Ultrastable Magnetic Nanoparticles Encapsulated in Carbon for Magnetically Induced Catalysis.

Magnetically induced catalysis using magnetic nanoparticles (MagNPs) as heating agents is a new efficient method to perform reactions at high temperatures. However, the main limitation is the lack of stability of the catalysts operating in such harsh conditions. Normally, above 500 °C, significant sintering of MagNPs takes place. Here we present encapsulated magnetic FeCo and Co NPs in carbon (Co@C and FeCo@C) as an ultrastable heating material suitable for high-temperature magnetic catalysis. Indeed, FeCo@C or a mixture of FeCo@C:Co@C (2:1) decorated with Ni or Pt-Sn showed good stability in terms of temperature and catalytic performances. In addition, consistent conversions and selectivities regarding conventional heating were observed for CO2 methanation (Sabatier reaction), propane dehydrogenation (PDH), and propane dry reforming (PDR). Thus, the encapsulation of MagNPs in carbon constitutes a major advance in the development of stable catalysts for high-temperature magnetically induced catalysis.

Introduction

Magnetic heating is an attractive alternative to conventional heating that has been gaining attention in catalysis during the last years. Indeed, magnetically induced catalysis using magnetic nanoparticles (MagNPs) has been successfully applied in solution[1−6] as well as in gas-phase reactions.[7−14] It is based on the principle that ferromagnetic materials can release heat to the environment through hysteresis losses in the presence of an oscillating magnetic field. The main advantage of magnetic heating relies on the very fast warming up of the system, which makes this system adapted to the storage of intermittent energies, as well as on the direct heating of the catalyst without the need for heating the whole reactor.[15] This makes magnetic heating a promising method to perform reactions that occur at medium/high temperatures, such as CO2 methanation (Sabatier reaction), propane dehydrogenation (PDH), or propane dry reforming (PDR). The heating power of MagNPs is commonly quantified by the specific absorption rate (SAR), which measures the energy released per unit of mass upon magnetic excitation. In particular, when applying an alternating magnetic field with a frequency f and an amplitude μ0Hrms to MagNPs, the area of the hysteresis loop is proportional to the dissipated energy.[16]

The current context of global warming has stimulated the development of catalytic reactions that utilize CO2 as a platform molecule for the production of fuels or chemicals, with the objective of reducing the global carbon footprint. CO2 hydrogenation (the so-called Sabatier reaction) (eq ) is one of the main synthetic routes to produce methane. Although it is an exothermic reaction, it is usually performed above 250–300 °C because of the high kinetic barrier for the activation of the CO2 molecule,[17−19] and development of catalysts able to work at lower temperature is an important issue.[20] Even if the Sabatier reaction is presently exploited at an industrial scale, it requires an important energy supply to reach operating conditions. This is one of the limitations of power-to-gas (P2G) technology, where the excess of electricity is used to produce methane.[21] Therefore, energetically efficient processes that can be rapidly started or stopped are of high interest because of the possibility to carry out the hydrocarbon production only when the weather conditions allow it. In this context, magnetic heating can be an interesting alternative for intermittent energy storage or P2G. On the other hand, a catalyst displaying good magnetic heating performances would enable it to more efficiently perform endothermic reactions that require high-energy supplies. As the catalyst itself, and not the reactor, is the heating source, the energy transfer is rapid and efficient. Another reaction of industrial interest is the endothermic dry reforming of propane (PDR), which can be a solution to the increasing demand of synthesis gas (CO, H2) or the hydrogen production for industrial applications.[22] Similarly, dehydrogenation of light alkanes (e.g., propane) has received great attention because of the growing need for light olefins for the production of chemicals.[23] These reactions occur at high temperatures, above 600 °C in the case of propane dehydrogenation (PDH; eq ) and from 550 to 900 °C for propane dry reforming (PDR; eq ).[22,23] Dry reforming is generally catalyzed by Ni-based catalysts,[24] whereas PDH is catalyzed by Pt-based catalysts reaching excellent selectivities to propene.[25−27]The Curie temperature (Tc) is the temperature at which a ferromagnetic material becomes superparamagnetic, hence losing its magnetic heating capacity. Thus, the composition of the heating agent is of high importance as the Tc of the MagNPs defines the highest temperature that can be theoretically attained by magnetic induction. Therefore, a precise control of the Tc of the heating agents will determine their further applicability in the magnetically induced catalysis of various reactions, when associated with appropriate catalytically active phases. In a recent publication, it has been reported that iron and iron carbide NPs are inefficient to activate high-temperature catalysis by magnetic heating, whereas FeCo NPs reach temperatures high enough to significantly activate PDR and PDH.[10] However, especially in cases where high temperatures are required, a main limitation of magnetically induced reactions is the lack of stability of the catalysts that need to be active for long reaction times under harsh conditions. It was observed that, for temperatures above 500 °C, massive sintering of both heating and catalytic nanoparticles occurred, together with carbon deposition, which led to deactivation of the catalyst.[10] These processes reduce both the catalytic and heating properties of the NPs and therefore the magnetically induced catalyst lifetime. In this respect, metal nanoparticles (MNPs) encapsulated in carbon have emerged as efficient sinter-resistant materials thanks to their high thermal stability and confinement properties.[28−30]

Herein, we present novel heating agents for magnetically induced catalysis based on magnetic Co and FeCo nanoparticles encapsulated in carbon (FeCo@C and Co@C). Encapsulation in carbon not only protects them from full oxidation, as was recently described,[28,29] but also confers to the materials the stability necessary to avoid sintering at such high temperatures. FeCo@C and Co@C were fully characterized by common techniques, such as transmission electron microscopy (TEM), high-resolution TEM (HRTEM), energy-dispersive X-ray spectrometry (EDX), inductively coupled plasma (ICP), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), X-ray powder diffraction (XRD), temperature-programmed reduction (TPR), and vibrating-sample magnetometer (VSM). In this work, we demonstrate that FeCo@C and Co@C (5 wt %) can be used as heating agents in magnetically induced reactions taking place at temperatures above 600 °C, maintaining their morphology and heating capacity. These ultrastable materials (FeCo@C and Co@C) have been decorated with Ni or Pt–Sn and evaluated in methanation, propane dry reforming and dehydrogenation of propane, presenting high stability in all the cases without modification of their catalytic performances.

Results and Discussion

Synthesis and Characterization

The methodology to encapsulate MagNPs in carbon is based on a two-step procedure. FeCo prepared by a previously reported procedure[31,32] and Co nanoparticles prepared by adaptation of the same procedure (see Experimental Section, section 2) were supported on activated carbon (FeCo/C and Co/C; 5 wt %). The MagNPs were then encapsulated in carbon through a pyrolysis process (600 °C, 2 h, 10 °C/min) to afford FeCo and Co NPs covered by a narrow carbon layer (Figure ; FeCo@C and Co@C). The detailed synthetic procedure is described in the experimental section (see Experimental Section, section 2). The metal contents of FeCo@C and Co@C were determined by ICP atomic emission spectroscopy (ICP-AES). Specifically, FeCo@C (Fe, 2.6 wt %; Co, 2.3 wt %; corresponding to a NP composition of Fe54:Co46) and Co@C (Co, 4.8 wt %) presented metal contents very close to the theoretical ones (5 wt %).

Figure 1

Synthesis of FeCo@C and Co@C following a two-step synthetic procedure: (i) formation of FeCo/C and Co/C by the immobilization of MagNPs on activated carbon (24 h stirring at room temperature) and (ii) pyrolysis of FeCo/C and Co/C (2h at 600 °C under a N2 flow, 10 °C/min).

TEM analysis of FeCo and Co NPs before encapsulation show spherical and well-distributed nanoparticles, with a good dispersion and a main diameter of ca. 10–11 nm (see Figures S1–S3). The size and morphology of FeCo@C (10.1 ± 2.0 nm; Figure a–c) and Co@C (10.5 ± 1.9 nm; Figure d–f) are very similar to those found in the NPs before encapsulation.

Figure 2

TEM micrographs and illustrations (insets) of (a, b) FeCo@C and (d, e) Co@C with their corresponding size histograms ((c)FeCo@C and (f) Co@C).

The morphology, composition, and crystallinity of FeCo@C and Co@C were studied by high-resolution TEM (HRTEM), energy-dispersive X-ray spectrometry (EDX), and scanning transmission electron microscopy (STEM) using a high-angle annular dark-field detector (HAADF). HRTEM micrographs indicate the existence of crystalline NPs encapsulated by a carbon layer with a thickness between 2 and 5 nm (Figure a, b and d, e). This carbon layer possesses a double function: (i) protecting the magnetic FeCo and Co NPs from full oxidation by air, and (ii) preventing their sintering under the high temperatures reached under catalytic conditions. However, the carbon layer is not continuous over the whole nanoparticle, as it presents some gaps or cracks (Figures b, e). The presence of these cracks was corroborated by a simple digestion experiment.[33] More specifically, Co and FeCo NPs encapsulated in carbon were dissolved in a H2SO4 solution suggesting that the NPs were not totally covered by carbon (for more details, see the Experimental Section, section 6). The carbon-encapsulated Co nanoparticles (Co@C) exhibit an fcc cobalt core with a lattice spacing of 2.06 Å that corresponds to the (111) plane and cobalt oxide (CoO) regions on their surface with a fringe spacing of 2.42 Å, which belong to (113) plane (Figure c). Similarly, the core of FeCo@C NPs presents a bcc structure with a lattice spacing of 2.03 Å with some oxide at the surface (Figure f). The cracked carbon layer explains the partial oxidation of Co@C and FeCo@C NP surface after exposure to air. In both systems, the space between the carbon layers is around of 3.6–3.4 Å (Figures c, f), which corresponds to a turbostratic ordering of the carbon.

Figure 3

HRTEM micrographs of (a–c) Co@C and (d–f) FeCo@C. The cracks of the carbon layer are highlighted by red arrows (b, e). The lattice spacing of Co and FeCo NPs are in white and the separation of the carbon layers in red (c, f).

Figures S4 and S5 show STEM-HADDF images and EDX analysis of Co@C and FeCo@C, respectively. EDX mapping of carbon, cobalt, and oxygen elements within Co@C (Figure S4a–d) confirms the partial oxidation of the enclosed-Co NPs. Elemental mapping of FeCo@C demonstrates the bimetallic nature of these NPs (see Figure S5a–d), with an atomic composition of Fe47:Co53 (see Figure S6), which agrees with the theoretical composition (Fe50:Co50) and matches with the ICP results obtained (vide supra). In the EDX composition profiles, we can observe that both MagNPs are embedded either in a thin carbon layer (e.g., 1–2 nm, Figures b and S4e) or in a thicker one (e.g., 4–5 nm; Figures e and S5e).

Figure S7 shows X-ray diffraction (XRD) diffractograms for Co@C and FeCo@C. Both materials present a (002) peak at 2θ = 26.4° with an interlayer separation of d = 3.36 Å, which corresponds to the carbon material. In addition, Co@C XRD exhibits fcc-Co and CoO peaks with relatively low intensities due to the low metal loading (∼5%). FeCo@C XRD also presents some minor peaks, which correspond to bcc FeCo and FeCo oxides (FeCoO). XRD confirms the partial oxidation of the MagNPs encapsulated by the cracked carbon layer when they are exposed to air. These results are in good agreement with HRTEM and EDX observations. Raman spectra of Co@C and FeCo@C NPs (see Figure S8) display two major bands at 1313–1307 cm–1 (D peak) and 1600–1595 cm–1 (G peak), associated with carbon materials.[34] The high intensity of the D peak is related to the large percentage of disorder present in these carbon materials. The region between 200 and 800 cm–1 of Co@C spectrum exhibits the typical vibrational modes of Co3O4 (3F2g, Eg, and A1g) at 194 (F2g), 474 (Eg), 517 (F2g), 619 (F2g), and 691 (A1g) cm–1.[35] The FeCo@C spectrum also shows the FeCo2O4 vibrational modes but with a lower intensity and slightly shifted to minor frequencies 189 (F2g), 465 (Eg), and 672 (A1g).[36]

The oxidation state and reducibility of these carbon-encapsulated MagNPs was investigated by X-ray photoelectron spectroscopy (XPS) and temperature-programmed reduction (TPR). By TPR we found that the reduction temperatures for Co@C and FeCo@C were 320 and 351 °C, respectively (see Figures S9). Figure S10 shows the Co 2p area of Co@C before and after reduction by H2 at 400 °C. The as-synthesized Co@C spectrum presents a main peak at 780.7 eV that can attributed to CoO species. After reaction at 400 °C under a H2 atmosphere for 4 h, the CoO surface was totally reduced to Co0 because the spectrum only exhibits a single signal at 778.4 eV, characteristic of metallic cobalt.[37] A similar behavior was observed for FeCo@C (see Figure S11), where we can see a clear reduction of Fe and Co in Fe 2p and Co 2p regions, respectively. These results confirm that the surface of the as-synthesized samples is fully oxidized but that it can be reduced under reductive conditions, which might increase the magnetic heating capacity of the materials under operating catalytic conditions.

The magnetic properties of Co@C and FeCo@C have been measured by vibrating sample magnetometry (VSM). Saturation magnetization (Ms) and coercive field (μ0Hc) have been determined from the hysteresis cycles (Figure ). For both samples, Ms is close to 120 A m2 g–1 at 300 K and 130 A m2 g–1 at 5 K. These values are below those reported for the bulk materials (230 A m2 g–1 for bulk FeCo and 165 A m2 g–1 for Co).[38] This reduction can be due to both surface oxidation and incorporation of carbon into the Co and FeCo lattice during the encapsulation of the NPs into carbon. Indeed, the formation of cobalt carbide[39] or the incorporation of carbon into FeCo NPs[40,41] is known to lead to species showing a depleted Ms. Partial oxidation was confirmed by the presence of a small exchange bias observed in the hysteresis loops at 5 K after a field cooling in the presence of a μ0H of 3 T, which is characteristic of the coupling between ferromagnetic and antiferromagnetic layers.[42] However, it is presumably not sufficient to explain the strong depletion of the magnetic properties that are likely to result from carbon incorporation.

Figure 4

Hysteresis loops measured by VSM on FeCo@C (red) and Co@C (green) at (a) 5 and (b) 300 K with zoomed region between −0.1 and 0.1 T.

The heating power of FeCo@C and Co@C was determined by calorimetry using our previously described procedure (see section S7 in the Supporting Information).[7,43,44] The specific absorption rate (SAR) was determined in ethanol due to the relatively good dispersibility of the NPs in this solvent (see Figure S12). In both cases, the NPs start to heat after application of a μ0Hrms of 25 mT with an f value of 93 kHz, with a quasi-linear increase in SAR with the field amplitude, reaching 165 W g–1 in the case of FeCo@C and 70 W g–1 for Co@C. These values are much lower than the previous ones reported in literature of 1600 W g–1 for FeCo alloys at 47 mT and 100 kHz.[8,10,45] This is in agreement with the lower Ms and the lower coercivity of these particles compared to reduced FeCo and CoNPs and are likely due to both the incorporation of carbon into the particles and their partial surface oxidation. In addition, the particles lack the necessary mobility that allows the formation of chains, which is responsible for the enhancement of the heating power of FeC and FeCo NPs.[10,46,47] However, the measured SAR values can be explained if we consider a metallic core of the NPs. As a whole, despite the lower SAR, the heating properties of FeCo@C and Co@C are suitable to heat the system and reach the required high temperatures under catalytic conditions (vide infra).

Catalytic Studies

Medium Temperature Catalysis

Sabatier Reaction

In a previous work, the catalyst used for the magnetically induced CO2 hydrogenation into CH4 was prepared by thermal decomposition of bis(1,5-cyclooctadiene)nickel(0), Ni(COD)2, in the presence of Fe2.2C NPs or a combination of Fe2.2C NPs and Co nanorods (Co NRs) supported on SiRAlOx.[11] The best catalytic results (CO2 conversions of 90% with a 100% selectivity to CH4 at a μ0Hrms of 16 mT and an f value of 300 kHz) were obtained by using a mixture of Co NRs and Fe2.2C NPs. Here, the softer Fe2.2C NPs act as preheating agents, which activate the Co NRs that have a higher anisotropy (shape and magnetocrystalline) and Tc (>1000 °C) and require higher values of magnetic field amplitude to be activated in magnetic heating. Upon preheating the Co-NRs, their coercive field decreases, which renders them active as heating agents. Following these results, 10 wt % Ni NPs were deposited on the surface of the carbon-coated heating agent by decomposition of Ni(COD)2 in the presence of FeCo@C (5 wt % in metal; see Experimental Section, section 4). This composition (FeCo@C/Ni) has been used in a continuous flow CO2 hydrogenation using a 1:4 CO2:H2 molar ratio (20 mL min–1 of H2 and 5 mL min–1 CO2). 96% of CO2 conversion with a 98% selectivity into CH4 was obtained at a magnetic field intensity μ0Hrms = 32 mT and a frequency f = 300 kHz (Figure , Table ). Decreasing the field amplitude led to lower conversions (91% at μ0Hrms 28 mT, 98% selectivity and 55% at μ0Hrms = 23 mT with 61% selectivity). The sample exhibited a reproducible cyclability over 4 h and the catalyst was stable for more than 7 h (Figure ). Although higher magnetic-field amplitudes were required to achieve the optimal catalytic conditions than in the case of the Fe2.2C NPs-Ni/SiRAlOx system previously reported, it should be highlighted that no activation step was required despite the low values of SAR, which might be related to the higher Tc of Co and FeCo. The temperature of the system, measured by an ultrathin thermocouple directly located in the catalytic bed,[48] increased with the field amplitude from 230 °C at 23 mT to 315 °C at 32 mT. TEM observation showed that the size of FeCo@C/Ni slightly increased up to ca. 16 nm after 7.5 h time on stream (TOS) (Figure d, e). EDX analysis of FeCo@C/Ni after catalysis shows signals corresponding to discrete FeCo particles, indicating the good stability of the heating agents during catalysis. Ni NPs were homogeneously dispersed thorough the material and did not exhibit extended agglomeration, which may be due to the global low temperature in the system (Figure d, e). Therefore, FeCo@C/Ni has demonstrated to be an active and stable catalyst capable of carrying out magnetically induced methanation without the necessity of an activation step. The most important aspect is that despite the lack of mobility of the encapsulated nanoparticles, they are able to heat efficiently and do not experience a massive sintering at these temperatures, as was observed in the same “naked” MagNPs.[10]

Figure 5

Catalysis of Sabatier reaction using FeCo@C/Ni. Gas flow: 25 mL/min CO2:H2 (1:4). X = conversion, S = selectivity, and Y = yield. (a) Catalytic performances over time, (b, c) TEM observation (b) before and (c) after 7.5 h TOS. (d) EDX mapping of Fe (red), Co (yellow), and Ni (blue) after preparation and (e) after 7.5 h on stream. Temperature was recorded by an ultrathin thermocouple directly located in the catalytic bed and confirmed by IR thermometry.[48]

Table 1

Magnetic Induced Catalysis Performances of Sabatier, Propane Dry Reforming (PDR), and Propane Dehydrogenation (PDH) Using FeCo@C and Co@C as Heating Agents

reaction	heating agent (HA)	HA loading (wt %)	catalyst nature	catalyst loading (%wt)	field amplitude (mT)a	X, conversion (%)	S, selectivity (%)	Tbulk(°C)	conversion at thermodynamic eq (%)b
						X(CO2)	S(CH4)
Sabatier	FeCo@C	5	Ni	10	23	48.6	55.2	235	99
					28	91.1	97.5	280	96
					32	96	97.5	315	94
						X(C3H8)	S(CO)
PDR	FeCo@C Co@C	5	Ni	10	32	9.1	100	425	15
					44	21.3	100	530	37
					53	43.9	97.9	600	60
					60	53.6	97.2	620	72
						X(C3H8)	S(C3H6)
PDH	FeCo@C Co@C	5	Pt	0,35
			Sn	0,5	44	2.7	88.1	530	27
					65	10.4	79.4	620	52

Applied magnetic field of 300 kHz.

Conversions at thermodynamic equilibrium were estimated employing the Gibbs free energy minimization method.[48,56−58]

Temperatures were measured by an ultrathin thermocouple directly located in the catalytic bed and validated with an IR camera. Notice that temperatures given are an estimation and the real temperatures at the surface of the catalyst are probably higher.

High-Temperature Catalysis

Propane Dehydrogenation (PDH)

Because of the high Tc values of both Co@C and FeCo@C systems, we then decided to evaluate them as heating agents for endothermic reactions operating at high temperatures.[10] With the intention of reaching the highest reaction temperature, a mixture of 2:1 FeCo@C:Co@C was used as heating agent. Following our previous observations, FeCo would act as the soft material displaying higher SAR values, whose role would be to activate hard Co, which possesses a Tc above 1000 °C. Thus, we experimentally observed that a mixture of 2:1 FeCo@C:Co@C was the optimum composition for the heating agent to activate catalysis at high temperatures. Concerning the catalyst, we chose Pt/Sn, prepared by a recently published procedure,[10] because recent works have demonstrated that Sn can not only enhance the activity of Pt NPs in propane dehydrogenation but also improve the selectivity and stability of the catalyst.[24,49−51] Therefore, Pt NPs (0.35 wt %) promoted with Sn (0.5 wt %)[52] were supported on a mixture of 2:1 FeCo@C:Co@C (see Experimental Section, section 4) and used as catalyst in propane dehydrogenation (PDH) (>700 °C). By using an incoming gas of pure C3H8 (flow of 20 mL min–1), the conversion reached values up to 11 with 81% selectivity to propene at μ0Hrms = 65 mT (300 kHz) and a monitored temperature of 620 °C (Figure a; Table ). The remaining 19% of selectivity is due to the formation of methane and propane in a 1:1 ratio. When only FeCo@C was used as heating agent in the same conditions, significantly lower performances were obtained (see Figure S13). TEM observations demonstrated that FeCo@C and Co@C were stable in time as their sizes slightly increased up to ca. 15 nm after 2 h of TOS (Figure b ,c). EDX analysis of the sample before catalysis could not clearly identify Pt and Sn because of the low signal as a result of the low catalyst loadings. However, EDX analysis after catalysis indicates that a slight agglomeration of Pt and Sn, mostly located on the heating agents, has taken place (Figure d). Similar aggregations of Pt over the heating agents have also been observed in other high-temperature magnetically induced catalytic reactions.[10,53] Under the same conditions, uncoated FeCo NPs, while being capable of activating PDH catalysis with slightly higher activity, have limited stability because of massive sintering of metallic NPs and carbon deposition.[10] Therefore, the main advantage of carbon encapsulation is the stabilization of the catalytic system, avoiding sintering of the heating agents.

Figure 6

PDH reaction using 2:1 FeCo@C:Co@C/PtSn (Pt 0.35 wt %, Sn 0.5 wt %) as catalyst. Gas stream: 20 mL/min C3H8. X = conversion, S = selectivity, and Y = yield. (a) Catalytic performances over time, (b, c) TEM observation (b) before and (c) after 4.5 h TOS. (d) EDX mapping of Fe (red), Co (green), Pt (yellow), and Sn (blue) after 4.5 h on stream. Temperature was recorded by an ultrathin thermocouple directly located in the catalytic bed and confirmed by IR thermometry.[48]

Propane Dry Reforming (PDR)

Propane dry reforming (PDR) was performed by using Ni (10 wt %) impregnated on 2:1 FeCo@C:Co@C as catalyst (FeCo@C:Co@C/Ni; see Experimental Section, section 4). At μ0Hrms = 32 mT (300 kHz), 9% of the incoming propane was converted into CO and H2 with 100% selectivity using a gas flow of 10 mL min–1 (3:1 molar ratio CO2:C3H8). However, upon increasing the field amplitude to 60 mT, propane conversion was increased to 54% with a 97% selectivity to CO, very close to the conversion value at the thermodynamic equilibrium (Figure , Table ).[54] The conversion values observed can be explained by the temperature increase from 425 to 620 °C, upon increasing the field amplitude. The selectivity toward CO is particularly high. We propose that this behavior is related to the encapsulation of FeCo and Co NPs and successful prevention of parallel reactions such as Fischer–Tropsch, which is a typical undesired side-reaction.[55] The temperature of the system remained stable during the 2.5 h of the catalytic test. The size of the MagNPs showed again a slight increase after 2 h TOS (ca. 14 nm). EDX analysis after catalysis revealed that Ni catalyst NPs slightly migrated toward the heating agents (Figure b–e) but kept a good size distribution.

Figure 7

PDR reaction using 2:1 FeCo@C:Co@C/Ni (10 wt %) as catalyst. Gas stream: 10 mL/min CO2:C3H8 (3:1). X = conversion, S = selectivity, and Y = yield. (a) Catalytic performances over time, (b, c) TEM observation (b) before and (c) after 2 h TOS. (d) EDX mapping of Fe (red), Co (blue), and Ni (light blue) after preparation and (e) after 2 h on stream. Temperature was recorded by an ultrathin thermocouple directly located in the catalytic bed and confirmed by IR thermometry.[48]

Stability Studies

To investigate in more detail the stability of our heating agents and of the catalyst under the harsh conditions employed, we ran PDR over 6 h. Specifically, we tested 2:1 FeCo@C:Co@C/Ni (10 wt % of Ni) and a gas mixture of 3:1 CO2:C3H8 with a total flow rate of 10 mL min–1 during 6 h. We previously showed that after 2 h on stream, 52% yield could be reached with nearly 100% selectivity for CO, whereas supported Ni NPs were slightly agglomerated. Two more cycles with variable magnetic-field amplitudes (32, 44, 53, and 60 mT) were performed and enabled to show a quasi-perfect cyclability in terms of temperature reached by magnetic induction of the heating agents (up to 620 °C, at μHrms 60 mT, 300 kHz) and CO yields (Figure ). This result demonstrates the very efficient encapsulation of the FeCo@C and Co@C even at high temperatures. The catalytic performances of the sample remained stable and reproducible for all the field amplitudes tested. However, EDX mapping after 6 h on stream shows a more important agglomeration of the Ni NPs, which, however, did not induce any significant decrease of the catalytic performances (see Figure S14a–c). The sample was then exposed to air and another reaction cycle was performed. This led to a slight decrease in the temperature reached in the system to 605 °C at a μHrms = 60 mT, but no significant effects on the catalytic performances were observed. In addition, the morphology of the sample was not significantly altered upon air exposure followed by two additional hours on stream (see Figure S14d). These results demonstrate the encapsulation in carbon, which remediates the catalyst instability issues associated with the sintering of MagNPs under high-temperature catalysis.

Figure 8

Magnetically induced propane dry reforming (PDR) performances and temperatures using 2:1 FeCo@C:Co@C/Ni (10 wt %) as catalyst at variable magnetic-field amplitudes. Gas stream: 10 mL/min CO2:C3H8 (3:1). X = conversion, S = selectivity, and Y = yield. Temperature was recorded by an ultrathin thermocouple directly located in the catalytic bed and confirmed by IR thermometry.[48]

Conclusions

Magnetically induced catalysis can indeed address various reactions and temperature ranges. Though the weakness of this technique is the agglomeration of the heating magnetic nanoparticles upon magnetic excitation, we have shown that this problem can be circumvented by confining the heating NPs in carbon. Thus, we have successfully encapsulated magnetic FeCo and Co NPs in carbon (Co@C and FeCo@C), which leads to ultrastable heating agents suitable for high-temperature magnetically induced catalysis. Combining a series of characterization techniques such as HRTEM, EDX, XRD, XPS, and VSM, we can conclude that Co@C and FeCo@C NPs have a metallic core and an oxidized surface encapsulated by a carbon layer, but that under reductive conditions similar to those used in the catalytic experiments, the oxidized surface is reduced.

Catalytic studies have shown that medium (Sabatier 300–400 °C) and high-temperature (PDR and PDH > 600 °C) catalysis can be activated by FeCo@C or a mixture of FeCo@C:Co@C (2:1) with Ni or Pt–Sn supported, with consistent conversions and selectivities compared to traditional heating sources. Under the conditions employed, we observed a good stability of the heating agents, in terms of temperature, and of the catalyst as a whole in terms of catalytic performances. However, a systematic agglomeration of the catalytic NPs (PtSn or Ni NPs) was observed at high temperature as in classical heterogeneous catalysis. Under the same conditions, the use of naked FeCo NPs in PDR or PDH did not show stability for more than one hour.[10] Thus, the encapsulation of the heating agent constitutes a major improvement toward the development of stable catalysts for high-temperature magnetically induced catalysis.

Finally, in the case of PDR, the catalyst showed remarkable cyclabilities of temperature and catalytic performance, up to 4 times over at least 6 h. Furthermore, excellent cyclability was achieved even after exposure to air of the catalysts and/or partial sintering of the Ni NPs.

Experimental Section

General Considerations and Starting Materials

Most of operations were carried out using standard Schlenk tubes or Fischer–Porter bottle techniques or in a glovebox under an argon atmosphere. Mesitylene, toluene, and tetrahydrofurane (THF) were obtained from VWR Prolabo then purified on alumina desiccant and degassed by bubbling Ar through the solution for 20 min. The commercial products, hexadecylamine (HDA, 99%), Ni(COD)2, (COD = cyclo-octadiene), and Bu3SnH (1 M in cyclohexane), were obtained from Sigma-Aldrich. Pt2(dba)3, [Fe{N(SiMe3)2}2]2 and [Co{N(SiMe3)2}2(THF)] precursors were obtained from Nanomeps. All the commercial compounds were used as received. All gases were supplied by Air–Liquide with the following purity: CO2 Alphagaz N48, H2 N55, Ar N56, CH4 N55.

Transmission Electron Microscopy (TEM) and High-Resolution TEM (HRTEM)

FeCo NPs, Co NPs, FeCo/C, Co/C, FeCo@C, and Co@C NPs were observed by TEM and HRTEM after deposition on a copper grid of a drop of a suspension of the material in THF. TEM analyses were performed at the “Servicio de Microscopía Electrónica” of Universitat Politècnica de València (UPV) by using a JEOL JEM 1010 CX-T electron microscope operating at 100 kV with a point resolution of 4.5 Å. The approximation of the particles mean size was made through a manual analysis of enlarged micrographs by measuring a number of particles on a given grid. HRTEM measurements were performed using a JEOL 2100F microscope operating at 200 kV, both in transmission (TEM) and scanning-transmission modes (STEM). Energy-dispersive X-ray spectrometry (EDX) and STEM images were obtained using a high-angle annular dark-field detector (HAADF), which allows Z-contrast imaging. FFT treatments have been carried out with Digital Micrograph version 3.7.4.

Inductively Coupled Plasma (ICP)

ICP analyses of FeCo@C and Co@C were performed by the ICP technique service of the Instituto de Tecnología Química (ITQ), using a Varian 715-ES ICP-Optical Emission Spectrometer. The samples for ICP were prepared following a digestion method recently reported.[59] In particular, 30 mg of catalyst sample was suspended in 21 mL of HCl-HNO3 (6:1). The solution was then sonicated for 90 min and the samples were digested at 180 °C for 15 h. Then, the solution was cooled down to room temperature (r.t.), diluted with 100 mL of water, and analyzed by ICP.

Raman

Raman spectra were recorded using an excitation of 785 nm in a Renishaw In via Raman spectrometer equipped with a Lyca microscopy. The samples (powder) were deposed on an Al support, and measured in the region between 0 and 3000 cm–1 with a resolution of <4 cm–1.

X-ray Photoelectron Spectroscopy (XPS)

XPS analyses were performed using a SPECS device equipped with a Phoibos 150–9MCD detector using Mg–Kα radiation (hν = 1235.6 eV) and Al–Kα radiation (hν = 1483.6 eV) from a dual source. The pressure during the measurements was kept under 1 × 10–9 Torr. The quantification and titration of the spectra was done with the help of the software CASA, referencing them in base of C 1s = 284.5 eV.

X-ray Powder Diffraction (XRD)

Powder samples were analyzed using a Cubix-Pro PANalytical diffractometer equipped with a detector PANalytical X′Celerator, employing a X-ray monochromatic radiation of CuKα (λ1 = 1.5406 Å, λ2 = 1.5444 Å, I2/I = 0.5).

Temperature-Programmed Reduction (TPR)

Micromeritics Auto-Chem 2910 catalyst characterization system with a thermal conductivity detector (TCD) was used for TPR analysis. Before analysis of the samples (50 mg), they were pretreated at room temperature in flowing He (10 mL/min) for 20 min. The samples were then heated from 25 to 600 °C at a rate of 5 °C/min in a flow (50 mL/min) of H2 in Ar (10% vol.).

Gas Chromatography Coupled to Mass Spectrometry (GC-MS)

GC-MS analyses were carried out in a PerkinElmer 580 gas chromatograph equipped with a TCD detector and coupled to a Clarus SQ8T mass spectrometer. Reactant conversions and product yields and selectivities were calculated by performing a C balance to the chromatograms. The areas of each peak were corrected with their response factor obtained after calibration of the TCD detector.

Vibrating Sample Magnetometer (VSM)

Magnetic measurements were performed on a vibrating sample magnetometer (VSM, Quantum Device PPMS Evercool II). VSM studies were carried out on compact powder samples that were prepared and sealed under an argon atmosphere.

Synthesis of FeCo and Co NPs

FeCo NPs

FeCo NPs were prepared according to the literature.[60] Fe and Co precursors, [Fe{N(SiMe3)2}2]2 (180,7 mg; 0.2 mmol) and [Co{N(SiMe3)2}2(THF) (150.5 mg; 0.4 mmol)], were reduced under 3 bar of H2 upon heating at 150 °C for 24 h in mesitylene (20 mL) and in the presence of long-chain surfactants, hexadecylamine (HDA; 386 mg; 1.6 mmol), and hexadecylamine hydrochloride (HDA·HCl; 333.5; 1.2 mmol). The precursor concentrations were kept at 20 mmol L–1. The HDA and HDA·HCl concentrations were 80 and 60 mmol L–1, respectively. After reaction, a black solution with the magnetic FeCo NPs stabilized by HDA/HDA·HCl was obtained. A drop of this solution was deposed on a cooper grid, and the size of the NPs was measured by TEM for at least 100 nanoparticles, which afforded a mean value of 10.7 ± 2.4 nm (Figure S1). ICP: Fe, 2.6 wt %; Co, 2.3 wt %.

Co NPs

Cobalt NPs were prepared as described for FeCo NPs but starting from the Co precursor, [Co(N(Si(CH3)3)2)2,THF] (361.4 mg; 0.8 mmol). The size of the NPs was measured by TEM for at least 100 nanoparticles, which afforded a mean value of 11.7 ± 2.6 nm (Figure S2). ICP: Co, 4.8 wt %.

Synthesis of FeCo@C and Co@C NPs

FeCo@C

Nine-hundred twenty grams of activated carbon dispersed in 10 mL of mesitylene were added to a Fischer–Porter bottle charged with a suspension of FeCo NPs (78 mg) in 20 mL of mesitylene. After 24 h of vigorous stirring, FeCo NPs were adsorbed on the activated carbon forming FeCo/C (see Figure S3). Then, FeCo/C was washed with hexane three times, and dried at 60 °C overnight. The resulting black powder was then subjected to a pyrolysis process (2 h at 600 °C under N2 pressure, with a heating ramp: 10 °C min–1), producing the FeCo NPs encapsulated in carbon (FeCo@C). The size of the NPs was measured by TEM for at least 100 nanoparticles, which afforded a mean value of 10.8 ± 3.2 nm (Figure a–c).

Co@C

Co@C was prepared as described for FeCo@C but using a suspension of Co NPs (76 mg) in 20 mL. The size of the NPs was measured by TEM for at least 100 nanoparticles, which afforded a mean value 10.7 ± 2.4 nm (Figure d–f).

Synthesis of FeCo@C and Co@C Decorated with Ni and PtSn NPs

FeCo@C/Ni and FeCo@C–Co@C/Ni catalyst (10% Ni loading) were prepared by the decomposition of Ni(COD)2 in the presence of FeCo@C or FeCo@C:Co@C (2:1) in mesitylene. In a typical preparation, 156 mg of Ni(COD)2 was dissolved in 20 mL of mesitylene and 300 mg of FeCo@C or FeCo@C:Co@C (2:1) was added to the solution. The mixture was then heated at 150 °C under an Ar atmosphere and vigorous stirring during 1 h followed by three washing steps with toluene. Drying under vacuum enables us to get Ni NPs with a mean size of ca. 2–4 nm immobilized on FeCo@C or FeCo@C:Co@C (2:1). The total amount of collected powder was of ca. 350 mg.

FeCo@C/Pt–Sn and FeCo@C:Co@C/Pt–Sn. Were prepared by adapting a published procedure.[10] For Pt–Sn catalysts (0.5% loading), the catalyst was prepared by mixing preformed Pt NPs with FeCo@C or a mixture of FeCo@C:Co@C (2:1). In a typical preparation, 6.5 mg of Pt2(dba)3 in 5 mL of THF are stirred in the presence of 1 bar CO for 20 min. The dark brown mixture formed is evaporated to a small volume (1–2 mL). Complete drying has to be avoided in order to facilitate redispersing nanoparticles. Three washing steps by pentane are then performed before diluting the obtained NPs in 10 mL of THF. Then, Bu3SnH is added to the suspension of the Pt NPs and the solution is left under magnetic stirring for 5–10 min. Finally, 450 mg of FeCo@C or 2:1 FeCo@C:Co@C are added and the solution is stirred for 10 min. The solvent is eliminated while sonicating to obtain a fine dark brown powder that is collected and stored.

Catalytic Reactions

Catalytic experiments were performed in a quartz tube continuous-flow reactor of 1 cm diameter, which was placed at the center of an inductor delivering an AC magnetic field oscillating at a frequency of 300 kHz with a root-mean-square (rms) amplitude adjustable between 0 and 65 mT. The induction coil is 3 cm wide and 2 cm high. We systematically used 400 mg of catalytic bed for Sabatier reaction and between 800 and 1200 mg of catalytic bed for PDH and PDR (800 mg of FeCo@C/Ni or FeCo@C/Pt–Sn + 400 mg of Co@C/Ni or Co@C/Pt–Sn). Mass flows of CH2:CO2 1:3 and 1:1 with 10 mL min–1 CH2 were supplied at the inlet for PDR and for MDR, respectively (weight hourly space velocity, WHSV, 7.5 L h–1 g(Ni)–1). Twenty milliliters per minute of pure propane was supplied for PDH (WHSV 300 L h–1 g(Pt)–1). For the Sabatier reaction, 25 mL min–1 of a 4:1 H2:CO2 molar mixture were flowed (WHSV 37.5 L h–1 g(Ni)–1). The outgoing gases are injected in a gas chromatography–mass spectrometry (PerkinElmer 580 gas chromatograph-thermal conductivity detector coupled to a Clarus SQ8T mass spectrometer). Calibration of the gas injection and the GC analysis method have been performed with pure gases. The response factor of the analytes i RF is determined by injecting known quantities of each analyte i into the column. The area of the peak of the species i on the chromatogram (A) enables us to determine the conversion of i (X) and selectivity toward j (S) on the basis of the calculations below (case of dehydrogenation of propane). For dry reforming of alkanes, the conversion is calculated on the alkane basis, as CO2 is expected to be supplied in excess.

Digestion Experiments

Following a similar procedure as Bao and co-workers,[33] who concluded that CoNi NPs were completely encapsulated by graphene because they were not soluble in a strong acid, we demonstrated that the carbon layer of FeCo@C and Co@C is discontinuous. Specifically, Co and FeCo NPs encapsulated in carbon (10 mg of FeCo@C/Co@C) were practically dissolved in a H2SO4 solution (2M, 5 mL) at 80 °C after 2 h. Analyzing the resulting solutions by ICP-AES, we observed a metal content of 3.55 wt % Co for Co@C and 2.45 wt % Co and 2.71 wt % Fe for FeCo@C.