Symmetric Metal Organic Framework‐Plasmonic Architectures for Reversible and High‐Sensitivity Optical Sensing

ABSTRACT

Plasmonic nanostructures have attracted considerable interest for their ability to manipulate light at the nanoscale. The optical response of these metallic architectures is greatly sensitive to changes in their surrounding environment, which makes them particularly interesting for sensing applications. Here, we report a versatile and highly sensitive plasmonic sensing platform that combines the surface lattice resonances (SLRs), sustained by 2D gratings of metal nanoparticles, and the exceptional refractive index tunability offered by ZIF‐8 metal–organic framework (MOF) films. The SLRs provide narrow resonances that strongly depend on the surrounding refractive index, which can be modulated via solvent exchange in the ZIF‐8. The plasmonic 2D arrays were fabricated using templated‐assisted assembly of metal colloids; a scalable and gentle process that enabled the study of different ZIF‐8/grating configurations. First, we studied ZIF‐8 films on top of the 2D arrays, where an optimized concentration of the colloidal dispersion allowed dynamic and reversible ON/OFF tuning of the SLR response. Second, we explored the role of the refractive index mismatch between substrate and ZIF‐8 superstrate. Finally, we fabricated 2D plasmonic gratings completely embedded within ZIF‐8 layers, achieving enhanced sensing performance with a refractive index sensitivity of 427 ± 19 nm·RIU^{− 1}. These results highlight the potential of MOF–plasmonic hybrids as responsive and tunable nanophotonic platforms for next‐generation optical sensing applications.

This work presents a plasmonic sensing platform combining surface lattice resonances (SLRs) in 2D metal nanoparticle gratings with ZIF‐8 MOF films. The plasmonic grating is easily and scalably produced from template‐assisted assembly of metal colloids, which is fully compatible with MOF growth. The resulting MOF‐plasmonic structure allows for a dynamic refractive index tuning through solvent exchange achieving enhanced sensing performance with a refractive index sensitivity of 427 ± 19 nm·RIU^{− 1}.

1. Introduction

Plasmonic resonances originate from the collective oscillations of free electrons in the conduction band of metallic nanostructures under optical excitation, giving rise to localized surface plasmon resonances (LSPRs). These resonances are confined to the metal‐dielectric interface and exhibit pronounced sensitivity to variations in the local refractive index. This high sensitivity, combined with the strong field enhancement near the nanostructure surface, renders plasmonic systems highly attractive for label‐free, real‐time optical sensing applications. In these sensing schemes, analyte detection is typically achieved by monitoring spectral shifts or intensity variations of the plasmonic resonance induced by changes in the surrounding dielectric environment.

Most conventional plasmonic sensors exploit the changes in the LSPRs, whose spectral bandwidth is restricted (low quality factors, Q‐factors) due to Ohmic losses and radiative damping [1]. The resulting broad resonance linewidths of the LSPR limit the spectral resolution and reduce detection sensitivity, especially when monitoring subtle refractive index variations [2]. To overcome this drawback, surface lattice resonances (SLRs) can be used as an alternative. SLRs arise from the coherent coupling between LSPRs and diffractive orders (Rayleigh anomalies (RA)) in periodically arranged plasmonic nanostructures [3]. The collective response of the array leads to loss compensation and spectrally narrow, high‐Q resonances with strong field confinement, significantly enhancing light‐matter interactions [4, 5, 6]. The resonance wavelength of SLRs can be engineered via various parameters, including nanoparticle (NP) size, shape, material composition, lattice spacing, and, most notably, they are influenced by the refractive index of the surrounding medium [7, 8, 9, 10]. The spectral position of the SLR is commonly governed (and slightly redshifted) by the RA condition, which for a square lattice is expressed as:

$${\lambda }_{\mathrm{RA}}=\frac{n·LP}{\sqrt{i^2+j^2}}$$

where λ _RA is the resonance wavelength, n is the refractive index of the surrounding medium, LP is the lattice period of the array, and (i, j) denote the diffraction order index [3, 11].

SLR band shift predictably in response to changes in the refractive index, typically manifesting as spectral red‐ or blueshifts [11, 12]. SLR‐based sensors already demonstrate remarkable sensitivity for bulk refractive index sensing. However, their optical response is typically fixed once fabricated, which present a major limitation for achieving dynamic tunability and analyte‐specific responsiveness [2]. To overcome the limitations of static plasmonic architectures, researchers have increasingly integrated functional materials with plasmonic arrays to achieve active tuning of SLRs [13, 14, 15]. Among the various candidates, metal‐organic frameworks (MOFs) have emerged as particularly promising due to their tunable porosity and chemical versatility. These materials combine metal centres such as zinc, copper, iron and zirconium, in networks mediated by organic ligands (like carboxylates) forming porous structures. Some of the most widely used examples include MOF‐5 (Zn), HKUST‐1 (Cu), UiO‐66 (Zr), and MIL‐101(Cr/Fe), which are extensively studied for applications in gas storage, catalysis and drug delivery [16].

Within this class of porous materials, zeolitic imidazolate frameworks (ZIFs), especially ZIF‐8, have garnered significant attention for their large surface area, high porosity, and structural robustness [17, 18]. Composed of Zn^{2 +} ions coordinated with 2‐methylimidazolate ligands, ZIF‐8 adopts a sodalite‐like topology with uniform micropores [19]. Among the widely utilized MOFs, ZIF‐8 stands out due to its good chemical and thermal stability, making it highly suitable for guest adsorption, molecular separation, and storage in liquid environments [20, 21, 22, 23, 24]. This stability has enabled its study as a complement to plasmonic nanostructures in hybrid structures across various applications, where its primary role is to act as a matrix for accumulating and “trapping” analytes in close proximity to the plasmonic structures for use in enhanced spectroscopies and optical sensors [25, 26]. Beyond its porous architecture, the well‐defined structure and tunable surface chemistry of ZIF‐8 enable its use as a self‐assembly platform, where size‐, shape‐ and charge‐complementary interactions facilitate the engineering of hierarchical nanostructures [27, 28].

ZIF‐8 has demonstrated broad applicability in plasmonic sensing platforms, particularly in surface‐enhanced Raman spectroscopy (SERS). For instance, Koh et al. developed a plasmonic nose using Ag nanocube arrays encapsulated in ZIFs for ppm‐level detection of volatile organic compounds via in situ adsorption kinetics [29]. Marchi et al. designed Au@Ag@ZIF‐8 SERS nanotags that integrate plasmonic enhancement and bioconjugation for multiplexed detection of cell surface biomarkers, whereas Liu et al. incorporated ZIF‐8 into nanoparticle‐on‐mirror plasmonic nanogaps to create tunable SERS substrates for sensitive detection of environmental pollutants [30].

The extraordinary porosity of ZIF‐8 also plays a pivotal role in refractive‐index‐based plasmonic sensing. The adsorption of analyte molecules in ZIF‐8 increases the effective local refractive index within its porous network, leading to detectable changes in the media surrounding the plasmonic arrays in the form of spectral shifts of the optical resonances. This combination enables dynamic refractive index tuning (1.35–1.6) through selective molecule uptake [20], resulting in precise and reversible resonance shifts for analyte‐specific detection. A notable example is the work by Chen et al., who developed a highly sensitive sensor by growing ZIF‐8 layers in situ on gold nanohole arrays. This system exhibited narrow‐band resonances and high sensitivity to ethanol vapor (50.1 pm/ppm), showcasing the practical advantages of ZIF‐8 in plasmonic sensing [31]. Similarly, Xiong et al. developed a ZIF‐8‐functionalized S‐tapered fiber sensor for the selective detection of polystyrene nanoplastics in aqueous environments, demonstrating high sensitivity (up to 114.1 nm/%(w/v)) enabled by electrostatic adsorption and π‐π stacking interactions [32]. These examples highlight the potential of ZIF‐8 plasmonic systems to enable high‐performance, analyte‐responsive optical sensing.

In this work, we explore various configurations that combine ZIF‐8 films with 2D plasmonic gratings to optimize refractive index sensing performance. The 2D periodic arrays are fabricated using a scalable, template‐assisted assembly of colloidal spherical Au NPs. This versatile nanofabrication approach enables the controlled deposition of ZIF‐8 layers both above and below the plasmonic gratings, allowing a systematic evaluation of different configurations. First, we characterize the optical response of the SLR sustained by our 2D grating in the presence of different solvents adsorbed into the ZIF‐8 layer grown on top. This configuration demonstrates excellent and reliable refractive index sensing, which can be further improved by using substrates with lower refractive indices, as we show here. To enhance the dynamic sensitivity of the system, we further embed the entire plasmonic grating within ZIF‐8, creating a sandwich‐like architecture. This is achieved by first growing a bottom ZIF‐8 film, followed by the assembly of the Au NP array, and finally growing a second ZIF‐8 layer on top, fully encapsulating the plasmonic structure. This later configuration enables a more pronounced modulation of the SLR position by tuning the local refractive index surrounding the gold nanostructures. As a result, we overcome the sensitivity limitations of previous configurations and achieve a figure of merit (FoM) of 17.1 ± 0.8 RIU⁻¹ around 905 nm. Our work provides a simpler and faster path toward the development of highly responsive optical sensors, laying the groundwork for next‐generation tunable optical sensors, plasmonic devices and other applications in nanophotonics.

2. Results and Discussion

To investigate the interplay between plasmonic gratings and MOFs, we leverage the straightforward growth of ZIF‐8 on solid substrates together with the scalable fabrication of 2D plasmonic arrays via template‐assisted assembly of metal colloids. The first configuration explored was the MOF coated plasmonic array, whose fabrication is summarized in Figure 1a. The 2D plasmonic grating was fabricated using our previously reported template‐assisted self‐assembly method [10, 33]. Briefly, a pre‐patterned elastomeric mold made of polydimethylsiloxane (PDMS) was used to guide the 30 nm Au NPs (with the Au⁰ concentration of approximately 20 mm, see related SEM image in Figure S2a) into a well‐ordered array of colloids, as confirmed by the SEM image in Figure 1b. The pattern was defined by the PDMS mold, which contained a 2D array of holes with a diameter (d) of 250 nm, a lattice parameter (LP) of 600 nm, and a feature depth of 390 nm. The Au NP array was heated on a hotplate at 60 °C overnight and then directly immersed into the ZIF‐8 growth solution for one cycle. SEM images (Figure 1c,d) revealed that ZIF‐8 nucleated preferentially on the Au NPs arrays, forming discrete dome‐like structures while avoiding the exposed glass regions. This selective growth likely stems from the low surface energy and chemical inertness of untreated glass, as well as the possible presence of organic contaminants or hydrophobic regions [18].

FIGURE 1.

Controllable deposition of ZIF‐8 thin films on Au NP arrays. Schematic illustration of the fabrication process (a). SEM image of the ordered Au NP array (b). ZIF‐8 growth on non‐plasma‐treated sample: “dome‐like” core‐shell structures (c,d). ZIF‐8 growth on plasma‐treated sample: continuous and uniform film (e,f).

To promote uniform ZIF‐8 coverage across the entire substrate, the plasmonic array was subjected to a 2 min oxygen plasma treatment prior to ZIF‐8 growth. This treatment removes organic residues and introduces polar functional groups (e.g., ─OH, ─COOH), thereby increasing surface energy and hydrophilicity [34]. As a result, even a single growth cycle yielded a continuous ZIF‐8 film (∼100 nm, Figure S3b) with uniform coverage over both the Au NPs and glass regions, as shown in Figure 1a‐ii. After three growth cycles, the film thickness increased to ∼300 nm [20] (Figure 1e,f, while maintaining a similar morphology.

Grazing incidence powder X‐ray diffraction (GIPXRD) analysis was performed on samples with and without oxygen plasma treatment (Figure S4). Both samples exhibited the characteristic crystalline peaks of ZIF‐8, confirming successful MOF formation. However, the plasma‐treated sample exhibited stronger ZIF‐8 signals and weaker gold features, consistent with thicker and more uniform MOF coating. These results support the SEM observations (Figure 1c–f) and underscore the role of plasma activation in enhancing MOF growth and interfacial uniformity.

Immersion of the Au NP array/ZIF‐8 hybrid structures in various solvent systems enables efficient molecular uptake by the ZIF‐8 framework, resulting in a dynamic and reversible shift of the SLRs. This allows tunable control of the optical response of the system. For these 2D plasmonic gratings to exhibit a well‐defined SLR peak in the transmittance spectrum, they must be surrounded by a homogeneous refractive index medium. The as‐prepared bare colloidal arrays on glass show a broad LSPR band centered at 700 nm in the transmittance spectra (Figure S5). This LSPR is due to the near field coupling between the NPs in each cluster. Upon index matching (achieved by adding a superstrate or a MOF coating, such as ZIF‐8 in this work), a sharp and intense resonance emerges near 900 nm, attributed to the SLR. The position and intensity of this SLR depends on the effective refractive index of the superstrate (ZIF‐8 with adsorbed solvent) with respect to that of the underlying substrate (glass, in this case).

Figure 2 shows the evolution of the transmittance spectrum of ZIF‐8 coated Au gratings, both plasma‐treated (Figure 2a) and untreated (Figure 2b), upon exposure to different solvents absorbed within the MOF porosity. The materials used and their corresponding refractive index within ZIF‐8 are as follows: Air (n _ZIF‐8 ≈ 1.34), isopropanol (IPA, n _IPA+ZIF‐8 ≈ 1.41), dimethyl sulfoxide (DMSO, n _DMSO+ZIF‐8 ≈ 1.45), toluene (n _{Toluene+ZIF‐8} ≈ 1.51), and chlorobenzene (n _PhCl+ZIF‐8 ≈ 1.54). The effective refractive index in each case reflects the interaction between the ZIF‐8 layer and the respective solvent, and was determined by ellipsometry measurements (Figures S6 and S7). A clear redshift in the SLR position is observed with increasing refractive index, shifting from approximately 880 nm for n = 1.34 to 910 nm for n = 1.54. Notably, the full width at half maximum (FWHM) of the resonance peak also decreases progressively, indicating a higher Q‐factor of the resonance at higher refractive indices.

FIGURE 2.

Transmittance spectra of Au NP array/ZIF‐8 structures. (a) and (b) present the transmittance spectra of the MOF layer, infiltrated with various solvents, both with (a) and without (b) O₂ plasma treatment. (c) and (d) display the Finite‐Difference Time‐Domain modelling results for these structures composed by a cluster of 56 nanoparticles on top of a glass substrate. The superstrate in (c) is a 300 nm flat layer atop the nanoparticle clusters, while (d) models it as hemi‐spheroids, 500 nm in diameter in the x, y plane and 150 nm height.

In contrast, the non‐plasma‐treated samples (Figure 2b) display significantly weaker and less responsive SLR behavior. At low refractive indices, no distinct SLR peak is observed. Even as the refractive index increases, the redshift remains minimal. A broad SLR feature only becomes evident at n = 1.54. The diminished response is likely due to the inhomogeneous nature of the ZIF‐8 layer in the untreated samples, which does not provide sufficient index matching (between superstrate and substrate) for the 2D grating to support a well‐defined SLR (see Figure S9).

The Finite Difference Time Domain (FDTD) calculated transmittance spectra and a representative scheme for both structures are displayed in Figure 2c,d. To isolate the effect of morphology, the same refractive index was used for the substrate and the superstrate. The continuous MOF layer (Figure 2c) and the MOF dome‐like structure (Figure 2d) induce markedly different optical responses in the gratings. In the thin‐layer model, a well‐resolved SLR is observed. Conversely, for the dome structures, only broad features are present, which are attributed to LSPR. This comparison highlights the importance of the proposed architecture: not only must the ZIF‐8 and NPs be in contact, so that the MOF acts as a homogeneous dielectric surrounding medium, but the formation of a continuous layer is also crucial to support collective plasmonic modes. It is worth noting that, while experimental modulation of transmittance can still be observed in the dome structures (Figure 2b), these broad features limit their applicability as sensors. However, the ability to confine analyte capture exclusively to the metal (sensing) regions, where the near field is strongest, could offer advantages in molecular sensing applications.

The well‐defined resonances of the SLR modes observed in the continuous‐layer systems (Figure 2a) allow a robust evaluation of their performance as refractive index sensors. These results highlight the importance of achieving uniform and dense MOF coverage, enabled by plasma treatment, for high‐sensitivity and well‐defined plasmonic resonance behavior.

An additional benefit of using metal colloid‐based 2D gratings is the possibility of tuning the grating morphology by varying the concentration of the Au NP solution during the self‐assembly process. Lowering the concentration of the colloidal dispersion (from 20 mm in Figure 2 to 10 mm in this case), while using the same PDMS stamp, results in 2D arrays with smaller clusters of NPs (corresponding SEM images are shown in Figure S2b), which directly impacts the scattering strength of the SLR. By optimizing the cluster size (both in terms of diameter and the number of colloids per cluster), we were able to fabricate ZIF‐8 coated gratings where no SLR is observed (black and grey lines in Figure 3a), with only the LSPR appearing around λ = 650 nm. However, upon adsorption of chlorobenzene (PhCl) molecules into the Au NP array/ZIF‐8 structure (red and pink lines), a sharp and well‐defined SLR peak emerges. This allows a clear identification of the presence of the solvent based on the appearance of the SLR peak in the transmittance spectrum.

FIGURE 3.

Optical switching and angular response of Au NP array/ZIF‐8 hybrid. Reversible transmittance switching between chlorobenzene‐infiltrated (red and pink) and chlorobenzene‐removed (black and grey) states (a), two cycles are shown first (curves 1–2) and second (curves 3–4). Corresponding Q‐factor values (b). Angle‐resolved transmittance maps in the ZIF‐8 without analyte case (c) and with the ZIF‐8 filled with chlorobenzene (PhCl), n = 1.54 (d).

Conceptually, the basis of this distinct differential behaviour lies in the refractive index contrast between the substrate and the superstrate. When the grating is constituted by small clusters of colloids, the amount of light scattered by each cluster is insufficient to support a collective mode. As a result, the optical quality of the collective mode becomes highly dependent on the refractive index contrast between the substrate and the superstrate. Specifically, when the refractive index of the substrate is higher than that of the superstrate, the optical quality of the SLR mode is poor, resulting in broad and low‐intensity resonances. Only when both refractive indices are similar, a sharp and intense SLR is observed [10] (see also Figure S10). In summary, optimizing the concentration of the metal colloid dispersion during the templated guided assembly enables the fabrication of 2D gratings where the SLR can be switched ON or OFF depending on the presence of solvent adsorbed within the MOF.

Furthermore, the solvent can be reversibly removed from the ZIF‐8 pores, demonstrating a robust optical switching behavior. This process, shown in Figure 3, remains effective even after five complete adsorption/desorption cycles of PhCl. Figure 3b illustrates how the Q‐factor of the SLR peak (Q = λ_SLR / FWHM) can be reversibly modulated between a high value of 50 (with PhCl adsorbed) and a low value of 3–5 (after desorption), with minimal hysteresis (see Figure S8).

Angle‐resolved transmittance measurements further reveal two distinct optical responses depending on the presence of PhCl (Figure 3c,d). In Figure 3d, the infiltrated MOF layer (n = 1.54) leads to the emergence of a pronounced SLR mode, evidence by a sharp “X”‐shaped dispersion centered around 900 nm. This pattern is characteristic of the angularly dispersive nature of SLRs [33]. The narrow linewidth reflects the high Q‐factor of the system when the refractive indices of the superstrate and the substrate are well matched [35]. In contrast, Figure 3c corresponds to the low‐index superstrate condition (n = 1.34, empty ZIF‐8), showing a broader angular dispersion feature and no discernible SLR. The transmittance map remains largely featureless, with only weak modulation across the spectral range, suggesting inefficient far field coupling to support SLR.

In summary, the observed modulation of resonance quality represents a promising approach for sensing applications. The Au NP array/ZIF‐8 plasmonic system offers an “ON/OFF” resonance sensing concept, which could offer a compelling alternative to traditional wavelength‐shift‐based strategies.

Since the SLR resonance strongly depends on the refractive index matching between the superstrate and the substrate, another way to tune the optical response of the ZIF‐8 plasmonic grating system is to vary the substrate's refractive index. Here, we tested substrates of n = 1.4 (PDMS), 1.5 (glass), and 1.6 (high‐index glass) by creating the plasmonic gratings (with an Au⁰ concentration of ∼ 10 mm) on each substrate, followed by ZIF‐8 growth.

The n ≈ 1.4 substrate was obtained by spin‐coating a thick PDMS layer onto standard glass substrate [36]. As illustrated in Figure 4a, plasmonic arrays assembled directly on PDMS without an MOF overlayer displayed no discernible SLR features. This absence is primarily due to the significant refractive index mismatch between the top medium (air, n = 1.0) and the PDMS substrate (n ≈ 1.4), which induces strong out‐of‐plane scattering and dampens coherent plasmonic modes.

FIGURE 4.

Transmission spectra of Au NP array/ZIF‐8 structure based on different refractive index substrate. n = 1.4 (a), n = 1.5 (b) and n = 1.6 (c). The vertical scale bar in each graph indicates the magnitude of the Transmittance (%) in each case.

Upon growing a ZIF‐8 layer atop the Au NP arrays, the local refractive index above the NPs increased to approximately 1.34. This improvement in index matching with the PDMS substrate significantly enhanced the in‐plane coupling of the Au NP arrays, leading to the emergence of a narrow and well‐defined SLR peak. To investigate the system's sensitivity to further dielectric changes, the PDMS/Au NP array/ZIF‐8 structures were incubated in a series of solvents with progressively higher refractive indices. While the SLR peak was maintained, a slight broadening of the resonance was observed, attributed to increasing asymmetry in the surrounding optical environment.

In Figure 4b, we show similar experiments using plasmonic arrays on glass (n = 1.5), also prepared with a Au° concentration ∼10 mm, resulting in smaller clusters than those in Figure 2c. The higher refractive index of the glass substrate, relative to PDMS, enables the ZIF‐8 plasmonic system to exhibit well defined and sharp SLR dips in transmittance. These are especially pronounced when the ZIF‐8 captures a solvent with a higher refractive index, resulting in a superstrate that better matches the index of the glass substrate. These results underscore the combined role of NP density and substrate refractive index in determining both the onset and quality of SLRs (see Figure S12).

Finally, Figure 4c displays the optical response of samples assembled on a high‐index glass substrate (n = 1.6). In this case, all solvent‐loaded MOF layers, despite having different refractive indices, remained lower than that of the substrate. Consequently, the large index asymmetry across the structure persisted, suppressing the formation of strong SLR modes.

In general terms, we can observe that the spectral position of the SLR is largely determined by the refractive index of the substrate. For arrays with a lattice parameter of 600 nm, the resonance is observed near 820 nm when PDMS is used as the substrate. Conversely, for conventional glass substrates, the resonance is found at λ_SLR = 900 nm. Although the resonance is not clearly resolved for the high refractive index glass (n = 1.6), it can be perceived that the wavelength will shift to even longer wavelengths. These results demonstrate that the formation of well‐defined SLR modes critically depends on achieving good refractive index matching on both sides of the NP array (see Figure S10). When the top layer has a lower refractive index than the substrate, the resulting optical asymmetry induces radiation losses and dephasing, leading to broad and weak resonances. Conversely, when the superstrate refractive index approaches or exceeds that of the substrate, coherent lattice modes are preserved, giving rise to sharp and high‐Q resonances. These dependence on the refractive indices of both substrate and superstrate can be used as a key design principle for engineering sensitive and tunable plasmonic sensors based on SLR modulation.

We have demonstrated that designing a structure in which the plasmonic array is embedded in a homogeneous dielectric medium is essential for enhancing the performance of SLR‐based optical sensing. To fully exploit the potential of the MOF‐plasmonic system, we engineered a plasmonic array fully encapsulated within ZIF‐8 (ZIF‐8/Au NP array/ZIF‐8). As illustrated in Figure 5a, this architecture was fabricated by first growing a bottom layer of ZIF‐8 on a glass substrate, followed by the template‐assisted self‐assembly of Au NP arrays, and then a secondary growth of ZIF‐8 as the top layer. This MOF encapsulation establishes a highly symmetric dielectric environment surrounding the plasmonic array, which is critical for maximizing mode coherence. The successful fabrication and uniform integration of the Au NPs within this trilayer structure were confirmed by SEM imaging (Figure 5b–d). Our fabrication protocol, which combines MOF growth with templated self‐assembly, is key to obtain well‐ordered structures embedded in soft materials, producing architectures that are difficult to realize using conventional methods.

FIGURE 5.

Sandwich structure for improved refractive index sensitivity. Schematic illustration (a), top view SEM images of Au array fabricated on a ZIF‐8 film (b), and after the growth of the second ZIF‐8 layer (c). Side view SEM of the ZIF‐8/Au NP array/ZIF‐8 structure (d). The inset show a close up of the side view of the sandwich structure. Transmittance spectra of the sandwich‐structures after embedding different solvents (e). Comparison of refractive index sensitivity of the plasmonic grating embedded between two ZIF‐8 layers (f) versus the ZIF‐8 coated grating on glass (g).

Figure 5e shows the transmittance spectra of the sandwich‐like structure. Upon solvent exposure, both MOF layers (above and below the Au NPs clusters) absorb molecules, maintaining the NP array embedded in a homogeneous medium whose refractive index dynamically changes with molecular uptake. The spectra reveal a pronounced redshift of the SLR peak as the refractive index of ZIF‐8 increases. Compared to previous designs using MOF superstrates, this fully encapsulated configuration exhibits larger peak shifts for the same refractive index change, indicating enhanced sensitivity. At lower refractive indices, the SLR peak broadens due to stronger coupling with the LSPR, which increases radiative damping and reduces lattice mode coherence.

To quantitatively evaluate the improved sensing performance, we compared the resonance wavelength shift as a function of refractive index for two configurations: ZIF‐8/Au NP array/Glass and ZIF‐8/Au NP array/ZIF‐8 (Figure 5f,g). The sandwich configuration exhibited a sensitivity of approximately S = 427 ± 19 nm·RIU⁻¹, three times higher than the S = 148 ± 28 nm·RIU⁻¹ measured for the single‐layer structure. Moreover, due to the sharpness of the SLR features, the FoM, defined as FoM = S/FWHM, reaches values of 17.1 ± 0.8 RIU⁻¹, which is over five times greater than that of the single layer structure and an order of magnitude compared to LSPR modes (see Table S1). Additionally, embedding the plasmonic structure within a homogeneous dielectric medium improves the linearity of the refractive index response. We observed a more linear dependence compared to simpler architectures, with only slight deviations attributed to the finite thickness of the MOF layer, which results in a residual influence from the underlying glass substrate (see Figure S13–S15).

Finally, we benchmarked our plasmonic detector against existing visible‐range refractive index sensors (see Table S2). Our architecture offers competitive performance while being simpler to fabricate. Additionally, it demonstrates excellent robustness and reproducibility. These results underscore the promise of sandwich‐type MOF–plasmonic hybrids for next‐generation optical sensors, offering high sensitivity, improved linearity, and tunable properties.

3. Conclusion

We have presented a versatile and highly sensitive plasmonic sensing platform based on gold nanoparticle arrays integrated with ZIF‑8 films. The ZIF‐8 solvent exchange capability has been exploited herein to modulate the effective refractive index surrounding the plasmonic grating resulting in adynamic and reversible tuning of the SLR response. The template‐assisted assembly of Au NPs enabled the fabrication of 2D plasmonic gratings intercalated with MOF films in different configurations, hardly achievable with conventional nanofabrication techniques. We demonstrated that SLRs could be selectively switched ON and OFF depending on the refractive index matching between the superstrate and the substrate and the scattering strength of the 2D grating. Furthermore, we have achieved a fully encapsulated ZIF‑8/Au NP array/ZIF‑8 sandwich structure, the optimal case for refractive index sensing with SLRs, with enhanced sensing performance respect the rest of the configurations, reaching a refractive index sensitivity of 427 ± 19 nm·RIU^{− 1}. These findings provide valuable guidelines for designing MOF‑assisted plasmonic systems and pave the way toward responsive, tunable nanophotonic devices for next‑generation sensing applications.

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supporting File: smll72632‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.