Ultrafine metal-polymer catalysts based on polyconjugated systems for Fisher–Tropsch synthesis

Introduction

The catalysts used for slurry reaction processes based on the synthesis gas (Fischer–Tropsch synthesis, methanol synthesis, dimethyl ether synthesis, etc.) with nanosized suspended particles have attracted special attention in recent decades, and in situ preparation methods (drop thermolysis, for example) represent one of the most important areas in the field [1], [2]. These catalysts exhibit some unique properties, for example, a high specific activity, a limited ability to precipitate, and a good miscibility, leading to the maintenance of the uniformity of their spatial distribution and the temperature range of the exothermic reaction. However, slurry systems can agglomerate easily [3], [4], ceteris paribus, and the smaller a catalyst particles is, the more pronounced the effect. This undesired effect can be eliminated by the introduction of a surfactant; however, many effective surfactants can affect the catalyst as strong poisons and are unacceptable for use during the Fischer–Tropsch synthesis or the preparation of catalysts.

Another possible solution to the agglomeration problem is the use of polymer systems as a stabilizing component. The concept of such a nanocomposite was implemented in [5], [6] and extended to the Fischer–Tropsch synthesis of Fe-containing slurry nanosized catalysts [7], [8], [9], [10]. The differences in the Fe/(polymer matrix) composite particle size and catalytic properties for several polymers were shown [7]; an interrelation between the high activity of several catalysts and their FT-IR features was established [8]. Additionally, the polymer chain flexibility and its ability to coil around a Fe nanoparticle, resulting in a complex metal-polymer particle, were determined by performing MD calculations for the polyacrylonitrile case [9]; the X-ray diffraction (XRD) and atomic force microscopy (AFM) data obtained for styrene-based matrices demonstrated the formation of both Fe₃O₄ and δ-FeOOH phases as well as differences in the crystallinity of the liquid paraffin slurry media [10]. However, the nature of the triple metal-paraffin-polymer (conjugated or not) interactions within a composite nanocatalyst and their influence on the nanosized Fe catalytic activity and selectivity have not yet been understood in depth.

The topic of the present paper, accordingly, is to more carefully investigate the interrelations between the characteristics and catalytic features of iron-polymer Fischer-Tropsch nanosized slurry catalysts. By following [10], AFM and XRD methods were used for this purpose.

Experimental

The catalyst preparation, testing and product analysis were performed according to literature procedures [7], [8]. The XRD patterns of the samples before synthesis were recorded on a Shimadzu XRD-7000 diffractometer using Cu Kα radiation (40 kV, 30 mA, scanning rate of 2 s, room conditions). AFM measurements were performed on a Titanium spectrograph with PX Ultra and HybriD^TM controllers by using the probes CSG10 (F_res = 18.5 kHz, k = 0.07 N/m) and NSG01 (F_res = 187.2 kHz, k = 3.9 N/m) and the HybriD method (DMT model) for the Fischer-Tropsch-used catalytic samples before and after hexane treatment (for the removal of the surface layer of heavy paraffins).

Results and discussion

For the iron-containing nanosized nonpromoted Fe catalyst prepared in situ in P-2 industrial paraffin, a mixture of long-chain alkanes (with a chain length C₁₉–C₃₃ and Gaussian-like length distribution), both free and with the addition of different polymer components (mass ratio of 6 Fe/10 polymer), the temperature dependencies for CO conversion and the selectivity were obtained (Figs. 1–3). Polyacrylonitrile (PAN), polyvinyl alcohol (PVA), styrene-divinylbenzene copolymer (PS-DVB) and polyamide-6 (PA) were used as the additives in [7]; subsequently, the list was supplemented with polyethylene (PE) (300 °C, [8]) and polystyrene (PS) [10]. For the purposes of this work, the experiments were supplemented with the data on PE for 220–280 and 320 °C to obtain temperature dependencies similar to those of the other samples.

Fig. 1:

Temperature dependencies of the total CO conversion: 1 – Fe/P-2 [7], 2 – Fe/P-2/PAN [7], 3 – Fe/P-2/PVA [7], 4 – Fe/P-2/PS-DVB [7], 5 – Fe/P-2/PA [7], 6 – Fe/ P-2/PE ([8] for 300 °C), 7 – Fe/P-2/PS [10].

Fig. 2:

Temperature dependencies of the CO₂ selectivity: 1 – Fe/P-2, 2 – Fe/P-2/PAN, 3 – Fe/P-2/PVA, 4 – Fe/P-2/PS-DVB, 5 – Fe/P-2/PA, 6 – Fe/ P-2/PE, 7 – Fe/P-2/PS.

Fig. 3:

Temperature dependencies of the C₅₊ (excluding CO₂) selectivity: 1 – Fe/P-2, 2 – Fe/P-2/PAN, 3 – Fe/P-2/PVA, 4 – Fe/P-2/PS-DVB, 5 – Fe/P-2/PA, 6 – Fe/ P-2/PE, 7 – Fe/P-2/PS.

For the CO conversion dependencies (Fig. 1), the polymer-containing samples were divided into two distinct groups, which was partially already shown in [7], [10] and has been supplemented and confirmed. The activities of the PAN-, PVA-, PS-DVB-containing samples are close to (for PAN and at elevated temperatures, the activity slightly exceeds) the activity of the polymer-free sample, and the activities of the PA-, PS-, and PE-containing systems are significantly lower than that of the polymer-free sample and almost coincide with one another for all three cases.

The selectivity data of the aforementioned samples presented in [7], [10] were related to the C₅₊ content, and the values ​​were determined based on all the carbon-containing products, including CO₂, and a small amount of higher alcohols. The comparison made in [7] for the PAN-, PS-DVB-, PVA-, and PA-based catalysts and in [10] for the PS- and PS-DVB-based catalysts showed that the samples with low activities revealed higher C₅₊ selectivity values than the samples with high activities. However, it is more logical to determine the tendency of any Fischer-Tropsch synthesis catalyst to form long-chain products by considering the selectivity values, which do not take into account that CO₂ does not directly affect the growth of the hydrocarbon chain. The temperature dependencies of the CO₂ selectivity are shown in Fig. 2; the C₅₊ selectivity (not taking into account the formation of CO₂) is shown in Fig. 3. For both cases, the difference between the catalysts containing the polymer matrices of the two different groups (as well as the proximity of the indicators for the PAN group and others to those of the polymer-free sample) is as obvious as it is for total CO conversion. However, for CO₂, this difference is not as pronounced as it is for CO conversion. For C₅₊, a clear distinction between the two groups is not revealed already – however, the CO₂ values of the samples with PA, PS, and PE differ (surpass) more from the CO₂ values of the polymer-free sample than the CO₂ values of the PAN, PVA, and PS-DVB samples differ from the those of the polymer-free sample.

As a result of introducing the aforementioned polymers into the nanosized slurry catalytic systems, as was established by the results of dynamic light scattering [7], [8], the unimodal distribution of the particles changes to a bimodal distribution due to the formation particle fractions that differ significantly in size and are smaller and larger with respect to the sizes of the polymer-free particles. However, a completely satisfactory relationship between these changes and the variation in the activity and selectivity has not been established. PE and PS caused the formation of extremely small iron particles (≈2 nm [8]), and it is well-known for Fischer–Tropsch catalysis that the formation of extremely small particles (nearly atomic clusters) does not always cause an increase in activity (for Fe-containing systems, a good example of such a behavior for pre-synthesized γ-Fe₂O₃ particles fixed on Al₂O₃ is presented in [11]). However, the PA-containing system does not follow this pattern – its activity is as low as that of PE- and PS-containing catalysts, but Fe particles are the largest among all the investigated samples [7], [8].

Several fundamental reasons for the manifestation of such a trend were presented in [8], namely:

1. the formation of peculiar π-complexes involved the polyconjugated bonds of the polymers used and the surface atoms of Fe particles. During in situ thermolysis (250–280 °C), the polymer structures transformed into polyconjugated structure; for PAN,

for PVA,

PS-DVB was confirmed from the well-defined (for PAN – more, for the other two polymers – less) changes of the catalyst FT-IR pattern in comparison with the free polymer FT-IR spectrum; for PAN, the appearance of the temperature-induced (>150 °C) polyconjugation of C≡N bonds was known already [12], [13]. On the other hand, no polyconjugation was observed for PA, PE, and PS, which was correlated with a low activity, and, therefore, it is not necessary to speak about any π-complexes. However, the role of such complexes, if they even exist, should be discussed at the moment, with caution. If the changes in the activity were caused only by such electronic reasons, one would expect a marked increase in the CO conversion for the PAN-, PVA-, and PS-DVB-containing catalysts, not only with respect to the other three polymer-containing systems but also with respect to a P-2-containing sample.

The following is not observed (Fig. 1):

1. the formation of specific spatial structures (Fig. 4) that limit/do not limit the diffusion of Fischer-Tropsch synthesis reagents in the immediate vicinity of active centers, which can probably have some effect on hydrocarbon chain growth. Apparently, a diffusion-limiting variant of such a structure can only be a product of complex triple metal–polymer–paraffin interactions. An Fe nanoparticle is “packed” inside the polymer molecule upon the interaction with PAN [9], so, in this case, it could be assumed that the PAN-containing catalyst, in contrast to the observations, will exhibit an activity comparable to the activities of the PA-, PE-, and PS-based catalysts but not comparable to the activity of a polymer-free sample. Signs of the existence of these structures and their formation, specifically, for low-activity samples, are revealed in the FT-IR pattern [8]: the spectra obtained for the PA-, PE-, PS-based systems are more similarity to the pure P-2 pattern than to the initial polymer pattern. The combinations of C-H lines indicate that the alkane chains are deformed, and their conformation is significantly different from the conformation manifested for P-2. This suggests that in the case of PA, PE, and PS, the P-2 alkane chains cover the polymer coils on the outer surface and, thus, create a “two-layer” shell that is, probably, slightly more hydrophobic than the polymer coils (so the shell can a slightly suppress the H₂O-to-CO₂ water gas shift; see Fig. 2) and can constrain the diffusion of the Fischer-Tropsch synthesis reagents and products (i.e., can slightly suppress the overall chain termination step of the Fischer-Tropsch mechanism and increase the C5+ selectivity; see Fig. 3) much more effectively than the “single-layer” shell in the case of PAN, PVA, and PS-DVB. Several different kinds of paraffin nucleation events (based on the crystallization degree, pour point, etc.) around the different polymer chains during the wax inhibition of crude oils and refined fuels were described [14], [15], [16], and one has no reason to reject the possibility of similar processes and differentiations according to the polymer nature (namely, in the scope of the presence/absence of chain polyconjugation) in the vicinity of the metal nanoparticles.

Fig. 4:

Schematic representation of typical combinations of a metal particle/polymer matrix in a composite nanoparticle: a – central (more diffusion limitations), b – peripheral (less diffusion complications).

Thus, the current experimental data provide quite important evidence in favor of the fact that the paraffin chains of the slurry medium which are involved in specific triple interactions with the polymer matrix and/or with the immobilized iron nanoparticles under the conditions of the Fischer–Tropsch synthesis method, can have a significant effect on the catalytic performance of such systems. The following XRD and AFM data confirm this conclusion.

The XRD pattern of a polymer-free catalyst (Fig. 5a) demonstrates a combination of crystalline and amorphous forms, both for the solid (under the XRD conditions) alkane medium and the iron-containing catalyst itself (crystalline Fe₃O₄ and the amorphous product of incomplete thermolysis – δ-FeOOН). The introduction of PAN (as well as PVA) makes the XRD pattern very similar to that presented in [10] for a PS-DVB-containing system (Fig. 5b) – the characteristic features of the amorphous state of the paraffin mixture almost disappear, while the phase composition of the Fe-containing particles varies slightly from that of the reference sample.

Fig. 5:

a – XRD pattern of Fe/P-2. b – XRD pattern of Fe/P-2/PAN. c – XRD pattern of Fe/P-2/PA.

If PA (or PE) is used as a polymer-containing medium, the XRD pattern changes and becomes very similar to the XRD spectrum of a system with PS, as presented in [10] (Fig. 5c). Paraffin exhibits an amorphous conformation, even to a somewhat greater extent than in the absence of a polymer, but the composition of the catalyst itself changes – δ-FeOOН is the dominant phase, while Fe₃O₄ is not observed practically. Thus, the combination of a paraffin medium with nonpolyconjugated polymers clearly leads to serious differences in the course of in situ catalyst formation. It is possible that this peculiarity had determined the activity and selectivity differences for nonconjugated polymers to a greater extent than the hypothetical formation of π-complexes or “double coating” of active particles.

The role of paraffin and its state in polymer-containing catalysts is additionally revealed using AFM immediately after Fischer–Tropsch synthesis and then after washing the sample with hexane, which seems to dissolve the surface layer of paraffin faster than any of the polymers used and removes it from the system. A similar approach was tested in [10] for catalysts based on PS-DVB and PS. In the first case, before hexane treatment, the surface of the sample turned out to be quite smooth, and the characteristic size of the irregularities very rarely exceeded 100 nm, reaching 300–350 nm (the average particle sizes of the catalyst, according to DLS, were larger – 180 and 669 nm – for the corresponding fractions of the bimodal particle distribution [7]). After hexane treatment, the relief becomes sharper, and the size of the irregularities significantly decreased (≈80 nm or less). In contrast, washing the catalyst with PS did not lead to any significant changes, either in the relatively smoothed surface morphology or in the scale of the irregularities (≈50 nm or less). These observations additionally support the hypothesis that the P-2 paraffin molecules interact with the Fe particles, and the nonpolyconjugated polymer is durable and resistant to dissolution. In the case of PS-DVB, a sufficient portion of the paraffin and, apparently, part of large Fe particles, are removed with a solvent. These particles most likely lost a tight contact with the polymer matrix during the Fischer-Tropsch synthesis procedure.

The AFM data obtained for the remaining samples, as presented below, supplement this picture. The polymer-free sample (Fig. 6a) demonstrates the diverse nature of the surface morphology (which can be attributed to the manifestations of both the amorphous and crystallized forms of the solid paraffin medium) and the presence of irregularities with very different sizes, the largest size of which practically corresponds to the average size of Fe particles for this system (455 nm [7]).

Fig. 6:

a – AFM morphology of Fe/P-2. b – AFM morphology of the Fe/P-2/PAN system (left – before, right – after hexane treatment). c – AFM morphology of the Fe/P-2/PVA system (left – before, right – after hexane treatment). d – AFM morphology of the Fe/P-2/PA system (left – before, right – after hexane treatment). e – AFM morphology of the Fe/P-2/PE system (left – before, right – after hexane treatment).

For the PAN-based system (Fig. 6b), as for the case of PS-DVB [10], the hexane treatment led to significant morphological changes of the sample surface, although the nature of the various surfaces turned out to be exactly the opposite: the relief was sharp, and the smaller scale of the irregularities gave way to a smoother one, with an increased size of irregularities. It can be suggested that in this case, small particles (rather than large ones for the PS-DVB case) are bound weakly with the polymer; however, as for PS-DVB, it can be stated that the ternary Fe-P-2-PAN interactions are “unstable”. The PVA-based system (Fig. 6c) turns out to be very similar to the PAN system.

The surface morphologies of the systems obtained using PA (Fig. 6d) or PE (Fig. 6e) differ markedly, as one would expect, from systems with polyconjugated polymers. However, the behavior of the systems obtained using PA or PE is not quite similar to that known for PS-containing catalysts. Hexane washing leads to a noticeable decrease in the size of the irregularities in both cases (for PE, the irregularities are much larger than the particle sizes known according to the DLS data – 2 nm for the numerically predominant fraction [8]; for PA, the irregularities are close to the particle size). However, the general character of the observed relief after washing remains more pronounced for the PA-containing system and generally smoothed for the PE-containing one. It can be suggested that the removal of alkanes, the polymer and some large Fe particles have occurred more or less simultaneously in both cases.

Thus, the AFM data additionally confirm the differences in the behavior of the paraffin medium in the solid state when combined with conjugated or nonconjugated types of polymers. These features most likely reflect differences in the behavior of alkane chains under Fischer-Tropsch synthesis conditions.

Conclusions

The new AFM and XRD data obtained for Fe nanosized Fischer–Tropsch slurry catalysts prepared with conjugated or nonconjugated polymer nanomatrices confirm that the liquid alkanes of the slurry medium not only act as a “hydrodynamic agent”, ensuring the maintenance of a more or less uniform spatial distribution of slurry particles but also probably can:

1. affect both the mobility and binding of Fe particles to polymer matrices;

2. impose (together with a nonconjugated polymer matrix) diffusion restrictions on the reagents and products of the Fischer–Tropsch synthesis, which caused the overall activity/conversion to decrease and C₅₊ selectivity to increase; and,

3. in some cases of a nonconjugated matrix, affect the phase composition of Fe particles during the in situ formation of the catalyst.

These results can serve as a starting point for experimental and theoretical studies on the creation of new Fischer–Tropsch catalysts with a pre-configured matrix environment of active metal nanoparticles. The development of models of these paraffin-polymer shells using analogies from the field of oil wax inhibition by polymers seems to be a crucial and fruitful component of the proposed research. It can be expected here that new highly active, selective, fine-tuning industrial Fischer–Tropsch catalysts as well as scientifically precious models of the complex behavior of nanocatalytic systems shall be elaborated.