Bioinspired neuromorphic memristor with logical operation for precision sperm detection in seminiferous tubules

Abstract

Clinically, even in patients diagnosed with non-obstructive azoospermia, spermatogenesis may be present in some seminiferous tubules, which gives the patient hope of having biological offspring of his own. However, there is still a blank for high-precision detection technologies to support accurate diagnosis and effective treatment. In this work, we successfully developed a minimally invasive fine needle detection memristive device that features a structure composed of Ag/C₄₀H₅₆-MnO₂/FTO by utilizes the organic-inorganic heterojunction as functional layer. The as-prepared memristor exhibits remarkable resistance switching (RS) performance, which serves as solid evidence of its efficacy in detecting sperm within the seminiferous tubules. Further, the RS characteristics of the memristor can be effectively modulated by adjusting the bias voltage amplitude, emulating the sensing dynamics of artificial neural synapses. By theoretical analyses, it is suggested that the RS mechanism involves synergistic effects of ion migration, Ohmic conduction, Fowler-Nordheim tunneling, and space-charge-limited current transport. Finally, a painless integrated detection device with micro needles was designed, which not only lays the foundation for novel reproductive health monitoring technologies, but also opens new avenues for bioelectronic integration in clinical diagnostics. Therefore, these findings highlight the potential of memristive systems to bridge neuromorphic engineering with precision medicine.

Graphical abstract

Highlights

• •It was developed a painless microneedle detection memristor with Ag/C₄₀H₅₆-MnO₂/FTO structure by utilizes the organic-inorganic heterojunction as functional layer.

• •The as-prepared memristor exhibits remarkable resistance switching •detecting sperm within the seminiferous tubules.

• •The RS characteristics of the memristor can be effectively modulated by adjusting the bias voltage amplitude, emulating the sensing dynamics of artificial neural synapses.

• •This work not only lays the foundation for novel reproductive health monitoring technologies, but also opens new avenues for bioelectronic integration in clinical diagnostics.

1. Introduction

Infertility represents a global health challenge, with male factors implicated in 30–50 % of cases among couples [1,2]. Notably, 10–15 % of male infertility patients present with azoospermia, clinically classified as obstructive azoospermia (OA) or non-obstructive azoospermia (NOA) subtypes [[3], [4], [5]]. The NOA phenotype, characterized by impaired spermatogenesis, accounts for 60 % of azoospermia cases and frequently eludes definitive etiological diagnosis despite multifactorial origins [[6], [7], [8]]. Clinically, therapeutic management of NOA primarily involves surgical sperm retrieval techniques, including testicular sperm extraction (TESE), aspiration (TESA), and microdissection TESE (micro-TESE), which combine with intracytoplasmic sperm injection (ICSI) for assisted reproduction [[9], [10], [11], [12], [13], [14], [15], [16], [17], [18]]. Although randomized controlled trials remain limited, comparative analyses indicate micro-TESE achieves superior sperm retrieval rates (SRRs) through microscopic identification of spermatogenically active tubules [19,20]. Nevertheless, this approach presents two critical limitations: ⅰ) SRR dependency on surgeon expertise in visual tubule selection. ⅱ) Iatrogenic testicular damage risk from excessive tissue resection [21]. These constraints underscore an urgent need for intraoperative detection technologies capable of real-time, specific sperm identification within seminiferous tubules. Such innovation would simultaneously enhance SRRs while minimizing surgical invasiveness and procedural duration through targeted tissue sampling.

The physical structure and signal transmission mechanism of the memristor closely resemble those of a neural synapse, demonstrating exceptional switching RS characteristics [[22], [23], [24], [25], [26], [27], [28], [29], [30]]. As an innovative dual-terminal nonlinear nanoscale electronic device, the memristor has found extensive applications in the field of biomedicine [28,[31], [32], [33], [34], [35], [36]]. Memristors have the capability to interact with biomolecules for the detection and analysis of disease markers, thereby demonstrating significant potential for advancement in the field of biosensing and diagnostic technologies [[37], [38], [39], [40], [41]]. For instance, it was developed a memristor with a structure of Ag/BaTiO₃/MnO₂/FTO, which exhibits exceptional stability. Following implantation in Sprague-Dawley rats, the device revealed three distinct phases of biological signals and retained both reversibility and relative stability after extraction from the host's body. This study provides new insights into the application of memristive devices in implantable biomedical detection and highlights their potential as sensors [42]. The application of memristors in ex vivo monitoring has broadened the scope of their sensing technologies. In addition, a blood-based biomemristor was developed and designed for the clinical detection of hyperglycemia and hyperlipidemia, thereby demonstrating the feasibility of this biomemristor for detecting these conditions and establishing a foundation for the advancement of memristors in the field of medical monitoring [43].

With the advancement of memristive-sensing capabilities, its coupling performance has also been enhanced in the field of tactile sensing. Zhang et al. utilized silk fibroin (SF) to develop an artificial intelligence injury detection and alarm system that integrates both sensing and monitoring functions with information transmission capabilities [44]. The memristor within this system exhibits high stability, low power consumption, and the ability to simulate synaptic plasticity, enabling it not only to detect pain but also to issue timely alerts when the carrier falls or remains immobile for extended periods. Furthermore, the integrated artificial pressure injury sensor can monitor pressure distribution across various parts of the carrier while providing direct feedback on the degree of long-term pressure experienced by the carrier. Xia et al. developed a W/VO₂/Pt-structured memristor that demonstrates all the characteristics of a nociceptor, including abnormal pain and relaxation phenomena in pain perception and hypersensitivity, while exhibiting high sensitivity and a broad sensing range [45]. Consequently, this device was employed to monitor various human movements and physiological signals. Anyway, the exploration of memristors in sensing applications is ongoing, offering researchers extensive opportunities for innovation in the biomedical field.

In this work, it is integrated the high sensitivity and stability of memristors in the sensing domain to fabricate a device with Ag/C₄₀H₅₆-MnO₂/FTO structure. This device employs organic and inorganic materials as dielectric layer for detecting the presence of sperm in seminiferous tubules. It was demonstrated that the reaction mechanism of C₄₀H₅₆ and MnO₂ as dielectric layers encompasses ionic reactions, Ohmic conduction, Shockley effect, and Fowler-Nordheim tunneling, enabling the detection of sperm presence. Consequently, we developed an accurately micro-needle memristor detector that has the potential to revolutionize the surgical sperm retrieval method for NOA patients, significantly enhancing the success rate of sperm retrieval and presenting promising commercial application prospects. The memristor exhibits rapid response capabilities and high integration potential, enabling precise and non-invasive monitoring of changes in reproductive system parameters. Therefore, this innovative device effectively addresses the issues related to biocompatibility and enables painless surgical procedures, which enables precise detection of sperm presence in the seminiferous tubules of the male reproductive system.

2. Experimental and methods

2.1. Materials and chemical reagents

Carrots harvested from Xianyang, Shaanxi Province were thoroughly washed with ultrapure water five times and subsequently dried in an oven at 55 °C for 1 h. Following this, the carrots were ground into powder using mechanical grinders and a ball mill (PULVERIZER 80, Nanjing Chishun Technology Development Co., Ltd.). The final particle size achieved through mechanical grinding was required to be less than 3 mm, while the ball mill parameters included a rotational speed of 400 rpm, a grinding duration of 50 min, and a material volume set at one-twentieth of the ball mill capacity, ultimately resulting in ultrafine C₄₀H₅₆ powder. The reagents utilized in this study comprised N-Methyl Pyrrolidone, polyvinylidene fluoride (PVDF), PEDOT:PSS, ethanol, FTO glass, and MnO₂ target materials were procured from Guoyao Group Chemical Reagents Co., Ltd.

2.2. Fabrication of device

To fabricate a memristor with the structure of Ag/C₄₀H₅₆-MnO₂/FTO, a 0.15 mm thick and 25 mm × 25 mm FTO glass substrate was initially cleaned using 75 % ethanol followed by ultrapure water in an ultrasonic cleaning apparatus for 20 min. Concurrently, 0.03 g of pure C₄₀H₅₆ powder and 0.02 g of PVDF powder were accurately weighed and subsequently transferred to a glass vial using a pipette. 5 ml of N-Methyl Pyrrolidone solvent were then added to the vial along with an appropriately sized stirring bar. The resulting mixture was stirred at a speed of 1000 rpm/min on a magnetic stirrer to achieve a homogeneous solution. After stirring for 2 h, the mixed solution was filtered using a filtration device to obtain the precursor solution for subsequent use. Following the completion of the preparatory steps, MnO₂ was deposited onto the FTO glass substrate via magnetron sputtering, with a sputtering pressure of 0.8 Pa, power set at 65 W, and a duration of 1 h. The FTO glass was then removed after MnO₂ deposition, and the precursor solution was applied dropwise onto its surface. After allowing sufficient time for the solution to diffuse uniformly, it was air-dried to form an intermediate layer that combines organic and inorganic materials. Finally, a 25 mm × 25 mm metal mask was affixed to the FTO glass substrate coated with functional layer. The magnetron sputtering parameters were then set to a pressure of 0.8 Pa, power of 65 W, and a deposition time of 12 min for the deposition of Ag electrode onto the surface of the functional layer. Thus, the Ag/C₄₀H₅₆-MnO₂/FTO memristor was successfully fabricated.

2.3. Characterization of materials

The cross-sectional morphology of the device was characterized using a ZEISS Sigma 300 scanning electron microscope (SEM), in conjunction with point and mapping techniques of energy-dispersive X-ray spectroscopy (EDS) for elemental analysis, as well as an X-ray photoelectron spectrometer (XPS) model K-Alpha from Thermo Scientific for comparative testing. Subsequently, the electrochemical characteristics of the memristor were evaluated using a Keysight 2901B device.

2.4. Tissue and samples

The samples utilized in this study include seminiferous tubules from patients diagnosed with both OA and NOA, as well as normal sperm samples. All of these samples were obtained from the Reproductive Medicine Department of the First Affiliated Hospital of Xi'an Jiaotong University. The Institutional Review Board of the First Affiliated Hospital of Xi'an Jiaotong University approved this study (approval number: XJTU1AF2025LSYY-596). All participants provided informed written consent.

3. Results and discussion

In this work, a memristor with the structure of Ag/C₄₀H₅₆-MnO₂/FTO was fabricated, as illustrated in Fig. 1a. Initially, fresh carrots were mechanically ground and extracted into an extremely fine C₄₀H₅₆ powder. Subsequently, this C₄₀H₅₆ powder was combined with polyvinylidene fluoride (PVDF) and the solvent N-methylpyrrolidone (NMP), followed by stirring to obtain the precursor solution. MnO₂ film was then deposited onto an FTO glass substrate using magnetron sputtering, after which an appropriate amount of the precursor solution was drop-cast onto the MnO₂ layer to form the intermediate layer. Finally, Ag was deposited on the intermediate layer via magnetron sputtering once more to successfully construct the sandwich structure of Ag/C₄₀H₅₆-MnO₂/FTO memristor. Fig. 1b illustrates a schematic representation of the Ag/C₄₀H₅₆-MnO₂/FTO memristor, while Fig. 1c presents a SEM image depicting the cross-section of the device. The SEM image clearly reveals that the thickness of the film formed by the integration of C₄₀H₅₆ and PVDF reaches 5 μm, whereas the MnO₂ layer deposited on FTO glass via magnetron sputtering has a thickness of 10 nm. The top and bottom electrodes of the device are connected to probes, with the FTO glass grounded. Following the application of a bias voltage to the top electrode, electrical measurements can be conducted.

Fig. 1.

Fabrication and morphological characterization of the Ag/C₄₀H₅₆-MnO₂/FTO memristor (a) Schematic representation of the memristor fabrication process. (b) Cross-sectional schematic illustration of the device. (c) Scanning electron microscopy (SEM) cross-sectional image of the device. (d, e) X-ray photoelectron spectroscopy (XPS) results for the memristor. (f) Energy-dispersive X-ray spectroscopy (EDS) results for the device. (g–j) The distribution of elements (N, O, C, and Mn) in the EDS analysis of the device.

Next, the chemical composition and elemental distribution of the functional layer were systematically characterized and analyzed. From the XPS results, the oxygen element is observed to exhibit C-O bonding characteristics at 532 eV, whereas at 530 eV, it demonstrates features indicative of metal-oxygen bonding. This phenomenon can primarily be ascribed to the interaction between Mn⁴⁺ ions and oxygen atoms within MnO₂, as illustrated in Fig. 1d. As shown in Fig. 1e, the carbon element is observed to exist in the forms of C=O, C-O, and C-C at binding energies of 288.68 eV, 286 eV, and 284.8 eV, respectively. As presented in Fig. 1f, the energy-dispersive spectroscopy (EDS) analysis reveals that the mass percentages of C, N, O, and Mn are 28.81 %, 0.99 %, 69.66 %, and 0.54 %, respectively. The film exhibits the highest proportions of C and O elements. Furthermore, EDS spectrum analysis (as shown in Fig. 1g–i) reveals that within a scanning range of 10 μm, the intermediate layer primarily consists of C, N, O, and Mn elements, with their distributions being relatively homogeneous.

Following the preparation of the Ag/C₄₀H₅₆-MnO₂/FTO device with carotene (C₄₀H₅₆) and MnO₂ as functional layer materials, it is performed the electrical tests, which exhibits favorable RS characteristics. As depicted in Fig. 2a, the functional layer of the device consists of MnO₂ and C₄₀H₅₆, with its structural composition and crystal structure clearly discernible. Notably, the crystal structure of MnO₂ exhibits an octahedral morphology and is interconnected with adjacent octahedra through shared vertices. The molecular formula of carotene is C₄₀H₅₆, and it exhibits multiple cis-trans isomeric forms. As depicted in Fig. 2a, the specific carotene presented here is β-carotene, which represents the most prevalent isomer form of carotene. Initially, to ascertain the optimal voltage window for RS performance and to evaluate the impact of various bias voltages on its electrical properties, we examined the I-V characteristic curves of the top electrode (Ag) under different bias conditions, as illustrated in Fig. S1 (see Supporting Information). After 20 cycles, the RS behavior of the device gradually diminishes and ultimately vanishes due to insufficient voltage that fails to provide adequate energy for ion mobility (Fig. S1b).

Fig. 2.

Electrical characteristics of the Ag/C₄₀H₅₆-MnO₂/FTO memristor. (a) The composition of the device structure and the crystal structure of the functional layer. (b) The typical I-V curve of the device over 100 scanning cycles with a voltage scanning range of 0 V → 2.0 V → 0 V → −2.0 V → 0 V). (c) The typical I-V curve of the device measured on a semi-logarithmic scale after the completion of 100 scanning cycles. (d) The stability test results of the memristor following 100 scanning cycles at 0.41 V. (e) The Gaussian frequency distribution characteristics of the device in both HRS and LRS. (f) Comparative analysis of the average values of the device in both HRS and LRS. (g) Variation curves of voltage and current signals within a defined time interval.

Subsequently, when a biasing sequence of (0 V → 1.5 V → 0 V → −1.5 V → 0 V) is applied to the top electrode, the device demonstrates favorable RS characteristics, with relatively stable high and low levels. When the applied bias voltages follow the sequences of 0 V → 2.0 V → 0 V → −2.0 V → 0 V, 0 V → 2.5 V → 0 V → −2.5 V → 0 V, and 0 V → 3.0 V → 0 V → −3.0 V → 0 V, the RS characteristics are outstanding in both positive and negative voltage regions, as illustrated in Fig. S1d-f. Moreover, a uniform distribution of high and low levels is observed at 0.41 V under the aforementioned biasing conditions. Analysis of the graph reveals that when the bias voltages are set to 0 V → 2.0 V → 0 V → −2.0 V → 0 V and 0 V → 1.0 V → 0 V → −1.0 V → 0 V, both high impedance and low impedance values are relatively elevated. However, under these conditions, the values of high and low impedance exhibit a continuous decline until they nearly converge, indicating the device has poor instability. In contrast, when the bias voltage is maintained at 0 V → 2.0 V → 0 V→ −2.0 V → 0 V, the stability is comparatively favorable. Consequently, we have opted to conduct subsequent experiments under a bias condition of ±2.0 V.

To further investigate the electrical properties of the device within the voltage range of ±2.0 V, Fig. 2b and c presents the I-V curves and their characteristics on a semi-logarithmic scale as the bias voltage gradually changes from 0 V → 2.0 V → 0 V → −2.0 V → 0 V. These results indicate that the Ag/C₄₀H₅₆-MnO₂/FTO structure possesses excellent electrical performance with good RS characteristics. Upon completing 100 cycles of I-V characteristic curve scanning, the experimental results demonstrate that the I-V curve maintains exceptional stability. Furthermore, as illustrated in Fig. 2c, the I-V characteristic curve displayed on a semi-logarithmic scale provided additional validation for this observation. To further examine the repeatability of the RS behavior, we conducted an analysis of the I-V curves scanned over 100 cycles at 0.41 V and extracted the corresponding data for both the high-resistance state (HRS) and low-resistance state (LRS). As the number of cycles increased, the LRS demonstrated excellent stability, whereas the HRS exhibited a decreasing trend during the first 50 cycles and subsequently began to gradually increase after the 50th cycle. The calculated ratio of HRS to LRS was found to be 1.77, with its distribution following a Gaussian distribution (as depicted in Fig. 2d and e). Overall, the average values of the HRS and LRS are concentrated at 147 Ω and 84 Ω, respectively (Fig. 2f). Fig. 2g illustrates the temporal evolution trends of voltage and current. Analysis of the I-V curves reveals that an Ohmic conduction mechanism dominates within the voltage range from 0 V to 1.0 V. Additionally, in the positive voltage region, the phase difference between the current and voltage peaks attains its maximum value. Considering that the Ag/C₄₀H₅₆-MnO₂/FTO device demonstrates superior resistive switching characteristics and stability, it is scientifically significant and practically valuable to conduct an in-depth investigation into the operational mechanism of memristor.

The working mechanism of the Ag/C₄₀H₅₆-MnO₂/FTO memristor primarily involves ionic reactions and migration processes, including Ohmic conduction, Shockley emission, and Fowler-Nordheim emission. Upon applying a bias voltage to the Ag/C₄₀H₅₆-MnO₂/FTO memristor, the ion migration process from the top electrode to the bottom electrode resembles the ionic reaction occurring. To facilitate a clearer analysis of the I-V curve mechanism, we conducted a voltage division simulation of the first I-V loop curve, with the segmented switching characteristic curve presented in Fig. S2d. Fig. 3a illustrates the distribution of ions related in the functional layer and oxygen vacancies in the initial state. In the absence of applied voltage, the oxygen ions released by MnO₂ exist in a dispersed and free state alongside the oxygen vacancies within the functional layer.

Fig. 3.

(A) Schematic representation of the conductive filaments formation process of Ag/C₄₀H₅₆-MnO₂/FTO device. (b) Schematic illustration of the conductive filaments formation process under forward bias voltage ranging from 0 V to 2.0 V at LRS. (c) Schematic illustration of the conductive filaments formation process under forward bias voltage ranging from 2.0 V to 0 V at HRS. (d) Ohmic conduction from 0 V to 1.0 V. (e) Shockley emission from 1.0 V to 2.0 V. (f) Direct tunneling from 2.0 V to −1.0 V. (g) Fowler-Nordheim emission ranging from −1.76 V to −2.0 V. (h) Schottky emission ranging from −2.0 V to −1.0 V. (i) Ohmic conduction ranging from −1.0 V to 0 V.

As illustrated in Fig. 3b, when a positive bias voltage (0 V → 2.0 V) is applied, oxidation occurs at the Ag electrode, resulting in the generation of Ag⁺ ions. Under the influence of the electric field, oxygen vacancies and oxygen ions migrate from the bottom electrode to the Ag electrode, while Ag⁺ ions migrate toward the bottom electrode. With Ag⁺ ions continuously accumulating at the interface between the functional layer and the bottom electrode, an Ag pillar is formed. When oxygen vacancies and oxygen ions reach the upper electrode, they collectively form a conductive filament with the Ag pillar. At this point, the device transitions into LRS. As the bias voltage decreases from 2.0 V to 0 V (Fig. 3c), the potential energy for ion migration driven by the electric field progressively diminishes, leading to the migration of oxygen vacancies and oxygen ions towards the bottom electrode. Concurrently, Ag⁺ ions undergo a reduction reaction and returns to the top electrode, where it is reduced back to Ag. At this juncture, the device transitions into HRS. When negative bias voltages (0 V → 2.0 V → 0 V) are applied to the top and bottom electrodes from the initial state, as illustrated in Fig. S2a-c, the process resembles that of applying a positive voltage. Initially, an Ag conductive filament is formed at the LRS, which subsequently transitions back to the HRS through a reduction process. Throughout this process, Ag undergoes an oxidation-reduction reaction, with the specific details outlined as follows [39,40]:

$$\text{Ag}\to {\text{Ag}}^{+}+{\mathrm{e}}^{-}$$ (1)

$${\text{Ag}}^{+}+{\mathrm{e}}^{-}\to \text{Ag}$$ (2)

Upon applying a bias voltage to the top and bottom electrodes of the device, the ion and electron migration process within the functional layer transitions from an analog signal to a digital signal. Further, we conducted the fitting for I-V curve to analyze the working mechanism. Fig. 3d presents the fitting results of Ln(I) versus Ln(V) over the bias voltage range of 0 V → 1.0 V, revealing a slope of approximately 1, which indicates that this process adheres to Ohmic conductive characteristics. When the voltage is increased to the range of 1.0 V → 2.0 V, electrons are able to surmount the potential barrier formed at the interface, resulting in the cessation of Ohmic conduction. At this juncture, the slope rises to ∼2.59, indicating that the electron conduction mechanism transitions to Schottky emission (Fig. 3e). When the bias voltage is decreased from 2.0 V to 1.0 V, the interface barrier is effectively eliminated, allowing electrons to tunnel directly, which results in an increase of the slope to ∼13.64, as illustrated in Fig. 3f.

Subsequently, as the negative voltage continued to increase within the range of −1.76 V to −2.0 V, Ag⁺ ions began forming conductive filaments, facilitating the transition from the HRS to the LRS. During this process, direct tunneling was insufficient to overcome the interface barrier. Thus, the Fowler-Nordheim emission became the predominant conduction mechanism, exhibiting a slope of −0.31 (Fig. 3g). As the voltage continues to decrease into the negative range (−2.0 V to −1.0 V), the conductive filaments have been broken, necessitating that electrons overcome the interface barrier in a reverse direction to revert to the HRS. During this phase, with a slope of ∼2.9, the predominant migration mechanism observed is Schottky emission. Ultimately, when the bias voltage ranges from −1.01 V to 0 V, electrons no longer need to overcome the barrier for movement, thereby returning to an ohmic conduction state. At this point, the slope is approximately 1, as illustrated in Fig. 3i. The difference in work functions induces significant band bending on the semiconductor side, forming a potential barrier that gives rise to various effects. Under zero bias, the band structure exhibits distinct electrical behaviors as a result of such work function differences. We have simulated the work functions and the zero-bias band structure, as illustrated in Fig. S3a, along with the band variation diagrams presented in Fig. S3(b-g). These simulated results correspond one-to-one to the electrical effects induced by the work functions.

This study employed the Ag/C₄₀H₅₆-MnO₂/FTO memristor to detect and differentiate the presence of sperm in seminiferous tubules. Initially, we evaluated the electrical characteristics of the memristor by analyzing normal seminiferous tubules (sourced from patients with obstructive azoospermia) and azoospermia seminiferous tubules (sourced from patients with the pathological diagnosis of Sertoli cell-only syndrome). As depicted in Fig. 4a, to assess the influence of physiological saline on the detection capability of the memristor, an Ag/C₄₀H₅₆-MnO₂/FTO memristor was employed to analyze physiological saline. Next, we examined the memristive electrical properties of sperm to explore their relevance in distinguishing between seminiferous tubules that contain sperm and those that do not throughout the detection process. Additionally, given the substantial presence of physiological saline within seminiferous tubules, we also evaluated physiological saline to ascertain its impact on the memristor's capability to determine whether sperm are present.

Fig. 4.

(A) Schematic diagram of the Ag/C40H56-MnO₂/FTO memristor for sample detection; (b) Current-voltage (I–V) electrical characteristic curve obtained from physiological saline assessment based on the Ag/C40H56-MnO₂/FTO memristor; (c) Current-voltage (I–V) curve of the physiological saline sample, presented on a logarithmic scale; (d) Current-voltage (I–V) electrical characteristic curve for detecting normal seminiferous tubules; (e) Current-voltage (I–V) curve for detecting normal seminiferous tubules, presented on a logarithmic scale; (f) Frequency distribution of high resistance state (HRS) and low resistance state (LRS) of the memristor at 0.41 V after 100 cycles when detecting normal seminiferous tubules; (g) Current-voltage (I–V) electrical characteristic curve for detecting seminiferous tubules from patients with non-obstructive azoospermia (NOA); (h) Current-voltage (I–V) curve for detecting non-obstructive azoospermia (NOA) samples, presented on a logarithmic scale; (i) Frequency distribution of HRS and LRS of the memristor at 0.41 V after 100 cycles when detecting seminiferous tubules from patients with non-obstructive azoospermia (NOA); (j) Current-voltage (I–V) electrical characteristic curve for detecting normal sperm; (k) Current-voltage (I–V) curve for detecting normal sperm, presented on a logarithmic scale; (l) Frequency distribution of HRS and LRS of the memristor at 0.41 V after 100 cycles in normal sperm detection.

Given the favorable memristive characteristics exhibited by the Ag/C₄₀H₅₆-MnO₂/FTO device at a bias voltage of ±2.0 V, we initially selected this voltage for testing azoospermia seminiferous tubules, normal seminiferous tubules, normal sperm, and physiological saline. The electrophysiological characteristics of normal seminiferous tubules tested using the Ag/C₄₀H₅₆-MnO₂/FTO device are presented in Fig. 4. Through the analysis of the I-V curves at ±2.0 V bias voltage and their semi-logarithmic representations, distinct RS characteristics can be identified (Fig. 4b and c). As illustrated in Fig. S4, which displays the distribution of HRS and LRS, the HRS is predominantly centered around approximately 98 Ω, whereas the LRS is concentrated around approximately 55 Ω. The testing and analytical results regarding the electrical properties of normal seminiferous tubules using the Ag/C₄₀H₅₆-MnO₂/FTO device is depicted in Fig. 4d–f. Fig. 4d illustrates the I-V curve of normal seminiferous tubules under memristor testing, indicating the presence of RS characteristics. Additionally, the semi-logarithmic I-V curve shown in Fig. 4e corroborates this characteristic. The HRS is primarily concentrated around 280 Ω, while the LRS predominantly occurs around 62 Ω (Fig. 4f).

To further differentiate the RS characteristics of normal seminiferous tubules from those of azoospermia seminiferous tubules under RRAM detection, we established a bias voltage of ±2.0 V and employed the Ag/C₄₀H₅₆-MnO₂/FTO device to detect and analyze the seminiferous tubules from patients with azoospermia. The seminiferous tubules of NOA patients obtained here do not contain sperm. The results of testing show that the seminiferous tubules of NOA patients exhibit certain RS characteristics, including the I-V curve in Fig. 4g and the I-V curve on the semi-log scale in Fig. 4h, both of which have obvious changes. The stability of detection results serves as a critical indicator for differentiation. As illustrated in Fig. 4i, the distribution of HRS and LRS reveals that LRS are predominantly concentrated around 265 Ω, while HRS are primarily centered at approximately 370 Ω. These values are significantly elevated compared to the average resistance levels observed in normal seminiferous tubules. Thus, it is provided an important basis for distinguishing azoospermia seminiferous tubules from normal seminiferous tubules using the Ag/C₄₀H₅₆-MnO₂/FTO memristor.

Clinically, some patients with NOA have focal spermatogenesis within their testes, which is precisely what surgeons aim to retrieve during surgery. Consequently, we employed the Ag/C₄₀H₅₆-MnO₂/FTO memristor to assess the presence of sperm within the seminiferous tubules. Following the application of a bias voltage of ±2.0 V, sperm continued to exhibit RS characteristics (Fig. 4j and k). As illustrated in Fig. 4l, HRS are predominantly concentrated around 320 Ω, while LRS are primarily centered at approximately 83 Ω. This distribution of HRS and LRS suggests that the impact of RS characteristics on NOA seminiferous tubules is not significantly different from that observed in normal seminiferous tubules. The results indicate that the Ag/C₄₀H₅₆-MnO₂/FTO memristor exhibits minimal impact on the memristive characteristics of seminiferous tubules when exposed to physiological saline. Through analysis, we can utilize the Ag/C₄₀H₅₆-MnO₂/FTO memristor to distinguish whether sperm exists within the seminiferous tubules.

To further differentiate the seminiferous tubules of patients with NOA from those of normal individuals, we applied bias voltages of ±1.0 V, ±1.5 V, ±2.0 V, ±2.5 V, and ±3.0 V between the top and bottom electrodes, as illustrated in Fig. 5a–d. The results indicate that NOA exhibits a significant response to varying bias voltages, characterized by a nonlinear relationship between the maximum and minimum currents across these voltages. In contrast, the maximum and minimum currents for normal seminiferous tubules, sperm samples, and physiological saline demonstrated a linear distribution. We conducted a detailed analysis of the electrical characteristics of the device following the application of a bias voltage of ±2.0 V, focusing on the discriminative capabilities of the memristor in detecting various samples. The typical I-V curves and logarithmic I-V curves indicate that the device demonstrates pronounced logarithmic characteristics when assessing different samples (Fig. 5e and f). By comparing the distributions of HRS and LRS in the seminiferous tubules of normal saline and those from patients with NOA, as illustrated in Fig. 5g and h, we observed a significant difference between the two groups. A detailed analysis indicates that the normal seminiferous tubules demonstrated remarkable stability after 100 cycles of scanning, as illustrated in Fig. 5g. We conducted a 400-cycle scan of the normal seminiferous tubules and recorded HRS and LRS at 0.41V, as shown in Fig. S5a. As the number of scanning cycles increased, the HRS exhibited regular fluctuations between 200 Ω and 320 Ω during the first 250 cycles but stabilized thereafter. Conversely, the LRS remain consistently stable throughout the scans.

Fig. 5.

Memristors distinguish samples by electrical characteristics. (a–d) Current-voltage (I–V) characteristic curves of seminiferous tubules from patients with non-obstructive azoospermia, seminiferous tubules from normal males, sperm from normal males, and saline solution under different bias voltages; (e) I-V curves of seminiferous tubules from patients with non-obstructive azoospermia, seminiferous tubules from normal males, sperm from normal males, and saline solution, compared via Ag/C40H56-MnO₂/FTO under a bias voltage of ±2 V. (f) I-V characteristic curves of seminiferous tubules from patients with non-obstructive azoospermia, seminiferous tubules from normal males, sperm from normal males, and saline solution, compared via Ag/C40H56-MnO₂/FTO under a bias voltage of ±2 V, presented on a semilogarithmic scale. (g) Comparison of the high resistance state (HRS) and low resistance state (LRS) retention characteristics of seminiferous tubules from patients with non-obstructive azoospermia, seminiferous tubules from normal males, and memristors after 100 scans at 0.41 V. (h) Comparison of HRS and LRS retention characteristics for saline and sperm alongside results from the memristor after 100 scans at 0.41 V. (i) Comparative analysis of the high resistance state/low resistance state (HRS/LRS) ratio between seminiferous tubules from patients with non-obstructive azoospermia and seminiferous tubules from normal male based on memristor detection.

The frequency distribution for HRS is approximately 280 Ω, while that for LRS is around 62 Ω (Fig. S5b). The calculated HRS/LRS ratio after completing 400 cycles is found to be 4.3. Under identical detection conditions, the RS characteristics of NOA samples in the Ag/C₄₀H₅₆-MnO₂/FTO memristor exhibits significantly lower stability compared to those of normal seminiferous tubules. Following 100 cycles of scanning at 0.41 V, both HRS and LRS manifested a scattered state. As the number of cycles increased, HRS gradually decreased while LRS displayed relatively minor fluctuations, resulting in much lower stability than that observed for normal seminiferous tubules after 400 cycles. In contrast, testing sperm and normal saline revealed that the RS characteristics of sperm remained relatively stable over 100 cycles, with HRS is decreased as cycle numbers increased and LRS exhibiting minimal fluctuation.

To facilitate the differentiation of detection results from the Ag/C₄₀H₅₆-MnO₂/FTO memristor, we assess the presence of sperm in the seminiferous tubules of patients with NOA. Fig. 5i presents the statistical and calculated HRS/LRS values for each sample after 100 scans, indicating that the HRS/LRS value for the Ag/C₄₀H₅₆-MnO₂/FTO memristor is 1.77. In contrast, normal seminiferous tubules exhibit an HRS/LRS value of 4.41 following detection by the memristor, while those from patients with NOA show a reduced value of 1.42 compared to pre-detection levels. Furthermore, the testing result reveals that sperm and normal saline yield HRS/LRS values of 4.2 and 3.11, respectively. However, this result does not impede distinguishing between normal and abnormal conditions. Consequently, we can assess the presence of sperm in the seminiferous tubules by measuring values within the range of 1.42–4.41.

In observations of seminiferous tubules from patients with non-obstructive azoospermia (NOA), a striking distinction is evident: sperm-free tubules display elevated reactive oxygen species (ROS) levels, while those containing sperm exhibit low ROS levels, with this difference being highly pronounced [[46], [47], [48]]. The Ag/C40H56-MnO₂/FTO memristor enables the identification of normal sperm presence in seminiferous tubules by leveraging the reaction between carotene within its dielectric layer and reactive oxygen species (ROS) [49]. The reaction extent between the dielectric layer and reactive oxygen species (ROS) dictates the resistive switching properties of the memristor. Carotene, via electron or hydrogen transfer, reacts with ROS of varying concentrations to interrupt the oxidative chain reaction, thereby inducing specific resistive switching behaviors in the memristor and facilitating the differentiation between normal and abnormal seminiferous tubules.

It is well known that the application of memristors in artificial vision and biomedical fields has attracted increasing interest [50]. Based on the aforementioned data analysis, the Ag/C₄₀H₅₆-MnO₂/FTO memristor can serve as an effective tool for determining the presence or absence of sperm in seminiferous tubules. To address the challenge faced by clinicians in treating patients with non-obstructive azoospermia, wherein the precise and rapid identification of normal seminiferous tubules within the testis is required for sperm retrieval through accurately detection, we have developed a microneedle-based accurately memristor device capable of detecting the presence of sperm in seminiferous tubules. The detailed structure of the device is presented in Fig. 6.

Fig. 6.

A microneedle noninvasive memristor detector for detecting the presence of sperm in seminiferous tubules.

Firstly, the circuit is designed based on the application scenario. The specific operational procedure is as follows: under the guidance of clinical experience, insert the needle into the seminiferous tubules of the testis. Using the pusher mechanism, bring the memristor located on the needle into contact with the sperm within the seminiferous tubules to enable real-time detection of the presence of sperm. Subsequently, transmit the acquired detection signal to the input terminal for intelligent medical diagnosis. Next, the fundamental concept of the circuit design is to enable physicians to more intuitively ascertain the precise location of sperm within the testis. This is achieved by analyzing the responses of three distinct samples (seminiferous tubules, physiological saline, and sperm) within the circuit, in conjunction with the status indicators of the blue and green lights in the digital circuit. Fig. 6a presents the logic display schematic of the circuit for scene application. In this diagram, the logic output signals are indicated by colored LEDs in the digital logic circuit.

Specifically, the LEDs arranged from left to right correspond to the detection results of three distinct samples: seminiferous tubules, physiological saline, and sperm. A gray LED signifies either the absence of a detected target or the presence of only the seminiferous tubule signal. The blue and green LEDs serve to identify the presence of specific targets, with the green LED indicating the detection of physiological saline and the blue LED representing the detection of sperm. When the device detects both the seminiferous tubules and sperm, the blue LED illuminates, indicating the presence of sperm at the micro-needle detection site. It is important to note that, as the green LED is exclusively used to indicate the saline signal, in cases where the detection target simultaneously comprises the seminiferous tubules, saline, and sperm, both the green and blue LEDs will illuminate. In practical applications, this can also serve to confirm the existence of sperm. Consequently, if the indicator LED displays gray or only the green LED is illuminated, it signifies that no sperm can be detected.

Based on the aforementioned application scenarios and the logical relationship of the output signal, Fig. 6b illustrates the equivalent symbol of the memristor within the circuit. When the current flowing from right to left is designated as the positive direction, the memristor exhibits the LRS in the positive direction and the HRS in the reverse direction. The characteristic of memristors can be utilized in digital logic circuits. By leveraging logical operations within digital logic circuits (such as AND, OR, and NOT), appropriate detection circuits may be developed. This work employs the switching voltage as the input signal (V_H = 2.0 V, V_L = 0.5 V) and the state of the lamp as the output signal. Through the interconnection of the switching voltage, memristor, and display module, a digital circuit capable of providing real-time feedback on the detection results is established. The truth table for the logical operations (AND, OR, NOT) presented in Fig. 6b illustrates the standard logical relationships between binary states (“0” and “1”).

To detect whether the seminiferous tubules contain sperm, the current study has developed the digital logic circuit depicted in Fig. 6c, which is constructed using AND, OR, and NOT logic gates. Specifically, V_A, V_B, and V_C correspond to the detection input signals of seminiferous tubules, sperm, and physiological saline, respectively. The output signal is indicated by the state of the lamp and serves to determine whether sperm are present at the detection site. The logic relationship is as follows: when all input signals are at a low level (“0”), the light remains off, indicating that no sperm was detected at the detection site. When the seminiferous tubule signal is high (“1”) and both the sperm and physiological saline signals are low (“0”), the light remains off, confirming the presence of only the seminiferous tubule without sperm. When both the seminiferous tubule and sperm signals are high (“1”) and the physiological saline signal is low (“0”), the blue light turns on, suggesting that the detection site is suitable for sperm retrieval surgery. When the seminiferous tubule and physiological saline signals are high (“1”) and the sperm signal is low (“0”), the green light turns on, indicating that no sperm was detected. When the sperm signal is high (“1”) and both the seminiferous tubule and physiological saline signals are low (“0”), the blue light turns on, indicating that the detection site is suitable for sperm retrieval surgery. When both the sperm and physiological saline signals are high (“1”) and the seminiferous tubule signal is low (“0”), both the blue and green lights turn on, confirming that the detection site is suitable for sperm retrieval surgery. When the seminiferous tubule and sperm signals are low (“0”) and the physiological saline signal is high (“1”), the green light turns on, confirming that no sperm is detected at the detection site. Finally, when all signals from the seminiferous tubule, sperm, and physiological saline are high (“1”), both the blue and green lights turn on, confirming that the detection site is suitable for sperm retrieval. Following the simulation using the software Multisim, the results of this logic circuit were found to be consistent with expectations, as depicted in Fig. 6d and Fig. S6-7.

According to the circuit, a microneedle-based accurately detection device is subsequently developed, as shown in Fig. 6e. This approach effectively minimizes the pain experienced by patients with non-obstructive azoospermia during sperm retrieval procedures and reduces potential damage to testicular tissue. Additionally, it provides an auxiliary detection mechanism for clinical embryologists conducting sperm retrieval operations, while establishing a solid foundation for the advancement of future detection technologies within the field of reproductive medicine.

4. Conclusions

Based on the experimental results and systematic analyses, we conclude that the novel Ag/C₄₀H₅₆-MnO₂/FTO memristive device exhibits enhanced sensitivity for sperm detection within seminiferous tubules. This technological advancement holds significant potential for optimizing clinical procedures in assisted reproductive medicine, particularly for improving sperm retrieval during in vitro fertilization (IVF) treatments for patients with NOA. The developed memristor integrated with microneedles showcases significant clinical value on account of its remarkable non-invasive monitoring ability. Its working mechanism is that Ag⁺ ions will undergo migration when spermatozoa exist within the seminiferous tubules, which subsequently triggers corresponding adjustments in the electrical conductivity of the intermediate layer material. These physicochemical alterations are capable of precisely differentiating between the seminiferous tubules that contain sperm and those that do not, and can also offer real-time feedback during the surgical process. This innovation boasts numerous outstanding advantages, which are specifically demonstrated as follows: Firstly, it can minimize tissue trauma by accurately pinpointing the location of spermatozoa. Secondly, it can effectively cut down the operation time through targeted retrieval guidance. Thirdly, it can genuinely enhance the final surgical outcomes by employing detection methods based on electrical conductivity. Fourthly, it can considerably improve patient comfort by averting unnecessary tissue exploration. The terminal feedback mechanism of this device has established a closed-loop monitoring system, which has entirely revolutionized the traditional sperm retrieval methods. By transforming biological information into quantifiable electrical signals, this technology can dynamically monitor the distribution patterns of spermatozoa during the microdissection testicular sperm extraction (micro-TESE) procedures. This breakthrough signifies a paradigm shift in the realm of reproductive medical instruments, mainly embodied in the following three key advancements: Firstly, the memristive technology has witnessed its first clinical application in gamete detection. Secondly, nanotechnology has gained a new application pathway in andrological surgery. Thirdly, intelligent medical devices equipped with biological sensing capabilities have been successfully developed. The successful application based on the memristor platform has laid a solid technological foundation for the next-generation medical devices in the field of reproductive health monitoring, and holds broad prospects for expansion. It is anticipated to be extended to other precision medicine application areas that demand detection at the cellular level and real-time biofeedback.

CRediT authorship contribution statement

Yu Cui: Writing – original draft, Data curation, Conceptualization. Wenting Yang: Methodology, Formal analysis. Xinyu Zhang: Investigation. Yiqing Chen: Software. Mengna Wang: Validation. Zelin Cao: Visualization. Kaikai Gao: Validation. Bai Sun: Writing – review & editing, Supervision, Funding acquisition. Yanmin Ma: Writing – review & editing, Supervision. Guoqing Tong: Supervision.

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.

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.

Appendix ASupplementary data

The following is the supplementary data to this article.

Data availability

Data will be made available on request.