Unified Metallic Catalyst Aging Strategy and Implications for Water Treatment

Abstract

Heterogeneous catalysis holds great promise for oxidizing or reducing a range of pollutants in water. A well-recognized, but understudied, barrier to implementing catalytic treatment centers around fouling or aging over time of the catalyst surfaces. To better understand how to study catalyst fouling or aging, we selected a representative bimetallic catalyst (Pd-In supported on Al₂O₃), which holds promise to reduce nitrate to innocuous nitrogen gas by-products upon hydrogen addition, and six model solutions (deionized water, tap water, sodium hypochlorite, sodium borohydride, acetic acid, and sodium sulfide). Our novel aging experimental apparatus permitted single-passage of each model solution, separately, though a small packed bed reactor containing replicate bimetallic catalyst “beds” that could be sacrificed weekly for off-line characterization to quantify impacts of fouling or aging. The composition of the model solutions led to the following gradual changes in surface composition, morphology, or catalytic reactivity: (i) formation of passivating species, (ii) decreased catalytic sites due to metal leaching under acid conditions or sulfide poisoning, (iii) dissolution and/or transformation of indium, (iv) formation of new catalytic sites by the introduction of an additional metallic element, and (v) oxidative etching. The model solution water chemistry (solutes, Eh, pH) captured a wide range of conditions likely to be encountered in potable or industrial water treatment. Aging-induced changes altered catalytic activity and provided insights into potential strategies to improve long-term catalyst operations for water treatment.

Graphical Abstract

1. Introduction

Aging and fouling of surfaces used in water treatment processes (e.g., membranes, granular media filters, UV quartz sleeves)^1–3, distribution systems, and premise plumbing (e.g., copper piping)^{4, 5} are well-recognized as critically important in influencing operational lifetime and infrastructure integrity and identifying remedial actions (e.g., antiscalants, backwashing, acid cleaning, corrosion inhibitors). Standardized tests or calculations (e.g., silt density index (SDI), modified fouling index (MFI), Langelier index)^6–8 exist to aid in predicting or understanding fouling potential. Unfortunately, similar approaches are lacking in the field of heterogeneous catalysis as applied to municipal or industrial water treatment (i.e., electro-, photo-, hydrogenation-catalysis). The lack of standardized methods and fundamental studies highlights catalyst surface modification as a specific barrier for technology readiness level advancement, which has been already recognized in several editorials.^{9, 10} Note that catalyst aging and surface fouling change surface characteristics, thus can deteriorate catalytic activity and ultimately decrease catalyst lifetime.^9–12

Palladium (Pd) and associated metal-based catalysts are widely studied because they can effectively activate hydrogen (H₂) and initiate chemical reactions that enable reduction of oxyanions (e.g., nitrate, chlorate, and perchlorate)^13–16 or halogenated organics (e.g., trichloroethylene).^{17, 18} For example, Pd-based bimetallic catalysts, such as alumina-supported bimetallic Pd-In (Pd-In/Al₂O₃) or Pd-Cu, have high reactivity with nitrate and selectivity towards innocuous by-products.^19–23 However, changes over operational time (i.e., aging) in oxidation state, surface foulants or poisoning surface species formation, leaching of active components, phase transformation, pore collapse of the supporting materials, and many other factors have not received as much attention as the development of novel catalyst materials themselves.^{9, 10} Understanding water quality factors that influence catalyst aging is essential to realize the potential of heterogeneous catalysts in water treatment.

For this study, we selected a representative bimetallic catalyst (i.e., Pd-In/Al₂O₃) used in hydrogenation processes. Most material discovery studies focus on maximizing reactivity of Pd-based catalysts to degrade target pollutants in simple solution (i.e., deionized water containing the pollutant) using short-term batch systems, and these contrast significantly to the actual intended application in engineered treatment processes where catalysts are immobilized on supporting materials and exposed to heterogeneous water chemistries containing a broad range of cations, anions, pH, and Eh.²⁴ Only a handful of studies investigated effects of solution chemistries on selectivity and regeneration of Pd-In/Al₂O₃ catalysts in continuous flow column apparatus over extended operational periods.^{25, 26} However, structural and compositional changes in catalysts over time in realistic waters impact catalytic activity but have not been elucidated in detail. There is a need for a systematic evaluation of common water treatment chemistries (i.e., pH, Eh, ion pairs, and hardness) on catalyst aging.

This paper develops and applies a methodology to systematically quantify catalyst aging and fouling using six model water chemistries spanning a range of Eh, pH, and ion pairs, aimed at improving our understanding of how to manage catalyst aging in water treatment applications. Water chemistries were selected to represent the following end-member water exposure scenarios for catalysts: (i) deionized water to simulate highly corrosive conditions, (ii) tap water to simulate the realistic working environments, (iii) sodium hypochlorite to simulate common oxidative conditions observed in water treatment applications, (iv) sodium borohydride to simulate reductive conditions present in H₂ catalyst or cathodic electrochemistry, (v) acetic acid to simulate the acidic conditions and common end-of-life leaching test, and (vi) sulfide to simulate sulfur poisoning conditions. As a representative catalyst, Pd-In/Al₂O₃ was loaded in multiple “plugs” in packed-bed columns through which the model test waters passed continuously for four weeks. At weekly time increments, exposed catalyst “plug” samples were sacrificed for material characterization and reactivity testing. Following a quantitative presentation of material characterization transformations, the discussion relates transformations in the chemical state catalyst surfaces with both its catalytic performance and chemical exposure. This study is timely as there is an unmet need for a standardized method to assess the long-term sustainability of catalysts with respect to catalyst properties.

2. Materials and methods

2.1. Catalyst preparation

The Pd-In/Al₂O₃ catalysts were prepared following previously published methods to achieve a Pd:In weight ratio of 10:1, and the surface coverages of In were set to 40%, as was previously optimized for nitrate reduction.¹⁴ The virgin catalyst had a metal loading of 1.0 wt% Pd and 0.1 wt% In, as determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES, Varian, 700-ES, CA, USA). TEM image revealed that the nanoparticles with an average size of 3.74 ± 0.04 nm were distributed on the alumina surface. Energy dispersive X-ray analysis confirmed that the nanoparticles were mainly composed of elemental palladium, as indium was only applied to the surface to create reactive “islands” for nitrate transformation (Figure S1).¹⁴ This virgin bimetallic Pd-In/Al₂O₃ catalyst served as a “baseline” against which aged catalyst characteristics and reactivity were compared.

2.2. Continuous-flow catalyst aging apparatus

Catalyst aging experiments were carried using parallel reactors fabricated from polytetrafluoroethylene (PTFE) tube (ID = 5 mm) containing a packed bed “plug” (1.2 cm in length) of Pd-In/Al₂O₃ catalysts (100 mg). The catalyst bulk density (ρ) and the reactor porosity (η) were calculated to be 0.49 g/mL and 0.56, respectively (see Supporting Information). As shown in Figure S2, The PTFE column packed with Pd-In/Al₂O₃ was connected to a peristaltic pump (Masterflex, Cole-Parmer, USA), and a feed solution was pumped through the column at 0.1 mL/min (residence time of 1.1 min). At fixed intervals (7 days), one “plug” was sacrificed for catalyst analysis and removed from the apparatus. The catalyst was removed from the sacrificed plug, dried under vacuum, and subjected to spectroscopic characterization and quantification of surface reactivity compared against the virgin catalyst (see SI for morphological analysis of the catalysts by transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD) and surface area).

The concentrations of the six aging solutions were strategically designed to represent different water chemistries to which catalysts could be exposed during water treatment (Table S1). Before conducting catalyst aging experiments under oxidative and reductive conditions, the oxidation-reduction potential (ORP) was measured for sodium hypochlorite (NaOCl) and sodium borohydride (NaBH₄) concentrations ranging from 10 to 100 mM (Figure S3). Above 10 mM, NaOCl and NaBH₄ led to similar ORP of +800 mV or −800 mV, respectively. Therefore, 10 mM concentrations of these solutions were used. A 0.l M acetic acid solution with a pH of 2.9 was used to mimic acidic toxicity characteristic leaching procedure (TCLP) extraction conditions. Because natural water may contain total sulfur up to 10⁻⁴ M,¹³ two sulfide concentrations (5 and 50 μM) were applied to the catalysts.

2.3. Catalytic reactivity assay

To enable use of small catalyst masses and avoid working with H₂ gas, the activity of virgin and aged Pd-In/Al₂O₃ catalysts was quantified using a colorimetric probe (methylene blue (MB) and reductant (NaBH₄)).²⁷ While other common catalyst reactivity assays have been reported,^{28, 29} the benefits of this assay include rapid and reproducible quantification using simple UV-vis spectrometry.

3. Results

3.1. Test water matrices age and alter catalytic activity

Figure 1a shows the catalytic assay responses to aging by plotting the absorbance (A⁶⁶⁴) against catalyst mass concentration. Mass-normalized reactivity coefficient (K) is determined by linear regression data fitting, and the reproducibility of the MB-BH₄ assay is within acceptable limits. Higher K indicates greater reactivity of the catalyst with the redox-probe chemicals in the MB-BH₄ assay. A baseline catalytic activity (K´ = 0.054 ± 0.005 L/cm·mg) was assigned for the virgin Pd-In/Al₂O₃ catalysts. The virgin bimetallic catalyst (Pd-In/Al₂O₃) had a 1.8× higher catalytic activity than Al₂O₃ catalyst coated only with Pd (K = 0.03 L/cm·mg). The alumina only (Al₂O₃) catalyst support showed negligible reactivity, indicating that the catalytic activity was primarily due to metals on the alumina.

Figure 1.

(a) Change in colorimetric absorption (A⁶⁶⁴) as a function of catalyst mass in the MB-BH₄ assay for alumina support alone, Pd catalyst (Pd/Al₂O₃), and bimetallic catalyst (Pd-In/Al₂O₃). (b) The mass-normalized reactivity coefficient (K) using the MB-BH₄ assay for the virgin Pd-In/Al₂O₃ (dashed line at K´ = 0.054 L/cm·mg) and after various exposure (1 to 4 weeks) to six aging solutions. The complete set of absorbance versus catalyst mass concentration data is in the supporting information (Figure S4).

Figure 1b shows K in the MB-BH₄ assay after exposure of individual catalyst “plugs” to one of the six aging solutions for durations of 1 to 4 weeks. Aging in the presence of a reductant, DI water, or sulfide resulted in lower K (i.e., less reactivity) relative to virgin Pd-In/Al₂O₃. Surprisingly, exposure to an oxidant, tap water, and acetic acid led to higher K (i.e., more reactivity) relative to the virgin Pd-In/Al₂O₃. For catalyst “plugs” exposed for different durations, there was a consistent change over time in catalyst reactivity. Overall, K varied from >95% reduction to >200% increase in the different matrices. Reasons for these changes in catalytic activity are discussed in subsequent sections where reactivity is related to spectroscopic measurements.

3.2. Aging solutions transform physical and chemical catalyst properties

3.2.1. Changes in surface speciation

Binding energies (BEs) of Pd 3d and In 3d spectra were analyzed by XPS to determine the oxidation state and bonding environment of metals after aging. The virgin catalyst exhibited two peaks at BE of 335.4 and 340.8 eV, which correspond to the Pd 3d_5/2 and Pd 3d_3/2, respectively (Figure S5a). In general, the BE of Pd 3d_5/2 peak varies from 335.0 to 335.3 eV for metallic Pd, 336.2 to 336.6 eV for PdO, and 337.2 to 338.0 eV for PdO₂.^{30, 31} Accordingly, the BE of 335.4 eV in the Pd 3d_5/2 spectrum of the virgin catalyst was assigned to slightly oxidized metallic Pd to the air during handling. The BE of 444.5 eV in the In 3d_5/2 spectrum of the virgin catalyst was assigned to In₂O₃ (Figure S5b). The BEs of In 3d_5/2 in the range of 444.5 to 444.9 eV were generally reported in the literature for In₂O₃, while metallic In range was reported at 444.0 to 444.3 eV.³²

Figure 2 summarizes the changes in BEs of Pd 3d_5/2 and In 3d_5/2 peaks across the six water matrices and weekly exposure samples in the aging tests. In this format, surface catalyst aging trends become apparent. The BE of the Pd 3d_5/2 increased consistently with exposure to DI water. The components at higher BEs (336.2–336.8 eV) were associated with oxidized Pd species. Similarly, the BE of the In 3d_5/2 peak increased, indicating that In was also oxidized in the presence of DI water. After 4 weeks, the BE of the In 3d_5/2 peak shifted from 444.4 eV to 444.9 eV, which was attributed to the In₂O₃.³² In contrast with DI water, the BE of the Pd 3d_5/2 and In 3d_5/2 peak shift decreased after exposure to tap water, indicating that certain components in tap water affected the electronic properties of the Pd and In.

Figure 2.

Representative XPS BEs of the (c) Pd 3d_5/2 and (d) In 3d_5/2 of aged Pd-In/Al₂O₃ catalyst (4 weeks exposure). Asterisks indicate the BE for PdS and In₂S₃ standard.³³

The most significant positive peak shift for Pd 3d_5/2 and In 3d_5/2 toward higher BE was observed upon exposure to the NaOCl solution, indicating oxidation of Pd and In. After 4 weeks of exposure to NaOCl, the BE of the Pd 3d_5/2 increased to 337.8 eV, suggesting formation of highly oxide species such as PdO₂. The In 3d_5/2 BE of 445.2 eV obtained after the oxidation treatments are indicative of In(OH)₃ formation. The BEs of the Pd 3d_5/2 and In 3d_5/2 decreased after prolonged exposure to the strong reductant solution. The BE of the In 3d_5/2 peak shifted to lower BE sides at 444.1 eV, which is close to the value of metallic In. Reduction also occurred upon exposure to acetic acid. Under acetic acid condition, as indicated from the Pd 3d_5/2 BE of aged catalyst, the oxidation state of Pd slightly decreased. However, In peaks were not detected, suggesting that In concentration went below the detection range of XPS presumably due to loss of In caused by acid-leaching. After 4 weeks of exposure to acetic acid, the BE of the Pd 3d_5/2 peak shifted to lower BE sides at 335.1 eV, which was similar to the value of metallic Pd, implying the Pd oxide layer was removed by leaching and the Pd surface reverted to the metallic state. After aging under sulfide condition, the BE of Pd 3d_5/2 and In 3d_5/2 shifted toward higher energy, which indicated the formation of metal-sulfur bonds.³⁴ Moreover, there was a large BE shift for Pd and In in high-sulfide-fouled catalysts compared with low-sulfide-fouled catalyst.

3.2.2. Changes in solution chemistry passing through aging system

The aging solution pH after passing through the column was monitored (Table S2), and the pH change indicated catalyst reactions with feed solution components. The susceptibility of the catalysts to leaching is a crucial issue because leaching of active metal causes catalyst deactivation, which threatens the process sustainability. Therefore, the stability of the Pd-In/Al₂O₃ catalyst was assessed by determining the amount of Pd, In, and Al leached from the catalyst during the aging process. Figure 3a shows the amount of leached Pd and In after various aging processes. The high standard deviations for In are a consequence of the very low concentrations in the ICP-OES. Except for the acetic acid condition, Pd leaching was negligible (<0.4%) during the aging processes (Figure 3a), indicating that the change in catalytic activity by Pd leaching can be ignored. For most of aging solutions, In was preferentially dissolved. This was expected as the standard redox potential of In³⁺/In (E° = −0.338 V) is much lower than Pd²⁺/Pd (E° = +0.915 V), implying that the oxidation of In and the reduction of Pd²⁺ is thermodynamically favorable. The acetic acid solution accelerated the corrosion of the Pd and In through the dissolution of the metal oxide. Pd and In concentrations in the column effluents decreased with the aging time, indicating that considerable metal dissolution occurred during the initial stage of acid aging. The total acid-leached Pd and In after acetic acid treatment was 5.1% and 40% of the initial load. For some test conditions, dissolution also damaged the alumina support and as such is important to consider if such supports are intended for used in full scale systems. The alumina support was stable without any dissolution in DI water. However, slight dissolution of the alumina (<1%) was observed in the NaOCl, NaBH₄, and acetic acid conditions (Figure 3b). This agrees with previous results that dissolved aluminium ions increased above pH 4 and above 9.³⁵

Figure 3.

Amount of leached (a) Pd and In, or (b) aluminum in effluent after various aging processes.

3.2.3. Morphological and textural changes in catalysts

The physical structural characteristics of Pd-In/Al₂O₃ changed upon exposure to some aging solutions. Figure 4a shows virgin Pd-In/Al₂O₃ catalyst as Pd nanoparticles uniformly distributed on the alumina surface. Figures 4b–e show that after 4 weeks of exposure in DI water, tap water, acetic acid, NaOCl, and sulfide solution, there were no apparent structural changes in size and morphology of Pd nanoparticles. Pd nanoparticles remained distributed on the alumina, with no remarkable agglomerations. However, the NaBH₄ condition caused major structural changes, including collapse and decomposition of the alumina support into many small debris particles (Figure 4f). The initial pH of the NaBH₄ solution was higher (pH = 9.6) than the DI water. This likely occurred due to hydrogen production as BH₄^– passed over the metallic catalysts. Dissolution of the alumina support likely occurred because the increase in alumina solubility led to formation of soluble products such as Al(OH)₄^– and Al³⁺. In general, Al³⁺ exists below pH 4, Al(OH)₂⁺ exists between pH 4 and pH 6, and Al(OH)₄^– is predominant above pH 8. Partial collapse of the alumina support was also observed upon exposure to the NaOCl and acetic acid solutions due to initial pH of 10.4 and 2.9, respectively. These results agree with the measured aluminum leaching in these water matrices (Figure 3b). Alumina dissolution accelerated with increasing pH in the same feed solution (Figure S7). Therefore, the catalyst support degradation was not due to the reducing environment, but it was instead due to the elevated pH where Al₂O₃ dissolved and released Al(OH)₄^– into the solution. The structure of Pd-In/Al₂O₃ did not change significantly with sulfide concentration (Figure 4g).

Figure 4.

(a) TEM image of a virgin Pd-In/Al₂O₃ catalyst. The inset of (a) shows the size distribution of Pd nanoparticles. TEM images of Pd-In/Al₂O₃ catalysts after 4-week aging process. (b) DI water, (c) tap water, (d) acetic acid, (e) NaOCl, (f) NaBH₄, and (g) sulfide (50 μM).

The surface area, pore volume, and pore size of virgin and aged catalysts are summarized in Table 1. The virgin catalyst had a surface area of 142 m²/g and a pore volume of 0.56 cm³/g. The average pore size of the Al₂O₃ was larger than 12 nm, which was larger than the Pd-In nanoparticles (3.74 nm), suggesting that the Pd and In nanoparticles can not only attach to the outer surface of the alumna, but can also be loaded into the pore channels. The surface area and pore volume were similar for the virgin and aged catalyst after exposure to DI water and tap water. However, porosity increased after exposure to other aging processes, as shown by increased surface area and pore volume. The increase in porosity indicated that part of the alumina support dissolved and some of the loaded metal nanoparticles occupying the pores were removed during the aging processes. Taken together with the TEM image (Figure 4f), the largest surface area and pore volume for NaBH₄-treated catalysts were attributed to the increased surface area caused by support collapse and pore exposure. To examine whether the surface area changes were caused by structural changes of alumina, we compared samples to the XRD pattern of the bare alumina support. The virgin and aged catalysts exhibited similar γ-Al₂O₃ diffraction patterns, indicating no change in the crystalline structure of the supports (Figure S8). Moreover, the pore size was not influenced by the different aging processes (Table 1), demonstrating that neither the crystallinity nor the pore size of the alumina were altered after aging processes.

Table 1.

Summary of changes in aged Pd-In/Al₂O₃ catalyst after 4 weeks

Aging solution	Textural properties			K value (L/cm·mg)	Oxidation state		Structural features	Metal leaching
	Surface area (m2/g)	Pore volume (cm3/g)	Pore size (nm)		Pd 3d5/2 BE (eV)	In 3d5/2 BE (eV)		Pd (%)	In (%)	Al (%)
DI water	150	0.62	14.8	0.017	336.79	444.89	-	0.3	17	0.02
Tap water	156	0.51	12.0	0.0972	334.93	443.73	-	0.2	18	0.11
10 mM NaOCl	235	0.92	13.6	0.1081	337.78	445.24	Pd nanoparticle shape change	N.D	36	0.72
10 mM NaBH4	382	1.56	14.2	0.0319	335.34	444.37	Support collapse	0.3	21	2.01
0.1M acetic acid	234	0.97	13.7	0.0701	335.09	N.D.	-	5.1	39	1.36
5 μM Na2S	198	0.77	13.4	0.0139	335.87	444.52	-	0.2	34	0.005
50 μM Na2S	209	0.81	13.5	0	336.25	444.91	-	0.4	40	0.006

4. Discussion on impacts of each aging solution

4.1. Exposure to DI water

Long-term exposure of the catalyst to DI water caused deactivation due to metal oxidation, which was similar with previous studies.³⁶ Thus, even mild oxidation of the catalysts can occur, thereby decreasing catalytic activity. In reactors using H₂, this suggests that intermittent interruption of H₂ gas feed or longer-duration process shut-downs could result in gradual oxidation of the catalysts. Thus, these changes should be expected to occur during continuous operation, and troubleshooting strategies should be developed to re-reduce the surface species. Such strategies could include re-pressurizing with H₂ prior to starting flow through the reactor or adding a chemical reductant (e.g., borohydride).

4.2. Exposure to tap water

The original intent of the tap water aging tests was to understand potential aging or fouling in a real water matrix representative of household point-of-use systems. Surprisingly, K increased for catalysts exposed to tap water. While the municipal tap water met all EPA regulations, a visual color change over time in the catalyst “plug” was observed. Specifically, it turned “green-bluish” in color, which suggested copper oxide solids may have formed on the surface. To support the visual observation, XPS spectra of the Cu 2p region of the aged catalysts were measured. As expected, no copper species were observed on the surface of the virgin catalyst. However, XPS of the catalysts exposed to tap water showed two main peaks centered at 933.9 (Cu 2p_3/2) and 953.8 eV (Cu 2p_1/2). These peaks were attributed to the presence of the Cu(II) state and indicated formation of CuO and/or Cu(OH)₂ on the catalyst surface (Figure S10). Moreover, the shake-up satellite peaks of the Cu 2p_3/2 and Cu 2p_1/2 at 942.3 and 962.5 eV confirmed the formation of Cu(II) on the surface.³⁷ As shown in Figure 2, there was a shift in the BE of Pd 3d_5/2 by ~0.5 eV toward lower BE in the tap-water-aged catalyst compared to the virgin catalyst. The negative shift in the BE of Pd 3d_5/2 with Cu content suggests a change in the electron density around Pd upon alloying with Cu. Noborikawa et al. reported that transition metals, including copper, could induce significant strain and charge transfer leading to an electronic effect on Pd.³⁸ Similar observations of negative BE shift for Pd with Cu addition has been reported previously.^39–41

Further, Cu concentration in tap water and effluent during the aging processes was measured. The effluent Cu decreased to 0 mg/L after 1-week exposure to tap water, indicating that Cu was deposited onto the catalyst. After 2 weeks, the effluent Cu increased to 0.3 mg/L, which was still lower than the Cu in tap water, indicating that the Cu deposition process lasted more than 2 weeks. After 3 weeks, the copper concentration recovered to the initial concentration, indicating that no more copper deposition occurred (Figure S11). We conducted an additional 1-week study using the tap water with a γ-alumina only. Visual observation of the “plug” showed that the color in the upper part of the γ-alumina started to change from white to green as the aging process progressed (Figure S12a). This confirmed that copper ions accumulation was independent of any reaction involving Pd or In. When the alumina support (without Pd and In) was subjected to the colorimetric reactivity assay, that tap-water-aged alumina with copper on the surface showed a high K compared with no reactivity from virgin alumina (Figure S12b). Therefore, it was proposed that surface-deposited copper which primarily came from plumbing pipes, acted as a catalyst and improved catalytic activity. This was an interesting and unexpected observation. In fact, these tests could suggest that modest copper addition over extended periods could be used to maintain or increase bimetallic catalyst reactivity.

4.3. Exposure to chemical oxidant (NaOCl) or reductant (NaBH₄) aging solutions

XPS analysis (Figures S9) showed that metal oxidation occurred in the presence of NaOCl, which should have decreased catalytic activity. However, NaOCl increased K relative to the virgin catalyst. Several studies have used unbuffered NaOCl solutions to regenerate sulfur-fouled Pd catalysts in packed-bed columns.^{36, 42} As oxidative regeneration with hypochlorite is effective for regeneration of catalysts deactivated by organic matter as well as sulfide, NaOCl treatment might recover the original catalytic activity by eliminating organic substances formed on the surface during the catalyst synthesis process or carbonaceous deposits formed during handling, which is presumed to lead to an increased catalytic activity in the assay system. Moreover, morphological changes in Pd nanoparticles were observed after the NaOCl oxidation treatment. The TEM micrograph shows that the majority of the Pd nanoparticles were spherical in shape. However, some of the Pd nanoparticles exhibited a hexagonal structure, consistent with cubo-octahedra (Figure S13),⁴³ indicating that NaOCl played a crucial role in the oxidative etching of Pd nanoparticles. We believe that during the catalyst activity assays, which contain the colorimetric dye and reductant, the borohydride re-reduced the oxidation imposed by NaOCl. In some catalyst applications (e.g., those involving electron donors such as hydrogen), periodic use of NaOCl to mitigate biofilm ground is required to allow solutes to contact catalyst surfaces. However, it may be necessary to “repair” detrimental impacts of NaOCl on the catalyst itself, and the above results suggest that addition of borohydride could be one path to “reactivate” the Pd-In nanoparticles.

Exposure to the reductant aging solution was accompanied by a gradual loss in catalytic activity over 4 weeks for the bimetallic catalysis. This was unexpected because XPS showed a decrease in the oxidation state of both Pd and In on the surface (Figure 2), which should improve catalytic activity. Based on TEM observations, the collapse of alumina and generation of fines were considered to be the main reasons for catalyst loss and resulting decrease in catalytic activity. Similar phenomena were observed in a catalyst where mesoporous Santa Barbara Amorphous-15 (SBA-15) was used as a support.^{44, 45} In the SBA-15-supported Pd catalyst, Pd nanoparticles on the interior render the catalyst inactive once the supporting material degrade because the collapse of silica mesopores blocks the access of reactants to Pd nanoparticles. The results suggest that structural collapse due to aluminum dissolution during the reaction and the resulting change in catalytic activity should be considered.

4.5. Exposure to acetic acid aging solution

Acetic acid is capable of increasing the corrosion rate by decreasing the solution pH and soluble metal in the oxide layer, thus reducing the oxide layer thickness, so it is commonly used to remove hardness scales from water treatment devices. After acetic acid treatment, the Pd oxide layer was removed, while In was preferentially leached by acetic acid.⁴⁶ Because acetic acid treatment did not cause serious structural changes of the catalyst, high catalytic activity despite In loss due to leaching was mainly attributed to increased surface exposure to metallic Pd, suggesting that (i) metallic Pd had higher catalytic activity than Pd oxide in our colorimetric assay, and (ii) the reactivity enhancement due to the formation of the metallic Pd overwhelmed the decrease in reactivity due to In leaching. Similarly, our previous study revealed that Pd nanoparticles show higher catalytic activity than other noble metal nanoparticles such as Au, Ag, and Pt in the assay system and that metal oxides have lower surface catalytic activity because of their lower electron mobility than pure metals.²⁷ However, in the case of a bimetallic catalyst system where the ratio of noble metal to catalyst-surface promoter is important, the catalytic activity (such as reaction rate) might be reduced after an acid aging process by changing surface atomic ratio of metal.¹⁹ Thus, if the bimetallic composition is required to achieve selective pollutant reduction (e.g., nitrate to N-gases¹⁴), periodic cleaning of catalysts with acetic acid is not recommended.

4.6. Exposure to sulfide in aging solutions

The accumulation of sulfur species on the catalyst surface affected the deactivation kinetics (Figure 1b). In the low sulfide-fouled catalyst, a broad S 2p peak at 162.5 eV was observed, which can be attributed to sulfur atoms bound to the Pd surface, while a low-intensity broad peak at around 168 eV was assigned to oxidized sulfur species (coming from the exposure to air during storage) (Figure S14a).⁴⁷ However, in the high sulfide-fouled catalyst, no distinct peak at 168 eV was visible due to strong intensity of the metal-sulfur bonds peak (Figure S14b), indicating that more sulfur was adsorbed onto the metal surface when the sample was treated with high concentration of Na₂S, forming the metal-sulfur bonds to fully deactivate the catalyst. The formation of metal sulfide was further confirmed by the BE shift of the Pd 3d_5/2 and In 3d_5/2 peak in the XPS measurement (Figure 2). In the case of the high-sulfide-fouled catalyst, the BE of Pd was much lower than literature BE for PdS (337.2 eV). However, the BE of In was similar to the literature BE for In₂S₃ (445.4 eV) after the 4-week sulfide aging process,^{33, 48} suggesting that sulfur complexed more of the In sites than the Pd sites. This observation is consistent with standard electrode potentials that indicate In₂S₃ formation is more thermodynamically-favorable than PdS.³³ The BE of sulfur (162.5 eV for S 2p) and indium (445.4 eV for In 3d_5/2 and 452.3 eV for In 3d_3/2) in the sulfide-fouled catalyst indicate formation of In₂S₃.⁴⁹

Clearly, treatment of groundwaters containing sulfides should be avoided. Strategies to minimize in-situ sulfide production in catalytic reactors should be also be avoided. Sulfides could be produced by sulfide reducing bacteria, which potentially thrive in hydrogen-rich environment that contain high sulfate such as in spent nitrate ion exchange (IX) brines where catalytic reduction of nitrate is desirable. Sulfate accumulates in sodium chloride IX brines because of limited selectivity of the resin for nitrate over sulfate by regenerable IX resins.

5. Environmental implications

This study addressed a critical need for a standardized methodology to characterize stability (aging and fouling) of catalysts in water matrices typically encountered during water treatment. While some aging solutions produced expected findings (e.g., higher sulfide-containing solution led to more rapid and greater reduction in catalytic activity), other findings were more complex (e.g., NaOCl solution oxidized surface species but increased catalytic activity) and sometimes unexpected and surprising (e.g., tap water containing low levels of copper increased catalytic activity). It was essential to perform XPS on aged catalyst “plugs” in order to understand the mechanism behind changes in catalytic activity. Furthermore, without TEM images comparing virgin against aged catalysts, neither transformation in Pd nanoparticle shape nor impact on the catalyst support would have been readily noticeable. The selected aging solutions could be further optimized, but they generally reflected the breadth of expected pH, Eh, and ion pair composition exposure environments for catalysts during potable water or industrial wastewater applications. These studies were intended to be screening experiments. The 4-week duration appeared to be essential for tracking relatively slow transformations in catalyst surface reactivity. Collecting samples at weekly intervals (Figure 2) provided valuable insights into these changes.

Using real tap water, rather than synthetic waters, proved interesting, and we may have serendipitously identified a new way to renew aged catalysts. Our results showed metal ion (i.e., copper) in the feed solution deposited on the alumina support or acid treatment could improve catalytic activity, suggesting that new catalyst development through alumina-metal interaction and pretreatment/regeneration of oxidized catalyst through acid treatment are both feasible.

It was not easy to separate the mechanisms that caused these catalytic activity changes. In most cases, the change in the catalytic activity was the result of more than one cause, even having the same effect. However, the study of the oxidation state of metals, crystalline phases, surface area, and other properties provided valuable insights of the possible deactivation phenomena occurring during the aging process. In the reaction based on electron transfer, the catalytic activity mainly depended on the oxidation state of the metals, but exposure to various aging solutions also led to catalyst deactivation. In particular, losing the uniform dispersion of active metal nanoparticles on the supporting material by pH-induced collapse of alumina significantly affected the catalytic activity. Here the support structure is alumina oxide, but other systems may use membranes, activated carbon, titania, or other materials. Thus, our observations reinforce the need to consider water chemistry impacts not only on the active catalyst but also on the support structure required to enable integration of the nanoparticles into macroscale water treatment systems.

Understanding the complex deactivation mechanisms, as well as understanding the tolerance of catalysts toward the poisoning species in the feed, is important to the design and optimization of durable catalysts and to more effective strategies to prolong catalyst lifetime. Furthermore, while this study evaluated Pd-In/Al₂O₃ catalysts, we believe the selection of aging solutions and testing apparatus/methodology is also suitable for hydrogenation catalysts, photocatalysts, and electrocatalysts. The approach makes it possible to evaluate catalyst tolerance and to predict catalyst lifetime in an industrial process based on the time required for catalyst deactivation under various reaction conditions. An advantage this is that the approach can be implemented early in the material design timeline, before evaluation in application-specific waters.