Introduction

The use of transition metal compounds containing elements such as Pd, Au, Ni, Cu, and Ru to catalyze different organic reactions is well known [1,2,3,4,5,6,7]. The Suzuki–Miyaura reaction, a reaction between a halogenated hydrocarbon and an arylboronic acid, was first reported in 1979 [8] and is widely used as a route to synthesize C–C bonds. It is mainly catalyzed by Pd owing to its high activity and selectivity and is one of the most widely accepted methods for the synthesis of pharmaceuticals [9]. The advantages of the Suzuki–Miyaura reaction include tolerance to aqueous media, functional group diversity and tolerance, high conversions, and mild reaction conditions.

Traditional reactions using homogeneous Pd complexes as catalysts have been described [10, 11]. The complexes exhibited high performance and stereoselectivity, but the formation of insoluble Pd black or aggregation of the insoluble species diminished its catalytic activity in the coupling reaction [12, 13]. In addition, it is very difficult to recycle and recover homogeneous catalysts as they remain in the reaction medium at low concentrations and cannot be separated from the synthesis product, which increases the cost of the process and potentially results in catalyst unavailability and product impurity [14]. Therefore, it is necessary to design economical and efficient environmental processes for the separation and reuse of catalysts. This approach has attracted enormous attention in the twenty-first century.

Using a heterogeneous catalyst can efficiently solve these problems and facilitate easy separation of the catalyst from the reaction mixture [15]. Pd has been immobilized on silica nanoparticles [16, 17], graphene [18], carbon nanotubes [19], metal–organic frameworks [20, 21], and metal oxides [22, 23]; however, these catalysts are expensive and usually require complicated and tedious preparation procedures. To overcome these drawbacks, the development of practical, economical, and environmentally friendly supported Pd catalysts is required.

The anchoring of catalysts to magnetic nanoparticles (MNPs) has been extensively studied in the recent years [24,25,26]. Iron oxide nanoparticles exhibit many advantages, including a high surface-to-volume ratio and facile magnetic recovery [27]. The magnetically recoverable catalyst has attracted significant attention because its recovery process is more economical and greener compared to centrifugation and filtration. Magnetite is one of the most commonly used metal oxide nanoparticles owing to its magnetic properties and biocompatibility. Pd-functionalized magnetite nanoparticles have been reported for application in the Suzuki–Miyaura reaction [16, 22, 28]. In general, the functionalization process involves difficult and time-consuming ligand synthesis and additional costs. To minimize the use of ligands in the Pd incorporation process, facile procedures based on sequestering processes were studied on materials capable of reducing Pd by oxidizing the surface of the sequestering material. Li et al. studied the sequestration of Ni(II) in a solution using Fe nanoparticles [29]. The process involved coordination of the metal ion to the surface of the reactive Fe, which acts as a reductant toward the less reactive metal ions. Applying this concept, FexOy@Fe was synthesized to recover a variety of ions, such as Co2+, Cu2+, Ni2+, Rhx+, Pd2+, Ag+, and Pt4+, from solution [30, 31].

The use of heterogeneous catalysts in the Suzuki–Miyaura reaction requires relatively elevated reaction temperatures, which can be achieved using a local heating procedure. The magnetic properties of magnetite nanoparticles enable the conversion of electromagnetic induction (EMI) into heat. The use of induction heating offers remarkable benefits compared with other heating methods, such as microwave irradiation and thermal baths, owing to the penetration power of the magnetic field [32]. In addition, EMI can quickly and directly induce high temperatures in catalyst particles [33, 34]. Thus, heating the system by EMI takes less time compared with external heating methods that involve a conventional heat source, such as a thermal bath, and provides greater control and safety over the reaction. The application of PEG- or oleic-acid-coated Fe@Fe3O4 nanoparticles with Pd adsorbed on the surface in Suzuki–Miyaura and Sonogashira reactions has been reported recently, demonstrating the feasibility of this approach [35]. Our study demonstrates the novel use of uncoated Pd/Fe@Fe3O4 nanoparticles with excellent catalytic activity in the Suzuki–Miyaura reaction heating by EMI. Nanoparticles are usually coated to prevent agglomeration and loss of their electrical and magnetic properties; however, we show that vigorous agitation does not result in a loss of properties, making it possible to apply lower frequency magnetic fields than those required for coated nanoparticles. The effect of different heat sources and temperatures on the efficiency of the heterogeneous catalytic Suzuki–Miyaura reaction was analyzed, and the results indicated that inductive heating improved the catalyst efficiency. The recovery and reuse of the catalysts was also tested.

Experimental section

Catalyst synthesis

The catalyst was prepared in two stages according to a previously reported procedure [31, 36]. Briefly, a solution of NaBH4 (2.24 g, 59 mmol) in 30 mL of deionized (DI) water was added dropwise (for a period of approximately 15 min) to a solution of FeCl3·6H2O (5.67 g, 21 mmol) in DI water (400 mL). The mixture was mechanically stirred at 600 rpm for 4 h, to form a black precipitate. The nanoparticles were separated using a magnet, and the supernatant was discarded. The solid was washed and sonicated with DI water three times, and finally with ethanol to produce a transparent solution. Finally, the resulting black paste was centrifuged and vacuum-dried overnight at 25 °C to yield Fe@FexOy powder (1 g). In the second stage, to obtain the Pd/Fe@FexOy catalyst, 200 mg of Fe@FexOy was reacted with Pd(NO3)2 (230 mg, 1 mmol) of in 250 mL of DI water at a pH of 10.5 under an argon atmosphere. The mixture was sonicated for 4 h increasing the temperature to 45–50 °C. The reaction mixture was subsequently centrifuged for 10 min at 10000 rpm, and the black precipitate was washed with distilled water several times, followed by vacuum drying for 12 h at 25 °C. The synthesis and purification steps were carried out in air, although the catalyst was stored in an inert atmosphere to prevent oxidation over time.

Material characterization

Phase identification analysis of the MNPs was conducted by X-ray diffraction (XRD) using a Bruker D8 powder diffractometer equipped with a primary monochromator and an ultrafast Lynxeye XE-T multichannel detector with Cu Kα1 radiation. The patterns were collected in the 2θ range of 5–80°, with an angular increment of 0.02° and an increment time of 1 s. The diffractograms were analyzed using the PANalytical X’Pert HighScore program, and their profile fitting was carried out to obtain more reliable peak parameters after deconvolution of the overlapped peaks. Peaks were assigned based on the International Center for Diffraction Data (ICDD) reference considering the peak positions and intensities. A semiquantitative XRD analysis of the composition based on the reference intensity ratio (RIR) was also performed. The morphology of the catalyst was studied using a transmission electron microscope (JEOL JEM 1010) operated at 100 kV. The samples were prepared by placing a single droplet (10 μL) of an aqueous solution of MNPs onto a copper grid coated with a carbon film. The grid was dried in air for several hours at room temperature (RT). Mössbauer spectroscopy was used to analyze the Fe phases in the samples. The spectra were recorded at RT in the triangular mode using a transmission spectrometer with a 57Co/Rh source. Spectral analyses were performed by nonlinear fitting using the NORMOS program [37]. Each spectrum was fitted by the convolution of subspectra corresponding to as many different crystallographic positions of Fe that could be identified in the material by XRD. For each Fe site, the isomer shift (δ), quadrupole splitting (ΔEQ), and hyperfine magnetic field (H) were fitted. We fixed equal full width at half maximum of the Lorentzian lines of each component, and the intensity ratio 3:2:1:1:2:3 for each sextet. The combination of these three parameters allows for the identification of a particular compound. The energy calibrations were performed using an α-Fe (6 µm) foil. The composition of the material was also examined by inductively coupled plasma optical emission spectroscopy (ICP-OES) using a Perkin-Elmer Optima 2100 DV system to determine the Pd loading on the catalyst before and after its use and the possible leaching of Fe and/or Pd; each sample was analyzed three times. X-ray photoelectron spectroscopy (XPS) was used to characterize the surface chemistry of the samples and the oxidation states of Fe and Pd. XPS experiments were performed in a UHV chamber with a base pressure of 10–10 mbar equipped with a hemispherical electron energy analyzer (SPECS Phoibos 150 spectrometer) and a 2D delay-line detector (Surface Concept), using a non-monochromatic X-ray source of Al Kα radiation (1486.6 eV) operated at 300 W. XPS spectra were recorded at the normal emission take-off angle, using an energy step of 0.50 and 0.10 eV and a pass-energy of 40 and 20 eV for survey spectra and detailed core-level regions (Fe 2p, Pd 3d, C 1 s, Fe 3p, Pd 4p, and O 1 s), respectively. The charging effects accumulated in the photoemission process were compensated using a low energy electron flood gun, in which a wide beam illuminates the sample surface during the measurements. On the one hand, this allows the binding energy calibration to be referenced to the C 1 s photoelectron peak at 284.5 eV, which was checked before and after the measurement of every selected core-level transition. On the other hand, and more importantly, charge compensation also provides the correct lineshape of a given core-level emission, which is significantly affected in the absence of flood-gun irradiation and unfortunately leads to contradictory conclusions in the determination of the oxidation states present. Data processing was performed using CasaXPS software (Casa Software Ltd., Cheshire, UK). Spectra are displayed after the subtraction of the contribution of the Al-Kα satellite emission; in some cases, the spectra were normalized to the maximum intensity to highlight the lineshape differences, which provides direct valuable insight into the oxidation states. 1H-NMR spectroscopy was performed on a Bruker Avance 300 MHz instrument to determine the conversion of the Suzuki–Miyaura reactions. To calculate the conversion, each 1H-NMR spectrum was integrated relative to the internal standard 1,3,5-trimethoxybenzene. The samples were prepared by dissolving 0.2 mL of the crude reaction in 0.4 mL of deuterated dimethylsulfoxide (DMSO-d6).

Suzuki–Miyaura coupling reactions

The efficiency of the Pd/Fe@Fe3O4 catalyst under magnetic induction heating was tested in the catalyzed coupling of phenylboronic acid with 4-bromophenol and 4-bromopyridine. The conversions of both reactions were compared using two heating sources: a conventional thermal bath and an alternating magnetic field. A schematic of the reactions and the magnetic system is shown in Fig. 1.

Figure 1
figure 1

Schematic of the Suzuki–Miyaura reaction conducted under induction heating

Full size image

For reaction 1a, 4-bromophenol (1.8 mmol), phenylboronic acid (2.7 mmol), and 1,3,5-trimetoxybenzene (1.8 mmol) as an internal standard were dissolved in 5 mL of ethanol. K2CO3 (3.6 mmol) was dissolved in 5 mL of DI water and added to the mixture with 30 mg of Pd/Fe@FexOy (5 mol% Pd). The reaction mixtures were mechanically stirred in a glass reactor in air over the required reaction times.

For reaction 1b, 4-bromopyridine hydrochloride (1.8 mmol), phenylboronic acid (2.7 mmol), and 1,3,5-trimetoxybenzene (1.8 mmol) were dissolved in 5 mL of ethanol. K2CO3 (7.2 mmol because 4-bromopyridine was added as hydrochloride) was dissolved in 5 mL of DI water and added to the mixture with 30 mg of Pd/Fe@FexOy (5 mol% Pd). The reaction mixture was treated as described in reaction 1a.

The conversions of the reactions were determined over time by comparing the concentrations of the final products through 1H-NMR analysis at the start of the reactions and after different reaction times. Some of these 1H-NMR spectra are provide in the Supplementary Information.

Magnetic induction heating was performed using an electromagnetic coil of an EASYHEAT LI 8310 apparatus (Ambrell) equipped with a water-cooling system. The fourth-turn coil diameter was 7.5 cm with a depth of 4.5 cm. Alternating currents between 200 and 550 A were applied at a constant frequency of 223 kHz to reach the desired temperature in the reaction medium (see Supplementary Information). A temperature signal was supplied to a temperature controller (Yokogawa) that could regulate the intensity of the alternating current flowing through the magnetic inductor coil to maintain the reaction at a constant temperature. The reaction temperature was measured using an optical fiber.

Results and discussion

Catalyst characterization

To obtain the Pd catalyst, an Fe@Fe3O4 solid support was fabricated, which acted as a reducing agent for the Pd2+ solution and resulted in its attachment to the surface. Figure 2 shows a characteristic X-ray diffractogram of the Fe@Fe3O4 precursor. All diffraction peaks match the standard JCPDS no. 01-087-0722, 01-089-2355 and 01–075-1594, which correspond to Fe, Fe3O4, and FeOOH, respectively. Figure 2 shows the positions of the Bragg peaks of the phases present as well as the difference between the observed and calculated patterns. Semiquantitative XRD analysis revealed the formation of 42 wt.% zerovalent Fe, 37 wt.% Fe3O4, and 21 wt.% FeOOH.

Figure 2
figure 2

XRD diffractogram of Fe@Fe3O4. The observed, calculated, and difference patterns are shown. Semiquantitative XRD analysis and the Bragg positions of the phases present are included

Full size image

Transmission electron microscopy (TEM) analysis of Fe@Fe3O4 (Fig. 3a) shows the formation of two types of nanoparticles. The larger particles have sizes of 20–100 nm and correspond to Fe and Fe3O4; these particles have a cuboid appearance and form aggregated, cross-linked chains owing to their magnetic nature (Figure S1, Supplementary Information) [30]. The smaller particles (with sizes less than 10 nm) would correspond to Fe(0) and are distributed over the surface of the Fe@Fe3O4 (indicated by arrows in Fig. 3a). XPS was used to analyze the surface of the material in detail and corroborate this hypothesis, as the previously reported results indicate the formation of an amorphous iron core and a shell formed mostly by Fe3O4 and FeOOH [31, 36]. Figure 3b shows the Fe 2p core-level emission corresponding to the Fe@Fe3O4 sample. The XPS spectrum shows the photoelectron emission of the complex lineshape mainly dominated by two wide peaks corresponding to the spin–orbit 3/2 and 1/2 doublet. The oxidation states can be determined by comparing the binding energy positions of both peaks, their energy splitting, and the presence/absence of characteristic satellites, in comparison with pure iron and standard iron oxide compounds [38, 39]. In our case, the analysis of the multiplet structure only reveals a very weak emission above the 2p1/2 component and a negligible signal between the doublet; it is noteworthy that the emission at lower binding energy comes from Fe(0), as discussed below. In accordance with this result, together with the binding energy values of 2p3/2 (710.2 eV) and splitting (13.6 eV), we suggest that the surface oxide signal arises mainly from Fe3O4. In addition, from the detailed analysis of the spectral lineshape, a prominent emission at 706.7 eV (arrow on the right in the figure) significantly shifted to a lower binding energy than the oxide can be observed. According to this binding energy value, the signal can be attributed to 2p3/2 from zerovalent Fe, which also has its spin–orbit counterpart (arrow on the left of the figure) shifted by 13.10 eV. Therefore, the presence of both components and the characteristic splitting provides a straightforward confirmation of the existence of a significant amount of Fe(0) in the sample surface.

Figure 3
figure 3

a TEM micrograph and b Fe2p XPS (Al-Kα) spectrum of Fe@Fe3O4

Full size image

Fe@Fe3O4 can act as a reducing agent, which adsorbs and reduces Pd2+ to zerovalent Pd during the synthesis process. Figure 4 shows the diffractogram obtained for the Pd/Fe@Fe3O4 sample, in which peaks corresponding to Fe, Fe3O4, FeOOH, and Pd (JCPDS no. 03-065-6174) are observed, demonstrating the successful anchoring of Pd to the substrate. Semiquantitative XRD analysis indicated that the sample contained 24 wt.% Pd, 5 wt.% Fe, 40 wt.% FeOOH, and 31 wt.% Fe3O4. The amount of Pd determined through semiquantitative analysis corresponded with ICP-OES results obtained for the digested Pd/Fe@Fe3O4 sample (35 ± 1 wt.% Pd).

Figure 4
figure 4

XRD diffractogram of Pd/Fe@Fe3O4. The observed (dots), calculated (blue line) and difference patterns are shown. Semiquantitative XRD analysis and the Bragg positions of the phases present are included

Full size image

The Pd/Fe@Fe3O4 TEM image (Figure S2) shows an increase in needle-like structures that are characteristic of FeOOH [40]. In addition, larger particles, probably corresponding to Fe3O4, and darker particles of Pd and Fe are also shown.

Figure 5a displays the normalized Fe 2p spectra of Fe@Fe3O4 and Pd/Fe@Fe3O4. Upon comparison, important lineshape differences can be observed clearly. First, the characteristic emission of Fe(0) is missing on the surface of the Pd/Fe@Fe3O4 sample. Second, there is a shift to higher binding energy values for both 2p3/2 and 2p1/2 oxide peaks, which is tentatively indicative of Fe3+ enrichment. In addition, a multiplet structure appears at higher energy values for each doublet peak. Following a similar analysis to that of Fig. 3b, the separation between the satellites and main peaks, 7.6 and 8.5 eV, respectively, is consistent with the majority presence of Fe3+, which can be understood as additional FeO–OH bonds present on the surface of this sample [39]. This result agrees with the increased emission of the OH component in the O 1 s core-level spectrum located at higher energy values from the signal ascribed to oxygen in the metallic oxide lattice (not showed). Figure 5b shows the XPS data of the Pd 3d core-level of Pd/Fe@Fe3O4. In contrast to the Fe species, the Pd 3d spectrum is basically composed of the well-defined and narrow 3d5/2 and 3d5/2 peaks, located at the right binding energy of 334.8 eV and with an energy splitting of 5.25 eV, ascribed to Pd(0) [36]. Note that there exists a subtle emission on the higher-energy side of the doublet owing to the residual formation of native oxide due to air exposure of the sample prior to XPS measurements. Therefore, a chemical reduction from Pd2+ to Pd(0) occurred during the synthesis of Pd/Fe@Fe3O4.

Figure 5
figure 5

XPS (Al-Kα) spectra: a normalized Fe 2p spectra of Fe@Fe3O4 and Pd/Fe@Fe3O4 b Pd 3d core level of Pd/Fe@Fe3O4

Full size image

The Mössbauer spectra of the catalyst and support, obtained at different stages of the synthetic process, are shown in Fig. 6. The Fe@Fe3O4 spectrum was fitted by overlapping a central doublet (with δ = 0.36 mm/s and ΔEQ = 0.7 mm/s) and a sextet (with δ = 0.02 mm/s, ΔEQ = 0.0 mm/s and H = 33.1 T), corresponding to FeOOH and Fe(0) detected by XRD, respectively. In addition, a broad sextet, corresponding to the Fe3O4 phase according to the XRD results, was also present. This subspectrum was fitted as a hyperfine field distribution (inset in Fig. 6a) that reflects the randomness of the environment of the Fe atoms owing to the wide size distribution of the nanoparticles or poor crystallinity. The most probable hyperfine magnetic field value is 25.2 T, which is much lower than values corresponding to bulk-sized magnetite. The spectrum of bulk-sized magnetite is formed by two sextets, of 45 and 49 T, which correspond to Fe in the tetrahedral and octahedral sites of the spinel lattice, respectively. The relative areas of the subspectra in the Fe@Fe3O4 sample were 31% Fe(0), 16% FeOOH, and 53% Fe3O4. Since the area of each Mössbauer subspectrum can be related to the population of Fe atoms, assuming that the Lamb-Mössbauer factor is identical for all sites and compounds, the sample composition determined by Mössbauer spectroscopy coincides with that semi-quantitatively determined by X-ray diffraction. The spectrum of Pd/Fe@Fe3O4 (Fig. 6b) is formed by the FeOOH doublet, Fe(0) sextet, and two additional sextets with H = 43.0 T and 48.4 T that correspond to magnetite. The transformation of the magnetic field distribution in the Fe@Fe3O4 spectrum into the two well-defined sextets characteristic of magnetite indicates that the increase in temperature due to Pd2+ reduction, which occurs during sonication of the solution, is sufficient for the growth of nanoparticles. In contrast, a decrease from 31 to 11% in the area of the Fe(0) subspectrum and an increase from 16 to 33% in the area of the FeOOH subspectrum are observed. Therefore, it can be concluded that under these reaction conditions, Fe(0) is mainly oxidized to FeOOH. The mechanism of incorporation of Pd into the catalyst could be explained by the reduction of Pd2+ with Fe(0) on the surface, as well as by the coordination of the metal in solution with the exposed hydroxyl groups on the particle surface, followed by reduction by the Fe(0) that contains the core. Both processes can occur simultaneously.

Figure 6
figure 6

Mössbauer spectra of a Fe@Fe3O4 and b Pd/Fe@Fe3O4

Full size image

Suzuki–Miyaura coupling reactions

The Suzuki–Miyaura reaction was performed as a model to compare the efficiency obtained by thermal-bath heating and heating by magnetic induction. The temperature of the reaction solution was considered as the reaction temperature (see Figure S3).

To optimize the reaction time required for maximum conversion, aliquots of the mixture taken from the reaction of phenylboronic acid with 4-bromophenol at 50 °C were analyzed at 2, 4, and 6 h (Table 1). The results indicate that for both heating systems, there is a slight increase in the conversion with an increase in the reaction time from 2 to 4 h; however, after 4 h, a further increase in the reaction time to 6 h resulted in slight fluctuations in conversion, rather than progress. Based on these results, reactions 1a–b were conducted over 4 h. The experiments were carried out three times, and the data are presented as the average of the independent reactions.

Table 1 Conversion (%) of Suzuki–Miyaura coupling reactions
Full size table

Table 1 shows the conversions of the coupling reactions between phenylboronic acid and 4-bromophenol and between phenylboronic acid and 4-bromopyridine as a function of the reaction temperature, using both heating methods.

The data indicate that heating by means of a magnetic inductor results in higher conversions for both reactions compared to that achieved using thermal bath. An increase in the temperature results in a slight increase in the substrate conversion under thermal-bath and induction heating conditions. On average, for the reaction between phenylboronic acid and 4-bromophenol, 25–30% higher conversions were observed with magnetic induction heating compared to thermal-bath heating. Although the solution temperature was identical, induction produced a higher temperature on the surfaces of the nanoparticles. We can assume that the temperature on the nanoparticle surfaces increases by more than 40 °C by comparing the conversion achieved at 70 °C using a thermal bath with that achieved at 30 °C with magnetic induction heating. It has been demonstrated that the local heating of MNPs can generate a gradient of 45 °C at a distance of 0.5 nm from the surface of the MNPs in order to decay exponentially with distance [41]. Published studies suggest that reactions catalyzed by solid-supported Pd occur in solution owing to leaching and subsequent redeposition of Pd [3]. Although we may assume that because of the increased temperature on the surface of the nanoparticles, where Pd is located, the reaction could occur superficially, we cannot rule out the possibility of some dissolution of Pd in the medium and subsequent redeposition. The temperature varies at a certain distance from the nanoparticles, but this decay is insignificant at a short distance. More precise experiments on the variation of temperature from the nanoparticles to the medium and the gradient in the concentration of Pd could provide insight into the mechanism. Regardless of the mechanism taking place, it is evident that the catalytic reaction occurs close enough to the particle surface for local heating to accelerate the reaction rate owing to the increase in temperature relative to the bulk solution. Although the reaction is temperature dependent, when the surfaces of the nanoparticles reach a certain temperature, an increase in the reaction-medium temperature does not considerably improve the conversion efficiency.

A relatively high conversion was achieved for the Suzuki–Miyaura coupling of phenylboronic acid and bromophenol, even at 30 °C, but the reaction with 4-bromopyridine requires a higher temperature; therefore, it does not occur at 30 °C in a thermal bath, and the yield obtained by magnetic induction heating is only 6% on average (Table 1). The conversions obtained at 70 and 80 °C for the reaction between phenylboronic acid and 4-bromopyridine heated by magnetic induction were 50% higher than those obtained at the same temperatures in a thermal bath (Table 1).

The heterogeneous Pd/Fe@Fe3O4 catalyst proved to be efficient in the Suzuki–Miyaura reactions studied, especially under magnetic induction heating. Subsequently, the reusability of the catalysts was assessed. Table 2 shows the evolution of the efficiency of the catalyst over three Suzuki–Miyaura coupling reaction cycles between phenylboronic acid and 4-bromophenol at 50 °C. Between each reaction cycle, the catalyst was recovered with a magnet, washed with distilled water and ethanol several times, and dried under vacuum.

Table 2 Conversion achieved for the reaction between 4-bromophenol and phenylboronic acid performed during 4 h at 50 °C under magnetic field induction (223 kHz)
Full size table

To determine the reason for the decrease in the reactivity of the catalyst in successive cycles, XPS measurements of the samples were carried out after reuse. Figure 7a, b compares the XPS spectra of Pd 3d and C 1 s corresponding to Pd/Fe@Fe3O4 fresh and reused samples. Note that the detailed XPS lineshape analysis reveals no significant differences between Fe 2p (not shown but similar to that displayed in Fig. 5a) and Pd 3d spectra in both samples. In addition, the Pd 3d spectrum includes more prominent asymmetric peaks with a minor contribution of surface native oxide, most likely because the surface of the nanoparticles is covered with a higher amount of C (see below), which reduces their exposure to air after preparation. The only difference between the spectra is the absolute emission intensity, in which signals are attenuated by the higher amount of C present on the surface. However, this modification did not change the relative surface atomic concentration ratio, suggesting low leaching of Pd. ICP-OES analysis of the catalyst revealed a slight variation in the composition of the MNPs with increased use; the Fe/Pd ratio changed by less than 5% after each cycle. In contrast, there exists a significant increase in the C concentration on the surface, which might be responsible of the poison effect observed in the catalytic behavior of the samples after use.

Figure 7
figure 7

XPS (Al-Kα) spectra of Pd/Fe@Fe3O4 fresh and reused. a Pd 3d core level, b C1 s core level of the catalyst

Full size image

The loss in catalyst efficiency observed for Pd/Fe@Fe3O4 over three cycles is slightly greater than that reported for catalysts consisting of Pd supported on coated Fe3O4, such as Fe3O4/PEG/Pd [35] or Fe3O4/DA-Pd [22]. However, the use of uncoated magnetite offers the advantage of a facile and economical synthetic process.

Conclusions

We have shown that it is possible to perform Suzuki–Miyaura reactions under magnetic induction heating using Pd-functionalized MNPs without the addition of an organic coating, which saves time and resources and makes the process more sustainable.

Magnetic induction increases the efficiency of the catalytic process by approximately 25–35% compared to the reaction in which conventional heating was applied; this is due to the temperature on the surface of the MNPs being higher than that of the solvent. This allows the reaction to proceed without the need to heat the solvent to the required temperature. Furthermore, owing to the magnetic properties of the nanoparticles, the catalyst can be easily recovered and reused. However, further studies are required to increase the number of reaction cycles in which the recovered catalyst can be used.