459663 Developing a Quantitative Spatial Resolution of Deactivation Effects on FT Synthesis in a Microchannel Reactor When Operating with Ultra-Low Sulfur Levels

Tuesday, November 15, 2016: 8:30 AM
Franciscan A (Hilton San Francisco Union Square)
Soumitra Deshmukh1, Henning Becker2, Sara Kampfe1, Kenneth Cowen1, Daniele Leonarduzzi2, Laura Barrio2, Jay Pritchard2 and Heinz J. Robota1, (1)Velocys, Plain City, OH, (2)Velocys, Milton Park, United Kingdom

Developing a quantitative spatial resolution of deactivation effects on FT synthesis in a microchannel reactor when operating with ultra-low sulfur levels

S. Deshmukh1, H. Becker2, S. Kampfe1, K. Cowen1, D. Leonarduzzi2, L. Barrio2, J. Pritchard2,
H.J. Robota1

Velocys

Plain City, OH1 and Milton Park, UK2

Introduction

Fischer-Tropsch (FT) is receiving new attention for its ability to generate conventional fuel from stranded, or unconventional natural gas, and alternative renewable biomass sources. Being a catalytic process, the catalyst is subject to deactivation via multiple mechanisms [1-2]. Sulfur is the foremost among the poisons causing permanent catalyst deactivation. Despite the impact of “S”, the number of published quantitative investigations related to practical conditions is quite modest.  While the recommendations for industrial syngas are typically <15 ppb S, [3-5] other publications would suggest unobservable impact at much higher exposures. [6,7]

In this paper, we present results of a detailed investigation of sulfur poisoning in a fixed bed microchannel reactor, with an emphasis on quantitative verification of delivered H2S concentrations. The poisoned catalyst was carefully extracted and analyzed to determine the spatial distribution of sulfur along the catalyst bed and within catalyst particles. This information is used to develop a quantitative model able to predict the impact in a commercially operating microchannel FT reactor.

Feeding H2S augmented syngas

The impact of contaminants, e.g. sulfur, on active Cobalt sites does not change with catalyst operating productivity. However, high catalyst productivity implies that the catalyst is exposed to a higher flux of these contaminants. With catalyst productivities exceeding 2000 v/v/h, the Velocys reactor passes up to 10-times the number of contaminant molecules per unit time compared to a conventional fixed bed reactor.

Detection and accurate quantification of syngas contaminants at low ppbv levels is challenging, as this approaches available equipment’s limit of detection. Nonetheless, it is critical to understanding the impact of poison concentration on catalyst deactivation rates. To accurately confirm the levels of sulfur delivered to our reactors, measurements were made using a Proton Transfer Reaction Mass Spectrometer (PTR-MS) at the point of delivery to the catalyst. [8] Figure 1 shows calibration of the instrument for the detection of ultra-low level H2S in syngas feed with the PTR-MS, indicating reliable measurements at <15 ppb detection levels.

A clean syngas feed was simulated by flowing desired amounts of pure CO, H2 and N2 streams from cylinders using mass flow controllers and blending to the target composition. Sulfur was introduced into the clean syngas by replacing part of the N2 feed with H2S-doped N2 feed.  All tubing and connections used in the contaminant flow path were treated with sulfur-resistant inert coatings to avoid transmission losses to the surfaces.

Figure  SEQ Figure \* ARABIC 1. PTR_MS calibration and response to ultra-low ppb level H2S

Microchannel fixed bed testing

The microchannel testing apparatus and methodology used to evaluate the catalyst and/or process performance has previously been described. [9,10] A microchannel reactor with 1 process channel and 2 adjacent coolant channels was loaded with a commercially manufactured, undiluted, particulate catalyst in the process channel as a packed bed. FT synthesis was performed by feeding clean and doped syngas as described earlier.

Figure 2 shows the impact of ultra-low levels of H2S in the syngas feed, 13 ppb in this case, on a catalyst operating under representative commercial conditions (H2:CO = 1.79, 28% inerts, 310 ms contact time). The deactivation rate, as measured by the decrease in CO conversion with time on stream, is higher even with such ultra-low levels of H2S present in the reactor feed.

Figure  SEQ Figure \* ARABIC 2. Decline in CO conversion with time during and after low level H2S poisoning.

In order to understand the quantitative impact of sulfur poisoning under practical extended operating conditions, an extended duration exposure of a catalyst to low level H2S in the feed was performed. As a trade-off between readily quantifiable impact and the required operating duration, the catalyst was exposed to 100 ppb H2S for a period of 300 hours, which is equivalent to 1 year exposure to H2S at a level of ~4 ppb.

Catalyst deactivation commensurate with expectations from the tests using lower levels of feed H2S was observed. Upon completion of the test, the reactor and catalyst were decommissioned to render the catalyst safe for handling and analysis.

Figure  SEQ Figure \* ARABIC 3. simulated long term exposure of catalyst to syngas operation in a laboratory single channel reactor.  The loss of CO conversion with time is readily visible during the 300 hour H2S exposure.

Spent catalyst analysis

Spent catalyst from the “1 year exposure” test was collected in aliquots of ~50 mg starting at the reactor inlet, with a spatial resolution of 1 cm. The samples were analyzed by ICP to quantify the sulfur accumulated in each section. Based on our prior experience with more heavily S-exposed catalyst charges, the top of the bed appears to be nearly saturated with sulfur (~5500 ppm)  with a concentration of S at approximately 4500 ppm. The sulfur content then decreases linearly to ~500 ppm at about 15% of the bed length. The subsequent axial sulfur profile is relatively, flat but there is still quantifiable sulfur content at 35% of the bed.

Figure  SEQ Figure \* ARABIC 4. Axial distribution of sulfur along the catalyst bed.

The recovered catalyst particles were characterized on a JEOL 6480LV SEM using a back-scattering detector and a beam voltage of 15 keV at Oxford Materials Characterization Service. A sample at 10% of the bed length (1875 ppm S) was first epoxy-embedded, polished, and coated with a 15 nm carbon thickness to see the internal topology and composition by EDX analyzer. At this point in the reactor, the examined catalyst granules had sulfur concentrated near the superficial surface and not detectable in their interiors. An example is shown in Figure 5 with the S:Co ratio as quantified by EDX at the locations noted in Figure 5 summarized in Table 1.

Model development

Quantifying the impact of spatially distributed poisons in the Velocys microchannel catalyst requires the use of a model which properly accounts for the effects of both reaction kinetics and mass transfer impact.  Such a model was developed using gPROMS. It accounts for the convective transport of mass and enthalpy along the catalyst bed and accounts for the diffusion of reactants into

Figure  SEQ Figure \* ARABIC 5. SEM/EDX analysis of catalyst particles, showing the areas of composition analysis.

Table 1.  EDX analysis of sulfur on catalyst particles LINK Excel.Sheet.12 "\\\\vmplab-dc01.vmplab.lan\\Data\\FT\\Post-operative analysis\\EXP-15-DK8435 - S poisoning (1 year equiv)\\SEM\\S_SEM_quanti.xlsx" "Sheet1!R4C1:R13C9" \a \f 5 \h  \* MERGEFORMAT

Molar ratio

Spectrum #

Co/Si

mmol S /mol Co

Spectrum 1

1.1

20.9

Spectrum 2

1.4

1.5

Spectrum 3

1.2

2.0

Spectrum 4

1.0

3.4

Spectrum 5

1.3

2.6

Spectrum 6

1.2

15.0

Spectrum 7

0.5

32.4

Spectrum 8

1.8

21.0

the pores of the catalyst particle. On top of this steady state model for the main FT reaction, a dynamic model describing the effects of deactivation and poisoning is superimposed. This allows for the prediction of the time dependent integral performance of the reactor. In addition to conversion and selectivities, it also enables one to quantitatively estimate the distribution of reactants, products, temperature, and catalyst activity. The reaction kinetics were modeled using a modified form of the Yates and Satterfield equation, with a factor, “F”, to account for the higher activity of the Velocys catalyst. [11-13] Deactivation was modelled as a result of the oxidation of metallic cobalt to inert cobalt oxide promoted by the presence of water as described in literature. [14] The poisoning was included by balancing the non-steady convective transport of H2S in the gas phase and non-steady diffusive transport inside the catalyst, where the sulfur adsorbs on the active cobalt sites. This leads to a combined deactivation due to oxidation and poisoning. An initial estimation of the deactivation parameters predicts a distribution of sulfur after poisoning with 100ppb for 300h consistent with the experimental observation and can simulate the distribution at for arbitrary entry concentrations and exposure times (Figure 6).

 

Figure  SEQ Figure \* ARABIC 6. Time evolution of the effective mass fraction of sulfur along the reactor axis (left) and mass fraction profiles in the intra-particle direction at 12% of the bed length (right).  Feed composition: 100 ppb H2S, H2:CO 1.79, 28% inert, 310 ms contact time, 205 °C.

The deactivation rate of 0.08% per day after sulfur has been switched off (Figure 2) was used to estimate the deactivation parameters assigned to cobalt oxidation. With this fitted parameter the loss in activity of 0.12% per day at 15 ppb sulfur was remarkably well predicted by the simulation with a value of 0.121% per day. This enables the predictive assessment of the effect of different levels of sulfur in the feed gas. Figure 7a shows how the CO conversion declines for different H2S feed levels after the same overall amount of sulfur has been fed to the catalyst bed. At the lowest sulfur level, a longer exposure time allows for the oxidation contribution to have the largest impact, yielding the lowest final conversion. This can also be seen in the final axial catalyst activity distribution shown in Figure 7b. Poisoning accounts for the loss of activity close to the inlet, which is similar for all H2S levels at equivalent exposure. Due to the longer run time at lower feed H2S levels, the downstream part of the bed, where the highest water partial pressures appear, is affected to the greatest extent by oxidation.

 

Figure  SEQ Figure \* ARABIC 7. Conversion as a function of time for different feed sulfur levels with total delivered sulfur equivalent to 100 ppb for 100 hours (left).  Final axial distribution of the relative catalyst effectiveness for each of the different exposure conditions.

Conclusions

Velocys has used a high sensitivity analytical method to accurately quantify the delivery of low ppb S levels in syngas feed to charges of FT synthesis catalyst in microchannel reactors. With this method, experiments were performed using doped and undoped syngas feeds to quantify the impact of S poisoning on the catalyst deactivation rates. Analysis of spent catalyst samples revealed gradients in concentration of S along the axial bed length and in the radial catalyst granule direction. Models were developed and validated to enable predictive assessment of the S poisoning in a commercial reactor.

References

1.   Rytter, E. & Holmen, A. Deactivation and Regeneration of Commercial Type Fischer-Tropsch Co-Catalysts—A Mini-Review. Catalysts 5, 478–499 (2015).

2.   Argyle, M. D. & Bartholomew, C. H. Heterogeneous Catalyst Deactivation and Regeneration: A Review. Catalysts 145–269 (2015).

3.   Duvenhage, D. J. & Coville, N. J. Deactivation of a precipitated iron Fischer-Tropsch catalyst - A pilot plant study. Appl. Catal. A Gen. 298, 211–216 (2006).

4.   Kritzinger, J. A. The role of sulfur in commercial iron-based Fischer-Tropsch catalysis with focus on C2-product selectivity and yield. Catal. Today 71, 307–318 (2002).

5.   Ehrensperger, M. & Wintterlin, J. In situ scanning tunneling microscopy of the poisoning of a Co(0001) Fischer–Tropsch model catalyst by sulfur. J. Catal. 329, 49–56 (2015).

6.   Sparks, D.E, Jacobs, G., Gnanamani, M.K., Pendyala, V.R.R., Ma, W., Kang, J., Shafer, W.D., Keogh, R.A., Graham, U.M., Gao, P. & Davis, B.H.; Poisoning of cobalt catalyst used for Fischer-Tropsch synthesis, Catalysis Today 215, 67-72 (2013)

7.   Pansare, S.S. & Allison, J.D.; An investigation of the effect of ultra-low concentrations of sulfur on a Co/γ-Al2O3 Fischer-Tropsch synthesis catalyst, Applied Catalysis A: General 387, 224-230 (2010)

8.   Herbig, J., Gutmann, R., Winkler, K., Hansel, a., & Sprachmann, G.. Real-Time Monitoring of Trace Gas Concentrations in Syngas. Oil & Gas Science and Technology – Revue d’IFP Energies Nouvelles. (2013). http://doi.org/10.2516/ogst/2012083

9.   Deshmukh, S. R.; Tonkovich A. L. Y.; McDaniel, J. S.; et al., 2011 Enabling cellulosic diesel with microchannel technology. Biofuels 2(3): 315.

10. Deshmukh, S. R.; Tonkovich, A. L. Y.; Jarosch, K. T.; et al., 2010 Scale-up of microchannel reactors for Fischer-Tropsch synthesis. Ind. Eng. Chem. Res. 49: 10883

11. Yates, I. C. & Satterfield, C. N. Intrinsic Kinetics of the Fischer-tropsch Synthesis on a Cobalt Catalyst. Energy & Fuels 5, 168-173 (1991).

12. Kaiser, P., Pöhlmann, F., Jess, A. Intrinsic Effective Kinetics of Cobalt-Catalyzed Fischer-Tropsch Synthesis in View of a Power-to-Liquid Process Based on Renewable Energy. Chem. Eng. Technol. 37, 964-972 (2014).

13. Vervloet, D., Kapteijn, F., Nijehuis, J., van Ommen, J. R. Fischer-Tropsch reaction-diffusion in a cobalt catalyst particle: aspects of activity and selectivity for a variable chain growth probability. Catal. Sci. Technol. 2, 1221-1233 (2012).

14. Sadeqzadeh, M., Chambrey, S., Hong, J., Fongarland, P., Luck, F., Curulla-Ferre, D., Schweich, D., Bousquet, J., Khodakov A. Y. Effect of different reaction conditions on the deactivationof alumina-supported cobalt Fischer-Tropsch catalysts in a milli-fixed-bed reactor: Experiments and modeling. Ind. Eng. Chem. Res. 53, 6913-6922 (2014).


Extended Abstract: File Not Uploaded