The Best of Two Worlds from the Gold Catalysis Universe: Making Homogeneous Heterogeneous

Abstract

Heterogeneous catalysts have well-known advantages in the gas or liquid phase, which include facile separation, possibility of recycling, and potential application to continuous flow processes. Homogeneous catalysts have remarkable properties such as high activity, enantioselectivity, and well-characterized structures. Heterogenization of metal complexes, through anchoring of inorganic/organometallic complexes on porous solid materials, is a way to combine the outstanding properties of both systems. This procedure has already been used with complexes of several metals such as chromium, cobalt, copper, manganese, molybdenum, nickel, or palladium, but heterogenization of gold complexes is still rare. Heterogeneous and homogeneous catalyses by gold have already established on their own, and the heterogenization of gold complexes is a very exciting and promising interface area of chemistry. This work reviews the attempts made so far by several authors on this important topic. The results obtained show that these hybrid materials can have improved activity and selectivity and are potential choice candidates for several catalytic processes.

Introduction

The development of efficient, stable, and selective catalysts is a continuous research challenge. Catalysis plays an important role in many areas such as energy, environment, pharmaceuticals, and materials. The industry also relies on catalysis; for example, the manufacture of plastics, detergents, paints, cosmetics, and food additives would benefit from the cleaner production of epoxides, ketones, aldehydes, and acids.1 Haruta pointed out the use of gold catalysts for the selective oxidation of hydrocarbons to achieve this “cleaner solution.”1 Also, Hughes et al. suggested gold to be part of a solution that lead to “green processing” by using atmospheric oxygen for oxidation reactions.2 The use of gold catalysts in other types of reactions, such as hydrogenations, is also a promising area because it can lead to several interesting reaction products with industrial impact.3 In 1998 and 1999, Thompson reviewed new advances in catalysis by gold, which were many at that time, clearly stating that those were yet “early days” in the use of the unique capabilities of gold as compared with other precious metals.4 Also, in 2002, Bond still considered gold as a “relatively new catalyst.”5 Even nowadays, in spite of all the recent developments, gold is definitely less used in catalysis as compared with other metals.6

Gold “big bang” and the current concept on gold catalysis

Gold nanoparticles have outstanding properties: Although bulk gold reveals great stability,7 nanosized gold has proved to be an excellent catalyst.6a,c,d, 8 The knowledge that gold can be catalytically active in the form of very small nanoparticles, small enough to chemisorb reactant molecules,6a,c,d, 8b was probably the “big bang” of gold catalysis.

Although heterogeneous catalysis is still predominant,6, 8b “explosive growth” in homogeneous gold catalysis took place in the beginning of the twenty-first century, and it is “still accelerating.”6a,c,d, 9 Before that, gold was regarded as a poor homogeneous catalyst, with very low turnover numbers (TONs) for the few examples of reactions catalyzed by soluble complexes. Developments in this field have, however, been equally dramatic since the publication of a paper by Teles et al. in 1998,10 which demonstrated that high TONs and turnover frequencies (TOFs) could be obtained for gold complexes in solution. In 2000, Hashmi et al. also showed the new and great performance of gold-catalyzed reactions, which combines CC and CO bonds formation.11 Since then, it has been found that soluble gold compounds can catalyze some organic reactions that cannot be performed in any other way. 6b,d, 12 Gold-catalyzed reactions that yield organic products through homogeneous or heterogeneous catalyses have been reviewed by Hashmi in several papers.6b, 9, 12, 13

Recently, Angelici published a review that describes catalysis on powdered unsupported gold.14 The author referred, namely, to the reaction between CO, RNH₂, and O₂ to produce isocyanates (RNCO), which react further with RNH₂ to give ureas (RNH)₂CO. Secondary amines HN(CH₂R)₂ can react with oxygen, in the presence of gold, to undergo dehydrogenation to the imine RCHN(CH₂R). Attempts to perform similar reactions with copper and silver metal powders were unsuccessful, which points to the fact that unsupported gold is a catalyst for organic synthesis and this metal is again unique, because others do not work.

Some works from Echavarren et al.15 refer to the high catalytic efficiency of gold, because it is known that Au^I complexes are the most active catalysts for the alkoxy-/hydroxycyclization and skeletal rearrangement of enynes,15b,e with a remarkable performance in reactions of these alkyne compounds with hetero- and carbo-nucleophiles,15c,d namely, 1,5- and 1,6-enynes.15d It is also known that Au^I complexes bearing bulky phosphines, phosphites, or NHC ligands usually have better performances than do the most active platinum complexes reported so far.15c,e

Apart from the supported gold nanoparticles and powdered unsupported gold mentioned above, the importance of cationic gold for catalytic processes is also known.10, 16 The Lewis acidity character of cationic gold species, their ability to selectively activate π systems toward the addition of a wide range of nucleophiles,17 and the relativistic and ligand effects explored by Toste et al.18 are crucial factors in homogeneous catalysis by gold.

The possibility of tuning a catalyst through ligand design is an interesting way to produce more effective gold compounds, just as some examples shown by Hashmi et al.,19 such as the case of Au^III complexes with pyridine derivatives, which showed satisfactory results in terms of activity, long-term stability, and product selectivity.19a The starting point of that work was the fact that AuCl₃ usually leads to good results with simple substrates, as a catalyst for phenol synthesis, but a loss of activity is seen with more complex structures.

In summary, there are recognized possibilities for homogeneous and heterogeneous catalyses. At present, the “gold rush” consists in the supported gold nanoparticles in the heterogeneous catalysis side and gold complexes in the homogeneous catalysis side,20 as shown in Figure 1.

Figure 1. Current state of catalysis by gold (black) and a less explored combination (gray).

Heterogenization

If heterogeneous catalysts are used in the gas or liquid phase, the separation of the reaction products from the catalyst, the possibility of its recycling, and the potential application to continuous flow processes are facilitated as compared with homogeneous systems. On the contrary, complexes used in homogeneous catalysis have remarkable properties such as high activity, enantioselectivity, and well-characterized structures, not found in heterogeneous systems. The heterogenization of complexes is thus an attractive and promising research area that can combine the advantages of homogeneous and heterogeneous catalyses (Figure 1). Naturally, the combination of these “two worlds” has some “gray” (unexplored) areas, such as the limited knowledge of the role of the support on the process, namely its influence on the catalytic activity due, for example, to the dispersion of the metal. Nevertheless, could the best of these two worlds be combined?

As early as 1998, it was shown that the use of heterogenized metal (Mn) complexes had similar or even higher activity than the respective homogeneous counterparts.21 In fact, heterogenization is currently an important area of research in organometallics, and so heterogeneous catalysts can be produced.22 Complexes of metals such as chromium,23 cobalt,24 copper,24, 25 iron,24a, 26 manganese,21, 24b, 26a, 27 molybdenum,27f, 28 nickel,27a, 29 rhodium,30 ruthenium,31 palladium,29d, 32 or platinum33 have already been successfully heterogenized. However, comparatively, examples of heterogenization of gold complexes are not very abundant. Nevertheless, due to the current importance of both homogeneous and heterogeneous catalyses by gold, heterogenization of gold complexes is an important topic that deserves to be reviewed because, to the best of our knowledge, no such attempt has ever been made so far. The efforts made by several authors on the subject are discussed below.

Heterogenization of gold catalysts

Studies with [Au(CH₃)₂(C₅H₇O₂)]

Several publications describe the work of Gates and co-workers, which deal with the [Au(CH₃)₂(C₅H₇O₂)] complex (CAu0) used as a “gold carrier” (see Figure 2). CAu0 consists in one atom of gold (Au^III), two methyl groups (CH₃), and an acetylacetonate moiety (acac, C₅H₇O₂).

Figure 2. Gold complex used in several works of Gates and co-workers. Adapted from Ref. [34].

It has been shown that some supported transition-metal complexes, analogues of the homogeneous counterparts, can establish bonds with the support that are strong enough to maintain the anchoring during catalytic reactions.35 Guzman and Gates described, in 2003, the preparation and characterization of the first structurally simple magnesium oxide (MgO)-supported gold complex.35 Later, in 2004, the same authors reported the use of this anchored complex on the same support for ethylene hydrogenation at 353 K.36 Although the initial complex was CAu0 (Figure 2), gold could be found on the magnesium oxide surface in the [Au(CH₃)₂(OMg)₂] form, the (OMg)₂ ligands being obtained from the support.35, 37 The existence of Au(CH₃)₂-supported species was confirmed by using infrared spectroscopy (IR) data, which also proved the occurrence of C₅H₇O₂ dissociation from the original compound during the adsorption process, being exchanged with oxygen from the support, and hydrogen bonding of surface OH groups to the organic moiety, which facilitates the formation of the anchored complex.35, 37 The X-ray absorption near-edge structure (XANES) data showed that gold remained in the oxidation state +3 (Au^III) after chemisorption. The extended X-ray absorption fine structure (EXAFS) data showed that the gold central atom was bonded to two oxygen atoms, presumably from the support, because the Au–O distance was within the typical values found for metal–O distances in other mononuclear metal complexes, and the bonding of two carbon atoms to the gold center (Au–C) was also confirmed.35 On the contrary, EXAFS showed no evidence of the existence of Au–Au bonds, which suggests the presence of fully isolated anchored complexes.35 However, it is possible that with different treatments, and also maybe during catalytic processes, supported gold can change its oxidation state from Au^III to a mixture of Au^I and Au⁰.38

Following catalytic activity tests for ethane hydrogenation (in which the catalyst was found to be stable during 12 h of measurements), after 30 min of steady state the essential data from XANES and EXAFS remained constant, with the exception of the Au–C contribution that disappeared, which shows that the methyl groups were removed before or during the catalytic reaction.35 Guzman and Gates confirmed with XANES, EXAFS, and IR spectroscopic data the stability of the Au^III species during this catalytic process and identified these complexes as the catalytically active species.36 In comparison with MgO-supported gold clusters of various sizes, the highest activity was observed for samples that contain only mononuclear gold complexes (Au–Au coordination number=0), which decreases with the increase in the Au–Au coordination number (clusters, Au–Au coordination number≥1) and consequently with the decrease in mononuclear gold complexes available.35

Guzman et al.37 confirmed and explored intensively the reaction of this metal acetylacetonate with an oxide surface by using the CAu0 and MgO models. These authors demonstrated that several situations related to the anchoring process or depending on reaction conditions or treatments can take place: 1) the C₅H₇O₂ moiety from the initial complex intervenes in a ligand-exchange process with the support, 2) the existence of hydrogen-bonding interactions between the surface Mg(C₅H₇O₂)₂ and H(C₅H₇O₂) species and the support OH groups, 3) the reactivity of CH₃ groups bonded to Au being removed as CH₄, 4) the decomposition of surface Mg(C₅H₇O₂)₂ and H(C₅H₇O₂) species by oxidation to CO₂ and traces of acetate ligands on the surface, and 5) structural and electronic changes of the supported gold.37 These results generically prove that during the heterogenization procedure, the anchoring process, the pretreatments, or the catalyzed reaction itself can change (permanently or not) the initial complex.

In another work, Guzman and Gates studied the different chemical species involved in the anchoring of CAu0 on γ-Al₂O₃.34 The data obtained showed that CAu0 reacted (at room temperature) with the OH surface groups and the aluminum sites available for coordination, which forms anchored mononuclear gold complexes and Al(C₅H₇O₂) species with Al from the support. The gold species predominant at the surface were isolated mononuclear complexes, as inferred from the absence of Au–Au bonding, confirmed through EXAFS experiments, and could be represented as (AlO)₂–Au(CH₃)₂, in which (AlO)₂ is the support.34 There was no evidence of Au–(C₅H₇O₂) interactions, which means that the gold complex was not adsorbed intact, but it underwent cleavage.34

The same authors tested the thermal decomposition (in helium) of ligands after the anchoring process.34 They found that the Al–(C₅H₇O₂) species were stable up to about 473 K, but other species (such as Al-acetate) were formed at higher temperatures. Also, at some point, the decomposition and removal of organic surface species were achieved, as shown by the results of the sample treatment at 573 K, which left no organic species on the surface. This type of data is important to show that despite a successful initial anchoring process at room temperature, the properties of the hybrid catalyst can change. Also, there are optimal or exclusive conditions, namely, temperature, for the reaction and the catalyst itself.

Another important change that can take place after treatments in helium at increasing temperatures, in spite of a perfect initial distribution of mononuclear gold complexes, is the formation of gold clusters with an average diameter of about 3 nm34 instead of primary forms of monodispersed species. The oxidation state of gold can change from Au^III to Au⁰ during the process.34 The same principle was described when the formation of gold nanoclusters was observed in zeolites with monodispersed gold after heating in flowing helium (these gold clusters were among the smallest reported in literature).39 It is also interesting to mention an important difference between the two zeolites used in that work: NaY (sodium Y zeolite) and DAY (dealuminated Y zeolite); the complex CAu0 was physisorbed on the former, but chemisorbed on the latter, with dissociation of the initial form of the complex to make the chemisorption possible.39 However, EXAFS studies with this complex and the NaY zeolite40 reported gold–support interface changes during CO oxidation; the Au–O distance changed from 2.08 (typical distance on the initial physisorbed complex) to 2.16 (typical Au–O distance, O being obtained from the zeolite support). The obtained data showed that the coordination of gold to the support could be tuned, which adjusts the composition of the reaction mixture.

CAu0 was also tested by Fierro-Gonzalez and Gates using the NaY zeolite as a support for the physisorbed complex.41 In this work, the oxidation of CO at 298 K and 760 Torr was the reaction model, these data being the first evidence of supported mononuclear cationic gold species as CO oxidation catalysts.41 The presence of Au^III on the initially prepared sample was inferred, but no Au^I or Au⁰ was found. Therefore, with the anchorage of the cationic gold species and the absence of zero valent gold being confirmed, data related to the influence of cationic gold in CO oxidation could be studied separately, unlike the case of heterogeneous catalysts containing supported gold nanoparticles (that might include cationic and zero valent gold).41 When the as-prepared catalyst was exposed to the reaction environment (CO, O₂, 760 Torr, 298 K, in He), the initial CO conversion rates fell to a considerably lower value and remained constant at that point for about 15 min of time on stream. EXAFS gave no evidence of Au–Au contributions, which means that the complexes remained isolated without the formation of gold clusters. However, XANES data showed that after the first 15 min, gold was at the Au^I state due to the reduction of Au^III by the CO+O₂ mixture. Au⁰ was not detected at any time in the process,41 unlike what was reported in another work, by using MgO as the support, which showed that CO played a dual role as a reactant and a reducing agent, converting Au^I to Au⁰.42 These reductions of gold seemed to be correlated with a decrease in the catalytic performance.

Using the same support, gold carbonyls, involving low-coordinated Au⁰ atoms—Au⁰(CO)₃ species, were formed.43 Mihaylov et al.44 also showed the formation of nonclassical gold carbonyls: Au^III(CO)₂, when Au³⁺ ions from complexes anchored to the NaY zeolite (such as those described above41), at temperatures higher than 220 K, migrated and formed Au^IIICO, which produced Au^III(CO)₂ at lower temperatures, in the presence of CO. This also suggests that the support has an important role because its structure is related to the existence (or not) of accessible Au^III to coordinate to the CO molecules.44 Moreover, the formation of gold carbonyls can be an important step to oxidize CO to CO_2.45

In another work of Fierro-Gonzalez et al.,45 the NaY zeolite was also used and it was demonstrated that the initial CH₃ ligands from the CAu0 complex can be released during the reaction. Steadily flowing or pulsed CO reacted in the active Au site, which produces CO₂ and CH₄, the latter probably being formed by the reaction of CH₃ groups, linked to the Au atom with H₂O in the stream. This type of data is very important to show that ligands of the original gold complex can be released not only during the anchorage process, as seen before, but also during the catalyzed reaction itself. Another important conclusion of the same study was that the type of CO stream can influence the change in the gold oxidation state: If CO is added in pulses, gold remained unreduced (Au^III), but in steadily flowing CO, the reduction of cationic gold (Au^III) to zero valent gold (Au⁰) was observed.45 In both cases, CO₂ was formed.

By using a different support—La₂O₃ nanoparticles—Fierro-Gonzalez et al.46 showed that mononuclear Au^III complexes can be highly active and stable CO oxidation catalysts at room temperature. These catalysts were found to be almost as active for CO oxidation at 298 K as the most active supported gold materials reported in the literature (up to 2005). Moreover, the prepared catalyst remained active up to more than 50 h of continuous operation. Like in other works mentioned above, the lack of Au–Au interactions in EXAFS spectra and the presence of two O atoms (from the support) bonded to each gold atom were observed. Mihaylov et al.47 also reported on the adsorption of CAu0 on La₂O₃, with gold being dispersed as mononuclear Au(CH₃)₂ complexes. It was shown that with increasing temperature, Au^III was reduced (to Au^I and Au⁰), while it aggregated and converted into nanoparticles.

Fierro-Gonzalez and Gates attempted to heterogenize the same CAu0 complex in TiO₂.48 Au(CH₃)₂ was supported through bonding with two O atoms from the support, as often happens with metal oxides. The EXAFS spectra showed that gold was initially dispersed as mononuclear complexes (Au^III); however, if the treatment implied a temperature increase, the previously nonexisting Au–Au contribution changed up to a coordination number of 5 at 390 K (but started to be higher than zero already at 335 K, which shows that the first Au–Au interactions were being formed at that temperature). The range of temperatures in which an increase in the gold coordination number was observed was similar to the range in which an increase in the CH₄ formation rate occurred. Both the reduction and the aggregation of supported gold occurred simultaneously with the release of linked CH₃ groups that reacted with molecular water available, which forms CH₄. This work effectively showed that the amount of cationic and zero valent gold can be tuned by experimental conditions, such as temperature.

In contrast to what was found by Gates and co-workers for many supports,16a,b, 34–39, 42, 46–49 Hisamoto et al.50 recently tested the adsorption of CAu0 on partially dehydroxylated silica (at 673 K), which shows that adsorption was possible and the organogold complex remained intact. Spectroscopic and computational results suggested that complexes were anchored through robust hydrogen bonds, namely, between surface silanol groups and the oxygen donor atoms of the acetylacetonate ligand.

However, not all supports are adequate for anchoring isolated mononuclear gold complexes. When the material commonly used by Gates and co-workers, CAu0, was used to generate isolated mononuclear gold complexes supported on cerium oxide (CeO₂),49b a low activity for CO oxidation was initially observed. Nevertheless, after a short period of operation, the catalytic activity increased due to the formation of gold clusters, which were more active than isolated mononuclear gold complexes in the CeO₂ support, as inferred from Gate and co-workers’49b and other49d works.

It is interesting to point out that only in 2009, the first images obtained by aberration-corrected scanning transmission electron microscopy were published, which showed isolated gold on a support, in the absence of gold clusters on MgO, a high area metal oxide support.49e

In a review of 2007, which deals with CO oxidation catalyzed by highly dispersed supported gold, Fierro-Gonzalez and Gates again stated that cationic gold, from isolated mononuclear gold cations on supports, is one of the “reaction channels” for this reaction.49c If under some conditions supported gold nanoparticles (zero valent gold) can be a good solution, under other conditions cationic gold can perform better. Therefore, all the common oxidation states of gold (Au⁰, Au^I, and Au^III) seem to be involved in CO oxidation catalysis.16b In another review, focused on supported cationic gold for CO oxidation, Fierro-Gonzalez et al.16a showed that cationic gold is involved in catalytic CO oxidation at room and higher temperatures, with or without the presence of detectable zero valent gold, under the form of nanoclusters.

The solutions presented above for the CAu0 complex are solid steps toward the possibility of obtaining highly dispersed gold on a support. They involve structures that are similar to the structure of the initial complex, or part of it, which, at the same time, can benefit from the advantages of having the gold catalyst in the heterogenized form. It can be noted that there are several tuning options for the catalysts and many factors that can change them. The analysis of these results is very important to avoid surprises that can compromise the initial goal of mimetizing some of the strongest points of homogeneous and heterogeneous catalyses.

Studies with Schiff bases

Corma and co-workers also published several works on the heterogenization of gold complexes that contain Au^III Schiff bases as described below. These complexes were active catalysts for homocoupling of aryl boronic acids,51 hydrogenations,32h, 51b,c, 52 hydrosilylations,53 hydroaminations,54 among others.

An example of heterogenization was presented by Gonzalez-Arellano et al., in which Au^III complexes supported on the mesoporous silica and laminar zeolitic material were used for the homocoupling of aryl boronic acids.51a,c Among the synthesized materials, the CAu1 complex, shown in Figure 3, gave better yields for several reagents/products when anchored on the organically modified silica-based supports (94–97 %) compared with when used as a homogeneous catalyst (80–93 %). Conditions used were as follows: aryl boronic acid 10 mmol, gold catalyst 0.3 equiv, and K₂CO₃ 20 mmol in xylene at 130 °C. Yields were determined from isolated samples of the reaction after 24 h.51a The Au^III complexes were also very active hydrogenation catalysts and their activity increased when supported on MCM-41.51c

Figure 3. Au^III Schiff base complexes synthesized by Corma and co-workers. Adapted from Refs. [51a,c, 52–54]. CAu homogeneous: R=CH₃; CAu heterogenized: R=CH₂- (linkage to the support); CAu3a: X=2-naphthyl; CAu3b: X=phenyl.

Later in 2007, Corma et al.53 used this type of gold complexes (CAu2 and CAu3a, in Figure 3) for hydrosilylations. Schiff base Au^III complexes anchored on MCM-41 have shown activity and chemoselectivity and could be recycled without decrease in activity or metal leaching.53

Other supports, namely, CeO₂ and Y₂O₃, were tested for the anchorage of Au^III species by Carrettin et al.51b These nanocrystalline materials stabilized gold in the Au^III state, which is active for the oxidation of CO and CC bond formation through the homocoupling of arylboronic acids. To evaluate the importance of Au^III as the catalytically active species, gold was deposited on nanoparticles of CeO₂ and Y₂O₃ by stirring the support slurries in a solution of HAuCl₄, for which the active surface species was considered to be Au^III. Although not a “classic heterogenization,” this procedure confirmed Au^III as the catalytic active species because it was found that the catalytic activity (for CO oxidation and homocoupling of arylboronic acids) of Au/CeO₂ was directly proportional to the Au^III surface species concentration.51b It was also known that homogeneous and immobilized Au^III Schiff base complexes (such as the CAu1 complex in Figure 3) were also very active and selective for the CC bond formation and hydrogenation of olefins and imines. This work also effectively established the strong activity of these materials when heterogenized on an ITQ-2-delaminated zeolitic material in the reaction of homocoupling of aryl boronic acids.

The reaction of homocoupling of aryl boronic acids is one of the most common reactions efficiently catalyzed by gold. A general (simplified) scheme is shown in Scheme 1, together with an example (reaction carried out under Suzuki reaction conditions).51 It was shown that Au^III homogeneous or heterogenized complexes catalyze the homocoupling of arylboronic acids or alkynes to produce symmetrical biaryls, whereas the analogue Au^I complex catalyzes the corresponding cross-coupling reaction.51c A general cross-coupling reaction is shown in Scheme 2.

Scheme 1. Homocoupling of aryl boronic acids (top), with an example (bottom). Adapted from Ref. [51].

Scheme 2. Cross-coupling of aryl boronic acids. Adapted from Refs. [51, 55].

Comas-Vives et al.52 showed that some types of supports can increase the activity of the homogeneous Au^III catalysts after anchorage. The complexes used by these authors were also Au^III Schiff bases (CAu1 and CAu3b, in Figure 3). The properties responsible for enhancing the performance were the polarity and proton-donating ability.52 Catalytic activities were compared between gold complexes, unsupported and supported on different materials, for hydrogenation reactions of diethyl ethylidene succinates under mild conditions. Table 1 shows the TOF results for the two different complexes (CAu1 and CAu3b), unsupported or silica supported (MCM-41, indicated here as “Silica”), zeolite (ITQ-2, indicated here simply as “Zeolite”), or silica enriched with surface protons (MCM-41 Si/Al=50, indicated here as “Silica H⁺”) for the catalytic hydrogenation of diethyl itaconate in ethanol under mild conditions.

Table 1. TOFs for two different gold complexes, homogeneous and anchored on different supports.^[a]

Complex	TOF [h^−1]
[a] Adapted from Refs. [32h, 52].
CAu1	3690
CAu1-Silica	10 520
CAu1-Zeolite	8980
CAu3b	3430
CAu3b-Silica	4920
CAu3b-SilicaH^+	6730

These results confirmed the influence of the support in the catalytic performance. It was shown that anchoring Au^III complexes on a high surface polarity support increased the catalytic activity, and this effect was even more pronounced if the surface contained Brønsted acid sites.52 Both the zeolite52 and the two types of silica32h, 52 showed to be suitable supports for heterogenizing metal–complex homogeneous catalysts. These materials maintained their catalytic performance after storage for 6 months at room temperature and could be recycled at least six times before any appreciable loss of activity.32h, 52

To find out what concerns the acid behavior of the supports, Corma et al.54 compared gold homogeneous catalysts (such as CAu2 and CAu3a, in Figure 3) with the respective gold complexes anchored on 1) pure silicate as a nonacidic model (MCM-41), 2) a tin silicate as a Lewis acid model (MCM-41 Sn), and 3) an aluminosilicate as a Brønsted acid model (MCM-41 Si/Al≈15). It was shown that both the homogeneous catalysts and the complexes heterogenized on the pure silicate were active alkyne and alkene hydroamination catalysts, but only after the addition of an acid promoter (NH₄PF₆) to the solution.54 Both heterogenized complexes with acidic properties showed activity as catalysts; however, the material that contains Brønsted acid sites rapidly deactivated in contrast with the catalyst with Lewis acid sites, which could be recycled at least four times without any loss of activity of selectivity. The authors concluded that from a chemical point of view, it would be of interest to avoid the use of NH₄PF₆ or any soluble acid promoter in solution, which is achieved by heterogenizing the complex.54

Parida et al. also synthesized a Schiff base complex of Au^III and anchored it on a MCM-41 silica support, as shown in Figure 4.51d A comparison of thermogravimetric analysis data from free and anchored complexes was made. It was shown that the decomposition of the latter is complete at 1173 K, with less defined steps of weight loss, compared with the free complex, where the main mass losses, corresponding to the splitting of the Schiff base molecules, are better defined and take place in the 473–973 K range. These results show mutual stabilization of propylamino groups (linkage to the support) and complexes and also the entrapment of complex moieties in the support pores.51d The catalytic activity for Suzuki coupling between X-Ph (X=Br, I) and aryl boronic acids originated moderate to excellent yields and good selectivity toward the homocoupling product (in comparison to Pd^II complexes that were highly selective toward cross-coupling reactions).51d From the recovery and reutilization point of view, the catalyst was removed by filtration from the reaction medium and several facts were observed: 1) no reaction took place after filtration, 2) no catalytic activity could be attributed to filtered solution portions when added to a new reaction mixture, 3) the catalyst showed almost no decrease in the catalytic activity at least after the fourth recycling process, and 4) no changes in the metal content of the recycled catalyst were found, which revealed no leaching from the support.

Figure 4. A Schiff base complex synthesized by Parida et al. Adapted from Ref. [51d].

The application of porous metal organic frameworks (MOFs) was considered by Zhang et al. as an intermediate solution between the use of homogeneous and heterogeneous gold catalysts.20 This type of structure represents a good dispersion of supported gold; however, the fact that this metal is, at the same time, the catalyst and a part of the structure can lead to limitations in the establishment of interactions with the reactants. The solution tested by these authors consisted of a postsynthesis modification of porous MOFs, which comprised covalent anchoring to a Au^III Schiff base complex to the pore walls,20 as shown in Figure 5, in contrast to MOFs with gold originally integrating the material itself.56

Figure 5. A Schiff base complex anchored on an MOF. Adapted from Ref. [20].

This hybrid catalyst had a much higher performance in catalyzed domino coupling and cyclization of N-protected ethylanilines, amine, and aldehyde as compared with oxide-supported gold and homogeneous gold complexes. It was also demonstrated that these materials are very active and selective for the hydrogenation of 1,3-butadiene.

More recently, the same principle was used by Corma et al.57 to catalyze alkene cyclopropanation reactions, such as the one schematized in Scheme 3. The goal was, as usual, to emulate the catalytic properties of the respective homogeneous complexes to promote the catalytic transfer of carbine fragments from diazo compounds, such as ethyl diazoacetate (EDA) or 2-phenyldiazoacetate (PhEDA), to olefins and obtain cyclopropane rings.

Scheme 3. Cyclopropanation reaction of styrene with a diazo compound. EDA for R=H; PhEDA for R=PhEDA. Adapted from Ref. [57].

The use of MOFs with the anchored gold complexes allowed moderate conversions (up to 42 %) for the cyclopropanation of styrene with EDA to be obtained in contrast to the use of cationic gold complexes only, with which no reaction occurred. In fact, this prepared gold catalyst was also active in other cyclopropanation reactions of several alkenes with EDA, although a copper-containing MOF was more active in some cases. A protocol was developed with cheap, reusable, easily recoverable, and relatively high performance catalysts.57 This hybrid material was also used by Kovtunov et al.58 to successfully describe, for the first time, parahydrogen-induced polarization of nuclear spins (in the products of heterogeneous gas-phase hydrogenations) carried out by a gold catalyst.

These examples on the use of Schiff base complexes are “less destructive” from the point of view of the initial complex integrity. Using this procedure for dispersing gold on a support, the electronic environment provided by the ligands can be better preserved, and better comparisons with the homogeneous analogues can be done.

Studies with other types of complexes

Corma et al.59 used mononuclear asymmetrical N-heterocyclic carbene–gold complexes (CAu6, in Figure 6) and also immobilized them on silica (gel and mesoporous MCM-41) and delaminated zeolite (ITQ-2). When the catalytic activity for hydrogenations was compared, again higher TOFs were found for supported complexes as compared with those for homogeneous analogues.59 These authors also focused on two important features of heterogenized catalysts: recovery and recycling. The possibility of recovering the catalyst as a stable species from the reaction mixture by an easy step of filtration and the possibility of reusing it are two of the most important points in heterogenization. To confirm this, the heterogenized catalyst was filtered and washed after the catalytic reaction, and then a fresh substrate was added without any further addition of the catalyst for several and consecutive experiments, and it was found that both yield and activity were retained. The filtrate was tested in a new reaction, and no catalyzed hydrogenation occurred.

Figure 6. N-heterocyclic carbene–gold complexes synthesized by Corma et al. CAu6a: R=H; CAu6b: R=(linkage to the support). Adapted from Ref. [59].

The heterogenization process can be advantageous or not, which depends on the enantioselectivity or activity point of view. One example is the work of del Pozo et al.,60 in which the activity (calculated by the determination of TOF) and the enantioselectivity (calculated by the determination of enantiomeric excess) were compared for homogeneous and heterogenized gold complexes. The compounds had ONN-tridentate pincer-type ligands (CAu7, in Figure 7). Palladium analogues were also used in this comparison, as shown in Table 2, for the asymmetric hydrogenation of diethyl 2-benzylidene succinate.60 Although all four catalysts were able to catalyze the complete conversion of substrates, it was shown that 1) the homogeneous catalysts were more active than the corresponding heterogenized analogues, but 2) the heterogenized complexes were more enantioselective than the homogeneous complexes, and 3) the most active and the most enantioselective samples had gold in their composition instead of palladium.60 For the supported gold catalyst (CAu7b), it was shown that no metal leaching from the silica surface occurred and that both the yield and the enantioselectivity showed high levels over up to at least four cycles.

Figure 7. Gold complexes synthesized by del Pozo et al. CAu7a: R¹=A; CAu7b: R¹=C; CAu8a: R²=B, R³=D; CAu8b: R²=C, R³=D; CAu8c: R²=B, R³=E; CAu8d: R²=C, R³=E. Adapted from Refs. [60, 61].

Table 2. TOFs and enantiomeric excesses (ee) for gold and palladium homogeneous and heterogenized complexes.^[a]

Complex	TOF [h^−1]	ee [%]
[a] Adapted from Ref. [60].
CAu (CAu7a)	34 800	80
CPd	33 900	15
CAu-Silica (CAu7b)	9960	98
CPd-Silica	4680	30

del Pozo et al.61 also tested CNN-tridentate pincer-type ligands (CAu8, in Figure 7) in a similar work. Using the same support, ordered mesoporous silica (MCM-41), the process of immobilization preserved the integrity of the organometallic and inorganic components in the final catalyst. The integrity of this hybrid catalyst persisted after consecutive cycles, which shows good activity and recycling capacity and combines the advantages of homogeneous and heterogeneous catalyses.61

In another work, Corma et al.62 explored, with success, the hydrothiolation of alkynes (Scheme 4) and electron-deficient olefins (Scheme 5), with heterogenized Schiff base Au^III complexes and N-heterocyclic carbine Au^I complexes (CAu1 and CAu2, in Figure 3; CuA5, in Figure 5), using silica MCM-41 as support. The reactions were performed in dry toluene at 313 K with 0.1 mol % of the catalyst. Up to three catalytic cycles were performed for heterogenized complexes.

Scheme 4. Hydrothiolation (with thiophenol) of alkynes. R=Ph, n-C₈H₁₇, or C(CH₃)₂OH. Adapted from Ref. [62].

Scheme 5. Hydrothiolation (with thiophenol) of electron-deficient olefins. R=H, Ph, or CH(OMe)O. R′=Me, OMe, Ph, O-nBu, or CH(OMe)O. Adapted from Ref. [62].

Again, the possibility of recycling the catalyst after several runs was proved by using a simple process of filtration, without any loss of efficiency. ICP measurements were taken to discard leaching, and TEM images of the catalyst before and after its utilization were obtained to evaluate the state of gold in the catalyst and no considerable changes were found.62

Ganai et al.63 also explored the silanol group chemistry of silica nanoparticles to support Au^III by using the 1,2,3-triazole linkage, as shown in Figure 8. The initial objective was achieved, and the catalytic efficiency was evaluated for the activation of the CC π bond of the alkynyl group in the Hashmi phenol synthesis (described in detail elsewhere64) with success. The initial development of these pyridine–carboxylate complexes for homogeneous catalysis was described by Hashmi et al. in 2004.19a

Figure 8. Gold complex anchored by a triazole linkage. CAu9a: X=N; CAu9b: X=CH. Adapted from Ref. [63].

Mallissery et al.65 used phosphinines Au^I complexes, shown in Figure 9, to study the possibility of immobilization of these complexes on TiO₂ or silica. These phosphabenzene compounds had been commonly used before in diverse catalytic reactions, but only in homogeneous catalysis, with no attempts to immobilize them on supports, which was done by these authors for the first time. The immobilization on TiO₂ resulted in complete degradation of the phosphine moiety, and in a mixture of phosphorous surface compounds when immobilized on chloropropyl-modified hexagonal mesoporous silica.65 This shows the need of successful immobilization protocols to avoid the destruction of important regions of the complex if the intention is to preserve them.

Figure 9. Phosphinines Au^I complexes synthesized by Mallisery et al. CAu10: R¹, R², R³, R⁴⇒CAu10a: OH, OH, H, H; CAu10b: H, H, OH, OH; CAu10c: OH, H, H, H; CAu10d: OCH₃, OCH₃, H, H. Adapted from Ref. [65].

The heterogenization approaches share some factors of the “pure” heterogeneous and homogeneous catalyses. Usually for heterogenized gold complexes, the “nanoenvironment” surrounding the gold atom comprises not only the support, but also the organic structure, which consists of the ligands of the chosen complex. Obviously, the electronic availability of the organometallic complex might influence the reactions.

With the exception of this last case, the complexes and heterogenization solutions described in this section are similar to the Schiff base materials, in which a nondestructive anchoring can take place, but it does not imply a physisorption process. This shows that the chemisorption is possible without complex cleavage through the establishment of strong covalent bonds in contrast to weaker interactions such as hydrogen bonds.

Summary

This Review summarizes the attempts made so far by several authors on the heterogenization of gold complexes onto solid porous materials. It mainly deals with the [Au(CH₃)₂(C₅H₇O₂)] complex used by Gates and co-workers, Schiff bases used by Corma et al., and other complexes used by different authors. In most cases, a nondestructive anchorage process took place. The advantages of heterogenization were proved, such as the easiness of the separation/recovery process and the possibility of recycling without loss of activity. Moreover, in general, higher activity rates for the heterogenized complexes were obtained as compared with those of the homogeneous analogues.

Furthermore, the merging of homogeneous and heterogeneous gold catalyses does not only concern concepts and ligands, because there is also an increasing overlap regarding analytical methods used. Recently, EXAFS and XANES spectra of homogeneous gold catalysis in operation have been reported to show that oxidative esterification of aldehydes occurred in a reaction medium in which only mononuclear gold species could be found, totally excluding the presence of gold nanoparticles.66 This was the first report on the use of these methods in homogeneous gold catalysis as they had been employed before only in heterogeneous systems, which shows the increasing synergy between these “two worlds” of the “gold catalysis universe.”

Outlook

Heterogenization is a mimetic way to facilitate contacts between the reactants and the catalytic sites, which probably is the most important factor for the success of heterogeneous gold catalysis, also combining the stronger points of the homogeneous catalysis, such as easy separation and full recycling of a precious metal such as gold. These hybrid materials, which join a huge variety of complexes and several available supports, result in a world of yet new possibilities that can be a solution for new catalytic processes or optimization of existing others. Where limitations were found earlier, new perspectives are now foreseen!

Acknowledgements

Authors are grateful to Fundação para a Ciência e a Tecnologia (FCT) for financial support (CIENCIA 2007 program for SACC and project PTDC/QUI-QUI/100682/2008, financed by FCT and FEDER in the context of program COMPETE).

Biographical Information

Miguel Peixoto de Almeida received his BSc and MSc degrees from the University of Porto, Portugal, in 2007 and 2009, respectively. He is currently a research fellow working on the heterogenization of gold complexes in carbon-based materials.

Biographical Information

Dr. Sónia A. C. Carabineiro received her PhD degree in catalysis in 2001 from the New University of Lisbon, Portugal. After two postdocs from Leiden University, the Netherlands, and the Technical University of Lisbon, she joined the University of Porto in 2007 as an assistant researcher. She has been working on catalysis by gold and more recently also on heterogenization of gold complexes.