Catalytic Reduction of Aqueous Perchlorate at Neutral pH

Abstract

Perchlorate (ClO₄ ^–) is a pervasive and inert pollutant threatening human health. Catalytic reduction of ClO₄ ^– is imperative for water environment remediation and promising for space exploration. Previous catalysts using oxo Mo^IV and Re^V species required acidic pH to reduce aqueous ClO₄ ^–. Herein, we developed a ligand-enhanced Ru⁰ nanoparticle catalyst for rapid ClO₄ ^– reduction by 1 atm H₂ at pH 7 and 20 °C. Molecular organic nitrogen ligands, particularly cis-1,2-diaminocyclohexane (cis-DACH), accelerated Ru⁰-catalyzed ClO₄ ^– reduction by up to 80-fold. Nitrogen-functionalized carbon (NC) support also provided up to 5-fold acceleration. The final [cis-DACH]­Ru/NC catalyst demonstrated unprecedented activity and robustness for ClO₄ ^– reduction at neutral pH. Kinetic analyses and instrumental characterizations found that the nitrogen ligands not only boosted the reactivity of exposed Ru⁰ sites by 63-fold (31-fold by cis-DACH and 2-fold by the NC support) but also enhanced ClO₄ ^– adsorption by 180%. This novel paradigm of tuning nanostructured catalysts by organic ligands will benefit environmental and energy applications for the degradation and utilization of ClO₄ ^–.

Introduction

Perchlorate (ClO₄ ^–) is widely distributed in the water environment due to its natural occurrence and the manufacture of energetic materials such as rocket fuels, munitions, and pyrotechnics. Excessive exposure to ClO₄ ^– can disrupt thyroid hormone production and thus impact the growth, development, metabolism, and mental function of humans. Recent studies identified new mechanisms of ClO₄ ^– toxicity in the thyroid, suggesting that its health threat was previously underestimated. ClO₄ ^– is also linked to more health issues via nonthyroidal mechanisms. In the United States, California and Massachusetts set the maximum contamination levels for ClO₄ ^– in drinking water at 6 and 2 μg L^–1, respectively. In 2023, China began nationwide regulation at 70 μg L^–1. In 2026, the United States Environmental Protection Agency resumed efforts toward federal regulation at 20 μg L^–1. Besides drinking water, high ClO₄ ^– contents in turnip, lettuce, spinach, and tea also indicate the need for addressing ClO₄ ^– contamination in agriculture. , Furthermore, the discovery of ClO₄ ^– on Mars, the moon, and meteorites suggests its widespread presence in the Solar System. − Therefore, ClO₄ ^– reduction technologies are of extensive interest for water purification, food safety, and human extraterrestrial exploration.

In aqueous solution, ClO₄ ^– is highly inert under various chemical reduction conditions and thus commonly used as an innocent electrolyte in electrochemical systems. Oxometallates of Mo, W, Re, and Os catalyzed ClO₄ ^– reduction by special electron sources (Figure a) via oxygen atom transfer (OAT) in the presence of concentrated 1–10 M acids. In contrast, anaerobic bacteria reduce ClO₄ ^– with hydrogen gas, methane, etc., using the Mo cofactor with molybdopterin ligands. , The reactivity with ClO₄ ^– at neutral pH is attributed to the hydrogen bonding between ClO₄ ^– and an amino acid residue in the enzyme pocket. Similarly, a hydrogen-bonding environment is critical for the high reactivity of biomimetic Fe complexes (moisture-sensitive) with ClO₄ ^–. , A biomimetic Mo complex uses Sc³⁺ as a Lewis acid to accelerate the reaction with ClO₄ ^–. The success of the Fe and Mo complexes has led to the recent advances of using ClO₄ ^– as a mild oxidizer in organic synthesis for C–H activation and functionalization. −

1.

Summary of aqueous perchlorate reduction systems.

In environmental remediation, despite the high activity of perchlorate reductase, microbial reactors typically undergo a long startup phase of several days to months. − To meet various needs of ClO₄ ^– reduction, an abiotic chemical system is highly desirable for both immediate functionality and excellent performance. Earlier environmental remediation efforts used Fe⁰ nanoparticles and Fe^II solution to reduce ClO₄ ^–, but high temperature and excessive reductants were required (Figure b). The enzyme-inspired Re–Pd/C , and Mo–Pd/C , catalysts (Figure c) realized ClO₄ ^– reduction under mild conditions, such as 1 atm H₂ as the electron donor, pH 3.0 (1 mM of H⁺), and 20 °C (eq ):

$${{\mathrm{ClO}}_4}^{-}+4{\mathrm{H}}_2\to {\mathrm{Cl}}^{-}+4{\mathrm{H}}_2\mathrm{O}$$ 1

These catalysts use Re^V–VII or Mo^IV–VI redox cycles to enable OAT from ClO₄ ^–. , In comparison to catalysts prepared from inorganic KReO₄ and Na₂MoO₄ precursors, the coordination with select pyridine (N), oxazoline-phenolate (N–O), and bipyridine (N–N) ligands (Figure d) accelerated ClO₄ ^– reduction by 12–5200 fold. However, the proton-assisted OAT suffers from significant activity loss at pH >4, ,,, preventing the application at neutral pH. Another major challenge is the hydrogenation of aromatic ligands , or decomposition of the ligand–metal complex sites, , resulting in an irreversible loss of catalyst activity. Interestingly, NO₃ ^– can deactivate both Re and Mo catalysts developed for ClO₄ ^– reduction. ,

Herein, we shift the catalyst design rationale to a new paradigm. Among platinum group metal (PGM) nanoparticle catalysts, Ru⁰ has unique reactivity at pH 7. A bimetallic Ru–Pd/C has exhibited the fastest ClO₃ ^– reduction to date but little activity in ClO₄ ^– reduction. A Ru/CeO₂ catalyst requires 80 °C to achieve a satisfactory rate of ClO₄ ^– reduction. But heating large volumes of water is a major technical obstacle for environmental applications. Although morphology control or support modification might optimize the activity of Ru nanoparticles, these strategies usually involve laborious procedures. In this study, we modified the properties of Ru⁰ nanoparticles with (i) molecular diamine ligands and (ii) nitrogen functional groups on the support material, enabling the unprecedented ClO₄ ^– reduction activity at pH 7. We elucidated the contributions of the two nitrogen ligand types and demonstrated the high catalyst robustness under real-world conditions.

Results

Substantially Enhanced Perchlorate Reduction by Nitrogen Ligands

Inspired by the in-situ preparation of our earlier Mo catalysts, we added various nitrogen ligands to the water suspension of a commercial Ru/C catalyst and observed substantially enhanced ClO₄ ^– reduction at 20 °C. To our surprise, the saturated cis-1,2-diaminocyclohexane (cis-DACH) provided the highest activity enhancement (Table and Figures S1–S3). Compared to the original Ru/C, the initial turnover frequency (TOF₀) was accelerated by 80-fold. The activity enhancement by cis-DACH was over twice that of its trans isomer (Table , entry 1 versus 2). The six-mmembered ring backbone also provided a higher activity than other cis-1,2-diamino analogs with five- and seven-membered rings (Table , entry 1 versus 7 and 8). The cis-DACH was also superior to the unsaturated, aromatic, and linear aliphatic diamine ligands (Table , entries 5, 6, and 9–11), and to all pyridine-containing ligands (Table , entries 12–22). In general, ligands with steric hindrance resulted in lower activity (Table , entry 2 versus 3 and 4; entry 9 versus 10 and 11; entry 13 versus 16 and 17), indicating the interaction between Ru and N atoms. The bipyridine (bpy) and phenanthroline (phen) ligand families also showed higher reaction rates from increased electron-donating character (Table , entries 12–15 and 18–20). Therefore, the nitrogen ligands appeared to enhance catalytic performance by altering the electronic properties of the Ru⁰ nanoparticle surface (as characterized in a later section).

1. Perchlorate Reduction Activity of Ru/C Enhanced by Various Ligands .

^aReaction conditions: 2.0 g L^–1 of 5 wt % Ru/C (Alfa Aesar No. 44338) with 1 mM ligand added in the water suspension, 1 mM ClO₄ ^–, pH 7.0, 1 atm of H₂, 20 °C. ^b The calculation assumed that each surface Ru atom reacted with all ClO _x ^– (x = 4, 3, 2, and 1) substrates. Details are described in Text S1 in the Supporting Information.

Having confirmed the positive effect of molecular ligands added from the aqueous solution, we further examined the effect of nitrogen functional groups from the support material. We prepared an amine- and amide-functionalized carbon support (NC) using a “soft nitriding” method with urea at 300 °C. Then we prepared Ru/C and Ru/NC by incipient wetness impregnation of RuCl₃ followed by H₂ reduction at an elevated temperature (detailed procedures in the Supporting Information). Elemental analyses found nitrogen contents of 0.75% and 3.32% in Ru/C and Ru/NC, respectively. Compared with Ru/C, the nitrogen groups in Ru/NC increased the ClO₄ ^– reduction activity by 460% (Figure a). Adding cis-DACH to the two catalysts further increased the total nitrogen contents in [cis-DACH]­Ru/C and [cis-DACH]­Ru/NC to 1.70 and 3.93%, respectively. In the presence of cis-DACH, the nitrogen groups in Ru/NC still increased the ClO₄ ^– reduction activity by 69% (Figure a).

2.

(a) Time profiles of ClO₄ ^– reduction by Ru catalysts; (b) chlorine balance for ClO₄ ^– reduction by [cis-DACH]­Ru/NC; (c) pH effect on [cis-DACH]­Ru/NC. For panels (a)–(c): 2.0 g L^–1 of lab-prepared catalysts (5 wt % Ru), 2 mM cis-DACH added in water. (d) Ligand-enhanced ClO₄ ^– reduction with Ru colloids (prepared from 1 mM RuCl₃, 10 mM NaBH₄, and 10 mM cis-DACH) and Ru–Pd/C (2.0 g L^–1 catalyst; 5 wt % Ru and 5 wt % Pd, 1 mM 2,2′-bipyridine (bpy) or 1,10-phenanthroline (phen)). For panels (a)–(d): 1 mM ClO₄ ^–, pH 7.0, 1 atm of H₂, 20 °C.

Catalyst Performance

At pH 7, the dually enhanced [cis-DACH]­Ru/NC catalyst outperformed all reported abiotic catalytic systems in ClO₄ ^– reduction (Table S1). The reaction rate at 20 °C was similar to that previously reported for Ru/CeO₂ at 80 °C. A good mass balance was observed between ClO₄ ^– and Cl^– (Figure b), indicating the complete reduction of ClO₄ ^– to Cl^– without accumulating ClO _x ^– intermediates. The [cis-DACH]­Ru/NC and [cis-DACH]­Ru/C showed similar responses to solution pH (Figure c and Figure S4), with the highest activity at pH 5–7. Because the original Ru/C preferred acidic pH (Figure S5), we attribute the lowered activity at acidic pH to the protonation of −NH₂, which impeded coordination to Ru⁰ nanoparticles. The reason for the inferior activity at pH 9 for all carbon-supported Ru catalysts remains elusive. However, the previously reported Ru/CeO₂ catalyst showed similar activity across pH 4–9, suggesting the promise of combining the organic ligand and novel support materials to maintain activity over a broader pH range. The optimized cis-DACH dose was 2 mM (Figure S6), with 59% immobilized on Ru/NC (41% in aqueous solution, determined by total nitrogen analysis). We note that the immobilized ligand (1.18 mM) is more than 10 times the surface Ru atoms in Ru/NC (0.09 mM, calculated using the metal dispersion, Table S2). Thus, the majority of cis-DACH was adsorbed on the carbon surface. Lowering the cis-DACH dose reduced the dissolved portion (Table S3) at the cost of catalytic activity (Figure S6). After filtration, the redispersion of the collected [cis-DACH]­Ru/NC solid in water still released 10% of the initial cis-DACH into the aqueous phase. Thus, cis-DACH was distributed in an equilibrium between the catalyst surface and water. However, the dissolved cis-DACH maintained the same concentration throughout ClO₄ ^– reduction (Figure S7). In practical application scenarios, the dissolved cis-DACH in the treated water could be protonated at low pH and captured by cation exchange resin. , The ligand structure could also be further modified with hydrophobic functional groups to substantially reduce the distribution in the aqueous phase.

Broader Examinations of the Ligand–Nanoparticle Design

We further evaluated the adaptability of the [cis-DACH]­Ru/NC design to other metals and systems. However, cis-DACH resulted in very limited ClO₄ ^– reduction for Rh/C and Ir/C and no activity for Pd/C and Pt/C (Figure S8), indicating the critical role of the Ru element. We also confirmed that Ru⁰, rather than Ru^II or Ru^III, is the active site for ClO₄ ^– reduction. First, mixing RuCl₃ and cis-DACH in water under 1 atm H₂ provided no activity. The direct reduction of aqueous Ru^III by H₂ under ambient temperature and pressure was sluggish. Instead, NaBH₄ effectively reduced Ru^III into Ru⁰ colloids (Figure S9) and the presence of cis-DACH enabled ClO₄ ^– reduction (Figure d). Second, adding presynthesized [Ru^II(phen)₃]²⁺, [Ru^II(bpy)₃]²⁺, Ru^II(bpy)₂Cl₂, and [Ru^II(NH₃)₆]²⁺ complexes (Table S4) into Pd/C under H₂ did not reduce ClO₄ ^–. In contrast, mixing RuCl₃, free phen or bpy ligands, and Pd/C under H₂ enabled ClO₄ ^– reduction (Figure d). The discrepancy between the two strategies can be attributed to the rapid formation of Ru⁰ nanoparticles from RuCl₃, which can be substantially accelerated by Pd/C. The results suggest the active site as ligand-coordinated Ru⁰ nanoparticles rather than single-atom Ru⁰(ligand) _x complexes. However, the performances of [phen]­Ru–Pd/C and [bpy]­Ru–Pd/C were very similar to that of the initially screened [phen]­Ru/C and [bpy]­Ru/C (Figures S1a and S1c). Hence, we did not further pursue the incorporation of Pd, which only accelerated the formation of Ru⁰ but not the reduction of ClO₄ ^–.

Mechanistic Elucidation

After finalizing the catalyst preparation strategy and procedure, we elucidated how cis-DACH and the nitrogen groups on NC interacted with Ru and enhanced ClO₄ ^– reduction. First, the nitrogen ligands did not change the morphology of Ru nanoparticles. Scanning transmission electron microscopy (STEM) characterization showed good dispersion of Ru particles and very similar size distribution (1.6 ± 0.3 nm for Ru/C and 1.5 ± 0.3 nm for Ru/NC; see Figure a and b, and Figures S10–S17). The small particle size resulted in no sharp Ru⁰ crystal peaks in X-ray diffraction (XRD) analysis (Figure S18). CO chemisorption revealed very similar Ru dispersion in Ru/C (10.2%) and Ru/NC (12.6%). Therefore, nitrogen groups on the NC support did not result in significantly improved Ru dispersion. Because cis-DACH was added to the Ru catalysts at room temperature, as expected, it did not alter the metal morphology (Figure c and d). However, the decrease of the apparent dispersion values from 12.6% (Ru/NC) to 3.8% ([cis-DACH]­Ru/NC) suggested that cis-DACH hindered the coordination of gaseous CO to ∼70% of surface Ru atoms in the dried catalyst. We further used suspended WO₃ particles in water as a probe of H₂ activation − for the effect of cis-DACH coordination. As expected, cis-DACH added to Ru/C slowed the reductive conversion of bright yellow WO₃ to dark blue H _x WO₃ considerably (Figure i versus Figure j and Figure S19), indicating that cis-DACH also hindered the access of dissolved H₂ to a significant fraction of surface Ru atoms. Therefore, it is reasonable to postulate that the cis-DACH-coordinated Ru atoms cannot bind and reduce aqueous ClO₄ ^–.

3.

(a–d) HAADF-STEM images of Ru catalysts, and EDX mapping of (e) Ru and (f) N on [cis-DACH]­Ru/NC; (g) initial rate of ClO₄ ^– reduction at various initial ClO₄ ^– concentrations and the LH model fits; (h) LH model-derived adsorption equilibrium constant K ₁ and rate constant k ₃ for Ru catalysts at 20 °C; color change of WO₃ powders (4 g L^–1) by (i) Ru/C and (j) [cis-DACH]­Ru/C under H₂ sparging at 60 °C. Reaction conditions: 2 g L^–1 of catalyst (5 wt % Ru), 2 mM cis-DACH, pH 7.0, 1 atm of H₂.

We conducted kinetic analysis with the Langmuir–Hinshelwood (LH) model (eq ) to differentiate the contributions of cis-DACH and the nitrogen groups on the NC:

$$r=\frac{K_1{k}_3[A]}{K_1[A]+1}$$ 2

where r is the apparent rate of ClO₄ ^– reduction in the bulk aqueous phase (mmol g_cat ^–1 h^–1), K₁ is the equilibrium constant for ClO₄ ^– adsorption (L mmol^–1), k ₃ is the rate constant for the reduction of surface-adsorbed ClO₄ ^– (mmol g_cat ^–1 h^–1), and [A] is the ClO₄ ^– concentration in the bulk aqueous phase (mmol L^–1). Text S2 provides mass transfer analysis and derivation of eq . The values of K ₁ and k ₃ (Figure h) were obtained by fitting the LH model with kinetic data from reducing 0.05–10 mM ClO₄ ^– (Figures g and S20). The [cis-DACH]­Ru/C catalyst exhibited 180% higher K ₁ and 940% higher k ₃ than Ru/C. Hence, the addition of cis-DACH ligand substantially enhanced both the adsorption and the reduction of ClO₄ ^–. In particular, considering the 70% decrease of exposed Ru sites and that the ClO₄ ^– reduction could only occur on Ru (see below), the cis-DACH coordination on Ru nanoparticles increased the ClO₄ ^– reduction activity of the remaining (30%) exposed Ru atoms in Ru/C by 31-fold. Further switching the support to NC doubled k ₃ but did not yield a higher K ₁ for [cis-DACH]­Ru/NC. Therefore, nitrogen groups on the NC support could synergistically accelerate the reduction of adsorbed ClO₄ ^–. Although NC materials have been well-known for enhancing the positive surface potential, − their role in enhancing ClO₄ ^– adsorption was overwhelmed by cis-DACH. More importantly, as suggested by CO chemisorption data, neither NC support nor cis-DACH increased the number of Ru sites. Thus, the increased k ₃ could only be attributed to the enhanced activity of Ru nanoparticles. The vastly different activities of Ru, Rh, and Ir catalysts (Figure S8) have suggested that direct interaction between ClO₄ ^– and Ru is necessary. In other words, “H spillover” from an H₂-activating metal particle to a distantly adsorbed substrate on the carbon surface is unlikely to reduce ClO₄ ^–.

X-ray photoelectron spectroscopy (XPS) characterization revealed only one set of Ru 3p spin–orbit doublets for both Ru/C (Figure a) and Ru/NC (Figure b), with the 3p_3/2 binding energies (BE) of 462.8 and 462.4 eV, respectively. Both peaks indicate metallic Ru⁰, , and the 0.4 eV difference probably reflects changes in the electronic properties of surface Ru atoms due to coordination with nitrogen groups on NC support. The same BE shift was observed between [cis-DACH]­Ru/C and [cis-DACH]­Ru/NC (Figure c versus d). In comparison, the addition of cis-DACH triggered a minor BE shift of 0.2 eV for both Ru/C (Figure c versus a) and Ru/NC (Figure d versus b). Regarding nitrogen speciation in the catalysts, only one N 1s peak (400.1 eV) corresponding to – NH₂ , was observed in [cis-DACH]­Ru/C (Figure e). Three N 1s peaks corresponding to −NH₂ (399.9 eV, major, 49.5% abundance), graphitic N (401.6 eV), and pyridinic N (398.2 eV) , were fit for Ru/NC (Figure f). [cis-DACH]­Ru/NC also contained three species (Figure g). The cis-DACH addition elevated the −NH₂ peak abundance from 49.5% to 64.3%.

4.

(a–d) Ru 3p XPS spectra (empty dots) and fits (solid lines) of Ru catalysts; (e–g) N 1s XPS spectra (empty dots) and fits (solid lines) of Ru catalysts; (h) DRIFT spectra of Ru catalysts; (i) time profiles of ClO₄ ^– reduction by Ru/Al₂O₃ with and without cis-DACH. Reaction conditions: 2 g L^–1 of 5 wt % Ru/Al₂O₃, 2 mM cis-DACH, 1 mM ClO₄ ^–, pH 7.0, 1 atm of H₂, 20 °C.

To further probe the ligand effects on the surface electronic properties of Ru⁰ nanoparticles, we used diffuse-reflectance infrared Fourier transform spectroscopy (DRIFTS) to characterize the CO adsorption at 25 °C (Figure h). Since the carbon support can fully absorb IR irradiation, we prepared Ru/Al₂O₃ and [cis-DACH]­Ru/Al₂O₃. On the Ru/Al₂O₃ platform, cis-DACH also accelerated ClO₄ ^– reduction by 24-fold (Figure i), suggesting that the simple ligand addition method can be applied to multiple catalyst support materials, particularly when nitrogen-doping into the specific solid is difficult. The IR bands corresponding to Ru-multicarbonyl (α, Ru­(CO)_n species), on-top adsorbed (β) and bridge-bonded (γ) CO on Ru sites, as well as bridge-bonded CO on interfacial Ru sites (δ) were observed regardless of the presence of cis-DACH. The red shift of all IR bands (8–26 cm^–1) upon adding cis-DACH suggests that the Ru⁰ nanoparticles become more electron-rich. The enhanced back-donation of electrons to the 2π* antibonding orbitals of CO resulted in a lower frequency of molecular stretching. This trend corroborates the XPS-observed shift in the Ru 3p_3/2 BE upon ligand coordination. Therefore, kinetic analysis of k ₃ and material characterization data collectively confirmed that nitrogen ligands donated electrons to Ru⁰ nanoparticles, thereby accelerating ClO₄ ^– reduction. Notably, this mechanism is distinct from the reported ligand-nanoparticle systems for the hydrogenation of organic substrates, in which the ligand was added to prevent deeper reactions rather than enhance the reactivity.

Catalyst Robustness in Real World Situations.

The robustness of the [cis-DACH]­Ru/NC catalyst was further demonstrated under various highly challenging water matrices. The catalyst achieved >99.9% reduction of 100 mM ClO₄ ^– within 50 h (Figure a) with the turnover number (TON) of 4350. In the presence of concentrated 0.1 M NaCl, 2 M NaCl, and 1 M Na₂SO₄, the activity decreased by 54%, 93%, and 75%, respectively (Figure b). After ten spikes of 10 mM ClO₄ ^– (one spike per day) into the [cis-DACH]­Ru/NC suspension, the activity loss observed at day 10 was merely due to the accumulated Cl^– from ClO₄ ^– reduction (Figure c). ¹H NMR analysis found no chemical transformation of cis-DACH throughout the catalytic process (Figure S21). The saturated structure of cis-DACH is inert to hydrogenation, which was a major challenge for the previously reported Mo–Pd/C catalyst. The latter used 4,4′-diamino-2,2′-bipyridine as the best ligand and was deactivated by 60% under continuous exposure to 1 atm H₂ for a week. Moreover, the catalyst prepared from the simple mixing of cis-DACH and Ru nanoparticles did not deactivate upon exposure to air. Upon resuming the H₂ supply, the ClO₄ ^– reduction activity was fully restored (Figure d). The H₂ atmosphere preserved Ru as metallic nanoparticles, resulting in no detected Ru in the aqueous phase. Hence, the catalyst showed high promise of treating ClO₄ ^– in real wastewater, particularly for the straightforward preparation, mild reaction conditions, robustness, longevity, and easy handling.

5.

Time profiles of the reduction of (a) 100 mM ClO₄ ^–; (b) 1 mM ClO₄ ^– in the presence of concentrated Cl^– and SO₄ ^2–; (c) 10 mM ClO₄ ^– by freshly prepared (cycle 1) and reused (cycle 10, with nine previous spikes of 10 mM ClO₄ ^–) catalyst. Each ClO₄ ^– spike was allowed 24 h to achieve complete reduction into Cl^–, with ten spikes over a total of 10 days to complete; (d) 1 mM ClO₄ ^– reduction by [cis-DACH]­Ru/NC before and after air exposure for 1 h. The preparation of [cis-DACH]­Ru/NC took 1 h under 1 atm H₂. After air exposure, the catalyst suspension was exposed to 1 atm H₂ for an additional hour before adding ClO₄ ^–; the reduction of mixed ClO₄ ^– and NO₃ ^– in (e) synthetic IX regenerant brine and (f) tap water. Default reaction conditions: 2 g L^–1 of 5 wt % Ru/NC, 2 mM cis-DACH, 1 mM ClO₄ ^–, pH 7, 1 atm of H₂, 20 °C.

We further tested the performance of [cis-DACH]­Ru/NC in real-world scenarios, including ClO₄ ^– treatment in (i) waste brines resulting from ion-exchange resin regeneration and (ii) drinking water containing ClO₄ ^– as a disinfection byproduct or from source water. We prepared a synthetic waste brine containing all anions identified in a previous brine treatment study (Table ). Notably, due to the low selectivity of early IX resins for ClO₄ ^–, the waste brines produced during resin regeneration often contained significantly higher concentrations of nitrate and sulfate than ClO₄ ^–. ,− In this scenario, Re- and Mo-based catalysts require a two-stage treatment for ClO₄ ^– reduction, as NO₃ ^– can deactivate them and thus needs to be reduced first by another catalyst (e.g., In–Pd/Al₂O₃). , In sharp contrast, Ru catalysts can be highly effective in reducing NO₃ ^–, obviating the need for pretreatment using other catalysts. More than 99% of the 38 mM of NO₃ ^– was reduced by [cis-DACH]­Ru/NC within 10 h (Figure e). Although the water matrix substantially slowed the reaction rate (Figure e versus 2a), > 99.99% of ClO₄ ^– was reduced within 84 h. In the tap water that contained 0.4 mM NO₃ ^– at pH 8 (Table ), [cis-DACH]­Ru/NC reduced >99.99% of NO₃ ^– and 99.7% of the spiked 1 mM ClO₄ ^– within 4 and 10 h, respectively (Figure f). These positive results further demonstrate the value of [cis-DACH]­Ru/NC in practical ClO₄ ^– treatment, where various case-specific pretreatment processes can be integrated to further enhance the catalyst performance. Notably, the reaction time can be proportionally shortened by increasing the catalyst loading. , Adequate H₂ supply and catalyst retention at pilot-scale can be achieved by a stirred-tank reactor with microfiltration membrane fibers (Figure S22). Even at a low catalyst loading, this ClO₄ ^– reduction method does not require intensive energy input, such as heat or UV, to drive the reaction.

2. Composition of Synthetic IX Regenerant Brine and Tap Water.

component	concentration in synthetic IX regenerant brine	concentration in tap water
chloride	0.9 M	0.92 mM
perchlorate	1 mM	1 mM
nitrate	38 mM	0.4 mM
sulfate	48 mM	0.66 mM
phosphate	0.22 mM	0.5 μM
fluoride	NA	0.025 mM
alkalinity	7 g/L as CaCO3	0.17 g/L as CaCO3
pH	7.4	8

^aNa⁺ was the sole cation introduced alongside the anionic species. Although K⁺, Ca²⁺, and Mg²⁺ are all present in real brines in small quantities, they had negligible effects on reaction kinetics. ,

^bThe ClO₄ ^– concentration in the real brine was at 0.02 mM (∼2 mg L^–1). To ensure accurate ClO₄ ^– quantitation within the concentrated salt matrix and to compare with the majority of experiments using 1 mM ClO₄ ^– as the probe, we increased the concentration to 1 mM. The catalyst can effectively reduce >99% of the 0.02 mM ClO₄ ^– (Figure S20).

^cData was not available in the previous study.

^d0.14 M sodium bicarbonate was added.

^eWhile the reduction of ClO₄ ^– does not consume H⁺, the reduction of NO₃ ^– consumes H⁺ and raises the pH. The experiment used a 50 mL double-neck flask with rubber stoppers on both necks. One stopper held two needles for H₂ inlet and outlet/sampling port, while the other accommodated a pH electrode to monitor pH changes. H₂SO₄ (0.1 M) was introduced via the sampling needle when the pH exceeded the initial value (i.e., 7.4 or 8).

^fThe tap water was sourced from Southern California, with information on the concentrations of other components available in the local water quality report.

^gThe ClO₄ ^– concentration in the tap water was below the detection limit of ion chromatography (0.1 μM). We raised it to 1 mM for the accuracy of quantitation and convenient comparison.

Conclusions

Simply adding an aliphatic cis-DACH ligand can substantially enhance the activity of Ru⁰ nanoparticles. This configuration is the first case of rapid ClO₄ ^– reduction at pH 7 at ambient temperature and pressure. Modifying the carbon support with nitrogen functional groups also enhances the reaction rate. The resulting [cis-DACH]­Ru/NC shows excellent performance with a wide range of ClO₄ ^– concentrations and water matrix conditions. The nitrogen groups in the molecular ligand or the solid support modified the electronic properties of the Ru⁰ nanoparticles, thereby enhancing their reactivity with ClO₄ ^–. The nitrogen groups also enhanced ClO₄ ^– adsorption to the catalyst surface. The unique activity of Ru, the structure specificity of cis-DACH, and the electronic structure of nitrogen-coordinated Ru⁰ nanoparticles warrant further investigation. This study exemplifies a new catalyst design rationale of using organic modifiers to enhance the catalytic performance of metal nanoparticles.

Materials and Methods

Chemicals and Materials

RuCl₃·xH₂O (99.98%) and NaClO₄ (≥99%) were used as received from Sigma–Aldrich. Detailed information on support materials and commercial catalysts is described in Table S5. The nitrogen ligands were used as received from Oakwood Chemical, Alfa Aesar, Ark Pharm, Combi–Blocks, Sigma–Aldrich, and TCI. All aqueous solutions were prepared with Milli-Q water (resistivity >18.2 MΩ cm).

Catalyst Preparation

Nitrogen-functionalized carbon (NC)

NC was prepared following a previously reported “soft nitriding” method. One gram of activated carbon and 1.5 g of urea were mixed and ground in a crucible. The mixture was annealed at 150 °C for 2 h and then at 300 °C for another 2 h. The resulting powder was first washed with ethanol and Milli-Q water to remove the unreacted urea and then dried in a 75 °C oven.

Supported Ru catalysts

The Ru^IIICl₃ precursor was impregnated into the support material (C, NC, or Al₂O₃) by incipient wetness. The resulting wet paste was first dried in an oven at 75 °C for 12 h and then reduced with Ar/H₂ (v/v = 95/5) at 450 °C for 6 h to yield Ru⁰/C, Ru⁰/NC, and Ru⁰/Al₂O₃. ,

Ligand-modified Ru catalysts

A 50 mL flask was sequentially loaded with 100 mg of Ru catalysts, 50 mL of DI water, and the desired amount of nitrogen ligand. Since the protonation of nitrogen ligands increased the solution pH, H₂SO₄ was used to adjust the pH to 7 (or other designated pH values). After being sealed with a rubber stopper, the flask was sonicated for 1 min to disperse the catalyst particles. The suspension was stirred at 360 rpm and sparged with H₂ (1 atm) at room temperature (around 20 °C) for 1 h to afford the ligand-modified Ru catalyst suspension (e.g., [cis-DACH]­Ru/NC) and allow the reduction of surface oxide on Ru nanoparticles. The H₂ gas supply was achieved by two 16-gauge stainless steel needles (introduced through the stopper), which served as gas inlet and outlet, respectively.

Perchlorate reduction and aqueous sample analysis

The reaction was initiated by adding the NaClO₄ stock solution to the catalyst suspension at 20 °C. Aliquots were collected through the H₂ outlet needle with a 3 mL plastic syringe and immediately filtered through a 0.22-μm cellulose acetate membrane. ClO₄ ^– concentration was measured by ion chromatography (Dionex ICS–5000) equipped with a conductivity detector and an IonPac AS16 column. The column was used at 30 °C, with 65 mM KOH eluent at 1 mL min^–1. The concentration of cis-DACH in aqueous samples was determined by total nitrogen analysis of the solution using a Shimadzu TOC-LCSH system equipped with a total nitrogen measurement unit (TNM-L). The control experiment confirmed that nitrogen leaching from the NC material is negligible.

Catalyst collection and characterization

The freshly prepared catalyst suspension was immediately transferred into an anaerobic glovebag (98% N₂, 2% H₂; Coy Laboratories) to avoid artifacts from air oxidation and then filtered under vacuum. The filter paper coated with the catalyst was transferred into a 20 mL scintillation vial and dried in a 110 °C sand bath for 8 h. The dried catalyst powder was stored in the glovebag before further characterization with X-ray photoelectron spectroscopy (XPS), diffuse reflectance FT-IR spectroscopy (DRIFTS), scanning transmission electron microscopy (STEM), and other techniques. Details of catalyst characterizations are described in Text S3.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscentsci.6c00532.

• Comparison with other reaction systems; full data of perchlorate reduction kinetics under varying configurations (ligand structure, pH, ligand concentration, metal species, perchlorate concentration), ligand immobilization and transformation (¹H NMR), and catalyst characterization (metal dispersion, XRD spectra, and HAADF-STEM-EDX images); Ru colloid formation; H₂ activation tests with the WO₃ indicator; proposed reactor design; texts for catalyst turnover calculation, mass transfer analysis, and the Langmuir–Hinshelwood kinetics modeling (PDF)

J. Gao designed the catalyst, conducted kinetic experiments, analyzed the data, and drafted the manuscript; S.X. and F.L. characterized the catalyst with chemisorption and DRIFTS; Q.L. helped with the H₂ activation test with WO₃; J.C. and J. Guo assisted in the preparation of supported Ru catalysts and XRD characterization; Y.Y. discussed and revised the manuscript; J.L. supervised the research and revised the manuscript.

The authors declare no competing financial interest.