Fischer–Tropsch Synthesis Over Zr-Promoted Co/γ-Al2O3 Catalysts

Vol.:(0123456789) 1 3 Top Catal (2017) 60:1285–1298 DOI 10.1007/s11244-017-0813-1 ORIGINAL PAPER Fischer–Tropsch Synthesis Over Zr-Promoted Co/γ-Al2O3 Catalysts JavierBarrientos1· VictoriaGarcilaso2· BaldassarreVenezia1· AtteAho3· JoséAntonioOdriozola2· MagaliBoutonnet1· SvenJärås1 Published online: 1 June 2017 © The Author(s) 2017. This article is an open access publication 1 Introduction The Fischer–Tropsch synthesis (FTS), i.e. the conversion of CO and H2 into hydrocarbons, is a widely known industrial catalytic process and key step in the transformation of carbonaceous sources (e.g. coal, natural gas and biomass) into liquid fuels [1, 2]. Supported cobalt catalysts have received great attention in FTS due to their high activity and selectivity to long chain hydrocarbons [3, 4]. Among different supports, Al2O3 is often used due to its high attrition resistance and its suitability in slurry bubble column reactors [5, 6]. Unfortunately, Co/Al2O3 catalysts present a hard reducibility due to the strong interaction between alumina and small cobalt oxide crystallites [7, 8]. A high degree of reduction is desired in order to maximize the Co0 site density and so, the activity of the catalyst. This problem can be overcome, to a certain extent, by using promoters. One solution is to promote Co/Al2O3 catalysts with small amounts of noble metals (e.g. Pt, Re and Ru) [9–11]. This promoting effect is generally explained by a H2 spillover mechanism, i.e. dissociative chemisorption of H2 and migration of H adatoms from the noble metal surface to the cobalt oxide surface [12–14]. Another solution is to modify the alumina support with an oxide promoter that offers a better cobalt-support interaction. Several studies have used promoters such as Zr [15–17], La [18], Mg [19], Mn [20], Ti [21] or Ce oxides [22] and most of them were reported to improve the catalyst activity. The effect of Zr on the performance of Co/Al2O3 catalysts has already been investigated by various researchers [15–17, 20, 23–25]. Yet, there is no consensus in what concerns its promoting effect. In the work of Rohr etal. [15] they evaluated the performance of Co/Al2O3 catalysts with Zr loadings varying from 0 to 10wt%. They found that Zr has no influence on the reducibility but Abstract Two Zr-modified alumina supports were synthetized containing the same amount of Zr but a different distribution of this modifier over the alumina surface. These supports, together with the unmodified alumina carrier, were used to prepare three cobalt-based catalysts which were characterized and tested under relevant Fischer–Tropsch conditions. The three catalysts presented very similar porosity and cobalt dispersion. The addition of Zr nor its distribution enhanced the catalyst reducibility. The catalyst activity was superior when using a carrier consisting of large ZrO2 islands over the alumina surface. The use of a carrier with a homogeneous Zr distribution had however, a detrimental effect. Moreover, a faster initial deactivation rate was observed for the Zr-promoted catalysts, fact that may explain this contradictory effect of Zr on activity. Finally, the addition of Zr showed a clear enhancement of the selectivity to long chain hydrocarbons and ethylene, especially when Zr was well dispersed. Keywords Fischer–Tropsch synthesis· Co/Al2O3· Cobalt· Zr· Activity * Javier Barrientos ja[email protected] 1 KTH (Royal Institute ofTechnology), Chemical Science andEngineering, Chemical Technology, Teknikringen 42, Plant 6, 10044Stockholm, Sweden 2 Química de Superficies y Catálisis, Instituto de Ciencia de los Materiales (US-CSIC), Av. Américo Vespucio no. 49, 41092Sevilla, Spain 3 Johan Gadolin Process Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, 20500Åbo/turku, Finland 1286 Top Catal (2017) 60:1285–1298 1 3 enhances the turnover frequency (TOF) and the selectivity to long chain hydrocarbons. Steady state isotopic transient kinetic analyses (SSITKA) revealed that the enhancement of the TOF was due to a change in the surface coverage of active intermediates and not to an increase of the intrinsic TOF, i.e. the TOF divided by the number of active surface intermediates. A similar study was later performed by Jongsomjit etal. [16] using catalysts containing between 0 and 11wt% of ZrO2. Nevertheless, they found that Zr enhances both the reducibility and the TOF. The authors also performed SSITKA experiments and obtained results comparable to those found by Rohr etal. [15]. However, they claimed that the number of active intermediates is probably a better measurement of cobalt active sites than that obtained from standard H2 chemisorption methods [26], and thus, suggested that Zr only enhances the reducibility but not the TOF. Other researchers [23] have showed that modification with 10wt% of ZrO2 can lead to smaller cobalt particles inside the pores and, as a result, limit the catalyst reducibility. Ma etal. [24] found that addition of 1–5wt% Zr enhances the reducibility of narrow pore Co/γ-Al2O3 catalysts (mean pore size: 4–5nm) but it does not have any influence on wide pore Co/γ-Al2O3 (mean pore size: 7–8nm). The reported TOF values were similar for all the catalysts. It may be worth mentioning that they also found that Zr increases the deactivation rate. Moreover, Shimura etal. [20] showed that the addition of 1wt% Zr to Co/θ-Al2O3 catalysts has no impact on the reducibility nor the activity of the catalyst. It may be noted that some discrepancies exist as well between studies promoting Co/SiO2 catalysts with ZrO2. Besides some studies [27], Zr has often been reported to increase the TOF [28–33]. However, some researchers [28–30] found that ZrO2 enhances the reducibility while others [31] claim that it does not have any influence. Finally, Hong etal. [32] found that Zr favors the reducibility in narrow pore Co/SiO2 catalysts (mean pore size: 2–3nm) but not in wide pore catalysts (mean pore size: 7–8nm). The present work aims to further investigate this effect of Zr on the reducibility and catalytic performance of Co/γ-Al2O3 catalysts at industrially relevant FTS conditions. The effect of the Zr distribution over the alumina surface was also evaluated in order to study the effect of this promoter from a different perspective. For that purpose, two Zr-modified alumina carriers were synthetized using different preparation methods. These two supports and the unmodified alumina carrier were further used to synthetize cobalt FTS catalysts with similar morphology and cobalt dispersion but a different cobalt-support interaction. 2 Experimental 2.1 Catalyst Preparation A ZrO2-modified alumina carrier was prepared using a conventional precipitation technique. For that purpose, a salt of ZrO(NO3)2·xH2O (Sigma Aldrich, 99%) was dissolved in EtOH (Merck, Abs Analytical grade). The alumina carrier was then added to the liquid solution and stirred for 30min. Afterwards, the mixture was treated with an NH3 aqueous solution 1:1 v/v H2O /NH3 (30% Panreac) under stirring for 30min. Finally, the solid was filtered, dried at 100 °C for 24h, and calcined at 500 °C (heating rate: 10 °C/min) for 4h in air. This support was named “ZrAl P”. Another ZrO2-modified support, using the same γ-Al2O3 carrier, was prepared by the microemulsion technique. For that purpose, an aqueous solution ZrO(NO3)2·xH2O (Sigma Aldrich, 99%) was dissolved in Milli-Q™ water. Then, this aqueous solution was used to prepare a water-in-oil microemulsion (ME) system. A water-in-oil ME is an optically transparent and thermodynamically stable solution which consists of aqueous nanodroplets stabilized by a layer of surfactant molecules [34]. Metal salts dissolved in the core of these nanodroplets can then be precipitated in form of particles in a nano-sized range [34]. The composition of the ME system is given in Table1. Once the ME was achieved, the alumina carrier was added. Then, NH3 was added to the mixture until achieving a pH 11. The mixture was kept under vigorous stirring for a period of 24h. The solids were then separated by centrifugation and cleaned with a 1:1 v/v solution of EtOH and CHCl3. The centrifugation-rinsing procedure was repeated a total of 3 times. Finally, the sample was dried at 70 °C for 24h and calcined at 500 °C (heating rate: 10 °C/min) for 4h in air. This support was named “ZrAl ME”. A third support was prepared by calcining the unmodified alumina carrier at 500 °C (heating rate: 10 °C/min) for 4h in air. This support was named “Al”. The three supports were prepared using the commercial γ-Al2O3 carrier Spheres Sasol® 1.8/210 crushed into a fine powder (mesh: 53–90μm). Table 1 Composition of the ME system Phase Compounds Composition (wt%) Oil (continuous phase) Isooctane 71 Surfactant Synperonic™ 91/5 (Polyoxyethylene C9-C11 alcohol) 22.9 Water (nanodroplets) Milli-Q™ water ZrO(NO3)2 5.53 0.57 1287Top Catal (2017) 60:1285–1298 1 3 The incipient wetness impregnation technique was used to deposit 12wt% of cobalt on the prepared supports. The supports were impregnated with an aqueous solution of Co(NO3)2·6H2O (ACS reagent, ≥98% Sigma Aldrich). Due to the large volume of water required to dissolve the salt, the aqueous solution was deposited batch wise, drying the samples at 100 °C for 2h after each deposition. A total of 3 batches were needed for complete deposition of the aqueous solution. Finally, the impregnated powders were dried at 120 °C for 3h (heating rate: 5 °C/min) and calcined at 300 °C (heating rate: 1 °C/min) for 16h in air. Thus, a total of 3 catalysts were prepared and named according to the names of the supports: “CoAl”, “CoZrAl P” and “CoZrAl ME”. 2.2 Catalyst Characterization 2.2.1 N2 Adsorption–Desorption N2 adsorption–desorption measurements were performed in a Micromeritics ASAP 2000/2010 unit in order to estimate the Brunauer–Emmett–Teller (BET) surface area and porosity of the catalysts and supports. The samples were evacuated and dried overnight at 250 °C prior to analysis. The BET surface area was estimated by adsorption at liquid nitrogen temperature at relative pressures between 0.06 and 0.2. The pore volume was estimated from a single point of adsorption at a relative pressure of 0.992. The mean pore diameter was estimated from the pore volume and the surface area assuming cylindrical pores. The pore size range was calculated with the Barrett–Joyner–Halenda (BJH) method using the desorption isotherm. 2.2.2 Inductively Coupled Plasma Mass Spectrometry The Zr-modified supports were analyzed by inductively coupled plasma mass spectrometry (ICP-MS) in order to determine the Zr loading. The analyses were performed by ALS Scandinavia. 2.2.3 Transmission andScanning Electron Microscopy Transmission electron microscopy (TEM) images presented on this paper were recorder on a PHILIPS CM-200 instrument with spot size of 15nm and resolution of 2.8Å. The samples were supported on a holey carbon-coated copper grid using an ethanoic suspension. The Digital Micrograph™ software from Gatan Inc. was used for treatment of the images. The surface of the supports was observed with a Hitachi® S-4800 SEM-FEG microscope (SEM) provided with secondary and retro-dispersed electron detector. An energy-dispersive X-ray spectroscopy (EDS) mapping was carried out by means of a Bruker X Flash Detector 4010, with resolution of 133eV. Compositional mapping measurements were carried out using an accelerating voltage of 15KV in order to evaluate the distribution of Zr and Al over the support surface. 2.2.4 Acidity Measurements Ammonia temperature-programmed desorption (NH3-TPD) analyses were performed in order to study the acidity of the supports. NH3-TPD measurements of ca. 0.2g of the support samples were performed in a Micromeritics AutoChem 2910 unit equipped with a thermal conductivity detector (TCD). Prior to analysis, the samples were pretreated at 500 °C (heating rate: 10 °C/min) for 1h in flowing He (30NmL/min). Afterwards, the samples were cooled down to room temperature and exposed to a flow 5% NH3 in He (30NmL/min) for 1h. Then, these were purged with He flow (100NmL/min) at 100 °C (heating rate: 10 °C) for 1h in order to remove physically adsorbed NH3. Finally, the samples were heated up in He (30NmL/min) from room temperature to 500 °C (heating rate: 10 °C/min) while monitoring the difference in thermal conductivity between the inlet and the outlet gas. The Brønsted and Lewis acidity of the catalysts was measured by infrared spectroscopy (ATI Mattson FTIR) using pyridine as a probe molecule. A thin (15–30mg) selfsupported wafer of the catalyst was pressed and then placed into the FTIR-cell. The cell was evacuated and the temperature was raised to 450 °C and kept for 1h. Thereafter, the temperature was decreased to 100 °C and the background spectra of the sample were recorded. Pyridine was adsorbed on the sample for 30min at 100 °C followed by desorption at 150, 250 and 350 °C for 1h and the spectra of the sample were recorded in between every temperature ramp. The scanning was performed under vacuum at 100 °C. Spectral bands at 1545cm−1 and at 1450cm−1 were used to identify Brønsted and Lewis acid sites, respectively. The quantitative amount of the acid sites was calculated with the constants of Emeis [35]. 2.2.5 Temperature‑Programmed Reduction The reducibility of the catalysts was studied by hydrogen temperature-programmed reduction (TPR) analyses in the same Micromeritics AutoChem 2910 unit. The analyses were performed by heating up the calcined samples up to 1000 °C (ramp: 10 °C/min) in a 5% H2 in Ar flow while monitoring the difference in thermal conductivity between the inlet and the outlet gas. The TCD signal was calibrated by complete reduction of a reference Ag2O powder in order to calculate the consumption of H2. 1288 Top Catal (2017) 60:1285–1298 1 3 The degree of reduction (DOR, %) was estimated by TPR after in situ reduction of the calcined catalysts at 350 °C (heating rate: 1 °C/min) for 16h in pure H2 flow. After reduction, the samples were flushed with inert gas for 30min at 350 °C. Afterwards, the abovementioned TPR procedure was performed in order to estimate the amount of H2 consumed when reducing the remaining cobalt oxides. The DOR was calculated assuming that the remaining cobalt oxides were in form of Co2+. 2.2.6 H2 andCO Chemisorption Static chemisorption measurements were performed on the fresh catalysts using both H2 and CO in order to estimate the cobalt dispersion (D, %) and the mean cobalt crystallite size [(dCo0)H, nm]. All the analyses were performed in a Micromeritics ASAP2020C unit after evacuation at 40 °C for 1h and insitu reduction at 350 °C (heating rate: 1 °C/ min) for 16h in pure H2 flow. After reduction, the samples were evacuated at 330 °C for 1h and for 30min at 100 °C prior to analysis. The method consists in obtaining an adsorption isotherm at 35 °C between 15 and 500mmHg. The H2 and CO uptakes were estimated by extrapolating the straight-line portion of the isotherm to zero pressure. After this analysis, the often called “reversible” and “irreversible” uptakes [26] were estimated by evacuating the samples at 10 μmHg for 1h and repeating this method to obtain a second adsorption isotherm. The reversible uptake was determined by extrapolating to zero pressure this second isotherm. The irreversible uptake was calculated by taking the difference between the total and the reversible uptake. The metal dispersion was calculated from the total H2 uptake assuming that the H:Co adsorption stoichiometry is 1:1 [26]. The d(Co0)H was estimated assuming spherical particles according to the following equation [36]: 2.2.7 X‑ray Diffraction X-ray diffraction (XRD) was performed on the calcined catalysts using a Siemens D5000 diffractometer with Cu-Kα radiation (2θ = 10°–90°, step size =0.02°) equipped with a Ni filter. The Co3O4 crystallite diameter [d(Co3O4)XRD, nm] was estimated by using the Scherrer equation and assuming spherical crystallites and applying a correction factor of 4/3 [37, 38]. The cobalt crystallite size [(dCo0)XRD, nm] was estimated from that of Co3O4 according to the relative molar volumes of Co0 and Co3O4 using the following formula [7]: d (Co0)H= 0.96 D × DOR d(Co0)XRD=0.75 ×d(Co3O4)XRD 2.3 Catalytic Testing The catalytic tests were performed in a down-flow stainless steel fixed bed reactor (i.d. 9mm). The reactor was heated by means of an electric oven and the temperature was regulated by cascade control with one sliding thermocouple inside a thermowell placed inside the reactor and another thermocouple placed in the oven. All the tests were performed with approximately 1.5g of dried catalyst samples diluted with ca. 10g of SiC (mean particle size: 75μm). This high catalyst dilution, together with an aluminum jacket placed between the reactor and the oven, allowed for an even temperature profile (210 ± 2 °C) along the catalyst bed. A catalyst particle size in the range of 53–90μm was used in order to suppress the influence of mass transfer on the catalyst performance [3, 8, 39, 40]. The samples were in situ reduced at 350 °C (heating rate =1 °C/min) for 16h in pure H2 flow [gas hourly space velocity =7.5NL/(h·gcatalyst)]. Afterwards, the reactor was cooled down to 180 °C and pressurized to 20bars with He flow. Then the He flow was replaced by syngas (containing 3% N2 as internal standard) and the temperature was increased to 195 °C (heating rate =0.5 °C/min) and subsequently to 210 °C (heating rate =0.2 °C/min). The tests were conducted at 210 °C, 20bar and using a H2/CO =2.1. The testing procedure consisted of four periods. In the first period (Period I) the catalysts were exposed to a GHSV of 10NL/(hgcatalyst) for approximately 24h. This first period served for determination of the catalyst initial activity and Site Time Yield (STY). In the second and third periods (Periods II and III) the GHSV was adjusted in order to expose the catalysts to 30 and 40% CO conversion, respectively. Each period had a duration of approximately 1day. Since the water partial pressure (and so, the CO conversion) has an effect on the FT product distribution [8, 41–43], these two periods served for a proper comparison of the selectivity of the catalysts. In the last period (Period IV), the GHSV was adjusted again to operate at ca. 50% CO conversion. These conditions were maintained for 3 days in order to compare the deactivation rate of the catalysts. The liquid hydrocarbons, waxes and water leaving the reactor were condensed in two consecutive traps at 140 °C and room temperature, respectively. The gases leaving the traps were depressurized and analyzed online with a gas chromatograph equipped with a TCD and a flame ionization detector (FID). H2, N2, CO, CH4 and CO2 were separated by a Carbosieve II packed column and analyzed on the TCD. C1-C6 products were separated by an aluminaplot column and quantified on the FID. The selectivity to C5+ products (SC5+) was used to represent the selectivity to long chain hydrocarbons. The S C5+ was calculated as follows: 1289Top Catal (2017) 60:1285–1298 1 3 3 Results andDiscussion 3.1 Characterization ofSupports andCatalysts 3.1.1 Surface Area andPorosity Measurements The BET surface area, pore volume and average pore diameter as estimated by N2 adsorption–desorption S C5+=100− ( SC1+SC2+SC3+SC4+SCO 2) measurements are presented in Table2. All the samples are characterized by a relatively high surface area and pore volume. It can be observed that the addition of ZrO2 increased the BET surface area, especially for the carrier prepared by ME. This enhancement may be due to added surface crystalline defects or by an enhancement of the thermal resistance of the carrier during calcination. As can be deduced, the mesopore volume (pores in the range of 1–50nm) constitutes between 80 and 90% of the total pore volume. Thus, the materials present mainly mesopores and possibly, in a minor extent, macropores. The mesopore size distribution of both supports and catalysts is presented in Fig. 1. As can be observed, all the samples present a very similar pore size distribution. The majority of the pores have a diameter in the range of 5–10nm. Moreover, the volume of these pores decreased after addition of cobalt, indicating that the metal particles were effectively deposited inside the carrier’s mesoporous structure. 3.1.2 Physicochemical Properties oftheSupports The surface of the Zr-modified carriers was analyzed with SEM in order to study the distribution of ZrO2 over the alumina carrier. Representative images of these two supports are presented in Fig.2. The SEM images reveal that Zr is heterogeneously distributed on the ZrAl P support, forming large islands rich in ZrO2 and poor in Al2O3. Contrariwise, the support prepared by ME shows a more homogeneous distribution. Further ICP-MS analyses of these two samples confirmed that both carriers contained 10wt% of Zr. These images demonstrate how the ME technique can be successfully applied to homogenously disperse oxide promoters over a mesoporous support. Table 2 Physical properties of supports and catalysts after calcination a Total pore volume as estimated by a single point of adsorption at a relative pressure of 0.992 b Mean pore diameter calculated according to: 4 · Pore volume/BET surface area c Mesopore volume (volume of pores in the range of 1–50nm) as estimated by using the BJH method Sample BET surface area (m2/g) Pore volume (mL/g)a Mean pore diameter (nm)b BJH mesopore volume (mL/g)c Supports Al 202 0.51 10.1 0.40 ZrAl P 215 0.52 9.7 0.45 ZrAl ME 241 0.51 8.5 0.46 Catalysts CoAl 164 0.39 9.5 0.32 CoZrAl P 176 0.37 8.4 0.32 CoZrAl ME 189 0.38 8.0 0.33 001011 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Pore volume (mL/g) Pore diameter (nm) CoAl CoZrAl P CoZrAl ME WO 001011 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Pore volume (mL/g) Pore diameter (nm) Al ZrAl P ZrAl ME WO Fig. 1 BJH pore size distribution of supports (left) and catalysts (right) 1290 Top Catal (2017) 60:1285–1298 1 3 In Fig.3 two representative TEM images of the ZrAl P and ZrAl ME samples are shown. As can be observed the supports presents high concentration of crystalline zones. The average interplanar spacing of crystals found in ZrAl P and ZrAl ME were respectively 4.62 and 4.68Å which correspond to {111} γ-Al2O3 [44]. In agreement with Rozita etal. [45], the γ-Al2O3 morphology corresponds to facetted and elongated crystals forming agglomerates. In that review, they also concluded that the cubeoctahedral morphology, also observed in our samples, is often dominant in γ-Al2O3 carriers probably due to a preferential growth of these {111} facets. In the case of the Al support such abundancy of these crystalline zones was not observed; fact that suggests that the incorporation of ZrO2 increases the support crystallinity. The NH3-TPD profiles are presented in Fig.4. The profiles suggest the existence of two main peaks which may be ascribed to two types of acid sites with different strength. The total acidity of the supports is similar. However, the results indicate that the addition of Zr increases the density Fig. 2 Representative SEM images of the ZrAl P (left) and ZrAl ME (right) support. The colors red and green indicate the presence of Al and Zr, respectively Fig. 3 Representative TEM images of: ZrAl P sample (left) and ZrAl ME sample (right). The arrows have been inserted to facilitate the identification of large crystallites 0100 200 300 400500 600700 80 09 NH 3 desorption (a.u./g) Temperature (°C) Al ZrAl P ZrAl ME Isothermal Fig. 4 NH3 temperature-programmed desorption profiles for the Al, ZrAl P and ZrAl ME supports 1291Top Catal (2017) 60:1285–1298 1 3 of weak acid sites (desorbing below 400 °C) and decreases the density of strong sites (desorbing above 400 °C). The profiles also suggest that the acidity of the ZrAl ME support is slightly higher than that of the ZrAl P. The acid strength distribution of the supports, as obtained by FTIR spectroscopy of adsorbed pyridine, is listed in Table3. The distribution of acid sites is fairly similar for all samples. No Brønsted acidity could be observed for any of the supports. The amount of Lewis acid sites of lower strength is slightly higher in the ZrAl ME support than in the ZrAl P, in line with the results obtained from NH3-TPD. Recent investigations have found that the support Lewis acidity plays an important role in cobalt-catalyzed FTS [46, 47]. Prieto etal. [46] found that the TOF and the selectivity to long chain hydrocarbons increases with increasing acidity of the oxide support. However, excessive acidity leads to a partial coverage of the cobalt particles by partially reduced oxide support species. This interplay between increasing the TOF and reducing the number of cobalt active sites results in a volcano dependence of the overall reaction rate with the carrier acid-base character. According to Johnson and Bell [47], this promoting effect of Lewis acid sites may be ascribed to their interaction with the O atom of adsorbed CO, which facilitates CO adsorption and dissociation. Unfortunately, the addition of Zr nor its distribution appear to enhance significantly the surface acidity of alumina. In this sense, no important changes in catalyst activity and selectivity should be expected. 3.1.3 Physicochemical Properties oftheCatalysts The temperature-programmed reduction (TPR) profiles of the three catalysts are presented in Fig.5. These profiles reveal the existence of various peaks. The first peak, situated at a temperature around 300 °C, is ascribed to the reduction of Co3+ to Co2+ (i.e. Co3O4 to CoO) [8, 12, 14, 48]. The second peak, situated at ca. 400 °C, corresponds to the reduction of big CoO crystallites to metallic cobalt. The broad H2 consumption area occurring above 600 °C is ascribed to the reduction of small CoO crystallites and cobalt-aluminate species [8, 12, 23, 49]. Two main conclusions can be derived from these TPR profiles. Firstly, the amount of Co3O4 in the catalysts is different. The results suggest that the addition of Zr increases the amount of Co3+ after calcination. This could be associated to a lower formation of CoAl2O4. Nevertheless, this does not necessarily imply that Zr improves the catalyst reducibility since cobalt aluminates can form during reduction. Secondly, the addition of Zr does not shift the reduction peaks to lower temperatures. This second observation suggests, in contrast, that Zr does not improve the reducibility of the catalyst. For instance the use of Pt, which is widely accepted to improve the reducibility, moves the peaks approximately 50–100 °C to lower temperatures [12, 14, 50]. The estimated degrees of reduction (DOR) after 16h at 350 °C in pure H2 flow are presented in Table4. The DOR’s are low (around 40%) and similar for the three catalysts. These results were expectable since the TPR profiles indicate that most of the Co2+ reduces at a temperature above 600 °C. These small differences in the DOR number may be ascribed to the partial reduction of cobalt species reducing slightly above 600 °C. Indeed, the TPR profiles present slight changes in the region between 600 and 700 °C, which may be in line with the observed DOR’s. Nevertheless, it cannot be discarded that such small differences are simply a result of instrument and procedure errors. In conclusion, our results suggest that Zr has no or almost no effect on the catalyst reducibility. The observed differences in reducibility are rather small and it is not clear whether these can be assigned to a Zr effect or to slight differences in Co particle size distribution. In any case, this possible improving effect of Zr appears to be very small and not comparable to that of noble metal promoters. These results may be somewhat surprising since it has been found Table 3 Distribution of Lewis acid sites in the supports as determined by FTIR spectroscopy of adsorbed pyridine Sample Lewis acid sites (µmol/g) 150 °C 250 °C 350 °C Al 116 54 15 ZrAl P 101 36 10 ZrAl ME 130 45 4 0200 400 600 8001000 1200 14 47% 33% 20% 45% 33% 22% 55% 33% mol H 2 consumed / g Temperature (°C) CoAl CoZrAl P CoZrAl ME Isothermal → 12% Fig. 5 Temperature-programmed reduction profiles of the three catalysts in 5% H2 in Ar flow 1292 Top Catal (2017) 60:1285–1298 1 3 that Co/ZrO2 catalysts present better reducibility than Co/Al2O3 catalysts [51]. A possible explanation to these results is that the surface of these Zr-modified carriers was not completely covered by ZrO2, as observed in the SEM images. The presence of superficial alumina may still lead to the formation of hardly reducible cobalt-support compounds. Therefore, we suggest that Zr may only improve the reducibility when the support is practically or totally coated with ZrO2 and thus, the catalyst behaves nearly as a Co/ZrO2 catalyst. This explanation may also explain the slightly higher reducibility of the CoZrAl P catalyst whose surface presented large islands rich in Zr. The X-ray diffractograms of the supports and the catalysts are presented in Fig.6. As can be observed, by comparing the supports, a ZrO2 peak appears at ca. 32° when ZrO2 is added to the alumina carrier. The small size of this peak suggests that ZrO2 did not form big agglomerates in any of the supports. Moreover, it can also be deduced by comparing the most intense γ-Al2O3 peaks {400} and {440} that these peaks become larger with the addition of ZrO2, especially for the ZrAl P support. The Co3O4 and Co mean particle sizes as estimated by X-ray diffraction are presented in Table 4. The H2 and CO chemisorption results are presented in Table5. As can be observed, the metal dispersion is similar but somewhat higher for the Zr-promoted catalysts. As a result, the mean Co particle size of these two catalysts is smaller than that of the CoAl. The results may be in Table 4 Catalyst physicochemical properties a Degree of reduction after 16h at 350 °C in H2 flow as estimated by temperature-programmed reduction b Mean Co3O4 particle size as estimated by X-ray diffraction using the isolated peak situated between 58° and 62° c Mean Co particle size as estimated from the Co3O4 mean particle size Catalyst DOR (%)ad(Co3O4)XRD (nm)bd(Co0)XRD (nm)c CoAl 42 13 10 CoZrAl P 44 15 11 CoZrAl ME 40 14 10 30 35 40 45 50 55 60 65 70 75 80 γ−Al 2 O 3 (222) Co 3 O 4 (400) Co 3 O 4 (440) Co 3 O 4 (511) Co 3 O 4 (220) Co 3 O 4 (220) γ−Al 2 O 3 (440) γ−Al 2 O 3 (400) Normalized intensity (a.u ) 2θ (°) CoAl CoZrAl P CoZrAl ME 30 35 40 45 50 55 60 65 70 75 80 γ−Al 2 O 3 (222) γ−Al 2 O 3 (311) ZrO 2 (111) γ−Al 2 O 3 (440) γ−Al 2 O 3 (400) Normalized intensity (a.u ) 2θ (°) Al ZrAl P ZrAl ME Fig. 6 X-ray diffractograms of the supports (left) and catalysts (right) Table 5 H2 and CO chemisorption results after insitu reduction at 350 °C for 16h in H2 flow a Metal dispersion as estimated using the total H2 uptake and assuming a H:Co stoichiometry equal to 1:1 b Mean Co particle size as estimated using the DOR and the metal dispersion and assuming spherical particles c Relationship between the irreversible CO uptake and the number of Co0 active sites estimated by H2 chemisorption Catalyst Total H2 uptake (μmol/g) Irreversible H2 uptake (μmol/g) D (%)ad(Co0)H (nm)bTotal CO uptake (μmol/g) Irreversible CO uptake (μmol/g) CO/Cos c CoAl 48 27 4.7 9 64 42 0.44 CoZrAl P 51 27 5.0 8 71 44 0.43 CoZrAl ME 58 32 5.7 7 74 47 0.41 1293Top Catal (2017) 60:1285–1298 1 3 disagreement with the Co particle size obtained by XRD. According to the XRD results, the Zr-modified catalysts should have presented a similar mean Co particle size. The Co particle size for the CoAl catalyst is, nevertheless, in better agreement with that obtained by XRD. A possible explanation to this discrepancy between XRD and H2 chemisorption is that XRD is performed on calcined samples and that the calculation of the Co particle size assumes particle integrity during reduction. Another explanation could be the adsorption of H2 on the support. Enache etal. [51] found by means of H2 temperature-programmed desorption analyses that hydrogen can chemisorb on ZrO2. In that case, the metal dispersion obtained for the Zr-modified catalysts would be overestimated. It may also be wrong to assume that the mean Co particle size can be estimated using the molar volumes of Co3O4 and Co metallic, especially if only part of the cobalt in the catalyst is present as Co3O4 after calcination. Finally, it cannot be discarded that the differences in Co particle size are due to other intrinsic errors of these techniques (e.g. assuming spherical particles). In order to offer a better understanding of the chemisorption results, it was found convenient to determine the irreversible H2 and CO uptakes. Since it is improbable that the reversibility of chemisorbed H2 on Co is identical to that on ZrO2, the ratio between irreversible and total uptakes should be different for these catalysts. Nevertheless, this ratio is very similar for the three samples (between 0.53 and 0.56). Therefore, H2 probably does not chemisorb on these supports. Furthermore, the CO uptakes are similar but also slightly larger for the Zr-promoted catalysts. The catalysts may however present different CO:Co adsorption stoichiometry due to a different metal dispersion and/or metal-support interaction. According to Reuel and Bartholomew [26], the relation between the irreversible CO uptake and the number of Co active sites estimated by H2 chemisorption (named CO/Cos in Table5) is representative of the CO adsorption stoichiometry. A CO/Cos value of ca. 0.4, which corresponds to poorly dispersed cobalt [26], was found for the three catalysts. Therefore, the presented CO uptakes may also be illustrative of the number of Co active sites. On the basis of the information gathered, it can be concluded that the porosity, reducibility and metal dispersion of the catalysts are similar. The main difference between these catalysts remains on the presence and dispersion of ZrO2. As a result, these three catalysts appear to be a convenient set for studying the influence of ZrO2 on the activity and selectivity of Co/γ-Al2O3 FT catalysts. 3.2 Catalytic Tests 3.2.1 Activity andStability oftheCatalysts The CO conversion as function of time on stream is presented in Fig.7 for the three catalysts. As explained previously, the tests consisted of four periods. In Period I, the catalysts were tested at a GHSV of 10 NL/h-gcatalyst in order to compare their activity. In Periods II, III and IV the GHSV was adjusted in order to operate at CO conversions of ca. 30, 40 and 50%, respectively. The GHSV employed in each period and the corresponding average CO conversions are presented in Table6. As can be deduced by comparing the CO conversions in the first period, the catalyst activity decreases in the order CoZrAl P > CoAl > CoZrAl ME. The initial activity (time on stream =5h) was further used for estimation of the turnover frequencies. The Site Time Yield (STY’s), based on the metal dispersion estimated by H2 chemisorption, are presented in Fig. 8. As can be seen, the STY of the catalysts also decreases in the order CoZrAl P > CoAl > CoZrAl ME; fact that could be expected since the catalysts presented similar cobalt dispersion. As explained previously, Period IV served for comparison of the catalyst deactivation rates. The results in Period IV suggest that Zr does not have any significant influence on catalyst stability (see Fig.7). Nevertheless, the CoAl catalyst exhibited a more stable performance than the other catalysts in Period I. The addition of ZrO2 may therefore enhance the deactivation rate at the beginning of the reaction. This observation is in line with the work of Ma etal. [24] who also found faster deactivation rates for Zr-modified catalysts, especially in the first 24h. 020406080100 120140 0 10 20 30 40 50 60 70 80 Period IV Period III Period II CoAl CoZrAl P CoZrAl ME CO conversion (%) Time on stream (h) Period I Fig. 7 CO conversion as function of time on stream. Operating conditions: 20bar, 210 °C and H2/CO =2.1