Activation Energy for Oxygen Chemisorption on Carbon at Low

that the activation energy for chemisorption is not constant and varies with surface coverage. This observation can be explained by a distributed kine...
0 downloads 0 Views 59KB Size
292

Ind. Eng. Chem. Res. 1999, 38, 292-297

GENERAL RESEARCH Activation Energy for Oxygen Chemisorption on Carbon at Low Temperatures Hsisheng Teng* and Chien-To Hsieh Department of Chemical Engineering, National Cheng Kung University, Tainan 70101, Taiwan

The kinetics of oxygen chemisorption on resin chars were investigated in this study. The Elovich equation was employed to facilitate the interpretation of the chemisorption process. It was found that the activation energy for chemisorption is not constant and varies with surface coverage. This observation can be explained by a distributed kinetic parameter model, here implemented as a distributed activation energy model. The energy distribution model reveals that the amount of mass uptake obtained in a typical chemisorption study does not involve full coverage of all of the active sites on carbon. It is concluded that the information on active sites obtained from low-temperature chemisorption cannot be directly applied to gasification at higher temperatures. Introduction Gasification of carbons in oxidizing gases has been extensively studied over the past 40 years; e.g., see reviews of work on O2, CO2, and H2O by Laurendeau (1978) and on NO by Teng et al. (1992). It is generally agreed that carbon gasification by the above species involves, first, their chemisorption on the carbon to form carbon oxide surface complexes. This is followed by desorption of the surface complexes as carbon monoxide or dioxide. In studies of actual gasification, because both chemisorption and desorption occur simultaneously on the carbon surface, separate characterization of each individual step in the process is difficult. Transient or temperature-programmed desorption of the oxide surface complexes has been widely used to investigate the kinetics of desorption processes separately from the gasification process of interest (e.g., Du et al., 1990; Radovic et al., 1991; Hu¨ttinger and Nill, 1990; Suuberg et al., 1990; Teng et al., 1992), and chemisorption of gases on carbon at low temperatures (at which condition desorption of surface complexes is negligible) has been widely used to explore this aspect of the process (e.g., Laine et al., 1963; Bansal et al., 1970; Waters et al., 1986; Khan, 1990; Floess et al., 1991). The present study is concerned with the interpretation of such oxygen chemisorption experiments. In the description of the heterogeneous reactions of gases with solids, including carbons, the reaction rate is generally assumed to be proportional to the accessible surface area of the solid. However, it has long been recognized (Laine et al., 1963) that the total surface area derived from physical adsorption does not reveal the actual number of reactive sites: there are many carbon atoms, such as those in the interior of a large, fusedring aromatic system, that are reactive under only extreme conditions. The proportionality between the * To whom correspondence should be addressed. Tel: 886-6-2385371. Fax: 886-6-2344496. E-mail: hteng@ mail.ncku.edu.tw.

accessible surface area and the reactivity is not destroyed, provided that the correlation is applied under the circumstance that the materials employed have a constant number of reactive sites per unit surface area and the sites have the same distribution of activity. Laine et al. (1963) applied a low-temperature oxygen chemisorption technique to determine the active surface area of a carbon. This classic study has led to the widespread use of oxygen chemisorption capacity for characterization of the “active sites” in carbons. The large number of attempts at developing correlations of reactivity with a number of active sites have been discussed in many later studies (e.g., Khan, 1987; Suuberg et al., 1989; Khan et al., 1990). The active surface area, determined from a consideration of the total oxygen chemisorption capacity alone, has been found to be only a crude correlative parameter for carbon reactivity at high temperatures. This comes as little surprise because the ability of a site to chemisorb oxygen does not ensure its ability to readily desorb oxide products. There are quite likely several different types of chemisorption sites within any carbon. Many studies have explicitly addressed the possibility of a variety of sites as determining the overall kinetics of the chemisorption process. One study explicitly identified “steps” corresponding to filling of different types of sites (Bansal et al., 1970). More typically, the chemisorption process is observed to follow the so-called Elovich kinetic law:

rc ) dq/dt ) b exp(-aq)

(1)

where rc is the rate of oxygen chemisorption, q is the total uptake of oxygen on the surface, t is the chemisorption time, and a and b are constants. The general explanation for this form of kinetic law involves a variation of the energetics of chemisorption with the extent of surface coverage (Waters et al., 1986; Suuberg et al., 1989; Khan et al., 1990). Another plausible explanation can be that the active sites are heteroge-

10.1021/ie980107j CCC: $18.00 © 1999 American Chemical Society Published on Web 12/12/1998

Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 293

neous in nature and, therefore, exhibit different activation energies for chemisorption. The objective of this paper is to further elucidate the kinetics of oxygen chemisorption on carbons, based on a number of new experimental analyses of the process as it occurs on a relatively pure microporous carbon derived from a phenol-formaldehyde resin. The approach will lead to an explicit activation energy distribution for the oxygen chemisorption process on this char. Experimental Section A standard thermogravimetric analyzer (TGA; Perkin-Elmer TGA 7) was used for the present study. Experiments were performed under an oxygen/helium flow of 40 cm3/min at 101 kPa total pressure. Pulverized char samples (30-50 mg) were placed in a sample pan suspended in the heated zone of the TGA. A small thermocouple probe (type K) placed within a few millimeters of the sample pan served to indicate the sample temperature. To minimize catalytic impurities, the chars used in the present study were derived from phenol-formaldehyde resins. The resin char was prepared by pyrolysis of the phenol-formaldehyde resin in a helium environment at 1223 K for 2 h and then ground and sieved to the desired particle size (200-300 µm). The surface of the char was cleaned of oxides prior to O2 chemisorption experiments by heating the sample to 1173 K in ultrahigh-purity helium for at least 1 h. Oxygen chemisorption experiments were performed immediately after surface cleaning. The sample temperature was quickly lowered from 1173 K to the desired chemisorption temperature, and then the desired O2/ He stream was quickly introduced. The chemisorption experiments in this study were carried out at an O2 partial pressure of 50.5 kPa and at temperatures ranging from 373 to 473 K. Oxygen uptake during chemisorption experiments was continuously monitored for at least 24 h, at which the rate of mass gain was negligible. Specific surface areas and pore volumes of the samples were determined by gas adsorption. An automated adsorption apparatus (Micromeritics, ASAP 2000) was employed for these measurements. Nitrogen adsorption at 77 K and CO2 adsorption at 273 K were measured. Before any such analysis, the sample was degassed at 573 K in a vacuum at about 10-3 Torr. The N2 and CO2 isotherms were analyzed by the BET equation and the Dubinin-Polanyi (DP) equation (Lowell and Shields, 1991), respectively, to determine the surface areas of the chars. Micropore volumes of the samples were determined from the application of the Dubinin-Radushkevich (DR) equation (Stoeckli, 1990; Carrott et al., 1991) to both of the N2 and CO2 adsorption isotherms. In type I isotherms (Lowell and Shields, 1991), the amount of N2 adsorbed at pressures near unity corresponds to the total amount adsorbed at both micropores (filled at low relative pressures) and mesopores (filled by capillary condensation at relative pressures above 0.2); consequently, the subtraction of the micropore volume (from the DR equation) from the total amount (determined at p/p0 ) 0.98 in this case) will provide the volume of the mesopore (Rodrı´guez-Reinoso et al., 1995). The adsorption isotherms are employed to deduce the BET surface area and the micropore and mesopore volumes. The average pore diameter can be determined

Table 1. Surface Characteristics of Phenol-Formaldehyde Resin Char with an Oxygen Burnoff Level of 20% N2 adsorption

CO2 adsorption

BET SA (m2/g)

micropore volume (cm3/g)

mesopore volume (cm3/g)

average pore diameter (Å)

DP SA (m2/g)

micropore volume (cm3/g)

610

0.287

0.003

19

510

0.186

according to the surface area and total pore volume (the sum of the micropore and mesopore volumes) if the pores are assumed to be cylindrical and have no intersection. Results Surface Structures of the Resin Char. In an effort to avoid any experimental artifacts due to differences in the nature or concentration of active sites caused by differing extents of burnoff, all of the resin chars used for the chemisorption experiments have similar extents of oxygen burnoff (about 20%). It has been shown that the chemisorption behavior can be sensitive to burnoff (Suuberg et al., 1989). N2 adsorption at 77 K was employed to characterize the physical structures of the char surface. However, it has long been recognized that if the micropores are very narrow, the entry of N2 molecules at 77 K may be kinetically restricted (Garrido et al., 1987). To avoid this problem, CO2 adsorption at a higher temperature (273 K) was also employed for surface characterization. The surface characteristics of the resin char at an oxygen burnoff level of 20% are shown in Table 1. The data show that the char is mainly microporous. The surface area and micropore volume determined from N2 adsorption are found to be larger than those determined from CO2 adsorption, indicating that there is no problem of restricted diffusion of N2 at 77 K at this extent of burnoff and the microporosity is relatively wide, because volume filling for CO2 adsorption at higher temperatures (273 K here) does not occur in larger pores (Ballal and Zygourakis, 1987). Under this circumstance, the area determined from the N2 adsorption was employed to represent the specific surface area of the resin chars used in the present study. Elovich Kinetic Analysis of Oxygen Chemisorption. The results of oxygen chemisorption experiments on the resin char are shown in Figure 1. Because chemisorptions were performed at temperatures (373473 K) which are low enough that desorption of surface oxides from this type of carbon can be neglected (Suuberg et al., 1988), the amount of mass increase during chemisorption is equivalent to the mass of oxygen in oxides formed on the surface. The total amounts of mass uptake for 24 h of chemisorption are shown in Table 2. It can be seen from Figure 1 that, as usual, chemisorption is initially fast and slows down markedly with increasing surface coverage; i.e., the results follow the usual type of Elovich law. After a long period of chemisorption (24 h), the rate of chemisorption becomes very low and the total amount of mass uptake varies little with time, although there is never a very distinct “endpoint” to the chemisorption, even at much longer times. The data of Figure 1 have been fit to the Elovich equation, eq 1, and the values of the parameters a and b are also shown in Table 2. The solid lines in Figure 1 represent the simulated data using the Elovich parameters determined. As seen in the figure, the simulated

294 Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999

Figure 1. Oxygen chemisorption on resin char in 50.5 kPa oxygen partial pressure at 373 (0), 423 (b), and 473 ([) K for 24 h. Solid lines represent simulated data from an evaluation using the Elovich parameters shown in Table 2. Table 2. Mass Uptake and Elovich Parameters for Oxygen Chemisorption on Resin Chars temperature (K) 24 h mass uptake (µg/m2) a (m2/µg) b (µg/m2 h)

373

423

473

10.4 0.63 88

13.1 0.50 99

15.8 0.44 120

mass uptake curves enjoy satisfying agreement with the experimental data for most of the coverage. The parameter b represents the rate of chemisorption at zero coverage (i.e., q ) 0). It is seen to increase with increasing temperature. There are some difficulties in actually measuring the rapid rate of mass gain very near the instant at which oxygen is introduced to the sample. This is apparent in plots such as Figure 1, from the steepness of the curves at times near zero. Thus, it is difficult to actually independently determine the zero time rate, and the value of b is obtained from the fitting of the overall mass gain curve. Nevertheless, the trend of b with temperature is reasonable. The activation energy implied by this trend is 4.4 kJ/mol. To confirm the correctness of this temperature dependence, a separate plot showing the mass uptake for short times (0-20 min) is presented in Figure 2. On the basis of the data in Figure 2, the average rate of mass uptake in the initial 2 min for chemisorption at different temperatures was determined, and the temperature dependence of the initial rate gave an apparent activation energy of 5.6 kJ/mol, which was fairly close to the value predicted by the b values. The quantity of oxygen mass uptake at long exposure times has often been used to determine the number of active sites on the carbon surface (e.g., Laine et al., 1963; Khan, 1987; Suuberg et al., 1989). Figure 1 shows very clearly the problem involved in this procedure. The amount of mass uptake, following 24 h of chemisorption, increases with chemisorption temperature. This demonstrates the well-known problem that the number of active sites on carbon is not well-defined by the procedure. However, it can be argued that the rate of chemisorption at lower temperatures is too low to reach site saturation in 24 h and that all of the sites could be filled up if the chemisorption time were long enough. This argument implies a system with distributed kinetic

Figure 2. Initial stage (0-20 min) of oxygen chemisorption on resin char in 50.5 kPa oxygen partial pressure at 373 (0), 423 (b), and 473 ([) K.

behavior. With high enough activation energies for the slow chemisorption steps, the process would, of course, not reach an endpoint in a reasonable amount of time, except if the temperature is increased. This strategy fails in that desorption processes set in before a temperature can be reached, which would ensure that all sites are filled. It can be seen from Table 2 that the value of a is a decreasing function of the chemisorption temperature in oxygen. This was reported earlier by others (Taylor and Thon, 1952; Allardice, 1966). A decreasing value of a with increasing chemisorption temperature implies an increasing apparent activation energy for chemisorption with increasing extent of surface coverage, as may be seen from the following:

Ec/R ) -[∂ ln(rc)/∂(1/T)]q ) -d(ln b)/d(1/T) + [da/d(1/T)]q ) 533 + 347q where Ec is the apparent activation energy for chemisorption at a surface coverage of q and R is the universal gas constant. The values of a and b are taken from Table 2. For a typical range of coverages from q ) 0 to 10 µg/ m2 (see Table 2), the activation energy rises from about 4.4 to about 33 kJ/mol, as reflected by the above equation. Such values are in the range typically reported for chemisorption. Note that this analysis made no allowance for the path dependence of site filling; i.e., at different temperatures, sites with different activation energies will fill with different relative rates to different extents, and thus the state of the surface is not fully described by the parameter q alone. The activation energy is thus an apparent activation energy. A more satisfactory modeling approach will be presented below. Distributed Activation Energy Model for Chemisorption. The fact that the chemisorption activation energy increases with increasing extent of surface coverage, observed in the present study and others (Suuberg et al., 1989; Khan et al., 1990), still remains only qualitatively understood. Use of explicit activation energy distribution models (as opposed to the Elovich law) has been made by several workers in the literature. Waters et al. (1986) found that Saran char could be characterized by a continuous distribution of activation energies for heterogeneous adsorption sites capable of oxygen chemisorption. Floess et al. (1991) have devel-

Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 295

oped a distributed activation energy model for oxygen chemisorption and Du et al. (1991) proposed a Gaussian form of activation energy distribution for oxygen adsorption on carbons, through mass balance calculations involving formation and desorption of surface oxides during pseudo-steady-state gasification of carbon in oxygen. Here, we seek to maintain generality in the allowable forms of the distribution function. The approach is thus initially similar to that of Floess et al. (1991), except that in their particular system the distribution function with respect to chemisorption activation energy was taken as flat. To keep the model to be developed here mathematically tractable, the following assumptions were made: I. Each chemisorption site on the char surface possesses a single value of activation energy for oxygen chemisorption, and the total number of chemisorption sites is constant. II. Each chemisorption site can be occupied by one oxygen atom to form an intermediate complex, C(O), on the carbon surface. III. The energy distribution for oxygen chemisorption is continuous. IV. At low temperatures (573 K) is much higher than that desorbed from chars oxidized at chemisorption temperatures (