1 Introduction

Fischer–Tropsch Synthesis (FTS), which is an alternative route for the production of diesel oil independent of petroleum material, has recently attracted much concern [1]. It is well known that supported cobalt catalysts are preferred for FTS because of their high activities and high selectivity to linear long chain hydrocarbons and also low activities for the competitive water–gas shift (WGS) reaction.

The performance of catalysts depends on many factors [2], such as preparation method, thermal treatment process, and the interaction between active species and support, catalyst compositions, metal dispersibility and types of inorganic supports used, etc. Thus, various materials were used as the supports of cobalt catalysts in past studies, including silica, alumina, kieselguhr, zeolite, titanium, carbon, and magnesia [3–7], which significantly influenced the reduction extent, the morphology, the adsorption capabilities and catalytic performance of the active phase, especially in well-dispersed catalytic systems. Among these supports, mesoporous silicas had attracted widespread attentions in FTS due to their high surface area, favoring the cobalt dispersion. Furthermore, the pore size could also control the cobalt particle size, improve the diffusion of reactants and products in the pore channel, and then influence the distributions of hydrocarbon products from the FTS.

Anderson et al. [8] reported that the FTS activity and selectivity of cobalt based catalyst could be affected by their pore sizes. And Xiong et al. [9] indicated that the pore size of alumina support could significantly influence the Co3O4 crystallite diameter, catalyst reducibility and FT activity. It was also reported [10] that the support with small pores could achieve a high dispersion of supported cobalt crystallites due to their high support surface area, and those supports with large pores could diminish the diffusion resistance and provide pathways for rapid molecular transport. Moreover, it was observed [11] that larger cobalt particles located in the wider pore silicas led to higher activity in FTS and lower methane selectivity than smaller cobalt particles situated in narrower pore supports.

Tsubaki et al. [12, 13] prepared a bimodal catalyst with mesopores and macropores by introducing silica or zirconia sols into large-pore silica gel. Such a catalyst showed high activity and favorable selectivity owing to the high dispersibility of the cobalt crystallite in the bimodal structure. But the effects of bimodal regular mesopores and macropores on Co-based FTS catalysts remained not clear yet. Thus, in the present study, the catalysts with 15 wt% cobalt loading on supports with different pore size distributions were prepared by the same method, and the FTS over the catalysts were carried out to investigate the influence of mono-modal and bimodal structure.

2 Experimental

2.1 Material Preparation

2.1.1 Support Preparation

The mono-modal support was prepared by traditional method [11]. The bimodal mesopores and meso-macropores supports were prepared using CTAB as surfactant, but using Na2SiO3 and TEOS as silicon source, respectively.

SM-MCM41: In a typical synthesis, 1 g CTAB was dissolved in 480 ml of distilled water. Afterwards 0.28 g NaOH was added, after the solution was stirred at 353 K for about 15 min, 5 ml TEOS was added dropwise and the solution was continually stirring for 2 h. The suspension was filtrated, dried in an oven at 333 K for 12 h and finally the sample was calcined at the rate of 5 K/min to 823 K and hold for 6 h. The obtained support was denoted as SM-MCM41(simple mesopores mono-modal MCM41).

MM-MCM41: The synthesis process of MM-MCM41 was similar to SM-MCM41. Under 353 K, 1 g CTAB was firstly dissolved in 480 ml of distilled water, afterwards 16 ml 35% ammonia was added. After stirring the solution for 15 min, 5 ml TEOS was added dropwise. When the solution was continually stirred for another 2 h, the suspension was filtrated, dried in an oven at 333 K for 12 h, and finally the sample was calcined at the rate of 5 K/min to 823 K and hold for 6 h. The obtained support was denoted as MM-MCM41 (mesopores-marcopores bimodal mesopores).

DM-MCM41: In a typical synthesis, 19.6 g CTAB and 10 g Na2SiO3 were dissolved in 350 ml distilled water at 353 K. After stirring for 30 min, 35 ml ethyl acetate was added dropwise. Then the suspension was stilled for 5 h, aged at 363 K for 48 h. The suspension was filtrated, dried in an oven at 333 K for 12 h, and finally the sample was calcined at the rate of 5 K/min to 823 K and hold for 6 h. The obtained support was denoted as DM-MCM41 (double mesopores bimodal mesopores).

2.1.2 Catalyst Preparation

The bimodal supported cobalt catalysts were prepared by incipient wetness impregnation (IWI) method using aqueous solution of cobalt nitrate as initial material. Firstly, the silicon supports were impregnated into the quantitative Co(NO3)2·6H2O solution and stilled for 24 h. Then, the samples were dried at 333 K for 12 h. Finally, the cobalt-supported catalysts were obtained by calcining the above samples in air at 773 K for 2 h. The synthesized Co/MCM41 catalysts with 15 wt% metal loading were denoted as Co/SM-MCM41, Co/DM-MCM41 and Co/MM-MCM41, respectively.

2.2 Characterizations of Catalyst

BET: N2 adsorption–desorption experiment was conducted at 77 K with a ASAP-2000 Micromeritics instrument. Nitrogen isotherms were obtained in both adsorption and desorption modes. The surface areas of supports and catalysts were determined by the BET method. The total pore volume (TPV) was calculated from the amount of vapor adsorbed at a relative pressure (P/P0) close to unity, where P and P0 were the measured and equilibrium pressures, respectively. Pore size distribution curves were established from the desorption branches of the isotherm using the BJH model. Before the analysis, the samples were outgassed at 393 K for 12 h.

XRD: Rigaku D/max-RA instrument with Cu-K radiation was used for the XRD measurement. The spectra were scanned at a rate of 2°/min in the range 2θ = 20–80°. The cobalt particle size (Co3O4 at 2θ = 36.9°) was calculated by using the full width at half maximum (FWHM) value with the help of Scherrer’s equation as well.

Raman: Raman analysis was performed on a LABRAM–HR 800 spectroscopy at room temperature with an excitation wavelength of 514 nm. A scanning range between 100 and 1000 cm−1 with a resolution of 2 cm−1 was applied.

TEM: The morphology of the samples was observed by transmission electron microscopy (TEM; Hitachi-600).

TPR: TPR experiments were carried out with a U-tube quartz microreactor heated by an electrical furnace to determine the reducibility of the catalyst sample. The reactor was loaded with 25 mg catalyst and heated at a temperature ramp from 333 to 1233 K at 10 K/min with a gas consisting of 5% H2 in N2. The gas flow rate was 60 ml/min. The H2 consumption (TCD signal) was recorded automatically by TCD detection.

2.3 Catalyst Evaluation

FTS reaction was performed in a fixed bed reactor (i.d. = 12 mm) at 2.0 MPa, 1000 h−1 and a H2/CO ratio of 2.0. The catalyst (60–80 mesh) was mixed with the same volume SiO2 to minimize the temperature gradient and reduced in a flow of hydrogen at 673 K for 6 h and then cooled down to ambient before switching to syngas. Data were taken at steady state after 24 h on-stream. The gas effluent was analyzed on a GC-920 chromatographs equipped with thermal conductivity detector (TCD) and flame ionization detector (FID). Liquid products and wax were collected in a cold trap and a hot trap, respectively, and were off-line analyzed on a GC-2010 chromatograph, which was equipped with a 35 m OV-101 capillary column.

3 Results

3.1 Characterizations

3.1.1 Texture

The isotherms of nitrogen adsorption and desorption, and the corresponding pore size distribution curves calculated using BJH method [1] of the catalysts were displayed in Figs. 1, 2, respectively. The isotherms of bimodal catalysts exhibited a typical irreversible IV type adsorption isotherm with two separate, well expressed H1 hysteresis loops at relative pressures P/P0 of 0.2–0.4 and 0.8–1. The first condensation step on the isotherm at 0.2–0.4 was similar to that for common MCM-41 materials with markedly higher saturation sorption capacity, though not very steep. The second condensation steps of the Co/DM-MCM41 and Co/MM-MCM41 on the isotherm at P/P0 > 0.8 were steeper than the first ones and the hysteresis loops were more wide. This indicated the presence of a significant amount of secondary pore [14]. As shown in Fig. 2, it was clearly evident that there existed three kinds of pore, the maximum distribution centered at 2.5 nm for Co/SM-MCM41, 2.5, 17 nm for Co/DM-MCM41, and 2.5, 55 nm for Co/MM-MCM41, respectively.

Fig. 1
figure 1

N2 adsorption–desorption of catalysts

Full size image
Fig. 2
figure 2

Pore size distribution of the catalysts

Full size image

The BET surface area, total pore volume and average pore diameter of the supports and corresponding catalysts were presented in Table 1. Compared with the respective supports, the cobalt loaded catalysts showed lower BET surface area, pore volume and smaller average pore size, but the pore size distribution has no obvious changes. The experimental BET surface area of Co/DM-MCM41 was 483 m2/g, which was close to its theoretical data (554 m2/g) [15], indicating that Co species might not block the pores, and a majority of cobalt species entered into the pore of support. This result was in agreement with the report [8] that cobalt species entered into the pore of support by incipient wetness impregnation method. Co/MM-MCM41 and Co/SM-MCM41 had the lower experimental surface areas than the theoretical data, and Co/DM-MCM41 had a slightly larger pore size than DM-MCM41, which might be due to the collapse of pore [15].

Table 1 Physico-chemical properties of the supports and Co/SiO2 catalysts
Full size table

3.1.2 Phase Structure

The X-ray powder diffraction patterns of the catalysts were presented in Fig. 3, and the Co3O4 crystallite diameters, calculated by the Scherrer equation [16] (2θ = 36.9°), were listed in Table 1. For all the catalysts, the diffraction peaks at 31.4°, 36.9°, 44.8°, 59.4°, 65.2° were corresponding to the spinel Co3O4 phase. The size of Co3O4 crystallite changed from 12 to 34 nm for all the catalysts. Clearly, Co/DM-MCM41 had the smallest Co3O4 crystallite size (12 nm) in these samples. This result was not consistent with the literatures [1, 13], which showed that the crystallite diameter increased with the pore size. Co/DM-MCM41 and Co/MM-MCM41 showed smaller Co3O4 crystallite sizes than the corresponding large pore size, implying that a majority of cobalt crystallites were located inside the pore and their growth was possibly confined by those pores. At the same time, Co/SM-MCM41 had the larger Co3O4 crystallites than the corresponding pore size, which indicated the cobalt species might mostly locate outside of the pores.

Fig. 3
figure 3

XRD patterns of the Catalysts

Full size image

3.1.3 Raman Spectroscopy

Raman spectroscopy is a powerful tool in the study of the microstructure of nanoparticles [17]. It is known that Co2+ and Co3+ ions of Co3O4 oxide crystallizes with the normal spinel structure (Co2+(Co3+)2(O2−)4) are located at tetrahedral and octahedral sites, respectively [18]. It has been known that the representative Raman peak positions of Co3O4 were 193, 475, 516, 615, and 680 cm−1, corresponding to all of the five Raman-active modes (A1g, Eg, and 3 F2g) of Co3O4, respectively. Those absorbance, 672, 691, and 682 cm−1 corresponded to the A1g active mode of Co/DM-MCM41, Co/SM-MCM41 and Co/MM-MCM41, respectively (see Fig. 4a). Compared to Co/SM-MCM41, the Co/MM-MCM41 had a blue shift of about 9 cm−1 and Co/DM-MCM41 about 19 cm−1 due to the size quantization effect, i.e., the blue shift showed the decrease in the crystalline size. Thus, it could be concluded that the Co3O4 sizes reduced gradually in the order of Co/SM-MCM41, Co/MM-MCM41, and Co/DM-MCM41, which was coincident with the XRD result.

Fig. 4
figure 4

Raman spectroscopy of the pre-catalysts and post-catalysts

Full size image

The post-catalysis Raman characterization was showed in Fig. 4b. After the reaction, those absorption peaks at 667, 672, and 675 cm−1 corresponded to the A1g active mode of Co/DM-MCM41, Co/SM-MCM41, and Co/MM-MCM41, respectively. The Raman peaks indicated the existence of cobalt oxide in the post-catalysis. The above data showed the blue shift character of post-catalysis, which indicated the cobalt particles size decreased after reaction. This might be due to the reduction of cobalt oxide.

3.1.4 TEM Micrographs

Figure 5 showed TEM micrographs of three kinds of catalysts with different pore distribution. TEM investigation provides the direct observation of the morphology and distribution of Co particles in the support. The bimodal catalysts showed obvious pore structure of support, whereas the mono-modal catalyst blocked the pore, which indicated the cobalt particles located outside the pore in clusters. It was clearly displayed that the bimodal catalysts had smaller particle size and higher dispersion than that of mono-modal one. A close look at Co/DM-MCM41 revealed the presence of a considerable number of small cobalt clusters and more homogeneous cobalt distribution (see Figs. 5c, d), which almost existed in the two kinds of pore. The dark areas in the images of Figs. 5e, f showed that cobalt oxide clusters with a wide distribution concentrated more towards the external surface of the Co/SM-MCM41 catalyst, which further confirmed the cobalt larger cobalt particles existed outside the pore channel in agglomeration.

Fig. 5
figure 5

The HRTEM images of catalysts. a, b Co/MM-MCM41; c, d Co/DM-MCM41; e, f Co/SM-MCM41

Full size image

3.2 Reduction Behavior

TPR curves of the catalysts with different pore distribution were showed in Fig. 6. The reduction process underwent three stages. This indicated the presence of several kinds of reducible cobalt species. The first reduction peak at 523–613 K could be attributed to the first reduction step of Co3O4 (Co3O4 → CoO) [19], and the second reduction region at 613–673 K was due to the reduction of intermediate CoO phase (CoO → Co) [1, 20]. These reduction peaks indicated that the cobalt oxides of all catalysts were reduced to metallic cobalt before 673 K. However, for Co/MM-MCM41 and Co/SM-MCM41 catalysts, the high temperature reduction was observed at 923–1023 K, which might be caused by the presence of Co-Si compound (Co2SiO3) [21]. It should be mentioned that the second reduction region of Co/MM-MCM41 was comprised of more than one peak, which could be explained by the different cobalt particles size [22]. The literature reported that different size particles gave rise to varying the interaction between cobalt and support.

Fig. 6
figure 6

H2-TPR curves of the catalysts

Full size image

The reducibility of the catalysts were also calculated (see Table 1). The order of reducibility ranked as Co/SM-MCM41>Co/MM-MCM41>Co/DM-MCM41 (see Table 1). This was in agreement with the crystallite size of cobalt oxides, which was similar to those in the literatures [11, 23]. Obviously, the large particles in wide pores were more easily reduced than small particles located in narrow pores.

3.3 Catalytic Performance in FTS

Table 2 and Fig. 7 show the activity and selectivity of catalyst in FTS. It should be noted that bimodal catalysts exhibited better catalytic activity than mono-modal one, especially the double mesopores catalyst with the pore size of 2.5 and 17 nm. Such a structure not only led to the highest FTS catalytic activity but also the highest selectivity to C5–18 hydrocarbons (as high as about 62%). Furthermore, the Co/DM-MCM41 catalyst presented the lowest methane selectivity, which was almost invariable with the increase of activity.

Table 2 Activity and selectivity of different catalysts for FTS
Full size table
Fig. 7
figure 7

Catalytic performances of the catalysts. a the relation of CO conversion and temperature; b the relation of CO conversion and CH4 selectivity

Full size image

4 Discussion

The pore size was one of the key factors for the catalyst performance, and it affected the size of cobalt oxide particles, the mass transfer of reactants and products, the re-adsorption of α-alkene, and the chemisorption ratio of H2 and CO on the surface active sites [4]. Table 2 and Fig. 7 showed the activity and selectivity of FTS catalysts were markedly depending on their pore structure. The cobalt particles size was also controlled by the pore size, and the relationship between cobalt nanoparticles size and FTS activity showed the obvious structure-sensitive character for FTS reaction, which was consistent with the literatures [6, 24]. The bimodal catalysts had smaller cobalt particles and higher dispersion than that of mono-modal catalyst. Especially the double mesopores catalyst had the smallest particle size about 12 nm by the pore confinement. Compared to the mono-modal catalyst, the bimodal catalysts showed both higher CO conversion and C5+ selectivity, but lower methane selectivity.

For the Co/SM-MCM41 catalyst, its first pore size was only 2.5 nm, but the size of cobalt particle was 30.4 nm as obtained by XRD. This indicated that only a small quantity of cobalt particles was located inside the smaller pore, and then the pore could not confine the growth of cobalt particles. It was well known that the reducibility increased with the size of cobalt particles [8], so the Co/SM-MCM41 catalyst had the highest reducibility. However, the catalyst showed the lowest activity due to the formation of Co2SiO3 and the existence of less active sites on the cobalt species surface [25]. This was consist with the report [26] that larger concentration gradient existed in the larger cobalt particle interior, which caused the restriction of inner diffusion, decreased the diffusion and desorption ratios of CO and H2. As a result, the desorption of product within the cobalt particle was prevented, and the FTS activity and selectivity of the catalyst were affected.

For the bimodal catalyst, the CO conversion was greatly increased. Due to larger and smaller pores coexisting in the bimodal catalysts, their methane selectivity was remarkably lower than that of mono-modal one, which was also proved by the literatures [22, 27, 28]. A catalyst having a small pore size tend to produce lighter hydrocarbons, and the larger pores can contribute to transport primary products,1-olefins, more effectively and to decrease the methane formation rate from secondary hydrocracking of olefins. For the Co/MM-MCM41 catalyst, both mesopores and macropores co-existed simultaneously. Little cobalt particles entered into the smaller pore, and a majority of cobalt particles were located in the macropores. In this case, the number of active sites increased, showing a higher FTS activity than Co/SM-MCM41.

Due to the double mesopores coexisting, the Co/DM-MCM41 catalyst not only had the highest FTS activity and C5+ selectivity, but also the lowest methane selectivity in three kinds of catalysts. Although the activity of Co/DM-MCM41 catalyst was highest, the reducibility obtained from TPR was lowest, the result was not consistant with the result [23]. This could be due to the pore distribution and confinement of cobalt particles by the second mesopores of 17 nm, which has been proved by the XRD, Raman and TEM. This resulted in a higher dispersion of cobalt species [20] (see Fig. 5) and formation of more active sites formed, so the Co/DM-MCM41 catalyst showed the highest FTS activity. Furthermore, the Co/DM-MCM41 catalyst had the lowest methane selectivity and higher middle distillate selectivity than other two catalysts. This was similar to the literatures [1, 29], which was due to the co-existence of large and small pores in bimodal support. Sun et al. [4] also reported that there were different chemisorption ratios of H2 and CO on the surface active sites for the catalysts with different pore size. The large molecules in the product could not get out of the appropriate pore, favoring to produce lighter hydrocarbons [23]. Consequently, the Co/DM-MCM41 catalyst showed a higher selectivity to the middle distillates.

5 Conclusion

Both mono- and bi-modal cobalt-based FT catalysts were prepared by incipient-wetness impregnation method. The results showed that the bimodal structure catalysts present higher CO conversion and C5+ selectivity, lower methane selectivity than mono-modal catalyst. This might be because the bimodal mesoporous structure strongly influenced the cobalt crystallite size, the dispersion, the reducibility and the FT catalytic performance. Amongst them, the bi-modal catalyst with 2.5–17 nm pores showed the high activity, the low selectivity to methane and the high selectivity to middle distillates due to the confinement effect of the bimodal mesoporous structure.