Biopropane from Syngas and Acyl Glycerides

By

Mark G. White, Ph. D. and Rafael Hernandez, Ph.D. Dave C. Swalm School of Chemical Engineering

Mississippi State University Mississippi State, MS 39762

white@che.msstate.edu, rhernandez@che.msstate.edu

Executive Summary

The present work examined the reaction products for the heterogeneous catalytic cracking of acylglycerides, the main components of vegetable oils and animal fats, and the catalytic conversion of synthesis gas (CO + H2) into light hydrocarbon gases and liquid transportation fuels. One objective of the project was to determine the reaction products as glycerides are cracked into gasoline, diesel range organics, and light gases such as propane. The composition of the gas (% by weight) phase after cracking of triglycerides, diglycerides, or monoglycerides was 9.9%, 20.1%, and 11.2%. The remaining compounds were associated with gasoline and diesel range compounds. Hydrogenation would have increased the yield of propane during the cracking reaction. This step was not evaluated as part of this project. Production of green fuels via lipid cracking would utilize the current petroleum refining processes and practices. Therefore, there would appear to be little changes to the production costs in producing green fuels versus traditional crude petroleum. The substantial differences arise in the cost of feedstock lipids. As long as lipid feedstocks are derived from row-crop plant oils, feedstock costs will be dependent upon farming practices and weather related issues. Therefore, cultivated lipids, either from oleaginous yeasts, algae, or municipal waste sludges, must be explored to maintain a steady supply of lipids to the refiners.

The catalytic conversion of synthesis gas, which may be obtained from gasifying lignocelluloses into light hydrocarbon gases and liquid transportation fuels, was accomplished under a variety of conditions and catalysts. The selectivity to propane was observed to vary from 3% to 17% over a range of CO conversions from 10% to 70%. A preliminary economic feasibility study based upon feedstock costs, only, showed that under some conditions, propane could be produced for just over $4 per gallon. A very attractive case was developed for which the production of gasoline, valued at $3 per gallon, offset the feedstock costs so that the price of propane was -50 cents/gallon of propane (i. e., selling gasoline at $3/ gallon completely underwrites the feedstock cost of producing propane).

1

Final Report – Production of Diesel Components and Propane from the Catalytic Cracking of Biocrude Components. PIs: Drs Rafael Hernandez and Mark White POC: Dr. Rafael Hernandez, Dave C. Swalm School of Chemical Engineering, PO Box 9595, Mississippi State, MS 39759, Tel. 662-325-0790.

Introduction The present work examined the reaction products for the heterogeneous catalytic

cracking of acylglycerides, the main components of vegetable oils and animal fats.

Another source of acyl glycerides is microorganisms. Oils generated from

microorganisms are labeled biocrude. Mississippi State Researchers in Dave C. Swalm

School of Chemical Engineering have proposed to use the wastewater treatment

infrastructure to grow microorganisms specialized on the generation of biocrude, which

can be catalytically cracked into renewable diesel and propane. It is estimated that the

use of 70% of the US water capacity for the generation of biocrude would generate

several billion gallons of these renewable feedstock.

The components of biocrude selected for these study were 1-monoolein, 1,3-

diolein, and triolein. These lipids have a glycerol backbone with oleic acid as the fatty

acid constituent. The catalyst used was H-ZSM-5, which is a highly crystalline, highly

acidic, and well studied zeolite. These glycerides were chosen as model reactants for the

identification of reaction products of catalytically cracked lipids.

The catalyst, H-ZSM-5, used in this study has an intrinsically high acidity, mostly

Bronsted, and was first synthesized by Mobil Research Laboratories (Princeton, NJ,

USA) in the mid-1970’s. It has a pore diameter of 5.4 Å and a surface area of ~ 425

m2/g. Although, H-ZSM-5 is not a commercially used catalyst, due to its instability at

high temperatures, it works well for reaction mechanistic studies.

2

The objective of the project was to determine the reaction products as glycerides

are cracked into gasoline, diesel range organics, and light gases such as propane. It was

an important factor of this work to operate the reactions in a hydrogen free environment.

Industrially operated hydrocrackers are fed hydrogen and operated at 2,000 psi and

1,400°F. To our knowledge, the use of model glyceride compounds has not been studied

to determine specific reaction steps during catalytic cracking.

For this study, the Quatra C was used for quick, online analysis of reaction

products. As described in Chapter VI, the Quatra C uses small amounts of catalyst,

usually 5 to 25 mg, and approximately 1 mg reactant for each experimental run. A

typical reaction sequence begins with the loading of catalyst into a glass tube using glass

wool to hold catalyst in place. The catalyst tube was then loaded into the reaction zone of

the Quatra C and the temperature was raised to 400°C. Air and water were monitored

using the mass spectrometer (MS), and reactions were not initialized until the air was

within 5 amu and water less than 15% of H2O+/H2O. Upon sufficiently low levels of air

and water, reactions were run by injection of reactant (monoolein, diolein, or triolein)

into the injection port.

Products were identified using MS and quantitated using TCD analysis. The

chromatographic column used for the MS was an Rxi-1ms (30m X 0.52mm X 1.25µm)

and the column used for the TCD was an RT-QPlot (30m X 0.53). The thick film,

megabore column to the MS was selected for retention of compounds from C1 to C30

hydrocarbons. The plot column to the TCD was selected for its ability to separate fixed

gases, as well as, gasoline range organics and naphthenic compounds that are precursors

to coke. The temperature profile for the GC oven started at -40°C, hold for 5 min, ramp

3

at 5 °C/min to 60 °C, then ramp at 8 °C/min to 300 °C, and finally hold for 12 min. The

segmentation parameters on the MS were set to scan 10 – 80 for the first 10 min and then

scan 50 – 200 for the remainder of the GC run.

Identification and Quantitation of Reaction Products

Figures 1 – 6 show the total ion chromatographs of reaction products for oleic

acid and the mono-, di-, and triolein. Once a peak has been identified using MS methods,

alignment of retention times using compound standards were used as further verification

of a peak’s identity. The overall product yields for the four reactants tested were (in

mg/g) 453 for oleic acid, 68 for 1-monoolein, 323 for 1,3-diolein, and 216 for triolein.

The largest difference between the compound identification of the oleic acid versus the

glycerides occurs at the front-end of the chromatograms where C1 – C6 hydrocarbon

gases elute. Due to the significantly higher yield, it appears from these results that oleic

acid cracks more readily. Also, oleic acid cracking had a higher product yield in the light

hydrocarbon gas range, which may be the result of functional group differences between

oleic acid and the esters of the acylglycerides.

Figures 7 – 10 show the product yields for all four reactants. Each of the

reactants was reacted at 3 levels of catalyst loading (5, 10, and 20 mg). While this may

not provide kinetic data, it demonstrates the change in products yields with increasing

residence time. Figure 7 shows that as reactor residence times for oleic acid increased, so

did the production of propylene and propane, in the light hydrocarbon region, and

benzene, toluene, naphthalene, and methylnaphthalene, in the aromatics region. Also, the

yield of propenylbenzene decreased with increased residence times. For the glycerides

(see Figures 8 – 10), the reaction yield for the remaining aromatics was low (less than 10

4

mg/ g reactant) and gave no specific trend in yield. Oleic acid, however, had relatively

high yields for trimethylbenzene (up to 118 mg/g) and the naphthenic compounds (up to

72 mg/g).

Another comparison can be seen in Figure 11, which shows the weight percent

composition for each product as fatty acid side chains are added to the glycerol backbone.

These results were derived from the 20 mg H-ZSM-5 catalyst runs. As can be seen from

these results, the addition of oleic acid side chains resulted in decreased production of

CO, CO2, ethane, and propenylbenzene and an increased production of propylene, C4

olefins, C4 alkanes, and C5 olefins. The decrease in inorganic carbon is due to an

increase in the ratio of carbons to oxygen in the reactant molecule. All other cracking

products remained relatively unchanged. The differences in the yield trends between

aliphatics and aromatics can be explained from slower diffusion of aromatics. Therefore,

at a given residence time, aromatic compounds are in contact with the catalyst for longer

periods of time. This allows for less time in which aliphatic compounds can oligomerize,

cyclisize, and aromatize. Table 1 shows the product compositions as divided between

gas products and organic liquid products. A composition comparison is made between

the glyceride reactants used in this study and canola oil [1] that had been reacted using H-

ZSM-5 for a temperature of 400°C and a conversion of 83.6%. Conversions for this

study are not available due to inherent limitations in determining reactant concentrations

using online analysis.

5

Table 1. Product Distribution of gas and organic liquid products.

Product Canola Oil

(Literature) Triolein Diolein Monoolein

Composition (wt %) in Gas Phase

CO + CO2 3.7 5.5 9.6 25.3

Methane 5.3

Ethylene 8.7 7.5 10.9 0.0

Ethane 6.9 9.0 17.2 35.8

Propylene 16.1 32.2 30.9 19.6

Propane 18.9 9.9 20.1 11.2

C4 Olefin 11 3.8 4.4 3.9

C4 Paraffin 17.1 3.1 3.5 3.0

C5 Olefin -- 3.1 3.4 1.1

C5 + 11

Composition (wt %) in Liquid Phase

Benzene 8.1 39.3 35.0 30.6

Toluene 18.7 22.8 19.0 19.2

Ethyl benzene 4.4 5.5 1.0 3.1

Xylenes 15 1.1 1.5 1.3

C9+ aromatics 8.8 31.3 43.5 45.9

6

Development of Reaction Mechanism

Based on the acylglyceride cracking results, it was hypothesized that cracking of

unsaturated lipids initiates at the double bond. Propenylbenzene and phenylbutene are

formed through cyclization and aromatization steps. Further cracking results in alkyl-

substituted aromatics, and coke precursors are formed from oligomerization of aromatic

compounds.

To test this hypothesis, a series of experiments were conducted using some of the

cracking products as reactants on the H-ZSM-5 catalyst. Toluene, m-xylene,

propenylbenzene, and phenylbutene were reacted separately to evaluate the secondary

product formation. Propenylbenzene was selected for further investigation due to its

behavior, with respect to yield, as residence time increases. It was observed that the yield

of propenylbenzene decreased with increased catalyst to oil ratios. Although,

phenylbutene did not have similar responses to propenylbenzene, with respect to yield, it

was chosen because of its structural similarity to propenylbenzene. Toluene and m-

xylene were examined to determine whether or not ring-opening and subsequent cracking

were occurring to produce paraffins and olefins.

As can be seen from Figures 12 and 13, toluene cracking on H-ZSM-5 produced

chiefly benzene, and m-xylene produced mostly toluene. It is believed that methyl

rearrangements and subsequent aromatic additions resulted in the formations of

trimethylbenzene and methylnaphthalene, for toluene and m-xylene reactions, and also

the formations of ethylbenzene and xylenes for toluene reactions. Table 2 illustrates the

overall yield for each of the four components studied. (Note that at the time of these

7

reaction runs no TCD was attached to the Quatra C; and therefore, only MS data was

obtained for both identification and quantitation of compounds.)

Table 2. Reaction Conditions and Results for Intermediate Product Cracking

Compound Mass H-ZSM-5

(mg)

Temperature

(°C)

Conversion

(%)

Predominant

Product

Toluene 15 400 50 Benzene

Toluene 25 400 59 Benzene

Toluene 50 400 75 Benzene

m-Xylene 15 400 46 Benzene

m-Xylene 25 400 83 Benzene

m-Xylene 50 400 88 Benzene

Propenylbenzene 20 400 81 Benzene

Phenylbutene 20 400 95 Benzene

Figures 14 and 15 show the cracking results as obtained through GC/TCD for

propenylbenzene and phenylbutene, respectively. While reacting on 20 mg H-ZSM-5,

reaction conversions were 81% for propenylbenzene and 95% for phenylbutene. These

results indicated significant breakdown of the parent compound to produce a variety of

paraffinic, olefinic, mono-aromatic, and di-aromatic compounds. Interestingly, in terms

of identified products, propenylbenzene had a mass balance of 58%, and only 34% for

phenylbutene. This perhaps might explain the anomaly seen in the lipid reactions

(Figures 7 – 10) in which propenylbenzene production decreased with increased

8

residence times but phenylbutene showed no specific trend. It is conceivable that

phenylbutene is readily transformed to naphthalene, which also has 10 carbons. Once

inside the catalyst pores of H-ZSM-5, phenylbutene can further cyclize and aromatize to

form naphthalene.

Proposed Reaction Mechanism

The reaction mechanism believed to be occurring on H-ZSM-5 begins with the

protonation of the double bond of the fatty acid constituent (Fig. 16). Figure 17 shows

the molecular geometry of triolein with dimensions determined from equilibrium

geometry calculations using Spartan software. Due to molecular size inhibitions,

protonation occurs on the outside surface of the catalyst and not within the pores. The

protonated charge migrates along the fatty acid moiety and β-scission results in cracking

of the fatty acid moiety. The resulting charged species are small enough to readily enter

the catalyst pores where additional reaction chemistry occurs. Cyclization steps result in

the formation of aromatic compounds (chiefly propenylbenzene and phenylbutene). The

propenylbenzene and phenylbutene are reaction intermediates that undergo series of

methyl shifts, hydride shifts, and isomerizations to form additional mono-aromatic

compounds. In conjunction with these intermediate reactionary steps, disproportionation

reactions are occurring. Disproportionation reactions are also responsible for the

formation of light olefins. The cyclization steps are believed to similarly to those

reported by Vedrine, et al. (1980) in which olefins are reacted with carbenium ion

intermediates to form benzene, toluene, and xylenes (demonstrated below).

9

Decarboxylation and decarbonylation steps are believed to be responsible for the

formation of CO2 and CO, respectively. As can be seen from Figure 7.11, production of

CO2 is about 2.5 times (molar basis) greater than CO. For this reason, it is believed that

decarboxylation is the dominant intermediate reaction for deoxygenation of the lipid

molecule.

Additions of mono-aromatic compounds result in the formation of naphthalene

and methylnaphthalene, which are precursors to coke. Coke is a rather ill-defined,

inherent byproduct of heterogeneous catalytic cracking reactions. Coke is a mixture of

polynuclear aromatics that form inside the catalyst pores. When a coke molecule

becomes too large to exit the pore, it poisons the catalyst by blinding off catalytic sites

for future carbenium ion formations. In industrial applications, the coke is burned off in

an air stream using higher temperatures than required for carbenium ion cracking

reactions.

Reactions using 13C labeled triolein (all 3 glycerol carbons were labeled) were

conducted to determine the pathway of the glycerol carbons. Figures 18 and 19 show the

total ion chromatograms and corresponding spectra for 13C Triolein and 12C Triolein,

respectively. When one compares the relative ratios of ions within each compound, there

are negligible differences between the two reaction series. Therefore, it is unclear

whether any of the carbons forming aromatics are derived from the glycerol backbone.

From a statistical approach, there is a 1 in 18 (3 glycerol carbons compared to 45 fatty

10

acid carbons) probability that an intermediate carbon atom stemming from the glycerol

backbone will be seen as part of an aromatic compound. Therefore, no conclusive

evidence in regards to the reaction mechanism can the obtained from this product

analysis.

CONCLUSIONS

• The unsaturated acylglycerides had the same cracking pattern with regards to the

amounts of product formed. Products yields were heavy with light hydrocarbon

gases (especially propylene) and included the formation of aromatic compounds.

• All acylglycerides studied showed the formation of propenylbenzene from

cracking on H-ZSM-5 catalyst. Propenylbenzene yield decreased from monoolein

to diolein to triolein. The product composition (wt %) of propenylbenzene

decreased with increased catalyst to oil ratios of 5, 10, and 20:1.

• Additional cracking experiments using toluene and m-xylene indicated that

transalkylation of the aromatic compound was the governing mechanism for the

formation of multi-substituted alkyl-aromatics.

• Additional cracking experiments using propenylbenzene and phenylbutene

indicated that mono-aromatic compounds with longer substituted side chains were

more likely to form light hydrocarbon gases along with multi-substituted alkyl-

aromatics than toluene and xylenes.

• The proposed cracking mechanism includes the cracking of the unsaturated fatty

acid side chains outside the catalyst pores as the ab initio step. Additional steps

include cyclization and aromatization to form mono- and di-aromatic compounds.

11

• The additional experiments using perceived aromatic intermediates which were

conducted to develop the cracking mechanism indicated that isomers of

propenylbenzene and phenylbutene were the secondary cracking products.

Tertiary products (paraffins, olefins, and aromatics) were then produced by

additional reactions of propenylbenzene and phenylbutene.

Albeit a complete economic analysis of green fuels from lipids is beyond the scope of

this work, a few ancillary thoughts should be mentioned here. Production of green fuels

via lipid cracking would utilize the current petroleum refining processes and practices.

Therefore, there would appear to be little changes to the production costs in producing

green fuels versus traditional crude petroleum.

The substantial differences arise in the cost of feedstock lipids. As long as lipid

feedstocks are derived from row-crop plant oils, feedstock costs will be dependent upon

farming practices and weather related issues. Therefore, cultivated lipids, either from

oleaginous yeasts, algae, or municipal waste sludges, must be explored to maintain a

steady supply of lipids to the refiners.

ACKNOWLEDGEMENTS

Funding for this work was provided by the Propane Research Council and the

Department of Energy Office of Energy Efficiency and Renewable Energy.

BIBLIOGRAPHY

[1] Idem, R.O., Katikaneni, S.P.R., Bakhshi, N.N. (1997) Catalytic conversion of canola oil to fuels and chemicals: roles of catalyst, acidity, basicity, and shape selectivity on product distribution, Fuel Processing Tech. V51, pp. 101 – 125.

12

Figure 1. Thermal vs Catalytic Comparison of Oleic Acid Reactions. NOTE: (Reaction Conditions: T = 400°C for both, Thermal reaction has no catalyst, Catalytic reaction had 20 mg H-ZSM-5)

13

Figure 2. GC/MS Total Ion Chromatogram for Oleic Acid Reaction. NOTE: (Reaction conditions: T = 400°C, Catalyst/oil = 20)

2-methylbutane

Trimethylbenzene

14

Figure 3. GC/MS Total Ion Chromatogram for Oleic Acid Reaction Emphasizing the Region of Aromatic Compounds. NOTE: Reaction conditions: T = 400°C, Catalyst/oil = 20

15

Figure 4. Total Ion Chromatogram for Monoolein Reaction. NOTE: Reaction conditions: T = 400°C, Catalyst/oil = 5

16

Figure 5. GC/MS Total Ion Chromatogram for Diolein Reaction. NOTE: Reaction conditions: T = 400°C, Catalyst/oil = 20

17

Figure 6. GC/MSTotal Ion Chromatogram for Triolein Reaction. NOTE: Reaction conditions: T = 400°C, Catalyst/oil = 20

18

Figure 7. Product Yield for Oleic Acid Cracking. NOTE: Reaction T = 400°C

19

Figure 8. Product Yield for Monoolein Cracking. NOTE: Reaction T = 400°C

20

Figure 9. Product Yield for Diolein Cracking. NOTE: Reaction T = 400°C

21

Figure 10. Product Yield for Triolein Cracking. NOTE: Reaction T = 400°C

22

Figure 11. Comparison of Fatty Acid Additions to the Glycerol Backbone. NOTE: 20 mg H-ZSM-5, 400°C

23

Figure 12. Product Yield for Toluene Cracking NOTE: Quantification by GC/MS, Reaction T = 400°C

24

Figure 13. Product Yield for m-Xylene Cracking NOTE: Quantification by GC/MS, Reaction T = 400°C

25

Figure 14. Product Yield for Propenylbenzene Cracking NOTE: Quantification by GC/TC D, Reaction T = 400°C

26

Figure 15. Product Yield for Phenylbutene Cracking NOTE: Quantification by GC/TCD, Reaction T = 400°C

27

Figure 16. Proposed Cracking Mechanism for the Transformation of Acylglycerides to Green Gasoline

28

Figure 17. Molecular Geometry of Triolein NOTE: Equilibrium Geometry from Semi-Empirical AM-1 Calculations

29

Figure 18. Aromatics Formation from 13C Triolein Cracking on H-ZSM-5. NOTE: Reaction conditions: T = 400°C, Catalyst/oil = 20

30

Figure 19. Aromatics Formation for 12C Triolein Cracking on H-ZSM-5. NOTE: Reaction conditions: T = 400°C, Catalyst/oil = 20

31

Biomass Conversion over Acidic Solids and Supported Metals Catalysts

Prepared by Prashant Daggolu, M.S., Shetian Liu, Ph. D. and Mark G. White, Ph. D.

Dave C. Swalm School of Chemical Engineering; Mississippi State University Mississippi State, MS 39762; white@che.msstate.edu

Executive Summary

A model compound study for the conversion of oxygenates over acidic solids was

completed to determine if these compounds can be efficiently converted to propane and other hydrocarbons. It was found that the family of alcohols (CnH2n+1OH; n = 1-4) which could be derived from processing biomass were easily converted to light hydrocarbon gases containing significant amounts of propane and hydrocarbon liquids, C6

+, over zeolites such as acidic ZSM-5 at modest reaction conditions (T = 400oC). Only modest amounts of coke were observed on the catalysts. These results prompted us to use metal-containing catalysts that would selectively convert synthesis gas (CO+H2), derived from biomass gasification, into higher alcohols and oxygenates. This catalyst did make copious amounts of light gases, including propane under certain conditions. A preliminary economic study showed that under some conditions, propane could be produced for just over $4 per gallon. A very attractive case was developed for which the production of gasoline offset the feed gas costs thus showing a net gain of ~50 cents/gallon of propane.

Background Literature Dehydration by acids. Two articles in the literature appear to be particularly helpful in determining how to remove oxygen from C2

+ oxygenates by a dehydration pathway. This pathway does not employ pressurized hydrogen nor a metal-activating catalyst to remove the oxygen. Instead, this reaction pathway involves a mechanism by which intra-molecular hydrogen combines with the oxygen to form the product water. The older articlei showed that conversion of organic oxygen compounds was possible using H-ZSM-5 catalysts (Si/Al=65) at the following conditions: T = 400oC, space-time = 3 g-cat-h/g carbon). For model compounds including alcohols, aldehydes, ketones, esters and organic acids, these researchers found that the ease of conversion to liquid hydrocarbons depended upon the length of the alkyl chain and the numbers of moles of water removed from the parent compound. For example,

▫ Compounds convert to aromatic hydrocarbons easily, if C/H ratio is less than 0.62 after water elimination. Hence alcohols are prime targets. Consider the stoichiometry of ethanol, for water elimination. One can rewrite the stoichiometry of ethanol as follows: C2H5OH C2H4[HOH]. Next examine the C/H in the hydrocarbon residue (C2H4) after the elimination of water: C/H = 0.5; which is less than the target value established by these researchers to ensure good liquids yields: < 0.62. Notice that all alcohols will show the same C/H after water elimination: CnH2n+1OH CnH2n[H2O].

▫ The yields of C5

+ liquids were the same for all alcohols tested.

32

▫ Coking is a dominant issue for compounds with C/H ratio greater than 0.62 after water elimination. We may illustrate this result best by considering the family of organic acids starting with acetic acid. We rewrite the stoichiometry of acetic acid as follows to show that two molecules of water must be eliminate in order to remove all of the water: C2H3OOH C2[HOH]2. When the waters are grouped together, all of the hydrogen originally in the molecule must be associated with the water thus leaving a residue devoid of hydrogen, and C/H = ∞; which is greater > 0.62. Coke necessarily is the primary product of this dehydration scheme. If we examine butyric acid, C4H7OOH, and then rewrite to show a hydrocarbon residue and two water molecules, we have: C4H4[HOH]2. The C/H ratio in this hydrocarbon residue is 1/1; thus, it will make coke as well according to the observations by Fuhse, et al. If these results are true, then it would seem that most of the C2

+ organic acids developed by partial hydrogenation of synthesis gases would not be good candidates for this method of de-oxygenation.

A more recent articleii describes the yields from a model compound study to represent the typical compounds often found in bio-oil products: acid, ester, alcohol, aldehyde, ketone, ether, and phenol chemical groups. Adjaye, et al. studied conversion of bio-oil model compounds on 2 g of H+-ZSM-5 at 330-410oC using a flow of liquid substrate at room pressure. The results for several substrates were as follows:

▫ 4-Methylcyclohexanol (4-MCH) used as a model compound for studying alcohols, resulted in maximum conversion of 98.2% at 410oC where aromatic hydrocarbons formed 39.9% of the products, and the coke formation was 7.8%.

▫ Propanoic acid, the model for organic acids, showed conversions from 24-100% over this temperature range, with coke yields ranging from 6 to 27% and gas yields from 5 to 38%. The gas product was mostly CO2. Very little liquid hydrocarbons were observed.

▫ 2-methylcyclopentanone (2-MCP) and cyclopentanone, models for ketones and aldehydes. The conversion of cyclopentanone was 50-97% over the range of temperatures tests whereas the conversion of 2-MCP was 60-95%. The coke and gas yields were modest for 2-MCP over this range of conditions (2-5%, and 5-9%, respectively). The liquid hydrocarbon yields were 30-53%; whereas, the “new aldehydes and ketones” created were ~3%. The light gas fractions were methane to pentane hydrocarbons with carbon oxides, mainly CO.

▫ Ethoxybenzene, EB, was the model for ethers. Unlike the other model compounds, the conversion of EB attained a maximum of only 52% at the intermediate temperature of 370oC. Unlike the other model compounds, very little water was observed suggesting that dehydration was not a dominant pathway. This result is not unexpected given the unusual inactivity of aromatic ethers even for superacidic catalysts such as HCl/AlCl3 or triflic acid. The coke yields were small (1-3%) but the light gas yields were high at all temperatures (11-14%) with moderate liquid yields (15-20%) containing oxygenates such as ethers and benzofurans. It would seem that this catalyst was not very effective for removing oxygen from the system.

33

▫ 2-Me-4-(2-propenyl) phenol, aka eugenol, was the model for phenolic compounds. The conversion increased from 57-60% over the temperature range showing small coke yields (1-4%) and smaller gas yields (<2%). The liquids were mainly the isomers of eugenol and very little water was formed. This catalyst appeared to be somewhat inactive for de-oxygenation.

These two articles suggest that some light oxygenates can be easily dehydrated over acidic zeolites to form liquid hydrocarbons (e. g., alcohols, ketones, and aldehydes); whereas other compounds are not amenable to this catalytic pathway (e. g., organic acids and esters, aromatic ethers, and phenolic compounds). Moreover, these articles reveal that decarbonylation/decarboxylation reactions also can occur in series and in parallel with dehydration reactions. In the case of the acids, aromatic oxygenates, the decarbonylation/decarboxlyation reactions often occurs to the exclusion of dehydration reactions and as may only partially de-oxgenate the substrate.  Supported metals catalysts for syngas conversion. Mo/HZSM-5 is an active catalyst for the dehydroaromatization of methane [iii,iv,v]. Dimerization of surface CHx or methylene (CH2=Mo) species was supposed to be a crucial step in the oligomerization-aromatization process. This is very similar to the supposed chain growing carbide mechanism for Ficher-Tropsch (FT) reaction [vi]. On the other hand, Mo/HZSM-5 is also an active catalyst in the aromatization of methanol and ethanol [vii,viii]. Dehydration of the alcohols on Mo/HZSM-5 was supposed to happen during the aromatization process, which is analogous to the CO insertion mechanism for FT synthesis. Furthermore, the significant stabilization effect of CO on Mo/HZSM-5 and its active incorporation into methane dehydroaromatization indicated its high reactivity in the oligomerization of surface hydrocarbon fragments (CHx, CxHy) [ix,x]. These literature results promoted us to design a ZSM-5 supported metal (M) as an innovative catalyst for the FT synthesis to produce liquid transportation fuels from plant biomass resources through gasification conversion route. The identification of the metal, M, will not be revealed until we can secure patent protection.

34

Experimental--zeolite. All studies were completed in a flow reactor holding 21 g of catalyst pellets. The zeolite was pressed into wafers and then broken into small particles that could be retained in the catalyst bed. The Helium carrier gas flow rate through the reactor was 100 STP cc/min He. A liquid pump displaced 0.5 cm3/min of EtOH feed into the reactor which was held at 400oC. The diameter of the tubular reactor was ¾”. Under these conditions, the liquid, hourly, space velocity (LHSV) was ~ 30 and the weight hourly space velocity (WHSV) was = 0.34 g C/g cat-h. The zeolite was calcined at 550oC prior to use to remove the template and the ammonium ions. The SiO2/Al2O3 ratio determined by elemental analysis was 23; whereas, the framework SiO2/Al2O3 ratio determined by the Hoffmann elimination reaction was 36. The Hoffmann reaction was followed by mass changes in a thermal gravimetric apparatus. The aluminum environment was examined in more detail by 27Al-MAS-NMR to determine the ratio of extra-framework aluminum to framework aluminum.xi This ratio was found to be 1/3 using the areas under the peaks found in this spectra corresponding to Al chemical shifts at 54 ppm and 0 ppm using an Al(OH)6(NO3)3 standard. These peak areas were used to find the framework SiO2/Al2O3 ratio determined by NMR = 30.7. The framework SiO2/Al2O3 ratios determined by the Hoffmann elimination reaction and by MAS-NMR of the aluminum nuclei are similar and thus reinforce one another. The BET surface area of the sample was 314 m2/g as determined by Quantachrome Autosorb 1-C. Experimental—metal/zeolite. M/HZSM-5 was prepared by impregnation of Mn+ aqueous solution with ammonium form ZSM-5 (SiO2/Al2O3 = 23, 50). The designated M loading amount was 5 and 10 wt.%. The samples were finally calcined in air at 773K for three hours and pelletized into 0.25-0.5 mm particles. The FT synthesis reaction was performed using a BTRS-JR Laboratory Reactor System. Before the reaction, the catalyst (1.0 g) was pretreated in methane gas flow at 923K for two hours, or in syngas (H2/CO = 1.0) flow at 673K for one hour. Gas hourly space velocity (GHSV) was 3,000 h-1. Liquid products were collected using a condenser kept at 271K, and the effluent gas from the condenser was analyzed with an on-line Gas Chromatograph (HP 5890) equipped with thermal conductive detector (TCD) and flame ionization detector (FID). Selectivity of lower hydrocarbons was calculated on carbon basis based on FID signal. The catalyst activity was calculated according to equation (1), where F0 and F are the flow rates of the syngas and effluent gas after the reaction, respectively, and Ci

0 and Ci are the concentrations of component i in the syngas and effluent gases. n is the carbon number in a product i molecular. N2 was used as the internal standard for the calculation.

(1)

(2)

 Results—dehydration of ethanol by a zeolite. The yields, Table 1, show that for an input of 100 g of EtOH, 18 g of liquid hydrocarbon were produced having the composition shown in the table. The remaining 40 g of liquid collected in the cold receiver was predominantly water which is in agreement with the theoretical yield of water when all of the ethanol was converted (100 g EtOH x 18 g H2O/46 g of EtOH = 39.1 g H2O). The amount of gas collected during this run was 41 g, of which 60 wt% was propane/propylene (~24.4 g) and 35 wt% was ethane/ethylene (14.4 g).

35

The remaining 5 wt% was higher molecular weight gases (2-3 g). Only 0.44 g of coke was produced during this run.

Results--reaction of syngas over M/HZSM-5. The FT reaction was performed on 5%M/HZSM-5 after ex-situ pretreatment in methane gas flow at 923 K for two hours. Significant amounts of benzene and naphthalene were produced during the pretreatment. The catalyst was activated 

Figure 1 Chromatograph of gas products of FT synthesis at 623K and 500 psi over 5%M/HZSM-5 after pretreated in methane gas flow at 923K for two hours.

Figure 2 GC-Mass spectra of liquid products of FT synthesis at 623K and 500 psi over 5%M/HZSM-5 after pretreated in methane gas flow at 923K for two hours.

36

under methane atmosphere was in view of the fact that methane-treated M/HZSM-5 is active for the dehydro-aromatization of lower alkanes and for the deoxy-aromatization of lower alcohols. It was thus expected that it maybe also active in the direct FT synthesis of aromatics rather than the aliphatic hydrocarbons via alcohols as the intermediate compounds. Actually, a CO conversion about 16% was obtained at 623 K and 500 psi on the methane activated M/HZSM-5. Methane, ethane, propane and a large number of other unidentified hydrocarbons as shown in Fig. 1 were detected using the on-line GC. The liquid product, which composed clearly water and oil phases, was also successfully collected from the condenser kept at 271 K. GC-Mass analysis of the liquid (dichloromethane as the solvent, Fig. 2) indicated the presence of three groups of compounds: (i) oxygenates mainly including methanol, ethanol, propanol, butanol and 2-methyl-butanol; (ii) aromatics including almost all the isomers of alkyl-substituted benzenes and naphthalenes, with xylene, trimethyl-benzene and tetramethyl-benzene as the most abundant components; (iii) aliphatic hydrocarbons (>C15), mostly in minor amounts. Lower alkanes, such as isobutane and methyl-pentane, were also detected, but only in trace amounts. The above results clearly indicated that M/HZSM-5 is active and selective for FT synthesis directly to aromatic compounds, which is distinctly different from that observed over the conventional Group VIII-based catalysts which produce aliphatic hydrocarbons as the dominant products.  Activity and selectivity of the catalyst. The pretreatment of M/HZSM-5 with methane at relatively higher temperature (923K) might produce coke in the zeolite channel, which decreases its activity towards CO conversion. Thus, the direct activation of M/HZSM-5 using syngas at lower temperature (673K) was investigated. As shown in Table 2, the CO conversion was almost doubled compared with that of methane-treated sample at 623 K reaction temperature and 500 psi pressure. Furthermore, the selectivity of liquid hydrocarbon products (Liq. CH’s), calculated based on assuming a 100% carbon balance, was also greatly increased. The negative selectivity listed in Table 2 for the liquid hydrocarbons was mainly due to the large analysis error of the internal standard N2 concentration (Ar was used as the carrier gas for 2.1% N2 analysis, and its peak was located on the tail of H2 peak.) Increasing the pressure to 1000 psi greatly increased both the CO conversion and liquid products formation. Lower temperature also favored the formation of liquid hydrocarbons, but significantly decreased the CO conversion. At 623K and 1000 psi, a liquid hydrocarbon selectivity of 66% was obtained at a CO conversion of 29%.

Table 2 Catalytic performances of M/HZSM-5 under various reaction conditions

Selectivity of Products (mol%) Catalyst SiO2/Al2O

3 Pretreat.

Gas Temp.

(K) Pressure

(psi) Conver.

of CO (%) Effluent H2/CO CO2 CH4 C2H6 C3H8 L. HCs

5%M/HZSM-5 50 methane 623 500 16.1 1.05 59.9 20.0 11.2 9.3 -0.3 623 500 31.2 1.13 51.8 19.2 8.6 10.1 10.3 623 1000 77.2 1.65 42.7 14.6 6.9 7.9 27.8 573 1000 28.8 1.15 21.9 6.4 2.5 3.0 66.1

5%M/HZSM-5 50 syngas

523 1000 8.3 1.07 10.0 3.5 1.1 3.8 81.6 673 500 77.3 1.55 52.0 24.2 12.1 15.3 -3.6 5%M/HZSM-5 23 syngas 623 500 46.4 1.26 53.3 22.4 9.2 16.1 -1.1 623 500 40.7 1.22 53.8 17.6 8.0 17.3 3.4 10%M/HZSM-5 23 syngas 573 500 15.8 1.10 55.3 12.2 3.5 13.8 15.2

37

Both the oil phase and water phase of the liquid products were analyzed with GC-Mass. Lower oxygenates including propanol and acetic acid were detected in the water phase. Methanol and ethanol were not identified most probably due to their overlap with water peak during the analysis. The composition of oil phase was different from that obtained with methane-treated sample. As shown in Fig. 3, the oil phase consisted mainly of branched alkanes as the minor fraction and aromatics as the dominant fraction. Aliphatic hydrocarbons were observed only in trace amounts. The noticeable formation of isomerized alkanes on M/HZSM-5 suggests a promising FT catalyst for high quality gasoline production. The product distribution is expected to be tunable by using different zeolite structure as the supports.

It was also noticed from the reaction results listed in Table 1 that the CO2 selectivity was very

high in almost all the cases and H2/CO ratio was greatly increased after the reaction, especially for reaction at higher temperature and pressure. Water gas shift may contribute to the increase of both the CO2 and H2. Decarboxylation of the formed acetic acid may also increase the formation of CO2 and methane. Effects of SiO2/Al2O3 and M loading. The FT reaction was also performed on a 5%M/HZSM-5 having a SiO2/Al2O3 of 23 as listed in Table 2. The CO conversion was increased, but the selectivity of liquid products was decreased and the selectivity of lower hydrocarbons (C1-C3) was increased. Assuming that the hydrocarbons were mainly formed via deoxygenation of alcohol intermediates, the zeolite with lower SiO2/Al2O3 has more acid sites, which may promote the deoxygenation reaction of lower alcohols and lead to more formation of lower alkanes as the final products. Increasing the M loading amount may also decrease and acid sites in the zeolite. The CO conversion on 10%M/HZSM-5 was very close to that over 5%Mo/HZSM-5 with the same SiO2/Al2O3 ratio of 23. Selectivity of liquid products was increased, and the selectivity of CH4 and C2H6 was decreased. But the selectivity of propane was slightly increased. It may

Figure 3 GC-Mass spectra of oil product of FT synthesis at 623K and 1000 psi over 5%M/HZSM-5 after pretreated in syngas at 673K for one hour.

38

suggest that the zeolite acidity affects the deoxygenation reaction of lower alcohols more than that of higher alcohols. Decreasing the acidity may increase the formation of higher alcohols in one hand and decrease the deoxygenation activity of lower alcohols. As a result, the lower activity of zeolite support may promote the formation of liquid hydrocarbons. Investigation of the effects of SiO2/Al2O3 and modification of the acid sites by K+ ion-exchange method is going on.

Active species of the catalyst. Literature studies on methane aromatization over M/HZSM-5 proposed MCx or MOxCy as the active species. Studies of F-T synthesis with the typical Fe- and Co-based catalysts also proposed the corresponding metal carbides (FeCx and CoCx??) as the active phases. It is thus assumed that the MCx/MOxCy possibly formed on M/HZSM-5 during methane or syngas pretreatments is active for the FT synthesis reaction. Acid sites in zeolite support may promote the deoxygenation, isomerization and aromatization reactions, but it may also promote the formation of lower hydrocarbons. Catalyst stability. The variation of CO conversion with reaction time on stream of 5%M/HZSM-5 for the FT reaction is illustrated in Fig. 4. For the reaction at 1000 psi the variation of CO conversion is roughly represented with the value of , where is the concentration of CO in the syngas and is the concentration of CO in product gas, because no internal standard gas was added to the high pressure syngas used for this test. However, two points obtained by measuring the product gas flow rates are illustrated in Fig.4. At of 60% the real CO conversion was around 77%. Clearly, an initial induction period of the reaction existed. But this initial variation in CO conversion was mainly due to the large volume of the condenser (150 mL) in our reaction system, which was located upstream of the backpressure valve causing a long, time delay when working at high pressure. However, an induction period was probably necessary for the catalyst to reach a stable working condition. The results showed in Fig. 4 indicate a quite stable performance of the catalyst within 100 hours reaction at two different pressures.

Figure 4 Conversion of CO with reaction time on stream of 5%M/HZSM-5 (SiO2/Al2O3 = 50, pretreated in syngas at 673K for one hour). (�) calculated based on internal standard or measurement of product gas flow rate; (▲) value of

39

The variation of products selectivity with reaction time on stream is showed in Fig. 5. The selectivity of CO2 and lower alkanes at 500 psi kept constant within 20 hours reaction, and also kept constant at a decreased level for 1000 psi pressure within 60 hours reaction. Other samples with Mo loadings of 5% and 10% and SiO2/Al2O3 = 23 also showed stable catalytic performances in the FT synthesis reaction with the syngas of low H2/CO ratio.

Preliminary Economics

We may estimate the preliminary economics, Table 3, for production of propane from

synthesis gas derived from biomass assuming that the cost of feed gas is $1/kg of synthesis gas derived from gasifying biomass costing $33.00 per dry metric ton. The price credit for gasoline is $3.00 per gallon. No credit was taken for the methane, ethane, and CO2 produced. Some of the best economic cases were developed for low conversion of CO for which some liquid product was observed. Even under these conditions, the price for propane derived from catalytic conversion of synthesis gas was $0.82--$1.05/kg of propane which could be much as $3.10--$3.95 per gal propane. Another very attractive case shows propane at -$0.53 per gallon (-53 cents/gal). For this case the yield of gasoline at very low conversion appears to offset the feed gas costs. This calculation does not include fixed capital/investment costs, nor does it consider the operating costs. This calculation is merely the cost of feed gas minus any price benefits.

Conclusions The more fully reduced oxygenates, such as the alcohols, are very easy to dehydrate and pose no significant problem for processing over the acidic acids according to pore sizes and acid strength. Moreover, the yields of propane/propylene are the major component of the gases produced. It is interesting that a C2 alcohol, derived from biomass, can be used to make C3 and larger hydrocarbons. Thus, the simple dehydration of ethanol to produce ethylene, is not the only reaction occurring here.  

1) M/HZSM-5 is an active catalyst for FT synthesis of branched lower alkanes and alkyl-substituted aromatics. Low temperature pre-reduction/carburization with syngas produces more active catalyst than high-temperature pretreatment using methane at higher temperature.

2) M/HZSM-5 is a stable catalyst for FT synthesis reaction. High pressure and lower reaction temperature favor to higher selectivity towards the formation of liquid

Figure 5 Selectivity of CO2 (�), methane (■), ethane (▲) and propane (●) with reaction time on stream of 5%M/HZSM-5 (SiO2/Al2O3 = 50, pretreated in syngas at 673K for one hour).

40

hydrocarbons. 3) Acidity of the zeolite support is a key factor controlling the products distribution. Lower

acidity favors to the formation of liquid hydrocarbons. HZSM-5 with higher SiO2/Al2O3 should be investigated.

4) Alcohols ranging from methanol to iso-butanol were detected in the products. Formation of the branched alkanes and alkyl-substituted aromatics were supposed through the deoxy-condensation of these intermediate alcohols. However, strong acidity of the catalyst may promote the formation of lower alkanes through hydro-dehydroxylation of these alcohol intermediates.

5) and formed during the FT synthesis Comprehensive evaluation of M/HZSM-5 based catalysts for F-T synthesis should be carried out.

6) Increase the M loading up to 10% on HZSM-5 did not greatly change the catalyst activity. A optimum M loading is probably in between 5% and 10%.

Several of the economic cases show very encouraging results suggesting that the gasoline product may offset the cost of the feed stock so that the price of propane could be less than $4 per gallon.

Table 3 Preliminary Economics for Synthesis Gas to Gasoline/Propane over M/Zeolite

T P Conv Selectivity of Products (mol%)

syngas cost per

kg propane

credit for

gasoline per kg

propane

net price

per kg propane

net price

per gal propane

(K) (psi) of CO (%) CO2 CH4 C2H6 C3H8

Liq. HC’s        

623 500 16.1 59.9 20 11.2 9.3 -0.3  $   1.44    $ (0.02)   $  1.46    $   5.50  623 500 31.2 51.8 19.2 8.6 10.1 10.3  $   3.18    $   0.95    $  2.22    $   8.39  623 1000 77.2 42.7 14.6 6.9 7.9 27.8  $13.33    $   8.15    $  5.18    $19.56  573 1000 28.8 21.9 6.4 2.5 3 66.1  $20.08    $ 19.04    $  1.05    $   3.95  523 1000 8.3 10 3.5 1.1 3.8 81.6  $   5.21    $   5.35    $(0.14)   $(0.53) 673 500 77.3 52 24.2 12.1 15.3 -3.6  $   3.77    $ (0.55)   $  4.31    $16.27  623 500 46.4 53.3 22.4 9.2 16.1 -1.1  $   2.33    $ (0.10)   $  2.42    $   9.15  623 500 40.7 53.8 17.6 8 17.3 3.4  $   2.17    $   0.24    $  1.93    $   7.28  573 500 15.8 55.3 12.2 3.5 13.8 15.2  $   1.34    $   0.52    $  0.82    $   3.10  573 1400 60.57 21.7 13 4 6 64.4  $21.28    $ 19.50    $  1.77    $   6.69  573 1000 51.96 21.2 12 3.4 5.5 59.7  $18.53    $ 16.91    $  1.62    $   6.11  

       

41

       Conclusions

The more fully reduced oxygenates, such as the alcohols, are very easy to dehydrate and

pose no significant problem for processing over the acidic acids according to pore sizes and acid strength. Moreover, the yields of propane/propylene are the major component of the gases produced. It is interesting that a C2 alcohol, derived from biomass, can be used to make C3 and larger hydrocarbons. Thus, the simple dehydration of ethanol to produce ethylene, is not the only reaction occurring here. For the zeolite-supported M syngas to gasoline/propane catalyst, we observe the following:

1. M/HZSM-5 is an active catalyst for FT synthesis of branched lower alkanes and alkyl-substituted aromatics. Low temperature pre-reduction/carburization with syngas produces more active catalyst than high-temperature pretreatment using methane at higher temperature.

2. M/HZSM-5 is a stable catalyst for FT synthesis reaction. High pressure and lower reaction temperature favor to higher selectivity towards the formation of liquid hydrocarbons.

3. Acidity of the zeolite support is a key factor controlling the products distribution. Lower acidity favors to the formation of liquid hydrocarbons. HZSM-5 with higher SiO2/Al2O3 should be investigated.

4. Alcohols ranging from methanol to iso-butanol were detected in the products. Formation of the branched alkanes and alkyl-substituted aromatics were supposed through the deoxy-condensation of these intermediate alcohols. However, strong acidity of the catalyst may promote the formation of lower alkanes through hydro-dehydroxylation of these alcohol intermediates and formed during the FT synthesis.

5. Comprehensive evaluation of M/HZSM-5 based catalysts for F-T synthesis should be carried out.

6. Increasing the M loading up to 10% on HZSM-5 did not greatly change the catalyst activity. An optimum M loading is probably in between 5% and 10%.

 

42

Table 1 Results from Reaction of Ethanol over

Zeolite

Masses, g Conversion% of EtOH 100

Input of EtOH, g 10

0 Total Liquid out, g 58 Total gas out, g 41 "Gasoline" out, g 18 Gasoline Composition, wt% Gas Composition, wt% Benzene 10 C2 35 Toluene 41 C3 60 Ethyl Benzene 6 C4, C5 4 o+p-xylene 35 others 1 others 8 Coke out, g 0.44

43

References [i] Fuhse, J., Bandermann, F., Chem. Eng. Technol. 10 (1987) 323-329 [ii] Adjaye, J.D., Bakshi, N.N., Biomass and Bioenergy 8 (1995) 131-149. [iii] “Methane activation without using oxidants over Mo/HZSM-5 zeolite catalysts” Yide Xu, Shetian Liu, Liansheng Wang, Maosong Xie, Xiexian Guo, Catalysis Letters, 30(1-4), 135-49 (1995) [iv] “Bifunctional catalysis of Mo/HZSM-5 in the dehydroaromatization of methane to benzene and naphthalene: XAFS/TG/DTA/MASS/FTIR characterization and supporting effects,” Shetian Liu, Linsheng Wang, Ryuichiro Ohnishi, Masaru Ichikawa, Journal of Catalysis, 181(2) (1999) 175-88 [v] “Direct conversion of methane under nonoxidative conditions”, Yide Xu, Xinhe Bao, and Liwu Lin, Journal of Catalysis, 216 (2003) 386-395 [vi] “Development of ideas concerning the mechanism of Fischer-Tropsch synthesis,” E. V. Slivinskii, Yu P. Voitsekhovskii, Russian Chemical Reviews 58 (1) (1989) 57-72 [vii] “Aromatization of methanol and methylation of benzene over Mo2C/ZSM-5 catalysts,” Róbert Barthos, Tamás Bánsági, Tímea Süli Zakar, Frigyes Solymosi, Journal of Catalysis, 247 (2007) 368-378 [viii] “Decomposition and Aromatization of Ethanol on ZSM-Based Catalysts,” R. Barthos, A. Sze´chenyi, and F. Solymosi, J. Phys. Chem. B 110 (2006) 21816-21825 [ix] “Unique promotion effect of CO and CO2 on the catalytic stability for benzene and naphthalene production from methane on Mo/HZSM-5 catalysts,” Shetian Liu, Qun Dong, Ryuichiro Ohnishi and Masaru Ichikawa, Chem. Commun., (1998) 1217-1218 [x] “Catalytic Dehydrocondensation of Methane with CO and CO2 toward Benzene and Naphthalene on Mo/HZSM-5 and Fe/Co-Modified Mo/HZSM-5,” Ryuichiro Ohnishi, Shetian Liu, Qun Dong, LinshengWang, and Masaru Ichikawa, Journal of Catalysis 182 (1999) 92–103 [xi] Dufreche, S. D.; Ph. D. thesis, Mississippi State University, 2007.