Palladium Catalysts Based on Porous Aromatic Frameworks, Modiﬁed with Ethanolamino-Groups, for Hydrogenation of Alkynes, Alkenes and Dienes

: The current work describes an attempt to synthesize hybrid materials combining porous aromatic frameworks (PAFs) and dendrimers and use them to obtain novel highly active and selective palladium catalysts. PAFs are carbon porous materials with rigid aromatic structure and high stability, and the dendrimers are macromolecules which can e ﬀ ectively stabilize metal nanoparticles and tune their activity in catalytic reactions. Two porous aromatic frameworks, PAF-20 and PAF-30, are modiﬁed step-by-step with diethanolamine and hydroxyl groups at the ends of which are replaced by new diethanolamine molecules. Then, palladium nanoparticles are applied to the synthesized materials. Properties of the obtained materials and catalysts are investigated using X-ray photoelectron spectroscopy, transmission electron microscopy, solid state nuclear magnetic resonance spectroscopy, low temperature N 2 adsorption and elemental analysis. The resulting catalysts are successfully applied as an e ﬃ cient and recyclable catalyst for selective hydrogenation of alkynes to alkenes at very high (up to 90,000) substrate / Pd ratios.


Introduction
Palladium is one of the most widespread metals used in the selective hydrogenation of unsaturated compounds [1][2][3]. It is commonly used in a form of nanoparticles, for this process, whose high surface area provides a large number of available active sites per unit area. This leads to greater reactivity of the nanoparticles than with bulk palladium [4]. Changes in the size of particles, their morphology and distribution significantly affect their catalytic properties. However, unstabilized metal nanoparticles (NPs) are prone to aggregation, which leads to the formation of thermodynamically stable large particles with lower catalytic activity [5]. The main method to avoid aggregation of NPs is the use of encapsulating ligands and polymers or porous materials [6][7][8][9][10][11].
One of the most effective supports for metal nanoparticles are dendrimers-regular, three-dimensional, spherically symmetric macromolecules. Since dendrimers have interior voids of nanometer dimensions, they can hold nanoparticles of suitable sizes, which can fit into those voids [12]. Encapsulation of metals in a structure of dendrimers allows for control of the size of the nanoparticles and provides for their uniform distribution [13]. Strong coordination of the chelating nitrogen-containing groups with metal prevents NPs from leaching and agglomerating during reaction, thus providing outstanding stability. However, the use of dendrimer-stabilized nanoparticles is limited

Synthesis and Characterization of Supports
Porous aromatic frameworks (PAF), PAF-20 and PAF-30, were synthesized according to the method described by Yuan Y. et al. [53] from tetrakis-(p-bromophenyl)methane and 1,4-phenylenediboronic acid or 4,4 -biphenyldiboronic acid. Consequently, PAF-20 and PAF-30 differed in the number of benzene rings between sp 3 carbon atoms in the nodes of the frameworks: PAF-20 had three benzene rings, and PAF-30 had four rings. Thus, PAF-20 typically had a smaller pore size, but a larger surface area compared to PAF-30 [24]. Subsequent modification of PAF structure with chloromethyl groups and diethanolamino groups was performed using methods by Gangadharan. D. et al. [54] and Lu.W. et al. [55] (Scheme 1). To determine the structural features of the structure of the obtained materials, they were studied using solid state nuclear magnetic resonance (NMR) spectroscopy, low-temperature N 2 adsorption and elemental analysis.  Figure 1 shows the solid-state cross polarization-magic angle spinning (CP-MAS) 13 C NMR spectra for PAF-based materials. All spectra contained typical for PAF signals in the region of 125-145 ppm for sp 2 -hybridized carbon atoms of aromatic rings and, at 63 ppm, corresponded to sp 3hybridized carbon atoms in the center of tetraphenylmethane units. After modification with chloromethyl groups, a new signal belonging to the -CH2Cl group appeared at 44 ppm, which was consistent with published data [56,57]. Replacing the chlorine atom in -CH2Cl with an ethanolamine group reduced the intensity of this peak due to screening of the carbon atom by branched ethanolamine groups and shifted its position from 44 ppm to 49-50 ppm. A similar dependence of the signal displacement upon the addition of various amines was observed during the earlier works [54,56]. Also, diethanolamino-modified materials PAF-20-G0, PAF-20-G1, PAF-30-G0 and PAF-30-G1 contained new signals at 56 and 58 ppm, which were assigned to carbon atoms of the ethanolamine groups.
(a) 13C CP-MAS NMR for PAF- 30 Figure 1 shows the solid-state cross polarization-magic angle spinning (CP-MAS) 13 C NMR spectra for PAF-based materials. All spectra contained typical for PAF signals in the region of 125-145 ppm for sp 2 -hybridized carbon atoms of aromatic rings and, at 63 ppm, corresponded to sp 3 -hybridized carbon atoms in the center of tetraphenylmethane units. After modification with chloromethyl groups, a new signal belonging to the -CH 2 Cl group appeared at 44 ppm, which was consistent with published data [56,57]. Replacing the chlorine atom in -CH 2 Cl with an ethanolamine group reduced the intensity of this peak due to screening of the carbon atom by branched ethanolamine groups and shifted its position from 44 ppm to 49-50 ppm. A similar dependence of the signal displacement upon the addition of various amines was observed during the earlier works [54,56]. Also, diethanolamino-modified materials PAF-20-G0, PAF-20-G1, PAF-30-G0 and PAF-30-G1 contained new signals at 56 and 58 ppm, which were assigned to carbon atoms of the ethanolamine groups.
Porous properties of PAFs were defined using low-temperature N 2 adsorption. Starting materials PAF-20 and PAF-30 had 578 m 2 /g and 506 m 2 /g Brunauer-Emmett-Teller (BET) surface area, respectively ( Table 1). The surface area of polymers decreased upon modification: after chloromethylation it was 472 and 436 m 2 /g for PAF-20-CH 2 Cl and PAF-30-CH 2 Cl, and after treatment with diethanolamine it was 64 and 31 m 2 /g, respectively. Transition from G0 to G1 generation resulted in an even greater reduction of surface area, which was explained by the blocking of pores by bulky functional groups. consistent with published data [56,57]. Replacing the chlorine atom in -CH2Cl with an ethanolamine group reduced the intensity of this peak due to screening of the carbon atom by branched ethanolamine groups and shifted its position from 44 ppm to 49-50 ppm. A similar dependence of the signal displacement upon the addition of various amines was observed during the earlier works [54,56]. Also, diethanolamino-modified materials PAF-20-G0, PAF-20-G1, PAF-30-G0 and PAF-30-G1 contained new signals at 56 and 58 ppm, which were assigned to carbon atoms of the ethanolamine groups.   The adsorption isotherms ( Figure 2) of the samples PAF-20 and PAF-30, as well as their chloromethylated derivatives, exhibited a sharp absorption of N 2 at a low relative pressure (p/p 0 <0.05), which indicated the developed microporous structure of these materials. The presence of a hysteresis loop, and the fact that the nitrogen sorption curve gradually rose without the appearance of a plateau, may indicate the presence of mesopores in the polymer structure [58]. Conversely, this species may be due to diffusion restrictions or polymer swelling [59]. Chloromethylation did not change the character of nitrogen adsorption, which may indicate a uniform distribution of -CH 2 Cl groups over the volume of the carrier and a slight change in the pore size.
<0.05), which indicated the developed microporous structure of these materials. The presence of a hysteresis loop, and the fact that the nitrogen sorption curve gradually rose without the appearance of a plateau, may indicate the presence of mesopores in the polymer structure [58]. Conversely, this species may be due to diffusion restrictions or polymer swelling [59]. Chloromethylation did not change the character of nitrogen adsorption, which may indicate a uniform distribution of -CH2Cl groups over the volume of the carrier and a slight change in the pore size.   Treatment of chloromethylated polymers with diethanolamine led to significant change in the character of nitrogen adsorption: there was no longer a sharp rise in the adsorption curve and, in the case of PAF-20-G0, there was practically no desorption of nitrogen from the pores. This shape of the curves can be associated with strong blocking of the pores by the diethanolamino-groups, which was more pronounced for PAF-20-G0 due to its smaller pore size. Further modification of materials up to generation G1 led to a further reduction in pore size: adsorption curves for PAF-20-G1 and PAF-30-G1 were typical for non-porous materials [58]. Thus, we can conclude for PAF-20-G0 and PAF-30-G0 materials, only a small fraction of the pores were available and. in the case of PAF-20-G1 and PAF-30-G1, the only pores available were in the immediate vicinity of the outer surface of the catalyst grain. Also, the decrease in surface area and porosity after grafting of the surface of porous materials is a well-known fact. Thus, after modification of PAFs with different polyamines [58], the surface area and free volume of pores decreased dramatically from 4023 m 2 /g for PPN-6 material to 555 m 2 /g for tris-(2-aminoethyl)amine-modified material. However, we should note that in the case of the modification of porous aromatic frameworks with diethanolamine led to too high a decrease in porosity.
Elemental analysis also suggests successful functionalization of aromatic rings with different functional groups ( Table 2). The averaged content of chlorine in the materials PAF-20-CH 2 Cl and PAF-30-CH 2 Cl was 3%-to be exact, about 7% of the benzene rings were modified. After treatment of chloromethylated polymers with diethanolamine, it decreased to 0.9%-1.2%, whereas nitrogen content in the resulting samples was 1.7%-1.8%. The presence of chlorine in these samples may indicate an incomplete modification: most likely, chloromethyl groups deep into the catalyst grain did not react with diethanolamine. Regarding the materials of the G1 series, the content of chlorine was even higher, and the nitrogen content increased only slightly. It indicates the occurrence of the substitution reaction of the hydroxyl groups for chlorine atoms during the treatment of materials of the G0 series with thionyl chloride, on the one hand. The completeness of this reaction was even lower than in the synthesis of materials of the G0 series, on the other hand.
Pd-PAF-20-G0 and Pd-PAF-30-G0 catalysts contained well-dispersed small particles with a relatively narrow size distribution (2-2.5 nm) based on the transmission electron microscopy (TEM) microphotographs ( Figure 3). This confirms the successful introduction of nanoparticles into the modified pores of materials. Table 3. Palladium content by atomic absorption spectroscopy in synthesized catalysts.

Material
Pd To contrast, the G1 series catalysts contained only a small number of palladium particles, which were larger (average size 7-8 nm) and the size distribution curves were wider ( Figure 4). It is seen in the micrographs that there are few nanoparticles and agglomerates are observed on the surface. This fact may be due to the blocking of the pores by the diethanolamine groups, which interfere with the diffusion of palladium ions inside the porous structure. To contrast, the G1 series catalysts contained only a small number of palladium particles, which were larger (average size 7-8 nm) and the size distribution curves were wider ( Figure 4). It is seen in the micrographs that there are few nanoparticles and agglomerates are observed on the surface. This fact may be due to the blocking of the pores by the diethanolamine groups, which interfere with the diffusion of palladium ions inside the porous structure.  According to the X-Ray photoelectron spectroscopy (XPS) data (Table 4), the nitrogen content was higher in the materials of the G1 series, which proves the process of modification of ethanolamine groups during the synthesis of materials PAF-20-G1 and PAF-30-G0. Concurrently, the chlorine content in the materials with modification of the G1 type remained higher than in the G0 series. That demonstrates the incomplete reaction of the substitution of chlorine for diethanolamine at the last stage of synthesis. The palladium content was higher in materials of the G0 series, which may be due to better immobilization of the palladium particles in the pores of the support. All XPS spectra demonstrated two sets of peaks related to reduced (Pd 0 ) and oxidized (PdOx) palladium forms ( Figure 5). The presence of oxides in the spectra of samples could be due to incomplete reduction of the initial palladium or oxidation of the nanoparticles with atmospheric oxygen. The binding energies of Pd 0 3d5/2 and Pd 0 3d3/2 for Pd-PAF-20-G0 and Pd-PAF-30-G0 catalysts were higher than that for free Pd 0 (335.6 and 341.1 eV, respectively). The changing of energy values indicates the presence of a coordination effect between nitrogen or oxygen atoms in the supports According to the X-ray photoelectron spectroscopy (XPS) data (Table 4), the nitrogen content was higher in the materials of the G1 series, which proves the process of modification of ethanolamine groups during the synthesis of materials PAF-20-G1 and PAF-30-G0. Concurrently, the chlorine content in the materials with modification of the G1 type remained higher than in the G0 series. That demonstrates the incomplete reaction of the substitution of chlorine for diethanolamine at the last stage of synthesis. The palladium content was higher in materials of the G0 series, which may be due to better immobilization of the palladium particles in the pores of the support. All XPS spectra demonstrated two sets of peaks related to reduced (Pd 0 ) and oxidized (PdOx) palladium forms ( Figure 5). The presence of oxides in the spectra of samples could be due to incomplete reduction of the initial palladium or oxidation of the nanoparticles with atmospheric oxygen. The binding energies of Pd 0 3d 5/2 and Pd 0 3d 3/2 for Pd-PAF-20-G0 and Pd-PAF-30-G0 catalysts Catalysts 2020, 10, 1106 9 of 17 were higher than that for free Pd 0 (335.6 and 341.1 eV, respectively). The changing of energy values indicates the presence of a coordination effect between nitrogen or oxygen atoms in the supports ( Table 5). The same binding energies for the materials Pd-PAF-20-G1 and Pd-PAF-30-G1 were practically no different from the binding energies for free palladium. These results confirm the assumption that there is no coordination between palladium nanoparticles and the diethanolamine groups. Nanoparticles in these catalysts are either in unmodified pores or on the surface of the support.
Catalysts 2020, 10, x FOR PEER REVIEW 9 of 17 (Table 5). The same binding energies for the materials Pd-PAF-20-G1 and Pd-PAF-30-G1 were practically no different from the binding energies for free palladium. These results confirm the assumption that there is no coordination between palladium nanoparticles and the diethanolamine groups. Nanoparticles in these catalysts are either in unmodified pores or on the surface of the support.

Catalytic Activity
The synthesized catalysts were examined in the hydrogenation of various C6 and C8 unsaturated compounds ( Table 6). The G0 series catalysts were more active than the G1 series catalysts, and catalysts based on the materials of the PAF-30 type were more active than those of the PAF-20 type.

Catalytic Activity
The synthesized catalysts were examined in the hydrogenation of various C 6 and C 8 unsaturated compounds ( Table 6). The G0 series catalysts were more active than the G1 series catalysts, and catalysts based on the materials of the PAF-30 type were more active than those of the PAF-20 type. Thus, the Pd-PAF-30-G0 catalyst showed the highest activity in the hydrogenation of linear alkynes and alkenes: its specific activity (Table 7) was more than 300,000 mol Sub × mol Me −1 × h −1 in the case of hexyne-1, hexene-1, octyne-1, and octene-1 (Table 7). Regarding 2,5-dimethyl-2,4-hexadiene, it was about 295,000 mol Sub × mol Me −1 × h −1 and, for more bulky substrates-phenylacetylene and styrene-it was much lower (116,000 and 79,000 mol Sub × mol Me −1 × h −1 , respectively), which may be due to conjugation between the benzene ring and double and triple bonds in these substrates.   Pd-PAF-20-G0 possessed lesser hydrogenation activity: high conversion was achieved only for hexyne-1, whereas yields of hydrogenation products for other substrates were much lower. This could be due to the smaller pore size in PAF-20 and, as a consequence, lower diffusion of substrates to palladium nanoparticles [33]. Catalysts Pd-PAF-20-G1 and Pd-PAF-30-G1 showed less activity. This may be due to the fact that further modification of the material with diethanolamine groups leads to a significant decrease in pore size, restricting the penetration of substrates, and the leaching of metal particles from the surface.
The stability of Pd-PAF-20-G0 and Pd-PAF-30-G0 catalysts was tested in recycle experiments regarding the hydrogenation of hexene-1. Both catalysts remained active for five reuse cycles. A slight reduction in conversions was observed only in the first two repetitions (Table 8). This fact is associated with the leaching of particles from the surface of the catalyst, as is seen in TEM micrographs, as well as in losses during catalytic experiments. A slight increase in the average particle size from 2.5 to 2.6 nm for Pd-PAF-30-G0, and from 2 to 2.9 nm for Pd-PAF-20-G0, can be noted; however, the distribution of the nanoparticles in the materials remained uniform ( Figure 6). The washing out of nanoparticles from the surface also is confirmed by XPS data, according to which, after five reuse cycles, the amount of surface palladium decreased from 2.8 at.% to 2.1 at.% for Pd-PAF-20-G0, and from 1.3 at.% to 1.0 at.% for Pd PAF 30 G0 (Table 9). Catalysts 2020, 10, x FOR PEER REVIEW 11 of 17 reduction in conversions was observed only in the first two repetitions (Table 8). This fact is associated with the leaching of particles from the surface of the catalyst, as is seen in TEM micrographs, as well as in losses during catalytic experiments. A slight increase in the average particle size from 2.5 to 2.6 nm for Pd-PAF-30-G0, and from 2 to 2.9 nm for Pd-PAF-20-G0, can be noted; however, the distribution of the nanoparticles in the materials remained uniform ( Figure 6). The washing out of nanoparticles from the surface also is confirmed by XPS data, according to which, after five reuse cycles, the amount of surface palladium decreased from 2.8 at.% to 2.1 at.% for Pd-PAF-20-G0, and from 1.3 at.% to 1.0 at.% for Pd PAF 30 G0 (Table 9).

Used Reagents
The following reagents were used in the work: benzene ( PAFs ware prepared according to published literature procedures [53]. The modification techniques for PAF-20 and PAF-30 are similar. Below are the methods for modifying the PAF-20 material.

Synthesis of PAF-20-CH 2 Cl
Chloromethylation was carried out according to the method modified from a previous work [55]. Paraformaldehyde (1 g) and hydrochloric acid (20 mL) were placed in a round-bottomed flask equipped with a stirrer and a reflux condenser. After dissolving of paraformaldehyde, phosphorus pentoxide (4 g) and glacial acetic acid (6 mL) were carefully added to the mixture. Subsequently, PAF-20 (200 mg) was placed in the flask and stirred at 90 • C for 3 days. The resulting solid was collected using filtration, washed 3 times with water (100 mL) and ethanol (100 mL), and dried in vacuo to produce PAF-20-CH 2 Cl as a yellow powder.

Synthesis of PAF-20-G0
Modification of PAF-20-CH 2 Cl was performed according to the procedure adapted from a previous work [54]. PAF-20-CH 2 Cl (100 mg), dioxane (6 mL), diethanolamine (0.4 mL) and catalytic amounts of potassium iodide (10 mg) were mixed in a round-bottomed flask and stirred at 70 • C for 3 days. The solid was collected using filtration, washed with acetone (3 × 50 mL), and dried in vacuo. The obtained material, PAF-20-CH 2 N(CH 2 CH 2 OH) 2 , was called PAF-20-G0 as an analogy with the numbering of the dendrimer generations.
Palladium acetate (4.26 mg) was dissolved in chloroform (6 mL) in a round-bottomed flask, equipped with a magnetic stirrer and a reflux condenser. Then, PAF-20-G0 (100 mg) was added to the resulting solution. The suspension was stirred at 60 • C for 24 h, and then a solution of sodium borohydride (7 mg) in a water-ethanol mixture (0.5 mL:0.5 mL) was added. The reaction mixture darkened and gas evolution was observed. Subsequently, the flask was closed with a stopper and left for 12 h with stirring. After the reaction, the resulting substance was collected by centrifugation, washed with water (3 × 50 mL) and ethanol (3 × 50 mL) to remove sodium tetraborate, the precipitate was isolated by centrifugation and dried in air.

Catalytic Experiments
The calculated amounts of substrates and catalyst (1 mg) were placed in a glass tube equipped with a magnetic stirrer. The tube was placed in a steel autoclave, then it was sealed and pressurized with hydrogen at a pressure of 10 atm. Reactions were carried out at 80 • C for 30 min. After completion of the reaction, the autoclave was cooled to room temperature and depressurized. Reaction products were analyzed by gas chromatography. Specific activity of the catalysts was calculated using following equation: where A is the specific activity, Conv is the conversion of the substrate, Sub/Me is the substrate to metal proportion, D is the metal dispersion, and t is the reaction time.

Low Temperature Nitrogen Adsorption
Nitrogen desorption/desorption isotherms were recorded at 77 K with a Micromeritics Gemini VII 2390 instrument (Micromeritics, Norcross, GA, United States). All samples were degassed at 110 • C for 6 h before measurement. The surface area (S BET ) was calculated using the Brunauer-Emmett-Teller (BET) method based on adsorption data in the relative pressure range P/P 0 = 0.05-0.2. The total pore volume (Vtot) was determined by the amount of nitrogen adsorbed at a relative pressure of P/P 0 = 0.995.

Transmission Electron Microscopy (TEM)
TEM analysis was carried out on a JEOL JEM-2100/Cs/GIF microscope (JEOL, Tokyo, Japan) with a 0.19 nm lattice fringe resolution and an accelerating voltage of 200 kV. The processing of the micrographs and the calculation of the average particle size were conducted using the ImageJ software program.

X-ray Photoelectron Spectroscopy (XPS)
XPS studies were performed on a VersaProbeII, ULVAC-PHI (ULVAC-PHI, Inc., Kanagawa, Japan) instrument using excitation with Al Kα X-ray radiation at 1486.6 eV. The calibration of photoelectron peaks was based on the Au 4f line with a binding energy of 84 eV and on the Cu2p3/2 line (932.6 eV). The transmission energy of the energy analyzer was 117.4 eV (survey scans) and 23.5 eV (individual lines).

Gas-Liquid Chromatography
Analysis of the reaction mixture was carried out on a Agilent 6890 G1530A chromatograph (Hewlett Packard, Santa-Clara, CA, United States) equipped with a flame-ionization detector and a HP-1 column (50 m × 0.32 mm × 1.05 µm, 100% dimethylpolysiloxane grafted phase). Helium was a carrier gas; the analysis was carried out in constant pressure mode (1.53 bar). Chromatograms were recorded and analyzed on a computer using the HP ChemStation Rev.A.06.01 (403) software.

Atomic Absorption Spectroscopy
The Pd content in the catalysts was determined via atomic absorption spectroscopy (AAS) on an iCE 3000 Series AA spectrometer (Thermo Scientific, Santa-Clara, CA, United States) with flame atomization. The data were processed using the SOLAAR software.

Conclusions
To conclude, we developed active and selective catalysts based on hybrid materials. It was shown that introduction of ethanolamine groups allowed for achievement of an efficient sorption of palladium ions and uniform distribution of palladium nanoparticles in size and in the pores of the carrier. Catalysts based on PAF-20-G0 and PAF-30-G0 materials demonstrated a high catalytic activity and stability. Nanoparticles in these catalysts were located both on the surface of the support and in the pore space.
Modification of the G1 type led to blockage of the pores of the aromatic framework, which prevented the diffusion of palladium ions into the pores of the carrier. The G1 series catalysts contained larger particles (average size 7-8 nm) and were characterized by a broader size distribution, and most of the nanoparticles were located on the material surface. The activity of the Pd-PAF-20-G1 and Pd-PAF-30-G1 catalysts turned out to be lower than that of the G0 series catalysts, which was due to the blockage of pores.
Thus, to enhance the activity of catalysts, it is necessary to solve the problem of a significant decrease in the porosity of hybrid materials, as well as to improve approaches to the modification of supports. Due to this, it will be possible to significantly expand the field of application of these materials, in particular, to use them to create catalysts for other catalytic processes (processing of petroleum fractions, bio-raw materials, fine organic synthesis, etc.)

Conflicts of Interest:
The authors declare no conflict of interest.