CO/H₂O Chemistry: An Alternative Approach to Fuels

Introduction

The Arab oil embargo of 1973 created the need for renewed interest in alternative fuels for the U.S. Since then, much research has been directed towards improving various processes having to do with the utilization of coal to produce the hydrocarbon fuels we now employ for many of our energy needs. The focus of this research has dealt with the Fischer-Tropsch reaction:

\[\ce{CO + H2 ->[catalyst][\Delta] (CH2)_n + CO2}\]

which uses the gaseous products obtainable from coal and natural gas to produce a wide range of hydrocarbons and alcohols. However, there are problems with the process, with a major one being the usefulness of H₂ in other areas. We shall focus on an alternative process which eliminates the requirement of hydrogen to synthesize fuels and chemical feedstocks from coal.

In a previous paper, we described how coal could be processed to almost exclusively CO, which could be used with water in the Kölbel-Englehardt reaction:

\[\ce{3 CO + H2O ->[catalyst][\Delta] (CH2)_n + 2 CO2}\]

Since Kölbel's pioneering work, very little has been done with CO/H₂O chemistry, excepting the water-gas shift reaction:

\[\ce{CO + H2O ->[catalyst][\Delta] H2 + CO2}\]

which has been used to alter the CO:H₂ ratios obtained from steam reformation of natural gas and coal gasification.

We sought to apply our knowledge of the mechanism of the water-gas shift reaction (WGSR) to develop catalysts active for Kölbel-Engelhardt (K-E) synthesis, also applying the oxide theory to achieve hydrocarbon as well as alcohol synthesis. These goals were achieved with high success, and research continues in these areas to fully characterize the mechanisms and catalysts involved.

Experimental

Several catalysts were prepared that consisted of a transition metal (Co, Ni, Fe) supported on a metal oxide (ZnO, Al₂O₃, MgO, Fe₂O₃) by the technique of impregnation. The nitrate of the transition metal was dissolved in water, and the solution was poured over the oxide. The mixture was rotoevaporated to remove the water, depositing the nitrate on the oxide support. This solid was dried overnight at 118°C and then calcined at 400°C in air for 4 hours. The product was then hydrogenated, if necessary, at 350°-400°C for several hours to yield the catalyst.

Alternatively, commercial catalysts were used, such as Raney cobalt (Alfa) and activated Raney nickel (Alfa). Urushibara catalysts were prepared by mixing the solid chloride of the transition metal with a slurry of zinc metal in water for 15 minutes at room temperature. The base-treated Urushibara catalysts were prepared by treatment of the Urushibara catalyst with 10% NaOH for several minutes, and filtration and washing of the resulting solid. Base-washed Raney catalysts were obtained in a similar fashion.

All reactions (except when noted otherwise) were run in a 310 mL Parr Mini-Bomb autoclave, with approximately 1 g of transition metal incorporated in the catalyst, 30-50 mL of LAH-distilled tetrahydrofuran (THF) solvent, 2.85 mL of distilled water, and approximately 600 psi of super-pure Matheson CO gas. The temperature was 245°-250°C, and the reaction time was 15-20 hours. Where CO and H₂ were used as reactants, the total initial pressure was approximately 600 psi.

The results of the experiments are shown in Tables I-V. The reactions were carried out by adding catalyst to the argon-purged autoclave, with subsequent solvent and water addition, sealing of the reactor, and pressurizing with CO, after purging the reactor with CO to eliminate any air or argon present. The reactants were gradually heated to 250°C and stirred via a mechanical overhead stirring device.

Analysis consisted of taking gas samples after cooling the bomb to 0°C, using a Linde 13X column for hydrogen analysis, a Linde 5A column for CH₄ and CO analysis, and a Chromosorb 102 column for CO₂, C₂H₄ and C₂H₆ analysis. A direct-flow GC was used to analyze for small (C₁-C₄) hydrocarbons using a Poropak Q column at 150°C. The bomb was then depressurized, and the mixture inside was filtered, and the solution was analyzed on a Poropak S column at 140°C for C₃-C₅ hydrocarbons and for methanol and ethanol. In some cases, the solution was evaporated, and the residue was analyzed by temperature-programmed GC for higher hydrocarbons and alcohols (up to C₃₀). The solid catalyst was collected, dried and saved for future use and/or characterization.

Results and Discussion

The following mechanism, which neatly puts together several different processes under one common intermediate – metal formate:

\[\ce{CO + H2O ->[MO] MO-CHO}\]

\[\ce{MO-CHO ->[M'] (CH2)_n}\]

\[\ce{MO-CHO -> H2 + CO2}\]

\[\ce{MO-CHO ->[M''] CH3OH}\]

is proposed for the production of hydrocarbons and alcohols, as well as H₂, from CO/H₂O mixtures, with the product dependent only on the catalyst used.

Metal formate has been shown to form on metal oxide surfaces by several groups, given CO/H₂O reactant mixtures. If we look at decomposition temperatures for some bulk metal formates, we obtain:

The oxides and metals above are potent water-gas shift catalysts at the temperature indicated by the formate decomposition temperature of the particular metal formate. This led us to believe that a metal formate was the active intermediate in metal-oxide WGSR, which are industrially significant.

If one conceives of a possible intermediate transition state with the transition metal–metal oxide catalyst described above, with CO/H₂O, using the oxide theory:

\[\ce{MO-CHO ->[M'] MO-CHO \bond{...} M'-H}\]

the interaction between the carbonyl oxygen and the supported metal (M') is what determines the decomposition scheme of the metal formate. We have shown that depending on the strength of the metal (M') – oxygen bond, different products may be obtained from batch thermal decomposition of several metal formates. Thus, Co, Ni and Fe would tend to hold onto the carbonyl oxygen strongly, permitting full reduction of the formate carbon to hydrocarbon by the supported metal, with subsequent reduction of the metal oxide formed by CO. On the other hand, with weak oxygen-bonders like Pt, Pd and Cu, we would expect to incorporate the carbonyl oxygen in the product, giving us alcohols. This is indeed the situation, as the results below indicate.

Table I (Cobalt-based catalysts)

^a measured as mol CO used per second x 10³; in parentheses, psi dropped per second x 10^3
b Urushibara cobalt, base-washed (rerun after washing catalyst with THF; gave same results)
^c corresponds to 1 g of cobalt metal deposited on 10 g of ZnO
^d L540 is a United Catalysts standard Cu•ZnO•Al₂O₃ methanol catalyst (ran unhydrogenated catalyst; gave same results)
^e Kieselguhr or diatomaceous earth consisting mainly of SiO₂
^f blank spaces indicate measurement was not made or was unable to be made

Table II (Effect of Temperature on Yield of Hydrocarbons from 10% Co/ZnO Catalysts)

^a see Table I
^b experiment run at 205°C had a large leak, but analysis indicated little hydrocarbon product and little activity was observed
^c see Table I, f

Table III (Nickel-based catalysts)

^a see Table I
^b Base-washed Urushibara nickel
^c corresponds to 1 g of nickel on 10 g of zinc oxide
^d standard hydrogenating catalyst from Alfa
^e same as d (also ran this catalyst for short time (5 hrs.), giving good activity, but with significant amounts of methanol)
^f see Table I

Table IV (Comparison of CO/H2O and CO/H2 Chemistry)

^a See Table I.
^b Reaction stopped when pressure stabilized.

Table V (Methanol Catalysts)

^a corresponds to 2.5 g of nickel on 2.5 g ZnO
^b same as a with iron

The results from Table I indicate that at 245°C, close to the decomposition temperature of bulk zinc formate, the most active K-E catalysts are those with ZnO as the support. Thus, U-Co-B, which consists mostly of cobalt metal with some residual ZnO, the prepared 10% Co/ZnO and the 10% Co/L540, which has ZnO as the main part of the support, had the highest K-E activity at 245°C. Catalysts which had Al₂O₃ as support had slightly less activity than for ZnO, as the decomposition temperature for aluminum formate is 280°C. Catalysts which contained a support which does not form a formate (Raney cobalt, 10% Co/Kies.) or for which the formate does not decompose below 300°C (15% Co/MgO, 10% Co/Fe₂O₃) had little or no K-E activity. These results strongly support the proposal that metal formate is the active intermediate in K-E chemistry, and suggest likely conditions to use to get other catalysts to work with CO/H₂O feeds.

The experiments in Table II are further convincing evidence for the formate intermediate route to K-E products. At temperatures significantly below the decomposition temperature of bulk zinc formate, very low yields of hydrocarbon were detected. As the reaction temperature approached the formate decomposition temperature, the yield increased significantly – suggesting that the yield of product is proportional to the amount of formate decomposed. To determine this effect for other metal oxides, further experiments will be undertaken at the temperature indicated by the chart on formate decomposition temperatures.

Comparing the results for cobalt (Table I) and nickel (Table III) catalysts, we can see some interesting differences. With similar oxygen-bond strength, these two metals exhibit widely different activities on the same support in the K-E reaction. We believe that the activity of these catalysts could be related to their ability to form solid solutions at different concentrations of metal for each case. Further surface and bulk analysis of these catalysts will determine their structural properties, which we will correlate with their activities in the K-E reaction.

Contrary to the inactivity of our prepared 10% nickel catalysts, an activated Raney nickel catalyst from Alfa was an extremely active catalyst in the K-E synthesis. Especially high yields of CH₄, C₂H₆ and C₃H₈ were observed in the gas phase (no alkenes observed). The liquid phase contained large amounts of butane, and also some oxygenated higher molecular weight products. With the cobalt catalysts, relatively small yields of C₁–C₄ hydrocarbons were obtained, and the higher molecular weight products were mostly saturated hydrocarbons having a waxy appearance. Thus, the metal as well as the metal oxide support used can have drastic effects on the products of the K-E reaction. All components of a catalyst should be scrutinized for their contribution to the overall activity of the catalyst.

It was necessary to eliminate the possibility that the K-E mechanism is just a succession of the WGSR and the Fischer-Tropsch reaction, as some have suggested. We therefore carried out the experiments in Table IV, to find out whether these claims could be true. This was done with one representative catalyst from the cobalt and nickel series. The results indicate that it would be impossible for these catalysts to produce the amount of hydrocarbon and CO₂ that they did under K-E conditions, using CO and any H₂ generated in situ. The catalysts used CO/H₂ in at least a 2:1 H₂:CO ratio, and the reaction stopped when the H₂ was used up (very quickly). Judging from the WGSR activity of the metal oxide support used (Al₂O₃), which is quite low at this temperature, there would be only enough H₂ produced (150 mmoles) to consume, at the most, 75 mmoles (given the above uptake ratio) to give the products of a Fischer-Tropsch reaction – whereas under K-E conditions, over 330 mmoles of CO were consumed. Given this data, it seems highly unlikely that the WGSR–Fischer-Tropsch mechanism for the K-E synthesis is valid.

As was referred to in the opening remarks, CO/H₂O can also be used to synthesize the important feedstock and fuel – methanol. We mentioned that the metal was the variable that could possibly alter the product composition by virtue of the its oxygen-bond strength. Thus, the weak oxygen-bonder copper is the metal component of the low temperature industrial methanol catalyst (L540). With synthesis gas feed, this catalyst produces a low yield of methanol per pass. Under CO/H₂O conditions, this catalyst gives a relatively high yield of methanol; however, the catalyst was destroyed (oxidized) in the process. No hydrocarbons were detected in the product. Similarly, a 5% Pt/ZnO catalyst at 245°C gave methanol as the only liquid product. Much research has been done in our lab which shows that methanol is the decomposition product of several metal formates when reacted with weak oxygen-bonding metals (Pt, Pd and Cu). This has led to catalysts capable of producing methanol from CO/H₂O; research we are actively engaged in at present, and will present soon.

As a consequence of the high activity of the Raney nickel catalyst from Alfa in the K-E synthesis, we decided to fabricate a catalyst of similar composition and see if it too would give hydrocarbons from CO/H₂O. The results (Table V) were quite different than for the Alfa catalyst, which is mostly finely dispersed nickel metal on up to 20w% Al₂O₃. With our catalyst, consistent production of only methanol in the liquid product was observed. The same results were obtained for a 50% Fe/ZnO catalyst. Since the 10% Ni and Fe on ZnO catalysts were ineffective in both hydrocarbon and alcohol production, it is thought that the structure of the catalyst (solid solution, chemisorbed metal, etc.) is very important to the outcome of the CO/H₂O reaction. A possibility is that the larger percentage metal catalysts are in actuality stabilized transition metal oxides, which are known to be methanol catalysts. As we obtain more results from structural analysis, we will be able to determine why these catalysts behave differently with different metal composition.

In summary, it has been shown that high yields of hydrocarbon product are possible using carbon monoxide and water as reactants in the Kölbel-Engelhardt synthesis, and that effective catalysts for this reaction can be made by impregnating a strong oxygen-bonding transition metal on a metal oxide support that forms a metal formate, which decomposes at a temperature compatible with the reducing ability of the supported metal. In addition, active methanol catalysts can be made which use carbon monoxide and water as starting materials, going through the same formate intermediate, giving us more choice as to the feedstocks for many important chemicals and fuels.