Comparison of Aqueous Phase Indices for Powdered Activated Carbon to Pore Size Distribution Measured via Gas Adsorption

The adsorptive capacity of powdered activated carbon (PAC) is related to the size of the target adsorbate compared with the quantity of pores with sufficient pore diameter into which the adsorbate may diffuse (i.e., the effective internal surface area). The adsorptive capacity of PACs are commonly characterized using aqueous-phase indices such as iodine (I2), methylene blue (MB), p-nitrophenol (PNP), molasses, and tannin. PACs may also be characterized using gas adsorption with nitrogen to develop detailed pore size distribution information. This paper describes the characterization of the internal porous nature of 12 PACs used in water treatment using both aqueousand gaseous-phase methods. The results showed that the internal surface areas of PACs tended to be ordered: wood>bituminous coal>lignite coal. The results showed that the bituminous and lignite carbons had a majority of their internal pores in the microporous range (<20Å) whereas wood-based carbons tended to have a much wider pore size distribution well into the mesoporous range. The results showed that the molasses and tannin numbers correlated well with the total pore volume of pores greater than about 11Å. The MB and I2 numbers correlated well with total surface area if the micropores below 11Å also were included. INTRODUCTION Powdered activated carbon (PAC) is often used in water treatment plants to remove a wide variety of organic chemicals of varied molecular weights and sizes such as synthetic organic chemicals (SOCs), taste and odor compounds such as methylisoborneol (MIB) and geosmin, and natural organic matter such as humic acid and fulvic acid (Lee et al., 1981). PACs are manufactured from source materials including bituminous coal, lignite coal, wood, coconut shells, peat, petroleum residue, bones, fruit seeds, and other sources (Greg and Sing, 1982). Due to the use of various source materials, and the common use of activation methods specific to each material, activated carbons may have different internal pore structures consisting of micropores (<20Å), mesopores (20-500Å), and macropores (>500 Å). Depending on the size of a target adsorbate molecule relative to the smallest pores, only a portion of a PAC’s total surface area or pore volume may be available for adsorption. PACs may be characterized experimentally using gas-phase and aqueous-phase methods. Gas-phase methods are particularly useful in characterizing the pore size distribution of a PAC as a function of surface area or pore volume because nitrogen molecules can move into very small pores, and are relatively inert with respect to interaction with the chemical functional groups on the surface of the carbon. More commonly used methods of PAC characterization, however, are the standardized adsorption tests with p-nitrophenol (PNP), iodine (I2), methylene blue (MB), molasses, and tannin as sorbate surrogates (Hassler, 1951; Galbraith et al., 1958; Kasaoka et al., 1989b). There is a fundamental question, however, regarding what pore sizes are being characterized by each standardized aqueous index. In this study, we examined 12 PACs commonly used in the water treatment industry using gas-phase and common aqueous-phase techniques. These data were then compared statistically to assess the correlations between each parameter. This information is useful for understanding the utility of commonly used indices, and may provide the basis for improving the specification and selection approach used by water utilities to choose a PAC for specific water treatment objective. Gas Adsorption Characterization Methods Gas adsorption is a standard method used to characterize the internal surface area and structure of porous materials. Different adsorbate gases that have been used to characterize adsorbents include nitrogen, carbon dioxide, krypton and argon (Gregg and Sing, 1982; Webb and Orr, 1997). Nitrogen is the most common adsorbate gas used to characterize surfaces because it is relatively inert and interacts with the adsorbent surface primarily through a mechanism of physical adsorption at liquid nitrogen temperatures (77 K). The data from the gas adsorption measurements can be used to determine the surface area using the BET method, the mesopore volume using the BJH method, and the micropore volume using the HK method. The density functional theory (DFT) method is used to develop pore size distributions based on statistical thermodynamics and is one of the newer models used to analyze gas adsorption data (Ravikovitch et al., 1998). When a gas is introduced into a pore space, the gas molecules experience an attractive force (van der Waal’s forces) by the adsorbent’s surface, which results in a higher average density of gas near the surface during equilibrium. To calculate the pore size distribution, it is necessary to develop a model for pore filling which relates the pore width to the condensation pressure. Most theories related to isotherms tend to take a relatively macroscopic approach and may not be able to accurately predict the isotherms across their entire region. The DFT method takes a more microscopic approach and is based on statistical thermodynamics, taking the fluid-fluid and fluid-solid interactions into account as well as pore size, geometry and temperature.. Several researchers have used this model in their analyses for pore size distributions of microporous materials (Lastoskie et al., 1993; Ravikovitch et al. 1998; Krupa and Cannon, 1996). Aqueous Indices Another way to characterize the pore structure of activated carbon is the adsorption of molecule out of aqueous solution (Galbraith, et al., 1958, Chern and Huang, 1998, Kasaoka et al,. 1989a, Kasaoka et al., 1989b, Krupa and Cannon, 1996, Linares-Solano et al., 1984, Pelekani and Snoeyink, 2000, Pelekani and Snoeyink, 2001). These methods of characterization often use PNP, I2, MB, molasses, and tannin. These methods are especially appropriate to characterize PACs because they specifically involve the aqueous-phase adsorption of sorbates from solution. The specific methods are described below. EXPERIMENTAL Materials In this study, the adsorption of PNP, I2, MB, molasses and tannin from their aqueous solutions on twelve different PACs was determined. The PACs were from source materials including bituminous coal, lignite coal and wood. Two different batches of Hydrodarco B were included in this research to study the variability of activated carbon as provided by the manufacturer. These replicate batches produced excellent reproducibility. The twelve activated carbons studied are shown in Table 1. Table 2 contains information about each liquid phase adsorbate. In the table the critical dimension (i.e., second largest dimension) are those reported in the literature (Linares-Solano et al., 1984) or calculated using the MacSpartan 3-D modeling software (Ver. 1.1, Wavefunction Inc., Irvine, CA, USA). Table 1. Powdered activated carbons (PACs) examined in study. Name of PAC Symbol Manufacturer Carbon source Reported I2 Number (mg/g) Reported Molasses Number Reported Tannin Number (mg of PAC/L) WPL Pulverized WPL Calgon 1 Bituminous coal WPH Pulverized WPH Calgon Bituminous coal PWA Pulverized PWA Calgon Bituminous coal 900 min 210 min Colorsorb GP COL Calgon Wood 900 min 550 Centaur Powder CEN Calgon Bituminous coal 800 min PAC 20 B 20B NORIT Bituminous coal 900 200 360-400 HDW HDW NORIT Lignite coal 530 350 Hydrodarco B (Batchs 1091 and 135-10C) HD-A NORIT 2 Lignite coal 560 450-650 200 Watercarb PAC WAT Acticarb Wood Bark 600 Nuchar SA-20 N-SA Westvaco Hard wood 900 min 225 Nuchar SA N-20 Westvaco Hard wood 900 min 225 1 Calgon Carbon Corp., Pittsburgh, PA 2 NORIT Americas Inc., Atlanta, GA 3 Acticarb, Inc., Dunnellon, FL 4 Westvaco Chemical Division, Westvaco Corp., Covington, VA Table 2. Data regarding the molecular weight and size for the adsorbates. Compounds M.W. Dimensions Critical Dimension (Å) Average Critical Pore Dimension (Å) †† Nitrogen 28 3† p-Nitrophenol 139 7.6 x 5 x 1.9* 4.6† 8 Methylene Blue 320 16 x 8.4 x 4.7 8.4 15.2 Iodine 254 7 x 4.4 x 4.4 4.4 < 14.8 Tannic Acid 1701 26.8 x 24.8 x 13.2* 24.8 Molasses ** Mixture Sucrose 35% 342 11.8 x 7.1 x 6.4* 7.1 Glucose 7 % 156 8.9 x 5.4 x 4.5* 5.4 Fructose 9 % 156 9.8 x 4.1 x 3.3* 4.1 Water 20 % 18 2.3 x 1.6 x 1.2* 1.6 Ash (K2O, SiO2, etc) 12 % ~64-94 Ex: SiO2 4.3 x 2.6 x 2.6* 2.6 * The dimensions of molecule and the critical dimension calculated using MacSpartan ** The general composition molasses, which is a mixture of several compounds, is given. † From (Linares-Solano et al, 1984) †† From (Kasaoka et al, 1989b) Gas Adsorption In the nitrogen adsorption tests, the samples were analyzed using static volumetric technique for nitrogen adsorption with the Quantachrome Autosorb-1C gas sorption apparatus. In the method, PAC samples are dosed with nitrogen at 77K to attain pressure equilibrium points during adsorption and desorption cycles to generate isotherms. The adsorption isotherm data are analyzed by a variety of techniques. Liquid Phase Adsorption The PNP capacity was determined for all the PAC samples using the procedure based on Hassler (1963), Linares-Solano et al. (1984), and Krupa and Cannon (1996). In this method, the PNP capacity is determined for a specific residual PNP concentration (100 mg/L) from isotherm data. The iodine number was determined for all the PAC samples using the standard method (ASTM D4607-94, 1994) with five different masses of activated carbon instead of the three required by the standard method. The I2 number is determined as the amount of I2 adsorbed per mass of PAC at a residual I2 concentration of 0.02 N from a log-log plot (0.95bituminous carbon>lignite carbon. Micropore analyses The HK volumes represent the microporous nature of activated carbons. Comparison of the total pore volume (mesopores plus micropores) with the HK pore volume indicates the relative amount of micropores. It was found that 43-44 percent of the pore volume is as micropores for the woodand lignite-coal carbons. Based on HK pore volume for the bituminous-coal carbon, however, the micropores account for 75 percent of the combined mesopore and micropore pore volume. Aqueous indices Activated carbon capacities for small molecules are frequently characterized by use of the PNP, I2 and MB methods, while capacities for large molecules are often characterized by the molasses and tannin methods (AWWA, 1996b). For the PNP sorption, it was found that bituminous carbon and wood-base carbon had significantly greater capacities than lignite carbons (α=0.05). For the I2 number results, bituminous carbon and wood-based carbons also had significantly greater capacities than for lignite coal. The WAT bark-based carbon had a similar I2 number to the lignite coal carbon. The PNP and I2 results agree with the gas adsorption results in that the bituminous and wood-based carbons tend to have much greater microporous structure than the lignite carbons. MB capacities determined using the isotherm methods versus the Hassler method correlated very well (r=0.992). Thus, these data provide justification for the use of the muchquicker and easier Hassler method versus the isotherm method. While the average MB values were ordered wood>bituminous coal>lignite coal, the differences were not significant (α=0.05). The molasses and tannin number results showed that wood-based carbons (excluding bark) exhibit a greater capacity for these compounds than bituminous or lignite carbons. This result is consistent with the fact that the wood-based carbons tend to have greater mesoporous nature than coal-based carbons. DFT Analysis The DFT surface area is computed for pores from 4 to 59Å, that is, on the order of the size of smaller molecules. The DFT surface area correlations with molasses and tannin are weaker than correlations with smaller molecule indices (PNP, I2, and MB) even though the DFT characterizes the lower range of the mesopore sizes. The correlation coefficient with DFT is weaker for the larger molecules because the DFT does not account for pores greater than 59 Å. It should be noted that the DFT method can be expanded to much larger pore diameters than it was applied in this study. Figure 1. DFT pore size distribution based on surface area for 12 PACs grouped by: (a) bituminous carbon, (b) lignite carbon, (c) wood-based carbon, and (d) group average values The pore size distributions based on surface areas were determined for the 12 PACs using the DFT method (Figure 1). The cumulative surface area was calculated from these data and plotted in Figure 2. These data show an interesting anomaly with very few pores being indicated at both 15 and 22Å. Similar anomalies have also been noted by other researchers studying six GACs (Krupa and Cannon, 1996). Because it is unlikely that there are no or few pores at these two distinct diameters, it has been suggested that the anomalies are a function of the DFT model itself. Krupa and Cannon (1996) suggested that the cause may be related 1.0 10.0 100.