545625 Effect of Potassium on Highly Dispersed Iron Nanoparticles Support on Carbon for CO Hydrogenation

Monday, June 3, 2019: 5:42 PM
Texas Ballroom EF (Grand Hyatt San Antonio)
Joakim Tafjord, Department of Chemical Engineering, Norwegian University of Science and Technology, Trondheim, Norway


Effect of potassium on highly dispersed iron nanoparticles support on carbon for CO hydrogenation

Joakim Tafjord1, Rune Myrstad2, Anders Holmen1, Jia Yang1*

1Department of Chemical engineering, Norwegian University of Science and Technology, NO-7491, Trondheim, Norway
2SINTEF Materials and Chemistry, N-7463 Trondheim, Norway
* Corresponding author: jia.yang@ntnu.no

1. Introduction
Fischer-Tropsch synthesis is a versatile process in terms of hydrocarbon production. Cobalt and iron are the most commonly employed catalysts in industry, where the former is used for long-chain paraffin production and the latter for wax formation as well as olefin production, depending on operation temperature. Iron also has the added benefit of water-gas shift (WGS) activity, utilizing the H2O by-product to form additional H2, thus enabling the use of H2/CO lower than the stoichiometric number of the reaction. The role of the different iron phases present during reaction remains controversial, but the presence of iron carbides are clearly linked with high FTS activity. The bulk carbon from iron carbides might participate in a Mars-van Krevelen-like reaction, as suggested by both experimental work and DFT. [1,2

In this work we employ an iron-carbon based catalyst (Fe-C) derived from pyrolysis of iron-containing polymers, with a narrow size distribution of iron particles. The effect of potassium promotion on CO hydrogenation is investigated using Steady-State Isotopic Transient Kinetic Analysis (SSITKA).  

2. Experimental
The unpromoted catalyst was prepared by pyrolysis of iron-containing polymers, resulting in iron nanoparticles with a narrow size distribution in a carbon matrix. Potassium promotion was introduced by subsequent incipient wetness impregnation of unpromoted catalyst with an aqueous solution of KHCO3. The sample was dried, then pyrolyzed in N2 at 500°C, with a ramp rate of 2°/min and dwell for 4 hours.

The supported Fe nanoparticles were investigated by powder X-ray diffraction (XRD) (Bruker D8 Advance Da Vinci, Cu Kα, 20-80 2θ, V6 slit) and Transmission Electron Microscopy (200k eV, Tecnai). The apparent surface areas of the catalysts were measured by N2-physisorption (Tristar-3020) applying the BET-equation, while Fe and K loadings were determined by ICP-MS (ELEMENT 2).

SSITKA experiments were carried out in a fixed-bed quartz reactor (44 mm i.d.). 150 mg of the catalyst (90-253 μm) was mixed with 150 mg of inert silicon carbide of the same fraction, to improve heat transfer in the reactor. The catalyst was reduced in H2/Ar (10/10 mL/min) with a ramping rate of 3º min-1 to 350°C. After 2 hours of reduction, the temperature was reduced to 340°C, and syngas H2/CO/Ar (15/3/7 mL/min) was introduced along with a pressure increase to 1.85 bar. The reactants and products were recorded with an online gas chromatograph (GC) equipped with a TCD and FID. After 12 hours on stream, an isotopic switch from H2/12CO/Ar to H2/13CO/Kr was performed, and the isotopic transient response was recorded with a Balmer QMG-422 quadrupole mass spectrometer (MS). A 16-loop storage was also used during isotopic switching, and subsequently analyzed with a GC-MS. The isotopic distribution of C2-C4 hydrocarbons was calculated by linear combination of the fragmentation patterns of the corresponding isotopic products. [3]

3. Results and discussion
The loading of the base catalyst (Fe-C_0.0) was 30 wt. %, with iron nanoparticles (Fe3O4 + Fe) well dispersed (average size of 3 nm) in a porous carbon matrix (350 m2g-1). The ICP-MS results confirmed that the desired potassium loading was achieved. These results are summarized in Table 1.

Table 1
. Physical characterization

Sample

Iron phase

Fe particle size
[nm]

Surface area
[m2g-1]

Fe loading
[wt. %]

K loading
[wt. %]

Fe-C_0.0

Fe3O4 + Fe

3.05

353

30.0

0.0

Fe-C_0.2

Fe3O4 + Fe

-         

350

31.5

0.2

 

The CO hydrogenation results show a relatively low selectivity towards methane given a H2:CO ratio of 5.0 and 1.85 bar pressure, and also good selectivity towards lower olefins (25-27% of total carbon, 45-51% of total hydrocarbons) (Table 2.). The potassium promotion leads to slightly higher olefin selectivity, but also lowers overall activity. The CO2 selectivity is also higher for the promoted sample, indicating higher WGS rates.

Table 2. Selectivity at 340ºC, H2/CO ratio of 5 after reaching steady state (16 h), averaged over 8 analysis.

XCO
[%]

CH4
[%]

C2-4p
[%]

C2-4o
[%]

C5+
[%]

CO2
[%]

C2 o/p

C3 o/p

C4 o/p

Fe-C_0.0

7.4

16.0

7.9

24.9

8.8

44.5

1.9

5.7

6.5

Fe-C_0.2

6.2

15.7

7.5

26.9

5.2

47.1

2.2

6.4

6.5

The SSITKA technique is able to measure site coverage and residence time of the intermediates, and therefore it is very powerful technique to distinguish the contribution of the two on the overall reaction rate. The residence time and site coverages of CO, CHx, C2H4, C3H6 and C4H8 intermediates are shown in Table 3.

Table 3. Residence times and site coverages of unlabeled species when switching from 12CO to 13CO. CH4 and CO measurements were performed 4 times with MS, while C2-C4 results are from one GC-MS loop analysis. All residence times are corrected for the chromatographic effect of Ar, and hydrocarbons are corrected by ½τCO. A dispersion of 10% was assumed for both samples when calculating surface coverages. Standard errors for the last digit are listed in parenthesis where applicable.

τCO
[s]

τCHx
[s]

τC2o
[s]

τC3o
[s]

τC4o
[s]

θCO

θCHx

θC2o

θC3o

θC4o

Fe-C_0.0

1.5(3)

12.5(5)

87.0

77.3

90.7

0.040(7)

0.0043(2)

0.019

0.015

0.008

Fe-C_0.2

2.2(3)

11.8(3)

74.7

67.1

81.1

0.058(9)

0.0032(1)

0.014

0.011

0.006

The effect of potassium promotion has been debated; it has been suggested to act as an electronic promotor through donation of negative charge, or to cause blocking of active sites. The effect could be differentiated by comparing the effect of potassium on residence times and site coverages. The difference between the two samples show a greater relative difference in CHx and olefin site coverage than residence time. These results suggest that electronic changes relating to surface adsorption are small, while the sites coverages are more severely reduced. The dispersion is important in the calculation of site coverage, and here we have assumed it to be equal for both catalyst. If the additional pyrolysis step performed on the promoted sample influences particle size, the comparison of site coverages falls short.

When performing a switch from 12CO to 13CO, the iron carbides will gradually increase the amount of interstitial 13C over time. The process of exchanging one isotope of interstitial carbon with another is a much slower process than exchanging surface species. With a Mars-van Krevelen like mechanism, one would expect a lingering effect on the isotopic product distribution. Here we observe a slow depletion of isotopic products, as seen on the left in Figure 1 with C3-olefins for the unpromoted catalyst, where particularly 12C13C2H6 still have a relatively significant contribution until 4 minutes. In cobalt where the active phase is mainly metallic, one would expect the depletion to be much quicker. This is illustrated in Figure 1 with the Fe-C_0.0 to the left, and 20 wt.%Co 0.5 wt.%Re/ γ-Al2O3 from Ledesma et al. on the right. [3] From these results it would appear that a Mars-van Krevelen-like mechanism is present, but it isn’t apparent to what degree it contributes to the overall activity.

At CO hydrogenation conditions, the effect of 0.2 wt. % potassium promotion is not very significant.  To confirm the observed effect, higher loadings (0.6 wt. %, 1.0 wt. %) of potassium will be tested to observe greater differences on the effect of promotion. In addition, we plan to investigate the effect of the pyrolysis treatment with operando XRD and EXAFS to investigate the nature of the iron particles.

Figure 1. Isotopic distribution of C3-olefin products for the Fe-C_0.0 and a 20 wt.%Co 0.5 wt.%Re/ γ-Al2O3 catalyst. [3]

4. Conclusion
The addition of potassium lowers methane selectivity and enhances olefin selectivity slightly, which is reflected in the residence times and surface coverages, with the largest observed impact on the surface coverages. The slow decay of 12C in the product distribution seem to indicate that a Mars-van Krevelen-like mechanism is present.

5. References

[1] Ordomsky V.V, Legras B, Cheng K, Paul S, Khodakov A.Y. Role of carbon atoms of supported iron carbides in Fischer-Tropsch Synthesis. Catal. Sci. Technol., 2015; 5, 1433-1437.
[2] Gracia J.M, Prinsloo F.F., Niemantsverdriet J.W. Mars-van Krevelen-like Mechanism of CO Hydrogenation on an Iron Carbide Surface. Catal. Lett. 2009; 133, 257-261
[3]
Ledesma C, Yang J, Blekkan E.A, Holmen A, Chen D. Carbon Number Dependence of Reaction Mechanism and Kinetics in CO Hydrogenation on a Co-Based Catalyst. ACS Catalysis. 2016; 6(10):6674-6686.


Extended Abstract: File Not Uploaded
See more of this Session: FTS on Co and Fe Catalysts Rapid Talks
See more of this Group/Topical: General Submissions