Chemistry

MagneticallyRecyclableCatalyticCarbonNanoreactors.pdf

Home >Education homework help >Chemistry

FULL PAPER www.afm-journal.de

Magnetically Recyclable Catalytic Carbon Nanoreactors

Mehtap Aygün, Thomas W. Chamberlain, Maria del Carmen Gimenez-Lopez,* and Andrei N. Khlobystov*

Multifunctional nanoreactors are assembled using hollow graphitized carbon nanofibers (GNFs) combined with nanocatalysts (Pd or Pt) and magnetic nanoparticles. The latter are introduced in the form of carbon-coated cobalt nanomagnets (Co@Cn) adsorbed on GNF, or formed directly on GNF from ferrocene yielding carbon-coated iron nanomagnets (Fe@Cn). High-resolution transmission electron microscopy demonstrates that Co@Cn and Fe@Cn are attached effectively to the GNFs, and the loading of nanomagnets required for separation of the nanoreactors from the solution with an external magnetic field is determined using UV–vis spectroscopy. Magnetically functionalized GNFs combined with palladium or platinum nanoparticles result in catalytically active magnetically separable nanoreactors. Applied to the reduction of nitrobenzene the multifunctional nanoreactors demonstrate high activity and excellent durability, while their magnetic recovery enables significant improvement in the reuse of the nanocatalyst over five reaction cycles (catalyst loss < 0.5 wt%) as compared to the catalyst recovery by filtration (catalyst loss <10 wt%).

DOI: 10.1002/adfm.201802869

Dr. T. W. Chamberlain Institute of Process Research and Development School of Chemistry University of Leeds Leeds LS2 9JT, UK Prof. A. N. Khlobystov Nanoscale & Microscale Research Centre University of Nottingham University Park Nottingham NG7 2RD, UK

Specifically, metal nanoparticles (NPs) supported in graphitized carbon nanofibers (GNFs) catalysts have recently been used for a variety of different reactions, including platinum nanoparticles (PtNPs) in CC cross coupling reactions,[21] IrNPs in hydrosilylations,[13,18,22] and CuNPs in click chemistry.[23] In a recent study, we dem- onstrated that RuNPs confined in GNFs result in dramatic changes to reactions with the highest observed activity and selectivity in single and competitive hydro- genations of norbornene and benzonorbor- nadiene compared to unconfined RuNPs supported on single-walled carbon nano- tubes (SWNTs) and commercial carbon black.[24] PtNPs confined within GNF were also investigated in the oxygen reduction reaction (ORR) by Gimenez-Lopez et al. and outstanding electrochemical stability was observed over 50 000 cycles of ORR, with the PtNPs stabilized by the step edges

significantly more strongly than commercial PtNPs on carbon black.[25] In addition to NP based catalysts, Lebedeva et al. syn- thesized fullerene containing and fullerene free Pd(II)Salen metal complexes and encapsulated both species on the step edges of the internal GNF surface to form catalysts which dis- played significantly higher activity and selectivity in several Heck reactions compared to the reactions in solution.[19]

Therefore, carbon nanotubes are of great interest for use as nanoreactors in a variety of different catalytic chemical reactions as they not only template the formation of catalytically active metallic nanoparticles but also influence the subsequent pathway of reactions.[11,14–27] However, despite the fact that carbon

Magnetic Nanoparticles

1. Introduction

Carbon nanotubes are mechanically robust, thermally and chemically stable cylinders of sp2-carbon that can be used to immobilize both molecules and nanoparticles which efficiently adsorb onto the nanotube walls and/or are encapsulated within the internal cavity of the nanotube via noncovalent interactions such as van der Waals forces.[1–8] Once the catalyst is immobi- lized in the hollow structure, catalytic chemical reactions which occur within the accessible nanoscale space of the nanore- actor interior can benefit from enhanced rates of reactions and selectivity.[9–27]

The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adfm.201802869.

Dr. M. Aygün, Prof. A. N. Khlobystov School of Chemistry University of Nottingham University Park Nottingham NG7 2RD, UK E-mail: [email protected] Dr. M. Aygün, Dr. M. C. Gimenez-Lopez Centro Singular de Investigación en Química Biolóxica e Materiais Moleculares (CIQUS) Universidade de Santiago de Compostela 15782 Santiago de Compostela, Spain E-mail: [email protected]

Adv. Funct. Mater. 2018, 1802869

http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadfm.201802869&domain=pdf&date_stamp=2018-07-11

www.afm-journal.dewww.advancedsciencenews.com

nanotubes are excellent support materials for heterogeneous catalyst systems, the inherent properties of carbon nanotubes, including their low density and hydrophobisity, make their sepa- ration from the reaction solution using conventional separation techniques, such as filtration and centrifugation challenging, meaning that currently expensive equipment and secondary processes are required.[28–30] In order to minimize the costs and technical challenges in conventional catalyst separation and facilitate the recycling of precious metal catalysts, intense research efforts have been focused on the development of mag- netic supported metal nanoparticle catalysts which could be con- trolled by an applied magnetic field.[31–36] This approach enables the selective separation of magnetic material supporting cata- lysts from the reaction mixture containing the products (non- magnetic species) by application of a magnetic field.

Ferromagnetic metal nanoparticles including Fe and Co combine high catalytic activity with a nonzero magnetic moment at room temperature due to unpaired electrons, which can be useful for many catalytic reactions.[34,35] However, as these magnetic metal nanoparticles are not stable in air and easily react with acid media, resulting in a change or loss of their magnetization, their use in preparative catalysis has been limited.[29] Therefore, an alternative route is to combine a fully protected magnetic nanoparticle component, coated with an appropriate inert material such as silica, polymers or carbon, with another, nonmagnetic but catalytically active metal nanoparticle component to perform the catalysis on a suitable support material.[37–40]

Preparation of carbon-coated magnetic nanoparticles has recently received increasing attention as carbon nanomaterials have been proven to be both chemically and thermally more stable and robust than silica or polymer coatings.[41] These materials consist of magnetic nanoparticles, providing a magnetic core, and a graphene-like outer shell which ensures that the magnetic material is completely coated and protected against oxidation and erosion by strong acids or bases, allowing their use under harsh reaction conditions. A number of methods have been applied to prepare carbon-coated magnetic nanopar- ticles, especially in the metallic phase which have higher mag- netic moments compared to metal oxides, using chemical vapor deposition and the sequential spraying and controlled pyrolysis of carbon sources at elevated temperatures.[37,42,43] Recently, Grass et al. developed a method to synthesize carbon-coated Co nanomagnets (Co@Cn) using reducing flame spray pyrolysis under an inert atmosphere.[37] This method allowed the pro- duction of nearly spherical magnetic particles with an onion- like sp2-carbon coating with a thickness of 2–3 nm and a mean particle diameter of about 4–100 nm. It was subsequently dem- onstrated that it is possible to modify the carbon coating using both covalent and noncovalent functionalization and this has been exploited to attach catalytic nanoparticles to the surface of the nanomagnets, creating magnetically recoverable heteroge- neous catalysts for a variety of catalytic applications.[44–49] On the other hand, Wittmann et al. demonstrated the preparation of a palladium complex noncovalently attached to Co@Cn based on strong π–π stacking interactions between pyrene units and the outermost graphene layer enabling efficient catalyst recovery.[45] Furthermore, the graphene-like outermost shell of carbon coated magnetic nanoparticles is very similar to the surface of carbon

nanotubes which enables combination of the two materials via adsorption of the nanoparticles onto the exterior sidewalls or into the internal channel of the nanotubes driven by van der Waals forces.[46] The combination of catalytic carbon nanoreac- tors with magnetic Co@Cn nanoparticles using noncovalent interactions (van der Waals forces) could potentially allow the separation of the carbon nanoreactors from reaction mixtures in a fast, easy and efficient way by simply applying a magnetic field. In this study, this idea of creating magnetically separable carbon nanoreactors which contain active metal nanoparticle catalysts confined in their channels is explored, and both the catalytic activity and the magnetically induced separation of the resultant hybrid materials are investigated (Scheme 1).

GNFs were chosen as the carbon nanoreactor support as they, unlike carbon nanotubes, have negligible residual metal content making analysis of the metal nanoparticle-nanocarbon hybrids easier to quantify.[50] Furthermore, GNFs have differently structured internal and external surfaces and wide, continuous internal channels, with an average internal diameter of ≈50 nm. Finally, unlike carbon nanotubes, the internal surface has a suc- cession of step edges which can act as anchoring points for guest species making GNF a highly effective nanoreactor for immobi- lization of catalytic nanoparticles and to perform catalytic reac- tions at the nanoscale.[7,18,19,24,25] With this aim, we developed two different procedures for forming magnetically recyclable GNF based carbon nanoreactors: (1) in situ formation of Fe@ Cn inside the GNF channels and (2) attachment of commercially available Co@Cn to GNF through noncovalent interactions. In addition, we explored two different methodologies to combine the formation step of the catalytically active palladium nanoparti- cles (PdNPs) or PtNPs with the magnetic functionalization step. As a result, catalytically active and magnetically separable hybrid materials were successfully designed and synthesized, and their activities in the reaction of nitrobenzene reduction were tested and compared.

2. Results and Discussions

2.1. Designing and Preparing Magnetically Recyclable GNF Based Carbon Nanoreactors

An experimental method was developed to make carbon-coated Fe nanoparticles directly in the GNF, (Fe@Cn)/GNF). In this method (Figure 1a), ferrocene was inserted from the vapor phase into GNF at 350 °C in vacuum and subsequently heated to 500 °C to decompose the ferrocene into FeNPs coated in gra- phitic shells (Fe@Cn) which deposit on the walls of the GNF (N.B. source of carbon is cyclopentadienyl ligand of ferrocene) (Experimental Section).

High-resolution transmission electron microscopy (HRTEM) imaging confirmed the presence of FeNPs coated by a graphitic shell in which the carbon interplanar distance was measured to be 0.34 nm, which is comparable to that of the interlayer spacing in graphite (Figure 1b–d). The Fe@Cn nanoparticles have an average diameter of 23.9 ± 14.9 nm (Figure 1e) with a carbon shell thickness of 5.84 ± 2.49 nm (corresponding to 17 ± 7 graphene-like carbon layers), and are adsorbed princi- pally to the step edges of the sidewalls within the cavity of the

Adv. Funct. Mater. 2018, 1802869

www.afm-journal.dewww.advancedsciencenews.com

GNF (>60%), with the remaining Fe@Cn absorbed on the outer surface of the GNF. This could be a result of the step edges pro- viding better adsorption sites for individual ferrocene molecules during the decomposition process. Thus, as the iron-containing material is already inside the channel, upon rapid thermal decomposition the resultant carbon-coated Fe nanomagnets are formed primarily inside the GNF channel. In addition, the concave surface of GNF interior is likely to assist nucleation of Fe@Cn. Powder X-ray diffraction (XRD) revealed the presence of a metallic Fe phase which is in good agreement with the dif- fraction pattern of α-FeNPs reported previously (Figure 1f).[52] The Fe loading (wt%) in the (Fe@Cn)/GNF was quantified using thermogravimetric analysis (TGA) by heating in air up to 1000 °C at a rate of 10 °C min−1 (Figure 1g). TGA showed that the presence of Fe in (Fe@Cn)/GNF led to a significant decrease in the oxidation temperature of the GNF from ≈700 to ≈500 °C. At ≈850 °C a small weight gain, presumably due to oxidation of the Fe, was observed. Therefore, the residual Fe content (wt%) was recorded as the average mass between 820 and 850 °C, and was observed to be 8.5 ± 0.6 wt% (Figure 1g).

To test the implications of the magnetic confinement on the development of magnetically recyclable GNF based carbon nanoreactors, commercially available magnetic Co@Cn were also attached to individual GNFs using noncovalent interac- tions. As carbon nanostructures are known to be attracted to each other by strong van der Waal forces (0.5 eV µm−1), a good solvent was required to obtain well-dispersed and separated GNFs[51] to ensure good mixing with Co@Cn. Therefore, GNFs were initially dispersed using ultrasonic treatment in hexane.

Co@Cn were also dispersed using the same method, and then added to the hexane/GNF dispersion very slowly while being continuously treated with ultrasonic waves to create a mate- rial in which the Co@Cn nanoparticles are adsorbed on the GNF, (Co@Cn)/GNF (Figure 2 and the Experimental Sec- tion). The minimum loading of Co@Cn required for com- plete separation of the composite material from solution was evaluated by changing the amount of magnetic Co@Cn in the (Co@Cn)/GNF material and exposing each sample, suspended in hexane, to an external magnet and evaluating the resulting solution by eye. (Co@Cn)/GNF was prepared in 1, 5, and 10 wt% of Co@Cn on GNF, and the resultant (Co@Cn)/GNF materials were separated from the solvent by applying an external magnetic field using a commonly available neodymium magnet with a magnetic strength of ≈0.1 Tesla (T). Complete separation for (Co@Cn)/GNF containing 10% of Co@Cn was achieved (Experimental Section). Lower Co@Cn loadings, however, resulted in incomplete separation compromising the recovery of all the catalyst material that is strictly required when pursuing recyclable catalytic materials (Figure S1, Sup- porting Information). The (Co@Cn)/GNF sample with 10 wt% loading was then characterized by HRTEM, TGA, and powder XRD. HRTEM confirmed the presence of very well distributed Co@Cn on both the outer and interior surfaces of the GNF with an average diameter of Co@Cn 29.7 ± 22.8 nm (Figure 2), with a carbon shell thickness of 2.87 ± 1.19 nm (corresponding to 7 ± 4 graphene-like carbon layers).

In contrast, to the Fe@Cn/GNF system, the majority of the Co@Cn nanomagnets are adhered to the GNF outer surface

Scheme 1. A schematic illustration of the recovery of catalytic GNF nanoreactors by magnetic separation from a liquid solution after a chemical reaction.

Adv. Funct. Mater. 2018, 1802869

www.afm-journal.dewww.advancedsciencenews.com

(>90%), presumably due to the more readily accessible and aromatic character of the outer surface of GNF resulting in a higher affinity for Co@Cn than the corrugated, step-edge con- taining internal channels. The smooth graphitic shell of the

Co@Cn is likely to be engaged in π–π stacking interactions with the smooth exterior of GNF and thus results in stronger van der Waals forces between the Co@Cn and the GNF outer surface. There will also undoubtedly be an energetic barrier

Adv. Funct. Mater. 2018, 1802869

Figure 1. a) Schematic illustration of the experimental procedure for synthesis of carbon coated FeNPs attached to GNF ((Fe@Cnn)/GNF), b–d) HRTEM images of (Fe@Cnn)/GNF where the graphene layers can be seen in the close-up of the particle, e) particle size distribution of Fe@Cn (the size of Fe and graphitic shell were measured together using more than 80 particles). f ) Powder XRD patterns and g) TGA measurements of (Fe@Cnn)/GNF in air at a heating rate of 10 °C min−1.

www.afm-journal.dewww.advancedsciencenews.com

Figure 2. a) Schematic illustration of the experimental procedure for synthesis of carbon coated CoNPs attached to GNF ((Co@Cnn)/GNF), b–c) HRTEM images of (Co@Cnn)/GNF where the graphene layers can be seen in the close-up of the particle, d) particle size distribution of Co@Cnn (the size of Co and graphitic shell were measured together using more than 80 particles). f ) Powder XRD patterns and g) TGA measurements of (Co@Cn)/GNF in air at a heating rate of 10 °C min−1.

www.afm-journal.dewww.advancedsciencenews.com

to diffusion of the larger Co@Cn down the internal channel of the GNF as a result of their similar sizes. The composition of the (Co@Cn)/GNF was determined by powder XRD showing the presence of a metallic cobalt phase which is in good agreement with the reference fcc-Co metal powder XRD pattern (Figure 2f).[52] Similar to (Fe@Cn)/GNF, TGA was used to identify the degree of magnetic metal loading in (Co@Cn)/ GNF after heating in air up to 1000 °C at a rate of 10 °C min−1 (Figure 2g). TGA studies of (Co@Cn)/GNF showed that the presence of Co led to a significant decrease in the oxidation temperature of the GNF from ≈700 to 500 °C. The weight gain observed between ≈800 and 1000 °C due to oxidation of the residual Co after the carbon shells have been removed was negligible, however, to ensure this was accounted for the residual weight was recorded as an average value between 800 and 850 °C and revealed the (Co@Cn)/GNF material to be 8.5 ± 0.5 wt%, which is comparable to the metal loading observed within (Fe@Cn)/GNF.

2.1.1. Evaluating the Efficiency of the Magnetic Separation of (Fe@Cn)/GNF and (Co@Cn)/GNF

Separation of the resultant functionalized GNF composite suspensions from the solvent was achieved by placing a magnet (0.1 T) on the external wall of the sample tube for a short period of time (90 s) (Figure 3a). The effect of varying the extent of loading of both Fe@Cn and Co@Cn on the separation of the GNFs was evaluated using an ultraviolet– visible (UV–vis) spectroscopy in which (Fe@Cn)/GNF and (Co@Cn)/GNF samples were suspended separately in hexane by stirring, and then exposed to an external magnet for dif- ferent lengths of time (Experimental Section). UV–vis spec- troscopy was used to evaluate the concentration of (Fe@Cn)/ GNF and (Co@Cn)/GNF that remained suspended in the hexane after application of the magnet. We used the inten- sity of optical density measured by UV–vis spectroscopy to determine the concentration of GNF-magnetic material com- posite remaining as a suspension after magnetic separation at the visible wavelength range, 350–700 nm, assuming that the optical density is directly proportional to the concentra- tion of GNF in solution, in accordance with the Beer-Lambert law.[53,54] UV–vis spectroscopy measurements for the sepa- ration of each material follow a linear trend over time at a single wavelength (500 nm) (Figure 3c). The optical density is observed to decrease over time upon application of the mag- netic field for each material demonstrating excellent separa- tion rates for both materials after exposure to the magnetic field for ≈90 s (Figure 3). However, (Co@Cn)/GNF is observed to reach lower optical densities faster than (Fe@Cn)/GNF, which indicates that (Co@Cn)/GNF is separated from hexane more rapidly.

In order to explain this difference on separation rates, we studied the magnetic properties of (Co@Cn)/GNF and (Fe@ Cn)/GNF (Experimental Section). The magnetic response of both systems under an applied magnetic field displayed hysteresis loops (Figure 3d,e) and exhibited the magnetic param- eters summarized in Table S1 (Supporting Information) typical for ferromagnetic systems. While at 5 T (Fe@Cn)/GNF reaches

higher magnetic saturation values than (Co@Cn)/GNF at both 2 K and 300 K, at low magnetic fields (<0.1 T) the magnetization values observed for (Co@Cn)/GNF are slightly higher than that of the Fe analogue. These observations are also in agreement with the thermal variation measurements of the magnetization performed for both materials at 0.1 T that showed higher mag- netization values for (Co@Cn)/GNF than that of (Fe@Cn)/ GNF in the temperature range 2–300 K (Figure S2, Supporting Information). These results are consistent with our UV–vis measurements for which we observed a better separation with (Co@Cn)/GNF in comparison to (Fe@Cn)/GNF when a small magnetic field (<0.1 T) was applied for the separation at room temperature.

2.2. Catalytic Chemical Reactions within Magnetically Recoverable Carbon Nanoreactors

2.2.1. Preparing and Testing Catalytic Carbon Nanoreactors

After successful demonstration of the magnetic separation of (Co@Cn)/GNF and (Fe@Cn)/GNF nanoreactors, the next step was to introduce catalytically active metal nanoparticles within the GNF nanoreactors in order to utilize these materials in a suitable catalytic reaction. The methods selected here must be compatible with the proposed magnetically recyclable GNF based carbon nanoreactors. To illustrate the catalytic activity of our magnetically separable hybrid materials, the reduction of nitrobenzene was chosen in this work as a model reaction, as it is very important reaction both in industry and academia, with aniline used as a key precursor in the synthesis of chemicals, dyes and pharmaceuticals.[56]

PdNPs and PtNPs were selected as they have been demonstrated as highly active catalysts for the solution phase reduction of nitrocompounds previously.[54–62] The formation of Pd and Pt nanoparticles supported by the GNF nanoreactor (MNPs@GNF; MNPs stands for metal nanopar- ticles where M = Pd or Pt) was initially investigated in the absence of the magnetic nanoparticles to optimize forma- tion conditions of PtNP and PdNP in GNFs using suitable metal precursors (Experimental Section for details), and then tested in the reduction of nitrobenzene using a high pres- sure H2 glass vessel (Scheme S1, Supporting Information). PdNPs@GNF-1 was produced by the thermal decomposi- tion of Pd(acac)2 to form Pd nanoparticles inside the GNF using vacuum filling conditions. The formation of PdNPs was confirmed by HRTEM revealing an average particle size of 10.79 ± 3.86 nm (Experimental Section, and Figure S3a,b in the Supporting Information) and the PdNPs to be located solely at the step edges in the GNF internal channel, while PdNPs@GNF-2 synthesized in solution from Pd2dba3

[66] in GNFs (Experimental Section) resulted in the formation of very small and well-distributed PdNPs, observed mostly inside the GNF attached to the step edges, as revealed by HRTEM, with an average PdNP diameter of 2.26 ± 0.56 nm (Figure 4b,d).

The powder XRD for PdNPs@GNF-1 showed the presence of metallic Pd (Figure S4, Supporting Information) but PdNPs@ GNF-2 did not exhibit clear Pd diffraction patterns due to very

Adv. Funct. Mater. 2018, 1802869

www.afm-journal.dewww.advancedsciencenews.com

small size of the PdNP in this material.[67] Therefore, we synthe- sized a control material, PdNPs@GNF-2*, with a higher metal loading (15 wt% Pd) using the same procedure as for PdNPs@ GNF-2 and observed distinctive Pd diffraction patterns by XRD

confirming decomposition of the starting complex to metallic palladium (Figure 4f).

PtNPs@GNF-1 and PtNPs@GNF-2 were produced using Pt(acac)2 and Pt(dba)3,

[68] using similar experimental

Adv. Funct. Mater. 2018, 1802869

Figure 3. a) Suspensions of (Co@Cn)/GNF (left) and (Fe@Cn)/GNF (right) after applying a magnetic field for 90 s. b) UV–vis measurements for the solutions of (Co@Cn)/GNF (left) and (Fe@Cn)/GNF (right) after magnetic separation at various times. c) Plot of optical density at 500 nm of the solu- tions from (Co@Cn)/GNF and (Fe@Cn)/GNF separations in hexane versus the length of time that the magnetic field (0.1 T) was applied for. d) Mag- netic hysteresis loops for (Co@Cn)/GNF and (Fe@Cn)/GNF recorded at 300 K (inset: expanded region at low magnetic fields between −0.5 and 0.5 T).

www.afm-journal.dewww.advancedsciencenews.com

procedures to those for PdNPs@GNF-1 and PdNPs@GNF-2, respectively (Experimental section). HRTEM imaging of PtNPs@GNF-1 showed Pt nanoparticles distributed along the

step edges of the nanoreactor with an average particle size of 4.21 ± 1.54 nm (Figure 4c,e), and for PtNPs@GNF-2 much smaller nanoparticles with an average size of 1.55 ± 0.48 nm

Adv. Funct. Mater. 2018, 1802869

Figure 4. a) Schematic and b,c) representative HRTEM images of b) PdNPs@GNF-2 and c) PtNPs@GNF-1 (0.5 wt% metal in both cases), and d,e) the histograms showing the size distribution of the PdNPs and PtNPs in their respective composites. Powder XRD patterns of f ) PdNPs@GNF-2 and g) PtNPs@GNF-1. Diffractograms for Pd, Ptreferences, and GNF (annealed at 450 °C for 1 h) are shown for comparison. All * composites show 15 wt% loading of Pt or Pd respectively for comparison.

www.afm-journal.dewww.advancedsciencenews.com

located mostly in the interior of the nanoreactor (Figure S3g,h, Supporting Information). The powder XRD did not exhibit clear diffraction peaks for metallic platinum in both PtNPs@ GNF-1 and PtNPs@GNF-2 due to the small size of Pt nano- particles (Figure S4, Supporting Information). We repeated the synthesis of these material using a higher metal loading (15% Pt by weight) resulting in bigger nanoparticles, allowing clear diffraction patterns of metallic platinum to confirm the presence of metallic PtNP for both materials (Figure S4, Supporting Information).[69]

The reduction of nitrobenzene was then carried out in the presence of all obtained catalysts using high pressure glass equipment and molecular H2, and quantified by

1H NMR (Exper- imental section). The lower catalytic activity of PdNP@GNF-1 compared to PdNPs@GNF-2 is attributed to the larger nanopar- ticle size, and thus lower catalytic surface area of the PdNPs in PdNPs@GNF-1. In contrast, no reactivity was observed for the smaller PtNPs in PtNPs@GNF-2 compared to the larger PtNPs in PtNPs@GNF-1, which were observed to be surprisingly active (Table 1). The reason for the lack of reactivity for the small PtNPs in PtNPs@GNF-2 requires further investigation but we propose that such small nanoparticles could be quite amorphous and therefore have poorly defined sites for catalysis, or may contain some residual dibenzylideneacetone (dba) ligand blocking the surface. The lack of crystallographic planes observed in XRD and HRTEM supports the former argument. Due to the lack of reactivity of PdNPs@GNF-1 and PtNPs@GNF-2, we did not fur- ther investigate these materials in the reduction of nitrobenzene. Therefore, PdNPs@GNF-2 and PtNPs@GNF-1, which both showed significant activity and high aniline selectivity (compared

to n-phenylhydroxylamine), were chosen for the design of cata- lytically active magnetically separable nanoreactors.

The Brunauer–Emmett–Teller (BET) surface area of PdNPs@GNF-2 and PtNPs@GNF-1 was investigated to enable the number of active sites for each catalyst to be approximated (Table S2 and Figure S5, Supporting Information). As we reported in our previous study,[24] empty GNF have a surface area of 12 m2 g−1 and contain mesoporous (2–50 nm) and some macroporous (>50 nm) pores in the carbon structure which is consistent with the size of the step edges (height = 3–5 nm) and the inner channel of the GNF (diameter = 10–100 nm). After adding Pd or Pt nanoparticles to the GNF, the BET surface area of the material increased in both cases, ≈16.0 m2 g–1 for PdNPs@GNF-2 and ≈15.7 m2 g–1 for PtNPs@GNF-1 attributed to the presence of metal nanoparticles in the GNF. Therefore, by subtracting the surface area of GNF from MNP@GNF, sur- face areas of PdNPs and PtNPs are estimated to be 4.0 and 3.7 m2 g–1, respectively.

2.3. Magnetically Recoverable Catalytic Nanoreactors

Two different methodologies were explored for fabrication of magnetically recoverable catalytically active carbon nanoreac- tors Figure 5.

In the first approach, catalytic PdNPs or PtNPs were encap- sulated within GNF using the solution method and gas phase filling method respectively (as described for PdNPs@GNF-2 and PtNPs@GNF-1 above) and then subsequently combined with Co@Cn in hexane using the ultrasound conditions previ- ously optimized (Experimental Section), thus yielding PdNPs@ ((Co@Cn)/GNF) and PtNPs@((Co@Cn)/GNF), with HRTEM analysis confirming the successful adsorption of Co@Cn on the outer surface of the PdNPs@GNF and PtNPs@GNF nanoreactors (Figure S7, Supporting Information). To fabri- cate catalytic magnetic nanoreactor from Fe@Cn, the magnetic component was produced initially to give (Fe@Cn)/GNF and then the catalytic PdNPs or PtNPs were encapsulated within (Fe@Cn)/GNF using solution or gas phase filling methods, respectively (Experimental Section). HRTEM images of PdNPs@((Fe@Cn)/GNF) and PtNPs@((Fe@Cn)/GNF) confirm successful formation of the two MNPs@(Fe@Cn)/GNF mate- rials (Figure S7, Supporting Information).

2.3.1. Catalytic Chemical Reactions within Magnetically Recoverable MNPs@((Fe@Cn)/GNF) and MNPs@ ((Co@Cn)/GNF) Catalysts

Performance of the magnetic catalytically active nanoreactors was tested in the nitrobenzene reduction reaction (Table 2). Both PdNPs@((Co@Cn)/GNF) and PdNPs@((Fe@Cn)/GNF) performed very similarly, exhibiting nitrobenzene TOFs of 69.5 and 69.1, respectively. This is also very similar to the cata- lytic performance of the PdNPs@GNF-2 in the absence of the magnetic nanomaterials (cf. TOF 69.1). PtNPs@((Co@Cn)/ GNF) and PtNPs@((Fe@Cn)/GNF) also exhibited similar reac- tivity to each other (TOFs of 24.1 and 23.2, respectively) and to the unmodified PtNPs@GNF-1 catalyst (cf. TOF of 25.2). All these experiments demonstrate that (Co@Cn)/GNF and

Adv. Funct. Mater. 2018, 1802869

Table 1. Reaction data for the reduction of nitrobenzene using PdNP@ GNF and PtNP@GNF catalytic nanoreactors using a high-pressure H2 glass reactor.

Catalyst Time Conversion of Ph-NO2 [%] Selectivity [%]

Ph-NHOH Ph-NH2

– 24 h 0 0 0

GNFa) 24 h 0 0 0

PdNPs@GNF-1 30 min 3.5 71 29

PdNPs@GNF-2 30 min 77 15 85

PdNPs@GNF-2 50 min 100 0 100

PtNPs@GNF-1 30 min 24 36 64

PtNPs@GNF-1 200 min 100 0 100

PtNPs@GNF-2 30 min 0 0 0

PtNPs@GNF-2b) 24 h 0 0 0

Reaction conditions: Nitrobenzene (0.78 mmol); ethanol (0.5 mL); catalyst (0.00047 mmol of metal); H2 (8 bar); room temperature. All reactions were performed in duplicate and nitrobenzene conversion was determined by 1H NMR with an error of ±2%. a)GNF were annealed at 450 °C for 1 h prior to use; b)PtNPs@GNF-2 was annealed under H2 flow for 5 h at 150 °C prior to the reaction to get rid of any impurities on the surface of Pt which can cause deactivation of the catalyst.

www.afm-journal.dewww.advancedsciencenews.com

(Fe@Cn)/GNF do not adversely affect the activity of the catalytic metal or product selectivity in nanoreactors.

Several studies have reported the reduction of nitro com- pounds under high pressures and temperatures in the presence of Pd and Pt catalyst supported by different materials.[56–65] Karwa et al. studied the effect of reaction temperature, hydrogen pressure and solvent on the selectivity of reduction of nitrobenzene in the presence of Pd and Pt catalysts on carbon supports and reported the formation of phenylhydroxylamine as a by-product at low temperatures, and a better selectivity for phe- nylhydroxylamine in the presence of Pt catalysts compared to Pd catalysts.[57] They also demonstrated that hydrogen pressures between 7–21 atm do not alter the selectivity of the reaction, however, the solvent significantly affected the selectivity for phe- nylhydroxylamine, especially in solvents with higher dielectric constants such as methanol (32.7), which gave lower selectivity

for aniline. This is rationalized as a result of the increased sol- ubility and thus desorption of phenylhydroxylamine from the catalyst into the solvent preventing further hydrogenation to aniline. Takenaka et al. studied Pt/C and Pt/SiO2 in the reduc- tion of nitrobenzene at room temperature using molecular hydrogen (1 and 10 bar) and observed very high selectivity for phenylhydroxylamine (>95%).[61] These results are con- sistent with our data in which we observed a higher selectivity for phenylhydroxylamine in the presence of PtNPs@GNF-1 compared to PdNPs@GNF-2, while getting higher aniline selectivity overall for each catalyst. Sangeetha et al. studied Pd supported on hydrotalcite (HT), MgO and γ-Al2O3 between 225 and 300 °C and obtained the best activity in the presence of Pd/HT with a maximum turnover frequency of ≈0.8 s–1 (48 min–1).[60] Gelder et al. investigated the catalytic ability of Pd supported on different active carbon materials in the reduction

Adv. Funct. Mater. 2018, 1802869

Figure 5. a) Schematic showing the two different approaches taken to form magnetic catalytically active carbon nanoreactors. TEM images of b) PtNPs@ ((Co@Cn)/GNF) and c) PtNPs@((Fe@Cn)/GNF); white and black arrows indicate positions of catalytic and magnetic nanoparticles, respectively.

www.afm-journal.dewww.advancedsciencenews.com

of nitrobenzene using methanol and isopropyl alcohol as sol- vents at 50 °C and observed better catalytic activity in methanol with a turnover frequency of 0.27 s–1 (16.2 min–1), significantly lower than to our PdNPs@GNF-2 (cf. our TOF = 72.3 min–1).[58]

The mechanism of nitrobenzene reduction is still not fully understood. However, the Haber mechanism is generally accepted in the literature and involves two different reaction routes: direct and indirect.[56–62,67] The direct route is based on the reduction of nitrobenzene to nitrosobenzene (Ph-NO) and consecutive formation to phenylhydroxylamine (Ph-NHOH) and aniline (Ph-NH2). In light of the Haber mechanism, sev- eral research groups have proposed different additions/altera- tions to the reduction mechanism.[70–73] Gelder et al.[71] recently proposed a new mechanism which is contrary to the Haber process and showed that nitrosobenzene cannot be an interme- diate in the formation of aniline. Our study revealed reduction of nitrobenzene to phenylhydroxylamine to form aniline which is consistent with the direct Haber process, however, we did not observe nitrosobenzene or any other side products. We there- fore propose that in our nanoreactors the reaction pathway is consistent with the Gelder mechanism (Scheme 2).

2.3.2. Catalyst Durability Test in the Magnetic Recovery Process of MNPs@(Co@Cn)/GNF and MNPs@(Fe@Cn)/GNF

We investigated the reusability and durability of PdNPs@(Co@ Cn)/GNF, PdNPs@(Fe@Cn)/GNF, PtNPs@(Co@Cn)/GNF, and PtNPs@(Fe@Cn)/GNF in the reduction of nitrobenzene. In each case the magnetic nanoreactor catalyst was separated

from the product after each run by applying a magnetic field and reused after washing with ethanol and drying in air without any other treatment and compared to the corresponding PdNPs@ GNF-2 and PtNPs@GNF-1 system recovered by traditional fil- tration using a PTFE membrane filter and washing with ethanol (Tables S3 and S4 in the Supporting Information and Figure 6a,b). Recyclability tests showed a decrease in nitrobenzene conversion during the five recoveries of each catalyst, however, in each case magnetic recovery resulted in slightly lower loss of activity during the five runs compared to traditional catalyst recovery by filtration.

To further explore the reasons for decrease in catalytic activity by traditional recovery compared to magnetic recovery, the PdNP@GNF catalysts, PdNPs@GNF-2 after traditional recovery, and PdNPs@((Co@Cn)/GNF) and PdNPs@((Fe@Cn)/GNF) by magnetic recovery, were weighted after each recovery step and an appreciable loss is observed, cf. 10% by traditional recovery, as compared to <0.5% by magnetic recovery during the five cycles (Figure 6c). The principle cause of catalyst loss during filtra- tion is that some of the material it is irreversibly absorbed onto the PTFE membrane filter. The reduced activity of each catalyst during the reaction, using both filtration and magnetic recovery, is most likely to be related to coarsening of the nanoparticles during the reaction, along with the leaching of Pd and Pt into the reaction medium. To probe whether there is a loss of catalytically active metal from the GNF by leaching, the precise metal loading of both PdNPs@GNF-2 and PtNPs@GNF-1 as synthesized and recovered after five cycles was determined by inductively cou- pled plasma optical emission spectrometry (ICP-OES) revealing a 3.76% leaching of the PdNPs from GNF compared to a 3.56% leaching of PtNPs (Table S5, Supporting Information).

These results imply that the leaching of Pd and Pt nanopar- ticles from the GNF is very low and therefore cannot be the reason for the reduction of the catalyst activity during the recy- cling. However, nanoparticle coarsening via Ostwald ripening or particle migration and coalescence, could also be occur- ring that would result in a decrease in active catalyst surface area, and thus a reduction of the activity of catalyst. Therefore, HRTEM imaging of the catalytic nanoreactors after five reac- tion cycles was performed (Figure S8, Supporting Information), showing some aggregation of the PdNPs, and excellent dis- persion of the PtNPs in PdNPs@GNF-2 and PtNPs@GNF-1, respectively. Interestingly, the average particle sizes after the fifth cycle measured by HRTEM of 2.4 ± 0.4 nm for the PdNPs, and 4.4 ± 1.0 nm for the PtNPs, were almost identical to the average size of the nanoparticles before the reaction. Therefore, the reduction in activity must be related to other factors, such as re-ordering of the nanoparticle structure or poisoning of the surface, with further work required to clarify this.

3. Conclusions

We have developed and compared several approaches for fab- ricating catalytically active nanoreactors, containing Pt or

Adv. Funct. Mater. 2018, 1802869

Ph-NO2 Ph-NOH adsorbed

Ph-N(OH)H Ph-NH2Ph-NH

Scheme 2. Proposed reaction pathway of the reduction of nitrobenzene.

Table 2. Reaction data for the reduction of nitrobenzene using (Co@ Cn)/GNF, PdNPs@((Co@Cn)/GNF) and PtNPs@((Co@Cn)/GNF), (Fe@Cn)/GNF, and PdNPs@((Fe@Cn)/GNF) and PtNPs@((Fe@Cn)/ GNF) catalyts using a high pressure H2 glass reactor.

Catalyst Conversion of Ph-NO2 [%]/TOF [min−1]

Selectivity [%]

Ph-NHOH Ph-NH2

PdNPs@GNF-2 77/72.3 15 85

PtNPs@GNF-1 24/25.2 36 64

(Co@Cn)/GNF 0 0 0

PdNPs@((Co@Cn)/GNF) 74/69.5 14 86

PtNPs@((Co@Cn)/GNF) 23/24.2 43 57

(Fe@Cn)/GNF 0 0 0

PdNPs@((Fe@Cn)/GNF) 72/69.1 16 84

PtNPs@((Fe@Cn)/GNF) 22/23.2 32 68

Reaction conditions: Nitrobenzene (0.08 mL, 0.78 mmol); ethanol (0.5 mL); cata- lyst (0.00051 mmol); H2 (8 bar); room temperature; 30 min. All reactions were per- formed in duplicate and nitrobenzene conversion was determined by 1H NMR with an error of ±2%. The TOFs were calculated as the ratio of the number of molecules of reactant consumed in the reaction to the number of true active catalyst sites calculated by BET measurements per minute.

www.afm-journal.dewww.advancedsciencenews.com

Pd nanoparticles confined within GNF, functionalized with carbon-coated Co or Fe nanomagnets attached to surfaces of the nanoreactors. Both, Co and Fe nanomagnets enable the facile separation of catalytic nanoreactors from the products mixtures in a fast, easy and efficient way by simply applying a magnetic field. This has significant advantages over cumber- some and energy consuming filtration methods, traditionally used for separation of catalysts.

The catalytic performance of Pt- and PdNPs encapsulated in GNF was probed in the reduction of nitrobenzene with both materials exhibiting excellent activity and selectivity, especially PdNPs@GNF. Importantly, adsorption of the magnetic nano- particles Co@Cn or Fe@Cn on the exterior of the catalytic nano- reactors was found to have no negative effects on the catalytic performance, while significantly reducing the loss of catalytic material over five cycles of reaction, compared to filtration.

Magnetic nanoreactors allow a combination of retention and recyclability of catalytically active metals offered by the GNF cavity, with the magnetic functionality enabling facile re-use of the catalytic material. This study lays the foundations for gen- eration of a diverse family of magnetically separable carbon nanoreactors and gives guidance for future development of metal-catalyzed reactions in magnetic carbon nanoreactors,

which in the long term can be scaled-up and applied for chem- ical processes of industrial importance streamlining catalysis and synthesis.

4. Experimental Section Chemicals: GNF were purchased from Pyrograf Products Inc

(PR19, chemical vapor deposition), USA. Co@Cn was purchased from Turbobeads LLC, USA. All other reagents and solvents were purchased from Sigma-Aldrich (UK) and used without further purification.

The glassware required to perform the experiments was thoroughly cleaned with “aqua regia” (concentrated hydrochloric and nitric acids (3:1)) and rinsed with deionized water prior to use.

Synthesis of (Co@Cn)/GNF: GNF (15 mg, annealed at 450 °C in air for 1 h prior to use) in hexane (20 mL) and Co@Cn (1.5 mg, corresponding to 10 wt% Co in the final (Co@Cn)/GNF material) in hexane (5 mL) were dispersed separately using an ultrasonication bath (3 L Ultrasonic cleaner, Agar Scientific, 100 W and 40 kHz) for 10 min. Once dispersed the Co@Cn hexane suspension was slowly added to the GNF dispersion in small portions (0.2 mL) while being treated with ultrasonic waves, the resultant dispersion was then sonicated for a further 10 min. Separation of resultant (Co@Cn)/GNF as a black powder was achieved by applying a magnet to the outside of the vial and decanting the hexane solvent.

Synthesis of (Fe@Cn)/GNF: Ferrocene (5 mg, corresponding to 10 wt% Fe in the final (Fe@Cn)/GNF) was combined with GNF (15 mg,

Adv. Funct. Mater. 2018, 1802869

Figure 6. a) Comparison of the catalyst TOF for nitrobenzene conversion during the five consecutive reduction reactions in which PdNPs@GNF-2 was recovered by filtration on a PTFE membrane and PdNPs@((Fe@Cn)/GNF) and PdNPs@((Co@Cn)/GNF) by magnetic separation. b) Comparison of the catalyst TOF for nitrobenzene conversion during the five consecutive reduction reactions in which PtNPs@GNF-1 was recovered by filtration on a PTFE membrane and PtNPs@((Fe@Cn)/GNF) and PtNPs@((Co@Cn)/GNF) by magnetic separation. c) Comparison of the loss of PdNPs@GNF-2 catalyst material during five consecutive nitrobenzene reduction reactions. The recovery of PdNPs@GNF-2 was achieved by filtration using a PTFE membrane, while PdNPs@((Fe@Cn)/GNF) and PdNPs@((Co@Cn)/GNF) were recovered by applying magnetic separation.

www.afm-journal.dewww.advancedsciencenews.com

annealed at 450 °C for 1 h prior to use) in a Pyrex tube and sealed under vacuum (10–6 bar) using a vacuum pump. The material was then heated at 350 °C for 1 d, and the temperature was then increased to 500 °C for a further day. The sample was then cooled and opened to yield the (Fe@Cn)/GNF material as black powder.

Synthesis of PdNPs@GNF-1: Pd(acac)2 (0.22 mg, corresponding to 0.5 wt% Pd in the final PdNPs@GNF-1) was combined with GNF (15 mg, annealed at 450 °C for 1 h prior to use) in a Pyrex tube and sealed under vacuum (10–6 bar) using a vacuum pump and heated at 150 °C for 3 d. After 3 d, the sample inside the Pyrex tube was cooled by immersing in an ice bath. The sample was then removed from the Pyrex tube and sealed in a separate Pyrex tube under argon atmosphere and heated at 550 °C for 3 h. The final material was then cooled to give PdNPs@GNF-1 as black powder.

Synthesis of PdNPs@GNF-2: GNFs (15 mg, annealed at 450 °C for 1 h prior to use) were dispersed in CHCl3 (2 mL) using ultrasound for 10 min. A solution of tris(dibenzylideneacetone)dipalladium(0)- chloroform adduct (Pd2(dba)3·CHCl3) (0.375 mg, corresponding to a 0.5 wt% of Pd in the final PdNPs@GNF-2 material) in CHCl3 (1 mL) was then slowly added to the GNF dispersion in small portions (0.1 mL) while being treated with ultrasonic waves and stirred at 40 °C for 4 h until the solution became colorless. PdNPs@GNF was then separated from the reaction mixture by filtration and washed repeatedly with acetone (20 mL) using a 0.2 µm PTFE membrane filter to remove free dba to yield the PdNPs@GNF-2 material as black powder.

Synthesis of PtNPs@GNF-1: Pt(acac)2 (0.30 mg, corresponding to a 1 wt% of Pt in the final PtNPs@GNF-1 material) was combined with GNF (15 mg, annealed at 450 °C for 1 h prior to use) in a Pyrex tube and sealed under vacuum (10−6 bar) using a vacuum pump and heated at 170 °C for 3 d. After 3 d, the sample was cooled by immersing in an ice bath and then removed from the Pyrex tube and sealed in a separate Pyrex tube under an argon atmosphere and heated at 550 °C for 3 h. The final PtNPs@GNF-1 was then recovered as black powder.

Synthesis of PtNPs@GNF-2: GNF (15 mg, annealed at 450 °C for 1 h prior to use) were dispersed in CHCl3 (2 mL) using ultrasound for 10 min. A solution of tris(dibenzylideneacetone)platinum(0) (Pt(dba)3) (0.69 mg, corresponding to a 1 wt% of Pt in the final PtNPs@GNF-2 material) in CHCl3 (1 mL) was then slowly added to the GNF dispersion in small portions (0.1 mL) while being treated with ultrasonic waves and stirred at 70 °C for 1 d until the solution became colorless. PtNPs@GNF was then separated from the reaction mixture by filtration and washed repeatedly with acetone (20 mL) using a 0.2 µm PTFE membrane filter to remove free dba and obtain the final PtNPs@GNF-2 material as black powder.

Synthesis of Magnetic PdNPs@((Co@Cn)/GNF): PdNPs@GNF-2 (15 mg) in hexane (20 mL) and Co@Cn (1.5 mg) in hexane (5 mL) were dispersed separately using ultrasound for 10 min. The Co@Cn hexane suspension was then added to the GNF dispersion in small portions (0.1 mL) while being treated with ultrasonic waves, the resultant dispersion was then sonicated for a further 10 min. The separation of catalyst from the solution was controlled by an external magnet (0.1 T) to give PdNPs@((Co@Cn)/GNF) as black powder.

Synthesis of Magnetic PdNPs@((Fe@Cn)/GNF): (Fe@Cn)/GNF (15 mg) was dispersed in CHCl3 (2 mL) using ultrasound for 10 min. Once GNF was dispersed, Pd2(dba)3·CHCl3 (0.34 mg) dissolved in CHCl3 (1 mL) were added to GNF dispersion in small portions (0.1 mL) while being treated with ultrasonic waves and stirred at 40 °C for 4 h until the solution became colorless. PdNPs@GNF was then separated from the reaction mixture by filtration using a 0.2 µm PTFE membrane filter and washed repeatedly with acetone (20 mL) to remove the free dibenzylideneacetone and give PdNPs@((Fe@Cn)/GNF) as black powder.

Synthesis of Magnetic PtNPs@((Co@Cn)/GNF): PtNPs@GNF-1 (15 mg) in hexane (20 mL) and Co@Cn (1.5 mg) in hexane (5 mL) were dispersed separately using ultrasound for 10 min. Once dispersed the Co@Cn in hexane were slowly added to GNF dispersion in small portions while being treated with ultrasonic waves, the resultant dispersion was then sonicated for a further 10 min. Separation of the resultant PtNPs@

((Co@Cn)/GNF) as black powder was achieved by applying a magnetic field (0.1 T) to the outside of the vial and decanting the hexane solvent.

Magnetic PtNPs@((Fe@Cn)/GNF): Pt(acac)2 (0.3 mg, 1wt% Pd) was combined with (Fe@Cn)/GNF (15 mg) in a Pyrex tube, sealed under vacuum (10–6 bar) using a vacuum pump and heated at 170 °C for 3 d. The sample inside the Pyrex tube was then cooled by immersing in an ice bath. The sample was removed from the Pyrex tube and sealed in a separate Pyrex tube under an argon atmosphere and heated at 550 °C for 3 h. PtNPs@((Fe@Cn)/GNF) was then isolated as black powder.

Reduction of Nitrobenzene Using a High Pressure H2 Glass Reactor: The selected catalyst (10 mg, equivalent to 0.00051 mmol of metal nanoparticles) and an ethanol (0.5 mL) solution of nitrobenzene (0.78 mmol) were stirred in a high-pressure reactor (10 mL volume). The reactor volume was then degassed thoroughly with H2 for 15 min. The reactor was then sealed and pressurized with H2 (8 bar) and left for 30 min at room temperature. At the end of the reaction, the reactor was slowly depressurized and analyzed via 1H NMR spectroscopy using CDCl3 solvent. Nitrobenzene (Ph-NO2):

1H NMR (300 MHz, 297 K, CDCl3, δ, ppm): 8.20–8.17 (m, 2H; CHCH), 7.69–7.63 (m, 1H;CH, 7.53–7.48 (m, 2H; CHCH). N-phenylhydroxylamine (Ph-NHOH): 1H NMR (300 MHz, 297 K, CDCl3, δ, ppm): 7.23–7.18 (m, 2H; CHCH), 6.95–6.93 (d, J = 7.54 Hz, 2H; CHCH), 6.90–6.85 (m, 1H; CH). Aniline (Ph-NH2):

1H NMR (300 MHz, 297 K, CDCl3, δ, ppm): 7.13–7.07 (t, J = 7.86 Hz, 2H, CHCH), 6.73–6.69 (m, 1H; CH), 6.68–6.64 (m, 2H; CHCH).

Characterization Techniques: 1H NMR spectra were recorded using a Bruker DPX300 NMR spectrometer. 1H NMR spectra were taken in CDCl3 and were referenced to residual trimethysilane (TMS) (0 ppm) and reported as follows: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, dd = doublet of doublet, m = multiplet). HRTEM analysis was performed on a JEOL 2100 Field emission gun transmission electron microscope with an information limit of 0.12 nm at 100 kV. Samples for HRTEM analysis were prepared by dispersing the materials in HPLC grade isopropanol using ultrasonication, then drop casting the resultant suspension onto a lacey carbon film coated copper grid. TGA analysis was performed on a TA Instruments TGA-SDTQ600 analyzer. Samples for TGA analyses were heated in air up to 1000 °C with a heating rate of 10 °C min−1. The powder X-ray diffraction patterns were obtained using a PANanalytical X’Pert PRO diffractometer equipped with a Cu-Ka radiation source (λ = 1.542) operating at 40 kV and 40 mA, with 0.05252° step size and a step time of 5925.18 s. Surface area analysis was performed using the BET method based on adsorption data in the relative pressure (P/Po) range 0.02 to 0.22 by measuring nitrogen sorption isotherms of the samples (50 mg) at −196 °C on a Micromeritics ASAP 2020 sorptometer. Before analysis, the samples were evacuated for 12 h at 200 °C under vacuum. The pore size distributions were obtained from a Nonlocal Density Functional Theory method using nitrogen-sorption data.

UV–Vis Spectroscopy Measurements: (Co@Cn)/GNF (10 mg) and (Fe@Cn)/GNF (10 mg) were suspended in hexane (10 mL) by stirring at 500 rpm using a magnetic stirrer for 2 min and then a magnetic field was applied for a set periods of time (10, 30, 60, and 90 s). For example, after applying a magnetic field for 10 s, 2 mL of solution were taken from the magnetically separated solution using a micro pipette then analyzed by UV–vis spectroscopy between 350–700 nm (wavelength step: 1 nm, scan speed: 240 nm min–1 ) using a Perkin Elmer Lambda 11 spectrophotometer. The 2 mL solution was then returned to the starting solution and the solution was re-dispersed and the same procedure was repeated.

Magnetic measurements on (Co@Cn)/GNF and (Fe@Cn)/GNF were carried out in a commercial Quantum Desing MPMS-XL5 Superconducting Quantum Interference Device magnetometer. Samples were carefully prepared using a plastic capsule with a negligible diamagnetic contribution. For both samples variable-temperature (1.8300 K, with 0.1 T applied field) and field dependent (at 2 K and 300 K with a maximum field of 5 T) magnetization measurements were carried and compared by dividing the magnetic signal per mass of the measured sample.

Adv. Funct. Mater. 2018, 1802869

www.afm-journal.dewww.advancedsciencenews.com

ICP-OES Measurements: PdNPs@GNF-2 and PtNPs@GNF-1 samples (3 × 2 mg) were burned in a boiling tube using a bunsen burner to fully oxidize and hence remove the GNF. The remaining residual metal was then digested in acid (1 mL, aqua regia) at room temperature using ultrasonication for 1 h and the resultant solution was diluted with ultrapure water to make a dilute aqua regia solution (10% by volume in water). ICP-OES was used to determine the Pd and Pt% loadings of the solutions, respectively using a Perkin Elmer, Optima 2000 DV ICP-OES with S10 autosampler with an axial detection method at wavelengths of 340.458 nm for Pd and 214.423 nm for Pt. Calibration Pd and Pt solutions (0.1, 0.2, 0.5, 1, and 5 mg L–1) were prepared using a Pd standard (Sigma-Aldrich) and a Pt standard (VWR Chemicals) and aqua regia (10% by volume in water), and gave a linear plot with an R coefficient of 0.999999. Blanks showed 0.00 mg L–1 of Pd and Pt, respectively. Corrected concentrations of Pd and Pt were then measured as number of mg of Pd and Pt per liter for each sample and correlated to Pd and Pt% loadings.

Catalyst Recovery by Applying a Magnetic Field: After each experimental cycle, the catalyst mixed with reaction products were extracted into ethanol (5 mL) and then applied magnetic field. As the catalyst accumulated on the wall of reaction vessel, the solution mixture was easily separated from the catalyst using a pipette. Ethanol (5 mL) was then added to the catalyst and the same procedure was repeated until no signs of starting materials or products could be observed by 1H NMR. The catalyst was then left to dry at room temperature.

Catalyst Recovery by Filtration: After each experimental cycle, the catalyst, mixed with reaction products, was extracted into ethanol (5 mL) and then washed with ethanol (20 mL) using a filtration assembly and a PFTE membrane followed by drying at ambient conditions. The washings were repeated until no signs of starting materials or products could be observed by 1H NMR. The catalyst was then collected from the filtration membrane using a spatula.

Supporting Information Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements The authors acknowledge The Ministry of National Education, The Republic of Turkey (studentship for M.A.), The Royal Society (DH110080 fellowship for M.C.G.), EPSRC and ERC (StG-277784 for A.N.K. and StG- 679124 for M.C.G.) for funding this work, the Nanoscale & Microscale Research Centre (nmRC) for access to TEM and Dr. Alessandro La Torre for technical assistance with acquisition of TEM data.

Conflict of Interest The authors declare no conflict of interest.

Keywords magnetic nanoparticles, nanocatalysis, nanoreactors, nanotubes, nitrobenzene reduction

Received: April 26, 2018 Revised: June 9, 2018

Published online:

[1] S. Iijima, Nature 1991, 354. [2] M. S. Dresselhaus, G. Dresselhaus, A. Jorio, Annu. Rev. Mater. Res.

2004, 34, 247.

[3] S. A. Hodge, M. K. Bayazit, K. S. Coleman, M. Shaffer, Chem. Soc. Rev. 2012, 41, 4409.

[4] A. Mamalis, Precis. Eng. 2004, 28, 16. [5] M. Terrones, Int. Mater. Rev. 2004, 49, 325. [6] M. S. Dresselhaus, G. Dresselhaus, P. Avouris, Carbon Nanotubes:

Synthesis, Structure, Properties, and Applications, Springer, New York 2001.

[7] G. A. Rance, D. H. Marsh, S. J. Bourne, T. J. Reade, A. N. Khlobystov, ACS Nano 2010, 4, 4920.

[8] M. Daenen, R. D. De Fouw, B. Hamers, P. G. A. Janssen, K. Schouteden, M. A. J. Veld, The Wondrous World of Carbon Nanotubes ‘A Review of Current Carbon Nanotube Technologies’, Eindhoven University of Technology, Eindhoven, the Netherlands 2003, pp. 1–96.

[9] H. Chu, L. Wei, R. Cui, J. Wang, Y. Li, Coord. Chem. Rev. 2010, 254, 1117.

[10] A. N. Khlobystov, ACS Nano 2011, 5, 9306. [11] P. Serp, J. L. Figueiredo, Carbon Materials for Catalysis, Wiley,

Hoboken, NJ 2009, 56. [12] G. G. Wildgoose, C. E. Banks, R. G. Compton, Small 2006, 2, 182. [13] W. A. Solomonsz, G. A. Rance, B. J. Harris, A. N. Khlobystov,

Nanoscale 2013, 5, 12200. [14] M. J. Ledoux, R. Vieira, C. Pham-Huu, N. Keller, J. Catal. 2003, 216,

333. [15] T. W. Chamberlain, J. H. Earley, D. P. Anderson, A. N. Khlobystov,

R. A. Bourne, Chem. Commun. 2014, 50, 5200. [16] P. Serp, M. Corrias, P. Kalck, Appl. Catal., A 2003, 253, 337. [17] S. A. Miners, G. A. Rance, A. N. Khlobystov, Chem. Commun. 2013,

49, 5586. [18] W. A. Solomonsz, G. A. Rance, M. Suyetin, A. La Torre,

E. Bichoutskaia, A. N. Khlobystov, Chem. – Eur. J. 2012, 18, 13180. [19] M. A. Lebedeva, T. W. Chamberlain, A. Thomas, B. E. Thomas,

C. T. Stoppiello, E. Volkova, M. Suyetind, A. N. Khlobystov, Nanoscale 2016, 8, 11727.

[20] X. Pan, Z. Fan, W. Chen, Y. Ding, H. Luo, X. Bao, Nat. Mater. 2007, 6, 507.

[21] B. Cornelio, A. R. Saunders, W. A. Solomonsz, M. Laronze-Cochard, A. Fontana, J. Sapi, A. N. Khlobystov, G. A. Rance, J. Mater. Chem. A 2015, 3, 3918.

[22] W. A. Solomonsz, G. A. Rance, A. N. Khlobystov, Small 2014, 10, 1866. [23] G. A. Rance, W. A. Solomonsz, A. N. Khlobystov, Chemm. Commun.

2013, 49, 1067. [24] M. Aygün, C. T. Stoppiello, M. A. Lebedeva, E. F. Smith,

M. D. C. Gimenez-Lopez, A. N. Khlobystov, T. W. Chamberlain, J. Mater. Chem. A 2017, 5, 21467.

[25] M. Gimenez-Lopez, A. Kurtoglu, D. A. Walsh, A. N. Khlobystov, Adv. Mat. 2016, 28, 9103.

[26] X. Pan, X. Bao, Chem. Commun. 2008, 6271. [27] X. Pan, X. Bao, Acc. Chem. Res. 2011, 44, 553. [28] A. Hutten, D. Sudfeld, I. Ennen, G. Reiss, W. Hachmann,

U. Heinzmann, K. Wojczykowaki, P. Jutzi, W. Saikaly, G. J. Thomas, J. Biotechnol. 2004, 112, 47.

[29] A. H. Lu, W. Schmidt, N. Matoussevitch, H. Bönnemann, B. Spliethoff, B. Tesche, E. Bill, W. Kiefer, F. Schüth, Angew. Chem. 2004, 116, 4403.

[30] H. Jia, Y. Lian, M. Ishitsuka, T. Nakahodo, Y. Maeda, T. Tsuchiya, T. Wakahara, T. Akasaka, Sci. Technol. Adv. Mater. 2005, 6, 571.

[31] L. M. Rossi, N. J. S. Costa, F. P. Silva, R. V. Gonçalves, Nanotechnol. Rev. 2013, 2, 597.

[32] J. Govan, Y. K. Gun’ko, Nanomaterials 2014, 4, 222. [33] L. M. Rossi, N. J. S. Costa, F. P. Silva, R. Wojcieszak, Green Chem.

2014, 16, 2906. [34] M. D. Lekgoathi, W. G. Augustyn, J. Heveling, J. Nanosci. Nanotechnol.

2011, 11, 7001. [35] H. M. Torres Galvis, J. H. Bitter, C. B. Khare, M. Ruitenbeek,

A. I. Dugulan, K. P. de Jong, Science 2012, 335, 835.

www.afm-journal.dewww.advancedsciencenews.com

[36] M. Melchionna, A. Beltram, A. Stopin, T. Montini, R. W. Lodge, A. N. Khlobystov, D. Bonifazi, M. Prato, P. Fornasiero, Appl. Catal., B 2018, 227, 356.

[37] R. N. Grass, W. J. Stark, J. Mater. Chem. 2006, 16, 1825. [38] I. K. Herrmann, R. N. Grass, D. Mazunin, W. J. Stark, Chem. Mater.

2009, 21, 3275. [39] Y. Deng, Y. Cai, Z. Sun, J. Liu, J. Wei, W. Li, C. Liu, Y. Wang, D. Zhao,

J. Am. Chem. Soc. 2010, 132, 8466. [40] L. Zhang, L. Pinhua, L. Hongji, L. Wang, Catal. Sci. Technol. 2012, 2,

1859. [41] A. H. Lu, E. L. Salabas, F. Schüth, Angew. Chem., Int. Ed. 2007, 46, 1222. [42] S. C. Tsang, V. Caps, I. Paraskevas, D. Chadwick, D. Thompsett,

Angew. Chem., Int. Ed. 2004, 43, 5649. [43] Z. Xu, H. Li, G. Cao, Z. Cao, Q. Zhang, K. Li, X. Hou, W. Lia, W. Cao,

J. Mater. Chem. 2010, 20, 8230. [44] M. Rossier, F. M. Koehler, E. K. Athanassiou, R. N. Grass,

B. Aeschlimann, D. Günther, W. J. Stark, Mater. Chem. 2009, 19, 8239. [45] S. Wittmann, A. Schätz, R. N. Grass, W. J. Stark, O. Reiser,

Angew. Chem., Int. Ed. 2010, 49, 1867. [46] G. U. Sumanasekera, B. K. Pradhan, H. E. Romero, K. W. Adu,

P. C. Eklund, Phys. Rev. Lett. 2002, 89, 166801. [47] I. K. Herrmann, M. Urner, F. M. Koehler, M. Hasler,

B. Roth-Z’Graggen, R. N. Grass, U. Ziegler, B. Beck-Schimmer, W. Stark, Small 2010, 6, 1388.

[48] A. Schätz, T. R. Long, R. N. Grass, W. J. Stark, P. R. Hanson, O. Reiser, Chem. Mater. 2010, 20, 4323.

[49] A. Schätz, R. N. Grass, Q. Kainz, W. J. Stark, O. Reiser, Chem. Mater. 2010, 22, 305.

[50] B. K. T. Kenneth, C. Singh, M. Chhowalla, W. I. Milne, Encycl. Nanosci. Nanotechnol. 2004, 1, 665.

[51] R. Saito, G. Dresselhaus, M. S. Dresselhaus, Physical Properties of Carbon Nanotubes, Imperial College Press, London 1998.

[52] A. W. Hull, Phys. Rev. 1917, 10, 661. [53] A. Combessis, C. Mazel, M. Maugin, L. Flandin, J. Appl. Polym. Sci.

2013, 130, 1778.

[54] S. H. Jeong, K. K. Kim, S. J. Jeong, K. H. An, S. H. Lee, Y. H. Lee, Synth. Met. 2007, 157, 570.

[55] I. M. L. Billas, A. Châtelain, W. A. de Heer, Science 1994, 265, 1682. [56] R. S. Downing, P. J. Kunkeler, H. van Bekkum, Catal. Today 1997,

37, 121. [57] S. L. Karwa, K. A. Rajadhyaskha, Ind. Eng. Chem. Res. 1987, 26,

1746. [58] E. A. Gelder, S. D. Jackson, C. M. Lok, Catal. Lett. 2002, 84, 205. [59] C. H. Li, Z. X. Yu, K. F. Yao, S. F. Ji, J. Liang, J. Mol. Catal. A 2005,

226, 101. [60] P. Sangeetha, K. Shanthi, K. S. Rama Rao, B. Viswanathan,

P. Selvam, Appl. Catal., A 2009, 353, 160. [61] Y. Takenaka, T. Kiyosu, J. Choi, T. Sakakuraa, H. Yasuda, Green

Chem. 2009, 11, 1385. [62] F. A. Westerhaus, R. W. Jagadeesh, G. Wienho, M. Pohl, J. Radnik,

A. Surkus, J. Rabeah, K. Junge, H. Junge, M. Nielsen, A. Bruckner, M. Beller, Nat. Chem. 2013, 5, 537.

[63] M. Shokouhimehr, T. Kim, J. S. Woojoo, S. Kwangsoo, Y. Jang, B. H. Kim, J. Kim, T. Hyeon, Appl. Catal., A 2014, 476, 133.

[64] M. Turáková, M. Králik, P. Lehocky, P. Pikna, M. Smrcová, D. Remeteiová, A. Hudák, Appl. Catal., A 2014, 476, 103.

[65] R. B. N. Baig, R. S. Varma, ACS Sustainable Chem. Eng. 2014, 2, 2155.

[66] E. O. Pentsak, A. S. Kashin, M. V. Polynski, K. O. Kvashnina, P. Glatzelc, V. P. Ananikov, Chem. Sci. 2015, 6, 3302.

[67] A. W. Hull, W. P. Davey, Phys. Rev. 1921, 17, 571. [68] E. O. Pentsak, V. P. Ananikov, Russ. Chem. Bull. 2014, 63, 2560. [69] H. E. Swanson, E. Tatge, Standard X-ray diffraction powder patterns

I, National Bureau of Standards, U.S., Circular, 1953. [70] Z. Haber, Z. Elektrochem. 1898, 4, 506. [71] E. A. Gelder, S. D. Jackson, C. M. Lok, Chem. Commun. 2005, 4, 522. [72] A. Corma, P. Concepcion, P. Serna, Angew. Chem., Int. Ed. 2007, 46,

7266. [73] M. Turákováa, T. Salmi, K. Eränenc, J. Wärnåc, D. Y. Murzinc,

M. Králik, Appl. Catal., A 2015, 499, 66.

Adv. Funct. Mater. 2018, 1802869