2.1 Materials

Epoxy resin bisphenol A epichlorohydrin medium molecular weight ≤ 700 Epidian 5 (EP5), epoxy equivalent 200, (viscosity 20000–30000 mPas at 23 °C), Organika Sarzyna, Poland and epoxy resin a high purity bisphenol A diglycidylether D.E.R.™332 (DER), epoxy equivalent 170, (viscosity 4000–6000 mPas at 25 °C), the Dow Chemical Company, USA, were used. Iodobenzene (Aldrich), triethylamine (Organika Sarzyna), methyl acrylate (Aldrich) and N-metylhylpyrrolidone (ABCR) were used for the Heck reaction. Tetrafluoroboric acid, theophylline, theobromine, caffeine, potassium hydroxide, didecyldimethylammonium chloride as well as all solvents were purchased from commercial suppliers (Sigma-Aldrich, Fluka, Lonza) and used for synthesis of ionic liquids without further purification. The ionic liquids applied are collected in Table 1.

2.2 Synthesis of ionic liquids

2.2.1 Synthesis of didecyldimethylammonium and 1-methyltheobromine (caffeine) salts

Method I – Didecyldimethylammonium chloride (0.05 mol) was dissolved in methanol and a stoichiometric amount of saturated methanolic solution of KOH was added. The solutions were stirred at room temperature for 5 min, after which the partially precipitated KCl was filtrated. Then a stoichiometric amount of an appropriate theophylline or theobromine or tetrafluoroboric acid was added. The solutions were stirred again at room temperature for 15 min and after evaporation of the solvent at a temperature of 40 °C, the product was dissolved in 20 ml of anhydrous acetone. The inorganic salt was separated by filtering and acetone was evaporated to give the product, which was finally dried under reduced pressure at 80 °C for 24 h (Scheme 1).

Method II – 1-methyltheobromine (caffeine) (0.1 mol) was dissolved in water at 40 °C. Then a stoichiometric amount of an appropriate tetrafluoroboric acid was added. The solutions were stirred at 40 °C for 24 h and after evaporation of the solvent at temperature of 40 °C, the product was finally dried under reduced pressure at 60 °C for 24 h (Scheme 1).

2.3 Preparation of the carriers

The carrier was prepared by mixing (3 g) the Epidian 5 or D.E.R™322 epoxy resin with (1.2 g) the relevant ionic liquid ([d₂m₂am][thb], [d₂m₂am][thp], [d₂m₂am][BF₄], [mthb][BF₄]). After obtaining homogenous consistency, the mixture was transferred onto a Teflon^® mold and cured at a temperature of 298 to 423 K over a period of 1 hour to 48 hours. The cured resin was then frozen by immersion in liquid nitrogen and mechanically ground to a powder. Average particle size of crushed resins was measured by sieve analysis. These measurements yield a median size ranging from 0.5 to 1.0 mm.

2.4 Catalyst preparation

A typical procedure was followed. To a known amount of polymer matrix (0.1 g) in a round-bottom flask, a known amount of PdCl₂(PhCN)₂ dissolved in 10 ml of toluene was added. The mixture was stirred at room temperature for 2–3 days when all Pd complex had reacted with the polymer. The yellow product was filtrated off, extracted with toluene under nitrogen in order to remove the complex that was non-chemically bound to the polymer and finally dried under vacuum. For all supported catalysts, the Pd content was fixed below the metal uptake capacity for a polymer. The dry catalysts could be stored in air for many months without change.

2.5 Catalytic tests

The typical Heck reaction was carried out in 50 ml laboratory reactor. The reactor was charged with 0.05 g of supported palladium catalyst, 0.28 ml (2.5 mmol) of iodobenzene, 0.23 ml (2.5 mmol) of methyl acrylate, 0.15 ml (1.4 mmol) of toluene, 0.35 ml (2.5 mmol) of triethylamine and 1.00 ml (10.4 mmol) of N-methylpyrrolidone (NMP). The reaction was conducted at a temperature of 80 °C for 1.5 hours. Samples were taken every 5, 15, 30, 60 and 90 minutes. The composition of the samples was analyzed by GC. The reaction products were quantitatively analyzed by gas chromatography (GC) on a Hewlett-Packard 5990 II gas chromatograph equipped with a thermal conductivity detector. The GC column was HP-50+ (cross-linked 50% Ph Me silicone) 30 m × 0.63 mm × 1.0 μm film thickness. The product was identified by matching retention times with those of authentic samples.

2.6 Catalyst recycling

In the recycling tests, when reaction was completed, the catalyst was filtered off from the reaction mixture, washed with 3 × 25 ml of toluene, vacuum dried and used for the next reaction.

2.7 Spectroscopy characterization

The FT-IR spectra were measured with a BIORAD 175 C spectrophotometer in an air atmosphere. XPS measurements were made on a VG ESCALAB 210 spectrometer with Mg Kα (hν = 1253.6 eV) excitation from an X-ray tube (reduced power 10 kV, 10 mA). The pressure in the spectrometer chamber was about 5 × 10^–9 mbar. Samples were pressed to pellets under a pressure of 100 kbar for 10 min before these measurements. The C 1s, N 1s, O 1 s, Cl 2p and Pd 3d core level spectra were recorded. The analyzer pass energy was set at 20 eV. A take-off angle of 90^o was used in all XPS studies. Curve fitting was performed using the ECLIPSE data system software. This software describes each of the components of a complex envelope as a Gaussian-Lorentzian sum function. The background was fitted using non-linear model function proportional to the integral of the elastically scattered electrons (Shirley backgrounds). All binding energies (BEs) were referenced to the C 1s neutral carbon peak at 284.6 eV.

2.8 Morphology studies

The morphology of samples was investigated with a scanning electron microscope JEOL 5500 LV, working in high vacuum and accelerating voltage of 10 kV. Energy-dispersive X-ray microscope (SEM–EDX), JEOL JSM 840A, Japan, was used to observe the elemental distribution. The white points in the figures denote Pd atoms.

Pore size distribution parameters were determined by application of the BET method on Sorptomatic 1900 FISONS Instrument.

2.9 Study of Pd leaching

TOF-SIMS measurements were performed using an ION-TOF GmbH instrument (TOF-SIMS IV) equipped with 25 keV pulsed Bi⁺ primary ion gun in the static mode (primary ion dose about 1.5 × 1011 ions/cm²). The analyzed area corresponds to a square of 500 × 500 μm in case of secondary ion mass spectra collecting and surface imaging. For each sample three spectra from different surface areas were made. To obtain the plain surface of catalysts (then better mass resolution could be achieved), powder samples were tableted before the measurements. In order to compare the quantity of palladium present on the surface of the catalysts before and after reaction, the number of counts of selected ions obtained from collected mass spectra were normalized on the basis of the value of total counts.

3.1 Spectroscopic characteristics of the polymer supports and the immobilized Pd catalysts

Currently, there is a growing number of works on the usage of cured polymers as potential carriers for immobilizing the catalytic complexes of the transition metals groups. This is due to a number of advantages they possess, such as: resistance to most organic solvents, ease of introduction of functional groups without additional functionalization or the possibility of shaping the final product. For the purpose of our research, we decided to select a new type of polymer supports based on epoxy resins with various epoxide numbers, cured with ammonium ionic liquids. We prepared two sets of carriers, one for the Epidian 5 resin, and the other for the D.E.R™322 resin (DER), which were cured with the ionic liquids presented in Table 1.For reasons of clarity and capacity of this article, we decide to focus on the carrier of the Epidian 5 epoxy resin (EP5) cured with didecyldimethylammonium theophyllinate (EP5/[d₂m₂am][thp]), which serves as an example of spectroscopic characterization.

The FTIR spectrum of the EP5/[d₂m₂am][thp] carrier (Fig. 1a) reveals a shift of the characteristic peaks observed in spectrum of pure didecyldimethylammonium theophyllinate (not presented here), such as the shift of the absorption bands from 1683 cm⁻¹ to 1705 cm⁻¹ and 1630 cm⁻¹ to 1661 cm⁻¹, corresponding to the stretching vibrations of the CN groups of the purine ring. There was also a new absorption band at 1743 cm⁻¹, corresponding to the carbonyl group resulting presumably from the polymerization of the epoxy resin. IR bands in the region of 1740–1750 cm⁻¹ have been found by Spychaj studying the epoxy resin curing mechanism with imidazolium ionic liquids [18,19]. These IR absorption bands were explained by the proposed mechanism of epoxy resin anionic polymerization initiated by thermal decomposition products of 1,3-dialkylimidazolium ionic liquids. In this mechanism, assuming regeneration of imidazole derivatives, unsaturated structure of epoxy polymer is formed and finally transformed by tautomerization into the structure with the carbonyl group. Our observations indicate similar mechanism of anionic polymerization initiated by the anion derived from the ionic liquid used. The ionic liquid appears to be incorporated into the network, resulting in a cross-linked polymer stabilized with bulky counterions.

By comparing the spectra of the EP5/[d₂m₂am][thp] carrier and the EP5/[d₂m₂am][thp]/Pd catalyst (Fig. 1a,b), we can assume that there are two ways of coordinating the metal to the matrix. They are shown in Scheme 2. At 1743 cm⁻¹, the absorption band, corresponding to the vibrations of the carbonyl group of epoxide chain, vanishes from the EP5/[d₂m₂am][thp]/Pd spectrum (Fig. 1b.), which indicates the bonding of the metal by the O atom with a free pair of electrons (Scheme 2b). We can also observe a decrease and the shift of the absorption bands from 1661 cm⁻¹ to 1656 cm⁻¹ and 1705 cm⁻¹ to 1701 cm⁻¹. This may attest to a particular disorientation of the purine ring under the influence of Pd(II), which indicates the possibility of the metal coordination to the N nitrogen atom (Scheme 2a). No change was observed for strong bands at 1581 cm⁻¹ and 1606 cm⁻¹, considered to be due to asymmetric and symmetric stretching vibrations of two carbonyl groups of theophylline moiety, which indicates that the palladium complex is not bonded to the epoxide support through theophylline carbonyl groups. The new vibrations band corresponding to the coordinate bond palladium-theophylline emerge in the FTIR spectra of the catalyst at 730 cm⁻¹, probably due to the metal-N valence vibrations.

X-ray photoelectron spectroscopy (XPS) analysis gives a better indication of the chemistry of the surface of the heterogenized catalyst particle, whereas the other analysis methods are more relevant to the composition of the particles as a whole. XPS analyses the outer surface of particulates and enables significant information to be extracted concerning the coordination of a catalytic metal to the polymer support and the mechanism of catalytic reaction. The surface composition of catalyst particles can be determined for species present down to 0.1 atomic percent, shallow from 1.5 to 10 nm, depth profiling. Major functional groups for carbon and oxygen can be identified as well as oxidation states for other species. These characteristics are important sources of information about the interfacial chemistry governing catalytic properties. The percentage data obtained by means of the XPS technique are summarized in Table 2. In the epoxy resin carrier sample cured with the [d₂m₂am][thp] ionic liquid, the highest percentage has carbon, which is determined by the use of organic compounds. The CO bond is less common than the others and amounts to 6%. A low percentage of nitrogen bonds with carbon and oxygen (nearly 3%) results from the application of the ammonium ionic liquid. As a result of the immobilization of PdCl₂(PhCN)₂, nearly 5% of PdCl bonds and 1.35% of the divalent palladium bonds emerged. There is also a low percentage (0.37%) of the electron-deficient Pd^δ+ bond, which formed due to the lateral overlapping of the d atomic orbitals. This valence state is also referred to Pd clusters [20]. They are not stable under normal conditions, but their prolonged life is related to the protection afforded by the polymer matrix. The lack of distinct changes in composition of catalyst surface after subsequent catalytic cycles indicates permanent binding of the ionic liquids particles.

The XPS data summarized in Table 3 show that three possible coordination sites participate in binding the Pd to the epoxide support. This is indicated by the increase of the binding energy BE for purine nitrogen atoms and O_CO 1s electrons after the Pd attachment. These data are consistent with the results of FTIR measurements. The XPS spectrum of the freshly prepared EP5/[d₂m₂am][thp]/Pd catalyst shows that palladium appears dominantly in the form of Pd(II). Minor fractions of Pd^δ+ nanoclusters are also observed. It indicates that this polymeric support chains might stabilize such a catalytically active monovalent metastable species with a binding energy larger than that of the Pd metal. The recovered EP5/[d₂m₂am][thp]/Pd catalyst after a Heck reaction contained also some minor fraction of Pd(0). Since the epoxide matrix cured with [d₂m₂am][thp] ionic liquid possesses coordination sites of different coordination ability, it seems very likely that the zero-valent palladium, Pd(0), comes from the reduction of Pd(II) coordinated to the carbonyl oxygen, the weaker of the three coordination sites, whereas new species formed after each reaction run, Pd clusters, come from the reduction of Pd(II) coordinated to the purine nitrogen atoms. The binding energy of Cl 2p_3/2 in the XPS spectra, in the region of 201.7–201.9 eV, indicates the presence of a –PdCl₂ structure.

3.2 Morphology analysis

The characterization of palladium distribution on the surface of the catalysts and the distribution of the size of the nanoparticles were performed by means of a scanning electron microscope (SEM) and energy-dispersive X-ray spectroscopy (EDX) (Figs. 2–5). The average size of the palladium nanoparticles for each sample was determined by counting the size of about 100 particles from several SEM pictures. The DER/[d₂m₂am][thp]/Pd X-ray microprobe (Figs. 2 and 3) indicates a uniform distribution of the palladium particles with an average size of 0.1 nm. A less regular distribution is observed for the DER/[d₂m₂am][thb]/Pd catalyst with an average particle size of 0.6 nm (Figs. 4 and 5). These differences may be explained by referring to the morphology of the polymeric carrier and the steric restrictions imposed by the different microporosities of these supports. The support microporous structure can influence the activity of catalyst. The steric demands of a catalytic centre bound to a support can be very different from those of its homogeneous analogue. The porous structure of the support affects the accessibility of the interior of the swollen resin by increasing diffusional restrictions for reaction species. The results determined by the nitrogen BET adsorption method indicate that DER and EP5-based supports have different pore size distribution which probably arise from differences in crosslink densities. These materials show different ratio of micropores and very low volume of meso- and macropores. The EP5 resins had pore size distribution with the majority of the micropores of radii below 2 nm. The DER resins exhibit the lower value of pore specific volume, the lower surface area and porosity than EP5-based supports.

3.3 Catalytic tests

The activity and stability of the palladium catalysts were researched during their multiple usage in the model Heck reaction (Scheme 3). The reaction entailed coupling iodobenzene with methyl acrylate, occurring in the solution of N-methylpyrrolidone (NMP) in the presence of triethylamine. The only observed product was methyl trans-cinnamate. Results in Table 4 show that for the samples of the EP5/[d₂m₂am][thb]/Pd and EP5/[d₂m₂am][thp]/Pd catalysts, where in both cases the organic anion is present in the used ionic liquid, a significant increase of activity can be seen in the second cycle of the reaction. This is probably caused by the increasing volume of the sample (swelling), hence allowing for an easier access to the active spots in the open pores of the catalyst. In the case of the catalysts based on EP5/[d₂m₂am][BF₄] and EP5/[mthb][BF₄] carriers which have an inorganic anion in their structure, a high activity in all reaction runs is observed. We ascribe the lower catalytic activity of the catalysts supported on the DER resins to fewer Pd sites being accessible to the reactants when these are attached to the polymer of lower microporosity. The better availability of the catalytic sites situated in a EP5 resin having a greater volume of micropores and higher surface area increases the activity of supported Pd systems.

The activities of most tested catalyst decrease with the subsequent usage probably due to the leaching of the palladium. We applied time-of-flight secondary ion mass spectrometry (TOF-SIMS) to investigate the deactivation process and to observe a deterioration of the Pd distribution on the surface of catalysts after a third reaction run in comparison with the fresh sample. This technique permits the evaluation of the chemical composition of the catalyst surface and the distribution of metal or other catalyst components on the surface. However, the quantitative analysis by TOF-SIMS method is relatively difficult. TOF-SIMS may be very useful in the studies of catalyst deactivation processes. In such a case changes in the composition of the upper layer of the investigated surface can be determined. Owing to TOF-SIMS, a considerable drop in the amount of surface accessible Pd was observed for EP5/[d₂m₂am][BF₄]/Pd and EP5/[d₂m₂am][thb]/Pd catalysts after the Heck reaction in comparison with the fresh samples (Table 5). The drop of Pd content was at safe level and slightly affects the catalysts activity within acceptable limits, and can be explained by the often reported enhancement of the activity for a catalyst containing mixed valence states of Pd(II) and Pd(0) [21]. Some metal leaching is often observed when supported catalysts are employed, usually higher in the first two runs and lower in the next ones. It mostly comes from the detachment of the weakly bound metal species via, for example, physical forces. These are still present after washing the catalyst during the preparation procedure. Observed decrease of activity during prolonged use can be also connected with poor chemical resistance of epoxy supports to polar solvents like NMP. However, this solvent was used for the model Heck reaction, which we chose only to preliminary characterization of catalytic properties of these catalysts. The performed catalysts showed good activity and stability comparing to other catalysts. For example, Cypryk et al. reported that the activity of the polysiloxane-supported Pd catalysts in the Mizoroki–Heck arylations of methyl acrylate, butyl acrylate, and acrylonitrile by iodobenzene and of styrene by bromobenzene was comparable to that of homogeneous PdCl₂(PhCN)₂ and maintained through five to ten reaction cycles [22]. Sudalai reported the use of Pd-exchanged montmorillonite clays in the Heck reaction. Investigated catalysts were reused three times without significant loss of activity [23].

The ionic liquid used at the curing stage has a decisive influence on the catalytic activity of the catalysts. This was also confirmed by the results of TOF-SIMS measurements of the EP5/[d₂m₂am][BF₄]/Pd and EP5/[d₂m₂am][thb]/Pd samples (Table 5). The disparity of the intensities of the palladium ions for the EP5/[d₂m₂am][BF₄]/Pd and EP5/[d₂m₂am][thb]/Pd catalysts just after the immobilization shows that more palladium complex is deposited on the carrier with an inorganic anion in the structure of the ionic liquid used as hardener.

To determine the nature of the active species, the effect of catalyst poisoning has also been investigated. To rule out the possibility of catalysis by microparticles (colloids or nanoclusters) of zero-valent palladium formed from leached and reduced Pd(II) precursor, a mercury poisoning experiment was carried out by adding mercury to the reaction mixture before the catalytic test. The mercury poisoning of metal(0) particles, by amalgamating the metal or adsorbing on the metal surface is a widely used test for the heterogeneity/homogeneity of catalysis [24]. An epoxide-supported palladium catalyst vigorously stirred with a large excess of metallic mercury for 24 h prior reaction shows almost the same catalytic activity for the Heck reaction as before poisoning. It is known that trace amounts of leached metal particles or soluble palladium complex can be very catalytically active [25]. As additional evidence for the identity of the true catalyst, hot filtration technique has also been employed, and no catalytic activity was observed, confirming the absence of active species (soluble Pd(0)) in the hot filtrate.