Introduction

Skin, the human body’s largest organ, harbors numerous secrets within its distinctive texture1,2,3,4,5,6, sweat gland distribution7, and biological properties1, awaiting precise capture and analysis2,8. For instance, fingerprints, as the most commonly used feature of the skin, are stable and unique for each individual due to the interplay of genetic and environmental factors during fetal development5,6. This makes them an excellent tool for personal identification9, crime scene investigations7,10,11,12, and security applications12. Thus, analytical techniques have been developed based on Raman spectroscopy13, mass spectrometry14, Fourier transform infrared spectroscopy (FTIR)15, luminescence imaging16,17,18 and other techniques19. Among the materials applied in these imaging techniques, luminescent materials such as carbon dots20,21, perovskite quantum dots22, aggregation-induced emissive materials23 and organic dyes24,25 exhibit good imaging performance. However, on-site analysis requires exceptional stability, accuracy, and safety imaging materials, posing significant challenges for technological advancements26. Furthermore, traditional skin imaging techniques are constrained by hardware limitations27, lens quality, image processing algorithms28 and shooting environments29,30,31,32, thereby impeding the attainment of high-resolution images of micrometer-scale skin textures9,10,14. Thus, a universal high-resolution skin imaging technique combined with a data analysis system for identity verification, forensic science and criminal investigations has not yet been fully developed.

To address these issues, we have developed a low-cost, portable nanofiber-based imaging technique (NFIT) for human skin texture photocopying, using CsPbBr3@HPβCD luminescent nanofibers. This NFIT facilitates in situ, multiregional imaging of skin microsurface morphology by harnessing a unique wavelength-dependent photoluminescence mapping technique, utilizing trace amounts of sweat residue. When benchmarked against real fingerprint tertiary feature data, the NFIT exhibits a remarkable micron-scale imaging similarity of 93.24 ± 4.6%, coupled with high-resolution fingerprint imaging (1450 dpi). This underscores its exceptional proficiency in capturing intricate skin texture details, ranging from the minute structure of individual sweat glands to extensive textures across the palm, sole, abdomen and dorsum of the hand. Notably, the NFIT adopts a noncontact imaging paradigm, eliminating chemical pre-/post-treatment requirements. This attribute ensures compatibility with DNA evidence collection while minimizing chemical interference, thereby presenting a dependable and convenient tool for forensic and security applications. Moreover, NFIT maintains high-definition imaging across a broad temperature range of –50 °C to +50 °C and delivers excellent long-term imaging stability (level 3 ≥ 81 days, level 2 ≥ 108 days). This robustness ensures reliable operation under diverse harsh conditions, offering robust support for on-site analysis.

Beyond its high precision, low cost (<$800), and portability (<400 grams), the NFIT’s extensive applicability enables users to extract and photocopy skin texture images in real-time, irrespective of location or time, thereby supporting 3D skin texture model reconstruction. Importantly, the entire digitization process necessitates minimal computational resources33, thereby reducing the complexity and cost associated with system development and maintenance. This functionality significantly enhances operational efficiency and brings unparalleled convenience and substantial contributions to high-precision 3D portrait construction, dermatology, regenerative medicine, forensic analysis, and other pertinent fields.

Results

Fingerprint imaging process and morphological characterization

CsPbBr3@HPβCD nanocrystals were synthesized via grinding (see details in the “Methods” section). In this study, amphiphilic thermoplastic polyurethane (TPU) was selected as the electrospinning matrix because of its lypohydrophilic character compared with polystyrene (Supplementary Fig. 1). This characteristic ensures reliable extraction of fingerprint residues, primarily consisting of sweat and sebum, and potentially enhances the luminescence response. The CsPbBr3@HPβCD nanocrystals were then blended with TPU to form a mixed solution prepared as an electrospinning ink. Consequently, a handheld mini-electrospinning device (ca. 10 cm in length) was used to spray CsPbBr3@HPβCD fibers onto the sample to form luminescent nanofibers for imaging. Herein, fingerprints were employed as a representative of detailed skin textures to examine the imaging capability of NFIT. As shown in Fig. 1a and Supplementary Movie 1, high-definition Level 3 fingerprint imaging can be obtained efficiently within 10 s. The Level 3 fingerprints offer the most intricate details, such as ridge shapes, pore positions, incipient ridges and scars5. To evaluate the accuracy of NFIT, we captured a fingerprint luminescence image via NFIT and compared it to the same zone of an actual fingerprint photo for the same individual (Fig. 1b). The luminescence image clearly revealed the sweat pore features, including the pore size, shape, location, distribution, frequency, and pore-to-pore interspacing (Fig. 1c). Overlaying Fig. 1b with 1c, NFIT evidently allows high-precision visualization of Level 3 fingerprint features, with a matching similarity of over 93.24 ± 4.6% compared to the real fingerprint (Fig. 1d, Supplementary Figs. 2, 3 and Supplementary Table 1), which is the highest accuracy as reported for sweat pore imaging to date16,19,23,24. Notably, unlike traditional methods in which sweat pore information is indirectly obtained after epidermal imaging, NFIT both directly responds to sweat pore and epidermal texture characteristics, providing a more precise and direct representation (Supplementary Fig. 4). Additionally, we compared the fingerprint images extracted by NFIT with those obtained via traditional ink-based fingerprinting and commercial capacitive fingerprint sensors. It can be seen that NFIT produces higher-quality fingerprint images with more abundant fingerprint features (Supplementary Fig. 5). Notably, NFIT employs a fully noncontact approach throughout the skin texture extraction process, allowing users to press their skin on any substrate and lift it without any contact with chemical reagents. This ensures both chemical safety and hygiene. In addition, the operational safety of the imaging process was thoroughly evaluated. The exposure dose per fingerprint area (ca. 4 cm2) was limited to DMF ≤ 4 μl (3.8 mg) and DMSO ≤ 0.48 μl (0.5 mg) (Supplementary Table 2), which are well below the 8-h weighted average exposure limits set by the Occupational Safety and Health Administration (OSHA)34. ICP-MS analysis revealed that Pb concentration leached during each fingerprint imaging process was 0.003 mg/L (Supplementary Table 3). This value is significantly lower than the 0.01 mg/L drinking water standard in the Guidelines for Drinking-water Quality published by the World Health Organization (WHO)35, indicating negligible risk to human health36. To elucidate the mechanism of the CsPbBr3@HPβCD nanofibers’ response to sweat residues, we investigated their morphology via scanning electron microscopy (SEM), high-resolution transmission electron microscopy (HRTEM) and energy dispersive spectroscopy (EDS) mapping. The TPU polymer forms a fibrous scaffold that effectively holds the CsPbBr3@HPβCD nanocrystals (Supplementary Figs. 6, 7). SEM images of real fingerprints and latent fingerprints (LFPs) obtained via NFIT are shown in Supplementary Fig. 8. In the absence of CsPbBr3@HPβCD nanofibers, the LFPs exhibit distinct regional contrast between the light and dark areas (Supplementary Fig. 6a). EDS mapping reveals that the darker regions correspond to organic components primarily composed of carbon, which is predominantly concentrated in the fingerprint ridges, except for at the sweat pores (Supplementary Fig. 9). In contrast, after electrospinning, the CsPbBr3@HPβCD nanofibers (diameter ~27.8 ± 10.2 nm) are uniformly distributed across the entire surface, without preferential accumulation in the luminescence-intense regions, namely sweat pores and ridges (Supplementary Fig. 8b–d). The crystalline nature of the CsPbBr3@HPβCD on TPU fibers was confirmed by HRTEM (Supplementary Fig. 8e, f). The crystalline regions display distinct lattice fringes, with an interplanar spacing of 0.210 nm, corresponding to the (110) crystal plane towards highly crystalline CsPbBr337. Excitingly, EDS analysis demonstrates that NFIT not only accurately maps the real fingerprint features but also reveals the sodium chloride (NaCl) distribution in the fingerprint sweat (Fig. 1e–g, and Supplementary Figs. 10–12), represented as the dark regions in the luminescence image (Fig. 1c). In contrast, in the EDS mapping of the corresponding fingerprint residues without CsPbBr3@HPβCD fibers, the Na and Cl contents are below the detection limit (Supplementary Fig. 9). We hypothesize that the capillary action of the CsPbBr3@HPβCD nanofibers allows rapid adsorption and enrichment of NaCl from sweat. To validate this hypothesis, we sprayed a sodium chloride aqueous solution onto the surface of the CsPbBr3@HPβCD fibers. TEM and EDS reveal Na and Cl distributions along the fibers, confirming that NaCl can be adsorbed onto the surface of the CsPbBr3@HPβCD nanofibers (Supplementary Fig. 13). Furthermore, the addition of NaCl significantly affects the crystalline quality of CsPbBr3 (Supplementary Fig. 14), potentially quenching its luminescence. The selective response to fingerprint residues enables the distinction between real and forged fingerprints, thus enhancing the security and reliability of fingerprint recognition. For example, when we prepared fake fingerprints using silicone molds, NFIT did not reveal any significant fingerprint features (Supplementary Fig. 15). Additionally, this noncontact approach addresses several issues associated with touch-based fingerprint collection, such as the nonlinear distortion caused by pressing fingers onto sensor plates and the need for regular sensor surface cleaning38. More importantly, NFIT achieves fingerprint image resolutions of up to 1450 dpi (Supplementary Fig. 16) without soaking, heating, or adding reagents for pre- or post-treatment. Furthermore, the fingerprint features obtained by NFIT belong to Level 3. This significantly improves the precision of fingerprint feature capture and recognition, offering promising prospects for practical applications.

Fig. 1: Schematic illustration of the fingerprint imaging process and morphological characterization.
Fig. 1: Schematic illustration of the fingerprint imaging process and morphological characterization.
Full size image

a Schematic diagram of the preparation of CsPbBr3@HPβCD nanofibers on LFPs. b Image of a real fingerprint. c Luminescence image obtained via NFIT for a real fingerprint marked with Level 3 fingerprint characteristics. d Overlay of the fingerprint luminescence image obtained via NFIT and the real fingerprint image. e SEM image of the LFP with electrospun CsPbBr3@HPβCD nanofibers. f Magnified SEM image of the LFP with electrospun CsPbBr3@HPβCD nanofibers. g SEM-EDS mapping of the distribution of sodium.

Selective spectral response and chemical interactions of CsPbBr3@HPβCD nanofibers to sweat components

Given that CsPbBr3@HPβCD nanofibers enable high-definition fingerprint imaging through a selective response to sweat residues, an in-depth study of their spectral response is essential. HPβCD enhances the photoluminescence (PL) intensity of CsPbBr3 (Supplementary Fig. 17). HRTEM images reveal that the CsPbBr3 nanocrystals without HPβCD exhibit irregular morphologies (Supplementary Fig. 18). In contrast, the addition of HPβCD improves the crystal quality of the CsPbBr3 nanocrystals (Supplementary Fig. 8e, f), which was further confirmed by X-ray diffraction results (Supplementary Fig. 19). This finding suggests that HPβCD molecules effectively passivate the defects in CsPbBr3, leading to enhanced PL performance. The luminescence property of the CsPbBr3@HPβCD fibers, coupled with their specific interactions with NaCl, prompted us to explore their potential for selective response to fingerprint residues.

Sweat residues are primarily composed of sweat and sebum. Among these components, sweat contains various chlorides, such as sodium chloride (NaCl), magnesium chloride (MgCl2) and potassium chloride (KCl), along with organic compounds, such as glucose, ascorbic acid, uric acid and urea39,40,41,42. Sebum, in contrast, primarily comprises triglycerides, squalene and other lipid-based compounds43. Therefore, we performed spectroscopic analysis to investigate the response mechanism of the CsPbBr3@HPβCD nanofibers to representative components (NaCl, KCl, MgCl2, laurostearin, ascorbic acid, urea, uric acid and squalene) of sweat residues. The ultraviolet-visible absorption (UV-Vis) spectra indicate that the addition of various components has little effect on the bandgap of the CsPbBr3 fibers without HPβCD. However, in the presence of HPβCD, the bandgap of the CsPbBr3@HPβCD fibers is affected to varying degrees by these components (Supplementary Fig. 20a, b). Correspondingly, the PL intensities of the CsPbBr3 fibers and CsPbBr3@HPβCD fibers exposed to the above components were characterized (Supplementary Fig. 20c, d). The patterns of their PL peak shifts are consistent with those for the UV-Vis spectra. A decrease in intensity accompanies this blueshift of the PL peaks. Owing to the good color purity of CsPbBr3@HPβCD (full width at half maximum = 20.86 nm), the color change upon the addition of chloride salts is highly noticeable and can be distinguished by the naked eye (Supplementary Fig. 21). In comparison, the CsPbBr3 fibers without HPβCD typically exhibit PL quenching and minimal PL peak shifts upon the addition of chlorides (Supplementary Fig. 22). Furthermore, exposure of the CsPbBr3 fibers to organic components (laurostearin, ascorbic acid, urea, uric acid and squalene) generally results in an increase or decrease in the PL intensity without regular changes (Supplementary Fig. 23). In contrast, for the CsPbBr3@HPβCD fibers, almost all, the aforementioned organic components cause a decrease in the PL intensity. Interestingly, glucose is an exception: trace amounts of glucose (mass ratio of glucose: CsPbBr3@HPβCD = 6 × 10⁻5: 1) increase the PL intensity (Supplementary Fig. 24). When the concentration is increased to 50 times, glucose still produces a detectable PL intensity. The above PL performance suggests that HPβCD plays a dual role: it enhances the interaction of CsPbBr3 with ionic species and selectively modulates the response to specific organic components, which facilitates the analysis of extracted fingerprints.

Considering the individual differences in sweat composition that could affect the specific response of the CsPbBr3@HPβCD nanofibers for imaging, we systematically analyzed the concentration-dependent PL intensity responses for nine fingerprint residue components (Supplementary Figs. 25–27 and Supplementary Tables 4–12). The relationship between these representative components and the PL peak shifts of the CsPbBr3 @HPβCD fibers is summarized in the box plot in Fig. 2a and Supplementary Table 13. The data indicate that introducing NaCl, KCl, and MgCl2 causes a blueshift in the PL peak, accompanied by a reduction in the PL intensity. In contrast, upon exposure to organic compounds, their influence on the PL peak wavelength of the CsPbBr3@HPβCD fibers is negligible, whereas a general decrease in the PL intensity was observed. Glucose, as an exception, increases the PL intensity at low concentrations but causes PL quenching at higher concentrations. The luminescence properties of the CsPbBr3@HPβCD fiber response to sweat residue components can be better understood via time-resolved photoluminescence (TRPL) spectra. As shown in Fig. 2b and Supplementary Table 14, glucose is the only component that significantly extends the average PL lifetime, whereas the other components reduce the lifetime to varying degrees.

Fig. 2: Spectral response and chemical interactions of CsPbBr3@HPβCD to sweat residue components.
Fig. 2: Spectral response and chemical interactions of CsPbBr3@HPβCD to sweat residue components.
Full size image

a Box plots of the dependence of the PL peak shift in CsPbBr3@HPβCD fibers on the concentration variation of sweat residue. b TRPL spectra of the CsPbBr3@HPβCD fibers exposed to sweat components. c XPS spectra of Pb 4f for CsPbBr3@HPβCD fibers with the sweat residue. ATR-FTIR spectra of CsPbBr3@HPβCD with the addition of d chloride and e glucose. f Adsorption energy simulation model of CsPbBr3@HPβCD with the addition of chloride. g Illustration of the response mechanism of CsPbBr3@HPβCD fibers to sweat residues.

To further investigate the chemical interactions between CsPbBr3@HPβCD and fingerprint residues, X-ray photoelectron spectroscopy (XPS), attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) and 1H nuclear magnetic resonance (1H-NMR) spectroscopy were used. Compared with pristine CsPbBr3 fibers, introducing various sweat components results in the Cs 3d, Pb 4f, and Br 3d peaks shifting to lower binding energies (Supplementary Figs. 28–30). In contrast, CsPbBr3@HPβCD exhibits a more systematic specific response upon exposure to fingerprint residues. For example, when ascorbic acid, laurostearin and squalene are introduced to CsPbBr3 @HPβCD fibers, the binding energy of the Pb 4f peak further decreases by 0.40, 0.50 and 0.40 eV, respectively, compared with that of pure CsPbBr3 fibers. This shift suggests a reduction in the electron affinity and an increase in the electron cloud density around the Cs, Pb, and Br atoms, which can be attributed to electron donation from the additive molecules, likely forming coordination bonds36. Conversely, the introduction of various chlorides to CsPbBr3@HPβCD fibers causes the Pb 4f and Br 3d peaks to shift to higher binding energies (Fig. 2c and Supplementary Fig. 31). This phenomenon may be ascribed to the partial incorporation of chloride ions (Cl⁻) into the CsPbBr3 lattice, facilitated by the presence of HPβCD (Supplementary Fig. 20)44. Interestingly, the introduction of glucose to CsPbBr3@HPβCD increases the binding energies of the Pb 4f and Br 3d peaks (Supplementary Tables 15, 16). This finding may be attributed to glucose synergistically strengthening hydrogen bonding interactions with CsPbBr3 through multiple hydroxyl groups45 and HPβCD. As a result, the local environment of CsPbBr3 is altered, and its electron density decreases. The above XPS results indicate that HPβCD plays a critical role in modulating the chemical interactions between CsPbBr3@HPβCD and fingerprint residues, particularly upon exposure to glucose and chlorides, thereby altering the photophysical properties of the material.

Owing to the ionic nature of CsPbBr3, which arises from the strong ionic interactions inherent to its crystal structure, the material exhibits distinctive characteristics when interacting with other species46. The incorporation of HPβCD further enhances these interactions, as its polar internal cavity and hydrophilic exterior provide an ideal microenvironment for interactions with metal ions and stabilizing agents47. Specifically, the ionic species interact with the hydroxyl and other functional groups on the HPβCD surface, altering the local electron density. This interaction, in turn, may affect the shielding of nearby protons, as reflected in the chemical shifts observed via 1H-NMR spectroscopy. The introduction of CsPbBr3 primarily causes more significant shifts in the exterior protons of HPβCD (H1, H2 and H7) than in the inner cavity protons (H4 and H6), indicating that the interaction predominantly occurs on the external surface of HPβCD48. Furthermore, the addition of chloride salts (NaCl, KCl, and MgCl2) to CsPbBr3@HPβCD results in distinct shifts in the exterior protons of HPβCD (Supplementary Tables 17, 18 and Supplementary Fig. 32). When chloride salts compete with CsPbBr3 for active sites on the HPβCD surface, the strong electronegativity of chloride ions may disrupt the original electrostatic field, leading to changes in the chemical environment of the exterior protons. Moreover, HPβCD can assist in capturing chloride ions from the system, thus promoting halogen exchange between the chloride ions and the perovskite. In contrast, when glucose is introduced to the CsPbBr3@HPβCD complex, the chemical shifts of the exterior protons remain largely unchanged, indicating that glucose does not compete with CsPbBr3 for the active binding sites of HPβCD. The multiple hydroxyl groups of glucose may provide additional active sites to stabilize CsPbBr3, further suppressing the exposure of CsPbBr3 defects41,49.

Given the above findings, we investigated the interactions between chlorides, glucose and CsPbBr3@HPβCD via ATR-FTIR spectroscopy. When CsPbBr3 is added to HPβCD, the -OH stretching frequency of HPβCD slightly increases, shifting from 3343 cm−1 to 3350 cm−1. Upon the addition of NaCl to CsPbBr3@HPβCD, the -OH stretching frequency increases dramatically to 3398 cm−1, suggesting that NaCl has a significant influence on the CsPbBr3@HPβCD complex (Fig. 2d). Similarly, the introduction of KCl and MgCl2 leads to an increase in the intensity and a blueshift of the -OH vibrational band (~3343 cm−1) of HPβCD, accompanied by a narrowing of the peak width. This phenomenon may arise from the alteration of the hydrogen bonding network of HPβCD by chloride salts through electrostatic interactions50, which affects the electron density distribution of HPβCD51,52. The impact of glucose can be observed in its interaction with CsPbBr3 (Fig. 2e), in which the -OH stretching frequency of glucose increases from 3243 cm−1 to 3264 cm−1 upon the introduction of CsPbBr3. This result suggests the formation of hydrogen bonds between the CsPbBr3 and -OH groups. Similarly, the introduction of CsPbBr3 causes a blueshift in the -OH stretching band of HPβCD, from 3343 cm−1 to 3350 cm−1, indicating a similar interaction of HPβCD with CsPbBr3 and glucose with CsPbBr3. The characteristic peaks of CsPbBr3@HPβCD and glucose@HPβCD are almost identical. However, the –OH stretching band of the CsPbBr3@HPβCD/glucose complex blueshifts from 3343 cm−1 to 3318 cm−1, accompanied by peak width broadening. This is likely due to the addition of glucose expanding the hydrogen bonding network of HPβCD and further enhancing the synergistic interaction between HPβCD and glucose acting on CsPbBr3.

To further elucidate the molecular-level interactions of the CsPbBr3@HPβCD fibers with NaCl, KCl, and MgCl2, we performed density functional theory (DFT) calculations53,54,55. First, we investigated Na+, K+, and Mg2+ doping at the Cs+ and Pb2+ sites, along with halogen exchange, and their effects on the CsPbBr3 bandgap (Supplementary Tables 19–21). The results indicate that doping Na+ and K+ at the Pb2+ site, and Cl⁻ substituted by Br⁻, produces a bandgap that closely matches the experimental results. This observation can be attributed to the ionic radii of Na+ (1.02 Å) and K+ (1.38 Å) being closer to that of Pb2+ (1.19 Å) than the ionic radius of Cs+ (1.67 Å), making them better suited for doping at Pb2+ sites56,57. Additionally, doping at the Pb2+ site introduces local charge imbalances, which are compensated by lattice defect mechanisms, thereby maintaining the overall material stability. Conversely, Cs+, as a large alkali metal ion, occupies the A-site in the octahedral perovskite structure, where it exhibits high chemical stability. Replacing Cs+ with smaller ions such as Na+ and K+ often destabilizes the structure, leading to collapse58. However, based on the aforementioned UV-Vis spectra, the transient surface adsorption of Na+ and K+ ions, along with halogen exchange, has a minimal effect on the bandgap of CsPbBr3, suggesting that such processes are unlikely to occur without additional external treatment. Compared to CsPbBr3, the CsPbBr3@HPβCD fibers demonstrate a stronger response to NaCl, KCl, and MgCl2, suggesting that HPβCD plays a crucial role in this system. We calculated the adsorption energies (Eabs) with and without HPβCD, revealing that HPβCD enhances the Eabs of NaCl (–6.42 eV), KCl (–6.64 eV), and MgCl2 (–6.01 eV) to –7.27 eV, –7.16 eV, and –7.14 eV, respectively (Fig. 2f and Supplementary Table 22)46. The enhanced Eabs values indicate stronger interactions between chloride salts and the CsPbBr3@HPβCD complex, particularly through Na+ and K+ doping at the Pb2+ site and halogen exchange, significantly influencing the PL properties. For comparison, we synthesized CsPbBr3@HPβCD, CsPbBr3@K-βCD, and CsPbBr3@Cl-βCD fibers, using HPβCD, potassium-enriched β-cyclodextrin (K-βCD), and chloride-enriched β-cyclodextrin (Cl-βCD), respectively (see “Methods” for details). The results further confirm that the halide exchange effect is more prominent as the emission peak of CsPbBr3@Cl-βCD blueshifts to ~475 nm (Supplementary Fig. 33).

The above computational and experimental results demonstrate that ion-dipole or electrostatic interactions between HPβCD and inorganic chloride salts facilitate halide exchange with CsPbBr3. In the case of glucose, the coexisting HPβCD and glucose may synergistically interact with surface defect sites on CsPbBr3, potentially inhibiting exposure of CsPbBr3 and thereby enhancing its fluorescence lifetime. Conversely, organic components other than glucose may undergo stronger interactions with HPβCD, acting as guests that compete with CsPbBr3. This competition could increase the exposure of CsPbBr3, leading to further quenching of the PL emission. The differential response of CsPbBr3@HPβCD fibers to fingerprint residue components is critical for achieving high-precision fiber-based imaging, as illustrated in the mechanism diagram (Fig. 2f).

Photoluminescence characterization and in situ imaging

Building on the above comprehensive characterization, we further conducted an in-depth study on the effects of NaCl, KCl, MgCl2 and glucose on the photophysical properties of CsPbBr3@HPβCD fibers via temperature-dependent PL spectroscopy, in situ PL mapping, and in situ fluorescence lifetime imaging microscopy (FLIM). During the temperature-dependent PL analysis, the PL intensities of the CsPbBr3 fibers, CsPbBr3@HPβCD fibers, and CsPbBr3@HPβCD fibers with added NaCl or glucose progressively decrease with increasing temperature (Fig. 3a and Supplementary Fig. 34). The reduction in the PL intensity can be attributed to the increased thermal energy, which induces exciton dissociation and promotes electron-phonon interactions, resulting in substantial PL quenching59,60. Notably, the temperature-dependent PL intensity of CsPbBr3@HPβCD/NaCl shows a more pronounced decline than that of CsPbBr3@HPβCD (Supplementary Fig. 35), which may be related to the lattice phase transition of CsPbBr3, the loss of surface HPβCD during heating61 and the defect formation induced by NaCl doping. The PL intensity of CsPbBr3@HPβCD/glucose exhibits a more gradual decline with increasing temperature, maintaining a detectable PL signal even at 350 K (Supplementary Fig. 34). Based on the earlier mechanistic investigations, we infer that glucose contributes to passivation effects, mitigating exciton dissociation and electron-phonon interactions under high thermal stress, thereby increasing the stability of CsPbBr3 perovskites. The exciton binding energy (Eb) was calculated from the Arrhenius plot via the following equation62:

$$I\left(T\right)=\frac{{I}_{0}}{1+A\exp \left(\frac{{E}_{b}}{{{K}_{B}}^{T}}\right)}$$
(1)

where I0 is the integrated PL intensity at 0 K, KB is the Boltzmann constant, and A is the Arrhenius coefficient. The CsPbBr3@HPβCD/glucose fibers exhibit the highest Eb (74.9 meV) compared to CsPbBr3 fibers (50.8 meV) and CsPbBr3@HPβCD fibers (63.4 meV) (Supplementary Fig. 36), which suggests a more effective inhibition of exciton dissociation due to the synergistic effects of glucose and HPβCD.

Fig. 3: PL characterization of and in situ fingerprint PL imaging with CsPbBr3@HPβCD nanofibers.
Fig. 3: PL characterization of and in situ fingerprint PL imaging with CsPbBr3@HPβCD nanofibers.
Full size image

a Temperature-dependent PL of CsPbBr3, CsPbBr3 @HPβCD, CsPbBr3@HPβCD/glucose and CsPbBr3@HPβCD/NaCl fibers. b Bright field image and peak wavelength distribution for CsPbBr3@HPβCD fiber fingerprint imaging, scale bar: 500 µm. c Wavelength-dependent PL mapping at 520 nm, 510 nm, 500 nm and 490 nm of CsPbBr3@HPβCD fibers in response to fingerprint residues, scale bar: 500 µm. d FLIM at 520 nm and 490 nm, scale bar: 1 mm.

Given the impact of chlorides and glucose on the emission wavelength and lifetime of CsPbBr3@HPβCD fibers, we further employed in situ PL mapping to investigate their potential for imaging the spatial distribution of specific components. Wavelength-dependent PL mapping reveals that the emission peaks of the CsPbBr3@HPβCD fibers on LFPs range from 490 nm to 520 nm, with distinct regional characteristics (Fig. 3b, c). The PL emission at 520 nm is predominantly concentrated in the furrows and sweat pores of the fingerprint. The distribution of the 500 nm luminescence accurately outlines the edges of ridges and pores. In comparison, the peak wavelength observed at the ridges surrounding the sweat pore, which corresponds to the region enriched with Na+ and Cl⁻, gradually blueshifts to 490 nm (Fig. 1g and Supplementary Figs. 10–12). Thus, FLIM was performed at PL emission wavelengths of 520 nm and 490 nm. Using the lifetime of the CsPbBr3@HPβCD fibers as a reference, a distinct spatial distribution of the relative lifetime was observed (Fig. 3d). At 520 nm, the long-lifetime regions are primarily concentrated around the pores and fingerprint valleys, corresponding to the areas of stronger emission. Conversely, chloride salts lead to a reduced lifetime, which is evident in the shorter lifetime of 490 nm compared to 520 nm. FLIM at various wavelengths provides additional details for Level 3 fingerprint imaging (Supplementary Fig. 37a–d). Consistent with the PL mapping and EDS results, FLIM indicates that chlorides are primarily distributed around the outer edge of the sweat pores on the ridges, where the emission undergoes a blueshift with a shortened lifetime. Regions with strong emission intensity and a relative lifetime exceeding 100% may provide clues for the distribution of glucose (Supplementary Figs. 37e, 38 and Table 23). Thus, FLIM can effectively visualize fingerprints and potentially provide glucose distribution. Due to precise measurements of sweat glucose concentrations can be utilized to estimate blood glucose levels63, there is potential for this NFIT to be integrated with sensing technology64 in the future, further promoting its application in biomedical diagnostics.

Evaluation of the universality, stability, and nondestructiveness of NFIT

In light of the high luminescence imaging performance of NFIT, we conducted skin texture extraction experiments under various environmental conditions to evaluate its stability, adaptability and broad applicability systematically. Specifically, clear and high-contrast visualization of LFPs on common substrates, including tinfoil, quartz, iron, glass and plastic (relatively clear Level-2 fingerprint imaging can be achieved on leather and paper), was achieved with a 10-second electrospinning process (Fig. 4a and Supplementary Figs. 39–41). These results demonstrate that NFIT can rapidly acquire high-resolution fingerprint information on a wide range of substrate surfaces, thus meeting the demands for swift, high-definition extraction of fingerprints on typical crime scene items (Supplementary Fig. 42). We further investigated the performance of NFIT in extracting Level 3 features under various finger pressures. As shown in Supplementary Figs. 43 and 44, an increase in pressure caused a reduction in the diameter of the sweat pores, consistent with previously reported results10. This finding suggests that NFIT can efficiently capture changes in the sweat pore diameter, providing insights into the force exerted by the individual and offering valuable forensic clues. The temporal stability and imaging consistency of NFIT in fingerprint extraction were also evaluated. Experiments demonstrate that even for aged LFPs left for 37 days, NFIT can still deliver clear Level 3 fingerprint features (Supplementary Figs. 45, 46). This result indicates that NFIT significantly outperforms traditional humidity-responsive fingerprint imaging techniques, which often struggle to extract complete Level 3 fingerprint information once the moisture in the fingerprint evaporates. Moreover, fingerprint imaging generated via CsPbBr3@HPβCD nanofibers exhibits impressive long-term stability (Level 3 ≥ 81 days, Level 2 ≥ 108 days); see Supplementary Fig. 47. The CsPbBr3@HPβCD nanofibers also demonstrate strong resistance to photodegradation. Under continuous high-frequency laser (20 MHz, 405 nm) irradiation at a single point, the PL intensity remains stable, with an estimated half-life of 158 min (Supplementary Fig. 48). The NFIT performance was further assessed under extreme environmental conditions with temperatures ranging from +50 °C to –50 °C. At PL mapping wavelengths of 520 nm and 490 nm, NFIT consistently captures clear Level 3 features (Fig. 4b, c). Combined with its portable design, it is suitable for on-site collection and analysis under extreme temperature conditions. Moreover, we expanded the extraction area from fingerprints to the entire palm to evaluate the ability of NFIT to capture the full ridge patterns. As shown in Fig. 4d, e, NFIT successfully captures whole palm print information, providing high-resolution details for sweat pores (Supplementary Fig. 49). This highlights the potential of NFIT for applications involving photocopying whole-body skin textures. In actual forensic and criminal investigation scenarios, using nondestructive techniques for DNA is crucial. Therefore, we investigated the impact of NFIT on DNA identification to ensure its applicability to nondestructive analysis. Based on short tandem repeat (STR) analysis of the data of extracted DNA with or without electrospun CsPbBr3@HPβCD nanofibers, NFIT has no negative effects on STR analysis, as the two samples yield consistent results (Supplementary Figs. 50, 51). This confirms the effectiveness of NFIT in extracting fingerprints without damaging DNA evidence. In summary, NFIT demonstrates adaptability across a wide range of extreme environments. Whether under the scorching heat of the Sahara or the freezing temperatures of polar regions, this technology can be easily applied to various substrate surfaces via a portable electrospinning device, enabling rapid, convenient and high-resolution onsite fingerprint imaging.

Fig. 4: Universality and stability characterization of NFIT.
Fig. 4: Universality and stability characterization of NFIT.
Full size image

a NFIT images of LFPs on various substrates, scale bar: 1 cm. Temperature-dependent PL mapping of CsPbBr3@HPβCD nanofiber-based fingerprint images at b 520 nm and c 490 nm. d Real palm image. e The palm imaging by NFIT.

Portable high-resolution fingerprint imaging and data analysis system

Leveraging the high-resolution imaging capability and broad applicability of NFIT, we self-developed a micro-image algorithm that integrates portable fingerprint imaging devices with rapid data analysis. This integrated system consists of a handheld mini-electrospinning device and a smartphone (Supplementary Fig. 52). The fingerprint analysis program can be embedded in a smartphone or a computer, which serves as the data processing platform (Fig. 5a and Supplementary Fig. 53). The high resolution of the extracted images ensures precise capture of fingerprint features. Coupled with the user-friendly program, the entire extraction and analysis process becomes highly efficient, allowing the collection and storage of 100 fingerprints within only 1 h. As a proof-of-concept, we used this system for fingerprint extraction and analysis, with the results shown in Fig. 5b. The images can then be imported into the program (see “Methods”) for image processing. This program can accurately extract grayscale values from multiple fingerprint regions within the image (Supplementary Fig. 54). The analysis regions can be user-defined, and the number of sampling points can be adjusted (Supplementary Movie 2). Unlike traditional methods that rely on “drawing lines to take points”, our integrated NFIT system directly converts large areas of image information into grayscale values. Additionally, depending on the number of sampling points, images of various resolutions can be generated (Supplementary Fig 55). Owing to the efficiency of the integrated NFIT system, extensive computational resources are not required, even when managing large datasets. For example, the data from the four charts in Fig. 5a occupy only 894 K bytes after processing by the integrated NFIT system. This substantial reduction in computational and storage demands significantly lowers operational complexity and maintenance costs. Simultaneously, the system converts digitized grayscale data into 2D/3D images, enabling swift visualization and comparison of fine fingerprint features.

Fig. 5: Integrated NFIT system for fingerprint data extraction and processing.
Fig. 5: Integrated NFIT system for fingerprint data extraction and processing.
Full size image

a Integrated NFIT system for fingerprint extraction and imaging. b Conversion and storage from fingerprint image to grayscale data. c Grayscale value distributions and the corresponding 3D models.

The integrated NFIT system can directly capture data and construct models from a photograph of a real fingerprint, allowing the obtainment of a 3D model that accurately aligns with the model generated through NFIT, thus validating the precision of the entire extraction and modeling process (Fig. 5c, Supplementary Fig. 56 and Supplementary Movie 3). Further comparisons reveal that the 3D reconstruction models obtained via NFIT provide more detailed fingerprint features, which is attributable to the high contrast achieved in fingerprint luminescence imaging (Supplementary Figs. 57–59). Additionally, the finger surface morphology can be reconstructed through mirroring inversion of the 3D fingerprint model obtained by the integrated NFIT system (Supplementary Fig. 60). This mirroring model provides clear information on the ridge patterns and more details, making it valuable for forensic identification and high-precision authentication applications. Thus, this integrated NFIT system provides comprehensive functionalities, including fingerprint extraction and imaging, image conversion and storage, data analysis and 3D model reconstruction. It has significant advantages over existing fingerprint collection techniques (Supplementary Table 24): (1) The device is compact, lightweight (<400 g) and easy to assemble, reducing the complexity and operational requirements compared with traditional laboratory instruments; (2) The entire system is low-cost (<$ 800, Supplementary Table 25), making it highly suitable for field operations and large-scale deployment; (3) The system is optimized for an efficient integrated workflow that encompasses fingerprint extraction, imaging, data storage, and analysis (Supplementary Movie 4); (4) The system employs a 3D reconstruction algorithm, thus providing high-precision features and a visually presented fingerprint morphology for forensic and identity authentication applications.

Integrated NFIT system for body skin photocopying and 3D model construction

Utilizing the NFIT system, we have achieved rapid, high-precision photocopying of skin textures from diverse body regions, including the forehead, abdomen, hand backs, and soles of the feet. This process consistently delivers high-definition skin texture images within just 5 min and it overcomes the challenges posed by areas with minimal sweat secretion, such as the hand back. Our system employs CsPbBr3@HPβCD as a luminescent imaging medium, uniquely responding to sweat secretions. By using an electrospinning mechanism, we enable swift and precise imaging without reliance on intricate feedback systems or chemical pre- or post-processing, ensuring operational simplicity and efficiency. Moreover, our NFIT system excels in generating 3D models from these high-resolution skin texture images (Fig. 6 and Supplementary Fig. 61). This capability extends even to regions like the hand back, where sweat secretion is limited, demonstrating the system’s exceptional sensitivity and adaptability for the whole human body skin. These 3D skin texture datasets have significantly enhanced the fidelity and detail optimization in 3D modeling and rendering. This advancement supports the creation of hyper-realistic works in game development, animation production, and film special effects, providing realism. Beyond creative applications, the system has propelled innovations in virtual try-on and personalized customization technologies. By enabling the accurate reproduction of individual skin textures, consumers can experience tailored services and convenience. Additionally, the system offers potential in medical diagnostics and health monitoring, providing a non-invasive approach to capturing skin textures for analysis of dermatological conditions, even in areas traditionally challenging for imaging. These advancements underscore the NFIT system as a transformative tool in high-resolution skin imaging and 3D modeling, redefining possibilities across industries.

Fig. 6: The workflow of the integrated NFIT system for body skin photocopying and 3D model construction.
Fig. 6: The workflow of the integrated NFIT system for body skin photocopying and 3D model construction.
Full size image

Integrated NFIT system for forehead, abdomen, hand backs, and soles of feet texture photocopying and 3D modeling, the scale bar: 1 cm.

Discussion

We developed a nanofiber-based imaging technique (NFIT) using CsPbBr3@HPβCD luminescent nanofibers, achieving real-time, multidimensional and multiregional imaging of human skin textures. The NFIT system utilizes unique chloride and glucose recognition mechanisms, enabling accurate sweat pore imaging with a similarity of 93.24 ± 4.6%. It supports large-area imaging across diverse materials while maintaining high-resolution (1450 dpi) image quality under extended storage (Level 3 ≥ 81 days, Level 2 ≥ 108 days) and extreme temperatures (–50 °C to +50 °C). Beyond capturing detailed 2D skin surface features, NFIT facilitates the construction of high-precision 3D skin morphology models. These data change biometric security systems, offering enhanced fingerprint and palm print recognition accuracy. Furthermore, the data serve as invaluable resources for educational and research purposes in biology and anthropology while advancing personalized medicine and dermatology. This technology sets a benchmark in human skin photocopying, addressing critical challenges in precision, safety, and usability and unlocking transformative applications across diverse fields.

Methods

Materials

Cesium bromide (CsBr, 99.9%) and lead bromide (PbBr2, 99.9%) were purchased from Advanced Election Technology Co., Ltd. The N, N-dimethylformamide (DMF, AR), and Dimethyl sulfoxide (DMSO, AR) were purchased from Sinopharm Chemical Reagent Co., Ltd. Hydroxypropyl-β-cyclodextrin (HPβCD) and deuterated dimethyl sulfoxide (DMSO–d6) were purchased from Shanghai Macklin Biochemical Co., Ltd. Thermoplastic Polyurethanes (TPU) was purchased from Qingdao Nuokang Environmental Protection Technology Co., Ltd. Sodium chloride (NaCl, AR), magnesium chloride (MgCl2, AR), potassium chloride (KCl, AR) and glucose were purchased from Sinopharm Chemical Reagent Co., Ltd., Lauro stearin (98%) and squalene (95%) were purchased from Shanghai Macklin Biochemical Co., Ltd. Ascorbic acid (99%) and uric acid (99%) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd.

Preparation of electrospinning solution and LFPs recognition process

CsBr (0.0213 g), PbBr2 (0.0367 g), and HPβCD (0.05 g) were combined and ground using a mortar and pestle for 1 min. DMF (0.25 ml) and DMSO (0.25 ml) were then added to the mixture, which was shaken and allowed to sit at room temperature for 1 h. Subsequently, 100 μl of this solution was mixed with 1 ml of a 16% TPU solution (the solvent is DMF) and stirred for 10 s to fabricate the electrospinning solution. The final solution was loaded into a handheld electrospinning device, allowing for the rapid generation of LFP images, with complete imaging achievable in approximately 10 s.

Preparation of K-βCD

To prepare K-βCD, a mixture of β-CD (0.68 g, 0.6 mmol) and KOH (0.269 g, 4.8 mmol) was dissolved in H2O (12 ml) in a beaker and sonicated for 5 min to ensure homogeneous dispersion. Methanol (12 ml) was subsequently added, and the solution was stirred continuously for 1 h. The mixture was then subjected to microwave irradiation at 100 W for 4 min and 30 s to facilitate crystal growth. Following this step, the reactant was carefully added dropwise to a large volume of hot methanol, which acted as a size-controlling agent. The resulting crystals were collected by centrifugation at 7000 rpm and washed three times with methanol. Finally, the crystals were dried under vacuum at 50 °C for 12 h. The dried product, K-βCD, was obtained by gentle grinding of the crystals.

Preparation of Cl-βCD

To synthesize Cl-βCD, a mixture of β-CD (1.9 g, 1.5 mmol) and imidazole (1.5 g, 22.5 mmol) was dissolved in DMF (60 ml) under a nitrogen atmosphere. Anhydrous phenylsulfonyl chloride (3.9 g, 22.5 mmol) was then added, and the reaction mixture was stirred at room temperature for 2 h. Subsequently, the reaction was heated to 70 °C and maintained at this temperature for 24 h. After completion, the solvent was removed by distillation, and 200 ml of water was added. The pH was adjusted by the addition of 2 mol L−1 NaOH solution until the solution was fully alkaline. The resulting mixture was stirred for an additional 2 h, and the precipitate was collected by filtration, washed thoroughly with H2O and acetone, and then recrystallized to yield purified Cl-βCD.

Characterization

A custom-built fluorescence microscopy system was employed to observe LFPs fluorescence imaging. The microscope used was a DO-BX53 (Olympus) equipped with a mercury lamp as the excitation source (300–500 nm) and ×2.5 – ×100 objective lens. The morphological images were captured using a Regulus 8230 (Hitachi) cold-field emission scanning electron microscope (SEM), which was equipped with a X flash 5060F (Bruker) detector for energy dispersive X-ray spectra and mapping (EDS). The high-resolution transition electron microscopy (HRTEM) images were captured with a Tecnai G2 F30 (S-TWIN) transition electron microscope at an accelerating voltage of 300 kV, which was equipped with an X Flash (STEM-HAADF) detector for EDS mapping and Gatan Ultrascan CCD camera for high-resolution imaging. The binding energy was characterized on an ESCALAB 250Xi+ (Thermo Fischer) X-ray Photoelectron Spectroscopy (XPS). Fourier-transform infrared (FTIR) spectra were obtained by Cary 610/670 (Agilent). The nuclear magnetic resonance (NMR) was performed on Quantum-I plus 600 (Oxford). A FLS1000 fluorescence spectrometer (Edinburgh Instruments) was used to test the PL spectra and TRPL. The temperature-dependent measurement was conducted with DN Optical Cryostats (Oxford Instruments) as an apparatus on FLS1000. Fluorescent lifetime imaging microscopy (FLIM) was performed on an ECLIPSE Ni-U microscope (Nikon) with a ×5 – ×10 objective lens, which was coupled with FLS1000 via optical fibers. A 405 nm picosecond pulsed diode laser was used as the excitation source for TRPL analysis and FLIM. The detailed parameters were as follows: a spatial resolution of 10 μm, a dwell time of 2 s per point, using the time-correlated single photon counting technique. PL mapping was performed on inVia Qontor (REINISHAW) with a ×40 objective lens. The sample was excited by a 325 nm laser, and the spatial resolution was 10 μm. Temperature-dependent PL mapping was conducted with a THMS600 temperature control stage equipped on inVia Qontor. The solid-state UV-Vis absorption spectra were conducted on Shimadzu with an integrating sphere. The LFP was spun using a handheld electrospinning device of E-01 (Foshan Qingzi Precision Measurement and Control Technology Co., Ltd).

Evaluation of the effect of CsPbBr3@HPβCD electrospinning treatment on the identification of DNA

Blood samples were collected from a volunteer. Samples were prepared by depositing blood drops onto filter paper, with or without electrospun CsPbBr3@HPβCD nanofibers. DNA was extracted from these blood samples using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). The final concentration of DNA was 3 ng per PCR. STR analysis was performed using the Goldeneye DNA Identification System 20 A (PeopleSpot, Beijing, China). The PCR products were analyzed using ABI 3130xl Genetic Analyzer (Applied Biosystems). Fluorescence was quantified, and the precise size of the DNA fragments was calculated with Genemapper software 3.2 (Applied Biosystems).

Computational details

First-principles density functional theory (DFT) calculations were performed using the Vienna Ab initio Simulation Package (VASP)47 with the projector augmented wave (PAW) method48. The exchange-functional was treated within the generalized gradient approximation (GGA) employing the Perdew-Burke-Ernzerhof (PBE) functional49. The long-range van der Waals interactions are accounted for through the DFT-D3 approach. A plane wave basis set with an energy cutoff of 500 eV was employed, and the geometry relaxation was performed until the forces on each atom were below 0.03 eV Å−1. The Brillouin zone was sampled using 1 × 1 × 1 k-point grid. Self-consistent calculations were conducted with an energy convergence threshold of 10−5 eV. A vacuum region of 15 Å was added along the z direction to prevent interactions between periodic structures.

Self-developed skin texture processing analysis program

The self-developed Level 3 fingerprint analysis program (see Supplementary Data 1) provides a comprehensive two-step data processing workflow, optimized for efficient analysis and visualization of high-resolution fingerprint data. In the first step, the software converts color images to grayscale, allowing for region-specific intensity analysis within user-defined regions of interest (ROI). The grayscale intensity data, including pixel coordinates and values, is exported as structured CSV or Excel files, ensuring compatibility with subsequent analysis stages. In the second step, this exported grayscale data is treated as point cloud input to generate detailed 2D grayscale images and 3D models, enhancing visualization and supporting interactive exploration of fingerprint features. The program’s graphical user interface (GUI) enables real-time processing and immediate feedback, improving usability across both analysis steps. Core implementation technologies include NumPy for efficient numerical computation, OpenCV and Pillow (PIL) for robust image manipulation, PyVista for 3D visualization, and Pandas for managing data exports. The program also employs Matplotlib to render 2D distribution charts and OpenPyXL to facilitate formatted Excel outputs with embedded visualizations. By ensuring smooth data flow between steps, the system minimizes manual data handling and enhances accuracy, reducing the risk of errors and enabling researchers to focus on analytical outcomes. This integrated approach supports a range of forensic and research applications, with potential future expansions to include automated data pipelines for even more efficient transitions from image processing to 3D visualization.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.