Phase-Tuning of Ru Nanostructures in Ru/ZrO₂ Catalysts for Controlling Radical and Nonradical Pathways of Peroxymonosulfate-Based Oxidation and Defluorination of a Fluorinated Anticancer Drug

Abstract

This study demonstrates that the catalytic performance of Ru catalysts is strongly influenced by the crystal facets and structures of Ru nanoparticles, with hexagonal close-packed (hcp)-Ru exhibiting superior activity (turnover frequency = 18.1 s^–1) compared to face-centered cubic (fcc)-Ru (5.27–10.6 s^–1). Notably, the crystalline phase of ZrO₂ (i.e., tetragonal (t-ZrO₂), monoclinic (m-ZrO₂), and mixed phase (mixed-ZrO₂)) preferentially directs the formation of distinct Ru nanostructures, resulting in fcc-Ru/t-ZrO₂, hcp-Ru/m-ZrO₂, and fcc-Ru/mixed-ZrO₂. These Ru nanostructures govern the dominant reaction mechanisms during peroxymonosulfate (PMS)-based oxidation of capecitabine, a fluorinated anticancer drug, where fcc-Ru favors a nonradical pathway involving direct electron transfer, while hcp-Ru mainly facilitates a radical pathway via PMS activation. This mechanistic variation between fcc-Ru and hcp-Ru led to different defluorination routes and degradation pathways. Among these catalysts, hcp-Ru/m-ZrO₂ exhibited a higher F^– selectivity (97.6 ± 1.42%) compared to fcc-Ru catalysts (40.9 ± 0.7–67.2 ± 1.2%). Density functional theory calculations further confirmed that the binding energy of Ru nanostructures toward PMS follows the order hcp-Ru > fcc-Ru, indicating that PMS interacts more strongly with hcp-Ru, thereby favoring the SO₄ ^•–-dominant pathway. Conclusively, these findings provide mechanistic insights to guide the rational design of catalysts with tailored reaction mechanisms and reactivity by controlling metal nanostructures through the specific crystallinity of the support material.

1. Introduction

The transition from H₂O₂-based oxidation (e.g., Fenton, Fenton-like, UV/H₂O₂, etc.) to peroxymonosulfate (PMS) and peroxydisulfate-based homogeneous (e.g., thermal, alkali, UV, ultrasound, etc.) and heterogeneous systems (transition metals, noble metals, and carbonaceous material-based catalysts) has marked a significant advancement in the oxidative degradation of emerging micropollutants in water. Despite the extensive application of Pt-group metals in heterogeneous catalysis, their potential to activate PMS for the degradation of micropollutants has only recently been recognized and remains less explored compared to non-noble metals (Co, Cu, Ni, Mo, Mn, Fe, etc.) and carbonaceous material-based catalysts. − Notably, Pt-group metals have been demonstrated to activate PMS through both radical (i.e., SO₄ ^•– and ^•OH) and nonradical pathways (i.e., direct electron transfer). Specifically, Pd and Rh–Pd were reported to facilitate radical pathway, ,− while Pt, Ru, Rh, Ir, and Au are known to induce nonradical pathway. , But their catalytic reactivities are strongly influenced by the type of support materials (ZrO₂, TiO₂, Al₂O₃, etc.) and even their crystalline phases. ,

Zirconia (ZrO₂) can exist in multiple crystalline phases, including tetragonal (t-ZrO₂), monoclinic (m-ZrO₂) phases, and the mixture of both phases (mixed-ZrO₂). This variability in crystalline phase has prompted research into its role as a support material influencing the catalytic performance of supported metal catalysts in a variety of reactions, such as the hydrogenation of natural fats and oils, , hydrogenation of CO₂, , the water–gas shift reaction, hydrodeoxygenation of m-cresol, hydrogenolysis of glycerol, benzene hydrogenation, and methanol oxidation. Indeed, these studies collectively demonstrate that the crystalline phase of ZrO₂ significantly influences the reactivity and selectivity of catalytic reactions, potentially by adjusting the metal–support interactions, the concentration of oxygen vacancies, and the distribution of acid–base sites in ZrO₂. However, we hypothesized that the observed differences in the catalytic performance may also be attributed to structural and electronic changes in the supported metal nanoparticles induced by the underlying crystalline phase of ZrO₂. This is based on the premise that the supported noble metals can adopt different nanostructures, such as hexagonal close-packed (hcp) and face-centered cubic (fcc) phases, which exhibit distinct catalytic properties.

Ruthenium (Ru), a member of the noble metals, is approximately ten times less expensive than Pt, and its unique catalytic properties have enabled its widespread application across various catalytic reactions as a substitute for other metal catalysts. While Ru typically adopts a hexagonal close-packed crystalline structure (hcp-Ru), recent studies have demonstrated that face-centered cubic Ru (fcc-Ru) can also be synthesized. , Particularly, most noble metals favor fcc, while Ru is stable when it takes a hcp structure. However, it was found that the decrease in metal size to nanoscale enables Ru to adopt another phase, i.e., fcc phase. Thus, Ru nanoparticles can exist in either hcp or fcc structures under ambient conditions, each exhibiting distinct surface geometries. This difference in surface geometry might influence the interaction between Ru and reactants, particularly at specific sites (e.g., O–O bond), thereby affecting the activation pathway and reaction rates. Indeed, fcc-Ru exhibits superior reactivity compared to hcp-Ru in reactions such as CO oxidation, , H₂O₂ activation, oxygen evolution reaction, hydrogen oxidation reaction, , hydrogen evolution reaction, and photocatalytic production of H₂. Conversely, hcp-Ru has demonstrated higher catalytic reactivity than fcc-Ru in the hydrogenation of benzene and the reduction of aromatic aldehydes and ketones, indicating the strong nanostructure-dependent reactivity of Ru nanoparticles. However, despite these findings, several critical questions remain regarding the behavior of Ru in PMS-based advanced oxidation processes: (i) do the reactivity and reaction mechanisms of Ru depend on their specific crystallinity or nanostructure? (ii) Does the crystalline structure of ZrO₂ influence the morphology and nanostructure of the supported Ru? (iii) If so, can the crystalline phase of the ZrO₂ support direct the formation of specific nanostructures or exposed facets in the resulting Ru, thereby modulating their catalytic reactivity and mechanistic pathways? To address these questions, a series of Ru catalysts were synthesized using ZrO₂ supports of distinct crystalline phases (i.e., Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru-mixed/ZrO₂). PMS was employed as the oxidant, while capecitabine (Xeloda), a fluorinated anticancer drug, was selected as a model compound to represent fluorinated pharmaceuticals, with emphasis placed on its degradation and defluorination rates, as well as the underlying mechanisms. While the overall degradation rate of capecitabine indicates the extent of parent compound removal, defluorination and F^– selectivity are critical for mitigating the stability and ecotoxicity of organofluorine compounds due to the high persistence of C–F bonds. Specifically, defluorination reflects the efficiency of catalytic systems in cleaving stable C–F bonds and suppressing the formation of persistent and potentially more toxic fluorinated byproducts.

In this study, it is hypothesized that fcc-Ru and hcp-Ru formed on specific phases of ZrO₂ may exhibit distinct catalytic reactivities and mechanistic pathways in PMS activation and the subsequent degradation of target micropollutants. To verify this hypothesis, the objectives of this study are as follows: (i) to identify ZrO₂ phase-dependent Ru nanostructures; (ii) to investigate the influence of Ru nanostructure on catalytic reactivity and performance; (iii) to elucidate Ru nanostructure-dependent reaction mechanisms (i.e., nonradical pathway vs. radical pathway); (iv) to explore nonradical and radical defluorination routes using density functional theory (DFT) calculations; and (v) to identify degradation pathways of capecitabine over hcp- and fcc-Ru catalysts through the characterization of transformation products (TPs).

2. Experimental Section

2.1. Catalyst Preparation

Tetragonal phase of ZrO₂ (t-ZrO₂) and mixed phase of ZrO₂ (mixed-ZrO₂) were synthesized by calcinating ZrOCl₂·8H₂O and Zr­(OH)₄, respectively. The monoclinic phase of ZrO₂ (m-ZrO₂) was obtained from the commercially available ZrO₂ powder (99%, Sigma-Aldrich Inc.). Subsequently, 1 wt % of Ru was doped onto t-ZrO₂, m-ZrO₂, and mixed-ZrO₂ via the modified impregnation method, wherein the Ru species were chemically reduced to metallic Ru using NaBH₄. Detailed information about the chemicals and synthesis procedures is provided in the Supporting Information (Texts S1 and S2).

2.2. Catalyst Characterization

The crystalline phases were further identified using powder X-ray diffraction (XRD, SmartLab, Rigaku) and PDF-2 from the ICDD database (International Centre for Diffraction Data). The XRD analysis was conducted over the range of 5–80° (2θ). The morphology of ZrO₂ with different phases was observed using a field emission scanning electron microscopes (FE-SEM, JSM-7800F Prime, JEOL Ltd., Japan).

Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/mixed-ZrO₂ were synthesized using a method previously reported. More details can be found in the Supporting Information (Text S2). The characterization of Ru catalysts was carried out using H₂-TPR, CO-pulse chemisorption, XPS, and FE-TEM/EDX. Details in characterization techniques are available in the Supporting Information (Text S3).

2.3. Catalytic Testing

The reactivity tests of Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/mixed-ZrO₂ for the oxidative degradation of capecitabine were carried out in a glass reactor with continuous stirring (1000 rpm) by using a glass-coated magnetic bar. For each reaction, 99.0 mL of a mixture consisting of the Ru catalyst and capecitabine was transferred into a reactor. Subsequently, 1.0 mL of PMS was injected into the reactor to initiate the oxidation reaction under unbuffered conditions (pH 3.7 ± 0.2), unless otherwise specified. The solution pH remained stable throughout the reaction. 1.0 mL of sample was taken from the reactor with a syringe and then filtered using a PES membrane syringe filter (FPE-224-013, Jetbiofil), immediately followed by quenching the reaction using sodium thiosulfate solution. The filtered sample was diluted in a glass vial for further analysis. The reactor was maintained at 25 °C throughout the reaction using an external bath circulator (MaXircu, DAIHAN Scientific).

To elucidate reaction mechanisms of Ru catalysts and assess reactive species (SO₄ ^•–, ^•OH, and ¹O₂), different quenchers and probes (e.g., methanol, tert-butanol, l-histidine, furfuryl alcohol, and p-benzoquinone) were added to the reaction with varying the molar ratios of [quencher]₀/[capecitabine]₀. Additionally, the evaluation of reactive species was conducted by using Pd/t-ZrO₂, a benchmark catalyst for the radical pathway, to gain further insights into the reaction mechanism.

2.4. Analytical Methods

The concentrations of capecitabine (m/z 360.1, [M + H]⁺) during catalytic oxidation by Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/mixed-ZrO₂ were quantified by using liquid chromatography–mass spectrometry (LC–MS).

To identify all unknown TPs with high mass accuracy (<5 ppm and 5 decimal places), nano liquid chromatography–high-resolution quadrupole orbitrap mass spectrometry (nano LC–HRMS/MS) was employed in this study. The nano LC–HRMS/MS system consisted of a high-pressure liquid chromatography instrument (Ultimate3000, Thermo Scientific Inc., USA) coupled to a high-resolution tandem mass spectrometer (Q Exactive, Thermo Fisher Scientific). Data were processed using Compound Discoverer (version 3.3.1.111, Thermo Fisher Scientific Inc., USA) and ChemCalc (online tool, https://www.chemcalc.org/mf_finder). Based on the exact chemical formulas and annotation of MS² spectra, possible molecular structures were proposed for the identified TPs in order to investigate the degradation pathways of capecitabine. Details of analytical methods for LC–MS and nano LC–HRMS/MS are provided in Supporting Information (Text S4).

The concentrations of F^– released from capecitabine during the reaction were determined by an Eco IC (Metrohm) configured with a 919 IC autosampler plus 1 (Metrohm) and an anion column (Metrosep A Supp 5-150/4.0, Metrohm). A mobile phase was 3.2 mM Na₂CO₃ and 1 mM NaHCO₃ solution pumped at a flow rate of 0.7 mL min^–1. Based on the results obtained from IC, defluorination and F^– selectivity are estimated as follows

$$\mathrm{d}\mathrm{e}\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{o}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}(\%)=\frac{C_{\mathrm{F},\mathrm{t}}}{1\times C_0}\times 100$$ 1

$${\mathrm{F}}^{-}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}(\%)=\frac{C_{\mathrm{F},\mathrm{t}}}{1\times (C_0-C_{\mathrm{t}})}\times 100$$ 2

where C _F,t denotes the concentration of F^– (μM) at time t released from oxidative degradation of capecitabine; C ₀ is the initial concentration of capecitabine; and 1 indicates the number of F in the parent compound. C _t is the concentration of capecitabine at reaction time, t. The concentration of PMS was determined by a UV–vis spectrophotometer (Implen) employing a previously reported method.

2.5. DFT Calculation

All calculations were conducted using the ORCA program suite (ver. 5.0.4). Details of the calculation methods can be found in Supporting Information (Text S5).

3. Results and Discussion

3.1. The Influence of ZrO₂ Crystalline Phases on Ru Nanostructures

The crystal structure, morphology, and purity of the ZrO₂ supports were characterized by using XRD (Figure ), FE-SEM (Figure S1), HR-TEM (Figure S2), and elemental mapping (Figure S3). The t-ZrO₂ and m-ZrO₂ samples exhibited pure tetragonal and monoclinic phases, respectively, while mixed-ZrO₂ showed the presence of both tetragonal and monoclinic phases (Figure ). Morphological and structural analyses revealed differences in the crystallite size, particle aggregation, and exposed lattice planes among the three ZrO₂ supports. Elemental mapping confirmed the high purity of ZrO₂ with a Zr/O atomic ratio close to 1:2. Detailed results are provided in the Supporting Information (Text S6).

1.

XRD pattern of (a) m-ZrO₂, (b) t-ZrO₂, and (c) mixed-ZrO₂.

In order to explore the lattice spacing in the Ru particles on ZrO₂ with different crystalline phases, elemental mapping of Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/mixed-ZrO₂ was first performed to identify Ru particles, and then FE-TEM analysis was conducted to observe the interplanar spacings of Ru (Figure ). Elemental mapping of Ru/t-ZrO₂ demonstrates that Ru particles were distributed on the surface of aggregated tetragonal ZrO₂ particles (Figure a). The lattice spacing of the Ru particle on the surface of t-ZrO₂ was observed to be 0.22 nm, which is assigned to the (111) plane of fcc-Ru (Figure c). ,, The exposed (111) facets of fcc-Ru placed on the surface of t-ZrO₂ suggested that the growth of fcc-Ru was highly favorable on t-ZrO₂. In addition, the lattice spacing of 0.19 nm was also observed in the Ru/t-ZrO₂, which confirmed the presence of the (200) plane of fcc-Ru (Figure d). Interestingly, the situations are different for Ru particles that are supported by m-ZrO₂. Relatively bigger Ru particles (50–60 nm) were identified on the surface of m-ZrO₂ (Figure f). The lattice spacing observed in the Ru/m-ZrO₂ was 0.21 nm, which could be determined as the characteristic (002) plane of the hcp-Ru structure (Figure g,h). , This result suggests that the hcp-Ru preferentially grows on the surface of the monoclinic phase of ZrO₂. Therefore, it can be concluded that the crystalline phase of ZrO₂ influenced the structure of supported Ru: fcc-Ru/t-ZrO₂ and hcp-Ru/m-ZrO₂. Similar observation is also noted for mixed-ZrO₂ where crystallinity of tetragonal ZrO₂ is relatively higher than that of the monoclinic phase. The lattice spacings of 0.22 and 0.17 nm are ascribed to (111) and (102) facets of fcc-Ru, indicating that the formation of fcc-Ru is related to the presence of the tetragonal phase in mixed-ZrO₂ (Figure k,l). While a detailed mechanistic explanation warrants further study, we hypothesized that the preferential formation of fcc-Ru on t-ZrO₂ and hcp-Ru on m-ZrO₂ could be attributed to the epitaxial growth of different Ru nanoparticles on specific lattice planes of the tetragonal or monoclinic ZrO₂ phases. ,, The t-ZrO₂ surface might provide more favorable crystal planes for the epitaxial growth and alignment of fcc-Ru (e.g., fcc-Ru (200) (d = 0.19 nm)//t-ZrO₂ (−120) (d = 0.16 nm); fcc-Ru (111) (d = 0.22 nm)//t-ZrO₂ (−110) (d = 0.25 nm)), despite moderate lattice mismatches (12–16%). In contrast, the m-ZrO₂ surface, particularly the (11–1) plane (d = 0.31 nm) (Figure S2d), might better accommodate or stabilize the thermodynamically favored hcp-Ru phase by providing more compatible interface for its growth of the (002) plane. However, further study is essential to verifying this hypothesis.

2.

FE-TEM/EDX analysis of Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/mixed-ZrO₂. Elemental mapping images of (a) Ru/t-ZrO₂, (e) Ru/m-ZrO₂, and (i) Ru/mixed-ZrO₂. FE-TEM images of (b–d) Ru/t-ZrO₂, (f, –) Ru/m-ZrO₂, and (j–l) Ru/mixed-ZrO₂ with identified lattice planes of Ru.

H₂-TPR results revealed that pure t-ZrO₂ was not reducible within the tested temperature range (Figure ), whereas m-ZrO₂ and mixed-ZrO₂ showed H₂ consumption at 500–600 °C due to the reaction between H₂ and lattice oxygen in the monoclinic phase. , Ru/ZrO₂ catalysts exhibited distinct peaks at 153–157 °C, corresponding to the reduction of RuO₂ to Ru(0), with additional peaks suggesting the presence of RuCl₃, aggregated RuO₂, or Ru^δ+ strongly anchored on the ZrO₂ surface (i.e., Ru–O–Zr species). Notably, Ru/t-ZrO₂ and Ru/mixed-ZrO₂ also showed peaks at 683–726 °C and 773–744 °C, respectively, indicating Ru-facilitated reduction of the ZrO₂ surface layer. Further discussion on the H₂-TPR results is provided in the Supporting Information (Text S6).

3.

H₂-TPR profile of (a) t-ZrO₂, Ru/t-ZrO₂, (b) m-ZrO₂, Ru/m-ZrO₂, (c) mixed-ZrO₂, and Ru/mixed-ZrO₂. XPS spectra of Ru 3d in (d) Ru/t-ZrO₂, (e) Ru/m-ZrO₂, and (f) Ru/mixed-ZrO₂ before reaction.

The oxidation states of surface Ru, Zr, and O on different Ru/ZrO₂ catalysts were further analyzed using XPS by deconvoluting Ru 3d, Zr 3d, and O 1s spectra (Tables , S2 and Figures d–f, and S4–S6). Considering that Ru 3d_3/2 and the C 1s peak are partially overlapped, a new peak at 280.2 eV emerged with splitting of 4.17 eV, verifying the formation of Ru(0) on the surface (Figure d–f). In contrast, the Ru(0) peak was not observed in the control sample (i.e., Ru/ZrO₂ without NaBH₄ activation) (Figure S4, Table S2), indicating that the formation of Ru(0) is attributed to the chemical reduction step using NaBH₄ during catalyst synthesis. Additional peaks were assigned to RuO₂ (281.1 eV) and RuCl₃ (282.4–5 eV), both exhibiting a spin–orbit splitting of 4.17 eV. It is because Ru­(III) and Ru­(IV) were incompletely reduced by NaBH₄ or Ru(0) was passivated after its exposure to air. , Quantitative analysis revealed that Ru/t-ZrO₂ and Ru/m-ZrO₂ exhibited 42.9% and 44.2% Ru(0), respectively, while Ru/mixed-ZrO₂ showed a slightly lower Ru(0) content of 36.1% (Table ). This result indicates comparable formation of Ru(0) across the ZrO₂ supports, regardless of their crystalline phases. The oxidation states of Zr and O in Ru/ZrO₂ samples were further monitored (Figures S5 and S6). Details of deconvolution of Ru 3d, Zr 3d, and O 1s spectra are provided in the Supporting Information (Text S6).

1. Oxidation States of Ru in Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/Mixed-ZrO₂ .

catalyst	before reaction
	RuO2 (281.1 eV)	RuCl3 (282.5 eV)	Ru(0) (280.2 eV)
Ru/t-ZrO2	28.5%	28.6%	42.9%
Ru/m-ZrO2	35.0%	20.8%	44.2%
Ru/mixed-ZrO2	38.5%	25.4%	36.1%

3.2. Catalytic Oxidation of Capecitabine by PMS and Ru Catalysts

No degradation of capecitabine was observed with PMS alone (Figure ), indicating that PMS (HSO₅ ^–, E _H ⁰ = 1.82 V) is unable to oxidize capecitabine under ambient conditions. Similarly, ZrO₂ supports (i.e., t-ZrO₂, m-ZrO₂, and mixed-ZrO₂) with PMS showed negligible capecitabine removal, suggesting that (i) sorption of capecitabine is not favorable on the tetragonal and monoclinic ZrO₂ surfaces and (ii) both t-ZrO₂ and m-ZrO₂ are unable to activate PMS for oxidation. Additionally, Ru catalysts alone can neither adsorb nor remove capecitabine. However, nearly complete degradation of capecitabine was achieved within 30 min using Ru/t-ZrO₂ (Figure a) and Ru/mixed-ZrO₂ (Figure c), while Ru/m-ZrO₂ resulted in 61.96 ± 1.65% removal (Figure b). The observed reaction rate constants (k _obs) follow the trend: Ru/t-ZrO₂ (0.183 ± 0.001 min^–1) > Ru/mixed-ZrO₂ (0.174 ± 0.003 min^–1) > Ru/m-ZrO₂ (0.011 ± 0.001 min^–1). To further identify different reactivities of Ru nanostructures, turnover frequencies (TOFs) were estimated to assess the rate of capecitabine oxidation (mol s^–1) per mole of Ru, based on CO pulse chemisorption data (Table ). Interestingly, Ru/m-ZrO₂ exhibited the highest TOF (18.1 s^–1), followed by Ru/t-ZrO₂ (10.6 s^–1) and Ru/mixed-ZrO₂ (5.27 s^–1). According to surface characterization results (Section 3.1), this higher TOF of Ru/m-ZrO₂ (despite lower k _obs) could be attributed to the hcp-Ru sites, which were predominantly observed in Ru/m-ZrO₂. Notably, despite its lower metal dispersion (2.03%) and smaller metallic surface area (7.42 m² per g_Ru), the Ru atom in the hcp-Ru structure appears to be more catalytically reactive toward capecitabine oxidation than that in fcc-Ru identified in both Ru/t-ZrO₂ and Ru/mixed-ZrO₂. These variations in reactivity among the Ru/ZrO₂ catalysts may also result from distinct reaction mechanisms that are dictated by the Ru nanostructure, as discussed in the following sections.

4.

Catalytic oxidation of capecitabine using (a) Ru/t-ZrO₂PMS, (b) Ru/m-ZrO₂PMS, and (c) Ru/mixed-ZrO₂PMS, compared to PMS only and Ru/ZrO₂ only. Conditions: [capecitabine]₀ = 10 mg L^–1; [catalyst]₀ = 0.1 g L^–1; [PMS]₀ = 0.5 mM.

2. CO Pulse Chemisorption of Ru/t-ZrO₂, Ru/M-ZrO₂, and Ru/Mixed-ZrO₂ .

catalyst	Ru (wt %)	cumulative uptake quantity (cm3 g–1)	metal dispersion (%)	metallic surface area (m2 g metal –1)	active particle diameter (nm)	cubic crystallite size (nm)	active metal sites (mmol)	capecitabine oxidation rate (mmol s–1)	TOF (s–1)
Ru/t-ZrO2	1	0.328	14.8	54.1	8.95	7.46	1.47 × 10–7	15.5 × 10–7	10.5
Ru/m-ZrO2	1	0.045	2.03	7.42	65.2	54.3	0.201 × 10–7	3.63 × 10–7	18.1
Ru/mixed-ZrO2	1	0.731	29.5	108	4.50	3.75	2.92 × 10–7	15.4 × 10–7	5.27

^aMetal dispersion represents the ratio of the number of active metal atoms (N _s) available for reaction to the total number of metal atoms (N _T) in the catalyst (metal dispersion (%) = (N _s/N _T) × 100%).

^bTOF of capecitabine degradation (TOF) = capecitabine removal rate (mmol·s^–1)/active metal sites (mmol).

3.3. Ru Nanostructure-Dependent Reaction Mechanisms

3.3.1. Radical Pathway Mediated by hcp-Ru/m-ZrO₂

The effect of quenchers (i.e., methanol and tert-butanol) on capecitabine degradation kinetics varied depending on the Ru nanostructures supported by different ZrO₂ crystal phases, suggesting that fcc-Ru and hcp-Ru may facilitate distinct reaction mechanisms. The addition of excess methanol at [methanol]₀/[capecitabine]₀ ratios of 1000 and 2000 significantly retarded both the degradation and defluorination kinetics of capecitabin by Ru/m-ZrO₂ (Figure b and S8b). Given that methanol effectively quenches both SO₄ ^•– (1.1 × 10⁷ M^–1 s^–1) and ^•OH (1.1 × 10⁷ M^–1 s^–1) (Table S1), these results indicate that hcp-Ru activates PMS to generate reactive radicals (e.g., SO₄ ^•– and ^•OH). This behavior is consistent with observations for the benchmark Pd/t-ZrO₂ catalyst, where the addition of excess methanol significantly suppressed capecitabine oxidation (Figure S7), confirming that PMS is activated by Pd(0) to generate SO₄ ^•– and ^•OH via hemolytic cleavage of the O–O bond. , To further assess the role of ^•OH, tert-butanol was used as a quencher due to its high reactivity with ^•OH ((5.9 ± 0.5)×10⁸ M^–1 s^–1) and much slower reaction with SO₄ ^•– (<1.6 × 10² M^–1 s^–1). As shown in Figure e, tert-butanol had a negligible inhibitory effect on capecitabine degradation by Ru/m-ZrO₂, while a moderate inhibition (∼30%) was observed for the Pd catalyst at a [tert-butanol]₀/[capecitabine]₀ ratio of 1000 (Figure S7). These findings suggest that SO₄ ^•– is the predominant radical generated by hcp-Ru, playing a key role in the oxidative degradation and defluorination of capecitabine (Figure g).

5.

Effect of methanol on catalytic oxidation of capecitabine by (a) Ru/t-ZrO₂, (b) Ru/m-ZrO₂, and (c) Ru/mixed-ZrO₂. Conditions: [capecitabine]₀ = 10 mg L^–1; [catalyst]₀ = 0.1 g L^–1; [PMS]₀ = 0.5 mM; [methanol]₀/[capecitabine]₀ = 1000 and 2000. Effect of tert-butanol on catalytic oxidation of capecitabine by (d) Ru/t-ZrO₂, (e) Ru/m-ZrO₂, and (f) Ru/mixed-ZrO₂. Conditions: [capecitabine]₀ = 10 mg L^–1; [catalyst]₀ = 0.1 g L^–1; [PMS]₀ = 0.5 mM; [tert-butanol]₀/[capecitabine]₀ = 1000 and 2000. (h) Schematic diagram of the nonradical pathway via fcc-Ru and (g) radical pathway via hcp-Ru.

3.3.2. Nonradical Pathway Mediated by fcc-Ru/t-ZrO₂ and Ru/mixed-ZrO₂

In contrast, the addition of excess methanol had a negligible effect on the capecitabine degradation rate in the case of Ru/t-ZrO₂. Specifically, the inhibition was limited to 9.8% at a [methanol]₀/[capecitabine]₀ ratio of 1000 (27.8 mM) (Figure a). Thus, these results indicate that capecitabine degradation by Ru/t-ZrO₂ proceeded predominantly through a nonradical pathway, with SO₄ ^•– playing a very minor role, as evidenced by the minimal quenching effect of methanol. Likewise, ^•OH involvement can be ruled out, consistent with the negligible inhibition by tert-butanol at the same molar ratio (Figure d). At a higher [methanol]₀/[capecitabine]₀ ratio of 2000, the increased inhibitory effect is likely due to the methanol blocking fcc-Ru reactive sites that are involved in PMS adsorption. , Notably, the defluorination of capecitabine was also only marginally affected by methanol, suggesting that alternative nonradical mechanisms other than SO₄ ^•–-mediated defluorination could be responsible for the observed defluorination (Figure S8a). A similar pattern was noted with Ru/mixed-ZrO₂ (Figure c,f), further supporting the conclusion that fcc-Ru preferentially promotes a nonradical oxidative pathway for capecitabine. Moreover, the bicarbonate ion (HCO₃ ^–) caused no inhibition with fcc-Ru/t-ZrO₂ (Figure S9) but significantly suppressed oxidation with hcp-Ru/m-ZrO₂. It is because HCO₃ ^– scavenges SO₄ ^•– (E ⁰ = 2.6–3.1 V) to yield CO₃ ^•–/HCO₃ ^•– with lower reactivity (E ⁰ = 1.78 V), resulting in more pronounced inhibitory effect on the radical pathway. , Hence, this further supports a radical-dominant pathway on hcp-Ru versus a nonradical dominant pathway on fcc-Ru. While singlet oxygenation (¹O₂) is often proposed as a nonradical pathway, its formation can be ruled out based on detailed analysis provided in the Supporting Information (Text S7, Figures S10 and S11). Therefore, fcc-Ru within Ru/t-ZrO₂ and Ru/mixed-ZrO₂ facilitates the direct oxidation of capecitabine without the involvement of free radicals, ¹O₂ or O₂ ^•–. In this nonradical pathway, electrons are transferred from the organic pollutant (electron donor) to PMS (electron acceptor) via the catalyst as a mediator (Figure h). ,, The mechanistic rationalization of the fcc-Ru-driven direct electron transfer and hcp-Ru-mediated PMS activation was elucidated with the aid of DFT calculations.

3.3.3. DFT Calculation of fcc-Ru and hcp-Ru

For the nonradical mechanism proposed for fcc-Ru, we hypothesized that capecitabine should have (i) sufficient electron-donating capacity and (ii) exhibit high affinity toward reactive fcc-Ru surface. To validate the first criterion, ionization potential (IP) was considered, since it is a well-known descriptor for assessing electron transfer reactivity. It has been reported that organic compounds with IP values higher than 9.0 eV are generally unable to serve as electron donors and thus show low reactivity in the nonradical oxidation pathway. To estimate the IP value of capecitabine, the previously reported method was employed in this study. The calculated IP value of capecitabine was 8.26 eV, which is far below the 9.0 eV threshold, indicating that capecitabine has a great potential to serve as an electron donor and to facilitate a nonradical pathway, namely, direct electron transfer on fcc-Ru.

For assessing the second criterion that the high affinity of fcc-Ru toward capecitabine could lead to a nonradical pathway via electron transfer (i.e., electrons shuttle), the E _binding of fcc-Ru₅₅ clusters were computationally determined and compared with that of hcp-Ru₅₆ (Figure S12). The calculated E _binding values for capecitabine followed the trend: fcc-Ru₅₅ (−78.4 kcal mol^–1) > hcp-Ru₅₆ (−65.8 kcal mol^–1). The more negative E _binding of fcc-Ru₅₅ indicates a higher affinity and stronger interaction with capecitabine compared to those of hcp-Ru₅₆. The fluoropyrimidine moiety of capecitabine was found to preferentially align with the fcc-Ru surface, forming close distances (d(N₇₅–Ru₂₀) = 2.05 Å, d(C₈₅–Ru₁₄) = 2.12 Å, and d(C₉₁–Ru₃₅) = 2.16 Å) with surface Ru atoms (Figure S12a1), whereas such interaction was not evident with hcp-Ru (Figure S12b1). These results suggest that capecitabine adsorbs more favorably onto the fcc-Ru surface to facilitate electron transfer.

In contrast, the E _binding values for PMS showed the opposite trend: hcp-Ru₅₆ (−45.6 kcal mol^–1) > fcc-Ru₅₅ (−32.0 kcal mol^–1) (Figure S12), implying that PMS has a stronger binding affinity for the hcp-Ru surface. This preferential interaction may promote the adhesion of PMS to hcp-Ru, facilitating the O–O bond scission and the subsequent generation of SO₄ ^•–. Indeed, when no structural constraints were applied to the O–O bond, spontaneous scission (d(O₅₉–O₆₃) = 3.72 Å) occurred during the optimization process of the hcp-Ru₅₆ + PMS complex, yielding SO₄ ^•– and ^•OH (Figure S12b3). In contrast, PMS is oriented vertically on the fcc-Ru surface without undergoing any structural deformation (Figure S12a2). These results demonstrate that the hcp-Ru surface structure not only enhances PMS adsorption but also facilitates selective activation of PMS, thereby promoting efficient radical production (Eq. )­

$${\mathrm{R}\mathrm{u}}^0(hcp)+{\mathrm{H}\mathrm{S}\mathrm{O}}_5^{-}\to {\mathrm{R}\mathrm{u}}^0(hcp)+{\text{SO}}_4^{·-}{+}^{·}\mathrm{O}\mathrm{H}$$ 3

3.3.4. Differences in Catalytic Behaviors between fcc-Ru/t-ZrO₂ and hcp-Ru/m-ZrO₂

Additionally, the nonradical pathway exhibited different behaviors in PMS consumption compared to the radical pathway. If electron transfer from capecitabine to PMS occurs, an increase in the concentration of the electron donor (capecitabine) would be expected to accelerate the decomposition of the electron acceptor (PMS). As illustrated in Figure S13a,b, during capecitabine oxidation over the fcc-Ru/t-ZrO₂, the PMS consumption rate increased significantly as the initial capecitabine concentration rose from 10 to 40 mg L^–1. Specifically, the k _obs for PMS decomposition increased by approximately 4.4-fold with the addition of 40 mg L^–1 of capecitabine (Figure S13b), indicating that the direct electron transfer process catalyzed by fcc-Ru is highly sensitive to the concentration of the substrate. In stark contrast, when hcp-Ru/m-ZrO₂ mediated the radical pathway, the PMS reduction kinetics exhibited only a marginal change as the capecitabine concentrations increased from 10 to 40 mg L^–1 (Figure S13e). This suggests that in the radical mechanism, PMS activation by hcp-Ru to generate reactive species was largely independent of the target pollutant. These findings align with established understandings that radical pathways exhibit less sensitivity to target pollutants, whereas nonradical pathways involving direct electron transfer are strongly dependent on the electron-donating substrate.

The difference between nonradical pathways and radical pathways was further highlighted with the addition of NOM to the reaction mixture. The nonradical pathway, facilitated by fcc-Ru in Ru/t-ZrO₂ and Ru/mixed-ZrO₂, demonstrated greater tolerance to NOM compared to the radical pathway mediated by hcp-Ru in Ru/m-ZrO₂. Notably, the fcc-Ru catalysts still achieved >99% of capecitabine removal and maintained defluorination efficiency in the presence of 1–5 ppm of NOM, whereas the hcp-Ru catalyst was deactivated at 5 ppm of NOM (Figure S14). It is because SO₄ ^•– produced by hcp-Ru was readily scavenged by electron-rich moieties within NOM. This observation is further corroborated by the quenching experiment conducted in the presence of NOM, wherein the sequential addition of NOM (1 ppm) and methanol (27.8 mM) resulted in complete inhibition of capecitabine oxidation by hcp-Ru/m-ZrO₂ (Figure S15). This suggests that the residual SO₄ ^•– species potently generated by hcp-Ru, even under NOM-rich conditions, were effectively scavenged by methanol (Figure S15b). The detailed analysis is provided in the Supporting Information (Text S8, Figures S14 and S15).

3.3.5. The Reactivity of fcc-Ru and hcp-Ru after Reaction

To investigate the recyclability and durability of Ru catalysts with respect to (i) reactivity, (ii) reaction mechanism, and (iii) crystallinity, additional recycling tests and quenching experiments were conducted (Figure S16), followed by corresponding surface characterizations. The used fcc-Ru/t-ZrO₂, Ru/mixed-ZrO₂, and hcp-Ru/m-ZrO₂ catalysts were analyzed by XPS (Figures S4) and FE-TEM/EDX (Figure S17). As shown in Figure S16, all used Ru catalysts maintained most of their catalytic reactivity, with a slight decrease in both the capecitabine removal rate and the defluorination rate. Specifically, the k _obs values for Ru/t-ZrO₂ and Ru/mixed-ZrO₂ decreased by 1.36- and 1.37-fold, respectively, relative to the pristine catalysts, while a 1.60-fold decrease was observed for hcp-Ru/m-ZrO₂. Consistently, defluorination decreased from 67.2 ± 1.2% to 55.3 ± 4.1% for Ru/t-ZrO₂ and from 60.8 ± 0.3% to 45.2 ± 0.9% for Ru/m-ZrO₂, with negligible changes for Ru/mixed-ZrO₂. This modest loss of activity can be attributed to changes in the Ru oxidation states, as indicated by XPS (Figure S4g–i). After reaction, Ru(0) remained but its fraction decreased across all catalysts from 36.2 to 44.2% to 25.6–31.1% (Table S2), indicating that partial oxidation of Ru(0) by PMS occurred. This passivation of hcp-Ru and fcc-Ru likely suppressed PMS activation and direct electron transfer, respectively. Nevertheless, quenching tests on the used catalysts revealed that, even with a reduced metallic fraction, the Ru catalysts retained both nonradical and radical pathways (Figure S16), despite a modest increase in methanol inhibition for the used fcc-Ru catalysts. These slight changes are also derived from the formation of oxidized Ru species after the first cycle; correspondingly, FE-TEM/EDX analysis still revealed the (111) lattice fringe of fcc-Ru (0.22 nm) in the Ru/t-ZrO₂ and Ru/mixed-ZrO₂ catalysts (Figure S17), alongside a lattice fringe in the spacing of 0.28 nm, which is assignable to fcc-RuO₂ (Figure S17a3). This suggests that PMS oxidation converted metallic fcc-Ru to the metastable fcc-RuO₂. In Ru/m-ZrO₂, hcp-Ru is well preserved, as evidenced by lattice spacings of 0.21 nm for (200) and 0.23 nm for (010) hcp-Ru. Therefore, it can be concluded that the nanostructures are largely retained after reuse without fcc → hcp interconversion, and the catalysts maintain their nanostructure-specific reactivity and mechanisms. The modest decrease in catalytic activity could be attributed to the partial oxidation of metallic Ru.

3.4. Influence of Ru Nanostructures on Defluorination of Capecitabine

The different reaction mechanisms of the fcc- and hcp-Ru catalysts also impacted capecitabine defluorination. Since defluorination was calculated based on the ratio of released F^– to the total fluorine content in the initial capecitabine concentration (eq , Figure a), it may not accurately reflect the extent of defluorination relative to the amount of capecitabine actually degraded. To address this limitation, we estimated F^– selectivity, as defined in eq . Interestingly, Ru/m-ZrO₂ exhibited the highest F^– selectivity (97.6 ± 1.4%) (Figure b), indicating that the decomposition of capecitabine consistently involved defluorination, with SO₄ ^•– generated by hcp-Ru preferentially attacking the C–F bonds. In contrast, Ru/t-ZrO₂ and Ru/mixed-ZrO₂ exhibited relatively lower 67.2 ± 1.2% and 40.9 ± 0.7%, respectively, implying that the nonradical pathway driven by fcc-Ru is less effective for the C–F bond cleavage than SO₄ ^•–. Consistently, Ru/m-ZrO₂ consistently exhibited high F^– selectivity (91.5–98.2%) compared to Ru/t-ZrO₂ and Ru/mixed-ZrO₂ (32.94–68.96%) across initial PMS concentrations ranging from 0.1 to 1.0 mM (Figure S18).

6.

(a) Time profiles of capecitabine defluorination during catalytic oxidation using Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/mixed-ZrO₂. (b) Capecitabine removal (%) and F^– selectivity (%) of Ru catalysts. Conditions: [capecitabine]₀ = 10 mg L^–1; [catalyst]₀ = 0.1 g L^–1; [PMS]₀ = 0.5 mM. Schematic diagram of the defluorination pathway for the (c) nonradical mechanism and (d) radical mechanism, with energy profiles. Optimized geometries of the reactant, transition state (TS), intermediate (IM), and product, along with the related bond lengths (Å), are provided. Gray, white, light green, yellow, blue, and red colors represent carbon, hydrogen, fluorine, sulfur, nitrogen, and oxygen atoms, respectively. Activation energies (ΔG‡, kcal mol^–1) are also noted.

This phenomenon can be attributed to SO₄ ^•–, which is highly active in cleaving the C–F bond, as evidenced by the DFT results. Using the nudged elastic band method, we simulated the possible defluorination route of capecitabine by SO₄ ^•–, locating the TS along the minimum energy path. The entire reaction pathway was divided into three steps. In the first step, SO₄ ^•– approached the C₁₆–F₁₇ site, decreasing the C₁₆–O₂₂ distance from 3.25 to 2.33 Å through TS1 (Figures d and S20). This step was barrierless (ΔG‡ = −1.52 kcal mol^–1), indicating that it can readily occur. In the second step, SO₄ ^•– was added to the C₁₆ site of capecitabine, forming the intermediate adduct, C₁₅H₂₂FN₃O₆–SO₄ ^•–, denoted as IM2, with a ΔG of −7.43 kcal mol^–1. During this step, the C₁₆–F₁₇ bond stretched from 1.33 (in IM1) to 1.47 Å (in IM2) via TS2, with a free energy of activation (ΔG‡) of 2.35 kcal/mol. Finally, the third step involved the cleavage of the C₁₆–F₁₇ bond, which increased to 3.54 Å via TS3. The ΔG and ΔG‡ values of this step were −5.20 and 4.65 kcal mol^–1, respectively. Therefore, it was confirmed that SO₄ ^•–-mediated defluorination is overall downhill and proceeds more easily by overcoming significantly lower energy barriers, indicating that it is both thermodynamically and kinetically favorable.

For the nonradical pathway, both capecitabine (C₁₅H₂₂FN₃O₆) and PMS (HSO₅ ^–) approached the fcc-Ru surface, where fcc-Ru facilitated electron capture and shuttled the electron to PMS (Figure c). This electron transfer resulted in the formation of the capecitabine radical cation (C₁₅H₂₂FN₃O₆ ^+•). The unstable radical could undergo defluorination via the S_N2 nucleophilic substitution reaction, which is also well established in hydrolysis and dichlorination of acyl chloride. Specifically, the reaction is initiated by the oxygen atom of the H₂O molecule attacking the C₁₆ site, which leads to the leaving of F₁₇. The leaving F^– then attracts H₅₀ by forming HF as a byproduct (Figure S19). This reaction pathway is thermodynamically favorable (ΔG = −14.5 kcal mol^–1) but requires a modest ΔG‡ of 33.6 kcal mol^–1, implying that the SO₄ ^•–-dominant radical pathway has greater advantage in defluorination than the nonradical pathway. Despite the lower F^– selectivity of the fcc-Ru catalysts, the overall abundance of fluorinated TPs was comparable to that observed with hcp-Ru/m-ZrO₂. This could be attributed to the formation of poorly ionizable or too small and early eluting fluorinated TPs that are difficult to detect using the LC–MS and nano LC–HRMS/MS methods employed in this study, warranting further investigations.

3.5. Influence of Ru Nanostructure on Oxidative Degradation Pathway of Capecitabine

The degradation pathways of capecitabine were investigated using nano LC–HRMS/MS. Initially, the identification and structural estimation of TPs were conducted by analyzing accurate masses and MS² fragmentation patterns, measured with high mass accuracy (to 5 decimal places) using nano LC–HRMS/MS (Table and Figures S21–S25). The interpretation of the MS² product ion spectrum of capecitabine is prerequisite for elucidating the molecular structures of unknown TPs (Figure S21a). Capecitabine shows characteristic fragmentations at m/z 117.0546 (C₅H₉O₃ ⁺), m/z 244.1088 (C₁₀H₁₅FN₃O₃ ⁺; via the C–N bond scission), m/z 174.03067 (C₅H₅FN₃O₃ ⁺; via the subsequent loss of C₅H₁₀), and m/z 130.04102 (C₄H₅FN₃O⁺; further loss of a CH₂O₂). This fragmentation pattern of the parent compound provides structure insights into the newly formed TPs. A detailed analysis of the TPs, along with interpretation of their MS2 product ion spectra (Figures S21–25), is presented in the Supporting Information (Text S9). Subsequently, the degradation pathways of capecitabine for each Ru catalyst were proposed based on the identified TPs (Figures , , and S26). Time profiles of the selected TPs were monitored by LC–MS (Figures S27–S29).

3. TPs Identified Using Nano LC–HRMS/MS.

name	formula	Δmass [ppm]	calc. MW	m/z ([M + H]+)	RT [min]	Ru/t-ZrO2	Ru/m-ZrO2	Ru/mixed-ZrO2
capecitabine	C15H22FN3O6	–2.6	359.1483	360.1556	11.087	parent compound	parent compound	parent compound
TP117 (I)	C5H8O3	–0.19	116.04732	117.0546	1.482	O	O	O
TP117 (II)	C5H8O3	–2.03	116.04711	117.05439	11.16	O	O	O
TP174	C5H4FN3O3	–1.54	173.02340	174.03068	10.204	O	O	O
TP85	C5H8O	1.66	84.05765	85.06493	2.542	O	O	O
TP151	C6H14O4	–1.67	150.08896	151.09624	1.470	O	O	O
TP173	C8H12O4	–2.01	172.0732	173.0805	7.361	O	X	O
TP242	C9H11N3O5	–2.17	241.0694	242.0766	9.177	O	O	O
TP243	C9H10N2O6	–2.26	242.0533	243.0606	1.006	O	O	O
TP244	C10H14FN3O3	–2.02	243.1014	244.1087	10.208	O	O	O
TP258	C10H15N3O5	4.05	257.10221	258.10949	0.87	O	O	X
TP356(I)	C15H21N3O7	–1.88	355.1373	356.1446	7.391	O	X	O
TP356(II)	C15H21N3O7	–2.08	355.1373	356.1449	9.274	O	O	O
TP358	C15H20FN3O6	3.16	357.1347	358.142	11.183	X	O	O
TP374(I)	C15H23N3O8	–1.78	373.1478	374.15513	7.306	O	O	O
TP374(II)	C15H23N3O8	–1.91	373.1478	374.15508	9.223	O	O	O
TP376	C15H22FN3O7	–1.41	375.1437	376.15092	11.145	O	O	O
TP390	C15H23N3O9	–1.44	389.1429	390.15	10.799	O	X	X
TP392(I)	C15H22FN3O8	–1.43	391.13853	392.14581	9.672	X	O	O
TP392(II)	C15H22FN3O8	–1.56	391.13848	392.14576	10.722	O	O	O
TP177	C6H12N2O4	–2.60	176.07925	177.08653	1.012	O	O	O
TP139	C6H6N2O2	–2.77	138.10408	139.11136	11.592	X	O	X
TP175	C6H10N2O4	–2.33	174.06365	175.07093	1.167	O	O	O
TP134	C5H11NO3	–0.83	133.07378	134.08106	1.011	O	O	O
TP334	C13H23N3O7	–1.96	333.15295	334.16022	11.173	O	O	O
TP346	C14H23N3O7	–1.67	345.15302	346.16030	10.498	O	O	O
TP332	C13H21N3O7	–1.86	331.13733	332.14461	11.555	O	O	O
TP232	C8H13N3O5	–1.92	231.08508	232.09235	1.012	O	O	O
TP203(I)	C7H10N2O5	–1.95	202.05858	203.06585	1.332	O	O	O
TP203(II)	C7H10N2O5	–1.94	202.05858	203.06586	1.726	O	O	O
TP203(III)	C7H10N2O5	–1.67	202.05863	203.06591	1.475	O	X	O
TP215(I)	C8H10N2O5	–2.02	214.05854	215.06581	1.013	O	X	O
TP215(II)	C8H10N2O5	–1.65	214.05862	215.06589	1.311	O	O	O
TP218	C8H15N3O4	–2.14	217.10579	218.11307	11.174	O	O	O
TP225	C9H8N2O5	–2.32	224.04280	225.05008	1.043	O	O	O
TP144	C6H9NO3	–1.42	143.05804	144.05632	1.475	O	O	O
TP220	C7H13N3O5	–1.68	219.08515	220.09243	1.467	O	X	O
TP123	C6H7N2O	–0.39	122.04796	123.05524	1.175	O	O	O

^aThe corresponding molecular structures are illustrated in Figures , , and S26.

7.

Proposed transformation pathways for capecitabine via nonradical mechanism catalyzed by Ru/t-ZrO₂. The TPs in the yellow and blue boxes represent fluorinated and defluorinated TPs, respectively. All TPs are categorized based on the number of C and F atoms.

8.

Proposed transformation pathways for capecitabine via radical mechanism catalyzed by Ru/m-ZrO₂. The TPs in the yellow and blue boxes represent fluorinated and defluorinated TPs, respectively. All TPs are categorized based on the number of C and F atoms.

Despite the mechanistic differences between fcc-Ru (direct electron transfer in nonradical pathway) and hcp-Ru (SO₄ ^•–-mediated oxidation), 27 out of 37 TPs were commonly identified (Table ), indicating a significant overlap in the degradation pathways. These shared degradation routes include hydroxylation, hydroxylative defluorination, C–C bond cleavage, C–N bond cleavage, and pyrimidine ring opening. However, certain TPs (i.e., TP356­(I), TP390, TP203­(II), TP215­(I), and TP220) were exclusively observed under the nonradical pathway mediated by fcc-Ru catalysts. In contrast, TP139 (C₆H₆N₂O₂) was detected only in the radical pathway using hcp-Ru/m-ZrO₂. Thus, the distinct appearance of these TPs likely arises from mechanistic differences between Ru nanostructures, resulting in partially divergent TP profiles.

The degradation pathways of capecitabine mediated by different Ru catalysts are illustrated in the Supporting Information (Text S9). Briefly, hydroxylation and hydroxylative defluorination were identified as the primary reactions common to both pathways. In the nonradical pathways of Ru/t-ZrO₂, capecitabine underwent (i) mono- and dihydroxylation and (ii) hydroxylative defluorination followed by mono- and dihydroxylation, leading to the formation of TP376 (C₁₅H₂₂FN₃O₇), TP392­(II) (C₁₅H₂₂FN₃O₈), TP356­(I–II) (C₁₅H₂₁N₃O₇), TP374­(I–II) (C₁₅H₂₃N₃O₈), and TP390 (C₁₅H₂₃N₃O₉) (Figure ). In the radical pathway of Ru/m-ZrO₂, similar TPs including TP376, TP392­(I–II), TP356­(II), and TP374­(I–II) were also identified (Figure ). These findings indicate that both reaction mechanisms promote hydroxylation and defluorination of the aminofluoropyrimidinone moiety in the early stage of reaction. However, the formation of distinct TPs suggests that each pathway exhibits partially diverging patterns in the hydroxylation position.

Subsequently, a progressive shortening of chain (C₁–C₅) within the pentyl carbamate (C₆H₁₃NO₂) moiety was observed for all Ru catalysts (Figures , , and S26). This resulted in TP346 (C₁₄H₂₃N₃O₇), TP334 (C₁₃H₂₃N₃O₇), and TP332 (C₁₃H₂₁N₃O₇₀) with the sugar moiety intact (Figures S23–24). Additionally, cleavage of the C₇–O₆ bond led to the formation of TP258 (C₁₀H₁₅N₃O₅), while further C–N bond (N₉–C₇ and C₁₀–N₉) scissions led to the formation of TP242 (C₉H₁₁N₃O₅), TP243 (C₉H₁₀N₂O₆), and TP225 (C₉H₈N₂O₅), which are detected in all pathways (Figures , , and S26). Interestingly, after the loss of the pentyl carbamate (C₆H₁₃NO₂) moiety, ring opening of the fluoropyrimidinone moiety was initiated, followed by additional hydroxylation to form TP215­(I–II) (C₈H₁₀N₂O₅) and TP232 (C₈H₁₃N₃O₅). Subsequently, TP232 underwent further destruction via the C–N bond cleavage and hydroxylation to become TP220 (C₇H₁₃N₃O₅) by only fcc-Ru catalysts (Figures and S26). Additionally, the C–N bond cleavage in TP215­(I–II) led to the formation of TP203­(I–II) (C₇H₁₀N₂O₅). Further cleavage of either C₁₅–N₁₄ or C₁₀–N₁₁ in TP203­(I–II) produced TP177 (C₆H₁₂N₂O₄) and TP175 (C₆H₁₀N₂O₄), both featuring a urea moiety linked to a sugar fragment. Oxidation of TP177 or TP175 led to TP144 (C₆H₆NO₃) via the destruction of the urea moiety. Lastly, TP177 and TP175 underwent dehydration to produce TP139 (C₆H₆N₂O₂) and TP123 (C₆H₇N₂O), or underwent C–N bond cleavage (i.e., C₁₂–N₁₄ or C₁₈–N₁₄) to yield TP134 (C₅H₁₁NO₃) and TP117­(I) (C₅H₈O₃) (Figures , , and S26).

Another degradation route involved initial cleavage of the sugar moiety, yielding TP244 (m/z 244.1087, C₁₀H₁₄FN₃O₃), followed by C₅–O₆ cleavage to form TP174 (C₅H₄FN₃O₃). Alternatively, TP244 underwent ring opening of the fluoropyrimidinone moiety via defluorination and hydroxylation, producing TP218 (C₈H₁₅N₃O₄). Oxidative degradation of the detached pentyl group gave rise to TP151 (C₆H₁₄O₄), TP117­(II) (C₅H₈O₃), and TP85 (C₅H₈O) (Figures , , and S26).

4. Conclusion

In this study, three ZrO₂ supports with distinct crystalline phases (i.e., monoclinic, tetragonal, and a mixed phase) were employed to synthesize Ru/t-ZrO₂, Ru/m-ZrO₂, and Ru/mixed-ZrO₂ catalysts, respectively. Surface characterization revealed that the crystalline phase of ZrO₂ strongly influences the resulting Ru nanostructures; fcc-Ru preferentially formed on the tetragonal phase and hcp-Ru on the monoclinic phase of ZrO₂. Interestingly, this ZrO₂ phase-dependent Ru nanostructure significantly influenced both the catalytic reactivity and the dominant reaction mechanism during PMS-based oxidation of capecitabine. Specifically, fcc-Ru/t-ZrO₂ and fcc-Ru/mixed-ZrO₂ mediated a nonradical pathway via direct electron transfer, whereas hcp-Ru/m-ZrO₂ induced a radical pathway through PMS activation to generate SO₄ ^•– for oxidative degradation. These mechanistic distinctions between fcc-Ru and hcp-Ru were well preserved after catalyst recycling and resulted in notable variations in the overall catalytic performance, TOF, defluorination, and F^– selectivity under various conditions. Although the TOF provides insight into the high intrinsic activity of hcp-Ru, fcc-Ru exhibits a higher degradation rate even under environmentally realistic conditions. Furthermore, nano LC–HRMS/MS enabled the detailed elucidation of TP structures, revealing that, despite similar degradation routes, fcc-Ru catalysts showed diverging patterns of TP formation compared to hcp-Ru/m-ZrO₂. Overall, this study is the first to demonstrate that the crystalline phase of ZrO₂ governs the formation of Ru nanostructures, which, in turn, dictates catalytic reactivity and the underlying reaction mechanism. While nonradical pathways involving noble metals such as Ru, Ir, Pt, and Au in persulfate-based AOPs have been recently reported, this work further unveils a Ru nanostructure-dependent mechanistic divergence. These findings highlight a promising strategy for engineering catalyst by controlling metal nanostructures through support crystallography, thereby selectively steering oxidation mechanisms. We believe that this work provides a new design principle for developing catalysts with tailored reaction mechanisms, enabling the construction of more advanced strategies for environmental water remediation.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.5c21822.

• Chemicals, catalyst preparation, characterization of ZrO₂ and Ru/ZrO₂ catalysts, analytical methods (LC–MS), DFT calculations, H₂-TPR and XPS analyses, scavenging tests, effects of natural organic matter, pH, bicarbonate, oxidant dosage, and initial concentration, identification of TPs by MS² spectra, catalyst recyclability and postreaction characterization, proposed reaction pathways, and time profiles of capecitabine degradation and defluorination (PDF)

§.

35-402, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea

The authors declare no competing financial interest.