Mechanochemical Synthesis of PdO2 Nanoparticles Immobilized over Silica Gel for Catalytic Suzuki–Miyaura Cross-Coupling Reactions Leading to the C-3 Modification of 1H-Indazole with Phenylboronic Acids

The C-3 modification of 1H-indazole has produced active pharmaceuticals for the treatment of cancer and HIV. But, so far, this transformation has seemed less available, due to the lack of efficient C-C bond formation at the less reactive C-3 position. In this work, a series of silica gel-supported PdO2 nanoparticles of 25–66 nm size were prepared by ball milling silica gel with divalent palladium precursors, and then employed as catalysts for the Suzuki–Miyaura cross-coupling of 1H-indazole derivative with phenylboronic acid. All the synthesized catalysts showed much higher cross-coupling yields than their palladium precursors, and could also be reused three times without losing high activity and selectivity in a toluene/water/ethanol mixed solvent. Although the palladium precursors showed an order of activity of PdCl2(dppf, 1,1′-bis(diphenylphosphino)ferrocene) > PdCl2(dtbpf, 1,1′-bis(di-tert-butylphosphino)ferrocene) > Pd(OAc, acetate)2, the synthesized catalysts showed an order of C1 (from Pd(OAc)2) > C3 (from PdCl2(dtbpf)) > C2 (from PdCl2(dppf)), which conformed to the orders of BET (Brunauer–Emmett–Teller) surface areas and acidities of these catalysts. Notably, the most inexpensive Pd(OAc)2 can be used as a palladium precursor for the synthesis of the best catalyst through simple ball milling. This work provides a highly active and inexpensive series of catalysts for C-3 modification of 1H-indazole, which are significant for the large-scale production of 1H-indazole-based pharmaceuticals.


Introduction
Indazoles refers to a group of bicyclic compounds containing an electron-rich pyrazole and a fused benzene ring [1], which can be regarded as nitrogen-substituted products of indole, but their studies have seemed much less extensive than indole for a long time [1].There is a ten-π electron aromatic heterocyclic structure on indazoles, naturally giving three tautomers such as 1H-, 2H-, and 3H-indazoles, due to the resonance structures of pyrazole nuclei of indazole (Figure 1a-c) [2].In practice, 1H-indazole was the most thermodynamically stable and abundant tautomer in both gas, aqueous, and metal-containing phases [3].Meanwhile, 1H-indazole derivatives have shown great medical values, including SE063 (HIV protease inhibitor, Figure 1d) [4], Lonidamine (anticancer agent, Figure 1e) [5], and YC-1 (inhibitor of platelet aggregation, Figure 1f) [6], which have aroused wide interests for many years.
Molecules 2023, 28, x FOR PEER REVIEW 2 of 24 including SE063 (HIV protease inhibitor, Figure 1d) [4], Lonidamine (anticancer agent, Figure 1e) [5], and YC-1 (inhibitor of platelet aggregation, Figure 1f) [6], which have aroused wide interests for many years.Although less stable and less common than the 1H-tautomer, 2H-indazole plays a key role in the synthesis of many highly useful pharmaceuticals for treatment of diseases like renal cell carcinoma, inflammation, as well as virus infection [3].On the other hand, probably due to the synthetic inconvenience of 3H-indazole or its derivatives, up to date there are few systematic explorations on pharmaceutical applications of 3H-indazole [7].
Therefore, synthetic organic chemists attempted to perform direct functionalization of 1H-indazole in order to improve both substrate and product diversities, particularly regarding the functionalization of the C-3 position, mainly due to the key pharmaceutical values of C-3 modified 1H-indazoles (Figure 1e-f).However, the C-H bond on the C-3 position of 1H-indazole was chemically inert compared to that on 2H-indazole [6,11]; metal-free transformation showed little effects [6], so most endeavors were focused on the design and use of metal catalysts for this conversion.For example, Kazzouli and co-workers reported the intermolecular C-H arylation of amino-protected 1H-indazole with (hetero)aryl bromide or iodide as a coupling partner, where palladium acetate in association with 1,10-phenanthroline was selected as catalyst, potassium carbonate was selected as a base, and dimethylacetamide was selected as solvent [11].Yamaguchi and co-workers also employed a catalytic combination of palladium chloride, 1,10-phenanthroline, silver carbonate, and potassium phosphate, along with dimethylacetamide for the direct C-H arylation of 1H-indazole with haloarene, eventually leading to a rapid synthesis of YC-1 (an antitumor agent, Figure 1f) [6].
Meanwhile, both Heck [12] and Suzuki-Miyaura cross-couplings [1] have been introduced to this transformation.The Heck reaction often showed poorer efficiency for C-3 functionalization, because the coupling partner demanded a tight (but sometimes irreversible) protection of the N-H group on 1H-indazole, otherwise unwanted by-products appeared [12].In comparison, the suitable coupling partners of Suzuki-Miyaura crosscoupling such as organoboronic acids just needed an easily removable protection of the N-H group, like that from di-tert-butyldicarbonate ((Boc)2O) [1].Furthermore, most Suzuki-Miyaura cross-couplings could be accomplished at mild temperature (lower than 80 Although less stable and less common than the 1H-tautomer, 2H-indazole plays a key role in the synthesis of many highly useful pharmaceuticals for treatment of diseases like renal cell carcinoma, inflammation, as well as virus infection [3].On the other hand, probably due to the synthetic inconvenience of 3H-indazole or its derivatives, up to date there are few systematic explorations on pharmaceutical applications of 3H-indazole [7]. Nowadays, attempts to synthesize and functionalize 1H-indazole attract wide and continuous attentions.These approaches can be divided into several types, including nitrosation of indole along with rearrangement [8], Staudinger-aza-Wittig reaction [9], as well as metal-catalyzed N-N bond formation [10].However, these profiles usually lack substrate and product diversity, obviously depressing the large-scale production of versatile 1H-indazole-based pharmaceuticals (Figure 1) [8][9][10].
Therefore, synthetic organic chemists attempted to perform direct functionalization of 1H-indazole in order to improve both substrate and product diversities, particularly regarding the functionalization of the C-3 position, mainly due to the key pharmaceutical values of C-3 modified 1H-indazoles (Figure 1e,f).However, the C-H bond on the C-3 position of 1H-indazole was chemically inert compared to that on 2H-indazole [6,11]; metal-free transformation showed little effects [6], so most endeavors were focused on the design and use of metal catalysts for this conversion.For example, Kazzouli and co-workers reported the intermolecular C-H arylation of amino-protected 1H-indazole with (hetero)aryl bromide or iodide as a coupling partner, where palladium acetate in association with 1,10-phenanthroline was selected as catalyst, potassium carbonate was selected as a base, and dimethylacetamide was selected as solvent [11].Yamaguchi and co-workers also employed a catalytic combination of palladium chloride, 1,10-phenanthroline, silver carbonate, and potassium phosphate, along with dimethylacetamide for the direct C-H arylation of 1H-indazole with haloarene, eventually leading to a rapid synthesis of YC-1 (an antitumor agent, Figure 1f) [6].
Meanwhile, both Heck [12] and Suzuki-Miyaura cross-couplings [1] have been introduced to this transformation.The Heck reaction often showed poorer efficiency for C-3 functionalization, because the coupling partner demanded a tight (but sometimes irreversible) protection of the N-H group on 1H-indazole, otherwise unwanted by-products appeared [12].In comparison, the suitable coupling partners of Suzuki-Miyaura crosscoupling such as organoboronic acids just needed an easily removable protection of the N-H group, like that from di-tert-butyldicarbonate ((Boc) 2 O) [1].Furthermore, most Suzuki-Miyaura cross-couplings could be accomplished at mild temperature (lower than 80 • C) within a short time (3 h) [13], producing no toxic pollutants [14], deeming them comparable to Heck reactions [12].
In practice, however, there are still some drawbacks on conducting Suzuki-Miyaura cross-coupling reactions, including the high cost of palladium precursors (salts) [13], the lengthy and costly synthesis of palladium catalysts [14], as well as the lack of effective catalyst recovery [13,14].Therefore, the immobilization of palladium into support was carried out in order to simplify catalyst synthesis and improve catalyst recovery.The support materials covered polymer [15], polysiloxane [16], covalent organic frameworks (COFs) [17], metal-organic frameworks (MOFs) [18], as well as carbon materials [19].Overall, the structure of a palladium complex immobilized over support usually differs from its homogeneous counterpart, giving lower activity [14].But, in rare cases, the support may act as an active ligand, reshaping the steric volume, size, and porosity of palladium precursors and showing positive influence on catalytic outputs [16,20].
In addition to traditional catalytic and synthetic protocols, some currently developed approaches like mechanochemistry appear to be a real breakthrough for Suzuki-Miyaura cross-coupling, mainly regarding clean and energy-saving catalytic reactions [21], as well as mechanochemical preparation of catalysts, along with effective recycling [22].The mechanochemistry means that the chemical transformation, propelled by mechanical energies involving friction, compression, forging, extrusion, and milling as well [23], which is high enough to break, rebuild, or connect chemical bonds, shows opportunities for establishing new reactions or advanced materials [23].
On the one hand, mechanochemistry appears to be a highly attractive technology for achieving a variety of chemical conversions, which may boost brand-new, highly chemo-or stereo-selective, and much safer reactions, because mechanical energies successfully circumvent the drawbacks derived from classical powers [24].For instance, classical heating such as artificially raising and decreasing temperature would produce unwanted by-products, and heating a closed reactor increases the risk of explosion [23].Secondly, it was reported that 85% of chemicals applied in the pharmaceutical industry were solvents, but even if recovered, 20-50% of solvents were lost by evaporation [24].However, mechanochemical reactions need marginal solvents or none at all, which save a huge amount of solvent, leading to much greener and cheaper processes [24].
On the other hand, mechanochemistry contributes to the synthesis or formation of micro-or nano-sized catalysts, nanomaterials, or other submicroscopic materials with great porosity [23].In comparison, conventional synthetic profiles not only demand lengthy processes, heating, or hazardous and expensive reagents, but also delivered amorphous catalysts (materials) with less activity [23].Herein, ball milling, as a typical mechanochemical operation, accumulates kinetic, potential, and thermal energies, along with shear and friction forces in a very short time, and then leaves a large variety of defects (changes, dislocations, or polymetallic mixed phases) in the final material, naturally improving the reactivity of synthesized materials [24].
Moreover, ball milling is beneficial to the formation of nanosized chemical structures rather than to degrading them into amorphous ones; the porosity might be increased, and active components would be highly dispersed even on the single-atom scale.For instance, palladium nanoparticles were ever immobilized over single-or multi-walled carbon nanotubes by using palladium acetate as a precursor and carbon nanotubes as support, through ball-milling [25].This protocol was solventless, rapid, and did not need any reducing reagents or electric current.The nanosized morphology of carbon nanotubes was well maintained where palladium nanoparticles had a size of 1-3 nm [25].Additionally, the synthesized palladium catalysts accomplished Suzuki-Miyaura crosscoupling within 50 min, giving almost 100% yield, which is much better than classical palladium/carbon catalysts, indicating the palladium nanoparticles were fully dispersed and highly active [25].
Gupton and co-workers prepared a 3D palladium nanoparticle-nickel-graphenecarbon nanotube hybrid catalyst by using ball milling, and synthesized a catalyst which was 10 times more active than commercial palladium/carbon catalysts in Suzuki-Miyaura crosscoupling [26].More than that, ball milling was also used to construct single-atom palladium components immobilized over MOF (metal-organic framework), and the resulting catalyst showed extremely high activity (turnover frequency of 13,043 h −1 ), selectivity (>99%), and yield (>99%) for Suzuki-Miyaura cross-coupling of bromobenzene with phenylboronic acid [27].Iron oxide was also employed as support for immobilizing single-atom palladium components through ball milling, and the catalyst showed great activity too [22].
On the other hand, although immobilization of palladium into porous supports may create highly active catalysts, most immobilizations cannot avoid leaching of palladium catalyst into solution, and then inactive palladium black appears [28].Therefore, ionic liquids (ILs) were always used as liquid support for immobilizing palladium catalyst in order for better recycling [29].
This work aims to construct an efficient, costly, eco-friendly, and recyclable profile for C-3 functionalization of 1H-indazole.In practice, various palladium(II) precursors were immobilized into silica gel through ball milling.The catalysts obtained were then employed in the Suzuki-Miyaura cross-coupling of amino-protected 3-iodo-1H-indazole with phenylboronic acid.Catalyst recycling was also tested with or without ILs.This work may show values on the manufacture of 1H-indazole-based pharmaceuticals.

Elemental Composition and Chemical State on Surface of Catalyst
X-ray photoelectron spectroscopy (XPS) was employed first to detect the elemental composition and chemical state on the surfaces of the synthesized catalysts (depth of 0-3 nm).The synthesis of catalysts is shown in Scheme 1.The XPS survey scan is summarized in Figure 2, and the corresponding binding energies and atomic compositions are in Table 1.At first, C0 showed a much lower content of carbon than C1-C3 (Figure 2a vs. Figure 2b-d; C0 vs. C1-C3, C 1s column, Table 1), while C1-C3 showed a considerably lower content of silicon than C0 (C1-C3 vs. C0, Si 2p column, Table 1), and palladium was detected on C1-C3 more than C0 (Figure 2b-d vs. Figure 2a), indicating that silica gel contained marginal carbon species; meanwhile, palladium components were successfully immobilized into silica gel through ball milling and subsequent centrifugation (Scheme 1).
cross-coupling [26].More than that, ball milling was also used to construct single-atom palladium components immobilized over MOF (metal-organic framework), and the resulting catalyst showed extremely high activity (turnover frequency of 13,043 h −1 ), selectivity (>99%), and yield (>99%) for Suzuki-Miyaura cross-coupling of bromobenzene with phenylboronic acid [27].Iron oxide was also employed as support for immobilizing single-atom palladium components through ball milling, and the catalyst showed great activity too [22].
On the other hand, although immobilization of palladium into porous supports may create highly active catalysts, most immobilizations cannot avoid leaching of palladium catalyst into solution, and then inactive palladium black appears [28].Therefore, ionic liquids (ILs) were always used as liquid support for immobilizing palladium catalyst in order for better recycling [29].
This work aims to construct an efficient, costly, eco-friendly, and recyclable profile for C-3 functionalization of 1H-indazole.In practice, various palladium(II) precursors were immobilized into silica gel through ball milling.The catalysts obtained were then employed in the Suzuki-Miyaura cross-coupling of amino-protected 3-iodo-1H-indazole with phenylboronic acid.Catalyst recycling was also tested with or without ILs.This work may show values on the manufacture of 1H-indazole-based pharmaceuticals.

Elemental Composition and Chemical State on Surface of Catalyst
X-ray photoelectron spectroscopy (XPS) was employed first to detect the elemental composition and chemical state on the surfaces of the synthesized catalysts (depth of 0-3 nm).The synthesis of catalysts is shown in Scheme 1.The XPS survey scan is summarized in Figure 2, and the corresponding binding energies and atomic compositions are in Table 1.At first, C0 showed a much lower content of carbon than C1-C3 (Figure 2a vs. Figure 2b-d; C0 vs. C1-C3, C 1s column, Table 1), while C1-C3 showed a considerably lower content of silicon than C0 (C1-C3 vs. C0, Si 2p column, Table 1), and palladium was detected on C1-C3 more than C0 (Figure 2b-d vs. Figure 2a), indicating that silica gel contained marginal carbon species; meanwhile, palladium components were successfully immobilized into silica gel through ball milling and subsequent centrifugation (Scheme 1).Scheme 1. Synthesis of catalysts.a Binding energy (eV), along with atomic percentage (at%) in parentheses.b Not found.
Furthermore, the palladium content of C1 seemed much higher than those of C2-C3 (Figure 2b vs. Figure 2c-d; C1 vs. C2-C3, Pd 3d column, Table 1), probably owing to the structural difference in palladium(II) precursors, which may affect the attachment of silica gel with palladium during catalyst synthesis (Scheme 1).Additionally, although both C2 and C3 employed phosphorous-containing palladium precursors (Scheme 1), there was no phosphorous found on C2 (Figure 2c; C2, Pd 3d column, Table 1), probably indicating that the dppf ligand of the Pd(II) precursor was leached into the solution rather than adsorbed by the silica gel during the synthesis of C2 (Scheme 1).
It was important to test the chemical state of the elements on the catalyst surface in order to understand the active component during catalysis.First of all, there were two components which appeared at 342.6 and 337.3 eV on the Pd 3d region of C1 (Figure 3a), corresponding to the 3d3/2 and 3d5/2 photoelectrons of Pd 4+ , respectively [30,31].For the purpose of comparison, it was previously reported that Pd 0 showed the peaks of 3d3/2 and 3d5/2 photoelectrons at 341.3-340.9 and 335.8-335.4eV [30,31].The Pd 3d regions of C2 and C3 showed very similar peak contours along with binding energies compared to C1 (Figure 3b,c vs. Figure 3a).Therefore, the palladium species formed on C1-C3 should have tetravalence.a Binding energy (eV), along with atomic percentage (at%) in parentheses.b Not found.
Furthermore, the palladium content of C1 seemed much higher than those of C2-C3 (Figure 2b vs. Figure 2c,d; C1 vs. C2-C3, Pd 3d column, Table 1), probably owing to the structural difference in palladium(II) precursors, which may affect the attachment of silica gel with palladium during catalyst synthesis (Scheme 1).Additionally, although both C2 and C3 employed phosphorous-containing palladium precursors (Scheme 1), there was no phosphorous found on C2 (Figure 2c; C2, Pd 3d column, Table 1), probably indicating that the dppf ligand of the Pd(II) precursor was leached into the solution rather than adsorbed by the silica gel during the synthesis of C2 (Scheme 1).
It was important to test the chemical state of the elements on the catalyst surface in order to understand the active component during catalysis.First of all, there were two components which appeared at 342.6 and 337.3 eV on the Pd 3d region of C1 (Figure 3a), corresponding to the 3d 3/2 and 3d 5/2 photoelectrons of Pd 4+ , respectively [30,31].For the purpose of comparison, it was previously reported that Pd 0 showed the peaks of 3d 3/2 and 3d 5/2 photoelectrons at 341.3-340.9 and 335.8-335.4eV [30,31].The Pd 3d regions of C2 and C3 showed very similar peak contours along with binding energies compared to C1 (Figure 3b,c vs. Figure 3a).Therefore, the palladium species formed on C1-C3 should have tetravalence.In order to further understand the composition of the palladium species on C1-C3, wide-angle (2θ = 10°-80°) X-ray diffraction (XRD) was performed as shown in Figure 4.A series of diffractions occurred at 2θ = 35.15°,58.15°, and 67.15° (grey cubes, Figure 4b), probably corresponding to 101, 220, and 112 indices derived from PdO2 (palladium oxide, PDF No. .The diffractions of PdO2 may also be found in the wide-angle XRD of C2-C3 (grey cubes, Figure 4c,d vs. Figure 4b).Therefore, it seemed that PdO2 probably emerged as the metal-containing phase after ball milling, no matter which kind of divalent palladium precursor was used (Scheme 1).Herein, Pd 2+ coming from various divalent palladium precursors was hydrolyzed into Pd(OH)2 under ball milling, which was further oxidized and dehydrated into PdO2 under drying (Scheme 1).In order to further understand the composition of the palladium species on C1-C3, wide-angle (2θ = 10 • -80 • ) X-ray diffraction (XRD) was performed as shown in Figure 4.A series of diffractions occurred at 2θ = 35.15• , 58.15 • , and 67.15 • (grey cubes, Figure 4b), probably corresponding to 101, 220, and 112 indices derived from PdO 2 (palladium oxide, PDF No. .The diffractions of PdO 2 may also be found in the wide-angle XRD of C2-C3 (grey cubes, Figure 4c,d vs. Figure 4b).Therefore, it seemed that PdO 2 probably emerged as the metal-containing phase after ball milling, no matter which kind of divalent palladium precursor was used (Scheme 1).Herein, Pd 2+ coming from various divalent palladium precursors was hydrolyzed into Pd(OH) 2 under ball milling, which was further oxidized and dehydrated into PdO 2 under drying (Scheme 1).In order to further understand the composition of the palladium species on C1-C3, wide-angle (2θ = 10°-80°) X-ray diffraction (XRD) was performed as shown in Figure 4.A series of diffractions occurred at 2θ = 35.15°,58.15°, and 67.15° (grey cubes, Figure 4b), probably corresponding to 101, 220, and 112 indices derived from PdO2 (palladium oxide, PDF No. .The diffractions of PdO2 may also be found in the wide-angle XRD of C2-C3 (grey cubes, Figure 4c,d vs. Figure 4b).Therefore, it seemed that PdO2 probably emerged as the metal-containing phase after ball milling, no matter which kind of divalent palladium precursor was used (Scheme 1).Herein, Pd 2+ coming from various divalent palladium precursors was hydrolyzed into Pd(OH)2 under ball milling, which was further oxidized and dehydrated into PdO2 under drying (Scheme 1).Next, the detection of carbon species on the synthesized catalysts shows additional information on immobilization.In fact, C1-C3 showed three very similar components on their C 1s regions, including the first one which appeared at 284.6-284.7 eV, the second at 285.7-285.9eV, as well as the last at 288.4-288.6 eV (Figure 5b-d), which can be ascribed to carbons stemming from the saturated hydrocarbon (sp 3 hybridization), C-O bond, and carboxyl group, respectively [32].These components may be derived from the organic ligands of the palladium precursors (Scheme 1).In comparison, however, C0 only showed two components that appeared at 285.4 and 287.2 eV, corresponding to a C-O bond and carboxyl group, respectively [32], actually representing the carbon residues of silica gel (C0, Scheme 1).
Molecules 2023, 28, x FOR PEER REVIEW 7 of 24 Next, the detection of carbon species on the synthesized catalysts shows additional information on immobilization.In fact, C1-C3 showed three very similar components on their C 1s regions, including the first one which appeared at 284.6-284.7 eV, the second at 285.7-285.9eV, as well as the last at 288.4-288.6 eV (Figure 5b-d), which can be ascribed to carbons stemming from the saturated hydrocarbon (sp 3 hybridization), C-O bond, and carboxyl group, respectively [32].These components may be derived from the organic ligands of the palladium precursors (Scheme 1).In comparison, however, C0 only showed two components that appeared at 285.4 and 287.2 eV, corresponding to a C-O bond and carboxyl group, respectively [32], actually representing the carbon residues of silica gel (C0, Scheme 1).Then, this aroused further interest to test other elements like silicon on the catalyst surface.In general, the binding energies of Si 2p photoelectrons can be found at 103.2-104.1 eV for C0-C3 (Figure S1a-d, Section S1, Supplementary Materials), which could be attributed to Si 4+ of the SiO2 phase [33].Accordingly, C0-C3 all showed a broad XRD band in the range of 2θ = 15°-35°, centered at 23° (Figure 4a), corresponding to the typical diffraction of the silicate backbone coming from C0 (Scheme 1) [34].
C2 showed a similar XRD contour to C1 (Figure 4c vs. Figure 4b), but the diffraction of grunerite might be detected (dark cube, Figure 4c; PDF No. 44-1401, Fe7Si8O22(OH)2), probably indicating that the dppf ligand was decomposed and transformed into Fe-Si mixed oxide during ball milling (Scheme 1).C3 showed a higher Si 2p photoelectron binding energy than C0-C2 (Figure S1d vs. Figure S1a-c), indicating that some new Si-containing phases appeared during immobilization (Scheme 1).On one hand, in addition to PdO2, the 002 diffraction of SiP may appear on the XRD of C3 (hollow cube, Figure 4d; PDF No. 29-1133, SiP).On the other hand, two peaks occurred at 129.2 and 122.8 eV on the P 2p region of C3 (Figure S2, Section S2, Supplementary Materials), probably corresponding to the 2p1/2 and 2p3/2 photoelectrons of P 4− [35], which were both lower than those of phosphorous with high valence coming from P-O or P=O bonds [36].Then, this aroused further interest to test other elements like silicon on the catalyst surface.In general, the binding energies of Si 2p photoelectrons can be found at 103.2-104.1 eV for C0-C3 (Figure S1a-d, Section S1, Supplementary Materials), which could be attributed to Si 4+ of the SiO 2 phase [33].Accordingly, C0-C3 all showed a broad XRD band in the range of 2θ = 15 • -35 • , centered at 23 • (Figure 4a), corresponding to the typical diffraction of the silicate backbone coming from C0 (Scheme 1) [34].
C2 showed a similar XRD contour to C1 (Figure 4c vs. Figure 4b), but the diffraction of grunerite might be detected (dark cube, Figure 4c; PDF No. 44-1401, Fe 7 Si 8 O 22 (OH) 2 ), probably indicating that the dppf ligand was decomposed and transformed into Fe-Si mixed oxide during ball milling (Scheme 1).C3 showed a higher Si 2p photoelectron binding energy than C0-C2 (Figure S1d vs. Figure S1a-c), indicating that some new Si-containing phases appeared during immobilization (Scheme 1).On one hand, in addition to PdO 2 , the 002 diffraction of SiP may appear on the XRD of C3 (hollow cube, Figure 4d; PDF No. 29-1133, SiP).On the other hand, two peaks occurred at 129.2 and 122.8 eV on the P 2p region of C3 (Figure S2, Section S2, Supplementary Materials), probably corresponding to the 2p 1/2 and 2p 3/2 photoelectrons of P 4− [35], which were both lower than those of phosphorous with high valence coming from P-O or P=O bonds [36].

Textural and Other Properties of Synthesized Catalyst
In order to further understand the difference in catalysts, the textural and other physicochemical properties of the synthesized samples were tested.At first, all the catalysts (C0-C3) showed typical type IV isotherms along with H3 type hysteresis loops (Figure 6a-d), clearly indicating they had mesoporous structures [37], which is further established by the pore size distributions (Figure 6a'-d') and high BET surface areas (Table 2).This result meant that ball milling would not destroy the mesoporous structure of silica gel during the immobilization of palladium (Scheme 1).

Textural and Other Properties of Synthesized Catalyst
In order to further understand the difference in catalysts, the textural and other physicochemical properties of the synthesized samples were tested.At first, all the catalysts (C0-C3) showed typical type IV isotherms along with H3 type hysteresis loops (Figure 6a-d), clearly indicating they had mesoporous structures [37], which is further established by the pore size distributions (Figure 6a'-d') and high BET surface areas (Table 2).This result meant that ball milling would not destroy the mesoporous structure of silica gel during the immobilization of palladium (Scheme 1).[39] using 101 diffractions of the PdO2 components for C1 (Figure 4b), C2 (Figure 4c), and C3 (Figure 4d).i Determined by n-butylamine titration (Section 3.2.).
C1-C3 showed a lower BET surface area, total pore volume, micropore volume, as well as higher bulk density than C0 (Table 2), indicating palladium was incorporated into both the mesopores and micropores of C0 during immobilization (Scheme 1).Furthermore, C1-C3 showed an order of BET surface areas of C1 > C3 > C2 (Table 2), proposing that the ligand of the palladium precursor played a key role in determining the adsorption amount of palladium over support (Scheme 1).Meanwhile, the order of acid amount was detected as C1 > C3 > C2 > C0 (Table 2), proposing that C0 had Lewis or Brönsted acidities by nature, but the immobilization of palladium would further increase the acidity, mainly because the palladium atom  ) size determined by XRD observation according to Scherrer's equation [39] using 101 diffractions of the PdO 2 components for C1 (Figure 4b), C2 (Figure 4c), and C3 (Figure 4d).i Determined by n-butylamine titration (Section 3.2).
C1-C3 showed a lower BET surface area, total pore volume, micropore volume, as well as higher bulk density than C0 (Table 2), indicating palladium was incorporated into both the mesopores and micropores of C0 during immobilization (Scheme 1).Furthermore, C1-C3 showed an order of BET surface areas of C1 > C3 > C2 (Table 2), proposing that the ligand of the palladium precursor played a key role in determining the adsorption amount of palladium over support (Scheme 1).Meanwhile, the order of acid amount was detected as C1 > C3 > C2 > C0 (Table 2), proposing that C0 had Lewis or Brönsted acidities by nature, but the immobilization of palladium would further increase the acidity, mainly because the palladium atom may provide more empty orbitals than silicon, and various ligands of palladium precursors showed different influences on product acidity (Scheme 1).

Functional Group and Thermal Stability of Synthesized Catalyst
FT-IR spectroscopy was carried out to detect the functional group of the synthesized catalysts.As shown in Figure S3 (Section S3, Supplementary Materials), all the tested samples (C0-C3) showed a broad band centered at 3450 cm −1 , representing the O-H stretching vibration of hydroxyl groups on the surface of silica gel (Figure S3a-d).Furthermore, the peaks that appeared at 1622 cm −1 on C0-C3 could be ascribed to the O-H bending vibration of hydroxyl groups (Figure S3a-d) [40].
Then, there were two following peaks which occurred at 1068 and 798 cm −1 (Figure S3a-d), corresponding to unsymmetric and symmetric Si-O stretching vibrations, respectively [41].Moreover, the contours around 1068 cm −1 for C0-C2 were much broader than that for C3 (Figure S3a-c vs. Figure S3d), probably indicating the backbone of C3 had been changed to a large extent, also in association with the XRD of C0-C2 vs. C3 (Figure 4d vs. Figure 4a-c).Additionally, both C3 and C2 showed the stretching vibration of the Pd-O bond at 492 cm −1 (Figure S3c,d), which was red-shifted slightly on C1 (Figure S3b), probably indicating that both dppf and dtbpf ligands had residues after ball milling, which affected the coordination environments of Pd 4+ (Scheme 1).
UV-Vis spectroscopy was employed to study the immobilization from another point of view.
Firstly, all the tested samples (C0-C3) showed adsorption at 284 nm (Figure S4a-d, Section S4, Supplementary Materials), corresponding to the charge transfer transition on the organic species of samples [42].C0-C2 showed no adsorptions above 325 nm (Figure S4a-c), but there were two bands centered at 362 and 450 nm on the UV-Vis spectrum of C3 (Figure S4d), corresponding to the ligand to metal charge transfer (LMCT) transition of the metal species [42] and the charge transfer transition of metals [42], respectively.This phenomenon indicated again that C3 had a quite different backbone than C0-C2, where a new Si-containing phase may appear, as shown in the XRD (Figure 4d) and FT-IR (Figure S3d).
It was also interesting to study the composition of the synthesized catalyst by using TGA.C0 showed a sharp weight loss of 5.62% at 35-150 • C (black line, Figure 7), proposing the release of adsorbed water.The following weight loss of 5.46% that occurred at 150-600 • C tended to be relatively flat (black line, Figure 7), indicating the decomposition of organic residues on C0.
Molecules 2023, 28, x FOR PEER REVIEW 9 of 24 may provide more empty orbitals than silicon, and various ligands of palladium precursors showed different influences on product acidity (Scheme 1).

Functional Group and Thermal Stability of Synthesized Catalyst
FT-IR spectroscopy was carried out to detect the functional group of the synthesized catalysts.As shown in Figure S3 (Section S3, Supplementary Materials), all the tested samples (C0-C3) showed a broad band centered at 3450 cm −1 , representing the O-H stretching vibration of hydroxyl groups on the surface of silica gel (Figure S3a-d).Furthermore, the peaks that appeared at 1622 cm −1 on C0-C3 could be ascribed to the O-H bending vibration of hydroxyl groups (Figure S3a-d) [40].
Then, there were two following peaks which occurred at 1068 and 798 cm −1 (Figure S3ad), corresponding to unsymmetric and symmetric Si-O stretching vibrations, respectively [41].Moreover, the contours around 1068 cm −1 for C0-C2 were much broader than that for C3 (Figure S3a-c vs. Figure S3d), probably indicating the backbone of C3 had been changed to a large extent, also in association with the XRD of C0-C2 vs. C3 (Figure 4d vs. Figure 4a-c).Additionally, both C3 and C2 showed the stretching vibration of the Pd-O bond at 492 cm −1 (Figure S3c,d), which was red-shifted slightly on C1 (Figure S3b), probably indicating that both dppf and dtbpf ligands had residues after ball milling, which affected the coordination environments of Pd 4+ (Scheme 1).
UV-Vis spectroscopy was employed to study the immobilization from another point of view.Firstly, all the tested samples (C0-C3) showed adsorption at 284 nm (Figure S4a-d, Section S4, Supplementary Materials), corresponding to the charge transfer transition on the organic species of samples [42].C0-C2 showed no adsorptions above 325 nm (Figure S4a-c), but there were two bands centered at 362 and 450 nm on the UV-Vis spectrum of C3 (Figure S4d), corresponding to the ligand to metal charge transfer (LMCT) transition of the metal species [42] and the charge transfer transition of metals [42], respectively.This phenomenon indicated again that C3 had a quite different backbone than C0-C2, where a new Si-containing phase may appear, as shown in the XRD (Figure 4d) and FT-IR (Figure S3d).
It was also interesting to study the composition of the synthesized catalyst by using TGA.C0 showed a sharp weight loss of 5.62% at 35-150 °C (black line, Figure 7), proposing the release of adsorbed water.The following weight loss of 5.46% that occurred at 150-600 °C tended to be relatively flat (black line, Figure 7), indicating the decomposition of organic residues on C0.After the immobilization of Pd(OAc)2 into silica gel (Scheme 1), the resulting C1 showed the first weight loss of 3.82% at 35-150 °C, along with the following one of 4.51% at 150-600 °C (red line, Figure 7).In comparison, C0 showed a higher total weight loss (11.08%) than C1 (8.33%) at 30-600 °C (black vs. red, Figure 7), obviously indicating that ball milling removed After the immobilization of Pd(OAc) 2 into silica gel (Scheme 1), the resulting C1 showed the first weight loss of 3.82% at 35-150 • C, along with the following one of 4.51% at 150-600 • C (red line, Figure 7).In comparison, C0 showed a higher total weight loss (11.08%) than C1 (8.33%) at 30-600 • C (black vs. red, Figure 7), obviously indicating that ball milling removed the adsorbed water and organic residues in silica gel, facilitating the immobilization of palladium (C1 vs. C0, Scheme 1).Furthermore, both C2 and C3 showed much lower total weight losses than C1 (green and blue vs. red, Figure 7), probably owing to their lower palladium contents compared to C1 (Table 1), reflecting the different effects of the palladium precursors used in the synthesis of C1-C3 (Scheme 1).Moreover, the TGA curve of C3 coincided with C2 at 30-200 • C but exhibited a much higher weight loss at 200-600 • C (blue vs. green, Figure 7), proposing that C3 contained the same amount of adsorbed water or volatile organic species as C2, while the less volatile components on C3 seemed more abundant than those on C2 (Scheme 1).

Morphology and Internal Structure of Synthesized Catalyst
C0 was commercial silica gel with size 300 mesh (about 8.47 µm on average).According to the SEM observation, C0 had a comparatively smooth and flat surface with a few wrinkles (Figure 8a,a'), reflecting the original morphology of silica gel (raw material, Scheme 1).
C3 (Scheme 1).Moreover, the TGA curve of C3 coincided with C2 at 30-200 °C but exhibited a much higher weight loss at 200-600 °C (blue vs. green, Figure 7), proposing that C3 contained the same amount of adsorbed water or volatile organic species as C2, while the less volatile components on C3 seemed more abundant than those on C2 (Scheme 1).

Morphology and Internal Structure of Synthesized Catalyst
C0 was commercial silica gel with size 300 mesh (about 8.47 µm on average).According to the SEM observation, C0 had a comparatively smooth and flat surface with a few wrinkles (Figure 8a,a'), reflecting the original morphology of silica gel (raw material, Scheme 1).
C1 turned out to be a mass of irregular particles of a size of 1-5 µm (Figure 8b,b'), obviously indicating that ball milling had at least halved the size of silica gel (C0) (Scheme 1).In association with the TEM observation, it can be seen that the much denser particles featuring a size of 10-20 nm could be ascribed to PdO2 (Figure 9a, Scheme 1).
C2 was composed of particles of a size of 1-3 µm, along with their agglomerates (Figure 8c,c').Simultaneously, C2 showed a similar internal structure compared to C1, including the sizes of both blocks and mesopores (Figure 9b vs. Figure 9a), which also conformed to those found through the N2 adsorption-desorption experiments (Table 2), probably suggesting the immobilization of PdCl2(dppf) would not degrade the backbone of silica gel as that of Pd(OAc)2 (Scheme 1).C1 turned out to be a mass of irregular particles of a size of 1-5 µm (Figure 8b,b'), obviously indicating that ball milling had at least halved the size of silica gel (C0) (Scheme 1).In association with the TEM observation, it can be seen that the much denser particles featuring a size of 10-20 nm could be ascribed to PdO 2 (Figure 9a, Scheme 1).
C2 was composed of particles of a size of 1-3 µm, along with their agglomerates (Figure 8c,c').Simultaneously, C2 showed a similar internal structure compared to C1, including the sizes of both blocks and mesopores (Figure 9b vs. Figure 9a), which also conformed to those found through the N 2 adsorption-desorption experiments (Table 2), probably suggesting the immobilization of PdCl 2 (dppf) would not degrade the backbone of silica gel as that of Pd(OAc) 2 (Scheme 1).
On the other hand, it was found that C1 had PdO2 particles featuring a size of 10-20 nm under TEM detection (dark dots, Figure 9a), and the calculated size was 25 nm on the basis of XRD (C1, dPdO2, Table 2), whose similarity approved the presence of PdO2 nanoparticles on C1 (Scheme 1).Furthermore, both C2 and C3 showed much larger PdO2 nanoparticle sizes than C1 according to XRD (dPdO2, Table 2); meanwhile, the PdO2 nanoparticles may have been dispersed into the denser blocks on their TEM images (Figure 9b,c).

Preparation of Iodinated 1H-indazole Substrate
In order to carry out the C-3 modification of 1H-indazole (compound 1) through catalytic Suzuki-Miyaura reaction, substrate 1 needed to be first iodinated almost stoichiometrically according to a previous method [43], whose amino group could be subsequently protected by di-tert-butyldicarbonate ((Boc)2O) [43].Herein, iodination can be conducted without affecting the amino group, but the amino group had to be protected Although C3 showed a similar morphology to C2 according to SEM (Figure 8d,d' vs. Figure 8c,c'), C3 exhibited a much looser internal structure than C2, particularly regarding the much larger pore size of C3 (Figure 9c vs. Figure 9b).Accordingly, it can be seen that C3 exhibited higher BET surface area and total pore volume, as well as lower bulk density than C2 (Table 2).Therefore, the ligand of the palladium precursor played an important role in controlling the product structure (Scheme 1).
With the data obtained so far, it seemed important to compare the sizes of both bulk and PdO 2 particles derived from different sources.On one hand, C0 showed a bulk particle size coming from BET surface area at 43 nm (C0, d S , Table 2), while that from SEM turned out to be 40 µm (C0, d SEM , Table 2; Figure 8a,a'), indicating that the particles of a size of 43 nm were the smallest units of N 2 adsorption and desorption, whose agglomeration actually constituted the much larger particles of a size of 40 µm.The same trends were also found in C1-C3 (d S and d SEM , Table 2).
On the other hand, it was found that C1 had PdO 2 particles featuring a size of 10-20 nm under TEM detection (dark dots, Figure 9a), and the calculated size was 25 nm on the basis of XRD (C1, d PdO2 , Table 2), whose similarity approved the presence of PdO 2 nanoparticles on C1 (Scheme 1).Furthermore, both C2 and C3 showed much larger PdO 2 nanoparticle sizes than C1 according to XRD (d PdO2 , Table 2); meanwhile, the PdO 2 nanoparticles may have been dispersed into the denser blocks on their TEM images (Figure 9b,c).

Catalytic Suzuki-Miyaura Cross-Coupling Reaction 2.5.1. Preparation of Iodinated 1H-indazole Substrate
In order to carry out the C-3 modification of 1H-indazole (compound 1) through catalytic Suzuki-Miyaura reaction, substrate 1 needed to be first iodinated almost stoichiometrically according to a previous method [43], whose amino group could be subsequently protected by di-tert-butyldicarbonate ((Boc) 2 O) [43].Herein, iodination can be conducted without affecting the amino group, but the amino group had to be protected before the cross-coupling, otherwise N-substituted products may appear [1,12].After the reaction, the Boc group can be removed in acid solution [43].

Reactions Using Two Substrates
According to Tables 3 and 4, the averaged conversion of entries 3-10 in Table 3 (60%) was lower than that of same entries in Table 4 (62%).
As for entries 11-15 in Tables 3 and 4, Table 3 (64%) still gave fewer outputs than Table 4 (69%).Additionally, the averaged conversion of entries 16-19 in Table 3 (66%) was also lower than that in Table 4 (72%).Therefore, 3-methoxycarbonylphenylboronic acid (compound 2, Table 3) seemed to be a less effective nucleophile than its para isomer (compound 4, Table 4) under various catalytic profiles.It was previously confirmed that methoxycarbonyl is an electron withdrawing group (EWG) [44].When methoxycarbonyl was ortho-, meta-, or para-oriented on phenylboronic acid, the homolytic bond dissociation enthalpy (BDE) of the C-B bond showed an order of ortho-< para-< meta-oriented, which was also the order of the C-B bond strength [44], meaning the cleavage of the C-B bond of compound 4 needed less energy than that of compound 3.
On the other hand, the palladium precursors showed an order of activity of PdCl 2 (dppf) > PdCl 2 (dtbpf) > Pd(OAc) 2 for transformations of both compounds 2 (entries 11 vs. 16 vs.3, Table 3) and 4 (entries 11 vs. 16 vs.3, Table 4), clearly indicating that the ligand of the palladium precursor showed key influences on the homogeneous catalytic efficiency.There were three steps in the Suzuki-Miyaura cross-coupling, including oxidative addition, transmetallation, and reductive elimination [13].The ligand of the palladium precursor played a key role in oxidative addition, directly affecting the catalytic outputs [13].
However, the activity order of the synthesized catalysts turned out to be C1 > C3 > C2 for the transformation of not only compounds 2 but also 4 (entries 4 vs. 17 vs. 2 in Tables 3 and 4).At first, the synthesized catalysts showed an order of BET surface areas of C1 > C3 > C2, while their bulk densities gave the opposite trend of C1 < C3 < C2 (Table 2).The porosity of the synthesized catalysts was directly controlled by the ligand of the palladium precursor that was used in ball milling (Scheme 1), which greatly affected the dispersion of the substrate into the catalytic center.
Secondly, the acid amounts of the synthesized catalysts exhibited an order of C1 > C3 > C2 (Table 2), indicating that ball milling of silica gel with Pd(OAc) 2 prepared the catalysts featuring the highest Lewis and Brönsted acidities; that those with PdCl 2 (dtbpf) did less so; while those with PdCl 2 (dppf) exhibited the lowest (Scheme 1; entries 4 vs. 17 vs. 2 in Tables 3 and 4).In synthesis, it was proposed that acetate anion (OAc − ) accelerated the hydrolysis of Pd 2+ into hydroxide, which was subsequently dehydrated to PdO 2 nanoparticle under ball milling.But in comparison, both dppf and dtbpf could not accelerate this process to the same extent (Scheme 1).
Furthermore, although the structures of dppf and dtbpf looked similar (Scheme 1), PdCl 2 (dppf) seemed better than PdCl 2 (dtbpf) in the catalytic conversion of compounds 2 and 4 (entries 11 vs. 16, Tables 3 and 4), meaning that dppf was able to stabilize the palladium intermediate [PhPdXLn] better than dtbpf in oxidative addition [13], probably because the phenyl group of dppf was much smaller than the tert-butyl of dtbpf, leaving less steric hindrance in the formation of [PhPdXLn] during oxidative addition [13].
But in heterogeneous catalysis, C3 coming from PdCl 2 (dtbpf) became more powerful than C2 from PdCl 2 (dppf) (entries 17 vs. 12, Tables 3 and 4).Therefore, ball milling actually changed the original palladium precursors into other catalytic components like PdO 2 nanoparticles (Scheme 1).In practice, dtbpf performed better than dppf during the oxidative hydrolysis of Pd 2+ and in dehydration under ball milling, which is supported by the higher BET surface area, total pore volume, and acid amount of C3 compared to those of C2 (Table 2).Additionally, C3 showed a different XRD pattern than C2 (Figure 4d vs. Figure 4c), where the phase of SiP may have appeared (Figure 4d and Figure S2), which may have decorated surface of the PdO 2 nanoparticle, leading to better catalytic outputs.
From another point of view, the yield stemming from the protocol of Table 4 catalyzed by C2 (entry 12, Table 4; yield of 88%) was much higher than that from the same reaction catalyzed by PdCl 2 (dppf) (yield of 65%) [43].Furthermore, C2 also showed a very close yield compared to the catalytic combination of PdCl 2 (dppf) with BMImPF 6 [45].These results clearly indicate that the ball milling of the divalent palladium(II) precursor with silica gel produces a highly active palladium catalyst.

Effects of Loading Amounts of Catalyst and Base
When the loading amount of C1 was halved from 4 mol% to 2 mol%, both the conversion and yield for the transformation of compound 2 were increased by 2% (entries 5 vs. 4, Table 3).If compound 2 was replaced with 4, the catalytic outputs including the conversion and yield were slightly decreased (entries 5 vs. 4, Table 4).Overall, the use of C1 (the heterogeneous catalyst with highest activity) was not very sensitive to the loading amount of the catalyst.
Herein, when C1 was reduced by half in the transformation of compound 2 (meta isomer), the agglomeration of C1 was weakened, and the dispersion of 2 into the internal pores of C1 became easier, leading to better outputs.If compound 2 was replaced with 4 (para isomer), the catalytic outputs were slightly decreased when the catalyst loading was halved, suggesting that the cross-coupling of 4 may occur on the surface of C1 more than in its internal pores, simultaneously indicating the textural structure of C1 preferred the meta isomer (2, smaller) to the para one (4, larger), mainly owing to the size of the substrate.
The loading amount of the base (Na 2 CO 3 ) also affected the catalytic outputs significantly.In practice, when the loading of Na 2 CO 3 was decreased by half, the catalytic outputs including the conversion and yield were degraded remarkably, not only for the transformation of the meta isomer (2, entries 6 vs. 4, Table 3), but also for that of the para isomer (4, entries 6 vs. 4, Table 4).According to the proposed mechanism [13], Na 2 CO 3 was dissolved in water and released OH − , and the concentration of OH − affected the appearance of the [PhPdOHLn] intermediate in transmetallation, involving the forming rate and concentration.Clearly, a high concentration of OH − in the catalytic solution contributed to the catalytic outputs.

Effect of Temperature
When the temperature was decreased gradually, the cross-coupling of compound 2 catalyzed by C1 decreased the conversions and yields sharply (entries 4 vs. 9 vs. 10, Table 3).The same tendency was also observed in the cross-coupling of compound 4 (entries 4 vs. 9 vs. 10, Table 4).Although much of the literature has reported that catalytic Suzuki-Miyaura cross-coupling reactions can be carried out at mild temperatures [13], C1 still needed 80 • C to guarantee high catalytic outputs, no matter which kind of phenyl substrate was used, supposing a higher temperature would facilitate the formation of an active intermediate during catalysis.

Effect of Ionic Liquid
In order to utilize the many great properties of IL, two imidazolium ILs such as BMImX (BMIm + = 1-n-butyl-3-methylimidazolium, X − = BF 4 − , PF 6 − ) were employed as co-solvents.However, the combination of BMImBF 4 with THE (toluene/H 2 O/EtOH, Tables 3 and 4) did not show any positive effects on the transformation of both compounds 2 and 4 (entry 2 for Tables 3 and 4), indicating that the BMIm + ions were completely inactive in catalyzing the present cross-coupling.
Next, when C1 was selected as a catalyst, the introduction of either BMImBF 4 or BMImPF 6 did not improve the catalytic outputs for the conversions of both compounds 2 and 4, and even depressed the reactions (entries 7-8 vs. 4, Tables 3 and 4).Moreover, neither BMImBF 4 nor BMImPF 6 showed positive effects on the reactions of compounds 2 and 4 catalyzed by C2 (entries 14-15 vs. 12, Tables 3 and 4) and C3 (entries 18-19 vs. 17, Tables 3 and 4).
It has been widely reported that the immobilization of palladium species into ionic liquid can provide more powerful catalysts for Suzuki-Miyaura cross-coupling, mainly because this kind of immobilization can increase the acidity of the palladium component, stabilize the active intermediate, inhibit the formation of non-catalytic palladium black, as well as improve catalyst recycling [45,46].However, the suitable palladium components may be limited to homogeneous palladium salts or complexes rather than supported palladium materials.Herein, the catalytically active components of C1-C3 were PdO 2 nanoparticles immobilized over silica gel, whose surface and internal pores may have been largely blocked by ILs, subsequently leading to very poor dispersion of the substrate into catalytic centers.
Furthermore, BMImPF 6 seemed better than BMImBF 4 in the C1 and C3 catalyzed reactions for two substrates (entries 8 vs. 7, 19 vs. 18, Tables 3 and 4), but C2-facilitated transformation showed a contrary tendency (entries 15 vs. 14, Tables 3 and 4).Herein, BF 4 − was water-soluble and PF 6 − was immiscible with water [46].Therefore, it seemed that the surfaces and internal pores of C1 and C3 were lipophilic, while C2 was hydrophilic, which affected their affinity to different ILs, leading to different outputs.

Effect of Catalyst Recycling
The results of catalyst recycling are summarized in Figures 10 and 11.At first, the recycling of synthesized catalysts including C1-C3 themselves showed slightly decreased or even improved yields in the cross-couplings of 1 with 2 (green columns, Figure 10b-d) and 4 (green columns, Figure 11b-d) during three instances of use.However, use of ILs as recycling media such as BMImBF 4 and BMImPF 6 gave sharply degraded or continuously low outputs for the cross-couplings of 1 with both 2 (red and blue columns, Figure 10b-d) and 4 (red and blue columns, Figure 11b-d).

Effect of Catalyst Recycling
The results of catalyst recycling are summarized in Figures 10 and 11.At first, the recycling of synthesized catalysts including C1-C3 themselves showed slightly decreased or even improved yields in the cross-couplings of 1 with 2 (green columns, Figure 10b-d) and 4 (green columns, Figure 11b-d) during three instances of use.However, use of ILs as recycling media such as BMImBF4 and BMImPF6 gave sharply degraded or continuously low outputs for the cross-couplings of 1 with both 2 (red and blue columns, Figure 10bd) and 4 (red and blue columns, Figure 11b-d).Therefore, the PdO2 nanoparticles fixed over silica gel not only showed great activity in the fresh round, but also exhibited considerable endurability during their continuous use.However, the introduction of imidazolium ILs as recyclable media was not effective; it blocked the dispersion of substrates into PdO2 nanoparticles, and also caused the decomposition of catalytic centers during continuous catalytic rounds.Therefore, the PdO 2 nanoparticles fixed over silica gel not only showed great activity in the fresh round, but also exhibited considerable endurability during their continuous use.However, the introduction of imidazolium ILs as recyclable media was not effective; it blocked the dispersion of substrates into PdO 2 nanoparticles, and also caused the decomposition of catalytic centers during continuous catalytic rounds.

Proposed Mechanism for Heterogeneous Suzuki-Miyaura Cross-Coupling
A proposed reaction process is summarized in Figure 12.The PdO 2 nanoparticles fixed on the surface of C1 were first activated and reduced to Pd(0) species by the internal electron transfer of C1, leading to catalytically active TS1 [13].Next, the organic iodide (electrophile, compound 1) was coordinated to TS1, meanwhile the surficial PdO 2 nanoparticles were oxidized to Pd(II) species under the present conditions, like oxidation of O 2 in air [13].In comparison, the ligand of the palladium(II) precursor stabilized the intermediate during oxidative addition, but this effect became marginal on the surface of the PdO2 nanoparticles, where surficial hydroxyl groups turned out to be great substitutes for those ligands.On the other hand, although the immobilization of homogeneous palladium(II) complexes into ILs enhanced both oxidative addition and transmetallation [45,46], the dispersion of PdO2 nanoparticles immobilized over silica gel into ILs was not successful, probably because the surface of the PdO2 nanoparticles was saturated and the internal pores were blocked by ILs.
In comparison, the ligand of the palladium(II) precursor stabilized the intermediate during oxidative addition, but this effect became marginal on the surface of the PdO 2 nanoparticles, where surficial hydroxyl groups turned out to be great substitutes for those ligands.On the other hand, although the immobilization of homogeneous palladium(II) complexes into ILs enhanced both oxidative addition and transmetallation [45,46], the dispersion of PdO 2 nanoparticles immobilized over silica gel into ILs was not successful, probably because the surface of the PdO 2 nanoparticles was saturated and the internal pores were blocked by ILs.

Synthetic and Analytical Instruments
Ball milling was carried out on a YJKS-Speediness Grind Machine, Foshan Tenghao Instrument Technology Co., Ltd.(Foshan, China), featuring two-vessel specification, a voltage of 220 V, as well as 370 W power.The milling beads of ZrO 2 had a size of 18 mm.X-ray photoelectron spectroscopy (XPS) was performed on a Kratos Axis Ultra DLD (Kratos Co., Ltd., Manchester, UK), whose lighting source selected monochromatic Al K α X-rays (1486.6 eV).During data processing, the binding energy scale was first calibrated by setting the C 1s peak at 284.8 eV.The peaks were then fitted by using the Gaussian-Lorentz (G/L) product function with 30% Lorentzian component.
The wide-angle (2θ = 10-80 • ) X-ray diffractions were performed on a Philips X'Pert Pro diffractometer (PANalytical B.V. Co., Ltd., Almelo, Holland), whose X-ray source was Cu-Kα radiation (λ = 1.5418Å), along with a diffraction angle interval of 0.05 • s −1 .The porosity parameters including BET surface area, total pore volume, micropore volume, and pore size were determined on Micromeritics ASAP 2020 (Micromeritics Instruments Corporation, Norcross, Atlanta, GA, USA).The nitrogen adsorption isotherms were measured at 77.35 K, and the samples were degassed in vacuum at 150 • C before testing.The surface area was calculated by using the multi-point Brunauer-Emmett-Teller (BET) method in light of the adsorption data with a relative pressure P/P 0 at 0.06-0.3.Total pore volume was derived from the nitrogen adsorbed at P/P 0 = 0.97.Both pore volume and pore size were calculated by using the Barrett-Joyner-Halenda (BJH) method.
The acid amounts including both Lewis and Brönsted acidities of synthesized samples were determined by n-butylamine titration in association with coeruleum bromocresolis as indicator.This protocol was described previously [47].The sample (300 mg) was first combined with n-butylamine solution (25.00 mL, 0.05 mol L −1 in toluene) into a conical flask (250 mL).After shaking for 5 min under cover, 2-proponal (100 mL) and coeruleum bromocresolis (one drop, diluted solution) were consequently added.The solution obtained was then titrated with HCl solution (0.025 mol L −1 in water), and the end point of titration was judged as when the solution color changed from blue to yellow.The acid amount of samples was evaluated as the amount of n-butylamine adsorbed, through the subtraction of the residues (n-butylamine) in the solution (determined by HCl titration) from the pre-added total n-butylamine.As shown in Figures 10 and 11, compound 1 (2.0 mmol, electrophile), compound 2 and 4 (2.0 mmol, nucleophile), Pd catalyst (C1-C3, Pd of 2-4 mol%, based on compound 1; C0, equal Si to C1; according to XPS, Table 1), and anhydrous Na 2 CO 3 (2.0-4.0 mmol) were combined with the THE solvent system (toluene, 8 mL; H 2 O, 4 mL; EtOH, 2 mL) into a round-bottom flask (100 mL) with a condenser sealed with a balloon.Then, the resulting mixture was magnetically stirred at 80 • C for 8 h.
After cooling to room temperature, the mixture was filtrated under reduced pressure, and the residues (catalyst) were collected, dried in air, and then recycled after consumables (substrates, solvent, base) were added.The filtrate was extracted with diethyl ether (3 × 50 mL); the organic layers obtained were washed with saturated NaHCO 3 solution (50 mL), distilled H 2 O (50 mL), and brine (50 mL).After being dried over anhydrous Na 2 SO 4 and filtered, the solution was concentrated under reduced pressure and the residue was purified by column chromatography (SiO 2 , 300 mesh; petroleum ether/ethyl acetate, 4/1, v/v; adding ten drops of triethylamine in eluent of 200 mL).The cross-coupling products (compound 3 or 5) were obtained as orange and white solids, respectively.

Ionic Liquid-Facilitated Reaction and Catalyst Recycling
As shown in Figures 10 and 11, compound 1 (2.0 mmol, electrophile), and Pd catalyst (C1-C3, Pd of 2-4 mol%, based on compound 1; C0, equal Si to C1; according to XPS, Table 1) were combined and fully dispersed into imidazolium ILs (BMImBF 4 or BMImPF 6 , 10 mL, respectively) under vigorous stirring at room temperature.Then, the temperature was raised to 80 • C, and the mixture was stirred at this temperature for 1 h to obtain a uniform solution.

Table 1 .
Binding energy and atomic composition of elements on catalyst surface.

Table 2 .
Textural and other properties of catalyst samples.

Table 2 .
Textural and other properties of catalyst samples.