Influence of the Particle Size of Activated Carbons on Their

Jun 14, 2012 - Synthesis of Magnetic Cobalt Nanoparticles Anchored on Graphene Nanosheets and Catalytic Decomposition of Orange II. Yunjin Yao , Chuan...
12 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/IECR

Influence of the Particle Size of Activated Carbons on Their Performance as Fe Supports for Developing Fenton-like Catalysts Filipa Duarte,† F. J. Maldonado-Hódar,‡ and Luis M. Madeira*,† †

LEPAE, Department of Chemical Engineering, Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal ‡ Department of Inorganic Chemistry, Faculty of Sciences, University of Granada, 18071 Granada, Spain ABSTRACT: In this work, a commercial activated carbon (AC) impregnated with iron (7 wt % Fe on Fe/AC) was used as catalyst for the removal of the azo dye Orange II (OII) by the heterogeneous Fenton-like process. The influence of the particle size of the AC support on adsorption and catalytic runs was evaluated, using different particle sizes below commercial pellet form (cylinders of ca. 3 × 5 mm). The materials were characterized using several techniques (N2 and CO2 adsorption, XRD, and HRTEM). It was found that the porosity of the extruded AC was liberated by milling, enhancing the adsorption capacity and the adsorption rate. Nevertheless, total discoloration was achieved with the four particle sizes tested, albeit taking ca. 24 h using pellets but only 2 h using the powder, under the tested conditions. The dispersion of iron in the Fenton-like catalysts was also improved with the decrease in AC particle size, which also favored the catalytic activity. However, leaching increased in the same way. This tradeoff between activity and stability points to the intermediate size of 0.80−1.60 mm as the best choice. To analyze the effect of the support particle size on the catalytic performance alone, catalysts were prepared with nearly the same iron dispersion (keeping the loading of iron per unit of surface area constant). The smaller the support, the higher the activity, and thus, the higher the effectiveness factor, because of the competition between the internal diffusion and reaction rates.

1. INTRODUCTION The textile industry uses large amounts of chemical reagents, as well as millions of tons of water, dyes, and other substances. These wastewaters constitute the effluents that are released into the water cycle, so their treatment and recovery are imperative from both environmental and social points of view. Dyes are an important type of pollutant present in textile effluents, and their destruction is not easy. To overcome the limitations of conventional processes, which are unable to oxidize these types of products efficiently, advanced oxidation processes (AOPs) have been developed and shown to be an effective solution for the elimination of several organic pollutants from water.1,2 Known for its simplicity, efficiency, and low investment cost,3 Fenton’s reaction is one of the most promising AOPs. The Fenton process exploits the reaction between iron(II) and hydrogen peroxide for the formation of hydroxyl radicals, species that are able to attack and oxidize organic matter nonselectively. When dissolved iron species are used as the catalyst (the so-called homogeneous Fenton process), final treatments of the effluents, such as precipitation, are needed to avoid contamination by the metal. On this basis, the development of active and stable heterogeneous Fe-based catalysts for water treatment is an interesting alternative, and different approaches using different types of supports (e.g., clays, carbon materials, zeolites) appear in the literature.4−11 Activated carbons (ACs) are largely used for wastewater treatments, as either adsorbents or catalyst supports.12 A support for the development of heterogeneous Fenton catalysts should be stable and highly porous and should provide a good dispersion of the metal phase, with a low degree of leaching and accessibility to the active sites. Different aspects of activated © 2012 American Chemical Society

carbons, such as low price, high porosity and surface area, stability in acidic and basic media, and easily tunable surface chemistry, can favor the use of ACs as alternative to inorganic supports. In previous works,6,13 the authors studied different types of carbon materials and tested some transition metals, among which iron was revealed to be a good choice and Norit RX 3 Extra a suitable AC support. Runs were carried out in slurry batch reactors, with the powder catalyst (containing 7 wt % Fe) in suspension, enabling promising results to be obtained.6,13 Nevertheless, the economy of the treatment process will greatly improve with the development of continuous flow reactors, for which the use of powder catalysts is problematic. This study focused on the removal of the azo dye Orange II using the commercial activated carbon Norit RX 3 Extra, milled and sieved into four different particle sizes. The influence of the particle size (dp) of the AC support on adsorption and catalytic runs was evaluated, taking into account the fact that, in Fenton’s oxidation, both adsorption and reaction coexist. The AC particle size was found to influence not only competition between internal mass transfer and chemical reaction, but also iron dispersion and, consequently, leaching and catalytic performance. The Fe catalysts were therefore extensively characterized by different techniques, and their catalytic performance was related to chemical and physical properties. Received: Revised: Accepted: Published: 9218

January 18, 2012 June 12, 2012 June 14, 2012 June 14, 2012 dx.doi.org/10.1021/ie300167r | Ind. Eng. Chem. Res. 2012, 51, 9218−9226

Industrial & Engineering Chemistry Research

Article

Table 1. Textural Characteristicsa of the Support in the Four Particle Sizes Used support

SBET (m2·g−1)

L0(N2) (nm)

W0(N2) (cm3·g−1)

Smic(CO2) (m2·g−1)

L0(CO2) (nm)

W0(CO2) (cm3·g−1)

VBJH (cm3·g−1)

SBJH (m2·g−1)

N powder 0.25−0.80 mm 0.80−1.60 mm N pellets

1405 1230 1110 1113

1.70 1.66 1.64 1.56

0.62 0.52 0.48 0.47

1173 942 893 984

0.81 0.73 0.72 0.73

0.48 0.35 0.32 0.36

0.592 0.482 0.430 0.358

1653 1339 1206 942

a

SBET, BET surface area obtained by N2 adsorption; L0(N2), mean micropore size obtained by N2 adsorption; W0(N2), micropore volume obtained by N2 adsorption; Smic (CO2), micropore surface area obtained by CO2 adsorption; L0(CO2), mean micropore size obtained by CO2 adsorption; W0(CO2), micropore volume obtained by CO2 adsorption; VBJH, Barrett−Joyner−Halenda (BJH) pore volume; SBJH, BJH surface area.

textile effluents.16−18 The initial pH of 3 was set by using a 1 M sulfuric acid solution, a thermostatic bath (from Huber) was used to keep the temperature constant during the runs at 30 °C, and 6 mM of H2O2 was used; these are the conditions of reference for Fenton’s reactions optimized in previous works.5,6,13 Continuous stirring of the reaction mixture was guaranteed by a magnetic bar, and time zero for the experiments was considered to be coincident with the addition of the activated carbon (for adsorption) or iron-impregnated activated carbon + hydrogen peroxide (for catalysis). Discoloration was measured in terms of absorbance at 486 nm (characteristic wavelength of the Orange II molecule) using a Philips PU8625 UV/vis spectrophotometer and was recorded throughout the reaction time using a LabVIEW 9.0 interface. Continuous and online measurements were guaranteed by the use of a flow-through cell and recirculation of the reaction mixture with a peristaltic pump. During the catalytic tests, samples of 10 mL were regularly taken and filtered through Reeve Angel microfiber glass filter paper (pore diameter of 0.8 μm). Thereafter, total organic carbon (TOC) analyses were undertaken in a TOC-500A apparatus to evaluate the degree of mineralization during the reaction time; iron leaching (i.e., the amount of iron that left the catalyst and went into the solution) was quantified by atomic absorption in a UNICAM 939/959 spectrophotometer. As these analyses were not carried out immediately, an excess of sodium sulfite was added to the sampling flasks to stop the reaction, because this reagent consumes the residual hydrogen peroxide instantaneously,5,19 and after that, the samples were kept in a refrigerator.

2. MATERIAL AND METHODS 2.1. Catalyst and Support Preparation and Characterization. Norit RX 3 Extra (herein called sample N), commercialized as extruded pellets of approximately 3 × 5 mm, was milled and sieved to obtain particles in powder form ( W0(CO2) (Table 1), indicating the occurrence of N2 condensation inside the larger micropores and/or mesopores. Clearly, the accessible porosity and surface area increased with milling, that is, from pellets to powder samples. Thus, the micropore volume (W0), determined either by N2 or CO2 adsorption, progressively decreased with increasing particle size, and simultaneously, the mean micropore size (L0) also decreased, leading to a progressive reduction in the SBET and Smic(CO2) values. The variation in the mesopore volume determined by the BJH method also showed the same trend. 9219

dx.doi.org/10.1021/ie300167r | Ind. Eng. Chem. Res. 2012, 51, 9218−9226

Industrial & Engineering Chemistry Research

Article

indicated mainly by the decrease in the BET surface areas and the W0(N2) values from the supports to the catalysts (Table 2). Moreover, when the support was used as a powder, there was a strong decrease (of 0.12 ± 0.01 cm3·g−1) in the micropore volume obtained by either N2 or CO2 adsorption (Table 2). Nevertheless, when the pellets support was impregnated, W0(CO2) was almost fully maintained, whereas W0(N2) showed a decrease similar to that observed in the case of the powder. This means that the Fe particles will be formed within a different porosity range: in the case of the powder, the small particle size and the larger mean micropore size of the support can favor access of the Fe solution to a narrower porosity than in the case of the pellets. Therefore, the accessibility of the iron solution used for impregnation of the support porosity strongly depends on the AC particle size, and thus, the Fe distribution and dispersion changed significantly among the four catalysts, decreasing in line with the corresponding increase in the support particle size (Fe particles formed progressively on the outer surface). This is appreciable even at a glance, because, in the case of the catalyst prepared with pellets, the high iron concentration on the external surface of the pellets provoked a yellow-green hue in comparison with the black aspect of the other samples (Figure 3). The HRTEM images (Figure 4) also reveal a better

Figure 1. Variation of the N2 adsorption isotherms with the particle size of the AC support (solid symbols for adsorption; open symbols for desorption).

However, the mesopore size distribution was not affected, as can be concluded from Figure 2. The diminution of the pore volume and surface area by pelletization of carbon powder has previously been described by other authors.20

Figure 3. Photos of N−Fe catalysts (series A) in the four different sizes: (a) powder, (b) 0.25−0.80 mm, (c) 0.80−1.60 mm, and (d) pellets.

Figure 2. Pore size distributions of the AC supports determined by the BJH method.

dispersion of the iron particles in the case of the catalyst prepared from powder, with an Fe particle size below 20 nm, whereas in the case of the catalyst prepared from pellets, the iron particles were more heterogeneous and formed some particles larger than 200 nm. The XRD patterns also

After the Fe impregnation of the supports to obtain the corresponding catalysts, the shape of the N2 adsorption isotherms was basically maintained (data not shown). However, Fe impregnation produced a certain micropore blockage,

Table 2. Textural Characteristicsa of the Support in Powder and Pellet Forms and Their Corresponding Catalysts (Series A Samples)

N powder N−Fe powder N pellets N−Fe pellets

SBET

L0(N2) (nm)

W0(N2) (cm3·g−1)

Smic(CO2) (m2·g−1)

L0(CO2) (nm)

W0(CO2) (cm3·g−1)

1405 1160 1113 893

1.70 1.75 1.56 1.38

0.62 0.51 0.47 0.37

1173 758 984 923

0.81 0.89 0.73 0.70

0.48 0.35 0.36 0.33

a

SBET, BET surface area obtained by N2 adsorption; L0(N2), mean micropore size obtained by N2 adsorption; W0(N2), micropore volume obtained by N2 adsorption; Smic (CO2), micropore surface area obtained by CO2 adsorption; L0(CO2), mean micropore size obtained by CO2 adsorption; W0(CO2), micropore volume obtained by CO2 adsorption. 9220

dx.doi.org/10.1021/ie300167r | Ind. Eng. Chem. Res. 2012, 51, 9218−9226

Industrial & Engineering Chemistry Research

Article

Figure 5. XRD patterns of (a) series A catalysts pretreated in N2 flow at 300 °C and (b) raw FeSO4·7H2O used as the iron salt precursor.

initial particle size of the raw sulfate, influencing the transformation of the metastable β-Fe2O3 into α-Fe2O3. Moreover, α-Fe2O3 is also transformed into γ-Fe2O3 during the Fe2(SO4)3 reduction process through the formation of magnetite (Fe3O4) as an intermediate.23 Both γ-Fe2O3 and Fe3O4 present the same inverse spinel structure, with some cation vacancies in octahedral positions for the former. Finally, the XRD patterns of γ-Fe2O3 present different diffraction peaks according to the shape, dimensions (axial ratio), and order/ disorder of the iron nanoparticle defects.23 Thus, in our case, clearly a mixture of phases is always present. The small Fe particles found in the N−Fe powder mainly match the peaks attributable to lepidocrocite (JCPDS file 8-98) with I100 at 14.2° corresponding to the (020) diffraction. This phase also seems to be present using the pellets as support. The peak at 33.1° in the powdered catalyst can be assigned to I100 of hematite (JCPDS file 33-0664). All of the samples present a diffraction peak at around 2θ = 26°, this being the main peak when using intermediate sizes of the AC support. However, the relative intensity with regard to the remaining peaks undergoes significant changes between samples. This peak can be assigned to the I100 (310 diffraction) of the akaneite (β-Fe2O3) oxide−hydroxide (JCPDS file 13157), but might also be due to the diffraction (120) of lepidocrocite. The most intense XRD peaks in the N−Fe pellet catalyst are attributable to the maghemite (γ-Fe2O3) phase, with I100 corresponding to the (220) diffraction at around 30°. At 35.5°, a small peak appears that can be attributed to I100 (311) of magnetite (Fe3O4) (JCPDS file 19-629). To minimize these differences in terms of iron dispersion and the nature of the iron species that change from sample to sample, the series of samples denoted as B was prepared. In this case and as mentioned above, the iron loading decreased proportionally to the BET surface area of each support fraction,

Figure 4. HRTEM images of N−Fe in (a) powder and (b) pellet forms; series A samples.

demonstrate this effect because the crystallinity of the Fe phases increased progressively with the support particle size, with the peaks of the iron phases becoming noticeably more intense (Figure 5a). However, the XRD patterns plotted in Figure 5a not only show an increase in the Fe particle size, but also a change in the nature of the Fe phases. Clearly, comparing the XRD patterns of the pretreated catalysts (Figure 5a) with those obtained for the raw FeSO4·7H2O used as precursor (Figure 5b), it is possible to conclude that the precursor salt was always decomposed. However, the assignment of the XRD diffraction peaks observed for the catalysts is a hard task because of the different oxides and allotropic forms. Major iron oxides include hematite (α-Fe2O3), maghemite (γ-Fe2O3), goethite (αFeOOH), lepidocrocite (γ-FeOOH), wustite (FeO), and magnetite (Fe3O4). Beta (β-Fe2O3) and epsilon (ε-Fe2O3) iron(III) oxides are less frequent polymorphs. Moreover, both are unstable, forming hematite as a product of their transformations. The thermal decomposition of iron sulfates was previously studied. According to Tagawa,21 the decomposition of hydrated FeSO4 does not yield anhydrate, but rather yields iron(III) oxide through previous oxidation to Fe2(SO4)3, although the crystal system of Fe2O3 was not specified. Zboril et al.22 identified the trigonal-hexagonal α-Fe2O3 and cubic βFe2O3 as Fe2(SO4)3 decomposition products. Nevertheless, the ratio between the two phases depends on the experimental conditions such as the decomposition temperature and the 9221

dx.doi.org/10.1021/ie300167r | Ind. Eng. Chem. Res. 2012, 51, 9218−9226

Industrial & Engineering Chemistry Research

Article

Figure 6. OII removal by adsorption using N supports (solid symbols) and N−Fe catalysts (open symbols) with different particle sizes, series A samples (COII= 0.1 mM, Csolid= 0.1 g·L−1, pH = 3, T = 30 °C).

summarized in Figure 7. The catalyst prepared by impregnation of N powder is therefore the most efficient in terms of the dye

that is, from 7 wt % for the powder to 5.5 wt % for the pellets, aiming keeping constant the iron dispersion. The XRD patterns of these samples are similar and coincident with the one shown in Figure 5a for the N−Fe powder catalyst. These results evidence that the differences in the metal dispersion and even the different phases formed during heat treatments in samples of series A could be a consequence of the progressively worse accessibility of the Fe solution to the AC porosity with increasing the AC particle size. For series B, all materials had nearly the same iron dispersion, and the nature of the iron oxides present was also similar; thus, only the support particle size changed from sample to sample. 3.2. Elimination of Orange II from the Solution. 3.2.a. Adsorption Experiments. The elimination of OII from solution by the heterogeneous Fenton-like process using this kind of porous sample is complex, with the coexistence of adsorption and reaction phenomena. Thus, adsorption experiments were first carried out under the same conditions as the catalytic tests (but in the absence of H2O2). The adsorption rate and the amount adsorbed after 24 h showed a significant decrease with increasing particle size (Figure 6). Only the support in powder form was able to decolorize the solution completely during this period, in fact after approximately 6−8 h. The evolution of the OII adsorption performance is in agreement with the smaller and narrower microporosity/ mesoporosity observed with the increase in the support particle size (Table 1). The pore blockage by Fe particles justifies the lower adsorption rate and capacity of the catalysts regarding their supports (cf. Table 2 and Figure 6). This effect is more evident when the AC particle size is being reduced, because, as mentioned previously, micropores are blocked to a large(r) extent by the greater accessibility of the Fe solution in the smaller particles. 3.2.b. Catalytic Fenton-like Oxidation Experiments. The catalytic behaviors of the four prepared catalysts were determined by their different porosities and the natures and dispersions of the iron species. During the heterogeneous process, the reactants should be adsorbed on the catalyst surface, which varies markedly depending on the support particle size, as described previously. Although the process is quite complex and is determined by several factors, we have already demonstrated the importance of the iron distribution and dispersion in the catalytic performance of Fe catalysts in previous studies.6,13 This clearly justifies the catalytic behavior

Figure 7. OII removal by heterogeneous Fenton reaction using the N−Fe catalyst in the four particle sizes, series A samples (COII= 0.1 mM, CH2O2= 6 mM, Csolid= 0.1 g·L−1, pH = 3, T = 30 °C).

oxidation rate, although the differences in discoloration ability are smaller between the three catalysts with smaller particle sizes than in the adsorption experiments (Figure 6), confirming that, under these conditions, the OII removal is due mainly to catalytic oxidation and not to the adsorption process. The poor ability of the pellets to disperse the Fe species is therefore the key factor with regard to the significant differences reported. The three catalysts with smaller support sizes decolorized the solution in 2 h, whereas the conversion obtained at this reaction time using the catalyst in pellet form was almost negligible. Nevertheless, total discoloration was also obtained in this case, but only after 24 h of reaction. In addition to dye degradation, quantification of the mineralization degree (i.e., level of oxidation) is also important. The TOC analyses (Figure 8a) showed that a large proportion of the degradation products remained in solution after 24 h, indicating a high stability of the intermediates/degradation products (i.e., a high refractory character) and also that they were not adsorbed. Once again, the most effective catalyst was 9222

dx.doi.org/10.1021/ie300167r | Ind. Eng. Chem. Res. 2012, 51, 9218−9226

Industrial & Engineering Chemistry Research

Article

Figure 8. (a) Mineralization degree and (b) leaching values achieved during 24 h of reaction for the different series A samples (experimental conditions as in Figure 7).

the most well dispersed one: N−Fe in powder form. However, leaching also presented the same trend (Figure 8b), and the less stable catalyst therefore corresponded to that supported on the powder, compromising its reuse. In fact, smaller Fe particles (better dispersion) lead to more active catalysts, but also favor Fe leaching. Nevertheless, this parameter can be improved, for instance, by using another iron salt precursor or more drastic thermal treatments, which will be the subject of future study. In any case, the final iron concentration found in solution was just slightly above the European Legislation values (2 ppm)24 in the worst cases and below that threshold for the supported catalysts with dp > 0.80−1.60 mm. Of course, if reaction is not prolonged for 24 h, iron leached into solution is even lower. The levels of dissolved iron increased progressively over time, as leaching from the supports proceeded. Homogeneous experiments performed with a ferrous salt employing the maximum values reported in Figure 8b (i.e., after 24 h of reaction) as the initial loading provided a TOC removal degree ca. 20% smaller, showing that the homogeneous contribution cannot be discarded and that, under such conditions, both processes are important. However, at the early stages of reaction, the process is exclusively heterogeneous because there is no iron in solution. A suitable balance between dye degradation, TOC removal, and leaching has to be made in the choice of the best particle size. The results suggest that intermediate sizes can be a good option, particularly in the case of the catalyst with 0.80 < dp < 1.60 mm, which also complies with the legislated value for iron in the final effluent. The catalytic performance of heterogeneous Fenton catalysts is related to their ability to decompose H2O2 molecules into hydroxyl radicals. For that purpose, a colorimetric method25 was used for the determination of the hydrogen peroxide concentrations remaining in solution after 4 and 24 h of reaction without dye. The H2O2 conversions are compiled in Table 3 and show an increase in H2O2 consumption with a decrease in dp due to the improvement in iron dispersion, which, together with the adsorptive behavior of the samples previously described, justifies the differences in the performances of the catalysts toward OII elimination. The reported decrease in the discoloration rate with increasing particle size (dp) might be due to the higher resistance to mass transfer (diffusion of the dye molecules within catalyst pores). To obtain the kinetic and mass-transport parameters, of importance for reactor design (e.g., in packed-

Table 3. H2O2 Conversions Achieved Using the Four Catalysts (Series A Samples) after 4 and 24 h of the Fenton Reaction XH2O2 (%) N−Fe N−Fe N−Fe N−Fe

powder 0.25−0.80 mm 0.80−1.60 mm pellets

after 4 h

after 24 h

91 58 28 14

99 99 84 23

bed applications), the effectiveness factors η were calculated as26,27 r η = obs rv,s (1) where robs is the observed reaction rate and rv,s stands for the true reaction rate, under catalyst surface conditions (both in mol·s−1·gcat−1). It is noteworthy that oxidation experiments with particles of dp < 0.10 mm, carried out under the same conditions as for Figure 7, confirmed that the catalyst with dp < 0.15 mm (powder sample) is in chemical regime, because hardly any differences for OII concentration histories were detected (data not shown). Thus, in this case, there are no internal resistances to mass transfer, so that η = 1 (robs = rv,s), and thus kobs = kv, where kobs and kv are the observed and true rate constants for dye consumption, respectively. To determine the observed rate constants and the apparent reaction orders (nobs) with respect to the dye for each run, the differential method27 was used for experiments carried out with the three catalysts of smaller particle sizes. Pellets were excluded from this study because of their extremely low reaction rates, particularly at the beginning of the experiments, making the analysis for the extraction of significant parameters difficult. A power rate law was thus assumed (r = kCOIIn), and from the logarithmic plot of robs versus COII, the observed reaction order and kinetic constant were computed, as well as the initial robs value (based on the initial dye concentration). Table 4 lists the results obtained, wherein the reported rate constants include the radical concentrations, which are assumed to be constant (pseudo-steady-state approach). 27,28 As expected, there was a significant decrease in the observed rate constants, as well as in the initial rates, with the increase in particle size. For the powder sample, which was in the chemical regime, the effectiveness factor was 1, and a reaction order of 1.4 was obtained. For the larger particles, which were in 9223

dx.doi.org/10.1021/ie300167r | Ind. Eng. Chem. Res. 2012, 51, 9218−9226

Industrial & Engineering Chemistry Research

Article

Table 4. Kinetic Parameters of the Three Samples with Smaller Particle Sizes (Series A Samples) dpa (cm)

N−Fe sample powder 0.25−0.80 mm 0.80−1.60 mm

kobs [(mol·g−1·s−1)·(cm3/mol)n]

−2

1.4 × 10 6.2 × 10−2 1.2 × 10−1

nobs

1.0 × 10 8.4 × 100 3.6 × 10−1 4

1.4 1.0 0.9

initial robs (mol·g−1·s−1) −6

1.6 × 10 5.2 × 10−7 2.9 × 10−7

η

ϕg

De (cm2·s−1)

De/r2 (s−1)

1.00 0.33 0.18

− 3.0 5.4

3.2 × 10−8 3.8 × 10−8

− 3.3 × 10−5 1.0 × 10−5

a

Average particle size diameters of powder and 0.25 < dp < 0.80 mm samples were measured with an LS 230 laser diffraction particle size analyzer from Beckman Coulter; for the sample with 0.80 < dp < 1.60 mm, the reported value is an average of the extreme values of dp.

intermediate/diffusion regime, the effectiveness factors were progressively lower, as were the observed orders of reaction. This is in line with the well-known falsification of the kinetics when competition between reaction and mass transfer becomes significant within catalyst particles, leading to a decreased nobs value.26 Based on the computed effectiveness factors, the generalized Thiele’s modulus (ϕg) was obtained26,27 η=

tanh ϕg ϕg

(2)

as well as the characteristic time constants, De/r2 ϕg =

Vp Ap

n−1 n + 1 ρap k vCOII,s 2 De

Figure 9. OII discoloration histories using the Fe−AC catalysts with the same iron dispersion (weight percentage Fe/SBET ratios, series B samples; experimental conditions as in Figure 7).

(3)

where Vp is the volume of the catalyst particle (cm3), Ap is the area of the catalyst particle (cm2), C OII,s is the OII concentration at the catalyst’s surface conditions (mol·cm−3), De is the effective OII diffusion coefficient (cm2·s−1), ρap is the catalyst apparent density (g·cm−3), r is the radius of the spherical catalyst particle (cm), and kv is the true rate constant [(mol·gcat−1·s−1)·(cm3·mol−1)n]. In this study, spherical geometry was considered for the catalysts; thus, in eq 3, Vp/Ap = r/3. As already mentioned (section 2.1), apparent densities were measured by mercury pycnometry. The apparent density essentially did not change with particle size, so a mean value of 0.60 ± 0.03 g·cm−3 was used. As indicated in Table 4, there was a clear loss of catalytic efficiency from the powder to the 0.80−1.60 mm sample, this being related to the increase in the Thiele modulus. It is also in line with the decrease in De/r2 from the sample with 0.25 < dp < 0.80 mm to that with 0.80 < dp < 1.60 mm, which justifies the poorer performance of the latter (Figure 7). The slight differences in the effective diffusion coefficients can be related to the changes in the textural properties from sample to sample, as well as to the iron dispersions, as discussed in section 3.1. As mentioned above, in addition to particle size, the Fe dispersion (and type of iron species) also varies from catalyst to catalyst. To minimize/eliminate this aspect, new samples were synthesized (identified with a B) with the same amount of iron per unit of surface area as the N−Fe powder (used as reference), as mentioned in sections 2.1 and 3.1. As a reminder, the XRD patterns obtained for all samples from series B were equal to that reported in Figure 5a for the powder; that is, the iron species were the same (as was the iron dispersion). Discoloration curves obtained with the new samples are shown in Figure 9. Again, the discoloration rate increased with the decrease in dp. However, in this case, the differences found in the OII elimination were due to particle size only, because the iron dispersion and nature were nearly the same in every sample. It is nevertheless remarkable that, even though the

amount of iron in each sample was reduced, compared with the 7 wt % Fe contained in the previous ones (except for the powder), their activities increased as a consequence of improved iron dispersion. This is particularly noteworthy for the pellets, which led to a dye conversion of ca. 70% after 2 h of reaction versus