic film, and DA-loaded PVA hydrogel constitute an integrated neural interface chip ( Fig. S21 ). 2.5. Incubation and response of cells The neural cell (PC12) was cultured under standard culture conditions (37 °C, 5% CO 2 ) in high-glucose Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. To assess whether PC12 cell was activated by DA released from the neural interface chip, Fluo-4 AM calcium indicator was applied to the cells for 30 min. After removing the probe solution and gently rinsing with phosphate buffered saline (PBS), the cells were incubated with the DA-containing solution released from the chip. Ca 2+ imaging was then observed at different times using a confocal laser scanning microscope (OLYMPUS BX60) to monitor dynamic changes in intracellular Ca 2+ levels. 3. Result and discussion 3.1. Structure and morphology of CDs Halogen-doped CDs (FCDs, ClCDs, and BrCDs) were synthesized via a solvothermal method utilizing phloroglucinol, urea, and various halogen dopants (NH 4 F, NH 4 Cl, and NH 4 Br) as reaction precursors. DMF and deionized water were selected as reaction solvents ( Fig. 1 A). The preparation method for UCDs as a control group is similar to that of the halogen-doped CDs, with the only difference existing in the absence of halogen dopant. As the carbon source, phloroglucinol contains a benzene ring and hydroxyl groups, which is favorable for dehydration and carbonization to form a larger conjugated structure [ 25 ]. As a high-boiling solvent, DMF is helpful for dehydration and carbonization of the reaction precursor to form a larger sp 2 conjugated structure [ 26 ], which is generally favorable to reducing fluctuations of threshold voltages in CDs-based memristors [ 12 ]. Fig. 1. Open in a new tab (A) The synthetic route and structure characterization of CDs. Transmission electron microscope (TEM), high resolution TEM (HRTEM), and particle size distribution histogram of (B) UCDs, (C)FCDs, (D) ClCDs, and (E) BrCDs. (F) FTIR spectra, (G) XRD spectra, (H) Raman spectra, and (I) XPS full survey spectra of the synthesized CDs. High-resolution XPS spectra for the (J) C 1s of UCDs, (K) F 1s of FCDs, (L) Cl 2p of ClCDs, and (M) Br 3d of BrCDs. The detailed structure and morphology of these CDs are exhibited in Fig. 1 . It can be seen that the four prepared CDs are spherical, evenly dispersed, and the particle size is mainly concentrated in a range of 1.6∼3.7 nm ( Fig. 1 B–E). HRTEM reveals that all CDs exhibit distinct lattice stripes with a lattice spacing ( d ) of approximately 0.21 nm, which corresponds to the (100) lattice plane of the graphitic structure [ 12 ]. The Fourier transform infrared (FTIR) result ( Fig. 1 F– Table S1 ) reveals that all CDs contain O–H/N–H (3219∼3342 cm −1 ), C=O (1649∼1653 cm −1 ), C=C (1589∼1598 cm −1 ), and C–N (1344∼1352 cm −1 ) bonds, in which –OH group is beneficial for constructing low-threshold-voltage memristors [ 27 ]. Moreover, the signals of C–F, C–Cl, and C–Br appear at 1159, 789, and 677 cm −1 , respectively, suggesting that the F, Cl, and Br elements have been doped into CDs [ [28] , [29] , [30] ]. The crystalline phase structure was further examined ( Fig. 1 G). All CDs exhibit a similar prominent diffraction peak, indicating that the doping of halogen atoms does not significantly change the CDs crystal structure but occurs primarily on their surface. This allows modulating device performance by influencing the conduction mechanism of the memristor. D bands (1355∼1361 cm −1 ) and G bands (1569∼1579 cm −1 ) of CDs are all emerged in Raman spectra ( Fig. 1 H), and the corresponding intensity ratios ( I D /I G ) of UCDs, FCDs, ClCDs, and BrCDs are 0.77, 0.75, 0.68, and 0.60, indicating that the synthesized CDs are well-ordered [ 12 ], and halogen dopants can improve structure ordering. The lower electronegativity of the doped halogen correlates with a higher I G /I D ratio, which may be attributed to F or Cl elements with greater electronegativity being more likely to cause electron cloud displacement in the carbon skeleton, leading to local structural distortion. The CDs are further characterized by X-ray photoelectron spectroscopy (XPS) ( Fig. 1 I), with particular focus on determining their compositions and atomic bonds. The full XPS spectrum of CDs reveals that they are primarily constituted of C, N, and O elements. Deconvolution analysis of high-resolution C 1s spectrum of UCDs provides three distinct constituent signals, including C=C/C–C (283.0 eV), C–N (284.3 eV), and C=O (286.4 eV) ( Fig. 1 J) [ 31 ]. The high-resolution N 1s spectrum of UCDs exists in two forms, corresponding to pyridinic N (398.3 eV) and pyrrolic N (399.2 eV), while the O 1s spectrum is mainly composed of C–O bond (531.4 eV), accompany with C=O signal (529.5 eV) ( Fig. S1 ). The main chemical compositions and atomic bonds of halogen-doped CDs are similar to those of UCDs ( Fig. S2–S4 ), but the high-resolution XPS reveals distinct characteristic peaks in the F 1s, Cl 2p, and Br 3d spectra, corresponding to their respective halogen dopants ( Table S2 ). Specifically, the F 1s spectrum exhibits a single characteristic peak at 687.1 eV, assigned to the covalent C–F bond ( Fig. 1 K) [ 32 ]. The Cl 2p spectrum is deconvoluted into peaks at 201.8 eV (C–Cl 2p 3/2 ) and 197.5 eV (C−Cl 2p 1/2 ) ( Fig. 1 L) [ 33 ]. Similarly, the Br 3d spectrum shows two fitted peaks at 71.6 eV (C–Br 3d 5/2 ) and 70.5 eV (C–Br 3d 3/2 ) ( Fig. 1 M) [ 34 ]. These features provide evidence for the successful incorporation of the heteroatoms (F, Cl, Br) into the CDs structure. Furthermore, the thermal stability of various CDs is also measured ( Fig. S5 ), which manifests almost no weight loss before 200 °C, indicating that they have a more thermally stable carbon core structure and implying that the corresponding memristors will have high memristance stability during multiple SET and RESET operations in normal operating environments. 3.2. Memristor performance All memristors based on CDs were fabricated with a sandwich structure, including an ITO bottom electrode, a CDs function layer, and an Al top electrode ( Fig. 2 A). The fabrication process was systematically optimized by investigating the parameters of CDs concentration, spin-coating revolution speed, and annealing temperature. Firstly, designing CDs concentration gradient of 1.0, 1.5, 2.0, 2.5 mg/mL to fabricate a functional layer. When employing CDs with a concentration of 2 mg/mL, the fabricated memristors show a lower threshold voltage and demonstrate more obvious memristive switching behavior than others ( Fig. S6A–D ). Then, selecting 2 mg/mL as the CDs concentration for optimizing spin-coating revolution speed. The memristors based on the four kinds of synthetic CDs hardly exhibit obvious hysteresis behavior at a revolution speed of 3000 rpm, which may be attributed to the fact that the functional layer is too thin to present a memristive switching phenomenon [ 12 ]. When the spin-coating speed is below 3000 rpm, devices fabricated at 2000 rpm exhibit the lowest threshold voltages and the smoothest bipolar current-voltage ( I-V ) curves, which is particularly evident in the memristor performance of halogen-doped CDs ( Fig. S6E–H ). Furthermore, the CDs films (2 mg/mL, 2000 rpm) are optimized by designing different annealing temperatures. The annealing process can effectively reduce defect density, and enhance film quality, adhesion, and electrical properties. The I-V curves of devices fabricated at different annealing temperatures (60, 80, and 100 °C) are also presented in Fig. S6I–L . Notably, devices annealed at 80 °C further demonstrate a smaller threshold voltage compared to those processed at lower (60 °C) and higher (100 °C) temperatures. This is primarily attributed to the improved surface structure and reduced defect density of the film, facilitating easier device conducting. Annealing at 60 °C, residual solvents and increased defect density result in suboptimal device performance. In contrast, annealing at 100 °C induces agglomeration of the CDs, caused by solvent evaporation and surface energy reduction, forming high-resistance regions that degrade overall conductivity and increase threshold voltages [ 35 ]. Therefore, the nonvolatile memristors fabricated at 2000 rpm, 80 °C and 2 mg/mL are selected for further investigation. Fig. 2. Open in a new tab (A) The fabrication process of CDs-based memristors. (B–E) Morphology images of the CDs films obtained from atomic force microscope (AFM). Each figure is composed of a 2D AFM image (top) and a 3D AFM image (bottom). (F–I) The cyclic I-V curve and (J–M) retention time of the devices. The result of atomic force microscope (AFM) images in Fig. 2 B–E reveals that CDs functional layer possesses particularly low root-mean-square (RMS) surface roughness (UCDs (1.16 nm), FCDs (1.15 nm), ClCDs (1.04 nm), BrCDs (1.10 nm)), which is conducive to improving the repeatability of device performance [ 12 ]. The cyclic tests (0 V → 4 V → −4 V → 0 V) of CDs-based memristors exhibit the device stable operation ( Fig. 2 F–I). In these memristors, the devices based on halogen-doped CDs are obviously more stable than UCDs-based memristor, and BrCDs-based memristor manifests as the lowest temporal (cycle-to-cycle) variability and the highest reproducibility. Regarding the retention time of the ON and OFF states ( Fig. 2 J–M), all other CDs-based memristors exhibit a certain degree of current decay or instability over a measurement period of 10 4 s, while the BrCDs-based device shows hardly any significant current decay over 1.2 × 10 5 s. Its performance is competitive comparing the other reported CDs-based memristors ( Table S3 ). The distribution statistics in Fig. 3 A–D clearly show threshold voltages in specific ranges for memristors based on different CDs, including UCDs ( V SET : 1.3∼2.5 V; V RESET : −3.6∼−2.4 V), FCDs ( V SET : 0.7∼1.6 V; V RESET : −3.2∼−2.0 V), ClCDs ( V SET : 0.7∼1.45 V; V RESET : −2.6∼−1.7 V), and BrCDs ( V SET : 0.45∼0.95 V; V RESET : −2.0∼−1.25 V). Notably, the distributions of both V SET and V RESET exhibit a progressive decrease and narrowing from FCDs to BrCDs, correlating directly with the decreasing electronegativity of the halogen dopant, which is beneficial for reducing device energy consumption. Threshold voltage fluctuation also decreases. Variation coefficient (the ratio of the standard deviation to the mean, σ / u ) is usually used to describe relative fluctuation. BrCDs-based memristors present the smallest cycle-to-cycle variability, with variation coefficients for SET and RESET operations of only 6.2% and 9.8%, respectively, lower than those of the other devices (7.9%∼16.1%) ( Fig. 3 E–H). Similarly, BrCDs-based memristors also exhibit smaller device-to-device variability (9.7%/8.7%) compared to the other devices (16.6%∼28.4%) ( Fig. S7 ). The influence of mechanical deformation on BrCDs-based memristor is also shown in Fig. S10 , and the stable current level after 1000 bending and twisting cycles confirms the device's mechanical flexibility and reliability. Moreover, the ON and OFF current values across different cycles are also measured ( Fig. 3 I–L). The high variability of the UCDs-based memristor (OFF: 159.6%; ON: 55.8%) is effectively reduced with the doping of F, Cl, and Br. Specifically, the variation coefficient for the ON current drops from 40.9% to 7.7%. These results indicate that with F, Cl, and Br doping, the distribution fluctuations of both ON and OFF currents become narrower and more stable. Fig. 3. Open in a new tab (A–D) V SET and V RESET cumulative probability distribution statistics and (E–H) relative fluctuations of threshold voltages. (I–L) Currents in ON and OFF states of different cycles (read at 0.1 V). 3.3. Memristor working mechanism Compared to UCDs, halogen-doped counterparts exhibit superior memristor performance. This enhancement correlates directly with halogen electronegativity, and arises from halogen-induced modifications to charge transport, CDs bandgaps, and energy levels. To examine the charge transport characteristics of the device, using representative BrCDs to illustrate, Kelvin probe force microscopy (KPFM) was employed to measure the surface potential, thereby evaluating the electron injection process into the film. Local carrier trapping behavior in the film can be assessed via KPFM using a Pt/Ir-coated probe to inject charge carriers on the surface ( Fig. 4 A). A voltage of −5 V was applied to the central region and +5 V to the periphery of the stamped area ( Fig. S11 ). It is observed that a 3 μm × 3 μm region exhibits reduced surface potential, in contrast to the edge of the stamped area ( Fig. 4 B). The potential profile along the purple line reveals a decrease from an initial 200 mV to roughly 160 mV after applying the negative voltage ( Fig. 4 C). This indicates that electrons are predominantly trapped in the BrCDs film, and holes serve as the main charge carriers across the active layer. Fig. 4. Open in a new tab (A) Schematic representation of surface potential measurement of the BrCDs film using the contact mode of KPFM using a Pt/Ir-coated probe. (B) Surface potential distribution diagram after the stamped voltage at the center (−5 V) and edge (+5 V). (C) Electrical potential distribution at the cross-section shown by the purple line in (B). (D) Summarized plots of the HOMO and LUMO energy levels of halogen-doped CDs obtained from the CV and Tauc curves. Visualization of (E) optimized model structure, (F, I) DFT-computed frontier orbitals (LUMO/HOMO), and (H) ESP-mapped van der Waals surface of BrCDs at B3LYP/6-31G (d, p) level using Multiwfn and VMD. (G) Double logarithmic fitting I-V curve of the BrCDs-based memristor. (J–L) Mechanism diagrams of memristors based on halogen-doped CDs. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Ultraviolet-visible (UV-vis) and cyclic voltammetry (CV) analyses provide complementary data that jointly allow for the determination of the bandgap and orbital energy level. The absorption peak blueshifts from 463 nm in BrCDs to 456 nm in FCDs, with ClCDs intermediate at 460 nm ( Fig. S13 ). This wavelength shift suggests intramolecular local donor-acceptor charge transfer from the electron-donating fused aromatic ring to electron-accepting halogen atoms of increasing electronegativity. Taking BrCDs as an example ( Fig. S13–S15 ), the optical bandgap is estimated as 2.24 eV utilizing the Tauc method. The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels, determined by cyclic voltammetry (CV) and optical bandgap results [ 12 ], are measured at −5.83 eV and −3.59 eV, respectively. Similarly determined bandgaps and energy levels for FCDs and ClCDs are presented in Fig. 4 D. These results demonstrate a gradual reduction in energy levels with halogen doping (F, Cl, Br), concomitantly lowering the energy barrier between the LUMO level and the Al electrode. Density functional theory (DFT) calculations at the B3LYP/6-31G (d,p) level further confirm this trend. The optimized structure, HOMO/LUMO distribution, and surface electrostatic potential (ESP) map of different CDs, are generated using Multiwfn and VMD ( Fig. 4E–I, S16 , S17) [ 36 , 37 ], yield calculated HOMO/LUMO energies of FCDs (−5.57 eV/–3.21 eV), ClCDs (−5.72 eV/–3.37 eV), and BrCDs (−5.93 eV/–3.59 eV), illustrating electrons are prone to inject into BrCDs from Al electrode. A CDs model with six halogen atomic sites was performed simulation calculations using DFT ( Fig. S17A ). The calculated ESP distributions for FCDs, ClCDs, and BrCDs are shown in Fig. S17B–D . The DFT-calculated ESP maps indicate that the electron-withdrawing strength of the atoms follows the order F (electronegativity χ = 3.98) > Cl (χ = 3.16) > Br (χ = 2.96) > C (χ = 2.55), consistent with the Pauling electronegativity trend. In our BrCDs molecular model, the ESP values at the six Br sites are −1.33, −0.34, −0.31, −0.66, −0.06, and −0.43 eV ( Table S4 ). These values are less negative than those at the corresponding sites in ClCDs and FCDs models. The F sites exhibit the most negative ESP values (−1.54, −0.38, −0.34, −1.00, −0.10, and −0.50 eV), representing the deepest charge traps in FCDs. This indicates that F atoms possess the strongest electron-withdrawing capability, making trapped electrons less prone to detrapping. This theoretical finding directly correlates with the device performance. FCDs-based memristors show the largest threshold voltage variation coefficients and the lowest stability, whereas BrCDs-based devices exhibit the smallest variation and highest stability. The relatively appropriate electron-withdrawing ability of Br promotes a more balanced process where charge trapping and detrapping can both occur efficiently. This supports the optimal performance of BrCDs-based memristors among halogen-doped CDs. Furthermore, the I - V curve is fitted as the double logarithmic curve to analyze electronic transmission mechanism. For UCDs-based memristor, the slopes of the double logarithmic I-V curve at ON and OFF states are 1.03 and 1.09 ( Fig. S18A ), respectively, both corresponding to the Ohmic conduction ( I ∝ V ), where charge transport occurs primarily via thermally generated free electrons. As for BrCDs-based memristor ( Fig. 4 G), the I - V curve of OFF state exhibits two distinct linear regions: a low-voltage region with a slope of 1.27, corresponding to the Ohmic region characterized by thermally generated free electrons, and a high-voltage region with a slope of 2.42, indicative of Child's region ( I ∝ V 2 ). In this Child's region, conduction is dominated by electrons injected from the electrode, which exceeds the equilibrium carrier concentration, signifying entry into the trap-filled limit region and supporting a space charge limited current conduction mechanism. The electronic transmission mechanism of FCDs- and ClCDs-based memristors is similar to that of BrCDs ( Fig. S18B–D ). Herein, an explanation for the working mechanism of devices in halogen-doped engineering is presented, taking the BrCDs-based memristor as an example. Device operation is attributed to the electron trapping/detrapping associated with Br doping. A significant presence of electron-withdrawing Br atoms, serving as negative potential points. Under a positive electric field, Br defects capture free electrons, with the conductive pathway forming upon defect saturation to facilitate the SET transition to the ON state. Subsequently, these trapped electrons can be released under a reverse voltage. Therefore, the device performance is dominated by the process of electron trapping and detrapping. Device operation is attributed to electron trapping/detrapping at halogen-induced defect sites within the functional layer. Under low positive bias, BrCDs act as electron trapping centers, where injected electrons from electrodes are captured by surface functional groups, maintaining the OFF state ( Fig. 4 J). As voltage increases, enhanced electron injection leads to progressive electron accumulation at defect sites ( Fig. 4 K). Upon saturation of these trapping sites, excess electrons form a continuous conductive pathway bridging the electrodes, triggering the SET transition to ON state ( Fig. 4 L). The trapped electrons can be gradually released from these defects before giving a RESET operation under a reverse voltage, demonstrating that electron trapping and detrapping dominate device performance. Owing to the higher electronegativity, FCDs and ClCDs exhibit larger electron affinity relative to BrCDs. However, their stronger charge trapping impedes both electron trapping and detrapping process, which consequently increases the threshold voltage and exacerbates memristance variability. Consequently, BrCDs-based devices demonstrate superior switching performance than others. 3.4. Brain-inspired neuromorphic computing and neural interface As an important artificial synapse device, memristor is suitable for brain-inspired neuromorphic computing. Its key significance lies in the ability to naturally process the complex, non-linear signals exchanged between the brain and external devices, assisting in achieving effective man-machine interaction. The three-layer structure of a memristor, composed of the top electrode, intermediate functional layer, and bottom electrode, corresponds to the pre-synapse, synaptic cleft, and post-synapse of the neural synapse, respectively ( Fig. 5 A). This structure makes them a potential choice for an artificial synapse in the artificial neural network (ANN). In biological neural synapses, the excitatory postsynaptic current (EPSC) plays a critical role in excitatory signals, which promotes neuronal activation and ultimately leads to the generation of action potential. Therefore, given the better performance of BrCDs-based memristor, its property as an artificial synapse is further explored. Specifically, the BrCDs-based memristor exhibits a current response that gradually decays until the next pulse is applied, demonstrating a typical EPSC behavior. The amplitude of the second EPSC consistently exceeds that of the first, which is characteristic of paired-pulse facilitation (PPF) ( Fig. 5 B). PPF phenomenon reflects the temporal interaction between consecutive neural spikes, and is like the Ca 2+ dynamics in biology [ 38 , 39 ]. PPF represents the temporal correlation between adjacent spikes [ 38 ], and can be triggered by two stimuli where EPSC induced by the second stimulus (A 2 ) is larger than that triggered by the first stimulus (A 1 ). A schematic diagram of PPF is inserted in Fig. 5 C. The PPF index is obtained by fitting the ratio across varying pulse intervals to the following equation: PPF index = 1 + C 1 exp ( − Δ t / τ 1 ) + C 2 exp ( − Δ t / τ 2 ) where C 1 and C 2 denote the initial magnitudes of facilitation, Δ t is interval time, τ 1 and τ 2 represent relaxation times. A shorter interval time between spikes corresponds to a higher PPF index ( Fig. 5 C), indicating that the device exhibits strong facilitation behavior. Besides, the spiking-timing-dependent plasticity (STDP), one of the representative behaviors of unsupervised learning in neural networks, is also studied. As an advanced form of synaptic learning, STDP refines the classical Hebbian rule by precisely modulating the interval time between the pre- and postsynaptic stimuli to induce either long-term potentiation (LTP) or long-term depression (LTD) [ 38 , 40 ]. The synaptic weight (ΔG) increases (potentiation) if the presynaptic spike arrives before the postsynaptic spike, and decreases (depression) if the order is reversed ( Fig. 5 D). Furthermore, spike-number-dependent plasticity (SNDP) is influenced by the number of applied pulses. As shown in Fig. 5 E, a heightened EPSC response is consistently observed as the number of applied pulses increases. This direct relationship demonstrates that the transition from short-term memory to long-term memory behaviors can be effectively modulated by increasing the number of stimuli. In contrast, a decrease in EPSC response under the reverse electric pulse exhibits the LTD characteristics ( Fig. 5 F), indicating a weakening of synaptic connections, which is linked to a phenomenon of forgetting. Fig. 5 G illustrates that the device displays gradual and repeatable LTP (60 positive pulses) and LDP (60 negative pulses), indicating potentiation and depression, respectively. To demonstrate the memristor stability and repeatability, ten cycles of synaptic potentiation and depression characteristics were repeated using the same pulse sequence, as illustrated in Fig. 5 H. After ten iterations of potentiation and depression cycles, the memristor's performance remains stable, indicating its excellent repeatability. Therefore, the BrCDs-based memristor can provide reliable operating repeatability and cycle stability for neuromorphic computing. Fig. 5. Open in a new tab (A) Schematic illustration of the memristor as an artificial synapse. (B) Current response of the BrCDs-based memristor under electrical pulses (+1.5 V, 50 μs) with different interval times. (C) PPF index and fitted curves of the device. Insert: a schematic diagram of PPF. (D) Implementation of STDP with BrCDs-based memristor (±1.5 V, 50 μs). Spike-number-dependent plasticity (SNDP) modulated under the stimulus spike of (E) +1.5 V and (F) −1.5 V with different numbers. (G) Tunable LTP and LTD achieved by a sequence of pulses (60 pulses at +1.5 V/50 μs followed by 60 pulses at −1.5 V/50 μs). (H) The long-term potentiation/depression characteristics of the device repeated for 10 cycles. (I) Schematic of ANN training for classification of the Fashion MNIST dataset. (J) Recognition results on Fashion MNIST images for the synapse device. (K) Confusion matrix of the average accuracy of the learning activities for Fashion·MNIST images. To further evaluate the performance of the synaptic devices, we constructed a convolutional neural network (CNN) for image classification tasks using the PyTorch framework, which is based on the Fashion MNIST dataset and employs a structure composed of alternately stacked convolutional and pooling layers, designed for hierarchical feature extraction at multiple scales ( Fig. 5 I). In the CNN, the weight parameter (convolution kernel) obtained via backpropagation training is used in forward propagation to compute the output of the feature layers, enabling efficient feature extraction. The BrCDs-based memristor achieves an overall accuracy of 97.72% on the Fashion MNIST after 100 training epochs, with the performance gradually converging. The recognition results are in complete agreement with the truth of selecting images ( Fig. 5 J). Fig. 5 K further demonstrates high per-class accuracy across various categories in the Fashion MNIST, with accuracy ranging from 84.2% for the pullover to 98.6% for the bag, and including strong performance on the trouser, sandal, sneaker, and other classes. These results demonstrate that the BrCDs-based memristor serves as an artificial synapse, showing its significant potential for brain-inspired neuromorphic computing. The synapse-like structure and neuromorphic functionality of memristors establish them as emerging artificial synapse components for neural interfaces ( Fig. 6 A) [ 9 , 41 ], enabling information communication with neural systems. In this work, neural signaling transmission was achieved by implementing the BrCDs-based memristor as the core component of a neural interface chip ( Fig. 6 B), leveraging its highly stable modulation of ON/OFF switching. The memristor-heater ensemble gates simulate neurotransmitter release from a temperature-responsive PVA hydrogel vesicle analog while processing input stimuli, forming an integrated synapse control unit. The hydrogel vesicle system (DA@PVA) is composed of neurotransmitter DA and PVA hydrogel ( Fig. S21 ). The heater temperature rises from 36 to 43 °C under a 0∼4 V bias voltage of the memristor ( Fig. S22 ), causing the hydrogel to melt due to the Joule heat and the encapsulated DA flow past a polydimethylsiloxane (PDMS) microfluidic film to nerve cells (PC12). Biocompatibility assessment of nerve cells-adjacent components, conducted via live/dead cell staining and CCK8 assays, confirms sustained cell viability of greater than 94.3% throughout a 24 h monitoring period, validating material biosafety ( Fig. 6 C). This mimics excitation and neurotransmitter release at the presynaptic neuron. The memristor was stimulated by a 4 V bias voltage and transferred into ON state, and its current keeps about 74 mA in the subsequent constant stimulation among 240 s, which causes a gradual elevation of released DA concentration to 61.6 μM ( Fig. 6 D). The released DA is bonded to its receptors and then triggers specific G protein-coupled signaling, which potentiates intracellular calcium release and thereby modulates neuronal excitability and synaptic transmission [ 42 , 43 ]. Thus, the information communication function can be verified through the Ca 2+ fluorescence imaging technique. Initially, the released DA concentration remains negligible, leading to minimal fluorescence intensity. As the memristor current continues to pass through the heater, DA is released into PC12, and the corresponding mean fluorescence intensity progressively increases from 1.9 to 43.5, accompanied by significant brightening of PC12 cells ( Fig. 6 E and F). This confirms successful neurotransmitter transmission from the neural interface chip to nerve cells, demonstrating that the memristor facilitates neurotransmitter-mediated communication with living neural cells, a function that corresponds to the information processing mechanisms of biological neural systems, which rely on chemical neurotransmission at synapses. In contrast to conventional neural interfaces that mainly depend on electrophysiological signals [ 44 ], this approach can bridge the fundamental mismatch between electrical and chemical signaling, paving the way for advanced neural interfaces. Fig. 6. Open in a new tab (A) Schematic diagram of communication between the neural interface chip and nerve cells. (B) Schematic diagram of neural signaling via the memristor-based neural interface chip. (C) Cell viability of the PET, PVA hydrogel, and BrCDs, determined by Cell Counting Kit 8 (CCK8). Data are expressed as the mean values, with error bars derived from three independent replicates. (D) The memristor current and the corresponding released DA concentration under a constant voltage of 4 V. (E) Mean fluorescence intensity of PC12 cells under progressive DA stimulation. (F) Ca 2+ fluorescence imaging response to varying DA release durations. Scale bar: 100 μm. Moreover, electrical stimulation studies further reveal that the neurotransmitter release is dynamically regulated in a learning-dependent manner governed by the memristor's synaptic plasticity. The device exhibits long-term potentiation, with both its current (from 165 mA to 202 mA) and the consequent DA release (from 63 to 115 μM) strengthening upon repeated stimulation, and implements parameter-dependent modulation via spike-duration-dependent plasticity, allowing modulation DA concentration by varying input pulse width (details in Fig. S24 ). These results demonstrate that the neural interface not only delivers signals but also adapts its output through intrinsic synaptic learning rules, thereby realizing activity-triggered and experience-modulated neurotransmitter release. In summary, the BrCDs-based memristor serves as both an efficient artificial synapse for neuromorphic computing and the central control unit of a neural interface. It realizes adaptive, plasticity-governed conversion of electrical signals into neurotransmitter release, advancing beyond conventional electrophysiological interfaces toward bio-adaptive systems that support information communication modulated by synaptic learning. 4. Conclusion In conclusion, a halogen doping engineering is proposed to enhance the performance of CDs-based artificial synapse integrating neuromorphic computing and neural interface functions. Firstly, a series of halogen-doped CDs (FCDs, ClCDs, BrCDs) and undoped CDs were successfully synthesized via a solvothermal method. The corresponding memristors were fabricated and optimized, revealing that the BrCDs-based memristor exhibits the lowest temporal and spatial variability, the most effective retention time within a 1.2 × 10 5 s period, and the narrowest distribution of threshold voltages, which is competitive comparing the other CDs-based memristors. Subsequently, the role of halogen (especially Br) was systematically investigated through comparative experiments (CDs structure and memristor performance) and simulated calculation results. It is demonstrated that the doped halogen elements, particularly Br with appropriate electronegativity, facilitate electron trapping/detrapping processes, thereby significantly improving device performance. Furthermore, the BrCDs-based memristor exhibited classic short-term and long-term plasticity, and demonstrated excellent performance in image classification tasks. Finally, the conceptual application of a BrCDs-based memristor as the core component of an artificial neural interface chip was successfully demonstrated, achieving effective neurotransmitter DA-mediated communication with neural cells, and adapting its output through intrinsic synaptic learning rules, which bridges the fundamental mismatch between electrical and chemical signaling. The halogen doping engineering of CDs will facilitate the development of memristor-based artificial synapses towards advanced neural interface. CRediT authorship contribution statement Haotian Hao: Funding acquisition, Methodology, Writing – original draft. Xiaochen Lang: Conceptualization, Data curation. Yanli Cao: Data curation. Lin Chen: Formal analysis. Mixue Wang: Conceptualization, Project administration. Zhibin Ding: Investigation. Yuhao Peng: Data curation. Bai Sun: Resources. Meiwen An: Funding acquisition, Supervision. Guangdong Zhou: Resources, Supervision, Writing – review & editing. Yongzhen Yang: Funding acquisition, Supervision, Writing – review & editing. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This work was funded by financial support from the Postdoctoral Fellowship Program of CPSF (GZC20251905), Foundational Research Project of Shanxi Province (202503021212068), National Natural Science Foundation of China (31870934, 12272251), Shanxi Scholarship Council of China (2024-058), Science and Education Cultivation Fund of the National Cancer and Regional Medical Center of Shanxi Provincial Cancer Hospital (TD2023003). We also thank Scientific Compass www.shiyanjia.com for providing invaluable assistance. Footnotes Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103088 . Contributor Information Haotian Hao, Email: [email protected]. Meiwen An, Email: [email protected]. Guangdong Zhou, Email: [email protected]. Yongzhen Yang, Email: [email protected]. Appendix A. Supplementary data The following is the Supplementary data to this article. Multimedia component 1 mmc1.docx (9.3MB, docx) Data availability Data will be made available on request. References 1. Wu J.C., Han Q.X., Gui D.K., Qian Y. Multidimensional advances in neural interface technology for peripheral nerve repair: from material innovation to clinical translation. Mater. Today Bio. 2025;34 doi: 10.1016/j.mtbio.2025.102092. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Villa J., Cury J., Kessler L., Tan X.D., Richter C.P. Enhancing biocompatibility of the brain-machine interface: a review. Bioact. Mater. 2024;42:531–549. doi: 10.1016/j.bioactmat.2024.08.034. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Wang L., Zhang T.Y., Lei J.X., Wang S.R., Guan Y.J., Chen K.T., Li C.C., Song Y.H., Li W.N., Wang S.M., Jia Z.B., Chen S.F., Bai J., Yu B.B., Yang C., Sun P.C., Wang Q.Y., Sheng X., Peng J., Fan Y.B., Wang L.Z., Zhang M.L., Wang Y., Yin L. A biodegradable and restorative peripheral neural interface for the interrogation of neuropathic injuries. Nat. Commun. 2025;16:1716. doi: 10.1038/s41467-025-56089-1. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Go G.T., Lee Y., Seo D.G., Lee T.W. Organic neuroelectronics: from neural interfaces to neuroprosthetics. Adv. Mater. 2022;34 doi: 10.1002/adma.202201864. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Han H., Yu H.Y., Wei H.H., Gong J.D., Xu W.T. Recent progress in three-terminal artificial synapses: from device to system. Small. 2019;15 doi: 10.1002/smll.201900695. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Dong L.Y., Yuan S., Wei G.D., Zhu P.F., Ma S.F., Xu B.S., Yang Y. Artificial optoelectronic synapse based on violet phosphorus microfiber arrays. Small. 2024;20 doi: 10.1002/smll.202306998. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Chen J.G., Zhu C., Cao C.M., Liu H.S., Bian R.J., Wang J.Y., Li C.C., Chen J.Q., Fu Q.D., Liu Q., Meng P., Li W., Liu F.C., Liu Z. Mimicking neuroplasticity via ion migration in van der waals layered copper indium thiophosphate. Adv. Mater. 2022;34 doi: 10.1002/adma.202104676. [ DOI ] [ PubMed ] [ Google Scholar ] 8. Liu X.R., Sun C., Ye X.Y., Zhu X.J., Hu C., Tan H.W., He S., Shao M.J., Li R.W. Neuromorphic nanoionics for human-machine interaction: from materials to applications. Adv. Mater. 2024;36 doi: 10.1002/adma.202311472. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Hao H.T., Yan L.P., Wang M.X., Cao Y.L., He J.T., Yang Y.Z. Research process of carbon dots in memristors. Adv. Electron. Mater. 2023;9 [ Google Scholar ] 10. He J.T., Zhou G.D., Sun B., Yan L.P., Lang X.C., Yang Y.Z., Hao H.T. Graphene quantum dots induced performance enhancement in memristors. Nanoscale. 2025;17:14082–140102. doi: 10.1039/d5nr00597c. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Wang T., Wang M., Wang J., Yang L., Ren X., Song G., Chen S., Yuan Y., Liu R., Pan L., Li Z., Leow W.R., Luo Y., Ji S., Cui Z., He K., Zhang F., Lv F., Tian Y., Cai K., Yang B., Niu J., Zou H., Liu S., Xu G., Fan X., Hu ✉ B., Loh X.J., Wang L., Chen X. A chemically mediated artificial neuron. Nat. Electron. 2022;5:586–595. [ Google Scholar ] 12. Hao H.T., Wang M.X., Cao Y.L., He J.T., Yang Y.Z., Zhao C., Yan L.P. Boron-doped engineering for carbon quantum dots-based memristors with controllable memristance stability. Small Methods. 2024;8 doi: 10.1002/smtd.202301454. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Zhang C., Chen M.H., Pan Y.L., Li Y., Wang K.B., Yuan J.W., Sun Y.Q., Zhang Q.C. Carbon nanodots memristor: an emerging candidate toward artificial biosynapse and human sensory perception system. Adv. Sci. 2023;10 doi: 10.1002/advs.202207229. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Sokolov A.S., Ali M., Riaz R., Abbas Y., Ko M.J., Choi C. Silver-adapted diffusive memristor based on organic nitrogen-doped graphene oxide quantum dots (N-GOQDs) for artificial biosynapse applications. Adv. Funct. Mater. 2019;29 [ Google Scholar ] 15. Ai L., Pei Y.F., Song Z.Q., Yong X., Song H.Q., Liu G.J., Nie M.J., Waterhouse G.I.N., Yan X.B., Lu S.Y. Ligand-triggered self-assembly of flexible carbon dot nanoribbons for optoelectronic memristor devices and neuromorphic computing. Adv. Sci. 2023;10 doi: 10.1002/advs.202207688. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Yang J.Q., Zhang F., Xiao H.M., Wang Z.P., Xie P., Feng Z.H., Wang J.J., Mao J.Y., Zhou Y., Han S.T. A perovskite memristor with large dynamic space for analog-encoded image recognition. ACS Nano. 2022;16:21324–21333. doi: 10.1021/acsnano.2c09569. [ DOI ] [ PubMed ] [ Google Scholar ] 17. Zhao X.N., Xu H.Y., Wang Z.Q., Lin Y., Liu Y.C. Memristors with organic-inorganic halide perovskites. InfoMat. 2019;1:183–210. [ Google Scholar ] 18. Heriyanto A.D.M., Cho Y., Okamoto N., Abe R., Pandey M., Benten H., Nakamura M. Influence of halogen elements in organic salts on n-type doping of CNT yarn for thermoelectric applications. RSC Adv. 2023;13:22226–22233. doi: 10.1039/d3ra03755j. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Kim S.J., Im I.H., Baek J.H., Choi S., Park S.H., Lee D.E., Kim J.Y., Kim S.Y., Park N.-G., Lee D., Yang J.J., Jang H.W. Linearly programmable two-dimensional halide perovskite memristor arrays for neuromorphic computing. Nat. Nanotechnol. 2025;20:83–92. doi: 10.1038/s41565-024-01790-3. [ DOI ] [ PubMed ] [ Google Scholar ] 20. Fang Y.T., Zhai S.B., Chu L., Zhong J.S. Advances in halide perovskite memristor from lead-based to lead-free materials. ACS Appl. Mater. Interfaces. 2021;13:17141–17157. doi: 10.1021/acsami.1c03433. [ DOI ] [ PubMed ] [ Google Scholar ] 21. Chen F., Zhang S., Hu L., Fan J., Lin C.-H., Guan P., Zhou Y., Wan T., Peng S., Wang C.-H., Wu L., Furlong T.M., Valanoor N., Chu D. Bio-inspired artificial perceptual devices for neuromorphic computing and gesture recognition. Adv. Funct. Mater. 2023;33 [ Google Scholar ] 22. Chen F., Zhou Y., Zhu Y., Zhu R., Guan P., Fan J., Zhou L., Valanoor N., von Wegner F., Saribatur E., Birznieks I., Wan T., Chu D. Recent progress in artificial synaptic devices: materials, processing and applications. J. Mater. Chem. C. 2021;9:8372–8394. [ Google Scholar ] 23. Chen F., Feng Z., Zhang S., Min J., Hu L., Huang T., Mei T., Kim J., Dong B., Ma X., Yin T., Wang J., Guan P., Wan T., Lin C.-H., Li Z., Chu D. Enhancing synaptic plasticity in strontium titanate-based sensory processing devices: a study on oxygen vacancy modulation and performance in artificial neural networks. Adv. Intell. Discov. 2025;1 [ Google Scholar ] 24. Chen F., Zhang S., Guan P., Xu Y., Wan T., Lin C.-H., Li M., Wang C., Chu D. High-performance flexible graphene oxide-based moisture-enabled nanogenerator via multilayer heterojunction engineering and power management system. Small. 2024;20 doi: 10.1002/smll.202304572. [ DOI ] [ PubMed ] [ Google Scholar ] 25. Wang Z.F., Yuan F.L., Li X.H., Li Y.C., Zhong H.Z., Fan L.Z., Yang S.H. 53% efficient red emissive carbon quantum dots for high color rendering and stable warm white-light emitting diodes. Adv. Mater. 2017;29 doi: 10.1002/adma.201702910. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Qu S.N., Zhou D., Li D., Ji W.Y., Jing P.T., Han D., Liu L., Zeng H.B., Shen D.Z. Toward efficient orange emissive carbon nanodots through conjugated sp2-domain controlling and surface charges engineering. Adv. Mater. 2016;28:3516–3521. doi: 10.1002/adma.201504891. [ DOI ] [ PubMed ] [ Google Scholar ] 27. Yu H.L., Zhou P.K., Chen X. Intramolecular hydrogen bonding interactions induced enhancement in resistive switching memory performance for covalent organic framework-based memristors. Adv. Funct. Mater. 2023;33 [ Google Scholar ] 28. Li T.Z., Shi W., Mao Q.X., Chen X.W. Regulating the photoluminescence of carbon dots via a green fluorine-doping-derived surface-statecontrolling strategy. J. Mater. Chem. C. 2021;9:17357–17364. [ Google Scholar ] 29. Nayak S., Aggarwal M., Das P. Bromine-doped carbon dot: concentration-dependent multicolor emission, nanozyme activity, and visible-light-induced photodynamic bacterial inactivation. Carbon Lett. 2025;35:593–605. [ Google Scholar ] 30. Sun X., Liu H.L., Yang L.L., Wang X.Y., Yang W.Q., Wei M.B., Liu X.Y., Cao J., Yang J.H., Xing S.G. Tailoring blue-green double emissions in carbon quantum dots via co-doping engineering by competition mechanism between chlorine-related states and conjugated π-domains. Nanomaterials. 2018;8:635. doi: 10.3390/nano8090635. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Annamalai K., Annamalai A., Ravichandran R., Jeevarathinam A., Annamalai P., Valdes H., Elumalai S. Simple devising of N-doped carbon dots (N-CDs) as a low-cost probe for selective environmental toxin detection and security applications. New J. Chem. 2024;48:216–227. [ Google Scholar ] 32. Liu Y.F., Al-salihi M., Guo Y., Ziniuk R., Cai S.T., Wang L.W., Li Y., Yang Z.G., Peng D.F., Xi K., An Z.F., Jia X.D., Liu L.W., Yan W., Qu J.L. Halogen-doped phosphorescent carbon dots for grayscale patterning. Light Sci. Appl. 2022;11:163. doi: 10.1038/s41377-022-00856-y. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Gong L., Sun J., Zheng P., Liu Y.S., Yang G.C. Yellow fluorescent nitrogen and bromine co-doped graphene quantum dots for bioimaging. ACS Appl. Nano Mater. 2021;4:8564–8571. [ Google Scholar ] 34. Zhao Y., Zhu X.X., Liu L., Duan Z.Q., Liu Y.P., Zhang W.Y., Cui J.J., Rong Y.F., Dong C. One-step synthesis of nitrogen/fluorine co-doped carbon dots for use in ferric ions and ascorbic acid detection. Nanomaterials. 2022;12:2377. doi: 10.3390/nano12142377. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. Lin H.K., Hong S.Z. Effects of Mg content and annealing treatment on optical and electrical properties of CuMg and ITO/CuMg metallic glass films. J. Alloys Compd. 2018;731:248–252. [ Google Scholar ] 36. Lu T., Chen F.W. Multiwfn: a multifunctional wavefunction analyzer. J. Comput. Chem. 2012;33:580–592. doi: 10.1002/jcc.22885. [ DOI ] [ PubMed ] [ Google Scholar ] 37. Zhang J., Lu T. Efficient evaluation of electrostatic potential with computerized optimized code. Phys. Chem. Chem. Phys. 2021;23:20323–20328. doi: 10.1039/d1cp02805g. [ DOI ] [ PubMed ] [ Google Scholar ] 38. Yang J.Q., Wang R.P., Ren Y., Mao J.Y., Wang Z.P., Zhou Y., Han S.T. Neuromorphic engineering: from biological to spike-based hardware nervous systems. Adv. Mater. 2020;32 doi: 10.1002/adma.202003610. [ DOI ] [ PubMed ] [ Google Scholar ] 39. Zeng T., Yang Z., Liang J.B., Lin Y., Cheng Y.K., Hu X.C., Zhao X.N., Wang Z.Q., Xu H.Y., Liu Y.C. Flexible and transparent memristive synapse based on polyvinylpyrrolidone/N-doped carbon quantum dot nanocomposites for neuromorphic computing. Nanoscale Adv. 2021;3:2623–2631. doi: 10.1039/d1na00152c. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 40. Yu F., Zhu L.Q., Xiao H., Gao W.T., Guo Y.B. Restickable oxide neuromorphic transistors with spike-timing-dependent plasticity and pavlovian associative learning activities. Adv. Funct. Mater. 2018;28 [ Google Scholar ] 41. Gao Y.L., Hao H.T., Chen L., Yang Y.Z. Recent advances of carbon dot-based memristors: mechanisms, devices, and applications. Appl. Mater. Today. 2024;36 [ Google Scholar ] 42. Liu C.L., Goel P., Kaeser P.S. Spatial and temporal scales of dopamine transmission. Nat. Rev. Neurosci. 2021;22:345–358. doi: 10.1038/s41583-021-00455-7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Li J., Sun C.-L., An P.R., Liu X.Y., Dong R.H., Sun J.H., Zhang X.Y., Xie Y.B., Qin C.G., Zheng W.F., Zhang H.-L., Jiang X.Y. Construction of dopamine-releasing gold surfaces mimicking presynaptic membrane by on-chip electrochemistry. J. Am. Chem. Soc. 2019;141:8816–8824. doi: 10.1021/jacs.9b01003. [ DOI ] [ PubMed ] [ Google Scholar ] 44. Xu S.M., Liu Y., Lee H., Li W.D. Neural interfaces: bridging the brain to the world beyond healthcare. Explorations. 2024;4 doi: 10.1002/EXP.20230146. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Multimedia component 1 mmc1.docx (9.3MB, docx) Data Availability Statement Data will be made available on request. Articles from Materials Today Bio are provided here courtesy of Elsevier ACTIONS View on publisher site PDF (15.0 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top