A bioinspired double-confining strategy enables highly practical target gas detection via promoted solvated ion transport†

Lijuan Wu^a, Guocheng Lv*^a, Lili Wang^a, Yi Zhou^a, Yupeng Chen‡ *^bcd and Cen Tang*^bc
aEngineering Research Center of Ministry of Education for Geological Carbon Storage and Low Carbon Utilization of Resources, Beijing Key Laboratory of Materials Utilization of Nonmetallic Minerals and Solid Wastes, National Laboratory of Mineral Materials, Hebei Key Laboratory of Resource Low-carbon Utilization and New Materials, School of Materials Science and Technology, China University of Geosciences, Beijing, 100083, China. E-mail: guochenglv@cugb.edu.cn
^bCAS Key Laboratory for Biomedical Effects of Nanomaterials and Nanosafety, CAS Center for Excellence in Nanoscience, National Center for Nanoscience and Technology, Beijing 100190, China. E-mail: ypchen0727@buaa.edu.cn; tang@nanoctr.cn
^cUniversity of Chinese Academy of Sciences, Beijing 100049, China
^dCollege of Materials Science and Technology, Beijing Forestry University, Beijing 100083, China

First published on 28th July 2025

---

---

New concepts This study proposes a bionic olfactory sensing strategy based on double-confined ionic channels, overcoming the technical limitations of conventional electronic signal-based sensors that require high operating temperatures, consume excessive power, and suffer from humidity interference. By confining [Bmim][Tf2N] within GO nanochannels and sub-nanometer TPU volumes, a novel bionic double-confined ion transport system is constructed for the first time. Experimental and molecular dynamics simulations demonstrate that this architecture enables efficient triethylamine (TEA) detection in ambient conditions through enhanced directional transport of solvated cations, achieving an exceptional response to 20 ppm TEA (1013.52 ± 14.31%), rapid response/recovery times (7.6 s/26.1 s), and a low detection limit (200 ppb) that significantly surpass those of conventional semiconductor sensors. Unlike the existing mechanisms relying on electron transfer, this work innovatively integrates multiscale confinement effects with ion transport kinetics through bioinspired ionic signaling pathways. The developed sensing platform demonstrates outstanding selectivity, humidity resistance, stability, and energy efficiency, establishing a new paradigm for designing intelligent sensing systems. This breakthrough advances the theoretical understanding of cross-scale synergistic effects in biomimetic materials and bridges critical gaps between biological inspiration and engineering implementation in ionic sensing technologies. The findings provide transformative insights for developing next-generation sensing platforms with environmental adaptability and ultra-low power consumption.

Introduction

As an intelligent frontier technology in information integration and management, the Internet of Things (IoT) has significantly propelled the gas sensing industry due to the necessary information perception and acquisition demand. Among the main sensing technologies, resistive sensing devices demonstrate the most outstanding commercial performance, with their material systems evolving from traditional metal oxides/metals,^1,2 carbon-based materials,³ and conductive polymers,⁴ to novel framework materials (e.g., metal–organic frameworks^5–7 and covalent organic frameworks⁸). Despite this material diversification, metal oxides (i.e., binary, ternary, composite metal oxides, etc.)^9–11 remain central to practical application.¹² However, their reliance on electron transport mechanisms imposes persistent challenges, such as high operational temperatures, high power consumption, and restricted environmental adaptability.¹³ Innovative, unconventional material designs are crucial to advancing gas sensing devices for precise physical/chemical signal detection in smart IoT applications. In nature, most organisms can achieve efficient information transfer (e.g., olfactory signal transduction) with minimal energy cost¹⁴ under a temperature lower than ca. 40 °C.^15,16 Notably, the human brain consumes only about 60 W for short-term neural processing.^14,17 This efficiency is attributed to the unique ionic signal transmission mechanism in living organisms.^18,19 The involved biological ion channels in liquid media offer high throughput, selectivity, and ultra-low resistance for ion transport,^20–22 which significantly reduces energy consumption. However, electronic signal transmission requires more energy to overcome losses from resistance and capacitance in solid media. At present, nanofluidic technology has emerged as a promising platform for achieving bionic sensing, effectively responding to various physical (light,^23–25 force,^26,27 and heat^28–30) and/or chemical (molecules^31–35 and ions^36,37) stimuli with high resolution.^38,39 This sensing platform provides greater flexibility and cost-effectiveness compared to complex microelectromechanical ones.^40,41 Previous research has demonstrated that polyethylene terephthalate nanochannels grafted with poly[1-(4-vinyl benzyl)-1H-imidazole] and polyimide nanochannels modified with 1-(4-amino-phenyl)-2,2,2-trifluoro-ethanone can be employed for CO₂ detection through finely mediated ion transport.^31,42 However, inorganic salt ion based nanofluidic systems were inevitably accompanied by hindrance in device integration due to high water volatility and complex encapsulation processes, which further limits their practical applications. Furthermore, water medium can impede gas diffusion to weaken sensing performance, leading to, for example, extended response and recovery times.^43,44

Unlike the dependence on inorganic salt electrolytes and/or hygroscopic substances to generate inorganic salt ions and/or H⁺/H₃O⁺ to realize ion transport,^45,46 ionic liquids (ILs) inherently contain abundant free anions and cations, exhibiting intrinsic ionic conductivity. Simultaneously, they possess wide electrochemical windows, high ionic density, and excellent thermal stability, which contribute to high environmental adaptation.⁴⁷ Their chemical structures can also be screened and designed to enhance affinity toward target gas molecules, while molecule-level wettability can be mediated to improve anti-humidity interference capacity.⁴⁸ Hence, ILs have emerged as an ideal candidate for realizing ion transport in artificial sensing platforms. Additionally, ILs can be readily encapsulated in polymer substrates and/or nanosheet assemblies for portable application. For instance, 1-ethyl-3-methylimidazolium bis[(trifluoromethyl)sulfonyl]imide ([Emim][TFSI]) was successfully confined in thermoplastic polyurethane (TPU) for toluene⁴⁹ and nitrogen dioxide⁵⁰ detection. Nanosheet assemblies can ensure high gas interaction efficiency due to the large specific surface area,⁵¹ and precise regulation of ion behavior because of abundant nanochannels with tunable interfacial structures,^52–55 as well as feasible promotion determined by large-scale manufacturing. Significantly, the non-covalent interactions between ILs and gas molecules in confined nanochannels may improve the dissociation and diffusion of ILs, promoting mass transport (e.g., ions and molecules) and achieving effective gas sieving.^49,51,56,57 It is anticipated that these confined channels can be employed for gas sensing taking advantage of the gas-induced alternation of the physicochemical properties of ILs.⁵⁸

The human olfactory perception process can provide unique design inspiration. Odorant receptors located on the cilia of olfactory neurons bind to odorants, leading to the opening of ion channels and propagation of ionic signals along the olfactory nerve to the olfactory center (Scheme 1a).^59,60 Inspired by this olfactory perception process, a highly efficient bionic olfactory film specifically for triethylamine (TEA) detection was constructed after confining acidic imidazolium ILs ([Bmim][Tf₂N]) within double-confined channels, which consist of the two-dimensional (2D) laminar nanochannels formed by graphene oxide (GO) nanosheets and sub-nanometer volumes of the TPU matrix (Scheme 1b). This double confinement facilitates the transport of solvated cations (TEA–[Bmim]⁺) for effective TEA detection, which was validated through both experimental studies and molecular dynamics (MD) simulations. The olfactory film with optimized compositions exhibits a rapid response, a low limit of detection (LOD), a minimal power consumption, a high cycling stability and an evident response at high humidity levels. It is anticipated that bionic sensing platforms with a high-level form can be achieved through ionic signal transmission-based nanofluidic technology in the future.

Results and discussion

Construction of double-confined ion channels

Monolayer GO nanosheets with a thickness of ca. 1.5 nm (Fig. S1, ESI†) are employed as the foundational building blocks to construct the GO/IL/TPU films. In detail, GO, ILs, and TPU were uniformly dispersed in a polar aprotic solvent (DMF) under ultrasonic dispersion. Strongly negatively charged GO (ζ = −32.6 mV at pH 7.4) underwent ionization of surface oxygen-containing functional groups in the weakly alkaline DMF medium,⁶¹ exposing its negative charge. Driven by electrostatic forces, IL anions and cations were adsorbed onto the GO surface in the form of an electric double layer. FTIR analysis of GO/TPU revealed splitting, shifting, and broadening of characteristic peaks (e.g., methylene, amide, ether, and ester groups) compared to individual components, thus confirming multiple non-covalent interactions, including π–π conjugation, hydrogen bonding, and van der Waals forces between GO and TPU (Fig. S2, ESI†).^62–65 During the layer-by-layer self-assembly process (Fig. S3, ESI†), GO nanosheets underwent pressure-driven orientation on the PTFE membrane under vacuum filtration, forming an ordered lamellar structure. Simultaneously, under the influence of non-covalent interactions such as electrostatic forces, van der Waals forces, and hydrogen bonds, ILs and TPU form an oriented arrangement together with GO, embedded within the 2D nanochannels. Driven synergistically by pressure and interchain interactions, TPU long chains entwined between GO layers, creating sub-nano volumes (Fig. 1a).^49,50,66,67 The superposition of nano/sub-nano spaces forms a double-confined architecture, providing a ‘host’ for the ILs (Fig. 1a). From a microscopic perspective, compared to the smooth surface of the pure GO film (Fig. S4a, ESI†), the obtained GO/IL/TPU film exhibits a flat yet slightly wrinkled surface (Fig. S4b, ESI†). The cross-section of the GO/IL/TPU film reveals a well-defined laminar structure, showing a uniform thickness of ca. 2.42 μm (Fig. 1b). The corresponding energy-dispersive spectroscopy (EDS) mapping images confirm the uniform distribution of carbon (C), nitrogen (N), and oxygen (O) from ILs and TPU, as well as sulfur (S) and fluorine (F) from [Tf₂N]⁻ (Fig. 1c). Their homogeneous integration within the interlayer channels between GO nanosheets is essential for facilitating ion migration and optimizing gas detection. In addition, the acidic protons on the imidazolium ring, combined with the moderate alkyl chain length of [Bmim]⁺, provide favorable chemical environments that can enhance the effective adsorption and desorption of alkaline substances.^68–71 The incorporation of TPU enhances the humidity resistance of the bionic olfactory films benefiting from intrinsic hydrophobicity.^50,72 X-ray diffraction (XRD) patterns of GO, GO/IL, and GO/IL/TPU indicate interlayer spacings of 8.46, 9.66, and 9.87 Å, respectively (Fig. 1d). The increased interlayer spacings between the GO nanosheets can be attributed to the sequential introduction of ILs and TPU. To elucidate the interactions between components, the FTIR spectra of IL, GO/IL, and GO/IL/TPU were analyzed. Characteristic peaks of ILs are observed at ca. 2960, 2870, and 2940 cm⁻¹ in both GO/IL/TPU and GO/IL films, corresponding to C–H stretching vibrations of the methyl and methylene groups in the imidazolium cations (Fig. 1e).⁷³ In the GO/IL films, interactions between [Bmim]⁺ and GO (e.g., π–π stacking, cation–π interactions, and electrostatic effects)⁷⁴ modify the charge distribution on the imidazolium ring. This alteration leads to a vibrational energy shift of the –CH₂ groups, splitting the original peak near 2939 cm⁻¹ into two distinct peaks at 2935 and 2925 cm⁻¹, while generating a new peak at 2854 cm⁻¹ (Fig. 1e, middle). Notably, the subsequent incorporation of TPU disrupts the interactions between GO and ILs, resulting in an FTIR spectrum of GO/IL/TPU that closely resembles that of bulk ILs. This weakened interaction liberates more active sites on the ILs upon reaction with target gases.

Gas sensing performance of bionic olfactory films

The composition of the biomimetic olfactory film was systematically optimized using the single variable method. The experimental results demonstrate that the GO/IL/TPU film exhibits optimal performance when the loading amounts of GO, ILs, and TPU are 3 mg, 1800 mg, and 5 mg, respectively (details in the ESI,† Fig. S5–S12). Upon exposure to 20 ppm TEA, the optimal GO/IL/TPU film exhibited a significant response of 1013.52 ± 14.31% (Fig. 2a), accompanied by a short t_Res of 7.6 s and a t_Rec of 26.1 s (Fig. 2b). This sensing behavior can be attributed to the efficient adsorption and desorption capability of [Bmim]⁺ for TEA.^69–71 As the TEA concentration increased from 2 to 30 ppm, the response linearly escalated from 306.48 ± 14.53% to 1375.08 ± 7.22%, achieving a high correlation coefficient (R²) of 0.9996 (Fig. 2c). Consequently, the theoretical LOD was estimated to be ca. 200 ppb (Table S1, ESI†). Throughout the entire concentration range, both t_Res and t_Rec remained consistently below 40 s (Fig. S13, ESI†). Compared with the existing literature reports, this dynamic response characteristic has reached the industry-leading level (Table S2, ESI†).

The selectivity towards various gases (e.g., methanol, ethanolamine, ethanol, acetone, ammonia, dichloromethane, TEA, toluene, benzene, cyclohexane, and n-hexane) was evaluated at a constant concentration of 20 ppm (Fig. 2d). Notably, evident responses occurred when TEA and ammonia were detected. A response of 1013.52 ± 14.31% to TEA was observed, which was ca. 14 times higher than that of 68.33 ± 7.31% to ammonia. In the air carrier, the GO/IL/TPU film also demonstrated consistently high selectivity towards TEA (Fig. S14, ESI†). It is suggested that this selectivity is attributed to the superior proton receptor role of TEA, which can do favors for effective interactions with the acidic hydrogen on the imidazolium ring of [Bmim]⁺ (Fig. 2d).^75,76

Underlying mechanism of gas detection

To elucidate the underlying mechanism, both experimental studies and MD simulations were conducted. The synergistic integration of 2D nanochannels with sub-nanometer polymer matrix volumes^49,50,67 establishes a double-confined architecture housing ILs (Fig. 3a, left), which remarkably enhances film sensing performance through optimized ionic transport dynamics. MD simulations indicate that [Bmim]⁺ and [Tf₂N]⁻ show symmetrically layered distribution across the width of the GO nanochannels ((Fig. 3k, left); boundary conditions are shown in Fig. S15, ESI†).^77–79 Analysis of the Z-axis distribution of [Bmim]⁺ molecules relative to the 0.64 nm boundary reveals that nitrogen atoms (N) of the imidazole ring are closer to the negatively charged interface (GO layer) (Fig. S16 and S17a, ESI†) and exhibit a narrower distribution compared to carbon atoms (C) of the alkyl chain (Fig. S17b, ESI†). This difference is ascribed to the greater rigidity of the imidazole ring, conferring enhanced positional stability, whereas the flexible carbon chain displays greater dispersion. ILs serve dual functions as ionic transport media and gaseous molecular capture. Their dynamic behaviors in confined spaces, particularly ion diffusion coefficient, migration number, and conductivity, directly determine the sensing performance of IL-based nanofluidic films. The intrinsically high viscosity of bulk ILs originates from intricate non-covalent interactions between anions and cations, and ionic mobility must overcome this internal frictional resistance—a critical determinant of ion transport efficiency.⁵⁰ Previous studies have established quantitative correlations among viscosity, diffusion coefficient, and molar conductivity in gaseous environments through the following three fundamental equations:

	(1)

Λη = k	(2)

	(3)

Here, D is the diffusion coefficient of the charged ions, k_B is the Boltzmann constant, r is its radius, and Λ is the molar conductivity. k is the Walden product constant that is temperature dependent. η_s and η represent the viscosity of pure ILs and ILs with dissolved gases at 20 °C, respectively. χ_cs is the gas mole fraction, and a is the empirical constant fitted to the viscosity data of selected ILs.⁸⁰ At a given temperature, both D and Λ are inversely proportional to η, indicating that a reduction in η can enhance both D and Λ. The Stokes–Einstein equation (eqn (1)) establishes a correlation between viscosity and ion diffusion.^50,81 Walden's rule (eqn (2)) elucidates the relationship between viscosity and conductivity.^73,82 Seddon's equation (eqn (3)) quantifies the impact of gas molecules on the viscosity of mixtures.^81,83

A sharp reduction in the Warburg coefficient (σ_W) after confining ILs indicates a significant increase in the D within double-confined channels (Fig. S18, ESI†). As governed by the inverse D–η correlation in eqn (1) (D ∝ 1/η), the intensified spatial confinement effects significantly reduce system viscosity by weakening cation–anion interactions to promote ion pair dissociation, thereby accelerating ion mobility.⁵³ The decreased η significantly enhances the conductivity of the film (eqn (2)). The progressive transition from polymer confinement (TPU/IL) to 2D nano-confinement (GO/IL) and ultimately a double-confinement architecture (GO/IL/TPU) enhances both baseline and response currents (Fig. 3b–d). Cation–anion interactions weakened by double confinement also contribute to the IL–TEA association (Fig. 3a, right). To systematically investigate the confinement-dependent gas sensing mechanisms, a comparative analysis of TEA (20 ppm) detection performance was conducted across samples with varying confinement configurations, where the optimized double-confined structure demonstrated a higher response amplitude than others (Fig. S19, ESI†). IL free films are non-conductive and are not able to respond to TEA (Fig. S19a–c, ESI†). The lack of the 2D confined channels provided by GO nanosheets significantly reduces the responses of TPU, IL and TPU/IL to 0%, 8.03 ± 1.64%, and 66.52 ± 11.65%, respectively (Fig. S19c–e, ESI†). They were lower than that with 2D confined channels (Fig. 2a and Fig. S19f, ESI†), which indicates the critical role of 2D confinement in facilitating effective TEA detection at low concentrations. Compared to the response of GO/IL/TPU, TPU free films (e.g., GO/IL and IL) showed reduced responses of 90.52 ± 3.81% and 8.03 ± 1.64%, respectively. This result highlights the role of further polymer confinement for improving detection performance.

The interaction between TEA and the GO/IL/TPU sensing system was further elucidated using FTIR spectroscopy. As is evident from the full spectrum of the GO/IL/TPU film (Fig. S20, ESI†), the spectral features are dominated by strong characteristic signals from IL, with partially observable characteristic peaks of GO. This indicates a significantly higher proportion of IL in the film. Due to the very minimal content of TPU compared to IL, the intense vibrational signals of the IL overshadow the vibrational information of TPU. Consequently, in the GO/IL/TPU film, the IL plays an overwhelmingly dominant role in gas interactions. Based on this observation, the mechanistic investigation focuses primarily on the interaction between the IL and TEA. New absorption peaks at 3358, 2922, 2850, 2807, 2690, and 2485 cm⁻¹ (Fig. 3e) suggest significant interactions between TEA and GO/IL/TPU. The –CH₂ groups’ stretching peaks (C–H bond) of bulk ILs and TEA appear at around 2940 and 2880 cm⁻¹ (Fig. S21 and S22, ESI†). TEA incorporation generates two distinct C–H stretching vibrations (–CH₂) at 2922 and 2850 cm⁻¹, suggesting localized electronic environment modifications through donor–acceptor interactions between the tertiary amine group of TEA and the imidazolium ring.⁸⁴ The vibrational signature at 2807 cm⁻¹ corresponds to the asymmetric stretching modes of –CH₂ groups in TEA. The characteristic absorption peak observed at 3358 cm⁻¹ is assigned to N–H asymmetric stretching vibrations, consistent with the typical spectral range of 3500–3200 cm⁻¹ for such modes and distinguishable from O–H stretching vibrations by its lower absorption intensity (Fig. S22, ESI†).⁸⁵ The characteristic absorption peaks observed at 2690 and 2485 cm⁻¹ are attributed to N–H⁺ stretching vibrations, consistent with protonated amine species.⁸⁶ The imidazolium ring acts as an effective proton donor, forming acid–base interactions with TEA.^68,69,87 These interactions induce vibrational splitting of methylene (–CH₂) stretching modes, simultaneous appearance of both neutral (N–H) and protonated (N–H⁺) species signatures, and formation of TEA–[Bmim]⁺ (Fig. 3g). Consequently, the interlayer spacing undergoes evident expansion, as quantitatively verified by XRD analysis (d-spacing increases from d₁ to d₂) and MD simulations (Fig. 3f and k). In contrast, FTIR analysis indicated no detectable interactions between [Tf₂N]⁻ and TEA. The interaction energy calculations further demonstrated that TEA preferentially interacts with [Bmim]⁺ while exhibiting repulsion toward [Tf₂N]⁻ (Fig. 3h), which aligns with the observed distribution patterns. Under TEA free conditions, both [Bmim]⁺ and [Tf₂N]⁻ exhibited symmetric spatial distributions within the nanochannels (Fig. 3i). Upon TEA introduction, however, [Tf₂N]⁻ adopted an asymmetric distribution that was inversely correlated with TEA localization (Fig. 3j), suggesting a mutually exclusive relationship. Upon TEA introduction, however, [Tf₂N]⁻ adopted an asymmetric distribution that was inversely correlated with TEA localization (Fig. 3j), suggesting a mutually exclusive relationship. Radial distribution function (RDF) analysis shows a weak peak corresponding to TEA–[Bmim]⁺ interactions, whereas no discernible peak is observed for TEA–[Tf₂N]⁻, indicating negligible interaction between them (Fig. S23, ESI†). The TEA-induced solvation of [Bmim]⁺ more easily promotes their dissociation from [Tf₂N]⁻ under double-confined conditions.⁸⁸ The observed decrease in σ_W upon TEA interaction indicates an enhanced D for both [Bmim]⁺ and [Tf₂N]⁻ in the film system, as evidenced by electrochemical analysis (Fig. S18, ESI†). Furthermore, MD simulations corroborate this finding (Fig. 3l). According to the relationship of D, η, and Λ in eqn (1) and (2), an enhanced D results in augmented Λ that manifests as quantifiable signal output in this double-confined sensing platform.

Application evaluation of GO/IL/TPU films

Cyclic stability and humidity resistance are crucial indicators for evaluating sensing materials, as they significantly influence long-term performance and accuracy in practical applications. The GO/IL/TPU films demonstrate an average response of 941% to TEA over 20 cycles. Only a 0.9% variation between the first and last cycles can be observed, representing exceptional cyclic stability (Fig. 4a). At ambient temperature, the responses to 20 ppm TEA at various humidity levels (18%, 40%, 60%, and 80% RH) were investigated in detail, which shows a decreasing trend as humidity increases (Fig. 4b). After penetration of water, a hydrogen bond association peak at 3396 cm⁻¹ appeared (Fig. 4c and Fig. S24, ESI†), which is not associated with TPU (Fig. S25a, ESI†). Water molecules interact with the oxygen-containing groups on GO to boost the infrared signal at 1623 cm⁻¹ that corresponds to the double bonds of the conjugated carbon rings (Fig. 4c and Fig. S25b, ESI†). Additionally, these water molecules interact with electronegative elements (N, O, and F) in [Bmim]⁺ and [Tf₂N]⁻. As a result, partial solvation of both GO⁸⁹ and ILs^83,90 induced by water occurred. During such a process, certain oxygen-containing groups on GO deprotonate,^89,91,92 while partial ILs form solvated ions.⁹⁰ The σ_W values from the N₂ and N₂ + H₂O curves suggest that the water induced solvation effect substantially enhances D to increase Λ (Fig. S26, ESI†). As humidity levels rise, the baseline current increases (Fig. S27a and b, ESI†), indicating that water molecules occupy partial active sites of the ILs. Then, the competitive adsorption between water and TEA weakens the response to TEA (Fig. 4d). Even so, the as-prepared bionic olfactory films maintain a high response (208.07%) to TEA at an 80% RH. Furthermore, σ_W for gas mixtures (i.e., TEA and H₂O) is significantly lower than that for gas mixtures (i.e., N₂ and H₂O) (Fig. S26, ESI†), because the solvated ions formed by TEA and ILs can further enhance D to improve responses under high humid conditions. The exceptional anti-humidity interference capability of GO/IL/TPU films can be attributed to the hydrophobic properties of the TPU matrix and [Bmim][Tf₂N], as well as the unique solvated ion transport.

The t_Res, t_Rec, and overall response of GO/IL/TPU films were evaluated and compared with those of the existing sensing materials for TEA detection at ambient temperature (Fig. 4e and Table S3, ESI†). Till now, a series of materials have been designed to achieve a tradeoff between high response, rapid response and recovery time. While some semiconductor materials (e.g., CuO foams and WO₃ flowers) exhibit a high response, typically along with extended t_Res and t_Rec, certain semiconductor composites (e.g., SnO₂/Co₃O₄ bilayer films, 3D rGO/Co₃O₄, SnS₂/Ti₃C₂T_x hybrids, and Cr₂O₃/TiO₂/Ti₃C₂T_x) show a lower response but shorter t_Res and t_Rec (less than 20 s). Although the PPy-WO₃ hybrids, SnS₂/Nb₄C₃T_x composites, and α-MoO₃/PANI hybrids exhibit relatively balanced performance with a t_Res less than 50 s, a t_Rec less than 160 s, and response between 260% and 760%, there remains a considerable room for improvement. Conversely, the as-prepared GO/IL/TPU films offer efficient TEA detection characterized by the high response and rapid response/recovery, which can suitable for timely and accurate TEA detection at varying concentrations. Furthermore, the power consumption (1.20 ± 0.12 to 8.04 ± 0.06 nW) of the GO/IL/TPU films for TEA detection at various concentrations (2 to 30 ppm) was calculated (Fig. S28, ESI†), which are significantly lower than that of biological systems under analogous conditions. In contrast to most low-power gas sensing materials with power consumption from milliwatts to microwatts, GO/IL/TPU films can operate at nanowatt levels with high performance (Table S4, ESI†).

The amines, such as trimethylamine and dimethylamine, produced by the reduction of trimethylamine oxide in fish, mainly contribute to the characteristic fishy odor and serve as key indicators of freshness.⁹³ Given the significant response of GO/IL/TPU films to amine gases, crucian was selected as a biological model to assess the potential application for fish freshness detection (Fig. 4f). Two fresh crucians were placed in a sealed glass vacuum drier (inset in Fig. 4f). To ensure that the released substances by the crucians were evenly mixed with the air in the drier, the initial gas extraction was performed after storage for 40 min. At this point, the crucians died due to oxygen deprivation, and this time was denoted as 0 h. After each gas extraction, the drier will be refilled with fresh air. Subsequently, gas extraction from the drier was carried out every 2 h. As the storage time of the crucian increased, the fishy odor in the drier gradually intensified, indicating a rise in amine levels. Correspondingly, the response to extracted gases (73–85% RH) increased over time (Fig. 4f and Fig. S29, ESI†). The abovementioned results suggest that the GO/IL/TPU film holds great promise for practical applications in fish freshness detection under wet conditions.

Conclusions

Inspired by human olfaction, an ionic signal transmission-based GO/IL/TPU film was developed for efficient TEA detection at ambient temperature. It exhibits a high response, rapid detection, excellent selectivity, consistent repeatability, anti-humidity interference, and low power consumption. As verified by both experimental studies and MD simulations, these advantages stem from the solvated cation transport within double-confined channels at the nano- and sub-nanometer scales. These findings highlight the exceptional performance of the GO/IL/TPU film for target gas sensing, which will pave the way for designing ionic signal transmission-based bionic integrated sensing systems to develop next-generation robots with intelligent perception toward various external stimuli.

Experimental section

Materials

1-Butyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide salt (C₁₀H₁₅F₆N₃O₄S₂, 99%) was purchased from Macklin Corporation. Monolayer graphene oxide (GO) with a diameter of 0.5–5 μm and a thickness of 0.8–1.2 nm was obtained from Jiangsu XFNANO Technology Co. Poly[4,4′-methylenebis(phenyl isocyanate)-alt-1,4-butanediol/di(propylene glycol)/polycaprolactone] (C₂₈H₄₄N₂O₁₁, denoted as TPU) was sourced from Sigma-Aldrich, USA. N,N-Dimethylformamide (DMF, ≥99%) was purchased from Anhui Zaisheng Technology Co. All reagents were utilized without further purification. A polytetrafluoroethylene microporous filter membrane with a pore diameter of 0.1 μm and a membrane diameter of 47 mm was obtained from Haining Delv New Material Technology Co.

Preparation of bionic olfactory films

A specific amount of TPU was dissolved in 12 mL DMF. Subsequently, GO and ILs were mixed in a predetermined ratio with the TPU-DMF solution. The specific ratios can be found in the Composition Optimization of Bionic Olfactory Films section of the ESI.† Ultrasound treatment was performed for 3 h to ensure adequate mixing of the components. Then, the bionic films were prepared on the PTFE filter membrane using vacuum filtration. Under negative vacuum pressure, ILs and TPU were encapsulated between GO nanosheets via a layer-by-layer self-assembly process. The resulting films were oven-dried at 60 °C for 24 h for subsequential gas detection.

Gas sensing testing

A gold-coated tungsten steel probe was used as an electrode. The films were connected to a Keithley 2612B digital source meter via electrodes, and current–time (I–t) curves were recorded in a constant voltage mode (2 V). The gas configuration was controlled using a FlexStream gas diluter (KIN-TEK, USA). During I–t curve monitoring, the film was exposed to a constant flow of TEA and N₂, alternating with a fixed-duration pulse pattern. Under the air carrier gas, the same testing procedure was also conducted. The response to TEA was determined by changes in current over time, reflecting gas sensitivity using the following equation:

Response = ΔI/(I0 × 100% = (It − I0)/I0) × 100%	(4)

where I_t is the real-time current at time t and I₀ is the baseline current. The response time (t_Res) is defined as the time required for the current in the carrier gas to reach 90% of its maximum value in the presence of the target gas. The recovery time is the duration needed for the current to return to 90% of its maximum value after the target gas is removed.

Characterization

An atomic force microscope (AFM) (Dimension ICON, Bruker, USA) was utilized to characterize the size and thickness of monolayer GO nanosheets. A field emission scanning electron microscope (SEM) (Carl Zeiss, Germany, Merlin Compact) was employed to examine the microscopic morphology of bionic olfactory films. X-ray diffraction (XRD) spectroscopy (Bruker, Germany, D8 focus) was performed to analyze the interlayer spacing of the films, using a Cu Kα (λ = 0.15406 nm) source with a scan step of 0.014302°. Fourier transform infrared (FTIR) spectra were obtained using a Nicolet iS20 FTIR spectrometer (Thermo Fisher Scientific, USA) with an attenuated total reflection (ATR) methodology, 64 scans, and a resolution of 2 cm⁻¹.

Computational details

MD simulations were performed using the GROMACS software package (version 2021.3).^94–97 The molecules were first optimized in ORCA, and the GO was optimized in CP2K. The system was constructed using Packmol.⁹⁸ Atomic interactions were parameterized using the optimized potentials for liquid simulations all-atom (OPLS-AA) force field,⁹⁹ and RESP2 charges obtained from Multiwfn¹⁰⁰ were applied in the calculations. After energy minimization, the systems were pre-balanced in the NPT ensemble using the Berendsen method for 5 ns. The production run was then conducted in the NPT ensemble at 300 K with a time step of 1 fs. The system temperature was controlled by a V-rescale thermostat (τ_T = 1 ps), and Parrinello–Rahman pressure coupling was used to control the pressure. After 20 ns of simulation, the particle distribution was analyzed using a GROMACS toolkit.

Author contributions

L. W.: methodology, investigation, data curation, visualization, formal analysis, writing – original draft, and writing – review and editing; G. L.: conceptualization, supervision, resources, and writing – review and editing; L. W.: methodology, investigation, and resources; Y. Z.: methodology and resources; Y. C.: conceptualization, supervision, resources, funding acquisition, and writing – review and editing; C. T.: writing – review and editing; under the project of “Bioinspired multiscale wet sensing interfaces based on three-phase interface regulation and ionic signal transmission”, Y. C. sincerely appreciates all contributors for their innovative suggestions and timely assistance.

Conflicts of interest

The authors declare no competing financial interest.

Data availability

The data supporting this article have been included as part of the ESI.†

Acknowledgements

This work was supported by the Beijing Natural Science Foundation (2232071), the National Natural Science Foundation of China (52003012), the National Key R&D Program Young Scientist Project of China (2022YFA1206800), the Category B of the One Hundred Talents Program (E3G451R1ZX), and the Natural Science Foundation of Jiangsu Province (BK20230274). The authors acknowledge the High-performance Computing Platform of China University of Geosciences Beijing for its support.

References

1. R. A. Potyrailo, Chem. Rev., 2016, 116, 11877–11923 CrossRef CAS PubMed .
2. S. Y. Jeong, J. S. Kim and J. H. Lee, Adv. Mater., 2020, 32, 2002075 CrossRef CAS PubMed .
3. H. Ding, Y. Wei, Z. Wu, K. Tao, M. Ding, X. Xie and J. Wu, ACS Mater. Lett., 2020, 2, 1381–1411 CrossRef CAS .
4. H. Luo, Y. V. Kaneti, Y. Ai, Y. Wu, F. Wei, J. Fu, J. Cheng, C. Jing, B. Yuliarto, M. Eguchi, J. Na, Y. Yamauchi and S. Liu, Adv. Mater., 2021, 33, 2007318 CrossRef CAS PubMed .
5. L. Kong, C. Yu, Y. Chen, Z. Zhu and L. Jiang, Small, 2024, 20, 2407021 CrossRef CAS PubMed .
6. Y. M. Jo, Y. K. Jo, J. H. Lee, H. W. Jang, I. S. Hwang and D. J. Yoo, Adv. Mater., 2023, 35, 2206842 CrossRef CAS PubMed .
7. L. T. Zhang, Y. Zhou and S. T. Han, Angew. Chem., Int. Ed., 2021, 60, 15192–15212 CrossRef CAS PubMed .
8. C. Yu, L. Xu, F. Zhao, L. Kong, Y. Chen, L. Li, Z. Zhu and L. Jiang, ACS Mater. Lett., 2024, 6, 5454–5478 CrossRef CAS .
9. H. Zhao, J. Li, X. She, Y. Chen, M. Wang, Y. Wang, A. Du, C. Tang, C. Zou and Y. Zhou, ACS Sens., 2024, 9, 2183–2193 CrossRef CAS PubMed .
10. K. Sivaperuman, A. Thomas, R. Thangavel, L. Thirumalaisamy, S. Palanivel, S. Pitchaimuthu, N. Ahsan and Y. Okada, Prog. Mater. Sci., 2024, 142, 101222 CrossRef CAS .
11. A. R. Patil, T. D. Dongale, R. K. Kamat and K. Y. Rajpure, Mater. Today Commun., 2023, 34, 105356 CrossRef CAS .
12. P. Lu, X. Guo, X. Liao, Y. Liu, C. Cai, X. Meng, Z. Wei, G. Du, Y. Shao, S. Nie and Z. Wang, Nano Energy, 2024, 126, 109672 CrossRef CAS .
13. J. Liu, L. Zhang, J. Fan and J. Yu, Small, 2022, 18, 2104984 CrossRef CAS PubMed .
14. L. Chen, D. Pranantyo and F. Xia, Sci. Bull., 2022, 67, 1509–1512 CrossRef CAS PubMed .
15. L. A. Schipper, J. L. Reeve, V. L. Arcus, T. Isson, E. J. Prentice, A. Williamson and A. D. Barnes, Ecol. Evol., 2024, 14(14), e70701 CrossRef PubMed .
16. A. I. Dell, S. Pawar and V. M. Savage, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 10591–10596 CrossRef CAS PubMed .
17. X. Zhang, M. Antonietti and L. Jiang, Sci. China Mater., 2019, 63, 167–171 CrossRef .
18. J. E. Lewis, K. M. Gilmour, M. J. Moorhead, S. F. Perry and M. R. Markham, J. Neurosci., 2013, 34, 197–201 CrossRef PubMed .
19. J. E. Niven, Curr. Opin. Neurobiol., 2016, 41, 129–135 CrossRef CAS PubMed .
20. L. Luo, S. Manda, Y. Park, B. Demir, J. Sanchez, M. P. Anantram, E. E. Oren, A. Gopinath and M. Rolandi, Nat. Commun., 2023, 14, 5364 CrossRef CAS PubMed .
21. S. Ren, Z. Zhang and Z. Dong, Trends Chem., 2024, 6, 726–738 CrossRef .
22. Z. Zhang, X. Huang, Y. Qian, W. Chen, L. Wen and L. Jiang, Adv. Mater., 2019, 32, 1904351 CrossRef PubMed .
23. L. Wang, Y. Zhang, Y. Chen and L. Jiang, ACS Nano, 2024, 18, 9043–9052 CrossRef CAS PubMed .
24. L. Wang, Y. Feng, Y. Zhou, Y. Zhang, X. Yang, J. Hu, L. Wu, X. Lei, Y. Chen, G. Lv and L. Jiang, Adv. Funct. Mater., 2024, 34, 2314165 CrossRef CAS .
25. L. Wang, Y. Zhang, Y. Chen and L. Jiang, Adv. Funct. Mater., 2024, 34, 2401411 CrossRef CAS .
26. Z. Ren, H. Zhang, N. Liu, D. Lei, Q. Zhang, T. Su, L. Wang, J. Su and Y. Gao, EcoMat, 2022, 5, e12299 CrossRef .
27. Q. Zhang, D. Lei, J. Shi, Z. Ren, J. Yin, P. Jia, W. Lu, Y. Gao and N. Liu, Adv. Mater., 2023, 35, 2308795 CrossRef CAS PubMed .
28. S. Hong, G. Zou, H. Kim, D. Huang, P. Wang and H. N. Alshareef, ACS Nano, 2020, 14, 9042–9049 CrossRef CAS PubMed .
29. P. Zhang, S. Chen, C. Zhu, L. Hou, W. Xian, X. Zuo, Q. Zhang, L. Zhang, S. Ma and Q. Sun, Nat. Commun., 2021, 12, 1844 CrossRef CAS PubMed .
30. K. Chen, L. Yao and B. Su, J. Am. Chem. Soc., 2019, 141, 8608–8615 CrossRef CAS PubMed .
31. Y. Xu, X. Sui, S. Guan, J. Zhai and L. Gao, Adv. Mater., 2015, 27, 1851–1855 CrossRef CAS PubMed .
32. Q. Ouyang, L. Tu, Y. Zhang, H. Chen, Y. Fan, Y. Tu, Y. Li and Y. Sun, Anal. Chem., 2020, 92, 14947–14952 CrossRef CAS PubMed .
33. X. Shang, G. Xie, X.-Y. Kong, Z. Zhang, Y. Zhang, W. Tian, L. Wen and L. Jiang, Adv. Mater., 2017, 29, 1603884 CrossRef PubMed .
34. Y. Huang, H. Xue, X. Lou and F. Xia, Sci. Bull., 2021, 66, 1599–1600 CrossRef PubMed .
35. J.-J. Hu, W. Jiang, Q. Chen, R. Liu, X. Lou and F. Xia, Anal. Chem., 2021, 93, 14036–14041 CrossRef CAS PubMed .
36. S. Zhang, Q. Du, J. Wang, Y. Huang and F. Xia, Anal. Chem., 2024, 96, 7163–7171 CrossRef CAS PubMed .
37. Y. Zhang, Y. Ma, L. Wang, C. Li, L. Wu, C. Zhong, B. Sun, Y. Chen and L. Jiang, Small, 2024, 20, 2403629 CrossRef CAS PubMed .
38. X. Zhang, M. Lin, Y. Dai and F. Xia, Anal. Chem., 2023, 95, 10465–10475 CrossRef CAS PubMed .
39. Y. Chen, Z. Zhu, Y. Tian and L. Jiang, Exploration, 2021, 1, 1–33 CrossRef .
40. M. Yin, Z. Alexander Kim and B. Xu, Adv. NanoBiomed Res., 2021, 2, 2100117 CrossRef .
41. H. Daiguji, Comprehensive Nanoscience and Nanotechnology, 2011, pp. 207–228 DOI:10.1016/b978-0-12-812295-2.00132-x .
42. X. Shang, G. Xie, X.-Y. Kong, Z. Zhang, Y. Zhang, W. Tian, L. Wen and L. Jiang, Adv. Mater., 2017, 29, 1603884 CrossRef PubMed .
43. Y. Ma, W. Li, W. Zhang, L. Kong, C. Yu, C. Tang, Z. Zhu, Y. Chen and L. Jiang, Mater. Horiz., 2024, 11, 3996–4014 RSC .
44. L. Wang, W. Zhang, J. Cao, S. Li, Y. Ma, Y. Meng, Y. Chen and L. Jiang, Device, 2024, 2, 100293 CrossRef .
45. Y. Zhou, Y. Wang, Y. Wang and X. Li, Anal. Chem., 2020, 92, 16033–16042 CrossRef CAS PubMed .
46. Y. Wang, Y. Zhou and Y. Wang, Sens. Actuators, B, 2020, 323, 128695 CrossRef CAS .
47. F. Borghi, C. Piazzoni, M. Ghidelli, P. Milani and A. Podesta, J. Phys. Chem. C, 2021, 125, 1292–1303 CrossRef CAS .
48. Y. Lu, W. Chen, Y. Wang, F. Huo, L. Zhang, H. He and S. Zhang, Phys. Chem. Chem. Phys., 2020, 22, 1820–1825 RSC .
49. M. L. Jin, S. Park, J. S. Kim, S. H. Kwon, S. Zhang, M. S. Yoo, S. Jang, H. J. Koh, S. Y. Cho, S. Y. Kim, C. W. Ahn, K. Cho, S. G. Lee, D. H. Kim and H. T. Jung, Adv. Mater., 2018, 30, e1706851 CrossRef PubMed .
50. M. L. Jin, S. Park, H. Kweon, H. J. Koh, M. Gao, C. Tang, S. Y. Cho, Y. Kim, S. Zhang, X. Li, K. Shin, A. Fu, H. T. Jung, C. W. Ahn and D. H. Kim, Adv. Mater., 2021, 33, e2007605 CrossRef PubMed .
51. D. Chen, W. Wang, W. Ying, Y. Guo, D. Meng, Y. Yan, R. Yan and X. Peng, J. Mater. Chem. A, 2018, 6, 16566–16573 RSC .
52. W. Wu, J. Wang, J. Liu, P. Chen, H. Zhang and J. Huang, Int. J. Hydrogen Energy, 2017, 42, 11400–11410 CrossRef CAS .
53. M. Dong, K. Zhang, X. Wan, S. Wang, S. Fan, Z. Ye, Y. Wang, Y. Yan and X. Peng, Small, 2022, 18, 2108026 CrossRef CAS PubMed .
54. I. Bou-Malham and L. Bureau, Soft Matter, 2010, 6, 4062–4065 RSC .
55. A. Fang and A. Smolyanitsky, ACS Appl. Mater. Interfaces, 2019, 11, 27243–27250 CrossRef CAS PubMed .
56. H. Lin, K. Gong, P. Hykys, D. Chen, W. Ying, Z. Sofer, Y. Yan, Z. Li and X. Peng, Chem. Eng. J., 2021, 405, 126961 CrossRef CAS .
57. H. Dou, M. Xu, B. Jiang, G. Wen, L. Zhao, B. Wang, A. Yu, Z. Bai, Y. Sun, L. Zhang, Z. Chen and Z. Jiang, Adv. Funct. Mater., 2019, 29, 1905229 CrossRef CAS .
58. C. Y. Li, H. Y. Liu, L. Y. Xu, F. R. Zhao, J. J. Zhu, L. J. Wu, C. Y. Yu, Y. P. Chen, Z. P. Zhu, C. Tang and L. Jiang, Adv. Funct. Mater., 2025, 35, 2418630 CrossRef CAS .
59. E. Strickland, IEEE Spectr., 2022, 59, 22–27 Search PubMed .
60. J. Fu, Z. Song, H. Wang, G. Xu, X. Li, W. Ding and Y. Zi, Nano Energy, 2024, 121, 109156 CrossRef CAS .
61. L. Zhang, W. Baslyman, P. Yang and N. M. Khashab, Chem. Commun., 2019, 55, 620–623 RSC .
62. J. S. Choi, S. J. Han and J.-H. Yim, ACS Appl. Electron. Mater., 2024, 6, 658–668 CrossRef CAS .
63. Y.-S. Joo, J.-R. Cha and M.-S. Gong, Mater. Sci. Eng., C, 2018, 91, 426–435 CrossRef CAS PubMed .
64. C.-H. Tsou, H.-T. Lee, M. De Guzman, H.-A. Tsai, P.-N. Wang, H.-J. Cheng and M.-C. Suen, Polym. Bull., 2015, 72, 1545–1561 CrossRef CAS .
65. S. L. Cooke and A. R. Whittington, Mater. Sci. Eng., C, 2016, 60, 78–83 CrossRef CAS PubMed .
66. J. Jiang, F. Liu, X. Yang, Z. Xiong, H. Liu, D. Xu and W. Zhai, Polymer, 2021, 228, 123872 CrossRef CAS .
67. J. Wang, Y. Zheng, T. Cui, T. Huang, H. Liu, J. Zhu, L. Song and Y. Hu, Adv. Funct. Mater., 2023, 34, 2312383 CrossRef .
68. H. Olivier-Bourbigou, L. Magna and D. Morvan, Appl. Catal., A, 2010, 373, 1–56 CrossRef CAS .
69. J. Tian and B. Liu, Crit. Rev. Environ. Sci. Technol., 2020, 52, 767–809 CrossRef .
70. G. Sun, W. Huang, D. Zheng, L. Dong and X. Wu, Chin. J. Chem. Eng., 2014, 22, 72–78 CrossRef CAS .
71. G. Li, Q. Zhou, X. Zhang, L. Wang, S. Zhang and J. Li, Fluid Phase Equilib., 2010, 297, 34–39 CrossRef CAS .
72. Y. Zhou, L. Zhao, Q. Jia, T. Wang, P. Sun, F. Liu, X. Yan, C. Wang, Y. Sun and G. Lu, ACS Appl. Mater. Interfaces, 2022, 14, 55109–55118 CrossRef CAS PubMed .
73. L. M. Ramenskaya, E. P. Grishina and N. O. Kudryakova, J. Mol. Liq., 2020, 312, 113368 CrossRef CAS .
74. K. Gupta, S. R. Yasa, A. Khan, O. P. Sharma and O. P. Khatri, J. Colloid Interface Sci., 2022, 607, 1973–1985 CrossRef CAS PubMed .
75. L. Sui, X. Song, X. Cheng, X. Zhang, Y. Xu, S. Gao, P. Wang, H. Zhao and L. Huo, CrystEngComm, 2015, 17, 6493–6503 RSC .
76. Q. Ma, H. Li, S. Chu, Y. Liu, M. Liu, X. Fu, H. Li and J. Guo, ACS Sustainable Chem. Eng., 2020, 8, 5240–5250 CrossRef CAS .
77. Y. Wang, F. Huo, H. He and S. Zhang, Phys. Chem. Chem. Phys., 2018, 20, 17773–17780 RSC .
78. H. Dou, B. Jiang, M. Xu, Z. Zhang, G. Wen, F. Peng, A. Yu, Z. Bai, Y. Sun, L. Zhang, Z. Jiang and Z. Chen, Angew. Chem., Int. Ed., 2019, 58, 13969–13975 CrossRef CAS PubMed .
79. S. Di Lecce, A. A. Kornyshev, M. Urbakh and F. Bresme, ACS Nano, 2020, 14, 13256–13267 CrossRef CAS PubMed .
80. N. Gao, Y. Yang, Z. Wang, X. Guo, S. Jiang, J. Li, Y. Hu, Z. Liu and C. Xu, Chem. Rev., 2023, 124, 27–123 CrossRef PubMed .
81. J.-Y. Jeon, S.-J. Park and T.-J. Ha, Sens. Actuators, B, 2022, 360, 131672 CrossRef CAS .
82. S. Panda, S. Ray, V. Losetty and R. L. Gardas, J. Mol. Liq., 2016, 224, 882–892 CrossRef CAS .
83. K. R. Seddon, A. Stark and M.-J. Torres, Pure Appl. Chem., 2000, 72, 2275–2287 CrossRef CAS .
84. I. B. Magdău and G. J. Ackland, Phys. Rev. Lett., 2017, 118, 145701 CrossRef PubMed .
85. A. Ezabadi and M. Salami, Res. Chem. Intermed., 2022, 48, 1287–1303 CrossRef CAS .
86. R. J. Warren, W. E. Thompson and J. E. Zarembo, J. Pharm. Sci., 1965, 54, 1554 CrossRef CAS PubMed .
87. Y. Huang and M. Sulkes, J. Phys. Chem., 1996, 100, 16479–16486 CrossRef CAS .
88. O. Nordness and J. F. Brennecke, Chem. Rev., 2020, 120, 12873–12902 CrossRef CAS PubMed .
89. F. Zhao, H. Cheng, Z. Zhang, L. Jiang and L. Qu, Adv. Mater., 2015, 27, 4351–4357 CrossRef CAS PubMed .
90. D. Wu, M. Wang, W. Yu, G.-G. Wang and J. Zhang, Chem. Eng. J., 2024, 486, 150121 CrossRef CAS .
91. Z. Wang, Q. Shen, J. Zhang, M. Jiang, W. Chen, P. Tao, C. Song, B. Fu, T. Deng and W. Shang, J. Mater. Chem. A, 2020, 8, 9248–9255 RSC .
92. W. Gao, N. Singh, L. Song, Z. Liu, A. L. M. Reddy, L. Ci, R. Vajtai, Q. Zhang, B. Wei and P. M. Ajayan, Nat. Nanotechnol., 2011, 6, 496–500 CrossRef CAS PubMed .
93. C. Jiang, A. J. X. Guo, Y. Li, Y. Wang, J. Sun, Z. Chen, X. Chen and Q. Ma, Chem. Eng. J., 2024, 484, 149283 CrossRef CAS .
94. D. Van der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E. Mark and H. J. C. Berendsen, J. Comput. Chem., 2005, 26, 1701–1718 CrossRef CAS PubMed .
95. S. Páll, M. J. Abraham, C. Kutzner, B. Hess and E. Lindahl, Tackling Exascale Software Challenges in Molecular Dynamics Simulations with GROMACS, Cham, 2015 Search PubMed .
96. M. J. Abraham, T. Murtola, R. Schulz, S. Páll, J. C. Smith, B. Hess and E. Lindahl, SoftwareX, 2015, 1–2, 19–25 CrossRef .
97. H. J. C. Berendsen, D. van der Spoel and R. van Drunen, Comput. Phys. Commun., 1995, 91, 43–56 CrossRef CAS .
98. L. Martínez, R. Andrade, E. G. Birgin and J. M. Martínez, J. Comput. Chem., 2009, 30, 2157–2164 CrossRef PubMed .
99. W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am. Chem. Soc., 1996, 118, 11225–11236 CrossRef CAS .
100. T. Lu and F. Chen, J. Comput. Chem., 2012, 33, 580–592 CrossRef CAS PubMed .

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5mh00718f

‡ Lead contact.

This journal is © The Royal Society of Chemistry 2025