12th Natural Gas Conversion Symposium
Deactivation of the Cobalt Fischer-Tropsch catalyst – a Kinetic Study
Ljubiša Gavrilovića, Jan Brandinb, Anders Holmena, Hilde J. Venvika, R. Myrstadc, Kumar R. Routc, Erling Ryttera, Magne Hillestada, Edd A. Blekkana,*
aNorwegian University of Science and Technology, Department of Chemical Engineering, Sem Sælands vei 4, 7491, Trondheim, Norway
bLinnæus University, Department of Built Environment and Energy Technology, 351 95 Växjö, Sweden
cSINTEF Materials and Chemistry, NO-7465 Trondheim, Norway
Biomass to liquid (BTL) via gasification and integrated Fischer-Tropsch synthesis (FTS) is an attractive process for production of green diesel and jet fuel. The first step involves biomass gasification to produce syngas (CO + H2)[1]. It was reported[2] that impurities in biomass feedstocks and partial gasification can lead to the contaminants (such as: tars, alkali, HCN, H2S, etc.) in the produced syngas, which might cause catalyst deactivation and downstream problems. Therefore, series of cleaning steps and novel process design must be applied to achieve the required purity of the gas before the FTS reactor[3]. Alkali salts (mainly potassium) are the dominant salts in the fly ash composition after biomass gasification. Due to poor plant design or operational upsets in the cleaning section, these alkali salts might be present in the effluent gas composition in the produced syngas[3].
In order to do a good and realistic design of the BtL plant, it is necessary to have a kinetic model that captures major variations of reaction rates and selectivity subject to changes in process parameters. There are numerous kinetic models available in literature predicting reaction rates of the FT and there is a diversity of different model structures, due to different catalyst type, support, promoter, pore size etc.[4]. The product distribution from the FT is often described by the Anderson-Schulz-Flory (ASF) distribution, where the distribution is defined by a single parameter. In order to perform a FT reactor design with realistic evaluation of the products produced, there is a need for a way to quantify the product distribution. Hydrocarbon selectivity depends on local conditions like temperature, pressure, species concentrations (CO and H2) and water produced. Therefore, a suitable growth model should include these variables. It is surprising that none of the models in literature have water partial pressure as a variable, which is found to have a significant positive impact on C5+ selectivity[5]. Recently, we have developed a chain growth model based on the data of Yang et al. [6](18 points), where effects of temperature, partial pressure of H2O, H2 and CO are taken into account. However, the model is not validated for a larger dataset covering the necessary range of variables, and also, the results from Yang et al.[6] do not include the effect of water and H2/CO variation. The aim of the current work is to provide a detailed kinetic study, including effects of pressures (also water), and temperature. It is established that oxidation of bulk cobalt metal does not occur under typical FTS conditions, however small metal particle properties may differ considerably from those of bulk cobalt metal. Thus, a deactivation mechanism by the FT water byproduct might be thermodynamically favorable[7]. The water could also facilitate mobility of alkali species to the specific active sites on the Co catalyst[8]. A kinetic model encompassing catalyst deactivation by alkali will be developed.
A 20%Co/0.5%Re/γAl2O3 Fischer-Tropsch catalyst was poisoned by potassium salt (KNO3) using the aerosol deposition technique, depositing up to 3500 ppm K as solid particles. Standard characterization techniques (H2 Chemisorption, BET, TPR) showed no significant differences between poisoned samples and the reference catalyst. The Fischer-Tropsch activity, investigated at industrially realistic conditions (210 °C, H2/CO = 2:1, 20 bar), showed a reduction in the site time yield (STY) with increasing potassium loading[8]. The selectivity towards heavier hydrocarbons (C5+) was slightly increased, while the CH4 selectivity was reduced with increasing potassium loading. In order to be able to predict the catalyst deactivation with potassium, the activity data were kinetically analyzed. The proposed simple (power rate law) kinetic model[9] developed for a supported cobalt catalyst is given in (1) :
(1)
In expression (1), 
(ml/h/gcat) is the rate
of CO consumption;
, 
(Pa) are the partial
pressures of reactants and k0 (ml/h/Pa0.5/gcat)
is the kinetic constant. To be able to predict the catalyst deactivation with
potassium, the expression (1) was modified with a term in the expression of the
kinetic constant depending on the K loading:
where K is the potassium concentration on the catalyst. Three simple empirical equations[8] were selected to determine the effect of potassium on the catalyst activity:
Kinetic model |
|
|
|
In this table K is the
potassium impurity loading expressed in mgK/kgcat, and 
is the kinetic constant
of the reference sample. The value of the parameters 
was estimated by means
of a linear regression of CO conversion data[8]
Figure 1. Linear regression of the deactivation rate equations
The results indicate that all the models fit the experimental results quite well, however the best fit was obtained with model (3), Figure 1. A hypothesis of selective K-poisoning of the Co FT catalyst is in the line with the strong effect of potassium on CO conversion. However, the number of data-points is low, so the apparent agreement does not constitute evidence of a mechanism, but the result can be used as a tool to predict the effect of potassium on the surface.
Further experiments are underway for a detailed kinetic study and kinetic model development including the deactivation behavior. The validated and fitted model can be implemented in a process simulation code thereby a realistic BtL plant can be designed in future.
[1] E. van Steen, M. Claeys, Chem. Eng. Technol. 31 (2008) 655–666.
[2] P.J. Woolcock, R.C. Brown, Biomass and Bioenergy 52 (2013) 54–84.
[3] H. Boerrigter, H.P. Calis, D.J. Slor, H. Bodenstaff, 2nd World Conf. Technol. Exhib. Biomass Energy, Ind. Clim. Prot. (2004) 10–14.
[4] M. Ostadi, E. Rytter, M. Hillestad, Chem. Eng. Res. Des. 114 (2016) 236–246.
[5] E. Rytter, A. Holmen, ACS Catal. 7 (2017) 5321–5328.
[6] J. Yang, S.B. Eiras, R. Myrstad, H.J. Venvik, P. Pfeifer, A. Holmen, in: Fischer-Tropsch Synth. Catal. Catal. Adv. Appl., 142 (2016) 223–242.
[7] C.E. Kliewer, S.L. Soled, G. Kiss, Catal. Today (2018) In Press, https://doi.org/10.1016/j.cattod.2018.05.021.
[8] L. Gavrilović, J. Brandin, A. Holmen, H.J. Venvik, R. Myrstad, E.A. Blekkan, Ind. Eng. Chem. Res. 57 (2018) 1935–1942.
[9] R. Zennaro, M. Tagliabue, C.H. Bartholomew, Catal. Today 58 (2000) 309–319.