Performance of a NiFe2O4@Co Core-Shell Fischer-Tropsch Catalyst: Effect of Low Temperature Reduction.

In situ TEM gas-cell imaging and spectroscopy with in situ XRD have been applied to reveal morphological changes in NiFe2O4@Co3O4 core-shell nanoparticles in hydrogen. The core-shell structure is retained upon reduction under mild conditions (180 °C for 1 h), resulting in a partially reduced shell. The core-shell structure was retained after exposing these reduced NiFe2O4@Co3O4 core-shell nanoparticles to Fischer-Tropsch conditions at 230 °C and 20 bar. Slightly harsher reduction (230 °C, 2 h) resulted in restructuring of the NiFe2O4@Co3O4 core-shell nanoparticles to form cobalt islands in addition to partially reduced NiFe2O4. NiFe2O4 underwent further transformation upon exposure to Fischer-Tropsch conditions, resulting in the formation of iron carbide and nickel/iron-nickel alloy. The turnover frequency in the Fischer-Tropsch synthesis over NiFe2O4@Co3O4 core-shell nanoparticles reduced in hydrogen at 180 °C for 1 h was estimated to be less than 0.02 s-1 (cobalt-time yield of 8.40 μmol.g-1.s-1) with a C5+ selectivity of 38 C-%. The low turnover frequency under these conditions in relation to the turnover frequency obtained with unsupported cobalt is attributed to the strain in the catalytically active cobalt.

Introduction

The Fischer–Tropsch synthesis can be used for the conversion of natural gas and coal[1] into a wide range of valuable hydrocarbon products, although biomass or waste[2] can also be used as the starting carbon-containing material. Cobalt-based catalysts are preferred for the low-temperature Fischer–Tropsch synthesis, yielding waxes and diesel, because of their higher activity, hydrocarbon productivity, and good stability.[2−8] The disadvantage of cobalt as the catalytically active material is its relatively high cost.[2,5] Core–shell nanoparticles may be used to reduce the required amount of catalytically active material[9,10] while exposing a large, catalytically active surface area.

Core–shell nanoparticles are composed of two or more metals or metal oxides and are made up of an inner layer material (core) and an outer layer material (shell).[9,11,12] The cost of the catalyst may be reduced by using a relatively cheap material as the core with the catalytically active material as the shell. These systems have been shown to be efficient catalysts for a variety of reactions, but their performance may differ from that observed over a monometallic catalyst or an alloy composed of the same metals.[9,11−13] Core–shell nanoparticles have also been shown to be promising catalysts for the Fischer–Tropsch synthesis when the active cobalt metal was present as a core surrounded by either a metal oxide or metallic species shell.[14−17] However, these systems may induce mass transfer limitations when demanding a high productivity from the catalyst.

Nanoparticles with a cobalt(II,III) oxide (Co3O4) shell around a core of an inverse spinel, Fe3O4, have been used as a precursor for a Fischer–Tropsch catalyst;[10] the use of an inverse spinel allows the preferential epitaxial growth of Co3O4 as a shell layer. We followed a similar strategy for the synthesis of NiFe2O4@Co3O4 core–shell nanoparticles.[18] The overall size and shell thickness observed for these nanoparticles correspond to a significant replacement of cobalt with approximately 82% of the crystallite having been replaced by a material cheaper than cobalt. However, reduction of these core–shell nanoparticles NiFe2O4@Co3O4 at temperatures as low as 230 °C for 2 h in hydrogen effectively destroyed the core–shell structure, resulting in the formation of cobalt islands on the core.[19]

There are in principle two strategies to ensure retention of the core–shell structure, i.e., increasing the shell thickness or reducing the severity of the activating reduction step. The former has the disadvantage that more catalytically active material is required. Here, the successful use of milder reducing conditions (180 °C) for maintaining the core–shell structure of NiFe2O4@Co3O4 and its effect on Fischer–Tropsch synthesis are reported.

Results and Discussion

The NiFe2O4@Co3O4 core–shell nanoparticles composed of a NiFe2O4 core with a diameter of ca. 14 ± 0.5 nm and a 2.5 ± 0.7 nm shell (see Figure S.1 in the Supporting Information) were synthesized. With a cobalt loading of 8.3 wt %, it can be easily seen that this synthesis method leaves quite a significant fraction of the NiFe2O4 particles uncovered (the ratio of covered to uncovered particles is estimated as 1:12 based on the cobalt loading and the observed shell thickness). The number of cobalt atoms in the surface region relative to the total number of cobalt atoms in the shell can be estimated to be ca. 7.5% if it can be assumed that cobalt is only present in epitaxially grown shells (this estimate is based on an average number of Co atoms in the surface region of Co3O4 of 6.8 atoms/nm2[20]).

As we showed previously, the reduction of NiFe2O4@Co3O4 core–shell particles in hydrogen starts at temperatures as low as 150 °C, as evidenced by temperature-programmed reduction.[19] At these low temperatures, NiFe2O4 does not yet reduce (it requires temperatures higher than 250 °C to start the reduction of NiFe2O4 in hydrogen).[19] Hence, the amount of hydrogen consumed at low temperatures below 200 °C has been attributed to the reduction of some cobalt in the shell.[19] The reduction in hydrogen at 180 °C for 1 h resulted in a rather low degree of reduction of cobalt of only 21% (see Table ). This would imply that only cobalt in the first few layers near the surface was reduced. Reduction at 230 °C for 2 h resulted in an increase in the degree of reduction from 21% to 62% (see Table , calculated assuming the re-oxidation of reduced cobalt only).

Table 1

Characterization of Reduction for NiFe2O4@Co3O4 Core–Shell Nanoparticles (8.3 wt % Co) in Hydrogen (Space Velocity = 6000 mLn.g–1.h–1)

Treduction (°C)	treduction (h)	O2 uptakea (mmol gcat–1)	DORb (%)	H2 uptakec (μmol gcat–1)	SMed (m2 gcat–1)
180	1	0.20	21	21.9	1.74
230	2	0.57	62	10.3	0.85

Amount of oxygen taken up in back titration after reduction.

Degree of reduction (DOR) assuming only the reduction of Co3O4.

H2 uptake as determined by pulse chemisorption.

Active metal surface area based on H2 uptake, assuming only metallic cobalt with 14.6 Co atoms/nm2.

Interestingly, the NiFe2O4@Co3O4 core–shell nanoparticles reduced at 180 °C show a higher hydrogen uptake than the NiFe2O4@Co3O4 core–shell nanoparticles reduced at 230 °C (despite a lower degree of reduction). The hydrogen uptake of the sample reduced at 180 °C can be related to a cobalt dispersion defined as the amount of surface cobalt relative to the total amount of cobalt in the catalyst, assuming that only cobalt will be reduced under the applied activation conditions (vide infra). A cobalt dispersion of ca. 3.1% is determined, implying that not all cobalt on the surface is reduced and not all reduced cobalt is at the surface. The lower hydrogen uptake for the sample reduced at 230 °C is consistent with a reduced exposed cobalt surface area as a consequence of the previously observed formation of cobalt islands over the surface of the core after reduction.[19] The restructuring may find its origin in the strain produced in the shell upon reduction due to the associated decrease in volume.

The reduction of the NiFe2O4@Co3O4 core–shell nanoparticles was investigated by in situ TEM starting from 180 °C (see Figure ). Clearly discernible changes in the morphology of the region of interest were not observed after reduction at 180 °C for 2 h. The average shell thickness after reduction (2.7 nm ± 0.7 nm) was statistically unaltered from the starting material as expected if the surface region was only partially reduced. The EELS spectrum image showed that cobalt was present around the nickel ferrite core (Figure b), implying that the core–shell structure remained intact after reduction at 180 °C in H2 at 1 atm.

Figure 1

STEM-HAADF images and the corresponding composite element maps obtained from EELS spectrum images of NiFe2O4@Co3O4 core–shell nanoparticles obtained (a) before reduction and after the in situ TEM reduction at (b) 180 °C in H2 (1 atm) for 2 h and (c) 230 °C in H2 (1 atm) for 2 h. The spectrum image region of interest is indicated on the STEM-HAADF images by the green rectangles. In the RGB maps, the green channel is the signal of the cobalt-L edge, while purple is produced by combination of red (Fe-L edge) and blue (Ni-L edge).

In a previous study, the restructuring of the cobalt shell yielding cobalt islands with a size between 1 and 3 nm was observed upon reduction in hydrogen at 230 °C for 2 h.[19] During the current in situ experiment, the hydrogen treatment at 180 °C was followed by treatment in hydrogen at 230 °C. As shown in Figure c, the restructuring of the cobalt shell after 2 h reduction at 230 °C was not observed in all particles, with some areas showing no evidence of restructuring. This is in contrast with our earlier work, where all cobalt shells were observed to break up into small cobalt islands during the same reduction step, indicating some stress release in the shell during slow reduction at a lower reduction temperature reduction.

Representative summed EELS spectra from the cobalt shell region of core–shell nanoparticles were obtained before reduction, after reduction at 180 °C for 2 h, and after reduction at 230 °C for 2 h (the latter was preceded by 2.5 h reduction at 180 °C) (see Supplementary Information, Figure S.2). These EELS spectra show the expected signals attributed to oxygen and cobalt. The EELS spectrum of NiFe2O4@Co3O4 core–shell nanoparticles (reduced at 230 °C for 2 h) also shows the presence of the Fe-L2,3 edge, suggesting the presence of iron in the near-surface region. EELS elemental quantification yielded the Co-to-Fe ratio in the region as 2:1.

A close examination of the O–K edge (Table and Figure S.3 in the Supporting Information) showed detectable changes in the edge fine structure after each reduction treatment. Three characteristic peaks were present at around (i) 530 eV, (ii) 542 eV, and (iii) 549 eV prior to reduction. These peaks are similar to what is expected for cobalt oxides.[21] After reduction, the O–K edge showed the presence of a single peak at approximately 538 eV. A similar peak was obtained from background EELS spectra of the gas cell with only SiN windows in place (i.e., in the absence of a sample), implying some oxidation of the SiN window. Nonetheless, the absence of the pre-edge peak around 530 eV, which is typically present in cobalt oxides,[21−23] suggests the presence of metallic cobalt after treatment in hydrogen at temperatures as low as 180 °C.

Table 2

Peak Positions Extracted from EELS (See Figure S.2 in the Supporting Information)

	EELS edge (eV)					white line ratio
pre-treatment	O–K	Fe-L3	Fe-L2	Co-L3	Co-L2	Co-L3/Co-L2a
before reduction	530.3; 540.3; 542.0; 549.0			780.8	796.0	2.7
Treduction 180 °C (2 h)	537.0			780.3	795.5	3.1
Treduction 230 °C (2 h)b	538.0	709.5	722.5	780.8	795.0	2.5
SiN window	538.5

White line ratio is the peak intensity ratio of the L3/L2 peaks in the fine structure; Co = 3.1, Co3O4 = 2.7, CoO = 4.5 obtained from reference spectra reported by Zhao et al.[24]

Following the reduction at 180 °C.

The fine structure of the Co-L edge can be used to gain further insights into the oxidation state of cobalt at various stages of reduction. No significant shift was observed in the relative peak position of the Co-L3 or Co-L2 peak at different stages of reduction. The average oxidation state of cobalt in the nanoparticle’s shell was quantified using the integrated peak intensity ratio of the L3 to L2 peaks (white line ratio) present in the edge fine structure and comparing it to reference materials.[25] An average Co-L edge spectrum was generated by averaging spectra obtained from all spectrum image pixels present within Co shell regions. This was achieved by thresholding the spectrum image data for each site by Co signal intensity using a binary spectrum image filter. Calculation of the integrated peak intensity ratios was done using a double arctangent function for background subtraction (see Figure S.4 in the Supporting Information), as implemented in the digital micrograph script.[26] For the unreduced material, the relative ratio of Co-L3 to Co-L2 was 2.7, consistent with the presence of Co3O4. After reduction at 180 °C for 2 h, an increased ratio of 3.1 was found, consistent with the expected increase in Co-L3 upon reduction. Interestingly, the ratio determined after reduction at 230 °C was 2.5, in which the lower value would seemingly indicate cobalt to be in a more oxidized state. However, the presence of Fe within the Co shell region may modify the availability of 3d electron states within cobalt since the electronic configuration of iron is nearly identical. This might lead to a mutual sharing of the additional outer-shell electron in cobalt between the atoms and thus a shift in the Co-L3/Co-L2 ratio.

The reduction in hydrogen may affect not only the cobalt shell but also the NiFe2O4 core. Hence, each activation regime was also studied by in situ XRD (see Table ). The XRD pattern recorded after reduction at 180 °C only showed the presence of NiFe2O4. Reflections attributable to metallic cobalt, Co(II)O, or Co3O4 were absent, but this is to be expected as the shell thickness is only ca. 2.5 nm. It can be further concluded that the NiFe2O4 shell is stable in hydrogen at 180 °C for 1 h. Increasing the reduction temperature to 230 °C resulted in some reduction of NiFe2O4 to form possibly an iron–nickel alloy. Some reduction of bare NiFe2O4 at 230 °C has been reported before.[19] The average crystallite domain size of the NiFe2O4 core increases upon reduction at 230 °C, which may indicate a preferential reduction of some NiFe2O4 in smaller cores/particles. The obtained size of the crystalline domain of the formed iron–nickel alloy is in line with the expected contraction upon removing oxygen from the lattice.

Table 3

Relative Phase Abundance and Average Crystallite Size of the Various Phases Present in NiFe2O4@Co3O4 after the Activation in H2 and Exposure to Fischer–Tropsch Conditions (Treaction = 230 °C, p = 16 bar) for 16 h Determined from the Rietveld Refinement of In Situ X-ray Diffraction Patterns

	H2 activation		after FT synthesis
phase	180 °C, 1 h	230 °C, 2 h	180 °C, 1 h	230 °C, 2 h
relative phase abundance (mass %)a
NiFe2O4	100 (0.0)	90.6 (0.3)	72.4 (0.5)	71.3 (1.5)
Ni			8.2 (0.4)
Fe2C			19.5 (0.5)	11.6 (1.6)
Fe0.8Ni0.2				13.6 (0.9)
Fe0.625Ni0.375		9.4 (0.3)		3.6 (0.6)
average crystallite size (nm)a,b
NiFe2O4	13.6 (0.1)	15.6 (0.3)	16.9 (0.2)	19.9 (0.7)
Ni			3.2 (0.2)
Fe2C			4.5 (0.2)	5.8 (1.0)
Fe0.8Ni0.2				8.1 (0.8)
Fe0.625Ni0.375		10.0 (0.5)		3.6 (0.6)

Uncertainty given in parentheses.

Volume-weighted average crystallite size determined from the integral breadth.

Exposing the NiFe2O4@Co3O4 core–shell nanoparticles reduced at 180 °C to Fischer–Tropsch conditions (230 °C, 16 bar, H2/CO = 2, 16 h) resulted in some reduction of the NiFe2O4 core with the formation of metallic nickel and iron carbide. This is accompanied by an increase in the crystalline domain size for NiFe2O4. This was also seen in a statistically significant shift in the crystallite size distribution toward larger crystallites from STEM-HAADF images (see the Supporting Information, Figure S.5/Table S.2). The crystallite size distribution of NiFe2O4@Co3O4 core–shell nanoparticles after activation in H2 at 180 °C (1 h) and exposure to Fischer–Tropsch synthesis conditions showed, in particular, a lower frequency of crystallites with sizes below 5 nm than in the fresh material. The increase in the average crystallite size could be ascribed to sintering of the nickel ferrite nanoparticles, but it is more likely that the smaller ferrite nanoparticles present in fresh NiFe2O4@Co3O4 core–shell nanoparticles are preferentially reduced at the low temperatures applied here.

After the Fischer–Tropsch synthesis, some iron carbide is present in the sample with some evidence also for the formation of iron–nickel alloy phases (although it should be stated that the assignment to the different alloys is tentative, seeing the broadness and low intensity of these diffraction lines). Interestingly, the amount of NiFe2O4 in the sample after the Fischer–Tropsch synthesis as determined by XRD is rather similar for the sample reduced at 180 °C and the sample reduced at 230 °C. It should be noted that the phases detected by XRD analysis present in the sample reduced at 230 °C for 2 h after the Fischer–Tropsch synthesis show a deficiency for nickel, implying the presence of some XRD-amorphous nickel.

The reduction of NiFe2O4@Co3O4 at 180 °C followed by exposure to Fischer–Tropsch conditions at 230 °C and 16 bar for 16 h results in cobalt over the ferrite surface, as evidenced by the EELS spectrum image elemental maps (see Figure ). The core also appeared to have remained intact, which would indicate that the core–shell morphology was retained.

Figure 2

STEM-HAADF image (white square: region from which the EELS spectrum image was generated) and elemental maps for Co (green), Fe (red), Ni (blue), and C (white) obtained from the EELS spectrum images of used NiFe2O4@Co3O4 core–shell nanoparticles after reduction and exposure to the Fischer–Tropsch synthesis conditions (Treaction = 230 °C, p = 16 bar) in an in situ XRD reactor.

In contrast, after reduction at 230 °C for 2 h and exposure to the Fischer–Tropsch synthesis conditions in an in situ XRD reactor, the STEM-HAADF images of NiFe2O4@Co3O4 nanoparticles showed subtle differences (see Figure ). Clusters of cobalt can be seen between the nanoparticles (indicated by a white arrow), suggesting that cobalt segregated and that the core–shell structure has been degraded. This STEM-HAADF image also shows the possible onset of the formation of a hollow Ni–Fe oxide nanoparticle (indicated by an orange arrow in the STEM-HAADF image), as evidenced from the presence of an oval core seemingly detached from the larger nanoparticle. The elemental maps suggest that this oval core is composed of both Fe and Ni, while the outer ring of the larger nanoparticles appears to contain predominantly iron. The observed effect may be explained by the nanoscale Kirkendall effect with iron diffusing faster than nickel in the reducing environment of the Fischer–Tropsch synthesis, generating additional porosity.[27] Furthermore, some carbon can be seen. It is tempting to ascribe this to the formation of iron carbide (as seen in the in situ XRD measurement, see Table ); however, there is currently not enough evidence to conclude this.

Figure shows another area studied (after reduction at 230 °C for 2 h followed by Fischer–Tropsch synthesis at 230 °C and 16 bar for 16 h in the in situ XRD cell) that exemplifies the clustering of Co over a larger field of view. Furthermore, there are signs of areas enriched with either iron (white circle in Figure b) or nickel (white square in Figure b), indicating segregation of these elements. All the metals present in this system are expected to have some mobility[28] under conditions of the Fischer–Tropsch synthesis since they may form (sub)-carbonyls under these conditions.[29] It can be further seen that carbon overlayers have been formed around the ferrite particles during the exposure to Fischer–Tropsch conditions.

Figure 3

(a) STEM-HAADF image with a white rectangle showing the region from where the EELS spectrum image was generated and (b) corresponding composite elemental map obtained from EELS spectrum image of another region of interest for the used NiFe2O4@Co3O4 core–shell nanoparticles after reduction at 230 °C (2 h) and exposure to the Fischer–Tropsch synthesis conditions (Treaction = 230 °C, p = 16 bar) in an in situ XRD reactor. Note: Co (green), Fe (red), Ni (blue), and Fe–Ni composite (pink).

Table shows the catalytic activity of NiFe2O4@Co3O4 core–shell nanoparticles as determined in a fixed bed reactor operating at 230 °C and 20 bar and a CO conversion of 3 ± 1%. The activity of the core–shell nanoparticles reduced at different temperatures is compared with that of a bulk Co3O4 reduced at 300 °C (see the Supporting Information for the synthesis procedure). The observed activity may be ascribed to the activity of metallic cobalt, iron carbide, nickel, or iron–nickel alloys. It was previously shown that the NiFe2O4 core contributes to the observed activity when reduction was performed at 230 °C.[19] The chosen reaction conditions (230 °C and 20 bar) are well-known conditions for CO hydrogenation over cobalt and it is further well known that cobalt is substantially more active than either nickel or iron when minimally affected by support effects.[30] However, the activity of small cobalt crystallites (<6 nm) is severely reduced (also resulting in a much lighter product being formed).[31,32] Furthermore, nickel may act as a methanation catalyst under these conditions.[30,33] Hence, the evaluation of the activity attributed to cobalt in the core–shell structure is not trivial.

Table 4

Activity and Selectivity in the Fischer–Tropsch Synthesis over Reduced Materials (Treaction = 230 °C, p = 20 bar at the Specified Syngas Space Velocity)

			activity				fraction in organic product (C-%)
catalyst	activation	syngas SVa mLn.gcat–1.h–1	rFTb μmol.gcat–1 s–1	rFTc μmol.gCo–1 s–1	TOFd (s–1)	SCO2e (%)	C1	C2–C4	C5+
NiFe2O4@Co3O4	180 °C, 1 h	6149	0.70	8.41	<0.02	4	34	29	37
NiFe2O4@Co3O4	230 °C, 2 h	7755	1.22	14.7	f	4	29	24	47
Co3O4	300 °C, 1 h	6350	4.76	6.43	0.18	0.5	9	10	81

SV refers to space velocity.

rFT is the integral rate or cobalt-time yield per gram of cobalt.

rFT is the integral rate per gram of catalyst.

TOF is the turnover frequency or site-specific activity based on H2 chemisorption: molCO.(2 molH2 adsorbed).s–1.

S refers to selectivity.

See text.

The integral rate (cobalt-time yield) obtained with the NiFe2O4@Co3O4 core–shell nanoparticles after activation at 180 °C (1 h) was 8.40 μmolCO.gCo–1.s–1 if all activity can be ascribed to the activity of cobalt, with a C5+ selectivity of 37 C-% (see Table ). The obtained cobalt-time yield is slightly higher than the cobalt-time yield over unsupported Co3O4 (after reduction at 300 °C for 1 h) with a similar crystallite diameter (ca. 14 nm) in its oxidized form. Of course, the activity per gram of catalyst for unsupported Co3O4 was much higher than that for NiFe2O4@Co3O4 core–shell nanoparticles (reduced at 180 °C for 1 h) due to the low cobalt content.

A better comparison is obtained by comparing the site-specific activity (TOF). The turnover frequency over the NiFe2O4@Co3O4 core–shell nanoparticles after activation at 180 °C (1 h) was determined to be 0.02 s–1 if the observed activity is attributed solely to the activity of cobalt. The site-specific activity was determined based on the H2 uptake after reduction, assuming that an adsorption stoichiometry of H2/active site of 1:2. Further reduction of the catalyst may occur during the Fischer–Tropsch synthesis, although the cobalt shell surrounding the NiFe2O4 core remained intact during the Fischer–Tropsch synthesis when the core–shell nanoparticles were reduced at 180 °C for 1 h (see Figure ). Some of the NiFe2O4 is transformed during the Fischer–Tropsch synthesis, yielding metallic nickel and iron carbide. Hence, the site-specific activity (0.02 s–1) obtained over NiFe2O4@Co3O4 reduced at 180 °C for 1 h represents an upper limit of the activity of cobalt in a shell as some activity may originate from other catalytically active materials formed during the Fischer–Tropsch synthesis. The observed turnover frequency is much lower than the turnover frequency observed over the unsupported Co3O4 (0.18 s–1 after reduction at 300 °C for 1 h). The low value for the upper limit of the TOF for cobalt in NiFe2O4@Co3O4 might be explained by a structural aspect of metallic cobalt embedded in a partially reduced shell: the lattice mismatch between the reduced cobalt metal shell, the metal oxide core, and the unreduced shell may result in the strain in metallic cobalt[34] by shrinkage of the reduced cobalt patch in the shell. The tensile strain in the metal surface results in a narrowing of the d-band while shifting the center of the d-band closer to the Fermi level.[35,36] Consequently, adsorbates, such as CO, C, O, and other reaction intermediates, are expected to bind more strongly to cobalt in the shell of NiFe2O4@Co3O4 core–shell nanoparticles after activation at 180 °C (1 h) than on extended planes. The strong adsorption of reaction intermediates could inhibit CO dissociation and O removal[37] from the surface, resulting in a lower activity.[38]

The selectivity obtained over NiFe2O4@Co3O4 core–shell nanoparticles reduced at 180 °C for 1 h differs from the selectivity obtained over unsupported Co3O4 after reduction at 300 °C for 1 h. A methane selectivity of 34 C-% was obtained over the core–shell nanoparticles reduced at 180 °C with a C5+ selectivity of only 38 C-%. The unsupported Co3O4 showed more the expected selectivity for the Fischer–Tropsch synthesis in a fixed bed reactor under the applied reaction conditions (9 C-% toward C1 and 81 C-% toward C5+). The observed shift in the product selectivity toward lighter products obtained with the core–shell nanoparticles reduced at 180 °C is consistent with the postulated change in the adsorption strength of the reactants and reaction intermediates on the shell. Stronger adsorption of, in particular, CO will reduce the likelihood for chain growth,[39] resulting in more methane and less C5+. It should be realized that the conclusion here remains tentative as the selectivity is determined not only by the presence of cobalt in the sample but also by the presence and activity of both iron carbide and nickel in the sample during the Fischer–Tropsch synthesis.

Comparing the activity obtained with the NiFe2O4@Co3O4 core–shell nanoparticles after activation at 180 °C (1 h) with the activity of the same material activated at 230 °C for 2 h shows that the increase in the reduction temperature increased the mass-specific activity. The cobalt-time yield for the sample reduced at 230 °C is higher than that for the sample reduced at 180 °C if the activity can be ascribed to the presence of metallic cobalt in the sample. It is difficult to even indicate an upper or lower limit for the site-specific activity for cobalt in this material. A turnover frequency (TOF) of 0.06 s–1 was estimated if cobalt was the only catalytically active material in the sample and if all hydrogen uptake takes place on reduced cobalt only. However, the sample reduced at 230 °C does contain not only reduced cobalt after reduction but also an iron–nickel alloy (Fe0.625Ni0.375), which may also adsorb hydrogen. Hence, the H2 uptake yields an upper limit on the number of surface cobalt atoms. The dispersion of cobalt will thus be lower, which will lead to an increased cobalt site-specific activity. On the other hand, this material also contains iron carbide after the Fischer–Tropsch synthesis and iron–nickel alloy synthesis (as evidenced by performing the Fischer–Tropsch synthesis in the in situ XRD cell). The surfaces of these phases may also contribute to the overall activity, thus reducing the site-specific activity of cobalt. Hence, the site-specific activity of the sample reduced at 230 °C cannot be indicated.

The selectivity obtained over the NiFe2O4@Co3O4 core–shell nanoparticles after activation at 230 °C for 2 h is quite similar to the selectivity obtained over the core–shell nanoparticles reduced at 180 °C for 1 h. The methane selectivity is slightly reduced to ca. 29 C-% with an increase in C5+ selectivity to 47 C-%. A relatively high methane selectivity in comparison to the selectivity obtained over bulk cobalt was expected for a sample containing small cobalt islands. The selectivity toward methane over small cobalt islands is expected to be higher, which has been attributed to an increase in the strength of adsorption of the reaction intermediates.[38]

Conclusions

The shell in NiFe2O4@Co3O4 core–shell nanoparticles can be retained by reducing at a relatively low temperature of 180 °C for 1 h. The resulting partially reduced shell remains intact after the Fischer–Tropsch synthesis performed at 230 °C and 20 bar. Reduction at a slightly higher temperature results in the partial destruction of the shell. Further transformation of NiFe2O4 takes place during the Fischer–Tropsch synthesis.

The resulting site-specific activity of NiFe2O4@Co3O4 in the Fischer–Tropsch synthesis after reduction at 180 °C for 1 h is rather low in comparison to unsupported cobalt and is even lower if some activity is attributed to other components in the catalyst, originating from reduction/carburization of the core. The low activity might be ascribed to the presence of tensile strain in the metal surface in the shell resulting in strong adsorption of the reactants and reaction intermediates. The reduced selectivity toward the desired C5+ products obtained over the core–shell NiFe2O4@Co is in accordance with the presence of strongly adsorbing CO induced by the tensile strain.

Experimental Section

A detailed description of the synthesis of the core–shell nanoparticles is given elsewhere.[18] Briefly, a two-step synthesis was followed. Initially, an aqueous solution (10 mL) of ammonium carbonate, (NH4)2CO3, (8.5 M) was mixed with a 25 wt % ammonia aqueous solution (3.6 mL). Cobalt carbonate (CoCO3) was added to obtain a cobalt concentration of 0.05 M at a pH of 11.5. The resulting cobalt-containing suspension was then stirred at 45 °C for 2 h. After this time, the solution was filtered by gravity. This filtrate was added to a round-bottom flask, and ∼0.05 g/mL NiFe2O4 was added; the suspension was stirred at 85 °C for 6 h. The temperature was then increased to 95 °C and kept there for another 24 h under stirring. The solvent was subsequently removed under vacuum, and the dried solid was rinsed with deionized water and again dried under vacuum. The solid obtained was oven-dried at 110 °C for 12 h. The thickness of the cobalt step layer was increased by repeating the synthesis using the dried solid instead of pure NiFe2O4.

The change in the phase composition in the NiFe2O4@Co3O4 core–shell nanoparticles upon reduction and subsequent Fischer–Tropsch synthesis was monitored by in situ X-ray diffraction (Panalytical X’Pert Pro multi-purpose diffractometer with an Anton Paar XRK900 chamber). In situ reduction was performed under a hydrogen flow (50 mL/min) with the sample being heated from room temperature to 180 °C or 230 °C using a heating rate of 5 °C/min. After the reduction, the reactor chamber was flushed with helium for 2 h and cooled to 40 °C. Subsequently, the catalyst was exposed to synthesis gas (H2/CO = 2, 50 mL/min) and the temperature was raised to 230 °C. At this temperature, the pressure was increased to 16 bar over a 5 h period. The reaction was maintained under these conditions for 16 h. At the end of the experiment, the system was de-pressurized, flushed with helium, and cooled to 40 °C. The catalyst was subsequently passivated in a mixture of helium (50 mL/min) and oxygen for 3 h (O2 flow rate initially 2 mL/min and gradually increased to 6 mL/min). The catalyst was unloaded into dry ice and referred to as the used catalyst.

The shell thickness and crystallite size in the NiFe2O4@Co3O4 core–shell nanoparticles were imaged by transmission electron microscopy (TEM; JEOL ARM 200F operating at 200 kV). A small amount of crushed sample dispersed in ethanol was transferred onto a holey carbon copper TEM grid (SPI Supplies, 300 mesh). High-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) as well as bright-field STEM imaging at atomic resolution was used. Electron energy loss spectroscopy (EELS) spectrum imaging was performed using the DualEELS mode on a Gatan GIF Quantum ERS spectrometer. A convergence semi-angle of 21.4 mrad was used for the STEM probe, and the collection semi-angle of the spectrometer was 54.3 mrad. Electron probe current densities used ranged from 78 to 200 pA with a probe size typically less than 0.1 nm.

The change in the morphology of the NiFe2O4@Co3O4 core–shell nanoparticles due to exposure to reduction conditions was imaged by in situ TEM (FEI Titan G2 80 – 200 (S)TEM ChemiSTEM instrument operating at 200 kV; University of Manchester). The instrument was equipped with the Gatan GIF Quantum ERS spectrometer. STEM-HAADF images were recorded using FEI TIA software, while EELS spectrum images were recorded using Gatan Digital Micrograph software. The gaseous e-Cell system used in the FEI Titan instrument was a Protochips Atmosphere holder. During the measurements, the e-Cell was filled completely with pure hydrogen at a nominal pressure of 1 bar. The temperature was controlled by on-chip pre-calibrated heater elements. The samples were crushed in ethanol, and then a drop was placed onto the atmosphere side of the electron transparent plasma cleaned SiNx windows of the e-Cell and allowed to dry in a clean ambient temperature Petri dish. The prepared e-Cell was sealed, and specimens were stabilized at room temperature. Thereafter, the sample was heated to the desired temperature and held there for 1 h at a time before imaging. Images of selected regions of interest were acquired before and after thermal treatments to ensure that electron beam damage did not influence the results. Furthermore, different areas in the sample that were not analyzed previously were also imaged after heating to ensure that the results presented herein are representative and reproducible.

The metal dispersion and the active metal surface area were determined by temperature-programmed desorption of hydrogen, which was performed on an AutoChem 2920 (Micrometrics, USA) equipped with a thermal conductivity detector (TCD). Before each measurement, the samples were dried at 120 °C for 1 h and subsequently reduced in pure hydrogen using the activation conditions listed in Table (heating rate of 5 °C/min). Thereafter, the samples were flushed under an argon flow (50 mL/min) for 1 h. H2 was adsorbed at 100 °C by pulse chemisorption in 10 pulses at a 2 min interval.

The degree of reduction was determined by oxygen (O2) back titration on a Micromeritics ASAP 2020 Unit (Micromeritics, USA). The sample (ca. 200 mg) was degassed at 120 °C for 2 h. After cooling to ambient temperature, the sample was reduced in pure H2 (50 mL/min) under the activation condition using a heating rate of 5 °C/min. Subsequently, the sample was cooled down to 200 °C and evacuated for 4 h. Then, the temperature was increased to 230 °C and O2 titration was carried out.

Catalytic activity was evaluated in a fixed bed reactor setup containing four independent parallel reactor tubes, each with 6.5 mm inner diameter. The catalyst (200 mg) was diluted with 3.1 g of SiC (320 grit), resulting in a catalyst bed length of ∼6 cm. The catalyst bed temperature was accurately controlled using a thermocouple in the middle of the bed. The catalyst was activated in situ in hydrogen prior to the Fischer–Tropsch synthesis. After reduction, the temperature was changed to 200 °C and the pressure was increased to 20 bar using hydrogen and argon. Synthesis gas (H2/CO = 2) was then introduced, and the temperature was increased to 230 °C (ramp rate, 0.5 °C/min). The space velocity during the reaction was varied to achieve a CO conversion of 3 ± 1%. The reaction effluent was passed through a hot trap (190 °C) to collect the high boiling waxes and a cold trap (15 °C) to collect water and low boiling organic product compounds. The outlet gases were analyzed on-line using an Agilent 7890A refinery gas analyzer equipped with one flame ionization detector and two TCD channels. Argon (∼8 vol %) was added to the feed as an internal standard to determine conversion and product selectivity. The selectivity values are reported in C-% from syngas converted to hydrocarbons (i.e., excluding CO conversion to CO2).