1 Introduction

Fischer–Tropsch synthesis (FTS) is an industrially proven process for the conversion of synthesis gas (CO + H2) into hydrocarbons, which is a key part of so-called XTL technologies (gas-to-liquid, coal-to-liquid, biomass-to liquid) [1,2,3]. As compared to oil-derived fractions, synthetic hydrocarbons contain little or no aromatics and extremely low level of sulfur and nitrogen contaminants. That is why FTS is considered as a promising source of clean motor fuel fractions, base oils and value-added chemicals including linear alpha-olefins, higher alcohols and ultra-pure waxes [4]. Cobalt and iron-based catalysts are the only those which ever used commercially. Ruthenium is known to be even more active in CO hydrogenation but it is too expensive and rare metal and therefore inappropriate for industrial implementation [5, 6]. All modern FTS facilities operate in fairly mild temperature regime, 200–240 °C, which is referred as low temperature Fischer–Tropsch (LTFT). Syncrude of LTFT is enriched by linear alkanes with broad molecular weight distribution and serves as a feed for subsequent mild hydrocracking into premium kerosene and diesel fractions [4]. Supported cobalt is especially suitable catalyst for LTFT due to its high intrinsic activity, high selectivity to target long chain hydrocarbons, low activity in water gas shift reaction and pretty good resistance to oxidation by product water [7].

Usually synthetic oxide materials with high surface area such as γ–Al2O3, silica, amorphous alumina-silica and zeolites are often used as supports for FTS cobalt catalysts. They are cheap, available in large scale and usually provide high dispersion of supported metal and therefore high specific activity [3, 5, 7]. However, known drawbacks of these supports are their low thermal conductivity (which may be critical because CO hydrogenation is highly exothermic process and demands fast heat removal from catalyst grains) and their prone to react with supported metal thus forming mixed oxides. The last ones are almost irreducible at standard activation temperatures (350–450 °C) and therefore lead to actual metal loss during catalyst preparation [8,9,10,11].

Recently, porous carbon materials have attracted much attention as supports for CO hydrogenation catalysts. Among them activated carbon, carbon nanotubes, carbon nanofibers, carbon spheres, graphene sheets, MOF-derived carbonaceous materials have been employed in FTS. Their advantages are chemical inertness to supported metal, thermal stability, tunable texture and possible chemical modification of surface in oxidative treatment [12,13,14,15,16,17,18,19]. Carbon-supported FTS catalysts demonstrated high activity and selectivity to C5+ hydrocarbons due to high dispersion and reducibility of metal species. As reported, possible electronic transfer between carbon and metal species could increase catalytic activity in CO hydrogenation [14].

Besides pure carbonaceous materials, carbon-coated inorganic oxides were also proposed as supports for FTS catalysts. They combine the advantages of both carbon and oxide supports such as low metal-support interaction and good mechanical strength. Carbon coating can be obtained for example via impregnation of oxide material with carbohydrate water solution followed by thermal decomposition in inert atmosphere [20]. It has been found that carbon-coating silica stabilized highly dispersed cobalt oxide with partial cobalt reduction during thermal treatment in inert atmosphere. The size of cobalt oxide nanoparticles decreased from 10–15 nm over the conventional silica supported catalyst prepared without carbon pre-coating to 5–6 nm over samples with high carbon content. The maximum FTS activity was almost twice higher compared to the reference uncoated catalysts. Higher selectivity to long chain hydrocarbons and low methane selectivity were also observed [20]. A single-step hydrothermal approach was also developed for supported iron catalyst preparation. Fructose dehydration/condensation over dissolved iron nitrate provides polymeric carbon on the surface of silica support. The enhanced catalytic performance in FTS compared to conventional silica-supported catalysts was observed [21]. Recently, we reported that support carbonization with glucose resulted in drastic changes in the transformations of supported iron phases during catalyst activation and FTS [22, 23].

In this work, magnesium aluminum spinel MgAl2O4 was used as a support for the preparation of series of Co-based, carbon-coated CO hydrogenation catalysts. We focus on the effect of support carbon coating on the formation of active cobalt phase and catalytic performance in CO hydrogenation. Carbonization of the support was performed by impregnation of the support with glucose water solution followed by drying and thermal treatment of the material in inert atmosphere.

2 Experimental

The catalysts were prepared by one or two-step incipient wetness impregnation method (IWI) of preliminarily carbonized magnesium aluminum spinel MgAl2O4 (Sasol Germany GmbH, PURALOX MG 30 Spinel, reported surface area of 23 m2/g, average particle size of 1.8 μm). Glucose monohydrate C6H12O6·H2O and cobalt (II) nitrate hexahydrate Co(NO3)2∙6H2O and aqueous solutions were applied for deposition of surface carbon, and active phase, correspondingly. The reference sample containing no carbon was prepared by impregnation of pristine support with cobalt nitrate solution. The material was dried under reduced pressure at 80 °C followed by calcining in nitrogen flow, while temperature was raised by 10 °C/min to 450 °C and then maintained for 2 h. This catalyst is denoted below as Co/MA. To prepare carbon-coated support, the spinel samples were impregnated by aqueous glucose solutions of pre-calculated concentration to provide 4, 20 and 40% carbon by weight of the support. The materials were dried under reduced pressure at 80 °C followed by calcining in nitrogen flow as pointed above. The resulting materials were used for cobalt deposition by IWI and subsequent drying and calcination procedures as described above. The resulted catalysts are denoted Co/xC/MA where x = 4, 20 and 40. Cobalt nominal loading was 10 wt.% for all the catalysts. Actual cobalt loading in the catalysts was determined by ICP-MS using ICPE-9800 spectrometer.

Textural properties of the catalysts were measured by low temperature nitrogen physisorption on a Micrometrics ASAP 2010 N gas sorption analyzer. Immediately before the measurements, 200–300 mg sample was degassed in vacuum (P < 10 mm Hg) at 300° C for 4 h. The nitrogen adsorption/desorption isotherms were obtained at − 196 °C by measuring with time interval of 5 s. The surface area was calculated from the Brunauer–Emmett–Teller (BET) equation. The total pore volume was determined at P/P0 = 0.995.

Scanning electron microscopy (SEM) was performed using a JEOL JSM–6000 NeoScope scanning electron microscope with an integrated EX-230 X-ray analyzer equipped with a backscattered electron detector and tungsten cathode. The accelerating voltage was 15 kV. The image was obtained in low vacuum mode.

Thermoanalytical study was carried out in flowing inert medium using a Netzsch STA 449 C Jupiter thermal analyzer. Argon flow rate was of 80 mL/min, heating rate was of 10 °C/min, temperature range was 40–600 °C. The gases released during thermal decomposition were analyzed using a Netzsch QMS 403C quadrupole mass spectrometer. The scanning range was from 10 to 50 m/z and the ion currents of NO, CO, NO2 and CO2 were reordered.

X-ray quantitative phase analysis of the studied samples was performed on a PANalytical Empyrean diffractometer. The immobile sample in the form of a powder layer of 1 mm thick and about 20 mm in diameter, on a non-reflective holder, was placed in the center of the vertical plane of the goniometer. The experiment was carried out in “θ–θ” geometry with synchronous movement of the PIXcel-3D detector and X-ray ceramic tube along the goniometric circle. Diffraction patterns were measured on Cu Kα-radiation in the scanning range for the angle 2 θ of 5–100 degrees. Mathematical processing of diffraction patterns, phase analysis of the samples, and particle sizes estimation by the Scherrer method were carried out using the High Score Plus program and the JSCD file cabinet. The average Co3O4 particle size in nm was converted to the corresponding Co0 particle size according to the relative molar volumes of metallic cobalt and Co3O4 [24]:

$${d}_{Co}=0.75\times {d}_{Co3O4}$$

Dispersion of metallic cobalt was calculated assuming spherical, uniform cobalt metal particles with atom surface density of 14.6 nm–2, giving the formula [24]:

$$D=\frac{96}{{d}_{Co}}\times 100\%$$

The magnetic measurements combined with isothermal catalyst reduction were performed in the vibrating sample magnetometer described elsewhere [25]. The catalyst sample (20 mg) was placed between two membranes of porous silica in a measurement cell of 0.3 cm3 in volume that also served as a plug-flow microreactor. The sample was heated in passing argon with a rate of 10 °C/min from ambient temperature to 450 °C and then the gas flow was switched to H2 (gas flow rate of 30 mL/min) keeping the temperature constant. The magnetization was continuously recorded at 3 kOe.

Catalytic performance in CO hydrogenation was assessed in a down flow fixed-bed reactor (stainless steel, 16 mm ID). Catalyst sample (400 mg of the catalyst diluted with 2 cc of quartz sand for better dissipation of the reaction heat) was placed in isothermal zone of the reactor. The sample was activated in situ by passing hydrogen through the reactor at 450 °C with a gas flow rate of 2 nL h–1 for 4 h. Then the reactor was cooled to 180 °C and CO hydrogenation was performed under 20 bar pressure with synthesis gas containing 27 vol% CO, 64 vol% H2 and 9 vol% Ar as an internal standard for analysis. The gas flow rate was maintained at 5 NL.gcat–1.h–1 and the temperature was gradually increased from 180 to 210 °C. After 48 h on stream at 210 °C the effluent gas was analyzed for CO, Ar, CH4, C2–C4 hydrocarbons and CO2 using Crystallux 4000 M gas chromatograph equipped with thermal conductivity detector and two 1.5 m × 2.1 mm columns packed with NaX molecular sieves (for Ar, CH4 and CO) and HayeSep phase (C2, C3, C4 and CO2), correspondingly. Synthesized heavy hydrocarbons were collected in heated trap (90 °C) and cold trap (ambient temperature, ~ 20 °C). The contents of the traps were combined and analyzed with the same instrument using 100 m × 0.25 mm capillary column with ZB-DHA-PONA phase and flame ionization detector. Helium is a carrier gas for the analysis.

Carbon monoxide conversion and selectivity to each gaseous product were calculated as follows:

$${X}_{CO}=\frac{moles\, CO\, converted}{moles\, of\, CO\, passed}\times 100\%$$
$$S=\frac{moles\, CO\, converted\, to\, the\, product}{moles\, of\, converted\, CO}\times 100\%$$

The C5+ selectivity was calculated as a difference between the total mass balance and the amount of C1–C4 and CO2 gases formed. The catalyst specific activity was evaluated as a cobalt-time yield (moles CO reacted per mole of cobalt loaded per second).

Cobalt-time yield was calculated as

$$Cobalt-time\, yield=\frac{W\times p\times {X}_{CO}\times 58.9}{b\times 22.4\times 3600}$$

where W is the flow rate of synthesis gas, p is a part of CO in synthesis gas, b is cobalt weight percentage in the catalyst measured by ICP-MS.

Turnover frequency (TOF) or site-time yield was calculated in proposition that each Co atom on the surface of metallic cobalt particle is conditional active site, giving the formula

$$TOF=\frac{Cobalt-time\, yield}{D}$$

3 Results and Discussion

The preparation process of the catalysts is schematically represented in Fig. 1. The reference sample Co/MA was prepared by impregnation of magnesium aluminum spinel with an aqueous cobalt nitrate solution followed by drying and calcining in inert atmosphere. To prepare carbonized supports, the spinel was impregnated by aqueous glucose solutions followed by thermal treatment in nitrogen flow. The prepared carbonized materials were further impregnated with aqueous solutions of cobalt nitrate followed by drying and calcining as pointed above. These samples are denoted as Co/xC/MA where x is the weight percentage of carbon per support (Fig. 1).

Fig. 1
figure 1

Schematic representation of the catalysts preparation procedure

Full size image

Nitrogen adsorption–desorption analysis was performed to evaluate the textural properties of the catalysts. The isotherm of Co/MA shows type II behavior, which reflects non-porous structure of the material. Deposition of 4 wt% carbon practically does not change the look of the isotherm curve of Co/4C/MA comparatively to that of reference Co/MA. However, increase in carbon loading leads to proportional increase in specific adsorption volume and changes isotherm type to IV, thus pointing on mesoporous texture of Co/20C/MA and Co/40C/MA. The hysteresis loop is the widest for the last one, indicating the most developed mesoporous structure of Co/40C/MA (Fig. 2).

Fig. 2
figure 2

Nitrogen adsorption–desorption isotherms of the catalysts

Full size image

Table 1 lists actual cobalt percentage in the catalysts determined by ICP-MS as well as some textural parameters derived from nitrogen adsorption–desorption measurements. The Brunauer–Emmett–Teller (BET) surface area, micropore area and pore volume increase with increasing carbon loading for Co/xC/MA series. Obviously, well-developed porous structure is formed in the course of thermal decomposition of glucose on the surface of spinel granules.

Table 1 Cobalt percentage in the catalysts and some textural parameters of the catalysts
Full size table

The scanning electron microscopy (SEM) micrographs of the catalysts are shown in Fig. 3. Co/MA displays grain-like surface structures while support carbonization leads to appearance of flakes and plate-like surface species obviously attributed to surface carbon. Also columnar structures appear in Co/20C/MA and their amount and size increase up to 2 × 0.5 μm in Co/40C/MA.

Fig. 3
figure 3

SEM images of Co/MA (a), Co/4C/MA (b), Co/20C/MA (c), Co/40C/MA (d)

Full size image

The processes taking place during the final step of catalyst preparation (calcination) were monitored using thermal analyzer connected with mass spectrometer. For this, the final calcination was omitted and the air-dried samples were heated in argon flow to 600 °C with a rate of 10 °C/min. Ion currents of CO+ (m/z = 28), NO+ (m/z = 30), CO2+ (m/z = 44) and NO2+ (m/z = 46) ions were detected (Fig. 4). Clear peaks of NO and NO2 with a maximum at 216° appear during linear heating of air-dried Co/MA precursor. Obviously, they are formed in the process of cobalt nitrate thermal decomposition [26, 27]. Note, the peak of m/z = 30 is much higher than other one because NO+ is the most intense fragment ion of both NO and NO2. Further heating of the sample is accompanied with slowly fading ion current up to 550 °C. Indeed, complete decomposition of supported cobalt nitrate reportedly requires a temperature above 450 °C [28].

Fig. 4
figure 4

Ion currents of small molecules released during thermal decomposition of air-dried catalyst precursors

Full size image

Carbonization of the support leads to two more ion currents appear, CO+ and CO2+. For Co/4C/MA air-dried precursor, peaks of NO+ and NO2+ are shifted by 20–25° to lower temperature while gas evolution begins already at 100 °C. Also, CO+ and CO2+ ion currents obviously indicate oxidation of surface carbon by nitrogen oxides released. Indeed, intensity of NO+ peak for Co/4C/MA is approximately a half of that for Co/MA. Notably, all four peaks have a maximum at the same temperature pointing on concerted processes of cobalt nitrate decomposition and surface carbon oxidation. Supposedly, physisorption of cobalt nitrate on carbon-coated spinel surface prevents strong metal-support interaction which is characteristic for oxide-supported cobalt catalysts [13, 14, 29]. This is why cobalt nitrate decomposition and gas evolution starts at lower temperatures for carbonized samples.

Increasing amount of surface carbon makes ion currents patterns more complex. Gas evolution for Co/20C/MA air-dried precursor proceeds in wide temperature range. Multiple overlapping peaks of NO+, CO+ and CO2+ ion currents represent this process. Notably, NO2+ ion current is very weak, likely due to increasing amount of reducing agent, carbon, in this sample. For the most carbon loaded Co/40C/MA, gas evolution pattern changes again and relatively narrow and intense peak of NO+ appears with maximum at 254 °C (Fig. 4). Supposedly, well-developed porous structure of Co/20C/MA and Co/40C/MA which includes both micro- and mesopores leads to diffusion restrictions during gas release, thus making ion currents profiles rather complex.

X-ray diffraction patterns of pristine MgAl2O4 spinel and support catalysts are shown in Fig. 5. Accurate analysis reveals the presence of Co3O4 dispersed phase in all the catalysts, although the characteristic reflexes are mixed with that of MgAl2O4 phase. Also, CoO phase appears in carbon-rich Co/20C/MA and Co/40C/MA. The average particle sizes of Co3O4 and CoO were calculated from the Scherrer’s equation and the amount of each component was determined from relative intensity of correspondent reflections (Table 2). Notably, the average size of Co3O4 crystallites in the catalysts is pretty large, presumably due to low porosity of the supported spinel. However, the size decreases almost lineally with BET surface area of the samples (Fig. 6). Thus, carbonization of the support gives rise to higher dispersion of active phase.

Fig. 5
figure 5

X-ray diffraction patterns of pristine support and catalysts

Full size image
Table 2 Phase composition based on XRD data
Full size table
Fig. 6
figure 6

Average Co3O4 particle size in Co/xC/MA calculated from XRD data as a function of BET surface area calculated from N2 adsorption–desorption analysis

Full size image

The presence of CoO phase in Co/20C/MA and Co/40C/MA clearly indicates the partial reduction of Co3O4 by surface carbon in the course of catalyst calcination:

$${\text{2Co}}_{{3}} {\text{O}}_{{4}} {\text{ + C}} \to {\text{6CoO + Co}}_{{2}}$$

Evolution of CO2 at the temperature above 300 °C was detected in the experiments on thermal decomposition of air-dried catalyst precursors (Fig. 4). It seems not to be a result of carbon oxidation by NO2 because it is not accompanied with NO release. Rather, CO2 appears due cobalt oxide reduction with carbon.

Conducting magnetic measurements in situ during catalyst reduction in hydrogen flow at 450 °C makes it possible to observe the evolution of ferromagnetic phases in the samples. For the studied catalysts, the only such phase is metallic cobalt. In the course of the reduction in hydrogen atmosphere, sample magnetization increases due to formation of metallic cobalt. Monotone appearance of the curves points on single mode Co0 particle size distribution. Based on kinetic curves, initial reduction rate increases with carbon percentage in the catalysts: Co/40C/MA > Co/20C/MA >  > Co/4C/MA ≈ Co/MA (Fig. 7). Surface carbon definitely accelerates cobalt oxides reduction at the start of the process, in line with the results listed above. However, fast growth of reduction of carbon-rich Co/40C/MA and Co/20C/MA quickly slows down while other catalysts demonstrate slower but steady reducibility rate.

Fig. 7
figure 7

Magnetization of the samples with time in catalyst reduction experiments

Full size image

Since metallic cobalt is the only ferromagnetic phases in the reduced catalysts, the saturation magnetization corresponds to its amount in the sample. After activation in hydrogen at 450 °C for 2 h, the sample were cooled down to ambient temperature and saturation magnetization values were measured by extrapolation the magnetization to zero field. Calculated amount of metallic cobalt seems to be 10–11 wt% for all the samples, as is of cobalt content determined by ICP-MS. Therefore, given activation parameters are sufficient for complete reduction of supported cobalt into metallic state which is known to be active phase in CO hydrogenation [1, 3, 8]. Calculated metallic cobalt particle size and cobalt dispersion in reduced catalysts are given in Table 3.

Table 3 Cobalt particles size and metallic dispersion in reduced catalysts
Full size table

For the reduced catalysts, dependence of magnetization on magnetic field strength (H) were measured. Hysteresis loop was recorded for all he samples indicting the presence of both single-domain and multi-domain ferromagnetic particles. Coercivity (HC) for reduced Co/MA and Co/4C/MA is almost the same but decreases with percentage of carbon in the samples (Fig. 8). This indicates decreasing average size metallic cobalt particles and correlates well with our XRD data for Co3O4 crystallite size and calculated cobalt metal particle size [28].

Fig. 8
figure 8

Effect of carbon weight amount on coercivity HC (A/m) for the reduced catalysts

Full size image

Catalytic performance in CO hydrogenation was measured using fixed-bed down-flow reactor under typical low temperature FTS condition: temperature of 210 °C, pressure of 20 bar, CO/H2 = 1/2. Carbon monoxide conversion slightly decreased during first 30–40 h on operation. Also, selectivity to methane decreased while selectivity to higher hydrocarbons increased for this time. These effects may result from gradual filling of pores in catalyst grains with condensed reaction products. Consequently, diffusion of reagents and products thought the liquid slows down thus affecting both CO conversion and selectivity (Fig. 9).

Fig. 9
figure 9

Catalyst performance with time on stream. T = 210 °C, P = 20 bar, feed gas flow rate 5 NL.gcat–1.h–1 and, feed gas composition (vol%): 27 CO, 64 H2, 9 Ar

Full size image

Catalyst performance indexes obtained after 48 h on stream are given in Table 4. Catalytic activity expressed as cobalt-time yield is related to carbon content in the sample. The most active catalyst is Co/4C/MA while increasing carbon loading leads to decrease in cobalt-time yield. However, the extent of support carbonization is in positive correlation with molar selectivity to target higher hydrocarbons. Indeed, the most carbonized sample Co/40C/MA exhibited the lowest selectivity to methane and the highest selectivity to C5+ fraction.

Table 4 Activity and selectivity data at T = 210 °C, P = 20 bar, time on stream 48 h, feed gas (vol%): 27 CO, 64 H2, 9 Ar
Full size table

Calculated metallic cobalt dispersion on the catalysts surface is rater low (Table 2), presumably due to low specific surface area of the spinel support. This is why cobalt-time yield in hydrocarbon synthesis is modest and lower than that reported for Co/γ–Al2O3 catalyst under similar experimental conditions [28, 30]. However, turnover frequency is comparable to that reported previously [28, 30]. Indeed, turnover frequency and product selectivity in CO hydrogenation were found to be independent on cobalt particle size for the particles larger than 6–8 nm [30, 31]. As far as metallic cobalt particles in our catalysts are much larger (Table 3), no expressed size-dependent effect should be observed. Nevertheless, somewhat decrease in turnover frequency and methane selectivity in the series Co/40C/MA > Co/20C/MA > Co/4C/MA are to be explained (Fig. 10). Average Co3O4 particle size in these samples decreases in the same sequence. Speculatively, most of the cobalt particles in Co/40C/MA and some of them in Co/20C/MA are indeed located inside mesopores. Therefore, diffusion in narrow pores restricts product transport to outer space of catalyst grains and fosters olefin readsorption on active sites. As a result, selectivity to higher hydrocarbons increases and selectivity to methane decreases (Table 4, Fig. 10).

Fig. 10
figure 10

Turnover frequency and methane selectivity after 48 on stream at T = 210 °C, P = 20 bar

Full size image

4 Conclusion

In summary, a series of cobalt catalyst supported on carbonized MgAl2O4 spinel were prepared, characterized and tested in CO hydrogenation. Support carbonization was performed by deposition of glucose from water solution followed by thermal treatment in inert atmosphere. The size of Co3O4 crystallites in calcined catalysts decreases with carbon content in the support. The same regularity is found for metallic cobalt particles in hydrogen-activated catalysts. The samples with highest carbon concentration provides partial reduction of Co3O4 into CoO during calcination in inert atmosphere. Reducibility of the catalyst was studied by in situ magnetic measurements. Cobalt is reduced completely within 2 h in typical activation conditions for FTS cobalt catalysts: H2 flow, 450 °C. Activity in CO hydrogenation in low temperature FTS conditions (210 °C, 20 bar) decreases while selectivity to higher hydrocarbons increases with carbon content.