Grantee Research Project Results

Final Report: Using Membrane Reactors to Reduce the Generation of Waste Products

EPA Grant Number: R822343
Title: Using Membrane Reactors to Reduce the Generation of Waste Products
Investigators: Lund, Carl R.F.
Institution: The State University of New York at Buffalo
EPA Project Officer: Aja, Hayley
Project Period: October 1, 1995 through September 1, 1997
Project Amount: $193,389
RFA: Exploratory Research - Engineering (1995) RFA Text |  Recipients Lists
Research Category: Safer Chemicals , Land and Waste Management

Objective:

Membrane reactors combine separation and reaction in a single process unit, and in so doing they offer some potentially unique opportunities for chemical processing. One such opportunity is to increase the yield of intermediate products in a consecutive reaction network, like that represented schematically in equations (1) and (2). In these equations, R denotes the primary reactant, C is a co-reactant, D is the desired intermediate product, B is a by-product, and U is an undesired final product.

(1) R + C D + B

(2) D + C U + B

When the undesired product U is a waste product, the use of a membrane reactor may represent an approach to reducing waste generation by selectively removing the desired product D from the reaction environment.

A modeling study conducted in the principal investigator's labs prior to this grant indicated that waste generation via these kinds of reactions could be reduced compared to a conventional plug flow reactor under certain conditions. That study used hypothetical chemicals such as in equations (1) and (2). Furthermore, the mathematical model, while quite reasonable, had not been validated experimentally. The purpose of the research reported here was to further explore this particular opportunity using a membrane reactor. The specific goals were to experimentally validate the model equations used in the previous simulations, to identify the primary limitations on the process, to explore alternative system configurations, and to extend testing to real chemical systems.

Summary/Accomplishments (Outputs/Outcomes):

A laboratory membrane reactor unit was constructed, in part to conduct the experimental component of this project. The information obtained from this recirculating batch membrane reactor system was the same as that obtained from a conventional concentric tube membrane reactor system. The major advantage of the new system was its ability to sample a wide range of reactor design parameters. It was shown that the mathematical description of this reactor system is fully equivalent to a conventional system. Experimental studies of the dehydrogenation of cyclohexane demonstrated that these mathematical descriptions did indeed provide very good quantitative agreement between experiment and simulation. On the basis of these results, the validity of the equations used in simulations prior to the present project was established.

The next phase of the project involved the identification of real chemical reactions in which membrane reactors might be used to improve the yield of intermediate products and consequently reduce waste generation. Models were used to determine which parameters of the system were most critical to successful improvement in intermediate product yield. When the reactant R enters the reaction side of the reactor, one of two processes can occur. It can react either to give the desired product D, or it can permeate through the membrane. This defines the first critical parameter for the system: the rate of permeation of reactant R and the rate of reaction of reactant R must be properly balanced. While this is a critical parameter, it is not a limiting parameter in terms of practical applications, as it can be adjusted through variations in the reactor geometry, specifically using the ratio of membrane surface area to total reactor volume.

Once the desired product D is formed, it too can either permeate through the membrane (desired event) or react giving the undesired product U. Thus, the second critical parameter is the permselectivity of the membrane for the desired product. This turns out to be the limiting constraint on the use of membrane reactors to enhance intermediate product yields in real applications. While other critical parameters usually can be adjusted through reactor design or operation, the permselectivity usually cannot be adjusted easily; one has to "take what is available." Available membrane materials can be divided into three groups: polymeric membranes, dense inorganic membranes, and porous inorganic membranes. Applications using polymer membranes were not examined in the present study, but initial considerations suggest that this may be a profitable area for future study. Dense inorganic membranes also typically have extremely high (or even perfect) permselectivity. In this case, membranes only exist for a few chemical species, so that while dense inorganic membranes possess the requisite permselectivity for use in membrane reactors to enhance intermediate product yield, there are no practical consecutive reactions that might use such reactors.

This leaves porous inorganic membranes as the remaining class to be considered. Here, the permselectivity is typically determined by the pore size relative to the species participating in the reaction. Smaller molecules permeate faster than larger ones, but the permselectivities typically differ by a factor of 10 or less. This is borderline for use in a membrane reactor to enhance intermediate product yield. While there are effectively no temperature limitations, the use of porous membranes requires that the desired product D be smaller than the reactant R. The hydrodechlorination of dichloroethane is one reaction system in which it is possible that a membrane reactor might have an impact. The intermediate product, HCl, can poison the catalyst for the primary reaction. If the HCl can be removed efficiently through a membrane, the catalyst may maintain activity for a longer period of time, or it may even display sustained activity. Prior to the present investigation, the kinetics of catalyst deactivation had not been studied in sufficient detail to permit a modeling analysis of the feasibility of this approach; hence, an experimental investigation was undertaken. In this work, a porous glass membrane was used; if Knudsen diffusion predominates, then the permselectivity for HCl can be expected to be ca.1.65. Thus, HCl can be removed selectively, but without kinetic data the extent to which catalyst lifetime would be affected is not clear. Therefore the kinetics were investigated first.

A number of runs were made under a variety of conditions. The rate of reaction was found to decrease over time. In a batch reactor this would be expected, since the amount of reactant is steadily decreasing. An analysis of the data, however, revealed that this was not the only reason for the decrease in reaction rate. In particular, when an attempt was made to fit a mass action rate expression to the data, it was found that the apparent rate coefficient decreased over time. The kinetic analysis was repeated with the assumption that chlorine adsorption is always equilibrated. It was then assumed that the apparent rate coefficient was diminished by a factor equal to the fraction of the sites that had been deactivated through adsorption of Cl. When this was done, it became possible to fit the data accurately with a rate coefficient that was invariant over time, as it should be. Additional experiments suggest that a part of this deactivation is not easily reversible, in that reduction of the catalyst at the end of one run and prior to the next run does not restore the catalyst's activity.

Experiments then were conducted using the membrane reactor. The ability to remove the HCl through the membrane was observed to extend the life of the catalyst. The HCl produced during the hydrodechlorination of the dichloroethane selectively permeates through the membrane, and thereby its concentration is reduced on the reaction side of the reactor. This, in turn, results in a lower rate of accumulation of chlorine on the catalyst surface and a longer catalyst life. At the same time, however, the system is additionally diluted by gas from the permeation side of the membrane. This dilution also contributes to the extended catalyst lifetime. The portion of the gain that can be attributed to selective product removal is only marginal, however, due to the small degree of permselectivity of the membrane. Simulation indicates that if the permselectivity can be increased, the benefits can be improved significantly.

In conclusion, a membrane reactor that affords some unique experimental advantages was designed and constructed. The reactor then was used to study the dehydrogenation of cyclohexane. Results from this study were described with very good quantitative accuracy by a simple mathematical model. The same model had previously been used to predict that membrane reactors could be used to reduce the generation of waste products during consecutive reactions. Further investigation of this application reveals that a critical requirement is that the membrane be highly permselective for the desired intermediate product. An experimental study of the hydrodechlorination of dichloroethane showed that positive results could be achieved with less highly selective membranes, but the greatest benefits will result if the membrane is more selective.

Conclusions:

Manuscripts describing this research are being prepared. Reprints will be provided as they become available.

Supplemental Keywords:

hydrodechlorination, chemical reactors, selectivity, membrane, Scientific Discipline, Sustainable Industry/Business, cleaner production/pollution prevention, Environmental Chemistry, Chemistry, New/Innovative technologies, Engineering, cleaner production, waste minimization, waste reduction, clean technologies, hydro dechlorination, membrane reactors, chemical reaction systems, environmental engineering, mathematical models, pollution prevention, source reduction, innovative technologies

Relevant Websites:

http://www.eng.buffalo.edu/~lund/Research/membrane.htm

Progress and Final Reports:

Original Abstract

1996