Introduction

Owing to the unique properties, the flexible electronics offer significant advantages over traditional silicon-based electronic materials in various fields including bending. The flexible electronics are found to be well-suited for various potential applications, particularly, in space exploration, human–machine interaction, and wearable electronics [1,2,3,4]. As a core component of flexible electronics, flexible sensors are capable of acquiring and converting various types of information (e.g., strain, pressure, temperature), serving as a crucial interface between the system and the external environment. The accuracy of sensor data acquisition is paramount, particularly for flexible pressure sensors [5,6,7,8]. Currently, extensive research has been conducted on achieving high sensitivity, wide range, and high stretchability of the flexible sensors [9,10,11,12,13]. For example, Gao et al. [14] constructed a pressure sensor based on a skin-like structure, achieving a sensitivity greater than 1000 kPa−1. They found that the excellent flexibility and sensitivity of the developed pressure sensor can be achieved by tuning the materials design and structure. However, the method is highly limited due to it difficult fabrication and high costs.

In recent years, the electrospun nanofibers with high specific surface area and controllable-structure has provided an attractive platform for various potential applications [15,16,17]. As a relatively mature commercial technology, electrospinning has obvious advantages in the controllable and large-scale production of various kinds of electrospun nanofibers [18]. The feasible electrospun nanofiber-based sensors have remarkable attractiveness and dominance in designing and manufacturing of thin, soft, and comfortable wearable products [19,20,21,22]. According to the previous research, three main sensing principles (piezoelectric, piezoresistive and capacitive) are required for pressure sensors [23]. Among them, piezoelectric materials have the potential to be self-powering due to their ability to collect various mechanical movements and convert them effectively into electrical energy [24]. To this end, researchers have devoted considerable research efforts to develop high-performance piezoelectric sensors [25]. Park et al. [26] successfully embedded lead-free (Na, K)(Nb, Sb)O3-BaZnO3-(Bi, K)ZrO3 in polycrystalline silicon into P(VDF-TrFE) (poly(vinylidene fluoride-co-trifluoroethylene)) electrospun nanofiber matrix. The nanofibers maintained an open-circuit voltage, a high maximum generated power, and an output voltage of approximately 12.2 V, 33.2 nW, and 1.25 V, respectively, despite being bent for 10,000 cycles. Li et al. [27]. reported that PDA-coated BaTiO3 nanowires (BTNW) were introduced into P(VDF-TrFE) electrospun nanofibers to form coaxial composite nanofibers, which produced an optimal piezoelectric output of 18.2 V under an impact force of 5 N with excellent sensitivity of 4.3 V N−1. Frank K. et al. [28] used electrospinning to grow CsPbI2Br perovskite crystals in situ on polyvinylidene fluoride (PVDF) nanofibers. CsPbI2Br spread over the nanofiber, leading to a rough surface, and complementing piezoelectric nanocomposite formation with PVDF for superior stress excitation. The output voltage of 8.3 V and current of 1.91 μA and used to charge capacitors for powering electronics.

It is evident that despite the inherently low piezoelectric performance of electrospun nanofibers, the introduction of inorganic nanomaterials through embedding or in-situ growth can leverage interfacial polarization between organic and inorganic components or enhances the material stress induction, thereby achieving heightened piezoelectric performance [29]. Nevertheless, the presence of inorganic materials poses issues such as sensitivity to environmental conditions and biological toxicity, restricting their practical applications [30]. Currently, achieving excellent piezoelectric performance without introducing inorganic materials remains a difficult challenging endeavor for electrospun nanofibers. Ding et al. [31, 32]. have developed electrospinning-jetting and electrospun mesh techniques to fabricate 2D nanonets with a continuous topology Steiner tree structure and a diameter of 20–35 nm. Due to the unconstrained nature of polymer materials, these nanonets exhibit outstanding performance in air and water filtration. Nevertheless, in the solution systems capable of producing ultrafine nanofibers in current research, achieving satisfactory formation necessitates strict requirements not only on the choice of the receiving substrate material but also on the substrate thickness, which significantly hinders their application in other fields. But inspired by it, if similar ultrafine nanofibers are introduced into conventional electrospun nanofibers, the resulting hierarchical structure are expected to have distinctly different pressure effects that will impact piezoelectric properties generation from the nanofiber membrane.

This work investigates, for the first time, the effect of ultrafine nanofibers on the piezoelectric properties of electrospun nanofibers. By varying the spinning time, PVDF composite nanofiber with different structures are created. This structure was confirmed to correlate with piezoelectric properties through a combination of experiments and finite element simulations. Additionally, by introducing dopamine during the electrospinning process and benefiting from hydrogen bonding and dipole interactions, it successfully induces PVDF and ultrafine PVDF nanofibers to form a higher β-crystal orientation. Moreover, the core–shell structure formed by dopamine (DA) and PVDF electrospun nanofibers enhances the material stability, ensuring high piezoelectric output and sensitivity. This study highlights the fact that the piezoelectric voltage output can be improved by 66.7% and high sensitivity by constructing a composite nanofiber with uniformly dispersed two-dimensional network topology. It is anticipated that the results presented in this work propose a new door for electrospun nanofibers to be used in the field of piezoelectric sensors and provide new ideas for the design and development of electrospun nanofiber membranes with high-performance piezoelectric materials.

Experimental

Materials

PVDF granules (Mr = 275,000) were obtained from Sigma-Aldrich (Japan). PVDF powder (Mr = 500,000) were purchased from Arkema (Japan). Tetrabutylammonium chloride (TBAC), Dopamine (DA), N,N-dimethylformamide (DMF, ≥ 99.9%) and Acetone (≥ 99.9%) were supplied by Sigma-Aldrich (Japan). All the chemicals used were of analytical grade.

Formation of Electrospun Nanofibers and Ultrafine Nanofibers

To prepare the electrospun nanofiber solution, at first, a 16 wt% of PVDF (Mr = 275,000) solution was prepared in a mixture of DMF and acetone (3:2). To the above mixture, 0, 1, 2 and 3 wt% (solute mass) of DA was dissolved and a homogenous spinning solution was obtained by stirring continuously for 2 h at 60 °C. In case of the preparation of spinning solution for obtaining the ultrafine nanofiber, TBAC powders with 2.5 wt% was first dissolved in DMF by magnetically stirring for 2 h at 40 °C. Subsequently, the DA of 0, 0.5, 1 and 2 wt% (solute mass) was dispersed in the TBAC solution and the mixture was stirred for 2 h at 50 °C. Finally, 3 wt% of PVDF (Mr = 500,000) powder was dissolved in the DA/TBAC solution. Electrospinning process was performed by using the above 2 mixed solutions into a 5 mL plastic syringe fitted with a 18G/20G stainless steel nozzle. In the process of electrospun nanofibers, a high voltage power supply was set to 15 kV between the nozzle tip and the collector at 20 cm. When preparing ultrafine nanofiber, the voltage is 22 kV and the tip to collector distance of 22 cm. During the electrospinning process, the temperature was maintained at 25 °C and the humidity was controlled between 35 and 40%. For the preparation of nanofiber membranes with hierarchical structures, PVDF/DA nanofiber membranes were covered on the drum rotator, of which the roller speed was selected to be 100 rpm for collecting the ultrafine nanofiber. The flow chart for the fabrication of composite nanofiber piezoelectric sensor with hierarchical structure is shown in Fig. 1.

Fig. 1
figure 1

Schematic of the fabrication of composite nanofiber piezoelectric sensor with hierarchical structure

Full size image

Configuration of the Sensor

A very lightweight, flexible, piezoelectric sensor (effective area of ~ 25 × 25 mm2) was fabricated by different structures and types of PVDF nanofiber membranes. PVDF nanofiber membranes was sandwiched between two copper (Cu) electrodes and two copper wires were soldered on the electrodes. The fabricated piezoelectric sensor, configured as Cu electrode/nanofiber membranes/Cu electrode, was encapsulated by PI (Polyimide) tapes to enhance its durability. For collecting biomechanical signals, the sensor was sealed with tape to make it fit better with human skin. Finally, the whole device was subjected to an appropriate compression, which helped to avoid the generation of triboelectric signal by eliminating the gap between different layers.

Characterizations and Measurements

The surface morphology of composite nanofiber piezoelectric sensors was observed using a JEOL JSM-6010 LA scanning electron microscope (FE-SEM). The FE-SEM was equipped with an energy-dispersive X-ray spectrometer (EDX), also from Japan called Inca Oxford, to image the surface morphology of individual nanofibers at an accelerating voltage of 10 kV. In addition, the internal and external structure and crystal orientation of the nanofibers were observed through transmission electron microscopy (TEM, JEOL JEM-2010). The attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy was used to analyze the chemical structures of nanofiber membranes using a spectrometer (IR Prestige-21, Shimadzu, Japan). The crystalline structures were analyzed via X-ray diffraction (MiniFlex300 Rigaku, Japan) equipped with Cu Kα radiation. The crystallinity of the electrospun nanofibers was determined by differential scanning calorimetry (DSC, DSC-8500, PerkinElmer, USA) with the heating rate of 10 °C min−1 and temperature ranging from 35 to 300 °C. Chemical modification of nanofiber membranes was evaluated by X-ray photoelectron spectroscopy (XPS, Axis-Ultra HSA SV, Kratos Analytical Inc., U.K.). The thickness of all the fiber films was determined using a digital thickness gauge (YHT780, China) by putting the film sample between two fixtures. A quasi-static d33 meter (ZJ-3AN, Chinese Academy of Sciences) was used for d33 values. The dielectric properties were tested by precision impedance analyzer (1260/1296, Solartron Analytical Inc., U.K.) in range 102–106 Hz. Electrical outputs of the packaged devices were measured by Oscilloscope (DS-5414A ViewGoll, IWATSU, Japan) and Keithley 6514 electrometer and the corresponding data were recorded by Labview software.

Results and Discussion

Morphology of the Nanofiber Membranes

The morphology of PVDF/DA nanofibers with different DA concentrations (0, 1, 2, 3 wt%) is presented (Fig. S1 in the supporting information), hereinafter referred to as PDS-0, PDS-1, PDS-2, PDS-3. The Representative images of PVDF nanofiber morphology with different DA concentrations are shown in Fig. 2a. As the concentration of DA increases to 2 wt% (PDS-2), the adhesion between fibers becomes more pronounced, accompanied by an increase in the average fiber diameter from 236 to 268 nm. From the TEM results in Fig. 2b, it is evident that these phenomena are not only attributed to the increase in DA concentration but also to the formation of typical core–shell structures between DA and PVDF. However, as the DA concentration increases to 3 wt% (PDS-3), the fiber diameter decreases, as observed from TEM (Fig. 2b), primarily due to the substantial embedding of DA into the PVDF matrix. Feng et al. [33] research suggests that the formation of the core–shell structure may be attributed to interactions between –NH2 and –CF2. Nevertheless, the molecular dynamics (MD) simulations were employed to compare the radial distribution functions (RDF) of fluorine atoms near –NH2 groups (N–H–F) and –OH groups (O–H–F) in the absence of an electric field. It was observed that within the range of 2.4–3.1 Å, the RDF for O–H–F exhibited an initial peak followed by a decrease, contrasting with N–H–F interactions (Fig. 2c). Notably, hydrogen bonding interactions predominantly occurred within the range of 2–3 Å. This implies that both –OH and –NH2 moieties exert an influence on –CF2 in PVDF in a solution state (Fig. 2d (i)). In addition, during the characterization of PDS-2 fibers post-electrospinning, the results obtained from Energy-Dispersive X-ray Spectroscopy (EDS), as depicted in Fig. 2e, bore a striking resemblance to prior research. With the sole exception of fluorine, the distribution of nitrogen (N) appeared notably more even than that of oxygen (O) within the core region. Moreover, the XPS findings, detailed in Fig. 2f–g, not only unequivocally confirmed the presence of carbon (C), nitrogen (N), fluorine (F), and oxygen (O) elements within PDS but also revealed an intriguing facet. Specifically, in the N 1s peak deconvolution, in addition to the discernible N–H and –NH2 characteristic peaks of DA, a distinct N–F characteristic peak (406.8 eV) emerged. This observation implies a robust interaction between nitrogen in DA and the –CF2 moieties on polyvinylidene fluoride, underscoring their strong interplay within the nanofiber structure. When amalgamating prior research with the findings from the molecular motion simulations presented in this article, a predominant factor behind the formation of the core–shell structure emerges the positive charge carried by –NH2 in DA under the influence of an electric field. This positive charge compels these entities to migrate to more distant positions from the nozzle, driven by electrostatic repulsion. However, the compelling dipolar interaction between these positively charged –NH2 and –CF2 groups subsequently exerts an attractive force, making it more probable for a protective shell structure to take shape on the surface of the fibers. Significantly, the alignment of PVDF molecules under this long-range interaction exhibits a pronounced level of symmetry (Fig. 2d (ii)), making it more predisposed to the β-type structure of CH2–CF2 [34](in an all-trans conformation referred to as TTTT structure).

Fig. 2
figure 2

Characterization of PVDF/DA electrospun nanofibers. a SEM image of PVDF with 0, 1, 2, 3 wt% DA (PDS-0, PDS-1. PDS-2, PDS-3). b TEM image of the core/shell PDS-1, PDS-2, PDS-3. c RDFs between the –CF2 groups in PVDF and the –NH2 (black) and –OH (red) groups in DA. The inset is the MD simulations of a PVDF/DA molecular structure in equilibrium, showing the core (PVDF)–shell (DA) arrangement. d Schematic of the PVDF chain arrangement and orientation induced by intermolecular interactions during the electrospinning process for the fabrication of PVDF/DA core/shell nanofibers. e EDS mapping of F (brown), O (blue), and N (petrol) elements distribution in a PVDF/DA core/shell NF shown in the FE-SEM image. f XPS survey spectra of PDS-2 (red) and pure PVDF (black) nanofibers. g The XPS N 1s spectrum of the PDS-2. h Digital images of PDS-2

Full size image

Figure 3a illustrates the X-ray diffraction (XRD) data of PDS. Notably, the prominent diffraction peak is positioned at 2θ = 20.5°, corresponding to the β (110) crystal plane. Additionally, relatively weaker diffraction peaks at 18.4° and 36° are attributed to the α (020) and α (200) crystal planes, respectively. XRD peak splitting (Fig. 3b–e) reveals that as the concentration of DA increases, the proportion of α characteristic peaks initially decreases and then rises, with PDS-2 displaying the lowest α characteristic peak ratio. This observation is also substantiated by ATR-FTIR data (Fig. 3f). Despite the presence of clear β-phase characteristic peaks at various DA concentrations, such as 840 cm−1 (–CH2 rocking), 1176 cm−1 (C–F stretching), and 1275 cm−1 (C–F out-of-plane deformation), the height of the α-phase characteristic peak at around 763 cm−1 (–CF2 bending and scissile bending) follows a pattern of initial reduction and subsequent increase. This is primarily attributed to the increased incorporation of DA into the PVDF matrix, which disrupts the orientation and arrangement of PVDF chains in the nanofibers (Fig. 1b), thereby diminishing crystallinity. By combining ATR-FTIR data, the β-phase content can be calculated using F(β) = Aβ/(1.26Aα + Aβ) (S2 in the Supporting Information), where Aα and Aβ represent the absorbance areas at 763 cm−1 and 840 cm−1, respectively (F(β)) [35,36,37]. Please note that a coefficient of 1.26 is assumed to be valid for electrospun nanofibers. The results indicate that PDS-2 boasts the highest β-phase content, reaching 89% (Fig. 3g). DSC analysis provides further validation of this outcome. PDS-2 exhibits the highest melting temperature peak (Fig. 3h), as the increase in β-phase content corresponds to a higher melting temperature of PDS. Additionally, the crystallinity content (Xc) of PVDF was computed from the DSC spectrum [38, 39], with PDS-2 achieving the highest value (Fig. 3i), totaling 67.5%. This PDS-2 is 22.14% greater than pure PVDF, making PDS-2 the preferred base to produce ultra-fine PVDF/DA nanofibers.

Fig. 3
figure 3

Crystallinity characteristics of PDS. a XRD spectra of PVDF/DA nanofibers with different DA contents. b–e Inset is the β-phase percentage as a function of the DA content, quantified from the XRD spectra by the curve deconvolution. f FTIR spectra and g β-phase contents (F(β)) of PVDF/DA nanofibers with different DA contents. h DSC thermograms and i the crystallinity contents (Xc) value as a function of the DA content, quantified from the FTIR spectra and the DSC thermograms, respectively. β-phase crystallinity (%) = Xc × F(β) × 100%

Full size image

Figure 4a displays FE-SEM images of ultrafine nanofibers with varying DA concentrations (0, 0.5, 1, 2 wt%, hereafter referred to as PDU-0, PDU-1, PDU-2, PDU-3). With increasing DA concentration, there is a slight enlargement in fiber diameter, reaching approximately 35 nm. Notably, in comparison to the twisted ultrafine nanofiber network observed in PDU-0, the structure of the ultrafine nanofiber network becomes more stable after the addition of DA, but it also leads to the emergence of more droplets. Among them, both PDU-1 and PDU-2 showed good conditions (Fig. S3 in the Supporting Information), and it is exciting that the self-oriented β phase of ultrafine PVDF nanofibers was confirmed by TEM for the first time (Fig. 4d). PVDF polymeric chains were aligned and oriented in parallel to the PVDF/DA interface. The inter-chain distance was 4.28 Å (upper inset of Fig. 4d), corresponding to the (110) plane.

Fig. 4
figure 4

Characterization of PVDF/DA ultrafine nanofibers. a SEM image of PVDF with 0, 0.5%, 1%, 2 wt% DA (PDU-0, PDU-1, PDU-2, PDU-3). b and c TEM image of the PDU. d High resolution TEM (HRTEM), the enlarged view of HRTEM (top inset) and FFT images (lower inset) of PDU-1 showing aligned PVDF chains along the fiber axis. e FTIR spectra and f XRD spectra of PVDF/DA ultrafine nanofibers with different DA contents. g ultrafine nanofiber forming mechanism diagram. h EDS mapping results of single ultrafine nanofibers of PVDF/DA containing 0.5 wt% DA. i RDFs between the –CF2 groups in PVDF and the –NH2 (black), –OH (red) and –N+(blue) groups in DA and TBAC. The inset is the MD simulations of a molecular structure in equilibrium, showing the PVDF, DA and TBAC arrangement. j XPS survey spectra and k N 1s spectrum of the PDU-1

Full size image

This can also be proven from the ATR-FTIR data (Fig. 4e) and XRD (Fig. 4f). Compared with PDS, the peak intensity of PDU at 1275 cm−1 is significantly higher, which is a unique characteristic peak of β. And the alpha phase peak intensity at 763 cm−1 is very weak. It can also be seen from XRD that the α characteristic peak at 36° disappears, and there is only the β (110) crystallization characteristic peak at 20.6°, reflecting the presence of a higher β phase in PDU. Through calculation, the β phase content of PDU-1 reaches the highest value of 98%, but with the DA content, the β phase content drops significantly to 88%, which is lower than the β phase content without adding DA. From the ATR-FTIR data (Fig. 4e), it is found that the main reason for the decrease is due to the increase in DA loading to an increase in the γ phase (intensity increase at 1234 cm−1) in the PDU. In addition, from the TEM data (Fig. 4b–c and Fig. S4 in the Supporting Information), no obvious core–shell structure was observed. The primary reason for the observed phenomenon may be closely related to the unique formation mechanism of an ultrafine nanofiber network (Fig. 4g). Presently, researchers posit that the critical factor for generating electrospun ultrafine nanofibers is the charge density in the solution, which must surpass the droplet threshold, denoted as Dc [40, 41], where \(Dc = \sqrt {288\varepsilon \gamma /\delta \rho^2 D^3 }\). Here, ε represents the ambient permittivity, γ is the surface tension of the fluid, δ is the correction factor, ρ denotes the fluid density, and D signifies the diameter of the charged fluid. Additionally, \(D = 1.46Q^{0.44} \varepsilon^{0.12} \eta^{0.32} K^{ - 0.12} \gamma^{ - 0.32} \) [42], where Q stands for the volumetric flow rate, η is the fluid viscosity, and K represents the fluid conductivity. Consequently, the formation of ultrafine nanofibers necessitates not only high electrical conductivity but also a low-concentration state in the solution. In this state, many droplets are ejected due to the high electrical conductivity and surface tension. These droplets disperse within the electric field and undergo a series of complex processes, including collision, stretching, and evaporation, which vary with changes in electric field intensity [43, 44]. This may result in the inability of DA to exist around PVDF ultrafine nanofiber stably and continuously.

Molecular dynamics simulations further validate the existence of this phenomenon. Analyzing the radial distribution function (RDF) reveals a similarity to the PDS process (Fig. 4i). Even though TBAC is present, the priority of DA is significantly higher than that of TBAC. This emphasizes that a core–shell structure will also form in the normal jet mode. Moreover, an examination of the highest β-phase PDU-1 surface scan and XPS data (Fig. 4j–k) indicates the presence of DA on the surface. However, the absence of N–F in the N 1s spectrum suggests that there is no strong bond between –NH2 and –CF2, demonstrating that the high β-phase of PVDF primarily relies on induction through high stretching and electrical fields. Nevertheless, due to the positive charge of –NH2 in DA, a small amount of DA may still promote the formation of the PVDF β-phase under the influence of hydrogen bonding and dipole forces. Moreover, an examination of the highest β-phase PDU-1 EDS and XPS data (Fig. 4h–k) indicates the presence of DA on the surface. However, the absence of N–F in the N 1s spectrum suggests that there is no strong bond between –NH2 and –CF2, demonstrating that the high β-phase of PVDF primarily relies on induction through high stretching and electrical fields. Nevertheless, due to the positive charge of –NH2 in DA, a small amount of DA may still promote the formation of the PVDF β-phase under the influence of hydrogen bonding and dipole forces. Nevertheless, due to the electric field inducing a positive charge on the –NH2 in DA, a small amount of DA can still promote the formation of the PVDF β-phase under the influence of hydrogen bonding and dipole forces. However, when the DA content within the droplets is excessive, leading to an uneven distribution, it may instead induce the formation of the γ phase.

Results and Analysis of Composite Nanofiber Piezoelectric Material

The high β-phase content in the NFs was expected to largely enhance their piezoelectricity. Using PDS-2 with the highest β phase as the base, PDU-1 was spun for 90 min, 180 min, and 240 min respectively, and three types of multi-scale PVDF/DA nanofiber membranes were prepared. For the convenience of recording, they are abbreviated as MPD-1 MPD-2 and MPD-3. From FE-SEM results (Fig. 5e, Fig. S5 in the Supporting Information), it was found that the density of the ultrafine fiber network in the pores of PDS-2 fibers did not increase with time. Specifically, after 240 min, numerous "film" and polymer spheres developed within the pores of PDS-2 fibers or adjacent to the fibers, without any increase in fiber density within the pores. It is crucial to emphasize that at the 180 min mark, a noticeable density of ultrafine nanofibers formed within the pores of the thicker nanofibers. These ultrafine nanofibers were closely interconnected with the thicker nanofibers, resulting in the formation of a complete two-dimensional topological network structure. Additionally, under higher magnification electron microscopy, it was observed that the thicker nanofibers and ultrafine nanofibers were closely connected, which was not only due to the adhesion of DA on the surface of the thicker nanofibers but also to Formation of "thin film" during spinning of ultrafine nanofibers. This helps to improve the mechanical properties of the composite nanofiber membrane (Fig. S6 in the Supporting Information).

Fig. 5
figure 5

a Dielectric constant, and b dielectric loss diagram of the PDS-2 and MPD. c Piezoelectric coefficient (d33) of the PDS-2 and MPD. d Optical photo and schematic diagram of flexible sensor structure. e typical FE-SEM images of PDU-1 with different spinning times (90 min, 180 min and 240 min) based on PDS-2 and corresponding output voltage. f Comparison of output voltages under different structures. g Output current and h data comparison under different structures

Full size image

To assess the impact of PDU-1 spinning time on the encapsulated electrical properties of MPD, a comprehensive investigation of MPD dielectric and piezoelectric properties under varying spinning durations was conducted. The relative dielectric constant and dielectric loss of MPD were measured across frequencies ranging from 102 to 106 Hz and were compared with the performance of PDS-3, as illustrated in Fig. 5a–b. It is noteworthy that the dielectric constant of MPD initially increases and then decreases as the spinning time of PDU-1 is extended. The initial increase can be attributed to differences in dielectric properties between PDS and PDU, which lead to an increase in Maxwell–Wagner–Sillars interface polarization and, subsequently, an increase in the dielectric constant. The subsequent decrease is linked to the agglomeration of TBAC due to the increased presence of "film" in the fiber membrane. In addition, with the addition of ultrafine nanofibers, the dielectric loss of MPD remains in a low range, which is due to the polymer matrix. The piezoelectric coefficient (d33) of MPD was determined using a static d33 meter, as depicted in Fig. 5c. The d33 value of MPD is found to be influenced by the content of PDU-1. With an increase in the spinning time of PDU-1, the d33 value of MPD rises from 32 to 52 pC N−1, with MPD-2 displaying the highest piezoelectric coefficient. The improvement in d33 can be attributed to its higher beta phase content compared to PDU-1. Notably, this high d33 value surpasses that of other reported PVDF-based nanostructures (films, nanoribbons, and nanofibers) and is comparable to certain ceramic-based micro/nanofibers (Table S1, Supporting Information) [17, 45,46,47,48].

To display its piezoelectric properties more intuitively, a piezoelectric test unit was made from the above-mentioned multi-scale structural fiber membranes at different times. The structure of the test unit is shown in Fig. 5d, with a size of 25 × 25 mm2 and an automatic stepper. The motor controller tested the film piezoelectric properties by repeatedly pressing and releasing it at 1.5 N and a low frequency of about 1.2 Hz. The performance results are shown in Fig. 5f–h. The open-circuit voltage and short-circuit current have increased, and both show a similar trend with d33, verifying that VOC and ISC are directly related to the d33 value. Especially for MPD-2, the open-circuit voltage output reaches 14.3 ± 0.2 V and the short-circuit current reaches 1.01 ± 0.1 μA.

It is of significance to highlight that despite the increase in PDU-1 content, the electrical output performance of MPD did not exhibit improvement; in fact, it showed a decline. To gain a clear understanding, it is imperative to elucidate the factors influencing the electrical output of piezoelectric materials during electromechanical operation. Presently, researchers have developed multiple theoretical models to elucidate the factors impacting the electrical output of piezoelectric materials. These theoretical models broadly delineate the microstructure of piezoelectric materials into alternating material layers and air layers. Following Kirchhoff’s second law, the formula for output voltage can be expressed as follows [49, 50]:

$$ V = \frac{Q}{\varepsilon_0 S}\left[ {\frac{d_e }{{\varepsilon_r }} + \frac{{d_{{\text{gas}}} }}{\varepsilon_g }} \right] - \frac{{d_{{\text{gas}}} }}{n\varepsilon_0 \varepsilon_g }\mathop \sum \limits_j^n \sigma_j $$
(1)

where V represents the open-circuit voltage, Q denotes the total charge density, ε0 signifies the vacuum permittivity, S stands for the effective operational area of the device, de signifies the total thickness of the solid layer, εr represents the relative permittivity of the material layer, dgas corresponds to the total thickness of the air layer, εg symbolizes the relative permittivity of air. σj represents the charge captured at the solid–gas interface after polarization, where j is the layer number of the air layer, and n represents the total number of layers.

The following corresponding boundary condition can be obtained at time t = 0:

$$ V\left( {t = 0} \right) = \frac{Q_0 }{{\varepsilon_0 S}}\left[ {\frac{d_e }{{\varepsilon_r }} + \frac{{d_{{\text{gas}}} \left( {t = 0} \right)}}{\varepsilon_g }} \right] - \frac{{d_{{\text{gas}}} \left( {t = 0} \right)}}{n\varepsilon_0 \varepsilon_g }\mathop \sum \limits_j^n \sigma_j = 0 $$
(2)

By combining Eqs. (1) and (2), the output voltage can be written as:

$$ V\left( t \right) = \frac{Q(t)}{{\varepsilon_0 S}}\left[ {\frac{d_e }{{\varepsilon_r }} + \frac{{d_{{\text{gas}}} \left( t \right)}}{\varepsilon_g }} \right] - \frac{{d_{{\text{gas}}} (t)Q_0 }}{\varepsilon_0 S}\left( {\frac{d_e }{{\varepsilon_r d_{{\text{gas}} - 0} }} + \frac{1}{\varepsilon_g }} \right) $$
(3)

where dgas(t = 0) = dgas_0, Q0 represents the initial charge density at t = 0, Q(t) expresses the quantity of electric charge on each electrode at the instant when the top and bottom electrodes are approaching each other immediately after the stimulus commences at time t = 0.

Under short-circuit (SC) current conditions, ISC can be obtained from Equation:

$$ I_{{\text{SC}}} = I\left( {V = 0} \right) = Q_0 \frac{{\varepsilon_g d_e + \varepsilon_r d_{{\text{gas}}} \__0 }}{{\varepsilon_g d_e \varepsilon_r d_{{\text{gas}}} \__0 }} \times \frac{{d_{{\text{gas}}} (t)}}{{d_{{\text{gas}} - 0} }} $$
(4)

Indeed, from the above formula, it is apparent that the electrical output performance of piezoelectric materials is positively correlated with the material charge density. In the case of organic materials, an elevation in the dielectric constant contributes to an increase in the peak of transferred charges after polarization [51], thus elucidating one of the factors behind the high electrical output performance of MPD-2.

Furthermore, apart from the inherent material properties, the enhancement of charge density in piezoelectric materials is also influenced by the distribution of mechanical stress. As the time duration of PDU-1 increases, the structural characteristics of MPD change, and this, in turn, affects the stress distribution within the material. To theoretically assess the impact of structure on piezoelectric performance, representative material models were established for stress and piezoelectric potential calculations using finite element simulations (COMSOL Multiphysics 6.1). The primary structure consisted of a 25 × 25 mm2 matrix of PVDF with a thickness of 30 mm and an 8 × 8 grid. Based on the structural variations within MPD, it was divided into three structures. In Fig. 6a, the top layer has no fine fibers within the pores, referred to as S-1. In Fig. 6b, the top layer exhibits an even distribution of fine fibers within the pores, known as S-2. In Fig. 6c, the top layer alternates between fine fibers and PVDF membranes within the pores denoted as S-3. All structures were anchored at the base and grounded, and an external force was applied parallel to the z-axis for stimulation. The corresponding schematic diagrams for the process are shown in Fig. S7 (in the Supporting Information).

Fig. 6
figure 6

Simulation results of composite nanofiber membranes with different structures. a–c Stress distribution under external force of three structures (S-1, S-2, S-3). d–f) Total output voltage induced by stress transfer variation of three structures (S-1, S-2, S-3)

Full size image

The piezoelectricity can be elaborated by coupling the electrical behavior of dielectric material and Hooke law, as described in the following equations [52]:

$$ S = s \times T $$
(5)
$$ D = \varepsilon \times E $$
(6)

where S is strain, s is compliance, T is applied stress, D is the electric charge density displacement, E is electric field strength and ε is permittivity; Their linear relations can be shown by the following equations [53]:

$$ S = \varepsilon^E \times T + {\text{d}}E $$
(7)
$$ D = {\text{d}}T + \varepsilon^T \times E $$
(8)

The above equations reveal the different piezoelectric outputs owing to strain variations. Based on the above concept, it is not difficult to find that the higher the internal the stress of the material, the higher the potential can be generated. In S-1, stress distribution is uniform. However, in S-2, due to the presence of fine fibers, stress within the pores of the thick fibers changes. Notably, stress concentration becomes prominent at the connections between the fine fibers and the thick fibers, resulting in significantly elevated stress levels (more than 30 times that of S-1). The stress concentration simultaneously promotes the piezoelectric response of the two-scale nanofibers and the stress-induced polarization under load, thereby feeding back the highest potential (Fig. 6d–f). In addition, although the content of fine fibers in S-3 (here it is assumed that the film is quite composed of fine fibers) is increased, the presence of the film weakens the stress concentration of the thin fibers and gives a lower potential than S-2. This also explains why MPD-2 has a higher piezoelectric constant.

Furthermore, structural changes induce different piezoelectric phenomena in MPD at different frequencies (2–10 Hz), as illustrated in Fig. 7a–c. Notably, when the frequency falls within the range of approximately 2–4 Hz, the voltage measured by the oscilloscope exhibits minimal fluctuations. As the frequency is increased to around 6–10 Hz, both MPD-2 and MPD-3 generate voltages that fluctuate regularly, with MPD-2 displaying shorter fluctuation periods. This change can be attributed to the differential response speed of ultrafine nanofibers compared to coarse fibers (Fig. 7d). In contrast to conventional piezoelectric materials, this periodic fluctuation is more conducive to providing feedback when the frequency changes under the same force [54].

Fig. 7
figure 7

Output voltage of a MPD-1, b MPD-2, c MPD-3 at different frequencies. d Possible charge change mechanism diagram

Full size image

Therefore, based on the results, it is concluded that MPD-2 exhibits the most favorable piezoelectric performance, and subsequent research has been focused on the sensing applications of MPD-2.

Practical Application and Stability of the MPD-2

When considering piezoelectric sensors, it is crucial to assess sensitivity and the linear pressure range. The sensitivity (S) is defined as \(S = (\Delta V/V_0 )\Delta P\), where \(\Delta V/V_0\) represents the relative variation in voltage [46], while \(\Delta P\) indicates the applied pressure. Figure 8a illustrates the sensitivity of the MPD-2 sample under applied pressure. The data reflects voltage stepwise changes in response to forces ranging from 1.5 N to 40 N. Notably, a substantial change in voltage amplitude is observed in the 1.5–10 N region, while the voltage change amplitude decreases in the 16.5–40 N range. Linear fitting (Fig. 8b) reveals that MPD-2 exhibits high sensitivity in both the low-pressure region (below 10 N) and the high-pressure region (above 10 N). Specifically, the sensitivity values in the low-pressure (0–4 N) region reach 7.29 V N−1, in the medium-pressure (5–10 N) region reach 4.11 V N−1, and in the high-pressure (30–40 N) region also reach 1.41 V N−1. These values demonstrate higher sensitivity and a broader pressure range when compared to PVDF-based piezoelectric materials and previously reported polymer composites (Table 1). Additionally, due to the stable structural network formed by DA adhesion between ultrafine nanofibers and thick fibers, consistent voltage output is maintained even after 10,000 taps at 2 Hz (Fig. 8c).

Fig. 8
figure 8

a Output voltage signals under different pressures. b Directional sensing behavior of MPD-1. c Long-term stability and durability of MPD-1 during 10,000 working cycles at the same location (4 Hz, 1.5 N). Real-time monitoring of human motions and subtle physiological activities, including d speaking, e wrist pulse, f finger tapping, g elbow bending, h knee bending and i foot stamping

Full size image
Table 1 Comparison of the various sensor device and their performances
Full size table

Certainly, besides its exceptional sensing capabilities and stability, its performance in practical applications must be taken into account. Thanks to the flexibility and lightweight nature of the nanofiber membrane, it finds utility as a wearable film sensor for detecting human movements and subtle physiological activities. When comfortably affixed to a speaker’s throat, it can capture delicate and intricate weak force signals, such as muscle movements and epidermal deformations during speech. In Fig. 8d, the output voltage curve is depicted as the word "Hello" is spoken five times in various languages. Consistent characteristic peaks are recorded for each pronunciation method, demonstrating that the sensor exhibits high sensitivity and reliability for speech recognition. This capability can be utilized for rehabilitating the speech abilities of individuals with damaged vocal cords. Additionally, Fig. 8e displays the real-time pulse signal when the sensor is attached to the wrist. In addition to detecting a wrist pulse frequency of 65 beats per minute, it's essential to note that, due to the presence of ultra-fine nanofibers, the single signal waveform of the wrist pulse reveals two distinct peaks, P1 and P2 [55]. Among these, P1 represents the pulse wave, while P2 corresponds to the reflected wave from the hand. Based on this parameter, the arterial augmentation index (AIr = P2/P1) can be computed, with an average value of 0.448, which aligns with that of a healthy 26-year-old woman [56, 57]. This pulse information can be harnessed for hypertension and cardiovascular disease detection. The MPD-1, with its exceptional sensitivity to weak force signals, holds significant potential in the realms of health monitoring and clinical diagnosis.

Additionally, with its exceptional sensitivity to weak force signals, the MPD-1 sensor is proficient in discerning variations in pressure, including those induced by finger, elbow movements, and foot joint motions. As exemplified in Fig. 8f, when the MPD-1 sensor is directly affixed to the fingertip, it can accurately capture real-time tapping actions across different contexts. Gentle taps yield an output voltage of 3.4 V, while forceful strikes result in an increased output voltage of 7.5 V. For the above sensing applications, the finger motion amplitude and speed can be extracted by decoupling the force and frequency dependences of the electric signal. Meanwhile, when the sensor is affixed to the elbow, as depicted in Fig. 8g, it can detect the bending of the elbow from a straight position to 45° and 90°, generating peak voltages of 7.1 V and 9.8 V, respectively. The positive and negative voltage values correspond to the bending [58] and angle of the elbow, respectively. Similarly, when the sensors are attached to corresponding body parts, such as the knee and foot, they can detect bending and movement, as shown in Fig. 8h–i. Therefore, the MPD-1 sensor exhibits rapid response and can be employed for motion capture and tactile perception, presenting potential applications in remote human–computer interaction and control.

To further, owing to the high piezoelectric performance of the MPD-1, it can be integrated into piezoelectric nanogenerator systems. When connected to a designated module, as delineated in Fig. 9a, and leveraging the internal rectification circuit, the MPD-1 alternating current (AC) output is efficiently converted into direct current (DC). This allows for the illumination of small LED lights through tapping actions on the MPD-1. It is worth noting that distinct tapping intensities yield varying numbers of illuminated LED lights, as exemplified in Fig. 9b and supplementary materials. In fact, under vigorous tapping, it can simultaneously power more than 20 LED bulbs (Movie S1 in Supporting Information).

Fig. 9
figure 9

a Circuit diagram for capacitor charging and discharging to lighten a LED lamp and b the number of LED bulbs lit under different tapping strengths. c Charging curves of 33 μF capacitors

Full size image

Moreover, the 33 μF capacitor within the module can be charged, reaching a voltage of 3 V in 280 s at a tapping frequency of 2 Hz, as illustrated in Fig. 9c. After the charging process, the timer can be activated through capacitor discharge via a switch within the module (Movie S2 in Supporting Information). Given the MPD-1’s impressive power generation capacity, particularly considering its effective working area of 2.5 × 2.5 cm2 square centimeters, it underscores its potential as a sustainable power source for portable electronic devices.

Conclusions

In summary, the electrospinning technique was effectively utilized to fabricate a composite nanofiber membrane with a coherent and uniformly dispersed two-dimensional network topology composed of PVDF/DA nanofiber membranes and ultrafine PVDF/DA nanofibers. This innovative configuration lays the foundation for the development of high-performance PVDF piezoelectric sensors. Initially, the PVDF/DA nanofiber membrane with a high β-phase content (89%) was prepared, which provides a suitable base material with excellent piezoelectric properties. A complete and homogeneous two-dimensional topology composed of ultrafine PVDF/DA nanofibers was successfully constructed. Experimental data and theoretical modeling analysis have demonstrated that the unique structure of the composite nanofiber membrane effectively enhances the material piezoelectric performance. Under a force of 1.5 N, the output voltage reaches 14.3 ± 0.2 V, while the output current reaches 1.01 ± 0.1 μA. Furthermore, the existence of this hierarchical structure enables systematic voltage fluctuations under varying impact frequencies, aiding in signal feedback amidst frequency changes. Additionally, due to the close distribution of DA and ultrafine nanofibers within the pores of electrospun PVDF/DA nanofibers, the material maintains the voltage stability even after approximately 10,000 cycles, demonstrating excellent durability. Benefiting from its exceptional piezoelectric performance, the sensor exhibits good sensitivity in the range of 0–4 N, 5–10 N and 30–40 N, which are 7.29 V N−1, 4.11 V N−1 and 1.41 V N−1. As a result, it can be utilized for detecting various pressure stimuli in the human body, including speech, pulse, and movement. Additionally, we further demonstrate its potential applications for energy harvesting devices. Overall, due to the simple fabrication and high piezoelectric performance, we believe that the present work would provide a beneficial idea for the construction of PVDF-nanofiber membranes based piezoelectric sensors for various wearable electronic products.