Cationic metal detection using an Al–Al 2 O 3 nanoengineered SPR sensor

This study reports a surface plasmon resonance (SPR) sensor concept for detecting heavy metal ions in water using a cost-effective multilayer stack based on aluminum (Al), aluminum oxide (Al 2 O 3), and 2D nanomaterials. The architecture is numerically analyzed using the transfer matrix method (TMM) under TM-polarized illumination at 633 nm. Key design parameters, including prism material, Al and Al 2 O 3 thicknesses, and 2D nanomaterial coatings (GO, rGO, sSWCNT, and graphene), are optimized to improve sensitivity and resonance definition. CaF 2 is identified as the optimal prism, and an Al thickness of 40–45 nm combined with a 6 nm Al 2 O 3 layer provides a favorable trade-off between resonance sharpness and field confinement. Among the evaluated coatings, rGO yields the best overall performance. For Pb 2+ sensing, the optimized configuration achieves an angular sensitivity of 195.67°/RIU, a detection accuracy (DA) of 0.334, and a figure of merit (FoM) of 481.62 RIU −1. Electric-field analysis indicates strong confinement at the sensing interface. Overall, the proposed Al–Al 2 O 3 /2D SPR platform supports real-time, label-free detection of toxic metal ions and highlights aluminum-based stacks as a scalable alternative to noble-metal configurations for environmental sensing.

Heavy-metal contamination of water is a major global concern because even trace levels of toxic metals can pose serious risks to human health and ecosystems (Jahanjooz et al., 2014; Singh et al., 2024). Metals such as lead (Pb 2+) and mercury (Hg 2+) are particularly concerning because of their high toxicity, persistence, and bioaccumulation potential, even at parts-per-billion levels (De Silva et al., 2025). These contaminants originate from both natural and anthropogenic sources, including mining, industrial effluents, agricultural runoff, and improper waste disposal. Chronic exposure to heavy-metal ions has been associated with kidney and liver damage, neurological disorders, developmental deficits, and cancer (Singh et al., 2024; Velusamy et al., 2022). Consequently, the very low permissible limits established for drinking water create an urgent need for sensitive and reliable monitoring methods (De Silva et al., 2025).

Compact sensors for point-of-need detection of heavy-metal ions have attracted growing interest because chemical sensors can provide cost-effective, robust, and selective detection with good reproducibility and stability. By converting ion–surface interactions into measurable signals, they can also complement laboratory analyses through rapid and potentially continuous in situ water-quality monitoring (Velusamy et al., 2022).

Among available transduction methods, surface plasmon resonance (SPR) sensors have attracted considerable attention because they enable label-free quantification, high sensitivity to refractive-index changes, and real-time response (Anas et al., 2020; Lambert et al., 2020). In the Kretschmann configuration, polarized light excites surface plasmons in a thin metal film, and the resonance condition shifts with refractive-index changes near the interface, enabling detection of analytes located at or near the surface through resonance tracking (Lambert et al., 2020). Gold (Au) has been widely used because of its chemical inertness and strong plasmonic response in the visible range (Ashrafi and Mohanty, 2025). Recent studies have also explored AI-driven SPR design and analysis for refractive-index sensing, including multilayer stacks with dielectric interlayers and 2D coatings, and have reported high FoM values under their respective interrogation schemes and metric definitions (Runthala et al., 2025a; Runthala et al., 2025b; Shukla et al., 2024; Shukla and Arora, 2022a; Shukla and Arora, 2022b; Arora and Awasthi, 2019). Examples include AI-assisted optimization of Al-based SPR stacks incorporating fluorinated graphene (Runthala et al., 2025a; Runthala et al., 2025b), Al-based refractive-index sensing in the near-infrared using antimonene (Shukla et al., 2024) and graphene or MXene-type coatings (Shukla and Arora, 2022a; Shukla and Arora, 2022b), and Al nanostructure platforms for general refractive-index/thickness sensing based on spectral features (Arora and Awasthi, 2019). Recent optical-material studies have also highlighted the broader role of engineered nanophotonic and luminescent platforms in advanced photonic applications, including plasmon-enhanced sensing and multifunctional optical responses (Lu et al., 2025; Lin et al., 2025).

Aluminum (Al) has recently gained attention as a plasmonic material for SPR sensing because it is abundant, inexpensive, and compatible with CMOS process flows (Lambert et al., 2020; Dasgupta et al., 2025; Li et al., 2016). Its high free-electron density and more negative permittivity than Au or Ag support plasmonic resonances over a broad spectral range, from the ultraviolet to the near-infrared, including the deep-UV region where Au is ineffective (Lambert et al., 2020; Naik et al., 2013; Dubey et al., 2022; Knight et al., 2014). Efficient plasmonic responses have been reported in Al thin films and nanostructures across multiple regimes, often comparable to those of Au under appropriate conditions, supporting scalable sensing and on-chip photonics applications (Lambert et al., 2020; Shukla and Arora, 2025; Al Mamun and Yuce, 2020). Beyond metal selection, sensor performance depends strongly on the dielectric environment and resonance linewidth. Therefore, Al-based SPR platforms should be evaluated using sensitivity together with resonance depth and sharpness. In practice, high sensitivity can be offset by resonance broadening, which reduces angular resolvability and increases tracking uncertainty. A sensor with moderate sensitivity but a narrow, well-defined dip may therefore provide more reliable readout than a design optimized for sensitivity alone. This sensitivity–linewidth trade-off is also evident in recent Al/2D-material SPR designs (Shukla et al., 2024; Shukla and Arora, 2022a; Shukla and Arora, 2022b) and in optimized multilayer stacks that favor sharper resonances to improve effective FoM (Runthala et al., 2025a; Runthala et al., 2025b).

A key challenge for Al is its reactivity: bare Al readily oxidizes in air or aqueous media, forming a thin, stable Al 2 O 3 passivation layer. In sensing architectures, the Al–Al 2 O 3 interface can be treated as a designed structure, where a nanometric Al 2 O 3 coating improves aqueous stability while maintaining plasmonic sensitivity and tuning the coupling condition (Lambert et al., 2020; Roy et al., 2021; Oliveira et al., 2017). Vapor-deposited Al films with native oxide have shown favorable SPR behavior in the visible range, including steeper resonance dips and higher volumetric sensitivity than conventional Au films under comparable conditions (Lambert et al., 2020). The ultrathin Al 2 O 3 layer also protects the metal and may reduce nonspecific adsorption because of its dielectric properties (Almawgani et al., 2023). In this study, Al 2 O 3 is therefore treated as an engineered dielectric spacer that protects Al while enabling balanced optimization of resonance depth, linewidth, and sensitivity.

Integrating 2D materials into SPR designs can further improve sensitivity and stability by combining refractive-index interrogation with high surface reactivity (Tene et al., 2024). A nanomaterial overlayer above Al 2 O 3 may also improve stability in water (Mahesh et al., 2023). Accordingly, Al–Al 2 O 3 –2D architectures can be engineered for improved performance while retaining the cost and manufacturability advantages of Al/oxide stacks (Zhou et al., 2016; Ku et al., 2023). In this work, a CaF 2 /Al/Al 2 O 3 /rGO-based SPR configuration is numerically investigated for cationic heavy-metal sensing at 633 nm. Beyond structural optimization, the study clarifies the role of Al 2 O 3 as a dielectric spacer that modulates the interfacial field and resonance condition, while rGO acts not only as a sensing overlayer but also as an optical functional layer affecting sensitivity, linewidth, and related performance metrics. The results also highlight the sensitivity–linewidth trade-off in Al-based SPR sensor design.

Numerical modeling approach

The reflectance of the proposed N-layer SPR structure is computed using the transfer matrix method (TMM) (Wu et al., 2010; Tene et al., 2025). The analysis applies the boundary conditions for the tangential components of the electromagnetic fields across the multilayer stack, with the first interface taken at Z = Z 1 = 0, and the final interface at Z n−1.

In Equation 1, E 1, E N−1, H 1, and H N−1 denote the tangential electric and magnetic field components at the first and (N−1)th interfaces, respectively. The elements M ij represent the transfer-matrix coefficients of the N-layer structure. The overall transfer matrix is obtained as: where each individual layer matrix M k is given by:

We define the following:

In Equations 3–5, λ 0 is the free-space wavelength, n 1 is the refractive index of the incident medium, ε k is the dielectric constant of the k th layer, β k is the phase constant, θ is the angle of incidence, and d k is the thickness of the k th layer. A wavelength of λ 0 = 633 nm (He–Ne line) is assumed throughout the simulations.

Following direct computation, the overall reflectance of the N-layer structure is given by:

Using Equation 6, the reflectance as a function of the angle of incidence—i.e., the SPR curve—can be obtained.

We now define the performance metrics used to evaluate the proposed SPR sensor. The first parameter is the sensitivity enhancement, which compares the refractive-index sensitivity of the baseline configuration (DIW) with that obtained after introducing the target metal-ion condition in the sensing region. This enhancement is defined as:

The sensitivity S RI with respect to variations in the refractive index can be defined as: where Δθ represents the angular shift and Δn denotes the change in refractive index.

The detection accuracy (DA) can be evaluated using the angular shift and the full width at half maximum (FWHM) of the surface plasmon resonance (SPR) curve:

The Quality Factor (QF) is defined as the ratio between the refractive index sensitivity and the FWHM:

The Figure of Merit (FoM) is expressed as: where R min represents the minimum normalized reflectance value in the SPR curve.

The Limit of Detection (LoD) is given by:

Here, LoD is reported as a comparative metric derived from the adopted optical model and should be interpreted within the assumptions of the simulated framework.

Finally, the Combined Sensitivity Factor (CSF) can be computed as:

Here, R max is the maximum reflectance value before resonance, typically observed at non-resonant wavelengths or incidence angles.

All calculations in this study were performed with a data resolution of 5 × 10 4 points. A TM-polarized light source was assumed, as it is required for efficient surface plasmon excitation at the metal–dielectric interface.

Transfer-matrix modeling (TMM)

The transfer-matrix formalism used in this work—tangential-field boundary conditions, per-layer characteristic matrices, the global reflection coefficient, and the performance-metric definitions—is presented in Section 2.1 (Equations 1–13). A representative SPR reflectance line-shape and a literature benchmark of the N-layer TMM implementation are provided in the Supplementary Material (Supplementary Figure S1) (Cheon et al., 2014). All computations assume TM polarization.

Design and operation of the Al–Al 2 O 3 /nanomaterial SPR sensor

The proposed sensor uses a multilayer surface plasmon resonance (SPR) architecture consisting of an Al plasmonic film coated with a thin aluminum oxide (Al 2 O 3) layer. A prism coupler (BK7, CaF 2, BaF 2, or CsF) excites surface plasmons at the metal–dielectric interface in the Kretschmann configuration. The top surface of the metal/oxide stack is coated with a 2D nanomaterial layer, including pristine graphene (G), semimetallic single-walled carbon nanotubes (sSWCNT), graphene oxide (GO), and reduced graphene oxide (rGO). These carbon-based layers act as functional coatings that modify the effective optical environment at the sensing interface and enhance near-field interaction within the evanescent region. The refractive-index values and thickness parameters used for the prism and dielectric layers are summarized in Supplementary Table S2, and the corresponding material data sources are provided in Refs. (Akib et al., 2024; Sayed et al., 2023; Kumar et al., 2022; Xue et al., 2013; Song et al., 2020; Hossea and Rugumira, 2024).

In the sensing configuration considered here, deionized water (DIW) containing divalent metal ions (Hg 2+, Pb 2+, Ni 2+, Zn 2+, and Cu 2+) is represented by effective refractive-index perturbations in the analyte region adjacent to the 2D layer. This treatment is adopted as a first-order optical approximation for comparing the SPR response of different multilayer configurations under controlled conditions. It does not explicitly model adsorption kinetics, ionic-strength effects, competing species, or other matrix-dependent interfacial phenomena that may influence the experimental response in real aqueous systems. Within this framework, the analyte-induced perturbation changes the local dielectric response and shifts the SPR resonance angle.

A monochromatic TM-polarized beam is directed through the prism toward the Al/Al 2 O 3 /2D stack at a controlled incidence angle. At resonance, collective oscillations of free electrons in the Al film couple with the evanescent field, producing a pronounced reflectance minimum. The Al 2 O 3 layer is treated as an ultrathin spacer that both protects the Al surface and tunes the coupling condition by shaping evanescent-field penetration into the analyte. As shown in Figure 1, the prism refractive index (BK7, CaF 2, BaF 2, or CsF) produces distinct resonance positions at 633 nm. The choice of 2D coating further affects the resonance characteristics by modifying the interfacial optical response, and the resulting performance is evaluated using sensitivity together with linewidth- and depth-related metrics.

Scheme 1 also defines the architectures compared throughout the study and fixes the stacking order used in the parameter sweeps. The incidence angle, θ, is scanned until the surface plasmon mode is excited, as indicated by the reflectance minimum. The Al film thickness is optimized because it governs coupling efficiency and strongly affects resonance linewidth, which directly impacts FoM-related metrics.

Proposed fabrication and implementation workflow

A conceptual fabrication and implementation workflow is provided to contextualize the proposed Al–Al 2 O 3 /nanomaterial Kretschmann SPR architecture and its intended operation. This study is based exclusively on numerical SPR modeling and multilayer optimization; therefore, Scheme 2 is not presented as an experimentally executed procedure, but as a representative pathway consistent with established thin-film deposition and oxide-formation approaches that could, in principle, realize the simulated stack.

As outlined in Scheme 2, the workflow begins with substrate/prism preparation and establishment of the Kretschmann coupling geometry for angular SPR interrogation under controlled optical alignment. The Al thin film is then deposited by a physical vapor deposition route, such as sputtering or thermal/e-beam evaporation, corresponding to the plasmonic layer used in the multilayer calculations. The subsequent Al 2 O 3 formation step, achievable through controlled oxidation or oxide deposition, represents the spacer/protective layer that governs coupling conditions in the simulations and provides a more suitable interface for aqueous exposure. The workflow then includes integration of the rGO overlayer by coating or transfer onto Al 2 O 3, reflecting the 2D nanomaterial layer used in the model to tailor the near-field distribution and resonance characteristics. An optional surface-functionalization step is included only to indicate a generic route for introducing ion-recognition chemistry when selectivity is required; however, recognition kinetics, selectivity mechanisms, and regeneration protocols are outside the scope of the present numerical analysis. Finally, the structure is integrated into a flow cell for angular SPR measurement, consistent with the interrogation configuration assumed in the simulations, where resonance shifts and linewidth changes are interpreted in terms of effective refractive-index variations at the sensing interface.

Prism optimization

SPR sensors rely on careful matching of the optical properties of the prism, metal film, and surrounding media to maximize field confinement at the metal–dielectric interface. In this work, an Al–Al 2 O 3 /2D-material platform was numerically optimized for heavy metal-ion detection. The sensor operates in the Kretschmann configuration, where p-polarized light is coupled through prisms of different materials, namely calcium fluoride (CaF 2), barium fluoride (BaF 2), cesium fluoride (CsF), and BK7 glass. The resonance angle depends strongly on both prism refractive index and the refractive-index contrast between reference water (n = 1.330) and contaminated water (n = 1.340). Supplementary Table S1 lists the abbreviated designations used for each prism-based system.

Figure 1a shows the simulated reflectance curves as a function of incidence angle for the four prism systems in contaminated water. Each curve exhibits a pronounced minimum at the resonance angle, where surface plasmons are excited. The blue curve corresponds to Sys 1 –CaF 2 (CaF 2 /Al/CW), the green to Sys 2 –BaF 2, the red to Sys 3 –CsF, and the black to Sys 4 –BK7. The resonance angle is highest for CaF 2 (≈74°) and decreases for BaF 2 (69.15°), CsF (68.79°), and BK7 (65.33°). This behavior is illustrated in Figure 1b, which plots resonance angle as a function of prism refractive index. An almost linear decrease in resonance angle is observed as prism refractive index increases from 1.43 (CaF 2) to 1.52 (BK7). This trend occurs because a lower-index prism requires a larger incidence angle to satisfy the momentum-matching condition for SPR excitation. As a result, small analyte-induced refractive-index variations produce larger angular displacements, which is favorable for sensitivity. In contrast, higher-index prisms satisfy the coupling condition at lower angles and therefore provide a smaller angular response window.

The quantitative metrics extracted from the reflectance spectra are presented in Supplementary Table S3. The SPR peak position for Sys 1 –CaF 2 is 73.89°, approximately 4–9° higher than for the other prisms. The angular shift (Δθ) induced by changing the analyte from pure to contaminated water reaches 9.51° for CaF 2, whereas BaF 2 and CsF show shifts of 4.77° and 4.42°, respectively, and BK7 only 0.96°. This translates into a sensitivity improvement of approximately 14.78% for CaF 2 relative to the BK7 reference, while BaF 2 and CsF reach 7.41% and 6.86%, respectively. The attenuation, which quantifies the reduction in reflected intensity at resonance, is also favorable for CaF 2 (3.98%), compared with 5.13%, 5.21%, and 6.32% for BaF 2, CsF, and BK7, respectively. Although the full width at half maximum (FWHM) is slightly larger for CaF 2 (0.54) than for the other prisms (0.40, 0.41, and 0.33), the gain in sensitivity outweighs this modest broadening and indicates a favorable balance between signal displacement and linewidth. This behavior suggests that the stronger angular response of CaF 2 is accompanied by only limited resonance broadening, preserving adequate dip definition for angular interrogation.

The role of prism material is further clarified by the correlation between refractive index and resonance angle, summarized in Supplementary Table S4. CaF 2, with a refractive index of 1.4329 at 633 nm, exhibits a resonance angle of 73.89°, whereas BaF 2, CsF, and BK7 show values of 69.15°, 68.79°, and 65.33°, respectively. This near-linear trend indicates that a low-index prism provides a greater dynamic range for detecting analyte-induced refractive-index changes, because the slope of resonance angle versus analyte refractive index increases as prism index decreases. Since CaF 2 has the lowest refractive index among the candidates, it maximizes this slope and therefore enhances detection sensitivity. An additional practical advantage of CaF 2 is its low absorption in the visible range, which reduces propagation losses and improves signal contrast.

Figure 2 summarizes the performance of the four prism materials across multiple metrics. Panel (a) compares the angular shift of the four systems and shows that CaF 2 provides nearly twice the shift of BaF 2 and CsF, and about an order of magnitude more than BK7. Panel (b) shows the corresponding improvement in relative sensitivity, further confirming the superior performance of CaF 2. Panel (c) presents the attenuation at resonance, where CaF 2 yields the lowest value, indicating stronger resonance contrast. Panel (d) compares the FWHM values; although CaF 2 exhibits a slightly broader resonance than BK7, this increase is small relative to the substantial sensitivity gain. Overall, these results support the selection of CaF 2 as the prism material offering the best balance between angular response and acceptable linewidth.

Thickness Al layer optimization

Al has attracted considerable interest as a cost-effective alternative to noble metals such as gold and silver in SPR sensors because of its favorable electronic, optical, and physical properties. Among plasmonic materials, Al offers a high free-electron density and strong interband transitions in the UV and visible regions, enabling sharp SPR resonances even in relatively thin films (Tene et al., 2025; Cheon et al., 2014). Its low cost, abundance, CMOS compatibility, and mechanical stability also make it attractive for scalable sensor platforms (Akib et al., 2024).

To evaluate the effect of Al thickness on SPR performance under water (Sys 0) and contaminated-water (Sys 1 –Sys 4) conditions, simulations were performed for Al thicknesses from 30 to 45 nm; the corresponding model coding is listed in Supplementary Table S5. Figure 3a shows the angular reflectance curves for increasing Al thickness. A slight shift in the SPR angle is observed as the thickness changes, reflecting thickness-dependent coupling conditions in the Al film. As summarized in Supplementary Table S6 and shown in Figure 3b, the SPR angle for 30 nm Al (Sys 1) is 73.98°, while for 45 nm Al (Sys 4) it is 73.89°, indicating progressive angular stabilization with increasing thickness. This behavior is relevant for resonance tuning and sensor optimization. Physically, the Al thickness governs the coupling efficiency between the incident TM mode and the surface plasmon mode. Thinner films allow stronger field penetration across the metal layer, but they also increase optical damping and reduce resonance definition. In contrast, thicker films improve confinement of the plasmonic mode and produce a more stable resonance condition.

To determine the optimal Al thickness for heavy metal-ion sensing, the key performance metrics listed in Supplementary Table S7 and summarized in Figure 4 were evaluated. These metrics are relevant for balancing sensitivity, signal quality, and resonance sharpness.

Figure 4a shows that Δθ decreases slightly from 1.58° at 30 nm to 1.48° at 45 nm, indicating a modest reduction in angular response. Sensitivity follows the same trend (Figure 4b), with Sys 1 showing the highest value (2.18%) and Sys 4 the lowest (2.05%). In contrast, attenuation decreases markedly from 35.78% at 30 nm to 3.98% at 45 nm, indicating improved resonance contrast and lower optical loss in thicker Al films (Figure 4c). The FWHM also decreases substantially, from 1.76° at 30 nm to 0.54° at 45 nm, producing narrower and better-defined resonance curves (Figure 4d). Although sensitivity decreases slightly with increasing thickness, the improvements in resonance sharpness and signal quality make the 40–45 nm range more favorable for stable, low-noise sensing. Overall, the 40 nm Al film (Sys 3) provides the best balance among sensitivity (2.07%), attenuation (0.87%), and FWHM (0.72°), making it the preferred thickness for heavy metal-ion detection.

Effect of aluminum oxide on plasmonic metal

Al 2 O 3 plays an important role in improving and stabilizing SPR sensor performance, especially in aqueous environments where bare aluminum is prone to oxidation. Because of its chemical and thermal stability and favorable dielectric properties, Al 2 O 3 serves as both a passivation and tuning layer in plasmonic architectures. When incorporated into Al-based SPR structures, it can sharpen the resonance, reduce excessive damping, and modulate evanescent-field decay in the sensing medium, all of which are relevant for heavy metal-ion detection in water (Sayed et al., 2023).

In this study, a CaF 2 /Al/Al 2 O 3 multilayer configuration was evaluated for Al 2 O 3 thicknesses of 2, 4, 6, and 10 nm to determine how the oxide layer affects SPR performance. As shown in Figure 5a, the reflectance spectra exhibit a clear shift of the SPR minimum toward larger angles as the oxide thickness increases. The CaF 2 /Al/Al 2 O 3 (2 nm) structure (Sys 0) shows a resonance angle of approximately 74.5°, whereas the 10 nm Al 2 O 3 configuration (Sys 4) shifts the resonance to 77.6°, consistent with Supplementary Table S8 and Figure 5b. This shift indicates that the Al 2 O 3 interlayer modifies the effective dielectric environment and changes the SPR coupling condition. As the oxide thickness increases, the evanescent field extends differently across the spacer and sensing region, which alters the resonance position and the sensor response.

The optimal Al 2 O 3 thickness was determined by evaluating Δθ, sensitivity improvement, attenuation, and FWHM, as summarized in Supplementary Tables S9, S10 and illustrated in Figure 6. Figure 6a shows that Δθ increases from 1.53° at 2 nm (Sys 1) to 4.64° at 10 nm (Sys 4), indicating a broader angular response with increasing Al 2 O 3 thickness. Sensitivity improvement follows the same trend, increasing from 2.09% to 6.35% (Figure 6b), which is attributed to stronger plasmonic coupling and greater analyte interaction volume. However, the FWHM also increases from 0.79° to 1.20° (Figure 6d), indicating resonance broadening that may reduce detection accuracy. In contrast, attenuation remains nearly constant across all configurations, ranging from 0.92% to 0.97% (Figure 6c), which indicates that the oxide layer does not introduce substantial additional loss. These results show that increasing the Al 2 O 3 thickness improves angular response but also broadens the resonance, creating a trade-off between sensitivity and spectral definition. Overall, the 6 nm Al 2 O 3 configuration (Sys 3) provides the best balance, with a sensitivity increase of 4.01% and a Δθ of 2.92°, while maintaining low attenuation (0.979%) and a moderate FWHM (0.97°).

Nanomaterial optimization

The integration of two-dimensional (2D) and carbon-based nanomaterials into SPR sensors has emerged as an effective strategy for enhancing light–matter interaction and improving refractive-index sensitivity (Supplementary Table S12). Figures 7a,b compare four carbon-based overlayers—graphene oxide (GO), semiconducting single-wall carbon nanotubes (sSWCNT), reduced graphene oxide (rGO), and graphene (G)—deposited on a CaF 2 /Al/Al 2 O 3 stack. Because the dielectric conditions were kept identical across the configurations listed in Supplementary Table S11, the analysis isolates the effect of each nanomaterial’s complex refractive index (n+ik) on plasmonic resonance behavior.

Figure 7a shows the reflectance spectra, where all systems retain a well-defined transverse magnetic (TM) resonance dip, confirming the compatibility of the Al-based platform with these carbon overlayers. Figure 7b correlates θ SPR with the real part of the nanomaterial refractive index. According to Supplementary Table S13, Sys 1 (GO) exhibits a resonance at θ SPR = 77.33° for n = 1.27. As the refractive index increases to n = 2.04 for Sys 2 (sSWCNT), the resonance shifts to θ SPR = 78.01°. The largest angular shift is observed for Sys 3 (rGO), with θ SPR = 79.21° at n = 2.75. A notable exception occurs for Sys 4 (graphene). Despite having the highest real refractive index (n = 3.0), its resonance position returns to 78.01°. This nonmonotonic behavior is attributed to its high extinction coefficient (k = 1.149), which increases optical damping, alters phase matching, and shifts the resonance toward lower angles. These results indicate that nanomaterial optimization requires balancing dielectric loading from the real refractive index with damping introduced by absorption.

The results summarized in Supplementary Tables S12, S13 and shown in Figures 8a–d demonstrate that the nanomaterial overlayer strongly influences the SPR response. Among the evaluated systems, rGO (Sys 3) exhibits the largest angular shift (Δθ = 3.10°) and the highest sensitivity improvement (4.07%), which is associated with its moderate real refractive index (n = 2.75) and relatively high extinction coefficient (k = 0.41). This stronger interfacial light–matter interaction enhances the local electromagnetic field and improves sensitivity to small refractive-index changes in contaminated water. However, this gain is accompanied by higher attenuation (6.33%) and a broader FWHM (2.11°), indicating increased damping and reduced spectral resolution. In contrast, GO (Sys 1) shows the lowest sensitivity (1.60%) and Δθ (1.22°), but also the lowest attenuation (0.39%) and the narrowest FWHM (1.21°), indicating better spectral definition but lower detection capability. sSWCNTs (Sys 2) provide a more balanced response, with moderate sensitivity (2.50%) and Δθ (1.90°), together with lower attenuation than rGO. Pristine graphene (Sys 4) shows performance comparable to that of sSWCNTs, but with slightly lower sensitivity and Δθ. Overall, these results show that the optical response of the nanomaterial overlayer is governed by the interplay between resonance displacement and damping, rather than by the real refractive index alone.

Sensor performance

To evaluate the performance of the proposed CaF 2 /Al/Al 2 O 3 /rGO-based SPR sensor, a numerical sensing analysis was performed for divalent metal ions (Hg 2+, Pb 2+, Ni 2+, Zn 2+, and Cu 2+). The sensing environment was modeled by assigning each analyte an effective refractive-index condition in deionized water (DIW) and quantifying the resulting changes in the SPR response. Sys 0 denotes the DIW baseline, whereas Sys 1 –Sys 5 correspond to DIW containing Hg 2+, Pb 2+, Ni 2+, Zn 2+, and Cu 2+, respectively, under the same sensing configuration. Figure 9 shows the angular reflectance curves for all systems. Relative to DIW, the presence of metal ions produces measurable resonance-angle shifts, with the largest shift observed for Pb 2+. The inset highlights the resonance region and resolves the angular separation among analytes. These shifts are associated with small refractive-index differences at the sensing interface (Supplementary Table S14), with modeled values ranging from 1.33284 to 1.33328 (Wu et al., 2010). Under the adopted optical model, these refractive-index perturbations modify the local dielectric environment near the rGO/analyte interface and therefore shift the SPR coupling condition.

Pb 2+ yields the highest Δθ (0.642°) and the largest sensitivity increase (0.83%), followed by Ni 2+ (0.592°, 0.76%), Cu 2+ (0.567°, 0.73%), and Hg 2+ (0.562°, 0.72%). These trends are consistent with the small refractive-index differences assigned to each analyte within the adopted optical model. Figure 10a shows the angular displacement (Δθ) for each system, where Pb 2+ exhibits the largest shift, while DIW shows the smallest value (0.327°) and serves as the reference condition. Figure 10b presents the corresponding sensitivity values, again showing Pb 2+ as the most optically responsive analyte under the modeled conditions. The attenuation values in Figure 10c remain close to 5.7% for all systems, indicating similar resonance contrast across the analyzed ion conditions. Likewise, the FWHM values in Figure 10d range from 1.884° to 1.923°, with only minor broadening. This indicates that the analyte-dependent response is dominated mainly by resonance-angle displacement rather than by substantial changes in damping or resonance sharpness.

Supplementary Table S15 summarizes the optical constants and thicknesses of the layers used in the simulations. The thin rGO layer (1.41 nm) contributes to field confinement at the sensing interface while limiting excessive optical loss. Combined with the plasmonic response of Al, the multilayer configuration produces stable resonance features across the analyzed ion conditions. Overall, the rGO-functionalized sensor shows consistent optical performance for the modeled heavy metal ions, supporting its potential as a numerically optimized platform for aqueous heavy metal-ion sensing.

Sensor metrics

For the CaF 2 /Al/Al 2 O 3 /rGO-based SPR sensor designed for heavy metal ion detection, three essential performance parameters were evaluated: angular sensitivity (Sθ), quality factor (QF), and detection accuracy (DA). These metrics are important for assessing response magnitude and resonance definition in aqueous environments, and the corresponding numerical results are reported in Supplementary Tables S16, S17. The refractive indices of the analyzed systems range from 1.3317 (DIW) to 1.333284 (Pb 2+), providing the basis for angle-based sensitivity calculations.

Figure 11a shows that all ion systems exhibit high angular sensitivity values close to 195°/RIU, indicating strong refractive-index responsiveness at the rGO-functionalized interface. Pb 2+ yields the highest sensitivity (195.67°/RIU), followed by Ni 2+ (195.18°/RIU) and Zn 2+ (195.07°/RIU). The DIW reference system (Sys 1) exhibits the lowest sensitivity (192.70°/RIU), supporting the ability of the configuration to resolve small refractive-index variations within the modeled conditions. The trend plot in Figure 11a follows the increase in refractive index across systems.

Figure 11b summarizes the QF results, which combine sensitivity and linewidth and therefore reflect the balance between responsiveness and spectral sharpness. All systems yield QF values above 101.7 RIU −1, with the highest value observed for DIW (102.283), followed by Zn 2+ (102.022) and Cu 2+ (101.920). Although Pb 2+ maximizes Sθ, its slightly broader resonance reduces QF relative to DIW and Zn 2+, as seen in Figure 11b. This behavior illustrates the sensitivity–linewidth trade-off and supports the use of multiple complementary metrics for sensor evaluation.

Figure 11c evaluates resonance resolvability through DA, where higher values correspond to a sharper SPR dip and improved discrimination. Pb 2+ again exhibits the highest DA (0.334), consistent with the combination of a large angular shift and a relatively narrow FWHM. Ni 2+ and Cu 2+ also show high DA values (0.308 and 0.296, respectively), whereas DIW yields the lowest DA (0.173), consistent with the smallest optical perturbation. The trend in Figure 11c follows the same general ordering.

The metric variations observed in Figure 11 can be interpreted from the combined optical roles of the Al 2 O 3 and rGO overlayers in the plasmonic stack. The Al 2 O 3 layer acts not only as a protective coating for Al but also as a thin dielectric spacer that modifies interfacial field confinement and the SPR coupling condition. In turn, the rGO overlayer contributes not only to analyte interaction at the sensing interface, but also to the effective optical loading of the structure, thereby influencing both the resonance shift and the resonance width. Consequently, conditions that maximize angular sensitivity do not necessarily maximize QF or DA, because enhanced refractive-index responsiveness may be accompanied by linewidth broadening.

Figure 12 reports three advanced metrics derived from the same simulated SPR responses. Figure 12a, shows that the FoM values remain consistently high, ranging from 476.00 to 481.62 RIU −1 (Supplementary Table S18). Pb 2+ yields the highest FoM (481.62 RIU −1), indicating a favorable balance between sensitivity and linewidth, while Ni 2+ (480.59 RIU −1) and Zn 2+ (480.45 RIU −1) also exhibit strong performance. Figure 12b, presents the LoD values, reported here as a comparative metric within the adopted optical model, where Pb 2+ provides the lowest LoD (2.5552 × 10 −5 RIU), followed by Ni 2+ (2.5616 × 10 −5 RIU) and Zn 2+ (2.5631 × 10 −5 RIU), all lower than the DIW baseline (2.5946 × 10 −5 RIU). Figure 12c, and Supplementary Table S18 show that Pb 2+ also achieves the highest CSF (485.97), followed by Ni 2+ (484.93) and Zn 2+ (484.80), whereas DIW yields the lowest CSF (480.30). Collectively, these results indicate that the rGO-functionalized CaF 2 /Al/Al 2 O 3 configuration provides strong overall performance across FoM, LoD, and CSF for the investigated ions under the modeled conditions.

Electric field distribution analysis

Figure 13 presents the normalized electric-field intensity profiles along the depth axis, extending from the Al layer through the dielectric layers into the analyte region, consisting of DIW and DIW containing different heavy metal ions. The layered structure is color-coded in the background: the green region corresponds to Al, the orange region to the Al 2 O 3 dielectric spacer, the narrow pink region to the rGO sensing layer, and the white region to the analyte medium, where refractive-index variations are introduced.

All analyte-dependent profiles show strong electric-field confinement at the Al/Al 2 O 3 interface, with a pronounced enhancement near the rGO/analyte boundary. This behavior indicates efficient SPR excitation and substantial evanescent-field penetration into the sensing region, both of which are favorable for refractive-index-based detection. The electric-field intensity decays rapidly beyond the rGO interface, confirming that most of the SPR field remains localized near the surface, which is desirable for interfacial sensing.

Differences among the analytes are more clearly observed in the zoomed inset of Figure 13. The DIW profile shows the lowest field intensity beyond approximately 150 nm, whereas Pb 2+ and Hg 2+ exhibit slightly higher values in the same region. Although these differences are small, they are consistent with the modeled refractive-index variations and with the relative trends in sensitivity and angular shift discussed in the previous subsections. Overall, the field profiles support the ability of the CaF 2 /Al/Al 2 O 3 /rGO structure to maintain strong plasmonic field localization at the sensing interface.

Comparation

To evaluate the performance of the proposed SPR sensor for cationic metal-ion detection, Table 1 compares the present design with representative SPR configurations incorporating low-dimensional materials. Rather than considering angular sensitivity alone, the comparison also includes resonance-shape metrics (QF and DA), since angular tracking precision depends strongly on linewidth and dip definition. Accordingly, a moderate sensitivity combined with a sharper resonance can provide more robust angular readout than sensitivity-only optimization. In addition, the proposed design treats Al 2 O 3 as an engineered dielectric spacer within an Al-based SPR platform, rather than as a purely passive native-oxide layer.

Wei et al. (Kumar et al., 2022) reported the highest sensitivity (626.00°/RIU) using a multilayer BK7/Si/Cu/Ni/Si/MXene (Ti 3 C 2 Tx) structure; however, QF and DA were not reported, which limits direct comparison beyond sensitivity. Jia et al. (Xue et al., 2013) achieved 201.00°/RIU using a Prism/TaS 2 configuration, indicating lower sensitivity under the reported conditions. Hosse et al. (Wu et al., 2010) and Malek et al. (Song et al., 2020) reported sensitivities of 339.45 and 298.55°/RIU, respectively, using CaF 2 /Ag/BaTiO 3 and Prism/Ag/Al 2 O 3 /Ni/BlueP/WSe 2 stacks; however, these studies do not consistently report resonance-shape metrics, which again limits sensitivity–resolution benchmarking. Tene et al. (Hossea and Rugumira, 2024) reported a CaF 2 /Cu/Si 3 N 4 /rGO configuration for Pb(II) with a sensitivity of 318.21°/RIU, QF of 75.97, and DA of 0.24. Because that study reports QF and DA in addition to sensitivity, it provides a useful reference for assessing the sensitivity–linewidth trade-off.

In the present work, Al is used as the plasmonic layer to enable a cost-effective platform, while Al 2 O 3 serves as a nanometric spacer and passivation layer, and rGO acts as the interfacial overlayer in the modeled sensing region. Across the evaluated systems, the sensor maintains consistently high QF (≈101.7–102.3 RIU −1) and DA (≈0.17–0.33), indicating well-defined resonances and improved angular resolution under the simulated conditions. For example, the DIW case yields 192.70°/RIU, QF = 102.283, and DA = 0.173, whereas the Pb(II) case yields 195.67°/RIU with the highest DA (0.334). The Hg(II), Ni(II), Zn(II), and Cu(II) cases also show strong performance, with QF values above 101 and DA values ranging from 0.289 to 0.308. These results are particularly relevant because several previous studies report sensitivity only, which limits broader performance comparison. Finally, the CaF 2 prism selected in this study provides favorable coupling in the modeled configuration and a broad optical transmission window for operation at 633 nm.

This study presents a numerical investigation of a multilayer Al–Al 2 O 3 SPR sensor functionalized with 2D nanomaterials for heavy metal-ion detection in water. The structure was optimized by varying prism material, Al thickness, Al 2 O 3 interlayer thickness, and 2D overlayer. Compared with conventional Au- and Ag-based SPR designs, the Al-based platform offers lower cost and greater compatibility with scalable fabrication.

Of the tested prism materials, CaF 2 provided the best response at 633 nm, yielding the largest resonance displacement and high sensitivity under the modeled conditions. An Al thickness of 40–45 nm provided a suitable balance between resonance definition and signal strength. The Al 2 O 3 interlayer improved interfacial stability and acted as a dielectric spacer that influenced field confinement and resonance coupling. A thickness of 6 nm provided the best trade-off among angular shift, sensitivity, and FWHM.

Among the evaluated 2D coatings, rGO showed the best overall performance, producing the largest angular shift (Δθ = 3.10°) and sensitivity increase (4.07%) relative to the other nanomaterials, although with slightly higher attenuation and FWHM. Overall, the optimized CaF 2 /Al/Al 2 O 3 /rGO configuration demonstrates the potential of Al-based multilayer SPR sensors as cost-effective and scalable platforms for practical water-monitoring applications. These results should be interpreted as optical design guidance under controlled modeling assumptions, while experimental response in real aqueous systems may also depend on adsorption, ionic-strength effects, competing species, and other matrix-dependent interfacial phenomena.