Influence of the surface chemistry of multi-walled carbon nanotubes on their activity as ozonation catalysts

CARBON

4 8 ( 2 0 1 0 ) 4 3 6 9 –4 3 8 1

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/carbon

Influence of the surface chemistry of multi-walled carbon nanotubes on their activity as ozonation catalysts Alexandra G. Gonc¸alves, Jose´ L. Figueiredo, Jose´ J.M. O´rfa˜o, Manuel F.R. Pereira

*

Laborato´rio de Cata´lise e Materiais (LCM), Laborato´rio Associado LSRE/LCM, Departamento de Engenharia Quı´mica, Faculdade de Engenharia, Universidade do Porto, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal

A R T I C L E I N F O

A B S T R A C T

Article history:

Multi-walled carbon nanotubes (MWCNTs) with different surface chemical properties were

Received 5 July 2010

prepared by oxidative treatments with HNO3, H2O2 and O2 to introduce oxygen-containing

Accepted 28 July 2010

surface groups and by thermal treatments for their selective removal. The texture and sur-

Available online 12 August 2010

face chemistry of the MWCNTs were characterized by nitrogen adsorption, temperature programmed desorption (TPD) and pH at the point of zero charge. A deconvolution procedure of the TPD spectra was used to quantify the oxygenated surface groups. These materials were used as catalysts for ozone decomposition, and for the ozonation of oxalic and oxamic acids. Generally, all these catalytic processes are favoured by carbon nanotubes with low acidic character. MWCNTs were shown to exhibit higher activity for the ozonation of oxalic and oxamic acids, compared to activated carbon. Successive experimental runs of oxalic acid removal carried out with a selected MWCNT sample show that the catalyst suffers some deactivation as a result of the introduction of oxygenated groups on the surface. Therefore, the effect of the surface chemistry is mainly observed for the fresh catalysts.  2010 Elsevier Ltd. All rights reserved.

1.

Introduction

Catalytic ozonation is an innovative technology for the elimination of organic pollutants in water and wastewater. Ozone preferentially attacks molecules containing unsaturated bonds, leading to the formation of saturated compounds such as aldehydes, ketones and carboxylic acids. Due to their low reactivity towards ozone, these compounds tend to accumulate in water. Therefore, ozonation by itself is not sufficient to achieve a high mineralization degree. To overcome this drawback, ozonation processes are being modified in order to increase their oxidizing capability. Heterogeneous catalytic ozonation reactions, which is one of the most attractive alternatives, aims to enhance the removal of highly refractory compounds by the transformation of ozone into more reactive species and/or adsorption and reaction of the pollutants on the surface of the catalyst [1]. It has been shown that several

metals in solution or in the solid phase under various forms (metal oxides, supported metals) may catalyze ozonation reactions, being able to destroy recalcitrant pollutants [2]. Activated carbon by itself has been proven to be an efficient ozonation catalyst [1,3–9] and a few papers focused on the decomposition of ozone in aqueous solutions in the presence of activated carbon [5,10–14]. The mentioned studies report that activated carbon accelerates the decomposition of ozone, leading to the formation of several radical species, including OH radicals. Carbon nanotubes have been appointed as a very promising material for catalytic applications [15]. However, there are few reports on their use in ozonation processes. Recently, a commercial sample of multi-walled carbon nanotubes (MWCNTs) was successfully used in the catalytic ozonation of oxalic acid in aqueous solution [16]. Moreover, Pt supported on MWCNTs has been tested in the ozonation of organic compounds [17].

* Corresponding author: Fax: +351 225081449. E-mail address: [email protected] (M.F.R. Pereira). 0008-6223/$ - see front matter  2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2010.07.051

4370

CARBON

4 8 ( 2 0 1 0 ) 4 3 6 9 –4 3 8 1

On the other hand, to the best of our knowledge, there are no reports in the literature on the decomposition of ozone in aqueous solutions in the presence of carbon nanotubes. The present work was split into three parts. The first part consists in the preparation of MWCNTs differing either in their surface chemical properties or/and in their textural properties, and their characterization by several techniques, like temperature programmed desorption (TPD), determination of the point of zero charge (pHPZC) and N2 adsorption at 196 C. The main goal of the second part is to investigate the role of surface chemistry and textural properties of MWCNTs in the decomposition of dissolved ozone in aqueous phase, at pH 3. The third part of this work aims at the assessment of modified MWCNTs as catalysts for the ozonation of oxalic and oxamic acids. These carboxylic acids were selected because they are common final oxidation products of several organic pollutants, and they are refractory to single ozonation. In order to evaluate the performance of the resulting materials, experiments in the presence of a commercial activated carbon were also carried out for comparative purposes.

2.

Experimental

2.1. Preparation and characterization of the modified MWCNT samples A commercial MWCNT sample (Nanocyl 3100) was used as the starting material (sample MWCNT-orig). According to the supplier, they have an average diameter of 9.5 nm, an average length of 1.5 lm and a carbon purity higher than 95%. In a recent work [18], it was shown that this material has average inner and outer diameters of 4 and 10 nm, respectively. In the same work, it was observed that Nanocyl 3100 contains growth catalyst impurities, mainly Fe and Co (0.19% and 0.07%, respectively), sulfur (0.14%), which is probably due to the purification process, and traces of Al (0.03%). In order to produce MWCNTs with a strongly acid character and a large amount of surface groups, the original sample was oxidized with HNO3 in the liquid phase (sample MWCNTHNO3). This oxidation was performed using a Pyrex roundbottom flask containing 300 mL HNO3 7 M and 4 g of MWCNTs, connected to a condenser. The liquid was heated to boiling temperature with a heating mantle during 3 h. MWCNTs were washed with distilled water to neutral pH, dried in an oven at 110 C for 24 h and stored in a desiccator for later use. This sample was used as the starting material for thermal treatments under N2, since it is important that it presents a large amount of surface groups in order to produce different MWCNTs with a successively lower acid character, by selectively removing those groups. About 3 g of sample MWCNT-HNO3 were placed in a fused-silica tubular reactor, heated to 400 C at 5 C/min under a flow of N2 (100 Ncm3 min1), and kept at this temperature for 1 h. The sample was cooled to room temperature under the same atmosphere, collected and stored in a desiccator (sample MWCNT-HNO3_N2_400). Sample MWCNT-HNO3 was also submitted to a similar thermal treatment at 600 or 900 C, originating samples MWCNT-HNO3_N2_600 and MWCNTHNO3_N2_900, respectively.

In order to introduce mainly basic and neutral oxygenated groups on the surface, about 3 g of sample MWCNT-orig were submitted to an oxidation in the gas phase with 5% O2 in N2 for 3 h at 500 C (sample MWCNT-O2). This treatment was carried out in the same reactor used for the thermal treatments. Sample MWCNT-orig was also treated by oxidation in liquid phase with H2O2 (30%) to produce a sample with only a slightly acid character compared to sample MWCNT-HNO3. This sample was prepared by mixing 1 g of the original material with 12.5 mL of H2O2 10 M at room temperature for 20 h (until complete degradation of the peroxide). The sample was washed with distilled water to neutral pH, dried for 24 h in the oven at 110 C and stored in the desiccator (sample MWCNT-H2O2). For comparative purposes, experimental results obtained with activated carbon Norit GAC 1240 PLUS were included in this study. The preparation methods for all the carbon materials used in this work are summarized in Table 1. The textural characterization of the materials, namely the Brunauer–Emmett–Teller (BET) surface area, was based on the N2 adsorption isotherms, determined at 196 C with a Quantachrome NOVA 4200e apparatus. The pore size distribution was obtained by using the non-local density functional theory (NLDFT) applying the kernel file provided by Quantachrome’s data reduction software, where a cylindrical-pore model is assumed. The surface chemistry of the carbon materials was characterized by the determination of pHPZC, in order to express the acid-basic character in a single number, and by TPD to identify and quantify the oxygenated groups [19]. TPD profiles were obtained in an AMI-200 (Altamira Instruments) apparatus. The total flow rate of the helium carrier gas (25 Ncm3 min1) and the temperature programme from room temperature to 1100 C at a heating rate of 5 C min1 were controlled with appropriate equipment. The amounts of CO and CO2 released from the MWCNTs samples (100 mg) were monitored with a Dymaxion mass spectrometer (Ametek Process Instruments). The masses monitored for all samples were 16 (O), 18 (H2O), 28 (CO) and 44 (CO2). The determination of the pHPZC of the samples was carried out as follows [20]: 20 mL of NaCl 0.01 M solution was placed in a closed Erlenmeyer flask; the pH was adjusted to a value between 2 and 12 by adding HCl 0.1 M or NaOH 0.1 M; then, 0.050 g of each sample was added and the final pH measured after 24 h under stirring at room temperature. A blank experiment (without the carbon material) was carried out in order to subtract the variation of pH caused by the effect of CO2 present in head space. The pHPZC is the point where the curve pHfinal vs. pHinitial crosses the line pHinitial = pHfinal.

2.2.

Ozone decomposition in the aqueous phase

The ozone decomposition experiments were carried out in a laboratory-scale reactor (ca. 1 L) equipped with stirring and a recirculation jacket. Ozone was produced from pure oxygen in a BMT 802X ozone generator. The concentration of ozone in the gas phase was monitored with a BMT 964 ozone analyser. Ozone leaving the reactor was removed in a series of gaswashing bottles filled with potassium iodide (KI) solution.

CARBON

4371

4 8 ( 20 1 0 ) 4 3 6 9–43 8 1

Table 1 – Carbon materials used in this study. Sample MWCNT-orig MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-HNO3_N2_900 MWCNT-O2 MWCNT-H2O2 AC

Starting material

Treatment

Nanocyl 3100 MWCNT-orig MWCNT-HNO3 MWCNT-HNO3 MWCNT-HNO3 MWCNT-orig MWCNT-orig Norit GAC 1240 PLUS

None Oxidation with HNO3 7 M at 130 C for 3 h Thermal treatment under N2 flow at 400 C for 1 h Thermal treatment under N2 flow at 600 C for 1 h Thermal treatment under N2 flow at 900 C for 1 h Oxidation with 5% O2 in N2 flow at 500 C for 3 h Oxidation with H2O2 (30%) for 20 h at room temperature None

Homogeneous and heterogeneous ozone decomposition experiments were performed at the temperature of 25 C, controlled by a thermostatic bath, and pH 3. In each experiment, 700 mL of 0.01 M phosphate buffer solution (NaH2PO4/ Na2HPO4) was introduced in the reactor. The solution was prepared from ultrapure water obtained in a Milli-Q Millipore system. Gaseous ozone was fed to the reactor for 30 min to saturate the solution and, at the same time, to remove any trace of organic matter that could further affect ozone stability. Prior to each heterogeneous ozone decomposition experiment, 100 mg of carbon material were placed inside the reactor. To keep the reactor content well mixed, stirring was maintained constant at 200 rpm. Liquid samples were taken regularly for quantification of dissolved ozone by the indigo method [21]. The initial concentration of dissolved ozone was around 8 mg L1.

2.3. Kinetic experiments of oxalic and oxamic acids ozonation Oxalic acid (99%) and oxamic acid (96%) were obtained from Fluka. Some properties of these compounds are presented in Table 2. In each ozonation experiment the reactor mentioned in the previous section was filled with 700 mL of carboxylic acid solution with a concentration of 1 mM, at the natural pH (values around 3). The solution was prepared from ultrapure water obtained in a Milli-Q Millipore system. In catalytic ozonation experiments, 100 mg of catalyst were introduced in the reactor. The experiments were performed at constant flow rate (150 Ncm3 min1) and constant inlet ozone concen-

tration (50 g Nm3). The agitation was maintained constant at 200 rpm. For comparative purposes, both adsorption on carbon materials and ozonation experiments in their absence (single ozonation) were performed in the same system, under identical experimental conditions. In cyclic experiments, the same procedure was followed. After each experiment, the solution was filtered and the MWCNT sample dried in order to be used in another run. This procedure was repeated two times. Textural and surface chemistry characterization of the MWCNTs was carried out after each run. Samples were collected using a syringe at selected times and centrifuged for further analysis. The concentration of organics was followed by HPLC using a Hitachi Elite LaChrom HPLC equipped with a diode array detector. The stationary phase was a BIO-RAD Aminex HPX 87H (300 mm · 7.8 mm) working at room temperature. Oxalic and oxamic acids were analyzed under isocratic elution with a solution of H2SO4 4 mM at a flow rate of 0.6 mL min1. The retention times for oxalic and oxamic acids were 6.8 and 9.2 min, respectively. The wavelength of 210 nm was used for quantitative measurements of the carboxylic acids. Six point calibration curves (0.05–1 mM) were made. Linear responses were obtained in this range. The R2 values of calibration lines were 0.99989 and 0.99995 for oxalic acid and oxamic acid, respectively. Limit of quantification (LOQ) = 0.0097 mM, limit of detection (LOD) = 0.00096 mM for oxalic acid, and LOQ = 0.0087 mM, LOD = 0.00294 mM for oxamic acid were obtained. Dissolved ozone was quantified by the indigo method, as mentioned before [21].

Table 2 – Properties of the carboxylic acids. Compound

M (g mol1)

pKa1

pKa2

Oxalic acid

90.03

1.23

4.19

Oxamic acid

89.05

2.50 (for the carboxylic group)

11.8 (for NH3+)

CARBON

3.1.

Characterization of MWCNTs

MWCNT-orig (1) MWCNT-HNO3 (2) MWCNT-HNO3_N2_400 (3) MWCNT-HNO3_N2_600 (4) MWCNT-HNO3_N2_900 (5) MWCNT-HNO3_O2 (6) MWCNT-H2O2 (7)

(a) 0.5 -1

Results and discussion

0.4

-1

3.

4 8 ( 2 0 1 0 ) 4 3 6 9 –4 3 8 1

CO2 (µmol g s )

4372

0.3 2

0.2 0.1

3 4

6

0.0

7

0

200

51

400 600 T (ºC)

800 1000

(b) 0.7 -1

-1

CO (µmol g s )

The final goal of the present work is to study the relationship between the surface chemical characteristics of MWCNTs and their performance on the removal of selected pollutants. For that purpose, a set of modified MWCNTs with different levels of acidity/basicity was prepared. Oxidations with HNO3 and H2O2 are known to generate acidic materials [19]. Treatments with HNO3 originate materials with large amounts of surface acidic groups, mainly carboxylic acids and, to a smaller extent, lactones, anhydrides and phenol groups [19], whereas H2O2 oxidation generates less acidic materials, in comparison to the HNO3 treatments [22]. The characterization results are shown in Table 3. TPD is a thermal analysis method that is becoming popular for the characterization of the oxygen-containing groups on the surface of carbon materials. In this technique, all oxygenated surface groups are thermally decomposed releasing CO and/or CO2, and in some cases H2O and H2, at different temperatures. The nature of the groups can be assessed by the decomposition temperature and the type of gas released, and their respective amounts determined by the areas of the component peaks, obtained by deconvolution techniques [19]. The major problem is the difficulty in identifying each surface group individually, because TPD results show composite CO and CO2 spectra. CO2 evolution results from the decomposition of carboxylic acids at low temperature, and lactones at higher temperatures; carboxylic anhydrides originate both CO and CO2; phenols and carbonyl/quinone groups originate CO. Fig. 1 shows the TPD spectra of the MWCNTs before and after the different treatments. An increase in the amount of surface oxygen-containing groups is evidenced by the larger amounts of CO and CO2 released. The CO2 spectra show that oxidation with HNO3 increases the CO2 evolution mainly at low temperatures (from 200 to 450 C). This results from the decomposition of carboxylic acid groups. However, the oxidation with HNO3 also introduces, in less extent, carboxylic anhydrides (CO and CO2 released from 450 to 650 C) and lactones (CO2 released from 550 to 700 C). Large amounts of phenols and carbonyls (CO released at high temperatures) are also introduced by this treatment.

0.6 0.5 0.4

2

0.3 0.2

3

6 4 7

0.1 0.0

1 5

0

200 400 600 800 1000 T (ºC)

Fig. 1 – TPD spectra: (a) CO2 evolution; (b) CO evolution.

Oxidation with hydrogen peroxide originates MWCNTs with much less acidic surface groups, compared to oxidation with nitric acid. On the other hand, oxidation in the gas phase induces a small CO2 peak at high temperatures, which may be attributed to lactones decomposition. The CO spectrum of this sample shows a large peak between 600 and 1100 C, which may originate from phenol (from 600 to 750 C) and carbonyl/quinone (from 700 to 950 C) groups. The groups introduced by oxidation with HNO3 were selectively removed in the thermally treated samples, as can be observed in the corresponding TPD spectra. Almost all carboxylic acids were removed after heat treating the sample at 400 C. After thermal treatment at 600 C, the CO2 releasing groups were almost completely removed (only a few lactones remain), and some CO releasing groups were also removed from the surface of the material, which can be assigned mainly to carboxylic anhydrides. Almost all oxygenated groups are removed at

Table 3 – Textural and chemical characterization of prepared materials. Sample MWCNT-Original MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-HNO3_N2_900 MWCNT-O2 MWCNT-H2O2 AC a b

CO2a (lmol g1) 25 1514 561 125 15 91 150 63

COa (lmol g1) 478 2435 2350 1494 308 1339 466 579

CO/CO2 19 1.6 4.2 12 20 15 3.1 9.2

% m0b

pHPZC

0.84 8.7 5.6 2.8 0.54 2.5 1.2 1.1

7.0 3.0 3.8 5.9 7.3 5.2 5.0 8.5

SBET (m2 g1) 331 476 483 504 529 508 337 909

Amounts of CO and CO2 released, obtained by integration of the areas under TPD spectra. Mass percentage of oxygen on the surface, obtained from TPD data assuming that all the surface oxygen is released as CO and/or CO2.

CARBON

4 8 ( 20 1 0 ) 4 3 6 9–43 8 1

900 C. It is interesting to note in Table 3 that the original MWCNT sample presents not much less oxygen-containing surface groups, especially CO releasing groups, compared to the activated carbon sample, in spite of the differences in their structure and surface area. This seems to indicate that these MWCNTs are easily oxidized. To determine the amount of each surface group, deconvolution of the CO and CO2 TPD spectra was carried out. A multiple Gaussian function was used for fitting each spectrum. The numerical calculations were based on a non-linear routine, which minimized the sum of squared deviations, using the Levenberg–Marquardt method to perform the iterations. The use of Gaussian functions is justified by the shape of the TPD peaks, which are a result of continuous random distributions of binding energies of the surface groups [23]. For the MWCNT samples, some assumptions (justified in previous works with activated carbons [19,24]) were followed: •

•

•

• •

The CO2 spectra are decomposed into three contributions, corresponding to carboxylic acids (CO2 peak #1), carboxylic anhydrides (CO2 peak #2) and lactones (CO2 peak #3). The samples heat treated above 400 C and sample MWCNTO2 do not present CO2 evolution at low temperature, because they were processed at temperatures higher than those corresponding to the decomposition of carboxylic acids. Each carboxylic anhydride group decomposes by releasing one CO and one CO2 molecule. The first component in the CO spectrum corresponds to the carboxylic anhydrides, and the corresponding peak (CO peak #1) has the same shape and equal magnitude to CO2 peak #2. This peak is pre-defined from the deconvolution of the CO2 spectra. In addition to carboxylic anhydrides (CO peak #1), the CO spectrum includes contributions from phenols (CO peak #2) and carbonyl/quinones (CO peak #3) [25]. This sequence of decomposition temperatures (first phenols, then carbonyl/quinones) is justified by the higher stability of the later groups. The same width at half-height was imposed for CO2 peaks #1, #2 and #3. The same width at half-height was considered for phenol and carbonyl groups. This was necessary mainly in the case of samples where there were no peak shoulders.

This deconvolution procedure proved to fit the data quite well for the CO and CO2 TPD spectra of samples MWCNTHNO3, MWCNT-HNO3_N2_400, MWCNT-HNO3_N2_600 and MWCNT-O2, as shown in Fig. 2. Deconvolution was not done for the remaining materials, since the amounts of oxygenated groups on the surface of those samples are very low. Tables 4 and 5 show the results obtained, where TM is the temperature of the component peak maximum, W is the width of the component peak at half-height and A is the integrated component peak area. The amounts of CO and CO2 released, obtained by integration of the areas under TPD spectra, the ratio CO/CO2 and the mass percentage of oxygen (% m0) on the surface of carbon materials are shown in Table 3. It was assumed that all oxygen-containing surface groups are thermally decomposed

4373

releasing only CO and CO2, which is perfectly acceptable since the H2O spectra show that there is practically no water released, and neglecting any groups that decompose above 1100 C. All samples present higher amounts of CO than CO2 releasing groups. The sample oxidized with nitric acid has the highest amount of surface oxygen. This sample also presents the lowest ratio CO/CO2, indicating that this is the most acidic sample. MWCNT-HNO3_N2_900 sample presents the highest CO/CO2 ratio, suggesting the less acidic characteristics. These observations are consistent with the pHPZC results (Table 3). The acid character of the samples decreases by increasing the thermal treatment temperature, because the acid groups are removed at lower temperatures than neutral and basic groups. This is clearly observed by analyzing the evolution of the individual peaks obtained by TPD deconvolution (see Table 4): carboxylic acids (CO2 peak #1) and anhydrides (CO2 peak #2 and CO peak #1) completely disappear for thermal treatments above 400 C; on the other hand, carbonyls (CO peak #3) only decrease significantly after heat treatments above 600 C. Oxidation in the gas phase introduced mainly neutral and basic groups on the surface, like phenols and carbonyls/quinones. The original sample and sample MWCNT-HNO3_N2_900 present almost no oxygenated groups on the surface. The prepared MWCNT samples present N2 adsorption isotherms of type II, which is typical of non-porous materials [26]. The surface areas of the samples were calculated by the BET method (SBET), and are included in Table 3. Oxidation treatments with HNO3 and O2 lead to an increase of the specific surface area. This occurs because these oxidative processes open up the endcaps of MWCNTs and create sidewall openings [27]. The specific surface areas of the samples slightly increase as the thermal treatment temperature increases, since carboxylic acids and other groups, introduced during oxidation with HNO3, are removed. The mild oxidation with H2O2 did not change the textural properties of the MWCNTs. The NLDFT results (Fig. 3) corroborate the previous observations. It is clearly observed that a new contribution appears in ˚ the pore size distribution for pore radius between 13 and 18 A only for the nitric acid treated samples (MWCNT-HNO3, MWCNT-HNO3_N2_400, MWCNT-HNO3_N2_600 and MWCNTHNO3_N2_900), which may be explained by the opening of the MWCNT tips. The same was expected for the sample oxidized with oxygen, because a substantial increase was observed in the BET surface area; nevertheless, the increase in the pore size distribution of this sample is displaced to higher ˚ ). This could be explained by pore radius (between 20 and 25 A the more severe oxidation conditions applied in the preparation of this sample, which could attack the interior of the tubes. Most of the pores observed in the pore size distribution result from the free space in the MWCNT bundles. It is interesting to note that the sample treated with H2O2 is the only one that does not present a large peak in the pore size distribution be˚ , which may indicate that this treatment tween 50 and 100 A has some effect in the arrangement of the MWCNT bundles. The commercially activated carbon, used for comparative purposes, presents a much higher BET specific surface area than MWCNTs, according to the respective microporosity.

4374

CARBON

0.7

0.5

0.6

-1 -1

0.4 0.3 0.2 0.1

0.6 0.5

0

200

400

600

0.2 0.1 0

200

800 1000

0.6

-1

0.2 0.1 0

200

400

600

0.5 0.4 0.3 0.2 0.1 0.0

800 1000

0.6

0.6

0.5

0.5

-1 -1

0.4 0.3 0.2 0.1 0

200

400

600

200

400

600

800 1000

T (ºC)

0.4 0.3 0.2 0.1 0.0

800 1000

0

0

200

400

600

800 1000

T (ºC)

T (ºC) 0.2

0.7

CO (µmol g s )

0.6 -1

-1

-1 -1

CO2 (µ mol g s )

600

T (ºC)

0.3

0.0

(d)

400

0.7

CO (µ mol g s )

-1 -1

0.3

T (ºC)

T (ºC)

CO2 (µmol g s )

0.4

0.0

800 1000

0.4

0.0

(c)

0.5

-1

(b) CO2 (µ mol g s )

0.0

-1 -1

CO (µmol g s )

0.6

CO (µ mol g s )

-1 -1

CO2 (µmol g s )

(a)

4 8 ( 2 0 1 0 ) 4 3 6 9 –4 3 8 1

0.1

0.0

0

200

400

600

0.5 0.4 0.3 0.2 0.1 0.0

800 1000

0

200

400

600

800 1000

T (ºC)

T (ºC)

Fig. 2 – Results of the deconvolution of TPD spectra using a multiple Gaussian function: (a) MWCNT-HNO3 (b) MWCNTHNO3_N2_400 (c) MWCNT-HNO3_N2_600 (d) MWCNT-O2. (h, TPD experimental data; - - -, individual peaks; —, sum of individual peaks).

Table 4 – Results of the deconvolution of CO2 spectra using a multiple Gaussian function. Sample

MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-O2

Peak #1

Peak #2

Peak #3

TM (K)

W (K)

A (lmol g1)

TM (K)

W (K)

A (lmol g1)

TM (K)

W (K)

A (lmol g1)

278 293 – –

140 127 – –

935 89 – –

440 501 – 606

140 127 – 165

298 288 – 39

512 658 748 658

140 127 147 166

194 115 77 44

In this section, several MWCNT samples were obtained. They were well characterized, including quantification of the oxygen-containing surface groups by TPD deconvolution and the acid-basic character by determination of the pHPZC.

These materials are subsequently used as catalysts for the decomposition of ozone and for the ozonation of oxalic and oxamic acids, with the main objective of studying the effect of the surface chemistry in those reactions.

4375

4 8 ( 20 1 0 ) 4 3 6 9–43 8 1

CARBON

Table 5 – Results of the deconvolution of CO spectra using a multiple Gaussian function. Sample

Peak #1

MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-O2

TM (K)

W (K)

440 501 – 606

140 127 – 165

Peak #2

A (lmol g1) 298 288 – 39

TM (K)

W (K)

640 676 758 712

244 245 167 212

Peak #3

A (lmol g1) 1386 1343 450 434

TM (K)

W (K)

800 790 844 794

244 245 167 212

A (lmol g1) 926 988 843 841

Fig. 3 – Pore size distributions obtained by NLDFT.

3.2. Homogeneous and decomposition in aqueous phase

heterogeneous

ozone

According to a few studies published in the literature, the presence of materials like activated carbon accelerates the decomposition of ozone in the aqueous phase [5,10–14]. In this section, the performance of MWCNTs in the decomposition of ozone in the aqueous phase is evaluated. Fig. 4 shows the results obtained, and it is observed that the presence of MWCNTs accelerates the decomposition of ozone. The experimental data were analyzed considering firstand second-order kinetic models. The second-order fit was

significantly better both for the homogeneous decomposition and for decomposition in the presence of all the materials considered. According to the second-order kinetic model, the evolution of ozone concentration is described by the following equation: 

dCO3 ¼ kd;hom C2O3 dt

ð1Þ

where kd;hom (L gO3 1 min1) represents the second-order apparent rate constant and CO3 (gO3 L1) is the concentration of dissolved ozone. Integration of Eq. (1), taking into consideration that CO3 ¼ CO3;0 when t = 0, leads to:

1.0

0.8

Single ozonation MWCNT-orig MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-HNO3_N2_900 MWCNT-O2 MWCNT-H2O2 nd 2 order fit

3

3

CO /CO ,0

0.6

0.4

0.2

0.0 0

5

10

15

20

25

30

t (min) Fig. 4 – Decomposition of ozone at 25 C and pH 3 (MWCNT = 0.14 g L1).

4376 CO3 1 ¼ CO3;0 1 þ kd;hom CO3;0 t

CARBON

4 8 ( 2 0 1 0 ) 4 3 6 9 –4 3 8 1

ð2Þ

In the presence of carbon materials both homogeneous and heterogeneous decomposition of ozone occur. Thus, the ozone decomposition rate is the sum of two contributions. Therefore, 

dCO3 ¼ ðkd;hom þ kd;het ÞC2O3 dt

ð3Þ

where kd;het (L gO3 1 min1) represents the second-order apparent rate constant for the heterogeneous decomposition. Integration of Eq. (3), considering CO3 ¼ CO3;0 when t = 0, leads to: CO3 1 ¼ CO3;0 1 þ ðkd;hom þ kd;het ÞCO3;0 t

ð4Þ

Apparent second-order rate constant values, obtained from non-linear fitting according to Eqs. (2) and (4) are listed in Table 6. For the fitting procedure in heterogeneous systems, the value of kd;hom , previously calculated, was fixed. The ratio between kd;het and SBET is presented in the same table. In order to evaluate the influence of the chemical surface properties of samples in ozone decomposition, kd;het /SBET is plotted against pHPZC (see Fig. 5). The correlation between the second-order heterogeneous apparent rate constants normalized by SBET and surface chemical properties of MWCNTs is significant at the 0.5% level. In general, the trend of the catalytic activity for the decomposition of ozone follows the decrease of acidity, according to what was observed for activated carbon [5,10]. It is interesting to note the good performance of sample MWCNT-H2O2, being the only sample above the expected values for that correlation. This oxidation treatment is much weaker than that with HNO3, since it did not change the textural properties of MWCNT. The only significant difference was the disappearance of the pores in ˚ (Fig. 3), which as already exthe range between 50 and 100 A plained may indicate that this treatment has some effect on the arrangement of the MWCNT bundles. In this case, the access of ozone to the carbon surface is easier than for the other samples. These observations can be explained in a similar way as previously proposed for ozone decomposition in the presence of activated carbon [10,28]: acid MWCNTs are characterized by having a high content of surface electron-withdrawing oxygenated groups and by the existence of delocalized p elec-

Fig. 5 – Correlation of the normalized rate constants for heterogeneous ozone decomposition (kd,het/SBET) with pHPZC.

trons on their surfaces. Therefore, the electron density on the surface is low when compared to that of less acidic MWCNTs. Thus, the later samples favour the adsorption step of ozone molecules, as a result of the dispersive interactions between the mentioned electrons and ozone molecules. On the other hand, they act as an electron source, favouring the reduction of ozone molecules, which have electrophilic properties. Consequently, MWCNTs with less acidic character have a higher catalytic activity for ozone decomposition.

3.3.

Oxalic acid removal

Oxalic acid has been identified among the most common oxidation products from organic pollutants degradation; additionally, it is usually refractory to single ozonation. Several catalysts were shown to improve oxalic acid ozonation in aqueous solution, as described in the literature [1,29–32]. Ozonation of oxalic acid leads directly to mineralization, i.e., there is no formation of organic intermediates. In the present study, the ozonation of this carboxylic acid, at the natural pH (approximately 3), in the presence of the MWCNT samples prepared was investigated. For comparative purposes, experimental results obtained with a commercial activated carbon are included. Fig. 6 shows the results obtained during the catalytic and non-catalytic ozonation of oxalic acid. In order to evaluate the adsorption capacity of the sam-

Table 6 – Apparent second-order rate constants for homogeneous and heterogeneous ozone decomposition. Homogeneous or heterogeneous experiments Homog MWCNT-orig MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-HNO3_N2_900 MWCNT-O2 MWCNT-H2O2 a

Homogeneous value.

kd,het (L gO3 1 min1) 0.81 ± 0.06a 17.2 ± 0.6 10.1 ± 0.8 14 ± 1 17 ± 1 27 ± 1 18.1 ± 0.7 15.6 ± 0.5

1 kd,het/SBET (L gMWCNT m2 g1 O3 min )

– 0.052 0.021 0.030 0.034 0.052 0.036 0.046

CARBON

4377

4 8 ( 20 1 0 ) 4 3 6 9–43 8 1

(a) 1.0 0.9 0.8 0.7

(b) 1.0

Single ozonation MWCNT-orig MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-HNO3_N2_900 MWCNT-O2 MWCNT-H2O2 AC zero order fit

0.5 0.4 0.3 0.2 0.1 0.0 0

0.8

0.6

C/C0

C/C0

0.6

0.4

0.2

0.0

20

40

60

0

20

40

60

80

100

120

140

160

180

t (min)

t (min) Fig. 6 – Evolution of the dimensionless concentration of oxalic acid at natural pH (3) during catalytic and non-catalytic ozonation (C0 = 1 mM, MWCNT or AC = 0.14 g L1): (a) until 1 h of reaction; (b) during 3 h of reaction.

ples towards oxalic acid, adsorption experiments were also carried out and are presented in Fig. 7. The results indicate that adsorption on MWCNTs scarcely contributes to the elimination of oxalic acid compared to catalytic ozonation, and can therefore be neglected. Thus, it can be considered that the removal of oxalic acid in the presence of ozone and MWCNTs mainly occurs by catalytic ozonation. However, it was observed that adsorption on sample MWCNTHNO3_N2_900 was higher (20% after 3 h) than on the other MWCNT samples, and presented a similar performance compared to activated carbon. The mineralization of oxalic acid was significantly enhanced by the addition of MWCNTs. All samples act efficiently as ozonation catalysts, even better than activated carbon. In fact, the simultaneous use of ozone and MWCNTs resulted in the removal of more than 70% of oxalic acid after 60 min, compared to only 40% with the commercial activated carbon. The better performance of MWCNTs compared to activated carbon may be explained not only by the higher amount of delocalized p electrons on the surface, which are

known to be active sites for ozone decomposition and formation of radicals [1], but also by the significant decrease of internal mass transfer resistances. In fact, contrary to activated carbons, MWCNTs do not present micropores, and the access of the reactants to the surface is almost free. The most efficient ozonation catalyst is MWCNTHNO3_N2_900, which removed more than 95% of oxalic acid, after 60 min. The experimental curves in Fig. 6 during the first 60 min suggest a zero-order reaction with respect to oxalic acid. According to the zero-order kinetic model, the evolution of oxalic acid concentration during single ozonation is described by the following equation: 

dCoxalic acid ¼ khom dt

ð5Þ

where khom (mmol L1 min1) represents the zero-order apparent rate constant and Coxalic acid (mmol L1) is the concentration of oxalic acid in each instant. Integration of Eq. (5), considering Coxalic acid ¼ Coxalic acid;0 when t = 0, leads to:

1.0

0.8 MWCNT-orig MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-HNO3_N2_900 MWCNT-O2 MWCNT-H2O2 AC

C/C0

0.6

0.4

0.2

0.0 0

20

40

60

80

100 120 140 160 180

t (min) Fig. 7 – Evolution of the dimensionless concentration of oxalic acid at natural pH (3) during adsorption experiments (C0 = 1 mM, MWCNT or AC = 0.14 g L1).

4378

CARBON

4 8 ( 2 0 1 0 ) 4 3 6 9 –4 3 8 1

Coxalic acid khom ¼1 t Coxalic acid;0 Coxalic acid;0

ð6Þ

In the presence of the prepared materials, both homogeneous and heterogeneous degradation occur. Thus, the oxalic acid removal rate is the sum of the two contributions. Therefore, 

dCoxalic acid ¼ ðkhom þ khet Þ dt

ð7Þ

where khet (mmol L1 min1) represents the zero-order apparent rate constant for the heterogeneous degradation. Integration of Eq. (7), considering Coxalic acid ¼ Coxalic acid;0 when t = 0, leads to: Coxalic acid khom þ khet ¼1 t Coxalic acid;0 Coxalic acid;0

ð8Þ

Apparent zero-order rate constants values are listed in Table 7. For the fitting procedure, the value of khom previously calculated was fixed. The most efficient ozonation catalyst was MWCNTHNO3_N2_900, according to its highest specific surface area and pHPZC. This observation could be explained by a sum of two effects: this material has a higher adsorption capacity for oxalic acid than other samples, and a superior ability to enhance the catalytic decomposition of ozone into surface radical species and/or oxygen-containing radicals in solution, such as OH.

Table 7 – Apparent zero-order rate constants for catalytic and non-catalytic ozonation of oxalic acid (determined from the first 60 min of reaction). Experiments Single ozonation MWCNT-orig MWCNT-HNO3 MWCNT-HNO3_N2_400 MWCNT-HNO3_N2_600 MWCNT-HNO3_N2_900 MWCNT-O2 MWCNT-H2O2 a

khet · 103 (mmol L1 min1) 1.22 ± 0.05a 13.8 ± 0.2 11.5 ± 0.2 12.3 ± 0.2 16.2 ± 0.2 18.0 ± 0.3 13.3 ± 0.1 15.8 ± 0.1

Homogeneous value.

According to Fig. 8, where khet is plotted against kd,het, the mineralization rate of oxalic acid is correlated with the decomposition rate of dissolved ozone in aqueous phase, being this correlation significant at the 2% level. This is a clear indication that the catalytic activity of MWCNTs in the ozonation of oxalic acid is related to the ability of these materials to decompose ozone producing oxygen-containing radicals on the surface and/or in solution; these species being known to play a key role in the ozonation mechanism [1]. In addition, this is also indicative of a direct relationship between the zero-order heterogeneous apparent rate constant and the surface chemical properties of the MWCNTs, namely that khet significantly increases with pHPZC. It can be argued that ozonation using MWCNTs as catalysts should have a mechanism similar to that observed using activated carbon [1]. This was also recently defended by Liu et al. [16] using MWCNTs as ozonation catalysts. Since the initial pH of the solutions is around 3.0, it is expected that the ozonation of oxalic acid occurs through a free radical mechanism involving surface and bulk reactions between oxalic acid and active species. According to the experimental results obtained in the present work, the presence of MWCNTs accelerates the decomposition of ozone. This performance can be explained by two pathways: the first one assumes that MWCNTs act as initiator of the decomposition of ozone yielding free radical species, such as hydroxyl radicals in solution [12]; another pathway is the adsorption and reaction of ozone molecules on the surface of the MWCNTs, yielding surface oxygenated radicals [14]. The oxidation of oxalic acid can then occur on the surface of the MWCNTs, between adsorbed reactant and surface radical species. It is necessary to consider that adsorbed species might also react with dissolved ozone or hydroxyl radicals from the aqueous phase. In addition, a homogeneous reaction mechanism must be considered, because the oxidation of oxalic acid in aqueous phase also occurs, mainly via a radical chain mechanism, where hydroxyl radicals are the main intervenient [1]. Therefore, the ozonation of oxalic acid in the presence of MWCNTs occurs both on the MWCNT surface and in solution [16]. The decrease of acidic groups on the MWCNT surface leads to an increase of their performance in the mineralization of oxalic acid, which is explained by the negative influence of the oxygen-containing surface groups in the decomposition of ozone, as already explained in section 3.2. Additionally, for the surface reactions, there is also a negative influence of the oxygenated groups, since they do not favour the adsorption of oxalic acid (see Fig. 7).

3.3.1.

Fig. 8 – Correlation between khet and kd,het.

Cyclic experiments

Reutilization of MWCNT-orig samples was carried out with the purpose of studying the influence of ozonation on the surface chemistry of the MWCNTs, and the eventual deactivation during mineralization of oxalic acid. The kinetic results obtained are shown in Fig. 9. The corresponding values of the heterogeneous apparent zero-order rate constants are presented in Table 8, as well as BET surface areas. A slight increase of SBET from the first to the second run was observed. By comparing the curves and the respective khet values for ozone decomposition in the presence of MWCNT-orig, it is evident that there is a decrease in the activity of the MWCNTs for oxalic acid removal, particularly from the first to the second run.

CARBON

4379

4 8 ( 20 1 0 ) 4 3 6 9–43 8 1

1.0

0.8

Single ozonation st 1 cycle nd 2 cycle rd 3 cycle

C/C0

0.6

0.4

0.2

0.0 0

20

40

60

80

100 120

140

160

180

t (min) Fig. 9 – Evolution of the dimensionless concentration of oxalic acid at natural pH (3) during catalytic cyclic ozonation experiments (C0 = 1 mM, MWCNT = 0.14 g L1).

Table 8 – Apparent zero-order rate constants for ozonation cyclic experiments of oxalic acid (determined from the first 60 min of reaction) and SBET of the materials used. SBET (m2 g1)

MWCNT-orig 1st cycle 2nd cycle 3rd cycle After 3rd cycle

-1

-1

CO2 (µmol g s )

(a)

13.8 ± 0.2 12.5 ± 0.3 10.7 ± 0.1

0.25 0.20

MWCNT-orig (1) st MWCNT-orig_after 1 cycle (2) nd MWCNT-orig_after 2 cycle (3) rd MWCNT-orig_after 3 cycle (4)

4

0.15 3

0.10

2

0.05 1

0

200

400 600 T (ºC)

800 1000

0.30 0.25

-1

-1

CO (µmol g s )

331 369 360 366

0.30

0.00

(b)

khet · 103 (mmol L1 min1)

0.20 4

0.15 0.10

3 2

0.05 0.00

1

0

200

400 600 T (ºC)

800 1000

Fig. 10 – TPD spectra for samples used in cyclic experiments: (a) CO2 evolution; (b) CO evolution.

0.042 0.034 0.030

ozone, but not to a great extent. The observed decrease in the activity of this sample in successive runs is probably due to the introduction of a limited number of oxygenated electron-withdrawing groups, which reduce the electron density on the carbon surface, thus decreasing the catalytic activity of the material for the decomposition of ozone. This effect is particularly evident from the first to the second run. The specific surface area of sample MWCNT-orig slightly increases from the first to second run, while for the following runs it remains constant (see Table 8). If the heterogeneous apparent zero-order rate constants are normalized by the specific surface areas, these values decrease with successive reutilizations of the catalyst, as presented in Table 8. It can be observed that these values seem to tend to the value of sample MWCNT-HNO3 (a value of 0.024 · 103 mmol gMWCNT m2 L1 min1 for khet/SBET can be obtained from Table 7), which is the sample with more oxygenated groups on its surface. Hence, even after three cycles of use, the MWCNTs continue to be active in the ozonation of oxalic acid. It is important to stress that the surface chemistry plays a role in oxalic acid removal. However, because of the surface oxidation that occurs during this process, the influence of the surface chemistry tends to be attenuated along successive reutilizations of the catalyst.

3.4. TPD spectra of samples before and after reaction are presented in Fig. 10. The increase in the amount of oxygen-containing surface groups in the reused MWCNT-orig sample confirms that this sample is, in fact, oxidized by dissolved

(khet/SBET) · 103 (mmol gMWCNT m2 L1 min1)

Oxamic acid removal

The oxidation of organic compounds containing nitrogen functional groups can also result in the formation of oxamic acid [33,34], which is more refractory to oxidation than oxalic

4380

CARBON

4 8 ( 2 0 1 0 ) 4 3 6 9 –4 3 8 1

1.0

0.8 1.0

Single ozonation MWCNT-orig MWCNT-HNO3 MWCNT-HNO3_N2_900 MWCNT-H2O2 AC

0.8

C/C0

0.6 C/C0

0.6

0.4

0.4 0.2

0.2 0.0 0

120

240

360

480

600

t (min)

0.0 0

30

60

90 t (min)

120

150

180

Fig. 11 – Evolution of the dimensionless concentration of oxamic acid at natural pH (3) during catalytic and non-catalytic ozonation (C0 = 1 mM, MWCNT or AC = 0.14 g L1).

acid. Thus, ozonation experiments of oxamic acid at natural pH were also carried out (see Fig. 11). The following samples were evaluated: MWCNT-orig, MWCNT-H2O2, MWCNTHNO3 and MWCNT-HNO3_N2_900. For comparative purposes, experimental results obtained with the commercial activated carbon are included. As expected, oxamic acid was found to be more recalcitrant to catalytic ozonation than oxalic acid. The best removal (about 30% after 3 h) was obtained using MWCNTHNO3_N2_900. Moreover, about 70% of oxamic acid was mineralized after 10 h of reaction in the presence of this catalyst. This sample is more active than the commercial activated carbon. The relative performances of catalysts in the mineralization of oxamic acid are similar to those found for degradation of oxalic acid, since the activity increases with the pHPZC of the samples.

creases the rate of degradation of both carboxylic acids, leading to complete mineralization of oxalic acid. A low acid character of the catalysts enhances the efficiency of this process. A significant correlation between the surface chemistry of MWCNTs and their performance for ozone decomposition, and indirectly as ozonation catalysts, was obtained. Successive experimental runs of oxalic acid ozonation carried out with a selected sample, show that the surface chemistry suffers a slight progressive oxidation by exposure to dissolved ozone, with a limited loss of activity, tending to the performance of the acid MWCNTs after a few runs. MWCNTs present a higher catalytic performance than activated carbons for ozonation, which was justified by their differences in surface chemistry and by the high internal mass transfer resistances expected for activated carbons (essentially microporous).

4.

Acknowledgements

Conclusions

MWCNTs with different surface chemical properties were prepared and characterized. Oxidation with nitric acid originated highly acidic materials with a large amount of oxygenated surface groups. The subsequent thermal treatments removed selectively those groups, the acid character of the samples decreasing as the thermal treatment temperature increases. Oxidation in the gas phase introduced mainly basic, neutral and weakly acidic groups on the surface. A deconvolution procedure of the TPD spectra was used to quantify the oxygenated surface groups. In general, oxidation induces an increase in the specific surface area of the MWCNTs. The decomposition of ozone in the aqueous phase at pH 3 is significantly enhanced in the presence of MWCNTs. Both textural and surface chemical properties control the process, which is favoured by MWCNTs with large specific surface areas and low acidic character. The simultaneous use of ozone and MWCNTs yields a significant improvement of oxalic and oxamic acid removal from water when compared to single ozonation and adsorption. Generally, the presence of MWCNTs during ozonation in-

This work was carried out with the support of Fundac¸a˜o para a Cieˆncia e a Tecnologia (FCT) and FEDER under Programme COMPETE (project NANO/NTec-CA/0122/2007). A.G.G. acknowledges the Grant received from FCT (BD/45826/2008). The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/ 2007-2013) under Grant Agreement No. 226347.

R E F E R E N C E S

´ rfa˜o JJM, Pereira MFR. Activated carbon catalytic [1] Faria PCC, O ozonation of oxamic and oxalic acids. Appl Catal B 2008;79(3):237–43. [2] Legube B, Karpel Vel Leitner N. Catalytic ozonation: a promising advanced oxidation technology for water treatment. Catal Today 1999;53(1):61–72. ´ rfa˜o JJM, Pereira MFR. Mineralisation of coloured [3] Faria PCC, O aqueous solutions by ozonation in the presence of activated carbon. Water Res 2005;39(8):1461–70.

CARBON

4 8 ( 20 1 0 ) 4 3 6 9–43 8 1

[4] Beltra´n FJ, Garcı´a-Araya JF, Gira´ldez I. Gallic acid water ozonation using activated carbon. Appl Catal B 2006;63(3– 4):249–59. [5] Sa´nchez-Polo M, von Gunten U, Rivera-Utrilla J. Efficiency of activated carbon to transform ozone into OH radicals: influence of operational parameters. Water Res 2005;39(14):3189–98. [6] Be´ltran FJ, Rivas FJ, Fernandez LA, Alvarez PM, Montero-deEspinosa R. Kinetics of catalytic ozonation of oxalic acid in water with activated carbon. Ind Eng Chem Res 2002;41(25):6510–7. [7] Lin SH, Lai CL. Kinetic characteristics of textile wastewater ozonation in fluidized and fixed activated carbon beds. Water Res 2000;34(3):763–72. [8] Rivera-Utrilla J, Sa´nchez-Polo M. Ozonation of 1,3,6naphthalenetrisulphonic acid catalysed by activated carbon in aqueous phase. Appl Catal B 2002;39(4):319–29. ´ rfa˜o JJM, Pereira MFR. Catalytic ozonation of [9] Faria PCC, O sulfonated aromatic compounds in the presence of activated carbon. Appl Catal B 2008;83(1–2):150–9. ´ rfa˜o JJM, Pereira MFR. Ozone decomposition in [10] Faria PCC, O water catalysed by activated carbon: influence of chemical and textural properties. Ind Eng Chem Res 2006;45(8):2715–21. ´ lvarez P, Montero-de-Espinosa R. Kinetics [11] Beltra´n FJ, Rivas J, A of heterogeneous catalytic ozone decomposition in water on an activated carbon. Ozone Sci Eng J Int Ozone Assoc 2002;24(4):227–37. [12] Jans U, Hoigne´ J. Activated carbon and carbon black catalyzed transformation of aqueous ozone into OH radicals. Ozone Sci Eng 1998;20(1):67–90. [13] Guiza M, Ouerdeni A, Ratel A. Decomposition of dissolved ozone in the presence of activated carbon: an experimental study. Ozone Sci Eng 2004;26(3):299–307. [14] Alva´rez PM, Garcı´a-Araya JF, Beltra´n FJ, Gira´ldez I, Jaramillo J, Go´mez-Serrano V. The influence of various factors on aqueous ozone decomposition by granular activated carbons and the development of a mechanistic approach. Carbon 2006;44(14):3102–12. [15] Serp P, Corrias M, Kalck P. Carbon nanotubes and nanofibers in catalysis. Appl Catal A 2003;253(2):337–58. [16] Liu Z-Q, Ma J, Cui Y-H, Zhang B-P. Effect of ozonation pretreatment on the surface properties and catalytic activity of multi-walled carbon nanotube. Appl Catal B 2009;92(3–4):301–6. [17] Liu Z-Q, Ma J, Cui Y-H. Carbon nanotube supported platinum catalysts for the ozonation of oxalic acid in aqueous solutions. Carbon 2008;46(6):890–7. [18] Tessonnier J-P, Rosenthal D, Hansen TW, Hess C, Schuster ME, Blume R, et al. Analysis of the structure and chemical properties of some commercial carbon nanostructures. Carbon 2009;47(7):1779–98. ´ rfa˜o JJM. [19] Figueiredo JL, Pereira MFR, Freitas MMA, O Modification of the surface chemistry of activated carbons. Carbon 1999;37(9):1379–89.

4381

[20] Rivera-Utrilla J, Bautista-Toledo I, Ferro-Garcı´a MA, MorenoCastilla C. Activated carbon surface modifications by adsorption of bacteria and their effect on aqueous lead adsorption. J Chem Technol Biotechnol 2001;76(12):1209–15. [21] Bader H, Hoigne´ J. Determination of ozone in water by the indigo method. Water Res 1981;15(4):449–56. [22] Moreno-Castilla C, Ferro-Garcia MA, Joly JP, Bautista-Toledo I, Carrasco-Marin F, Rivera-Utrilla J. Activated carbon surface modifications by nitric acid, hydrogen peroxide, and ammonium peroxydisulfate treatments. Langmuir 1995;11(11):4386–92. [23] Calo JM, Hall PJ. Fundamental issues in the control of carbon gasification reactivity. Dordrecht, The Netherlands: Kluwer Academic Publishers; 1991. ´ rfa˜o JJM. [24] Figueiredo JL, Pereira MFR, Freitas MMA, O Characterization of active sites on carbon catalysts. Ind Eng Chem Res 2007;46(12):4110–5. [25] Zielke U, Huttinger J, Hoffman WP. Surface-oxidized carbon fibers. I: Surface structure and chemistry. Carbon 1996;34(8):983–98. [26] Sing KSW, Everett DH, Haul RAW, Moscou L, Pierotti RA, Rouque´rol J, et al. Reporting physisorption data for gas/solid systems with special reference to the determination of surface area and porosity. Pure Appl Chem 1985;57(4): 603–19. [27] Monthioux M, Smith BW, Burteaux B, Claye A, Fischer JE, Luzzi DE. Sensitivity of single-wall carbon nanotubes to chemical processing: an electron microscopy investigation. Carbon 2001;39(8):1251–72. [28] Leon y Leon CA, Solar JM, Calemma V, Radovic LR. Evidence for the protonation of basal plane sites on carbon. Carbon 1992;30(5):797–811. ´ rfa˜o JJM, Pereira MFR. A novel ceria-activated [29] Faria PCC, O carbon composite for the catalytic ozonation of carboxylic acids. Catal Commun 2008;9(11–12):2121–6. [30] Andreozzi R, Insola A, Caprio V, Marotta R, Tufano V. The use of manganese dioxide as a heterogeneous catalyst for oxalic acid ozonation in aqueous solution. Appl Catal A 1996;138(1):75–81. [31] Beltra´n FJ, Rivas FJ, Montero-de-Espinosa R. A TiO2/Al2O3 catalyst to improve the ozonation of oxalic acid in water. Appl Catal B 2004;47(2):101–9. ´ rfa˜o JJM, Pereira [32] Orge CA, Sousa JPS, Gonc¸alves F, Freire C, O MFR. Development of novel mesoporous carbon materials for the catalytic ozonation of organic pollutants. Catal Lett 2009;132(1):1–9. [33] Skoumal M, Cabot P-L, Centellas F, Arias C, Rodrı´guez RM, Garrido JA, et al. Mineralization of paracetamol by ozonation catalyzed with Fe2+, Cu2+ and UVA light. Appl Catal B 2006;66(3–4):228–40. [34] Leitner V, Karpel N, Berger P, Legube B. Oxidation of amino groups by hydroxyl radicals in relation to the oxidation degree of the a-carbon. Environ Sci Technol 2002;36(14):3083–9.