Due to increasingly severe regulations and economic constraints, the development of cost-effective production technology with minimal energy consumption, environmental impact and CO₂ emissions is a major drive at the global level ^{[1, 2]}. This is particularly urgent in the fine chemical sector where stoichiometric reactions, toxic additives, protecting groups, sophisticated promoters and the number of processing steps required to achieve high selectivity ultimately results in the highest E-factors ^{[3, 4]}. A solution is to leverage catalysis to increase resource and energy efficiency, by designing novel active materials and green processes through a smarter use of feedstocks and reagents. Among the strategies developed over the last decades, the use of chemical catalysts immobilized on insoluble materials has been preferred by industry owing to the significant benefits in terms of recover and reuse of expensive catalysts, and easier separation and purification procedures ^{[5, 6]}. Solid-supported metal nanoparticles (MNP) are perfectly suited to this purpose owing to their versatility and because they mimic metal surface activation at the nanoscale, thereby bringing selectivity and efficiency to heterogeneous catalysis ^{[7, 8]}. However, achieving 100% selectivity and long-term productivity at low energy costs requires careful tuning of the MNP properties, as well as of the interplay with their active environment, including stabilizers, support material and reagents ^{[9, 10]}. This task is not trivial because the preparation of efficient heterogeneous MNP catalysts is often complicated by multiple issues: (i) lack of control over MNP size and distribution, (ii) lack of reproducibility, (iii) lack of stability under reaction conditions, and (iv) applicability only to specific reaction-support combinations ^{[11, 12]}. Further drawbacks in MNP synthesis relate to the need of hazardous reducing reagents, toxic modifiers, harsh conditions or sophisticated equipment ^{[13, 14]}.

Compared with other materials, ion-exchange resins offer several advantages for the manufacture of supported MNP catalysts, including:

- commercial availability at low cost in various chemical compositions, ionic forms and morphology,

- satisfactory chemical, mechanical and thermal resistance,

- ease of handling and recovery,

- straightforward, non-covalent metal anchoring,

- MNP stabilization due to the dual effects of charged functional groups and porosity,

- potential to engineer bifunctional catalysts, e.g., containing metal and acid sites ^[15],

- facile integration into existing reactor equipment.

These favourable features explain the variety of ion- exchange resin-supported metals (MNP@resin) described in the literature and the successful application of some of those catalysis in industrial processes ^{[16, 17]}.

This paper reviews the recent achievements in the field of immobilization of MNP on ion-exchange resin and the related catalytic application. The focus is on the production processes for fine and commodity chemicals for which a low environmental impact has been demonstrated. The matter has been partially reviewed up to the end of 2010 ^{[18, 19, 20, 21]}. Therefore, the present manuscript covers the most notable papers appearing in the literature from January 2010 to July 2014. Their use in unselective processes (e.g., extraction, metal recovery, pyrolysis, pollutants abatement) ^{[22, 23]}, bulk chemical production (e.g., H₂O₂, H₂) ^{[24, 25]}, fuel cell components ^{[26, 27]}, as well as the use of metal-free ion-exchange resins in acid / base-catalysed reactions (e.g., esterifications, hydrolysis, isomerizations) ^{[28, 29]}, were not included.

A comprehensive description of ion-exchange resins is out of the scope of this review ^{[30, 31]}. Herein a short description of the main features affecting MNP catalyst preparation and performance is provided.

Most ion-exchange resins consist of functionalized cross- linked polystyrene-divinylbenzene copolymers and are conventionally classified into two main groups (Scheme 1) ^[32,33]:

cation-exchange resins (with anionic functionalities and positively charged mobile ions)

- strong acid exchangers (e.g., containing sulfonic acid groups or the corresponding salts)

- weak acid exchangers (e.g., containing carboxylic acid groups or the corresponding salts)

anion-exchange resins (with cationic functionalities and negatively charged mobile ions)

- strong base exchangers (e.g., containing quaternary ammonium groups)

- weak base exchangers (e.g., containing ammonium groups)

Other ion-exchanging materials include acrylic resins ^[34] and perfluorinated chain polymers bearing sulfonic acid heads, like Nafion® ^[35] and Aquivion ^[36].

Cross-linking, typically from 0.5% to 20%, controls resin porosity with low and high cross-linked resins having a gel (microporous) and macroporous structure, respectively ^{[37, 38]}. In turn, porosity regulates some of the resin properties affecting their catalytic application: swelling, exchange capacity, equilibration rate, and selectivity. Usually, the lower the cross-linking, the higher the moisture content, equilibration rate, loading capacity (typically 1.5-10 meq/g on a dry basis) and the ability to accommodate larger ions.

Swelling of microporous resins is crucial for their catalytic activity. Swelling volumes up to 800% have been found for low cross-linked resins, either in water or methanol ^{[39, 40]}. Gel type resins are thus generally preferred over macroporous resins owing to better active-sites accessibility to the reactants in solution ^[41]. However, internal (diffusive) mass transfer limitations, and pressure drops in the case of flow processes ^{[42, 43]}, are usually lower for macroreticular resins ^{[44, 45]}.

The number, type and strength (e.g., acidic) of the functional groups have a direct consequence on both the ease of MNP growth within the resins (see section 3 “Synthesis”) and the activity of the MNP@resin catalysts (see section 4 “Application”). Functional groups are crucial in determining site- inhibition ^[46] or accelerating effects ^[47] and on the potential to carry out reaction sequences in cascade ^[48].

Ion-exchange resins are commercially available in the solid state as powders, pellets (16-400 mesh, 1200-40-μm diameter) ^{[49, 50]} or membranes ^{[51, 52]}. Foams and monolithic ion-exchanging polymers have also been reported for use in laboratories ^{[53, 54]}. The shape and size of these materials allow for their easy and quantitative recovery for reuse by simple filtration or decantation. This is often not the case for other powdered materials commonly used for catalyst support (e.g., silica, zeolites, carbon, etc.). When the particle size is less than 1 μm, the catalyst may not settle out of solution over sufficiently short time, thus catalyst separation may require centrifugation or ultrafiltration. Very fine powders may also clog or poison the reactors or autoclaves employed in catalytic experiments. In a few instances, the recovery of resin beads may require not using magnetic stirring to avoid grinding ^[46].

Common uses of ion-exchange resins include water purification, metal recovery and separation, ion substitution, acid-base catalysis, and as sensors or as solid electrolytes (e.g., in fuel cells, electrolysers, electrodialysis devices) ^[55].

Ion-exchange resin-supported MNP catalysts have been reported with a variety of noble metals (Pd, Ru, Au, Ag Rh, Pt), with palladium and rhodium being the most and the least explored, respectively ^[56]. Regardless of the metal and the resin type, the synthetic procedures usually involve two steps: immobilization of the metal onto the resin by ion-exchange using an appropriate soluble metal precursor, i.e., a (ML_n)^z+ cationic species for cation-exchange resins and a (ML_n)^z- anionic species for anion-exchange resins, followed by metal reduction (Scheme 1).

Ion-exchange is an equilibrium process driven by electrostatic interactions. The position of this equilibrium depends on the size and charge of the metal ion; wherein for a given resin, the affinity is greater for large ions with high valence, for instance Fe³⁺ > Ca²⁺ > Li⁺ for strong cation-exchange resins ^{[57, 58]}. The immobilization step is usually achieved within a few minutes by immersing the resin in a metal precursor solution at room temperature resulting in an atomic level distribution of the metal within the support. Typical metal uptakes are in the range of 50%-80% ^[59].

Metal reduction can be carried out by a variety of reagents and conditions, where the use of “clean” dihydrogen is preferred over an excess of harmful, strong reducing agent (borohydrides, hydrazine, or formaldehyde), provided that high pressures and temperatures can be avoided, as these would make the process uneconomical ^[60]. The ease of reduction depends on several factors, including the metal, the precursor, the resin type and its ionic form ^[61]. For example, in the case of gel-type sulfonic resins, harsher conditions are generally required in the order Pd < Rh < Ru < Pt ≤ Au, while metal reduction is easier from the lithium or sodium form rather than the protonated form of the resins. This may be attributed to the change in the reduction potential of the metal because of the different acidity of the surrounding medium (vide infra). A selection of recent, synthetic friendly methods for MNP@resin catalysts are summarized in Table 1. The choice of the reducing agent may have consequences on MNP size and distribution within the support and, in turn, on the catalytic activity. Egg-shell distributions may be obtained if the rate of metal cation diffusion within the solid matrix (due to the concentration gradient generated by its reduction) is higher than the rate of reduction which, in turn, depends on the rate of diffusion of the reducing agent (vide infra) ^[62].

Overall, ion-exchange resins allow for the synthesis of supported MNP catalysts to be achieved through simple, one-pot methods, under mild conditions, whereby MNP are generated within the pores of the solid matrix, avoiding cumbersome heterogenization procedures of preformed MNP. Well-dispersed, small and stable metal particles are usually obtained, thanks to the concurrent effect of charged functional groups (electrostatic stabilization) ^[63] and porosity (steric stabilization) ^{[64, 65]}.

PdNP-containing solids are among the most used and versatile heterogeneous catalysts, so they have frequently been used to probe the effect of the support on the catalytic activity ^{[66, 67]}. PdNP@resins are effective catalysts, including as bifunctional catalysts ^{[68, 69]}, for a variety of processes and are at the core of the large scale manufacture of methyl isobutyl ketone and methyl-tert-butyl ether ^{[70, 71]}. Common precursors employed for the PdNP@resin synthesis are either [Pd(NH₃)₄]²⁺ or Pd(CH₃CO₂)₂, for cation exchange resins, and PdCl₄^2-, for anion exchange resins, which require high temperature treatment or large excesses of sodium borohydride for their conversion to Pd metal ^{[72, 73]}.

Recently, fast palladium reduction under the mild conditions of H₂ (1 x 10⁵ Pa) and room temperature was reported using Pd(NO₃)₂ ^[74] or [Pd(CH₃CN)₄](BF₄)₂ ^[75] in conjunction with commercial, gel-type sulfonic resins (Dowex® 50WX2) irrespective of the ionic form (H⁺, Li⁺), wherein [Pd(CH₃CN)₄]²⁺ is preferred owing to superior reproducibility (Table 1, entry 1-4). No H₂ reduction was observed for the [(dppp)Pd(CH₃CN)₂]²⁺ or PdCl₄^2- precursors under analogous conditions (dppp = 1,2 diphenylphosphino propane). In line with the higher reduction potential of Pd²⁺ compared with PdCl₄^{2- [76]}, these results indicate that smooth reduction can be achieved, provided that a metal species bearing weakly coordinated ligands is used. Importantly, this strategy allows for the in-situ synthesis of PdNP@resin under conditions appropriate for catalytic hydrogenation, with significant benefits both in terms of minimization of synthetic steps and enhanced catalytic activity. This was attributed to the smaller size of the embedded PdNP (see later “Application”). Typical PdNP size was around 3 and 4 nm for the in-situ and H₂ pre-reduced species, respectively, at ca. 1%Pd (mass fraction; the same below) loading (TEM, XRD, SAXS) ^{[74, 75]}, whereas, according to previous reports, larger metal crystallites (5 nm) were obtained by sodium borohydride reduction ^{[77, 78]}. Consistent with the fast metal reduction, PdNP obtained by H₂ treatment were found to be evenly distrib uted within the support (260-μm beads), as confirmed by EDS spectroscopy (Fig. 1) ^[74].

Using the same strategy, palladium immobilization was also reported on an unconventional polymeric borate macroporous monolith (Monobor ^[79]) in which cation exchange tetraphenyl borate anions are incorporated into a highly cross-linked styrene-divinylbenzene matrix ^[80]. Thus, PdNP were grown in one-pot by flowing a THF solution of Pd(NO₃)₂ through a monolith using an HPLC pump, followed by reduction under a stream of H₂ (1 x 10⁵ Pa, 2 ml/min, r.t., 3h) (Fig. 2 and Table 1, entry 5). The procedure afforded homogeneously distributed PdNP, with 0.67% metal loading and a diameter of 2.5 nm (TEM).

Immobilization of PdNP on macroporous Amberlite IRA-900 anion-exchange resin was recently obtained by chemical vapour deposition of the volatile precursor Pd(C₃H₅)(C₅H₅) at 30 °C, followed by H₂ (1 x 10⁵ Pa) -reduction (Table 1, entry 7) ^[81]. Embedded 2.6 nm particles were obtained with 1.7%Pd loading.

Bimetallic Pd catalysts have also been explored, aimed at tuning the performance of Pd in various process (Heck, Sonogashira and Stille coupling) by a slight modification of its electronic and/or morphological properties through interaction with an additional metal. Thus, a library of catalysts with fixed Pd loading (1% (m/m)) and variable content of Pt (0.1%-1%) or Au (0.25%-1%) was prepared by the immobilization of [Pd(NH₃)₄](SO₄)₂, [Pt(NH₃)₄](NO₃)₂ and [Au(en)₂]Cl₃ precursors (en = 1,2-diaminoethane) on the macroreticular sulfonated resin Lewatit K2621, followed by treatment with concentrated aqueous formaldehyde at reflux for 3 h (Table 1, entries 8,10) ^[82] or under H₂ (5 x 10⁵ Pa) at 60 °C (Table 1, entries 9,11) ^[83]. Pd/Pt NP were found to be of smaller size when obtained by H₂ (3-4 nm) rather than by formaldehyde reduction (ca. 5 nm), with bimetallic particles smaller than the monometallic Pd particles. A broader size distribution was found for the bimetallic Pd/Au species (4-30 nm).

Pd/Cu bimetallic composites (ca. 1.5%) comprised homogeneously distributed 4.9-nm Pd/PdO/CuO nanoparticles have been reported on a cationic macroporous Amberlite resin bearing a formate counter anion (ARF) ^[84]. Partial reduction of Pd(CH₃CO₂)₂ and Cu(CH₃CO₂)₂ precursors was attributed to the embedded formate in this case (Table 1, entry 12).

Core-shell Pd/Co NP on ion-exchange resins were also obtained by a sequence of metal anchoring-reduction steps using alternate loading of Co(NO₃)₂ and Pd(NH₃)₄Cl₂ precursors and NaBH₄ or Na₂S₂O₄ reducing agents ^[85]. Two materials were used as support: a DVB-crosslinked carboxylate polyacrylic acid (macroporous type) and a sulfonate polypropylene-styrene-DVB copolymer (PS, gel type) (Table 1, entry 13). The procedure led to a pronounced location of metal on the polymer surface (SEM) at 8% total metal content.

It is worth mentioning that Pd-doped macroreticular Amberlyst type sulfonic acid resins with (0.1%-0.3%) Pd content are now commercially available from the Dow Chemical Company ^[86].

Rhodium nanoparticles are pivotal in catalysis because of the unsurpassed performance in oxidation, hydroformylation, carbonylation and hydrogenation reactions, particularly of monocyclic arenes to cyclohexane derivatives ^{[87, 88]}. Previously reported methods for solid-supported RhNP synthesis involve impregnation-reduction methods using RhCl₃, Rh(NH₃)₆Cl₃, Rh(CH₃CO₂)₂ or Rh(NO₃)₃ precursors followed by sodium borohydride or H₂ high-temperature (400-1000 °C) reduction, or dispersal methods using preformed RhNP ^{[89, 90]}, vapour deposition ^[91], flame spray pyrolysis ^[92] and laser ablation ^[93] techniques. Support materials include montmorillonite ^[94], silica ^[95], CNT ^[96], titania ^[97], alumina ^[98], ceria ^[99], zirconia ^[100], and MgAl₂O₄ ^[101]. The use of ion-exchange resins has gone almost unexplored.

Following the same strategy described above for the immobilization of PdNP, rhodium NPs were recently prepared within the pores of a gel-type, cation-exchange sulfonic resin Dowex, using [Rh(NBD)₂]BF₄ as the soluble metal precursor ^[102]. Choice of the precursor was motivated by the known propensity to H₂ metal reduction via norbornene elimination ^[103]. Indeed, treatment of the rhodium(I)-resin with hydrogen under very mild conditions (RT, H₂ (1 x 10⁵ Pa), CH₃OH, ca. 10 min.) provided rhodium(0)-containing black beads with 1%Rh loading, corresponding to ca. 80% metal uptake (Table 1, entry 14). Importantly, the success of the H₂ reduction was determined by the ionic form, only working with the lithium form of the resins, and the use of NaBH₄ was required to achieve metal reduction on the protonated polymer (Table 1, entry 15). This can be attributed to the different reduction potentials of rhodium in an acidic medium. Irrespective of the reducing agent or the resin type, XRD and TEM analyses revealed the presence of agglomerates of up to 30 nm of embedded RhNP with a mean particle size of 3.0 ± 0.7 nm. Both peripheral and homogeneous metal distributions with the support (260-μm beads) were observed to depend on the reduction conditions.

The approach was also successful in generating RhNP within the Monobor monolith using a continuous flow of [Rh(NBD)₂]BF₄ solution in methanol followed by reduction under a stream of H₂ (1 x 10⁵ Pa, RT) ^[104]. Homogeneously distributed RhNPs, 3.9 nm in diameter (TEM, Fig. 3), were obtained with a 0.82% rhodium loading (Table 1, entry 16).

An alternative in situ reduction-deposition protocol for RhNP@resin was reported using Amberlite IRA-900 resin partially exchanged with borohydride anions and subsequently treated with a THF solution of RhCl₃ at 80°C for 1 h ^[105]. HRTEM and EDX analyses showed the presence of 2-4-nm RhNP and 0.4% metal content (Table 1, entry 17). As expected, harsher reaction conditions were required for RhCl₃ compared with the reduction of [Rh(NBD)₂]⁺.

RuNPs on ion-exchange resins are quite novel in catalysis. One early report described the immobilization of ruthenium on Amberlyst-15 anionic resin by thermal decomposition of supported [Ru(NH₃)₆]Cl₃ ^[68]. More recently, RuNPs on a sulfonated cation exchange resin Dowex were obtained by impregnation with an aqueous solution of RuCl₃ followed by treatment with NaBH₄ (Table 1, entry 18) ^[106]. Slow metal reduction by borohydride resulted in 2.8-nm embedded Ru particles (TEM), 0.85%Ru loading (ICP-OES) and an egg-shell metal distribution within the solid matrix (EDS) (Fig. 4). An outer shell of depth 3.7 ± 0.6 μm, with a mean Ru content of ca. 3.1%, and an inner sphere with a mean Ru loading of ca. 0.7%, were estimated from EDS analyses on sections of 276 ± 2-μm diameter Ru@Dowex beads.

Several AuNP@resin catalysts have been reported in the recent years, mostly for use in oxidative and coupling reactions. The most commonly used gold precursor for strong cation exchange resin is [Au(en)₂]Cl₃. Gold nanoclusters inside gel-type Amberlyst and macroreticular Lewatit resins have been formed either by solvent-free H₂ reduction (40 x 10⁵ Pa, 80 °C) or by aqueous NaBH₄ treatment of [Au(en)₂]³⁺ containing resins, resulting in ca. 1%Au polymers (Table 1, entry 19-20) ^[107]. The AuNP dispersion is dependent on the preparation method. As previously discussed for Pd, Rh and Ru, and in line with relatively fast (H₂) and slow (BH₄^-) metal reduction, AuNPs have been found to be smaller (3-5 nm, Fig. 5) and evenly distributed within the support for H₂-based reduction, whereas larger (5-7 nm) and peripheral particles (ca. 30-μm size shell for ca. 400-μm size beads, Fig. 6) were observed for the borohydride synthesis.

For anion-exchange resins, the usual precursor is an AuCl₄^-salt. Thus, treatment of Dowex Marathon MSA in its chloride form with HAuCl₄, followed by reduction with an excess of NaBH₄ and washing with a saturated solution of Na₂CO₃, gave supported 4.8-nm AuNP at 0.1% metal content (Table 1, entry 21) ^[108].

Similarly, Dowex-M43 resin treated with NaAuCl₄ and NaBH₄ resulted in 8.8-nm AuNP at 1%Au loading (Table 1, entry 22), while smaller AuNP (5.4 nm) were immobilized by contacting the resin with a preformed colloidal solution of tetrakishydroxypropylphosphonium chloride (THPC)- stabilized AuNP (Table 1, entry 23) ^[109].

AuNP have also been supported on imidazolium salt (SILLPs) ion-exchange resins. SILLP polymers containing styrene, divinylbenzene (at various cross-linking degrees) and alkylimidazolium chloride subunits (at various loadings) were recently prepared from commercial Merrifield resins and used to explore the effect of these ionic-liquid-like resins on Au NP stabilization ^[110].The synthetic procedure is sketched in Scheme 2 for the methylimidazolium group. The conventional two-step procedure using HAuCl₄ and a NaBH₄ reduction resulted in fully adsorbed metal and 2-21-nm AuNP (TEM), depending on the polymer composition (Table 1, entry 24). However, the lack of application prevented any study of sintering or leaching of Au particles under catalytic conditions.

In summary, to achieve the immobilization of noble MNP on ion-exchange resins, an appropriate combination of metal precursor, resin type, solvent and reaction conditions must be selected. Table 2 reports some parameters found in the literature enabling green protocols for MNP synthesis on gel-type, strong cation exchange sulfonated resins. Cells highlighted in green indicate the combinations for which MNP immobilization is possible under the mild and clean conditions of H₂ (1 x 10⁵ Pa) and room temperature. Cells highlighted in yellow indicates those metals whose reduction can be accomplished only under stronger conditions, i.e., H₂ > 5 x 10⁵ Pa and T > 60 °C or require the use of NaBH₄ as a reducing agent. Choice of the reducing agent directly affects the MNP@resin properties with potential consequences on their catalytic performance. As a general trend, the use of H₂ results in a narrow dispersion of smaller MNP compared with NaBH₄. The distribution of the MNP within the solid support is more difficult to predict because it is regulated by a subtle combination of several factors: the swelling ability of the resin (which depends on the solvent used, the ionic form and the porosity of the resin), the concentration of the reducing agent (or pressure in case of H₂), the temperature, the metal loading and the intrinsic mobility of the metal ions to be reduced. It is expected that easily reduced metals with low mobility (e.g., Pd²⁺) will be evenly distributed within the support under mild reaction conditions, whereas ions with higher reduction potentials (e.g., Au³⁺) would require harsher conditions, resulting in more peripheral distributions.

Selective hydrogenation of hydrocarbons with multiple C=C and/or C≡C bonds to achieve partially hydrogenation products is a highly desired but challenging process in the chemical industry ^[111]. Particularly, the catalytic semi-hydrogenation of substituted alkynes to obtain (Z)-alkenes is a reaction at the core of the production of many important fine chemicals, including bioactive molecules, food additives and fragrances ^{[112, 113]}. It is also crucial in the polymerization industry for the complete elimination of alkynes from alkene feedstocks ^[114]. The conventional Pd-based heterogeneous catalysts used for this purpose are generally problematic because of selectivity issues, overhydrogenation to alkanes, isomerization, and poor selectivity with respect to other functional groups ^{[115, 116]}. Strategies have been developed to enhance the selectivity of these systems by the addition of variable amounts of contaminants, either organic bases, carbon monoxide, sulfides, dimethyl sulfoxide or metal ions (Cu, Pb), whose purpose is invariably to lower the hyperactivity of the metal ^{[117, 118]}. On the industrial scale, the hydrogenation processes are usually carried out using the Lindlar catalyst (5%Pd on CaCO₃ doped with (2%-3%) Pb) ^[119], which has serious drawbacks in terms of reuse, deactivation, the presence of toxic lead, and the need for excess amine modifier ^{[120, 121]}.

Various alkynes have been recently hydrogenated by Pd@resin catalysts, both under batch and continuous flow conditions, with the latter offering considerable advantages in terms of space-time-yield productivity (STY), purification, environmental impact, automation and energy consumption ^{[122, 123]}.

The reduction of 3-hexyn-1-ol (1) is of particular interest because the cis-partial hydrogenation product leaf alcohol 1a is a market flavour ingredient ^{[124, 125]}, currently manufactured at 400 t/y and ca. 96% selectivity at 99% conversion by a batch process using the Lindlar catalyst ^{[126, 127]}. The hydrogenation of 1 was first scrutinized using the lithium form of Pd@Dowex under batch conditions, showing excellent catalyst activity and selectivity under very mild condition without additives ^[74]. An alkene selectivity of 99.8% (97% of which the cis isomer) was obtained at 98.5% conversion (TOF 880 h^-1) under H₂ (1 x 10⁵ Pa) and RT (Table 3, entry 1). The catalyst could be reused with no significant loss of activity and selectivity, nor leaching of Pd species into solution, over six cycles. The excellent catalyst performance was attributed to the restricted growth and high stabilization of the embedded PdNP obtained in-situ by catalytic hydrogenation, which was attributed to an ‘‘excess of substrate stabilizing effect’’ ^[128], in combination with the appropriate swelling and resistance of the support.

The catalyst was thus packed (150-300-μm size beads) into commercial glass columns and tested under continuous flow and similar conditions, showing slightly lower efficiency but better STY, compared with batch operations (80% ene and 89% cis selectivity at 75% conversion, TOF = 352 h^-1, STY = 1.02 kg/(L·h)) ^[75] (Table 3, entry 2).

Much better catalyst performance was observed under continuous flow using the monolithic Pd@Monobor system (Fig. 2) ^[80]. An alkene selectivity of 95% (93% cis) was obtained at 99% conversion (TOF = 829 h^-1, STY = 0.85 kg/(L·h)), was detected with no conversion nor selectivity decay over 14 h time-on-stream under H₂ (1.3 x 10⁵ Pa) and RT (Fig. 7) (Table 3, entry 3). This is the first example of a flow process for the production of cis-3-hexen-1-ol with a selectivity comparable to that of the industrial process and the best result so far reported under continuous flow. The efficiency and durability of Pd@MonoBor were attributed to the strategic combination of non-coordinating and chemically inert borate functional groups in the polymer ^{[129, 130]}, with the rigid, homogeneous macroporous structure of the monolith ^{[131, 132]}. This results in the formation of small and stable PdNP, as well as a flat flow profile and scarce flow resistance over prolonged reaction time.

Pd@Monobor allowed for the effective, continuous partial hydrogenation of a variety of substituted alkynes, including 2-methyl-3-butyn-2-ol (2) to give the alkene 2a ^[80], an important intermediate in the synthesis of vitamins and perfumes, which is currently produced in 95%-97% yield using the Lindlar catalyst ^{[133, 135]}, with 93.9% selectivity at 92% conversion (Scheme 3).

Similarly to the hydrogenation of alkynes, the reduction of a variety of α,β-unsaturated carbonyl compounds was reported using Dowex and Monobor-supported Pd or Rh MNP catalysts. Scheme 4 and Table 4 summarizes the reaction products and catalytic efficiency for trans-4-phenyl-3-buten-2-one (3), isophorone (4) and (R)-carvone (5) substrates. All of the catalysts were active under H₂ (1 x 10⁵ Pa) and room temperature for all substrates, with remarkable selectivity toward C=C versus C=O bond hydrogenation ^{[134, 135]}.

Thus, the use of Pd@Dowex under batch conditions provided the monohydrogenated products 3a and 4a and the di-hydrogenated product 5d, in excellent to good selectivity at very high conversion levels (Table 4, entry 1, 4, 7) ^{[74, 75]}. Better selectivity were observed under flow conditions using the monolithic Pd@Monobor and Rh@Monobor catalysts, albeit with lower TOFs, which is common when comparing continuous flow with batch operations ^[136]. Thus, Pd@Monobor gave 3a in 96% selectivity at nearly full conversion (Table 4, entry 2), which is the best result so far reported for the catalytic hydrogenation reaction of 3 under continuous flow ^{[131, 137]}, while complete selectivity towards the important fine chemical intermediate dihydroisophorone 4a ^[138] was observed at 92% conversion (Table 4, entry 5) ^[104]. An unprecedented 85.7% selectivity towards cis-carvomenthone (5d) was observed at full conversion in the hydrogenation of 5 using Rh@Monobor (Table 4, entry 9) ^[104]. The best selectivity observed under continuous flow was attributed to the shorter contact time of the intermediate hydrogenation products, compared with the corresponding batch systems ^[139]. Differences observed in the activity and selectivity of the Rh and Pd-catalysed reactions are more difficult to explain. Previous studies invoked different adsorption mechanisms on the metal surface ^[140]. MNP size may play a role, although the lack of direct connection between size and shape of RhNP and their catalytic activity has been demonstrated ^[141].

The selective hydrogenation of substituted, monocyclic arenes to cyclohexane derivatives is a reaction of interest to the polymer, fuel and fine-chemicals industries, and is usually achieved under mild conditions using homogeneous Rh catalysts, and more drastic conditions with heterogeneous Rh catalysts ^[142]. In-situ prepared Rh@Dowex catalysts (Table 1, entry 14) efficiently catalyse the hydrogenation of arenes under the very undemanding conditions of H₂ ((1-10 x 10⁵ Pa) at room temperature in the liquid phase, providing excellent chemo- and cis/trans selectivity to the corresponding cyclohexane derivatives at full substrate conversion. However, this reaction has very poor tolerance for ketone groups toward hydrogenation (Table 5) ^[102].

The different rates of hydrogenation observed for the various substrates were attributed to the steric hindrance and electronic properties of the substituents, as has been already described for other metallic catalysts ^{[143, 144]}. Interestingly, the parent catalyst obtained by NaBH₄ pre-reduction of rhodium onto the protonated form of Dowex (Table 1, entry 15), did not show any significant arene hydrogenation activity. This was ascribed to the deactivation of aromatic substrates toward metal hydrogenation by Brӧnsted acidic supports.

Conversion of vegetable biomass-derived products by heterogeneous bifunctional catalysts, which combine metal and acid sites, is of current interest in the sustainable production processes of fine chemicals ^[15]. Conventional catalysts for this purpose usually need drastic reaction conditions, organic solvents and/or the addition of strong soluble acids to work ^[145]. The achievement of high selectivity requires the combination of well-defined supported acid and metal sites acting in concert under the same reaction conditions ^[146].

The one-pot, hydrogenation-condensation reaction of cellulose-derived levulinic acid (LA) to γ-valerolactone (GVL) via γ-hydroxyvaleric acid was recently reported using Ru@Dowex under continuous flow (Scheme 5) ^[106]. Very mild conditions (70 °C, H₂ (7 x 10⁵ Pa)) using 84-μm packed catalyst beads (4.8 mmol/g acidic sites, Ru 0.85%) afforded > 97% conversion to pure GVL (TOF = 117 h^-1, 1.16 mol/(g·h), STY = 0.12 kg/(L·h)), with no appreciable activity decay nor Ru leaching into solution over 35 h time-on-stream. Previously reported methods for the catalytic production of GVL from LA under mild conditions include the use of a mechanical mixture of 5% Ru/C and Amberlyst co-catalyst, demonstrating the positive contribution of the acidic resin to the metal hydrogenation activity (TOF = 80 h^-1, 0.7 mol/(g·h)) ^[147]. The superior performance of the Ru@Dowex catalyst was attributed to the favourable combination of well-defined acid and RuNP hydrogenation sites on the support, with the optimal resistance and swelling of the resin in water.

The Suzuki-Miyaura is a well-established cross-coupling reaction between aryl-boronic acids and aryl-halides catalysed by homogeneous palladium(0) complexes, which is performed in polar organic solvents (e.g., DMF) or in water and in the presence of a base (Scheme 6). The palladium catalyst precursors used usually involve toxic and expensive phosphine ligands ^{[148, 149]}. In a heterogeneous phase, Pd catalysts on a variety of supports have been reported with major problems arising from catalyst deactivation due to metal leaching, PdNP sintering, poisoning by carbonaceous deposits, and support degradation ^[150].

1.7%Pd@IRA-900 on ion exchange resin catalysed the Suzuki coupling under phosphine-free, aerobic conditions in the presence of K₂CO₃ in a mixture ethanol:water = 1:1 (V/V) ^[81]. The best results were observed for the coupling of iodobenzene and phenylboronic acid at 70 °C (conversion > 99%, TOF = 2486 h^-1). The catalyst could be reused without metal leaching or requiring regeneration for five cycles, with only a minor decrease in activity. Curiously, no catalytic activity was observed when the chloride form was replaced by the OH^- form of the resin with the Suzuki coupling being performed under base-free conditions (Table 1, entry 7). Core-shell Pd/Co NP on sulfonate polystyrene resins (Table 1, entry 13) were active in the coupling of 4-bromoacetophenone with phenylboronic acid under comparable conditions (1%Pd, DMF:water = 4:1, K₂CO₃, 70 °C) ^[85].

The coupling of terminal alkynes and aryl halides (Sonogashira reaction ^[151]) was explored using PdCu@ARF catalysts (Scheme 6) (Table 1, entry 12) ^[84]. Thus, the reaction of 4-iodotoluene with propargyl acetate gave the expected product in 90% yield in the presence of K₂CO₃, CH₃CN solvent and at 80 °C. No products due to homo-coupling of alkynes, a common side reaction, nor Pd in solution were detected under a N₂ atmosphere. The catalyst could be reused four times but exhibited a ca. 20% loss in activity.

Nano-structured Au catalysts supported on a variety of insoluble materials have been shown to be effective heterogeneous catalysts for aerobic oxidations ^[152], including the carboxylation of amines and epoxides and polyol oxidation to give carboxylates under basic reaction conditions ^[153]. In particular, the potential lactones products are of considerable interest because of their insecticide and pharmaceutical activity ^[154]. From an industrial point of view, the most advantageous synthetic approach to lactones is the oxidative lactonization of α,w-diols.

The one-pot aerobic conversion of 1,4- and 1,5-diols to lactones in toluene has been performed using a Au@MSA catalyst bearing 0.1%AuNP on a macroreticular anion- exchange resin in its carbonate form (Scheme 7) ^[108]. Aromatic diols were chemoselectively converted into the corresponding lactones, while aliphatic diols, which are prone to overoxidation reactions, showed selectivity in the range of 92% to 97% owing to the formation of the corresponding dicarboxylic acid. Thus, benzene-1,2-dimethanol gave the desired lactone in 100% selectivity and conversion at 50 °C and with air (5 x 10⁵ Pa) (TOF = 500 h^-1), whereas butane-1,4-diol was less reactive (TOF = 285 h^-1) and selective (95 %) under the same reaction conditions. Recycling the catalyst without regeneration with aromatic substrates showed only a slight drop in the catalytic activity from 96% to 93% on the 4th cycle, whereas with aliphatic substrates experienced an immediately decrease in the catalytic conversion from 83% to 15%. In the former case this small drop was attributed to Au leaching into solution (1.6 x 10^-6 in each cycle, ICP-OES), while in the latter it was ascribed to catalyst deactivation by the dicarboxylic acid side product. Washing this deactivated catalytic system with Na₂CO₃ in water after each cycle restored the initial catalytic performance for at least four consecutive catalytic runs. Interestingly, the carbonate form of the Au@MSA catalyst was three times more active than the chloride form, while an acetate counter-anion completely suppressed the catalytic activity, which was attributed to coordination to surface Au-atoms.

The liquid phase oxidation of biomass-derived glycerol by the Au@M43 and Au_THPC@M43 catalysts (Table 1, entry 22 and 23) was tested under batch conditions, and the latter case showed higher activity (50 °C, O₂ pressure = 3 x 10⁵ Pa, 4 eq. NaOH, 90% conversion, TOF = 2098 h^-1), which was attributed to the smaller size of the supported AuNP ^[109]. Scheme 8 shows the main products observed in the selectivity data (%, in brackets). The system was also tested as a continuous flow system, in a packed bed reactor, but exhibited poor selectivity and low conversions.

From this review it is clear that the use of ion-exchange resins offers significant advantages in the development of a sustainable process industry. Benefits are expected in both catalyst manufacture and performance. Thus, noble and non-noble metals may be supported with tuned loadings and composition by means of clean reagents and low energy procedures, starting from low-cost, commercial available products. These catalysts are applicable to a large variety of processes in the liquid phase, often showing better efficiency under milder reaction conditions, compared with other supported metal catalysts. Stability and morphology of resins complement the viability of the catalysts, allowing for effective recover and reuse, reduced metal leaching and ease of integration into batch and flow reactors.

Potential fields of future development include one-pot catalytic conversion of vegetable biomass thanks to their ability to combine metal and strong acid functionalities in one solid material ^[155,156].