A sweat-responsive covalent organic framework film for material-based liveness detection and sweat pore analysis

Fingerprints and their level 3 features (e.g., sweat pore distribution) are widely used in personal identification. However, existing systems are vulnerable to attacks by artificial fake fingerprints, threatening safety and privacy. Prior liveness detection approaches predominantly rely on image processing and computer algorithms to discriminate live versus fake fingerprints, analyzing patterns and level 3 features post-collection. From a materials perspective, a simpler route is to develop a collection medium that inherently distinguishes living fingers from fakes at the time of contact. Because fake fingerprints cannot secrete sweat, a sweat-responsive material could naturally differentiate real from fake during collection. The authors propose dividing fingerprinting into collection and recognition steps and achieving liveness detection through a sweat-responsive covalent organic framework (COF) film that produces a visible pattern only when touched by a living finger, thereby enabling material-based liveness detection at the collection stage.

The paper situates its contribution against prior work in fingerprint liveness detection, which largely employs computational methods (e.g., multi-scale texture features, PCA, pore-based analysis) to detect presentation attacks after image acquisition. Level 3 features such as sweat pores are valuable for robust identification and have been explored using optical, fluorescence, mass spectrometry, and electrochemical methods that typically require instruments. Antispoofing surveys underscore ongoing challenges with fake fingerprints. Separately, covalent organic frameworks (COFs) are an active materials class with tunable porosity, crystallinity, and stimuli-responsiveness, previously applied in sensing, separations, and electronics. Leveraging COF solvatochromism/hydrochromism for biometric liveness detection represents a novel materials-based solution that complements and potentially simplifies existing software-centric approaches.

Synthesis and film preparation: COF_TPDA-TFPY powders were synthesized solvothermally at 120 °C for 72 h using o-dichlorobenzene (o-DCB) and n-butanol (n-BuOH) as solvents with acetic acid (HOAc) as catalyst, from monomers N,N,N',N'-tetrakis(4-aminophenyl)-1,4-benzenediamine (TPDA) and 1,3,6,8-tetrakis(4-formylphenyl)pyrene (TFPY). For subsequent studies, COF_TPDA-TFPY films were grown on transparent glass using the same solvothermal method. Characterization: Crystallinity and structure were confirmed by PXRD (strong reflections, unit cell parameters a ≈ 23.5 nm, b ≈ 2.43 nm, c ≈ 0.43 nm, β ≈ 90.59°, γ ≈ 78°). FTIR indicated imine formation (peak ~1630 cm−1) with reduction of precursor peaks (~1690 and ~1600 cm−1). 13C NMR showed a signal near 153 ppm consistent with imine carbons. Nitrogen adsorption (BET) showed a type-I isotherm (microporosity), with pore size distribution centered at ~1.32 nm. SEM/AFM revealed films of stacked quadrilateral nanosheets with ~100 nm thickness. TEM showed periodic pore patterns with ~1.6 nm pore-to-pore spacing. GIWAXS of films produced diffraction rings with peaks matching powder PXRD (e.g., 0.38 (110), 0.54 (200), 0.72 (310), 1.43 (001)). Optical response: In situ UV–Vis absorption (Cary 60) under controlled dry/wet N2 demonstrated a humidity-induced absorbance shift (from ~457 nm to ~480 nm), with reversible color change for water exposure and persistent red coloration for sweat exposure. Fingerprint and sweat pore collection: Films were stored in ethanol (75%) then dried with desiccant to ensure yellow appearance before use. Human fingers were cleaned with 75% ethanol and dried. For fingerprint imaging, living fingers gently touched films for ~10 s to generate visible red fingerprint patterns; for sweat pore imaging, contact time was reduced to ~1–5 s to visualize discrete active pores. Patterns were imaged by smartphone; simple image processing (e.g., Photoshop) was used to extract features. Reusability was assessed by washing films with ethanol between uses (tested >50 cycles). Liveness evaluation and matching: Artificial fake fingerprints lacking sweat secretion were used as negative controls and did not produce patterns. Collected fingerprints were compatible with fingerprint matching software; the authors used an open-source code (Fingerprint Matching: https://github.com/allilou63/fingerprint). A database consisting of 12 new samples from a single donor plus 60 additional fingerprints (from FWC2002) was used for similarity testing.

- The COF_TPDA-TFPY film exhibits a sweat-induced, persistent yellow-to-red color change, enabling naked-eye capture of fingerprints from living fingers after ~10 s contact. In contrast, water-induced color changes are reversible upon drying, indicating distinct mechanisms for sweat versus water. - Artificial fake fingerprints that cannot secrete sweat leave no imprint on the film, enabling material-based liveness detection at the collection step with a reported 100% accuracy rate in blocking fake fingerprint attacks before recognition. - Fingerprint images collected on the film are compatible with standard matching: a representative pair from the same donor yielded a similarity factor of 0.8596 (threshold >0.84 for confirming same person). Across a database of 72 samples (12 from the donor, 60 others), all 12 donor samples were correctly identified and all 60 others correctly discriminated. - The film is reusable: ethanol washing erases stable sweat-induced images; after >50 touch-and-wash cycles, clear images were still obtained. The estimated per-collection material cost is < $0.02. - Sweat pore analysis: Reducing contact time to ~1–5 s yields images of active sweat pores that can be directly seen and processed with simple tools, providing level 3 features and an additional liveness cue. Non-active pores can be identified via superimposition analyses. - Structural/optical data supporting performance: TEM shows periodic pores with ~1.6 nm spacing; AFM/SEM indicate ~100 nm film thickness. UV–Vis absorbance peak shifts from ~457 to ~480 nm under humid N2, consistent with solvatochromic/hydrochromic response; sweat produces long-lasting red coloration, suggesting synergistic effects of sweat components. - The approach is instrument-free for end users (smartphone imaging and simple software), quick, and compatible with clinical or field settings.

The study demonstrates that a sweat-responsive COF film can shift liveness detection from software-based post-processing to a material-enabled collection step. Because fake fingerprints cannot secrete sweat, they fail to produce any pattern, effectively preventing spoofing attempts before recognition. The persistent red coloration triggered by sweat, contrasted with reversible water-induced changes, supports a distinct interaction likely due to the combined components of sweat, enabling stable capture of fingerprint ridges. The collected images integrate seamlessly with existing recognition software, achieving high similarity scores for same-donor pairs and full discrimination in the tested database. Additionally, rapid-contact imaging enables capture of active sweat pore distributions, providing level 3 features and a further liveness indicator. This material-based paradigm reduces reliance on complex instrumentation and algorithms during acquisition, potentially enhancing the robustness and usability of fingerprint systems, and opens avenues for assessing sweat pore function relevant to dermatological and endocrine conditions.

This work introduces a sweat-responsive COF film that enables material-based fingerprint liveness detection and direct, naked-eye sweat pore imaging. The film generates persistent red fingerprints only from living fingers, blocks fake fingerprints at the collection stage with 100% reported accuracy, supports compatibility with standard matching algorithms, and can be rapidly reset by ethanol washing for >50 reuse cycles at low cost. By shortening contact times, the method also captures active sweat pore distributions as level 3 features, expanding both identification and potential clinical assessment of sweat function. Future research could optimize COF compositions for enhanced sensitivity and durability, quantify long-term stability and environmental robustness, integrate the films into practical collection devices, and explore clinical correlations between sweat pore activity patterns and disease states.