DEVELOPMENT OF ALTERNATIVE FUELS AND CHEMICALS FROM SYNTHESIS GAS

Final Technical Report

Contractor: Air Products and Chemicals, Inc. Allentown, PA 18195

Prepared for the United States Department of Energy Under Cooperative Agreement No. DE-FC22-95PC93052

May 2003

DISCLAIMER This work was prepared as an account of work sponsored by the United States Government. Neither the United States, nor the United States Department of Energy, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately-owned rights. Reference herein to any specific commercial product, process, or service by trade name, mark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

Table of Contents Disclaimer ........................................................................................................................... Table of Contents ............................................................................................................... Acronyms/Abbreviations ................................................................................................... Abstract ............................................................................................................................... Executive Summary ........................................................................................................... 1. Introduction .................................................................................................................. 2. Program Description.................................................................................................... 3. Results and Discussion ................................................................................................. 3.A Tasks 1 and 2 Activities at LaPorte Alternative Fuels Development Unit............................................................................................. 3.A.1 Introduction ..................................................................................................... 3.A.2 1995 Liquid Phase Fluid Dynamic / Methanol Run........................................ 3.A.3 Liquid Phase Fischer-Tropsch (III and IV) Demonstration Tests................... 3.A.4 1999 Liquid Phase Dimethyl Ether Demonstration Run................................. 3.B Task 3 Research and Development Activities............................................. 3.B.1 Introduction ..................................................................................................... 3.B.2 Improved Processes for Dimethyl Ether Synthesis ......................................... 3.B.3 Value-Added Chemicals from Synthesis Gas Vinyl Acetate Monomer ...... 3.B.4 Isobutanol Studies ........................................................................................... 3.B.5 Conversion to Fuels and Chemicals ................................................................ 3.B.6 Construction of Alternative Fuels Field Test Unit.......................................... 3.B.7 New Kinetic Models for the LPMEOH Process ......................................... 3.C Task 4 Program Support .............................................................................. 3.C.1 Introduction ..................................................................................................... 3.C.2 Catalyst-Wax Separation Study for Fischer-Tropsch Synthesis ..................... 3.C.3 Economics for Production of Methyl tert-Butyl Ether.................................... 3.C.4 Trace Metal Contaminants: Identification, Quantification and Removal Study ................................................................................................ 4. Conclusions ................................................................................................................... 5. General References ...................................................................................................... List of Quarterly Reports submitted by Air Products.................................................... List of Topical Reports submitted by Air Products........................................................ Topical Report Summaries................................................................................................ Appendix LaPorte Alternative Fuels Development Unit (AFDU) - Process Descriptions for Operating Campaigns (1995-1999) 1. Introduction .................................................................................................................. 2. Process Description - 1995 Methanol/Fluid Dynamic Run ...................................... 3. Engineering and Modifications - Fischer-Tropsch III Run (1996).......................... 4. Engineering and Modifications - Fischer-Tropsch IV Run (1998).......................... 5. Process Description - 1999 LPDME Run ............................................................... 6. Reference....................................................................................................................... i ii iii 1 3 13 13 15 15 15 15 22 30 36 36 36 49 55 64 78 85 90 90 90 94 98 102 104 105 106 107

A-1 A-2 A-3 A-7 A-8 A-9

ii

Acronyms / Abbreviations Ac2O acetic anhydride AcOH acetic acid AcH acetaldehyde AET Advanced Engines Technology AFDU - Alternative Fuels Development Unit AFFTU Alternative Fuels Field Test Unit Air Products Air Products and Chemicals, Inc. APCI Air Products and Chemicals, Inc. BPR backpressure regulator CETANER oxygenates that function as cetane enhancers CN cetane number CSTR continuous stirred tank reactor DEGME diethylene glycol methyl ether DME dimethyl ether DOE U.S. Department of Energy DP differential pressure Eastman Eastman Chemical Company EDS electron dispersion spectroscopy FP flash point F-T Fischer-Tropsch GC gas chromatograph GHSV gas hourly space velocity gm gram GTL gas to liquids HC hydrocarbons hr hour IGCC Integrated Gasification Combined Cycle ID internal diameter LPDME Liquid Phase Dimethyl Ether LPG liquefied petroleum gas LPMAS Liquid Phase Mixed Alcohol Synthesis LPMEOH Liquid Phase Methanol MEP methanol equivalent productivity MS mass spectroscopy MTBE Methyl tert-Butyl Ether NDG nuclear density gauge NETL National Energy Technology Laboratory PFR plug flow reactor ppb parts per billion ppbv parts per billion by volume ppm parts per million ppmv parts per million by volume

iii

Acronyms / Abbreviations (cont'd) ppmw parts per million by weight R&D research and development SBCR slurry bubble column reactor SEM scanning electron microscopy SSFI Shell Synthetic Fuels, Inc. STP standard temperature and pressure SwRI Southwest Research Institute syngas abbreviation for synthesis gas TAME tert-Amyl Methyl Ether TGA Thermogravimetric Analysis TPD tons per day uv ultraviolet VAM vinyl acetate monomer vol % volume percent WTM wet test meter

iv

Abstract This Final Report for Cooperative Agreement No. DE-FC22-95PC93052, the "Development of Alternative Fuels and Chemicals from Synthesis Gas," was prepared by Air Products and Chemicals, Inc. (Air Products), and covers activities from 29 December 1994 through 31 July 2002. The overall objectives of this program were to investigate potential technologies for the conversion of synthesis gas (syngas), a mixture primarily of hydrogen (H2) and carbon monoxide (CO), to oxygenated and hydrocarbon fuels and industrial chemicals, and to demonstrate the most promising technologies at the LaPorte, Texas Alternative Fuels Development Unit (AFDU). Laboratory work was performed by Air Products and a variety of subcontractors, and focused on the study of the kinetics of production of methanol and dimethyl ether (DME) from syngas, the production of DME using the Liquid Phase Dimethyl Ether (LPDME) Process, the conversion of DME to fuels and chemicals, and the production of other higher value products from syngas. Four operating campaigns were performed at the AFDU during the performance period. Tests of the Liquid Phase Methanol (LPMEOH) Process and the LPDME Process were made to confirm results from the laboratory program and to allow for the study of the hydrodynamics of the slurry bubble column reactor (SBCR) at a significant engineering scale. Two campaigns demonstrated the conversion of syngas to hydrocarbon products via the slurry-phase FischerTropsch (F-T) process. Other topics that were studied within this program include the economics of production of methyl tert-butyl ether (MTBE), the identification of trace components in coal-derived syngas and the means to economically remove these species, and the study of systems for separation of wax from catalyst in the F-T process. The work performed under this Cooperative Agreement has continued to promote the development of technologies that use clean syngas produced from any one of a variety of sources (including coal) for the production of a spectrum of alternative fuels (hydrocarbons and oxygenate fuels), octane enhancers, and chemicals and chemical intermediates. In particular, the data from the 1995 LPMEOH campaign provided confirmation of assumptions used in the design of the catalyst reduction system at the Kingsport LPMEOH Commercial Demonstration Project, and the alternate methanol catalyst has been in use there since late 1998. The kinetic model was also expanded to allow for more accurate prediction of methanol production and carbon dioxide (CO2) conversion, and more accurate modeling of by-product formation for the alternate methanol catalyst. The outstanding performance results of the LPMEOH Process at Kingsport can be attributed in large part to the body of work performed since 1981 in collaboration between the U.S. Department of Energy (DOE) and Air Products. In addition, a pilot-plant-tested LPDME Process has been demonstrated, and the product cost of DME from coal-derived syngas can be competitive in certain locations and applications. The need for liquid fuels will continue to be a critical concern for this nation in the 21st century. Efforts are needed to ensure the development and demonstration of economically competitive, efficient, environmentally responsible technologies that produce clean fuels and chemicals from 1

coal under DOE's Vision 21 concept. These liquids will be a component of the fuel mix that will provide the transition from the current reliance on carbon-based fuels to the ultimate use of H2 as a means of energy transport. Indirect liquefaction, which converts the syngas (H2 and CO) produced by the gasification of coal to sulfur- and nitrogen-free liquid products, is a key component of the Vision 21 initiative. The results from this current program provide continued support to the objectives for the conversion of domestic coal to electric power and co-produced clean liquid fuels and chemicals in an environmentally superior manner.

Executive Summary This Final Report for Cooperative Agreement No. DE-FC22-95PC93052, the "Development of Alternative Fuels and Chemicals from Synthesis Gas," was prepared by Air Products and Chemicals, Inc. (Air Products), and covers activities from 29 December 1994 through 31 July 2002. The full details of laboratory programs and operations at the LaPorte, Texas Alternative Fuels Development Unit (AFDU) are to be found in various topical reports. The overall objectives of this program were to investigate potential technologies for the conversion of synthesis gas (syngas) (a mixture primarily of hydrogen (H2) and carbon monoxide (CO), along with other compounds such as carbon dioxide (CO2)) to oxygenated and hydrocarbon fuels and industrial chemicals, and to demonstrate the most promising technologies at the AFDU. The program involved a continuation of the work performed under Cooperative Agreement No. DE-AC22-91PC90018, entitled "Alternative Fuels from Coal-Derived Synthesis Gas." The work was also expected to draw upon information and technologies generated in parallel current and future Department of Energy (DOE)-funded contracts. The Statement of Work for activities under this Cooperative Agreement can be divided into three categories: Activities at the LaPorte AFDU, Research and Development Activities, and Program Support. At the AFDU, modifications were made to the AFDU so that selected technologies could be evaluated at the proof-of-concept scale in one of two slurry bubble column reactors (SBCR's). Shakedown testing of the AFDU was then performed prior to operations, and four different campaigns were completed to confirm technologies and performance of catalysts developed in this or in other DOE-funded programs, as well as by the private sector. The AFDU was temporarily deactivated following each of the operating campaigns. 1995 Liquid Phase Fluid Dynamic / Methanol Run The first test of the AFDU under this Cooperative Agreement was a fluid dynamic study that was successfully completed in the high-pressure, 18-in. internal diameter (ID) SBCR in MayJune 1995. The study was conducted with the Liquid Phase Methanol (LPMEOH) Process, for which additional objectives were pursued: demonstrate operation at high-velocity conditions (1.2 ft/sec) to improve commercial SBCR design, demonstrate improved reduction procedures developed for use at the Clean Coal Technology Programs Kingsport, Tennessee LPMEOH Commercial Demonstration Project, evaluate an alternate methanol catalyst, and produce methanol for end-use testing. Significant fluid dynamic information was gathered during three weeks of operations. In addition to the usual nuclear density gauge (NDG) and temperature measurements, differential pressure measurements (DP) were made to better understand the hydrodynamics of the system. The DP measurements worked very well mechanically, without the anticipated plugging problems, throughout the run. Gas holdup estimates based on DP measurements followed the same trends as those indicated by NDG readings. However, the NDG-based gas holdups were 15-20% higher than the DP-based holdups. The difference can be explained if a radial profile for gas holdup exists in the SBCR with higher holdup in the center. Such a radial profile is expected to be prominent at high velocities studied in this run.

Interesting differential pressure data were collected using Sandia National Laboratory's highspeed data acquisition system to gain insight on flow regime characteristics and bubble size distribution. Analysis of the data suggested that the SBCR was being operated in the churn turbulent regime at most of the velocities considered. High-speed differential pressure measurements were also used to perform dynamic gas disengagement experiments. The dynamic gas disengagement curves showed a single slope compared to two distinct slopes seen in low-pressure, cold-flow work corresponding to two classes of bubble sizes. A tracer study was conducted during the run to evaluate mixing in both gas and liquid phases at three different conditions. High-velocity conditions were demonstrated during this run. Operation with a linear velocity of 1.2 ft/sec was achieved with stable hydrodynamics and catalyst performance. Acceptable oil carry-over from the reactor was observed at this velocity. The magnitude of the velocity was limited only by the recycle gas compressor capacity, as the plant was designed for 1 ft/sec maximum velocity. Improvements included for catalyst activation in the design of the Kingsport LPMEOH Commercial Demonstration Project, were also confirmed. An alternate catalyst was demonstrated for the LPMEOH Process. Expected catalyst activity, by-product formation, and stability were obtained with the alternate catalyst. Overall, the catalyst appeared very comparable to the baseline catalyst. Stable performance was obtained at both high and very low (turndown) velocities. Dephlegmator testing was conducted at various conditions during the run. Liquid Phase Fischer-Tropsch (III and IV) Demonstration Tests During this program, slurry phase Fischer-Tropsch (F-T) technology was successfully demonstrated in the low-pressure 22.5-in. ID SBCR. Earlier work at LaPorte, with iron catalysts in 1992 and 1994, had established proof-of-concept status for the slurry phase process. The third campaign (F-T III), in October 1996, aimed at aggressively extending the operability of the SBCR using a proprietary cobalt catalyst. Due to an irreversible plugging of catalyst-wax separation filters as a result of unexpected catalyst fines generation, the operations had to be terminated after seven days on-stream. Following an extensive post-run investigation by the participants, the campaign was successfully completed in March-April 1998 (F-T IV), with an improved proprietary cobalt catalyst. The DOE, Air Products, and Shell Synthetic Fuels, Inc. (SSFI) sponsored these runs. A productivity of approximately 140 grams (gm) of hydrocarbons (HC) / hour (hr)-liter of expanded slurry volume was achieved during the F-T IV campaign at reasonable system stability. The productivity ranged from 110-140 gm HC/hr-liter at various conditions during the 18 days of operations. The catalyst/wax filters performed well throughout the demonstration, producing a clean wax product. For the most part, only one of the four filter housings was needed for catalyst/wax filtration. The filter flux appeared to exceed the design flux. A combination of use of a stronger catalyst and some innovative filtration techniques was responsible for this success. There was no sign of catalyst particle attrition, and very little erosion of the slurry pump was observed, in contrast to the F-T III operations.

The reactor operated in a hydrodynamically stable manner, with uniform temperature profile and gas holdups. Nuclear density and differential pressure measurements indicated somewhat higher than expected gas holdup (45-50 volume % (vol %)) during F-T IV operations. The high gas holdup was confirmed by a dynamic gas disengagement test conducted at the end of the run. Heat transfer in the reactor was better than expected. Heat, mass and elemental balance calculations indicated excellent closure. After the initial learning curve with system dynamics, the plant was restarted very quickly (24 hours and 17 hours) following two plant trips. The quick restarts demonstrate the ease and flexibility of the slurry technology. The selectivity to wax was lower than expected, with higher methane selectivity. Returning to the baseline condition indicated a productivity decline from 135-140 to 125-130 gm HC/hr-liter of reactor volume in two weeks of operation. This may be a result of some catalyst loss from the reactor, as well as initial catalyst deactivation. The participants collected significant quantities of product and samples for further processing and analysis. Gas, liquid and solid phase mixing were studied as planned at two operating conditions using radioactive materials. Using 43 detectors around the reactor, ICI Tracerco collected a large amount of data. Washington University in St. Louis analyzed the data as part of the Engineering Development of Slurry Bubble Column Reactor Technology project with DOE (SBCR Hydrodynamics).1999 Liquid Phase Dimethyl Ether Demonstration Run A demonstration of the production of dimethyl ether (DME) by Air Products' Liquid Phase Dimethyl Ether (LPDME) Process was successfully completed in the high-pressure SBCR in October-November 1999. The demonstration was conducted at a pilot scale of 10 tons per day (TPD) to evaluate the commercial viability of the LPDME Process. Air Products' LPDME Process uses a physical mixture of a commercial methanol catalyst and a commercial dehydration catalyst in a single SBCR to co-produce DME and methanol. This process provides high syngas conversion and efficient heat transfer, and directly converts a variety of feed gas compositions. While the proof-of-concept was demonstrated in 1991 at the AFDU, the catalyst system needed improvements in stability. Researchers at Air Products significantly improved the life of the catalyst system in a study that began in 1994, and the next step was to demonstrate these improvements in a SBCR at a larger scale. This run was co-funded by this program and the Clean Coal Technology Program's Kingsport LPMEOH Commercial Demonstration Project. The plant was operated for 25 days to compare catalyst aging in a pilot-scale SBCR with that in a laboratory autoclave. Instead of a planned process variable study, the catalyst life study was extended to obtain additional data on catalyst aging. The proportion of two catalysts corresponded to a 95:5 methanol to dehydration catalyst ratio by weight. The activities of the two catalysts had a similar deactivation rate of about 0.7% per day, which is lower than the 1.2% per day observed for both the catalysts in the autoclave. However, the rate was slightly higher than the 0.5% per day rate achieved for the LPMEOH Process after 3 weeks of operation at LaPorte in a proof-of-concept test in 1988/89. This demonstration represented a significant step forward in the development of the LPDME Process. The 0.7% per day aging rate achieved in the AFDU was a large improvement over the 4% per day autoclave deactivation of the previous catalyst system (81:19 methanol to dehydration catalyst ratio by weight).

The initial productivities of methanol and DME were higher than that achieved in the laboratory, perhaps due to the effect of multiple continuous stirred tank reactors (CSTR's) in the SBCR. The SBCR operated in a hydrodynamically stable manner, with uniform temperature profile and gas holdups. Differential pressure measurements indicated about 42 vol % gas holdup and 36 wt % catalyst concentration. Gas-, liquid-, and solid-phase mixing was studied using radioactive tracer injections. A large amount of data was collected using 34 detectors around the reactor. Manganese oxide-doped gamma alumina was injected at four different locations to examine the dehydration catalyst mixing. Both short-term and long-term observations of irradiated dehydration catalyst suggested no alumina settling in the reactor. A post-run inspection of the reactor bottom head did not show any settled catalyst around the sparger, in contrast to the 1991 DME run when significant settled catalyst was found at the bottom. Washington University in St. Louis analyzed the tracer data as part of the SBCR Hydrodynamics Project with DOE. The results of an economic screening study indicated that DME can be produced in an integrated gasification combined cycle (IGCC) system at costs approaching liquefied petroleum gas (LPG) in China. The objective of the tasks covering Research and Development Activities was to conduct laboratory research and development to investigate new technologies for the conversion of syngas to chemicals, chemical intermediates, oxygenated fuels, fuel components and hydrocarbon fuels, with the goal of demonstrating the technologies on a larger scale. A significant effort within Air Products continued earlier work to develop catalysts and processes for the economic production of DME, including: 1) improvement of catalyst stability, 2) development of a kinetic understanding of the LPDME reaction system, and 3) application of this understanding to process optimization and economics. DME has shown potential as a fuel in a variety of applications. The LPDME Process is aimed at lowering DME production cost required for the fuel market. There are three simultaneous reactions in this system, namely, Methanol synthesis: Water gas shift: Methanol dehydration: CO + 2H2 CH3OH CO + H2O CO2 + H2 2CH3OH CH3OCH3 + H2O (1) (2) (3)

Reactions 1 and 2 are catalyzed by the methanol synthesis catalyst, and Reaction 3 by the dehydration catalyst. Two essential characteristics of this process may lead to low-cost DME production. First, a chemical synergy among the three reactions gives the process greater single-pass syngas conversion or reactor productivity compared to the traditional two-step process, in which methanol is formed from syngas in a methanol synthesis reactor, followed by dehydration into DME in a second reactor. The other characteristic of the process is the SBCR. It provides superior heat management to accommodate the large amount of heat release from the reactions as a result of the enhanced productivity. 6

The LPDME Process uses catalysts that are in powdered form and are suspended in hydrocarbon oil to form a catalyst slurry. During work performed under an earlier DOEsponsored program, the stability of the catalyst system was shown to be insufficient to yield a viable process. A novel mechanism, detrimental interaction between the two catalysts, was identified. Migration of Cu- and, possibly, Zn-containing species from the methanol catalyst to the dehydration catalyst under LPDME conditions causes both catalysts to deactivate. The existence of this mechanism was supported by the correlation between the rate of catalyst aging and the rate of Cu accumulation on the dehydration catalyst. Further investigation revealed that the rate of catalyst aging under LPDME conditions depends on the reaction conditions. Studies over a dual-catalyst system with aluminum phosphate as the methanol dehydration catalyst were instrumental in developing this understanding. Quantitative correlations between the rate of catalyst deactivation and reaction conditions were established. Methanol was shown to be a stabilizing reagent for the catalyst system. The correlations were used to identify stable LPDME operating conditions, especially those relevant to commercial applications. These correlations were used to identify operating conditions for the 1999 operating campaign at the AFDU (described above). Reactions 1, 2 and 3 shown above form a complex, highly non-linear reaction network. Understanding the interplay among these reactions and its net effects on the reactor performance is important for optimizing the LPDME Process and economics. One of the net effects of the interplay among the reactions is that the increase in the methanol equivalent productivity (methanol productivity plus two times the DME productivity) from LPMEOH to LPDME is a strong function of the H2:CO ratio in the reactor feed. The largest productivity enhancement is observed at the CO-rich end of the feed. This is the region where all of the factors that contribute to the chemical synergy work the best. These factors include consuming the product from methanol synthesis by methanol dehydration, and lowering the water level and replenishing H2 through the water gas shift reaction. This analysis also demonstrated that the reactor productivity of the LPDME Process is a strong function of the H2:CO ratio in the reactor feed. The maximum productivity is observed at a H2:CO ratio of around 1 for the given reaction conditions. As a matter of fact, the maximum reactor productivity for the LPDME Process varies with reaction conditions. A mathematical approach was developed to understand this dependence. The approach breaks up the rate of a reaction into its kinetic and thermodynamic components and separately studies their dependence on the reactor feed composition. The analysis reveals that the best H2:CO ratio for the thermodynamics of the LPDME reaction system is 1, while that for the kinetics of the reaction system shifts from 2 to 1 with decreasing space velocity. The very possibility of low-cost DME production by the LPDME Process lies in its enhanced reactor productivity, compared to the syngas-to-methanol reactor. This enhanced productivity translates into significantly lower capital and operating cost associated with the synthesis loop. To fully utilize the economic benefit from the enhanced productivity, the reactor should be run with 1:1 H2:CO feed. For coal-derived CO-rich syngas, this can be achieved by water injection to the LPDME reactor to increase the H2:CO ratio from the coal gasification system. For 7

natural gas-derived, H2-rich syngas, process schemes were developed to achieve overall H:C balance, while the reactor is operated with the 1:1 feed. CO2 is formed in the LPDME reactor due to the water gas shift reaction. The amount of CO2 formation closely tracks the enhancement of reactor productivity. This indicates that there is an inherent trade-off in the LPDME Process between the potential cost saving (enhanced reactor productivity) and cost addition (CO2 separation). Therefore, in developing an economic LPDME Process, one needs to take both factors into consideration. One approach is to operate the reactor in the best kinetic regime (H2:CO in the reactor feed around 1), while devising an economic way for CO2 separation. Such a process package was developed. Another area of investigation of interest to industry has been to replace the existing ethylenebased vinyl acetate monomer (VAM) process with an entirely syngas-based process. Although there are a large number of process options for the conversion of syngas to VAM, Eastman Chemical Company undertook an analytical approach, based on known chemical and economic principles, to reduce the potential candidate processes to a select group of eight processes. The critical technologies that would be required for these routes were: 1. 2. 3. 4. the esterification of acetaldehyde (AcH) with ketene to generate VAM, the hydrogenation of ketene to acetaldehyde, the hydrogenation of acetic acid to acetaldehyde, the reductive carbonylation of methanol to acetaldehyde.

This analysis showed that the cost of production of VAM from syngas was about 15% higher than the conventional oxidative acetoxylation of ethylene, primarily due to higher capital associated with the syngas-based processes. Production of isobutanol was examined at two universities well known for their work in this field. The University of Aachen designed and built two high-pressure units for isobutanol synthesis from syngas for use with both a continuous stirred tank reactor (CSTR) and a plug flow reactor (PFR). The influence of CO2 in PFRs and CSTRs on isobutanol synthesis from syngas could be shown for different catalytic systems. From these experiments, it could be demonstrated that the C1C2 step is not the only rate-limiting reaction step in isobutanol synthesis. Catalysts were tested in each unit; however, the severe reaction conditions needed will make it difficult to operate this catalyst commercially. At Lehigh University, the concept of employing a double-bed reactor with a pronounced temperature gradient to enhance higher alcohol synthesis was explored, and it was found that utilization of a Cs-promoted Cu/ZnO/Cr2O3 catalyst as a first lower temperature bed (598K) and a Cs-promoted ZnO/Cr2O3 catalyst as a second high-temperature bed (678K) significantly promoted the productivity of isobutanol from syngas. However, the higher reaction temperatures led to lower CO conversion levels and lower yield of alcohols, especially of methanol, because of equilibrium limitations. With the double catalyst bed configuration, the effect of pressure in the range of 7.6-12.4 MPa on catalyst activity and selectivity was studied. High pressure was found to increase CO conversion to oxygenated products, although the increase in isobutanol productivity did not keep pace with that of methanol. It was also shown that the 8

Cs/Cu/ZnO/Cr2O3 catalyst could be utilized to advantage as the second-bed catalyst at 613-643K instead of the previously used copper-free Cs-ZnO/Cr2O3 catalyst at higher temperature. With double Cs/Cu/ZnO/Cr2O3 catalysts, an increase in space time yields of up to 45% over the double catalyst bed configuration, with high selectivity to isobutanol, was achieved. As with the work at the University of Aachen, the economics for the production of higher alcohols using this approach were not favorable, and the program efforts in this area were concluded in September 1996. Development and commercialization of the LPDME Process will require a substantial growth in the demand for DME. A task within this program studied the potential for use of DME as a feedstock for the synthesis of liquid diesel fuel additives. These additives, dubbed "CETANER," consist of a mixture of oxygenates that function as cetane enhancers. The resulting diesel fuel- CETANER blends have improved fuel ignition properties that yield environmental and economic advantages. The major CETANER components were envisioned to be glymes, principally monoglyme (CH3OCH2CH2OCH3) and diglyme (CH3OCH2CH2OCH2CH2OCH3). Starting with DME, the most reasonable and straightforward route to such oxygenates is by oxidative coupling involving reaction with oxygen gas to yield a glyme and water. The challenge was to identify a catalyst that was selective for the oxidation coupling reaction over the thermodynamically more favorable total or partial oxidation. Numerous catalysts were examined, and the results with a literature catalyst, SnO2/MgO, reported to be active for coupling, at 300C resulted in the highest monoglyme selectivity (5.9%), a value significantly lower than the 34.5% reported in the literature for the same catalyst. Thus, identification of a viable oxidative coupling catalyst remains a challenge. Encouraging results were obtained for synthesis of diglyme by the a nonoxidative route, dehydration of 2-methoxyethanol (CH3OCH2CH2OH) over Nafion resin at 120C with a 2-methoxyethanol conversion of 6.0% and a diglyme selectivity of 77%. A second major effort was the evaluation of CETANER and its diesel fuel blend properties including fuel miscibility, water tolerance, volatility, phase change characteristics, conductivity, flash point, and cetane number. CETANER was completely miscible with diesel fuel to at least 40 vol %. Flash point requirements for U.S. diesel fuel will limit the monoglyme concentration to 4 vol % in fuel blends, while diglyme concentration is not limited. The glyme components of CETANER have relatively low toxicity; however, some glymes, in particular diglyme, may have teratogenic properties. Monoglyme and diglyme were found to have low biodegradability. Economic studies have shown that the cost of CETANER is likely to be competitive with or superior to other technologies for achieving equivalent increases in diesel fuel ignition properties, but with the unique advantage of reduced particulate emissions. The Alternative Fuels Field Test Unit (AFFTU) is a portable laboratory designed specifically to provide on-site evaluation of potential feedstocks for processes that produce alternative fuels from indigenous raw materials such as coal, natural gas or environmentally disadvantaged carbonaceous feedstocks. The AFFTU was designed and constructed during this program within budget and on schedule. The AFFTU was first utilized to provide long-term testing at the Kingsport LPMEOH Commercial Demonstration Unit. Two forms of testing were employed: 9

(1) a life test of the LPMEOH reaction in a 300-mL reactor using the actual feed streams and (2) semi-continuous analysis of those same feed streams using gas chromatographs equipped with detectors sensitive to targeted poisons. Stable LPMEOH catalyst activity was demonstrated over a 28-day life test using the actual syngas feed streams for the Kingsport LPMEOH Commercial Demonstration Unit. Subsequent operation of the LPMEOH Commercial Demonstration Unit indicated that the levels of trace contaminants in the balanced syngas feed adversely impacted the catalyst performance within the SBCR. The nature of operating with small quantities of catalyst in the laboratory autoclaves and the high ratio of surface area to volume in these units have been noted as potential causes of this elevated baseline deactivation. The AFFTU has provided important information on the presence and levels of trace contaminants in syngas, and the results from the autoclave can yield immediate information on catalyst performance as means to lower the baseline rate of catalyst deactivation are developed. Efforts were made to develop new kinetic models for the LPMEOH Process. This development was needed to meet the requirements for more accurate process simulations over a wide range of reaction conditions. It was also prompted by the fact that the existing kinetic models were developed based on a methanol catalyst that is no longer commercially available. The main areas for improvement included a more robust methanol synthesis model and more accurate prediction of CO2 conversion. In addition, all models for by-product formation needed to be revamped, since the currently used commercial methanol catalyst exhibited very different selectivity. During this effort, fifty-three different reaction conditions, including different feed gas compositions from a gas blending station, space velocities, pressures and temperatures, were examined. A statistical program developed by Air Products was used to carry out data regression and design supplemental kinetic experiments. A number of kinetic models for CO hydrogenation to methanol, from both literature and in-house development, were examined, and the best one was selected and re-parameterized. A series of new reaction by-products were identified and incorporated into the reaction network. New rate models for 12 different byproducts were developed. A new reaction (CO2 hydrogenation to methanol) was introduced into the reaction network to add the needed degree of freedom for modeling CO2 conversion. Further improvement in predicting CO2 conversion was obtained via a new water gas shift rate model and better prediction of water formation from the side reactions. Within the task on Program Support, Air Products and Bechtel Corporation performed various process, research, and economic studies when necessary throughout the program. Based upon the results from work in the earlier Alternative Fuels from Synthesis Gas project Bechtel conducted a study to describe state-of-the-art techniques for the separation of F-T wax from iron catalysts, discuss the potential for a commercially viable separation method, and present follow-up recommendations. Bulk removal of up to 50% of the catalyst can be achieved inexpensively by several of the processes. However, based upon a targeted product purity of 2 to 5 parts per million by weight (ppmw) of catalyst in the wax, only the filtration process using microfiltration membranes offered the possibility of removing the particles to the level required in a single step. Other separation options could be used if the particle size of the catalyst is large and catalyst attrition during handling and use is minimal. 10

A study was also conducted to examine the use of a liquid phase mixed alcohol synthesis (LPMAS) plant to produce gasoline-blending ethers such as methyl tert-butyl ether (MTBE). The LPMAS plant was integrated into three utilization scenarios: a coal-fed IGCC power plant, a petroleum refinery using coke as a gasification feedstock, and a standalone, natural gas-fed, partial-oxidation plant. The objective of the study was to establish targets for the development of catalysis for the LPMAS reaction. In the IGCC scenario, syngas conversions need only be moderate because unconverted syngas is utilized by the combined cycle system. A once-through LPMAS plant achieving syngas conversions in the range of 38-49% was found to be suitable. In the petroleum refinery scenario, high conversions (~95%) are required to avoid overloading the refinery fuel system with low Btu-content, unconverted syngas. To achieve these high conversions with the low H2/CO ratio syngas, a recycle system was required (because of the limit imposed by methanol equilibrium), steam was injected into the LPMAS reactor, and CO2 was removed from the recycle loop. In the standalone LPMAS scenario, essentially complete conversions are required to achieve a fuelbalanced plant. The economics of this scenario are highly dependent on the cost of the natural gas feedstock and the location of the plant. For all three case scenarios, the economics of a LPMAS plant were marginal at current ether market prices. Large improvements over demonstrated catalyst productivity and alcohol selectivity will be required. A third study by Bechtel considered the impact of trace contaminants from coal gasification systems on catalytic processes. Catalyst systems tend to have more stringent requirements for presence of these species than the environmental regulations when the syngas is burned in a gas turbine. Bechtel carried out a trace contaminant literature search in two directions. The first search surveyed the experimental work being performed under DOE sponsorship at various research organizations to understand the mechanisms that impact the levels of trace elements during combustion and gasification. A survey of data published by the major supplier of coal gasification equipment was also performed. After determining the type and concentration of the catalyst poisons, Bechtel contacted more than 30 vendors to obtain information on the cost and performance of systems to remove these compounds. Six vendors responded, recommending contaminant removal systems based on various adsorbents such as activated carbon, molecular sieves, metal oxides, and proprietary adsorbents. The incremental production cost to upgrade a fuel-grade syngas containing 5 ppmv total sulfur to the requirements that could be expected for methanol synthesis catalysts ranged from 4-5 cents per million Btus for an activated carbon system, to 23-46 cents per million Btus for a metal adsorbent system. A lower cost, metal-based system had a production cost in the range of 12-18 cents per million Btus. _____________________________ In summary, the overall objectives of this program to investigate potential technologies for the conversion of syngas to oxygenated and hydrocarbon fuels and industrial chemicals, and to demonstrate the most promising technologies at the AFDU were met.

11

Laboratory work was performed by Air Products and a variety of subcontractors, and focused on the study of the kinetics of production of methanol and DME from syngas, the production of DME using the LPDME Process, the conversion of DME to fuels and chemicals, and the production of other higher-value products from syngas. Four operating campaigns were performed at the AFDU during the performance period. Tests of the LPMEOH Process and the LPDME Process were made to confirm results from the laboratory program and to allow for the study of the hydrodynamics of the slurry reactor at a significant engineering scale. Two campaigns demonstrated the conversion of syngas to hydrocarbon products via the slurry-phase F-T Process. Other topics that were studied within this program include the economics of production of MTBE, the identification of trace components in coal-derived syngas and the means to economically remove these species, and the study of systems for separation of wax from catalyst in the F-T Process. The work performed under this Cooperative Agreement has continued to promote the development of technologies that use clean syngas produced from any one of a variety of sources (including coal) for the production of a spectrum of alternative fuels (hydrocarbons and oxygenate fuels), octane enhancers, and chemicals and chemical intermediates. In particular, the data from the 1995 LPMEOH campaign provided confirmation of assumptions used in the design of the catalyst reduction system at the Kingsport LPMEOH Commercial Demonstration Project, and the alternate methanol catalyst has been in use since late 1998. The kinetic model was also expanded to allow for more accurate prediction of methanol production and CO2 conversion, and more accurate modeling of by-product formation for the alternate methanol catalyst. The outstanding performance results of the LPMEOH Process at Kingsport can be attributed in large part to the body of work performed since 1981 in collaboration between the DOE and Air Products. In addition, a pilot-plant-tested LPDME Process has been demonstrated, and the product cost of DME from coal-derived syngas can be competitive in certain locations and applications. The need for liquid fuels will continue to be a critical concern for this nation in the 21st century. Efforts are needed to ensure the development and demonstration of economically competitive, efficient, environmentally responsible technologies that produce clean fuels and chemicals from coal under DOE's Vision 21 concept. These liquids will be a component of the fuel mix that will provide the transition from the current reliance on carbon-based fuels to the ultimate use of H2 as a means of energy transport. Indirect liquefaction, which converts the syngas (H2 and CO) produced by the gasification of coal to sulfur- and nitrogen-free liquid products, is a key component of the Vision 21 initiative. The results from this current program provide continued support to the objectives for the conversion of domestic coal to electric power and co-produced clean liquid fuels and chemicals in an environmentally superior manner.

12

1. Introduction The approach taken to achieve energy independence and to improve the global competitiveness of U.S. energy conversion technology has been to promote the development of technologies that use clean syngas produced from coal, natural gas, and other non-traditional feedstocks to produce a variety of alternative transportation fuels and chemicals. Since 1981, Air Products and Chemicals, Inc. (Air Products) has led the development of LiquidPhase technology and demonstrated the efficacy of this technology to the production of lowestcost, clean-burning, alternative fuels from domestic fossil energy sources. Continued developments in the areas of advanced gasifiers, slurry bubble column reactors (SBCR's), and catalyst technology have further enabled the development of new liquid-phase technologies based upon Fischer-Tropsch (F-T) and other syngas reactions to efficiently produce alternative fuels and chemicals. The success of the liquid-phase program has centered on the unique U.S. Department of Energy (DOE)-owned Alternative Fuels Development Unit (AFDU) located in LaPorte, Texas, which is operated by Air Products. The LaPorte AFDU has unique capabilities to provide the proper springboard for commercialization of new technology. The facility is capable of processing syngas of widely varying compositions in SBCRs -- the heart of liquid-phase technology -- to demonstrate production at an industrially relevant engineering scale of 5-15 tons per day (TPD). The AFDU is adjacent to a commercial plant that produces both hydrogen (H2) and carbon monoxide (CO) to provide the flexibility to simulate syngas from any alternative energy source. More detailed information on the AFDU is provided in the Appendix to this Final Report. This Final Report for Cooperative Agreement No. DE-FC22-95PC93052, the "Development of Alternative Fuels and Chemicals from Synthesis Gas," was prepared by Air Products, and covers activities from 29 December 1994 through 31 July 2002. A description of the development work performed in the laboratory and operations at the AFDU under this Agreement is provided. 2. Program Description The overall objectives of this program were to investigate potential technologies for the conversion of syngas to oxygenated and hydrocarbon fuels and industrial chemicals, and to demonstrate the most promising technologies at the AFDU. This work is a follow-on to an earlier program undertaken by Air Products and DOE under Cooperative Agreement No. DEAC22-91PC90018, entitled the "Alternative Fuels from Coal-Derived Synthesis Gas." During this initial program, laboratory work focused on the production of oxygenated fuels and fuel additives, and the AFDU was also used to demonstrate the capability to manufacture these chemicals at a significant engineering scale. Additionally, the F-T Process, which produces primarily straight-chained hydrocarbons (which can be further refined to a clean-burning diesel fuel) from syngas, was also demonstrated for the first time at the LaPorte unit. The successes achieved in this program identified the strong positive potential of the SBCR technology. Using 13

the understanding gained and the advantages provided by this technology, several important areas were identified for future laboratory research, process development, and proof of concept demonstration. Clean burning alternative fuels and fuel additives remained as the targeted products, but the technology applications were expanded to include the production of industrial chemicals and chemical intermediates of strategic national importance. Conversion of the syngas to useful products focused on high efficiency and low capital systems, of which the SBCR is a prime example. The focus of the commercialization effort continued to be the AFDU facility at LaPorte. Interesting new syngas conversion technologies developed in the laboratory by the project participants were required to first demonstrate sufficient technical and economic potential before being considered for testing at LaPorte. The unique capabilities of the AFDU provide the proper springboard for commercialization of new technology. If the concept works in the AFDU, and the process economics and market assessments still hold, then the technology can be ready for commercial demonstration. This program fostered a government/industry partnership that built upon DOE-supported technical expertise to provide the country with a degree of energy independence and to provide U.S. industry with advanced routes to strategically important fuels and chemicals. A broad involvement of universities, national laboratories, and industry was utilized in the basic R&D tasks, and industrial partners were used where possible in support of LaPorte operations. Under this program, Air Products assembled a team including Eastman Chemical Corporation, Bechtel Corporation, and the Universities of Aachen and Lehigh to perform supporting laboratory research in both catalysis and process development to enable cost-effective syngas technology and to conduct proof-of-concept operations at the LaPorte AFDU for those technologies that appear to be economically attractive in the marketplace. The work was also expected to draw upon information and technologies generated in parallel current and future DOE-funded contracts. For example, ongoing analyses performed under Cooperative Agreement No. DE-FC22-95PC95051, entitled "Engineering Development of Slurry Bubble Column Reactor Technology," (herein called: SBCR Hydrodynamics) were utilized to assess the performance of the SBCR during operation of the LaPorte AFDU. Similarly, results gathered from this project and the SBCR Hydrodynamics project were utilized under Cooperative Agreement No. DE-FC22-92PC90543, a Clean Coal Technology Round III project in Kingsport, Tennessee entitled Liquid Phase Methanol Demonstration (herein called the Kingsport LPMEOH Commercial Demonstration Project).

14

3. Results and Discussion 3.A. Tasks 1 and 2 Activities at LaPorte Alternative Fuels Development Unit 3.A.1 Introduction The U.S. Department of Energy (DOE)-owned Alternative Fuels Development Unit (AFDU) is located in LaPorte, Texas, and is operated by Air Products. A photograph of the AFDU is provided in Figure A-1 of the Appendix. The facility is capable of processing synthesis gas (syngas) of widely varying compositions in slurry bubble column reactors to demonstrate liquid phase technology at an industrially relevant engineering scale of 5-15 tons per day (TPD). The AFDU is adjacent to a commercial plant that produces both hydrogen (H2) and carbon monoxide (CO) to provide the flexibility to simulate syngas from any alternative energy source. The objectives of Task 1, Engineering and Modifications, were to perform modifications to the AFDU so that selected technologies could be evaluated at the proof-of-concept scale, to provide analytical support to operations at the AFDU, and to test and procure catalysts that would be used in the plant trials. Modifications to the AFDU are described in the sections that deal with each of the operating campaigns (Sections 3.A.2 through 3.A.4); catalyst testing associated with the development of the Liquid Phase Dimethyl Ether (LPDME) Process is summarized in Section 3.B.3. The objectives of Task 2, AFDU Shakedown and Operations, were to perform shakedown testing of the AFDU prior to continued operations, to conduct various tasks to confirm technologies and performance of catalysts developed in Task 3 or in other DOE-funded programs, as well as by the private sector, and to temporarily deactivate the facility following the operating campaigns. 3.A.2 1995 Liquid Phase Fluid Dynamic / Methanol Run A fluid dynamic study was successfully completed in the high-pressure, 18-in. internal diameter (ID) SBCR in May-June 1995. The main objective of this run was to perform a fluid dynamic study in a SBCR, including differential pressure (DP) measurements along reactor height to estimate gas holdup, dynamic gas disengagement measurements during shutdown tests to understand flow regime and bubble size distribution, and radioactive tracer studies to evaluate mixing in both liquid and gas phases. The study was conducted with the Liquid Phase Methanol (LPMEOH) Process, for which additional objectives were pursued: demonstrate operation at high-velocity conditions (1.2 ft/sec) to improve commercial reactor design, demonstrate improved reduction procedures developed for use at the LPMEOH Commercial Demonstration Unit at Kingsport, Tennessee, evaluate an alternate methanol catalyst, and produce methanol for end-use testing. Modifications were conducted in the AFDU to measure relevant fluid dynamic parameters during the operation: two nozzles (N1 and N2) were added to the high-pressure SBCR for DP measurements, as well as liquid tracer injections. A schematic of the reactor is given in Figure A.1. Six new DP transmitters were added and connected to both the existing Distributed Control System and a new high-speed data acquisition system from Sandia National 15

Laboratories. A new stronger 8 curie Cs-137 source was installed for the reactor nuclear density gauge (NDG) to improve the resolution of the NDG reading by a factor of four, and the NDG was calibrated with nitrogen (N2). Figure A.1 LaPorte AFDU Oxygenates High-Pressure Reactor Differential Pressure Transmitter and Tracer Study Detector Positions
N W S
Syngas/Products Out

DET

0.46 m

7
1.52m

DET

6
1.83m

DET

9.66 m

Liquid/Catalyst Injection to Center (N1-Center)

5
2.74m 13.25m 2.74m

DET

N1 Liquid/Catalyst Injection to Sidewall (N1-Sidewall)

DET

3
0.61m

DET

3.56 m N2

Liquid/Catalyst Injection to Center (N2-Center)

2
1.74m

DET

Liquid/Catalyst Injection to Sidewall (N2-Sidewall)

DET

1
1.52m

Recycle Fresh Feed Syngas In

DET

Gas Tracer Injection

DET - detector

16

Significant fluid dynamic information was gathered during three weeks of liquid phase methanol operations. In addition to the usual NDG and temperature measurements, DP measurements were made to better understand the hydrodynamics of the system. The DP measurements worked very well mechanically, without the anticipated plugging problems, throughout the run. Results from the alternate catalyst run are shown in Figure A.2. Gas holdup estimates based on DP measurements followed the same trends as those indicated by NDG readings. However, there appeared to be a systematic difference between gas holdup estimates from the two methods. To check the accuracy of the DP readings, the DP transmitters were calibrated at the end of the run by filling the reactor with water. The calibrations resulted in only minor corrections in the zero and the span. The NDG based gas holdups remained relatively 15-25% higher than the DP-based holdups (37-51 volume % (vol %) vs. 28-40 vol %). Holdups based on shutdown tests conducted at two different conditions matched well with the DP based holdups. The holdups based on shutdown tests were estimated using liquid levels measured by NDG before and after the shutdown tests. The NDG is considered highly accurate in measuring liquid levels. The estimated holdups from correlations based on NDG data at low velocities also had a better match with holdups from DP data. The systematic error in the densitometry readings can be linked to the measurement technique that relies on data obtained at a single chord (diameter). This method would be accurate if the gas holdup is uniform radially. However, a radial distribution of gas holdup is generally observed in two-phase and three-phase flows, such that gas holdup is highest at the centerline and decreases toward the wall. For such a profile, the data averaged along the diameter would give too much weight to the area with highest gas holdup at the center, and therefore would overestimate the average gas holdup. This appears to be the case for the gamma densitometer measurements at the AFDU. The effect was more prominent during this run, as most of the operations were at higher velocities. Figure A.2

Fluid Dynamic / Methanol Run at LaPorte Alternate Catalyst (Run No. AF-R14)
70 Gas Hold-up (vol%) 60 50 40 30 20 10 0 50 100 150 200 250 Time On-stream (hrs)
R14.1 R14.2 R14.3 R14.4 R14.5

from NDG from DP from SD Test from Correltn.

17

The high-speed data acquisition system installed by Sandia National Laboratories personnel to monitor the DPs on the SBCR was operated throughout the run. Sandia personnel conducted detailed analysis of the data. Statistical analysis was performed on the gas holdup data to discern flow regime transitions. The standard deviation of the gas holdup increased with velocity as expected, possibly showing that the largest gas bubbles are increasing in size and/or number. A frequency spectrum obtained from Fourier transform analysis of the DP data at high velocity showed a wide band of frequencies, but also the existence of a discernible peak at about 0.05 Hz, suggesting that a large pocket of gas either enters or leaves the region between the pressure nozzles every 20 seconds. The strongest frequency was found to increase with velocity, and there did not appear to be a dominant frequency for velocity less than 0.5 ft/sec. The beginning of the appearance of a dominant frequency is an indication that large bubbles are present and that a transition to churn-turbulent flow has begun. Thus, the flow appears to be in the churnturbulent regime for all conditions, except for the two low-velocity cases. A dynamic gas disengagement analysis was performed on high-speed DP data collected during the shutdown tests to determine bubble size distributions. The dynamic gas disengagement curves showed a single slope compared to two distinct slopes seen in low-pressure, cold-flow work, corresponding to two classes of bubble sizes. One explanation for the difference in the curves could be that the gas shutdown at the AFDU was too slow to measure bubble classes. During both operations, attempts were made to operate at superficial gas velocities higher than the 1 ft/sec design velocity. In each run, the NDG readings indicated high fluctuations compared to those typically observed at lower velocities, but average readings were stable. Also, the oil loss rate from the reactor was moderate. A superficial gas velocity of 1.13 ft/sec was achieved at 720 psig during the baseline run, limited by the capacity of the recycle compressor. The reactor performance was stable, with a production rate of about 18 TPD. A modest loss rate of about 10 gallons per hour was estimated from level rises in vessels downstream of the reactor. During the run with the alternate catalyst, a superficial gas velocity of 1.18 ft/sec was achieved at a lower pressure (520 psig). The limitation for this case was CO supply. The plant performed steadily at this condition, with expected catalyst performance. CO conversion was about 33% compared to a 2-CSTR expectation of 32.5%. The two runs at high velocities demonstrated that we had not reached SBCR limitations at 1.2 ft/sec, and operations at higher velocities were possible. Improvements included for catalyst activation in the design of the Kingsport LPMEOH Commercial Demonstration Unit, were also confirmed. Successful activations were achieved using dilute CO as a reductant, a faster temperature ramp, and lower gas flow, compared to previous "standard" activation procedures. Reduction of the baseline catalyst was conducted using 4% CO in N2 with the reactor pressure at 67 psig. The heatup proceeded from 197 to 464F at a rate of 15F/hr. The temperature ramp was significantly faster than the previous "standard" ramp to save time in commercial applications. The reduction under CO was quite rapid, and the total uptake peaked out very close to the theoretical maximum value of 2.82 SCF/lb oxide. The reduction was essentially complete at 360F, or 12-13 hours on-stream, which was an encouraging result for the Kingsport project. Reduction in the SBCR was faster compared to the autoclave. Despite the rapid uptake, the reactor internal heat exchanger was easily able to control temperature, and the ramp rate proceeded on schedule with no evidence of an exotherm. At 392F, the reduction gas flow was reduced as planned to reduce oil loss from the reactor and conserve on N2 usage. Gas holdup during the reduction was close to expected: 18

27-30 vol % at 12,500 SCFH and 24 vol % at 9,375 SCFH. The catalyst concentration was in the 39-41 wt % range. An alternate catalyst was demonstrated for the LPMEOH Process. Expected catalyst activity, by-product formation, and stability were obtained with the alternate catalyst. Very similar CO conversion and methanol production rate were obtained for the two catalysts at two different conditions. Lower gas holdup occurred with the alternate catalyst. CO conversions obtained with both catalysts at different conditions are shown in Figure A.3. In addition to the similarity of the two catalysts, the plot shows stable operation with the alternate catalyst, when conversion for R14.5 is compared with that for R14.1. By-product data were analyzed more closely as increased levels of higher alcohols, methyl formate and methyl acetate were observed with the alternate catalyst at low space velocity conditions. The baseline catalyst was not operated at this condition; however, as shown in Table A.1, comparison of the two catalysts was available at two other conditions, and the by-product formation was very similar for the two catalysts at these conditions. Overall, the catalyst appeared very comparable to the baseline catalyst. Stable performance was obtained at both high and very low (turndown) velocities. Figure A.3

Fluid Dynamic / Methanol Run at LaPorte

CO Conversion to Methanol (%) 60 50 40 30 20 10 0 0 50 100 150 200 250 Time On-stream (Hrs)
R13.1 R14.1 R14.4 R14.5 R13.2 R14.2 R13.3

Baseline Catalyst
R14.3

Alternate Catalyst

19

Table A.1 By-Product Analysis Comparison for Baseline and Alternate Catalysts in LPMEOH Process

CATALYST RUN NO. SYNGAS COMPOSITION SPACE VEL, SL/HR-KG PRESSURE, PSIG TEMPERATURE, DEG F PRODUCT ANALYSIS, WT% methanol ethanol 1-propanol iso-propanol 1-butanol 2-butanol iso-butanol 2-Methyl 1Buoh 1-pentanol 2-Methyl 1-Peoh 1-hexanol 2-Methyl 1-Isobutyrate meAc etAc meFm DME CO2 water oil Total

BASELINE AF-R13.1B TEXACO 7000 750 482

ALTERNATE AF-R14.1B TEXACO 7000 750 482

BASELINE AF-R13.2B KINGSPORT 4000 735 482

ALTERNATE AF-R14.2B KINGSPORT 4000 735 482

96.72 0.89 0.25 0.02 0.17 0.06 0.06 0.00 0.08 0.00 0.04 0.00 0.20 0.00 0.91 0.00 0.00 0.43 0.19 100.00

96.73 0.91 0.26 0.02 0.17 0.05 0.05 0.00 0.08 0.00 0.04 0.00 0.22 0.00 0.99 0.00 0.00 0.42 0.06 100.00

97.31 0.32 0.08 0.01 0.09 0.03 0.01 0.00 0.03 0.00 0.00 0.00 0.06 0.00 0.42 0.00 0.00 1.46 0.19 100.00

97.42 0.35 0.10 0.00 0.08 0.03 0.02 0.00 0.03 0.00 0.00 0.00 0.07 0.00 0.44 0.00 0.00 1.30 0.16 100.00

Prior to the three-phase slurry operations, the reactor was loaded with oil and heated up to conduct a carbonyl burnout and two-phase dephlegmator testing. The carbonyl levels were extremely low during the entire burnout: 2-10 parts per billion by volume (ppbv) iron carbonyl and undetectable (<10 ppbv) nickel carbonyl. During the burnout period, extensive testing of the 21.11 dephlegmator was also conducted. The dephlegmator had been added during the 1993 modifications of the plant as a possible replacement for the cyclone, the feed-product economizer and the vapor-liquid separator. During the 1994 isobutanol run, the dephlegmator did not perform as designed. Data were collected in the two-phase system during the current burnout period to rule out fouling. The dephlegmator continued to perform below expectations, indicating catalyst fouling was not the main reason for lack of adequate performance. The heat transfer performance of the dephlegmator was lower than expected. In addition, the oil carryover was significantly higher than expected at the operating temperatures. Further measurements were made during the three-phase operations. The heat transfer performance of the dephlegmator continued to be lower than expected with high oil carry-over. Although flooding was ruled out by calculations, variability in oil capture was still apparent. Further data analysis and additional tests are needed before a final decision can be made on inclusion of the dephlegmator in a commercial flowsheet. 20

Approximately 64,300 gallons of methanol were produced during this demonstration, which was available for product testing. Summary and Conclusions The following conclusions were drawn based upon the results of this operating campaign: A fluid dynamic study was successfully completed in a SBCR, gathering significant information at pilot scale. DP measurements made to better understand the dynamics of the system worked very well mechanically throughout the run, without the anticipated plugging problems. Significant challenges have been anticipated in future attempts to obtain accurate gas holdup measurements in large-diameter SBCR's; the successful use of DP transmitters during this run provides a method of obtaining data on this important parameter. Gas holdup estimates based on DP measurements followed the same trends as those indicated by NDG readings. However, the NDG-based gas holdups were 15-20% higher than the DP-based holdups. The difference can be explained if a radial profile for gas holdup exists in the SBCR, with higher holdup in the center. Such a radial profile is expected to be prominent at the high velocities studied in this run. The model for the design of the SBCR was updated based upon this data; future designs will benefit from this improvement. Differential pressure data collected using Sandia's high-speed data acquisition system provided insight into flow regime characteristics and bubble size distribution. Standard deviation and a Fourier spectrum analysis of the DP fluctuations suggested that the SBCR was being operated in the churn turbulent regime at most of the velocities considered. This result is important, given that large-diameter SBCR's should also operate in this flow regime. Operation with a superficial gas velocity of 1.2 ft/sec was achieved with stable fluid dynamics and catalyst performance. Acceptable oil carry-over from the reactor was observed at this velocity. Improvements included for catalyst activation in the design of the Clean Coal III LPMEOH Commercial Demonstration Unit at Kingsport, Tennessee, were also confirmed. Successful activations were achieved using dilute CO as reductant, a faster temperature ramp, and smaller gas flow, compared to the previous "standard" activation procedure. An alternate catalyst was demonstrated for the LPMEOH Process, providing a competitive environment for procurement of catalyst supplies in the future. Expected catalyst activity, by-product formation, and stability were obtained with the alternate catalyst. Overall, the catalyst appeared very comparable to the baseline catalyst. Stable performances were obtained at both high and very low (turndown) velocities.

21

3.A.3 Liquid Phase Fischer-Tropsch (III and IV) Demonstration Tests Slurry phase Fischer-Tropsch (F-T) technology was successfully demonstrated in the lowpressure 22.5" ID SBCR. During October 1996, operations were carried out at the AFDU to evaluate further improvements to the slurry process for F-T synthesis. Earlier work at LaPorte, with iron catalysts in 1992 and 1994, had established proof-of-concept status for the slurry phase process (3, 4). The first campaign in 1992 (F-T I) was a 19-day demonstration of the technology at 1 TPD product scale and addressed scale-up issues such as catalyst activation, catalyst performance and hydrodynamics. The scale-up of the technologies involved demonstration in the 22.5" ID SBCR based on laboratory bench scale investigations. A very high level of reactor productivity (more than five times the F-T I productivity) was demonstrated for slurry phase F-T synthesis in 1994. Reactor productivity of 136 grams of hydrocarbons/hr-liter of three-phase slurry volume (gm HC/hr-liter) was achieved, which was within the target of 120-150 gm HC/hrliter. The productivity was constrained by mass transfer limitations, due to slurry thickening. With an improved catalyst, if carbon formation can be avoided, there appeared to be significant room for further improvements. The third campaign (F-T III), in October 1996, aimed at aggressively extending the operability of the SBCR using a proprietary cobalt catalyst. Due to an irreversible plugging of catalyst-wax separation filters as a result of unexpected catalyst fines generation, the operations had to be terminated after seven days on-stream. Following an extensive post-run investigation by the participants, the campaign was successfully completed in March-April 1998 (F-T IV), with an improved proprietary cobalt catalyst. Key issues such as catalyst-wax separation, reactor productivity improvements, reactor temperature control, and insitu activation were addressed. The DOE, Air Products, and Shell Synthetic Fuels, Inc. (SSFI) sponsored these runs. The principal objective of these runs was to conduct F-T synthesis in a large diameter SBCR and demonstrate sustainable high productivity - a space time yield of 150 gm HC/hr-liter, activity and selectivity of SSFI's proprietary catalyst, catalyst-wax separation by external cross-flow filtration, and in-situ reduction of catalyst pre-cursor. The runs would also allow the participants to study other issues such as large-scale fluid dynamics, erosion, catalyst stability, and catalyst attrition. The catalyst-wax separation has been recognized as a challenge. No single proven technology existed in the public domain. An external system of tangential (cross) flow filters was used at LaPorte based on SSFI's pilot plant experience. Filtration was preferred at reactor pressure to avoid catalyst attrition that may occur if a control valve is used to reduce the pressure. The existing filtration system at LaPorte was designed for low pressure, with limited capacity. So, the entire filtration system was redesigned and replaced. The new system was rated at higher pressure (1,000 psig) and higher temperature (600F), with significantly higher capacity. It included four new cross-flow filters in series, a catalyst-wax slurry circulation pump, a slurry cooler and a slurry degasser. A layout of the filtration system is shown in Figure A.4. The degasser was installed close to the top of the reactor to obtain almost the same liquid level in the two vessels. The slurry cooler was located at a level near the bottom of the reactor. The pump and the filters were installed at the ground level. There was no back-up system for filtration. If the filtration did not work as designed, the reactor would have to be shut down.

22

Figure A.4 Fischer-Tropsch III Filtration System Layout

During the F-T III operations, the cross-filters functioned adequately at the initial low productivity condition. However, while transitioning to the high condition, a large increase in the pressure drop across the filters was observed, and the filters were not able to keep up with the wax production. Backflushing the system seemed to increase the pressure drop further. The reactor temperature was then reduced to lower the productivity. It was assumed that catalyst fines, perhaps created by the slurry pump, were gradually plugging the filter elements. The longitudinal as well as membrane DPs had been rising since two days on-stream. The longitudinal DPs for each pair of filters increased from about 12 psi to 25-30 psi for a slurry circulation flow in approximately the same range. The membrane DPs started in the 0-20 psi range and ended up in the 30-70 psi range for the same flux. A membrane DP of about 100 psi was required to get a slightly higher flux. The unit was held on a stand-by until the filter issues could be resolved. Mott Metallurgical was contacted to see if arrangements could be made to replace the current elements with a smaller filter grade (i.e., 0.2 microns instead of 1 micron). In addition, Sundstrand was contacted to see if modifications could be made so that the slurry velocity through the discharge throat of the pump could be reduced. Several attempts were made to backflush the catalyst/wax filters without a significant improvement in filter performance. As 23

the plant could not run without filtering the product wax out of the reactor slurry, it was shut down after being on a stand-by for a day. The plant was cooled, purged, and partially drained so that the filters could be examined internally. Examination of the filters could provide some clues about the cause of the plugging problem. If the plugging was caused by a buildup of the filter cake, and particle size distribution tests confirmed that catalyst attrition was not a problem, we would have some hope of mechanically cleaning the filters and restarting. However, if the catalyst had broken down, penetrated into the annular region of the filter, and plugged, a replacement filter with a smaller micron grade was not available within the current time window. After cooling down, the reactor slurry was drained and the reactor was flushed with oil. Flush oil from the filter system was also drained; it did not show significant solid content. The filters were then taken apart. Although no plugging was visible, backflushing with steam pushed out significant quantities of solids. It was difficult to judge whether the solids came from a cake on the slurry side (inside) of the filters or from the filter membrane annulus. Overnight, the filters were kept on a longitudinal steam flush through the slurry side. A N2 purge was maintained from the clean wax (shell) side. Mechanical cleaning was attempted the next day with a nylon brush to clean the filter elements from inside. Looking inside through a boroscope showed that the brush action was only scraping the layer partially and just moving the material around. Steam backflushing was then resumed, as it was the only method that had worked. Highpressure steam and N2 were used to get higher temperature and pressure on the elements. Inspection of the filters with boroscope indicated that most of the filter cake was flushed out. The system was put together, and a hot oil flush was conducted to remove residual water from the system. Some of the stickiness observed with the solids could be due to agglomeration of particles caused by condensate from steam. The flush oil was then replaced with clean oil, and a flux test was conducted on all the filters. The test revealed a significantly higher pressure drop across the membranes than expected. The pressure drop was in the range of 10 to 40 psi at a flux of 1 lb/min compared to an expectation of 0.1 psi for a new clean element. The longitudinal pressure drops were lower than observed during the run. It seemed that while the filter cake was removed, the membrane remained blocked with particles. Samples taken from the clean wax side appeared to contain fines, indicating breakthrough. Particle size distribution results on various samples from the filter system showed a bi-modal distribution, indicating particle breakdown. The particle size ranged from 0.5 to 100 microns with peaks at 3 and 35 microns. This compares with the fresh catalyst particle size range of 10 to 100 microns with a single peak at 35 microns. According to Mott Metallurgical, a 1 micron membrane can only retain 100% of the particles above ~3 micron particles. The clean oil test clearly showed that the existing elements did not have adequate capacity for a high productivity condition. The particle size distribution provided an explanation of membrane blockage due to smaller than expected particle size. With replacement filters not immediately available, the participants decided to terminate the run at that point and regroup for a second trial later. A plan of action was developed and followed prior to the second trial; it included investigation of catalyst particle breakdown in the laboratory, additional filter tests and slurry pump improvements. SSFI followed up several possible causes of filter problems during F-T III: (1) Strength of their proprietary catalyst used in F-T III vs. strength of their improved proprietary catalyst. (2) Effect of liquid medium on catalyst attrition. (3) Effect of pumping velocity on attrition. (4) Use of 24

different type of filter elements. The initial laboratory test results were inconclusive for (1), (2) and (3). The improved catalyst appeared to be stronger when slurried with water. However, in oil, the two catalysts showed similar attrition. Also, surprisingly, no effect of pumping velocity on attrition was evident. For filtration, the plan was to test three types of filter elements (sintered metal, ceramic membrane and woven metal) in parallel with activated improved catalyst in wax. A Sundstrand centrifugal pump would be used in these tests to evaluate erosion. Commissioning of the filter loop in SSFI's pilot plant was completed and slurry pump/filtration tests started during May 1997. SSFI's improved proprietary catalyst was activated in-situ in wax. Initially, the slurry circulation was started at high pump velocity to generate fines and test the pump for erosion. A more erosion resistant material -- manganese alloy -- was utilized for pump internals. Reliable and steady operation was obtained with only a small sign of erosion. Also, very limited catalyst attrition was observed. For filtration, woven metal filter elements were tested. These elements were most promising as they are uniform. Successful filtration was achieved with flux rate at twice the design rate at low membrane and longitudinal differential pressures. As a result of successful filtration tests, the F-T IV modifications were kicked-off. The modifications included new filter elements, new element bundle arrangement, rebuilding the slurry pump and individualizing the filtration control. The slurry pump modification involved new internals (diffuser and cover plate) made up of manganese alloy, opening the throat to 0.446," use of a differential pressure regulator to maintain proper differential pressure between process and buffer fluid, and installation of a shut-off valve on the buffer system. The cross flow filter system consisted of four 10' filters (four parallel elements in each) in series. The new filter elements were woven metal elements, 14 mm ID, 10 ft long, 10-micron grade stainless steel. A tangential velocity of 8.7 ft/sec, corresponding to 26 gpm of pump flow, would be maintained through the elements. The elements would be back-flushed with clear wax, as needed. For 61 gph of filtrate wax production rate, the design flux through the elements required would be 0.044 gpm/ft2. The total filtration area of the system was 23.1 ft2. With expectation of lower wax production associated with the new catalyst and higher filter capacity as measured by SSFI, it appeared that there was a 100% spare capacity. So it was decided to put only two housings online initially, and withdraw wax from only one of them. For additional backup, 16 extra elements were purchased and constructed into 4 additional bundles. The new bundle arrangement would allow easier on-site replacements of bundles, if the four bundles in service got plugged. Four control valves were installed on the product wax line to improve filtration control by achieving an individual control of each housing. These valves replaced the existing manual throttle valves. The single existing larger control valve was also removed. During F-T IV operations, a productivity of approximately 140 gm HC/hr-liter was achieved at reasonable system stability. The productivity ranged from 110-140 gm HC/hr-liter at various conditions during the 18 days of operations. The catalyst/wax filters performed well throughout the demonstration, producing a clean wax product. For the most part, only one of the four filter housings was needed for catalyst/wax filtration. The filter flux appeared to exceed the design flux. An average filtrate flux of 0.031 gpm/ft2 was demonstrated, compared to the design flux of 0.044 gpm/ft2. However, the system showed significant higher capacity, as the filtrate was withdrawn intermittently during the run. The average flux was calculated based on amount of wax drained from the system, since instantaneous flux data were not available. With the slurry being recycled back from the filtration loop, the upward liquid velocity through the reactor was 25

about 0.024 ft/sec. A combination of use of a stronger catalyst and some innovative filtration techniques was responsible for this success. There was no sign of catalyst particle attrition, and very little erosion of the slurry pump was observed, in contrast to the F-T III operations. The particle size distribution data for the F-T IV run are shown in Figure A.5. The particle size of the catalyst in the slurry samples was hardly different from that of the fresh catalyst precursor and, more importantly, constant in time. Moreover, the production of fines was very limited and seemed to occur mainly at the start-up. These data suggest very good mechanical properties of the catalyst, which was consistent with the excellent filtration performance observed and represents a large improvement over F-T III. The particle size distribution data for F-T III run are plotted in Figure A.6 and show a substantial reduction in average particle size accompanied by high fines formation right from the first sample. This is fully in line with the observed filter plugging during that run.

Figure A.5
Catalyst Particle Size Variation and Fines Formation for FT IV
1 0.9 0.8 0.7 Size / Size Fresh 0.6 0.5 0.4 0.3 0.2 0.1 fresh catalyst precursor size/size fresh fines (%m/m) 10 9 8 7 Fines, wt% 6 5 4 3 2 1

0 0 03/28/98 03/30/98 04/01/98 04/03/98 04/05/98 04/07/98 04/09/98 04/11/98 04/13/98 04/15/98 04/17/98 Date

26

Figure A.6
Catalyst Particle Size Variation and Fines Formation for FT III
1 0.9 0.8 0.7 Size / Size Fresh 0.6 0.5 0.4 0.3 0.2 0.1 0 10/14/96 Fresh catalyst precursor diameter (vol. avg)/ avg) fresh fines, %w diameter (vol.
100 90 80 70 60 50 40 30 20 10 0 Fines, wt%

10/15/96

10/16/96

10/17/96 Date

10/18/96

10/19/96

10/20/96

10/21/96

The reactor operated in a hydrodynamically stable manner, with uniform temperature profile and gas holdups. Nuclear density and differential pressure measurements indicated somewhat higher than expected gas holdup (45 to 50 vol %) during F-T IV operations. The high gas holdup was confirmed by a dynamic gas disengagement test conducted at the end of the run. Heat loss from the reactor was estimated at about 50,000 Btu/hr from data obtained during the 2-phase hot function test. Heat balance during the run was in the 96 to 102% range based on the heat of reaction. Heat transfer coefficients were calculated based on the data obtained during the run. The heat transfer in the reactor was significantly better than expected. Mass and elemental balance calculations also indicated excellent closure. After the initial learning curve with system dynamics, the plant was restarted very quickly (24 hours and 17 hours) following two plant trips. The quick restarts demonstrate the ease and flexibility of the slurry technology. Close-to-expected syngas conversion was obtained at the beginning of the run. The selectivity to wax was lower than expected, with higher methane selectivity. Returning to the baseline condition, towards the end of the run, indicated a 7% productivity decline from 135-140 to 125130 gm HC/hr-liter in two weeks of operation. This may be a result of some catalyst loss from the reactor, as well as initial catalyst deactivation. Gas-, liquid- and solid-phase mixing was studied as planned at two operating conditions using radioactive materials. Synetix (formerly known as ICI Tracerco) set up 43 detectors around the reactor during the last day of the process variable study. The first tracer run was conducted at the baseline condition. Argon-41 was injected into the inlet gas line for the vapor residence time distribution study. In addition, two injections of radioactive manganese oxide were made in the reactor slurry to study liquid phase mixing. Solid injections were also done with radioactive manganese-doped catalyst support to evaluate solid mixing. The second tracer run was 27

conducted after the switch to the high-velocity condition. Washington University in St. Louis analyzed the large amount of data collected by Synetix as part of the SBCR Hydrodynamics project with DOE. The participants collected significant quantities of product and samples during this run for further processing and analysis. Summary and Conclusions The slurry phase F-T technology was successfully demonstrated: A productivity of approximately 140 gm HC/hr-liter of reactor volume was achieved at reasonable system stability during the second trial. As this was more than 90% of the goal of 150, and within the success criteria of 120-150 gm HC/hr-liter, it was decided not to push any further and risk unacceptable system instability. The productivity ranged from 110 to 140 gm HC/hr-liter at various conditions during the 18 days of F-T IV operations. The catalyst/wax filters performed well throughout the F-T IV demonstration, producing a clean wax product. For the most part, only one of the four filter housings was needed for catalyst/wax filtration. The filter flux appeared to exceed the design flux. A combination of use of a stronger catalyst and some innovating filtration techniques was responsible for this success. There was no sign of catalyst particle attrition, and very little erosion of the slurry pump was observed. This was in contrast to the F-T III operations, when the run had to be terminated after seven days on-stream, as the filter membranes apparently plugged with catalyst fines. The reactor operated hydrodynamically stably, with uniform temperature profile and gas holdups. Nuclear density and differential pressure measurements indicated somewhat higher than expected gas holdup (45 to 50 vol %) during F-T IV operations. The high gas holdup was confirmed by a dynamic gas disengagement test conducted at the end of the run. Heat transfer in the reactor was better than expected. Heat, mass and elemental balance calculations indicated excellent closure. After the initial learning curve with system dynamics, the plant was restarted very quickly (24 hours and 17 hours) following two plant trips. This demonstrated the ease and flexibility of the slurry technology. Returning to the baseline condition indicated a productivity decline from 135-140 to 125-130 gm HC/hr-liter of reactor volume in two weeks of operation. This may be a result of some catalyst loss from the reactor, as well as initial catalyst deactivation. The participants collected significant quantities of product and samples for further processing and analysis.

28

Gas, liquid and solid phase mixing were studied as planned at two operating conditions using radioactive materials. Synetix collected large amounts of data using 43 detectors around the reactor. The data were analyzed by Washington University in St. Louis as part of the SBCR Hydrodynamic project with DOE. Specific knowledge gained from this study includes, but is not limited to, the following: Excellent reproducibility was obtained for gas phase tracer runs at a given set of operating conditions due to gas injection prior to the sparger, which ensures a high degree of cross-sectional uniformity. This means that in the future, multiple gas tracer tests at a chosen condition are unnecessary. In contrast, due to the point nature of the liquid (catalyst) tracer injection, multiple injections are needed to properly assess the liquid tracer response. The engineering models developed as part of this contract predicted very well tracer responses without adjustable parameters for runs in churn turbulent flow (run 16.7 in this study). Model predictions were less accurate for flows that may not be churn turbulent (run 16.6), which is in line with the fact that the physics incorporated in the models assures churn-turbulent conditions. The differences in responses for the catalyst and fine powdered manganese oxide tracer injections were minimal, indicating the validity of the pseudo-homogeneous assumption for the liquid (F-T wax) plus solids (catalyst) phase. In contrast, the responses to injections of coarse manganese oxide tracer particles differed dramatically with the responses of the fine catalyst, indicating settling of large particles. The current procedures used in the field for gamma ray scanning lead to significant uncertainties in the chordal average gas holdup estimates, which makes quantitative determination of gas holdup radial profiles difficult and highly inaccurate. A protocol for improved scanning that should lead to accurate assessment of the holdup profile was proposed and specified. This study confirmed that the gas-liquid recirculation and mixing model developed at Washington University in St. Louis does indeed possess the ability to predict tracer responses in F-T synthesis and provides a valuable engineering tool.

Bibliography Bhatt, B. L., "Liquid Phase Fischer-Tropsch Demonstration in the LaPorte Alternative Fuels Development Unit," Topical Report Prepared for DOE by Air Products and Chemicals, Contract No. DE-AC22-91PC90018, June 1994. Bhatt, B. L., "Liquid Phase Fischer-Tropsch (II) Demonstration in the LaPorte Alternative Fuels Development Unit," Topical Report prepared by Air Products and Chemicals for U. S. DOE, Contract No. DE-AC22-91PC90018, September 1995.

29

3.A.4 1999 Liquid Phase Dimethyl Ether Demonstration Run A demonstration of the production of dimethyl ether (DME) by Air Products' Liquid Phase Dimethyl Ether (LPDME) Process was successfully completed in the high-pressure SBCR in October-November 1999. The demonstration was conducted at a pilot scale of 10 TPD to evaluate the commercial viability of the LPDME Process. Syngas produced from coal as well as natural gas can be converted to form DME. DME has potential applications as a diesel substitute, a domestic fuel or a chemical building block. Alternatively, DME synthesis can be incorporated into an IGCC system. Air Products' LPDME Process uses a physical mixture of a commercial methanol catalyst and a commercial dehydration catalyst in a single slurry bubble column reactor (SBCR) to co-produce DME and methanol. This process provides high syngas conversion and efficient heat transfer, and directly converts a variety of feed gas compositions. While the proof-of-concept was demonstrated in 1991 at the AFDU, the catalyst system needed improvements in stability. Researchers at Air Products significantly improved the life of the catalyst system in a study that began in 1994, and the next step was to demonstrate these improvements in a SBCR at a larger scale. This run was co-funded by the Alternative Fuels and Chemicals from Synthesis Gas project and the Clean Coal Technology Program's Kingsport LPMEOH Commercial Demonstration Project. The objectives of the run were to demonstrate the operation of the LPDME Process with improved catalyst life at a 10 TPD scale, using commercially produced catalysts, obtain information to correlate the scaleup of catalyst aging from autoclave to SBCR, conduct process variable testing, and perform experiments to better understand the reactor fluid dynamics. The plant was in the F-T mode after the F-T IV run, and the oxygenates system was last used in 1995 for the fluid dynamic/methanol run; as a result, a significant set-up effort was needed, even though no major modifications were required for this run. Since this was the first DME run with the high-pressure reactor system, as well as with the distributed control and data acquisition systems, heat and mass balance spreadsheets had to be developed, and the data acquisition system had to be set up for DME synthesis. In addition, the analytical set-up required reconfiguration, blinds needed to be switched from F-T to oxygenates, common pumps needed to be relocated, and the reduction circuit and the CO2 removal systems required reactivation. The plant was operated for 25 days to compare catalyst aging in a pilot-scale SBCR with that in a laboratory autoclave. The proportion of two catalysts corresponded to a 95:5 methanol to dehydration catalyst ratio by weight. The methanol and dehydration catalyst activities, expressed as the ratio of the rate constant at any point in time to the rate constant for a freshly reduced catalyst (as determined in the laboratory autoclave), are plotted in Figure A.7. These normalized rate constants were estimated based on a reaction model developed from laboratory data. After the expected initial aging, the catalysts appeared to be stabilizing, but there was significant scatter in the data. Gas chromatographic as well as sampling problems were discovered with methanol analysis that required use of liquid balance for calculations, contributing to the scatter. The problems were corrected in two days. The initial deactivation rate appeared high: 0.08% per hr (2% per day) for methanol catalyst with a 0.05% per hr standard error and 0.03% per hr (0.6% per day) for dehydration catalyst with a 0.009% per hr standard error. It was

30

decided to extend the aging run to get a better estimate on catalyst deactivation rate, which was the main objective. Figure A.7
LPDME at LaPorte (1999) - Estimated Catalyst Actiivity
1.60

Normalized Rate Constant

1.40

MeOH Cat Data Dehy Cat Data MeOH Cat Fit Dehy Cat Fit

1.20

1.00

0.80

0.60 0 100 200 300 400 500 600

Time On-stream, Hours

In order to stay within the budget, the process variable study was eliminated. Operations continued at the baseline conditions for a total of nearly 600 hours on-stream. Overall, the activities of the two catalysts had a similar deactivation rate of about 0.7% per day, which is lower than the 1.2% per day observed for both the catalysts in the autoclave. However, the rate was slightly higher than the 0.5% per day rate achieved after 3 weeks of operation during the 4-month proof-of-concept run of the LPMEOH Process at the LaPorte AFDU in 1988-89. Due to the initial scatter, the standard error was still high for the methanol catalyst: 0.34% per day. This demonstration represented a significant step forward in the development of the LPDME Process. The 0.7% per day aging rate achieved in the AFDU was a large improvement over the 4% per day autoclave deactivation of the previous catalyst system (81:19 methanol to dehydration catalyst ratio by weight). The results of an economic screening study indicated that DME can be produced in an IGCC system at costs approaching LPG pricing in China. The methanol catalyst was successfully activated at the beginning of the LPDME operations with dilute H2, resulting in an expected H2 uptake. The reduction was conducted with 3 vol % H2 in N2 at 67 psig reactor pressure. The heat-up proceeded from 200 to 464F, as planned. The reduction appeared normal, and a cumulative uptake very close to the theoretical maximum value of 2.68 SCF H2/lb oxide was obtained. The reduction was essentially complete at 390F or 17 hours on-stream. The uptake curve was always above the minimum curve, as the reduction in the SBCR was faster compared to the autoclave. NDG measurements indicated an average gas

31

holdup of 36.8 vol %, with a catalyst concentration of 40.1 wt % at 392F during the reduction. The catalysts appeared to have good initial activity, with DME and methanol productivity slightly exceeding expectations. This confirmed that the catalyst activation was proper. The initial DME production rate was 5.1 TPD compared to an expectation of 4.8 TPD, while the methanol production rate was 3.6 TPD vs 3.5 TPD expected. The higher productivities were perhaps due to the effect of multiple CSTRs in the SBCR. The AFDU data appeared to follow the autoclave trends, with somewhat higher conversions than the autoclave throughout the run. The methanol productivity level remained relatively constant, while the DME productivity showed a slight decline. The DME and methanol production rates through the run are shown in Figure A.8. The DME production rate declined from 5.1 TPD to 4.1 TPD in 25 days on-stream, while methanol production showed a scatter within the 3.1 to 3.8 TPD range through the run. The scatter in data decreased significantly after the gas chromatograph (GC) and sampling problems were resolved (350 hours on-stream). The performance also met the productivity and selectivity targets. The initial methanol equivalent productivity exceeded the target of 28 gmole/hr-kg catalyst, and the DME selectivity was at the target of 65% on a carbon basis. The reactor operated in a hydrodynamically stable manner, with uniform temperature profile and gas holdups. DP measurements indicated about 42 vol % gas holdup and 36 wt % catalyst concentration. Figure A.8
LPDME at LaPorte (1999) - Production Results
6.0

5.0

Production, Tons/Day

4.0

3.0

2.0

1.0

Methanol DME

0.0 0 100 200 300 400 500 600

Time On-stream, Hours

Gas-, liquid-, and solid-phase mixing was studied using radioactive tracer injections. An extensive study was conducted at the baseline condition. Several repeat injections were made during the gas and liquid injections to evaluate variability with time. A schematic of the reactor is given in Figure A.1. Synetix (formerly known as ICI Tracerco) set up 34 detectors at various locations outside the reactor. Sets of four detectors at 90 angles were set up at seven different 32

heights. In addition, detectors were set up at the reactor inlet, reactor outlet, vapor space near the reactor top, and recycle feed line. An extra detector was set up on the reactor outlet piping to measure gas velocity. During the liquid injection, a detector was set up at the liquid injection nozzle to monitor the injection pulse. A vapor residence time distribution study was conducted by injecting Argon-41 into the inlet gas line and monitoring its progress through the reactor. The first injection indicated a bad detector. The detector was repaired and seven additional injections were made. Time was allowed between injections for the recycle to stabilize. Excellent pulses were obtained at the inlet, and sharp responses were observed at other locations. Five injections of radioactive manganese oxide mixed in Drakeol-10 were made in the reactor slurry to study liquid phase mixing. All the injections were made at the same location to evaluate variability with time. The location used was nozzle N2-4.5" from wall. Data periods following the injections were varied in the 5-60 minutes range to look for longer term trends, as shown below. Four injections of a slurry of radioactive manganese oxide-doped gamma alumina in Drakeol-10 were made to study gamma alumina mixing. The injections were made at four different locations: nozzle N2-4.5" from wall, nozzle N2-wall, nozzle N1-4.5" from wall, and nozzle N1-wall. Both short-term and long-term observations of irradiated dehydration catalyst suggested no alumina settling in the reactor. A post-run inspection of the reactor bottom head did not show any settled catalyst around the sparger, in contrast to the 1991 DME run when significant settled catalyst was found at the bottom. Washington University in St. Louis analyzed the tracer data as part of the SBCR Hydrodynamics Program with DOE. The oxygenates system containing the high-pressure reactor was operated after a 4-year hiatus; the operations were smooth. The initial start-up was very quick, with the baseline condition reached in 12 hours after the introduction of syngas. A re-start after a syngas outage took only 4 hours. The speed of the start-up and re-start demonstrates the ease and flexibility of the slurry technology in response to changes. Approximately 27,200 gallons of liquid product (89 wt % methanol, 7 wt % DME) were collected during the run. Summary and Conclusions The following conclusions can be drawn based upon the results of this operating campaign:

Operation of the LPDME Process with improved catalyst life was successfully demonstrated on a 10-TPD scale, using commercially produced catalysts. The AFDU was operated for 25 days to compare catalyst aging in a pilot-scale SBCR with that in a laboratory autoclave. The catalyst life study was extended, replacing a planned process variable study, in order to obtain additional data on catalyst aging. Hydrodynamic information was obtained at the baseline conditions by conducting a detailed survey of the reactor with radioactive tracer injections. The catalysts were activated successfully with an expected H2 uptake. The initial productivities of methanol and DME were higher than those found in the laboratory, perhaps due to the multiple-CSTR effect in the SBCR. The DME production rate started at 5.1 TPD and declined to 4.1 TPD in 25 days on-stream, while the methanol production rate showed a scatter within the 3.1 to 3.8 TPD range through the run. The scatter in data decreased significantly after a GC problem and a sampling problem were discovered and resolved (350 hours on-stream). 33

The deactivation rate for both the catalysts was estimated at 0.7% per day. This was lower than the 1.2% per day rate observed for both the catalysts in the autoclave. However, the rate was slightly higher than the 0.5% per day rate achieved for the LPMEOH Process after 3 weeks of operation at LaPorte. The methanol productivity level remained relatively constant throughout the run, while the DME productivity showed a slight decline, consistent with laboratory observations. The standard error for the methanol catalyst deactivation rate was high (0.25% per day) due to scatter in methanol data. The dehydration catalyst activity data were much tighter, with a standard error of 0.06% per day. Spent catalysts from the run showed expected properties. The activity results of the spent catalysts supported the stable performance observed during the run. Elemental analysis of the spent catalyst samples showed a typical composition, with no evidence of accumulation of catalyst poisons. The reactor operated hydrodynamically stably, with uniform temperature profile and gas holdups. DP measurements indicated about 42 vol % gas holdup and 36 wt % catalyst concentration. Mass balance calculations showed good closure. The initial start-up was very quick; the baseline condition was reached in 12 hours after the introduction of syngas. A re-start after a syngas outage required only 4 hours. This demonstrated the ease and flexibility of the slurry technology. Gas-, liquid-, and solid-phase mixing was studied at the baseline conditions using radioactive materials. A large amount of data was collected using 34 detectors around the reactor. Several repeat injections were made during the gas and liquid injections to evaluate variability with time. Manganese oxide-doped gamma alumina was injected at four different locations to examine dehydration catalyst mixing. Both short-term and longterm observations of irradiated dehydration catalyst suggested no alumina settling in the reactor. Results from analysis of these tracer results by the DOE's Hydrodynamics Program reinforce earlier findings regarding the application of the model for the SBCR. Best model predictions are achieved for churn turbulent flows at the highest gas velocities and the lowest pressures. The 1999 LPDME run seemed to fall in the transition zone, since pressure was high and exit gas superficial velocity was relatively low. Also, there were no differences in detected tracer responses to fine powder and doped catalyst injections, confirming that the slurry can be treated as a pseudo-homogeneous phase, as is assumed in the model. The oxygenates system was operated after a 4-year gap; the operations were smooth.

This demonstration represents a significant step forward in the development of the LPDME Process. The 0.7% per day aging rate achieved in the AFDU is a large improvement over the 4% per day autoclave deactivation of the previous catalyst system. 34

Bibliography "Development of Alternative Fuels from Coal Derived Syngas, Task 2.2: Demonstration of a One-step Slurry-Phase Process for the Production of Dimethyl Ether/Methanol Mixtures at the LaPorte Alternative Fuels Development Unit," Topical Report Prepared for DOE by Air Products and Chemicals, Contract No. DE-AC22-91PC90018, 1 June 1993.

35

3.B Task 3 Research and Development Activities 3.B.1 Introduction The objective of Task 3, Research and Development, was to conduct laboratory research and development to investigate new technologies for the conversion of syngas to chemicals, chemical intermediates, oxygenated fuels, fuel components and hydrocarbon fuels, with the goal of demonstrating the technologies on a larger scale. The areas of study included: improved catalysts and processes for dimethyl ether (DME) synthesis; catalysts and processes for conversion of syngas into value-added chemicals; catalysts and processes for higher alcohols synthesis; DME conversion to fuels and chemicals; gas cleanup and catalyst poisons testing in a newly constructed, transportable laboratory; and mathematical modeling of reaction pathways. 3.B.2 Improved Processes for Dimethyl Ether Synthesis Under the previous Alternative Fuels From Synthesis Gas project with DOE, Air Products started research and development of the Liquid Phase Dimethyl Ether (LPDME) Process in the late 1980s. The objective was to develop a new technology that could produce DME from syngas at a lower cost compared with traditional and other alternative processes. The potential cost saving lies in the two essential characteristics of the LPDME Process: high reactor productivity and superior heat management (see detail below). These two characteristics were proven in the laboratory and demonstrated at the LaPorte demonstration plant in 1991 (Bhatt, 1992; Air Products, 1993). However, this early work also revealed that the stability of the catalyst system was insufficient to warrant a viable commercial process. In the current project we investigated the possible causes of catalyst deactivation under LPDME conditions, and identified a novel catalyst deactivation mechanism. We further developed the methods to mitigate catalyst deactivation and identified stable operating conditions for commercial applications. Stable LPDME operation was successfully demonstrated at the LaPorte AFDU in 1999. In the current program, we also developed a kinetic understanding of the LPDME reaction system, and applied this understanding to optimize the LPDME Process and improve its economics. At the end of the program, we arrived at a pilot-plant-tested LPDME Process and a competitive product cost. Air Products Laboratories Air Products maintains centralized R&D laboratories and dedicated R&D support service groups located in Allentown, Pennsylvania to support the company's Alternative Fuels and Chemicals projects. Three laboratories contain four 300-cc continuous-stirred-tank reactors, each with pressure limits of 2000 psi and temperature limits of 325C. All are capable of providing benchscale data that can be used to evaluate the performance of the slurry reactors at the AFDU. In addition, there are fully equipped laboratories to study heterogeneous catalysts and gas/solid 36

catalytic reactions, which are dedicated to studies of methanol and DME catalysts for the liquidphase processes. A catalyst preparation lab provides bench-scale quantities of new catalysts that can be sent to commercial catalyst vendors to produce the 1-ton quantities required for testing in the AFDU. Why DME and Why Does LPDME Offer Potentially Low-Cost DME Production? DME has been heralded as a fuel with great potential. Among the proven and developing applications are power generation, home use (e.g., as a LPG substitute for cooking) and transportation (as a diesel substitute) (Fleisch et al., 2001). Production of DME from syngas is one of a very few technologies that are under development for natural gas utilization (gas to liquids or GTL). Syngas-to-DME may also provide a route for converting coal into clean energy in countries that are poor in oil and natural gas, but rich in coal (Chen et al., 1994). One of the key issues in developing DME into a fuel is the manufacturing cost. The LPDME Process is aimed at lowering DME production cost. In the LPDME Process, syngas is converted into DME in a single SBCR over a dual-catalyst system consisting of a methanol synthesis catalyst and a methanol dehydration catalyst. There are three simultaneous reactions in this system, namely, Methanol synthesis: Water gas shift: Methanol dehydration: CO + 2H2 CH3OH CO + H2O CO2 + H2 2CH3OH CH3OCH3 + H2O (1) (2) (3)

The methanol synthesis catalyst catalyses Reactions (1) and (2), and the dehydration catalyst catalyzes Reaction (3). Both catalysts are micron-sized powders suspended in a slurry fluid (e.g., hydrocarbon oil). Two essential characteristics of this process may lead to low-cost DME production. First, chemical synergy among the three reactions gives the process greater single-pass syngas conversion or reactor productivity (Peng et al., 1999a) compared to the traditional two-step process, in which methanol is formed from syngas in a methanol synthesis reactor, followed by dehydration into DME in a second reactor. This greater reactor productivity translates into low capital and operating cost for the syngas conversion reactor and recycle loop. The second characteristic of the process is the Slurry Bubble Column Reactor that provides superior heat management to accommodate the large amount of heat released from the reactions as a result of the enhanced productivity, making the cost saving practical. Investigation of the Mechanism of Catalyst Deactivation under LPDME Conditions A stable LPDME catalyst system should improve the economic viability of the LPDME Process. Our first step in developing stable operation was to develop an understanding of the cause of catalyst deactivation that had been seen in our early developmental work. The early LPDME work used a dual-catalyst system that contained a physical mixture of a commercially available, Cu-based methanol catalyst and -alumina as the methanol dehydration catalyst. This catalyst system deactivated rapidly under LPDME conditions (Bhatt, 1992). As shown in Figure B.1, line (a), the methanol equivalent productivity, defined as methanol productivity plus two times the DME productivity, dropped 50% after 400 hours on stream. This drop was due to

37

the simultaneous deactivation of both the methanol synthesis catalyst and the methanol dehydration catalyst; and both catalysts deactivated more rapidly when used in a dual-catalyst system than when they were used separately for methanol synthesis and methanol dehydration (Peng, 1997a and 1997b). A systematic investigation was carried out to understand this accelerated catalyst aging. The factors that were investigated included hydrothermal sintering, leaching, coking, the effect of DME and water, and the compatibility between the methanol synthesis and methanol dehydration catalysts. Figure B.1 Improvement in the Stability of the LPDME Catalyst System: (a) Previous and (b) Current
1.0

Relative MEOH Equiv. Productivity

0.9

(b)

0.8

0.7

0.6

0.5

250 C, 750 psig, 6000 GHSV, Shell gas

(a)

0.4 0 100 200 300 400 500

Time on stream (hr)

Hydrothermal sintering was eliminated as the cause of accelerated catalyst deactivation under LPDME conditions by the following reasoning. Figure B.2 displays normalized catalyst activity as a function of copper crystallite size. A linear correlation is observed for the spent methanol catalysts from the stable LPDME experiments (both laboratory and LaPorte plant) and laboratory LPMEOH experiments. This characteristic decay curve is taken as being indicative of a normal deactivation process for the methanol catalyst by hydrothermal sintering of copper. However, the results from the LPDME experiments with accelerated catalyst deactivation fall outside this correlation. Catalyst activity is much lower for a given copper crystallite size. We infer that a mechanism other than hydrothermal sintering is responsible for the accelerated catalyst aging under LPDME conditions. Reports covering the great number of experiments carried out to investigate the other possible causes of accelerated catalyst deactivation under LPDME conditions were issued (Peng, 1997a and 1997b). Leaching of active components from the methanol catalysts into the slurry oil was eliminated as a possible cause of deactivation by careful and repeated analysis of Cu and Zn in

38

the spent slurry oil. Carbon or coke deposition, either due to the Boudouard reaction on the methanol catalyst or acid-catalyzed reactions on the dehydration catalyst, was investigated and ruled out. No coke deposition was detected on the spent LPDME catalysts, even using stateof-the-art techniques. Since the catalyst samples were opaque, a new uv-Raman technology that was then under development at Northwestern University was employed for this study, but revealed no coke deposition. Water was eliminated as a possible source of accelerated deactivation, since no correlation between the water level and the rate of catalyst deactivation was observed. To ensure that this observation was correct, the effect of water on catalyst stability was investigated by changing the H2:CO ratio in the feed gas. These experiments resulted in different levels of water in the reactor via the water gas shift reaction, but no correlation with deactivation was found. The effect of DME per se on the stability of the methanol catalyst was studied by injecting DME during an otherwise standard LPMEOH experiment. The increase in DME had little effect on the kinetics or the stability of the methanol catalyst. Since most of the DME sources contain trace amounts of halogen species, which are potent poisons to the methanol catalyst, considerable efforts were made during the DME injection experiments to realize this artifact and identify a clean DME source. Figure B.2 Methanol Catalyst Activity vs. Copper Crystallite Size

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 70 80 90 100 110 120

Lab LPDME, stable Lab LPDME, unstable LaPorteLPDME, stable Lab LPMEOH

km/k

130

140

Copper Crystallite Size (Angstrom)

Finally, having rejected the standard explanations for catalyst deactivation, we proposed a novel mechanism, detrimental interaction between the two catalysts. An experiment in which the compatibility of the two catalysts was isolated from all other factors that may contribute to the catalyst deactivation (Peng, 1997a) was performed. The experiment showed that the mere co39

existence of the two catalysts caused each other to lose considerable activity. It was speculated that the detrimental interaction between the two catalysts is due to migration of Cu- and, possibly, Zn-containing species from the Cu-based methanol catalyst to the dehydration catalyst (e.g., -alumina) under LPDME conditions. This migration hypothesis can explain the simultaneous deactivation of the two catalysts. That is, the methanol synthesis catalyst loses its active components in this process, and the acid sites on the dehydration catalyst are poisoned by the migrating species. The likelihood of this hypothesis was supported by the correlation between the rate of catalyst aging and the rate of Cu accumulation on the dehydration catalyst, as shown in Figure B.3. Analysis of Cu content in the spent dehydration catalyst required a tedious procedure. It included cutting open the micron-sized, spent -alumina powders to expose the cross sections, analyzing hundreds of spots on many cross sections using scanning electron microscopy (SEM) and energy dispersion spectroscopy (EDS), and integrating hundreds of weak Cu EDS peaks manually. Figure B.3 Catalyst Deactivation vs. Copper Accumulation
1.0

0.8

Dehydration catalyst Methanol catalyst

Catalyst Aging Rate (%/hr)

0.6

0.4

0.2

0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6

correlat ion

0.7

Cu Accumulation Rate (a.u.)

Developing Stable LPDME Operation Identification of the mechanism of accelerated catalyst deactivation under LPDME conditions laid the foundation for solving the catalyst stability problem. However, developing solutions proved to be as challenging. It took a somewhat tortuous road and considerable efforts for us to finally achieve stable LPDME operation. These efforts included (1) extensive screening of alternative dehydration catalysts, (2) the work with aluminum phosphate as the dehydration catalyst, (3) development of the understanding and correlation between catalyst stability and reaction conditions using the aluminum phosphate-containing dual catalyst system, and (4) 40

application of this correlation to the -alumina-containing dual catalyst system. All of this work, as described in more detail below, led to the successful demonstration of LPDME catalyst stability at the LaPorte AFDU in 1999 (Bhatt, 2000 and Section 3.A.4). Since the mechanistic study discussed above demonstrates that the Cu-based commercial methanol catalyst and -alumina in our original dual-catalyst system are not compatible with each other, the solution we pursued initially was to find an alternative dehydration catalyst that is compatible with the methanol catalyst. Twenty-nine different dehydration materials were tested, but none showed satisfactory stability in combination with good activity. However, this screening process demonstrated that the strength and type of the acid sites on dehydration materials have different effects on the stability of the dual catalyst system. Strong and Brnsted -type acid materials (e.g., zeolites) impair the methanol catalyst more severely than the weak and Lewis-type acid materials. The former are also more vulnerable under LPDME conditions themselves (Peng, 1997a). These observations furthered our understanding of the interaction of methanol synthesis and dehydration catalysts under LPDME conditions. Stable LPDME performance was finally observed when amorphous aluminum phosphate was tested as the dehydration catalyst. A large amount of work was followed to improve the activity of the aluminum phosphate catalyst and understand the relationship between the preparation parameters and the performance of the aluminum phosphate-containing, dual-catalyst system. It was found that the preparation affects the activity and stability of the aluminum phosphate catalyst and its impact on the stability of the methanol catalyst (Peng, 2000a). The extensive synthesis and testing work led to an optimal preparation and formulation. A plan for demonstrating LPDME stability at the LaPorte AFDU using the aluminum phosphatecontaining, dual-catalyst system was formed, and scaleup of the aluminum phosphate catalyst was carried out accordingly (Wang, 2002). However, this plan and aluminum phosphate scaleup were discontinued owing to the further understanding we developed as discussed below. During the process of scaling up the aluminum phosphate catalyst, screening the conditions for the demonstration and switching to a new commercial Cu-based methanol catalyst, unstable LPDME performance was observed from time to time, even with the best aluminum phosphate material. A series of experiments was designed to determine whether the unstable performance is due to something "physical" (catalyst agglomeration or loss of catalyst from the slurry due to laboratory artifacts) or "chemical" (chemical processes that lead to real deactivation). The investigation showed that the deactivation is real. This led to an important realization in our LPDME catalyst research: the rate of catalyst deactivation under LPDME conditions depends on the reaction conditions. A systematic study was then initiated to understand the relationship between the rate of catalyst deactivation and reaction conditions, using the aluminum phosphate-containing, dual-catalyst system. Data were collected under 32 different conditions. With this database, a quantitative correlation between the rate of catalyst deactivation and reaction conditions was established as shown in Figure B.4; the catalyst aging rates predicted from the reaction conditions agree well with those observed from the experiments. The development of the correlation proved to be an important milestone in our research. First, it showed that methanol is a stabilizing reagent for the catalyst system. This led to a process scheme for achieving stable LPDME operation 41

(Peng, 2000). Second, further experiments showed that the correlation applies to a -aluminacontaining, dual-catalyst system; stable operation is possible with a -alumina-containing, dualcatalyst system if it is in a stable regime. One such stable operation is shown in Figure B.1, line (b). This understanding brought -alumina, a commercially available and inexpensive material, back to our LPDME Process. Third, the correlation can be used quantitatively in the CO-rich regime to map out stable LPDME operating conditions, including those relevant to commercial applications. Therefore, it serves as a tool for process design. Figure B.4 Correlations between Catalyst Aging Rate and Reaction Conditions

0.6

Predicted Catalyst Aging Rate (a.u.)

0.5

0.4

0.3

0.2

0.1

0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6

Measured Catalyst Aging Rate (a.u.)

Other approaches were also attempted to solve the catalyst stability problem, including

modifying the methanol catalyst and developing a dual-functional LPDME catalyst system. However, neither proved to be fruitful. The modifications included doping the commercial methanol catalyst with promoters (e.g., SiO2) aimed at impeding the mobility of Cu and Zn species in the methanol catalyst, and acid washing aimed at removing mobile species in the methanol catalyst. Development of a dual-functional catalyst was a three-year joint effort between Air Products and a commercial methanol catalyst manufacturer. The intent was to form both methanol synthesis and methanol dehydration functionalities in a single catalyst. It was hoped that some meta-stable equilibrium could be established between the two functionalities in a single catalyst, therefore, mitigating the detrimental interaction between them. Nearly 60 preparations were conducted in the labs of the methanol catalyst manufacturer and tested at Air Products; none yielded better performance than the dual-catalyst system.

42

Preparation for the 1999 LaPorte LPDME Demonstration The 1999 LaPorte LPDME Process demonstration was planned according to the stable laboratory LPDME results discussed above. It would use a -alumina-containing, dualcatalyst system, perform a catalyst life study under the baseline LPDME conditions (250oC, 750 psig, 6,000 gas hourly space velocity (GHSV) and a feed gas simulating the Shell gasifier), and test several potentially commercially relevant conditions. Therefore, our preparation consisted of three parts: (1) identify a commercial -alumina material, (2) conduct a simulating baseline life run in the laboratory and (3) identify a number of stable, commercially relevant LPDME conditions for demonstration using the correlation and confirm them in the laboratory. Both (2) and (3) would use the demonstration materials. -alumina materials from three different suppliers were screened. The results showed that not all -alumina materials are equal and the material still is important to the catalyst stability. Even under the stable operating conditions, some -alumina samples would give unstable performance. It depended on both the supplier and the batch of preparation. A commercial -alumina material was finally identified. In addition to the baseline life run, several stable, commercially relevant conditions were identified using the correlation and tested successfully in the laboratory using the demonstration materials. These included the baseline syngas with different space velocities, a different CO-rich syngas and a syngas feed containing equal moles of H2 and CO. These preparations were instrumental in the success of the 1999 LPDME Process demonstration. We also spent considerable time resolving a so-called "30 gram problem" during the preparation for demonstration. The stable LPDME operations we have discussed thus far were all observed in the laboratory 300 cc autoclave using the standard catalyst loading, 10 grams. To match the high catalyst loading that would be used in the LaPorte LPDME demonstration, we tested the baseline life study conditions several times using 30 grams of catalyst. In contrast to the experiments using 10 grams of catalyst, the experiments using 30 grams of catalyst all exhibited unstable performance. A series of experiments finally identified this "30 gram problem" as a laboratory artifact. Although this effort had no direct bearing on the demonstration, it showed our honest approach for preparing the demonstration -- anticipating and resolving possible problems. Furthermore, the knowledge of the laboratory artifact warned us about possible misleading results from laboratory reactors, and eventually led to a later study on laboratory artifacts. Kinetic Understanding of the LPDME Reaction System Solving the catalyst stability problem was an important step in developing a viable, low-cost process for DME production from syngas. Further efforts were made to improve LPDME Process economics by developing and optimizing process schemes. The kinetic understanding of the LPDME reaction system we developed during this project provided the guidelines for these efforts. The first insight from the kinetic study to the LPDME Process is the sensitivity of reactor productivity toward reactor feed composition. This sensitivity is a result of the interplay among the three reactions in the LPDME system: Reactions (1), (2) and (3) shown above. These three reactions form a complex, highly non-linear network. One of the net effects of the interplay among the reactions is the dependence of the chemical synergy in the LPDME 43

reaction system on the composition of reactor feed. Figure B.5 displays the methanol equivalent productivity of the LPMEOH and LPDME Processes as a function of the H2:CO ratio in the reactor feed. The productivities of the two processes are compared on a common basis, the methanol equivalent productivity or MEP, defined as the methanol productivity plus two times the DME productivity. Also plotted in Figure B.5 is the increase in MEP from LPMEOH to LPDME (i.e., the chemical synergy) for the same feed composition. It can be seen that the greatest synergy is observed at the CO-rich end of the feed. A thorough analysis has been made to understand this dependence (Peng, 1999a and 2002b). It shows that the CO-rich region is where all of the factors that contribute to the chemical synergy work the best. These factors include consuming the product from methanol synthesis by methanol dehydration, and lowering the water level and replenishing H2 through the water gas shift reaction. This study has direct implications on how the LPDME Process should be used in once-through applications. Figure B.5 The Methanol Equivalent Productivity (MEP) from LPDME ( ), the MEP from LPMEOH (), and the % Increase in MEP from LPMEOH to LPDME ( ) as a Function of H2:CO Ratio in the Syngas Feed
160 140 120 100 80 20 60 40 20 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 4.5

40

Inlet H2:CO

The kinetic study also showed why the productivity of the LPDME reactor depends on the feed gas composition and where the maximum productivity is located for a given set of reaction conditions. For the conditions shown in Figure B.5, the maximum productivity is observed at a H2:CO ratio of around 1. In general, this maximum varies with reaction conditions. For example, it shifts from a H2:CO ratio of 2 to 1 with decreasing space velocity. A mathematical approach was developed to understand this dependence. The approach breaks up the rate of a reaction into its kinetic and thermodynamic components and separately studies their dependence on the reactor feed composition. The analysis reveals that the best H2:CO ratio for the thermodynamics of the LPDME reaction system is 1, while that for the kinetics of the reaction

44

MEP Increase from LPMEOH to LPDME (%)

MEOH Equiv. Prod. (mol/kg-hr)

250 C, 750 psig, 6000 GHSV Once-through operation Simulated results

system shifts from 2 to 1 with decreasing space velocity (Peng, 2002b). The understanding obtained from this type of study provided us with guidelines for optimizing the reactor productivity, and therefore, the LPDME Process (see below). The approach was also used to analyze the optimal reactor feed composition for the syngas-tomethanol process (Peng, 2002b). The optimization is captured in gas phase literature by a single parameter expressing the relationship between feed components called the modulus, M. It was determined that the optimal feed moduli are:
0 0 N H 2 1 N CO 2 =2 M= 0 0 N CO + 1 N CO 2

for H2-rich syngas with NoH2O = 0

(4)

and

M=

0 0 N H 2 + 2 N H 2O =2 0 0 N CO 2 N H 2O

for CO-rich syngas with NoCO2 = 0

(5)

Ni0 in Eqs. 4 and 5 is the moles of species i in the reactor feed 1=2/(2+Kw) and 2 = Kw /(2+Kw) with Kw being the equilibrium constant of the water gas shift reaction. This is the first time, to our knowledge, that the optimal feed modulus for methanol synthesis has been analyzed from the reaction rate point of view, as opposed to the empirical module based on material balance. It is also the first time that an optimal feed modulus for methanol synthesis from CO-rich syngas has been proposed and derived. It provides guidelines for methanol synthesis directly from CO-rich syngas. Development of Optimal LPDME Process Schemes and Economics The very possibility of low-cost DME production by the LPDME Process lies in the chemical synergy, i.e., the enhanced reactor productivity compared to the syngas-to-methanol reactor. This enhanced productivity translates into significantly lower capital and operating cost associated with the synthesis loop. For stand-alone, high-conversion applications, its economic impact can be represented by the MEP curve shown in Figure B.6. This curve depicts the enhancement in LPDME reactor productivity as a function of the H2:CO ratio in the reactor feed, compared to the productivity of the syngas-to-methanol reactor under its optimal feed composition (H2:CO = 2), for a set of typical high-conversion conditions. It can be seen that the largest enhancement is achieved with the H2:CO ratio in the reactor feed around 1.

45

Figure B.6 The Trade-off between Enhanced Productivity and Resulting CO2 Separation

[CO 2 ]
20

15

10

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

H 2 :CO in Reactor Feed

This result indicates that to fully utilize the economic benefit of the LPDME Process, the reactor should be run with 1:1 H2:CO feed. How to apply this understanding in practice offered technical challenges, since most commercial syngas generation technologies do not provide syngas with equal molar H2 and CO. For coal-derived CO-rich syngas, the shortage of H2 can be supplemented by water injection to the LPDME reactor. For natural gas-derived, H2-rich syngas, process schemes have been developed to achieve overall H:C balance, while the reactor is operated with the 1:1 feed (Peng, 1999b). A general scheme is shown in Figure B.7; it is based on the integration between the LPDME synthesis loop and the syngas generation unit. A further advantage of this integration is zero CO2 emissions. Simulations using realistic kinetics and commercially relevant conditions verified the feasibility of integrating the LPDME Process with different syngas generation technologies. Further improvement in LPDME Process economics came from our efforts in minimizing the cost associated with CO2 separation. CO2 is formed in the LPDME reactor due to the water gas shift reaction. Separation of CO2 from the reactor recycle stream and final DME product adds cost to the process. Figure B.6 shows that the amount of CO2 formation closely tracks the enhancement of reactor productivity. This indicates that there is an inherent trade-off in the LPDME Process between the potential cost saving (enhanced reactor productivity) and cost addition (CO2 separation). Therefore, in developing an economic LPDME Process, one needs to take both factors into consideration. One approach is to operate the reactor in the optimal kinetic regime (H2:CO in the reactor feed around 1), while devising an economic way for CO2

46

[CO2] in Reactor Effluent (mol%)

40

MEP

%Enhancement in MEP

2000 GHSV o 250 C 52 atm once-through

20

Figure B.7 Integration of Syngas to DME with Syngas Generation

Reformer purge 3 CH4 CO2 H2O O2 Methane Methane to reformer syngas CO, H2 2 H2, CO Coal gasifier 4 CO, H2 5 H2, CO, CH3OH, H2O 7 DME LPDME reactor Product separation 8 Reactor purge CO2 9

CH3OCH3

Coal, O2

separation. This approach provides the projected lowest cost DME process industry-wide (Peng, 2002c). Summary and Conclusions In summary, the R&D efforts in the Alternative Fuels and Chemicals From Synthesis Gas project have improved the technical and commercial viability of the LPDME Process:

The rapid catalyst deactivation observed during the 1991 LPDME run at the LaPorte AFDU has been mitigated, and stable LPDME operation under various conditions, including those relevant to commercial applications, has been achieved in the laboratory. This improvement in catalyst performance was demonstrated during the 1999 LPDME campaign at the LaPorte AFDU (detailed in Section 3.A.4). New process schemes have been developed to fit the LPDME technology to potential markets for coal as well as natural gas utilization. The R&D activities in this program were also marked by the development of a more fundamental and thorough understanding of the catalysis, kinetics and process of the LPDME technology. A systematic study uncovered the novel deactivation mechanism of the LPDME dual-catalyst system, and eliminated many others. Considerable efforts were made to develop an understanding of the catalyst deactivation rate as a function of reaction conditions. The correlation that was further developed on the basis of this understanding provided a map of the stable LPDME operation schemes and conditions. We also developed a fundamental understanding of the kinetics of the LPDME reaction system and a mathematics approach for analysis of the optimal reactor feed composition. The knowledge acquired from these efforts informed us of the optimal reactor operating

47

conditions, revealed the inherent trade-off between the two most important factors in the LPDME Process (reactor productivity and CO2 separation), and led to optimal process schemes and good economics. Bibliography Air Products and Chemicals, Inc., "Demonstration of a One-Step Slurry-Phase Process for the Production of Dimethyl Ether/Methanol Mixtures at the LaPorte Alternative Fuels Development Unit," APCI Topical Report to DOE under Contract No. DE-AC22-91PC90018, 1993. Bhatt, B., "Synthesis of Dimethyl Ether and Alternative Fuels in the Liquid Phase from Coal Derived Synthesis Gas," APCI Topical Report to DOE under Contract No. DE-AC2290PC89865, 1992. Bhatt, B. L., Toseland, B. A., Peng, X. D., and Heydorn, E. C., "Catalyst and Process Development for Liquid Phase DME Synthesis," Symposium on "C1 Chemistry," 17th Annual International Pittsburgh Coal Conference, Pittsburgh, PA, September 11-15, 2000. Chen, Z. H., and Niu, Y. Q., "The Status Quo and Trends of the Domestic Liquid Fuel in China," Coal Chemical Industry 67 (1994) 1. Fleisch, T.H., Puri, R., Sills, R. A., Basu, A., Gradassi, M., and Jones, G. R. Jr., "Market Led GTL: The Oxygenate Strategy," Stud. Surf. Sci. Catal. 136 (2001) 423. Peng, X. D., "Catalyst Activity Maintenance for the Liquid Phase Synthesis Gas-To-Dimethyl Ether Process: Part I: An Investigation of the Cause and Mechanism of Catalyst Deactivation," APCI Topical Report to DOE under Contract No. DE-FC22-94 PC93052, 1997a. Peng, X. D., Toseland, B. A. and Underwood, R. U., "A Novel Mechanism of Catalyst Deactivation in Liquid Phase Synthesis Gas-to-DME Reactions," Stud. Surf. Sci. Catal. 111 (1997b) 175. Peng, X. D., Toseland, B. A. and Tijm, P. J. A., "Kinetic Understanding of the Chemical Synergy under LPDME Conditions Once-Through Applications," Chem. Eng. Sci. 54 (1999a) 2787. Peng, X. D., Wang, A. W., Toseland, B. A. and Tijm, P. J. A., "Single-Step Syngas-to-Dimethyl Ether Processes for Optimal Productivity, Minimal Emissions, and Natural Gas-Derived Syngas," Ind. Eng. Chem. Res. 38 (1999b) 4381. Peng, X. D., Wang, A. W., and Toseland, B. A., "Single Step Synthesis Gas-To-Dimethyl Ether Process with Methanol Introduction," U.S. Patent 6,069,180 (2000). Peng, X. D., "Catalyst Activity Maintenance for the Liquid Phase Synthesis Gas-To-Dimethyl Ether Process: Part II: Development of Aluminum Phosphate as the Dehydration Catalyst for the Single-Step Liquid Phase Syngas-To-DME Process," APCI Topical Report to DOE under Contract No. DE-FC22-94 PC93052, 2002a. Peng, X. D., "Kinetic Understanding of the Syngas-To-DME Reaction System and Its Implications to Process and Economics," APCI Topical Report to DOE under Contract No. DEFC22-94 PC93052, 2002b.

48

Peng, X. D., Diamond, B. W., Tsao, T.-C., and Bhatt, B. L., "A Separation Process for One-Step Production of Dimethyl Ether from Syngas," U.S. Patent, Application No. 10/008261, allowed (2002c). Wang, A.W., "Scaleup of Aluminum Phosphate Catalyst for Pilot Plant LPDME Run," APCI Topical Report to DOE under Contract No. DE-FC22-94 PC93052, January 2002. 3.B.3 Value-Added Chemicals from Synthesis Gas Vinyl Acetate Monomer Introduction There has been a long-standing desire on the part of industry to replace the existing ethylenebased vinyl acetate monomer (VAM) process with an entirely syngas-based process. Although there are a large number of process options for the conversion of syngas to VAM, Eastman Chemical Company (Eastman) undertook an analytical approach, based on known chemical and economic principles, to reduce the potential candidate processes to a select group of eight processes. The critical technologies that would be required for these routes were: 1. 2. 3. 4. the esterification of acetaldehyde (AcH) with ketene to generate VAM, the hydrogenation of ketene to acetaldehyde, the hydrogenation of acetic acid to acetaldehyde, the reductive carbonylation of methanol to acetaldehyde.

Historically, the esterification of acetaldehyde with acetic anhydride (see Reaction (6)) played a very important role in the generation of vinyl acetate: Ac2O + (acetic anhydride) AcH VAM + (acetaldehyde) (vinyl acetate) AcOH (acetic acid) (6)

For example, the Celanese Corporation of America operated a facility for this conversion from 1953 until 1968 in Pampa, Texas. (It was ultimately displaced by the current commercially practiced oxidative acetoxylation of ethylene.) A closer examination of the reaction reveals that the conversion depicted in Reaction (7) actually is composed of two reactions, both of which are equilibria. These are shown below: Ac2O + AcH (AcO)2CHCH3 (ethylidene diacetate, EDA) VAM + AcOH K140C = 25 (7)

(AcO)2CHCH3

K140C = 0.01 (8)

(The equilibrium constants were obtained from Halcon SD.) The fact that these equilibria are unfavorable for the formation of vinyl acetate has always been problematic. Further, as one heats the reaction, despite the favorable equilibrium, Reaction (7) has a tendency to reverse quickly, since the very low boiling acetaldehyde is preferentially distilled out of the reaction mixture

49

while any trace of catalyst is still present. Celanese used large excesses of acetic anhydride to push the equilibria further in favor of EDA to reduce this problem, but this practice required the recycle of large volumes of acetic anhydride. Prior to participating in this DOE program, Eastman had conceived a novel approach to improving this process and reduced it to practice on a bench scale. The concept sought to simplify the overall process and increase the per-pass yields by applying principles of reactive distillation. A reactive distillation column is best applied to processes involving two reactants that are undergoing dynamic equilibria (one of the reactants must be higher boiling than the desired final product). The column consists of two parts, a reaction zone and a separation zone. The two reactants are introduced to the column at different points: the highest boiling reactant and, if necessary, catalysts are introduced at a point above the introduction of lower boiling starting material. Introduction of the higher boiling reactant above the lower boiling component creates a unique component profile in the column. Essentially, this is an inversion of a normal column, in which a high-boiling component is now richest at a midpoint, not at the base, and a low boiler is now richest in the base, not at the top of the column. As a consequence, as the lower boiling starting material begins to distill up the column, it encounters an increasingly rich concentration of the higher boiling starting material. Conversely, the higher boiling starting material sees an increasing concentration of the more volatile reactant as it descends the column, forcing its consumption to rise: thus the consumption of both components is actually enhanced by the process. This effectively forces the equilibrium toward the product side without the use of large excesses of one starting material. The product(s) distills into the refinement zone where the product(s) is distilled from the higher boiling starting material. (Remember that the desired product must boil at a point lower than the highest boiling starting material.) In their application to the esterification of acetaldehyde with acetic anhydride, shown in Figure B.8, Eastman added the acetic anhydride to the mid-point of a distillation column, along with an acid catalyst, and added the other reactant, acetaldehyde, to the base. If one examines the different boiling points of the materials, which are included in the figure, the consequences of this mode of operation become apparent. The use of large excesses of acetic anhydride is circumvented since, as acetaldehyde rises up the column, it will encounter a continuously enriched acetic anhydride environment, effectively shifting equilibrium in Reaction (7) toward the formation of EDA without requiring the use of huge excesses of acetic anhydride. The EDA, being the highest boiling component in the reaction mixture (167C), descends the column. In the second reaction, the cracking of EDA to vinyl acetate and acetic acid (Reaction (8)), the equilibrium is independent of acetic anhydride, so the vinyl acetate and acetic acid are eventually distilled from the reaction zone. Work conducted prior to the contract did indicate that, due to the volatility of acetaldehyde, we would be hard pressed to operate the reactive distillation without purging some excess acetaldehyde. Further, to enhance the concentration within the reactive section and reduce the amount of acetaldehyde distilled overhead, it would be best to operate under moderate pressure.

50

The demonstration column used in the work preceding the contract was operated for extended periods over two pressure conditions and was used as a source of "kinetic" data to further develop the process. Within the term of the contract, Eastman continued to optimize the reactive distillation using a series of computer simulation techniques that utilized the data obtained from their bench-scale unit before they entered the joint research program. Figure B.8 Idealized Reactive Distillation for the Conversion of Acetaldehyde and Acetic Anhydride to Vinyl Acetate and Acetic Acid [component boiling point in bold] Unreacted AcH (if any) [19C] VAM [72-73C] Separation Zone AcOH [118-119C] Ac2O [140C] + Acid (cat.) React Reaction Zone AcH [19C] High boilers ("Tars") One important difference between the Eastman process and the earlier Celanese process emerged from this optimization. To reduce column sizes, and thus the capital investment, Eastman had to move toward increasingly larger purge of acetaldehyde from the reactive distillation column as it drove toward complete conversion of the acetic anhydride. Ultimately, the Eastman process involved the use of excess acetaldehyde, rather than the large excess of acetic anhydride used by Celanese. Initial Equipment Costs and Utility Usages Table B.1 includes an equipment list for each column, as well as rough capital cost estimates (generated using ASPEN) and utility usages for the initial case. The high utility usages for the

51

reactive distillation column reboiler and condenser are somewhat alarming. These high usages are a result of moving essentially all of the material entering the column out as an overhead stream, and appear to be required with the current design. Process Optimization In an effort to improve both the performance and costs of the reactive distillation block, the effects of a number of variables were studied. These variables included the locations of the feed stages, volume of liquid holdup in the reactive portion, the number of trays for both the reactive and non-reactive portions, bottoms flow rate, reflux ratio, tray spacing, and the acetaldehyde recycle flow rate (hence the ratio of acetaldehyde to acetic anhydride in the column). The most important variables appear to be the reflux ratio and the size of the acetaldehyde recycle loop. Table B.1 Eastman VAM Process Equipment Capital and Utility Usage Original Basis Section Reactive Distillation Equipment Item Accumulator Column Condenser Reboiler Accumulator Column Condenser Reboiler Recycle Cooler Accumulator Column Condenser Reboiler Product Tank Net

Equipment Capital $163,600 $1,372,600 $299,200 $1,779,600 $63,400 $539,300 $161,300 $212,500 $25,100 $44,900 $316,900 $48,600 $306,600 $1,300,000 $6,633,600

Utility Type

Utility Usage (Mbtu/hr)

Cooling Water 600 psi Steam

-130 140

Acetaldehyde Recovery

Cooling Water 100 psi Steam Cooling Water

-32 36 -2.1

Vinyl Acetate Recovery

Cooling Water 100 psi Steam

-17 17

Reflux ratio has a considerable effect on the diameter of the reactive distillation column. The initial model had a reflux ratio of 8.0, requiring column diameters of 20-25 feet under the flow conditions (acetaldehyde recycle flow rates) necessary to achieve high vinyl acetate yields. Sensitivity studies revealed that a reflux ratio of 2.0 requires columns only 10-15 feet in diameter under the necessary conditions.

52

Figure B.9 illustrates the dependence of vinyl acetate production on the acetaldehyde feed rate (fresh + recycle) and reflux ratio. With a reflux ratio of 8.0, only 1,000-1,200 lbmol/hr of acetaldehyde is required to achieve quantitative yields of VAM (715 lbmol/hr). With a reflux ratio of 2.0, quantitative yields of VAM can be achieved with acetaldehyde flows of 2,8003,000 lbmol/hr. Lowering the reflux ratio to 1.0 requires much larger acetaldehyde flows. It can therefore be seen that optimum conditions for high vinyl acetate production and low column diameter exist with a reflux ratio of 2.0 and a total acetaldehyde feed of 2,800 lbmol/hr. Figure B.9 Dependence of Column Diameter on Total Acetaldehyde Fed to the Reactive Distillation Column and Reflux Ratio
800 700 VAM Flow, lbmol hr
-1

600 500 400 300 200 100 0 0 500 1000 1500 2000 2500
-1

VAM (RR=1) VAM (RR=2) VAM (RR=4) VAM (RR=8)

3000

3500

4000

HAc Flow, lbmol hr

Economic Comparison The positive impacts of this optimization work are readily seen by an equipment capital cost comparison for the initial and optimized versions of the reactive distillation model (Figure B.10). The cost of the acetaldehyde recovery column has increased from roughly $500,000 to $1,000,000 due to increases in the size of the acetaldehyde recycle loop. The benefits of the modifications are readily seen in the 40% capital reduction for the reactive distillation column. There was little change in the cost of the vinyl acetate cleanup column. Overall, there was a 26% equipment capital reduction for the reactive distillation block due to modifications in the reflux ratio and size of the acetaldehyde recycle loop. In addition, the utility capital has been reduced by roughly 10%. Summary and Conclusions A reactive distillation process model for the synthesis of vinyl acetate has been developed, optimized, and costed using ASPEN. Significant improvements in equipment capital costs have been made through modifications of the reflux ratio for the reactive distillation column and the

53

size of the acetaldehyde recycle loop. With the technology developed in this study, VAM may be produced from syngas, but the cost of production is about 15% higher than the conventional oxidative acetoxylation of ethylene, primarily due to higher capital associated with the syngasbased processes. Figure B.10 Capital Costs from Reactive Distillation Model VAM Production

$8,000,000 $7,000,000 $6,000,000 $5,000,000 $4,000,000 $3,000,000 $2,000,000 $1,000,000 $0 HAc Column Reactive Distillation VAM Column Total

Original Optimized

54

3.B.4 Isobutanol Studies The development of catalysts to generate C4-compounds directly from syngas can provide an alternative source to imported crude oil of these important chemical building blocks. As part of the work within this program, isobutanol was targeted as a key product. Two laboratory programs focused on this area of study. Aachen University One prominent C4-based compound in the domestic fuel mix has been MTBE (methyl tert butyl ether), which is typically produced by the reaction of methanol with isobutene. A promising path to isobutene is the heterogeneously catalyzed CO-hydrogenation to isobutanol followed by dehydration (Figure B.11). As shown by thermodynamic studies, the heterogeneously catalyzed CO-hydrogenation to isobutanol is not expected to experience any thermodynamic constraints. However, heterogeneous hydrogenation of CO is a very exothermic process, a problem that can only be partly solved when work is conducted in a plug flow reactor (PFR). When work is carried out in reaction vessels with moving catalyst beds such as a continuous stirred tank reactor (CSTR), heat transfer problems can be resolved, while additional benefits connected with this reactor type are offered. Figure B.11 Syngas-Based Synthesis of MTBE

CH3
CH3-OH
(methanol)

CH3-C-O-CH3

CO/H2/CO2
(syngas)

CH3
(methyl t-butyl ether)

+
CH3
CH3-CH-CH2-OH
(isobutanol)

- H2O

CH3
CH3-C=CH2
(isobutylene)

Several heterogeneous catalytic systems have been under investigation for their capability of isobutanol synthesis from syngas. The most promising catalysts for an active and selective isobutanol synthesis from CO are modified high-temperature methanol catalysts. In 1982, work at the Institut fr Technische Chemie und Petrolchemie at the University of Aachen studied the production of isobutanol from syngas at severe conditions (pressure of 325 bar and temperature of 450C). The catalyst (ZrO/In2O3/CuO/ZnO/K2O) produced isobutanol in yields up to 22%, which exceeded the performance of a BASF ZnO/Cr2O3/K2O catalyst. In 1986, additional studies were performed to develop a catalyst that operated at more moderate

55

conditions. This improved catalyst (Zr-Zn-Mn-Li-Pd), which at 41% CO conversion resulted in an isobutanol production of 750 grams/liter-hour, was the basis for the collaboration with Air Products under this DOE program. Two high-pressure laboratory units were designed and built for the investigation of heterogeneously catalyzed CO-hydrogenation to isobutanol. One unit was designed to test a large number of catalyst formulations, while the second unit (design conditions of 400 bar and 450C) allowed for testing of catalysts in either a CSTR or a PFR. Given the backmixing that is inherent in the CSTR, a consequence of producing isobutanol directly from syngas is that the CO2 concentration within the reactor will be higher than levels present in a PFR. One set of experiments in this program showed the impact of CO2 on performance of the Pd-Zr/Zn/Mnoxide catalyst system (Figure B.12), and that the effect was more pronounced in the CSTR (Figure B.13). However, for other catalysts such as Zn/Cr-oxide, the impact of CO2 was less pronounced. Another area of work attempted to reproduce the results from the 1986 study. The catalyst formulation had to be reproduced, and the formation of methane caused by production of metal carbonyls had to be minimized. A study of the construction of the PFR used in the 1996 work showed that an axial temperature gradient was present, raising temperature at the reactor outlet. This led to a decrease in methanol activity; this conclusion was verified by reproducing the axial temperature gradient on the newly designed PFR by the use of external heating (Figure B.14). Figure B.12 Influence of CO2 Addition in PFR on Main Products (Pd-Zr/Zn/Mn-oxide, PFR, T = 400C, P = 250 bar, GHSV = 27,000 h -1, Vcat = 2 ml, Dcat = 0.63- 0.71 mm)

56

Figure B.13 Influence of CO2 Content of Gas Phase at Reactor Outlet on Isobutanol Content in Gas Phase (Pd-Zr/Zn/Mn-oxide, PFR, T = 400C, P = 250 bar, GHSV = 27,000 h-1 ,Vcat = 2 ml, Dcat = 0.63-0.71 mm)

Figure B.14 Influence of Axial Temperature Gradient on Yield to Main Products in the PFR

57

Other catalyst formulations were made in an attempt to reduce the severity of the reaction conditions. A Zn/Zr/Mn-oxide catalyst was impregnated with a variety of metals (Cu, Co, Ru, Rh, Pd, and Pt) that have either hydrogenation activity or are able to catalyze the reaction from methanol to ethanol. Only copper showed improved results in isobutanol activity as well as selectivity, and the catalyst showed a remarkable thermal stability. When copper content was increased to high levels, linear chain growth to ethanol and n-propanol dominated over the preferred branching chain growth, especially at high temperature. Cu/Zr/Zn-aerogel-based catalysts were also investigated, and were shown to have a high selectivity and activity to isobutanol. However, deactivation of these catalysts was found to be adversely impacted by both thermal effects and the presence of H2 and CO at the elevated operating temperature. Tests were also performed by combining a catalyst active in linear chain growth at low reaction conditions (IFP Cu/Co-based catalyst system) with one active in branched chain growth (BASF Zn/Cr catalyst system), in order to increase isobutanol activity and selectivity at milder reaction conditions. Activity of the Zn/Cr-oxide catalyst can be greatly enhanced by addition of the Cu/Co-based catalyst, and the activity and selectivity towards isobutanol were enhanced with increased Cu/Co content. At a high Cu/Co-content of the mixed catalysts, however, the preferred products in the C2+-phase were linear alcohols. At higher temperatures, methane formation increased to undesired levels. Summary and Conclusions The University of Aachen designed and built two high-pressure units for isobutanol synthesis from syngas for use with both a CSTR and a PFR. Previously obtained results in isobutanol synthesis from syngas could be reproduced. The influence of CO2 in PFRs and CSTRs on isobutanol synthesis from syngas could be shown for different catalytic systems. From these experiments, it could be demonstrated that the C1C2 step is not the only rate-limiting reaction step in isobutanol synthesis. The activity of the Zr/Zn/Mn-oxide catalyst system could be enhanced at lower reaction conditions. Copper-containing catalysts showed high activity towards isobutanol with a remarkable thermal stability. The results from both the current program and other research work in the field did not exceed the performance of the catalyst formulation from the 1986 study. Given the required severity of the reaction conditions, the economics of isobutanol production were found to be unfavorable. Bibliography W. Keim, W. Falter, Catalysis Letters 1989, 3, 59. C.H. Finkeldei, B. Jaeger, W. Keim, K.A.N. Verkerk, Isobutanol Synthesis from Syngas, in "Designing Transportation Fuels for a Cleaner Environment," ed. J.B. Reynolds, M.R. Khan, Applied Energy Technology Series 1957.

58

Lehigh University Over the traditional higher alcohol synthesis catalysts, including the low-temperature Cu-based catalysts and the high-temperature zinc chromite catalysts, the branched alcohols are, along with methanol, the preferred products of the chain growth process. Due to their intrinsic low reactivity, they manifest a terminal behavior: once formed through the evolution of the intermediate species, isobutanol and higher branched oxygenates do not undergo further reactions and are accumulated in the reacting system. However, productivities of isobutanol higher than 50 g/kg cat/h have not been reported in the literature for these conventional catalysts. Appreciably higher yields are needed to be economically feasible. The slowest step of the reaction is the low rate of the C1 C2 step that initiates the formation of higher oxygenates. The optimization of higher alcohol synthesis kinetics through proper manipulation of the operating variables, copper-based catalyst composition, and the study of new catalyst formulations had been pursued in order to improve isobutanol productivity. In the work under this DOE program, high productivities and selectivities to isobutanol and higher 2-methyl alcohols were achieved by an alternative engineering design of higher alcohol synthesis. This consists of separating the process into two consecutive steps: (1) the initial synthesis of intermediate oxygenates and (2) their subsequent conversion to the terminal branched species. It is already known from the results of chemical enrichment experiments that the addition of C2 and C3 intermediates to the reacting system causes a general promotion of higher alcohol synthesis and, in particular, the production of branched alcohols. The purpose of the present research was the initial optimization of the production of short-chain alcohols over a Cs-promoted Cu/ZnO/Cr2O3 catalyst, the study of a Cs-promoted ZnO/Cr2O3 catalyst for the selective synthesis of isobutanol, and the engineering of double-bed experiments for the synergic utilization of the two catalysts. Herein, the optimized Cu-based catalyst (used as the first bed) is expected to supply intermediate reactants to the high-temperature catalyst (used as the second bed), and the reactants can be efficiently converted to isobutanol. The results are discussed in light of well-established knowledge about the mechanistic, kinetic, and thermodynamic aspects that govern higher alcohol synthesis over both the Cu-based and Cu-free catalysts. Catalyst testing was carried out in a tubular fixed-bed reactor. Special arrangements were made to minimize iron carbonyl formation from contact of CO with iron-containing surfaces, since deposition of iron from Fe(CO)5 over Cu-based catalyst sites causes irreversible deactivation. Charcoal and molecular sieve traps were employed for filtering the CO stream fed from highpressure stainless steel tanks. In addition, a copper-lined stainless steel reactor and copper thermocouple wells for reading the catalyst temperature were adopted, and a water-cooling system was used to keep the temperature of the stainless steel tubing upstream from the reactor below 338K. Indeed, elemental analysis of tested catalysts (500 hr on stream) indicated iron content <60 ppm. The catalyst pretreatments in the double-bed reactor required a stepwise procedure. Initially, the Cu-based catalyst was reduced by heating the top portion of the reactor at 523K under a flowing H2/N2 = 2/98 vol% mixture, while the bottom portion of the reactor was kept unheated. When the reduction was completed, the first bed was cooled to room temperature under flowing N2. Subsequently, the treatment of the zinc chromite catalyst was carried out by heating the bottom portion of the reactor to 723K under a flowing H2/N2 mixture. During this phase of the reduction 59

process, in order to "protect" the Cu-based catalyst from possible Cu sintering or over-reduction, the temperature of the first bed was kept below 343K by an external water circulating system. Table B.2 summarizes key elements of parametric studies using a 3 mol% Cs-promoted Cu/ZnO/Cr2O3 catalyst, a 4 mol% Cs-promoted ZnO/Cr2O3 catalyst, and the double bed arrangement using both catalysts. Table B.2 Effect of Catalyst and Reactor Configuration on Higher Alcohol Synthesis (H2/CO = 0.75)
Space Velocity, l(STP)/kg cat-hr 5,450 12,000 18,000 3 mol% Cs/Cu/ZnO/Cr2O3 (2 g) (598K, 7.6 MPa) Methanol, g/kg cat-hr Isobutanol, g/kg cat-hr Total 2-methyl oxygenates, g/kg cat-hr % CO Conversion (CO2-free) 4 mol% Cs/ZnO/Cr2O3 (2 g) (678K, 7.6 MPa) Methanol, g/kg cat-hr Isobutanol, g/kg cat-hr Total 2-methyl oxygenates, g/kg cat-hr % CO Conversion (CO2-free) Double bed arrangement (1 g each) (598/678K, 7.6 MPa) Methanol, g/kg cat-hr Isobutanol, g/kg cat-hr Total 2-methyl oxygenates, g/kg cat-hr % CO Conversion (CO2-free)

268.0 52.1 73.7 19.7

844.5 75.9 121.1 13.8

1,200.0 65.6 114.5 11.7

52.8 39.5 46.3 8.1

132.8 69.1 84.7 4.9

173.4 74.1 88.6 4.5

56.6 76.8 109.4 12.0

114.3 114.6 175.1 9.4

178.8 138.8 217.4 6.6

The kinetic investigation over the ternary Cs-doped, Cu-based catalyst indicated that the overall yield of higher oxygenates was enhanced with increasing temperature and was optimal for a H2/CO syngas ratio of 0.75. As did methanol formation, the production of C2+ species increased with gas hourly space velocity and total pressure. Under the optimized operating conditions (598K, H2/CO = 0.75, GHSV (gas hourly space velocity) = 12,000 l(STP-standard temperature and pressure)/kg cat-hr, and 7.6MPa), the productivities of both the higher alcohols and methanol were significantly higher than those reported in the literature for other higher alcohol synthesis catalysts. Conversely, the product distribution observed over the Cs-promoted commercial ZnO/Cr2O3 catalyst was characterized by low selectivities to ethanol and propanol and high selectivities to isobutanol. The chemical pathway

60

+C1 C2 C3

+C1 iC4 (9)

appeared to prevail over any competitive route, as indicated by the low quantities of C4-C6 oxygenates produced via cross condensations of the C2 and C3 intermediates. The high rate of the Cn + C1 steps was most probably related to the high reaction temperature employed. Nevertheless, it must be noted that neither the ternary Cu-based catalyst tested at the same temperature nor did other Zn/Cr/O catalysts with different compositions achieve the selectivity for isobutanol as did the currently used Cs-promoted commercial catalyst. The double-bed configuration combines the complementary features of the two catalytic systems, wherein each is allowed to operate under its own optimal temperature regime for the production of higher oxygenates. The 3% Cs/Cu/ZnO/Cr2O3 catalyst, operating at 598K as the first bed, produced methanol, ethanol and propanol in abundance; the 4% Cs/ZnO/Cr2O3 catalyst, operating at 678K as the second bed, efficiently converted the supplied C2 and C3 intermediates to the terminal branched alcohols. In other terms, the double-bed design takes advantage of the separation of the overall higher alcohol synthesis process into two subsequent steps: the initial low-temperature formation of the first C-C bond and the following hightemperature carbon chain growth process. The former step controls the rate of the overall oxygenates formation, while the latter effects the high selectivity to the 2-methyl alcohols, primarily isobutanol. As a result, at GHSV = 18,375 l(STP)/kg cat-hr, the overall formation of 2-methyl oxygenates over the double-bed configuration amounts to over 215 g/kg cat-hr, which represents an improvement of 90% with respect to the performance of the single Cs/Cu/ZnO/Cr2O3 catalyst, and 145% with respect to the single Cs/ZnO/Cr2O3 catalyst. It must be noted that the double-bed configuration is penalized with respect to the single Cu-based catalyst by a considerable "loss" of CO conversion. A comparison of results obtained over the double-bed system at GHSV = 5,450 l(STP)/kg cat-hr with those obtained over the single Cu-based catalyst at approximately twice the space velocity (12,000 l(STP)/kg cat/hr) allows the estimate that almost 80% of methanol produced over the first bed is decomposed over the high-temperature bed to CO and H2. However, as already pointed out, the net result of the reduced amount of methanol and the increased formation of branched species over the doublebed configuration is a very favorable methanol/2-methyl-oxygenates molar ratio. It was finally observed that the enhanced formation of higher oxygenates over the double-bed configuration is accompanied by an increase in C2+ hydrocarbons, probably due to the dehydration and decarboxylation reactions of higher alcohols that occur over the hightemperature catalysts [Forzatti and Tronconi, 1991]. However, the molar carbon selectivity to C2+ oxygenates is still more than one order of magnitude higher than the carbon selectivity to C2+ hydrocarbons. The effects of pressure were also investigated for the double-bed arrangement; Table B.3 provides results for a syngas with a H2:CO ratio of 0.75 at a GHSV of 18,375 l(STP)/kg cat-hr. Methanol productivity showed the largest increase at the higher operating pressure, and overall CO conversion was also greater. 61

The cesium-doped, copper-based catalyst has shown reasonable stability at high temperatures, along with its high selectivity toward isobutanol synthesis, indicated that this catalyst could be used to improve upon the previous double-bed experiment by increasing the isobutanol yield. Table B.4 shows the results of using the 3 mol % Cs/Cu/ZnO/Cr2O3 catalyst at a temperature of 598K in the top bed and two different temperatures in the bottom bed. By modifying the doublebed higher alcohol synthesis by using the Cs/Cu/ZnO/Cr2O3 catalyst in both beds but at different temperatures, isobutanol productivity increased significantly, as did the production of methanol. When the copper-based catalyst was used as the bottom bed and maintained at 613K, isobutanol productivity was 202 g/kg cat-hr; this was in contrast to 125 g/kg cat-hr (see Table B.3) when the bottom bed consisted of the copper-free catalyst at 678K, an increase of 62%. Similarly, methanol production was greatly increased by the copper-containing bottom bed, reaching a value of 574 g/kg cat-hr. Table B.3 Effect of Pressure on Higher Alcohol Synthesis (H2/CO = 0.75, GHSV = 18,375 l(STP)/kg cat-hr)
Top Bed: 3 mol% Cs/Cu/ZnO/Cr2O3 (1 g), 598K Bottom Bed: 4 mol% Cs/ZnO/Cr2O3 (1 g), 678K Pressure , Mpa 7.6 10.3 Methanol, g/kg cat-hr Isobutanol, g/kg cat-hr Total 2-methyl oxygenates, g/kg cat-hr % CO Conversion (CO2-free) 173 125 179 6.1 398 166 234 8.3

Table B.4 Higher Alcohol Synthesis over 3 mol% Cs/Cu/ZnO/Cr2O3 Catalyst (H2/CO = 0.75, GHSV = 18,375 l(STP)/kg cat-hr, Pressure = 7.6 MPa)
Top = 598K Top = 598K Bottom = 613K Bottom = 643K Methanol, g/kg cat-hr Isobutanol, g/kg cat-hr Total 2-methyl oxygenates, g/kg cat-hr % CO Conversion (CO2-free) 574 202 240 8.5 369 180 211 6.2

62

Summary and Conclusions The concept of employing a double-bed reactor with a pronounced temperature gradient to enhance higher alcohol synthesis was explored, and it was found that utilization of a Cspromoted Cu/ZnO/Cr2O3 catalyst as a first lower temperature bed and a Cs-promoted ZnO/Cr2O3 catalyst as a second high-temperature bed significantly promoted the productivity of isobutanol from syngas. While the conversion of CO to C2+ oxygenates over the double-bed configuration was comparable to that observed over the single Cu-based catalyst, major changes in the product distribution occurred by the coupling to the zinc chromite catalyst; that is, the productivity of the C1-C3 alcohols decreased dramatically, and 2-methyl branched alcohols were selectively formed. The target molar ratio of methanol to 2-methyl oxygenate molar ratios of approximately 1 (desirable for feedstock for high-octane and high-cetane synthesis) was obtained in the present double-bed system, and the isobutanol productivity was as high as 139 gm/kg cat-hr. Higher alcohol synthesis has been investigated over a Cs/Cu/ZnO/Cr2O3 catalyst at temperatures higher (up to 703K) than those previously utilized, and no sintering of the catalyst was observed during the short-term testing. However, the higher reaction temperatures led to lower CO conversion levels and lower yield of alcohols, especially of methanol, because of equilibrium limitations. With the double catalyst bed configuration, the effect of pressure on catalyst activity and selectivity was studied. High pressure was found to increase CO conversion to oxygenated products, although the increase in isobutanol productivity did not keep pace with that of methanol. It was also shown that the Cs/Cu/ZnO/Cr2O3 catalyst could be utilized to advantage as the second-bed catalyst at 613-643K instead of the previously used copper-free Cs-ZnO/Cr2O3 catalyst at higher temperature. With double Cs/Cu/ZnO/Cr2O3 catalysts, high space time yields of up to 202 gm/kg cat-hr, with high selectivity to isobutanol, were achieved. As with the work at the University of Aachen, the economics for the production of higher alcohols using this approach were not favorable, and the program efforts in this area were concluded in September 1996.

63

3.B.5 DME Conversion to Fuels and Chemicals Introduction Development and eventual commercialization of the LPDME Process will require a substantial growth in the demand for DME. An alternative, large market to current uses for DME is its use as a feedstock to the synthesis of liquid diesel fuel additives. These additives when blended with diesel fuel result in improved fuel properties, in particular, increased cetane number. Cetane number is a measure of the ignition quality of a diesel fuel. For reference, typical U.S. diesel fuel has a cetane number of about 45. High cetane number fuels have several advantages over low cetane number fuels. These include shorter ignition delays, lower emissions, better cold starting properties, lower smoke/particulate emissions, and higher fuel efficiencies, which means reduced CO2 emissions. The Air Products approach uses DME as a feedstock to produce a liquid mixture of oxygenates. This mixture is envisioned to have a cetane number of 70 or greater and, when blended with diesel fuel, to yield a fuel blend of cetane number 50 to 60. The liquid mixture of oxygenates was dubbed "CETANER." The potential market will have to be sufficiently large to support the development of a new technology. If 10% of the U.S. diesel market, estimated to be 2.7x1010 gal in 1995, could be captured by CETANER technology and it were blended at 15 vol %, production of 4.1x108 gallons or nearly 3 billion pounds of CETANER would be required per year. The most reasonable and straightforward method for production of higher molecular weight oxygenates from DME is by oxidative coupling. Oxidative coupling of DME by reaction with O2 (Reaction (10)) produces monoglyme (CH3OCH2CH2OCH3) and water. 2 CH3OCH3 + O2 = CH3OCH2CH2OCH3 + H2O (10)

DME oxidative coupling to monoglyme is thermodynamically favorable, largely because water is a product. At 25C, the following were calculated from literature data for Reaction (10): H = -52.5 kcal; G = -48.0 kcal; K = 2x1035. DME combustion (Reaction (11)), however, is even more favorable: H = -317.5 kcal; G = -325.4 kcal; K = 4x10238 at 25C. CH3OCH3 + 3 O2 = 2 CO2 + 3 H2O(g) (11)

The chemical challenge is to find catalysts and conditions under which the coupling reaction is favored over total or partial oxidation reactions. Surveying the literature revealed little about DME oxidative coupling catalysts. Only for a SnO2/MgO catalyst was monoglyme produced selectively. Some catalysts, particularly supported noble metals, were examined under conditions that favored total oxidation products. Lacking knowledge of what DME conversion and monoglyme selectivity would be required for a practical, economically viable process, those reported to be necessary for the analogous methane coupling process were used: 25% DME conversion, 75% monoglyme selectivity. No known catalysts could meet these targets. Thus, the primary effort of this project was to find a 64

catalyst that achieved target DME conversion and monoglyme selectivity. A second major effort was the evaluation of the properties of the expected CETANER product and of its diesel fuel blends. Catalyst Development Clues to the types of potential catalysts can be obtained by considering a reaction scheme for oxidative coupling. The initial step is likely abstraction of a hydrogen atom to produce a DME radical, CH3OCH2. Subsequent radical coupling gives rise to the desired product, monoglyme. Hydrogen atoms are oxidized to water by reaction with O2 or its equivalent on the catalyst surface. Therefore, basic and redox active metal oxides are reasonable choices as potential catalysts. Included among these is the literature catalyst SnO2/MgO. The similarities of DME and CH4 oxidative coupling suggest that the examination of methane coupling catalyst might be fruitful. Finally, general oxidation catalysts were considered. As a result of this analysis, more than 40 catalysts were examined within this program. Thermogravimetric analysis (TGA) offers a quick screening technique to gauge the DME reactivity of potential catalysts in the absence of O2. Reaction (12) (M = metal), for example, might occur in the absence of O2, and the resulting weight loss can be monitored in the TGA. Alternatively, deposition of elemental C or various CxHy fragments could result in weight increases, as would formation of carbonates. MO + 2CH3OCH3 = CH3OCH2CH2OCH3 + H2O + M (12)

Many of the oxides examined exhibited a weight loss when heated under N2, and such losses can complicate the TGA experiment. These weight losses were usually the result of water desorption, but some oxides liberated O2 gas upon heating. To minimize the contribution of weight losses under N2, each sample was heated to 450C under N2 to achieve a constant weight prior to exposure to DME. A few oxides that were known to evolve O2 at elevated temperatures were heated to lower temperatures. The reactivity of each sample was evaluated between 200 and 450C. The TGA results were such that the catalysts examined could be grouped into one of four classes described below and listed in Table B.5. Table B.5 DME Reactivity Groupings by TGA
Reactivity class Class 1: Unreactive Substances PbO, Gd2O3, CeO2, MoO3, Bi2O3, NaWO4/CeO2, Mn/Na2WO4/MgO, Bi-Mo-Fe oxide, Bi2MoO6, "Bi24Fe2O39," Bi2Mo3O12, Bi3FeMo2O12, Bi2Mo2O9, BiFeO3a, BiFeO3b, Bi6Mo2O15, "Bi4MoO9" MgO, Li-MgO, -Al2O3, AlPO4, SnO2/MgO, SrSnO3, Ca2V2O7, Sr2V2O7, Ce0.8Gd0.2O1.9, Pd/CeO2, Pd/Al2O3, Pt/CeO2, Pd/C, MgMoO4/MgO SnO2, Nd2O3/MgO, Nd2O3, Sm2O3 Fe2O3, NiO, SrNiO2.5, LaNiO3, V2O5, La2O3, Ag-Cs-Sr oxide, Pr6O11, Pr0.55Ce0.45O2

Class 2: MgO-like

Class 3: MgO-like, less reactive Class 4: Very reactive or unique reactivity

65

Many of the oxides examined proved to be unreactive with respect to DME, that is, almost no weight change under DME (Class 1). This is surprising for some of the oxides in this group. For example, PbO is one of the most selective CH4 coupling catalysts, but it appears to show no reactivity with DME in the absence of O2. A mixed Bi-Mo-Fe oxide that was found to be active for DME oxidative coupling (see below) showed no reactivity in the absence of O2. Class 2 includes non-redox active oxides such as MgO, some redox active mixed oxides, SnO2/MgO, and supported Pd and Pt catalysts. For each, a significant weight increase was observed at 200C. Increasing temperature resulted in smaller weight increases. The source of the weight increase is unknown. Class 3 oxides exhibited a weight change pattern like that of MgO, but the oxides were less reactive as judged by the magnitude of the weight increases. The last group (Class 4) of oxides showed substantial weight changes upon exposure to DME. Catalysts were then evaluated at pressures less than 40 psig using a fixed-bed reactor system. Typically, 1 cc of catalyst (500-850 m) was used, and the reactor volume above and below the sample was filled with quartz chips to minimize gas phase reactivity. Feed gas was flowed from the top of the reactor down. SnO2/MgO is the only substance known to catalyze the oxidative coupling of DME. The literature preparation involves impregnation of MgO with aqueous Sn(NO3)2 followed by drying at elevated temperatures. This route was avoided because of the potential explosive nature of Sn(NO3)2 upon dehydration. The use of SnCl2 as a precursor was also avoided to eliminate possible complications associated with HCl formation during calcination. A preparation using tin(II) acetate, Sn(OAc)2 was found to be problematic as well. Finally, a preparation using SnO2 as the tin precursor gave an acceptable product, and it is this material that was evaluated as a catalyst. The literature evaluation of SnO2/MgO was performed using a feed consisting of 73 mol% DME and 17 mol% O2 at 235 psia. A monoglyme selectivity of 34.5% and DME conversion of 10.8% were achieved at 200C and a contact time of approximately 0.6 seconds. Reproduction of such conditions with the current testing apparatus was not feasible. As detailed in Table B.6 (entry 1), SnO2/MgO was evaluated at feed pressures ranging from 6 to 38 psig, feed DME/O2 molar ratios of 5 and 10, and temperatures of 200 to 400C. For none of the conditions were any oxidative coupling products observed. As expected, the catalyst activity increased with temperature, but the largest increase was in the C1 products, CO and CO2. Methanol and methyl formate (HCOCH3) were also observed, but at lower concentrations. The most likely route to methyl formate is the oxidation of DME. These results are significantly different from those of the literature. At the relatively low literature reaction temperatures, 200-230C, no reactivity of any kind was observed in the above experiments. Even at 250C, only trace quantities of CO2 were observed. Dimethoxymethane, reportedly produced with 13.5% selectivity using the literature catalyst, was not detected in the product stream, but the unreported methyl formate was at temperatures of 300C or greater. These differences were ascribed to the lower total pressure and O2 partial pressure of these experiments.

66

Table B.6 Results for the Reaction of DME and O2 over SnO2/MgO and SnO2
(MeOH = methanol; MeF = methyl formate, unk = unknown)

Entry no. Catalyst 1 SnO2/MgO CO 19.9 1.09 7.20 2.60 CO2 100 68.2

Selectivity, % C CH4 MeOH MeF

unk 1.09

54.7 29.5 24.1 18.4 34.5 16 10.9 25.8 25.0 1.42 0.36 0.24 0.94 0.19

100 36 1.44

5.75

1.31

0.80 0.19 23.2 16.4 16.7 46.3 6.9 3.1 16.5 3.9 0.49 0.28 0.89 0.79

2 SnO2

SnO2/MgO

T, C 250 400 200 250 300 400 250 400 300 320 340 250 300 320 250 300 350 250 300 350 68.9 100 45 61.7 100 30.6 32.9 100 100 88.9 99.1 69.5 73.3 0.2 0.86 0.07 0.24 0.58 1.46 3.86

P, psig 6 6 10 10 10 10 10 10 20 20 20 38 36 34 20 20 20 20 20 20

DME/ contact % conversion carbon O2 time, s DME O2 balance 5 7.0 0 0 1.21 5 7.0 0 87.6 1.07 5 1.7 0 0 1.21 5 1.7 0 0 5 1.7 0 0 1.30 5 1.7 0 85.9 1.17 10 1.7 0 0 1.14 10 1.7 0 92.6 1.17 5 2.2 0 6.8 1.02 5 2.2 4.6 65.0 1.01 5 2.2 1.5 79.7 1.04 5 2.2 0 1.8 1.02 5 2.2 3.3 61.4 1.01 5 2.2 1.0 83.5 1.09 5 2.4 6.5 1.8 0.94 5 2.4 2.8 2.9 0.98 5 2.4 10.3 100 0.96 5 2.4 6.4 1.9 0.94 5 2.4 8.6 57.4 0.96 5 2.4 11.8 100 0.95

0.20

67

A few catalysts, however, yielded trace quantities of the coupling products 2-methoxyethanol, monoglyme, and diethyleneglycol methylether. These included a mixed Bi-Mo-Fe oxide, MoO2, and V2O3. Encouraged by trace activity, these catalysts, along with the literature catalyst, were examined at high pressure (> 200 psi), at which the bimolecular coupling reaction was expected to be favored. Monoglyme selectivities for these three catalysts were only slightly greater than that for an empty reactor. Also, a major increase in selectivity for coupled products was not observed for the literature catalyst, SnO2/MgO. Table B.7 lists results for this catalyst as a function of feed pressure, reactor temperature, and two different feed DME/O2 ratios. At the higher DME/O2 feed ratio of 16.3 (less O2 in feed), the highest monoglyme selectivity of 5.9% was observed at 300C and 300 psig. This value is significantly lower than the literature value for the same catalyst, 34.5%, but is in agreement with earlier results at Air Products. Most surprising were the large methyl formate selectivities, 45-69%, observed under these experimental conditions. Contrary to the literature results, the use of higher feed O2 concentrations would appear to offer no advantage, at least at 250C or higher. It had been assumed that lack of reactivity was due to differences in experimental conditions. The literature experiments were performed within the DME explosive range; these experiments could not be performed at Air Products. As an approximation, a feed containing the literature partial pressures of DME and O2 (195.0 and 40.0 psia, respectively) was generated by blending air and DME to a total pressure greater than that in the literature. Thus, DME and O2 partial pressures were comparable to the literature values, but the mole fraction of DME was outside the explosive range. A comparison of the results from these experiments with the literature at 200C illustrated striking differences (Table B.8). Another potential source of differences was the catalyst preparation. Preparation of the literature catalyst was not well documented. It is merely stated that magnesium oxide was impregnated with the metal nitrate. Because of potential hazards associated the preparation of Sn(NO3)2, our catalyst was prepared by slurrying together SnO2 and MgO, followed by calcination. In a final attempt, Sn(NO3)2 was prepared by reaction of aqueous AgNO3 and SnCl2 and used to prepare a catalyst by the literature method. However, monoglyme selectivity was still low (2.41%). Based on flash point requirements (see below), the most abundant glyme component of CETANER will necessarily be diglyme or higher. Several routes to diglyme can be envisioned. One route is the two-step oxidative coupling of DME to monoglyme and subsequently to diglyme. Such a scheme would require the two selective oxidation steps (Reactions (13), (14)), which is highly unlikely: 2CH3OCH3 + 0.5 O2 = CH3OCH2CH2OCH3 + H2O CH3OCH2CH2OCH3 + CH3OCH3 + 0.5 O2 = CH3O(CH2CH2O)2CH3 + H2O (13) (14)

A more reasonable approach is the intermolecular dehydration of 2-methoxyethanol, a species that was observed in some DME oxidative coupling runs: 2 CH3OCH2CH2OH = CH3OCH2CH2OCH2CH2OCH3 + H2O (15)

68

Table B.7 Evaluation of SnO2/MgO

Feed A: 72.6 mol % DME, 4.7 mol % O2, 22.7 mol % N2; DME/O2 = 16.3; contact time, 1.6 sec. Feed B: 57.0 mol % DME, 8.75 mol % O2, 34.2 mol % N2; DME/O2 = 8.9; contact time, 1.6 sec. Feed order run T, C P (psig) A 1 250 300 1.3 1.04 100 14.7 10.5 1.41 0.55 8.68 54.5 0.60 1.39 0.23 0.50 0.49 4.06 0.37 0.17 0.54 2.00 85 0.66 3.36 0.69 0.13 0.26 0.63 1.83 83 0.83 4.06 1.39 0.18 9.22 57.7 0.35 1.35 0.35 0.67 9.49 56.6 0.44 1.58 0.45 0.61 11.4 10.1 1.21 11.3 8.98 1.32 17.2 12.1 3.39 0.12 0.053 9.41 45.4 1.61 1.79 15.2 12.2 3.14 0.090 10.0 46.1 0.98 2.32 0.38 0.46 0.48 5.93 0.31 0.29 0.15 3.55 93 2.04 88 1.88 89 0.72 5.55 0.62 0.28 12.9 12.7 2.67 0.064 11.1 47.9 0.60 2.23 0.54 0.61 0.88 4.37 1.06 0.26 3.06 92 0.24 1.38 102 101 1.52 97 0.000 14.2 4.14 0.088 0.17 9.67 60.3 3.10 0.73 0.64 0.98 0.35 3.87 A 2 250 500 3.1 1.02 100 A 3 250 750 33.0 0.72 99.5 A 5 275 300 9.0 0.96 100 A 6 275 500 12.2 0.93 100 A 4 275 750 13.2 0.93 100 A 10 275 750 14.0 0.92 100 A 7 300 300 15.4 0.90 100 A 8 300 500 16.6 0.90 100 A 9 300 750 19.6 0.87 100 B 13 250 500 * 1.13 100 B 12 275 500 15.7a 1.04 100 B 11 300 500 57.5 0.55 100 23.6 18.6 3.37 0.079 8.08 37.1 1.06 0.73 0.85 0.77 0.47 2.57

DME conv. (%) C balance O2 conv. (%) Product selectivity (%C) CO 11.7 17.4 CO2 13.1 9.00 9.96 12.4 CH4 4.66 0.43 1.23 ethylene 0.11 0.61 ethane 0.25 methanol 10.9 8.30 8.35 8.25 methyl formate 58.0 69.3 63.6 53.1 methyl ethyl ether 4.48 1.94 0.78 ethanol 0.84 0.93 0.53 0.82 diethyl ether 0.30 0.12 dimethoxymethane 0.57 0.32 0.57 0.42 methyl acetate 0.20 2-methoxyethanol 0.27 0.34 0.36 0.26 monoglyme 5.45 5.50 1.83 3.22 trioxane 0.33 1,4-dioxane 0.20 0.15 0.17 methyl methoxyacetate 0.28 0.45 dimethyl oxalate 0.15 0.22 DEGME 0.15 0.46 0.30 0.41 unknowns 0.82 1.22 1.09 1.39 reference: 1679780 81 82 84 a. Data at 275C suspect and selectivities not included in table.

69

Feed: 54.0 mol % DME, 8.86 mol % O2, 37.1 mol % N2; DME/O2 = 6.10; contact time, 0.6 sec. Literature feed: 73 mol % DME, 17 mol % O2

Table B.8 Evaluation of SnO2/MgO Catalyst at High O2 Concentration at 200C

DME conversion, % O2 conversion, % % C selectivity Methyl formate Dimethoxymethane Monoglyme

Literature resultsa 11 78.1 0 13.5 34.5

Current results ~0 4.8 53.8 0 0

a. Calculated literature heats and free energy of formation for monoglyme at 25C estimated by Benson's method: H = -82.8 kcal/mol; G = -47.4 kcal/mol. Other thermodynamic calculations in this report were performed using the program "HSC Chemistry for Windows 3.0" version 3.02. An experimental heat of formation of monoglyme, -81.4 kcal/mol, is reported: Luocks, L.F.; Laidler, K.J. Can. J. Chem. 45, 2785-2793 (1967).

The synthesis of diglyme by the dehydration of 2-methoxyethanol, a non-oxidative route, in the gas and liquid phases was investigated using oxide catalysts and acidic resins, respectively. In the gas phase, the most promising results were obtained for SrHPO4, which gave a diglyme selectivity of 6-7%. Results for the liquid phase reaction over acidic resins were more encouraging. For example, Nafion resin gave a 2-methoxyethanol conversion of 6.0% and a digylme selectivity of 77% at 120C. DME oxidative coupling products, like DME itself, are susceptible to unwanted oxidation reactions. Monoglyme if produced might decompose or undergo further oxidation before exiting the reactor. To gauge this potential reactivity, feeds consisting of gaseous monoglyme and N2 or air were passed over SnO2/MgO. The stability of monoglyme was found to decrease with increasing temperature over SnO2/MgO catalyst in the presence of O2. About 9% monoglyme decomposed at 250C, while at 300C, perhaps as much as 50% decomposed. The major organic product was methyl formate. Evaluation of Properties of CETANER and CETANER Blends with Diesel Fuel A second major effort was the evaluation of CETANER and CETANER diesel fuel blend properties. CETANER was shown to be completely miscible with diesel fuel to at least 40 vol %. The alcohols that are expected by-products of DME oxidative coupling, methanol, 2-methoxyethanol, and diethylene glycol methyl ether (DEGME), were far less miscible. At approximately 10 vol %, mixtures of each alcohol with diesel fuel resulted in two phases. Flash point minimums for European and U.S. diesel fuels are 55 and 38C, respectively. As shown by the flash points of potential CETANER components listed in Table B.9, meeting such targets will require minimizing the concentration of low flash point compounds, particularly monoglyme, in CETANER. The only component with a flash point greater than that of European diesel is diglyme, and it is therefore the only component that will not decrease the flash point upon blending. 2-methoxyethanol has a relatively high flash point, but its cetane 70

number is low, 13. The flash point of dimethoxymethane is very low, and its presence in CETANER needs to be avoided. Methanol also has a low flash point, a low cetane number and, based on its potential corrosivity in the presence of water, its concentration needs to be minimized as well. Table B.9 Flash Points of Possible CETANER Components Compound Methanol Dimethoxymethane Monoglyme 2-methoxyethanol Diglyme
* Ferro product literature

Flash point, C 11 -17 0 or -6* 46 57

To estimate the maximum permissible monoglyme concentration in U.S. and European diesel fuel, a literature method was used to calculate flash points. The method assumes that the flash point of a blend depends only on the vapor pressure of the most volatile component, in this case, monoglyme, and uses the equation T1/T2 = 1 +(T1Rlnx/H) (16)

where T1 is the flash point of monoglyme (273 K), T2 is the calculated flash point of the blend, x is the mole fraction of monoglyme in the blend, and H is the heat of vaporization of monoglyme (39.1 kJ/mol). To calculate a mole fraction, the average molecular weight of diesel fuel was assumed to be 225 (molecular weight of hexadecane), and a density of 0.825 g/ml was used. The flash points (FP) of a series of monoglyme and/or diglyme-diesel fuel blends were determined by the closed cup method. Figure B.15 provides the results of experiments using monoglyme with a No. 2 diesel fuel (FP = 61C), as well as the results from Equation (16). The maximum allowable monoglyme concentration in U.S. diesel fuel is less than 4 vol %. Even at 1 vol % monoglyme, the blend FP was below the European minimum of 55C. Blend flash points were approximately the same for equal concentrations of added monoglyme, whether diglyme was present or not. Experimental flash points, especially at lower monoglyme concentrations, were usually less than the corresponding calculated values. Based on flash point considerations, diglyme is the most beneficial CETANER component. Central to the use of CETANER is its ability to improve the cetane number of diesel fuel. Cetane number is a measure of the ignition delay in a diesel engine and, as such, is a measure of the ignition quality of a fuel. The higher the cetane number, the shorter the ignition delay. Cetane number (CN) is measured with respect to two reference compounds, n-hexadecane (cetane), a linear hydrocarbon, with an assigned value of 100, and a highly branched hydrocarbon, heptamethylnonane, CN of 15. Because of the lack of suitable standards, differentiation among cetane numbers greater than 100 is thought to be unwise. 71

Figure B.15 Dependence of Fuel Blend Flash Point on the Concentration of Monoglyme

60

50

40

Flash point, C

30

20

10

0 5 10 vol % monoglyme

15

20

Note: Data from experimental flash point for blends containing 61C flash point diesel fuel; curve represents calculated flash points

72

Cetane number determinations were performed by Southwest Research Institute (SwRI) and Advanced Engines Technology (AET). Relevant data is reported below. Table B.10 lists potential CETANER components and other compounds studied at SwRI. Cetane numbers of various CETANER compositions were determined by SwRI and are listed in Table B.11. Table B.10 Cetane Numbers of Pure Substances determined by SwRI Substance Monoglyme Diglyme Trigylme Dimethoxymethane 2-methoxyethanol 1,2-dimethoxypropane dioctyl ether Cetane number 105 228 119 24.4, 28.9 13.2 109 100.7

Table B.11 Cetane Numbers of Various CETANER Compositions

Components, vol % CETANER designation A C E F G H I J CNa 75.6, 67.8 90.2, 118 85.9, 107 139 70.1 37.4 57.0 39.4 monoglyme 72.9 95.9 91.9 70.0 51.0 77.2 87.4 82.1 dimethoxymethane 24.2 4.1 3.9 0 24.3 12.2 3.7 3.6 methanol 2.9 0 4.2 0 2.9 10.6 8.9 14.3 diglyme 0 0 0 30.0 21.9 0 0 0

a. Those with two valves are the result of two different determinations.

The highest cetane number was obtained for a monoglyme-diglyme blend containing no methanol or dimethoxymethane. Consistent with the pure compound data, compositions containing relatively large concentrations of methanol or dimethoxymethane, such as CETANER H, have relatively low cetane numbers. The real test of CETANER is its ability to increase the cetane number of diesel fuel. Several studies were performed at SwRI and AET to quantify this increase. To more thoroughly understand the relationship between blending concentration and cetane number, a study was conducted by SwRI in which CETANER B (90 vol % monoglyme, 5 vol % dimethoxymethane, 5 vol % methanol) was blended with No. 2 diesel fuel between 5 and 40 vol %. Surprisingly, there was not a steady increase in CN with increasing CETANER concentrations. As shown in Figure B.16, between 10 and 25 vol % there was virtually no 73

change in CN. An independent investigation into the effects of CETANER concentration on CN was conducted by AET. The fuels used in their study were low cetane number diesel fuels, including a Straight Run Light Gas Oil, CN 33.7, a "low cetane conventional fuel," CN 36.4, and a "synthetic oil sands fuel," CN 36.7. Cetane numbers were determined for each fuel blend with CETANER B at about 5 to 35 vol %. As shown in Figure B.16, there is a linear relationship between percent increase in CN and CETANER concentration. The disparity between the AET and SwRI studies may arise from differences in the fuels used. Monoglyme and other potential CETANER components are generally regarded as stable molecules with respect to hydrolysis or other decomposition reactions. However, the glymes are known to form hydroperoxides (ROOH) upon storage in ambient air. A year long study, however, showed that hydroperoxides were not formed in diesel-CETANER blends stored at ambient temperature. The absence of hydroperoxides is likely attributable to reaction with antioxidants often found in diesel fuel. Monoglyme and digylme-diesel fuel blends were shown to be stable with respect to hydrolysis or other decomposition reactions over a 98-day period. The glyme components of CETANER have relatively low toxicity; however, some evidence suggests that they may have teratogenic properties. In particular, diglyme was found to be a teratogen in mice because of its metabolism to methoxyacetic acid, a known teratogen. Monoglyme and diglyme were found to have low biodegradability. There is a huge potential market for CETANER technology. Capture of 10% of the U.S. diesel fuel market would require 4,000 tons/day of CETANER, along with 2,100 to 2,600 tons/day of O2 for its production. A process analysis showed that if target catalyst selectivity to monoglyme (75%) and DME conversion (25%) can be achieved, CETANER could be produced at $1.03/gal based on using DME from the LPDME Process at $0.45/gal (if diglyme is required as the major CETANER component, additional synthetic steps will incur additional costs). As shown in Table B.12, the cost of CETANER is competitive with or superior to other technologies for achieving improved diesel fuel properties. In addition, CETANER technology offers the unique advantage of reduced particulate emissions, which may not be achievable using the others technologies.

74

Figure B.16 Percentage Increase in Cetane Number as a Function of Monoglyme Concentration

50.0 AET Fuels

40.0

30.0

SwRI

20.0

10.0

0.0 10 CETANER
TM

20 30 B vol% in Fuel Blend

40

Triangles, diamonds, squares - blends using AET fuels; circles - blend using No. 2 diesel fuel (SwRI). CETANER B = 90 vol % monoglyme, 5 vol % dimethoxymethane, 5 vol % methanol.

75

Table B.12 Cost Comparison for Equivalent Improvement in Cetane Number Fuel composition or treatment Diesel fuel Diesel fuel/cetane improver Diesel fuel/10 vol % CETANER Diesel fuel/20 vol % FT liquids Deep hydrogenation Cost per gallon $0.500 $0.512 $0.553 $0.576 $0.900

Summary and Conclusions The development of a series of liquid additives as cetane enhancers for diesel fuel was studied within this program. These additives, dubbed "CETANER," consist of a mixture of oxygenates that could be produced by routes such as the oxidative coupling of DME. Fuels with high cetane number, which reflects the fuel ignition quality, offer several environmental and economic advantages. The potential market for CETANER technology is huge, perhaps 4,000 tons/day of CETANER. A process analysis shows that CETANER can be competitively priced with alternative cetane enhancer technology provided that target selectivities and conversions can be achieved. The major objectives of this program were (1) to identify catalysts that permit a practical, cost-effective route to the oxidative coupling of DME to CETANER and (2) to evaluate the properties critical to the use of CETANER as a blending agent in diesel fuel. More than forty catalysts were evaluated for the oxidative coupling of DME at 200 to 400C. Most catalysts exhibited no coupling activity. The major products were CO, CO2, methanol, and often methyl formate. A few catalysts when examined at low pressures showed trace formation of the coupling products 2-methoxyethanol, monoglyme, and diethylene glycol methylether. These included a mixed Bi-Mo-Fe oxide, MoO2, and V2O3. However, at higher pressures where coupling is favored, only the literature catalyst SnO2/MgO showed selectivity for the coupling product monoglyme, 5.9%, significantly better than that for an empty reactor, 2.4%. Monoglyme selectivities of 3.4% were obtained using the catalysts Bi-Mo-Fe oxide and Pr6O11, and 2.7% using MoO2, and V2O3. A wide range of CETANER-diesel fuel blend properties, including fuel miscibility, water tolerance, volatility, phase change characteristics, conductivity, flash point, and cetane number were studied. CETANER was shown to be completely miscible with diesel fuel to at least 40 vol %. Cetane numbers of diesel fuel blends generally increased with increasing CETANER concentration. Based on the flash point requirements, the maximum monoglyme concentration in a U.S. fuel blend is probably 4 vol %. Blends of CETANER with diesel fuel were found to be stable upon storage and free of hydroperoxides after one year. The glyme components of CETANER have relatively low toxicity; however, some evidence suggests that they may have teratogenic properties. The introduction of any new fuel additive into the marketplace requires that different constituencies accept the product, including those from the viewpoints of health, safety, the impact on the environment, and engine performance. Along with any future catalyst and process development, close alignment with stakeholders in each of these areas is recommended. 76

Bibliography Automotive Fuels Reference Book, 2nd edition, Owen, K; Coley, T.; Society of Automotive Engineers, Inc. (1995). Peckham, J. Hart's Fuel Technology & Management, 26-28 (July/August 1998). Yagita, H.; Asami, K.; Muramatsu, A.; Fujimoto, K.; Appl. Cat. 53, L5-L9 (1989). Estimates based on: Matherne, J.L; Culp, G.L. in "Methane Conversion by Oxidative Process. Fundamental and Engineering Aspects" Wolf, E.E., ed.; Van Nostrand Reinhold, Chapter 14 (1992). Donalson, J.D.; Moser, W. J. Chem. Soc. 1996-2000 (1961).

77

3.B.6 Construction of Alternative Fuels Field Test Unit (AFFTU) Introduction The Alternative Fuels Field Test Unit (AFFTU) is a portable laboratory designed specifically to provide on-site evaluation of potential feedstocks for processes that produce alternative fuels from indigenous raw materials such as coal, natural gas or environmentally disadvantaged carbonaceous feedstocks. Since conversion of these raw materials into feed gas streams can produce a variety of bulk gas compositions, which furthermore can contain a myriad of trace components, it is necessary to evaluate each new feedstock on an individual basis. While it is possible to prepare blended gas mixtures to simulate the bulk composition of a known feedstock, it is neither possible nor cost-effective to simulate adequately the variety of trace chemicals present in that feedstock -- some of which may not even be detected by routine analysis. Additionally, the transient composition of the gas during upsets or routine process changes may have an impact on the proposed process that is not foreseen in standard design. The AFFTU was constructed to address these critical design questions by providing an accurate, on-site measurement of the quality of syngas streams via: 1. 2. 3. 4. The capability to run the desired reaction for an extended period using syngas as feed. State-of-the-art trace gas analysis targeting specific, known catalyst poisons. Highly automated data acquisition and storage. Equipment and operating procedures designed with the safety of the operators and the equipment as the top priority.

Description To achieve portability, the AFFTU was constructed in a commercial 48-foot trailer. Roughly half of the trailer is dedicated as "office" space, and it contains three personal computers that serve as an interface to the process control and handles data acquisition and analysis. The other half houses the laboratory, which is highly automated and designed for unattended operation. A photograph of the AFFTU is provided in Figure B.17. The utility requirements of the AFFTU are minimal, allowing for maximum flexibility in siting. The specific requirements are: 1. 2. 3. 4. 5. 480V-3Ph-60Hz electrical supply (75 kVA, 90A transformer) Potable water Instrument air: 150 psig Instrument nitrogen: 150 psig Municipal drain/sewer for sink

The experimental apparatus (Figure B.18) consists of a feed manifold, adsorption (pretreatment) system and autoclave system. The feed manifold allows the blending of two feed streams, as well as any of several cylinder gases (2% H2 in N2 is shown in Figure B.18). Flows are

78

Figure B.17 Photograph of Alternative Fuels Field Test Unit (AFFTU)

Figure B.18 Simplified Overview of AFFTU Experimental Apparatus

Ads o rpt io n Sy s t e m BPR Gas -Liquid Se parat o r

1 Feed

BPR

H2/N2
WTM1 3 0 0 m L Aut o c lave

2 Feed

Nitrogen
Stars represent GC sampling points; squares represent mass flow controllers

79

controlled with mass flow controllers. The adsorption system is designed to permit up to four adsorbent beds to be placed in series in the feed stream for selective removal of various catalyst poisons potentially present in the syngas. A compressor is available if adsorption at pressures greater than the supplied feed pressure is desired. A back-pressure regulator (BPR) maintains the pressure in the adsorption system. The treated feed is sent to a 300-mL stirred autoclave. This reactor is equipped with a gas-liquid separator, maintained at 145C, to return any entrained slurry to the reactor. Feed pressure to the reactor may be boosted using a compressor; reactor pressure is set with a regulator and is maintained using a separate BPR. The reactor effluent is vented through a wet test meter (WTM) to obtain an accurate measurement of the reactor exit flow. The AFFTU was designed with the following experimental capabilities: 1. A state-of-the-art gas chromatograph system to perform semi-continuous monitoring of both bulk composition and the concentration of key trace poisons down to one part per billion by volume (ppbv). 2. A 300-mL reactor system that can accept up to two feed streams, allowing a true-life test with the actual gas projected for use in the proposed facility. 3. A manifold of four adsorbent beds, located upstream of the reactor, which permits the testing of adsorbents for the removal of contaminants from the feed stream. The effectiveness of these adsorbents may be evaluated either by analysis of the gas upstream and downstream of the bed (or at an intermediate point within the bed) or by observing the impact of the presence or absence of that bed on the actual stability of the catalyst activity. When not in use, the AFFTU has been located at Air Products' research facility, where it became an effective extension of the DOE program's research laboratories. During the design phase, the AFFTU was subjected to Air Products' Process Hazards Review protocol. This protocol reviews the equipment and procedures in terms of chemical hygiene/exposure, strategies to handle potential hazards, risk minimization, required training and personal protective equipment, operability and acceptable design practices. Following construction of the unit (which was completed on time and under budget), an Operational Readiness Inspection at Air Products verified that the equipment had been constructed in accordance with the Hazards Review and that all alarms and safety interlocks were calibrated and functioning properly. Shakedown testing was then performed to confirm that the results from the AFFTU were similar to those from the existing in-house laboratory systems. Several critical problems with the data acquisition software surfaced during this test run. These were corrected, allowing unattended data acquisition to be achieved. Testing at Eastman's Chemicals-from-Coal Complex The AFFTU was used at Eastman Chemical Companys Chemicals-from-Coal complex in May 1996, during the design phase of the LPMEOH Commercial Demonstration Project (this project is sponsored by the DOE's Clean Coal Technology program). The AFFTU was transported from Air Products' research facilities in Allentown, Pennsylvania. to the Eastman site in Kingsport, Tennessee to provide long-term (6 weeks) testing on the presence of trace species

80

that would be expected to enter the LPMEOH Commercial Demonstration Unit from Eastman's coal gasification system. Figure B.19 is a simplified flow diagram of the LPMEOH Commercial Demonstration Unit and the location of the AFFTU. Three gas streams were being made available as feedstocks to the process: "balanced syngas" from Eastman's coal gasifiers, "makeup CO" from a cold box and "H2 makeup" from Eastman's existing methanol unit. The reactor was expected to operate at 250C and 750 psig. Downstream of the reactor, unreacted syngas is separated from the products, and roughly 90% of it will be recycled. For most of the AFFTU experiments at Kingsport, a feed mixture of 75% balanced syngas and 25% makeup CO was used. This corresponded to a condition in the demonstration plan that used the highest quantity of makeup CO. This case was chosen since our objective was not to mimic the final plant design, but rather to evaluate the poisons concentrations for both streams. The adsorption system (Figure B.18) consisted of five beds -- four operating beds and a fifth bed filled with alumina, which is used to decompose toxic metal carbonyls desorbed from the other four beds during regeneration. Each bed was approximately 12 inches long and 0.65 inches inside diameter. Sampling ports were located at three intermediate positions along the bed. The choice of adsorbents and their position in the pretreatment sequence (Table B.13) was based on results from laboratory studies and the previous work with an earlier field test unit in Beulah, N.D. During the design phase for the AFFTU, details were supplied to Eastman for their review. Once the AFFTU was installed at Kingsport, a second ORI was performed by a combined team of Air Products and Eastman personnel. Minor changes in the tie-in tubing that connected the AFFTU to Eastman's syngas headers were implemented as a result. Results A chronology of the operation at the Eastman chemicals-from-coal complex is provided in Table B.14. Four materials that are known to poison the methanol synthesis catalysts were observed during at least portions of the testing period: iron carbonyl, nickel carbonyl, hydrogen sulfide and carbonyl sulfide. No other compounds (other than the bulk components of the feed and product streams) were observed with either an Electron Capture Detector or a Sulfur Chemiluminescence Detector. The packed bed of methanol synthesis catalyst was effective for the removal of <20 ppb iron carbonyl, <3 ppb hydrogen sulfide and <20 ppb carbonyl sulfide, but was saturated rapidly by the 10-200 ppb nickel carbonyl initially present in the feed stream. The activated carbon bed was effective for the removal of both iron and nickel carbonyl and the trace hydrogen sulfide in the feed. The BPL carbon did not remove carbonyl sulfide. The nickel carbonyl was clearly demonstrated to be an artifact of the fresh tie-in tubing between Eastman's piping and the AFFTU. Most, if not all, of the iron carbonyl was likely also an artifact of the tiein.

81

Figure B.19 Simplified Process Flow Diagram Kingsport LPMEOH Commercial Demonstration Unit

Table B.13 Adsorbents Used for Kingsport Testing Bed 1 2 3 4 5 Adsorbent Ground Methanol Catalyst Zeolite BPL Carbon Mixed CuO/ZnO Alumina Target Contaminant H2S, COS Ni(CO)4 Fe(CO)5, Ni(CO)4 AsH3 Decomposing Metal Carbonyls Weight 80.71 g 36 g 29.14 g 69.53 g 42.8 g Temp. Room Room Room 40C 250C

82

Table B.14 Chronology of Kingsport AFFTU Run Event Beginning of Run Bed #4 Dropped Bed #2 Dropped Bed #1 Dropped Bed #3 Dropped First Gasifier Outage Restored Second Gasifier Outage Restored Run Terminated Date & Time 5/15 1440 5/20 830 5/21 930 5/22 1000 5/26 1500 5/30 1430 5/31 400 6/6 730 6/10 1000 6/17 1700 Time Onstreama 0 114 139 163 264 360 360 503 503 678 Days of Operationb -0.35 4.4 5.44 6.56 10.67 14.65 15.21 21.36 25.47 32.76

a. The cumulative time (in hours) of exposure of the catalyst to feed gas. b. The number of days since the beginning of the experiment. The negative value at the beginning results from differing definitions of when the experiment began. Continuous syngas poisons monitoring ("Days of Operation") began several hours after the point we adopted as the basis for our "Time Onstream" datum.

The overall productivity history and the results from the analysis of this data in Air Products' reaction model are shown in Figure B.20. During the initial 50 hours on stream, fairly rapid loss of catalytic activity was observed. This is typical behavior for methanol production runs using this catalyst. After this initial period, activity decline was gradual and fairly steady. Several abrupt productivity losses (for example, at roughly 370, 390, 500, 540 and 660 hours on stream) are not reflected in the normalized rate constant km because these productivity fluctuations were due to changes in the feed CO2 content and not to changes in the actual activity of the catalyst. The overall rate of deactivation (-0.037% per hour) was similar to the results that have been calculated for blended syngas compositions in the absence of trace contaminants.

83

Figure B.20 Comparison of Normalized Rate Constant km and Methanol Productivity

1.2

0.8 Normalized Value

0.6

Dea ct iva t ion =0 .0 3 7 %/ h r

Rate Constant Productivity Best-Fit Line

0.4

0.2

0 0 100 200 300 400 500 600 700 Time On Stream (hours)

Summary and Conclusions The AFFTU was constructed on schedule and at the budgeted cost. The capabilities of the AFFTU to perform on-site evaluation of catalyst performance and stability coupled with state-ofthe-art gas chromatography for ppb level analysis of sulfides and metal carbonyls were demonstrated during the extended test at Eastman's chemicals-from-coal complex. The performance results from catalyst testing were similar to those achieved with blended syngas compositions in the absence of trace contaminants such as sulfur. Subsequent operation of the Kingsport LPMEOH Commercial Demonstration Unit indicated that the levels of trace contaminants in the balanced syngas feed adversely impacted the catalyst performance within the SBCR. The nature of operating with small quantities of catalyst in the laboratory autoclaves and the high ratio of surface area to volume in these units have been noted as potential causes of this elevated baseline deactivation. The AFFTU has provided important information on the presence and levels of trace contaminants in syngas, and the results from the autoclave can yield immediate information on catalyst performance as means to lower the baseline rate of catalyst deactivation are developed. Bibliography "Liquid Phase Methanol LaPorte Process Development Unit: Modification, Operation, and Support Studies. Task 3.8, Catalyst Poisons Field Demonstration," Topical Report No. DOE/PC/90005-T38, (9 Nov 1990).

84

3.B.7 New Kinetic Models for the LPMEOH Process The LPMEOH Process is the first liquid phase reaction technology developed at Air Products under DOE contract. During the term of this project, the LPMEOH Process was in the commercial demonstration stage at Eastman Chemical Company's chemicals-from-coal complex at Kingsport. Development of new kinetic models for the LPMEOH Process became necessary as the LPMEOH technology moved forward to the commercial front. The potential applications of this technology expanded from the original solid feedstock-based (e.g., coalbased) niche market to the large natural gas-based market. Engineering studies and commercial evaluations required more and more accurate simulations. The existing reaction network and corresponding rate models are shown in Table B.15. The methanol catalyst from which the original kinetic models were developed was no longer commercially available, and a new commercial methanol catalyst had been qualified in its place during the 1995 operation at the LaPorte AFDU (see Section 3.A.2). The recent commercial developments demanded more robust and accurate kinetic models over a wide range of reaction conditions. The main areas for improvement included more accurate prediction of methanol production and CO2 conversion, especially in the H2-rich syngas regime to cover the natural gasbased applications, and more accurate modeling of by-product formation for the new methanol catalyst. The first part of the development was to build a reaction network that correctly represents all of the important reactions under LPMEOH conditions. The on-line GCs were calibrated to detect all possible products. Further product identification was carried out by off-line GC/MS analysis of the reactor exit gas and condensable samples. All of the products in the original reaction network were observed with the new methanol catalyst, and the corresponding reactions remained in the new reaction network. In addition, two more reactions were identified. The first one is associated with the detection of a group of high alcohols (C6 to C12) that were not included in the original reaction network. These alcohols were lumped into a pseudo-compound, C8.6H18.1OH or Hicohol, and the following reaction of Hicohol formation was added to the network: 16.7 H2 + 8.6 CO Hicohol + 7.6 H2O (17)

No attempt was made to develop a specific rate model for Hicohol. Laboratory results indicated that the rate of Hicohol formation could be estimated by multiplying the rate of 1-pentanol formation by a factor of ten. Table B.16 shows that the addition of this reaction helped to better close the mass balance, as much as 2% under some reaction conditions. It also helped in modeling the water gas shift reaction, since water is formed along with these alcohols, and accurate modeling of water formation is crucial for modeling the water gas shift reaction.

85

Table B.15 Reactions in the Existing LPMEOH Reaction Network Product methanol ethanol 1-propanol 1-butanol isobutanol 1-pentanol DME methyl acetate methyl formate methane ethane propane water gas shift Reaction CO + 2H2 CH3OH CO + 2H2 + CH3OH EtOH + H2O CO + 2H2 + EtOH C3OH + H2O CO + 2H2 + C3OH C4OH + H2O CO + 2H2 + C3OH IBOH + H2O CO + 2H2 + C4OH C5OH + H2O 2CH3OH CH3OCH3 + H2O CO + 2CH3OH MeAc + H2O CO + CH3OH MeFm CO + 3H2 C1 + H2O 2CO + 5H2 C2 + 2H2O 3CO + 7 H2 C3 + 3H2O CO + H2O CO2 + H2 Table B. 16 Effect of Hicohol on Mass Balancea Feed Gas Shellb Shell Shell Texacoc Texaco Mass Balance w/o Hicohols C H O 6000 98.6 98.7 98.9 3000 98.3 97.7 98.9 9000 99.1 99.3 99.3 3000 98.3 99.1 98.6 6000 98.7 99.5 98.9 SV Mass Balance with Hicohols C H O 99.1 100.0 99.0 99.1 99.7 99.0 99.4 100.0 99.4 99.0 100.4 98.6 98.9 99.9 98.9

Reaction # 1 2 3 4 5 6 7 8 9 10 11 12 13

a: All at 250oC and 750 psig b: Shell gas = 30 vol % H2, 66 vol % CO, 3 vol % CO2, 1 vol % N2 c: Texaco gas = 35 vol % H2, 52 vol % CO, 12 vol % CO2, 1 vol % N2

The second reaction that was added to the new reaction network was CO2 hydrogenation to methanol: CO2 + 3H2 CH3OH + H2O (18)

The existing model underestimated the rate of CO2 conversion for all syngas compositions. An analysis based on the thermodynamic equilibrium demonstrated that this reaction provided the needed degree of freedom for modeling CO2 conversion.

86

The second part of the model development was to build a kinetic database that covers a wide range of reaction conditions. The laboratory reactor system was tuned, and GCs were calibrated frequently to provide consistent and high-quality kinetic data. Fifty-three different reaction conditions, including different feed gas compositions from a gas blending station, space velocities, pressures and temperatures, were examined. A statistical program developed by Air Products (StatisticStudio) was used to design supplemental kinetic experiments. Small, but significant, catalyst aging was observed in the kinetic experiments. All data were corrected for aging prior to being used for model development via an iterative procedure. The third part of the model development work was to formulate rate expressions. For the main reactions in the network, methanol formation from CO hydrogenation, methanol formation from CO2, and water gas shift, a number of Langmuir-Hinshelwood type rate expressions, from both the literature and in-house development, were examined, and the best ones were selected and reparameterized. The formation of 12 different by-products was modelled by empirical power-law rate expressions. Air Products' Statistic Studio program provided robust data regression. All fifteen new rate models exhibited much improved fit to the experimental data over the entire range of conditions compared to the existing models. Two examples are given in Figures B.21 and B.22, the parity charts for the CO-hydrogenation-to-methanol reaction and CO2 conversion, respectively. It can be seen that the new models fit the data more accurately. A by-product of this model development work is a new understanding of the effect of different rate models on catalyst life studies; this understanding provides useful guidelines for such studies. It was found that, when the catalyst aging rate is expressed by the change in the preexponential factor of the rate constant with time on stream, the aging rate depends on the sensitivity of the model to variations in reaction conditions (e.g., temperature, pressure, and gas composition). For a given reaction, different models could have different sensitivities. Furthermore, the sensitivity for a given model could vary from one reaction regime to another. More sensitive models or a model in a more sensitive regime would give a higher rate of catalyst aging. One needs to keep this artifact in mind while comparing the aging rates from different models or from different reaction regimes. Another effect related to the model sensitivity is how the errors in the raw data propagate to the calculated pre-exponential factor. Any errors in the raw data are magnified to a greater extent for a more sensitive rate model, appearing as greater scatter in the catalyst life data (i.e., pre-exponential factor vs. time on stream).

87

Figure B.21 Comparison of Experimental and Simulated CO Hydrogenation Rates of New and Existing Rate Models
120

Simulated CO hydrogenation Rate (mol/kg-hr)

100

New Model Existing model

80

60

40

20

0 0 20 40 60 80 100 120

Measured CO Hydrogenation Rate (mol/kg-hr)

Figure B.22 Comparison of Experimental and Simulated CO2 Hydrogenation Rates of New and Existing Rate Models
15

10

New model Existing model CO2 consumption

RCO2, simulated

-5

-10

-15 -15 -10 -5 0 5 10 15

RCO2, exprim.

88

Summary and Conclusions A more robust kinetic model for the LPMEOH reaction system was developed during this project. The development was needed to meet the requirements for more accurate process simulations over a wide range of conditions. To this end, kinetic experiments were designed based on commercial needs and a D-Optimal design package. A database covering 53 different conditions was built. Two new reactions were identified and added to the LPMEOH reaction network. New rate models were developed for all 15 reactions in the system. The new rate models are more robust than the original ones, showing better fit to the experimental results over a wide range of conditions. Related to this model development are some new understandings about the sensitivity of rate models and their effects on catalyst life study. With the improvements in the calculation of by-product selectivity and CO2 conversion, a more accurate prediction of the performance of the LPMEOH Reactor in future applications is available. Although some of these changes appear to be small in nature, they can have a significant impact on the per-pass conversion and required recycle ratio of unreacted syngas, which can lead to smaller plant sizes and costs.

89

3.C. Task 4 Program Support 3.C.1 Introduction Within Task 4, Program Support, Air Products and Bechtel Corporation performed various process, research, and economic studies when necessary throughout the program. Specifically, Bechtel provided support in the following areas: 1) Study of catalyst-wax separation systems for F-T synthesis; 2) Economic analysis for the production of Methyl tert-Butyl Ether (MTBE); and 3) Assessment of the levels of trace contaminants present in coal-derived syngas that could be contaminants to catalytic processes. 3.C.2 Catalyst-Wax Separation Study for Fischer-Tropsch Synthesis Introduction Bechtel Corporation conducted a catalyst separation. Development of a reliable and costeffective method of wax/catalyst separation is a key step toward a commercially viable slurry reactor process with iron oxide-based catalyst for F-T synthesis of hydrocarbon transportation fuels. Although a variety of suitable catalysts are available, iron oxide-based catalysts are preferred for coal-derived, CO-rich syngas because, in addition to catalyzing the F-T reaction, they simultaneously catalyze the reaction shifting CO to H2, obviating a separate shift process block and associated costs. Because of the importance of developing this wax/catalyst separation, a study was initiated in 1991 in which Burns and Roe reviewed the status of F-T wax/catalyst separation techniques, leading to the selection of a filtration system for the separation. Separation of the catalyst solids from the wax still represents a challenge. In 1992 testing of the selected filter yielded poor performance, and it was therefore recommended that the wax/catalyst separation be developed further. The objectives of this study were to 1) describe state-of-the-art techniques for the separation of the F-T wax from iron catalysts, 2) discuss the potential for a commercially viable separation method, and 3) present follow-up recommendations. Results Tests that have been performed with fresh catalysts have shown that certain techniques such as cross-flow filtration can produce a clean wax product that may require little to no after-treatment prior to use as feedstock to the F-T refinery. However, the technical challenge of producing a clean wax product from a F-T reactor using an iron oxide-based catalyst is complicated by the existence of a wide range of particle sizes that changes with time due to either physical damage within the SBCR or chemical changes to the iron. For this study, the "spectrum" in Figure C.1 was used; the particle size ranged from 100 to 0.08 micron, with 16 wt % smaller than 1 micron. Given this design basis, and the need to limit the feed from the catalyst-wax separation system to a downstream catalytic cracking unit to 2-5 ppm by weight of iron, a two-stage system was envisioned; Figure C.2 provides design parameters for a 50,000 barrel-per-day F-T plant. The two most important design parameters for selection of the separation process are required 90

product purity and catalyst size distribution. There are many methods to separate the wax and the catalyst. These methods are based on particle size, density differences, alteration of properties, magnetic differences, electrical charge, solubility, wettability and vapor pressure. Given the required product purity of 2 to 5 ppm of catalyst in the wax, this means that all particles smaller then 0.1 micron should be removed if the wax is used in a hydrocracker. The results of the comparison of the processes considered in this study are presented in Table C.1. Filtration with microfiltration membranes offers the possibility of removing the particles to the level required. This may be accomplished in one process step. Summary and Conclusions This study considered a variety of techniques to separate the heavy wax product from an ironbased F-T reactor. Bulk removal of up to 50 % of the catalyst can be achieved inexpensively by several of the processes. However, only the filtration process using membranes has the potential to produce a wax product at the specified solids level in a single step. Not enough is known about the particle size spectra of the catalyst, its constituents and the desired particle size to maintain overall performance in the reactor. This is critical for selecting a separation process. Accordingly, more measurements of the constituents and size distribution are needed, and a specification must be developed for use in the pilot testing. Membrane systems, as well as asymmetric inorganic membranes should be further evaluated. Other approaches should be followed if the membrane system does not appear to be able to achieve the testing objectives. Separation of the catalyst from the wax should become easier as the particle size of the catalyst increases. Catalyst manufacturers and researchers should be encouraged to develop more robust catalysts that resist attrition during handling and use.

91

Figure C.1 Particle Size Distribution of Iron Oxide Catalyst

Figure C.2 Catalyst-Wax Separation System 50,000 Barrels-per-Day F-T Plant

92

Table C.1 Comparison of Separation Processes

Suitable for: Capital Cost Bulk Residual Both

Process

Recommendation Backup design after modifications Drop

Comments Backup design may not work with present particle distribution Large insulated pressure vessel

Kerr-McGee ROSE Sedimentation Hydroclone Centrifuge Coagulation/Filter Aids Porous Medium Screen Filters Membrane Filters High-Gradient Magnetic Separation Chemical Enhanced Sedimentation

High Medium Low Medium Low Low Low Low High X X X X

Determine if it enhances membranes Possible wear of the hydroclone X X X Drop Consider for polishing filter if required Drop Drop X Testing X Test if others fail Drop Does not meet product purity Either lost catalyst or separation of chemical/catalyst requirements Lost catalyst Lost catalyst Experimentation required Demonstration requried Additional complications and chemical separation

93

3.C.3 Economics for Production of Methyl tert-Butyl Ether Study Parameters A DOE/NETL-funded study was conducted to examine the use of a liquid phase mixed alcohol synthesis (LPMAS) plant to produce gasoline-blending ethers (MTBE and tert-amyl methyl ether (TAME)). The LPMAS plant was integrated into three utilization scenarios: a coal-fed IGCC power plant, a petroleum refinery using coke as a gasification feedstock, and a standalone, natural gas-fed, partial-oxidation plant. The objective of the study was to establish targets for the development of catalysts for the LPMAS reaction. The conditions from earlier work on synthesis of mixed alcohols at Air Products were selected for the LPMAS reactor (GHSV = 5,000 standard liters/kg catalyst-hour, temperature = 315C, pressure = 1,800 psig, catalyst slurry concentration = 40 wt %). The optimum ratio of methanol to isobutanol was selected at 1.03 (a small amount of methanol is required to convert iso-amylene to TAME). In the IGCC scenario (Case 1, Figure C.3), syngas conversions need only be moderate because unconverted syngas can be utilized by the combined cycle system. The power production from the IGCC plant, 385 net MW, was based on the production from a single 3,000 TPD Shell gasifier without the addition of a LPMAS system. Since the LPMAS plant consumes a portion of the syngas, the coal feed rate would be higher than 3,000 TPD to maintain the 385 net MW production level. Figure C.3 IGCC/LPMAS Block Flow Diagram (Case 1)

Coal

Coal Gasification

Syngas

Acid Gas Removal

Clean syngas

LPMAS

Unconverted syngas

Combined Cycle

Power

Mixed alcohols

Alcohol Separation

C4+ alcohols

Alcohol Dehydration
C4+ Olefins

Methanol

MTBE/TAME Plant

MTBE/TAME

94

In the petroleum refinery scenario (Case 2, Figure C.4), syngas is produced from an 1,800 TPD coke gasification plant. High syngas conversions (~95%) would be required to avoid overloading the refinery fuel system with low Btu-content, unconverted syngas. To achieve these high conversions with the low H2/CO ratio syngas, a recycle system was required (because of the limit imposed by methanol equilibrium), steam was injected into the LPMAS reactor, and CO2 was removed from the recycle loop.

Figure C.4 Petroleum Refinery/LPMAS Block Flow Diagram (Case 2)

Coke

Coke Gasification

Syngas

Acid Gas Removal

Clean syngas

LPMAS

CO2

Mixed alcohols

Alcohol Separation

C4+ alcohols

Alcohol Dehydration

Methanol

H2 S

Steam

C4+ Olefins

Power

Unconverted syngas

Crude Methanol Iso-butane Power MTBE Fuel gas Natural gas

MTBE/TAME Plant

MTBE/TAME

Reformulated gasolines Conventional gasolines Jet, Diesel

285, 000 BPD Refinery

C4+ n-olefins

Alkylation

Alkylate

Sulfur Coke

95

In the standalone LPMAS scenario (Case 3, Figure C.5), 230 MMSCFD of natural gas is partially oxidized with steam and oxygen. Essentially complete conversions would be required to achieve a fuel-balanced plant. The cost of natural gas was assumed to be $1.00 per million Btu. Given that the CO2 recovery system represented 25% of the capital cost for the facility, a sensitivity study (Case 3S) considered the impact of eliminating this equipment. Process conditions were adjusted in order to maintain the high syngas conversion requirement.

Figure C.5 Standalone LPMAS Block Flow Diagram (Case 3)

CO2 Unconverted syngas to fuel

Natural Gas
Steam

Partial Oxidation

Syngas

CO2 Removal

LPMAS
H2O

Mixed alcohols

Alcohol Separation

C4+ alcohols

Alcohol Dehydration

Methanol

C4+ Olefins

MTBE/TAME Plant

MTBE/TAME

96

Table C.2 Summary of Conditions and Economic Analysis for LPMAS Study
Case Feed Syngas H2/CO ratio LPMAS reactor feed H2/CO ratio MeOH:isobutanol ratio LPMAS syngas recycle ratio Per pass conversion, % Overall conversion, % Productivity, g iBuOH/Kg-hr g MeOH/Kg-hr Reqd ether price @ 13% Internal Rate of Return, cents/gal Internal Rate of Return at 85 cents/gal ether, % 1 Coal 0.5 0.5 1.03 0 38-49 38-49 370-460 165-205 85-76 2 Coke 0.4 0.5 1.03 2.0 48 95 265 118 3 NG @ $1 1.54 1.8 1.03 2.3 52 98 285 126 68 3S NG @ $1 2.02 3.3 4.0 1.9 45 98 181 314 53

13.9

Summary and Conclusions The results of the economic analysis for this study are provided in Table C.2. A 13% internal rate of return was used to determine the required catalyst performance. The analysis indicated that:

For the coal-based IGCC facility (Case 1), a once-through LPMAS plant achieving syngas conversions in the range of 38-49% was found to be suitable, and the target catalyst productivity ranged from 370 to 460 gm isobutanol/kg catalyst-hr. At the most economical recycle ratio for the refinery application (Case 2), the target catalyst productivity was 265 gm isobutanol/kg catalyst-hr. At the most economical recycle ratio, the target catalyst productivity in the standalone LPMAS facility (Case 3) was 285 gm isobutanol/kg catalyst-hr. The economics of this scenario are highly dependent on the cost of the natural gas feedstock and the location of the plant. If the CO2 removal system can be removed, the potential exists to achieve the target economics at the lowest catalyst productivity (181 gm isobutanol/kg catalyst-hr).

For all case scenarios, the economics of a LPMAS plant were marginal at ether market prices. Large improvements over demonstrated catalyst productivity and alcohol selectivity would be

97

required. These results were used to direct the focus of the laboratory work on mixed alcohol synthesis (Section 3.B.4). Bibliography Stein, V.E., "Demonstration of a One-Step Slurry-Phase Process for the Co-Production of Methanol and Isobutanol," Topical Report Prepared for DOE by Air Products and Chemicals, Contract No. DE-AC22-91PC90018, June 1996.

3.C.4 Trace Metal Contaminants: Identification, Quantification and Removal Study Study Parameters Coal gasification and the associated syngas cleaning technology were developed to produce syngas that is clean enough to meet the environmental regulations when burned in a gas turbine. However, concentrations of certain trace constituents in fuel-grade syngas may be unacceptably high for catalytic chemical synthesis. Therefore it is necessary to identify the type and concentration of trace constituents in fuel-grade syngas and to determine whether they exceed or meet the specifications for catalytic processing. Bechtel carried out a trace contaminant literature search in two directions. The first search showed that under DOE's sponsorship, various research organizations are doing experimental work to understand the mechanisms that affect the fate of feed trace elements during combustion and gasification. This research will ultimately lead to the development of simulation models. For given coal, gasifier and operating conditions, these simulation models will predict concentrations of trace element compounds in syngas. The second search path found data published by Shell, BGL, Texaco and Dow for their respective gasification technologies. These data included concentrations of various trace constituents present in fuel-grade syngas. Table C.3 shows the classification of trace components that has been assembled in combustion systems, and the current work developed a similar methodology for gasification systems using the concept of the partition coefficient (ppmw in syngas/ppmw in coal * 106). For syngas streams produced via Shell or Texaco gasification, the partition coefficients for Na, K, Mg, Ca, Mn, Fe, Al, and Si are low. These are the most abundant trace elements in coal, and, given their low volatility at gasification conditions, it is reasonable to assume that the majority leave the gasifier with the bottom slag; the small fraction that does partition to the gas phase is readily removed in one of the downstream purification steps. For pulverized coal systems, these elements would be considered as refractory in Table C.3.

98

Table C.3 Classification of Elements in Pulverized Coal Boilers

Classification of Elements, Ratafia-Brown Meij Eu, Hf, La, Mn, Rb, Sc, Sm, Th, Zr Al, Ca, Ce, Cs, Eu, Fe, Hf, K, La, Mg, Sc, Sm, Si, Sr, Th, Ti Ba, Cr, Mn, Na, RB Be, Co, Cu, Ni, P, U, V, W As, Cd, Ge, Mo, Pb, Sb, Tl, Zn B, Br, C, Cl, F, Hg, I, N, S, Se

Class

Volatility Characteristics

I IIC IIB IIA III

refractory

Ba, Be. Bi, Co, Cr, Cs, Cu, Mo, semi-volatile to refractory Ni, Sr, Ta, Tl, U, V, W semi-volatile volatile to semi-volatile volatile As, Cd, Ga, Ge, Pb, Sb, Sn, Te, Ti, Zn I, Se, B Hg, Br, Cl, F

Partition coefficients for Ag, Cd, Sn, Sb, Hg, Tl, and Pb are two to four orders of magnitude greater that the coefficients for Na, K, etc. These elements have moderate volatilities at gasification conditions, but once volatilized are less easily recovered due to either the low solubility of their salts or the formation of submicron particles through homogeneous condensation, which makes it difficult to completely remove them in the downstream scrubbing stages. A parallel can been seen between high-temperature gasifiers and pulverized coal boilers, where these elements are volatile enough to be depleted in the slag/bottom ash, but enriched in the fines recovered from electrostatic precipitators. Non-metals F, Cl, Br, and B are all classified as very volatile, but are reported as having low partition coefficients in gasification processes. This is most likely due to the solubility of these species both elements and salts in the scrubbing stages. Examination of these data showed that fuel-grade syngases contain constituents like H2S, COS, HCL, HCN, PH3, Fe(CO)5, AsH3, and H2Se all known or potential poisons of methanol synthesis catalysts. As shown in Table C.4, the concentration of sulfur compounds would be the largest trace component in the feed to a methanol plant. After determining the type and concentration of the catalyst poisons, Bechtel contacted more than 30 vendors were contacted to obtain information on the cost and performance of systems to remove these compounds. Six vendors responded, recommending contaminant removal systems based on various adsorbents such as activated carbon, molecular sieves, metal oxides, and proprietary adsorbents. Based on vendor information, sizes of equipment, direct capital cost and operating cost for six syngases and six removal systems were prepared. Three systems of the initial six were considered to be promising:

99

Table C.4 Concentrations of Trace Contaminants in Fuel-Grade Syngas Streams

Expected Concentrations in Syngas Concentration Limit from Modern Coal for Methanol Catalysts, Compound Gasifiers, ppmv ppmv H2 S COS HCl HCN PH3 Fe(CO)5 Ni(CO)4 AsH3 H2Se 2 3 0.05 0.75 0.15-0.20 0.02-1.0 0.2 0.02 0.02 0.03 0.03 0.01 0.01 0.01 0.01 0.01 0.01 0.01

United Catalysts Inc. recommended the use of three adsorbents: G-132, a proprietary copperzinc adsorbent for removing H2S, COS, PH3, and AsH3; C125, composed of CaO and ZnO to remove HCl and HCN; and C8-1, a copper-promoted activated carbon to remove H2Se, Fe(CO)5 and Ni(CO)4. Two vessels (one operating and one in hot standby mode) were recommended to achieve a 1-2 month bed life in this service. ICI Katalco offered three adsorbents to clean the fuel-grade syngas stream: Puraspec 2110 for HCl removal, Puraspec 2132 for COS hydrolysis, and Puraspec 2220 for adsorption of the remaining components (including the H2S produced across the Pursapec 2132). Two sets of four beds (one set operating, the other on standby) were specified to achieve a life of 12 months for HCl and sulfur. Nucon International specified three activated carbon adsorbents for removal of all trace contaminants with the exception of COS, which had to be converted to H2S before it could be adsorbed: A-3, impregnated with a proprietary material to remove HCl, HCN, AsH3, PH3, and H2Se; FC-3, impregnated with metal oxides to remove H2S; and GC60-3, an unimpregnated material to remove Fe(CO)5 and Ni(CO)4. Including the separate COS hydrolysis unit, a total of three beds (one vessel on standby) were recommended to achieve a bed life of 1 month.

100

Table C.5 summarizes the production cost to upgrade a fuel-grade syngas containing 5 ppmv total sulfur to the requirements that could be expected for methanol synthesis catalysts. The incremental production cost ranged from 4-5 cents per million Btus for the activated carbon system, to 23-46 cents per million Btus for a metal adsorbent system. A lower cost metal-based system had a production cost in the range of 12-18 cents per million Btus. Table C.5 Production Cost to Upgrade Fuel-Grade Syngas to Expected Requirements for Methanol Synthesis Catalysts
Gasification Technology Shell SCGP Texaco TGP BGL Shell SCGP Texaco TGP BGL Production Cost, $/MMBtu United Nucon Catalysts ICI Katalco International 0.28 0.39 0.23 0.29 0.46 0.23 0.14 0.15 0.12 0.14 0.18 0.12 0.05 0.07 0.04 0.05 0.08 0.04

Coal Illinois No. 6 Illinois No. 6 Illinois No. 6 Wyoming PRB Wyoming PRB Wyoming PRB

Summary and Conclusions The available measured data on trace contaminants in fuel-grade syngas indicate that, even after conventional cleaning steps, components would still be present in concentrations that would exceed the anticipated recommended levels for methanol synthesis catalysts. These species would be expected to be present in the syngas from any of the modern entrained-bed coal gasification systems. Adsorption of trace contaminants on a fixed bed, by physisorption or chemisorption, was recommended as the most proven and effective technology for reducing the concentration of these species. The incremental purification costs were calculated to range from 4-46 cents per million Btus. Three vendors (United Catalysts, ICI Katalco, and Nucon International) offered systems that met the technical requirements. Additional testing with actual fuel-grade syngas was recommended. Bibliography Ratafia-Brown, J.A., "Overview of Trace Element Partitioning in Flames and Furnaces of Utility Coal-Fired Boilers," Trace Element Transformations in Coal Fired Power Systems Workshop, Scottsdale, AZ, 19-22 April 1993, page 139. Meij, R., "Trace Element Behavior in Coal-Fired Power Plants," ibid, page 199.

101

4. Conclusions This Final Report for Cooperative Agreement No. DE-FC22-95PC93052, the "Development of Alternative Fuels and Chemicals from Synthesis Gas," covers activities from 29 December 1994 through 31 July 2002. The overall objectives of this program were to investigate potential technologies for the conversion of syngas to oxygenated and hydrocarbon fuels and industrial chemicals, and to demonstrate the most promising technologies at the AFDU. Within this program, significant advancements toward achieving the goal of commercializing several of these technologies were made at the laboratory and proof-of-concept scale: Methanol: A campaign at the AFDU was performed to demonstrate operation at high velocity conditions (1.2 ft/sec) to improve commercial reactor design, to provide additional supporting information for the design and operation of the LPMEOH Commercial Demonstration Unit at Kingsport, to evaluate an alternate methanol catalyst, and to produce methanol for possible enduse testing. The data from this campaign provided confirmation of assumptions used in the design of the catalyst reduction system at Kingsport, and the alternate methanol catalyst has been in use since late 1998. The kinetic model was also expanded to allow for more accurate prediction of methanol production and CO2 conversion, and more accurate modeling of byproduct formation for the alternate methanol catalyst. The outstanding performance results of the LPMEOH Process at the Kingsport LPMEOH Commercial Demonstration Project can be attributed in large part to the body of work performed since 1981 in collaboration between the DOE and Air Products. Dimethyl Ether: During this program, a greater understanding of the performance and interaction of the LPDME catalyst system was achieved. The possible causes of catalyst deactivation under the LPDME conditions were investigated, and a novel catalyst deactivation mechanism was identified. Methods were developed to mitigate catalyst deactivation and identified stable operation conditions for commercial applications. The stable LPDME operation was successfully demonstrated at the LaPorte AFDU in 1999. The 0.7% per day aging rate achieved in the AFDU during this latest campaign was a large improvement over the 4% per day autoclave deactivation of the previous catalyst system. In the current program, a kinetic understanding of the LPDME reaction system was developed and applied to optimize the LPDME Process and improve its economics. DME has potential applications as a chemical building block. In addition, since the characteristics of DME are similar to those of LPG and given its higher heating value, it has been speculated that DME could be used in large scale power production, in home heating, in replacement of LPG (liquefied petroleum gas) for automobiles, and as a diesel fuel substitute or combustion supplement. Economic studies performed under the Clean Coal Technology Projects Kingsport LPMEOH Commercial Demonstration Project have shown that the cost of dimethyl ether produced from coal using the current catalyst system can be competitive with LPG in certain markets. At the end of this program, a pilot-plant-tested LPDME Process has been demonstrated, and the product cost of DME from coal-derived syngas can be competitive in certain locations and applications. Fischer-Tropsch: Significant advancements in the production of straight-chain hydrocarbons from a slurry-phase F-T process were made. Two operating campaigns were undertaken at the 102

AFDU, the second of which met the aggressive technical goals for the demonstration. A catalyst productivity of approximately 140 gm HC/hr-liter was achieved at reasonable system stability. The productivity ranged from 110-140 gm HC/hr-liter at various conditions during the 18 days of operations. The cross-flow filter arrangement performed well throughout the second demonstration test, producing a clean wax product that had not been achieved during previous operating campaigns. A separate study was performed to evaluate other systems to separate a clean wax product from the catalyst slurry. In addition to the developments for technologies at or near the proof-of-concept scale, other laboratory efforts were undertaken to determine if the processes and catalysts could be developed to produce other higher-value products from syngas. The production of isobutanol directly from syngas was continued from the earlier Alternative Fuels program, new routes for the production of vinyl acetate monomer were investigated, and the production of an oxygenated liquid additive to diesel fuels (CETANER) was developed. None of these areas of interest met the requirements for testing at the scale of the AFDU during the performance period. A survey was also performed which identified the categories of trace components in coal-derived syngas and the means to economically remove these species. The work performed under this Cooperative Agreement has continued to promote the development of technologies that use clean syngas produced from any one of a variety of sources (including coal) for the production of a spectrum of alternative fuels (hydrocarbons and oxygenate fuels), octane enhancers, and chemicals and chemical intermediates. The need for liquid fuels will continue to be a critical concern for this nation in the 21st century. Efforts are needed to ensure the development and demonstration of economically competitive, efficient, environmentally responsible technologies that produce clean fuels and chemicals from coal under DOE's Vision 21 concept. These liquids will be a component of the fuel mix that will provide the transition from the current reliance on carbon-based fuels to the ultimate use of H2 as a means of energy transport. Indirect liquefaction, which converts the syngas (H2 and CO) produced by the gasification of coal to sulfur- and N2-free liquid products, is a key component of the Vision 21 initiative. The results from this current program provide continued support to the objectives for the conversion of domestic coal to electric power and co-produced clean liquid fuels and chemicals in an environmentally superior manner.

103

5. General References 1. Bhatt, B. L., "Liquid Phase Fluid Dynamic (Methanol) Run in the LaPorte Alternative Fuels Development Unit," Topical Report Prepared for DOE by Air Products and Chemicals, Contract No. DE-FC22-95PC93052, May 1997. 2. Degaleesan, S., Dudukovic, M. P., Toseland, B. A. and Bhatt, B. L., "Tracer Studies of the LaPorte AFDU Reactor During Methanol Synthesis," Topical Report Prepared for DOE by Washington University and Air Products and Chemicals, Contract No. DE-AC2295PC95051, September 1996. 3. Bhatt, B. L., "Liquid Phase Fischer-Tropsch Demonstration in the LaPorte Alternative Fuels Development Unit," Topical Report Prepared for DOE by Air Products and Chemicals, Contract No. DE-AC22-91PC90018, June 1994. 4. Bhatt, B. L., "Liquid Phase Fischer-Tropsch (II) Demonstration in the LaPorte Alternative Fuels Development Unit," Topical Report prepared by Air Products and Chemicals for U. S. DOE, Contract No. DE-AC22-91PC90018, September 1995. 5. Bhatt, B. L., "Liquid Phase Fischer-Tropsch (III & IV) Demonstration in the LaPorte Alternative Fuels Development Unit," Topical Report Prepared for DOE by Air Products and Chemicals, Contract No. DE-FC22-95PC93052, June 1999. 6. "Development of Alternative Fuels from Coal Derived Syngas, Task 2.2: Demonstration of a One-step Slurry-Phase Process for the Production of Dimethyl Ether/Methanol Mixtures at the LaPorte Alternative Fuels Development Unit," Topical Report Prepared for DOE by Air Products and Chemicals, Contract No. DE-AC22-91PC90018, 1 June 1993. 7. "Liquid Phase Dimethyl Ether Demonstration in the LaPorte Alternative Fuels Development Unit," Topical Report Prepared for DOE by Air Products and Chemicals, Contract Nos. DEFC22-92PC90543 and DE-FC22-95PC93052, January 2001.

104

List of Quarterly Reports submitted by Air Products to DOE

Report # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Calendar Quarter Oct - Dec, 1994 Jan - Mar 1995 Apr - June, 1995 July - Sept., 1995 Oct - Dec, 1995 Jan - Mar, 1996 Apr - June, 1996 July - Sept, 1996 Oct - Dec, 1996 Jan - Mar, 1997 Apr - June, 1997 July - Sept, 1997 Oct - Dec, 1997 Jan - Mar, 1998 Apr - June, 1998 July-Sept., 1998 Oct. - Dec., 1998 Jan.- Mar., 1999 Apr - June, 1999 July-Sept., 1999 Oct-Dec., 1999 Jan.- Mar., 2000 Apr - June, 2000 July-Sept., 2000 Oct-Dec., 2000 Jan.- Mar., 2001 Apr - June, 2001 July-Sept., 2001 Oct Dec., 2001 Jan Mar 2002 Apr - June, 2002 OSTI Document # 621748 569013 674613 1669 750406 780890 Date Report Submitted 2 Oct 1995 1 Nov 1995 5 Feb 1996 29 Mar 1996 9 May 1996 13 June 1996 27 Aug 1996 26 Nov 1996 23 Apr 1997 18 July 1997 20 Aug 1997 8 Dec 1997 27 Feb 1998 28 May 1998 17 Aug 1998 10 Nov 1998 19 Feb 1999 28 May 1999 24 Aug 1999 19 Nov 1999 23 Feb 2000 15 June 2000 10 April 2001 9 April 2001 9 April 2001 23 May 2001 31 Aug 2001 27 Dec 2001 18 Feb 2002 22 May 2002 01 Aug 2002

9060 9316 780893 2008

780891 780892 780894 774950 794367

801224

105

List of Topical Reports submitted by Air Products to DOE

Report # 1 2 3 4 5 Title Isobutanol from Syngas in a Three-Phase System - University of Aachen Fischer-Tropsch Wax/Catalyst Separation Study - Bechtel Economics of MTBE via Mixed Alcohol Synthesis - Bechtel Oxygenates via Syngas - Lehigh University Design and Construction of the Alt. Fuels Field Test Unit and Liquid Phase Methanol Test at Eastman Chemical Co. Air Products Liquid Phase Fluid Dynamic (Methanol) Run in the LaPorte Alternative Fuels Development Unit Air Products Vinyl Acetate Monomer Air Products Liquid Phase Fischer-Tropsch (III & IV) Demonstration in the LaPorte Alternative Fuels Development Unit Air Products Catalyst Activity Maintenance for the Liquid Phase Synthesis Gas-to-Dimethyl Ether Process - Part I: An Investigation of the Cause and Mechanism of Catalyst Deactivation Air Products Molecular Sieves as Catalysts for Methanol Dehydration in the LPDME Process Air Products Scaleup of Aluminum Phosphate Catalyst for Pilot Plant LPDME Run Air Products Catalyst Activity Maintenance for the Liquid Phase Synthesis Gas-to-Dimethyl Ether Process - Part II: Development of Aluminum Phosphate as the Dehydration Catalyst for the Single-Step Liquid Phase Syngas-to-DME Process Air Products Cetane Enhancers by Oxidative Coupling of Dimethyl Ether Air Products Development of Kinetic Models for the Liquid Phase Methanol (LPMEOH) Process Air Products Kinetic Understanding of the Syngas to DME Reaction System and Its Implications to Process and Economics Air Products OSTI Document #

750407 750390 750374 794368

750387

7 8 9

750388

10 11 12

801223

13 14 15

106

Topical Report Summaries 1) Isobutanol from Syngas in a Three-Phase System - University of Aachen Two high-pressure units for isobutanol synthesis from syngas were designed and built for usage with the CSTR, as well as the PFR. Previously obtained results in isobutanol synthesis from syngas could be reproduced. Isobutanol synthesis from syngas could be performed in a two- and three-phase CSTR, even at high reaction conditions. Isobutanol yields up to 80 g/(Icat h) were reached. The influence of carbon dioxide in the PFR and the CSTR on isobutanol synthesis from syngas could be shown for different catalytic systems. From these experiments, it could be demonstrated that the C1C2-step is not the only rate-limiting reaction step in isobutanol synthesis. Heat and mass transfer calculations and experiments in isobutanol synthesis from syngas were performed in the PFR and the CSTR. The sol-gel catalyst synthesis method was extensively studied, leading to comparable catalysts. The activity of the Zr/Zn/Mn-oxide catalyst system could be enhanced at lower reaction conditions. Copper-containing catalysts showed high activity towards isobutanol with a remarkable thermal stability. Comparing the work of the various other research groups in this field showed that Falter's catalyst is the best developed so far. However, the severe reaction conditions needed will make it difficult to operate this catalyst commercially. 2) Fischer-Tropsch Wax/Catalyst Separation Study - Bechtel The two most important design parameters for selection of the separation process are required product purity (<2 to 5 ppm) and catalyst size distribution. The size distribution is much smaller than in the earlier Mott laboratory testing. The fine particle size makes this a difficult separation process. In the baseline design report, the Kerr-McGee ROSE process was evaluated. This is a two-step process with a hydrocyclone to separate larger particles before the Kerr-McGee ROSE separation. This design can lead to a buildup of higher-molecular-weight waxes in the reactor. SASOL has a patent application for wax/catalyst separation that is based upon filtration within the reactor. There are some major differences from the baseline design. These include a much larger catalyst size and wax that is sold as a by-product and not used further to produce a liquid fuel (i.e., the wax may have less stringent product purity limits). There are many methods to separate the wax and the catalyst. These methods are based on: particle size, density differences, alteration of properties, magnetic differences, electrical charge, solubility, wettability, and vapor pressure. The product purity should be 2 to 5 ppm of catalyst in the wax. This means that all particles smaller than 0.1 micron should be removed if the wax is used in the hydrocracker.

107

Filtration with microfiltration membranes offers the possibility of removing the particles to the level required. This may be accomplished in one process step. 3) Economics of MTBE via Mixed Alcohol Synthesis - Bechtel A study was also conducted to examine the use of a liquid phase mixed alcohol synthesis (LPMAS) plant to produce gasoline-blending ethers such as methyl tert-butyl ether (MTBE). The LPMAS plant was integrated into three utilization scenarios: a coal-fed IGCC power plant, a petroleum refinery using coke as a gasification feedstock, and a standalone, natural gas-fed, partial-oxidation plant. The objective of the study was to establish targets for the development of catalysis for the LPMAS reaction. In the IGCC scenario, syngas conversions need only be moderate because unconverted syngas is utilized by the combined cycle system. A once-through LPMAS plant achieving syngas conversions in the range of 38-49% was found to be suitable. In the petroleum refinery scenario, high conversions (~95%) are required to avoid overloading the refinery fuel system with low Btu-content, unconverted syngas. To achieve these high conversions with the low H2/CO ratio syngas, a recycle system was required (because of the limit imposed by methanol equilibrium), steam was injected into the LPMAS reactor, and CO2 was removed from the recycle loop. In the standalone LPMAS scenario, essentially complete conversions are required to achieve a fuelbalanced plant. The economics of this scenario are highly dependent on the cost of the natural gas feedstock and the location of the plant. For all three case scenarios, the economics of a LPMAS plant were marginal at current ether market prices. Large improvements over demonstrated catalyst productivity and alcohol selectivity will be required. 4) Oxygenates via Syngas - Lehigh University Methanol synthesis from H2/CO has been carried out at 7.6 MPa over zirconia-supported copper catalysts. Catalysts with nominal compositions of 10/90 mol% and 30/70 mol% Cu/ZrO2 were used in this study. Additionally, a 3 mol %, cesium-doped, 10/90 catalyst was prepared to study the effect of doping with heavy alkali, and this promoter greatly increased the methanol productivity. The effects of CO2 addition, water injection, reaction temperature, and H2/CO ratio have been investigated. Both CO2 addition to the synthesis gas and cesium doping of the catalyst promoted methanol synthesis, while inhibiting the synthesis of dimethyl ether. Injection of water, however, was found to slightly suppress methanol and dimethyl ether formation as it was converted to CO2 via the water gas shift reaction over these catalysts. There was no clear correlation between copper surface area and catalyst activity. Surface analysis of the tested samples revealed that copper tended to migrate and enrich the catalyst surface. The concept of employing a double-bed reactor with a pronounced temperature gradient to enhance higher alcohol synthesis was explored, and it was found that utilization of a Cspromoted Cu/ZnO/Cr2O3 catalyst as a first lower temperature bed and a Cs-promoted ZnO/Cr2O3 catalyst as a second high-temperature bed significantly promoted the productivity of 2-methyl-1propanol (isobutanol) from H2/CO synthesis gas mixtures. While the conversion of CO to C2+ oxygenates over the double-bed configuration was comparable to that observed over the single Cu-based catalyst, major changes in the product distribution occurred by the coupling to the zinc chromite catalyst; that is, the productivity of the C1-C3 alcohols decreased dramatically, and 2methyl branched alcohols were selectively formed. The target molar ratio of methanol to 2108

methyl oxygenate molar ratios of approximately 1 (desirable for feedstock for high-octane and high-cetane synthesis) was obtained in the present double-bed system, and the isobutanol productivity was as high as 139 gm/kg cat-hr. Higher alcohol synthesis has been investigated over a Cs/Cu/ZnO/Cr2O3 catalyst at temperatures higher (up to 703K) than those previously utilized, and no sintering of the catalyst was observed during the short-term testing. However, the higher reaction temperatures led to lower CO conversion levels and lower yield of alcohols, especially of methanol, because of equilibrium limitations. With the double catalyst bed configuration, the effect of pressure in the range of 7.612.4 MPa on catalyst activity and selectivity was studied. The upper bed was composed of the copper-based catalyst at 598K, and the lower bed consisted of a copper-free Cs-ZnO/Cr2O3 catalyst at a high temperature of 678K. High pressure was found to increase CO conversion to oxygenated products, although the increase in isobutanol productivity did not keep pace with that of methanol. It was also shown that the Cs/Cu/ZnO/Cr2O3 catalyst could be utilized to advantage as the second-bed catalyst at 613-643K instead of the previously used copper-free CsZnO/Cr2O3 catalyst at higher temperature. With double Cs/Cu/ZnO/Cr2O3 catalysts, high space time yields of up to 202 g/kg cat/hr, with high selectivity to isobutanol, were achieved. 5) Design and Construction of the Alternative Fuels Field Test Unit and Liquid Phase Methanol Test at Eastman Chemical Co. Air Products The Alternative Fuels Field Test Unit (AFFTU) is a portable laboratory designed specifically to provide on-site evaluation of potential feedstocks for processes that produce alternative fuels from indigenous raw materials such as coal, natural gas or environmentally disadvantaged carbonaceous feedstocks. The AFFTU was designed and constructed during this program within budget and on schedule. The AFFTU was first utilized to provide long-term testing at the LPMEOH Commercial Demonstration Unit at Kingsport. Two forms of testing were employed: (1) a life test of the LPMEOH reaction in a 300-mL reactor using the actual feed streams and (2) semi-continuous analysis of those same feed streams using gas chromatographs equipped with detectors sensitive to targeted poisons. Stable LPMEOH catalyst activity was demonstrated over a 28-day life test using the actual syngas feed streams for the Kingsport LPMEOH Commercial Demonstration Unit. 6) Liquid Phase Fluid Dynamic (Methanol) Run in the LaPorte Alternative Fuels Development Unit - Air Products A fluid dynamic study was successfully completed in a bubble column at the LaPorte AFDU. Significant fluid dynamic information was gathered at pilot scale during three weeks of LPMEOH operations in June 1995. In addition to the usual nuclear density and temperature measurements, unique differential pressure data were collected using Sandia's high-speed data acquisition system to gain insight on flow regime characteristics and bubble size distribution. Statistical analysis of the fluctuations in the pressure data suggests that the column was being operated in the churn turbulent regime at most of the velocities considered. Dynamic gas disengagement experiments showed a different behavior than seen in low-pressure, cold-flow work. Operation with a superficial gas velocity of 1.2 ft/sec was achieved during this run, with stable fluid dynamics and catalyst performance. Improvements included for catalyst activation in the design of the Clean Coal III LPMEOH Commercial Demonstration Unit at Kingsport,

109

Tennessee, were also confirmed. In addition, an alternate catalyst was demonstrated for the LPMEOH Process. 7) Vinyl Acetate Monomer Eastman Chemical Company Eastman Chemical Company studied the possible pathways to replace the existing ethylenebased VAM process with an entirely syngas-based process. Although there are many options for the conversion of syngas to VAM, Eastman undertook an analytical approach, based on known chemical and economic principles, to reduce the potential candidate processes to a select group of eight processes. The critical technologies that would be required for these routes were: 1. 2. 3. 4. the esterification of acetaldehyde with ketene to generate VAM, the hydrogenation of ketene to acetaldehyde, the hydrogenation of acetic acid to acetaldehyde, the reductive carbonylation of methanol to acetaldehyde.

This analysis showed that the cost of production of VAM from syngas was about 15% higher than the conventional oxidative acetoxylation of ethylene, primarily due to higher capital cost associated with the syngas-based processes. 8) Liquid Phase Fischer-Tropsch (III & IV) Demonstration in the LaPorte Alternative Fuels Development Unit Air Products Slurry phase Fischer-Tropsch technology was successfully demonstrated in the LaPorte AFDU. Earlier work at LaPorte, with iron catalysts in 1992 and 1994, had established proof-of-concept status for the slurry phase process. The third campaign (Fischer-Tropsch III), in 1996, aimed at aggressively extending the operability of the slurry reactor using a proprietary cobalt catalyst. Due to an irreversible plugging of catalyst-wax separation filters as a result of unexpected catalyst fines generation, the operations had to be terminated after seven days on-stream. Following an extensive post-run investigation by the participants, the campaign was successfully completed in March-April 1998, with an improved proprietary cobalt catalyst. A productivity of approximately 140 grams (gm) of hydrocarbons (HC)/ hour (hr)-liter (lit) of expanded slurry volume was achieved at reasonable system stability during the second trial (Fischer-Tropsch IV). The productivity ranged from 110-140 at various conditions during the 18 days of operations. The catalyst/wax filters performed well throughout the demonstration, producing a clean wax product. For the most part, only one of the four filter housings was needed for catalyst/wax filtration. The filter flux appeared to exceed the design flux. A combination of use of a stronger catalyst and some innovative filtration techniques were responsible for this success. There was no sign of catalyst particle attrition, and very little erosion of the slurry pump was observed, in contrast to the Fischer-Tropsch III operations. The reactor operated in a hydrodynamically stable manner, with uniform temperature profile and gas holdups. Nuclear density and differential pressure measurements indicated somewhat higher than expected gas holdup (45 - 50 vol %) during Fischer-Tropsch IV operations. The high gas holdup was confirmed by a dynamic gas disengagement test conducted at the end of the run. Heat transfer in the reactor was better than expected. Heat, mass and elemental balance calculations indicated excellent closure. After the initial learning curve with system dynamics, 110

the plant was restarted very quickly (24 hours and 17 hours) following two plant trips. This demonstrated the ease and flexibility of the slurry technology. Close to expected syngas conversion was obtained at the beginning of the run. The selectivity to wax was lower than expected, with higher methane selectivity. Returning to the baseline condition indicated a productivity decline from 135-140 to 125-130 gm HC/hr-lit. of reactor volume in two weeks of operation. This may be a result of some catalyst loss from the reactor as well as initial catalyst deactivation. The participants collected significant quantities of product and samples for further processing and analysis. Gas-, liquid- and solid-phase mixing were studied as planned at two operating conditions using radioactive materials. ICI Tracerco collected a large amount of data using 43 detectors around the reactor. The data were being analyzed by Washington University as part of the Hydrodynamic Program with DOE. 9) Catalyst Activity Maintenance for the Liquid Phase Synthesis Gas-to-Dimethyl Ether Process - Part I: An Investigation of the Cause and Mechanism of Catalyst Deactivation Air Products In the LPDME Process under development at Air Products, synthesis gas is converted into DME in a slurry phase reactor over a catalyst system with both methanol synthesis and methanol dehydration functionalities. Three reactions take place simultaneously in the reactor, namely, methanol synthesis, methanol dehydration and water gas shift reactions. A chemical synergy among these three reactions leads to substantial syngas conversion per pass and makes the process especially suitable for CO-rich gas. Furthermore, the slurry phase operation provides the superior heat management that makes the high conversion and direct use of CO-rich gas possible in practice. All these merits were proven by our laboratory experiments and a demonstration at the LaPorte pilot plant in 1991. Among the catalyst systems examined in our laboratory was a physical mixture containing a commercial methanol synthesis catalyst and -alumina as the dehydration catalyst. While the methanol catalyst and -alumina were stable when used separately in a slurry reactor for methanol synthesis and methanol dehydration reactions, both catalysts deactivated rapidly under LPDME conditions. This report describes our investigation into the cause of this accelerated catalyst deactivation and the underlying mechanism. Traditional causes for catalyst deactivation, such as leaching, hydrothermal sintering, coking, and poisoning, were investigated and subsequently ruled out. A novel cause of catalyst deactivation, namely, detrimental interaction between the methanol synthesis and methanol dehydration catalysts, was identified. Intimate, solid-state contact between the two catalysts in the slurry phase is necessary for the interaction to take place. The nature of this interaction is most likely inter-catalyst migration under the reaction conditions. Zn- and/or Cu-containing species may migrate onto the alumina to poison the acid sites, therefore destroying the dehydration activity. In the meantime, the methanol catalyst loses its activity by losing its active components. The reverse process, i.e., migration of alumina onto the methanol catalyst to poison its active sites, may also occur. There are a number of possible driving forces for this intercatalyst migration: 1) solid-state reaction between the base in the methanol catalyst (e.g., ZnO) 111

and the acid sites on the alumina, 2) solid-state ion exchange of Zn- or Cu-containing species with the protons (Brnsted acid) on the dehydration catalyst, and 3) the concentration gradient between the two catalysts, i.e., spontaneous dispersion, considering that alumina is a good support for metal, metal oxides, and salts. Some evidence in support of the migration hypothesis was obtained from SEM/EDS analysis. Further understanding of the mechanism was obtained from screening experiments for alternative dehydration catalysts. The results show that acid properties of dehydration catalysts play an important role in the interaction. Acid sites of greater strength on a dehydration catalyst resulted in more rapid deactivation of the methanol catalyst, while the strong sites themselves also had a higher rate of deactivation. Brnsted acid sites appeared to be extremely vulnerable under LPDME conditions. Good stability was observed from dual-catalyst systems with low dehydration activity. These understandings and observations paved the road for the development of our current stable dual-catalyst system. 10) Molecular Sieves as Catalysts for Methanol Dehydration in the LPDMETM Process Air Products Several classes of molecular sieves were investigated as methanol dehydration catalysts for the LPDME Process. Molecular sieves offer a number of attractive features as potential catalysts for the conversion of methanol to DME. These include (1) a wide range of acid strengths, (2) diverse architectures and channel connectivities that provide latitude for steric control, (3) high active site density, (4) well-investigated syntheses and characterization, and (5) commercial availability in some cases. We directed our work in two areas: (1) a general exploration of the catalytic behavior of various classes of molecular sieves in the LPDME system and (2) a focused effort to prepare and test zeolites with predominantly Lewis acidity. In our general exploration, we looked at such diverse materials as chabazites, mordenites, pentasils, SAPOs, and ALPOs. Our work with Lewis acidity sought to exploit the structural advantages of zeolites without the interfering effects of deleterious Brnsted sites. We used zeolite Ultrastable Y (USY) as our base material because it possesses a high proportion of Lewis acid sites. This work was extended by modifying the USY through ion exchange to try to neutralize residual Brnsted acidity. We confirmed that many molecular sieves possess very high intrinsic activity for methanol dehydration to DME. However, no molecular sieve catalysts were found that provided a combination of activity, stability, and compatibility with methanol catalyst superior to our existing LPDME catalysts. Therefore, we concluded that zeolites and related molecular sieves are not suitable for use in the LPDME Process. Most materials deactivated very rapidly even in the absence of methanol catalyst. This deactivation was caused by formation of nonvolatile hydrocarbons via a Brnsted acid catalyzed mechanism. We had some success in suppressing this deactivation by eliminating Brnsted acid sites through ion exchange, but at the cost of activity.

112

Molecular sieves with weaker Brnsted acidity, such as SAPOs, ALPOs, and boron-substituted zeolites, were not active enough, deactivated rapidly, or exhibited both behaviors. We were not able to find materials that could trade off balance activity and stability. Using different molecular sieve structures significantly altered the activity and deactivation behavior. Catalysts with one-dimensional channel systems generally had much lower, but more stable, activity. In addition, in some cases these materials also appeared to have a less deleterious effect on the methanol catalyst. Three-dimensional channel structures favored high activity. However, even using zeolites with very constrained channels (such as chabazite or zeolite rho) did not suppress deactivation appreciably. 11) Scaleup of the Aluminum Phosphate Catalyst for Pilot Plant LPDME Run Air Products In order to improve the stability of the LPDME catalyst system, non-stoichiometric aluminum phosphate was proposed as the dehydration catalyst for the LPDME process. This aluminum phosphate material is a proprietary catalyst. This catalyst system of a standard methanol catalyst and the aluminum phosphate provided stable process performance that met the program targets under the standard test process conditions in the laboratory. These targets were (1) an initial methanol equivalent productivity of 28 gmol/kg/hr, (2) a CO2-free, carbon selectivity of 80% to dimethyl ether and (3) stability of both catalysts equivalent to that of the methanol catalyst in the absence of the aluminum phosphate. A pilot plant trial of the LPDME Process using the aluminum phosphate catalyst was originally planned for March 1998 at the LaPorte AFDU. Because the aluminum phosphate catalyst was not commercially available, a scaleup project was initiated with a commercial catalyst vendor. A total of 800 pounds of aluminum phosphate catalyst was ordered to provide two reactor charges and some additional material for testing. Although the scaleup was never completed, the effort yielded valuable information about the nature of the catalyst and the nature of the LPDME Process. This information is documented in this topical report. 12) Catalyst Activity Maintenance for the Liquid Phase Synthesis Gas-to-Dimethyl Ether Process - Part II: Development of Aluminum Phosphate as the Dehydration Catalyst for the Single-Step Liquid Phase Syngas-to-DME Process Air Products At the heart of the LPDME Process is a catalyst system that can be active as well as stable. In the Alternative Fuels and Chemicals From Synthesis Gas project, a dual-catalyst system containing a Cu-based commercial methanol synthesis catalyst and a commercial dehydration material (-alumina) was demonstrated. It provided the productivity and selectivity expected from the LPDME Process. However, the catalyst system deactivated too rapidly to warrant a viable commercial process. The mechanistic investigation in the early part of this DOE program revealed that the accelerated catalyst deactivation under LPDME conditions was due to detrimental interaction between the methanol synthesis catalyst and methanol dehydration catalyst. The interaction was attributed to migration of Cu- and/or Zn-containing species from the synthesis catalyst to the dehydration catalyst. Identification of a dehydration catalyst that did not lead to this detrimental interaction while retaining adequate dehydration activity was elusive. Twenty-nine different dehydration materials were tested, but none showed the desired performance.

113

The search came to a turning point when aluminum phosphate was tested. This amorphous material is prepared by precipitating a solution containing Al(NO3)3 and H3PO4 with NH4OH, followed by washing, drying and calcination. The aluminum phosphate catalyst has adequate dehydration activity and good stability. It can co-exist with the Cu-based methanol synthesis catalyst without negatively affecting the latter catalysts stability. This report documents the details of the development of this catalyst. These include initial leads, efforts in improving activity and stability, investigation and development of the best preparation parameters and procedures, mechanistic understanding and resulting preparation guidelines, and the accomplishments of this work. 13) Cetane Enhancers by Oxidative Coupling of Dimethyl Ether Air Products Development and eventual commercialization of the LPDME Process will require a substantial growth in the demand for DME. An alternative, large market for DME is as a feedstock to the synthesis of liquid additives for diesel fuel. These additives, dubbed "CETANER," consist of a mixture of oxygenates and function as cetane enhancers. Fuels with high cetane number, which reflects the fuel ignition quality, offer several environmental and economic advantages. The potential market for CETANER technology is huge, perhaps 4,000 tons/day of CETANER along with 2,100 to 2,600 tons/day of O2 for its production. A process analysis shows that CETANER can be competitively priced with alternative cetane enhancer technology provided that target selectivities and conversions can be achieved. The major objectives of the current program were (1) to identify catalysts that permit a practical, cost-effective route to the oxidative coupling of dimethyl ether to CETANER and (2) to evaluate the properties critical to the use of CETANER as a blending agent in diesel fuel. More than forty catalysts were evaluated for the oxidative coupling of DME at 200 to 400C. Most catalysts exhibited no coupling activity. The major products were CO, CO2, methanol, and often methyl formate. A few catalysts when examined at low pressures showed trace formation of the coupling products 2-methoxyethanol, monoglyme, and diethylene glycol methylether. These included a mixed Bi-Mo-Fe oxide, MoO2, and V2O3. However, at higher pressures where coupling is favored, only the literature catalyst SnO2/MgO showed a selectivity for the coupling product monoglyme, 5.9%, significantly better than that for an empty reactor, 2.4%. Monoglyme selectivities of 3.4% were obtained using the catalysts Bi-Mo-Fe oxide and Pr6O11 and 2.7% using MoO2, and V2O3. A wide range of CETANER-diesel fuel blend properties including fuel miscibility, water tolerance, volatility, phase change characteristics, conductivity, flash point, and cetane number were studied. CETANER was shown to be completely miscible with diesel fuel to at least 40 vol %. Cetane numbers of diesel fuel blends generally increased with increasing CETANER concentration. Based on the flash point requirements, the maximum

114

monoglyme concentration in a U.S. fuel blend is probably 4 vol %. Diesel fuel CETANER blends were found to be stable upon storage and free of hydroperoxides after one year. The glyme components of CETANER have relatively low toxicity; however, some evidence suggests that they may have teratogenic properties. 14) Development of Kinetic Models for the Liquid Phase Methanol (LPMEOH) Process Air Products The major part of the report concerns the development of more robust kinetic models for the LPMEOH reaction system. The development was needed to meet the requirements for more accurate process simulations over a wide range of conditions. To this end, kinetic experiments were designed based on commercial needs and a D-Optimal design package. A database covering 53 different conditions was built. Two new reactions were identified and added to the LPMEOH reaction network. New rate models were developed for all 15 reactions in the system. The new rate models are more robust than the original ones, showing better fit to the experimental results over a wide range of conditions. Related to this model development are some new understandings about the sensitivity of rate models and their effects on catalyst life study. The last section of this report covers a separate topic: water injection to the LPMEOH reactor and its effects on the LPMEOH Process. An investigation was made of whether water injection can enhance the reactor productivity and how this enhancement depends on the composition of the major syngas feed. A water injection condition that resulted in 32% enhancement in productivity was observed. A catalyst life test under this water injection condition was conducted and showed no negative effects of water injection on catalyst stability. 15) Kinetic Understanding of the Syngas to DME Reaction System and Its Implications to Process and Economics Air Products In a single-step synthesis gas-to-dimethyl ether process, synthesis gas (or syngas) is converted into DME in a single reactor. The three reactions involved in this process, methanol synthesis, methanol dehydration and water gas shift, form an interesting reaction network. The interplay among these three reactions results in excellent syngas conversion or reactor productivity. A fundamental understanding of this interplay helps to explain many experimental and simulation observations, to identify optimal reaction conditions, and to provide guidelines for process development. The higher syngas conversion or reactor productivity in the syngas-to-DME reaction system, compared to that in the syngas-to-methanol reaction system, is referred to as chemical synergy. This synergy exhibits a strong dependence on the composition of the reactor feed. To demonstrate the extent of this dependence, simulations with adjusted activity for each reaction were performed to reveal the relative rate of each reaction. The results show that the water gas shift reaction is the most rapid, being practically controlled by the equilibrium. Both methanol synthesis and methanol dehydration reactions are kinetically controlled. The kinetics of the dehydration reactions is greater than that of the methanol synthesis reaction in the CO-rich regime. However, the rates of these two reactions come closer as the H2 concentration in the reactor feed increases.

115

The role of the dehydration reaction is to remove the equilibrium barrier for the methanol synthesis reaction. The role of the water gas shift reaction is more complex; it helps the kinetics of methanol dehydration by keeping the water concentration low, which in turn enhances methanol synthesis. It also readjusts the H2:CO ratio in the reactor as the reactions proceed. In the CO-rich regime, the water gas shift reaction supplements the limiting reactant, H2, by reacting water with CO. This enhances both the kinetics and thermodynamic driving force of the methanol synthesis reaction. In the H2-rich regime, water gas shift consumes the limiting reactant, CO, which harms both the kinetics and thermodynamics of methanol synthesis. An understanding of these complex roles of the methanol dehydration and water gas shift reactions and of their dependence on the syngas composition explains why the synergy is high in the COrich regime, but decreases with increasing H2 or CO2 content in the reactor feed. The methanol equivalent productivity of the syngas-to-DME reactor is also a strong function of the reactor feed. A mathematical approach was developed to understand this dependence. The approach divides a power law type of rate equation into two terms, the kinetic term (the rate of the forward reaction) and the thermodynamics or driving force term (1- approach to equilibrium). The equations for the best feed composition for each term were derived. The approach was developed for the single reaction system, and then extended to the syngas-to-DME reaction system. The equations provide insights into why and how the methanol synthesis in the syngas-to-DME system depends on the other two reactions. They can also be used to calculate the best feed composition for a given conversion. The analysis shows that for typical commercial syngas conversion, the optimal H2:CO ratio for the LPDME reactor is around 1to-1, in good agreement with the results from the simulation. While the 1-to-1 feed provides a good foundation for some process configurations, it does not match the composition of natural gas-derived syngas, which typically has a H2:CO ratio of 2:1 or greater. The process would also produce one CO2 molecule for every DME product, both a materials utilization and an environmental problem. However, recycling CO2 to the syngas generation unit can solve all of these problems. Integration schemes with different syngas generation technologies (dry reforming, steam methane reforming and partial oxidation) were developed. The feasibility of these schemes was illustrated by simulations using realistic kinetics, thermodynamics, and commercial conditions. Finally, this report discusses the implications of the kinetic understanding and the resulting process schemes to the process economics. It was recognized that, for the overall process, the cost saving in the synthesis loop due to the reaction synergy is counteracted by the cost addition due to CO2 formation and the resulting costly separation. This counteraction occurs in the entire H2:CO range of commercial interest. The curves that showed the enhancement in productivity and CO2 formation as a function of H2:CO ratio in the reactor feed were used to discuss qualitatively how the economics of the syngas-to-DME process (1-step) compares to the syngasto-methanol-plus-dehydration process (2-step). While the 1-step process has clear advantages over the 2-step process for CO-rich syngas derived from coal and other carbonaceous solids and liquids, its advantage for the natural gas-derived syngas is not so clear. Process optimization appears to be an important factor in tilting the balance.

116

APPENDIX LaPorte Alternative Fuels Development Unit (AFDU) Process Descriptions for Operating Campaigns (1995-1999) 1. INTRODUCTION The success of the liquid-phase program has centered on the unique U.S. Department of Energy (DOE)-owned Alternative Fuels Development Unit (AFDU) located in LaPorte, Texas, which is operated by Air Products. A photograph of the AFDU is provided in Figure A-1. The LaPorte AFDU has unique capabilities to provide the proper springboard for commercialization of new technology. The facility is capable of processing synthesis gas (syngas) of widely varying compositions in slurry bubble column reactors -- the heart of liquid-phase technology -- to demonstrate production at an industrially relevant engineering scale of 5-15 tons per day (TPD). The AFDU is adjacent to a commercial plant that produces both hydrogen (H2) and carbon monoxide (CO) to provide the flexibility to simulate syngas from any alternative energy source.

Figure A-1 LaPorte Alternative Fuels Development Unit (AFDU)

A1

2. PROCESS DESCRIPTION 1995 METHANOL/FLUID DYNAMIC RUN The process flow diagrams for this operating campaign (described in Section 3.A.2 of this report) are shown on Pages A-10 and A-11. The operation of the plant is described as follows (refer to Page A-10): Hydrogen (H2), carbon monoxide (CO), carbon dioxide (CO2), and nitrogen (N2) are blended and compressed in the 01.10 feed gas compressor. This stream then mixes with recycle gas and additional H2 from a high-pressure pipeline to obtain the desired syngas composition and flow. (The 01.30 booster compressor was bypassed, as higher pressure operation was not desired.) The mixed feed then passes through the 01.34 aftercooler used to control the inlet temperature to the 21.11 feed/product economizer, which preheats the feed against the reactor effluent. (The 10.95 high-pressure liquid injection pump was not used during this run.) The mixed feed can then be further preheated (if necessary) against high-pressure steam in the 02.63 high-pressure feed steam preheater before the syngas blend is introduced to the bottom of the 27.20 highpressure slurry reactor. The syngas flows upward through the slurry of catalyst and mineral oil as the reaction proceeds. The heat of reaction is absorbed by the slurry and removed through the internal heat exchanger, which also uses mineral oil as its heat transfer fluid. The product gas passes through the reactor freeboard with the unconverted syngas, and the gross reactor effluent cools against the feed in the 21.11 economizer. Any traces of slurry oil entrained or vaporized in the effluent condense and are returned to the bottom of the reactor by the 10.52.02 pumps. The vapor leaving the 21.11 de-pressurizes across a valve to less than 1,000 psig; chills against cooling water in the 21.30 hairpin exchangers; and passes into the 22.10 separator where liquid products (methanol, water, higher alcohols) collect. The liquids flash to near atmospheric pressure in the 22.11 degasser and collect in the 22.15 low-pressure separator before passing on to the 22.16 day tank and eventually a trailer for storage. To minimize the amount of gas sent to the flare, most of the syngas leaving the 22.10 separator is recycled to the reactor. A small portion of this gas is purged to flare to prevent the buildup of inerts. Bubble Column Reactor The 27.20 bubble column reactor for oxygenate synthesis measures 50 ft flange-to-flange and 18 in. inside diameter. Its design slurry level is 40 ft, with the remainder being vapor disengagement space. The reactor contains an internal heat exchanger consisting of twelve -in. U-tubes occupying 8% of the reactor cross section. In addition, 13 thermocouples measure the longitudinal temperature profile at 4-ft intervals. A nuclear density gauge, mounted on an external hoist mechanism, spans the space occupied by the internal exchanger to measure slurry level and gas holdup. The design pressure of the reactor is 2,000 psig at 700F.

A2

3. ENGINEERING AND MODIFICATIONS FISCHER-TROPSCH III RUN (1996) Process Description Simplified process flow diagrams for the Fischer-Tropsch III campaign (described in Section 3.A.3 of this report) are shown on Pages A-12 and A-13. The operation of the plant is described as follows (refer to Page A-12): CO, H2, and N2 are blended and compressed using the 01.10 compressor to obtain the desired fresh syngas composition and flow. The fresh feed is then mixed with recycle feed from the 01.20 recycle compressor. High-pressure H2 is used to supplement the fresh feed. The highpressure H2 may be compressed using the recycle compressor if its pressure is not adequate. The combined feed gas is preheated in the 21.38 feed/product economizer and the 02.61 feed gas steam heater. The preheated feed gas is introduced to the bottom of the slurry reactor, 27.10. The syngas flows upward through the slurry and is partially converted to hydrocarbons, water and CO2. The heat of reaction is absorbed by the slurry medium and then rejected to an internal heat exchanger. The heavier hydrocarbon fraction of the product (heavy wax) is liquid at reaction conditions and accumulates in the reactor. The reactor effluent is first sent through the 27.11 cyclone separator to remove entrained slurry and then cooled using the 21.38 economizer to condense light waxes which are separated in the 22.14 separator. The reactor effluent is subsequently chilled against cooling water in the 21.65 hairpin exchangers. Condensed hydrocarbons and water are separated from the vapor phase in the 22.10 separator. After analysis, part of the uncondensed vapor is sent to the flare as a purge stream; most of this stream is recycled using the 01.20 recycle compressor. The liquids from the 22.10 are de-pressurized and sent to vessels 22.11, 22.15 and 22.16 in sequence. The product from the 22.16 is sent to a tank trailer in batches, periodically. The excess slurry from the reactor is drained into the 27.15 slurry degasser. Following degassing, the slurry is cooled by about 36F in the 21.70 catalyst-wax slurry cooler. The slurry is then pumped using the 10.62 catalyst-wax circulation pump to the 22.62 cross-flow filter for wax removal. The product wax from the filter is collected in the 28.30 prep tank and then drained into trailer or drums and sampled periodically. After filtration, the concentrated slurry is sent back to the reactor. The liquid level in the reactor is measured by the nuclear density gauge DIC-585 and is controlled by position of the control valve, which directly controls wax withdrawal from the filters. In the reactor, particles are kept fully suspended by the upward liquid flow (also in the absence of gas flow), as the liquid velocity is well beyond the particle settling velocity. A liquid level is maintained in the 27.12 slurry carryover surge tank, which receives slurry from the 27.11 cyclone separator. Excess liquid from the 27.12 is sent to the 27.10 reactor using the 10.52.02 carryover oil pump. The light wax from the 22.14 separator is de-pressurized into the 27.13 tank. Light wax in the 27.13 tank is circulated using the 10.60 pump and kept warm by flowing it through the 21.85 heat exchanger. Heavy wax in the 28.30 prep tank is circulated by

A3

the 10.52.01 pump. The pressurized wax is used to back-flush the 22.62 filters when needed. Waxes from the 27.13 and the 28.30 are drained into trailer or drums. Flows and compositions, including feed and product gas, are measured at various strategic points in the process. Bubble Column Reactor The 27.10 bubble column reactor is 28.3 ft top to bottom and 22.5 in. inside diameter. The maximum slurry level is about 20 ft, with the remainder being vapor disengagement space. A productivity goal of 150 grams hydrocarbons /liter reactor vol - hr, which was the same as that for F-T II, was set. The heat exchanger, which was installed prior to F-T II (see Reference), was evaluated for F-T III and found to be adequate. The heat exchanger consists of 22 vertical -in. U-tubes with an internal header. Twelve of the U-tubes are near the wall and ten are near the center. Detailed drawings of the heat exchanger were included in the Fischer-Tropsch II topical report (Bhatt, 1995). The external surface area of the U-tubes is 217.7 ft2 based on a 36-ft length. The heat exchanger occupies 9.6% of the reactor cross section. The reactor is fitted with a number of thermocouples, located at various elevations. A nuclear density gauge is mounted on an external track and spans the space occupied by the internal heat exchanger. The maximum temperature for the reactor is 315C at the maximum pressure of 1,000 psig. Operations at higher temperature are feasible by lowering the operating pressure. Catalyst-Wax Separation System Catalyst-wax separation has been recognized as a challenge. No single proven technology exists in the public domain. An external system of tangential (cross) flow filters was used at LaPorte based on Shell Synthetic Fuels, Inc.s (SSFIs) pilot plant experience. Filtration was preferred at reactor pressure to avoid catalyst attrition that may occur if a control valve is used to reduce the pressure. The existing filtration system at LaPorte was designed for low pressure, with limited capacity. Therefore, the entire filtration system was redesigned and replaced. The new system was rated at higher pressure (1,000 psig) and higher temperature (600F), with significantly higher capacity. It included four new cross-flow filters in series, a catalyst-wax slurry circulation pump, a slurry cooler and a slurry degasser. A layout of the filtration system is shown on Page A-14. The sketch is not to scale, but does show an approximate elevation of the equipment. The degasser was installed close to the top of the reactor to obtain almost the same liquid level in the two vessels. The slurry cooler was located at a level near the bottom of the reactor. The pump and the filters were installed at the ground level. There was no backup system for filtration. If the filtration did not work as designed, the reactor would have to be shut down. An extra charge of catalyst was available on site for another start-up if the problem could be identified and addressed. 27.15 Slurry Degasser The degasser was used to separate gas and solid-liquid from the three-phase reactor slurry. The degasser would protect the slurry pump from gas and minimize any further reaction in the loop. The degasser diameter was based on the liquid velocity being half the bubble rise velocity. A tubular vessel with 8-in. internal diameter and 8 ft height was specified. A sketch of the vessel is shown on Page A-15. The instrumentation on the vessel included three thermocouples and a differential pressure transmitter to measure the liquid level, which was maintained at the 5-ft

A4

level. A conical-shaped head was used for the bottom to avoid slurry accumulation. The layout of the degasser was very important. Liquid level in the degasser was maintained at the same level as the reactor through nozzle T1 and a 3-in. pipe. A new nozzle was installed and used to return the gas to the reactor. In case of gas shutdown, valve NV-1752-S on nozzle T1 would close and valve NV-1751-S on nozzle P would open to provide continuous liquid flow to the pump. The bypass line with two shut-off valves (NV-1756-S and 3553-S) was provided for start-up. If the degasser did not degas adequately, the pump and filter would see gas in the slurry, which could affect their performance. 21.70 Slurry Cooler Slurry cooling of 36F was desired to significantly lower the reaction rate. Utility oil exiting from the reactor heat exchanger was used on the shell side to cool the slurry (Page A-13). Use of cooler oil going into the reactor heat exchanger was avoided so that the reactor temperature stability would not be impacted. HTRI (Heat Transfer Research Institute) heat exchanger simulations were performed for the slurry cooler. Oil temperatures based on realistic calculations (2.14 MMBtu/hr heat load) for the reactor heat exchanger were specified. A conservative heat load (2.50 MMBtu/hr) on the reactor heat exchanger would lead to an underdesign of this heat exchanger. If the reactor heat exchanger worked much better than designed, the oil would be hotter than expected. In that case, the cooler would not have adequate capacity. The cooler was not substantially over-designed to keep slurry inventory/residence time in the filtration system to a minimum. A horizontal, multi-pass flow heat exchanger with segmental baffles on the shell side was specified. It contained an 8-in. internal-diameter, single-pass shell with 6 cross passes and 10 tubes with 3 passes, 11.5 ft long and 0.527 8 in. in internal diameter. 10.62 Slurry Pump Due to an anticipated long delivery time for the slurry pump, a process specification for the pump was quickly issued. A centrifugal pump was chosen based on SSFI's experience. The centrifugal pump is more reliable in operation compared to a disc type pump considered, which has lower catalyst attrition rate. SSFI's testing showed an acceptable level of catalyst attrition. The high-temperature, high-pressure service was considered severe. There was no backup for this pump. In case of mechanical problems, the reactor would have to be shut down and the pump repaired on site. A design flow rate of 26 gpm was specified with a head of 107.5 psi. A variable-speed motor was utilized to allow flow rate changes during the run. 22.62 Cross-flow Filters The cross-flow filter system consisted of four 10-ft filters (four parallel elements in each) in series. The elements were -in. inside diameter, 5/8-in. outside diameter, 1-micron-grade stainless steel. A tangential velocity of 9 ft/sec would be maintained through the elements. The elements would be back-flushed with clear wax, as needed. The filters were designed for 62 gph of filtrate wax, which was the anticipated maximum production rate. The design flux through the elements was 0.059 gpm/ft2, which was a conservative number based on SSFI's experimental work. The total filtration area of the system was 20.9 ft2, about 20% higher than required. The filtration performance depended on the catalyst strength. Laboratory testing at SSFI indicated that the catalyst was much harder than those used in F-T I and II. The F-T III catalyst was a supported catalyst, which is typically more resistant to attrition. Differential pressure measurements capabilities were installed for individual filter housings. This was necessary to control pressure drop across each filter separately, as significant pressure drop was expected for the 40 ft of total length due to the slurry flow through the inside of the elements. The transmembrane pressure drop would be controlled by throttling manual valves. A5

Miscellaneous Modifications Preliminary heat and mass balance calculations indicated that at high conversion (~80%) and high pressure (750 psig), the dew point for water (~235C) was very close to the operating temperature (240-250C). The heat exchanger tubes were expected to be significantly colder (170C), at which point water could condense out. Water condensation would be an obvious problem for catalyst activity. Hence, it was decided to reduce per-pass conversion to 40% with a recycle of unconverted syngas. The Fischer-Tropsch train was connected to the existing 01.20 recycle compressor to allow the recycle. It was decided not to modify the system for light hydrocarbon/water separation due to funding limitations. Miscellaneous changes include: (1) (2) (3) (4) (5) Differential Pressure (DP) taps and transmitters on the 27.10 reactor, Radial thermocouples in the reactor, Erosion test pieces, Removal of once-through connections for the reduction of catalyst precursor. Relocation of water analysis sample port for catalyst activation from the reactor outlet to the 22.14 vapor-liquid separator outlet to avoid significant hydrocarbon condensation in the water analyzer, which could cause interference.

The reactor was modified to accommodate SSFIs proprietary sparger and optical fiber probe. The probe was installed to measure radial bubble size distribution. A number of signals were connected to SSFI's high-speed data acquisition system. New instrumentation needing specification included slurry/wax flow meters, differential pressure transmitters, and automatic shut-off valves. In addition, specifications were developed for two new relief valves (PSV-236A/B and PSV-1766) and a rupture disc (PSE-1769).

A6

4. ENGINEERING AND MODIFICATIONS FISCHER-TROPSCH IV RUN (1998) A simplified process flow diagram for the Fischer-Tropsch IV campaign (described in Section 3.A.3 of this report) is given on Page A-16. The process description from the F-T III Run (Section 3 of this Appendix) describes the operation of the majority of the AFDU for this campaign. Other improvements included purchase of significant hardware and software for the distributed control system to both fix a data acquisition problem and upgrade the system, and a cooling water pump to boost the cooling water pressure. In preparation for installing alternate filter elements, four filter housings from LaPorte were shipped to SSFI. 10.62 Slurry Pump The slurry pump modification involved new internals (diffuser and cover plate) made up of manganese alloy, opening the throat to 0.446 in., use of a differential pressure regulator to maintain proper differential pressure between process and buffer fluid, and installation of a shutoff valve on the buffer system. 22.62 Cross-Flow Filters The cross-flow filter system consisted of four 10-ft filters (four parallel elements in each) in series. The new filter elements were woven metal elements, 14-mm inside diameter, 10-ft long, 10-micron-grade stainless steel. A tangential velocity of 8.7 ft/sec, corresponding to 26 gpm of pump flow, would be maintained through the elements. The elements would be back-flushed with clear wax, as needed. For 61 gph of filtrate wax production rate, the design flux through the elements required would be 0.044 gpm/ft2. The total filtration area of the system was 23.1 ft2. With expectation of lower wax production associated with the new catalyst and higher filter capacity as measured by SSFI, it appeared that there was a 100% spare capacity. Therefore, it was decided to put only two housings on-line initially, and withdraw wax from only one of them. For additional backup, 16 extra elements were purchased and constructed into 4 additional bundles. The new bundle arrangement would allow easier on-site replacements of bundles, if the four bundles in service became plugged. Four control valves were installed on the product wax line to improve filtration control by achieving an individual control of each housing. These valves replaced the existing manual throttle valves. The single existing larger control valve (LV203) was also removed. Heat Transfer Fluid An alternate oil was evaluated for the utility oil system to improve heat transfer. Better heat transfer in the reactor tubes translates into improved reactor temperature control. Therminol-59, a heat transfer fluid used in the Kingsport Liquid Phase Methanol (LPMEOH) Commercial Demonstration Units catalyst reduction vessel, was found to be superior to the current Drakeol10. The log mean temperature difference (LMTD) was reduced by about ~15%. The improvement was due to better heat transfer properties: higher thermal conductivity, lower viscosity and higher density. The only negative effect was due to lower heat capacity.

A7

5. PROCESS DESCRIPTION 1999 LPDME RUN The process flow diagrams for the 1999 Liquid Phase Dimethyl Ether (LPDME) campaign (described in Section 3.A.4 of this report) are shown on Pages A-17 and A-18 (Page A-13 shows the configuration of the utility oil system). The operation of the plant is described as follows (refer to Page A-17): H2, CO, and N2 are blended and compressed in the 01.10 feed gas compressor to about 800 psig. This stream is then mixed with recycle gas and additional H2 from a high-pressure pipeline to obtain the desired syngas composition and flow. (The 01.30 booster compressor was bypassed, as higher pressure operation was not desired.) The mixed feed then passes through the 01.34 aftercooler used to control the inlet temperature to the 21.11 feed/product economizer, which preheats the feed against the reactor effluent. (The 10.95 high-pressure liquid injection pump was not used during this run.) The mixed feed is further preheated against high-pressure steam in the 02.63 high-pressure feed steam preheater before the syngas blend is introduced to the bottom of the 27.20 high-pressure slurry reactor. The syngas flows upward through the slurry of catalyst and mineral oil as the reaction proceeds. The heat of reaction is absorbed by the slurry and removed through the internal heat exchanger, which uses a heat transfer fluid. The product gas passes through the reactor freeboard with the unconverted syngas, and the gross reactor effluent cools against the feed in the 21.11 economizer. Any traces of slurry oil entrained or vaporized in the effluent condense and are returned to the bottom of the reactor by the 10.52.02 pumps. The vapor leaving the 21.11 economizer flows through the 27.14 vapor-liquid separator to remove any leftover condensed oil; chills against cooling water in the 21.30 hairpin exchangers (Page A-18); and passes into the 22.10 separator where liquid products (methanol, water, higher alcohols, some DME) collect. The liquids flash to near atmospheric pressure in the 22.11 degasser and collect in the 22.15 lowpressure separator before passing on to the 22.16 day tank and eventually to a trailer for storage. To minimize the amount of gas sent to the flare, most of the syngas leaving the 22.10 separator is recycled to the reactor. (Since CO2 is a by-product of DME synthesis with CO-rich syngas, it was necessary to remove CO2 from the 22.10 separator overheads before recycling this stream.) The closed-loop CO2-removal system uses methanol to preferentially absorb the CO2 and DME from the syngas. The vapor from the 22.10 separator cools against returning CO2-lean syngas in the 21.10 gas-gas economizer. It then feeds into the bottom of the 07.10 absorber and contacts against chilled methanol introduced at the top of the column. The CO2-lean syngas leaves the top of the absorber and rewarms to ambient temperatures in the 21.10 gas-gas economizer before being recompressed in the 01.20 recycle compressor. A small portion of this gas is purged to flare to prevent the buildup of inerts. The CO2-rich liquid collects in the bottom of the 07.10 absorber, de-pressurizes across a valve, and heats up against returning methanol in the 21.45 hairpin exchangers. This liquid then passes into the top of the 07.20 stripper, where it is reboiled to remove the dissolved gases such as CO2 and DME. The overhead cooling water condenser reduces the amount of methanol solvent lost in the overhead stream, which goes to flare. The liquid from the bottom of the 07.20 stripper cools in the 21.45 hairpin exchangers prior to recompression in the 10.80 pump. The methanol then chills against liquid CO2 in the 21.80 kettle evaporator before recycling to the top of the 07.10 absorber. The syngas from the 22.10 separator includes equilibrium amounts of methanol, water, and other hydrocarbons which will build up in the methanol solvent. Methanol is also lost in the A8

07.20 stripper overheads. As a result, the CO2-removal system operates in an unsteady state as the composition of the solvent changes. Since this change affects the level of CO2 removal, the system includes a solvent purge and fresh methanol makeup lines. Bubble-Column Reactor The high-pressure 27.20 bubble-column reactor for oxygenate synthesis measures 50 ft flangeto-flange and 18 in. inside diameter. Its design slurry level is 40 ft from the bottom flange, with the remainder being vapor disengagement space. The reactor contains an internal heat exchanger consisting of twelve -in. U- tubes occupying 8% of the reactor cross section. In addition, 13 thermocouples measure the longitudinal temperature profile at 4-ft intervals. A nuclear density gauge (NDG), mounted on an external hoist mechanism, spans the space occupied by the internal exchanger to measure slurry level and density. Six differential pressure (DP) transmitters have been installed on the reactor wall for more accurate slurry density measurements. Detectors have been temporarily set up at various locations outside the reactor for a radioactive tracer injection study. The design pressure of the reactor is 2,000 psig at 700F. 6. REFERENCE Bhatt, B. L., "Liquid Phase Fischer-Tropsch (II) Demonstration in the LaPorte Alternative Fuels Development Unit," Topical Report prepared by Air Products and Chemicals for U. S. DOE, Contract No. DE-AC22-91PC90018, September 1995.

A9

A-10

A-11

A-12

A-13

A-14

A-15

A-16

A-17

A-18