Effect of Surface Modification by Chelating Agents on FischerâTropsch

Article pubs.acs.org/IECR

Effect of Surface Modification by Chelating Agents on Fischer− Tropsch Performance of Co/SiO2 Catalysts Ashish S. Bambal,†,‡,§ Edwin L. Kugler,†,‡ Todd H. Gardner,‡ and Dady B. Dadyburjor*,†,‡ †

Department of Chemical Engineering, West Virginia University, Morgantown, West Virginia, 26506 United States National Energy Technology Laboratory, U.S. Department of Energy, Morgantown, West Virginia, 26505 United States

‡

ABSTRACT: The silica support of a Co-based catalyst for Fischer−Tropsch (FT) synthesis was modified by the chelating agents (CAs) nitrilotriacetic acid (NTA) and ethylenediaminetetraacetic acid (EDTA). After the modification, characterization of the fresh and spent catalysts shows reduced crystallite sizes, a better-dispersed Co3O4 phase on the calcined samples, and increased metal dispersions for the reduced samples. The CA-modified catalysts display higher CO conversions, product yields, reaction rates, and rate constants. The improved FT performance of CA-modified catalysts is attributed to the formation of stable complexes with Co. The superior performance of the EDTA-modified catalyst in comparison to the NTA-modified catalyst is due to the higher affinity of the former for complex formation with Co ions.

1. INTRODUCTION Fischer−Tropsch (FT) synthesis has been recognized as one of the most promising technologies for the conversion of coal, natural gas, and biomass-derived syngas into liquid fuels and chemicals.1 Limited oil reserves, energy supply security concerns, carbon credits,1 pollution abatement laws, and, most notably, uncertainty about fuel prices have increased the prospect of commercializing the FT process. Catalysts that are typically used for FT synthesis include supported Co or Fe. Cobased catalysts have the advantage of higher syngas conversion, more high-molecular-weight products (higher alpha numbers), and greater liquid product selectivity at lower operating temperatures compared to Fe-based catalysts. However, the performance of Co catalysts, activity, alpha number, product selectivity, and stability is affected by sintering and poor dispersion of active phases.2 Therefore, efforts are ongoing to optimize the performance of Co catalysts for commercial applications. It is axiomatic2 that the performance of FT catalysts is a strong function of the number of active sites on the catalyst surface and the crystallite size. Therefore, the study and control of such properties become crucial in the improvement of catalyst performance. Co catalysts are generally synthesized by impregnation of a metal precursor on an oxide support (e.g., SiO2 and Al2O3) followed by heat treatments, which include drying, calcination and reduction. Hence, the overall catalyst synthesis procedure plays an important role in determining the properties of a Co catalyst. The addition of a noble-metal promoter2 (e.g., Ru, Rh, or Pt) and the surface treatment of the oxide support3,4 are among the widely researched approaches that can be used to engineer the properties of FT catalysts. However, the performance enhancement due to the addition of a metal promoter seems to be more pronounced for Fe-based catalysts. On the other hand, surface treatments of the support, e.g., organic modifications of the support, are thought to be more effective in the synthesis of improved Co-based catalysts. The use of various chelating agents (CA) has been reported 5−7 to be a useful approach to improve the © 2013 American Chemical Society

performance of Co-based FT catalysts. Early CA-modified catalysts3 show a notable enhancement in FT performance, with an improved selectivity for C10−C20 hydrocarbons. These improvements were found for catalysts prepared by coimpregnation of metal precursor and CAs. However, the metal loading in the early work was limited to 5 wt %, due to difficulties associated with flow inside the pores of the support caused by the increased viscosity of the impregnating solution at higher metal loadings. It is desirable to investigate the effectiveness of the CA for conventional Co loadings of around 20 wt %. While Koizumi and co-workers5−7 have recently reported the effect of CA addition on the conversion and overall hydrocarbon selectivity over Co-based FT catalysts prepared by stepwise impregnation with 20 wt % metal loading, the effect of CA modification on individual product yields and distribution of hydrocarbon selectivities has not been reported to date. The present work aims to quantify in detail the changes in the FT performance and stability of Co/SiO2 catalysts when CAs are used. A cobalt loading of 20 wt % is used, consistent with the conventional loading. The silica support is modified with one of two chelating agents, nitrilotriacetic acid (NTA) and ethylenediaminetetraacetic acid (EDTA), before impregnation of the cobalt salt. The activity of the base catalyst and the modified catalysts are evaluated under FT-relevant conditions in a fixed-bed reactor. The catalysts are compared based on the conversion of syngas and selectivities of individual products. In addition, in order to establish the structure− activity relationship, the catalysts are characterized before and after the reaction. Received: Revised: Accepted: Published: 16675

June 21, 2013 October 1, 2013 October 28, 2013 November 14, 2013 dx.doi.org/10.1021/ie4019676 | Ind. Eng. Chem. Res. 2013, 52, 16675−16688

Industrial & Engineering Chemistry Research

Article

Figure 1. Fixed-bed reactor setup to perform Fischer−Tropsch synthesis.

samples were dried again at the same conditions, 110 °C for 12 h. The catalyst samples in this study were denoted as Co/X/ SiO2, representing modification of Co/SiO2 by a chelating agent X. 2.1.1. Calcination. The catalysts prepared as above were then calcined in a muffle furnace. Calcination decomposes the nitrate precursor into the oxide phase. Nitrate decomposition is an exothermic reaction, and hence phase transformations during calcination are accompanied by the release of heat. Therefore, both the final calcination temperature and the rate of decomposition influence the properties of the cobalt catalysts. However, the available furnace setup did not allow control of the ramp rate, only of the desired set-point. Initially, when catalysts were subjected to calcination at 250 or 350 °C, the furnace was operated with the desired temperature setpoint. However, the resulting high rates of decomposition resulted in complete disintegration of the materials. In fact, the 20−40 mesh particles turned into fine powder after the calcination step. Therefore, more-effective releases of heat and mass from the catalyst pores were desired during the decomposition. Accordingly, a procedure was developed in which the final calcination temperature was reached with stepwise increments of temperature. In the absence of any control over the ramprate, the furnace temperature was manually increased by 30 °C every 30 min. Therefore, the desired calcination temperature of 250 °C was reached in ca. 250 min, and similarly 350 °C was achieved in ca. 350 min. After reaching the final desired value,

2. MATERIALS AND METHODS 2.1. Catalyst Preparation. Commercial technical-grade SiO2 support (SS61138) was received from Saint-Gobain Norpro. Before the synthesis step, SiO2 pellets were thoroughly washed with deionized water and dried overnight at 110 °C, in order to remove any soluble impurities on the support. The washing/drying step was followed by heating at 400 °C to remove adsorbed species from the surface. The pretreated silica was sieved to 20−40 mesh and was used as a standard SiO2 support for all the catalysts prepared in this study. Cobalt nitrate (Co(NO3)2.6H2O, 99 +% pure), NTA (C6H9NO6, 99 +% pure), and EDTA (C10H16N2O8, 99 +% pure) were obtained from ACROS Organics. Ammonium hydroxide (NH4OH, 99 +% pure) was obtained from Fisher Scientific. All of these chemicals were used without further purification. The catalysts investigated here were prepared by the incipient-wetness method.8 A Co loading of 20 wt % was used for all catalysts, consistent with industrial practice. In the stepwise synthesis, if the catalyst was to be modified, an aqueous solution of a chelating agent (NTA or EDTA) was first impregnated on the silica support. The molar ratio of the chelating agent to Co was selected as unity for all these catalysts. The pH of the aqueous solution was maintained at 5.5 by using aqueous NH4OH. The samples were subsequently dried at 110 °C for 12 h. In the next step, the solids were impregnated by an aqueous solution of cobalt nitrate in two steps, each step equivalent to 10 wt % of cobalt loading. The 16676

dx.doi.org/10.1021/ie4019676 | Ind. Eng. Chem. Res. 2013, 52, 16675−16688


Article

through PID loops. The computer control system acts to maintain the system parameters. Therefore, the operating flow rate, temperature, and pressure can be controlled from the computer. The electrical cabinet controls process shutdown, independent of the computer, to protect the process during any malfunction. The process is protected from overpressure by a relief valve and from overtemperature by temperature-safety switches. The settings on these protection devices are based on standard practices. The switches could automatically shut down the entire system, independent of a computer shut-down. LabVIEW software by National Instruments was chosen to perform the monitoring, data acquisition, and operator interface for the system. The software was configured for the control strategies, data acquisition, and process display to suit this application best. 2.3. Experimental Procedure. Prior to the FT reaction, the calcined catalyst (approximately 1000 mg) is reduced in situ by H2 at 400 °C, 1 atm, and 200 sccm for 12 h. At the end of the reduction step, the H2 is replaced by inert He to flush the lines, while cooling to 170 °C. The system is then pressurized under syngas to 20 bar. Thereafter, the temperature of the bed is slowly increased at 1 °C/min to the desired temperature. Slow temperature ramping is desired in order to avoid run-away situations due to the exothermic CO hydrogenation reaction. The nominal operating conditions for FTS used in the study were: H2/CO = 2 (mol ratio) at 220−230 °C, 20 bar, and GHSV of 9000 sccm/h/gcat for 72 h time on stream (TOS). Incondensable gases leaving the cold separator include He, H2, CO2, Ar, CH4, CO, and lower hydrocarbons. As mentioned earlier, these were analyzed every 0.5 h by an online GC. Liquid and wax phases which contain higher hydrocarbons and alcohols were collected every 24 h during the entire run and are analyzed by an offline GC. The liquid products recovered from the hot separator contain an oil layer and an aqueous layer. These are allowed to settle for 24 h, after which the top layer (oil) is removed using a single-channel pipet from Eppendorf. Oil-phase samples for the GC analysis are prepared by mixing 200 μL of the oil sample with 800 μL of CS2 as an internal standard. Aqueous-phase samples for the GC analysis are prepared with cyclohexanol as an internal standard and deionized water as a solvent. More details on the qualitative and quantitative product analysis are reported elsewhere.9 The following expressions were used to calculate the values of the performance parameters of the catalysts used:

the temperature was held constant for 4 h. The slow temperature rise during the calcination assists in efficient heat and mass release from the materials. Therefore, with the controlled calcination approach, the catalysts retained their structural characteristics without disintegration. 2.2. Reaction Equipment. The FT synthesis was carried out in a reaction system designed in-house and built by Altamira Instruments. The block diagram of the reactor system is shown in Figure 1. The system was designed to operate at up to 500 °C or 103 bar and contains four inlet gas lines for feeding CO, H2, N2, and He/Ar. The system uses Brooks 5850S mass-flow controllers (MFCs) to feed gases to the process manifold. The setup also includes a liquid delivery system consisting of a syringe pump followed by an evaporator and a volume expander. Heat tapes or trace heaters wrap the plumbing from the mixing point of the feeds to the reactor. A dial pressure gauge is located on each end of the reactor for constant visual display of the pressure. Additionally, pressure transducers, one each located before and after the reactor, are used to measure and record the pressure in the system. The reactor pressure is controlled using a Brooks 5866E pressure control valve (PCV). The unit was periodically pressure-leak tested to 24 bar, using either helium or argon and a bubble leakdetection fluid. The system includes a stainless-steel reactor tube with specifications 1″ O.D. × 0.7″ I.D. × 35″ L. A three-point thermocouple with points at 0″, 2″, and 4″ is inserted into the top of the reactor to read the temperature within the catalyst bed. A set of two three-way valves is used to bypass the reactor. The furnace is used to maintain the temperature of the reactor tube. Heat tapes or trace heaters also wrap the plumbing from the reactor exit to the liquid separation section. The separation section incorporates three stages. The first stage is the wax separator, with a volume of 500 mL, designed to operate at a temperature up to 200 °C. The wax separator condenses the high-molecular-weight waxes formed during the Fischer− Tropsch process. The second stage, the hot separator, has a volume of 1000 mL and is designed to operate up to 100 °C. Most of the aqueous phase, along with some liquid hydrocarbons, would condense in this separator. The final stage is the cold separator. For efficient condensation, the cold separator is operated with water as the cooling medium at a flow rate of 1− 2 L/min at 4 °C or colder. This cold separator serves to handle traces of any condensable products escaping from the first two condensers. Typically, during the experiment, the wax separator was operated at 190 °C, the hot separator at 40 °C, and the cold separator at 4 °C. A Perkin-Elmer GC unit (Claurus 500) is attached to the FT reactor setup for online analysis of the noncondensible products. The outlet gases are analyzed every 0.5 h, using a HaySep packed column with a dual thermal-conductivity detector (TCD) and a capillary column with a flame-ionization detector (FID). The condensed products are collected every 24 h during the entire run and analyzed off-line by a HP 3400 GC. The organic phase of the condensed product is analyzed with a capillary column and a FID, whereas the aqueous phase is analyzed by a Propak-Q packed column and a FID. More details are provided in section 2.3 below. The computer system accepts the process signals and maintains a digital log of all of the process variables. All process components are controlled through the computer by simple ON/OFF commands, by analog voltage signals, or

CO conversion (%) = (mol rate of CO consumption)/(mol rate of CO inlet) × 100

(1a)

H 2 conversion (%) = (mol rate of H 2 consumption)/(mol rate of H 2 inlet) × 100

(1b)

degree of reduction (DOR, %) = (metal Co)/(total Co) × 100

(2)

turnover frequency (TOF, s−1) = (mol rate of CO consumption) /(mol of surface Co0 atoms) 16677

(3a)



Article

4f7/2) = 83.8 eV. The binding energy (BE) of the reference C1s peak was 284.6 eV. 2.4.4. Temperature-Programmed Reduction. The reducibility of the supported cobalt oxides in the presence of H2 was evaluated by temperature-programmed reduction (TPR). The experiments were performed using a Micromeritics Autochem 2950 HP. The calibration was performed by the reduction of a reference Ag2O sample. Prior to the H2-TPR experiment, the sample (0.1 g, 100−200 μm) was first heated at 120 °C for 1 h to remove moisture. The sample was subsequently cooled to 50 °C. The heating and cooling steps were performed in the presence of inert Ar gas. In the next step, the temperature was ramped to 1000 °C at 5 °C/ min in the presence of 50 cc/min of a 10 vol % H2/Ar mixture. The product gases from the reactor outlet were passed through a condenser. A cold trap containing a mixture of 2-propanol and dry ice was used to control the condenser temperature. The condenser eliminated traces of water from the product gas. The dry gas from the condenser outlet was subsequently analyzed with a TCD. The H2 consumption was measured by comparing the thermal conductivity of the exit gas with that of a reference gas stream containing a 10 vol. % H2/Ar mixture. The degree of reduction (DOR) was calculated from eq 2 as the ratio of the amount of H2 consumed to the theoretical H2 required for complete reduction of Co3O4 to Co. 2.4.5. H2-Chemisorption. Hydrogen-adsorption isotherms allow estimation of the chemisorbed H2 and metal dispersion. The experiments were performed with a Micromeritics ASAP 2020 unit. Initially, the sample was loaded in a sample tube and evacuated while heating to 350 °C at 10 °C/min. This was followed by cooling to 100 °C. The sample was then reduced by increasing the temperature to 350 °C at 10 °C/min under H2 flow. The sample was subsequently evacuated and cooled to 35 °C. An adsorption isotherm was recorded at 35 °C in the pressure interval from 75 to 400 mmHg. The amount of chemisorbed H2 was determined by extrapolating the straightline portion of the isotherm to zero pressure. In order to calculate the metal dispersion (D), it was assumed that two cobalt sites were covered by one H2 molecule.2 2.5. Kinetic Modeling. The rate expressions proposed in the literature for FT over cobalt catalysts range from simple power-law expressions to complex Langmuir−Hinshelwood expressions.10 Zennaro et al.11 have proposed a kinetic model for a FT catalyst with a Co loading of 11.7 wt % from their work with a differential fixed-bed reactor over a wide range of operating conditions at 20 bar: 180−240 °C; H2/CO feed ratios of 1.0−4.0; space velocity between 3000 and 5000 h−1; partial pressures of H2 and CO from 4 to 14 and 2 to 6 bar, respectively; and TOF between 0.0071 and 0.095 s−1. These operating conditions are close to the conditions used in the present work. Hence, their power-law rate expression

total TOF (TT, s−1) = (mol rate of CO consumption)/(mol of total Co) (3b)

hydrocarbon or alcohol yield (gC/kgcat/h) = (hydrocarbon or alcohol production rate, gC/h) /(catalyst mass, kgcat)

(4)

selectivity of hydrocarbon or alcohol with carbon number of n (wt %) = (rate of hydrocarbon or alcohol produced with carbon no. of n) /(rate of total hydrocarbon or alcohol produced) × 100 (5)

As mentioned earlier, the liquid products were collected every 24 h during the 72 h TOS. The product selectivities were determined based on the liquid products collected between 24 and 48 h TOS. This assumes that the catalyst activity is stabilized after the first 24 h TOS and that deactivation could be significant after 48 h TOS. 2.4. Characterization Equipment and Procedure. 2.4.1. Surface Area and Pore Volume. The surface area and pore volume of the support and calcined catalysts were measured on a Micromeritics ASAP 2020 instrument. The sample (0.15 g) was evacuated at 400 °C for 4 h prior to measurement. In the next step, N2 adsorption and desorption isotherms were measured at liquid N2 temperature. The Brunauer−Emmett−Teller (BET) area was estimated by N2 adsorption at relative pressures (P/P0) between 0.06 and 0.2, where P and P0 were the measured and equilibrium pressures, respectively. The total pore volume was calculated from the amount of vapor adsorbed at a relative pressure close to 0.995. The N2 desorption branch was chosen for the pore size analysis. The pore size distribution was established by applying the Barrett−Joyner−Halenda (BJH) method. 2.4.2. X-ray Diffraction. X-ray diffraction (XRD) was used to obtain the crystallographic structure and chemical composition of solids. The XRD patterns of the support and the prepared catalysts were obtained using a PANalytical diffractometer (XPert PRO) with Cu Kα radiation (1.5418 Å) operated at 40 kV and 25 mA. The catalysts were crushed to fine powders prior to measurement. The samples were scanned from 2θ = 5 to 120° at a rate of 0.05°/s. The observed peaks were assigned by referring to the International Centre for Diffraction Data (ICDD) database. The phase identification of silica and cobalt oxide was performed by comparing with the standard diffraction patterns available in database PDF-00-0040379 and PDF-01−073−2750, respectively. 2.4.3. X-ray Photoelectron Spectroscopy. X-ray photoelectron spectroscopy (XPS) allows the chemical state of elements to be obtained. XPS spectra were recorded with a Physical Electronics model 590 equipped with a cylindricalmirror analyzer and a 15 kV X-ray source from Physical Electronics. A magnesium anode was used as the source of xradiation (Mg Kα: 1253.6 eV). The system was routinely operated within a pressure range from 10−9 to 10−8 Torr (1.3 × 10−7 to 1.3 × 10−6 Pa). The instrument was calibrated using the photoemission lines EB (Cu 2p3/2) = 932.4 eV and EB (Au

−0.24 −rCO = kpH0.74 pCO 2

(6)

was adopted in this study. Here, rCO is the rate of CO consumption (mol/h/gcat), pH2 and pCO are the partial pressures of reactants (Pa), and k is the rate constant (mol CO/h/Pa0.5/ gcat). The inlet molar ratio of H2 to CO M = 2.0

(7a)

is used in the present study. Further, based on the reaction stoichiometry: 16678



Article

Table 1. Physical Characteristics of SiO2 Support and As-Prepared SiO2-Supported Co Catalysts

BET analysisa catalyst

H2-TPRa

250 °C 350 °C

degree of reduction, DOR (%)

metal surface area (m2/g metal)

dispersion, D (%)

H2-chemisorptiona

surface area (m2/g)

pore volume (cm3/g)

pore diameter (nm)

224 163 187

0.93 0.67 0.73

12.4 12.5 11.6

15.1 13.1

17.7 16.0

89.0 78.4

7.5 16.2

1.1 2.4

170

0.65

11.7

13.8

11.0

60.1

16.2

4.7

SiO2 supportb Co/SiO2 Co/NTA/ SiO2c Co/EDTA/ SiO2c a

Crystallites size (nm) by XRD after calcination at

Catalysts calcined at 350 °C. bStandard pretreated support before impregnation. cCA/Co = 1.0.

(mol of H 2 reacted) = 2(mol of CO reacted)

impregnating metal precursor flows inside the pores of the support and grows with subsequent drying and calcination treatment. The isotherms for N2 adsorption for the SiO2 support and the supported cobalt catalysts are shown in Figure 2a, and the

(7b)

Accordingly, a design equation for a fixed-bed reactor in terms of the CO conversion, XCO, becomes υ0CCO,in

dXCO (1 − XCO) = 20.74kCCO,in 0.5 dWcat (1 + εXCO)

(8)

where υ0 is the volumetric flow rate (m /h), Wcat is the weight of the catalyst (g), and the expansion factor, ε, accounts for the change in total molar volume. The value of ε was determined by considering the paraffin formation reaction as a representative one, especially for cobalt catalysts. The average value of ε was calculated to be 3

(9)

ε = − 0.67

for carbon numbers (n) from 1 to 40. In the estimation of ε, the lower paraffins (n < 5) were assumed to be gaseous, and the higher paraffins (n > 5) were assumed to be liquid. It must be noted that ε is negative in the present case due to the reduction in the total molar volume during the FT synthesis. For a negative value of ε, eq 8 can be integrated to obtain exit conversions XCO,out for a fixed-bed reactor ⎛ ⎜1 − ⎜ ⎝

(1 − XCO,out)(1 + εXCO,out) +

(1 + ε) −ε

⎡ (1 + εX −ε(1 − XCO,out) CO,out) + ln⎢ ⎢ 1 + −ε ⎣ =

⎤⎞ ⎥⎟ ⎥⎟ ⎦⎠

20.74kWcat υ0CCO,in 0.5

(10)

This equation is used to calculate the rate constant in the Results.

Figure 2. (a) N2 adsorption isotherms and (b) pore size distribution (PSD) curves for a SiO2 support before impregnation and for SiO2 supported Co catalysts.

3. RESULTS 3.1. Characterization of Support and Fresh Catalyst. The BET surface area, pore volume and average pore diameter measured by N2-physisorption for the SiO2 support, Co/SiO2, Co/NTA/SiO2, and Co/EDTA/SiO2 catalysts are presented in columns 2−4 of Table 1. Impregnation of the silica support with cobalt precursor results in the reduction of surface area and pore volume. This trend is observed for the base-case catalyst and the CA-modified catalysts. On average, the three catalysts show a ca. 23% drop in surface area and a ca. 27% drop in pore volume after impregnation. The drop in surface area and pore volume can be attributed to the fact that the

pore size distribution (PSD) curves are shown in Figure 2b. The shapes of the isotherms for the support and all catalysts are qualitatively similar, as are the shapes of the PSD curves. This suggests that structural characteristics are retained even after the chelation step. The isotherms of the three catalysts do not reveal any hysteresis, suggesting that the majority of the active sites are accessible on these catalysts. Additionally, there is a considerable drop in the maximum of the PSD curves for the Co catalysts in comparison to the SiO2 support; however, the differences among the catalysts themselves are only marginal. 16679



Article

in the energy region of Co 2p, with spin−orbital splitting Co 2p3/2 and Co 2p1/2, are given in Figure 4. The decomposition of

The X-ray diffraction patterns of a silica support and the three supported cobalt catalysts for the concentration ratio of chelating agent to Co (i.e., CA/Co) of 1.0 are shown in Figure 3. Calcinations were carried out at 250 °C (top figure) or at

Figure 4. Comparison between the Co 2p XPS spectra of Co3O4 for the base catalyst and CA-modified catalysts.

the Co 2p spectrum shows the contributions from CoIII, CoII, and a CoII satellite for each of the Co 2p3/2 and Co 2p1/2 peaks. The binding energies (BEs) after peak decomposition, given in Table 2, are in good agreement with the reported values6,13−16 for all three supported cobalt catalysts. From Table 2, based on the comparison of the BEs of the standard Co3O4 and the BEs of the Co 2p spectra of three catalysts, Co3O4 is confirmed as the dominant cobalt phase present in the calcined catalysts for all three cobalt catalysts. In general, the higher intensity of the CoII satellite peak is an indication of the stronger interaction of the Co2+ with the silica surface.13,14,17 Such stronger interactions are more pronounced for the cobalt-silicate-like species (e.g., α-Co2SiO4) than for the cobalt oxide phases (e.g., Co3O4).17,18 Also, as noted from Table 2, the BEs of the CoIII peaks of both the Co 2p3/2 and Co 2p1/2 peaks are higher for α-Co2SiO4 phase than those for the Co3O4 phase. Therefore, the qualitative comparison of intensity of a satellite peak and BEs of the CoIII peaks can be used to evaluate the extent of metal−support interaction due to CA modification.5 For the NTA-modified catalyst, the BEs of both the CoIII peaks (780.1 and 795.2 eV) are similar to the corresponding values for the base catalyst (780.0 and 795.0 eV). However, in the case of the EDTA-modified catalyst, the corresponding values of the BEs are higher (781.1 and 796.0 eV). Further, the intensity of the CoII satellite peak for the EDTA-modified catalyst is higher compared to the base catalyst and the NTA-modified catalyst. The higher BE of the CoIII peak and the higher intensity for the CoII satellite peak are characteristics of cobalt-silicate-like phases (see BE values of αCo2SiO4 in Table 2). Therefore, such cobalt-silicate-like species are expected to be present on the surface of at least the EDTAcontaining catalyst, and perhaps less so for the NTA-modified and base catalysts. Recall that the Co3O4 species on the EDTAcontaining catalyst was earlier shown (by XRD) to have the smallest crystallite size at 350 °C. This is consistent with the XPS study, as the smaller crystallite could interact more easily with the support during the drying and calcination steps, resulting in the formation of cobalt-silicate-like structures.19,20 The values of the ratio of intensities of the Co 2p3/2 and Si 2p peaks (ICo/ISi) are also listed in Table 2. Higher values of ICo/ISi are characteristic of higher cobalt dispersion on the surface of a calcined cobalt catalyst.14,19 The values of ICo/ISi in

Figure 3. X-ray diffraction patterns: (a and e), SiO2 support; (b and f), calcined Co/SiO2; (c and g), calcined Co/NTA/SiO2; (d and h), calcined Co/EDTA/SiO2. Calcinations were at 250 °C (curves e−h) and 350 °C (curves a−d), for CA/Co = 1.

350 °C (bottom figure). In all cases, a broad diffraction peak, a characteristic of amorphous silica, is located at 2θ = 22°. Furthermore, after calcination at either temperature, the presence of the Co3O4 phase on Co/SiO2, Co/NTA/SiO2, and Co/EDTA/SiO2 catalysts is confirmed (curves b−d and f− h, respectively). The average particle sizes of the Co3O4 phases were calculated from the Scherrer equation using the full width at half-maximum (fwhm) of the most intense diffraction peak, (311) located at 2θ = 36.8°. The average crystallite sizes for the catalysts with and without CA addition are shown in columns 5 and 6 of Table 1. From Table 1, the CA-modified catalysts, whether calcined at 250 °C or at 350 °C, contain smaller Co3O4 crystallites than does the corresponding base catalyst (without any CA modification). At the higher calcination temperature, 350 °C, the Co3O4 crystallites are larger for the base-case and the NTAmodified catalysts than the corresponding values at 250 °C. On the other hand, the EDTA-modified catalyst shows smaller crystallites at the higher calcination temperature. Formation of larger crystallites at higher calcination temperatures has been previously reported12 for the Co/SiO2 catalyst. The base-case catalyst and CA-modified catalysts calcined at 350 °C were subjected to surface analysis by XPS. The spectra 16680



Article

Table 2. Binding Energies of Co 2p3/2 and Co 2p1/2 for the Co3O4 Phase of As-Prepared Cobalt Catalystsa binding energy Co 2p3/2 (eV) catalyst

CoIII

5

778.6 781.5 780.0 780.1 781.1

(Co3O4) (α-Co2SiO4)5 Co/SiO2 Co/NTA/SiO2b Co/EDTA/SiO2b a

binding energy Co 2p1/2 (eV)

CoII

CoII satellite

CoIII

CoII

CoII satellite

intensity ratio, ICo/ISi

781.4 781.6 782.0

788.0 786.1 787.4 787.6 787.9

794.1 797.9 795.0 795.2 796.0

796.5 796.6 797.5

801.5 801.6 802.7

0.6 1.5 1.7

Catalysts calcined at 350 °C. bCA/Co = 1.0.

SiO2 interaction). Finally, peak δ (at ca. 685 °C) corresponds to the reduction of an oxide with stronger interactions, such as the reduction of silicates or hydrosilicates. In fact, the most pronounced difference between the catalysts is observed in this region. For the EDTA-chelated cobalt, the presence of a sharp peak at 685 °C suggests the reduction of very small Co oxide particles which could be strongly interacting with the SiO2 support. Such a higher-temperature reduction peak is also observed for the Co/SiO2 and Co/NTA/SiO2 catalysts. However, the lower intensity of the peak in the latter two cases suggests a weaker interaction in comparison to the EDTA-modified samples. Based on the TPR results, catalyst reduction prior to FTS experiments was performed at 400 °C to ensure optimum reduction. Note this means that small particles in the pores of the EDTA-modified catalyst are not reduced to Co. The values of the degree of reduction (DOR) measured for three catalysts are shown in column 7 of Table 1. The base catalyst with larger crystallites shows higher reducibility in comparison to a lower reducibility measured for the CAmodified catalysts, even though they contain smaller crystallites. Higher DOR values for the base and NTA-modified catalyst are consistent with the TPR results. The metal surface area and the metal dispersion of the catalysts, as measured by H2-chemisorption, are presented in Table 1, columns 8 and 9. Both of the CA-modified catalysts display higher amounts of chemisorbed H2, and hence higher densities of active sites, compared to the base catalyst. Accordingly, the metal surface areas and dispersions of the NTA- and EDTA-modified catalysts are higher than those of the base catalyst. These findings are in agreement with the other characterization results in this study. 3.2. Catalytic Activity. The FT performances of the CAmodified cobalt catalysts were evaluated under industrially relevant conditions and compared with the base catalyst. Preliminary runs were devoted to obtaining the carbon balance after careful quantification of each component in the products. Typically, the closure on carbon balance was 5% for all runs performed here. Only after establishing a satisfactory carbon balance for each run were the product selectivities determined. 3.2.1. CO Conversions versus Time on Stream. The CO conversions for three catalysts are plotted versus TOS in Figure 6, for bed temperatures of 220 and 230 °C. For the base catalyst (Co/SiO2) at 230 °C, the CO conversion slowly increases and reaches 33%. Thereafter, the CO conversion shows a gradual drop and steadies at ca. 21%. The CA-modified cobalt catalysts also show similar time dependences of the conversion. The NTA-modified catalyst at 230 °C shows a slightly higher conversion than the base catalyst and steadies around 25% at the end of 72 h of TOS. The EDTA-modified catalyst, at the same temperature, shows an improved performance, with a nearly 2-fold increase in the CO

Table 2 are clearly higher for the CA-modified catalysts than that for the base-case catalyst. This further confirms that the surface of the CA-modified catalyst possesses a better-dispersed cobalt oxide phase, mainly due to the presence of smaller crystallites. In the nomenclature of Girardon et al.,19 there is a more-uniform repartition of the cobalt on the support surface. The temperature-programmed reduction (TPR) profiles of the catalysts are shown in Figure 5. The occurrence of multiple

Figure 5. Temperature-programmed reduction profiles of various cobalt catalysts (experimental conditions: rate, 5 °C/min; 10% H2/Ar mixture; total flow rate, 50 sccm; catalyst wt., 0.15 g).

reduction peaks indicates the presence of a number of reducible cobalt species. A sharp peak in TPR generally indicates a fast reduction process, whereas broad peaks are attributed to difficult-to-reduce phases. For the Co/SiO2 catalyst, two strong peaks are located at 229 and 291 °C and a low-intensity hightemperature peak at 665 °C. Similarly, the NTA-modified catalyst shows two peaks at 230 and 293 °C and a peak at 642 °C. However, the EDTA-modified sample shows a single weak peak at 170 °C, followed by a single strong peak at 215 °C and another strong peak at the higher temperature of 685 °C. In addition, for the Co/EDTA/SiO2 catalyst, a broad region of H2 consumption is observed between 300 and 500 °C. The two lower-temperature peaks (α at ca. 230 °C and β at ca. 291 °C) are assigned to a two-step reduction of Co3O4. The reduction is known16 to proceed via the formation of various CoO species according to Co(III)2 Co(II)O4 → Co(II)O → Co

(11)

where the loss of O species in each step is not explicitly indicated for simplicity. The broad region of H2 consumption between 300 and 500 °C (peak γ) could be ascribed to a very slow reduction of CoO species present inside the pores (i.e., reduction with a mild Co16681



Article

At the bed temperature of 220 °C, the CO conversions follow a similar trend: an initial increase in the CO conversion and thereafter a steady conversion. Further, the higher CO conversion for the NTA−modified catalyst compared to the base catalyst is consistent with the conversion measurements at 230 °C. Even at the lower operating temperature, the EDTAmodified catalyst shows the highest CO conversion (ca. 39%) in comparison to the NTA-modified and the base catalyst. The higher CO conversion for the CA-modified catalysts in comparison to that of the base catalyst indicates that larger numbers of active sites are available on the reduced CAmodified catalysts during FT synthesis. Therefore, for these catalysts, each containing 20 wt % Co, the CA modification improves the active site densities desirable for the CO hydrogenation reaction. This can also be supported by the higher metal dispersions measured on the CA-modified catalysts after the reduction (see Table 1), even though the DOR values are somewhat greater for the base catalyst. For instance, the EDTA-modified catalyst with the highest metal dispersion displays the highest CO conversions during FT synthesis. 3.2.2. Average Conversions and Hydrocarbon Selectivities. Figures 7(a) and (b) show the average CO and H2 conversions for the three catalysts during the period between 24 and 48 h TOS, along with the product selectivities of C1, C2−C5, C5+,

Figure 6. CO conversions vs TOS of the base catalyst and the CAmodified catalysts at 220 and 230 °C (experimental conditions: catalyst load 1 g, diluted with 1:27 wt/wt SiC; P = 20 bar; H2/CO = 2.0; GHSV = 9000 sccm/h/gcat).

conversions compared to the base catalyst. At the end of 72 h TOS, the CO conversions for the EDTA-modified catalyst are notably higher than those of the NTA-modified catalyst and the base catalyst.

Figure 7. Syngas conversions and hydrocarbon selectivities of base catalyst and CA-modified catalysts at (a) Tcat = 230 °C and (b) Tcat = 220 °C (experimental conditions: P = 20 bar; H2/CO = 2; GHSV = 9000 sccm/h/gcat). 16682



Article

and a diesel fraction C10−C20 at 230 and 220 °C, respectively. At 230 °C, the NTA- and EDTA-modified catalysts show an increase in the average CO conversion and the H2 conversion compared to values for the base catalyst. However, CH4 selectivity values are ca. 20% for all three catalysts. In addition, the overall selectivity values, ca. 42% toward the C2−C5 fraction and ca. 38% toward C5+, are nearly the same for all catalysts. However, selectivities toward the diesel fraction (i.e., C10−C20) show some changes, with the base catalyst displaying the highest value and Co/EDTA/SiO2 the lowest at both the reaction temperatures. When panels a and b in Figure 7 are compared, the average CO and H2 conversions drop with the decrease of the reaction temperature. At 220 °C, the CH4 selectivity shows a marginal increase for all three catalysts, compared to the corresponding values at 230 °C. The hydrocarbon selectivities are noticeably different at 220 than at 230 °C, with an increase in the selectivities for shorter-chain hydrocarbons at the lower temperature, accompanied by a decrease in the C5+ selectivities and the diesel fraction selectivities. 3.2.3. Product Yields. The yields of paraffins, olefins, and alcohols on a carbon basis, i.e., in gC/kgcat/h, are presented in Table 3. The total yield is the sum of all hydrocarbon and Table 3. Product Yields of Modified Catalystsa catalyst

Co/SiO2

temp, °C 220 CO conver., % 18.9 product yield, gC/kgcat/h paraffins 95.1 olefins 15.1 alcohols 4.4 total 114.6 a

Co/NTA/SiO2

Co/EDTA/SiO2

230 26.0

220 19.8

230 32.0

220 42.1

230 66.7

141.2 12.2 16.3 169.7

132.7 21.5 10.5 164.7

239.9 24.0 36.7 300.6

287.5 45.0 25.3 357.8

523.5 85.3 37.1 645.9

Figure 8. Paraffins selectivity as a function of carbon number over CAmodified catalysts at (a) Tcat = 230 °C and (b) Tcat = 220 °C (experimental conditions: P = 20 bar; H2/CO = 2; GHSV = 9000 sccm/h/gcat).

selectivities are noted for all three catalysts, and all display a secondary maximum around C7. Figure 9a,b illustrates the olefin selectivities as a function of carbon numbers for the three catalysts at 230 and 220 °C, respectively. At 230 °C, the base catalyst and the NTAmodified catalyst show increases in the olefin selectivity from C2 to C3 and thereafter show a drop in the selectivities with increasing carbon number till C6 (NTA) or C7 (base case). The EDTA-modified catalyst, however, shows a progressive drop from C2 to C6. For all three catalysts, after reaching the minimum, the olefin selectivities increase with carbon number before reaching a secondary maximum. Thereafter, the selectivity drops for higher carbon numbers. Similar to the distribution of paraffin selectivities, the secondary maximum at 230 °C for the base catalyst shifts from C12 to a lower carbon number of C10 for the NTA-modified catalyst and to C8 for the EDTA-modified catalyst. Both the CA-modified catalysts show better selectivities for the olefins in the range of C6 to C14, compared to the base-case catalyst. At 220 °C, the EDTA-modified catalyst shows increases only from C2 to C3, with a progressive drop till C6. The NTAmodified catalyst and the base catalyst, however, show a progressive drop from C2 till C6. The olefins selectivity trends for the higher carbon numbers are comparable for all the three catalysts, with a secondary maximum around C8. Alcohol selectivities as a function of carbon number are compared in Figure 10a,b for the three catalysts at 230 and 220 °C, respectively. The selectivities are qualitatively the same at both temperatures over all catalysts, showing an increasing trend from C1 to C2 and then dropping. Therefore, at both

Averaged values for TOS of 24−48 h.

oxygenate yields. Based on the noted difference in activity of the three catalysts, it is worthwhile to determine the overall product yields. For all three catalysts, the product yields at 220 °C are lower compared to those at 230 °C. At either reaction temperature, the total product yields of both the CA-modified catalysts are higher than those of the base catalyst, with the EDTA-modified catalyst showing the highest value. Both the CA-modified catalysts display higher yields of paraffins, olefins, and alcohols at both the temperatures. The higher product yields of the CA-modified catalysts are consistent with the higher conversions observed over these catalysts. Therefore, the modification of silica by a CA not only improves the CO conversions but also results in higher yields of the FT products with only slight changes in selectivities. 3.2.4. Distribution of Paraffin, Olefin, and Alcohol Selectivities. Paraffin selectivities as a function of carbon number over the three catalysts at 230 and 220 °C are presented in Figure 8a,b, respectively. At both temperatures, the maximum selectivity is at C1 for all three catalysts. The selectivity drops with increasing carbon numbers before showing an increase at around C6. The selectivity reaches a secondary maximum before showing a final decrease for the higher carbon numbers. The secondary maximum at 230 °C for the base catalyst is around C10, shifting to C9 for the NTAmodified catalyst and to C7 for the EDTA-modified catalyst. In contrast, at 220 °C, no significant changes in paraffin 16683



Article

Figure 9. Olefin selectivity as a function of carbon number over CAmodified catalysts at (a) Tcat = 230 °C and (b) Tcat = 220 °C (experimental conditions: P = 20 bar; H2/CO = 2; GHSV = 9000 sccm/h/gcat).

Figure 10. Alcohol selectivity as a function of carbon number over CA-modified catalysts at (a) Tcat = 230 °C and (b) Tcat = 220 °C (experimental conditions: P = 20 bar; H2/CO = 2; GHSV = 9000 sccm/h/gcat).

Table 4. Rate Constant and Reaction Rates of CA-Modified Catalystsa

temperatures, C2H5OH is the dominant alcohol in the aqueous products. The observed trends of the alcohol selectivities are consistent with those previously reported.21 3.3. Determination of Rate Constant and Turnover Frequency. For all three catalysts, the values of reaction rates and rate constants calculated using eqs 6, 7a, 9, and 10 are presented in columns 3 and 4 of Table 4. The highest value of the rate constant and reaction rate are found for the EDTAmodified catalyst, with the base catalyst having the lowest values. In fact, the value of k for the EDTA-modified catalyst is more than twice that of the base catalyst, further emphasizing the extent of enhancement in the overall FT rates due to CA modification. Finally, the values of TT and TOF, as defined in eq 3, for the three catalysts are shown in columns 5 and 6, respectively, of Table 4. The values of TT are comparable to generally reported11 values for the cobalt catalysts. As expected, the TT values track with the reaction rate (which is on the basis of the total weight of the catalyst): the highest TT value is noted for the EDTA-modified catalyst and the lowest one for the basecase catalyst. The selectivity toward the C10−20 hydrocarbons also tracks with TT. The TOF values (on the basis of reduced Co atoms on the crystallite surface) do not change significantly with CA pretreatment. This implies that the reaction mechanism is not appreciably changed by CA pretreatment. 3.4. Characterization of Spent Catalyst. Characterization of the spent catalyst provides information about the structural changes on the catalysts during the FT reaction. Accordingly, the FT catalysts from the runs at 230 °C were

catalyst

XCO, %

k (mol/h/Pa0.5/gcat)

rCO (mol/h/gcat)

TTb (s−1)

TOFb (s−1)

Co/SiO2 Co/NTA/ SiO2 Co/ EDTA/ SiO2

26.0 32.0

0.017 0.020

0.058 0.072

0.026 0.032

2.7 1.7

66.7

0.040

0.151

0.065

2.3

FT runs at 230 °C, TOS of 24−48 h. bTT calculations are on total Co basis; TOF on metal Co basis.

a

recovered from the reactor after 72 h TOS and physically separated from the diluent quartz chips. The spent catalysts were characterized by N2-physisorption and XPS without further processing. The measured pore size distributions, BET surface areas and pore volumes of the spent catalysts are presented in Table 5. Table 5. N2-Physisorption Results of the Spent Catalystsa catalyst Co/SiO2 Co/NTA/SiO2 Co/EDTA/ SiO2

surface area (m2/g)

pore volume (cm3/g)

pore diameter (nm)

134 149 135

0.53 0.57 0.54

12.0 11.9 12.6

a

Measurements carried out on spent catalysts without further processing; FT runs at 230 °C.

16684



Article

For the spent base catalyst and the spent CA-modified catalysts, the N2 adsorption isotherms and the pore size distribution are shown in Figure 11, panels a and b, respectively. The analogous

Figure 12. Comparison between the Co 2p XPS spectra of Co3O4 for the spent catalysts (experimental conditions: Tcat = 230 °C; P = 20 bar; H2/CO = 2; GHSV = 9000 sccm/h/gcat).

results for the fresh catalysts can be found in Table 2 and Figure 4. For the spent base catalyst, however, the lower intensity of the Co 2p results in very broad peaks, and peak decomposition to CoIII, CoII, and the CoII satellite is problematical. Therefore, the binding energies of Co 2p3/2 and Co 2p1/2 for the base catalyst are not tabulated in Table 6, only the intensity ratio (ICo/ISi). The intensity ratios (ICo/ISi) are lower for all spent catalysts than for the corresponding fresh ones, indicating masking of exposed surface sites by FT products. Interestingly, both the spent CA-modified catalysts show higher intensity ratios compared to the spent base catalyst. These ratios indicate first that there are fewer accessible surface sites available on all three catalysts after reaction; that is, that repartitioning of the Co is less uniform after reaction, in the nomenclature of Girardon et al.19 Further, there are more accessible sites for the CA-modified catalysts than in the base case; that is, the CA modifications lead to more uniform repartitioning of the Co. These findings are well supported by the higher FT activity observed at the end of 72 h for both of the CA-modified catalysts.

Figure 11. (a) Nitrogen adsorption (BET) isotherms and (b) pore size distribution curves (PSD) for the spent Co catalysts (experimental conditions: Tcat = 230 °C; P = 20 bar; H2/CO = 2; GHSV = 9000 sccm/h/gcat).

results for the fresh catalysts (before the FTS run) can be found in Table 1 and Figure 2. From the comparisons of Tables 1 and 5, a drop of ca. 20% in the surface area and the pore volume is evident during FT synthesis for all catalysts. This could be attributed to the heavy products formed during FT partially blocking the inaccessible portions of the pores of catalysts. However, the isotherms for the spent catalysts do not reveal any noticeable hysteresis; this trend is similar to the one noted for the fresh catalysts. Therefore, the reaction intermediates effectively access the majority of the active sites available on the catalyst during the FT reaction. In addition, all the spent catalysts display comparable values for the pore diameter even after 72 h under FT conditions. Therefore, the catalysts in the present work did not undergo any structural collapse. Instead, they retained their structural integrity during the hightemperature and high-pressure FT operation. The surfaces of the spent catalysts recovered from the FT run were analyzed using XPS. Now the heavy products formed during the FT reaction cover the surface of the catalyst and analysis of XPS over the spent catalysts is less straightforward than over the fresh catalysts. In any case, the intensities of the Co 2p peaks for all three catalysts can be compared in Figure 12. The binding energies of CoIII, CoII, and a CoII satellite after peak decomposition and the intensity ratios (ICo/ISi) are listed in Table 6, for the spent CA-modified catalysts. The analogous

4. DISCUSSION It is evident from the characterization study that the modification of a silica support by a chelating agent affects the structure of the cobalt catalysts. More precisely, the presence of the smaller crystallites and the better-dispersed Co3O4 on the calcined samples, leading to the higher metal dispersion for the reduced samples (more-uniform repartitioning), are the observed outcomes of CA modification. The role of CAs in altering the properties of Co catalysts can be explained based on the interaction between CA and Co precursors. It has been previously hypothesized,5 using a variety of characterization techniques, that the formation of Co−CA complexes mainly affects the phase transformations during the drying and the calcination steps. As mentioned earlier, the heat treatments are necessary to decompose the Co nitrate into the Co oxides (Co3O4 or CoO). During the drying step, the Co− CA complexes on the CA-modified catalysts spread over the surface of the support by forming coordination bonds with the surface ions present on the support. These Co−CA complexes on the dried sample, having bulky structures, are moderately stable toward the heat treatment during calcination.7 In the 16685



Article

Table 6. Binding Energies of Co 2p3/2 and Co 2p1/2 for the Co3O4 Phase of the Spent Catalystsa binding energy Co 2p3/2 (eV)

a

binding energy Co 2p1/2 (eV)

catalyst

CoIII

CoII

CoII satellite

CoIII

CoII

CoII satellite

ICo/ISi ratio

Co/SiO2 Co/NTA/SiO2 Co/EDTA/SiO2

780.7 780.6

782.3 782.9

787.1 787.9

795.9 795.8

797.4 797.8

801.5 801.9

0.3 0.8 0.7

Measurements carried out on spent catalysts without further processing; FT runs at 230 °C.

by the higher values of rate constants and reaction rates at same operating conditions. The trends noted for the hydrocarbon and alcohol selectivities can be explained based on the individual reaction mechanisms for the product formation. It is widely accepted that the formation of the hydrocarbons during the FT synthesis follows the surface carbide mechanism.10 The monomers of the carbide mechanism are methylene (−CH2−) species,10,23,24 which are assumed to originate from the interaction of the (dissociatively) adsorbed CO species and the (dissociatively or associatively) adsorbed H2 species. Next, the insertion of the monomer (−CH2−) into a growing alkyl species is believed to cause chain growth of the intermediates.25 Chain-growth termination can occur due to an elimination of hydrogen to form α-olefins or due to the addition of either methyl (−CH3) species or hydrogen to form paraffins.26 The highly active α-olefins can further readsorb on the catalytic sites and undergo secondary reactions (hydrogenation, reinsertion, hydrogenolysis, or isomerization) to form olefins and paraffins.27 The rate of readsorption and the secondary reactions of α-olefins are thought28 to be mainly responsible for the formation of long-chain hydrocarbons. Also, the diffusion rate of the long-chain α-olefin intermediates out of the pores is slower, which increases the probability for readsorption and thereby enhances the secondary reactions.29,30 Accordingly, the consumption of short-chain α-olefins leads to formation of the long-chain intermediates, which results in relatively high yields of long-chain paraffins and olefins. This explains the secondary maximum noticed for the paraffin and olefin selectivities at both the reaction temperatures in the present work. The mechanism for the formation of oxygenates differs from that for the hydrocarbons. For the formation of the primary alcohols, the chain termination is thought10,21 to occur by the addition of the adsorbed CO species and subsequent hydrogenation. However, the observed selectivity trends in the present work cannot be supported exclusively by the alcohol formation mechanism. The higher selectivity of the C2H5OH relative to the other alcohols may be as a result of this sort of termination of chains of three or more allyl species being less favored. In any case, under FT conditions, the C2H5OH has been reported21 to be the most stable species among the alcohols.

initial stage of calcination, the Co−nitrate species from the complexes decompose to form small Co−oxide clusters. Next, the CA species from the complexes slowly start decomposing, and while doing so they facilitate the dispersion of Co−oxides on the support. It has been speculated7 that these phase transformations during calcination can also result in the formation of cobalt−silicate-like structures, to some extent. These structures may act as anchoring sites for Co3O4 and can minimize the subsequent sintering of Co3O4, resulting in the improved dispersion with smaller crystallites. The formation of smaller crystallites and the existence of a better-dispersed Co3O4 phases on the CA-modified catalysts prepared in his study are consistent with this. However, the extent of alterations is different for the NTAand EDTA-modified catalysts. The EDTA-modified catalyst has the smallest crystallite size (11 nm) and the highest metal dispersion (4.7%) of all catalysts, while the NTA-modified catalyst has bigger crystallites (14 nm) and a lower metal dispersion (2.4%). These differences can be explained based on the values of the complex formation constant, Kco. This constant is a measure of the affinity of the corresponding CA for complex formation with Co ions. The reported5 value of this constant is log Kco,NTA = 10.4

(12a)

for NTA, and for EDTA the value is log Kco,EDTA = 16.3

(12b)

This quantifies that EDTA has the higher affinity for complex formation with Co ions. Therefore, larger numbers of Co−CA complexes are expected to be present on the EDTA-modified catalyst than on the NTA-modified catalyst. Hence, the extent of the alteration in the crystallite sizes and dispersions due to CA modification is greater for the EDTA-modified catalyst than for the NTA-modified one. Further, XPS data (such as the higher BEs of CoIII peaks and the greater intensity of the satellite peak) and TPR data (the higher intensity of the high-temperature peaks) confirm the increased presence of cobalt−silicate-like structures on the fresh EDTA-modified catalyst. These structures may contribute in altering the Co sizes and the dispersions. More details on the role of CA in the enhancement of FT activity are provided by Koizumi and co-workers.5,7,22 The effect of CA modification on the FT performance of the cobalt catalyst is explained based on the densities of active sites. Surface modification by CA forms a larger number of adsorption sites, as noted by the improved conversions. As the site density increases for the CA-modified catalysts, the increased probability for CO hydrogenation results in the higher syngas conversions and the higher yields of the product paraffins, olefins, and alcohols. From a kinetic point of view, the activity enhancement due to CA modification is well supported

5. CONCLUSIONS The effect of two CAs, NTA and EDTA, on the FT performance of Co-based catalyst was studied. Regarding the procedure, each CA was added to a silica support, and catalysts with a conventional loading of 20 wt % Co on silica were prepared by stepwise impregnation. The impregnating metal precursor flows inside the pores of the support and grows with subsequent drying and calcination. A slow temperature rise during calcination is crucial to minimize the disintegration of 16686



Article

Present Address

the material. In general, the CA modification was seen to alter the physical characteristics and FT activity and selectivity of the catalyst. For fresh catalysts, N2 isotherms show no hysteresis, suggesting that most active sites are accessible (at least to N2). XRD confirms the presence of Co3O4 in the base case and CA-modified catalysts. The CA-modified catalysts contain smaller crystallites of Co3O4 than the base case. XPS indicates that catalysts modified by EDTA contain more cobalt−silicatelike species on the surface than do those modified by NTA or the base case. The modification also appears to result in a morehighly dispersed Co-containing phase, i.e., a more uniform repartitioning of the Co. TPR results show that EDTA leads to a relatively weakly adsorbed cobalt oxide outside the pores (compared to the base case and modification by NTA) and small Co particles inside the pores interacting strongly with the SiO2 support. Values of DOR obtained from these characterization data are higher for the base case than for the CAmodified catalysts. The differences between the catalysts can be explained on the basis of values for the equilibrium constant for complex formation for the different CAs. Reaction results show that the EDTA-modified catalyst with the greatest value of dispersion D displays the highest CO conversion. However, the TOF values do not change significantly with CA pretreatment, implying that the reaction mechanism for each individual surface Co metal is not appreciably changed by the pretreatment, just the numbers of these species. Hydrocarbon selectivity values are little affected by CA modification, except that the C10−20 fraction is highest for the base case and lowest for the EDTA-modified catalyst. With increasing reaction temperature, the selectivities for shorter-chain HCs decrease and the C5+ and C10−20 fractions increase. The yields of FT products mirror the activity behaviors of the catalysts. At lower temperatures, the distributions of the paraffin selectivity are the same for all three catalysts: all display a secondary maximum around C7. At higher temperatures, however, the secondary maximum shifts to higher carbon numbers. This is because of readsorption and secondary reactions, consistent with the carbide mechanism of FT. For the spent catalysts, N2 physisorption reveals a drop in surface areas and pore volumes relative to the fresh catalysts. This is due to partial blocking of the pores by the heavy FT products. However, an absence of hysteresis in the isotherms of both fresh and spent catalysts suggests that the reaction intermediates effectively access the majority of the active sites during the FT reaction. Also, comparable pore diameters of the fresh and spent catalyst imply that all three catalysts retain their structural integrity. XPS of the spent catalysts shows a drop in peak intensity ratios in comparison to the corresponding fresh catalyst. This is likely due to masking of the cobalt sites by FT products. Nevertheless, the spent CA-modified catalysts show higher values of the peak intensity ratio than does the spent base catalyst. This confirms the availability of larger numbers of accessible sites and more-uniform repartition of cobalt on the CA-modified catalysts relative to the base-case catalysts.

■

§

UOP LLC., Des Plaines, IL.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was performed with partial support obtained under the National Energy Technology Laboratory, RDS Contract No. DE-AC26-04NT41817. The authors would like to thank James Poston and Gabriela Perhinschi for their assistance with the physical characterization.

■

REFERENCES

(1) Williams, R. H.; Larson, E. D.; Jin, H. Synthetic Fuels in a World with High Oil and Carbon Prices, 8th Intern. Conf. GHG Control Technologies; Trondheim, Norway, June 2006. (2) Iglesia, E.; Soled, S. L.; Fiato, R. A. Fischer−Tropsch Synthesis on Cobalt and Ruthenium - Metal Dispersion and Support Effects on Reaction-Rate and Selectivity. J. Catal. 1992, 137, 212. Iglesia, E.; Soled, S. L.; Fiato, R. A.; Via, G. H. Dispersion, Support, and Bimetallic Effects in Fischer−Tropsch Synthesis on Cobalt Catalysts. Nat. Gas Convers.-II 1994, 81, 433. (3) Mochizuki, T.; Hara, T.; Koizumi, N.; Yamada, M. Surface structure and Fischer−Tropsch synthesis activity of highly active Co/ SiO2 catalysts prepared from the impregnating solution modified with some chelating agents. Appl. Catal. A: Gen. 2007, 317, 97. (4) Wang, Y.; Hou, B.; Chen, J.; Jia, L.; Li, D.; Sun, Y. Ethylenediamine modified Co/SiO2 sol−gel catalysts for non-ASF FT synthesis of middle distillates. Catal. Commun. 2009, 10, 747. (5) Koizumi, N.; Mochizuki, T.; Yamada, M. Preparation of highly active catalysts for ultra-clean fuels. Catal. Today 2009, 141, 34−42. (6) Mochizuki, T.; Koizumi, N.; Hamabe, Y.; Hara, T.; Takizawa, H.; Yamada, M. Preparation of Highly Active Co/SiO2 Fischer−Tropsch Synthesis Catalyst with Chelating Agents: Effect of Chelating Agents on Structure of Co Species during Preparation Steps. J. Jpn. Petrol. Inst. 2007, 50, 262. (7) Koizumi, N.; Ibi, Y.; Hongo, D.; Hamabe, Y.; Suzuki, S.; Hayasaka, Y.; Shindo, T.; Yamada, M. Mechanistic aspects of the role of chelating agents in enhancing Fischer−Tropsch synthesis activity of Co/SiO2 catalyst: Importance of specific interaction of Co with chelate complex during calcination. J. Catal. 2012, 289, 151. (8) Perego, C.; Villa, P. Catalyst preparation methods. Catal. Today 1997, 34, 281. (9) Bambal, A. Study of the Effect of Surface Modification and Sulfur Impurities in Syngas on the Fischer−Tropsch Performance of Cobalt Catalysts. Ph.D. Dissertation; West Virginia University: Morgantown, WV, 2012. (10) Van der Laan, G. P.; Beenackers, A. Kinetics and selectivity of the Fischer−Tropsch synthesis: A literature review. Catal. Rev.: Sci. Eng. 1999, 41, 255. (11) Zennaro, R.; Tagliabue, M.; Bartholomew, C. H. Kinetics of Fischer−Tropsch synthesis on titania-supported cobalt. Catal. Today 2000, 58, 309. (12) Khodakov, A. Y. Enhancing Cobalt Dispersion in Supported Fischer−Tropsch Catalysts Via Controlled Decomposition of Cobalt Precursors. Br. J. Phys. 2009, 39, 171. (13) Ernst, B.; Hilaire, L.; Kiennemann, A. Effects of highly dispersed ceria addition on reducibility, activity and hydrocarbon chain growth of a Co/SiO2 Fischer−Tropsch catalyst. Catal. Today 1999, 50, 413. (14) Ernst, B.; Libs, S.; Chaumette, P.; Kiennemann, A. Preparation and characterization of Fischer−Tropsch active Co/SiO2 catalysts. Appl. Catal. A: Gen. 1999, 186, 145. (15) Castner, D. G.; Watson, P. R.; Chan, I. Y. X-ray absorption spectroscopy, x-ray photoelectron spectroscopy, and analytical electron microscopy studies of cobalt catalysts. 1. Characterization of calcined catalysts. J. Phy. Chem. 1989, 93, 3188.

AUTHOR INFORMATION

Corresponding Author

*Email: [email protected]. Phone: (304) 2939337. Fax: (304) 293-4139. 16687



Article

(16) Sexton, B. A.; Hughes, A. E.; Turney, T. W. An XPS and TPR study of the reduction of promoted cobalt-kieselguhr Fischer−Tropsch catalysts. J. Catal. 1986, 97, 390. (17) Puskas, I.; Fleisch, T. H.; Full, P. R.; Kaduk, J. A.; Marshall, C. L.; Meyers, B. L. Novel aspects of the physical chemistry of Co/SiO2 Fischer−Tropsch catalyst preparations: The chemistry of cobalt silicate formation during catalyst preparation or hydrogenation. Appl. Catal. A: Gen. 2006, 311, 146. (18) Dai, X. P.; Yu, C. C.; Li, R. J. Deactivation of CeO2-Promoted Co/SiO2 Fischer−Tropsch Catalysts. Chin. J. Catal. 2007, 28, 1047. (19) Girardon, J. S.; Quinet, E.; Griboval-Constant, A.; Chernavskii, P. A.; Gengembre, L.; Khodakov, A. Y. Cobalt dispersion, reducibility, and surface sites in promoted silica-supported Fischer−Tropsch catalysts. J. Catal. 2007, 248, 143. (20) Khodakov, A. Y.; Griboval-Constant, A.; Bechara, R.; Zholobenko, V. L. Pore Size Effects in Fischer−Tropsch Synthesis over Cobalt-Supported Mesoporous Silicas. J. Catal. 2002, 206, 230. (21) Gall, D.; Gibson, E. J.; Hall, C. C. The Distribution of Alcohols in the Products of the Fischer−Tropsch Synthesis. Appl. Chem. 1952, 2. (22) Mochizuki, T.; Hara, T.; Koizumi, N.; Yamada, M. Novel preparation method of highly active Co/SiO2 catalyst for Fischer− Tropsch synthesis with chelating agents. Catal. Lett. 2007, 113, 165. (23) Chaumette, P.; Courty, P.; Kiennemann, A.; Ernst, B. Higher Alcohol and Paraffin Synthesis on Cobalt-based Catalysts-Comparison of Mechanistic Aspects. Top. Catal. 1995, 2, 117. (24) Davis, B. H. Fischer−Tropsch synthesis: current mechanism and futuristic needs. Fuel Proc. Technol. 2001, 71, 157. (25) Lox, E. S.; Marin, G. B.; De Grave, E.; Bussière, P. Characterization of a promoted precipitated iron catalyst for Fischer−Tropsch synthesis. Appl. Catal. 1988, 40, 197. (26) Saxena, S. C. Bubble Column Reactors and Fischer−Tropsch Synthesis. Catal. Rev.: Sci Eng 1995, 37, 227. (27) Kuipers, E. W.; Vinkenburg, I. H.; Oosterbeek, H. Chain Length Dependence of α-Olefin Readsorption in Fischer−Tropsch Synthesis. J. Catal. 1995, 152, 137. (28) Dwyer, D. J.; Somorjai, G. A. The role of readsorption in determining the product distribution during CO hydrogenation over Fe single crystals. J. Catal. 1979, 56, 249. (29) Madon, R. J.; Reyes, S. C.; Iglesia, E. Primary and Secondary Reaction Pathways in Ruthenium-Catalyzed Hydrocarbon Synthesis. J. Phys. Chem. 1991, 95, 7795. (30) Madon, R. J.; Iglesia, E. Hydrogen and CO Intrapellet Diffusion Effects in Ruthenium-Catalyzed Hydrocarbon Synthesis. J. Catal. 1994, 149, 428.

16688


Effect of Surface Modification by Chelating Agents on FischerâTropsch

Recommend Documents

Effect of Surface Modification by Chelating Agents on FischerâTropsch