Plasmonic Nanoparticle-Metal–Organic Framework Hybrid Nanostructures for Biomedical Applications: From Design to Function

Abstract

Plasmonic nanoparticle-metal–organic framework (plasmonic NP-MOF) hybrid nanostructures have emerged as distinctive platforms that synergistically combine the physical, chemical, and biological properties and functions of plasmonic nanoparticles and MOFs, enabling biomedical capabilities inaccessible to either component alone. This Perspective focuses on a central design-to-function relationship: the structural paradigm between the plasmonic nanoparticle and the MOF governs how the properties of the two components complement one another and interact synergistically. We consider four principal configurations: plasmonic NPs embedded within MOF pores, plasmonic NPs decorate the MOF exterior, and a plasmonic NP serves as the core of a plasmonic@MOF core–shell structure or forms the shell of a MOF@plasmonic NP core–shell structure. Each configuration offers distinct functional advantages suited to specific biomedical applications. Here, we first review the synthetic strategies and formation mechanisms underlying these four hybrid nanostructures. We then examine how these structure-dependent properties manifest across three major biomedical application areas: biosensing, drug delivery, and therapeutic applications. Finally, we identify the synthetic limitations constraining structural precision, reproducibility, and material versatility, as well as translational barriers that must be addressed to advance plasmonic nanoparticle–MOF hybrids toward robust and clinically relevant biomedical probes and platforms.

1. Introduction

Plasmonic nanoparticles (plasmonic NPs) exhibit optical properties arising from localized surface plasmon resonance (LSPR), which results from the coherent excitation of the collective oscillation of conduction electrons in a metal nanostructure by an incident electromagnetic field. This resulting resonance depends on the material composition, size, shape, and local dielectric environment of the nanoparticle, concentrates electromagnetic energy near its surface, and can be synthetically tuned across the visible-to-near-infrared (NIR) window. −

Metal–organic frameworks (MOFs) are porous crystalline materials self-assembled from metal ions or clusters as nodes and organic linkers through coordination bonds. − Their pore architecture and surface chemistry can be rationally tailored through deliberate selection of building blocks, including metal nodes and organic linkers, as well as postsynthetic linker functionalization. Combined with their tunability, their exceptionally high surface areas and porosities give rise to a molecular-enrichment effect, whereby target analytes, drugs, and reporters are efficiently concentrated within the pores. Pore apertures with appropriate dimensions and chemical functionality can also confer a molecular-sieving effect by selectively admitting target molecules while excluding interfering species. Furthermore, although many MOFs exhibit considerable thermal and structural stability, − such stability is not universal and strongly depends on the node-linker chemistry, framework defects, surface modification, and the surrounding chemical environment. Accordingly, for biomedical applications, it is essential to select an MOF with the stability and physicochemical characteristics appropriate for its intended purpose.

Integrating plasmonic NPs with MOFs into structurally defined hybrid nanoarchitectures produces synergistic properties unattainable by either component alone. A unifying design principle is the deliberate spatial positioning of analytes or functional cargo relative to the plasmonic near-field. Because electromagnetic field enhancement decays rapidly within nanometers of the metal surface, the MOF serves not as a passive support but as an active structural mediator that concentrates molecules near hotspot regions and can stabilize the plasmonic nanoparticle against aggregation under physiological conditions. The compatibility of the MOF pore architecture and surface chemistry with the target molecules governs molecular confinement, surface immobilization and stimuli-responsive cargo release in a given application. This design-to-function relationship manifests across three principal biomedical domains: biosensing, drug delivery, and therapeutic applications. Each area leverages a distinct combination of the complementary facets provided by the hybrid architecture. In biosensing, the MOF enriches analytes near plasmonic hotspots and provides molecular-sieving selectivity, while the plasmonic component enables ultrasensitive surface-enhanced Raman scattering (SERS) and LSPR-based detection. In drug delivery, the stimuli-responsive lability of MOF coordination chemistry can be coupled with NIR-absorbing plasmonic core to enable multistimuli-triggered, spatiotemporally controlled cargo release within the tumor microenvironment (TME). In therapeutic applications, the plasmonic component can drive photothermal conversion, reactive oxygen species (ROS)-generating catalytic activity, and photoacoustic contrast generation, while the MOF can supply stimuli-responsive cargo release and catalytic amplification of oxidative stress. The structural motif determines how these roles are partitioned and coupled into ROS-mediated, photothermal, or image-reporting therapeutic modalities. On this basis, we classify plasmonic NP-MOF hybrid nanostructures into four structural categories according to the spatial relationship between two components: plasmonic NP-embedded MOF, plasmonic NP-decorated MOF, plasmonic NP@MOF core–shell, and MOF@plasmonic NP core–shell structures. Table summarizes the relationship among structural motifs, functional attributes and biomedical applicability by highlighting the characteristic properties and predominant biomedical applications of each motif.

1. Structure–Property–Application Relationships of Plasmonic NP-MOF Hybrid Nanostructures.

Hybrid structure	Synthetic strategy	NP/MOF	Intrinsic properties	Biomedical applications	References
Plasmonic NP-embedded MOF	Metal-organic chemical vapor deposition (MOCVD)	Au/MOF-5, Au/ZIF-8	LSPR, SERS, Catalytic/nanozyme activity, Photothermal conversion (NIR), MRI contrast (Fe-based MOF nodes)	SERS biosensing, Electrochemical sensor, Antibacterial/infected tissue regeneration, Drug delivery and chemo-photothermal therapy, MRI-guided theranostics, Stimuli-responsive MOF degradation	–
	Liquid impregnation (external reductant)	Au/MIL-101(Cr), Au/PCN-224(Fe), Au/MOF-199, Au/UiO-66, Au/UiO-67, Au@Ag/ZIF-8
	Liquid impregnation (redox-active reduction)	Ag/[{Ni(C10H26N6)}3(bpdc)3]
	Liquid impregnation (photochemical reduction)	Ag/ZIF-8, Ag/UiO-66-NH2, Ag/MIL-125-NH2
	Liquid impregnation (Post-synthetic pore functionalization)	Au/MIL-101-NH3 +, Ag/NU-902, Au/NU-1000, Au/MOF-1, Au/MOF-2, Au/MOF-3
Plasmonic NP-decorated MOF	Ligand-bridged coordination anchoring	MIL-101/AgS, MOF@Au or Ag (MIL-101, UiO-66, MOF-808)	LSPR, SERS, Magnetic separation, Interfacial charge separation,Nanozyme, Photothermal (NIR, PDA-assisted)	SERS biosensing, Photocatalysis, Chemodynamic (CDT)/Fenton cascade therapy, Antibacterial therapy	,,–
	Electrostatic assembly	Fe3O4/MOF-867/Au, HKUST-1/AgNPs, ZIF-8/AuNPs
	Node-templated in situ-nucleation	MIL-101(Fe)/Ag, ZIF-8/Au
MOF@Plasmonic NP Core–Shell	Node-templated in situ shell growth	ZIF-8@Au (ZACS NCs)	LSPR, SERS, Quantitative molecular fingerprinting, Sub-nm facet selectivity	Quantitative SERS detection, Molecular fingerprinting, SERS sensing platform	,
	Ligand-bridged facet-selective growth	ZIF-8 RD@Au/Ag, ZIF-8 NC@Au/Ag
Plasmonic NP@MOF Core–Shell	Capping-mediated nucleation (ionic surfactant: CTAB/CTAC)	Au nanosphere/nanorod/nanobipyramid/nanostar@ZIF-8, AuNR@HKUST-1	LSPR, SERS (MOF shell-gated analyte access to plasmonic core), Molecular sieving, Photothermal conversion (NIR), Stimuli-responsive cargo release (pH/ GSH/NIR), High drug loading capacity, Single-crystal lattice alignment, Tunable shell thickness, High-yield single-core encapsulation	SERS biosensing, Dual-trigger drug delivery, PTT and PDT combination therapy, PA imaging + PTT theranostics, GSH-triggered self-assembly (PA + PTT)	,– ,,
	Capping-mediated nucleation (nonionic polymer: PVP/PEG)	AuNBP@ZIF-8 (end/waist/full), Au@ZIF-8 (single/multicore)
	Capping-mediated nucleation (PDA)	Au@ZIF-8, Au@UiO-66
	Kinetically controlled nucleation	AuNR@NU-901, Au@ZIF-8, Au@ZIF-67, Au@FJU-30, Au@MIL-88(Fe), Au@HKUST-1, Au@MOF-74(Co)

2. Structural Paradigms of Metal–Organic Frameworks Hybridized with Plasmonic Nanoparticles

In this section, we classify plasmonic NP-MOF hybrid nanostructures according to synthetic sequence and resulting structural configuration and review the synthetic strategies and formation mechanisms associated with each paradigm (Figure ). We consider four principal configurations. In plasmonic NP-embedded MOF structures, plasmonic NPs are formed within the pore network of a preformed MOF, whereas in plasmonic NP-decorated MOF structures, the NPs are deposited on the external surface of a preformed MOF. The remaining two configurations are core–shell architectures: MOF@plasmonic NP structures, in which a MOF core is surrounded by a conformal plasmonic shell, and plasmonic NP@MOF structures, in which a MOF shell is grown around a presynthesized plasmonic NP core.

1.

Schematic overview of the four structural paradigms for plasmonic nanoparticle–MOF hybrid nanostructures. a. Plasmonic NP-embedded MOF structures: metal precursors are infiltrated into the preformed MOF host and subsequently reduced in situ within the internal pores, yielding metal nanoparticles confined within the MOF cavities. b. Plasmonic NP-decorated MOF structures: metal precursors directly anchor to Lewis-acidic metal nodes on the MOF exterior and undergo subsequent reduction to generate surface-bound nanoparticles (top pathway), while preformed metal nanoparticles are anchored onto the MOF surface through coordination between the nanoparticle surface and a bifunctional ligand bound to CUSs (bottom pathway), as detailed in the magnified insets. Both routes converge to yield metal nanoparticles localized on the outer MOF surface. c. MOF@plasmonic NP core–shell structure: preformed MOF crystals are exposed to a metal precursor solution, followed by reduction to yield a conformal plasmonic metal nanoshell grown on the outer surface of the MOF, producing MOF@plasmonic metal core–shell architectures in which the porous MOF interior is preserved as the core and the plasmonic component forms the encapsulating shell. d. Plasmonic NP@MOF core–shell structures: preformed metal nanoparticles of varying morphology (sphere, rod, and cube) nucleate MOF crystal growth at their surface, directing conformal MOF shell formation around the nanoparticle core to yield MOF-encapsulated Plasmonic NP structures.

2.1. Plasmonic NP-Embedded MOF Structure

In embedded structures, a number of small plasmonic NPs are confined within the cavities and pore network of a preformed MOF host. Synthesis typically involves infiltration of gas- or liquid-phase metal precursors into the pore network of a presynthesized MOF, followed by in situ decomposition or reduction to form plasmonic NPs within the framework. Three key factors govern synthetic success: (1) the driving force for precursor penetration, which influences precursor loading capacity; (2) the affinity of plasmonic metal precursors or nuclei for the MOF pore walls, which controls spatial confinement; and (3) the reduction conditions, including method, reductant identity, and reaction time, which influence not only the size and morphology of the resulting plasmonic NPs but also the structural integrity of the MOF. Insufficient MOF–metal interaction or partial framework degradation during reduction can cause plasmonic NPs to migrate to the external surface, where they agglomerate and compromise performance.

2.1.1. Metal–Organic Chemical Vapor Deposition (MOCVD)

MOCVD is a solvent-free, gas-phase loading method pioneered by Fischer and coworkers that uses volatile organometallic precursors as the metal source, with the vapor pressure generated during sublimation serving as the driving force for precursor penetration into the MOF pores. In their initial demonstration, they exposed preformed MOF-5 to the vapor of the volatile organometallic precursor [(CH₃)­Au­(PMe₃)] (PMe₃ = trimethylphosphine) under static vacuum, followed by treatment with H₂ gas as a reactive gas (Figure a). A key advantage of MOCVD is that it does not require displacement of solvent molecules occupying the MOF pores, affording a higher loading capacity than solvent-based liquid impregnation methods. However, this overall process was relatively long, requiring about 5 h. Moreover, the weak affinity of Au precursors for the MOF-5 pore walls drove precursor diffusion toward the external surface, leading to low internal Au loading and the formation of agglomerated Au nanoparticles (AuNPs) with a broad size distribution of 5–20 nm.

2.

Synthesis and characterization of embedded plasmonic NP-MOF hybrid structures. a. Powder X-ray diffraction patterns of a) MOF-5 and b) Au/MOF-5 after reductive H₂ treatment, with an inset TEM image confirming AuNP formation within the MOF-5 framework via MOCVD. Adapted with permission from ref . Copyright 2005, John Wiley and Sons. b. Schematic illustration of Au@ZIF-8 embedded structure synthesis via solid grinding of [(Me)₂Au­(acac)] with ZIF-8 crystals followed by H₂ reduction, and representative TEM image of AuNPs/ZIF-8. Adapted with permission from ref . Copyright 2009, American Chemical Society. c, d. TEM image and energy-dispersive X-ray (EDX) spectrum, respectively, of AuNPs/MIL-101 fabricated via liquid impregnation, confirming successful AuNP incorporation within the MIL-101 framework. Adapted with permission from ref . Copyright 2014, American Chemical Society. e. Schematic representation of postsynthetic amine grafting onto the CUSs of MIL-101: perspective view of the mesoporous cage with hexagonal windows (i); generation of CUSs from Cr trimers upon vacuum treatment at 423 K (ii); selective amine (ethylenediamine) grafting onto CUSs (iii); and subsequent embedding of noble metal NPs within the amine-functionalized MIL-101 (iv). Yellow, pale gray, and red represent Cr atoms/octahedra, C atoms, and O atoms, respectively. Adapted with permission from ref . Copyright 2008, John Wiley and Sons.

To accelerate precursor penetration relative to sublimation-driven MOCVD, Xu and coworkers introduced a solid-grinding approach. Figure b illustrates the solid-grinding process. The volatile organometallic precursor [(CH₃)₂Au­(acac)] (acac = acetylacetonate) was physically mixed with ZIF-8 crystals, generating localized vapor that rapidly diffused into the pores. This process yielded embedded 3–4 nm AuNPs with a substantially reduced penetration time to approximately 30 min compared with conventional MOCVD. Nevertheless, both MOCVD and solid-grinding methods have a limited precursor scope because many volatile organometallic compounds are air- and moisture-sensitive and therefore require stringent handling conditions. Reactive metal precursors may also compromise framework integrity, while MOCVD requires relatively long process times.

2.1.2. Liquid Impregnation

Liquid impregnation offers a practical and versatile alternative to MOCVD. In this method, dissolved plasmonic metal salt precursors are impregnated into the pores of a preformed MOF by capillary pressure and subsequently converted into plasmonic NPs by chemical or photochemical reduction. As a representative example, AuNP-embedded MIL-101­(Cr) was fabricated by immersing presynthesized MIL-101­(Cr) in an HAuCl₄ solution, followed by reduction with sodium citrate as an external reducing agent. The hydrophilic, zeolite-like solid framework and high surface area of MIL-101­(Cr) promoted preferential infiltration of the HAuCl₄ precursor into the internal pore network (Figure c,d). Notably, AuNP loading could be tuned by varying the HAuCl₄ concentration, while the AuNP diameter could be independently controlled by adjusting the sodium citrate concentration, maintaining a narrow size distribution. Furthermore, the inherent water stability of MIL-101­(Cr) allows the Au/MIL-101­(Cr) hybrid composite to be used in aqueous media.

Liquid impregnation has been applied to diverse plasmonic NP-MOF combinations, including Au/PCN-224­(Fe) prepared using NaBH₄, and Au/MOF-199, Au/UiO-66, and Au/UiO-67 prepared using sodium citrate. Reductant-free strategies have also been demonstrated. The intrinsic redox activity of Ni²⁺/Ni³⁺ macrocyclic nodes in [{Ni­(C₁₀H₂₆N₆)}₃(bpdc)₃] (bpdc = 4,4′-biphenyldicarboxylate) enables Ag⁺ reduction without an external reductant, yielding Ag nanoparticles (AgNPs). UV-driven photoreduction has likewise been used to prepare Ag/ZIF-8, Ag/UiO-66-NH₂, and Ag/MIL-125-NH₂. By eliminating the need for external chemical reductants, these approaches can minimize reductant-derived chemical impurities. Moreover, sequential impregnation of Au and Ag precursors followed by NaBH₄ reduction yielded small bimetallic Au@Ag core–shell NPs with a narrow size distribution. Liquid impregnation offers considerable flexibility in the selection of MOF building blocks, allowing water-stable frameworks to serve as the host lattice and potentially broadening the applicability of the resulting composites to solution-phase sensing and in vivo biomedical applications. Nevertheless, capillary forces and additional hydrophilic interactions alone may provide insufficient driving force for metal precursor infiltration, resulting in limited metal loading. Therefore, this limitation has motivated the development of the postsynthetic functionalization strategies that enhance metal precursor uptake and promote the formation of well-dispersed plasmonic NPs with narrow size distributions within MOFs.

Postsynthetic pore functionalization addresses this limitation by modifying coordinatively unsaturated metal sites (CUSs) or introducing functional organic sites on the pore surfaces. In the former approach, amine grafting at the CUSs of MIL-101 generated −NH₃ ⁺ sites that electrostatically attract anionic Au precursors, which were subsequently reduced to yield Au–NH₃ ⁺/MIL-101. Postsynthetic solvent-assisted ligand incorporation (SALI) installed −SO₃ ^– K⁺ groups at the metal nodes of NU-902, enabling Ag⁺ loading through K⁺/Ag⁺ ion exchange to yield Ag/NU-902. In NU-1000, grafting 4-carboxy-phenylacetylene onto the Zr₆ nodes enabled metalation of terminal alkyne sites with monometallic Au­(I)­PEt₃ ⁺ ions, thereby yielding Au/NU-1000 (PEt₃ = triethylphosphine) (Figure e). , In the latter approach, terminal alkyne functionalization of the organic linkers exploited the alkynophilicity of Au³⁺, enabling AuNP loadings of approximately 50 wt % in MOF-1, MOF-2, and MOF-3.

Building on the conventional liquid impregnation, reaction–diffusion coupling can direct spatially selective NP deposition by exploiting differences in precursor deposition rate. In addition, confinement within anisotropic one-dimensional MOF channels can template the growth of ultrathin metal nanowires, with their diameter and crystallographic orientation dictated by the channel geometry. These examples demonstrate that liquid impregnation can extend beyond simple NP embedding to provide control over the morphology of the resulting hybrid nanostructures.

Plasmonic NP-embedded MOF structures can suppress NP aggregation, constrain NP growth within the available pore dimensions, and concentrate a large number of closely spaced plasmonic NPs within a single MOF scaffold. However, precise control over the spatial distribution and number of plasmonic NPs within each MOF particle remains challenging. Occupation of the pore network by embedded plasmonic NPs can also reduce the accessible surface area and pore volume, potentially attenuating the molecular enrichment and sieving functions relevant to biomedical applications.

2.2. Plasmonic NP-Decorated MOF Structure

In plasmonic NP-decorated MOF hybrid nanostructures, a number of metal NPs are attached to the external surface of a preformed MOF. In this section, we focus on the central synthesis strategies by which plasmonic metal precursors or preformed plasmonic NPs interact with the external MOF surface during synthesis. We categorize these attachment modes into three classes: (1) Ligand-bridged coordination anchoring, in which metal ions or preformed plasmonic NPs are immobilized via molecular bridging ligands; (2) Electrostatic assembly, in which difference in surface charge drives physisorption-dominated attachment; (3) Ligand-free precursor adsorption and reduction, in which Lewis-acidic CUSs or specific MOF surface facets serve as nucleation sites for the formation and growth of plasmonic NPs from solution-phase metal precursors.

Understanding the mechanisms by which plasmonic NPs are anchored to the MOF exterior provides a basis for controlling NP loading and spatial distribution, thereby enabling rational tuning of the functional performance of plasmonic NP-decorated MOF hybrids.

2.2.1. Ligand-Bridged Coordination Anchoring

Ligand-bridged coordination anchoring describes a strategy in which a molecular linker provides defined coordination sites for both the MOF framework and the plasmonic metal domain, thereby translating the crystalline order of the MOF surface into spatially defined attachment sites. Lin et al. exploited cysteamine as a bifunctional molecular linker. Its amine group coordinates to the CUSs of MIL-101, while its thiol group serves as a docking site for plasmonic metal cations such as Cu²⁺ or Ag⁺. Subsequent S^2– anion exchange converts the coordinated metal ions into surface-bound metal sulfide NPs. This coordination-assisted strategy lowers the interfacial energy barrier and promotes surface-attached nucleation and epitaxial growth with intimate interfacial contact under mild conditions. Sun et al. coated several MOFs, including MIL-101, UiO-66, and MOF-808, with a conformal polydopamine (PDA) layer in Tris buffer (Figure a). The PDA coating provides abundant redox-active catechol groups that enable the localized reduction of [Ag­(NH₃)₂]⁺ or HAuCl₄, leading to the formation of Ag or Au seeds decorated on the MOF surface. These strategies anchor plasmonic NPs through well-defined interfacial chemistry, including coordination linkages and redox-active interlayers.

3.

Synthesis and characterization of decorated plasmonic NP-MOF hybrid structures. a. Schematic illustration of a PDA-mediated ligand-bridged anchoring strategy for the synthesis of MOF/Plasmonic nanoparticles (MOF/PNPs): MOF crystals are first coated with a conformal PDA layer to generate MOF/PDA seeds, onto which plasmonic nanoparticles are subsequently reduced and grown to yield MOF/PNPs. Reproduced with permission from ref . Copyright 2025, American Chemical Society. b. Schematic illustration (top) of the preparation process for HKUST-1/Ag plasmonic films by electrostatic assembly and SEM images (bottom) of (i) Ag colloids, (ii) HKUST-1, (iii) HKUST-1/Ag NPs. Reproduced from ref . Copyright 2023, Elsevier. c. HAADF image and corresponding EDX element mapping (Cu, Zn, O, C) of ZIF-8/Cu prepared by ligand-free precursor adsorption–reduction via aqueous NaBH₄, showing Cu⁰ nanoparticles (bright contrast) localized on the outer ZIF-8 crystal surface with Zn retained in the framework interior. Reproduced with permission from ref . Copyright 2023, American Chemical Society. d. Schematic illustration of ligand-free in situ Au decoration on ZIF-8: Ce₆-loaded ZIF-8 nanocrystals (prepared from Zn²⁺ and 2-methylimidazole with PVP as stabilizer) are decorated with ∼ 4 nm AuNPs through adsorption of HAuCl₄ onto the ZIF-8 surface followed by rapid NaBH₄ reduction. Adapted with permission from ref . Copyright 2019, Royal Society of Chemistry.

2.2.2. Electrostatic Assembly

Electrostatic assembly exploits complementary surface charges between plasmonic NPs and MOF surfaces to drive spontaneous adsorption without covalent modification; when charge contrast is sufficient, the resulting Coulombic attraction yields uniform decoration while preserving the intrinsic framework structure. In citrate-capped Au or Ag nanoparticles, adsorbed citrate anions confer a net negative surface charge that stabilizes the particles through electrostatic repulsion in colloidal solution. When these negatively charged plasmonic NPs are mixed with MOFs whose external surface is positively charged, Coulombic attraction drives spontaneous adsorption of plasmonic NPs onto the MOF surface. The balance between NP-MOF electrostatic attraction and NP-NP repulsion leads to uniform decoration rather than uncontrolled aggregation. Zhao et al. demonstrated that this principle extends even to MOFs lacking sufficient native surface charge: functionalizing Fe₃O₄/MOF-867 with (3-aminopropyl)­trimethoxysilane (APTES) introduced surface amine groups that enabled the electrostatic adsorption of preformed AuNPs, producing Fe₃O₄/MOF-867/Au composites without requiring direct coordination between the plasmonic NPs and MOF nodes. Beyond colloidal mixtures, the same electrostatic principle extends to thin film and substrate architectures: Zhang et al. fabricated HKUST-1/Ag composite substrates by first electrodepositing HKUST-1 onto copper foil, then electrostatically adsorbing Ag colloids onto the HKUST-1 surface (Figure b). And Thao Anh et al. employed an immersion-based electrostatic self-assembly approach to deposit preformed AuNPs onto ZIF-8 thin films to yield a homogeneous plasmonic surface with uniformly distributed AuNPs.

Electrostatic assembly offers a conceptually simple and synthetically mild route to decorate MOF surfaces with plasmonic NPs, because it relies only on surface charge contrast rather than specific ligand chemistries or redox conditions. Nevertheless, electrostatic assembly is intrinsically limited by the weak, reversible nature of physisorption because small variations in pH, ionic strength, or solvent polarity can readily disrupt the Coulombic attraction at the MOF-nanoparticle interface.

2.2.3. Ligand-Free Precursor Adsorption–Reduction

This strategy relies on direct adsorption of plasmonic metal precursors onto MOF surfaces, followed by in situ reduction to generate plasmonic NPs without additional anchoring ligands. Plasmonic NP growth is therefore governed primarily by precursor adsorption and reduction conditions, rather than pre-existing anchoring chemistry. Zan et al. employed Cu­(II)-impregnated ZIF-8 precursors to demonstrate that the reduction protocol alone determines the final location of CuNPs (Figure c). High-temperature H₂ reduction at 340 °C produced CuNPs mainly localized within the ZIF-8, whereas aqueous NaBH₄ reduction generated CuNPs predominantly on the external crystal surface together with ZnO impurities arising from partial framework degradation. This direct comparison highlights how ligand-free precursor adsorption followed by in situ reduction can selectively yield either embedded or surface-decorated ZIF-8/Cu architectures by adjusting the reduction conditions. Building on this principle that reduction conditions alone dictate NP localization, subsequent studies applied ligand-free adsorption–reduction to decorate MOF surfaces with a single plasmonic metal. Ma et al. synthesized ZIF-8 nanocrystals and decorated them with Au nanoparticles by directly adding HAuCl₄ followed by rapid NaBH₄ reduction in the ZIF-8 dispersion (Figure d). HAuCl₄ was reduced in situ on the ZIF-8 surface to generate ∼4 nm AuNPs without the use of additional anchoring ligands. A related but mechanistically distinct route replaces the external reductant with a linker that both anchors and reduces the precursor: Jiang et al. functionalized MIL-101­(Fe) with tannic acid through complexation with coordinatively unsaturated Fe­(III) sites, after which tannic acid reduced Ag⁺ in situ to generate AgNPs directly on the outer surface of the MOF. Extending this strategy from single- to multimetal systems, Dash et al. applied a adsorption–reduction strategy in which UiO-66-NH₂ was dispersed in a water/methanol mixture and subsequently exposed to AgNO₃ or mixed Ag/Pd precursor solutions, allowing the metal precursors to adsorb onto the MOF prior to reduction. NaBH₄ solution was then added dropwise under ice-bath conditions to reduce the adsorbed Ag⁺ and/or Pd²⁺ in situ, producing mono- and bimetallic nanoparticles deposited on the UiO-66-NH₂ surface without introducing additional anchoring molecules.

In ligand-free precursor adsorption–reduction, the MOF surface transitions from a passive support into an active nucleation template. Taken to its logical extreme, this nucleation-directing capacity of the MOF surface can be extended to enable not merely isolated particle attachment but laterally continuous shell growth, a structural transition that defines the distinct motif examined in the following section.

2.3. MOF@Plasmonic NP Core–Shell Structure

MOF@plasmonic NP core–shell structures are formed through continuous plasmonic shell growth across the MOF surface. This architecture requires heterogeneous nucleation to remain confined to the MOF surface, where laterally growing nuclei merge into a conformal shell. CUSs provide the structural basis for this process by serving as preferential adsorption and nucleation sites for metal precursors.

This principle is illustrated by the ZIF-8@Au core–shell nanocube (ZACS NC) developed by Shin et al. Highly crystalline ZIF-8 nanocubes expose reactive {100} facets that direct the growth of a uniform Au nanoshell. This work represents one of the first structurally well-defined plasmonic shells grown directly on a MOF surface and establishes a core–shell motif distinct from conventional nanoparticle decoration. AuCl₄ ^– ions preferentially adsorb onto the coordinatively unsaturated Zn²⁺ sites on the {100} facets of ZIF-8 nanocubes. Sequential addition of HAuCl₄ and sodium ascorbate then triggers reduction and nucleation at these sites, accompanied by partial diffusion of Zn atoms to generate an interfacial AuZn₃ alloy layer; this is followed by lateral growth into a continuous, conformal Au nanoshell (Figure a). The formation of the AuZn₃ alloy was evidenced by additional X-ray diffraction (XRD) reflections at 2θ = 22.68°, 25.2°, 27.14°, and 42.69°, indexed to the (200), (210), (211), and (321) planes of the alloy, together with a corresponding shift in the Au 4f X-ray photoelectron spectroscopy (XPS) binding energies, including an additional Au 4f_5/2 ³⁺ peak at 90.77 eV consistent with an Au–Zn alloyed environment. In contrast, Li and Zeng demonstrated that 3-mercaptopropionic acid (MPA) can bridge AuNPs to ZIF-8 via carboxylate coordination to surface Zn²⁺ sites, enabling single-layer decoration of ∼1–3 nm ultrafine plasmonic NPs selectively on the {110} facets of rhombic dodecahedral ZIF-8 or the {100} facets of ZIF-8 nanocubes (Figure b). While this represents an exceptionally facet-precise formation of a decorated structure rather than a conformal shell, it highlights how crystallographic surface periodicity can be exploited across both motifs to direct NP assembly with subnanometer spatial control.

4.

Synthesis and characterization of MOF@plasmonic NP core–shell structures. a. Three-step mechanism for ZACS NC formation and structural characterization. (top) Schematic of conformal Au nanoshell growth on ZIF-8 nanocubes: Step 1, selective adsorption of AuCl₄ ^– onto coordinatively unsaturated Zn²⁺ sites on the {100} facets; Step 2, Au nucleation accompanied by partial Zn diffusion to form an interfacial AuZn₃ alloy layer; Step 3, continued Au reduction and lateral growth to yield a uniform, conformal Au nanoshell enclosing the ZIF-8 core (ZACS NC). EDS elemental mapping confirming the presence of a gold nanoshell on the ZIF-8 nanocubes (bottom left) and line profiles across the center of the nanocubes, demonstrating the composition and distribution of elements within the core–shell structure (bottom right). Scale bar, 100 nm. Reproduced with permission from ref . Copyright 2025, American Chemical Society. b. Facet-selective AuNP deposition on ZIF-8 via MPA-bridged coordination anchoring. (left) Schematic illustrations for the deposition of AuNPs on the exterior surfaces of two different types of ZIF-8 crystals (RD and NC crystals in light purple) to form the ZIF-8@Au hybrid nanocomposites, and the epitaxial growth of outer ZIF-8 shells to produce various ZIF-8@Au@ZIF-8 hybrid nanocomposites. (right) TEM images of the resulting ZIF-8 RD@Au (i–iii) and ZIF-8 NC@Au (iv–vi) hybrid nanocomposites at increasing magnification, confirming preferential NP localization on {110} and {100} facets, respectively. Reproduced with permission from ref . Copyright 2013, American Chemical Society.

When surface-confined nucleation is sustained sufficiently for lateral coalescence, a continuous film forms rather than isolated nuclei. In this regime, the MOF’s crystallographic topology and node chemistry encode hotspot density and plasmonic mode structure into the framework itself, affording single-particle-level control over light–matter interaction and structural reproducibility.

2.4. Plasmonic NP@MOF Core–Shell Structure

In plasmonic NP@MOF core–shell structures, single- or multicore plasmonic NPs are covered by the MOF as a shell, which preserves its intrinsic porosity. The synthesis of plasmonic NP@MOF core–shell structures rests on a single thermodynamic condition: heterogeneous nucleation at the metal nanoparticle surface must be kinetically and energetically preferred over homogeneous nucleation in the bulk solution. When this condition is met, the MOF crystal grows outward from the plasmonic metal surface to produce a conformal shell. The strategies to enforce this preference fall into two mechanistically distinct categories: (1) Capping-mediated approaches, which exploit the chemistry of the preadsorbed surface layer to lower the interfacial free energy barrier at the plasmonic metal surface; and (2) Kinetically controlled approaches, which manipulate the supersaturation profile in time and space, ensuring that the local linker concentration at the plasmonic NP surface transiently exceeds the heterogeneous nucleation threshold.

2.4.1. Capping-Mediated Nucleation

In capping-mediated encapsulation, the surfactant or polymer preadsorbed on the plasmonic NP surface lowers the interfacial free energy barrier at the plasmonic NP-MOF contact and thereby biases MOF nucleation toward the nanoparticle surface. Three families of capping agents have been widely exploited for engineering plasmonic NP@MOF core–shell structures: ionic surfactants, which enable electrostatic templating; nonionic polymers, which mediate steric and hydrogen-bond interactions; and PDA, which provides covalent or coordination bridging.

2.4.1.1. Ionic Surfactants: Electrostatic Templating

Cetyltrimethylammonium surfactants such as cetyltrimethylammonium bromide (CTAB) and cetyltrimethylammonium chloride (CTAC) form bilayer-like or hemimicellar adlayers on plasmonic metal surfaces, projecting positively charged head groups into solution that attract anionic MOF precursors and bias heterogeneous nucleation at the metal–solvent interface. In these systems, the balance between surface-directed and bulk nucleation correlates strongly with surfactant concentration relative to the critical micellar concentration (CMC): below the CMC, conformal shells form preferentially at the coated surface, whereas at higher concentrations excess surfactant increasingly redirects crystallization to the bulk solution. Zheng et al. demonstrated this principle using CTAB-stabilized plasmonic NPs across various geometries including nanospheres, nanorods, and nanoprisms, achieving single-particle encapsulation within ZIF-8, thereby demonstrating that the accessible encapsulation size range extends beyond the nominal pore size of ZIF-8. Liu et al. extended this to a 3–60 nm continuous range by covarying CTAB with precursor stoichiometry (Figure a). Hu et al. further revealed that the CTAB adlayer can mediate crystallographic alignment, directing Au{100}//ZIF-8{110} single-crystal-to-single-crystal orientation (Figure b).

5.

Synthesis and characterization of plasmonic NP@MOF encapsulated structures. a. Plasmon-coupled SERS substrate in which ZIF-8 fills plasmonic nanogap hotspots on an Au film to enrich organic molecules and enhance Raman readout. Reproduced from ref . Available under a CC-BY 4.0 license. Copyright 2024, Yi Liu et al. b. CTAB-directed metal@ZIF-8 core–shell nanocomposites showing one-to-one encapsulation of single metal nanocrystals within single-crystalline ZIF-8 and lattice alignment between metal {100} and ZIF-8 {110} facets. Reproduced with permission from ref . Copyright 2014, American Chemical Society. c. PVP-mediated controlled encapsulation of one or two types of nanoparticles into ZIF-8 crystals, enabling tunable spatial distributions (central vs off-center, single vs dual components) via timed nanoparticle addition during MOF growth. Reproduced with permission from ref . Copyright 2012, Springer Nature. d. Layer-by-layer growth of HKUST-1 shells on Au nanorods, where increasing Cu²⁺/BTC deposition cycles yields progressively thicker and more uniform MOF coatings while preserving rod morphology. Reproduced with permission from ref . Copyright 2018, American Chemical Society. e. Au@ZIF-8 core–shell nanostructure viewed along a representative ZIF-8 crystallographic axis, highlighting a single metal nanoparticle encapsulated within a crystalline ZIF-8 lattice that serves as a porous diffusion barrier and structural scaffold. Reproduced from ref . Available under a CC-BY 4.0 license. Copyright 2025, Guoming Lin et al.

2.4.1.2. Nonionic Polymers: Interfacial Engineering

In nonionic polymer-mediated encapsulation, the polymer layer modifies the plasmonic NP-MOF interface to direct heterogeneous nucleation at the nanoparticle surface. Under CTAB-mediated conditions, ZIF-8 preferentially nucleates at the tips of Au nanobipyramids (AuNBP) to form AuNBP/end-ZIF and, upon continued growth, fully encapsulated AuNBP@ZIF structures. By introducing thiol-terminated methoxy poly­(ethylene glycol) (mPEG-SH), which preferentially adsorbs onto the side facets of Au nanobipyramids, Yang et al. redirected heterogeneous nucleation to the waist region, enabling the synthesis of waist-selective AuNBP/waist-ZIF architectures. Another representative nonionic polymer, poly­(vinylpyrrolidone) (PVP), forms hydrated coordination that lowers the plasmonic NP-MOF interfacial free-energy barrier through steric accommodation and mild hydrogen bonding with Zn²⁺/linker precursors. This soft interfacial layer biases heterogeneous nucleation toward the plasmonic NP surface. Chen et al. demonstrated that Zn²⁺/linker stoichiometry controls the heterogeneous-to-bulk nucleation balance, confirming the active nucleation-directing role of PVP. Subsequent work on Au@ZIF-8 core–shell nanoparticles showed that, under PVP-mediated soft-interface conditions, reaction time and temperature can be used to switch deliberately between single- and multicore architectures while independently tuning ZIF-8 shell thickness over a range of approximately 3–50 nm. Lu et al. demonstrated that PVP-functionalized nanoparticles can be uniformly encapsulated within individual ZIF-8 crystallites by exploiting the enhanced affinity between the PVP-modified nanoparticle surface and the growing coordination-polymer spheres (Figure c). The adsorbed PVP layer facilitates nanoparticle incorporation through coordination interactions between the pyrrolidone carbonyl groups and the growing MOF framework, together with hydrophobic interactions. Importantly, the spatial distribution of encapsulated plasmonic nanoparticles can be programmed by controlling the timing of nanoparticle addition during MOF crystallization: nanoparticles introduced at the onset of crystal growth become confined to the crystal center, whereas sequential addition during crystal growth results in their incorporation into the outer regions of the growing crystal, enabling controlled multicore architectures within a single ZIF-8 crystallite.

2.4.1.3. Polydopamine Interlayers: Universal Adhesive Encapsulation

PDA deposits conformally on any substrate under mild alkaline conditions, with catechol groups anchoring to the plasmonic NP surface and amine/catechol moieties coordinating MOF metal nodes to nucleate heterogeneous shell growth. Zhou et al. showed that this strategy could be adapted to Au nanospheres and nanostars, establishing PDA as a substrate-independent MOF-nucleating interfacial platform rather than a conventional capping ligand. Moreover, the redox activity of PDA enables localized plasmonic metal deposition at the interlayer, allowing construction of sandwich-type plasmonic metal–PDA–MOF architectures that integrate plasmonic components within the encapsulated structure.

Beyond simple thickness tuning, layer-by-layer (LbL) deposition on suitably functionalized Au cores enables truly conformal HKUST-1 shells whose thickness increases linearly with the number of deposition cycles. Hinman et al. used CTAB-capped Au nanorods that were subsequently converted into COOH-terminated, negatively charged substrates via ligand exchange with 11-mercaptoundecanoic acid (MUA). Alternating exposures to Cu­(OAc)₂ and the tetralkylammonium salt of benzene-1,3,5-tricarboxylate (BTC) then deposited HKUST-1 shells in a truly LbL fashion, with Cu²⁺ coordinating to MUA/BTC carboxylates at the interface while BTC^3– propagated the framework and maintained a strongly negative surface potential that suppressed aggregation over many growth cycles. This sequential CTAB to MUA to BTC surface modification provides a practical strategy for transforming generic plasmonic nanoparticles into LbL-ready substrates for conformal MOF shell growth while enabling Å-scale control over MOF shell thickness (Figure d).

Beyond simply forming a MOF shell, this capping-mediated nucleation strategy enables precise tuning of shell thickness, site-selectivity, and crystallographic registry.

2.4.2. Kinetically Controlled Nucleation

In kinetically controlled nucleation, precursor kinetics are regulated to favor heterogeneous nucleation at the plasmonic NP surface while suppressing competing homogeneous nucleation. Lin et al. provided the most direct mechanistic evidence for this strategy. At low MOF precursor concentrations, ZIF-8 nucleated at a single nucleation site on the AuNP surface and grew as a single-crystalline cuboidal shell at a rate of ∼40 nm² s^–1. Raising the precursor concentration increased the growth rate to ∼160 nm² s^–1, triggering simultaneous nucleation at multiple sites and generating a polycrystalline, nonuniform shell alongside an increased population of empty ZIF-8 crystals (Figure e). At the process-design level, Osterrieth et al. demonstrated that slow syringe-pump addition of the H₄TBAPy linker into a gold nanorod (AuNR)/Zr⁴⁺ precursor solution controlled the temporal linker concentration, thereby favoring MOF nucleation on the AuNR surface over homogeneous nucleation in solution. This strategy increased the yield of single-AuNR@NU-901 core–shell particles to >99%, substantially outperforming conventional rapid addition. Vapor-assisted triethylamine (TEA) deprotonation provides a broadly applicable alternative, in which TEA vapor gradually diffuses into the reaction medium to enable controlled linker deprotonation and heterogeneous MOF nucleation. The generality of this strategy was demonstrated across seven MOF shell types and six nanoparticle core types.

Taken together, these studies highlight three principal kinetic handles for establishing conditions that favor core–shell growth in plasmonic NP@MOF hybrid nanostructures: precursor concentration, which governs the instantaneous supersaturation and the number of nucleation events initiated on the plasmonic NP surface; precursor addition rate, which controls the temporal evolution of supersaturation and enables heterogeneous growth to outcompete homogeneous nucleation; and linker activation rate, which decouples the supply of coordinating species from their chemical activation.

3. Biomedical Applications of Plasmonic Nanoparticle-Metal-Organic Framework Hybrid Nanostructures

The structural paradigms described in Section translate into distinct functional advantages across three biomedical domains, each defined by a complementary division of roles between the plasmonic and MOF components. Critically, the choice among the embedded, decorated, MOF@plasmonic NP core–shell, or plasmonic NP@MOF core–shell motif is not an incidental synthetic choice but a primary determinant of which functional modes are accessible, as discussed follow. In biosensing, the MOF sieves and concentrates target analytes while the plasmonic component transduces molecular recognition into amplified SERS or LSPR readouts. In stimuli-responsive drug delivery, the MOF provides a programmable cargo reservoir whose release is gated by photothermal or chemical stimuli generated by the plasmonic NP. In theranostics, the plasmonic component simultaneously enables imaging contrast through photoacoustic signal generation or SERS readout and therapeutic activation via photothermal therapy (PTT) or ROS-mediated cytotoxicity. Meanwhile, the MOF integrates stimuli-responsive cargo release and catalytic amplification of oxidative stress, thereby coupling diagnostic feedback to therapeutic output within a single nanoplatform. Across all three domains, performance is determined not only by the precise functional partitioning at the plasmonic NP-MOF interface but also by either component independently.

3.1. Biosensing

Biosensing most clearly illustrates the functional division inherent to plasmonic NP-MOF hybrids: the MOF regulates molecular access to the plasmonic metal surface, enriches low-abundance analytes near electromagnetic hotspots, and filters interfering species, while the plasmonic component converts this gated molecular recognition into amplified optical readouts. Within this sensing paradigm, the embedded and plasmonic NP@MOF core–shell structural motifs operate through distinct mechanisms: embedded architectures exploit the dense inter-NP hotspot network formed by multiple small plasmonic particles confined within a single MOF scaffold, whereas plasmonic NP@MOF core–shell structures leverage the conformal MOF shell as a molecular sieve that simultaneously controls analyte proximity to the plasmonic core. Early embedded-structure Au/MIL-101 systems established the core design principle: AuNPs provide LSPR-mediated Raman enhancement while MIL-101 concentrates analytes near hotspots, enabling the detection of rhodamine 6G and benzidine at a limit of detection (LOD) of 41.75 and 0.54 fmol, respectively, as well as the direct detection of alpha-fetoprotein in human serum (Figure a). Au@ZIF-8 core–shell particles demonstrated that ZIF-8 shell thickness directly controls analyte-hotspot proximity. Among shells ranging from 3 to 50 nm in thickness, the 3 nm shell produced the maximized SERS response for volatile organic compounds (VOCs) including toluene, ethylbenzene, and chlorobenzene.

6.

MOF-Gated Signal Transduction: Plasmon-Enhanced Biosensing. a. Au/MIL-101 composite for SERS-based biosensing in human serum. AuNPs embedded within MIL-101 concentrate analytes near plasmonic hotspots while the MOF pore aperture excludes interfering macromolecules, enabling detection of rhodamine 6G (LOD 41.75 fmol), benzidine (0.54 fmol), and alpha-fetoprotein directly in human serum (blank, blue; serum, red; spiked at 20 ng/mL, black). Reproduced with permission from ref . Copyright 2014, American Chemical Society. b. ZACS NCs for spatially aligned quantitative SERS. Unlike randomly oriented 2-methylimidazole (2-mIM) on Au nanocubes (2-mIM/AuNC), 2-mIM linkers within the ZIF-8 lattice maintain fixed dihedral angles (75° < ϕ < 104.5°) relative to the Au nanoshell, enabling concentration-independent molecular orientation. SERS intensities at eight 2-mIM vibrational modes (574–1468 cm^–1) show statistically significant changes upon doxorubicin (DOX) loading at 928, 1135, and 1468 cm^–1. Reproduced with permission from ref . Copyright 2025, American Chemical Society. c. Au nanocrystal@ZIF-8 particles placed on Au films produced nanoparticle-on-mirror nanogap hotspots, enabling quantitative detection over eight orders of dynamic range. Reproduced from ref . Available under a CC-BY 4.0 license. Copyright 2024, Yi Liu et al. d. AAO@Ag/Ag@ZIF-8 particle-in-cavity (MMPIC) architecture for wearable in situ SERS sensing in human biofluids. Three cascaded enhancement mechanisms, namely ZIF-8 molecular enrichment (I), ZIF-8 shell light confinement within an Ag nanolayer cavity (II), and AAO@Ag conical nanocavity light-trapping (III), concentrate hotspots within the ZIF-8 matrix (maximum field intensity 34.4 × E₀). Concentration-dependent Raman spectra for NH₃ detection in exhaled breath. The measured NH₃ levels distinguish healthy volunteers from individuals with respiratory conditions across morning, noon, and evening time points. Reproduced with permission from ref . Copyright 2026, American Chemical Society.

Spatial alignment of MOF-mediated enrichment with plasmonic hotspots is decisive for quantitative sensing performance. This principle is most directly realized in core–shell architectures through two complementary mechanisms. In plasmonic NP@MOF core–shell structure, the MOF shell enables analyte to access proximately to the plasmonic core. In MOF@plasmonic NP structures, the periodic and spatially aligned arrangement of organic linkers within the MOF core enables quantitative SERS measurements by producing higher linear relationships between particle concentration and SERS intensity. In AuNBP@ZIF-8 hybrids, ZIF-8 was selectively deposited at the tips, the sites of maximum electromagnetic field enhancement. This tip-selective deposition increased the SERS enhancement factor for the 4-aminothiophenol (4-ATP) from 5.33 × 10⁶ to 3.15 × 10⁷, with the improvement attributed to tip-localized analyte enrichment. ZIF-8 nanocubes with a conformal Au nanoshell grown from {100} Zn²⁺ Lewis acid sites (ZACS NCs) yielded linear SERS–concentration relationships with R² = 0.99 across eight vibrational modes for doxorubicin, crystal violet, and brilliant blue G (Figure b). Au nanocrystal@ZIF-8 particles coupled to an Au film were developed for SERS detection of VOCs. The ZIF-8 shell serves as both a nanogap spacer and a molecular enrichment layer: VOC molecules diffuse into the porous ZIF-8 and are trapped within the plasmon-coupled nanogap hotspot between the Au nanocrystal and Au film, where the electric field is enhanced by factors of 10²–10⁵. Concentration-dependent SERS intensities of benzene (993 cm^–1), toluene (1000 cm^–1), o-xylene (733 cm^–1), and formalin (1071 cm^–1) span a dynamic range of more than 8 orders of magnitude with an LOD down to 10^–2 mg m^–3 (Figure c). This design was further integrated into wearable formats using anodic aluminum oxide (AAO)@Ag/Ag@ZIF-8 particle-in-cavity arrays. These arrays combine ZIF-8-mediated molecular enrichment and light scattering with light trapping by the AAO@Ag cavities. This platform extended plasmonic NP-MOF SERS to NH₃ detection in exhaled breath and pH/glucose sensing in sweat.

Plasmonic NP-MOF hybrids can further integrate catalytic signal generation with SERS amplification. This approach can be achieved through post-anchoring transformation of plasmonic NP@MOF structures, where the chemical evolution of the anchored metal phase yields nanozyme activity alongside plasmonic enhancement. AuNBP@CuZn-MOF core–shell nanozymes leverage AuNBP tip hotspots for SERS enhancement, while the CuZn-MOF shells catalyze the peroxidase-like oxidation of 3,3′,5,5′-tetramethylbenzidine (TMB); glutathione (GSH) scavenges oxidized TMB, enabling ratiometric GSH quantification over a concentration range of 1 to 150 μM with an LOD of 0.091 μM and a simultaneous colorimetric readout. Similarly, decorated AgNPs/MIL-101­(Fe) systems couple MOF-catalyzed H₂O₂-mediated 2,2′-azinobis­(3-ethylbenzithiazoline-6-sulfonic acid) (ABTS) oxidation with plasmonic signal readout for indirect dopamine detection over a concentration range from 1.054 pM to 210.8 nM with an LOD of 0.32 pM in human urine. In this system, AgNPs are grown in situ at the MOF outer surface via node-templated nucleation, generating dense surfactant-free hotspots. These nanozyme-coupled platforms demonstrate that the MOF can function not only as a passive filter or concentrator but as an active catalytic signal-generating layer, expanding plasmonic NP-MOF biosensing beyond direct molecular fingerprinting.

3.2. Drug Delivery

In drug delivery applications, the functional division between plasmonic NP and MOF components is most clearly realized in the plasmonic NP@MOF core–shell structural motif. The conformal MOF shell simultaneously serves as a high-capacity cargo reservoir, a stimuli-responsive release gate, and a colloidal stabilizer for the plasmonic core, while the encapsulated NP provides an externally controllable photothermal trigger inaccessible to MOF-only carriers. The mass-transport advantage of plasmonic NP@MOF over embedded architectures is particularly consequential here: whereas embedded NPs occlude MOF pores and reduce drug loading capacity, plasmonic NP@MOF core–shell structures preserve the full internal pore volume of the MOF shell for cargo storage while the plasmonic NP core remains structurally isolated. This pore accessibility translates directly into high drug loading capacity: the exceptional surface area of ZIF-8 (>1,300 m² g^–1) enables loading densities difficult to achieve with conventional carriers. The inherent lability of metal–organic coordination bonds further provides a stimuli-responsive release mechanism without additional chemical modification, since mildly acidic pH (∼6.5) and elevated GSH in the TME weaken these bonds.

For example, AuNR@ZIF-8 core–shell nanostructures exemplify dual-trigger design. NIR irradiation excites the longitudinal LSPR mode of the AuNR, generating localized heat at the plasmonic NP-MOF interface, thermally weakening Zn²⁺-2-methylimidazolate bonds; TME acidity independently protonates the linker through a distinct mechanism, and both thermal and chemical stimuli act synergistically (Figure a). Doxorubicin (DOX)-loaded AuNR@ZIF-8, with a loading efficiency of 0.358 mg DOX per mg of carrier, released substantially more cargo under acidic condition (pH 5.5) combined with NIR irradiation than under either stimulus alone or at physiological pH 7.4, yielding in vivo tumor suppression of approximately 90% for combined PTT and chemotherapy versus approximately 58% for PTT alone and approximately 30% for chemotherapy alone, as demonstrated in a mouse tumor model. Au nanostar@ZIF-8 architectures extend this design. Capping-mediated heterogeneous nucleation at the branched nanostar surface yields a conformal ZIF-8 shell directly at the plasmonic NP-MOF interface. The sharp-tip geometry concentrates photothermal energy at this interface more efficiently than rod geometries, triggering shell degradation at reduced NIR power densities with demonstrated intracellular light-triggered release in vitro (Figure b).

7.

Representative encapsulated Au/MOF hybrid nanostructures for stimuli-responsive drug delivery. a. Synthesis and dual-stimuli-responsive drug release mechanism of AuNR@ZIF-8 core–shell nanostructures. PVP-stabilized AuNRs are encapsulated within a DOX-loaded ZIF-8 shell; NIR excitation of the longitudinal AuNR LSPR generates localized heat that weakens Zn²⁺-2-methylimidazolate bonds, while tumor microenvironmental acidity independently protonates the linker. Both thermal and chemical stimuli act synergistically, achieving ∼95% cumulative DOX release at pH 5.5 under NIR irradiation and ∼90% tumor suppression in vivo for combined PTT and chemotherapy. Reproduced with permission from ref . Copyright 2018, Springer Nature. b. Gold nanostar (NS)@ZIF-8 nanocomposites coated with an amphiphilic polymer (PMA) for NIR light-triggered intracellular cargo release. The multiple apexes of the nanostar core concentrate photothermal energy efficiently under NIR excitation (∼770 nm), driving thermodiffusion-mediated release of encapsulated bisbenzimide (HOE), while PMA suppresses nonspecific leakage in aqueous and physiological media. Reproduced with permission from ref . Copyright 2019, John Wiley and Sons. c. Au@ZIF-8 for tumor microenvironment-responsive photoacoustic (PA) imaging and photothermal therapy (PTT). In vivo PA imaging of a 4T1 tumor-bearing mouse following intratumoral injection of Au@ZIF-8: the tumor site (red ellipse) shows progressively increasing PA signal at 10, 30, and 60 min post-injection, while the subcutaneously injected normal tissue site (green ellipse) remains dark, confirming that Au nanosphere release and self-assembly into coupled plasmonic resonance (CPR)-active aggregates are selectively triggered by the synergistic action of overexpressed H⁺ and GSH in the tumor microenvironment. Reproduced with permission from ref . Copyright 2020, American Chemical Society.

Endogenous chemical stimuli native to the TME provide additional selectivity. In AuNP@ZIF-8 core–shell systems, tumor-relevant GSH concentrations trigger ZIF-8 decomposition via Zn²⁺ corrosion; released AuNPs self-assemble through interparticle interactions under low pH, generating NIR-absorbing plasmonic aggregates that activate photoacoustic imaging and PTT selectively within tumor tissue in a 4T1 tumor-bearing mouse model, mechanistically coupling diagnostic signal to therapeutic activation (Figure c). Extending these platforms into the 1,000–1,350 nm region of the NIR-II window could improve tissue penetration. Exploring biocompatible MOF frameworks beyond ZIF-8, including MIL-101­(Fe) and UiO-66, with distinct degradation kinetics and plasmonic metal-ion release profiles represents natural directions for the field.

3.3. Therapeutic Applications

Plasmonic NP-MOF hybrid nanostructures offer a versatile platform for therapeutic applications because the plasmonic and MOF components contribute distinct yet complementary functional roles that, when structurally integrated, enable therapeutic modalities inaccessible to either component alone. The plasmonic component provides photothermal conversion, ROS-catalytic activity, and photoacoustic contrast, all arising from the same LSPR absorption, while the MOF contributes stimuli-responsive cargo release, catalytic amplification of oxidative stress, and structural confinement that directs therapeutic activity to the target site. Critically, the choice of structural motif is not incidental but determines which of these roles are expressed and how they are coupled, a principle that underlies the organization of the therapeutic examples discussed in the sections that follow.

3.3.1. ROS-Mediated and Catalytic Therapy

Conventional ROS-based cancer therapies are constrained by insufficient ROS yield in the hypoxic, GSH-rich TME and by poor spatial confinement of oxidative damage. Plasmonic NP-MOF hybrids address both limitations through a division-of-labor architecture in which the structural motif determines how the plasmonic and MOF components partition their catalytic roles. In decorated architectures, externally accessible NP surfaces can participate directly in catalytic reactions for H₂O₂ decomposition and Fenton-active metal ion reduction, while the MOF exterior simultaneously anchors photosensitizers and supplies Fenton-active nodes.

In chemodynamic therapy based on Fenton-type cascades, decorated Fe-MOF/Au nanoparticles, in which AuNPs are anchored at the Fe-MOF surface, enable concurrent camptothecin (CPT) chemotherapy and Fe-mediated Fenton-mediated •OH generation, with AuNP surfaces accelerating Fe³⁺ → Fe²⁺ reduction to regenerate the active Fenton catalyst (Figure a). In vivo antitumor efficacy was further demonstrated in HepG2 tumor-bearing nude mice, with significant tumor growth suppression. Intracellular GSH scavenging is the primary barrier to Fenton efficacy. ZIF-67@AuNPs@MnO₂ nanoprobes overcome this through three amplification mechanisms: MnO₂ depletes GSH and releases Mn²⁺, bimetallic Co²⁺/Mn²⁺ accelerates Fenton kinetics beyond either ion alone, and AuNPs catalyze glucose oxidation to self-supply H₂O₂ when endogenous peroxide is limiting. Through these combined mechanisms, the hybrid nanoprobes decrease cell viability to 9.3% in vitro and achieved complete in vivo tumor elimination in tumor-bearing mice (Figure b). For photodynamic therapy (PDT), AuNP-decorated ZIF-8 nanostructures in which chlorin e6 (Ce6)-loaded ZIF-8 nanostructure decorated with approximately 4 nm AuNPs function as catalase-mimetic nanozymes, converting endogenous H₂O₂ to O₂ to relieve tumor hypoxia and sustain ¹O₂ yield under 660 nm irradiation; in vivo tumor suppression reached 94.1% versus 66.5% for ZIF-8/Ce₆ alone, with the improvement entirely attributable to AuNP-mediated hypoxia relief.

8.

Au/MOF hybrid nanostructures for multimodal cancer therapy integrating chemotherapy and chemodynamic therapy (CDT). a. PEG-Au/FeMOF@CPT nanoparticles are fabricated by in situ growth of Au nanoparticles on TCPP­(Fe)/Zr-based FeMOF, followed by CPT loading and PEG-SH/C₁₂SH surface modification. Upon cellular internalization, elevated intracellular phosphate triggers MOF collapse and CPT release, while Au nanoparticles oxidize glucose to generate H₂O₂ that is converted to •OH via Fenton reaction catalyzed by TCPP­(Fe). CLSM images (DCF, green; DAPI, blue) confirm enhanced intracellular ROS generation in PEG-Au/FeMOF-treated cells, with the iron-free PEG-Au/HMOF control demonstrating the essential catalytic role of TCPP­(Fe). Reproduced from ref . Available under a CC-BY 4.0 license. Copyright 2020, Yuan Ding et al. b. ZIF-67@AuNPs@MnO₂ nanoprobes achieve triple-amplified CDT through four sequential TME-responsive reactions: MnO₂ depletes GSH to relieve •OH scavenging and generate Mn²⁺, AuNPs catalyze glucose oxidation to self-supply H₂O₂, and bimetallic Co²⁺/Mn²⁺ cooperatively drive Fenton-like •OH generation. MB degradation assays confirm stepwise enhancement of •OH yield across ZIF-67, ZIF-67@AuNPs, and ZIF-67@AuNPs@MnO₂, with cell viability reduced to 9.3% and complete in vivo tumor elimination. H&E staining confirms biocompatibility across major organs. Reproduced with permission from ref . Copyright 2023, Royal Society of Chemistry.

3.3.2. Photothermal Therapy and Theranostics

Embedded Au/ZIF-8 nanocomposites couple acid-triggered ZIF-8 collapse with NIR-activated hyperthermia in a self-reinforcing chemo-PTT response: the same acidic stimulus that releases DOX simultaneously exposes the AuNP photothermal surface, and elevated temperature potentiates intracellular DOX uptake and DNA intercalation synergistically. This chemo-photothermal response was demonstrated at the cellular level. AuNR@porphyrin-MOF core–shell hybrids integrate PTT and PDT within a single nanostructure. AuNRs provide NIR photothermal output, while porphyrin linkers in the MOF shell generate¹ O₂ under the same irradiation and Fe-cluster nodes catalytically regenerate O₂. The PTT-to-PDT ratio can be tuned independently by adjusting MOF shell thickness, a structural handle unique to the plasmonic NP@MOF core–shell motif, since shell growth is controlled by the number of layer-by-layer deposition cycles. Therapeutic efficacy and biocompatibility were further evaluated in both cell and animal experiments. Yolk–shell engineering extends Au–MOF photothermal platforms into the NIR-II window. Au nanostar@ZIF-8 yolk–shell nanoparticles combine a NIR-II-responsive Au nanostar with a hollow cavity and porous ZIF-8 shell for DOX delivery. The broad NIR absorption of the Au nanostar enabled photothermal activation at 1064 nm, where the hybrid exhibited a photothermal conversion efficiency of 48.5%, compared with 30.2% at 808 nm. NIR-II-induced heating further promoted DOX release, enabling synergistic chemo-photothermal therapy. Therapeutic efficacy was validated both in vitro and in H22 tumor-bearing mice. Ag@cyclodextrin-based MOF@PDA (Ag@CD-MOF@PDA) nanoplatforms extend the design to antibacterial applications: PDA-mediated NIR hyperthermia provides immediate thermal membrane disruption, while the CD-MOF scaffold functions as a controlled Ag⁺ -release reservoir that sustains bactericidal activity after the light source is removed, a temporal complementarity inaccessible to monofunctional systems. Efficacy was demonstrated both in vitro and in bacterial- and biofilm-infected wound models in vivo.

For theranostics, the NIR LSPR absorption of AuNRs generates photoacoustic (PA) signals under pulsed laser excitation, making AuNR@MOF core–shell nanostructures inherently dual PA-imaging/PTT agents without additional modification. Fe-based MOF frameworks add a T₂-weighted magnetic resonance imaging (MRI) channel through the paramagnetic relaxivity of framework Fe³⁺ centers; Au/MIL-88 composites demonstrated continuously increasing intratumoral MRI signal intensity from 1 to 24 h post-injection in vivo, with concurrent AuNP-driven PTT and heat-induced drug release. Mechanistic coupling of imaging readout to therapeutic activation transforms colocalization into a self-reporting platform: in AuNR@ZIF-8 systems, GSH-triggered ZIF-8 degradation simultaneously releases cargo and assembles NIR-absorbing AuNP aggregates whose PA signal intensity tracks the extent of framework degradation and thus the therapeutic progress. Building on these demonstrations, further development of NIR-II Au–MOF platforms may benefit from anisotropic Au nanostructures with red-shifted plasmon resonances, such as high-aspect-ratio Au nanorods and Au nanostars, to exploit the deeper light penetration associated with reduced tissue attenuation in this window. Their long-term photothermal stability under repeated irradiation, however, should be systematically evaluated alongside irradiation wavelength and power density to ensure therapeutic efficacy within clinically acceptable laser exposure limits.

4. Perspectives

Among the four structural paradigms reviewed, the evolution from embedded to core–shell architectures reflects a broader and more diverse synthetic maturation: whereas embedded structures suffer from reduction-induced MOF degradation, poor size uniformity of plasmonic NPs, and limited spatial control over plasmonic NP number and distribution, plasmonic NP@MOF core–shell structures invert the synthetic sequence to decouple plasmonic NP synthesis from MOF growth, affording independent control over core multiplicity, shell thickness, and spatial distribution while preserving intact pore volumes. These structural advantages have driven a trend toward plasmonic NP@MOF architectures across all three biomedical domains reviewed here.

Despite this structural progress, the biomedical evidence discussed here spans markedly different levels of validation that should not be interpreted equivalently. In vitro and cellular studies establish mechanistic feasibility, but require subsequent validation of biodistribution, stability, therapeutic efficacy, and off-target effects in vivo. Studies using human serum, urine, breath, or sweat demonstrate analytical compatibility with clinically relevant matrices, but do not constitute clinical validation; larger patient cohorts, appropriate disease controls, and assessment of interindividual variability and assay reproducibility are required to establish diagnostic utility. Animal studies provide stronger evidence of therapeutic feasibility, yet most rely on short-term mouse tumor models and therefore require systematic pharmacokinetic, biodistribution, clearance, and long-term toxicological evaluation, followed by validation in more physiologically representative and larger-animal models. Importantly, clinical validation of plasmonic NP-MOF hybrids has not yet been established, and eventual translation will additionally depend on reproducible and scalable manufacturing, storage stability, and regulatory assessment. Looking forward, synthetic frontiers remain critical for advancing core–shell nanostructures toward broader biomedical utility. In plasmonic NP@MOF core–shell nanostructures, precise MOF shell thickness control at the subnanometer scale remains elusive. While CTAB concentration modulation offers a practical handle on shell growth, reproducible tuning below ∼10 nm is nearly unattainable by surfactant concentration alone, and layer-by-layer deposition is too complex and time-consuming for general use. Thinner shells are essential not only for maximizing analyte access to the plasmonic surface and interparticle hotspot formation in biosensing, but also for enabling direct interaction between the metal surface and external stimuli or the TME in therapeutic applications. A parallel challenge arises at the metal-MOF interface in both core–shell configurations. Current capping-ligand approaches serve merely to partially relieve mismatch-induced stress rather than to fully compensate for lattice parameter mismatch or to actively direct the facet-selective epitaxial shell growth onto a specific core surface, which would be required to establish crystallographically registered core–shell interfaces. To broaden the structural design space, future efforts should therefore pursue versatile and scalable methods for reproducible nanometer-to-subnanometer MOF shell thickness control, strategies that sufficiently reduce interfacial surface energy to stabilize arbitrary plasmonic NP-MOF combinations, and ultimately approaches that achieve facet-selective epitaxial shell growth on defined core nanoparticle facets. These advances would collectively expand the structural paradigm of plasmonic NP-MOF hybrids far beyond what is currently accessible.

Biomedical plasmonic NP@MOF core–shell hybrids encompass diverse material combinations because the optical, catalytic, transport, and degradation properties required for a given application can be distributed between the plasmonic and MOF components. Au and Ag have been employed as plasmonic core components, while HKUST-, NU-, porphyrin-, CuZn-, CD-, and ZIF-based frameworks have provided distinct pore architectures, node chemistries, and stability profiles. Within the plasmonic NP@MOF core–shell motif, which has emerged as a dominant configuration, however, the selection of both components has often been influenced by the availability and reproducibility of well-established synthetic protocols rather than being systematically guided by the functional requirements of the target application. The prevalence of ZIF-8 in the drug delivery and biosensing literature does not necessarily indicate that it outperforms alternative MOF frameworks under equivalent conditions. Instead, it appears to reflect, at least in part, the availability of well-characterized protocols, such as capping agent-mediated ZIF-8 encapsulation, that are broadly applicable to diverse core NP morphologies. Another important advantage is that ZIF-8 can be grown under mild, room-temperature conditions. Such conditions are particularly important for constructing plasmonic NP-MOF hybrids because the elevated temperatures or chemically harsh environments required for the synthesis of some MOFs may thermally anneal or etch the plasmonic cores, disrupt their surface ligands, and consequently alter their structural and plasmonic properties. AuNPs are similarly prevalent because reproducible, morphology-controlled synthesis enables tunable plasmonic response from the visible to the NIR region, while their comparatively high chemical stability supports consistent performance under physiological conditions. Thus, the predominance of Au and ZIF-8 primarily reflects their methodological maturity and synthetic accessibility rather than their demonstrated superiority over alternative materials. Advancing toward application-driven material selection will require robust and reproducible synthetic strategies for a broader range of plasmonic materials and MOF frameworks. These strategies should enable precise control over morphology and physicochemical properties under mild, broadly compatible conditions. Expanding this synthetic toolbox would allow each component to be selected according to the functional requirements of the intended biomedical application, thereby enabling the more rational design of plasmonic NP-MOF hybrids.

A parallel constraint arises from the conditions under which MOF components are synthesized. Many of the MOFs employed in the hybrid systems reviewed here are assembled under organic solvent conditions that bear little resemblance to the physiological environment in which they are ultimately expected to function: UiO-66 requires N,N-dimethylformamide (DMF) as the primary synthesis solvent, and ZIF-8, while synthesizable in aqueous media, achieves optimal crystallinity in methanol or DMF. The structural integrity and porosity established during synthesis in organic media are not necessarily preserved upon transfer to aqueous, ion-rich, protein-containing physiological environments. Competitive coordination by phosphate and chloride ions can disrupt metal–linker bonds, serum proteins can occlude pore apertures through nonspecific adsorption, and the pH fluctuations inherent to endosomal trafficking can accelerate framework degradation at unintended sites. These stability gaps between synthetic and physiological conditions represent a largely unaddressed design challenge. Developing MOF synthesis protocols that either operate directly in aqueous, physiologically relevant media or that yield frameworks whose structural and functional properties are explicitly validated under physiological conditions, rather than inferred from dry-state characterization, is therefore a critical direction for translating the performance demonstrated in idealized in vitro settings toward reliable in vivo function.

‡.

H.S., E.P., and K.L. contributed equally.

The authors declare no competing financial interest.