**1. Introduction**

Healthcare-associated infections (HCAIs) impact on millions of patients annually, and hundreds of thousands of deaths worldwide [1–4]. Current models predict that global mortality rates will approach 10 million annual deaths as a direct result of HCAIs by 2050 [5]. Approximately one-third of all such HCAIs are preventable [6], and while recent improvements in hygiene standards across the medical sector have for example significantly reduced the number of HCAIs per year in the UK, certain groups remain at very high risk of infection during hospital visits, and infection rates for some bacteria such as *Escherichia coli* have even increased [7–9].

The rise of drug-resistant bacteria which are disproportionately responsible for HCAIs, such as methicillin-resistant *Staphylococcus aureus* (MRSA) and extended-spectrum beta-lactamases (ESBL)-producing *Enterobacteriaceae* [2], has prompted a resurgence in the search for broad-spectrum antibiotics to tackle HCAIs. Silver is one such broad spectrum antibiotic, effective against Gram-positive (e.g., MRSA, *B. subtilis*) and Gram-negative bacteria (e.g., *E. coli*, *P. aeruginosa*) [10], and is consequently the subject of commercial [11] and academic interest for the treatment of existing HCAIs and drug-resistant organisms through its incorporation in surface coatings [12], nanotechnology [13–16], and pharmaceuticals [17–19].

A major advantage of nanoparticulate silver (and silver-based solid compounds) compared with salts or complexes is that oxidative dissolution is confined to exposed surfaces, offering a potential route to regulate the release of biologically active (commonly held to be Ag+) [20] species permitting the design of long lasting antimicrobial therapies [15,21–23]. Control over silver dissolution kinetics in Ag-nanoparticle-(NP) based systems has been investigated through tuning particle size [23–26], the use of inert coatings to retard release [26], or control of silver speciation in nanocomposite to accelerate dissolution and hence antibacterial activity [13,14]. Silver has also shown recent promise as a promoter of conventional antibiotic therapies [27].

Here we explore the use of high surface area and porous metal oxide frameworks to control the dispersion and subsequent dissolution of surface silver. The impact of framework porosity and oxide termination on silver speciation, dissolution, and antibacterical efficacy, was studied using ordered mesoporous and hierarchical macroporous-mesoporous SBA-15 silicas [28,29], and titania functionalized analogues. Conformal titania surface coatings, and hierarchically porous architectures promote silver NP dissolution and activity against Gram-positive (*Staphylococcus aureus*) and Gram-negative (*Pseudomonas aeruginosa*) bacteria.

#### **2. Results and Discussion**

#### *2.1. Materials Synthesis and Characterisation*

Three families of ordered porous frameworks were prepared using soft or hard-soft dual templating approaches and subsequent titania functionalization (Figure 1). Mesoporous and macroporous-mesoporous (MM) SBA-15 silicas were prepared using a Pluronic 123 (P123) surfactant either alone or in conjunction with 400 nm colloidal polystyrene nanospheres respectively, according to literature methods. Titania modified variants were subsequently prepared adapting the procedure of Landau et al. [30] to preactivate surface silanols before the self-limited grafting of titanium isopropoxide under anhydrous conditions, and calcination. Repeated grafting cycles were used to create conformal TiO2 monolayers encapsulating the silica supports. Silver nanoparticles were introduced to either the parent SBA-15, or TiO2-grafted SBA-15 and TiO2-grafted MM-SBA-15, frameworks by wet impregnation with varying silver nitrate concentrations.

**Figure 1.** Synthesis of Ag-doped mesoporous silica and titania-functionalized mesoporous and mesoporous-macroporous silica materials.

Textural properties of the parent SBA-15 and MM-SBA-15 supports, before and after titania functionalization, were first characterized by transmission electronic microscopy (TEM), N2 porosimetry, X-ray diffraction (XRD), and X-ray photoelectron spectroscopy (XPS). The expected hexagonally close-packed channels of the SBA-15 mesopore network were observed in all cases, with surface areas and mesopore diameters decreasing with successive TiO2 grafting cycle (Figure 2 and Figure S1), reflecting pore narrowing and partial blockage of micropores in the silica framework [31]. Uniform mesopores of 6.3 nm diameter were in accordance with previous reports [29]. No titania crystallites were visible by TEM following grafting, indicating the formation of highly dispersed phase. Assuming a TiO2 monolayer thickness of 0.355 nm [32], mesopore shrinkage across three grafting cycles from 6.3 to 5.3 nm for SBA-15 was consistent with deposition of a 1.4 monolayer (ML) titania coating, and from 6 to 5.2 nm for MM-SBA-15 consistent with 1.1 ML titania.

**Figure 2.** Bright-field transmission electron microscopy (TEM) images of (**a**) TiO2/SBA-15 (3r<sup>d</sup> cycle grafting), and (**b**) TiO2/MM-SBA-15 (3r<sup>d</sup> cycle grafting) highlighting ordered hexagonally close-packed mesopores and uniform macropores, and the evolution of textural properties and calculated TiO2 monolayer thickness as a function of grafting cycle for (**c**) TiO2/SBA-15 and (**d**) TiO2/MM-SBA-15.

Low-angle X-ray diffraction (XRD) (Figure 3) confirmed that titania-functionalized frameworks retained the *P6mm* symmetry of the parent SBA-15 mesopore network. Wide angle XRD (Figure S2) showed no evidence for crystalline titania phases, consistent with TEM and the formation of conformal monolayers over the SBA-15 template. The surface chemical environment of Ti atoms within the monolayer coatings was examined by XPS, which reveals the presence of Ti-O-Si (associated with the silica-titania interface) and Ti-O-Ti species over both SBA-15 and MM-SBA-15 with Ti 2p5/2 binding energies of 459.9 eV and 458.5 eV respectively. The shift to higher binding energy for Ti coordinated to Si (through a bridging oxygen) reflects the higher electronegativity of the latter, and hence higher induced initial charge on the former. We attribute the small shift between the Ti-O-Ti environments in P25 and the titania monolayers to quantum size (initial and/or final state) effects.

**Figure 3.** (**a**) Low angle X-ray diffraction (XRD), and (**b**) fitted Ti 2p XP spectra of TiO2/SBA-15 and TiO2/MM-SBA-15, alongside pure silica (SBA-15) and titania (P25) references.

Silver was subsequently deposited over the unfunctionalized SBA-15, and TiO2/SBA-15 and TiO2/MM-SBA-15 at three different nominal loadings spanning 0.3–3 wt% (see the Supplementary Materials (Table S1); elemental analysis revealed almost identical bulk loadings for the mesoporous materials, but systematically lower loadings for the hierarchical material. Surface and bulk silver loadings were generally in good agreemen<sup>t</sup> for all materials (see the Supplementary Materials Table S1), indicating a homogenous distribution throughout the porous frameworks. Physicochemical properties of the resulting silver species were determined by TEM, XPS, XRD, and X-ray absorption near edge spectroscopy (XANES). Only fcc metallic silver (JCPDS no. 04-0783) reflections were observed by XRD, which sharpened with increasing loading (see the Supplementary Materials Figure S3), indicative of crystallite growth. Volume averaged silver NP sizes were smaller for MM-SBA-15 (3.5–6.5 nm) than SBA-15 (4.5–8.5 nm) at comparable loadings (~0.3 wt%), evidencing higher metal dispersion over the hierarchically porous framework, and indeed fell below the threshold of detection for the lowest 0.25 wt% Ag/TiO2/MM-SBA-15. Ag NPs were directly visualized by TEM (see the Supplementary Materials Figures S4–S6), which revealed a relatively broad size distribution spanning 1–20 nm over all three supports. The mean particle size increased (and size distribution narrowed) with Ag loading (see the Supplementary Materials Figure S7), with the smallest NPs observed over the hierarchical support in all cases, in accordance with XRD. Ag 3d XP spectra showed the existence of two distinct chemical environments with 3d5/2 spin-orbit split component binding energies of 368.0 eV and 370.5 eV consistent with Ag<sup>0</sup> and possibly a surface Ag2CO3 respectively (see the Supplementary Materials Figure S8) [33].

Note that discrimination of silver oxides and carbonate by XPS alone is complicated due to their similar core-level binding energies, and anomalous negative binding energy shift of the oxides relative to the metal [14,34,35] and therefore requires careful fitting to well-defined standards (difficult to guarantee due to the tendency of Ag2O to oxidise, and of both oxides to adsorb atmospheric CO2 and form surface carbonates [36]), or Auger parameter analysis employing the Ag M4,5NN Auger transition in conjunction with 3d5/2 core-level spectra [13]. Silver speciation as metal, oxide, or carbonate was therefore examined by linear combination fitting of Ag K-edge X-ray absorption near-edge spectroscopy (XANES, see the Supplementary Materials Figures S9–S11). In all cases, good spectral fits could only be obtained using Ag and Ag2CO3 components. Nanoparticle growth is accompanied by a continuous decrease in the proportion of surface and bulk silver carbonate for all frameworks (Figure 4). This switchover from electron deficient to metallic silver with particle size likely reflects the higher surface energy of the latter [37], which is thus favoured for larger particles. The bulk Ag2CO3 content determined by XANES was systematically higher than that of the surface determined by XPS, likely reflecting the distribution of particle sizes present in all materials; XPS is expected to more sensitive to larger (more metallic) particles preferentially located on the external surface. AgO and Ag2O could not be detected by linear combination fitting of the XANES data.

**Figure 4.** Surface (XPS) and bulk (XANES) silver speciation as Ag2CO3 of Ag/SBA-15, Ag/TiO2/SBA-15, and Ag/TiO2/MM-SBA-15 as a function of Ag loading.

#### *2.2. Materials Performance Assaying*

The release rate of ionic silver (Ag+) from the preceding SBA-15 and MM-SBA-15 frameworks in a 0.5 M NaNO3 solution at 37 ◦C was subsequently determined by ICP-MS analysis (Figure 5). Dissolution rates normalised to the mass of silver were inversely proportional to particle size (loading) for all frameworks, as reported for citrate stabilized silver metal nanoparticles [23] and core-shell silver-silica nanoparticles [26] indicating that the release of ionic silver occurred by a common mechanism, being solely dependent on the geometric surface area of the silver nanoparticles. Coating of the mesoporous SBA-15 by a conformal titania monolayer slightly decreased the average particle size over the bare silica framework, and hence increased the rate of silver dissolution, a phenomenon further enhanced by the introduction of macropores. The rate constant for Ag+ dissolution was determined as 5244 <sup>μ</sup>mol·h−1·nm<sup>−</sup>1·<sup>g</sup>1Ag by fitting the data (see the Supplementary Materials Figure S12) in accordance with the method of Zhang et al. [23] This is significantly higher than that found for Ag@SiO2 core-shell nanoparticles (14.7 <sup>μ</sup>mol·h−1·nm<sup>−</sup>1·g<sup>−</sup>1Ag) [26], while the absolute release rate

of the 0.3 wt% Ag/TiO2/MM-SBA-15 is consistent with that previously observed for 0.05 wt% high surface area Ag-hydroxyapatite nanoparticles [14].

**Figure 5.** Ag+ dissolution rates normalized to mass of Ag determined by XANES for Ag/SBA-15, Ag/TiO2/SBA-15 and Ag/TiO2/MM-SBA-15.

Antibacterial activity was subsequently assessed against a range of bacteria (*Pseudomonas aeruginosa*, *Staphylococcus aureus*, *Clostridium difficile*, *Escherichia coli*, *Bacillus subtilis* and MRSA) by zone plate inhibition (ZoI) (Figure 6) using simulated body fluid (SBF), in which the antimicrobial efficacy is proportional to the size of the colony-free zone surrounding the silver functionalized porous frameworks. None of the parent (silver-free) materials exhibited bactericidal properties. In all cases the ZoI values normalized to the mass of silver were inversely proportional to silver loading, mirroring the ionic silver release rates, and superior to those reported for hydroxyapatite (HA) supported Ag3PO4 composites or Ag@SiO2 core-shell nanocomposites. For example, for *S. aureus* Ag/TiO2/MM-SBA-15 achieved 70 mm·gAg<sup>−</sup><sup>1</sup> versus 40–60 mm·gAg<sup>−</sup><sup>1</sup> for a high area Ag-HA [14] or 23 mm·gAg<sup>−</sup><sup>1</sup> for 4.5 nm Ag nanoparticles encapsulated by a 10.5 nm mesoporous silica shell) [26]. The zone size (antibacterial performance) decreased in the sequence Ag/TiO2/MM-SBA-15 > Ag/TiO2/SBA-15 > Ag/SBA-15.

The logarithmic reduction method was also employed to obtain a more quantitative evaluation of antibacterial activity for representative Gram-positive and Gram-negative bacteria (*Staphylococcus aureus* and *Pseudomonas aeruginosa* respectively). Experiments were performed in the absence of light to eliminate any possible influence from photogenerated reactive oxygen species by the semiconducting titania coating. All three frameworks were inactive in the absence of silver (see the Supplementary Materials Figure S13). Corresponding decimal reduction times (D-values), i.e., the time to kill 90% of the bacteria, normalised to the mass of silver, also evidenced an inverse relationship with silver loading for all frameworks (Figure 7) consistent with their Ag+ release rates. (Figures S14–S15). As for the ZoI assays, antibacterial activity followed the order Ag/TiO2/MM-SBA-15 > Ag/TiO2/SBA-15 > Ag/SBA-15, i.e., titania functionalization, and the introduction of macroporosity, both enhanced the potency of silver nanoparticles dispersed throughout an SBA-15 framework. Ag/MM-TiO2/SBA-15 outperformed Ag/SBA-15 by ~105% against *Staphylococcus aureus* and ~103% against *Pseudomonas aeruginosa*. The present observations are consistent with previous reports of augmented antimicrobial efficacy arising from the preferential formation of highly soluble silver carbonate versus silver metal over γ-alumina supports [13].

**Figure 6.** Zone of inhibition plots, normalized to mass of Ag, against a range of Gram-positive and Gram-negative bacteria for (**a**) Ag/SBA-15, (**b**) Ag/TiO2/SBA-15 and (**c**) Ag/TiO2/MM-SBA-15.

**Figure 7.** Decimal reduction times for Ag/TiO2/MM-SBA-15, Ag/TiO2/SBA-15 and Ag/SBA-15 against (**a**) *Staphylococcus aureus*, and (**b**) *Pseudomonas aeruginosa.*
