Oxidized contaminants, such as haloacetic acid, chlorophenol, bromate, chlorate, nitrate and perchlorate, commonly appear in drinking water following chlorine or ozone disinfection, polluted surface water/groundwater and industrial wastewater. Most of the oxidized contaminants possess carcinogenic, teratogenic and mutagenic properties. Once the concentration of the oxidized contaminants exceeds a certain limit, they will cause serious damage to the ecological environment and human health1,2,3,4,5. Electrocatalytic hydrogenation (ECH) can transform all kinds of oxidized contaminants into low-toxic, even non-toxic species or industrial raw materials at room temperature and atmospheric pressure using environmentally benign water as the hydrogen donor, which has aroused great interest in the field of water pollutant control in recent years6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33.

ECH in a conventional electrochemical reactor (Fig. 1a) requires a reaction medium with good conductivity, and therefore the reaction system must contain a certain concentration of supporting electrolyte. The coexistence of supporting electrolyte and oxidized contaminants in water greatly limits the competitiveness and range of applications of ECH. For example, it is difficult to remove oxidized contaminants from drinking or surface water by ECH owing to the very poor conductivity of the water34,35. Additionally, in the treatment of oxidized contaminants in industrial wastewater containing only a small amount of supporting electrolyte, the added supporting electrolyte will not only increase the cost of treatment, but also unavoidably lead to secondary pollution. To mitigate these problems, it is common to reduce the amount of added supporting electrolyte, but this leads to an increase in the energy consumption of the ECH process (due to the increased cell voltage). Another issue that restricts the practical application of ECH for the treatment of oxidized contaminants is the production of a high proportion of unsafe intermediates. For example, ECH of trichloroacetic acid (TCAA) produces a high proportion of monochloroacetic acid (MCAA; >20%)21,29,30, which has a toxicity (dose lethal to 50% of animals tested (LD50) = 76 mg per kg body weight (hereafter given as mg kg−1)) 43 times higher than that of TCAA (LD50 = 3,300 mg kg−1)36. The main product from the ECH of various chlorophenols is phenol22,23, which is also carcinogenic and highly toxic (LD50 = 317 mg kg−1).

Fig. 1: Comparison of two electrochemical reactor configurations.
figure 1

a,b, Conventional electrochemical reactor (a) and the Rh NPs-PdMER reactor (b).

To overcome these limitations, we have designed a rhodium nanoparticle-modified palladium membrane electrochemical reactor (Rh NPs-PdMER; Fig. 1b and Supplementary Fig. 1) to treat oxidized contaminants in drinking water and industrial wastewater. In a conventional electrochemical reactor (Fig. 1a), hydrogenation occurs in the cathode compartment, and it is impossible to avoid the coexistence of supporting electrolyte and reactants in the same system. Unlike conventional electrochemical reactors, hydrogenation in a Rh NPs-PdMER takes place in a third compartment (the chemical compartment). The chemical compartment and the cathode compartment are separated by a dense Pd membrane, which also acts as the cathode. Strictly speaking, the hydrogenation occurs on Rh nanoparticles (NPs) on the backside of the Pd cathode. The hydrogen atoms generated in the reduction of protons or water are adsorbed on the Pd membrane surface in the cathode compartment and can be transferred to the chemical compartment through the Pd membrane37,38,39,40 where they react with the oxidized contaminants in water without the need for a supporting electrolyte.

The catalytic material used in ECH includes Pd (refs. 8, 10, 12, 17,21,22,23,24), Pt (ref. 25), Rh (refs. 16, 25,26), Ru (refs. 9,27), Ag (refs. 14, 15,28), Fe (ref. 6), Co (ref. 7), Ni (refs. 18,29), Cu (refs. 13, 14,30), Pd–In (ref. 19), Pd–Ni (ref. 31), Pd–Fe (ref. 32), Pd–Rh (ref. 33) and carbon (ref. 11). Of these, Pd has attracted the most attention8,10,12,17,19,20,21,22,23,24,31,32,33. Pd not only shows exceptionally high catalytic activity in general, but it can also catalyse the hydrogenation of most oxidized contaminants in water20. Under an applied cathodic potential, protons or water molecules are reduced on Pd, generating large numbers of adsorbed hydrogen atoms (Hads) on the Pd surface. These Hads are very reductive and can react quickly with most oxidized contaminants. However, surprisingly, Hads on most Pd catalysts show very low or almost no activity for the further hydrogenation of some hydrogenated intermediates, such as MCAA21,32 and phenol22,23. Compared with Pd, hydrogenated intermediates are adsorbed much more strongly on Rh, which can hydrogenate fluorophenol to cyclohexanone and cyclohexanol26. In this work, a Pd membrane was modified with Rh NPs to construct a Rh NPs-PdMER. The Rh NPs-PdMER enabled the electrochemical hydrogenation of 12 different oxidized contaminants to safe products with ≥99% conversion and ≥95% yield at a very low operating voltage of only 1.4 V in the absence of a supporting electrolyte. The excellent performance of the Rh NPs-PdMER has been attributed to the ingenious configuration of the reactor and the unique adsorbability of the hydrogenated intermediates of oxidized contaminants on the Rh NPs.


Hydrogenation performance in different palladium membrane electrochemical reactors

Chloroacetic acids are common disinfection byproducts (oxidized contaminants) in drinking water, and their concentrations have to be strictly controlled16. In this study, chloroacetic acids were chosen as representative oxidized contaminants to investigate the influence of palladium membrane electrochemical reactor (PdMER) configuration on the performance of electrochemical hydrogenation in the absence of supporting electrolyte. As shown in Fig. 2a,b, TCAA can be hydrogenated in a PdMER using an unmodified Pd membrane, and the main hydrogenation products are dichloroacetic acid (DCAA), MCAA and acetic acid (AA). Modifying the Pd membrane with Pd nanoparticles (Pd NPs-PdMER) greatly increased the rate of hydrogenation (Fig. 2c,d and Supplementary Fig. 2a), but a high proportion of MCAA was still present at the end of the reaction, consistent with the results of the Pd-catalysed electrochemical hydrogenation of TCAA reported in the literature21,32,41. However, if the Pd membrane was modified with Rh NPs (Fig. 2e,f and Supplementary Fig. 2b), although the rate of hydrogenation of TCAA was slightly lower than that in the Pd NPs-PdMER, almost all the TCAA was fully converted into AA by the end of the reaction.

Fig. 2: TCAA hydrogenation performance in different PdMERs.
figure 2

TCAA hydrogenation was explored in PdMERs with different configurations at 25 °C in pure water (pH 7) at a current density of 3.4 mA cm−2. a,c,e, Configurations of the PdMERs prepared with different membranes: an unmodified Pd membrane (a), a Pd membrane modified with Pd NPs (c) and a Pd membrane modified with Rh NPs (e). Scale bars, 5 μm. b,d,f, TCAA hydrogenation performance in PdMERs prepared with an unmodified Pd membrane (b), a Pd membrane modified with Pd NPs (d) and a Pd membrane modified with Rh NPs (f). The solution (17 ml) in the chemical compartment was vigorously stirred. The projected area of the working electrode was 1.76 cm2. The graphs in Fig. 2b,d,f are shown with error bars in Supplementary Fig. 3a–c.

Source data

To further illustrate the unique hydrogenation performance in the Rh NPs-PdMER, we replaced TCAA with DCAA and MCAA as reactants: the results showed that both DCAA and MCAA could be rapidly and almost completely hydrogenated to AA in the Rh NPs-PdMER (Supplementary Fig. 4a,b), and the rates of hydrogenation of DCAA and MCAA were only slightly lower than that of TCAA (Fig. 2f). In the Pd NPs-PdMER, however, the rate of hydrogenation of DCAA was much lower than that of TCAA, and a high proportion of MCAA was formed at the end of the reaction (Supplementary Fig. 4c). Additionally, the hydrogenation of MCAA was almost impossible in the Pd NPs-PdMER (Supplementary Fig. 4d). It should be noted here that the complete hydrogenation of low concentrations of MCAA to AA by electrochemical methods is very challenging21,29,30,32,41,42.

Effect of operating parameters

We then investigated the effect of operating parameters, including the current density, chemical compartment pH, TCAA concentration, type of electrolyte and absence/presence of ion membrane, on the hydrogenation performance in the optimal Rh NPs-PdMER reactor. As summarized in Table 1, at a current density of 1.2 mA cm2, the conversion of TCAA after 1 h of hydrogenation reached 44.5%, but the system still contained small amounts of DCAA and MCAA (entry 1). The hydrogenation of TCAA was greatly accelerated by increasing the current density, with over 96% of TCAA being hydrogenated to AA at current densities of 3.4 and 6.0 mA cm2 (entries 2 and 3). Alkaline conditions not only inhibited the hydrogenation reaction, but also caused the production of a high proportion of DCAA (entry 5). In contrast, the hydrogenation of TCAA proceeded quickly under weakly acidic or neutral conditions with AA as the main product (entries 2 and 4). Even at industrially relevant concentrations, TCAA could still be rapidly and selectively hydrogenated to AA (entry 6), which means that the Rh NPs-PdMER is suitable for the hydrogenative removal of oxidized contaminants from industrial effluents. At very low concentrations of TCAA, the selectivity for AA decreased (entries 7 and 8); however, it should be pointed out that after the hydrogenation reaction, the total chloroacetic acid concentration was still less than the maximum contaminant level (60 μg l−1) set by the US Environmental Protection Agency (US EPA)3. Rapid and highly selective hydrogenation of TCAA to AA could also be achieved using alkaline electrolyte in place of sulfuric acid in the catholyte and anolyte, and without an ion membrane (entries 9 and 10). These results indicate that the Rh NPs-PdMER can be operated using a very wide range of parameters. It should be emphasized that the possibility of removing the ion membrane as well as using alkaline electrolyte in the Rh NPs-PdMER could greatly simplify the reactor structure and thus reduce reactor cost, which is crucial for practical applications. Additionally, the Pd membrane of the Rh NPs-PdMER can also effectively prevent various molecules and ions from transferring from the chemical to the cathode compartment, and vice versa, when H2SO4 or NaOH aqueous solution is used as the catholyte (Supplementary Fig. 5), which not only ensures the long-term stability of the electrolyte composition, but also avoids secondary pollution of the water in the chemical compartment by the catholyte.

Table 1 Effect of operating parameters on the hydrogenation of TCAA in the Rh NPs-PdMER at 25 °C

Performance comparison with a conventional electrochemical reactor

Using the Rh NP-modified Pd membrane as the cathode, we further compared the reaction efficiency, product distribution and energy consumption of TCAA hydrogenation in the Rh NPs-PdMER and a conventional electrochemical reactor (the distances between the cathode, ion membrane and anode in the two reactors were kept the same). As shown in Fig. 3, the rate of hydrogenation of TCAA was very similar in the two reactors. However, the concentration of MCAA in the Rh NPs-PdMER was lower than that in the conventional electrochemical reactor during and at the end of the reaction. It should be noted that even in the conventional electrochemical reactor, the proportion of generated MCAA was much lower than that reported in the literature21,29,30, which is a result of the unique catalytic performance of the Rh NPs in the hydrogenation of chloroacetic acids. Furthermore, compared with the conventional electrochemical reactor, the cathode potential of the Rh NPs-PdMER shifted positively by about 400 mV, while the cell voltage and energy consumption decreased by about 1.5 V and 40%, respectively (owing to the high concentration of H+ in the catholyte of the Rh NPs-PdMER, which not only can reduce the overpotential of the electrochemical reduction of H+ to Hads, but also enhance the conductivity of the electrolyte), which is of great practical importance.

Fig. 3: Comparison of the TCAA hydrogenation performance in different reactors.
figure 3

af, Comparison of TCAA hydrogenation performance in a conventional electrochemical reactor with a Pd membrane modified with Rh NPs as the cathode (ac) and the Rh NPs-PdMER (df): configuration of the reactor (a,d), TCAA hydrogenation performance (b,e) and electrochemical properties (c,f). Unless otherwise noted, the reaction conditions were the same as those in entry 2 of Table 1. The solution in the cathode compartment of the conventional electrochemical reactor was vigorously stirred. SCE, standard calomel electrode. The graphs in Fig. 3b,e are shown with error bars in Supplementary Fig. 3c,d.

Source data

Universality and practicability of the Rh NPs-PdMER

To investigate the universality and practicability of the Rh NPs-PdMER for removing oxidized contaminants from water, we examined the reduction of 15 common oxidized contaminants in this reactor. As shown in Table 2, with the exception of perchlorate (entry 8) and nitrate (entry 11), the oxidized contaminants experienced a conversion of more than 99% within 120 min, with ten of the contaminants reaching more than 99.9% conversion. It should be pointed out that the main hydrogenation products of six different haloacetic acids (entries 1–6), chlorate (entry 7) and bromate (entry 9) were the almost non-toxic AA or halide ions, with yields of more than 95%. The main hydrogenation product of chlorinated phenols, even at an industrially relevant concentration (entries 12–16), was not phenol (the main hydrogenated product of chlorophenols in palladium-catalysed electrochemical reactions22,23,24), but instead cyclohexanol, which is not carcinogenic and is much less toxic than phenol (by a factor of 6.5)43. Similarly to the three chloroacetic acids, the hydrogenation of three bromoacetic acids (Supplementary Fig. 6), chlorate (Supplementary Fig. 7a), bromate (Supplementary Fig. 7b) and four chlorophenols (Supplementary Fig. 8) also produced very low amounts of intermediates. The practicability of the Rh NPs-PdMER for removing chloroacetic acid contaminants from real tap water was investigated in both batch and continuous processes. In the batch process (Fig. 4a and Supplementary Fig. 1), the hydrogenation of TCAA dominated. After reaction for 120 min, the total concentration of the three chloroacetic acids (TCAA, DCAA and MCAA) had been reduced to about half of the maximum contaminant level (60 μg l−1) set by the US EPA3. In the continuous process (Fig. 4b,c and Supplementary Figs. 9 and 10), the concentration of TCAA was reduced from 120 μg l−1 at the inlet of the reactor to about 20 μg l−1 at the outlet of the reactor with a very low operating voltage of approximately 1.4 V during 80 h of operation. The concentrations of DCAA and MCAA at the outlet of the reactor were as low as <5 μg l−1, such that the total concentration of all chloroacetic acids (CAAs) at the outlet of the reactor was well below the maximum contaminant level (60 μg l−1) set by the US EPA.

Table 2 Universality of the Rh NPs-PdMER for the hydrogenation of oxidized contaminants in pure water
Fig. 4: Hydrogenation performance of CAAs in real tap water.
figure 4

a, Hydrogenation performance in a batch Rh NPs-PdMER with a current of 6 mA. The total concentration of CAAs was 320 μg l−1. Unless otherwise noted, all reaction conditions were the same as those in entry 2 of Table 1. b, Hydrogenation performance in a continuous Rh NPs-PdMER (without an ion membrane) with a current of 4 mA. Aqueous 0.5 M H2SO4 was used as the electrolyte solution. The flow rate was set to 10 cm h−1. The concentration of TCAA at the inlet of the reactor was 120 μg l−1. c, Architecture of the continuous Rh NPs-PdMER.

Source data

Mechanistic origin of the catalytic performance

Rh NPs show a much higher selectivity for AA in the hydrogenation of CAAs than Pd catalysts (including the Pd NPs used in this work and the Pd catalyst reported in the literature21) in both the PdMER and conventional reactor. Specifically, the three CAAs were highly selectively hydrogenated to AA on the Rh NPs, while TCAA and DCAA were hydrogenated to MCAA and AA over the Pd catalysts, and MCAA barely underwent any hydrogenation. Similar phenomena were also observed in the hydrogenation of chlorophenols on the two catalysts. Additionally, nitrite and chlorate were rapidly hydrogenated over the Rh NPs, while the hydrogenation of nitrate or perchlorate was slow or even impossible. To identify the unique catalytic properties of the Rh NPs, we first calculated the adsorption energies of hydrogen and MCAA on Rh and Pd using density functional theory (DFT). As shown in Supplementary Table 1, the calculated adsorption energy of the hydrogen atom on the different facets of Rh and Pd decreases in the order Pd(110) > Rh(110) > Pd(100) > Rh(111) > Pd(111), indicating that Rh has no obvious advantage in adsorptive and reductive capacity for hydrogen atoms compared with Pd. Therefore it is believed that the difference in the adsorption energies of the hydrogen atom on Rh and Pd is not the main reason why Rh is superior to Pd in producing AA from CAAs. Unlike the hydrogen atom, the adsorption energy of MCAA on the three facets of Rh is greater than or equal to that on the same facets of Pd (Supplementary Table 1 and Supplementary Fig. 11). This indicates that MCAA is likely to be more strongly adsorbed on Rh than on Pd. This conclusion is further supported by MCAA adsorption (Supplementary Fig. 12a) and cyclic voltammetry (CV) measurements (Supplementary Fig. 15a,b). Hence, it can deduced that the stronger adsorption of MCAA on Rh NPs is one of the reasons why TCAA can be hydrogenated directly to AA rather than just MCAA in the Rh NPs-PdMER. MCAA was not hydrogenated on Pd NPs due to its weak adsorption on Pd. A similar principle also applies to the hydrogenation of chlorophenols in the Rh and Pd NPs-PdMERs, and the hydrogenation of perchlorate and nitrate on Rh NPs. In other words, the hydrogenation of chlorophenol to cyclohexanol instead of phenol in the Rh NPs-PdMER is due to the strong binding of phenol to the Rh NPs (Supplementary Figs. 12b and 16), which allows it to be further hydrogenated to cyclohexanol (Supplementary Fig. 17a). The lack of hydrogenation of perchlorate and the slow hydrogenation of nitrate are caused by the weak adsorption of the two oxidized contaminants on Rh NPs (Supplementary Fig. 15c,d).


We have developed a Rh NPs-PdMER without the need for a supporting electrolyte for the hydrogenative treatment of oxidized contaminants in drinking water and industrial effluents. The hydrogenation compartment in the Rh NPs-PdMER is separated from the electrolyte by a dense palladium membrane that reduces protons or water to hydrogen atoms. The hydrogen atoms can permeate through the palladium membrane to hydrogenate oxidized contaminants over the Rh NPs in the hydrogenation compartment, allowing electrochemical hydrogenation to proceed in non-conductive water while using a high concentration of electrolyte in the electrochemical compartment. This not only broadens the range of applications of ECH, but also greatly reduces the energy consumption of the hydrogenation treatment process. Furthermore, owing to the strong adsorption of hydrogenation intermediates on the Rh NPs, the Rh NPs-PdMER can hydrogenate most of the oxidized contaminants in drinking water and industrial effluents to meet the requirements set by the US EPA, making this electrochemical hydrogenation technology, which has no need for a supporting electrolyte, appealing for water treatment.


Materials and chemicals

The palladium membrane (with a purity of ~99.9 wt% and thickness of 5−20 μm), Nafion 117 membrane, Pt sheet electrode (1 × 1 cm2) and Ag/AgCl reference electrode were provided by Hangzhou Saiao Technology ( All oxidized contaminants, including haloacetic acids, chlorophenols, bromate, chlorate, nitrate and perchlorate with purities of 98–99 wt%, were purchased from Aladdin. RhCl3 (99.5 wt%) and Na2PdCl4 (99.5 wt%) were obtained from Adamas-Beta. Methanol (HPLC grade), used for HPLC analysis, was obtained from National Medicines. Deionized water with a resistivity of 18 MΩ cm, obtained from a Millipore Milli-Q system, was used to prepare all reaction solutions.

Preparation of Rh NP- and Pd NP-modified Pd membranes

An electrodeposition method was used to prepare Rh NP- and Pd NP-modified palladium membranes (Rh NPs-Pd and Pd NPs-Pd membranes) with a projection area of 1.76 cm2 used in batch Rh NP-PdMER and Pd NPs-PdMER reactors (Supplementary Fig. 1). A 15 ml solution containing 1 M NaCl and 1.5 mM RhCl3 or Na2PdCl4 at pH 2 was used as the electrodeposition solution. A Pd membrane with a projection area of 1.76 cm2 and a Pt sheet with a projection area of 1 × 1 cm2 were used as the cathode and anode, respectively. A current of 4 mA was applied to the cathode at 25 °C in air atmosphere until the colour of the electrodeposition solution changed from rose-red or yellow to colourless. A similar electrodeposition method was used to prepare Rh NPs-Pd membranes with a projection area of 3 × 3 cm2 used in the continuous Rh NPs-PdMER (Supplementary Fig. 9), except that a 30 ml solution containing 1.5 mM RhCl3 and 1 M NaCl at pH 2 was used as the electrodeposition solution. The prepared Rh NP- and Pd NP-modified Pd membranes were placed in deionized water for later use.

Characterization of Rh NP- and Pd NP-modified Pd membranes

The morphology and size distribution of the as-prepared Rh NPs-Pd and Pd NPs-Pd membranes, as well as the Rh and Pd NPs ultrasonically removed from the two Pd membranes, were examined by scanning electron microscopy (Hitachi S-4700 microscope) and transmission electron microscopy (Tecnai G2 F30S-Twin microscope). The real surface areas of the Rh NPs-Pd and Pd NPs-Pd membranes were calculated using an underpotential deposition (UPD) method28, which involved measuring the Faradaic charge consumed in the anodic removal of a monolayer of the deposited metal. An electrochemical workstation (CHI660B, CH Instruments) and a three-electrode cell (Supplementary Fig. 18) were used to perform the UPD.

The Faradaic charge (Q, C) was calculated using the following equation:

$${{Q}} = {{I}} \times \Delta {{t}} = \frac{{{{I}} \times \Delta {{E}}}}{{{v}}}$$

where I is the current (A), ∆t is the sweep time (s), ∆E represents the potential range (mV) and v is the sweep rate (mV s1) for anodic removal of the UPD Cu layer deposited in the linear sweep current–potential scan (Supplementary Fig. 13e,f). The total number (N) of copper atoms in the UPD Cu layer was calculated using the following equation:

$${{N}} = {{n}} \times {{N_{\mathrm{a}}}} = \frac{{{{Q}} \times {{N_{\mathrm{a}}}}}}{{{{z}} \times {{F}}}}$$

where n is the number of moles of Cu in the UPD Cu layer, Na is Avogadro’s number (6.02 × 1023 mol−1), Q (C) is the Faradaic charge consumed in the anodic removal of a monolayer of UPD Cu, z is the number of electrons required to oxidize one copper atom and F is Faraday’s constant (96,500 C mol−1). The following equation was used to calculate the real surface area (S, nm2) of the two catalytic membranes:

$${{S}} = {{L}}_1{{{\mathrm{ \times }}}}L_{{{\mathrm{2}}}} = {{d}}_1 \times \sqrt {{N}} \times \frac{{\sqrt 2 }}{2}{{d}}_2 \times \sqrt {{N}}$$

where L1 and L2 are the length and width, respectively, of the UPD Cu layer deposited with N copper atoms, and d1 and d2 are the distances between adjacent copper atoms in the UPD Cu layer deposited in different directions (Supplementary Fig. 14).

Hydrogenation experiments

The oxidized contaminants were hydrogenated in two types of palladium membrane electrochemical reactors (Supplementary Figs. 1 and 9) and an H-type electrochemical reactor (a conventional electrochemical reactor). The pure water containing oxidized contaminants in the chemical compartment of the batch Rh NPs-PdMER and Pd NPs-PdMER reactors (Supplementary Fig. 1) and the catholyte of the H-type electrochemical reactor were vigorously stirred during the electrochemical hydrogenation reaction. The rate of flow of tap water containing oxidized contaminants through the continuous Rh NPs-PdMER (Supplementary Figs. 9 and 10) was set to 10 cm h−1 during the electrochemical hydrogenation reaction. Aqueous 0.5 M H2SO4 was used as the electrolyte solution in the continuous Rh NPs-PdMER. Each hydrogenation experiment was repeated at least three times under exactly the same conditions to guarantee reproducibility. Unless otherwise noted, all experiments were carried out at 25 °C in air atmosphere.

The product yield was calculated as the molar ratio of the product produced to the initial amount of oxidized contaminant. The conversion of oxidized contaminant was determined as the ratio of the amount of oxidized contaminant eliminated to the initial amount.

The electrical energy consumption (E, kWh m3 water) of the hydrogenation process was calculated using the following formula:

$${{E}} = \left( {{{I}} \times {{U}} \times {{t}}} \right)/\left( {1,000 \times {{V}}} \right)$$

where I is the applied current (A), U is the average cell voltage (V), t is the reaction time (h) and V is the volume of the treated water (m3).

Analysis of oxidized contaminants and their hydrogenated products

The concentrations of chlorophenols and PhOH during the reactions were quantified using a Thermo Scientific Dionex ultimate 3000 HPLC instrument. The concentrations of CYC-one and CYC-ol were determined by gas chromatography (GC) using an Agilent 7890A device. The HPLC system was composed of a symmetry column (250 mm length × 4.6 mm internal diameter, 5 μm particle size), an injection valve with a 20 μl sample loop and an ultraviolet detector (λ = 280 nm). A mobile phase of methanol–water (4:1, v/v) with 0.03 M H3PO4 at a flow rate of 0.06 l h−1 was used for isocratic elution. The gas chromatographic system was composed of an Agilent 7683B autosampler coupled to a flame ionization detector and a capillary column (HP-INNOWAX, 30 m × 0.32 mm internal diameter, 0.5 μm particle size). The column temperature was increased from 50 to 250 °C at a rate of 15 °C min−1. The temperature of the injector was controlled at 250 °C. High-purity nitrogen (99.999%) was used as the carrier gas at a flow rate of 6.5 ml min−1.

Reaction solutions were filtered through an injection filter and then directly subjected to HPLC analysis after appropriate dilution with deionized H2O. For GC analysis, the organic products were obtained from the reaction mixture by (1) extraction with CH2Cl2, (2) desiccation over anhydrous sodium sulfate, (3) filtration using a sand core funnel and (4) appropriate dilution with CH2Cl2. Standard calibration curves were used to determine the concentration of 4-CP and its hydrogenated products. The precision of the mass balance measurements was 100 ± 5% for GC analysis and 100 ± 3% for HPLC analysis.

The concentrations of the haloacetic acids, AA, ClO3, ClO4, BrO3, NO2, NO3, SO42−, Cl and Br, and NH4+ were determined by ion chromatography (IC) using Dionex ICS-2100 and Dionex ICS-900 instruments, respectively, and standard calibration curves. The Dionex ICS-2100 IC and Dionex ICS-900 instruments were equipped with Dionex IonPac AS11-HC and IonPac CS12A columns, respectively. The precision of the mass balance measurements was approximately 100 ± 3%. The reaction solutions were filtered through an injection filter and then directly subjected to IC analysis after appropriate dilution using deionized H2O. Supplementary Fig. 19 shows the typical evolution of the IC peaks of reactants and products over time during the hydrogenation of TCAA in pure water in the Pd NPs-PdMER and Rh NPs-PdMER.

DFT calculations

All DFT calculations were carried out using the CP2K/Quickstep package44,45. The 1s electrons of H, the 2s and 2p electrons of O and C, the 3s and 3p electrons of Cl, and the 5s and 4d electrons of Rh and Pd atoms were treated as valence electrons. The core electrons were represented by analytic Goedecker–Teter–Hutter pseudopotentials46. Gaussian double-ζ basis sets were used with one set of polarization functions (DZVP)47.

Pd(100), Pd(110), Pd(111), Rh(100), Rh(110) and Rh(111) were modelled by a six-layer slab of p(3 × 3), p(3 × 3), p(4 × 4), p(3 × 3), p(3 × 3) and p(4 × 4) supercells and the sizes of the simulation boxes were 7.7814 × 7.7814 × 24.7267, 11.6721 × 8.2533 × 21.8779, 11.0046 × 9.5303 × 26.2315, 8.0703 × 8.0703 × 24.5110, 11.4132 × 8.0703 × 21.7253 and 10.7604 × 9.3188 × 25.9824 Å3, respectively. Neighbouring slabs were separated by a vacuum of 15 Å to avoid spurious self-interactions. The convergence threshold for ionic steps in geometry optimization was 5 × 10−6 eV. Geometries were deemed converged when the forces on each atom were below 4.5 × 10−4 eV Å−1.

The adsorption energies (Eads) of MCAA and hydrogen are defined as follows:

$${{E}}_{{{{\mathrm{ads}}}}} = {{E}}_{{{{\mathrm{sub}}}}/{{{\mathrm{M}}}}}-{{E}}_{{{{\mathrm{sub}}}}} - {{E}}_{{{\mathrm{M}}}}$$

where Esub/M, Esub and EM denote the total energy of an adsorbed system, a clean substrate and an MCAA molecule in the gas phase or a hydrogen atom, respectively, each of which can be obtained directly from DFT calculations.