SUPERCRITICAL FLUID REGENERA TION OF EXHAUSTED GRANULAR ACTIVATED CARBON -POTENTIAL APPLICATION TO WATER PURIFICATION
Report No 923/1/00

EXECUTIVE SUMMARY

Activated carbon, a material with excellent adsorption capability, is used, among other applications, in water purification processes for the removal of organic substances. For economical reasons activated carbon is regenerated rather than discarded once the adsorbing surface is exhausted. Until now, surface recovery has been achieved mainly by thermal regeneration, with the disadvantages of high energy demand, micropore burn-out and carbon mass loss. Other possibilities include chemical and biological regeneration, but these methods are applicable in a few special cases only.

During recent years supercritical fluid extraction proved to be a potential alternative method for the regeneration of exhausted activated carbon. Promising experiments have been performed with samples of activated carbon exhausted under clinical laboratory conditions. In this study samples were investigated which had been exhausted under "real-world" conditions in water purification installations in southern Africa. The objectives were to attempt the regeneration of such samples with supercritical carbon dioxide (sc-CO2), to optimise the regeneration with regard to the adjustable parameters controlling the limiting processes and to gain insight into the mechanism of supercritical fluid regeneration.

The excellent adsorption properties of activated carbon stem from a large surface area of up to 1 500 m2/g. A surface area of this dimension is obtained in the manufacture of activated carbon by subjecting all conceivable forms of carbonaceous material (wood, coal, nut shell) to high temperatures and thus transforming it into a labyrinth-like pore structure. Especially organic substances are adsorbed onto this active surface either by van der Waals forces (physisorption) or by chemical bonds (chemisorption). Activated carbon is classified according to the type of raw material, the magnitude of the surface area, the size, shape, density and hardness of the particles and the nature of the pore structure.

The properties of gases under supercritical conditions (31C and 73 bar for CO2) are considered ideal to extract substances from a solid matrix as is required for the regeneration of exhausted activated carbon. These supercritical fluids exhibit densities similar to those of liquids (high solvent strengths) and diffusion coefficients similar to those of gases (excellent transport characteristics), enabling them to effectively dissolve and/or desorb contaminants from the carbon surface and to easily enter/exit even the smallest pores and carry away any released material. On restoring ambient conditions, the supercritical fluid is reconverted into a gas and all formerly dissolved/desorbed substances fallout instantly.

The samples investigated were obtained from water purification plants situated in Windhoek (Namibia), at Vereeniging (South Africa) and in Durban (South Africa). The selection was based, on the one hand, on the different raw water types and operational conditions at the respective installations in order to acquire a representing overview and, on the other hand, on the common type of GAG used at these installations in order to exclude type of carbon as a variable affecting the regeneration.

The regeneration runs on the spent activated carbon samples were performed with a bench-top supercritical fluid extractor featuring a microprocessor controlled system for sc-CO2 at temperatures of up to 150C and pressures ranging between 130 and 450 bar, and a temperature controlled variable flow restrictor for the reinstatement of ambient conditions and the collection of the extracted substances.

The extent of regeneration of all samples was monitored in terms of a titrimetrically determined iodine number based on milligram of iodine adsorbed per gram of activated carbon. Since the adsorbed amount of iodine depends on the available active carbon surface, it was possible to correlate surface recovery during regeneration with a corresponding increase in iodine number.

A statistical method based on surface response analysis was utilised to design a minimum of ten experimental runs for each sample by which the temperature and pressure dependence of the regeneration at a fixed flow rate (2 ml/min) and extraction time (10 to 90 minutes) could be determined. The resulting three-dimensional surface response graphs revealed that a pressure of 450 bar and a temperature of 150C are optimum parameter values for the supercritical fluid regeneration in all cases and that both desorption of substances from the carbon surface and dissolution of substances in the sc-CO2 play a role in the regeneration process. The species removed from the carbon surface were not analysed since an identification of all possible types of contaminants and their respective influence on the studied regeneration process were considered beyond the scope of this investigation.

The samples selected for investigation were all "real-world" samples, in contrast to clinical samples purposively exhausted for the majority of studies on the regeneration of activated carbon by other researchers.

These clinical samples are exhausted in a laboratory under ideal conditions with a limited number of contaminants in high concentration. The exhaustion features low-energy adsorption and capillary condensation, with the result that regeneration is fairly easy. For "real-world" samples the exhaustion comprises the adsorption of different contaminants in minute concentrations over an extended period of time, so that the subsequent desorption process is complicated by high-energy demand, possible biological growth and ageing of the adsorbed species. The differentiation between these two types of samples is important to fully understand the regeneration results for the samples investigated .

The samples obtained from Durban were activated carbon which had not been subjected to any kind of regeneration. These were exhausted over a period of 18 months while being utilised to purify weakly polluted surface water and were therefore strictly "real-world" samples. The sc-CO2 regeneration of these samples was disappointing (2-5% recovery of active surface only), probably as a result of ageing of the adsorbed species, the varying environmental conditions and possible microbial surface growth.

The samples obtained from Vereeniging were exhausted only for about 100 hours by purifying highly concentrated filter backwash water. These samples had not been regenerated before and closely resembled samples clinically exhausted in a laboratory .The sc-CO2 extraction performed at optimum parameter values resulted in an increase of up to 90% of the original active surface area, in agreement with results previously obtained for laboratory prepared samples.

The samples obtained from Windhoek had been previously regenerated thermally, as a result of which "dead" material had been burnt onto the active surface. The treatment with sc-CO2 resulted in a significant mass loss (over 20%) due to the successful removal of a double layer of "dead" material, but without a significant increase in iodine number as a result of the destroyed active surface and micropore structure. It could thus be concluded that a prerequisite for supercritical fluid regeneration is that the activated carbon should not have been regenerated thermally before.

From the results of this investigation the requirements for the supercritical fluid regeneration of activated carbon exhausted in "real-world" water purification systems could be derived. These include that samples should not have been subjected to thermal regeneration before, that samples should be free of impeding biological growth and aged adsorbed species and that optimum supercritical conditions (high temperature for effective desorption, high density for guaranteed dissolution) should be selected.