Applied Surface Science 200 (2002) 196±202 Characterization of synthetic carbons activated with phosphoric acid A.M. Puziy a,*, O.I. Poddubnaya a, A. MartõÂnez-Alonso b, F. SuaÂrez-GarcõÂa b, J.M.D. TascoÂn b a Institute for Sorption and Problems of Endoecology, Naumov St. 13, 03164 Kiev, Ukraine b Instituto Nacional del CarboÂn, CSIC, Apartado 73, 33080 Oviedo, Spain Received 13 November 2001; received in revised form 13 November 2001; accepted 28 July 2002 Abstract The structuralheterogeneity of synthetic phosphoric acid activated carbons has been analyzed using pore-size distributions (PSDs) obtained from nitrogen at 196 8C and carbon dioxide at 0 8C isotherms. PSDs where obtained by the BET±Kelvin method. It is shown that the BET±Kelvin method is in good agreement with DFT and provides a fast means for assessment of the porous structure of adsorbents. PSDs obtained by the BET±Kelvin method using different adsorbates give results consistent with each other. Due to the restricted pressure range for carbon dioxide adsorption isotherm the PSD gives information only about pores in the micropore range. The agreement between different methods is better for small micropores. # 2002 Elsevier Science B.V. All rights reserved. PACS: 81.20, methods of materials synthesis and materials processing; 81.05, porous materials, granular materials; 81.40, treatment of materials and its effects on microstructure and properties Keywords: Synthetic carbons; Chemicalactivation; Adsorption modeling; Pore-size distribution 1. Introduction Phosphoric acid activation is widely used for the production of activated carbons. The use of various precursors including wood, nut shells, viscose rayon and coalhas been reported. It has been shown that maximum surface area is obtained at carbonization temperatures of 350 8C for white oak wood [1], 450 8C for coconut shell and subbituminous coal [2] and 500 8C for bituminous coal [1]. However, other precursors such as porous polymers have not been investigated in detail [3]. In the present communication, * Corresponding author. Fax: 380-44-452-9327. E-mail address: alexander.puziy@ispe.kiev.ua (A.M. Puziy). the porous structure of synthetic phosphoric acid activated carbons has been investigated using a modi cation of the recently proposed BET±Kelvin method [4] applied to nitrogen adsorption isotherms at 196 8C and carbon dioxide adsorption isotherms at 0 8C. The results are compared with the density functional theory (DFT) method. 2. Experimental 2.1. Samples Synthetic carbons activated with phosphoric acid were prepared according to a procedure described 0169-4332/02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0169-4332(02)00883-8
A.M. Puziy et al. / Applied Surface Science 200 (2002) 196±202 197 elsewhere [5]. Brie y, a porous chlormethylated and then sulfonated copolymer of styrene and divinylbenzene [6] was impregnated with 60% phosphoric acid to the desired ratio, dried in air at about 220 8C and then carbonized in a quartz reactor at a temperature of 400± 1000 8C for 30 min. The carbonization was done in a ow of argon (1 ml/min). After carbonization, the carbon was cooled down to room temperature in a ow of argon. The weight ratio of H 3 PO 4 to copolymer on dry basis was 0.75. To remove the excess of H 3 PO 4, the carbons after carbonization were extensively washed with hot water in a Soxhlet extractor until neutralph. Then the samples were dried in an oven at 110 8C. Samples are designated as ``SP'' followed by their respective carbonization temperatures (in 8C). 2.2. N 2 and CO 2 adsorption Information on the carbon pore structure was derived from nitrogen adsorption isotherms obtained at 196 8C on an ASAP 2010 apparatus (Micromeritics, USA) and from carbon dioxide adsorption isotherms obtained at 0 8C on a NOVA 1200 apparatus (Quantachrome, USA). Samples were outgassed at 250 8C overnight prior to all adsorption measurements. 3. The model Analysis of physical adsorption data has become a standard method for assessment of heterogeneity of solid adsorbents. The theoretical description of adsorption on heterogeneous adsorbents is almost exclusively interpreted by superposition of adsorption on independent homogeneous adsorption sites. This concept leads to a Fredholm integral equation of the rst kind: Z Y t P ˆ Y l P; w dw (1) O (local isotherm) describes the adsorption on homogeneous sites characterized by property w (pore size, energy, etc.). Since this distribution function is widely used for characterization of adsorbents, many methods have been developed for calculating the distribution function from adsorption data. Comprehensive reviews of the methods for solving the adsorption integralequation with respect to the distribution function may be found elsewhere [7±11]. Due to their potentialaccuracy and high resolving power, rigorous numericalmethods are widely used as advanced methods for assessment of the distribution function [11±14]. The major problems in application of numericalmethods involve the choice of local isotherm and solution of adsorption integral Eq. (1). A new simple method for characterization of porous materials was recently proposed [4]. In this method, local isotherms are calculated on the basis of n-bet equation [15] describing a molecular layering process followed by a pore lling mechanism represented by Kelvin equation. In the proposed method the pore-size distribution (PSD) is calculated by solving the adsorption integral Eq. (1) using non-negative least-squares method. The method provides fast means for calculating the local isotherms as compared to sophisticated methods, such as molecular dynamics, Monte Carlo or density functionaltheory (DFT). The method was claimed to be comparable with DFT method when applied to nitrogen adsorption data at 196 8C. In this research, we have used the proposed BET± Kelvin method with some modi cation for calculation of the local isotherms. The PSD was calculated by solution of adsorption integral equation using a generalpurpose constrained regularization method CON- TIN [11]. Here, we brie y outline the method. Brunauer and co-workers proposed the n-bet equation [15±17] for describing the molecular layering in pores of nite dimension: t ˆ t m Cx 1 x 1 nf =2 n=2 xn 1 nf 1 x n nf =2 n=2 x n 1 1 C 1 x Cf =2 x n Cf =2 C=2 x n 1 (2) where the desired distribution function, F(U), is to be estimated from experimentaladsorption data, Y t (P), of limited accuracy. The kernel function, Y l P; w, where x ˆ P/P 0 is the relative pressure, n ˆ 2t=w the number of layers that can be accommodated by one wall of pore, f is the capillary condensation constant
198 A.M. Puziy et al. / Applied Surface Science 200 (2002) 196±202 and t m ˆ v m =N 1=3 monolayer thickness is calculated from the liquid molar volume v m and Avogadro's number N. Adsorption in micropores is enhanced due to overlapping of potential from opposite walls. Nguyen and Do [4] suggested to account for the enhancement of adsorption in micropores by the use of BET C constant as a function of heat of adsorption, which depends on pore size: C ˆ A exp Q p l (3) RT where A is the preexponentialfactor, Q p is heat of adsorption and l is the heat of liquefaction. The heat of adsorption is calculated as the minimum of the solid± uid potential. In the present work, we used Steele's 10-4-3 potential [18] for rigid chemically homogeneous carbon slit pores. The size of adsorbed molecule s sf and energy parameter e sf are taken from [19]. Carbon/adsorbate well depth e sf and intermolecular diameter s sf are calculated using Berthelot mixing rules. The number of layers n and adsorbed lm thickness t in Eq. (2) are treated as statisticaland may take non-integer values. As the pressure of adsorbate increases, the adsorbed layer in the pore grows according to Eq. (2). For small pores merging of two adsorbed layers from opposite walls may occur. However, for larger pores the condensation occurs before layers can meet. The capillary condensation in slit pores is described by the Kelvin equation: w c ˆ t 2gv mcos Y (4) RT ln P=P 0 where w c is the criticalpore size where condensation occurs, g the surface tension and t the adsorbed lm thickness described by the n-bet equation. Then local isotherm describing adsorption in uniform pores of width w is Y ˆ 2t=w; for w > w c or 2t < w (5) 1; for w < w c or 2t > w Parameters of adsorption systems used in our calculations are listed in Table 1. 4. Results and discussion Pore-size distributions of synthetic phosphoric acid activated carbons are presented in Figs. 1±7. PSDs were calculated by the BET±Kelvin method using adsorption isotherms of nitrogen at 196 8C and adsorption isotherms of carbon dioxide at 0 8C. The results are compared with the standard DFT method supplied by Micromeritics as DFT Plus software [20]. Nitrogen adsorption isotherms were measured in wide rangeðfrom pressures as low as 10 6 P/P 0 up to saturation pressure. Thus, these isotherms cover all possible mechanisms of adsorption taking place in a wide range of poresðvolume lling for small pores and capillary condensation in larger pores (5±1200 AÊ in the present study). However, carbon dioxide adsorption isotherms were measured up 1 atm, which corresponds at 0 8C to relative pressure range of 0.0005± 0.03 P/P 0. This region covers only adsorption in micropores. At this point, the question arises as to Table 1 Parameters of adsorption systems Parameter N 2 at 196 8C CO 2 at 0 8C Heat of liquefaction, l (kj/mol) 5.59 25.1 Surface tension, g (N/cm) 8.5 4.62 Collision diameter, s sf (AÊ ) 3.599 3.67 Interaction energy, s sf (K) 50.75 73.93 Molar volume, v m (cm 3 /mol) 34.67 47.58 Capillary condensation constant, f 40 10,000 Fig. 1. Pore-size distributions of synthetic phosphoric acid activated carbon SP400.
A.M. Puziy et al. / Applied Surface Science 200 (2002) 196±202 199 Fig. 2. Pore-size distributions of synthetic phosphoric acid activated carbon SP500. Fig. 4. Pore-size distributions of synthetic phosphoric acid activated carbon SP700. the upper limit of micropores for carbon dioxide adsorption at 0 8C and relative pressure of 0.03. Although there are some evidences that carbon dioxide lls pores up to 9±10 AÊ at such conditions [21±23], in our calculations we adopt an upper limit of micropores of 20 AÊ. This is because the CONTIN method uses all local isotherms with adsorption greater than zero to t the experimentaldata. According to BET±Kelvin modelwith parameters listed in Table 1, the largest pore size which has adsorption greater than zero at 0.03 P/P 0 is about 20 AÊ. That is why upper limit of micropores was set to 20 AÊ. This does not mean that 20 AÊ pores are completely lled with carbon dioxide, but only that adsorption at 0.03 P/P 0 is greater than adsorption on a at surface. In other words, partial lling of 20 AÊ pores at 0.03 P/P 0 is a signi cant indication of existence of such pores. For this reason it was possible to calculate PSD only in the range from 5 to 20 AÊ. Pore-size distributions of synthetic phosphoric acid activated carbons are similar to each other indicating Fig. 3. Pore-size distributions of synthetic phosphoric acid activated carbon SP600. Fig. 5. Pore-size distributions of synthetic phosphoric acid activated carbon SP800.
200 A.M. Puziy et al. / Applied Surface Science 200 (2002) 196±202 Table 2 Parameters of porous structure of synthetic phosphoric acid activated carbons obtained by the DFT method using N 2 at 196 8C adsorption isotherms a Carbon V mi S mi V tot S tot SP400 0.059 116 0.274 128 SP500 0.094 209 0.391 226 SP600 0.106 260 0.331 276 SP700 0.132 357 0.379 374 SP800 0.126 339 0.334 354 SP900 0.167 457 0.486 479 SP1000 0.134 337 0.468 361.13 a Volumes in cm 3 /g; areas in m 2 /g. Fig. 6. Pore-size distributions of synthetic phosphoric acid activated carbon SP900. Fig. 7. Pore-size distributions of synthetic phosphoric acid activated carbon SP1000. that the porous structure is already formed at a temperature as low as 400 8C. Both PSDs obtained by the BET±Kelvin and by the DFT method using nitrogen adsorption isotherms at 196 8C show a developed micropore structure (from 5 to 20±30 AÊ ) as well as a developed mesopore structure (from 200 to 1200 AÊ ). All distributions obtained by both BET±Kelvin and DFT methods using two different adsorbates show a bimodalstructure of micropores for a lcarbons. It should be noted that the bimodal distribution of micropores with a minimum near 10 AÊ is often observed for carbon adsorbents and is probably an artifact introduced by modeling assumptions [24]. What these two models have in common is that they use the rigid, homogeneous graphite-based slit pore model described by Steele's widely used 10-4-3 formula. Taking into account the heterogeneity of pore walls (edge effects, closed sides, etc.) would probably affect the resulting PSDs. The data presented in Figs. 1±7 show that the BET± Kelvin method is in good agreement with the DFTone. These gures also show that PSDs obtained by the BET±Kelvin method using different adsorbates (nitrogen at 196 8C and carbon dioxide at 0 8C) are in close agreement with each other. The agreement between different methods is better for small micropores. Parameters of PSDs obtained by the different methods are presented in Tables 2±4. Changes of these parameters with temperature are shown in Figs. 8 and 9. It follows from these results that a common Table 3 Parameters of porous structure of synthetic phosphoric acid activated carbons obtained by the BET±Kelvin method using N 2 at 196 8C adsorption isotherms a Carbon V mi S mi V tot S tot SP400 0.104 180 0.430 204 SP500 0.154 288 0.603 319 SP600 0.154 336 0.453 358 SP700 0.186 434 0.505 457 SP800 0.171 400 0.440 418 SP900 0.246 557 0.664 587 SP1000 0.198 429 0.619 458 a Volumes in cm 3 /g; areas in m 2 /g.
A.M. Puziy et al. / Applied Surface Science 200 (2002) 196±202 201 Table 4 Parameters of porous structure of synthetic phosphoric acid activated carbons obtained by the BET±Kelvin method using CO 2 at 0 8C adsorption isotherms a Carbon V mi S mi SP400 0.116 232 SP500 0.175 340 SP600 0.196 428 SP700 0.208 502 SP800 0.233 531 SP900 0.244 556 SP1000 0.298 638 a Volumes in cm 3 /g; areas in m 2 /g. trend is observed with some deviationsðpore volumes (V mi, V tot ) and surface areas (S mi, S tot ) increase with the increase in heat treatment temperature. Deviations from the generaltrend are observed for data obtained from nitrogen adsorption isotherms. This may be due to restricted diffusion of nitrogen at low temperature. The total pore volume obtained by BET±Kelvin from nitrogen adsorption isotherms is systematically higher than that obtained by the DFT method (see Fig. 8) because the molar volume of nitrogen is variable in DFT while in BET±Kelvin it is constant. The contribution of micropores to the totalporosity increases with the increase of temperature treatment up to 800 8C and then decreases. Fig. 9. Temperature dependence of totalsurface area of synthetic phosphoric acid activated carbons. The temperature dependence of surface area and micropore volume is different from other carbonaceous precursors where maximum development of surface area was observed at 450 8C for coconut shell [1] and subbituminous coal [2], 350±500 8C for white oak [2], and 500 8C for bituminous coal [2]. This is probably due to lower reactivity of polymer precursor than lignocellulosic materials or coals. Although the total pore volume steadily increases with the increase of heat treatment temperature the contribution of the micropore volume to the total porosity increases up to 800 8C and then decreases (see Fig. 8. Temperature dependence of totalpore volume of synthetic phosphoric acid activated carbons. Fig. 10. Temperature dependence of micropore share of synthetic phosphoric acid activated carbons.
202 A.M. Puziy et al. / Applied Surface Science 200 (2002) 196±202 Fig. 10). This offers some possibilities of altering the micro/mesoporous structure of the carbon adsorbent. 5. Conclusions Results demonstrate that the BET±Kelvin method is in good agreement with DFTand provides a fast means for assessment of porous structure of adsorbents. PSDs obtained by the BET±Kelvin method using different adsorbates give results consistent with each other. Due to the restricted pressure range for carbon dioxide adsorption isotherms the PSD gives information only about micropores in the range from 5 to 20 AÊ. The agreement between different methods is better for small micropores. Acknowledgements This research was made possible in part by NATO Collaborative Linkage Grant EST.CLG 974769 and by support from CICYT (project 1FD1997-1915) and Ministry of Education and Science of Ukraine (project 2M/117-2001). References [1] M. Jagtoyen, F. Derbyshire, Carbon 31 (1993) 1185. [2] J. Laine, A. Calafat, M. Labady, Carbon 27 (1989) 191. [3] A.M. Puziy, O.I. Poddubnaya, A. MartõÂnez-Alonso, F. SuaÂrez- GarcõÂa, J.M.D. TascoÂn, Carbon'2000, Berlin, Germany, 2000, pp. 657±658 (extended abstracts). [4] C. Nguyen, D.D. Do, Langmuir 15 (1999) 3608. [5] O.M. Puziy, O.I. Piddubna, M.T. Kartel, Phosphorus containing carbonaceous cation exchanger and process for production thereof, Ukrainian patent 42910A, 1998, bulletin 10. [6] N.T. Kartel, A.M. Puziy, Carbon'96, Newcastle-upon-Tyne, UK, 1996, pp. 525±526. [7] M. Jaroniec, R. Madey, PhysicalAdsorption on Heterogeneous Solids, Elsevier, Amsterdam, 1988. [8] M.M. Nederlof, W.H. Van Riemsdijk, L.K. Koopal, J. Colloid Interface Sci. 135 (1990) 410. [9] W. Rudzinski, D.H. Everett. Adsorption of Gases on Heterogeneous Surfaces, Academic Press, New York, 1992. [10] G.F. Cerofolini, N. Re, Riv. Nuovo Cimento Soc. Ital. Fis. 16 (1993) 1. [11] A.M. Puziy, T. Matynia, B. Gawdzik, O.I. Poddubnaya, Langmuir 15 (1999) 6016. [12] M. Von Szombathely, P. BrauÈer, M. Jaroniec, J. Comput. Chem. 13 (1992) 17. [13] L.K. Koopal, C.H.W. Vos, Langmuir 9 (1993) 2593. [14] J. Jagiello, Langmuir 10 (1994) 2778. [15] S. Brunauer, The Adsorption of Gases and Vapours, Oxford University Press, London, 1945. [16] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. [17] S. Brunauer, L.S. Deming, W.E. Deming, E. Teller, J. Am. Chem. Soc. 62 (1940) 1723. [18] W.A. Steele, The Interaction of Gases with Solid Surfaces, Pergamon Press, Oxford, 1974. [19] R.C. Reid, J.M. Prausnitz, B.E. Polling. The Properties of Gases and Liquids, McGraw-Hill, New York, 1987. [20] J.P. Olivier, J. Porous Mater. 2 (1995) 217. [21] D. Cazorla-AmoroÂs, J. AlcanÄiz-Monge, A. Linares-Solano, Langmuir 12 (1996) 2820. [22] D. Cazorla-AmoroÂs, J. AlcanÄiz-Monge, M.A. de la Casa- Lillo, A. Linares-Solano, Langmuir 14 (1998) 4589±4596. [23] P.I. Ravikovich, A. Vishnyakov, R. Russo, A.V. Neimark, Langmuir 16 (2000) 2311. [24] J.P. Olivier, Carbon 36 (1998) 1469.