- Email: [email protected]
Dynamic pesticide removal with activated carbon fibers
PII: S0043-1354(00)00262-1
Wat. Res. Vol. 35, No. 2, pp. 516–520, 2001 # 2000 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/00/$ - see front matter
DYNAMIC PESTICIDE REMOVAL WITH ACTIVATED CARBON FIBERS I. MARTI´N-GULLO´N* and R. FONT Chemical Engineering Department, Universidad de Alicante, PO Box 99, 03080 Alicante, Spain (First received 10 June 1999; accepted in revised form 25 April 2000) Abstract}Rapid small-scale minicolumn tests were carried out to simulate the atrazine adsorption in water phase with three pelletized pitch-based activated carbon fibers (ACF) and one commercial granular activated carbon (GAC). Initial atrazine solutions were prepared with pretreated ground water. Minicolumn tests showed that the performance of highly activated carbon fibers (surface area of 1700 m2/g) is around 7 times better than the commercial GAC (with surface area at around 1100 m2/g), whereas carbon fibers with medium activation degree (surface area of 1500 m2/g) had a removal efficiency worse than the commercial carbon. The high removal efficiency of the highly activated ACF is due to the wide-opened microstructure of the material, with an appreciable contribution of the low size mesopores, maintaining at these conditions a fast kinetic adsorption rate rather than a selective adsorbent for micropollutants vs. natural organic matter. # 2000 Elsevier Science Ltd. All rights reserved Key words}ACF, atrazine, natural organic matter, minicolumn, pore size distribution
INTRODUCTION
Environmental regulations in developed countries have become very strict for drinking water treatment over the last few years, especially regarding pesticide compounds (levels decreased to only 0.1 mg/l). Usually, the drinking water treatment ends in adsorption with granular activated carbon (GAC) beds and/or ozonization–oxidation treatment, which removes taste, odour and other micropollutants. GAC beds have a short life due to these stricter regulations, although the taste and odour levels are apparently well removed (Smetham, 1994; Horner et al., 1998). Consequently, there is a growing interest in developing new carbon adsorbents which are more effective for the micropollutants (Horner et al., 1998), as well as to deepen the understanding of the micropollutants adsorption in water phase, where there are interferences with natural organic matter (NOM) which negatively decreases effectiveness of GAC (Smith and Weber, 1985). Activated carbon fibers (ACF) have received increasing attention in recent years as a better product than GAC in several adsorption fields, and normally present much higher adsorption capacity and adsorption kinetics than regular GAC (Suzuki, 1994). Whereas the adsorption onto GAC is
*Author to whom all correspondence should be addressed. Tel.: +34-96-5903400 ext. 2426; fax: +34-96-5903826; e-mail: [email protected] 516
produced through an intraparticle porous network consisting of macro-, meso- and micropores, ACF present by nature only micropores (no meso- and macronetwork), which are directly accessible from the external surface fiber. The adsorption kinetics is then obviously faster than that of the GAC (Suzuki, 1994). Several recent works tested the performance of ACF vs. GAC in the aqueous removal of micropollutants. Brasquet and Le Cloriec (1997) studied the performance of one ACF (from rayon, 1500 m2/g) vs. GAC (from coconut, 1200 m2/g) for the static adsorption of aromatic micropollutants, reporting that ACF had higher adsorption kinetics and capacity than GAC, highlighting that the presence of humic substances did not alter the micropollutant adsorption onto the ACF, and concluding that ACF were selective for the organics respecting to humic substances. Pelekani and Snoeyink (1999) studied the pesticide atrazine adsorption onto two phenolic resin based-ACF (885 and 2312 m2/g specific surface area). These authors carried out adsorption isotherms of atrazine in deionized water and untreated ground water. These authors found that ACF with pores bigger than atrazine and smaller than NOM (NOM was not adsorbed) adsorbed much less atrazine than ACF with a widened micropore structure (where both were adsorbed), i.e. in the first case there is a pore blocking by NOM, and in the latter a direct adsorption competition where the micropollutant is preferred. This is in disagreement with previous
Dynamic pesticide removal with activated carbon fibers
findings of Brasquet and Le Cloriec (1997), who worked with humic substances as NOM instead of with natural ground waters. Hopman et al. (1995, 1996) studied the pesticide adsorption of several ACF (surface area from 1500 to 2500 m2/g) vs. GAC (peat based, 1000 m2/g) in both static and dynamic runs. There was a higher adsorption of all the ACF vs. GAC in the static runs; nevertheless, dynamic tests in minicolumn showed that only ACF with a surface area above 2000 m2/g performed better than GAC, but not those ACF with surface areas below 2000 m2/g. Additionally, they reported that ACF with a surface area below 2000 m2/g did not adsorb NOM molecules, and that those samples with a surface area above 2000 m2/g, although giving the best pesticide removal performance, yielded some kind of NOM adsorption. Finally, they attributed the higher performance of ACF vs. GAC to the absence of transport network pores, being a selective adsorbent for micropollutants and not influenced by NOM. Although the results are not contrary to those reported by Pelekani and Snoeyink (1999), the explanation fully disagrees with them. The present work studies the performance of three different ACF and one commercial GAC in the dynamic adsorption of atrazine. Dynamic tests were run in a minicolumn, in order to simulate a full-scale adsorption tower of water treatment. ACF were produced in the laboratory and were fully characterized. The analysis of the performance in the pesticide removal is fundamental in the adsorption and structure characteristics of the sample. The results of the present work may help to complete the understanding of the adsorption behavior of pesticides over ACF.
517
Specific surface areas were calculated applying the BET equation to the butane adsorption isotherm data. Micropore volume was determined applying the Dubinin–Raduskievich (DR) equation (Dubinin and Stoeckli, 1980) to the adsorption isotherm data, and the mesopore volume was evaluated semiquantitatively by mercury porosimeter (Jankowska et al., 1991). Additionally, other analytical methods, more common in industry such as iodine and methylene blue indexes, were done to characterize the adsorption capacity (Jankowska et al., 1991). Table 1 shows the results of the sample characterization, and Fig. 1 shows the adsorption isotherms for the three pelletized ACF produced (ACF-48, ACF-58 and ACF-68) and the commercial GAC (GAC1240), showing the evolution of the porosity in the pelletized ACF when increasing the activation degree. As expected, micropore structure widens as the activation proceeds, and the micropore volume and specific surface area increases. The lowactivated ACF (ACF-48) is essentially microporous, with a good iodine number and surface area and low values of both mesopore volume and methylene blue. This fact can be observed in the isotherm plot of ACF-48, which presents a close knee at low relative pressures (close micropore structure) and a low adsorption difference at relative pressures between 0.1 and 0.42, which reveals low contribution of the low size mesopores (Rodriguez-Reinoso and Linares-Solano, 1988). The other two ACF present excellent high values of the adsorption capacity (1400 iodine and 1600 m2/g), and their isotherms have a more opened knee (up to relative pressures of 0.1) and a big difference in the amount adsorbed at relative pressures between 0.1 and 0.42, which implies a widened microstructure with higher contribution of the high size micropores and the low size mesopores, especially for the highest activated ACF-68. This fact is corroborated by the mesopore volume obtained by mercury porosimetry (0.89 vs. 0.20 cm3/g of the ACF48). The commercial GAC1240 has a similar adsorption capacity as low activated ACF-48, but a more opened microstructure towards the mesoporosity, which can be observed in a higher difference between the 0.1 and 0.42 relative pressure and by mercury porosimetry (0.35 vs. 0.20 cm3/g). In the physical properties, GAC1240 presents a good apparent density of 500 kg/m3, much higher than the pelletized ACF. However, the apparent density reached by the pelletized ACF is relatively high, taking into account the shape of the milled fibers which form the pellets.
MATERIALS AND METHODS
ACF preparation and characterization
Experimental setup
Commercial pitch-based carbon fibers (from Donacarbo Ltd. ) were milled, mixed with an appropriate lubricant and pelletized to 1 mm diameter. Afterwards, these pelletized carbon fibers were steam activated at 800–10008C at three different times, obtaining three different samples with burnoff degrees of 48, 58 and 68%. NORIT GAC1240 (from Norit N.V.), coal-based carbon typical for drinking water treatment, was used to compare the performance of the three different ACF vs. a commercial GAC. 298 K butane adsorption isotherms and mercury porosimetries were used to characterize the carbon samples.
Since the objective of the present work was to simulate the pesticide removal in a full-scale tower, an appropiate type of water was used. Pretreated ground water which was coagulated, sedimented and sand filtrated only (not ozonized or GAC treated) was collected in a water-works plant. This pretreated water was free of pesticides. The remaining NOM was measured through the dissolved organic compounds (DOC, Standard Method 5910), with a value of 1.5 mg/l. Initially, a solution of 2.5 mg/l (ppb) atrazine was prepared spiking the pesticide on the pretreated water.
Table 1. Adsorption characteristics of the pelletized ACF and the commercial GAC1240 Carbon sample
ACF-48
ACF-58
ACF-68
GAC1240
Micropore volume (cm3/g) BET surface area (m2/g) Mesopore volume (cm3/g) Iodine number Methylene blue Apparent density (kg/m3)
0.38 1061 0.20 1045 18.5 345
0.57 1570 0.21 1425 29.0 276
0.61 1665 0.89 1365 27.0 254
0.38 1062 0.45 1028 } 495
I. Martı´ n-Gullo´n and R. Font
518
permitted for any pesticide. The runs are stopped after the outflow stream reaches the breakthrough point.
RESULTS
Minicolumn test with ACF and GAC
Fig. 1. 298 K adsorption isotherms of the three ACF samples and the commercial GAC1240.
Fig. 2. Scheme of the experimental apparatus used to carry out the dynamic adsorption test.
Figure 2 shows the experimental apparatus used to carry out the atrazine adsorption in water phase. The 2.5 ppb atrazine solution was conducted by a HPLC pump through a 1/800 tube to the carbon bed, placed in a tube of 2.1 mm internal diameter, with fine grids at the top and the bottom to avoid carbon entrainment. The atrazine solution crossed down the carbon bed, and was finally collected in an autosampler, which took water samples every 20 min. The atrazine in samples collected was analyzed by Envirograd Triazine Plate Kits (Millipore Corp). This equipment setup follows the scaling method proposed by Crittenden et al. (1991) to simulate GAC adsorption towers in minicolumn rapid scale runs. Pelletized ACF and GAC were milled to a given size (around 100 mm), and the water stream was fixed to a given flow (1–2.5 ml/ min), and a specific weight of carbon (40–80 mg) was placed in the tube bed in order to simulate a typical adsorption GAC tower, with an empty bed contact time (EBCT) of 10 min and a spatial velocity of 3 m/h. The breakthrough point is considered 0.1 mg/l because this is the limit
Table 2 shows the operating conditions of the different dynamic adsorption tests carried out in a minicolumn, as well as the corresponding parameters in the full-scale simulating conditions. Figure 3 shows the breakthrough plots (atrazine concentration of the outflow water stream vs. the amount of water passed) of the dynamic tests done for ACF-48, ACF-58, ACF-68 and GAC1240. ACF-48 reaches the breakthrough nearly immediately, giving a very poor performance; ACF-58 also reaches the breakthrough point nearly immediately, much worse than the commercial granular carbon. In retrospect to the high activated ACF (ACF-68), it performs 7 times better in the pesticide adsorption than the commercial GAC1240: ACF-68 reaches 0.1 mg/l level when over 240 liters of atrazine solution per gram of carbon passed through the carbon bed, while the commercial GAC1240 reaches the breakthrough point at around 34 l/g. Minicolumn tests with NOM preloaded ACF and GAC Two additional runs were carried out with those samples which gave better performance, ACF-68 and GAC1240.. The difference with respect to the previous runs is that, before pumping the atrazine solution through the carbon bed, pretreated water without spiked atrazine was pumped through the carbon bed at the same water flow rate during 3 days (around 150 l/g). Pretreated water contains common levels of NOM, and in this way the carbon samples were absolutely saturated with NOM prior to the atrazine adsorption test. The results will indicate if NOM, and especially the preloading of NOM, interferes in the atrazine adsorption onto the highest activated ACF, and how important this interference is. Figure 4 shows that effectively preloaded NOM interferes in the atrazine adsorption. GAC1240 drastically diminishes the performance, yielding an immediate breakthrough, while the performance of the ACF-68 is reduced from 240 to 80 l/g, although it
Table 2. Operating parameters of the minicolumn tests carried out and the corresponding simulation conditions for full scale (Crittenden et al., 1991)a Sample ACF-48 ACF-58 ACF-68 ACF-68-preloaded GAC1240 a
Q (ml/min)
W (mg)
EBCTmc (s)
EBCTfs (min)
Vmc (m/h)
Vfs (m/h)
1.7 1.7 1.7 1.7 1.7
55.7 46.5 42.1 41.3 84.7
5.7 5.9 5.8 5.7 6.0
9.5 9.9 9.7 9.6 10.1
29.4 29.4 29.4 29.4 29.4
2.9 2.9 2.9 2.9 2.9
Notation: Q, atrazine solution influent flow; W, carbon sample weight in minicolumn; subscripts mc and fs are minicolum and full-scale, respectively; V, spatial velocity.
Dynamic pesticide removal with activated carbon fibers
Fig. 3. Atrazine adsorption dynamic minicolumn test breakthrough plots of pelletized ACF and the commercial GAC, represented vs. amount of solution passed per mass of carbon.
Fig 4. Breakthrough plots of ACF-68 and GAC1240 in dynamic minicolumn test preloaded firstly with NOM.
is still higher than the performance of the GAC1240 without NOM preloaded.
DISCUSSION
The experimental results obtained in this work show that in dynamic runs simulating full scale conditions, ACF can perform better than a typical GAC for water treatment. Nevertheless, the experimental results also show that to perform better than GAC, the ACF must be highly activated, in order to have a wide-opened microporosity with an appreciable contribution of the low size mesopores. Finally, it was shown that the presence of NOM considerably affects the efficiency of the ACF, decreasing the adsorption by around 65% for the ACF-68. Atrazine molecular size is between 0.80 and 0.85 nm, and it can then be adsorbed in the micropores, which have pore width up to 2 nm. On the other hand, NOM in an aquatic environment is comprised of organic and bioorganic substances formed by the breakdown of animal and vegetable matter (Horner et al., 1998). Regardless of whether NOM varies in composition among different raw waters, these organic substances are a mixture of complex compounds (sugars and amino sugars, fulvic
519
and humic acids), with molecular weights comprising typically between 500 and 250,000. A 500 molecular weight molecule presents a medium size of around 1.4 nm, so NOM compounds start to interfere with the adsorption in the mesopores and bigger micropores, which was observed by Pelekani and Snoeyink (1999) and Hopman et al. (1995, 1996). ACF-48 presents a narrow microporous structure (with adsorption capacities similar to the commercial GAC1240) with a negligible contribution of the mesopores. Furthermore, this sample is the only one where the micropores are bigger than atrazine and smaller than NOM, and in principle, with molecular sieve effect of the atrazine vs. NOM. However, it yields nearly no atrazine adsorption, with an immediate breakthrough. This result agrees with those of Pelekani and Snoeyink (1999). ACF-58 has a much wider pore structure and adsorption capacity (1500 m2/g), with high contribution of high size micropores and a not negligible amount of mesopores (which come from the micropore widening). In this case, it can be assumed that adsorption pores are bigger than atrazine and of similar size to low NOM compounds. ACF-58 performed worse than the commercial granular activated carbon too. On the other hand, ACF-68 gave the best performance in the atrazine removal, and there is no doubt that in ACF-68 (with an absolutely wide-opened pore structure and similar adsorption capacity to ACF58), NOM molecules fit in the pore structure and, consequently, interfere with the atrazine adsorption. This fact was tested when the sample was first preloaded with NOM (65% reduction in efficiency). These results show that adsorbents with a fiber-type structure present different activities in the atrazine removal depending on their pore development. In the present work, the atrazine removal activity is ACF-68>GAC1240>ACF-58>ACF-48, which can be related with the more opened micropore structure of the carbon adsorbents, regardless of the total adsorption capacity. Following the above explanation, ACF-48 is a microporous adsorbent with a good adsorption capacity, but it gives a poor performance in dynamic atrazine tests very probably due to its low kinetics and not because the adsorption capacity is ended. The same explanation is valid for ACF-58; the structure is more opened, but not enough to give higher kinetics, and consequently, at EBCT of only 6 s, the performance is still very poor in spite of its excellent adsorption capacity and fiber shape structure. The ACF-68, the most activated sample studied with a pore size distribution in the border micropore–mesopore, gives an excellent performance in the atrazine adsorption, because its wide opened structure allows a very high adsorption rate. The commercial GAC presents a wide-opened structure, and its efficiency is good, but lower than ACF-68 because the structure is not so open and consequently the adsorption kinetic rate is not so fast.
I. Martı´ n-Gullo´n and R. Font
520 CONCLUSIONS *
*
Pitch-based ACF are more effective in the atrazine removal dynamic tests than GAC if they are highly activated. The rapid adsorption kinetics of the atrazine (in the presence of NOM) with the highly activated ACF seems to be the main reason for its having a better performance than GAC. This means that a fiber-type structure with micropores directly accesible from the surface is not enough reason to justify the good efficiency of ACF.
REFERENCES
Brasquet C. and Le Cloriec P. (1997) Adsorption onto activated carbon fibers: application to water and air treatments. Carbon 35, 1307–1313. Crittenden J. C., Reddy P. S., Arora H., Trynoski J., Hand D. W., Perran D. L. and Summers S. (1991) Predicting GAC performance with rapid small-scale column test. J. Am. Water Works Assoc. 83, 77–87. Dubinin M. M. and Stoeckli H. F. (1980) Homogeneous and heterogeneous micropore structures in carbonoceous materials. J. Colloid Interface Sci. 75(1), 34–42.
Hopman R., Siegers W. G. and Kruithof J. C. (1995) Organic micropollutant removal by activated carbon fiber filtration. Water Supply 13, 257–261. Hopman R., Siegers W. G. and Kruithof J. C. (1996) Activated carbon fiber filtration: an innovative technique for pesticide removal. Proceedings of Water Quality Technology Conference, Pt. 1, pp. 511–522. Amer. Water Works Assoc. Horner D. J., Streat M., Hellgart K. and Mistry B. (1998) Selective sorption of atrazine from aqueous solutions using activated carbon. Proceedings of Icheme Research Event. Jankowska H., Swiatkowki A. and Chroma J. (1991) Active Carbon. Ellis Horwood, London. Pelekani C. and Snoeyink V. L. (1999) Competitive adsorption in natural water: role of activated carbon pore size. Water Res. 33(5), 1209–1219. Rodriguez-Reinoso F. and Linares-Solano A. (1988) Microporous structure of activated carbons as revealed by adsorption methods. In Chemistry and Physics of Carbons, Vol. 21, ed. P. A. Thrower, pp. 1–146. Marcel Dekker, New York. Smetham N. B. (1994) Activated Carbon in Drinking Water. South Water Services Ltd., London. Smith E. H. and Weber W. J. (1985) The effect of the dissolved organic matter on the adsorption capacity of organic compounds on activated carbon. Proceedings of American Water Works Association Annual Conference, p. 553. Suzuki M. (1994) Activated Carbon Fiber: Fundamentals and Applications. Carbon 32, 577–586.
Wat. Res. Vol. 35, No. 2, pp. 516–520, 2001 # 2000 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/00/$ - see front matter
DYNAMIC PESTICIDE REMOVAL WITH ACTIVATED CARBON FIBERS I. MARTI´N-GULLO´N* and R. FONT Chemical Engineering Department, Universidad de Alicante, PO Box 99, 03080 Alicante, Spain (First received 10 June 1999; accepted in revised form 25 April 2000) Abstract}Rapid small-scale minicolumn tests were carried out to simulate the atrazine adsorption in water phase with three pelletized pitch-based activated carbon fibers (ACF) and one commercial granular activated carbon (GAC). Initial atrazine solutions were prepared with pretreated ground water. Minicolumn tests showed that the performance of highly activated carbon fibers (surface area of 1700 m2/g) is around 7 times better than the commercial GAC (with surface area at around 1100 m2/g), whereas carbon fibers with medium activation degree (surface area of 1500 m2/g) had a removal efficiency worse than the commercial carbon. The high removal efficiency of the highly activated ACF is due to the wide-opened microstructure of the material, with an appreciable contribution of the low size mesopores, maintaining at these conditions a fast kinetic adsorption rate rather than a selective adsorbent for micropollutants vs. natural organic matter. # 2000 Elsevier Science Ltd. All rights reserved Key words}ACF, atrazine, natural organic matter, minicolumn, pore size distribution
INTRODUCTION
Environmental regulations in developed countries have become very strict for drinking water treatment over the last few years, especially regarding pesticide compounds (levels decreased to only 0.1 mg/l). Usually, the drinking water treatment ends in adsorption with granular activated carbon (GAC) beds and/or ozonization–oxidation treatment, which removes taste, odour and other micropollutants. GAC beds have a short life due to these stricter regulations, although the taste and odour levels are apparently well removed (Smetham, 1994; Horner et al., 1998). Consequently, there is a growing interest in developing new carbon adsorbents which are more effective for the micropollutants (Horner et al., 1998), as well as to deepen the understanding of the micropollutants adsorption in water phase, where there are interferences with natural organic matter (NOM) which negatively decreases effectiveness of GAC (Smith and Weber, 1985). Activated carbon fibers (ACF) have received increasing attention in recent years as a better product than GAC in several adsorption fields, and normally present much higher adsorption capacity and adsorption kinetics than regular GAC (Suzuki, 1994). Whereas the adsorption onto GAC is
*Author to whom all correspondence should be addressed. Tel.: +34-96-5903400 ext. 2426; fax: +34-96-5903826; e-mail: [email protected] 516
produced through an intraparticle porous network consisting of macro-, meso- and micropores, ACF present by nature only micropores (no meso- and macronetwork), which are directly accessible from the external surface fiber. The adsorption kinetics is then obviously faster than that of the GAC (Suzuki, 1994). Several recent works tested the performance of ACF vs. GAC in the aqueous removal of micropollutants. Brasquet and Le Cloriec (1997) studied the performance of one ACF (from rayon, 1500 m2/g) vs. GAC (from coconut, 1200 m2/g) for the static adsorption of aromatic micropollutants, reporting that ACF had higher adsorption kinetics and capacity than GAC, highlighting that the presence of humic substances did not alter the micropollutant adsorption onto the ACF, and concluding that ACF were selective for the organics respecting to humic substances. Pelekani and Snoeyink (1999) studied the pesticide atrazine adsorption onto two phenolic resin based-ACF (885 and 2312 m2/g specific surface area). These authors carried out adsorption isotherms of atrazine in deionized water and untreated ground water. These authors found that ACF with pores bigger than atrazine and smaller than NOM (NOM was not adsorbed) adsorbed much less atrazine than ACF with a widened micropore structure (where both were adsorbed), i.e. in the first case there is a pore blocking by NOM, and in the latter a direct adsorption competition where the micropollutant is preferred. This is in disagreement with previous
Dynamic pesticide removal with activated carbon fibers
findings of Brasquet and Le Cloriec (1997), who worked with humic substances as NOM instead of with natural ground waters. Hopman et al. (1995, 1996) studied the pesticide adsorption of several ACF (surface area from 1500 to 2500 m2/g) vs. GAC (peat based, 1000 m2/g) in both static and dynamic runs. There was a higher adsorption of all the ACF vs. GAC in the static runs; nevertheless, dynamic tests in minicolumn showed that only ACF with a surface area above 2000 m2/g performed better than GAC, but not those ACF with surface areas below 2000 m2/g. Additionally, they reported that ACF with a surface area below 2000 m2/g did not adsorb NOM molecules, and that those samples with a surface area above 2000 m2/g, although giving the best pesticide removal performance, yielded some kind of NOM adsorption. Finally, they attributed the higher performance of ACF vs. GAC to the absence of transport network pores, being a selective adsorbent for micropollutants and not influenced by NOM. Although the results are not contrary to those reported by Pelekani and Snoeyink (1999), the explanation fully disagrees with them. The present work studies the performance of three different ACF and one commercial GAC in the dynamic adsorption of atrazine. Dynamic tests were run in a minicolumn, in order to simulate a full-scale adsorption tower of water treatment. ACF were produced in the laboratory and were fully characterized. The analysis of the performance in the pesticide removal is fundamental in the adsorption and structure characteristics of the sample. The results of the present work may help to complete the understanding of the adsorption behavior of pesticides over ACF.
517
Specific surface areas were calculated applying the BET equation to the butane adsorption isotherm data. Micropore volume was determined applying the Dubinin–Raduskievich (DR) equation (Dubinin and Stoeckli, 1980) to the adsorption isotherm data, and the mesopore volume was evaluated semiquantitatively by mercury porosimeter (Jankowska et al., 1991). Additionally, other analytical methods, more common in industry such as iodine and methylene blue indexes, were done to characterize the adsorption capacity (Jankowska et al., 1991). Table 1 shows the results of the sample characterization, and Fig. 1 shows the adsorption isotherms for the three pelletized ACF produced (ACF-48, ACF-58 and ACF-68) and the commercial GAC (GAC1240), showing the evolution of the porosity in the pelletized ACF when increasing the activation degree. As expected, micropore structure widens as the activation proceeds, and the micropore volume and specific surface area increases. The lowactivated ACF (ACF-48) is essentially microporous, with a good iodine number and surface area and low values of both mesopore volume and methylene blue. This fact can be observed in the isotherm plot of ACF-48, which presents a close knee at low relative pressures (close micropore structure) and a low adsorption difference at relative pressures between 0.1 and 0.42, which reveals low contribution of the low size mesopores (Rodriguez-Reinoso and Linares-Solano, 1988). The other two ACF present excellent high values of the adsorption capacity (1400 iodine and 1600 m2/g), and their isotherms have a more opened knee (up to relative pressures of 0.1) and a big difference in the amount adsorbed at relative pressures between 0.1 and 0.42, which implies a widened microstructure with higher contribution of the high size micropores and the low size mesopores, especially for the highest activated ACF-68. This fact is corroborated by the mesopore volume obtained by mercury porosimetry (0.89 vs. 0.20 cm3/g of the ACF48). The commercial GAC1240 has a similar adsorption capacity as low activated ACF-48, but a more opened microstructure towards the mesoporosity, which can be observed in a higher difference between the 0.1 and 0.42 relative pressure and by mercury porosimetry (0.35 vs. 0.20 cm3/g). In the physical properties, GAC1240 presents a good apparent density of 500 kg/m3, much higher than the pelletized ACF. However, the apparent density reached by the pelletized ACF is relatively high, taking into account the shape of the milled fibers which form the pellets.
MATERIALS AND METHODS
ACF preparation and characterization
Experimental setup
Commercial pitch-based carbon fibers (from Donacarbo Ltd. ) were milled, mixed with an appropriate lubricant and pelletized to 1 mm diameter. Afterwards, these pelletized carbon fibers were steam activated at 800–10008C at three different times, obtaining three different samples with burnoff degrees of 48, 58 and 68%. NORIT GAC1240 (from Norit N.V.), coal-based carbon typical for drinking water treatment, was used to compare the performance of the three different ACF vs. a commercial GAC. 298 K butane adsorption isotherms and mercury porosimetries were used to characterize the carbon samples.
Since the objective of the present work was to simulate the pesticide removal in a full-scale tower, an appropiate type of water was used. Pretreated ground water which was coagulated, sedimented and sand filtrated only (not ozonized or GAC treated) was collected in a water-works plant. This pretreated water was free of pesticides. The remaining NOM was measured through the dissolved organic compounds (DOC, Standard Method 5910), with a value of 1.5 mg/l. Initially, a solution of 2.5 mg/l (ppb) atrazine was prepared spiking the pesticide on the pretreated water.
Table 1. Adsorption characteristics of the pelletized ACF and the commercial GAC1240 Carbon sample
ACF-48
ACF-58
ACF-68
GAC1240
Micropore volume (cm3/g) BET surface area (m2/g) Mesopore volume (cm3/g) Iodine number Methylene blue Apparent density (kg/m3)
0.38 1061 0.20 1045 18.5 345
0.57 1570 0.21 1425 29.0 276
0.61 1665 0.89 1365 27.0 254
0.38 1062 0.45 1028 } 495
I. Martı´ n-Gullo´n and R. Font
518
permitted for any pesticide. The runs are stopped after the outflow stream reaches the breakthrough point.
RESULTS
Minicolumn test with ACF and GAC
Fig. 1. 298 K adsorption isotherms of the three ACF samples and the commercial GAC1240.
Fig. 2. Scheme of the experimental apparatus used to carry out the dynamic adsorption test.
Figure 2 shows the experimental apparatus used to carry out the atrazine adsorption in water phase. The 2.5 ppb atrazine solution was conducted by a HPLC pump through a 1/800 tube to the carbon bed, placed in a tube of 2.1 mm internal diameter, with fine grids at the top and the bottom to avoid carbon entrainment. The atrazine solution crossed down the carbon bed, and was finally collected in an autosampler, which took water samples every 20 min. The atrazine in samples collected was analyzed by Envirograd Triazine Plate Kits (Millipore Corp). This equipment setup follows the scaling method proposed by Crittenden et al. (1991) to simulate GAC adsorption towers in minicolumn rapid scale runs. Pelletized ACF and GAC were milled to a given size (around 100 mm), and the water stream was fixed to a given flow (1–2.5 ml/ min), and a specific weight of carbon (40–80 mg) was placed in the tube bed in order to simulate a typical adsorption GAC tower, with an empty bed contact time (EBCT) of 10 min and a spatial velocity of 3 m/h. The breakthrough point is considered 0.1 mg/l because this is the limit
Table 2 shows the operating conditions of the different dynamic adsorption tests carried out in a minicolumn, as well as the corresponding parameters in the full-scale simulating conditions. Figure 3 shows the breakthrough plots (atrazine concentration of the outflow water stream vs. the amount of water passed) of the dynamic tests done for ACF-48, ACF-58, ACF-68 and GAC1240. ACF-48 reaches the breakthrough nearly immediately, giving a very poor performance; ACF-58 also reaches the breakthrough point nearly immediately, much worse than the commercial granular carbon. In retrospect to the high activated ACF (ACF-68), it performs 7 times better in the pesticide adsorption than the commercial GAC1240: ACF-68 reaches 0.1 mg/l level when over 240 liters of atrazine solution per gram of carbon passed through the carbon bed, while the commercial GAC1240 reaches the breakthrough point at around 34 l/g. Minicolumn tests with NOM preloaded ACF and GAC Two additional runs were carried out with those samples which gave better performance, ACF-68 and GAC1240.. The difference with respect to the previous runs is that, before pumping the atrazine solution through the carbon bed, pretreated water without spiked atrazine was pumped through the carbon bed at the same water flow rate during 3 days (around 150 l/g). Pretreated water contains common levels of NOM, and in this way the carbon samples were absolutely saturated with NOM prior to the atrazine adsorption test. The results will indicate if NOM, and especially the preloading of NOM, interferes in the atrazine adsorption onto the highest activated ACF, and how important this interference is. Figure 4 shows that effectively preloaded NOM interferes in the atrazine adsorption. GAC1240 drastically diminishes the performance, yielding an immediate breakthrough, while the performance of the ACF-68 is reduced from 240 to 80 l/g, although it
Table 2. Operating parameters of the minicolumn tests carried out and the corresponding simulation conditions for full scale (Crittenden et al., 1991)a Sample ACF-48 ACF-58 ACF-68 ACF-68-preloaded GAC1240 a
Q (ml/min)
W (mg)
EBCTmc (s)
EBCTfs (min)
Vmc (m/h)
Vfs (m/h)
1.7 1.7 1.7 1.7 1.7
55.7 46.5 42.1 41.3 84.7
5.7 5.9 5.8 5.7 6.0
9.5 9.9 9.7 9.6 10.1
29.4 29.4 29.4 29.4 29.4
2.9 2.9 2.9 2.9 2.9
Notation: Q, atrazine solution influent flow; W, carbon sample weight in minicolumn; subscripts mc and fs are minicolum and full-scale, respectively; V, spatial velocity.
Dynamic pesticide removal with activated carbon fibers
Fig. 3. Atrazine adsorption dynamic minicolumn test breakthrough plots of pelletized ACF and the commercial GAC, represented vs. amount of solution passed per mass of carbon.
Fig 4. Breakthrough plots of ACF-68 and GAC1240 in dynamic minicolumn test preloaded firstly with NOM.
is still higher than the performance of the GAC1240 without NOM preloaded.
DISCUSSION
The experimental results obtained in this work show that in dynamic runs simulating full scale conditions, ACF can perform better than a typical GAC for water treatment. Nevertheless, the experimental results also show that to perform better than GAC, the ACF must be highly activated, in order to have a wide-opened microporosity with an appreciable contribution of the low size mesopores. Finally, it was shown that the presence of NOM considerably affects the efficiency of the ACF, decreasing the adsorption by around 65% for the ACF-68. Atrazine molecular size is between 0.80 and 0.85 nm, and it can then be adsorbed in the micropores, which have pore width up to 2 nm. On the other hand, NOM in an aquatic environment is comprised of organic and bioorganic substances formed by the breakdown of animal and vegetable matter (Horner et al., 1998). Regardless of whether NOM varies in composition among different raw waters, these organic substances are a mixture of complex compounds (sugars and amino sugars, fulvic
519
and humic acids), with molecular weights comprising typically between 500 and 250,000. A 500 molecular weight molecule presents a medium size of around 1.4 nm, so NOM compounds start to interfere with the adsorption in the mesopores and bigger micropores, which was observed by Pelekani and Snoeyink (1999) and Hopman et al. (1995, 1996). ACF-48 presents a narrow microporous structure (with adsorption capacities similar to the commercial GAC1240) with a negligible contribution of the mesopores. Furthermore, this sample is the only one where the micropores are bigger than atrazine and smaller than NOM, and in principle, with molecular sieve effect of the atrazine vs. NOM. However, it yields nearly no atrazine adsorption, with an immediate breakthrough. This result agrees with those of Pelekani and Snoeyink (1999). ACF-58 has a much wider pore structure and adsorption capacity (1500 m2/g), with high contribution of high size micropores and a not negligible amount of mesopores (which come from the micropore widening). In this case, it can be assumed that adsorption pores are bigger than atrazine and of similar size to low NOM compounds. ACF-58 performed worse than the commercial granular activated carbon too. On the other hand, ACF-68 gave the best performance in the atrazine removal, and there is no doubt that in ACF-68 (with an absolutely wide-opened pore structure and similar adsorption capacity to ACF58), NOM molecules fit in the pore structure and, consequently, interfere with the atrazine adsorption. This fact was tested when the sample was first preloaded with NOM (65% reduction in efficiency). These results show that adsorbents with a fiber-type structure present different activities in the atrazine removal depending on their pore development. In the present work, the atrazine removal activity is ACF-68>GAC1240>ACF-58>ACF-48, which can be related with the more opened micropore structure of the carbon adsorbents, regardless of the total adsorption capacity. Following the above explanation, ACF-48 is a microporous adsorbent with a good adsorption capacity, but it gives a poor performance in dynamic atrazine tests very probably due to its low kinetics and not because the adsorption capacity is ended. The same explanation is valid for ACF-58; the structure is more opened, but not enough to give higher kinetics, and consequently, at EBCT of only 6 s, the performance is still very poor in spite of its excellent adsorption capacity and fiber shape structure. The ACF-68, the most activated sample studied with a pore size distribution in the border micropore–mesopore, gives an excellent performance in the atrazine adsorption, because its wide opened structure allows a very high adsorption rate. The commercial GAC presents a wide-opened structure, and its efficiency is good, but lower than ACF-68 because the structure is not so open and consequently the adsorption kinetic rate is not so fast.
I. Martı´ n-Gullo´n and R. Font
520 CONCLUSIONS *
*
Pitch-based ACF are more effective in the atrazine removal dynamic tests than GAC if they are highly activated. The rapid adsorption kinetics of the atrazine (in the presence of NOM) with the highly activated ACF seems to be the main reason for its having a better performance than GAC. This means that a fiber-type structure with micropores directly accesible from the surface is not enough reason to justify the good efficiency of ACF.
REFERENCES
Brasquet C. and Le Cloriec P. (1997) Adsorption onto activated carbon fibers: application to water and air treatments. Carbon 35, 1307–1313. Crittenden J. C., Reddy P. S., Arora H., Trynoski J., Hand D. W., Perran D. L. and Summers S. (1991) Predicting GAC performance with rapid small-scale column test. J. Am. Water Works Assoc. 83, 77–87. Dubinin M. M. and Stoeckli H. F. (1980) Homogeneous and heterogeneous micropore structures in carbonoceous materials. J. Colloid Interface Sci. 75(1), 34–42.
Hopman R., Siegers W. G. and Kruithof J. C. (1995) Organic micropollutant removal by activated carbon fiber filtration. Water Supply 13, 257–261. Hopman R., Siegers W. G. and Kruithof J. C. (1996) Activated carbon fiber filtration: an innovative technique for pesticide removal. Proceedings of Water Quality Technology Conference, Pt. 1, pp. 511–522. Amer. Water Works Assoc. Horner D. J., Streat M., Hellgart K. and Mistry B. (1998) Selective sorption of atrazine from aqueous solutions using activated carbon. Proceedings of Icheme Research Event. Jankowska H., Swiatkowki A. and Chroma J. (1991) Active Carbon. Ellis Horwood, London. Pelekani C. and Snoeyink V. L. (1999) Competitive adsorption in natural water: role of activated carbon pore size. Water Res. 33(5), 1209–1219. Rodriguez-Reinoso F. and Linares-Solano A. (1988) Microporous structure of activated carbons as revealed by adsorption methods. In Chemistry and Physics of Carbons, Vol. 21, ed. P. A. Thrower, pp. 1–146. Marcel Dekker, New York. Smetham N. B. (1994) Activated Carbon in Drinking Water. South Water Services Ltd., London. Smith E. H. and Weber W. J. (1985) The effect of the dissolved organic matter on the adsorption capacity of organic compounds on activated carbon. Proceedings of American Water Works Association Annual Conference, p. 553. Suzuki M. (1994) Activated Carbon Fiber: Fundamentals and Applications. Carbon 32, 577–586.